Opening the Egg Box: NMR spectroscopic analysis of the interactions between s-block cations and kelp monosaccharides

Abstract

The best-known theory accounting for metal-alginate complexation is the so-called “Egg Box” model. In order to gain greater insight into the metal-saccharide interactions that underpin this model, the coordination chemistry of the corresponding monomeric units of alginate, L-guluronate (GulA) and D-mannuronate (ManA) have been studied herein. GulA and ManA were exposed to solutions of different s-block cations and then analysed by ¹H and ¹³C NMR spectroscopy. It was found that the α/β ratio of the pyranose anomeric equilibria of GulA showed large pertubations from the starting value (α/β = 0.21 ± 0.01) upon contact with 1.0 M Ca²⁺, Sr²⁺, and Ba²⁺ (α/β = 1.50 ± 0.03, 1.20 ± 0.02, and 0.58 ± 0.02, respectively) at pD 7.9, but remained almost constant in the presence of Na⁺, K⁺, and Mg²⁺ (α/β = 0.24 ± 0.01, 0.19 ± 0.01, and 0.26 ± 0.01, respectively). By comparison, no significant changes were observed in the α/β ratios of ManA and related mono-uronates D-glucuronate (GlcA) and D-galacturonate (GalA) in the presence of all of the metal ions surveyed. Analysis of the ¹H and ¹³C coordination chemical shift patterns indicate that the affinity of α-GulA for larger divalent cations is a consequence of the unique ax–eq–ax arrangement of hydroxyl groups found for this uronate anomer.

---

Introduction

Kelps are large brown seaweeds that are found commonly across the shores of Northern Europe and contain the biopolymer alginate at high levels (up to 40% of the dry weight).¹ Alginates are a class of anionic polysaccharide comprised of linear chains of (1 → 4)-linked (¹C₄)-α-L-gulopyranuronate and (⁴C₁)-β-D-mannopyranuronate units,^2,3 and are well-known for their metal-binding properties.^4,5 Notably, in the presence of divalent metal cations alginates form gels,⁶ which are used extensively in food preparation,⁷ cosmetics,⁸ and biomedical applications.^9–12 Of particular importance are the high affinities of alginates for specific metal ions (such as Pb²⁺, Cu²⁺, and Cd²⁺),^13,14 which enables use of kelp biomass for the bioremediation of contaminated water.⁴ This technological importance of metal-alginate coordination interactions has driven considerable research interest.^15–17

The most well-known theory for the nature of metal-alginate binding is the so-called Egg Box model, originally formulated by Morris, Rees, and co-workers,¹⁸ and independently by Angyal¹⁹ and Smidsrød et al.²⁰ This model remains the subject of much research interest even after almost 50 years of its original proposal. The premise of the model is that the buckled structure of poly-α-L-guluronate chains give rise to anionic cavities, in which divalent metal ions can nestle like eggs in an egg box. Key to the Egg Box binding motif is the axial–equatorial–axial (ax–eq–ax) arrangement of oxygen atoms on the α-L-gulopyranuronate units, which enables a di-pentadentate coordination of a metal ion (Fig. 1). The model successfully explains why some metals (such as Ca²⁺) readily form gels, whilst others (such as Mg²⁺) do not, as a result of the size match/mis-match of the pseudo 3D metal ion cavity.^21,22 This Egg Box model also offers an explanation for why the flatter, ribbon-like poly-β-D-mannuronate chains do not show the same gelation behaviour as their guluronate counterparts.^14,18,23,24

Fig. 1 Ca²⁺ coordinated within an Egg Box site formed by two buckled, parallel poly-α-L-guluronate chains of an alginate. The ax–eq–ax arrangements of hydroxyl groups are highlighted (bold).

Nevertheless, since its initial suggestion, many authors have questioned the validity of the Egg Box binding model, particularly the high-coordination numbers of the central metal ion that have been proposed.²⁵ Consequently, a number of alternative models have been put forward supported by computational studies.^26–33 However, experimental validation of these models remains a challenge, as techniques such as NMR spectroscopy and X-ray diffraction are not easily applied and interpreted for disordered metal-alginate gels.^34–37 To circumvent these problems Plazinski and Drach have sought recently to refine computational investigations of metal alginate systems by first modelling the interactions of the mono- and oligo-saccharide units involved.^28,38 Again, however, supporting experimental data for such endeavours is scarce owing to the, often prohibitively, expensive nature of the requisite isolated sugars.^39–44

In this present manuscript, we provide detailed experimental data on metal-uronate interactions. This has been achieved by contacting aqueous solutions of various s-block metal cations with the monomeric units of which alginate is comprised, namely L-guluronate (GulA) and D-mannuronate (ManA) (Fig. 2), which we have recently prepared and characterised.⁴⁵ By studying the resulting solutions by NMR spectroscopy, changes to the anomeric equilibria and chemical shifts of the sugars upon metal coordination can be determined, which are indicative of the different metal-saccharide binding interactions established.²⁵ In addition, parallel experiments were also carried out with commercially available D-galacturonate (GalA) and D-glucuronate (GlcA) (Fig. 2). Both GalA and GlcA are found abundantly in nature (in pectins and hemicelluloses, respectively), and their metal-binding properties have been much more widely investigated than that of the algal uronates, hence providing a useful comparison.^{42–44,46–63}

Fig. 2 The pyranose forms of the four mono-uronates studied in this work.

To our knowledge, the work presented herein represents the most extensive investigation of the complexation of metal ions to mono-uronates to-date, and will help to inform future studies of the behaviour and applications of their parent polysaccharides.

