Therapeutic and Prophylactic Vaccines to Counteract Fentanyl Use Disorders and Toxicity

Abstract

The incidence of fatal overdoses has increased worldwide due to the widespread access to illicit fentanyl and its potent analogues. Vaccines offer a promising strategy to reduce the prevalence of opioid use disorders (OUDs) and to prevent toxicity from accidental and deliberate exposure to fentanyl and its derivatives. This study describes the development and characterization of vaccine formulations consisting of novel fentanyl-based haptens conjugated to carrier proteins. Vaccine efficacy was tested against opioid-induced behavior and toxicity in mice and rats challenged with fentanyl and its analogues. Prophylactic vaccination reduced fentanyl- and sufentanil-induced antinociception, respiratory depression, and bradycardia in mice and rats. Therapeutic vaccination also reduced fentanyl intravenous self-administration in rats. Because of their selectivity, vaccines did not interfere with the pharmacological effects of commonly used anesthetics nor with methadone, naloxone, oxycodone, or heroin. These preclinical data support the translation of vaccines as a viable strategy to counteract fentanyl use disorders and toxicity.

Graphical Abstract

INTRODUCTION

The prevalence of opioid use disorders (OUDs) and incidence of opioid-related overdoses have increased at an alarming rate in the past decade in North America and worldwide.¹ In the United States, at least 2.5 million people suffer from an OUD, and 46 802 opioid-related fatal overdoses occurred in 2018 alone.^1,2 These statistics are supported by the widespread access to prescription opioids as well as illicit use of heroin, counterfeit prescription opioids, or street mixtures of heroin and psychostimulants laced with synthetic opioids such as fentanyl and its potent analogues.^3–5 In the United States, roughly 66% of opioid-related fatal overdoses reported in 2018 involved fentanyl,⁶ a synthetic opioid 50- to 100-fold more potent than heroin.⁷ In addition to the well-established role of fentanyl in OUD and drug-related overdoses, fentanyl and its potent analogues could be potentially involved in accidental or deliberate poisoning in high-risk occupations such as airport security, custom officials, law enforcement, and military, or used as chemical threat agents in mass casualty incidents.^8,9 These data highlight the need for safe, long-lasting, and effective medical interventions to counteract toxicity and overdose from fentanyl and its analogues.

Approved medical interventions to counteract OUD and overdose consist of opioid receptor ligands, including the agonist methadone, partial agonist buprenorphine, antagonists naltrexone and naloxone, and combinations thereof (e.g., suboxone). These medications are safe and effective, but their clinical outcome is still suboptimal due to side effects, administrative hurdles, and the potential for abuse and diversion.^10–17 Although methadone, buprenorphine, and naltrexone are known to prevent opioid overdoses, their ability to reduce fentanyl-induced fatalities is not clear.^18–20 In addition, these medications may not be prescribed to those diagnosed with other substance use disorders (SUDs) related to cocaine or methamphetamine, which places these patients at higher risk for overdose if accidentally exposed to fentanyl-contaminated psychostimulants.⁵ Naloxone is currently approved to reverse acute opioid overdose and can be administered intranasally, intravenously, subcutaneously, or intramuscularly.¹⁰ However, due to both the potency of fentanyl and the limited half-life of naloxone (~90 min), multiple doses are often required to rescue patients exposed to fentanyl and other opioids.^14,21–23 Because of the limitations of current medication-based treatments, vaccines have been proposed as an alternative or complementary strategy to treat OUD and to prevent overdose.^24–30 Vaccines against OUD consist of conjugates containing drug-based haptens linked to an immunogenic carrier protein, which stimulate the innate and adaptive immune systems to generate polyclonal antibodies against the selected opioid. Opioid-specific IgG antibodies bind the target opioid in serum and reduce its distribution to the brain, thus preventing opioid-induced behavioral and pharmacological effects, as well as opioid-related toxicity, including overdose, in animal models.^{24,29,31–36} Vaccines have shown preclinical proof of efficacy against heroin, its metabolites 6-acetylmorphine (6-AM) and morphine, oxycodone, and morphine, oxycodone, hydrocodone, fentanyl, and its analogues.^37–41

Recent preclinical studies have shown that vaccines are effective in reducing the pharmacological, behavioral, and toxic effects of fentanyl and its analogues in animal models of OUD.^38,42,43 In this context, our team has also reported a vaccine against fentanyl composed of a fentanyl-based hapten containing a tetraglycine linker [F(Gly)₄ or F₁ in Figure 1] conjugated to either a native keyhole limpet hemocyanin (KLH) or a GMP-grade subunit KLH (sKLH) carrier protein. Active immunization with either F₁–KLH or F₁–sKLH reduced fentanyl-induced antinociception in the hot plate test, a measure of centrally mediated behavior, in both mice and rats.⁴⁴ Vaccination of rats with F₁–sKLH also protected against fentanyl-induced respiratory depression and bradycardia while preserving the ability of naloxone to reverse the pharmacological effects of fentanyl.⁴⁴

Figure 1.

Series of fentanyl-based F_1–6 haptens. The first-generation fentanyl-based F₁ hapten containing a tetraglycine linker (F(Gly)₄) was compared to the novel next-generation F_2–6 hapten series. Proof of efficacy for F₁ hapten was previously established.⁴⁴ The F_1–6 haptens were conjugated to either the sKLH or diphtheria toxin cross-reacting material (CRM) carrier proteins to generate conjugate vaccines for in vivo testing in mice and rats.

To further improve upon the first-generation F₁–sKLH conjugate vaccine, the current study generated a library of fentanyl-based haptens to probe structural variations of synthetic opioids with the 4-anilinopiperidine core, as well as investigating its impact on vaccine efficacy against fentanyl and its analogues. Our screening strategy was designed to first identify promising haptens and conjugate vaccines in mice and then to further characterize leads in rat models of opioid-induced behavior and toxicity. In mice, lead selection was based upon analysis of vaccine-induced antibody titers, affinity for fentanyl and their efficacy in reducing fentanyl-induced antinociception in the hot plate test, increasing fentanyl serum concentration and decreasing the distribution of fentanyl to the brain. Lead vaccines were then evaluated in rats for their efficacy in blocking physiological and behavioral effects of fentanyl and other 4-anilinopiperidine-related analogues, such as sufentanil.

The first-generation F₁–sKLH vaccine did not interfere with the reversal of fentanyl’s effects by naloxone, supporting the use of anti-fentanyl vaccines in clinical settings.⁴⁴ In a broader context, it has been shown that anti-heroin and anti-oxycodone vaccines do not interfere with naloxone and that the vaccines’ ability to induce anti-opioid antibodies is not affected by the concurrent administration of extended-release naltrexone or morphine.^45,46 However, additional studies are needed to investigate the impact of vaccination against fentanyl and its analogues on medication therapies for OUD or other medications used in pain management or critical care. To provide further preclinical proof of selectivity and safety, the current study tested whether vaccination against fentanyl would interfere with commonly used anesthetics, opioid agonists, and opioid antagonists to demonstrate that vaccination against synthetic opioids does not interfere with emergency medicine, surgical procedures, or medications for OUD and pain management.

RESULTS

Vaccine Formulation: Hapten Synthesis, Conjugation, and Characterization.

Generation of effective vaccines against drugs of abuse requires the use of haptens that display molecular mimicry with the target compound. Unlike other natural or semisynthetic opioids, fentanyl and related synthetic opioids display no intrinsic reactive handles for conjugation to carrier proteins for the purpose of vaccine generation. To generate drug-based small-molecule haptens suitable for conjugation, a series of molecularly distinct lysine-reactive fentanyl analogues were created by varying the structure and position of linking groups on the pharmacophore (Figure 1). Although the F₁ hapten containing a tetraglycine linker was previously reported,⁴⁴ in this study, synthesis of the F₁ hapten yielded improved hapten purity and enhanced conjugation to carriers as shown by its matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) and dynamic light scattering (DLS) profiles (Supporting Table 1 and Figure 3). The synthetic approach employed to develop the F_1–6 haptens can be classified into four fentanyl derivative categories: (a) replacement of 2-ethyl-benzyl group with lysine-reactive linking species consisting of a tetraglycine linker (F₁), (b) modification of the para position on the 2-ethyl-benzyl with an acrylic acid moiety (F₃), (c) modification of the 4-amino-phenyl ring at the para position (F₂, F₄, and F₅), and (d) modification of the 4-aminophenyl ring at the meta position (F₆). These haptens, displaying a synthetic handle capable of conjugation to carrier proteins, were readily prepared using straightforward techniques (synthesis of the novel F₂–F₆ is described in Schemes 1,2 and the Experimental Section) yielding purity exceeding 95% as confirmed via high-performance liquid chromatography (HPLC, Supporting Figure 1). The F_1–6 series displaying the fentanyl core structure (Supporting Figure 2) is expected to offer a range of structural diversity sufficient to provide different antigen presentation to the immune system.

Figure 3.

Vaccine efficacy against fentanyl in rats. Sprague Dawley rats (n = 6, each group) were vaccinated i.m. on days 0, 21, 42, and 63 with conjugates containing the F_1–3 haptens conjugated to sKLH, CRM₁, or CRM₂ and on day 49 challenged s.c. with 0.075 mg/kg fentanyl: (A) antinociception in the hot plate test, (B) respiratory depression reported as oxygen saturation (%), and (C) bradycardia reported as heart rate (beats per minute, bpm) measured by pulse oximetry. Following challenges with fentanyl analogues (shown in Figure 4), rats were vaccinated on day 63 and challenged s.c. on day 77 with 0.1 mg/kg fentanyl: (D) oxygen saturation, (E) heart rate, and (F) total fentanyl in the brain at 60 min post drug challenge. Data are mean±SEM. Statistical symbols: * or # p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, and **** p ≤ 0.0001 compared to either control (*) or F₁-sKLH (#).

Scheme 1. Schematic Depicting the Synthesis Steps for the F₂ and F₃ Haptens^a.

^bReagents and conditions: (a) tert-butylacrylate, Pd(OAC)₂, dppe, Et₃N, 140 °C, 24 h; (b) trifluoroacetic acid (TFA), dichloromethane (DCM), 25 °C, 24 h; and (c) 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC), 4-dimethylaminopyridine (DMAP), N-hydroxysuccinimide, DCM, 25 °C, 24 h.

^aThe F₂ and F₃ haptens were conjugated to carrier proteins using N-hydroxysuccinimide-activated hapten esters. Hapten synthesis and conjugation are described in the Experimental Section.

Scheme 2. Schematic Depicting the Synthesis Steps of the F₄, F₅, and F₆ Haptens^a.

^bReagents and conditions: (a) 1-phenethyl-4-piperidone, PMHS, SnCl₂, MeOH, 70 °C, 16 h; (b) propionic anhydride, CH₂Cl₂, rt, 24 h; (c) TFA, CH₂Cl₂/H₂O, rt, 18 h; (d) (CCl₃O)₂CO, Et₃N, CH₃CN, NH₂CH₂COOCH₃·HCl, THF; (e) LiOH, MeOH/THF/H₂O, rt, 24 h; and (f) succinic acid monomethylester chloride, Et₃N, CH₂Cl₂, 0 °C-rt, 1 h.

^aThe F_4–6 haptens were conjugated to carrier proteins using carbodiimide chemistry. Hapten synthesis and conjugation are described in the Experimental Section.

