Published on Web 03/20/2004
Photochemical Reaction Mechanisms of 2-Nitrobenzyl Compounds: Methyl Ethers and Caged ATP Yuri V. Il’ichev,† Markus A. Schwo¨rer,‡ and Jakob Wirz* Contribution from the Departement Chemie der UniVersita¨t Basel, Klingelbergstr. 80, CH-4056 Basel, Switzerland Received October 15, 2003; E-mail:
[email protected]
Abstract: The mechanism of methanol photorelease from 2-nitrobenzyl methyl ether (1) and 1-(2nitrophenyl)ethyl methyl ether (2), and of ATP release from adenosine-5′-triphosphate-[P3-(1-(2-nitrophenyl)ethyl)] ester (‘caged ATP’, 3) was studied in various solvents by laser flash photolysis with UV-vis and IR detection. In addition to the well-known primary aci-nitro transients (A, λmax ≈ 400 nm), two further intermediates preceding the release of methanol, namely the corresponding 1,3-dihydrobenz[c]isoxazol1-ol derivatives (B) and 2-nitrosobenzyl hemiacetals (C), were identified. The dependencies of the reaction rates of A-C on pH and buffer concentrations in aqueous solution were studied in detail. Substantial revision of previously proposed reaction mechanisms for substrate release from 2-nitrobenzyl protecting groups is required: (a) A novel reaction pathway of the aci-tautomers A prevailing in buffered aqueous solutions, e.g., phosphate buffer with pH 7, was found. (b) The cyclic intermediates B were identified for the first time as the products formed by the decay of the aci-tautomers A in solution. A recently proposed reaction pathway bypassing intermediates B (Corrie et al. J. Am. Chem. Soc., 2003, 125, 8546-8554) is shown not to be operative. (c) Hemiacetals C limit the release rate of both 1 (pH < 8) and 2 (pH < 10). This observation is in contrast to a recent claim for related 2-nitrobenzyl methyl ethers (Corrie et al.). Our findings are important for potential applications of the 2-nitrobenzyl protecting group in the determination of physiological response times to bioagents (‘caged compounds’).
Introduction
The 2-nitrobenzyl functionality is the most widely used photoremovable protecting group for application in synthesis,1 photolithography (DNA microarrays),2 and biochemistry (“caged compounds”).3 Physiological response times to bioagents such as neurotransmitters may be determined, if the release of a signaling molecule following pulsed excitation of the protected precursor is faster than the response time under investigation.4 The primary photoreaction of 2-nitrobenzyl compounds is intramolecular H-atom transfer affording aci-nitro tautomers that are readily detected by their strong absorption around 400 nm. Their decay is often taken to indicate the release rate of the protected agents. However, the subsequent reactions leading to release of the protected leaving groups are not well understood. † Present address: Wichita State University, Department of Chemistry, 1845 Fairmount St., Wichita, KS 67260-0051. ‡ Present address: MDL Information Systems GmbH, Theodor-HeussAllee 108, D-60486 Frankfurt/Main, Germany.
