thermally induced irreversible conformational changes in collagen probed by optical second harmonic...

Lasers Med Sci 2002, 17:34–41Ownership and CopyrightSpringer-Verlag London Limited © 2002

Thermally Induced Irreversible ConformationalChanges in Collagen Probed by Optical SecondHarmonic Generation and Laser-inducedFluorescence

T. Theodossiou1, G.S. Rapti1, V. Hovhannisyan2, E. Georgiou1, K. Politopoulos1 and D. Yova1

1Biomedical Optics and Applied Biophysics Laboratory, Department of Electrical and Computer Engineering, Division ofElectromagnetics, Electrooptics and Electronic Materials, National Technical University of Athens, Zografou, Athens, Greece2Laboratory of Biomedical Optics, Applied R&D division, Yerevan Physics Institute, Yerevan, Armenia

Abstract. Irreversible thermal conformational changes induced to collagen have been studied by opticalmethods. More specifically, second harmonic generation (SHG) from incident nanosecond Ng:YAG 1064 nmradiation and laser-induced fluorescence by 337 nm, pulsed nanosecond nitrogen laser excitation, at 405, 410and 415 nm emission wavelengths were registered at eight temperatures (40�, 50�, 55�, 60�, 65�, 70�, 75� and 80�C)and normalised with respect to the corresponding values at the ambient temperature of 30�C. The heatingprotocol used in this work, was selected to monitor only permanent changes reflecting in the optical propertiesof the samples under investigation. In this context, the SHG, directly related to the collagen fibril populationin triple helix conformation, indicated on irreversible phase transition around 64�C. On the other handfluorescence related to the destruction of cross-linked chromophores in collagen, some of which are related tothe triple helix tertiary structure, also indicated a permanent phase transition around 63�C. These results arein agreement with previous results from studies with di#erential scanning calorimetry. However SHG andfluorescence, being non-invasive optical methods are expected to have a significant impact in the fields of laserablative surgery and laser tissue welding.

Keywords: Collagen; Fluorescence; Gelatine; Second harmonic generation; Thermal denaturation

INTRODUCTION

Several recent reports have dealt with opticalharmonic generation in various types of bio-ligical tissue [1–7]. Collagen appears to be thecommon tissue component mainly responsiblefor these second order non-linear opticalphenomena [2,4,5,7]. Collagen comprisingabout 6% of body weight in mammals, qualifiesas the most widespread structural protein inhigher vertebrates and the main means oftheir structural support. Its molecules consistof three amino acid chains convoluted in arod-shaped, triple helix, and this tertiarystructure of collagen is of microcrystallinetriclinic nature, as demonstrated by the Braggreflections in its X-ray di#raction pattern [8].

The fluorescence of collagen in the UV andvisible spectral regions, has been investigatedin many papers [9,10] and several attemptshave been made to clarify its nature and origin[11,12]. The existence of at least four photo-labile fluorescent chromophores found toabsorb optical radiation in the region between300 and 400 nm has also been shown [13]. Inconsequent work [14], several types of chromo-phores absorbing and emitting throughoutthe UVA and visible spectral regions wereobserved. Some collagen chromophores wereisolated and identified as pentosidine and pyri-dinoline [10,12]. Both these chromophores aresuggested to be products of cross-linking,pentosidine being age related. There are alsovarious ‘endogenous’ chromophores in col-lagen found in the amino acid chains liketyrosine (fluorescence maximum at 303 nm)and phenylalanine (fluorescence maximum at285 nm).

Correspondence to: Dr T. Theodossiou, 92 Ymittou st.,15561 Holargos, Athens, Greece. Tel: +301 6520717, e-mail:theo–[email protected]

Helix-coil transition in collagen induced byheating has been thoroughly studied [15–18].More specifically it has been shown with theuse of di#erential scanning calorimetry, thatthe position and shape of the denaturationendotherm of rat tail tendon collagen fibrilsare governed by the kinetics of an irreversibleprocess around 62�C, as shown by measure-ments of change of the specific heat capacity[15], whereas the thermal transitions of type Icollagen were investigated again with theuse of di#erential scanning calorimetry andspectrophotometry of turbidity, indicating anirreversible process centred around 52�C andcorroboated by both methods used [16].

