Biomacromolecules 2001, 2, 1089-1096 1089

Molecular Basis of Ca2+-Induced Gelation in Alginates and

Pectins: The Egg-Box Model Revisited
Isabelle Braccini and Serge Pérez*
Centre de Recherches sur les Macromolécules Végétales,† CNRS, BP 53 X,
38041 Grenoble Cedex, France
Received January 8, 2001; Revised Manuscript Received July 16, 2001

For many ionic polysaccharides, the ability to form gels in the presence of divalent cations such as calcium
is the key to biological functions and technological applications. This is particularly true for alginates and
pectins, where the regular occurrence of respectively R-L-(1-4)-guluronate residues and R-D-galacturonate
residues generates ordered templates for polymer chain associations that are involved in physical gels. The
molecular basis responsible for the strength and the stereospecificity of calcium interactions for the two
polysaccharides were investigated in a previous paper (Braccini; et al. Carbohydr. Res. 1999, 119). In the
present work, a novel molecular modeling procedure has been developed; it involves a pairing procedure
that evaluates all the possible associations of the ordered polyuronate chains with calcium ions to form
dimers. Starting from the stable ordered forms of polygalacturonate and polyguluronate, all possible ways
to form Ca2+-bridged dimers were computed; the parallel and antiparallel relative arrangements of the chains
were also considered. Despite the structural analogy between polyguluronate and polygalacturonate chains,
significant differences at the level of chain-chain associations are found. The popular “egg box model”
can still be referred to in the case of polyguluronate. However, it cannot be used to describe a pectate
junction zone as the unique feature of two consecutive chelation site per repeat, that provides a favorable
entropic contribution to the interchain association is not reproduced by this pioneering model. The body of
these results corroborates the two-stage process in the mechanism of calcium gelation, where the formation
of strongly linked dimer associations is followed by the formation of weak inter-dimer associations mainly
governed by electrostatic interactions.

Introduction
Pectins1 and alginates2 are natural ionic polysaccharides
with many applications in the food and pharmaceutical
industries. The extensive use of these polymers is derived
from their ability to form gels in the presence of divalent
cations such as calcium ions. Gelation has been demonstrated
to result from specific and strong interactions between
calcium ions and blocks of galacturonic and guluronic acid
residues for pectins and alginates, respectively (Figure 1).
Polygalacturonate and polyguluronate chains bind calcium
strongly, as demonstrated by the high selectivity coefficients.3
In the case of alginate, gel strength is also related to the
level of polyguluronate present.4 Numerous physicochemical
studies have been performed in order to obtain information
regarding the mechanisms and the structural features involved
in the gelation process. Morris et al. for alginates5 and
Ravanat and Rinaudo for pectins6 have shown that calcium
ions induce chain-chain association. These associations
would constitute the junction zones responsible for gel
formation. A model for the junction zone has been derived,
popularly known as the “egg box model”.7 In this model,
pairs of 21 helical chains are packed with the calcium ions
located between them (Figure 2). Even if this model is Figure 1. Schematic drawings of (a) galacturonate and (b) guluronate
chains.
* To whom correspondence should be addressed. Telephone: 33-476-
03-76-30. Fax: 33-476-03-76-29. E-mail: perez@cermav.cnrs.fr. commonly adopted, it has been questioned several times and
† Affiliated with the Université Joseph Fourier, Grenoble, France. is still a subject of debate. The original model proposed 10
10.1021/bm010008g CCC: $20.00 © 2001 American Chemical Society
Published on Web 09/29/2001
1090 Biomacromolecules, Vol. 2, No. 4, 2001
Figure 2. Schematic drawing and calcium coordination of the “egg
box model” as described for the pair of guluronate chains in calcium
alginate junction zones. Dark circles represent the oxygen atoms
involved in the coordination of the calcium ion.
oxygen atoms from the guluronate chains involved in the
coordination of the calcium ions. This does not seem very
likely as calcium-carbohydrate complexes8 generally exhibit
7- to 9-fold coordination of Ca2+ (mainly 8-fold), including
one to four oxygen atoms from water. A more realistic
coordination pattern has been proposed by Mackie et al.,9
supported by molecular modeling.
Conventional experimental techniques used to gather
detailed structural information are not well suited to study
intermediate states such as a polysaccharide gel where
ordered and disordered conformations coexist. Considering
the discrepancies in the experimental data and the speculative
nature of the proposed models, we decided to investigate
the structural features of the junction zones at the molecular
level by studying the associations of galacturonate and
guluronate chains with calcium ions using molecular model-
ing. In a preceding paper,10 we documented the molecular
basis responsible for the strength and stereospecificity of
calcium interactions for the two polymers. In the present
study, using these previous results, we computed all the
possible associations of two ordered uronate chains with
calcium ions to form dimers. In particular, we compared 31
and 21 helical conformations in pectate chain-chain associa-
tions and evaluated the relevance of the “egg box model” as
commonly invoked for both galacturonate and guluronate
systems in terms of feasibility and stability. Finally, dimer-
dimer interactions were also evaluated in order to explore
their role, if any, in the formation of a three-dimensional
network.
