Brought to you by Portland Press Ltd.
Published on behalf of the International Federation for Cell Biology
Cancer Cell death Cell cycle Cytoskeleton Exo/endocytosis Differentiation Division Organelles Signalling Stem cells Trafficking
Cell Biology International (2006) 30, 56–65 (Printed in Great Britain)
Evidence that collagen and tendon have monolayer water coverage in the native state
Gary D. Fullerton* and Maxwell R. Amurao
Radiology Department, University of Texas Health Science Center at San Antonio, San Antonio, Floyd Curl Drive, TX 78229-3900, USA


This paper investigates an alternative explanation for widely reported paradoxical intracellular water properties. The most frequent biological explanation assumes water structure extending multiple layers from surfaces of compactly folded macromolecules to explain large amounts of perturbed water. Long range water structuring, however, contradicts molecular models widely accepted by the scientific majority. This study questions whether the paradoxical cell water could result from larger than expected amounts of first layer interfacial water on internal protein surfaces rather than structured multilayers. Native mammalian tendon is selected for the study because (1) the organ consists of highly compact structures of a single macromolecular protein – collagen, (2) molecular structure and geometry of collagen is well characterized by X-ray diffraction, (3) molecular structure extends to the macroscopic tendon level and (4) perturbed water behavior similar to cellular water is reported on tendon. Native tendon holds 1.6gwater/gdrymass. The 62% native water content simulates the water content of many cell types. MicroCT studies of tendon dilatometry as a function of hydration are measured and correlated to X-ray diffraction measurements of interaxial separation. Correlations show that native tendon has sufficient water for only a single monolayer of interfacial water. Thus the paradoxical properties of water in native tendon are first-layer interfacial water properties. Similar water behavior on globular proteins suggests that paradoxical cell water behavior could be caused by larger than expected amounts of first layer interfacial water on internal and external macromolecular surfaces of cell components.

Keywords: Collagen, Tendon, Cell hydration, Cell water content, Dilatometry, Intracellular water, Paradoxical cell water, Water structuring, Protein hydration.

*Corresponding author.

1 Introduction

Cell biologists have frequently proposed phenomenological models suggesting that water inside cells has properties differing from water in the bulk state. Ling (1984, 2001) and more recently Pollack (2003) have reviewed this literature in 3 comprehensive monographs. Paradoxical observations and alternative models are invoked to explain a wide variety of anomalous osmotic pressure, optical, NMR, absorption isotherm and other observations. The models, however, lead frequently to explanations logically inconsistent with firmly held beliefs accepted by the majority of the scientific community. The explanatory models are rejected by the scientific majority but data continue to accumulate indicating failures in present understanding. The primary biomolecular constituents of the cell are water and proteins. The source of the paradoxical observations must be related to the interactions between these molecules. It is also widely accepted that protein hydration and protein induced water structuring are fundamental sources of protein chemical behavior and collagen is frequently selected as the model protein to elucidate these relationships due to its unique molecular structure (Leikin et al., 1994, 1995, 1997; Leikina et al., 2002; Privalov, 1982). This report investigates the hypothesis that the source of paradox may come from larger than expected amounts of first monolayer interfacial water on internal protein surfaces. Collagen provides the molecular model to test the idea. Molecular predictions are extrapolated to native tendon where the collagen content in some instances approaches 100% of the dry mass to allow macroscopic testing of concepts.

1.1 Stochastic description of collagen

Both chemical and X-ray diffraction analyses of collagen show that the collagen molecule is highly conserved among mammalian species with glycine evenly distributed along the molecule at every third position and with exceptionally high hydroxyproline content. For most purposes stochastic calculations may use the species specific mean amino acid residue mass (MAAmean) for type I collagen as calculated from the chemically determined amino acid sequences using 2 α-1 chains and 1 α-2 chain compositions measured by Fietzek and Rexrodt (1975). The bovine type I collagen mean molecular weight per amino acid residue is 91.2Da. This consistency allows accurate calculation of critical hydration fractions for structural water (h=0.0658gH2O/gcollagen; Note: The hydration of tendon is measured gravimetrically in units of “grams of water/gram dry mass” which are assumed identical with “grams of water/gram collagen” due to the high concentration of collagen in tendon. Throughout the paper hydration is expressed as “g/g” but should be read as described here.), cleft water (h=0.197g/g) and monolayer surface (h=1.315g/g) waters as shown in the companion paper in this volume (Fullerton et al., 2006b). We use these predicted hydration fractions to identify transitions in water properties at critical hydration values hc=0.0658, 0.264 and 1.584g/g. The values are nearly identical for warm blooded mammals which allow intercomparison of data from multiple mammalian species.

1.2 Axial hydration chain calculation

Significant new collagen structural information is available from high resolution X-ray diffraction studies (Bella and Berman, 1996; Bella et al., 1994, 1995, 1996; Fraser et al., 1979; Kramer et al., 2001, 1999, 2000, 1998; Miller and Scheraga, 1976; Okuyama et al., 1977; Ramachandran, 1967; Rich and Crick, 1961; Yonath and Traub, 1969) that localize water molecules in a cylindrical hydrogen bonded network or layer surrounding each molecule. These observations are consistent with the Berendsen (1962) linear water chain hypotheses summarized in Fig. 1. The Berendsen model implies that a chain of water in each groove would require a water content of 0.262 g/g. Measurements on fully hydrated bovine tendon in the native state show 61.8% water content or hydration 1.62g/g (Fullerton et al., 1985). As shown in Fig. 2 fully hydrated native tendon would thus require 1.62/0.263=6.2 chains of water per groove or slightly more than 18 chains in the first monolayer to completely hydrate the cylindrical surface of the molecule. Hydration of 18 axial water chains would be adequate to form a first monolayer network over large portions of native tendon while maintaining the minimum spacing necessary to accommodate hexagonal chains possessing geometries for full 4-hydrogen bonds per molecule as found in bulk water.

