If a given DNA oligo is an aptamer, will the corresponding RNA oligo with the same sequence be an aptamer aswell?

Functionality of aptamers depends upon the oligo's sequence and secondary structure. So, if I take a DNA aptamer and make an RNA oligo of the same sequence (T replaced with U obviously), will this RNA oligo also act as an aptamer?

There is at least one known case where both the DNA and RNA versions of the same aptamer sequence bind the same target (Lauhon and Szostak, 1995).

But you can't generalize this, there are two differences in structure between RNA and DNA, the Uracil/Thymine change and the missing 2'-OH in DNA. Those differences can affect the binding and overall structure of the aptamer.

It is possible that the RNA version of a DNA aptamer binds the same ligand, but this won't be always the case.

Aptamers are single-stranded sequences that bind to proteins or other molecular partners. Obviously, ssDNA and ssRNA have very different (in general) binding partners. E.g. DNA is interacting with transcription factors, RNA polymerase, histones etc, while RNA is interacting with ribosomes, reverse transcriptase and others.

So, shortly, it is more likely that protein A that binds DNA aptamer will not bind RNA with corresponding sequence.

Also, DNA and this DNA transcript will probably have very different secondary structure.

LNA/DNA chimeric oligomers mimic RNA aptamers targeted to the TAR RNA element of HIV-1

1 INSERM U386, Université Victor Segalen, 33076 Bordeaux cx, France, 2 Institut Europພn de Chimie et Biologie, 2 rue Escarpit, 33607 Pessac cx, France and 3 Santaris Pharma A/S, Bøge Allé 3, DK-2970 Copenhagen, Denmark

Jens Bo Hansen

1 INSERM U386, Université Victor Segalen, 33076 Bordeaux cx, France, 2 Institut Europພn de Chimie et Biologie, 2 rue Escarpit, 33607 Pessac cx, France and 3 Santaris Pharma A/S, Bøge Allé 3, DK-2970 Copenhagen, Denmark

Henrik Orum

1 INSERM U386, Université Victor Segalen, 33076 Bordeaux cx, France, 2 Institut Europພn de Chimie et Biologie, 2 rue Escarpit, 33607 Pessac cx, France and 3 Santaris Pharma A/S, Bøge Allé 3, DK-2970 Copenhagen, Denmark

Carmelo Di Primo

1 INSERM U386, Université Victor Segalen, 33076 Bordeaux cx, France, 2 Institut Europພn de Chimie et Biologie, 2 rue Escarpit, 33607 Pessac cx, France and 3 Santaris Pharma A/S, Bøge Allé 3, DK-2970 Copenhagen, Denmark

Jean-Jacques Toulmé

1 INSERM U386, Université Victor Segalen, 33076 Bordeaux cx, France, 2 Institut Europພn de Chimie et Biologie, 2 rue Escarpit, 33607 Pessac cx, France and 3 Santaris Pharma A/S, Bøge Allé 3, DK-2970 Copenhagen, Denmark

A ssDNA Aptamer That Blocks the Function of the Anti-FLAG M2 Antibody

Using SELEX (systematic evolution of ligands by exponential enrichment), we serendipitously discovered a ssDNA aptamer that binds selectively to the anti-FLAG M2 antibody. The aptamer consisted of two motifs (CCTTA and TGTCTWCC) separated by 2-3 bases, and the elimination of one or the other motif abrogated binding. The DNA aptamer and FLAG peptide competed for binding to the antigen-binding pocket of the M2 antibody. In addition, the aptamer eluted FLAG-tagged proteins from the antibody, suggesting a commercial application in protein purification. These findings demonstrate the feasibility of using SELEX to develop ssDNA aptamers that block the function of a specific antibody, a capability that could lead to the development of novel therapeutic modalities for patients with systemic lupus erythematosus, rheumatoid arthritis, and other autoimmune diseases.

1. Introduction

Antinuclear antibodies are diagnostic markers of systemic lupus erythematosus, rheumatoid arthritis, and other autoimmune diseases [1]. In these B lymphocyte disorders, a large variety of autoantibodies are made against nuclear self-antigens, including ribonucleoproteins, nucleosomes, chromatin, and polynucleotides (RNA, ssDNA, and dsDNA). Among these, anti-DNA antibodies have been the most extensively studied [2]. Anti-DNA antibodies bind with high-affinity to either single- or double-stranded DNA and many tend to favor association with pyrimidine bases [3, 4]. Several reports have also described antinuclear antibodies cross-reacting with peptide self-antigens and depositing in the brain, kidneys, and skin [5–9]. As proposed by several investigators, this deposition may be a cause of inflammation-mediated tissue damage, especially in the kidneys where nephritis is a major source of morbidity [1, 2]. In mouse models of systemic lupus erythematosus, attempts were made to block the function of these cross-reacting antibodies using peptide aptamers, derived either from their cognate peptide self-antigens or from phage display libraries [10, 11]. In some cases, the peptide aptamer competitively associated with the antinuclear autoantibodies, thereby preventing antibody-mediated tissue damage [10, 11]. Thus, direct antibody inhibition might be an effective therapy in patients with autoimmune diseases driven by the presence of antinuclear antibodies. Another viable approach to block antinuclear antibodies might be to use DNA aptamers, given the high-affinity of these antibodies for DNA and evidence of nucleotide base specificity. But this approach has clearly been underexplored, perhaps due to the lack of reports on the feasibility of developing DNA aptamers to block the function of specific antibodies.

An adaptive technique employed to define the sequence specificity of DNA/RNA-binding proteins is SELEX (systematic evolution of ligands by exponential enrichment). In SELEX, the protein of interest is used as a selection matrix to capture high-affinity DNA binding sites from a pool of randomized DNA molecules [12, 13]. This pool is comprised of an oligonucleotide that contains a randomized core (up to 35 bases in size) flanked by PCR priming sequences. The randomized core is made during chemical synthesis using a mixture of all four nucleoside phosphoramidites at each of the random positions. Following their capture, the selected DNA molecules are reamplified by PCR and then further enriched through successive rounds of selection. After 4–6 rounds, the selected DNA molecules are cloned and sequenced to identify any common DNA motifs recognized by the protein of interest. SELEX can be applied to the selection of ssDNA, dsDNA, or even RNA molecules [12, 13]. It is a powerful tool that has been used to optimize nucleic acid ligands for a multitude of proteins, even some which do not normally interact with DNA or RNA. As an example, SELEX was utilized to develop RNA aptamers that bind to blood coagulation factors, including thrombin [14], Von Willebrand factor [15], and Factor IXa [16]. In all three cases, the selected RNA aptamers interacted selectively with their corresponding protein targets and, in the process, inhibited their blood coagulation activities. A second generation of aptamers was developed, and, among these, some have entered clinical trials in patients with blood coagulation disorders [15].

Using SELEX, we serendipitously discovered a ssDNA sequence that binds selectively to the M2 antibody, a commonly used reagent that recognizes the Flag epitope (DYKDDDDK). The DNA aptamer and Flag peptide competed for binding to the M2 antibody, thereby allowing the aptamer to elute Flag-tagged proteins from an immobilized M2 antibody, a commonly employed procedure in protein purification. Aside from this immediate application in protein purification, identification of this DNA aptamer demonstrates the feasibility of using SELEX to develop aptamers that block specific antibodies. Applying this approach to antinuclear autoantibodies could lead to the development of novel therapeutic strategies for patients with systemic lupus erythematosus, rheumatoid arthritis, and other autoimmune diseases.

2. Materials and Methods

2.1. Materials

Oligonucleotides were synthesized by the Eppley Core Facility (University of Nebraska Medical Center, Omaha, NE). Plasmid pTetFLAGhTRF2 45-501 was a gift from Dr. Titia de Lange (Rockefeller University, New York, NY) [17]. The polynucleotide kinase and the Platinum Pfx and Taq DNA polymerases were purchased from Invitrogen (Carlsbad, CA). All other enzymes were obtained from Fermentas (Hanover, MD), New England BioLabs (Beverly, MA), Promega (Madison, WI), or Invitrogen (Carlsbad, CA). The TnT Quick coupled Transcription/Translation System was purchased from Promega (Madison, WI). The γ-[ 32 P]-ATP (4500 Ci/mmol) was purchased from MP Biologicals (Solon, OH), and the L-[ 35 S]-Methionine (1000 Ci/mmol) was obtained from PerkinElmer (Boston, MA). M-450 magnetic beads coated with a sheep anti-mouse IgG antibody were received from Dynal Biotech, Inc., (Lake Success, NY). 3XFLAG peptide and all other chemicals were from Sigma-Aldrich (St. Louis, MO). The Mini-PROTEAN 4%–20% SDS-PAGE gels were purchased from Bio-Rad (Hercules, CA).

2.2. Antibodies

Mouse monoclonal antibodies against the Flag tag (IgG1 clone M2) and StnI/OBFC1 (IgG2ak clone 3G12-1B7) were purchased from Sigma-Aldrich (St. Louis, MO). Normal mouse IgG (cat. # sc-2025) and anti-vimentin mouse monoclonal antibody (IgG1 clone sc-6260) were obtained from Santa Cruz (Santa Cruz, CA), as was the rabbit polyclonal antibody against TRF2 (cat. # H-300). Also purchased were mouse monoclonal antibodies against PTOP (IgG1 clone 1D8-1B6 Novus Biologicals, Littleton, CO) and TIN2 (IgG1 clone 59B388, AbCam, Cambridge, MA).

2.3. Expression Vectors

The plasmid pcDNA3.1-Flag-Stn1 was made by insertion of the human Stn1 sequences into plasmid pcDNA3.1-Flag. The coding sequence of Stn1 was amplified from plasmid pOTB7-Stn1 (GenBank # BC017400 American Type Culture Collection, Manassas, VA) using the Platinum Pfx DNA polymerase and primers 5′-GACTGACAATTGGGTGGTATGCAGCCTGGATCCAG-CC-3′and 5′-GACTGAAGATCTTCAGAACGCTGTGTAGTAGTG-3′. The PCR product was then cut with MfeI/BglII and inserted into the EcoRI/BamHI sites of pcDNA3.1-Flag, in frame with the Flag tag. pcDNA3.1-Flag is a pcDNA3.1(−) vector encoding a Flag epitope located downstream of a T7 promoter and immediately followed by an EcoRI site. Plasmid pcDNA3.1-Flag-TRF2 ΔB was made from its pCMV1 equivalents by transfer of its TRF2 cassette to vector pcDNA3.1(−) (Invitrogen, Carlsbad, CA). pCMV1-Flag-TRF2 ΔB was in turn made by the transfer to pCMV1 of a SacII/BamHI fragment from plasmid pTetFLAGhTRF2 45-501 (a gift from Titia de Lange, Rockefeller University, NY) [17].

2.4. Protein Expression

The Flag-Stn1 and [ 35 S]-Flag-TRF2 ΔB proteins were made by in vitro transcription/translation in a rabbit reticulocyte lysate. In a final volume of 50 μL, one microgram of pcDNA3.1-Flag-Stn1 was transcribed/translated using the TnT Quick Coupled system, according to the manufacturer’s instructions (Promega, Madison, WI). A water-programmed lysate (mock) was produced in parallel to serve as a negative control. After translation, aliquots of the two reactions were analyzed by western blotting using both the Stn1/OBFC1 antibody and anti-Flag M2 antibody. A single species of 45 kDa was detected by the two antibodies in the Stn1-programmed lysate but not the Mock lysate (See Figure S1 in Supplementary Material available online at doi: 10.4061/2011/720798). The [ 35 S]-labeled Flag-TRF2 ΔB protein was produced similarly with the exception that the unlabeled methionine was replaced with 2 μL of L-[ 35 S]-methionine (1000 Ci/mmol, 10 μCi/μL).

2.5. Preparation of Beads Coated with the Anti-Flag M2 Antibody

M2 antibody-coated beads were prepared by mixing 200 μL of M-450 magnetic beads coated with a sheep anti-mouse IgG antibody (Dynal Biotech, Inc., Lake Success, NY) with 15 μg of anti-Flag antibody (M2 mouse monoclonal, Stratagene, La Jolla, CA) in 5 mL of PBS containing 0.1% each of NP-40 and BSA. After overnight rotation at 4°C, M2-coated beads were washed 3 times in PBS containing 0.1% each of NP-40 and BSA, after which beads were suspended and stored in 200 μL of the same buffer. Before use, beads were washed 3 times with ice-cold 1X binding buffer containing 0.1% BSA. Control beads coated with normal mouse IgG (cat. # sc-2025, Santa Cruz) were prepared following the same exact protocol.

2.6. SELEX

An oligonucleotide containing a random core of 35 nucleotides flanked by PCR priming sequences was made: 5′-GCGTCGACAAGCTTTCTAGA(N)35GAATTCGGATCCCTCGAGCG-3′. In the first round of selection, 5 μg of this randomized oligo (215 pmoles, 130 trillion molecules) was incubated with 12.5 μL of rabbit reticulocyte lysate (programmed with either water or Flag-Stn1) and 5 μg of sonicated denatured E. coli genomic DNA in a 50 μL reaction containing 1X binding buffer (4% glycerol, 1 mM MgCl2, 1 mM DTT, 50 mM NaCl, 10 mM Tris-HCl pH 7.5). After 30 minutes at room temperature, 10 μL of M2-coated beads were added and the samples were rotated at room temperature for 1 hour. Magnetic beads were washed 3 times with ice-cold 1X binding buffer containing 0.1% BSA, after which the selected oligos were eluted for 10 minutes at room temperature in the presence of an excess of 3XFLAG peptide (34 μM in 1X binding buffer). Next, the eluted DNA was reamplified by PCR using Platinum Taq DNA polymerase and primers 5′-GCGTCGACAAGCTTTCTAGA-3′ (forward) and 5′-CGCTCGAGGGATCCGAATTC-3′ (reverse). Aliquots taken after 10, 15, 20, and 25 cycles of PCR were resolved on a 3% agarose gel. The most optimally amplified aliquot (no smear, no supershift, within the exponential range) was selected, cut, and gel purified using the GENECLEAN III kit (MP Biologicals, Solon, OH). Next, the gel-purified products were subjected to asymmetric PCR to regenerate ssDNA molecules needed for the following rounds of SELEX. Sixteen cycles of asymmetric PCR were performed using Platinum Taq DNA polymerase and the forward primer, after which the PCR products were extracted with phenol : chloroform (1 : 1), chloroform only, and then ethanol precipitated. The ssDNA pellet was dissolved in water and was ready to be used in the next round of SELEX. Additional round of SELEX were done identically, except that the randomized oligo was replaced with the previously selected and reamplified ssDNA.

After the sixth round of selection, the symmetrically amplified PCR product was gel purified and then prepared for TA-cloning into vector pCR2.1-Topo (Invitrogen, Carlsbad, CA). Chemically transformed TOP10 E. coli cells were spread onto plates containing kanamycin (25 μg/mL). A total of 50 white colonies were picked and sent for sequencing. Over 30 sequences were obtained for each of the two SELEX procedures performed (mock versus Flag-Stn1).

2.7. Radiolabeling of DNA Probes

In 20 μL of forward reaction buffer (10 mM MgCl2, 5 mM DTT, and 70 mM Tris-HCl, pH 7.6), 30 pmoles of oligonucleotide were labeled for 15 minutes at 37°C with 75 μCi of γ-[ 32 P]-ATP (4500 Ci/mmole) and 10 units of T4 polynucleotide kinase (Invitrogen, Carlsbad, CA). Next, the probes were resolved by electrophoresis on an 8% polyacrylamide gel and were gel purified. Lastly, the eluted probes were desalted on a G50 spun column (GE Healthcare, Piscataway, NJ).

2.8. Electrophoretic Mobility Shift Assay (EMSA)

In 25 μL of 1X binding buffer, reactions contained 1 μg of the indicated antibody and 80,000 cpm of the designated [ 32 P]-labeled oligonucleotide. Reactions were incubated at room temperature for 30 minutes, and were then loaded onto a 4% native polyacrylamide gel containing TBE buffer (45 mM Tris-borate, 2 mM EDTA, pH 8.3). Gels were run at 4°C for 1 hour at 180 Volts. Gels were then transferred to a DE81 anion exchange chromatography paper (Whatman International, Maidstone, England), dried, and exposed to a PhosphorImager screen (Molecular Dynamics, Sunnyvale, CA). In competition experiments, the indicated competitors were added 5 minutes prior to the addition of the [ 32 P]-labeled ABA probe. The competitors used included an increasing concentration of 3XFLAG peptide (2, 5, 10, 20, 50, 100, 200, or 500 nM) or an excess of either the ABA or CTR oligo (500 nM each).

2.9. Binding Isotherm of the ABA Probe

Experiments were performed with [ 32 P]-labeled ABA as described above, except that the M2 antibody was added at an increasing concentration (0, 50, 100, 200, 350, 500, 750, 1000, 2000, and 4000 ng). After exposure of the gel to a PhosphorImager screen, the amount of free [ 32 P]-ABA and amount of [ 32 P]-ABA bound to the M2 antibody were quantified by volume integration using ImageQuant program (Molecular Dynamics, Sunnyvale, CA). The amount of [ 32 P]-ABA bound, expressed as a fraction of the total, was plotted as a function of total amount of M2 antibody. The resulting curves were fitted by nonlinear regression to a “one-site” saturation binding curve (

). Fitting, performed using SigmaPlot version 11.0, allowed calculation of the apparent dissociation constant ( ).

2.10. Release of the Captured [ 32 P]-ABA Oligo by the 3XFLAG Peptide

The [ 32 P]-labeled ABA oligo was first immobilized onto the M2-coated beads. Briefly, 150 μL of M2-coated beads were added to 375 μL of 1X binding buffer containing 600,000 cpm of [ 32 P]-ABA oligo. Capture of the oligo was allowed to proceed for one hour at room temperature, after which the beads were washed 3 times with 500 μL of ice-cold 1X binding buffer (containing 0.1% BSA). The radioactive beads were then resuspended in 450 μL of the same buffer. To assess the ability of the 3XFLAG peptide to elute the captured oligo, an increasing concentration of 3XFLAG peptide (0, 2, 5, 10, 20, 50, 100, 200, 500, 1000, 2000, and 5000 nM) was added to 30 μL of beads carrying the [ 32 P]-ABA oligo. After 20 minutes at room temperature, the beads were pulled down and the amounts of radioactivity remaining on the beads and in the supernatant were counted by scintillation and expressed as a fraction of the total.