Methodology and Computational Methods
In the representation of a junction zone, we consider two
chains interacting via calcium ions; the two chains are either
in a parallel or antiparallel arrangement. The calculated chain
pairings refer to the first-stage process in the formation of
the junction zones, i.e., the dimerization. In agreement with
the models of physical gels, we assume that the chain
segments involved in the pairwise associations exhibit a
regular helical conformation and that the calcium ions are
periodically distributed along the chains.
Single Chain Conformation. The stable regular helical
conformations of polyguluronate and polygalacturonate were
Braccini and Pérez
Figure 3. Interhelical parameters computed by CHACHA program:
µA and µB, rotation of chains A and B; ∆Z, translation of chain B with
respect to chain A; ∆X, distance between the helical axis of chains A
and B after the contacting procedure.
reported and fully documented in our prior paper,10 which
was focused on the conformational and configurational
features of four acidic polysaccharides along with their ability
to bind calcium ions. The regular helices were extrapolated
from the low energy regions of the adiabatic maps of the
disaccharide subunits using the program POLYS11 (see ref
10 for the detailed description of the procedure). Using the
geometrical features of these stable helices which were
previously determined, we generated corresponding dodecam-
ers. The structures are 21 and 31 helical conformations for
polygalacturonate and two 21 helical conformations for
polyguluronate. As a prerequisite for the chain-chain
interaction procedure, the helix axes of all the individual
chains are coincident with the Z-axis.
Positions of Calcium Ions. Our previous results10 regard-
ing the interaction between calcium ions and polygalactur-
onate and polyguluronate have demonstrated the high
specificity of the interaction with the presence of a well-
defined periodic and unique chelation site for calcium along
the uronate chains. These results were established with the
program GRID,12 which computes the favorable positions
near a target molecule (i.e., the uronate chain) with which a
small chemical moiety (i.e., the calcium ion) would interact
favorably. Consequently, the program GRID was used to
place the calcium ions along the helical uronate dodecamers.
Chain-Chain Interaction Procedure. To evaluate the
favorable chain pairing configurations in the presence of
calcium ions for polygalacturonate and polyguluronate, a
computer program capable of evaluating chain-chain inter-
actions was utilized.13 In this procedure, the inter-helical
distance, ∆X, along with the interaction energy between two
chains is computed for all the relative positions of one chain
(A) with respect to the other (B) (Figure 3).
In the chain-chain interaction procedure, the two chains
in the system are defined as follows: (A) uronate chain with
Ca2+-Induced Gelation
calcium ions placed at the favorable periodic chelation sites
(see above); (B) the same uronate chain without calcium ions.
The final dimer system is electrically neutral.
Initially, the two chains are superimposed in either a
parallel or in an antiparallel arrangement, with their helical
axes coinciding with the Z-axis. Then, the calculation is
performed by rotating A (µA) and B (µB) by 10° increments
over the angular range 360°/n (using the symmetry of the
system) and by translating B (∆Z) over the length of a whole
fiber repeat, by about 0.25 Å increments. For each combina-
tion of the µA, µB, and ∆Z parameters, the chain B is moved
away from chain A along the X axis, until the van der Waals
surface of the A and B chains are in contact, with no
interpenetrating atoms. As this criteria is not valid for two
atoms involved in a hydrogen bond, the hydroxyl protons
cannot be considered in the interaction procedure, to not
violate the above-mentioned van der Waals radii conditions.
In the same manner, a minimum Ca‚‚‚O distance of 2.3 Å
was allowed in accordance with experimental data.8
The same procedure was applied to dimer-dimer interac-
tions of the uronic acids (total of four chains). This time,
the A and B systems are identical and represent the best
dimer structures of galacturonate and guluronate systems.
In this case, µA and µB are explored over the whole angular
range (0-360°).
Interchain Energy Calculations. The interaction energy
of the two chains is considered on an atom-atom potential
basis and is calculated as the sum over all contributions of
pairwise interactions. The expression used in the evaluation
of the interchain interaction energy is
NA NB
EAB ) ∑ ∑ Elj + Ehb + Eel
i)1 j)1
where NA and NB are the number of atoms in chains A and
B respectively, Elj, Ehb, and Eel are the three enthalpic
components Lennard-Jones, hydrogen bonds, and electrostat-
ics of the interaction, respectively. The van der Waals
contribution is represented by the 6-12 Lennard-Jones
potential:
Elj(i,j) ) A/rij12 - B/rij6
where rij is the interatomic distance, and A and B are
coefficients related to the atomic van der Waals radii,
polarizabilities, and effective numbers of electrons. The
hydrogen bonds are defined on the oxygen-oxygen distance
criteria, and the energy contribution is calculated using:
Ehb(i,j) ) 33.14(rij - 2.55)(rij - 3.05)
The distance between the two oxygen atoms should lie
between 2.55 and 3.05 Å. The electrostatic term is defined
following the expression used in the GRID program:
( )
2 1
Eel(i,j) ) K*qiqj* * molecules fill the remaining chelation sites.
