Reports
O N
Systematic selection of modified aptamer pairs for diagnostic sandwich assays
Supplementary material for this article is available at www.BioTechniques.com/article/114134.
S IO N
Urs A. Ochsner, Louis S. Green, Larry Gold, and Nebojsa Janjic SomaLogic, Inc., Boulder, CO BioTechniques 56:125-133 (March 2014) doi 10.2144/000114134 Keywords: aptamer; modified nucleotides; SELEX; sandwich assay; diagnostics; epitope selection
LY
Reports
W
IT
H
P
E
R
M
IS
Protein diagnostic applications typically require pairs of analyte-specific reagents for capture and detection. We developed methods for the systematic isolation of slow off-rate modified aptamer (SOMAmer) reagents that bind to different epitopes and allow efficient pair-wise screening of multiple ligands. SOMAmers were generated via a second systematic evolution of ligands by exponential enrichment (SELEX), using complexes of target proteins with a primary, non-amplifiable SOMAmer and employing different modified nucleotides (e.g., naphthylmethyl- or tryptaminocarbonyl-dU) to favor alternate binding epitopes. Non-competing binding of primary and secondary SOMAmers was tested in radiolabel competition and sandwich binding assays. Multiplexed high-throughput screening for sandwich pairs utilized the Luminex platform, with primary SOMAmers as capture agents attached to different types of LumAvidin beads, which were then pooled for testing the secondary SOMAmers individually as detection agents. Functional SOMAmer pairs were obtained for Clostridium difficile binary toxin (CdtA) and for a panel of human proteins (ANGPT2, TSP2, CRDL1, MATN2, GPVI, C7, PLG) that had been previously identified as promising markers for cardiovascular risk. The equilibrium dissociation constants (Kd values) ranged from 0.02–2.7 nM, and the detection limits were in the low picomolar range for these proteins in SOMAmer sandwich assays. These results indicate that SOMAmer pairs hold promise for the development of rapid tests or specific diagnostic panels.
R E
P
R
IN
T
W hi le antibodies have been ver y successful diagnostic reagents, they can be difficult to procure in adequate quality and quantity (1) and have limited utility in arrays for multiplexed or high-content proteomic applications due to their inherent crossreactivity and non-universal assay conditions (2–4). Slow off-rate modified aptamer (SOMAmer) reagents have already demonstrated their utility in a highly multiplexed proteomic assay (SOM Ascan)(5), in immunohistochemical staining (6), and in antibodySOMAmer sandwich assays (7). The SOM Ascan assay utilizes only one
specific SOM A mer per target but employs two distinct bead-based kinetic selections for slow off-rates that characterize specific binding interactions (5). To harness proteomic signatures or disease-specific biomarkers to a maximal extent for point-of-care clinical use, transition from a discovery platform such as SOMAscan to other more streamlined assay formats will often be needed. Most diagnostic assays, however, rely on two reagents (generally antibodies) to achieve the desired specificity—typically one for capture and one for detection. SOMAmer pairs have not yet been described for any type of proteomic
assay, and few examples of proteins bound by two standard RNA or DNA aptamers exist to date (8–13). Special selection methods were required to force the selection toward non-overlapping epitopes for known aptamer pairs to the fibrinogen-recognition and heparin binding exosites of thrombin (9,14,15), the aV or b3 subunits of integrin aVb3 (11), TATA binding protein (TBP)(10), prion protein (PrP)(12), VEGF-165 (13), and bovine diarrhea whole virus (BVDV) (8). Conventional aptamers tend to bind to predominantly cationic epitopes on protein targets, which drives the best
Method summary: We developed methods for the isolation of modified aptamers that bind to different epitopes, using complexes of target proteins with primary, non-amplifiable SOMAmers and employing different modified nucleotides in a second SELEX. Multiplexed pair-wise screening utilized the Luminex platform, with primary SOMAmers as capture agents attached to different bead types and secondary SOMAmers in sandwich format for detection. Vol. 56 | No. 3 | 2014
125
www.BioTechniques.com
Reports
ligands to common surfaces. We have recently shown that epitopes engaged by SOMAmers are significantly more hydrophobic compared with those engaged by conventional aptamers (16) because they contain deoxyuridine residues that are modified at their 5-position with hydrophobic aromatic functional groups that mimic similar amino acid side-chains (17). This observation led us to postulate that SOMAmers may be less confined to electropositive epitopes compared with conventional aptamers. Here we exploit the different physicochemical properties (hydrophobicity, size, and shape) of various side chains in the modified nucleotides to isolate non-competing SOMAmer pairs.
Materials and methods
Proteins and SOMAmers Clostridium dif f icile binar y toxin (CdtA) was produced in a recombinant, His10 -tagged form (7). Human proteins available in recombinant form (R&D Systems, Minneapolis, MN)
included His-tagged angiopoietin-2, TSP2, CRDL1, MATN2, GPVI, and ESAM fused to Fc. C7 (Quidel, San Diego, CA) and plasminogen (Athens Research & Technolog y, Athens, GA), native proteins purified from human plasma, were biotinylated using EZ-Link NHS-PEG4-Biotin (Thermo, Rockford, IL) as described (5). Menu SOMAmers as primary binding agents to all protein targets had been previously isolated via SELEX (5,7). Truncated synthetic SOMAmers that contained the 4 0 -nucleotide target bind ing region and 5 nucleotides on each end were prepared via standard phosphoramidite chemistry using modified nucleotides (17). AB-H 50-mers contained a 5´-biotin-dA hexaethyleneglycol spacer for easy coupling to streptavidin (SA) and a 3´ inverted dT nucleotide (idT) for improved exonuclease stability. The SOMAmers were heat-cooled to ensure consistent renaturation and to minimize aggregation by heating to 95°C for 3 min in SB18T buffer (40 mM HEPES pH 7.5, 0.1 M NaCl, 5 mM KCl, 5 mM
Unlock your assay with custom aptamers. Data available on successful aptamer development projects at www.basepairbio.com
Why choose us: Ten years of successful aptamer development Recipient of multiple awards from NIH, NSF, EPA and others Experience with difficult targets, aptamers validated prior to delivery Patented methods can generate custom validated nucleic acid ligands in weeks not months
The Aptamer Discovery Company
Base Pair Biotechnologies, Inc. 8058 El Rio St. • Houston, TX 77054 (281) 829-8876 • [email protected] Visit us on the web: www.BasePairBio.com
