Neurochemical Research, Vol. 17, No. 2, 1992, pp. 157-166
Is Myelin Basic Protein Crystallizable? Jan Sedzik 1,2 and Daniel A . Kirschner 1
(Accepted June 18, 1991)
Myelin basic protein (MBP) is the predominant extrinsic protein in both central and peripheral nervous system myelins. It is thought to be involved in the stabilizing interactions between myelin membranes, and it may play an important role in demyelinating diseases such as multiple sclerosis. In spite of the fact that this abundant protein has been known for almost three decades, its threedimensional crystal structure has not yet been determined. In this study we report on our extensive attempts to crystallize the major 18.5 kDa isoform of MBP. We used MBP having different degrees of purity, ranging from crude MBP (that was acid or salt extracted from isolated myelin), to highest purity single isoform. We used conventional strategies in our search for a suitable composition of a crystallization medium. We applied both full and incomplete factorial searches for crystallization conditions. We analyzed the available data on proteins which have previously resisted crystallization, and applied this information to our own experiments. Nevertheless, despite our efforts which included 4600 different conditions, we were unable to induce crystallization of MBP. Previous work on MBP indicates that when it is removed from its native environment in the myelin membrane and put in crystallization media, the protein adopts a random coil conformation and persists as a population of structurally non-identical molecules. This thermodynamically preferred state presumably hinders crystallization, because the most fundamental factor of protein crystallization - homogeneity of tertiary structure - is lacking. We conclude that as long as its random coil flexibility is not suppressed, 18.5 kDa MBP and possibly also its isoforms will remain preeminent examples of proteins that cannot be crystallized. KEY WORDS: Myelin basic protein; random coil; X-ray diffraction; protein crystallization; beta-conformation.
ture of MBP could provide insight regarding its role in myelin stability and degeneration. Therefore, after the successful crystallization of P2 basic protein of myelin (3) we chose to crystallize MBP. Compared with P2, which is predominantly (--70%) in the 13-conformation in both aqueous solution (4) and in the crystal (5), MBP has been determined by circular dichroism (CD) to have little, if any, a-helix or 13-conformation (6). By contrast with this experimental evidence, secondary structure predictions for MBP reveal the potential for a series of 13-strands (7-9) that are conserved phylogenetically and could direct the folding of the molecule (10). In the current paper we report on our extensive efforts to crystallize 18.5 kDa MBP, the major isoform in human, bovine and rabbit CNS myelin. Our inability to obtain crystals is likely due to the extreme conforma-
INTRODUCTION Myelin basic protein (MBP) is a major extrinsic protein of the myelin sheath where it is thought to underlie membrane compaction at the cytoplasmic apposition in CNS myelin (1) and to play a central role in demyelinating disease such as multiple sclerosis (2). MBP is easily extracted from myelin under acidic or hypertonic conditions, and its amino acid sequence has been known for about 20 years; however, MBP has never been crystallized. Knowing the three-dimensional struci Division of Neurology Research, The Children's Hospital and Department of Neuropathology, Harvard Medical School, 300 Longwood Ave, Boston, Massachusetts 02115. 2 To whom correspondence should be addressed.
157 0364-3190/92/0200-0157506.50/09 1992PlenumPublishingCorporation
158
tional flexibility of the protein. Unless such flexibility is suppressed, MBP, as well as its homologous isoforms (10), may serve as an example of a class of non-crystallizable protein. This work has been presented in preliminary form (11).
