Characterization of Nitrazine Yellow as a Photoacoustically Active pH

Mar 5, 2015 - Department of Chemistry and Physics, Florida Gulf Coast University, 10501 FGCU Boulevard South, Fort Myers, Florida 33965-6565, United ...
1 downloads 3 Views 1MB Size
Article pubs.acs.org/ac

Characterization of Nitrazine Yellow as a Photoacoustically Active pH Reporter Molecule Jordan E. Brown,† Lilibet Diaz, Ty Christoff-Tempesta,† Kathryn M. Nesbitt,†,‡ Julia Reed-Betts,† John Sanchez,† and Kevin W. Davies* Department of Chemistry and Physics, Florida Gulf Coast University, 10501 FGCU Boulevard South, Fort Myers, Florida 33965-6565, United States S Supporting Information *

ABSTRACT: Throughout the fields of biomedical imaging, materials analysis, and routine chemical analysis, it is desirable to have a toolkit of molecules that can allow noninvasive/ remote chemical sensing with minimal sample preparation. Here, we describe the photophysical properties involved in photoacoustic (PA) measurements and present a detailed analysis of the requirements and complications involved in PA sensing. We report the use of nitrazine yellow (NY) as a wellbehaved PA pH reporter molecule. Both the basic and acidic forms of NY are photoacoustically well-behaved and allow for rapid and noninvasive measurement of pH in either transparent or turbid media. We also find that the serum protein-bound form of NY is photoacoustically well-behaved and should permit applications in noninvasive 3D imaging (e.g., the lymphatic system). molecules absorbing a photon (molar absorptivity, ε, and absorbance, A); (ii) the energy contained in the photon (excitation energy, Eex), and the relative intensity of the light pulse (Ep); (iii) the fraction of molecules that utilize each relaxation pathway (quantum yield of each process, Φ) and the total fraction that is converted into heat (f h); (iv) how quickly these processes relax the molecule (lifetime for the process, τ); (v) how much the solvent expands upon heat deposition (thermal expansivity of the solvent, χs); (vi) if there is a change in the solvated volume of the molecule stemming from isomerization or photoinitiated reactions (ΔVrxn); (vii) the radial expansion and diminishment of this acoustic wave en route to the detector; and (viii) the sensitivity of the apparatus (empirical instrument response factor, κ′). Equation 1 is relevant for a single measurement, where the contributions of ΔVheat and ΔVrxn (and the resulting signals: Sheat, the signal associated with heat deposition, and Srxn, the signal stemming from chemical processes such as isomeration and fragmentation) occur concurrently and require additional experimental work to disentangle. When present, both contribute to the observed signal (Sobs, eq 2). In this work, we will use the terms f obs (the apparent fraction of energy h deposited as heat, which may contain a nonthermal volumechange) and f h (the actual fraction of energy converted to heat, after ΔVrxn has been controlled for or found to be absent) in

I

n transparent systems, spectroscopic measurements allow various properties to be measured routinely and noninvasively. However, these measurements are less reliable in turbid media and rarely will permit three-dimensional measurement or detection. The photoacoustic (PA) effect has been used to measure energetic,1 spectral,2−4 or kinetic information5,6 (often simultaneously) for processes with lifetimes in the nanoseconds to microseconds range7−20 (references illustrative, not exhaustive). In the past decade or so, the PA effect has been developed into a measurement and imaging modality, allowing chemical measurements within turbid media, including living tissue. Here, we examine the PA behavior of nitrazine yellow (NY), a known pH indicator dye. NY demonstrates ideal PA behavior for use as a reporter molecule, including large molar absorptivity, spectral characteristics that change in direct correlation with a useful property (pH), well-behaved photoacoustic properties, water solubility, and negligible toxicity. In this work, we demonstrate NY’s PA behavior to be well-suited to high-quality pH measurements. Additionally, we find that the NY-serum albumin and NY-poly-Lysine (poly-Lys) complexes are similarly well-behaved and should permit application to in vivo PA imaging. Photoacoustic Effect. When a molecule absorbs a photon, some (or all) of the absorbed energy is converted into heat. When this occurs from a short-lived light pulse, the sudden heating of the molecule’s solvent shell results in a local volume expansion (ΔVheat) and propagates as an acoustic wave. The magnitude of this wave will depend on a number of factors, summarized here and shown in eqs 1−3: (i) the number of © XXXX American Chemical Society

