On-demand opioid effect reversal with an injectable light-triggered polymer-naloxone conjugate

Abstract

Misuse of opioids can lead to potential lethal overdose. Timely administration of naloxone is critical for survival. Here, we designed a polymer-naloxone conjugate that can provide on-demand photo-triggered opioid reversal. Naloxone was attached to the polymer poly (lactic-co-glycolic acid) via a photocleavable coumarin linkage, and formulated as injectable nanoparticles. In the absence of irradiation, the formulation did not release naloxone. Upon irradiation with a blue (400 nm) light, the nanoparticles released free naloxone, reversing the effect of morphine in mice. Such triggered events could be performed days and weeks after the initial administration of the nanoparticles, and could be performed repeatedly.

Graphical Abstract

There are more than 4000 deaths from opioid overdose each year in the US¹. Timely treatment with an opioid antagonist such as naloxone (NLX) can be life-saving in opioid overdose.² Even though the importance of naloxone is recognized, factors limit access:³ naloxone not being stocked by pharmacies,^4–5 restrictions to access by lay people,⁶ affordability⁷ and payment method, the stigma surrounding opioid use disorder,⁸ and other socioeconomic factors. Even when acquired, naloxone may be not be on the person needing it at the time of overdose.

Naloxone could be delivered by a sustained release system placed into the patient during contact with the healthcare system. However, prolonged constant release of naloxone could interfere with normal usage of opioids, such as in acute perioperative use, or in treatment of opioid use disorder. A system that was injected into the patient at the time of contact with the health care system and only released naloxone when needed would be a major advance.

Stimuli-responsive materials have been used to provide temporal control of drug delivery, in contrast to the extended release from more conventional drug delivery systems.^9–12 Stimuli such as light,^13–14 temperature^15–16 and ultrasound,^17–18 have been employed in various applications. Light has been commonly used because of its ease of spatiotemporal control and tunability of wavelength and intensity.¹⁹ As one example of an analogous application, we have used photo-triggering to provide on-demand local anesthesia, where the timing, intensity, and duration of local anesthesia could be adjusted by modulating the intensity and/or duration of irradiation of an injected depot.^{13, 20}

Here, we designed a light-responsive polymer-naloxone conjugate, and formulated it into injectable nanoparticles. Blue (400 nm) light was used since the drug depot would be injected subcutaneously (deep penetration by light would not be necessary) and because light of shorter wavelengths has higher energy to cleave chemical bonds. Coumarin (CM) was used as the photo-cleavable linkage due to its good photocleavage efficiency with blue light.^{13, 19} The subcutaneously injected polymer-drug conjugate would not release naloxone unless the site were irradiated with light of a specific wavelength, at an irradiance far greater than that of ambient light.²¹ Poly(lactic-co-glycolic acid) (PLGA) was selected as the polymeric carrier for naloxone. PLGA is an FDA-approved polymer that has been widely used in drug delivery.^22–23 Since the drugs were attached at the polymer chain ends, a low molecular weight (2000 Da.) was selected to achieve a high drug loading (i.e., the ratio of total drug mass to polymer mass would be high).

Synthesis and characterization.

A polymer-naloxone conjugate (PLGA-CM-NLX) was synthesized (Fig. 1a). The synthesis is detailed in Supplementary Fig. 1. In brief, we synthesized a coumarin-naloxone conjugate with a dibenzocyclooctyne (DIBO) moiety, which was reacted with azide-terminated PLGA (molecular weight 2 kDa, lactide:glycolide 50:50) through strain-promoted azide–alkyne cycloaddition (SPAAC). Disappearance of the azide group at 2100 cm⁻¹ in Fourier-transform infrared (FT-IR) spectra confirmed the coupling (Supplementary Fig. 2).^24–25 Representative peaks of polymer and drug were observed by nuclear magnetic resonance (NMR) spectroscopy (Supplementary Fig. 3), e.g., 5.2 ppm and 4.8 ppm from the lactic and glycolic moieties of PLGA, respectively, and 5.8 ppm from the vinyl moiety of naloxone. The retention time by gel permeation chromatography (GPC) of PLGA-CM-NLX was also shortened compared to PLGA-azide, indicating increased molecular size and weight (Supplementary Fig. 4). We used SPAAC rather than copper-catalyzed azide-alkyne cycloaddition because it is highly efficient for macromolecules, and reaction conditions are mild and do not require catalysts, without metal residue. ^26–28

Fig. 1. Synthesis, characterization and In vitro photocleavage.