Results

Changes in mono-pyranuronate anomeric equilibria in the presence of s-block metal ions

Initially, the interactions of the four mono-uronates (GulA, ManA, GlcA, and GalA) with different s-block cations were probed. In each case, the changes to the pyranose anomeric equilibria (expressed as the ratio of α-anomer to β-anomer, α/β) on addition of the metal ions was determined. Perturbations of the α/β ratio are indicative of preferential complexation of the metal by one of the two anomers. To this end, the ¹H NMR spectra (295 K, 400 MHz) of the sodium salts of GulA, ManA, GlcA, and GalA were recorded in D₂O solutions (42 mM, pD 7.9) containing the desired metal chloride salt (1.0 M). A control experiment in which no additional metal salt was added was run in parallel, (referred to as the “no metal” case). Following the acquisition of the NMR spectra at pD 7.9, the pD was lowered to 1.4 (well below the pK_a values of the uronates),⁴⁵ and the spectra re-recorded. In addition to the influence of pD, the impact of other variables (ionic strength and metal counter-anion) were also evaluated (ESI sections 3.1 and 3.2†). The measured α/β ratios are summarised in Fig. 3. Changes to the pyranose/furanose equilibria were generally small and are disregarded for this analysis, but are reported in full in ESI section 3.4.†

Fig. 3 The α/β ratios of (a) L-guluronate, (b) D-mannuronate, (c) D-glucuronate, and (d) D-galacturonate in 1.0 M aqueous solutions of different metal chloride salts at 295 K and pD 7.9 or 1.4.

GulA was found to exhibit the lowest α/β ratio of all of the four uronates studied (α/β = 0.21 ± 0.01 when measured at pD 7.9 in the metal-free case). It was found that addition of Na⁺, K⁺, or Mg²⁺ had virtually no effect, with α/β ratios all staying within ± 0.05 of the metal-free value (Fig. 3(a)). However, addition of the ions Ca²⁺, Sr²⁺ and (to a lesser extent) Ba²⁺ all perturbed the anomeric equilibrium strongly in favour of the α-pyranose configuration, yielding α/β values of 1.50 ± 0.03, 1.20 ± 0.02, and 0.58 ± 0.02, respectively.

In contrast to the effects determined for GulA, changes to the α/β ratios for the other mono-uronates on addition of metal ions were much smaller. For ManA, the only mono-uronate to have an inherent energetic preference for the α-anomer over the β-configuration under the metal-free conditions (α/β = 2.36 ± 0.04), addition of Na⁺, Ca²⁺, Sr²⁺, or Mg²⁺ had little impact on the α/β ratio (Fig. 3(b)). However, ions with larger radii than that of Sr²⁺ did promote a slight increase in the proportion of the β-ManA (with Ba²⁺ and K⁺, giving rise to α/β values of 1.36 ± 0.02 and 1.77 ± 0.02, respectively). The observation of a small increase in the proportion of mannofuranuronate upon the addition of Ca²⁺ is also noteworthy (ESI Table S.13†) as the formation of Ca²⁺-mannofuranose adducts has been identified previously.⁶⁴

Irrespective of the metal ions added GlcA showed no significant deviation from the α/β value of the free uronate (metal-free α/β ratio of 0.75 ± 0.01 at pD 7.9) (Fig. 3(c)). Similarly, GalA (metal-free α/β ratio of 0.70 ± 0.01 at pD 7.9) was not affected significantly by the presence of excess Na⁺, K⁺, Mg²⁺ or, Ba²⁺ ions (Fig. 3(d)). However, a stabilisation of α-GalA was detected upon addition of Ca²⁺ (α/β = 1.24 ± 0.01) and Sr²⁺ (α/β = 1.02 ± 0.01), albeit much smaller effects than those for α-GulA discussed above.

The trends in α/β ratios of the four mono-uronates remained the same upon changing the pD from 7.9 to 1.4. However, one point of difference was seen for GulA and GalA in the presence of Ca²⁺. At the lower pD, the stabilisation of α-GalA is virtually lost (with the pD 1.4 metal-free solution exhibiting an α/β ratio of 0.89 ± 0.01 compared to 0.91 ± 0.01 with Ca²⁺), in contrast to α-GulA (pD 1.4 metal-free α/β = 0.21 ± 0.01, compared to 1.11 ± 0.02 with Ca²⁺). These results indicated a notable degree of complexation between α-GulA and Ca²⁺, which persisted even upon protonation of the carboxylate moiety, something that is explored further in the following sections.

Notably, the lower pD conditions employed herein also allowed for analysis of the behaviour of the uronates in the presence of Zn²⁺. To keep it in solution to prevent Zn hydroxide. Here, the results were found to be almost exactly analogous to those of Mg²⁺, with only very minor changes detected in the α/β anomeric equilibria upon addition of Zn²⁺ relative to the no metal case (see ESI section 3.4†).

Changes to mono-uronate ¹H NMR chemical shift values in the presence of s-block metal ions

Having established the metals for which the α- and β-mono-pyranuronates show affinities (summarised in Fig. 3), the nature of the corresponding coordination modes were analysed by studying changes to ¹H chemical shifts (δ¹H) of the respective ¹H NMR signals of the uronates. In order to interpret the chemical shift data, changes in δ¹H of the uronate protons in the metal-containing solutions relative to the metal-free and (i.e. coordination chemical shifts, Δδ_obs) were first recorded. As large values for Δδ_obs were found for all uronate signals in the metal chloride solutions, it was convenient to define a second parameter, Δδ_rel, determined by subtracting the value of Δδ_obs of H4 of the particular anomer under investigation. The H4 signal was chosen arbitrarily as it was often the resonance most shifted following addition of the metal salts; see Fig. 2 for numbering schemes.