Sedzik and Kirschner
A
B
EXPERIMENTAL PROCEDURE Pre-Crystallization Set-Up. Before designing the major crystallization experiments, we compiled from the literature: (1) the physicochemical characteristics of MBP, (2) all information regarding its precipitation, aggregation and solubility in aqueous and organic solutions, and (3) any salt and pH dependent conformational changes of this protein. Because physicochemical data for MBP (particularly the 18.5 kDa form) is thoroughly documented, including post-translational modifications (12) we have not made any attempt here to replicate such data. Reagents and Source of MBP. Polyethylene glycols (10 and 20 kDa) and 2-mcthyl-3,4--pentanediol (MPD) were from Aldrich (Milwaukee, Wisconsin). All other chemicals were of analytical grade (Sigma, St. Louis, USA). The water used was either Millipore-Q standard or glass distilled. All aqueous solutions were filtered (45 l~m) and degassed. MBP used for crystallization was from bovine or rabbit tissue, after different degrees of final purification: (1) concentrated acid extract (bovine); (2) mixture of MBP isoforms (bovine, rabbit); and (3) highest purity single isoform (rabbit). To purify MBP from bovine brain or spinal cord, we used the published protocol (13). In general, MBP was extracted at pH 2-3 (30 mM HC1) from delipidated tissue. The crude protein was precipitated from the acid extract by ethanol, acetone or methanol. In some experiments MBP was extracted directly from purified, lyophilizedmyelin by using low pH or high ionic strength. The final purification was by gel filtration (Sephadex G50; Pharmacia) and/or on ion exchange (CM-52; Whatman). Purified fractions containing MBP were pooled, dialyzed against water or buffer, and lyophilized or concentrated on Amicon membranes. SDS-PAGE and urea-PAGE was used to check MBP purity. As expected, only one band of M~ 18.5 kDa was detected in the presence of SDS, and 4-5 bands (charge isoforms) were detected in the presence of urea at pH 10.6 (Figure 1A,B). We also used MBP (rabbit) commercially available (Sigma, product #M-1891 and #M-2016). Highly purified single isoform rabbit MBP was kindly provided by Dr. R.E. Martenson (NIH, Bethesda, MD). MBP Assay. The protein content of aqueous solutions was determined either gravimetrically (when using lyophilized MBP, after extensive dialysis against water) or by measuring the absorption of the aqueous solution and knowing the MBP extinction coefficient, E(l%,l cm) = 5.44 (14). Physicochemical Modifications of MBP. Binding of Mgz+, I-Ig2+, and Znz§ was performed according to the method of Berlet (15). Unbound ions were removed by dialysis against the buffer. Methanol, ethanol, trifluoroethanol,surfactants, glycerol, and spermine were added to buffered protein solutions before crystallization. It was expected that either the added counterions or the other reagents would generate some degree of secondary structure in MBP, suppress its flexibility, or increase its conformational adaptability to crystallization (16-21). General Crystallization Strategy. Following conventional crystallization strategy, our objective was to avoid precipitation of MBP from solution as its solubility in the crystallization medium gradually
~!i84 ~
~ ~ ~ ~i~
Fig. 1. Purity of isolated bovine myelin basic protein. (A) SDS-PAGE (12% acrylamide, 5% bis-acrylamide) shows a single band for the protein. (B) Urea-PAGE shows 4-5 charge isoforms (5% acrylamide, 8M urea, glycine-NaOH buffer, pH = 10.4). The polypeptide used for crystallization is indicated by the arrows.
decreased to a minimum. It was expected that achieving some degree of supersaturation would result in stable nucleation sites followed by crystallization. To decrease the solubility of MBP, we used salts, polyethylene glycols (2-20 kDa), and organic solvents. The concentration of MBP was in the range of 10-50 mg/ml in 20-50 mM buffers at pH 3-10. Crystallization Techniques. Our crystallization techniques included the conventional procedures of bulk crystallization, dialysis, free interface diffusion and vapor diffusion (22-24). For practical reasons vapor diffusion in hanging or sitting (10-30 M) drops was most often used. Crystallization experiments were performed routinely at 4~ 20~ and sometimes at 35~ with or without the counterions and other additives described above. We developed a computer program to generate crystallization experiments based on a full factorial design, incomplete design or saturated factorial design (25-28). Detailed examples of typical pre-crystallization searches for determining minimum MBP solubility under particular sets of factors are included in Results. X-ray Diffraction Techniques. Crystalline material (dimensions at
Crystallizability of Myelin Basic Protein least 0.2 x 0.2 x 0.2 ram) were transferred into glass capillaries of diameter 0.5-1.0 mm (Charles Supper Company, Natick, MA), sealed with paraffin and mounted for a still exposure on a precession (Buerger) camera (kindly loaned by Dr. L. Makowski, Boston University) that was set up either on the double-mirror focussing camera of an Elliott GX-20 rotating anode X-ray generator (operated at 35 kV, 35 mA; spot size 250 ~m) or on a Rigaku sealed tube X-ray generator (run at 40 kV, 25 mA; colimator 0.4 mm). In both cases the X-ray wavelength was 1.542 ~. The distance between the film and crystal was 75-85 mm, and the temperature during exposure was ambient (1820~ Diffraction patterns were recorded on Kodak DEF film. Scoring of Crystallization Experiments. Each crystallization experiment was periodically monitored for at least 6 months and scored according to the Carters' scale (25).
