Size Exclusion Chromatography - Analytical Chemistry (ACS


Size Exclusion Chromatography - Analytical Chemistry (ACS...

21 downloads 121 Views 4MB Size

Anal. Chem. 1994,66, 595R-620R

Size Exclusion Chromatography Howard G. Barth,'lt Barry E. Boyes,* and Christian Jacksont DuPont Company, Central Research and Development, Experimental Station, P.0. Box 80228, Wilmington, Dela ware, 19880-0228, and Rockland Technobgies, Inc., 538 First State Boulevard, Newport, Delaware 19804 Review Contents Books and Symposia Reviews Theory Band Broadening Calibration General Information Universal Calibration Data Processing Non-Size-Exclusion Effects Shear Degradation/Concentration Effects Adsorption Effects/Mobile-Phase Selection Detectors Light-Scattering Detectors Viscometers Combination Light-Scattering/Viscosity Detectors Conductivity Electrochemical Evaporative Light Scattering Inductively Coupled Plasma Mass Spectrometry and Atomic Emission Infrared Spectrometry Mass Spectrometry NMR Raman Spectroscopy X-ray Fluorescence Packings Inorganic-Based Packings Polymeric-Based Packings Compositional Heterogeneity SEC with Selective Detectors Interactive HPLC Temperature Rising Elution Fractionation Orthogonal Chromatography Physicochemical Studies Synthetic Polymers Biopolymers Microcolumn SEC Preparative SEC Coupled Columns/Column Switching Automation/Quality Control Selected Applications Synthetic Polymers Asphalt, Bitumens, Fossil Fuels, and Related Products Plant Polysaccharides and Cellulosics Lignins and Tannins Natural Oils, Fats, and Lipids Humic Acids and Related Substances Inorganic Compounds and Particles Membrane Characterization Sample Cleanup/Pretreatment Biopolymers 0003-2700/94/0366-0595814.0010 0 1994 American Chemical Society

596R 596R 597R 597R 597R

598R

599R

601R

602R

603R

609R 609R 609R 610R 610R

The term size exclusion chromatography (SEC), rather than gel permeation, gel filtration, steric exclusion, exclusion, or gel chromatography, has been used in this review. SEC is an entropically controlled separation technique that depends on the relative size or hydrodynamic volume of a macromolecule with respect to the average pore size of the packing. It is important to realize that SEC is a relative technique that requires column calibration in order to determine statistical average molecular weights and the molecular weight distribution of polymers. Absolute molecular weight measurements are possible, however, with the use of either an on-line light scattering detector or an on-line viscometer with universal calibration. With these detectors, it is possible also to determine molecular conformation and long-chain branching. In addition to being able to obtain molecular parameters, SEC is useful also for preparative fractionation of polymers and for separating small molecules from complex polymeric or biogenic matrices as an aid to sample cleanup. SEC has become a mature and well-accepted technique for characterizing both synthetic polymers and biopolymers. We see increased usage of high-performance columns, as compared to soft-gel conventional packings, by life scientists. High-performance columns are now favored because of their speed and high resolution. For aqueous SEC,either silicaor organic-based packings are employed. Silica-based packings seem to be preferred for quality control and quality assurance because of their higher efficiency (smaller particle size), shorter analysis time, and more robust nature; however, unwanted solute-packing interaction can be a major concern. For organosoluble polymers, polystyrene gels remain the packing of choice. During this review period, there continues to be less emphasis placed on theoretical aspects of SEC,including band broadening, and more focus on applications, especially the use of on-line light-scattering detectors and viscometers. With these molecular weight-sensitive detectors, one can now determine branching, molecular size, and conformation as a function of molecular weight in a single analysis. The use of both these detectors, together with a concentration-sensitive detector, has greatly improved the accuracy and precision of these measurements. We see increased publications on the determination of compositional heterogeneity of oligomers and polymers using SEC with on-line spectrophotometric detection, such as UV and FT-IR. The most important advancement in this area is the use of mass spectrometry, such as matrix-assisted laser+ DuPont Co. 8

Rockland Technologies, Inc.

Anal'icalChemlstry,

Vol. 66, No. 12, June 15, 1994 S9SR

Howard G. Barth is a research associate of the Corporate Center for Analytical Sciences at the DuPont Experimental Station, Wilmington, DE. Before joining the DuPont Co. in 1988, he was a research scientist and group leader at Hercules Research Center. He received his B.A. (1969) and Ph.D. (1973) in analytical chemistry from Northeastern University. His speciatties include polymer characterization, size exclusion chromatography,and high-performance liquid chromatography. He has published over 60 papers in these and related areas. Barth has also edited the book Modern Methods of Particle Size Analysis (Wiley, 1984) and coedited Modern Methods of Polymer Characterization (Wiley, 1991). He has also edited five symposium volumes on polymer characterizationpublished in the JournalofApplied Polymer Science. Barth was on the Instrumentation Advisory Panel of Analyfical Chemistry and was Associate Editor of the Journal of Applied Polymer Science. He is cofounder and Chairman of the International Symposium on Polymer Analysis and Characterization. He has been appointed recently editor-in-chief of the International Journal of Polymer Analysis and Characterization. Barth is past Chairman of the Delaware Section of the American Chemical Society where he presently serves as councilor. Dr. Barth is a member of the American Chemical Society divisions of Analytical Chemistry, Polymer Chemistry, and Polymeric Materials Science and Engineering,the Society of Plastics Engineers, and the Delaware Valley Chromatography Forum. He is also a Fellow of the American Institute of Chemists and a member of Sigma Xi. Barry E. Boyes is a research and development scientist with Rockland Technologies Inc., Newport, DE. He received his B.Sc. in biochemistry in 1982 at the University of Alberta (Edmonton, Alberta). Barry carried out his graduate work at the University of British Columbia (Vancouver, BC), where he received the M.Sc. in neurological research ( 1985), working with Professor S.-C.Sung, and the Ph.D. in neurosciences(199 1)under Professor Edith G. McGeer. Before joining Rockland T e c h nologies in 1991, he spent 11/* years as a visiting scientist in the Analytical Division of the Central Research and Development Department at the DuPont Experimental Station in Wilmington, DE. At Rockland Technologies, Barry is responsible for the development of new products for analytical biochemistry. The research activities of Dr. Boyes have included the development of analytical techniques for molecular neurobiology and methods for the purification and analysis of biopolymers by chromatography, hydrodynamics, and electrophoresis. Christian Jackson is a research scientist in the Corporate Center for Analytical Sciences of the DuPont Co. at the Experimental Station, Wilmington, DE. He received a B.A. in physics from Oxford University (1983)and an M.S. in philosophy of physics from the University of London (1984). He is currently completing a Ph.D. in chemistry with Professor Julia S. Higgins at Imperial College, University of London. Before joining DuPont Co. in 1990, he worked in the research division of Wyatt Technology Corp. His research interests are in the area of polymer solutions, polymer characterization, and the study of polymer architecture and conformation. He has published over 20 papers in these and related areas.

desorption ionization and other soft ionization techniques, for characterizingboth the absolute molecular weight of polymers and their chemical identity. The application of MS to SEC is rapidly expanding, especially as laser-desorption methodologies improve. We were surprised, however, that on-line NMR detection has not experienced any growth, possibly because of the low sensitivity of the technique. 596R

Analytical Chemistry, Vol. 66, No. 12, June 15, 1994

The use of HPLC (adsorption, precipitation, and critical chromatography) for separating polymers and oligomers on the basis of chemical composition has continued to increase. Although these chromatographic methods are non-SEC separations, they are included in this review because of their importance. In terms of applications, SEC has proven valuable for the characterization of block copolymer micelles, as well as for the determination of biopolymer association, ligand binding, and conformational changes. Long-chain branching measurement as a function of molecular weight is now a routine determination with on-line viscometry. With the proper choice of mobile phase and packing type, most polymers and biopolymers can be characterized by SEC. Exceptions include amphipathic biopolymers which contain highly hydrophobic and ionic groups, such as membrane proteins. This review covers fundamental developments in SEC (and HPLC of polymers) abstracted by Chemical Abstracts and Biosis from 1992and 1993 and is a continuationof our previous review (BZ). Topics that are included in this review are listed in the Table of Contents. Writing these reviews is a long and laborious process that involves developing search routines, reading all abstracts and often full papers, and categorizing this information in a convenient format. To this end, we are grateful to Carol R. Perrotto (DuPont) for developing an excellent and thorough search routine, and to Rebecca Pennington (DuPont) and Marie Tobasso (Rockland Technologies) for their excellent typing skills. We also thank our respective companies for their support and encouragement in the preparation of the manuscript. A. BOOKS AND SYMPOSIA For this review period, only one book on SEC, written in Japanese by Mori ( A I ) ,has been published. The proceedings of the 1991 ACS Symposium on SEC and Field Flow Fractionation, edited by Provder ( A 2 ) , was recently issued. Proceedings of the Fourth and Fifth International Symposia on Polymer Analysis and Characterization (ISPAC) have appeared (A3, A4). These proceedings contain a number of contributions on SEC, in addition to other methodsof polymer characterization. The proceedings of ISPAC-6, held in Crete, Greece, in 1993, are in press. A new journal, International Journal of Polymer Analysis and Characterization, published by Gordon and Breach Publishers, will be launched in 1995. Interested readers should contact H. G. Barth for further information. B. REVIEWS Reviews of specific topics are listed in the appropriate sections. Barth and Boyes (BZ) presented a comprehensive coverage of the SEC literature on SEC from 1990 to 1991, this present review being a continuation of that format. Hagel (B2) presented a detailed review on SEC including the influence of operating parameters, properties and evaluation of packings, and guidelines for optimizing separations. Hagel (B3)also reviewed the resolving power and limitations of SEC column packings and the need to use well-calibrated columns. Francois and Sarazin (B4) discussed the possibilities and limitations of SEC.

