Skip to main content
NCBI home page
Protein Science : A Publication of the Protein Society logoLink to Protein Science : A Publication of the Protein Society
. 2002 Jul;11(7):1825–1833. doi: 10.1110/ps.0208102

Interactions between PEG and type I soluble tumor necrosis factor receptor: Modulation by pH and by PEGylation at the N terminus

Bruce A Kerwin 1, Byeong S Chang 1, Colin V Gegg 1, Margherita Gonnelli 2, Tiansheng Li 1, Giovanni B Strambini 2
PMCID: PMC2373643  PMID: 12070334

Abstract

The effects of polyethylene glycol (PEG) on protein structure and the molecular details that regulate its association to polypeptides are largely unknown. These issues were addressed using type I soluble tumor necrosis factor receptor (sTNF-RI) as a model system. Changes in solution viscosity established that a truncated form of sTNF-RI bound free PEG in a pH-dependent manner. Above pH 5.3, the viscosity escalated as the pH increased, while no effect occurred below pH 5.0. Conjugation of 2 kD, 5 kD, or 20 kD PEG to the N terminus attenuated the viscosity at the higher pH values. Tryptophan phosphorescence spectroscopy correlated changes in the protein structure about Trp-107, at the C terminus, with the pH-dependent and PEGylation-dependent attenuation of the viscosity. The results indicate that specific interactions between PEG and the truncated form of sTNF-RI are elicited by an increased flexibility of the truncated protein combined perhaps with removal of steric or charge barriers. Covalently bound PEG at the N terminus reduced the protein affinity for the free polymer and induced a more rigid and polar configuration around Trp-107. Deprotonation of His-105, which is perpendicular to Trp-107, was integral to the binding mechanism producing a pH-dependent switching mechanism. These findings stress the importance of surface charge and structural plasticity in determining macromolecular binding affinities and demonstrate the ability of conjugated PEG to modify the localized surface structure in proteins away from the site of conjugation.

Keywords: PEG, polyethylene glycol, soluble tumor necrosis factor receptor, phosphorescence, viscosity

PEG is a water-soluble, biocompatible polymer used extensively as a stabilizer in protein formulations (Cleland and Jones 1996), as a cosolvent in X-ray crystallography (McPherson 1976) and protein purification (Andrews et al. 1996; Harris et al. 1997). Covalently bound to biomedical devices, it prevents the adhesion of proteins and cells to exogenous surfaces (Deible et al. 1999; Jo and Park 2000), while its conjugation to proteins increases their water solubility, in vivo lifetimes, and immunocompatibility (Molineux et al. 1999; Pepinsky et al. 2001).

Despite its numerous applications, the molecular bases for its action are not always understood and, therefore, its behavior on individual protein systems is largely unpredictable. According to solution studies on model proteins, the polymer, owing to its size, tends to be excluded from the protein surface (Lee and Lee 1981; Bhat and Timasheff 1992), a process known as preferential exclusion. The addition of PEG to protein solutions then will increase the free energy of the system causing the polypeptide to adopt a more compact globular shape that minimizes the surface area exposed to the solvent (Lee and Lee 1981; Bhat and Timasheff 1992). This effect of molecular crowding by the polymer is expected to prompt polypeptides to fold and associate, the latter reaction leading eventually to phase separations in which proteins either precipitate (Atha and Ingham 1981) or crystallize (McPherson 1976) out of the solution.

For any given protein, however, the degree of preferential exclusion will be influenced also by the balance between repulsive interactions with surface charges and attractive interactions with hydrophobic, nonpolar patches. The latter clusters have been found to be composed of the side-chains of Trp, Phe, and Leu (Sasahara and Uedaira 1993) and the methylene backbones of ionizable amino acids such as Asp, Arg, and Lys (Furness et al. 1998). In general, PEG binding is weak (e.g., Kd = 76 mM, between PEG and Trp-62 of lysozyme (Furness et al. 1998) but can become stronger following large structural changes exposing previously buried hydrophobic residues, such as in the case with the molten globule state (Cleland and Randolph 1992) or denatured proteins (Lee and Lee 1987).

Beyond these broad considerations, our understanding of the molecular details governing protein-PEG interactions or the polymer influence on the conformation of the macromolecule is still inadequate. In particular, little is known about the factors that preside over PEG binding to specific protein sites and, more generally, on the effects that free or covalently bound PEG may have on the structure of folded proteins. Perturbations of the native fold may arise from proximity or direct contact with the polymer, replacing or altering the aqueous interface, as well as from PEG conjugation to specific, Cys (Kuan et al. 1994), or N terminus (Kinstler et al. 1996) protein sites. A largely unanswered question in the field of protein PEGylation is to what extent altered immunogenic response or macromolecular recognition is the result of modifications of the native protein conformation. The fact that PEG-induced structural changes have not been detected by common spectroscopic methods, which are mostly sensitive to changes in secondary structure, suggests that any perturbation is probably subtle and limited to the tertiary and quaternary structure.

