Spatially Compact Arrangement of Larval Zebrafish Sections for Spatial Transcriptomic Analysis

Here, we present a method for aligning and cryosectioning multiple Zebrafish (Danio rerio) larvae samples and collecting them on a single slide for spatial transcriptomic analysis.

Spatial transcriptomic techniques are a sophisticated tool in biomedical research to visualize spatially registered gene expression patterns. Imaging and analysis of multiple samples with spatial imaging platforms can be costly. Performing these tests over multiple experimental conditions, as seen in developmental studies, further increases costs. To reduce costs, this study sought to optimize the techniques and strategies of spatial transcriptomic specimen arrangement for developmental studies. Here, the study utilized zebrafish, which are a well-established developmental vertebrate model that is transparent during development, have ~70% genetic homology to humans, and a highly annotated genome ideal for transcriptomic analysis. Because of their small size, developing zebrafish also allows for compact placement of serial sections across several biological replicates. Herein, we report optimized fixation, cryosectioning, and reliable alignment of multiple fish samples within the imaging area of a multiplex in situ hybridization spatial imaging platform. With this method, zebrafish as young as 15 days post fertilization (dpf) from at least 4 different molds and up to 174 sections can be successfully cryosectioned, collected within the imaging area of 22 mm 10.5 mm (for an in situ spatial transcriptomic slide), and processed simultaneously. Based on section quality, sample alignment, and sample size per slide, this method in zebrafish optimizes output and per sample cost of spatial transcriptomic techniques.

Assessment of spatially distinct expression patterns in tissue remains critical for our understanding of genomic influences in development, cancer, and disease^1,2,3. Spatial transcriptomics combines multiplexed expression techniques with the spatial registration of expression in tissues. "Spatial transcriptomics" was first coined by Ståhl and colleagues⁴, where mounted cancer specimens were probed using in situ next-generation sequencing. Since that time, "spatial transcriptomics" has been used as a catch-all for high throughput expression studies combined with spatial registration. While these are powerful tools, they are also expensive undertakings that often require large institutional investment and laboratory costs before data can be generated⁵. Strategies to minimize cost while preserving high-quality data are in high demand.

Zebrafish, Danio rerio, have become an important model system for studying developmental biology and offer a means of multiplying vertebrate whole organ (and organism) analyses in limited space. Zebrafish are small (4-6 mm as larvae and 2-3 cm as adults) and can lay hundreds of transparent eggs at a time⁶. Zebrafish embryos are fertilized externally and develop rapidly, allowing researchers to introduce transgenes at early stages of development to readily generate gain- and loss-of-function alleles⁷. Fitting multiple specimens on a single slide is an appealing strategy to reduce costs. Their high fecundity and small size make zebrafish an ideal candidate for multiplexing spatial transcriptomic assays which have restricted space for specimens⁸.

Cryosectioning zebrafish larvae is a challenging technique. Many spatial transcriptomic platforms have not been optimized for zebrafish paraffin sections and require cryosections when working with zebrafish as a model organism to preserve tissue structure and retain RNA transcripts. Additionally, the small size of zebrafish makes it difficult to obtain quality cryosections and analyze multiple samples effectively. This task becomes more difficult when working with zebrafish larvae that are smaller and more fragile than their adult counterparts. To overcome these challenges, we describe a method that reliably aligns multiple samples and utilizes the imaging area of spatial imaging platforms efficiently to obtain many high-quality sections on a single slide that can then be imaged and analyzed by spatial imaging platforms (Figure 1). In this instance, this method is applied to a spatial transcriptomic imaging platform.

This protocol follows the guidelines of Dartmouth College's institutional animal care and use committee.

1. Preparing cryostat

1. Cool the cryostat to -22 °C and clean the interior surfaces of the cryostat by brushing debris into the receptacle. Place all necessary brushes and tools inside the chamber.

