Abstract
The accurate localization of proteins in fixed cells is important for many studies in cell biology, but good fixation is often antagonistic to good immunolabeling, given the density of well-preserved cells and the size of most labeled antibody probes. We therefore explored the use of single-stranded oligonucleotides (aptamers), which can bind to proteins with very high affinity and specificity but which are only ∼10 kD. To evaluate these probes for general protein localization, we sought an aptamer that binds to a widely used protein tag, the green fluorescent protein (GFP). Although this quest was not successful, we were able to solve several practical problems that will confront any such labeling effort, e.g., the rates at which oligonucleotides enter fixed cells of different kinds and the extent of nonspecific oligonucleotide binding to both mammalian and yeast cell structures. Because such localization methods would be of particular value for electron microscopy of optimally fixed material, we also explored the solubility of aptamers under conditions suitable for freeze-substitution fixation. We found that aptamers are sufficiently soluble in cold organic solvents to encourage the view that this approach may be useful for the localization of specific proteins in context of cellular fine structure.
Keywords
T
Novel methods for labeling proteins are therefore under investigation (e.g., Gaietta et al. 2002). Here we explored the possibility of using single-stranded nucleic acids, or “aptamers”, as probes for protein localization in cells. Aptamer labeling has the potential to overcome several of the problems that currently limit antibody labeling. Like antibodies, these macromolecules can bind tightly and specifically to their ligands (Ellington and Szostak 1990; Tuerk and Gold 1990). Such binding is sometimes even tighter than antigen–antibody interaction (Jenison et al. 1994), opening the possibility that aptamer labeling could be even more specific. Aptamers have already been found that show promise for labeling of both surface proteins (Blank et al. 2001) and cytoplasmic enzymes (Bianchini et al. 2001), so there is a precedent for success with this approach. The molecular mass of a high-affinity aptamer is often <10 kD, significantly smaller than Fab fragments (∼50 kD). They are about the size of the smallest engineered single-chain antibodies (∼12 kD), so they should diffuse comparatively rapidly into cells. Unlike antibodies, however, short fragments of DNA/RNA are soluble in both aqueous buffers and high concentrations of organic solvents, so probe–target binding might even be accomplished at low temperatures during freeze-substitution. Indeed, aptamers could be selected to bind a chosen tag, analogous to an epitope, in any preferred organic solvent, so the quality of the labeling could be optimized by probe variation and selection in vitro (Tuerk and Gold 1990). If this approach could be made to work, labeling could be carried out on material that had been immobilized by rapid freezing and fixed by freeze substitution, which seems to be the most reliable way to preserve cell or tissue structure for subsequent morphological study at high resolution (reviewed in McIntosh 2001). We have therefore set out to explore the utility of aptamers for protein localization by light and electron microscopy.
The green fluorescent protein (GFP) has been used by many labs as a chimera with a protein of interest to provide information on protein localization in vivo (reviewed in Tsien 1998). It follows that an aptamerlabeling protocol that recognizes GFP would be of immediate and widespread use. Moreover, the β-can structure of GFP shows remarkable stability, including a resistance to both heat and protease denaturation (Bokman and Ward 1981; Ward and Bokman 1982; Yang et al. 1996). It therefore seemed likely that GFP would retain its structure during freeze-substitution and might continue to bind an appropriate aptamer during preparation for visualization in the electron microscope (EM). Here we explore the feasibility of aptamer tagging by measuring the solubility of short, single-stranded DNAs (ssDNAs) under conditions that are suitable for freeze-substitution fixation. We also characterize the diffusion of ssDNA into fixed cells and assess various treatments for their ability to reduce nonspecific aptamer binding. EM is used to assess the impact of these conditions on the preservation of cell fine structure. Finally, we explore the use of GFP as a protein target for aptamer binding by describing the results of our efforts to select a high-affinity aptamer to EGFP.
