Nanopore-based kinetics analysis of individual antibody-channel and antibody-antigen interactions
- 3.8k Downloads
The UNO/RIC Nanopore Detector provides a new way to study the binding and conformational changes of individual antibodies. Many critical questions regarding antibody function are still unresolved, questions that can be approached in a new way with the nanopore detector.
We present evidence that different forms of channel blockade can be associated with the same antibody, we associate these different blockades with different orientations of "capture" of an antibody in the detector's nanometer-scale channel. We directly detect the presence of antibodies via reductions in channel current. Changes to blockade patterns upon addition of antigen suggest indirect detection of antibody/antigen binding. Similarly, DNA-hairpin anchored antibodies have been studied, where the DNA linkage is to the carboxy-terminus at the base of the antibody's Fc region, with significantly fewer types of (lengthy) capture blockades than was observed for free (un-bound) IgG antibody. The introduction of chaotropic agents and its effects on protein-protein interactions have also been observed.
Nanopore-based approaches may eventually provide a direct analysis of the complex conformational "negotiations" that occur upon binding between proteins.
KeywordsHide Markov Model Capture Event Channel Blockade Hide Markov Model Profile Murine IgG1
The highly stable Alpha-Hemolysin protein channel
The alpha-Hemolysin toxin is produced by the bacteria Staphylococcus aureus. The alpha-Hemolysin channel is a heptamer, a seven member molecular complex. Each alpha-hemolysin monomer is water soluble and on the membrane surface these monomers self assemble, in an ATP-independent process, into the functional heptamer geometry. The oligomerization that completes the formation of the heptamer provides the energy to punch through the membrane to form the highly stable alpha-Hemolysin channel. From crystallographic results , we know that the alpha-hemolysin water filled channel ranges in diameter from 2.6 nm at the cis-side opening to 1.5 nm at the limiting aperture. The length of the channel along its line of axial symmetry is approximately 10 nm. The channel widens in the middle creating a chalice shaped cross section along its axis. This channel widening provides a cavity for a captured molecule to wiggle about. Many different molecules have been examined on the nanopore detector platform, including biopolymers like ssDNA, dsDNA, ethylene glycol, and a variety of sugars and proteins (see Background for more details).
Previous nanopore detector measurements involving hairpin DNA molecules with varying base stem lengths have shown a relationship between the number of base pairs and the occurrence of a bi-level dominated current signal or "toggle signal" . These experiments also serve to directly confirm the channel geometry described above, where the DNA hairpins can be viewed as "depth gauges" of varying length. A model for the mechanism of the toggle signal, that is observed for 9 base pair DNA hairpins, is proposed as an interaction between the terminus of the DNA hairpin stem and the limiting aperture's border amino acids (see ). Upon introduction of antibodies to the same system, similar blockage signals have been observed suggesting a similar mechanism is responsible for the antibody toggle signal.
Nanopore blockade detector
The original and prevailing method of characterizing DNA oligonulceotides is based on analyzing the depth and duration of the static channel blockade created by ssDNA freely passing, also referred to as "translocating," through the channel . The method employed in this study and similar studies is quite different in that the shape of our specially designed dsDNA molecules makes them unable to fully translocate, and thus, the blockade signal produced corresponds to that of a (partially) trapped, non-translocating molecule [2, 3, 4, 5, 6, 7, 9, 10]. The direct approaches offer prospects for DNA sequencing (via translocation observations [8, 11, 12, 13, 14, 15, 16, 17, 18, 19]) and single nucleotide polymorphism (SNP) analysis (via non-translocation observations [2, 3, 4, 5, 6, 7]). In one direct study of molecular event statistics , the binding of an individual DNA oligonucleotide, covalently tethered within the lumen of the alpha-hemolysin pore to free-floating ssDNA, caused changes in the ionic current flowing through a nanopore that allowed discrimination between individual DNA strands up to 30 nucleotides in length. This is a very brief and limited synopsis of the Nanopore Detector background relevant to this paper. For other references on Nanopore Detectors use is made of a Nanopore Detector review presented in : early work involving alpha-Hemolysin Nanopore Detectors can be found in [2, 3, 4, 5, 6, 7, 11, 13, 14, 15, 16, 17, 18, 19]; rapidly growing research endeavors on Nanopore Detectors based on solid-state, and other synthetic, platforms can be found in [20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30].