2. Preparing the disposable base mold

1. Prepare a base mold (37 mm 24 mm 5 mm disposable plastic mold) for sample alignment by drawing a straight line across the inside of a base mold with a permanent marker to use as a reference point for sample alignment. Place a piece of lab tape on the inside of the base mold where the straight line should be before drawing a line for best results (Figure 2A).
2. Draw a dot on the inside of the base mold near the left or right wall to ensure proper sample orientation during cryosectioning (Figure 2B).
3. Measure the desired cutting angle with a protractor and mark this on the inside of the base mold for each sample (Figure 2C).
4. Apply a shallow layer of freezing (optical coherence tomography [OCT]) medium to the prepared base mold. Ensure there is just enough freezing medium to cover the samples.
   1. Avoid air bubbles when applying the freezing medium by priming the nozzle of the medium bottle and adding the necessary amount of medium to one corner of the base mold before shifting the horizontal plane of the base mold so that the medium is distributed evenly across the entire surface.
   2. Squeeze the bottle slowly when dispensing the freezing medium.
5. Add ice to a 1 L beaker, place the base mold with freezing medium in the ice bath, and incubate for at least 10 min to cool the medium.

3. Preparing dry ice:100% ethanol bath

1. Prepare a dry ice and 100% ethanol bath in a fume hood by adding one part of 100% ethanol to one part of dry ice in an ice bucket.
2. Use a disposable aluminum dish or fold aluminum foil into a boat large enough to fit a disposable base mold. Ensure that the dish or boat is large enough for the base mold to lay completely flat.
3. Place the dish or boat in the bath and cover the bucket. Allow 5-10 min for the bucket to cool before freezing samples.

4. Euthanizing samples

1. Randomly select zebrafish for sectioning. If samples vary in size, separate them into groups by relative size to make accurate alignment easier (see discussion for details). Place larger fish and smaller fish in separate dishes.
2. Fill a beaker with fish system water and place the beaker in an ice bucket. Surround the beaker with ice.
3. Monitor the temperature of the water with a thermometer. Let the temperature stabilize between 2-4 °C.
4. Use a net or strainer to put one group of fish into the 4 °C water. Fish should be fully immersed in water and not in contact with ice. Once opercular movement has ceased, add ice to the water to ensure it remains below 4 °C. Leave fish in 4 °C water for 10 min.
   NOTE: Continue with the embedding, alignment, and flash freezing of each group of euthanized samples before euthanizing the next group. Replace the water each time.

5. Embedding and alignment

1. Collect the euthanized fish after they have been submerged in 4 °C water for 10 min. Remove the fish from the water with fine-tipped forceps by grabbing them by the caudal fin and dry them by gently pressing them against an absorbent, lint-free wipe.
   NOTE: For high-quality sections, it is critical to limit the time between removing fish from the 4 °C water and flash-freezing
2. Working with the prepared base mold in an ice bath under a stereomicroscope (10x magnification), place each sample into the base mold along the reference points in the correct orientation and gently cover the samples with another thin layer of freezing medium.
   NOTE: Do not fill the whole mold with freezing medium. Instead, only use a thin layer, just enough to cover all the samples.
3. Use an anatomical reference point to precisely align the fish to the lines marked on the inside of the base mold. Use fine-tipped forceps to adjust the orientation of each fish so that they are aligned and in the same orientation. Avoid creating bubbles in the freezing medium by moving slowly.
4. Apply a piece of dry ice on the bottom of the base mold under the samples until they are locally frozen into position. Keep the horizontal plane of the base mold level to avoid shifting the samples off of their reference points before freezing into position with dry ice.
5. Place the base mold with the samples onto the aluminum boat or dish in the dry ice:100% ethanol bath. Ensure the boat is floating at the surface of the bath and the base mold remains dry. Cover the bath and allow samples to float for 10 min.
6. Wrap the frozen base mold in foil and store in -80 °C freezer until ready to section. Repeat steps 4.2-5.6 for any remaining groups.

