Routine Characterization of DNA Aptamer Affinity to Recombinant Protein Targets Ilavarasi Gandhi, Research Assistant, Base Pair Biotechnologies, Inc., George W

Interactions • SPRING 2012

Routine Characterization of DNA Aptamer Affinity to Recombinant Protein Targets Ilavarasi Gandhi, Research Assistant, Base Pair Biotechnologies, Inc., George W. Jackson, Ph.D., Founder and Chief Scientist, Base Pair Biotechnologies, Inc.

Nucleic acid aptamers are high affinity, ing to a variety of targets such as proteins, tion of the protein itself requires less total high-selectivity ligands produced in vitro by a peptides, and even small molecules with affin- protein, but immobilization (primarily through process commonly known as SELEX . While the ity and specificity rivaling that of antibodies . lysine residues) may be perturbing to protein selection of DNA and RNA aptamers has been They are typically selected in vitro by a process epitopes and may require protein-to-protein described for some time, the SELEX process commonly referred to as SELEX 1, 2 . The initial optimization depending on the nature of was traditionally performed against a single randomized library applied to an immobilized the target . Ultimately, both approaches are target at a time requiring weeks to months target comprises approximately 1015 unique complementary and we therefore present for successful execution . We have developed oligonucleotide sequences of 30–40 bases in facile protocols for each below . a proprietary process for multiplexing SELEX length bracketed by constant regions for PCR to discover aptamers against multiple targets priming . The output of the process is therefore METHOD 1: APTAMER simultaneously, thereby greatly increasing the several (5–20) clonal DNA sequences likely to IMMOBILIZATION throughput of the process . Our aptamer dis- have specific affinity for the target molecule . covery services include validation of aptamer As mentioned above, a quantitative valida- We have developed two methods for ready binding by characterizing the aptamer:target tion of the binding of such clones is of critical aptamer immobilization to ForteBio’s Dip and Read biosensors: dissociation constant (kd) before delivery of importance to our process and business . aptamer materials to the customer for further • Direct immobilization of biotinylated DNA testing . With an increasing customer base and GENERAL WORKFLOW on streptavidin biosensors . a longer list of targets, it becomes increasingly In general, we can take two approaches to important to implement higher throughput • Biotinylated polyA capture method of affinity characterization of our aptamer prod- methods for aptamer validation . The ForteBio aptamer immobilization ucts — either immobilizing the aptamer or the Octet platform provides a fast, simple, and target itself . Each approach may have specific In the first method of aptamer immobiliza- cost-effective means to characterize multiple advantages and disadvantages . Immobiliza- tion, a DNA clone is appended at the 5'- or aptamer clones at a throughput that meets tion of the aptamer itself allows a modular 3'-end and offered directly to ForteBio’s our customers’ demands . approach in which each of our aptamers is Streptavidin Biosensor . For most aptamers treated identically . In other words, there are no we have observed minimal perturbation or Intro to Aptamers protein-specific immobilization conditions to detrimental effect on binding due to such im- optimize . This approach has the disadvantage, mobilization . Indeed, we can choose aptamer Aptamers are single-stranded DNA or RNA however, of requiring additional protein mate- clones based on secondary structure that oligonucleotides selected to have unique rial when offering multiple protein analyte should be minimally affected by tethering at three-dimensional folding structures for bind- concentrations to the biosensor . Immobiliza- either end .

0.14

BaselineSELEX in bu er 1μM hCGaptamer-biotin BaselineSELEX in bu er hCG portein Dissociationin SELEX bu er 0.13 0.10 0.12 0.05 0.11 0.00 0.10 -0.05 0.09 -0.10

-0.15 0.08

-0.20 0.07

-0.25 0.06 Binding, nm

Binding, nm -0.30 0.05

-0.35 A9 0.04 B9 -0.40 C9 0.03 -0.45 D9 0.02 E9 -0.50 0.01 -0.55 0.00 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 2,000 0 100 200 300 400 500 600 700 800 900 1,000 1,100 1,200 T ime (sec) T ime (sec) FIGURE 1: Kinetic assay set up for direct immobilization of FIGURE 2: Processed kinetic data for 1 µM #387 hCG aptamer-biotin and hCG biotinylated DNA to streptavidin biosensors . protein analyte showing overlaid fits with KD = 56 .6 nM .

