A Sensitive Alternative for MicroRNA In Situ Hybridizations Using Probes of 2′-O-Methyl RNA + LNA
Abstract
The use of short, high-affinity probes consisting of a combination of DNA and locked nucleic acid (LNA) has enabled the specific detection of microRNAs (miRNAs) by in situ hybridization (ISH). However, detection of low–copy number miRNAs is still not always possible. Here the authors show that probes consisting of 2′-O-methyl RNAs (2OMe) and LNA at every third base (2:1 ratio), under optimized hybridization conditions, excluding yeast RNA from the hybridization buffer, can provide superior performance in detection of miRNA targets in terms of sensitivity and signal-to-noise ratio compared to DNA + LNA probes. Furthermore, they show that hybridizations can be performed in buffers of 4M urea instead of 50% formamide, thereby yielding an equally specific but nontoxic assay. The use of 2OMe + LNA–based probes and the optimized ISH assay enable simple and fast detection of low–copy number miRNA targets, such as miR-130a in mouse brain.
MicroRNAs (miRNAs) are a class of short (18–25 nt) non-coding RNAs involved in posttranscriptional regulation of gene expression. Several miRNAs have been linked to various types and subtypes of cancers (Sempere et al. 2007; Sempere et al. 2010; Ventura and Jacks 2009); therefore, the ability in situ to visualize miRNA allows researchers to address the complexity and heterogeneity of cancer diseases in a new dimension. In recent years, significant improvements to miRNA and, in general, short-target in situ hybridization (ISH) techniques have been achieved with locked nucleic acid (LNA) modified DNA probes (Kloosterman et al. 2006; Nuovo et al. 2009; Obernosterer et al. 2007; Pena et al. 2009; Silahtaroglu et al. 2007). LNAs are RNA nucleotides containing a methylene bridge between the 2′-oxygen and 4′-carbon, locking the ribose in a C3′-endo conformation (Koshkin et al. 1998; Vester and Wengel 2004). Performing miRNA ISH using DNA probes with LNA substitutions (LNA + DNA) has been shown to stabilize the resulting duplexes compared to unmodified DNA probes and simultaneously improve discrimination between perfect matched and mismatched targets (Kloosterman et al. 2006; McTigue et al. 2004). More elaborate and laborious procedures have been published for the detection of low–copy number miRNAs, including the introduction of an EDC fixation step to crosslink the 5′ end of miRNA to the tissue section (e.g., for detection of miR-130a; Pena et al. 2009), and lengthy hybridizations (7–15 hr), which use either low-stringency washing (4C) with probe concentrations of 50 nM or, alternatively, high-stringency washing (50C) with probe concentrations of 400 to 2000 nM (Nuovo et al. 2009).
In our lab, we have not been able to reproduce the positive effect of the EDC fixation step, which is in line with the observation by Lu and Tsourkas (2009), who showed that there was no significant loss of miRNA in the absence of EDC fixation in LNA–fluorescence in situ hybridization (FISH). To avoid using long hybridization times or high probe concentrations, which results in long assay times, potentially higher nonspecific background signals (Darnell et al. 2010), and, from an economical perspective, increased assay expenses, we sought to circumvent these procedures and achieve equally good sensitivity in a more simplified way.
In this context, we focused our interest on new probe alternatives for miRNA ISH such as 2′-O-methyl RNA (2OMe)–based oligonucleotide probes. 2OMe nucleotides, like LNA, have several advantages compared to DNA as building blocks for probes suitable for ISH assays. 2OMe compared to DNA bases have shown 1) increased stability of the resulting probe:RNA duplex, 2) faster kinetics of hybridization, 3) capability of binding structured targets under conditions where DNA probes will not, and 4) improved specificity (Majlessi et al. 1998). 2OMe probes have previously been used for in situ detection of RNA in living cells (Molenaar et al. 2001) and of transfer RNA (tRNA; Budde et al. 2008) and miRNA (Aronica et al. 2010) in brain tissue sections. In addition to unmodified 2OMe probes, LNA substitutions have been shown to increase the stability of probe:RNA duplex formation, with most stability being conferred when LNA oligonucleotides are separated by at least one 2OMe nucleotide (Kierzek et al. 2005). Considering that probes of DNA + LNA (2:1 ratio) are generally used for miRNA ISH, it is evident that substituting DNA for 2OMe bases provides a probe design possibility that potentially can improve the resulting probe:RNA duplex formation and thereby yield assays of higher sensitivity and specificity.
However, a potential undesired property of 2OMe-based compared to DNA-based probes are their higher affinity for nonspecific RNA species (Majlessi et al. 1998), which are routinely included in ISH buffers as a blocking agent. Lacking a common standard for ISH blocking, we used hybridization buffers with yeast RNA (yRNA) at a concentration of 500 µg/ml as previously published (Wienholds et al. 2005; Kloosterman et al. 2006). We hypothesized that to obtain the expected full advantages of 2OMe + LNA probes in miRNA ISH, it was necessary to exclude yRNA in the hybridization step, to make a higher quantity of probe available for hybridization to its complementary miRNA target.
Here we report on ISH using probes consisting of 2OMe and LNA (2OMe + LNA) and compare the performance, in terms of signal intensity, signal-to-noise ratio (SNR), and probe specificity, to commercially available DNA + LNA probes. We investigated the effect of excluding the commonly used blocking agent, yRNA, in the hybridization step of ISH assays and demonstrate that improved ISH performance can be achieved through the combination of using probes consisting of 2OMe + LNA and excluding yRNA in the hybridization step compared to using DNA + LNA probes with yRNA included in the hybridization step. We also show that exclusion of yRNA in the hybridization step does not compromise the specificity of the assay. Finally, we also show that replacing the commonly used toxic denaturant formamide in the hybridization buffer can be accomplished by non-toxic urea, again without compromising specificity of the assays.