6. Cryosectioning

1. Bring frozen base molds to the pre-cooled cryostat for cryosectioning and place them in the cryostat chamber. Transport frozen blocks in a box with dry ice to prevent thawing.
2. Remove in situ spatial imaging slide from storage and place it in a pre-chilled slide holder. Store the slide holder with spatial imaging slide in the -22 °C cryostat chamber until ready to collect sections from the region of interest.
3. Remove the frozen samples from the base mold and then freeze them in a chuck with a fresh freezing medium. Freeze them onto the chuck so that the cutting surface will face the blade.
4. Place a fresh, fine microtome blade in the cryostat.
5. Align the mold to the blade and trim the region of interest (recommended trim thickness 20-50 µm). Ensure the markings made inside the mold are transferred to the frozen sample block to help identify the location within the samples.
6. Adjust the mold during the trimming phase so that the cutting surface is parallel to the reference markings in the freezing medium.
7. Collect sections onto a standard positively charged microscope slide and check them by brightfield to confirm when trimming is no longer necessary.
8. Remove the spatial imaging slide from the cryostat chamber and place it in a 4 °C ice bath. Keep the slide in the slide holder. Ensure that the slide does not get wet.
9. Start cryosectioning (recommended 10-14 µm; Figure 3) and collect sections onto positively charged slides until the region of interest in the samples is reached. Check sections by brightfield to confirm the region of interest will be the next section from the mold.
10. Bring the single-cell spatial imaging slide back to the cryostat chamber. Remove the slide from the slide holder and collect sections from the exact area of interest onto the spatial imaging slide row by row using a fine-tipped paintbrush to keep sections from rolling up.
    1. Press a corner of the empty, freezing medium into the knife stage with the backside of the paintbrush so the section remains flat when grabbing the slide for collection.
    2. Use the top of the slide as a pivot point, slowly lower the slide onto the section, and allow sections to adhere to the slide for 3 s before lifting the slide off the knife stage.
    3. Work from left to right when collecting sections in the imaging area of the slide and overlap layers of empty, freezing medium when possible.
    4. Use colored paper borders as a reference to place sections within the imaging area of the slide if tissue is hard to see.
11. After collecting sections from the region of interest, place the slide back into the slide holder. If there are no more samples to be collected onto the spatial imaging slide, store the slide at -80 °C for up to 2 weeks until it is ready for instrument analysis with the in situ spatial imaging platform. If sections need to be collected from multiple molds, put the spatial imaging slide back in the 4 °C ice bath and repeat steps 6.3-6.11 with the next mold.
12. Collect sections before and after region of interest sections from each mold on a standard positively charged microscope slide for hematoxylin and eosin (HE) staining to check that sample alignment and section quality are sufficient before proceeding with analysis.

7. Fixing the sample

1. Remove the reference slides from the cryostat and air dry at RT for 30 min to adhere sections to the slide.
2. Fix sections by placing them in a slide container with 4% paraformaldehyde (Table 1) for 20 min.
3. Wash sections by placing them in a slide container with distilled water for 3 min.
4. Continue with HE staining or dry and store slides at -80 °C for future staining.

8. HE staining of the sections

1. Dehydrate and clean the sections by incubating the slides in 100% ethanol for 2 min, 95% (Table 2) ethanol for 2 min, and then tap water for 1 min. Use a microscope slide staining rack to transfer slides from bath to bath.
2. Stain the nuclei and differentiate by incubating the slides in Hematoxylin for 2 min 45 s, tap water for 1 min, 0.3% acidified alcohol (Table 3) for 1 min, and then running tap water for 1 min. Use a microscope slide staining rack to transfer slides from bath to bath.
3. Stain the cytoplasmic components and dehydrate by incubating the slides in Eosin Y 1% for 45 s, 50% ethanol (Table 4) for 1 min, 95% ethanol for 1 min, and 100% ethanol for 1 min. Use a microscope slide staining rack to transfer slides from bath to bath.
4. Clear the sections by incubating slides in Xylene for 1 min. Mount and cover the slides by applying a drop of mounting medium to the top third of the slide with a transfer pipet and slowly lowering a coverslip on top of the mounting medium with forceps.

9. Spatial transcriptomic imaging and analysis of the sections

1. Remove the imaging slides from -80 °C storage and image with an in situ spatial imaging platform for spatial transcriptomic analysis.
   NOTE: The exact steps involved in imaging will be determined by the spatial imaging platform.
2. Review the quality control metrics of the spatial transcriptomic platform. Important metrics to check are the number of cells detected, median transcripts per cell, nuclear transcripts per 100 µm², and total high-quality decoded transcripts of each gene in the probe set.
   NOTE: These quality control metrics do not have universal thresholds, and expectations for these thresholds will vary depending on the sample and gene panel being used.
3. Read the data output of detectable RNA transcripts, determine which transcripts are low quality based on the experiment's quality control metrics, and filter out low-quality transcripts. Analyze the remaining high-quality transcripts in relation to their spatial arrangement within the section.
4. View the cellular segmentation of the sections and cluster cells based on experimental interests. Compare the RNA transcripts within cell clusters of the region of interest across groups of different ages of zebrafish on the same slide.