2.11. Inhibition of the Capture of [ 35 S]-Labeled Flag-TRF2 ΔB

Prior to these experiments, beads were blocked at 4°C for 30 minutes with 5% milk in 1X binding buffer to reduce nonspecific binding. In 1X binding buffer, M2-coated beads were first incubated with the indicated competitor (30 μM each 3XFLAG, 3XABA, or 3XCTR) or with no competitor (No Comp). After 10 minutes at room temperature, 2–10 μL of [ 35 S]-labeled FLAG-TRF2 ΔB were added. After 1 hour of rotation at room temperature, each reaction was washed 3 times with 500 μL of ice-cold 1X binding buffer (containing 0.1% BSA), after which the amount of [ 35 S]-Flag-TRF2 ΔB captured was quantified. In one experiment, the amount captured was quantified by SDS-PAGE electrophoresis and exposure to a PhosphorImager screen. In a second experiment done in triplicate, the amount of [ 35 S]-Flag-TRF2 ΔB captured was quantified by scintillation counting. In both experiments, beads coated with normal mouse IgG were included as a negative control for the capture (IgG).

2.12. Release of Precaptured [ 35 S]-Labeled Flag-TRF2 ΔB

The [ 35 S]-labeled Flag-TRF2 ΔB protein was first immobilized onto the M2-coated beads. Briefly, 80 μL of M2-coated beads were added to 800 μL of 1X binding buffer containing 32 μL of [ 35 S]-Flag-TRF2 ΔB . Capture of the protein was allowed to proceed for one hour at room temperature, after which the beads were washed 3 times with 500 μL of ice-cold 1X binding buffer (containing 0.1% BSA). The radioactive beads were then resuspended in 80 μL of the same buffer. To assess the ability of the 3XABA oligo to elute the captured protein, five microliters of these beads were mixed with 45 μL of 1X binding buffer containing the indicated competitor (30 μM each 3XFLAG, 3XABA, or 3XCTR) or no competitor (No Comp). After 20 minutes at room temperature, the beads were pulled down and the amount of radioactivity in the supernatants was quantified by SDS-PAGE electrophoresis and exposure to a PhosphorImager screen. In an experiment done in triplicate, the amount of [ 35 S]-Flag-TRF2 ΔB released was also quantified by scintillation counting. In both experiments, beads boiled to release all of the captured [ 35 S]-labeled protein were included as positive control for the elution (Total).

3. Results

3.1. SELEX Driven by the Anti-Flag M2 Antibody

SELEX was used to characterize the DNA-binding specificity of a human Flag-tagged Stn1 protein [18, 19], in this case produced by in vitro translation in a rabbit reticulocyte cell-free system. The in vitro translated Flag-Stn1 was mixed with a pool of randomized ssDNA molecules and then, the anti-Flag M2 antibody was used to immunoprecipitate the Flag-Stn1/DNA complexes to select for high-affinity targets. After six successive rounds of selection, the selected ssDNA molecules were cloned and sequenced. Analysis of a related unselected random pool revealed a slight bias for G-rich sequences, but no evidence of any recurrent motifs [20]. In contrast, analysis of the Flag-Stn1 selected oligonucleotides revealed a bipartite consensus motif made of two distinct elements: CCTTA and TGTCTWCC (where W = A/T Figure 1(a)). These elements were separated by 2-3 bases of poorly conserved sequences. However, the same bipartite consensus motif was also produced when SELEX was performed with the same antibody but using a mock-programmed reticulocyte lysate as a control source of proteins (Figure 1(b)). Based on these results, we concluded that the bipartite consensus had been selected, not by the Flag-Stn1 protein, but by the M2 antibody itself or proteins of the lysate that the antibody cross-reacted with.

(b) SELEX with or without the Flag-Stn1 protein yields the same bipartite consensus. A randomized pool of ssDNA molecules was incubated with an in vitro translation system programmed with either a Flag-Stn1 plasmid (Flag-Stn1 present) or water (Flag-Stn1 absent), after which protein/DNA complexes were recovered by immunoprecipitation using the anti-Flag M2 antibody. After six successive rounds of selection, the remaining oligonucleotides were cloned and sequenced. (a) Oligonucleotides selected in the presence of the Flag-Stn1 protein. The selected ssDNA molecules share a common bipartite consensus, which consists of a CCTTA (pink) and TGTCTWCC (green) motifs. Nonnucleotide letters denote ambiguous sequencing reads (R = A/G, Y = C/T, S = C/G, W = A/T, K = T/G, M = A/C). (b) Oligonucleotides selected in the absence of the Flag-Stn1 protein. The same bipartite consensus is shared among the ssDNA molecules selected by the mock-programmed in vitro translation system.
3.2. The M2 Antibody Binds the Consensus ssDNA Motifs

To test the possibility of direct interactions between the bipartite consensus and M2 antibody, EMSA was performed. A series of equal length [ 32 P]-labeled ssDNA probes containing or lacking each element of the consensus were made (Figure 2(a)). Probes AB, BA, and ABA carried both the CCTTA and TGTCTWCC motifs, whereas probe 2XB and 4XA consisted of multiple copies of one or the other element. The unrelated CTR probe was used as negative control. Figure 2(b) shows that the M2 antibody could form a stable complex with the ABA probe and to a lesser extent, the AB probe. None of the other probes interacted with the M2 antibody. Also noteworthy, neither the BA probe lacking the CCTTA element nor the AA probe lacking the TGTCTWCC element interacted with the antibody, thereby indicating that both elements are essential for binding. Figure 2(b) also shows that among 7 different antibodies, the M2 antibody only interacted with probe ABA. These results show that probe ABA binds very selectively to the M2 antibody, thereby suggesting that these interactions are involving the variable regions of the antibody, whether part of the light chain, heavy chain, or both.

(d) The bipartite ssDNA consensus binds directly to the anti-Flag M2 antibody. (a) Sequence and graphical representation of the probes used. All probes were labeled at the 5′-end with [ 32 P]. Probes were designed to carry none, one, or both of the identified motifs. Motif A = CCTTA (pink). Motif B = TGTCTWCC (green). (b) The anti-Flag M2 antibody binds selectively to probe ABA. The indicated probes (80,000 cpm) were incubated with the listed antibodies (1 μg), after which protein/DNA complexes were resolved by electrophoresis in a native polyacrylamide gel. (c) Titration of the M2 antibody. A constant amount of probe ABA was incubated with an increasing concentration of M2 antibody, and the protein/DNA complexes were resolved by native electrophoresis. At the higher antibody concentrations, a larger protein/DNA complex is observed (top arrow), which may represent the product of antibody oligomerization. (d) Binding isotherm of probe ABA interacting with the anti-Flag M2 antibody. The scatter plot shows the amount of ABA bound (amount present in both protein/DNA complexes) as a function of antibody concentration. Dotted line shows fitting of the data to a “one site” saturation binding curve. Nonlinear regression of the data allowed for the determination of the dissociation constant (

nM). Error on the value of the

To measure the affinity of the M2 antibody for probe ABA, the apparent dissociation constant ( ) of the complex was determined. A binding isotherm was generated by EMSA using a constant amount of [ 32 P]-ABA incubated with an increasing concentration of M2 antibody (Figure 2(c)). Plotting the amount of ABA bound as a function of the total concentration of antibody generated a binding isotherm. Fitting of this isotherm to a “one site” saturation binding curve allowed for calculation of the . In this case, the apparent was determined to have a value of nM (Figure 2(d)). The value is comparable to the apparent of other previously reported DNA aptamer/antibody complexes [21]. This result shows that the M2 antibody binds with high affinity to its preferred DNA ligand, the ABA probe.

3.3. The Flag Peptide Competes with the Binding of the ABA Oligo to the M2 Antibody

The M2 antibody binds to both the Flag peptide (DYKDDDDK) and the ABA probe, so we sought to determine whether the two ligands could compete for each other. In a first series of experiments, we asked if an excess of 3XFLAG peptide (MDYKDHDGDYKDHDIDYKDDDDK) could block the binding of [ 32 P]-labeled ABA to the M2 antibody. In the EMSA shown in Figure 3(a), an increasing concentration of 3XFLAG peptide was incubated with the M2 antibody prior to adding the [ 32 P]-ABA. At concentrations of 50 nM and higher, the 3XFLAG peptide could completely block the binding of ABA to the M2 antibody. A similar inhibition was produced following the inclusion of an excess of unlabeled ABA oligo, but not after the addition of the CTR oligo. These results show that the 3XFLAG peptide competes with the binding of probe ABA to the M2 antibody.

(b) The 3XFLAG peptide blocks the binding of ABA to the M2 antibody. (a) The 3XFLAG peptide prevents formation of the ABA/M2 complex. The M2 antibody was incubated with an increasing concentration of 3XFLAG peptide after which the [ 32 P]-ABA probe was added. Protein/DNA complexes were resolved by electrophoresis in a native polyacrylamide gel. (b) The 3XFLAG peptide elutes the ABA probe already bound to the M2 antibody. The [ 32 P]-ABA was first captured by magnetic beads coated with the M2 antibody. The beads were then incubated with an increasing concentration of 3XFLAG peptide and the amount of [ 32 P]-ABA released and remaining on the beads was counted by scintillation (

In a second series of experiments, we asked whether an excess of 3XFLAG peptide could elute a [ 32 P]-ABA oligo already bound to the M2 antibody. For this purpose, [ 32 P]-ABA was first captured by magnetic beads coated with the M2 antibody, after which an increasing concentration of 3XFLAG peptide was added. In Figure 3(b), the amount of ABA bound to the beads and the amount released from the beads were plotted as a function of the concentration of the 3XFLAG peptide. The data show that the ABA oligo was very efficiently released by the addition of the 3XFLAG peptide. A small fraction of oligo, representing 36%, could not be displaced, but the other fraction was almost completely released by 2000 nM of 3XFLAG peptide. Fitting the data to a “one site” competition curve allowed for the calculation of an IC50. In agreement with the values of the , the IC50 was determined to be 96 nM (95% confidence interval: 72–127 nM). These results show that the 3XFLAG peptide can elute the ABA oligo when bound to the M2 antibody.

3.4. The 3XABA Oligo Blocks the Association of Flag-Tagged Proteins with the M2 Antibody

If the ABA oligo and the Flag peptide compete for binding to the M2 antibody, then an excess of ABA oligo could potentially allow the elution of Flag-tagged proteins from the M2-coated beads, a commonly used procedure in protein purification. To address this possibility, we first determined whether an excess of ABA oligo could block binding of [S 35 ]-labeled Flag-TRF2 ΔB to M2-coated beads. In these experiments, the M2-coated beads were first incubated with the competitors (30 μM of 3XFLAG peptide, 3XABA or 3XCTR oligos) or with no competitor (No Comp). Then, the [S 35 ]-labeled Flag-TRF2 ΔB was added and the amount of [S 35 ] captured by the beads was measured by either SDS-PAGE electrophoresis (Figure 4(a)) or scintillation counting (Figure 4(b)). As Figures 4(a)–4(b) show, the 3XFLAG peptide abolished the capture of the [S 35 ]-labeled protein whereas the 3XCTR oligo had no effect. Although not as active as the 3XFLAG peptide, the 3XABA oligo inhibited the capture of Flag-TRF2 ΔB by 62%. The shorter ABA aptamer was also tested but found to be less active (Figure S2), with ABA blocking the capture by 45% only.

(d) 3XABA oligonucleotide blocks the interaction of Flag-tagged proteins with the M2 antibody. (a)-(b) The 3XABA oligo blocks the binding of Flag-TRF2 ΔB to M2-coated beads. Magnetic beads coated with the M2 antibody were incubated in the absence (No Comp) or presence of the indicated competitor (3XABA, 3XCTR, 3XFLAG). In vitro translated [ 35 S]-labeled Flag-TRF2 ΔB was then added and the amount captured by the beads was determined by SDS-PAGE electrophoresis and exposure to a PhophorImager cassette (a). In a second experiment done in triplicate, the amount of [ 35 S]-labeled protein captured was counted by scintillation (b). The amount of [ 35 S]-labeled protein captured in the absence of competitor (No Comp) was arbitrarily set to 100%. In both experiments, beads coated with normal mouse IgG were included as negative control for the capture (IgG). Data represent the mean ± S.D. (

). (c)-(d) The 3XABA oligo elutes the Flag-TRF2 ΔB proteins already bound to M2-coated beads. The [ 35 S]-Flag-TRF2 ΔB protein was first captured by magnetic beads coated with the M2 antibody. The beads were then incubated in the absence (No comp) or presence of the indicated competitor (3XABA, 3XCTR, 3XFLAG). The amount of [ 35 S]-Flag-TRF2 ΔB released was determined by SDS-PAGE electrophoresis and exposure to a PhophorImager cassette (c). In a second experiment done in triplicate, the amount of [ 35 S]-labeled protein released was counted by scintillation (d). The amount of [ 35 S]-labeled protein released by the boiling (total) was arbitrarily set to 100%. In both experiments, beads boiled to release to all of the captured [ 35 S]-labeled protein were included as positive control for the elution (Total). Data represent the mean ± S.D. (

Next, we asked whether an excess of 3XABA oligo could elute an [S 35 ]-labeled Flag-TRF2 ΔB protein already bound to the M2-coated beads. In these experiments, the [S 35 ]-labeled protein was first captured by the M2-coated beads. Next, the radioactive beads were exposed to the competitors (30 μM of 3XFLAG peptide, 3XABA or 3XCTR oligos) or to no competitor (No Comp), and the amount of [S 35 ] released was quantified by either SDS-PAGE electrophoresis (Figure 4(c)) or scintillation counting (Figure 4(d)). As Figures 4(c)-4(d) show, the 3XFLAG peptide caused the release of 54% of the captured protein whereas the 3XCTR oligo had no activity. The 3XABA oligo was almost as effective as the 3XFLAG peptide, causing the release of 36% of the bound protein.

4. Discussion

The initial intent of this study was to use SELEX to characterize the ssDNA-binding specificity of Flag-tagged Stn1, a subunit of the CST complex [18, 19]. Instead, the SELEX procedure enriched for ssDNA molecules that bind with high affinity to the anti-Flag M2 antibody. How was this selection by the antibody even possible? First, it may be that the in vitro translated Flag-Stn1 protein did not possess the minimum required DNA-binding activity. Improper folding could have contributed to this lack of activity. Alternatively, it may be that the OB-fold domain (oligonucleotide/oligosaccharide binding) of Stn1 is involved in mediating protein-protein interactions rather than DNA-binding [18, 19]. Second, the Flag epitope (DYKDDDDYK) is highly negatively charged, making it possible for an optimized DNA molecule to mimic the electrostatic signature of the peptide. Third, ssDNA is far more flexible than dsDNA, and in that respect, could more readily adapt to the positively charged surface of the antigen-binding pocket of the M2 antibody [22]. In other words, we propose that selection mediated by the antibody would not have been possible if the Flag epitope had been neutral or positively charged, if SELEX had been performed using dsDNA, or if the Flag-tagged Stn1 protein had exhibited a higher affinity for ssDNA. In fact, we have successfully used SELEX under similar conditions to define the ssDNA binding specificity of POT1, a protein that recognizes telomeric DNA. This SELEX, also performed using the M2 antibody, almost exclusively selected for telomeric sequences with no evidence of the bipartite motif detected (Choi, K. H. and Ouellette, M. M., manuscript in preparation). The selection by SELEX of off-target aptamers is not uncommon, especially when selecting single-stranded nucleic acids [23]. Off-target aptamers that bind to the selection matrix (e.g. avidin, M2 antibody), as we have observed here, have also been reported by others [23]. Hence, caution and experimental controls are necessary, especially when selecting ssDNA molecules for their binding to a protein of unknown functionality. Under these conditions, an important control to perform, as we have done here, is a mock selection to verify that the molecules were indeed selected by the protein of interest.

Our findings show that the identified bipartite consensus binds selectively and with high affinity to the M2 antibody. An important finding was the specificity of this interaction. For example, the consensus did not bind to any of the other antibodies tested, many of which belonging to the same isotype as the M2 antibody (mouse IgG1). This specificity implied that the interactions involved the variable regions of the antibody, which together form the antigen-binding pocket. This assumption was confirmed by the results of the competition experiments. As Figure 4 shows, the ssDNA consensus could very efficiently displace the Flag peptide from the antigen-binding pocket of the M2 antibody. Conversely, the Flag peptide could similarly displace the DNA aptamer from the antibody as well (see Figure 3). This competition between the Flag peptide and DNA aptamer implies that the two are recognized by overlapping binding determinants at the surface of the antibody. A shared feature of the Flag peptide and ABA aptamer is their negative electrostatic charges. On the surface of the antibody, these negative charges on the two ligands are likely to interact with some of the same binding determinants. The binding pocket of the M2 antibody appears to be specific only for the first 4 amino acids of the Flag peptide, namely, DYKD [22]. Two clusters of positively charged residues (lysines, arginine, histidine) are present in the binding pocket, each capable of coordinating an aspartate within the DYKD motif. The same two clusters could potentially interact with phosphate groups present within the backbone of the ABA oligo. Also important for the binding are two tyrosine residues, each one capable of forming stacking interactions with the tyrosine of DYKD. When bound to the ABA oligo, the same tyrosines could provide stacking interactions to some of the nucleotide bases. Yet, these limited interactions would not suffice to explain the sequence specificity observed. The selected consensus consisted of two separate motifs (CCTTA and TGTCTWCC) and mutating one or the other was sufficient to abrogate binding. To allow recognition of this large bipartite consensus, other residues must be implicated that make contact with several of the nucleotide bases. Hence, it must be that the Flag peptide interacts with a subset only of all the residues implicated in the binding of ABA. If so, then how could the Flag peptide be so efficient at displacing ABA from the M2 antibody? It is possible that the binding determinants that the two ligands share are especially critical for the binding of ABA. A second possibility would be that the binding of the Flag peptide causes changes in the conformation of the antibody, and that these allosteric alterations are incompatible with the binding of ABA. How the ABA aptamer is able to mimic the structure of the Flag peptide remains to be determined, but this form of mimicry between peptide and DNA has previously been reported to play an important role in autoimmune diseases and may also have important applications in drug design [5–9]. The ABA oligo and Flag peptide offer an ideal system with which to study the structural basis of this form of molecular mimicry.

