Analysis of the stability of amino acids derivatized with naphthalene-2,3-dicarboxaldehyde using high-performance liquid chromatography and mass spectrometry

ANALYTICAL BIOCHEMISTRY Analytical Biochemistry 322 (2003) 68–78 www.elsevier.com/locate/yabio

Analysis of the stability of amino acids derivatized with naphthalene-2,3-dicarboxaldehyde using high-performance liquid chromatography and mass spectrometry Drew P. Manica, Julie A. Lapos, A. Daniel Jones, and Andrew G. Ewing* Department of Chemistry, The Pennsylvania State University, 152 Davey Laboratory, University Park, PA 16802, USA Received 11 May 2003

Abstract The stability of amino acids derivatized with naphthalene-2,3-dicarboxaldehyde (NDA) was investigated using a combination of high-performance liquid chromatography, solid-phase extraction, photodiode array spectrophotometric detection, and mass spectrometric (MS) characterization. The degradation of amino acid derivatives, generated using b-mercaptoethanol as a nucleophile, was characterized under a variety of environmental influences, with a focus on understanding the degradation kinetics and identifying the degradation products. The predominant degradation product observed under most reaction conditions was the nonfluorescent lactam form of the originally fluorescent isoindole derivative. First, the time-dependent degradation of the isoindole derivative L -serine–NDA–b-mercaptoethanol was found to follow pseudo-first order kinetics with a half-life of 2.0 min at pH 9.2 and room temperature. The isoindole derivative was observed to react further with methanol to form a more stable fluorescent methoxy–isoindole, shedding new light on the basis for enhanced stability of these derivatives in methanol. Tandem mass spectrometry (MS/MS) experiments were used to demonstrate unimolecular degradation of the protonated isoindole in the absence of solvent or atmosphere, suggesting an intramolecular reaction mechanism involving the hydroxyethylthio group. Finally, in photobleaching studies, NDA derivatives rapidly degraded into a variety of products within the first 2 min of photobleaching versus timed controls, with the predominant product being the lactam. These results suggest that the degradation pathway for NDA derivatives is similar to the previously reported pathway for o-phthalaldehyde derivatives and clearly identifies the reaction and degradation products under a variety of conditions. Ó 2003 Elsevier Inc. All rights reserved. Keywords: Amino acid derivative stability; Fluorescence labeling; Liquid chromatography; Mass spectrometry; Photobleaching

The importance of monitoring amino acids, endogenous primary amines, and peptides in biological processes has encouraged the expansion of analytical methodologies for observing amines. In most cases, selective detection of these compounds is achieved only after attaching some type of label to the amine moiety. Increased interest in amino acid and peptide analysis has led to the introduction of a wide range of agents for labeling amines, including ninhydrin [1,2], fluorescamine [2], dansyl chloride [3], 7-fluoro-4-nitro-

* Corresponding author. Fax: 1-814-863-8081. E-mail address: [email protected] (A.G. Ewing).

0003-2697/$ - see front matter Ó 2003 Elsevier Inc. All rights reserved. doi:10.1016/j.ab.2003.07.002

benz-2,1,3-oxodiazole (NBD)1 [4], o-phthalaldehyde (OPA) [5], and naphthalene-2,3-dicarboxaldehyde (NDA) [6,7]. The requirements of the derivatization reaction and the stability of the derivative are deciding factors in choosing a derivative. Therefore, in addition to optimizing derivatization parameters—such as stoichiometry, reaction time, temperature, and dilution effects—a thorough understanding of factors influencing derivative stability is critical.

1 Abbreviations used: NBD, 7-fluoro-4-nitrobenz-2,1,3-oxodiazole; OPA, o-phthalaldehyde; NDA, naphthalene-2,3-dicarboxaldehyde; CN, cyanide; BME, b-mercaptoethanol; SPE, solid-phase extraction; DAD, diode array detector.

