Logic of designing

biochemical experiments

Biochemistry/MCB 568 -- Fall 2007
John W. Little--University of Arizona

The following gives the logic of distinguishing two models of the types given in the problem set. The actual models chosen here differ from the ones you are asked to distinguish.

Many students are not clear about the meaning of an in vitro experiment. The following should help clarify how in vitro experiments differ from in vivo ones.

"in vitro":

The term "in vitro" has different meanings in different areas of science. To a cell biologist working with tissue culture of cells from multi-cellular organisms such as ourselves, it refers to an experiment using tissue culture cells. Sometimes this is called ex vivo. To a biochemist, however, the term "in vitro" means an experiment done in the absence of living cells. This latter usage will be adopted in this course.

Many in vitro biochemical experiments involve the use of purified components, although this is not a necessary condition. Such experiments can involve the use of crude cell extracts, or partially purified components. In any case, the essential feature is the absence of living materials.

A simple in vitro experiment is an enzyme assay, like a beta-galactosidase assay. In its simplest form, such an assay would include three types of components: Buffer and salts (to maintain the proper pH and ionic environment), enzyme, and substrate. When the enzyme is added, it begins to convert the substrate to product. The essential point is that the experimenter has control over the constituents of the incubation.

The experimenter can measure the amount of product formed after a given time, or the rate of its production. How the product is quantified differs with the system. In the beta-galactosidase assay, it's pretty simple: the substrate is colorless and the product is yellow. In more complicated situations, such as an in vitro transcription system, the substrate is a labeled nucleotide and the product is a labeled run-off RNA, which must be quantified by running a gel and counting the amount of label in the product band.

A common type of in vitro experiment we've discussed in class is one designed to detect specific protein-DNA interactions. In such assays, there may be no covalent modification of any component (as in a gel shift assay), or the DNA may be treated lightly with DNase I (as in a footprinting experiment). Consider a gel shift assay: Again, the incubation contains three types of components: buffer and salts, labeled template, and DNA-binding protein. The latter can be highly purified, or it can be a crude extract, or anywhere in between. The point, again, is that the experimenter controls the composition of the incubation.

With this background, consider how to distinguish two models, here termed A and D. The models:

Separate image for printing out the figure:

Model A

1. Now consider the types of experiments used to test the models. Your goal is to distinguish between models A and D. First, let's consider model A. You were asked to use purified components to test this model. What do you think these components are? Answer:

2. What experiments could be done with these components? Let's list the simplest ones first. Answer

3. Call the proteins I and II, and the binding sites IIBS and TBS. II is defined as the one that binds to the target gene. Considering the simplest interaction first, what does the model predict about the DNA-binding properties of II (or the activator)? Answer:

4. What does the model predict about the DNA-binding properties of I (the repressor)? Answer:

5. How would you test this prediction, using a gel shift assay? What is the design of your experiment? Answer:

6. What other predictions does this model make about the biochemical properties of the constituents? For the moment, we are not trying to distinguish it from Model D, but to list its predictions; when we do the same for Model D, we will see which ones actually do distinguish. Let's consider protein I first. What other prediction does the model make about this protein? Answer:

7. How would you detect this? What would be the design of your experiment? Answer:

8. Does the model make any other predictions about I? Answer:

9. Bearing this in mind, what would be the design of your experiment? Answer:

10. What predictions does the model make about the effects of II (or activator) on the in vitro transcription system? Answer:

Model D

11. Now consider Model D. What predictions does it make about DNA-binding properties of the proteins? Answer:

12. And what predictions can you make about the effects of these proteins in the in vitro transcription system? What would be the design of your experiments? Answer:

Model A vs Model D

13. Now we are in a position to distinguish models A from D. It may help you to make a list of the predictions each one makes. How would you distinguish them? Let's start with protein II (which you recall we define as the one that binds to the target gene). Can we distinguish the models on the basis of its properties? Answer:

14. Now let's consider the properties of protein I and its interactions with IIBS (regulatory region of gene II). Can we use these properties to distinguish the models? List all the ways you can think of. Answer:

Mutant Proteins

Often an important set of tests involves the use of mutant proteins. Sometimes these can distinguish models from one another, but even if they don't they are still useful. They help you understand the molecular basis for the mutant protein's in vivo behavior, and they provide further tests of the model, even if there is no competing model that is being ruled out.

15. The last part of the question asked you to consider the properties of dominant mutant proteins. Each model predicts one such mutant: A predicts a repressor that can no longer bind inducer, and is dominant non-inducible; D predicts an activator that no longer needs inducer to bind DNA, and is dominant constitutive. Could you use these mutant proteins to distinguish the models? Answer:

16. What would you predict the biochemical properties of these mutant proteins to be? Answer:

To summarize: The approach is to ask what are the properties of the various components in a system that's as simple as you can make it; then to make it somewhat more complex so that you can ask about molecular interactions. The simplest system here is the DNA-binding protein and the site to which it binds (and, for protein I, the effect of inducer). A more complicated system involves adding RNA polymerase and assessing effects of the protein on transcription.