RESULTS
159 Table
Condition No. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12.
I. Precipitation of MBP by Salt 1
Volume (ixl) 200 210 220 230 240 250 260 270 280 290 300 310
Ixl Added ppt H20 10 10 10 10 10 10 10 10 10 10 10 10
Observation transparent transparent transparent little cloudy very cloudy cloudy cloudy cloudy less cloudy transparent transparent transparent
1 The protein was lyophilized bovine MBP, and at a starting concentration of 14 mg/ml. The buffer was 50 mM imidazole at pH 6.5. The precipitant (ppt) was 3 M ammonium sulfate in water. The temperature was 20~
Overview and Description of Typical Experiments MBP was available in sufficiently large amounts to permit us to design numerous experiments on a large scale (i.e., using I ml sample sizes). In a typical crystallization trial, with a particular precipitant, pH and temperature, we first searched for a minimum solubility of MBP. The subsequent crystallization experiments (for particular sets of factors) were based on those initial observations. A few examples of our precrystallization searches are described below. Example I: Precipitation of MBP by Salt
Rationale and Design. Ammonium sulfate (AmS) was chosen as the precipitant, because historically it has been the most effective in the crystallization of other proteins (22,24). We used lyophilized MBP solubilized in aqueous solution at pH
Is Myelin Basic Protein Crystallizable? Jan Sedzik 1,2 and Daniel A . Kirschner 1
(Accepted June 18, 1991)
Myelin basic protein (MBP) is the predominant extrinsic protein in both central and peripheral nervous system myelins. It is thought to be involved in the stabilizing interactions between myelin membranes, and it may play an important role in demyelinating diseases such as multiple sclerosis. In spite of the fact that this abundant protein has been known for almost three decades, its threedimensional crystal structure has not yet been determined. In this study we report on our extensive attempts to crystallize the major 18.5 kDa isoform of MBP. We used MBP having different degrees of purity, ranging from crude MBP (that was acid or salt extracted from isolated myelin), to highest purity single isoform. We used conventional strategies in our search for a suitable composition of a crystallization medium. We applied both full and incomplete factorial searches for crystallization conditions. We analyzed the available data on proteins which have previously resisted crystallization, and applied this information to our own experiments. Nevertheless, despite our efforts which included 4600 different conditions, we were unable to induce crystallization of MBP. Previous work on MBP indicates that when it is removed from its native environment in the myelin membrane and put in crystallization media, the protein adopts a random coil conformation and persists as a population of structurally non-identical molecules. This thermodynamically preferred state presumably hinders crystallization, because the most fundamental factor of protein crystallization - homogeneity of tertiary structure - is lacking. We conclude that as long as its random coil flexibility is not suppressed, 18.5 kDa MBP and possibly also its isoforms will remain preeminent examples of proteins that cannot be crystallized. KEY WORDS: Myelin basic protein; random coil; X-ray diffraction; protein crystallization; beta-conformation.
