The technique of rate-zonal centrifugation was first proposed by Brakke (1951). In essence, the technique is very simple. A small volume of a suspension is layered over a shallow density gradient. The latter is required to stabilize the sedimentation of the particles. On centrifugation, particles move away from the starting zone with velocities determined both by their size and shape and by the centrifugal force to which they are subjected. After centrifugation for a certain time, particles will be found in a series of zones spaced according to the relative velocities of the particles. In this way particles differing in sedimentation rate by 20% or less can be separated without undue difficulty. Rate-zonal centrifugation thus complements differential centrifugation. Rate-zonal separations cannot be carried out in angle head rotors as the sample mixes with the gradient during acceleration of the rotor (0 3.2) and until recently, the capacity of swing-out rotors was severely limited. As the sedimenting zones are as broad as, or broader than, the starting zone, the volume of material which can be loaded onto a rotor is limited if any degree of separation is to be achieved. In addition, the concentration of material in the sample must not be too high, or the entire band will mix with the top part of the gradient. Rate-zonal centrifugation was initially used mainly for analytical separations such as the analysis of the size distribution of samples of polysomes or of RNA although very soon after the introduction of the technique Thomson and his colleagues used rate sedimentation to separate mitochondria and lysosomes. Improvement in the design of swingout rotors and especially the introduction of zonal rotors has greatly extended the application of rate-zonal centrifugation and it is with this method that we shall be mainly concerned. Vertical tube rotors may also be used.
Preparation of the density gradient
Most density gradient separations in the tubes of swing-out rotors are performed on simple linear density gradients. If the aim is to study the distribution of particles through the gradient rather than prepare a specific fraction, the gradient is normally underlaid with a sufficient volume of a cushion of dense and viscous liquid to fill the curved portion at the bottom of the tube and to prevent any loss of particles into a pellet.
In general, complicated gradients are not used with swing-out rotors. Some will only produce linear gradients; curved gradients may be prepared by use of a closed mixing chamber.
The gradient must be introduced into the tube with the minimum of disturbance. The properties of the plastic walls of the centrifuge tube determine how the gradient should be introduced. Polypropylene is reasonably wettable by aqueous solutions and it is, therefore, simplest to touch the end of the tube leading from the gradient maker onto the wall of the centrifuge and allow the stream of liquid to run slowly down the wall onto the top of the forming gradient. In this case, the gradient should be prepared ‘heavy end first’ and the cushion placed in the bottom of the tubes before starting the preparation of the gradient. This method is not suitable for use with polycarbonate or polyallomer tubes. These plastics are not easily wettable by water so that the gradient material tends to gather into large drops around the tip of the delivery tube coming from the gradient maker. These drops finally detach, and cannon down the side of the tube and would disturb (or even destroy) the gradient. If Polyallomer tubes are soaked in spent chromic acid for 24 hr and then washed extensively with distilled water the plastic becomes readily wettable. This also works with polycarbonate.
Layering of sample onto the gradient
In rate-zonal separations, the layering of the sample onto the gradient and the acceleration of the rotor are the two most critical steps. The sample must be stable. Also one must avoid mixing the sample with the top part of the gradient. Everybody develops their own favorite method for layering samples, but it is advisable to practice layering dyes before attempting real separations.
The sample is drawn up into a relatively large pipette, for example, if a 0.1 ml aliquot were to be layered, one would draw up 0.11 ml aliquot into a 1 ml pipette. The pipette is then immersed in the less dense of the gradient solution and a volume of liquid equal to the volume of sample to be layered allowed to flow into the pipette. Mixing in the pipette forms a small gradient. The liquid is then layered on the main gradient in the normal way. Although this technique is useful where small volumes of concentrated solutions are to be layered onto density gradients, for theoretically layering in a gradient is of advantage when a separation is limited by spreading of the sample band due to sedimentation in droplets. The authors have found that samples are rarely so concentrated that this is the case. When other factors limit resolution, there is no advantage to be gained by layering in a gradient. Thus we would recommend anyone using rate sedimentation routinely to try running in parallel samples layered as described in the first paragraph of this section and samples layered in a gradient to see whether they obtain any improvement by using the latter technique.