4 Interactions • SPRING 2012

Direct Immobilization of Biotinylated DNA to Streptavidin Biosensors (Single Reference Well) on Octet RED96 System

Blocking with 1% NFDM Baseline in SELEX buffer 0.60 Baseline in SELEX buffer 1 μM 5’biotin polyA Baseline in SELEX buffer 10 μM Hb aptamer with 3’14T Baseline in SELEX buffer Association with Hb protein Dissociation in SELEX buffer Sample Plate Preparation 1 Prepare a 96-well plate with required 0.40 ligands, analytes and wash buffers .

0.20 2 All steps in the following example (Figures 1–2) were performed with a 0.00 shake speed of 1000 rpm . Optimal shake speeds may vary . -0.20 Prepare and Load Biosensor Surface with -0.40 Biotinylated Aptamer Binding, nm 1 Place the biosensors in the SELEX buffer -0.60 and equilibrate for 1 minute .

-0.80 2 Load the biosensors with 1 μM biotinyl- A5 B5 ated hCG aptamer for 10 minutes . -1.00 D5 3 Perform a wash/baseline step in SELEX buffer for 1 minute . -1.20 0 600 1,200 1,800 2,400 3,000 3,600 4,200 4,800 5,400 T ime (sec) Assess Association with Analyte Target Figure 3: Kinetic assay set up for immobilization of aptamer with biotinylated polyA approach . Protein 1 After the baseline step, place the biosensors in a dilution series of hCG In the second method of aptamer immobi- biosensor preparation . Finally, this configura- protein for 10 minutes . For some analytes lization, a DNA aptamer clone is appended tion exactly matches some of our customers’ with a fast binding rate, even 3–5 minutes with a 14-mer poly-thymidine sequence at bead-based applications, allowing for large of association time will be enough to the 5'-end . This allows for hybridization to a batches of polyA beads to be prepared before perform kinetic analysis . biosensor pretreated with a biotin-polyA . The aptamer-specific functionalization . 2 Place the biosensors in the SELEX buffer advantage of this approach is further steric for 10 minutes for dissociation . If inter- spacing of the aptamer from the biosensor step correction is required, the same well surface and, again, a modular approach to used for baseline should also be used for dissociation .

A B 0.80 0.80

0.70 0.70

0.60 0.60

0.50 0.50

0.40 0.40 Binding, nm Binding, nm 0.30 0.30

0.20 0.20 1:1 binding 1:2 (bivalent K = 25.6 nM analyte) binding 0.10 D 0.10 KD = 12.3 nm

0.00 0.00 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 Time (sec) T ime (sec)

FIGURE 4: Processed kinetic data showing overlaid fit and KD values . A) Load 1: 1 µM load of 5'biotin-polyA (14A) . Load 2: 10 µM 3'14T gly Hb aptamer 72H09 #439 . Non-gly Hb protein at 1 and 0 .5 µM analyte . B) Load 1: 5'biotin-polyA . Load 2: 3'14T #439 Hb aptamer 72 H09 . Hb protein analyte .

5 Interactions • SPRING 2012

Equilibratein water EDC/NHSActivation LoadingHsp27 Protein Quench1 withM Tris, pH 8.0 Wash inSELEX bu er AssociationHsp27 with Aptamer LoadingStreptavidin 1:2000 HRP LoadingSubstrate 1X TMB

4 Binding, nm 0

-4

-8

-12

0 300 600 900 1,200 1,500 1,800 2,100 2,400 2,700 3,000 3,300 3,600 3,900 4,200 4,500 4,800 5,100 5,400 5,700 6,000 6,300 T ime (sec)

FIGURE 5: Kinetic assay set up for immobilization of protein to AR2G biosensors .

