X-ray diffraction analysis of tendon collagen at ambient and cryogenic temperatures: role of hydration

Biological ~olecules ELSEVIER

International Journal of Biological Macromolecules 20 (1997) 23-33

~ R t J C ' I X ~ WONCTIONAND ~

X-ray diffraction analysis of tendon collagen at ambient and cryogenic temperatures: role of hydration Roger I. Price a.,, Sidney Lees

b,

Daniel A. Kirschner 1, c

Department of Medical Technology and Physics, Sir Charles Gairdner Hospital, Verdun St., Nedlands, Western Australia 6009, Australia b Department of Bioengineering, Forsyth Dental Center, 140 Fenway, Boston, MA 02115, USA Department of Neurology, Children "s Hospital and Harvard Medical School, Longwood Avenue, Boston, MA 02115, USA

Received 19 February 1996; received in revised form 10 December 1996; accepted 23 December 1996

Abstract

Equatorial (d) and meridional (D) spacings of native rat tail tendon (RTT) and unmineralized native turkey leg tendon (UTLT) were measured at ambient and liquid-nitrogen temperatures, using X-ray diffraction. Cooling of air-dried RTT or UTLT caused little change in d, which was approximately equal (1.1 nm) in the two tissues before and after cooling. For fully hydrated RTT or UTLT, cooling caused the familiar broad equatorial diffraction pattern to increase in sharpness to more resemble the pattern seen in dehydrated tissue. The d-spacings of hydrated RTT and UTLT fell by 0.12 nm (8.5%) and 0.19 nm (13%), respectively, to values seen (at ambient temperature) when the tissue water content is the maximum possible in the absence of unbound water (0.5g water/g dry collagen). These results can be explained by the movement of water within the fibril. In tissue with a water content of greater than 0.5 g/g dry collagen the spacings reflecting the lateral packing of the axially-linked tropocollagen molecules comprising a collagen fibril are determined partly by the unbound component of intermolecular hydration. As the bulk water between the fibrils freezes, this mobile component remains initially unfrozen as the tissue is cooled below 0°C. It diffuses from the intermolecular spaces into the interfibrillar spaces where it also rapidly freezes. This allows the d-spacing to decrease to the value appropriate for the presence of bound intermolecular water only. The mechanism is likely to be an energetically favourable relaxation of the lateral positions of the tropocollagen molecules from a quasi-hexagonal arrangement to hard-disk liquid-like packing. The results and methods of this study may be applicable in the elaboration of more complex collagenous systems. © 1997 Elsevier Science B.V. Keywords: Collagen; Hydration; Low-temperature; X-ray diffraction

* Corrresponding author. Tel.: + 61 9 3462866; fax: + 61 9 3463466; e-mail: [email protected] Present address: Department of Biology, Boston College, 140 Commonwealth Avenue, Chestnut Hill, MA 02167-3811, USA. 0141-8130/97/$17.00 © 1997 Elsevier Science B.V. All rights reserved. PII S0141-81 30(97301 148-3

24

R.I. Price et al. ,/International Journal o[' Biological Macromolecules 20 (1997) 23 33

1. Introduction

Bone fracture risk is attributable to force and frequency of impact, abundance of mineral, and 'structural' (or 'quality') related factors independent of mineral abundance [1]. While many structural investigations have examined the relationship between the defining variables of trabecular networks, e.g. trabecular number, separation and thickness, there is theoretical and experimental evidence that bone quality may depend also on 'ultrastructural' factors operating at the nanometre level. For example, the mechanical properties of collagen (the main component of bone matrix) depend on its interaction with water. This is to be expected, given its fibrous nature and the abundance of hydrophilic groups within its structure [2]. Katsura [3] has proposed that the early stages of biomineralization are governed by the size of spaces between tropocollagen molecules and between collagen fibrils, together with the relative abundances of bound and unbound water within these spaces. The fibrils, which are comprised of numerous parallel chains of longitudinally-linked tropocollagen units, are the principal structural components of connective tissues. They are approximately cylindrical in shape with diameters typically in the range 20500 nm [4]. Ferris et al. [5] have suggested that the abundance of unbound water in bone is associated with increased fragility. Unbound water is comprised of water molecules that have sufficient mobility to form clusters large enough to behave as ice and to undergo a phase transition near 0°C [6]. The role, structure, abundance or mobility of water in collagenous tissue has been studied using techniques including nuclear magnetic resonance [7-9], calorimetry [10-12], dielectric relaxation [13], dynamical mechanical measurements [6], neutron diffraction [14,15] and X-ray diffraction [16 18]. There is agreement that at least two moieties of bound water are associated with the tropocollagen molecule, depending on the degree of tissue hydration. The most tightly bound fraction helps to link the three helical strands comprising tropocollagen by creating a 'bridge' of two water molecules for every three amino acid

