Toxicant Isomers Characteized Using Fluorescence Emission of Coumarin Attached to Beta-Cyclodextrin

Chapter 1 reports a highly sensitive and selective array-based sensing strategy for classifying isomeric and analogous analytes based on their differential interactions with three supramolecular cyclodextrin-fluorophore sensors. Each analyte-sensor interaction results in a distinct fluorescence modulation response, and these variable responses are then statistically classified via linear discriminant analyses (LDA) into clusters of maximum separation. Three classes of isomeric analytes (aromatic alcohols, aliphatic alcohols, and hexanes) and two classes of analogous analytes (analogues of dichlorodiphenyltrichloroethane (DDT) and congeners of polychlorinated biphenyls (PCBs)) have been successfully classified with 100% accuracy. High sensitivity of this sensor is demonstrated as well, with limits of detection approaching or surpassing known levels of concern, and preliminary efforts at successfully classifying binary analyte mixtures using this sensor system are also

responses are then statistically classified via linear discriminant analyses (LDA) into clusters of maximum separation. Three classes of isomeric analytes (aromatic alcohols, aliphatic alcohols, and hexanes) and two classes of analogous analytes (analogues of dichlorodiphenyltrichloroethane (DDT) and congeners of polychlorinated biphenyls (PCBs)) have been successfully classified with 100% accuracy. High sensitivity of this sensor is demonstrated as well, with limits of detection approaching or surpassing known levels of concern, and preliminary efforts at successfully classifying binary analyte mixtures using this sensor system are also reported. I also want to thank my family for their love and support that they have given me.
I can say that your investment in me throughout the years has not been in vain.
I also want to thank my wonderful fiancée, Abigail Johnson, for helping me through these stressful times. I know that moving up to Rhode Island was tough on both of us and I would not have made it without you.
Finally, I want to thank all of the friends that I have made up here at URI.
Whether it was studying for an exam or trying to relax after a hard day, I will always cherish the time that we've spent together.    Table 4. Calculated limits of detection and comparisons to known levels of concern 18 Fluorescence Modulation Summary    accuracy. High sensitivity of this sensor is demonstrated as well, with limits of detection approaching or surpassing known levels of concern, and preliminary efforts at successfully classifying binary analyte mixtures using this sensor system are also reported.

INTRODUCTION
The selective detection and accurate quantification of structurally similar analytes is a major environmental challenge for chemists and toxicologists, as structurally similar analytes often have widely disparate toxicities and environmental degradation pathways. 1 The most common strategy to address this challenge is to use mass spectrometry based methods such as liquid chromatography-mass spectrometry (LC-MS) 2 or gas chromatography-mass spectrometry (GC-MS). 3 However, there are significant drawbacks associated with this approach, including the significant costs and time necessary to conduct such analyses, 4 which limits the ability to conduct high throughput assays. 5 An alternate strategy is to use array-based sensing systems, which have gained in popularity in recent years. 6 This approach relies on the development of a chemical signature for each analyte based on analyte-specific interactions with a series of sensors. In a multi-component system, each individual analyte develops a unique response pattern, which is then compared against known samples to enable accurate identification.
Array-based sensing systems can be combined with supramolecular sensors, which rely on differential non-covalent interactions of isomeric analytes with supramolecular hosts, including cyclodextrins, 7 fluorescent polymers, 8 molecularly imprinted polymers, 9 and metal-organic frameworks (MOFs). 10 Cyclodextrin-based detection systems in particular have used either covalent 11 or non-covalent attachment 12 of a spectroscopically active unit to achieve a read-out signal. However, often the detection specificity in this kind of sensing is limited due to structural similarities among related groups of analytes.
Although supramolecular array-based systems overcome many challenges associated with mass-spectrometry based detection methods, the analyte scope explored in most of these reports have been limited to aromatic small molecules such as xylenes 13 and nitrotoluenes. 14 In a real-world contaminated environment, the nature of the various pollutants is highly complex, 15 and includes complex mixtures of aromatic and non-aromatic compounds. 16 This kind of situation requires the development of a sensing system which is rapid, simple and efficient in classifying a broad range of persistent organic pollutants (POPs). 17 The proper identification of the contaminants using such a system would provide knowledge that would then inform the rational development of a decontamination strategy.
Our group has previously employed both β-cyclodextrin and γ-cyclodextrin in the development of array-based detection systems for the sensing and classification of a wide variety of environmental toxicants and POPs. 18 The sensing strategy is based on cyclodextrin promoted analyte-to-fluorophore energy transfer as well as on analyteinduced fluorescence modulation. This fluorescence modulation relies on doping of the free fluorophore into the cyclodextrin solution prior to analyte addition, which often leads to binding of the fluorophore in the cyclodextrin and reduces the cyclodextrin's ability to bind the target analyte. As such, introduction of the analyte to the fluorophore-cyclodextrin solution requires the analyte-cyclodextrin association constants to be higher than the fluorophore-cyclodextrin constants in order to achieve binding ( Figure 1A), or it requires the formation of higher order association complexes between the analyte, cyclodextrin and fluorophore (i.e. ternary complex formation) ( Figure 1B). Such higher order association complexation is probable only for γcyclodextrin. 19 β-cyclodextrin, by contrast, has been extensively reported to participate in the formation of binary association complexes, 20 and ternary inclusion complexes with β-cyclodextrin are less reported. 21 Although in our previous work we have successfully distinguished between classes of analytes, the results tend to cluster structurally similar analytes and analyte derivatives near each other. Herein, we report the development of a highly selective array-based detection system using fluorophore-functionalized perbenzylated βcyclodextrin sensors as the key components, which directly enables binary complex formation between the fluorophore-cyclodextrin and target analyte ( Figure 1C). Each individual sensor is highly selective towards a specific isomer/analogue within a group of structurally similar analytes, which enables the array to distinguish isomers and structural analogues with high efficiency. Three classes of isomeric analytes and two classes of structurally similar analytes have been successfully classified based on this strategy, with a classification accuracy of 100% in every case. High sensitivity is demonstrated as well, with limits of detection approaching or surpassing literature-reported levels of concern. Finally, preliminary efforts at using this system for accurate identification of binary analyte mixtures are also reported.

