Team:MichiganState/Results
Results
Bioinformatics
Growth of Pseudomonas putida EM371 With Various Concentrations of Imidacloprid
From our literature review, Pseudomonas putida EM371 is expected to grow with imidacloprid (IMI). We grew P. putida EM371 with increasing concentrations of IMI to determine the concentration when IMI becomes toxic to P. putida EM371. P. putida EM371 with 8.8μM, 17.6μM, 23.5μM, and 35.2μM IMI in LB media grows at the same rate as P.putida EM371 without IMI. The growth of P. putida EM371 drops by an OD600 of 0.1 when inoculated in LB media with 70.4μM IMI. The growth is reduced, but not completely inhibited.
Transformation of Imidacloprid by P. putida EM371
From our literature review, we found that imidacloprid can be metabolized by three different pathways.
To determine if P. putida EM371 metabolizes IMI in any way, we conducted LC-MS/MS and uploaded the data to GNPS. GNPS identified the presence of three compounds from the above diagram: imidacloprid (IUPAC name: (NE)-N-[1-[(6-chloropyridin-3-yl)methyl]imidazolidin-2-ylidene]nitramide), imidacloprid-guanidine (IUPAC name: 1-[(6-chloropyridin-3-yl)methyl]-4,5-dihydroimidazol-2-amine), and imidacloprid-urea (IUPAC name: 1-[(6-chloropyridin-3-yl)methyl]imidazolidin-2-one). The xanthine dehydrogenase, a protein in the aldehyde oxidase family, in P. putida Z-4 has been associated with the nitro reduction pathway[1]. Therefore, it is highly suspected that P. putida EM371 uses the nitro-reduction pathway for metabolizing imidacloprid. RT-qPCR on xanthine dehydrogenase could verify this claim.
Nitroso-imidacloprid and 6-CNA were not detected. Nitroso-imidacloprid and 6-CNA are possibly short lived intermediates. Metabolites from the hydroxylation pathway were not detected. Therefore, the hydroxylation is most likely not used by P. putida EM371 under these conditions.
Device
The final results for the device team resulted from the synthesis of hardware and software components of our research: the physical design of a 3D printed bee feeder, and the development of image recognition software to control feeder access.
Image recognition
For image recognition, we needed to use an open-source image recognition API that was available. This is where the Google Vision API would come in. The Google Vision API allows for both remote and local processing of images and comes with a variety of vision capabilities. It is able to provide confidence intervals of object recognition and label recognition for images. Additionally, the vision API allows for both local and remote image processing.
In Figure 3, the initial approach to image recognition would be done locally on the Raspberry Pi. The idea was to take a picture, process the image on the Raspberry Pi and return labels and their confidence intervals, and from the list of labels and confidence intervals, decide whether or not there would be a motor output. The vision API was tested on a personal laptop to experiment and test for desired outputs. We found that the best way to increase confidence intervals and to reduce unnecessary labels outputted was to have a solid color background to reduce the amount of noise in the image. The benefit of this method is that it allows the user to explore programming and the Raspberry Pi features in a simple manner. The immediate problem was that a network connection would be required at all times. Another problem is that all the data would be stored locally, where in the event of an accident, data would be irretrievable. This is where Figure 4 is introduced.
The behavior is similar to Figure 3, but data is sent to the private web server and the server would return desired values, reducing the amount of data stored locally on the Raspberry Pi. This method is by far more challenging to learn for those who have not learned to use a Raspberry Pi for web hosting. However, it relieves the need for unnecessary packages to be installed and can simplify the amount of storage needed to process an image.
Unfortunately, the team was not able to create a finalized private web server with the Google Vision API implemented to carry out the desired behavior. However, we hope future teams can improve this approach or apply image recognition in a similar fashion.
3D Printed Feeder
The Device subteam designed a bee feeder that could be used to selectively administer the Bee-tox probiotic. This device was designed to be attractive to bees, while also featuring image recognition technology and motor components that would stop other pests from accessing the probiotic. After significant research detailed in our Project Design page, we successfully created our feeder. The motor cover and feeder were printed in resin.
This final feeder prototype includes 4 feeding sites with diameters of 3mm.External dimensions of this feeder are 10cm x 10cm x 6.5 cm. The filling and drainage holes are both 0.6 cm in diameter. The internal tank dimensions are 9.5 cm x 2.9 cm x 9.5 cm, leading to an approximate maximum volume of over 260 cm^3. The feeder and motor cover .stl files to download can be found on our Contribution Page.
Final device with motor components
The complete device includes an outer acrylic cover to restrict access to potential predators while allowing entry for bees and smaller insects. It also includes a sliding cover over the feeding sites to restrict access to the probiotic for non-target organisms such as agricultural pests. The sliding cover is designed to open only when the image recognition system detects a bee present, and to close only when the bee is no longer close to the feeding sites.
The device was able to hold and dispense water successfully. In future iterations, however, we would modify the design to have a tapered channel that would reduce air bubble formation near the feeding holes.
Procedure for testing device
As mentioned in our Project Design page, our initial plan was to conduct experimental trials to determine the ideal color, feeding angle, and other feeder parameters that would optimize bee attraction. Unfortunately, due to several setbacks in the approval process, both through iGEM and our university, there was not enough time to conduct these trials. For future testing, we would use the procedure outlined in our Experiments Page for the Device team.
References
- Lu, T. Q., Mao, S. Y., Sun, S. L., Yang, W. L., Ge, F., & Dai, Y. J. (2016). Regulation of hydroxylation and nitroreduction pathways during metabolism of the neonicotinoid insecticide imidacloprid by Pseudomonas putida. Journal of agricultural and food chemistry, 64(24), 4866-4875.