residues. Analysis of the hydration structure of a collagen-like peptide [18] suggests that these repetitive water bridges link oxygen atoms within a peptide chain and between peptide chains. Hydroxyprolyl groups have a crucial role in the buttressing of the triple helix by water. Hydrogenbonded intrahelical water corresponds to a hydration of about 0.1 g/g dry collagen [6,13,14]. With increasing water content a less tightly bound fraction appears on the surface of tropocollagen, consisting of chains of water molecules linked by hydrogen bonding and lying along the fibril axis. Some of these water molecules are bound to the tropocollagen by hydrogen bonding [13]. Such water is hindered from forming ice upon cooling because insufficient intermolecular space is available to form a three-dimensional lattice. Therefore intermolecu[ar bound water maintains liquid-like properties with cooling until approximately - 1 0 0 ° C , by which temperature it has become immobilised in a vitreous state [11]. Though reported categories and concentrations of bound water vary somewhat according to the particular physical effect used to identify them, there is broad evidence [6,7,11-13] that water content in excess of about 0.5 g/g dry collagen in tendon is sufficiently mobile that it can diffuse out of the intermolecular spaces and exhibit the properties of bulk water, including formation of ice crystals. X-ray diffraction (XRD) measurements on unmineralized, well-oriented collagen fibrils yield one or more equatorial (lateral) diffraction spacings (d, perpendicular to the axial or fibre stretch direction) and the meridional spacing (D, parallel to the axial or stretch direction) [19]. The sharply defined D-spacing arises from the precise registration of adjacent tropocollagen molecular chains, which are staggered axially. This leads to distinct and highly regular 'gap' and 'overlap' regions along the macromolecular assembly. This successful model was first proposed by Hodge and Petruska [20], and refined subsequently to the amino-acid level by Miller [21]. The most prominent d-spacing reflects an average lateral intermolecular separation within the fibril, and is derived from a broad equatorial peak consisting of diffuse scattering superimposed on sharp Bragg

R.I. PHce et al. ,International Journal of Biological Macromolecules 20 (1997) 23 33

reflections [22]. The equatorial X-ray pattern of tendon collagen is less well understood than the multiple-order meridional peaks representing the 'D' axial macroperiod. A number of models have been proposed to explain the lateral arrangement of collagen, including quasi-hexagonal packing schemes [23,24]. It is generally accepted that the equatorial Bragg reflections are consistent with a three-dimensional triclinic unit cell derived from quasi-hexagonal crystalline molecular packing in the lateral plane [25,26]. However, these conclusions have been determined mainly from the Xray scattering patterns of rat tail tendon. Hydrated unmineralized turkey leg tendon exhibits broad diffuse maxima without sharp Bragg peaks, indicating a less crystalline (long-range) lateral order than seen in rat tail tendon [17]. This has been interpreted as fluctuations in the Bragg spacings in the tendon and has been described in terms of a liquid-like short-range order in lateral packing. Such a model was proposed by Woodhead-Galloway and Machin [27] and studied theoretically by Fratzl et al. [17]. Recently Hulmes et al. [28] have combined quasi-hexagonal longrange ordering and liquid-like short-range ordering in an energy-minimised lateral packing model of a cylindrical fibril. In their scheme radially oriented crystalline domains are separated by liquid-like disordered regions in a weighted combination determined by the axial position of the lateral plane under analysis. There is strong agreement between the calculated equatorial diffraction pattern of this model and the experimental results for rat tail tendon. It is reasonable to assume that intermolecular (mobile) unbound water contributes to the broadening of the equatorial peak in hydrated tissue. At ambient temperature there may also be a contribution to lateral disorder (and broadening of the diffraction pattern) from thermal motion of the tropocollagen molecules. Comparison of the diffraction peaks at ambient and liquid-nitrogen (cryogenic) temperatures should enable estimation of the sensitivity of d and D to thermally-induced disorder, and to any mobilisation of unbound intermolecular water as the tissue is cooled. Using XRD in this study, measurements were made of meridional and equatorial diffraction maxima in

25

rat tail tendon and unmineralized turkey leg tendon at ambient temperature (22°C) and at cryogenic temperatures (approximately - 193°C). Tissues were measured in their air-dried dehydrated and fully hydrated states.

