Continuous Nicotine Monitors for Personal Nicotine Pharmacokinetics: A Receptor-Aware Research Agenda

Abstract

A minimally invasive “continuous nicotine monitor” (CNM) would resolve the dynamic nicotine concentration, [nicotine]_t, faced by high-sensitivity nicotinic acetylcholine receptors (nAChRs) during and after nicotine intake by individual subjects. Motivations: “Know the potential enemy at an individual level.” Smoking or vaping produces an initial “bolus” of nicotine in the blood and brain, lasting ~5 minutes with a peak concentration of ~100–200 nM. The bolus largely governs reinforcement, reward, and cognitive enhancement. A prolonged declining phase of [nicotine], with a half-time of 1–4 hours, largely suppresses withdrawal symptoms and governs the known cell biology of addiction. Next, “Know the potential therapy,” because individual [nicotine]_t records will be useful during research on the effectiveness of nicotine replacement therapy. Finally, “Know the physiology.” The only three known effects on nAChRs in cerebrospinal fluid at the relevant [nicotine]_t are activation, desensitization, and chaperoning/upregulation. Therefore, additional mechanistic insights will arise from correlating [nicotine]_t with readily measurable physiological data on those effects in molecular, cellular, and brain slice systems and animal models. Interstitial fluid is the appropriate compartment for a CNM. The molecular sensor technology could employ fluorescence, as shown by progress on measuring [nicotine]_t with improved variants of intensity-based nicotine-sensing fluorescent reporters (iNicSnFRs). Electrochemical measurements of [nicotine]_t may also be possible. Studies like the Population Assessment of Tobacco Health would contextualize [nicotine]_t measurements during each subject’s ad libitum nicotine intake, hopefully at a cost <$100 for a 24-hour record.

Implications.

We propose the first research agenda leading to a wearable continuous nicotine monitor (CNM). We explain that this interdisciplinary agenda must be publicly funded and not-for-profit. The CNM would measure the time-varying concentration, [nicotine]_t, experienced by nicotinic receptors. We define the challenging specifications for a CNM, and we show how present techniques can nearly meet these challenges. With the appropriate CNM, individual [nicotine]_t can be measured in populations of 10,000 or more subjects, addressing many present and future hypotheses about individual nicotine pharmacokinetics and receptor physiology for all modes of human nicotine intake, including emerging nicotine replacement therapies.

Introduction

Ever since Columbus’s crew sampled tobacco in 1492, nicotine research has led the way in new concepts and techniques for neuroscience. With that history in mind, we intend this article to inform two communities. The nicotine and tobacco research community may wish to learn about the prospects for new generations of wearable, minimally invasive or noninvasive devices to measure the pharmacokinetics of nicotine at an individual level.

The wearable device community may wish to learn about the specific opportunities and challenges associated with nicotine. The opportunities are great: at any given time, some 60% of the world’s >1 billion smokers would like to quit. The challenge: nicotine is present at concentrations ~ 1 million times smaller than glucose, the target for the most successful continuous, minimally invasive monitor, yet the required time resolution resembles that for glucose!

We wish to measure nicotine concentration [nicotine] versus time, t, near the relevant receptors, or [nicotine]_t. Since tobacco produces primarily the S-enantiomer and only the charged species binds to nicotinic acetylcholine receptors (nAChRs), one would more precisely write [(S)-nicotine⁺]_t.

Three Motivations for the Proposed Continuous Nicotine Monitor

1. “Know the Potential Enemy”: Personal Pharmacokinetics

It is already known that cigarettes and electronic nicotine delivery systems (ENDS) produce two phases of [nicotine]_t. The first phase occurs while nicotine is entering the blood. It is the relatively high, initial peak or “bolus” of [nicotine] (100–200 nM for 5–10 minutes, Figure 1, area highlighted mins 0–10). This phase appears to provide reinforcement via a sense of well-being, stress relief, or cognitive boost. The bolus results from the efficiency of smoking/vaping: (1) the large area (50–75 m²) of the lungs and (2) the high membrane permeability of the nicotine free base. During a puff, nicotine crosses the apical and basolateral membranes of alveolar endothelial cells, reaching arterial blood within a few seconds, then the brain a few seconds later.^3–6 When nicotine is no longer present in the alveoli, [nicotine]_t begins to decline.

Figure 1.

Collected data on time course of plasma [nicotine]_t, using the intravenous (IV) blood-draw method. Measurements on cigarette smoking cover a wide range, mostly falling between the two examples given (Cigarette1¹ and Cigarette2²). Juul, Cigarette1, first-generation (pre-2015) vape¹; IQOS, Cigarette2²; Nasal spray, Gum, Inhaler, Patch¹¹⁸; Zyn.¹¹⁹ Time zero is the first puff, insertion of the pouch, or attachment of the patch. Measures of variability have been omitted for clarity. All studies comprised at least 24 subjects at least 19 y of age. In one study,² the subjects had an average age of 34 y and were 52.5% male. Study¹¹⁸ is a review that compiled several other studies. Two studies^1,119 did not report on average age or % male.

Oral nicotine products (ONPs) include nicotine pouches, lozenges, gums, and inhalers and involve permeation through the buccal tissues rather than the alveoli. Transdermal patches eponymously use transfer through the skin.

The prolonged, declining phase of [nicotine]_t (Figure 1, area highlighted mins 10–60) occurs with a half-time of 1–4 hours after a single cigarette. The prolonged phase both suppresses withdrawal symptoms and maintains the cellular biological aspects of addiction.⁷ For a given individual, the prolonged declining phase probably differs less than the bolus among methods of intake, because the flux of nicotine into the blood and cerebrospinal fluid (CSF) has nearly stopped. An exception is the transdermal patch; after removal, there may be a plateau of 1–2 hours before the decline begins.