A recent example of indirect statistical observation, the biologically relevant interactions between ssDNA and a protein, the bacterial enzyme Exonuclease I, were examined . If the free-floating ssDNA bound the enzyme in solution, it then formed a complex too large to pass through the channel. The ssDNA remained in the channel until it finally detached from the enzyme. The time it takes for the bonds between the DNA and the enzyme to break informs about the nature of their kinetic interaction. Contrast the  indirect kinetic study with the direct kinetic study done by , where kinetic features are directly measured in the sense that the molecular event presented there directly modulates the channel environment flow, and with  it does so by indirectly releasing the molecule and allowing it to translocate.
The immunoglobulin molecule IgG is often described as a bifunctional molecule: one region for binding to target antigen, the other region for mediating effector function. Effector functions include binding of the antibody to host tissues, to various cells of the immune system, to some phagocytic cells, and to the first component (C1q) of the classical complement system. Activation of the immune system in response to a specific antigen is an amazing example of how a series of protein phosphorylation and dephosphorylation reactions convert a cell surface event to changes in DNA transcription and cell replication.
The structure of the IgG antibody forms three globular regions that are attached to each other in the middle of its grouping (see image of IgG embedded in Fig. 2). The overall shape of the structure forms a Y configuration. At the base of this structure is the Constant (Fc) region where the effector functions take place and at the tips of the two arms, both referred to as the variable region (Fab), are the antigen binding sites. These variable regions are tethered to the trunk of the Y shaped molecule by a flexible hinge which allows for a high degree of arm movement. The relative size of the antibody is about three times the size of the alpha-hemolysin channel. Its length from base (Fc) to arm tip (Fab) is 25 nm and the width of each globular arm ranges from 6–10 nm [34, 35, 36, 37].
The forces binding antigen to antibody are an important and difficult area of study. Hydrophobic bonds, in particular, are very difficult to characterize by existing crystallographic and other means, and often contribute half of the overall binding strength of the antigen-antibody bond. Hydrophobic groups of the biomolecules exclude water while forming lock and key complementary shapes. The importance of the hydrophobic bonds in protein-protein interactions, and of critically placed waters of hydration, and the complex conformational negotiation whereby they are established, may be accessible to direct study using nanopore detection methods in future developments of this technology. In the preliminary work that follows, however, we focus on the recognition of a binding event, not the nature of that binding event (future work to decipher the binding mechanism will require far more extensive data gathering and analysis and will not be discussed further).
Ab-antigen binding and Ab-pore interaction
Channel current cheminformatics
The signal processing architecture (Fig. 1, Bottom Panel) is designed to rapidly extract useful information from noisy blockade signals using feature extraction protocols, wavelet analysis, Hidden Markov Models (HMMs) and Support Vector Machines (SVMs). For blockade signal acquisition and simple, time-domain, feature-extraction, a Finite State Automaton (FSA) approach is used  that is based on tuning a variety of threshold parameters. A generic HMM can be used to characterize current blockades by identifying a sequence of sub-blockades as a sequence of state emissions [2, 3, 4, 5, 6, 7]. The parameters of the generic HMM can then be estimated using a method called Expectation/Maximization (or 'EM") , to effect de-noising. The HMM method with EM, denoted HMM/EM, is used in what follows (further Background on these methods can be found in [2, 3, 4, 5, 6, 7]). Classification of feature vectors obtained by the HMM for each individual blockade event is then done using SVMs, an approach which automatically provides a confidence measure on each classification.
Indirect detection experiments
Ab-antigen binding and Ab-pore interaction in presence of urea and MgCl2
The clarity of the current blockade signal was examined by varying the composition of working buffer. In one series of experiments, mentioned above, we used free antibody molecule interacting with the nanopore detector, where the antibody (anti-biotin) molecule is introduced to our nanopore device to produce the characteristic two-state telegraph signal (Fig. 11). The blockade signal for the antigen is practically unaltered by excess antigen: even 100 fold excess of biotin does not change the blockade signal considerably (Fig. 11, panel 2). The signal changes greatly in presence of urea, however, in a relatively small concentration. Here the duration of any event to occupy upper state level becomes shorter and the total probability value of upper level decreases with urea concentration rise. In the other series of experiments the antibody was conjugated to the 9GC DNA hairpin. Primary amine crosslinked using 1-Ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC) to the peptide carboxyl terminus of the antibody heavy chain. This crosslinkage results in a covalent bond between the primary amine and the carboxyl. With this arrangement, two transient states were observed in the current signal resulting from the IgG-DNA molecule blockades (see figure in Additional File 1). Additional File 1 shows the Current blockade signal change induced by presence MgCl2 (0.4 M). Antigen binding to Ab-DNA hairpin appears to result in a more complex signal (lower panel). With MgCl2 concentration increase, there is an increased occurrence of the upper-level blockade state, at the same time the current signal becomes nosier and the open channel current increases to 130 pA. The lower panel corresponds to 0.4 M concentration of MgCl2. The total time scale is 15 seconds.
