Mar 23, 2018
Single-molecule mRNA detection and counting in mammalian tissue
- Anna Lyubimova1,
- Shalev Itzkovitz2,
- Jan Philipp Junker1,
- Zi Peng Fan3,
- Xuebing Wu3,
- Alexander van Oudenaarden3
- 1Massachusetts Institute of Technology, Hubrecht Institute—KNAW;
- 2Massachusetts Institute of Technology, Weizmann Institute of Science;
- 3Massachusetts Institute of Technology
- Human Cell Atlas Method Development Community
External link: http://www.nature.com/doifinder/10.1038/nprot.2013.109
Protocol Citation: Anna Lyubimova, Shalev Itzkovitz, Jan Philipp Junker, Zi Peng Fan, Xuebing Wu, Alexander van Oudenaarden 2018. Single-molecule mRNA detection and counting in mammalian tissue. protocols.io https://dx.doi.org/10.17504/protocols.io.mc7c2zn
Manuscript citation:
Single-molecule mRNA detection and counting in mammalian tissue
Anna Lyubimova, Shalev Itzkovitz, Jan Philipp Junker, Zi Peng Fan, Xuebing Wu, Alexander van Oudenaarden.
License: This is an open access protocol distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Protocol status: Working
We use this protocol and it's working
Created: December 20, 2017
Last Modified: March 28, 2018
Protocol Integer ID: 9343
Keywords: Fluorescence in situ hybridization, Molecular imaging, Oligonucleotide probes, Transcription
Abstract
We present a protocol for visualizing and quantifying single mRNA molecules in mammalian (mouse and human) tissues. In the approach described here, sets of about 50 short oligonucleotides, each labeled with a single fluorophore, are hybridized to target mRNAs in tissue sections. Each set binds to a single mRNA molecule and can be detected by fluorescence microscopy as a diffraction-limited spot. Tissue architecture is then assessed by counterstaining the sections with DNA dye (DAPI), and cell borders can be visualized with a dye-coupled antibody. Spots are detected automatically with custom-made software, which we make freely available. The mRNA molecules thus detected are assigned to single cells within a tissue semiautomatically by using a graphical user interface developed in our laboratory. In this protocol, we describe an example of quantitative analysis of mRNA levels and localization in mouse small intestine. The procedure (from tissue dissection to obtaining data sets) takes 3 d. Data analysis will require an additional 3–7 d, depending on the type of analysis.
Guidelines
The protocol workflow timing is as follows:
Steps 1–5, coupling FISH probe libraries to fluorophores: 2–4 h
Steps 6–10, probe precipitation, washing, and purification: 1–2 d
Step 11, probe reconstitution and storage: 5–10 min
Steps 12–16, tissue dissection, fixing and cryoprotection: 4 h to overnight
Steps 17–20, preparation of frozen blocks: 30–40 min
Steps 21–26, cryosectioning, fixation and permeabilization of the sections: 5 h 30 min to overnight
Steps 27–29, proteinase K digestion (optional): 20 min
Steps 30–32, hybridization: overnight
Steps 33–36, washing and mounting: 1 h 40 min
Steps 37–42, microscope setup and data collection: 4–6 h
REQUIRED EQUIPMENT
- Microcentrifuge tubes, 1.8 mg
- Six-well plates (BD Biosciences, cat. no. 351146)
- Petri dishes (BD Biosciences, cat. no. 351029)
- Parafilm (Sigma-Aldrich, cat. no. P7793-1EA)
- Sharp-tip forceps (ROBOZ, cat. no. RS-5095)
- Filter paper (Schleicher & Schuell, cat. no. 10310992)
- Speed vacuum concentrator (Eppendorf Concentrator Plus)
- Tabletop centrifuge (Eppendorf, cat. no. 5424R)
- Vortex Genie (Scientific Industries)
- NanoDrop 2000 spectrophotometer (Thermo Scientific)
- Falcon tubes, 15 mg (BD Biosciences, cat. no. 352196)
- Falcon tubes, 50 mg (BD Biosciences, cat. no. 352070)
- Tube rotator (Stuart, cat. no. SB2)
- Nutating mixer (VWR)
- Centrifuge (Eppendorf, cat. no. 5810R)
- RNase-free, low-retention pipette tips (GeneMate, cat. no. P-1237)
- Sonicator (VWR, Ultrasonic Cleaner)
- Desiccator
- Plastic molds for tissue embedding, cat. no. 6254-10 (10 × 10 × 5 mm); cat. no. 6254-15 (15 × 15 × 5 mm); cat. no. 6254-25 (25 × 20 × 5 mm) (frozen blocks; Tissue Tek Cryomold, Electron Microscopy Sciences)
- Cryostat (Cryostar NX70, Thermo Scientific)
- HPLC (Gilson GX271 liquid handler, with C18 Targa column and UV-visible 155 spectrometer, Higgins Analytical)
- Pre-column frit (Chrom Tech, cat. no. A-332X)
- Hybridization oven (UVP, HB-1000 hybridizer)
- Microscope glass slides (Thermo Scientific, Gold Seal, cat. no. 3031-002)
- Microscope cover glass, 22 × 22 mm, no. 1 (Thermo Scientific, cat. no. D10143263NR1)
- Sterile syringe filters, 0.2 μm (Millipore, cat. no. 6900-2502)
- Cover glass staining rack (Electron Microscopy Sciences, cat. no. 72241-01)
- Cover glass wash jar (Electron Microscopy Sciences, cat. no. 70312-23)
- Inverted fluorescence microscope (Nikon, Eclipse-Ti)
- Numerical aperture (NA) 1.4 objective, ×100 (Nikon)
- Immersion oil
- Metal halide lamp (Prior Lumen 220)
- High sensitivity CCD camera (Photometrics Pixis 1024B)
- Motorized optical shutter (Sutter SMARTSHUTTER, Sutter Instruments)
- Motorized stage controller (TI-S-E motorized stage)
- Filter cube for Cy5.5 (Chroma Technology, cat. no. 41023)
- Custom filter cube for Alexa Fluor 594: Omega Optical (exciter: cat. no. 2017901-590DF10, dichroic filter: cat. no. XF2014-610DRLP and emitter: cat. no. XF3028-630DF30)
- Filter cube for TMR (Omega Optical, cat. no. XF204)
- Computer of minimum 4 GB RAM
- Microscope managing and imaging software (Metamorph, Molecular Devices)
- MATLAB (The MathWorks)
- KNF Teflon pump
- Vacuum pump
- Styrofoam boxes with lids
- ImageM and other MATLAB codes developed in our lab are available upon request (requests can be addressed to S.I. (shalev.itzkovitz@weizmann.ac.il), J.P.J. (j.junker@hubrecht.eu) and A.v.O. (a.oudenaarden@hubrecht.eu))
Applications of the method
This protocol enables researchers to determine the absolute numbers of mRNA molecules in single cells within a tissue. The absolute numbers obtained by our technique are in a close correspondence to those measured by different means, for instance, via quantitative reverse-transcription PCR (qRT-PCR)1,2. As few as one mRNA molecule per cell can be detected. The maximum number of individual molecules that can be resolved microscopically is of the order of several thousand. This upper limit is dictated by the ratio between the cell volume and the typical voxel taken by a diffraction-limited mRNA dot (~0.2 × 0.2 × 0.3 μm).
Three different mRNA species can be simultaneously visualized (using probe libraries differentially labeled with Cy5, Alexa Fluor 594 and tetramethylrhodamine (TMR)) and quantified in parallel by implementing the present protocol; this upper limit is determined by the number of different fluorophores that can be spectrally resolved. Simultaneous staining with a DNA-binding dye and an antibody against cell-cell adhesion proteins enables the visualization of cell nuclei and borders,
which is a requirement when the aims are to assign mRNA spots to cells within intact tissue.