ture of MBP could provide insight regarding its role in myelin stability and degeneration. Therefore, after the successful crystallization of P2 basic protein of myelin (3) we chose to crystallize MBP. Compared with P2, which is predominantly (--70%) in the 13-conformation in both aqueous solution (4) and in the crystal (5), MBP has been determined by circular dichroism (CD) to have little, if any, a-helix or 13-conformation (6). By contrast with this experimental evidence, secondary structure predictions for MBP reveal the potential for a series of 13-strands (7-9) that are conserved phylogenetically and could direct the folding of the molecule (10). In the current paper we report on our extensive efforts to crystallize 18.5 kDa MBP, the major isoform in human, bovine and rabbit CNS myelin. Our inability to obtain crystals is likely due to the extreme conforma-
INTRODUCTION Myelin basic protein (MBP) is a major extrinsic protein of the myelin sheath where it is thought to underlie membrane compaction at the cytoplasmic apposition in CNS myelin (1) and to play a central role in demyelinating disease such as multiple sclerosis (2). MBP is easily extracted from myelin under acidic or hypertonic conditions, and its amino acid sequence has been known for about 20 years; however, MBP has never been crystallized. Knowing the three-dimensional struci Division of Neurology Research, The Children's Hospital and Department of Neuropathology, Harvard Medical School, 300 Longwood Ave, Boston, Massachusetts 02115. 2 To whom correspondence should be addressed.
157 0364-3190/92/0200-0157506.50/09 1992PlenumPublishingCorporation
158
tional flexibility of the protein. Unless such flexibility is suppressed, MBP, as well as its homologous isoforms (10), may serve as an example of a class of non-crystallizable protein. This work has been presented in preliminary form (11).
Sedzik and Kirschner
A
B
EXPERIMENTAL PROCEDURE Pre-Crystallization Set-Up. Before designing the major crystallization experiments, we compiled from the literature: (1) the physicochemical characteristics of MBP, (2) all information regarding its precipitation, aggregation and solubility in aqueous and organic solutions, and (3) any salt and pH dependent conformational changes of this protein. Because physicochemical data for MBP (particularly the 18.5 kDa form) is thoroughly documented, including post-translational modifications (12) we have not made any attempt here to replicate such data. Reagents and Source of MBP. Polyethylene glycols (10 and 20 kDa) and 2-mcthyl-3,4--pentanediol (MPD) were from Aldrich (Milwaukee, Wisconsin). All other chemicals were of analytical grade (Sigma, St. Louis, USA). The water used was either Millipore-Q standard or glass distilled. All aqueous solutions were filtered (45 l~m) and degassed. MBP used for crystallization was from bovine or rabbit tissue, after different degrees of final purification: (1) concentrated acid extract (bovine); (2) mixture of MBP isoforms (bovine, rabbit); and (3) highest purity single isoform (rabbit). To purify MBP from bovine brain or spinal cord, we used the published protocol (13). In general, MBP was extracted at pH 2-3 (30 mM HC1) from delipidated tissue. The crude protein was precipitated from the acid extract by ethanol, acetone or methanol. In some experiments MBP was extracted directly from purified, lyophilizedmyelin by using low pH or high ionic strength. The final purification was by gel filtration (Sephadex G50; Pharmacia) and/or on ion exchange (CM-52; Whatman). Purified fractions containing MBP were pooled, dialyzed against water or buffer, and lyophilized or concentrated on Amicon membranes. SDS-PAGE and urea-PAGE was used to check MBP purity. As expected, only one band of M~ 18.5 kDa was detected in the presence of SDS, and 4-5 bands (charge isoforms) were detected in the presence of urea at pH 10.6 (Figure 1A,B). We also used MBP (rabbit) commercially available (Sigma, product #M-1891 and #M-2016). Highly purified single isoform rabbit MBP was kindly provided by Dr. R.E. Martenson (NIH, Bethesda, MD). MBP Assay. The protein content of aqueous solutions was determined either gravimetrically (when using lyophilized MBP, after extensive dialysis against water) or by measuring the absorption of the aqueous solution and knowing the MBP extinction coefficient, E(l%,l cm) = 5.44 (14). Physicochemical Modifications of MBP. Binding of Mgz+, I-Ig2+, and Znz§ was performed according to the method of Berlet (15). Unbound ions were removed by dialysis against the buffer. Methanol, ethanol, trifluoroethanol,surfactants, glycerol, and spermine were added to buffered protein solutions before crystallization. It was expected that either the added counterions or the other reagents would generate some degree of secondary structure in MBP, suppress its flexibility, or increase its conformational adaptability to crystallization (16-21). General Crystallization Strategy. Following conventional crystallization strategy, our objective was to avoid precipitation of MBP from solution as its solubility in the crystallization medium gradually
~!i84 ~
~ ~ ~ ~i~
Fig. 1. Purity of isolated bovine myelin basic protein. (A) SDS-PAGE (12% acrylamide, 5% bis-acrylamide) shows a single band for the protein. (B) Urea-PAGE shows 4-5 charge isoforms (5% acrylamide, 8M urea, glycine-NaOH buffer, pH = 10.4). The polypeptide used for crystallization is indicated by the arrows.