(1) Pillai, V. N. R. Synthesis 1980, 1-26. Greene, T. W.; Wuts, P. G. M. ProtectiVe Groups in Organic Synthesis, 3rd ed.; Wiley-Interscience: New York, 1999. (2) Chee, M.; Yang. R.; Hubbell, E.; Berno, A.; Huang, X. C.; Stern, D.; Winkler, J.; Lockhart, D. J.; Morris, M. S.; Fodor, S. P. A. Science 1996, 274, 610-614. Pirrung, M. C. Chem. ReV. 1997, 97, 473-488. Pirrung, M. C. Angew. Chem., Int. Ed. 2002, 41, 1276-1289. (3) Pelliccioli, A. P.; Wirz, J. Photochem. Photobiol. Sci. 2002, 1, 441-458. (4) Breitinger, H.-G. A.; Wieboldt, R.; Ramesh, D.; Carpenter, B. K.; Hess, G. P. Biochemistry 2000, 39, 5500-5508. Givens, R. S.; Weber, J. F. W.; Conrad, P. G.; Orosz, G.; Donahue, S. L.; Thayer, S. A. J. Am. Chem. Soc. 2000, 123, 2687-2697. Pollock, J.; Crawford, J. H.; Wootton, J. F.; Corrie, J. E. T.; Scott, R. H. Neurosci. Lett. 2003, 338, 143-146. 10.1021/ja039071z CCC: $27.50 © 2004 American Chemical Society
Here, we report a study of the model compounds 2-nitrobenzyl methyl ether (1) and 1-(2-nitrophenyl)ethyl methyl ether (2) by picosecond pump-probe spectroscopy, nanosecond laser flash photolysis (LFP), and time-resolved infrared (TRIR) spectroscopy. Two additional intermediates that are formed from the primary aci-nitro photoproducts were identified and the rate of methanol photorelease from nitrobenzyl-protected substrates was found to be orders of magnitude slower than that for the decay of the primary aci-nitro transients in aqueous solution at pH ≈ 7. ‘Caged ATP’,5 the disodium salt of adenosine-5′triphosphate-[P3-(1-(2-nitrophenyl)ethyl)] ester (3) was also briefly investigated. A related study on derivatives of 1 and 2 was published6 after the completion of our work,7 and will be referred to in the discussion. A well-known problem with the use of 2-nitrobenzyl and 2-nitrophenethyl protecting groups in biological preparations is that the release of the desired bioactive compounds is accompanied by the formation of 2-nitrosobenzaldehyde or 2-nitrosoacetophenone, respectively, which may inactivate the biological preparations or give rise to further, undesired but harmful photoreactions. Secondary photoreactions of the nitroso products were not investigated here. In a previous publication, we have reinvestigated the largely reversible (5) Kaplan, J. H.; Forbush, B.; Hoffman, J. F. Biochemistry 1978, 17, 19291935. (6) Corrie, J. E. T.; Barth, A.; Munasinghe, V. R. N.; Trentham, D. R.; Hutter, M. C. J. Am. Chem. Soc. 2003, 125, 8546-8554. (7) Schwo¨rer, M. ‘Mechanismen der lichtinduzierten Freisetzung von Abgangsgruppen aus 2-Nitrobenzylverbindungen’, Ph.D. thesis, University of Basel, 2002, and ref 3. J. AM. CHEM. SOC. 2004, 126, 4581-4595
9
4581
ARTICLES
Il’ichev et al.
Scheme 1. Phototautomerization of 2-Nitrotoluene
phototautomerization of parent 2-nitrotoluene (Scheme 1) to identify the elementary steps of the thermal back reaction in aqueous solution as a benchmark.8 In a forthcoming paper, we will discuss the photoreactions of 2-nitrobenzyl alcohols.9
Experimental Section Methods. Reaction quantum yields were determined by spectrophotometric monitoring of the absorbance changes. A medium-pressure mercury lamp equipped with a 365-nm band-pass filter was used as a light source. Actinometry was done with a solution of azobenzene in methanol.10 The nanosecond kinetic and spectroscopic laser flash photolysis setup was of standard design. An excimer laser operated mostly on KrF (248 nm), but also on XeCl (308 nm) or XeF (351 nm), was used as an excitation source with a pulse duration of about 25 ns and pulse energies of 5. The amplitudes of the two components were approximately equal.28 The decay of the shortlived component was accompanied by a 15-nm blue shift of the absorption maximum (measured in phosphate buffer, pH 6.8). This indicates that transient A is a mixture of two kinetically distinct species, A1 and A2. (28) The decay traces clearly deviated from a simple first-order rate law and gave excellent fits with two exponentials, but the two close-lying rate constants and their amplitudes were not defined with high accuracy. (29) This value is close to the time resolution of our instrument that was limited by the 30-ns duration of the laser pulse. 4584 J. AM. CHEM. SOC.
9
VOL. 126, NO. 14, 2004
Figure 4. pH-Rate profiles of reactions ab1 (b), ab2 (O), bc1 (+), bc2 (×), and c4 (0) initiated by LFP of 1. The solid lines were obtained by nonlinear least-squares fitting of eq 1 to the data points (Table S1)30 for ab1 and ab2 and of eq 3 to those for bc1, bc2, and c4. The resulting parameters are given in Table 1.