As an example of quantitative and spatial analysis of gene expression, here we use the crypt of Lieberkühn, a functional unit of the intestinal epithelium, as a target tissue. First, the numbers of diffraction-limited spots are automatically detected using algorithms previously developed in our laboratory2,3. Next, cells are manually segmented on the basis of DAPI nuclear staining and E-cadherin lateral membrane staining, and spots are then assigned to individual cells. The resulting single-cell gene-expression profiles can be presented as a spatial profile (Fig. 1), which we termed ‘cryptogram’, from ‘crypt’ and ‘histogram’. This approach enables the experimenter to assess the precise spatial location of the expression of genes of interest in multiple crypts in the tissue of interest3 (Fig. 1h).
In addition to the intestine3–6, the three-color smFISH protocol can be applied to a variety of tissues, including mammary glands7, kidney8, stomach, lymph nodes, mouse embryos at different gestational stages, human breast tumors9 and possibly many more. Although the quality of FISH signal is consistently high in the different tissues tested, autofluorescence and/or nonspecific probe binding may emerge in some tissues, and it may need to be addressed if it interferes with spot detection and/or data analysis. Examples of tissues that we have found to be particularly high in autofluorescence are skin (especially hair shafts and cornified layers of epidermis; A.L., unpublished observation), stomach (parietal cells contain highly autofluorescent cytoplasmic vesicles; A.L., unpublished observation) and small intestine (Paneth cells with highly autofluorescent cytoplasmic granules3). smFISH can also be combined with other techniques, such as EdU staining4, immunofluorescence4 and DNA FISH9 for probing single cells in intact tissues.
Given that hybridization conditions for different probe sets are similar, this method has great potential for the simultaneous detection of several mRNA species. As mentioned, the number of mRNA species to be detected simultaneously is limited by the availability of fluorophores that can be spectrally resolved using current technology. We routinely detect three different transcripts at the same time, using Cy5, Alexa Fluor 594 and TMR fluorophores. Other fluorophores emitting light at different wavelengths can be used as well, thus increasing the number of mRNAs that can be simultaneously detected10. Furthermore, combinatorial labeling, in which different probe subsets are coupled to different fluorophores, can further increase the number of different color combinations that can be computationally resolved10,11. This approach may enable truly multiplexed single-molecule mRNA imaging, which would greatly facilitate large-scale gene expression profiling in single cells, in the context of tissues.
Cell border detection (segmentation) may become a challenge in tissues for which no good antibody against cell membrane– associated proteins is available. In these cases, cell borders may be visualized by other means, such as by using fluorescently labeled phalloidin or GFP-fusion membrane-associated proteins. An example of the latter is the use of tissues from a knock-in mouse expressing E-cadherin fused to cyan fluorescent protein (CFP)12. Alternatively, when cell border labeling is impossible, too laborious or not crucial, tissue surface can be divided into arbitrary (virtual) sectors (‘cell proxies’) and mRNA spots can be assigned to these proxies instead of cells. After the segmentation step, spatial analysis of mRNA distributions can be performed according to the topological features of the tissue.
In addition to tissue sections, our protocol can be applied to whole-mount organs or embryos with no substantial modifications. The only challenging part of such an application is imaging depth, as the smFISH signal quality from parts lying deeper than 10 μm below the objective will decrease substantially. This problem can be alleviated by using confocal microscopy. The best results so far are obtained on a spinning-disk microscope (quantifiable signal from tissue layers 30–50 μm deep tissue).
Experimental design
Oligonucleotide libraries are designed in agreement with the set of requirements established previously2, and then purchased. By the term ‘library’, we mean a set of oligonucleotides, each of which hybridizes to the same mRNA molecule. The ‘collective’ binding of singly labeled oligonucleotides to the target mRNA molecule results in the appearance of a single bright fluorescent spot. A typical library consists of 48 oligonucleotides, each 20 bp long. The number of oligonucleotides in a library should exceed 20 (ref. 2), and it often is as large as 96.
The minimum size of a library depends on the length of the target mRNA. This target molecule should be sufficiently long to accommodate at least 20 oligonucleotides, plus a two-base space between neighboring oligonucleotides. Other important parameters are the G-C content of the library, which should be as close to 45% as possible, and the uniqueness of the target sequence (to minimize binding of the probes to anything other than the target mRNA). Conveniently, the ‘Probe Designer’ tool (freely available at http://www.biosearchtech.com/stellarisdesigner/) will guide a user through probe design, taking all the important parameters into consideration.
The 3′-end of each library oligonucleotide is coupled to a single fluorophore (Steps 1–5), and the resulting labeled oligonucleotide is purified (Steps 6–10). Alternatively, the sets of fluorophore-coupled and purified oligonucleotides can be purchased from Biosearch Technologies under the brand name Stellaris, or from another company of choice.
The tissue is dissected right after euthanizing the animal, and it is immediately fixed (Steps 12–16). After a cryoprotection step (Step 16), frozen blocks are prepared (Steps 17–20). Cryosections of 5–10-μm thickness are mounted onto cleaned and poly-l-lysine–coated glass coverslips (Steps 21–26). The attached sections are fixed again, rinsed and permeabilized (Steps 24–26), and then rehydrated and hybridized with the probes and antibody overnight (Steps 30–32). An antibody can be used to stain the cell borders, so as to visualize tissue structure. However, if hybridization conditions are not optimal for staining with the necessary antibody, alternative counterstaining methods should be considered, such as staining with fluorescently labeled phalloidin (Step 35). After hybridization, sections on coverslips are washed, counterstained with nuclear dye and equilibrated with the imaging buffer. If phalloidin staining is chosen to visualize the cell border, this staining is performed at this point, and sections are subsequently rinsed and mounted (Steps 33–36).
Images are acquired using an inverted epifluorescence microscope equipped with an automated stage, a cooled back-illumined charge-coupled device (CCD) camera and a ×100 oil-immersion objective with a numerical aperture of 1.4 (Steps 37–39). At each position, a 5–10-μm-thick z-stack of images with 0.2–0.4 μm of distance between individual planes is acquired in order to collect the information from the whole thickness of the section.
The data are analyzed using MATLAB software (Steps 40–42). This software enables the automatic detection of the mRNA molecule spots, assists in manually segmenting cell borders and assigns the spots to individual cells, labeled according to position within the tissue relative to a reference (for instance, the bottom of an intestinal crypt). All automated and manual steps are integrated into ImageM, a straightforward and user-friendly GUI developed in our lab (Box 1 and Figs. 2 and 3).
Key analysis of the gene expression signatures created can include spatial mapping of gene expression, single-cell gene expression profiles and single-cell correlations of pairs of genes3. In intestinal crypts, the position of cells above the crypt apex defines a natural one-dimensional axis. For each gene, a spatial gene expression profile can be created. Pooling profiles from different analyzed crypts enables one to obtain the average gene expression profile and its variance, which we term a cryptogram (Fig. 1g,h). Such an analysis is particularly suited to genes broadly expressed in a consistent manner between crypts. Separate analysis can be applied to other genes that show a more variable expression pattern in terms of the position of the expressing cells along the crypt axis.
In the mouse intestine, some genes are highly specific to particular differentiated cell types, which can appear in variable positions in different crypts. Examples include markers for Goblet cells (Clca3) and Tuft cells (Dclk1, also known as Dcamkl1)3. An additional analysis, which can be performed using smFISH on tissue sections, is the characterization of single-cell gene expression signatures for such rarer cell types. The numbers of transcripts of other genes in such cells, defined by a marker of interest, can be compared to cells, which are directly above or below the cell of interest. This analysis stringently tests the upregulation or downregulation of particular genes in the cell type of interest, while controlling the spatial pattern of gene expression of the tested genes3. In addition, the Pearson or Spearman correlation coefficients of
the single-cell gene expression, correlated transcription patterns of pairs of genes may indicate the presence of common regulatory modules for these genes.
More advanced and sophisticated analyses can be performed on the resulting data sets. MATLAB codes are available upon request (see MATERIALS).
Finally, please note that although positive and/or negative controls for the expression of specific mRNAs are not covered in the procedure detailed below their use is likely to greatly improve the reliability of the protocol’s results. A more in-depth discussion on this subject is reported in the ANTICIPATED RESULTS (Step 42).