a. Synthesis of PLGA-CM-NLX. b. DLS of PLGA-CM-NLX nanoparticles. c. TEM image of PLGA-CM-NLX nanoparticles. d. UV-vis of PLGA-CM-CM-NLX in acetone and nanoparticles in water. e. Scheme of photocleavage releasing naloxone from PLGA-CM-NLX. f. Mass spectrum of the cleaved naloxone. g. Release of naloxone by photocleavage of NLX-NP (10 mg mL⁻¹) over time with a 400 nm LED at different irradiances (n = 4, data are means ± SD). h. Repeated photo-triggered naloxone release from NLX-NP (arrows represent two minutes of irradiation with a 400 nm LED at 30 mW cm⁻², n = 4, data are means ± SD).

PLGA-CM-NLX was dissolved in acetone and 103 ± 29 nm (by dynamic light scattering (DLS; Fig. 1b) nanoparticles (NLX-NP) prepared by nanoprecipitation into water. Spherical nanoparticles were observed by transmission electron microscopy (TEM; Fig. 1c). PLGA-CM-NLX in acetone showed a representative UV-vis absorption peak at 378 nm from the coumarin moiety. The peak shifted to 392 nm when nanoparticles formed (Fig. 1d). This red-shift can be attributed to aggregation of molecules in the solid form.²⁹

NLX-NP irradiated with 400 nm light cleaved to a naloxone derivative, which released unmodified naloxone by self-immolative cyclization^30–31 (Fig. 1e). Release of unmodified naloxone (m/z 328.2; Fig. 1f) was confirmed by liquid chromatography-mass spectroscopy (LC-MS). Naloxone release from NLX-NP increased with duration or intensity of irradiation with 400 nm LED (Fig. 1g). Naloxone release from NLX-NP could be repeatedly triggered by irradiation (Fig. 1h).

In vivo efficacy.

The effectiveness of light-triggered naloxone was assessed in mice using a hotplate test, widely used for testing opioid analgesia.^32–34 Dosing was constant throughout: 10 mg kg⁻¹ for morphine, 0.1 mg kg⁻¹ for naloxone, 100 mg kg⁻¹ for NLX-NP, and 300 mW cm⁻² for 2 min for irradiation events with a 400 nm LED.

At predetermined intervals (every 30 min) animals were placed on a 50 ^oC hotplate, and the latency (time in seconds) to a response (flinching, licking their hind paws or jumping) was recorded every 30 min except where indicated otherwise. Latency of animals without treatment with morphine (baseline) was 27.8 ± 4.7 s. This baseline was used to calculate Δ latency: the difference between the latency in treated animals and baseline. Latency of animals treated with morphine (subcutaneously between the scapulae) rose to 60 s by 30 min after injection, with a Δ latency of 32.2 s (Fig. 2a). (Since animals were removed from the hotplate at 60 s to avoid thermal injury, 60 s was the maximum latency. Therefore, the maximum Δ latency from the test was 32.2 s.) The Δ latency remained 32.2 s at 60 min, and gradually returned to baseline at a difference in 174.5 ± 50.0 min.

Fig. 2. In vivo reversal of opioid effect by NLX-NP.

Hotplate Δ latency of mice injected subcutaneously with morphine a. without (M, black line) or with (M+NLX, red line) naloxone injection at 30 min b. with injected NLX-NP at t = 0 min, without ((M+NLX-NP hv (−)), green line) or with ((M+NLX-NP hv (+)), blue line) subsequent irradiation at 30 min c. Δ Latency at 60 min, from data in panels a and b. d. Illustration of experimental timeline for delayed triggering; see text for details. e. Hotplate Δ latency of mice treated with morphine at the indicated time after injection of NLX-NP, followed by irradiation. f. Δ Latency at 60 min after morphine and 30 min after irradiation (from data in panels b and e) as a function of days after injection of NLX-NP. p values in yellow compare the Δ latencies of irradiated animals to animals 60 min after morphine; p values in blue compare the Δ latencies of irradiated animals to those 30 min after free naloxone. g. In Vivo Imaging System (IVIS) images of mice injected with NLX-NP containing PLGA covalently labeled with the fluorescent dye cyanine5.5 (PLGA-cy5.5). h. Quantitation of relative fluorescent intensity over time from IVIS data in panel g; see Supplementary Fig.9 for images for the free dye. All irradiations (hv) in this figure were 2 min with 400 nm LED at 300 mW cm^-2. The black dashed lines in panels a, b, e, are the curves for morphine alone. All data are means ± SD, n = 4.

