Conformational impurity of disulfide proteins: Detection, quantification, and properties

ANALYTICAL BIOCHEMISTRY Analytical Biochemistry 342 (2005) 78–85 www.elsevier.com/locate/yabio

Conformational impurity of disulWde proteins: Detection, quantiWcation, and properties Jui-Yoa Chang ¤, Bao-Yuan Lu, Li Li Center for Protein Chemistry, Brown Foundation Institute of Molecular Medicine for the Prevention of Human Diseases, University of Texas, Houston, TX 77030, USA Received 3 February 2005 Available online 9 April 2005

Abstract The conformations of native proteins are in principle, and in most cases, dictated by the law of thermodynamics. Accordingly, a native protein must always exist in equilibrium with a minor concentration of nonnative (denatured) conformational isomers even at nondenaturing conditions. The presence of an inWnitesimal quantity of nonnative conformational isomers at physiological conditions is biologically relevant due to their propensity to aggregate, which is an underlying cause of many neurodegenerative diseases. However, their detection and quantiWcation are inherently diYcult. In this article, we describe a simple strategy using the technique of disulWde scrambling to identify and quantify such minute concentrations of nonnative isomers. It is demonstrated that even for small stable proteins such as epidermal growth factor and hirudin, approximately 1% of heterogeneous nonnative isomers coexist with the native proteins under physiological conditions.  2005 Elsevier Inc. All rights reserved. Keywords: Conformational impurity; DisulWde isomerization; Scrambled isomers of hirudin; Scrambled isomers of EGF; Scrambled isomers of
-lactalbumin; Ribonuclease A; Tick anticoagulant peptide

The mechanism of protein folding is governed by the rule of thermodynamics. The native structure is typically represented by the conformation that possesses the lowest free energy [1]. In this setting and in view of thermodynamic considerations, one can hypothesize that no polypeptide chain will fold 100% mathematically into one single (native) structure. Under physiological conditions, a native protein must always exist in a state of equilibrium with a small fraction of nonnative isomers. One may further stipulate that the concentration of nonnative proteins must be a function of the stability of the native structure. The more stable the native conformation, the lower the concentration of the nonnative isomers. This phenomenon can be observed during the

*

Corresponding author. Fax: +1 713 500 2424. E-mail address: [email protected] (J.-Y. Chang).

0003-2697/$ - see front matter  2005 Elsevier Inc. All rights reserved. doi:10.1016/j.ab.2005.03.038

course of studying the oxidative folding pathway of disulWde proteins. It has been shown that some proteins are unable to form the native structure quantitatively (>99%) even under highly favorable folding conditions. An extreme example is insulin-like growth factor (IGF1).1 Oxidative folding of IGF-1 at physiological pH generates two isomers, the native IGF-1 (N-IGF) and a scrambled species (IGF-s), with a molar ratio of 60:40 [2–5]. Another outstanding case is potato carboxypeptidase inhibitor (PCI), in which approximately 3–4% of nonnative isomers exist in equilibrium with the native 1 Abbreviations used: IGF-1, insulin-like growth factor; N-IGF, native IGF-1; IGF-s, IGF scrambled species; PCI, potato carboxypeptidase inhibitor; TAP, tick anticoagulant peptide; RNase A, ribonuclease A;
LA,
-lactalbumin; TFA, triXuoroacetic acid; DTT, dithiothreitol; EGF, epidermal growth factor; PBS, phosphate-buVered saline.

Conformational impurity of disulWde proteins / J.-Y. Chang et al. / Anal. Biochem. 342 (2005) 78–85

