Quantitative Factors Introduced in the Feasibility Analysis of Reactive Oxygen Species (ROS)-Sensitive Triggers

Abstract

Reactive oxygen species (ROS) are critical for cellular signaling. Various pathophysiological conditions are also associated with elevated levels of ROS. Hence, ROS-sensitive triggers have been extensively used for selective payload delivery. Such applications are predicated on two key functions: (1) a sufficient magnitude of concentration difference for the interested ROS between normal tissue/cells and intended sites and (2) appropriate reaction kinetics to ensure a sufficient level of selectivity for payload release. Further, ROS refers to a group of species with varying reactivity, which should not be viewed as a uniform group. In this review, we critically analyze data on the concentrations of different ROS species under various pathophysiological conditions and examine how reaction kinetics affect the success of ROS-sensitive linker chemistry. Further, we discuss different ROS linker chemistry in the context of their applications in drug delivery and imaging. This review brings new insights into research in ROS-triggered delivery, highlights factors to consider in maximizing the chance for success and discusses pitfalls to avoid.

1. Introduction

The important roles of reactive oxygen species (ROS) in cellular signaling are widely recognized.^[1] ROS concentration differences between physiological and pathological states have been exploited for the development of selective payload delivery approaches.^[2] The design principle almost always depends on the use of a ROS-sensitive functional group that gets “activated” upon reaction with various types of ROS. Fundamentally, such reactions are bimolecular events with the activation kinetics dependent on the reactivity and local concentrations of the two components: the ROS-sensitive group and the ROS. This is an important issue because the rate of activation by ROS needs to be fast enough to overcome the equalizing effects of diffusion, re-distribution, metabolism, and degradation of the ROS itself. Further, the term “reactive” in ROS does not mean equal reactivity of all the ROS. Hydrogen peroxide (H₂O₂) is fairly stable under normal physiological conditions. In contrast, hydroxyl radical (OH^•) reacts with almost everything instantaneously upon formation. In addition, some ROS are one-electron oxidants including OH^•, superoxide (O₂^•−), alkoxyl radical (RO^•), nitrogen dioxide (NO₂^•), and alkyl peroxyl (ROO^•); while others are two-electron oxidants, including H₂O₂, hypochlorous acid or hypochlorite (HOCl/OCl⁻), and peroxynitrite (ONOO⁻).^[3] Within the same class, the reactivity of the individual ROS can be very different. For instance, the reaction of H₂O₂ with glutathione has a low second order constant of 0.9 M⁻¹s⁻¹, whereas the same reaction is very fast with HOCl (k = 3×10⁷ M⁻¹s⁻¹).^[3] Therefore, treating all ROS as a uniform group is an over-simplification. There is a need to consider the interplay among concentration of ROS, reaction type and kinetics, and desired time scale of payload activation for optimal outcomes. These points are congruent with issues raised in a recent consensus paper on the need to understand molecular details in studying ROS biology and payload activation.^[4] In our own research related to ROS-sensitive chemistry^[5] and ROS-signaling in biology,^[6] we have realized that ROS represents a diverse group of reactive species, and ROS-signaling is more than a “binary event,” as it is portraited sometimes. ROS is present in healthy and pathological cells at varying concentrations depending on cell type and culturing conditions. Further, literature reports indicate that ROS is associated with a number of pathological conditions such as inflammation,^[7] cancer,^[8] diabetes,^[9] neurodegenerative conditions,^[10] cardiovascular diseases,^[11] and so forth. Then, there is a question as to how much difference in ROS concentration is sufficient for “selective” ROS-sensitive payload release. Of course, there is not a universal answer, given the complexity of the interplay among activation chemistry, ROS type involved, reaction kinetics, and desired outcome. Herein, we critically analyze the literature in a quantitative context whenever possible and propose issues to consider. We organize this review by first providing an overview of the various ROS species and their production. Second, we analyze their reactivity and reaction kinetics with various ROS-sensitive moieties. This aspect is important for applications. Third, we provide a section on ROS concentration determination methods and their associated issues. Fourth, we provide a critical summary of concentrations of various ROS under different conditions. Finally, we provide an analysis of several reported ROS-sensitive triggers and their applications. The last section is not meant to be comprehensive, but to highlight key issues.

2. ROS Generation and Reactivity

2.1. The origin of ROS.

ROS are formed by the partial reduction of oxygen^[13] with the most abundant being hydrogen peroxide (H₂O₂), hypochlorous acid or hypochlorite (HOCl/OCl⁻), superoxide anion (O₂^•−), hydroxyl radicals (OH^•), and singlet oxygen species (¹O₂). ROS are byproducts of various redoxreaction in different cellular compartments such as mitochondria, cytoplasm, endoplasmic reticulum, and peroxisomes.^[14] Consequently, elevated ROS levels in these organelles are explored as an endogenous stimulus for subcellular targeted delivery of payloads.^[15] Specifically, molecular oxygen (O₂) serves as the final electron acceptor in the electron transport chain, leading to the formation of H₂O as the final form as the result of a four-electron reduction. Often, partial reduction of oxygen results in the formation of highly reactive superoxide anion (O₂^•−),^[16] a primary ROS species which is a precursor of different secondary ROS species. Spontaneous dismutation from two molecules of O₂^•− produces one molecule of H₂O₂ and O₂.^[17] Similarly, NADPH-oxidase in phagocytic cells produces a “respiratory burst” by the secretion of O₂^•− and spontaneous dismutation to form H₂O_2.^[12] The dismutation reaction is ultrafast with a rate constant of 5×10⁵ M⁻¹s⁻¹, which can be further increased by 10⁴-fold by superoxide dismutase (SOD) to a rate constant of 5×10⁹ M⁻¹s⁻¹.^[18] This implies that any other reactions of O₂^•− must outperform such dismutation rate in order to be kinetically relevant. Formation of peroxynitrite (ONOO⁻) from NO^• and O₂^•− is the only reaction that could match this kinetic requirement (k = 1.9 × 10¹⁰ M⁻¹s⁻¹).^[19] Unlike O₂^•−, H₂O₂ is the most stable ROS and is diffusible across the cell membrane. Although less reactive than O₂^•−, H₂O₂ undergoes metal-catalyzed partial reduction to produce the most reactive ROS: OH^•.^[20] Moreover, myeloperoxidase (MPO) catalyzes the reaction of H₂O₂ with chlorine to produce HOCl (Figure 1).^[21] H₂O₂ is also reduced to water by cellular peroxidases such as catalase (Cat), glutathione peroxidase (GPX), and peroxiredoxins (PRDX). Additionally, ROS are also produced in the endoplasmic reticulum by CYP-450 mediated detoxification and biotransformation of different xenobiotics.^[22] Similarly, the peroxisomes are enriched with H₂O₂generating enzymes such as D-amino acid oxidase and acetyl-Co-A oxidase, leading to subsequent peroxidative reactions and beta oxidation of fatty acids.^[23] Under physiological conditions, ROS serves as signaling messenger(s).^[24] However, when levels of ROS production exceed the physiological needs, that condition causes “redox-stress,”^{[13b, 14a]} leading to oxidative damage of biomacromolecules such as DNA, lipids, and proteins.^{[14, 25]} Hence, failure to maintain redox homeostasis is involved in the pathophysiology of different diseases such as tumor,^[26] acute kidney injury,^[27] autoimmune diseases,^[28] diabetes,^[29] and aging.^[30]

Figure 1:

Different types of ROS and their formation and transformation pathways.^[12]

2.2. The need for quantitative analysis of ROS-triggered events.

Blood concentrations of H₂O₂ have been reported as ~1–5 μM,^[31] whereas the cells under stress may have as high as 5 mM H₂O₂.^[32] Such kind of high concentration needs further validation, but provides a boundary condition of H₂O₂. Because of the higher ROS level under pathological conditions, ROS has been widely explored as an endogenous stimulus for targeted payload delivery,^{[13a, 33]} through a two-step process: an initial ROS-mediated oxidation of a “linker” or protective moiety and subsequent payload release. Herein, the initial oxidation step is a bimolecular reaction and the rate of the reaction depends on the reactivity and concentrations of both reactants.^[34] Thus it is critical to analyze the reactivity of the intended “linker” toward the ROS in question to ensure timely activation. It is understood that each application is different. Therefore, this section is meant to provide an analysis of boundary conditions, not specific requirements for various applications. To put all reaction kinetic questions in perspective, methionine undergoes rapid oxidation by hypochlorite^[35] with a rate constant of 3.4 ×10⁷ M⁻¹s⁻¹, but sluggish oxidation by hydrogen peroxide^[36] with a rate constant of 8.8 × 10⁻³ M⁻¹s⁻¹. In other words, the first half-life (t_1/2) of the reaction is calculated to be 0.29 ms for OCl⁻ and ~315 h for H₂O₂ when each reactant is at 100 μM. The rate could be even slower with a decrease in concentration (10 μM): 2.9 ms and 3150 h for OCl⁻ and H₂O_2, respectively. These t_1/2 values make it readily apparent that reaction kinetics are very important for feasibility consideration.

