Skip to main content
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2008 Sep 15.
Published in final edited form as: J Am Chem Soc. 2006 May 31;128(21):7015–7024. doi: 10.1021/ja0613714

Polymorphism of the Signaling Molecule c-di-GMP

Zhaoying Zhang 1, Seho Kim 1, Barbara L Gaffney 1, Roger A Jones 1,*
PMCID: PMC2535936  NIHMSID: NIHMS63147  PMID: 16719482

Abstract

Using UV, CD, and NMR, we demonstrate that the important bacterial signaling molecule involved in biofilm formation, cyclic diguanosine monophosphate (c-di-GMP), exists as a mixture of five different but related structures in an equilibrium that is sensitive both to its concentration and to the metal present. At the lower concentrations used for UV and CD work (0.05 – 0.5 mM) Li+, Na+, Cs+, and Mg2+ favor a bimolecular self-intercalated structure, while K+, Rb+, and NH4+ favor formation of one or more guanine quartet complexes as well. At the higher NMR concentrations (∼30 mM), the bimolecular structures associate and rearrange to a mixture of all-syn and all-anti tetramolecular and octamolecular quartet complexes. With K+ the octamolecular complexes predominate, while with Li+ the tetramolecular and octamolecular quartet complexes are present in approximately equal amounts, along with the bimolecular structure. We also find that both guanine amino groups in c-di-GMP are essential for formation of the quartets, since substitution of inosine for one guanosine allows formation of only the bimolecular structure. Further, two molecules of c-di-GMP tethered together are constrained in such a way that limits their ability to form these quartet complexes. The polymorphism we describe may provide different options for this signaling molecule when performing its functions in a bacterial cell, with K+ and its own local concentration controlling the equilibrium.

Introduction

The importance of cyclic diguanosine monophosphate, c-di-GMP, 1 (Fig. 1), as a ubiquitous intracellular bacterial signaling molecule responsible for regulating biofilm formation, cell differentiation, and a variety of pathogenic processes has recently been recognized.1 Biofilms are complex, multicellular aggregations of bacteria attached to surfaces and encased in a matrix of polysaccharides.2 Formation of such communities is an ancient adaptation that allows bacteria to survive a range of environmental challenges in a protected state. The medical, industrial, and environmental threat of biofilms for human life is profound, since they are far more resistant than free-living bacteria to antimicrobials as well as to natural host defense mechanisms.3 Not only are antibiotics typically ineffective in treating chronic infections protected by biofilms, but low doses of aminoglycosides have recently been shown to actually induce biofilm formation in Pseudomonas aeruginosa and Escherichia coli through a mechanism controlled by c-di-GMP.4

Figure 1.

Figure 1

Structures of c-di-GMP, 1, c-GMP-IMP, 2, and oligoethyleneoxide tethered c-di-GMP, 3a (n=4) and 3b (n=12).

Specific c-di-GMP cyclases with GGDEF domains in many different bacteria are responsible for formation of c-di-GMP from two molecules of GTP,5,6 and a crystal structure has been determined for the cyclase from Caulobacter crescentus.7 c-di-GMP levels in P. aeruginosa, E. coli, Vibrio cholerae, Salmonella enterica, and Yersinia pestis have been shown to be inversely regulated by such cyclases along with specific phosphodiesterases associated with EAL domains.8,9 Many bacterial proteins contain both GGDEF and EAL domains,10 and in C. crescentus, a phosphodiesterase with an active EAL domain has recently been found to be regulated by the binding of GTP to its neighboring catalytically inactive GGDEF domain.11 Thus, control of c-di-GMP concentrations in bacteria is extremely complex, and may also involve gradients as well as spatial localization.12

As an intracellular signaling molecule, c-di-GMP is known to regulate cell mobility,5 production of adhesive components,8,13 and other features of multicellular behavior in response to environmental cues.12 It also plays an important role in the pathogenic virulence of Salmonella in mice14 and virulence gene expression in V. cholerae.15 A group of extracellular signaling molecules termed autoinducers also regulate biofilm formation and virulence in bacteria through a process known as quorum sensing.16 Examples include acyl homoserine lactones,17 modified oligopeptides,18 modified quinolones,19 and a possible universal autoinducer, AI-2.20 The integration of these extracellular signals with the c-di-GMP pathways is only now being explored.21 To date the molecular mechanisms for these complex pathways remain unknown and represent a major challenge. Although several crystal structures for c-di-GMP published some time ago revealed a self-intercalated arrangement of two molecules,22,23 structures in solution have not been determined. Understanding the structure and behavior of c-di-GMP in solution would provide valuable insight into how it might interact with the multitude of other molecules involved in its pathways.

We recently reported a new method for the efficient synthesis of c-di-GMP, 1, as well as UV, CD, and 1D NMR results that demonstrated a monovalent metal ion-dependent polymorphism.24 We found evidence for a stacked structure with Li+ and Na+, and the suggestion of a quartet structure with K+. We now describe UV and CD data for additional metals that further define these forms, and we have initiated more detailed structural studies using 2D NMR. We have also synthesized a mixed dimer containing guanosine and inosine, c-GMP-IMP, 2 (Fig. 1), and found that the elimination of one amino group precluded quartet formation. Further, in an effort to evaluate the consequences of molecular constraints on the behavior of 1, we have connected two units together with a flexible oligoethylene oxide tether of two different lengths, 3a where n=4 and 3b where n=12 (Fig. 1). This constraint also has a significant effect on the equilibrium, shifting it substantially to the self-intercalated structure. The new data support the existence of five distinct but related complexes for 1 that exist in an equilibrium that is sensitive to both concentration and the metal present.

Experimental Procedures

Synthesis

The detailed procedures for synthesis of c-GMP-IMP, 2, and the tethered c-di-GMP, 3a and 3b, are described in the Supporting Information.

UV

UV melting curves were obtained on an Aviv 14 UV Spectrophotometer using 1, 2, 5, and 10 mm path length cells. After preparation in the appropriate buffer, samples were degassed, heated to 80 °C, allowed to cool slowly to room temperature, and then placed in a refrigerator for at least 15 hours. UV absorbance spectra were obtained using samples in the 1 mm cells.

CD

CD spectra were obtained on an Aviv Model 60DS CD Spectrometer using a 1 mm cell. The samples were prepared as described for UV. For time-dependent CD experiments, data at a given wavelength were acquired immediately after the temperature was changed. The effective temperature change rate of the instrument was 10 °C/min.

