Citation

Cao Xiaowen, Zhu Yuheng, Fan Siyue, Chen Feng, Zhao Yongxi. Accurate counting of RNA in tissues with highly denoising amplified imaging[J]. Chin. Chem. Lett., 2025, 36(12): 111091  

Xiaowen Cao^a, Yuheng Zhu^a, Siyue Fan^a, Feng Chen^a,*, Yongxi Zhao^a,b,*     

Tissues are complex systems that comprise vast cells arranged in a manner. These cells may present significant differences in functional states and cellular subtypes depending on their intricate spatial organizations. Characterization of the distribution and expression level of RNA molecules in the context of tissues is necessary to understand this complexity and heterogeneity in both health and disease [1,2].

Fluorescence in situ hybridization (FISH) is the most common technology used to offer information of the RNA spatial distribution in fixed tissue samples. And the single-molecule FISH is achieved by tiling over RNA sequence with more than thirty DNA probes for enough signal [3–6]. However, this technology can detect only long-strand target with limited specificity. Conversely, in situ amplification assisted FISH can detect targets with little difference even more single-base. These methods are mainly divided into enzymatic-based and DNA assembly-based technologies [7–11]. Among them, RCA-assisted FISH is widely applied due to its simple design and high rate of reaction. In this approach, RNAs are recognized by a circle or padlock probe, triggering RCA with a primer. And the RCA amplicons can be imaged as bright fluorescent spots by hybridization with fluorescently labeled probes. However, this circular probe/primer probe design tends to cause significant amplification background (noise) due to off-target binding or well-known nonspecific adsorption of probes in fixed samples. And the back-ground is typically more pronounced in complex tissues [12,13]. This decrease in the signal-to-background ratio makes specific visualization of RNAs in intact tissue more challenging. Especially, in situ circularization of padlock probe would avoid this non-specific amplification as the circular template is not formed. However, several previous works reported that the enzyme-based ligation in situ is limited by the accessibility of bulky enzymes which need to diffuse through the crowded cellular environment to reach their target molecules [9,14]. This additional enzymatic ligation reaction in fixed cells reduces the detection efficiency.

To eliminate the non-specific amplification background, the optimal RCA-assisted FISH should use the target RNA molecule as the amplification primer. RNA-primed approach avoids the addition of DNA primer probe. And the adsorption or off-target hybridization of circular probe could not cause non-specific amplification. We used the target RNA rather than additional probe as primer to avoid this noise in imaging before [15]. Briefly, a DNA probe was used to hybridize part of RNA, which was then digested by RNase H. This RNA-primed approach achieved nearly zero background and offered comparable detection efficiency. Despite the improvement, this method requires an additional DNA probe binding the target RNA sequence for its cleavage by RNase H. Though this additional sequence hybridization can further increase the detection specificity, it may fail to detect the target sequences bound with proteins. Compared to DNA probe-assisted RNA ribonuclease, the emerging sequence-specific endoribonuclease and guide RNA-assisted Cas enzyme are better choices to cut RNA at interested sites [16–20]. Especially, the sequence-specific endoribonuclease can cut all RNA molecules containing the same restriction site without using any additional DNA probes. It holds important advantage on both multi-site amplification and multiplexed imaging.

Herein, we develop highly denoising amplified imaging, which we called specific RNA cleavage-amplifying FISH (SCA-FISH) for accurate RNA counting in tissues. This method is mainly based on endoribonuclease-mediated RNA direct cutting and phi29 DNA polymerase-catalyzed RNA-primed RCA (Fig. 1). The MazF endoribonuclease can specifically cut the ACA motif in single-stranded RNA to generate a new 3′-OH end. In the RNA-primed RCA, one circular probe is designed to be complementary with RNA cleaved fragment, which will form a short 3′ RNA overhang. The phi29 DNA polymerase first digests this 3′ overhang, and then uses this RNA as primer to catalyze the RCA reaction along the circular probe template. Followed by FISH reaction, the interested RNA molecules can be visualized as bright spots for digital analysis. Our method avoids the additional DNA primers, and could eliminate the false RCA reaction ascribed to non-specific adsorption of DNA probes on fixed samples, reducing nonspecific spots of single cells from 7 to nearly zero. Especially, our method achieves nearly zero fluorescence background imaging of RNA in tissue sections.

Figure 1

	Download: JPG larger image
Fig. 1. Schematic diagrams highly denoising amplified imaging for accurate counting of RNA in tissues.

The first step in the developing SCA-FISH was to determine the specific cleavage of RNA restriction enzyme. The structure of MazF-ssDNA analogue complex (Fig. 2A) suggested that a series of four main chain and three side chain hydrogen bonds with the bases of the (ACA) sequence in the complex define substrate specificity [21]. Then the MazF was used to cleave single strand RNA (ssRNA) and control substrates as described above (dsDNA from in vitro extension, ssRNA from in vitro transcription, ssDNA from purchasing, Fig. S1 in Supporting information). As shown in Fig. 2B, we observed distinct cleavage product in assay with ssRNA. And the assays with other substrates showed limited evidence of cleavage. These analyses identified that MazF appeared to specifically cleaves the ssRNA. We then demonstrated the successful preparation of circular probes (Fig. S2 in Supporting information). And we employed T4PNK to improve the efficiency of RCA (Fig. 2C). Meanwhile, the fluorescence showed no raise when performing the RCA with the cleavage product of control substrates. The images in electrophoretic analysis showed the same results with that in tubes, indicating that the cleaved RNA can efficiently trigger RCA as primer (Fig. 2D).