M + W rij
where K is a constant, qi and qj are the partial atomic charges
of the i, j interacting atoms, and M and W are the dielectric
constants of the target system and the surrounding medium,
Biomacromolecules, Vol. 2, No. 4, 2001 1091
respectively. In the present study, we used M ) 4 and W
) 80 (water). A cutoff distance of 12.5 Å was used in
computing the electrostatic contribution.
All of the parameters used in these calculations were
extracted from, or computed (partial charges, qi) by the GRID
program, to be consistent with the previous GRID calcula-
tions.10
Optimization of the Favorable Associations. For the
most favorable pairwise associations (dimers or association
of dimers), the positions of calcium ions were reoptimized
with respect to the whole system. This was accomplished
by removing the calcium ions and subsequently performing
a GRID calculation on the negatively charged structures,
using a calcium probe.
Computational Problem and Approximations
In calcium pectate and alginate gels, the junction zones
involve three components: uronate chains, calcium ions, and
water. However, using molecular mechanics calculations, it
is not possible to evaluate the relative positions and energetics
of the chains, the calcium ions, and at the same time, the
number and the position of the participating water molecules
since calcium coordination is very flexible with respect to
the number of ligands and to the direction of Ca‚‚‚O vectors.
Consequently, water molecules were not explicitly included
in the calculations. It may be expected that the conformation
of the pectate chain depends on the presence or absence of
water. To explore these two possibilities, both 31 (solid form)
and 21 (a priori aqueous and dilute gel forms) helical
conformations were considered in these calculations. The
remaining question to address is if the water molecules have
an influence on the way the chains interact to form a junction
zone. The experimental data tend to indicate that, for both
systems, calcium-uronate chain interaction is the predomi-
nant parameter. Selectivity coefficients indicated a strong
calcium binding for both polygalacturonate and polygulur-
onate and weak calcium binding for polymannuronate.3
Therefore, differences in ion binding properties of polyur-
onates are caused by differences in the steric arrangement
of the active groups in the polymer chain. Studies of
calcium-carbohydrate structures demonstrated that calcium-
carbohydrate interactions display considerable stereospeci-
ficity.8 Angyal identified some very efficient arrangements
of hydroxyl groups in sugars for calcium binding.14 Thibault
and Rinaudo reported that calcium activity coefficients
suggest the occurrence of some specific interactions of the
pectin molecules with calcium ions.15 Furthermore, in our
preceding paper, we evaluated calcium-uronate interaction
for several polysaccharides and, in agreement with the above
experimental data, observed a strong and very specific
calcium binding for both polygacturonate and polyguluronate
chains. In view of these experimental and theoretical data,
we assume that calcium-uronate chain interaction is pre-
dominant in the formation of a junction zone and that water
Results
A chain pairing procedure was developed to evaluate the
efficiency of packing two chains together and to predict
1092 Biomacromolecules, Vol. 2, No. 4, 2001 Braccini and Pérez

Table 1. Calculated Interchaina and Energyb Parameters for the bond network with two complementary hydrogen bonds:
Best Chain-pairing Situations of Calcium Guluronate Oligomer (DP O2‚‚‚O6, 2.73 Å (2.88 Å), and O3‚‚‚O5, 2.87 Å (2.91 Å). It
12) in Both Parallel and Antiparallel Arrangements and for the Egg
Box Model also provided a unique periodic chelation site for calcium
parameters
ions, as predicted by GRID. It is not possible to discriminate
between the two forms of these chain pairings based on total
µA µB ∆Z ∆X ET Elj Ehb Eel
energy; however, the antiparallel arrangement is probably
21
favored in the gel since it has been reported that urea
| 120 120 8.12 6.24 -122 -31 -21 -70
substantially weakens the gel20 indicating a notable contribu-
anti-| 60 110 3.36 6.37 -122 -36 -47 -39
egg box 70 110 2.26 6.56 -107 -37 -23 -47 tion of hydrogen bonds to gel strength. Moreover, the
antiparallel association of 21 helical chains is the arrangement
a µ , µ in deg; ∆Z, ∆X in Å. b In kcal/mol (E , total energy; E , Lennard
A B T lj
Jones energy contribution; Ehb, hydrogen bond energy contribution; Eel,
found in the solid state. The shape of this structure is close
electrostatic energy contribution). to that of the popular “egg box model”. This model, as
commonly represented, was evaluated by searching for the
Table 2. Calculated Interchaina and Energyb Parameters for the
Best Chain-pairing Situations of Calcium Pectate Oligomer (DP best way to connect two antiparallel chains having the
12) in Both Parallel (|) and Antiparallel (anti-|) Arrangements and glycosidic oxygen atoms face to face (no shift of one chain
for the Egg Box Model with respect to the other). The results are given in Table 1