Materials and Methods
Fixation of Cells for Oligonucleotide Diffusion
The wild-type fission yeast Schizosaccharomyces pombe was grown to mid-log phase in yeast extract medium plus supplements (YES) (Moreno et al. 1991). The culture was centrifuged for 3 min at 250 × g and re-suspended in fresh YES, then fixed by the double aldehyde method (Hagan and Hyams 1988). Cells were stored at 4C in 0.1 M PIPES, pH 6.9, 1 mM EGTA, and 1 mM MgCl2 (PEM) plus 0.1% sodium azide to prevent bacterial growth. In some experiments the cell wall was digested by re-suspending cells in 0.5 ml of PEM plus 1 M sorbitol and adding 0.3 mg/ml zymolyase from Arthrobacter lutens (ICN Biochemicals; Irvine, CA) and 0.5 mg/ml lysing enzyme (Sigma–Aldrich Chemicals; St Louis, MO). The cells were incubated at 37C until shells from the cell wall were visible in the light microscope and the wall's birefringence was lost. Cells were then washed three times in PEM with 1 M sorbitol and stored as described above.
PtK1 cells were grown on glass coverslips to 75% confluence and fixed for 15 min using 4% paraformaldehyde in PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na2PO4, 2 mM KH2PO4, pH 7.4). The cells were washed three times with PBS for 5 min each. Then their membranes were permeabilized in a 1:1 mixture of methanol:acetone for 30 sec, followed by five washes with PBS for five min each. Fixed cells were stored at 4C in PBS plus 0.1% sodium azide.
ssDNA Diffusion into Cells
Fixed S. pombe cells were incubated overnight in 100–300 nM fluorescent ssDNA from IDT (Chicago, IL), using the desired buffer at room temperature (RT) with rocking. The cells were then centrifuged as described above and re-suspended in the same buffer without fluorescent oligonucleotides. Images were collected on a Zeiss Axiophot II using epifluorescence optics and a Cooke SensiCam charge coupled device (CCD) camera. Captured images were analyzed using Slidebook software (3I Inc; Denver, CO). When appropriate, either S. pombe or PtK1 cells were treated with RNase A (Sigma–Aldrich) at 40 μg/ml in 10 mM Tris, pH 7.5, 5 mM EDTA, 3 M NaCl for 2 hr at 37C. The cells were washed twice with a 1:1 mixture of formamide with 2 × standard sodium citrate (SSC, i.e., 0.3 M NaCl, 30 mM sodium citrate, pH 7.0) and then washed twice with 2 × SSC for 30 min each. After the fixed cells had been treated with RNase, they were incubated in ssDNA for 10–30 min and then washed as described above.
Fixed PtK1 cells were incubated in 100–300 nM ssDNA for 10–30 min before being rinsed in buffer and examined in the light microscope. Cellular nucleic acids were stained with 4 μg/ml propidium iodide in PEM at RT for 30 min with rocking. When RNase treatment of PtK1 cells was needed, the protocol used was the same as that described for S. pombe above. Buffer components used to reduce background ssDNA binding were tested at the concentrations shown in Table 1.
Assessing the Solubility of ssDNA in Organic Solvents
Unless otherwise stated, nucleic acids were handled by standard methods, as described in Sambrook and Russell (2001). Lyophilized ssDNAs (Genosys Biotechnologies; Woodlands, TX) were re-suspended in sterile Milli-Q purified water and their concentrations measured by A260. The oligonucleotides were end-labeled with [γ-32P]-ATP and purified by electrophoresis on denaturing polyacrylamide gels. Salt, buffer, and either methanol or acetone were added to the required concentrations and the solution was vortexed, then equilibrated to the desired temperature (25, 4, −70C). When necessary, the oligonucleotide strands were separated or denatured by heating for 2 min at 95C, then added to the aqueous/organic solution and vortexed again. These solutions were allowed to stand overnight at the desired temperature and then spun in a microcentrifuge at 14,000 rpm for 10 min to pellet precipitated ssDNA. The supernatant liquid was aspirated and the pellet was re-suspended in 100 μl sterile water. Amounts of ssDNA in both samples were measured by adding an aliquot to a vial containing 5 ml Ecolume (ICN Biochemicals; Irvine, CA) and measuring their radioactivity with a scintillation counter. ssDNA dissolved in water was used as a positive control for solubility. The negative control was ssDNA in 300 mM sodium acetate, pH 5.2, precipitated with 2.5 × volumes of ethanol.
