Controlling oligomerization of pharmaceutical proteins

PHARMACEUTICA ACTAHELVETIAE ELSEVIER Pharmaceutica Acta Helvetiae 69 (1994) 119-126 Review Controlling oligomerization of pharmaceutical proteins Catherine H. Schein Swiss Institute for Alternatives to Animal Testing (SIA T), Technopark, Pfingstweidstrasse 30, CH-8005 Ziirich, Switzerland Received 10 July 1994 Abstract The degree of oligomerization (or in some cases aggregation) often determines the physiological half-life and uptake rate of a protein preparation. High-resolution crystal structures of insulin and other pharmacologically interesting proteins have aided in the design of mutants with altered quaternary structure and physiological uptake rates. Analysis of the contacts between natural oligomers and protein complexes can indicate sequences that may enhance protein oligomerization. These sequences can be altered to produce monomeric protein. Keywords: Interprotein contacts; Protein aggregation; Protein oligomerization; Site-directed mutagenesis 1. Introduction The limited solubility of proteins presents a major difficulty in their pharmaceutical use. Aggregation of proteins lowers the activity of preparations and their shelf life, and it interferes with accurate dosaging. Until recently, the only way to deal with this problem was to optimize the solution conditions to reflect the solubility spectrum of the protein and to use additives and cosolvents that decrease the surface tension of the buffer (Schein, 1990). The usefulness of additives is specific for both the protein and the storage conditions. For example, the irreversible aggregation of porcine growth hormone, induced by heat denaturation, vortex agitation or during renaturation after denaturing in guanidium, could be prevented by adding Tween 20 to the buffer. A detergent-like molecule, hydroxypropyl-/3-cyclodextrin, lowered aggregation induced by the first two methods but had no effect on aggregation during renaturation (Charman et al., 1993). A further problem is that suitable co-solvents may be too toxic for use in pharmaceutical preparations. If one is dealing with a cloned protein, it is now possible to alter the primary structure of the protein itself to reduce its tendency to aggregate. Even single 0031-6865/94/$07.00 © 1994 Elsevier Science B.V. All rights reserved SSDI 0 0 3 1 - 6 8 6 5 ( 9 4 ) 0 0 0 3 8 - 7 amino acid changes in very large proteins can improve solubility (Schein, 1993). However, just as the ideal solution conditions for a protein must be empirically deternfined, there is still no accurate guide to altering a protein to lower its tendency to precipitate. This is mostly due to an incomplete understanding of the molecular interactions that lead to aggregation. Protein aggregates can be studied with methods as diverse as polyacrylamide gels and molecular sizing columns to laser light scattering (Schein, 1991). However, the fine structure of protein aggregates is difficult to determine, as high resolution methods require a uniform molecular ensemble. High-resolution X-ray data of oligomeric proteins and various protein complexes, coupled with more recent results from NMR spectroscopy, are the most valuable source of information on sequences that interact. Comparative sequence analysis of well-described inter-protein interactions (i.e., from the X-ray structure of multi-domain or subunit proteins) have shown some common features of quaternary interactions (Argos, 1988). Studies of complexes with very high binding constants, such as between proteins and specific protein inhibitors or activators, or antibodies and antigens, can also help to understand irreversible aggregation. 120 C.tl. Schein / Pharmaceutica Acta Heh'etiae 69 (1994) 119-126 As every protein crystallographer knows, most proteins precipitates will never yield a crystal. Thus deriving empirical rules for interacting sequences from crystal structures that can be applied to aggregation would seem to be difficult. However, there are several examples of how a protein's aggregation tendencies can be altered by mutations predicted from the structure. Examples of how information derived from these models can be used to design proteins with altered quaternary structure and tendency to aggregate are shown below. 