Probing alkali metal–π interactions with the side chain residue of tryptophan

Abstract

Feeble forces play a significant role in the organization of proteins. These include hydrogen bonding, hydrophobic interactions, salt bridge formation, and steric interactions. The alkali metal cation-π interaction is a force of potentially profound importance but its consideration in biology has been limited by the lack of experimental evidence. Our previous studies of cation–π interactions with Na⁺ and K⁺ involved the side arms of tryptophan (indole), tyrosine (phenol), and phenylalanine (benzene) as the arene donors. The receptor system possesses limiting steric constraints. In this report, we show that direct interactions between alkali metals and arenes occur at or within the van der Waals contact distance.

Molecular frameworks are determined by covalent links. The energies of such linkages are typically 100–200 kcal/mol depending on the extent of unsaturation in the bonds. Strong as these bonds are, numerous other forces affect overall molecular structure in any flexible molecule. Feeble forces exert a modest individual influence on chemical structure but their cumulative effect can be profound (1). An example is the formation of a hydrogen bond, such as the N-H⋅⋅⋅O interaction common in peptides. Individually, the stabilization is typically less than 10 kcal/mol. In a molecular assembly in which there are a dozen or more weak interactions, the energetic sum can exceed that of a covalent bond.

Additional interactions whose cumulative effect may be great include salt bridge formation and van der Waals contacts. The charge–charge (pole–pole) interaction that exists in the contact between, for example, aspartate and lysine (—COO⁺⁻H₃N—), is estimated to amount to ∼10 kcal/mol each. Hydrophobic interactions or van der Waals contacts are often estimated to be worth 1–2 kcal/mol in stabilization energy although the origins and magnitudes of these effects remain more difficult to assess.

The alkali metal cation–π interaction has been known as an additional feeble force for many years but its importance remains poorly documented. Most of the experimental evidence that is available falls into three categories. These are computational experiments (2–10), mass spectrometric studies (11–15), and solid-state structure analyses of charged species. Examples of the latter abound and date from the 1960s (16). For example, the solid-state structure of RbBPh₄ appeared in a Russian crystallographic journal in 1962 (17). The structures of numerous lithium arene salts have also been reported during the ensuing four decades (18–21). Almost no alkali metal cation–π complexes involving a neutral arene (22–24) were available until quite recently, however (25).

The major focus of our previous efforts (26) has been on the neutral arenes benzene (27), phenol (28), and tryptophan (29). These comprise three of the four aromatic amino acid side chains. We have recently expanded our efforts to include the extremely interesting neutral double (30) and triple (31) bonds but these clearly do not have a counterpart among the essential amino acids.

Among the 20 common amino acids, four side chains possess aromatic residues. These are phenylalanine (Phe, F), tyrosine (Tyr, Y), tryptophan (Trp, W), and histidine (His, H). The arenes are, respectively, benzene, phenol, indole, and imidazole. The latter is expected to be both electron-poor and a σ-donor to Lewis acids and we have thus far not studied it. Calculations suggest that the order of Lewis basicity for the other three arenes is indole > phenol > benzene (2–10).

Materials and Methods

General.

¹H-NMR were recorded at 300 MHz in CDCl₃ and are reported in ppm (δ, Me₄Si). Melting points (mps) were determined on a Thomas Hoover (Philadelphia) apparatus in open capillaries and are uncorrected. All reactions were conducted under dry N₂ unless otherwise stated. All reagents were the best (non-LC) grade commercially available and were used without further purification. Combustion analyses were performed by Atlantic Microlab, Atlanta, GA, and are reported as percents.

N-(2-(3-Indolyl)ethyl)aza-12-crown-4, 7.