2.3. An intuitive presentation of the reaction time scales for various known ROS-sensitive linkers.

We have generated a set of plots of the relationship among ROS concentration, second-order rate constant, and the estimated t_1/2 (Figure 2) to provide an intuitive way of estimating reaction timescale. Figure 2a shows t_1/2 estimation for different linkers when ROS is in excess, using published rate constants (Table 1) and assuming pseudo-first order kinetics. For example, thioketals have been widely used as a ROS-sensitive trigger for payload release. With the reported second order rate constant of 1.18 × 10⁻⁵ M⁻¹s⁻¹,^[37] the t_1/2 is calculated to be 136 days at a concentration of 5 mM of H₂O₂. However, for hypochlorite, the rate constant for an analogous reaction is on the order of 10⁷ M⁻¹s⁻¹,^[38] leading to a t_1/2 of 0.69 ms even at 100 μM of the HOCl. Such an analysis indicates that any in-vivo activation of such a trigger is probably through oxidation by hypochlorite (or other more reactive ROS) instead of H₂O₂. Figure 2a also highlights examples of boronate oxidation being the ROS-trigger with t_1/2 on the scale of min or ms even at 100 μM ROS. The above examples emphasize the need to consider all ROS species when designing a ROS-trigger for payload delivery. Further, the selectivity of a linker for various ROS species is likely entirely dependent on the chemical reactivity; and it will be very challenging to design H₂O₂-selective triggers in the presence of other more reactive species such as hypochlorite. Figure 2b shows scenarios when the concentrations of ROS and probe are comparable, leading to true second order kinetics; and Figure 2c shows scenarios when, the probe concentration is in excess, leading to pseudo-first order kinetics with t_1/2 being dependent on probe concentration. We hope these figures allow application scientists in various fields to obtain intuitive estimates of the reaction timescale for feasibility assessment. Further, we summarize the literature second-order rate constants in Table 1 for various ROS-sensitive groups.

Figure 2:

Relationship among concentration (M, x-axis), second-order rate constant (M⁻¹s⁻¹, yaxis) and half-life (s, z-axis) of reported ROS-triggered reactions (all three axes are in log scale). (a) ROS concentration is in excess, and half-life was calculated by assuming pseudo-first order kinetics. (b) Equal concentration of ROS and probe; half-life was calculated by second order kinetics. (c) Probe concentration is in excess, and half-life was calculated by assuming pseudofirst order kinetics. (d) ROS triggered reactions discussed in (a-c). Second order rate constants were obtained from literature reports (Table 1); the graphs were generated by using MATLAB R2023b.

Table 1:

ROS-sensitive groups and their second-order rate constants with different ROS.

No	Reaction	Rate constant (M-1s-1)	Conditions
i		2.2	pH 7.4, room temperature (RT)[39]
2		6.2 × 103	pH 7.4, RT[39]
3		1.6 × 106	pH 7.4, RT[39]
4		1.18 × 10−5 (−OMe) 6.67 × 10−6(−Me) 5.14 × 10−6 (−H) 3.19 × 10−6 (−F) 2.22 × 10−6 (−Cl)	N/A[37]
5		107	pH 7.4 (n.(a) [38]
6		3.4 × 107 3.8 × 107	pH 7.4, 22 °C [35] pH 7.4, 37 °C (predicted)[40]
7		8.8 × 10−3 10−2	pH 6.0, 25 °C [36] pH 4.0, 37 °C [41]

Because of the importance of ROS concentration in determining the outcome, the next two sections describe ROS concentration determination using various methods. We should also note that no detection method is completely specific for a particular ROS. One has to look at the reported ROS concentration in the context of the detection method used. Further, most reaction probe-based detection methods actually determine the “cumulative amount” of the relevant ROS over a period of time instead of the true concentration, even though it is often communicated as a concentration concept. This is a very important distinction in concept for designing the appropriate ROS-activation chemistry. Section 3 discusses the various detection methods in detail.

3. Commonly Used ROS Detection Methods and the Issue of Selectivity

Due to the signaling role of ROS,^{[1, 28–29]} there has been extensive work in developing methods for ROS detection. Due to the high reactivity, short half-life, and low steady-state level of ROS, its monitoring in tissue or a biological setting is an especially difficult task.^[42] To highlight these issues, a group of international experts has published a consensus statement with guidelines for measuring ROS concentration and oxidative damage in cells and in-vivo.^[4] This article critically analyzes frequently encountered issues and provides a set of recommendations to avoid common pitfalls. Readers are recommended to read these guidelines. The data and analyses provided below address some of the chemistry questions that are congruent with many of the issues raised in this consensus publication.

The detectable concentration of any ROS is the difference between the total ROS produced and its removal by consumption, diffusion, and degradation. Among the different ROS, the most extensively studied are H₂O₂ and HOCl/OCl⁻, partially because of their relative stability and high concentration.^[43] Despite challenges, several methods have been reported for determining biologically relevant ROS concentrations, including fluorescent probes,^{[43a, 44]} electrochemistry,^[45] and chemical derivatization.^[46]

3.1. ROS sensitive fluorescent probes

Among all the available methods, the most common is the use of fluorescent probes, for their fast response, high sensitivity, spatio-temporal control, and non-invasive nature.^{[43a, 47]} Briefly, a masked fluorophore “selectively” reacts with the ROS of interest, leading to a change in the fluorescent properties through the cleavage of a masking moiety or change to a pair of chromophores capable of Förster Resonance Energy Transfer (FRET). Probes have been reported for monitoring H₂O₂, OCl⁻, ONOO⁻, OH^•, and O₂^•−.^{[43a, 48]} Although these probes are generally stated as selective, often they show cross reactivity.^{[44a, 49]} As one example, boronate-based probes (Figure 3) are often used for the “selective” detection of H₂O₂.^{[44a, 49]} However, later studies have found that the reaction of a boronate is 10⁶ and 10³ times faster with ONOO⁻ and OCl⁻, respectively, than with H₂O₂.^[39] These findings are very significant because screening against peroxynitrite has not been a common practice in developing H₂O₂ probes. Therefore, possible interference from peroxynitrite is an issue that may not have been adequately studied. Such non-specific interactions can be addressed by using proper controls with H₂O₂-consuming enzymes such as catalase.^[4] Further, HEPES and Tris buffer have been found to scavenge OCl⁻.^[50] As such, the choice of buffer is an important factor for accurate determination.

Figure 3:

Boronate-based probes for the detection of H₂O_2.^{[5c, 44a, 49a]}

Another issue in describing selectivity for H₂O₂ probes derives from how selectivity is studied and presented.^[5c] For example, the H₂O₂ selectivity of boronate-based probes is often often-studied by testing against the “physiological concentrations” of the various ROS.^[5c] However, the results may not necessarily mean what the face value represents. For example, 5 μM of probe 1 in PBS was separately incubated with H₂O₂ (20 eq.), O₂^•− (10 eq.), t-BuO. (10 eq.), and OCl⁻ (0.3 eq) for different time intervals within 0–120 min. The fluorescence after 2 h indicates a higher intensity for the reaction with H₂O₂. However, even with their lower concentrations, O₂^•−and OCl⁻ produced ~50% and ~5% of the intensity of H₂O₂. Therefore, if one looks at the true selectivity for these ROS at the same concentration, compound 1 should not be considered H₂O₂ selective. Furthermore, if the incubation was set up as competition experiments, then the probe almost certainly would offer selectivity toward the highly reactive ROS such as O₂^•−and OCl⁻. Nevertheless, this probe was reported as a H₂O₂ probe. This one example represents a somewhat general issue in the field.

In using a boronate-based H₂O₂ probe, another issue is the choice of time point for data sampling. Because a probe-based method leads to the determination of the cumulative amount of a ROS over a defined period of time, not true concentration. Again, the reaction of the boronatebased probe is very fast with ONOO⁻ and OCl⁻. Therefore, a shorter incubation time would “amplify” the relative signal of these very reactive species. On the other hand, a longer reaction time would abrogate the effect of the different reaction rates and “amplify” the signal resulting from less reactive ROS species such as H₂O₂. Another issue is the choice of fluorophore, because highly reactive species such as OCl⁻ may cause oxidative degradation of the fluorophore, leading to fluorescence reduction, which can be misinterpreted as a lack of reaction with OCl⁻.^{[38, 51]} All these mean that in interpreting the information on the concentration of different ROS, there is a need to examine the experimental details in terms of probes used, time point of the study, and how the experiments are set up. Again, we refer to Figure 2 to allow for a quick glance of t_1/2 differences arising from variations in reaction rate constant and ROS concentration.

Another critical concern about probe-based methods is their antioxidant nature: the probe consumes stoichiometric amounts of ROS, decreases its level, and thereby perturbs the redox balance of cellular environment. Further, if an accurate number is needed from a fluorescent probe, one also needs to consider the ease of photobleaching of the fluorophore used^[52] and variable intracellular concentrations of the probe. Often, probe diffusion can also lead to weak signal. In general, ratiometric probes offer improved internal calibration.^[53]

3.2. ROS detection by electrochemical methods

Another widely used ROS detection method is electrochemistry, through the generation of an electric signal by the redox reaction of a ROS sensor on an electrode.^[54] Though still reaction based approach, the main advantages of this method are high specificity, fast response and real time monitoring. Enzymatic and non-enzymatic sensors have been reported for the detection of H₂O₂ in biological fluids. For instance, an enzymatic biosensor using horse radish peroxidase (HRP) was used for the detection of exogenous H₂O₂ in human plasma.^[55] However, the durability of such kind of probes is limited by enzyme stability. Moreover, enzyme inhibition by other biological components can reduce the catalytic turnover and thereby alter the results.^[54] For example, HRP can interact with ascorbate and N-acetyl cysteine (NA(C) and can be inactivated by O₂^•−.^[4] To a great extent, such drawbacks can be overcome by specificity testing with these components, or by using their scavengers such as SOD for O₂^•−. An alternative is the use of a nonenzymatic sensor. For example, a peroxidase-mimicking AuPt/ZIF-8-rGO nano sensor was reported for the detection of H₂O₂ in exogenously added H₂O₂ in serum.^[56] However, neither method was able to detect endogenous H₂O₂ in blood/plasma.