In this method (Figure 1), zebrafish is used as an animal model to probe for spatially resolved gene expression patterns. Cryosectioning larval zebrafish efficiently for spatial imaging is challenging. Sections must be high quality to retain tissue structure and detectable genes (Figure 4). Sections containing multiple samples for spatially efficient imaging must be aligned precisely to analyze regions of interest across all samples (Figure 2 and Figure 5). Finally, sections must be collected and spaced efficiently to maximize the number of potential data points within the imaging area of a spatial transcriptomics slide (Figure 3 and Figure 6). Experiments were measured using these three parameters to determine whether they were successful or sub-optimal.

Many protocol modifications contributed to our ability to collect high-quality sections that are good candidates for spatial transcriptomic imaging. Adjusting the protocols for euthanization, fixation, and embeddingled to improvements in section quality (Figure 4). In the first protocol (Figure 4, protocol 1), the section quality was sub-optimal, and tissues did not retain their structure. To correct this, in the second protocol (Figure 4, protocol 2), a flash-freeze step was introduced after embedding samples and fixed tissue after collecting sections onto slides. This improved the overall tissue structure of the sections but failed to reach what was thought to be acceptable quality for spatial transcriptomics. Finally, in the third protocol (Figure 4, protocol 3), The euthanization techniques were refined by shortening the amount of time it took to euthanize fish and embedding fish on ice before flash-freezing and subsequent sectioning and fixation post-collection. This resulted in the highest quality samples that were good candidates for spatial transcriptomic imaging.

Aligning multiple samples to a region of interest required some modifications to the protocol as well. One of the modifications that improved alignment across samples the most was introducing reference points to the base molds that zebrafish are embedded in (Figure 2). In the experiments conducted in this study, the samples were aligned by comparing structures across multiple sections in the same cut (Figure 5).

The last important step in developing the method for efficient spatial transcriptomics was to fit as many of these high-quality, aligned samples of different ages onto the same slide for imaging. Gaining experience and improving technical skills contribute to an improved number of samples being collected. These improvements were also largely due to slight adjustments to the embedding and collection steps in the protocol. These panels in Figure 6 highlight the number of sections collected in the imaging area of each slide over time. Figure 6A (40 sections of 15 dpf fish) is an early run that uses one mold during sectioning. Figure 6B (90 sections of 15 dpf fish and 47 sections of 26 dpf fish) introduces a second mold and shows improvement in spatial arrangement by reducing the amount of space between samples during embedding. Figure 6C (54 sections of 15 dpf fish, 80 sections of 19 dpf fish, 24 sections of 23 dpf fish, and 21 sections of 26 dpf fish) introduces a third and fourth mold and shows continued improvement in a spatial arrangement by trimming the cutting surface of the block around the samples and overlapping empty, freezing medium. These improvements allowed us the opportunity to place 179 sections in the imaging area of our spatial transcriptomic slide and test fish of 4 different ages.

At the completion of the protocol, we had samples that were readily analyzable both with the off-the-shelf multiplexed in situ software and with custom methods. The multiplex software used here employed a pseudo-phred score, loosely based on phred quality scores in sequencing data⁹. The scores ranged from 59%-70%, with 60% being the cutoff for a warning of lower-quality reads. Upon inspection, the low pseudo-phred score was due to the low signal complexity in the empty space between sections. Though the slide area was efficiently used, the slide area was still >50% empty. When accounting for the proportion of slide area occupied by tissue, where the signal was sufficiently complex, the signal quality was excellent.

Low-intensity, nonspecific signals were also seen in two general regions on the slide: outside the specimens and within the empty notochord (Figure 7A). Outside the sections, there was a pattern consistent with a wash artifact, which would decrease moving away from the specimens (Figure 7B). A nonspecific signal was also seen within the empty regions of the specimens occupied by the notochord. This was considered a likely wash artifact where amplification reagents could get trapped. Transcripts in both regions were largely cytoplasmic (i.e., ribosomal proteins L3 and L4) or mitochondrial (aspartate transferase and isocitrate dehydrogenase). When these regions were included in Baysor segmentation, which identifies cells based on localization and clustering of transcripts around nuclei¹⁰, the segmentation analysis included the cell borders, which sometimes included these. This signal was low enough that it was easily distinguishable from the specimen.

