An open-source computational tool for measuring bacterial biofilm morphology and growth kinetics upon one-sided exposure to an antimicrobial source |…

The following section describes a series of morphological measurements of B. subtilis macrocolonies response to one-sided CHX exposure. The first subsection details changes related to macrocolony growth and expansion, followed by a second subsection which focuses on GFP signal intensity and additional phenomena.

Macrocolony growth and expansion. (a) Fluorescent images of B. subtilis macrocolony development over a period of 3days. CHX droplet is located horizontally to the right of the macrocolonies in each image, at a distance of 1cm from the macrocolony center. (b) Total coverage area ((upmu)m(^2)) of macrocolonies.

Figure1a shows the original macrocolony images, as obtained by fluorescent microscopy 24, 48 and 72h after initial seeding. On a macro-scale, Fig.1b demonstrates that there is an inverse relationship between the distance of CHX from the seeding point to the expansion rate of the macrocolony. Macrocolonies that were seeded with CHX at 1cm (closest) distance exhibited statistically significant reductions in expansion over all 3days. In contrast, macrocolonies with CHX at 1.5cm were smaller in a statistically significant manner only on day 3, while 2cm macrocolonies did not differ from control on any of the days. Table1 summarizes the relevant p-values (two-sided t-Test).

The morphological changes that occur as a result of CHX proximity can be seen on day 2 and 3-colony periphery on the exposed (i.e., right-hand) side of the macrocolony is notably thinner than that on the unexposed (i.e., left-hand) side (Fig.1a). In order to quantify the morphological changes that occur in B. subtilis macrocolonies as a result of proximity to CHX source during maturation, a series of computational measurements were applied to the images (Fig.2a): firstly, the macrocolony was segmented into an exposed and unexposed sides by a vertical cut through the macrocolony that directly passes through the colony center (i.e., seeding point)the separating line is shown in yellow. For each macrocolony, a binary image was obtained using Otsus thresholding method. For each macrocolony, an outer contour surrounding the entire macrocolony was determined using a border following algorithm applied on binary images from the previous stepthe resulting contour is shown in red. For both the exposed and unexposed sides, a half-contour was mirrored around the separating line. The resulting mirrored contours can be seen in Fig.2a, middle column-top image shows the unexposed side contour, as it was mirrored onto the exposed side, while the bottom image does the same for the contour of the exposed side. Each one of the two contours is then fitted to an ellipse, shown in whiteFig.2a, rightmost column. The semi-major and semi-minor axes of the fitted ellipses were measured.

Illustration of inhibition measurement at the periphery. (a) The macrocolony is divided vertically into unexposed (left) and exposed (right) halves. The CHX spot is horizontal to the right of the macrocolony in each image. Each macrocolony half is separately mirrored and the resulting contour fitted to an ellipse. Red background in leftmost image reflects the Euclidean distance of each pixel from the CHX source. Outer contours are shown in bright red. (b) Colony periphery deformation analysis. At each distance from CHX source (control and 1/1.5/2cm) the ratio between horizontal (left) and vertical (right) radii between the unexposed and exposed halves is shown.

Figure2b demonstrates the differences in morphology that occur between the exposed and unexposed sides, both in the horizontal (left) and vertical (right) planes. The loss of symmetry that occurs in macrocolonies as a result of CHX proximity on day 3 is statistically significant in the horizontal and vertical planes only in macrocolonies with CHX placed at a distance of 1cm. Thus, changes in morphology are directly correlated to the distance from the CHX source. Table2 summarizes the relevant p-values (two-sided t-Test).

Illustration of inhibition measurement at the core. (a) Illustration of inner core segmentation with mirroring and fitting to ellipse. (b) Colony core deformation analysis. At each distance from CHX source (control and 1/1.5/2cm) the ratio between horizontal (left) and vertical (right) radii between the unexposed and exposed halves is shown.

Figure3a illustrates the same image processing pipeline, applied to the colony core, rather than the periphery. Figure3b demonstrates that no comparable changes in morphology occur at the colony core, whether in the horizontal (left) or vertical (right) planes. Indeed, no statistically significant loss of symmetry was observed at the colony core, regardless of distance from CHX source.

Figure3b shows that on day 3, macrocolony core did not differ in a statistically significant manner from the control, regardless of CHX proximity. The colony core is therefore more preserved in structure than colony periphery (or more resistant to CHX). Table3 summarizes the relevant p-values (two-sided t-Test).

Figure 4a illustrates the relevant regions of the macrocolony - the exposed and unexposed (control) periphery and core. Figure4b demonstrates how pixel intensity is affected by proximity to CHX source: average pixel intensity at the exposed/control areas is shown for both periphery (orange) and core (blue) regions on day 3. In other words, for each macrocolony, the ratio between average pixel intensity of the exposed and unexposed halves was calculated and compared at the periphery and core regions. Statistically significant differences in values were found in periphery of macrocolonies that were grown at distance 1cm from CHX, as well as core of macrocolonies that were grown at distance 1.5cm from CHX. Thus, at these distances, the macrocolony is affected both by morphological deformation as well as changes in GFP intensity.