Within a given study (15–30 subjects), [nicotine]_t measured with a single method of nicotine intake shows great variability among subjects; a typical coefficient of variation (CV) is at least 0.5.⁸ When nicotine is administered intravenously in controlled doses (mg/kg), the CV is usually much smaller.⁸ The variability in [nicotine]_t among users is a strong reason to study individual [nicotine]_t records. Experiments will require orders of magnitude more subjects than is possible with existing pharmacokinetic methods to address open problems in addiction and therapy:

• Pioneering studies suggest that menthol prolongs the lifetime of [nicotine] in mice⁹ and humans,¹⁰ increases conditioned place preference to nicotine in mice,¹¹ and potentiates upregulation of nAChRs in mice and humans (see “Know the Physiology,” below).¹² Menthol might exert these effects via (1) its inhibition of cytochrome P450 2A6 (CYP2A6),¹³ (2) its inhibition of TrpA1 channels in respiratory pathways,¹⁴ or (3) its action as a chemical chaperone for nAChRs.¹¹ Mechanism (1), but neither (2) nor (3), would increase and prolong [nicotine]_t.

• On average, women experience more difficulty in smoking cessation than men.^15–17 Experiments with deuterium-labeled nicotine show that the prolonged declining phase is faster in women.¹⁸ Men seem to value the reinforcing effects of the bolus more than women¹⁵; Does this arise from sex differences in the bolus? How extensive are the individual variations? Does estrogen accelerate nicotine metabolism solely by inducing CYP2A6¹⁸?

• Do the varied effects of nicotine across the lifespan¹⁹ correlate with changes in nicotine pharmacokinetics, for instance slowed pharmacokinetics in elderly people^20,21?

• Do smokers titrate the bolus to achieve 100–150 nM nicotine, but no more?

• How does the bolus change with reduced-nicotine cigarettes^22–24 or adjustable e-cigarettes²⁵?

• How do individual vapers control [nicotine]_t during the day? Do some “seek the peak” and others “avoid the trough?”

2. “Know the Potential Therapy”: Personal Nicotine Replacement Therapy

Nicotine replacement therapy (NRT) will play a role in smoking cessation for the foreseeable future. Various national regulatory agencies (for instance, the US Food and Drug Administration [FDA]) have approved nicotine for smoking cessation via NRT in the form of transdermal patches, inhalers, and gum.

This approval does not yet extend to ENDS, which now receive authorization for marketing from the FDA’s Center for Tobacco Products (CTP) as “reduced risk products” (RRPs) under the Tobacco Control Act of 2009. Like many previous authors, we acknowledge the potential abuse liability of RRPs, including the possibility that RRPs could cause increases in dual use (vaping and smoking).²⁶

However, the FDA’s Center for Drug Evaluation and Research (CDER) has now cleared at least one “investigative new drug” (IND) application of an ENDS for investigation as a prescription-only device leading to smoking cessation. In an early iteration of the rationale for such devices, previous CTP and FDA leaders stated,²⁷ “We are examining possible steps the agency could take to address the pharmacokinetic performance of FDA-approved medicinal nicotine products to help more smokers quit. Factors for consideration may include the speed with which nicotine is delivered... [from] products capable of delivering nicotine without having to set tobacco on fire.” In a promising technology, piezoelectrically driven ultrasonic mesh inhalers produce controllable droplet size, require no carrier molecules such as propylene glycol and vegetable glycerin, and do not heat the liquid at all.^28,29 This could represent a step forward in individually controllable NRT.^30–32 The mesh nebulizer that received the IND clearance showed [nicotine]_t kinetics approaching that of a cigarette and much faster than a conventional inhaler.³³

Several metrics have been proposed as guidelines for effective NRT. One suggestion is to measure flux from the device to the subject³⁴; this can be measured in the device itself or by collecting nicotine with an impactor. A continuous nicotine monitor (CNM) measuring [nicotine]_t near the relevant nAChRs would be much more useful by capturing the best time resolution. It has been suggested³⁵ that the most successful means of smoking cessation involves the greatest rate of rise of [nicotine]_t, $\frac{d\left[nicotin{e}_t\right]}{dt},$presumably duplicating the “bolus.” An aspirational CNM must resolve the bolus, on an individual basis.

Newer techniques for delivering NRT also raise the specter of abuse liability in two ways. First, it must be noted that many individuals fail to quit smoking with buccal or transdermal NRT because they miss the bolus from a cigarette. If a prescription-only device, such as an ultrasonic nebulizer, perfectly copies the bolus of a user’s favorite cigarette, after a prescribed time course of NRT, a smoker might simply switch to a nicotine salt ENDS, which does optimize the bolus. We acknowledge that vaping manufacturers could use a CNM to enhance the bolus and, therefore, increase abuse liability.

Second, nicotine may well remain in the CSF for the prolonged declining phase, desensitizing nAChRs and provoking the cell biology of nicotine addiction. Newer ONPs also have abuse liability.

Despite these concerns about the abuse liability of NRT, one cannot deduce either the bolus or the prolonged declining phase during therapeutic NRT from first principles; we must perform dynamic, quantitative, individual measurements of [nicotine]_t. Individual measurements of [nicotine]_t are a highly appropriate component of research to test hypotheses that various forms of NRT benefit distinct subgroups of people defined by “tobacco abuse.”³⁶

3. “Know the Physiology”: Activation, Desensitization, and Chaperoning/Upregulation of Receptors

The CNM concept interfaces directly with nearly a century of research on time-resolved nAChR responses. Nicotine acts on nAChRs in only three ways, according to present knowledge. Nicotine activates nAChRs, producing ion flux (including Ca²⁺ flux) and depolarization. After a few minutes at the relevant nAChRs, nicotine desensitizes nAChRs. Slowest of all—hours to days—nicotine upregulates nAChRs via one or more post-translational processes, primarily chaperoning of nascent nAChRs within the endoplasmic reticulum (ER). The first two processes have received intense study since the identification of nAChRs, including experiments on isolated cells, brain slices, in vivo recordings on animal models, imaging of nAChR availability, functional magnetic resonance imaging, and heterologous expression, mostly via electrophysiological and Ca²⁺ recording.