BTN_Mar_BasePair.indd 1
Vol. 56 | No. 3 | 2014
Complete intellectual property transfer possible Free aptamer development through NCI grant, contact us for more details
126
2/19/14 9:37 AM
MgCl 2 , 0.05% Tween-20) and slowly cooling (0.1°C/second) to 37°C. Activity of the SOMAmers was confirmed in equilibrium binding and in pull-down assays (5). Sandwich SELEX For sandwich SELEX, the published selection protocol (5) was modified as follows: The proteins were complexed with AB-H versions of the primary SOMAmers immediately prior to each round of SELEX. For the first round of SELEX, 100 µL of 500 nM protein (50 pmol) were mixed with 5 µL of 5 µM (25 pmol) heat-cooled SOMAmers and incubated for 30 min at 37°C to allow complex formation. For subsequent rounds, 10 µL of 500 nM protein (5 pmol) and 2 µL of 5 µM (10 pmol) SOMAmers (2-fold excess) were used. Specific counter-selection beads were prepared fresh in each round of SELEX to reduce background due to non-specific SOMAmer-SOMAmer interactions. In brief, 2 µL of 5 µM (10 pmol) heat-cooled AB-H SOMAmers were added to 40 µL of 2.5 mg/mL SA beads in SB18T buffer and shaken for 15 min to allow immobilization. The beads were then washed to remove residual free SOMAmers, resuspended in 50 µL SB18T buffer, and added to the counter-selection plate along with 10 µL Hexa-His (AnaSpec, Fremont, CA) coated beads or with 50 µL SA beads or Protein G beads, according to the downstream partitioning method. SB18T buffer was used for SELEX and for all subsequent binding assays. The starting library consisted of 1 nmol (1014 –1015) of modified DNA sequences containing 40 consecutive randomized positions flanked by fixed sequences for PCR amplification. Separate libraries with different modified nucleotides were used, including TrpdU, 2NapdU, and PEdU. A kinetic challenge with 5 mM dextran sulfate was performed from SELEX round 2 forward to favor slow off-rates. Partitioning of the target– SOMAmer complexes was achieved with paramagnetic His-tag 2 or Protein G Dynabeads (Life Technologies, Carlsbad, CA) for His-tagged and Fc-fusion proteins, respectively, and with MyOne Streptavidin C1 beads (Life Technologies) for biotinylated targets. Selected DNA was eluted from the beads with buffered sodium perchlorate (1.8 M NaClO 4 , 40 mM PIPES, 1 mM EDTA, 0.05% Triton X-100, pH 6.8) for 5 min, then captured on primer beads and processed for PCR and primer extension to obtain sensestrands with the modified nucleotides using KOD EX DNA polymerase (EMD www.BioTechniques.com
Figure 1A_BT5549
A
Reports
B
Free target
1st
SELEX Random modified ssDNA library #1
25
SOMAmers
Bn
Nap
Competitive
20
SOMAmer pairs
Target−SOMAmer Complex
# sequences
Free target
= Bn Nap Trp PE Tyr 2Nap
Non-competitive
screen
40-nt random 5’ Fixed 3’ Fixed T TT T TTT T T T=
Pool 5551 (free CdtA SELEX) Pool 5579 (CdtA–4758-6 SELEX)
15 10
2nd
SELEX Random modified ssDNA library #2
Trp
SOMAmers
screen
5 0
PE
Tyr
2Nap
Figure 1. Strategy for the isolation and validation of a SOMAmer pair for a protein target. (A) A free target is used in a first SELEX with a modified random ssDNA library to isolate a set of SOMAmers that can be screened directly for pairs of non-competing clones. If no pairs are present, the SOMAmer with the best binding properties is allowed to form a complex with the target, which is then used in a second SELEX with a different modified library. The new SOMAmer clones are then screened for paired sandwich binding to the target. Bn, 5-benzylaminocarbonyl; Nap, 5-naphthylmethylaminocarbonyl; Trp, 5-tryptaminocarbonyl; PE, 5-phenylethyl-1-aminocarbonyl; Tyr, 5-tyrosylaminocarbonyl; 2Nap, 5-(2-naphthylmethyl)aminocarbonyl (B) Representation of sequence patterns (shared motifs) and multicopy sequences (equivalent clones with 100-fold excess (10 nM) of unlabeled primary SOMAmer as competitors. Sandwich filter binding assays Two va riations of f i lter bind ing sandwich assays were performed to obtain 12-point binding curves. The first method involved equilibrium binding of protein and radiolabeled, fulllength detection SOMAmer, followed by partitioning (30 min with intermittent shaking) with specific capture beads that were prepared by immobilizing AB-H primary SOMAmers on SA beads. The second method was based on the formation of tripartite complexes during equilibrium binding of protein, radiolabeled detection SOMAmer (fulllength, non-biotinylated) and excess (10 nM) unlabeled A B-H primar y SOMAmer, followed by partitioning (5 min with intermittent shaking) with SA beads. Both methods yielded comparable results. As controls, all sequences were tested in separate assays where the capture agent was omitted in order to identify non-specific background due to SA bead binding. Multiplexed sandwich screening assays The Luminex platform (Luminex Corporation, Austin, TX) was used for the multiplexed, pair-wise screening of SOM Amers (A B-H 50-mers) in sandwich binding format. Capture bead preparation and binding assays
127
were performed in MSBV N12 filter plates (EMD Millipore) pre-wet with SB17T buffer (SB18T supplemented with 1 mM EDTA). Different types (colors) of LumAvidin Microspheres (Luminex Corporation) were dispensed into separate wells (100,000 beads per well) and washed 3 × 1 min with 180 µL SB17T by vacuum filtration. SOMAmers were heat-cooled, 80 µL of 50 nM stocks of each capture SOMAmer for a given protein target were added to a different bead type, and the plate was shaken (20 min at RT, 1100 rpm) to allow for immobilization. The beads were then washed for 5 min each with 100 µL 50 nM streptavidin and with 10 mM biotin in SB17T, then 4 × 1 min with 180 µL SB17T. All capture beads for each protein were pooled, diluted to 1.7 mL with SB17T, dispensed (50 µL) into the wells of a pre-wet MSBVN12 filter plate, and mixed with 50 µL of 20 nM protein in SB17T supplemented with 1% BSA (SB17TB) or 50 µL SB17TB buffer for controls. After a 30–60 min incubation with shaking at 1100 rpm, the plate was vacuum-washed 2 × 1 min with 180 µL SB17TB and the beads were resuspended in 50 µL SB17TB. Heat-cooled detection SOMAmers were added to individual wells, using 50 µL of 12.5 nM stocks, and incubation was continued for 30 min with shaking. The beads were washed and resuspended as above. As a reporter, 50 µL of 10 µg/mL streptavidin-phycoerythrin (SA-PE) conjugate (Moss, Inc., Pasadena, MD) in SB17TB were added, and the beads were again shaken, washed, www.BioTechniques.com
Reports
and resuspended as above. The plate was read on a Luminex 100 analyzer (time out: enabled 120 s, DD gating: 7500–18000, reporter gain: high PMT).