parameters and the van der Waals representation of the corresponding
µA µB ∆Z ∆X ET Elj Ehb Eel structure is given in Figure 6a. Both energetic and structural
21 considerations indicate that the guluronate “egg box” struc-
| 100 50 0.75 8.07 -117 -32 -37 -48 ture corresponds to a favorable association. It provides a
anti-| 130 50 1.25 5.83 -166 -38 -55 -73 compact dimer, with cavities well adapted to calcium ions
egg box 140 60 2.42 6.33 -90 -19 -14 -57 (Figure 6a), resulting in an efficient junction zone. Neverthe-
31 less, the model is wrong in the description of calcium
| 70 120 1.57 7.03 -89 -19 -22 -48 coordination: only four ligands from the chains, O2 and O6
anti-| 70 110 9.22 6.57 -102 -29 -22 -51
of both chains and not 10 oxygen atoms, bind the calcium
a µ , µ in deg; ∆Z, ∆X in Å. b In kcal/mol (E , total energy; E , Lennard
A B T lj ion. The other postulated ligands, O3, O4, and O5, are at
Jones energy contribution; Ehb, hydrogen bond energy contribution; Eel,
electrostatic energy contribution).
distances longer than 3 Å. The involvement of these atoms
in the coordination sphere of calcium ions would lead to an
crystalline structures.16 It has been applied to the neutral inter-penetration of the two chains. In fact, as described by
polysaccharides amylose,17 chitin, agarose,18 and cellulose19 Smidrød and Draget,21 “the egg box model may be regarded
for which it gave significant insights about the stability of as principally correct, giving an intuitive understanding of
crystal structures and polymorphic transitions. In the present the characteristic chelate-type of the ion-binding properties
study, the procedure was applied to uronate chains to predict of alginates”.
the strongest chain-Ca2+-chain interactions. This was Pectate Chain Pairings. The results reported in Table 2
accomplished by modifying the method to include electro- indicate that, irrespective of the helical form, the antiparallel
static contributions to the interaction energy in the calcula- arrangement was the most favorable one. It allowed for an
tions. efficient connection of the chains, as illustrated by the shorter
Starting from the stable ordered forms of polygalacturonate interhelical distance ∆X, and thus optimized the van der
(21 and 31) and polyguluronate (21), all possible ways to form Waals contacts. The optimum antiparallel structures for both
Ca2+-bridged dimers were computed. The geometric and the 21 and 31 helical forms are illustrated in Figure 5, parts
energetic parameters of the best associations obtained for a and b. In these associations, and in most of the other
the parallel and antiparallel arrangement of the different favorable ones, the two chains are related by a 2-fold
helical forms are reported in Tables 1 and 2 for guluronate symmetry operation (µA ) -µB + 180). The association of
and galacturonate oligomers, respectively. Data for the egg antiparallel pectate chains with the 2-fold (21) helical
box model, as commonly described, are also reported. conformation provided the best chain-pairing configurations.
Guluronate Chain Pairings. The results reported in Table One chain pairing (out of about 12 600) appeared to be
1 indicate that both parallel and antiparallel arrangements particularly favorable (Figure 5a) as it corresponded to the
of the chains provided efficient associations of guluronate best configuration for all the individual energy contributions
chains. The specific interactions between the chains may be (Table 2). In particular this structure, which looks like its
very different as illustrated by the best structures of each guluronate homolog, exhibited an efficient interchain hy-
type shown in Figure 4, parts a and b. These represent the drogen bond network with two periodic and reciprocal
two general kinds of favorable association of 21 helical hydrogen bondssO6‚‚‚O3, 2.78 Å, and O5‚‚‚O3, 2.67 Ås
chains, and are symbolized “|” and “\” (by reference to the that alternate with the calcium interaction sites. The opti-
shape of the structures as viewed from the top). The energetic mization of calcium ion positions with respect to the dimer
contributions were different in these two cases. The parallel using GRID provided a chelation site, represented by two
chain pairing provided a good electrostatic contribution with symmetrical positions in the cavities between the chains
calcium ions coordinated by five uronate oxygen atoms at (Figure 5a, van der Waals representation); wherein the
distances ranging from 2.24 to 2.51 Å (Figure 4a). The calcium ions have three oxygen uronate ligands with Ca‚‚‚
antiparallel chain pairing gave rise to an efficient hydrogen O distances lying between 2.43 and 2.82 Å (Figure 5a). Four
Ca2+-Induced Gelation Biomacromolecules, Vol. 2, No. 4, 2001 1093

Figure 4. Representations (stick and van der Waals structures) of the best [chain-Ca2+-chain] associations of 2-fold guluronate chains: (a)
parallel arrangement; (b) antiparallel arrangement. Positions of calcium ions have been reoptimized with respect to the dimer structures. Dark
circles represent calcium ions. Key: (s) calcium coordination; (- - -) hydrogen bonds.

