Explanation

Proteins are the main building unit of all living creatures and essential components of all information and energy processing involved in life. Hence, understanding of genome-structure-function correlations for this group of natural compounds [1] has emerged as a focus of intense recent investigations. Although advances in nuclear magnetic resonance techniques continue to shift upwards the limit of protein molecular sizes accessible to this method [2], the most widely used means for protein structure studies is still diffraction of x-rays, electrons, or neutrons, by protein crystals. To resolve atoms that are, typically, 1.5 - 2 Å apart, these diffraction methods require single crystals as large as several tenths of a millimeter in all three dimensions, of low defect contents and high compositional and structural uniformity. Recent advances in protein expression and characterization and purification techniques, as well as beam and detector technology and in computational crystallography greatly accelerated the investigation steps that precede and follow crystallization [3]. Thus, the preparation of diffraction-quality crystals has emerged as the bottleneck of macromolecular structure determinations [4,5].

Beyond protein single crystal growth, progress in various biochemical and biomedical research and production tasks is impeded by a lack of insight into protein nucleation and growth. For instance, the slow dissolution rate of protein crystals is utilized to achieve sustained release of medications, such as insulin and interferon-a [6,7,8,9,10]. Work on the crystallization of other proteins for this purpose, e.g. human growth hormone, is currently under way. Steady release rates can be maintained longer if an administered medication dose consists of fewer, larger, equi-sized crystallites. To achieve such size distributions, crystal nucleation must be limited to a short time interval so that all crystals grow at the same decreasing supersaturation.

Furthermore, some pathological conditions are related to the formation of crystals or amorphous solid aggregates in the human body. An often cited example is the crystallization of hemoglobin C, and the polymerization of hemoglobin S that cause, respectively, the CC and sickle-cell anemia [11,12,13].

An unrecognized application of protein crystallization is that these systems represent a convenient model for the crystallization phenomena that occur in a variety of systems: from water freezing in clouds and oceans, through magma solidification in the Earth's interior, to the pulling of 12 and 18 inch semiconductor dies, to casting of metals and alloys. Given the resolution limits of the modern surface characterization techniques, proteins are a particularly attractive systems for studies of the fundamental crystal growth mechanisms: on the one hand, the size of the protein molecules, of the order of a few nanometers, and the typical growth time scales, with few seconds between sequential discrete growth events, are within the reach of the advanced experiment devices; on the one hand, the protein molecular mass leaves the thermal equilibration times short. This latter argument makes the conclusions of such model studies meaningful for the other crystallization systems, and proteins a better model than, for instance, often cited colloids [14,15,16].

1. Darby N.J.; Creighton, T.E. Protein Structure; Oxford Univ. Press: Oxford, 1993.
2. Glaser, S. J.; Schulte-Herbrüggen, T.; M. Sieveking, O. Schedletzky, N. C. Nielsen, O. W. Sørensen, C. Griesinger, Science 1998, 280, 421.
3. Chayen, N.E.; Boggon, T.J.; Casseta, A.; Deacon, A.; Gleichmann, T.; Habash, J.; Harrop, S.J.; Helliwell, J.R.; Neih, Y.P.; Peterson, M.R.; Raftery, J.; Snell, E.H.; Hädener, A.; Niemann, A.C.; Siddons, D.P.; Stojanoff, V.; Thompson, A.W.; Ursby, T.; Wulff, T. M. Quarterly Reviews of Biophysics 1996, 29, 227.
4. DeLucas, L.J ; Bugg, C.E. Trends in Biotechnology 1987, 5, 188.
5. Weber, P. in Advances in protein chemistry, Vol. 41, edited by Afinsen, C.B.; Richards, F.M.; Edsal, T.J.; Eisenberg, D.S.; Academic Press: New York, 1991.
6. Matsuda, S.; Senda, T.; Itoh, S.; Kawano, G.; Mizuno, H.; Mitsui, Y. (). J. Biol. Chem. 1989, 264, 13381.
7. Peseta, S.; Langer, J.A.; Zoon, K.C.; Samuel, C.E. in Annual Review of Biochemistry, Vol. 56, edited by Richardson, C.C.; Boyer, P.D.; Dawid, I.B.; Meister, A.; Annual Reviews: Palo Alto, 1989; p. 727.
8. Brange, J. Galenics of Insulin. Berlin: Springer, 1987.
9. Reichert, P.; McNemar, C.; Nagabhushan, N.; Nagabhushan, T.L.; Tindal, S.; Hruza, A. Metal-interferon-alpha crystals. US Patent No. 5,441,734, 1995.
10. Long, M.L.; Bishop, J.B.; Nagabhushan, T.L.; Reichert, P.; Smith, G.D.; DeLucas, L.J. J. Crystal Growth 1996, 168, 233.
11. Charache, S.; Locksard-Conley, C.L; Waugh, .D.F.; Ugoretz, R.J.; Spurrell, J.R. J. Clinical Investigations 1967, 46, 1795.
12. Hirsch, R.E.; Raventos-Suarez, C.; Olson, J.A.; Nagel, R.L. Blood 1985, 66, 775.
13. Ferrone, F.A.; Hofrichter, J.; Eaton, W.A. J. Mol. Biol. 1985, 183, 591; ibid, 611.
14. Pusey, P.N.; van Megen, W. Nature, 1986, 320, 340.
15. Van Mewgen, W; Underwood, S.M. Nature 1993, 362, 616.
16. Harland, J.L.; Henderson, S.J.; Underwood, S.M.; van Megen, W. Phys. Rev. Lett 1995, 75, 3572.

