Abstract
Two major components of rRNA (18S and 28S rRNA) were separated by electrophoresis in injection-molded acrylic chips with a microchannel 100 μm in width, 40 μm in depth, and with 1 cm of separation distance. Microchannels were filled with 4 g/L hydroxypropylmethylcellulose as sieving polymer and 5 mg/L ethidium bromide for RNA staining. The fluorescent signals were detected by a fluorescent microscope equipped with a photometer and 590 nm emission filter. The assay is rapid (<3 min), reproducible, RNase-free, and requires only 1–2 μL of sample. The detection limit was ∼10 mg/L (10 ng/μL), 100-fold lower than that for conventional agarose gel electrophoresis. Because only 0.1 nL of the loaded sample was used for electrophoresis, the detectable peaks of rRNA in the separation were derived from less RNA than in a single cell. Because the quality of RNA is critical for RNA-related diagnostic tests, disposable plastic chips will be useful for quality assessment of RNA.
mRNA is widely used in the biosciences for cDNA cloning, construction of expressed sequence tag databases, and gene expression analysis. In diagnostic molecular pathology, mRNA is also frequently used to detect or quantitate the levels of specific gene expression, such as bcr-abl translocation of the Philadelphia chromosome in leukemia (1), point mutation of the p53 gene in breast cancer (2), or prostate-specific antigen transcript for micrometastasis of prostate cancer (3). Many technologies and commercial kits are available for the purification of mRNA or total RNA from cells and tissues. However, the quality of mRNA is a major concern for scientists because of its extreme instability against contaminating RNases. This is particularly important when the goal is to isolate full-length cDNA or to analyze gene expression quantitatively, as in clinical diagnostics.
The ratio of optical absorbance at 260 nm and 280 nm is the most common technique for the quality assessment of RNA; however, this method provides only information about whether proteins are contaminating the samples. Although Northern blot or reverse transcription-PCR is used to detect housekeeping genes, the existence of PCR products of these genes in samples does not guarantee that the RNA is entirely intact. We previously developed an assay to determine the amount of total mRNA by capturing mRNA onto oligo(dT)-immobilized microplates, followed by YOYO-1 (Molecular Probes) fluorescence measurement (4) or colorimetric detection of incorporated biotin-mononucleotides during cDNA synthesis on the microplate (5). However, quantification of mRNA does not guarantee that the mRNA is free from degradation, because partially digested mRNA may be captured by oligo(dT). When purified mRNA is separated by agarose gel electrophoresis and stained with ethidium bromide, one can see the smear of mRNA. If the smear is distributed to the large molecular weight region of the gel, the mRNA can be considered to be of good quality. However, this assay is not quantitative. Therefore, no suitable procedure is available for the analysis of the quality of mRNA.
Interestingly, a gold standard method exists for total RNA: agarose gel electrophoresis to identify two to three major bands: 28S, 18S, and 7S rRNA (6). If these bands disappear, the RNA is considered useless as a sample because the RNA has been digested by contaminating RNases during the purification procedure. Although agarose gel electrophoresis is easy, one should take extra care against RNase contamination in the electrophoresis chamber, loading buffer, separation buffer, and agarose gel. Moreover, because of the low sensitivity of ethidium bromide toward RNA, a relatively large quantity of purified RNA is consumed by agarose gel electrophoresis. Therefore, a demand exists for RNase-free easy-to-use assay tools for sensitive detection of rRNA. Here, we report on disposable plastic chips for very sensitive rRNA analysis, which require small sample volumes (<1–3 μL) for analysis.
Materials and Methods
materials
Cell culture media and appropriate antibiotics, fetal calf serum, DNA marker φX174 HaeIII (Life Technologies), λ HindIII (Promega), and U937 cells (American Type Culture Collection) were purchased from designated suppliers. All other chemicals were purchased from Sigma Chemical Co. Plastic chips for microchannel electrophoresis (RNA chips) were supplied at Goshomiya Works, Hitachi Chemical (Shimodate, Japan; Fig. 1) and at Soane BioSciences [Hayward, CA; see McCormick et al. (8)].