Figure 5. Representations (stick and van der Waals structures) of the best [chain-Ca2+-chain] associations of antiparallel galacturonate
chains: (a) 31and (b) 21 helical conformations. Positions of calcium ions have been reoptimized with respect to the dimer structures. Dark
circles represent calcium ions. Key: (s) calcium coordination; (- - -) hydrogen bonds.

water oxygen atoms may complete the calcium coordination. Table 2 with the van der Waals representation of the structure
Once again, this best antiparallel association is similar to illustrated in Figure 6b. The energetics data indicate that this
the “egg box structure”. This model, described by Rees and model did not allow for an efficient association. It led to a
co-workers, has been evaluated and the data are reported in weak chain-chain interaction with only a few contacts per
1094 Biomacromolecules, Vol. 2, No. 4, 2001
Figure 6. van der Waals representation of the calculated egg box
models for (a) guluronate and (b) galacturonate chains, respectively.
Dark circles represent the calcium ions.
repeat unit. More importantly, the main argument for this
model is not valid in the case of the pectate system: the
cavities formed by the joined chains do not have the proper
size for calcium ions; they are far too large (Figure 6b). The
present calculations clearly demonstrate that the “egg box
model” cannot correspond to the pairwise associations of
pectate chains that form the junction zones in calcium pectate
gels. Solid calcium pectate is known to exhibit a 3-fold
helical conformation and due to poor X-ray data, a specula-
tive model derived from that of pectic acid has been proposed
by Walkinshaw and Arnott.22 The proposed model was found
among the favorable associations of 31 pectate chains (E )
-87.3 kcal/mol, data not shown) and the calcium coordina-
tion was in agreement with that predicted by GRID.
Nevertheless, the most efficient way to associate two 31
chains was roughly side-by-side, as illustrated by the best
antiparallel association represented in Figure 5b. This
structure exhibits an alternating network of hydrogen bonds
and chelation sites but exhibits the unique feature of two
consecutive similar chelation sites per repeat (Figure 5b). A
comparison between the energetic data of the two possible
helical conformations (21 and 31) shows that pairwise
association of galacturonate chains is far less energetically
favorable in the 3-fold screw symmetry than in the 2-fold
case. This result supports the 21 form as the conformation
of the galacturonate chain segments involved in the junction
zones of pectate gel. In this particular case, the stable ordered
conformation would not correspond to the solid state
conformation. This is not necessarily surprising if we
consider that, as suggested by numerous experimental results,
gelation is effectively initiated by a dynamic process of
dimerization in which no packing constraints may exist.
Braccini and Pérez
Discussion and Conclusions
The “egg box model” is commonly used to describe the
junction zones of alginate and pectate gels but its application
to pectate gels has never been reported by any direct
structural information. The present modeling study indicates
that despite the structural analogy between polyguluronate
and polygalacturonate chains, the alginate system cannot be
directly transposed to the pectate system as demonstrated
by the large difference in the shape of the corresponding
egg box models represented in Figure 6, parts a and b. Even
if this model is conceptually valid for alginates, it cannot
represent a pectate junction zone.