The thermal denaturation process in polarcod skin was investigated with the use ofscanning microcalorimetry and intrinsic spec-trofluorometry and was found to occur in threeindependent stages [17]. The e#ect of thermaldenaturation on water collagen interactionswere studied by use of NMR and di#erentialscanning calorimetry [18] at four temperaturesranging from 40� to 70�C and the heat-inducedstructural changes were discussed.

In the current work second harmonicgeneration (SHG) from collagen and visiblefluorescence are proposed as probes for theinvestigation of the degree of thermallyinduced denaturation. As SHG intensity ispresumably proportional to the percentage ofcollagen molecules with triple helix of micro-crystalline structure, it can directly indicatethe denaturation ratio, i.e. the percentage ofcollagen fibrils in denatured form, lackingcrystal structure and concomitant non-linearoptical properties. Characteristics collagenfluorescence around the 410 nm spectral regionon the other hand, induced by 337 nm nano-second pulsed later excitation, can be attrib-uted mainly to cross-linked chromophores.Here both SHG and characteristic fluor-escence of collagen at 405, 410 and 415 nm aresuggested as non-invasive probes of irrevers-ible thermal changes to the conformation ofnatural collagen. As SHG is dependent on theundenatured collagen fibrils it can relatedirectly to the denaturation ratio.

EXPERIMENTAL PROCEDURE

Type I collagen from Achilles bovine tendon(Fluka 27662) in fibrous form was placedbetween glass microscope slides to form analmost uniform 1 cm�1 cm square sample ofless than 0.5 mm thickness. The microscope

slides did not exhibit any absorption andemission in the spectral regions of interest(320–1100 mm). Quartz slides were avoided, asquartz is known to produce second harmonicradiation quite e$ciently.

Sample preparation was based on a previouslysuggested protocol [18]. More specifically thesamples were heated for 60 min in a water bathat temperatures ranging from 40 to 80�C (40�, 50�,55�, 60�, 65�, 70�, 75� and 80�C). Each sample wasthen cooled to 4�C for 15 min. Finally thesamples were brought to room temperaturewhere all optical measurements were performed.In this manner only irreversible structuralchanges were investigated. Each sample alsoserved as a control specimen, as it was immersedin water at 30�C for 60 min and then cooled to4�C prior to heating to a specific temperature.Fluorescence spectra and SHG intensity valuewere registered for each individual sample bothat the control temperature of 30�C and afterheating according to the protocol.

The set-up for SHG measurements is shownin Fig. 1(a). A Nd:YAG pulsed Q-switched

Fig. 1. Experimental set-up for (a) SHG measurements (b)characteristic fluorescence registration.

Thermally Induced Irreversible Conformational Changes in Collagen 35

Spectra Physics GCA 16 with 12 ns pulse dur-ation was used as the source of 1064 nm radi-ation. The pulse energy E was set to 215 mJ. Alens with +75 nm focal length was used tofocus the laser spot down to a 5 mm diametercircular spot on the sample surface. The inter-action region was precisely marked formeasurement repeatability after heating. Alongpass filter cutting o# at �<800 nm wasplaced between the laser and the sample toreject spurious radiation (e.g. flashlamp origi-nating). The resulting SHG signal was col-lected by a 600 �m-diameter optical fibre at 45�angle to the incident beam and directed to anOptometrics SDMC1-03 grating monochroma-tor set to 532 nm. At the exit of this monochro-mator an IR-blind Hamamatsu R4220 PMTinsensitive to 1064 nm radiation produced anelectronic signal which was then collected by aTektronix TDS 540D digital oscilloscope, inter-faced to a computer where the data werefinally stored.

The SHG intensity was registered from eachsample first at 30�C and then after heating to aspecific temperature (40�, 50�, 55�, 60�, 65�, 70�,75� and 80�C) and re-cooling according to thedescribed protocol. The signal from eachsample heat processed at a specific tempera-ture as per our protocol was then normalisedwith respect to the signal obtained before-hand from the same sample at the controltemperature of 30�C, to form the data set:

The exponential errors in the case of theSHG signal, were of a statistical nature, in theform of the standard deviation of the meanvalue of the SHG signal obtained from eachsample at the control temperature of 30�C, andwere found to be of the order of 8%. Theseerrors reflected the uncertainty in the signalintensity as measured and registered by ourdetection system.