2. Materials and methods

Strands of native rat tail tendon (RTT), approximately 0.9 mm in diameter were obtained from freshly sacrificed animals. Strips of unmineralized turkey leg tendon (UTLT) of the same diameter were dissected from frozen commercially-obtained material. Dehydrated specimens were prepared by drying in air for 5 h at 37°C while slightly stretched. Hydrated specimens were prepared by placing in physiological saline (0.85°/,, w/v, buffered to pH 7.4) for 3 h, immediately after dissection. For the ambient-temperature experiments, decrimped, i.e. lightly-stretched [29], specimens were tied at both ends with cotton threads, drawn into 1 mm internal-diameter thinwalled glass capillary tubes (Charles Supper, Natick, MA, USA) and sealed with wax. Hydrated specimens were sealed so that their flanking cotton threads were in contact with saline. Specimens for cryogenic-temperature experiments were stretched lightly and placed in a groove across a 3 mm diameter hole in a 3 mm thick copper block, on which was placed another grooved 3 mm thick copper block with a conical coaxial hole, so that the specimen was surrounded predominantly by metal. The X-ray beam path was coaxial with the holes. There was a direct thermal link between the specimen holder and a small cryostat which was maintained with liquid nitrogen. Hydrated samples for cryogenic experiments were maintained in saline after mounting, adherent fluid was removed by rapid blotting and vigorous shaking, then the sample was frozen rapidly in liquid nitrogen. To prevent formation of ice crystals on the specimen during the exposure to X-rays, the region containing the beam path, from incident-beam port to film, was encased in a chamber fed constantly with dry nitrogen gas at a low level. The cryostat also vented into this chamber. The gas flow was maintained at a constant rate and not directed at the sample.

26

R.I. Price {'l al.

h#l~'rmtli~mal Journal ¢!/Biolot,,ical tld~nmT¢deculc~

X-ray diffraction patterns were obtained using nickel-filtered, double-mirror focussed CuK;~ radiation from a rotating anode generator operated at 45 kV and 35 mA (Elliott GX-20: GEC Avionics, Herts, UK). The specimen-to-film distance was approximately 71 ram. images were recorded on flat direct-exposure X-ray film (Eastman Kodak, Rochester, NY), the beam diameter on the film was 0.2 ram, and the final (0.5 mm square) guard slit lay 3 mm anterior to the specimen. Exposure time varied from 2 to 25 h. A heliumfilled tunnel with aluminium entrance and Mylar exit windows was placed between the specimen and film holder to reduce X-ray scatter. The effective specimen-to-film distance was calibrated for each experiment by very lightly dusting a fine powder of crystalline alumina on the specimen (cryogenic temperature) or on the anterior and posterior faces of the specimen capillary tube (ambient temperature). Two prominent lines of alumina, 0.349 and 0.245 nm, provided calibration rings. The water contents of the wet cooled samples were not measured before and after each experiment: however repeated cryogenic measurements over the relevant range of exposure times (2 12 h) did not show a dependence on exposure time. Indeed the coefficients of variation in cryogenic d-values obtained from replicate samples of either hydrated RTT or hydrated U T L T were less than 2% of the mean. Therefore the precision with which the cryogenic wet-collagen d-values were able to be measured, and their significant difference from the equivalent dry-collagen d-values (see Section 3), are indications that dehydration of cooled wet samples was not significant. The positions of paired diffraction pattern features (symmetrical about the beam stop) were determined using two separate methods. Method I employed a CCD camera (Cohu, Model 4015, San Diego) and a frame grabber (EPIX, Silicon Video, Chicago). The diffraction pattern on film was captured and represented by a two-dimensional array of pixel values (712 pixels/line: 240 lines/ frame). An engraved glass millimetre scale was used for calibration. The positions of paired diffraction maxima were determined from the coordinates (in two dimensions) of the two pixel groups representing the highest optical densities.

21) (1097~ 23 .;.;

Method II used a ,Ioyce Loebl (JL) 3CS scanning microdensitometer. Analogue optical density curxes were gene.'ated by raster scans performed on each film. along the meridional and equatorial axes, using an acceptance aperture of 491 × 37 /,m, and a pantographic expansion factor of 10. Repeated measurements on six separate equatorial patterns revealed a reproducibility error of less than 0.5%. Results obtained using Methods 1 and 11 did not differ significantly for any equatorial diffraction peak, but since Method 11 yielded a consistently smaller standard deviation and resolved the meridional diffraction peaks more ell fectively, it is the JL data which are presented in this study. For each set of films representing a particular combination of tissue type, temperature and hydration, all visible meridional diffraction orders not affected by the central scatter region were included in the calculation of D. The diffraction patterns were analysed using a linear regression of ,;./(2 sin 0) on 1/N, where ),=0.154 nm, O is the Bragg angle calculated from measurements of the X-ray film and N is the expected order of the diffraction peak. The diameters of the calibration diffraction rings were obtained from multiple measurements using precision digital needlepoint callipers. Reproducibility error for calibration was less than 0.1%. Identical results were obtained using Jk microdensitometry.