The present results help clarifying the body of experi-
mental data gathered on the significance of calcium binding
to polyguluronate and polygalacturonate chains. Equilibrium
dialysis studies5 of the stoichiometry of calcium binding to
alginate chains revealed that the level of bound calcium could
be consistent with pairs of chains having 21 symmetry. As
syneresis was observed with increasing calcium concentra-
tion, aggregation of the preformed dimers was postulated to
occur when enough calcium is available. The same conclu-
sions were drawn for pectins. Using methods such as
measurement of activity coefficients and transport param-
eters, calcium ions were shown to dimerize pectate chains
even at very low polymer concentrations and appreciable
aggregation was observed.15,23-25 The cooperative effect in
calcium binding was also observed for both oligouronates26,27
with a threshold of DP ∼ 15-20 for strong interaction.3 Such
cooperativity was correlated with a conformational transi-
tion25,28 and could be explained by the presence of an array
of specific binding sites regularly distributed along the chain
which presents an ordered conformation.29 In the solid state,
polyguluronic acid adopts a 21 conformation in the acidic30
as well as in the salt forms.31 Circular dichroism (CD)
experiments indicated no conformational changes between
the gel and the solid states.32 Furthermore, inspection of
molecular models of the dimer structure indicated that the
corrugated chains provide cavities between them having the
proper size and environment for binding calcium ions. This
information led to a model for the junction zones, popularly
known as the “egg box model”,33 in which pairs of 21 helical
chains are packed with the calcium ions located between
them (Figure 2). By analogy, considering that polyguluronate
and polygalacturonate are nearly mirror images in their
structure (with the exception at C3) (Figure 1) and in their
dilute gel CD spectra, the same model was proposed for the
junction zones in pectate gels.34
For pectate gels, the situation was even more complicated
as the conformation of the pectate chains could not definitely
established. In the solid state, pectic acid adopts a right-
handed 3-fold helical conformation in both H+ and Na+
forms.35 The same conformation is maintained in oriented
fibers made from highly concentrated calcium pectate gel,22
and thus a model based on 31 helical chains for pectate
junction zones was proposed. At the same time, CD
experiments showed that a major conformational change
takes place when the gels are dried to solid films.36 On the
basis of the similarities between galacturonate and gulur-
onate, the transition was suggested to be a result of the
Ca2+-Induced Gelation
Figure 7. Representation of the four most favorable associations (1 to 4) of the best dimer structure for (a) guluronate chains and (b) galacturonate
chains.
conversion from the 21 to the 31 helical form, and, as a result,
the “egg box model” was postulated for Ca-pectate gels.
EXAFS studies37 confirmed the transition between gel and
solid states. Jarvis and Apperley38 reported the occurrence
of both 31 and 21 forms, along with intermediate forms
between them, in Ca-pectate gels using solid state 13C NMR
spectroscopy. Furthermore, in dilute solutions, a conforma-
tional change was also observed upon neutralization of pectic
acid,39,40 consistent with a 31 (H+) to a 21 (Na+ or Ca2+ form)
transition.6 Recently, high-resolution NMR data of dilute
sodium pectate were best fitted with a 2-fold conformation.41
All these experimental data pointed toward a 21 conformation
of pectate chains in the junction zones of dilute gels.
While providing a detailed description of the molecular
basis of Ca2+ interacting with polygalacturonate and poly-
glucuronate chains, the present work establishes the validity
of the egg box model for alginate and assesses new structural
features for pectate gels. By analogy, the most favorable
antiparallel associations of galacturonate and guluronate
chains obtained may, at best, be considered as “shifted egg
boxes”. In the guluronate system, the shift is small (1.1 Å)
and has weak effects on the structure. It simply allows for
establishing more interchain hydrogen bonds, but at the
expense of electrostatic interactions. In the galacturonate
system, starting from the “egg box”, the shift of one chain
with respect to the other is more pronounced (1.7 Å) and
has three important consequences: (1) It leads to an efficient
association with numerous van der Waals contacts (the chains
fit into each other). (2) It reduces the original large cavity
and provides two symmetrical sub-cavities of appropriate size
for binding a calcium ion (see Figure 5a, van der Waals
representation). (3) It creates an efficient periodic intermo-
lecular hydrogen bonding network. This last result suggests
that hydrogen bonds have a noticeable contribution to pectate
gel strength, as it is the case for alginate gels. A comparison
between the interaction energy of the best dimers for
galacturonate and guluronate chains indicates that the as-
sociations of the former are stronger. This result suggests
that the critical length of galacturonate residue sequences in
pectin to form a stable junction zone is shorter than that of
guluronate residue sequences in alginate. This is in agreement
with experimental data. For example, the ratio of calculated
interaction energies (-122 to -166, see Tables 1 and 2)
closely corresponds to the observed ratio of the critical DP
(∼15-20) for cooperative binding.3 The computed favorable
associations exhibit intermolecular hydrogen bonds, which
Biomacromolecules, Vol. 2, No. 4, 2001 1095
involve hydroxyl groups at C2 and/or C3 at the inside faces
of the dimers. Any substitution at these sites is expected to
prevent an efficient packing of the chains and to decrease
the gelling ability, as is typically observed with acetylation
in sugar beet pectins.42 It has been reported that some degree
of methyl esterification can be accommodated within calcium
pectate junction zones.43 The structure in Figure 5a (top view)
shows that methyl ester groups may be incorporated without
steric hindrance at any C6 by only slight modification of
the ester group’s torsion angles. On the other hand, the
presence of these groups weakens the calcium-pectate
interaction and might hamper subsequent dimer-dimer
interaction.
