Date Published: April 19, 2018
Publisher: Springer Berlin Heidelberg
Author(s): Oscar B. Torres, Alexander J. Duval, Agnieszka Sulima, Joshua F. G. Antoline, Arthur E. Jacobson, Kenner C. Rice, Carl R. Alving, Gary R. Matyas.
We describe for the first time a method that utilizes microscale thermophoresis (MST) technology to determine polyclonal antibody affinities to small molecules. Using a novel type of heterologous MST, we have accurately measured a solution-based binding affinity of serum antibodies to heroin which was previously impossible with other currently available methods. Moreover, this mismatch approach (i.e., using a cross-reactive hapten tracer) has never been reported in the literature. When compared with equilibrium dialysis combined with ultra-performance liquid chromatography/tandem mass spectrometry (ED-UPLC/MS/MS), this novel MST method yields similar binding affinity values for polyclonal antibodies to the major heroin metabolites 6-AM and morphine. Additionally, we herein report the method of synthesis of this novel cross-reactive hapten, MorHap-acetamide—a useful analog for the study of heroin hapten–antibody interactions. Using heterologous MST, we were able to determine the affinities, down to nanomolar accuracies, of polyclonal antibodies to various abused opioids. While optimizing this method, we further discovered that heroin is protected from serum esterase degradation by the presence of these antibodies in a concentration-dependent manner. Lastly, using affinity data for a number of structurally different opioids, we were able to dissect the moieties that are crucial to antibody binding. The novel MST method that is presented herein can be extended to the analysis of any ligand that is prone to degradation and can be applied not only to the development of vaccines to substances of abuse but also to the analysis of small molecule/protein interactions in the presence of serum.
Drug abuse and misuse continue to be at epidemic levels the world over. According to the 2017 World Drug Report, approximately 70% of the global burden of disease resultant of total drug use disorders (29.5 million) was attributable to opioids (~ 20.7 million) [1, 2]. Incidentally, heroin is a drug with one of the highest mortality rates . In the United States alone, the number of deaths from heroin has spiked in the past decade with a 6.2-fold increase from 2002 to 2015 , and in October of 2017, the opioid crisis was declared a Public Health Emergency. Among various psychoactive substances, heroin ranks among the worst in terms of the physical harm and strong dependencies that it generates . Therefore, there is an urgent need to develop alternative heroin abuse treatments. Recently, vaccines have been explored as a potential treatment modality for substances of abuse because they do not produce unwanted neurological side effects and they have the potential to be utilized as preventive therapeutics against drug overdose or as synergistic therapies for substance-use disorders [5, 6]. Vaccines to substances of abuse function by generating antibodies that sequester the substance in the blood, thereby preventing it from crossing the blood–brain barrier, engaging its receptor in the brain, and inducing its subsequent psychoactive effects. The primary component of such a vaccine is the hapten–carrier conjugate. In general, substances of abuse are small molecules and consequently do not evoke an immune response by themselves. Thus, an analog (hapten) that structurally mimics the substance is covalently linked to an immunogenic carrier, such as tetanus toxoid (TT), to allow for the substance’s presentation to immune cells [7, 8].
Sulfo-cyanine5 maleimide (≥ 95%, Cy5) was purchased from Lumiprobe Corporation (Hallandale Beach, Florida, USA). Trifluoroacetic acid (TFA), triethylsilane (Et3Si), dimethyl sulfoxide (DMSO), 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid) (HEPES), triethylamine (Et3N), tert-butyl-(chloro)diphenylsilane (TBDPCl), imidazole, 2-bromoacetamide, tetrabutylammonium fluoride hydrate (TBAF), cesium fluoride (CsF), chloroform (CHCl3), dimethylformamide (DMF), Tween® 20, tetraisopropyl pyrophosphoramide (iso-OMPA), bis(4-nitrophenyl) phosphate (BNPP, 99%), acetanilide (≥ 99.9%), and N-methylpiperidine (99%) were purchased from Sigma-Aldrich (Saint Louis, MO, USA). RediSep® Rf Reversed-phase C18 column/CombiFlash Rf+ PurIon Flash chromatography system was purchased from Teledyne Isco (Lincoln, NE, USA). Anti-morphine antibody (ab1060) was purchased from Abcam (Cambridge, MA, USA). Standard treated glass capillary tubes for MST measurements were purchased from NanoTemper Technologies GmbH (Munich, Germany). Drug solutions (1 mg/mL) of 3,6-diacetylmorphine•HCl•H2O (heroin•HCl•H2O), 6-AM•HCl, morphine•H2O, morphine-3-β-glucuronide•H2O, morphine-6-β-glucuronide•H2O, hydromorphone, naloxone•HCl, normorphine•HCl•H2O, nalorphine•HCl, and oxymorphone were purchased from Lipomed (Cambridge, MA, USA). Drug solutions (1 mg/mL) of levorphanol tartrate, desomorphine, thebaine, and meperidine; and drug solutions (100 μg/mL) of morphine N-oxide and 10-hydroxymorphine were purchased from Cerilliant Corporation (Round Rock, TX, USA). All drug solutions were Certified Reference Materials. Optima™ LC/MS grade ammonium formate (NH4HCOO), methanol (MeOH), acetonitrile (ACN), and water (H2O) were purchased from Fischer Scientific (Suwanee, GA). Waters XBridge® BEH C18 column and screw neck total recovery vial with polytetrafluoroethylene/silicone septa were purchased from Waters (Cambridge, MA, USA). Dulbecco’s phosphate buffered saline (DPBS, 10 mM Na2HPO4, 1.8 mM KH2PO4, 2.7 mM KCl, 137 mM NaCl, pH 7.4) was purchased from Quality Biological Inc. (Gaithersburg, MD, USA). All MST experiments were performed in DPBS with 0.05% Tween® 20 (DPBS-Tween).