TROUBLESHOOTING
| Step | Problem | Possible reason | Solution | |
| 25-35 | Tissue sections detach from the glass | Insufficient cleaning and/or coating of the coverslips | Increase the length of coverslip sonication in detergent and ethanol to 20 min or add a second sonication step in ethanol. Make sure the coated coverslips are used immediately, or that they are stored in the desiccator only | |
| Insufficient drying at Step 23 | Increase drying time by 1–2 min. Try not to exceed 8 min | |||
| 37 | Tissue integrity is poorly preserved, with frequent rips and holes throughout the samples | Tissue is cryodamaged | Cryoprotection step is necessary (Step 16) | |
| Tissue is twisted and/or ripped | Sample was damaged during mounting (Step 36) | Do not let coverslip move during mounting | ||
| 38,39 | Insufficient signal intensity (low signal-to-background ratio) | Nonspecific probe binding owing to a suboptimal formamide concentration, leading to a high background fluorescence | Increase formamide concentration up to 25% (vol/vol) in wash and hybridization buffers (the required formamide concentration can be probe dependent) | |
| Insufficient number of binding probes due to the association of target mRNA with proteins | Subject tissue sections to proteinase K digestion (Steps 27–29). Increase proteinase K concentration, the duration of digestion and the incubation temperature, if needed, being careful not to affect tissue integrity | |||
| Insufficient number of binding probes due to a small oligonucleotide library size | If possible, increase the size of the library, such as by adding UTRs to the target sequence | |||
| High autofluorescence | Use red dyes, such as Cy5 instead of AlexaFluor594 and TMR. Autofluorescence is generally weaker at longer wavelengths, improving signal-to-noise | |||
| Photobleaching (signal quickly fades during acquisition) | Antiphotobleaching medium is not working properly | Prepare fresh GLOX buffer. Make sure glucose oxidase and catalase are fresh, properly reconstituted and stored. Activity of these enzymes is crucial to buffer performance. Use a lower intensity of excitation light, in particular when using laser illumination in (spinning-disk) confocal microscopy | ||
| Intensity of fluorescent dots decays toward the edges of the image, rendering accurate quantification difficult | Uneven illumination of the sample (vignetting) | Optimize alignment of the illumination, apply digital shading correction, or exclude peripheral areas from analysis |
For Comparison with other methods, Limitations, References, and more information please see the attached full manuscript.
Materials
MATERIALS
Animals. Tissue samples can be collected from a broad spectrum of animals, from Drosophila melanogaster to mammals.
DNA oligonucleotide probe set, with 3′-amine modification or coupled to a fluorophore (Biosearch Technologies, http://www.singlemoleculefish. com, or another company of your choice). Sequences of the three mRNAs detected in an example experiment (mRNAs for Ki67, Lgr5 and Wipi1) and the three corresponding probe libraries are listed in Supplementary Table 1
6-Carboxytetramethylrhodamine succinimidyl esterInvitrogen - Thermo FisherCatalog #C6123
Cy5 succinimidyl esterGE HealthcareCatalog #PA25001
Alexa Fluor 594 carboxylic acid succinimidyl esterInvitrogen - Thermo FisherCatalog #A37572
FITC mouse anti-E-cadherin antibodyBecton Dickinson (BD)Catalog #612130
Alexa Fluor 488 phalloidin (optional, see Experimental design)Life TechnologiesCatalog #A12379
Sodium bicarbonateMerck MilliporeSigma (Sigma-Aldrich)Catalog #S5761
DMSOMerck MilliporeSigma (Sigma-Aldrich)Catalog #D1435
Triethylammonium acetate buffer, 1 MMerck MilliporeSigma (Sigma-Aldrich)Catalog #90358
AcetonitrileMerck MilliporeSigma (Sigma-Aldrich)Catalog #34998
Sodium acetate, pH 5.5, 3 MAmbionCatalog #AM9740
Tris, 1 M, pH 8.0AmbionCatalog #AM9855G
Ethanol, absoluteJ.T. BakerCatalog #8025
Tris-EDTA, pH 8.0AmbionCatalog #AM9849
PBS, pH 7.4, RNase-free, 10×AmbionCatalog #AM9625
SSC, RNase-free, 20×AmbionCatalog #AM9763
Nuclease-free water or water filtered using a Milli-Q filtering systemAmbionCatalog #AM9932
Formamide, deionized, nuclease-freeAmbionCatalog #AM9342
Escherichia coli tRNARocheCatalog #10109550001
BSA, nuclease-free, 50 mg ml − 1AmbionCatalog #AM2616
Dextran sulfate, sodium saltMerck MilliporeSigma (Sigma-Aldrich)Catalog #D8906
Ribonucleoside-Vanadyl Complex (200 mM) - 10 mlNew England BiolabsCatalog #S1402S
Formaldehyde, 37% (wt/vol)Merck MilliporeSigma (Sigma-Aldrich)Catalog #F15587
Paraformaldehyde, 16% (wt/vol)Electron Microscopy SciencesCatalog #15710
Proteinase K solution, 20 mg ml − 1AmbionCatalog #AM2546
4′,6-Diamidino-2-phenylindoleMerck MilliporeSigma (Sigma-Aldrich)Catalog #D9542
Poly-l-lysine, 0.1% (wt/vol)Merck MilliporeSigma (Sigma-Aldrich)Catalog #P8920
Jung tissue freezing medium (Leica Microsystems) or OCT compound (TissueTek)
SucroseMerck MilliporeSigma (Sigma-Aldrich)Catalog #S7903
GlucoseMerck MilliporeSigma (Sigma-Aldrich)Catalog #G8270
Glucose oxidase Merck MilliporeSigma (Sigma-Aldrich)Catalog #G2133
Catalase Merck MilliporeSigma (Sigma-Aldrich)Catalog #C3515
TroloxMerck MilliporeSigma (Sigma-Aldrich)Catalog # 238813
RBS-35 Honeywell FlukaCatalog #83461
Rubber cement (Fixo Gum)Catalog #290117000
Dry ice
TetraSpeck microspheres, 0.2 μm 3M corporationCatalog #T-7280
Sodium hydroxide, 5 N
Methanol
STEP MATERIALS
Dry ice
Dry ice
Protocol materials
SSC, RNase-free, 20×AmbionCatalog #AM9763
6-Carboxytetramethylrhodamine succinimidyl esterInvitrogen - Thermo FisherCatalog #C6123
SucroseMerck MilliporeSigma (Sigma-Aldrich)Catalog #S7903
Alexa Fluor 594 carboxylic acid succinimidyl esterInvitrogen - Thermo FisherCatalog #A37572
DMSOMerck MilliporeSigma (Sigma-Aldrich)Catalog #D1435
TroloxMerck MilliporeSigma (Sigma-Aldrich)Catalog # 238813
Dry ice
Animals. Tissue samples can be collected from a broad spectrum of animals, from Drosophila melanogaster to mammals.