Configuration of the RNA chip.
(A) Each number represents the length in millimeters. The dimensions of each chip is ∼25 × 75 mm. I–IV indicate four independent reservoirs, and A–C are locations of physical examination. The depth of the injection (vertical channel between II and IV) and separation (horizontal channel between I and III) channels is 40 μm, and sample was applied to reservoir IV. (B) The RNA chip is covered with thin plastic film on the side with the channels.
analysis of surface characteristics of microchannels
The RNA chip was loaded into a vacuum chamber, and the profiles of three different portions of the microchannel on the chip were analyzed by three-dimensional scanning electron microscopy (ERA-8000, Elionix) (7).
rna sample preparation
U937 cells were grown in RPMI 1640 containing 100 000 units/L penicillin, 100 mg/L streptomycin, and 100 mL/L fetal calf serum at 37 °C in CO2:air, 5:95 (by volume), and were subcultured two to three times a week, as described previously (4). Viability was always >90%, as assessed by the exclusion of trypan blue. The number of cells was determined with a hemocytometer. Fresh lung tissues were collected from rats and immediately processed for RNA preparation. Samples were mixed with chaotropic buffer (Toyobo) and applied to an automatic RNA extractor (MFX-2000, Toyobo). RNA was bound to silica beads; unbound materials were then removed by magnetic separation, according to the manufacturer’s protocol.
conventional agarose gel electrophoresis
RNA samples were mixed with loading buffer (2.5 g/L bromphenol blue, 2.5 g/L xylene cyanol FF, 1 mmol/L EDTA, and 400 g/L sucrose in diethylpyrocarbonate-treated water) and applied to a 1.0% agarose gel, which was prepared in 1× Tris-borate-EDTA buffer. Electrophoresis was conducted in 1× Tris-borate-EDTA buffer containing 0.1 mg/L ethidium bromide, at a constant voltage of 100 V for ∼30 min.
equipment for microchannel electrophoresis
RNA chips were mounted on a PlexiglasTM electrophoresis stage (Soane BioSciences) and placed on a fluorescent microscope (Microphot-FXA, Nikon) equipped with a 20× CF Plan ELWD objective lens (Technical Instruments), a 546/10 nm excitation filter, a 580 nm dichroic mirror, a 590 nm G-1B emission filter (Nikon), and a 100 W mercury lamp (model HB 10101AF). Fluorescent signals were collected by a photometer (D104B, Photon Technology International), and digital data were collected on a personal computer equipped with interface boards (DT 2837 A/D and DT 2815 D/A board, Data Translation), as described previously (8). The high voltage power supply was purchased from Soane BioSciences.
procedure for microchannel electrophoresis
Separation buffer consisting of 4 g/L hydroxypropylmethylcellulose (HPMC), 44.75 mmol/L Tris, 44.75 mmol/L boric acid, pH 8.0, and 5 mg/L ethidium bromide was loaded into reservoirs I, II, and III; the buffer filled the channels by both capillary action and vacuum from reservoir IV until all air bubbles were removed. RNA samples were then applied to reservoir IV. To apply samples into the injection channel (vertical channel in Fig. 1A), electrophoresis was conducted with voltages of 100, 300, 0, and 0 V at reservoirs I, II, III, and IV, respectively. By viewing the cross section between the injection and separation channels (horizontal channel in Fig. 1A) with the fluorescent microscope, we confirmed that sample had crossed the separation channel. The microscope was then moved to the detection point, 0.5–5 cm downstream from the cross section between the injection and separation channels. Electrophoretic separation was started by applying voltages of 0, 500, 1000, and 500 V at reservoirs I, II, III, and IV, respectively.
Results
characteristics of channels
As shown in Fig. 2 , the surface of the channel was smooth, and the shape of the channel was quite similar among three different locations, with the size of the orifice being 3133.8, 2975.2, and 3461.9 μm at locations of A, B, and C, respectively (see Fig. 1A).