RESULTS AND DISCUSSION
We employed a series of three cyclodextrin-based supramolecular   The synthesis of supramolecular hosts S2 and S3 is shown in Scheme 1.
Perbenzylated β-cyclodextrin is obtained from reacting β-cyclodextrin with excess benzyl chloride at room temperature in DMSO in the presence of excess sodium hydride. 22 Regioselective debenzylation of the primary rim is effected by treating the perbenzylated β-cyclodextrin with DIBAL-H in toluene (ways to control the selectivity to achieve mono vs. di-debenzylation are discussed in the ESI). 23 This is followed by Steglich esterification 24 with the acid derivative of fluorophore 4, yielding mono-and di-functionalized sensors S2 and S3. New supramolecular compounds S2 and S3 were fully characterized by 1 H-NMR, 13 C-NMR, MALDI-TOF mass spectrometry, and UV-visible and fluorescence spectroscopy.

Scheme 1. Synthesis of supramolecular hosts S2 and S3.
The fluorescence emission responses of sensors S1, S2 and S3 were  The choice of perbenzylated β-cyclodextrin as a receptor over that of βcyclodextrin is due to the stronger binding of organic guest molecules as a result of its extended hydrophobic cavity. 25 In particular, a comparison of association constant values of analyte 5 revealed a 1000-fold increase in the binding constant with perbenzylated β-cyclodextrin over that of the naturally occurring β-cyclodextrin, and this binding constant is even higher in the fluorophorefunctionalized cyclodextrins S2 and S3 (Table 1). These binding constants are orders of magnitude higher than highest literature-reported binding constants for analyte 5 in β-cyclodextrin (K a = 50-215 M -1 ). 26 In general, higher association constants for the binding of an analyte in a sensor are known to lead to improved sensor performance. 27 Table 1. Association constants of analyte 5 in β-cyclodextrin a and in Although the fluorescence response was essentially unchanged with analyte addition in the case of S1 (leading to modulation values near 1.00 in every case), significant differences in the response patterns of sensors S2 and S3 with analyte addition were observed (Table 2). An example of analyte-induced fluorescence modulation for analyte 8 is shown in Figure 5 and highlights the small but distinct fluorescence changes observed for S2 and S3. The fluorescence signals of sensors S1-S3 in the presence of analytes 5-8 were subjected to linear discriminant analysis, and enabled 100% selectivity between the different aromatic alcohol isomers ( Figure 6).
This selectivity is particularly noteworthy as such isomers are challenging to separate using other chemical techniques. 28   The binding of other structural isomers and analogues in supramolecular hosts S1-S3 also led to noticeable, analyte-specific changes in the fluorescence emission (Table 3), with some key results highlighted in Figures 7-10. Table 3. Fluorescence modulation of sensors S1 -S3 in the presence of analytes 9 -

