The particle-induced formation of hydroxyl radicals has gained considerable attention by toxicologists, environmental scientists, and geochemists 1. The hydroxyl radical is a molecule with an unpaired electron, which will react nonspecifically with most organic molecules within nanoseconds after their formation 2. It is included in a group of molecules termed reactive oxygen species (ROS), which also includes hydrogen peroxide (H₂O₂) and superoxide (O₂^•-). Hydroxyl radicals (^•OH) have been implicated in genotoxicity 34 and oxidative stress 456 that may contribute to lung disease upon exposure to asbestos 7, silica 38, and other airborne particulate matter 910. Hence, ^•OH formation in vitro and in vivo has been used as an indicator for particulate-induced toxicity potential 34511121314.

The detection of hydroxyl radicals is not a trivial process; detection methods require high sensitivity, high selectivity, and low detection. The extreme reactivity of ^•OH precludes its direct detection by conventional spectroscopic methods, and instead, techniques employ reaction of ^•OH with a target molecule (probe). Upon reaction, the probe's characteristics such as light absorption, fluorescence, and/or electron spin resonance may change. These changes in the probe's characteristics can be correlated with estimated concentrations of ^•OH (or another ROS) generated under controlled conditions. The resulting calibration curve can then be used to quantify particulate-generated ^•OH.

Several techniques have been employed for detecting particle-induced formation of ^•OH. Electron spin resonance (ESR) spectroscopy is a technique for detecting molecules with unpaired electrons. By adding a probe molecule such as 5,5-dimethyl, 1-pyrroline N-oxide (DMPO), commonly referred to as a spin trap, the lone electron on ^•OH can be captured (trapped) on the DMPO molecule (i.e., DMPO-OH). The reaction product, DMPO-OH, is also a radical species, but it has a sufficiently long half life so that it is possible to record its distinctive electron paramagnetic resonance spectrum 15161718. In other techniques involving cells or tissues, biological products of ^•OH-induced oxidation which can be subsequently quantified include lipid hydroperoxides and aldehydes 19, DNA strand-breaks 320, products of RNA degradation 21, nucleotide base oxidation products 222324 and multiple signaling molecules such as cytokines 2526 and the transcription factor, p53 27, the upregulation of which is indicative of inflammation and apoptosis (i.e., programmed cell death). Still other methods employ a probe molecule which may undergo changes in absorbance (e.g., leuco crystal violet) 28 or fluorescence [e.g., 2',7'-dichlorofluorescein (DCFH)] 293031 when oxidized by hydroxyl radicals.

Although these techniques are suitable for certain experimental objectives, they all have limitations. Some techniques are better adapted to use with viable cells and others to cell-free solutions, some methods require specialized equipment and others, additional probe preparation steps. The electron spin resonance method with DMPO has been used extensively but it requires specialized equipment (i.e., an ESR spectrometer) and the method is susceptible to artifacts when ferric iron is present [e.g., Fe(III) associated with a particle] 32. In addition, the requirement for tuning the ESR spectrometer and the short half-life of DMPO-OH (i.e., 2.9 minutes 33) may lead to irreproducible results and difficulties in generating calibration curves. 2',7'-dichlorofluorescein (DCFH) has been used widely, but it will react with several reactive oxygen species, not just hydroxyl radicals 31. Since the diacetate ester of DCFH (H₂DCFDA), which is frequently employed because of its high permeability to viable cells, will not react with ^•OH unless it undergoes a de-esterification step either within cells or chemically 30, its use with particle suspensions in the absence of cells requires an extra step in the method protocol. Evaluating the fate of nucleic acids after exposure to particles is another technique that has been employed for the detection of particle-generated hydroxyl radicals 32021. Although this method is useful for estimating a genotoxic potential from exposure to certain particles, it is not well adapted for quantifying hydroxyl radical concentration, especially when the multiple products of reaction with nucleic acid are formed in situ within viable cells. In addition, when cellular DNA degradation is evaluated, DNA strand repair mechanisms may reduce the level of particle ROS-induced strand breaks 34.

