Case A: Pesticides and viruses – multiplex PCR and BLAST

Case B: Mites and virus diversity – sequence analysis and tree-building

Case C: Compare relative amounts of viral DNA in honey bee hives in relation to mite loads – qPCR

Case D: Determine absolute quantities of DNA from samples via standard curves – qPCR

Case A: Pesticides and viruses

Background: Honey bees are commonly exposed to pesticides as they forage for pollen and nectar. Some pesticides are known to affect the central nervous system of bees and thus impact their behavior. Sub-lethal exposures of some pesticides are considered possible contributing factors to to the decline in honey bee health.

Scenario: Dr. Muskiver was curious if pesticide exposure was linked to virus infection, another possible contributing factor to decreased honey bee health. To test this question, Dr. Muskiver set up test colonies, and fed the honey bees either with untreated pollen or pollen treated with sub-lethal doses of pesticides. She then tested the bees for the presence of several viruses using multiplex PCR on cDNA isolated from the bees.

DNA samples:

• Negative control – bee sample with no viruses present
• Positive control – bee sample containing all four viruses
• Hive 1 – exposed to pesticides
• Hive 2 – exposed to pesticides
• Hive 3 – no pesticides exposure
• Hive 4 – no pesticide exposure

Primers	PCR product size
Actin	120 bp
Deformed wing virus	203 bp
Black queen cell virus (BQCV)	322 bp
Sac brood virus (SBV)	487 bp
Israeli acute paralysis virus (IAPV)	719 bp

Procedure: To analyze this case, open the DNA sequences and multiplex primer, and run multiplex PCR. Then load and run the gel. Click on fragments and BLAST the associated sequence to verify that the fragments are correctly associated with the viruses.

1. Do the control samples produce the results you expected?
   
   Yes, the positive and negative controls appeared to work properly.
   
   
2. What are the results for each experimental hive, in terms of viruses that are detected?
   
   The pesticide treated-hives (4 and 5) had 4 and 2 of the viruses present, respectively, whereas the no-pesticide treated hives (6 and 7) had 3 and 2 viruses present, respectively.
   
   
3. Is there any correlation between pesticide exposure and viruses detected?
   
   Sample size would need to be increased to see if statistical correlation present.
   
   
4. How would you explain these results to Dr. Muskiver?
   
   There may be a correlation, but more data needs to be collected to take into account natural variability in the results.
   
   
5. What changes would you make to the experiment design if this is repeated?
   
   The case description does not indicate if the ‘test colonies’ were set up so as to isolate all factors other than the ones of interest. Are levels of pesticide sufficient to cause increase in viral infection, even if a correlation is potentially present?
   
   
6. What would you suggest that these researchers do next?
   
   Redesign the experiment to increase sample size and control extraneous variables.
   
   
7. What are some other tests that could be done to address this question?
   
   Vary levels of pesticide exposure, use bees from different geographical areas, determine if bees are in suboptimal health because of other factors, use bees of different genotypes, etc.
   
   

Case B: Mites and virus diversity

Background: Recent declines in honey bee populations have given rise to the syndrome named Colony Collapse Disorder (CCD). Several potential stressors have been identified. It has recently been reported that V. destructor  transmits certain strains of DWV more effectively, and that long-term mite infection reduces virus diversity and leads to the prevalence of more pathogenic viruses.

Scenario: A team of research scientists, funded by the North American Honey Bee Council, decide to survey colonies from around North America for two of the notable stressors – Deformed Wing Virus (DWV), a virus that causes wing deformation, and Varroa destructor, a parasitic mite that feeds on the bee. The scientists are interested in testing the relationship between DWV strains and the Varroa mite in North America.