This report attempts to address these issues by employing sTNF-RI as a model protein system. The full length polypeptide is composed of 162 amino acids arranged as an elongated (2.5 nm × 7.5 nm) β-sheet, cross-linked by 10 disulfide bonds spread along the length of the macromolecule (Rodseth et al. 1994; Naismith et al. 1996) (Fig. 1 ▶). The macromolecule contains a single, partially exposed Trp residue at position 107 that, together with nearby Phe-115, Tyr-106, and Leu-111, forms a hydrophobic/aromatic cluster with PEG-binding propensity. In this paper, we report that free PEG binds specifically to a truncated form of sTNF-RI. The interaction, which is absent in the full length protein, 4D sTNF-RI (Solorzano et al. 1998), was found to be sharply modulated by pH and to be for the most part attenuated when PEG was conjugated to the N terminus, a site far removed from the putative binding site. Taking advantage of the fine sensitivity of Trp phosphorescence to the nature and to the dynamic features of the Trp environment (Gonnelli and Strambini 1995; Strambini and Gonelli 1995; Cioni and Strambini 1998), it was found that an increased plasticity of the local structure plays an important role in conferring PEG-binding selectivity to the truncated protein. Most significantly, while protonation of a nearby His apparently is responsible for pH modulation of the binding interaction, an attenuated affinity for the free polymer following conjugation with PEG at the far N terminus was associated to subtle changes in conformation about the aromatic cluster.

Fig. 1.

Fig. 1.

Open in a new tab

Structure of sTNF-RI, highlighting Trp107 and His105. The yellow C-terminal segment is removed upon truncation of 4D sTNF-RI to yield the 2.6D sTNF-RI. The X-ray crystallographic structure coordinates were retrieved from the Protein Data Bank (classified as 1NCF) and published by Naismith et al. (1996).

Results

The macroscopic interaction between 2.6D sTNF-RI and unconjugated PEG was manifested by a large enhancement of the solution viscosity of protein/polymer mixtures (Fig. 2A-C). The phenomenon is pH-dependent as the viscosity of solutions containing 2.6D sTNF-RI and free PEG increased sharply only above pH 5.2 (Fig. 2A ▶). At the highest pH examined (5.7), viscosity is enhanced fivefold over that of PEG or protein alone. Higher pH values were not examined because of the reduced solubility of 2.6D sTNF-RI. Because viscosity is a direct function of molecular size, its increase is indicative of higher-order macromolecular interactions. It is also noted that the viscosity change is not instantaneous but grows over an extended period of time in a sigmoidal fashion (Fig. 2B ▶). This suggests that a concerted change in the structure of protein, polymer, or both was necessary before equilibrium was reached.

Fig. 2.

Fig. 2.

Open in a new tab

Effect of pH, PEGylation, and protein truncation, on the viscosity of 20 kD PEG/sTNF-RI solutions. (A) Viscosity of 2.6D sTNF-RI at varying pH. The protein concentration was 20 mg/mL and that of PEG was 50 mg/mL. (B) Time-dependent rise of the solution viscosity of 20 kD PEG plus 2.6D sTNF-RI at pH 5.46. The protein and polymer concentrations were the same as in 2A. (C) Viscosity enhancement by various sTNF-RI species at pH 5.7. The bars indicate the protein alone and hatched bars indicate the protein plus free 20 kD PEG. The protein concentration of PEGylated sTNF-RI was 20 mg/mL while that of 4D sTNF-RI was increased to 26 mg/mL to maintain an equivalent PEG to protein molar ratio. The PEG was 50 mg/mL.

In principle, viscous solutions can result from the formation of large polymer-polymer, protein-polymer, and protein-protein aggregates or protein unfolding. The invariance of the protein secondary structure as well as of disulfide and Trp Cβ-C3 bond angles, observed by Fourier transform Raman spectroscopy (data not shown), ruled out both global unfolding and major changes in the protein tertiary structure. Indirect evidence, inferred from the protein-to-protein crosslinking yield, points to a specific association of 2.6D sTNF-RI to PEG. The rationale behind the crosslinking data is that the crowding effect of PEG is expected to enhance protein self-association, and thereby promote intermolecular crosslinking, binding of the polypeptide to PEG would decrease the probability of protein-protein encounters and inhibit the reaction. The effect of 20 kD PEG on the remaining uncrosslinked monomer fraction at pH 4.5 and 5.5 is shown in Figure 3 ▶. The results show that increasing the PEG concentration enhances the crosslinking yield at pH 4.5 but has the opposite effect at pH 5.5. The decreased yield over the pH range of the viscosity enhancement therefore is consistent with a pH-induced protein-PEG association. Based on the known affinity of PEG for Trp (Eiteman and Gainer 1991; Sasahara and Uedaira 1993; Lu and Tjerneld 1997; Furness et al. 1998) and the spectroscopic data presented below, the viscosity effect is attributed to specific binding of PEG to the hydrophobic cluster formed by Trp-107. Figure 2C ▶ emphasizes that the viscosity effect is lacking with full-length sTNF-RI and that it is considerably attenuated on PEGylation of 2.6D sTNF-RI at the N terminus with 2 kD, 5 kD, and to a lesser degree 20 kD PEG (Fig. 2C ▶).

Fig. 3.

Fig. 3.

Open in a new tab

Effect of 20 kD PEG on the intermolecular crosslinking of 2.6D sTNF-RI by EDC, at pH 4.5 (○) and pH 5.5(•). A decrease in the monomer fraction indicates an increase in the degree of crosslinking.