Biotinylated PolyA Capture Method 5 Perform a wash/baseline step in SELEX The reference well is subtracted from the for Aptamer Immobilization (Single buffer . analyte wells for buffer artifacts . Then y-axis Reference Well) on Octet RED96 System alignment, inter-step correction and Savitzky- 6 Load the biosensors with 10 μM Hb ap- Golay filtering are also applied to the data . Sample Plate Preparation tamer having 3'14T, for 10 minutes . 1 Prepare a 96-well plate with required 7 Wash the biosensors with SELEX buffer . The processed data is then allowed to fit a curve ligands, analytes and wash buffers . for association and dissociation using 1:1 model Assess Association with Analyte Target fitting with either global or local fitting . 2 All steps in the following example Protein (Figures 3–4) were performed with a shake speed of 1000 rpm . Optimal shake 1 After the baseline step, place the biosen- METHOD 2: PROTEIN speeds may vary . sors in a dilution series of Hb protein for IMMOBILIZATION 10 minutes . For analytes with a fast bind- For protein immobilization we use Forte- ing rate, even 3–5 minutes of association Prepare and Load Biosensor Surface with Bio’s Dip and Read AR2G biosensors . In time will be enough to perform kinetic Biotinylated PolyA Oligo and PolyT Aptamer this particular example, hsp27, a protein analysis . 1 Place the biosensors in the SELEX buffer implicated in breast cancer3, was covalently and equilibrate for 1 minute . 2 Place the biosensors in SELEX buffer for 10 immobilized to amine-reactive second gen- minutes for dissociation . If inter-step correc- 2 Block the biosensor surfaces with 1% eration (AR2G) biosensors using standard tion is required, the same well used for base- NFDM for 300 seconds . EDC/NHS coupling . Four different concen- line should also be used for dissociation . trations were immobilized on different bio- 3 Wash the biosensors in SELEX buffer for sensors . A novel biotinylated aptamer clone ~600 seconds to ensure unbound NFDM Data Processing selected by Base Pair Biotechnologies was is removed . The data obtained is processed to determine then applied to each biosensor at 1 μM . 4 Load the biosensors with 1 μM 5' biotinyl- To obtain additional signal, the biosensors the overlaid fits and the KD, kon and koff values . ated polyA oligo for 10 minutes . were then briefly washed in buffer and sub- sequently dipped in an equal concentration of streptavidin-HRP conjugate .

6 Interactions • SPRING 2012

0.8 4.8 R2 = 0.9648 4.4 log 10 0.6 4.0 Log10 [conc] vs log binding rate 0.4 3.6 10

3.2 0.5

2.8 0.0 2.4 -4 -3 -2 -1 0 1 2 3

Binding, nm 2.0 -0.2 1.6 -0.4 1.2

0.8 -0.6 0.4

0.0 -0.8 0 100 200 300 400 500 600 700 800 900 1,000 1,100 1,200 T ime (sec) FIGURE 6: Quantitating the streptavidin–HRP step of kinetics assay . FIGURE 7: Standard curve of log hsp27 concentration vs . streptavidin-HRP 1:2000 streptavidin-HRP conjugate by known concentration (µg/mL) . binding rate .

