Protein isolated from biopharmaceutical formulations cannot be used for comparative studies: Follow-up to "a case study using Epoetin Alfa from Epogen and EPREX".

In the biotechnology area, the issue of comparability with an innovator product is complex. Ideally, a side-by-side comparison of physical properties would be part of the demonstration of comparability. However, biogeneric companies do not have access to the bulk drug substance from the innovator company for biophysical comparison, and isolation of protein from marketed product cannot be guaranteed to produce material that is identical to the bulk drug substance from which it was prepared. In a recently published study, protein was isolated from marketed product and comparative studies performed. In a follow-up investigation of the published work, we demonstrate here that even a simple isolation procedure can significantly compromise the protein, which raises serious questions about the interpretation of that study, and in a broader context the value of any studies done with such "out-of-process" protein.

Sites of iodination in recombinant human brain-derived neurotrophic factor and its effect on neurotrophic activity.

Recombinant human brain-derived neurotrophic factor (BDNF) is now under extensive investigation because of its potential clinical applications. Radioactively labeled proteins are usually required to study receptor binding and pharmacokinetic properties of proteins. This study was undertaken to see if iodination affects the biological and conformational properties of a recombinant BDNF. BDNF was iodinated using a stoichiometric amount of nonradioactive cold NaI to minimize multiple iodinations. Of the four tyrosines present in BDNF--Tyr-52, Tyr-54, Tyr-63, and Tyr-86--only Tyr-63 and Tyr-86 were iodinated under the experimental conditions used. Iodination of Tyr-63 resulted in modification without alteration of the biological activity, whereas iodination of Tyr-86 resulted in a molecule with highly compromised biological activity. Similar inactivation was observed if both Tyr-63 and Tyr-86 were iodinated. These modified proteins exhibited conformation and dimerization apparently identical to those of the native protein, as demonstrated by analytical ultracentrifugation, gel filtration, light scattering, and circular dichroism. From these results, we concluded that Tyr-52 and Tyr-54 are not accessible to the reagent and are probably buried in the hydrophobic core, whereas Tyr-63 and Tyr-86 are exposed on the surface of the molecule; of the two exposed residues, only Tyr-86 contributes to the biological activity.

Complete sequence, subunit structure, and complexes with pancreatic alpha-amylase of an alpha-amylase inhibitor from Phaseolus vulgaris white kidney beans.

The complete amino acid sequence of a white kidney bean (Phaseolus vulgaris) alpha-amylase inhibitor (PHA-I), which is composed of two kinds of glycopolypeptide subunits, alpha and beta, was established by conventional methods. The polypeptide molecular weight of PHA-I determined by the light-scattering technique, considered together with the sequence molecular weights revealed for the subunits, indicated that PHA-I has the subunit stoichiometry of (alpha beta)2 complex. Inhibition test of PHA-I with increasing amounts of porcine pancreatic alpha-amylase (PPA) suggested that an inactive 2:1 complex is formed between PPA and PHA-I. In fact, two complexes differing from each other in the molar ratio of PPA to PHA-I were separated by gel filtration, and molecular weight estimation by the light-scattering technique confirmed that they are complexes of PHA-I with one or two PPA molecules. The binding of PPA to PHA-I appeared to follow simple binomial statistics, suggesting that two binding sites on PHA-I are independent and of high affinity for PPA.

Structural characterization of an alpha-amylase inhibitor from a wild common bean (Phaseolus vulgaris): insight into the common structural features of leguminous alpha-amylase inhibitors.

The primary structures of two subunits of an alpha-amylase inhibitor (alpha AI-2) from a wild common bean (Phaseolus vulgaris) were revealed by a comparison of the amino acid sequence previously deduced from the nucleotide sequence with the amino- and carboxyl-terminal amino acid sequences determined by conventional methods. The polypeptide molecular weight of alpha AI-2 obtained by the light-scattering technique, considered together with the sequence molecular weights revealed for the subunits, indicated that alpha AI-2 has the subunit stoichiometry of an alpha 2 beta 2 complex. These structural features were closely similar to those recently elucidated for a white kidney bean (P. vulgaris) alpha-amylase inhibitor, which is quite different in the inhibitory specificity from alpha AI-2. The post-translational processing of the precursor glycoproteins to form the tetrameric structure appeared to require an Arg residue close to the processing site. Further, the proper associations of the subunits into the tetrameric structures seemed to be strictly controlled by a few amino acids on the subunit interfaces.

Dimerization of granulocyte-colony stimulating factor receptor: the Ig plus CRH construct of granulocyte-colony stimulating factor receptor forms a 2:2 complex with a ligand.