Materials and Methods
Probe Design and Hybridization Temperature Conditions
Probes consisting of 2OMe and LNAs at every third position (2:1 ratio), complementary to the full length of the target miRNAs (miR-1, -9*, -124, -130a, -138, -195, -205, and let-7b), were designed and synthesized (RiboTask, Odense, Denmark). This simple design scheme was believed to provide the best unbiased comparison to commercially available DNA + LNA probes (Exiqon, Vedbæk, Denmark) of which the exact probe compositions were not disclosed. Negative controls included 2OMe + LNA probes with 2–base pair (bp) mismatches for miR-124, miR-138, and miR-195, as well as probes (2OMe + LNA and DNA + LNA) consisting of a scrambled sequence with no specific targets. All probes were singly labeled in the 5′ end with either digoxyginin (DIG) or fluorescein (FITC).
Hybridization temperature conditions used for the DNA + LNA probes were 18–24C below melting temperature in accordance with the manufacturer’s instructions and previously published protocols (Bak et al. 2008; Kloosterman et al. 2006; Sempere et al. 2007; Silahtaroglu et al. 2007). The manufacturer did not provide information about the melting temperature of the 2OMe + LNA probes. To predict optimal hybridization temperatures for the probes, we consulted the Kierzek website (http://rnachemlab.ibch.poznan.pl/calculator2.php), from which it was possible to calculate melting temperatures for 2OMe + LNA probes at a concentration of 10−4 M. However, as probe concentration affects melting temperature and as probe concentrations in the range of 5–66 nM are commonly used for miRNA ISH (Bak et al. 2008; Jørgensen et al. 2010; Kloosterman et al. 2006; Nuovo 2008; Obernosterer et al. 2007; Pena et al. 2009; Sempere et al. 2007; Silahtaroglu et al. 2007; Wienholds et al. 2005), of which 40 nM was used in this study, this software did not provide accurate data. Such data could potentially be used in further probe optimization. Therefore, the hybridization temperatures were experimentally determined for the 2OMe + LNA probes to yield specific signals. Using 2OMe + LNA probes for which a corresponding mismatch probe was available, hybridization temperatures were selected such that a significantly reduced signal was observed for the mismatch probe compared to the perfect match. Table 1 shows details for probe compositions and the individual hybridization conditions that were used.
| Targeta | Probe Name | Composition | Probe Sequence (5′ to 3′)b | Product Numberc | 5′ Label | Tm (C)d | Hybridization Temperature (C) | %GC | ΔG Hairpin |
|---|---|---|---|---|---|---|---|---|---|
| hsa-miR-1 | miR-1 | DNA + LNA | ATACATACTTCTTTACATTCCA | 18008-01 | DIG | 64 | 45 | 27 | 2.5 |
| 2OMe + LNA | AUACAUACUTCUTUACAUTCCA | — | DIG | — | 45 | ||||
| hsa-miR-9* | miR-9* | DNA + LNA | ACTTTCGGTTATCTAGCTTTAT | 38890-01 | DIG | 65 | 45 | 32 | −0.5 |
| 2OMe + LNA | ACUTUCGGUTAUCUAGCUTUAT | — | DIG | — | 45 | ||||
| mmu- miR-124 | miR-124 | DNA + LNA | GGCATTCACCGCGTGCCTTA | 39452-01 | DIG | 80 | 60 | 60 | −3.0 |
| 2OMe + LNA | GGCAUUCACCGCGUGCCUTA | — | DIG | — | 60 | ||||
| miR-124-2mm | 2OMe + LNA | GGCAUUCAAAGCGUGCCUTA | — | DIG | — | 60 | 50 | −3.0 | |
| mmu-miR-130a | miR-130a | DNA + LNA | ATGCCCTTTTAACATTGCACTG | 39031-01 | DIGe | 74 | 50 | 41 | 0.0 |
| 2OMe + LNA | AUGCCCTUUTAACAUTGCACUG | — | FITCe | — | 50 | ||||
| mmu- miR-138 | miR-138 | DNA + LNA | CGGCCTGATTCACAACACCAGCT | 39418-04 | FITC | 84 | 60 | 57 | 1.9 |
| 2OMe + LNA | CGGCCUGAUTCACAACACCAGCT | — | FITC | — | 65 | ||||
| miR-138-2mm | 2OMe + LNA | CGGCCUGAUTUATAACACCAGCT | — | FITC | — | 65 | 48 | 1.9 | |
| mmu- miR-195 | miR-195 | DNA + LNA | GCCAATATTTCTGTGCTGCTA | 39083-01 | DIG | 72 | 50 | 43 | 0.6 |
| 2OMe + LNA | GCCAAUAUUTCUGUGCUGCUA | — | DIG | — | 50 | ||||
| miR-195-2mm | 2OMe + LNA | GCCAAUAUUAAUGUGCUGCUA | — | DIG | — | 50 | 38 | 0.8 | |
| hsa-miR-205 | miR-205 | DNA + LNA | CAGACTCCGGTGGAATGAAGGA | 18099-04 | FITC | 81 | 60 | 55 | −0.4 |
| 2OMe + LNA | CAGACUCCGGUGGAATGAAGGA | — | FITC | — | 65 | ||||
| hsa-let7b | let-7b | DNA + LNA | AACCACACAACCTACTACCTCA | 38001-01 | DIG | 77 | 55 | 45 | NAf |
| 2OMe + LNA | AACCACACAACCTACTACCUCA | — | DIG | — | 55 | ||||
| — | Scrambled | DNA + LNA | GTGTAACACGTCTATACGCCCA | 99004-01 | DIG | 78 | 55 | 50 | −0.9 |
| 2OMe + LNA | GTGUAACACGTCUAUACGCCCA | — | DIG | — | 55 |
a
Target sequences can be found at the miRBase homepage (www.mirbase.org).