Chaperoning/upregulation are both more complex and less well studied.^7,37–41 How is the prolonged phase crucial for nicotine dependence? Nicotine enters the ER, beginning a pathway that we term “inside-out.”⁷ This entry occurs within a few seconds after nicotine appears near cells, and at a concentration within 2-fold of the extracellular value.⁴² Importantly, even [nicotine] as low as 10 nM activates the “inside-out” pathway,⁴⁰ and in some people, [nicotine] continues to exceed 10 nM for ~6 hours after smoking/vaping.^42–46 A major effect of the inside-out pathway is to produce upregulation of plasma membrane nAChRs, and this upregulation is selective at every level studied—brain region, cell type, axonal versus somatic versus dendritic localization, and subunit stoichiometry of the upregulation nAChRs.⁷

Important simplifications are now in hand: It is generally agreed that the relevant nAChRs are α4β2*, α6β2*, and possibly α2β2* (the * denotes the possible presence of an additional subunit, such as α5, in the assembled pentameric receptor). Also, the relevant peak [nicotine]_t values are those achieved by smoking, <200 nM, in part because higher [nicotine] activates other nAChRs in aversive pathways. In another major simplification, measurements in CSF are unnecessary: nicotine reaches the CSF within seconds from brain capillaries because of its high logD_pH7.4.⁴⁷

Some aspects of nicotine pharmacokinetics are shaped by the effects of inhaled nicotine in the mouth, airways, and the alveolar epithelium; the nicotine concentrations in these compartments are much higher than considered above, and they take place at additional targets. (1) Deprotonated (free base) nicotine activates TrpA1, a channel abundantly expressed in airways.^14,48 This irritation could induce “braking” of respiration.⁴⁹ (2) Aerosolized nicotine in the alveoli equilibrates with the cytosol of alveolar epithelial cells, and this could lead to cytosolic concentrations of hundreds of millimolar (mM; Henry’s law). In alveolar epithelial cells, these concentrations may be high enough to chaperone and upregulate additional nAChRs, such as α3β4*.^50,51 These complications are additional reasons to measure [nicotine]_t in CSF-like compartments, downstream from measurements on nicotine flux from devices.

The CNM Concept Is Directly Relevant to Social and Behavioral Studies of Nicotine Intake

Because nicotine consumption—certainly for the abusive aspects, and probably also for the smoking cessation aspects—involves many cues, social support, and other activities such as eating and drinking, flavorings, and diurnal rhythms, it is crucial that measurements of [nicotine]_t occur during ad libitum consumption in daily life. The CNM we envision would satisfy this requirement.

The Population Assessment of Tobacco Health (PATH) study has resulted in 737 research papers as of mid-2024. The PATH study’s biomarker and biospecimen components have varied. PATH-like studies would be enhanced by 24-hour CNM records of [nicotine]_t, and we argue below that such records would be more useful than the currently employed urinary nicotine equivalents. Distributing and supporting the CNM could become as straightforward as the activities associated with continuous glucose monitors (CGMs; see below). If a study also collects tissue, blood, or saliva, the DNA analyses could further define additional components of nicotine metabolism, including CYP2B6⁵² and CYP2E1.⁵³

Present Measurements of [nicotine]_t

The classical ideas introduced so far have motivated measurements of [nicotine]_t on hundreds of subjects gathered in several recent reviews.^8,35,54 Two methods now provide “gold standard” measurements of nicotine pharmacokinetics. The intravenous blood-draw method is performed in a clinic on a subject with an intravenous catheter.^1,2,55–57 It costs thousands of dollars per subject, including tens of dollars for each sample in a typical series of 10–25 samples repeated at 5-minute intervals. The positron emission tomography (PET) method, which mixes ¹¹C-nicotine with inhaled smoking or vaping, has a better temporal resolution (a few seconds).^58,59 For both methods, a study typically includes 15–30 subjects. Both methods are too tedious and expensive for routine personal use. Neither can be routinely applied to youths.

Slower methods have been especially useful in showing that the prolonged, declining phase varies up to 10-fold among individuals, partially due to polymorphisms in cytochrome P450 2A6 (CYP2A6). CYP2A6 partially governs the conversion of nicotine to (nearly inactive) cotinine.⁶⁰ The “nicotine metabolite ratio” (NMR) measures CYP2A6 activity. NMR reveals that ~15% of the population are “slow” metabolizers: individuals with defective CYP2A6 have up to a severalfold longer [nicotine] half-life, and the increase depends on the genotype.⁶⁰ Slow metabolizers smoke fewer cigarettes per day; but especially for moderate to heaving smokers, they score just as strongly “dependent” on Fagerstrom-like tests.⁶¹

NMR does not measure the key data stream, [nicotine]_t. Furthermore, genetic risk scores capture only 34.5% of the variability in NMR.^62,63 Proxy measures for [nicotine]_t in blood include urinary nicotine metabolites. The only presently available physiological proxy for [nicotine]_t, increased heart rate,⁴⁵ is complicated for ad libitum intake because heart rate is influenced by several activities of daily living: exercise, simultaneous consumption of caffeinated beverages and nicotine, and tolerance to subsequent cigarettes.^64–66 Some ENDS measure puff strength, duration, and frequency (“smoking topography”).⁶⁷

Thus, [nicotine]_t is the relevant metric and

conventional intravenous (IV) and PET methods have too few subjects to resolve most confounding factors. In contrast, we envision a CNM that costs <100 USD (2024) to scalably study tens of thousands of subjects, transforming “confounding factors” for either therapy or abuse into “identifiable risk factors” or “personal smoking cessation therapy.”