Figure 1: Schematic outline of cryosectioning for spatial transcriptomics in zebrafish. Juvenile zebrafish are randomly collected and then grouped by relative size. Samples are then embedded in a freezing medium and aligned by reference points on the base mold corresponding to a region of interest. Each mold is flash-frozen, cryosectioned, and collected one row at a time on the same slide. Slides are stored at -80 °C after sectioning and imaged and analyzed using a spatial transcriptomic imaging platform. Please click here to view a larger version of this figure.

Figure 2: Establishing reference points for precise sectioning. A series of images demonstrates the process of adding reference points to a disposable base mold to align multiple samples for cryosectioning. The first step (A) is to use lab tape to mark a line on the inside of the mold that all samples will use as a reference for alignment. (B) Place a dot on the inside of the mold near the left or right wall to inform proper sample orientation during cryosectioning, and then (C) mark lines for each sample at a desired angle. Please click here to view a larger version of this figure.

Figure 3: Schematic of sectioning location. Fish are sectioned coronally, from caudal to rostral, at 14 µm intervals. Please click here to view a larger version of this figure.

Figure 4: Section quality improvement with protocol modifications (10x). The figure compares the first two protocols, which resulted in sub-optimal tissue quality, with the final protocol, which resulted in high-quality sections (which retained morphology) that are good candidates for spatial transcriptomics. Protocol 1 utilizes Penn State bio-atlas' comparative analysis⁸, highlighted by 4% paraformaldehyde fixation prior to sectioning. Protocol 2 introduces dry ice/ethanol flash freeze after euthanization and fixation after sectioning, which preserves tissue structure much better with the caveat of freezing artifacts throughout the section. Protocol 3 refined the euthanization techniques and embedding on ice before flash-freezing and subsequent sectioning and fixation post-collection, which resulted in the highest-quality samples. The scale bar is 500 µm. Please click here to view a larger version of this figure.

Figure 5: Verification of multiple section alignment (20x). Representative HE images of 26 dpf sections that highlight the alignment between multiple samples in a single cut. The scale bar is 500 µm. Please click here to view a larger version of this figure.

Figure 6: Improved spatial arrangement of specimens. These panels highlight the number of sections collected in the imaging area of each slide over time. Gaining experience and improving technical skills contribute to an improved number of samples being collected. These improvements were also largely due to slight adjustments to the embedding and collection steps in the protocol. Panel A (40 sections of 15 dpf fish) is an early run that uses one mold during sectioning. Panel B (90 sections of 15 dpf fish and 47 sections of 26 dpf fish) introduces a second mold and shows improvement in a spatial arrangement by reducing the amount of space between samples during embedding. Panel C (54 sections of 15 dpf fish, 80 sections of 19 dpf fish, 24 sections of 23 dpf fish, and 21 sections of 26 dpf fish) introduces a third and fourth mold and shows continued improvement in a spatial arrangement by trimming the cutting surface of the block around the samples and overlapping empty, freezing medium. The brackets in each panel represent different molds used during sectioning. Sections are on a 10.5 mm 22 mm area. The scale bar is 4 mm. Please click here to view a larger version of this figure.

Figure 7: Representative results of our spatial transcriptomic experiment. (A) A single image of a slide processed with the protocol as seen in Figure 6C. (B) 14 µm sections of a zebrafish at 15 dpf after spatial transcriptomic imaging displaying nonspecific signal adjacent to sections and within the notochord. The scale bar is 4 mm in panel A. Please click here to view a larger version of this figure.

Table 1: 4% Paraformaldehyde formula. Composition of 4% paraformaldehyde solution to fix samples.

Table 2: 95% Ethanol formula. Composition of 95% ethanol solution to dehydrate sections during H&E staining.

Table 3: 0.3% Acidified alcohol formula. Composition of 0.3% acidified alcohol to differentiate during HE staining.

Table 4: 50% Ethanol formula. Composition of 50% ethanol to dehydrate sections during HE staining.