One commonly employed procedure in protein purification is the capture of Flag-tagged proteins by the M2 antibody and their subsequent release by incubation with an excess of 3XFLAG peptide [24]. This procedure allows the elution of the purified proteins under nondenaturing conditions. However, the procedure leaves the eluted protein in solution with an excess of 3XFLAG peptide, which can pose a problem if the Flag tag needs to remain functional. A DNA aptamer capable of eluting the Flag-tagged proteins, such as 3XABA, could alleviate these potential drawbacks. Once the protein is eluted, the DNA aptamer could conveniently be inactivated by an antisense oligo or eliminated using trace amounts of DNAse. Alternatively, if the aptamer is biotinylated, its removal could be accomplished using streptavidin-coated beads. Finally, a DNA aptamer would also offer the possibility of using mutated oligos as negative controls or for the differential elution of captured proteins.

These results described here may also have implications for the treatment of autoimmune diseases driven by antinuclear antibodies. Previous attempts to block the function of these autoantibodies have been made with peptide mimics, either isolated from phage display libraries or derived from peptide self-antigens with which these antinuclear antibodies cross-react [10, 11]. These results described therein show that SELEX can be used to identify DNA aptamers that block the function of a particular antibody. Could this approach be used to develop aptamers that block the function of antinuclear autoantibodies? In certain autoimmune diseases, blocking the function of antinuclear autoantibodies has been shown to limit tissue damage [10, 11]. The sequence specificity of antinuclear autoantibodies has not been systematically investigated [3, 4]. SELEX would represent an ideal method for characterizing the sequence specificity of antinuclear autoantibodies. The SELEX data could serve as starting point for the design of second-generation aptamers that might be useful to block the function of these autoantibodies. The development of these DNA aptamers could therefore represent an alternative approach to the treatment of patients afflicted with systemic lupus erythematosus, rheumatoid arthritis, and other autoimmune diseases.

5. Conclusion

In summary, SELEX identified a DNA aptamer that binds directly to the antigen-binding pocket of the anti-Flag M2 antibody. The DNA aptamer and Flag peptide competed for binding to the M2 antibody and the aptamer eluted Flag-tagged proteins from an immobilized M2 antibody. This new reagent therefore offers an alternative method for the elution of Flag-tagged proteins bound to the M2 antibody, a commonly employed procedure in protein purification. Aside from this immediate application in protein purification, identification of this bipartite consensus demonstrates the feasibility of using SELEX to develop DNA aptamers that block specific antibodies. Applying this approach to autoantibodies, particularly the antinuclear antibodies, could lead to the development of novel therapeutic strategies for patients with systemic lupus erythematosus, rheumatoid arthritis, and other autoimmune diseases.


The authors wish to thank Titia de Lange (Rockefeller University, New York, NY) for the pTetFLAGhTRF2 45-501 plasmid. Appreciation is extended to Asserewou Honoré Etekpo for the quality of his technical assistance. They also wish to acknowledge the UNMC Eppley Cancer Center Molecular Biology Core and the UNMC DNA Sequencing Core Facility. This paper was supported by Grants from the National Institute of Health (P50CA127297, P30CA036727) and a GAANN fellowship from US Department of Education (P200A090064) to A. S. Lakamp.

Supplementary Materials

Figure S1: Western blot analysis of the in vitro Flag-Stn1 protein.

Figure S2: Relative efficacy with which the different aptamers block the interaction of Flag-tagged proteins with the M2 antibody.


  1. R. Lyons, S. Narain, C. Nichols, M. Satoh, and W. H. Reeves, “Effective use of autoantibody tests in the diagnosis of systemic autoimmune disease,” Annals of the New York Academy of Sciences, vol. 1050, pp. 217–228, 2005. View at: Publisher Site | Google Scholar
  2. Y. J. Jang and B. D. Stollar, “Anti-DNA antibodies: aspects of structure and pathogenicity,” Cellular and Molecular Life Sciences, vol. 60, no. 2, pp. 309–320, 2003. View at: Publisher Site | Google Scholar
  3. D. S. Pisetsky and S. A. Caster, “Binding specificity of a monoclonal anti-DNA antibody,” Molecular Immunology, vol. 19, no. 5, pp. 645–650, 1982. View at: Publisher Site | Google Scholar
  4. T. Sasaki, T. Muryoi, Y. Sekiguchi, E. Tamate, K. Yoshinaga, and Y. Kitagawa, “Monoclonal human anti-DNA antibodies from EB virus-transformed lymphocytes of systemic lupus erythematosus (SLE) patients,” Journal of Clinical Immunology, vol. 5, no. 4, pp. 246–253, 1985. View at: Google Scholar
  5. L. A. DeGiorgio, K. N. Konstantinov, S. C. Lee, J. A. Hardin, B. T. Volpe, and B. Diamond, “A subset of lupus anti-DNA antibodies cross-reacts with the NR2 glutamate receptor in systemic lupus erythematosus,” Nature Medicine, vol. 7, no. 11, pp. 1189–1193, 2001. View at: Publisher Site | Google Scholar
  6. B. Deocharan, X. Qing, E. Beger, and C. Putterman, “Antigenic triggers and molecular targets for anti-double-stranded DNA antibodies,” Lupus, vol. 11, no. 12, pp. 865–871, 2002. View at: Publisher Site | Google Scholar
  7. T. W. Faust, E. H. Chang, C. Kowal et al., “Neurotoxic lupus autoantibodies alter brain function through two distinct mechanisms,” Proceedings of the National Academy of Sciences of the United States of America, vol. 107, no. 43, pp. 18569–18574, 2010. View at: Publisher Site | Google Scholar
  8. J. B. Lefkowith, M. Kiehl, J. Rubenstein et al., “Heterogeneity and clinical significance of glomerular-binding antibodies in systemic lupus erythematosus,” Journal of Clinical Investigation, vol. 98, no. 6, pp. 1373–1380, 1996. View at: Google Scholar
  9. J. S. Rice, C. Kowal, B. T. Volpe, L. A. DeGiorgio, and B. Diamond, “Molecular mimicry: anti-DNA antibodies bind microbial and nonnucleic acid self-antigens,” Current Topics in Microbiology and Immunology, vol. 296, pp. 137–151, 2005. View at: Google Scholar
  10. B. Gaynor, C. Putterman, P. Valadon, L. Spatz, M. D. Scharff, and B. Diamond, “Peptide inhibition of glomerular deposition of an anti-DNA antibody,” Proceedings of the National Academy of Sciences of the United States of America, vol. 94, no. 5, pp. 1955–1960, 1997. View at: Publisher Site | Google Scholar
  11. H. B. Lee, B. A. Diamond, and M. D. Blaufox, “In vivo detection of deposition of radiolabeled lupus antikidney antibody and its inhibition by soluble antigen,” Journal of Nuclear Medicine, vol. 42, no. 1, pp. 138–140, 2001. View at: Google Scholar
  12. M. M. Ouellette and W. E. Wright, “Use of reiterative selection for defining protein-nucleic acid interactions,” Current Opinion in Biotechnology, vol. 6, no. 1, pp. 65–72, 1995. View at: Publisher Site | Google Scholar
  13. B. Zimmermann, I. Bilusic, C. Lorenz, and R. Schroeder, “Genomic SELEX: a discovery tool for genomic aptamers,” Methods, vol. 52, no. 2, pp. 125–132, 2010. View at: Publisher Site | Google Scholar
  14. D. M. Tasset, M. F. Kubik, and W. Steiner, “Oligonucleotide inhibitors of human thrombin that bind distinct epitopes,” Journal of Molecular Biology, vol. 272, no. 5, pp. 688–698, 1997. View at: Publisher Site | Google Scholar
  15. R. C. Becker, S. Oney, K. C. Becker, and B. Sullenger, “Antidote-controlled antithrombotic therapy targeting factor IXa and von willebrand factor,” Annals of the New York Academy of Sciences, vol. 1175, pp. 61–70, 2009. View at: Publisher Site | Google Scholar
  16. C. P. Rusconi, E. Scardino, J. Layzer et al., “RNA aptamers as reversible antagonists of coagulation factor IXa,” Nature, vol. 419, no. 6902, pp. 90–94, 2002. View at: Publisher Site | Google Scholar
  17. B. van Steensel, A. Smogorzewska, and T. de Lange, “TRF2 protects human telomeres from end-to-end fusions,” Cell, vol. 92, no. 3, pp. 401–413, 1998. View at: Publisher Site | Google Scholar
  18. C. M. Price, K. A. Boltz, M. F. Chaiken, J. A. Stewart, M. A. Beilstein, and D. E. Shippen, “Evolution of CST function in telomere maintenance,” Cell Cycle, vol. 9, no. 16, pp. 3157–3165, 2010. View at: Publisher Site | Google Scholar
  19. Y. Miyake, M. Nakamura, A. Nabetani et al., “RPA-like mammalian Ctc1-Stn1-Ten1 complex binds to single-stranded DNA and protects telomeres independently of the Pot1 pathway,” Molecular Cell, vol. 36, no. 2, pp. 193–206, 2009. View at: Publisher Site | Google Scholar
  20. A. Prakash, A. Natarajan, L. A. Marky, M. M. Ouellette, and G. E. Borgstahl, “Identification of the DNA-binding domains of human replication protein A that recognize G-quadruplex DNA,” Journal of Nucleic Acids, vol. 2011, Article ID 896947, 14 pages, 2011. View at: Publisher Site | Google Scholar
  21. S. D. Mendonsa and M. T. Bowser, “In vitro evolution of functional DNA using capillary electrophoresis,” Journal of the American Chemical Society, vol. 126, no. 1, pp. 20–21, 2004. View at: Google Scholar
  22. T. P. Roosild, S. Castronovo, and S. Choe, “Structure of anti-FLAG M2 Fab domain and its use in the stabilization of engineered membrane proteins,” Acta Crystallographica Section F, vol. 62, no. 9, pp. 835–839, 2006. View at: Publisher Site | Google Scholar
  23. G. F. Joyce, “Directed evolution of nucleic acid enzymes,” Annual Review of Biochemistry, vol. 73, pp. 791–836, 2004. View at: Publisher Site | Google Scholar
  24. A. Einhauer and A. Jungbauer, “The FLAG peptide, a versatile fusion tag for the purification of recombinant proteins,” Journal of Biochemical and Biophysical Methods, vol. 49, pp. 455–465, 2001. View at: Publisher Site | Google Scholar


Copyright © 2011 Amanda S. Lakamp and Michel M. Ouellette. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

2 Synthesis of DNA Amphiphiles

A DNA amphiphile is based on hydrophilic DNA that contains a covalently connected hydrophobic segment. 19 Usually, the hydrophobic moiety is a polymer or a small molecule. The lipophilic modifications of DNA can be achieved by conjugation at either the 3′- or 5′-terminal, or within the DNA sequence, allowing the construction of complex structures. 21-24

These hydrophobic moieties can be conjugated to DNA, either on a solid support during DNA synthesis or by coupling to already synthesized DNA units in solution. The first successful chemical synthesis of a dinucleotide was achieved in 1955. 25 Stable deoxynucleoside phosphoramidites were introduced as synthons in 1985, opening up the field. 26 Nowadays, solid phase synthesis (SPS) allows generating DNA fragments of up to 200 nucleotides. This technology allows functionalization or introduction of non-natural nucleotides. 27 The fully automated synthesis can be precisely controlled, monitored, and is characterized by a high reproducibility. To broaden the scope of synthesis robots by introducing special solvents, catalysts, extreme reaction conditions or long reaction times, the automated process can be replaced by the syringe synthesis technique or in-flask reactions to realize various modifications of the DNA with hydrophobic units. 20

Coupling of DNA with specific motifs in solution phase has been demonstrated as another highly versatile strategy, which was reviewed by our group before. 19 Solution phase synthesis is used for covalent bond formation between functional groups such as amines 28 or thiols, 29 with groups such as carboxylic acids 30 or maleimides. 31 However, aqueous solution coupling of DNA with hydrophobic molecules often results in low yields due to the solvent incompatibility of starting materials. To overcome this limitation, we reported a conjugation protocol for coupling of hydrophobic molecules to DNA with high efficiency. 32 By complexing DNA with positively charged quaternary ammonium surfactants, we neutralized the charge on the DNA, making it soluble in organic solvent. The organic phase coupling technique expands the number of possibilities to generate amphiphilic DNA hybrids.

One of the most commonly used lipids in DNA amphiphiles is cholesterol. In addition to cholesterol or one of its derivatives, other synthetic single-chain fatty acids, 33 steroid molecules, 34 α-tocopherol, 35 hydrophobic polymers, such as poly(propylene oxide) (PPO), 21 or the π-conjugated system porphyrin 36, 37 have been successfully introduced to DNA (Figure 1). Hence, synthetic protocols to introduce a wide range of hydrophobic moieties into DNA at various positions are available, allowing for the exploration of new functionalities in nanotechnology. 38

Download and print this article for your personal scholarly, research, and educational use.

Buy a single issue of Science for just $15 USD.

Science Translational Medicine

Vol 3, Issue 66
19 January 2011

Article Tools

Please log in to add an alert for this article.

By Charles Preston Neff , Jiehua Zhou , Leila Remling , Jes Kuruvilla , Jane Zhang , Haitang Li , David D. Smith , Piotr Swiderski , John J. Rossi , Ramesh Akkina

Science Translational Medicine 19 Jan 2011 : 66ra6

A dual-function aptamer that targets both a HIV-1 surface protein and a critical messenger RNA can inhibit HIV infection in humanized mice.


Sample preparation for single-molecule measurements

The following method was used to prepare fluorescently labeled RNA and RNA-DNA complexes 27 . Briefly, all steps involving fluorescently labeled oligomers were performed in the dark. Unlabeled DNA oligomers, primers, and plasmids (Supplementary Table 1) were purchased from Eurofins Genomics LLC and Integrated DNA Technologies. Fluorescently labeled oligonucleotides were purchased from Trilink (San Diego, CA), including the Cy3-labeled 5′-end segment of the F. ulcerans ZTP riboswitch (nt 1–33, Supplementary Table 1) and the Cy5-labeled and biotinylated tether DNA (biotin-DNA-Cy5, Supplementary Table 1). Fluorophores were attached to these oligomers via 6-carbon amino linkers at the C5 position of U32 and at the 3′-end of the biotinylated DNA, respectively. ZTP riboswitch RNAs lacking the first 33 nucleotides (Δ1–33) were transcribed and purified from DNA templates PCR amplified from plasmids 25,42 . Mutations to the terminator stem were made via PCR amplification using a reverse primer containing the mutations. Δ1–33 RNAs were phosphorylated with T4 polynucleotide kinase (New England Biolabs), phenol-chloroform extracted, and ethanol precipitated. The Cy3-labeled nt 1–33 RNA and Δ1–33 RNAs were annealed to a DNA splint identical to the cDNA-dT20 used in single-molecule measurements and ligated using T4 RNA ligase 2 (New England Biolabs). Ligated RNAs were purified by 8% denaturing urea-PAGE, recovered via elution in Whatman Elutrap electroelution systems, concentrated, washed with DEPC-treater water, and filtered in 0.22 μm spin filters. Concentrations were calculated by the absorbance of Cy3 at 552 nm (ε552 = 0.15 μM −1 cm −1 ) and the absorbance of the RNAs at 260 nm (ε260 = 1.16 μM −1 cm −1 for the terminator-containing RNAs and ε260 = 0.982 μM −1 cm −1 for the aptamer-only RNA) after correcting for the contribution of Cy3 absorbance at 260 nm (

5% correction). Ligated RNAs were aliquoted and stored at −20 °C in the dark until further use.

Fluorescently labeled RNAs for double PIFE helicase rate measurements (Fig. 5d, e) were constructed in a similar manner, resulting in RNAs identical to the WT sequence and containing two Cy3 labels. For these experiments, three RNAs were ligated, consisting of segments spanning riboswitch nucleotides 1–33, 34–71, and 72–94. The segment 1–33 contained a Cy3 label either at position U12 or position U32. The segment 72-94 contained a Cy3 label at position U84 and also contained the RNA sequence complementary to the biotinylated DNA.

To form annealed heteroduplexes for single-molecule measurements, 20 μl reaction aliquots containing a 1:0.8:10 ratio of Cy3-RNA:biotin-DNA-Cy5:cDNA-dT20 (160 nM Cy3-RNA, 128 nM biotin-DNA-Cy5, and 1.6 μM cDNA-dT20) were prepared in the buffer containing 50 mM HEPES-KOH (pH 7.4) and 150 mM KCl. The reaction aliquots were slowly cooled from 95 °C to 20 °C in a thermocycler (30 s per 5 °C). Annealing success was evaluated on 8% nondenaturing PAGE (in 1X TBE) by comparison to reactions lacking either biotin-DNA-Cy5, cDNA-dT20, or both. Reactions lacking cDNA-dT20 were also saved for single-molecule measurements. Gels were scanned with a Typhoon Trio Variable Mode Imager (GE Healthcare) to identify bands containing fluorescent labels.

For dropoff-PIFE experiments (Fig. 5b, c), a longer cDNA-dT20 (cDNA-speedmer-dT20) was used for annealing reactions in the presence of an 18-nt RNA oligo labeled at the 3´ end with Cy3 (18-speedmer), the WT Cy3-labeled RNA, and biotin-DNA-Cy5. Reactions contained a 1:0.8:9:10 ratio of Cy3-RNA:biotin-DNA-Cy5:cDNA-speedmer-dT20:18-speedmer to saturate cDNA-speedmer-dT20 with 18-speedmer. Annealing success was evaluated via nondenaturing PAGE as above.

For dropoff experiments (Supplementary Fig. 5), 20 μl reaction aliquots containing a 1:1.2:1.2:1.2 ratio of biotin-DNA:cDNA-dT20:Cy3-oligo:Cy5-oligo (1 μM biotin-DNA, 1.2 μM cDNA-dT20, 1.2 μM Cy3-oligo and 1.2 μM Cy5-oligo) were prepared in the buffer containing 50 mM HEPES-KOH (pH 7.4) and 150 mM KCl. The reaction aliquots were slowly cooled from 95 °C to 20 °C in a thermocycler (30 s per 5 °C). Selecting doubly labeled surface-tethered molecules ensured that the molecules being imaged contained all four oligos.