Kuo and Provder (B5)covered the applications of SEC to polymers and coatings, including calibration methods and determining branching. Teramachi (B6)reviewed recent developments and applications of SEC. Mourey and Schunk (B7) discussed the use of LC to separate synthetic polymers. Ettre (B8) listed IUPAC-approved definitions for chromatography, including SEC. Ueda and Hasegawa (B9)reviewed the use of both mass spectrometry and SEC for determining molecular weights of polymers.

C. THEORY Potschka ( C I ) presented a comprehensive treatment of the role of convective and obstructed diffusion in SEC. A theory was established to describe obstructed diffusion and convection within the packing. Gibbs and colleagues (C2) determined the intradiffusion coefficient of ovalbumin in a polymeric SEC packing using pulsed-field gradient N M R as a function of protein concentration. Moussaoui et al. (C3, C4) measured the reduced diffusion of globular proteins in AcA-34 and Sepharose Cl-B packings using proteins labeled with fluorescein isothiocyanate. Reduced diffusion was measured from the fluorescence recovery after photobleaching the labeled proteins in the free state and in the gels. The two gels exhibited different retardation diffusion effects. Pavlov et al. (C5) correlated the retention volumes of polymers with translational diffusion coefficients, sedimentation coefficients, intrinsic viscosity, and molecular weight. Athalye et al. (C6) demonstrated that packings of narrow particle size distribution perform as well as monodisperse packings in terms of separation efficiency, confirming theoretical predictions. Gorbunov and Skvortsov (C7) established a relationship between resolution and pore size of the packing as applied to flexible-chain polymers and proteins. Renn and Synovec (C8)presented theoretical relationships for separation efficiency as a function of temperature for high-speed (10 mm/s) separations. Hagel (C9) introduced an equation for determining SEC peak capacity. He found that the practical peakcapacity of a high-performance column is approximately 13 for completely resolved peaks, which is roughly equal to the peak capacity of conventional SEC columns, the reason being the small pore volume of high-performance packings and short column lengths. Porcar et al. (C10) tested the fractal nature of silica packings using SEC of polystyrene in pure and mixed mobile phases. Fractal parameters depended on the eluent and were related to secondary separation effects, i.e., partition and adsorption contributions. Unger and eo-workers (CI I) determined the surface fractal dimensions of polymer-coated packings regarding protein separations. From fractal analysis, the authors proposed that the coated polymer is situated predominantly in inclusions and the pore volume is filled randomly with bulk polymer. Laurent (CI2) gave a historical perspective on his theory of gel filtration first published in 1964. Dubin et al. (CI3) studied the dependence of the SEC partition coefficient of branched and linear polysaccharides and a carboxylated dendrimer. All three solutes displayed a congruent dependence of the partition coefficient K on solute radius R . In accord with a simple geometric model for SEC, these data conform to the linear plot of K1/2 vs R . Comparison of these results

with the elution volume of globular proteins showed deviations from which protein-packing interactions can be measured. Van Steveninck et al. (C14)studied SEC retention behavior of small molecules in aqueous SEC to explain why K values sometimes deviate from unity. Previously accepted explanations have been solute-packing adsorption if K > 1 or inaccessibility to very small pores if K < 1. However, these authors claim that deviations from unity are caused by anomalous properties of vicinal water layers at the gel matrixwater interface. Kazakevich and McNair (CI5)discussed the definition of dead volume in HPLC, including SEC, and its determination from thermodynamics. The existence of negative values of capacity factors, Le., K < 1, fox some compounds is also explained. Malhorta et al. (C16) described a method for determining the accessible bed volume of gels for SEC of proteins. Potschka (C17) discussed interfacial mutual repulsion between the pore wall and surface of a particle and concluded that some interfacial repulsion is crucial in obtaining a pure SEC separation mechanism. Guszczynski et al. (CI8)studied the separation of poly(styrene sulfate) particles using electrophoresis and polymer solutions as the media for separation. Retardation of electrophoretic migration by either a sieving or a permeation mechanism was observed. Choi et al. (C19) described a method for the semicontinuous separation and concentration of two solutes based on the opposite swelling responses of two gels to a change in temperature.

D. BAND BROADENING Yau and colleagues (DI) presented an improved algorithm for characterizing skewed band broadening. The method produced results that were more precise and less sensitive to baseline noise than previous methods. Cheng and Zhu (D2) presented band broadening corrections for SEC-viscometry and Bielsa and Miera ( 0 3 ) presented a correction for instrumental band broadening in dual detector SEC. E. CALIBRATION General Information. Dubin et al. ( E l )used carboxylated starburst dendrimers as calibration standards for aqueous SEC. They found that the dendrimers acted as noninteracting spheres under appropriate solvent conditions and that there was a linear correlation between the dendrimer generation number and the chromatographic partition coefficient. The calibration was used to determine the size of quasi-spherical polysaccharides. Volpi and Bolognani (E2) studied the influence of the conformation of globular proteins and glycosaminoglycans in SEC. Although the exclusion limit molecular weight was much higher for globular proteins, at relative molecular weights below 1000 the calibration curves coincided. Kinugasa and Hattori (E3) used preparative SEC to prepare pure oligo(ethylene glycols) for calibration standards. Spychaj et al. (E4) used anionically polymerized oligomeric styrenes for calibration. Polystyrene calibration was found to give good results for polyimide-siloxane and polyimide-polyketone siloxane copolymers (E5). Bruessau et al. (E@ used a broad molecular weight distribution sodium polyacrylate calibration to characterize sodium polyacrylates and sodium acrylic acidmaleic acid copolymers. The influence of dialysis on the measured values of the refractive index increment and the Analytical Chemistry, Vol. 66, No. 12, June 15, 1994

5Q7R

necessity of achieving osmotic equilibrium were demonstrated. Holtzhauer and Rudolph used colloidal gold to characterize and calibrate SEC packing materials ( E 7 ) . Danielsson and Malmquist (E8)used multidimensional simplex interpolation to determine calibration curves, and Kang et al. (E9)used a nonlinear step regression method. The advantages of using chemometric methods for data treatment in the low molecular weight range was demonstrated (EIO). A number of methods were presented for combining the intrinsic viscosity with SEC data to determine the molecular weight distribution (E11-E17). Piskareva and Kartasheva (E18)combined SEC data with the average molecular weight and size obtained by light scattering to characterize branched polyisoprene. Wu (E19)combined the hydrodynamic volume determined by dynamic light scattering with SEC data to characterize gelatin. Universal Calibration, Lew et al. (E20) used SECviscometry to evaluate five different SEC calibration techniques. The use of weighting factors for fitting local calibration curves was shown to improve greatly the precision of the calculated molecular weight distributions. Sanayei and colleagues (E21)presented a method for constructing the calibration curve using the Stockmayer-Fixman equation rather than the Mark-Houwink equation to relate intrinsic viscosity to molecular weight. For high molecular weight fractions of polydisperse polyethylene, this calibration gave more accurate results than those obtained using molecular weight-sensitive detectors. Netopilik and colleagues (E22) found that universal calibration was not valid for polymethacrylates with mesogenic pendant groups and that the dilute solution properties were at variance with conventional polymer solution theory but could be explained by wormlike chain models. Bahary and Jilani (E23)used SEC-viscometry to assess the applicability of universal calibration to aqueous SEC. Universal calibration was used to characterize star block copolymers (E24),phenol-formaldehyde resin (E25),poly(ethylene glycol)-modified proteins (E26),and poly(viny1pyrrolidone) (E27). Data Processing. Vickory et al. (E28)found that the molecular weight distribution from polyethylene samples could be deconvoluted into five to seven individual Flory distributions. Deconvolution procedures were also used with SEC in the analysis of breast cancer estrogen and progesterone receptor isoforms (E29). A number of computer programs for processing SEC data were presented (E30,E31);in one case the program was developed for use with a pocket calculator

(E32). F. NON-SIZE-EXCLUSION EFFECTS Shear Degradation/Concentration Effects. Chubarova et al. ( F I ) studied shear degradation of high molecular weight polystyrene on macroporous glass membranes and related these results to SEC. Wills and co-workers (F2) reported on the macromolecular concentration dependence of virial coefficients. The implications of these findings are discussed in relation to results obtained by a number of techniques including SEC. Podosenova and Lebedev (F3) derived equations showing the effect of sample concentration on SEC elution behavior. S98R