We have previously shown that the extracellular domain of granulocyte-colony stimulating factor receptor (soluble G-CSFR), prepared from CHO cell conditioned media, dimerizes upon binding its ligand, G-CSF. The most stable ligand-receptor complex occurs at a 2:2 stoichiometry, unlike the growth hormone and erythropoietin systems. In the latter cases, each ligand uses two sites to bring two receptors together. In this study, we have generated a truncated G-CSF receptor, known to be sufficient for high affinity ligand binding, which consists of an Ig-like domain and a cytokine receptor homology module. With an affinity purified receptor, sedimentation equilibrium experiments clearly demonstrated that this truncated form of the receptor behaves very similarly to the entire extracellular domain. The sedimentation equilibrium data are consistent with the model that the truncated receptor has a weak tendency to self-associate into a dimer in the absence of a ligand, this receptor-receptor interaction is enhanced by ligand binding, and the most stable complex occurs at a 2:2 stoichiometry. These results are very different from those described by others for various murine G-CSF receptor constructs from either Escherichia coli or insect expression systems.

A chimera-like alpha-amylase inhibitor suggesting the evolution of Phaseolus vulgaris alpha-amylase inhibitor.

White kidney bean (Phaseolus vulgaris) contains two kinds of alpha-amylase inhibitors, one heat-stable (alpha AI-s) and one heat-labile (alpha AI-u). alpha AI-s has recently been revealed to be a tetrameric complex, alpha(2)beta(2), with two active sites [Kasahara et al. (1996) J. Biochem. 120, 177-183]. The present study was undertaken to reveal the molecular features of alpha AI-u, which is composed of three kinds of subunits, alpha, beta, and gamma. The gamma-subunit, in contrast to the alpha- and beta-subunits that are indistinguishable from the alpha- and beta-subunits of alpha AI-s, was found to correspond to a subunit of an alpha-amylase inhibitor-like protein, which has been identified as an inactive, evolutionary intermediate between arcelin and the alpha-amylase inhibitor in a P. vulgaris defense protein family. The polypeptide molecular weight of alpha AI-u determined by the light-scattering technique, together with the polypeptide molecular weights of the subunits, suggests that alpha AI-u is a trimeric complex, alpha beta gamma. The inhibition of alpha AI-u by increasing amounts of porcine pancreatic alpha-amylase (PPA) indicates that an inactive 1:1 complex is formed between alpha AI-u and PPA. Molecular weight estimation of the complex by the light-scattering technique confirmed that it is a complex of alpha AI-u with one PPA molecule. Thus it seems probable that alpha AI-u is an evolutionary intermediate of the P. vulgaris alpha-amylase inhibitor.

Kinetics of oxygen binding and subunit assembly for the hemoglobin alpha subunit.

A thorough kinetic characterization of the O2-binding and self-association reactions of alpha-subunits of human hemoglobin A has been performed. All of the rate constants for a five step reaction model linking the monomer-dimer reaction to the O2-binding steps have been determined for the first time. Our analysis of the ligand binding reaction shows that both monomer and dimer have nearly identical intrinsic O2-association and dissociation rate constants and therefore identical affinities for oxygen. During this investigation we discovered a small absorbance difference between the oxy-monomer and oxy-dimer alpha-subunits. This difference spectrum enabled direct measurements of the alpha O2 self-association reaction. We find an association rate constant of, 2.0 10(5) M(-1)s-1, similar to that for other subunit assembly processes in the hemoglobin system. Our results also suggest that the deoxy-subunit assembly kinetics must be similar to that for the oxy-subunit. These kinetic results together with the equilibrium constants obtained for these solution conditions by Ackers and coworkers provides, for the first time, a complete kinetic and thermodynamic description of all the intrinsic ligand binding and association reactions for alpha-subunits.

[Applications of analytical ultracentrifuge to molecular biology and pharmaceutical science].

Analytical ultracentrifuges, XL-A and XL-I, developed by Beckman Company now find a broad application not only in universities but also in industries. Especially they are utilized conveniently in industries aiming at the development of proteins as a therapeutic drug or in those targeting drugs composed of small molecules developed on the basis of the structures of proteins. Sedimentation techniques can be used 1) to determine the molecular weight of proteins in solution, 2) to examine protein aggregation, 3) to evaluate the molecular shape of proteins, 4) to study the interaction of proteins, e.g. between ligands and receptors, and 5) to obtain insight into biological functions of homologous proteins. Application of this technique to molecular biology and pharmaceutical science will be reviewed with ample examples.

Overview of the quantitation of protein interactions.