Electrochemical methods have been used for ROS detection in cell culture.^{[32, 57]} For example, silver nanosheets (AgNSs) have been used in HeLa and SH-SY5Y cells, giving H₂O₂ concentration of 0.52 μM and 0.41 μM respectively.^[57] In another example, the intracellular concentration of H₂O₂ in SV-transformed human fibroblast cells was determined as ~5 mM within 5 s.^[32] Similarly, a Pt-nanoelectrode has been reported for the quantification of total reactive species (ROS+ reactive nitrogen species (RNS)) in RAW 264.7 macrophage cells.^[45c] Materials based on metal organic framework (MOF) have been used to mimic superoxide dismutase for the ultra-fast monitoring of short-lived ROS species such as superoxide.^[58] Most of such reported experiments finish within 5–10 s, allowing for a snapshot of ROS concentration. However, in-vivo detection of ROS by an electrochemical method is often challenging due to its invasive nature, and interference by other electroactive species in the biological milieu.^[59] Hence, proper specificity studies and control experiments are needed to avoid potential pitfalls. Another important drawback of electrochemical methods is the technical difficulty in manufacturing the electrodes and the requirement of sophisticated machinery, which may not necessarily be readily available in non-specialty labs.

With all the ROS determination methods available, one needs to consider the specific needs of a project in terms of time scale, the approximate ROS concentration ranges, and the type of desired information (e.g., a snapshot vs. a cumulative amount) in making a selection.

4. ROS Concentrations: in Health and Sickness

To truly assess the utility of ROS-sensitive delivery methods, concentration information is critical. Clearly, achieving a systematic and comprehensive understanding of this issue is not a trivial task. Nevertheless, there have been many studies in determining ROS concentrations in various pathophysiological states. In the subsequent section, we critically summarize concentration data available for ROS concentration from various studies and provide the context appropriate for feasibility considerations. We should also note that such data are not systematic enough for side-by-side comparisons, due to variations in experimental conditions, stimulant used, time points for determination, and detection methods used. As previously discussed, the “concentration” determined using a ROS-reactive probe only represents the “cumulative concentration” over a period of time, accounting for the continuous production of ROS and perturbation of the redox balance by the probe. As a result, meaningful comparisons of the concentration numbers may require examination of detailed experimental procedures and conditions. We hope this review will help spur work in the direction of generating directly comparable data and future studies can be benchmarked against some commonly acceptable standards. Below, we present such data.

4.1. Hydrogen peroxide concentration in animal models

H₂O₂ is the most stable among physiological ROS.^{[1, 60]} Due to the presence of different metabolizing enzymes (catalase, peroxiredoxins, and glutathione peroxidase),^[61] the intracellular concentration of H₂O₂ in healthy cells is generally maintained in nanomolar to low micromolar range.^{[1, 43a, 60, 62]} There have been extensive studies in assessing H₂O₂ concentrations under various scenarios and in different cell types and animal models. Tables 2–6 provide a summary of data available with data in animal models in Table 2 and from various cell culture models in Tables 36. Human plasma H₂O₂ concentrations have been reported to be in the range of 0.8–6 μM (Entry 1, Table 2). In healthy humans, the blood concentration of H₂O₂ above 10 μM causes irritation and cell death.^[63] Hence H₂O₂ concentration in the range of 0.8–6 μM seems reasonable for a healthy individual. However, blood H₂O₂ concentration under pathological conditions may exceed this threshold. For instance, plasma H₂O₂ concentration was found to be 3.36 μM in hypertensive patients (slightly higher than the normal value in the same study, 3 μM; Entry 2), 8.3 μM in hyperglycemic patients (versus 1.9 μM in healthy controls, Entry 3, Table 2), and 15.6 μM in metastatic pancreatic cancer patients (Entry 4, Table 2).

Table 2:

Hydrogen peroxide concentrations from animal models

Entry #. Models (Condition)	Concentration & conditions	Analytical Method (Time)	Comment
1. Human Blood (No stimulant)	0.8–6 μM	Amplex red assay[70] (15 min)	Deproteinized/diluted plasma was used to avoid interference from plasma proteins
2. Human Plasma (No stimulant)	3.00 μM (normal) 3.36 μM (hypertension)	Electrochemistry (Clarks Electrode) [71] (15 min)	H2O2 concentration is calculated as the difference of current response from plasma + catalase subtracted from current response of plasma + azide (inhibitor of catalase and MPO)
3. Rat plasma (No stimulant)	1.9 μM (normal) 8.3 μM (hyperglycemic)	FOX-1 method using xylenol orange oxidation[72] (30 min)	FOX-1 is sensitive to other ROS and oxidants.
4. Blood (metastatic pancreatic cancer patients)	15.6 μM	Boronate-naphthalimide fluorophore[73] (20 min)	Boronate esters can also be cleaved by HOCl & ONOO−
5. Blood (septic shock patients)	558 μM	Phenol red oxidation[74](n.a)	Control experiments with healthy individuals show 8 out of 20 having H2O2 concentrations as high as 200 μM.
6. Aqueous humor (cataract patients)	69 μM, (15–30 min after isolation)	2,6- Dichloroindophenol (DCPIP) oxidation[75] (n.a)	The assay mixture itself can produce H2O2 by auto oxidation of DCPIP. Measurements in cataract patients (n = 17) show H2O2 concentration in a wide range of 10–660 μM.
Aqueous humor (Retinal blastoma/glaucoma patients)	24 μM (15–30 min after isolation)
7. Healthy rat brain (No stimulant)	50 μM (1 h)	Scintillation spectrometer[65] measuring 14CO2 from H2O2promoted decarboxylation of 14C- αketoglutarate. (1 h)	Interference from other oxidative species is addressed by a control experiment with catalase.
Ischemic rat brain	160 μM (1 h)
8. Healthy human urine (No stimulant)	15 μM	Modified FOX-2 method[67] using xylenol orange (30 min)	Modified FOX-2 method, nonspecific oxidation of xylenol orange is addressed by control experiments with catalase
Cancer patient urine (No stimulant)	56.3 μM
9. Breath condensate (No stimulant)	0.03 μM (normal) 0.60 μM (COPD)	Spectrophotometry after treatment of breath condensate with TMB and HRP[76] (20 min)	H2O2 level increased with exacerbation of disease
10. Neutrophils (PMA-25 ng/mL)	12 μM (30 min) (Extracellular)	Electrochemistry (oxidase probemodified electrode)[77]	The identity of the measured species was confirmed with control experiments using catalase and MPO. Peak concentration of H2O2 produced by fMLP and zymosan are less than PMA, but at a more rapid pace than PMA. H2O2 concentration in the presence of RBC is less than the concentration in the absence of RBC (70% and 30% decreased for PMA and zymosan.) One possible contributor to the above effects is the stronger stimulation by zymosan for MPO release than PMA, leading to the consumption of H2O2 for HOCl production.
	615 μM (Cum. 1.5 h) (Extracellular)
Zymosan (1.25 mg/mL)	3.6 μM (8 min) (Extracellular)
	68.2 μM (Cum. 1.5 h) (Extracellular)
fMLP (0.1 μM)	1 μM (3 min) (Extracellular)
	8.4 μM (Cum. 1.5 h) (Extracellular
11. PMN PMA (100–200 ng/mL) PMA (100 ng/mL)	5–15 μM (3–8 min) (Extracellular) 160 μM(+Azide) (1 h) (Extracellular)	Electrochemistry[78] (o-PD membrane coated Pt electrode). (8 min)	Inhibition of H2O2-consuming enzymes by azide increased the H2O2 concentration to 160 μM

o-PD (ortho-phenylenediamine dihydrochloride)

Table 6:

Hydrogen peroxide concentrations from single cell experiments

Entry #, Cell lines (Stimulus)	H2O2 Concentration	Analytical Method & (Time)	Comment
1. RAW 264.7 (PMA-(n.a))	81 μM (Intracellular)	Electrochemistry (Prussian blue coated nanoring electrode)[45a] (snapshot)	Did not study the interference from other electroactive species.
2. HeLa (PMA-800 ng/mL)	5 mM (Intracellular)	Electrochemiluminescence (Au-luminol microelectrode) [107] (snapshot)	Besides ONOO−, interference from other ROS species is a possibility
3. MRC5 VI (Cell membrane rupture)	>5 mM (Intracellular)	Electrochemistry (Pt Carbon fiber microelectrode)[32] (snapshot)	Used a catalase-deficient fibroblast cell. Not able to detect H2O2 in normal cells. The identity of the measured species is confirmed with control experiments with catalase or peroxidase.

Elevated H₂O₂ concentrations in other bodily fluids have also been reported. For instance, elevated level of H₂O₂ is reported to cause oxidative degradation of human lens proteins, leading to cataract.^[64] The average concentration of H₂O₂ in aqueous humor has been reported as 69 μM (n = 17, with one patient peaking at 663 μM) in cataract patients and 27 μM in non-cataract controls (Table 2, Entry 6). However, one needs to note the case of 663 μM H₂O₂, which adds 39 μM to the average of 17 patients and suggests caution in interpreting the data. Along the same line, H₂O₂ concentration in ischemic rats 1 h after reperfusion is reported in the range of 160 μM, compared to the control group of ~50 μM (Table 2, Entry 7).^[65] Similarly, chemiluminescence imaging is used to demonstrate the increased production of H₂O₂ in the nasal mucosa of mice infected with influenza virus (IAV WS/33 (H1N1)).^[66] However, the study did not quantify the H₂O₂ produced.

Moreover, increased H₂O₂ excretion is associated with different diseases. For instance, in a study of 25 patients with cancer, urinary H₂O₂ concentration is reported as 56.3 μM, nearly 4 times higher than that of the healthy controls (15 μM, Table 2, Entry 8).^[67] However, correlating the elevated level of H₂O₂ with tumor needs additional experiments. For instance, drinking coffee can also lead to increased urinary H₂O₂ (103 μM),^[68] suggesting the need for caution in such studies. Similarly, expired breath condensates of patients with severe chronic obstructive pulmonary disease (COPD) have been reported to be 20 times higher than that of the healthy controls (0.6 μM vs 0.03 μM respectively; Table 2, Entry 9). All these data indicate that pathological conditions can elevate H₂O₂ levels in the body very significantly. However, all these entries except for Entry 2 are actually the cumulative amounts of H₂O₂ because of the use of chemical probes. Therefore, these numbers cannot be compared with other true concentration data. Such kind of difference may not be significant in the case of plasma/body fluid when there is no continuous accumulation of H₂O₂.