Evaluation of physical configuration of a microchannel.
The structure of three different portions of microchannel (A, B, and C in Fig. 1A) were analyzed by three-dimensional scanning electron microscope (ERA-8000, Elionix). Each graph represents an area 0.5 × 0.5 mm in size, with 200× magnification.
separation profile of small fragments of dna
A DNA size marker, HaeIII digest of φX174, was applied to the chip, and separation was detected 5 cm downstream from the cross section between the injection and separation channels. As shown in Fig. 3 , all DNA fragments from 72 to 1353 bp were clearly separated with a linear relationship between size (log scale) and separation time (linear scale; Fig. 3 , inset). Moreover, the fluorescence intensity of the larger fragments was always higher than that of the smaller fragments. This is reasonable because each fragment in the mixture exists in the same molar concentration, and large DNA fragments will react with more ethidium bromide, giving more-intense bands. Another point of emphasis is the separation of the 271- and 281-bp fragments, suggesting that the chip provides at least 10-bp resolution under these conditions.
Microchannel electrophoresis of the DNA marker φX174 HaeIII in the injection-molded plastic microchip.
Three microliters of DNA marker (40 mg/L) was applied to reservoir IV, and electrophoresis was conducted with voltages of 100, 300, 0, and 0 V at reservoirs I, II, III, and IV, respectively. The microscope was set 5 cm downstream from the cross section between the injection and separation channels. Electrophoretic separation was started by applying voltages of 0, 500, 1000, and 500 V at reservoirs I, II, III, and IV, respectively, and the fluorescence of 590 nm (relative fluorescence unit, RFU) was measured. Each number represents the size of fragments of HaeIII digests, and the inset is a different representation (size in log scale vs time in linear scale) of the same data.
To determine the detection limit, the same DNA marker was diluted from 1 g/L to 1 mg/L (1 μg/μL to 1 ng/μL), and electrophoresis was conducted in the chip. We set the detection point at 4 cm downstream from the cross section between the injection and separation channels. As shown in Fig. 4 , fluorescent signals of the higher molecular weight fragments (>603 bp in length) were detected for DNA samples at concentrations as low as 0.01 g/L (0.01 μg/μL).
Dose dependency of the DNA marker φX174 HaeIII on microchannel electrophoresis.
Different concentrations of the DNA marker φX174 HaeIII [A, 1 g/L (1 μg/μL); B, 100 mg/L (100 ng/μL); C, 10 mg/L (10 ng/μL); and D, 1 mg/L (1 ng/μL)] were applied to the microchip, and electrophoresis was conducted as described in Materials and Methods and in Fig. 3. The detection point was 4 cm downstream from the cross section between the injection and separation channels.
separation profile of rna
In general, RNA is separated conventionally by denaturing gel electrophoresis to prevent complicated secondary structure (9)(10). However, two rRNA bands are clearly and reproducibly identified in conventional agarose gel electrophoresis (Fig. 5 , inset). Therefore, in this study RNA was separated in the chip with 4 g/L HPMC as the sieving polymer without any denaturing agent. As shown in Fig. 5 , both 28S and 18S RNA were separated in the chip even with a separation length as short as 1 cm (Fig. 5A , upper trace), although a longer separation channel (4 cm) exhibited better resolution (Fig. 5A , lower trace). Small RNA fragments containing 7S rRNA, 4–5S tRNA, and possibly RNA fragments migrated earlier than 18S rRNA and formed additional peak(s). The RNAs were not separated without using sieving polymers, and the optimal concentration of HPMC was 4 g/L. HPMC >4 g/L is very viscous and is difficult to fill into entire channels. HMPC <0.4 g/L exhibits poor resolution for the separation of 18S and 28S rRNA (data not shown). To determine the approximate size of each peak, a λ HindIII marker was also separated in the same chip. As shown in the lower trace of Fig. 5B , the largest RNA peak migrated at the spot similar to 2027–2322 bp, suggesting that the peak was 28S rRNA. Interestingly, the second largest peak (probably 18S rRNA) migrated at the spot similar to 564 bp in RNA chips, whereas in agarose gel electrophoresis this RNA migrated more slowly than 564 bp.