The metric Δδ_rel is helpful because it is close to zero in most cases, enabling rapid analysis of the large data sets (tables of all Δδ_obs and Δδ_rel values are provided for each experiment in the ESI section 3.3†). Where Δδ_rel is non-zero, it implies that the protons around the saccharide ring are affected differently by the presence of the metal ion, indicative of binding with different hydroxyl groups. Notably in this regard, the values of Δδ_rel at pD 7.9 for the protons of α-GulA, which were very small in the presence of Na⁺, K⁺, and Mg²⁺, were up to 10-fold larger when Ca²⁺, Sr²⁺, and Ba²⁺ ions were present (Table 1). These results therefore provide further evidence of a mode of complexation between α-GulA and larger divalent cations that is not accessible for other smaller metals. In contrast, the Δδ_rel values for β-Gul (Table 1), α- and β- ManA (ESI Tables S.6–7†), GlcA (ESI Tables S.8–9†) and GalA (ESI Tables S.8–9†), are all much smaller for all of the metals studied. The closest comparison to α-GulA is α-GalA, but even then, the differences are significant. For example, whilst α-GulA showed large Δδ_rel values in the presence of Ca²⁺ at pD 7.9 for H1 (0.27 ppm), H2 (0.19 ppm), H3 (0.13 ppm), and H5 (0.20 ppm), α-GalA only displayed significant Δδ_rel values for H1 (0.18 ppm) and H5 (0.13 ppm). Taken as a whole, these results indicate binding mode(s) for Ca²⁺, Sr²⁺, and Ba²⁺ in α-GulA, that are not operative for other metals or uronates.

Table 1 Changes in the ¹H chemical shift of protons of α- and β-pyranose anomers of Na-L-guluronate in the presence of different metal chloride salts (1.0 M), measured relative to H4 in each case (Δδ_rel). Values were recorded in D₂O at either pD 7.9 (upper) or pD 1.4 (lower) at 295 K and 400 MHz. Full sets of data for all of the mono-uronates studied, including uncorrected Δδ_obs values, are provided in the ESI (section S.3†)

Solution	pD	Alpha pyranose					Beta pyranose
Change in chemical shift (ppm) compared to metal-free solution relative to H4 (Δδrel)
		αH1	αH2	αH3	αH4	αH5	βH1	βH2	βH3	βH4	βH5
No metal	7.9	0.00	0.00	0.00	—	0.00	0.00	0.00	0.00	—	0.00
NaCl	7.9	0.04	0.03	0.04	—	0.04	0.01	0.01	0.02	—	0.02
KCl	7.9	0.03	0.02	0.03	—	0.01	0.01	0.01	0.01	—	0.01
CaCl2	7.9	0.27	0.19	0.13	—	0.2	0.05	0.04	0.01	—	0.07
SrCl2	7.9	0.28	0.20	0.14	—	0.22	0.07	0.06	0.03	—	0.08
BaCl2	7.9	0.28	0.20	0.14	—	0.22	0.12	0.07	0.03	—	0.11
MgCl2	7.9	0.06	0.00	0.02	—	0.04	0.01	0.00	0.01	—	0.02
Ca(NO3)2	7.9	0.24	0.15	0.09	—	0.19	0.05	0.04	0.02	—	0.06
Cal2	7.9	0.28	0.21	0.15	—	0.21	0.09	0.06	0.05	—	0.10

---

Solution	pD	Alpha pyranose					Beta pyranose
Change in chemical shift (ppm) compared to metal-free solution relative to H4 (Δδrel)
		αH1	αH2	αH3	αH4	αH5	βH1	βH2	βH3	βH4	βH5
No metal	1.4	0.00	0.00	0.00	—	0.00	0.00	0.00	0.00	—	0.00
NaCl	1.4	0.02	0.02	0.02	—	0.02	0.00	−0.01	0.00	—	0.01
KCl	1.4	0.02	0.01	0.01	—	0.01	0.00	0.00	0.00	—	0.01
CaCl2	1.4	0.13	0.15	0.10	—	0.11	0.01	0.00	0.02	—	0.01
SrCl2	1.4	0.14	0.16	0.10	—	0.12	0.01	0.00	0.02	—	0.01
BaCl2	1.4	0.15	0.15	0.10	—	0.14	0.04	0.01	0.01	—	0.04
MgCl2	1.4	0.00	0.00	0.01	—	−0.02	−0.01	−0.01	−0.01	—	0.00

---

Upon lowering the pD from 7.9 to 1.4, a decrease in the magnitude of the Δδ_rel values for α-GulA was observed in all cases, though not uniformly for all protons. Notably, in the presence of Ca²⁺, Δδ_rel(α-GulA-H1) and Δδ_rel(α-GulA-H5) were lowered by 0.14 and 0.09 ppm, respectively, on decreasing the pD from 7.9 to 1.4, whilst Δδ_rel(α-GulA-H2) and Δδ_rel(α-GulA-H3) were only reduced by 0.04 and 0.03 ppm, respectively. Again, this is evidence for a coordination mode of α-GulA that persists even after protonation. In contrast, the Δδ_rel values for β-GulA, α- and β-ManA, GlcA, and, GalA were all reduced to near zero at pD 1.4, demonstrating the importance of the carboxylate anion in the main binding modes of all of these saccharides.