D.P. Manica et al. / Analytical Biochemistry 322 (2003) 68–78

The first automated amino acid analyzer was described in 1958 by Spackman et al. [1] and used ninhydrin for post-column derivatization with absorbance detection. Throughout the years, dramatic improvements in instrumentation and synthesis have produced a myriad of tags available for various detection schemes, perhaps the most widespread of which were developed for fluorescence, amperometric, and radiometric detection. Development of the family of isoindole-forming derivatives—composed of OPA and NDA—has been a critical advance in the analysis of amino acids and other biogenic amines. Both reagents react with primary amines in the presence of various nucleophiles including thiols [8], cyanide (CN) [7], and sulfite [9] to form fluorescent and electroactive derivatives [10–13]. One important advantage of OPA and NDA is that they do not intrinsically fluoresce [5], allowing sensitive detection of derivatives without interference from the derivatizing reagent. The drawbacks of OPA are its selectivity for only primary amines, low fluorescence quantum yields for peptides, excitation wavelength requirement (UV), and the relative instability of the derivative, which degrades to nonfluorescent products in a matter of minutes [11]. To avoid some of the limitations associated with OPA, NDA was developed [6,7]. NDA, like OPA, reacts with primary amines in the presence of a nucleophile to create fluorescent isoindole derivatives. Derivatives of NDA are typically more stable than OPA derivatives, are excited in the visible spectrum, exhibit higher fluorescence quantum yields (especially with peptides), and are electroactive [10–12]. Derivatization of primary amines with NDA has been demonstrated for separation applications using either precolumn [14], postcolumn [15], or on-column [16,17] derivatization. Understanding derivative stability can be especially important for separation and imaging applications in which fluorescence detection is used. For imaging applications, the derivatized sample may be detrimentally exposed to the excitation beam for extended periods of time, often photobleaching the derivative. For separation applications, individual analytes spend different amounts of time in the detection region, thus the excitation beam may inconsistently alter detection sensitivity due to photobleaching. In contrast, some analytical methods, such as optically gated sample introduction, take advantage of the instability of the fluorophore to create a novel sample introduction method [18,19]. Additionally, Orwar et al. [20] demonstrated that knowledge of the photochemical properties of an analyte assists in peak characterization for capillary electrophoresis. Despite the importance of understanding the stability (or the degradation) of derivatives and the applications that benefit from this property, there is surprisingly little about the degradation products of NDA derivatives and the factors that influence degradation rates reported.

69

Several degradation mechanisms have been discussed in the literature for derivatives of OPA and b-mercaptoethanol (BME) [8,20–24]. Simons and Johnson [24] identified the decomposition product of n-propylamine– NDA–BME and postulated a mechanism for equimolar quantities of primary amine, NDA, and BME. Nakamura et al. [22] reported on the effect of the primary structure of the amine on derivative stability, with more sterically hindered amines providing greater stability. Nakamura et al. [22] also proposed that excess OPA served to destabilize the derivative by directly attacking the isoindole ring, with OPA serving as a dienophile or even a nucleophile. These mechanisms were rejected by Stobaugh et al. [8], who proposed an alternate mechanism involving the isoindole forming a hemiacetal, followed by rearrangement and hydrolysis to form the nonfluorescent product. Clearly, there has been much discussion about the details of the conversion of OPAlabeled amines to nonfluorescent products; however, to our knowledge, the discussion has not carried over into the degradation of NDA derivatives. This paper focuses on the stability and degradation of amino acids derivatized with NDA and BME, and comparisons with OPA–BME and NDA–CN derivatives are made. High-performance liquid chromatography (HPLC) and solid-phase extraction (SPE) were used to separate samples before spectrophotometric and mass spectrometric (MS) characterization. Degradation mechanisms are proposed with a focus on identifying and verifying the degradation products. Such products were studied following a variety of degradation (or reaction) conditions: time-dependent degradation involving hydrolysis, reaction in the presence of methanol, degradation in the absence of solvent, and photodestruction. Across these conditions, the primary degradation product is a similar if not identical nonfluorescent product, and it is speculated that there are several conditions that can accelerate its formation. From an analytical perspective, it is also noteworthy that this study represents the first demonstration of the use of soft ionization mass spectrometry to characterize NDA–BME derivatives.

Materials and methods Reagents Reagents were obtained from Sigma Chemical Co. (St. Louis, MO) unless otherwise noted. Borate buffer (10 mM, pH 9.2) was prepared for use in the derivatization reaction. Amino acids (15 mM) and potassium cyanide (80 mM) solutions were prepared in borate buffer. Stock solutions of NDA (55 mM, Fluka, Milwaukee, WI) were prepared in either methanol or acetonitrile as indicated. BME stock (J.T. Baker,

70

D.P. Manica et al. / Analytical Biochemistry 322 (2003) 68–78

Phillipsburg, NJ) was mixed in a 1:1 molar ratio with the NDA solution for labeling with NDA–BME.

solvent delivery of acetonitrile (100 lL/min) from a LC10ADvp pump (Shimadzu, Columbia, MD).