Nuclease-free water or water filtered using a Milli-Q filtering systemAmbionCatalog #AM9932
Formamide, deionized, nuclease-freeAmbionCatalog #AM9342
BSA, nuclease-free, 50 mg ml − 1AmbionCatalog #AM2616
Ribonucleoside-Vanadyl Complex (200 mM) - 10 mlNew England BiolabsCatalog #S1402S
Glucose oxidase Merck MilliporeSigma (Sigma-Aldrich)Catalog #G2133
TetraSpeck microspheres, 0.2 μm 3M corporationCatalog #T-7280
Dry ice
DNA oligonucleotide probe set, with 3′-amine modification or coupled to a fluorophore (Biosearch Technologies, http://www.singlemoleculefish. com, or another company of your choice). Sequences of the three mRNAs detected in an example experiment (mRNAs for Ki67, Lgr5 and Wipi1) and the three corresponding probe libraries are listed in Supplementary Table 1
Sodium acetate, pH 5.5, 3 MAmbionCatalog #AM9740
Tris, 1 M, pH 8.0AmbionCatalog #AM9855G
Cy5 succinimidyl esterGE HealthcareCatalog #PA25001
FITC mouse anti-E-cadherin antibodyBecton Dickinson (BD)Catalog #612130
Paraformaldehyde, 16% (wt/vol)Electron Microscopy SciencesCatalog #15710
Methanol
4′,6-Diamidino-2-phenylindoleMerck MilliporeSigma (Sigma-Aldrich)Catalog #D9542
Poly-l-lysine, 0.1% (wt/vol)Merck MilliporeSigma (Sigma-Aldrich)Catalog #P8920
GlucoseMerck MilliporeSigma (Sigma-Aldrich)Catalog #G8270
Catalase Merck MilliporeSigma (Sigma-Aldrich)Catalog #C3515
Sodium hydroxide, 5 N
Dry ice
Triethylammonium acetate buffer, 1 MMerck MilliporeSigma (Sigma-Aldrich)Catalog #90358
PBS, pH 7.4, RNase-free, 10×AmbionCatalog #AM9625
Dextran sulfate, sodium saltMerck MilliporeSigma (Sigma-Aldrich)Catalog #D8906
Jung tissue freezing medium (Leica Microsystems) or OCT compound (TissueTek)
Rubber cement (Fixo Gum)Catalog #290117000
Alexa Fluor 488 phalloidin (optional, see Experimental design)Life TechnologiesCatalog #A12379
Sodium bicarbonateMerck MilliporeSigma (Sigma-Aldrich)Catalog #S5761
Ethanol, absoluteJ.T. BakerCatalog #8025
Formaldehyde, 37% (wt/vol)Merck MilliporeSigma (Sigma-Aldrich)Catalog #F15587
RBS-35 Honeywell FlukaCatalog #83461
AcetonitrileMerck MilliporeSigma (Sigma-Aldrich)Catalog #34998
Tris-EDTA, pH 8.0AmbionCatalog #AM9849
Escherichia coli tRNARocheCatalog #10109550001
Proteinase K solution, 20 mg ml − 1AmbionCatalog #AM2546
Dry ice
Safety warnings
- Researchers should follow appropriate procedural regulations when performing animal experiments. These regulations may vary depending on the legislation of the country in which the research is being conducted. Commonly accepted guidelines, such as the ARRIVE21 guidelines, should be followed.
- Sodium hydroxide is highly caustic. Handle it carefully; wear protective gloves and a lab coat.
- Triethylammonium acetate is volatile; handle it in a fume hood while wearing protective gloves.
- Paraformaldehyde is highly toxic and carcinogenic. Handle it in a fume hood while wearing a lab coat and protective gloves.
- Formaldehyde is highly toxic and carcinogenic. Handle it in a fume hood while wearing a lab coat and protective gloves.
- Formamide is toxic and teratogenic. Handle it in a fume hood while wearing protective gloves and a lab coat.
Before start
REAGENT SETUP
Definition of the target sequence Choose a region of the target mRNA that is sufficiently long so that at least 20 probes—but preferably 48—each 20 bp long, can anneal to it, with at least two bases present between neighboring probe-binding sites (which amounts to at least 1,054 bp). If possible, avoid using untranslated regions (UTRs). ΔCRITICALThe presence of multiple binding sites for some of the probes in the library (other than the mRNA of interest) will lead to a high fluorescent background, which will decrease the quality of the FISH signal and may cause problems in spot detection and data analysis. If possible, exclude tandem repeats, microsatellites and other repeating sequences from the target sequence.
Design of the library of probe oligonucleotides Use the Probe Designer tool (http://www.biosearchtech.com/stellarisdesigner/) developed by Arjun Raj for this task. This tool is freely available, and no downloads are required; users only need to create their own login name and password, and then they can start using the Probe Designer immediately and free of charge. Follow stepwise instructions to upload the target sequence, to enter the information on the origin of the sequence, to specify the masking level and to create the list of oligonucleotide probes. Order the oligonucleotides on the list from Biosearch Technologies or another company of your choice. Please make sure to order sufficient quantities of the oligonucleotides in the most convenient format. In this protocol (probe-fluorophore coupling procedure, Steps 1–4), we use 10 μl of each oligonucleotide at the concentration of 100 μM (in water) per labeling reaction. In general, DNA oligonucleotides dissolved in water are stable for several years if they are stored below −20 °C. Repeated freeze-thaw cycles do not detectably affect the performance of our libraries in this method. However, please consider the manufacturer’s instructions for reconstitution and storage of the oligos.
Sodium bicarbonate, 1 M, pH 8.0 In a 50-ml Falcon tube, dissolve 4.2 g of the powder in 40 ml of distilled or deionized, nuclease-free water. Adjust the pH to 8.0 with 5 N sodium hydroxide. Adjust the volume to 50 ml. !CAUTION Sodium hydroxide is highly caustic. Handle it carefully; wear protective gloves and a lab coat. ΔCRITICAL prepare this solution fresh and discard it after 2 weeks. Alternatively, divide the solution into 5-ml portions and freeze them. These aliquots can be stored at −20 °C for several months.
Sodium hydroxide, 5 N Dissolve 10 g of sodium hydroxide pellets in 50 ml of distilled or deionized water. Add the pellets in small quantities to the water, allow them to dissolve, cool the solution down, and then add the next portion of pellets. This solution can be stored indefinitely at room temperature (20–25 °C). !CAUTION Sodium hydroxide is highly caustic. Handle it carefully and wear gloves and a lab coat. Sodium hydroxide pellets absorb water from the air; weigh them quickly using glass or plastic weighing cups (not weighing paper). Dissolving sodium hydroxide is highly exothermic; keep the tube on ice. Do not pour water onto the pellets, as it will boil and splash out of the tube.
Sodium bicarbonate, 0.1 M, pH 8.0 Combine 5 ml of the 1 M sodium bicarbonate and 45 ml of distilled or deionized nuclease-free water. ΔCRITICAL Freshly prepare this solution each time or make sure to use it within 1 week of preparation.
HPLC buffer A Mix 100 ml of 1 M triethylammonium acetate with 900 ml of deionized water. Freshly prepare the buffer every week and store it at room temperature. !CAUTION Triethylammonium acetate is volatile; handle it in a fume hood while wearing protective gloves.
HPLC buffer B HPLC buffer B is undiluted acetonitrile. !CAUTION This is an organic solvent; handle it in a fume hood while wearing protective gloves.
DAPI, 20 mg ml −1 Add 500 μl of deionized, nuclease-free water to 10 mg of DAPI powder. Divide the solution into 20- to 50-μl portions and store them at −20 °C for up to several years.
DAPI, 100 μg ml − 1 Add 990 μl of deionized, nuclease-free water to 10 μl of 20 mg ml −1 DAPI. Store the solution at 4 °C and use it within a month.
Formaldehyde, 3.7% (vol/vol) For a 50-ml solution, combine 40 ml of distilled or deionized, nuclease-free water, 5 ml of 37% formaldehyde (wt/vol) and 5 ml of 10× PBS. Freshly prepare the solution just before use;
alternatively, prepare it in advance, divide it into 25- to 50-ml portions and freeze the portions. Store formaldehyde at −20 °C for up to several months and thaw it just before use. !CAUTION Formaldehyde is highly toxic and carcinogenic. Handle it in a fume hood while wearing a lab coat and protective gloves.
Paraformaldehyde, 4% (vol/vol) Combine 10 ml of 16% (wt/vol) paraformaldehyde, 4 ml of 10× PBS and 26 ml of deionized, nuclease-free water. Freshly prepare the solution just before use or prepare it in advance, divide it into 25- to 50-ml portions and store the portions at −20 °C for up to several months. Thaw the solution in a 37 °C water bath just before use. !CAUTION Paraformaldehyde is highly toxic and carcinogenic. Handle it in a fume hood while wearing a lab coat and protective gloves.