1D NMR

The 31P NMR spectra were acquired on a Varian Mercury 300 MHz spectrometer and referenced to 10% phosphoric acid in D2O; the 1H NMR spectra were acquired on a Varian Unity 400 MHz spectrometer.

2D NMR

2D NOESY (nuclear Overhauser effect spectroscopy), HMBC (heteronuclear multiple bond correlation), and HMQC (heteronuclear multiple quantum coherence) spectra were acquired at 25 °C on a Varian Inova 500 MHz spectrometer. NOESY data were collected by 4096 (t2) times 512 (t1) complex data points with spectral widths of 8000 Hz in both dimensions and transformed to spectra with 2048 (D1) times 1024 (D2) real data points. The mixing time for NOESY spectra was 150 ms, the number of scans per each t1 increment was 16, and the relaxation delay for each scan was 2 s. Natural abundance 1H-13C HMQC and HMBC were acquired by 2048 (t2) times 256 (t1) complex points with spectral widths of 8000 Hz (t2) and 18000 Hz (t1) and transformed to spectra with 2048 (D1) times 512 (D2) real points. The DEPT transfer delay in HMQC was chosen as 0.5/1JHC with 1JHC = 150 Hz and the transfer delay in HMBC was chosen as 0.5/2,3JHC (2,3JHC = 12 Hz) which is close to multiples of 0.5/1JHC to suppress HMQC peaks. The HMBC pulse sequence and data collection mode (phase sensitive mode) were the same as HMQC with decoupling during 1H data acquisition. The number of scans was 32 for HMQC and 256 for HMBC and the relaxation delay for each scan was 1s. All 2D NMR spectra used 1 s 1H pre-saturation during the relaxation delay to suppress water.

Results

1. c-di-GMP, 1

UV Melts (Fig. 2)

Figure 2.

Figure 2

Normalized UV melting profiles (A) at 250 nm of four different concentrations of the Mg2+ form of 1 in 10 mM Mg(OAc)2 w ith 0.1 M MgCl2, pH 7.0, (B) at 250 nm of the K+ (black), Rb+ (blue), and NH4+ (red) forms of 1 in 10 mM potassium phosphate with 0.1 M KCl, 10 mM rubidium phosphate with 0.1 M RbCl, and 10 mM ammonium phosphate with 0.1 M NH4Cl, respectively, all pH 7.0, and (C) at 305 nm of three different concentrations of the K+ form of 1 in 10 mM potassium phosphate with 0.1 M KCl.

UV melting profiles at 250 nm for four different concentrations of 1 with Mg2+ (Fig 2A) show single conformational transitions similar to those with Li+ and Na+,24 but reflect a more thermally stable structure. First derivative curves for the most concentrated sample with each metal (Supporting Information) show one clear maximum, with Tmax values of 8 °C for Na+, 10 °C for Li+, and 45 °C for Mg2+. The concentration dependence of the transitions with these metals indicates that the structured forms have a molecularity of at least 2.25 Further, the significant hypochromicity of the UV curves indicates that stacking of the guanines plays an important role in the structure.

The UV melting profile at 250 nm for the Rb+ form of 1 (Fig. 2B, blue) shows two broad conformational transitions, somewhat similar to those of the K+ form24 (included in Fig. 2B, black), but with lower stabilities. Tmax values from first derivative curves are 30 °C/54 °C for K+ and <5 °C/20 °C for Rb+. Although the UV melting profile for the NH4+ form appears to show a single conformational transition (Fig. 2B, red), its first derivative curve shows two broad overlapping maxima, with Tmax values of 10 °C/20 °C.

The UV spectrum of the K+ form of 1 displays a distinct shoulder between 295 and 305 nm that is not present in the Li+, Na+ or Mg2+ forms, and is weak in the Rb+ and NH4+ forms (Supporting Information). Since similar shoulders were found to be characteristic of intramolecular guanine quartet structures formed by a series of oligodeoxynucleotides,26 we now report UV melting experiments at 305 nm for a series of three K+ samples with different concentrations (Fig. 2C). As seen with the quartet-forming oligonucleotides,26 the profiles at 305 nm display hyperchromic changes rather than the typical hypochromic changes seen at 250 nm. Together, the UV results support the presence of one or more additional structures for 1 with K+ and Rb+ that contain guanine quartets. Further, since intramolecular guanine quartet oligonucleotides that exhibit hyperchromic behavior all contain some guanines with syn conformations,27 by extension, at least one of the quartet structures of 1 in K+ and Rb+ may also contain some syn guanines.

CD Spectra (Fig. 3)

Figure 3.

Figure 3

CD spectra of 1 at 5°C in 0.1 cm cells of (A) 0.384 mM Li+ form (red), 0.395 mM Na+ form (blue), and 0.437 mM Mg2+ form (green) in 10 mM LiOAc with 0.1 M LiCl, 10 mM sodium phosphate with 0.1 M NaCl, and 10 mM Mg(OAc)2 with 0.1 M MgCl2, respectively, all at pH 7.0, and (B) 0.409 mM K+ form (black), 0.383 mM Rb+ form (blue), 0.382 mM Cs+ form (green), and 0.378 mM NH4+ form (red) in 10 mM potassium phosphate with 0.1 M KCl, 10 mM RbOAc with 0.1 M RbCl, 10 mM CsOAc with 0.1 M CsCl, and 10 mM NH4OAc with 0.1 M NH4Cl, respectively, all at pH 7.0.

A CD spectrum of the Mg2+ form of 1 (Fig. 3A, green) closely resembles those of Li+ (red)24 and Na+ (blue).24 Upon heating to 75°C, the bands all quickly diminish to small residual values (not shown). The positive band at 257 nm and the negative band at 281 nm form a pattern that is inverted relative to right-handed DNA and RNA, but is similar to CD spectra of cyclic dithymidine monophosphate,28 5′-GMP and 3′-GMP in 0.1 M K+,29 left-handed 5′-5′ GpG,29 left-handed Z DNA,30 and diadenosine and diguanosine 5′,5‴-oligophosphates.31 For the Li+, Na+ and Mg2+ forms of 1, the pattern for the 257 and 281 nm bands is consistent with the possibility of a left-handed alignment of the guanine transition moments.