Fig. 1

The Berendsen (1962) water chain hypothesis assumed that the water network conforms to ice-like spacing adjusted to room temperature 22°C as shown at the left. He then showed that such a chain matches perfectly with the spacing of hydrogen bonding sites on the collagen main chain with a spacing of 0.47nm between bonds or dc=0.235nm per water molecule chain link. This model was adapted by Lim (1981) to show (drawing on the right adapted from Lim) that there are 4 water (18Da) molecules per every 3 protein residues (mean 91.2Da) or 0.262g/g for a single chain of water in each groove or cleft of the collagen triple helix. The hydration fraction for a single chain of water on the collagen molecule is thus 0.262/3=0.0877.

Fig. 2

Fully hydrated tendon has 1.62g/g which implies 6 chains of water per cleft as shown in this cross-sectional cartoon of a collagen fibril consisting of 7 associated tropocollagen molecules. This requires 3×6=18 water molecules in a circumferential water chain around each tropocollagen molecule.

1.3 Circumferential hydration chain calculation

The calculation of monolayer surface hydration can be approached in a completely independent manner starting from measurements of axial or equatorial spacing as a function of hydration. Three authors have reported equatorial (Bragg) spacing for dry collagen from kangaroo tail (Fraser and MacRae, 1973; Rougvie and Bear, 1953, see p. 354) and rat tail tendon (Nomura et al., 1977) with consistent results 10.5±0.1Å. Assuming quasihexagonal molecular spacing and the Bragg separation one can calculate the dry interaxial separation or dry molecular diameter of collagen 12.0Å (Kuznetsova et al., 1997; Leikin et al., 1994). High resolution X-ray diffraction measurements (Berisio et al., 2002) on collagen analog molecules grown on the space station allow calculation of the separation of water molecules from atoms on the collagen surface. There are two characteristic values shown in Fig. 3. The mean separation of water on polar surfaces is 2.7Å while separation on hydrophobic surfaces increases to 3.6Å with nearly equal amount of both on the molecular surface. Division of the resulting water chain circumference by dc=0.235nm gives 20.2 water molecules in the circumferential chain. This agrees well with 18.2 circumferential water molecules calculated from the axial method with the assumption of native monolayer water coverage. The monolayer hydrated diameter=12.0+2.7+3.6=18.3Å is only slightly larger than 17Å separation reported by Leikin and his colleagues for collagen fully equilibrated in various water solutions (Kuznetsova et al., 1997; Leikin et al., 1994, 1995, 1997).

Fig. 3

The synchrotron X-ray diffraction measurements of water separation from the collagen surface yield first shell (polar hydration) and second shell (hydrophobic hydration) water separation from the protein surface (Berisio et al., 2002). There are nearly equal amounts of polar and hydrophobic surface which allow estimation of the mean hydrated diameter, 18.3Å. The circumferential water chain thus has a diameter of 15.1Å. This yields an estimate of 20.2 water molecules in the circumferential water chain in good agreement with the axial calculation method.

1.4 Predictions for tendon dilatometry

On the basis of the axial and circumferential calculations of monolayer hydration on the collagen molecule we reasoned that collagen rich tendons should dilate in the radial dimension as a function of hydration such that the ratio of tendon diameters [d(h)/do where d(h)=tendon diameter at hydration h and do=dry diameter] equals the ratio of interaxial separations s(h)/so measured by X-ray diffraction. Values of the interaxial separations as a function of hydration are available from the literature (Katz and Li, 1973; Nomura et al., 1977; Rougvie and Bear, 1953; Sasaki et al., 1983) while tendon dilation can be measured directly in 3 dimensions on large bovine tendons using microCT methods newly available in our laboratory.

2 Methods and materials

2.1 Tendon samples

Fresh bovine hind limbs were obtained from a local slaughterhouse. The common extensor, superficial flexor, and deep flexor tendons were excised in the laboratory and washed with normal saline (0.9% sodium chloride injection USP, Baxter Healthcare Corp., Illinois). Tendon sections approximately 5cm long were extracted from the midsection of each tendon. The tendon size was chosen to optimize measurement accuracy for the microCT field of view (see below). Four tendon samples of each type were initially obtained, from which 3 common extensor tendons, 2 superficial flexor tendons and 2 deep flexor tendons were selected for the experiment. Each sample was placed separately in a thin walled glass sample tube to minimize physical perturbation of tendon dimensions from factors other than hydration h.

2.2 Dilatometry measurements

MicroCT imaging (X-SPECT, Gamma-Medica, Northridge, Calif.) was performed on each tendon immediately after preparation, and then at 24-h intervals to measure the cross-sectional area and mean diameter at each position as shown in Fig. 4. The sample tubes were stored in a vacuum chamber at 22.5°C and at pressure 0.3Torr between measurements. Samples were removed periodically from the chamber for repeated CT geometric measurements at room temperature and atmospheric pressure. The mass and shape of the tendon initially decreased rapidly in a fashion dependent on the initial diameter but all asymptotically approached an equilibrium value although in different periods as shown in Table 1. After at least 5 consecutive days at equilibrium mass in 22.5°C and 0.3Torr, the temperature in the vacuum chamber was raised to 90°C and the measurement procedure continued until the mass approached a final equilibrium value. Prior experiments have shown that all structural waters are removed by this procedure to yield a direct measurement of the dry protein mass. The sample tube's plastic caps were attached every time the tube was taken out of the vacuum oven for imaging and mass measurement to minimize sample re-hydration from atmospheric humidity.