b
DNA + locked nucleic acid (LNA) probes were purchased at Exiqon (Vedbæk, Denmark); the specific composition of DNA + LNA oligonucleotides was not disclosed by the manufacturer. 2OMe–RNA + LNA probes were designed and synthesized at RiboTask (Odense, Denmark); bold refers to LNA oligonucleotides, and italic refers to 2OMe–RNA oligonucleotides.
c
Product numbers of DNA + LNA probes.
d
Melting temperature for DNA + LNA probes was provided by the manufacturer (Exiqon) upon purchase. To our knowledge, servers do not exist for melting temperature calculations of 2OMe + LNA probes at conditions for microRNA ISH (i.e., salt and probe concentrations). Hence, hybridization temperatures were experimentally determined to yield a specific ISH signal.
e
DNA + LNA probes complementary to miR-130a were labeled with DIG, and 2OMe–RNA + LNA probes were labeled with FITC. In our experience, the DIG label provides a stronger binding to its antibody–enzyme conjugate compared to the analogous FITC label antibody–enzyme conjugate binding, resulting in a higher in situ hybridization (ISH) signal intensity. In this regard, we expect the miR-130a data presented in Figure 3, Supplementary Figure S1, and Supplementary Figure S2A,B to give even higher ISH signal intensity and signal-to-noise ratio (SNR) had the miR-130a 2OMe–RNA + LNA probe been DIG labeled.
f
No folding could be calculated.
In addition, Table 1 also includes the calculated lowest ΔG values of possible hairpin structures of each probe (i.e., corresponding to self-annealing ability). ΔG values were predicted using mFold RNA folding version 2.3 (http://mfold.rna.albany.edu/?q=mfold) with unmodified RNA versions of the individual probe sequence at their indicated hybridization temperatures.
In Situ Hybridization (FFPE)
Formalin-fixed paraffin-embedded (4% paraformaldehyde [PFA]; Sigma Aldrich, St. Louis, MO) sections of mouse brain and normal breast tissue sections were prepared as follows. Sections were serially cut to 4 µm thicknesses, mounted on SuperFrost Plus (Menzel-Gläser, Braunschweig, Germany) microscope glass slides, baked at 60C for 30 min, and stored at 4C until use. Upon use, sections were pretreated by dewaxing (2 × 3 min in xylene), rehydrating (2 × 3 min in 100% ethanol followed by 3 min in 70% ethanol), and postfixation (10 min in 4% PFA). Sections were acetylated for 10 min (mix of 98 ml milliQ water, 1.35 ml triethanolamine [Sigma Aldrich], 175 µl 12M HCl, and 250 µl acetic anhydride [Sigma Aldrich]). Demasking was necessary to provide a detectable miRNA ISH signal and optimized for different tissue types. A stock solution was prepared from 200 µg proteinase K (PK), 200 µl 1 M Tris-HCl (pH 7.4), 400 µl 1 M CaCl2, 10 ml glycerol, and 9.4 ml milliQ, aliquoted and kept at −20C until use. The stock demasking solution was diluted in PBS to a final PK concentration of which 25 µg/ml proved optimal for brain sections, whereas breast sections performed best with 50 µg/ml PK (Sigma Aldrich). Sections were incubated for 40 min at 30C. Certain sections required quenching of endogenous peroxidase, which was performed for 10 min (0.3% H2O2 in methanol). Sections were washed in PBS three times between each of these steps. A delimiting pen (Dako, Glostrup, Denmark) was used to encircle the sections.
In Situ Hybridization (Cryo)
Mouse heart cryo sections were generously donated, stored at −20C until use, and processed as follows (M. Schneider, personal communication 2010; Silahtaroglu et al. 2007). Slides were thawed for 10 min at room temperature. Sections were encircled with a delimiting pen and dried for 10 min at 50C. Sections were fixed in 4% PFA for 10 min at room temperature, demasked in PK (4 µg/ml), postfixed in 4% PFA for 10 min, and washed three times in PBS between each step. Sections were then acetylated as previously described and blocked for endogenous peroxidase activity (10 min in 0.3% H2O2 in methanol).
Hybridization Step
Prehybridization buffer (4 M urea [Sigma Aldrich] or 50% formamide [Merck, Whitehouse Station, NJ], 5× SSC, 1× Denhardt’s solution [Sigma Aldrich], 500 µg/ml yRNA [Ambion, Austin, TX]) and hybridization buffer (2 M, 4 M, 6 M urea or 50% formamide, 5× SSC, 1× Denhardt’s solution, with 0, 50, 500, or 2500 µg/ml yRNA) were prepared, aliquoted, and kept at −20C until use. For optimal ISH performance, we used hybridization buffers of 5× SSC, 1× Denhardt’s solution, and 4 M urea, without yRNA. Tissue sections were incubated in prehybridization buffer in a lab oven for 30 min at the hybridization temperatures shown in Table 1. Hybridization mixtures were prepared by adding probe to hybridization buffers to a final concentration of 40 nM and heated to 80C prior to being applied to the tissue section. Hybridizations were performed at the hybridization temperature for 60 min and subsequently stringency washed in 0.1× SSC for 30 min at the hybridization temperature. Sections were washed in PBS three times, incubated in blocking buffer (0.1 M Tris [pH 7.5] + 0.15 M NaCl + 10% FBS [Biological Industries, Kibbutz Beit-Haemek, Israel], and 0.05% Tween-20) for 30 min, and washed in PBS three times.