3-(2-Bromoethyl)indole (0.67 g, 3 mmol), aza-12-crown-4 (0.53 g, 3 mmol), and Na₂CO₃ (1.59 g, 15 mmol) were stirred in refluxing CH₃CN (25 ml) for 24 h. After work up and column chromatography (0–5% Et₃N, Me₂CO on silica gel), 7 was obtained as an oil. The crude material was kept under vacuum overnight during which time the product (0.7 g, 73%) solidified. After crystallization from CH₃OH, the product had an mp of 122–123°C. ¹H-NMR: 2.83 (t, 4H, J = 4.8, −CH₂NCH₂−), 2.86–2.96 (m, 4H, −CH₂CH₂−), 3.66–3.71 (m, 12H, −CH₂OCH₂−), 7.08–7.10 (m, 1H, indole-H5), 7.13 (s, 1H, indole-H2), 7.18 (t, 1H, J = 7.2, indole-H6), 7.35 (d, 1H, J = 7.5, indole-H7), 7.59 (d, 1H, J = 7.5, indole-H4), 8.03 (s, 1H, indole-H1). ¹³C-NMR: 22.98, 55.02, 57.60, 70.41, 70.47, 71.19, 111.04, 114.52, 118.75, 119.10, 121.80, 121.83, 127.59, 136.18. Anal. Calcd for C₁₈H₂₆N₂O₃, C, 67.90; H, 8.23; N, 8.80. Found: C, 68.01; H, 8.20; N, 8.86.

N-(2-(3-Indolyl)ethyl)aza-15-crown-5, 8.

3-(2-Bromoethyl)indole (0.67 g, 3 mmol), 1-aza-15-crown-5 (0.55 g, 2.5 mmol), and Na₂CO₃ (3.18 g, 30 mmol) were stirred in refluxing CH₃CN (40 ml) for 48 h. After work up, column chromatography (0–5% Et₃N in Me₂CO on silica gel) afforded 8 as a yellow oil (0.85 g, 93%). IR: 3183, 3053, 2868, 1456, 1353, 1233, 1119 cm⁻¹. ¹H-NMR: 2.86–2.90 (m, 8H, −CH₂CH₂N−, −CH₂NCH₂−), 3.65–3.71 (m, 16H, −CH₂OCH₂−), 7.08–7.20 (m, 3H, indole-H2,5,6), 7.35 (d, 1H, J = 7.5, indole-H7), 7.59 (d, 1H, J = 7.5, indole-H4), 8.05 (s, 1H, indole-H1). ¹³C-NMR: 23.17, 54.69, 57.59, 70.19, 70.47, 71.07, 111.04, 114.47, 118.75, 119.09, 121.76, 121.83, 127.60, 136.16. Anal. Calcd for C₂₀H₃₀N₂O₄, C, 66.27; H, 8.34; N, 7.73. Found: C, 66.33; H, 8.43; N, 7.44.

N-(2-(3-Indolyl)ethyl)aza-18-crown-6, 9.

3-(2-Bromoethyl)indole (0.34 g, 1.5 mmol), 1-aza-18-crown-6 (0.26 g, 1 mmol), and Na₂CO₃ (1.06 g, 10 mmol) were stirred in refluxing butyronitrile (20 ml) for 48 h. After work up, column chromatography (0–5%Et₃N in Me₂CO on silica gel) afforded 9 as a yellow oil (0.30 g, 73%). IR: 3204, 3053, 2871, 1458, 1454, 1354, 1111 cm⁻¹. ¹H-NMR: 2.83–2.94 (m, 8H, −CH₂CH₂N−, −CH₂NCH₂-), 3.61–3.72 (m, 20H, −CH₂OCH₂−), 7.06–7.18 (m, 3H, indole-H2,5,6), 7.33 (d, 1H, J = 7.5, indole-H7), 7.57 (d, 1H, J = 7.5, indole-H4), 8.48 (s, 1H, indole-H1). ¹³C-NMR: 22.99, 54.17, 56.42, 69.88, 70.34, 70.61, 70.70, 70.81, 111.07, 114.05, 118.57, 118.84, 121.49, 122.29, 127.57, 136.13. Anal. Calcd for C₂₂H₃₄N₂O₅, C, 65.00; H, 8.43; N, 6.89. Found: C, 65.26; H, 8.19; N, 6.87.

Crystallization of 7⋅NaI.

Equivalent amounts of 7 and NaI were mixed and dissolved in boiling acetone. After vapor diffusion of this solution with hexane for several weeks, 7⋅NaI was obtained as colorless rhombohedroids (mp 205–206°C).

Crystallization of 8⋅KPF₆.

Equivalent amounts of 8 and KPF₆ were mixed and dissolved in boiling ethyl acetate. After vapor diffusion of this solution with hexane for several weeks, 8⋅KPF₆ was obtained as colorless parallelepipeds (mp 174–175°C).

Crystallization of 8⋅NaBPh₄.

Equivalent amounts of 8 and NaBPh₄ were mixed and dissolved in boiling, anhydrous ethanol. After vapor diffusion of this solution with hexane for several weeks, 8⋅NaBPh₄ was obtained as colorless parallelopipeds (mp 177–178°C).