3. Results

At ambient temperature, the diffraction patterns of hydrated R T T and U T L T were of the characteristic 'dumbbell' shape, with the central region in the equatorial directions obscured by scattered radiation, as shown in Fig. l(a) and (c). In this study it was not possible to distinguish 'equatorial' and 'near-equatorial' peaks, as reported by others for R T T (Ref. [29], Table 2). The patterns for hydrated tissues became more narrowly defined in the equatorial direction at cryogenic temperature, as shown in Fig. l(b) and (d). In contrast, disorder in the equatorial direction was much less evident in dehydrated tissues at ambient temperature, as shown in Fig. 2(a) and

R.I. Price et a l . / International Journal of Biological Macromolecules 20 (1997) 23 33

Ambient Temp

Cryogenic Temp

(a)

(b)

27

Hyd RTT

UTLT (c)

/Ill (d)

Fig. 1. X-ray diffractionpatterns for fully-hydrated(Hyd) native rat tail tendon (RTT) and unmineralizedturkey leg tendon (UTLT) at ambient and cryogenictemperatures. The alumina calibration rings corresponding to 0.349 and 0.245 nm lie well outside these patterns and are therefore not shown. (c). A reduction in temperature (and therefore thermally-induced disorder) produced no significant increase in the sharpness of the equatorial diffraction pattern in the dehydrated case, as shown in Fig. 2(b) and (d). Equatorial (d) diffraction spacings for fully-hydrated and air-dehydrated tissues, at ambient and cryogenic temperatures, are shown in Table 1, with representative reported ambient-temperature values. The most significant change with cooling occurred in the principal d-spacings of hydrated RTT and UTLT, which decreased by 0.12 nm (8.5%) and 0.19 nm (13%), respectively. For hydrated RTT at ambient temperature, reflections corresponding to 1.83 and 3.84 nm were observed also, though not consistently in all experiments. These spacings (which are in accord with values reported by Brodsky et al. (1.89 nm, 3.8 nm; Ref. [29] Table 2), were not seen at cryogenic temperature. No peak corresponding specifically to the

1.28 nm peak of hydrated RTT at ambient temperature was seen at cryogenic temperature. Cooling of dehydrated RTT or UTLT produced no change in their d-spacings, except a marginal (P < 0.02) increase in the principal d-spacing of UTLT. The d-spacings to which hydrated RTT and UTLT relax at cryogenic temperatures are approximately equal (1.26 and 1.28 nm; Table 1) and quite distinct from the value for dehydrated RTT or UTLT tissue at either ambient or cryogenic temperatures (approximately 1.1 nm). It is useful to compare the cryogenic d-values with those corresponding to a water content of 0.51 g/g dry collagen at ambient temperature. This is the 'critical' hydration above which unbound (freezable) water begins to accumulate [6,7,11-13] in tendon collagen and is the mean of the values determined from magnetic resonance [7], calorimetric [10] and mechanical [6] measurements

R.I. Pri~ u cl ~t/.

Intcrnali~,na/ ,Imu'na/ q/ Bio/~.,.:t~a/ ~/u~r,m~du~tdc~

20 t l')U't ~.~ .;.~

Ambient Temp

Cryogenic Temp

(a)

(b)

Dehyd RTT

~,i%,~!i!ii!i17I¸¸!

Dehyd UTLT (c)

....~,~!!!~ii!~

(d)

Fig. 2. X-ray diffraction patterns for air-dried dehydrated (Dehyd) native rat tail tendon (RTT) and umnineralized turkey leg tendon (UTLT) at ambient and cryogenic temperatures.

(range 0.48 0.54). The critical ambient-temperature d-spacings (de) at this hydration for RTT and UTLT may be obtained by interpolating the data of Katz and Li [41] and Eanes et al. [31], respectively, as shown in Fig. 3. For RTT this yields (from quadratic regression of data from Fig. 4(b) of Ref. [41]) d~ = 1.248 rim. This falls within the 95% confidence interval for d of hydrated RTT (1.245 1.265 nm) at cryogenic temperature (Table 1). The data of Eanes et al. were obtained by rehydration of lyophilised specimens. We calculated the 'true' hydration of UTLT as a function of d-spacing from their Table 1 by assuming that, following preparation of lyophilised samples, only 0.1 g water/g dry collagen remained (corresponding to the tightly bound intrahelical water) [6,13,14]. Linear regression of these corrected data yielded d,, = 1.268 nm for UTLT. This value falls within the range of measurements for the d of hydrated UTLT (1.268 1.286 nm) at cryogenic temperature (Table 1).