As part of the immune response, polymorphonuclear leukocytes (PMNs) produce an “oxidative burst” of ROS for the destruction of phagocytosed pathogens.^[69] Electrochemical methods have been used for the real-time detection of H₂O₂ from neutrophils using different stimulants. For example, the peak H₂O₂ concentration reached 12 μM by phorbol myristate acetate (PM(A) stimulation for 30 min, 3.6 μM after 8 min zymosan stimulation, and only 1 μM after 3 min of formyl-methionyl-leucyl-phenylalanine (fMLP) stimulation. However, because of the difference in experimental time points, it is not clear whether these data can be directly compared. Moreover, the cumulative amount of H₂O₂ produced within 2 h was estimated to be 614.9 μM, 68.2 μM, and 8.4 μM after stimulation with PMA, zymosan, and f-MLP respectively (Table 2, Entry 10). PMA and zymosan are generally considered powerful stimulants; the ~4-fold lower response with zymosan was attributed to its higher induction level of MPO (an H₂O₂-degrading enzyme) than PMA. Similarly, H₂O₂ produced by PMA-stimulated PMNs from a group of healthy individuals was reported as ~1–12 μM within 3–8 min. Whereas the addition of NaN₃ increased the H₂O₂ concentration gradually to a peak level of 160 μM within 1 h, presumably due to the inhibition of H₂O₂-consuming enzymes such as catalase and MPO (Table 2, Entry 11). Such a high concentration (160 μM) indeed represents a cumulative amount of H₂O₂ produced by PMA stimulated PMNs in the presence of NaN₃ at a 1-h time point.

4.2. Hydrogen peroxide concentrations from cell culture experiments

Many reports are available on H₂O₂ concentrations in various cell culture models. Table 3 shows H₂O₂ concentrations in normal cells; Table 4 includes data from different types of cancer cells; Table 5 lists comparative studies of H₂O₂ levels in normal and cancer cells; and Table 6 has concentrations from single-cell experiments. Entries reporting cumulative amounts of H₂O₂ are specified with their time points in parentheses.

Table 3:

Hydrogen peroxide concentration from healthy cell culture experiments

Entry #, Cell lines, (Stimulus)	H2O2 Concentration	Analytical Method (Time)	Comments
1. RAW 264.7 (LPS-10 μg/ml)	97 nM (Extracellular)	Electrochemical transducer (NiCo2S4@CoS2 @CC modified electrode)[43d] (Snapshot)	Results are validated with a commercially available fluorimetry assay kit.
2. RAW 264.7 No stimulant (LPS-200 ng/mL)	1.88 μM (Extracellular)	Electrochemistry (Pt microelectrode array)[80] (Snapshot)	Did not study the interference by other electroactive species
	6.0 μM (Extracellular)
3. RAW 264.7 (LPS −10 μg/mL)	0.286 μM (Extracellular)	Electrochemistry (g-CNTs/PB MCs modified electrode)[81] (Snapshot)	-
4. RAW 264.7 PMA (200 ng mL− 1)	0.2 μM (Extracellular)	Electrochemistry (NiMn LDH GCE)[82] (Snapshot)	-
5. RAW 264.7 PMA (0.5 μg/mL)	10 μM (Extracellular)	Colorimetry (P\peroxidasemimicking polypyrrole nanoparticle) [83] (10 min)	Polypyrrole triggered oxidation of TMB is studied by colorimetry. Did not study the other interfering species/ROS; the identity of measured species is not confirmed.
6. RAW 264.7 (Starvation)	34 μM (Intracellular)	4-Amino naphthalimide based ratiometric fluorescent probe[43a] (14 h)	Cells were incubated in protein-free HBSS buffer. Fluorescence intensity with H2O2 is 150-fold higher than other ROS. Pseudo first order rate constant between H2O2 and probe was found to be 1.3 ×10−2 s. −1 Reported concentration is cumulative amount of H2O2 in 14 h
7. Human mononuclear cells No stimulation	0.25 μM (10 min) (Extracellular)	Spectrofluorometry[62] (HRP catalyzed oxidation of p-HP(A) (10 min, 30 min)	-
PMA (0.1 μg/mL)	0.80 μM (10 min) (Extracellular)
Thymocytes No stimulation	35 nM (30 min) (Intracellular)
Thapsigargin (5 μM)	75 nM (30 min) (Intracellular)

Abbreviations: g-CNTs (graphenated carbon nanotubes), PB MCs (Prussian blue microcubes), LDH (layered double hydroxide), p-HPA (p-Hydroxyphenylacetic acid) HBSS (Hanks’ Balanced Salt Solution)

Table 4:

Hydrogen peroxide concentration from cancer cell culture experiments

Entry #, Cell lines (Stimulus)	H2O2 Concentration	Analytical Method (Time)	Comments
1. HeLa (No stimulation)	2.29–2.43 μM (Intracellular)	Circular dichroism (CD)/Upconversion luminescence (UCL) responsive nanoparticle[84] (4 h)	H2O2 detected with two different analytical methods (CD, UCL) provided similar results.
2. PC-3 No stimulation	2.36 μM (Extracellular)	Electrochemical Biosensor[85] (HRP-AuNP-PEG Au electrode) (2.5 h)	-
3. A549 (No stimulation)	12 μM (Extracellular)	Surface enhanced Raman spectroscopy[86] (4MPBE modified AuNP) (1 h)	Arylboronate based AuNP nano sensor, specificity study of the probe 0 showed high response to ONOO-, (Response from 5 nM ONOO- matches with that of 10 μM H2O2).
4. SK-Ov 3, SK-Mel-28, SK-Mel 30 (No stimulation)	(Up to 0.5 nmol/104 cells/h) (Extracellular H2O2 in presence of 1mM azide)	Scopoletin fluorescence assay[87] (2 h, 4 h)	H2O2 concentration by SK-Mel-30 tumor cells in the absence of azide was found to be 0.25 nmol/104 cells/h. The rate of H2O2 production at 2 h was higher for PMNs than stimulated cancer cells, but cumulative H2O2 after 4 h was higher for cancer cells.
PMN (PMA-10 ng/mL)	1.4 nmol/104 cells/h (Extracellular H2O2 in presence of 1mM azide)
5. SW620 (PMA −1μg/mL)	0.7–1.45 μM (Intracellular)	Surface enhanced Raman spectroscopy(SERS)[88] (3-MPBE modified AuNPs) (1 h)	Boronic acid oxidation with H2O2 was monitored by SERS, but did not study the possible interference by ONOO−
6. SKOV3 MCF-7 A431 (PMA- 10 μL, 1μg/mL)	1.98 μM (Extracellular) 2.21 μM (Extracellular) 2.34 μM (Extracellular)	Electrochemistry (PDDA)-AuPtAg/RGO-GCE) [89] (Snapshot)	In the original article, concentration was reported as intracellular, however, the experimental protocol is consistent with extracellular measurement.
7. HeLa (PMA-0.3 μL, 1 μg/ mL)	0.8 μM (Extracellular)	Electrochemistry (RGO-PMS@AuNPs GCE) [68] (Snapshot)	No current change with normal cell (HEK 293)
			The identity of H202 was confirmed by catalase experiments
8. HeLa (PMA-100 μL, 5 mg/L)	5.8 μM (Extracellular)	Electrochemistry (Ce02/HCC0/MWCNTs/GCE)[90] (Snapshot)
9. PC 12- (PMA (n.a))	6.32 μM (Extracellular)	Electrochemistry (Co-N-C-800 SAN-RDE)[91] (Snapshot)
10. HeLa (PMA-5 (μg /mL)	610 μM (Extracellular)	Photoelectrochemistry (CuO Nanoflower coated FTO electrode)[92] (Snapshot)	H2O2 concentration increased with an increase in cell density. The identity of H2O2 was confirmed using catalase experiments. ROS generation due to phototoxicity is a concern.
11. HeLa (PMA-2 μM)	1.99 mM (Extracellular)	Colorimetry[93] (Fe-N-C SAN) (5 min)	(Fe-N-C single-atom nanozyme promotes the conversion of OH• from H2O2 and causes oxidation of TMB and causes color change), Interference by native OH• is a possibility.
12. MDA-MB-231 (PMA-2 μM)	535 μM (Extracellular)	Colorimetry (Fe-N-C SAN)[94] (5 min)	(Fe-N-C single-atom nanozyme promotes the conversion of OH• from H202 and causes oxidation of TMB and color change). Interference by native OH• is a possibility
13. MCF-7 (Ascorbic Acid-400 (μM)	1.04 μM (Extracellular)	Electrochemistry (Pt/cMIL-68/MoS2/GCE)[95] (Snapshot)	Ascorbic acid itself can produce H2O2 in pure culture media such as MEM, DMEM, RPMI 1640 H2O2 was measured in PBS, and control experiment by PBS (without cell) and ascorbic acid could not significantly change the current signal
14. PC12 (Ascorbic Acid-1 (μM)	2.98 μM (Extracellular)	Electrochemistry (Pt-Au-bimetallic nano porous electrode)[96] (Snapshot)	Ascorbic acid itself can produce H202 in pure culture media such as MEM, DMEM, RPMI 1640 Control experiment by PBS (without cell) and ascorbic acid could not significantly change the current signal
15. MCF-7 (fMLP-1.5 μM)	33.1 μM (Extracellular)	Electrochemistry (AuPd@FexOy NPs/GCE) [97] (Snapshot)	-
16. Rat Hepatocyte (Ethanol-100 mM)	1.16 μM (Extracellular)	Electrochemistry (PEG-HRP Au electrode array)[98] (Snapshot)	Prior treatment with antioxidant (NAC), and ALDH inhibitor (4-MP) reduced the current signal. Did not study the interference by other electroactive species
17. SH-SY5Y (6-QHDA-50 μM)	51.2 μM	Electrochemistry (-SPE, CB/PB-SPE)[99] (Snapshot)