Single-molecule measurements

Vectorial folding (VF) was performed with the following method 23 . Briefly, Rep-X was incubated in the loading buffer for 5 min with heteroduplexes that were immobilized on the imaging surface. The loading buffer contained 50 nM Rep-X, 50 mM HEPES-KOH (pH 7.4), 150 mM KCl and 10 mM MgCl2. Free Rep-X was washed out and unwinding was initiated by adding the unwinding buffer. The unwinding buffer contained 50 mM HEPES (pH 7.4), 150 mM KCl, 10 mM MgCl2, 1 mM ATP, and different concentrations of ZMP. To stabilize the remaining aptamer conformation and distinguish it from the terminator conformation, 1 mM ZMP was supplied into the imaging channel after 1 min of vectorial folding. To observe the heteroduplex unwinding and riboswitch folding in real-time, imaging was started

5 s before the addition of the unwinding buffer. The loading and unwinding buffers used during imaging contained additional 4 mM Trolox, 0.8 % wt vol −1 glucose, 165 U ml −1 glucose oxidase, and 2170 U ml −1 catalase to reduce the photobleaching rate 43 .

For VF multiple turnover experiments with Rep-X (Fig. 2d), 50 nM Rep-X was included in the unwinding buffer and no separate Rep-X loading step was performed. When PcrA-X was used in place of Rep-X, 100 nM of PcrA-X was added in the loading buffer. Other buffer components were kept the same.

To refold RNAs for single-molecule measurements, 20 μl reaction aliquots containing a 1:0.8 ratio of Cy3-RNA:biotin-DNA-Cy5 (160 nM Cy3-RNA and 128 nM biotin-DNA-Cy5) were prepared in the buffer containing 50 mM HEPES-KOH (pH 7.4), 150 mM KCl, 10 mM MgCl2 and different concentrations of ZMP. The reaction aliquots were slowly cooled from 95 °C to 20 °C in a thermocycler (30 s per 5 °C). When refolded at 1 mM ZMP, RNAs were imaged within 5 min after refolding to minimize the loss of ZMP-bound aptamer due to terminator hairpin formation.

Single-molecule data acquisition and analysis

Single-molecule data acquisition and analysis were performed with the following method 23,44 . Briefly, fluorescently labeled samples were immobilized on DT20 or PEG surfaces through a biotin-NeutrAvidin linkage 45,46 . Samples were imaged with a prism-based total internal reflection fluorescence (TIRF) imaging microscope 47 . A 532-nm laser (Coherent Compass 315 M) and a 633-nm laser (Research Electro-Optics) were used for Cy3 and Cy5 excitation, respectively. A water immersion objective (NA 1.2, 60X, Olympus) and an EMCCD camera (Andor Technology IXon 897) were used to collect and record signals with 0.075 s time resolution. The fluorescence emission was filtered by a long pass filter (Semrock BLP02-561R-25) and a notch filter (Chroma ZET635NF) to block the excitation lasers. Spatially separated single molecules were picked by a custom IDL code and their intensities were extracted for further analysis. To plot the trajectory overlay, we synchronized trajectories according to the PIFE peak center. The percentage of aptamer-like fold is calculated as the ratio of the aptamer population (magenta) over the sum of the terminator (green) and aptamer populations.

For traces containing multiple PIFE peaks (Fig. 5d, e), peak centers were determined by Gaussian fitting in Origin (OriginLab, Northampton, MA), and time differences between PIFEs (Δt) were calculated by the difference between the peak centers. Traces appearing to contain too many PIFE peak were excluded from analysis (7 traces excluded from 12–84 dataset, 8 traces from 32–84 dataset, and 9 traces from drop-PIFE dataset). Unwinding rates are reported as the mean ± s.e.m. with the n being the number of individual molecules.

For dropoff-PIFE traces (Fig. 5b, c), the drop in Cy3 intensity was determined by eye, and time was recorded. PIFE peak centers were determined by Gaussian fitting as above. The time difference (Δt) between these two events was calculated for each trace and used to determine a helicase unwining rate (in nt s −1 ) with the known distance of 34 nt separating the two events. Traces containing more than one PIFE peak were excluded from analysis (9 traces excluded). The average rate is reported as the mean ± s.e.m. with n being the number of individual molecules (91 total).

For dropoff experiments (Supplementary Fig. 5), the decrease in FRET value and decrease in Cy3 intensity were determined by eye. The time difference (Δt) between these two events was calculated for each trace, and the average unwinding rate was calculated using all traces for which Δt was less than 2 s for Rep-X (or 10 s for PcrA-X), which includes traces contained within the Gaussian distribution of rates.

Single-round transcription termination assays

PCR templates were amplified from a plasmid containing a 100 nt leader sequence, a λ Pr promoter, 26-nt C-less region, the F. ulcerans ZTP riboswitch sequence, which contains the aptamer domain, expression platform and 51 nucleotides beyond the terminator U8 sequence. After amplification, templates were purified by 2% agarose gel electrophoresis. Mutations to the templates were made by site-directed mutagenesis to the plasmid and confirmed by sequencing.

Transcription reactions were performed by halting transcription by omission of CTP and restarting transcription by adding NTPs and ZMP 25 . Halted transcription complexes contained 80 pmol uL −1 DNA template, 20 mM Tris-HCl, pH 8, 20 mM NaCl, 4 mM MgCl2, 0.1 mM DTT, 0.1 mM EDTA, 4% glycerol, 0.14 mM ApU, 1 μM GTP, 2.5 μM ATP, 2.5 μM UTP,

1 μCi mL −1 [α- 32 P]-ATP, 0.04 U uL −1 Escherichia coli RNA polymerase holoenzyme (Epicenter), and 10 μM of a 26 nt oligonucleotide complementary to the C-less region. Transcriptions were restarted by addition of various concentrations of NTPs and ZMP prior to incubation at 37 °C for the duration of the time course, removing aliquots as needed and quenching reactions by addition of loading dye containing 8 M Urea, 20% sucrose, 0.1% SDS, 0.01% bromophenol blue, 0.01% xylene cyanol and placing them on ice or storing them at −20 °C. Reactions were then separated via 8% denaturing urea-PAGE and analyzed 48,49 . To determine apparent reaction rates for appearance of terminated and readthrough transcription products, time courses were fit by single exponential fits using Origin. For single time point titrations, reactions containing different concentrations of ZMP were incubated for 20 min at 37 °C, and results were fit to a 1:1 binding isotherm to determine T50 values.

Sequential folding in silico was performed by using Mfold 50 with different lengths of ZTP riboswitch transcripts as inputs.

Isothermal titration calorimetry (ITC)

All ITC measurements were performed and analyzed with the following method 51 , using a MicroCal iTC200 (Malvern, Egham, UK). Briefly, RNAs were refolded by heating at 95 °C for 2 min and immediately placed on ice. MgCl2 was added to 10 mM, the sample was brought up to volume, and the RNA was incubated at 37 °C prior to performing titrations at 37 °C in a buffer containing 50 mM Hepes-KOH, pH 7.4, 150 mM KCl, and 10 mM MgCl2. Typically the cell containing 20 μM RNA was titrated with 200 μM ZMP, but higher concentrations were used for weaker binders.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Computational methods

Single-cell data processing

Fastq files from the 10x Genomics libraries with four distinct barcodes were pooled together and processed using the standard Drop-seq pipeline (Drop-seq tools v1.0, McCarroll Lab). Reads were aligned to the hg19-mm10 concatenated reference, and we included the top 50,000 cell barcodes in the raw digital expression matrix as output from Drop-seq tools. For ADT and HTO quantification, we implemented our previously developed tag quantification pipeline [18] as a python script, available at, and run with default parameters (maximum hamming distance of 1).

Demultiplexing with genotyping data using demuxlet

We first generated a VCF file that contained the individual genotype (GT) from the Infinium CoreExome 24 array output, using the PLINK command line tools (version 1.07). This VCF file (which contained genotype information for the 8 PBMC donors as well as HEK293T cells) and the tagged bam file from Drop-seq pipeline were used as inputs for demuxlet [13], with default parameters.

Single-cell RNA data processing

Normalization and downstream analysis of RNA data were performed using the Seurat R package (version 3.0, Satija Lab [29]) which enables the integrated processing of multi-modal (RNA, ADT, HTO) single cell datasets [31, 32]. We collapsed the joint-species RNA expression matrix to only include the top 100 most highly expressed mouse genes (along with all human genes) using the CollapseSpeciesExpressionMatrix function.

We first considered a set of 20,854 barcodes where we detected at least 200 UMI in the transcriptome data. Since the HEK293T and NIH-3T3 cells were not labeled with HTOs, we identified these cells based on their transcriptomes. We performed a low-resolution pre-clustering by performing PCA on the 500 most highly expressed genes, followed by k-medoid clustering on a distance matrix based on the first 2 principal components [33,34,35]. Based on this clustering, we identified 160 NIH-3T3 cells and 2233 HEK293T cells, with the remainder representing PBMCs.

As a separate test of HEK293T identity, we examined the demuxlet genotype for possible HEK293T cells. We observed 225 barcodes classified as HEK by the demuxlet algorithm but whose transcriptomes clustered with PBMCs. These cells expressed tenfold fewer UMI compared to transcriptomically classified HEK293T cells and did not express HEK293T-specific transcripts (i.e., NGFRAP1), both consistent with a PBMC identity. We therefore excluded these barcodes from all further analysis.

Classification of barcodes based on HTO levels

HTO raw counts were normalized using centered log ratio (CLR) transformation, where counts were divided by the geometric mean of an HTO across cells and log-transformed:

Here, xi denotes the count for a specified HTO in cell i, n is the total cell number, and log denotes the natural log. Pairwise analysis of either normalized or raw HTO counts (Fig. 1B) revealed mutually exclusive relationships, though determining the exact cutoffs for positive and negative signals required further analysis. We reasoned that if we could determine a background distribution for each HTO based on “negative” cells, outliers from this distribution would represent positive signals.

To assist in the unsupervised identification of “negative” cells, we performed an initial k-medoids clustering for all cells based on the normalized HTO data. We set k = 9 and observed (as expected) that eight of the clusters were highly enriched for expression of a particular HTO, while the ninth cluster was highly enriched for cells with low expression of all HTO. This represents an initial solution to the demultiplexing problem that suggests likely populations of “positive” and “negative” cells for statistical analysis.

Following clustering, we performed the following procedure independently for each of the eight HTOs. We identified the k-medoids cluster with the highest average HTO expression and excluded these cells. We next fit a negative binomial distribution to the remaining HTO values, after further excluding the highest 0.5% values as potential outliers. We calculated the q = 0.99 quantile of the fitted distribution and thresholded each cell in the dataset based on this HTO-specific value.

We used this procedure to determine an “HTO classification” for each barcode. Barcodes that were positive for only one HTO were classified as singlets. Barcodes that were positive for two or more HTOs were classified as multiplets and assigned sample IDs based on their two most highly expressed HTO. Barcodes that were negative for all eight HTOs were classified as “negative.”

We expect that barcodes classified as “singlets” represent single cells, as we detect only a single HTO. However, they could also represent doublets of a PBMC with a HEK293T or NIH-3T3 cell, as the latter two populations were unlabeled and represent negative controls. Indeed, when we analyzed the “HTO classification” of cells that were transcriptomically annotated as HEK293T or NIH-3T3 cells, we found that 60.1% were annotated as “negative,” while 32.1% were annotated as singlets, in agreement with our expected ratios in our “super-loaded” 10x Genomics experiment. These cells appear in the heatmap in Fig. 1C, but all HEK293T and NIH-3T3 cells were excluded from further analysis.

For 2D visualization of HTO levels (Fig. 1D), we used Euclidean distances calculated from the normalized HTO data as inputs for tSNE. Cells are colored based on their HTO classification as previously described. For visualization and clustering based on transcriptomic data (Fig. 1F), we first performed PCA on the 1000 most highly variable genes (as determined by variance/mean ratio) and used the distance matrix defined by the first 10 principal components as input to tSNE and graph-based clustering in Seurat (Fig. 1E). We annotated the seven clusters based on canonical markers for known hematopoietic populations.

Comparison with demuxlet

Demuxlet classifications were labeled as singlets (SNG), doublets (DBL), or ambiguous (AMB) according to the BEST column in the *.best output file. In Fig. 2e, we plot the posterior probability of a doublet assignment from the PRB.DBL column in the same file.

Calculation of staining index for antibody titrations

To assess the optimal staining efficiency for CITE-seq experiments, we considered ADT levels for cells across a range of antibody concentrations as multiplexed in a titration series. ADT levels were normalized using a CLR transformation of raw counts using an identical approach to the normalization of HTO levels as previously described.

After normalization, we computed a staining index based on standard approaches in flow cytometry, which examine the difference between positive and negative peak medians, divided by the spread (i.e., twice the mean absolute deviation) of the negative peak.

In order to avoid manual classification of positive and negative peaks, we implemented an automated procedure that can scale to multiple antibodies and concentrations. To approximate the negative peak, we leveraged unstained control cells (donor H). To approximate the positive peak, we clustered the ADT data in each titration experiment (donor A through donor G). To perform clustering, we computed a Euclidean distance matrix across cells based on normalized ADT levels and used this as input to the FindClusters function in Seurat with default parameters. We examined the results to identify the cluster with the maximally enriched ADT signal and referred to the distribution of ADT levels within this cluster as the positive peak.

Discriminating low-quality cells from ambient RNA

We performed HTO classification of low-quality barcodes (expressing between 50 and 200 UMI), using the previously determined HTO thresholds. For each barcode, we classified its expression as 1 of our previously determined 7 hematopoietic populations using random forests, as implemented in the ranger package in R [36]. We first trained a classifier on the 13,954 PBMCs, using the 1000 most variable genes as input and their clustering identities as training labels. We then applied this classifier to each of the low-quality barcodes. We note that this classifier is guaranteed to return a result for each barcode.

Capture-SELEX: Selection of DNA Aptamers for Aminoglycoside Antibiotics

Small organic molecules are challenging targets for an aptamer selection using the SELEX technology (SELEX—Systematic Evolution of Ligans by EXponential enrichment). Often they are not suitable for immobilization on solid surfaces, which is a common procedure in known aptamer selection methods. The Capture-SELEX procedure allows the selection of DNA aptamers for solute targets. A special SELEX library was constructed with the aim to immobilize this library on magnetic beads or other surfaces. For this purpose a docking sequence was incorporated into the random region of the library enabling hybridization to a complementary oligo fixed on magnetic beads. Oligonucleotides of the library which exhibit high affinity to the target and a secondary structure fitting to the target are released from the beads for binding to the target during the aptamer selection process. The oligonucleotides of these binding complexes were amplified, purified, and immobilized via the docking sequence to the magnetic beads as the starting point of the following selection round. Based on this Capture-SELEX procedure, the successful DNA aptamer selection for the aminoglycoside antibiotic kanamycin A as a small molecule target is described.

1. Introduction

Aptamers are a very attractive class of targeting molecules being in great demand in many fields of application, like medicine (as diagnostic and therapeutic agents) and pharmaceutics (for drug discovery and validation), as well as environmental or food analytics (as biological recognition elements). Aptamers are by nature nucleic acid molecules, but their functionality is based on their complex three-dimensional structure different from the conventional view on nucleic acids as carrier of genetic information. The complex folding of the short, single stranded DNA or RNA aptamers according to their primary sequence enables them to bind with high affinity and specificity to the given target. The intermolecular interactions between aptamer and target are characterized by a combination of complementarity in shape, stacking interactions between aromatic compounds and the nucleobases of the aptamers, electrostatic interactions between charged groups, and hydrogen bonds [1–3]. Since the first publication of aptamers in 1990 [4, 5], they have been described for a wide variety of different classes of targets from small molecules, like nucleotides, cofactors or amino acids, over peptides, polysaccharides and proteins to complex structures like whole cells, viruses and single cell organisms [6]. The growing number of aptamer publications over the years describing their selection andapplication shows the high interest in this research field. One of the challenges in this area is the optimization of the methodology to get new aptamers with outstanding binding abilities for a certain target. Aptamers are usually generated by an in vitro selection and amplification technology called SELEX—Systematic Evolution of Ligands by Exponential enrichment. [4, 5]. The SELEX process is an iterative process. From a DNA oligonucleotide library comprising a large sequence diversity and structural complexity only those oligonucleotides are selected and enriched during several SELEX rounds which can bind very tightly to the specific molecular target [7, 8]. Basic steps of the SELEX process are the binding reaction between oligonucleotides and target, washing steps to remove unbound oligonucleotides, enzymatic amplification of target-bound oligonucleotides, and purification of the selected oligonucleotide pool to subsequently start the next selection round. The best fitting sequences survive the selection procedure and represent the target-specific aptamer pool as the result of a successful SELEX process. Since the early phase of the application of the SELEX technology, it was often modified to make the selection process more efficient and less time consuming and to select aptamers with particular binding features (affinity and specificity) for different target molecules and for different applications. Variations often concern the binding conditions (buffer, temperature, and time), the efficient separation of target binding and nonbinding oligonucleotides [9], the introduction of additional selection steps, like negative or counter selection steps to remove nonspecifically binding oligonucleotides or to discriminate between closely related target molecules, the stringency of washing steps of the oligonucleotide-target complexes to improve the specificity of the aptamers to be selected, the labeling of the oligonucleotides for quantification, or the elution of the oligonucleotides from the binding complex before amplification [7]. An important aspect of an aptamer selection also is the design of the SELEX library and the use of modifications at the nucleotide level or at the ends of the oligonucleotides. Such modifications can be used to enhance the stability of the oligonucleotides necessary for the selection of RNA aptamers or to introduce new features like functional groups to provide new possibilities for the interaction with target molecules [6, 10, 11]. The numerous SELEX variants demonstrate that there is no standard selection protocol applicable to any kind of target. The selection conditions have to be carefully adapted with regard to the specific target features and the desired application of the aptamers. Medical and analytical applications for instance can have different requirements for the aptamer working conditions like buffer composition or temperature. The probe matrix often poses challenges to the aptamer stability and specificity, which can be influenced during the aptamer selection process by nucleic acid modifications and counterselection steps.