For this study, our objective was to evaluate several hydroxyl radical detection techniques that could overcome some of the challenges described in the preceding paragraph. Several relatively new compounds have been developed for ROS detection using fluorescence. Some of them will detect many ROS and others have been designed to react only with highly reactive oxygen species such as ^•OH. Three of the five probes evaluated will react directly with ^•OH, while the other two require reaction of ^•OH with another molecule that can directly react with the probes.

In order to determine hydroxyl radical generation from hydrogen peroxide in the presence of HRP or from ferrous iron solutions and particulate suspensions, calibration curves were first produced. In the presence of HRP, the fluorescence intensity of all of the probes increases with increasing hydrogen peroxide concentration (Fig. 1). The gain on the fluorimeter was adjusted in this series of experiments to display a relative fluoresence intensity of 1000 in the presence of 1000 nM H₂O₂ along with 0.2 μM HRP and each probe at its concentration as listed above in Methods. The baseline of the instrument was adjusted to display a relative intensity of zero in the presence of water alone. It is evident from Figure 1 that in the absence of H₂O₂, all of the probes except for proxyl fluorescamine and tempo-9-ac show relative fluorescence intensities that are less than 5% of the signals they generate in the presence of 1000 nM H₂O₂ and HRP. With addition of HRP alone, in the absence of exogenously added H₂O₂, APF, DCFH, and amplex ultrared show small increases in fluorescence intensity which may reflect spontaneous formation of H₂O₂ but these signals are still less than 20% of the signals generated in the presence of 1000 nM exogenous H₂O₂ and HRP. The two fluorogenic free radical scavengers, tempo-9-ac and proxyl fluorescamine, however, clearly undergo significantly smaller increases in fluorescence with increasing H₂O₂ concentration than the other probes. We hypothesize that these smaller increases reflect the indirect mechanisms by which hydroxyl radicals modify the two spin traps (through phenoxy- and thiyl radicals in the case of tempo-9-ac and through methyl radicals in the case of proxyl fluorescamine), resulting in suboptimal conversion of the probes to the spin-neutral species.

In order to measure hydroxyl radicals generated by ferrous iron solutions and pyrite suspensions, a different calibration curve was used. The gain was still adjusted to display a relative fluorescence intensity of 1000 in the presence of 1000 nM H₂O₂ and HRP, but the signal from each probe in the absence of H₂O₂ and HRP (i.e., values to the left of the vertical axis), rather than the signal from water alone, was set to a relative fluorescence intensity of zero.

In an aqueous solution containing dissolved oxygen, the addition of soluble ferrous salts or ferrous iron-containing minerals such as pyrite results in the formation of both hydrogen peroxide and hydroxyl radicals (eqs 1 and 2). All of the probes were incubated with multiple concentrations of ferrous iron in 25 mM phosphate buffer, pH 7.4 (Fig. 2). The results show differences in each probe's sensitivity to hydroxyl radicals generated in ferrous iron solutions. When less than 50 μM ferrous iron salts were added to the probes, the fluorescence intensities of all of the probes were observed to increase. The apparent concentration of detected hydroxyl radicals declined when greater concentrations of ferrous iron were added to tempo-9-ac. Since the calibration curve generated for proxyl fluorescamine has a very shallow slope, small changes in the fluorescence intensity of this probe equate to very large apparent changes in hydroxyl radical concentration, reflected in the vertical axis for proxyl fluorescamine on the right side of Figure 2. An additional complication in detection of hydroxyl radicals generated by ferrous salts and minerals using amplex ultrared as a probe arises from the capacity of the dye to be oxidized by H₂O₂ to the fluorescent resorufin in the presence of HRP. It is evident that H₂O₂ is indeed formed in addition to hydroxyl radicals upon addition of ferrous iron salts or minerals because addition of HRP to the incubation mixtures containing the iron samples and amplex ultrared results in a significant increase in fluorescence.