Protein Immobilization on AR2G Associate Aptamer to the Immobilized considerable satisfaction . Because the Forte- Biosensors on Octet RED96 System Protein Bio instrument does not employ a flow cell, 1 Load 1 µM hsp27 aptamer (Oligo #410) we do not experience the instrument down Sample Plate Preparation with 3'biotin for 300 seconds . time associated with microfluidic clogging, 1 Prepare a 96-well plate with required etc . Additionally, the use of largely modular ligands, analytes and wash buffers . Assess Association with Streptavidin-HRP workflows requiring little protein-to-protein 2 All steps in the following example and TMB Substrate optimization have allowed us to address (Figures 5–7) were performed at a shake 1 After loading aptamer, add 1:2000 increasing customer demand for our aptamer speed of 1000 rpm . Optimal shake streptavidin-HRP enzyme conjugate for discovery services . speeds may vary . 1200 seconds . 2 Load the final step, 200 µL 1X TMB References Prepare and Load Biosensor Surface with substrate, in each of the wells for 1200 1 Tuerk C, Gold L: Systematic evolution of Protein seconds . ligands by exponential enrichment: RNA li- 1 Place the biosensors in water and equili- gands to bacteriophage T4 DNA polymerase. brate for 30 seconds . Data Processing Science 1990, 249:505–10. 2 Load 20 mM EDC and 10 mM NHS for Taking the initial slope of the response (Forte- 2 Ellington AD, Szostak JW: In vitro selection 300 seconds . Bio’s usual approach) results in a log-log of RNA molecules that bind specific ligands. 3 Load a dilution series of hsp27 protein standard curve . As can be seen in Figure 7, Nature 1990, 346:818–822. the linear fit is quite good (R2 = 0 .96) . Good (0 .2 µM, 2 nM, 20 pM, 0 .2 pM) in 10 mM 3 Kang SH, Kang KW, Kim K-H, Kwon B, Kim signal-to-noise was obtained at an hsp27 acetate buffer, pH 6 .0, for 1200 seconds . S-K, Lee H-Y, Kong S-Y, Lee ES, Jang S-G, Yoo concentration of 0 .2 pM which compares Since the amount of protein loading is BC: Upregulated HSP27 in human breast can- favorably with even the best ELISAs . For ad- low, the loading time is increased . cer cells reduces Herceptin susceptibility by ditional signal and sensitivity, we can readily 4 Quench the remaining unreacted surface increasing Her2 protein stability. BMC Cancer add a standard HRP substrate . with 1 M Tris, pH 8 .0, for 300 seconds . 2008, 8:286.

5 Move the biosensors into SELEX running Download Application Note 5 for CONCLUSION buffer (20 mM Tris, 100 mM NaCl, 0 .005% more experimental details, reagent Tween20, pH 7 .4) for 1800 seconds . In a short amount of time we have adapted requirements and operational tips at our laboratory workflow from surface plas- www .fortebio .com/literature .html mon resonance (SPR) to BLI analysis with

7 Interactions • SPRING 2012

BioLayer Interferometry (BLI) – How Does it Work? Darick Dayne, Ph.D., Product Manager, Marketing, [email protected]

Two waves in phase produces Two waves 180° out of phase constructive interference produceForteBio destructive introduced interference its first BLI breakthrough Two waves in phase produces Two waves 180° out of phase constructive interferencein 2005 with the Octet instrument,produce destructive interference and one Re ected Light question we’ve been asked quite often since White Light Re ected Light White then is “how does it work”? So we thought Light it was high-time to answer that question for a a Constructive interference a a BLI biosensor Constructive both new and experienced users alike . interference tip surface BLI biosensor b b tip surface b b The technology employed by the Octet Partially Partially Optical layer Biocompatible constructive platform is based on the principles of optical surfaceOptical layer Biocompatible interferenceconstructive interference interferometry, namely the interaction of surface Immobilized light waves . The basic phenomenon is quite Immobilized DestructiveDestructive molecules molecules interference simple . When two propagating waves are interference perfectly in phase (e .g . the peaks and troughs of the waves match up exactly), the resulting Figure 2 Figure 3 Figure 4 wave has an amplitude equal to the sum of the two waves due to constructive interfer- Octet or BLItz platforms, white light (inclusive plitudes of the red, green, violet and all other ence (Figure 1A) . When the two propagating of all wavelengths in the visible spectrum) wavelengths were plotted, an interferometric waves are completely out of phase (e .g . the is sent down the glass fiber (Figure 3) and is profile would be derived (Figure 7) . This forms peaks of one wave matches up exactly with reflected back up to the instrument from two the basis of BLI measurements on the Octet the trough of the other), the resulting wave interfaces: 1) the interface between the glass and BLItz instruments . will have zero amplitude due to destructive fiber and the proprietary bio-compatible When molecules bind to the surface of the interference (Figure 1B) . layer, and 2) the interface between the surface chemistry and solution (Figure 4) . biosensor, the path length of the reflection (the one reflecting from the interface