Bees tested from:

• Central Ontario – low mite levels
• Northwestern Washington –  low mite levels
• Southeast Florida – high mite levels
• Oahu, Hawaii – high mite levels
• Northern Arizona –  moderate mite levels
• Southern British Columbia – moderate mite levels

Procedure: The file contains a total of18 sequences, 6 from Florida, 6 from Ontario, and 6 from Washington (Arizona and Hawaii sequences are not included in the file). The tree can be built three ways. If using MEGA software, a single menu command will open MEGA and build the tree via the Analyze button of the Opened & Processed window. If using MABL or MAFFT, then the sequences need to be transferred to the Export field of the Sequence Analysis window (using the Analyze button). The Analyze button can then be used again to opent the MABL or MAFFT web site, after which the contents of the Export field (copied to the clipboard automatically) can be pasted into the input fields of either web site. Instructions for using the web sites are included in the menus of the Analyze button.

Results from MEGA software

Results from MAFFT site

Results from MABL site

Analysis

Why are the three trees different? Or are they? 

Trees are analogous to mobiles that hang from the ceiling – the nodes can rotate without changing the overall relationships among the samples. So the three methods (MAFFT, MABL, and MEGA) would give the same overall results, assuming that the default settings are the same in all three cases.

The hypothesis is that long term mite infection reduces viral diversity. Is that hypothesis supported by the data?

It appears that there is greater diversity in the Ontario and Washington samples (low mite infection) than there is in the Florida samples (high might infection). But care must be used in interpreting the data, for the reason given above. Depending on how the ‘mobile’ rotates, name labels that are in close proximity might appear to be more closely related than name labels that are not. It is important to look at branching patterns to determine the closeness of relationships among the samples.

What else do we need to know before concluding that relatively high mite infections are associated with relatively low viral diversity?

Correlation does not imply cause and effect; this could be a spurious correlation. That is, there could be reasons for lower diversity in Florida that have nothing to do with mite infection rates. We would need to know much more about the experimental design before making any conclusions from these data. It would also be informative to have data from Arizona and Hawaii, as well as data from other regions, before drawing conclusions.

Case C: Comparing relative amounts of viral DNA in honey bee hives in relation to mite loads

Background: Quantitative PCR (qPCR) is a method for determining both relative and absolute quantities of DNA in samples. In this procedure, DNA amplification is monitored over time, as the DNA doubles each cycle. The point at which the amount of DNA present (measured as a fluorescence value) crosses a predetermined threshold is called the ‘Ct’ value (see the qPCR tutorial for a more detailed explanation of Ct and Delta Ct).

Scenario: A local beekeeper is experiencing declines in honey bee production from his hives and has asked biology instructors if their classes can study the problem. This was a ‘single-blind’ study, so the instructors knew which hives had high, moderate and low mite infestations (loads) and good, moderate and poor overwintering success, but the student researchers did not. The students were told that 4 of the hives had high mite loads, 4 had low mite loads, and 2 had moderate mite loads, but they were not told which hives fell into each of these categories.

Procedure: The students determined relative amounts of DWV and BQCV viral levels for the 10 hives using the reverse-transcriptase qPCR procedure (via amplification of cDNA). Resulting data is a file (“qPCR honeybee”) containing fluorescence levels resulting from amplification of cDNA for DWV and BQCV over time. The students were then asked to analyze the data to see if any trends were present. They were also asked to search the literature for known causes of DWV and BQCV, that might offer an explanation for any trends that they found.

Hint: With the Case It software, qPCR can be run for subsets of the data by selecting wells (by clicking or dragging) so that they turn purple in color, and then qPCR can be run for “purple wells only” (or “orange wells only”). This makes it easier to examine the data for any trends that might be present.

Results of data analysis:

A trend is clearly present for DWV at a threshold level of 4051. All four hives with high mite loads (2,3,5,7) had relatively high levels of DWV, compared to the hives with low mite loads (4,6,9,10). The two hives with moderate mite loads (1,8) were not selected (they were colored purple, and the procedure was run for ‘orange wells only’).

No definitive trend is present for BQCV at a threhold level of 4051, since Hive 5 (high mite load) has a relatively low amount of BQCV. Amplification of cDNA for Hive 7 did not cross the threshold, at a threshold level of 4051, hence the question mark on the graph for S7. The four low-mite loads do have relatively low levels of virus, and two of the four high-mite loads are positive delta Ct values.