A
Figure 2C_BT5549
Figure 2B_BT5549
Bmax >50%
Bmax=25-50%
Bmax=10-25%
C
Bmax=50%; light-green, 25%–50%; yellow, 10%–25%; orange, 10
P10643
2888–49
2.71
BndU
7579–67
3.16 (0.07)
2NapdU
1.35
Human
PLG
P00747
4151–6
4.86
BndU
7567–64
13.1 (1.94)
TrpdU
2.70
C. difficile
CdtA
Q9KH42
4758–6
0.86
TrpdU
5579–12
0.97 (0.28)
2NapdU
0.52
Values in parentheses are for the complex of the target with SOMAmer 1 b Determined in Luminex bead-based sandwich assay a
parallel with the same library. Comparative sequence analysis of the 2NapdU clones from affinity-enriched pools 5579 (CdtA-4758–6 complex) and 5551 (free CdtA) revealed clear differences in patterns and abundance, although there were some shared sequence motifs as well (Figure 1B). The new SOMAmers bound the free CdtA protein with affinities ranging from 0.05 nM-14 nM (Supplementary Figure S1) and were also tested in competition assays and in sandwich format together with the existing 4758–6 TrpdU SOM Amer (Supplementar y Figure S2). For 1 clone (5579–12) that possessed affinity comparable to 4758–6 (Kd = 0.67 nM), binding to CdtA was not affected the presence of a >100-fold Figure by 3_BT5549 excess (10 nM) or 4758–6 competitor, or when 4758–6 was used as a capture agent, indicating that the 2 SOMAmers bind to distinct CdtA epitopes (Supplementary Figure S3A). In contrast, binding of
Fraction F r a c tio n B B max m ax
1 .0 0 1.00
0 .7 5 0.75
0 .5 0 0.50
CdtA was much reduced for another clone (5579–11) when 4758–6 was used as a competitor or capture agent, in spite of superior affinity (K d = 0.05 nM), indicating overlapping binding epitopes (Supplementary Figure S3B). Representative sequences from each SELEX were also screened on the Luminex platform, where the first SOMAmer 4758–6 served as the capture agent on beads. Most clones from complex SELEX (5579–7, 5579–10, 5579–12, 5579–21) as well as one clone from free CdtA SELEX (5551–81) confirmed their binding when used as detection agent (Supplementary Figure S4A). Capture and detection agents were interchangeable for 4758–6 and 5579–12, producing similar binding curves in the Luminex sandwich assay (Supplementary Figure S4B). SELEX with the CdtA-4758–6 complex clearly resulted in a higher fraction of sequences useful for sandwich assays in conjunction
ANG PT2 ANGPT2 TSP2 TSP2 C R D L1 CRDL1 M ATN2 MATN2 C7 C7 PLG PLG GPVI GPVI ESAM ESAM
0 .2 5 0.25
0 .0 0 0.00 0 .0 0 0 1 0.0001
0 .0 0 1 0.001
0 .0 1 0.01
0 .1 0.1
1 1
1 0 10
1 00 100
Protein P ro te in (nM) (n M ) Figure 3. Sandwich binding curves obtained in the Luminex assay. The CV risk panel proteins were spiked in buffer and tested with the best-performing SOMAmer pairs identified in the pairwise screening assay listed in Table 1. In all cases, the sequence that had served to form the complex with the target during the sandwich SELEX was used for capture, and a new clone was used for detection. The maximum signals (RFU at Bmax) were 23046 (ANGPT2), 16623 (TSP2), 23349 (CRDL1), 25586 (MATN2), 26000 (C7), 13927 (GPVI), 7103 (PLG), and 3000 (ESAM), respectively.
Vol. 56 | No. 3 | 2014
130
with 4758–6, although free CdtA SELEX also produced a few non-competing SOMAmers (Supplementary Table S1). We next focused on eight human proteins that were promising markers for cardiovascular (CV) risk, including angiopoietin-2 (A NGPT2), thrombospondin-2 (TSP2), chordin-like 1 (CRDL1), matrilin-2 (MATN2), glycoprotein VI (GPVI), endothelial cellselective adhesion molecule (ESAM), complement 7 (C7), and plasminogen (PLG) (18). First, we screened archived SOMAmers with different sequences than the menu SOMAmer used in SOMAscan, and this data mining yielded functional SOMAmer pairs for three of these targets (C7, MATN2, PLG), all with BndU modifications (Supplementary Figure S5). Second, we performed SELEX with target–SOM Amer complexes, employing other modified nucleotide libraries (TrpdU and 2NapdU), and in parallel SELEX with free protein. In all cases, we obtained active clones that bound the protein-SOMAmer complexes, and using protein-SOMAmer complex targets during SELEX coupled with different modified libraries increased the likelihood of finding sandwich candidates. Exceptions were TrpdU clones selected with ESAM-2981–9 complex, which interacted with SOM A mer 2981–9 of the complex. The observed binding of the new clones to the protein-SOM A mer complexes did not prove the existence of a true sandwich, since they might simply displace the primary SOMAmer and bind to the same epitope. To make this distinction, the new clones were further evaluated in competition and sandwich binding assays (Supplementary Tables S2–S9). In addition, we set up a more global, multiplexed sandwich screening method using the Luminex platform (Figure 2A). In brief, each of the clones for a given target was separately immobiwww.BioTechniques.com
Reports
lized on a specific LumAvidin bead type. The beads were then pooled and used for capture of the target, and each of the same clones was used individually for detection. With respect to the 7–16 SOMAmer sandwich candidates for the 8 CV risk panel proteins, this approach reduced the 2-dimensional matrix of pair-wise screening to a single dimension, from 1116 assays (15 2 +14 2 +16 2 +14 2 +9 2 +7 2 +8 2 +7 2) to 90 assays (15+14+16+14+9+7+8+7). The best SOMAmer pairs were rapidly identified by plotting the net signals of the multiplexed Luminex sandwich screening assay as heat maps, as shown as an example for the 16 CRDL1 SOMAmers (Figure 2B). The menu SOMAmer 3362–61 for CRDL1 performed well when used as a capture or detection agent in conjunction with any of the new clones. This is not surprising, since the new clones had been selected to bind the complex of CRDL1 with 3362–61, and in fact, two clones (7575–6, 7575–19) performed better as detection agents than capture agents. In contrast, sandwich assays using one of the new clones (7575–2) with most other new clones produced much lower signals (Figure 2C). As expected, no signals were obtained with the same SOMAmer used for capture and detection. Sandwich binding curves on the Luminex platform were generated for seven of the eight CV risk panel proteins: ANGPT2, TSP2, CRDL1, MATN2, C7, PLG, and GPVI (Figure 3). Without any further assay optimization, the apparent equilibrium binding constants (Kd values) of the new SOMAmers for the protein complexes with the cognate primary SOMAmer ranged from 0.02–2.7 nM, and the lower limits of detection (5% of maximum signal) ranged from 1–100 pM (Table 1). Interestingly, 2 of the C7 SOMAmers (7579–65 and 7579–67) that had been selected with the C7–2888–49 complex showed much better affinity for the complex than for the free protein, with Kd = 0.13 nM versus 11.3 nM for 7579–65 and K d = 0.07 nM versus 3.16 nM, respectively (Table 1 and Supplementary Table S8). It is possible that the primary SOMAmer 2888–49 induces a conformational change of the target protein, thereby creating or improving the epitope for the second SOMAmer. No SOMAmer pairs were obtained for ESAM, which is the most basic (pI = 9.3) of the proteins and the only Fc-fusion we used here. All clones from SELEX with ESAM–2981–9 complex using a TrpdU library contained a Vol. 56 | No. 3 | 2014
common sequence pattern and bound not only the complex, but also the free primary SOMAmer 2981–9, indicating that the counter-selection with 2981–9 to remove such sequences had failed in this case. Other clones from SELEX with ESAM–2981–9 complex or free ESAM using a 2NapdU library (e.g., 7581–41) showed apparent binding of the complex or binding of ESAM in the presence of 2981–9 competitor, but in fact they were out-competing the primary SOMAmer for the same or overlapping binding site (Table 1, Supplementary Table S7). The SOM Amer pairs were also evaluated in plate-based sandwich assays, using biotinylated SOMAmers as capture agents immobilized on streptavidincoated plates, and as detection agents in conjunction with streptavidin-HRP conjugate (Supplementary Figure S6). Target titrations indicated some differences in assay performance compared with the bead-based test, emphasizing the need for future platform-specific assay development. We were successful in isolating SOMAmer pairs for 7 of 8 targets, which is in agreement with our typical success rate of 75%–80% with standard SELEX. Our new approach of applying a second SELEX with target-SOMAmer complexes strongly supports the notion that the use of different types of modified nucleotides enabled this breakthrough in the efficiency of finding separate SOMAmers for distinct epitopes. In our highly multiplexed diagnostic platform (SOMAscan), we have shown that background due to non-specific binding can be reduced by polyanionic competitors such as dextran sulfate as a consequence of differential off-rates between specific and non-specific binding events. Dextran sulfate, when present in excess (1–5 mM), will readily occupy binding sites vacated by SOMAmers with fast off-rates for non-specific binding and prevent them from rebinding to these sites. This feature, combined with the added specificity inherent in two-reagent sandwich-type measurements, may facilitate the development of sandwich assays with greater specificity and potential for higher multiplexing abilities compared with what can be achieved with antibodies. SOMAmer pairs hold promise for the development of specific panels in various areas of medical diagnostics for which a large installed base of instruments is already in place, using traditional sandwich assays or novel ligand proximity assays (19, 20), and also for small molecular sensing platforms (21,22).