conformation at nondenaturing conditions [6,7]. For most proteins, the nonnative isomers may account for less than 0.1–1% under physiological conditions. However, attempts to identify and detect such minute concentrations of nonnative conformational isomers are complicated by a number of factors. First, separation of the native and nonnative proteins is a major hurdle. Nonnative proteins consist of heterogeneous isomers [8]. They usually exist in rapid equilibrium with the native structure (there are exceptions in cases of kinetically partitioned isomers) and elute collectively by most chromatographic techniques. Second, identiWcation of nonnative conformational isomers is elusive even if they can be chromatographically fractionated from the native species. Most nonnative isomers are devoid of biological function and not recognizable by antibodies directed against the native structure. Third, the detection limit of the techniques currently available is another key barrier to the identiWcation of nonnative isomers. For the majority of proteins that do fold into multiple structures, the nonnative species may constitute less than 1% of the total protein under physiological conditions. This 1% of nonnative proteins further comprises heterogeneous isomers. Ultimately, a single isomer may represent far less than 0.1% of the total protein analyzed. It is extremely diYcult for most chromatographic systems to pick up fractions that are less than 0.1–0.2% of the predominant one unless their identity is known and the sample is heavily overloaded. Although trace concentrations of conformational impurities can be detected by the methods of NMR hydrogen exchange [9–12], their isolation is unfeasible. In this article, we describe a simple method for measuring the minor nonnative isomers equilibrated with the predominant native conformation at physiological pH and temperature. The method involves the technique of disulWde scrambling along with HPLC analysis of protein isomers [13,14]. It is applicable only to proteins containing more than two disulWde bonds. This method is demonstrated here with seven disulWde-containing proteins. The heterogeneity and concentration of detectable nonnative scrambled isomers depend on the stability of the native conformation and vary from protein to protein. The detection limit of a single nonnative isomer relative to the native species is approximately 0.01%.

Materials and methods Materials Recombinant human epidermal growth factor was supplied by the Protein Institute (Houston, TX, USA). Recombinant hirudin and tick anticoagulant peptide (TAP) were kindly provided by Novartis (Basel, Switzerland). Hen egg lysozyme (L-6876), ribonuclease A (RNase A), and calcium-depleted bovine
-lactalbumin

79

(
LA, L-6010) were obtained from Sigma (St. Louis, MO, USA). All proteins were repuriWed by HPLC using the conditions described below with a Wnal purity of greater than 99.8%. Cysteine (Cys), trypsin, and chymotrypsin were also purchased from Sigma. Thermal denaturation of disulWde proteins at physiological pH Native disulWde proteins (25 g/50 l) were dissolved in phosphate buVer (50 mM, pH 7.4, 140 mM NaCl) containing 200, 75, or 25 M of Cys as thiol catalyst to promote disulWde isomerization. Samples containing native disulWde proteins but devoid of Cys were analyzed in parallel as controls. The samples were incubated at 22, 37, 50, and 60 °C for the indicated time period. Isomerization of disulWde bonds was quenched by the introduction of an equal volume of 4% aqueous triXuoroacetic acid (TFA). Acid-quenched samples were analyzed by HPLC. Thermal denaturation of disulWde proteins at pH 8.4 (Tris–HCl, 50 mM) was performed using the same protocol. Cys was used here as the thiol catalyst because the reduced form of Cys represents the most abundant endogenous thiols in human serum [15]. Thermal denaturation of disulWde proteins in human serum Native intact hirudin (Hir 1–65) and native hirudin core domain (Hir 1–49) (0.5 mg/ml) were dissolved in undiluted human serum. The serum samples were incubated at 37, 50, and 60 °C for a time period up to 24 h. A serum sample without inclusion of hirudin was processed as the control. All samples were quenched in a time course manner by mixing aliquots of reaction samples with a double volume of 4% aqueous TFA and were further diluted with another 2 volumes of 0.1% TFA. The samples were centrifuged at 16,000g for 10 min. The supernatant was Wltered by passing through a 0.2-m polyvinylidene Xuoride syringe Wlter (4 mm diameter). Then the Wltrate was analyzed by HPLC or stored at ¡20 °C. Mild reduction of scrambled proteins A mixture of native and scrambled proteins (10 g/ 20 l) generated by thermal denaturation in the phosphate buVer (pH 7.4) were treated directly with 1 mM of dithiothreitol (DTT) at 22 °C for 10 min. The samples were quenched by mixing with an equal volume of 4% aqueous TFA and were subsequently analyzed by HPLC. Proteolysis of scrambled proteins A mixture of native and scrambled proteins (10 g/ 20 l) generated by thermal denaturation were treated