Cryoprotecting solution Combine 5 ml of 10× PBS, 5 ml of 37% (vol/vol) formaldehyde and 15 g of sucrose. Add distilled or deionized, nuclease-free water to a final volume of 50 ml. Incubate the solution with gentle agitation at room temperature to allow the sucrose crystals to dissolve. Freshly prepare this solution just before use or prepare it in advance, divide it into 25- to 50-ml portions and store the portions at −20 °C for up to several months. Thaw the solution in a 37 °C water bath just before use. !CAUTION Formaldehyde is highly toxic and carcinogenic. Handle it in a fume hood while wearing a lab coat and protective gloves.
RBS-35, 2% (vol/vol) Combine 20 ml of RBS and 980 ml of distilled or deionized water. Store the solution at room temperature for up to several months.
Ethanol, 70% (vol/vol) Combine 700 ml of absolute ethanol and 300 ml of distilled or deionized nuclease-free water. Store it indefinitely in a tightly closed container at room temperature.
Poly-l-lysine Combine 20 ml of 0.1% (wt/vol) poly-l-lysine solution and 180 ml of deionized or distilled, nuclease-free water. Freshly prepare the solution before use and discard the remaining solution after use.
Ribonucleoside vanadyl complex Incubate the purchased solution in a 65 °C water bath for 10 min, divide it into 100-μl aliquots and store them at −20 °C for up to several months.
Glucose oxidase, 100× stock Dissolve the enzyme in 50 mM sodium acetate (pH 5.0), to a concentration of 3.7 mg ml −1. Divide the solution into 50- to 100-μl aliquots and store them at −20 °C for up to several months.
Trolox, 200 mM (100× stock) Dissolve 5 mg of the Trolox powder in 1 ml of pure ethanol. Divide the solution into 50- to 100-μl portions and store them at −20 °C for up to several months.
Glucose 10% (wt/vol) Dissolve 5 g of glucose in 50 ml of deionized, nuclease-free water. Sterilize the solution by passing it through a 0.2-μm syringe filter and store it at 4 °C for up to a few weeks.
GLOX buffer Combine 100 μl of Tris (1 M, pH 8.0), 1 ml of 20× SSC, 400 μl of 10% (wt/vol) glucose and 8.5 ml of deionized, nuclease-free water. This solution needs to be prepared freshly at least every week. For best results, prepare the solution just before use.
GLOX buffer with Alexa Fluor 488 phalloidin Reconstitute Alexa Fluor 488 phalloidin as described in the manufacturer’s manual. Briefly, add 1.5 ml of methanol to 300 units of Alexa Fluor 488 phalloidin and dissolve it by passing it several times through a pipette tip. Add 4 μl of the reconstituted Alexa Fluor 488 phalloidin to 2 ml of GLOX buffer. Freshly prepare this solution before use.
Imaging buffer Add 1 μl of glucose oxidase (stock 3.7 mg ml − 1), 1 μl of catalase (vortex catalase suspension just before pipetting, do not spin) and 1 μl of Trolox (200 mM stock) to 100 μl of GLOX buffer. Use this solution immediately, although it may be stored on ice or at 4 °C for up to 8 h.
Formamide Before opening the formamide bottle, bring it to room temperature. Divide the solution into 10- to 50-ml portions and store them in the dark at 4 °C for up to several months. !CAUTION Formamide is toxic and teratogenic. Handle it in a fume hood while wearing protective gloves and a lab coat.
Wash buffer, 10% (vol/vol) formamide Combine 5 ml of 20× SSC, 5 ml of formamide and 45 ml of deionized, nuclease-free water. Freshly prepare the solution before use. !CAUTION Formamide is toxic and teratogenic. Handle it in a fume hood while wearing protective gloves and a lab coat.
Wash buffer with DAPI Prepare the wash buffer, 10% (vol/vol) formamide, as described above. Add 50 μl of DAPI (100 μg ml − 1 stock) to 10 ml of wash buffer (1:200 dilution). Use this solution immediately. !CAUTION Formamide is toxic and teratogenic. Handle it in a fume hood while wearing protective gloves and a lab coat.
Coupling FISH probe libraries to fluorophores
Coupling FISH probe libraries to fluorophores
Pool the oligonucleotides belonging to one library (a set of oligonucleotides that can hybridize to a single mRNA molecule) in a 1.8-ml microcentrifuge tube (10 μl of each 100 μM solution per coupling reaction). The final solution volumes will vary depending on oligonucleotide library size. For the experiment described as a template in this protocol, the probe library against mouse Lgr5 (Supplementary Table 1) consists of 96 oligonucleotides, and therefore the final volume will be 960 μl. The Ki67 probe library is of the same size; therefore, the final volume will be the same, that is, 960 μl. By contrast, the Wipi1 library consists of 48 oligonucleotides; therefore, the final volume of the pooled oligonucleotides will be 480 μl. As the solvent will be evaporated, the volume does not need to be adjusted. Mix the solution and dry it in a vacuum concentrator to complete evaporation. To reduce the drying time, split the oligonucleotide mixture into small portions (50–100 μl) and heat them up to 60 °C.
60 °C
Resuspend each dried library in 70 μl of 0.1 M sodium bicarbonate, pH 8.0. In the experiment detailed here, prepare three tubes containing the libraries: the first against Lgr5, the second against Ki67 and the third against Wipi1.
70 µL 0.1 M sodium bicarbonate
Note
CRITICAL STEP A pH value of 8.0–8.5 is a prerequisite for an efficient coupling.
Resuspend the dry Alexa Fluor 594 and Cy5 in 0.1 M sodium bicarbonate. When you reconstitute the dyes, consider that one vial of Alexa Fluor 594 (1 mg) contains enough dye for up to 10 coupling reactions (enough to label 10 separate, 48-oligonucleotide-sized libraries, with the same color), whereas one vial of Cy5 contains enough dye for up to 5 coupling reactions (i.e., sufficient to label up to 5 separate, 48-oligonucleotide-sized libraries, with the same color). The volume of 0.1 M sodium bicarbonate used for dye reconstitution depends on the numbers of libraries or oligos that have to be coupled to the same dye. The volume for labeling one 48-oligonucleotide library is 30 μl. For example, to couple five libraries of 48 oligos to Cy5, add 150 μl of buffer to the dry dye, resuspend it and use 30 μl per library reaction (each library in a separate tube). To couple ten separate 48-oligo libraries to Alexa Fluor 594, resuspend the dye in 300 μl of buffer and use 30 μl per reaction. To couple a larger library, the quantity of dye needs to be increased. For example, a 96-oligonucleotide library will need 60 μl of the dye solution. In the present experiment, Ki67 library requires 60 μl of the Alexa Fluor 594 solution, and the Wipi1 library requires 30 μl of the Cy5 solution.
Note
CRITICAL STEP Once they have dissolved, the dyes quickly lose reactivity, so please proceed to Step 5 of the PROCEDURE quickly. The reconstituted dye cannot be stored and used later; therefore, all unused dye must be discarded.
Dissolve 1–3 mg of TMR powder (measuring 1 mg of TMR powder is difficult; any amount between 1 and 3 mg will suffice) in 5 μl of DMSO. As with other dyes, add 30 μl of 0.1 M sodium bicarbonate per coupling reaction to the dye. For example, to couple three oligonucleotide libraries, add 90 μl to the dry dye, resuspend it and use 30 μl per reaction. In the present experiment, 60 μl of TMR solution is required to label the Lgr5 library, which consists of 96 oligonucleotides.
1 mg TMR powder
5 µL TMR powder
30 µL 0.1 M sodium bicarbonate per coupling reaction
Add 30 μl of the dye solutions prepared in Steps 4 and 5 to each tube containing the pooled probe oligonucleotides (70 μl, one library per tube). Mix Alexa Fluor 594 with the Ki67 probe library mix, Cy5 with the Wipi1 probe library mix and TMR with the Lgr5 probe library mix. The final volume of the reaction will be 100–130 μl. The final volumes of the coupling reactions for the probes for Ki67 and Lgr5 are 130 μl each, whereas the Wipi1 coupling reaction volume is 100 μl. Mix and incubate the tubes at room temperature for 1–3 h, protecting the tubes from light (there is no need to agitate the tubes).
