A Relationship between Amide Hydrogen Bond Strength and Quinone Reduction Potential; Implications for Photosystem 1 and Bacterial Reaction Center Quinone Function

Abstract

A series of 11 simple phylloquinone derivatives, each lacking the extended phytyl side chain but featuring H-bond donor amides at one or both peri positions, were prepared and some salient physical properties were measured. A correlation between both IR frequency and NMR peak position, as indicators of internal H-bond strength, and the quinone half-wave reduction potential, was observed. These data are consistent with the prevailing hypothesis that quinone carbonyl H-bonding in general, and stronger H-bonds in particular, favorably bias the endogenous quinone's electrochemical potential toward easier reduction.

The complete structural characterization of the photosynthetic reaction centers Photosystem 1 (PS1)¹ and bacterial reaction center (bRC)² has enabled unparalleled advances in understanding the mechanistic underpinnings of photosynthesis.³ Even as Angstrom-level features have come into focus, the relationship between structure and function, in many cases, has yet to be clarified. One pivotal feature of the light-driven electron transfer from each system's primary chlorophyll acceptor to the terminal repository, a quinone (bRC) or Fe₄S₄ cluster (PS1), involves reaction through an intermediate quinone transfer junction (ubiquinone (1) in bRC and phylloquinone (2) in PS1), Fig. 1. The critical reduction potentials of these quinones are thought to be responsive to local electrostatic effects, with proximal negatively charged amino acid side chains and a juxtaposed negatively charged iron-sulfur cluster contributing to a lower value (harder to reduce) for phylloquinone in PS1 compared with an adjacent and reduction-facilitating Fe²⁺ site in bRC.⁴ In addition to these charge effects, both H-bonding and π-stacking also are cited as influential factors in determining the quinone reduction potential. Ubiquinone is pinioned by the protein matrix between two H-bonds as shown, and so it is not surprising that it's estimated reduction potential in vivo is higher than (more easily reduced than) the same quinone in the non-protic media DMF. Whereas the inexact modeling of a protein interior by DMF does not allow too firm of a conclusion to be drawn, it is likely that at least some of the diminished reduction potential of the in vivo version can be attributed to dual H-bond activation of the quinone's carbonyls. It is quite surprising, then, that the analogous quinone 2 of PS1 exhibits a reduction potential that is substantially lower than various phylloquinone models in DMF solution, given that it, too, presumably benefits from the single H-bond shown.⁵ The role that the enveloping protein plays in modulating the reduction potential of this quinone has been a matter of speculation, and one theory in current ascendancy focuses on the protein-mediated modulation of a π-stacking interaction with the adjacent indole ring of TrpA697.⁶ From this perspective, the π-stacking interaction is viewed as energetically favorable for the parent quinone, but destabilizing (and yet structurally unavoidable) for the reduced species. Nevertheless, the contributions of the singular H-bond to 2 cannot be ignored, especially when circumstantial evidence suggests that this H-bond, in contrast to the dual H-bonded arrangement exhibited for the analogous quinone 1 in bRC, is particularly strong as a consequence of an extended H-bonding network provided by the protein matrix and associated waters.⁷ Thus, the overarching question emerges, how can amide H-bonding, and more precisely, the strength of amide H-bonding, influence a quinone's reduction potential?

Figure 1.

Structures and figures of merit for the quinones of bRC (left) and PS1 (right). All reduction potentials are reported vs. the normal hydrogen electrode.¹⁰