Single Fluorescent ISH Detection
Fluorescent detection of a single miRNA was performed by incubating sections with anti-fluorescein-POD (Roche Applied Science, Mannheim, Germany) or anti-digoxygenin-POD (Roche Applied Science) diluted to 0.375 U/ml in blocking buffer for 60 min and washing in PBS three times, after which tyramide signal amplification was performed using the TSA Plus Cyanine 3 kit (PerkinElmer, Waltham, MA) for 10 min according to the manufacturer’s instructions. Sections were washed in PBS three times, rinsed briefly in milliQ water, and dried and mounted with ProLong Gold containing DAPI (Invitrogen, Carlsbad, CA).
Double fluorescent ISH detection
For double fluorescent ISH assays, two probes with different labels were hybridized, simultaneously or sequentially, and then detected sequentially. Simultaneous hybridizations were performed by including the two probes (miR-124-DIG and miR-138-FITC) in the hybridization buffer, followed by hybridization and stringency washing at 60. Fluorescent detection of miR-124-DIG was performed first using Cy3-TSA-based deposition. The antibody-conjugated peroxidase from the detection was inactivated by 0.3% H2O2 in methanol for 10 min. Detection of miR-138-FITC was then performed using FITC-TSA-based deposition. Sequential hybridizations and detections were performed by hybridizing first the miR-124-DIG probe, stringency washing, and detection using Cy3-TSA-based deposition. The antibody-conjugated peroxidase was inactivated, sections were then hybridized with the miR-130a-FITC probe, and detection was performed using FITC-TSA-based deposition. Sections were finally washed, dried, and mounted as previously described for single fluorescent detection. Sequential hybridizations allowed for hybridization at the optimal temperature for each probe, whereas simultaneous hybridizations, in cases where the optimal hybridization temperatures of the probes were not identical, required that probes to be hybridized at a suboptimal temperature.
Chromogenic ISH Detection
Chromogenic detection was performed by incubating sections with anti-fluorescein-alkaline-phosphatase (AP) (Roche Applied Science) or anti-DIG-AP (Roche Applied Science) diluted to 0.375 U/ml in blocking buffer and washing in PBS three times. NBT/BCIP reagent was applied according to the manufacturer’s instruction for a minimum of 6 hr and a maximum of 33 hr. Sections were washed in PBS. Co-staining of nucleic acids was optionally performed by immersing sections in hematoxylin (VWR—Bie & Berntsen, Herlev, Denmark) for 5 min and rinsing for 10 min in demineralized water. Sections were dried and mounted with Aqua-Mount (Lerner Laboratories, Pittsburgh, PA) reagent.
Quantitative Image Analysis
Sections were imaged with an Axio Imager Z1 epi-fluorescent microscope (Zeiss, Oberkochen, Germany) using an EC Plan-NEOFLUAR 10×/0.3 or a Plan-APOCHROMAT 20×/0.8 objective and an AxioCam MRm CCD camera (Zeiss, Oberkochen, Germany). Segmentation of the images was performed using the VIS software (Visiopharm, Hørsholm, Denmark). Bayesian classification using the quadratic method was applied in the analysis on images acquired at their optimal exposure times (full dynamic range). This provided labeling and definition of positive cells (cells) and their surroundings (tissue). This labeling scheme was subsequently used to quantify intensity on images of the same area but taken at identical exposure times for a given set of probes. This provided comparable intensities for ISHs for a specific miRNA detected using different experimental conditions (variation in probe or hybridization buffer composition). Quantification data were collected for multiple areas for each section and, for some probes, multiple hybridizations. Mean specific signal intensities were determined as cells minus tissue, whereas SNRs were calculated as intensity of cells divided by the intensity of tissue. Error bars are standard error of the mean.
Results
Probes consisting of 2OMe and LNAs (2:1 ratio) were designed and synthesized for detection of eight different miRNAs. ISH performance of the 2OMe + LNA probes was compared to commercially available DNA + LNA probes on formalin-fixed paraffin-embedded (FFPE) mouse brain sections, normal human breast FFPE sections, and cryo-fixed (cryo) mouse heart sections. Three probes with 2-bp mismatches as well as a scrambled probe with no specific targets were included in the study as negative controls. Table 1 shows details of probe compositions and the individual hybridization conditions that were used.
Qualitative ISH Performance of 2OMe + LNA Probes
Tissue sections were blocked with yRNA during the prehybridization step and subsequently hybridized with either 2OMe + LNA or DNA + LNA probes in hybridization buffer with (500 µg/ml) or without yRNA. “yRNA included or excluded from the hybridization buffer” will in the following be referred to simply as “with or without yRNA.” Figure 1 shows chromogenic in situ detection of miR-138 in Purkinje cells of mouse brain cerebellum. A number of observations were made in this initial experiment: 1) both 2OMe + LNA and DNA + LNA probes showed staining in Purkinje cells, indicating that the probes have affinity for the same miRNA target; 2) under standard hybridization buffer conditions with yRNA, 2OMe + LNA and DNA + LNA probes showed comparable staining intensity; 3) stronger staining was observed for 2OMe + LNA probes when excluding yRNA, indicating an inhibitory effect of yRNA on the formation of a duplex between the probe and its target; 4) the inhibitory effect of yRNA was not evident when using DNA + LNA probes; 5) 2-bp discrimination was possible with the 2OMe + LNA probes. The inhibitory effect from the yRNA was also observed with probes specific for miR-130a (see Supplementary Fig. S1) as staining was only visible using 2OMe + LNA–based probes without yRNA and not when using 2OMe + LNA–based probes with yRNA or using DNA + LNA–based probes with or without yRNA. It was attempted to prolong the hybridization time from 1 to 3 or 22 hr, but this did not provide a detectable hybridization signal using the DNA + LNA probes targeting miR-130a (Supplementary Fig. S1).