¹H and ¹³C NMR spectroscopic titrations of mono-uronates with Ca²⁺

From the initial screening of the behaviour of mono-uronates in the presence of different metal ions, the interaction between α-GulA and Ca²⁺ was found to be unique. Consequently, a more in-depth series of experiments was carried out, whereby solutions of sodium mono-uronates (0.26 M, pD 7.9) were titrated with increasing amounts of calcium chloride (0.0–1.3 M) and studied by both ¹H and ¹³C NMR spectroscopy (295 K). As previously, the analysis was restricted to the α- and β-pyranose forms of the GulA and ManA uronates, as the furanose isomers are not present in sufficiently high concentrations. Although GlcA was again studied for comparison, GalA was excluded since precipitation occurs during the titration sequence (the reader is instead referred to the work of Jacques et al. for a similar study of GlcA and GalA together).⁵⁷

The values of Δδ_obs(¹³C) and Δδ_obs(¹H) of the ¹³C/¹H signals from the α- and β-anomers of GulA, ManA, and GlcA were plotted against the number of equivalents of Ca²⁺ added (Fig. 4 and Fig. 5, respectively). Representative spectra for GulA are provided in ESI section 3.5,† and show no line broadening across the range of the titration. Full data-sets for the titrations (including the corrected anomer concentrations at each value of [Ca²⁺]) are also provided in ESI section 3.6.†

Fig. 4 Uncorrected changes in chemical shifts, Δδ_obs (ppm), of carbon signals in the ¹³C NMR spectra of sodium mono-uronates upon the addition of aliquots of CaCl₂. The nominal concentration of uronate is 0.26 M in each case (though the individual concentration of each anomer must be calculated separately for every value of [Ca²⁺], see ESI section 2.6†). Spectra recorded in D₂O at pD 7.9, 295 K, 100 MHz. Each plot represents the average of two independent runs. Individual error bars are omitted for clarity, but an average error, σ, is stated in parentheses for each plot.

Fig. 5 Uncorrected changes in chemical shifts, Δδ_obs (ppm), of proton signals in the ¹H NMR spectra of sodium mono-uronates upon the addition of aliquots of CaCl₂. The nominal concentration of uronate is 0.26 M in each case (though the individual concentration of each anomer must be calculated separately for every value of [Ca²⁺], see ESI section 2.6†). Spectra recorded in D₂O at pD 7.9, 295 K, 400 MHz. Each plot represents the average of two independent runs. Individual error bars are omitted for clarity, but an average error, σ, is stated in parentheses for each plot.

The data presented in Fig. 4(a) show that the Δδ_obs(¹³C) values for α-GulA at 5.0 eq. Ca²⁺ span 1.71 ppm, whilst those of β-GulA vary by only 0.71 ppm (Fig. 4(b)). Similarly, the range of Δδ_obs(¹H) values for α-GulA at 5.0 eq. Ca²⁺ spans 0.24 ppm (Fig. 5(a)), compared to only 0.06 ppm for β-GulA (Fig. 5(b)). Whilst C4, C5, C6, and H4 all display similar Δδ values upon the addition of Ca²⁺ in both α- and β-GulA anomers, the large negative shift changes for αC1 (−0.87 ppm), αC3 (–1.30 ppm) and αC2 (–1.42 ppm) and positive shift changes for αH3 (0.05 ppm), αH2 (0.09 ppm), αH5 (0.11 ppm), and αH1 (0.17 ppm) contrast strongly with the equivalent β-shifts and, indeed, with all shifts determined in the experiments employing ManA and GlcA. In the cases of ManA and GlcA, Δδ_obs(¹³C) for C1, C2, C3, C4, and C6 and Δδ_obs(¹H) for H1-H5 all remain small and tightly clustered with increasing Ca²⁺ concentration, relative to those from α-GulA. In the case of C5, both α- and β-GulA show a positive Δδ_obs(¹³C) value (both + 0.30 ppm) corresponding to a deshielding of the C5 nucleus, whilst α- and β-ManA and α- and β-GlcA show negative shifts (−0.57 to −0.84 ppm) for the same carbon atom. These distinct patterns of chemical shift changes can be used to discern the nature of the binding modes between Ca²⁺ and the various uronates, which are evaluated synoptically in the discussion section.

Finally, the Ca²⁺/uronate NMR titrations also enabled the calculation of tentative stability constants for the metal-carbohydrate complexes, using the open source Bindfit model developed by Thordarson et al. (see ESI section 2.6†).^65–67 Using the data for C2-C4 for each anomer, the 1 : 1 Ca²⁺:uronate stability constants (K_1,1) were calculated: α-GulA 2.20 M⁻¹, β-GulA 0.61 M⁻¹, α-GlcA 0.38 M⁻¹, β-GlcA 0.25 M⁻¹, α-ManA 0.30 M⁻¹, β-ManA 0.26 M⁻¹, all with errors < ± 5%. Attempting to fit the data arising from carbons close to the carboxylate and ring oxygens (C1, C5, C6) to the same model as C2–C4 was unsuccessful, indicating that additional binding modes utilising the carboxylate moiety were also present.