The relationship between H-bond strength and quinone reduction potential¹¹ was probed using simple model 5-amido and 5,8-diamidonaphthoquinones. This approach is precedented by a long history of studying the relationship between the reduction potential of ortho hydroxy- and amino substituents on benzo- and naphthoquinones,¹² and peri-positioned hydroxy, and in a few examples amino,^13a groups on naphtho- and 9,10-anthroquinones.¹³ The upshot of these earlier studies is that a peri-substituted hydroxyl (i.e., 5-hydroxy or 5,8-dihydroxy in the naphthoquinones series) renders the quinone more easily reduced as a consequence of the LUMO-lowering effect of a geometrically favorable¹⁴ internal H-bond. The ortho-hydroxylated (aminated) benzoquinones, on the other hand, do not consistently follow this trend. Unfortunately, the H-bond strength of the peri-hydroxyls cannot be modulated, and so these types of substrates cannot be employed to test the hypothesis that reduction potential scales with H-bond strength. However, the use of a series of naphthoquinones bearing peri-positioned amide units with deliberately varying H-bond donor capabilities can overcome this limitation and represents one approach by which the effect of varying H-bond strength on quinone reduction potential can be probed.

This premise was explored by synthesis of both unsymmetrically and symmetrically peri-functionalized 2,3-dimethylnaphthoquinones 5−9 and 11−13, respectively, and correlation of salient molecular attributes reflecting H-bond strength with their half-wave reduction potentials. The syntheses of the target quinones were accomplished by modification of Diels Alder/oxidation chemistry reported by Fillion¹⁵ using the mono amidobutadiene 4¹⁶ and the bis amidobutadiene 10,¹⁷ Scheme 1 (see Supplementary Data for details). In this chemistry, the efficiency of naphthoquinone formation was increased over prior work by consolidating the oxidation-cycloaddition-oxidation sequence into a single pot operation. Simply exposing the dihydroquinone 3 and the requisite diene to excess oxidant (MnO₂) furnished good yields of the desired naphthoquinone products 5 and 11. These BOC-protected intermediates then served as branch points for the preparation of all of the other aminonaphthoquinone derivatives. All quinone products were characterized fully,¹⁸ including single crystal X-ray for 8a (Supplementary Data).¹⁹ The X-ray derived structure clearly indicated the dominating effect of the N–H--O=C H-bond, which held the amide side chain in a coplanar orientation with the quinone framework.

Scheme 1.

Synthesis of the target naphthoquinones 5− 9, and 11−13; X-ray structure of 8a. The thermal ellipsoid diagram (50% probability) was obtained with Bruker XSHELL software.

Measurement of half-wave reduction potentials of the substituted naphthoquinones in CH₂Cl₂ led to the data shown in Table 1. These values were obtained under the following standard conditions: Pt disc electrode (1.6 mm diameter), Ag/AgNO₃ reference (0.01 M in CH₃CN), 2 mM in quinone, 100 mM in TBAP, 3 V/s, 0 to −2496 mV range, T = 20 °C. Under these conditions, ferrocenium ion exhibited a reduction potential of +200 mV (lit. 206 mV vs. Ag/AgPF₆)²⁰. The IR frequency of the H-bond, which is well known to scale with H-bond strength, and the ¹H NMR H-bond signal position, a value also related to H-bond strength,²¹ are reported in Table 1 as well. Concentration studies with 8a over the span 1–100 mM revealed that neither the IR nor the ¹H NMR signal values varied significantly (± 1 cm⁻¹, and ± 0.05 ppm, respectively) throughout the experimental range.

Table 1.

Half-wave reduction potentials, salient IR frequencies, and ¹H NMR signal positions for the naphthoquinones ¹⁴.

entry	cmpd. #	R	R1	Red. Pot. (mV) CH2Cl2a	N-H IR CCl4 (cm-1)	N-H 1H NMR C6D6 (δ)
a		H	H	-623	---	---
b	6	NH2	H	-721	3339	not observed
c	7	NHCH3	H	-697	3304	9.29
d	5	NHBOC	H	-551	3263	11.77
e	8a	NHAc	H	-484	3262	11.92
f	8b	NHTFA	H	-347	3109	13.08
g	9a	N(CH3) Ac	H	-560	---	---
h	9b	N(CH3)TF A	H	-554	---	---
i	12	NH2	NH2	-819	3288	not observed
j	11	NHBOC	NHBOC	-419	3240	12.06
k	13a	NHAc	NHAc	-359	3201	12.18
l	13b	NHTFA	NHTFA	-108	3076	13.18

^aAll reduction potentials have been converted to NHE values via the formula: E_(NHE) = E_(AgNO3) + 541 mV; see Handbook of Analytical Chemistry; Meites, L., Ed.; McGraw-Hill: New York, 1963.