80

Conformational impurity of disulWde proteins / J.-Y. Chang et al. / Anal. Biochem. 342 (2005) 78–85

with trypsin or chymotrypsin in the Tris–HCl buVer (50 mM, pH 8.0) at 37 °C for 30 min. The weight ratio of enzyme/substrate ranged from 1:50 to 1:5000. The reactions were quenched by mixing the sample with an equal volume of 4% aqueous TFA and were subsequently analyzed by HPLC. HPLC analysis of conformational isomers Protein isomers (native and scrambled) were analyzed by HPLC using the following conditions, but with diVerent gradient systems. The column was Zorbax 300XBC18 (250 £ 4.6 mm, 5 m). BuVer A was 0.05% TFA in water. BuVer B was 0.042% TFA in acetonitrile/water (9:1 by volume). The Xow rate was 0.5 ml/min. The column temperature was 22 °C. The gradient of elution diVered in each case. For Hir(1–49), it was 14 to 36% B in 60 min. For epidermal growth factor (EGF), it was 14 to 32% B in 15 min and 32 to 46% B from 15 to 60 min. For Hir(1–65), it was 24 to 40% B in 27 min. For
LA, it was 22 to 37% B in 15 min and 37 to 56% B from 15 to 45 min. For TAP, it was 29 to 37% B in 40 min. For RNase A, it was 20 to 50% B in 32 min. For lysozyme, it was 20 to 34 to 53% B from 0 to 15 to 50 min.

Results and discussion The method of disulWde scrambling The technique of disulWde scrambling is useful for analysis of reversible conversion between the native and nonnative (scrambled) protein isomers. In the presence of thiol catalyst and under denaturing conditions, a native disulWde protein denatures and unfolds by shuZing its native disulWde bonds and converting to a mixture of fully oxidized scrambled isomers [13,14]. Scrambled isomers are stabilized by nonnative disulWde bonds and, in most cases, can be well separated and characterized by reversed-phase HPLC [13,14,16,17]. The heterogeneity of scrambled isomers depends on the number of disulWde bonds. Proteins containing three, four, and Wve disulWde linkages may potentially generate 14, 104, and 944 scrambled isomers, respectively. A unique advantage of this method is that it permits simultaneous and independent quantiWcation of the extent of denaturation and unfolding [14] because a denatured disulWde protein may adopt varied extent of unfolding. The extent of denaturation is deWned and quantiWed by the simple conversion of the native structure to nonnative structures (scrambled isomers). Unfolding is deWned by the state of denatured protein and is structurally characterized by the composition (relative concentration) among the scrambled isomers. This method is used here to trap and quantify conformational impurity of disulWde proteins that may be pres-

ent at physiological conditions (37 °C, pH 7.4). HPLC repuriWed proteins with a minimum purity of 99.8% were used for demonstration here. It is relevant to mention that the HPLC condition for protein repuriWcation has no eVect on the true level of nonnative scrambled isomers detected because native and scrambled isomers always exist in the same equilibrium state under a deWned condition. The study demonstrates that even for stable proteins such as EGF and hirudin, approximately 0.2–1% of denatured isomers coexist with their native structures at physiological pH and temperature if adequate thiol catalyst is present. Denaturation of native disulWde protein under physiological pH and temperature Native hirudin (both intact and core domain) and EGF were Wrst allowed to incubate in the Tris–HCl buVer (pH 8.4) containing 200 M Cys. The incubation was carried out at diVerent temperatures for up to 1 h. Samples were then quenched by acidiWcation and analyzed by HPLC. The results, given in Fig. 1 and Table 1, show that even at room temperature (22 °C), approximately 0.03–0.04% of hirudin and EGF exist as scrambled isomers. The concentration of nonnative isomers increases to approximately 1.0–1.5% at 37 °C and 14–21% at 50 °C. The proWles of denatured isomers are roughly indistinguishable at various temperatures for each protein. The denatured isomers of Hir(1–65) overlap extensively. The disulWde structures of 11 diVerent scrambled isomers of Hir(1–49) and 6 diVerent isomers of EGF displayed in Fig. 1 have been characterized previously [14,18]. The experiments were further performed in the phosphate buVer (pH 7.4). The decrease of one unit of pH reduces the recovery of scrambled isomers by a factor of four- to eightfold (Table 1) at 37 °C. The yield of nonnative scrambled isomers was further diminished when a lower concentration of Cys was used. At pH 7.4 and 37 °C (1 h incubation), the yield of nonnative Hir(1–65) declined from 0.23% (200 M Cys) to 0.11% (75 M Cys) to 0.06% (25 M Cys). In the cases of less stable proteins, the amount of nonnative isomers existing under nondenaturing conditions can be signiWcantly higher. PCI is a single-domain, threedisulWde protein. At room temperature, approximately 3–4% of PCI, which disperses among eight detectable fractions [7], was shown to be present as nonnative isomers.
LA is an extreme and unique example. In the absence of calcium, approximately 22–27% of nonnative
LA was shown to coexist in equilibrium with the native
LA at 22 °C (Fig. 2) when HPLC-puriWed, calciumdepleted
LA was incubated for 24 h at 22 °C in the Tris–HCl buVer (pH 7.4 and 8.4) containing 0.1 mM of -mercaptoethanol as thiol catalyst. Substitution of -mercaptoethanol with Cys (0.2 mM) yielded