Concentrations of compounds used to reduce background oligo binding
a6.67 nM fluorescent oligo used to probe.
Purification of EGFP
EGFP, the F64L, S65T allele of GFP that is both brighter and more thermostable than wild-type GFP (Heim and Tsien 1996), was synthesized in bacteria and purified using the IMPACT-CN system (New England Biolabs; Beverly, MA). DNA containing this gene was amplified by the polymerase chain reaction (PCR) from pREP42-EGFP (Craven et al. 1998), using PFX DNA polymerase (Invitrogen; Carlsbad, CA) and primers with the sequences 5'GCAAAAAACCCCTCAAGACCCG3’ and 5'GGTGGTTGCTCTTCCAACATGAGTAAAGGAGAAGAACTTTTCACT3'. Conditions for PCR were 95C for 2 min, followed by 25 cycles of 94C for 1 min, 50C for 1 min, 72C for 30 sec. This reaction added a Sap1 site at the 5’ of the product and a Pst1 site at its 3’ end. The PCR products were purified on a QiaQwick column (Qiagen; Valencia, CA), digested with Sap1 and Pst1, ligated into the pTYB11 plasmid, and transformed into competent IMPACT E. coli strain ER2566 for expression (Chong et al. 1997).
Sequencing of the expression constructs revealed two alleles of EGFP: the expected sequence and a second sequence carrying a mutation that changed the protein product by S147P (EGFP-mut1). This allele has previously been isolated and was shown to display no differences from wild-type in its fluorescence spectrum, but it has an increased stability to elevated temperature (Kimata et al. 1997). It was therefore used for studies of protein fluorescence in organic solvents.
Cells carrying plasmids that contained either pTYB11-EGFP or pTYB11-EGFP-mut1 were grown at 37C in the presence of 100 μg/ml ampicillin to OD600 0.5–0.8. The culture was then induced by the addition of IPTG to 0.5 mM and grown at 25C for 6 hr. Cells were collected by centrifugation at 5000 × g for 10 min at 4C and sonicated with a Digital Sonifier (Model 450; Bransen Ultrasonics, Danbury, CT) for 7 min at 50% amplitude using 10 sec on, 60 sec off at 4C in column buffer (20 mM Tris, pH 8.0, 500 mM NaCl). This lysate was clarified by centrifugation at 12,000 × g and 4C for 30 min, then loaded onto a column of chitin beads that had been equilibrated with column buffer and washed with 10 bed volumes of the same buffer. The protein was cleaved from the chitin-binding domain by flushing the column with 50 mM DTT, then incubating it at RT overnight. EGFP or EGFP-mut1 was eluted by washing the column with 20 ml of column buffer. The fractions containing protein were combined and dialyzed against 0.1 × column buffer, then concentrated using a Centricon-30 (Millipore; Bedford, MA). Protein concentrations were determined by the Bradford assay (BioRad Laboratories; Hercules, CA).
Purified Protein in Organic Solvents
Purified EGFP-mut1 protein was added directly to acetone or methanol solutions containing salts and buffers, and the solutions were equilibrated for several hours. Fluorescence and emission spectra were collected with a spectrofluorimeter (model 48000S from SLM–Aminco; Champagne–Urbana, IL).