Relationship to Conventional Pharmacological Concepts

Unbound Concentrations

All the techniques we envision will monitor the unbound concentration of nicotine⁺. This is the form sensed by the binding site of nAChRs, the active site of enzymes such as CYP2A6, and the binding site of biosensor proteins described below. However, the TrpA1 channel probably binds the deprotonated (free base) form.¹⁴

Volume of Distribution

The unbound concentration of any drug should be distinguished from the total amount of the drug in the organism.⁶⁸ This concept is usually associated with apparent volume of distribution. Volume of distribution in humans is conventionally inferred from measurements of the so-called clearance.⁶⁸ Most drugs have an apparent volume of distribution >1 L/kg, thought to arise either via drug binding to proteins, accumulation within lipids, or accumulation within acidic organelles (“acid trapping”). Nicotine has a relatively low volume of distribution, 1.8–4.2 L/kg,⁶⁹ with a consensus value of 2.6 L/kg.⁷⁰ That is, nicotine becomes bound or otherwise sequestered to the extent that the total amount of nicotine in the body would be appropriate to the volume of water equal to ~2.6 times the body’s weight (“corporal water volume”).

Most of these possibilities and mechanisms can influence [nicotine]_t. Therefore, some of these mechanisms would shape the data stream produced by a CNM—but none would challenge the validity of the [nicotine]_t data as they apply to actions on nAChRs.

Area Under the Curve of Concentration × Time

Area under the curve (AUC) is often invoked to compare actions of therapeutic drugs. For nicotine, abuse potential is sometimes assumed to increase with AUC,^54,71,72 but this idea has little experimental proof. More generally, the goal to “know the physiology” will transcend the AUC metric by showing how the bolus and the prolonged phase affect activation, desensitization, or chaperoning of nAChRs.

Interstitial Fluid Is the Most Relevant Compartment

CSF bathes highly nicotine-sensitive nAChRs in several distinct brain regions; therefore, [nicotine]_t in CSF governs activation, desensitization, and upregulation of nAChRs. However, no minimally invasive or noninvasive, nonradioactive technique is available for measuring [nicotine]_t in CSF. For all means of nicotine intake, the blood–brain barrier, formed by tight junctions in cerebral capillaries and by the end-feet of astrocytes, provides a delay of at most a few seconds in achieving a CSF [nicotine]_t that essentially equals blood [nicotine]_t.^8,47

As in most organs, brain cells lie within 50 µm of capillaries. Nicotine can diffuse passively over 50 µm within 5 seconds (assuming a diffusion constant of 0.5 µm²/ms). Smokers and vapers feel a “buzz” within ~20 seconds, showing that nicotine has appeared in the CSF and reached nAChRs. Rat measurements confirm that [nicotine]_t in CSF is close to that in plasma.⁷³ Special intra-arterial [¹¹C]nicotine PET measurements in humans achieve a resolution of ~5 seconds,^4,6,58 which is unnecessary for most studies of [nicotine]_t.

Interstitial fluid (ISF) is minimally invasive. Nicotine is predicted to be slightly more permeable to capillaries in peripheral tissue than to brain capillaries; but the difference is likely to be just a few seconds. In this context, time-resolved measurements of [nicotine]_t in ISF are presumably more relevant than blood to [nicotine]_t in CSF.

Commercial tobacco manufacturers modify the taste and and/or pH of cigars and cigarettes; and this may also modify [nicotine]_t.^74,75 In nicotine salt ENDS, the pH of the e-liquid and of the aerosol is lower than cigarettes; but the higher solubility of nicotine⁺ than of free base nicotine allows increased [nicotine⁺] in the e-liquid and in the inhaled aerosol. Manufacturers of nicotine salt ENDS may also choose an anion that affects airway flow.^6,76 All NRT strategies also manipulate nicotine pharmacokinetics.^8,35

On the one hand, the various effects on nicotine entry into the blood cannot yet be predicted simply either from the amount of e-liquid left in the reservoir, from the volume of each puff, or from the nicotine deposited on an impactor after a controlled number of mechanical puffs. On the other hand, once the nicotine enters the blood, its concentration (<1 μM) is so low that buffers in the blood maintain its pH at ~7.4. If a nicotine salt is ingested, the blood similarly dilutes the anion, which therefore need not be considered further. As noted above, the nicotine passes within seconds from the blood to the CSF and other compartments such as ISF. Therefore, [nicotine]_t in the blood or ISF is an excellent measurement of the delivery system’s total pharmacologically relevant, dynamic result—whether the delivery system is a combustible cigarette, an ENDS, an ONP, or NRT.

In summary, we strongly favor interstitial fluid as the compartment of choice for measurements of personal [nicotine]_t.

Analogy to CGMs

All FDA-approved CGMs reside in ISF, are inserted with minimally invasive techniques, transmit their data, [glucose]_t, wirelessly, and have a temporal resolution of ~5 minutes. One FDA-approved glucose monitor uses fluorescent detection,^77,78 although via a different molecular mechanism than the one described below.

Most CGMs now in use employ electrochemical detection via glucose oxidase positioned at the intradermal tip of a 5-mm-long gold wire. The latest electrochemical CGMs last ≥10 days. CGMs have become commodity items, available over the counter ($50) for use by prediabetic patients and by nondiabetic consumers to satisfy their curiosity about their [glucose]_t. Several labs are now attempting to develop CGMs in essentially noninvasive form: arrays of microneedles a few hundred micrometers long and tens of micrometers wide, with their tips in ISF.