The CD spectrum that we previously reported for the K+ form of 1 displays additional bands at 215 nm (strong) and 309 nm (weak)24 (Fig. 3B, black). Further, its central positive band appears at 252 nm rather than 257 nm. We now report CD spectra for Rb+ (blue), Cs+ (green), and NH4+ (red) forms (Fig. 3B). The presence of a similar 215 nm band for the Rb+ form (although only half the intensity), but none for the Cs+ form, demonstrates the same dependence on ion size shown by guanine quartets.27,32 The Rb+ form, but not the Cs+ form, also displays a shift to lower wavelength for the central positive band, although only about half as much as for the K+ form. The NH4+ form shows a weak 215 nm band and a small shift of its central band. Positive CD bands near 300 nm have been observed for oligodeoxynucleotides with intramolecular quartet structures that have a mixture of syn and anti orientations, while those with intermolecular quartets that are all anti do not have this 300 nm band.27,33

We have previously reported the slow appearance and disappearance of the 215 and 309 nm bands for the K+ form of 1 (∼20 hr).24 We now report similar behavior for the Rb+ form, where nearly 8 hr are required for the full appearance of the 215 nm band upon cooling from 75 °C to 5 °C (Fig. 4A, blue). Further, a similar cooling curve for the 254 nm band (red) displays a fast increase over minutes to a maximum value, followed by a slow decrease to a somewhat smaller value. This slow decrease of the 254 nm band correlates with the slow increase of the 215 nm band, and reflects the shifting of the central positive band that presumably occurs upon quartet formation. Our previously reported data for the K+ form at 250 nm upon cooling24 also shows a fast increase followed by a similar slow decrease that is less pronounced than with Rb+. These results support the initial formation of a bimolecular stacked structure in K+ and Rb+, which then undergoes a slow association and rearrangement to one or more quartet structures.

Figure 4.

Figure 4

Plots of molar elliptici ty at a given wavelength as a function of time after (A) cooling or (B) heating pre-equilibrated samples of the Rb+ form of 1 in a 0.1 cm cell.

We had already shown the slow loss of the 215 and 250 nm bands upon heating the K+ form from 5 °C to 40 °C (∼20 hr), consistent with both bands being associated with the quartet structure.24 Heating the Rb+ form from 5 °C to 40 °C also shows a slow loss of the 215 and 254 nm bands (Fig. 4B), but over only about 10 minutes, consistent with the lower thermal stability of the Rb+ quartet structure.

NMR Spectra

We previously reported 1D 1H and 31P NMR spectra for the Li+, Na+, and K+ forms of ∼35 mM samples of 1 in 0.1 M LiCl, NaCl, or KCl, respectively, at pH 7.24 The data for the K+ form were consistent with two different types of quartet structures that were both stable even at 55 °C: one type (∼50%) displayed two guanine H8 resonances and two upfield 31P resonances, and another type (∼44%) displayed only one H8 resonance and only one downfield 31P resonance. Two addi tional minor (∼2 and 4%) H8 and 31P resonances were also present. 31P NMR spectra at 5 °C for the Li+ and Na+ forms of 1 also displayed upfield and downfield resonances, but in more complex patterns.24 These signals had significantly diminished at 55 °C, leaving a prominent central resonance near −1.0 ppm. 1H NMR spectra at 5 °C for the Li+ form showed six guanine H8 resonances. These results are consistent with the presence of a group of structures having lower stabilities than those of the K+ form. In spite of their different stabilities, the Li+, Na+, and K+ forms in 90% H2O all displayed 1H resonances near 11 ppm consistent with hydrogen bonded guanine H1s, as well as broad resonances near 9 and 6.5 ppm consistent with hydrogen bonded and non-hydrogen bonded guanine amino protons, respectively.34 As described more fully in the Discussion section, these previous1D NMR results and new 2D NMR data introduced below for both the K+ and Li+ forms of 1 support the presence in varying amounts of five different but related structures shown in the cartoons at the bottom of Figure 5: one bimolecular self-intercalated structure (shown in more detail at the top left of Figure 5), two tetramolecular and two octamolecular structures, all of which contain guanine quartets (shown at the top right of Figure 5, with the H8 to H-bonded amino NOE highlighted in red). The 2D NMR data described below demonstrate that the larger complexes exist in both all-syn and all-anti conformations.

Figure 5.

Figure 5

Cartoons of proposed structures of 1

The 1D spectrum of a 37 mM sample of the K+ form of 1 in KCl at pH 7 at 25 °C is shown at the top of Figure 6, with a portion of its superimposed 1H-13C HMBC/HMQC spectrum at the center, and a portion of its 1H-1H NOESY spectrum at the bottom. Region A in the NOESY spectrum shows a set of H8 to H1′ crosspeaks. The three major H8 resonances show crosspeaks to H1′ resonances, of which one is significantly stronger than the other two. The different NOE intensities are consistent with a syn guanosine conformation for the stronger NOE and anti conformations for the weaker NOEs.35 The H8 resonances at 7.75, 8.20, and 7.95 ppm, and the H1′ resonances at 6.05, 6.19, and 5.77 ppm, are therefore labeled Os, Oa, and Oa′ for octamolecular/syn and anti, respectively. There are no crosspeaks between the syn and anti resonances, while crosspeaks are observed between the two anti resonances. Of the two minor H8 resonances, one shows a stronger crosspeak to the H1′ region than the other, again indicating syn and anti conformations. The H8 resonances at 7.85 and 8.26 ppm, and the H1′ resonances at 6.12 and 5.96 ppm are therefore labeled Ts and Ta, for tetramolecular/syn and anti, respectively. These NOESY crosspeaks had the same positive sign as the auto-peaks due to negative NOE enhancements, and built up quite fast at a mixing time of 150 ms, which indicates that the molecular species giving rise to them are in the slow tumbling regime.36 Since neither the monomeric (MW=689) nor the dimeric (MW=1378) forms of 1 in water are large enough to generate strong negative NOE enhancements, our data suggest that the four species present in this sample with K+ are larger than the bimolecular form.

Figure 6.

Figure 6

NMR spectra of a 37 mM sample of the K+ form of 1 in 0.1 M KCl in 90/10 H2O/D2O at 25 °C, with its 1D 1H spectrum at the top, a portion of its superimposed 1H-13C HMBC/HMQC spectrum at the center, and a portion of its 1H-1H NOESY spectrum at the bottom. Regions of interest are labeled to correspond with the text.

