|Home | About | Journals | Submit | Contact Us | Français|
In contrast to several proposed mechanisms for prebiotic selection of l-amino acids in proteins, formation of the gaseous magic-number cluster (Ser8+H)+ achieves extensive homochirality; d,l-Ser selectively yields (d-Ser8+H) + and (l-Ser8+H)+. Using processes that simulate prebiotic conditions, serine octamers are also selectively formed as neutrals and anions and from solid and solution phases. (Ser8+H)+ is far more stable than homochiral octamers of any other of the 20 naturally-occurring amino acids (X), and single substitutions of the latter into (l-Ser8+H)+ are enantioselective to form (l-Ser7-l-X+H)+. However, high chiral selectivity of an enzyme is usually found to result from a unique binding-site conformation, while recent infrared photodissociation (IRPD) spectra from our Cornell and Taiwan laboratories found contrasting structures containing no and multiple free aliphatic –OH groups, respectively. Herein we reexamine (Ser8+H)+ IRPD spectra from ions formed under a variety of conditions.
(Ser8+H)+ ions have been extensively studied by experimental[1–9] and theoretical[6b,7,8] methods. Although all (Ser8+H)+ ions studied have nearly identical cross sections (~190 Å2),[6b,7] two isomeric forms have been distinguished by their H/D exchange rates.[2c,9] The faster-exchanging isomer B shows no chiral selection and is selectively dissociated in collisional SWIFT isolation of the A/B mixture.[2,9] Theory predicts a zwitterionic structure for isomer A, with the free (i.e. without H-bonding) aliphatic O–H group absent from that of Beauchamp et al.[6b] and present in the Schalley et al. structure.
IRPD spectra in the O–H/N–H stretching region were measured on electrosprayed (Ser8+H)+ ions (Figure 1) stored in a Fourier-transform mass spectrometer.[4,5] The Cornell 2003 spectrum[4,11] (as well as pre-2006 spectra, not shown, measured by Oh in Korea) did not absorb at ~3675 cm–1 (isomer i, Table 1), indicating the absence of free aliphatic O–H groups. However, this peak was strong in the 2006 Taiwan spectrum (isomer ii). Negligible absorption at ~3565 cm–1 (Figure 1) shows the absence of free carboxyl O–H,[4,5] supporting a zwitterionic structure. Some Cornell 2004 spectra (Figure 1) do have a new absorption at 3565 cm–1 that indicates a new isomer iii (Table 1).
Additional Cornell IRPD spectra (termed “2007”) were measured under what was thought to be the same experimental conditions, but without SWIFT isolation that removes isomer B. Although our 2007 spectra are quite similar, now the 3675 cm–1 peak has one-third the relative intensity of the corresponding Taiwan 2006 peak. This decrease is not due to a ~1:2 mixture of isomer ii with i (which does not absorb at 3675 cm–1), because IR irradiation at this frequency causes (Figure 2) essentially complete dissociation of these 2007 ions without forming i. The data in Figure 2 show a first-order correlation, indicating that both 2007 forms A and B (Figure 3 a, without SWIFT isolation) exhibit similar dissociation kinetics over this 200 s time period. The 2007 ions from SWIFT isolation, presumably isomer iv, gave an IRPD spectrum whose main features were closely similar to those of the iv/v mixture.
In contrast to these equivalent A/B photodissociation rates, initial SWIFT isolation of the 2007 ions (Figure 3 b), in which both the A and B forms at 10–9 torr are translationally excited by off-resonance frequencies followed after 25 s by a collisionally-activating gas pulse, reduces the intensity of B to a far lower value (~10%) than that of A (~60%).[2c,9] In further apparent contrast, conventional collisional activation (sustained off-resonance irradiation) that uses far faster (~0.1 s), higher-energy collisions (Figure 3 c) gave more comparable values: 46% for A and 33% for B (as shown by Figure 3 d, this is not due to the isomerization of A to B). Further, for the IRPD spectrum (not shown) of these product ions, the 3675 cm–1 free O–H absorption has been dramatically reduced to <10% of its original value (Figure 4), with that at 3325 cm–1 halved. Thus, the spectrum is closely similar to that of isomer i, suggesting i as a collisional activation product of the 2007 ions. Its B form counterpart is then the new isomer vi (Table 1; 2003 ions did not contain the B form).