Upon completion and characterization of the F_1–6 series, haptens were conjugated to lysine residues of carrier proteins using either in situ activation of the haptens’ carboxylic acid groups via standard carbodiimide (EDAC) chemistry (F₁, F₄, F₅, F₆) or conjugation using N-hydroxysuccinimide-activated hapten esters (F₂, F₃). To support future development of clinically viable GMP-grade formulations, the F_1–6 haptens were conjugated to Escherichia coli-expressed CRM (EcoCRM or CRM₁), CRM₁₉₇ (CRM₂), and sKLH. Carrier proteins were selected based upon our team’s experience using CRM and sKLH to generate vaccines against oxycodone and heroin, which are now being prepared for first-in-human clinical trials.^31,47 F_1–6 haptens were also conjugated to bovine serum albumin (BSA) to generate reagents to detect antibodies by enzyme-linked immunosorbent assay (ELISA). The hapten–protein conjugates were purified using size exclusion-based filtration methods (either dialysis or centrifuge filtration). When possible, hapten–protein conjugates were characterized using MALDI-TOF to determine the total number of haptens per protein (Supporting Table 1) or DLS to determine size and aggregation status (Supporting Figure 3). Because of the high molecular weight of sKLH, hapten–sKLH conjugates were not characterized by MALDI-TOF. The haptenization ratio of conjugates varied according to the choice of protein and conjugation method (Supporting Table 1). For instance, conjugation using N-hydroxysuccinimide-activated hapten esters yielded similar haptenization ratios as EDAC coupling in the case of F₃–CRM₂ and F₆–CRM₂. Yet, EDAC chemistry provided a viable method to achieve higher haptenization ratios (up to 18) when the F₁ hapten was conjugated to either CRM₁ or CRM₂. Despite their relatively low haptenization ratio, CRM conjugates containing the F_1–6 series showed in vivo efficacy, which is consistent with the first-generation F₁–sKLH.⁴⁴ All conjugates containing the F_1–6 series were formulated in the FDA-approved aluminum adjuvant (Alhydrogel) and tested for efficacy in mice and rats.

Vaccine Efficacy against Fentanyl in Mice.

Vaccines containing the F_1–6 series were first tested for efficacy against fentanyl in vivo in two independent cohorts of mice (Figure 2). Mice were immunized i.m. on days 0, 14, and 28, and 1 week after the third immunization, mice were challenged s.c. with fentanyl (0.05 mg/kg, Figure 2A,C, or 0.1 mg/kg, Figure 2B,D). Vaccine efficacy was assessed via attenuation of antinociception on a hot plate, a centrally mediated behavior commonly used to screen anti-opioid vaccines. In the first cohort, all conjugates effectively reduced fentanyl-induced antinociception after the 0.05 mg/kg fentanyl challenge (Figure 2A) and altered fentanyl distribution to the brain as measured by liquid chromatography paired with mass spectrometry (LC/MS, Figure 2C). As seen in the hot plate, F₄–sKLH was not as effective as the other vaccine formulations in altering the distribution of fentanyl in the brain. In the second cohort, mice were vaccinated with F₁–sKLH, F₁–CRM₁, or F₃–CRM₂. Here, all fentanyl-based haptens produced significant reductions in fentanyl-induced analgesia (Figure 2B). These effects were confirmed by analysis of brain fentanyl concentration (Figure 2D).

Figure 2.

Vaccine efficacy against fentanyl in mice. Conjugates containing the F_1–6 haptens were tested in two independent cohorts of BALB/c mice (n = 5–10, each group). Haptens were conjugated to either sKLH, CRM₁, or CRM₂, adsorbed on aluminum adjuvant and injected i.m. on days 0, 14, and 28. (A) A week after the last immunization, the first cohort of mice receiving conjugates containing F_1–6 was challenged with 0.05 mg/kg s.c. fentanyl. Vaccination reduced fentanyl-induced antinociception in the hot plate test at 30 min post challenge. (B) Conjugates containing the F₁ and F₃ haptens were evaluated in a second independent cohort challenged with 0.1 mg/kg s.c. fentanyl and showed reduced fentanyl-induced antinociception in the hot plate test. (C, D) In both cohorts, vaccines decreased the distribution of fentanyl to the brain compared to control. Data are mean±SEM. Statistical symbols: * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, and **** p ≤ 0.0001 compared to control.

These data suggest that the retention of the core structure of norfentanyl with a nonamide-based linking chemistry was sufficient to generate an effective vaccine response. Previous studies demonstrated the feasibility of targeting both fentanyl and carfentanil with the same hapten structure,^38,48 supporting the feasibility of synthesizing haptens that target multiple compounds within the fentanyl-like family.

Vaccine Efficacy against Fentanyl and Its Analogues in Rats: Antinociception, Respiratory Depression, and Bradycardia.

Testing vaccine efficacy in rats allows for a wider range of behavioral and physiological measures than mice. Two independent studies tested the efficacy of vaccines containing either F_1–3 or F_4–6 haptens against fentanyl and selected analogues in two cohorts of rats. Fentanyl analogues are increasingly involved in opioid-related fatal overdoses, and it is imperative to understand the cross-reactivity of lead haptens with various analogues to assess the potential for cross-protection. In this study, rats were vaccinated i.m. on days 0, 21, and 42. Starting a week after the last immunization, rats were challenged weekly with fentanyl, alfentanil, or sufentanil. During the first week, the first cohort of rats was challenged s.c. with 0.075 mg/kg fentanyl. Vaccine efficacy was measured as the reduction of antinociception via the hot plate test and reduction of respiratory depression and bradycardia via pulse oximetry (Figure 3A–C). All vaccines, except for F₁–sKLH, reduced opioid-induced antinociception to nearly baseline levels (Figure 3A) throughout the testing period. In this assay, the new vaccine formulations were more effective than F₁-sKLH (Figure 3A). All vaccines, including F₁–sKLH, were effective in preventing opioid-induced respiratory depression measured as oxygen saturation compared to control (Figure 3B). While F₁–sKLH was not effective in reducing bradycardia in rats, F₁–CRM₁, F₁–CRM₂, F₂–sKLH, and F₃–sKLH were effective in this test (Figure 3C). These data suggest that the F₁ hapten is just as effective as F₂ and F₃ when conjugated to CRM, instead of sKLH. Indeed, it is possible that the hydrophobicity of F₁ may affect the quality of the conjugation to sKLH.

During the second week, rats were challenged s.c. with 0.5 mg/kg alfentanil, which is a dose that induced antinociception on the hot plate assay (MPE% = 100), as well as respiratory depression (up to SaO₂% = 50), and bradycardia in both control and vaccinated rats (data not shown). These data suggest that the F_1–3 hapten series does not induce antibodies cross-reactive with alfentanil, as confirmed by competitive binding ELISA (Supporting Table 2). Similar data displayed lack of cross-reactivity for remifentanil in vitro (Supporting Table 2).

During the third week, vaccinated rats were challenged s.c. with 0.008 mg/kg of sufentanil. F₁ conjugated to either sKLH or CRM₁, as well as F₃–sKLH, were ineffective at reducing MPE% at 15 min post challenge, while F₁–CRM₂ and F₂–sKLH significantly reduced antinociception compared to either control or F₁-sKLH (Figure 4A). At 30 and 45 min post challenge, F₂–sKLH was more effective than either control or F₁-sKLH (Figure 4B,C). All new vaccine formulations were more effective than F₁–sKLH in reducing sufentanil-induced antinociception at 60 min post challenge (Figure 4D). This time-dependent increase in vaccine efficacy may reflect the short half-life of sufentanil, which is illustrated by the trend toward reduced antinociception over time in the control rats (Figure 4D). These data provide evidence that conjugates containing the F_1–3 series elicit antibodies cross-reactive with sufentanil, although their affinity for fentanyl was greater (Supporting Table 2). None of the vaccine formulations were effective in reducing sufentanil-induced respiratory depression or bradycardia in this first cohort of rats (data not shown).

Figure 4.

Vaccine efficacy against sufentanil in rats. On week 2, rats immunized with conjugates containing the F_1–3 haptens were challenged s.c. with 0.008 mg/kg sufentanil. Selected conjugates containing F₁, F₂, and F₃ haptens were effective in reducing sufentanil-induced antinociception in the hot plate test measured at (A) 15, (B) 30, (C) 45, and (D) 60 min post drug challenge. These rats are the same subjects as shown in Figure 3 (n = 6). Data are mean±SEM. Statistical symbols: * or # p ≤ 0.05, ** or ## p ≤ 0.01, ### p ≤ 0.001 compared to either control (*) or F₁-sKLH (#).

During the fourth week, rats received a final fentanyl challenge of 0.1 mg/kg, s.c. At this higher fentanyl dose, the F₁–sKLH vaccine was not as effective as the other formulations at preventing respiratory depression and bradycardia (Figure 3D,E). During this fourth week of testing, fentanyl-induced respiratory depression in control rats (Figure 3D) was reduced compared to the previous fentanyl challenge (Figure 3B), indicating that these animals likely developed tolerance to fentanyl upon repeated exposure. Nevertheless, all vaccines blocked fentanyl-induced respiratory depression, bradycardia, and reduced fentanyl distribution to the brain compared to controls (Figure 3F).

In a second cohort of rats, the F_4–6 hapten series was tested in vivo following a similar strategy as for F_1–3 above, and the first-generation F₁–sKLH was included as a positive control. After the final immunization, rats were challenged s.c. once a week with either 0.1 mg/kg of fentanyl or 0.008 mg/kg of sufentanil. Each week, rats were randomized for drug allocation and then challenged a week later with the alternate drug. While F₁–sKLH displayed modest efficacy against fentanyl, vaccines containing F_4–6 haptens were more effective than F₁–sKLH in reducing fentanyl-induced antinociception as well as respiratory depression and bradycardia (Figure 5A–C). Only F₆-CRM₂ showed efficacy (p ≤ 0.05) in reducing sufentanil-induced antinociception at 15 and 30 min post drug challenge (Figure 5D). Although vaccination with F₆–CRM₂ showed a trend toward protection from sufentanil-induced respiratory depression and bradycardia, these data were not statistically significant (Figure 5E,F).

Figure 5.

Efficacy of vaccines containing haptens F_4–6 against fentanyl and sufentanil in rats. Conjugates containing the F_4–6 haptens conjugated to either sKLH or CRM₂ were tested in Sprague Dawley rats (n = 6, each group). Conjugates were injected i.m. on days 0, 21, 42, and 63. A week after the third vaccination, rats were challenged s.c. weekly with either 0.1 mg/kg fentanyl or 0.008 mg/kg sufentanil. Vaccination reduced the effects of fentanyl: (A) antinociception in the hot plate test, (B) respiratory depression reported as the percentage (%) of oxygen saturation measured by oximetry, and (C) bradycardia reported as heart rate (bpm) measured by oximetry. (D) F₆–CRM₂ was effective in reducing sufentanil-induced antinociception in the hot plate test. Conjugates showed a trend toward reducing sufentanil-induced: (E) respiratory depression and (F) bradycardia. Data are mean±SEM. Statistical symbols: * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, and **** p ≤ 0.0001 compared to control.

On the third week, all rats in the F_4–6 cohort were challenged s.c. with a final dose of 1 mg/kg fentanyl and were monitored by pulse oximetry at 15 and 30 min post drug challenge, followed by collection of blood and brain for measurement of fentanyl concentrations (Supporting Figure 6). While this higher dose was intended to assess vaccine efficacy in the event of an overdose of fentanyl, no evidence of vaccine efficacy against bradycardia was detected (data not shown) and limited evidence of vaccine efficacy against respiratory depression was found. Here, rats immunized with F₆–CRM₂ showed a significant degree of degree of protection against fentanyl-induced reduction in breath rate at 15 min post challenge (Supporting Figure 6A,B). Although vaccination did not reduce the concentration of fentanyl in the brain compared to control at this higher dose (Supporting Figure 6C), vaccinated rats displayed a significant decrease in the ratio of brain-to-serum drug concentration compared to control (Supporting Figure 6D), providing further evidence for vaccine efficacy.