The biological function of many proteins involves reversible interactions with other proteins, nucleic acids, or other non-protein ligands. Such interactions play many different roles in a wide range of cellular processes. A few examples are: (1) storing or transporting key metabolites (e.g., O(2) storage by myoglobin); (2) forming and maintaining the quaternary structure of multi-subunit enzymes; (3) specific binding and recognition events (antigen-antibody, hormone-receptor, transcription factor-promoter); and (4) self-assembly of large structures (microtubules, chromatin). Thus, the quantitative characterization of such interactions represents an important part of understanding the function of such proteins and their role in these cellular events. This unit sets the tone for the rest of the chapter, and gives important information necessary to understand some of the topics that will be covered in future supplements, such as sedimentation equilibrium (analytical and micro-preparative), surface plasmon resonance (SPR), size-exclusion chromatography (SEC) with on-line light scattering, and chemical cross-linking.

An improved function for fitting sedimentation velocity data for low-molecular-weight solutes.

Many traditional approaches to the analysis of sedimentation velocity data work poorly with data for low-molecular-weight solutes, which have sedimentation boundaries that are severely broadened by diffusion. An approach that has previously had some success is to directly fit these broad boundaries to approximate solutions of the Lamm equation that directly account for the high diffusion. However, none of the available approximate solutions work well at times both early and late in the run, or give boundary shapes that are highly accurate, especially for species of molecular weight < 10,000. An improved fitting function has been developed to overcome some of these limitations. The new function adds two correction terms to the Fujita-MacCosham solution. The optimum coefficients for these new correction terms were determined by a least-squares approach. The accuracy and limitations of fitting with this new function were tested against synthetic data sets obtained by finite-element methods, for analysis of samples containing either single species or several noninteracting species. We also compare the strengths and weaknesses of this method of analysis, and its ability to work with noisy data, relative to recently developed time-derivative methodologies.

Kinetics of hemoglobin-carbon monoxide reactions measured with a superconducting magnetometer: a new method for fast reactions in solution.

A new technique for measuring fast reactions in solution has been demonstrated. The changes in magnetic susceptibility during the recombination reaction of human hemoglobin with carbon monoxide after flash photolysis have been measured with a new high-sensitivity instrument using cryogenic technology. The rate constants determined at 20 degrees (pH 7.3) are in excellent agreement with those obtained by photometric techniques [Gray, R.D. (1974) J. Biol. Chem. 249, 2879-2885]. A unique capability of this new method is the determination of the magnetic susceptibilities of short-lived reaction intermediates. The magnetic moment of the intermediate species Hb4(CO)3 was found to be 4.9+/-0.1 muB in 0.1 M phosphate buffer by partial photolysis experiments. This value agrees with the predictions of two-state allosteric models of cooperativity in hemoglobin. Possible applications and improvements in this technique are discussed.

A method for directly fitting the time derivative of sedimentation velocity data and an alternative algorithm for calculating sedimentation coefficient distribution functions.

The time-derivative method for deriving the sedimentation coefficient distribution, g(s*), from sedimentation velocity data that was developed by Walter Stafford has many advantages and is now widely used. By fitting Gaussian functions to the g(s*) distribution both sedimentation and diffusion coefficients (and therefore molecular masses) for individual species can be obtained. However, some of the approximations used in these procedures limit the accuracy of the results. An alternative approach is proposed in which the dc/dt data are fitted rather than g(s*). This new approach gives improved accuracy, extends the range to sedimentation coefficients below 1 S, and enhances resolution of multiple species. For both approaches the peaks from individual species are broadened when the data cover too wide a time span, and this effect is explored and quantified. An alternative algorithm for calculating g(s*) from the dc/dt curves is presented and discussed. Rather than first averaging the dc/dt data for individual scan pairs and then calculating g(s*) from that average, the g(s*) distributions are calculated for every scan pair and then subsequently averaged. This alternative procedure yields smaller error bars for g(s*) and somewhat greater accuracy for fitted hydrodynamic properties when the time span becomes large.

Quaternary structure has little influence on spin states in mixed-spin human methemoglobins.