Abbreviations: 4-MPBE (4-mercaptophenylboronic acid pinacol ester), )(3-MPBE (3-mercaptophenylboronic acid pinacol ester), PDDA-(poly (diallyldimethylammonium chloride) RGO (reduced graphene oxide) PMS (periodic mesoporous silic(a) GCE (glassy carbon electrode), HCCO (Holey CuCo₂O₄), MWCNTs (multi-walled carbon nanotubes), SAN (single-atom nanozyme), RDE (rotation disk electrode), FTO (Fluorinedoped Tin Oxide), SPE (screen printed electrode), PEG (polyethylene glycol), 4-MP (4-methyl pyrazole), CB/PB (Prussian blue on carbon black), 6-OHDA (6-Hydroxydopamine)

Table 5:

Hydrogen peroxide concentrations from healthy vs disease cell culture experiments

Entry #, Cell lines (Stimulus)	H2O2 Concentration	Analytical Method & Time	Comments
1. HaCaT (Ascorbic Acid-(n.a))	0.23 μM (Extracellular)	Electrochemistry[82] (Ni Mn LDH/GCE) (Snapshot)	H2O2 concentration in the cell was determined in PBS. Control experiment by PBS (without cell) and ascorbic acid did not significantly change the current signal
HeLa (Ascorbic Acid-(n.a))	0.64 μM
2. HaCaT (Ascorbic Acid-10 μM)	0.18μM (Extracellular)	Electrochemistry (Ni-Co LDHmodified GCE) [103] (Snapshot)	H2O2 concentration in the cell was determined in PBS. Control experiment by PBS (without cell) and ascorbic acid did not significantly change the current signal
HeLa (Ascorbic Acid-10 μM)	2.04 μM (Extracellular)
3. Adrenal medulla cells (Ascorbic Acid-1 μM)	0.079 μM (Extracellular)	Electrochemistry (Acidified-MnO2/GCE) [104] (Snapshot)	H2O2 concentration in the cell was determined in PBS. Needed further experiments to address the issue of interference from media
PC-12 (Ascorbic Acid −1 μM)	0.79 μM (Extracellular)
4. HeLa (PMA-400 ng/mL)	1.08 nmol/104 Cells /h	Raman scattering (Au NPs/4-C(A) [105] Aryl boronate oxidation monitored by SERS spectra.	Specificity to H2O2 is confirmed by screening with OCl−, ONOO−
HL-7702 (PMA-400 ng/mL)	0.85 nmol/104 Cells /h (Intracellular)

LDH (layered double hydroxide), 4-carboxyphenylboronic acid (4-CA)

Macrophages produce an oxidative burst and release an excess of ROS/RNS for digesting phagocyted pathogens.^[79] Macrophage can be stimulated by using appropriate immunogenic agents such as lipopolysaccharide (LPS) and PMA. For instance, the extracellular concentration of H₂O₂ in LPS-stimulated macrophages (RAW 264.7 cells) was found to be about 97 nM (Table 3, Entry 1). In a different report, extracellular H₂O₂ concentration was found to be ~1.88 μM in non-stimulated RAW 264.7 and 6.0 μM upon stimulation with LPS (Table 3, Entry 2). As one indication of experimental variations in such studies, another study reported 0.286 μM H₂O₂ from LPS stimulated RAW 264.7 cells (Table 3, Entry 3) and 0.2–10 μM from treatment with a more potent stimulant PMA (Table 3, Entry 4–5). Moreover, the intracellular H₂O₂ concentration in nutrient starved culture media (protein free HBSS buffer) was reported as 34 μM using a probebased method, leading to the report of a cumulative amount instead of a concentration (Table 3, Entry 6). In another example, extracellular H₂O₂ in human mononuclear cells stimulated with PMA has been reported to be 0.80 μM, 3 times higher than non-stimulated controls (Entry 7). Similarly, an endoplasmic stress-inducing agent thapsigargin was reported to increase intracellular H₂O₂ production in thymocytes to 75 nM from 35 nM in non-treated thymocytes (Table 3, Entry 7). It should be noted that comparing the various extracellular concentrations is hard because of variations in medium volume/cell density and other experimental conditions.

Table 4 shows H₂O₂ concentrations in cancer cells being in the range of 2.29 to 12 μM (Entries 1–5). For instance, the intracellular H₂O₂ concentrations in HeLa cells were found to be 2.29–2.43 μM (Table 4, Entry 1) using a ROS-responsive nanoprobe after 4-h incubation. Hence, the value reported is indeed the cumulative “concentration.” Other studies measured extracellular concentrations of H₂O₂, which presumably is dependent on medium volume and cell density. Specifically, the extracellular H₂O₂ concentration in PC-3 (Table 4, Entry 2) and A549 (Table 4, Entry 3) cells was found to be 2.36 μM and 12 μM, respectively. In another way of assessment, H₂O₂ production rates in tumor cells were compared with human PMNs. Briefly, 10⁴ cells of SKOv-3, SK-Mel-30, and SK-Mel-28 were found to produce up to 0.5 nmol/h of H₂O₂, in comparison to the rate of 1.4 nmol/h by PMA-stimulated PMNs (Table 4, Entry 4).

Mechanical or chemical stimulations of cells are known to lead to inflammatory responses, resulting in the overproduction of ROS (Table 4, Entries 5–17). For example, upon stimulation with PMA, cumulative intracellular H₂O₂ “concentration” at 1-h in SW620 cells was estimated to be 0.7–1.45 μM (Table 4, Entry 5). In another report, H₂O₂ concentrations in different cells were measured as 1.98–2.34 μM (Table 4, Entry 6). Moreover, the extracellular concentrations of PMAstimulated cells were found to be 0.8–6.32 μM (Table 4, Entries 7–9). H₂O₂ concentrations in entries 10–11 seem to be very high in HeLa cells (610 μM and 1.99 mM, respectively). In contrast, H₂O₂ in HeLa cells with the same stimulant PMA (Entry 8) but a different electrochemical method was determined as 5.8 μM H₂O₂. Entries 11 and 12 show results by using a single atomic site nanozyme, which mimics peroxidase and catalyzes the conversion of H₂O₂ to OH^• for subsequent detection using 3,3′,5,5′-tetramethylbenzidine (TM(B) as a reporter. As such, Entry 12 shows a high 535 μM of extracellular H₂O₂ concentration in MDA-MB-231 cell culture. However, control experiments using HeLa (Entry 11)/MDA-MB-231 cells (Entry 12) without any stimulation also led to a high background. Thus, these numbers probably need further validation. Ascorbic acid has also been widely used as a stimulant for H₂O₂ production in cell culture (Entries 13, 14, Table 4). However, ascorbate is known as an antioxidant and the exact mechanism for ascorbate to promote H₂O₂ production is yet to be studied. Widely stated reasons include the prooxidant effects of metalmediated (e.g., Fe³⁺ and Cu²⁺) reactions of high doses of ascorbate.^[100] Moreover, there are multiple reports of H₂O₂ production by ascorbic acid in cell-culture media without cells, and the quantity of H₂O₂ is dependent on the culture media (Section 4.3).^[100–101] Hence, additional experiments are needed to examine the detailed mechanism(s) responsible for “ascorbate induced” H₂O₂ generation in cell culture. Another commonly used stimulant for ROS production is fMLP. For instance, the concentration of H₂O₂ produced by fMLP-stimulated MCF-7 cells was reported as 33.1 μM (Table 4, Entry15).

Alcohol induced liver injury is associated with elevated levels of ROS and RNS via increased CYP2E1.^[102] H₂O₂ concentration in ethanol-treated (100 mM) hepatocytes was found to be 5 times higher (1.16 μM) than the controls (Table 4, Entry 16), though the identity of the monitored species needs to be further confirmed. Similarly, nigral dopaminergic neurons in Parkinson’s disease (PD) is associated with elevated levels of ROS. For example, H₂O₂ concentration in Parkinson model cells (SHSY5Y cells stimulated with 6-hydroxy dopamine) was reported as 51.2 μM (Table 4, Entry 17). Table 4 highlights the wide variations of H₂O₂ concentrations (up to 1.99 mM) under different pathological conditions. However, some of the differences might simply be due to experimental conditions and methods used and may not be directly comparable. We urge caution in comparing these numbers.

Table 5 lists results from comparative studies of H₂O₂ levels in cancer vs normal cells. Because the same experimental protocol was used in each study for determining the H₂O₂ concentrations in both healthy and cancer cells, the results offer the chance for valid direct comparisons. For instance, H₂O₂ concentration was found to be 0.23 μM in ascorbic acid-treated normal cells (HaCaT) and 0.64 μM in HeLa cells (Table 5, Entry 1). However, in another study, H₂O₂ concentration was found to be 0.18 μM and 2.04 μM in ascorbate treated HaCaT and HeLa cells respectively (Table 5, Entry 2). It should be noted that there are variations of H₂O₂ concentrations for the same cell type under the same conditions. Often, one can trace back to different detection methods used, though there is no concrete evidence to make such an attribution. Moreover, H₂O₂ concentrations in tumor cells were found to be 10μ higher than the corresponding healthy cells. Specifically, H₂O₂ concentration was found to be 0.079 μM in ascorbate-treated healthy adrenal medulla cells and 0.79 μM in adrenal medulla cancer cells (PC12) (Table 5, Entry 3). As discussed in the previous section, ascorbic acid itself generates H₂O₂ in culture media, thus complicating results interpretation. Entries 1 and 2 performed a control experiment in ascorbic acid-treated culture media (PBS) without cells, but Entry 3 seems to be missing such control studies. In a different study, the rate of H₂O₂ production by 10⁴ Hela cells was determined to be 1.08 nmol/h in comparison to 0.85 nmol/h in healthy liver cells (HL-7702, Table 5, Entry 4). All these provide very useful information in terms of the magnitude of differences between cancer and normal cells.