Figure 2

Download: JPG larger image

Fig. 2. Cleavage of ssRNA with MazF and in vitro specific RNA cleavage-initiated amplification. (A) Model of MazF-ssDNA analogue complex structure (PDB ID: 5CR2). (B) Electrophoretic analysis of cleavage of different substrates by MazF. Substrates and products were resolved by 15% denaturing PAGE. Arrows indicate RNA product of MazF cleavage. (C) Real-time fluorescence of RCA with different cleavage products. (D) Electrophoretic analysis of RCA products with 0.8% agarose gel. Arrows indicate RCA product.

We then investigated the in situ feasibility of our method in fixed cells. As shown in Fig. S3 (Supporting information), we observed bright fluorescence spots marking individual molecules with matched probe. Besides, the number of spots was dramatically reduced with mismatched circular probe and almost none spot was observed when MazF or circular probe was absent. These indicated the specificity of our method. Furthermore, we compared our method to the additional primer-assisted RCA FISH. As shown in Fig. 3, bright fluorescence spots were observed in both samples with the matched probe and mis-matched probe. This indicated that unintended hybridization or absorption of probes caused undesired amplification in additional primer-assisted RCA-FISH (ADA RCA-FISH). Furthermore, we used our method to detect three different mRNA at the same time in single cells. As shown in Fig. S4 (Supporting information), fluorescent spots corresponding to individual RNAs were clearly detected. Then these RNAs were investigated in different cell lines. As shown in Fig. S5 (Supporting information), these mRNAs show diverse expression levels and subcellular distribution in these four cell lines.

Figure 3

	Download: JPG larger image
Fig. 3. Comparation of our method with ADA RCA-FISH. (A) Representative images of target RNA in single cells (red, Alexa Fluor 555 for TK1; blue, DAPI). The scale bar is 10 µm. (B) Statistical analysis of amplicon spot counts for each sample in single cells (N = 30).

Then our method was performed in tissue sections, which suffer from severe background noise caused by off-target binding and cellular autofluorescence. Animal experiments were conducted with approval from the Animal Ethics Committee of Xi'an Jiaotong University (Approval number: 2021–270). For a comparative analysis, we performed two types of in situ RCA-FISH method. The mismatched circular probe was used as negative control. And we performed the statistical analysis of average spot counts per cell through choosing three ROIs for each sample. As shown in Fig. 4 and Fig. S6 (Supporting information), the bright spots were observed by these two methods when using matched probes. Additionally, we compared targeted amplification FISH methods [22–24] for target sequences, which achieved in situ visualization by targeting several sequences of the target. The detection efficiency of these methods is about 10%−30% due to the inaccessibility of some sequences in RNA. Our method has consistent detection rates with traditional RCA methods. The background noises were more intense when using the ADA RCA-FISH than using the SCA-FISH. Meanwhile, the bright spots and background noise were observed in the samples by the ADA RCA-FISH with the mis-matched circular probe, indicating the remarkable non-specific amplification. Oppositely, no fluorescence was showed in section by our method with mismatched probe.

Figure 4

	Download: JPG larger image
Fig. 4. Schematic and representative images of RNA with matched probe (above) and mismatched probe (bottom) in mouse myocardium tissue sections using two different methods. The images showed full field of view and a zoomed-in region (white box in inset). Scale bars: 10 µm.

In theory, the background noise in tissue is getting worse with the rise of section thickness. We chose tissue sections with thickness of 5, 10 and 15 µm of mouse lung to investigate the denosing capacity of our method. As shown in Fig. S7 (Supporting information), the fluorescent spots corresponding to TK1 mRNA were clearly detected with extremely low background in all three sections. Besides, statistical analysis of amplicon spot counts per ROI in three kinds of sections were performed. As shown in Fig. 5, the average spot counts in three sections range from 76.4 to 135.7. And the number of spots in tissue sections with thickness of 10 and 15 µm were similar. The results indicated that the thick tissue section contains more RNA with more cells and our method can reveal the abundance and distribution of RNA even in the thick tissue sections.

Figure 5

	Download: JPG larger image
Fig. 5. SCA-FISH in different thickness of tissue sections. (A) Representative images of RNA in mouse lung tissue sections with different thickness. Scale bar, 10 µm. (B) Statistical analysis of amplicon spot counts per ROI (region of interest) for each sample (N = 10).

In summary, we have developed a highly denoising amplified imaging with specific RNA cleavage, which achieves accurate counting of single-molecule RNA in tissue sections. Our method avoids the additional DNA primers, and eliminates the false RCA reaction ascribed to non-specific adsorption of DNA probes. Comparing to other RCA-based amplification methods, the MazF in our method can specifically cut all of ACA motif in single-stranded RNA, presenting a variety of options for interested region in targeted RNA. Meanwhile the nonspecific spots of single cells were decreased from 7 to nearly zero. And the RNA imaging in uncleared tissue sections showed nearly zero noise. And the efficacy of our method on various thickness of mouse tissues was performed and the results showed that extremely low background was observed in all three sections. We envision this approach will be suitable for single molecular imaging in the thick tissue sections and serve as a tool to reveal the information content from patient samples for biomedical purpose.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Xiaowen Cao: Writing – original draft, Methodology, Investigation, Data curation. Yuheng Zhu: Methodology, Investigation. Siyue Fan: Methodology, Investigation. Feng Chen: Writing – review & editing, Writing – original draft, Methodology, Conceptualization. Yongxi Zhao: Writing – review & editing, Visualization, Project administration, Funding acquisition, Conceptualization.

This work was supported by the National Natural Science Foundation of China (Nos. 22125404, 92068118, 21874105), the Natural Science Basic Research Program of Shaanxi Province (Nos. 2023-JC-JQ-13, 2020JQ-021), the Innovation Capability Support Program of Shaanxi Province (No. 2023-CX-TD-62).

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2025.111091 .