Sweat

Although sweat measurements have the great advantage of being noninvasive, other aspects lead to uncertainties that sweat can provide quantitative, time-resolved measurements relevant to CSF. Sweat is secreted through a complex duct at varying pressures. According to a biofluidic model of sweat glands, a biomolecule secreted by the gland reaches the skin surface within <5 minutes, fulfilling the temporal resolution criterion of a CNM.⁷⁹ Active microfluidics may then be required in the sensor device to transport the sweat to the sensing surface or chamber. This would introduce a delay; but if the flow is laminar, the time of actual secretion can be determined.⁸⁰ Primary sweat (in the sweat coil) is quite acidic,⁷⁹ leading to concerns about acid trapping, which may explain initial measurements that [nicotine] in sweat is >10 times higher than in blood.⁸¹ It is not understood whether the walls of the sweat duct would then allow reabsorption of nicotine. Researchers in the sweat sensor field are working to eliminate these complications, with the goal of realistic measurements on a known time scale.^82–85

Specifications for a CNM

Concentration Resolution: ~10 nM

Although the EC₅₀ for nicotine activation of nAChRs is typically several hundred nanomolar (nM), much lower concentrations desensitize or upregulate nAChRs.⁷ For instance, the EC₅₀ for upregulation is 37 nM.⁴⁰ We therefore consider it necessary to measure [nicotine]_t as it declines to values as low as ~10 nM.

Absolute [Nicotine] Calibration

The signal from all copies of the CNM should have an invariant, simple relationship to [nicotine]_t. In Figure 2, we show that a candidate fluorescent sensor molecule displays a linear response in the relevant range of [nicotine], and we refer to the sensitivity as δ-slope. We would accept an absolute accuracy of ±20%. Modern electrochemical CGMs do have absolute calibrations, in part because a few CGMs from each production batch are tested at varying [glucose]; the resulting calibrations are then embedded in all CGMs from the batch.⁸⁷

Figure 2.

Molecular basis of the iNicSnFR approach to a CNM. A, Schematic view of the single-chain fluorescent sensing molecule such as iNicSnFR12. B, A single frame from Supplementary Video S1, depicting the details of the conformational change that underlies sensing of nicotine by iNicSnFR12. C, Concentration–response curve at [nicotine] in the range 0–180 nM, the expected plasma, CSF, or ISF range during nicotine intake via smoking, vaping, transdermal patches, inhalers, or oral nicotine products.⁸ Measurements were conducted in 96-well microtiter plates, and each well contained 100 nM purified iNicSnFR12 in 100 μL of solution. F₀ and ∆F are the resting fluorescence and additional nicotine-induced fluorescence, respectively. The excitation and emissions wavelengths are 496 and 535 nm, respectively. From Haloi et al.⁸⁶ CSF = cerebrospinal fluid; CNM = continuous nicotine monitor; iNicSnFR = intensity-based nicotine-sensing fluorescent reporter; ISF = interstitial fluid.

Temporal Resolution: 5 Minutes

The CNM must have the temporal resolution to measure the bolus of [nicotine]_t (~100 nM for ~5 minutes, Figure 1, area highlighted mins 0–10). This implies a temporal resolution of ~300 seconds: the typical 10-puff period for a single cigarette or a typical ENDS “vaping session.” Juul and IQOS achieve 60–150 nM nicotine during this 300- to 600-second time frame (Figure 1).

Selectivity: At Least 100-Fold for Nicotine

The ideal CNM would have zero response to other molecules in the ISF. We aim to render the sensing molecules 100-fold more sensitive to nicotine than to all other ligands. According to this specification, even if an endogenous interfering molecule has a 100-fold higher concentration higher than nicotine, this would not markedly distort the [nicotine]_t signal.

Durability: At Least 24 Hours

A suitable first-generation CNM would measure [nicotine]_t for roughly 24 hours. This would allow 40 or more cigarettes or vaping sessions, providing an excellent view of the prolonged phase. Including a full 24 hours has the additional advantage that [nicotine]_t is expected to fall to nearly zero during the night, providing an important set of control readings.

Measurements of [nicotine]_t during research on smoking cessation would require much longer records. By analogy with CGMs, a 10-day period seems approachable. Mice with brain expression of cpGFP-based biosensors provide signals for as long as 1 year.⁸⁸ Mice expressing an adeno-associated virus encoding iFentanylSnFR, which differs by only a dozen amino acids from intensity-based nicotine-sensing fluorescent reports (iNicSnFRs), continue to give useful responses to fentanyl for at least 3 weeks.⁸⁹

Available Molecular Technology for a CNM

Nicotine-Binding Protein-Based Approaches: The iNicSnFR Family

The iNicSnFR family^42,86,90 is based on ~50 years of research on periplasmic binding proteins (PBPs), comprising some 20 000 genes in bacterial and archaeal species. The PBP binds a small molecule (ligand) of importance to the microorganism (in an intact bacterium, the PBP would then present the ligand to other proteins). More than 450 atomic-scale structures for PBPs now exist in the protein data bank RCSB. In all known cases, the PBP responds to the ligand binding by undergoing a conserved “Venus flytrap” or “clamshell” conformational change.⁹¹ Most known PBPs exist as monomers, and the conformational change depends on the ligand conformation with a Hill coefficient near 1. This well-understood change requires <1 second in most cases, rendering the PBP suitable to become the molecular basis of a real-time, continuous, reversible, reagent-less sensor (Figure 2).