Fixation of Yeast Cells by Freeze-substitution
S. pombe was grown to mid-log phase in Edinburgh minimal medium (EMM; Moreno et al. 1991) plus leucine. The strain used (DW551) expressed EGFP conjugated with the nuclear protein Mcl1p (Williams and McIntosh 2002). The cells were collected on a membrane by filtration under vacuum, transferred to the sample chambers of a BAL-TEC HPM-010 high-pressure freezer (Technotrade International; Manchester, NH), and rapidly frozen under high pressure, all as described previously (Ding et al. 1997), then stored in liquid nitrogen. Frozen cells were freeze-substituted in acetone for 24 hr at −90C and stored at −20C until use. Such samples were serially substituted into mixtures of water with methanol or acetone at 4C and equilibrated in each mixture for 30 min with rocking. The water concentration in successive mixtures was increased by no more than 10% in each substitution step. After a solution containing the desired concentrations of water and organic solvent had been reached, the cells were stored for 24–72 hours at 4C.
Imaging Cells Treated with Aptamers
Light microscope images were recorded and measured with the instruments described above. To evaluate the effect on cell ultrastructure of treatments that reduced nonspecific binding of oligonucleotides, mammalian cells were prepared for EM. PtK1 cells were cultured on coverslips as described above, then fixed for 1 hr with 2% paraformaldehyde and 0.1% glutaraldehyde in PEM at 37C. Fixed samples were rinsed three times in PBS, then soaked in either PBS for 4 hr at RT (control) or for an equal time in the buffer to be tested. They were then rinsed twice in PBS and postfixed in 0.5% OsO4 with 0.8% Fe(CN)6 for 5 min on ice (McDonald 1984). After three more rinses in PBS, the cells were treated with 0.15% tannic acid in PBS for 3 min, rinsed in distilled water, dehydrated in an acetone series, and embedded in Epon/Araldite. Sections (70 nm) were cut on a Leica Ul-tracut E, post-stained in uranyl acetate and lead citrate, then imaged at 80 KeV on a Philips CM10 electron microscope.
Primary and secondary structures of the three aptamers used in this study. A, T, G, and C represent the bases used. Aptamer 1 is compact, thanks to its hairpin conformation, whereas aptamers 2 and 3 are extended. CY5 indicates the position of a cyanine-5 fluorescent dye, and FC marks a fluorescein.
Using SELEX to Find a GFP-binding Aptamer
No aptamers with a high affinity for GFP have previously been described, so we sought our own through two rounds of in vitro selection. An ssDNA library containing a 30-nucleotide (nt) region with random sequence was synthesized by Operon Technologies (Alameda, CA) as described in Fitzwater and Polisky (1996). For each selection, EGFP at a range of concentrations from 1 to 1000 nM was bound to nitrocellulose, then incubated with the pool of single-stranded oligonucleotides. Bound ssDNA was extracted using 1:1 phenol:chloroform at pH 7.9 with 7 M urea. The selected ssDNA strands were amplified by PCR using a biotinylated primer. The strands were then incubated with 20 μg strepavidin and separated by electrophoresis on a 12% acrylamine, 8 M urea gel. Selections were attempted at both pH 6.5 and 5.2, and each employed counter-SELEX to prevent the amplification of nitrocellulose-binding aptamers.
Entry and nonspecific binding of fluorescently labeled ss-DNA into fixed, permeabilized S. pombe cells. (
Results
Aptamer Entry into Fixed Cells
A useful probe for protein localization should diffuse readily into fixed cells. We have studied the diffusion of oligonucleotides into aldehyde-treated, permeabilized mammalian cells and fission yeast, using the fluorescent ssDNAs diagramed in Figure 1. The time required for oligonucleotide entry varied with cell type. Diffusion of ssDNA into fixed, permeabilized PtK1 cells required only a few seconds (data not shown) and therefore was not studied further. Entry into S. pombe, on the other hand, was very slow (Figure 2). Even after a day of incubation, oligonucleotides were not detectable in fixed cells with intact walls (Figure 2A). The negative charges in the wall probably inhibited ssDNA entry by electrostatic repulsion. After digestion of the walls with lysozyme and zymolyase, oligonucleotide entry was still slow, but an intracellular signal was detectable by 8 hr (Figure 2B).