Received: September 18, 2014 Accepted: March 5, 2015

A

DOI: 10.1021/ac503515k Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

ments, in turbid systems. For example, Wang and co-workers26 have used the PA effect to perform 3D chemical sensing. In their work, they have measured oxygenation levels in single blood vessels within living tissue via the photoacoustic determination of the formal concentration of Hb and the measurement of the concentration of Hb-O2; multicomponent Beer’s Law analysis allowed noninvasive measurements of hemoglobin’s oxygen saturation level at depths of up to 8 mm. The biomedical imaging community has developed instrumentation and algorithms that exploit the PA response of hemoglobin to recover 3D images within turbid media (both tissue phantoms and living tissue). After a sample is flashed with a short pulse of light (typically 10 ns) is used to calibrate the instrument. When ΔVrxn ≠ 0, eqs 2 and 3 allow both the thermal and reaction volumes to be determined by varying the solvent expansivity (usually by varying solvent or temperature), allowing the actual f h to be measured without contributions from ΔVrxn. S obs = κ′Ep(1 − 10−A)χs f hobs

(1)

S obs = S heat + Srxn

(2)

f hobs hνχs = fh hνχs + ΔVrxn Φ

(3)

Typically, the volume change associated with heat deposition (Sheat) is the predominant signal source, but when present ΔVrxn can cause a significant contribution to Sobs. Previous work on biomolecules has shown the volume increase associated with the photodissociation of CO-hemoglobin and CO-myoglobin to be in the range of 5−15 mL/mol1,21,22 and the photoisomerization of rhodopsin-lumirhodopsin and isorhodopsinlumirhodopsin to be 29 and 39 mL/mol, respectively.20 When the PA effect is used for chemical sensing, it becomes increasingly important to account for any ΔVrxn from all species involved in the measurement. For example, if the results in ref 3 (Figure 1 of their work) had failed to account for ΔVrxn,

Figure 1. Absorptivity of NY/phosphate buffer solutions. Inset values refer to the buffered pH of the solution; wavelengths used for photoacoustic measurements shown as vertical lines.

would have been incorrectly determined to 0.83, then f obs h representing a 7% underestimation vs their correct finding of f h = 0.89 for human oxyhemoglobin (Hb-O2). Other molecules have been found to have larger ΔVrxn magnitudes, and both expansions and contractions have been reported.4,23−25 In general chemical sensing (e.g., distinguishing between oxygenated and deoxygenated blood in imaging), this error may be acceptable but for absolute quantitation (or pH measurements, where the deviation will be magnified by the log scale) this volume change must be accounted for in order to yield highquality concentration measurements. Depth-Resolved Measurements, Photoacoustic Imaging, and ΔVrxn. Photoacoustic measurements can permit depth-resolved measurements, as well as 3D chemical measureB

