B T Sewell and M Jaffer
Electron Microscope Unit
University of Cape Town
In eukaryotes (organisms whose genetic material is contained in a cell nucleus), the nuclear DNA, RNA and various key proteins are are found in fibrous structures called chromatin. These structures are clearly visible in electron microscopic studies. Studies on the structure of chromatin obtained from model eukaryotes are important insofar as they shed light on structure-activity relationships for the eukaryotic genetic material.
In eukaryotes, the nuclear DNA is packaged by nuclear proteins called histones, resulting in the formation of entities called nucleosomes. Nucleosomes present as oblate spheroids with a height of 6 nm and a diameter of 11 nm. The nucleosomes, by interacting with each other under physiological conditions, form fibres in which the nucleosomes are thought to stack with their long axis aligned along the fibre axis.
A full understanding of the events occurring in transcription and replication can only be achieved if the compact structure of the chromatin fibre is known. Various models for the arrangement of the nucleosomes in the fibre have been proposed. A number of techniques, including two-dimensional electron microscopy, x-ray scattering, fluorescence, sedimentation, enzyme digestion, antibody labelling and circular dichroism studies have been brought to bear on this problem, but so far, they have failed to unambiguously define the structure of compact chromatin fibres.
In the study reported here, we have attempted to solve the structure of the chromatin fibre by a direct approach. We have tackled the problem by tomographic reconstruction of several electron microscopic tilt series of negatively stained chromatin fibres isolated from sea urchin sperm. Sea urchin sperm represents a system in which the packaging function is paramount. Although the structure of sea urchin sperm chromatin up to the level of the nucleosome has properties similar that described for the conventionally studied chicken erythrocyte and rat liver chromatin, it does exhibit a large repeat length and a fibre diameter of 45 nm. The fibres can be unravelled by exposure to low ionic strength conditions, resulting in a "beads-on-a-string" appearance. The refolding of the unravelled chromatin can be seen to result in a zig-zag type structure before being condensed into fibres.
When an object is tilted around a single axis, the reconstruction of the three-dimensional volume from the two-dimensional projections can be thought of as a buildup of slices where each slice is reconstructed from a combination of [Xn;Y1...n] at the various tilt angles. This is termed tomography. Tomography can be applied to crystalline objects, objects having some symmetry and to amorphous objects. However, strictly speaking, electron microscopic tomography, only deals with the latter class of objects.
This technique can yield important structural data, but for optimal results, great care must be taken at all steps of the investigation, from the initial preparation and selection of the the object, the correct alignment of the image, the collection of the data, to the choice of the algorithm used in the reconstruction process.
The sequence of steps involved in obtaining the three dimensional structure of the chromatin fibres, from the isolation stage to the interpretation of the reconstructed images, is listed below:
Chromatin fibres were isolated from sea urchin sperm under conditions which maintained the fibres in a condensed state. These fibres were fixed with formaldehyde prior to adsorbing them to a carbon support on copper grids to which gold beads had been attached. The fibres were negatively stained with methylamine tungstate.
In order to achieve the desired resolution in the reconstruction (i.e., the resolution of objects of nucleosomal size), it was necessary to take over twenty one micrographs covering the tilt range ± 60°. Photography of the chromatin fibres was done under minimum dose conditions in a JEM 200CX (JEOL Ltd.)
The micrographs were converted to digital format by densitometry on a Joyce Loebl MDM 6 two-dimensional, flat bed scanning microdensitometer. Images were usually scanned to give a pixel size of 0.83 nm or 0.54 nm.
Methods described for determining a common coordinate system can be divided into two classes, namely those that use correlation techniques and those that use fiducial markers.
Our alignment strategy involved the use of fiducial markers, which are typically colloidal gold particles that are distributed on the support film around the specimen of interest. In this method, the relationship between the specimen and a digital raster coordinate system is determined from a least-squares analysis of the measured positions of the digital markers in the scanned images. The images are transformed to a common coordinate system and the pixel densities are resampled by interpolation, prior to any reconstruction taking place.
Depending on the nature of the specimen to be reconstructed, the commonly used reconstruction techniques can create a problem, since in order to obtain creditable reconstructions, filters, which are obtained in an ad-hoc manner, have to be applied. These filters which effectively exclude high resolution data (e.g. noise), also remove any non-noise data thereby leading to artifacts in the reconstruction. The generation of these artifacts however can be overcome if the object lends itself to some sort of symmetry or if many copies of the object being reconstructed have the same structure. However, chromatin fibres do not lend itself towards this type of analysis and hence in order to get reconstructions which are devoid of artifacts, all the information present in the sample has to be preserved. Thus an alternate method namely the maximum entropy method, which does not involve the use of filters, was developed.
The maximum entropy method is used extensively in astronomy and medical tomography and is particularly useful for the restoration of images from incomplete and noisy data.
This method of image restoration or reconstruction is an iterative method that models the process of image formation in which a trail structure is matched to the experimental data at every attempt to satisfy certain criteria in that the image, with the highest entropy is selected subject to preset constraints. Only the forward operation in the reconstruction process is modelled. The problems associated with the inversion techniques are not encountered. The definition of entropy also makes it impossible for values of the density to become non-negative.
The reconstructed fibres were examined by looking at slices either parallel or perpendicular to the carbon film using the programs SEMPER and SPIDER.
As an aid to interpreting the reconstructed images, contour maps of the reconstructed images were obtained and displayed on an interactive graphics display workstation using the program FRODO.
Evaluation of the above fibre showed that the nucleosomes lie on two intertwined helical ramps. Constrictions along the length of the fibre are the cross-over points. The properties exhibited by the reconstructed chromatin fibres appear to conform with the proposed fibre crossed-linker model. Tomographic studies at increased resolution are required to gain further insight into fibre structure.
Fundamental studies in structural biology have always been dependent on the study of suitable model systems, ultimately leading up to more general and more complex systems. In the study of chromatin, it is found that echinoderm (such as sea urchins and starfishes) sperm is the only tissue which is completely transcriptionally inactive, packaged in the same way as somatic tissue. All other sperm cells have chromatin packaged with protamines, which greatly complicate the structural investigations.