In this paper we describe the development of a SELEX variant called Capture-SELEX. It is partially derived from the FluMag-SELEX procedure published in 2005 [12], which is characterized by the immobilization of the target molecules on magnetic beads and the fluorescence labeling of the oligonucleotides for quantification. The main difference between both is the presentation of the target molecules and the oligonucleotides. The aim was to select DNA aptamers for targets which are not suitable for immobilization on solid surfaces, like small organic molecules, for example, pharmaceuticals. Therefore, a special SELEX library was developed according to Nutiu and Li, 2005 [13] to enable the immobilization of the oligonucleotides instead of the target molecules during the aptamer selection process. This paper also describes the successful application of the Capture-SELEX procedure using a mixture of pharmaceuticals (kanamycin A, sulfacarbamide, sulfamethoxazole, and sotalol hydrochloride) as aptamer selection targets. Pharmaceutical residues are found in surface, ground, and drinking water. They arise mostly from human and animal treatment. Antibiotics are an increasing issue in this context. Their introduction into the environment promotes the proliferation of genes for antibiotic resistance in bacteria. Aptamers with specificity for pharmaceuticals can be used in biosensors and assays for fast and easy detection of these substances [14]. Keeping in mind this requirement we assembled the target mixture mentioned above consisting of three antibiotics and the beta blocker sotalol hydrochloride. Using the Capture-SELEX process, an aptamer pool with specificity for kanamycin A was enriched.

2. Materials and Methods

2.1. Chemicals

All chemicals for preparing buffers and solutions were obtained from Merck (Germany) if not mentioned otherwise. Kanamycin A disulfate salt dihydrate, sulfacarbamide, sulfamethoxazole, and sotalol hydrochloride were purchased from Sigma-Aldrich (Germany). PCR (polymerase chain reaction) components like 10×reaction buffer, 25 mMMgCl2, and HOTFire polymerase were purchased from Solis BioDyne (Estonia). 100 mM stock solutions of dNTPs were from GE Healthcare (Germany).

2.2. Capture-SELEX Library, Capture Oligo and Primers

The Capture-SELEX library BANK-S4 was synthesized by Microsynth (Switzerland) including a PAGE (polyacrylamide gel electrophoresis) purification step. The library consists of a multitude of different oligonucleotides. Each of them contains specific primer binding sites (PBSs) of 18 nt at the 5′ and 3′ ends and between both two different-sized random regions (N) separated by a specific docking sequence of 12 nt: 5′-ATACCAGCTTATTCAATT—N10—TGAGGCTCGATC—N40—AGATAGTAAGTGCAATCT-3′ (Figure 1) [12, 15]. The following primers were used for amplification of the oligonucleotides during the aptamer selection process and were synthesized by (Germany): AP10:5′-ATACCAGCTTATTCAATT-3′, AP60: the modified variant of AP10 with 5′-fluorescein, AP20:5′-AGATTGCACTTACTATCT-3′, and TER-AP20:the modified variant of AP20 with 5′-poly-dA20-HEGL [12]. The capture oligo i-ODN2Sp is characterized by a complementary sequence to the docking sequence of the oligonucleotides in the library: 5′-Bio-GTC-HEGL-GATCGAGCCTCA-3′ (Figure 1). It also contains three additional nucleotides and a hexaethylene glycol spacer (HEGL) at the 5′-end. The capture oligo is modified with 5′-biotin and was also synthesized by Microsynth (Switzerland) including a PAGE purification step.

Specific composition of the Capture-SELEX library BANK-S4 and the capture oligos i-ODN2Sp and i-ODN4Sp. PBS1 and PBS2 represent the primer binding sites of each oligonucleotide in the library.
2.3. Coupling of Capture Oligos to Streptavidin-Coated Magnetic Beads

Superparamagnetic Dynabeads M-270 Streptavidin (diameter 2.8 μm) were purchased from Invitrogen/Life Technologies (USA). An appropriate amount of these streptavidin-coated magnetic beads were transferred from stock solution to a sample tube and washed three times with 500 μL B&W buffer (binding and washing buffer, 10 mM Tris-HCl pH7.5, 1 mM EDTA, and 2 M NaCl). The beads were separated by placing the tube in a magnet stand. After washing the beads they were resuspended in B&W buffer to a concentration of 2×10 9 beads/mL, and an equal volume of the biotinylated capture oligo i-ODN2Sp was added (600 pmol biotinylated oligo/1×10 8 beads). This immobilization mixture was then incubated at room temperature for 1 h with gentle rotation using an overhead shaker (Intelli-Mixer RM-2, neoLab, Germany). The beads were separated and washed three times with 500 μL B&W buffer and afterwards three times with 500 μL selection buffer (100 mM NaCl, 20 mM Tris-HCl pH7.6, 2 mM MgCl2, 5 mM KCl, and 1 mM CaCl2). After resuspension in selection buffer to a final concentration of 1×10 9 beads/mL, the streptavidin-coated magnetic beads, which are now modified with the biotinylated capture oligo i-ODN2Sp, are ready for use in the Capture-SELEX process. The bead concentration can be determined more exactly by microscopic counting (microscope Olympus BX60, Olympus Europa Holding GmbH, Germany) using a Neubauer-improved counting chamber.

2.4. Target Solution

Four pharmaceuticals (kanamycin A disulfate salt dihydrate, sulfacarbamide, sulfamethoxazole, and sotalol hydrochloride, see Table 1) were used as aptamer selection target mixture. Individual stock solutions of the pharmaceuticals were prepared, diluted in selection buffer, and mixed to the final concentration of 1 mM for each substance. This mixture was sterile filtered using a syringe filter with the pore size of 0.22 μm (VWR, Germany), aliquoted and stored at −18°C.

2.5. Capture-SELEX Process

Each Capture-SELEX round starts with the thermal equilibration of the oligonucleotide pool in selection buffer. 2-3 nmol of the library oligonucleotides in the first round and, in each of the following rounds, the total quantity of selected oligonucleotides from the previous round, respectively, were heated to 90°C for 8 min, immediately cooled, and kept at 4°C for 10 min followed by a short incubation at room temperature. In parallel, an aliquot of the streptavidin-coated magnetic beads modified with the capture oligos (1×10 9 beads in the first round and 1×10 8 beads in each of the following rounds) was washed three times with 500 μL selection buffer. The beads were resuspended in 300 μL of the pretreated oligonucleotide pool and incubated overnight at 21°C with mild shaking. This step serves for the immobilization of the oligonucleotides from the pool on the magnetic beads by hybridization between the docking sequence within the oligonucleotides and the capture oligos on the beads. Unbound oligonucleotides were removed by washing the beads nine times with 500 μL selection buffer. In the temperature step all of the remaining unhybridized oligonucleotides or weakened DNA duplex structures were eliminated by incubation of the DNA-bead-complexes in 500 μL selection buffer at 28°C for 15 min with mild shaking and subsequently washing seven times with 500 μL selection buffer. The following incubation of the DNA-bead-complexes in 300 μL selection buffer at 21°C for 45 min with mild shaking serves as a control for background elution of hybridized oligonucleotides from the beads caused by the incubation procedure. After washing again seven times with 500 μL selection buffer, the DNA-bead-complexes were incubated at 21°C for 45 min with mild shaking in 300 μL of the target mixture (see Section 2.4). Oligonucleotides with affinity to the selection target are able to fold into a specific three-dimensional structure for binding to the target in solution and therefore are released from the DNA-bead-complexes during this target binding step. They can be collected in the supernatant by magnetic separation from the beads.

These selected oligonucleotides bound to the target were directly amplified in 15 parallel PCR reactions. Each contained 1 μM of primers AP60 and TER-AP20, 0.2 mM dNTPs each, 1.9 mM MgCl2, and 5 U HOTFire polymerase in PCR reaction buffer (80 mM Tris-HCl pH9.5, 20 mM (NH4)2SO4, and 0.02% Tween20) in a volume of 100 μL. Amplification conditions were 15 min at 95°C and 30 cycles of 1 min at 95°C, 1 min at 51°C, 1 min at 72°C, and a final step of 10 min at 72°C after the last cycle. As a result, dsDNA products were obtained with a fluorescein modification at the 5′-end of the relevant sense strand and a poly-dA20 extension at the 5′-end of the antisense strand. Electrophoresis on 2.5% agarose gel was used to monitor the successful amplification and the correct size of the amplified DNA. An influence of the pharmaceuticals on the PCR reactions was checked and could not be observed.

All PCR products were pooled, precipitated with ethanol in presence of linear polyacrylamide [19], and resuspended in 100 μL TE buffer (10 mM Tris-HCl pH7.4, 1 mM EDTA). The two strands of the dsDNA differ in length due to the poly-dA20 extension of the antisense strand [20]. This was utilized for the separation of the two DNA strands in a preparative denaturing PAGE with an 8% polyacrylamide gel containing 7 mM urea and 20% formamide in TBE buffer (90 mM Tris-HCl, 90 mM boric acid, and 2 mM EDTA). The fluorescein-labeled sense strands could be identified in the gel by using a UV transilluminator. The corresponding DNA bands were cut out, and the single stranded DNA (ssDNA) was eluted from the gel with 2 mM EDTA, 300 mM sodium acetate, and pH7.8 at 80°C for 150 min with mild shaking. After removing of the gel residues by filtering through silanized glass wool, the eluted ssDNA was precipitated with ethanol in presence of linear polyacrylamide [19] and resuspended in selection buffer. A new pool of selected and fluorescein-labeled oligonucleotides was now ready for the next round of the Capture-SELEX process starting with the immobilization of the oligonucleotide pool on magnetic beads.

The fluorescein label attached to the oligonucleotides from round two onwards enables the quantification of the oligonucleotides present in the SELEX fractions like washing steps, temperature step, background elution, and target binding. This is important in order to assess the selection progress over several SELEX rounds. By this way enrichment of specific target-binding oligonucleotides was monitored. In total, 13 rounds of this Capture-SELEX procedure were performed, and a pharmaceutical-specific aptamer pool was selected.

The selected aptamer pool from round 13 of the Capture-SELEX was amplified with the unmodified primers AP10 and AP20 and subsequently cloned into the vector pCR2.1-TOPO (TOPO TA Cloning Kit from Invitrogen/Life Technologies, USA). The resulting recombinant vectors were transformed into chemically competent Escherichia coli TOP10 cells (also provided by the TOPO TA Cloning Kit). Several positive transformants could be analyzed by colony PCR using a combination of a vector-specific primer (M13 forward primer or M13 reverse primer) and an aptamer-specific primer (e.g., primer AP10). This method enables a fast screening for correct plasmid inserts directly from E. coli colonies. The plasmid DNA of 96 clones was isolated using the QIAprep 96 Turbo Miniprep Kit from QIAgen (Germany), and the inserted aptamer DNA of each clone was sequenced (Microsynth, Switzerland). The obtained sequences were analyzed and aligned by using the web-based tool ClustalW provided by the EBI web server ( [21–23].

2.6. Specificity and Affinity Tests of Aptamers

Individual aptamer candidates chosen for further characterizations were synthesized with a 5′-fluorescein label by Microsynth (Switzerland) including a PAGE purification step. Comparative binding tests with individual aptamers were performed according to the Capture-SELEX conditions using the target mixture (see Section 2.4) as well as the individual pharmaceutical solutions. Selection buffer without any target substance was used as a background control.

Several binding tests were performed in parallel. A fresh aliquot of an appropriate amount of streptavidin-coated magnetic beads modified with capture oligos (5× 10 7 beads for each binding test) was first washed three times with 500 μL selection buffer. 50pmol aptamer per 5× 10 7 beads was suspended in 300 μL selection buffer and pretreated by heating it to 90°C for 8 min, immediately cooling to 4°C for 10 min, and keeping it at room temperature for 5-6 min before adding it to the washed beads. Incubation was performed overnight at 21°C and with mild shaking for immobilization of the aptamers to magnetic beads. On average, in each test, an amount of

pmol of the aptamers (determined as the difference between the initially employed ssDNA and the ssDNA in the supernatant after the overnight immobilization step) was immobilized on (

)×10 7 streptavidin-coated magnetic beads modified with capture oligos. The following steps up to the background elution were identical to those described in the Capture-SELEX section. After background elution the beads were washed six times with 500 μL selection buffer and evenly distributed on reaction tubes according to the number of planed binding tests. The DNA-bead-complexes were then resuspended in 300 μL pharmaceutical mixture (final concentration for each pharmaceutical 1 mM), in 300 μL individual pharmaceutical solutions (final concentration 1 mM) for specificity tests, in 300 μL differently concentrated individual pharmaceutical solutions for affinity tests (e.g., concentration series of Kanamycin A in the range of 0–1.5 mM), or only in 300 μL selection buffer as a control. Incubation was carried out at 21°C for 45 min with mild shaking. All further steps were identical to those described in the Capture-SELEX section.

The amount of aptamer released during the target binding step from the beads was determined by fluorescence detection and calculation using a calibration curve. Fluorescence conditions were as described in Section 2.7. Solutions were centrifuged prior to pipetting them into the 96 microwell plate in order to eliminate most of the remaining beads. Aptamers used for the determination of the calibration curves were subjected to the same thermal pretreatment as the aptamers used for the specificity and affinity tests.

2.7. Fluorescence Detection

All fluorescence measurements of fluorescein-labeled DNA were performed on a Wallac 1420 Victor 2 V Multilabel Counter (PerkinElmer, Germany) with excitation at 485 nm and emission at 535 nm (prompt fluorometry, time 1 s, CW-lamp energy 22500). The readings were performed in black 96microwellplates from NUNC/Thermo Fisher Scientific (Germany) with a sample volume of 100 μL/well. A calibration curve in the range of 0.4–40 pmol/mL of fluorescein-labeled ssDNA prepared from the Capture-SELEX library (BANK-S4) was used to calculate the DNA concentration in samples. In the case of purchased fluorescein-labeled aptamers, a calibration curve for each of them had to be determined. No influence of the pharmaceuticals on the fluorescence readings was detected.

3. Results and Discussion

3.1. Design of the Capture-SELEX Library and the Capture Oligos

Starting point of a selection process of target-specific aptamers is the SELEX library, comprising a huge amount of different oligonucleotides (approximately 10 15 unique sequences), which are able to fold into distinct three-dimensional structures. This functionality is achieved through a randomized region of the oligonucleotides. A typical SELEX library consists of oligonucleotides with the same length and a central, randomized region of 20–60 nucleotides, flanked by specific sequences at the 5′- and 3′-ends. The latter serves as primer binding sites for amplification of the oligonucleotides by PCR during the aptamer selection process. This fully random library can be transformed into a partially randomized library for giving it new features, for example, by introducing defined sequences into the randomized region. Here we describe the use of such an additional, defined sequence region as docking sequence to construct a Capture-SELEX library. In a typical SELEX experiment, the target molecules are usually immobilized on a solid surface to enable an efficient separation of target-binding and nonbinding oligonucleotides in each SELEX round. This is a very crucial step for a successful aptamer selection. Conventional methods are affinity chromatography with target immobilization on different column material or the use of magnetic beads as immobilization matrix. But this methodological approach is not applicable to all potential aptamer targets, especially to very small, organic target molecules with only few or no suitable functional groups. Additional functionalization of such molecules often affects their structural features. However, retaining the native biomolecule structures of the targets is very important for the binding features of the aptamers to be selected. On the other hand, size-dependent separation techniques like filtration or centrifugation are often not practicable for small, organic target molecules. The Capture-SELEX library provides an alternative approach by immobilization of the oligonucleotides to a solid matrix instead of the targets. This is done by hybridization between the docking sequence within the oligonucleotides and a complementary capture oligo, which is coupled to the solid matrix by affinity binding or covalent binding. The selection concept is based on release of those oligonucleotides from the matrix which show an affinity to the selection target and therefore undergo a specific conformational change for binding to the target in solution. One possible risk of this concept is that not all oligonucleotides of a SELEX library are productive. Those of them which can bind to the target but do not undergo a conformational change including the docking sequence are not released and therefore are not selected, especially during the first SELEX round.

The design of the Capture-SELEX library called BANK-S4 (see Figure 1) was derived from designs described in Nutiu and Li, 2005 [13]. They firstly described an aptamer selection procedure based on the structure switching idea, with the aim to immediately transform the selected aptamers into fluorescence signaling reporter molecules for the detection of the binding complexes.

BANK-S4 is characterized by an internally unstructured 12-nucleotide docking sequence flanked by two random regions in an asymmetric arrangement. One of them was constructed relatively large with 40 nt for more structural complexity of the library. The other consists of only 10 nt. A short randomized region (20–25 nt) of a SELEX library seems to be sufficient for a successful aptamer selection, because post-SELEX optimizations like truncations often reveal a relatively short minimal functional sequence of many aptamers. However, longer random regions (at least 60–70 nt) give the libraries a higher structural complexity, for example, including high-order junctions [24] and may provide better opportunities for an interaction between the DNA molecules and the targets over an extended domain of both binding partners. The design of the docking sequence (length and nucleotide composition) has to be aimed to achieve the balance between duplex formation with the capture oligo for a stable immobilization of the library and target-dependent release of oligonucleotides from the duplex structure to form specific oligonucleotide-target complexes. The primer binding sites at the 5′- and 3′-end of the BANK-S4 oligonucleotides were derived from the FluMag-SELEX library described in Stoltenburg et al. 2005 [12].

The second part of the Capture-SELEX library concept is the capture oligo (Figure 1), which consists of 12nucleotides complementary to the docking sequence and is biotinylated for the convenient coupling to a streptavidin-modified matrix, for example, magnetic beads. Hybridization between the docking sequence and the matrix-coupled capture oligo permits the immobilization of the Capture-SELEX library. The two kinds of capture oligo design concerning the biotinylation site, at the 5′ or 3′-end of the oligos (i-ODN2Sp or i-ODN4Sp, Figure 1), implicate two alignments of the oligonucleotides contained in the SELEX library due to the asymmetric arrangement of the random sequences. Theoretically, two alignments are possible: when hybridizing to i-ODN2Sp, the longer randomized region (N40) is directed towards the surface of the magnetic bead, whereas when hybridizing to i-ODN4Sp, the shorter randomized region (N10) is directed towards the bead. The capture oligo furthermore contains a hexaethylene glycol spacer (HEGL) between the nucleotides and the biotinylation site. An alternative poly-dA9 spacer has proved to be less suitable for this application, because of an unwanted co-selection of T-rich oligonucleotides from the library, which are able to hybridize with the capture oligo over an extended domain including the poly-dA9 spacer.