DOI: 10.1021/ac503515k Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

2.5 mJ/pulse max), firing at 1 Hz to pump a tunable dye laser (GL-301, Photon Technology International, Birmingham, NJ). A small portion of the beam is diverted to an optical trigger; this serves as a trigger signal to the recording electronics. Laser intensity and beam shape are controlled by passing it through both a firmly clamped attenuation cell and a round aperture (≈1−2 mm). The beam then passes through a 50:50 cube beamsplitter, with each resulting beam directed through a photoacoustic cell (referred to as either the Sample or Reflected positions). Our photoacoustic cells consist of a 10 mm quartz cuvette spring-loaded against a custom-built microphone (500 MHz) described in ref 35, with a thin layer of vacuum grease used as an acoustic couplant between the cuvette and microphone. In the Sample position, either NY or a PA reference solution is measured; in the Reflected position, an aqueous sample of absorptive dye (typically Brilliant Black Bn, BBBn) allows the PA cell to measure Ep and to account for shot-to-shot intensity variations. Signals from both positions are amplified 60 dB (5660B, Panametrics-NDT, Olympus Corp., Waltham, MA), and 64 shots are averaged and digitized on a digital oscilloscope (TDS 2024B, Tektronix, Inc., Beaverton, OR). The transducer waveforms are analyzed by Microsoft Excel spreadsheets, developed in-house. Spectral Measurements. Absorbance spectra were collected with either a Shimadzu UV-2450 (Shimadzu Scientific Instruments, Columbia, MD) or Cary 50 (Agilent Technologies, Santa Clara, CA) spectrometer, using a 0.1 nm spectral bandwidth. Individual absorbance measurements were similarly recorded, with dwell times sufficient to ensure the stability of the measurement. After measurement, the solutions were kept in sealed containers for days (sometimes with the UV−vis cuvette); the glassware was drained, rinsed with methanol, and the rinse was checked spectrally to determine if the dye adhered to the glassware. No significant solute adhesion was observed on either the glassware or cuvettes. In order to determine the pKa for our NY samples, a series of solutions at various pHs were prepared following the same approach as in our PA samples and their spectra were recorded. Photoacoustic Measurements. The detector response parameter (κ′) was determined within each experiment by measuring the PA response of one or more standards ( f obs h,standard = 1.00 and known to convert absorbed light to heat faster than the instrument’s time-resolution) with high absorptivity at the chosen wavelength. The signals for NY were compared to this standard, allowing the measurement of f obs h,NY as in eq 1. In each experiment, 7 different laser intensities (Ep, controlled by changing attenuator solutions) were used, and the reflectedposition photoacoustic signal (Sreflected) served as a shot-to-shot measure of incident laser intensity. Alternating samples of NY and a PA standard were each measured twice, followed by a baseline measurement using the buffer solution, and the results were checked for any additional effects (e.g., evaporation, dye aggregation). PA measurements were conducted at 465 nm (λmax for the acidic form), 590 nm (λmax basic), and 532 nm (λintermediate, corresponding to the commonly used Nd:YAG second harmonic). The dye laser’s wavelength was monitored by a CCD array spectrometer (Quest X, B&W Tek, Inc., Newark, DE) and benchmarked against a DPSS Nd:YAG laser at 532 nm. For measurements conducted at 22 °C, no temperature control was needed to keep samples within ±0.5 °C. For temperature-controlled experiments (see ΔVrxn section in the Results and Discussion), a brass-block (1.5 kg) PA

are indistinguishable, “slow” events (τ > 100 μs for our detector) occur after the analysis time window, and the delay time between the firing of the laser and the arrival of the PA signal reports the depth of the feature. The peak width of the PA wave reports the thickness of the region that absorbed light while containing photoacoustically active molecules. In the absence of time-resolved photoacoustic behaviors, these measurements are much like traditional ultrasonic imaging or sonar-style methods. However, if the lifetime of the excited state is such that heat is deposited to the solvent on a time scale between that of the “fast” and “slow” cases, a more complex waveform is generated.5 Time-resolved PA calorimetry (TR-PAC) exploits this by measuring the superpositioned waves caused by distinct processes with different lifetimes, then disentangling the superpositioned waves to simultaneously measure the energetics and kinetics of each. TR-PAC waveforms are shifted to longer times; the peak maximum arrives later and has a broader peak shape when compared to the “fast” case. This causes the region with PA reporter molecules to appear thicker, further away from the microphone, and less-dyed from the microphone than it actually is, complicating the interpretation of both qualitative and quantitative imaging applications. An effective PA reporter molecule should be biologically/ environmentally compatible, yield large signals at low concentrations, generate no reactive species, and provide useful information either for object location or about the local chemical environment. NY was selected as a candidate dye because (a) it has a high molar absorptivity in both acidic and basic forms, allowing efficient detection with both low concentrations and incident light levels; (b) it operates as an indicator dye in a pH range that is relevant for a number of turbid sample types, such as living tissue, groundwater, and aqueous chemical processes; (c) it is currently used to visualize the human lymphatic system during surgeries, where it binds to serum proteins and is known to be biocompatible.34 Prior to this work it was not known whether the dye would be photoacoustically well-behaved. A well-behaved dye should (i) deposit all absorbed energy as heat faster than the temporal resolution of the detector; (ii) remain photostable over a useful number of laser pulses, without photodegradation into toxic species or generation of reactive products (e.g., singlet oxygen); (iii) exhibit simple heat-relaxation photochemistry, so that routine analysis of pH or spatial imaging is straightforward; (iv) remain soluble in water over a range of experimental conditions. In this work, we investigate these photoacoustic behaviors for NY molecules in both acidic and basic environments and for NY when bound to small molecules (poly lysine) or large proteins (bovine serum albumin). Though we anticipate a wide range of applications for PA reporter molecules, we recognize that biomedical applications will commonly have more stringent criteria than needed in much analytical work (e.g., singlet oxygen formation is problematic in living tissue, though it may be acceptable in groundwater analysis); it is not, however, our aim to deemphasize the broader applications of PA measurements in turbid media.