Possible changes of the protein structure in the region of Trp-107 that may underlie the specificity of PEG interactions to truncated sTNF-RI and its fine modulation by pH and PEG conjugation were investigated by Trp phosphorescence. This emission provides a uniquely sensitive probe of the chromophore's microenvironment. It reports on the polarity and the fluidity of the local matrix (Hershberger et al. 1980; Strambini and Gonelli 1995), on the relative accessibility of the indole ring to the solvent and to solvent-mediated reactions (Cioni and Strambini 1998), and on its potential association to His and Tyr side-chains (Gonnelli and Strambini 1995).

The phosphorescence spectrum obtained in low-temperature (140 K) glasses is related to the polarity of the Trp environment, while the bandwidth of the 0,0 vibronic band reports on its structural homogeneity. The spectral energy of Trp-107 (Fig. 4 ▶) was indicative of a polar matrix similar to the aqueous solvent. However, the emission bandwidth of the 0,0 vibronic band was significantly smaller (5.4 nm) than for fully solvent-exposed residues (8.5 nm). Likewise, spectral invariance with respect to the excitation wavelength indicates homogeneity in ground-state energy and again excludes substantial solvation of the indole ring. Overall, the spectrum indicated that Trp-107 is located in a relatively polar but conformationally restricted environment. These spectral features were common to 2.6D and full-length sTNF-RI, and were invariant to PEGylation or pH from 4.5–10.

Fig. 4.

Fig. 4.

Open in a new tab

Phosphorescence spectrum of sTNF-RI at pH 7.5 in a propylene glycol/buffer (50:50, V:V), at 140 K. The spectrum of N-acetyltryptophanamide was included as a reference for a completely exposed Trp residue while that of Trp-84 of glyceraldehyde-3–phosphate dehydrogenase represents an example of a completely buried, structurally homogeneous Trp site.

Considerable shielding of the aromatic ring from the solvent was confirmed by restricted spectral relaxation in fluid solutions. In the temperature interval 160 K–210 K, the glass formed by propylene glycol/water (50/50, V/V) turns into a fluid solution. During this transition, the spectrum of free Trp (NATA) red shifts by 10 nm as a result of rapid solvent reorientation about the excited chromophore. In the case of Trp-107, the red shift is significantly smaller, ∼1.3 nm, and again independent of truncation, pH, or PEGylation. From the correlation between spectral shift and degree of solvent exposure, it was estimated that ∼80% of the indole ring surface was not in contact with the mobile solvent. That is in fair agreement with the crystallographic structure of full-length sTNF-RI, in which only the carbon atoms CE3, CZ3, CH2, and CE2 of the indole ring can contact the solvent with the remainder of the structure being inaccessible (Naismith et al. 1996).

The static accessibility of Trp-107 to the aqueous solvent can be inferred from the decrease in phosphorescence lifetime caused by heavy-atoms species, like I−, relegated to the aqueous medium (Li et al. 1989). In glasses, the phosphorescence lifetime of Trp-107 was 6.4 sec. This value is typical of Trp residues not perturbed by the proximity of disulfides, which are numerous in sTNF-RI. In the presence of I−, the emission is considerably shorter-lived and heterogeneous (Fig. 5 ▶), the multiplicity in phosphorescence lifetimes reflecting the random, static distribution of Trp-107-iodide distances. The accessibility of the indole ring to I− was estimated from the reduction in the average lifetime of Trp-107 relative to that of NATA free in solution (Table 1). For the 4D sTNF-RI, the phosphorescence lifetime was about 2.3 times greater, at both pH 4.5 and 7.4, than for the free chromophore confirming substantial shielding of the aromatic ring from the solvent. The I−effect for truncated 2.6D sTNF-RI was similar to 4D sTNF-RI at acidic pH, but increased significantly on raising the pH to neutrality. PEGylation enhanced the interaction with I−, the effect increased with the molecular weight of PEG. In this case, the I− effect was similar at both acidic and neutral pH. The increased I− accessibility of the 2.6D sTNF-RI observed on raising the pH or subsequent to PEGylation reports subtle variations in the environment of Trp-107. These changes are consistent with either an increased exposure of the indole ring to the solvent and/or a more positive (attractive for I−) surface potential.

Fig. 5.

Fig. 5.

Open in a new tab

Effect of 1 M KI on the phosphorescence decay of 2.6D sTNF-RI at 140 K. The solvent conditions were the same as in Figure 4 ▶. The effect of iodide on the phosphorescence decay of N-acetyltryptophanamide is included for comparison. All intensities are normalized to their initial value.

Table 1.

Effect of 1M potassium iodide on the average phosphorescence lifetime (τ) of Trp-107 of sTNF-RI at acidic and neutral pH

Sample pH τ/τNATA
2.6D sTNF-RI 4.5 2.24 ± 0.04
7.4 2.04 ± 0.03
2kD PEGsTNF-RI 4.5 1.95 ± 0.05
7.4 1.82 ± 0.04
5kD PEGsTNF-RI 4.5 1.84 ± 0.04
7.4 1.9 ± 0.04
20kD PEGsTNF-RI 4.5 1.72 ± 0.04
7.5 1.70 ± 0.04
4D sTNF-RI 4.5 2.29 ± 0.04
7.4 2.27 ± 0.04
Open in a new tab

All measurements were conducted in triplicate. The reproducibility of phosphorescence lifetime ratios was better than 2%.