A larger probe set was included in the study to investigate if the inhibitory effect of yRNA on ISH performance using 2OMe + LNA probes was a general phenomenon or if it was associated with specific probes. ISH targeting the eight different miRNAs was compared at four different conditions (i.e., DNA + LNA and 2OMe + LNA probes tested with or without yRNA). Figure 2 shows overlay images of fluorescent ISH signals in red and nuclei in blue at representative areas of the tissue. ISH images for miR-9*, miR-130a, and miR-138 show a strong signal for 2OMe + LNA probes when excluding yRNA, whereas including yRNA resulted in very low or absent signals. For other probes (miR-124, miR-195, let-7b, and miR-205), there was no observable effect on the corresponding signals from hybridizations with and without yRNA.
The lowest ΔG values of potential hairpin formations for individual probe sequences are shown in Table 1. We observed no correlation between ΔG values of hairpins and susceptibility of the probes to be inhibited by yRNA. These data and the ISH signals indicate that the inhibitory effect previously described is likely to be associated with specific probe sequences.
Quantitative ISH Analysis
The miRNA ISH assays presented in Figure 2 were quantitatively analyzed to provide a clearer understanding of how ISH performance in terms of signal intensity (Supplementary Fig. S2A) and SNR (Supplementary Fig. S2B) varied with the tested conditions.
The effect of using 2OMe instead of DNA in the probes was evaluated for in situ hybridizations performed with yRNA (Table 2). The results showed that three ISHs (targeting miR-138, miR-195, and miR-130a) benefited from the substitution of DNA to 2OMe, four performed equally well (targeting miR-9*, miR-124, miR-205, and let-7b), and one performed worse (targeting miR-1).
| Type 1a | Type 1b | Type 2 | Type 3 | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Calculation | Data | miR-9* | miR-130a | miR-138 | miR-195 | let-7b | miR-124 | miR-205 | miR-1 | |
| Probe comparison | (2OMe + LNA [+])/(DNA + LNA [+]) | Intensity | = | ↑ | ↑ | ↑ | = | = | = | ↓ |
| SNR | = | ↑ | = | ↑ | ↑ | = | = | ↓ | ||
| Blocking agent effect | (DNA + LNA [–])/(DNA + LNA [+]) | Intensity | = | = | ↓ | = | ↓ | = | ↓ | = |
| SNR | = | = | = | = | = | = | ↓ | = | ||
| (2OMe + LNA [–])/(2OMe + LNA [+]) | Intensity | ↑ | ↑ | ↑ | = | = | = | = | = | |
| SNR | ↑ | ↑ | ↑ | = | = | = | = | = | ||
| New method vs gold-standard | (2OMe + LNA [–])/(DNA + LNA [+]) | Intensity | ↑ | ↑ | ↑ | ↑ | ↑ | = | = | ↓ |
| SNR | = | ↑ | ↑ | ↑ | ↑ | = | = | = | ||
Three criteria were evaluated: the effect of changing probe from DNA + locked nucleic acid (LNA) to 2OMe + LNA with yeast RNA (yRNA) in the hybridization buffer, the effect of 50 µg/ml yRNA in the hybridization buffer [+] or without yRNA in the hybridization buffer [–], and the direct comparison between the proposed new method, using 2OMe + LNA probes without yRNA in the hybridization buffer, and the hitherto gold-standard method, using DNA + LNA probe with yRNA in the hybridization buffer. The performance was evaluated by calculating the fold changes for a specific effect and indicating this with ↑ if the hybridization assay was improved by 50% or more, with ↓ if it was impaired by 50% or more, or with = if the measured effect was in between these borderlines (i.e., ±50%). A conservative cutoff of 50% was used to disregard variation in intensity caused by using serial sections, performing manual handling of sections, or from imaging, segmentation, and quantification analysis.
Next, the effect of including or excluding yRNA was evaluated independently for both DNA + LNA and 2OMe–LNA probes. ISH using DNA + LNA probes showed a decrease in signal intensity for three probes (miR-138, let-7b, and miR-205) when yRNA was excluded, of which only the miR-205 probe also showed a decrease in SNR. No effect of yRNA was observed for the five remaining DNA + LNA probes. In the case of the 2OMe + LNA probes, the ISH results have been described previously.
Finally, since DNA + LNA probes with yRNA in the hybridization buffer are considered a gold standard for ISH of miRNA, it was interesting to directly compare the signal intensities and SNR between the gold standard and the new method presented here, which is based on using 2OMe + LNA probes and excluding yRNA from the hybridization step. From this comparison (Table 2), we can conclude that in all cases except one (miR-1), the signal intensity and/or SNR was better using the new protocol based on 2OMe + LNA probes without yRNA in the hybridization step compared to the gold standard.
In Situ Localization of miRNAs in Different Tissues
By evaluating the localization of miRNA in situ signals, we showed that DNA + LNA and 2OMe + LNA probes when detectable had a similar localization, indicating highly specific affinity for the same miRNA targets (see Fig. 2). MiR-1 was detected in myocytes in mouse heart cryo sections (see Fig. 2) as previously published (Kloosterman et al. 2006). ISH signals for miR-9* and miR-130a (only detectable using 2OMe + LNA probes) showed expression in some (but not all) cells in the cortex and primarily in the Purkinje cells in the cerebellum (data not shown) (Nuovo et al. 2009; Pena et al. 2009). MiR-138 and miR-195 showed expression in the Purkinje cells of the cerebellum and very little in the neuronal cells. MiR-195 has previously been shown to have neuronal expression (Bak et al. 2008), which is in contrast to our observation performed with both DNA + LNA and 2OMe + LNA probes. MiR-124 showed strong expression in the neuronal cells and less in the Purkinje cells as previously published (Silahtaroglu et al. 2007).