Conformational impurity of disulWde proteins / J.-Y. Chang et al. / Anal. Biochem. 342 (2005) 78–85

81

Fig. 1. Thermal denaturation of EGF, hirudin core domain (residues 1–49), and intact hirudin (residues 1–65), all comprising three disulWde bonds. Hir(1–49) was dissolved in phosphate buVer (50 mM, pH 7.4, 140 mM NaCl) containing 200 M of Cys. EGF and Hir(1–65) were incubated in Tris– HCl buVer (50 mM, pH 8.4), also containing 200 M of Cys. The protein concentration was 25 g/50 l. Samples containing corresponding native proteins without inclusion of Cys were analyzed in parallel as controls. The samples were incubated at 22, 37, 50, and 60 °C for 1 h; quenched by acidiWcation; and analyzed by HPLC using the conditions described in Materials and methods. Denatured Hir(1–49) consists of 11 scrambled isomers. Denatured EGF consists of 6 major isomers. Denatured isomers of Hir(1–65) are largely overlapped. The most predominant fraction of denatured isomers in each case is marked as “b.” “N” indicates the native conformation. Table 1 Percentages of hirudin and EGF present in denatured (scrambled) form Protein and Cys

22 °C (1 h)

37 °C (1 h)

50 °C (1 h)

60 °C (1 h)

Hir(1–49), 0.2 mM Cys (in Tris, pH 8.4) Hir(1–65), 0.2 mM Cys (in Tris, pH 8.4) EGF, 0.2 mM Cys (in Tris, pH 8.4) Hir(1–49), 0.2 mM Cys (in PBS, pH 7.4) Hir(1–65), 0.2 mM Cys (in PBS, pH 7.4) EGF, 0.2 mM Cys (in PBS, pH 7.4)

0.4 § 0.03 0.03 § 0.003 0.04 § 0.003 0.1 § 0.01 0.01 § 0.001 0.03 § 0.002

6.7 § 0.7 0.9 § 0.05 1.5 § 0.2 1.05 § 0.1 0.2 § 0.1 0.2 § 0.02

44.4 § 5.5 13.6 § 1.1 21.1 § 2.7 18.8 § 2.0 6.8 § 0.7 3.01 § 0.3

76.6 § 6.0 52.0 § 4.2 58.0 § 4.8 72.3 § 5.8 45.3 § 3.6 11.0 § 1.5

comparable results. The fractions of nonnative
LA diminished drastically to less than 1% on stabilization of
LA with an equal molar amount of CaCl2 (data not shown). Because 22% of nonnative
LA is distributed among 40 fractions of scrambled isomers, they are visibly detectable only when the sample is overloaded for HPLC analysis. Among nonnative
LA , isomer “c” (X
LA-c, where X stands for scrambled) represents the most predominant structure and constitutes approximately 20–35% of the total nonnative
LA, depending on the pH (Fig. 2). X-
LA-c consists of a native-like
helical domain and a disordered -sheet domain [14]. Its structure is consistent with the well-characterized molten globule state of
LA [19–22]. It is also interesting to notice that the concentration of X-
LA-c as a fraction of the total nonnative
LA increases signiWcantly as the pH is lowered from 8.4 to 7.4 (Fig. 2). These data indicate that an essential prerequisite to the detection of diminutive concentration of conformational impurity is the information provided in HPLC proWles of denatured isomers. In the case of thermal denaturation, this information can be acquired by treating disulWde