With progressive MgCl2 concentration, there exist a tendency of higher probability for the upper state, at the same time the current signal becomes noisier and the open channel current rises up to 130 pA. The effect of increase of KCL on a DNA hairpin blockade signal is shown in the figures in Additional Files 1 and 3. Additional File 2 shows the current blockade signal change with KCl increase. A nine base-pair DNA hairpin with a distinctive upper level toggle is shown in the top panel. The middle and bottom panels show the blockade patterns at 1.9 and 2.5 M KCl, correspondingly. As the concentration of background electrolyte increases, the signal keeps its highly structured profile, although becoming noisy. Additional File 3 shows the "Upper level toggler" DNA hairpin profiles at different KCl concentrations. The 150-component profiles of the signals for a nine base-pair DNA hairpin, known to exhibit a "fine-structure" in its upper-level, at 1.0, 1.9 and 2.5 M KCl (corresponding to the signal shown in the previous image). Each blockade signature is de-noised by 5 rounds of Expectation-Maximization (EM) training on the parameters of the HMM. After the EM iterations, 150 parameters are extracted from the HMM. The 150 feature vectors obtained from the 50-state HMM-EM/Viterbi implementation in are: the 50 dwell percentage in the different blockade levels (from the Viterbi trace-back states), the 50 variances of the emission probability distributions associated with the different states, and the 50 merged transition probabilities from the primary and secondary blockade occupation levels (fits to two-state dominant modulatory blockade signals).
A discrete level change after initial current blockade is a distinctive feature of channel blockade detection at this scale (as previous work with hairpin DNA has shown [2, 3, 4, 5, 6, 7]) and supports the assumption of protein-loop or terminus insertion into the channel as a possible mechanism for this process. Assuming that a very small portion of the globular antibody has inserted itself into the channel, this type of signal could be produced by a protein branch dangling over the limiting aperture. The DNA hairpin molecules studied in [2, 3, 4, 5, 6, 7] produce such a signal as its nine base pair stem dangles over the limiting aperture typically creating a two state dominant current blockade. The fixed blockade levels are thought to represent short lived ionic bonds forming between the residues of the alpha-hemolysin channel and the terminal base pair of the hairpin stem. The Hidden Markov Model (HMM) signal profiles in the figures, have as their first 50 components, a histogram-like profile of the main blockade region that usually resides between 40% and 60% of the baseline current.
The IgG antibody may vary in net charge and is nowhere near as negatively charged as the DNA hairpin molecules examined. Differences in channel interaction are often attributed to its net charge and its electrophoretic mobility [2, 16]. To improve the antibody's affinity for the channel and to aid in signal classification, a complex of antibody and DNA hairpin is linked together. The result is the increase in channel affinity and a significant reduction in capture class configurations (see Figs. 2, 9, 10 and  for further details), while, evidently, still retaining binding detection sensitivity.
For the antigen-binding studies, different versions of copolymer (Y, E)-A – K are prepared which vary in molecular weight and valency (the one-letter codes are for respective the amino acids). Previous studies have demonstrated that the epitope to which the antibodies bind may be represented by the synthetic peptide EYYEYEEY [34, 35, 36, 37]. Thus we have the following antigens: 1. The synthetic polypeptide (Y, E)-A – K, which is highly multivalent and with different preparations has a molecular of either 50,000, 125,000, or 300,000 daltons. 2. EYYEYEEY which is monovalent and has a molecular weight of 1183, and 3. Ovalbumin, molecular weight 43,000 which has been conjugated to EYYEYEEY at ratios of 1, 3, and 10 peptides per molecule of OVA. These different antibody preparations allow study of the effect of antigenic mass and valency of binding upon the observations.
For the experiments on antibody (anti-biotin) interaction with alpha-hemolysin channel, we assume as a possible interpretation that the hypervariable loops of the antibody can be captured by the nanopore. Since the blockade signal for the antigen is practically unaltered by excess antigen, even the 100 fold excess of biotin does not change the blockade signal considerably (see Results), biotin is not thought to interact greatly with the (blocked) channel. The strong signal changes observed in presence of urea and MgCl2, in relatively small concentrations normally insufficient to result in protein denaturing, on the other hand, allows us to hypothesize significant channel-Ab binding interference by urea.
