Use of two novel approaches to discriminate between closely related host microRNAs that are manipulated by Toxoplasma gondii during infection

Abstract

MicroRNAs (miRNAs) are a class of small, endogenously encoded regulatory RNAs that function to post-transcriptionally regulate gene expression in a wide variety of eukaryotes. Within organisms, some mature miRNAs, such as paralogous miRNAs, have nearly identical nucleotide sequences, which makes them virtually indistinguishable from one another by conventional hybridization-based approaches. Here we describe two inexpensive, sensitive methods for rapidly discriminating between paralogous miRNAs or other closely related miRNAs and for quantifying their abundance. The first approach is a sequential ribonuclease-protection and primer-extension assay; the second approach is a primer-extension assay that employs short oligonucleotide probes to exacerbate the instability of mismatched probe:miRNA hybrids. Both approaches are rapid and can be easily performed in their entirety using common laboratory equipment. As a proof of concept, we have used these methods to determine the exact identities of the human miR-17 family members that are increased by infection with the intracellular parasite Toxoplasma gondii. These methods can be used to rapidly and inexpensively discriminate between any closely related miRNAs in any organism.

INTRODUCTION

MicroRNAs (miRNAs) bestow specificity to a post-transcriptional gene-regulatory mechanism that is both ancient and highly conserved. Since their discovery (Lee et al. 1993; Wightman et al. 1993), miRNAs have emerged as versatile gene regulators and are now known to orchestrate a multitude of developmental, cellular, and physiological processes. miRNAs now rank as one of the most abundant classes of gene-regulatory molecules in the cell (Hobert 2008), and in silico target prediction suggests that miRNAs are potentially involved in the regulation of up to 30% of the human transcriptome (Grimson et al. 2007). Mature miRNAs are ∼23-nucleotide (nt) RNA molecules that base pair with target mRNAs and generally function to negatively influence the translation and/or stability of their target mRNAs (Du and Zamore 2005). Closely related miRNAs that share identical “seed” sequences (which comprise the first 6–8 nt of their mature 5′ ends) are grouped into families; as these seed sequences are instrumental in target mRNA discrimination (Lewis et al. 2003), individual miRNAs that are grouped into the same family are believed to have related functions (Leaman et al. 2005; Mendell 2008; Xiao and Rajewsky 2009).

The experimental approaches that have been crafted for the identification and quantification of miRNAs fall into two broad classes; those that are hybridization-based approaches, such as primer-extension (Gottwein et al. 2007), Northern blotting (O'Donnell et al. 2005), miRNA microarray profiling (Thomson et al. 2004), in situ hybridization (Pena et al. 2009), RNase protection (Ventura et al. 2008), and reverse-transcription PCR (He et al. 2005), or those based on high-throughput sequencing of prepared miRNA libraries (Lu et al. 2005). High-throughput sequencing has several notable advantages over the hybridization-based approaches, such as the a priori identification and quantification of all miRNAs within a sample with single-nucleotide resolution, but has limitations in that it is expensive, technically difficult, and time consuming to perform and that the high-volume of resultant data is often a challenge to analyze. These features make high-throughput sequencing impractical for the interrogation of a small number of miRNAs.

For the above reasons, hybridization-based approaches are the most commonly used methods for querying known miRNAs. Compared with high-throughput sequencing, these approaches are inexpensive, expedient, and simple to perform. One of the main limitations of conventional hybridization-based approaches is that when they are performed under standard hybridization conditions, they fail to distinguish between highly conserved miRNAs that differ at one or two nucleotide positions. Through careful adjustment of hybridization temperatures with specific probes, Northern blot analysis has been successfully used to distinguish between the members of the Let-7 family (Chang et al. 2008), but this method has limitations. First, it is time consuming, requiring >3 d when the hybridization temperature is already optimized and longer if the hybridization temperature has to be empirically determined; second, Northern blotting requires that the location of the probe:miRNA mismatches be near the center of the probe:miRNA hybrid, which due to the short length of miRNAs, is not always achievable. Similarly, conventional RNase protection has been used to distinguish closely related miRNAs (Ventura et al. 2008), but generating high specific-activity radiolabeled RNA probes requires a significant time investment and large amounts of input RNA (up to 100 μg total RNA) (see Lee et al. 2002) are required.