proteins at 50–60 °C in the presence of catalytic thiol, as shown in Fig. 1. Most native disulWde proteins undergo at least partial denaturation under these conditions. Denaturation and unfolding of native hirudin in human serum To evaluate conformational impurity that may potentially generate in vivo, disulWde proteins were incubated in undiluted human serum at 37 °C and elevated temperature. However, these experiments can be complicated by the abundance of serum proteins during the HPLC analysis. Hirudin (both intact and core domain) was chosen for demonstration here because it is an acidic and highly soluble protein. Both properties favor its rapid and early elution ahead of most serum proteins on the reversed-phase HPLC. These unique properties, together with acid quenching and precipitation of mass serum proteins, enable the analysis of small amounts of hirudin isomers without interference of serum proteins. When the native hirudins (both intact and core domain) were incubated in human serum at 37 °C,

82

Conformational impurity of disulWde proteins / J.-Y. Chang et al. / Anal. Biochem. 342 (2005) 78–85

Fig. 2. Presence of nonnative (scrambled) isomers of
LA at room temperature and nondenaturing condition. HPLC-puriWed, calciumdepleted native
LA was used in this study. (A) Native
LA was incubated at 22 °C for 24 h in the buVer (pH 7.4) containing 0.1 mM mercaptoethanol. (B) Native
LA was incubated at 22 °C for 24 h in the buVer (pH 7.4) (control). (C) Native
LA was incubated at 22 °C for 24 h in the buVer (pH 8.4) containing 0.1 mM -mercaptoethanol. (D) Native
LA was incubated at 22 °C for 24 h in the buVer (pH 8.4) (control). Samples were acidiWed and analyzed by HPLC using conditions described in the text.
LA consists of 4 disulWde bonds, and denatured
LA may adopt 104 possible disulWde isomers. Approximately 40 fractions of scrambled
LA (denoted collectively as “X” and marked individually in lowercase) were generated in the two samples that permit disulWde scrambling (A,C). They amount to 22–27% of the total
LA. The most predominant denatured isomer (“c”) consists of 2 native disulWde bonds at the
-helical domain and 2 nonnative disulWde bonds within the -sheet domain. Two minor peaks immediately adjacent to the native
LA are two unknown purities that were not removed by HPLC repuriWcation.

denatured hirudin isomers were not detectable (Fig. 3 and Table 2). Nonetheless, protracted incubation at 50 °C does lead to the formation of scrambled hirudins. Approximately 3.5–5.5% of scrambled hirudins were identiWed after 4 h of sample incubation at 50 °C. The slow formation of nonnative hirudin is due to the relatively low concentration of the reduced form of endogenous thiols available in human serum, which totals approximately 10 M [15]. These results, however, clearly imply the possibility of generating in vivo conformational impurity at locales with a high concentration of free thiols and/or high temperature. It is also relevant to mention that the composition of denatured isomers of hirudin generated in human serum (Fig. 3) is indistinguishable from that obtained in phosphate-buVered saline (PBS, Fig. 1). Distinctively diVerent stability of native and nonnative disulWde bonds DisulWde bonds of the native and denatured scrambled isomers diVer signiWcantly in their chemical stability. In general, nonnative disulWde bonds of scrambled isomers can be completely reduced at 22 °C by 1 mM of

Fig. 3. Denaturation of Hir(1–49) in human serum. Hir(1–49) (25 g/ 50 l) was incubated in the undiluted human serum for the indicated conditions. Samples were then acidiWed and centrifuged to remove the mass of serum protein and were analyzed by HPLC.

Table 2 Percentages of hirudin present in denatured (scrambled) form after sample incubation in the undiluted human serum Temperature and time

Hir(1–49)

Hir(1–65)

37 °C (4 h) 37 °C (24 h) 50 °C (1 h) 50 °C (4 h) 50 °C (8 h) 50 °C (24 h) 60 °C (1 h)

ND ND ND 5.6 § 0.9 22.9 § 2.1 34.8 § 4.2 11.9 § 1.0

ND ND ND 3.8 § 0.5 8.5 § 1.1 13.3 § 1.2 5.9 § 0.5

Note. ND, not detected.

DTT within 10 min, leading to the formation of the fully reduced protein (designated as “R” in Fig. 4) [23]. Under the same conditions, the native disulWde bonds that are stabilized by surrounding noncovalent interactions usually remain intact. This is demonstrated here using four diVerent proteins that are contaminated with a minor amount of scrambled isomers (Fig. 4). In each case, conversion of the scrambled proteins (marked by “a” and “b” in EGF or collectively by “X” in hirudin and RNase A) to “R” is shown to be near quantitative.