132
Author contributions
U.A.O. designed and performed the sandwich SELE X method. L . S.G. conducted the sandwich screening and titration assays on the Luminex platform. L.G. and N.J. provided essential contributions on experimental strategy and manuscript preparation.
Acknowledgments
SOMAmer and SOMAscan are trademarks of SomaLogic, Inc. We thank Jeff Carter, Steve Wolk, Tim Fitzwater, Allison Weiss, Evaldas Katilius, Glenn Sanders, Nick Saccomano, Fintan Steele, and Jim Reed for their contributions regarding DNA synthesis, quality control, sequence analysis, assay development, and graphic design. No external funding was obtained for this work.
Competing interests
The authors are employees and shareholders of Soma L og ic , I nc. a nd co-inventors of patents relating to SOMAmer technology.
References
1. Marx, V. 2013. Calling the next generation of affinity reagents. Nat. Methods 10:829833. 2. Zichi, D., B. Eaton, B. Singer, and L. Gold. 2008. Proteomics and diagnostics: Let’s Get Specific, again. Curr. Opin. Chem. Biol. 12:78-85. 3. Perlee, L., J. Christiansen, R. Dondero, B. Grimwade, S. Lejnine, M. Mullenix, W. Shao, M. Sorette, et al. 2004. Development and standardization of multiplexed antibody microarrays for use in quantitative proteomics. Proteome Sci. 2:9. 4. Brody, E.N., L. Gold, R.M. Lawn, J.J. Walker, and D. Zichi. 2010. High-content aff init y-based proteomics: u n lock ing protein biomarker discovery. Expert Rev. Mol. Diagn. 10:1013-1022. 5. Gold, L., D. Ayers, J. Bertino, C. Bock, A. Bock, E.N. Brody, J. Carter, A.B. Dalby, et al. 2010. Aptamer-based multiplexed proteomic technology for biomarker discovery. PLoS ONE 5:e15004. 10.1371/ journal.pone.0015004. 6. Gupta, S., D. Thirstrup, T.C. Jar vis, D.J. Schneider, S.K. Wilcox, J. Carter, C. Zhang, A. Gelinas, et al. 2011. Rapid histochemistry using slow off-rate modified aptamers with anionic competition. Appl. Immunohistochem. Mol. Morphol. 19:273278. 7. Ochsner, U. A ., E . K ati l ius, a nd N. Janjic. 2013. Detection of Clostridium difficile toxins A, B and binary toxin with slow off-rate modified aptamers. Diagn. Microbiol. Infect. Dis. 76:278-285. 8. Park, J.W., S. Jin Lee, E.J. Choi, J. Kim, J.Y. Song, and M. Bock Gu. 2014. An ultra-
www.BioTechniques.com
Reports
sensitive detection of a whole virus using dual aptamers developed by immobilization-free screening. Biosens. Bioelectron. 51:324-329. 9. Tasset, D.M., M.F. Kubik, and W. Steiner. 1997. Oligonucleotide inhibitors of human thrombin that bind distinct epitopes. J. Mol. Biol. 272:688-698. 10. Shi, H., X. Fan, A. Sevilimedu, and J.T. Lis. 2007. RNA aptamers directed to discrete functional sites on a single protein structural domain. Proc. Natl. Acad. Sci. USA 104:3742-3746. 11. Gong, Q., J. Wang, K.M. Ahmad, A.T. Csordas, J. Zhou, J. Nie, R. Stewart, J.A. Thomson, et al. 2012. Selection strategy to generate aptamer pairs that bind to distinct sites on protein targets. Anal. Chem. 84:5365-5371. 12. Xiao, S.J., P.P. Hu, X.D. Wu, Y.L. Zou, L.Q. Chen, L. Peng, J. Ling, S.J. Zhen, et al. 2010. Sensitive discrimination and detection of prion disease-associated isoform with a dual-aptamer strategy by developing a sandwich structure of magnetic microparticles and quantum dots. Anal. Chem. 82:9736-9742. 13. Nonaka, Y., K. Sode, and K. Ikebukuro. 2010. Screening and improvement of an anti-VEGF DNA aptamer. Molecules 15:215-225. 14. Sosic, A., A. Meneghello, E. Cretaio, and B. Gatto. 2011. Human thrombin detection through a sandwich aptamer microarray: interaction analysis in solution and in solid phase. Sensors (Basel). 11:9426-9441. 15. Tennico, Y.H., D. Hutanu, M.T. Koesdjojo, C.M. Bartel, and V.T. Remcho. 2010. On-chip aptamer-based sandwich assay for thrombin detection employing magnetic beads and quantum dots. Anal. Chem. 82:5591-5597. 16. Davies, D.R., A.D. Gelinas, C. Zhang, J.C. Rohloff, J.D. Carter, D. O’Connell, S.M. Waugh, S.K. Wolk, et al. 2012. Unique motifs and hydrophobic interactions shape the binding of modified DNA ligands to protein targets. Proc. Natl. Acad. Sci. USA 109:1997119976. 17. Vaught, J.D., C. Bock, J. Carter, T. Fitzwater, M. Otis, D. Schneider, J. Rolando, S. Waugh, et al. 2010. Expanding the chemistry of DNA for in vitro selection. J. Am. Chem. Soc. 132:4141-4151. 18. Gill, R.D., S. Williams, A. Stewart, R. Mehler, T. Foreman, and B. Singer. 2012. Cardiovascular risk event prediction and uses thereof, p. PCT/US2012/058060. SomaLogic, Inc., USA. 19. Fredriksson, S., M. Gullberg, J. Jarvius, C. Olsson, K. Pietras, S.M. Gustafsdottir, A. Ostman, and U. Landegren. 2002. Protein detection using proximity-dependent DNA ligation assays. Nat. Biotechnol. 20:473-477. 20. Hu, J., T. Wang, J. Kim, C. Shannon, and C.J. Easley. 2012. Quantitation of femtomolar protein levels via direct readout with the electrochemical proximity assay. J. Am. Chem. Soc. 134:7066-7072. 21. Jo, M., J.Y. Ahn, J. Lee, S. Lee, S.W. Hong, J.W. Yoo, J. Kang, P. Dua, et al. 2011. Development of single-stranded DNA aptamers for specific Bisphenol A detection. Oligonucleotides 21:85-91. 22. Lee, J., M. Jo, T.H. Kim, J.Y. Ahn, D.K. Lee, S. Kim, and S. Hong. 2011. Aptamer sandwich-based carbon nanotube sensors for singlecarbon-atomic-resolution detection of non-polar small molecular species. Lab Chip 11:52-56.
Fast & Gentle Concentrator Vivaspin® Turbo 15 Concentrate your protein gently with the fastest tool on the market: Vivaspin® Turbo 15 Differentiation by design
Received 20 October 2013; accepted 24 January 2014. Address correspondence to Urs A. Ochsner, SomaLogic, Inc., Boulder. E-mail: [email protected] To purchase reprints of this article, contact: [email protected]
Vol. 56 | No. 3 | 2014
www.sartorius.com/vivaspin-turbo www.BioTechniques.com
O N
Systematic selection of modified aptamer pairs for diagnostic sandwich assays
Supplementary material for this article is available at www.BioTechniques.com/article/114134.