Crystallization of 9⋅KPF₆.

Equivalent amounts of 9 and KPF₆ were mixed and dissolved in boiling ethanol. Vapor diffusion of this solution with hexane for 1 day afforded colorless needles (mp 148–149°C).

X-Ray Crystallography.

Intensity data for all crystals reported here were collected at 173(1) K on a Bruker SMART CCD diffractometer (θ scan mode Mo-K_α radiation, λ = 0.7107 Å). Data were corrected for absorption by using the program sadabs (32). Structure solution and refinement proceeded similarly for all structures [SHELX-97 software (33) with the x-seed interface http://x-seed.net; ref. 34]. Direct methods yielded all nonhydrogen atoms of the asymmetric unit. These atoms were refined anisotropically (full-matrix least-squares method on F²). Hydrogen atoms were placed in calculated positions with their isotropic thermal parameters riding on those of their parent atoms. Figs. 1 and 2 were prepared with x-seed and pov-ray (http://www.povray.org; ref. 35).

Figure 1.

Structures of receptor complexes. First row, KI complexes of 1, 2, and 3. Second row, 5⋅NaI, 5⋅KI, 6⋅NaPF₆. Third row, 7₂⋅NaI, (8⋅KPF₆)₂. Fourth row, 8⋅NaBPh₄, 9⋅KPF₆.

Figure 2.

Top and side views of 8⋅NaPBh₄.

Crystal Data for 7⋅NaI.

C₃₆H₅₂O₆N₄NaI: M = 786.71, colorless rhombohedroids, 0.30 × 0.25 × 0.20 mm³, orthorhombic, P2₁2₁2₁ (No. 19), a = 12.7533(17), b = 15.003(2), c = 19.558(3) Å, V = 3742.3(9) ų, Z = 4, D_calcd = 1.396 g cm⁻³, F(000) = 1632, λ (Mo-Kα) = 0.71073 Å, T = 173 K, reflections collected/unique: 23,675/8,224 [R(int) = 0.0289]. Final goodness of fit = 0.988, R1 = 0.0285, wR2 = 0.0569, R indices based on 7,242 reflections with I > 2σ(I) (refinement on F²), 433 parameters, 0 restraints, Lorentz polarization and absorption corrections applied, μ = 0.916 mm⁻¹.

Crystal Data for 8⋅KPF₆.

C₂₀H₃₀O₄N₂KPF₆: M = 546.53, colorless parallelepipeds, 0.45 × 0.35 × 0.20 mm³, monoclinic, P2₁/c (No. 14), a = 10.721(2), b = 11.716(2), c = 19.263(3) Å, β = 93.727(3) °, V = 2414.3(6) ų, Z = 4, D_calcd = 1.504 g cm⁻³, F(000) = 1136, λ (Mo-Kα) = 0.71073 Å, T = 173 K, reflections collected/unique: 14,266/5,260 [R(int) = 0.0278]. Final GooF = 1.049, R1 = 0.0586, wR2 = 0.1528, R indices based on 3,959 reflections with I > 2σ(I) (refinement on F²), 307 parameters, 0 restraints, Lp and absorption corrections applied, μ = 0.362 mm⁻¹.

Crystal Data for 8⋅NaBPh₄.

C₄₄H₅₀O₄N₂BNa: M = 704.66, colorless parallelepipeds, 0.30 × 0.25 × 0.20 mm³, triclinic, P-1 (No. 2), a = 11.985(1), b = 16.956(2), c = 19.402(2) Å, α = 78.245(2), β = 89.663(2), γ = 85.009(2) °, V = 3845.3(6) ų, Z = 4, D_calcd = 1.217 g cm⁻³, F(000) = 1504, λ (Mo-Kα) = 0.71073 Å, T = 173 K, reflections collected/unique: 24,592/16,675 [R(int) = 0.0420]. Final GooF = 0.928, R1 = 0.0642, wR2 = 0.1207, R indices based on 7,837 reflections with I > 2σ(I) (refinement on F²), 937 parameters, 0 restraints, Lp and absorption corrections applied, μ = 0.086 mm⁻¹.

Crystal Data for 9⋅KPF₆.