Fig. 4

Deep flexor tendon diameter versus longitudinal position at different hydrations h. For decreasing hydration levels approaching 0.26, the sample diameter of the end proximal to the tube opening decreased faster than the diameter on the closed end. At hydration levels below 0.26, sample diameter decreased uniformly across the length of the sample. These results are representative of all tendon samples (3 common extensor, 2 deep flexor and 2 superficial flexor tendons).

Table 1.

This table summarizes temperature and pressure conditions used for dehydration of the 7 tendon samples

TendonDayTemperature (°C)Pressure (Torr)
3 Common extensor1–1322.50.3
2 Deep flexor1–1622.50.3
2 Superficial flexor1–2522.50.3

2.3 Hydration

Sample tendon hydration (h) was calculated from the sample mass Ms measured at the time of each CT experiment and its measured dry mass Mp. That is,Mp was measured following equilibration of tendon mass within the ±0.0016g balance measurement error for at least 5 consecutive days, with the sample stored in a vacuum oven at 0.3Torr and 90°C between measurements.

2.4 CT measurement of tendon diameter

Imaging-based tendon cross-sectional areas were measured from the reconstructed microCT image data using the Analyze program (Mayo, Rochester, Minn.). Cross-sectional area was measured at 5mm intervals along the longitudinal axis of the tendon. The average cross-sectional-area spanning the length of the phantom (Amean) was calculated for &007E;10 area measurements (number of measurements depending on the sample length). The mean tendon diameter d(h)=(4Amean/Π)0.5 was calculated for the hypothetical average circle with the mean area for each value of hydration h. The geometric accuracy of CT measurements of diameter was determined by comparison of CT measures with calibrated caliper measurements on acrylic cylinders encompassing the range of tendon diameters between 10 and 15 mm. Accuracy and precision are approximately equal ±0.02mm and image-based diameter measurement accuracy and precision are better than 2 parts per thousand.

3 Results

3.1 Tendon diameter

Typical diameter measurements as a function of hydration are shown in Fig. 4 for one of the 7 samples. All samples showed similar systematic decrease in diameter as the hydration level was decreased from near native hydration h=1.6 to completely dry h=0.0g/g. Inhomogeneous dehydration is visible in the diameter measurement as a function of longitudinal position with fastest drying near the opening of the glass tube. The use of the mean diameter calculation method compensates for this inhomogeneity as shown in Fig. 5 which summarizes the data for all 7 tendon samples.

Fig. 5

The relative mean diameter or diameter expansion factor (d/do) as a function of hydration for 3 common extensor tendons, 2 deep flexor tendons and 2 superficial flexor tendons are plotted together to show that factors such as inhomogeneous hydration, initial diameter and uneven dehydration rates do not effect the hydration induced expansion of the diameter. The linear regression equation (solid line) for the combined data d/do=0.383h+1.005 for n=122, r2=0.993 fits equally as well as the nonlinear regression (dashed line) model for continuous fluid hydration of a cylindrical molecule d/do=(0.960 h+1)0.5 for n=122, r2=0.993. This supports the contention that native hydration can be described as monolayer hydration. Longitudinal hydration effects are neglected in both model calculations.

4 Discussion

4.1 Transition between cleft and monolayer surface waters

Measurement of the transition between cleft water and monolayer first layer water was not part of the experimental design. Larger than expected hydration inhomogeneity along the tendon as shown in Fig. 4 demonstrates that the extent of cleft water can be measured from these data. There are highly significant slopes of the measured diameter as a function of longitudinal position until h=0.26g/g where the slope is 0. The inhomogeneity was observed on all 7 samples. The distribution results from the short hydration equilibration times in our protocol combined with the smaller free energy and faster diffusion of monolayer surface waters relative to the immobile cleft and structural water fractions. The transition is more accurately measured by NMR titration (Fullerton et al., 2006b) and will not be evaluated here further. As discussed below by comparison with interaxial measurement the issue of long equilibration periods at laboratory temperature is important but results for the present experiments are greatly simplified by using a mean diameter measurement insensitive to such effects.

4.2 Mathematical models of diameter expansion

We consider 2 mathematical models of mean diameter expansion with the assumption of negligible change in tendon length. The most general model of radial expansion describes accumulation of a continuous fluid of unit density on the surface of a cylinder of fixed density and length with the equation d/do=(k1h+1)0.5 to fit the data with k1=0.960 for n=122, r2=0.993 while the molecular or linear model shown in Fig. 5 assumes partial filling of a single water monolayer to predict the linear equation d/do=k2h+k3. The linear model fits native tendon hydration data equally well with k2=0.3834 and k3=1.005 for n=122, r2=0.993 and thereby supports the hypothesis of monolayer hydration on native tendon. A number of water molecules in the circumferential chain can be calculated by the molecular model nf=2dw/(0.0877k2do) as shown in Fig. 6. The measurements predict a full monolayer of 15.4 parallel water chains around the circumference using dw=3.17Å calculated from Berisio's data but this increases to 20.5 using the zig-zag dw=2.37Å from Berendsen.