Discussion

Coordination of α-GulA to Ca²⁺ and similar dications

From the preceding NMR spectroscopic experiments, it is clear that α-GulA shows a preferential affinity for the larger divalent cations (Ca²⁺r₊ = 114 pm, Sr²⁺r₊ = 132 pm, Ba²⁺r₊ = 149 pm) as evidenced by the significant changes in anomeric equilibria (α/β ratio) upon coordination of these ions compared to the monovalent (Na⁺r₊ = 113 pm, K⁺r₊ = 151 pm) and smaller divalent ions (Mg²⁺r₊ = 71 pm, Zn²⁺r₊ = 74 pm).⁶⁸ In contrast to GulA, the α/β ratio for ManA, GlcA, and GalA show minimal perturbations upon contact with all of the cations studied (Fig. 3). Together these observations suggest that α-GulA adopts a binding mode with Ca²⁺ (and similar dications) that lowers the free energy of the molecule relative to the β-form, which is unique amongst the eight uronate anomers studied.

It is proposed that the ax–eq–ax arrangement of hydroxyl groups found for C1, C2, and C3 of α-GulA is responsible for the Ca²⁺ coordination. It has previously been established that the 1,3-di-axial clash of hydroxyl oxygen atoms renders the ax–eq–ax motif particularly unstable,^69,70 yet here we observe that the dipolar charges are screened by the presence of a suitable cation, which stabilises the structure significantly. The identification of ax–eq–ax binding by α-GulA is consistent with the stability trends elucidated by Angyal, for metal ion coordination to different arrangements of hydroxyl groups in cyclic polyols.^{19,25,71–75}

If the change in α/β ratio is used as the metric to determine the favourability of coordination between divalent cations and a monopyranuronate then a metal ion affinity series can be established, which for α-GulA is Ca²⁺ > Sr²⁺ > Ba²⁺ ≫ Mg²⁺ ≈ Zn²⁺. Thus, whilst Ca²⁺ may be regarded as being well suited to binding to the ax–eq–ax arrangement, Mg²⁺ and Zn²⁺ are either too small or too strongly solvated by H₂O to be able to coordinate. In this context, it is tempting to suggest that the notably lower affinity for Ba²⁺ of α-GulA is a result of a combination of its large ionic radius and low charge density, which minimise the strength of its binding to the ax–eq–ax arrangement of hydroxyl groups. However, from the NMR spectroscopic experiments described herein alone, it is not possible to rule out that the smaller perturbation of the equilibrium towards the α-anomer of GulA by Ba²⁺ may simply arise because this metal ion can also coordinate to the β-anomer in a favourable manner (discussed further below).

Further proof of the Ca²⁺ coordination to the ax–eq–ax arrangement of hydroxyl groups on α-GulA is given by the coordination chemical shift data Δδ_obs(¹³C) and Δδ_obs(¹H). Shielding of carbons C1, C2, and C3 and deshielding of protons H1, H2, and H3 can be considered to be diagnostic of ax–eq–ax binding, owing to the polarisation of the respective C–H bonds by the electric field of the cation.^57,72 Whilst qualitatively this trend is observed for α-GulA in the presence of Ca²⁺ (Fig. 4(a) and 5(a)), the numerical results do not match perfectly with those observed by Angyal for other Ca²⁺/ax–eq–ax complexes, such as epi-inositol.⁷² Furthermore, the large deshielding of the αH5 signal observed here for α-GulA would not be predicted by a pure ax–eq–ax binding arrangement. Together these anomalies indicate the presence of a second coordination mode in the α-GulA system.

It seems likely that such a secondary mode would be similar to that previously found in α-GalA, as the two anomers possess the same relative orientation of functional groups on C1, C2, C4, and C5. α-GalA is believed to bind Ca²⁺ through the carboxylate moiety and ring oxygen.^43,49 Comparison of the Δδ_obs(¹H) values of α-GalA in the presence of Ca²⁺ (see ESI Table S.10†) show a deshielding of H1 and H5, which is also be observed in the α-GulA system. Hence, it can be concluded that Ca²⁺ binding to α-GulA can occur through both (i) the ax–eq–ax hydroxyl groups and (ii) the carboxylate plus ring oxygen (Fig. 6). It was calculated that the tentative stability constant of Ca²⁺ binding through the first mode K_1,1(α-GulA, ax–eq–ax) = 2.20 M⁻¹, which is roughly similar to K_1,1(ax–eq–ax) determined for the analogous saccharides α-L-gulose (K_1,1 ≈ 3.7 M⁻¹)^25,76 and epi-inositol (K_1,1 ≈ 3.0 M⁻¹).⁷² The stability constant for Ca²⁺ binding to the second site, K_1,1(α-GulA, carboxylate), could not be calculated here in this qualitive study as the data in the early stages of the titration are insufficiently granular, but is likely to be higher than K_1,1(α-GulA, ax–eq–ax) based on literature values for K_1,1(GlcA) and K_1,1(GalA) at pD = 7–8.^44,55–59 At low pD however, the magnitude of K_1,1(carboxylate) is heavily reduced due to protonation, as evidenced by the loss of affinity of α-GalA for Ca²⁺ observed here, and also by other authors.⁵⁴ In contrast, the extent of binding of Ca²⁺ by α-GulA is only slightly diminished by acidifying the solution, as the ax–eq–ax coordination mode remains operative.

Fig. 6 The two modes for Ca²⁺-coordination by (¹C₄)-α-L-guluronate detected in this work. (Upper) A coordination mode involving the ax–eq–ax arrangement of hydroxyl groups unique to the α-GulA anomer, and (lower) the carboxylate/ring-oxygen binding observed generally for α-mono-uronates.