Conformational impurity of disulWde proteins / J.-Y. Chang et al. / Anal. Biochem. 342 (2005) 78–85

83

Fig. 4. DiVerentiation between native and nonnative disulWde bonds by mild reduction. Four diVerent proteins contaminated with a minor amount of nonnative scrambled isomers (25 g in 50 l of Tris–HCl buVer, pH 8.4) were reduced with 1 mM of DTT at 22 °C for 10 min. Samples were then acidiWed and analyzed by HPLC. Under this mild reducing condition, nonnative disulWde bonds were quantitatively reduced, and all scrambled isomers (denoted collectively as “X” or marked individually in lowercase) were converted to the fully reduced state (R), whereas the native isomers (N) remained intact.

This speciWc property facilitates the detection and quantiWcation of conformational impurity in two ways. First, it serves to amplify the signal of the presence of nonnative isomers and permits a better accuracy of their quantiWcation. In addition, it assists identiWcation of the presence of conformational impurity that may otherwise escape detection. For some proteins, especially large molecular weight proteins, denatured scrambled isomers might not be separated from the native isomer by HPLC. They coeluted in one single peak and exhibit indistinguishable molecular mass. In these cases, part of conformational impurity may evade detection by sensitive HPLC or mass spectrometry analysis. For instance, a fraction of scrambled RNase A coelutes with the native protein (Fig. 4), and their concentration can be measured more precisely by this method. In executing this strategy, it is essential that the mechanism of reductive unfolding of the native protein already has been well characterized because not all native disulWde bonds can withstand 1 mM of DTT. Multiple disulWde bonds of many proteins are reduced in a sequential manner in the presence of a mild concentration of DTT. For instance, Cys6–Cys20 of the native EGF is partially reduced by 1 mM of DTT (10 min) [24], generating a two-disulWde intermediate (marked as IIA in Fig. 4) that is eluted immediately before scrambled isomer “a” of denatured EGF.

Proteolytic susceptibility of nonnative scrambled proteins Another unique yet well-expected property of denatured isomers is their increased susceptibility toward proteolysis. Two cases are demonstrated here. A mixture of native (“N”) and two predominant scrambled isomers (“a” and “b”) of EGF was treated with a serially diluted solution of trypsin or chymotrypsin at 37 °C for 30 min. At an enzyme/substrate weight ratio of 1:1000, the relative concentration of (a + b)/N reduces from 2.1 to 0.75 (trypsin) and 1.1 (chymotrypsin) (Fig. 5). The two scrambled isomers appear to respond to trypsin with varied vulnerability in which the degradation rate of isomer “b” is nearly twofold faster than that of isomer “a.” In the case of lysozyme, the diVerence of enzymatic susceptibility between native and scrambled isomers is far more striking. Under the same reaction conditions, the native lysozyme remains intact, whereas assorted scrambled isomers are completely degraded (Fig. 5). Implications of the presence of conformational impurity at physiological conditions The ability to detect and quantify small amounts of nonnative conformational isomers and to further identify conditions of their emergence is imperative for several key reasons. First, denatured nonnative proteins

84

Conformational impurity of disulWde proteins / J.-Y. Chang et al. / Anal. Biochem. 342 (2005) 78–85

Acknowledgments The authors acknowledge support from the Protein Institute and the Robert Welch Foundation.

References

Fig. 5. Proteolysis of mixtures of native and scrambled isomers of EGF and lysozyme generated by thermal denaturation. (A) Control (incubated in the buVer). (B) Trypsin. (C) Chymotrypsin. The two major nonnative isomers of EGF and lysozyme are marked by “a” and “b.” Reactions were carried out in the Tris–HCl buVer (50 mM, pH 8.0) at 37 °C for 30 min. The protein concentration was 10 g/20 l. The ratio of substrate/enzyme was 1:1000. Samples were quenched by adding an equal volume of 4% aqueous TFA and were analyzed by HPLC using the conditions described in the text.