To determine the delta Ct level for Hive 7, the threshold is reduced to 1038. No trend is present, as Hive 7, with a high mite load, has a relatively low BQCV level.

When the threshold is reduced to 1038 for DWV, the same trend is apparent that was present for a threshold value of 4051 – all four high mite hives (2,3,5,7) still have relatively high levels of DWV, and all four low mite hives (4,6,9,10) still have relatively low levels of DWV.

No trend is apparent for moderate infection hives (1 and 8), since levels of DWV and BQCV are relatively low for hive 1, but relatively high for hive 8.

Questions:

•Are there relationships among mite infestation levels and viral levels in Mr. Smith’s honey bee hives? 

Based on this data set, DWV is associated with high mite loads, but no relationship is apparent for BQCV.

•If there are relationships, are they dependent on the threshold level set before the qPCR procedure is run? 

No, the same trend (or lack thereof) is apparent for both threshold levels selected.

•From the literature, what hypotheses have been advanced about how DWV and BQCV are transmitted? Did the class data support or not support these hypoetheses? 

Other studies have suggested that DWV is spread by V. destructor mites, whereas BQCV is associated with Nosema infection of hives (Nosema is a fungal parasite spead by ingestion of spores in fecal matter).

•What additional information would you need to know before drawing conclusions from the results of this study? How could the experimental design be improved?

 Although the samle size is small, results for DWV agree with other studies suggesting a relationship between DWV and high mite infestations. The students did not determine Nosema levels, so no conclusions can be drawn concerning the impact of this parasite on Mr. Smith’s hives, in terms of BQCV. No information is provided as to how hives were classified as low, moderate or high, which is especially problematic when interpreting results from the ‘moderate’ hives. No information is provided on other factors that might impact results (other possible sources of mortality, pesticide use, etc.)

•What other viruses and environmental factors have been implicated in honey bee declines in the U.S.? How important are honey bees, both ecologically and economically? Are honey bees used commerically in the U.S. native to this country? What is their impact on native bee populations? 

Considerable information is available on the web regarding these and other student-generated questions. The impact on non-native bees on native bees is particularly interesting, as it is less well known then problems associated with non-native species used commercially for pollination purposes.

Using other sets of fluorescence data with the Case It simulation

The qPCR module will work with any set of raw fluorescence data, as long as it is in the proper format. See the file Cases ->qPCR ->”qPCR honeybee.csv” for an example. Contact mark.s.bergland@uwrf.edu with any questions regarding formatting and use of qPCR data with the Case It simulation.

Disclaimer: Case It is designed to teach the basic principles of qPCR. Although the simulation will display accurate curves and Ct values for any set of data, it is not designed to statistically analyze difference among delta Ct values for control versus treatment, etc. For that, use the software that came with the qPCR machine that generated the data.

The qPCR module will work with any set of raw fluorescence data, as long as it is in the proper format. See the file Cases ->qPCR ->”qPCR honeybee.csv” for an example. Contact mark.s.bergland@uwrf.edu with any questions regarding formatting and use of qPCR data with the Case It simulation. Disclaimer: Case It is designed to teach the basic principles of qPCR. Although the simulation will display accurate curves and Ct values for any set of data, it is not designed to statistically analyze difference among delta Ct values for control versus treatment, etc. For that, use the software that came with the qPCR machine that generated the data.

Case D. Determination of absolute quantities of DNA from samples using standard curves

Standard curve data is included in the Cases ->qPCR folder, to demonstrate how standard curves are used to quantify the amount of DNA present in samples. An outlier is deliberately included to show how outliers can influence results. See the qPCR tutorial for more detail on how to generate and use a standard curve. [Note: The standard curve data set was not generated from honey bee viruses, but rather is a generic data set included here to demonstrate the concept.]

Note to instructors: Analysis of outliers is critical to proper interpretation of data sets. The outlier may be due to experimental error, or the outlier may be the most important part of the data set.