Introduction

  Our original research objective was to investigate if overexpressing or under-expressing specific genes would improve the memory of D. melanogaster presenting with Alzheimer's Disease (AD), a condition known for decreasing memory capabilities in individuals. However, we had some difficulty creating a well-balanced Y-maze, which is essential for accuracy in our testing. We ended up spending most of the semester investigating how can we change the Y-maze (assay for testing memory) in order to have it balanced to where there is a 50/50 chance the flies will go into each side of the Y-maze. In addition to testing Y-mazes, we analyzed previously published data to find genes that are known to be altered by AD and setup crosses to get progeny that express AD in neurons. This will enable us to begin testing with genetically altered flies as soon as we can get well-balanced assays.

Experiments:

The goal of the first run of our experiments was to demonstrate that flies will choose from two scents indiscriminately, thus giving us a rough average of 50/50 preference for each scent with minimal deviation.

First Set of Experiments:

We first implemented the olfactory test using a y-maze setup, which consists of a starting tube where the flies are dispensed, two tubes where the scents are diffused, and small connectors piecing the tubes together. The small connectors were made from cut-up pipettes at the base of the bulbous area so that the wider end faced the starting tube and the smaller end faced the scent-diffusing tubes. The pieces were 3-D printed and come out a bit sticky and rough, so we had to sand down some of the edges and use tape for a tighter fit for some of the connecting pieces.

For our first couple of trials, we first knocked out the Drosophila melanogaster (wild-type flies, specifically W118) with carbon dioxide and then separated out 50 uninjured species and swept them into an empty vial. We then transferred the flies into a 1% agar vial after a couple of hours.

We then prepped the training vials by using 0.8 µL of octan-3-ol (OCT) and 0.9 µL of 4-methyl cyclohexanol (MCH) diluted in 1.25 mL of white oil. After the solutions have mixed thoroughly and had been diluted, we placed 20 mL of MCH on filter paper and placed it within one of the tubes of the y-maze. We did the same for OCT and placed it in the second scent-diffusing tube of the y-maze. We let the odor diffuse for about 20 minutes. 

We then placed the flies into the starting tube and reattached it to the y-maze configuration. We placed the y-maze horizontally (laying down) to reduce the expenditure of the flies which would allow more flies to make a decision. After 24 hours, we counted the number of flies in each tube to see the distribution of choices.

Second Set of Experiments:

We were not seeing the desired results, so we switched up various aspects of our y-maze experiment design. We switched the scent tubes to shorter tubes in order to make it easier for the flies to make a decision. We also decided to set the y-maze upright, so the flies had to climb to make their decision because the flies have a tendency to climb against gravity. We also changed the amount of time from 24 hours for the flies to make a decision to 16 hours.

Data and Coding:

The main goal of our data and coding is to find genes in drosophila with human orthologs that are either underexpressed or overexpressed in Alzheimer's Disease (AD). To start, we downloaded a list of human orthologs from the Biomart database. We then found the overlap* between human orthologs and genes expressed in the plasma membrane and genes expressed in the extracellular region using Flybase.org**. We combined the two lists*. Next, we found significant expression data from fly neurons in healthy flies. We determined the expression of a gene to be significant if the expression factor was 140 or greater (see code here). We then found the overlap* between human orthologs in the extracellular and plasma membrane regions and expression in neurons. We took the resulting overlap and compared it with the genes that were altered in AD (as listed in this article). The result was a 19 gene long list of neuronally-expressed human orthologs in the plasma membrane and extracellular region in drosophila that are altered in AD. We determined which genes were overexpressed and underexpressed by taking the average expression at the 50-day mark. See data and information on genes below:

*Merging and Overlap code

** FBgn to Gene IDs : flybase.org exports genes as FBgns. We used this site to convert them to standard gene IDs.

Fly Stocks of Interest:

We also set up crosses for different fly types in order to express Alzheimer's in all of the neurons cells. We made two crosses of specific fly types for control and test groups. Below are the specific crosses that we created:

Control: Canton (males) x vglutGal4 (female virgins)

Test: vglutGal4 (female virgins) x UAS-AB42 (males)

We successfully created the crosses we wanted to produce based on the amount of progeny produced. The test group should express Alzheimer's as the UAS-AB42 which is thought to cause some misfolding of proteins and accelerates the amyloid build-up. The control group should be normal flies with normal memories. We were not able to test our control group against the test group due to time constraints and y-maze complications. 

