Revealing real-time 3D in vivo pathogen dynamics in plants by label-free optical coherence tomography

Understanding plant-pathogen interactions is crucial for developing disease-resistant crops. Traditional microscopic imaging methods often rely on destructive techniques or the generation of transgenic pathogens, limiting in vivo studies. This research aimed to overcome these limitations by developing a label-free imaging method for real-time 3D visualization of pathogen dynamics within plant tissue. The use of label-free imaging, specifically dynamic optical coherence tomography (dOCT), offers several advantages. It eliminates the need for labeling, providing simultaneous structural information and enabling prolonged imaging in the pathogen's native biological state. This approach is applicable to all pathogens, regardless of their transformability, eliminating a significant hurdle faced by current techniques. While conventional OCT offers 3D imaging of plant tissue, distinguishing pathogens from plant tissue due to limited intrinsic contrast remains a challenge. Dynamic OCT, however, utilizes sub-resolution motion inside cells to create cell-specific contrast, providing label-free contrast images based on the spectral content of these fluctuations. This study hypothesized that dOCT would provide high-contrast, label-free in vivo imaging of plant pathogens, enabling detailed analysis of pathogen structures and their interactions with plant tissue.

Current methods for studying plant-pathogen interactions often involve destructive techniques like clearing and staining, or require the generation of transgenic pathogens expressing fluorescent proteins. These methods present limitations. Destructive methods prevent longitudinal in vivo studies. Generating transgenic pathogens is time-consuming, species-specific, and necessitates specialized facilities for genetically modified organisms. Many pathogens, especially biotrophs such as downy mildew, are difficult or impossible to transform. Label-free imaging methods, based on intrinsic scattering or absorption contrast, have been explored as an alternative, offering advantages such as applicability to all pathogens, simultaneous structural information, and the ability to conduct longitudinal studies in the pathogen’s natural state. However, previous label-free imaging methods struggled to visualize small pathogens within highly scattering plant tissue due to limited optical contrast. This study builds on previous work utilizing OCT for plant imaging, which demonstrated the potential for deep tissue imaging, particularly when combined with water infiltration. However, conventional OCT lacked the resolution and contrast necessary to differentiate pathogens from plant tissue. This research leverages advancements in dynamic OCT, which uses sub-resolution motion to generate contrast, to overcome these challenges.

The study used a spectral domain OCT system to generate dynamic OCT (dOCT) images. Leaf disks were infiltrated with water or perfluorodecalin to reduce light scattering. A series of B-scans were acquired at the same location over 1.5 seconds. The temporal fluctuations in OCT amplitude were analyzed using a Fourier transform to obtain the amplitude spectrum. This spectrum was divided into three frequency bands (low, medium, high) which were optimized for visualization of biological tissue. The average amplitude in each band determined the intensity for the blue, green, and red channels, generating a false-color dOCT image. Repeating this process for parallel frames yielded a 3D image. The study used lettuce downy mildew (*Bremia lactucae*) infection in three lettuce genotypes with different resistance levels as a model system. Infection levels were quantified by segmenting the *B. lactucae* hyphae in the 3D dOCT images. Hyphal volume and length were measured and compared with qPCR data for pathogen DNA content. Longitudinal imaging was also performed on *B. lactucae*-infected Salinas lettuce to track pathogen growth over time. The dOCT data were processed in Python using custom algorithms for filtering, denoising, segmentation, and quantification of hyphal structures. Statistical analyses were performed using two-sided Mann-Whitney U tests for comparisons of hyphal volume and length between genotypes, and two-sided t-tests for qPCR data, with P-value correction for multiple testing.

The dOCT method successfully generated high-contrast, label-free 3D images of *B. lactucae* hyphae within lettuce leaf tissue. Different biological features were clearly distinguishable, including hyphae, haustoria, living and dead plant cells, and stomata. The method also successfully imaged other plant-pathogen systems, such as *Hyaloperonospora arabidopsidis* in Arabidopsis and *Meloidogyne incognita* in pepper roots. Quantification of *B. lactucae* infection in three lettuce genotypes revealed significant differences in hyphal volume and length that correlated with the genotypes' resistance levels. These results were consistent with qPCR data measuring pathogen DNA content. Longitudinal imaging revealed the rapid colonization of lettuce leaves by *B. lactucae*, with hyphal growth rates ranging from 10.5 to 54 µm/h. The study observed a dynamic change in the dOCT signal of *B. lactucae* hyphae over time, suggesting potential insights into pathogen physiology and development. The segmentation process for quantifying infection was optimized, with a weighted combination of signals from different frequency bands used to distinguish between plant and pathogen tissues. Manual correction was performed to remove non-pathogenic structures such as veins and stomata. The results demonstrate the accuracy and potential of the dOCT method for quantitative plant pathogen imaging and digital phenotyping.

This study successfully demonstrated the feasibility of using label-free dOCT for high-resolution, in vivo 3D imaging of plant pathogens. The results validate the hypothesis that dOCT can provide high-contrast images of pathogens within plant tissue without the need for labeling or genetic modification. The correlation between dOCT-based quantification of hyphal volume and length with qPCR data suggests the accuracy and reliability of the dOCT method for assessing disease severity. The longitudinal imaging capabilities of dOCT provide valuable insights into the dynamics of pathogen growth and development within the host plant. The ability to image different plant-pathogen systems indicates the broad applicability of this technique for studying various plant diseases. This method has significant potential for applications in plant pathology research, breeding programs for disease-resistant crops, and the development of novel strategies for disease management. The observed changes in dOCT signal intensity over time warrant further investigation, potentially revealing valuable information about pathogen physiology and interactions with the host plant.

This study presents a novel label-free imaging technique using dynamic optical coherence tomography (dOCT) for visualizing plant-pathogen interactions in 3D in vivo. The high-contrast images obtained allow for the quantification of pathogen colonization and the tracking of growth dynamics. The results demonstrate the potential for dOCT to revolutionize the study of plant disease, offering significant advancements over traditional methods. Future research directions include exploring the potential of machine learning for automated image analysis, extending the methodology to a wider range of plant-pathogen systems and employing more sophisticated image analysis techniques to further understand the molecular processes underlying plant-pathogen interactions and resistance mechanisms.

The accuracy of the method depends on the spatial resolution of the dOCT system and the accuracy of image segmentation. The segmentation process, while optimized, still involved some manual steps, which could introduce a degree of subjectivity. The current study focused on a limited number of plant-pathogen combinations; further research is needed to validate the generalizability of the method. Some hyphae may be missed or misidentified during segmentation, potentially underestimating the extent of infection, while occasional artifacts need to be manually excluded. The dynamic range of the dOCT images can slightly influence segmentation accuracy.