Examination of these data reveals a clear correlation between the half-wave reduction potential and both H-bond strength metrics. However, a dissection of the two related effects that result when an amide is introduced onto the naphthoquinone framework is necessary in order to make further sense out of the data.

Both the (1) strong and directional H-bond between N–H and O=C and the (2) contributions of the nitrogen's lone pair via resonance with the carbonyl can play a role. Much precedent from the peri-hydroxylated cases suggest that H-bond activation of the carbonyl exerts the dominant influence, between these two interactions, on quinone reduction potential.^12a,13a-c,e In an effort to probe the relative contributions of these two features in the amide substituted series examined herein, the controls 6, 7, 9a, and 9b were evaluated. In the absence of any electron-engaging carbonyl (amide) fu nctions, the nitrogen's lone pair certainly has the expected effect on the quinone's reduction potential, making it harder to reduce by 98 mV; compare entry a (R = H) with entry b (6, R = NH₂). In this instance, the N-H--O=C bond (IR absorption at 3339 cm⁻¹) can only be described as modest at best, and so this comparison is closest to a case where the H-bonding is turned off whereas the nitrogen lone pair contribution is turned on. Designing the complementary control (H-bonding turned off, nitrogen lone pair turned on) cannot be assured in this system due to uncertainty in the degree of overlap between the nitrogen's lone pair and the quinone, but entries 9a and 9b, where H-bonding has been abolished by methyl incorporation, come closest to approximating this goal. In these instances, the quinone reduction potential is depressed compared to the H-bonding analogues 8a/8b by 76 mV (8a vs. 9a) and 207 mV (8b vs. 9b), respectively. In fact, the large difference between these values appears to be attributable primarily to the increased strength of the H-bond in the NHTFA case 8b, as the reduction potentials of the H-bond incapable species 9a and 9b are essentially equivalent. From these data, it is apparent that whereas the nitrogen lone pair definitely contributes to depression of the quinone's reduction potential, its effects are relatively smaller that the H-bond effect and consistent throughout the relevant substrates. Therefore, the differences in measured reduction potential between the different acylated aminoquinone substrates should to a large measure reflect the impact of the H-bond, much as they do in the more well-studied peri-hydroxylated series.

Overall, the data for substrates 5, 8a, 8b, 11, 13a, and 13b do support the contention that quinone reduction potential correlates with the strength of the H-bond, but the imperfect measures of H-bond strength used for this correlation do little to encourage linearity. Double activation of the quinone function leads to increases in the reduction potential, as expected, but the amount of increase does not scale with increased H-bond strength. Thus, introducing an additional H-bond in the BOC species leads to an increase in reduction potential of 132 mV (entries d vs. j), whereas in the NHAc series (stronger H-bond), the difference is 125 mV (entries e vs. k) and in the NHTFA analogues (strongest H-bond), 239 mV (entries f vs. l). Extrapolation from these model compounds to photosynthetically relevant quinones embedded in protein matrices must be viewed with caution, but at the very least, the following hypotheses are supported by the data: (1) stronger N–H--O=C hydrogen bonds, as might be provided by H-bond networks in the enveloping proteins, makes the quinones more easily reduced, and (2) simultaneous N–H--O=C hydrogen bonding to both quinone carbonyls makes the quinones even more easily reduced than simple additivity might suggest.

Supplementary Material

01

Supplementary Data. Experimental procedures and full spectral data for 5−13. Crystal structure data for 8a, and a plot of ¹H NMR and IR signal position as a function of concentration for 8a.