26.
Analyte S1 S2 S3 9 1.01 ± 0.00 0.89 ± 0.00     The sensor S1 shows a fluorescence modulation value close to 1.00 for all the tested analytes (see Figure 5A for an example), indicating minimal to no effect on the fluorescence emission of the fluorophore with the introduction of the analyte. In highlighting the power of the cyclodextrin-based supramolecular sensor in differentiating even small structural changes. Overall, the uses of sensors S1-S3 in combination with these analytes enabled 100% differentiation using linear discriminant analysis ( Figure 6).
Interestingly, although sensor S3 demonstrated nearly identical fluorescence modulation values in response to analytes 13 and 15, sensor S2 was able to clearly differentiate between those two analytes. These results illustrate the fact that altering the degree of functionalization of the supramolecular sensor alters its response for a target analyte.
Analytes 17-21 represent aliphatic n-hexane (compound 17), its commonly occurring structural isomers (compounds 18-20, generated in 10-30% yield from industrial production of hexane) 35 and its cyclopentane analogue (compound 21). The fact that hexanes co-occur as isomeric mixtures of compounds 17-20 complicates their accurate characterization as well as fuel applications that rely on such characterization. 36 Using this supramolecular sensing strategy, 100% accurate classification between these analytes has been achieved.
Analytes 22-26 represent a class of (POPs) called polychlorinated biphenyls (PCBs), which cause neurotoxicity 37 and endocrine disruption. 38 As a result, the use of PCBs has been banned in many countries; however, their extreme environmental persistence means that significant amounts of PCBs are still found in the environment today. 39 100% accurate classification has been achieved for these analytes via this strategy, which is particularly relevant because these analytes have widely disparate toxicities.
The ability of this detection method to generate well-separated signals was further investigated by generating an array with all analytes from all classes. In this case, the array exhibited well-separated clusters based on compound class, as well as excellent separation within each class. Overall, 100% accurate identification was obtained (see ESI for more details).
The limits of detection for each sensors S1, S2 and S3 for each class of analytes were calculated to determine their ability to sense analyte concentrations at or near environmental levels of concern and literature-reported levels of toxicity. In every case, the calculated limits of detection were at or lower than the literature reported limits of concern for these analytes (Table 4). These results illustrate the high sensitivity of this detection method.
Practical applications of this system require the capability to identify analyte mixtures, because environmental contamination scenarios almost never involve a single chemical contaminant. To that end, preliminary work focused on identification of 1:1 binary mixtures of aromatic alcohol analytes 5-8. Using the same supramolecular sensors and the same linear discriminant analytical techniques, 83% accurate identification of the 1:1 binary mixtures was obtained ( Figure 11).
Interestingly, the mixture of analytes 5 + 7 is grouped near the mixtures of analytes 6

S3
4.59 1.00 44 a Limits of concern have not been established for these compounds.   Both the excitation and emission slit widths were 3 nm. All of the fluorescence spectra were integrated vs. wavenumber on the X-axis using Origin Pro Version 9.1 software. All arrays were generated using SYSTAT Version 13.

DETAILED PROCEDURES DETAILED SYNTHETIC PROCEDURES
Overall Synthetic Scheme: To a stirred solution of oven-dried β-cyclodextrin (2.00 g, 1.76 mmol, 1.0 eq.) in DMSO (100 mL) under nitrogen, sodium hydride (2.60 g, 65 mmol, 36 eq.) was added carefully. The solution was allowed to stir for one hour at room temperature, after which time benzyl chloride (18.5 mL, 65 mmol, 36 eq.) was added over the course of one hour. The reaction mixture was stirred for 18 hours at room temperature, followed by the addition of methanol (20 mL

DETAILED PROCEDURES FOR FLUORESCENCE MODULATION EXPERIMENTS
Fluorescence emission spectra were obtained using a Shimadzu RF-5301PC spectrophotofluorimeter with 3 nm excitation and 3 nm emission slit widths. In a quartz cuvette, 0.5 mL of S1, S2, or S3 solutions (5 μM in DMSO) and 2 mL of deionized water were combined. Then, the solution was excited at 320 nm, and the fluorescence emission spectra were recorded. Repeat measurements were recorded for four separate trials.
The fluorescence emission spectra were integrated vs. wavenumber on the X- All of the fluorescence emission spectra were integrated vs. wavenumber on the X-axis, and calibration curves were generated. The curves plotted the analyte concentration in μM on the X-axis, and the fluorescence modulation ratio on the Yaxis. The curve was fitted to a straight line and the equation of the line was determined.
The limit of detection is defined according to Equation 2: Where SD blank is the standard deviation of the blank sample and m is the slope of the calibration curve. In cases where the slope of the trendline was negative, the absolute value of the slope was used to calculate the LOD. In all cases, the LOD was calculated in μM.

