Spatial transcriptomics enables us to study gene expression from a new perspective by providing the localization information of detected genes. Previously, spatial gene expression profiling could be achieved with methods such as laser-capture microdissection (LCM) that allow to analyze regions of interest in tissue samples (Emmert-Buck et al., 1996). However, it can’t provide an overview of the transcriptional atlas of the target region and its surrounding regions. Thus, it may lead to missing information on how genes are synergistically expressed in their native context. Currently, the mainstreamed spatial transcriptomic technologies are often referred to as methods that can offer localization-indexed gene expression information in a defined area of tissue samples. These methods can be categorized into two approaches, one is based on the sequencing of spatially-indexed transcripts, and the other is based on imaging in situ (Moses and Pachter, 2022). The first approach relies on mapping back the detected transcripts to the original compartmentalized regions in tissue by decoding the barcoded cDNA primers after the next-generation sequencing (NGS) (Koboldt et al., 2013). The second approach includes a series of in situ hybridization (ISH) (Gyllborg et al., 2020) or in situ sequencing (ISS) (Ke et al., 2013) methods that depend on cyclic imaging to decode the barcodes for individual transcripts directly in their original places, achieving highly multiplexed in situ RNA detection by using combinatorial color-coding strategies to overcome the limitation of spectrally-distinguished labels for ISH assays. The NGS-based sequencing approach has a high coverage of genes that includes almost all expressed mRNAs but at a lower spatial resolution because each compartmentalized spot typically contains more than one cell (Stahl et al., 2016). The imaging-based approach includes targeted spatial transcriptomic methods that usually analyze a panel of pre-selected genes at a much higher resolution because RNA molecules are detected at single molecule resolution (Wang and Guo, 2021; Le et al., 2022).

In situ sequencing is a type of imaging-based targeted spatial transcriptomic technology. In ISS, rolling circle amplification (RCA) is used to amplify circularized probes designed for specific genes or cDNA generated by in situ reversed transcription (Ke et al., 2013; Lee et al., 2014a; Liu et al., 2021), forming rolling circle amplification products (RCPs) that can be sequenced by NGS chemistry to identify the detected genes based on their barcodes or mapping the sequences to the reference genome (Ke et al., 2016). The classical ISS is based on barcoded padlock probes and RCA for signal amplification combined with the combinatorial probe anchor ligation chemistry (cPAL) (Drmanac et al., 2010) for decoding the amplification products. As a popular and proven versatile, targeted spatial transcriptomic method, ISS has been applied for region-specific gene expression profiling (Yu et al., 2021), in situ cell typing as well as novel histopathological discovery (Svedlund et al., 2019; Qian et al., 2020), etc. Here, we present improved in situ sequencing (IISS) for high-resolution targeted spatial transcriptomics analysis. Our method is based on the probe designing principle from amplification-based single molecule fluorescent in situ hybridization (asmFISH) (Lin et al., 2021) combined with an improved combinatorial probe anchor ligation (icPAL) chemistry for RCA-based in situ sequencing library preparation. By using a new cell segmentation pipeline based on Cellpose (Stringer et al., 2021), we showed that IISS could be applied to perform in situ cell typing both in fresh frozen and formalin-fixed paraffin-embedded tissue sections. We also demonstrate the construction of the invasive carcinoma trajectory and investigation of cell-to-cell communications between malignancy and normal cells of cancer tissue using spatial gene expression data generated by IISS.