Pathogens can alter host miRNA levels as we previously demonstrated with the obligate, intracellular parasite Toxoplasma gondii; this protozoan specifically increases the expression of host miR-17 family members in primary human cells (Zeiner et al. 2010). In that study, however, we were unable to distinguish between the individual mature miR-17 family members using conventional methods because of their nearly identical sequences. To address this, we describe here a pair of simple primer-extension-based approaches to successfully discriminate miRNAs that differ by as little as 1 nt.

RESULTS

A sequential RNase-protection/primer-extension assay distinguishes between synthetic miR-17 family templates that differ by a single nucleotide

The miR-17 family members are encoded in three paralogous loci in mammals and birds (Tanzer and Stadler 2004), miR-17-92, miR-106a-363, and miR-106b-25 (Fig. 1). As miR-17-92 and miR-106a-363 both encode the same four families of miRNAs (Fig. 1) and a there is no mature miRNA type that provides a unique signature to differentiate them, conventional hybridization-based methods do not readily distinguish between activation of one cluster over the other. Previously, we concluded that one or both of these loci are affected by Toxoplasma infection but could not determine which. Although the primary transcript (pri-) for miR-17-92 increased in abundance in Toxoplasma-infected cells, we were unable to detect pri-miR-106a-363 (Zeiner et al. 2010), precluding any conclusion on its contribution to the increased levels of mature miR-17 family members; this leaves open the possibilities that pri-miR-106a-363 transcription peaks at time points not examined and/or that pri-miR-106a-363 is so rapidly processed into mature miRNAs that it is not present at levels high enough to be detected.

FIGURE 1.

Genomic organization of the miR-17 family encoding paralogs. MiR-17 family members are shown in white.

To determine if miR-17-92 and/or miR-106a-363 contributes to the rise in miR-17 family members upon Toxoplasma infection, we developed an assay that is sensitive to the single-nucleotide differences that distinguish the individual miR-17 family members unique to each cluster. Specifically, we assayed for the presence of mature miR-17 (which is derived from miR-17-92) and mature miR-106a (which is derived from miR-106a-363) (see Figs. 1, 2B). This was achieved by combining RNase protection and primer-extension analyses (for overview of the assay, see Fig. 2A), an approach that was originally designed to probe RNA secondary structure (Ziehler and Engelke 2001). The assay works by exploitation of RNase A and RNase T1 substrate specificities; under high-salt conditions (≥300 mM NaCl), RNase A and RNase T1 only cleave single-stranded RNAs at 5′-(C/U)↓N, or 5′-G↓N, respectively, leaving perfectly base paired RNA:DNA (or RNA:RNA) hybrids intact. Hence, following hybridization to a miR-specific probe, miRNAs that base pair imperfectly with the probe expose RNase-labile nucleotides, and these miRNA templates will yield no subsequent primer-extension product after RNase digestion. Additionally, as miRNAs bound to the probe are double-stranded hybrids containing unpaired 5′ miRNA overhangs (see Fig. 2B), RNase-labile nucleotides within these unpaired miRNA overhangs will also be cleaved and will produce primer-extension products of predictably different lengths for different miRNA templates.

FIGURE 2.