C₂₂H₃₄O₅N₂KPF₆: M = 590.58, colorless needles, 0.35 × 0.25 × 0.20 mm³, monoclinic, P2₁/c (No. 14), a = 11.524(1), b = 15.178(1), c = 15.624(1) Å, β = 95.913(2) °, V = 2718.4(4) ų, Z = 4, D_calcd = 1.443 g cm⁻³, F(000) = 1232, λ (Mo-Kα) = 0.71073 Å, T = 173 K, reflections collected/unique: 17,051/6,000 [R(int) = 0.0623]. Final GooF = 0.859, R1 = 0.0486, wR2 = 0.0869, R indices based on 3,012 reflections with I > 2σ(I) (refinement on F²), 371 parameters, 0 restraints, Lp and absorption corrections applied, μ = 0.330 mm⁻¹. Four of the PF₆ fluorine atoms were disordered over two distinct sites and modeled accordingly.

Results and Discussion

The Experimental Paradigm and Strategy.

Most of the reported arenes that exhibit alkali metal cation–π interactions are anionic (36, 37). In the majority of cases, the anion was produced by reduction, using the alkali metal in question. Such reactions and the resultant aromatic anions are unlikely to exist in proteins; the arenes are expected to be neutral. Our experimental approach was therefore to use the lariat ether system (38, 39) that we developed a number of years ago to appropriately position the cation and the arenes. Because the arenes are present on flexible side arms, they can either interact with the ring-bound cation or not, according to the ultimate energetics of the overall system. Diaza-18-crown-6 accommodates both Na⁺ and K⁺ within its central hole and two arene-terminated side arms can readily be attached to the nitrogen atoms. Symmetrical sidearm substitution may be achieved either by alkylation of the parent diazamacrocycle or by a single-step approach that begins with the amine form of the incipient sidearm (Scheme S1).

Scheme 1.

We prepared compounds 1-3 and formed a variety of alkali metal salt complexes from them. The reported structures of 1⋅KI, 2⋅KI, and 3⋅KI are shown in the first row of Fig. 1 (26). All three complexes show clear evidence for alkali metal cation–π complexation. Both benzene and phenol in the complexes of 1 and 2 are in the “sandwich” arrangement; the two arenes are parallel (or nearly so), and potassium is on or near the vertical line connecting their centroids.

The observation that 1-3 all showed apparent cation–π complexation raised the concern that crystal-packing forces might account for the structures observed. We therefore prepared 4, the decafluorophenyl analog of 1. If steric forces controlled complexation, a complex analogous to 1⋅KI should form. Instead, iodide was in contact with ring-bound K⁺ in 4⋅KI as expected in the absence of a cation–π interaction (27).

Sidearm triple- and double-bond complexes were also prepared and isolated. They (5⋅NaI, 5⋅KI, and 6⋅NaPF₆) are shown in the second row of Fig. 1. They serve to illustrate the flexibility of the cation–π interaction but will not be discussed further in the present report.

Complexes of Indole-Sidearmed Macrocycles.

The K⁺ complex of receptor 3 differs from complexes of 1 and 2 in that the pyrrolo subcyclic unit of indole coordinates the cation rather than the benzene ring. Theoretical calculations suggest that the benzo, rather than the pyrrolo, unit is preferred for coordination (2–10). Of course, these calculations are for an unencumbered indole in the gas phase. In 3⋅KI, the thickness of the macroring hinders the approach of the indole to the cation. Although the pyrrolo residue is calculated to be a weaker donor than is the benzo group, its closer approach to the cation may be more favorable than the better donor interacting at a greater distance. If so, this means that tryptophan is potentially a very flexible donor in the low polarity environment of globular proteins.

The strategy for the studies described here was to decrease the macroring size and/or the number of indole-terminated side arms. The goal was to complex either Na⁺ or K⁺ by a macrocycle too small for cation insertion. A “perching” conformation would permit the side arm free access to the indole. This conformation, we hypothesized, would permit a closer examination of the cation–π interaction between indole and M⁺. The results of these studies follow (Scheme S2).

Scheme 2.

Syntheses.