The following sections describe the Capture-SELEX procedure in more detail and its successful application for selection of DNA aptamers for the aminoglycoside antibiotic kanamycin A.

3.2. General Procedure of the Capture-SELEX

The Capture-SELEX process using BANK-S4 as SELEX library was established and optimized as a variant of the SELEX technology for the selection of target-specific aptamers. It is characterized by capturing the oligonucleotides (SELEX library or selected oligonucleotide pool) on magnetic beads. By this way, the target molecules do not have to be immobilized and can be used in dissolved form. The magnetic separation technology offers a convenient handling and an efficient separation of bead bound components from other components of the solution [12]. The Capture-SELEX strategy is shown in Figure 2. To initiate a selection round of the Capture-SELEX process, the oligonucleotides of the library in the first round and of the selected pool in subsequent rounds, respectively, were immobilized on streptavidin-coated magnetic beads. These beads are additionally modified by coupling of the biotinylated capture oligos (Figure 1). Hybridization between both DNA partners during an overnight incubation results in the immobilization of the SELEX library or the selected oligonucleotide pool on the beads. Afterwards, the DNA-bead-complexes were incubated twice in selection buffer to reduce the unspecific release of oligonucleotides from the complexes. At first, we introduced the temperature step at an elevated temperature to eliminate unhybridized oligonucleotides or weakened DNA duplex structures. The next incubation step, the background elution step, is characterized by the same incubation conditions (time and temperature) as used for the following target binding step. We used this background elution in selection buffer to determine the amount of oligonucleotides released from the complexes caused only by the incubation procedure. Moreover, this step can easily be combined with a negative selection step by addition of appropriate substances to the selection buffer. Negative selection steps are often useful and recommended for an aptamer selection to avoid not only the enrichment of oligonucleotides that bind nonspecifically, but also to direct the selection of aptamers to a specific epitope of the target, for example, or to distinguish between closely related target molecules. After the immobilization and the different incubation steps we had to wash the DNA-bead-complexes extensively to widely remove unbound oligonucleotides. In the next target binding step the DNA-bead-complexes were exposed to the target solution. The target concentration can typically be in the range of 0.1 mM–1 mM. There is the possibility to reduce this concentration during the SELEX process for more stringent selection conditions, which can affect the affinity of the aptamers to be selected. There is also the possibility to use a target mixture (2 or more different target molecules) for a parallel selection of aptamers in one SELEX process. Oligonucleotides with affinity to the target can fold into a specific three-dimensional structure to bind to the target. These oligonucleotides are able to form a stable binding complex and are therefore released from the bead-bound state into solution during the target binding step. The remaining DNA-bead-complexes are separated, and the target-eluted DNA is directly transferred to the following amplification and purification steps of the selected oligonucleotides of one SELEX round. These steps are the same as described for the FluMag-SELEX process [12]. After the first selection round all oligonucleotides are labeled with fluorescein during amplification by PCR using a 5′-modified primer (sense primer). This enables a quantification of DNA by direct fluorescence detection and thereby the monitoring of the enrichment of target-binding oligonucleotides. We used the ratio of the fraction of oligonucleotides eluted by the target molecules (target binding step) compared to that eluted by selection buffer (background elution step) to obtain information about the aptamer selection progress in each SELEX round. A second modified primer (antisense primer) is used for PCR to introduce a nucleotide extension to the antisense strand, which permits the size-dependent separation of both strands by denaturing PAGE followed by the purification of the relevant sense strand. The resulting new pool of oligonucleotides is used for the next SELEX round starting with their immobilization on streptavidin-coated magnetic beads. Many of such Capture-SELEX rounds (typically 10–15) are necessary to enrich a target-binding oligonucleotide pool (aptamer pool). Each SELEX process ends with cloning of the PCR products of the last round to get individual aptamers. We usually choose up to 100 aptamer clones for their characterization by sequencing and sequence analysis. Binding tests with individual aptamers are then performed to screen for the best binding aptamer candidates, which are chosen for further characterization in order to get information about affinity, specificity, and minimal binding domain.

3.3. Selection of Aptamers for Pharmaceuticals Using the Capture-SELEX Process

In total, 13 rounds of Capture-SELEX were performed as described above (see also Section 2.5). A constant mixture of four pharmaceuticals was used as target in a concentration of 1 mM in selection buffer each (see Section 2.4.). A mixture of pharmaceuticals was chosen in order to enhance the probability to find binding sequences to at least one of the target substances as there is theoretically no hindrance of the development of aptamers to different targets during a multi-target SELEX procedure [25]. The background elution step was carried out with selection buffer without additional substances and therefore without a combination with a negative selection step. Concerning the question which one of the both possible capture oligos i-ODN2Sp and i-ODN4Sp is to prefer, the hybridization efficiencies were tested, determined as the ratio of immobilized ssDNA to the number of used beads. As the hybridization efficiencies were very similar in both assemblies, the use of i-ODN2Sp as capture oligo was decided.

In the first round, about 2 nmol of ssDNA of the SELEX library BANK-S4 was added to 7.9×10 8 streptavidin-coated magnetic beads modified with capture oligos. An amount of 600 pmol (determined as the difference between the initially employed ssDNA and the ssDNA in the supernatant after the overnight immobilization step) was immobilized on these beads. In the following rounds, on average 45 pmol (30–70 pmol) of ssDNA from the preceding SELEX round was immobilized on 1× 10 8 modified beads. It was determined that an amount of—on average—37 pmol (10–75 pmol) of this immobilized ssDNA remains on the beads after all washing steps as well as the temperature and background elution steps and is available for the target binding step. The concentration of the target molecules was 1 mM in 300 μL each. Therefore the concentration of each target present in the sample during the target binding step was about 8,000 times higher than that of the immobilized ssDNA. Therefore the probability of oligonucleotides, once released from the capture oligos on the beads by the presence of a binding partner, to return to the capture oligos, is negligible. As can be seen in Figure 3, from round 10 on, an increase of the amount of eluted ssDNA in the target binding step hinted at the enrichment of aptamers in the process. The total amount of target eluted ssDNA increased from approximately 1 pmol to more than 8 pmol in round 12. In round 13, it decreased to 7.4 pmol and indicated that the end of the SELEX procedure was reached. The ratio of the amounts of oligonucleotides eluted in the target binding step compared to the background elution step is another parameter that can be considered for the assessment of the progress in Capture-SELEX. At first, when there is only a very small amount of specifically binding oligos in the solution, this ratio should equal unity as the conditions for target binding and background elution are the same. As a successful selection of aptamers proceeds and the amount of binders grows, the value will increase. This can be concluded from Figure 3, where the ratio of the amounts of oligonucleotides eluted in the target binding step compared to the background elution step increased from

Enrichment of fluorescein-labeled aptamers during 13 rounds of Capture-SELEX for the target mixture consisting of kanamycin A disulfate salt dihydrate, sulfacarbamide, sulfamethoxazole, and sotalol hydrochloride, each 1 mM in selection buffer. Amounts of eluted single stranded DNA (ssDNA) in the target binding step (dark red) and in the background elution step (blue) are shown. As the starting oligonucleotide library initially was not fluorescein labeled, there are no bars for round 1.

After cloning and transformation of the selected aptamer pool, 99 E. coli clones were further examined. Of these, 96 positive transformants were determined, and the contained plasmid DNA was prepared for sequencing the inserted aptamer DNA of each clone. The 79 sequences that could be unambiguously identified [26] were grouped according to their sequence similarity into 10 groups and 9 orphans (Figure 4). The first group is the one with the largest number of members. It consists of 17 sequences that are 97 nucleotides long (one deletion occurred during the SELEX process inside the docking sequence). These 17 sequences differ from one another only in one, at the most two, base exchanges marked in yellow. In addition, in group 1 there are two sequences that are 126 nucleotides long. They obviously derived from clone #3_7 (representative of 4 identical sequences) of this group by sequence doubling of the 5′ primer binding site and the additional insert of 11 nucleotides.

Groups and orphans of oligonucleotides (5′ → 3′) obtained after cloning and sequencing of the single stranded DNA that was selected by the target mixture (kanamycin A disulfate salt dihydrate, sulfacarbamide, sulfamethoxazole, and sotalol hydrochloride, each 1 mM in selection buffer) during 13 Capture-SELEX rounds. The representative of each group is printed in red. Each group could be divided into subgroups of totally identical sequences. The number of sequences in each subgroup is given (qty), and the representative is printed in black. Green: binding site for primer AP60 (18mer), magenta: docking sequence, originally: TGAGGCTCGATC, blue: binding site for primer AP20 (18mer), yellow marking: sites of base exchange or deletion, and underlined: possible G-quadruplex area.

Group 2 with its 15 members (12 identical sequences and 3 sequences with single base exchanges) is much more homogeneous than group 1 as is group 3 with 10 identical clones. Additionally, there is a 124 nucleotide sequence in group 3 that again emerged from the parent sequence (clone #13_83 as representative) by doubling of the 5′ primer binding site and an additional insert of eight bases. Groups 4 (6 members), 5 (5 members), 6 (4 members), 7 and 8 (3 members each), and 9 and 10 (2 members each) are more or less homogeneous. However, no consensus sequence could be found between groups when using the web-based tool ClustalW [23].

In the sequences of group 1 an accumulation of four stretches of two to three guanine residues, respectively, separated by one to two other bases is noticeable (underlined in Figure 4). This gives the possibility for the formation of a G-quadruplex structure during the three-dimensional folding of the aptamers.

One interesting feature of the derived sequences is the variations in the original docking sequence (TGAGGCTCGATC, printed in magenta) that are marked yellow in Figure 4. In most of the groups, changes of the original docking sequence can be found, like base exchanges, or deletions. Especially noticeable in the docking sequence of group 10, five nucleotides are lost compared to the original docking sequence.

Due to the design of the Capture-SELEX process it is necessary for potential binders to undergo conformational changes when switching from the duplex structure with the capture oligo to the specific target-binding structure. Therefore, there is a selection pressure towards docking sequences that are less stably attached to the capture oligos. On the other hand, the docking sequences have to be specific enough as hybridization has to be stable enough to survive the elution steps precedent to the target binding step. The balancing of these two demands could be observed in our selection as an increase in eluted ssDNA during the temperature elution step starting in round 3 from

17 pmol in round 6. This amount remained high (

12 pmol) until it decreased again to

7 pmol in rounds 12 and 13 (data not shown). Furthermore, it is not astonishing that the base exchanges in the docking sequences lead in 11 cases to G·T and in 2 cases to C·A mismatches, as these are among the most stable mismatches in DNA [28, 29], but of course, they are less strong than perfect matches.

One representative (Figure 4, printed in red) of each of the ten groups was chosen for further examination. Those ten selected aptamers were synthesized, fluorescein tagged (Microsynth, Switzerland), and were subsequently characterized by specificity and some of them by affinity tests.

3.4. Specificity and Affinity of Selected Aptamers for Kanamycin A

Firstly, the binding ability of the representatives of the ten aptamer groups to the target mixture was checked by binding tests comparable to the conditions in the selection rounds (SELEX conditions, see materials and methods section). Of the ten sequences tested, only six exhibited binding to the target mixture that exceeded the range of the negative control (as a threshold, elution of 1 pmol ssDNA was chosen to define a sequence as a binding sequence). These six sequences were the representatives of groups 1 (#3_7), 4 (#4_30), 6 (#13_82), 7 (#3_18), 8 (#11_76), and 9 (#3_19). As four possible target substances were used in parallel, it was particularly necessary to determine in specificity tests the target(s) to which the aptamers are binding. Therefore each of the four target substances was examined. In parallel, selection buffer without any target substance was tested as a negative control (Figure 5). It could be shown that most of the binding sequences were eluted only by kanamycin A (and the mixture of all four pharmaceuticals), whereas sequence #13_82 also showed weak binding to sulfacarbamide and sulfamethoxazole. Of the four pharmaceuticals, Kanamycin A is the most hydrophilic and possesses the most amino groups (see Table 1) which favor electrostatic interaction with the negatively charged phosphate backbone of the aptamer DNA and the formation of hydrogen bonds between aptamer and target. Stacking interaction of the aromatic rings of the other three pharmaceuticals with the nucleobases seems to be of minor importance here.

Specificity tests on the representative of each of the groups of DNA oligonucleotides derived from Capture-SELEX against the mixture of pharmaceuticals. Conditions were chosen comparably to the steps in the SELEX procedure. The four pharmaceuticals were tested individually at a concentration of 1 mM in selection buffer as well as their mixture and selection buffer alone as a negative control. The amount of eluted ssDNA was normalized to 5×10 7 beads. Experiments were performed in duplicate or triplicate in the case of binding sequences (aptamers), the representatives of groups 1, 4, 6, 7, 8, and 9.

Besides from the binding sequences, there are four groups (groups 2, 3, 5, and 10) of oligonucleotides that do not bind to one of the four targets or the mixture. Two of them even are strong groups with many members (groups 2 and 3). The reason for this can certainly be found in the design of the Capture-SELEX. Although there are the temperature and background elution steps and a multitude of washing steps, it is still possible that some nonspecific elution during the target binding step occurs. However, the specific sequences should finally dominate the selection process. Therefore, it is especially important in the Capture-SELEX to thoroughly perform the specificity tests in order to eliminate nonbinders from the received set of sequences.

In Figure 5 standard deviations (error bars) are sometimes quite large as the experiments are very complex in its many steps, and it is therefore not easy to receive constant absolute amounts of eluted DNA. Even after the preceding centrifugation step, beads still remaining in the solution that is to be detected with fluorescence reading interfere by reflection of the excitation beam. Also the slightly different amount of beads used in each preparation, even after normalizing the amount of beads to a fixed value of 5×10 7 beads, adds to the error. Nevertheless, this design of binding assay was chosen as it was closest to the SELEX conditions and therefore the binding reaction of the resulting aptamers to the target was expected to be unaffected by any changes in assay configuration. Furthermore, the pharmaceuticals could still be employed in solution—preferable for a future application of the developed aptamers in the detection of those pharmaceuticals in environmental samples—and without the need for immobilization and the possible risk of losing binding properties. However, in future examinations, different binding assay designs have to be tested.

In order to determine the affinity of some selected aptamers to kanamycin A free in solution, affinity tests were performed similar to the Capture-SELEX procedure as described in the materials and methods section. Upon target binding of the selected aptamer to kanamycin A in different concentrations, certain amounts of fluorescein-labeled aptamer are released from the streptavidin-coated magnetic beads modified with capture oligos. These amounts are then quantified by fluorescence detection (Figure 6), and the resulting data is fitted by the model of one-site direct binding using a rectangular hyperbola also known as binding isotherm or saturation binding curve (OriginLab Corproration, OriginPro 8G SR2). The equation used for this model describes the equilibrium binding of a ligand to a receptor as a function of increasing ligand concentration, and

is the equilibrium dissociation constant. When the target concentration equals , half of the binding sites (aptamers) are occupied at equilibrium. The derived dissociation constants, , for aptamer representatives of the groups 1, 6, 8, and 9 are in the low micromolar range ((#3_7 (gr. 1): 24 μM, #13_82 (gr. 6): 5.1 μM, #11_76 (gr. 8): 9.6 μM, and #3_19 (gr. 9): 3.9 μM).

Affinity tests on the representatives of aptamer groups evolved from Capture-SELEX. Affinity tests were performed similar to the Capture-SELEX procedure, and ssDNA was eluted by different concentrations of kanamycin A in selection buffer. Data was fitted by the model of one site direct binding using a rectangular hyperbola for the saturation curve (OriginLab Corporation, OriginPro 8G SR2).

Using a different SELEX method (affinity chromatography with kanamycin-immobilized sepharose beads), Song et al., 2011 only recently selected DNA aptamers for kanamycin [27]. They started with a synthetic ssDNA library of 90 nt in length with a random sequence of 40 nt flanked by two primer binding sites. Kanamycin was immobilized on CNBr-activated sepharose beads. These beads were used for the preparation of an affinity column for kanamycin binding oligonucleotides. After nine rounds of selection, they obtained 16 individual sequences out of 48 clones. A T-GG-A motif with a stem-loop folding pattern was present in 11 of the sequences and was assumed to be the binding region of the aptamers. The affinity of aptamer Kana2 to kanamycin immobilized at magnetic beads was tested, and a value of 85.6 nM is stated.

In order to find possible sequence similarities, we checked the sequences of the representatives of the ten groups of oligonucleotides that we found in our Capture-SELEX against all those 13 sequences shown in the publication of Song et al. [27], using the web-based tool ClustalW [23]. We compared the regions that were given beforehand (the primer binding sequences and in our case the docking sequence, resp.) and the random regions. The highest similarity—especially in the random regions—showed one pair of sequences (our sequence #3_18 (gr. 7) and their Kana18). However, no consensus was found between the sequence that was determined as binding region by Song et al. and our sequences (Figure 7). Therefore, a different binding motif has to be assumed in our case. The experiments in order to find the binding region within our selected aptamers will follow.

ClustalW [23] comparison of the sequences for kanamycin binding aptamers derived from different types of SELEX. #3_18 (gr. 7): aptamer developed by Capture-SELEX as described here. Kana18: aptamer developed by Song et al., 2011 [27] using affinity chromatography with kanamycin-immobilized sepharose beads. Fixed regions are marked with green (5′-primer binding regions), blue (3′-primer binding regions), and magenta (docking sequence). Random regions are printed in black. Red: binding region as assumed by Song et al., 2011 [27] and underlined: stem-loop forming sequence as given in [27].
3.5. Aptamer Assay Development Based on Capture-SELEX Principle

Analytical applications of aptamers often require some post-SELEX modifications. Typical modifications are the attachment of functional groups, for example, for immobilization of the aptamers on sensor surfaces, the attachment of reporter molecules for monitoring the aptamer-target binding, sequence alterations, or a combination of them. Different strategies have been developed to transduce aptamer-target interactions into a recordable signal [30–33]. However, the risks of such modifications are the possible loss of the affinity and specificity of the aptamers for their targets. To circumvent this problem, a few alternative methods for a direct selection of fluorescence-signaling aptamers have been described [13, 34–36]. These in vitro selection processes are characterized by the inclusion of signaling components like reporter molecules for fluorescence-based detection, special sequences, or additional oligonucleotides, so that large post-SELEX modifications can be avoided.