AnatYticalChedstty, Vol. 66,No. 12, June 15, 1994

Adsorption Effects/Mobile-Phase Selection. Aqueous Systems. Brussau (F4) reviewed SEC of polyelectrolytes and difficulties associated with these types of polymers. Wu et al. (F5)discussed SEC of cationic, anionic, and nonionic copolymers of vinylpyrrolidone. Kuehn and colleagues (F6) presented approaches for mobile-phase optimization for quaternized poly(2-vinylpyridines) and poly(styrene sulfonates). Ono used acidic buffers containing tetrabutylammonium compounds for SEC of polymers containing a quaternary ammonium functionality (F7) and a phosphate buffer for the SEC of polyacrylamide (F8).Mori et al. (F9) studied the elution behavior of poly(ethy1ene glycols) using water, methanol, THF, and their mixtures and a hydrophilic polymethacrylate gel column. The SEC behavior of proteins and amino acids using TSKGEL Toyopearl was examined by Inouye (F10).Proteins having isoelectric points higher than the pH of the mobile phase are retained, but can be eluted in the SEC mode by the addition of 0.3-0.5 M NaCl or 30%ethanol, depending on the nature of the protein. Klyushnichenko and Vul’fson (FI I) studied the mechanisms of exclusion-adsorption of insulincontaining proteins on a TSK G 3000 SW column. Golovchenko et al. (F12)reported on the effect of mobile-phase ionic strength and pH on the SEC behavior of strongly hydrophobic and weakly hydrophobic proteins using a Superose 12 column. Edwards and Dubin (F13a)used Superose 6 to determine the chromatographic behavior of globular proteins as a function of pH. The elution behavior of hybridoma antibodies on Superose 12 was studied by Michaelson et al. (F13b)who speculated that retention on the column might have been caused by aromatic interactions between amino acid residues in the hypervariable region of the antibody and the gel matrix. Schlueter and Zidek (F14)used non-SEC effects with Sephacryl S-100 to purify and desalt amino acids and metabolites. Kang et al. (F15)studied protein adsorption on Sephacryl gels and Fractogel TSK-HW65 in the presence of high concentrations of ammonium sulfate. Porath and coworkers (F16)investigated the adsorption of albumin on Sepharose 6B using different buffers. Adsorption of aromatic amino acids on Bio-Gel P 2 during SEC of protein hydrolyzates was reported by Chorbanov and colleagues (F17). Herold (F18)tested the salt-dependent elution of proteins on different SEC packings. A binary-layered-phase packing, which contains a hydrophilic upper phase and a hydrophobic lower phase, was evaluated by Nimura et al. (F19).Large molecules, such as proteins, interact only with the hydrophilic surface and are size excluded. Small molecules are retained on the internal hydrophobic phase and are separated on the basis of reversedphase LC. Micallef et al. (F20) studied the effect of pH and electrolyte concentration on the elution of metal-binding proteins. The addition of SDS to the mobile phase improved the anomalous SEC behavior observed for soluble intercellular adhesion protein (F21a).This improvement was considered to be caused by a tertiary structural rearrangement of the protein. An SDS micellar mobile phase was used to determine high molecular weight impurities in ceftiofur (F21b). Ono (F22)found that the addition of sodium perchlorate to an aqueous mobile phase prevented interaction between SEC packings and many different types of polysaccharides.

Feste and Khan (F23) used SEC columns to separate maltooligosaccharides by interaction chromatography using an acetonitrilewater gradient. The chromatographic behavior of various &adrenoblockers with Sephadex G-25 was investigated as a function of mobile-phase pH and ionic strength (F24). Hu et al. (F25) studied the SEC behavior of cephalosporins using Sephadex G- 10with an ion-pair reagent. The SEC behavior of humic substances was studied using Sephadex G-100 and eluents containing disodium tetraborate (F26),porous glass packings and aqueous buffers (F27), and XAD resins and aqueous buffers (F28). Warwicket al. (F29) used SEC and a salt gradient to study nickel complexation with humic and fulvic substances. Sulfate lignin and lignosulfonate were chromatographed on Sephadex G-75 using as eluents DMSO or aqueous solutions, respectively (F30). Inamura and Uchida (F31)reported on the SEC behavior of chlorophyllin in water and in an aqueous poly(N-vinylpyrrolidinone) solution. The latter mobile phase was found to suppress aggregation of chlorophyllin. Wesslen and Wesslen (F32) added sodium lauryl sulfate or inorganic salts to lowionic-strength eluents to dissociate aggregates seen during SEC of amphiphilic graft copolymers of poly(ethy1ene glycol) grafted onto methacrylic copolymers. Okada (F33) studied the role of polymeric ligands added to the mobile phase with respect to secondary equilibrium SEC of small ions. Poly(ethy1eneglycol) and polybrene were added to the eluent to separate barium and alkali metal salts and salts of iodide, nitrate, and nitrite, respectively. Polymeric ligands altered the elution order and partition coefficients of the solutes. Nonaqueous Systems. Murakoshi (F34)discussed approaches for SEC measurements of slightly soluble polymers. Tokunaga (F35)used 20 mM triethylaminein T H F to prevent adsorption of tertiary amine-containing acrylic resins. Oguri et al. (F36) used chloroform containing 0.5% triethylamine for SEC of polyamide hardeners for epoxy resins, polyaniline, and organic tin. N-Methyl-2-pyrrolidinone was used as the mobile phase for high-temperature SEC of poly(4-vinylpyridine) (F37), and poly(pheny1ene sulfide) (F38). Hightemperature SEC of aggregate-free poly(viny1 chloride) solutions was reported by Pang and Rudin (F39) using trichlorobenzene as the mobile phase. These authors used an on-lineviscometer for molecular weight determinations. Balke and co-workers (F40)investigated the use of o-chlorophenol as the mobile phase for poly(ethy1ene terephthalate), and Onishi and Kawamoto (MI)used hexafluoro-2-propanol containing 0.1-2% water as the eluent for SEC of poly(buty1ene terephthalate). SEC of low molecular weight polyurethanes was accomplished using 0.01 M LiBr in T H F (F42). Radic et al. (F43) reported on non-size-exclusion effects during SEC of poly(n-alkyl itaconates) using T H F and cross-linked polystyrene packings. Elution of these polymers was higher than predicted by universal calibration. Mukoyama et al. (F44) investigated the SEC elution properties of polyamic acid and polyamideimide using a hydrophilic polystyrene packing and a hydrophobic polystyrene packing and DMF, THF, and a mixture of the two with and without phosphoric acid and lithium bromide as mobile phases.

Lafleur and Plummer (F45)examined the effects of column and mobile-phase modifications on retention behavior in SEC of polycylic aromatic hydrocarbons on cross-linked polystyrene packings. Interactions with a sulfonated polystyrene packing was studied including the addition of methanol to the mobile phase. The elution behavior of polystyrene, poly(methy1 methacrylate), polybutadienes, PS-PMMA blockcopolymers, and PS-PB star-shaped copolymers was studied with polystyrene packings and THF, toluene, chloroform, methylene chloride, and THF-cyclohexane mixtures as mobile phases (F46). Ono was awarded Japanese patents for the use of dichloroacetic acid as the mobile phase for the SEC of polyamides (F47),poly(ary1 sulfones) (F48), and polyketones (F49). Aromatic polyesters were also dissolved in dichloroacetic acid at elevated temperatures prior to SEC (F50).Dichloroacetic acid was used as the solvent for preparing polyketone or polysulfone-poly(thioether) solutions for subsequent analysis using chloroform and/or dichloroaceticacid as the eluent ( H I , F52). Ono also patented the use of DMF as the eluent for polyarylates (F53),polysulfones (FM),poly(viny1idene fluoride) (F55),and additives in fluoro rubbers (F56). Ethylenediamine was used as a solvent for a liquid crystalline polyester, and T H F as the eluent (F57).Lastly, Ono (F58) received a patent for using carbon tetrachloride as the eluent for SEC of polyethylene. Using acetone as the mobile phase, Funaki et al. (F59) concluded that the multimodal peaks seen in cellulose acetate were not aggregates but chains containing an anionic functionality introduced during acetylation. Acetone containing 0.05-0.2% phosphoric acid was reported by Ono (F60) for SEC of cellulose acetate. This author was also received a patent for the SEC of hydroxypropyl methyl cellulose phthalate using a chloroform-methanol mobile phase (F61). Hasegawa et al. (F62) used 5% LiCl in dimethylacetamide as the mobile phase for the successful SEC of cellulose and chitin, and Kennedy and co-workers (F63) also reported a detailed study of this eluent for underivatized cellulose. Schmidl et al. (F64) investigated the use of different eluents for high-temperature SEC of kraft lignins and concluded that DMF was an appropriate solvent. Eremeeva and Khinoverova (F65) studied the SEC elution behavior of 4-0-methylglucuronoxylan on Separon HEMA 1000 with DMSO and DMFA mobile phases. Polyelectrolyteeffects were suppressed by the addition of acetic acid and lithium bromide (0.03 M).

G. DETECTORS Light-ScatteringDetectors. Jenget al. (GI, G2) evaluated low-angle (LALLS) and multiangle laser light scattering (MALLS) photometers for SEC. They found that both instruments demonstrated equivalent precision and accuracy. The main source of inaccuracy was found to be thesensitivity differences between the light-scattering and the concentrationsensitive detectors. They also evaluated the use of different light-scattering equations for interpreting multiangle lightscattering data. Cotts and Siemens (G3) used SEC-LS to characterize random copolymers and discussed the applicability of the technique to copolymer characterization. Lederer and Hoellwarth (G4) used light scattering to extend the measurement range of SEC to higher molecular weights than Analytical Chemistty, Vol. 66,No. 12, June 15, 1994

59gR

those detected by the concentration detector. Claes and colleagues (G5) reported on the development of an on-line dynamic light-scattering photometer for LC applications. Roger and Colonna (G6) used SEC-MALLS to measure the molecular weight and radius of gyration distributions of amyloses. Using this method, they were able to detect large aggregates in the amylose solutions. SEC-MALLS was also used to characterize structural differences in hydroxyethyl starches (G7),the dimensions of polyimide precursors (G8), and branching in chitosans (G9). Zeng et al. (G10) used SEC-LALLS with both UV and differential refractive index detectors to measure the refractive index increment of poly(amino acids) as a function of molecular weight. The same instrument was also used to study alkaline phosphatase aggregation (G11). Sedlacek et al. (G12) used SEC-LALLS to characterize polymer degradation in poly(pheny1acetylene). Kat0 et al. (G13)used SEC-LS to study the molecular weight distributions of dextran-protein conjugates. Dollinger et a1 (G14a) demonstrated the use of an HPLC fluorometer as a right-angle light-scattering detector. They found that this detector was sufficient to determine molecular weight distributions for two example protein samples. This approach was also used by Rosenfeld (G14b) and Benedek (G14c) to study unfolding of recombinant human brain-derived neurotrophic factor. Jumel and colleagues (G15) reviewed the application of SEC-LS to the characterization of biopolymers, and Cotts (G16) reviewed its use in the characterization of polyimide precursors. SEC-LS was also used to characterize polyethylene (GI 7 ) ,poly(methylpheny1)silane (GI@, polyamic acid (G19), dextran (G20), polysaccharides (G21, G22), cereal P-glucans (G23), hydroxylated amylopectin (G24), carrageenans (G25), proteins (G26), and heparins (G27). Viscometers. A number of papers were published on data interpretation in SEC-viscometry. Balke and co-workers ( G 2 8 4 3 0 ) outlined a systematic approach to data analysis. They drew attention to the similar effect that band broadening and the interdetector volume have on calculated results and proposed that an effective interdetector volume could be used as an axial dispersion correction. Methods for correcting for the interdetector volume were also presented by Sagar et al. (G31)and Suddabyet al. ((33.2). Goldwasser (G33)presented a new method for calculating the number-average molecular weight of copolymers and polymer blends using SEC with a viscometer as the sole detector. The method gave good agreement with expected values for different polymer mixtures. The use of the intrinsic viscosity distribution for polymer characterization was demonstrated by Sanayei et al. (G34) using model molecular weight distribution functions. Guaita and Chiantore (G35) presented a simulation of SECviscometry which illustrated the importance of correctly determining sample concentration and interdetector volume. Mori (G36) developed a new design for a differential capillary viscometer for SEC, and Kraemer-Lucas et al. (G37) modified a Ubbelohde capillary viscometer for use as an SEC detector. Bartle and colleagues (G38) used an evaporative mass analyzer with a viscometer to characterize asphalt; the evaporative mass detector gave a more universal response than a UV detector. SEC-viscometry was also used in the analysis of branched polymer standards (G39), poly(4-methyl-lBOOR

