Spatial transcriptomics: mapping brain cellular organization

in collaboration with Dr Alvaro Crevenna, Head of Microscopy, EMBL Rome

Due to their capacity to allow both proteomics and transcriptomics profiling, spatial biology techniques are advancing rapidly as powerful tools to obtain an in-depth understanding of tissues. With the primary goal of creating complete maps of large and complex tissues by studying individual cells, these techniques represent a game-changer in learning about their function. The reconstruction of spatial information is changing the way of addressing unmet challenges in areas like neurobiology, developmental biology and cancer biology.

Spatially resolved transcriptomics studies can be carried out by microscopy through, for example, in situ sequencing or multiplexed fluorescence in situ hybridization. The analysis of different RNA species within the tissue can help us understand cellular heterogeneity, define cell types and provide physiological context to this spatial information. 

Since spatial transcriptomics involves high-throughput and large-scale scanning of the sample, complex multidimensional acquisitions must be performed, including large area stitching, Z-stacks and multichannel imaging.

To support such complex acquisitions, a set-up capable of acquiring the sample in its entirety and in a short time is needed, without having to compromise on the resolution of the image. Large field of view (FOV) microscopes can assist in providing high-resolution imaging with high data throughput.

In this application note, we present a mouse brain tissue analysis of four different RNA species using the CrestOptics X-light V3 spinning disk confocal (CF). Thanks to its seamless stitching capacity and to the great disk pattern flexibility, we recommend our system as particularly well-suited to high-throughput imaging for mapping individual RNA molecules within the brain cellular architecture.

Matching disk pattern to application

While most commercial systems offer limited options in terms of disk patterns, the spinning disk box is one of the most versatile components of the CrestOptics X-light V3 spinning disk CF which offers the freedom to choose the most suitable disk geometry for the application and imaging requirements. In other words, our disk can be customized with different pinhole sizes and distances to provide the best possible balance between pinhole cross-talk and 3D sectioning capability (for more information on how to match the disk pattern to your application, visit our webpage Technology. In this application note, we compare the acquisitions made with two disks with different patterns in terms of imaging speed, confocality and image quality (Figure 1A). In particular, we used a disk with slits (HT-Slits), i.e containing continuous spirals instead of pinholes (Figure 1A, left), designed for samples that require high throughput, high speed and low excitation power. Contextually, we used the standard (ST) disk (Figure 1A, right), containing 50 um diameter pinholes with a spacing of 250 um, which represents the ideal balance between brightness and confocality for a wide range of applications during routine research.

A

B

Figure 1: High throughput Slits disk (HT-Slits) and Standard (ST) disk (A); axial resolution data, expressed in um, measured on 200 nm fluorescent beads.

To check how the use of the two disks affected the axial resolution of the microscope with respect to the WF, we analyzed images of 200 nm fluorescent beads (Figure 1B) with the PSFj software to obtain information about axial resolutions. As shown in Figure 1B, the use of these two patterns allows to reach a great improvement in the axial resolution (600 nm with the ST and 720 nm with the HT-Slits) compared to widefield (WF, 880 nm). At the same time, a good trade-off with the throughput of the system can be obtained with the HT-Slits disk (>13% expressed as CF to WF ratio) with respect to the ST disk (>3.1% expressed as CF to WF ratio), enabling high-speed acquisitions. Depending on the specimen brightness and thickness, it is crucial to match the disk pattern to the sample and, in the following paragraph, we will show how the use of the two disks impacts the level of light collected from the specimen and therefore also the acquisition speed, without neglecting confocality and image quality.

Uniform illumination for seamless stitching of a mouse brain section

Spatial transcriptomics experiments consist of sequential multi-markers imaging on large and complex samples and here we present the study of individual RNA molecules organization within a mouse brain section. With this purpose, automated image acquisitions, consisting of 80 FOVs and 16 Z-planes (0.9 um steps), were performed for five channels (DAPI for nuclei visualizations and four channels for as many RNA species). The whole-brain slice was imaged with a Plan Apo Lambda 20X air objective (NA 0.75, WD 1) and the same type of acquisition was performed with both ST and HT-Slits disk, comparing acquisition times and image quality (Figure 2).