30 µL Dye solutions
01:00:00 Incubation
Probe precipitation, washing and purification
Probe precipitation, washing and purification
Add a 10–12% volume of 3 M sodium acetate (pH 5.5) and at least 2.5 volumes of 100% ethanol to each tube and mix the contents. In the experiment described, add 15 μl of 3 M sodium acetate (pH 5.5) and 350 μl of ethanol to each Lgr5 and Ki67 libraries’ coupling reactions; add 12 μl of 3 M sodium acetate (pH 5.5) and 300 μl of ethanol to the Wipi1 coupling reaction. Incubate the mixture at −80 °C for at least 1.5 h and up to overnight.
15 µL 3 M sodium acetate
350 µL Ethanol
12 µL 3 M sodium acetate
300 µL Ethanol
-80 °C Incubation
01:30:00 Incubation
Centrifuge the tubes at maximum speed in a tabletop centrifuge at 4 °C for 20 min.
4 °C Centrifugation
00:20:00 Centrifugation
Remove and discard the supernatant, and then wash the pellet with cold 70% (vol/vol) ethanol. Do not vortex. Centrifuge the mixture at maximum speed in a tabletop centrifuge at 4 °C for 10 min. Remove the supernatant carefully and completely. Air-dry the pellet.
Note
PAUSE POINT The dried pellet can be stored for up to a year at −20 °C.
4 °C Centrifugation
00:10:00 Centrifugation
HPLC-purify the fluorophore-labeled oligonucleotides. Various setups can be used for this purpose. We use a Gilson GX271 liquid handler with a C18 column and a UV-visible 155 spectrometer. We use a buffer A composed of 0.1 M triethylammonium acetate (pH 6.5), and a buffer B composed of acetonitrile (pH 6.5) for the gradient, which ranges from 7 to 23% buffer B in buffer A over the course of 45 min, followed by 80% buffer B for 5 min and another 15 min with 7% buffer B before running another sample.
Note
CRITICAL STEP Use a pre-column frit (A-332X) to prevent clogging.
00:45:00 7 to 23% buffer B in buffer A
00:05:00 80% buffer B
00:15:00 7% buffer B
Dry the samples in a vacuum concentrator, first by using a KNF Teflon pump for 2–5 h to remove acetonitrile, and then by drying overnight with a vacuum pump. Labeling efficiency routinely achieved is close to 90%. The resulting yield of labeled oligonucleotides is ~200 μg.
02:00:00 KNF Teflon pump
Probe reconstitution and storage
Probe reconstitution and storage
Reconstitute the labeled oligonucleotide library in 30–50 μl of TE (pH 8.0). This probe stock (2–4 μg μl − 1) is normally used at a final dilution of 1:1,000–1:5,000. It is convenient to make at this stage a 1:100 intermediate dilution (in TE buffer, pH 8.0). To determine which probe dilution will give the highest FISH signal intensity with the lowest background fluorescence, we recommend doing a coarse probe titration by adding 2–10 μl of the intermediate dilution to 100 μl of hybridization buffer (used for one 22 × 22 mm coverslip with tissue sample on it), to find an optimal dilution range for each probe library.
Note
PAUSE POINT Both the stock and the 1:100 dilution can be stored at −20 °C for several months. However, we do not recommend storing working dilutions.
50 µL TE (pH 8.0)
2 µL Intermediate dilution
100 µL Hybridization buffer
Tissue dissection, fixing and cryoprotection
Tissue dissection, fixing and cryoprotection
Place enough 3.7% (vol/vol) formaldehyde on ice for more than 20 min. The solution’s volume should be more than fivefold that of the tissue to be fixed.
Note
CRITICAL Please note that consulting basic guides on mouse tissue pathology [22] (http://ctrgenpath.net/static/atlas/mousehistology/), dissection [23] and cryoprotection (http://mousepheno.ucsd.edu/histo.shtml) might be helpful.
Note
CRITICAL STEP The fixative solution (formaldehyde, 3.7% (vol/vol)) should be chilled, and collecting tubes and instruments should be prepared before euthanizing the mouse and dissecting the organ.
00:20:00 3.7% (vol/vol) formaldehyde on ice
To repeat the experiment detailed herein, collect the most proximal section of the mouse intestine (duodenum) into a 15-ml conical Falcon tube containing 6–7 ml of fixative solution.
7 mL Fixative solution
Dissect the tissue, and immediately place it into the chilled 3.7% (vol/vol) formaldehyde. Larger organs may be cut to smaller (3 × 3 × 3 mm) pieces to facilitate quick penetration of the fixative solution into the depth of the tissue. Hollow organs, such as stomach, intestine or heart, should be flushed with cold fixative solution or cut open before they are submerged in the fixative solution in the tube. In this experiment, we cut the duodenum into 3- to 4-cm pieces, flushed them with cold 3.7% (vol/vol) formaldehyde and then transferred the pieces into a 15-ml conical tube containing 6–7 ml of cold 3.7% (vol/vol) formaldehyde.
Note
CRITICAL STEP Keep the dissected tissue on ice at all times. Perform all necessary tissue manipulations (such as flushing with fixative or cutting) in a sterile Petri dish with 2–3 ml of cold PBS in it, placed on ice.
To achieve tissue fixation, incubate the tubes for ~3 h at 4 °C with gentle agitation. If you are working with a thick tissue piece, consider longer incubation times, as the fixative penetration speed is ~1 mm h-1. In the current experiment, fix the mouse duodenum for 3 h.
4 °C Fixation
03:00:00 Fixation
(Optional) To improve the structural integrity of the sample, transfer the tissue into the tube filled with the cryoprotection solution (volume requirements are the same as for fixation). Incubate the mixture overnight at 4 °C with gentle agitation. We strongly encourage experimenters to perform this step.
4 °C
Preparation of frozen blocks
Preparation of frozen blocks
Label plastic molds for frozen block preparation with a permanent marker. If necessary, make marks on the mold to indicate the orientation of the tissue. Place the molds on ice, fill them with the Jung tissue freezing medium and let them chill for >5 min. Prechill 10 ml of PBS in a Petri dish placed on ice. Place another empty, open Petri dish on ice for manipulating the tissue. Prepare pieces of dry ice (flat slabs are the most convenient) in an appropriate container. We use styrofoam boxes with lids.
00:05:00 Chill
10 mL PBS
Dry ice
Remove the tissue from the fixative or cryoprotection solution, quickly rinse it by dipping it in the dish containing cold PBS, and then place it in the empty dish on ice. Cut the tissue to smaller pieces if necessary: for instance, we cut the intestine to pieces 3–5 mm long. The size of the embedded tissue pieces may vary depending on the tissue type and the researcher’s preferences.
Note
CRITICAL STEP Keep the tissue cold at all times.
Place the tissue into the chilled molds filled with freezing medium. If necessary, fill the tissue cavities with medium by using a pipette. Orient the tissue as required for the most efficient sectioning. We position the intestine pieces to optimize obtaining cross-sections of the organ. When you pipette the freezing medium, avoid making air bubbles. If some bubbles are accidentally introduced, remove them by sucking them with a pipette into a 200-μl tip.
Place the molds with the tissue in Jung tissue freezing medium on dry ice. Allow it to solidify. If necessary, cover still exposed parts of the tissue with more medium. Allow it to stand on dry ice until the block is fully white and solid. Wrap the mold with the block in aluminum foil and label it.
Note
PAUSE POINT The mold can be stored at −80 °C for years. Please note that after freezing, the block should never be allowed to thaw.
Cryosectioning, fixation and permeabilization of the sections
Cryosectioning, fixation and permeabilization of the sections
Prepare the cover glasses and sharp-tip forceps (Equipment Setup). Use 22 × 22 (no. 1) coverslips and place them into six-well culture plates. Prepare dry ice (flat slabs are most convenient) in an appropriate container. Use styrofoam boxes with lids.