Analyt/calChemistry, Vol. 66, No. 12, June 15, 1994

pentene) (G40), poly(amino acids) ( G I I ) , cellulose (G42), and carbohydrates (G43-G45). Combination Ligbt-Scattering/Viiity Detectors. Tripledetection systems, combining both a light-scattering detector and a viscometer with SEC and a concentration-sensitive detector have been used for the analysis of a wide range of complex polymers. Gore and Kilz (G46) demonstrated how such an instrument could be used to characterize block and random copolymers as well as comb- and star-branched polymers. Cheng (G47) used a triple-detector SEC to study the applicability of universal calibration to star-branched polystyrene, and Vilenchik and Ayotte (G48)studied random branching in EPDM rubber. Mourey and Balke (G49) extended their systematic approach to data interpretation to a triple-detector SEC instrument. Jackson and Yau (G50) developed a computer simulation of the triple-detector SEC experimental data. Triple-detector systems were used also to study polyolefins (G51), polystyrene (G52, G53), irradiated polymers (G54), and carbohydrates (G55). Conductivity. Rinaudo et al. (G56) used a conductivity detector in series with a differential refractometer to determine the charge distribution of carboxymethyl cellulose as a function of molecular weight. Ploeger (G57) also employed these two types of detectors to obtain the degree of esterification of pectin as a function of molecular weight. A patent was issued on the conductometric determination of phytic acid using either ion exclusion chromatography or SEC (G58). Electrochemical. Soucaze-Guillous et al. (G59)detected fullerenes from an SEC column using amperometry and fastscan-rate cyclic voltammetry. Evaporative Light Scattering. Lipids from edible oils and fats were analyzed by SEC with an evaporative light-scattering detector (G6O). Operating parameters of the detector were optimized to minimize differences in detector response among various types of lipids. These authors (G61) also used this procedure to study autoxidation of saturated triacylglycerols. Steryl glycosides in phospholipids were determined by SEC and an evaporative light-scattering detector (G62). Inductively Coupled Plasma Mass Spectrometry and Atomic Emission. Shum and Houk (G63) used a direct injection nebulizer with packed SEC microcolumns with on-line ICPMS for determining selenium in human plasma proteins. SEC/ ICP-MS was used in environmental toxicology studies involving metal-ligand interactions in cytoplasmic samples (G64). SEC with on-line ICP-MS was reported for the speciation of cadmium-binding metallothionein-like protein in a cyanobacterium (G65) and aluminum in tea infusions (G66). Reynolds and Biggs (G67) analyzed bitumens extracted from tar sands for iron using SEC/ICP-AE. Infrared Spectrometry. Markovich et al. (G68) coupled an FT-IR spectrometer with high-temperature SEC for the characterization of ethylene-based polyolefin copolymers for determining chemical composition as a function of molecular weight. Wheeler and Willis (G69)reported on a commercially available solvent-elimination SEC interface, capable of operating at 145 OC, for subsequent FT-IR analysis. Sample is deposited continuously onto a rotating germanium disk which is then scanned using an FT-IR spectrometer. Mass Spectrometry. Genuit and de Boer ((370) constructed and evaluated a dual-beam thermospray system for

use as an SEC detector. Li and co-workers (G71) used an on-line SEC-ion spray MS detector for determining solution dimerization of leucine zipper peptides. Prokai and Simonsick (G72) applied electrospray ionization MS for the analysis of an octylphenoxypoly(ethoxy)ethanol mixture using a T H F mobile phase containing a sodium salt for the formation of intact molecular ions through cationization. SEC with tandem MS was used to determine hydrophobic extractables in white waters from paper mills (G73). Several studies compared results obtained from MS to molecular weight distributions determined from SEC. Herod etal. (G74)reported that SECshowed the presenceofmaterial in tar fractions in excess of 10 000 g/mol, while FAB-MS indicated ions up to 4000 g/mol. Matrix-assisted laserdesorption ionization (MALDI) MS results of ferredoxin and myoglobin agreed with their expected molecular weights, as compared to poor agreement for ferredoxin and good agreement for myoglobin from SEC measurements (G75). With some limitations, good agreement was obtained between molecular weight results of oligomers derived from poly(3hydroxybutanoate) using MALDI and those results from SEC (G76). NMR. Ute and Hatada (G77) examined the feasibility of determining the molecular weight of poly(methy1 methacrylate), polymethacrylates, poly(viny1 alcohol), and a chloral oligomer using on-line SEC-NMR with a 500-MHz instrument as a detector. Raman Spectroscopy. Edwards et al. (G78)evaluated the use of an FT-Raman spectrometer as an SEC detector for measuring microstructural variations in polybutadiene. Detector sensitivity was assessed for a range of samples with different cis- 1,4, trans- 1,4, and vinyl- 1,2 contents and was found to lack sufficient sensitivity for these measurements. X-ray Fluorescence. Off-line coupling of a total reflectance X-ray fluorescence spectrometer to a Sephadex column was used for multielement speciation of elements in vegetables

(G79).

H. PACKINGS Inorganic-Based Packings. Kirkland ( H I ) described superficially porous Poroshell particles which consisted of a I-pm shell of approximately 300-Apores on a 7-pm-diameter silica core. This packing had the separation characteristics of 1-2-pm totally porous particles for macromolecules. Compared to 5-pm totally porous particles, it is expected that the larger Poroshell particles should result in less shear for large macromolecules. Cohen et al. (H2-H4) produced and evaluated graft-polymerized poly(vinylpyrro1idone)-silica packings for aqueous SEC and studied the permeability of this material. Frere and Gramain (H5) coated porous silica using in situ polymerization of poly(ethy1ene oxide) macromonomers followed by cross-linking and evaluated these columns with protein standards. The chromatographic properties of siliceous supports with thermally immobilized Carbowax 20M for SEC of biopolymers were reported by Choma and Dawidowicz (H6). Ahmed and Modrek (H7)reported on the preparation and evaluation of a hydrophilic-coated silica for SEC of proteins and peptides. A patent was awarded for a glucose-coated silica in which (3-aminopropy1)silica is treated with an aldose

or ketose mono- or disaccharide in the presence of sodium cyanoborohydride ( H 8 ) . This packing was applied to the separation of proteins. Komiya and Kat0 (H9)received a patent for an aqueous SEC packing produced by copolymerizing acrylamide with a vinyl-polymerizable silane bonded to silica particles. A 8- phenylethyltrichlorosilane-modified silica was prepared and evaluated for SEC of influenza virus particles (HIO). Vilenchik and co-workers (HI1) were issued a patent for an alumina SEC packing. Gewehr et al. ( H I 2 )coupled poly(N-isopropylacrylamide) with porous glass containing an amino group to produce an interesting temperature-responsive SEC packing. By the use of dextran as test solutes, elution times could be varied between 25 and 32 OC due to a change in the effective pore size of the packing brought about by the transition of the stationary phase from random coils to a globular conformation. Takagi (H13) presented a review on the development of hydroxylapatite as an SEC packing for protein separation. Polymeric-Based Packings. Ma et al. ( H I 4 modified cross-linked methyl methacrylate gels by reacting them with poly(4-vinylpyridine) microgels. Once reacted, the microgels were subsequently quaternized. The packing was evaluated for aqueous SEC using dextran and pullulan standards. Porous particles filled with polymer gels was described by Wulff et al. (H15). Smigol and Svec ( H I 6 )synthesized,and evaluated SEC packings based on hydrolyzed macroporous poly(glycidy1 methacrylate-methylene dimethacrylate). Hosoya et al. (H17)prepared and evaluated packings made by copolymerization of methyl methacrylate with a divinyl monomer. Juhl and Heitz (HI8) developed SEC packings based on 4-[2(heptafluoropropoxy)- 1,2,2-trifluoroethoxy) Jstyrene-divin ylbenzene which are compatible with fluorinated mobile phases, such as hexafluorobenzene, (trifluoromethyl)benzene, and 1,1,2-trichlorotrifluoroethane. Patents have been awarded for the preparation of porous poly(amino acid) particles (HI9, H20), cross-linked hydroxyethyl methacrylate packings (H21),and hydrophilic particles made from the copolymerization of 3-(allyloxy)- 1,2-propanediol and N,N’-methylenebisacrylamide(H22). Meehan et al. (H23) evaluated the use of PL aquagel-OH columns for the SEC of poly(viny1 alcohols). The pore size distribution of Separon HEMA 1000 was examined with mercury porosimetry by measuring the advancing and receding contact angles of mercury in pores (H24). The characteristics of Ashipak GS columns, composed of vinyl alcohol copolymer hydrophilic gels, with both aqueous and nonaqueous mobile phases, was described by Masaki et al. (H25). These packings were used for the separation of poly(viny1 acetate) and poly(vinyl alcohol) (H26). Van Dam and Daenens (H27) examined the suitability of Fractogel packings for SEC of poly(ethy1ene glycols) and compared these results to data obtained from Superdex columns. Glucomannan hard gel packing, which was 2-diethylaminoethylated, was evaluated for the preparation of bloodcoagulation factor VI11 ( H 2 8 ) . A series of patents was awarded for the preparation of cross-linked cellulose SEC packings (H29-H32). Motozato et al. (H33) prepared gel particles from luffa, a vegetable material from a tropical vine, and used it in an aqueous SEC system. A vesicular packing Analytical Chemistry, Vol. 66, No. 12, June 15, 1994

601R

material, consisting of microcapsules from plant cells, which act as negatively charged ultrafiltration membranes, was prepared and evaluated for the separation of nucleic acids (H34, H35)and proteins (H36). Seliskoet al. (H37)compared SEC analysis and purification of monomethoxypoly(eth1ene glycol) using vesicle chromatography and Superose 12 and concluded that SEC analysis with Superose packing gave superior results.