Through our merged data list, we know what genes are of interest in Alzheimer's flies that also correspond to human genes in Alzheimer's patients. We found out if each gene is underexpressed or overexpressed and the corresponding genes we would have to manipulate to balance the overexpression or underexpressed out, which are listed above.

We expected to see the flies choose the scent indiscriminately, thus giving us a rough average of 50/50 preference for each scent, MCH and OCT. We expected this to happen because both scents are attractive scents to the flies so we would expect to see no preference thus resulting in roughly 50/50. 

The actual results we received tended to skew towards one scent randomly, oscillating between MCH and OCT. We did a total of 13 trials, where we only utilized 11 trials for the data analyzed due to the two sets of trials being contaminated. These trials were unusable due to the connectors being blocked by left-over blockage from the 3-D printing thus inhibiting the flies from making a choice for the scents. From our total of 11 trials, we determined an average of 39.02% preference for OCT and a 60.62% preference for MCH with a standard deviation of 22.27%. This technically means that a 50/50 preference is within our standard deviation, but the large standard deviation is a bit worrisome because it means that our error bar is within 22.27% of our data points which does not make our data robust or reliable to conclude a 50/50 preference. 

Details of the more precise data can be seen in the graph and table below, specifically referencing each test time for the y-maze configuration (24 hours vs 16 hours). According to our data, our tests were technically balanced as the standard deviation bar covered the 50 percent mark for each scent preference in each test. The averages expressed in percentages were also close to 50%, but the problem lies in the standard deviation. The standard deviation is very high meaning that the results of the averages are not robust and likely not repeatable. The results must be replicable in order for this to be balanced and useful for testing. This is the main reason why we are switching our y-maze configuration to a t-maze configuration to hopefully reduce the standard deviation whilst also receiving scent averages around 50%.

We were successful in creating the test and control groups we desired but were unable to test their memories through the y-maze. We know that our test and control groups can create progeny and thus create a stable line for future tests. The progeny that we made expressed Alzheimer's disease in our test group and we had normal flies in our control group.

From our results and experiments, the one thing we can conclude is that there was something wrong with our y-maze in the sense that the flies were not showing a 50/50 preference, despite us accounting for different variables that could affect the y-maze. Ultimately we could not find the cause of this error so we can only conclude there is an error with the Y-maze itself. We can also conclude that we were able to make progeny for both our control group and our test groups. 

For future experiments, we will be moving to a T-maze for our memory assay (see design outline below). This is due to the errors we ran into with the y-maze as stated above. This configuration will most likely force the flies to make a defined, clear choice of scent, which was a slight issue in the y-maze as many flies stayed in the starting tube. By switching to a T-maze, we will hopefully be able to recreate the 50/50 preference that typical (non-diseased, wild-type) flies should make. We also hope to reduce our standard deviation by switching to this configuration, to make our results more robust and repeatable.

Our first experiment will be balancing the t-maze in order to set up future experiments. We need to ensure balance with the T-maze in order to test the flies with Alzheimer's against a healthy-fly baseline. We will use the W118 flies to balance, using the same initial techniques we tried with the Y-maze. Once we balance our T-maze, we will be able to actually test our memory assay, in which we train the flies to prefer and specific scent and then test our control group (W118) against the test group (flies with AD and under- or over-expressed target genes).

Group Members: Rachel Becker, Catherine Calma, Michelle English