Whilst the coordination behaviour of the monomeric form of α-L-guluronate does not necessarily translate to that of its polymeric alginate analogue, it is nevertheless interesting to note that both binding modes detected here (ax–eq–ax and carboxylate) are featured in the Egg Box model as it is classically depicted (Fig. 3).^18–20 This affinity of α-GulA for larger divalent metal ions may also influence the interactions of oligo-uronides with enzymes such as bacterial alginate lyases, which are believed to employ Ca²⁺ for catalysis.⁷⁷

Coordination of other uronate anomers to Ca²⁺

Compared to the effects of Ca²⁺ ions on α-GulA (discussed above), the influence of this dication on the other uronates is more subtle. The non-carboxylate K_1,1 values calculated for Ca²⁺ with β-GulA, α-GlcA, β-GlcA, α-ManA, and β-ManA were around 3–7 times smaller than K_1,1(α-GulA, ax–eq–ax). The small K_1,1 values (≪1.0 M⁻¹) for all anomers except α-GulA are consistent with the much weaker coordination by less distinct mono-/bi-dentante arrangements of hydroxyl groups on those saccharides,^25,55,78 which are not unambiguously identifiable from the chemical shift data.

With the exception of α-GulA therefore, it can be concluded that Ca²⁺ binding occurs predominantly through the carboxylate moiety of most uronates. This is something that is in line with previous reports of Ca²⁺ coordinating preferentially to the α-anomers of GlcA and GalA through the carboxylate and ring oxygens, and less favourably to their β-anomers through just the carboxylate moiety.^{43,44,49,57,58} These literature results are further supported by the results presented here. For example, the α/β ratio measurements are consistent with slightly higher affinities of α-GlcA and α-GalA for Ca²⁺ than the affinities of their β-anomers (Fig. 3a). Similarly, Δδ_rel for H1 and H5 were greater than those of H2 and H3 for α-GlcA and α-GalA, but in the β-anomers, these values were more similar (see ESI†). The same Δδ_rel patterns were also shown by α- and β-ManA, and β-GulA. From this, we conclude that α-ManA predominantly binds to Ca²⁺via the carboxylate and ring oxygens (like α-GalA and α-GlcA), whilst β-ManA and β-GulA predominantly bind through just the carboxylate group (like β-GalA and β-GlcA). From this qualitative study we have not been able to provide K_1,1 values for these carboxylate binding modes, but note that they are likely to be much higher than the K_1,1 values for the hydroxyl-binding modes reported above, based on previous literature reports.

Other notable metal/uronate interactions

A final observation that is worthy of note relates to the coordination of cations larger than Sr²⁺ to GulA and ManA. Despite the affinity of α-GulA for Ca²⁺ and Sr²⁺, changes to the α/β ratio in the presence of Ba²⁺ do not appear to be as large, possibly indicating a more favourable interaction of β-GulA with this larger ionic radii dication compared to that of α-GulA as a result of unfavourable binding of the latter. Additionally, ManA, which showed minimal response to the addition of metal ions, tended towards the formation of the β-anomer in the presence of Ba²⁺ and K⁺. Hence, it can be concluded that larger ions such as Ba²⁺ or K⁺ are able to coordinate to the wide, open faces of the pyranose rings of β-ManA and β-GulA, specifically giving rise to interactions with β-GulA-O2 and β-GulA-O3, and β-ManA-O1, β-ManA-O2, and β-ManA-O_ring (Fig. 7). Indeed, cis-hydroxyl groups (as found on β-ManA and β-GulA) were identified by Angyal as being favourable binding sites for metal ions in carbohydrates, although less so than the ax–eq–ax arrangement.²⁵

Fig. 7 Possible coordination modes between β-anomers of (upper) GulA and (lower) ManA and the large Ba²⁺ ion. Both complexes involve a cis arrangement of hydroxyl oxygen atoms.

Conclusions

The interactions of the two monomeric units of alginate, GulA and ManA, and two related biomass-derived monosaccharides (GlcA and GalA) with environmentally abundant s-block metal cations have been investigated. Particular focus centred on the interaction of the α-pyranose anomer of GulA, which is of relevance to the metal coordination proposed by the long-standing Egg Box model of metal-alginate binding.^18–20

Here, we have shown that one of the monomeric units of alginate, α-GulA, shows a preferential affinity for the coordination of larger divalent cations (Ca²⁺, Sr²⁺, and Ba²⁺). This is attributed to the greater stabilisation conferred to this anomer over that achieved with its lower energy β-GulA counterpart as a result of the coordination of these dications. The origin of this stabilisation is ascribed predominantly to be the result of an interaction between the divalent cation and the ax–eq–ax arrangement of hydroxyl atoms that is unique to α-GulA. The ax–eq–ax binding mode of α-GulA has a value of K_1,1 of 2.20 M⁻¹ with Ca²⁺, which is around 3-times higher than any such hydroxyl-binding mode on the β-GulA counterpart. Furthermore, this coordination motif remains operative even at low pH. A second binding mode of Ca²⁺ with α-GulA was also detected, which is generic to all α-mono-uronates, and involves the carboxylate moiety and ring oxygen. In contrast, the β-anomers principally coordinate to metal ions through just their carboxylate groups. Additionally, we report evidence for weak complexation of very large cations by the cis hydoxyl groups of β-ManA and β-GulA.

Together, this study represents the most comprehensive comparative analysis to-date of the coordination of metal ions to mono-uronates. Whilst the results presented for metal binding to GulA, ManA, GalA, and GlcA do not necessarily reflect the coordination environment present in the parent biopolymers, the results provide valuable data for the benchmarking of computational models. Such models will ultimately facilitate a fuller understanding of metal coordination by alginates and related polyuronides, both in their natural environments and in their important technological applications.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

The staff of the Durham University NMR service, Drs Alan Kenwright and Juan Aguilar, are thanked for their assistance in the running of experiments for this work. Funding: This work was supported by the Centre for Process Innovation (CPI), Durham University, and the Engineering and Physical Sciences Research Council (EPSRC).