The dimers are thought to be strongly associated structures
that govern gelation and if sufficient calcium is available,
subsequent dimer-dimer aggregation may occur. Several
authors observed that these subsequent associations are easily
disrupted by competing ions while the initially formed,
2-chain dimers are not. Therefore, the one issue that remains
is why the association between dimers is weaker and more
tenuous than that between participating chains within dimers.
With respect to this question, we have studied dimer-dimer
interaction using the best chain pairings obtained for galac-
turonate (best antiparallel dimer) and guluronate (best
antiparallel dimer and “egg box” structure) chains. It was
found that the favorable dimer-dimer associations exhibited
a different behavior from that of the chain-chain associa-
tions. Chain-chain associations were strongly specific while
dimer-dimer associations were not. If there are roughly two
favorable types of chain-chain association, symbolized as
“|” and “\” and illustrated in the guluronate case (Figure
4a,b, top views), favorable dimer-dimer associations pre-
sented many different geometries (examples given in Figure
7, parts a and b). In these associations, the energetic
contributions exhibit significant variations from one structure
to another. In most of the energetically favored systems,
electrostatic interactions dominated the energy partitioning
with little or no contribution arising from hydrogen bonding
(Table 3). Both the number of carboxylate groups (twice
that of an isolated chain) and the geometry of the dimers
allow for numerous arrangements for efficient COO--
Ca2+--OOC electrostatic interactions.
The theoretical results confirm the experimental observa-
tions and corroborate the two-stage process that was pro-
posed23 for Ca-induced gelation: (1) initial dimerization
corresponding to strong associations; (2) subsequent ag-
1096 Biomacromolecules, Vol. 2, No. 4, 2001 Braccini and Pérez

Table 3. Calculated Interchaina and Energyb Parameters for the (12) Goodford, P. J. J. Med. Chem. 1985, 28, 849-857.
Best Associations of Dimers for Both Pectate and Guluronate Best (13) Scaringe, R. P.; Pérez, S. J. Phys. Chem. 1987, 91, 2394-2403.
Dimers (14) Angyal, S. J. AdV. Carbohydr. Chem. Biochem. 1989, 47, 1-43.
(15) Thibault, J. F.; Rinaudo, M. Biopolymers 1985, 24, 2131-2143.
parameters
(16) Pérez, S. In Electron Crystallography of Organic Molecules; Fryer,
µA µB ∆Z ∆X ET Elj Ehb Eel J. R., Dorset, D. L., Eds., 1990; pp 33-53.
pectate 0 0 0.00 11.60 -92 -26 -33 -32 (17) Pérez, S.; Imberty, A.; Scaringe, R. P. In Computer Modeling of
Carbohydrate Molecules; French, A. D., Brady, J. W., Eds., ACS
70 250 7.50 8.24 -76 -34 -14 -27
Symposium Series 430; American Chemical Society: Washington,
150 190 0.50 12.60 -71 -18 -35 -18 DC, 1990; pp 281-299.
180 100 0.00 8.56 -67 -16 0 -51 (18) Kouwijzer, M. L. C. E.; Pérez, S. Biopolymers 1988, 46, 11-29.
guluronate 290 100 1.68 6.80 -78 -27 -4 -47 (19) Vietor, R. J.; Mazeau, K.; Lakin, M.; Pérez, S. Biopolymers 2000,
160 250 3.78 7.73 -73 -39 -24 -10 54, 342-354.
0 0 0.00 13.13 -69 -24 -15 -30 (20) Morris, E. R.; Norton, I. T. In Aggregation Processes in Solution;
100 310 1.26 6.71 -67 -14 0 -53 Wyn Jones, E., Gormaly, J.; Elsevier Science Publishing Co.:
Amsterdam, 1983; Chapter 19, 549-593.
a µ , µ in deg; ∆Z, ∆X in Å. b In kcal/mol (E , total energy; E , Lennard
A B T lj (21) Smidsrød, O.; Draget, K. I. Carbohydr. Eur. 1996, 14, 6-13.
Jones energy contribution; Ehb, hydrogen bond energy contribution; Eel, (22) Walkinshaw, M. D.; Arnott, S. J. Mol. Biol. 1981, 153, 1075-1085.
electrostatic energy contribution). (23) Rees, D. A. Carbohydr. Polym. 1982, 2, 254-263.
(24) Thibault, J. F.; Rinaudo, M. Biopolymers 1986, 25, 455-468.
gregation of these dimers. The most convincing argument (25) Malovı́ková, A.; Rinaudo, M.; Milas, M. Biopolymers 1994, 34,
for the preferential formation of dimers came from competi- 1059-1064.
tive inhibition studies where the addition of polygalactur- (26) Garnier, C.; Axelos, M. A. V.; Thibault, J. F. Food Hydrocolloids
1991, 5, 105-108.
onate23,43 and polyguluronate44 chain segments dramatically
(27) Braudo, E. E.; Soshinsky, A. A.; Yuryev, V. P.; Tostoguzov, V. B.
reduced the observed gel strength. Chain-chain interactions Carbohydr. Polym. 1992, 18, 165-169.
lead to strong dimer associations with important contributions (28) Lips, A.; Clark, A. H.; Cutler, N.; Durand, D. Food Hydrocolloids
from the van der Waals and hydrogen bonding interactions 1991, 5, 87-99.