Overall, these data indicate that vaccines containing F_1–6 haptens are effective against fentanyl and their efficacy may be improved upon conjugation to alternative carrier proteins, as demonstrated by the increased efficacy of F₁–CRM₁ and F₁–CRM₂ over F₁–sKLH. Furthermore, F₁–CRM₂, F₂–sKLH, F₃–sKLH, and F₆–CRM₂ show the potential to generate antibodies that cross-react with sufentanil. While these vaccines were equally effective in blocking fentanyl-induced antinociception, respiratory depression, and bradycardia, the same vaccines were only effective in blocking sufentanil-induced antinociception but not some of its other pharmacological effects. It is possible that these differences in efficacy are related to distinct pharmacodynamic/pharmacokinetic (PD/ PK) profiles of fentanyl and sufentanil,^7,49,50 which may differentially influence analgesia and respiratory depression. Future studies will focus on further improvement of these lead vaccine formulations and in-depth characterization of their efficacy in counteracting behavioral, physiological, and toxic effects of fentanyl and analogues over a wider dose range and across routes of administration.

Therapeutic Vaccination Reduced Intravenous Fentanyl Self-Administration.

Drug self-administration assays in nonhumans are considered the gold standard for evaluating the efficacy of medications for SUD. Preclinical studies have shown that vaccines can reduce acquisition, maintenance, and reinstatement or relapse of heroin and oxycodone intravenous self-administration in rats and nonhuman primates.^36,51–53 Furthermore, vaccines attenuated fentanyl self-administration (FSA), fentanyl vs food choice, and other operant behaviors in rats and nonhuman primates.^54,55 In the current study, the lead F₁–CRM₁ vaccine was used as a model to test whether therapeutic vaccination would reduce FSA in rats. To this end, rats were trained to self-administer fentanyl (1 μg/kg/infusion) and then immunized with either CRM₁ or F₁–CRM₁ (n = 8/group) on days 0, 21, 42, and 63. Baseline mean infusion rate, active responses, and inactive responses were 43.8 (±4.7 SEM), 282.9 (±93.7 SEM), and 3.8 (±1.6 SEM), respectively, for the CRM₁ control group and 47.3 (±8.7 SEM), 261.4 (±92.9 SEM), and 1.6 (±1.1 SEM), respectively, for the F₁–CRM₁ group (Figure 6A and Supporting Table 3). Rats immunized with F₁–CRM₁ showed a significant decrease in FSA after immunization, whereas the control CRM₁ group showed no change (Figure 6A). Because the vaccine only produced a partial reduction in FSA, rats received a fifth immunization on day 84 and were then exposed to a progressive weekly reduction in the fentanyl unit dose (Figure 6B,C) to determine the extent to which vaccine effects increase at lower unit doses. In rats vaccinated with F₁–CRM₁, FSA continued to decrease as the fentanyl dose decreased and did not differ from saline (0 μg/kg/infusions) at any dose. In contrast, FSA was maintained above saline at all fentanyl doses in control rats immunized with CRM₁. Thus, all fentanyl doses were reinforcing in control rats but not rats immunized with F₁–CRM₁. Although FSA tended to increase in control rats as the fentanyl dose decreased, those increases were not statistically significant. These data provide initial supporting evidence that active immunization with anti-fentanyl vaccines could be potentially initiated in patients with ongoing OUD related to fentanyl as part of a strategy to treat OUD. In addition, these data provide early evidence that vaccinated individuals may not show a sustained increase in their fentanyl intake to overcome the effects of vaccination.

Figure 6.

Active immunization reduced fentanyl intravenous self-administration (FSA) in rats. Rats were trained to self-administer fentanyl (1 μg/kg/infusion) and then immunized with either CRM₁ (open bars) or F₁–CRM₁ (closed bars, n = 8/group) on days 0, 21, 42, and 63. (A) Mean number of infusions (±SEM) earned in each group at baseline prior to immunization (Pre) and 3 weeks after the fourth immunization (Post). Rats immunized with F₁–CRM₁ showed a significant decrease in FSA compared to baseline, whereas the control group showed no change. Statistical symbols: * p < 0.05 compared to Pre, ^∧ p < 0.1 compared to CRM₁. Rats were then immunized once more and exposed to a progressive weekly reduction in the fentanyl unit dose. (B) Mean number of infusions (±SEM) in each group during dose reduction. FSA did not differ from saline (0 μg/kg/infusions) at any fentanyl dose in rats vaccinated with F₁–CRM₁, whereas FSA was maintained above saline at all fentanyl doses in control CRM₁ rats. Although FSA tended to increase in controls as fentanyl dose decreased, those increases were not statistically significant. Statistical symbols: * p ≤ 0.05 or ^∧ p < 0.1 compared to F₁–CRM₁, ^# p < 0.05 or ^## p < 0.01 compared to 0.0 μg/kg/infusion. (C) Data from (B) expressed as a percentage of baseline. Statistical symbols: ** p < 0.01 compared to F₁–CRM₁, ## p < 0.01 or ### p < 0.001 compared to 0.0 μg/kg/infusion.

Pre-Existing Immunity against Carrier Proteins Does Not Interfere with Vaccination against Fentanyl.

Because it is likely that human subjects have been previously exposed to common carrier proteins through standard pediatric or occupational immunization (e.g., diphtheria or tetanus prophylaxis), we tested whether pre-exposure to carrier proteins would interfere with vaccination. A similar study previously found that pre-existing immunity to carrier proteins may affect the efficacy of antinicotine vaccines.⁵⁶ Here, mice were first immunized i.m. on day −14 with either saline control or 60 μg of carrier protein and then immunized using the corresponding F₁–carrier protein conjugate on days 0, 14, and 28. Pre-exposure to either sKLH or CRM₂ had no effect on fentanyl-specific serum IgG antibody titers, while pre-exposure to CRM₁ negatively impacted the development of fentanyl-specific antibody titers (Supporting Figure 4). These data indicate that both sKLH and CRM₂ are viable carrier proteins for future vaccine development as pre-exposure in animals did not significantly diminish vaccine efficacy. Yet, it is critical to acknowledge that it is challenging to predict the potential effect of pre-existing immunity on vaccine efficacy in human subjects based upon preclinical data. In regard to sKLH, patients are not likely to have been previously exposed to KLH-based products, which is considered a neoantigen.⁵⁷ In regards to CRM or other commonly used carriers such as tetanus toxoid (TT), patients are likely to have been previously exposed to vaccine formulations containing CRM or TT from different sources. In determining which carrier to use for clinical development, choices will be based on manufacturing or regulatory aspects and ultimately be determined in clinical studies.

Vaccination against Fentanyl Does Not Interfere with Anesthesia.

To provide proof of safety and to guide clinical use of anti-fentanyl vaccines, it is critical to identify lead vaccines that do not interfere with medications for OUD, standard critical care, and pain management. To determine whether anti-fentanyl vaccines would interfere with anesthesia, rats were immunized with either CRM₁ or F₁–CRM₁ and then challenged weekly with a series of anesthetics. Anesthetic efficacy was measured by respiratory depression and bradycardia. Vaccination with F₁–CRM₁ did not interfere with anesthesia induced by dexmedetomidine (0.25 mg/kg, Figure 7A–C) and did not prevent its reversal by the standard dexmedetomidine-reversal agent atipamezole (1 mg/kg). Accordingly, no differences between control and active vaccine groups were found when rats were anesthetized with ketamine (75 mg/kg, Figure 7D,E), propofol (100 mg/kg, Figure 7F), or isoflurane (Figure 7G,H). Following multiple challenges with anesthetics, the F₁–CRM vaccinated rats retained efficacy against fentanyl (0.05 mg/kg, Figure 7I–K). These data demonstrate that F₁–CRM₁ is selective for fentanyl and does not interfere with commonly used anesthetics. While F₁–CRM₁ is expected to provide results predictive of other F₁-containing vaccines, future studies will focus on the characterization of lead vaccine formulations containing other promising haptens (e.g., F₂ or F₆).

Figure 7.

Immunization against fentanyl does not interfere with anesthesia protocols. Sprague Dawley rats (n = 6, each group) were immunized i.m. with either CRM₁ or F₁–CRM₁ on days 0, 21, 42, and 63. From day 49, rats were challenged weekly with 2% inhaled isoflurane, 0.25 mg/kg dexmedetomidine (reversed by 1 mg/kg atipamezole), 75 mg/kg ketamine, 100 mg/kg propofol, or 0.05 mg/kg fentanyl as control. Induction of anesthetic efficacy was initially monitored by the loss of righting reflex and then measured by respiratory depression reported as percent (%) oxygen saturation and breath rate (brpm), and bradycardia reported as heart rate (bpm), both measured by pulse oximetry. Results were equivalent in CRM₁ or F₁–CRM₁ groups, confirming the selectivity of these vaccines. (A–C) dexmedetomidine, (D, E) ketamine, (F) propofol, (G, H) isoflurane, and (I–K) fentanyl as positive controls. Data are mean±SEM. Statistical symbols: *** p ≤ 0.05 and ** p ≤ 0.01.

Vaccination against Fentanyl Does Not Interfere with the Pharmacological Activity of Agonists and Antagonists.

To test whether anti-fentanyl vaccines would interfere with opioid receptor agonists and antagonists used in the treatment of OUD and pain management, rats were immunized with either CRM₁ or F₁–CRM₁ and challenged weekly with a series of agonists or antagonists. Efficacy of opioids was measured by antinociception on the hot plate. In this experiment, rats were challenged sequentially with either oxycodone (2.25 mg/kg, s.c.) or heroin (0.9 mg/kg, s.c.), and antinociception was measured 30 min post challenge. Immediately after, rats were given naloxone (0.1 mg/kg, s.c.) to reverse opioids’ effects. Vaccination with F₁–CRM₁ did not interfere with antinociception induced by either oxycodone (Figure 8A,C) or heroin (Figure 8B,D) and did not impact the efficacy of naloxone in reversing either oxycodone or heroin effects (Figure 8A–D). Next, rats were challenged with methadone (2.25 mg/kg, s.c.) and showed that methadone-induced antinociception was not different between CRM₁ and F₁–CRM₁ (Figure 8E). As a positive control, a final challenge with fentanyl (0.1 mg/kg, s.c.) confirmed that the efficacy of F₁–CRM₁ was preserved against its target opioid (Figure 8F). These data indicate that vaccination with F₁–CRM₁ does not interfere with the pharmacological activity of oxycodone and methadone, as well as naloxone reversal of the effects of oxycodone and heroin. In vivo data were further supported by in vitro competitive binding studies demonstrating that polyclonal antibodies induced by conjugates containing the F_1–6 hapten series did not cross-react with buprenorphine, methadone, naloxone, or naltrexone (Supporting Table 2, panel C). Hence, it is expected that the clinical implementation of vaccines will not impact the use of current medications for OUD.

Figure 8.

Vaccination against fentanyl does not interfere with the effects of methadone, heroin, oxycodone, and naloxone in rats. Sprague Dawley rats (n = 6, each group) were immunized i.m. with either CRM₁ or F₁–CRM₁ on days 0, 21, and 42. One week after the third vaccination, rats were challenged s.c. weekly with (A, C) oxycodone (2.25 mg/kg), (B, D) heroin (0.9 mg/kg), (E) methadone (2.25 mg/kg), and (F) fentanyl (0.1 mg/kg) as controls. (A–D) Oxycodone and heroin effects were reversed by naloxone (0.1 mg/kg, s.c.). Drug-induced antinociception, or its reversal by naloxone, was assessed in the hot plate test of analgesia. Vaccination did not interfere with the antinociceptive effects of oxycodone, heroin, or methadone. The effect of naloxone was preserved in both control and vaccinated groups. Data are mean±SEM. Statistical symbols: * p ≤ 0.05, ** p ≤ 0.01, and **** p ≤ 0.0001.

Characterization of F₁ Hapten at Mu Opioid Receptor (MOR).