Experimental efforts have found that different forms of channel blockade can be associated with the same antibody, presumably associated with different orientations of "capture" of an antibody in the detector's nanometer-scale channel (i.e., the antibody presents numerous epitopes for nanopore capture). To get around the non-specific capture orientation limitation, DNA-hairpin anchored antibodies have also been an area of study , where the DNA linkage is to the carboxy-terminus at the base of the antibody. On-binding is easily observed in all of the above experiments, as shown in the Results. So the difficulty is the observation of Off-binding in the timescale of the experiment. For this reason weaker binding-pairs are sought, or are being induced by introduction of divalent ions (MgCl2 is added to the buffer, for example) and chaotropic agents in general.
In addition to a murine IgG1, a human IgG1, and a human IgG4 (see Methods for details), we plan on testing the following antibodies: murine and human monoclonal antibodies representing all of the IgG isotypes, murine IgG1 antibodies having different pIs (determined experimentally), and polyclonal protein A purified antibodies from human, rabbit, goat and mouse. These and other antibodies are going to be used to determine whether the ability of antibodies to be captured is a general phenomenon, and whether the blockade patterns obtained with antibodies are dependent on pI, isotype, species origin, or other predictable characteristics of the antibody or of the method of preparation.
Insofar as an analysis of the binding affinity between two complex proteins is concerned (like the antibody and its target antigen), this study also provides a test case for nanopore-based protein-protein interaction studies. It might prove possible to rank the different antibody binding strengths to target antigen, for example, according to the observed lifetimes of their bound states. Efforts are underway to observe full binding histories to get very precise measurements of Kon and Koff. Nanopore-based approaches may eventually provide a direct analysis of the complex conformational "negotiations" that occur upon binding between proteins.
Each experiment is conducted using one α-hemolysin channel inserted into a diphytanoyl-phosphatidylcholine/hexadecane bilayer across a 25-micron-diameter horizontal Teflon aperture, as described previously [2, 3, 4] (Fig. 1). Two seventy microliter chambers on either side of the bilayer contain 1.0 M KCl buffered at pH 8.0 (10 mM HEPES/KOH), except in the case of buffer experiments where the salt concentration, pH, or other aspects of composition may be varied (introduction of glucose, urea, etc.). Voltage of 120 mV is applied across the bilayer between Ag-AgCl electrodes. DNA control probes are added to the cis chamber at 10 or 20 μM final concentration. All experiments are maintained at room temperature (23 ± 0.1°C), using a Peltier device.
Control probe design
Since the five DNA hairpins studied in the prototype experiment have been carefully characterized , they are used in the antibody (and other) experiments as highly sensitive controls. (The probes are used, periodically, to test the channel for proper operation, or if a channel is suspected to be performing abnormally, they are used to test right away. If recognized blockade signals don't result upon introduction of control it is assumed that the channel is not operating under desirable conditions.) The nine base-pair hairpin molecules examined in the prototype experiment share an eight base-pair hairpin core sequence, with addition of one of the four permutations of Watson-Crick base-pairs that may exist at the blunt end terminus, i.e., 5'-G•C-3', 5'-C•G-3', 5'-T•A-3', and 5'-A•T-3' denoted 9GC, 9CG, 9TA, and 9AT, respectively. The full sequence for the 9CG hairpin is 5' CTTCGAACG TTTTCGTTCGAAG 3', where the base-pairing region is underlined. The eight base-pair DNA hairpin is identical to the core nine base-pair subsequence, except the terminal base-pair is 5'-G•C-3'. The prediction that each hairpin would adopt one base-paired structure was tested and confirmed using the DNA mfold server http://bioweb.pasteur.fr/seqanal/interfaces/mfold.html.
Antibody/antigen design, synthesis, and purification
For most of the experiments, a panel of native and genetically engineered antibodies to a well defined synthetic polypeptide antigen are used, (Y, E)-A – K, [34, 35, 36, 37]. The antigen-binding characteristics, ability to form immune complexes, and effector functions of these antibodies have been carefully studied. Three different antibodies from this set are utilized in the experiments in this experimental effort. All have identical variable domains of murine origin, but one is a murine IgG1, one a human IgG1, and the other a human IgG4. Only the murine IgG1 data is shown in this paper.