Fig. 6

Simple conceptual derivation of the monolayer dependence of the average hydrated diameter of the collagen molecule with water giving a linear dependence of the diameter on the hydration. Using the collagen single water chain hydration ratio hsc=0.263/3=0.08772g/g derived in Fig. 2 and the slope calculated in Fig. 5 the number is nf=5.986/0.3836=15.4 circumferential water molecules for bovine tendon using dw=3.15Å but increases to 20.2 using Berendsen's dc=2.37Å.

4.3 Correlation of tendon expansion with molecular expansion measurements

Fig. 7 compares the calculated linear tendon expansion regression line for d/do as a function of h from Fig. 5 to X-ray diffractions measurement of relative Equatorial (Bragg) Bragg spacing s/so as a function of h measured by a number of laboratories (Katz and Li, 1973; Leikin et al., 1997; Rougvie and Bear, 1953; Sasaki et al., 1983). The plot shows general agreement of the four data sets such that the average slope does not differ sufficiently to change the prediction of monolayer coverage from the last section. On the other hand the very good data set from Sasaki and colleagues (1983) shows a different initial slope and a change slope near h=0.8 that is interpreted by the authors as an equilibration separation with increasing hydration that requires “hidden water” as it was named by Rougvie (Rougvie and Bear, 1953). Can we describe these observations in terms of monolayer hydration? First we must define first monolayer.

Fig. 7

A plot of the ratio of equatorial or Bragg spacing (s/so) for wet collagen relative to dry collagen as a function of hydration shows general agreement for data on collagen from bovine Achilles tendon (BAT), Kangaroo tail tendon, rat tail tendon and bovine skin as well as agreement with the relative tendon diameter ratio (d/do) as predicted. There is, however, significant disagreement in the detailed relationship that is most evident with data from Sasaki. This discrepancy is thought due to comparison of “inhomogeneous” tendon diameter results with equilibrated samples as shown below.

4.4 Primary and secondary hydration components of the first monolayer

As demonstrated in the studies of Berisio et al. (2002) the first monolayer of water on collagen consists of nearly half primary hydration adjacent to hydrophilic surfaces and half secondary hydration adjacent to hydrophobic surfaces. The proposed model proposes a total monolayer collagen hydration h=1.584g/g of which half is associated with polar surfaces hp=0.792g/g and half with hydrophobic or nonpolar surfaces hn=0.792g/g. The mean separation for polar water according to Berisio's measurement (see Fig. 2) is 2.7Å while the separation for nonpolar water is 3.6Å or 0.9Å surface displacement. The function of water as the dielectric to reduce the electrostatic energy dictates that the polar portions of the surface will be hydrated first followed by bridge hydration over the hydrophobic regions when h>0.792g/g. Thus the surface hydration will resemble a pebbled surface as shown in Fig. 8.

Fig. 8

This drawing demonstrates the larger impact of primary hydration on collagen separation caused by hydrophilic sites relative to secondary hydration on hydrophobic regions. As the hydrophilic sites are preferentially filled first there should be a change in slope as shown in Fig. 7 for measurements on BAT. The use of inhomogeneous dehydration removing both polar and nonpolar waters and measuring the global mean diameter measurements would mask this effect as shown for the data in Fig. 5.

4.5 Reevaluation of BAT spacing

Fig. 9 presents a reevaluation of the separation expansion ratio for bovine Achilles tendon from the measurements of Sasaki et al. (1983). The division of the hydration into primary and secondary hydrations successfully describes the observations with the equations d/do=1.010+0.4747h for primary hydration (h0.792) and d/do=1.3860+0.08183 (h−0.792) for secondary hydration (h>0.792). For the primary hydration region we know that do=12Å and d/do=1.010+0.4747×0.792=1.3860 when the primary hydration is completely filled which implies that d=16.63Å. Thus the effective diameter of the water molecule is dw=(16.63Å−12.0Å)/2=2.315Å and neighboring molecules are separated by twice that 4.63Å which agrees well with the water bridge length 4.7Å calculated by Berendsen (1962) using the crystalline water chain bond length described in Fig. 2. Substituting this value in the expression nf=2dw/(slope dohsc) with the primary hydration slope=0.4747 gives nf=9.26 for the number of polar water molecules on the surface. The total circumferential water count is twice this or 18.5 water molecules when nonpolar water molecules are counted as well. This agrees with the original model circumferential calculation 18.2 total water molecules. The slope of the secondary hydration region allows calculation of the increase in collagen hydrated diameter due to waters on the hydrophobic surface 2Δdw=0.08183×0.792=0.0648Å. This increase is much smaller than anticipated from the measurements of Berisio. The results, however, clarify why authors coined the term “hidden waters”; waters bridging over the hydrophobic surfaces fit between the “polar bound waters” with little increase in interaxial spacing. This effect was masked in the tendon expansion experiments reported in the present study due to inhomogeneous or mixed removal of both primary and secondary hydrations from one end of the sample.

Fig. 9

Reevaluation of the data of Sasaki et al. (1983) using the interpretation methods developed in this paper based on monolayer molecular hydration of the collagen molecule. The regression fit to the bi-modal model has goodness of fit r2=0.992 for 29 values and 26 degrees of freedom.