Region B in the HMBC/HMQC confirms the assignments, since the H1′ resonances labeled Os and Ts show stronger crosspeaks to their C4 than to their C8 resonances, while the H1′ resonances labeled Oa and Oa′ do not show crosspeaks to their C4 but do to their C8 resonances. Such differences have been shown to support a syn assignment for the former and an anti assignment for the latter.37

The HMBC/HMQC spectrum also allows assignment of the amido H1 resonances, as shown in Region C. The large downfield H1 resonance at 11.27 ppm (Os) correlates only with the syn H8 resonance Os, while the pair of large upfield H1 resonances at 11.18 ppm (Oa and Oa′) both correlate only with the two anti H8 resonances, Oa and Oa′. From Region D in the NOESY spectrum, the broad hydrogen bonded amino resonances can then be assigned in terms of syn and anti. The syn H1 resonance Os correlates only with the more upfield hydrogen bonded amino resonance at 8.81 ppm (labeled s) and the anti H1 resonances Oa and Oa′ correlate with the more downfield hydrogen bonded amino resonance at 9.45 ppm (labeled a). The minor downfield H1 resonance at 11.42 ppm, labeled Ts,a since it presumably represents these two overlapping resonances, shows a crosspeak to the region of the syn amino resonance at 8.81 ppm.

Since in a given quartet, each hydrogen bonded amino is close to the H8 of its neighbor (Figure 5), the pattern of their NOE crosspeaks can define the arrangement of syn and anti conformations within each quartet.38,39 Region E in the NOESY spectrum shows that the broad anti hydrogen bonded amino region (a) has crosspeaks only to the two anti H8 resonances Oa and Oa′, and the broad syn hydrogen bonded amino region (s) has a crosspeak only to the single syn H8 resonance Os. Thus we conclude that with K+, c-di-GMP exists primarily as two major different quartet-containing species, with the Oa/Oa′ structure being all-anti and the Os quartet structure being all-syn. The different intensities of the Os and Oa resonances do not support the presence of a mixed species with one anti quartet and one syn quartet. All-anti quartets are well known in parallel guanine-containing oligodeoxynucleotides,27,32 and all-syn quartets have been found in lipophilic guanine quartet assemblies.38,40

A 1D NMR spectrum of a 33 mM sample of the Li+ form of 1 in 0.1 M LiCl at pH 7 at 25 °C is shown at the top of Figure 7, with a portion of its superimposed 1H-13C HMBC/HMQC spectrum at the center and a portion of its 1H-1H NOESY spectrum at the bottom. Although the relative intensities of the resonances are quite different from those of the K+ form, the 2D spectra are strikingly similar. Region A in the NOESY spectrum shows significant crosspeaks from the two most upfield H8 resonances (Ts at 7.83 and Os at 7.74 ppm) to H1′ resonances, but weak crosspeaks from the three more downfield H8 resonances (Ta at 8.25, Oa at 8.19, and Oa′ at 7.95 ppm) to H1′ resonances. These NOE crosspeaks were also positive with strongly negative NOE enhancements, similar to those with K+. However, no crosspeak is seen for the sixth resonance at 7.90 ppm, labeled Ba for bimolecular/anti. These results indicate the presence of the same syn and anti structures as with the K+ form, although in more similar amounts, and with one additional structure. The HMBC/HMQC spectrum again confirms these assignments, as shown in Region B. Further, for the sixth H8 resonance at 7.90 ppm (Ba), a crosspeak is seen between its corresponding H1′ resonance at 5.84 ppm to its C8, but not to its C4 resonance, indicating this additional structure has an anti conformation.

Figure 7.

Figure 7

NMR spectra of a 33 mM sample of the Li+ form of 1 in 0.1 M LiCl in 90/10 H2O/D2O at pH 7 at 25 °C, with its 1H 1D spectrum at the top, a portion of its superimposed 1H-13C HMBC/HMQC spectrum at the center, and a portion of its 1H-1H NOESY spectrum at the bottom. Regions of interest are labeled to correspond with the text.

Region C of the HMBC/HMQC spectrum again allows assignment of the amido H1 resonances. The most downfield H1 resonance at 11.38 ppm (Ta) correlates only with the H8 resonance Ta at 8.25 ppm; the center two H1 resonances at 11.36 ppm and 11.23 ppm (Ts and Os) correlate with H8 resonances Ts at 7.83 ppm and Os at 7.74 ppm; and the upfield H1 resonance at 11.14 ppm (Oa,a′) correlates with H8 resonances Oa at 8.19 ppm and Oa′ at 7.95 ppm. From Region D in the NOESY spectrum, the hydrogen bonded amino resonances again can be assigned in terms of syn and anti. The anti H1 resonance Ta correlates with the more downfield hydrogen bonded amino region at 9.42 ppm (a); the syn H1 resonance Ts correlates with the more upfield hydrogen-bonded amino region at 8.85 ppm (s); the syn resonance Os correlates with the s amino region; and the anti H1 resonance Oa correlates with the a amino region. Region E in the NOESY spectrum shows two crosspeaks for the broad syn hydrogen bonded amino region (s), one to the H8 resonance Os and one to the H8 resonance Ts. The anti hydrogen-bonded amino (a) is apparently too weak to show analogous crosspeaks. These data are again consistent with an Oa/Oa′ quartet structure that is all-anti and an Os quartet structure that is all-syn.

The NOESY data described above suggest that the two major and the two minor species with K+ all have molecular weights larger than the bimolecular form. In order to obtain additional information, NMR diffusion experiments were performed. The data (Supporting Information) show that with K+, the translational diffusion rates of the two major forms, Oa and Os, are similar to each other, but smaller than those of the two minor forms, Ta and Ts, which are also similar to each other. These results demonstrate that Oa and Os have a higher molecular weight than Ta and Ts. The measured diffusion rates of these four species agree well with the published range of DNA diffusion rates.41 Diffusion data for the Li+ form show a similar pattern in that the rates for Ta and Ts are larger than those for Oa and Os. The rate for the fifth form, Ba, was considerably larger than any of the other species, demonstrating a smaller molecular weight that is consistent with a bimolecular complex. The absolute values of the rates for the analogous species in the two samples are slightly different, perhaps because the samples are not identical.