Notes and references

1. I. Donati and S. Paoletti, in Microbiology Monographs, 2009, vol. 13, pp. 1–53.
2. E. D. T. Atkins, I. A. Nieduszynski, W. Mackie, K. D. Parker and E. E. Smolko, Biopolymers, 1973, 12, 1879–1887.
3. E. D. T. Atkins, I. A. Nieduszynski, W. Mackie, K. D. Parker and E. E. Smolko, Biopolymers, 1973, 12, 1865–1878.
4. T. A. Davis, B. Volesky and A. Mucci, Water Res., 2003, 37, 4311–4330.
5. B. Gyurcsik and L. Nagy, Coord. Chem. Rev., 2000, 203, 81–149.
6. P. Gacesa, Carbohydr. Polym., 1988, 8, 161–182.
7. W. J. S. Peschardt, US2403547A, 1946, 1–2.
8. M. Wilkinson, Aquat. Conserv. Mar. Freshw. Ecosyst., 1992, 2, 209–210.
9. K. Y. Lee and D. J. Mooney, Prog. Polym. Sci., 2012, 37, 106–126.
10. K. I. D. Skja, in Renewable Resources for Functional Polymers and Biomaterials : Polysaccharides, Proteins and Polyesters, 2011, pp. 186–209.
11. K. G. Mandel, B. P. Daggy, D. A. Brodie and H. I. Jacoby, Aliment. Pharmacol. Ther., 2000, 14, 669–690.
12. S. S. Pedersen, A. Kharazmi, F. Espersen and N. Høiby, Infect. Immun., 1990, 58, 3363–3368.
13. A. Haug, J. Bjerrum, O. Buchardt, G. E. Olsen, C. Pedersen and J. Toft, Acta Chem. Scand., 1961, 15, 1794–1795.
14. R. Kohn, Pure Appl. Chem., 1975, 42, 371–397.
15. Q. Yu, J. T. Matheickal, P. Yin and P. Kaewsarn, Water Res., 1999, 33, 1534–1537.
16. W. M. Antunes, A. S. Luna, C. a. Henriques and A. C. A. Da Costa, Electron. J. Biotechnol., 2003, 6, 174–184.
17. P. Lodeiro, B. Cordero, J. Barriada, R. Herrero and M. Sastredevicente, Bioresour. Technol., 2005, 96, 1796–1803.
18. G. T. Grant, E. R. Morris, D. A. Rees, P. J. C. Smith and D. Thom, FEBS Lett., 1973, 32, 195–198.
19. S. J. Angyal, Pure Appl. Chem., 1973, 35, 131–146.
20. O. Smidsrød, A. Haug, S. G. Whittington, E. Sjöstrand and S. Svensson, Acta Chem. Scand., 1972, 26, 2563–2566.
21. F. Topuz, A. Henke, W. Richtering and J. Groll, Soft Matter, 2012, 8, 4877.
22. A. Haug, O. Smidsrød, B. Högdahl, H. A. Øye, S. E. Rasmussen, E. Sunde and N. A. Sørensen, Acta Chem. Scand., 1970, 24, 843–854.
23. A. Penman and G. R. Sanderson, Carbohydr. Res., 1972, 25, 273–282.
24. R. Kohn, B. Larsen, L. J. Sæthre, E. Sjöstrand and S. Svensson, Acta Chem. Scand., 1972, 26, 2455–2468.
25. S. J. Angyal, Adv. Carbohydr. Chem. Biochem., 1989, 47, 1–43.
26. C. DeRamos, A. Irwin, J. Nauss and B. Stout, Inorg. Chim. Acta, 1997, 256, 69–75.
27. I. Braccini and S. Pérez, Biomacromolecules, 2001, 2, 1089–1096.
28. W. Plazinski and M. Drach, J. Phys. Chem. B, 2013, 117, 12105–12112.
29. W. Plazinski and M. Drach, Appl. Surf. Sci., 2012, 262, 153–155.
30. M. B. Stewart, S. R. Gray, T. Vasiljevic and J. D. Orbell, Carbohydr. Polym., 2014, 102, 246–253.
31. P. Agulhon, V. Markova, M. Robitzer, F. Quignard and T. Mineva, Biomacromolecules, 2012, 13, 1899–1907.
32. K. Panczyk, K. Gaweda, M. Drach and W. Plazinski, J. Phys. Chem. B, 2018, 122, 3696–3710.
33. L. Bekri, M. Zouaoui-Rabah, M. Springborg and M. S. Rahal, J. Mol. Model., 2018, 24, 312.
34. C. A. Steginsky, J. M. Beale, H. G. Floss and R. M. Mayer, Carbohydr. Res., 1992, 225, 11–26.
35. Z.-Y. Wang, Q.-Z. Zhang, M. Konno and S. Saito, Biopolymers, 1993, 33, 703–711.
36. L. Li, Y. Fang, R. Vreeker, I. Appelqvist and E. Mendes, Biomacromolecules, 2007, 8, 464–468.
37. P. Sikorski, F. Mo, G. Skjåk-Braek and B. T. Stokke, Biomacromolecules, 2007, 8, 2098–2103.
38. W. Plazinski and M. Drach, New J. Chem., 2015, 39, 3987–3994.
39. F. Mo, T. J. Brobak and I. R. Siddiqui, Carbohydr. Res., 1985, 145, 13–24.
40. R. Kohn, I. Furda, A. Haug and O. Smidsrød, Acta Chem. Scand., 1968, 22, 3098–3102.