The Capture-SELEX principle, described in this paper, also offers the possibility to develop detection assays based on the duplex formation between the docking sequence of the selected aptamer and the capture oligo. The target-dependent conformational change leads to the release of the aptamer from the duplex structure with the capture oligo. This switch from the hybridized to the dehybridized stage of the aptamer can be used to generate a recordable signal. Figure 8 shows potential strategies for an assay development based on fluorometry using well-known methods like changes in fluorescence intensity, fluorescence resonance energy transfer (FRET), molecular beacon, or fluorescence polarization (FP). The strategies shown in Figures 8(a) and 8(b) enable the direct detection of aptamer-target complexes. The aptamer is labeled with a fluorophore and therefore the binding complex formed after addition of the target can be measured directly by fluorescence detection. The strategies in Figures 8(c) and 8(d) are characterized by indirect detections of aptamer-target complexes. This means the aptamer is unlabeled, whereas the capture oligo is labeled with one or two fluorophores. After addition of the target the aptamer is released from the duplex structure with the capture oligo causing changes in the fluorescence behavior of the capture oligo, which can be detected instead of the aptamer-target complex. Figure 8(a) represents the same strategy as used during the Capture-SELEX process. We have applied this assay strategy for screening for the best binding aptamer candidates after cloning and sequencing of the selected aptamer pool and also for specificity and affinity tests (see Section 3.4.). All other assay strategies are based on different modifications mostly of the capture oligos (attaching of fluorescence reporter molecules and sequence alterations). Figures 8(c) and 8(d) show strategies for the detection of aptamer-target complexes without any modifications of the aptamers. Our future work aims at the development of an aptamer assay for kanamycin A using the described strategies.

Potential assay strategies using aptamers selected by the Capture-SELEX principle. The switch from the hybridized to the dehybridized stage of the aptamers after target addition can be used to generate a recordable signal based on fluorometric methods like changes in fluorescence intensities, fluorescence resonance energy transfer (FRET), FRET with molecular beacon formation, or fluorescence polarization (FP).

4. Conclusions

A special variant of the SELEX technology, the Capture-SELEX process, has been established for the selection of DNA aptamers for those targets, which cannot be immobilized on a solid surface, for example, small organic target molecules. But the immobilization of one of the binding partners is a very important and convenient method to enable an efficient separation of target-binding and nonbinding oligonucleotides during the aptamer selection procedure. Therefore, a partially randomized DNA library with a defined docking sequence within the random region has been designed to hybridize with capture oligos on magnetic beads. By this way, the oligonucleotides of the library can be immobilized instead of the target molecules. An extensive washing of the resulting DNA-bead-complexes is recommended to remove all unhybridized DNA molecules. Moreover, two special elution steps are used prior to the target binding step to minimize the amount of oligonucleotides that is released nonspecifically from the beads, for example, caused by incorrect duplex formation or by mechanical forces. For the assessment of the progress of selection in each SELEX round, the direct comparison between the background elution step and the target binding step is used. The amount of oligonucleotides eluted in presence of the target should significantly exceed the background elution and should further increase (until reaching a maximum level) during the Capture-SELEX process. A fluorescein label serves for the quantification of the DNA in the different SELEX fractions of each SELEX round (from the second round on) by fluorescence detection.

The configuration of the Capture-SELEX process is very flexible. The background elution step can easily be combined with a negative or counterselection step in order to influence the specificity of the aptamers to be selected. In addition to single target substances, the Capture-SELEX process is applicable to target mixtures containing two or more different substances for a parallel selection of differently targeted aptamers. Moreover, the Capture-SELEX principle is equally attractive for other classes of target molecules, like polysaccharides, peptides, or proteins.

The successful application of the Capture-SELEX process for the selection of DNA aptamers for kanamycinA was demonstrated, and dissociation constants were determined. For a more precise examination of the received aptamers, binding specificities towards other aminoglycoside antibiotics have to be determined. This is under investigation and will be subject of a following publication.

The Capture-SELEX principle also provides the possibility to develop fluorescence-based detection assays without extensive additional modifications of the selected aptamers. The switch of the oligonucleotides between the duplex structure (with the capture oligos) and the aptamer-target complex upon addition of target molecules can be used to generate a measurable signal. Therefore, future work will aim at the assay development for a sensitive detection of kanamycin A based on the particular structural features of the selected aptamers, as depicted in Section 3.5.


This work was supported by the German Federation of Industrial Research Associations (AiF) in context of the funding Program PRO INNO II (KF 011004DA6) and by the Federal Ministry of Education and Research, Germany (BMBF) within the Program BIONA (01RB0805B). The authors thank their colleagues Anja Prange, Carina Flores de Looß, and Christine Reinemann for technical assistance and discussion.


  1. T. Hermann and D. J. Patel, “Adaptive recognition by nucleic acid aptamers,” Science, vol. 287, no. 5454, pp. 820–825, 2000. View at: Publisher Site | Google Scholar
  2. D. J. Patel, “Structural analysis of nucleic acid aptamers,” Current Opinion in Chemical Biology, vol. 1, no. 1, pp. 32–46, 1997. View at: Google Scholar
  3. D. J. Patel, A. K. Suri, F. Jiang et al., “Structure, recognition and adaptive binding in RNA aptamer complexes,” Journal of Molecular Biology, vol. 272, no. 5, pp. 645–664, 1997. View at: Publisher Site | Google Scholar
  4. A. D. Ellington and J. W. Szostak, “In vitro selection of RNA molecules that bind specific ligands,” Nature, vol. 346, no. 6287, pp. 818–822, 1990. View at: Publisher Site | Google Scholar
  5. C. Tuerk and L. Gold, “Systemic evolution of ligands by exponential enrichment: RNA ligands to bacteriophage T4 DNA polymerase,” Science, vol. 249, no. 4968, pp. 505–510, 1990. View at: Google Scholar
  6. S. Klussmann, The Aptamer Handbook, Functional Oligonucleotides and Their Applications, Wiley-VCH, Weinheim, Germany, 2006.
  7. R. Stoltenburg, C. Reinemann, and B. Strehlitz, “SELEX-A (r)evolutionary method to generate high-affinity nucleic acid ligands,” Biomolecular Engineering, vol. 24, no. 4, pp. 381–403, 2007. View at: Publisher Site | Google Scholar
  8. G. Aquino-Jarquin and J. D. Toscano-Garibay, “RNA aptamer evolution: twodecades of selection,” International Journal of Molecular Sciences, vol. 12, no. 12, pp. 9155–9171, 2011. View at: Publisher Site | Google Scholar
  9. S. C. B. Gopinath, “Methods developed for SELEX,” Analytical and Bioanalytical Chemistry, vol. 387, no. 1, pp. 171–182, 2007. View at: Publisher Site | Google Scholar
  10. S. D. Jayasena, “Aptamers: an emerging class of molecules that rival antibodies in diagnostics,” Clinical Chemistry, vol. 45, no. 9, pp. 1628–1650, 1999. View at: Google Scholar
  11. W. Kusser, “Chemically modified nucleic acid aptamers for in vitro selections: evolving evolution,” Journal of Biotechnology, vol. 74, no. 1, pp. 27–38, 2000. View at: Google Scholar
  12. R. Stoltenburg, C. Reinemann, and B. Strehlitz, “FluMag-SELEX as an advantageous method for DNA aptamer selection,” Analytical and Bioanalytical Chemistry, vol. 383, no. 1, pp. 83–91, 2005. View at: Publisher Site | Google Scholar
  13. R. Nutiu and Y. Li, “In vitro selection of structure-switching signaling aptamers,” Angewandte Chemie, vol. 44, no. 7, pp. 1061–1065, 2005. View at: Publisher Site | Google Scholar
  14. B. Strehlitz, C. Reinemann, S. Linkorn, and R. Stoltenburg, “Aptamers for pharmaceuticals and their application in environmental analytics,” Bioanalytical Reviews, vol. 4, no. 1, pp. 1–30, 2012. View at: Publisher Site | Google Scholar
  15. A. Crameri and W. P. C. Stemmer, “1020-fold aptamer library amplification without gel purification,” Nucleic Acids Research, vol. 21, no. 18, p. 4410, 1993. View at: Google Scholar
  16. Santa Cruz Biotechnology Inc, “Kanamycin acid sulfate data sheet,” 2007-2012, View at: Google Scholar
  17. SRC, “Interactive physProp database demo,” 2011, View at: Google Scholar
  18. Sigma-Aldrich, “Sotalol hydrochloride specification sheet (n. d.),” View at: Google Scholar
  19. C. Gaillard and F. Strauss, “Ethanol precipitation of DNA with linear polyacrylamide as carrier,” Nucleic Acids Research, vol. 18, no. 2, p. 378, 1990. View at: Google Scholar
  20. K. P. Williams and D. P. Bartel, “PCR product with strands of unequal length,” Nucleic Acids Research, vol. 23, no. 20, pp. 4220–4221, 1995. View at: Google Scholar
  21. M. A. Larkin, G. Blackshields, N. P. Brown et al., “Clustal W and Clustal X version 2.0,” Bioinformatics, vol. 23, no. 21, pp. 2947–2948, 2007. View at: Publisher Site | Google Scholar
  22. J. D. Thompson, D. G. Higgins, and T. J. Gibson, “CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice,” Nucleic Acids Research, vol. 22, no. 22, pp. 4673–4680, 1994. View at: Google Scholar
  23. R. Chenna, H. Sugawara, T. Koike et al., “Multiple sequence alignment with the Clustal series of programs,” Nucleic Acids Research, vol. 31, no. 13, pp. 3497–3500, 2003. View at: Publisher Site | Google Scholar
  24. X. Luo, M. Mckeague, S. Pitre et al., “Computational approaches toward the design of pools for the in vitro selection of complex aptamers,” RNA, vol. 16, no. 11, pp. 2252–2262, 2010. View at: Publisher Site | Google Scholar
  25. B. Vant-Hull, A. Payano-Baez, R. H. Davis, and L. Gold, “The mathematics of SELEX against complex targets,” Journal of Molecular Biology, vol. 278, no. 3, pp. 579–597, 1998. View at: Publisher Site | Google Scholar
  26. N. Nikolaus, A. Prange, R. Stoltenburg, and B. Strehlitz, Aminoglykosid-spezifische Aptamere. patent, application number: DE10 2011 006612. 1-41, Germany, 2011.
  27. K. M. Song, M. Cho, H. Jo et al., “Gold nanoparticle-based colorimetric detection of kanamycin using a DNA aptamer,” Analytical Biochemistry, vol. 415, no. 2, pp. 175–181, 2011. View at: Publisher Site | Google Scholar
  28. H. T. Allawi and J. SantaLucia, “Nearest neighbor thermodynamic parameters for internal G·A mismatches in DNA,” Biochemistry, vol. 37, no. 8, pp. 2170–2179, 1998. View at: Publisher Site | Google Scholar
  29. H. T. Allawi and J. SantaLucia, “Nearest-neighbor thermodynamics of internal A·C mismatches in DNA: sequence dependence and pH effects,” Biochemistry, vol. 37, no. 26, pp. 9435–9444, 1998. View at: Publisher Site | Google Scholar
  30. E. J. Cho, J. W. Lee, and A. D. Ellington, “Applications of aptamers as sensors,” Annual Review of Analytical Chemistry, vol. 2, pp. 241–264, 2009. View at: Publisher Site | Google Scholar
  31. K. Sefah, J. A. Phillips, X. Xiong et al., “Nucleic acid aptamers for biosensors and bio-analytical applications,” Analyst, vol. 134, no. 9, pp. 1765–1775, 2009. View at: Publisher Site | Google Scholar
  32. K. Han, Z. Liang, and N. Zhou, “Design strategies for aptamer-based biosensors,” Sensors, vol. 10, no. 5, pp. 4541–4557, 2010. View at: Publisher Site | Google Scholar
  33. R. E. Wang, Y. Zhang, J. Cai, W. Cai, and T. Gao, “Aptamer-based fluorescent biosensors,” Current Medicinal Chemistry, vol. 18, no. 27, pp. 4175–4184, 2011. View at: Publisher Site | Google Scholar
  34. D. P. Morse, “Direct selection of RNA beacon aptamers,” Biochemical and Biophysical Research Communications, vol. 359, no. 1, pp. 94–101, 2007. View at: Publisher Site | Google Scholar
  35. M. Rajendran and A. D. Ellington, “In vitro selection of molecular beacons,” Nucleic Acids Research, vol. 31, no. 19, pp. 5700–5713, 2003. View at: Publisher Site | Google Scholar
  36. R. Nutiu and Y. Li, “Aptamers with fluorescence-signaling properties,” Methods, vol. 37, no. 1, pp. 16–25, 2005. View at: Publisher Site | Google Scholar


Copyright © 2012 Regina Stoltenburg et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Consensus Opinions

This document is intended to serve as a focal point of discussions between industry and regulatory agency representatives, to attempt to identify best practices for assessing EP with oligonucleotides due to uncertainties in discerning regulatory expectations regarding the scope of EP assessment.

The primary focus was on the ON subclasses that impact expression of “host” gene products (antisense, siRNAs, and related ONs that target mRNA), with later and more limited discussions on the aptamer and IS subclasses. Therefore, the outcome of the discussions is presented for each of these types of ONs. Many of the principles are expected to be relevant to other subclasses but will need to be carefully considered as those development programs advance toward clinical trials.

ONs affecting gene product expression

Species selection

A key question initially addressed was how many and which animal species should be used for assessment of EP. Characterization of cross-species pharmacologic activity is important, particularly among the species commonly used for toxicity studies, and with an eye to those species that have been used historically for safety assessment of ONs, such as non-human primates, to enable selection of pharmacologically relevant species. However, the above-described species specificity of ONs often precludes ubiquitous activity in the commonly used animal species. In some cases, activity can be documented or predicted (based on sequence homology) for 1 species (e.g., monkey) but may be lacking in other common laboratory species. Or, there may be no cross-species activity. Under such circumstances, concerns may arise about the validity of employing animal-active analogues (see below). In order to make judgments about the appropriate scope of EP assessment, the likelihood of encountering adverse effects stemming from EP should be considered. In this regard, only a few examples were identified where EP was expressed in animals that translated into significant toxicity and the Subcommittee could not readily identify ONs for which such expression of EP was a key safety issue that impacted the selection of the starting clinical dose. The absence of numerous examples of serious toxicity attributable to EP with ONs may be related, at least in part, to the often incomplete or modest impact on gene expression from ASOs, which is the subclass that has been most widely investigated. In addition, the toxicity profile of many types of ONs is often dominated by nonspecific “class effects,” particularly for ONs that contain chemically modified backbones that impart a strong anionic character, most notably, those with a phosphorothioate modification. The phosphorothioate modification has been widely employed to confer nuclease resistance and promote longer in vivo tissue persistence. For such molecules, the class effects typically manifest at lower dose levels than those required to produce complete inhibition of gene product expression, thereby obscuring or precluding any adverse effects stemming from EP.

More intensive investigation of EP might be justified based on the nature of the target and from well-founded concern over the consequences of knockdown of the specific target, or from the complete absence of information of the consequences of such knockdown. The newer generations of ASOs, as well as the newer subclasses targeting gene product expression (such as siRNAs and microRNAs) appear to be generally more pharmacologically potent, owing to greater in vivo stability and/or to their mechanism of action. For these newer types of ONs, greater attention to assessment of EP may be warranted. In general, the level of effort directed towards assessing the safety implications of EP with ONs should be based on the available body of information (case by case), and there should not be rigid requirements about the number of species and conditions for testing. The types of information to be considered in making a judgment about the scope of EP assessment should include: (1) the role of the target gene product and what is known about loss of its function (e.g., from knockout models) (2) the potency and persistence of the ON-induced inactivation (3) the route of administration (and likelihood of extensive systemic exposure) (4) the expression profile of the specific target (5) the dosing frequency and duration and (6) the clinical indication (risk–benefit considerations).

As an example, the level of concern about possible EP-related toxicity may be elevated when the ON is targeting a ubiquitous key regulatory protein, whereas there may be less concern when the ON is directed against a protein that is expressed only in diseased tissue or abnormal cells (e.g., cancer cells). Similarly, there may be more concern about the consequences of EP when the ON is delivered systemically at relatively high doses and is widely distributed into tissues, as opposed to an ON that is administered topically to a discrete area of skin and is not absorbed systemically. Obviously, the number of possible scenarios is quite large, and more extensive discussion of specific circumstances and considerations is beyond the scope of this document.

In general, it may be appropriate to consider the options for addressing EP in 2 species (i.e., both the rodent and non-rodent species) for cases where ONs act by novel mechanisms and/or where the toxicity profile of the subclass is not well characterized. However, for those ON subclasses under discussion, investigation of EP in 1 species should suffice, unless further investigation is warranted by compelling theoretical concerns or by the results of general toxicity studies that revealed a novel toxicity apparently stemming from EP. Apart from those reservations, assessment in 1 species is consistent with International Conference on Harmonisation (ICH) S6 (R1), where similar limitations in cross-species activity may exist, especially considering the fact that the dose-limiting toxicity of biopharmaceuticals is most often an extension of pharmacologic activity, whereas this has rarely been the case for the ON subclasses under discussion.