The signal-to-noise ratio available from an aptamer-labeling protocol will, of course, depend in part on the level of nonspecific background binding of ssDNAs. We found significant background binding in both cell types, regardless of the oligonucleotide sequence or fluorescent tag used. However, the patterns of binding were quite different in the two cell types. Oligonucleotides bound mostly to the nucleus of PtK1 cells (Figure 3), whereas the diffusing oligonucleotides bound nonspecifically throughout S. pombe cells (Figure 2B). These patterns were seen despite variation in the length, sequence, and secondary structure of the ssDNA. Moreover, fluorescein or CY5 label resulted in similar binding, and neither aptamer binding nor fluorescence stability was affected appreciably by pH. Free fluorescein dye did not bind significantly to the cells, nor did it inhibit the binding of fluorescent oligonucleotides (data not shown). Therefore, the binding patterns appear to be a general property of ssDNA.
Nonspecific binding of fluorescently labeled ssDNAs to fixed, permeabilized PtK1 cells. (
With short oligonucleotide probes, nonspecific binding to cellular DNA or RNA is a particular concern. The distributions of ssDNA binding seen in both cell types corresponded roughly to the location of high concentrations of RNA, so RNase was tested as a way to decrease background binding. RNA is, of course, located throughout both the cytoplasm and the nucleus, whereas DNA is restricted to the nucleus and the mitochondria. Before RNase treatment, most of the fission yeast cell stained brightly with propidium iodide. After RNase treatment, only the nucleus was stained. DNase treatment was not attempted because the distribution of nonspecific oligo binding corresponded better to the pattern of cellular RNA than of DNA. RNase A treatment of PtK1 cells reduced background binding by almost 50% (not shown). In fixed S. pombe, a 2-hr treatment with RNase A rid the cytoplasm of RNA, as indicated by the propidium iodide staining pattern, but it increased the background binding of oligonucleotide by more than ninefold. The target of this nonspecific binding of ssDNA is not known, but there is probably a good deal of it because excess non-fluorescent oligonucleotides did not compete significantly with labeled ones to reduce this fluorescence.
We therefore sought buffer conditions that would reduce the extent of nonspecific oligonucleotide binding. The goals of in situ hybridization (ISH) resemble those of our study: to localize intracellular macromolecules by binding a nucleic acid probe specifically to a fixed ligand (complementary RNA or DNA for ISH and a protein for aptamer labeling). Components of such protocols therefore seemed worth testing for their effects on nonspecific aptamer binding. However, some of these components are detrimental to protein conformation or stability and would therefore be unsuitable for the aptamer-labeling methodology. Additional chemical treatments, such as competitors and various buffer components, were also tested for their efficacy in reducing nonspecific binding of ssDNA to aldehyde-fixed PtK1 and yeast cells. Chemicals tested included zwitterions, lysine, free phosphate ions, and nonfluorescent ssDNA.
In PtK1 cells, 0.5 M HEPES/0.5 M PIPES buffer was exceptionally effective at decreasing nonspecific binding, giving an 80% reduction, whereas lysine had a negligible effect. Effects on nonspecific binding in S. pombe cells were similar (data not shown). The use of free phosphate (500 mM, pH 7.5) or excess non-fluorescent ssDNA before, during, or after binding of the fluorescent ssDNA had a small but detectable effect (data not shown).
Ability of individual buffer components to reduce nonspecific background staining a
aComponents were evaluated at the concentrations described in Table 1 and in Materials and Methods. + and - indicate the relative strength of the effects seen. Lysine was not evaluated in S. pombe. nd, not done.