2. Oligomeric proteins Oligomeric proteins are ubiquitous in nature. From phage coats to collagen, the ability of individual protein chains to form higher-order structures determines the shape of the natural world. Oligomerization also controls the activity of proteins. Some enzymes (e.g., the RNase H activity of HIV reverse transcriptase (Restle et al., 1992)) and binding proteins (A cro (Mossing and Sauer, 1990) and other D N A binding proteins and transcription factors (Lamb and McKnight, 1991)) are only active as dimers or oligomers. In other cases, oligomerization serves .to either lower or alter the spectrum of activity. For example, the plasmid initiator protein R e p A is converted from an inactive dimer to a DNA-binding monomer in an ATPdependent reaction by the heat shock protein DnaK (Wickner et al., 1991). The classic example Of the control that oligomerization can exert on how a protein functions is hemoglobin (Hb), which can be described as a myoglobin tetramer. While myoglobin in muscle cells is essentially an oxygen storage protein, Hb's ability to bind and release O 2 in the tissues and CO 2 in the lungs depends on pH in the different parts of the body inducing alterations at its subunit interface (Perutz, 1970; Perutz et al., 1987). Single point mutations that disturb these changes have been demonstrated in several diseases (Nathan, 1973). The active form of some growth factors are oligomeric. IFN-y, for example, but not IFN-a or/3, is a dimer with the monomers tightly enmeshed in one another (Ealick et al., 1991). Laboratory conditions may, however, induce mulfimer formation of proteins that does not indicate their active form. Human growth hormone, for example, dimerizes in llae presence of Zn (Cunningham et al., 1991). Interleukin-8 is dimeric at the high concentrations needed for determining the N M R structure, but a chemically synthesized form that only forms a monomer is just as active (Rajarathnam et al., 1994). Aggregation is favored by a too high concen- tration of reductant in the sample buffer and incubation temperature, which may complicate interpretation of r e d u c i n g / S D S P A G E (Schein and Noteborn, 1988; Hyman and Arp, 1993). One indication that oligomerization is essential for activity is inactivation upon dilution that is reversed when the sample is reconcentrated (Thornberry et al., 1992). There are many ways that a protein may form contacts with itself and other proteins. A single mechanism will not describe even the "controlled" aggregation of proteins in Nature. The interface contacts of a single viral coat protein between itself and other proteins depend on the final position of the subunit within the shell structure (Casjens, 1985). Indeed, the original postulate of "quasi-equivalent" binding at lattice points in virus capsules was modified to "non-equivalency" when the first structure was solved at atomic dimensions. For example, the coat of Tomato Bushy Stunt Virus consists of 180 identical subunits of a 43-kDa protein which self-interact at lattice points in at least three distinct ways. Human growth hormone (huGH), a monomer, binds to each subunit of its homodimeric receptor at different positions (DeVos et aI., 1992). 3. Mechanisms for stabilizing interprotein contacts As oligomerization is a useful way of controlling activity and form, Nature has many ways of altering the quaternary structure of proteins in response to environmental changes. One of the most important points in complex formation is how the multimer structure is maintained after the initial contacts occur. Oligomers and protein complexes are usually stabilized after forming by one of the following mechanisms: 1. Multiply coordinated interaction at specific sequences based on hydrophobic interaction (van der Waals forces), opposite charge attraction, or hydrogen bonding. Hydrophobic regions are often found to mediate oligomer formation. For example, the interface region of the tyrosyl-tRNA synthetase dimer is largely hydrophobic and can be destabilized by mutating a single phenylalanine residue (Ward et al., 1987); mutating lysine 97 to the apolar residue valine in interleukin-1/3 greatly increases its aggregation in E. coli (Churynk et al., 1993). Heat shock protein trimer formation at high temperatures is dependent on hydrophobic heptad repeats at its amino terminus that are masked at lower temperatures by sequences at its C-terminus (Rabindran et al., 1993). A study of the amino acid contacts at C.H. Schein/Pharmaceutica Acta Helvetiae 69 (1994) 119-126 subunit and protein domain interfaces (Argos, 1988) indicated that phenylalanine side chains are likely to self-interact and are frequently found at subunit interfaces. Other amino acids with a high probability of self-interaction are methionine and histidine. Hydrophobic interactions stabilize many of the protein complexes with high binding constants. Most of the residues at the interface of the complex of cyclosporin with cyclophilin, for example, are apolar (Spitzfaden et al., 1992). Several protein complexes have been particularly well defined. The basis of the tight binding between paired molecules is an assembly of individual residue interactions, each of which is multiply coordinated. Thus alteration of a single amino acid residue may affect several different binding sites. For example, in the h u G H / r e c e p t o r complex, the receptor Arg43 forms hydrogen bonds with two or three residues on the other receptor subunit and on huGH. Many of the interactions are between hydrophobic side chain residues (De Vos et al., 1992). Hirudin binds to and inhibits thrombin in a 1 : 1 M complex with a binding constant of 2 x 10 -14 M. Nearly half of hirudin's residues are in direct contact with multiple sites on thrombin (Rydel et al., 1991). 