Use of fentanyl-based haptens can present regulatory challenges for vaccine development. Haptens displaying reduced or no activity at the MOR may offer advantages during preclinical manufacturing and clinical development. Because the F₁ hapten yields some of the most promising vaccine candidates in this series, this compound was used as a model to test the ability of haptens to activate the human MOR in both in vitro and in silico assays. The F₁ hapten did not have activity at MOR in vitro in calcium mobilization assays (Supporting Figure 5). The lack of activity at MOR is most likely due to the lack of the N-phenylethyl substituent and the extended tetraglycine linker (Supporting Figure 5). Instead, the control F₄ hapten retaining the N-phenylethyl substituent showed activity at MOR in the same assay (Supporting Figure 5). Future studies will expand upon these results to test whether there is a predictive relationship between individual hapten’s activity at MOR and its efficacy against the target opioid in vivo. Such information may guide the rational design and screening of haptens and hapten–carrier conjugates. Next, the F₁ was characterized for its potential to interact and activate the MOR in silico (Figure 9). To this end, the F₁ hapten was docked into the Kolbilka crystal structure (PDB: 5C1M)⁵⁸ for the agonist μ opiate receptor complexed with BU72 stabilized by a G-protein camelid antibody fragment Nb39, which increases the agonist affinity of BU72 from 470 to 16 pM.⁵⁸ Consistent with Figure 9A,B, the F₁ hapten showed the lowest Gscore pose. F₁ loosely fits the binding site with its core buried in the same region as prototypical MOR agonists fentanyl (Figure 9B) and BU72 (Figure 9A) and with its [Gly]₄ linker extending toward the extracellular loops. It is noteworthy that most F₁ poses did not possess the strong hydrogen bonding and π–cation interactions with D147/Y148 that BU72 and fentanyl induce. In 1–2 cases of less favorable Gscore/Emodel F₁ docking poses, F₁ picked up a hydrogen bond interaction with D147 but only at the expense of considerable ligand strain energy, as indicated by positive Emodel Glide score values. The Gscore/Emodel scores parallel the EC₅₀ values of the ligands, as shown in Figure 9D. Although docking scores are not robust predictors of activation, they often parallel molecular mechanics generalized Born surface area (MMGBSA) relative binding free-energy scores. The qualitative conclusion that can be drawn from these results is that F₁ has some of the shape prerequisites to recognize the orthosteric binding site but does not possess key interactions that would induce activation changes in the receptor ensemble. Figure 9D details the MOR functional activity for F₁, fentanyl, and BU72, which demonstrate the lack of functional activation for F₁ at the MOR, indicating a favorable profile for progression. This preliminary information will guide future studies testing whether docking scores and limited favorable MOR interactions may predict hapten activity in vivo.

Figure 9.

Characterization of F₁ hapten at Mu opioid receptor (MOR). The crystal structure of the MOR receptor stabilized by the Nb39 antibody (PDB: 5C1M) and top GLIDE-XP poses for (A) BU72 and (B) fentanyl agonists juxtaposed with (C) the best scoring pose for the F₁ hapten and the interaction legend for panels (A–C) is shown. (D) GLIDE-XP Gscore, Emodel scores for these poses, and corresponding EC₅₀ for MOR activation in calcium mobilization assays in vitro.

DISCUSSION

Prophylactic or therapeutic vaccination offers a promising strategy to treat OUD and to prevent overdose, which will hopefully extend current treatment options available to patients and physicians. Although a first-generation F₁–sKLH vaccine displayed promising efficacy against fentanyl in mice and rats, the current study focused on improving upon F₁–sKLH by developing a series of next-generation vaccines containing the previously reported F₁ and the new F_2–6 haptens conjugated to sKLH and CRM carrier proteins and formulated in the clinically approved aluminum adjuvant. Compared to a previous report,⁴⁴ the use of the F₁ hapten in combination with alternative carriers resulted in conjugates that showed a reduced particle size distribution and improved haptenization ratio. Conjugate vaccines containing F₁ were compared to conjugates containing the F_2–6 haptens in vivo. In mice, all vaccines were equally effective in blocking fentanyl-induced antinociception and reducing fentanyl distribution to the brain, except for a less effective conjugate containing the F₄ hapten. Vaccines were advanced to testing in rats, which allowed for screening efficacy against clinically relevant parameters such as fentanyl-induced respiratory depression (measured as oxygen saturation or breath rate) and bradycardia (measured as beats per minute). In rats, vaccines containing F_1–6 haptens were effective in reducing or blocking fentanyl-induced antinociception as well as respiratory depression and bradycardia, which are primary contributors to opioid-induced fatal overdoses. Proof of efficacy against respiratory depression for F₁–sKLH was previously demonstrated using a cumulative dosing paradigm involving multiple doses up to 0.1 mg/kg of fentanyl in rats.⁴⁴ The present study identified vaccines that were more effective than F₁–sKLH in reducing fentanyl-induced antinociception, respiratory depression, and bradycardia in rats challenged with acute fentanyl doses (0.075–0.1 mg/kg, s.c.). Furthermore, in a subset of vaccinated rats challenged with an additional higher fentanyl dose (1 mg/kg, s.c.), F₆–CRM₂ showed encouraging efficacy against fentanyl-induced reduction in breath rate and reduced the magnitude of change in the brain-to-serum fentanyl concentration ratio (Supporting Figure 6). In control rats, this dose resulted in serum concentrations of fentanyl of 245 ± 38 ng/mL (n = 5), which are 10- to 200-fold greater than the reported peripheral blood drug concentrations (1–20 ng/mL, with a median of 5–10 ng/mL) in fentanyl-related fatal overdoses in humans.⁵⁹ Overall, these data support proof of efficacy for vaccines against fentanyl at clinically relevant doses and warrant further investigations.

Because the goal of this study was to identify vaccines targeting fentanyl and possibly its structurally related analogues, the efficacy of vaccines containing F_1–6 haptens against fentanyl analogues was also assessed by challenging vaccinated rats with sufentanil and alfentanil. These target compounds are prescription drugs classified by the Drug Enforcement Agency (DEA) as Schedule II Controlled Substances, and are also commonly misused or diverted for illicit use. While the range of analogues tested may appear limited, one of the strengths of the current study is that the efficacy of vaccines against fentanyl and its analogues (and off-target opioids) was examined primarily in vivo. While in vitro characterization of binding (e.g., K_d or IC₅₀) for target opioids and off-target compounds may provide useful information, in vitro assays may not necessarily account for in vivo potency when comparing compounds with different behaviorally active doses (e.g., heroin vs fentanyl, or fentanyl vs carfentanil). Despite no protective effects or evidence of in vitro binding found against alfentanil or remifentanil, the F₁–CRM₂, F₂–sKLH, and F₆–CRM₂ vaccines showed promising efficacy against sufentanil, probably because it is closer in structure to fentanyl than to alfentanil. Future studies will expand our vaccine development and screening strategy to target other selected analogues of interest (e.g., acetylfentanyl or carfentanil) and possibly include new fentanyl analogues becoming increasingly abused or found in illicit street mixtures. For instance, it has been shown that vaccines can be designed to target carfentanil, cyclopropyl fentanyl, (±)cis-3-methyl fentanyl, para-fluorofentanyl, and furanyl fentanyl.^42,43 The high selectivity of these vaccines for specific analogues within the fentanyl-like chemical family suggests that multivalent vaccine formulations may be required to target structurally unique fentanyl analogues, such as alfentanil or carfentanil, which are not as structurally similar to fentanyl as other analogues.

A qualitative assessment of vaccine performance was conducted using a simple 0–3 scoring system for various parameters such as efficacy in reducing drug distribution to the brain, antinociception, respiratory depression, and bradycardia (Supporting Table 4). Because only F₁–CRM₁ was tested in the FSA test, this parameter was not included in this overall evaluation of vaccine candidates. Based on these scores and the additional proof of efficacy for the F₆ conjugate in the 1 mg/kg fentanyl challenge (Supporting Figure 6), the F₁–CRM₂, F₂–sKLH, and F₆–CRM₂ adsorbed on alum adjuvant were identified as the lead candidates in terms of in vivo efficacy against fentanyl and sufentanil. Future studies will compare the efficacy of F₁, F₂, and F₆ conjugated to either sKLH or CRM and delivered in various adjuvant formulations or other immunomodulatory platforms. These vaccines will be further tested to establish leads for manufacturing and clinical evaluation.

One caveat of this study is that vaccines were screened using single acute doses of fentanyl in mice (0.05–0.1 mg/kg) or multiple acute doses within a limited dose range in rats (0.075–1 mg/kg, fentanyl, and 0.008 mg/kg sufentanil). In this context, acute dosing provided a high-throughput testing strategy to identify promising candidates for more advanced studies. Future studies will employ more exhaustive challenge models to determine which vaccine formulation is the most effective in shifting the drug’s potency.^54,60,61 One of the strengths of this study is that vaccines were screened over several weeks, using different in vivo assays. While it would be ideal for rats to be screened over a period of several months to identify the window of protection offered by these vaccines, this study showed that repeated fentanyl challenges resulted in tolerance to its physiological effects in control rats and could complicate the interpretation of data and determination of vaccine efficacy. To mitigate some of the concerns regarding tolerance, rats were exposed to repeated weekly randomized drug challenges (e.g., fentanyl or other analogues).

The different performances of the various conjugates containing the F_1–6 series may be attributed to (1) hapten structure, (2) linker chemistry, composition, and anchor points, and (3) choice of carrier (e.g., sKLH vs CRM). First, the properties of structurally distinct haptens may affect the physical and chemical properties of the conjugate such as haptenization ratio, size, aggregate status, stability over time, and interaction with adjuvants or other delivery platforms. These features may affect antigen presentation to the innate and adaptive immune systems. Second, the linker properties likely contributed to antigen presentation. For instance, the F₁ hapten includes a long flexible tetraglycine linker, which has been successfully used in other opioid-based haptens and vaccines targeting heroin and oxycodone. In contrast, the F₂ and F₃ haptens utilized relatively “rigid” linking chemistry, which may force the antigen away from the carrier protein and help with presentation. It has been previously demonstrated that vaccine efficacy changes according to flexible or constrained hapten conformations.^62,63 Although F₄ and F₅ are the same molecule with different linking groups, F₄-sKLH performed differently than F₅-sKLH in vivo. The linker for F₄ is shorter and more hydrophobic compared to the more water-soluble F₅. Overall, it is possible that minor structural differences in the hapten structure, such as the presence of hydrogen bond donors or acceptors, may affect hapten recognition at the B cell receptor (BCR) on the surface of B cell lymphocytes and, subsequently, the immunogenicity of the hapten–carrier conjugate. Differing only by a single nitrogen, F₄ and F₅ may display such functional differences. Similarly, attachment of the same linker to either the para (F₅) or meta (F₆) position may result into different in vivo profiles. Although no structural data are available for opioids or opioid-based haptens bound to the BCR of opioid-specific B cells, it has been shown that the binding of an antiphencyclidine (PCP) IgG monoclonal antibody fragment (Fab) to PCP is determined by structural features including the formation of a hydrogen bond between a carbonyl oxygen atom on the Fab and the lone nitrogen atom of PCP.⁶⁴ Similarly, the binding of nicotine to a nicotine-specific Fab was largely driven by hydrophobic contact with the nicotine rings including the charged pyrrolidine nitrogen.⁶⁵ In support of this hypothesis, minimal modifications of hapten structures such as fluorination or deuteration^66,67 seem to impact vaccine efficacy against their respective drug targets. It has been consistently shown that oxycodone is a more immunogenic hapten core than hydrocodone, despite a minor structural difference consisting of a -OH vs -H at the C14 position on the morphinan structure.^32,34 Differences between the original pharmacophore and the cognate hapten structure may alter their ability to engage the BCR of hapten-specific B cells, as supported by evidence that a higher frequency of B cells specific for oxycodone-based haptens is found in the murine preimmunization repertoire compared to hydrocodone-based haptens.³² These data support the hypothesis that hapten design should take into account the effect of hydrogen bonds, and that donor or acceptor moieties could be rationally incorporated into the hapten structure to improve antigen presentation.