In fluid solutions, a measure of the dynamic accessibility of Trp-107 to the solvent, that is largely independent of the surface electrostatic potential, can be obtained from the ease with which neutral quenching solutes like acrylamide diffuse to it and quench its phosphorescence by a relatively short-range, near-contact quenching reaction (Cioni and Strambini 1998). The bimolecular rate constant (kq) for acrylamide quenching, at 274 K and pH 9, was determined from the slope of the phosphorescence lifetime Stern-Volmer plot (1/τ = 1/τ0 + kq[acrylamide], τ0 the unperturbed lifetime) (Cioni and Strambini 1998). For the 2.6D sTNF-RI, a magnitude of kq = 9.6 ± 0.5 × 108 M−1s−1, similar to that for free Trp (kq = 1.3 × 109 M−1s−1), confirmed the superficial location of the aromatic ring, if not its partial exposure to the aqueous solvent. Relative to the 2.6D sTNF-RI, acrylamide quenching decreased nearly twofold in the 4D sTNF-RI. This implies that following truncation of the polypeptide, Trp-107 becomes more accessible to solvent-mediated interactions through either a better exposure of the aromatic ring to the solvent or increased mobility of the local protein structure. Interestingly, PEGylation of 2.6D sTNF-RI, with 2 kD, 5 kD, and 20 kD PEG offered little protection (a 10% reduction of kq) against quenching. This suggests that the greater iodide effect induced by PEGylation is likely because of an increased polarity of the region rather than a greater exposure of the indole ring to the solvent.

In fluid-aqueous solutions, the phosphorescence lifetime of Trp (τ0) is drastically reduced by the motional freedom of the protein-solvent matrix surrounding the chromophore (Strambini and Gonelli 1995). Beyond this effect of matrix dynamics, τ0 can be decreased further by internal quenching interactions with Cys, His, and Tyr side-chains coming in contact with the chromophore (Gonnelli and Strambini 1995). For all sTNF-RI forms, the phosphorescence decay in buffer at neutral pH was short-lived and remarkably nonuniform (Fig. 6 ▶). The magnitude of phosphorescence lifetimes, τi, amplitudes, αi, and the average phosphorescence lifetime, τav = Σαiτi, are reported in Table 2. For the 2.6D sTNF-RI, a major fraction (84%) of the sample has a τ shorter than that exhibited by highly flexible, unstructured peptides (τ = 200–250 μs) (Gonnelli and Strambini 1995). This demonstrates that in sTNF-RI, τ is largely dominated by intramolecular quenching reactions. The X-ray structure points out that the imidazole of His-105 is perpendicular to and within interaction range (3.6 Å) of the indole ring of Trp-107 (Fig. 1 ▶). His quenching can be discriminated from other reactions by the pH dependence of τ in that its protonation is known to enhance up to 50-fold its quenching efficiency (Gonnelli and Strambini 1995). The lifetime pH profile of 2.6D sTNF-RI (Fig. 7 ▶) confirmed a sharp decrease of τav at acid pH, the transition midpoint being around the expected pKa (6–6.5) of His. Both the acid transition and the pKa value strongly indicate proximal His-105 as the quencher of phosphorescence.

Fig. 6.

Fig. 6.

Open in a new tab

Examples of phosphorescence decay of 2.6D and 4.0D sTNF-RI in buffer, at pH 7.5, and 274 K. The decay of N-acetyltryptophanamide is included for comparison.

Table 2.

Lifetimes (τi, microseconds) and amplitudes (αi) derived from the phosphorescence decay of various sTNF-RI species in buffer (pH 7.5), at 274K

Sample τ1 (α1) τ2 (α2) τ3 (α3) τavg
2.6D sTNF-RI 90 (0.84) 250 (0.16) — 115
2kD PEGsTNF-RI 87 (0.61) 187 (0.39) — 126
5kD PEGsTNF-RI 80 (0.58) 187 (0.42) — 125
20kD PEGsTNF-RI 90 (0.55) 190 (0.45) — 135
4D sTNF-RI 63 (0.34) 201 (0.58) 711 (0.08) 195
Open in a new tab

All measurements were conducted in triplicate, the reproducibility of the phosphorescence lifetime measurements typically better than 5%.

Fig. 7.

Fig. 7.

Open in a new tab

pH dependence of the average phosphorescence lifetime of sTNF-RI: 4D sTNF-RI (•), 2.6D sTNF-RI (▪), 5 kD PEGsTNF-RI (□) and 20 kD PEGsTNF-RI (○).

Relative to truncated 2.6D sTNR-RI, the phosphorescence decay of the full-length 4D sTNF-RI was slower and more heterogeneous, implying a more restricted mobility for Trp-107. There appears to be a significant reduction in His quenching in that the amplitude of the short-lived fraction decreased from 80% in the truncated form to 34% in the full-length molecule. More importantly, for the 4D sTNF-RI, τav is largely independent of pH (Fig. 7 ▶). This clear distinction in the pH profile suggests that imidazole-indole orientations that are efficient for quenching by proton transfer are attained only after truncation of the polypeptide and may be a consequence of the in gain in conformational freedom. PEGylation of the 2.6D sTNF-RI did not affect the acid transition (Fig. 7 ▶). It simply caused a moderate lengthening of τav (Table 2); the change increased with the size of bound PEG. The effects of N-terminal PEGylation on the phosphorescence lifetime are consistent, a slightly more rigid structure about Trp at the far C-terminal region, the magnitude of the change being correlated to the size of the attached PEG.