Microchannel electrophoresis of total RNA extracted from rat lung.
RNA [4 mg/L (4 μg/mL)] was purified from fresh rat lung as described in Materials and Methods and applied to two different microchips (A and B) for electrophoresis. Fluorescent signals were detected at 1 cm (A, upper trace; B, all 3 traces) or 3 cm (A, lower trace) from the cross section between the injection and separation channels. The DNA marker λ HindIII [50 mg/L (50 μg/mL)] was also analyzed in the same chip (B). (Inset), both RNA (2.5 μg in 5 μL) and DNA marker (500 ng in 6 μL) were analyzed by 1.0% agarose gel electrophoresis and stained with ethidium bromide.
The separation pattern of these three peaks (28S, 18S, and small fragments) was reproducible when the samples were repeatedly applied to the same chip (intraassay; Fig. 5B). The chip-to-chip variation was also negligible (Fig. 5). We have also analyzed various RNA samples derived from different tissues and cultured cells. The results were always consistent and equivalent to those of conventional agarose gel electrophoresis. When RNA was prepared from RNase-rich pancreas, small intestine, or granulocytes, rRNA was not detected by the chip or by agarose gel electrophoresis (data not shown).
RNA samples were diluted and applied to the chip to determine the detection limit of this assay. Interestingly, the fluorescent signals of rRNA were detected from an RNA sample equivalent to 200 cells/μL (Fig. 6). In parallel experiments, these diluted samples were also analyzed by agarose gel electrophoresis. When conventional settings of an ultraviolet illuminator and ethidium bromide stain were used, rRNA bands were visible from samples equivalent to 20 000 cells/μL (data not shown). For the RNA chip assays, the actual injection volume for each separation was ∼0.1 nL (0.0001 mm at the cross section between the injection and separation channels), suggesting that each rRNA peak was derived from less than 1/50 of the rRNA in a single cell (Fig. 6).
Dose dependency of RNA on microchannel electrophoresis.
RNA was purified from 106 U937 human leukemia cells as described in Materials and Methods, and various dilutions (A, equivalent to 104 cells/μL; B, 103 cells/μL; C, 102 cells/μL; and D, 10 cells/μL) were applied to the same microchip for electrophoresis.
Discussion
Two major components of rRNA (18S and 28S rRNA) were separated in plastic chips. The assay is rapid (<3 min), reproducible, RNase-free, disposable, easy to use, and consumes only 1–2 μL of sample. More interestingly, the detection limit of this chip-based assay was approximately 100-fold lower than that of conventional agarose gel electrophoresis, because rRNA bands were detected from RNA samples equivalent to 200 cells/μL in the chip, whereas 20 000 cell/μL was required to visualize rRNA in conventional agarose gel electrophoresis stained with ethidium bromide. This may be attributable to the difference in light source (mercury lamp vs ultraviolet lamp), detection method (photometer vs Polaroid), and sharpness or diffusion of rRNA bands. Because only 0.1 nL from the loaded samples was actually used for electrophoresis, the detectable rRNA peaks in Figs. 5 and 6 were derived from rRNA equivalent to 1/20 to 1/100 of a single cell. In Fig. 2 , the actual detection limit of HaeIII digest was 10 mg/L (10 ng/μL, or 1 pg/0.1 nL). These two independent results (1/50 of a cell vs 1 pg) are in good agreement, because we usually obtain ∼10 μg of total RNA from 10 cells. Furthermore, the detection limit of this assay can be improved even more by replacing the mercury lamp with a high powered laser and switching from ethidium bromide to more sensitive dyes, such as PicoGreen and YO-PRO-1 (Molecular Probes). Moreover, we apply 1–10 μL of samples in the chip, although only ∼0.1 nL of the sample is consumed for the separation. If one can handle nanoliter quantities of liquid, this chip will become extremely useful.