Pixel intensity calculation. (a) Image illustrating the different areas within the macrocolony. CHX source lies directly horizontally to the right. (b) Ratio of intensity average between unexposed and exposed sides of the macrocolony is shown separately for the periphery (orange) and the core (blue). For control images, unexposed and exposed sides were determined via data augmentation as average ratio of left vs. right halves, top vs. bottom halves and a combination of upper left and bottom right quadrants vs. upper right and bottom left quadrants. The shorthand ns indicates non-significant p-value (>0.05).

Figure5a illustrates how distance from CHX is determined for each pixel in the macrocolony. Euclidean distance was used for the calculations. In Fig. 5b, bacterial cells at the leading edge of the macrocolonies are those that are located at the outermost layer of the macrocolony periphery. Due to the curvature of the macrocolony, points along the leading edge are located at varying distances from the CHX source (Fig.5b). In order to characterize the nature of relationship between pixel intensity and distance to CHX, pixel intensities along the leading edge were plotted in Fig.5b: red dots represent pixels along the leading edge of the exposed side of a macrocolony grown at 1cm from CHX, while blue dots represent pixels along the leading edge of the exposed side of a macrocolony grown at 2cm from CHX. Given both sets of pixel intensity values, a linear regression model was applied to bothas can be seen in Fig.5b, there is a linear correlative relationship between Euclidean distances and pixel intensities. This relationship is stronger when CHX is located closer to the macrocolony centerfor example, in the images that are shown in Fig.5b, linear approximation revealed that 1cm macrocolonies are characterized by a slope that is significantly higher (red) than that of the 2cm macrocolonies (blue). This finding signifies the linear relationship between GFP signal intensity of cells located at the leading edge of the macrocolony to their distance from the CHX source.

Linear regression model for pixel intensity at the leading edge as function of Euclidean distance from CHX source. (a) Illustration demonstrating the distance calculation between the CHX source (red dot) to each pixel within the macrocolony. (b) (Top) B. subtilis macrocolonies on day 3 at 1cm (left) and 2cm (right) distances from CHX source. (bottom) Intensities of pixels located at the leading edge (highlighted 20 pixels-wide section from the outer rim) of the exposed half of the macrocolony: red pixels originate from 1cm macrocolony, blue pixels originate from 2cm macrocolony. Linear regression lines demonstrate that at 1cm, pixel intensity is correlated to the distance from CHX source, while no such effect is seen at 2cm macrocolony.

Crescent-shaped morphology occurring at CHX distances of 0.5cm. (a) Top row illustrates the macrocolony morphology over a period of 3daysthe change in morphology appears in the form of crescent-shaped colonies. (b) Expansion comparison with control macrocolonies. (c) Illustration depicting several relevant distances when CHX is placed at 0.5cm-average radius of a mature B. subtilis macrocolony on day 1, average radius of a CHX droplet. (d) Representative image of macrocolony on day 1, with CHX (color corrected for visual clarity) shown to the right.

Bright field images of expanding macrocolonies. (a) Bright field images of expanding B. subtilis macrocolonies grown in proximity to CHX at 1/1.5/2cm. CHX droplet is seen to the right of the macrocolonies. (b) Cross-section of agar substrate seeded with a macrocolony and CHX droplet.

As CHX is placed closer to the macrocolony, it exerts greater inhibitory effect, resulting in increasing deformation of the macrocolony on the side closer to the antimicrobial source. However, in the case when CHX is placed at 0.5cm from the initial point of seeding the macrocolony only develops towards the unexposed sideFig.6a shows the growth of a sample macrocolony over a period of 3days (left-to-right). Starting from day 1, the macrocolony appears to grow only on the side opposite CHX location. On average, control macrocolonies expand on day 1 to a radius of 0.3cm. CHX droplet is on average 0.2cm in radius. Hence, even when CHX is placed at a distance of 0.5cm, the macrocolonies have enough potential space to expand to 0.3cm. However, Fig.6d demonstrates that despite the fact that there is sufficient unoccupied space in front of the macrocolony to expand into (indeed, equal to that required by control macrocolonies which are uninhibited by CHX), the macrocolony does not expand towards the exposed side at all. Rather, it expands towards the opposite side and consequently assumes a unique crescent shape starting from day 1 onwards.

Figure7a shows bright field images of B. subtilis macrocolonies, with CHX droplets seen to their right. This visualization reveals a bright formation in the agar substrate, between macrocolony and CHX, undetected in the fluorescent images. This structure is embedded into the agar throughout its entire width, as seen in Fig.7b. Over a period of 3days, its shape changes from concave to convex, with it seemingly engulfing the CHX droplet. More interestingly, the appearance of the agar at both sides of the formation is uneven, best visualized in Fig.7b, where the agar on the CHX side appears muddy , unlike the one on the macrocolony side.

Read this article:
An open-source computational tool for measuring bacterial biofilm morphology and growth kinetics upon one-sided exposure to an antimicrobial source |...