S IO N
Urs A. Ochsner, Louis S. Green, Larry Gold, and Nebojsa Janjic SomaLogic, Inc., Boulder, CO BioTechniques 56:125-133 (March 2014) doi 10.2144/000114134 Keywords: aptamer; modified nucleotides; SELEX; sandwich assay; diagnostics; epitope selection
LY
Reports
W
IT
H
P
E
R
M
IS
Protein diagnostic applications typically require pairs of analyte-specific reagents for capture and detection. We developed methods for the systematic isolation of slow off-rate modified aptamer (SOMAmer) reagents that bind to different epitopes and allow efficient pair-wise screening of multiple ligands. SOMAmers were generated via a second systematic evolution of ligands by exponential enrichment (SELEX), using complexes of target proteins with a primary, non-amplifiable SOMAmer and employing different modified nucleotides (e.g., naphthylmethyl- or tryptaminocarbonyl-dU) to favor alternate binding epitopes. Non-competing binding of primary and secondary SOMAmers was tested in radiolabel competition and sandwich binding assays. Multiplexed high-throughput screening for sandwich pairs utilized the Luminex platform, with primary SOMAmers as capture agents attached to different types of LumAvidin beads, which were then pooled for testing the secondary SOMAmers individually as detection agents. Functional SOMAmer pairs were obtained for Clostridium difficile binary toxin (CdtA) and for a panel of human proteins (ANGPT2, TSP2, CRDL1, MATN2, GPVI, C7, PLG) that had been previously identified as promising markers for cardiovascular risk. The equilibrium dissociation constants (Kd values) ranged from 0.02–2.7 nM, and the detection limits were in the low picomolar range for these proteins in SOMAmer sandwich assays. These results indicate that SOMAmer pairs hold promise for the development of rapid tests or specific diagnostic panels.
R E
P
R
IN
T
W hi le antibodies have been ver y successful diagnostic reagents, they can be difficult to procure in adequate quality and quantity (1) and have limited utility in arrays for multiplexed or high-content proteomic applications due to their inherent crossreactivity and non-universal assay conditions (2–4). Slow off-rate modified aptamer (SOMAmer) reagents have already demonstrated their utility in a highly multiplexed proteomic assay (SOM Ascan)(5), in immunohistochemical staining (6), and in antibodySOMAmer sandwich assays (7). The SOM Ascan assay utilizes only one
specific SOM A mer per target but employs two distinct bead-based kinetic selections for slow off-rates that characterize specific binding interactions (5). To harness proteomic signatures or disease-specific biomarkers to a maximal extent for point-of-care clinical use, transition from a discovery platform such as SOMAscan to other more streamlined assay formats will often be needed. Most diagnostic assays, however, rely on two reagents (generally antibodies) to achieve the desired specificity—typically one for capture and one for detection. SOMAmer pairs have not yet been described for any type of proteomic
assay, and few examples of proteins bound by two standard RNA or DNA aptamers exist to date (8–13). Special selection methods were required to force the selection toward non-overlapping epitopes for known aptamer pairs to the fibrinogen-recognition and heparin binding exosites of thrombin (9,14,15), the aV or b3 subunits of integrin aVb3 (11), TATA binding protein (TBP)(10), prion protein (PrP)(12), VEGF-165 (13), and bovine diarrhea whole virus (BVDV) (8). Conventional aptamers tend to bind to predominantly cationic epitopes on protein targets, which drives the best
Method summary: We developed methods for the isolation of modified aptamers that bind to different epitopes, using complexes of target proteins with primary, non-amplifiable SOMAmers and employing different modified nucleotides in a second SELEX. Multiplexed pair-wise screening utilized the Luminex platform, with primary SOMAmers as capture agents attached to different bead types and secondary SOMAmers in sandwich format for detection. Vol. 56 | No. 3 | 2014
125
www.BioTechniques.com
Reports
ligands to common surfaces. We have recently shown that epitopes engaged by SOMAmers are significantly more hydrophobic compared with those engaged by conventional aptamers (16) because they contain deoxyuridine residues that are modified at their 5-position with hydrophobic aromatic functional groups that mimic similar amino acid side-chains (17). This observation led us to postulate that SOMAmers may be less confined to electropositive epitopes compared with conventional aptamers. Here we exploit the different physicochemical properties (hydrophobicity, size, and shape) of various side chains in the modified nucleotides to isolate non-competing SOMAmer pairs.
Materials and methods
Proteins and SOMAmers Clostridium dif f icile binar y toxin (CdtA) was produced in a recombinant, His10 -tagged form (7). Human proteins available in recombinant form (R&D Systems, Minneapolis, MN)
included His-tagged angiopoietin-2, TSP2, CRDL1, MATN2, GPVI, and ESAM fused to Fc. C7 (Quidel, San Diego, CA) and plasminogen (Athens Research & Technolog y, Athens, GA), native proteins purified from human plasma, were biotinylated using EZ-Link NHS-PEG4-Biotin (Thermo, Rockford, IL) as described (5). Menu SOMAmers as primary binding agents to all protein targets had been previously isolated via SELEX (5,7). Truncated synthetic SOMAmers that contained the 4 0 -nucleotide target bind ing region and 5 nucleotides on each end were prepared via standard phosphoramidite chemistry using modified nucleotides (17). AB-H 50-mers contained a 5´-biotin-dA hexaethyleneglycol spacer for easy coupling to streptavidin (SA) and a 3´ inverted dT nucleotide (idT) for improved exonuclease stability. The SOMAmers were heat-cooled to ensure consistent renaturation and to minimize aggregation by heating to 95°C for 3 min in SB18T buffer (40 mM HEPES pH 7.5, 0.1 M NaCl, 5 mM KCl, 5 mM
Unlock your assay with custom aptamers. Data available on successful aptamer development projects at www.basepairbio.com
Why choose us: Ten years of successful aptamer development Recipient of multiple awards from NIH, NSF, EPA and others Experience with difficult targets, aptamers validated prior to delivery Patented methods can generate custom validated nucleic acid ligands in weeks not months
The Aptamer Discovery Company
Base Pair Biotechnologies, Inc. 8058 El Rio St. • Houston, TX 77054 (281) 829-8876 • [email protected] Visit us on the web: www.BasePairBio.com
BTN_Mar_BasePair.indd 1
Vol. 56 | No. 3 | 2014