2. Interaction with a metal ion, some other cofactor, or nucleic acids. Divalent cations can influence solubility even at very low concentrations. Both Ca 2+ and Mg 2+ at concentrations between 1-20 mM encourage self-association of sea urchin hyalin (Robinson, 1988). Zn 2+ aids in insulin solubilization as well as crystallization (Markussen et al., 1988). Some transcription factors do not dimerize until they come in contact with their DNA recognition site. 3. Chemical alteration of surface residues, proteolysis, or partial denaturation of the protein to allow usually shielded residues to be exposed to intermolecular contact. A common form of dimerization is of course, the oxidation of cysteine residues to form a disulfide bond. However, other posttranslational modifications can stabilize dimeric structures. For example, in the presence of IFN-y, the transcription factor Stat91 is phosphorylated at a tyrosine residue. This phosphorylation mediates the formation of an active, DNA-binding dimeric form of Stat91 from the inactive monomers (Shuai et al., 1994). Removal of a single tyrosine residue of a protein that regulates cell adhesion abolishes its ability to enhance cell-matrix adhesion. The tyrosine is located at the C-terminus of the molecule in a consensus 121 site for tyrosine kinase phosphorylation (in the one letter code, RIVEILY) (Pullmann and Bodmer, 1992). Proteolysis is, of course, the basis for blood clotting. Fibrinogen is cleaved by thrombin to reveal the interactive sites that lead to formation of fibrin multimers. Fibrin clots can be dissolved by adding GPRG, the N-termini of fibrinogen molecules released by thrombin cleavage (Dietler, 1985) and the peptide RGDS reverses platelet aggregation (Krishnamurthi et al., 1989). Proteolyzed proteins can also be less likely to aggregate. For example, serine protease inhibitors (the "Serpins") undergo a drastic conformational change when proteolytically cleaved at their active site. Proteolytically modified Serpins are both more stable (the guanidinium concentration required for denaturation is about twice that required for the non-cleaved molecules) and have a lower tendency to precipitate at high temperatures than the uncleaved proteins. It is believed that the conformational change is part of the inhibitory mechanism of the Serpins, as the structurally related protein ovalbumin, which is not a protease inhibitor, does not change its conformation or stability after protease treatment (Bruch et al., 1988). Normal vimentin filament assembly can be prevented by proteolysis, deimination of arginine residues, or phosphor,lation (Horkovics-Kovats and Traub, 1990). All three mechanisms control the degree of oligomerization of insulin in vivo and in vitro. All three of these mechanisms can induce the formation of slowly dissolving or insoluble oligomeric forms of insulin. Insulin in preparations used for the control of human diabetes is usually in the form of a zinc-containing hexamer (Brader and Dunn, 1991) (mechanism 2 above), which is slowly absorbed into the blood stream. In the absence of Zn, human insulin is a trimer or tetramer. Recent studies have shown that even the dimeric form of insulin characterized from crystal studies is in a "locked" state and is incapable of binding to the insulin receptor until it is converted to an "unlocked" monomer with a less ordered B-chain carboxy terminal (Hua et al., 1991). Although slow uptake of oligomers is desirable for daily controlled use, active monomers are needed for rapid absorption during emergency situations (glucose shock). Faster acting, monomeric preparations of insulin have been prepared by single amino acid changes at the oligomer interface seen from crystal structures of several oligomeric forms. One way to avoid oligomerization was found to be incorporation of negative charges at positions that are in close contact at the C.H. Schein / Pharmaceutica Acta Heh:etiae 69 (1994) 119-126 122 O~ I H , , cii _. "1'--.