Derivatization procedures Results and discussion Precolumn derivatization reactions were performed manually as follows unless otherwise noted. Reactions were carried out in amber vials to prevent detrimental exposure to light and were performed at room temperature. For labeling with NDA–CN, amino acid (5 lL) and CN (5 lL) were added to 1 mL borate buffer, and NDA (10 lL) was added last, at which point the reaction was carefully timed to ensure consistent yields. For labeling with NDA–BME, amino acid (5 lL) was added to 500 lL borate buffer, and NDA–BME (10 lL) was added last, while timing was initiated. Separation and mass spectrometry procedures Following reaction, the sample was either injected onto the HPLC column or drawn through a solid-phase extraction column (as indicated in the text) prior to MS detection. The separation and detection system consisted of an 1100 Series quaternary pump and diode array detector (DAD) (Hewlett–Packard, Palo Alto, CA) with a 60 nL flow cell, coupled in series to a Mariner time-of-flight mass spectrometer (PerSeptive Biosystems, Framingham, MA) operated using positive mode electrospray ionization. A polymer-based reversephase HPLC column (PRP1 column, 50
2.1 mm, 3-lm particles, Hamilton Co., Reno, NV) was employed for chromatographic analysis. The solvent was delivered at a flow rate of 0.25 mL/min with gradient composition as follows: solvent A, water; B, acetonitrile (75% A (0– 5 min) followed by a step to 100% B at 5 min, held at 100% B until 15 min). Alternately, extraction of the derivative from the aqueous derivatizing reagent was accomplished using a 1-mL SupelClean C18 SPE column (Supelco, Bellefonte, PA) followed by a direct flow injection introduction into the mass spectrometer. The derivatized sample was loaded onto the SPE column, and aqueous eluent was collected by applying an aspirator vacuum to a 125-mL side-arm flask. The SPE column was then rinsed with 500 lL deionized water to remove buffer salts and nonretained reagents, and the aqueous eluent was discarded. The fluorescent derivative was collected into a clean flask by loading 1 mL of methanol or acetonitrile (as indicated) onto the SPE column and applying the vacuum. Occasionally, solid-phase extraction and chromatographic separation were both performed. For tandem MS experiments, the fluorescent derivatives were first extracted using a SPE column as described above, and 10 lL was injected onto a Quattro II tandem quadrupole mass spectrometer (Micromass, Beverly, MA) using positive mode electrospray ionization and

The less stable BME derivatives were the focus of this study because of their poorer stability and because their rapid rates of formation offer advantages for fast capillary electrophoretic analyses [15,25,26]. The degradation of NDA–BME-labeled amino acids was investigated under a variety of conditions, and NDA– CN-labeled amino acids were used for comparison. The following conditions were investigated for their effect on derivative stability: time-dependent decomposition, the presence of methanol, the absence of solvent, and photobleaching. For each environmental influence, the molecular decomposition most likely follows multiple pathways to form multiple products; however, for simplicity, only the most abundant degradation products were investigated. Time-dependent decomposition The phenomenon of time-dependent loss of fluorescence has been well documented for some isoindoles in the literature [7,11]; however, the degradation mechanisms and products of NDA–BME-labeled amines have not. It is possible that short derivative lifetimes have hindered efforts to isolate them. In this study, HPLCMS was utilized to capture the decay of the L -serine derivative (Ser–NDA–BME) and the simultaneous formation of degradation product at different incubation times. Fig. 1A illustrates the resolution and detection of the fluorescent isoindole product ([M + H]þ at m=z 332) and the primary nonfluorescent degradation product ([M + H]þ at m=z 272), most likely the lactam structure shown. The derivatization reaction mixture characterized in Fig. 1 was allowed to react for approximately 2 min before solid-phase extraction of the derivatives with acetonitrile. By extracting the derivatives from the aqueous reaction matrix and redissolving in acetonitrile, it was observed that the lifetime of the fluorescent derivative was greatly extended. This effect is attributed to suppression of hydrolysis of the isoindole in the aprotic acetonitrile as will be explained later. Fig. 1B shows the background-subtracted mass spectrum accumulated over the elution times for the isoindole and lactam products, emphasizing that these are the predominant compounds present at their retention times. The broadly increasing profile of the reconstructed ion chromatogram for m=z 272 suggests that it may be the product of on-column decomposition, as would be expected for a rapid hydrolysis reaction in relatively aqueous buffer or mobile phase. To verify that the m=z 272 peak is the nonfluorescent product, an absorbance chromatogram

D.P. Manica et al. / Analytical Biochemistry 322 (2003) 68–78

A

71

% Intensity % Intensity

Mass 332.1±0.5 100

100

T0.9

2308

50 0

0

3

6

9

12

0 15

Mass 272.1±0.5 T3.3

50 0

1389

T0.9 0

3

6

9

12

0 15

Retention Time (Min)

% Intensity

B

100 179.0 90 80 70 60 50 40 30 20 198.0 10 0 165.0 225.8

332.1

S

OH

O

1717

N CHCOOH

CH2OH

N CHCOOH CH2OH

272.1

333.1 273.0

286.6

347.4

408.2

0 469.0

Mass (m/z) Fig. 1. Separation and detection of Ser–NDA–BME and its primary degradation product, 2 min after mixing. (A) Reconstructed ion chromatograms: (top) fluorescent isoindole derivative (m=z 332), (bottom) primary nonfluorescent degradation product (m=z 272). The sloping profile of m=z 272 suggests that its formation is the result of on-column decomposition. (B) Combined mass spectrum (minus background) at the elution times of the isoindole and lactam products.

