Fig 1.
Microbial enumeration using HCA methods.
Example images from SytoBC stained Bradyrhizobium japonicum are presented in panel A. For comparison, images obtained from un-coated, standard tissue culture plates are presented with images obtained from plates coated with Syntrix-25. To demonstrate HCA-enumeration methodology using a specific microbe, we utilized the B. japonicum strain, graphed data in panel B., represent the average of 3 separate samples taken from two separate cryopreserved samples from the same fermentation, error bars indicate the standard deviation. Test 1 viability was measured with PI, for test 2 we used SYTOX Orange to determine viable cells. Panel C. is the graph of average values obtained from 21 different microbial fermentation enumerated via CFU/mL, HCA, or Flow. All values are the mean of triplicate measurements.
Table 1.
B. japonicum enumeration results, traditional CFU/mL assay format compared to high content methods.
Fig 2.
HCA protein synthesis viability determination.
Two-color images of microbes incorporating HPG (A). Total DNA was stained with SYTOX Green and detected in the FITC channel, synthesized protein with incorporated HPG and ligated Alexa Fluor 594 was detected in the Cy3 channel. Arrows indicate cells positive for DNA staining, but negative for HPG incorporation. Three representative data sets demonstrating the ability to separate individual cell populations into viable and non-viable populations based on a control sample without the artificial amino acid HPG compared to HPG-containing samples. Green lines demonstrate the approximate Cy3 intensity value used to determine HPG incorporation as opposed to autofluorescence. Individual object data are represented by single dots, summary data is reported with box and whisker plots (B). Statistical analysis of enumeration results from 52 (largely Bacillus) samples determined by HCA measured cell viability via CLICK chemistry and CFU/mL assays (C).
Fig 3.
Representative images of microbes treated with antibiotics.
Representative images of control and antibiotic-treated microbes stained with the DNA dye SYTOX Green are presented. Images were normalized for veiwing using ImageJ within each genus for comparison, mixed microbe samples were normalized to each other.
Fig 4.
Object feature based strain identification in mixed culture.
Data represented is based on the Bootstrap random forest analysis of the total objects (indicated by N) for each microbe species tested. First, subsamples of data from each species in the un-treated control group were used to train the algorithm. Following training, accuracy of the algorithm was determined by re-analysis of all data from each microbe. Data is presented as the percent of objects identified within each training set, so that correctly identified microbes match the identified sub-group, mis-identified microbes do not. Once we validated the rate of correct identification, the algorithm was applied to the mixed-microbe sample object data set, values returned correspond to the percent objects identified as either B. megaterium, E. coli, or P. fluorescens.