The PBP of most interest for a CNM is a mutated variant of OpuBC from a hyperthermophilic bacterial species which uses choline or betaine as an osmolyte. We mutated OpuBC to favor binding of acetylcholine⁹² or nicotine.⁴² Each member of the iNicSnFR family is a single protein chain. The chain represents a nearly modular merger of two moieties: the PBP binding moiety and a fluorescent readout moiety. We use the term “merged” to avoid implying that these two moieties are simply concatenated, C-terminal of the first connected to N-terminal of the second. The GFP moiety is circularly permuted (cpGFP), as in GCaMP sensors for Ca²⁺. The circularly permuted cpGFP is then inserted into the PBP sequence, near the hinge region. Recently, the iNicSnFR project was aided by adding computational predictions^93–95 to the directed evolution pipeline. iNicSnFR12 achieves nearly the desired sensitivity in 96-well saline-based lab tests (Figure 2C).⁸⁶ The concentration–response relation is linear in the range shown, because the EC₅₀ for the entire relation is >1 μM.⁸⁶

The iNicSnFR approach exploits the amplification produced by modulating the fluorescence on the basis of nicotine binding to the PBP moiety. During each second that the single nicotine molecule is bound to the PBP, the fluorophore absorbs as many as 10⁸ blue photons, then emits ~0.7 times as many green photons.⁸⁶ The optical elements resemble those of a modern fluorescence microscope, simplified further by measuring with a single photodetector (“fiber photometry”). Figure 3 presents a schematic for our planned next-generation CNM.

Figure 3.

Design for a miniaturized fiber photometric CNM, based on initial progress with iNicSnFR12 entrapped in a hydrogel.⁹⁶ A, The iNicSnFR-containing hydrogel is placed in ISF, at the end of a fiber optic inserted by a spring apparatus like those in some CGMs.¹²⁰ The miniature fiber photometer is 14 mm tall, including the protruding fiber optic. There are two time-multiplexed excitation LEDs. Excitation at 470 and 405 nm yield nicotine-dependent and nicotine-insensitive fluorescence, respectively,^42,121 a common tactic to control for movement artifacts and bleaching. The SolidWorks file is available from the authors. B, Simulations of radial diffusion in a cylindrical hydrogel.^122,123 A hydrogel 400 μm in diameter contains 20 μM iNicSnFR. We assume that nicotine has a diffusion constant, D, of 0.3 μm²/ms in the hydrogel containing no iNicSnFR (~2/3 of the free-solution D). Effective D is further reduced by rebinding to iNicSnFR molecules with a K_d of 10 μM. Following a jump from [nicotine] = 0 to C₀ at time 0 in the external solution, [nicotine]_t within the gel approaches 90% of C₀ within 300 s, satisfying the criterion for temporal resolution. C, In the design, the hydrogel contains 6.28 pmol of iNicSnFR, 20 μM. This is close to the 10 pmol of our standard assay condition in which a 100 μL microtiter well contains 100 nM iNicSnFR12 (see Figure 2). CGM = continuous glucose monitor; CNM = continuous nicotine monitor; iNicSnFR = intensity-based nicotine-sensing fluorescent reporter; ISF = interstitial fluid.

We presented preliminary data for an approach in which the purified iNicSnFR molecules are trapped within a hydrogel, at the tip of a fiber optic.⁹⁶ The hydrogel would be surrounded by ISF. The hydrogel approach (1) prevents other proteins from interacting with sensor molecules, distorting F₀ or ∆F, (2) avoids increasing F₀ by endogenous cellular fluorescence, and (3) minimizes contact with immune cells. The purified protein-hydrogel approach differs from most cpGFP sensors, which are expressed in animal models via viral vectors or transgenes. Points (1) and (2) also allow us to plan ways for absolute calibrations of δ-slope (see Figure 2C).

The Redox-Modified PBP Method

This method (Figure 4A) was suggested as early as 2001,⁹⁷ then lay fallow until 2021. A recent report confirmed the concept, for continuously sensing glutamine with the eponymously named glutamine binding protein.⁹⁸

Figure 4.

Electrochemical approaches to a CNM. A, PBP bearing a redox group. The exemplar group is methylene blue. B, Conceptual view of a PBP in a nanopore. The apo-PBP from iNiCSnFR3a (excised from pdb file 7S7W) was manually positioned within the vestibule of hemolysin E from E. Coli K12 (pdb 2WCD), using ChimeraX. The complex was conceptually positioned in a bilayer membrane. CNM = continuous nicotine monitor; iNicSnFR = intensity-based nicotine-sensing fluorescent reporter; PBP = periplasmic binding protein.

Site-selective modification would attach the redox probe to the nicotine-binding PBP at a sequence position that experiences an acceptably large movement (10–20 Å) upon ligand binding (see Figures 2B and 4A). We have defined several candidate residues, based on liganded and unliganded structures of iNicSnFR12. The original previous study, with other ligand–PBP pairs, coupled redox-active groups to introduced cysteine residues^97,98; and we have begun with this tactic. Our preliminary experiments have utilized Cys variants of iNicSnFR12 and also the PBP alone, without the cpGFP moiety (“PBP12”).

For the redox probe, one can choose among more than a dozen molecules.⁹⁹ We have begun experiments with methylene blue. Phenazine derivatives⁹⁸ and ruthenium (Ru(II))⁹⁷ have also been used with PBPs. In the most general tactic, one would site-selectively introduce noncanonical azido amino acids. Our preliminary studies have produced signals that show dose dependence on nicotine. However, we are still experimenting with conditions that yield satisfactory control data and also yield reversibility suitable for continuous sensing.