Complete intellectual property transfer possible Free aptamer development through NCI grant, contact us for more details
126
2/19/14 9:37 AM
MgCl 2 , 0.05% Tween-20) and slowly cooling (0.1°C/second) to 37°C. Activity of the SOMAmers was confirmed in equilibrium binding and in pull-down assays (5). Sandwich SELEX For sandwich SELEX, the published selection protocol (5) was modified as follows: The proteins were complexed with AB-H versions of the primary SOMAmers immediately prior to each round of SELEX. For the first round of SELEX, 100 µL of 500 nM protein (50 pmol) were mixed with 5 µL of 5 µM (25 pmol) heat-cooled SOMAmers and incubated for 30 min at 37°C to allow complex formation. For subsequent rounds, 10 µL of 500 nM protein (5 pmol) and 2 µL of 5 µM (10 pmol) SOMAmers (2-fold excess) were used. Specific counter-selection beads were prepared fresh in each round of SELEX to reduce background due to non-specific SOMAmer-SOMAmer interactions. In brief, 2 µL of 5 µM (10 pmol) heat-cooled AB-H SOMAmers were added to 40 µL of 2.5 mg/mL SA beads in SB18T buffer and shaken for 15 min to allow immobilization. The beads were then washed to remove residual free SOMAmers, resuspended in 50 µL SB18T buffer, and added to the counter-selection plate along with 10 µL Hexa-His (AnaSpec, Fremont, CA) coated beads or with 50 µL SA beads or Protein G beads, according to the downstream partitioning method. SB18T buffer was used for SELEX and for all subsequent binding assays. The starting library consisted of 1 nmol (1014 –1015) of modified DNA sequences containing 40 consecutive randomized positions flanked by fixed sequences for PCR amplification. Separate libraries with different modified nucleotides were used, including TrpdU, 2NapdU, and PEdU. A kinetic challenge with 5 mM dextran sulfate was performed from SELEX round 2 forward to favor slow off-rates. Partitioning of the target– SOMAmer complexes was achieved with paramagnetic His-tag 2 or Protein G Dynabeads (Life Technologies, Carlsbad, CA) for His-tagged and Fc-fusion proteins, respectively, and with MyOne Streptavidin C1 beads (Life Technologies) for biotinylated targets. Selected DNA was eluted from the beads with buffered sodium perchlorate (1.8 M NaClO 4 , 40 mM PIPES, 1 mM EDTA, 0.05% Triton X-100, pH 6.8) for 5 min, then captured on primer beads and processed for PCR and primer extension to obtain sensestrands with the modified nucleotides using KOD EX DNA polymerase (EMD www.BioTechniques.com
Figure 1A_BT5549
A
Reports
B
Free target
1st
SELEX Random modified ssDNA library #1
25
SOMAmers
Bn
Nap
Competitive
20
SOMAmer pairs
Target−SOMAmer Complex
# sequences
Free target
= Bn Nap Trp PE Tyr 2Nap
Non-competitive
screen
40-nt random 5’ Fixed 3’ Fixed T TT T TTT T T T=
Pool 5551 (free CdtA SELEX) Pool 5579 (CdtA–4758-6 SELEX)
15 10
2nd
SELEX Random modified ssDNA library #2
Trp
SOMAmers
screen
5 0
PE
Tyr
2Nap
Figure 1. Strategy for the isolation and validation of a SOMAmer pair for a protein target. (A) A free target is used in a first SELEX with a modified random ssDNA library to isolate a set of SOMAmers that can be screened directly for pairs of non-competing clones. If no pairs are present, the SOMAmer with the best binding properties is allowed to form a complex with the target, which is then used in a second SELEX with a different modified library. The new SOMAmer clones are then screened for paired sandwich binding to the target. Bn, 5-benzylaminocarbonyl; Nap, 5-naphthylmethylaminocarbonyl; Trp, 5-tryptaminocarbonyl; PE, 5-phenylethyl-1-aminocarbonyl; Tyr, 5-tyrosylaminocarbonyl; 2Nap, 5-(2-naphthylmethyl)aminocarbonyl (B) Representation of sequence patterns (shared motifs) and multicopy sequences (equivalent clones with 100-fold excess (10 nM) of unlabeled primary SOMAmer as competitors. Sandwich filter binding assays Two va riations of f i lter bind ing sandwich assays were performed to obtain 12-point binding curves. The first method involved equilibrium binding of protein and radiolabeled, fulllength detection SOMAmer, followed by partitioning (30 min with intermittent shaking) with specific capture beads that were prepared by immobilizing AB-H primary SOMAmers on SA beads. The second method was based on the formation of tripartite complexes during equilibrium binding of protein, radiolabeled detection SOMAmer (fulllength, non-biotinylated) and excess (10 nM) unlabeled A B-H primar y SOMAmer, followed by partitioning (5 min with intermittent shaking) with SA beads. Both methods yielded comparable results. As controls, all sequences were tested in separate assays where the capture agent was omitted in order to identify non-specific background due to SA bead binding. Multiplexed sandwich screening assays The Luminex platform (Luminex Corporation, Austin, TX) was used for the multiplexed, pair-wise screening of SOM Amers (A B-H 50-mers) in sandwich binding format. Capture bead preparation and binding assays
127
were performed in MSBV N12 filter plates (EMD Millipore) pre-wet with SB17T buffer (SB18T supplemented with 1 mM EDTA). Different types (colors) of LumAvidin Microspheres (Luminex Corporation) were dispensed into separate wells (100,000 beads per well) and washed 3 × 1 min with 180 µL SB17T by vacuum filtration. SOMAmers were heat-cooled, 80 µL of 50 nM stocks of each capture SOMAmer for a given protein target were added to a different bead type, and the plate was shaken (20 min at RT, 1100 rpm) to allow for immobilization. The beads were then washed for 5 min each with 100 µL 50 nM streptavidin and with 10 mM biotin in SB17T, then 4 × 1 min with 180 µL SB17T. All capture beads for each protein were pooled, diluted to 1.7 mL with SB17T, dispensed (50 µL) into the wells of a pre-wet MSBVN12 filter plate, and mixed with 50 µL of 20 nM protein in SB17T supplemented with 1% BSA (SB17TB) or 50 µL SB17TB buffer for controls. After a 30–60 min incubation with shaking at 1100 rpm, the plate was vacuum-washed 2 × 1 min with 180 µL SB17TB and the beads were resuspended in 50 µL SB17TB. Heat-cooled detection SOMAmers were added to individual wells, using 50 µL of 12.5 nM stocks, and incubation was continued for 30 min with shaking. The beads were washed and resuspended as above. As a reporter, 50 µL of 10 µg/mL streptavidin-phycoerythrin (SA-PE) conjugate (Moss, Inc., Pasadena, MD) in SB17TB were added, and the beads were again shaken, washed, www.BioTechniques.com
Reports
and resuspended as above. The plate was read on a Luminex 100 analyzer (time out: enabled 120 s, DD gating: 7500–18000, reporter gain: high PMT).