Rate constants determined with buffer solutions (pH 3.510.5) were found to increase linearly with buffer concentration indicating general acid and/or base catalysis. Data points for the pH-rate profile were, therefore, obtained by linear extrapolation of buffer dilution plots (vide infra) to zero buffer concentration. The pH-rate profiles for reactions ab1 and ab2 are shown in Figure 4. The data points are given in Table S1 of the Supporting Information.30 The rates of reactions ab1 and ab2 are constant at pH > 10, and rise steadily with increasing acid concentration at pH < 10. The slope of log kobs for reaction ab2 clearly drops below -1 in the pH range of 4-6. The solid lines were obtained by nonlinear least-squares fitting of the parameters kab′′, kab′, and kab/Ka,1 in eq 1 to the observed first-order rate constants of reactions ab1 and ab2. Equation 1 will be derived in the Discussion. The acidity constants Ka,2 of A1 and A2 were more accurately defined by the buffer dilution plots (vide infra) and were thus fixed to the values Ka,2(A1) ) 3.2 × 10-5 M and Ka,2(A2) ) 1.0 × 10-4 M for fitting of eq 1.
log (kobs/s-1) ) log{(kab′′Ka,2 + kab′[H+] + kab[H+]2/Ka,1)/ (M s-1)} - log{([H+] + Ka,2)/M} (1) In pure water with no buffer added, a small time-resolved increase in absorbance was observed following the sudden rise in absorption at 420 nm. This is attributed to ionization of the aci-nitro transients A to the conjugate anions A-, k-H ) (2.7 ( 0.3) × 106 s-1 (ca 10% of the total amplitude), which appear to show stronger absorption at 420 nm. This rate constant is nearly an order of magnitude lower than that of the aci-tautomer of 2-nitrotoluene,8 indicating that the acidity constants of the aci-tautomers A, Ka,2 ) k-H/ kH, are about an order magnitude smaller than that of the latter, pKa,2 ) 3.6. Combination of k-H with a typical rate constant for the reaction of protons with a weak oxygen base, kH ≈ 5 × 1010 M-1 s-1, provides a first estimate of the acidity constant of the aci-tautomers A, pKa,2 ≈ (30) Supporting Information: See paragraph at the end of this paper regarding availability.
Photoreactions of 2-Nitrobenzyl Compounds
Figure 5. Kinetic traces obtained by LFP of 1 in aqueous solutions. Trace 1: ionization of A in 5 × 10-4 M NaOH. Trace 2: decay of A1 and A2 in phosphate buffer (0.0143 M KH2PO4, 0.0286 M Na2HPO4, pH 7.08). Trace 3: reaction bc1 and bc2 in acetate buffer (0.0072M HAc, 0.00144 M NaAc, pH 3.87). Trace 4: reaction c4 in 0.1 M HClO4.