Two waves in phase produces Since theTwo two waves reflections 180° out of come phase from the same between surface and solution) increases A constructive interference white lightproduce source destructive in the interference instrument, they while that of the other reflection remains the both contain the same wavelengths in the same (Figure 8) . This changes the interference visible spectrum (Figure 5) . When the two patterns for all wavelengths . In this example, Re ected Light reflections are taken apart and the same color the two waves of the red wavelength from White Light channel (e .g . wavelength) from each reflec- the two reflections are no longer perfectly tion is monitored and analyzed together, an in phase, which causes the slight drop in the a interferencea pattern emerges . In the example resulting amplitude . The two waves of the Constructive interference shown in Figure 6, the two red light waves green wavelengthBLI biosensorare further out of phase, tip surface b might bundergo complete constructive inter- causing a further drop in the resulting ampli- ference, resulting in a doubling of amplitude . tude . However, the two waves of the violet Partially The two green light waves might undergo wavelength are no longer completely out of Two waves in phase produces Two waves 180° out of phase Optical layer Biocompatible constructive constructive interference B produce destructive interference partially constructive interference, resulting phase and thus the resulting amplitude is no surface interference in a certain amplitude smaller than that of the longer zero (Figure 9) . Plotting this for all the red channel . The two violet light waves might wavelengths results in a new interferometric Immobilized Re ected Light Destructive undergo complete destructive interference, profile that is shifted to the right from the moleculesWhite interference Light Two waves in phase produces Two waves 180° out of phase resulting in zero amplitude . If the relative am- constructive interference produce destructive interference original profile (Figure 10) .

Re ected Light Constructive a White a interference Light BLI biosensor tip surface b b a a Constructive interference BLI biosensor tip surface Partially b b Figure 1 Optical layer Biocompatible constructive Partially surface interference Optical layer Biocompatible constructive interference surface Figure 2 shows a magnified view of a ForteBio glass fiber-based biosensor, where the Immobilized Destructive Immobilized moleculesDestructive molecules surface chemistry occurs at the very tip of interference interference the glass fiber . When the biosensor is in the Figure 5 Figure 6

8 Interactions • SPRING 2012

1.0 Molecules bind, Δλ spectral shift due to change in thickness creating a thicker surface { 100% 100% 0.8 t f

0.6 i s h

50% 50% m n

Relative Intensity Relative 0.4 Relative Intensity Relative Relative Intensity Relative

0.2 0% 0%

Wavelength (nm) Wavelength (nm) Wavelength (nm) Time

Figure 7 Figure 8 Figure 9

1.0 Molecules bind, Δλ spectral shift due to change in thickness creating a thicker surface { 100% 100% 0.8 t f

0.6 i s h

50% 50% m n

Relative Intensity Relative 0.4 Relative Intensity Relative Relative Intensity Relative

0.2 0% 0%

Wavelength (nm) Wavelength (nm) Wavelength (nm) Time Figure 10 Figure 11 Figure 12

This optical construct allows the real-time time and its magnitude plotted as a function market today require samples to flow across a monitoring of molecular binding events of time, a classic association/dissociation sensor in finicky and hard-to-maintain micro- occurring on the surface of the biosensor . curve is obtained (Figure 12) . fluidic systems that are also prone to frequent As more molecules bind to the surface, the breakdowns . However, Octet and BLItz plat- The Octet and BLItz platform’s superior tech- interferometric profile shifts further to the forms create flow without microfluidics and nology harnesses the real-time label-free de- right (Figure 11) . Conversely, as molecules place biosensors into samples directly for true tection power of BLI in a manner that makes dissociate from the surface, the interferomet- Dip and Read simplicity, enabling scientists them simple to use and easy to maintain for ric profile shifts back left to its initial position . to accelerate their research by maximizing researchers in a variety of laboratory settings . When this shift is measured over a period of productivity . Almost all surface-based technologies on the

Technical Tip: Measuring High-Affinity Interactions on the Octet System Darick Dayne, Ph.D., Product Manager, Marketing, [email protected]