was collected simultaneously, as shown in Fig. 2. Fluorescent NDA derivatives are known to absorb strongly near 460 nm [11,14] which is the region used for excitation. Therefore, absorbance in the 460-nm region (Fig. 2A) is used as one indication of the fluorescent derivative, while absorbance at 280 nm (Fig. 2B) is characteristic of various aromatic derivatives of NDA. Because the diode array detector is located upstream from the mass spectrometer, the absorbance retention times slightly precede those of the ion chromatograms. Separate LC/MS and LC/DAD analyses to ensure that diode array detection does not produce significant amounts of photodecomposition products have been performed. A comparison of Fig. 2 with Fig. 1A clearly indicates that the 460-nm absorber corresponds to the isoindole derivative and not to the lactam degradation product, while both products and NDA absorb at 280 nm. These data provide solid evidence for the degradation of an NDA–BME derivative to an identifiable nonfluorescent product and suggest a predictable behavior in different solvent environments, as will be explained next. Characterization of the time-dependent degradation of Ser–NDA–BME was performed using mass spectrometry by monitoring the decrease in peaks at m=z

332, which are characteristic of isoindole derivative, and the increase in peaks at m=z 272, corresponding to nonfluorescent lactam, over the lifetime of the degradation event. The peak height ratios (n ¼ 3) of the isoindole form vs the total isoindole plus lactam I332 =ðI332 þ I272 Þ were monitored for several different reaction times and compared with the ratios of peak heights of the lactam form vs the total (I272 =½I272 þ I332 ), as shown in Fig. 3A. The data near 0 min represent a brief mixing time of approximately 5 s, followed by immediate solid-phase extraction. For each point thereafter, the time axis represents the exact amount of time between addition of NDA and solid-phase extraction. The decrease in the isoindole ratio and concomitant increase in the lactam ratio supports the assertion that the isoindole undergoes conversion to the lactam product. The proposed mechanism of the degradation reaction in aqueous solvent is shown in Fig. 3B, which depicts the nucleophilic attack of a hydroxide ion, the donation of a proton by water, and the release of the BME group. Possible intermediates include an isoindole with a hydroxyl moiety, which rapidly tautomerizes to a lactam (at m=z 272). The lactam apparently lacks the conjugation required for fluorescence. Identical decomposition pathways have been tested and observed for

72

D.P. Manica et al. / Analytical Biochemistry 322 (2003) 68–78

Absorbance at 460 nm (mAU)

0.497

A 5

isoindole

4 3 2 1 0 0

2

4

8

6

10

12

14

12

14

Time (min)

7.797

500

400

isoindole

NDA

300

0 0

2

4

6

11.575

9. 208

5.399

3.070

100

6.137 6.611

lactam

0.515

200

0.937

Absorbance at 280 nm (mAU)

B

8

10

Time (min) Fig. 2. Background-subtracted absorbance chromatograms at (A) 460 nm and (B) 280 nm for Ser–NDA–BME derivatization reaction. Absorbance data were obtained simultaneously with the MS data in Fig. 1. The isoindole derivative (0.515 min) and the degradation product (maximum at 3.070 min) both absorb at 280 nm, indicating aromaticity. Only the isoindole absorbs at 460 nm, near the fluorescence excitation wavelength of NDA derivatives. Background subtraction was accomplished by subtracting the absorbance at 600 nm.

Val–NDA–BME and Phe–NDA–BME (data not shown), and it is expected that many other primary amines behave similarly. The mechanism and degradation products proposed herein are consistent with those proposed for the decomposition of OPA derivatives under similar conditions [8]. The kinetics of the degradation event has been characterized further by monitoring the absorbance at 460 nm of Ser–NDA–BME at various reaction times. A plot of the natural logarithm of the absorbance vs reaction time is shown in Fig. 4. The high correlation coefficient (0.9998, n ¼ 3) for this relationship indicates that the derivative undergoes a pseudo- first order degradation reaction to form nonfluorescent products. The slope (m) of the line is related to the half-life (s) of a first order decay according to the equation, s ¼ ðln 2Þ=m. Hence, under these conditions, the fluorescent half-life, s, is calculated to be 2.0 min for Ser–NDA–BME. Additional observations of the labeling reaction further elucidate the stability of NDA derivatives. The nonfluorescent lactam product appears to be highly stable over time in aqueous or organic solvents.

Ser–NDA–BME samples that are allowed to react for as long as 60 min show complete disappearance of the isoindole peak (m=z 332), with no reduction in the m=z 272 peak corresponding to the lactam. In contrast, Ser– NDA–CN samples that are allowed to react for as long as 60 min show no reduction in cyanobenz[f]isoindole signal. The dramatic difference in the stability of NDA– CN and NDA–BME derivatives is most likely due to the fact that BME is a better leaving group than CN. Rapid formation of fluorescent derivatives makes BME the optimal choice for applications that require very rapid reaction times such as on-column or postcolumn labeling; however, the stability achieved with CN makes it the preferred nucleophile for precolumn labeling despite lengthier reaction times. Formation of methoxybenz[f]indole derivatives from reaction with methanol Methanol is a common solvent used to dissolve NDA and widely used for chromatographic separations and solid-phase extraction. The use of methanol in NDA