The set-up for fluorescence registration isshown in Fig. 1(b). A Laser Science VSL-337ND nitrogen laser, with 4 ns pulse widthoperating at 30 Hz, delivered the excitationradiation of 0.3 mJ/pulse at 337 nm to a 0.8 cmdiameter circular spot on the smaple via afused silica optical fibre with 600 �m diameter.The same number of pulses were delivered toall the samples, summing to an overall inci-dent energy per sample of 0.4 J. Again theexcitation region was carefully marked for the

precise experiment repetition after sampleheating. Another fused silica optical fibre with1000 �m core diameter collected the fluor-escence and delivered it to the entrance of a1024 CCD linear array grating spectrometer. Ahighpass filter, with cuto# at 350 nm, wasplaced there to reject any residual excitingradiation. The CCD was calibrated withrespect to the wavelength with the use of astandard Ocean Optics HG-1 mercury–argonlamp, taking the advantage of its distinct spec-tral lines. The signal of the CCD was againstored in a computer interfaced to it via anA/D card.

From the collection of the fluorescence spec-tra at 30�C from all the collagen samples it isevident that we have a very good agreement inthe spectral region between 405 and 420 nm.From this region where the signal to noiseratio is best, three wavelengths were selected,namely 405, 410 and 415 nm, at which analysiswas carried out.

For each sample, elevated to a discrete tem-perature according to the protocol describedabove, the fluorescence intensities at thesewavelengths were normalised with respect tothe intensity at 30�C, to a new data set:

The experimental errors in the case of fluor-escence registration were calculated in thesame statistical manner as in the case of SHGsignal registration. In this case they werefound to be in the order of 10%.

In both SHG measurements and fluorescenceregistration the set-up geometry as well asexperimental conditions were carefully main-tained identical throughout the experiments.

The denatured counterpart of collagen, gela-tine (Fluka 48722) in granular form, was usedin an experiment aimed to explain someaspects of collagen fluorescence due to part ofits cross-linked chromophores also existingin gelatine, as well as to illustrate reversibleand irreversible changes due to heating asmonitored by fluorescent registration.

The fourth harmonic of the Nd-YAG laserdescribed earlier in the experimental section,was used according to the setup in Fig. 1(a)for the experiments with gelatine and wasobtained by consecutive doubling (1064�532 nm and 532�266 nm) in two non-linearKDP crystals (�=41� and �=78�, oo-e inter-action). The output parameters of the fourth

36 T. Theodossiou et al.

harmonic, used for UV excitation of tyrosine,were: �=266 nm, E=20 mJ. The fluorescentemission from this experiment was registeredfor the case of dry gelatine, gelatine in gel formafter the addition of 10 ml bidistilled water andheating, and the same sample after drying inambient temperature for two days.

RESULTS

The SHG signal was registered as described inthe experimental procedure first at 30�C andthen after heating to a specific temperatureand recooling. The signal from each samplelikewise heat processed was then normalisedwith respect to the corresponding signalobtained for that specific sample at the controltemperature of 30�C according to Equation (1).The results are shown in Fig. 2. In this plotthe values at 70�, 75� and 80�C can withinexperimental error be considered as zero.

Fluorescence measurements were againmade with the set-up described in the previoussection both at 30�C and each of the investi-gated temperatures. A characteristic collagenfluorescence spectrum can be seen in Fig. 3.

It can be seen that the maximum of thisspectrum lies at 395 nm. As also explained inthe previous section due to the good signal-to-noise ratio in the spectrum region 405–420 nm,

three wavelengths were selected from thatregion namely 405, 410, 415 nm for the analysisto be carried out. The fluorescence intensitiesat these wavelengths were normalised withrespect to the intensity at 30�C forming a newdataset according to Equation (2).

The set of these data versus temperature isgraphically presented for the three wave-lengths along with the normalised SHG inten-sities in Fig. 2. From this figure it can be seenthat the normalised SHG intensities and thenormalised fluorescence intensities follow asimilar pattern, however, the fluorescence datatend asymptotically to a value di#erent thanzero from 70�C upwards.