These new UV, CD, and NMR results together present a compelling case for the presence of a mixture of five different structures that 1 can adopt, whose ratio is both metal and concentration-dependent. As discussed more fully below, a bimolecular structure (Ba) is dominant with Li+, Na+, Cs+, and Mg2+ at dilute concentrations of 1, but is accompanied by one or more quartet structures with K+ and Rb+. At the higher NMR concentrations, with K+ the bimolecular structures associate and rearrange primarily to two stable higher molecular weight quartet structures, one all-syn (Os) and one all-anti (Oa), with two minor lower molecular weight quartet structures also present, Ts and Ta. With Li+, the bimolecular structure (Ba) predominates, but both quartet structures (Oa, Os, Ta, and Ts) are also present.

2. c-GMP-IMP, 2

We have now synthesized a mixed cyclic dimer containing one guanosine and one inosine, 2 (Figure 1), by a method similar to what we previously reported for 1.24 In brief, an intermolecular coupling using phosphoramidite chemistry was followed by an H-phosphonate cyclization, with the H-phosphonate group also serving as a 3′ protecting group during the initial coupling. The detailed procedures are described in the Supporting Information. We describe below UV, CD, and NMR spectra for this molecule that has one fewer amino groups than 1.

UV Melts (Fig 8A)

Figure 8.

Figure 8

(A) Normalized UV melting profiles at 250 nm of four different concentrations of Li+ (blue) and K+ (red) forms of 2 in 10 mM LiOAc with 0.1 M LiCl, and 10 mM potassium phosphate with 0.1 M KCl, respectively, both pH 7.0, and (B) CD spectra at 5 °C in 0.1 cm cells of 0.537 mM Li+ form (blue) and 0.544 mM K+ form (red) of 2 in 10 mM LiOAc with 0.1 M LiCl, and 10 mM potassium phosphate with 0.1 M KCl, respectively, both at pH 7.0.

UV melting profiles at 250 nm for different concentrations of 2 with both Li+ and K+ show single concentration-dependent conformational transitions that are identical for the two cations and reflect a more stable structure than the Li+ form of 1. The Tmax values of the most concentrated samples of both forms of 2 were found to be 18 °C, while the Tmax for the Li+ form of 1 was only 10 °C. These results suggest that loss of one of the guanine amino groups prevents formation of quartet structures in K+, but allows a bimolecular stacked structure.

CD Spectra (Fig. 8B)

CD spectra for the Li+ and K+ forms of 2 are nearly identical, and display much lower intensities than the Li+ form of 1, perhaps due to the different absorption properties of inosine. The lack of any bands near 215 and 305 nm supports the absence of quartet structures.

NMR Spectra

The 1D NMR spectra for 61 mM samples of the Li+ and K+ forms of 2 at 5, 30, and 55 °C (Supporting Information) are much simpler than those for 1. Both 31P NMR spectra of 2 at 5 °C show only two resonances near −1.5 ppm, one for each of the different phosphates in the molecule. Neither of the 1H NMR spectra in 90% H2O at 5 °C shows H1 resonances near 11 ppm or amino resonances near 9 and 6.5 ppm, demonstrating the exchange of these protons with solvent. Further, single sharp resonances are present for the guanine H8 and the inosine H8 and H2. As seen in Figure 9, the 1H-1H NOESY spectrum for 2 with K+ does not show crosspeaks between the H8 and H1′ resonances, but the 1H-13C HMBC/HMQC spectrum does show crosspeaks between the two H1′ resonances and their C8 resonances, but not their C4 resonances, indicating anti conformations. Thus, absence of one amino group precludes quartet formation but allows a bimolecular anti stacked structure in which the amido and amino protons exchange readily with solvent.

Figure 9.

Figure 9

NMR spectra of a 60.8 mM sample of the K+ form of 2 in 90/10 H2O/D2O at 25 °C, with its 1D spectrum at the top, a portion of its superimposed 1H-13C HMBC/HMQC spectrum at the center, and a portion of its 1H-1H NOESY NMR spectrum at the bottom.

3. Oligoethyleneoxide tethered c-di-GMP, 3a and 3b

We have also linked two molecules of 1 together with flexible oligoethyleneoxide tethers of four (3a) or twelve (3b) units (Fig. 1), thereby constraining their interactions. Their syntheses and 1D NMR spectra are described in the Supporting Information, and their UV and CD data are presented below.

UV Melts (Fig. 10A)

Figure 10.

Figure 10

(A) Normalized UV melting profiles at 250 nm of two different concentrations each of the Li+ (dark and light green) and K+ (red and blue) forms of 3a and 3b in 10 mM LiOAc with 0.1 M LiCl, and 10 mM potassium phosphate with 0.1 M KCl, respectively, both pH 7.0, and (B) CD spectra at 5°C in 0.1 cm cells of 0.227 mM Li+ form (dark green) and 0.227 mM K+ form (red) of 3a in 10 mM LiOAc with 0.1 M LiCl, and 10 mM potassium phosphate with 0.1 M KCl, respectively, and 0.209 mM Li+ form (blue) and 0.212 mM K+ form (light green) of 3b in 10 mM LiOAc with 0.1 M LiCl, and 10 mM potassium phosphate with 0.1 M KCl, respectively, all at pH 7.0.

UV melting profiles at 250 nm for two different concentrations each of 3a and 3b with Li+ and K+ show hypochromic conformational transitions that have very little concentration dependence. Further, for both tether lengths, the K+ forms are only slightly more stable than the Li+ forms. The transitions are thus primarily intramolecular and cation independent. The structured forms of 3b have Tmax values of 24 °C with Li+ and 26 °C with K+, significantly lower than those of 3a, which have Tmax values of 42 °C with Li+ and 44 °C with K+. The similar hypochromicity in both metals indicates that the tethers allow intramolecular formation of related stacked structures, while the lower stabilities with the longer tether may be due to entropic factors.

CD Spectra (Fig 10B)

CD spectra for the Li+ and K+ forms of 3a and 3b are nearly identical to each other, and similar to the CD spectrum of 1 with Li+. The lack of any bands near 215 or 305 nm shows that the tether precludes formation of guanine quartets at this concentration.