Meridional (D) spacings for hydrated and dehydrated tissue, at ambient and cryogenic temperatures, are shown in Table 2 with representative (ambient-temperature) published values for comparison. Those multiple diffraction orders (N) of the 'D' macroperiod which were identified clearly, and thus used in the calculation of D, are also shown. The visible orders at ambient temperature are in qualitative agreement with published relative intensities for hydrated native RTT [34,35], dehydrated native RTT [34] and hydrated native UTLT [36]. Except for dehydrated UTLT (which remained unchanged), cooling produced a small (2.1 2.7%) significant decrease in D. The only clearly observed change in the meridional diffraction orders upon cooling occurred in hydrated RTT, where the N = 12 order (visible at ambient temperature) did not appear on the film and the N = 11 order (unseen at ambient temperature) became detectable. Though the short (71 ram)

R.L Price et al./International Journal of Biological Macromolecules 20 (1997) 23 33

29

Table 1 Equatorial diffraction spacings (d, in nm) for fully-hydratedand air-dried dehydrated native rat tail and unmineralized turkey leg tendon (RTT and UTLT, respectively)at ambient (22°C) and cryogenic(approximately -193°C) temperatures Tendon

Hydrated RTT

UTLT Dehydrated RTT UTLT

Ambient temperature

Cryogenic temperature

Published values

This study

This study

1.26 '~ 1.33-1.37 ~ 1.89 a 3.8 a 1.448, 1.47 e, 1.51 c

1.28 (1.27 1.298) r 1.37 (1.367 1.377) 1.838 3.848 1.46 (1.455-1.47) h

1.1 d 1.12 b, 1.08 ~, 1.16e

1.08 (1.076 1.086) 1.11 (1.099-1.113) 4.90 (4.87 4.932)

1.26 (1.245-1.265)***

1.28 (1.268-1.286) h** 1.07 (1.059 1.087) 1.12 (1.114-1.124)* 4.90 (4.68 5.121)

Data expressed as mean (95% confidence interval of the mean). Significance of comparison between ambient and cryogenic values: *P < 0.02; **P < 0.005; ***P < 0.0001 ; two-tailed unpaired t-test. a Ref. [19]. b Ref. [30]. Ref. [31]. d Ref. [32]. e Ref. [33].

t~95%confidenceintervals are calculated from multiple experiments for each tissue/hydration/temperaturecombination. g Not seen consistently in all experiments. h Values in parens are the range. specimen-to-film distance was not suited ideally to determination of multiple-order meridional peaks, reflection N = 20 was always clear. Calculated D-values using this reflection alone were not significantly different from the data of Table 2.

4. Discussion

4.1. Hydrated tissues The reduction in the diffuse scatter between the principal equatorial peaks and the central region on cooling of hydrated R T T or U T L T has the appearance of an increase in the regularity of the lateral packing of the collagen molecules within each fibril [28]. There are two possible reasons for the temperature-induced reduction in the d-spacing in the hydrated case. The effect might arise from thermal contraction, or from loss of water within the fibril, leading to a rearrangement (compaction) of the lateral molecular packing. The

fractional change in a linear dimension of a fixed shape of unbound water, then ice, with cooling from ambient to cryogenic temperature is a 2.2% increase [39,40]. Therefore the reduction in dspacing of 8.5-13% with cooling (Table 1) must arise from a spatial redistribution of the water in the tissue, rather than from a physical change in (otherwise immobile) unbound water. The agreement (for either tissue) between dc the ambient-temperature d-spacing corresponding to a water content of 0.51 g/g dry collagen, and the cryogenic-temperature d-spacing for fully-hydrated tissue (Fig. 3) suggests that unbound water is removed from the intermolecular spaces with cooling, permitting d to decrease to the value (de) appropriate for the presence of inter- and intramolecular bound water only. A suggested mechanism is as follows. As the sample cools, the interfibrillar unbound water behaves as bulk water and freezes close to 0°C. The intermolecular unbound water does not freeze at 0°C as the sample is cooled from ambient temperature. Be-

R.I. Price c/ a/.