BENESI-HILDEBRAND PLOTS FOR NMR TITRATION
Accordingly, 15 Py values can be used for the determination of solvent polarity due to the strong S 0 S 2 absorption between electronic states and the relaxation between S 1 S 0 . Band I corresponds to the fluorescence relaxation between S 1 v=0 S 0 v=0 and band III corresponds to the fluorescence relaxation between S 1 v=0 S 0 v=1 . The Py value has been found to be independent of the presence of oxygen . 15 Py values have been used to calculate the binding constant of pyrene to a host as described in

Equation 1.
Pyrene is found to have a 2:1 CD:pyrene stoichiometry rather than a 1:1 complex when mixed with high concentrations of beta-cyclodextrin (β-CD). 11 A 1:1 complex is more favorable than 2:1 CD:pyrene complex for pyrene in gamma-cyclodextrin (γ-CD) solutions, although at high concentrations of γ-CD, 2:2 complexation may be a more favorable state. 11,16 The 2:1 complex may be favored due to the limited space in the β-CD cavity, 11,9 which would decrease the polarity around pyrene (giving a Py value less than 1). Limited research has been conducted on PAH binding with β-CD

RESULTS
The fluorescence intensity of pyrene increased as the amount of pyrene added to the solutions increased; however, the Py value was consistent with small concentrations of pyrene (Table 1). Py values only changed significantly when the solvent polarity changed. At concentrations of β-cyclodextrin greater than 1 mM, the Py value was below 1, similar to what was found in the non-polar solutions (n-octane).
Pure γ-cyclodextrin solutions showed Py values greater than 1 at all concentrations. of the excimer emission is seen in pure γ-cyclodextrin, but excimer emission is higher in the β:γ mixture (from cyclodextrin ratios above No significant excimer peaks were seen with pure β-cyclodextrin solutions.

DISCUSSION
Pyrene is sterically constrained as to how it can occupy the cavity of cyclodextrin. β-cyclodextrin has a cavity size of 7.8 Å, 18 which doesn't allow pyrene to fit along its z-axis ( Figure 8) and restricts it to the y-axis, 19 and γ-cyclodextrin has a cavity size of 9.5 Å, 18 which is large enough to fit pyrene along the z-or y-axis. 19 The steric limitations of pyrene fitting into β-cyclodextrin would leave pyrene partially exposed to the polar solvent. However, as the concentration of β-cyclodextrin increases, there is evidence that it moves from a 1:1 pyrene:cyclodextrin environment to a 1:2 pyrene:cyclodextrin environment, which would encapsulate pyrene and represent a more non-polar microenvironment (seen by the shift in the I/III ratio from > 1 to < 1). Since γ-cyclodextrin has a larger cavity size, our results suggest that water may also occupy the cavity along with the pyrene, maintaining a more polar microenvironment (I/III is still > 1 at all concentrations of γ-cyclodextrin). As γcyclodextrin concentration increases above 1 mM, the excimer peak increased (470 nm), which indicates the likelihood of a 2:2 pyrene:cyclodextrin complex. 16 There is a continual increase in the excimer/monomer ratio as the γ-cyclodextrin concentration increases, however when mixed with β-cyclodextrin, there is a ~30% increase in the excimer/monomer ratio (table excimer). This could be due to the formation of 1:1:1 pyrene:β-cyclodextrin:γ-cyclodextrin, but the different number of cyclodextrin monomers in between γ and β-cyclodextrin (β -7 glucose units, γ -8 glucose units) may sterically restrict this complexation. Methyl-β-cyclodextrin and 2-hydroxypropyl-β-cyclodextrin both I/III band ratios that were greater than 1 and the absence of any notable excimer peak (Figure 8).
Even as the concentration of each β-cyclodextrin derivative increases, there is no change in the Py values. This suggests that pyrene maintains a polar microenvironment when occupying the cavity of methyl-β-cyclodextrin or 2hydroxypropyl-β-cyclodextrin. No significant change in the Py value was seen when either β-cyclodextrin derivative was mixed with β-cyclodextrin (Table S1). The binding constant of pyrene-d10 to β-cyclodextrin increased compared to pyrene-h10 with β-cyclodextrin, however, the opposite was seen with γ-cyclodextrin.
Previous studies have reported that deuterated guest molecules decrease in binding with cyclodextrin host. 21  Although benzo[a]pyrene does not have a change in its band ratios with a change in solvent polarity, the change in the excimer peak ratios is evidence that benzo[a]pyrene has a higher affinity for γ-cyclodextrin compared to β-cyclodextrin, which is the opposite that we see with pyrene.