Traditionally, a fluorophore is conjugated to a ligand or protein to generate the fluorescent tracer for MST measurements. It is noteworthy to mention that proteins have intrinsic fluorescence due to tryptophan residues. This intrinsic fluorescence, however, is difficult to utilize when MST measurements are conducted in the presence of dilute serum. Conversely, the intrinsic fluorescence becomes negligible in the presence of a strong fluorophore such as Cy5. In this study, MorHap [6, 12, 24–27], a cross-reactive analog of 6-AmHap, was conjugated to Cy5. The resulting novel fluorescent tracer (MorHap-Cy5) was used in all the MST experiments. 6-AmHap-Cy5 could not be used as a fluorescent tracer because of its high binding affinity to 6-AmHap-Abs (data not shown).
A novel compound, MorHap-acetamide, was used to measure the binding affinity of monoclonal antibody ab1060 to MorHap, a heroin hapten. The two-step synthesis of MorHap-acetamide was problematic (Fig. 3, Scheme a), hence, MorHap-acetamide was synthesized in four steps (Fig. 3, Scheme b). The phenol functionality of MorHap was protected using TBDPCl to yield silyl-protected MorHap, 1. The trityl (Trt) group of MorHap was deprotected using 10% TFA to yield 2. The free thiol was then reacted with 2-bromoacetamide to yield silyl-protected MorHap-acetamide, 3. Finally, the silyl-protecting group was removed using CsF or TBAF to yield MorHap-acetamide. The detailed synthesis and analytical characterization of 1, 2, 3, and MorHap-acetamide is described in ESM.
To confirm that MST can accurately measure antibody binding affinities, the Ki of morphine to ab1060 was measured. This mouse mAb binds morphine with a manufacturer-reported Kd value of 2 nM, as determined by a proprietary ultraviolet-visible (UV-VIS) absorption spectroscopy method [28, 29] and by ED-UPLC/MS/MS . Since neither ab1060 nor morphine contain strong fluorophores to allow the direct measurement of binding interactions by MST, and since the presence of a fluorophore on the antibody or the ligand had the potential to produce inaccurate binding data, our strategy was to measure the Ki in a two-step process. First, the fluorescent tracer MorHap-Cy5 was synthesized and an increasing concentration of ab1060 was used to measure the Kd to the tracer (conventional MST). Second, in a separate trial, an increasing concentration of morphine was used to displace MorHap-Cy5 in the ab1060:MorHap-Cy5 complex to measure IC50 (heterologous MST). The Kd of ab1060 to the tracer and the IC50 of morphine to the complex were used together to calculate the Ki of ab1060 to morphine (vide infra).
Post-immune sera (week 8) from the mouse study described by Sulima et al.  were used in this study. The Kds of 6-AmHap-Abs to 6-AM and morphine, which were subsequently used to approximate the concentration of antibody binding sites, were determined by ED-UPLC/MS/MS in accordance with the method outlined by Torres et al. . Briefly, post-immune serum was diluted in DPBS (1:1600). Dialysis buffer was prepared by diluting the corresponding pre-immune serum (week 0) in DPBS (1:1600). Competitive inhibitors, 6-AM or morphine, were prepared in the dialysis buffer. A 100-μL dilution of week 8 serum containing 5 nM D3-tracer was pipetted into the sample chamber, and a 300-μL aliquot of the competitive inhibitor solution was added to the buffer chamber. Both week 8 and week 0 were run against each other at the same sera dilutions. For negative controls, the same setup was employed except both sides of the dialysis chamber contained only week 0 serum. Equilibrium dialysis, preparation of samples for UPLC/MS/MS, and calculation of Kd were performed as described .
To accurately determine the binding affinity of polyclonal serum to heroin, heroin degradation must be suppressed during the execution of the heterologous MST assay. Since competition MST is performed at a constant concentration of antibody and a varying concentration of drug competitor, the heroin degradation studies were performed with a constant dilution of serum and a varying concentration of heroin to mimic the heterologous MST assay conditions.
The MST measurements were done on a Monolith™ NT.115Pico instrument from NanoTemper Technologies GmbH. The instrument was equipped with an IR laser (wavelength, 1475 ± 15 nm; power, 120 mW maximum) and a red fluorescence channel that can detect red dyes, such as Cy5 and Alexa Fluor 647. The samples were measured at an LED power of 20% and an MST power of 25% with a laser-on time of 30 s and a laser-off time of 5 s. MST measurements of 15–16 samples typically take ~ 20 min.