If you apply the above line of thinking to all the models in the 10/20/03 handout, you should have a clear idea of how to distinguish models on the basis of biochemical tests.

If you had trouble with these concepts before, let me know (jlittle@u.arizona.edu) if you found the present approach helpful.

Back to 568 home page

Answers to Questions above:

Ans. 1: The model involves five components: Two DNA-binding proteins, two sites to which they bind, and inducer, which modulates the binding of one of the proteins. Return

Ans. 4: The model predicts that I will bind to IIBS, but this case is more complicated. The model predicts that it will bind in the absence of inducer, but not in its presence. Return

Ans. 11: Model D predicts that protein I (AI in the model) binds to IIBS, but only in the presence of the inducer. Similarly, protein II binds to TBS, with or without the inducer. Return

Ans. 6: The model predicts that when I (or repressor) binds to IIBS, it will repress transcription of the gene for II. This would be detected using an in vitro transcription system. Return

Ans. 2: The simplest experiments would be ones that ask whether the proteins bind the DNA templates. Return:

Ans. 10: It predicts that it will have no effect on transcription of its own gene, but that it will activate transcription of the target gene. Moreover, this activity should be independent of the inducer. So this can be tested by a set of incubations like the ones used to test the repressor activity. Return

Ans. 8: It predicts that I will repress only in the absence of inducer. If inducer is added to an incubation also containing I, the level of transcription should be the same as if repressor was not present at all. Return

Ans. 5: To test this, you need two incubations: One lacking inducer, and one containing inducer. The prediction is that in the absence of inducer binding will be detected, while in the presence of inducer no binding will be detected. To be rigorous, you might need to run a separate gel in which the gel itself contains inducer for the second incubation (otherwise, the protein and the DNA might associate when they migrate away from the inducer). An alternative approach would be to do footprinting, in which the desired components are present during the footprinting reaction (that is, they can't separate during the time you're detecting the complex). Return

Ans. 9: The ideal design of the experiment would be to have four incubations: One with no repressor, one with repressor, and one with repressor plus inducer. You would also have a control with inducer but no repressor to convince yourself that inducer had no effect on the RNA polymerase. Return

Ans. 14: There are two ways we can distinguish the models:

1. Model A predicts that I binds to IIBS only in the absence of the inducer; D predicts that it binds only in the presence of the inducer (recall you need to test both in the absence and presence of inducer to be able to interpret the result).

2. Model A predicts that when I binds it represses transcription; Model D predicts that when I binds it activates transcription. Return

Ans. 3: The model predicts that II will bind to TBS. This can readily be detected by a gel shift assay, or by footprinting. Return

Ans. 15: No, but the reason isn't what you might think. If you had a mutant like the dominant non-inducible one, model D would already be ruled out; conversely, if you had the dominant constitutive mutant, model A would be ruled out. One way to say this is that each test involves a type of mutant that the other model doesn't predict. Another way is to say that you've already used this distinction in your genetic tests. If you think about it from the point of view of someone actually doing a set of experiments, it isn't possible (for example) to do an experiment with a protein from a dominant constitutive mutant if you haven't isolated such a mutant. Return

Ans. 12: It predicts that protein I will activate transcription of the II gene, but only in the presence of the inducer; it also predicts that protein II will activate transcription of the target gene, with or without inducer. Again, to test these predictions, you would need incubations with and without activator protein, and with and without inducer, so you can make the proper comparisons. Return

Ans. 7: To satisfy yourself that the repressor is repressing transcription, you need two incubations: One containing the repressor, and one without it. If you have only a tube with repressor, you will get a certain level of transcription, but you won't know what to compare it to; for example, you don't know how strong the promoter is, so a particular level of transcription is hard to interpret. So the components of both incubations are: buffer and salts, NTP's, RNA polymerase, and template; one incubation also includes repressor. The prediction is that, if I is a repressor, the latter incubation will give less transcription than the former. Return

Ans. 13: No, both models predict that II binds to the target gene with or without inducer, and that it activates transcription. Return

Ans. 16: Model A: The mutant protein would bind IIBS with or without inducer, and would repress under both conditions.

Model D: The mutant protein would bind IIBS with or without inducer, and would activate under both conditions. Return

Back to 568 home page

Back to Methods page

Biochemistry/MCB 568 -- University of Arizona

Logic of designing

biochemical experiments

Biochemistry/MCB 568 -- Fall 2007 John W. Little--University of Arizona

"in vitro":

Biochemistry/MCB 568 -- Fall 2007
John W. Little--University of Arizona