Set up the cryostat (Equipment setup) and clean the stage with 70% (vol/vol) ethanol. Use a new knife for sectioning. Place the tissue blocks inside the chamber, mount them and allow them to equilibrate to the temperature (15–30 min in case of small blocks, up to 1 h, if the blocks are large).
00:15:00 Small blocks
01:00:00 Large blocks
Note
CRITICAL STEP Do not let the blocks thaw or even soften.
Cut 5- to 10-μm sections and mount them onto the coated coverslips. Use sharp-ended forceps to pick up the coverslips. Let the sections air-dry for 3–5 min.
00:03:00 Air-drying
Note
CRITICAL STEP Drying time should be sufficient to allow the water present in the freezing medium to evaporate (insufficient drying will cause section detachment), but drying time should not exceed 5 min, as doing so may lead to RNA degradation. However, in a humid environment, it may be necessary to increase drying time to up to 8 min.
Place dried sections (on coverslips) into the wells of six-well plates on dry ice. It is important to keep the lids of the plate and the ice box closed, so as to prevent contact with warm room air and formation of water condensate on the coverslips. This condensation may cause section detachment. Keep sections on dry ice during sectioning procedure until fixation (Step 25). Although sections can be kept at −80 °C or on dry ice for weeks, temperature in a freezer can accidentally rise (as a result of opening the door, for instance) and the sections may temporarily thaw, which will affect RNA integrity. We therefore do not recommend pausing the protocol at this point.
Remove the plates with sections on coverslips from the dry ice. Fix the sections by adding 2 ml of room-temperature 4% (wt/vol) paraformaldehyde into each well. At this stage, add 4% (wt/vol) paraformaldehyde dropwise, not directly onto each section, so as not to dislodge it from the glass. Incubate the sections for 15–20 min at room temperature.
2 mL 4% (wt/vol) paraformaldehyde
00:15:00 Incubation
Safety information
Paraformaldehyde is highly toxic and carcinogenic; work should be carried out in a fume hood while you are wearing a lab coat and protective gloves.
Rinse the sections twice with room-temperature PBS, add 3 ml of 70% (vol/vol) ethanol to each well and incubate the mixture at 4 °C for a minimum of 1 h (seal the plates with Parafilm to prevent ethanol evaporation). Proceed directly to Step 30, unless the resulting FISH signal is suboptimal, which may become evident at Steps 37 and 38.
3 mL 70% (vol/vol) ethanol
4 °C
01:00:00 Incubation
Note
CRITICAL STEP Add liquids to the sections slowly, so as not to dislodge them from the glass.
PAUSE POINT Sections can be stored in 70% (vol/vol) ethanol at 4 °C for several months.
Proteinase K digestion (optional)
Proteinase K digestion (optional)
Rehydrate the sections by placing them into 2× SSC for 3–5 min.
Note
CRITICAL Steps 27–29 are optional, and proteinase K digestion should only be implemented if an insufficiently strong FISH signal is detected at Steps 37 and 38.
00:03:00 Rehydration
Replace 2× SSC with proteinase K working solution and incubate the resulting mixture at room temperature for 5–10 min.
00:05:00 Incubation
Wash the sections twice with 2× SSC.
Hybridization
Hybridization
Pipette 2 ml of freshly made wash buffer into each well of a six-well plate, one well per coverslip, to rehydrate. Rehydrate the sections by picking them up with sharp-tip forceps and placing them, section side up, into wash buffer for 5 min.
2 mL Wash buffer
00:05:00 Wash buffer
Safety information
Formamide is toxic and teratogenic. Handle it in a fume hood while you are wearing a lab coat and protective gloves.
Thaw the required quantity of hybridization buffer (100 μl per 22- × 22-mm coverslip) and set up the hybridization reactions according to option A, if you are counterstaining cell membranes with an antibody, or according to option B, if you are using Alexa Fluor 488 Phalloidin for that purpose instead (see also relevant discussion in the Experimental design section).
(A) Hybridization in the presence of an antibody for cell counterstaining
(i) In a 1.5-ml microcentrifuge tube, combine the hybridizaton buffer, probes (from 1:100 intermediate dilutions, to the final dilution of 1:1,000–1:5,000) and the antibody solution. For instance, the mouse FITC–anti-E-cadherin antibody (1:100 dilution) may be used for counterstaining cell membranes in the intestine. In the case of the specific experiment described herein, prepare 1:100 intermediate dilutions in TE buffer of the Lgr5-TMR library, Ki67-Alexa Fluor 594 library and Wipi1-Cy5 library. Next, add 2 μl of the intermediate dilution of the Lgr5-TMR intermediate dilution, 5 μl of each intermediate dilution of libraries Ki67-Alexa Fluor 594 and Wipi1-Cy5 and 1 μl of the mouse FITC–anti-E-cadherin antibody (undiluted) to 100 μl of hybridization buffer. Mix and spin the mixture briefly.
(B) Hybridization with cell counterstaining without an antibody
(i) Proceed as detailed in Step 31A. However, prepare the hybridization mix without the antibody.
Safety information
Formamide is toxic and teratogenic. Handle it in a fume hood while wearing wearing a lab coat and protective gloves.
Spread a piece of Parafilm on the bottom of a Petri dish. Place 100-μl drops of the hybridization mix onto a piece of Parafilm (Fig. 4a). Invert the coverslips, tissue side down, onto the drops (Fig. 4b,c). Avoid producing air bubbles in the process. Close the lid of the Petri dish. It is not necessary to humidify the inside of the dish. Perform hybridization overnight at 30 °C.
100 µL Hybridization mix
30 °C Hybridization
Washing and mounting
Washing and mounting
Add drops of wash buffer under the coverslips and then lift them off the hybridization solution carefully. Place the coverslips back into the six-well plate, rinse them with 2 ml of wash buffer per well, add fresh wash buffer and wash them by incubating at 30 °C for 30 min. Please note that agitating the plate is not necessary.
2 mL Wash buffer
30 °C Incubation
00:30:00 Incubation
Safety information
Formamide is toxic and teratogenic. Handle it in a fume hood while wearing a lab coat and protective gloves.
Replace the first wash buffer with fresh wash buffer containing DAPI and incubate it for 30 min at 30 °C.
30 °C Incubation
00:30:00 Incubation
Safety information
Formamide is toxic and teratogenic. Handle it in a fume hood while wearing a lab coat and protective gloves.
Equilibrate the hybridization mixtures according to option A, if you opted to stain cell borders with mouse FITC–anti- E-cadherin antibody, or according to option B, if you opted to use Alexa Fluor 488 Phalloidin.
(A) Equilibration in the presence of an antibody for cell counterstaining
(i) Place each coverslip in 2 ml of GLOX buffer (with no enzymes) and allow it to equilibrate for >5 min.
(B) Equilibration in the presence of Alexa Fluor 488 phalloidin
(i) Place each coverslip in 2 ml of GLOX with Alexa Fluor 488 phalloidin. Incubate the coverslips for 15 min, and then transfer each coverslip to 2 ml of fresh GLOX buffer with no enzymes. Incubate the coverslips for 5 min.
00:15:00 Incubation 15 min
00:05:00 Incubation 5 min
Note
PAUSE POINT The hybridized tissue sections can be kept in GLOX buffer, without enzymes, at 4 °C for up to 24 h without any adverse effects on smFISH signal quality.
Place 15 μl of imaging buffer onto a microscope slide. Lay each coverslip, hybridized tissue section down, onto the drop. Avoid producing air bubbles in the process. Suck any excess fluid off with a piece of filter paper, being careful not to move the coverslip. Seal the edges with rubber cement, and let it air-dry for 10–20 min.
15 µL Imaging buffer
00:10:00 Air-drying
Note
CRITICAL STEP Imaging buffer must be freshly prepared and used immediately or kept on ice for a maximum of 8 h.