I. COMPOSITIONAL HETEROGENEITY The first part of this section deals with the use of SEC with selective detectors to determine the chemical compositional heterogeneity or compositional drift of polymeric materials as a function of molecular weight. The second section covers the application of interactive chromatography for measuring compositional heterogeneity. In this approach, molecular weight fractions can first be prepared by SEC and then characterized by interactive HPLC or fractions can first be prepared from interactive HPLC and then characterized by SEC. The third section reviews studies involving temperature rising elution fractionation (TREF), in which the separation is based on solubility differences as a function of temperature. This technique is useful for separating polymers on the basis of crystallinity, short-chain branching, tacticity, and/or chemical differences. The last section deals with orthogonal SEC, in which the nature of the mobile phase is changed to effect polymer conformation or size. The reader should also consult section G on detectors and section M on coupled columns and column switchingfor related studies on determining chemical compositional distribution of polymers. SEC with Selective Detectors. SEC with UV detection was used for the characterizationof blockcopolymers of styrene and methyl methacrylate (ZI), random copolymers of styrene and isobutylene and triblcckcopolymers of polyacenaphthylene outer segments and polyisobutylene midsegment (12), functionalized groups in styrenic polymers (13), carbohydrate portion of glycoproteins (14),and ethanol fermentation broths (15). To determine hydroxyl-terminated polybutadiene, the hydroxyl group was converted to a UV-absorbing urethane derivativeand determined with a UV detector and a differential refractometer (16). Trathnigg and Yan (17)analyzed triblock copolymers of styrene and butadiene and mixtures of the corresponding homopolymers using SEC with an on-line density detector and refractometer. FT-IR was used off-line with SEC for the characterization of end-capped polyesters (18) and hydrocarbon oils (19). Interactive HPLC. Gloeckner (110,111)reviewed the use SEC and gradient elution HPLC for the chromatographic cross-fractionation of copolymers in terms of chemical composition. Gloeckner and Wolf (112)described the analysis of block copolymers of styrene and tert-butyl methacrylate by methanol-THF gradients on reversed-phase columns. Gloeckner (113) also studied adsorption and solubility control in gradient HPLC of styrene and methacrylate copolymers. Mori (114, 115) described an HPLC method for identifying polymer types based on either hydrogen bonding between the polymer and silica packing or precipitation of the polymer in the column. Thus, some classes of polymers were retained 602R

9

Ana/ytcal Chemistry, Vol. 66, No. 12, June 15, 1994

and others were not eluted, depending on the separation mechanism. Jansen et al. ( 1 1 6 ~fractionated ) polymer blends in a precipitation-redissolution process. Following precipitation in a nonsolvent, fractionation was achieved according to solubility by multiple solvent gradient elution. Kilz (116b) reviewed the use of two-dimensional chromatography, gradient HPLC coupled to SEC, for the characterization of complex polymers. An example using butadienestyrene block copolymer was given. Shalliker et al. (117) compared SEC and gradient elution HPLC for determining the molecular distribution of polystyrenes. Both methods gave similar results, but the reversedphase method was less concentration sensitiveand gave higher resolution. Northrop and co-workers (118) studied isocratic retention of polystyrenes on a bimodal pore diameter, reversedphase column. Ogino and colleagues (119) investigated the separation of styrenemethacrylate copolymers by composition using normal and reversed-phase HPLC. Teramachi et al. (120) determined the chemical composition distribution of poly(methy1 methacrylate)-graft-polystyrene by adsorption HPLC. Elomaa and Lehtinen (121)fractionatedpoly(styreneco-methyl methacrylate) by preparative, nonaqueous reversedphase LC. Reversed-phase HPLC was used to characterize the statistical composition distribution of 4-vinylpyridine-Nvinylpyrrolidone copolymers (122) and Novolac resins (123). HPLC was used to determine the composition of oligomeric poly(m-phenyleneisophthalamide) (124), poly(ethy1ene glycols) (125),nonionic surfactants of polyethoxylated octylphenol , alkyl phenol and ethylene oxide oligomers ( 1 2 ~ 9ethoxylated surfactants (127),ethoxylated nonylphenol nonionic surfactants (128),hydroxy-terminated polysulfone oligomers (129), and trifunctional methyl phenyl siloxane oligomers (130). Rissler et al. (131) separated hydroxy-terminated polyethers [poly(ethylene glycol), poly(propy1ene glycol), and poly(butylene glycol)] using reversed-phase HPLC and an evaporative light-scattering detector; the corresponding 3 3 dinitrobenzoyl derivatives were monitored by UV detection. Liautard (132) studied the separation of high molecular weight nucleic acids by reversed-phase HPLC and proposed a multiple-point interaction model based on chain length, base composition, and secondary structure. Maltooligosaccharides were separated by hyrophilic interaction chromatography on aqueous SEC columns using gradient elution with pulsed amperometricdetection (133). Fractionation of y-carrageenan was carried out by stepwise elution with decreasing concentrations of ammonium sulfate on a Sepharose CL-4B column (134). Using isocraticreversed-phaseHPLC at critical conditions, Trathnigg (135, 136) developed a procedure for analyzing poly(oxya1kenes) with on-line density and refractive index detectors. In a series of papers, Pasch and colleagues investigated the use of HPLC at critical conditions for the characterizationof poly(ethy1ene oxide-block-propyleneoxide) (137), polymer blends (138), poly(styrene-block-methyl methacrylate) (139),and block copolymers of decyl and methyl methacrylate (140). Hunkeler and co-workers (141) studied the solvent composition necessary for critical conditions for LC of polystyrene and poly(methy1 methacrylate) using silica packing. The ratios of a nonpolar thermodynamically good

solvent (toluene) and a polar nonsolvent (methanol) were systematically varied. Zamina et al. (142)separated methyl methacrylate-styrene block copolymers and these polymers copolymerized with tertbutyl methacrylate on wide-pore silica at critical conditions using binary and ternary mixtures of different solvents. This group (143) described retention characteristics of diblock copolymers of polystyrene and poly(methy1methacrylate) and mixtures of related homopolymers using the SEC mode of separation, critical conditions for one of the blocks, and critical conditions for the entire block copolymer. Prudskova et al. (144) proposed a general approach for LC analysis of linear and cyclic polymers using critical chromatography. Belenkii et al. (145) used TLC at critical conditions to separate a diblock copolymer consisting of polystyrene and poly(tert-butyl methacrylate). Temperature Rising Elution Fractionation. Wild (146) presented a review of TREF for the characterization of polyolefins based on crystallinity. Karbashewski et al. (147) reported that TREF calibration curves must take into account sequence distribution to adequately predict crystallinity and to relate '3C-NMR data on branch content to short-chain branching distribution of linear low-density ethylene-1 -octene copolymers. Bonner and co-workers (148) used linear paraffins and low molecular weight polyethylene as TREF standards to determine the length of the methylene sequences between branches. Housaka (149) described and evaluated a TREF method for the analysis of polyethylenes and found that the short-chain distribution of linear low-density polyethylene did not depend on molecular weight. Karoglanian and Harrison (150) compared TREF data with results obtained from DSC of solution- and meltcrystallized ultra-low-density polymers of ethylene and 1-octene. Although there appeared to be a reasonable shape correlation between the compositional distribution curves of the two techniques, there were significant differences. Preparative TREF was used to fractionate ethylene-1-butene copolymers,and fractions were subsequently analyzed by DSC, FT-IR, and SEC (151). Usami and colleagues (152) used TREF-SEC, 13C-NMR,and transmission electron microscopy to characterize impact-resistant poly(propylene-ethy1ene) copolymers. Orthogonal Chromatography. Using techniques related to orthoganal chromatography with a single SEC column, Cheng and Zhao (153)determined thecompositiondistribution of a siloxane-styrene graft copolymer and a butyl acrylatesiloxane-styrene graft copolymer with the use of different solvents. J. PHYSICOCHEMICAL STUD I ES Synthetic Polymers. Branching/Mark-Houwink CoefficientslConformation. Rudin (JI) reviewed thecurrent state of SEC techniques for measuring the variation of long-chain branching. Bauer and Burchard (J2) used a combination of a light-scattering detector and a viscometer to study the scaling behavior of the molecular weight of branched polycyanurates. They found good agreement with the predictions of threedimensional percolation theory. Pang and Rudin (J3)'studied long-chain branching frequencies in polyethylene and found reasonable agreement between estimates of long-chain branch-