A
Figure 2C_BT5549
Figure 2B_BT5549
Bmax >50%
Bmax=25-50%
Bmax=10-25%
C
Bmax=50%; light-green, 25%–50%; yellow, 10%–25%; orange, 10
P10643
2888–49
2.71
BndU
7579–67
3.16 (0.07)
2NapdU
1.35
Human
PLG
P00747
4151–6
4.86
BndU
7567–64
13.1 (1.94)
TrpdU
2.70
C. difficile
CdtA
Q9KH42
4758–6
0.86
TrpdU
5579–12
0.97 (0.28)
2NapdU
0.52
Values in parentheses are for the complex of the target with SOMAmer 1 b Determined in Luminex bead-based sandwich assay a
parallel with the same library. Comparative sequence analysis of the 2NapdU clones from affinity-enriched pools 5579 (CdtA-4758–6 complex) and 5551 (free CdtA) revealed clear differences in patterns and abundance, although there were some shared sequence motifs as well (Figure 1B). The new SOMAmers bound the free CdtA protein with affinities ranging from 0.05 nM-14 nM (Supplementary Figure S1) and were also tested in competition assays and in sandwich format together with the existing 4758–6 TrpdU SOM Amer (Supplementar y Figure S2). For 1 clone (5579–12) that possessed affinity comparable to 4758–6 (Kd = 0.67 nM), binding to CdtA was not affected the presence of a >100-fold Figure by 3_BT5549 excess (10 nM) or 4758–6 competitor, or when 4758–6 was used as a capture agent, indicating that the 2 SOMAmers bind to distinct CdtA epitopes (Supplementary Figure S3A). In contrast, binding of
Fraction F r a c tio n B B max m ax
1 .0 0 1.00
0 .7 5 0.75
0 .5 0 0.50
CdtA was much reduced for another clone (5579–11) when 4758–6 was used as a competitor or capture agent, in spite of superior affinity (K d = 0.05 nM), indicating overlapping binding epitopes (Supplementary Figure S3B). Representative sequences from each SELEX were also screened on the Luminex platform, where the first SOMAmer 4758–6 served as the capture agent on beads. Most clones from complex SELEX (5579–7, 5579–10, 5579–12, 5579–21) as well as one clone from free CdtA SELEX (5551–81) confirmed their binding when used as detection agent (Supplementary Figure S4A). Capture and detection agents were interchangeable for 4758–6 and 5579–12, producing similar binding curves in the Luminex sandwich assay (Supplementary Figure S4B). SELEX with the CdtA-4758–6 complex clearly resulted in a higher fraction of sequences useful for sandwich assays in conjunction
ANG PT2 ANGPT2 TSP2 TSP2 C R D L1 CRDL1 M ATN2 MATN2 C7 C7 PLG PLG GPVI GPVI ESAM ESAM
0 .2 5 0.25
0 .0 0 0.00 0 .0 0 0 1 0.0001
0 .0 0 1 0.001
0 .0 1 0.01
0 .1 0.1
1 1
1 0 10
1 00 100
Protein P ro te in (nM) (n M ) Figure 3. Sandwich binding curves obtained in the Luminex assay. The CV risk panel proteins were spiked in buffer and tested with the best-performing SOMAmer pairs identified in the pairwise screening assay listed in Table 1. In all cases, the sequence that had served to form the complex with the target during the sandwich SELEX was used for capture, and a new clone was used for detection. The maximum signals (RFU at Bmax) were 23046 (ANGPT2), 16623 (TSP2), 23349 (CRDL1), 25586 (MATN2), 26000 (C7), 13927 (GPVI), 7103 (PLG), and 3000 (ESAM), respectively.
Vol. 56 | No. 3 | 2014
130
with 4758–6, although free CdtA SELEX also produced a few non-competing SOMAmers (Supplementary Table S1). We next focused on eight human proteins that were promising markers for cardiovascular (CV) risk, including angiopoietin-2 (A NGPT2), thrombospondin-2 (TSP2), chordin-like 1 (CRDL1), matrilin-2 (MATN2), glycoprotein VI (GPVI), endothelial cellselective adhesion molecule (ESAM), complement 7 (C7), and plasminogen (PLG) (18). First, we screened archived SOMAmers with different sequences than the menu SOMAmer used in SOMAscan, and this data mining yielded functional SOMAmer pairs for three of these targets (C7, MATN2, PLG), all with BndU modifications (Supplementary Figure S5). Second, we performed SELEX with target–SOM Amer complexes, employing other modified nucleotide libraries (TrpdU and 2NapdU), and in parallel SELEX with free protein. In all cases, we obtained active clones that bound the protein-SOMAmer complexes, and using protein-SOMAmer complex targets during SELEX coupled with different modified libraries increased the likelihood of finding sandwich candidates. Exceptions were TrpdU clones selected with ESAM-2981–9 complex, which interacted with SOM A mer 2981–9 of the complex. The observed binding of the new clones to the protein-SOM A mer complexes did not prove the existence of a true sandwich, since they might simply displace the primary SOMAmer and bind to the same epitope. To make this distinction, the new clones were further evaluated in competition and sandwich binding assays (Supplementary Tables S2–S9). In addition, we set up a more global, multiplexed sandwich screening method using the Luminex platform (Figure 2A). In brief, each of the clones for a given target was separately immobiwww.BioTechniques.com
Reports
lized on a specific LumAvidin bead type. The beads were then pooled and used for capture of the target, and each of the same clones was used individually for detection. With respect to the 7–16 SOMAmer sandwich candidates for the 8 CV risk panel proteins, this approach reduced the 2-dimensional matrix of pair-wise screening to a single dimension, from 1116 assays (15 2 +14 2 +16 2 +14 2 +9 2 +7 2 +8 2 +7 2) to 90 assays (15+14+16+14+9+7+8+7). The best SOMAmer pairs were rapidly identified by plotting the net signals of the multiplexed Luminex sandwich screening assay as heat maps, as shown as an example for the 16 CRDL1 SOMAmers (Figure 2B). The menu SOMAmer 3362–61 for CRDL1 performed well when used as a capture or detection agent in conjunction with any of the new clones. This is not surprising, since the new clones had been selected to bind the complex of CRDL1 with 3362–61, and in fact, two clones (7575–6, 7575–19) performed better as detection agents than capture agents. In contrast, sandwich assays using one of the new clones (7575–2) with most other new clones produced much lower signals (Figure 2C). As expected, no signals were obtained with the same SOMAmer used for capture and detection. Sandwich binding curves on the Luminex platform were generated for seven of the eight CV risk panel proteins: ANGPT2, TSP2, CRDL1, MATN2, C7, PLG, and GPVI (Figure 3). Without any further assay optimization, the apparent equilibrium binding constants (Kd values) of the new SOMAmers for the protein complexes with the cognate primary SOMAmer ranged from 0.02–2.7 nM, and the lower limits of detection (5% of maximum signal) ranged from 1–100 pM (Table 1). Interestingly, 2 of the C7 SOMAmers (7579–65 and 7579–67) that had been selected with the C7–2888–49 complex showed much better affinity for the complex than for the free protein, with Kd = 0.13 nM versus 11.3 nM for 7579–65 and K d = 0.07 nM versus 3.16 nM, respectively (Table 1 and Supplementary Table S8). It is possible that the primary SOMAmer 2888–49 induces a conformational change of the target protein, thereby creating or improving the epitope for the second SOMAmer. No SOMAmer pairs were obtained for ESAM, which is the most basic (pI = 9.3) of the proteins and the only Fc-fusion we used here. All clones from SELEX with ESAM–2981–9 complex using a TrpdU library contained a Vol. 56 | No. 3 | 2014
common sequence pattern and bound not only the complex, but also the free primary SOMAmer 2981–9, indicating that the counter-selection with 2981–9 to remove such sequences had failed in this case. Other clones from SELEX with ESAM–2981–9 complex or free ESAM using a 2NapdU library (e.g., 7581–41) showed apparent binding of the complex or binding of ESAM in the presence of 2981–9 competitor, but in fact they were out-competing the primary SOMAmer for the same or overlapping binding site (Table 1, Supplementary Table S7). The SOM Amer pairs were also evaluated in plate-based sandwich assays, using biotinylated SOMAmers as capture agents immobilized on streptavidincoated plates, and as detection agents in conjunction with streptavidin-HRP conjugate (Supplementary Figure S6). Target titrations indicated some differences in assay performance compared with the bead-based test, emphasizing the need for future platform-specific assay development. We were successful in isolating SOMAmer pairs for 7 of 8 targets, which is in agreement with our typical success rate of 75%–80% with standard SELEX. Our new approach of applying a second SELEX with target-SOMAmer complexes strongly supports the notion that the use of different types of modified nucleotides enabled this breakthrough in the efficiency of finding separate SOMAmers for distinct epitopes. In our highly multiplexed diagnostic platform (SOMAscan), we have shown that background due to non-specific binding can be reduced by polyanionic competitors such as dextran sulfate as a consequence of differential off-rates between specific and non-specific binding events. Dextran sulfate, when present in excess (1–5 mM), will readily occupy binding sites vacated by SOMAmers with fast off-rates for non-specific binding and prevent them from rebinding to these sites. This feature, combined with the added specificity inherent in two-reagent sandwich-type measurements, may facilitate the development of sandwich assays with greater specificity and potential for higher multiplexing abilities compared with what can be achieved with antibodies. SOMAmer pairs hold promise for the development of specific panels in various areas of medical diagnostics for which a large installed base of instruments is already in place, using traditional sandwich assays or novel ligand proximity assays (19, 20), and also for small molecular sensing platforms (21,22).