As more drug candidates are affinity-ma- the other binding partner is in solution (the the 1x10-12 molar range indicate very strong tured through the development process, analyte) . When the interaction between the interactions . The challenge of measuring a accurately measuring their affinity becomes target and the analyte is expected to be very very high affinity interaction arises from the a crucial part of characterization . In this strong (e .g . very high affinity), precautions slow dissociation of target-bound analyte article, we discuss some tips and tricks that must be taken to ensure the kinetic mea- back into solution . For interactions with KD can help Octet and BLItz platform users surement is set up and performed correctly . values in the picomolar range, the dissocia- measure high-affinity molecular interactions tion is so slow that the slope of the dissocia- The two kinetic terms that describe affin- more successfully . tion curve is extremely small . To measure ity of an interaction are the association and fit this slope accurately for an off-rate There are always two binding partners constant (k ) and dissociation constant (k ) . a d determination, some necessary precautions involved in any kinetic measurement, and The ratio of these two terms (k / k ) give d a should be taken . the essence of the measurement is to char- rise to the affinity constant KD, which is acterize how strongly the two molecules used for systematic comparison of affinity • Ensure that the slope only comes from the interact . Using a surface-based technology between different molecular interactions . dissociation of the target-bound analyte platform such as the Octet system, one of In the overall affinity spectrum, KD values and nothing else . Signal drift of any kind -3 the two binding partners is immobilized generally in the 1x10 molar range or above during measurement, whether a result onto the biosensor surface (the target) while indicate very weak interactions . K values in D continued on page 10

9 Interactions • SPRING 2012

Technical Tip Measuring High-Affinity Interactions on the Octet System continued from page 9

of the surface-bound target detaching tion, start with a concentration that is • Binding capacity of the assay and thus the

from the biosensor surface or other buf- ~10–20X above the expected KD value resulting signal response of association fer/detergent effects, should be strictly and titrate down by 2–3X . For example, if should be maximized so that the drop in

minimized or controlled well enough to the expected KD value of the interaction is signal during dissociation is still greater be corrected during analysis (i .e . reference 10 pM, then the concentration series of the than the instrument background noise . subtraction) . analyte would be 200 pM, 100 pM, 50 pM, For users of the Octet RED96/384 instru- 25 pM, and so forth . ments the average background noise is • The method of target immobilization approximately 0 .005 nm . Higher signal should be firm and non-reversible, ide- • The dissociation step should be sub- response during the association step will ally done either through amine coupling stantially lengthened depending on the always increase the reliability of the statis- chemistry with the AR2G Biosensor or expected K value of the interaction being D tically calculated dissociation constant k through biotin-avidin binding with the measured . For interactions with K values d D and therefore the accuracy of the affinity Streptavidin Biosensor . in the sub-nanomolar range, dissociation constant K . steps longer than one hour are generally D • The measurement should include a refer- required . ence biosensor immobilized with the same Summary target molecule but taking measurement • The stronger the interaction, the longer in running buffer during analyte associa- it will take to measure the dissociation Biomolecules with very high binding af- tion/dissociation steps to account for any step . Figure 1 shows a reference-corrected finities are becoming more common in drift . This reference measurement can be kinetic measurement of a very high-affinity biopharmaceutical R&D, and their accurate used for subtraction during data analysis interaction, where the dissociation is kinetic determination is very important in to eliminate any drift due to background so tight that a dissociation time of 900 drug discovery . These technical tips will effects . seconds is insufficient to observe any ap- enable researchers to accurately and easily preciable drop in signal . In such cases, a measure high-affinity molecular interactions • When measuring a titration series of much longer dissociation such as one hour while also taking advantage of the stream- the analyte for a rigorous K determina- D (3600 seconds) or more is needed . lined workflow and ease-of-use provided by the Octet and BLItz platforms .

2.0

1.8

1.6

1.4

1.2

1.0

0.8 Binding, nm

0.6

0.4

0.2

0.0 0 200 400 600 800 1,000 1,200 1,400 1,600 T ime (sec)

Figure 1: Kinetic measurement of a high-affinity interaction in which the detachment of the analyte from the surface-bound target is so slow that the off-rate cannot be reliably determined with a 900-second dissociation step .