3.1. In vitro imaging of bacterial colonies
To assess the sensitivity of the system to image fluorescent bacteria, colonies of bacteria expressing tdTomato were imaged in vitro.
shows an image of bacteria expressing tdTomato. Based on the signal and resolution, the image indicates that bacterial colonies expressing tdTomato can be detected and resolved well in vitro. Fluorescent protein colonies display strong fluorescent signal in comparison to the control that carries vector alone . Based on these results, tdTomato was used as the fluorescent protein expressed by bacteria for subcutaneous and intra-tracheal infection experiments.
Bacterial colonies grown on agar (a) expressing tdTomato, and (b) negative control (BCG carrying the vector backbone).
3.2. In situ imaging of subcutaneous infection
Imaging bacteria subcutaneously inoculated beneath the skin of mice provides controlled conditions and ease of access for characterizing the sensitivity of this system in mammalian tissue. Subcutaneously inoculated fluorescent bacteria expressing tdTomato at concentrations ranging from 106 to 101 CFU were imaged in situ. An image demonstrating greater than average signal within the 106 CFU inoculated region is shown in . Representative images of 106, 105, and 104 CFU sites, with average fluorescence signal within one standard deviation of the mean fluorescence for each inoculum, are shown in , respectively. Similarly, a representative image of the negative control is shown in . The corresponding macroscopic image taken with the IVIS Spectrum is shown in . Similarly, representative images of 103, 102, and 101 CFU sites are shown in , respectively, and an image of the negative control is shown in . Bacterial infection foci are visible at concentrations of fluorescent bacteria from 106 to 104 CFU; whereas, the negative control displays only background autofluorescence. These images are consistent with data from the macroscopic system image . Tissues that had been inoculated with less than 105 CFU did not have visible signal when imaged with the IVIS system. For bacterial concentrations lower than 104 CFU, infection foci appear to be visible down to 102 CFU in microendoscopic images; however it is unclear whether the fluorescence is from bacteria or auto-fluorescence, especially at low inocula.
Due to the heterogeneous distribution of bacteria within the region of inoculation, there is significant variability in signal intensity depending on where the 750 µm field of view imaging probe is placed. This is demonstrated by the variation in intensity between and , two imaging sites in the region of 106 CFU subcutaneous infection. In addition, there are imaging locations in close proximity to the site of inoculation that have no detectable bacterial signal. In order to average out this variability, 20 images were acquired randomly for each 1 cm region surrounding an inoculation site. It is also important to note that the actual number of bacteria detected within the fiber bundle’s field of view is significantly less than the actual inoculum, since the probe’s field of view is much smaller than the region of inoculation and infection.
To assess the increase in fluorescence signal with increasing inoculum dose and variation within the signal intensities of each individual inoculum, the fluorescence intensity of each image obtained for the skin samples were averaged over the field of view. The mean of the average fluorescence signals for all images in each inoculum is plotted in a bar graph in
. After removal of samples that did not have statistically significant signal, the mean fluorescence intensity was plotted versus CFU for each inoculum in .
Subcutaneous infection. (a) Bar plot of average fluorescence intensity for each subcutaneous bacterial inoculum. (b) Regression fit for average fluorescence intensity versus CFU levels determined from bacterial inoculum titer.
An increasing trend is seen in the average fluorescence intensity with increasing inocula levels. A Student’s t-test reveals 104 CFU to be the lowest inoculum with a statistically significant mean as compared to the negative control (p-value = 0.0176) as shown with a single asterisk in . The 105 and 106 CFU show much lower p-values (0.005 and 0.0004, respectively) as compared to the negative control. According to the statistical analysis, we do not find a significant signal for inocula lower than 104 CFU; however, what appear to be a small number of bacteria in the field of view for lower inocula can be visualized. It is possible that signal from bacteria at these lower concentrations is masked by tissue autofluorescence. Therefore, this system may be able to qualitatively detect regions of infection for 103 and 102 CFU, but quantitative assessment using our current image analysis is only possible at higher inocula. Consequently, our limit of detection of subcutaneous infection of fluorescent bacteria using this microendoscope is 104 CFU, in comparison to a detection limit of 105 CFU for the same sample using the IVIS.
Variation in fluorescence signal increases with increasing inoculum, as indicated by the increasing size of the error bars in . This variation is visible in different imaging sites from the same inoculum dose that display varying sizes of tissue infection foci, as is seen in and . Due to heterogeneity in the local distribution of bacteria and the small field of view, variation in numbers of bacteria within the imaging area are expected, but it should be possible to control for this variation by sampling more tissue sites.