Fig. 2. The normalised SHG (1064→532 nm) values [ISHGnorm=ISHG(T)/ISHG(30°C)] of the collagen samples at eight tempera-tures as per the experimental protocol (.) and the corresponding normalised fluorescence values [IFnorm=IF(T)/IF(30°C)] for threeemission wavelengths, 405 nm (h), 410 nm (C) and 415 nm (m).

Fig. 3. Characteristic fluorescence from type I collagen frombovine Achilles tendon.


The results of the fluorescence registrationexperiments on gelatine, performed accordingto the description in the experimental pro-cedure section, appear in Fig. 4. Curve 2(dashed line) in Fig. 4 depicts the characteris-tic fluorescence of dry gelatine in granularform at ambient temperature. There is a maxi-mum for that fluorescence at 314 nm and thisis tyrosine-related fluorescence. However, themaximum is shifted with respect to curve 1(dotted line) which depicts the fluorescencespectrum of tyrosine and has a maximum at303 nm. The fluorescence around 410–420 nm ofdry gelatine (curve 2 dashed line) is almost aplateau and the ratio of 410/420 nm values ispractically one. The corresponding ratio of themaximum of tyrosine-like fluorescence overthat at 420 for the same curve is 0.77. Curve 3(solid line) on the other hand depicts the fluor-escence of gelatine gel produced after additionof 10 ml bidistilled water to 2 mg dry gelatineand heating. The tyrosine-like spectral part ofthis matches almost exactly curve 1 with amaximum at 303 nm. Around the 410–420 nmregion the spectral form is no longer a plateaubut has a distinct maximum at 410 and a410/420 nm value ratio of 1.2. The respectiveratio of the tyrosine-like fluorescence at maxi-mum over the value at 420 nm, is in this caseapproximately 1.

In curve 4 (solid circles) the fluorescencespectrum of the gel of curve 3 after dryingin ambient temperature for two days, is

presented. The tyrosine-like fluorescence maxi-mum is again shifted to longer a wavelength(311 nm) and the fluorescence at 410–420 nmtends to a plateau again, with a ratio of valuesat 410/420 nm of 0.95. The ratio of the maxi-mum of the tyrosine-like fluorescence over thevalue at 420 nm is 0.75.

DISCUSSION

SHG in collagen originates mainly from colla-gen macroscopic polymerised fibrils and theirmicrocrystal structure, which is dependent ontriple helix conformation. Gelatine, which hasa similar chemical composition to collagenbut lacks its tertiary structure, both at themolecular and fibril level does nto exhibit anydetectable SHG.

Unlike inorganic non-linear crystals, colla-gen fibrils have isotropic distribution and ori-entation. Hence the resulting SHG emission isalso isotropic (non-directional) and is not gov-erned by phase matching conditions butdepends on the coherence length between thefundamental and second harmonic radiationcomponents [2]. This being the case, if collagenfibrils have a diameter smaller or comparableto this e#ective length, then the second har-monic radiation emitted is proportional to thefibril population in triple helix conformation.On the other hand if the fibril diameter isbigger than the coherence length the above

Fig. 4. Normalised spectra of gelatine fluorescence at 266 nm nanosecond Nd:YAG laser irradiation.


proportionality is violated. In our experimentssince the SHG measurements are taken in thereflection geometry, it can be safely assumed tobe proportional to the fibril population in tri-ple helix form. The normalised second har-monic intensity of Fig. 2 with respect to the30�C denotes the percentage of collagen mol-ecules remaining in triple helix form in a givenprotocol temperature with respect to thetemperature of 30�C.