Mature miR-17 is present in Toxoplasma-infected HFFs; mature miR-106a is absent. (A) The sequential RNase protection/primer extension assay. Paralogous miRNAs are shown as thick black lines, and RNase-labile nucleotides are shown as asterisks. (Top box) The [5′-³²P]-labeled oligonucleotide probe is hybridized to its targets in high-salt buffer, which forms either perfectly base-paired miRNA:probe duplexes (left column) or miRNA:probe duplexes that are mismatched (right column); local areas of very short and thus unstable base-pairing are indicated by dotted lines. (Middle box) The miRNA:probe hybrids are exposed to single-strand-specific RNases (scissors), which cleave RNase-labile nucleotides in single-stranded regions of the miRNA:probe hybrids. After the initial cleavage, the mismatched miRNA:probe hybrid (right column) is further destabilized, which exposes more single-stranded RNase-labile nucleotides in the miRNA, which are subsequently digested. (Bottom box) The RNases and salt are removed, and the remaining miRNA:probe hybrids are subject to primer-extension analysis; extended primers are indicated by white dots. Shown are the expected extension lengths of matched and mismatched hypothetical miRNA:probe duplexes in the presence or absence of RNase. (B) Sequence alignment of miR-17 family members (bracket). The miR-17/106a probe sequence is shown in base-pairing orientation. Bases in bold type are RNase A– or RNase T1–labile nucleotides of each miRNA that do not base pair with the miR-17/106a probe; underlined uridines form dG:U base pairs with the miR-17/106a probe. (C) Primer-extensions were performed on synthetic templates corresponding to specific miR-17 family members. Specific primer-extension products derived from RNase-treated probe:miRNA hybrids are present only for the synthetic miR-17 and miR-106a templates (indicated with asterisks). The RNase-treated probe:miR-17 hybrid produces a primer-extension product that migrates 1 nt shorter than the corresponding product derived from the probe:miR-106a hybrid due to the presence of an RNase A–labile single-stranded C at the 5′ terminus of the probe:miR-17 hybrid. (D) As in C except that the template RNA was from mock-infected HFFs or from HFFs infected with Neospora or Toxoplasma for 24 h, as indicated, in the absence (−) or presence (+) of RNase A and RNase T1. Specific primer-extension products in the presence of RNase and free probe are indicated by arrows. The (−) RNA lane corresponds to primer-extensions performed in the absence of template RNA. The position of the predicted extension product specific for miR-17 (22 nt) in the presence of RNase is indicated. Size markers at 20 and 30 nt are shown.

As a test of this assay, a [5′-³²P]-labeled oligonucleotide probe (miR-17/106a probe) that was perfectly complementary to both miR-17 and miR-106a, but that had mismatches with the other miR-17 family members (see Fig. 2B) was hybridized to in vitro-synthesized miR-17 family members (Fig. 2C), or to total RNA derived from uninfected primary human foreskin fibroblasts (HFFs) or HFFs infected with Neospora or Toxoplasma (Fig. 2D). The resulting hybrids were incubated in the presence of high salt and RNase A and RNase T1 (RNase mix). The RNase mix was inactivated, and a primer-extension was performed on the miRNA:[³²P]-miR-17/106a probe hybrids to assay for the presence of a primer-extension product and to determine the 5′ ends of the remaining miRNA templates.

The data demonstrate that in the absence of RNase mix, all in vitro synthesized templates were extended to a major product with the expected size of 23nt (Fig. 2C, − RNase lanes), confirming that the miR-17/miR-106a probe fails to distinguish between the individual miR-17 family members under conventional primer-extension reaction conditions. An additional, minor primer-extension product that is 1 nt larger than the major product is also observed and is probably due to nontemplated nucleotide addition by reverse transcriptase (Schmidt and Mueller 1999; Efimov et al. 2001), as discussed further below. When the mix of RNase A and T1 was added to the miRNA:probe hybrids prior to primer-extension, however, miRNA templates that had internal mismatches (miR-93, miR-106b, and miR-20b) or non-Watson–Crick base pairs such as dG:U (miR-20a) with the miR-17/106a-specific probe yielded no primer-extension products, presumably due to cleavage by RNase near the middle of the miRNA templates at these bulged, labile residues (Fig. 2C, + RNase lanes). In contrast, miR-17 and miR-106a templates that were hybridized to the miR-17/106a probe each yielded a distinct primer-extension product following RNase mix addition (Fig. 2C, asterisks). MiR-17 and miR-106a differ from one another by a single nucleotide at their 5′ terminus (C or A, respectively), which does not base pair with the miR-17/106a probe (Fig. 2B); this 5′ terminal nucleotide should have different sensitivities to the RNase mix (the 5′-C of miR-17 should be RNase A-labile, whereas the 5′-A of miR-106a should not). As a result, primer-extension after RNase treatment of the duplex would be expected to yield an extension product for miR-17 that is 1 nt shorter than for miR-106a. Consistent with this, in the presence of RNase mix, the size of the major primer-extension product using the synthetic miR-17 template was indeed 22 nt, whereas the size using synthetic miR-106a was 23 nt (Fig. 2C, asterisks).