4.6 Dependence of surface hydration on the collagen coiling

A third independent calculation method based on axial water chains for various types of native and synthetic collagen analogues is shown in Table 2 using 5 different X-ray diffraction models for synthetic collagen molecules (Miller and Scheraga, 1976; Okuyama et al., 1977) as well as three different type I collagens from a range of mammals (Ramachandran, 1967; Rich and Crick, 1961; Yonath and Traub, 1969). This table summarizes a range of collagen models (Fraser et al., 1979) with pitch from 60.2 to 104.8Å. The axial hydration mass ratio for a single water chain is 0.241±0.0044g/gfor all 5 models. Correction for a helical water chain in the groove of the collagen increases this value to approximately 0.26g/g as calculated previously. The calculation emphasizes the point that first layer hydration remains equivalent for different collagen coiling relationships. The separation of hydrogen bonding sites on the collagen depends only on the separation of bonding sites on the protein main chain. This implies that hydration characteristics could be similar for other proteins as they also have similar geometric separations along the protein main chain.

Table 2.

Calculation of water chain to protein mass fraction for as single water chain per molecular cleft in various collagen models

Model selection, Fraser et al. (1979)Residues/turnPitch (Å)Residue length (Å)α-Chain (amu/Å)Water chain (amu/Å)Water chain mass fraction (g/g)
Okuyama et al. (1977)2160.20.34831.817.590.238
Miller and Scheraga (1976)24.573.20.33430.527.590.248
Rich and Crick (1961)3085.80.34931.887.590.238
Yonath and Traub (1969)3086.10.34831.777.590.239
Ramachandran (1967)36104.80.34331.327.590.242

4.7 Collagen hydration force experiments

The hydration force experiments of Leikin and colleagues (1994) on rat tail collagen support the conclusion of monolayer hydration as shown in Fig. 4. These experiments use immersion of collagen in solvents of varying concentrations of a non-penetrating osmolyte with measurement of interaxial molecule separation using X-ray diffractions. The osmotic pressure is applied with solutions of polyethylene glycol and NaCl salts to remove water from the surface of the collagen and thereby force the collagen molecules closer to one another. The change in osmotic force per length of the collagen molecule versus the interaxial separation is plotted in Fig. 4 and shows changes slope occurs abruptly at s=16.8Å. This is equivalent to s/so=16.8Å/12=1.4, which is slightly above the primary hydration expansion limit s/so=1.38 where the slope changes in Fig. 9 for bovine Achilles tendon. Leikin reports the hydration h=1.1g/g but the slope is so small in the secondary hydration range that osmolyte concentration and equilibration time become confounding factors in directly measuring monolayer hydration. Leikin et al. (1995) demonstrated attractive contribution to the hydration force for s/so>1.4 that is temperature dependent. The present study shows that the attractive region is associated with secondary hydration over the hydrophobic surfaces. Thus attractive forces are likely due to hydrophobic bonding while repulsive forces are due to electrostatic interactions related to charges on the protein surface that dominate the primary hydration region (Fig. 10).

Fig. 10

Hydration force measurements on rat tail tendon (Leikin et al., 1994) using 8000 molecular weight polyethylene glycol (PEG) to apply osmotic compression reveal that hydration force is the sum of repulsive and attractive forces that come to equilibrium at 16.8Å in agreement with d=do+2×dc=12+2×2.35=16.7 as sufficient for monolayer water coverage of primary hydrophilic sites; neighboring molecules are connected by short double water bridges. It is important to note that salt contributes to the osmotic pressure with an effective pressure due to only partial exclusion of salt as shown on the right. At physiologic levels 0.2M NaCl the equilibrium separation is nearly constant at monolayer separation over a wide range of lower PEG applied forces.

4.8 Monolayer hydration of native collagen

The measured and calculated monolayer hydration values derived in this study are summarized in Table 3 using the single water chain hydration fraction hf=0.263×nf derived from the molecular model of collagen hydration (Fullerton et al., 2006b). The calculated monolayer values agree with direct gravimetric measures of native hydration. Mechanical stress and/or elevated osmotic pressure easily reduce native hydration as much as half the native 1.63g/g by removing water over hydrophobic patches. Native tendon will, however, recover to the equilibrium monolayer coverage following removal of the stress with sufficient time (hours to tens of hours) to re-equilibrate by re-hydrating the surfaces. Removal of the primary hydration requires more elevated stress or pressure as shown by Leikin and colleagues.

Table 3.

Summary of calculated and measured values of monolayer collagen hydration for mammals (see text for details of methods)

MethodData sourceCircumferential chains – nfNative hydration – hf (g/g)
Axial water chains in 4-bond geometryBerendsen, 1962 and Fullerton et al., 1985181.58
Circumferential chains, X-ray diffractionBerendsen (1962)Nomura et al. (1977)Rougvie and Bear (1953)20.21.77
Native gravimetric, assume monolayerFullerton et al. (1985)1.62
Expansion slope, MicroCT dilatometryPresent study15.4–20.41.35–1.79
Bragg spacing X-raySasaki et al. (1983)18.51.62
Mean18.61.62 ± 0.16 (SD)

4.9 Calculation of monolayer hydration from solvent accessible surface (SAS) area

The theoretical maximum SAS for collagen can be calculated from summations of the tabulations of solvent accessible surface areas of individual amino acid residues calculated by Miller et al. (1987) using the rolling ball method. Miller et al. calculated the SAS contribution for each amino acid residue side chain using tripeptides Gly-X-Gly in an extended configuration to calculate the polar and nonpolar surface areas of the X-residue in a protein chain by subtraction of the area calculated separately for Gly-Gly-Gly. The maximum possible SAS for the protein is Smax=iniAi where i is the index number 1 through 22 of a specific amino acid residue, ni=the number of the residue in collagen and Ai=the residue specific SAS (separate values for polar, hydrophobic and total area for each residue). The maximum possible SAS for collagen was calculated in Table 4 using the chemical composition and the chemically determined amino acid content of the α-1 and α-2 type I bovine collagen (Ayad et al., 1998) with the assumption of three polar sites on each residue main chain and the remainder hydrophobic main chain area. The native protein SAS is reduced from the maximum SAS by folding and twisting of the protein chains to reduce the free energy.

