Teaching the pathway: does a drug change a protein?

Using the pathway

The learning pathway follows one question in six steps, and step 4 uses the Western blotting guided experiment. It suits students who know what cells, genes and proteins are and are beginning to learn experimental design. Students can work alone or in pairs; pairs tend to discuss each decision, which is the point of the exercise.

• Before the class, work through the Western blotting experiment once, choosing the drug question, so that its stations and messages are familiar.
• The time the pathway takes has not yet been measured with students. The guided experiment can be paused and resumed, because progress is saved in the browser, so the work can be split across sessions.
• Each student’s answers stay in their own browser. Ask students to save the worksheet as a file or print it, and to bring it to the discussion; a completed worksheet is a record of reasoning, not proof of laboratory competence.
• Useful questions to ask during the work: What is n, and why? What would the result look like if the drug did nothing? Which result would make you change your mind? What else could explain a difference?

The worked example below is public, so it cannot keep the answers confidential in a graded assessment. For assessment, the transfer problem or a new scenario of the teacher’s own is more informative than the pathway questions.

A worked example

One good set of answers. Other answers can be equally good if they are justified. The numbers in step 5 are illustrative and do not come from the guided experiment, which generates its own data.

1. The question

Question. Does seven days of treatment with the drug change the amount of Gx protein in the mouse hippocampus, compared with the vehicle?

Prediction. If the drug changes Gx, the mean amount of Gx relative to the loading control will differ between the drug and vehicle groups. The prediction is not supported if the estimated difference is close to zero and its confidence interval is narrow enough to exclude a difference large enough to matter.

2. What to measure

Method. Western blotting, because the question concerns the amount of protein. RT-qPCR would measure Gx mRNA, which need not change with the protein; immunofluorescence shows where Gx is rather than how much of it the tissue contains.

What it cannot show. In which cells Gx changed, its absolute amount, or whether its function or the behaviour of the mice changed.

3. The plan

Unit and number. The mouse is the independent unit, because the treatment is given to each mouse; lanes and repeated measurements of a band are not independent. The number per group should come from a power calculation made before the study; with three mice per group only large differences can be detected.

Controls. The vehicle group; a loading control or total-protein stain, checked for any change with treatment; and evidence that the antibody is specific, such as the absence of the band in tissue from a knockout.

Allocation and blinding. Mice allocated to drug or vehicle at random; tubes coded by a colleague who keeps the key until quantification is finished; lane order randomised across the gel.

Analysis. One value per mouse (Gx relative to the loading control); the groups compared with Welch’s t-test, reporting the difference, its 95% confidence interval and an effect size, as planned before the data are seen.

5. The result

Illustrative result. Gx relative to the loading control was 25% lower in mice given the drug than in mice given the vehicle (95% confidence interval 5% to 45% lower; six mice per group). The interval is wide: the data are compatible with a small reduction or a large one.

Checks. Equal loading confirmed by the total-protein stain, which did not differ between groups; a single band at about 55 kDa, absent from knockout tissue; the strongest bands below saturation in the chosen exposure.

6. The limits

Not established. How the drug lowers Gx (by changing its production or its breakdown, directly or through other effects), in which cells, whether the function of the hippocampus changes, whether the effect occurs at other doses, durations, in both sexes or in other species, and whether the drug would alter any disease.

Next experiment. RT-qPCR for Gx mRNA in the same tissue: if the mRNA is unchanged while the protein falls, the drug more probably acts on the production from the mRNA or the breakdown of the protein; a fall in both points to an effect on transcription or mRNA stability. A replication in an independent cohort would come first if the confidence interval were wide.

Misconceptions

• Counting lanes or repeated measurements as n. Several lanes from one mouse measure the same animal; treating them as independent understates the uncertainty (pseudoreplication).
• Treating a change in mRNA as a change in protein. The two can change differently, because protein levels also depend on translation and degradation.
• Assuming the loading control cannot change. A treatment can alter the protein used as a loading control; it should be checked, or total protein used instead.
• Quantifying saturated bands. Once the signal stops increasing with the amount of protein, differences between strong bands are hidden.
• Reading p > 0.05 as “no effect”. A non-significant result with a wide confidence interval is inconclusive; the interval shows which effects remain compatible with the data.
• Moving from a change in a protein to a mechanism or a treatment. A difference in Gx after seven days does not show that the drug acts on Gx directly, or that it would help any disease.

A transfer problem

A research group asks whether chronic stress changes the amount of the synaptic protein PSD-95 in the prefrontal cortex of rats. They plan to take cortex from three stressed and three control rats, run each sample in three lanes, and compare the groups with n = 9 lanes per group. The blots will be quantified by the person who did the stress procedure.

Students identify the problems and redesign the study: the unit of analysis (the rat, not the lane, so n = 3 per group as planned, which is too few to detect anything but a large difference); the lack of blinding; how animals are allocated to groups; the loading control and its possible change with stress; antibody specificity; and what the result could and could not show about stress and synapses.

Assessment rubric

For formative feedback on a completed worksheet or on the transfer problem. The descriptions are a starting point for a teacher’s own judgement.

Criterion	Not yet	Developing	Secure
Question and prediction	The question is vague or untestable	A testable question and a prediction	Also states which result would count against the prediction
Choice of method	The method does not measure what the question asks	A suitable method, with a partial justification	Justified against the alternatives, including what it cannot show
Design	The unit of analysis is wrong or missing	The correct unit, with some controls, allocation or blinding	Unit, sample size reasoning, controls, random allocation, blinding and a pre-planned analysis
Interpretation	Reports only significance, or ignores the checks	Reports an estimate, or the checks, but not both	An estimate with its uncertainty, after checking loading, specificity and saturation
Limits and next step	Claims more than the data show	Names some limits, or a next experiment	Bounded conclusion and a next experiment whose possible results are stated