This report provides detailed solutions to many of the technical challenges associated with zebrafish as a model organism in spatial transcriptomic analysis during development. In addressing these challenges, our compact specimen arrangement optimizes costs on the emerging spatial transcriptomic platforms¹. Cryosectioning larval zebrafish for spatial imaging is challenging. Sections should retain sufficient tissue structure and transcript quality for satisfactory experimental execution and downstream spatial analysis⁸. Slices containing multiple samples for spatially efficient imaging must be aligned precisely to analyze common regions of interest across all samples. Finally, sections must be collected and spaced efficiently to maximize the number of potential data points within the designated area of an imaging slide.

The protocol permits the collection of high-quality sections that are excellent candidates for single-cell spatial imaging platforms. The first modification was to flash-freeze zebrafish and then fix sections in 4% paraformaldehyde after collecting them onto a slide. Originally, zebrafish were fixed with 4% paraformaldehyde at 4 °C overnight and embedded and sectioned afterwards¹¹. Choosing to fix samples after they were sectioned helped preserve internal tissues that were not effectively preserved when fixing a whole sample (Figure 4).

The second critical modification was refining the euthanization method to preserve tissue integrity. Zebrafish were originally placed in 4 °C water for 20 min to ensure that they were sufficiently euthanized. This is 10 min beyond the minimum requirements for the IACUC guidelines. This extra time in 4 °C water gave specimens more time to decompose and contributed to less-than-ideal section quality of very fine or delicate internal structures. Reducing the amount of time in 4 °C water to the minimum requirement of 10 min helped preserve delicate internal structures that were missing from previous sectioning attempts. Other modifications not shown in Figure 4 include sectioning on the same day as euthanization, reducing the amount of time between euthanization and flash freezing from 15 min to under 5 min, and working with a pre-chilled freezing medium when embedding. The longer the samples were stored at -80 °C before sectioning, the worse the section's tissue quality ended up being. The differences were subtle, but it is ideal for sectioning on the same day as euthanization and block creation. Shortening the amount of time it takes to align samples after euthanizing and before flash freezing to no more than 5 min is critical to section quality.

Aligning multiple samples to a region of interest required some modifications to the protocol as well. The precision of alignment that was achieved (Figure 5) grouped samples by relative size and provided the greatest contribution to an efficient spatial arrangement. The primary strategy for aligning the samples is overlaying the zebrafish onto reference points that are drawn onto the base molds in which they are embedded. Specimens are aligned to these reference points by anatomical structures within the specimens of the zebrafish that can vary depending on the desired region of interest. Developing fish of the same age can vary in size up to 50%¹². Samples with drastically different sizes can misalign due to varying distances between the anatomical reference point and region of interest. Grouping samples of the same age by relative size and embedding each size into their molds improved the ability to align multiple samples within their mold. This method can reliably align approximately 2/3 of samples depending on the window allowed by the region of interest and choice of anatomical reference point.

Reducing the space between samples when embedding and aligning leads to a greater number of samples per slide. However, this tight configuration is further constrained by the fiducial limits on the imaging area. In addition to experience at the cryostat, several strategies for reconciling these are described. First, the precise placement of overlapping embedding medium between sections and trimming of the frozen mold can have profound impacts on the number of sections that can fit in a specific imaging area. Figure 6 highlights the number of sections obtained before and after optimizing the embedding arrangement. Second, a boundary was placed on the outside of the outermost samples during embedding and before freezing to assist with proper placement onto the imaging slide. Third, the size and transparent nature of larval zebrafish can make it difficult to know where samples are located within a slide so that they can be precisely collected onto a slide. We found that using colored paper to contrast the white freezing medium to create a boundary on the outside of the outermost samples led to a much more precise placement of the sections onto the slides.

The nuances of local temperature help with sample preservation during the cutting process. Once samples from multiple molds are collected onto a single slide, the imaging slide is stored at 4 °C, the next mold is trimmed to the desired region of interest before collecting again. The imaging slide must be slightly warmer than the sections for it to adhere properly to the surface of the slide. So, storing the imaging slide at -80 °C or in the cryostat at -20 °C between sample collection from different molds is not ideal. It is also critical that the collected samples on the imaging slide do not undergo multiple freeze/thaw cycles, as this can negatively impact the quality of RNA transcripts within each sample. Keeping the imaging slide with previously collected sections at 4 °C adequately preserves RNA transcripts while maintaining a lower temperature for proper adherence without negative side effects associated with freeze/thaw cycles¹³.