One of our goals of having an ROS detection method is for its implementation for the detection of particulate-generated hydroxyl radicals. From previous research, we know that aqueous suspensions of pyrite particles will form hydroxyl radicals 21. In this study, all of the probes were incubated with several different concentrations of pyrite particles and then evaluated for the generation of hydroxyl radicals (Fig. 3). The results show that for all of the probes except amplex ultrared (in the presence of HRP) and tempo-9-ac, fluorescence intensity generated by the probes increases proportionally to the total pyrite surface area exposed to the aqueous solvent. In other words, the probes record an increase in the formation of hydroxyl radicals with an increase in available surface area. The increased fluorescence from tempo-9-ac in the presence of pyrite particles shows a similar iron dose dependence to that observed in the presence of soluble ferrous salts: the fluorescence increases with small amounts of ferrous iron or pyrite and then decreases when more of the iron-containing species is added. The results from amplex ultrared are similar but when no HRP is added, the fluorescence increases nearly linearly. As in the studies with soluble ferrous iron salts described above, the sensitivity of proxyl fluorescamine to hydroxyl radicals from pyrite suspensions is so low that the concentrations of hydroxyl radicals formed cannot be detected with precision, and the apparent values, shown on the right vertical axis, are clearly outside the range of those detected by the other probes.

A major goal of these studies was to evaluate several fluorogenic probes for their sensitivity to hydroxyl radicals, especially at the lower limit of detection, as well as their suitability for use with particulate mineral suspensions. Specifically, we have used the probes to determine the apparent concentrations of hydroxyl radicals generated in a 24 hour time period, at room temperature, in phosphate buffer at pH 7.4. Changing any of these variables would probably lead to different apparent concentrations of hydroxyl radicals being detected. However, by keeping these variables constant, a comparison of all the probes could be performed for three systems: (1) H₂O₂ and HRP, (2) Fe(II), and (3) pyrite suspensions.

After reaction with H₂O₂ and HRP, each probes fluorescence' is different (Fig. 1). The curves from DCFH, APF, and amplex ultrared are similar but different from those recorded for tempo-9-ac and proxyl fluorescamine. This is expected because the former three will directly react with H₂O₂ in the presence of HRP and hence their greater sensitivity (APF is the most sensitive of all probes). Because both tempo-9-ac and proxyl fluorescamine will not directly react with H₂O₂ in the presence of HRP, their responsiveness' is reduced. In the presence of HRP but absence of H₂O₂, all of the probes exhibit some level of fluorescence. This fluorescence level is higher than the background fluorescence level for each probe (listed on the left side of the vertical axis, Fig. 1). This indicates that either there is some H₂O₂ in the water that was used to dilute the solutions in the experiments, or that HRP directly reacts with the probes.

Calibrating the fluorometer is necessary but it presents some challenges.

The fluorometer was calibrated with water as zero fluorescence and a reaction of H₂O₂/HRP/glutathione & phenol (for tempo-9-c)/DMSO (for proxyl fluorescamine) as a fluorescence value of 1000. Compared to the fluorescence of only DCFH, APF, and amplex ultrared in a pH buffered water solution, the addition of H₂O₂ & HRP results in much greater fluorescence intensity. When the fluorometer is calibrated with this large difference in fluorescence intensities, subsequent measurements have a greater accuracy. Both tempo-9-ac and proxyl fluorescamine do not respond with as great of an increase in fluorescence when incubated with H₂O₂ and HRP. Therefore, the sensitivity of tempo-9-ac and proxyl fluorescamine are lower than the sensitivities of DCFH, APF, and amplex ultrared. The background fluorescence for proxyl fluorescamine is only slightly less than the fluorescence of proxyl fluorescamine reacted with 1 μM H₂O₂ and HRP (i.e., 890 compared to 1000). When proxyl fluorescamine is incubated with either Fe(II) (Fig. 2) or pyrite (Fig. 3), the larger fluorescence intensities measured are calculated as significant concentrations of hydroxyl radicals. This problem may be resolved by using more than 1 μM H₂O₂ in the calibration of proxyl fluorescamine, but that would defeat our objective of comparing all of the probes uniformly.