Discussion

PEG binding to truncated sTNF-RI

The solution viscosity demonstrates that free PEG binds to the 2.6D fragment but not to the full-length protein and, further, that the affinity of this association decreases sharply below pH 5.7. Based on the spectroscopic data gathered in this study and the propensity of PEG for binding specifically to Trp (Sasahara and Uedaira 1993) and to hydrophobic patches (Lee and Lee 1987; Cleland and Randolph 1992; Furness et al. 1998), it is proposed that the association of PEG to the hydrophobic cluster formed by Trp-107, Phe-115, Tyr-106, and Leu-111 accounts for both the selectivity toward the truncated protein and the pH modulation of binding.

As in the 4D protein, the indole ring of 2.6D sTNF-RI is not fully available for an interaction with the polymer (being still largely shielded from the aqueous phase); the conformational plasticity of the local structure is probably crucial to permit, through an induced-fit type mechanism, a strong association between aromatic ring and polymer. The crystallographic structure of the 4D sTNF-RI (Naismith et al. 1996) predicts an indole ring with a restricted mobility and moderate exposure to the aqueous solvent. The relatively high phosphorescence spectral resolution (BW), ground-state homogeneity, and limited spectral relaxation agree with an ordered, rigid local structure where the indole ring is largely shielded from the solvent. Truncation of the C-terminal part of the polypeptide affects neither the side-chain configuration about Trp-107 nor the secondary and tertiary structures of the remaining 2.6D portion (FT-Raman and FTIR data not shown).

The main effects of truncation are an increased fluidity of the Trp environment with greater accessibility of the indole ring to solvent-mediated interactions and pronounced intramolecular phosphorescence quenching at acidic pH. A greater flexibility of the 2.6D fragment, relative to the parent macromolecule, is inferred from both a shorter phosphorescence lifetime (Table 2) and the twofold increase in the acrylamide-quenching constant. On the other hand, the acid transition in the lifetime pH profile of 2.6D sTNF-RI emphasizes efficient quenching by neighboring His-105 in the protonated state, as the pH profile is similar to that observed for His titration (pKa = 6.6) (McNutt et al. 1990). It is significant that the acid transition is practically abolished in the full-length protein as it suggests that effective indole-imidazole quenching geometries are not attained in this case, presumably through restricted mobility of the side-chains. We note that the direct interaction between PEG and Trp-107 also provides a simple switch mechanism for the pH modulation of PEG binding. When His-105 is protonated, the imidazole ring can form a strong complex with indole (Loewenthal et al. 1992) and lead to partial withdrawal of the ring from the aqueous phase at acidic pH. A greater accessibility of Trp to the solvent at neutral/alkaline pH also is consistent with a larger I−effect (Table 1).

Effects of PEGylation on protein structure

Covalently bound PEG at the N terminus reduces the interaction with the free polymer as it largely abolishes the viscosity enhancement. This occurs without changes to the secondary structure of the protein, as generally observed with other PEGylated proteins (Tuma et al. 1995; Kinstler et al. 1996). In principle, the decreased affinity for free PEG could be ascribed to subtle alterations of the protein conformation at the binding site, or to steric hindrance by bound PEG either shielding the binding site from the solvent or preventing a close approach of the polymer to the protein. Phosphorescence data show that in the conjugated protein, the chromophore is not screened from the solvent as the accessibility of acrylamide is practically unaltered and that of iodide is actually increased. The phosphorescence lifetime and the iodide accessibility of Trp-107 report that PEG conjugated to the N terminus induces a more rigid and polar configuration about the aromatic residue in the far C-terminal region. Thus, both an electrostatic repulsion to PEG as a result of the increased local charge and/or a more rigid local structure (as in the case of the full-length protein) could lead to a decreased affinity for the free polymer.

Steric hindrance by the bulky polymer also may be important. In general, its contribution is difficult to assess because, although it is assumed that covalently bound PEG surrounds the polypeptide, the area occupied by it in relation to the protein is unknown. Assuming a random coil configuration for PEG, the Flory radius estimates that 2 kD, 5 kD, and 20 kD PEGs occupy spheres with diameters of 35, 70, and 153 Å, respectively (Jeppesen et al. 2001). Because the length of 2.6D sTNF-RI is ∼35 Å, this rough calculation shows that steric interference by attached PEG can play a role in decreasing the interaction between sTNF-RI and free PEG, especially for the larger polymers. However, it may be significant in this respect that steric hindrance does not inhibit the association of PEGylated sTNF-RI to TNF-α (Edwards 1999), although its binding site is close to the N terminus (Banner et al. 1993). Furthermore, the smallest 2 kD PEG is even more effective in inhibiting the interaction with the free polymer than the 10 times bulkier 20 kD PEG.