Statistical analyses were performed using GraphPad Prism version 7.0a. A one-way analysis of variance (ANOVA), Kruskal–Wallis test with Dunn’s correction for multiple comparisons was used to compare the Kd values of ab1060 to MorHap-Cy5 with different fluorescent tracer concentrations, the Ki values of ab1060 to morphine derived from the 15 various assay conditions, and the Ki values of ab1060 to morphine derived from UV-VIS, MST, and ED-UPLC/MS/MS. A Mann–Whitney non-parametric T test was used to compare the Ki values of 6-AmHap-Abs to heroin in the presence and absence of esterase inhibitors. A Mann–Whitney non-parametric T test was also used to compare the binding affinities of 6-AmHap-Abs to 6-AM and morphine derived from MST and ED-UPLC/MS/MS. The binding curves were re-plotted using GraphPad Prism version 7.0a for presentation purposes.
The measurement of antibody:antigen affinities in the presence of biological matrices (e.g., serum) by solution-based methods has long been an arduous task [13, 14, 17]. One approach to solve this problem is to use labeled antibodies, which provide the chemical properties necessary for the detection of antibody:antigen binding. In a typical setup, the antibodies are purified from sera or produced by cell culture, treated with an excess amount of tracer, and then further purified to isolate the labeled antibodies [23, 33, 34]. The exhaustive purification steps and non-specific labeling of the antibodies invariably lead to the production of labeled antibodies that have varying amounts of binding loss and/or may not reflect the binding properties of the same antibodies in their native milieu. To date, competition ED-UPLC/MS/MS is the only solution-based assay that can indirectly measure polyclonal antibody affinities without the need for antibody labeling or extensive purification due to its use of a deuterated competitor. Previously, we demonstrated competition ED-UPLC/MS/MS as a novel analytical method that can measure drug–antibody interactions . Although it provides accurate Kd values for most analytes in the nanomolar range, competition ED-UPLC/MS/MS has some major drawbacks including the long hours needed for dialysis and LC/MS runs, as well as the relatively large volume of serum that is needed per experiment (5–10 μL). Alternatively, solid-based heterologous and homologous competition ELISAs require much less time to execute. Competition ELISAs, however, similarly have high sample volume requirements and have the potential to conflate antibody concentration variables with antibody affinity variables. Although the binding affinities of antibody:protein antigen that are derived from ELISA are comparable to solution-based methods [35, 36], we found that the IC50 values of antibody:opioid antigen interactions were platform-dependent. For example, the IC50s derived from heterologous ELISA were comparable to the IC50s derived from heterologous MST when the competitors have a relatively tight binding affinity to 6-AmHap-Abs. However, unlike heterologous MST, overestimation of IC50 from heterologous ELISA was observed when the competitors had moderate to weak binding affinities to the polyclonal antibodies (ESM Table S3). On the other hand, homologous ELISA tends to numerically overestimate IC50 by a factor of 103 [13, 37] for antibody:heroin-derived molecule interactions, frequently returning IC50 values in the micromolar range. Collectively, homologous ELISA and homologous MST were both unsuccessful in measuring the apparent binding affinities of 6-AmHap-Abs to various opioids because they require high concentrations of the competitors to disrupt the inherent tight binding of the homologous pairs (i.e., 6-AmHap-Abs:6-AmHap ligand). It is noteworthy to mention that the overestimation of IC50 by homologous ELISA and by heterologous ELISA with weak competitors might only occur in ELISA platforms utilizing protein carrier-haptens with anti-hapten serum IgG. BSA is reported to be an ellipsoid with dimensions of 8 × 4 nm  while IgG’s antigen-binding arms are separated by a 10–15 nm spread [39–41]. Despite the fact that we have previously used a BSA-hapten coating antigen with a hapten density of three to five per protein , it is possible that the two arms of IgG are engaged in divalent interactions between neighboring BSA-haptens. Immobilization of the antigen in a solid support undoubtedly places protein carrier-hapten molecules within the 10–15 nm distance, which results in a high local concentration of hapten. This could therefore provide opportunity for more divalent interactions than when the antigen is in solution. It is possible that the geometry constraints of the protein carrier-hapten and IgG complex combined with the surface effects associated with solid-based ELISA  become more pronounced in heterologous ELISA when the competitor is weaker. In homologous ELISA, these geometry constraints and surface effects compounded with the strong association of antibody:hapten homologous pairs might drive the need for high concentrations of competitor drugs to dissociate the antibody:antigen interaction, thereby further overestimating the IC50 values to 103-fold. These same effects may also occur with other assays, such as SPR when the protein carrier-hapten is used as the capture antigen. The shortcomings of ELISA, ED-UPLC/MS/MS, and other binding affinity assay methods thus prove that an alternative solution-based assay—offering nanomolar accuracy with a short assay time and low sample requirement—is needed.