A key feature of the Perutz stereochemical model for cooperativity in hemoglobin is a strong coupling between quaternary structure and the spin state of the heme iron [Perutz, M. F. (1979) Annu. Rev. Biochem. 48, 327-386]. While this coupling appears to be present for carp azide methemoglobin, it should also be present for all liganded forms of human methemoglobin that exhibit a thermal high-spin in equilibrium low-spin equilibrium. To test this hypothesis, we have measured the changes in spin equilibria upon conversion of six mixed-spin forms of human methemoglobin from the R (high-affinity) to the T (low-affinity) quaternary structure by addition of inositol hexaphosphate. These experiments were done with a sensitive superconducting magnetic susceptibility instrument on solutions at 20 degrees C in 20 mM maleate buffer, pH 6. The data show zero or small increases in high-spin content upon switching from R to T, changes that are equivalent to a relative stabilization of the high-spin form by only 0-300 cal mol-1 heme-1. These changes in energy are far less than the 1200 cal mol-1 heme-1 predicted from the Perutz stereochemical model [Cho, K. C., & Hopfield, J. J. (1979) Biochemistry 18, 5826-5833]. That is, these data do not support a view that the low affinity of the T state is due to restraints acting through the iron-proximal histidine linkage. The mechanistic implications of these results and the differences between species and ferric ligands are discussed.

Kinetic investigations of the quaternary enhancement effect and alpha/beta differences in binding the last oxygen to hemoglobin tetramers and dimers.

Analysis of O2 binding equilibria by two independent groups has suggested that the affinity for binding the fourth O2 to Hb tetramers is very high, about 800-1200 cal/mol higher than that of dimers (Chu, A. H., Turner, B. W., and Ackers, G. K. (1984) Biochemistry 23, 604-167; Di Cera, E., Robert, C. H., and Gill, S. J. (1987) Biochemistry 26, 4003-4008). Recently, Gibson and Edelstein challenged the reality of the quaternary enhancement effect, based on kinetic data (Gibson, Q. H., and Edelstein, S. J. (1987) J. Biol. Chem. 262, 516-519). However, these studies failed to directly address the key issue of the relative affinities of dimers and alpha 2 beta 2(O2)3. Furthermore, the extent to which alpha/beta differences influence these results remains an open question. Using partial laser photolysis and O2/CO replacement techniques we have, for the first time, resolved the rates of O2 association and dissociation to both alpha and beta chains within "R state" tetramers and dimers. We find that the beta chains are faster than alpha for both O2 binding (approximately 2-fold) and release (approximately 3-fold). The kinetically determined O2 affinities derived from these data are essentially identical for dimers and alpha 2 beta 2(O2)3. That is, the data do not show significant quaternary enhancement and suggest that the equilibrium data have both overestimated the affinity of alpha 2 beta 2(O2)3 and underestimated the affinity of dimers. The significance of and possible origins for the discrepancy between equilibrium and kinetic data are discussed.

Densimetric determination of equilibrium binding of sucrose octasulfate with basic fibroblast growth factor.

Fibroblast growth factors (FGFs) strongly bind to heparin and are thereby stabilized against deactivation and proteolytic cleavage. Sucrose octasulfate (SOS), which has a chemical structure resembling the repeating unit of heparin, has also been shown to enhance stability of basic FGF against thermal denaturation and to induce a small conformational change. We have examined SOS binding to bFGF using equilibrium dialysis. The difference in SOS concentration across the dialysis membrane was measured using a precision density meter, since the density of SOS differs greatly from that of water. With care, this densimetric technique can measure binding with a precision of +/- 0.1 mol/mol using about 2 mg/ml of protein. These results show that the binding saturates at 2 mol of SOS per mole of bFGF as the SOS concentration increases to 3.6 mM or higher. The effect of SOS on the thermal stability of bFGF was examined using denaturation at a constant heating rate, by both turbidity and differential scanning calorimetry. Since the thermal denaturation is irreversible, the temperature where aggregation abruptly increases was taken to indicate the onset of denaturation. This temperature increased by approximately 12 degrees C as the SOS concentration increased from 0.018 to 3.6 mM and remained constant above 3.6 mM, consistent with our binding data if the binding is specific to the native state.

Association-dependent absorption spectra of oxyhemoglobin A and its subunits.

A number of factors which lower the oxygen affinity of hemoglobins are also known to produce shifts of the absorption band maxima of the oxyheme. We have studied the variation of the absorption spectra of oxygenated alpha SH and beta SH subunits of human Hb A as a function of their state of association (i.e. alpha, alpha 2, beta, beta 4, alpha beta, or alpha 2 beta 2), and have attempted to correlate the spectral changes with changes to O2 affinity. These studies were carried out under solution conditions (0.1 M Tris, 0.1 M NaCl, 1 mM Na2EDTA, pH 7.4, 10 degrees C) where detailed thermodynamic data for subunit association and oxygen binding are available (Ackers, G. K. (1980) Biophys. J. 32, 331-346). Concentration-difference spectra reveal that the visible and Soret absorption band maxima of beta O2 are slightly red shifted relative to beta 4O8. A unique feature of this spectral change is that the red shift is accompanied by an increase in the ratio of the peak absorbances of the visible alpha and beta spectral bands. By measuring the spectral change as a function of concentration, an association constant of 6.4 +/- 1.9 X 10(15) M-3 was determined for the 4(beta O2) in equilibrium beta 4O8 equilibrium. In contrast, no spectral differences were found between alpha O2 and alpha 2O4 or between oxy alpha beta dimers and oxyHb. Mixing experiments show that the spectrum of oxyHb differs from the average of either alpha O2 + beta O2 or alpha O2 + beta 4O8, but is closer to the former. A comparison between these spectral data and the reported O2 affinities of these species shows that affinity and oxyheme spectra are not correlated.