30

hller/lalio;la] ,lore'hal o~ Biological Ma~romo/ccu/c,s 20 (1997) 23 33

tween 0°C and minus 30°C the intermolecular u n b o u n d water undergoes freezing-point depression, arising from its storage in small capillaries of n a n o m e t r e dimensions and its inability to assume the tetrahedral clustering characteristic of bulk water [7]. This c o m p o n e n t is identified as a broadening of the water-ice phase transition (i.e. heat capacity) curve, obtained from differential scanning calorimetry, over the range 0°C to minus 30°C in samples where the hydration is above 0.51 g/g dry collagen [12]. This still-liquid water diffuses into the interfibrillar region, because of a reduction in the v a p o u r pressure there, and freezes rapidly. This mechanism has been described by Hoeve et al. [11] who cited X-ray diffraction evidence for the existence of large interfibrillar ice crystals [2]. Such a mechanism can proceed if a portion of the intermolecular water is mobile and the free energy increase arising from 1.4 UTLT

'-: o

o. I ~ _

.~o~

1.3

~

]

A:)

" ° " ~ ~ 'RTT

1.2

2

~-, o" LId

1.1

bound water 1,0

i

i

0

0.1

,

I

f

i

0.2

0.3

0.4

l ~ '

unbound water -ere

0.5

,

i

r

0.6

0.7

,

J

0.8

Water content (g H20/g dry collagen] Fig. 3. Interpolation of published data for the functional relationship between principal d-spacings and degree of hydration in rat tail tendon (RTT) and unmineralized turkey leg tendon (UTLT) at ambient temperature. This enables determination of d~, the value of d for a water content of 0.51 g water/g dry collagen; the maximum hydration in the absence of unbound water. For RTT the interpolated value (d~,= 1.248 nm) was obtained from a quadratic fit to the data of Katz and Li [41] [c)]. For UTLT the data of Eanes et al. [31] [el were corrected prior to linear interpolation (see text), yielding d~ = 1.268 nm. For each of UTLT and RTT, dc derived by interpolation (horizontal arrow on y-axis) was not significantly different from the principal d-spacing obtained in fully hydrated tissue at cryogenic temperature (B and II, for RTT and UTLT respectively; error bars are 95% confidence interval [RTT] and range [UTLT]).

the lateral adjustment of the collagen lattice is less in magnitude than the free energy decrease of ice formation [11]. ll is more likely that ice crystals are sequestered between the fibrils than in "pockets' within the fibril created by lateral distortion of the liquid-like lattice. The effect of this latter mechanism on the diffraction pattern has been considered by Fratzl et al. [17] in connection with the location of mineral crystallites in turkey leg tendon. Creation of large ice crystals within the fibrils of hydrated R T T or U T L T would lead to a broadening and reduction in definition of the equatorial diffraction pattern on cooling. As seen in Fig. 1, this is the opposite of the observed effect. When considering the alteration in lateral packing caused by cooling of hydrated tissue it is probable that the predominantly quasi-hexagonal arrangement prevailing at ambient temperature is lost, with relaxation to a hard-disk liquid-like packing [29]. This would permit tropocollagen molecules to coexist with m i n i m u m lateral intermolecular distances constrained by full occupancy of sites for bound water, at a total water concentration of 0.51 g/g dry collagen. This new packing would produce a m o r e sharply defined (but still basically diffuse) equatorial X-ray scattering pattern. The long-range molecular order implicit in crystalline quasi-hexagonal packing has in fact been replaced by the short-range order of the liquid-like packing. Therefore the apparently increased order implied by the 'sharper' equatorial patterns of Fig. l(b) and (d) must be interpreted with caution. The sharper definition of the equatorial pattern m a y also arise in part from a reduction in thermal broadening with cooling. The data of this study are unable to distinguish this effect from a change in lateral packing.

4.2. Dehydrated tissue Because dehydration was achieved by air-drying at atmospheric pressure, the expected residual hydration was a b o u t 0.1 g water/g dry collagen. This corresponds to the presence of intra-helical water (e.g. Regimes I + II of Pineri et al. [6] ). Reducing hydration below this level does not affect the equatorial spacing [2]. This suggests

R.L Price et al./ International Journal of Biological Macromolecules 20 (1997) 23-33

31

Table 2 Meridional diffraction spacings (D, in nm) for fully-hydrated and air-dried dehydrated native rat tail and unmineralized turkey leg tendon (RTT and UTLT, respectively) at ambient and cryogenic temperatures Tendon

Hydrated RTT

Ambient temperature

Cryogenic temperature

Published values

This study

This study

67.5% 66.8b

67.7 (67.0-68.3) r [9,12,20] 67.1 (66.0-68.2) [9,20]

66.3 (65.8 66.8)** [9,11,20] 65.3 (64.7-65.8)** [9,11,14,20]

65.9 (65.3 66.5) [6,9,11,17,20,27,30] 65.0 (64.6-65.4)

64.3 (63.1-65.5)*

UTLT

67.0c'e

Dehydrated RTT

64.0d

UTLT

64.0~

[6,9,11,14,17,20,27,30]