Table 4.

Calculation of collagen solvent accessible surface (SAS) areas and relation to monolayer hydration fractions for polar, nonpolar and total coverage

SurfaceSASmax2)hmax (g/g)hdis (g/g)hnat (g/g)
Polar205,2181.060.26 (17%)0.794 (49%)
Nonpolar402,1972.071.25 (83%)0.826 (51%)
Total607,4163.131.51 (100%)1.62 (100%)

The calculated SAS areas are converted to mass of “unfolded” first monolayer water using the mean surface area occupied by a water molecule. The projected surface area of a water molecule Aw=πrw2. This area is increased by a packing factor assuming hexagonal close packing to give a total effective area . Using the mean water occupancy diameter 3.15Å measured by Berisio gives Aew=8.592. Thus the maximum or extended protein chain hydration fractions for collagen are calculated with the equation hmax=(18×SAS/Aew)/(MWcoll) as summarized in Table 4 for polar, nonpolar and total SAS values on the collagen molecule.

Collagen models from X-ray diffraction studies show polar water displaced by one direct bond per tripeptide (2 water molecules displaced), one structural water bridge (one water molecule displaced) and one weakly bonded water between negative to negative sites (one water molecule displaced) or a total of 4 water molecules displaced from polar surfaces for every tripeptide (Fullerton et al., 2006b; Ramachandran and Ramakrishnan, 1976). Thus the polar hydration displaced is hpdis=4×18/4×91.2=0.2631g/g. The measured monolayer or native total hydration value hTnat=1.62 in Table 4 is the mean value calculated in Table 3. Using these values as a starting point the hydration distribution of first monolayer primary and secondary water on native tendon can be calculated.

The total water displaced hTdis due to folding is calculated as hTdis=hTmaxhTnat=3.13−1.62=1.51g/g. The amount of water displaced from nonpolar surfaces is calculated hndis=hTdishpdis=1.25 (83% of the displaced water). Only 17% of the displaced water is from polar surfaces. These values allow calculation of the amount water on the native or folded protein for both the polar hpnat=hpmaxhpdis=0.794 (49% of native surface) and nonpolar fractions hnnat=hnmaxhndis=0.826 (51%). While the accuracy of these calculations is limited by the lack of accounting for post-synthesis changes expected in native tendon, such changes are sufficiently small to offer useful information. Firstly, the division of native hydration into 49% polar and 51% hydrophobic hydration on collagen agrees with the 50/50 ratio measured by Berisio on an analog collagen crystal (Berisio et al., 2002). Secondly, the calculated polar hydration 0.794g/g agrees well with the limit of primary hydration 0.792g/g demonstrated from the X-ray scatter experiments of Sasaki shown in Fig. 9. Additionally, one notes the large fraction of hydrophobic surface hidden from water exposure (83% of displaced monolayer water) by folding and winding about neighboring protein chains relative to only 17% polar water displaced. Such ratios are consistent with predictions of the importance of hydrophobic bonding for protein folding (Kauzmann, 1959).

4.10 Paradoxical properties of first monolayer water

A wide variety of experiments on tendon and collagen confirm paradoxical water properties using optical (Leikin et al., 1997; Susi et al., 1971), isothermic (Luescher et al., 1974), calorimetric (Luescher et al., 1974; Zhang et al., 1985), NMR orientational (Fullerton et al., 1985; Henkelman et al., 1994; Krasnosselskaia et al., 2005), and NMR titration (Fullerton et al., 2006b). Comparisons show water on tendon has paradoxical properties similar to those reported for cellular water. In addition hydration force experiments show that collagen interfacial water has reduced free energy relative to bulk water but can be removed by applying osmotic stress or mechanical stress (Kuznetsova and Leikin, 1999; Kuznetsova et al., 1997; Leikin and Parsegian, 1994; Leikin et al., 1993, 1997). In related articles in this volume we show that globular proteins and cells have similar osmotic properties (Cameron et al., 2006; Fullerton et al., 2006a). We conclude that native tendon has paradoxical properties of water similar to cellular water but in the case of tendon the perturbation occurs at the molecular interface in the first water monolayer on the collagen molecule. This suggests that the paradoxical properties of cellular water could also be due to first monolayer interfacial relationships rather than extended water structuring as proposed by some investigators.

5 Conclusions

From microCT measurements of tendon diameter as a function of tendon hydration and comparison to studies of collagen molecular structure we conclude that native tendon has near monolayer water coverage of 1.62g/g (water 62% of native tendon mass). Concurrence between diameter expansion ratio as a function of hydration and the molecular collagen separation as a function of hydration demonstrate monolayer coverage. Relation of monolayer hydration measurements to solvent accessible surface (SAS) area calculations shows agreement that allows calculation of monolayer water hydration fractions. The monolayer water layer consists of primary hydration over the hydrophilic surfaces 0.794g/g (49% of native water) and a nearly equal amount of water over hydrophobic surfaces 0.83g/g (51% of native water). A large fraction of the hydrophobic surface 1.25/2.07=60% is buried by hydrophobic bonding induced folding and winding of the collagen molecule to remove exposure to water while only 0.26/1.06=25% of the hydrophilic surface is dehydrated by electrostatic interactions through hydrogen bonds. Comparisons with molecular separation experiments as a function of hydration show that secondary hydration 0.826g/g is removed during an initial dehydration phase while much larger forces are necessary to remove primary hydration 0.794g/g. A small subset of primary hydration 0.26g/g consists of structural (single water bridge=Ramachandran water bridge) and cleft (double water bridges=Berendsen water chain) water bound directly to the main chain of the protein. All the remaining polar surface water is bound to the surface of side chains. The sum of the two polar main chain subsets is frequently called bound water in the protein hydration literature. The monolayer hydration characteristics of collagen are likely the general properties of proteins as they reflect interfacial interactions with charge groups identical to those on every protein differing only in proportions. First layer interfacial water could therefore account for many of the paradoxical properties reported for cellular water.