Figure 2: Comparison between HT-Slits disk (top) and ST disk (bottom) acquisitions. Large images of a mouse brain section marked with four different RNA species (green, red, cyan and white dots); single cell nuclei are stained in blue. These images represent intensity projections of a mouse brain slice acquired with a CFI Plan Apo Lambda 20X air objective (NA 0.75, WD 1) in a stack with z-step of 0.9 um and with 18 um of Z range. Scale bar: 500 um.

In both cases, the slice stitching was without artifacts, and this is due to the uniform illumination of the system. The X-light V3 illuminator is based on micro-lens technology which can turn a high-power laser from a multi-mode fiber into a uniform square collimated beam with 90% homogeneous illumination over the entire 25 mm FOV. This peculiar feature allows the periphery of the FOV to be properly illuminated resulting in reliable data without artifacts, avoiding any post-processing correction and increasing speed by minimizing tiles overlap. 

Nevertheless, thanks to the high throughput of the Slits geometry, using the HT-Slits disk it was possible to considerably reduce the exposure time during acquisitions, obtaining a substantial time saving compared to the experiments conducted with the ST disk. As a matter of fact, considering only the acquisition time and the Z-motion, with the HT-Slits disk it took 23 minutes to acquire the entire sample (80 tails, 18 planes, five channels) compared to 52 minutes for the ST disk.

These data demonstrate that for this type of application, the HT-Slits disk represents the right balance between throughput and confocality allowing to obtain very complex images in a short period of time with respect to the ST disk while maintaining excellent image quality. Moreover, if we consider that these experiments were not conducted with a piezo stage or even using the X-light V3 dual-camera configuration, these acquisitions could be further speeded up when leveraging these advanced configurations.

Higher magnification: confocality and image quality

Once a general overview of the entire tissue area was obtained, we moved to higher magnification to appreciate cellular details and investigate RNA dots on individual cells.

In Figure 3 a comparison between WF, HT-Slits and ST disks CF acquisitions is reported. In particular, these images represent a 16 um Z-stack (0.2 um steps) intensity projection of a mouse brain slice acquired with a CFI Plan Apo Lambda 60X oil objective (NA 1.4, WD 0.13) and can easily show a great improvement in terms of resolution obtained with CF acquisitions.

Figure 3: Comparison between WF (left), CF-HT-Slits (middle) and CF-ST acquisitions. These images represent intensity projections of a mouse brain slice acquired with CFI Plan Apo Lambda 60X oil objective (NA 1.4; WD 0.13) in a stack with 0.2 um Z-step and 16 um of Z-range. Scale bar: 50 um.

In Figure 4 we focused on the axial resolution showing XY, XZ and YZ orthogonal views (Figure 4A) and 3D volume views (Figure 4B) of images acquired with a CFI Plan Apo Lambda 60X oil objective in 16 um Z-range (0.2 um steps). To make visualization easier and optical sectioning clearer, we focused on two single cells and one single RNA species. The comparison between WF and the acquisitions made with HT-Slits and ST spinning disks clearly show how it is important the use of a CF technique to image such small and closely spaced dots. In fact, the orthogonal projections and 3D views in Figure 4 show a clear improvement in the Z-sectioning moving from WF to CF modality and similar image quality between the acquisitions taken with the two disks.

A

B

Figure 4: Comparison between WF (left), CF-HT-Slits (middle) and CF-ST (right) acquisitions. XY, XZ and YZ orthogonal views (A) and 3D volume views, 16 um thickness (B), showing different optical sectioning and single cells details. These images were acquired with a CFI Plan Apo Lambda 60X oil objective (NA 1.4, WD 0.13) in a stack with 0.2 um Z-step and with 16.8 um of Z range.