2-(3-Indolyl)ethyl derivatives of aza-12-crown-4 (7), aza-15-crown-5 (8), and aza-18-crown-6 (9) were prepared by alkylation of the macrocycle with 3-(2-bromoethyl)indole (Na₂CO₃, CH₃CH₂CH₂CN solvent). The respective macrocycles were obtained in 73 (7, white powder), 93 (8, oil), and 73% (9, oil) yields, respectively. The alkali metal salt complexes were obtained by vapor diffusion methods. Equivalent amounts of receptor and salt (0.1 mmol each) were dissolved in a boiling solvent such as ethanol or acetone (about 10 ml), and the solution was transferred to two or three vials (5 ml). Each vial was placed in a bottle with a cap that was filled with hexane (5 ml) then allowed to stand at ambient temperature for several days or weeks to afford crystals suitable for x-ray analysis.

N-(2-(3-Indolyl)ethyl)aza-12-crown-4 Complex of NaI, 7₂⋅NaI.

We have reported the solid-state structure of (aza-12-crown-4)₂⋅NaI (40). Numerous (12-crown-4)₂⋅MX complexes have been reported as well (41–78). Two-to-one complexes typically form when the cation is too large to fit within the hole of the macrocycle. An added advantage of the 2:1 structure is that the coordination number of the cation is increased and metal-donor distances are extended as well.

The complex 7₂⋅NaI is a nonsymmetrical complex in which Na⁺ is octacoordinated by six oxygen and two nitrogen atoms. The indolylethyl side arms are turned away from the complex and no H-bond interaction involving their NH residues is apparent. The average N-Na (2.70 Å) bond distances are slightly longer than the average O-Na (2.56 Å) lengths, as expected. The O-Na⁺ distances are slightly longer than the average bond distances for 8-coordinated Na⁺ in the known (12-crown-4)₂⋅NaClO₄: 2.495 Å (41). Both values are typical for Na-donor distances in complexes of this sort. The mean planes of the two macrocyclic rings are each separated from the cation by 1.64 Å.

An interesting feature of the structure is that the two rings are offset by ∼52^o, close to the 45^o value expected for a staggered arrangement. The two side arms are on adjacent positions on the macrorings. In the solid-state structure of (aza-12-crown-4)₂⋅NaI, the nitrogen atoms were offset from each other by 43.3^o. In that work, it seemed that a weak (3.73 Å) H-bond interaction between the hydrogen atoms on nitrogen and iodide accounted for this orientation. In the present case, there is no hydrogen atom on nitrogen. No H-bond interaction involving the indolyl hydrogen is apparent, and the Na⁺-I distance is >8 Å.

N-(2-(3-Indolyl)ethyl)aza-15-crown-5 Complex of NaBPh₄ and KPF₆, 8₂⋅2KPF₆.

A similar steric situation exists for Na⁺ complexes of 12-membered crowns and for K⁺ complexes of 15-crown-5. The cation cannot fit within the macrocycle in either case and the cation typically “perches” on the ring donor atoms. 15-Crown-5 does, however, readily complex Na⁺ in the “nesting” configuration. N-(2-(3-Indolyl)ethyl)aza-15-crown-5, 8, gave complexes with both Na⁺ and K⁺. As expected, the K⁺ complex was dimeric and the Na⁺ complex was monomeric. Both, however, revealed interesting properties.

Aza-15-crown-5 derivative 8 forms a dimeric complex with two molecules of KPF₆. Typically, 15-crown-5 complexes K⁺ to form sandwich structures in which a single cation is held between two macrocycles. In the present case, two lariat ethers, two cations, and two PF ions form an extended stack. Each K⁺ is bound by five macroring donors, two fluorides from one PF ion, and one fluorine from a second PF. As shown at the right of the third row in Fig. 1, the indolylethyl side arms are on opposite sides of the complex. The symmetrical complex has average O-K distances of 2.80 Å and N-K contacts of 2.95 Å. The F-K-F interaction involving the single anion is unsymmetrical; K-F contact distances are 2.799 and 3.248 Å. The third K-F contact comes from the adjacent anion and is 2.715 Å. No H-bond interaction involving the indole nitrogen is apparent.

It is interesting to note that a sandwich complex is also observed for indole-sidearmed aza-15-crown-5 derivative <15N>COCH₂CH₂CH₂Indole (not shown). In this case, the side arm is linked to the macrocycle by an amide residue, which flattens the macroring, and the amide donor is turned outward. The complex with Na⁺ is a crown₂⋅Na⁺ sandwich in which the cation is in contact with the eight macroring oxygen atoms of two crowns but not with either amidic nitrogen. The anion in the case described here was tetraphenylborate and did not contribute any obvious interaction to the complex.