All monoclonal antibodies are grown in tissue culture because ascites preparations are inflammatory exudates subjecting the antibodies to the potential of proteolytic digestion, attachment of complement components and so forth. Cells are either grown in medium containing fetal calf serum adsorbed on protein G to remove remaining Ig, or in serum free hybridoma medium. To test the effect of preparation method, murine IgG1 antibody is either purified by ammonium sulfate precipitation, antigen-affinity purification or protein G chromatography. All other antibodies are routinely purified on protein G and eluted with 0.5 M glycine-HCl pH 2.5, immediately neutralized, and dialyzed into phosphate buffered saline (PBS). Once antibodies are purified, they are run on SDS-PAGE to confirm purity and run on IEF prepoured gels (Biorad) to determine PI. Antigen binding is confirmed by the immunoassay technique of ELISA (enzyme-linked immunosorbent assay, a biochemical technique to identify the presence of antibody or antigen) in PBS and in 1 M KCl (so long as that buffer is used).
Mouse anti-Biotin monoclonal antibody
Anti-Biotin Antibody monoclonal IgG1 from Stressgen was used at the concentration of 1.0 mg/mL. Horseradish peroxidase (HRP) was conjugated with affinity purified mouse immunoglobulin in phosphate buffered saline (PBS) at pH 7.2 with 0.1 mM PMSF and 50% glycerol. The Immunogen was un-bound Biotin.
Federal funding was provided by NIH K-22 (PI, 5K22LM008794), NIH NNBM R-21 (co-PI), and NIH R-01 (sub-award). State funding was provided from a LaBOR Enhancement (PI), a LaBOR Research Competitiveness Subcontract (PI), and a LaBOR/NASA LaSPACE Grant (PI). Many thanks to Dr. Seth Pincus for discussions and providing the monoclonal antibodies studied (except for the commercially obtained anti-Biotin Antibody). Thanks to Matthew Landry for helping to run the cheminformatics software. We thank the Reviewers of the manuscript (anonymous) for their excellent and helpful suggestions.
This article has been published as part of BMC Bioinformatics Volume 8 Supplement 7, 2007: Proceedings of the Fourth Annual MCBIOS Conference. Computational Frontiers in Biomedicine. The full contents of the supplement are available online at http://www.biomedcentral.com/1471-2105/8?issue=S7.
- 3.Vercoutere W, Winters-Hilt S, DeGuzman VS, Deamer D, Ridino S, Rogers JT, Olsen HE, Marziali A, Akeson M: Discrimination Among Individual Watson-Crick Base-Pairs at the Termini of Single DNA Hairpin Molecules. Nucl Acids Res 2003, 31: 1311–1318. 10.1093/nar/gkg218PubMedCentralCrossRefPubMedGoogle Scholar
- 5.Winters-Hilt S: Nanopore detection using channel current cheminformatics. SPIE Second International Symposium on Fluctuations and Noise, 25–28 May, 2004Google Scholar
- 6.Winters-Hilt S, Akeson M: Nanopore cheminformatics. DNA and Cell Biology 2004.Google Scholar
- 7.Winters-Hilt S: Highly Accurate Real-Time Classification of Channel-Captured DNA Termini. Third International Conference on Unsolved Problems of Noise and Fluctuations in Physics, Biology, and High Technology 2003, 355–368.Google Scholar
- 9.Winters-Hilt S: Nanopore Detector based analysis of single-molecule conformational kinetics and binding interactions. BMC Bioinformatics 7(Suppl 2):S21. 2006, Sept. 26, 2006, Sept. 26, 10.1186/1471-2105-7-S2-S21Google Scholar
- 12.Ken Healy K: Nanopore-Based DNA Analysis. In PhD Thesis. The National University of Ireland – University College Cork; 2006.Google Scholar
- 20.Heins E, Albertorio F, Yang T, Siwy Z, Baker L, Cheley S, Bayley HP, Cremer PS, Martin CR: "Development of a supported lipid bilayer on porous polymeric support". presented at the 49th Biophysical Society Meeting, Long Beach, CA, USA 2005.Google Scholar
- 27.Heng J, Dimitrov V, Grinkova Y, Ho C, Kim T, Muller D, Sligar S, Sorsch T, Twesten R, Timp R, Timp G: "The detection of DNA using a silicon nanopore". Proc IEDM 32.2.1–32.2.4. 8–10 Dec. 2003 8–10 Dec. 2003Google Scholar
- 32.Cormen TH, Leiserson CE, Rivest RL: Introduction to Algorithms. MIT-Press, Cambridge, USA; 1989.Google Scholar
This article is published under license to BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.