Growth of 3D Protein Crystals. Insulin.

Why are we interested in studying the growth of 3D protein crystals?

Large, well-ordered 3D protein crystals are required for solving protein structure by X-ray crystallography. Formation of such crystals remains the botteleneck of crystallographic studies, even for soluble proteins. Understanding the mechanisms of crystal growth and defect formation will lead to rational approach to crystallization and an improvement in the success rate of crystallization trials.

Proteins are excellent model systems with which fundamental physical processes of self-assembly can be studied.

In vivo, proteins perform multitudes of functions. Crystalline proteins may be useful for biotechnological purpouses.

They (the crysals) are beautiful.

Even larger crystals (milimeters in size) are required for neutron scattering studies, which are aimed at the elucidation of the locations of hydrogen atoms in the crystals (recall that hydrogens are invisible to X-rays and their locations are inferred from the X-ray difraction studies). Take a look for yourself here. To grow larger crystals, exquisit control over the growth conditions is required. The ability to control the growth of crystals depends on the understanding of the various factors that affect the crystal growth.

Crystals of proteins are already used in pharmaceudical industry. Insulin preparations used to treat diabetis are suspensions of crystals.

How do 3D crystals grow?

Below the roughening transition, 3D crystals grow by spreading of layers.

Molecules attach to the edges of these layers (called steps) at specific sites. These attachement sites are called kinks.

Layers can be generated at screw dislocations (see below), by two-dimensional nucleation, or by three-dimensional nucleation. Here is an excellent web page by James De Yoreo from Lawrence Livermore National Laboratory with some atomic force microscopy images of these processes.

A sequence of steps, such as one shown on the right, is called "step train".

Diagram showing how molecules join growing crystals at the kink sites along the steps.

Growth of rhombohedral (R3) insulin 3D Crystals

Optical Image of an R3 insulin crystal

An optical micrograph of a rhombohedral (R3) insulin 3D crystal is shown on the left. The corner indicated with solid arrows is closest to the observer. The corner indicated with dashed arrows is furthest from the observer. The green line connecting the two corners is drawn to indicated one of the three-fold symmetry axes. The area imaged by AFM is outlined in white (not to scale). The edge of the crystal is about 200 um long.

You can learn about crystal symmetries and crystal habit here, here, and here. Specific information pretaining to the rhombohedral lattice can also be found here. The pdb code of this form of insulin is 4INS.

These crystals grow by spreading of layers generated at screw dislocations. The AFM image on the left shows a group of screw dislocations, pointed to by arrowheads, on the <100> face of a rhombohedral insulin crystal (an optical microscopy image of such a crystal is shown above). The image is 9.5 x 9.5 um2, Z-scale (black-to-white) is 30 nm. Dislocations act as sources of steps, giving the whole face a staircase-like appearence (especially clear along the black line. See also the image below). These steps are one molecule high. New molecules join the crystal at the edges of the steps. Thus the crystal grows by spreading of the existing layers and generation of new ones, at the screw dislocations. This is one of several possible modes of growth of 3D crystals.

Height profile of the image above, measured along the black line. The total height is ~ 30 nm, while the height of the individual steps is 3.4 nm - corresponding to thickness of one molecule of insulin.

Dislocations in an insulin crystal

A movie showing the same group of screw dislocations in action. Each dislocation generates single-molecule steps, which propagate away from the source. A series of AFM images (9.5 x 9.5 um2) of an insulin 3D crystal growingin situ taken at approximately 50 s intervals, were used to construct this movie.

Absence of step bunching in rhombohedral (R3) insulin 3D crystals

Step trains in insulin crystals look like this:


Equidistant Step Train in an Insulin Crystal

The image is 11 x 11 um2. The width of the terraces is almost the same throughout. This is an example of an equidistant step train.

More often, step trains in crystals look like this:


Step Bunches in an Instulin Crystal

The image is 20 x 20 um2.
Step density is seen to vary. (In this case, the variation is caused by a defect that is located upstream (not shown)).

The areas of high step density seen in the image on the right are called "step bunches". Formation of step bunches is a manifistation of instabilities that occur during step generation or growth. It appears that equidistant step trains - step trains without bunches, such as the one shown in the left image - are stable in the case of insulin. This is due to the regularity of step generation at screw dislocations, lack of step-step interactions, and the fact that the growth of these crystals is predominantly under transport control. These features make insulin crysals a very interesting model system for studying crystal growth.
On one hand, step bunching lowers the quality and utility of the material. On the other hand, patterns of step density may be useful for applications in separation methods or as nano-templates.
See Gliko, Reviakine and Vekilov (2003), PRL 90, 225503-1 - 225503-4 for detailed discussion of these issues.

Dislocation Hollow Cores

The core of a screw dislocation is strained. Strained areas dissolve at higher supersaturations than non-strained areas. Image on the right shows a hollow core of a screw dislocation. The concentration of insulin was chosen in such a way that the solution was still supersaturated with respect to the non-strained areas (step edges) - i.e., the crystal was still growing - but undersaturated with resepct to the strained areas, such as cores of screw dislocations. They dissolved, leaving behind hollow channels. One such channel is visible in the center of the image (green arrowheads). Turquoise arrowheads point to insulin hexamers that make up the crystal, while the black arrowhead in the lower left points to the step that winds around the disclocation core. Notice that the channel is not isotropic. The anisotropy of the hollow core is the manifistation of the anisotropy of the line tension of the steps with respect to their crystallographic orientation (steps with different orientations have different line tensions).
The image on the right was obtained in contact mode.


125 nm2

Crystal Dissolution: Etch Pits

Etch Pits in a dissolving insulin crystal

30 um2


When the concentration of insulin in solution is lowered to below that needed to sustain crystal growth (the solution is said to become undersaturated), the crystal begins to dissolve. Etch pits (image on the left) form in the place of the dislocations as they unwind.
The image on the left was obtained in tapping mode.

The AFM studies shown here build on the body of knowledge developed by several groups over the past years. You may be interetsted in looking up the work of prof. A. McPherson and colleagues at UC Irvine, James De Yoreo and colleagues from Lawrence Livermore National Laboratory, as well as other papers on crystal growth by prof. Vekilov.