The electrodes for the intradermal version of the electrochemical CNM would be subcutaneous gold wires (0.5 mm diameter, 5 mm length) resembling those in CGMs—though with the more sophisticated analog processing required by pulsed techniques such as square-wave voltammetry. In another variation, the PBP could be coupled to a graphene monolayer that acts as the gate of a field-effect transistor.¹⁰⁰

The PBP-Gated Nanopore Approach

A single nicotine binding event could, in principle, be amplified electrically if the binding gates a channel or a nanopore that passes 10⁸ ions/s (a typical single-channel current) while the nicotine is bound (Figure 4B).^101,102 In a pioneering report, Salmonella typhi nanopores were embedded in planar bilayers. Conventional electrophysiological amplifiers and software were used in single-channel mode to analyze the nanopore current. Each of 13 different metabolite-binding PBPs proved suitable for ligand detection.¹⁰³ These were expressed in bacterial systems, purified, and added to the chambers separated by the membrane. The PBPs produced a range of fluctuations in the nanopore current, typically on a time scale of seconds. Adding the cognate ligand for the PBP then produced additional current fluctuations. Usually the liganded (“closed”) conformation of the PBP blocked the current more completely, suggesting that “as the structure of the protein becomes more compacted, the protein penetrates deeper inside the nanopore, resulting in more current being blocked.” Importantly, the EC₅₀ of the ligand-induced current fluctuations was similar in all cases to the ligand–PBP interaction in solution.

Thus, the PBP-gated nanopore approach is biophysically tractable and pharmacologically understandable. Although no PBP was studied in the class F family which contains OpuBC (the molecule that was evolved to yield PBP12),⁹¹ these data are encouraging.

A deployable CNM need not be constrained by fragile lipid bilayers and single-channel measurements. The membrane would be a synthetic polymer, as used in pore sequencing; and many PBP12–nanopore complexes would be embedded in parallel, providing a macroscopic signal proportional to [nicotine]_t. This would utilize presently available high-throughput planar electrophysiological methods, for assaying candidate variants in a protein engineering campaign that develops and optimizes the PBP–nanopore complex. The supported membrane and electrodes would then be fixed to the tip of CGM-like probe.

Nicotine-Binding Aptamer-Based Approaches

In the present context, an aptamer (a synthetic single-stranded oligonucleotide) would be attached to an electrode. The aptamer would be evolved or selected to bind a nicotine molecule, changing shape. This would change the distance between an attached redox-active group and the electrode, resulting in a signal analogous to the redox-modified PBP.¹⁰⁴ Electrochemical aptamer-based sensors have been developed for metabolites, therapeutic drugs, and abused drugs; but unfortunately, there is not yet a reported aptamer with selective affinity for nicotine. Furthermore, we expect a poorer dynamic range and lifetime from aptamers compared to fluorescent hydrogels.¹⁰⁵

Nicotine-Oxidizing Enzymes

In part because amperometric glucose sensors have found widespread use, investigators have considered oxidoreductase enzymes that already use nicotine as a substrate. The major challenge arises because we have specified that [nicotine]_t must be measured at 10⁶-fold lower levels than [glucose]_t, and each nicotine molecule that becomes oxidized by the enzyme contributes only a single elementary charge to the current.

Nicotine oxidase from Pseudomonas putida NicA2, was studied with amperometry.¹⁰⁶ Selectivity for nicotine is excellent. A mutated version became the basis for an issued patent,¹⁰⁷ aimed primarily at measurements in sweat. The limit of detection was 1 µM, yielding a current of 0.01 µA for an electrode area that was probably 3 mm². An intradermal electrode would presumably have a 100-fold smaller area. This would give a signal to 10 nM nicotine of just 1 pA, a challenge to measure. Studies on Shinella sp. HZN7 NctB, another nicotine-oxidizing enzyme, show that mutations can increase turnover number only ~2-fold from the wild-type enzyme.¹⁰⁸

The study of Tai et al.⁸¹ claimed to measure the current associated with CYP2B6 oxidation of nicotine.¹⁰⁹ There was no control for the additional oxidative currents; it is likely that the enzyme simply increased previously measured faradaic currents¹⁰⁹ by ~2-fold, leaving the sensitivity at ~100-fold less than our goals.

Data Analysis

Pioneering instruments generally produce suboptimal data. It may not be possible for a CNM to detect when a subject takes a puff or otherwise ingests nicotine. We plan to equip the initial iNicSnFR12-based CNMs with three pushbuttons. Buttons 1, 2, and 3 “time-stamp” the signal at the beginning of a cigarette, a vaping session, or a nicotine pouch, respectively. In simulations, vast reductions in noise can result (Figure 5).

Figure 5.

Synchronized averaging will reduce noise in CNM recordings. A, A simulated noisy CNM data stream. The dose table for the simulator⁷² contained observed ad libitum times for one smoker¹²⁴ (21 cigarettes, total; we assumed 1 mg doses). We added white noise, single-pole filtered at τ = 5 min (the target temporal resolution) with an RMS deviation of 29 nM (roughly the present resolution of iNicSnFR12). The resulting [nicotine]_t signal is poorly discernible. B, Using the dose table, we retrospectively synchronized and averaged the traces during 10 min before and 60 min following the beginning of each cigarette. As expected, this decreased the noise by a factor $\sqrt{21}$. The resulting denoised average trace approximates the peak and waveform of the noiseless simulated [nicotine]_t for a single cigarette. CNM = continuous nicotine monitor; iNicSnFR = intensity-based nicotine-sensing fluorescent reporter; RMS = root mean square.