Funding for much of this work came from discretional funds supplied by Malcolm Jones Distinguished Professorship for which I am grateful to Drs. Reuter, Dodd and my other colleagues from the Radiology Department in San Antonio.


Ayad S, Boot-Handford, RP, Humphries, MJ, Kadler, KE, Shuttleworth, CA. The extracellular matrix facts book. 1998:
1st Citation  

Bella J, Berman, HM. Crystallographic evidence for C alpha-H.OC hydrogen bonds in a collagen triple helix. J Mol Biol 1996:264:4:734-42
Crossref   Medline   1st Citation  

Bella J, Brodsky, B, Berman, HM. Hydration structure of a collagen peptide. Structure 1995:3:9:893-906
Crossref   Medline   1st Citation  

Bella J, Brodsky, B, Berman, HM. Disrupted collagen architecture in the crystal structure of a triple-helical peptide with a Gly→Ala substitution. Connect Tissue Res 1996:35:1–4:401-6
Crossref   Medline   1st Citation  

Bella J, Eaton, M, Brodsky, B, Berman, HM. Crystal and molecular structure of a collagen-like peptide at 1.9 A resolution. Science 1994:266:5182:75-81
Crossref   Medline   1st Citation  

Berendsen HJ. Nuclear magnetic resonance study of collagen hydration. J Chem Phys 1962:16:12:3297-305
1st Citation   2nd   3rd   4th   5th  

Berisio R, Vitagliano, L, Mazzarella, L, Zagari, A. Crystal structure of the collagen triple helix model [(Pro-Pro-Gly)(10)](3). Protein Sci 2002:11:2:262-70
Crossref   Medline   1st Citation   2nd   3rd   4th  

Cameron IL, Kanal, KM, Fullerton, GD. Role of protein conformation in pumping water into and out of a cell. Int J Cell Biol 2006:30:78-85
Crossref   1st Citation  

Fietzek PP, Rexrodt, FW. The covalent structure of collagen. The amino-acid sequence of alpha2-CB4 from calf-skin collagen. Eur J Biochem 1975:59:1:113-8
Crossref   Medline   1st Citation  

Fraser RD, MacRae, TP. . Conformation in Fibrous Proteins 1973:620 p
1st Citation  

Fraser RD, MacRae, TP, Suzuki, E. Chain conformation in the collagen molecule. J Mol Biol 1979:129:3:463-81
Crossref   Medline   1st Citation   2nd   3rd  

Fullerton GD, Cameron, IL, Ord, VA. Orientation of tendons in the magnetic field and its effect on T2 relaxation times. Radiology 1985:155:2:433-5
Medline   1st Citation   2nd   3rd   4th  

Fullerton GD, Kanal, KM, Cameron, IL. Osmotically unresponsive water fraction on proteins and in cells. Int J Cell Biol 2006:30:86-92
Crossref   1st Citation  

Fullerton GD, Kanal, KM, Cameron, IL. . Fullerton GD, Nes E, Amarau M, Rahal A, Krasnosselskaia L, Cameron IL. An NMR method to characterize three water compartments on mammalian collagen. Int J Cell Biol 2006b;30:66–73.
1st Citation   2nd   3rd   4th   5th  

Henkelman RM, Stanisz, GJ, Kim, JK, Bronskill, MJ. Anisotropy of NMR properties of tissues. Magn Reson Med 1994:32:5:592-601
Crossref   Medline   1st Citation  

Katz EP, Li, ST. The intermolecular space of reconstituted collagen fibrils. J Mol Biol 1973:73:351-69
Crossref   Medline   1st Citation   2nd  

Kauzmann W. Some factors in the interpretation of protein denaturation. Advances in protein chemistry 1959:1-63
Medline   1st Citation  

Kramer RZ, Bella, J, Brodsky, B, Berman, HM. The crystal and molecular structure of a collagen-like peptide with a biologically relevant sequence. J Mol Biol 2001:311:1:131-47
Crossref   Medline   1st Citation  

Kramer RZ, Bella, J, Mayville, P, Brodsky, B, Berman, HM. Sequence dependent conformational variations of collagen triple-helical structure. Nat Struct Biol 1999:6:5:454-7
Crossref   Medline   1st Citation  

Kramer RZ, Venugopal, MG, Bella, J, Mayville, P, Brodsky, B, Berman, HM. Staggered molecular packing in crystals of a collagen-like peptide with a single charged pair. J Mol Biol 2000:301:5:1191-205
Crossref   Medline   1st Citation  

Kramer RZ, Vitagliano, L, Bella, J, Berisio, R, Mazzarella, L, Brodsky, B. X-ray crystallographic determination of a collagen-like peptide with the repeating sequence (Pro-Pro-Gly). J Mol Biol 1998:280:4:623-38
Crossref   Medline   1st Citation  