These observations were further validated by double in situ hybridizations performed with two different probes on the same tissue section. Figure 3 shows differential expression of miR-124 and miR-130a detected in the mouse cortex (top row) and of miR-124 and miR-138 in the mouse cerebellum (bottom row). In the normal human breast tissue samples, miR-205 showed expression in the myoepithelial/basal cell layer (see Fig. 2), whereas let-7b localized predominantly to the luminal cell layer (see Fig. 2) (Avril-Sassen et al. 2009; Sempere et al. 2007; Sempere et al. 2010).
Competitive Binding of 2OMe-Containing Probes to yRNA
A quantitative comparison of the inhibitory effect of yRNA observed for ISH using 2OMe + LNA probes for miR-138 detection was performed. Hybridizations were done for 1 or 3 hr with varying concentrations of yRNA in the hybridization buffer. Increasing concentrations of yRNA resulted in a clear decrease in signal intensity for both lengths of hybridization time (see Fig. 4A). Figure 4B shows the gain in signal intensity yielded by hybridization for 3 hr compared to 1 hr. The increase in signal intensity due to longer hybridizations increased along with the concentration of yRNA, indicating that hybridization kinetics between the 2OMe + LNA probe and miRNA is slower in the presence of yRNA. In summary, lengthy hybridizations are likely to provide higher intensity (~1.8-fold for 3 hr vs 1 hr), although this positive gain is minor in comparison to exclusion of yRNA from the hybridization buffer (~8.6-fold [47.1 vs 5.5 arbitrary units] for hybridizations of 1 hr).
Specificity in Hybridization
Specificity in the hybridization was obtained with 4 M urea instead of toxic 50% formamide. Figure 5 shows images of ISH targeting miR-124 using DNA + LNA probes at 4 M urea and 50% formamide (FA). Quantification of signal intensity showed a 25% intensity increase using 4 M urea, whereas the SNR ratio was the same. Simard et al. (2001) have previously shown that hybridization buffers with urea concentrations of 2–4 M urea performed as good as 50% formamide in reducing background hybridization in Northern blotting experiments, and Poulsen et al. (2008) demonstrated in DNA microarray experiments that 4 M urea increased stringency of SSC buffers, resulting in increased specificity. The unchanged specificity of replacing formamide with urea was confirmed in the ISH assays by demonstrating that 2OMe + LNA probes with 2-bp mismatches against miR-124 (miR-124–2 mm), miR-138 (miR-138–2 mm), and miR-195 (miR-195–2 mm) without yRNA resulted in signal intensity decreases of 19-, 14-, and 16-fold, respectively, and decreases in SNR of 2.6-, 2.2-, and 6.2-fold, respectively, compared to the perfect match probe (Supplementary Fig. S3). These results show that exclusion of yRNA from the hybridization step does not compromise specificity of the assay.
Discussion
Nonspecific RNA is routinely included in ISH buffers as a blocking agent in the form of varying concentrations of yRNA (Kloosterman et al. 2006; Thompson et al. 2007; Wienholds et al. 2005), yeast tRNA (ytRNA) (Sempere et al. 2010; Silahtaroglu et al. 2007), yRNA in combination with salmon sperm DNA (Obernosterer et al. 2007), or ytRNA in combination with salmon sperm DNA (Pena et al. 2009; Thomsen et al. 2005). The ISH signals that we have observed for 2OMe + LNA probes targeting miR-9*, miR-130a, and miR-138 indicate that there is a non-negligible affinity-based interaction between specific 2OMe + LNA probes and yRNA that apparently restricts high amounts of probe from hybridizing to the specific target miRNA of interest. A similar mechanism was observed for detection of miR-138 using a 2OMe + LNA probe in a hybridization buffer containing ytRNA (500 µg/mL) (data not shown). The comparably high signals observed for miR-1, miR-124, miR-195, miR-205, and let-7b using 2OMe–LNA probes independent of the presence or absence of yRNA in the hybridization step indicates that it is the probe sequence that most likely influences duplex formation between probe and yRNA. Prediction of this inhibitory binding will require sequencing of the yRNA using quantitative methods such as RNA-seq, which is out of the scope of this article. Instead and as a general recommendation, it appears warranted to exclude yRNA from the hybridization step to avoid nonspecific and unpredictable binding of 2OMe + LNA probes to nonspecific RNA such as yRNA.
The optimal hybridization temperatures of the 2OMe + LNA probes were generally very close to those of the corresponding DNA + LNA sequence (see Table 1). Before a server is available that can calculate the melting temperature of 2OMe + LNA probes at the concentrations commonly used for miRNA ISHs, we propose calculating the melting temperature of the corresponding DNA + LNA probes and using this as reference for predicting temperatures that provide specific hybridization signals. In this report, hybridization temperatures of 2OMe + LNA probes were 16–24C below the melting temperature of the corresponding DN A +LNA probes. For the two probe sets targeting miR-138 and miR-205, hybridization temperatures of the 2OMe + LNA probes were higher than the DNA + LNA probes. These temperatures were selected to decrease unspecific bindings, thereby providing a specific hybridization signal.
In the following, it is speculated that the beneficial effects of using 2OMe + LNA probes and exclusion of yRNA from the hybridization step compared to DNA + LNA probes with yRNA can be explained by a combination of the inhibition effect previously described, a potential difference in the affinity between probe and target, and the relative abundance of the miRNA in the tissue.