When animals treated with morphine (Δ latency = 32.2 s) were administered free naloxone subcutaneously 30 min after the morphine (but at a different site in the lower back), the Δ latency decreased to 1.1 ± 3.3 s at the next time point (60 min; p = 0.000001, Fig. 2a), and remained at baseline afterwards, indicating that free naloxone reversed the effect of morphine for the entirety of the opioid’s duration of effect.

We tested the ability of NLX-NP to provide triggered opioid reversal. NLX-NP (100 mg kg⁻¹_, containing 300 μg NLX) were injected subcutaneously in the lower back of mice, followed immediately by injection of morphine subcutaneously between the scapulae. In the absence of irradiation of the site of NLX-NP injection, Δ latency remained at 30.7 ± 3.0 s (Fig. 2b) 30 minutes after morphine injection, similar to the latencies in mice injected with morphine only, indicating that naloxone linked to the polymer was not released. When the site of NLX-NP injection was irradiated with a 400 nm LED for two minutes at 300 mW cm⁻² 30 min after morphine, Δ latency decreased to 4.8 ± 5.7 s at 60 min (Fig. 2b; p = 0.00007), and similar to that in morphine-treated animals treated with free naloxone at the same time point (p = 0.3). These data indicated that irradiation released naloxone and reversed the effect of the opioid. We measured that approximately 6.7 ± 0.7% of light from the LED could penetrate skin to the subcutaneous site of injection.

To determine effect onset time, animals were injected with morphine, then 30 min later injected with free naloxone or NLX-NP and then triggered with 400 nm light and tested every 5 min (Supplementary Fig. 7). The onset time of reversal by naloxone triggered from NLX-NP was slightly slower than from free naloxone. With free naloxone, it took 5 min for Δ latency to become different from that of animals treated with morphine alone (p = 0.03); with triggered release of naloxone, this took 10 min (p = 0.04). With free naloxone, latency was similar to baseline within 10 min (p = 0.08); with triggered release of naloxone, this took 20 min (p = 0.06).

In clinical applications, the depot would be injected subcutaneously by healthcare practitioners, and only triggered in emergencies at a later time. We assessed the potential for late triggering by injecting NLX-NP on Day 0 without irradiation. At predetermined intervals after injection of NLX-NP (2 days, 1 week or 1 month, Fig. 2d), separate groups of animals were first injected with morphine, then 30 min later, after the Δ latency had risen to the maximum (32.2 s) they were irradiated with a 400 nm LED at 300 mW cm⁻² for 2 min at the site of NLX-NP injection. Latency was measured again 30 minutes later, to assess the effect of photo-triggered release of naloxone (Fig. 2e, f). 2 days after injection of NLX-NP, irradiation reduced Δ latency from morphine from 32.2 s to 5.8 ± 7.9 s (p = 0.0006), similar to the Δ latency with morphine treated with free naloxone (p = 0.3, from Fig. 2a) (Fig. 2e and f). Irradiation at 1 week reduced Δ latency from 32.2 s to 2.5 ± 5.1 s (p = 0.00002; p = 0.6 compared to the free naloxone group, from Fig. 2a), and at 1 month from 32.2 s to 23.1 ± 2.7 s, a lesser but statistically significant reduction in Δ latency (p = 0.0005; p = 0.00005 compared to the free naloxone group). NLX-NP did not show triggered opioid reversal if irradiation was performed two months after injection (Supplementary Figure 8). The ability to trigger long after injection was consistent with the demonstration with an in vivo imaging system (IVIS) that subcutaneously injected NLX-NP containing PLGA covalently labeled with the fluorescent dye cyanine5.5 (PLGA-cy5.5) remained in tissue for at least a month (Fig. 2g and h). This prolonged retention can be attributed to the nanoparticles, since most of the fluorescent signal of injected free y5.5 disappeared within one day (Supplementary Fig. 9).

Repeated dosing with naloxone could be important for eventual clinical application, to extend the useful lifetime of the device. Repeated dosing was studied by injecting animals with NLX-NP and testing the ability to trigger reversal of morphine on days 2, 5, 10, and 20 in the same animals (Fig. 3). The Δl atencies after triggering at 2, 5, and 10 days were 5.8 ± 7.9 s, 10.9 ± 6.7 s, and 15.5 ± 12.9 s, respectively, which were all statistically significantly reduced compared to Δ latencies before irradiation (32.2 s, p = 0.0005, 0.0007, 0.04, respectively). Triggering on day 20 (26.5 ± 11.4 s) did not reduce Δ latency (p = 0.36 compared to Δ latency before irradiation).