NMR Spectra (Supporting Information)

As with 2, the 1D 1H and 31P NMR spectra of a 15 mM sample of the Li+ form of 3a are much simpler than those of 1. The 31P NMR spectrum shows only three resonances near −2.0 ppm, one for each type of phosphate in the molecule. The 31P NMR spectrum of a 13 mM sample of the K+ form also gives three dominant resonances near −2.0 ppm, but smaller clusters of signals are present upfield and downfield as well. The 1H NMR spectrum of the Li+ form shows no H1 signals near 11 ppm or amino signals near 9 and 6.5 ppm, signifying their exchange with solvent. Single sharp resonances are present for each of the two types of guanine H8, one of which is closer to the tether and the other farther away. The K+ form also shows two dominant sharp H8 resonances, but a cluster of several smaller H8 resonances is also present. Further, there are several small H1 resonances near 11 ppm and amino resonances near 9 and 6.5 ppm. The UV, CD, and NMR results for 3a with Li+ are consistent with the presence of only a unimolecular self-intercalated structure related to the bimolecular self-intercalated structure of 1, Ba. With K+, the results support the presence of a mixture with the major component being the self-intercalated structure, but with small amounts of quartet complexes also present. Since the concentration of 3a with K+ (26 mM in terms of guanine) is similar to that of 1 with K+ (37 mM in guanine), we conclude that the tether makes formation of a quartet structure significantly more difficult, but allows the self-intercalated structure to readily form.

Discussion

Based on the UV, CD, and NMR data presented above and in our earlier report,24 we propose that c-di-GMP can adopt five different but related structures, all participating in an equilibrium that is sensitive to both concentration of 1 and the metal present. The cartoons at the bottom of Figure 5 depict these five structures: an anti bimolecular self-intercalated complex, syn and anti versions of a tetramolecular quartet complex, and syn and anti versions of an octamolecular self-intercalated quartet complex. The UV concentration dependence and hypochromicity, as well as the strong CD spectra, demonstrate the presence of intermolecular guanine stacking. At dilute concentrations with Li+, Na+, Cs+, or Mg2+, 1 forms the self-intercalated bimolecular complex that had been found in crystal structures reported for 1 in the presence of Mg2+ or Co2+,22,23 as well as for 1 associated with its cyclase from Caulobacter crescentus.7 At the higher NMR concentrations with Li+, single H8 and 31P resonances are seen from this structure. Although this complex has two different types of phosphates and guanines (inner and outer), and the amido H1s and aminos of the inner guanines may be oriented so as to form hydrogen bonds with a phosphate of its partner, our data are consistent with this self-intercalated structure being in fast exchange on the NMR timescale with an unstructured form. Thus no amido or amino resonances are present and only single H8, H1′, and 31P resonances are observed. The lack of positive NOE crosspeaks and the relatively large translational diffusion rate support the lower molecular weight of a bimolecular complex.

Our data indicate that the tethered c-di-GMP (3a and 3b) as well as c-GMP-IMP (2) are also able to adopt this self-intercalated structure, in spite of the constraints of the tethers and lack of one amino, respectively. Their UV spectra show significant hypochromicity, and the CD spectra of 3a and 3b are nearly identical to those of 1. The smaller intensity of the CD spectrum of 2 is most likely because of the altered absorption properties of inosine. Both the 1D and 2D NMR data are consistent with the predominant presence of this self-intercalated structure in the Li+ and K+ forms of 2 and 3a.

At low concentrations of 1 with K+, Rb+, and NH4+, our UV and CD data strongly support the presence of one or more additional structures containing guanine quartets. The UV melt of the K+ form of 1 at 250 nm displays two transitions, and its melt at 305 nm shows characteristics of oligodeoxynucleotides containing quartets with syn conformations. The CD spectra with these metals display new bands at 215 and 309 nm that are slow to appear and diminish, consistent with a higher molecularity. At the higher NMR concentrations of the K+ form, 1H and 31P NMR resonances distinct from those corresponding to the bimolecular structure persist at 55 °C. The 1H NMR spectra, both 1D and 2D, are consistent with two major structures, both of which have similar and relatively high molecular weights. One has two sets of H8, H1′, and 31P resonances, to which we can assign anti conformations, and the other has only one set to which we can assign the syn conformation. The NMR spectra also show two minor structures with similar and relatively smaller molecular weights that are still greater than the bimolecular form, one of which is anti and one of which is syn.

In Wang's reports on the self-intercalated crystal structure of 1 with Mg2+ and Co2+,23 he had speculated about the possibility of its rearrangement to a cage-like complex of four molecules of 1 containing two parallel guanine quartets 7Å apart (bottom of Figure 5, center). Although he described only an all-anti version, it might also occur in an all-syn version, and our data support the presence of these two tetramolecular quartet structures as minor components in the K+ form of 1 at high concentrations. However, if two such all-anti or all-syn quartet complexes (a total of eight molecules of 1) were able to self-intercalate, they should be stabilized by favorable stacking (3.5 Å apart) and additional interactions with the K+ ions (bottom of Figure 5, right). The presence of inner and outer guanosines would account for the two sets of NMR resonances in the all-anti form, in which models show that one set of H8 atoms appears to be significantly closer to nearby phosphates than the other. In the all-syn form, models show the H8 atoms all appear to be in more similar environments, which may explain the failure to observe more than one set of resonances at the field strength we used. Presumably, these large octamolecular complexes would most easily form by gradual association and readjustment of four of the self-intercalated bimolecular structures.

A slow equilibrium would exist among these five different complexes at high concentrations of 1, with K+ highly favoring the two octamolecular complexes. Since the smaller, less easily dehydrated Li+ is less able to stabilize quartets, the resulting mixture has approximately similar amounts of the bimolecular, the two tetramolecular, and the two octamolecular complexes. Although Li+ does not normally stabilize quartet structures, in this case the presence of two aligned guanines in one molecule presents unique circumstances. c-GMP-IMP (2) is unable to form any of these quartet structures because the lack of one amino prevents the full quartet hydrogen bonding. The tethered c-di-GMP (3a, and presumably 3b) are restricted in their abilities to form quartet structures with K+. An equilibrium among these five structures for c-di-GMP in a bacterial cell might provide a range of options for the actions of this signaling molecule that would significantly depend on its own concentration as well as local K+ concentrations, or the concentrations of other as yet unidentified effectors. The subcellular compartmentalization and gradients that are known to affect the signaling ability of c-di-GMP12 may be taking advantage of the equilibrium we describe here. As first proposed by Wang,23 the tetramolecular complexes may participate in biological pathways by intercalating aromatic groups between the two quartets. We further speculate that the octamolecular complexes may serve as a way to sequester c-di-GMP in an inactive state, with the bimolecular complex providing the active form for signaling. Additional work in our laboratory is underway to further characterize the structural details of these five different complexes, as well as explore in more depth the equilibrium among them.