have a propensity to aggregate [25–27], which is an underlying chemical event associated with numerous neurodegenerative diseases [28–32]. The presence of minute amounts of conformational impurity is critical because denatured isomers typically exist in a state of equilibrium with the native protein. In a scenario of progressive aggregation and a steady depletion of nonnative isomers, the dynamic equilibrium may continue to convert the native protein to nonnative isomers. For instance, one of the major models [33,34] explaining the infectivity of prion disease stipulates that the trace concentration of the scrapie form of prion protein (PrPSc) exists thermodynamically in equilibrium with the benign cellular prion protein (PrPc) under physiological conditions. Second, defective or denatured nonnative proteins are targets of prompt degradation by proteolysis. Thus, their relative concentration to the native structure under physiological conditions may play a vital role in determining the metabolic rate of functional proteins. Third, the presence of minor impurity is a major concern of proteins intended for therapeutic application. In general, contaminants greater than 0.1% are required by the Food and Drug Administration to be thoroughly characterized. Currently, characterization of minor contaminants of most biopharmaceuticals focuses mainly on covalently degraded and modiWed products of the native protein [35–38].

[1] C.B. AnWnsen, Principles that govern the folding of protein chains, Science 181 (1973) 223–230. [2] F. Raschdorf, R. Dahinden, W. Maerki, W.J. Richter, J.P. Merryweather, Location of disulphide bonds in human insulin-like growth factors (IGFs) synthesized by recombinant DNA technology, Biomed. Mass Spectrom. 16 (1988) 3–8. [3] H. Meng, B.D. Burleigh, G.M. Kelly, Reduction studies on bacterial recombinant somatomedin C/insulin-like growth factor-1, J. Chromatogr. 443 (1998) 183–192. [4] S. Hober, G. Forsberg, G. Palm, M. Hartmanis, B. Nilsson, DisulWde exchange folding of insulin-like growth factor I, Biochemistry 31 (1992) 1749–1756. [5] J.A. Miller, L.O. Narhi,, Q.X. Hua, R. Rosenfeld, T. Arakawa, M. Rohde, S. Prestrelski, S. Lauren, K.S. Stoney, L. Tsai, M.A. Weiss, Oxidative refolding of insulin-like growth factor I yields two products of similar thermodynamic stability: a bifurcating proteinfolding pathway, Biochemistry 32 (1993) 5203–5213. [6] J.-Y. Chang, F. Canals, P. Schindler, E. Querol, F.X. Aviles, The disulWde folding pathway of potato carboxypeptidase inhibitor, J. Biol. Chem. 269 (1994) 22087–22094. [7] J.-Y. Chang, L. Li, F. Canals, F.X. Aviles, The unfolding pathway and conformational stability of potato carboxypeptidase inhibitor, J. Biol. Chem. 275 (2000) 14205–14211. [8] K.A. Dill, D. Shortle, Denatured states of proteins, Annu. Rev. Biochem. 60 (1991) 795–825. [9] Y. Bai, J.S. Milne, L. Mayne, S.W. Englander, Protein stability parameters measured by hydrogen exchange, Proteins 20 (1994) 4–14. [10] M.M. Krishna, L. Hoang, Y. Lin, S.W. Englander, Hydrogen exchange methods to study protein folding, Methods 34 (2004) 51–64. [11] S.L. Mayo, R.L. Baldwin, Guanidinium chloride induction of partial unfolding in amide proton exchange in RNase A, Science 262 (1993) 873–876. [12] D.R. Shortle, Structural analysis of non-native states of proteins by NMR methods, Curr. Opin. Struct. Biol. 6 (1996) 24–30. [13] J.-Y. Chang, Denatured states of tick anticoagulant peptide: compositional analysis of unfolded scrambled isomers, J. Biol. Chem. 274 (1999) 123–128. [14] J.-Y. Chang, L. Li, The structure of denatured alpha-lactalbumin elucidated by the technique of disulWde scrambling: fractionation of conformational isomers of alpha-lactalbumin, J. Biol. Chem. 276 (2001) 9705–9712. [15] R.H. Williams, J.A. Maggiore, R.D. Reynolds, C.M. Helgason, Novel approach for the determination of the redox status of homocysteine and other aminothiols in plasma from healthy subjects and patients with ischemic stroke, Clin. Chem. 47 (2001) 1031–1039. [16] J.-Y. Chang, The property of scrambled hirudins, J. Biol. Chem. 270 (1995) 25661–25666. [17] J.-Y. Chang, L. Li, The disulWde structure of denatured epidermal growth factor: preparation of scrambled disulWde isomers, J. Protein Chem. 21 (2002) 203–213. [18] J.-Y. Chang, P. Schindler, B. Chatrenet, The disulWde structures of scrambled hirudins, J. Biol. Chem. 270 (1995) 11992–11997. [19] J.-Y. Chang, A. Bulychev, L. Li, A stabilized molten globule protein, FEBS Lett. 487 (2000) 298–300.