[6,9,11,20] 64.6 (64.1-65.1) [6,9,11,14,17,20]

Data expressed as mean (95% confidence interval of the mean). Integers in brackets N are the orders (N) of the 'D' macroperiod, sufficiently clearly identified to be used to calculate D. Not all peaks were seen on all films in a replicate set. Significance of comparison between ambient and cryogenic values: *P<0.02; **P<0.005, two-tailed unpaired t-test. a Ref. [19]. b Ref. [34]. c Ref. [37]. d Ref. [38]. Ref. [33]. f95% confidence intervals are calculated from multiple experiments for each tissue/hydration/temperature combination, each yielding multiple orders of D diffraction peaks. e

that the equatorial packing of the collagen molecules has reached a limiting level. The sharpness of the principal equatorial patterns of both tissues at ambient temperature implies that random variations in lateral intermolecular separation are restricted to small amplitudes, even at ambient temperature. This is in accord with the absence of a significant increase in sharpness of the equatorial pattern, or a marked change in equatorial spacings, with cooling. Therefore it is likely that the lateral packing o f dried collagen at either ambient or cryogenic temperatures is best described by the hard-disk liquid model, with the distance of closest approach of the molecules constrained only by the 'hard-disk diameter' of the tropocollagen [17]. As in the case of cooled hydrated tissue the liquid-like lateral packing of dried ambient-temperature or cooled collagen lacks the long-range order of quasi-hexagonally packed hydrated tropocollagen molecules at ambient temperature.

4.3. D-spacings This study was not designed to measure D optimally, given limitations of the optics and the specimen-to-film distance. This restricted the number of orders of the ' D ' macroperiod that could be identified. Since the small (2%) decrease in the D of R T T occurred independently of degree of hydration, it is possible that the effect arose from an alteration of the secondary structure of the collagen molecule. In hydrated R T T at least, the absence of the N = 12 reflection and the appearance of the N = 11 reflection with cooling suggests that the axial molecular packing in hydrated tissue at cryogenic temperature is more similar to that of the ambient-temperature dehydrated case (where N = 11 is dominant over N = 12) than the ambient-temperature hydrated configuration (where N = 12 predominates over N = 11) [34]. A possible explanation is that the electron-density 'contrast' between the 'gap' and 'overlap' regions [20] of the macromolecular as-

32

R.I. Price el al. International Journal o~ Biolo,~ica/ Ila(romo[eculex 20 (19~)7~ 23 .?3

sembly is diminished by cooling. At ambient temperature the contrast is significant for wet collagen because of the different water contents of these regions [17,37], but the removal of intermolecular unbound water with cooling may cause the gap and overlap regional water contents to become comparable. In turn this may cause a slight decrease in D (Table 2). Dehydration of wet collagen at ambient temperature also causes a decrease in D, possibly for a similar reason. Tendon is comprised almost entirely of Type I collagen, which is also a major component of bone matrix and virtually all other connective tissues [4]. Therefore the results and methods of this study may be of use in elaborating on the abundance, location and structural role of water in more complex (including mineralized) collagenous systems.

Acknowledgements A major portion of this research was carried out in the Mental Retardation Research Center, Children's Hospital, Boston, supported by NIH Core Grant HD 18655. We are grateful for the technical advice and assistance of Richard Alquist (Charles Dana Research Center, Northeastern University, Boston), Elizabeth Page (Forsyth Dental Center, Boston), and Peter Henson (Royal Perth Hospital, Perth, Australia). One of us (RP) acknowledges the receipt of an Australasian College of Physical Scientists and Engineers in Medicine Travelling Scholarship. The comments of a reviewer regarding possible changes in collagen packing with cooling are gratefully acknowledged.

References [1] Melton, L.J. 1II and Cummings, S.R. Bone Min. 1987; 2: 321. [2] Nomura, S., Hiltner, A., Lando, J.B. and Baer, E. Biopolymers 1977; 16: 231. [3] Katsura, N. In: Mechanisms and Phylogeny o f Mineralisation in Biological Systems (Suga, S. and Nakahara, H. eds.), Springer, New York, 1991, 193 pp. [4] Hulmes, D.J.S. Essays" Bioehem. 1992; 27: 49.