-....,/~0 HzC) or H2N-R Hydrolysis Dimerformation ]Fig. 1. Mechanism of dirner formation in insulin through dearnination of asparagine A-21 (adapted from Markussen ¢ta]., ]988). oligomer surface in the crystal structure of hexameric insulin (SerB9Asp or ThrB27Glu), as the like charges then repel each other. These mutants were as active as the wild-type insulin; however, their absorption times were considerably lower, and they remained monomeric even at pharmaceutically useful concentrations (0.6 mM) (Brange et al., 1988). Other changes in the Cterminal region of the B-chain, predominately replacing Pro28, also reduce self association (Brems et al., 1992b). Finally, the tendency of insulin to form less-active dimers during storage is at least partially due to chemical alteration of its primary structure (mechanism 3). Insulin's half-life in acidic solution can be increased by replacing asparagine at position A-21, which deamidates and leads to dimer formation (Fig. 1), with Gly, Ser, Thr, Asp, His, and Arg (Markussen et al., 1988). Many single amino acid changes that lower the conformation stability of insulin increase formation of a disulfide-linked multimer during storage at 50°C. Stability can be increased by mutating histidine B10 to aspartic acid (DBI°). The most stable insulin construct was a triple mutant combining the D Bt° mutation with replacement of proline at B28 with aspartic acid and lysine at B29 with proline (Brems et al., 1992a). This last example illustrates that although one can identify contact areas of the protein for mutation, the best amino acid for the alteration must still be determined experimentally. 4. Controlling hemoglobin subunit interaction and aggregation Normal human hemoglobin is an extremely soluble protein; the concentration in red blood cells is about 60% by weight. Many natural mutants of human hemoglobin (Hb) that increase aggregation (or reduce the concentration at which the protein forms a gel) are known (Nathan, 1973). Replacement of glutamic acid /36 to a valine near the amino terminus of the hemoglobin /3-chain (VHLTP(E)EKAVTA), leads un- der the correct oxygenation conditions to the hemoglobin polymerization and red blood cell shape change of sickle cell anemia. The sickle cell /3Glu6Val point mutation can be suppressed by an /3Asp73Asn mutation. The concentration of protein needed for gelation of the double mutant is approximately that of when the protein is in the deoxy-conformation (Nathan, 1973). The cooperativity between the four subunits of mammalian hemoglobin is essential for its ability to respond appropriately to differences in 0 2 tension and pH between the lungs and the tissues to which it must supply oxygen and remove excess CO 2. Maintaining these cooperative interactions is the biggest problem in using pure hemoglobin as a blood substitute. In red blood cells, hemoglobin contains one molecule of 2,3diphosphoglycerate (DPG), which binds at the tetramer interface and lowers the oxygen affinity of the protein. Free hemoglobin cannot bind DPG. Thus, the oxygen affinity of the protein is too high to release oxygen efficiently in the tissues. In addition, the free tetrameric protein dissociates into a/3-dimers, which have a very short in vivo half-life. Both of these problems have been dealt with to some extent in a mutant protein produced in E. coil. To stabilize the tetramer structure, a single protein coding an a-chain-fused dimer was made by ligating two copies of the gene sequence to each other (with the N-terminal of the second attached to the C-terminal of the preceding chain). A single point mutation in the /3-chains (Asnl08 to Lys) reduced the affinity for oxygen to nearly that of red blood cell hemoglobin (Looker et al., 1992). Despite the progress, the half-life of this hemoglobin blood substitute in studies in dogs was only about 1 h, compared to the 120-day average life span of normal red blood cells. 5. Preventing oligomerization Although there are few oligomeric proteins as well studied as insulin or hemoglobin, there are many examples of specifically mutating a protein to eliminate dimer formation. If a structure is available, mutation of the contact site can be used to generate monomeric proteins. The general categories of replacement are: Remouing surface phenylalanines or other hydrophobic residues. Dimerization of T4 endonuclease V was prevented by changing one or two adjacent Phe residues, predicted from a model structure to be at the dimer C.H. Schein/ PharmaceuticaActa Helvetiae 69 (1994) 119-126 interface, to Leu (NicKell and Lloyd, 1991). Similarly, replacement of Phe164 in the dimeric enzyme tyrosyltRNA synthetase with charged residues causes reversible dissociation of the protein into inactive monomers (Ward et al., 1987). Altering charged residues that specifically interact. Certain side chains are most likely to form a pair with only one or two other amino acids (Argos, 1988). For example, arginine is much more likely to pair with aspartate than with glutamate, while lysine pairs equally often with either negatively charged residue. Perhaps the Asp-Arg pairing is more stable, as amino acid comparisons of thermophilic proteins with their mesophilic counterparts indicates a tendency to replace Lys and Glu with Arg and Asp (as well as a preference for the hydrophobic amino acids Phe, Val and lie over Leu, Ala and Met) (Zuber, 1988). However, when charged residues were inserted in place of the surface residue Phe164 of tyrosyl tRNA transferase (see above), heterodimers of the mutants Glu164Lys164 and Asp164-Lys164 