back to the tutorial main page
For a detailed direct boundary analysis, there are a few requirements that need to be taken into account before doing the experiment.
This tutorial assumes general familiarity with the experimental setup. For detailed information, see the experimental protocols on the website of the PBR/DBEPS at NIH.
One of the most important points is that the rotor is thoroughly equilibrated with respect to temperature in order to avoid convection. (If you have convection, the only rigorous analysis methods possible are the Lamm equation initialized with an experimental scan [single macromolecular component only], or the differential second moment method [gives sw only].) I had most success with the following procedure: After inserting the rotor and the optics into the ultracentrifuge, I set the desired temperature on the control panel and start evacuating the rotor chamber. Note that the temperature reading of the rotor may not be very accurate, and there can be a large jump when the vacuum goes below 100 micron. I do not start the rotor yet. With the rotor sitting at zero rpm, I observe the rotor temperature. If the vacuum is far below 100 micron, and the reading of the rotor temperature coincides with the set-point of the temperature, I wait 1 hour to have good thermal equilibration of the rotor, before I start the run. Please note that there is no 3000 rpm phase, because sedimentation at the low rpm although small, dependent on the system studied, may lead to boundary deformations and propagate systematic error into the data. (This is apparently not as important for methods like g(s*), which primarily use the w2t information from the files, and do not model the shape of the boundary). If an initial phase at low rotor speed is needed (e.g. for setting up the optics), make sure to stop the run, shake the cells up, and restart the run. Unfortunately, usually that increases the rotor temperature by a few degrees, so that thermal equilibration must be reestablished by waiting at zero rotor speed.
I use the rotor acceleration at the maximal setting, as during the acceleration clearly sedimentation takes place (this can be easily observed with the interference optical system). Although Sedfit can calculate the Lamm equation solutions with a uniform acceleration (see Loading Options), it seems best to approximate as good as possible the ideal assumption of instantaneous acceleration to the full speed.
Scanning should be started as soon as possible, because the complete sedimentation process is informative and important for an optimal analysis. In direct boundary modeling, there is no need for the molecules to clear the meniscus. If setting up the parameters in the control software of the optics is complicated, it should be done before the start of the run (and cells should be mixed after that before starting the run). Similarly, the scans should proceed until the sedimentation is complete, i.e. until the smallest species vanishes into the optical artifacts in the bottom region. Unless it is known that the sedimentation is very slow, I usually set the scan rate to the highest possible value. It is not recommended to use Sedfit for the analysis of small data subsets, for example those in the early part of the sedimentation, as this will only have limited information.
Besides these considerations how to do the ultracentrifuge run, there are also requirements for the sample in order to apply c(s) analysis. Because c(s) is based on Lamm equation solutions describing mixtures of independent ideally sedimenting species, the sample should not exhibit non-ideal sedimentation (i.e. concentration-dependent s and D), and should not be in rapid association equilibrium.
An example data set with BSA suspended in phosphate buffered saline is described in the following, and can be downloaded here. Please note that in this particular example, the experiment was not conducted to completion, and ideally one would want to have approximately double the number of scans.
This data was acquired with the interference optics. Most of the following will be the same with absorbance data, with the exception of the few steps regarding the subtraction of systematic noise (although, occasionally, this can be useful for absorbance data, too).