132
Author contributions
U.A.O. designed and performed the sandwich SELE X method. L . S.G. conducted the sandwich screening and titration assays on the Luminex platform. L.G. and N.J. provided essential contributions on experimental strategy and manuscript preparation.
Acknowledgments
SOMAmer and SOMAscan are trademarks of SomaLogic, Inc. We thank Jeff Carter, Steve Wolk, Tim Fitzwater, Allison Weiss, Evaldas Katilius, Glenn Sanders, Nick Saccomano, Fintan Steele, and Jim Reed for their contributions regarding DNA synthesis, quality control, sequence analysis, assay development, and graphic design. No external funding was obtained for this work.
Competing interests
The authors are employees and shareholders of Soma L og ic , I nc. a nd co-inventors of patents relating to SOMAmer technology.
References
1. Marx, V. 2013. Calling the next generation of affinity reagents. Nat. Methods 10:829833. 2. Zichi, D., B. Eaton, B. Singer, and L. Gold. 2008. Proteomics and diagnostics: Let’s Get Specific, again. Curr. Opin. Chem. Biol. 12:78-85. 3. Perlee, L., J. Christiansen, R. Dondero, B. Grimwade, S. Lejnine, M. Mullenix, W. Shao, M. Sorette, et al. 2004. Development and standardization of multiplexed antibody microarrays for use in quantitative proteomics. Proteome Sci. 2:9. 4. Brody, E.N., L. Gold, R.M. Lawn, J.J. Walker, and D. Zichi. 2010. High-content aff init y-based proteomics: u n lock ing protein biomarker discovery. Expert Rev. Mol. Diagn. 10:1013-1022. 5. Gold, L., D. Ayers, J. Bertino, C. Bock, A. Bock, E.N. Brody, J. Carter, A.B. Dalby, et al. 2010. Aptamer-based multiplexed proteomic technology for biomarker discovery. PLoS ONE 5:e15004. 10.1371/ journal.pone.0015004. 6. Gupta, S., D. Thirstrup, T.C. Jar vis, D.J. Schneider, S.K. Wilcox, J. Carter, C. Zhang, A. Gelinas, et al. 2011. Rapid histochemistry using slow off-rate modified aptamers with anionic competition. Appl. Immunohistochem. Mol. Morphol. 19:273278. 7. Ochsner, U. A ., E . K ati l ius, a nd N. Janjic. 2013. Detection of Clostridium difficile toxins A, B and binary toxin with slow off-rate modified aptamers. Diagn. Microbiol. Infect. Dis. 76:278-285. 8. Park, J.W., S. Jin Lee, E.J. Choi, J. Kim, J.Y. Song, and M. Bock Gu. 2014. An ultra-
www.BioTechniques.com
Reports
sensitive detection of a whole virus using dual aptamers developed by immobilization-free screening. Biosens. Bioelectron. 51:324-329. 9. Tasset, D.M., M.F. Kubik, and W. Steiner. 1997. Oligonucleotide inhibitors of human thrombin that bind distinct epitopes. J. Mol. Biol. 272:688-698. 10. Shi, H., X. Fan, A. Sevilimedu, and J.T. Lis. 2007. RNA aptamers directed to discrete functional sites on a single protein structural domain. Proc. Natl. Acad. Sci. USA 104:3742-3746. 11. Gong, Q., J. Wang, K.M. Ahmad, A.T. Csordas, J. Zhou, J. Nie, R. Stewart, J.A. Thomson, et al. 2012. Selection strategy to generate aptamer pairs that bind to distinct sites on protein targets. Anal. Chem. 84:5365-5371. 12. Xiao, S.J., P.P. Hu, X.D. Wu, Y.L. Zou, L.Q. Chen, L. Peng, J. Ling, S.J. Zhen, et al. 2010. Sensitive discrimination and detection of prion disease-associated isoform with a dual-aptamer strategy by developing a sandwich structure of magnetic microparticles and quantum dots. Anal. Chem. 82:9736-9742. 13. Nonaka, Y., K. Sode, and K. Ikebukuro. 2010. Screening and improvement of an anti-VEGF DNA aptamer. Molecules 15:215-225. 14. Sosic, A., A. Meneghello, E. Cretaio, and B. Gatto. 2011. Human thrombin detection through a sandwich aptamer microarray: interaction analysis in solution and in solid phase. Sensors (Basel). 11:9426-9441. 15. Tennico, Y.H., D. Hutanu, M.T. Koesdjojo, C.M. Bartel, and V.T. Remcho. 2010. On-chip aptamer-based sandwich assay for thrombin detection employing magnetic beads and quantum dots. Anal. Chem. 82:5591-5597. 16. Davies, D.R., A.D. Gelinas, C. Zhang, J.C. Rohloff, J.D. Carter, D. O’Connell, S.M. Waugh, S.K. Wolk, et al. 2012. Unique motifs and hydrophobic interactions shape the binding of modified DNA ligands to protein targets. Proc. Natl. Acad. Sci. USA 109:1997119976. 17. Vaught, J.D., C. Bock, J. Carter, T. Fitzwater, M. Otis, D. Schneider, J. Rolando, S. Waugh, et al. 2010. Expanding the chemistry of DNA for in vitro selection. J. Am. Chem. Soc. 132:4141-4151. 18. Gill, R.D., S. Williams, A. Stewart, R. Mehler, T. Foreman, and B. Singer. 2012. Cardiovascular risk event prediction and uses thereof, p. PCT/US2012/058060. SomaLogic, Inc., USA. 19. Fredriksson, S., M. Gullberg, J. Jarvius, C. Olsson, K. Pietras, S.M. Gustafsdottir, A. Ostman, and U. Landegren. 2002. Protein detection using proximity-dependent DNA ligation assays. Nat. Biotechnol. 20:473-477. 20. Hu, J., T. Wang, J. Kim, C. Shannon, and C.J. Easley. 2012. Quantitation of femtomolar protein levels via direct readout with the electrochemical proximity assay. J. Am. Chem. Soc. 134:7066-7072. 21. Jo, M., J.Y. Ahn, J. Lee, S. Lee, S.W. Hong, J.W. Yoo, J. Kang, P. Dua, et al. 2011. Development of single-stranded DNA aptamers for specific Bisphenol A detection. Oligonucleotides 21:85-91. 22. Lee, J., M. Jo, T.H. Kim, J.Y. Ahn, D.K. Lee, S. Kim, and S. Hong. 2011. Aptamer sandwich-based carbon nanotube sensors for singlecarbon-atomic-resolution detection of non-polar small molecular species. Lab Chip 11:52-56.
Fast & Gentle Concentrator Vivaspin® Turbo 15 Concentrate your protein gently with the fastest tool on the market: Vivaspin® Turbo 15 Differentiation by design
Received 20 October 2013; accepted 24 January 2014. Address correspondence to Urs A. Ochsner, SomaLogic, Inc., Boulder. E-mail: [email protected] To purchase reprints of this article, contact: [email protected]
Vol. 56 | No. 3 | 2014
www.sartorius.com/vivaspin-turbo www.BioTechniques.com