It would be desirable to automatically detect smoking/vaping episodes, so that subjects can ingest ad libitum without the added burden of time-stamping the dataset.¹¹⁰ Machine leaning algorithms are being developed to identify meal times and predict individual patients’ glucose trajectory from present CGM measurements.^110,111 How much of this effort is useful for the 24-hour measurements we envision for the CNM? We suggest that ~500 subject-days could provide an initial time-stamped dataset analogous to the glucose sensor time series in the OhioT1DM dataset, which contain 56 days × 8 type 1 diabetic patients.¹¹² This dataset could then provide a benchmark for developing new smoothing and curve-fitting algorithms.

The Caltech authors include a group with expertise in developing wearable devices. Like several other groups, we have reported on-device analog and digital signal processing, Bluetooth-based wireless communication, and battery power. These seem suitable for a CNM.^113,114

A Not-for-Profit, Collaborative Research Agenda: Summary, Economics, and Support

Figure 6 summarizes this review in a cartoon form. Developing a CNM is a research agenda; and the outcome will be primarily a wearable instrument for research on the neuroscience of human nicotine use.

Figure 6.

A cartoon summary of this review’s stated motivation for a measurements of individual [nicotine]_t, our present opinions on the most likely path to achieve a wearable CNM, and our suggestions for further analysis of [nicotine]_t. The lungs, especially the bronchioles and alveoli, are the most frequent route for nicotine to reach the bloodstream. Because ISF [nicotine]_t strongly resembles blood and cerebrospinal fluid (CSF) [nicotine]_t (see text), a minimally invasive intradermal CNM provides appropriate measurements of [nicotine]_t. Entrapping a purified protein of the iNicSnFR family in a hydrogel has promise as a molecular sensing strategy with the stated specifications. During ad libitum smoking or vaping, a photometric CNM can also incorporate compensation and normalization algorithms (not shown in detail) to result in the [nicotine]_t signal. The wearable CNM communicates [nicotine]_t via Bluetooth to a portable device. The [nicotine]_t trace shown is simulated from the smoking times observed for ad libitum smoking by one exemplar subject over the time course of one day.^124,125 In the future, further time series analysis of many actual individual [nicotine]_t records can enlighten research on nicotine and tobacco. Created partially in https://BioRender.com. CNM = continuous nicotine monitor; iNicSnFR = intensity-based nicotine-sensing fluorescent reporter.

The example we have given, PATH-like research programs, might call for 30 000 CNMs/year. At these quantities, in the design of Figure 3, each thin-film wafer would be cut to yield 75 optical components (excitation filter, dichroic mirror, or emission filter). The total cost of these three components would be ~$18 per CNM. The electronics, including a Bluetooth chip with sufficient processing power to normalize and correct the signal, would cost ~$22 per CNM. Other components, including purified iNicSnFR protein, plastic fiber optic, lens, 3D-printed body, and battery, would have a total cost of ~$20. Therefore, we envision a cost of <$100 per CNM. If each subject in a PATH-like study wears a CNM for 24 hours in association with each annual interview, the additional cost will be $3 M/year—a single-digit percentage of the PATH budget.

We advocate a research agenda analogous to the ongoing, highly collaborative, international development of two other neuroscience instruments: miniature fluorescent microscopes¹¹⁵ and high-density multielectrode probes.¹¹⁶ Initial experiments can and must use rodent models; but unlike the two examples given, the research agenda must progress to measurements in humans with all deliberate speed. The medical device industry is largely driven by clinician uptake and insurance reimbursement; however, no reimbursement path for a CNM exists today. Therefore, the tobacco control, smoking cessation, and regulatory communities must decide whether public sources or a nonprofit startup¹¹⁷ provide the most appropriate route for supporting the development of a CNM.

Supplementary Material

Supplementary material is available at Nicotine and Tobacco Research online.

References (101–125) are available as Supplementary Material.

Funding

H.A.L. was supported by TRDRP Grant 27IP-0057, National Institute on Drug Abuse (NIDA) Grant DA049140, and National Institute of General Medical Sciences (NIGMS) Grant GM-123582. N.L.B. is supported National Institute on Drug Abuse (NIDA) Grant DA039264. N.J.F. was supported by the Biotechnology Leadership Program through the Rosen Bioengineering Center at Caltech. D.A.D., H.A.L., H.L., S.L.M., A.K.M., and W. G. were supported by internal Caltech grants from the Merkin Institute for Translational Research, the Rosen Biotechnology Center, the Carver Mead New Adventures Fund, and the Sensing to Intelligence Fund.

Declaration of Interests

Dr. Benowitz is a consultant to Achieve Life Sciences and Qnovia, companies that are developing smoking cessation medications, and has been a expert witness in litigation against tobacco companies. This review was written without active participation or encouragement by any person associated with the Population Assessment of Tobacco Health (PATH). The references to PATH exemplify the type of study that could complement use of a continuous nicotine monitor for understanding [nicotine]_t during ad libitum nicotine intake. Dr. Wei Gao is co-founder and advisor at Persperity Health. Healther Lukas is an employee at Persperity Health.

Author Contributions

Aaron Nichols (Funding acquisition [equal], Investigation [equal], Writing - review & editing [equal]), Christopher Marotta (Investigation [equal], Writing—original draft [equal]), Heather Lukas (Investigation [equal], Writing - review & editing [equal]), Nicholas Friesenhahn (Investigation [equal]), Daniel Wagenaar (Conceptualization [equal], Software [equal]), Stephen Mayo (Conceptualization [equal], Funding acquisition [equal]), Dennis Dougherty (Conceptualization [equal], Supervision [equal]), Neal Benowitz (Conceptualization [equal], Funding acquisition [equal], Writing - review & editing [equal]), Wei Gao (Conceptualization [equal], Supervision [equal]), Anand Muthusamy (Conceptualization [equal], Writing—original draft [equal], Writing - review & editing [equal]), and Henry Lester (Conceptualization [equal], Funding acquisition [equal], Software [equal], Writing—original draft [equal])