In addition to the expected products, we periodically observed the previously characterized nontemplated terminal-transferase activity of reverse-transcriptase (Schmidt and Mueller 1999; Efimov et al. 2001) in some reactions performed on the synthetic oligonucleotide templates (Fig. 2C). This nontemplated band was not the result of low template RNA concentration or target homogeneity, as similar experiments were performed in the presence of 10 μg of carrier yeast tRNA and showed the same periodic nontemplated band addition (data not shown). The nontemplated addition was also not dependent on the presence or absence of a 5′ phosphate group on the template miRNAs (data not shown). The presence of this additional primer-extension product did not alter the interpretation of these data, as the shadow “+1” band paralleled the behavior of the major primer-extension product in all respects. With miR-17, it appears that both products were reduced by 1 nt if the duplex was exposed to RNase treatment, whereas with the synthetic miR-106a template, both extension products were reduced in intensity but not size. Overall, then, this assay faithfully discriminates between miR-17 family members that differ by just 1 nt. Although a significant fraction of each template miRNA is degraded in the presence of the RNase mix, perhaps due to “breathing” of the duplex, comparisons can be made between different miRNA samples that were hybridized to the same probe. We believe that this assay has advantages over conventional RNase protection for the assay of miRNAs in that closely related miRNAs can be distinguished very rapidly since no cloning is required for probe preparation and [5′-³²P]-labeling of oligonucleotide probes is extremely rapid and efficient.

MiR-17 family members derived from miR-17-92, but not from miR-106a-363, are increased in Toxoplasma-infected primary human cells

We next performed the sequential RNase protection/primer-extension assay with the miR-17/106a probe on RNAs extracted from uninfected HFFs, Neospora-infected HFFs, or Toxoplasma-infected HFFs (Fig. 2D). For all total RNA samples, the results demonstrated that the only primer-extension product seen after treatment with the RNase mix was 22 nt in length (Fig. 2D, arrow), indicating the presence of mature miR-17 in these RNA samples. The absence of a 23-nt primer-extension product after RNase mix treatment indicates that the levels of miR-106a are either below the level of detection for this assay or altogether absent in these RNA samples. In contrast to the RNase protection/primer-extensions performed on synthetic templates (Fig. 2C), we did not observe the nontemplated primer-extension products in any reaction performed on total RNA. Based on our previous observation that pri-miR-106a-363 is undetectable in uninfected HFFs or in HFFs infected with Toxoplasma (Zeiner et al. 2010) and on the results presented in Figure 2D, we conclude that miR-106a-363 is likely not affected by Toxoplasma infection. In contrast, based on our observation that pri-miR-17-92 is significantly increased in Toxoplasma-infected HFFs and that the mature miR-18 and miR-19 members derived from pri-miR-17-92 are also increased (Zeiner et al. 2010), and upon the results of Figure 2D, it appears that a contributor to the rise in miR-17 family levels is miR-17, itself.

Primer-extension using short oligonucleotide probes discriminates between synthetic miR-17 family members and shows that Toxoplasma increases the levels of mature miR-106b-25-derived miRNAs

We have previously reported that one of the effects of Toxoplasma infection is to increase the level of the pri-miR-106b-25 (Zeiner et al. 2010). To determine whether the levels of mature miR-106b and miR-93, which are both derived from pri-miR-106b-25, also increase, we designed two short (12-nt) gel-purified primer-extension probes that are perfectly complementary to miR-106b or miR-93 but have 3′-bp mismatches with all other miR-17 family members (Fig. 3A). To determine if these short probes were capable of distinguishing between the miR-17 family members, in vitro synthesized miR-17 family members were subject to primer-extension with the miR-106b long, miR-106b short, and miR-93 short probes. Independent primer-extensions were performed while adjusting the hybridization temperature (T_m) from 37°C to 48°C in a mix of 2°C–3°C increments to fine-tune the specificity of these short probes for their cognate templates (data not shown). At 48°C, both the miR-106b short and miR-93 short probes were highly specific for only miR-106b or miR-93 synthetic templates, respectively, while the miR-106b long probe was unable to completely discriminate between miR-17 family members at any T_m between 37°C and 50°C (Fig. 3B).

FIGURE 3.