If the probes reacted equally with a ferrous iron solution, then they would all detect the same concentration of hydroxyl radicals and all of the curves would overlap. The results shown in Figure 2 show different calculated concentrations of hydroxyl radicals being detected. The trend of tempo-9-ac is unique because it's fluorescence increases with Fe(II) up to 50 μM, but then the fluorescence decreases when the concentration of Fe(II) is further increased (Fig. 2). The same result occurs when tempo-9-ac is incubated with pyrite particles (Fig. 3). The increasing fluorescence intensities of the other probes with increasing loadings of pyrite and Fe(II), suggests that tempo-9-ac is either only capable of detecting small concentrations of hydroxyl radicals or another factor is limiting it's fluorescence such as an interaction between iron and the probe. The addition of HRP to amplex ultrared results in a greater detected concentration of ROS. A ferrous iron solution will generate both hydrogen peroxide and hydroxyl radicals (eqs 1 to 3). In the presence of HRP, the relative contribution to amplex ultrared fluorescence' due to hydroxyl radicals versus hydrogen peroxide remains unknown.

In the presence of pyrite particles (Fig. 3), each probe's increase in fluorescence shows similar trends to the results when the probes are incubated with ferrous iron (Fig. 2). The main differences are in the reactivities of amplex ultrared with HRP and tempo-9-ac. For these probes, increasing the pyrite loading above 3 m²/L results in a decrease in the detected hydroxyl radical concentration. This may be due to several factors: (1) a decrease in the stability of amplex ultrared in the presence of greater concentrations of hydroxyl radicals, (2) amplex ultrared adsorption to the larger surface areas of pyrite present in higher loadings, (3) an interaction between iron and the probe or one of the reactants (e.g., glutathione), or (4) alteration of solution pH in the presence of higher loadings of pyrite. Although we evaluated pyrite particles for generation of ROS for 24 hours, the time span of incubation with the probes could have been shorter or longer. The probes appear to increase in fluorescence over time when incubated with particles suggesting that the detection of ROS is cumulative.

Throughout the study, the solutions containing APF were the easiest to prepare, contained a minimal amount of additional reactants and the results were the easiest to interpret. Unlike DCFH, APF was used directly from the purchased vial without further treatment. Both proxyl fluorescamine and tempo-9-ac required addition of DMSO and glutathione/phenol, respectively. Proxyl fluorescamein and tempo-9-ac also required preparation from powders and quantification using UV-Vis spectroscopy – APF was purchased as a 5 mM solution.

The objective of these studies was to evaluate several fluorescence-based techniques for measuring hydroxyl radicals in solution with the ultimate goal of using one or more of the techniques to compare the concentration of hydroxyl radicals that are generated in particulate suspensions. The results show that some probes are more suitable for these goals. APF, DCFH, and amplex ultrared do not require extra reactants other than water and a pH buffer to quantify hydroxyl radicals, which results in their higher sensitivity and ease of use. Proxyl fluorescamine and tempo-9-ac both require other reactants and their use is limited to high precision measurements and solutions with low concentrations of reactants such as ferrous iron and pyrite, respectively. DCFH and amplex ultrared would also be suitable but DCFH requires an additional de-esterification step for its use in cell-free solutions and both DCFH and amplex ultrared will react with ROS other than hydroxyl radicals. APF is both sensitive and selective for only the most highly reactive ROS, including hydroxyl radicals. The fluorogenic probes that were evaluated here were chosen because they can be purchased and used with little further treatment. This study is not an extensive study of all methods available and many other probes are available for the detection of ROS; for reviews, see 616263.

CAC designed the study, performed the experiments and drafted the manuscript. SRS participated in the design of the study. MAAS coordinated the study and provided funding. All authors read and approved the final manuscript.