The affinity between probe and miRNA target is critical in the in situ detections. Thermodynamic studies indicate that the binding affinity of miRNA to 2OMe + LNA probes is greater than between miRNA and DNA + LNA probes (Kierzek et al. 2005). Ideally, software should be able to calculate melting temperatures for DNA + LNA as well as 2OMe + LNA probes in molar concentrations commonly used in ISHs, thereby providing melting temperature predictions. As such calculations, are presently not available, the GC content of a probe can be used as a simplistic predictive value in comparing the relative affinity of different probes. The only probe performing worse using the new method was the miR-1 targeting probe, which also has the lowest GC content of all the tested probes. The binding affinity of the 2OMe + LNA probe targeting miR-1 could likely be increased by including a higher percentage of LNA bases. It can be speculated that this is already the case for the commercially purchased DNA + LNA probe of undisclosed base composition.
The relative abundance of a specific miRNA is considered a key parameter in predicting the ability of a given ISH assay to be sensitive enough. When trying to detect highly abundant miRNAs, the sensitivity issue is not critical as it is likely that a high number of probes will bind, even under non-optimal hybridization conditions, to reach a certain saturation level where there is no more space for further probes to bind. This will provide the basis for obtaining equally good signals from probes with different affinity. Hence, the use of probes with an increased affinity for their target, such as the 2OMe + LNA compared to the DNA + LNA probes, is unlikely to provide better ISH performance under these circumstances. Our data support this hypothesis since 2OMe + LNA probes without yRNA did not positively affect detection of highly expressed miRNAs compared to DNA + LNA probes with yRNA (Table 2). For instance, miRNAs categorized as type 2 and type 3 included probes targeting some of the most abundant miRNAs for their specific tissue. MiR-1 is the most abundant miRNA in the heart (Lagos-Quintana et al. 2002), miR-124 in the brain (Bak et al. 2008; Lagos-Quintana et al. 2002), and miR-205 in myoepithelial/basal cells of the breast (Sempere et al. 2007). In comparison, type 1 includes probes targeting low expressed miRNAs such as miR-130a (Pena et al. 2009; Nuovo et al. 2009) and miR-195 (Bak et al. 2008; Pena et al. 2009), which benefited from analysis with 2OMe + LNA probes without yRNA. In these cases, the benefit of the increased hybridization affinity of 2OMe + LNA probes compared to DNA + LNA probes in the absence of yRNA is likely to provide the basis for a higher number of duplex formations and hence more sites of substrate conversion, yielding a higher signal and more sensitive assay.
We believe that even further improvements of the miRNA ISH protocol presented here are likely to be obtainable. The 2OMe + LNA probe design used in this study was based on a simplistic design scheme of replacing every third 2OMe nucleotide with LNA, disregarding potential hairpin formations and melting temperature predictions. LNA nucleotide substitutions in a probe have been shown to have an additive effect on the melting temperature (Kierzek et al. 2005), and hence altering the 2OMe:LNA ratio of a probe can provide probes of desired melting temperature. If lack of specificity is encountered in a 2OMe + LNA ISH miRNA targeting assay, probes can be shortened to 19 nucleotides, thereby increasing specificity without reducing sensitivity (Majlessi et al. 1998). Similarly, the incorporation of haptens in both ends of a probe (e.g., double-DIG labeling) has been shown by us (unpublished results) and Sempere et al. (2010) for miRNA ISH detection and by Darnell et al. (2010) for mRNA ISH detection to provide significantly higher ISH signals compared to singly labeled probes.
In conclusion, we have obtained significant increases in the sensitivity of miRNA detection by ISH with a concomitant increase in SNR that has enabled detection of low–copy number miRNAs (e.g., miR-130a in mouse brain) without the need for extra fixation or cross-linking steps, using reduced hybridization time (1 hr) and low probe concentrations of 40 nM. The improved ISH protocol, based on a simple, non-optimized probe design using 2OMe + LNA probes and a hybridization buffer of 4 M urea without yRNA, enables sensitive, fast, and non-toxic assays.
Acknowledgments
The authors thank the Ministry of Science, Technology and Innovation—Danish Agency for Science, Technology and Innovation for funding to MJS; Lise H. Christensen for providing access to normal breast tissue sections; and Mikael Schneider for providing mouse heart sections and the cryo ISH protocol.
Competing Interests
The author(s) declared no potential conflicts of interest with respect to the authorship and/or publication of this article.
Funding Information
The author(s) disclosed receipt of the following financial support for the research and/or authorship of this article: support from the Ministry of Science, Technology and Innovation—Danish Agency for Science, Technology and Innovation to MJS.
References
Aronica E, Fluiter K, Iyer A, Zurolo E, Vreijling J, van Vliet EA, Baayen JC, Gorter JA. 2010. Expression pattern of miR-146a, an inflammation-associated microRNA, in experimental and human temporal lobe epilepsy. Eur J Neurosci. 31:1100–1107.
Avril-Sassen S, Goldstein LD, Stingl J, Blenkiron C, Le Quesne J, Spiteri I, Karagavriilidou K, Watson CJ, Tavare S, Miska EA, et al. 2009. Characterisation of microRNA expression in post-natal mouse mammary gland development. BMC Genomics. 10:548.
Bak M, Silahtaroglu A, Moller M, Christensen M, Rath MF, Skryabin B, Tommerup N, Kauppinen S. 2008. MicroRNA expression in the adult mouse central nervous system. RNA. 14:432–444.
Budde BS, Namavar Y, Barth PG, Poll-The BT, Nurnberg G, Becker C, van Ruissen F, Weterman MAJ, Fluiter K, Beek ETT, et al. 2008. tRNA splicing endonuclease mutations cause pontocerebellar hypoplasia. Nat Genet. 40:1113–1118.