The NaBPh₄ complex of 8 is monomeric and shows clear and convincing evidence for Na⁺–π interaction. The complex is illustrated in Fig. 2. Fig. 2 Top shows a view of the macrocycle and cation (tetraphenylborate anion excluded for clarity) from above. The interaction between Na⁺ and the pyrrolo centroid of indole is apparent. Interactions between Na⁺ and the ring donor atoms are strong. The average Na-O distance is 2.37 Å and the Na-N contact is 2.60 Å. The Na-O distances are shorter than those noted above for other complexes, but the coordination number for Na⁺ is typically 8 in those cases and the individual interactions are correspondingly weaker.

The Na⁺-centroid distance is 2.60 Å. The two views of the complex shown in Fig. 2 illustrate that indole is approximately parallel to the plane of the macroring and that the pyrrolo unit, rather than the benzo residue of indole, is positioned directly over Na⁺. This arrangement differs from that observed for complexes of 3, in which the arene was tilted into the macroring cavity. Cation–π contact with the cation was unsymmetrical in those cases, the cation and C2 being closest.

N-(2-(3-Indolyl)ethyl)aza-18-crown-6 Complex of KPF₆, 9⋅KPF₆.

Potassium fits almost perfectly within an 18-membered ring crown ether. The D_3d arrangement of the macroring is nearly planar, and the cation resides near the mean plane of the donors. In 9⋅KPF₆ the K⁺ ion is displaced from the calculated center of the macroring by 0.129 Å. The O-K distances are typical of 18-crown-6 complexes, averaging 2.76 Å. The N-K distance is 2.99 Å, and the closest contact between K⁺ and the indole is with C2 at a distance of 3.22 Å. Indole is offset from the center of K⁺ as indicated by the K-C3 and K-N(indole) distances of 3.61 and 3.57 Å, respectively.

Further solvation is provided to ring-bound K⁺ by the PF anion. Although it is disordered, the F-K contacts are both ≈3 Å. This value corresponds to the average observed for (8⋅KPF₆)₂. If the centroid is counted as a single donor element, K⁺ is 9-coordinate in this complex.

Comparison of Indole-Terminated Receptor Complexes with Alkali Metals.

We have previously reported the complexation of both Na⁺ and K⁺ by 3. The M⁺-arene distances for the NaI and KI complexes were, respectively, 3.50 and 3.45 Å. When the anion was PF, the K-arene distance was 3.48 Å. In these three cases and when the anion was SCN⁻, the sidearm indoles were tilted inward such that the shortest distances were observed with C2. We concluded that the distance of approach between cation and indole reflected, in part, the steric volume of the macroring. If this speculation is correct, we would expect the closest M⁺-arene contact to be observed for a 1:1 K⁺ complex of 8. Unfortunately, the only K⁺ complex that has yet crystallized is 8₂⋅2KPF₆.

The monomeric complex 8⋅NaBPh₄ showed the perching conformation between macroring and cation. In this case, the approach of indole to the cation was not sterically limited; the Na-arene centroid distance was 2.60 Å. The van der Waals radius of aromatic carbon is reported to be 1.72–1.80 Å (79). We have previously used the value of 1.18 Å as the radius for octacoordinated Na⁺. This value certainly applies to 3⋅NaI, but it is difficult to know what coordination number to assign to a 5-membered ring centroid. In 8⋅NaBPh₄, Na⁺ contacts 4 oxygens, nitrogen, and the pyrrolo subunit of indole. If the latter is counted as 1 or 5, the coordination number ranges from 6–10. The radius reported for 6-coordinate Na⁺ is 0.99, therefore the range of contact distances limited by van der Waals volumes is 2.71–2.98 Å. The observed contact is only 2.60 Å, suggesting a very strong interaction.

Conclusion

Theoretical studies have predicted that the interaction of an alkali metal cation with indole would afford a benzo, rather than pyrrolo, complex when a cation–π interaction was observed. We have now expanded the family of structures that reveal cation to indole–π interactions. In all cases, the pyrrolo residue is the favored donor site in this experimental receptor system. In the clearest example of such an interaction, i.e., 8⋅NaBPh₄, the cation is located at a distance less than the sum of the van der Waals radii and located precisely in the center of the pyrrolo subunit. Steric factors may once again be playing a role in this receptor system, but we conclude that indole and, by extension, tryptophan may be a flexible donor for Na⁺ and K⁺ in proteins that can afford substantial stabilization, especially in a low-polarity environment.