Toxoplasma infection increases the levels of mature miR-106b and miR-93. (A) Sequence alignment of miR-17 family members (upper bracket) and primer-extension probes (lower bracket). The miR-106b-long, miR-106b-short, and miR-93-short probe sequences are shown in base-pairing orientation. (B) Primer-extensions were performed at T_m=48°C using synthetic templates corresponding to miR-17 family members. The probes used were as follows: miR-106b long (upper panel); miR-106b short (middle panel); miR-93 short (lower panel). Major primer-extension products are indicated by arrows. (C) Primer-extensions were performed at T_m=48°C using total RNA templates harvested from mock-infected HFFs or from HFFs infected with Toxoplasma using either the miR-106b-short, miR-93-short, or 5S rRNA probes. (−) RNA lanes were primer-extensions performed in the absence of template RNA. Free probe was present in >100-fold excess in all primer-extension lanes (data not shown). Primer-extension product sizes were determined using a 10-bp ladder.

Using these optimized conditions, we performed primer-extension with the miR-106b short and miR-93 short probes on total RNA harvested from mock- or Toxoplasma-infected HFFs at 24 h post-infection (Fig. 3C). The results demonstrate that mature miR-106b and mature miR-93 are both increased in Toxoplasma-infected HFFs relative to mock-infected HFFs, and therefore, miR-106b-25-derived mature miRNAs are affected by Toxoplasma infection.

DISCUSSION

Here we describe two sensitive assays that enable rapid single-nucleotide discrimination of paralogous miRNAs and have used these methods to demonstrate that Toxoplasma infection increases the levels of mature miR-17 family members derived from host miR-17-92 and miR-106b-25 clusters, but not from miR-106a-363. We believe that these two complementary methods can easily be adapted to rapidly and sensitively discriminate between any closely related miRNAs using common laboratory reagents.

For the sequential RNase protection/primer-extension assay, the proof-of-concept case of the miR-17 family is a stringent test; among the most difficult paralogous miRNAs to distinguish are miR-17 and miR-106a since they differ from each other by only 1 nt located at their 5′ terminus. The only other assays capable of distinguishing these miRNAs are direct sequencing and, potentially, conventional RNase protection, which are less economical and expedient than the sequential RNase protection/primer-extension assay presented here. It is unclear whether conventional RNase protection would be capable of distinguishing miR-17 and miR-106a, as a full-length RNase protection probe would be base paired immediately 3′ to the mismatched nucleotide; thus RNase A digestion might not be able to efficiently access and subsequently cleave the labile 5′-C of miR-17 (in the sequential RNase protection/primer-extension approach, the 5′-terminal 4 nt of miR-17 and miR-106a are unpaired with the miR-17/106a probe in the presence of RNase mix, which provides access to RNase A).

The main limitation of the sequential RNase protection/primer-extension approach is that some template miRNA is lost during the RNase digestion (perhaps due to miRNA:probe hybrid “breathing” as discussed above), thus rendering the assay not purely quantitative across experiments, a limitation that is shared with conventional RNase protection analysis. Thus, this assay is most suitable for determining which paralogous miRNAs are present in a complex mixture but is not as well suited for assaying the absolute abundance of these miRNAs across samples. By decreasing the time or temperature of the RNase digestion step, template loss could potentially be minimized, although the assay conditions presented here were sufficient to address our question.

The short oligonucleotide probe primer-extension assay enables the straightforward, inexpensive and, importantly, quantitative measurement of closely related miRNAs. This assay does not rely on RNase digestion and thus does not suffer from the same limitation as the sequential RNase protection/primer-extension assay; however, the short oligonucleotide probe primer-extension approach would be unable to distinguish between miR-17 and miR-106a (due to the location of the sole nucleotide difference between them). Paralogous miRNAs like miR-17 and miR-106a thus provide an example of the limitations of this assay. In general, however, this approach can be used to rapidly and quantitatively distinguish between most closely related miRNAs, which contain nucleotide differences in exploitable locations. These are nucleotide differences in the miRNAs that can be targeted during design of short oligonucleotide probes such that a perfect hybrid will form with only one member of a paralogous miRNA family. In summary, the short oligonucleotide probe primer-extension and the sequential RNase protection/primer-extension assays make it possible to determine the full complement of virtually all paralogous miRNAs.