Pretreatment of PtK1 cells with 0.1 M triethanolamine (TEA) plus 0.25% acetic anhydride (freshly prepared) reduced the average background signal by 50%, but this protocol had no effect on ssDNA binding in S. pombe cells (not shown). The buffer for ISH (1.2 M NaCl, 20 mM Tris-HCl, pH 7.5, 4 mM EDTA, 1 mg/ml yeast tRNA, 0.4 mg/ml Ficol, 0.4 mg/ml polyvinylpyrrolidone, and 0.4 mg/ml bovine serum albumin plus 50% formamide) had an even greater effect on nonspecific binding in PtK1 cells (79% reduction), and this effect was not dependent on TEA/acetic anhydride pretreatment. Denhardt's solution and yeast tRNA from the in situ buffer had negligible effects on non-specific binding in PtK1 cells. In S. pombe cells, formamide and ISH buffer had no effect individually, but together they almost completely eliminated nonspecific ssDNA binding (data not shown).
Quantification of background oligonucleotide binding (oligo 3) in RNase-treated S. pombe cells. (
Electron micrographs of fixed PtK1 cells, showing the effects of incubations in a control buffer (
The most effective treatments were identified and combined to form the final binding buffer. The comparative effect of each treatment is summarized in Table 2. The final buffer (HEPIN) consisted of zwitterions and selected components from the ISH buffer (0.25 M HEPES, 0.25 M PIPES, 0.05 M NaCl, 1.6 mM EDTA, 4 mg/ml yeast tRNA, 0.16 mg/ml each of Ficoll, BSA, and polyvinylpyrrolidone). This mixture was extremely effective in reducing background staining in PtK1 cells and in S. pombe cells pretreated with RNase (Figure 4). Because the nonspecific binding is apparently independent of aptamer sequence or dye chemistry, this buffer should be effective for labeling with diverse aptamers and labels. Despite its usefulness, formamide was omitted from the final buffer formulation because it can disrupt hydrogen bonding and thus secondary structure formation within oligonucleotides.
Effects of salt (
Electron Microscopy of Cells Treated to Reduce Nonspecific Binding of ssDNA
To examine the effect of conditions that reduced nonspecific binding of oligonucleotides in cultured mammalian cells, we compared the fine structure of fixed PtK1 cells treated either with a control buffer (PBS) or with HEPIN (Figure 5). Control cells at low (Fig. 5A) and medium magnification (Figure 5B) showed the expected collection of organelles and cytoskeletal fibers. HEPIN-treated cells at corresponding magnifications (Figures 5C and 5D) looked quite similar. Some vesiculation of cytoplasmic membranes was seen (Figure 5C, arrows), but both the overall organization of the cells and the fine structure of their cytoplasm were quite well preserved by a solution that minimized the nonspecific binding of aptamer probes to experimental material.
Aptamer Solubility Under Conditions for Fixation by Freeze-substitution
For aptamers to be useful probes of protein localization, they should work after the use of optimal fixations. We therefore examined the solubilities of ssDNA under conditions suitable for freeze-substitution fixation: high concentrations of methanol or acetone at low temperatures, such as −90C. Because of the effect of organic solvents on the dielectric constant of a solution, low concentrations of salt had large ionic effects, inducing the precipitation of ssDNA. This effect was assessed by determining the impact of salts on nucleotide solubility in 97% organic solvent (Figures 6A and 6B). Although the ionic strengths of 1 μM MgCl2 and 3 μM NaCl are equivalent, ssDNA was soluble in methanol at up to 3 mM NaCl but only to 10 μM MgCl2. In acetone, however, the effects of MgCl2 or NaCl were much the same, and the precipitation of ssDNA was negligible only below 1 μM MgCl2 or 3 μM NaCl.
The solubility of ssDNA was much more sensitive to buffer concentration in acetone than in methanol. Tris and HEPES (0.5–500 μM) were tested for their ability to precipitate a 108-mer ssDNA in 97% organic solvent (Figures 6C and 6D), and even at 500 nM oligonucleotide, the precipitation of ssDNA in methanol was minimal. Acetone-containing solutions showed significant precipitation of ssDNA at only 5 μM HEPES or Tris. HEPES was used for further experiments because HEPES and Tris gave similar results, but the pH of HEPES is much less temperature-dependent.