Third, another factor for determining vaccine efficacy relates to the use of different carrier proteins. Compared to the first-generation F₁ vaccine conjugated to sKLH, F₁ conjugated to CRM demonstrated higher efficacy against fentanyl and sufentanil and performed similarly to other hapten–carrier conjugates. These data illustrate the importance of optimizing each hapten–carrier combination by comparing the efficacy of a lead hapten conjugated to alternative proteins or other immunogenic carriers. For instance, the identity of the carrier protein may affect both the quantity and quality of the antibody response by differential activation of B cells and engagement of their cognate CD4⁺ T cell populations. It has been shown that the preimmunization frequency of carrier-specific CD4⁺ T cells correlates with vaccine efficacy against oxycodone⁶⁸ and that an oxycodone-based hapten conjugated to sKLH, KLH, CRM, or TT had varying efficacy in inducing oxycodone-specific B cell population subsets and had significantly different impact in decreasing oxycodone-induced effects in vivo.^35,47 Further characterization of these interactions will allow us to identify promising hapten–carrier combinations suitable for clinical evaluation.

A critical finding in this study was that F₁–CRM₁ vaccine reduced the maintenance of fentanyl self-administration. This is consistent with numerous studies by us and others showing that similar opioid-specific vaccines can reduce acquisition, maintenance, and reinstatement/relapse of oxycodone, morphine, heroin, or fentanyl self-administration in rats.^{36,52,53,61,69} This effect is attributed to the binding of fentanyl to antibodies in serum, resulting in lower fentanyl concentrations in brain and a decrease in activation of MORs in brain areas that mediate reinforcement. The effect of vaccination on fentanyl pharmacokinetics in mice and rats (e.g., Figures 2 and 3) supports this interpretation. Additionally, vaccination significantly suppressed the dose–response curve for FSA, suggesting a reduction in the reinforcing potency of fentanyl. However, it may be necessary to extend FSA across a wider range of unit doses, including those that can surmount the binding capacity of the antibodies, to confirm that vaccination shifts the dose–response curve to the right. Furthermore, studies using higher clinically relevant doses will provide a more rigorous demonstration of efficacy.

One potential adverse effect of vaccination is a compensatory increase in drug intake to surmount the binding capacity of antibodies. This was observed in our prior study of heroin self-administration (HSA), in which vaccination with a morphine-conjugate vaccine (M-KLH) produced an increase in HSA at the training dose and a rightward shift in the HSA dose–response curve.⁵² Similar effects have been reported with nicotine and cocaine vaccines.^70,71 In contrast, FSA was reduced in vaccinated rats in the present study. Because much lower doses of fentanyl can maintain self-administration compared to heroin (i.e., fentanyl is more potent), the elicited antibodies are less likely to become saturated or surmounted under FSA conditions. Hence, it is predicted that fentanyl vaccines may be more effective than heroin vaccines against their targeted opioid in human studies. The present findings also suggested that compensatory increases in the consumption of fentanyl or other highly potent opioids (e.g., sufentanil) may not be as much of a concern in humans vaccinated against those opioids compared to vaccination against other less potent opioids. Future preclinical studies will examine this issue across a wider range of fentanyl unit doses and longer access periods resulting in levels of fentanyl intake that are more comparable to humans.

To provide proof of selectivity and safety as well as to inform clinical implementation, this study included an evaluation of the vaccination medications used in the treatment of OUD, pain management, and critical care. While prophylactic or therapeutic vaccination against fentanyl or other opioids will likely lead to overall improved outcomes in OUD treatment, prevention of relapse, or overdose, it is not intended as a single line of intervention. It is likely that patients who receive anti-fentanyl vaccines may seek additional medications for OUD, such as methadone or buprenorphine. The current study demonstrated that vaccination did not interfere with other opioids including methadone, oxycodone, and heroin in vivo. The opioid overdose reversal agent naloxone also retained its ability to reverse the effects of oxycodone and heroin. These in vivo data were further substantiated by demonstration that vaccine-induced polyclonal antibodies did not bind methadone, buprenorphine, naltrexone, and naloxone in vitro. These results support the notion that anti-fentanyl vaccines can be safely combined with medications for OUD, life-saving opioid antagonists, and analgesics such as oxycodone. Clinicians would be able to vaccinate patients with OUD de novo or in combination with other treatment medications for OUD and still maintain the ability to adequately manage pain. Furthermore, this study also evaluated vaccination in the context of common anesthetics, and found that vaccination did not interfere with the pharmacological effects of injectable and volatile anesthetics (i.e., dexmedetomidine, ketamine, propofol, and isoflurane) nor with a model anesthetic reversal agent, atipamezole. These data demonstrate preclinical selectivity and safety for vaccines in the context of critical care or emergency medicine, such as a surgical procedure.

To further characterize critical vaccine components, this study tested whether haptens would activate MOR in vitro and in silico. These experiments focused on the well-characterized F₁ hapten to provide initial proof of concept for this approach. First, the ability of the F₁ hapten in activating the human MOR was evaluated using a cell-based calcium mobilization assay. The F₁ hapten lacking the N-phenylethyl moiety did not activate the MOR compared to the MOR agonist morphine (Figure 9 and Supporting Figure 5). In contrast, the F₄ hapten retained activity at the MOR, which may be related to the presence of the N-phenylethyl moiety. These data suggest that hapten screening and vaccine development may benefit from incorporation of in vitro evaluation of hapten’s activity at the MOR, which may provide additional information for selecting haptens. For instance, knowledge of the hapten activity at MOR may facilitate its DEA classification status (or lack thereof), manufacturing, and regulatory path. Docking studies of the F₁ hapten into the crystal structure of the MOR confirmed that F₁ displays some of the shape prerequisites to recognize the MOR orthosteric binding site but does not engage in the significant key interactions, such as hydrogen bonds, necessary to induce activation changes in the receptor ensemble. Because activation of MOR (a 7-transmembrane receptor) by an opioid requires interaction with an active conformation of the receptor, any modifications to the molecular structure during hapten design have the potential to disrupt drug–receptor interaction. Future studies will expand this strategy beyond F₁ to test whether MOR activation by a series of haptens in vitro or in silico may be predictive of their ability to generate opioid-specific antibodies capable of sequestering opioids in vivo.

CONCLUSIONS

The present study focused on the development of vaccines targeting fentanyl and its structurally related analogues. Together, these data establish the preclinical efficacy, selectivity, and safety of anti-fentanyl vaccines in vivo and support their use as a viable prophylactic or therapeutic strategy against OUD and opioid-induced toxicity. In addition, our findings, taken together with other published studies,^42,61 highlight the high selectivity of these vaccines for the targeted opioid, suggesting that the use of multivalent vaccination strategies to target multiple fentanyl analogues or mixtures of fentanyl combined with other opioids (e.g., heroin/fentanyl mixtures) could be effective. In support of this concept, it has been previously demonstrated that coadministration of vaccines targeting heroin and oxycodone,^31,72 or heroin and fentanyl³⁸ offers a viable strategy to offer protection against multiple opioids. Future studies will focus on optimization of multivalent vaccination strategies against OUD.

EXPERIMENTAL SECTION

Chemistry.

Synthesis and Characterization of the F_1–6 Haptens.

The F₁ hapten was synthesized as previously described.⁴⁴ F₂ was prepared starting from 4-bromo fentanyl derivative 1 (Scheme 1). Palladium-catalyzed cross-coupling of 1 with tert-butylacrylate afforded alkene 2 in good yield. Deprotection of tert-butyl ester 2 with trifluoroacetic acid gave the free acid 3. The active hapten F₂ was then prepared by coupling N-hydroxysuccinimide with free acid 3 and was used without further purification. F₃ was prepared following a similar route as F₂ but starting with free acid 4.⁷³ Haptens F₅ (para) and F₆ (meta) followed an identical synthetic pathway (Scheme 2). Mono tert-butoxy protected anilines 5a,b were subjected to reductive amination using tin(II) chloride with 4-phenylethyl piperidone to yield anilines 6a,b. Propionic anhydride was used to acylate the free amines 6a,b providing propionamides 7a,b in excellent yield. Subsequent deprotection of the aniline moieties using trifluoroacetic acid gave free amines 8a,b. Urea formation was then accomplished using triphosgene and glycine methylester providing 9a,b. Final hydrolysis of the methyl esters using lithium hydroxide afforded target haptens F₅ and F₆ in excellent yield. Target hapten F₄ was generated from key intermediate 8b (Scheme 2). Acylation of 8b with succinic acid monomethyl chloride gave ester 10 in good yield. Final hydrolysis using lithium hydroxide afforded target hapten F₄ in moderate yield.

Purity and characterization of compounds were established by a combination of MS, NMR, HPLC, and thin-layer chromatography (TLC) analytical techniques described below. NMR spectra were recorded on a Bruker Avance DPX-300 (300 MHz) spectrometer and were determined in chloroform-d (7.26 ppm) or methanol-d₄ (3.31 ppm) with tetramethylsilane (0.00 ppm) or solvent peaks as the internal reference unless otherwise noted. Chemical shifts are reported in ppm relative to the solvent signal, and coupling constant (J) values are reported in hertz (Hz). Thin-layer chromatography (TLC) was performed on EMD precoated silica gel 60 F254 plates. TLC spots were visualized with UV light or I₂ detection. Low-resolution mass spectra were obtained using a single quadrupole PE Sciex API 150EX (electrospray ionization (ESI)). Unless stated otherwise, all test compounds were at least 95% pure as determined by HPLC. HPLC method (F₂ and F₃ as measured by their hydrolytically stable precursors 3 and 4, respectively): spectra were collected using a 1260 Infinity Isocratic Pump with a ZirChrom-PBD column (50 mm × 2.1 mm i.d., 3 μm). The mobile phase was acetonitrile/10 mM ammonium acetate (v/v = 45/55) with 0.1 mM citrate (pH 4.4), and the temperature was 25 °C. The flow rate was set at 0.3 mL/min. HPLC method (F₄, F₅, F₆): an Agilent-Varian system equipped with Prostar 210 pumps, a Prostar 335 Diode UV detector, and a Phenomenex Synergi 4 μm Hydro RP 80A C18 250 mm × 4.6 mm column using a 20 min gradient elution of 5–95% solvent B at 1 mL/min followed by 5 min at 95% solvent B (solvent A, water with 0.1% TFA; solvent B, acetonitrile with 0.1% TFA and 5% water; absorbance monitored at 220 and 280 nm).

2,5-Dioxopyrrolidin-1-yl (2E)-3-(4-{N-[1-(2-phenylethyl)-piperidin-4-yl]propanamido}phenyl)prop-2-enoate (F₂).

(2E)-3-(4-{N-[1-(2-Phenylethyl)piperidin-4-yl]propanamido}phenyl)prop-2-enoic acid (3) (20 mg, 0.05 mmol) was added to a solution of 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide-HCl (12.5 mg, 0.064 mmol) and N-hydroxysuccinimide (8.5 mg, 0.074 mmol) in CH₂Cl₂ (10 mL), and the reaction was allowed to stir for 18 h at room temperature. EtOAc (15 mL) and H₂O (25 mL) were added to the reaction mixture, and the mixture was allowed to stir for 10 min. The mixture was transferred to a separatory funnel; the organic layer was extracted with EtOAC (25 mL), washed with brine, and dried with anhydrous MgSO₄. The solvent was removed under reduced pressure to give (F₂) (19 mg), which, due to hydrolytic instability of the NHS ester to analytical processes, was used without further purification or characterization. The purity of F₂ was indirectly assessed via HPLC analysis of its hydrolytically stable precursors 3, which free acid showed a percent purity of 97.0% (Supporting Figure 1).