had a much lower K m for tyrosine (which was taken to indicate that the dimer was more stable) than either Glu164-Arg164 or Asp164-Arg164 (Ward et al., 1987). Altering cysteine residues. Covalent dimers frequently form through the intersubunit oxidation of Cys residues to cystines. Alteration of Cys residues known to be at the surface of a protein may thus prevent irreversible aggregation. For example, autoreduction and dimerization of yeast cytochrome c was eliminated by substituting a Thr for Cys at position 107 (Shaw, 1987). During high-level periplasmic secretion of humanized Fab fragments from E. coli, Carter et al. (1992) observed that the heavy and light chains were covalently linked (to what they call Fab'), but almost no hinge region dimer (Fab~) formed spontaneously if a single hinge Cys residue (Cys-AlaAla) was encoded. When the original hinge region of human IgG1 (Cys-Pro-Pro-Cys) was used, about 25% of the Fab' molecules formed dimers. Alternatively, a dimer structure can be stabilized by adding surface cysteine residues. Sauer et al. (1986), for example, were able to increase the stability of the phage A-repressor dimer by replacing Tyr-88 at the dimer interface with Cys. The control construct, where the interior residue Tyr85 was replaced with Cys, altered and destabilized the structure of the protein. Bovine seminal ribonuclease forms a covalent dimer mediated by disulfide bridges at adjacent cysteines (Capasso et al., 1983). A dimer of the closely related 123 RNaseA can be induced to form by incorporating the double cysteine residues into its structure at the same position (Sunai Raillard and Steven S. Benner, Organic Chemistry, ETH Ziirich, unpublished). Altering other distinct residues. The half-life at 100°C and pH 6 of yeast triose phosphate isomerase, a dimeric enzyme, could be nearly doubled by changing asparagine residues at the subunit interface to threonine or isoleucine. Conversion of the asparagine residue to aspartic acid (the product of a suspected deamidation reaction), however, greatly decreased the dissociation temperature. This was presumably because the negative charges destabilize the dimer, as dilution-induced dissociation of the mutant dimer was also enhanced (Ahern et al., 1987). A single amino acid change (D152H) in the extracellular domain of the human growth hormone (hGH) receptor prevents its dimerization and reduces its ability to bind hGH. Patients identified with this mutation (Laron syndrome) have high serum hGH levels but show most of the symptoms of hGH deficiency (Duquesnoy et al., 1994). Both the crystal and NMR structures of IL-8 indicated a homodimeric structure. Fully active, monomeric IL-8 was synthesized with a methylated amino group on l_eu25 (L25NMe-IL8). The methylation prevented formation of one of the hydrogen bonds seen across the dimer interface with IL-8 (Rajarathnam et al., 1994). Replacing whole areas of proteins with non-aggregating areas. Very small areas of certain proteins have been shown to mediate tight contacts between large protein molecules. These sequences are often referred to as "sticky" and in some cases as "protein velcro" (even when the partner "hooks" have not been identified). The best studied of these were originally identified in fibronectin and laminin, proteins required for the adherence of contact-dependent mammalian cells to surfaces. Adherence sequences typically contain at least two charged residues separated or surrounded by glycines and hydrophobic residues; the two most common ones are -LRE- and RGD (V, S or T) (D'Souza et al., 1991; Yamada, 1991). The solution structure of kistrin, a 68-residue protein that contains an RGD motif and inhibits platelet aggregation, has recently been solved by NMR. The RGD residues occur on the surface of the protein in an extended loop region which should allow them free access to complementary sequences on the cell surface (Adler et al., 1991). Deletion of larger areas of two other proteins offers 124 C.H. Schein / Pharmaceutica Acta Helvetiae 69 (1994) 119-126 A ~,,'~N O(~ ~4 (~ ^ A /~) / N- O~ -~- ~ } ,~ N N 171 c) H 6 ~ ~~ J ~ l ~ v ~ Dimer formedthrough the C-terminalresidues of two identicalmonomers. {DEGKNRS} Monomerformed by duplicationof residues and linker insertionat {AGST} the C-terminus. Fig. 2. Making a dimer into a monomer by duplicating the dimer interface region in the subunit (adapted from Mossing and Sauer, 1990). indirect evidence for the importance of surface phenylalanine residues in dimer contacts. R e p l a c e m e n t of the hydrophobic (-Glu-Gly-Asn-PhePhe-Gly-Lys-Ile-IleAsp-Tyr-Ile-Lys-Leu-Met-Phe-His-His-Trp-Phe-Gly) carboxy terminal amino acids of E. coli penicillin-binding protein 5 with a shorter hydrophilic sequence (-IleArg-Arg-Pro-Ala-Ala-Lys-Leu-Asp) made the protein water soluble and allowed crystallization (Ferreira et al., 1988). Deleting 13 residues (Asp-Val-Leu-Asn-AspAsn-Leu-Leu-Arg-PhePhe-Val-Ala) f r o m a-casein makes the molecule more soluble (Farrell et al., 1988). A d d i n g " solubilizing linkers" to proteins. It is sometimes possible to alter the solubility of the whole protein by adding a linker to it. For example, two IgG-binding domains added to h u m a n I G F - I increased its solubility and improved refolding (Samuelsson et al., 1991) and m a m m a l i a n proteins produced in E. coil were more soluble if fused to ubiquitin (Koken et al., 1993) or thioredoxin (LaVallie et al., 1993). Dimer formation can also be prevented by duplicating intramolecularly a sequence that forms (or is proposed to form) the dimer contact area, so that the individual chain reacts with itself instead of a second molecule. One example of this is alteration of the A-cro's Cterminus so that the protein remains monomeric (Fig. 2) (Mossing and Sauer, 1990). A model of the dimer proposed that residues 52-58 could form a /3-sheet contact region. Alteration of each residue indicated that the only one important for activity is the Phe58. A duplicate of residues 54-56 preceded by two " l o o p " residues (i.e., ones that would be expected to form a "/3-turn" in a protein) was inserted, so the end of the protein would loop back on itself instead of being free to form a dimer with another monomer. Sedimentation equilibrium studies showed the dimer no longer formed. The mutant m o n o m e r was considerably more stable than the wt dimer (Td of 58°C vs 45°C for the wt) but much weaker in binding to DNA. Initial N M R studies indicated a solution structure identical with that of the wt-dimer protein. 6. Conclusions Despite the many examples where a protein has been successfully mutated to control aggregation, the actual solubilization of problem proteins still requires trial and error. It should be noted that unexpected and drastic changes in the protein may occur even when residues known from a crystal structure to be at the interface surface are altered. For example, conversion of a single residue, aspartic acid 199, to asparagine at the trimer interface of chloramphenicol acetyl transferase (CAT) resulted in changes up to 20A away from the site of the mutation that nearly eliminated catalytic activity. The mutated inactive trimer was, however, nearly as thermostable as the wild-type (WT). A crystal structure revealed that Argl8, which forms a salt bridge with Asp199 in the WT-CAT, formed new hydrogen bonds with Glul01 and a water molecule. The A r g l 8 amide groups were replaced at the active site by two water molecules, accounting for the loss of activity (Gibbs et al., 1990). Replacement of a single amino acid residue (histidine 111 with aspartic acid) in human interferon-y completed eliminated biological activity by eliminating receptor recognition (Lunn et al., 1992). This histidine residue is not conserved; it would be interesting to see if it could be responsible for the species specificity of interferon-ys. There are some general principles that can be used to predict and change sequences that are involved in oligomerization. A good structure can greatly reduce the number of experiments needed by defining the areas of the protein at the contact site. Developing a screening system for solubility can also shorten the time required by allowing the use of linker scanning and other random methods of amino acid replacement. References Adler, M., Lazarus, R.A., Dennis, M.S. and Wagner, G. (1991) Solution structure of Kistrin, a potent platelet aggregation inhibitor and GP lib-Ilia antagonist. Science 253, 445-448. Ahern, T.J., Casal, J.I., Petsko, G.A. and Klibanov, A.M. (1987) Control of oligomeric protein stability by protein engineering. Proc. Natl. Acad. Sci. USA 84, 675-679. C.H. Schein / Pharmaceutica Acta Helvetiae 69 (1994) 119-126 Argos, P. (1988) An investigation of protein subunit and domain interfaces. Protein Eng. 2, 101-113. Brader, M.L. and Dunn, M.F. (1991) Insulin hexamers: new conformations and applications. Trends Biochem. Sci. 16, 341-345. Brange, J., Ribel, U., Hansen, J.F., Dodson, G., Hansen, M.T., Havelund, S., Melberg, S.G., Norris, F., Norris, K., Snel, L., S~renson, A.R. and Voigt, H.O. (1988) Monomeric insulins obtained by protein engineering and their medical implications. Nature 333, 679-682. Brems, D.N., Brown, P.L., Bryant, C., Chance, R.E., Green, L.K., Long, H.B., Miller, A.A., Millican, R., Shields, J.E. and Frank, B.H. (1992a) Improved insulin stability through amino acid substitutions. Protein Eng. 5, 519-525. Brems, D.N., Alter, L.A., Beckage, M.J., Chance, R.E., DiMarchi, R.D., Green, L.K., Long, H.B., Pekar, A.H., Shields, J.E. and Frank, B.H. (1992b) Altering the association properties of insulin by amino acid replacement. Protein Eng. 5, 527-533. Bruch, M., Weiss, V. and Engel, J. (1988) Plasma serine proteinase inhibitors (serpins) exhibit major conformational changes and a large