EXPERIMENTAL SECTION Photoacoustic Calorimeter. A schematic and detailed account of our instrument is provided in the Supporting Information. Our instrument uses a nitrogen laser (LN1000, PRA International Inc., London, Canada; 337.1 nm, τ = 800 ps, C

DOI: 10.1021/ac503515k Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry sample holder with embedded flow-through water channels was used with a Neslab RTE 7 recirculating water bath. All cuvette temperatures were monitored via a thermocouple and reader (Omega TrueRMS Supermeter, Newport Corp., Stamford, CT), with the thermocouple positioned in a corner of the cuvette away from the beam path and opposite to the microphone. Preparation of NY Samples. A stock solution of NY (disodium salt, 85% dye content, Sigma-Aldrich, St. Louis, MO; 0.1 g NY/100 mL deionized water, 1.6 × 10−3 M) was used to prepare all working solutions, ranging from pH 4.65−8.74. Details regarding sample purity can be found in the Supporting Information; concentrations are presented here both in molarity (accounting for 85% w/w nominal purity) and g/ mL (total mass NY and impurities, from bottle) for clarity. The NY/buffer solutions had absorbances of 0.000−0.200, and NY concentrations of 6.6 × 10−6 M (4.22 × 10−6 g of 85% NY/ mL). Spectra at pH 4.65 and 8.74 (pKa = 6.70, vide inf ra) lacked additional peaks, and the spectra of solutions between these pHs showed only one isosbestic point. We note that in remote-sensing applications, concentration may be locally variable but can be determined by an absorbance (or photoacoustic) measurement at the isosbestic point, where absorbance is related to the formal concentration of the solute by the Beer−Lambert equation. Additionally, if a photoacoustically modified version of the Henderson−Hasselbalch equation is used to determine local pH, and if f obs h is the same for both acid and base forms, the ratio of signals and pKa alone will report pH without requiring local concentrations. The pH of the NY/buffer working solutions was measured both potentiometrically (Accumet AR25 pH meter with Orion 910600 pH probe, Fisher Scientific, Waltham, MA) and spectrally. In order to ascertain whether ionic strength would affect either spectral or PA behaviors over these pH ranges, KCl-spiked solutions were measured photoacoustically and compared to unspiked signals; the spectra were unchanged as were the PA wave shapes, peak heights, and transit times. As such, in all subsequent measurements the formal concentration of the solution, not the ionic strength, was controlled. Preparation of PA Standards. A number of PA standards have been reported;36,37 in our work, we typically used New Coccine (NC; trisodium salt, 75% dye content, Sigma-Aldrich, St. Louis, MO) or Brilliant Black Bn (BBBn; tetrasodium salt, 60% dye content, Sigma-Aldrich, St. Louis, MO) at 465 nm, NC or BBBn at 532 nm, and BBBn at 590 nm. Stock solutions of our standards were diluted until their absorbance was wellmatched (within 0.005 au and usually within 0.002 au) to the NY/buffer solutions. This absorbance-matching served to ensure that the two solutions had similar light penetrations; dramatically different absorbances would result in different acoustic intensities and transit times, resulting in deviant PA intensities and waveforms that could be mistaken for timedependent signals. In our experiments at 465 and 532 nm, the absorbance of each set of solutions was typically below 0.150 au. At 590 nm, the light output from our dye laser was weak, and in order to collect high-quality PA signals we increased the absorbance as needed ( 1.00 is not known. It is important to note that under conditions where serum proteins are present, NY will over-report the basicity of the solution since the serum protein-bound NY (blue, deprotonated form) appears photoacoustically identical to NY in its deprotonated unbound form. In these environments, two measurement approaches can be attempted: PA response can be used to provide pH gradient measurements or an alternate modality could be used to measure the pH in one location, and the PA response can be calibrated to this measurement. pH Measurements in Turbid Media. PA data was collected for 4 calibration solutions of NY in reconstituted powdered milk. This solution was sufficiently turbid that a 10.0 mm cuvette was visually opaque while containing the solution, and when illuminated with our laser, a significant portion of the light directly illuminated the microphone, resulting in an immediate signal that was unrelated to NY in solution (Figure 4). Analysis of the set of waveforms showed that this initial