Conclusion

We have demonstrated that c-di-GMP, an important bacterial signaling molecule involved in biofilm formation and other pathogenic processes, can adopt a mixture of five different but related structures in ratios that are sensitive to both its concentration and the metal present. A bimolecular self-intercalated structure is dominant at dilute concentrations with Li+, Na+, Cs+, and Mg2+, while it is accompanied by one or more quartet complexes with K+, Rb+, and NH4+. At higher concentrations with K+ the bimolecular structures associate and rearrange primarily to two stable octamolecular self-intercalated quartet complexes, one all-syn and one all-anti. Two minor tetramolecular quartet complexes are also present, one all-syn and one all-anti. At the higher concentrations with Li+ the tetramolecular and octamolecular quartet complexes (each with syn and anti versions) are present in approximately equal amounts, along with the bimolecular structure.

Supplementary Material

1si20060406_10. Supporting Information Available.

Synthesis of 2, 3a, and 3b; first derivative curves of UV melting profiles; full UV spectra of 1 in Li+, Na+, K+, Rb+, NH4+, and Mg2+; translational diffusion rates for 1 with K+ and Li+; 1D NMR of 1, 2, 3a, and 3b. This material is available free of charge via the Internet at http://pubs.acs.orb.

Acknowledgments

This work was supported by a grant from NIH (EB002809). We thank Kenneth J. Breslauer, James Elliott, and Jens Völker in this Department for assistance with the CD and helpful discussions.