The elution time was consistent when samples of the same concentration were applied; however, the elution time was shifted, depending on the concentration of DNA or RNA. This was confirmed at three different locations: the Hitachi Chemical Research Center in Irvine, CA; Soane Biosciences in Hayward, CA; and the Goshomiya Works, Hitachi Chemical Co., Shimodate, Japan. The amount of incorporated ethidium bromide may be different, depending on the concentrations of DNA or RNA, and this induces different electrophoretic characteristics as well as electroosmotic flow. The DNA:DNA, RNA:RNA, or DNA/RNA:HPMC interaction may be altered in different concentration of nucleic acids, and this may induce different migration patterns. Although migration time varied among different concentrations of DNA or RNA, two peaks of rRNA (18S and 28S) were always detected separately in any concentrations.
We have also separated rRNA in conventional capillary electrophoresis equipped with a laser (Beckman). However, the electrophoresis on the chip exhibited major advantages over capillary electrophoresis. First, we can view the condition of channels anywhere within the chip, using the fluorescent microscope or even by eye, which helps us to confirm electrophoresis and to solve many mechanical troubles, such as bubbling and plugging. We can also observe fluorescent signals during electrophoresis and confirm that samples are migrating in the right direction. In addition, we can easily change the length of electrophoresis by moving the detection point anywhere along the channels. Because capillary electrophoresis reuses the same capillary over and over again, extensive washing steps are unavoidable between electrophoretic runs. This is problematic because the surface characteristics of the inner wall of the capillary may change after the washing process and may produce irreproducible results. It also requires time-consuming steps to exchange capillaries. More importantly, if an RNase-rich sample is applied, one should take extra care to remove any contaminating RNases from the capillaries. These problems can be entirely eliminated by using a disposable plastic chip.
Recently, microfabricated chip-based electrophoresis has been reported from several laboratories (11)(12)(13). However, these chips utilized glass or silica as substrates, which are not suitable for mass production. Application of these glass/silica chips is also limited to DNA at this time. In contrast to DNA applications, rRNA separation does not require high resolution and allows us to use channels 100 μm in width. We have also evaluated narrower plastic channels; however, narrow channels produced many technical problems. The chips are made by injection-molded technology, similar to that of McCormick et al. (8), and can be produced in large quantities at low cost. Appropriate acrylic substrates were also selected to minimize autofluorescence. Furthermore, during the manufacturing process, thin plastic films are automatically adhered on the surface of the chip to cover the injection and separation channels. This process makes the chip RNase-free. Therefore, microchannel electrophoresis on a plastic chip as described in this study is ideal for rRNA analysis.
The presence of rRNA in the sample does not always guarantee that mRNA is free from degradation. Some RNases may degrade mRNA more than rRNA. The complicated secondary structure of rRNA may prevent the attack of RNases. In fact, substantial degradation of specific mRNA is sometimes observed in a Northern blot, in spite of the presence of rRNA. We do not know whether mRNA degradation occurs physiologically in the living cells or during the purification process. However, if rRNA disappears from the purified RNA samples, there is no argument that the sample is useless for further analysis. Therefore, we believe that the present chip is useful as an initial quality-control method for any RNA-related experiments and diagnostics.
We thank T. Shimayama, H. Watanabe, and R. Kato (Hitachi Chemical Co., Tokyo, Japan), and M.G. Alonso-Amigo, R.J. Nelson, and D.J. Benvegnu (Soane BioSciences, Hayward, CA) for their technical support and helpful comments and suggestions.
References
Gala J-L, Heusterspreute M, Loric S, Hanon F, Tombal B, Van Cangh P, et al. Expression of prostate-specific antigen and prostate-specific membrane antigen transcripts in blood cells: implications for the detection of hematogenous prostate cells and standardization.
Sambrook J, Fritsch EF, Maniatis T. Extraction, purification and analysis of messenger RNA from eukaryotic cells. In: Molecular cloning, a laboratory manual, 2nd ed. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press, 1989:7.1–7.87..