41. J. T. Triffitt, Nature, 1968, 217, 457–458.
42. T. Anthonsen, B. Larsen, O. Smidsrød, Å. Pilotti, S. Svensson and C.-G. Swahn, Acta Chem. Scand., 1973, 27, 2671–2673.
43. S. J. Angyal, D. Greeves and L. Littlemore, Carbohydr. Res., 1988, 174, 121–131.
44. J. W. Haas, Mar. Chem., 1986, 19, 299–304.
45. J. S. Rowbotham, J. A. Aguilar, A. M. Kenwright, H. C. Greenwell and P. W. Dyer, Carbohydr. Res., 2020, 495, 108087.
46. S. Thanomkul, J. A. Hjortås and H. Sørum, Acta Crystallogr., Sect. B: Struct. Crystallogr. Cryst. Chem., 1976, 32, 920–922.
47. S. E. B. Gould, R. O. Gould, D. A. Rees and W. E. Scott, J. Chem. Soc., Perkin Trans. 2, 1975, 237.
48. H. S. Isbell and H. L. Frush, J. Res. Natl. Inst. Stand. Technol., 1944, 32, 77–94.
49. K. Izumi, Agric. Biol. Chem., 1980, 44, 1623–1631.
50. B. J. Kvam, H. Grasdalen, O. Smidsrød, T. Anthonsen and R. Kivekäs, Acta Chem. Scand., 1986, 40b, 735–739.
51. H. Grasdalen, T. Anthonsen, B. Larsen, O. Smidsrød, O. Dahl, O. Buchardt and G. Schroll, Acta Chem. Scand., 1975, 29b, 99–108.
52. L. Fuks, D. Filipiuk and W. Lewandowski, J. Mol. Struct., 2001, 563–564, 587–593.
53. H. Grasdalen, T. Anthonsen, O. Harbitz, B. Larsen, O. Smidsrød, C.-O. Pontchour, P. Phavanantha, S. Pramatus, B. N. Cyvin and S. J. Cyvin, Acta Chem. Scand., 1978, 32a, 31–39.
54. T. Anthonsen, B. Larsen, O. Smidsrød, M. J. Tricker and S. Svensson, Acta Chem. Scand., 1972, 26, 2988–2989.
55. B. Kutus, X. Gaona, A. Pallagi, I. Pálinkó, M. Altmaier and P. Sipos, Coord. Chem. Rev., 2020, 417, 213337.
56. A. Pallagi, C. Dudás, Z. Csendes, P. Forgó, I. Pálinkó and P. Sipos, J. Mol. Struct., 2011, 993, 336–340.
57. L. W. Jaques, J. B. Macaskill and W. Weltner, J. Phys. Chem., 1979, 83, 1412–1421.
58. R. O. Gould and A. F. Rankin, J. Chem. Soc. D, 1970, 489–490.
59. D. M. Whitfield, S. Stojkovski and B. Sarkar, Coord. Chem. Rev., 1993, 122, 171–225.
60. J. A. Rendleman, Food Chem., 1978, 3, 47–79.
61. L. Fuks and J.-C. G. Bünzli, Helv. Chim. Acta, 1993, 76, 2992–3000.
62. L. DeLucas, C. E. Bugg, A. Terzis and R. Rivest, Carbohydr. Res., 1975, 41, 19–29.
63. J. Hjortås, B. Larsen, S. Thanomkul, I. Szabo-Lin, C. Guthenberg and B. Mannervik, Acta Chem. Scand., 1974, 28b, 689–689.
64. D. C. Craig, N. C. Stephenson and J. D. Stevens, Carbohydr. Res., 1972, 22, 494–495.
65. http://supramolecular.org .
66. D. B. Hibbert and P. Thordarson, Chem. Commun., 2016, 52, 12792–12805.
67. P. Thordarson, Chem. Soc. Rev., 2011, 40, 1305–1323.
68. R. D. Shannon, Acta Crystallogr. Sect. A, 1976, 32, 751–767.
69. A. E. Vickman, D. C. Ashley, M.-H. Baik and N. L. B. Pohl, J. Mol. Model., 2017, 23, 214.
70. S. J. Angyal, in Advances in Carbohydrate Chemistry and Biochemistry, 1984, pp. 15–68.
71. S. J. Angyal, Chem. Soc. Rev., 1980, 9, 415–428.
72. S. J. Angyal and K. P. Davies, J. Chem. Soc. D, 1971, 500–501.
73. S. J. Angyal, Aust. J. Chem., 1972, 25, 1957.
74. M. E. Evans and S. J. Angyal, Carbohydr. Res., 1972, 25, 43–48.
75. S. J. Angyal, Tetrahedron, 1974, 30, 1695–1702.
76. H. S. Isbell, Bur. Stand. J. Res., 1930, 5, 741–755.
77. F. Xu, F. Dong, P. Wang, H.-Y. Cao, C.-Y. Li, P.-Y. Li, X.-H. Pang, Y.-Z. Zhang and X.-L. Chen, J. Biol. Chem., 2017, 292, 4457–4468.
78. B. Kutus, D. Ozsvár, N. Varga, I. Pálinkó and P. Sipos, Dalton Trans., 2017, 46, 1065–1074.

---

Footnote

† Electronic supplementary information (ESI) available: Experimental details and NMR parameters, full datasets for chemical shift changes, observed equilibrium populations, and titration measurements. See DOI: 10.1039/d0dt04375c

---