Stoichiometry of heparin binding to basic fibroblast growth factor.

Fibroblast growth factors (FGFs) strongly bind to heparin and are thereby stabilized against deactivation and proteolytic cleavage. We have investigated the interactions of basic fibroblast growth factor (bFGF) with low- and high-molecular-weight heparin using size exclusion chromatography with on-line light scattering, absorbance, and refractive index detection. When heparin-bFGF mixtures with excess heparin are chromatographed using eluant that does not contain heparin, essentially all the protein is seen to elute as a complex with the heparin, indicating strong binding such that the complex does not dissociate significantly during chromatography (approximately 20 min). Combining the data from the light scattering, absorbance, and refractive index chromatograms allows us to determine the molecular weight of the protein component of the complex, and therefore to measure the number of bFGF molecules bound per heparin. A series of samples were prepared with a constant concentration of bFGF and variable amounts of a low-molecular-weight heparin (LMWH, M(r) = approximately 5000). At bFGF: heparin ratios above 1.5, a mix of complexes containing 3, 2, and 1 bFGF molecules is observed, with an average of 2.2 bFGF molecules per complex. Since the amount of bFGF incorporated into complexes implies an average of 2.5 +/- 0.3 bFGF molecules per heparin, there is only one heparin molecule per complex. The coexistence of complexes of different size when bFGF is in excess implies that the LMWH molecules are heterogeneous with respect to their ability to bind bFGF. When a high-molecular-weight heparin (HMWH, M(r) = 15,000) is used, complexes averaging 6.3 bFGF molecules per HMWH molecule are seen, while the overall amount of bFGF appearing in complexes implies six to seven sites per HMWH. These data show that the protein molecules can be packed very closely together. Both types of heparin give a heparin mass of 2300 Da per bFGF binding site, which corresponds approximately to an octasaccharide.

Size-exclusion chromatography with on-line light-scattering, absorbance, and refractive index detectors for studying proteins and their interactions.

Techniques of using size-exclusion chromatography (SEC) with on-line light-scattering, uv absorbance, and refractive index detectors to characterize the polypeptide molecular weights of simple proteins or glycoproteins or to determine the stoichiometry of protein complexes are described. Two unique advantages of this approach over conventional SEC are that the molecular weight measurement is independent of elution position and can exclude the contributions from carbohydrates. When a protein or complex contains no carbohydrates, a two-detector method, i.e., light scattering combined with refractive index, can be used to calculate the molecular weight. When a protein contains carbohydrates, a three-detector method is used to calculate the molecular weight of polypeptide alone. Finally, a self-consistent three-detector method is used to determine the stoichiometry of a protein complex containing carbohydrates. Example applications for all these methodologies are described.

Quaternary structure dynamics and carbon monoxide binding kinetics of hemoglobin valency hybrids.

The kinetics of CO binding and changes in quaternary structure for symmetric valency hybrids of human hemoglobin have been extensively studied by laser photolysis techniques. Both alpha+beta and alpha beta+ hybrids were studied with five different ferric ligands, over a broad range of CO concentrations and photolysis levels. After full CO photolysis, the hybrid tetramers switch extensively and rapidly (< 200 microseconds) to the T quaternary structure. Both R --> T and T --> R transition rates for valency hybrid tetramers with 0 and 1 bound CO have been obtained, as well as the CO association rates for alpha and beta subunits in the R and T states. The results reveal submillisecond R reversible T interconversion, and, for the first time, the changes in quaternary rates and equilibria due to binding a single CO per tetramer have been resolved. The data also show significant alpha-beta differences in quaternary dynamics and equilibria. The allosteric constants do not vary with the spin states of the ferric subunits as predicted by the Perutz stereochemical model. For the alpha beta+CN hybrid the kinetics are heterogeneous and imply partial conversion to a T-like state with very low (seconds) R reversible T interconversion.