In the absence of detectable changes in protein structure, steric effects often have been invoked to explain the decreased immunogenicity of some PEGylated proteins (Moreland et al. 2000; Reddy 2000). However, the results with sTNF-RI emphasize the ability of bound PEG to induce subtle changes in the conformation of surface residues in regions remote from the site of PEGylation, effects that may have important consequences for macromolecular recognition and the immunogenic response. Overall, the outcome of the present investigation emphasizes the importance of specific side-chains, surface charges, and structural plasticity in determining the binding affinity between proteins and PEG. Additionally, it demonstrates the ability of conjugated PEG to alter the surface structure of proteins and their interaction with other macromolecules such as may occur with the immune system. The ability to modify localized protein conformation away from the site of PEGylation expands the realm of possibilities for preventing unwanted macromolecular interactions. This will become increasingly important as, with the availability of the human genome, more proteins are brought forth as potential drugs.

Materials and methods

All chemicals were of the highest purity grade available from commercial sources. All PEG products were purchased from Shearwater polymers. Free Trp as NATA was purchased from Sigma and prior to use was recrystallized three times from ethanol/water. Acrylamide (>99.9% electrophoretic purity) was from Bio-Rad Laboratories. Spectroscopic-grade propylene glycol was from Merck & Co. and prior to use was treated with the reducing agent sodium borohydride and distilled under vacuum.

The 2.6D and 4D sTNF-RI were produced at Amgen, Inc. as previously described (Solorzano et al. 1998) and provided by Dr. Jim Seely. PEGylation and purification of the 2.6D sTNF-RI with 2 kD PEG-aldehyde, 5 kD PEG-aldehyde, and 20 kD PEG-aldehyde was achieved through reductive alkylation as previously described (Kinstler et al. 1996).

The viscosity was measured at 26°C using a Haake fallingball micro viscometer. All proteins were fully equilibrated with 20 mM sodium acetate at pH 5.6. The protein concentration was 20 mg/mL for 2.6 D sTNF-RI, 2 kD PEGylated sTNF-RI, 5 kD PEGylated sTNF-RI, and 20 kD PEGylated sTNF-RI. The concentration was increased to 26 mg/mL for 4D sTNF-RI to maintain a constant protein/polymer molar ratio. Solid 20 kD MPEG was added to a final concentration of 50 mg/mL. Measurements were recorded after the system reached equilibrium as judged by a constant viscosity for more than three consecutive measurements. The values are the average of three measurements. Standard deviations were <5% of the value.

Chemical crosslinking was achieved using the following procedure. 2.6D sTNF-RI was equilibrated in 50 mM 2–(N-morpholino) ethane sulfonic acid (MES) and diluted to a final concentration of 1 mg/mL. Ethylendiamine carbodiimide (EDC) was added at an EDC to protein ratio of 2:1 and reacted for 4 h at room temperature. Samples were diluted 20-fold with buffer equilibrated at 4°C to stop the reaction. Reaction mixtures were separated using reverse-phase HPLC. An aliquot, 15 μL, of the quenched samples were loaded on a 2.1 × 300 mm Phenomenex Jupiter C4 column equilibrated with 18% acetonitrile/0.1% trifluoroacetic acid at a flow rate of 0.2 mL/min. Protein was eluted with a 1%/min gradient of 90% acetonitrile/0.1% trifluoroacetic acid. The uncrosslinked and crosslinked sTNF-RI was quantified from the A215 of the chromatograms.

Tryptophan phosphorescence was measured on thoroughly deoxygenated samples using a procedure and instrumentation previously described (Strambini and Gonelli 1995). Low-temperature phosphorescence emission spectra were measured in a propylene glycol/buffer (50:50, V:V), at 140 K. The excitation wavelength was 295 nm and the bandwidth was 6 and 1 nm for excitation and emission, respectively. In these glasses, the sTNF-RI concentration was typically 20 μM, in 20 mM sodium phosphate, 100 mM sodium chloride, at the specified pH. Phosphorescence decays in buffer, above freezing temperatures, were obtained using pulsed excitation at 292 nm while the emission was collected through a filter combination transmitting between 410–440 nm. All proteins were dissolved at a concentration of 8–10 μM in a buffer composed of equimolar (2 mM) sodium acetate, sodium phosphate, and sodium carbonate plus 100 mM sodium chloride. The presence of various buffering salts allowed pH variations between 4 and 11. All phosphorescence decays were analyzed in terms of a sum of exponential components by a nonlinear least-squares–fitting algorithm (Global Unlimited, LFD, University of Illinois). The reported lifetime data are averages of two or more independent measurements. The reproducibility of phosphorescence lifetimes was typically better than 5%. Acrylamide-quenching experiments were carried out as described (Cioni and Strambini 1998). The phosphorescence emission of sTNF-RI is intrinsically heterogeneous and remains so even when the average phosphorescence lifetime is considerably reduced by acrylamide. For convenience, lifetime Stern-Volmer plots were all constructed from the average lifetime, τav = ΣαiτI, obtained in general from a biexponential fit of phosphorescence decays. Consequently, the value of the bimolecular quenching rate kq derived from these plots is an averaged quantity.

Acknowledgments

We thank Dr. Jim Seely for kindly providing the 4D sTNF-RI, Dr. Izydor Apostol for many useful discussions and critical readings of the manuscript, and Dr. Margaret Speed for critical reading and suggestions of the manuscript. A portion of the work was paid for by a grant to G.S. from Amgen Inc.

The publication costs of this article were defrayed in part by payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 USC section 1734 solely to indicate this fact.