D.P. Manica et al. / Analytical Biochemistry 322 (2003) 68–78

73

Fig. 3. Monitoring the time-dependent conversion of the fluorescent isoindole form of Ser–NDA–BME to the nonfluorescent lactam form. (A) Plot of the ratio (r) of peak height of isoindole over total (lactam + isoindole) vs reaction time, and the ratio () of peak height of lactam over total (lactam + isoindole) vs reaction time (n ¼ 3). (B) Proposed mechanism for the degradation of Ser–NDA–BME.

derivatization reactions has been reported to have a stabilizing effect on OPA–BME derivatives [23]; however, the interaction of methanol with NDA–BME has not been extensively characterized, and its effect on the derivative is debated. Therefore, some clarification of the interaction between methanol and isoindole derivatives is necessary. The effect of methanol on Ser– NDA–BME has been studied by comparing samples of derivative extracted in methanol vs acetonitrile. In each case, the labeling reaction has been carried out as previously described with a reaction time of 1 min. Solidphase extraction with either methanol or acetonitrile has been carried out and loop injections onto the mass spectrometer have been made. The predominant product remaining after extraction in acetonitrile (Fig. 5A) is the expected thio-isoindole (m=z 332), while the product after extraction in methanol (Fig. 5B) is m=z 286, which

corresponds to a methoxybenz[f]isoindole. (Because the NDA stock solution is made with methanol in both cases, some m=z 286 is present after extraction in acetonitrile.) Based on its conjugation, the methoxy–isoindole is expected to fluoresce. Prior reports have claimed that methanol serves to enhance the stability of OPA derivatives [23]; however, based on the data presented here, this idea may require modification because a unique derivative (methoxybenz[f]isoindole) is formed. As such, methanol enhances the fluorescence lifetime of the derivatized amine; however, it may be more accurate to claim that, although the product is labile in methanol, such a reaction yields a new fluorophore with greater stability owing to slower rates of hydrolytic displacement. Furthermore, in light of the susceptibility of some isoindoles to hydrolysis, it can be suggested that the sustained fluorescence after methanol extraction may be

74

D.P. Manica et al. / Analytical Biochemistry 322 (2003) 68–78 7

ln (Abs[460 nm] - Abs[600 nm])

6.5

y = -0.3501x + 6.1494 2 R = 0.9998

6 5.5 5 4.5 4 3.5 3 -2

0

2

4

6

8

Reaction time (min)

Fig. 4. Linear plot of the natural log of the integrated absorbance at 460 nm vs time (n ¼ 3) indicates a pseudo-first order decay of the fluorescent isoindole to nonfluorescent products. The slope was used to calculate the fluorescent half-life of Ser–NDA–BME under conditions used for derivatization (s ¼ 2:0 min).

% Intensity

A

100 90 80 70 60 50 40 30 20 10 0 165

126 OMe OH

S

332.1

N CHCOOH CH2OH

N CHCOOH CH2OH

207.0 185.0

286.1 226

287

377.1

333.1

348

433.2 409

0 470

Mass (m/z)

% Intensity

B

100 90 80 70 60 50 40 30 20 10 0 165

286.1

126 OMe N CHCOOH CH2OH

345.1 287.1 226

287

348

409

0 470

Mass (m/z) Fig. 5. Background-subtracted spectra comparing the derivatives remaining after solid-phase extraction in (A) 100% acetonitrile and (B) 100% methanol. The thiobenz[f]isoindole (m=z 332) and methoxybenz[f]isoindole (m=z 286) are expected to exhibit similar fluorescence yields, although dissimilar fluorescence lifetimes. In both cases, the NDA was dissolved in methanol prior to reaction, leading to a small amount of methoxybenz[f]isoindole in (A).

D.P. Manica et al. / Analytical Biochemistry 322 (2003) 68–78

attributed in part to the elimination of water and salts, as suggested above. Degradation in the absence of solvent The examples of degradation described previously have occurred in the presence of solvents such as water, methanol, acetonitrile, and oxygen-containing atmosphere. To understand the stability of NDA–BME derivatives without such environmental influences, tandem MS/MS experiments have been performed. Ser– NDA–BME samples have been prepared as described previously and extracted in acetonitrile before MS/MS analysis. Examination of the daughter peaks from the parent m=z 332 (data not shown) reveals a strong peak at m=z 272. This loss of 60 Da yields the same mass as the protonated lactam and corresponds to loss of C2 H4 S. This decomposition may be best explained by invoking an intramolecular rearrangement of the hydroxyethylthioether group to a mercaptoethylether, followed by elimination of ethylene sulfide. A similar formation of ethylene sulfide has been reported by Simons and Johnson [27] as a spontaneous degradation product of OPA–BME derivatives in solution. The facile gas-phase rearrangement observed in this study suggests that an intramolecular sulfur-to-oxygen rearrangement could contribute to the instability of thio-isoindoles independent of solvent. Photodestruction Photobleaching or photodestruction are terms broadly given to the chemical alteration of a molecule in the presence of light that irreversibly eliminates fluorescence of that molecule. ‘‘Photobleaching’’ does not refer to a specific mechanism and is generally agreed to be a complex process involving a variety of intermolecular and intramolecular interactions. It is sensitive to the specific molecule, the microenvironment, and the type of irradiation [28]. Photobleaching is involved to a greater or lesser extent in any fluorescence-based technique. Moreover, many imaging applications [28,29] and separation applications [18–20,30] directly rely upon photobleaching for specific modes of analysis. Despite the widespread awareness and use of the phenomenon, very little is known about the photodestruction of most fluorophores and, while mechanisms have been suggested for specific fluorophores under certain conditions [28,29], more is known about the conditions that cause photobleaching than the actual products of photobleaching. In this study, mass spectrometric detection has been used to monitor the photodestruction of fluorescent NDA–BME derivatives. Fresh Ser–NDA–BME was mixed (515 lL total volume) in a clear vial as previously described. The vial was exposed to the excitation beam