Possible reactions of these A and B isomers are given by Equations (1)–(3), respectively: Dissociation :
. Isomerization :
. Dissociation :
At the low energy of Figure 2, the dissociation rates of Equations (1a) and (1b) are equivalent, but at the higher-energy SWIFT excitation (Figure 3 b) reaction (1b) is substantially faster, indicating a looser transition state. However, both (1a) and (1b) could involve an initial tight-complex isomerization, as even looser transition states are indicated for the those of Equations (2a) and (2b) that become competitive when activated at the highest Figure 3 c energy. Here their rates are greater than those of dissociations (3a) and (3b) [as well as (1a) and (1b)], indicating that isomers i and vi (no free O–H) are substantially more stable than iv and v. Similarly, ii with the most free O–H groups could be even less stable, avoiding isomerization to i or iv only under gentle ion introduction conditions. Quite different H/D exchange rates have also been noted for other gaseous conformers of similar structure.[3a,13] The A forms ii and iii could have B counterparts vii and viii, while other stable A forms could give similar IRPD spectra.
Of the (Ser8+H)+ structures proposed previously from theory, the zwitterionic form I (Figure 5) of Beauchamp et al.[6b] has been found by Schalley to be the most favorable. Further calculations here found at least seven stable isomers. The most stable (I, II, III) contain 0, 1, and 6 free aliphatic O–H groups, respectively, as reflected in the relative intensities of the 3675 cm–1 peaks in their computed IR spectra (Figure 5), with relative energies of 0, –13.5, and –11.9 kJ mol–1. However, I–III only reflect general properties of the Table 1 isomers i–vi, not specific characteristics. The absence of a free O–H peak from structure I is consistent with isomers i, iii, and vi, the weak peak of II with iv and v, and the strong peak of III with ii. However, i and vi (supposedly like I) are produced by activation of iv and v (II) by Equations (2a) and (2b), inconsistent with the lower relative energy of II. The 3565 cm–1 peaks in I and II are due to CH2O–H···(OCO)– (not free COO–H), suggesting this assignment for isomer iii as well. No free COO–H in any of I–III suggests that all have zwitterionic structures.
In summary, experiment and theory indicate at least six stable (Ser8+H)+ isomers. these have eight aliphatic O–H groups variously H-bonded around a zwitterionic core. However, only kinetic dissociation data [Eq. (1a)–(3b)], not IRPD spectral features, distinguish between the enantiomerically active and inactive forms A and B. A singular conformational structure is often cited as basic for an enzyme's unusually high specificity, but here the specificity of serine octamer is also of unusually broad applicability, as it acts for all the essential l-amino acids (X) to select l-X, not d-X, in (l-Ser7-l-X+H)+. The structural variability of the –OH groups on the outside of the cluster could optimize the conformation for substituting any amino acid for a single serine while retaining the unique enantioselectivity of its singular zwitterionic center.
IRPD spectra of (Ser8+H)+ ions were obtained at Cornell University as described previously[4,11] using a 6T Fourier-transform MS; its resolving power of ~105 shows no more highly charged ions such as (Ser16+2H)2+. Ions from nano-electrospray ionization of 5 mm L-serine (Sigma, St. Louis) in 49:49:2 MeOH:H2O:AcOH were trapped in the ion cell (N2 pulse, ~10–6 Torr). Where noted, all but the (Ser8+H)+ ions were ejected with a SWIFT waveform at 10–9 Torr, followed 25 s later by a 10–6 torr N2 pulse. Sustained off-resonance irradiation used 38 db, –1 kHz off resonance, for 0.1 s at ~10–6 Torr N2. H/D exchange used CH3OD in the FTMS cell at 6×10 7 Torr for 50 s, then a 300 s pump down.
IRPD spectra were measured at ~10–9 Torr with a pulsed infrared laser (3025–3775 cm–1, IR OPO 2732, OPOTek, Carlsbad, CA) pumped by a ND:YAG laser, output 4–9 mJ at 10 Hz. Intensities of the (Ser8+H)+ ions, relative to the laser power, were determined before and after photodissociation. Irradiation time at each frequency, typically 15 s, was controlled by a mechanical shutter (UNI-BLITZ, Vincent Assoc., Rochester, NY).
Theoretical calculations at the Computer Center, Academia Sinica, used the Gaussian 03 program package. All structures and frequency calculations were optimized by DFT at the B3LYP/6-31G(d) level. Final energies were refined by B3LYP/6-311 ++ G(d,p) single-point calculations and B3LYP/6-31G(d) zero-point energy corrections. IR spectra were calculated at B3LYP/6-31G(d) and scaled with a single factor of 0.97.
We thank to Drs. J. L. Beauchamp, S. Castro, R. G. Cooks, J. C. Jiang, and C. A. Schalley for helpful discussions and the Academia Sinica and the National Institute of General Medical Science, NIH (GM16609) for financial support.