ing based on the Zimm-Stockmayer relation and NMR results. A simulation of the size distribution of branched polymers measured by SEC was developed by Kidera and Kohjiya (J4). SEC was also used in the characterization of branching and network formation in polychloroprene (J.5, J6),poly(viny1 chloride) (J7),polystyrene (J8),and polyurethane (J9, JIO). Hennessey et al. ( J I l ) used on-line light scattering and viscometry to determine the Mark-Houwink coefficients of polyribosyl ribitol phosphate. They found values for the exponent a close to 1.5 in two buffered mobile phases, indicating a rodlike conformation. SEC-viscometry was used to determine the Mark-Houwink coefficients of hydrolyzed poly(viny1 alcohol) (JI2). Mark-Houwink coefficients were determined for acrylic copolymers (J13), poly(1actic acid) (J14), polybutadiene (J15), liquid crystalline copolyester (J16),and polysaccharides (JI7, J I 8 ) . Hu and Song (JI9, J20) studied concentration effects in SEC and developed methods for determining the second virial coefficient from the concentration dependence of elution volume of a polymer. They also showed how to calculate the radius of gyration from SEC and viscometry data. Results were in good agreement with those obtained by light scattering. Magiera and Krull (J2I) used SEC-LS to examine the formation of aggregates of alkaline phosphatase in different buffer systems. Theeffects of buffer conditions, concentration, and various batch preparations were studied. SEC was used by Boochathum et al. (522) to measure the stem length and stem length distribution in trans-1,C polyisoprene crystals. The isoprene units at the crystal surface were selectively degraded by ozonolysis and then the crystals were dissolved and the degraded molecular weight distribution was determined by SEC. Association/Complexation/Solvation Studies. During this review period, there has been increased activity regarding the study of micelle formation using SEC. Funasaki (J23) reviewed theuseof SEC to study self-association of surfactants and related compounds, including the estimation of hydrodynamic radius and determination of sphere-rod transition of surfactant micelles. Funasaki and co-workers (524) presented retention mechanisms of self-associating heterocyclic compounds and surfactants on Sephadex gels. This group (J25) also used frontal SEC to study micelle formation of hepta(ethy1ene glycol) decyl ether. Booth and co-workers (J26) introduced a new method for studying micellization in which the block copolymer solution is used as the eluent, and the equilibrium behavior of the eluent is probed by injecting an additional small volume of copolymer into the system. This approach was used to study the micellization behavior of ethylene oxide-propylene oxide triblockcopolymers (J26,J27). This group (J28)also studied micellization of diblock ethylene oxide-butylene oxide copolymers. D'Oliveira et al. (J29) examined the interaction between polystyrene-poly(ethy1ene oxide) diblock micelles and polystyrene latex particles in water using dynamic light scattering and SEC. Pu et al. (J30) investigated the micelleforming behavior of poly(styrene-isoprene) diblock copolymers in selective solvents and different temperatures with SEC. The aggregation number and aggregation energy of micelles were determined. Analytical Chemistty, Vol. 66,No. 12, June 15, 1994

605R

SEC was used to analyze micelle formation of a drugblock copolymer conjugate [Adriamycin-conjugated poly(ethylene glycol)-poly(aspartic acid) block copolymer] (531). Stone et al. (J32) separated bile vesicles and micelles by SEC in which the eluent contained a bile salt. Self-association of methylene blue, a cholic acid derivative, chlorpromizine hydrochloride, and octaethylene glycol decyl ether was investigated with frontal SEC (J33). SEC was used to study complex formation between Fe3+ and carboxymethyl-cellulose (534) and a three-component complex of quercetin, pectin, and glucose (J35). Cheng and Zhao (536) investigated the self-association and adduct formation of transition metal chelates of acetylacetone. Preferential solvation parameters for the ternary system poly(dimethylsiloxane)-benzene-methanol were determined by comparing areas of the solvated and nonsolvated SEC peaks (537). Kinetic Studies. SEC was used to study the kinetics of cross-linking of a brominated epoxy resin with dicyandiamide using benzyl-N,N-dimethylaminecatalyst as functions of reaction temperatures and concentrations cross-linking agent and catalyst (538). Photochemical-induced radical reactions of hyaluronic acid in aqueous solutions were investigated using SEC and EPR spin trapping (J39). Badamshina et al. (J40) described the use of SEC to determine the kinetics of bulk copolycyclotrimerization of 1,6-hexamethylenediisocyanate and 1-chlorohexamethylene-6-isocyanate. The pulsed-laser technique, in combination with SEC, was used to study the temperature dependence of the propagation rate coefficient for radical polymerization of tert-butyl methacrylate at low conversion (J41). Inverse SEC. Inverse SEC is a technique in which the material being characterized is packed into a column, and solutes of different size are injected to probe the pore size and specific surface area of the packing. Jerabek et al. (J42) reviewed the use of SEC for pore structure characterization of organic and inorganic materials, recommended guidelines for these studies, and discussed data treatment. Inverse SEC was used to determine the pore size distribution of a macroporous poly(styrene-co-divinylbenzene) rod (J43), microporous isoporousdivinylbenzene-styrene copolymers (J44), mesoporous carbon adsorbents (J45,546),dietary fiber (J47), cellulose fibers (J48),calcium alginate beads (J49),and even anaerobic sludge granules (J50). Jerabek et al. (J51)compared a fluorescent probe method with inverse SEC for characterizing the gel phase in macroporous network polymers. Peter and co-workers (J52) determined the pore volume distribution curve for precoated silica gel TLC plates using polystyrene standards. The morphological structure of several chromatographicpackings was evaluated using the inverse SEC approach (J53). Nakazawa and Imae (J54)characterizedvesicles formed from alkenesuccinic acids by measuring the retention of an anionic dye in the inner aqueous phase of the vesicle. Biopolymers. Structure/Conformation Studies. Estimation of the molecular mass and/or size of native proteins by aqueous SEC has become commonplace because of the rapid and convenient determinations that can be conducted using high-resolution HPLC columns. A number of examples of the estimation of protein native molecular mass is included 604R

Analytical Chemistty, Vol. 66, No. 12, June 15, 1994

in the section on Selected Applications and also appears in Table 1. Examination of these studies illustrates the vast range of experimental conditions which may be used for SEC analyses. Analysis of the subunit structure of native proteins and enzymes can be accomplished by SEC under native conditions, followed by comparison of molecular mass using denaturing conditions. The determination of dissociated (denatured) subunit molecular mass can be achieved by SEC, for example, with a mobile phase containing 6-8 M guanidine hydrochloride (GuHCl) or urea or 0.05-O.2% sodium dodecyl sulfate (SDS) or by using SDS-polyacrylamide gel electrophoresis (SDSPAGE) or mass spectral analysis of selected peaks. Examples of the determination of protein subunit structure by this approach include the following: human aminopeptidase P (J55),a single-chain Fv antibody fragment (J56),chorismate mutase of Bacillus subtilis (J57), a mixture of peptide conjugates to Fab fragments of monoclonal antibodies (J58), ethanol dehydrogenase of Salmonella typhimurium (J59), the ribonuclease reductase small subunit of vaccinia virus (J60), and the aggregational state of synthetic A@-peptides (J61). Analysis of monoclonal IgG F(ab’)z fragments and a thioether cross-linked bispecific (hybrid) F(ab’)* fragment, before and after reduction of disulfide bridges, permitted proof of the desired polypeptide structures (J62). Matsushita et al. (563) purified two forms of methanol dehydrogenase (MD-I and MD-11) from Acetobacter methanolicus. Analysis of MD-I and MD-I1 subunits showed good agreement between estimates of molecular weights by SDS-PAGE and SEC using an SDS-containing mobile phase, but presumptive Coloumbic interactions of the native enzymes with the column packing prevented accurate estimation of the aggregate molecular mass of the enzymes. Accurate assessment of native molecular weight, and thus subunit composition, required the use of SEC with on-line LALLS. Representative examples of preformulation studies of a recombinant protein were described by Hageman et al. (J64, J65), employing SEC in the presence of SDS to analyze the formation of covalently linked dimers and oligomers of recombinant somatotropins, which were formed during storage of the protein under a variety of conditions. A number of studies employed SEC determination of Stokes radius as a hydrodynamic measurement of particle asymmetry and to assess the effects of experimental manipulations on protein conformation. Examples of the hydrodynamic characterization of proteins, combining Stokes radius (as determined by SEC) with the sedimentation coefficient, include the following: the extracellular domains of the macrophage scavenger receptors (J66), propylamine transferase from a thermophilic archebacterium (J67), mouse CAMPphosphodiesterase (J68), and a core complex (glutamylprolyl-) aminoacyl-tRNA synthetase (J69). Hackney et al. (J70) combined SEC and sedimentation analyses for the investigation of the ionic strength- and pH-dependent conformational transition of kinesin from a compact (low ionic strength, globular) form to an extended conformation. Sato et al. (J71) used SEC-LALLS to establish that the anion-induced decrease in SEC retention volume of mitochondrial electron-transferring protein was due to an increased molecular volume, rather than resulting from a shift in the aggregational state of the

+

Table 1. SEC Applications of Biochemical Interest'

Target

Matrix

amyloid @Acore peptides amyloid proteins opioid peptides calmodulin-melittin complex salmon calcitonin brain-derived neurotrophic factors human macrophage CSF bovine somatotropin retinoic acid-binding proteins retinol-binding protein insulin-like growth factors cytokinin-binding protein MDGI-related polypeptide HIV-1 reverse transcriptase chicken liver pyruvate carboxylase a-amylase inhibitor human serum biotinase human aggrecan proline-rich mycobacterial protein immunoglobins IgM, IgG, IgA immunoconjugates cystic fibrosis TR fusion proteins 8-( 1-4)-galactosyltransferase torpedo VAT-1 avian zyxin myofibrillar proteins rat albumin tetanus toxoids cholera enterotoxin buffalo casein wheat flour proteins glandless cottonseed proteins

crude extract, purified synthetic peptides crude tissue extracts bovine hemoglobin peptic hydrolysate binding characterization of pure components aerosol formulations purified recombinant polypeptides purified recombinant polypeptide purified recombinant polypeptide rat epididymal cytosol extract human small intestine extract human serum extract of mung bean seedlings mammary gland extracts purified mutant recombinant polypeptides purified protein barley kernal isolate serum isolate aggrecan hydrolysis by stromelysin- 1 crude extract of culture medium milk, commercial prepns, ascites purified IgGs, various derivatives purified recombinant polypeptides detergent-solubilized golgi membranes solubilized cholinergic synaptic vesicles purified yzxin mixed with structural proteins soluble muscle extract rat urine commercial preparations Vibrio cholera culture filtrate isolate isolated caseins sonicated extract of Australian cultivars extracted cottonseed fractions