The proteins that we work with

Lysozyme from hen-egg white has been the model protein for crystallization studies for more than 15 years (see, e.g., Shall et al., 1996; 1996a, and refs. therein). Hence, the parameters needed for interpretation of the results, such as diffusivity, crystallization kinetic coefficients and surface energies, are well known. A protocol for high purity based on FPLC has been developed in our lab. The typical impurities for this protein have been identified. We have shown that its most typical impurity, the covalently bound lysozyme dimer, is incorporated into crystals (Thomas, Vekilov & Rosenberger, 1998). Furthermore, the dimer effects on averaged growth kinetics (growth deceleration and cessation) and on the time-dependent kinetics (increase of fluctuation amplitudes) depend on the rate of convective supply to the interface (Vekilov, Thomas & Rosenberger, 1998). A correlation was found between the transport conditions (microgravity versus ground growth), amount of dimer incorporated into crystals grown from an unpurified solution and their diffraction resolution (Carter et al., 1999). Thus, this protein is the prime candidate for tests of the hypothesis that the microgravity reduction of impurity supply will lead to more stable growth and, hence, to more perfect crystals.

The lysozyme molecule, with the active center in the cleft on the upper surface

Insulin, recombinant human
Preparations of micro-crystals of this hormone are one of the commercially available forms of the medication (Brange, 1991). A narrow crystal size distribution is necessary for steady sustained release. Insulin is used as a model for crystalline pharmacological protein materials (Reichert, 1996). Insulin will be provided by Lilly Research Center, see the appended Statement of Investigator. Solubility is dependent on temperature; it is currently crystallized by the temperature method (Brange, 1991, Long et al., 1996).

The insulin hexamer, with 2 Zn atoms (shown in yellow) at the axis; sub-chains colored in different colors

Ferritin/apoferritin from horse spleen
This protein has been extensively studied in connection with possible use of the ferritin iron-containing core in information storage devices (Higo & Nagayama, 1993; Takahashi et al., 1996; Dominiguez-Vera et al., 1996). This protein strongly differs from lysozyme in molecular weight (~30 higher) and structure (24 subunits, spherical shape). A protocol for high purity based on FPLC has been developed in our lab. The crystals, grown from this material diffracted to 1.8 Å, better by about 1 Å than earlier results for this protein (Thomas et al., 1998a), which indicates a high sensitivity to impurities of the growth processes and the crystal quality.

Ferritin molecule consist of 24 sub-chains, shown here in different color. They are arranges in pairs at the 12 faces of a cubeoctahedron. One of six channels along the fourfold axes is seen on top. There are also eight channels along the threefold axes. The structures of these channels is of prime interest to scientist as possible pathways to introduce pharmaceuticals in the protected space under the protein shell.

Hemoglobin C, human
Crystallization, including co-crystallization with other hemoglobin mutant types, has been studied both in vivo and in vitro in connection to investigations of causes of Hb CC disease and the SC disease. Retrograde temperature dependence of solubility was established in our laboratory in collaboration with the laboratories of Drs. Nagel and Elison Hirsch at the Albert Einstein College of Medicine.

Human hemoglobin molecules contain four sub-chains, two a and two b, indicated with different colors; the oxygen-binding hemes are highlighted in brown. Association of O2 molecules, one per heme, lead to a conformational change. The sixth aminoacid in the b-chains, green and red, are prone to a mutation that causes crystallization or aggregation of the protein, associated with deadly anemia, the sickle cell disease.

We are only able to study a few macromolecules with the detail required to obtain insight into the unsteady kinetics of protein crystals. Then, the applicability of the reached conclusions for the multitude unstudied proteins becomes a legitimate concern. To address this issue, we have selected models that cover a wide range of molecular sizes (Mr from 6.5 to 1,450 kDa), structures (1 polypeptide chain in lysozyme, 2 and 4 different chains in insulin and hemoglobin, 24 similar chains in apoferritin, viral arrangement of a nucleic acid and protein molecules in STMV), sources (plant, avian, mammalian), and functions (enzyme, storage proteins, inhibitor and virus).