There are reports of single-cell experiments estimating H₂O₂ concentrations (Table 6). For instance, the intracellular H₂O₂ concentration in isolated PMA-stimulated single RAW 264.7 cells was estimated to be 81 μM (Table 6, Entry 1). Herein, the cell was lysed using 3 mM NaOH, and H₂O₂ in the supernatant was determined by a Prussian blue coated electrode. Unlike the cases discussed before, herein H₂O₂ in the cell lysate was interpreted as intracellular concentration. The same research group also reported the intracellular H₂O₂ concentration in PMA-stimulated single HeLa cell as 5 mM (Table 6, Entry 2). Here, an electro-chemiluminescent method by the electrochemical oxidation of luminol coated electrode was used for the visualization and quantification of H₂O₂. The detection limit for the electrode was found to be > 1 mM. Electrode insertion into the cell is said not to cause any membrane deformation through the formation of a tight seal around the electrode. Further, cell membrane integrity was confirmed by propidium iodide (PI) staining. Similarly, the intracellular concentration of H₂O₂ in MRC5 VI (SV transformed human fibroblasts) cells was found to be higher than 5 mM (Table 6, Entry 3). One possible reason for such a high concentration could be due to depleted catalase activity in SV₄₀transformed cells, an enzyme responsible for the dismutation of H₂O₂ to water and molecular oxygen.^[106]

The concentration of total ROS +RNS produced by single macrophage cells have been quantified by using a platinum coated nanoelectrode or microelectrode.^[45c] The response inside the cell was smaller with a shorter duration (t_1/2 = 0.5 s) whereas the extracellular current was higher with a longer duration (t_1/2 = 5.8 s). The current response corresponds to sub-millimolar concentrations of ROS/RNS produced inside the cells, which rapidly diffuse out of the cell to cause extracellular accumulation. However, this concentration is a combination of different reactive species, including H₂O₂, NO, ONOO ⁻, and NO₂⁻.

In short, H₂O₂ concentrations from single-cell experiments have been reported to be ~1000 times higher than that in human plasma.^[31] One possible reason for such high concentrations might be attributed to the cellular environment and the experimental conditions used. Specifically, isolation of single cells from its culture probably imparts stress to the cell, leading to elevated production of ROS. Additionally, penetration of micro/nanoelectrodes into the cell membrane (Entries 2–3) may contribute to additional stress conditions along with PMA.

4.3. What is the physiologically relevant concentration of H₂O₂?

Even after extensive work, there is not a single number reported for “physiologically relevant” H₂O₂ concentration, partially due to the naturally variable and dynamic nature of H₂O₂ concentration.^[78] Major factors affecting H₂O₂ concentrations include the dismutation of superoxide, production in organelles like peroxisomes, mitochondria, and continuous consumption by different enzymes like peroxidases and catalases.^{[78, 108]} Hence, reported concentrations of H₂O₂ are largely varied. Even the maximally attainable physiological concentration of H₂O₂ is still a point of debate. Nevertheless, there are some generally accepted ranges, which are presented graphically in Figure 4.

Figure 4:

Summary of literature reported concentrations of H₂O₂ in different environments. *The dotted line represents an outlier with a surprisingly high concentration.

In cell culture experiments, many factors contribute to variations in H₂O₂. It is well known that cells after multiple passages could behave very differently. Further, antioxidants such as various thiol species (e.g.; cysteine, N-acetyl cysteine, etc.), polyphenols, and ascorbic acid themself have been reported to generate significant amounts of H₂O₂ in popular cell culture media (MEM, DMEM, RPMI).^{[101a, 109]} For instance, addition of 1 mM cysteine thiol derivatives in DMEM could produce ~15– 20 μM H₂O₂ within 2 h. The effect is more obvious with ascorbic acid and gallic acid, producing 102 μM and 278 μM H₂O₂ within 2 h, respectively.^[109a] Moreover, the quantity of H₂O₂ produced from ascorbate and thiol species varies with the culture media. For instance, 1 mM of ascorbic acid generates 161 μM of H₂O₂ in DMEM medium, but only 83 μM in RPMI 1640 medium.^[101a] However, ascorbic acid in the blood did not produce detectable levels of H₂O₂, presumably due to the presence of a high level of catalase and glutathione peroxidase in red blood cells.^[110] Moreover, generation of H₂O₂ includes the formation of an ascorbate radical intermediate and depends on the presence of 0.5–10% serum as well as the concentration and incubation time with ascorbate.^[110] One possible reason for H₂O₂ production in the oxidation of ascorbates and thiols is metal catalysis in the culture media.^[111] For instance, DMEM is supplied with added iron (III) nitrate, Fe(NO₃)_3. In other culture media, the source of metal ions is attributed to the contamination of different components of the media. For example, FBS supplemented culture media could be contaminated with free iron released from transferrin upon storage.^{[111a, 111c]} Beyond hypothesis, further experiments are needed to confirm the mechanism of oxidation. Cell culture experiments with thiol species, polyphenols, and ascorbic acid need to consider artifacts from the culture media itself. Moreover, ascorbate is a usual interfering component for peroxidasebased electrodes.^[110]. Hence, proper control experiments or screening with ascorbate should be done to avoid such pitfalls.

4.4. HOCl concentrations in animal models

Besides H₂O₂, hypochlorite is also produced under normal and pathological conditions and upon stimulation (Tables, 7, 8). For example, cumulative extracellular OCl⁻ concentration after 1 h from isolated human neutrophils was determined to be 39 μM (Table 7, Entry 1).^[46b] Moreover, elevated OCl⁻ production upon stimulation has also been studied. For example, at pH 7.4, the OCl⁻ concentration in PMA-treated neutrophils was found to be about 76 μM and 117 μM or 167 μM upon co-incubation of PMA with SOD or MPO, respectively. Incubation with zymosan significantly increased the hypochlorite level (316 μM); co-treatment with SOD increased it to 398 μM. Co-stimulation with PMA+SOD+MPO led to OCl⁻ concentration peaking at 381 μM. These results are aligned with the fact that zymosan is a much more powerful stimulant than PMA for MPO secretion.^[112]

Table 7:

Hypochlorite concentrations from animal models

Entry #, Cell lines /Models (Stimulus)	Concentration	Analytical Method (Time)	Comment
1. Neutrophils No stimulation PMA (80 nM) PMA+ SOD PMA + MPO Zymosan Zymosan + SOD	39 μM (Extracellular) 76 μM (Extracellular) 117 μM (Extracellular) 167 μM (Extracellular) 316 μM (Extracellular) 398 μM (Extracellular)	UV-Vis Spectroscopy[46b] (HOCl trapped with taurine, the product was quantified by oxidation of TN(B) (1 h)	It is the cumulative concentration of HOCl within 1 h Other ROS and oxidants can interfere the assay process for the quantification of taurine chloramine
2. Neutrophils (PMA- 100 ng/mL)	51.5 μM (Extracellular)	UV-Vis Spectroscopy[46a] Similar to Entry 1(But oxidation of TNB is catalyzed by iodine) (30 min)	-
3. Mouse xenograft No stimulation APAP (20 μM) APAP (40 μM)	51 μM 138.3 μM 421.7 μM	Mitochondria targeted ratiometric probe[43b] (Rhodamine as fluorophore and indole as mitochondrion-targeting moiety) (1 h)	The mechanism of fluorophore release needs to be studied
4. Mouse liver injury No stimulation Alcohol	45 μM 100 μM	Aggregation Induced Emission from a ratiometric fluorescent nanoprobe[47b] (Probe composed of phenothiazine donor and quinoline malononitrile acceptor) (na)	In a mouse liver injury model, time point for probe incubation is not specified
5. Alveolar fluid of cystic fibrosis patients (No stimulation)	2.65–8.15 mM	Calculation based on previous data.[114] (n.a)	The concentration is not experimental, estimated based on previous report and seems very high

Table 8:

Hypochlorite concentrations from cell culture models

Entry #, Cell lines (Stimulus)	Intracellular HOCl Concentration	Analytical Method (Time)	Comment
1. LO2 No stimulation APAP (20 mM) APAP (40 mM) APAP+NAC	33.2 μM 96.3 μM 285.4 μM 38.3 μM	Mitochondrion-targeted ratiometric Probe[43b] (Rhodamine as fluorophore and indole as mitochondria targeting moiety) (30 min)	The mechanism of fluorophore release needs to be confirmed. The method yields the “cumulative concentration” of HOCl within 30 min.
293T No stimulation APAP (20 mM) APAP (40 mM) APAP (40 mM)+NAC (40 mM)	22.6 μM 85.1 μM 288.3 μM 15.9 μM
2. MCF-7 (No stimulation) HT-29 (No stimulation) HepG2 (No stimulation) LO2 (No stimulation)	50 μM 70 μM 85 μM 60 μM	Aggregation induced emission from a ratiometric fluorescent nanoprobe.[47b] (Probe composed of phenothiazine donor and quinoline malononitrile acceptor) (30 min)	The method yields the “cumulative concentration” of HOCl within 30 min.