[5] Ferris. B.D.. Klenerman. I.. [)odds. R.A.. Bitensk 5, I and Chayen. J. Bone 1987: 8: 285. [6] Pineri. M.H.. Escoubes. M. and Roche. G. Biop,/vnwr,~ 197H: 17: 2799. [7] Dehl, R.E. Scicn('c 1970: 170: 73~. [8] Pelo, S., Gillis. P. and Henri, V.P. Biophys J. 1992: 57: 71. [9] Renou. J.P.. Bonnet, M., Bielicki. G., Rochdi, A. and Gatellier. P. Biopolvmers 1994: 34:1615. [10] Mrevlishvili, G.M. and Privalov, P.L. In: Water m Bio/o~ical Systenzs (Kayushin. [_.P. ed.}, Consultants' Bureau, New York. 1969, 63 pp. [I1] Hoe,e, C.A.J. and Tats, A . S . J . Phys. (Tzem. 1978: 82: 1660. [12] Kopp, 3.. Bonnet, M. and Renou, 3.P. Matrix 1989: 9: 443. [13] Shinyoshiki, N., Asaka, N., Mashimo, S., Yagihara, S. and Sasaki, N. Biopo&mers 1990, 29: 1185. [14] MiddendorL H.D., Hayward, R.L., Parker, S.F., Bradshaw. J., and Miller, A. Biophys. J. 1995: 69: 660. [15] Lees, S.L., Bonar, C. and Mook, H.A. Int. J. Biol. Macromol. 1984; 6: 321. [16] Migchelsen, C. and Berendsen, H . J . C . J . Chem. Phys. 1973: 59: 296. [17] Fratzl, P., FratzI-Zelman. N. and Klaushofer, K. Bi{qgtvs. J. 1993; 64: 260. [18] Bella, J., Brodsky, B. and Berman, H.M. Structure 1995: 3: 893. [19] Brodsky, B. and Eikenberry, E.F. In: Structural and Con*ra{'tile Proteins: (Part A) E.vtraeellular MatrLv. MethmLs in Eno,mo[ogy (Cunningham, L.W. and Fredericksen, D.W. eds.), Academic Press, New York, 1982, 127 pp. [20] Hodge, A.J. and Petruska, J.A. In: Aspects of Protein Structure (Ramachandram G.N. ed.), Academic Press, New York, 1963, 289 pp. [21] Miller, A. Phil. Trans. Roy. Soc. set'. B. 1984; 304: 455. [22] Miller. A. and Wray, J.S. Nature 1971; 230: 437. [23] Hulmes, D.J.S. and Miller, A. Nature (Lond.) 1979: 282: 878. [24] Lees, S., Pineri, M. and Escoubes, M. lnl. J. Biol. Macrotool. 1984; 6: 133. [25] Fraser. R.D.B., MacRae, T.P., Miller, A. and Suzuki, E. J. Mol. Biol. 1983: 167: 497. [26] Wess, T.J., Hammersley, A., Wess, L. and Miller, A. ,/. Mol. Biol. 1995: 248: 487. [27] Woodhead-Galloway, J. and Machin, P. Acts Crr,~t. A. 1976: 32: 368. [28] Hulmes, D.J.S., Wess, T.J., Prockop, D.J. and Fratzl. P. Biophys. J. 1995: 68: 1661. [29] Brodsky, B., Eikenberry. E.F., Belbruno, K.C. and Sterling, K. Biopolymers 1982: 21: 935. [30] Eanes. E.D., Lundy, D.R. and Martin, G.N. Calcill Tiss. Re.~. 1970: 6: 239. [31] Eanes, E.D., Martin, G.N. and Lundy, D.R. Calcgl Tis,s. Re,s. 1976; 20:313. [32] Eanes, E.D. and Miller, E.J. Arch. Biochem. Biophys. 1969: 129: 769.

R.1. Price et al./ International Journal of Biological Macromolecules 20 (1997) 23-33

[33] Bigi, A., Ripamonti, A., Koch, M.H.J. and Roveri, N. Int.J. Biol. Macromol. 1988: 10: 282. [34] Eikenberry, E. F. and Brodsky, B. J. Mol. Biol. 1980; 144: 397. [35] Hulmes, D.J.S., Miller, A., White, S.W. and Doyle, B.B. J. Mol. Biol. 1977; 110: 643. [36] Berthet-Colominas, C., Miller, A. and White, S . W . J . Mol. Biol. 1979; 134: 431. [37] White, S.W., Hulmes, D.J.S., Miller, A. and Timmins, P.A. Nature 1977; 266: 421.

33

[38] Ericson, L.G. and Tomlin, S.G. Proc. R. Soc. A 1959; 252: 197. [39] Hobbs, P.V. Ice Physics, Clarendon Press, Oxford, 347, 1974. [40] Weast, R.C. In: Handbook of Chemistry and Physics (Weast, R.C. ed.), CRC Press, Boca Raton, USA, 54th edn., Table F-5, 1974. [41] Katz, E.P. and Li, S.-T. J. Mol. Biol. 1973; 80: 1.