Abbreviations

  • sTNF-RI, soluble tumor necrosis factor receptor

  • PEG, polyethylene glycol

  • MPEG, methoxypolyethylene glycol

  • NATA, N-acetyltryptophanamide

  • EDC, ethylenediamine carbodiimide

  • HPLC, high pressure liquid chromatography

Article and publication are at http://www.proteinscience.org/cgi/doi/10.1110/ps.0208102.

References

  1. Andrews, B.A., Nielsen, S., and Asenjo, J.A. 1996. Partitioning and purification of monoclonal antibodies in aqueous two-phase systems. Bioseparation 6 303–313. [PubMed] [Google Scholar]
  2. Atha, D.H. and Ingham, K.C. 1981. Mechanism of precipitation of proteins by polyethylene glycols. Analysis in terms of excluded volume. J. Biol. Chem. 256 12108–12117. [PubMed] [Google Scholar]
  3. Bhat, R. and Timasheff, S.N. 1992. Steric exclusion is the principal source of the preferential hydration of proteins in the presence of polyethylene glycols. Protein Sci. 1 1133–1143. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Cioni, P. and Strambini, G.B. 1998. Acrylamide quenching of protein phosphorescence as a monitor of structural fluctuations in the globular fold. J. Am. Chem. Soc. 120 11749–11757. [Google Scholar]
  5. Cleland, J.L. and Jones, A.J. 1996. Stable formulations of recombinant human growth hormone and interferon-γ for microencapsulation in biodegradable microspheres. Pharm. Res. 13 1464–1475. [DOI] [PubMed] [Google Scholar]
  6. Cleland, J.L. and Randolph, T.W. 1992. Mechanism of polyethylene glycol interaction with the molten globule folding intermediate of bovine carbonic anhydrase B. J. Biol. Chem. 267 3147–3153. [PubMed] [Google Scholar]
  7. Deible, C.R., Petrosko, P., Johnson, P.C., Beckman, E.J., Russell, A.J. and Wagner, W.R. 1999. Molecular barriers to biomaterial thrombosis by modification of surface proteins with polyethylene glycol. Biomaterials 20 101–109. [DOI] [PubMed] [Google Scholar]
  8. Edwards, C.K., 3rd. 1999 PEGylated recombinant human soluble tumour necrosis factor receptor type I (r-Hu-sTNF-RI): Novel high affinity TNF receptor designed for chronic inflammatory diseases. Ann. Rheum. Dis. 58 I73–81 [DOI] [PMC free article] [PubMed]
  9. Eiteman, M.A. and Gainer, J.L. 1991. Partition of isomeric dipeptides in poly(ethylene glycol)/magnesium sulfate aqueous two-phase systems. Biochim. Biophys. Acta. 1073 451–455. [DOI] [PubMed] [Google Scholar]
  10. Furness, E.L., Ross, A., Davis, T.P. and King, G.C. 1998. A hydrophobic interaction site for lysozyme binding to polyethylene glycol and model contact lens polymers. Biomaterials 19 1361–1369. [DOI] [PubMed] [Google Scholar]
  11. Gonnelli, M. and Strambini, G.B. 1995. Phosphorescence lifetime of tryptophan in proteins. Biochemistry 34 13847–13857. [DOI] [PubMed] [Google Scholar]
  12. Harris, D.P., Andrews, A.T., Wright, G., Pyle, D.L., and Asenjo, J.A. 1997. The application of aqueous two-phase systems to the purification of pharmaceutical proteins from transgenic sheep milk. Bioseparation 7 31–37. [DOI] [PubMed] [Google Scholar]
  13. Hershberger, M.V., Maki, A.H., and Galley, W.C. 1980. Phosphorescence and optically detected magnetic resonance studies of a class of anomalous tryptophan residues in globular proteins. Biochemistry 19 2204–2209. [DOI] [PubMed] [Google Scholar]
  14. Jeppesen, C., Wong, J.Y., Kuhl, T.L., Israelachvili, J.N., Mullah, N., Zalipsky, S., and Marques, C.M. 2001. Impact of polymer tether length on multiple ligand-receptor bond formation. Science 293 465–468. [DOI] [PubMed] [Google Scholar]
  15. Jo, S. and Park, K. 2000. Surface modification using silanated poly(ethylene glycol)s. Biomaterials 21 605–616. [DOI] [PubMed] [Google Scholar]
  16. Kinstler, O.B., Brems, D.N., Lauren, S.L., Paige, A.G., Hamburger, J.B., and Treuheit, M.J. 1996). Characterization and stability of N-terminally PEGylated rhG-CSF. Pharm. Res. 13 996–1002. [DOI] [PubMed]
  17. Kuan, C.T., Wang, Q.C., and Pastan, I. 1994. Pseudomonas exotoxin A mutants. Replacement of surface exposed residues in domain II with cysteine residues that can be modified with polyethylene glycol in a site-specific manner. J. Biol. Chem. 269 7610–7616. [PubMed] [Google Scholar]
  18. Lee, J.C. and Lee, L.L. 1981. Preferential solvent interactions between proteins and polyethylene glycols. J. Biol. Chem. 256 625–631. [PubMed] [Google Scholar]
