Chemical address tags of fluorescent bioimaging probes

Abstract

Chemical address tags can be defined as specific structural features shared by a set of bioimaging probes having a predictable influence on cell-associated visual signals obtained from these probes. Here, using a large image dataset acquired with a high content screening instrument, machine vision and cheminformatics analysis have been applied to reveal chemical address tags. With a combinatorial library of fluorescent molecules, fluorescence signal intensity, spectral, and spatial features characterizing each one of the probes' visual signals were extracted from images acquired with the three different excitation and emission channels of the imaging instrument. With multivariate regression, the additive contribution from each one of the different building blocks of the bioimaging probes toward each measured, cell-associated image-based feature was calculated. In this manner, variations in the chemical features of the molecules were associated with the resulting staining patterns, facilitating quantitative, objective analysis of chemical address tags. Hierarchical clustering and paired image-cheminformatics analysis revealed key structure-property relationships amongst many building blocks of the fluorescent molecules. The results point to different chemical modifications of the bioimaging probes that can exert similar (or different) effects on the probes' visual signals. Inspection of the clustered structures suggests intramolecular charge migration or partial charge distribution as potential mechanistic determinants of chemical address tag behavior.

Figure 1

Correlation coefficients between actual and predicted values for the 32 image-based features analyzed in this study, using additive factors for the two styryl components as predictors. The bars represent the correlation value (estimated unbiasedly using cross-validation) for each specific feature.

Figure 2

Array of images acquired from styryl compounds, sorted based on the regression effects of different aldehyde building blocks (rows) and pyridinium/quinolinium building blocks (columns) to the total intensity feature of the styryl molecules. The effect increases from left to right for the pyridinium/quinolinium building blocks, and top to bottom for the aldehyde building blocks. Each image is a color composite of the signal obtained from the Hoechst™ channel (blue), FITC channel (green) and TRITC channel (red). For this figure, the images shown were acquired in the presence of the styryl molecules in the extracellular medium, so some of the images have high background fluorescence. A white, filled square in the upper left hand corner of an image indicates that the image was not used to train the regression model. For display purposes, the images were grouped into an array, and the array of images was then scaled (as a whole) to preserve the relative differences in the staining intensity of each probe.

Figure 3

The calculated, relative contribution of the pyridinium/quinolinium building blocks (A) and the aldehyde building blocks (B) towards the 32 image-based features analyzed in this study. The vertical bars represent the partial R² value, capturing the additional information in A groups not contained in B groups, or vice-versa, based on the regression model. Error bars show the 95% confidence interval for the partial R² value.

Figure 4

Correlation coefficients between the degree of chemical structure variation in the building blocks of the styryl molecules and their contributions towards each image-based feature analyzed in this study. The bars represent the calculated correlation coefficient between Tanimoto structure similarities and the absolute differences in image feature regression coefficients over all pairs of structures, obtained while individually varying the pyridinium/quinolinium (A) or aldehyde building block (B), while keeping the other building block constant. Error bars show the 95% confidence interval for the partial R² value.

Figure 5

The relative effect of a pyridinium vs. quinolinium building block (left box plot in each feature) and a phenyl vs. naphthalene aldehyde building block (right box plot in each feature) on each one of the 23 noncontrol image-based features analyzed in this study. For the left box plots, the difference in each image feature value was calculated for every possible pair of molecules possessing a pyridinium vs. quinolinium building block, with the aldehyde building block held constant. For the right box plots, the pyridinium/quinolinium building block was held constant, and the difference in each image feature value was calculated for all possible phenyl vs. naphthalene aldehyde building block pairs. The boxes extend from the 25^th to the 75^th percentiles of the data, with the central line denoting the median. The “whiskers” lie at the median plus and minus 1.5 times the median. All points outside the whiskers are plotted explicitly with '+' symbols.

Figure 6

The relative effect of a pyridinium/quinolinium building block isomers (left box plot in each feature) and aldehyde building block isomers (right box plot in each feature) on each one of the 23 non-control image-based features analyzed in this study. For the left box plots, the difference in each image feature value was calculated for every pair of molecule possessing a pyridinium/quinolinium building block isomer, with the aldehyde building block held constant. For the right box plots, the pyridinium/quinolinium building block was held constant, and the difference in each image feature value was calculated for every aldehyde building block isomer pair. See the caption to figure 5 for a description of the boxplots.

Figure 7

Hierarchical cluster analysis of the relationship between the regressed contribution of all the styryl building blocks to the 23 noncontrol image-based features analyzed in this study. A) Dendrogram and heat map visualization of the global relationship between different building blocks of the styryl molecules. The columns correspond to different aldehydes (numbers) or pyridinium/quinolinium (letters A–H) building blocks. Clusters I, II, III and IV are highlighted since they contain one or more pyridinium/quinolinium building block (from the dendrogram, the major branch point effectively separates cluster IV from the rest of the styryl building blocks). Colors of the heat map correspond to the regression coefficient values transformed to the unit interval, as indicated in the scale bar at the bottom of the heat map. B) Chemical structures of the selected pyridinium/quinolinium building blocks (letters) and surrounding aldehyde building blocks (numbers) associated with clusters I, II, III and IV.