those at 532 and 590 nm, the absence of a trend between these measurements suggests that this is an artifact in the data and that it can be assumed that f obs h,NY = 1.00 for these wavelengths in NY’s acidic and basic forms. ΔVrxn. In order to test for possible PA signals resulting from photoinitiated isomerizations or reactions, as in eqs 2 and 3, we measured f obs h,NY for the highest-energy light used here (465 nm) at pH 4.6 and 9.1 and varied the solvent expansivity by controlling the temperature of the solution. Since χs,H2O ≈ 0 at 4 °C, any signal observed at this temperature would be entirely attributable to ΔVrxn; the signal was indistinguishable from baseline under these conditions. Measurements done at the higher temperatures gave f hobs ≈ 1.00, and when both temperatures were analyzed by eq 3, no ΔVrxn was found for this system. Both measurement approaches confirmed that no chemical volume change occurs for either form of NY, and PA measurements of concentration and pH may be carried out without concern for this potential complication. NY-BSA and NY-poly-Lys Measurements. NY is known to bind efficiently to serum proteins and has been investigated by Tsopelas and Sutton as a vital stain for lymphatic mapping of the sentinel lymph nodes.34 In their work, they determined that NY has an optimal spacing between two of its sulfonate groups, allowing it to undergo a sulfonation reaction between the amino acids with terminal side-chain amines and the NY’s sulfonic acid groups,34 leaving NY as a blue chromophore. In order to compare the PA behavior of free-NY and its proteinbound form, we compared measurements of f obs h,NY for NY in buffered solution (free dye) and in buffered BSA (“large” molecules). In order to determine if heat deposition might lead to different processes when bound to large molecules (e.g., some energy might be used in melting a portion of the protein), we also measured solutions of NY in buffered poly-Lys to compare the effect when NY is bound to small and large molecules. These experiments were done under conditions similar to our other measurements (465 nm, 22 °C, pH 7.65 phosphate buffer), and all solutions were measured together in each experiment to allow direct comparisons between these solution compositions. Data were analyzed in various combinations to examine the PA behavior, and determine f obs h for NC across all three solution compositions: NY across the three solution compositions; and between NC/NY for each solvent composition (Table 2).

Figure 4. PA waves in turbid media. Averaged waves from 64 laser shots at a single attenuation; uncorrected for Ep. Integration limits shown as vertical dotted lines.

a Table 2. f obs h for NY and NC

microphone absorption signal could generate a meaningful calibration curve; the intensity of the microphone’s PA response was attenuated as the absorbance of the NY in solution increased. However, to better demonstrate applicability in both highly- and moderately turbid media, we instead integrated a later portion of the waveform to use as Sobs. Otherwise, we used recording and analysis methods similar to those in previous sections. In order to determine the pH of the unknown sample, the normalized Sobs were plotted against the fraction of molecules in the deprotonated form in order to build a calibration curve, shown in Figure 5. This permitted the determination of the fraction of NY in the deprotonated form and the pH of the solution. The unknown sample was found to be 46.6% deprotonated, which corresponds to pH = 6.64, which is closely matched with the potentiometric pH reading of 6.62.

f obs h solvent system

NC/buffer vs NC

NY/buffer vs NY

NY vs NC

buffer poly-Lys/buffer/buffer BSA/buffer

1.00b 0.99 ± 0.02 1.14 ± 0.02

1.00b 1.01 ± 0.02 1.05 ± 0.02

0.98 ± 0.02 0.98 ± 0.02 0.90 ± 0.02

a In buffer, buffer and poly-lysine, and buffer and bovine serum albumin. bPA standard ( f h = 1.00); NY is demonstrated here to be f h = 1.00 (vide supra).

Measurements of f obs h involving NC-BSA (as either sample or reference) gave surprising results; when NC/buffer was used as a standard, f obs h,NC−BSA/buffer for this system was ≈15% higher than the standard; similarly, when NC-BSA/buffer was used as a standard against NY-BSA/buffer, f obs h,NY−BSA appeared to be 10% lower than the standard.Our initial analysis was that the NC F

DOI: 10.1021/ac503515k Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry





CONCLUSION NY has been found to be an ideally behaved photoacoustic reporter molecule for pH measurements in turbid media and to have well-behaved photoacoustic behavior when bound to serum proteins. It is highly efficient at converting light into heat (100% conversion in