It should be noted that OCl⁻ concentration in Entry 1 has been determined by a trapping method with the addition of 10 mM taurine for conversion to taurine-N-chloramine, which can be quantified spectrophotometrically by the oxidation of 5-thio-2-nitrobenzoic acid (TN(B) at 1 h. As a result, all the values represent a “cumulative concentration” of OCl⁻ within 1 h. Further, the oxidation step is not specific and TNB can be oxidized by ONOO⁻ and H₂O₂.^[113] Later, such nonspecific oxidation issue was addressed by the use of iodine catalyzed oxidation of TNB and subsequent measurement of absorbance. Using this modified method, the cumulative concentration of HOCl in neutrophils was determined to be 51.5 μM after 30 min PMA stimulation (Table 7; Entry 2).^[46a]

Elevated levels of hypochlorite upon injury (inflammation) have also been reported in animal models. For example, OCl⁻ concentration in a drug-induced liver injury model was estimated by using a rhodamine-based mitochondria-targeted probe; NRH-O (Figure 5).^[43b] The probe is made of a rhodamine analog as the fluorophore and a positively charged benzoxazole group as a mitochondria-targeting moiety. The fluorophore and targeting group are linked via a conjugated double bond, which undergoes rapid oxidation to epoxide and subsequent hydrolysis to release the fluorescent rhodamine. The cumulative HOCl concentration after 1 h of incubation was found to be 51 μM in control mice and 138.3–421.7 μM in mice treated with acetaminophen (20–40 mM) (Table 7, Entry 3). Moreover, control experiments with ROS scavenger NAC inhibited the fluorescence induced by OCl⁻ in vivo, supporting the role of OCl⁻ in fluorescence production.

Figure 5:

Mitochondrion-targeted ratiometric fluorescent probe for the detection of HOCl.^[43b]

A ratiometric fluorescent probe TCPQ based on aggregation induced emission (AIE) has been used for the quantification of HOCl in alcohol-induced liver injury (Figure 6). Briefly, mice injected with alcohol showed dark green fluorescence, corresponding to 100 μM of intracellular OCl⁻, which is 2.2-fold that of the control group (45 μM) (Table 7, Entry 4). Further, injection of silymarin (a drug for liver injury) was found to reverse the OCl⁻ concentration to the normal level. Altogether, these experiments indicate the elevated amount of hypochlorite in liver injury.

Figure 6:

A ratiometric fluorescent probe (TCPQ) for the quantification of OCl⁻. The figure is reprinted from ref^[47b] with copyright permission

In another report, PMNs stimulated by PMA or zymosan produced ~100 nmol OCl⁻ per 2×10⁶ PMNs.^[115] Later, Guo et al., calculated the concentration of OCl⁻ in the alveolar fluid of cystic fibrosis patients to be in the range of 2.65–8.15 mM (Table 7, Entry 5).^[114b] These numbers seem very high and can be very damaging to tissues if proven true. Further, it should be noted that the number in Entry 5 is estimated based on literature reports^[115] and the actual concentration must be verified experimentally.

4.5. HOCl concentration in cell culture models

There have also been studies examining the concentration of hypochlorite in cell culture. For example, using the NRH-O probe discussed in the previous section (Figure 5),^[43b] the OCl⁻ level was estimated in hepatocytes (LO2) and human embryonic kidney (293T) cells after acetaminophen (APAP)-induced injury. Cumulative OCl⁻ concentration within 30 min in the control group of LO2 and 293T cells was found to be 33.2 and 22.3 μM, respectively. In contrast, APAP (20–40 mM) led to cumulative OCl⁻ levels of 96.3–285.4 μM in LO2 cells and 85.1–288.3 μM in 293T cells within 30 min (Table 8, Entry 1).

Similarly, TCPQ (Figure 6) was used for the intracellular imaging of HOCl in non-stimulated cells such as HT-29, MCF-7, HepG2, and LO2,^[47b] leading to the determination of cumulated intracellular concentration of OCl⁻ being 70 μM, 50 μM, 85 μM, and 60 μM, respectively, after 30 min of incubation (Table 8, Entry 2).

5. Reactions Commonly Used as ROS-sensitive Triggers: The Need for Considering Reaction Kinetics in Deciding the Reaction to Use

The principle of ROS triggered payload delivery is to exploit the elevated level of ROS in the targeted site. Since different ROS have their unique chemical reactivity, several ROS-sensitive reactions have been reported for various applications.

5.1. Aryl boronic acid/boronate oxidation

Aryl boronic acid/boronate is a widely used ROS trigger for drug delivery and imaging applications (Figure 7).^{[2b, 116]} Boronates are prone to oxidation by H₂O₂, leading to the release of an alcohol functionalized payload.^[2b] Such chemistry alone or in-combination with a quinone methide-based linker have been widely used in H₂O₂-triggered delivery of payloads.^{[2a, 117]} Nevertheless, this approach has some key issues to be addressed including protodeboronation and non-specific reaction with other ROS species.^[39] Prior to 2009, boronic acids were considered as a specific trigger for H₂O₂. Later, studies by Sikora et al revealed rapid reactions of boronates with peroxynitrite (k = 1.6 × 10⁶ M⁻¹s⁻¹) and HOCl (k = 6.2 × 10³ M⁻¹s⁻¹) in comparison to H₂O₂ (k = 2.2 M⁻¹s⁻¹).^[39] Therefore, the reaction of boronates is 10⁶ and 3,000 times faster with ONOO⁻ and HOCl, respectively, than with H₂O₂. As discussed in a previous section, the highest reported cellular concentration of OCl⁻ and H₂O₂ is 400 μM, and 5 mM, respectively (Tables 2–8). Hence, the t_1/2 for each probe-ROS pair can be calculated. If we consider the concentration of ONOO⁻ as 70 nM^[118] and probe as 10 μM, complete reaction would only consume <1% of the probe. However, the scenario is different in the case of OCl⁻ and H₂O₂. If we consider the physiological concentration of OCl⁻ and H₂O₂, the half-life for 10 μM boronate probes can be calculated as 0.28 s for HOCl (400 μM) and 63 s for H₂O₂ (5 mM), with respect to the probe. These numbers are critical because the boronate molecule is expected to react first with hypochlorite present at a meaningful concentration and release the payload, regardless of H₂O₂ concentration. Considering the wide range of applications of aryl boronic esters in reported H₂O₂ probes,^{[49b, 119]} the results have to be interpreted in the context of reaction kinetics involving various species. As already discussed in Section 3.1, the choice of time point in sampling can skew the “selectivity” in different directions because of the differing kinetics of the boronate reaction with various ROS. Incidentally, there have been efforts to improve the oxidation kinetics with H₂O₂ by using a borinic acid derivative^[120] with a 10,000-fold increase in reaction rate (to k = 1.9 × 10⁴ M⁻¹s⁻¹) with H₂O₂ in comparison to the boronic acid derivative (k = 1.8 M⁻¹s⁻¹). However, the reactivity towards two important ROS, ONOO⁻ and HOCl, were not examined in this study. It is anticipated that an increase in electrophilicity of the boron atom can increase the reactivity towards ONOO⁻ and HOCl as well.

Figure 7:

Mechanism of payload release from prodrugs via different linker chemistry to transduce the signal of boronate oxidation. Reprinted with permission from ref.^[2b] Copyright ©2021 American Chemical Society.

5.2. Oxidation of chalcogens

Another approach to ROS-sensitive delivery is to exploit the facile oxidation of chalcogens (Group-16 elements) like sulfur, selenium, and tellurium.^{[33a, 42, 121]} Several types of nanoparticles have been reported to take advantage of phase transition as a trigger for material disassembly and payload release upon oxidation (Figure 8). For example, the thioether undergoes oxidation by ROS to sulfoxide or sulfone, leading to release the payload from pre-assembled materials due disassembly because of increased polarity (phase transition).^[121–122] In general, thioether oxidation is slow and kinetic studies need to be performed at mid-millimolar concentrations of H₂O₂ for meaningful results within hours.^[123] One should be cautious in extrapolating results from experiments using such high concentrations of H₂O₂ (>10 mM), which is beyond physiologically relevant concentrations.¹⁰⁴ Oxidation kinetics can be improved by using more electropositive selenides or tellurides as the ROS triggers. For instance, telluride-polymer can be fully oxidized by 100 μM H₂O₂ within 1 h.^[124]

Figure 8:

ROS-triggered phase transition of chalcogenide polymer. Reprinted with permission from ref.^[123b] Copyright ©2019 American Chemical Society.

ROS-mediated oxidation has also been used to generate a good leaving group for the purpose of payload release. For example, phenyl selenide oxidation by ROS to selenoxide and its subsequent elimination was used for the delivery of carbon monoxide (CO) (Figure 9).^[5b] Briefly, norbornadiene-7-one (2) is an unstable scaffold under near physiological condition and undergoes spontaneous cheletropic reaction to release carbon monoxide. Hence, the development of prodrugs for norbornadiene-7-one provides an opportunity for the delivery of CO, which has shown therapeutic effects in animal models of inflammation, cancer, and organ injury.^[6c] The structural analog (4) with a single bond between C5 and C6 is very stable and can be isolated without CO release. Thus, structure (5) tethered with appropriate leaving group on the C5 or C6 position serves as a pro-prodrug for CO delivery under near physiological conditions.^{[6b, 125]} Specifically, tethering a ROS sensitive group such as phenylselenyl ether at C6 position leads to ROS triggered delivery of CO (Figure 9).