(29) Rees, D. A.; Welsh, E. J. Angew. Chem., Int. Ed. Engl. 1977, 16,
and for which calcium ions have specific positions in well- 214-224.
adapted cavities. The subsequent association of dimers (30) Atkins, E. D. T.; Mackie, W.; Parker, K. D.; Smolko, E. E. Polym.
involves weaker associations that display no particular Lett. 1971, 9, 311-316.
specificity and seem to be mainly governed by electrostatic (31) Mackie, W. Biochem. J. 1971, 125, 89P.
(32) Bryce, T. A.; McKinnon, A. A.; Morris, E. R.; Rees, D. A.; Thom,
interactions. D. Faraday Discuss. Chem. Soc. 1974, 57, 221-229.
(33) Grant, G. T.; Morris, E. R.; Rees, D. A.; Smith, P. J. C.; Thom, D.
Acknowledgment. Financial support form Hercules In- FEBS Lett. 1973, 32, 195-198.
corporated is acknowledged. The authors thank Dr. Robert (34) Rees, D. A. Pure Appl. Chem. 1981, 33, 1-14.
P. Grasso (Hercules Research Center, Wilmington, DE) for (35) Walkinshaw, M. D.; Arnott, S. J. Mol. Biol. 1981, 153, 1055-1073.
(36) Morris, E. R.; Powell, D. A.; Gidley, M. J.; Rees, D. A. J. Mol.
his help and comments. Biol. 1982, 155, 507-516.
(37) Alagna, L.; Prosperi, T.; Tomlinson, A. A. G.; Rizzo, R. J. Phys.
References and Notes Chem. 1986, 90, 6853-6857.
(1) Thakur, B. R.; Singh, R. K.; Handa, A. K. Crit. ReV. Food Sci. Nutr. (38) Jarvis, M. C.; Apperley, D. C. Carbohydr. Res. 1995, 275, 131-
1997, 37, 47-73. 145.
(2) Challenge in Alginate Research. Carbohydr. Eur. 1996, 14, 5-26. (39) Cesaro, A.; Ciana, A.; Delben, F.; Manzini, G.; Paoletti, S. Biopoly-
(3) Kohn, R. Symp. Carbohydr. Chem., BratislaVa 1974, 371-397. mers 1982, 21, 431-449.
(4) Smidsrød, O.; Haug, A. Acta Chem. Scand. 1972, 26, 79-88. (40) Paoletti, S.; Cesaro, A.; Delben, F.; Ciana, A. In Chemistry and
(5) Morris, E. R.; Rees, D. A.; Thom, D.; Boyd, J. Carbohydr. Res. 1978, Function of Pectins; Fischman, M. L., Jen, J. J., Eds.; ACS
66, 145-154. Symposium Series 310; American Chemical Society: Washington,
(6) Ravanat, G.; Rinaudo, M. Biopolymers 1980, 19, 2209-2222. DC, 1986; pp 73-87.
(7) Grant, G. T.; Morris, E. R.; Rees, D. A.; Smith, P. J. C.; Thom, D. (41) Catoire, L.; Derouet, C.; Redon, A. M.; Goldberg, R.; Hervé du
FEBS Lett. 1973, 32, 195-198. Penhoat, C. Carbohydr. Res. 1997, 300, 19-29.
(8) Dheu-Andries, M. L.; Pérez, S. Carbohydr. Res. 1983, 124, 324- (42) Rombouts, F. M.; Thibault, J. F. Carbohydr. Res. 1986, 154, 177-
332. 187.
(9) Mackie, W.; Pérez, S.; Rizzo, R.; Taravel F.; Vignon, M. Int. J. Biol. (43) Powell, D. A.; Morris, E. R.; Gidley, M. J.; Rees, D. A. J. Mol.
Macromol. 1983, 5, 329-341. Biol. 1982, 155, 517-531.
(10) Braccini, I.; Grasso, R. P.; Pérez, S. Carbohydr. Res. 1999, 317, 119- (44) Morris, E. R.; Rees, D. A.; Robinson, G.; Young, G. A. J. Mol. Biol.
130. 1980, 138, 363-374.
(11) Engelsen, S. B.; Cros, S.; Mackie, W.; Pérez, S. Biopolymers 1996,
39, 417-433. BM010008G