MATERIALS AND METHODS

Parasite strains and cell culture

Tachyzoites of T. gondii (RH88 strain) and Neospora caninum (NC1 strain) were used in this study. RH88 and NC1 were maintained in vitro by serial passage on confluent monolayers of primary HFFs at 37°C with 5% CO₂ in CDMEM (DMEM [GIBCO], 10% fetal calf serum [GIBCO], 2 mM glutamine, 50 μg/mL penicillin, 50 μg/mL streptomycin). All primary HFFs were used between passages 7 and 10.

For total RNA derived from parasite-infected cells, 150-mm dishes of confluent HFFs were infected with either Toxoplasma or Neospora. When parasites reached the 64–128 parasites/vacuole stage, the infected HFF monolayers and uninfected control plates were washed twice in DMEM and then scraped and lysed by repeated passage through 27-gauge syringe needles. Parasite numbers were quantified using a hemocytometer. The following steps were performed for both mock-infected and infected cultures. Cell debris was removed by centrifuging for 1 min at 700 rpm; supernatant from the low speed spin was decanted and centrifuged for 10 min at 2000 rpm in a tabletop centrifuge at room temperature, and the medium was aspirated. Parasite pellets were resuspended in 40 mL DMEM and centrifuged in a tabletop centrifuge for 10 min at 1400 rpm at room temperature, and the medium was aspirated again.

The parasite pellet and mock-infected HFF cell debris pellet were resuspended in CDMEM and added to a fresh monolayer of confluent, recently passed primary HFFs in 150-mm dishes at a multiplicity of infection of 5. At 4 h post-infection, the medium was replaced with fresh CDMEM prewarmed to 37°C. At 24 h post-infection, the medium from mock + infected plates was aspirated and 10 mL of TRIzol (Invitrogen) containing 10 μg glycogen (Roche) was added to each plate. TRIzol-solublized cell lysates were scraped and passaged once through an 18-gauge syringe needle, and total RNA was extracted according to the manufacturer's instructions. This total RNA from TRIzol extraction was used in both the sequential RNase protection/primer-extension and short-oligonucleotide primer-extension experiments. Total RNAs were resuspended in water and stored in siliconized 1.7 mL tubes (Axygen Maxymum Recovery) at −80°C until used.

Probe labeling

Oligonucleotide probes (Integrated DNA Technologies) were 5′-end labeled with γ-[³²P]-ATP (6000 Ci/mmol, MP Biomedicals) and T4 polynucleotide kinase (New England Biolabs) according to the manufacturer's instructions; unincorporated γ-[³²P]-ATP was removed by passage of the reaction mixture through a Micro-Spin 6 column (Bio-Rad) according to the manufacturer's instructions. Twenty nanograms of 10-base-pair (bp) DNA ladder (Invitrogen) was 5′-end [³²P]-labeled under exchange-reaction conditions for 10 min at 37°C (50 mM imidazole at pH 6.4, 12 mM MgCl₂, 1 mM 2-mercaptoethanol, 50 μM ADP, 10 μCi of fresh γ-[³²P]-ATP, 10 units T4 polynucleotide kinase); unincorporated γ-[³²P]-ATP was removed by passage of the reaction mixture through a Micro-Spin 6 column. Probe sequences are as follows: 5S rRNA, 5′-AGATCGGGCGCGTTCAG; miR-17/106a, 5′-CTACCTGCACTGTAAGCAC; miR-106b long, 5′-ATCTGCACTGTCAGCAC; miR-106b short, 5′-ATCTGCACTGTC; and miR-93 short, 5′-CTACCTGCACGAAC.