2, 5-Dioxopyrrolidin-1-yl (2E)-3-(4-{2-[4-(N-Phenylpropanamido)piperidin-1-yl]ethyl}phenyl)prop-2-enoate (F₃).

(2E)-3-(4-{2-[4-(N-Phenylpropanamido)piperidin-1-yl]ethyl}-phenyl)prop-2-enoic acid (4 HPLC = 98% at 0.99 min) (100 mg, 0.25 mmol) was added to a solution of 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide-HCl (62.5 mg, 0.32 mmol) and N-hydroxysuccinimide (42.5 mg, 0.37 mmol) in CH₂Cl₂ (10 mL), and the reaction was allowed to stir for 18 h at room temperature. EtOAc (50 mL) and H₂O (50 mL) were added to the reaction mixture, and the mixture was allowed to stir for 10 min. The mixture was transferred to a separatory funnel; the organic layer was extracted with EtOAc (50 mL), washed with brine, and dried with anhydrous MgSO₄. The organic solvent was removed under reduced pressure to give F₃ (99 mg), which, due to hydrolytic instability of the NHS ester to analytical processes, was used without further purification or characterization. As described above for F₂, the purity of F₃ was indirectly assessed via HPLC analysis of its hydrolytically stable precursors 4, which free acid showed a percent purity of 97.7% (Supporting Figure 1).

Lithium 4-Oxo-4-((4-(N-(1-phenethylpiperidin-4-yl)-propionamido)phenyl)amino)-butanoate (F₄).

The procedure for the synthesis of F₄ was followed starting with ester 7 (14.8 mg, 0.03 mmol) to provide lithium salt 8 (11.6 mg). ¹H NMR (300 MHz, DMSO) δ 12.94 (s, 1H), 7.60 (d, J = 8.6 Hz, 2H), 7.29–7.19 (m, 2H), 7.19–7.12 (m, 3H), 7.08 (d, J = 8.6 Hz, 2H), 4.52–4.22 (m, 1H), 2.95–2.88 (m, 2H), 2.76–2.70 (m, 1H), 2.69–2.60 (m, 3H), 2.59–2.54 (m, 1H), 2.46–2.39 (m, 2H), 2.31–2.25 (m, 1H), 1.99 (t, J = 11.4 Hz, 3H), 1.83 (q, J = 7.4 Hz, 3H), 1.65 (d, J = 12.8 Hz, 2H), 0.90–0.82 (m, 3H). MS (ESI) m/z 452.00 [M + H]⁺. HPLC = 95% at 10.80 min (Supporting Figure 1).

Lithium 2-(3-(4-(N-(1-Phenethylpiperidin-4-yl)propionamido)-phenyl)ureido)acetate (F₅).

The procedure for the synthesis of 6a was followed starting with ester 5b (37 mg, 0.08 mmol) to provide lithium salt 6b (38.2 mg). ¹H NMR (300 MHz, DMSO) δ 7.62–7.44 (m, 2H), 7.29–7.07 (m, 5H), 6.97 (d, J = 8.6 Hz, 2H), 4.45–4.31 (m, 1H), 2.91 (d, J = 11.0 Hz, 2H), 2.68–2.59 (m, 2H), 2.47–2.39 (m, 3H), 1.98 (t, J = 11.1 Hz, 2H), 1.84 (q, J = 7.4 Hz, 2H), 1.69–1.56 (m, 3H), 1.31–1.14 (m, 2H), 0.87 (t, J = 7.4 Hz, 3H). MS (ESI) m/z 453.00 [M + H]⁺. HPLC = 96% at 10.70 min (Supporting Figure 1).

Lithium 2-(3-(3-(N-(1-Phenethylpiperidin-4-yl)propionamido)-phenyl)ureido)acetate (F₆).

To a solution of ester 9a (9.7 mg, 0.02 mmol) in THF/MeOH/H₂O (1 mL:1 mL:0.1 mL), LiOH (1 mg, 0.04 mmol) was added and the reaction was allowed to stir at room temperature for 16 h. The reaction was evaporated to dryness under a stream of N₂ to provide the lithium salt F₆ (11 mg, >100% crude). ¹H NMR (300 MHz, DMSO) δ 7.42 (s, 1H), 7.33–7.07 (m, 7H), 6.70–6.55 (m, 1H), 4.49–4.29 (m, 1H), 3.01–2.86 (m, 4H), 2.76–2.70 (m, 1H), 2.68–2.54 (m, 2H), 2.32–2.24 (m, 2H), 2.07–1.79 (m, 4H), 1.70–1.56 (m, 3H), 0.88 (t, J = 7.4 Hz, 3H). MS (ESI) m/z 453.00 [M + H]⁺. HPLC = 100% at 10.98 min (Supporting Figure 1).

tert-Butyl (2E)-3-(4-{N-[1-(2-Phenylethyl)piperidin-4-yl]-propanamido}phenyl)prop-2-enoate (2).

N-(4-Bromophenyl)-N-[1-(2-phenylethyl)piperidin-4-yl]propanamide (1) (270 mg, 0.6 mmol) and tert-butylacrylate (768 mg, 6 mmol, 10 equiv) were dissolved in Et₃N (2 mL) and N,N-dimethylformamide (DMF, 3 mL). The solution was deoxygenated by bubbling with dry N₂ gas for 30 min, and then a DMF solution containing Pd(OAc)₂ (13.5 mg, 0.06 mmol, 0.1 equiv) and ethylenebis(diphenylphosphine) (23.9 mg, 0.06 mmol, 0.1 equiv) was injected though a needle. The reaction mixture was further bubbled with N₂ for 30 min before being heated to 140 °C for 8 h. The reaction mixture was cooled to room temperature, and Et₂O (100 mL) was added. The organics were passed through a neutral alumina column to remove the Pd catalyst and were washed with 1 M NaOH, H₂O, and brine. The organic extracts were dried over MgSO₄ and concentrated under reduced pressure. Additional Et₂O was added (10 mL), and the product was purified by precipitation in Et₂O using 2.5 M HCl in EtOH (1 mL). The precipitate was filtered, dried, and isolated as a white solid (240 mg) (75%). ¹H NMR (500 MHz, CDCl₃): ¹H NMR: (500 MHz, CDCl₃): δ 7.56 (d, J = 16.0 Hz, 1H), 7.5–7.4 (m, 5H), 7.26 (d, J = 7.0 Hz, 2H), 7.12 (dd, J = 8.5, 8.5 Hz, 2H), 6.36 (d, J = 16.0 Hz, 1H), 4.82 (tt, J = 12.0, 4.0 Hz, 1H), 3.65 (d, J = 12.0 Hz, 2H), 3.27 (m, 2H), 3.14 (m, 2H), 2.84 (td, J = 12.0, 1.5 Hz, 2H), 2.2–1.9 (m, 6H), 0.55 (s, 9H), 1.04 (t, J = 7.5 Hz, 3H); δ MS: m/z 463.989 [M + H]⁺.

(2E)-3-(4-{N-[1-(2-Phenylethyl)piperidin-4-yl]propanamido}-phenyl)prop-2-enoic Acid (3).

tert-Butyl (2E)-3-(4-{N-[1-(2-phenylethyl)piperidin-4-yl]propanamido}phenyl)prop-2-enoate 2 (330 mg, 0.6 mmol) was dissolved in CH₂Cl₂ (40 mL) with slow addition of trifluoroacetic acid (2 mL). The solution was stirred at room temperature for 24 h. 1 N HCl (80 mL) was added, and the reaction mixture was extracted with CH₂Cl₂ (3 × 40 mL). The organic extracts were combined, washed with H₂O and brine, and concentrated under reduced pressure to afford 3 (275 mg (95%)). ¹H NMR: (500 MHz, CDCl₃): δ 7.56 (d, J = 16.0 Hz, 1H), 7.5–7.4 (m, 5H), 7.26 (d, J = 7.0 Hz, 2H), 7.12 (dd, J = 8.5, 8.5 Hz, 2H), 6.36 (d, J = 16.0 Hz, 1H), 4.82 (tt, J = 12.0, 4.0 Hz, 1H), 3.65 (d, J = 12.0 Hz, 2H), 3.27 (m, 2H), 3.14 (m, 2H), 2.84 (td, J = 12.0, 1.5 Hz, 2H), 2.2–1.9 (m, 6H), 11.04 (t, J = 7.5 Hz, 3H);MS: m/z 407.453 [M + H]⁺. HPLC = 97% at 0.99 min.

tert-Butyl (3-((1-Phenethylpiperidin-4-yl)amino)phenyl)-carbamate (6a).

SnCl₂ (72 mg, 0.38 mmol) was added to a room-temperature solution of N-boc-m-phenylenediamine (5a) (400 mg, 1.92 mmol), 1-phenethyl-4-piperidone (390 mg, 1.92 mmol), and poly(methylhydrosiloxane) (7.30 g, 3.84 mmol) in MeOH (20 mL). The reaction was then heated to 70 °C and stirred for 16 h. The reaction mixture was concentrated to dryness, and the residue was subjected to chromatography on silica gel using 0–100% MeOH in CH₂Cl₂ to furnish amine 6a (357 mg, 47%). ¹H NMR (300 MHz, CDCl₃) δ 7.35–7.15 (m, 4H), 7.05 (t, J = 8.0 Hz, 1H), 6.86 (bs, 1H), 6.48 (dd, J = 7.9, 1.4 Hz, 1H), 6.38 (bs, 1H), 6.28 (dd, J = 8.0, 1.7 Hz, 1H), 3.57–3.47 (m, 2H), 3.38–3.27 (m, 1H), 3.00–2.90 (m, 2H), 2.87–2.76 (m, 2H), 2.66–2.56 (m, 2H), 2.30–2.15 (m, 2H), 2.13–2.03 (m, 2H), 1.51 (s, 9H). MS (ESI) m/z 396.00 [M + H]⁺.

tert-Butyl (4-((1-Phenethylpiperidin-4-yl)amino)phenyl)-carbamate (6b).

An analogous procedure for the synthesis of 6a was followed starting with N-boc-p-phenylenediamine 5b to provide amine 6b (504 mg, 66%). ¹H NMR (300 MHz, CDCl₃) δ 7.32–7.28 (m, 1H), 7.23–7.11 (m, 5H), 6.56 (d, J = 8.8 Hz, 2H), 6.22 (bs, 1H), 3.56–3.48 (m, 2H), 3.32–3.22 (m, 1H), 2.96 (d, J = 11.8 Hz, 2H), 2.87–2.77 (m, 2H), 2.65–2.57 (m, 2H), 2.20 (t, J = 11.3 Hz, 2H), 2.13–2.02 (m, 2H), 1.50 (s, 9H). MS (ESI) m/z 396.00 [M + H]⁺.

tert-Butyl (3-(N-(1-Phenethylpiperidin-4-yl)propionamido)-phenyl)carbamate (7a).