75

for a selected amount of time before solid-phase extraction in acetonitrile and injection onto the mass spectrometer. A control was run separately in an amber vial and allowed to sit for the selected duration before extraction and analysis. Comparative mass spectra from this procedure are shown in Fig. 6 for an experiment where the duration was 2 min for (A) control and (B) photobleached samples. One of the predominant species detected after photobleaching is the nonfluorescent lactam product (m=z 272) characterized previously. It is reasonable that the attachment of the hydroxide ion (as depicted in Fig. 3B) would be accelerated for the excited-state derivative. There is experimental precedent [31] that a moleculeÕs excited state is often more reactive to nucleophiles than its ground state, resulting in a photonucleophilic reaction. The additional reactivity has been attributed to an increase in internal energy and dipole, which may help the molecule overcome reaction barriers [31]. Given this precedent and the proportion of lactam vs isoindole following photobleaching, it may be expected that a significant population of the isoindole reacts via the established degradation mechanisms—the formation of the lactam and the methoxybenz[f]isoindole (m=z ¼ 286)—when photobleached. Furthermore, Orwar et al. [20] proposed that the degradation of select OPA–BME derivatives via photobleaching might follow a photoinduced intramolecular reaction. For amidecontaining amino acids (e.g., Asn, Gln) in solution, the isoindole serves as an excited-state electron donor that interacts with nearby amide groups (electron acceptors), accelerating the degradation of the isoindole. The increased reactivity of the excited state observed by Orwar et al. [20] is consistent with the observations described here. The transition to the lactam is not the only barrier overcome by photobleaching. The number of ions detected in the photobleached sample (vs the nonphotobleached sample) increases, whereas their relative abundance decreases. Clearly, there are multiple degradation pathways available under photobleaching conditions. One obvious possibility involves the direct photolysis of the isoindole into its constituents, which may explain the increase in NDA (m=z ¼ 185) observed following photobleaching. Another photobleaching mechanism is evident from the presence of the peak at m=z ¼ 348. This oxidation product is consistent with the formation of reactive oxygen species, a common product of photochemical reactions. Additional photodestruction mechanisms may include the conversion of energy from the excited-state isoindole to the nearby lactam, resulting in the photo-sensitization of the lactam. In this way, the degradation of a nonabsorbing molecule may be accelerated, as will be discussed later. To demonstrate the effects of photobleaching, the ratios of peak heights for isoindole vs the sum of isoindole and lactam forms are plotted (Fig. 7) for timed

76

D.P. Manica et al. / Analytical Biochemistry 322 (2003) 68–78

Fig. 6. Mass spectra obtained for (A) a derivative that was not exposed to light for 2 min before injection and (B) a derivative that was photobleached for 2 min before injection. The predominant photobleached product is the lactam product also produced by hydrolysis (or a variation thereof). Other suggested products are shown. Photobleaching conditions: Arþ laser, 500 mW, 488 nm. Spectra have been set to the same absolute ion abundance.

0.7

Ratio (isoindole / lactam + isoindole)

0.6

non-photobleached photobleached

0.5

0.4

0.3

0.2

0.1

0.0 0

2

4

6

8

Duration of Influence (min) Fig. 7. Comparison of the relative proportions of isoindole and lactam forms of Ser–NDA–BME for durations of photobleaching vs no photobleaching. Samples were mixed as described previously and either were allowed to react in an amber vial for the specified time or were photobleached in a clear vial for the specified time before solid-phase extraction and MS analysis. Each bar represents the average (n ¼ 3, 1 standard deviation) ratio of the peak height of the isoindole form divided by the sum of the isoindole and lactam forms after the specified time and duration of influence (photobleaching or no photobleaching).