Purpose

refs

Proteins/Peptides/Conjugates 0203 0204-5 0206 0207 0208 0209-12 0213 0214 0215 0216 0217 0218 0219 0220 0221 0222 0223 0224 0225 0226-9 023C-2 0233 0234 0235 0236 0237 0238 0239 0240 0241 0242 0243

Particles/ Assemblies buffalo semen liposomes a

1, A review of SEC methods for size distribution

P

0244 0245

Legend: I, isolation or purification; A, assay or activity determination; P, physical characterization.

protein. Heyduk et al. (J72) studied the effects of cAMP binding on the dimer-monomer equilibrium and conformation of N-terminal fragments of a bacterial cAMP receptor protein. SEC analysis was also applied to the investigation of environmental effects on the conformation of disulfide-reduced serum albumin ( 5 7 4 , and a variety of N-terminal fragments and derivatives of serum albumin ( J 7 4 ) . AssociationslProtein Folding. During this review period, the largest growth in biochemical SEC analyses has been in characterizing protein and peptide folding, associations, and aggregation. Frequently, these topics are interrelated, especially in cases where the studies address folding patterns of multisubunit proteins or where monomeric proteins exhibit a folding intermediate which passes through a self-associable state. The subclassification of such studies, for this review, is somewhat arbitrary. One of the reasons for growth in this area has been the availability of proteins expressed by recombinant DNA techniques. Biophysical characterization of proteins still requires a relatively large quantity of protein, relative to the quantities that could previously be obtained from biological sources, particularly for the case of polypeptides produced by highly specialized animal cell types or tissues. Thus, the production of low-abundance proteins by recombinant DNA techniques has permitted protein structure/ function studies that were not before practical. Endo and colleagues (J75) studied the heat-induced aggregation of intact and deglycosylated recombinant eryth-

ropoietins using SEC-LALLS. The propensity to form covalently cross-linked erythropoietins varied with pH, and the molecular weight distribution of oligomers varied with salt concentration. Deglycosylation of erythropoietin greatly increased the rate of heat-induced aggregate formation. Comparison of human and recombinant (nonglycosylated) a 1-antitrypsin aggregational properties revealed an increased rate of aggregate formation for the recombinant protein; the two proteins were otherwise highly similar (J76). Conversely, deglycosylated human erythrocyte band 3 protein showed a minor increase in its tendency to aggregate, relative to the normally glycosylated protein (J77). The presence of polysaccharides on proteins does not necessarily reduce aggregate formation. For example, analyses of dextran-ovalbumin and dextran-lysozyme conjugates were conducted by SEC-DRILALLS to characterized the degree of polysaccharide substitution in the conjugates and to demonstrate the ionic strength-dependent formation of aggregates (578);the parent proteins do not exhibit appreciable tendency to aggregate under the same conditions. A series of studies described the folding, self-association, and aggregation of human insulin and compared a variety of semisynthetic and recombinant insulin mutants ( J 7 9 4 8 1 ) . Amino acid substitutions of critical residues identified a highaffinity interaction site for dimer formation, which appears to be requisite for the formation of higher molecular weight aggregates. Analytical Chemistry, Vol. 66, No. 12, June 15, 1994

605R

The CAMP-dependent protein kinase is a remarkable enzyme, insofar as its catalytic activity is controlled by a readily reversible associationdissociation reaction. Beuchler et al. (J82) produced mutant polypeptides, introducing amino acid substitutions in the autoinhibitor site of the regulatory subunit (R) of the enzyme, and then used an SEC method to determine the apparent binding affinity of the R2 dimer for the catalytic subunit dimer (C2). The dissociation constants for the RzC2 holoenzyme were much lower for the wild-type holoenzyme than for the mutant, both in the presence and in the absence of MgATP, which is an accessory modulator of the holoenzyme. The study by Johnson and colleagues (J83) used SEC analysis to prove that chemically modified (fluophore-labeled) Rand C subunits were able to associate to form an appropriate CAMP-dissociable holoenzyme complex. A method for the analysis of collagen monomers and oligomers, their incorporation into collagen fibrils, and the effects of solution treatment on the oligomerization rate of collagen monomers was the subject of a study by Condell et al. (J84).Other SEC studies of protein aggregation, involving proteins of probable commercial interest, included the following: the effects of heating on whey protein aggregation (585),freezing- and low-temperature-storage induction of aggregates of carbonic anhydrase (J86), and an analysis of the mass distribution of covalently cross-linked and saltdissociable hemoglobin aggregates in glutaraldehyde-treated hemoglobin [polyhemoglobin (J87)].An interesting study by Litzen et al. (J88) compared the favorable capabilities of asymmetrical flow-field flow fractionation to SEC for the analysis of monoclonal IgG antibodies and aggregates thereof. The aggregation properties of bovine lens crystallins and component subunits were investigated by SEC. A method for the fractionation of a crude protein mixture for isolation of the individual a-, @-, and y-crystallins was described by Vlaanderen et al. (589).These authors observed the apparent oligomerization state of the subtypes of the @-crystallinsto be highly concentration dependent, suggesting reasonably rapid dissociation kinetics, and calling into question previous conclusions regarding the native state of the protein. An analysis of the reaggregation of urea denatured a-crystallin subunits (aA and BA) and small heat-shock protein (HSP25) subunits was carried out by Merck et al. (J90). The close structural and functional relationships between these proteins were demonstrated by the formation of functional hybrids of crystallin subunits with HSP25 subunits, as determined by SEC and various biochemical assays. The assembly of several types of cytoskeletal polypeptides underwent analysis by SEC. The kinetics and thermodynamics of dimer and tetramer formation of a-and @-subunitsfrom human red cell spectrin were investigated by DeSilva et al. (J91).Related studies using truncated recombinant spectrin a-subunits permitted mapping of binding sites which lead to the dimeric and tetramericforms of spectrin (592). The effects of pressure on the dissociation of polymerized G-actin were studied by Garcia et al. (J93). Features of the association of bovine brain 7 proteins, varying in the degree of phosphorylation, were investigated by Garcia de Ancos and colleagues (594). In this study, the authors point out that absorbance monitoring of column effluent greatly overestimates the amount of high molecular weight aggregates, due to their 606R

Analytical Chemistty, Vol. 66, No. 12, June 15, 1994

light-scattering properties. The formation of homo- and heterodimers of desmin and vimentin, in various chemically modified forms, was the subject of a study by Traub et al. (J95). An increasingly popular approach to the study of protein domain interactions employs analysis of the association of polypeptide fragments, produced by fragmentation of intact proteins or by assembly of preformed protein fragments. A number of studies have employed limited proteolytic cleavage of monomeric and multisubunit proteins, followed by SEC determination of the interactions between polypeptide fragments. This approach was used to demonstrate the stable association of the two tryptic fragments of prolyl oligopeptidase under native conditions and their dissociation in a ureacontaining mobile phase (396). Analysis of proteolytically treated yeast phosphofructokinase (a multisubunit enzyme) by SEC and SDS-PAGE permitted analysis of the internal cleavage sites and measurement of the effects of the proteases on the quaternary structure of the enzyme (J97).Wyss et al. (J98) demonstrated stable association of the proteolytically generated polypeptide fragments of mitochondrial creatine kinase. The effects of various environmental conditions on the association of the dimeric protomer, and its ability to form the octameric form of the enzyme, were studied. Recombinant expression of the pepsin polypeptide chain as separate continguous fragments was described by Lin et al. (J99). SEC was employed to characterize the refolded heterodimeric pepsin and to evaluate refolding and association conditions which reduced aggregative loss of the polypeptide fragments. Rather than express and purify a series of mutant polypeptides, Edgerton and Jones (J100)utilized SEC analysis of the transcription/translation products corresponding to recombinant phytochrome polypeptides. By this novel approach, the amino acid sequences involved with subunit associations of the homodimeric protein were identified. Expression of a recombinant 5s subunit of bacterial transcarboxylase resulted in the appropriate dimeric protomer, which was observed by Shenoy et al. (JIOI) to be incapable of reconstituting the complete 26s enzyme when added to additional appropriate subunits of the enzyme. Reconstitution experiments, using the 5 s protomers with SEC-separated components of the disassembled native enzyme, revealed a low molecular weight component of the native enzyme (named the assembly-promoting factor) which was not previously known. Analysis of the reconstitution mixtures by SEC determined the minimum subunit stoichiometry of the catalytically active enzyme complex. Synthetic peptides are being studied as model systems to characterize protein associations and folding. Monera et al. (J102)prepared several 35-residue synthetic peptides, which were disulfide cross-linked, to form dimeric antiparallel and parallel cy-helical coiled coils. SEC demonstrated authentic dimeric structure and was employed for characterization of the aggregation state of the antiparallel coiled coil. Similar measurements were employed by Graddis et al. (5103)for the analysis of associations between their heptad-repeat synthetic a-helical coiled-coil polypeptides. Narrow-bore SEC with on-line ion spray mass spectrometry was used to characterize the formation of noncovalent dimers of two 33-residue leucine zipper peptides (J104).Interestingly, the associations between