No differences were observed in the solubilities of either a 45-mer or a 108-mer ssDNA in 97% acetone at 25C and −90C (Figure 7). The 45-mer ssDNA was freely soluble at 500 nM. The 108-mer was entirely soluble at 125 nM, but even at 500 nM less than 25% of the ssDNA precipitated.
Protein Stability Under Conditions of Freeze-substitution
For an aptamer-labeling strategy to work, a protein target, such as EGFP, must retain its structure during fixation. Because fluorescence is a good indicator of EGFP's conformation, we have used this readily measurable parameter to assess the effects of different treatments on the conformation of EGFP.
The ability of cellular EGFP to maintain its fluorescence in the presence of methanol or acetone was studied using the nuclear protein Mcl1p, expressed as a chimera with EGFP in fission yeast (Williams and McIntosh 2002). Log-phase cells were fixed as described in Materials and Methods, then warmed to 4C and serially substituted into the solution of choice. Nuclear fluorescence was reduced by more than 50% in 100% acetone but it was still visible. In 70% methanol the fluorescence was virtually gone (Figure 8). As the acetone samples were rehydrated their fluorescence reappeared, suggesting that acetone is better than methanol for preserving protein structure. Acetone is also a preferred solvent for freeze-substitution fixation, so it should be satisfactory for an aptamer-labeling strategy.
Effects of ssDNA oligonucleotide length and concentration on the extent of precipitation in 96% acetone. (
Fluorescence of Mlm1p-GFP in fission yeast cells as seen after rapid freezing, freeze-substitution fixation, and rehydration in mixtures of water and organic solvent. (
The fluorescence of purified EGFP-mut1 in solution was analyzed in the presence of methanol and acetone to see whether the retention of fluorescence in cells was due to protection by other cell components. Fluorescence in 70% acetone dropped to approximately half that in water (Figure 9A), but in 70% methanol it became undetectable (Figure 9B). The only difference between these data and those obtained from cells was the lack of solution fluorescence in 100% acetone. We infer that under virtually anhydrous conditions, some cellular component (e.g., salt, additional bound water, fixation) helps to stabilize protein structure. In general, the results of the two studies were very similar, indicating that EGFP is stable even in high concentrations of acetone.
Effects of (
Efforts to Select Aptamers Specific for EGFP
To evaluate aptamer labeling for protein localization, we worked to identify an ssDNA with high affinity and specificity for EGFP. A library of single-stranded DNA molecules with 30 nts of random sequence was purchased from Operon Technologies. This collection of many different ssDNAs (in principle, 430 but in practice closer to 1010) was used from which to seek an aptamer that bound tightly to EGFP. Two aqueous selections were performed using the SELEX procedure (Tuerk and Gold 1990; Morris et al. 1998). One was at pH 5.2 and another at pH 6.5. Both selections failed to identify a suitable nucleotide sequence, even though they were carried out with the help of Soma-Logic (Boulder, CO), a company that specializes in such work.
Discussion
We have begun a study of aptamers as probes for protein localization in cells. On the basis of their size and possible affinity, these probes have significant promise as tools for cell biology. Our results suggest, however, that unusual treatments are necessary to permit the rapid entry of aptamers into some fixed cells; special buffers and even treatments with enzymes may be required. Our results with PtK1 cells are moderately encouraging, but S. pombe does not appear to be a good candidate for the aptamer-labeling technique. For example, the presence of large quantities of ribosomal RNA in these cells appears to limit the rate of aptamer entry, and this effect will probably reduce the extent of both specific and nonspecific aptamer binding. Diffusion times of ∼24 hr are inconvenient, and the temperatures used for freeze-substitution fixation may exacerbate this problem. For such cells it may be necessary to use chemical fixation and RNase, followed by treatments with special buffers to maximize rates of entry and to minimize nonspecific aptamer binding.