D.P. Manica et al. / Analytical Biochemistry 322 (2003) 68–78

reactions under conditions of photobleaching and no photobleaching. A comparison of the relative amount of fluorescent isoindole in the photobleached vs the nonphotobleached samples reveals that the photodestruction of the isoindole is rapid within the first 2 min and less rapid thereafter. The ratio of isoindole to the sum of isoindole and lactam is reduced by 63.2% (vs 2 min control) after 2 min of photobleaching, with a difference that is statistically significant at the 97.5% confidence level. Therefore, the process of photobleaching appears to accelerate the already established method of decomposition to the lactam and to cause causing the formation of multiple new compounds. In summary, the general trend is that both the isoindole and the lactam decrease in abundance under photobleaching conditions (Fig. 6), but the isoindole degrades much more quickly (Fig. 7). Because the lactam does not absorb at the excitation wavelength, its degradation might be explained as an intermolecular reaction, wherein the isoindole acts as a photo-sensitizer. Thus, it appears that the energy of the excited-state isoindole is transferred to the nearby lactam, giving it energy to undergo further reactions. For very long durations (i.e., 8 min), the photobleached sample appears to have a higher relative amount of isoindole than the nonphotobleached sample. This is not an indication of a greater amount of fluorescence in 8min photobleached samples than in 8-min nonphotobleached samples, as the absorbance at 460 nm for all photobleached samples is significantly less than that of their nonphotobleached counterparts (data not shown). Rather, the higher ratio of isoindole after 8 min of photobleaching may be attributed to continued photodestruction of the more abundant lactam (in the denominator) to unrecognized products, thereby raising the apparent ratio.

Conclusions The stability of NDA derivatives has been characterized and the degradation mechanism and predominant degradation products have been identified. These experiments demonstrate the role of mass spectrometry as a means of elucidating complex degradation pathways. The reaction mechanism proposed is consistent with experimental observations and analogous to the degradation of OPA derivatives. The degradation pathway can be carried out with a number of different initiators and under different conditions, such as timedependent hydrolysis, degradation in the absence of solvent, and photodestruction. Additionally, a quantitative measure of the fluorescent half-life of Ser–NDA– BME is given. Photobleaching of Ser–NDA–BME appears to accelerate the degradation of the isoindole and, to a lesser extent, the lactam form, possibly by increasing their reactivity with nucleophiles in either

77

intermolecular or intramolecular interactions. In addition, formation of the methoxy-substituted benz[f]isoindole derivative maintains the structural requirements for fluorescence. This observation supports the assertion that methanol stabilizes fluorescence in general, even though it does not necessarily preserve the original derivative.

Acknowledgments This work was supported by the National Science Foundation. The Mariner mass spectrometer was purchased in part using funds from NIH grant 1S10RR11318.

References [1] H. Spackman, W.H. Stein, S. Moore, Automatic recording apparatus for use in the chromatography of amino acids, Anal. Chem. 30 (1958) 1190–1206. [2] M. Rubinstein, S. Chen-Kiang, S. Stein, S. Udenfriend, Characterization of proteins and peptides by high-performance liquid chromatography and fluorescence monitoring of their tryptic digests, Anal. Biochem. 95 (1979) 117–121. [3] C. DeJong, G.J. Hughs, E. Van Wieringen, K.J. Wilson, Amino acid analyses by high-performance liquid chromatography: an evaluation of the usefulness of pre-column Dns derivatization, J. Chromatogr. 241 (1982) 345–359. [4] Y. Watanabe, K. Imai, High-performance liquid chromatography and sensitive detection of amino acids derivatized with 7-fluoro-4nitrobenzo-2-oxa-1,3-diazole, Anal. Biochem. 116 (1981) 471–472. [5] M. Roth, Fluorescence reaction for amino acids, Anal. Chem. 43 (1971) 880–882. [6] R.G. Carlson, K. Srinivasachar, R.S. Givens, B.K. Matuszewski, New derivatizing agents for amino acids and peptides. 1. facile synthesis of N-substituted 1-cyanobenz[f]isoindoles and their spectroscopic properties, J. Org. Chem. 51 (1986) 3978–3983. [7] P. de Montigny, J.F. Stobaugh, R.S. Givens, R.G. Carlson, K. Srinivasachar, L.A. Sternson, T. Higuchi, Naphthalene-2,3-dicaion: ion: a rationally designed fluorogenic reagent for primary amines, Anal. Chem. 59 (1987) 1096–1101. [8] J.F. Stobaugh, A.J. Repta, L.A. Sternson, K.W. Garren, Factors affecting the stability of fluorescent isoindoles derived from reaction of o-phthalaldehyde and hydroxyalkylthiols with primary amines, Anal. Biochem. 135 (1983) 495–504. [9] W.A. Jacobs, o-Phthalaldehyde-sulfite derivatization of primary amines for liquid chromatography-electrochemistry, J. Chromatogr. 392 (1987) 435–441. [10] M.D. Oates, J.W. Jorgenson, Determination of naphthalene-2,3dicarboxaldehyde-labeled amino acids by open tubular liquid chromatography with electrochemical detection, Anal. Chem. 61 (1989) 432–435. [11] B.K. Matuszewski, R.S. Givens, K. Srinivasachar, R.G. Carlson, T. Higuchi, N-substitiuted 1-cyanobenz[f]isoindole: evaluation of fluorescence efficiencies of a new fluorogenic label for primary amines and amino acids, Anal. Chem. 59 (1987) 1102–1105. [12] S.M. Lunte, T. Mohabbat, O.S. Wong, T. Kuwana, Determination of desmosine, isodesmosine, and other amino acids by liquid chromatography with electrochemical detection following precolumn derivatization with naphthalenedialdehyde/cyanide, Anal. Biochem. 178 (1989) 202–207.