  19. Lee, L.L. and Lee, J.C. 1987. Thermal stability of proteins in the presence of poly(ethylene glycols). Biochemistry 26 7813–7819. [DOI] [PubMed] [Google Scholar]
  20. Li, Z., Lee, W.E., and Galley, W.C. 1989. Distance dependence of the tryptophan-disulfide interaction at the triplet level from pulsed phosphorescence studies on a model system. Biophys. J. 56 361–367. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Loewenthal, R., Sancho, J., and Fersht, A.R. 1992. Histidine-aromatic interactions in barnase. Elevation of histidine pKa and contribution to protein stability. J. Mol. Biol. 224 759–770. [DOI] [PubMed] [Google Scholar]
  22. Lu, M. and Tjerneld, F. 1997. Interaction between tryptophan residues and hydrophobically modified dextran. Effect on partitioning of peptides and proteins in aqueous two-phase systems. J. Chromatogr. A. 766 99–108. [DOI] [PubMed] [Google Scholar]
  23. McNutt, M., Mullins, L.S., Raushel, F.M., and Pace, C.N. 1990. Contribution of histidine residues to the conformational stability of ribonuclease T1 and mutant Glu-58—Ala. Biochemistry 29 7572–7576. [DOI] [PubMed] [Google Scholar]
  24. McPherson, A. 1976. Crystallization of proteins from polyethylene glycol. J. Biol. Chem. 251 6300–6303. [PubMed] [Google Scholar]
  25. Molineux, G., Kinstler, O., Briddell, B., Hartley, C., McElroy, P., Kerzic, P., Sutherland, W., Stoney, G., Kern, B., Fletcher, F.A., Cohen, A., Korach, E., Ulich, T., McNiece, I., Lockbaum, P., Miller, M.-M.A., Gardner, S., Hunt, T., and Schwab, G. 1999. A new form of Filgrastim with sustained duration in vivo and enhanced ability to mobilize PBPC in both mice and humans. Exp. Hematol. 27 1724–1734. [DOI] [PubMed] [Google Scholar]
  26. Moreland, L.W., McCabe, D.P., Caldwell, J.R., Sack, M., Weisman, M., Henry, G., Seely, J.E., Martin, S.W., Yee, C.L., Bendele, A.M., Frazier, J.L., Kohno, T., Cosenza, M.E., Lyons, S.A., Dayer, J.M., Cohen, A.M., and Edwards, C.K., 3rd. 2000. Phase I/II trial of recombinant methionyl human tumor necrosis factor binding protein PEGylated dimer in patients with active refractory rheumatoid arthritis. J. Rheumatol. 27 601–609. [PubMed] [Google Scholar]
  27. Naismith, J.H., Devine, T.Q., Kohno, T., and Sprang, S.R. 1996. Structures of the extracellular domain of the type I tumor necrosis factor receptor. Structure 4 1251–1262. [DOI] [PubMed] [Google Scholar]
  28. Pepinsky, R.B., LePage, D.J., Gill, A., Chakraborty, A., Vaidyanathan, S., Green, M., Baker, D. P., Whalley, E., Hochman, P.S., and Martin, P. 2001. Improved pharmacokinetic properties of a polyethylene glycol-modified form of interferon-beta-1a with preserved in vitro bioactivity. J. Pharmacol. Exp. Ther. 297 1059–1066. [PubMed] [Google Scholar]
  29. Reddy, K.R. 2000. Controlled-release, pegylation, liposomal formulations: New mechanisms in the delivery of injectable drugs. Ann. Pharmacother. 34 915–923. [DOI] [PubMed] [Google Scholar]
  30. Rodseth, L.E., Brandhuber, B., Devine, T.Q., Eck, M.J., Hale, K., Naismith, J.H., and Sprang, S.R. 1994. Two crystal forms of the extracellular domain of type I tumor necrosis factor receptor. J. Mol. Biol. 239 332–335. [DOI] [PubMed] [Google Scholar]
  31. Sasahara, K. and Uedaira, H. 1993. Solubility of amino acids in aqueous poly(ethylene glycol) solutions. Colloid and Polymer Science 271 1035–1041. [Google Scholar]
  32. Solorzano, C.C., Kaibara, A., Hess, P.J., Edwards, P.D., Ksontini, R., Abouhamze, A., McDaniel, S., Frazier, J., Trujillo, D., Kieft, G., Seely, J., Kohno, T., Cosenza, M.E., Clare, S.-M., MacKay, S.L., Martin, S.W., Moldawer, L.L., and Edwards, C.K., 3rd. 1998. Pharmacokinetics, immunogenicity, and efficacy of dimeric TNFR binding proteins in healthy and bacteremic baboon. J. Appl. Physiol. 84 1119–1130. [DOI] [PubMed] [Google Scholar]
  33. Strambini, G.B. and Gonelli, M. 1995. Tryptophan phosphorescence in fluid solution. J. Am. Chem. Soc. 117 7646–7651. [Google Scholar]
  34. Tuma, R., Russell, M., Rosendahl, M., and Thomas, G.J., Jr. 1995. Solution conformation of the extracellular domain of the human tumor necrosis factor receptor probed by Raman and UV-resonance Raman spectroscopy: Structural effects of an engineered PEG linker. Biochemistry 34 15150–15156. [DOI] [PubMed] [Google Scholar]

Articles from Protein Science : A Publication of the Protein Society are provided here courtesy of The Protein Society

RESOURCES