increase in conformational stability upon cleavage at their reactive sites. J. Biol. Chem. 263, 16626-16630. Capasso, S., Giordano, F., Mattia, C.A., Mazzarella, L. and Zagari, A. (1983) Refinement of the structure of bovine seminal ribonuclease. Biopolymers 22, 327-332. Carter, P., Kelley, R.F., Rodrigues, M.L., Snedicor, B., Cavarrubias, M., Velligan, M.D., Wong, W.L.T., Rowland, A.M., Kotts, C.E., Carver, M.E., Yang, M., BoureU, J.H., Shepard, H.M. and Henner, D. (1992) High level Escherichia coli expression and production of a bivalent humanized antibody fragment. Biotechnology 10, 163-167. Casjens, S. (Ed.) (1985) Virus Structure and Assembly. Jones and Bartlett Publishers, Boston, MA. Charman, S.A., Mason, K.L. and Charman, W.N. (1993) Techniques for assessing the effects of pharmaceutical excipients on the aggregation of porcine growth hormone. Pharm. Res. 10, 954-962. Churynk, B.A., Evans, J., Lillquist, J., Young, P. and Wetzel, R. (1993) Inclusion body formation and protein stability in sequence variants of interleukin-lfl. J. Biol. Chem. 268, 18053-18063. Cunningham, B.C., Mulkerrin, M.G. and Wells, J.A. (1991) Dimerization of human growth hormone by zinc. Science 253, 545-548. De Vos, A.M., Ultsch, M. and Kossiakoff, A. (1992) Human growth hormone and the extracellular domaill of its receptor: crystal structure of the complex. Science 255, 306-312. Dietler, G. (1985) Fibrin polymerization: a combination of light scattering with measurements of fibrinopeptide release. Dissertation ETH No. 7819, Department of Physics, Swiss Federal Institute of Technology, Ziirich. D'Souza, S.E., Ginsberg, M.H. and Plow, E.F. (1991) Arginyl-glycylaspartic acid (RGD): a cell adhesion motif. Trends Biochem. Sci. 16, 246-250. Duquesnoy, P., Sobrier, M.L., Duriez, B., Savage, M.O., Preece, M.A., Craescu, C.T., Blouquit, Y., Goosens, M. and Amselem, S. (1994) A single amino acid substitution in the exoplasmic domain of the human growth hormone receptor confers familial GH resistance (Laron syndrome) with positive GH-binding activity by abolishing receptor homodimerization. EMBO J. 13, 1386-1395. Ealick, S.E., Cook, W.J., Vijay-Kumar, S., Carson, M., Nagabhushan, T.L., Trotta, P.P. and Bugg, C.E. (1991) Three-dimensional structure of recombinant human interferon-3,. Science 252, 698-701. Farrell, H.M., Kumosinski, T.F., Pulaski, P. and Thompson, M.P. (1988) Calcium-induced associations of the caseins: a thermody- 125 namic linkage approach to precipitation and resolubilization. Arch. Biochem. Biophys. 265, 146-158. Ferreira, L.C.S., Schwarz, U., Keck, W., Charlier, P., Dideberg, O. and Ghuysen, J.-M. (1988) Properties and crystallization of a genetically engineered, water-soluble derivative of penicillin-binding protein 5 of Escherichia coli K12. Eur. J. Biochem. 171, 11-16. Gibbs, M.R., Moody, P.C.E. and Leslie, A.G.W. (1990) Crystal structure of the aspartic acid-199-asparagine mutant of chloramphenicol acetyltransferase to 2.35-,~ resolution: structural consequences of disruption of a buried salt bridge. Biochemistry 29, 11261-11265. Horkovics-Kovats, S. and Traub, P. (1990) Specific interaction of the intermediate filament protein vimentin and its isolated Nterminus with negatively charged phospholipids as determined by vesicle aggregation, fusion, and leakage measurements. Biochemistry 29, 8652-8657. Hua, Q.X., Shoelson, S.E., Kochoyan, M. and Weiss, M.A. (1991) Receptor binding redefined by a structural switch in a mutant human insulin. Nature 354, 238-341. Hyman, M.R. and Arp, D.J. (1993) An electrophoretic study of the thermal and reductant-dependent aggregation of the 27 kDa component of ammonia monooxgenase from Nitrosornas europea. Electrophoresis 14, 619-627. Koken, M.H.M., Odijk, H.H.M., van Duin, M., Fornerod, M. and Hoeijmakers, J.H.J. (1993) Augmentation of protein production by a combination of the T7 RNA polymerase system and ubiquitin fusion: overproduction of the human DNA repair protein ERCC1 as a ubiquitin fusion protein in E. coli. Biochem. Biophys. Res. Commun. 195, 643-653. Krishnamurthi, S., Dickens, T.A., Patel, Y., Wheeler-Jones, C.P.D. and Kakkar, V.V. (1989) The fibrinogen-derived peptide (RGDS) prevents proteolytic degradation of protein kinase C in platelets by inhibiting platelet aggregation. Biochem. Biophys. Res. Commun. 