78

D.P. Manica et al. / Analytical Biochemistry 322 (2003) 68–78

[13] M.A. Nussbaum, J.E. Przedwiecki, D.U. Staerk, S.M. Lunte, C.M. Riley, Electrochemical characteristics of amino acids and peptides derivatized with naphthalene-2,3dicarboxaldehyde pH effects and differences in oxidation potentials, Anal. Chem. 64 (1992) 1259–1263. [14] O. Orwar, H.A. Fishman, N.E. Ziv, R.H. Scheller, R.N. Zare, Use of 2,3-naphthalenedicarboxaldehyde derivatization for single-cell analysis of glutathione by capillary electrophoresis and histochemical localization by fluorescence microscopy, Anal. Chem. 67 (1995) 4261–4268. [15] S.D. Gilman, A.G. Ewing, Post-column derivatization for capillary electrophoresis using naphthalene-2,3-dicarboxaldehyde and 2-mercaptoethanol, Anal. Methods Instrum. 2 (1995) 133–141. [16] S.D. Gilman, A.G. Ewing, Analysis of single cells by capillary electrophoresis with on-column derivatization and laser-induced fluorescence detection, Anal. Chem. 67 (1995) 58–64. [17] Y. Liu, R.S. Foote, S.C. Jacobson, R.S. Ramsey, J.M. Ramsey, Electrophoretic separation of proteins on a microchip with noncovalent, postcolumn labeling, Anal. Chem. 72 (2000) 4608– 4613. [18] C.A. Monnig, J.W. Jorgenson, On-column sample gating for highspeed capillary zone electrophoresis, Anal. Chem. 63 (1991) 802– 807. [19] J.A. Lapos, A.G. Ewing, Injection of fluorescently labeled analytes into microfabricated chips using optically gated electrophoresis, Anal. Chem. 72 (2000) 4598–4602. [20] O. Orwar, H.A. Fishman, M. Sundahl, V. Banthia, R. Dadoo, R.N. Zare, Determination of photodestruction quantum yields using capillary electrophoresis: application to o-phthalaldehyde/bmercaptoethanol-labeled amino acids, J. Liquid Chromatogr. 18 (1995) 3833–3846. [21] J.D.H. Cooper, G. Ogden, J. McIntosh, D.C. Turnell, The stability of the o-phthalaldehyde/2-mercaptoethanol derivatives

[22]

[23]

[24]

[25]

[26]

[27]

[28]

[29] [30]

[31]

of amino acids: an investigation using high-pressure liquid chromatography with a precolumn derivatization technique, Anal. Biochem. 142 (1984) 98–102. H. Nakamura, A. Matsumoto, Z. Tamura, On the stability of isoindole-type fluorophores derived from o-phthalaldehyde, primary amino compounds and thiols, Anal. Lett. 15 (1982) 1393– 1410. K. Saito, M. Horie, H. Nakazawa, Kinetic study of the stability of the o-phthalaldehyde-spermine fluorophore formed by on-column derivatization, Anal. Chem. 66 (1994) 134–138. S.S. Simons Jr., D.F. Johnson, Reaction of o-phthalaldehyde and thiols with primary amines: formation of l-alkyl(and aryl)thio-2alkylisoindoles, J. Org. Chem. 43 (1978) 2886–2891. J.S. Rossier, R. Ferrigno, H.H. Girault, Electrophoresis with electrochemical detection in a polymer microdevice, J. Electroanal. Chem. 492 (2000) 15–22. J. Wang, M.P. Chatrathi, B. Tian, Micromachined separation chips with a precolumn reactor and end-column electrochemical detector, Anal. Chem. 72 (2000) 5774–5778. S.S. Simons Jr., D.F. Johnson, The structure of the fluorescent adduct formed in the reaction of o-phthalaldehyde and thiols with amines, J. Am. Chem. Soc. 98 (1976) 7098–7099. L. Song, E.J. Hennink, I.R. Young, H.J. Tanke, Photobleaching kinetics of fluorescein in quantitative fluorescence microscopy, Biophys. J. 68 (1995) 2588–2600. G.H. Patterson, D.W. Piston, Photobleaching in two-photon excitation microscopy, Biophys. J. 78 (2000) 2159–2162. C.A. Monnig, D.M. Dohmeier, J.W. Jorgenson, Sample gating in open tubular and packed capillaries for high-speed liquid chromatography, Anal. Chem. 63 (1991) 807–810. H. Hou, Y. Huang, S.J. Gulding, C.T. Rettner, D.J. Auerbach, A.M. Wodtke, Enhanced reactivity of highly vibrationally excited molecules on metal surfaces, Science 284 (1999) 1647–1650.