Krasnosselskaia L, Fullerton, GD, Dodd, SJ, Cameron, IL. Water in tendon: orientational analysis of the free induction decay. J Mag Reson Med 2005:
1st Citation  

Kuznetsova N, Leikin, S. Does the triple helical domain of type I collagen encode molecular recognition and fiber assembly while telopeptides serve as catalytic domains? Effect of proteolytic cleavage on fibrillogenesis and on collagen–collagen interaction in fibers. J Biol Chem 1999:274:51:36083-8
Crossref   Medline   1st Citation  

Kuznetsova N, Rau, DC, Parsegian, VA, Leikin, S. Solvent hydrogen-bond network in protein self-assembly: solvation of collagen triple helices in nonaqueous solvents. Biophys J 1997:72:1:353-62
Crossref   Medline   1st Citation   2nd   3rd  

Leikin S, Parsegian, VA. Temperature-induced complementarity as a mechanism for biomolecular assembly. Proteins 1994:19:1:73-6
Crossref   Medline   1st Citation  

Leikin S, Parsegian, VA, Rau, DC, Rand, RP. Hydration forces. Ann Rev Phys Chem 1993:44:369
Crossref   1st Citation  

Leikin S, Parsegian, VA, Yang, W, Walrafen, GE. Raman spectral evidence for hydration forces between collagen triple helices. Proc Natl Acad Sci USA 1997:94:21:11312-7
Crossref   Medline   1st Citation   2nd   3rd   4th   5th  

Leikin S, Rau, DC, Parsegian, VA. Direct measurement of forces between self-assembled proteins: temperature-dependent exponential forces between collagen triple helices. Proc Natl Acad Sci USA 1994:91:1:276-80
Crossref   Medline   1st Citation   2nd   3rd   4th   5th  

Leikin S, Rau, DC, Parsegian, VA. Temperature-favoured assembly of collagen is driven by hydrophilic not hydrophobic interactions. Nat Struct Biol 1995:2:3:205-10
Crossref   Medline   1st Citation   2nd   3rd  

Leikina E, Mertts, MV, Kuznetsova, N, Leikin, S. Type I collagen is thermally unstable at body temperature. Proc Natl Acad Sci USA 2002:99:3:1314-8
Crossref   Medline   1st Citation  

Lim VI. A novel structural model for collagen: water-carbonyl helix. FEBS Lett 1981:132:1-5
Crossref   1st Citation  

Ling GN. In search of the physical basis of life. 1984:
1st Citation  

Ling GN. Life at the cell and below – cell level. 2001:
1st Citation  

Luescher M, Ruegg, M, Schindler, P. Effect of hydration upon the thermal stability of tropocollagen and its dependence on the presence of neutral salts. Biopolymers 1974:13:12:2489-503
Crossref   Medline   1st Citation   2nd  

Miller MH, Scheraga, HA. . J Polymer Sci Symp 1976:54:171-200
1st Citation   2nd   3rd  

Miller S, Janin, J, Lesk, AM, Chothia, C. Interior and surface of monomeric proteins. J Mol Biol 1987:196:641-56
Crossref   Medline   1st Citation  

Nomura S, Hiltner, A, Lando, JB, Baer, E. Interaction of water with native collagen. Biopolymers 1977:16:231-46
Crossref   Medline   1st Citation   2nd   3rd  

Okuyama K, Takayanagi, M, Ashida, T, Kakudo, M. . Polym J 1977:9:341-3
Crossref   1st Citation   2nd   3rd  

Pollack GH. Cells, gels & the engines of life. 2003:
1st Citation  

Privalov PL. Stability of proteins. Proteins which do not present a single cooperative system. Adv Protein Chem 1982:35:1-104
Medline   1st Citation  

Ramachandran GN. . Treatise on collagen 1967:103-83
1st Citation   2nd   3rd  

Ramachandran GN, Ramakrishnan, C. Molecular structure. Biochemistry of collagen 1976:45-84
1st Citation  

Rich A, Crick, F. The molecular struture of collagen. J Mol Biol 1961:3:483
Medline   1st Citation   2nd   3rd  

Rougvie MA, Bear, RS. An X-ray diffraction investigation of swelling by collagen. J Am Leather Chem Assoc 1953:48:735-51
1st Citation   2nd   3rd   4th   5th  

Sasaki N, Shiwa, S, Yagihara, S, Hikichi, K. X-ray diffraction studies on the structure of hydrated collagen. Biopolymers 1983:22:2539-47
Crossref   Medline   1st Citation   2nd   3rd   4th   5th   6th  

Susi H, Ard, JS, Carrol, RJ. The infrared sectrum and water binding of collagen as a function of relative humidity. Biopolymers 1971:10:1597-604
Crossref   Medline   1st Citation  

Yonath A, Traub, W. Polymers of tripeptides as collagen models. IV. Structure analysis of poly(l-proly-glycyl-l-proline). J Mol Biol 1969:43:3:461-77
Crossref   Medline   1st Citation   2nd   3rd  

Zhang J, Zhang, Z, Zhang, W, Fu, Y, Ye, G, Nin, Z. Hydration and thermotransition of collagen fibre. Water and ions in biological systems 1985:197-214
1st Citation  

Received 30 May 2005; accepted 30 September 2005


ISSN Print: 1065-6995
ISSN Electronic: 1095-8355
Published by Portland Press Limited on behalf of the International Federation for Cell Biology (IFCB)