Up till now, we have assumed that the sample is to be layered on top of the gradient. However, rate flotation can sometimes be a useful alternative to rate sedimentation. In this case, the sample must be introduced under the gradient. It must be emphasized that on no account must air bubbles be allowed to enter the gradient, for these may pick up small portions of the sample and carry them through the gradient. A second point is that when, as in the illustration, a liquid underlay is used, ‘sedimentation in droplets’ of the samples into the underlay will cause severe spreading of the sample band. It may, therefore, be preferable to fill the curved portion of the gradient with an epoxy plug.
The centrifuge rotor should be accelerated slowly and decelerated without the use of the brake. The powerful motors of modern centrifuges give very high initial acceleration. The resultant Coriolis force can cause considerable mixing of the sample with the upper part of the gradient. Naturally, these effects are more pronounced with low- viscosity salt gradients, but also occur on sucrose gradients. The greatest angular acceleration, and hence the greatest damage to the sample zone, occurs in acceleration through the first few hundred revs/min and cannot be controlled by the normal speed control fitted to ultracentrifuges. If the centrifuge does not have a control per-mitting slow acceleration, a rheostat should be fitted in the motor circuit. This should only be done with the advice and approval of the centrifuge manufacturer. An alternative approach is to use a rotor with a horizontal as well as a vertical pivot so that the line of buckets does not have to lie along a radius of the rotor. In this case, the bucket may swing so that the resultant centrifugal field will always act along the axis of the bucket so that there will be no force tending to cause mixing. Such a modified rotor has been made for use with I.E.C. centrifuges. As the bands of separated particles always broaden somewhat during centrifugation, controlled deceleration is less important than controlled acceleration. Nevertheless, we prefer to decelerate without the use of the brake.
Monitoring the displaced gradient
In practice, the only monitoring which is carried out on material separated in the tubes of swing-out or angle-head rotors is the measurement of the extinction, usually at 260 or 280 nm, to determine the overall distribution of material through the gradient. Such monitoring is usually performed during displacement of the gradient (Beckman market quartz centrifuge tubes, which may be spun up to about 400,00Og, in which the distribution of the separated fractions can be monitored in situ). The geometry of the flow cell is most important. Ideally, a tubular cell should be used. Gradient inversion which may cause mixing and turbulence in the flow cell must be avoided at all costs. As a second choice, a cell allowing liquid to enter at the bottom and leave at the top is to be preferred. If the geometry of the spectrophotometer demands that liquid should enter and leave at the top of the cell there should be a short ‘upwards’ section leading to the light path. The least favorable design is the simple ‘U’ tube. Readers are strongly recommended to check the geometry of flow cells with manufacturers before they order.
When monitoring a gradient, one must observe certain precautions if the separation achieved in the rotor is not to be spoilt during measurement. Firstly, mixing in the flow lines should be minimized by keeping them as short and narrow as possible. Secondly, gradient inversions must be avoided. For example, if the gradient is being fractionated by upward displacement (i.e. light end first) the flow lines should run upwards all the way to the flow cell. It is for this reason that upward displacement is preferable when monitoring, for the whole liquid stream must move upwards through the flow cell air bubbles are to be effectively removed. One cannot emphasize too strongly that air bubbles are the major cause of trouble in monitoring liquid effluents and every care should be taken to avoid their entering the liquid stream. If the gradient is displaced dense end first, there is a tendency for the dense liquid to enter the cell and remain as a puddle in the base of the cell so interfering with the measurement of later displaced portions of the gradient. This is not so important with sucrose gradients, where the viscosity of the solutions ensures efficient clearance but can cause major degradation of separations when low viscosity salt gradients are employed.
Separation of viruses
Continuous-flow rotors must be used for the separation of large volumes of liquid, batch rotors may be very useful for trial experiments involving only a few liters or less of suspension. A great deal of attention was given to the problem of separating viruses. A large number of viruses have a combination of size and density which is distinct from that of any subcellular particles, hence they fall within what they call the virus window of an S-p diagram. Thus it should be possible to separate viruses by rate-zonal centrifugation. It has been found that many viruses can be banded on CsCl gradients and sucrose gradients. Potassium tartrate has been proposed as an alternative to CsCl for the fractionation of denser viruses, but the pH of tartrate solutions changes markedly with the concentration. All these ionic gradients may, however, damage viruses during separation and hence sucrose, when it can be used, is probably the safest as well as the cheapest solute. When considering which gradient material to use, one should bear in mind that viruses, like other nucleoprotein particles, a band at lower densities in gradients of non-ionic solutes such as sucrose than in CsCl. For example, a virus that bands in CsCl at a density of 1.38 bands on sucrose gradients at a density of 1.20.