Propionic anhydride (excess) was added at room temperature to a solution of amine 6a (249 mg, 0.48 mmol) in dry CH₂Cl₂ (10 mL). The reaction, which resulted, was allowed to stir for 24 h at room temperature. At that time, the reaction was quenched with saturated aq NaHCO₃ (5 mL). The layers were separated, and the aqueous layer was extracted with additional CH₂Cl₂ (2 × 20 mL). The combined organic layers were washed with brine (3 × 20 mL), dried (Na₂SO₄), and concentrated under reduced pressure. The crude amide 7a was used in the next step without further purification. ¹H NMR (300 MHz, CDCl₃) δ 7.31–7.13 (m, 7H), 6.76–6.71 (m, 1H), 6.69 (bs, 1H), 4.76–4.57 (m, 1H), 3.12–2.96 (m, 2H), 2.81–2.69 (m, 2H), 2.64–2.53 (m, 2H), 2.41–2.14 (m, 3H), 2.04–1.94 (m, 2H), 1.90–1.71 (m, 2H), 1.52 (s, 9H), 1.18–1.10 (m, 1H), 1.02 (t, J = 7.4 Hz, 3H). MS (ESI) m/z 452.20 [M + H]⁺.

tert-Butyl (4-(N-(1-Phenethylpiperidin-4-yl)propionamido)-phenyl)carbamate (7b).

An analogous procedure to 7a was used for the synthesis of 7b starting with amine 6b. ¹H NMR (300 MHz, CDCl₃) δ 7.39 (d, J = 8.6 Hz, 2H), 7.29–7.13 (m, 4H), 6.99 (d, J = 8.7 Hz, 2H), 6.61 (bs, 1H), 4.79–4.58 (m, 1H), 3.55–3.47 (m, 2H), 3.18–3.07 (m, 2H), 2.84–2.73 (m, 2H), 2.67–2.57 (m, 2H), 2.30–2.19 (m, 2H), 1.94 (q, J = 7.4 Hz, 2H), 1.84–1.73 (m, 2H), 1.53 (s, 9H), 1.00 (t, J = 7.4 Hz, 3H). MS (ESI) m/z 452.20 [M + H]⁺.

N-(3-Aminophenyl)-N-(1-phenethylpiperidin-4-yl)propionamide (8a).

Trifluoroacetic acid (5 mL) was added at room temperature to a solution of 7a (crude, 0.48 mmol) in CH₂Cl₂ (10 mL, containing 0.1 mL H₂O). The reaction was allowed to stir at room temperature for 18 h. The solvent was removed under reduced pressure, and the residue was redissolved in EtOAc and washed with 1 N NaOH (2 × 10 mL). The residue was subjected to medium-pressure column chromatography on silica gel using 0–100% CMA-80 (CHCl₃/CH₃OH/aq NH₄OH, 80:18:2) in CH₂Cl₂ to furnish deprotected amide 8a (10 mg). ¹H NMR (300 MHz, CDCl₃) δ 7.31–7.12 (m, 6H), 6.65 (dd, J = 8.0, 1.6 Hz, 1H), 6.45 (d, J = 7.7 Hz, 1H), 6.40–6.35 (m, 1H), 4.70–4.55 (m, 1H), 3.72 (s, 2H), 3.00 (d, J = 11.1 Hz, 2H), 2.79–2.68 (m, 2H), 2.61–2.48 (m, 2H), 2.19–2.08 (m, 2H), 2.05–1.97 (m, 2H), 1.85–1.68 (m, 2H), 1.60–1.38 (m, 2H), 1.02 (t, J = 7.5 Hz, 3H). MS (ESI) m/z 352.00 [M + H]⁺.

N-(4-Aminophenyl)-N-(1-phenethylpiperidin-4-yl)propionamide (8b).

An analogous procedure to that of 8a was used for the synthesis of 8b starting with amide 7b (84 mg, 38%). ¹H NMR (300 MHz, CDCl₃) δ ¹H NMR (300 MHz, CDCl₃) δ 7.32–7.09 (m, 5H), 6.88–6.79 (m, 2H), 6.71–6.58 (m, 2H), 4.72–4.56 (m, 1H), 3.75 (s, 2H), 2.99 (d, J = 11.6 Hz, 2H), 2.78–2.69 (m, 2H), 2.59–2.47 (m, 2H), 2.21–2.08 (m, 2H), 1.97 (q, J = 7.5 Hz, 2H), 1.76 (d, J = 11.8 Hz, 2H), 1.53–1.37 (m, 2H), 1.01 (t, J = 7.5 Hz, 3H). MS (ESI) m/z 352.00 [M + H]⁺.

Methyl 2-(3-(3-(N-(1-Phenethylpiperidin-4-yl)propionamido)-phenyl)ureido)acetate (9a).

Triethylamine (excess) was added at 0 °C to a solution of triphosgene (20 mg, 0.67 mmol) in CH₃CN (10 mL). Aniline 8a (10 mg, 0.03 mmol, free base) in CH₃CN (5 mL) was added via a syringe over 10 min. The reaction mixture was warmed to room temperature, and methyl glycinate hydrochloride (25 mg, 0.28 mmol) in THF (10 mL) was added and the reaction was allowed to stir at room temperature for 16 h. The reaction was quenched with MeOH (5 mL) and concentrated under reduced pressure. The residue was subjected to chromatography on silica gel using 0–100% CMA-80 (CHCl₃/CH₃OH/aq NH₄OH, 80:18:2) in CH₂Cl₂ to furnish 9a (9.7 mg, 73%). ¹H NMR (300 MHz, DMSO) δ 8.30 (s, 1H), 7.55–7.29 (m, 2H), 7.25–6.99 (m, 8H), 4.47–4.30 (m, 1H), 4.06–3.94 (m, 2H), 3.40–3.17 (m, 3H), 3.09 (s, 1H), 2.96–2.78 (m, 3H), 2.67–2.49 (m, 2H), 2.11–1.87 (m, 3H), 1.85–1.71 (m, 2H), 1.69–1.50 (m, 2H), 1.08–1.00 (m, 1H), 0.82 (t, J = 7.3 Hz, 3H). MS (ESI) m/z 467.00 [M + H]⁺.

Methyl 2-(3-(4-(N-(1-Phenethylpiperidin-4-yl)propionamido)-phenyl)ureido)acetate (9b).

An analogous procedure for the synthesis of 9a was followed starting with aniline 8b (37 mg, 79%). ¹H NMR (300 MHz, DMSO) δ 9.30 (s, 1H), 7.48 (d, J = 8.7 Hz, 2H), 7.36–7.16 (m, 5H), 7.05 (d, J = 8.6 Hz, 2H), 6.69 (t, J = 5.9 Hz, 1H), 4.61–4.43 (m, 1H), 3.89 (d, J = 5.8 Hz, 2H), 3.65 (s, 3H), 3.21–3.11 (m, 2H), 2.75 (s, 4H), 2.56–2.36 (m, 2H), 1.91–1.70 (m, 4H), 1.44–1.29 (m, 2H), 0.87 (t, J = 7.4 Hz, 3H). MS (ESI) m/z 467.00 [M + H]⁺.

Methyl 4-Oxo-4-((4-(N-(1-phenethylpiperidin-4-yl)-propionamido)phenyl)amino)butanoate (10).

Succinic acid monomethylester chloride (15.4 μL, 0.125 mmol) was added to an ice-cold solution of compound 8b (35 mg, 0.10 mmol) in dry CH₂Cl₂ (10 mL) and triethylamine (70.0 μL, 0.50 mmol). The reaction mixture was allowed to warm to room temperature and stirred for 1 h. The reaction was quenched with saturated aq NaHCO₃ (5 mL). The layers were separated, and the aqueous layer was extracted with additional CH₂Cl₂ (2 × 20 mL). The combined organic layers were washed with brine (3 × 20 mL), dried (Na₂SO₄), and concentrated under reduced pressure. The residue was subjected to chromatography on silica gel using 0–100% CMA-80 in CH₂Cl₂ to furnish 10 (32.0 mg, 69%) as a white solid. ¹H NMR (300 MHz, CDCl₃) δ 9.39 (s, 1H), 7.61 (d, J = 8.6 Hz, 2H), 7.30–7.22 (m, 2H), 7.21–7.12 (m, 2H), 7.05 (d, J = 8.6 Hz, 2H), 6.73 (dd, J = 54.0, 8.6 Hz, 1H), 4.78–4.54 (m, 1H), 3.83 (s, 3H), 3.52 (s, 2H), 3.02 (d, J = 11.3 Hz, 2H), 2.80–2.69 (m, 2H), 2.62–2.47 (m, 2H), 2.18 (t, J = 11.4 Hz, 2H), 2.00–1.87 (m, 2H), 1.80 (d, J = 10.6 Hz, 2H), 1.54–1.36 (m, 2H), 1.25 (s, 1H), 1.01 (t, J = 7.4 Hz, 3H), 0.91–0.82 (m, 1H).

Conjugation of the F₁, F₄, F₅, and F₆ Haptens to Carrier Proteins via Carbodiimide (EDAC) Chemistry.

The conjugations were performed according to the protocol previously described for either OXY(Gly)₄OH or M(Gly)₄OH haptens with minor modifications.³¹ Briefly, the F₁, F₄, F₅, and F₆ haptens were dissolved at a concentration of 5.2 mM in 0.1 M MES buffer pH 5 containing 10% dimethyl sulfoxide (DMSO) and were activated by carbodiimide coupling chemistry using N-ethyl-N′-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDAC, Sigma-Aldrich, St. Louis, MO) cross-linker at a final concentration of 208 mM. The mixture was left reacting for 10 min at room temperature (RT). BSA, sKLH, CRM₁, or CRM₂ were added at a final concentration of 2.8 mg/mL, and the reactions were stirred for the following 3 h at RT. The final conjugates were ultrafiltered using Amicon filters with 30 or 100 kDa molecular cutoff depending on the carrier protein dimensions: after having replaced MES buffer with phosphate-buffered saline (PBS) 0.1 M pH 7.2, the resulting solutions were stored at +4 °C. CRM₁ and CRM₂ conjugates required the presence of 250 mM sucrose in both reacting and storage buffer as a stabilizing agent.

Conjugation of the NHS-Esters of F₂ and F₃ Haptens to Carrier Proteins.

F₂ or F₃ (Figure 1) were dissolved in 1 mL of DMSO (final concentration 5 mg/mL), which was added at a rate of 20 μL per min to 5 mL of CRM or sKLH (1 mg/mL) under magnetic stirring in 1× PBS, pH = 7.4 (Corning, NY). After 1.5 h, samples were diluted with 5 mL of buffer and purified using a dialysis membrane (50k MWCO) in 1× PBS for 20 h, 2 exchanges at 4 °C. The sample concentration was measured using a Direct Detect spectrophotometer.

Characterization of the F_1–6 Conjugates via MALDI-TOF and Dynamic Light Scattering (DLS).

The molecular weight (MW) of BSA and CRM-containing conjugates was assessed before and after conjugation using an Applied Biosystems/MDS SCIEX 5800 MALDI TOF/TOF analyzer (Foster City, CA) and TOF/TOF 5800 System Software (SCIEX, Concord, Ontario, Canada). Each sample was assigned a spot and mixed with a saturated sinapinic acid matrix on a 384 opti-tof 123 × 81 mm Rev A plate. Data were acquired in the linear high mass positive acquisition mode. The resulting haptenization ratio (number of haptens per mole of protein) of each conjugate was calculated as follows: [(MW_conjugate − MW_{carrier protein})/MW_hapten]. Due to their large molecular weight, the haptenization ratio of sKLH conjugates could not be determined by MALDI-TOF and was consequently characterized for size by DLS on a Zetasizer (Malvern Panalytical Inc., United Kingdom) as described previously.⁴⁷

Drugs and Reagents.

Fentanyl, sufentanil, alfentanil, remifentanil, and other controlled substances were obtained from the University of Minnesota Pharmacy. Drug doses and concentrations are expressed as weight of the free base. Subunit Keyhole Limpet Hemocyanin (sKLH) was purchased as GMP-grade source (Biosyn, Carlsbad, CA), and cross-reactive material (CRM) from diphtheria toxin was purchased as either E. coli-expressed CRM (EcoCRM) (Fina Biosolutions, Rockville, MD) or CRM₁₉₇ (PFEnex, San Diego, CA). CRM from these different sources is described as CRM₁ and CRM₂, respectively.