Figure 9:

ROS triggered oxidation of selenides to selenoxide and subsequent elimination as an approach to CO prodrugs.^[5b]

Without getting into the details, a similar approach has been used in oxidizing a thioether for the preparation of CO prodrugs (Figure 10).^[5a]

Figure 10:

ROS triggered oxidations of a thioether to sulfoxide as an approach for CO delivery.^[5a]

Reported selectivity studies of thioether polymers show rapid oxidation by hypochlorite, but not by H₂O₂. As a result, thioether oxidation has also been used in designing a turn-on fluorescent probe for OCl⁻.^{[44c, 47a, 126]} For example, fluorescence from a BODIPY probe is suppressed by the lone pair electrons of S via photoinduced electron transfer (PET) (Figure 11). ROS oxidation of sulfur heteroatom results in suppression of the PET effect and thus fluorescence turn-on.^{[44c, 47a, 126]} Importantly, the oxidation kinetics depends on the electronic nature of the thioether. Generally, aliphatic chalcogenides are more reactive and sensitive to most of the ROS than aryl chalcogenides. For instance, the oxidation of thiomorpholine with HOCl has a high rate-constant of 10⁷ M⁻¹s⁻¹.^[38] As such, 10 μM of the probe is estimated to have a t_1/2 of 173 μs with 400 μM of HOCl.

Figure 11:

An example of chalcogen oxidation based fluorescent probe.

5.3. Thioketals

Thioketals are an important carbonyl protecting group used in synthetic applications^[127] and have been reported as an ROS-sensitive group.^{[2c, 15a, 128]} Therefore, thioketals have been explored in ROS-sensitive drug delivery (Figure 12). A Pubmed search in April 2024 generated 256 publications using such chemistry. The proposed mechanism is through the production of two thiol fragments and acetone after thioketal oxidation by ROS.^{[15a, 129]} However, this release of a thiol group actually involves a more sophisticated redox reaction with ROS than proposed (Figure 12). Further, thiols in the product are known to undergo rapid oxidation to disulfide in the presence of peroxide.

Figure 12:

A proposed mechanism of ROS triggered thioketal cleavage and release of camptothecin.^[15a]

In 2020, Liu et al. elucidated the mechanism as shown in Figure 13.^[37] The first step involves the oxidation of one of the thioether to sulfoxide, facilitating the cleavage of the first C-S bond to give reactive sulfenic acid and thiocarbenium intermediate. Subsequent hydrolysis of thiocarbenium intermediate produces the corresponding ketone/aldehyde together with a free thiol, which could undergo cyclization to release the payload (Figure 13) or further oxidation to be trapped as a disulfide product.

Figure 13:

A proposed mechanism of thioketal oxidative degradation.^[37]

Another issue with the thioketal linker is its sluggish oxidation kinetics. For instance, the very first publication on thioketal chemistry reported slow degradation kinetics of a thioketal in presence of H₂O₂ (t_1/2 ~ 48 h). The kinetic study was performed using NMR with an initial concentration of ~17 mM thioketal and 50 mM H₂O₂ in presence of 1.6 μM CuCl₂ at 37 °C.^[130] The physiological relevance of such experiments needs to be considered. Liu et al., studied the effect of substituents on H₂O₂-mediated oxidation of thioketals.^[15a] An analog with the sulfur substituted with a 4-methoxy phenyl group (k = 42.5×10⁻³ M⁻¹h⁻¹) has a ~5-times faster kinetics than 4-chloro phenyl substituted thioketals (k = 8×10⁻³ M⁻¹h⁻¹).^[37] However, the kinetics of oxidative cleavage H₂O₂ is probably still too slow to be practical value as already discussed in Section 2 (Figure 2). Despite the sluggish kinetics with H₂O₂, some thioketals have been used in in-vitro and in-vivo applications. One reason for this success could be attributed to the improved reaction kinetics of thioketal with other ROS species. For instance, the degradation kinetics with different ROS species have been reported as O₂^•− < H₂O₂ < OH^•.^{[2c, 130]} Thioketal oxidation by OCl⁻ and ONOO⁻ is expected to be much faster than with H₂O₂ (Figure 2).^[15a] Moreover, the thioketal linker chemistry is widely used for photodynamic therapy (PDT). Singlet oxygen generated by the exogenous photoirradiation can cleave the thioketal linker within minutes, which might be the reason for the observed cleavage.^{[129, 131]}

5.4. Diphenylphosphine oxidation

Recently, diphenylphosphine-based fluorescent probes for the detection of OCl⁻ have been reported (Figure 14).^{[43c, 132]} Again, the lone-pair electrons on the trivalent phosphorus can act as an electron donor and inhibit the fluorescence from BODIPY through PET. However, oxidation of phosphorus to pentavalent phosphine oxide makes it an electron withdrawing group and prevents PET, leading to fluorescence turn-on. Fluorescence studies using different ROS show high selectivity towards OCl⁻ with a rapid response of 10 s. Further, the probe was used for imaging of hypochlorous acid in RAW 264.7 cells and mouse models of rheumatoid arthritis.^[43c] However, there are some critical points that need to be discussed. The phosphine groups are highly nucleophilic and can undergo reductive ligation with species like HNO^[133] and S-nitrosothiols^[134] via a mechanism similar to that of Staudinger ligation. Specifically, HNO is a highly reactive RNS formed by enzymes such as nitric oxide synthase, peroxidase, and xanthine oxidase,^[135] whereas S-nitrosothiols are formed by the S-nitrosation of cysteine thiols.^[136] These two reactive species need to be included in future selectivity studies.

Figure 14:

Diphenyl phosphine-based “turn on” fluorescent probe for OCl⁻.^[43c]

5.5. Hypochlorite-catalyzed hydrolysis of hydrazone linkers

Another widely used approach is the hypochlorite-catalyzed hydrolysis of hydrazone linkers (Figure 15). For instance, an aggregation induced ratiometric fluorescent probe was used for imaging HOCl in HepG2 cells.^[137] Specifically, the hydrazones are non-fluorescent. However, the HOCl-triggered cleavage of the hydrazone C=N bond leads to turn-on fluorescence. For this application, a hydrophobic tetraphenyl ethylene fluorophore (THG-1) was conjugated with a hydrophilic lactose unit via a hydrazone linker. The purpose of lactose moiety is to improve the solubility of the probe and prevent aggregation. In the presence of OCl⁻, the hydrazone linker is cleaved off, resulting in the removal of hydrophilic group and abolishment of the solubilization effect. Such kind of solubility change promotes aggregation of the hydrophobic fluorophore, and consequently results in the AIE fluorescence (Figure 15). Further, changes in the electronic nature of the fluorophore due to linker cleavage cause a change in the wavelength and facilitate ratiometric imaging work. However, it is well known that hydrazone can be cleaved by peroxynitrite as well.^[138] Further, probe specificity has not been confirmed by incubation with peroxynitrite and other ROS species. Later, the probe was used for imaging OCl⁻ in LPS treatedHepG2 cells. Along the same line, the OCl⁻cleavable hydrazone linker in conjugation with different targeting molecules was used for the organelle-specific imaging of hypochlorite. For instance, a ptoluene sulfonamide conjugated hydrazone-based turn-on fluorescent probe was used for imaging of hypochlorite in the live endoplasmic reticulum.^[15c] It is interesting to note that the proposed mechanism for OCl⁻ mediated cleavage does not seem to involve redox chemistry: it is purely hydrolysis. It is not clear why OCl⁻ is able to catalyze such a hydrolytic reaction.

Figure 15:

Hypochlorite-triggered hydrolysis of hydrazone linker as proposed in the original publication.^[137]

Besides the methods discussed here, there are other approaches for ROS triggered activation, including HOCl-induced cyclization of hydrazone to 1,2,4 triazole,^[139] catechol oxidation,^[140] naphthol deprotection followed by chlorination,^[47c] and ROS triggered decarboxylation of aryl oxalates.^[141] Further, ROS triggered nanomaterial disassembly has been explored for diagnostic and therapeutic applications.^{[13a, 142]} With the intention of staying focused on the theme of this review, these examples are not discussed in detail. Comprehensive information on ROS triggered activation chemistry can be found in some excellent review articles.^{[13a, 25, 33a, 42]}

6. Summary and Outlook

Clearly ROS-triggered delivery has tremendous potential for diverse applications. Recent research has led to findings that impact the analysis of the likelihood of success. We highlight a few factors to consider in designing approaches to ROS-sensitive delivery. First of all, ROStriggered activation starts with a bimolecular reaction. Therefore, the concentration of the intended ROS and the second order rate constant are critical factors to consider. As shown in Figure 2, some reaction pairs only allow slow activation with t_1/2 in the range of days. One will have to consider whether such kinetics are fast enough for intended applications. Second, if the delivery application is intended to take advantage of enhanced ROS levels under certain conditions, then it is important to understand the magnitude of difference. In some cases, a 2–3-fold difference in the level of an ROS may not give a theoretical chance for success. Third, if the intended application is to take advantage of the reactivity of a given ROS, one needs to examine the reaction rate issue in the context of the presence of other ROS because their reactivity can be different by order of magnitude. Such differences in reactivity could pose serious selectivity issues even if the “interfering species” is present at a much lower concentration. Furthermore, the reaction time used could amplify or diminish the observed selectivity. Therefore, some theoretical calculation could help optimize experimental conditions. Fourth, earlier reports of “selective probes” for a given ROS may not be based on comprehensive screening against all physiologically relevant ROS (and RNS). It is important to examine the experimental details before accepting the “face value” of a statement on selectivity. Fifth, a large percentage of reported ROS concentrations is based on its reaction with a probe over a period of time. This approach gives a “cumulative concentration” and could under or over-estimate the concentration depending on the reaction rate constant, continuous production of a given ROS, and whether the reaction went into completion. Therefore, in designing a ROS-sensitive delivery approach, one will need to consider these technical details. Sixth, the information available on the ROS levels under various pathophysiological conditions is far from ideal or complete. There is much more high-quality work needed in this area. Finally, we hope this review will stimulate discussions and lead to additional research to allow for more quantitative assessments of the feasibility of a given ROS-sensitive delivery approach.