Determination of pharmacologic relevance

Another topic of extensive discussion was the level of information that is needed to “validate” a species for investigation of EP. If there is 100% sequence homology between the human and animal target mRNA sequences, the human ON will most likely be active in the animal species, particularly if the human ON had been screened for optimal activity. Under such circumstances, documentation of activity of the ON in the animal species is probably not needed. However when the degree of non-homology is moderate (e.g., 1 or 2 mismatches), the animal species may still be a valid model for assessment of EP (with the human ON), but the uncertainty warrants investigations to substantiate pharmacologic activity, either by demonstrating decreased target gene expression or some other measure of intended activity reflecting decreased gene expression (e.g., efficacy in a relevant animal model). Pharmacologic activity could be confirmed by a relatively simple in vitro assay, such as demonstration of reduced target mRNA or protein expression following incubation of the ON in a relevant in vitro system that contains the target gene (e.g., a monkey peripheral blood leukocyte preparation). The potency of the human ON in the whole animal or in vitro system need not precisely match that of the ON in a human system to be able to extrapolate the findings. This view is supported by the fact that the ON will usually be tested at high clinical-multiple doses in toxicity studies, such that target gene product inhibition will likely be achieved. However, for those programs in which a high clinical-multiple dose level cannot be evaluated in the toxicity studies (e.g., ocular studies), the pharmacologic potency of the human ON in the animal species should be considered in making a judgment about whether EP can be adequately assessed in that species.

When no activity can be documented for a human ON in any of the animal species that are commonly used for toxicity studies, the use of animal-active analogues (surrogates) for assessment of EP should be considered in 1 species (see below).

Case studies

Many cases are intermediate between the extremes of ubiquitous cross-species activity for the human ON and no cross-species activity, and several of these were discussed. One scenario commonly encountered is when the human ON has 100% sequence homology (i.e., complementarity) with the target mRNA region for 1 animal species (e.g., monkey) and/or documentation of pharmacologic activity in the non-rodent species to be used for toxicity investigations is adequate, and the sponsor additionally has utilized a rodent analogue for pharmacology investigations. In this circumstance, documentation of EP in a single species (e.g., non-human primate) was considered sufficient. Inclusion of the rodent analogue in the good laboratory practice (GLP) toxicity study should not be needed, as EP of the clinical candidate should be adequately assessed in the non-rodent study. This position is also consistent with ICH S6 (R1). However, several developers have set a precedent for investigating the toxicity of a rodent analogue in such circumstances, either because such testing was requested by a Food and Drug Administration reviewing division, or because dialogue with the Agency could not be accomplished prior to initiating toxicity studies and the developer was concerned that assessment of EP in only 1 species (non-rodent) might be perceived as inadequate by the Agency. Importantly, in all of these cases the additional testing with the rodent analogue did not inform about the risks of EP (i.e., no new manifestations of toxicity stemming from EP were identified with the rodent analogue), and in some instances, unclear results were obtained that were difficult to interpret for clinical decision-making (e.g., no adverse findings with the clinical candidate in a relevant non-human primate study and toxicity with the rodent analogue in a rodent study).

In some development programs, the human ON has been inactive in the non-rodent species but active in rodents. In general, the same considerations about the potential adequacy of a 1-species assessment of EP would be applicable, such that EP investigation in a non-rodent species should not absolutely be required. However, in such cases where a non-rodent species analogue is available (e.g., because it had been developed for use in pharmacology studies), the inclusion of that analogue in the non-rodent toxicity study may be considered.

Another scenario is the case of an ON that is not pharmacologically active in the non-rodent species (e.g., non-human primate) or in rodents, but a rodent analogue is available. This circumstance may warrant utilization of the rodent analogue in the GLP toxicity study to address EP (i.e., tested in parallel with the human sequence), but developers should give careful consideration to the concerns outlined below regarding the use of analogues.

Concerns about the use of animal-active analogues

There have been numerous programs with ONs for which an animal-active analogue was developed as a pharmacology tool and was ultimately tested for safety in the same species in a GLP toxicity study. These analogues often have a substantially or completely different nucleotide sequence than the human ON, and hence, they are distinct molecular entities. One concern is that the intensity of class effects can vary substantially among ONs of different sequences, which can result in different manifestations of toxicity between the human ON and analogue that may be construed as evidence of EP but is actually a reflection of some non-pharmacology-based sequence-related difference. Companies with extensive experience in the safety assessment of antisense ONs have learned over the years that certain ONs exhibit dramatically greater toxicity than other sequences for reasons unrelated to EP, and this anomalous toxicity may occur in approximately 10%–20% of the ONs tested (personal communication among Subcommittee members). The structure–activity relationship for this type of unexpected toxicity is not well understood, and hence, it would behoove sponsors to select active analogues that do not exhibit such properties before including them in a side-by-side comparison with the human ON in toxicity studies. However, because of the sense of urgency for most ON development programs, many sponsors elect to forgo preliminary toxicity screening of active analogues.

In addition, the fact that the analogue possesses a different sequence than the human ON poses a risk that the analogue could, by chance, elicit some type of mechanism-based (e.g., antisense) effect (i.e., an “off-target” inhibition of another gene product) that translates into toxicity. If unexpected toxicity occurred with an analogue, discerning whether the effect reflected EP or an off-target effect may be difficult. There are other possible differences in properties between the human ON and an analogue that are unrelated to EP but could yield differential toxicity that could be construed as evidence of EP. These include differences in pharmacokinetic (PK), biodistribution and tissue stability. Although the PK and tissue distribution profile is typically similar among ONs that are structurally related, there may be some influence of nucleotide sequence on ON disposition that could affect toxicologic potency.

In the event that the human molecule was not species cross-reactive and EP needed to be evaluated in at least 1 species, it may generally be more appropriate to conduct focused safety evaluations with analogues in a disease model in which the target may be over-expressed, with specific investigations of endpoints that would be expected to be affected by the EP, rather than subject the analogue to a general toxicity evaluation in parallel with the human ON. Studies in disease models might be challenging to accomplish and usually cannot always be performed under GLP-compliant conditions. However, the absence of a GLP setting should not be a deterrent from conducting such studies, as long as the execution and record-keeping practices for such studies are sound. However, even with more focused safety studies, the potential to encounter off-target toxicities with analogues exists and could yield results that are difficult to interpret.

The possible utility of analogues for later-stage investigations of reproductive toxicity and/or carcinogenicity was also discussed. For those human ONs that have activity in non-human primates, but not rodents, a rodent analogue may be used to assess EP-related reproductive toxicity, but such judgments should be based, at least in part, on the level of concern or uncertainty about whether inhibited expression of the targeted gene product would be expected to affect reproductive performance. Therefore, while the use of animal active analogues for general toxicity investigations should be approached with caution, it is recognized that such analogues may be the best option for addressing requirements for reproductive toxicity testing. In addition, such analogues may have a role in the carcinogenicity evaluation, considering that these studies are conducted only in rodents however, such an undertaking should be carefully considered as per the criteria discussed above (i.e., regarding the level of concern about the consequences of target inactivation and other factors).

When evaluating the safety of an animal-active analogue, there are a number of tactical considerations about how to accomplish this. At the outset, it is important to understand the additional resources needed to manufacture and characterize a second test item. The most common paradigm has been to include 1 or more groups to be treated with the analogue in the GLP toxicity study, in parallel with the human sequence. For many programs, sponsors have elected to test only 1 dose level of the analogue, typically matching the high or middle dose level of the human sequence. However, the evaluation of a single dose level presents a risk that, should the group exhibit a unique or more pronounced toxicity that is regarded as a manifestation of EP, a regulatory issue may arise from not having characterized a no-adverse-effect level (NOAEL) for that effect. Hence, careful consideration needs to be given to the number of groups required for proper evaluation of an analogue. As mentioned, prescreening a range of doses may aid this decision process, although it may be more time-efficient to simply include multiple groups in a GLP toxicity study.

Questions were also addressed about whether the use of an active analogue as a test article in a GLP toxicity study should be subject to the same analytical stringency as the primary (human ON) test article. It was the Subcommittee's opinion that the prestudy analysis of the analogue need not be conducted under GLP or GMP conditions, as long as the material is well characterized by methodology similar to that employed for the primary test article (human ON) and the analysis results are well documented. However, as is expected for any study element that is not fully GLP compliant, use of non-GMP material in a GLP study should be addressed in the compliance section of the study report.

Similarly, regarding the dosimetry and/or exposure verification for analogues, some concessions should be allowed relative to full GLP compliance. It was acknowledged that the resource allocation is not trivial, especially for those programs involving multiple ONs with corresponding analogue combinations. For GLP-complaint determination of dosimetry, a sponsor would need to undertake method development and validation to enable analysis of the analogue concentration, homogeneity, and stability in dosing solutions under GLP-compliant conditions, in much the same manner as is done for the human sequence. Hence, most sponsors have elected to forego assessment of dosimetry for analogues. The rationale for this decision is based largely on the fact that the dosing solutions of the analogue are prepared and delivered in the same manner as the primary (human) test article. The absence of dosimetry verification for an animal-active analogue should be cited as an exception to GLP compliance in the study report.

Determination of exposure of the animals to the analogue in the context of a GLP toxicity study would require additional resource allocation in the form of bioanalytical method development and validation prior to study sample analysis. However, because most ONs are administered by intravenous or subcutaneous routes, and because absorption is complete with intravenous dosing and known to be extensive with subcutaneous dosing, exposure of the test system can be assumed to be somewhat similar for an analogue and the human sequence, thereby justifying reliance on the toxicokinetic data obtained with the primary test article (human sequence) to represent the likely exposure of the analogue. The primary caveat to this assumption is that, following systemic absorption, the in vivo stability of the analogue may differ substantially from that of the human sequence, which could translate into a different degree of tissue accumulation with repeated dosing that could dictate a different severity of toxicity. Such differences in toxicity between the analogue and human sequence may be inappropriately construed as evidence of EP. Therefore, it may be prudent to include additional animals in the analogue-treated groups for toxicokinetic (plasma and tissue) sample collection, to be stored frozen and possibly analyzed if an analogue exhibited a different degree of toxicity than the human sequence.

The use of inactive analogues as control articles

Inactive analogues have been used to permit the distinction of EP-related effects from non-pharmacologic effects. This approach is becoming increasingly prevalent for siRNA programs, particularly for those programs in which the siRNA is delivered via a specialized formulation and formulation-based toxicity is expected. This circumstance presents challenges in distinguishing between the effects of the formulation excipients, the non-specific ON class effects, and effects stemming from EP. In such cases, the inclusion of an inactive analogue can be very informative. However, the concerns expressed about the use of active analogues are relevant to the use of inactive analogues (i.e., that anomalous toxicity related to off-target mechanism-based activity or other sequence-dependent effects could manifest and could confound interpretation of the study data). In those cases in which the human ON is devoid of activity in the animal species, the human ON could serve as an inactive control ON and may be the best choice for such a control because it avoids the addition of another inactive sequence to the study design. However, it is important to ensure that the human ON is truly inactive in the animal species.

The role of formulations

Many ON development programs are moving forward with specialized delivery systems, which could impact the likelihood of encountering significant toxicity related to EP. On the one hand, formulations that afford targeted delivery to specific cell types could dramatically increase the potential for adverse manifestations of EP in those cell types. However, if such targeted delivery is directed against cancer cells or cells containing non-host pathogens, and if the pharmacologic objective is cellular destruction, adverse consequences of EP in those cell types may be moot. For several types of specialized formulations, it appears that the toxicity stemming from excipients or other properties of the formulation apart from the ON content may be much more pronounced than any EP effect of the ON, particularly if the formulated ON possesses little or no xenobiotic chemical modification. However, it is difficult to know apriori whether the formulation-based toxicity or the ON-related toxicity will be predominant. Hence, the increasing use of delivery formulations presents new considerations about the strategies and scope of EP assessment, and sponsors are encouraged to examine the circumstances of their program(s) and make appropriate proposals for regulatory interactions.

Aptamer ONs

Aptamers are a unique subclass of ONs that are designed to interact with a protein, as opposed to mRNA. The tertiary structure of ONs is sufficiently complex to provide a rich array of conformations for interacting with the proteins, and a complex progressive selection process is applied to yield a molecule with very high affinity for the target protein, such that binding of the aptamer ON to the protein inhibits the function of the protein. Hence, the anti-human aptamer that is selected possesses a unique structure that is highly specific for the target protein. In this respect, aptamers behave like monoclonal antibodies, in that their cross-species activity may be limited if conservation of the protein sequence and structure across species is poor. Therefore, most aptamers exhibit activity in primate species but typically not in rodents or other non-primate species such as rabbits and dogs.

The leading developers in this space have often relied on pharmacology data from human or non-human primate models, and they have occasionally developed a rodent analogue aptamer to enable in vivo pharmacology studies when non-human primate models are not available or feasible. However, because aptamer discovery is a sequential selection process rather than a molecular design process, the analogue will always be substantially different in length, composition, and structure than the anti-human aptamer, and the only common properties may be the homologous target binding and basic nucleic acid structure. The analogue is a valuable tool for performing pharmacology studies of the target biology, but it may not completely mimic the clinical candidate.

Many of the aptamers that are currently undergoing clinical and nonclinical development are polyethylene glycol conjugates (PEGylated), which is commonly introduced to confer desirable pharmacokinetic properties. PEGylation can profoundly affect the activity of an aptamer, which dictates that the final stages of selection of an optimally active human aptamer must be done with PEGylated molecules. However, since the rodent analogue that is employed for target biology investigations is typically not PEGylated, there may be significant differences in the property of the analogue that preclude its valid use in toxicity studies. With PEGylated aptamers, the primary chemistry-related toxicity that is observed is vacuolation of various cells, reflecting uptake of the PEGylated molecule. Hence, the toxicity profile of a non-PEGylated rodent analogue aptamer cannot be compared directly to the PEGylated human aptamer.

For all of these reasons, aptamer companies have generally avoided the use of analogue aptamers in toxicity studies, and they have relied mainly on the testing done with the human aptamer in primate toxicity studies (i.e., 1-species evaluation of EP). As discussed for the other subclasses, judgments about the appropriate scope of EP assessment for aptamers should be made on a case-by-case-basis.

Immunostimulatory ONs

Immunostimulatory (IS) ONs are much less species-specific than other subclasses and can exhibit exaggerated pharmacology across most species. The IS ONs interact with Toll-like receptors (TLRs) on immunocompetent cells and trigger intracellular events that translate into various cytokine-mediated responses. Common anatomic pathology findings stemming from the intended IS activity include injection site reactions, lymphoid hyperplasia, cellular infiltration in various organs, and related responses. These effects invariably dominate the toxicity profile of IS ONs and are dose limiting. Thus, EP is quite evident from pharmacology and toxicity studies of IS ONs, and the main challenge is to determine which animal species is the most appropriate model for human responses.

In this regard, several recent publications have documented the differences between rodent and primate species with respect to the cellular distribution of TLRs, the downstream cytokine responses and other sequelae, and the structure-activity relationships for TLR activation. Discussion of the details of those species differences is beyond the scope of this summary document (this has been addressed by the Immunomodulatory Subcommittee of the OSWG). However, there is a growing body of evidence indicating that the severity and diversity of the anatomic pathology elicited by IS ONs in rodents may not be representative of human responses. Non-human primates appear to be a better model for human responses to IS ONs. Further discussions are needed to help determine whether the rat provides any additional value in the context of general toxicity assessment and first-in-human dose selection.


Several new subclasses of ONs are emerging that have unique properties and will present new issues for assessment of EP. One of the most promising new subclasses is microRNA (miR). Various strategies to deliver anti-miRs or miR mimetics are being employed, but such programs are currently at an early stage. The mechanism for the modulation of gene regulation by miRs shares some features with the RNAi pathway, but with some important distinctions. MicroRNAs function by using the RNA-induced silencing complex to bind to short “seed regions” in mRNAs, which are typically present within the 3’-untranslated region of mRNAs for multiple interrelated genes. Therefore, unlike other approaches to blocking gene expression that are aimed at 1 target (e.g., antisense and RNAi), the introduction of a miR mimetic or anti-miR typically affects a constellation of largely interrelated gene products. The direct effect of an anti-miR is the derepression of expression (increase), but downstream genes may change in a positive or negative direction (quite commonly in both directions for different gene products). The aim of miR-based therapies is to produce some degree of overall modulation of a biological response by mimicking or inhibiting a disease-related miR. Another important distinction between miR-based therapy and siRNAs and ASOs is that the magnitude of modulation of any particular gene product is modest (typically no more than a 2-fold change), in contrast to the more dramatic inhibition intended for other ONs that affect gene expression. MicroRNAs function by fine tuning levels of expression. Although the experience is limited thus far, the conservation of miRs across species, combined with their mechanism of action, translates into robust cross-species activity, such that most miRs exhibit the desired changes in biomarkers in species commonly used for safety assessment of ONs (i.e., monkeys and rodents). Hence, animal-active analogues are generally not needed to assess EP for this class.

The extent to which the pattern of miR-induced modulation of gene product expression will be analogous across species is uncertain, although, theoretically, there should be considerable commonality. Because of the broader spectrum of mRNA targets for an individual miR, it may be difficult to distinguish whether any specific change in gene expression reflects intended targeting versus off-targeting versus a compensatory response of the cell. Micro arrays and other means of characterizing changes in gene product expression can be used to obtain snapshots of the modulation profile, as such profiling is central to understanding the miR's pharmacologic response. From such information, one might also glean insights into whether certain ancillary effects may be anticipated based on the observed pattern of alterations in gene expression. However, it is questionable whether this type of information should serve as an impetus for special, dedicated toxicity investigations, apart from standard regulatory safety studies.

In summary, the current approach to identifying potential safety issues associated with the intended pharmacologic action of miRs needs to be considered, particularly because of the novelty of such molecules and the uncertainty about the array of gene targets affected and any downstream consequences of such effects. However, because the miRs currently under development have shown cross-species activity, and because they typically produce only a moderate degree of modulation in gene product expression, there should not be a heightened concern about the likelihood of encountering exaggerated pharmacology with these molecules, and there is no reason to expect that conventional toxicity studies will fail to reveal effects in the EP category that would be relevant to human safety.


This work was supported by the Swedish Medical Research Council and by The Swedish Cancer Society. We are grateful to Samir El Andaloussi, Anna Berglöf, and Roger Strömberg, Karolinska Institutet Mark A. Behlke, Integrated DNA Technologies (IDT), Inc. Peter B. Dervan, California Institute of Technology Scott Henry, ISIS Pharmaceuticals Per Trolle Jørgensen, Erik B. Pedersen, and Jesper Wengel, University of Southern Denmark and Rula Zain, Karolinska University Hospital, for valuable comments. We apologize to all those who have contributed in many ways to the development of ON therapies that we were unable to cite their work owing to space limitations.

Watch the video: Riboswitches (January 2022).