Activity of fentanyl-based haptens at the Mu Opioid Receptor (MOR).

To assess whether the F₁ hapten had any functional activity at the MOR, the F₁ hapten was tested in vitro in a calcium mobilization assay involving Chinese Hamster Ovary (CHO) cells coexpressing the human MOR and Gα16, a promiscuous G-protein, as described before.⁴⁵ In these studies, the F₄ hapten and morphine were used as controls.

Animals.

All studies were approved by the University of Minnesota and the Hennepin Healthcare Research Institute Animal Care and Use Committee. Both institutions are AALAC-certified. Adult male BALB/c mice (Jackson Laboratories) and adult male and female Sprague Dawley rats (Envigo) were housed in a standard 12/12 h light/dark cycle and fed ad libitum. Mice were 6 weeks old on arrival. Rats were 2 months old on arrival for drug challenge studies, and 9–10 weeks old for drug self-administration studies. Most animals were immunized immediately after 1 week of habituation. Rats in the self-administration experiment were immunized after self-administration behavior was trained and stable (see below).

Immunization.

Mice were immunized i.m. with 60 μg of conjugates containing the F_1–6 hapten series or unconjugated sKLH, CRM₁, or CRM₂ as control. Conjugates were adsorbed on 300 μg of alum (Alhydrogel85, Brenntag) and PBS to a final volume of 60 μL and delivered 30 μL per hind leg. Mice were immunized on days 0, 14, and 28. Rats were immunized i.m. with 60 μg of conjugates containing the F_1–6 hapten series or unconjugated carrier as control. Conjugates were adsorbed on 90 μg of alum and PBS to a final volume of 150 μL and delivered to one hind leg. Rats were immunized on days 0, 21, and 42, or days 0, 21, 42, 63, and 84 as detailed in individual experiments.

Antibody Analysis.

Serum antibody analysis was performed via indirect ELISA after blood collection using facial vein sampling in mice or tail vein sampling in rats. Ninety-six-well plates were coated with 5 ng/well of the corresponding F_1–6–BSA conjugate or unconjugated BSA as a control diluted in 50 mM Na₂CO₃, pH 9.6 (Sigma-Aldrich, St. Louis, MO) and blocked with 1% porcine gelatin. Plates were incubated with serum samples and then washed and incubated with a HRP-conjugated goat antimouse IgG or goat antirat IgG (Jackson ImmunoResearch Laboratories) to assess hapten-specific serum IgG antibody levels as previously described.⁶⁸ For determination of affinity by competitive binding ELISA, 96-well plates were coated and blocked as described above, and fentanyl or other compounds were added to the wells with concentrations ranging from 1 × 10⁻⁴ to 1 × 10⁻¹⁰ M. Individual immune sera were diluted to subsaturating concentrations, incubated with a competitor on the plate, and washed and incubated with HRP-conjugated antibody as above.

Vaccine Efficacy against Antinociception Induced by Target and Off-Target Opioids in Mice and Rats.

The effect of candidate vaccines against antinociception induced by either target or off-target opioids was evaluated in the hot plate test of centrally mediated analgesia. Mice and rats were allowed to acclimate to the testing environment for 1 h prior to measuring baseline. Rodents were placed on a hot plate (Columbus Instruments, Columbus, OH) set to 54 °C and removed after displaying a lift or flick of the hindpaw or reaching the maximal cutoff of 60 s in mice or 30 s in rats to avoid thermal tissue damage. In mice, testing was initiated after the third immunization (day 35). After measurement of baseline latencies, mice were challenged with fentanyl (0.05–0.1 mg/kg, s.c.) and their hot plate responses were recorded at 30 min post drug challenge. In rats, testing was initiated after the third immunization (day 49). To determine the efficacy of candidate vaccines against selected target opioids, rats were tested repeatedly once a week with drug challenges including fentanyl (0.075–1.0 mg/kg, s.c.), sufentanil (0.008 mg/kg, s.c.), and alfentanil (0.5 mg/kg, s.c.) as detailed in each experiment, and the latency to respond was measured at 15, 30, 45, and 60 min post drug administration. At completion of the behavioral assessment (mice = 31 min and rats = 61 min, which includes the 60 s maximal cutoff threshold), trunk blood and brain were collected for assessment of fentanyl concentrations. To test whether vaccination interfered with selected opioid agonists, and the reversal of their effects by naloxone, rats immunized with either CRM₁ or F₁–CRM₁ were challenged weekly with oxycodone (2.25 mg/kg, s.c.), heroin (0.9 mg/kg, s.c.), methadone (2.25 mg/kg, s.c.), and fentanyl (0.1 mg/kg, s.c., positive control) and hot plate responses were recorded at 30 min post drug challenge. After receiving oxycodone or heroin, rats were given naloxone (0.1 mg/kg, s.c.) and their responses were recorded for an additional 15 min. In both mouse and rat studies, data are displayed as the maximal possible effect (MPE%) calculated as (postdrug latency − baseline latency)/(maximal cutoff − baseline latency) × 100.

Vaccine Efficacy against Respiratory Depression and Bradycardia Induced by Target Opioids and Off-Target Anesthetics in Rats.

To assess the effect of candidate vaccines against respiratory depression and bradycardia induced by either target opioids or off-target anesthetics, pulse oximetry was used to measure oxygen saturation (SaO_2%), breath rate (breaths per minute, brpm), and heart rate (beat per minute, bmp) before and after drug challenges. Oximetry was measured using a MouseOx Plus pulse oximeter (Starr Life Sciences, Oakmont, PA). Rats were allowed to acclimate to the testing environment for 1 h prior to measuring baseline. After baseline recordings, rats were challenged s.c. repeatedly once a week with fentanyl (0.075–1.0 mg/kg), sufentanil (0.008 mg/kg), and alfentanil (0.5 mg/kg) as detailed in each experiment. In studies involving anesthetics, rats immunized with either CRM₁ or F₁–CRM₁ were first challenged with dexmedetomidine (0.25 mg/kg, i.p.), whose effects were reversed by atipamezole (1 mg/kg, i.p.). In subsequent weeks, rats were challenged with ketamine (75 mg/kg, i.p.), propofol (100 mg/kg, i.p.), or isoflurane (2%, Drager Vapor 2000, Telford, PA). Measurements were obtained at 15, 30, 45, and 60 min post drug administration.

Analysis of Fentanyl Concentration.

Brain tissue was homogenized using Agilent ceramic beads with a Beadblaster 24 homogenizer (Benchmark Scientific, Sayreville, NJ) and placed at −20 °C until extraction. Brain homogenate, serum, and standards were processed in acetonitrile at 4 °C, and then the supernatant was transferred, evaporated, and diluted in phosphate buffer. Samples were extracted using Bond Elut Plexa PCX extraction cartridges (Agilent, Santa Clara, CA), evaporated, and reconstituted in a solution of water, ammonium formate, and formic acid. Samples were injected onto a reversed-phase Agilent (Santa Clara, CA) Zorbax Eclipse plus C18 column (2.1 mm × 50 mm i.d., 1.8 μm) and then analyzed on an Agilent G6470A triple quadrupole LCMS/MS system consisting of an Infinity II 1290 G7116B Multicolumn Thermostat, G7120A High Speed Quad Pumps, and a G7267B Multisampler. Data acquisition and peak integration were analyzed using Mass Hunter software (Tokyo, Japan). Detailed methods are provided in the Supporting Materials.

Efficacy of Vaccine against Fentanyl Intravenous Self-Administration.

To provide proof of efficacy for therapeutic vaccination to reduce ongoing fentanyl intake, rats were trained in a fentanyl self-administration (FSA) assay using standard two-lever operant conditioning chambers (Med Associates, St. Albans, VT) prior to initiation of the immunization regimen. Rats were implanted with jugular catheters and then trained to FSA (1 μg/kg/infusion) under a fixed-ratio (FR) 1 schedule during the daily 120 min sessions (5 days/week) according to a previously described standard protocol.^36,52 This unit dose was chosen because it maintains robust self-administration in rats and lies near the peak of the FSA dose–response curve,^74–78 providing a sensitive initial screen for vaccine efficacy. Responses on one lever produced fentanyl infusions at a rate of 0.1 mL/kg/s, while responses on the other lever had no programmed consequence. After at least 10 sessions and when robust fentanyl intake was established, the FR was gradually increased to three over several sessions. After at least 10 sessions at FR 3 and once FSA stabilized (at least 20 infusions per session, greater than 2:1 ratio of active:inactive lever presses, and no significant trend across three consecutive sessions), rats were immunized i.m. with either CRM₁ (n = 8) or F₁–CRM₁ (n = 7) on days 0, 21, 42, and 63 while their FSA sessions continued. This course of injections allowed assessment of the effects of vaccination on the maintenance of FSA at the training dose. Then, rats received a fifth vaccination on day 84 (3 weeks after the fourth vaccination) to allow testing the effects of vaccination on the FSA dose–response curve. For this phase, the fentanyl dose per infusion was reduced each week on Mondays using the following doses: 0.75, 0.50. 0.25, and 0 μg/kg/infusion. Three rats did not complete this phase because of catheter failure or death, resulting in a final sample size of seven CRM₁ rats and six F₁–CRM₁ rats.

Computational Methods.

GLIDE docking grids were prepared based on the initial N-terminal truncated crystal structure coordinates with PropPKA protonation assignments. Two full-length structures were prepared using MODELLER⁷⁹ simulated annealing with topological constraints, nudged elastic band pulls of the complete sequence N- and C-terminal ends employing AMBER 18 parameterization⁸⁰ to generate more compact termini conformations followed by equilibration in a DOPC/K+/Cl−/water environment. A LIPID 14 membrane-component AMBER parameterization was used.⁸¹ Protein preparation, including protonation assignment, was performed on these initial structures with Small Molecule Discovery Suite software (Schrödinger, INC, New York, NY) prior to preparation of Glide⁸² docking grids centered on residues near the extracellular region and including D147, Y148, and H54 -proximate to the orthosteric binding site. We particularly included H54 in the docking region because we were intrigued by the N-terminal loop insertion into the orthosteric binding site of the 5C1M structure in a manner “analogous” to that observed in the initial CB1 antagonist crystal structures. The role of such insertions has not been explored, to the best of our knowledge, in stabilization of the bound ligand binding conformation, and we wanted to at least probe this facet in a preliminary fashion. For this reason, we docked BU72(4V0 in the 5C1M structure file) into the truncated original coordinates and two membrane equilibrated full-length MOR structures, one with an elastic band pull of the N-terminus removing that N-terminal loop insertion and a second retaining the N-terminal loop insertion.

Statistical Analysis.

Fentanyl brain and serum concentrations, latency to respond in the hot plate nociception test, oxygen saturation, breath and heart rates, and serum antibody titers at single time points were analyzed using either an unpaired t-test with Welch’s correction or a one-way analysis of variance (ANOVA) paired with Dunnett’s multiple comparison test. Comparisons over multiple time points were analyzed by two-way ANOVA, using the Geisser-Greenhouse correction, paired with Tukey’s or Sidak’s multiple comparison test as appropriate. Reporting of statistical differences was focused on comparisons to either control or the first-generation lead F₁-sKLH vaccine formulation. For the self-administration study, the mean number of infusions during the last three sessions at baseline and 3 weeks after the fourth vaccine injection was used to assess vaccine effects on the maintenance of FSA. To assess vaccine effects on the FSA dose–response curve, the mean number of infusions during the last two sessions at each unit dose was used. FSA data were log-transformed prior to statistical analysis due to non-normality of the data and heterogeneity of variance. The transformed data were analyzed using mixed-model ANOVA followed by Bonferroni’s multiple comparison tests. All statistics were performed using Prism (version 8.3.0; GraphPad, San Diego, CA).