Darnell DK, Stanislaw S, Kaur S, Antin PB. 2010. Whole mount in situ hybridization detection of mRNAs using short LNA containing DNA oligonucleotide probes. RNA. 16:632–637.
Jørgensen S, Baker A, Møller S, Nielsen BS. 2010. Robust one-day in situ hybridization protocol for detection of microRNAs in paraffin samples using LNA probes. Methods. 52:375–381.
Kierzek E, Ciesielska A, Pasternak K, Mathews DH, Turner DH, Kierzek R. 2005. The influence of locked nucleic acid residues on the thermodynamic properties of 2′-O-methyl RNA/RNA heteroduplexes. Nucleic Acids Res. 33:5082–5093.
Kloosterman WP, Wienholds E, de Bruijn E, Kauppinen S, Plasterk RHA. 2006. In situ detection of miRNAs in animal embryos using LNA-modified oligonucteotide probes. Nat Methods. 3:27–29.
Koshkin AA, Singh SK, Nielsen P, Rajwanshi VK, Kumar R, Meldgaard M, Olsen CE, Wengel J. 1998. LNA (locked nucleic acids): synthesis of the adenine, cytosine, guanine, 5-methylcytosine, thymine and uracil bicyclonucleoside monomers, oligomerisation, and unprecedented nucleic acid recognition. Tetrahedron. 54:3607–3630.
Lagos-Quintana M, Rauhut R, Yalcin A, Meyer J, Lendeckel W, Tuschl T. 2002. Identification of tissue-specific microRNAs from mouse. Curr Biol. 12:735–739.
Lu J, Tsourkas A. 2009. Imaging individual microRNAs in single mammalian cells in situ. Nucleic Acids Res. 37:10.
Majlessi M, Nelson NC, Becker MM. 1998. Advantages of 2′-O-methyl oligoribonucleotide probes for detecting RNA targets. Nucleic Acids Res. 26:2224–2229.
McTigue PM, Peterson RJ, Kahn JD. 2004. Sequence-dependent thermodynamic parameters for locked nucleic acid (LNA)–DNA duplex formation. Biochemistry. 43:5388–5405.
Molenaar C, Marras SA, Slats JCM, Truffert JC, Lemaitre M, Raap AK, Dirks RW, Tanke HJ. 2001. Linear 2′ O-methyl RNA probes for the visualization of RNA in living cells. Nucleic Acids Res. 29:E89–E90.
Nuovo GJ. 2008. In situ detection of precursor and mature microRNAs in paraffin embedded, formalin fixed tissues and cell preparations. Methods. 44:39–46.
Nuovo GJ, Elton TS, Nana-Sinkam P, Volinia S, Croce CM, Schmittgen TD. 2009. A methodology for the combined in situ analyses of the precursor and mature forms of microRNAs and correlation with their putative targets. Nat Protocols. 4:107–115.
Obernosterer G, Martinez J, Alenius M. 2007. Locked nucleic acid-based in situ detection of microRNAs in mouse tissue sections. Nat. Protocols 2:1508–1514.
Pena JTG, Sohn-Lee C, Rouhanifard SH, Ludwig J, Hafner M, Mihailovic A, Lim C, Holoch D, Berninger P, Zavolan M, Tuschl T. 2009. miRNA in situ hybridization in formaldehyde and EDC-fixed tissues. Nat Methods. 6:139–141.
Poulsen L, Søe MJ, Snakenborg D, Moller LB, Dufva M. 2008. Multi-stringency wash of partially hybridized 60-mer probes reveals that the stringency along the probe decreases with distance from the microarray surface. Nucleic Acids Res. 36:E132.
Sempere LF, Christensen M, Silahtaroglu A, Bak M, Heath CV, Schwartz G, Wells W, Kauppinen S, Cole CN. 2007. Altered microRNA expression confined to specific epithelial cell Subpopulations in breast cancer. Cancer Res. 67:11612–11620.
Sempere LF, Preis M, Yezefski T, Ouyang H, Suriawinata AA, Silahtaroglu A, Conejo-Garcia JR, Kauppinen S, Wells W, Korc M. 2010. Fluorescence-based codetection with protein markers reveals distinct cellular compartments for altered microRNA expression in solid tumors. Clin Cancer Res. 16:4246–4255.
Silahtaroglu AN, Nolting D, Dyrskjot L, Berezikov E, Moller M, Tommerup N, Kauppinen S. 2007. Detection of microRNAs in frozen tissue sections by fluorescence in situ hybridization using locked nucleic acid probes and tyramide signal amplification. Nat Protocols. 2:2520–2528.
Simard C, Lemieux R, Côté S. 2001. Urea substitutes toxic formamide as destabilizing agent in nucleic acid hybridizations with RNA probes. Electrophoresis. 22:2679–2683.
Thompson RC, Deo M, Turner DL. 2007. Analysis of microRNA expression by in situ hybridization with RNA oligonucleotide probes. Methods. 43:153–161.
Thomsen R, Nielsen PS, Jensen TH. 2005. Dramatically improved RNA in situ hybridization signals using LNA-modified probes. RNA. 11:1745–1748.
Ventura A, Jacks T. 2009. MicroRNAs and cancer: short RNAs go a long way. Cell. 136:586–591.
Vester B, Wengel J. 2004. LNA (locked nucleic acid): high-affinity targeting of complementary RNA and DNA. Biochemistry. 43:13233–13241.
Wienholds E, Kloosterman WP, Miska E, Alvarez-Saavedra E, Berezikov E, de Bruijn E, Horvitz HR, Kauppinen S, Plasterk RHA. 2005. MicroRNA expression in zebrafish embryonic development. Science. 309:310–311.