Conformational impurity of disulWde proteins / J.-Y. Chang et al. / Anal. Biochem. 342 (2005) 78–85 [20] L.C. Wu, P.S. Kim, A speciWc hydrophobic core in the alpha-lactalbumin molten globule, J. Mol. Biol. 280 (1998) 175–182. [21] L.C. Wu, Z.Y. Peng, P.S. Kim, Bipartite structure of the alphalactalbumin molten globule, Nat. Struct. Biol. 2 (1995) 281–286. [22] B.A. Schulman, C. RedWeld, Z.Y. Peng, C.M. Dobson, P.S. Kim, DiVerent subdomains are most protected from hydrogen exchange in the molten globule and native states of human alpha-lactalbumin, J. Mol. Biol. 253 (1995) 651–657. [23] J.-Y. Chang, A two-stage mechanism for the reductive unfolding of disulWde containing proteins, J. Biol. Chem. 272 (1997) 69–75. [24] J.-Y. Chang, L. Li, P.H. Lai, A major kinetic trap for the oxidative folding of human epidermal growth factor, J. Biol. Chem. 276 (2001) 4845–4852. [25] E.Y. Chi, S. Krishnan, T.W. Randolph, J.F. Carpenter, Physical stability of proteins in aqueous solution: mechanism and driving forces in nonnative protein aggregation, Pharm. Res. 9 (2003) 1325–1336. [26] C.M. Dobson, Protein folding and misfolding, Nature 426 (2003) 884–890. [27] M. Bucciantini, E. Giannoni, F. Chiti, F. Baroni, L. Formigli, J. Zurdo, N. Taddei, G. Ramponi, C.M. Dobson, M. Stefani, Inherent toxicity of aggregates implies a common mechanism for protein misfolding diseases, Nature 416 (2002) 507–511. [28] C.A. Ross, M.A. Poirier, Protein aggregation and neurodegenerative disease, Nat. Med. 10 (2004) S10–S17.

85

[29] M.P. Mattson, M. Sherman, Perturbed signal transduction in neurodegenerative disorders involving aberrant protein aggregation, Neuromol. Med. 4 (2003) 109–132. [30] M. Hashimoto, E. Rockenstein, L. Crews, E. Masliah, Role of protein aggregation in mitochondrial dysfunction and neurodegeneration in Alzheimer’s and Parkinson’s diseases, Neuromol. Med. 4 (2003) 21–36. [31] E.H. Koo, P.T. Lansbury Jr., J.W. Kelly, Amyloid diseases: abnormal protein aggregation in neurodegeneration, Proc. Natl. Acad. Sci. USA 96 (1999) 9989–9990. [32] A.L. Fink, Protein aggregation: folding aggregates, inclusion bodies, and amyloid, Fold Des. 3 (1998) R9–R23. [33] P.T. Lansbury Jr., B. Caughey, The chemistry of scrapie infection: implications of the “ice-9” metaphor, Chem. Biol. 2 (1995) 1–5. [34] F.E. Cohen, S.B. Prusiner, Pathologic conformations of prion proteins, Annu. Rev. Biochem. 67 (1998) 793–819. [35] M.C. Manning, K. Patel, R.T. Borchardt, Stability of protein pharmaceuticals, Pharm. Res. 11 (1989) 903–918. [36] R.L. Garnick, N.J. Solli, P.A. Papa, The role of quality control in biotechnology: an analytical perspective, Anal. Chem. 60 (1988) 2546–2557. [37] D.T. Liu, Deamidation: a source of microheterogeneity in pharmaceutical proteins, Trends Biotechnol. 10 (1992) 364–369. [38] L.J. Janis, G.C. Davis, Analytical strategies for the determination of protein modiWcations, Dev. Biol. Stand. 83 (1994) 135–142.