Although satisfactory results were obtained, the design of this method does come with possible limitations. The first limitation is that we are constrained to the imaging area of a spatial transcriptomic slide, which puts a ceiling on the total number of sections that can be imaged on the same slide. Given that we still have empty space between sections, it is likely that we have not yet reached the bounds of what is possible for this method, but we recognize that there is a point where it is no longer possible to fit any more sections onto the imaging slide. Working with larger fish at older ages will also limit the total sample size but should not affect the total amount of tissue that can fit into the imaging area of a slide. Technical improvements by spatial imaging platforms could possibly increase the total amount of tissue that could possibly be imaged by enlarging the imaging area on the imaging slides. Another limitation of this method is that the protocol has not been optimized for fish younger than 15 dpf. It was found that fish are increasingly more fragile the younger they are and that fish younger than 15 dpf were extremely difficult to handle and were easily damaged during the embedding process. This resulted in increasing the amount of time between euthanization and flash-freezing, which was detrimental to tissue quality. This also reduced the percentage of viable samples left for cryosectioning, and the few viable fish that remained for sectioning were not rigid enough to retain tissue structure for quality analysis. Finding tools that are better suited for handling more fragile fish and working with a less-dense freezing medium that also increases the rigidity of samples during sectioning could allow for this method to be applied to fish younger than 15 dpf. The last major limitation of this method is sample alignment for very small regions of interest. We found that the smaller the region of interest in the sample, the less likely it was that we were able to align them properly. We worked with a region of interest that is approximately 50 µm in size, and this resulted in about 2/3 of the sections being properly aligned. Samples with a smaller region of interest than 50 µm could result in less than 2/3 of sections being aligned. A majority of misaligned samples come from the embedding step in the protocol. It is difficult to adjust one sample in the freezing medium without moving other samples that have already been aligned. There is also a limited amount of time that the researcher has to align these samples before negatively impacting tissue quality. Lastly, molds are moved from the stereomicroscope to the flash-freezing bath, and it is possible for samples to shift in the mold before they are completely frozen. It is possible that working with a colder freezing medium could allow the researcher to move one sample more easily without affecting other samples that are already aligned. Flash-freezing the mold without having to move it after alignment could eliminate sample shifting before they are completely frozen in place. We did identify nonspecific signals outside of the specimens.

The final product and initial post-analysis presented additional lessons. Nonspecific signals were present, which required consideration. Per the manufacturer, some low signal transcripts are a known artifact and ambient RNA can be seen in single-cell RNA-seq that can be managed post-analysis¹⁴. The protocol used here may have contributed to ambient transcripts as the fixation occurred after the placement of the specimens. However, it was found that early fixation led to poor and inconsistent morphology for these young ages. Alternatively, amplification reagents that pooled could create such an effect. Qualitatively, the nonspecific signal seen in the notochord was most pronounced. During processing, the notochord was the only truly empty region as it did not contain mounting media and transcript signals, which resembled the cytoplasmic and mitochondrial profiles of adjacent notochord lining tissues. This suggests that physical "backing" (by tissue or mounting media) is helpful in reducing this signal. Ultimately, we reason that the exclusion of the signal in post-processing is a satisfactory strategy. The nonspecific signal features were readily identifiable after experiments and could be mitigated during downstream analysis, for example, through manual exclusion of nonspecific signals of regions outside the specimen.

Zebrafish as young as 15 dpf from at least 4 different molds (and ages) can be successfully cryosectioned and collected within the spatial transcriptomic imaging area, and up to 179 sections can be processed simultaneously. Based on section quality, sample alignment, sample size, and quality of RNA transcripts during imaging, we conclude that this method is an effective protocol for conducting efficient spatial transcriptomic assays. Researchers can use this method to reduce costs associated with spatial imaging, increase sample size during spatial transcriptomic assays, and even achieve large experiments on a single run.

The authors have no disclosures or conflicts of interest regarding this report.

Sectioning and imaging were performed with instruments provided by shared resources at the Dartmouth Cancer Center, funded by NCI Cancer Center Support Grant 5P30CA023108, and the Center for Quantitative Biology at Dartmouth College (NIGMS COBRE).