Figure 5 shows a comparison between WF and CF images, focusing on a single cell and a single Z plane (8 um inside the sample). As demonstrated by these images, the use of a CF acquisition modality clearly improves the sectioning in Z; moreover, the intensity profiles related to 3 single RNA dots prove that the use of the HT-Slits disk allows obtaining a similar image quality with respect to that achieved with the ST disk.

Figure 5: Comparison between WF (top-left), CF-HT-Slits (top-middle) and CF-ST (top-right) acquisitions of single cell and a single Z plane, 8 um inside the sample. Scale bar: 5 um. Intensity profile comparison related to three RNA fluorescent dots (bottom). These images were acquired with a CFI Plan Apo Lambda 60X air objective (NA 1.4; WD 0.13) in a stack with 0.2 um Z-step and 16.8 um of Z-range.

Finally, 3D movies of HT-Slits (Figure 6A) and ST (Figure 6B) are reported in Figure 6 showing, again, similar resolution.

A

B

Figure 6: 3D movie comparison of CF-HT-Slits (top) and CF-ST (bottom) volume views. These images were acquired with a CFI Plan Apo Lambda 60X air objective (NA 1.4; WD 0.13) in a stack with 0.2 um Z-step and 16.8 um of Z-range.

Altogether these data demonstrate that there is a significant improvement in the axial resolution using CF approaches. The comparison between HT-Slits and ST disks 60X acquisitions shows that an excellent level, in terms of image quality and confocality, can also be achieved with the HT-Slits. At the same time, the slits pattern allows to obtain a good trade-off with the throughput of the system, meaning that a sufficient level of light can be collected from the specimen, therefore enabling high-speed and high-resolution acquisitions.

In conclusion, the CrestOptics X-light V3 CF spinning disk provides a robust solution for a high-throughput single-cell resolution imaging for spatial biology analysis. Thanks to its homogeneous illumination over the entire 25 mm field of view, it gives the opportunity for seamless stitching of images of very large samples, collecting reliable data without artifacts and increasing speed. Moreover, CrestOptics offers tailor-made solutions with great flexibility in choosing the disk geometry that best suits applications, including complete disk customization. The HT-Slits geometry is designed for dim specimens and samples that require high speed and low excitation power; this represents a powerful solution for spatial transcriptomics and proteomics research, ensuring throughput increases (typically 3-4 times more compared to pinholes) while maintaining good axial resolution in samples of no excessive thickness (< 30 um).

For more information on how to match the disk pattern to your application, visit our webpage Technology.

Microscopy Methods

All the acquisitions of this application note were performed through a Nikon Eclipse Ti2 microscope equipped with CrestOptics X-Light V3 spinning disk system, Celesta laser source (Lumencore) and Kinetix sCMOS camera with 6.5 um pixels and 29.4 mm FOV (Photometrics). The images were acquired by using NIS-Elements Microscope Imaging software (Nikon).

HT-Slits spinning disk consists of continuous spirals with a slit aperture of 50 um and a spacing of 380 um; ST spinning disk contains 50 um pinhole diameter and a spacing of 250 um.

Axial resolution data shown in Figure 1B results from the analysis of 200 nm fluorescent beads acquired with a CFI Plan Apo Lambda 100X oil objective (NA 1.45, WD 0.13). Images were analyzed with PSFj software.

Figure 2 represents stitched images of 80 FOV and 16 Z-planes shown as intensity projections. The mouse brain slice was acquired with a CFI Plan Apo Lambda 20X air objective (NA 0.75, WD 1) in a stack with z-step of 0.9 um. Scale bar: 500 um.

Figure 3, 4, 5 and 6 shows different visualization of a mouse brain slice acquired with CFI Plan Apo Lambda 60X oil objective (NA 1.4; WD 0.13) in a stack with 0.2 um Z-step and 16.8 um of Z-range.

The sample used in this Application note is a mouse brain section marked with 4 different RNA species and DAPI for nuclei visualization. This sample was kindly provided by Dr Alvaro Crevenna, Head of Microscopy, EMBL Rome.

Share this Article

Fill the form and download the application note

Subscribe to MicroLife to receive a monthly update on CrestOptics application notes