4.3.31 Ionization of the aci-tautomers A was accelerated by the addition of NaOH or buffers. Measurements in the range of 5 × 10-4 (Figure 5, trace 1) to 1 × 10-3 M NaOH gave a rate constant of kOH ) (8 ( 2) × 109 M-1 s-1 for the deprotonation of A to A- by hydroxyl ions. A very weak transient absorption that was observed at 500 nm decayed with a rate constant of ca. 5 × 106 s-1, independent of pH (up to 1 M NaOH) and oxygen (up to 1 atm). We cannot offer a convincing assignment for this transient. The decay of transient A was clearly biexponential at pH > 5 (trace 2 in Figure 5). With increasing acidity the decay rate of the slower component A2 increased more rapidly than that of A1, such that the two decays merged to an essentially single exponential with a rate constant of about 6 × 105 s-1 at pH 3 (cf., however, the buffer dilution plots discussed below). The absorption due to transients A1 and A2 decayed to the baseline in aqueous solutions with pH < 6, leaving no residual absorption above 300 nm. Subsequently, a slow, biexponential growth in absorbance was observed at 310-330 nm (trace 3 in Figure 5: formation of intermediate C). These observations require intervention of some intermediates, B1 and B2, which are transparent above 300 nm. The pH-dependence of the two firstorder rate constants for the formation of C from B1 and B2, determined at 330 nm, is shown in Figure 4 (reactions bc1 and bc2). Formation of C is also accompanied by the growth of a very weak absorbance at 740 nm, which is characteristic for the formation of nitroso compounds.32 However, the measurements at 330 nm gave more accurate results, because the absorbance changes associated with the reactions B f C are much larger at this wavelength. The decay of transients B is catalyzed by base. At pH > 6 the decay rates of B1 and B2 exceed the rates of formation of these intermediates, so that the decay rates of A1 and A2 coincide with the rates of formation of C. In general, the time-dependent concentration of a product C that is formed in a sequential reaction scheme of the type A f B f C with rate constants kab and kbc is given by eq 2. Thus, the formation of C by the two parallel reaction sequences A1 (31) A similar estimate based on the ionization kinetics of 2-nitrotoluene’s acinitro tautomer in pure water gave pKa ≈ 3.4, in good agreement with pKa ) 3.6 determined independently.8 (32) Walker, J. W.; Reid, G. P.; McCray, J. A.; Trentham, D. R. J. Am. Chem. Soc. 1988, 110, 7170-7177.
ARTICLES
Figure 6. UV-vis difference absorbance spectra monitoring reaction c4 in aqueous solution. The first spectrum was recorded about 10 s after LFP of 1 in 5 × 10-4 M HClO4.
f B1 f C and A2 f B2 f C is expected to obey a fourexponential rate law.
kbc(1 - e-kabt) - kab(1 - e-kbct) [C]t ) [A]t)0 kbc - kab
(2)
However, eq 2 rapidly approaches a single-exponential rate law, [C]t ≈ [A]t)0(1 - e-kt), as the two rate constants kab and kbc become different in magnitude, with the observable rate constant k becoming that of the slower (rate limiting) step. A reliable determination of four rate constants was not possible from the absorbance growth waveforms monitored at 330 nm. Growth curves determined in the pH range of 5.5-6.5, where kab ≈ kbc and a four-exponential fit would have been required, were disregarded in the analysis. Traces obtained outside that range were adequately fitted by a sum of two exponentials (Figure 5, trace 3). The two rate constants for the formation of C that were determined at 330 nm in solutions of pH > 6.5 are limited by and equal to the rate constants of the reactions ab1 and ab2, kab1 and kab2, whereas those at pH < 5.5 are slower and provide the rate constants of reactions bc, kbc1, and kbc2. Finally, the lifetime of product C is around 40 s in the pHrange of 3-6, and the absorbance changes arising from reaction c4 were sufficiently slow to be recorded with a diode array UV spectrometer. The absorbance at 320 nm due to C decreases, while that at 230 nm increases strongly (Figure 6). A global fit of the data obtained with 5 × 10-4 M aqueous HClO4 gave a rate constant of (2.57 ( 0.01) × 10-2 s-1 and provided UV spectra of the initial (t ) 0) and final species. These were nearly superimposable with the UV spectra of 2-nitrosobenzyl alcohol (as a stable analogue for the hemiacetal C) and authentic 4, respectively, in the same solvent. The higher rate constants kc4 outside the range of 3 < pH < 6 were determined by kinetic flash photolysis analyzed at 320 nm. The pH-rate profiles for reactions bc1, bc2, and c4 (Figure 4) indicate acid catalysis at pH < 3 and base catalysis at pH > 5. Equation 3 describes a reaction that has a pH-independent
log (kobs/s-1) ) log {(kH[H+] + kOHKW/[H+] + k0)/s-1} (3) component, k0, but also exhibits catalysis by protons, kH, and by hydroxide ions, kOH. The hydroxyl ion concentration was replaced by KW/[H+], where KW ) 1.59 × 10-14 M2 is the J. AM. CHEM. SOC.
9
VOL. 126, NO. 14, 2004 4585
ARTICLES
Il’ichev et al.