Fig. 3. Repeated triggering.

a. Illustration of experimental timeline. Hotplate Δ latency of mice injected with NLX-NP then, at predetermined intervals, administered morphine followed 30 min later by irradiation b. latencies on day 2; c. day 5; d. day 10; e. day 20 (hν: 2 min irradiation by 400 nm LED at 300 mW cm⁻²; the black dashed line is the curve for morphine; data are means ± SD, n = 4). f. Δ latency at 60 min after morphine injection (from panel b-e) as a function of time after injection. p values in yellow compared to Δ latency in the morphine only group; p values in blue are compared to morphine treated with free naloxone.

Repeated dosing with naloxone would be important in case of opioid overdose with a dose that had a prolonged effect. We mimicked that situation by repeated administration of morphine. After the initial dose of morphine, a second dose was given at 90 min, while the latency remained near-maximal. The combined morphine doses produced analgesia for five hours (the time until latency was no longer statistically significantly different from baseline), compared to two hours from a single dose of morphine (Fig. 4a). A dose of free naloxone injected 30 min after the first dose of morphine decreased Δ latency from 32.2 s to 1.7 ± 5.0 s (Fig. 4b and e, p = 0.00002). With a second dose of morphine at 90 min, Δ latency rose to 32.2 s, i.e., the effect of morphine was not impeded by the previously administered naloxone (Fig. 4b). Similarly, in animals injected with NLX-NP at the same time as the initial dose of morphine (t = 0 min), irradiation of the nanoparticles at 30 min reversed the effect of morphine but did not prevent the effect of the second dose of morphine (Fig. 4c). A second irradiation at 90 min prevented the effect of the second dose of morphine (Fig. 4d and f) with the Δ latency remaining at 7.0 ± 7.0 s (p = 0.0003 compared to 32.2s for animals that did not receive the second irradiation) at the next time point (i.e., 120 min), demonstrating that prolonged morphine effect could be reversed by repeated triggering.

Fig. 4. Triggered reversal of repeated doses of morphine.

a. latency of mice receiving two sequential doses of morphine without other injections (M+M, the grey lines in all panels) vs mice receiving a single dose of morphine (M, dashed line). b. Same experiment as panel a but with a single injection of naloxone at 30 min (M+M+NLX); c. Same experiment as panel a but with NLX-NP injected at time 0 and one irradiation at 30 min (M+M+NLX-NP hv (+), hν: 2 min irradiation by 400 nm LED at 300 mW cm⁻²); d. Same experiment as panel c but with irradiations at 30 and 90 minutes (M+M+NLX-NP hv (+ +). e, f. Δlatencies at 60 min (e) and 120 min (f) from data in panels a-d. n =4, data are means ± SD.

Cytotoxicity and biocompatibility.

NLX-NP showed minimal or no cytotoxicity in 3T3 fibroblasts or C2C12 muscle cells (Supplementary Fig. 10, p > 0.05 compared to the saline group). Naloxone also showed minimal cytotoxicity below 0.5 mg mL⁻¹ in both cell lines (p > 0.05 compared to the saline group), but cell survival was reduced to 76.0 ± 10.5 % at 1 mg mL⁻¹ for 3T3 cells (p = 0.007 compared to the saline group) and 80.0 ± 5.2 % cell survival for C2C12 cells (p = 0.0005 compared to the saline group).

To assess tissue reaction in vivo, animals were euthanized 4, 14, or 30 days following subcutaneous injection of NLX-NP. The tissues around the site of injection, including epidermis, dermis, dermal white adipose tissue (dWAT), panniculus carnosus (a subcutaneous muscle layer found in mice but not humans), and interstitial connective tissue, were processed into hematoxylin and eosin (H&E) stained slides (Fig. 5).

Fig. 5. Tissue reaction to subcutaneously injected PLGA-CM-NLX NP.

a, b. normal tissues. c-h. Tissue reaction to the materials 4, 14 and 30 days after the injection. dWAT: dermal white adipose tissue; PC: panniculus carnosus; ICT: interstitial connective tissue; Mtox: myotoxicity; Infl: inflammation; M: macrophages; N: neutrophils; L: lymphocytes.