The most important challenges for the labeling of protein in cells are to achieve high levels of probe binding with a good signal-to-noise ratio in well-fixed material. To this end, we have explored the suitability of aptamers for use in organic solvents, such as those used in freeze-substitution. Our results suggest that aptamers are sufficiently soluble under these conditions to serve as effective probes. For example, other labeling techniques with DNA/RNA, such as ISH, use probe concentrations ≤100 nM. The oligonucleotide solubilities determined here suggest that almost any aptamer of 30–100 nucleotides should have a solubility of ∼500 nM under conditions suitable for freeze-substitution. Therefore, aptamers at concentrations of approximately one fifth saturation should provide ample ligand and avoid problems with precipitation.
Considerations of oligonucleotide solubility in organic solvents dictate that the final aptamer labeling buffer should contain small amounts of HEPES and MgCl2. In spite of the decreased solubility of ssDNA in MgCl2 as opposed to NaCl, MgCl2 appears preferable for the final buffer because divalent metal cations can stabilize the tertiary structures of ssRNA or ssDNA (Brion and Westhof 1997). This stabilization may be the reason for lower oligonucleotide solubility in the presence of MgCl2, since the more stable and rigid a RNA/DNA structure, the more rapidly it precipitates. Our results are therefore consistent with solubility data previously published (Gardiner et al. 1985).
Blank et al. (2001) have identified a fluorescent-tagged aptamer that binds with good avidity and specificity to a surface component of the blood vessels in a specific part of the brain. Their work shows that aptamers can be selected with a signal-to-noise ratio that is useful for cytology. Bianchini et al. (2001) have used a novel two-step selection procedure to identify a fluorescent aptamer with somewhat lower affinity for its target but with impressive specificity of binding, at least under conditions suitable for blotting. Their work suggests that an aptamer can localize a protein to the cytoplasm of a fixed, cultured mammalian cell.
Our work has sought a more widely usable aptamer probe, such as one that would bind to a commonly used protein tag like GFP. The further development of this labeling strategy will obviously require the identification of an appropriate aptamer. Our efforts with SELEX (Tuerk and Gold 1990) failed to identify a high-affinity ssDNA aptamer to EGFP. It is possible that we did isolate low-affinity aptamers; the binding assay used had only 50 μM sensitivity. However, EGFP has two characteristics that reduce the chances of finding a high-affinity aptamer: an acidic isoelectric point and a smooth surface topography. Aptamers have a marked tendency to bind preferentially to active sites or recognition surfaces of proteins, and they tend to fit into the binding pocket of large proteins or create a binding pocket for smaller structures such as ATP (Tian et al. 1995). EGPF binds nothing and is bound by nothing, except by itself when at high concentrations. It is possible that a larger aptamer could make more contacts and therefore have a higher affinity, but a small aptamer is necessary for rapid diffusion into whole fixed cells. A ligand other than GFP is likely to be preferable for pursuit of aptamer labeling. Once a suitable aptamer has been found, it will, of course, be necessary to tag it with something visible in the EM. Initial experiments suggest that either conversion of fluorescence by photobleaching in the presence of diaminobenzidine or tagging with a small electron-dense probe, such as NanoGold, will provide the necessary contrast.
Our evaluation of rates for ssDNA diffusion into fixed cells and of conditions that minimize nonspecific aptamer binding has been useful as an initial study, but the evaluation of our approach will require an aptamer that binds selectively to a cell protein. The signal-to-noise ratio achieved with such an aptamer will be the final arbiter of the usefulness of such cell preparations. To test our approach, we are now seeking high-affinity aptamers for small proteins that may work better than GFP as ligands for cellular localization.
Footnotes
Acknowledgements
Supported in part by a grant from the Keck Foundation to the University of Colorado and in part by NIH grants RR00592 and PO1GM61306 to JRM, who is a Research Professor of the American Cancer Society.
We thank Drew Smith of SomaLogic (Boulder, CO) for his help with selecting aptamers.