Table 1. Rate Coefficients Determined by Fitting of Eqs 1 and 3 to the pH-Rate Profiles of 1 to 3 reaction
kab′′/s-1
ab1(1) 2.1 ( 0.4 ab2(1) 0.23 ( 0.08 ab1(2) ab2(2) (9 ( 2) × 102 ab(3) 1.5 ( 0.5
kab′/s-1
(kab/Ka,1)/s-1
Ka,2/M
(8.7 ( 0.8) × 105 (5.0 ( 0.6) × 109 ∼1.0 × 10-4a,b (1.4 ( 0.3) × 104 (7.8 ( 1.5) × 109 ∼3.2 × 10-5a,b (5.1 ( 0.2) × 107 ∼7 × 109 (1.1 ( 0.2) × 106 (1.8 ( 0.8) × 1010 1 × 10-4b (7 ( 2) × 104 ∼5 × 109 3.2 × 10-5b
reaction
kH/M-1 s-1
k0/s-1
bc1(1) bc2(1) c4(1) bc(2) c5(2) bc(3)
(2.6 ( 0.4) × (3.6 ( 0.3) × 104 23 ( 4 ∼1 × 103 (2.3 ( 0.9) × 102 (2.0 ( 0.2) × 107 104
kOH/M-1 s-1
(1.8 ( 0.2) × 13 ( 2 (1.6 ( 0.3) × 10-3 (4.0 ( 2.7) × 102 ∼1 × 10-2 (7.7 ( 0.9) × 103 102
∼1 × 1010 (1.1 ( 0.2) × 1010 (1.2 ( 0.2) × 106 ∼1 × 1010 ∼1.3 × 106
a Value taken from a fit of buffer slopes to eq 5. b Value was kept constant in fitting.
Figure 7. Buffer dilution plot for the decay rate constants of transients A1 (b) and A2 (O) in acetic acid buffer 1:1 (pH 4.57).
ionization constant of water at ionic strength I ) 0.1 M.33 Nonlinear least-squares fitting of eq 3 to the observed rate constants for reactions bc1, bc2, and c4 provided the parameters k0, kH, and kOH for each reaction, which are given in Table 1. The coincidence of the reaction rates for ab1 and bc1 and for ab2 and bc2 at pH > 6 (Figure 4) proves that intermediate A1 is the precursor of B1, and A2 of B2. Consistently, the amplitude ratio of reactions ab1 and ab2 in the kinetic traces measured at 420 nm was about equal to that of reactions bc1 and bc2 at 320 nm. Buffer Dilution Plots. Measurements in aqueous solutions of pH 3.5-10.5 require the addition of buffers, which, however, may influence the reaction rates by general acid and general base catalysis. These measurements were made in series of solutions of varying buffer concentration but constant buffer ratio and, hence, constant pH. Catalysis by buffers was found for all reactions and buffer dilution plots at constant buffer ratio were linear (e.g., Figure 7). The intercepts represent the decay rates at zero buffer concentration, which were used for the pHrate profiles (Figure 4). The buffer catalytic coefficients, kbu, which correspond to the slopes of the buffer dilution plots, are given in Table 2. As shown in Figure 7, the slopes kbu determined for the decay of A1 were about 10-fold higher than (33) Bates, R. G. Determination of pH, Theory and Practice; Wiley: New York, 1973. 4586 J. AM. CHEM. SOC.