Four days after injection and irradiation, mild to moderate inflammation with occasional neutrophils and more lymphocytes and macrophages occurred in tissues adjacent to the area of particle deposition, specifically the panniculus carnosus and interstitial connective tissue (Fig. 5c, d). The depth of injection was measured as 664 ± 79 μm. More superficial tissues, including the epidermis, dermis, and dWAT, did not demonstrate inflammation or cell injury. Inflammation was diminished by days 14 and 30 (Fig. 5e–h), consisting principally of lymphocytes and foamy macrophages (reflecting the uptake of injected particles). Tissues from animals injected with NLX-NP without irradiation showed similar inflammation (Supplementary Fig. S11), indicating that irradiation was not the source of inflammation.

On day 4, there was very mild myotoxicity of the panniculus carnosus, with regenerating myocytes adjacent to the area of particle deposition (i.e., interstitial connective tissue) as well as myocytes with nuclear internalization (Fig. 5d). The superficial layers of the panniculus carnosus (i.e., those further away from NLX-NP) had no myotoxicity. The panniculus is not found in humans. By days 14 (Fig. 5f) and 30 (Fig. 5h), myotoxicity was overall diminished with myocytes demonstrating regeneration and no necrosis was observed. Histology from animals injected with NLX-NP but not irradiated were similar to those that were (Supplementary Fig. S11).

We developed a polymer-naloxone depot to provide light-triggered reversal of opioid effects. For many externally triggered drug delivery systems (e.g., those based on hydrogels or particles such as liposomes), substantial untriggered drug release is not uncommon; it may result in untriggered therapeutic effects. Here, the NLX-NP was inert and did not interfere with the effect of morphine. The system only released naloxone when irradiated, and could reverse opioid effect. Minimal untriggered drug release from a depot is desirable in the design of an externally triggered naloxone-releasing system, to minimize loss of drug that could be used in triggered events, and to avoid interfering with the therapeutic usage of opioids.

We used morphine analgesia in mice as a model and a proxy for opioid overdose. Opioid overdose occurs by the same pharmacological mechanism (i.e., the same opioid receptors) as opioid-induced analgesia. Moreover, since, the dose used here to induce analgesia in mice (10 mg kg⁻¹) is approximately 100-fold greater than a substantial analgesic dose in humans (0.1 mg kg⁻¹), the ability to reverse murine opioid analgesia reflects the ability to reverse a very large opioid dose.

The therapeutic index of naloxone is high.² There would therefore be few adverse effects of over-triggering other than possibly inducing opioid withdrawal in addicted users, a necessary risk of treating overdoses. Accidentally triggering by ambient light is unlikely since the irradiance required for triggering is ~100-fold greater.^{21, 35} If necessary, in clinical practice the device might be protected from ambient light. The LED could be readily incorporated into a personal adornment, such as jewelry, or at MedicAlert bracelet, possibly overlying the site of injection.

The device would be administered by health care professionals and would subsequently be on the patient’s person, obviating issues with access to naloxone,^4–5 and removing the need to find or carry intranasal naloxone delivery systems.

The materials remained triggerable for up to a month after initial administration, due to the long retention of the PLGA. In clinical practice, the system would be administered by the healthcare system. The naloxone depot would only be activated when needed at a remote and unpredictable time. Overdose might require more than one dose of naloxone for successful rescue. Since the half-life of the opioid used may be longer than that of naloxone,³⁶ repeated triggering may be necessary. Overdose could also occur on more than one occasion. We have demonstrated that our system could provide rescue in those circumstances. In this proof-of-principle study, opioid reversal could not be triggered two months after injection. An eventual clinical system could be engineered to provide more doses of naloxone over a longer period.

US FDA-approved intranasal, intramuscular, and subcutaneous systems for naloxone delivery exist.³⁷ All three methods can achieve effective bioavailability within 15–20 min. The onset of triggered reversal of morphine with NLX-NP occurred within that time frame. Further optimization in dosage, formulation or irradiation could accelerate naloxone release.

Tissue reaction to NLX-NP was consistent with reaction to injected polymeric materials,^38–39 except that PLGA itself generally does not cause myotoxicity; the myotoxicity might due to the low molecular weight polymer or the small size of the particles,^40–41 or combination of PLGA with even low doses of therapeutics.⁴² NLX-NP would be used subcutaneously in humans, where there is no equivalent to the murine panniculus carnosus (where the myotoxicity was seen). Therefore, the myotoxicity is likely a non-issue from the clinical point of view.

In summary, we have developed a light-responsive polymer-naloxone conjugate that was processed into injectable nanoparticles and could release naloxone in its native form upon blue light irradiation. The material did not affect opioid effect in the absence of photo-triggering, but enabled opioid reversal by light irradiation. Triggering could be performed weeks after administration, over multiple days and multiple times. The material could provide on-demand opioid antidote administration to reverse overdoses.