the leucine zipper peptide were observed to be reasonably stable in the gas phase, where the hydrophobic interaction driving the association would be expected to be very small. Several studies were reported using SEC for the investigation of associations of membrane proteins or soluble fragments of membrane receptors. The concentration-dependent association of cloned wild type and several mutant, cytoplasmic fragments of the Escherichia coli aspartate receptor were characterized by Long and Weis (5105) using SEC and light scattering . Titration of the ligand-induced dimerization of a recombinant extracellular domain of the bovine growth hormone receptor was used to determine the stoichiometry of complex formation for receptor binding to bovine growth hormone and bovine placental lactogen (5106). Examples of solubilized membrane protein studies included determination of the monomers and dimers of rat intestinal butyrylcholinesterase (5107) and investigation of the dissociation and reconstitution of the light-harvesting complex of Rhodospirillium rubrum (5108). In bacteria, translocation of newly synthesized precursor polypeptides to the extracellular environment requires a number of cytosolic factors. The interactions of one such factor, SecB, with a precursor protein, prePhoE, was investigated by Breukink et al. (5109). SEC analysis of ureaunfolded prePhoE demonstrated rapid and random aggregation of the polypeptide upon dilution to a low concentration of denaturant. Addition of SecB to the renaturation mixture prevented aggregation of prePhoE by the formation of a prePhoE-SecB complex, demonstrating a chaperonin-like activity for SecB. Current ,thoughts on protein-folding pathways recognize that a variety of schemes may apply to describe the transition of unfolded polypeptide (U) to the native folded structure (N). In recent years, evidence has accumulated for the existence of folding intermediates, which may appear under certain environmental conditions; the best known of these is the molten globule (MG) state, which is also variously known as the “compact intermediate”, or “collapsed form”. Other intermediate states (I) are known toexist. A number of model schemes have been described in recent studies using SEC for the analysis of protein folding, including N-U

- -

(1)

NQMG-U

N

MG

I,

(2) Q

U

(3)

where I, respresents intermediate state(s), x = 0 ... n. Associations or aggregates may also form, complicating these reactions, generically;

where subscript a represents associated species. Similarly, multisubunit proteins may exhibit much more complex folding equilibria, for example, with a homodimer, the simplest scheme would be N,

Q

21,

w

2MG

--

2IY 2U

(5)

where subscripts x and y = O... n, and the intermediates I, and I,, are different. The folding equilibria for multisubunit proteins may also be greatly complicated by association reactions involving intermediate species. Experimentally, protein folding patterns are frequently investigated by physiochemical analysis of equilibrium intermediates formed by varying the concentration of denaturing agents, such as urea or GuHCl, by altering the pH, or by elevated temperature. Many previous studies have proven the utility of SEC for analyzing the two-state transition of monomeric proteins, as represented by eq 1. Studies by Uversky (5110)and colleagues (5111) have compared the features of SEC monitoring of both two-state denaturant-dependent equilibrium unfolding (exemplified by myoglobin and lysozyme) and those which unfold via the molten globule state, as described by eq 2 (e.g., bovine and human a-lactalbumins, bovine carbonic anhydrase, and @-lactamasefrom Staphylococcus aureus). These studies demonstrate the wealth of information that can be obtained by SEC measurements and place such measurements in context with the information that can be obtained by near- and farUV circular dichroic spectroscopy, the ANS fluorescence binding assay, and enzyme activity measurements. In general, these studies also confirm the similarity of the hydrodynamic volumeof the molten globule state to the native state; Uversky (5110) estimated an approximate 10-20% increase for the Stokes radius for MG, which compares well with results using other analytical methods. The unfolding of monomeric acidic fibroblast growth factor by lithium perchlorate was investigated by Mach et al. (.TI 12). This protein appears to exhibit a number of partially structured intermediates, similar to the situation described by eq 3. The combined results of spectroscopic and SEC experiments suggest that at least one intermediate, and possibly several related intermediates, exhibit properties of a molten globule. Further investigation of the possibly several molten globule intermediates would be of considerable interest. DeFelippis and colleagues (5113) reexamined the folding pathway of recombinant human growth hormone, which had previously been considered to fold by a simple two-state reaction. Analysis of folding using higher concentrations of protein revealed a self-associating intermediate, which can lead to formation of insoluble aggregates, essentially like the scheme described by eq 4. Associated forms of the intermediate could be seen as high molecular weight species in SEC. It was observed also that the formation of aggregates, and subsequent precipitate formation, could be greatly reduced by the addition of peptide fragments from the third helical segment of growth hormone. An interesting parallel was discussed relating the aggregation-protective effects of such peptide binding to the ability of the chaperonins to prevent aggregation and precipitative loss of proteins in vivo. Herrold and Leistler (5114) continued their analysis of the folding pathway of the dimeric aspartate aminotransferase of E . coli. This apoenzyme binds and utilizes pyridoxal phosphate as a coenzyme, upon which the native fluorescence of the apoenzyme is quenched. SEC was employed to monitor the equilibrium unfolding of the native enzyme by guanidine hydrochloride, using on-line fluorescence monitoring for detection of pyridoxal phosphate binding. In the process of unfolding, the enzyme first exhibits dissociation of the dimer Analytical Chem/stty, Vol. 66, No. 12, June 15, 1994

007R

to monomers, which retain both a small hydrodynamic volume and bound pyridoxal phosphate, revealing a highly structured monomeric intermediate. At higher guanidine concentrations, binding of pyridoxal phosphate is lost (as revealed by the loss of fluorescence quenching), but the small hydrodynamic volume is retained, consistent with the formation of the MG state. Subsequently, the MG state unfolds as revealed by a marked decrease in SEC retention. The folding pathway is thus described by eq 5 above, in which the I, state is a highly structured monomer intermediate. SEC investigation of the unfolding of the dimeric human glutathione transferase (GST, Pi class) was reported by Aceto et al. (5115) to follow a pattern of dissociation, formation of a pact monomer, and then transition to unfolded chains, essentially as indicated by scheme 5. These investigators extended their analysis to GST B1-1 from Proteus mirabilis to compare the folding pattern with that of the mammalian enzyme (JI 16). Unlike the mammalian enzyme, dimeric native GST B1-1 first formed an inactive dimer (in a fully reversible fashion), which then dissociated to an MG-like monomeric intermediate before assuming the chain-unfolded state. Thus, significant differences were found in the folding patterns between these structurally and catalytically similar enzymes. Analysis of protein-folding intermediates is frequently made difficult by their low concentration at equilibrium, under any particular environmental condition. This condition results from the necessarily high energy difference that is required for a stable native protein. This biological necessity complicates the direct characterization of folding intermediates in many cases. Sanz and Fersht (J117)describe the use of site-directed mutagenesis to shift the folding equilibria to favor the accumulation of folding intermediates, essentially by introducing mutations which destabilize the folded (native) proteins. With Bacillus amylolique-faciens ribonuclease (barnase) as a model system, several mutant proteins were expressed which exhibited decreased stability and altered folding equilibria in urea. This effect was demonstrated, in part, by urea unfolding of the wild-type protein, which showed coincident two-state tertiary structural changes (monitored by intrinsic fluorescence) and hydrodynamic changes (monitored by SEC). In comparison, a mutant barnase exhibited a noncoincident, probably three-state (Le., eq 2 above) tertiary and hydrodynamic unfolding pattern. The unfolding characteristics of the mutant(s) were interpreted to demonstrate a greater contribution of the MG state to the observed physiochemical behavior of the equilibrium mixture. The unfolding of both mutant and wild-type nucleotide diphosphate kinase (NDP kinase) from Dictyostelium discoideum was studied by Lascu et al. (J118). SEC analysis of the wild-type hexameric enzyme exhibits a single transition from hexamer directly to unfolded chain; in contrast, the less stable PlOOG mutant exhibits both the disaggregation and a discrete compact monomeric unfolding intermediate. Mechanistically, the unfolding scheme is similar to that in eq 5. In analogy with the site-directed mutagenesis experiments described for barnase, the effect of the introduction of the P1OOG mutation in NDP kinase is to shift the folding equilibria to favor the MG folding intermediate, also at the expense of the stability of the folded (native) form of enzyme. 608R

Analytical Chemistry, Vol. 86, No. 12, June 15, 1994

Unfolding of recombinant human stefin B, induced by GuHCl, low-pH, and high-temperature treatments was investigated by Zerovnik et al. (JI 19). In all three denaturing treatments, conditions were found where an apparent MG folding intermediate was formed, exhibiting the appropriate hydrodynamically compact character. The unfolding of monomeric stefin B by guanidine hydrochloride exhibited an apparent two-state transition a t low temperature, but a threestate pattern at 25 OC, with a discernable monomer to dimer transition, followed by unfolding of the dimer. The spectroscopic features of the dimeric form were consistent with an associated (dimeric) MG state, suggesting the folding scheme described by eq 4, but with the possibility of a direct transition of MG2 to unfolded chains. The kinetic resolution of the experiments was presumably inadequate to detect a probable transient monomeric MG intermediate. In addition to equilibrium analyses, several publications have dealt with the use of SEC to describe the kinetics of protein associations and protein folding, in which species conversion is relatively fast. The two greatest challenges to analyses of rapidly converting species by SEC are that (1) analytes passing through the SEC column suffer dilution in the mobile phase and (2) species which convert with kinetics faster than, or comparable to, the time scale of the chromatographic separation will continuously change their concentrations during the separation process. Although a variety of calculation procedures have been employed to reconstruct or model the band shape and area of eluting peaks, in order to obtain quantitative information on equilibrium concentrations and the kinetics of interconversion, evaluation of such systems remains a challenging task. Patapoff and colleagues(J120)studied the monomeraimer equilibrium of recombinant human growth hormone (hGH) and determined the kinetics of dimer dissociation. In this study, computer simulation of the eluting bands was employed to estimate the dissociation rate constant and equilibrium constant. Elution of the partially separated monomers and dimers of hGH could be rapidly conducted by the use of short (4-cm) columns operated at elevated flow rates, effectively reducing the chromatographic run times to