References

  • 1.Jenal U. Curr Opin Microbiol. 2004;7:185–191. doi: 10.1016/j.mib.2004.02.007. [DOI] [PubMed] [Google Scholar]; Romling U, Gomelsky M, Galperin MY. Mol Microbiol. 2005;57:629. doi: 10.1111/j.1365-2958.2005.04697.x. [DOI] [PubMed] [Google Scholar]
  • 2.Hall-Stoodley L, Costerton JW, Stoodley P. Nature Rev Microbiol. 2004;2:95–108. doi: 10.1038/nrmicro821. [DOI] [PubMed] [Google Scholar]
  • 3.Stewart PS, Costerton JW. Lancet. 2001;358:135–138. doi: 10.1016/s0140-6736(01)05321-1. [DOI] [PubMed] [Google Scholar]; Mah TFC, O'Toole GA. Trends Microbiol. 2001;9:34–39. doi: 10.1016/s0966-842x(00)01913-2. [DOI] [PubMed] [Google Scholar]
  • 4.Hoffman LR, D'Argenio DA, MacCoss MJ, Zhang Z, Jones RA, Miller SI. Nature (London) 2005;436:1171–1175. doi: 10.1038/nature03912. [DOI] [PubMed] [Google Scholar]
  • 5.Paul R, Weiser S, Amiot NC, Chan C, Schirmer T, Giese B, Jenal U. Genes Dev. 2004;18:715–727. doi: 10.1101/gad.289504. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Ryjenkov DA, Tarutina M, Moskvin OV, Gomelsky M. J Bacteriol. 2005;187:1792–1798. doi: 10.1128/JB.187.5.1792-1798.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Chan C, Paul R, Samoray D, Amiot NC, Giese B, Jenal U, Schirmer T. Proc Natl Acad Sci USA. 2004;101:17084–17089. doi: 10.1073/pnas.0406134101. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Simm R, Morr M, Kader A, Nimtz M, Romling U. Mol Microbiol. 2004;53:1123–1134. doi: 10.1111/j.1365-2958.2004.04206.x. [DOI] [PubMed] [Google Scholar]; Tischler AD, Camilli A. Mol Microbiol. 2004;53:857–869. doi: 10.1111/j.1365-2958.2004.04155.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Bobrov AG, Kirillina O, Perry RD. FEMS Microbiol Lett. 2005;247:123–130. doi: 10.1016/j.femsle.2005.04.036. [DOI] [PubMed] [Google Scholar]
  • 10.Galperin MY, Nikolskaya AN, Koonin EV. FEMS Microbiol Lett. 2001;203:11–21. doi: 10.1111/j.1574-6968.2001.tb10814.x. [DOI] [PubMed] [Google Scholar]
  • 11.Christen M, Christen B, Folcher M, Schauerte A, Jenal U. J Biol Chem. 2005;280:30829–30837. doi: 10.1074/jbc.M504429200. [DOI] [PubMed] [Google Scholar]
  • 12.Romling U. Cell Mol Life Sci. 2005;62:1234–1246. doi: 10.1007/s00018-005-4557-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Ross P, Weinhouse H, Aloni Y, Michaeli D, Weinberger-Ohana P, Mayer R, Braun S, de Vroom E, van der Marel GA, van Boom JH, Benziman M. Nature. 1987;325:279–281. doi: 10.1038/325279a0. [DOI] [PubMed] [Google Scholar]
  • 14.Hisert KB, MacCoss M, Shiloh MU, Darwin KH, Singh S, Jones RA, Ehrt S, Zhang Z, Gaffney BL, Gandotra S, Holden D, Murray D, Nathan C. Mol Microbiol. 2005;56:1234–1245. doi: 10.1111/j.1365-2958.2005.04632.x. [DOI] [PubMed] [Google Scholar]
  • 15.Tischler AD, Camilli A. Infect Immun. 2005;73:5873–5882. doi: 10.1128/IAI.73.9.5873-5882.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Waters CM, Bassler BL. Annu Rev Cell Dev Biol. 2005;21:319–346. doi: 10.1146/annurev.cellbio.21.012704.131001. [DOI] [PubMed] [Google Scholar]
  • 17.Fuqua C, Parsek MR, Greenberg EP. Annu Rev Gen. 2001;35:439–468. doi: 10.1146/annurev.genet.35.102401.090913. [DOI] [PubMed] [Google Scholar]
  • 18.Kleerebezem M, Quadri LE, Kuipers OP, de Vos WM. Mol Microbiol. 1997;24:895–904. doi: 10.1046/j.1365-2958.1997.4251782.x. [DOI] [PubMed] [Google Scholar]
  • 19.Pesci EC, Milbank JBJ, Pearson JP, McKnight S, Kende AS, Greenberg EP, Iglewski BH. Proc Natl Acad Sci USA. 1999;96:11229–11234. doi: 10.1073/pnas.96.20.11229. [DOI] [PMC free article] [PubMed] [Google Scholar]; Deziel E, Lepine F, Milot S, He J, Mindrinos MN, Tompkins RG, Rahme LG. Proc Natl Acad Sci USA. 2004;101:1339–1344. doi: 10.1073/pnas.0307694100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Chen X, Schauder S, Potier N, Van Dorsselaer A, Pelczer I, Bassler BL, Hughson FM. Nature (London) 2002;415:545–549. doi: 10.1038/415545a. [DOI] [PubMed] [Google Scholar]; McKenzie KM, Meijler MM, Lowery CA, Boldt GE, Janda KD. Chem Commun. 2005:4863–4865. doi: 10.1039/b509396a. [DOI] [PubMed] [Google Scholar]
  • 21.Camilli A, Bassler BL. Science. 2006;311:1113–1116. doi: 10.1126/science.1121357. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Egli M, Gessner RV, Williams LD, Quigley GJ, van der Marel GA, van Boom JA, Rich A, Frederick CA. Proc Natl Acad Sci USA. 1990;87:3235–3239. doi: 10.1073/pnas.87.8.3235. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Liaw YC, Gao YG, Robinson H, Sheldrick GM, Sliedregt LAJM, van der Marel GA, vB AJ, Wang AHJ. FEBS. 1990;264:223–227. doi: 10.1016/0014-5793(90)80253-f. [DOI] [PubMed] [Google Scholar]; Guan Y, Gao YG, Liaw YC, Robinson H, Wang AHJ. J Biomol Struct Dyn. 1993;11:253–276. doi: 10.1080/07391102.1993.10508725. [DOI] [PubMed] [Google Scholar]
  • 24.Zhang Z, Gaffney BL, Jones RA. J Am Chem Soc. 2004;126:16700–16701. doi: 10.1021/ja0449832. [DOI] [PubMed] [Google Scholar]
  • 25.Marky LA, Breslauer KJ. Biopolymers. 1987;26:1601–1620. doi: 10.1002/bip.360260911. [DOI] [PubMed] [Google Scholar]
  • 26.Mergny JL, Phan AT, Lacroix L. FEBS Lett. 1998;435:74–78. doi: 10.1016/s0014-5793(98)01043-6. [DOI] [PubMed] [Google Scholar]
  • 27.Keniry MA. Biopolymers (NAS) 2001;56:123–146. doi: 10.1002/1097-0282(2000/2001)56:3<123::AID-BIP10010>3.0.CO;2-3. [DOI] [PubMed] [Google Scholar]
  • 28.Cantor CR, Fairclough RH, Newmark RA. Biochemistry. 1969;8:3610–3617. doi: 10.1021/bi00837a018. [DOI] [PubMed] [Google Scholar]
  • 29.Chantot JF, Haertle T, Guschlbauer W. Biochemie. 1974;56:501–507. doi: 10.1016/s0300-9084(74)80065-9. [DOI] [PubMed] [Google Scholar]
  • 30.Pohl FM, Jovin TM. J Mol Biol. 1972;67:375–396. doi: 10.1016/0022-2836(72)90457-3. [DOI] [PubMed] [Google Scholar]
  • 31.Scott JF, Zamecnik PC. Proc Natl Acad Sci USA. 1969;64:1308–1314. doi: 10.1073/pnas.64.4.1308. [DOI] [PMC free article] [PubMed] [Google Scholar]; Holler E, Holmquist B, Vallee BL, Taneja K, Zamecnik P. Biochemistry. 1983;22:4924–4933. doi: 10.1021/bi00290a008. [DOI] [PubMed] [Google Scholar]
  • 32.Hardin CC, Perry AG, White K. Biopolymers (NAS) 2001;56:147–194. doi: 10.1002/1097-0282(2000/2001)56:3<147::AID-BIP10011>3.0.CO;2-N. [DOI] [PubMed] [Google Scholar]; Davis JT. Angew Chem Int Ed Engl. 2004;43:668–698. doi: 10.1002/anie.200300589. [DOI] [PubMed] [Google Scholar]
  • 33.Jin R, Gaffney BL, Wang C, Jones RA, Breslauer KJ. Proc Natl Acad Sci USA. 1992;89:8832–8836. doi: 10.1073/pnas.89.18.8832. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Wang Y, Patel DJ. Biochemistry. 1992;31:8112–8119. doi: 10.1021/bi00150a002. [DOI] [PubMed] [Google Scholar]; Smith FW, Feigon J. Biochemistry. 1993;32:8682–8692. doi: 10.1021/bi00084a040. [DOI] [PubMed] [Google Scholar]
  • 35.Wang Y, de los Santos C, Gao X, Greene K, Live D, Patel DJ. J Mol Biol. 1991;222:819–832. doi: 10.1016/0022-2836(91)90513-6. [DOI] [PubMed] [Google Scholar]; Smith FW, Feigon J. Nature (London) 1992;356:164–1168. doi: 10.1038/356164a0. [DOI] [PubMed] [Google Scholar]; Wang KY, McCurdy S, Shea RG, Swaminathan S, Bolton PH. Biochemistry. 1993;32:1899–1904. doi: 10.1021/bi00059a003. [DOI] [PubMed] [Google Scholar]
  • 36.Cavanagh J, Fairbrother WJ, Palmer AG, III, Skelta NJ. Academic Press, Inc.; San Diego, CA: 1996. pp. 289–290. [Google Scholar]
  • 37.Zhu G, Live D, Bax A. J Am Chem Soc. 1994;116:8370–8371. [Google Scholar]
  • 38.Marlow AL, Mezzina E, Spada GP, Masiero S, Davis JT, Gottarelli G. J Org Chem. 1999;64:5116–5123. doi: 10.1021/jo9901440. [DOI] [PubMed] [Google Scholar]
  • 39.Phan AT, Patel DJ. J Am Chem Soc. 2003;125:15021–15027. doi: 10.1021/ja037616j. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Forman SL, Fettinger JC, Pieraccini S, Gottarelli G, Davis JT. J Am Chem Soc. 2000;122:4060–4067. [Google Scholar]
  • 41.Lapham J, Rife JP, Moore PB, Crothers DM. J Biomol NMR. 1997;10:255–262. doi: 10.1023/a:1018310702909. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1si20060406_10. Supporting Information Available.

Synthesis of 2, 3a, and 3b; first derivative curves of UV melting profiles; full UV spectra of 1 in Li+, Na+, K+, Rb+, NH4+, and Mg2+; translational diffusion rates for 1 with K+ and Li+; 1D NMR of 1, 2, 3a, and 3b. This material is available free of charge via the Internet at http://pubs.acs.orb.

RESOURCES