9
VOL. 126, NO. 14, 2004
those of A2. This led to a clear-cut separation of the two reaction rates with increasing buffer concentration in acetic and formic acid buffers, and proves that the near coincidence of the decay rate constants kab1 and kab2 in wholly aqueous solutions at pH < 5 is fortuitous. Solvent Kinetic Isotope Effects. The decay rates of A1 and A2 were determined in H2O and D2O solutions with equal nominal concentration of added acid or base: kH/kD ) 0.97 ( 0.03 in 1.1 × 10-3 M aqueous perchloric acid (single exponential), and kH/kD ) 1.08 ( 0.18 (ab1) and 0.84 ( 0.18 (ab2) in 0.1 M sodium hydroxide. Relatively large error limits in the latter values arise because of the uncertainty associated with fitting decay traces with a sum of two close-lying exponentials. TRIR-Measurements of 1. The IR-spectrum of 1 in CD3CN is dominated by strong bands arising from the symmetric (1343 cm-1) and antisymmetric (1526 cm-1) stretching vibrations of the NO2 group. Medium intensity bands are seen in the range 2800-3000 cm-1 (C-H stretch) and at 1109 cm-1 (antisymmetric C-O-C stretch). The corresponding frequencies predicted by the DFT calculations (1329 and 1353 cm-1 for the symmetric NO2 stretch, 1551 and 1564 cm-1 for the antisymmetric NO2 stretch, 1109 and 1111 cm-1 for the C-OMe stretch in the anti- and syn-conformations of 1, respectively) are in satisfactory agreement with the experimental data. CD3CN and CD2Cl2 were used as solvents for the TRIR experiments. Measurements with the same (nondeuterated) solvents were done by optical LFP for comparison. Reaction A f B in CD3CN was followed by the step-scan method with a 9 × 10-3 M solution of 1. The kinetic analysis of TRIR spectral data is less precise than that of kinetic traces obtained by optical LFP and the TRIR data for reactions ab and bc were adequately fitted with a single exponential function. Hence, only a single IR spectrum is available for A (representing a mixture of A1 and A2), and one for B. Reaction A f B. The spectral matrix covering a time range from 2.5 to 500 µs after flash photolysis (Figure 8) was subjected to factor analysis. Two components were sufficient to reconstruct the spectra within experimental accuracy. Prominent features are the decaying bands in the region of 1100-1400 and 15401650 cm-1. A band at 1080 cm-1 grows in simultaneously, but otherwise no strong absorption bands, particularly no NdO or CdO bands, are present in the last spectra. The strong negative bands in the difference spectra did not show a significant change in position or intensity during the time span covered. The positions of these negative bands correspond to those of the absorption bands of 1 indicating depletion of the starting material 1. Intermediates A. The data matrix represented in Figure 8 was adequately fitted by a single-exponential rate law. The resulting rate constant kab ) (6.6 ( 0.2) × 103 s-1 is in good agreement with that for the major component, kab1 ≈ 6 × 103 s-1,28 of the biexponential decay observed by LFP of 1 in acetonitrile. The negative bands due to the depletion of 1 were removed by adding an appropriate amount of the IR spectrum of 1 to the first spectrum of Figure 8 to obtain the absorption spectrum of transients A, Figure 9. It is in qualitative agreement with the much more highly resolved spectrum that was reported by Dunkin et al.34 for the same product formed by irradiation of 1 in Ar and N2 matrixes at 12 K.
Photoreactions of 2-Nitrobenzyl Compounds
ARTICLES
Table 2. Buffer Slopes kBu Obtained with 1 by Linear Regression of Buffer Dilution Plotsa buffer
phthalic formic formic formic formic acetic acetic acetic acetic acetic phosphate phosphate phosphate phosphate phosphate tris borate d
xHB
1/2 4/5 2/3 1/3 1/5 5/6 2/3 1/2 1/3 1/9 6/7 3/4 2/3 1/2 1/3 1/2 1/2
cbu/Mb
kbuab1/(M-1s-1)
0.01-0.05 0.05-0.50 0.03-0.30 0.015-0.15 0.0125-0.125 0.011-0.60 0.02-0.12 0.02-0.2 0.015-0.15 0.01-0.11 0.008-0.078
(2.70 ( 0.45) × (4.8 ( 0.3) × 106c (6.2 ( 1.9) × 107 (3.1 ( 0.3) × 106 (1.3 ( 0.1) × 107 (1.18 ( 0.03) × 107 (7.6 ( 1.1) × 106 (3.5 ( 0.1) × 106 (1.1 ( 0.1) × 106 (5.8 ( 1.0) × 104 (8.6 ( 3.8) × 103
c c (5.0 ( 0.3) × 106 (1.9 ( 0.2) × 106 (1.4 ( 0.1) × 106 (1.16 ( 0.14) × 106 (4.5 ( 1.2) × 105 (1.46 ( 0.13) × 105
