Abstract
Intravital Multiphoton Microscopy (MPM) is an extremely powerful imaging modality for the investigation of renal pathophysiology in mice[1]. It uniquely allows to assess renal function and morphology at the same time and in the living animal. Following the implementation of an Abdominal Imaging Window (AIW)[2], it is possible to assess the same kidney regions by MPM serially for up to several weeks and to obtain longitudinal information regarding cellular behavior and changes in tissue structure. The conduction of serial MPM studies in live animals is however technically challenging. Kidneys are located in the abdominal cavity immediately below the diaphragmatic muscle and are thus subjected to significant breathing movements, which can create image artefacts. To protect renal tissue from phototoxicity during repeated scanning of the same regions, low laser excitation power is advisable, which however results in low signal to noise (SNR). Lastly, in order to detect structural changes over time in an unbiased manner, individual time points need to be registered to create an overlay of renal structures imaged on individual time points.
Here, we present a workflow for the acquisition and image processing of serial MPM volumetric data of transgenic mouse kidneys, expressing a multicolor fluorescent protein reporter system on endothelial cells (ECs).
A custom 3D-printed animal holder was developed to mechanically decouple the kidney from breathing movements. Data acquisition was performed following the implantation of an abdominal imaging window on low-dose Tamoxifen-induced Cdh5-CreERT2-Confetti mice, which identify endothelial cells by demonstrating 1 out of possible 10 color combinations through the specific co-expression of 2 out of 4 fluorescent proteins (CFP, GFP, YFP, RFP)[3]. To serially image the same fields of view (FOVs) for up to 3 weeks, the surface of the kidney was glued to the AIW and mapped to create an atlas for long-term navigation. A femtosecond laser was used as a micromanipulator to induce a localized thermal ablation in renal ECs. The entire post-acquisition image processing protocol was performed in the FIJI ImageJ distribution[4]. Images were denoised with a 3D Noise2Void deep neural network[5], [6]. The Descriptor-based series registration plugin from Preibisch et al[7] was used to align the single focal planes in each volume. Finally, volumes were registered to the data from the first scanning day with a rigid Euler transform using the FIJI Elastix wrapper[8], to run the external intensity-based Elastix registration software[9].
Our custom 3d-printed animal holder provided a good imaging stabilization while largely removing image artefacts due to breathing. Noise2Void image denoising, significantly increased the SNR of the data while not requiring ground truth “clean” training datasets. The first 2D registration step minimized the effect of residual breathing movements. Rigid 3D registration with Elastix provided a way to accurately align volumes acquired at different time points. The algorithm paired to the use of binary masks proved quite robust when challenged with substantial progressive changes in tissue morphology in response to laser injury.
Thanks to its ability to provide biological information both at sub-cellular as well as to image tissue regions up to several square millimeters, serial intravital microscopy exists at the intersection between “traditional” fluorescence microscopy and medical imaging. MPM data analysis can benefit from the integration of solutions developed for both fields. Our workflow exploits multiple open-source tools running under a common FIJI framework to provide a freely available and fully reproducible approach to study the plasticity of ECs and renal tissue remodeling in response to injury.
Here, we present a workflow for the acquisition and image processing of serial MPM volumetric data of transgenic mouse kidneys, expressing a multicolor fluorescent protein reporter system on endothelial cells (ECs).
A custom 3D-printed animal holder was developed to mechanically decouple the kidney from breathing movements. Data acquisition was performed following the implantation of an abdominal imaging window on low-dose Tamoxifen-induced Cdh5-CreERT2-Confetti mice, which identify endothelial cells by demonstrating 1 out of possible 10 color combinations through the specific co-expression of 2 out of 4 fluorescent proteins (CFP, GFP, YFP, RFP)[3]. To serially image the same fields of view (FOVs) for up to 3 weeks, the surface of the kidney was glued to the AIW and mapped to create an atlas for long-term navigation. A femtosecond laser was used as a micromanipulator to induce a localized thermal ablation in renal ECs. The entire post-acquisition image processing protocol was performed in the FIJI ImageJ distribution[4]. Images were denoised with a 3D Noise2Void deep neural network[5], [6]. The Descriptor-based series registration plugin from Preibisch et al[7] was used to align the single focal planes in each volume. Finally, volumes were registered to the data from the first scanning day with a rigid Euler transform using the FIJI Elastix wrapper[8], to run the external intensity-based Elastix registration software[9].
Our custom 3d-printed animal holder provided a good imaging stabilization while largely removing image artefacts due to breathing. Noise2Void image denoising, significantly increased the SNR of the data while not requiring ground truth “clean” training datasets. The first 2D registration step minimized the effect of residual breathing movements. Rigid 3D registration with Elastix provided a way to accurately align volumes acquired at different time points. The algorithm paired to the use of binary masks proved quite robust when challenged with substantial progressive changes in tissue morphology in response to laser injury.
Thanks to its ability to provide biological information both at sub-cellular as well as to image tissue regions up to several square millimeters, serial intravital microscopy exists at the intersection between “traditional” fluorescence microscopy and medical imaging. MPM data analysis can benefit from the integration of solutions developed for both fields. Our workflow exploits multiple open-source tools running under a common FIJI framework to provide a freely available and fully reproducible approach to study the plasticity of ECs and renal tissue remodeling in response to injury.
| Originalsprog | Engelsk |
|---|---|
| Publikationsdato | 2021 |
| Status | Udgivet - 2021 |
| Begivenhed | Danish Bioimaging Network Meeting 2021 - Copenhagen, Danmark Varighed: 27 okt. 2021 → 28 okt. 2021 https://www.conferencemanager.dk/danish-bioimaging-network-fall-meeting-2021/conference |
Konference
| Konference | Danish Bioimaging Network Meeting 2021 |
|---|---|
| Land/Område | Danmark |
| By | Copenhagen |
| Periode | 27/10/2021 → 28/10/2021 |
| Internetadresse |