Microscope setup and data collection
Microscope setup and data collection
Mount the slide onto the microscope stage and find the right focal plane for the z-middle of the tissue section under DAPI illumination. Define xy positions to image the regions of interest (intestinal crypts, in our case). Set up acquisition of z-stacks, covering a thickness of the tissue in an optimal way. In the case of intestines, acquire 2–5-μm-thick stacks.
Note
CRITICAL STEP The distance between individual optical planes should be 0.2–0.4 μm
Acquire z-stacks at each xy position in each wavelength channel, starting with Cy5, then Alexa Fluor 594, then TMR, then FITC and then DAPI. Typical illumination times are 0.5–3 s for Cy5, Alexa Fluor 594 and TMR, 50–200 ms for FITC and 20–30 ms for DAPI. However, illumination times should be adjusted for each slide to obtain optimal results. Please avoid excessive exposure times, which will be defined by an observed saturation of pixel intensities. For the experiment described here, implement the following exposure times: 1 s for Cy5, 3 s for Alexa Fluor 594, 500 ms for TMR, 200 ms for FITC and 25 ms for DAPI. The number and relative locations of xy positions at which z-stacks will be acquired are determined by tissue architecture, by the number of cells expressing mRNA of interest and by the numbers of mRNA molecules of interest. The number of acquired xy positions should be optimal to enable statistical analysis, and it may vary in different tissues and for different mRNA species. The xy positions at which z-stacks will be acquired can be chosen either manually or automatically (for example, scan and image certain areas of the sample using the imaging software’s scanning tool; we use Metamorph, in this case). In the case of intestinal crypts, record images of 30–40 crypts in stacks of 2–3 μm thickness, with 0.3 μm between individual planes.
Note
CRITICAL STEP In any smFISH experiment, make sure to always acquire light emission data from less-stable fluorophores (or from probes, experimentally shown to give weaker signal, fused to any fluorophore) first. With the channels used in this protocol, we recommend the following order: Cy5, Alexa Fluor 594, TMR, FITC, DAPI, bright field. This order can be modified empirically, by evaluating the performance of each fluorophore-fused probe library in a single-color experiment. Please note that here we acquire FITC data after Cy5, Alexa Fluor 594 and TMR, even though FITC is less stable than the other fluorophores. The reason for doing so is that FITC is only used for cell boundary staining, and not for the more sensitive smFISH data acquisition. This order of acquisition is chosen to minimize the effect of bleaching, which results from repeated illumination during stack acquisition in other channels.
Record data in TIFF format. For automated FISH signal detection (single-molecule RNA counting), we use ImageM, our custom-written MATLAB-based software (see Box 1 in Guidelines for directions). Automatic transcript detection uses algorithms described previously2,3. The spot stack images are first filtered with a 3D Laplacian of Gaussian filter of 15 pixels in size and a s.d. of 1.5 pixels. The number of connected components in binary-thresholded images is then recorded for a uniform range of intensity thresholds, and the threshold for which the number of components was least sensitive to threshold selection is used for spot detection. At this stage, dots appearing in multiple fluorescence channels are automatically detected and can be removed by the software (Box 1). These dots are often nonspecific aggregates of probes or autofluorescent organelles. We advise removing such dots. Please note that images from different fluorescence channels may have a slight shift, depending on the imaging setup. The use of zero-shift filter cubes can alleviate this problem, however, and if a shift exists it can be corrected by software. Beads fluorescing in all channels can be used to detect and characterize the shifts between channels. Additional filtering of spots, which are not single transcripts, can include removal of spots exceeding a certain size threshold, the size expected from a diffraction-limited object, which is associated with a single mRNA molecule.
Automatic cell segmentation can be performed by relying on DAPI channels in some cases (e.g., using CellProfiler24 or similar software). In tissues such as intestinal crypts, the nonisotropic cell morphology limits the precision of such approach and requires manual cell border segmentation. We advise readers to use our ImageM software (Box 1) to perform manual cell segmentation, on the basis of both DAPI and E-cadherin antibody staining. In intestinal crypt tissue, we segment only one cell for each vertical position along the crypt-villus axis, as was done in the example experiment (Fig. 1f).
Assign the detected mRNA-molecule spots to cells to generate single-cell gene expression signatures. Use either the discrete integer number of transcripts per cell or, alternatively, the transcript concentration, which is defined as the number of transcripts divided by cell volume and estimated as the segmented cell area times the number of analyzed optical z-sections (planes). Please note that transcript concentration estimation can be a more robust feature when you pool results from several experiments in which the numbers of analyzed planes are different.
Perform gene expression analysis of choice. An example of such analysis in the intestinal crypt is creation of spatial expression profiles along the axis of the crypt (bottom to top). The result of the example experiment (a typical result of such profiling of multiple crypts) is represented in Figure 1g,h.
Expected result
This protocol enables users to implement smFISH in a variety of tissues. It enables assessing the absolute numbers of mRNA molecules in single cells in the context of a tissue. The mRNA expression is automatically quantified, semiautomatically assigned to single cells in the tissue and analyzed, resulting in high-resolution gene expression maps of the tissue.
Results of a typical three-color smFISH experiment (with mouse intestine as a target tissue) are shown in Figure 1. Diffraction-limited spots are clearly observed by microscopy. First, the smFISH spots are recognized and counted by our MATLAB software (ImageM). Only the signal coming from the binding of the probe library to a single mRNA molecule (specific signal) is recognized and taken into account by the software. A specific signal (i.e., coming from the binding of the probe library to a single mRNA molecule) must meet the following criteria: (i) there must be sufficient signal-to-background ratio (of ~2 in a typical experiment); (ii) the size of the signal spot must correspond to the size of a diffraction-limited spot (200–300 nm, which in our setup corresponds to ~4–6 pixels); and (iii) the signal spot must appear only in the appropriate fluorescence channel. Fluorescent spots that appear in multiple channels are filtered out by ImageM. For example, in the experiment used as template in this protocol, large blots of fluorescence, resulting mostly from nonspecific binding of the probes, are present (Fig. 1a–c, arrows) alongside specific signals (Fig. 1a–c, arrowheads). These spots are clearly distinguishable from the specific signal by their size (larger than diffraction-limited spots), and by the fact that, unlike specific signal, they usually appear in all three fluorescent channels. Therefore, these nonspecific fluorescent spots are not present in the software-generated representation of the results (Fig. 1f).
The use of positive and negative controls will greatly improve reliability of the protocol. Fluorescence spots, resulting from nonspecific binding of the probes, will be present in both controls. Different cell types, one expressing and another not expressing the mRNA of interest, can serve as positive and negative controls, respectively. For example, intestinal crypt base columnar cells express Lgr5 mRNA, whereas enterocyte or Goblet cells do not express Lgr5 mRNA (Fig. 1a). Another example includes Ki67, a proliferation marker expressed only in dividing cells within the crypt, and not in the differentiated cells in the villi (Fig. 1b).
Tissues known to express or not to express an mRNA of interest can be used as a positive and negative control, respectively. We present here an example of such controls for the smFISH probe library against Bmi1 mRNA, where Bmi1 is a gene expressed in intestinal crypts. A wild-type crypt is used as a positive control, and a crypt derived from a Bmi1-knockout mouse is used as a negative control (Fig. 5a)3. smFISH against an exogenous transcript (such as GFP), expressed in mammalian cells or tissue, can serve as an excellent positive or negative control for the protocol. GFP is especially convenient in this context, as its presence is easily detected by green fluorescence. GFP-positive fluorescent cells also contain GFP mRNAs (Fig. 5b, arrows), whereas cells devoid of fluorescence are GFP transcript negative (Fig. 5b, arrowheads)3.
Next, manual definition of cell borders (cell segmentation) is performed (Figs. 1f and 2). Although the procedure is made user-friendly by our GUI (ImageM), it is the most time-consuming step in the process of the analysis. Finally, each of the segmented cells is assigned a number and detected spots are assigned to the cells. On the basis of this assignment, expression of the three genes can be presented in the context of the crypt (Fig. 1g,h).