163, 1256-1264. Lamb, P. and McKnight, S.L. (1991) Diversity and specificity in transcriptional regulation: the benefits of heterotypic dimerization. Trends Biochem. Sci. 16, 417-422. LaVallie, E.R., DiBlasio, E.A., Kovacic, S., Grant, K.L., Schendal, P.F. and McCoy, J.M. (1993) A thioredoxin gene fusion expression system that circumvents inclusion body formation in the E. coli cytoplasm. Biotechnology 11, 187-193. Looker, D., Abbott-Brown, D., Cozart, P., Durfee, S., Hoffman, S., Mathews, A.J., Miller-Roehrich, J., Shoemaker, S., Trimble, S., Fermi, G., Komiyama, N.H., Nagai, K. and Stetler, G.L. (1992) A human recombinant haemoglobin designed for use as a blood substitute. Nature 356, 258-260. Lunn, C.A., Fossetta, J., Dalgarno, D., Murgolo, N., Windsor, W., Zavodny, P.J., Narula, S.K. and Lundell, D. (1992) A point mutation of human interferon-y abolishes receptor recognition. Prot. Eng. 5, 253-257. Markussen, J., Diers, I., Hougaard, P., Langkjaer, L., Norris, K., Snel, L., Sorensen, A.R., Sorensen, E. and Voigt, H.O. (1988) Soluble, prolonged-acting insulin derivatives. III. Degree of protraction, crystallizability, and chemical stability of insulins substituted in positions A21, B13, B23, B27 and B30. Protein Eng. 2, 157-166. Mossing, M.C. and Sauer, R.T. (1990) Stable, monomeric variants of A cro obtained by insertion of a designed /3-hairpin sequence. Science 250, 1712-1715. Nathan D.G. (1973) Hemolytic anemias. III. Hemoglobinopathic 126 C.H. Schein / Pharmaceutica Acta Heh.,etiae 69 (1994) 119-126 disorders. In Beck, W.S. (Ed.), Hematology. Harvard Pathophysiology Series Volume 1. MIT Press, Cambridge, pp. 199-216. Nickell, C. and Lloyd, S. (1991) Mutations in endonuclease V that affect both protein-protein association and target site location. Biochemistry 30, 8638-8648. Perutz, M.F. (1970) Stereochemistry of cooperative effects in hemoglobin. Nature 228, 726-734. Perutz, M.F., Fermi, G., Luisi, B., Shaanan, B. and Liddington, R.C. (1987) Stereochemistry of cooperative mechanisms in hemoglobin. Acc. Chem. Res. 20, 309-321. Pullmann, W.E. and Bodmer, W.F. (1992) Cloning and characterization of a gene that regulates cell adhesion. Nature 356, 529-532. Rabindran, S.K., Haroun, R.I., Clos, J., Wisniewski, J. and Wu, C. (1993). Regulation of heat shock factor trimer formation: role of a conserved leucine zipper. Science 259, 230-234. Rajarathnam, K., Sykes, B.D., Kay, C.M., Dewald, B., Geiser, T., Baggiolini, M. and Clark-Lewis, I. (1994) Neutrophil activation by monomeric interleukin 8. Science 264, 90-92. Restle, T., Miiller, B. and Goody, R.S. (1992) RNase H activity of HIV reverse transcriptases is confined exclusively to the dimeric forms. FEBS 300, 97-100. Robinson, J.J. (1988) Roles for Ca 2+, Mg 2+ and NaCI in modulating the self-association reaction of hyalin, a major protein component of the sea-urchin extraembryonic hyaline layer. Biochem. J. 256, 225-228. Rydel, T.J., Tulinsky, A., Bode, W. and Huber, R. (1991) Refined structure of the birudin-thrombin structure. J. Mol. Biol. 221, 583-601. Samuelsson, E., Wadensten, H., Hartmanis, M., Moks, T. and Uhl6n, M. (1991) Facilitated in vitro refolding of human recombinant insulin-like growth factor I using a solubilizing fusion partner. Biotechnology 9, 363-366. Sauer, R.T., Stearman, R.S., Weiss, M.A., Jeitler-Nilsson, A.. Suchanek, E.G. and Pabo, C.O. (1986) An engineered intersubunit disulfide enhances the stability and DNA binding of the N-terminal domain of 1 repressor. Biochemistry 25, 5992-5998. Schein, C.H. (1990) Solubility as a function of protein structure and solvent components. Biotechnology 8, 308-317. View publication stats Schein, C.H. (1991) Physical methods and models for the study of protein aggregation. In Georgiou, G. and de Bernardez-Clark, E. (Eds.), Protein Refolding. A.C.S. Books, Washington, DC, pp. 21-34. Schein, C.H. (1993) Solubility and secretability. Curr. Opinions Biotechnol. 4, 456-461. Schein, C.H. and Noteborn, M.H.M. (1988) Production of soluble recombinant mammalian proteins in Escherichia coli is favored by lower growth temperature. Biotechnology 6, 291-294. Shaw, W.V. (1987) Protein engineering. The design, synthesis and characterization of factitious proteins. Biochem. J. 246, 1-17. Shuai, K., Horvath, C.M., Tsai Huang, L.H., Qureshi, S.A., Cowburn, D. and Darnell, J.E. (1994) Interferon activation of the transcription factor Stat91 involves dimerization through SH2phosphotyrosyl peptide interactions. Cell 76, 821-828. Spitzfaden, C., Weber, H.-P., Braun, W., Kallen, J., Wider, G., Widmer, H., Walkinshaw, M.D. and Wiithrich, K. (1992) Cyclosporin A-Cyclophilin complex formation. FEBS Letters 300, 291.-300. Thornberry, N.A., Bull, H.G., Calaycay, J.R., Chapman, K.T., Howard, A.D., Kostura, M.J. et al. (1992) A novel heterodimeric cystein protease is required for interleukin-1/3 processing in monocytes. Nature 356, 768-774. Ward, W.H.J., Jones, D.H. and Ferscht, A.R. (1987) Effects of engineering complementary charged residues into the hydrophobic subunit interface of tyrosyl-tRNA synthetase. Biochemistry 26, 4131-4138. Wickner, S., Hoskins, J. and McKenney, K. (1991) Monomerization of RepA dimers by heat shock proteins activates binding to DNA replication origin. Proc. Natl. Acad. Sci. USA 88, 7903-7907. Yamada, K.M. (1991) Adhesive recognition sequences. J. Biol. Chem. 266, 12809-12812. Zuber, H. (1988) Temperature adaptation of lactate dehydrogenase. Structural, functional and genetic aspects. Biophys. Chem. 29, 171-179.