SCRINSHOT, a spatial method for single-cell resolution mapping of cell states in tissue sections

Changes in cell identities and positions underlie tissue development and disease progression. Although, single-cell mRNA sequencing (scRNA-Seq) methods rapidly generate extensive lists of cell-states, spatially resolved single-cell mapping presents a challenging task. We developed SCRINSHOT (Single Cell Resolution IN Situ Hybridization On Tissues), a sensitive, multiplex RNA mapping approach. Direct hybridization of padlock probes on mRNA is followed by circularization with SplintR ligase and rolling circle amplification (RCA) of the hybridized padlock probes. Sequential detection of RCA-products using fluorophore-labeled oligonucleotides profiles thousands of cells in tissue sections. We evaluated SCRINSHOT specificity and sensitivity on murine and human organs. SCRINSHOT quantification of marker gene expression shows high correlation with published scRNA-Seq data over a broad range of gene expression levels. We demonstrate the utility of SCRISHOT by mapping the locations of abundant and rare cell types along the murine airways. The amenability, multiplexity and quantitative qualities of SCRINSHOT facilitate single cell mRNA profiling of cell-state alterations in tissues under a variety of native and experimental conditions.

has been used to supplement scRNA-Seq data with spatial information. It utilizes multiple 60 fluorophore-labeled probes, which recognize the same RNA molecule along its length and 61 visualize single RNA molecules as bright fluorescent dots (9, 10). Nonetheless, this method still 62 retains some limitations such as low signal-to-noise ratio, reduced sensitivity on short transcripts, 63 false positive signal due to unspecific-binding of the labeled probes and low capacity for multiplex 64 detection of many RNA molecules (11)(12)(13). Multiplex detection with smFISH was initially levels because it can be proportionally competed with increasing concentrations of a synthetic 192 oligo masking the hybridization site. It also highlights the importance of proper padlock design to 193 achieve similarly high Tm values and hybridizations conditions for the different probes.

196
To evaluate SCRINSHOT applicability to other tissues, we performed the assay using PFA-fixed 197 sections from adult mouse kidney and heart and human embryonic lung. On murine tissues, we

218
We first tested the quantitative power of SCRINSHOT by correlating its detection performance 219 with the fluorescence of a transgenic red fluorescent protein (RFP) in mouse lung tissue sections.

220
In the Sftpc-CreER;Rosa-Ai14 reporter mouse, the RFP expression is activated in AT2 cells, upon

246
Apart from the AT2 cells, the lung alveolus contains additional cells types, including endothelial,

293
Spatial mapping of tracheal cell heterogeneity using SCRINSHOT

294
Recently, two studies addressed the heterogeneity of tracheal epithelium using sc-RNA Seq.

295
These studies identified a new pulmonary cell type, which expresses Cftr and therefore 296 considered to play a role in cystic fibrosis pathophysiology (48,49). They also provided the 297 detailed transcriptomic state of additional cell types, like basal, tuft and secretory cells, including 298 club and two classes of goblet cells.

299
We tested the ability of SCRINSHOT to detect the above cell types and analyze tracheal epithelial

307
For assignment of cell positions we utilized structural landmarks that separate the tracheal airway 308 epithelium in three parts, the proximal, which extends until the end of the submucosal gland, the 309 intermediate part, which spans eight cartilage rings deeper and the distal, which includes the 310 remaining part of trachea epithelium, up to bronchial bifurcation (carina). We also assigned 311 positions to proximal intra-lobar airway epithelial cells, which extend up to the L.L3 branching 312 point (51) and to distal airway epithelial cells located at terminal bronchioles ( Figure 7A). 14 313 For cell-type annotation, we initially applied the following threshold criteria for the selected marker 314 genes. We considered club cells, only the Scgb1a1 pos cells, which were negative for the 315 neuroendocrine cell marker Ascl1 and expressed up to one goblet, tuft, ionocyte and basal-cell 316 markers. Ascl1 pos cells were considered as neuroendocrine cells. Similarly, we annotated the 317 analyzed cells as goblet, basal, tuft cells and ionocytes, if they were positive for at least two 318 characteristic genes of the respective type and only expressed up to one marker of the others.

319
The previously described similarities in gene expression between goblet and club cells in (48),

357
More than 99% of them also contained more than 3 dots of Lyz2, another AT2 marker (54), which

405
For lung tissue collection, the mice were anesthetized with a lethal dose of ketamine/xylazine.

438
The 40-45 Taqman probe sequences were also used to prepare the fluorophore-labelled oligos.

439
Using the IDT OligoAnalyzer tool the length of the sequences was adjusted to Tm 56 o C. To 440 remove the fluorescent oligos after each detection cycle, we exchanged "T" nucleotides with "U"  Thiophosphate-modified bounds to inhibit the 3-5 exonuclease activity of Φ29 polymerase (23).

472
A fixation step with 4% PFA for 15 minutes was done to ensure stabilization of the RCA-products 473 on the tissue. Sections were thoroughly washed with PBS-Tween20 0.05% before next step.

476
The visualization of the RCA products was done with hybridization of the 3'-fluorophore labeled

483
After image acquisition, the coverslips were removed by placing the slides in 70% ethanol in 45 o C.

484
Then, sections were dehydrated using a series of ethanol to mount the hybridization chambers.

485
After tissue rehydration with PBS-Tween20 0.05%, the detection-oligos were digested with Uracil- The nuclear staining was used to align the images of the same areas between the hybridizations 498 and multi-channel *.czi files, containing the images of all genes, were created using Zen2.5 (Carl

499
Zeiss Microscopy GmbH). The images were analyzed as 16bit *.tiff files, without compression or A cell-ROI size criterion was applied to remove the outliers with very small or big surface. In 508 particular, only cells included between two standard deviations of the mean size of the analyzed 509 cells were further processed. In SCRINSHOT we considered a cell-ROI positive for an analyzed 510 gene, we used a threshold strategy. First, we determined the maximum numbers of signal-dots 511 per cell-ROI for all analyzed genes. A cell-ROI was considered as positive, if it contained more 512 than 10% of the maximum number of signal-dots for the specific gene. The higher threshold was 513 set to 3, which was applied for highly abundant genes with maxima over 31 signal dots.

515
Curation of the data

516
In general, the 2 μm nuclear expansion provides an underestimation of real signal dots and 517 provides satisfactory results for airway cells (e.g. Figure 3A' merge-images with cell-ROI outlines) at RT for 30 min, following by 6 washes with PBS-Tween 0.05%.

562
Evaluation of SCRINSHOT specificity using mutated padlock probes