decreased to a minimum. It was expected that achieving some degree of supersaturation would result in stable nucleation sites followed by crystallization. To decrease the solubility of MBP, we used salts, polyethylene glycols (2-20 kDa), and organic solvents. The concentration of MBP was in the range of 10-50 mg/ml in 20-50 mM buffers at pH 3-10. Crystallization Techniques. Our crystallization techniques included the conventional procedures of bulk crystallization, dialysis, free interface diffusion and vapor diffusion (22-24). For practical reasons vapor diffusion in hanging or sitting (10-30 M) drops was most often used. Crystallization experiments were performed routinely at 4~ 20~ and sometimes at 35~ with or without the counterions and other additives described above. We developed a computer program to generate crystallization experiments based on a full factorial design, incomplete design or saturated factorial design (25-28). Detailed examples of typical pre-crystallization searches for determining minimum MBP solubility under particular sets of factors are included in Results. X-ray Diffraction Techniques. Crystalline material (dimensions at
Crystallizability of Myelin Basic Protein least 0.2 x 0.2 x 0.2 ram) were transferred into glass capillaries of diameter 0.5-1.0 mm (Charles Supper Company, Natick, MA), sealed with paraffin and mounted for a still exposure on a precession (Buerger) camera (kindly loaned by Dr. L. Makowski, Boston University) that was set up either on the double-mirror focussing camera of an Elliott GX-20 rotating anode X-ray generator (operated at 35 kV, 35 mA; spot size 250 ~m) or on a Rigaku sealed tube X-ray generator (run at 40 kV, 25 mA; colimator 0.4 mm). In both cases the X-ray wavelength was 1.542 ~. The distance between the film and crystal was 75-85 mm, and the temperature during exposure was ambient (1820~ Diffraction patterns were recorded on Kodak DEF film. Scoring of Crystallization Experiments. Each crystallization experiment was periodically monitored for at least 6 months and scored according to the Carters' scale (25).
RESULTS
159 Table
Condition No. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12.
I. Precipitation of MBP by Salt 1
Volume (ixl) 200 210 220 230 240 250 260 270 280 290 300 310
Ixl Added ppt H20 10 10 10 10 10 10 10 10 10 10 10 10
Observation transparent transparent transparent little cloudy very cloudy cloudy cloudy cloudy less cloudy transparent transparent transparent
1 The protein was lyophilized bovine MBP, and at a starting concentration of 14 mg/ml. The buffer was 50 mM imidazole at pH 6.5. The precipitant (ppt) was 3 M ammonium sulfate in water. The temperature was 20~
Overview and Description of Typical Experiments MBP was available in sufficiently large amounts to permit us to design numerous experiments on a large scale (i.e., using I ml sample sizes). In a typical crystallization trial, with a particular precipitant, pH and temperature, we first searched for a minimum solubility of MBP. The subsequent crystallization experiments (for particular sets of factors) were based on those initial observations. A few examples of our precrystallization searches are described below. Example I: Precipitation of MBP by Salt
Rationale and Design. Ammonium sulfate (AmS) was chosen as the precipitant, because historically it has been the most effective in the crystallization of other proteins (22,24). We used lyophilized MBP solubilized in aqueous solution at pH