As the virus is separated from cell debris on the basis of banding density, the size of the sample does not affect resolution. Thus Leach separated influenza virus from 1 1 of culture medium in a single centrifugation step using a B-XV zonal rotor. The suspension containing the virus was pumped into the rotor and followed by a buffer layer of 20% (w/v) sucrose and a steep gradient extending from 20% to 60% (w/v) sucrose. After centrifugation, the virus was found in a band only 20-40 ml wide. When the sample region contained a large concentration of protein, this tended to contaminate the virus band due to sedimentation in droplets. This contaminating protein could be removed by taking the virus-rich region, increasing its density by the addition of further sucrose and layering it under a second density gradient; upon centrifugation, the virus particles will float up and so separate from the protein. Alternatively, the sucrose may be dialyzed out and the virus purified by rate sedimentation.
The major problem in separating live viruses probably lies not in the separation techniques themselves, but in the precautions which must be taken to prevent the escape of potentially dangerous material. It should be realized that aerosols tend to form at the seals of zonal rotors and that particles in such aerosols may travel a considerable distance.
The most usual method for fractionating RNA is by rate- zonal centrifugation on sucrose gradients, although, as discussed earlier better analytical separations are achieved by gel electrophoresis. Nevertheless, density gradient centrifugation is still a convenient method for analyzing radio-actively-labeled RNA, as it is much easier to fractionate liquid gradients than polyacrylamide gels. Problems do, however, arise in the counting of 3H in the presence of appreciable amounts of sucrose.
For special purposes, media other than sucrose may be used. A complex density gradient in which a gradient of phenol overlies a sucrose gradient has been proposed by Hastings for the direct analysis of very small amounts of RNA (and DNA). The cells are lysed and the nucleic acids are deproteinized in the phenol-containing layer. This may sometimes be overcome by centrifuging at a higher temperature than normal or by using conditions in which the RNA is totally denatured. Gradients of dimethyl sulfoxide and hexa-deutero- dimethylsulfoxide and of methanol/ methoxyethanol have been used for this purpose. The rate of sedimentation in such gradients is directly proportional to molecular weight, but very odd patterns may be obtained due to the revelation of ‘hidden breaks’ which exist naturally in the chains of high molecular weight RNA
Polyacrylamide gel electrophoresis is replacing gradient centrifugation as the method of choice for analyzing RNA but rate zonal centrifugation is still useful for preparing pure RNA fractions. Preparative gel electrophoresis can also be used, so can column chromatography, but zonal rotors are useful when reasonable amounts of even minor components are to be separate. Cs2SO4 must be used as the density gradient solute since RNA pellets through saturated CsCl.
DNA is normally separated on the basis of its density, which reflects its base composition, but rate separation may be useful in separating viral DNA or, with more difficulty, bacterial DNA. In the latter case, great care must be exercised if the DNA is not to be degraded; the bacteria must be lysed directly onto the surface of the gradient, for example by preparing protoplasts and treating them in situ with sodium dodecyl sulfate. This detergent should be included in the gradient to dissociate DNA and protein. When the gradient is fractionated one must avoid the use of narrow flow tubes and the displacement should be carried out slowly to minimise shearing. It is not possible to maintain mammalian DNA molecules intact, but rate sedimentation in alkaline sucrose gradients has been used for the isolation of fragments of sheared DNA of homogeneous size for hybridization experiments. An interesting feature of these experiments was the use of very shallow (5-1 1 % w/v) sucrose gradients and of long centrifugation (up to 52 hr) to separate small molecules.
Fractionation of ribonucleoprotein particles
The fractionation of polysomes and of ribosome subunits by rate sedimentation on sucrose gradients was among the earliest applications of the technique. Separations may be carried out either on swing-out or zonal rotors. The separation of polysomes, in particular, is one of the best tests for technique in density gradient centrifugation, as polysomes form a nicely graded series in which each particle is slightly more similar in size to the next largest than it is to the next smallest. Thus, trimers (1 54 S) are 25% larger than dimers (123 S) but tetramers (183 S) are only 18% larger than trimers and so on. Polysomes up to the 12-mer have been resolved as separate peaks but most workers are satisfied if they can separate up to the 7-mer or 8-mer. Density gradient centrifugation can also be used to reveal slight differences between newly formed small ribosome subunits and ribosome subunits recycled from polysomes and to fractionate ribonucleoprotein particles extracted from the nucleoplasm or the nucleolus.