Sequential RNase protection/primer-extension analysis

Six RNA oligonucleotides with nucleotide sequences identical to each human miR-17 family member were synthesized and gel-purified to obtain full-length synthetic templates (Integrated DNA Technologies) (for sequences, see Figs. 2B, 3A). We incubated 2 nmol of these RNA oligonucleotides, or 10 μg of total RNA derived from uninfected HFFs, or Neospora- or Toxoplasma-infected HFFs, with 50 nM of a [5′-³²P]-labeled DNA oligonucleotide probe specific for miR-17/106a in 1× primer-extension buffer (50 mM Tris-HCL at pH 8.3; 75 mM KCL, 3 mM MgCl₂,) in a volume of 10 μL; reaction mixes were assembled in duplicate to use in +/− RNase-treated pairs. These RNA:[5′-³²P]-labeled DNA oligonucleotide hybrids were incubated for 3 min at 95°C and slow-cooled in a heat block to 40°C. The tubes were removed, spun quickly to recover the accumulated condensation, and added to a second heat block pre-equilibrated to 30°C, where the samples were allowed to equilibrate to the block temperature for 1 min. Three hundred ninety microliters of 1× RNase digestion buffer (10 mM Tris-HCl at pH 7.5, 300 mM NaCl, 5 mM EDTA) prewarmed to 30°C was added to annealed RNA:probe mixes. For RNase-treated samples, 3.4 U RNase T1 (Roche) and 10 μg of RNase A (Sigma) were added to each 400 μL reaction mixture; and ± RNase-treated reactions were incubated for 30 min at 30°C. Reactions were terminated by addition of 400 μL phenol:chloroform (Ambion), vortexed for 1 min, and centrifuged for 15 min at maximum speed in a tabletop microfuge. Three hundred fifty microliters of the aqueous phase was carefully decanted to a fresh 1.7 mL siliconized microfuge tube. Nucleic acids were ethanol precipitated in the presence of 10 μg glycogen and rinsed once with 75% ethanol. Pellets were resuspended in 5 μL 1× primer-extension reaction buffer (which is 1× primer-extension buffer supplemented with 200 μM dNTPs [Invitrogen] and 10 mM DTT) and heated to 50°C for 5 min, and 20 units of Moloney Murine Leukemia Virus reverse transcriptase (MMLV-RT; Invitrogen) was added (20 units of MMLV-RT is 1 μL of a 1:10 dilution of reverse transcriptase in 1× primer-extension buffer). Primer-extension reactions proceeded for 5 min at 50°C, and the reactions were terminated by addition of 4 μL stop mix (45% formamide, 20 mM EDTA, 0.05% bromophenol blue, 0.05% xylene cyanol). To identify any nontemplated artifacts of primer-extension, one primer-extension reaction was performed in parallel for the miR-17/106a probe in the absence of template RNA. Samples were heated for 3 min to 95°C, snap-cooled in an ice water bath, and resolved by electrophoresis through a 12% acrylamide (Bio-Rad) + 8 M urea (Invitrogen) + 1× TBE (89 mM Tris-borate at pH 8.0, 2 mM EDTA) sequencing gel capped at 50 W for 1 h. The gel was disassembled, cut, dried, exposed to a phosphorimager screen (Amersham) for 18–24 h, and scanned on a PhosphorImager Storm (Molecular Devices).

Short-oligonucleotide primer-extension analysis

W mixed 2 nmol of the synthetic miR-17 family RNA oligonucleotides or 4 μg replicate aliquots of each total RNA sample with 50 nM of the indicated [5′-³²P]-labeled oligonucleotide probe (except 5S probe, see below) in 1× primer-extension reaction buffer. In the case of 5S primer-extensions, a mixture of 50 nM of [5′-³²P]-labeled 5S probe plus 2 μM of cold 5S probe was added to each 4 μg RNA pellet in 1× primer-extension reaction buffer. At these probe and template miRNA concentrations, after the primer-extension reaction, the free probe was present in >100-fold excess of the major primer-extension products (data not shown). Each primer-extension reaction mixture was heated for 5 min to 95°C and slow-cooled in a heat block to 60°C; the tube was removed and spun quickly to remove condensation and added to a second heat block pre-equilibrated to 48°C, where the samples were allowed to equilibrate to the block temperature for 1 min. Twenty units of MMLV-RT was added to each mixture, and the reactions were allowed to progress for 5 min at T_m=48°C. To identify any nontemplated artifacts of primer-extension, one primer-extension reaction was performed in parallel for each probe in the absence of template RNA. These 6 μL reactions were terminated by addition of 4 μL of stop mix. After primer-extension, samples were heated for 3 min to 95°C, snap-cooled in ice water, spun down in a microcentrifuge, and resolved by electrophoresis through a 12% acrylamide + 8 M urea + 1× TBE sequencing gel capped at 50 W for 1 h. The gel was disassembled, cut, dried, exposed to a phosphorimager screen for 18–24 h, and scanned.