A linear regression fit to the average fluorescence values from positive imaging sites for each inoculum versus CFU levels obtained from the inocula titer  has an increasing trend with an R2 value of 0.9953 and a slope p-value of 0.0434. The regression line has a slope of 0.00022 ± 0.00012. There is a linear relationship between the average fluorescence intensities for each inoculum and bacterial CFU. These observations indicate that this imaging system may be capable of measuring bacterial numbers directly in mammalian tissues.
3.3. In situ imaging of intra-tracheally infected lungs
The lungs of the mice inoculated with BCG at various concentrations were excised and imaged with the microendoscope system. Representative images for 108, 107, and 106 CFU sites, with an average fluorescence signal within one standard deviation of the mean fluorescence, are shown in . A representative image of the negative control is shown in . These images demonstrate that regions of fluorescent bacterial infection can be detected in the lung tissue. An increasing trend in fluorescence intensity is visible in the images with inocula ranging from 106 to 108 CFU. The corresponding macroscopic fluorescence images are shown in
. Bifurcation points in the trachea are seen to have high levels of fluorescent bacteria, but bacteria are also more homogeneously distributed in deeper regions within the lobes of the lung.
Fig. 8 Macroscopic IVIS images of lungs intra-tracheally inoculated with (a) 108, (b) 107, and (c) 106 CFU of bacteria expressing tdTomato, and (d) 106 CFU of non-fluorescent vector backbone (negative control). Field of view is 6.5 cm and scale bar units are (more ...)
The average fluorescence intensity for each intra-tracheal dose is plotted in . An inoculum of 107 CFU produced a statistically significant signal in the lung (p-value = 0.0048) as compared to the negative control. The higher detection limit for bacteria in the lungs is most likely due to the greater distribution of bacteria in the lung in comparison to a more localized subcutaneous inoculation. Following an intra-tracheal inoculation, bacteria are distributed throughout the lung tissue, making localization difficult and offering an important potential application for microendoscopic systems.
Prior to performing a correlation analysis, images within one standard deviation of the average intensity of the negative control were removed from the data set for each inocula. Although the average signal of all the images acquired from the 106 CFU inoculum site was not found to be statistically significant, removal of negative images resulted in analysis of images with statistically significant signal compared to the negative control (p-value = 0.0026). The regression fit was performed for data between 106 and 108 CFU. An increasing trend in average fluorescence signal of positive sites for each inoculum versus CFU levels is seen in lung tissue . The regression fit with an R2 value of 0.9998 and a slope p-value of 0.0094 indicate a linear relationship between the average signal for each inoculum and CFU levels. The slope of the regression line has a value of 0.0000061 ± 0.0000017. These results demonstrate potential for use of this system to image and measure bacterial numbers during pulmonary infection for intra-tracheal inocula down to 106 CFU. Improvements in quantification of signal and quantification of bacteria in small tissue samples may enable measurement of bacterial numbers in individual images.
3.4. Confocal microscopy of in vitro bacterial samples and punch biopsies of intra-tracheally infected lung tissue
Imaging in vitro bacterial smear samples and sections from punch biopsies of infected lung tissue was performed with a confocal microscope to provide a high resolution image reference. In vitro samples of tdTomato expressing bacteria were prepared by smearing bacteria on glass slides and imaging with a confocal microscope. The results in
show both isolated and clumps of fluorescent bacteria, 4-6 µm in length, in the TRITC (red) channel, while no fluorescence was observed for the negative control in .
Confocal microscope images of in vitro smear samples prepared using (a) tdTomato expressing BCG bacteria and (b) non-fluorescent BCG bacteria
Confocal images of sections of infected lung punch biopsies are shown in
. The nuclei of cells in lung tissue were counter stained with DAPI for ease of visualization. Images of tdTomato expressing bacteria in the TRITC and DAPI + TRITC channels in and reveal fluorescent bacteria lodged within tissue. Bacterial fluorescence is absent in the TRITC and DAPI + TRITC channels for images of the non-fluorescent negative control in .
Fig. 11 Confocal microscope images of histology slides prepared using punch biopsies of lung tissue infected with (a), (b), and (c): tdTomato expressing bacteria; (d), (e), and (f): non-fluorescent BCG bacteria. (a) and (d): DAPI (blue) channel; (b) and (e): (more ...)