In this context SHG emerges as a powerfulprobe of the degree of thermal denaturation ofthe bovine Achilles tendon type I collagenunder investigation. The fact that the values ofthis curve are practically zero at 70�C andabove, indicates complete denaturation of oursample at these temperatures. In the samegraph (Fig. 2) the curves of fluorescence vs.temperature appear at the three emissionwavelengths selected, namely 405, 410 and415 nm. It can be readily seen that the fluor-escence curves are in very good agreementwith each other; however between 70 and 80�Cunlike the SHG curve there is some residualfluorescence still undergoing thermal quench-ing but at a di#erent rate. In natural formcollagen has a number of chromophore cross-links that emit around the 400 nm spectralregion. A big part of them is related to thetriple helical structure of the collagen and itsfibrils, so when collagen undergoes thermaldenaturation these cross-links are destroyed.However, random cross-links remain in colla-gen between individual amino acid strandseven after complete denaturation. These ran-dom cross links are also present in gelatine,the denatured counterpart of collagen and areresponsible for its characteristics fluorescencein the visible region. It must, however, benoted that upon heating of gelatine in anaqueous environment, these cross-links aredestroyed and their characteristic fluorescenceis quenched. Upon registration of this fluor-escence with 266 nm nanosecond 20 mJ/pulseirradiation, as detailed in previous sections,it was shown that whereas the cross-link-originating fluorescence was quenched, theendogenous (mainly tyrosine) fluorescencerose and this e#ect was reversible afterwater evaporation at ambient temperature(Fig. 4).

The results of the experiment on gelatine,can lead us to conclude that a part of gelatinerelated cross-link chromophores exist in tissueeven after complete thermal denaturation andcontinue to undergo thermal damage some of

which is reversible. This fact accounts for theresidual tail appearing in the fluorescencedata of Fig. 2. It can further be suggested thatthe chromophore species emitting around420 nm, is tyrosine-related and could be dityro-sine or polytyrosine as proposed elsewhere[13].

However, various other cross-link-chromo-phores, such as pentosidine or pyridinolinefound to exist in collagen fluorescence aroundthe 400 nm spectral region under UVA exci-tation [10,13]. All cross-link-chromophores ofcollagen are destroyed upon thermal denatur-ation and this is a result of the helix coiltransition as the amino acid strands becomedetached from one another. For some of thesechromophores this process is irreversiblewhereas for others it is reversible to a certainextent (e.g. gelatine-associated cross-links) asshown above.

Figure 5 shows the di#erentials of the fourcurves of Fig. 2, with their sign changed sothat the biggest rate of destruction corre-sponds to a maximum instead of a minimum.Curve 1 depicts the di#erential of the SHGsignal of Fig. 2 with temperature. Curves 2, 3and 4 show the corresponding di#erentialcurves of collagen fluorescence vs. tempera-ture for the three selected wavelengths 405, 410and 415 nm, respectively. Curve 1 starts risingat about 36�C monotonically, indicating a con-stantly increasing rate of permanent destruc-tion of triple helix formations in collagen, upto about a temperature of 55�C. From thistemperature upward, there is an even steeperincrease of the rate of destruction up to64�C where there is a maximum. Followingthat there is a very steep decrease of the rateof helix-coil transition up to about 75�Cwhere the rate becomes zero due to completedenaturation of our sample.

Since our experimental protocol wasdesigned to detect permanent changes incollagen, this further indicates practically anirreversible phase transition centred about64�C of collagen into gelatine. The fluor-escence curves on the other hand practicallyshow a steady destruction of flurophores from30 to 55�C which indicates recombination ofsome destroyed fluorophores and that is whythe curve is not in complete agreement withthe corresponding curve of SHG in that region.From 55 to 77�C, however, there is a notableincrease in the permanent destruction of colla-gen fluorophores which reaches a maximum atabout 63�C and then drops again to a steady


state of destruction of the residual collagenfluorophores after complete denaturation.

Both the SHG and fluorescence di#erentialcurves show a phase transition of collagencentred around 63–64�C, into its denaturedform, which is in complete agreement withprevious experimental results from di#erentialcalorimetry [15,16]. In Tiktopulo and Kajava[16] the phase transition point (52�C) is di#er-ent from that of the present work (63–64�C) andthat of Miles et al. [15] (62�C). This can beexplained by the fact, that skin tissue was usedwhich is expected to have a lower phase tran-sition point than tendon tissue, as used in thetwo later cases.

The SHG measurements presented here aremore sensitive to thermal damage of tissuesince they directly monitor destruction oftriple helix of fibril formations in tissue.Fluorescence on the other hand, can beinduced by quite low energy density of incidentradiation due to its one photon nature butcannot directly associate with the thermaldenaturation. Instead it measures chromo-phore destruction, a part of which is revers-ible. In addition, photobleaching of thechromophores during resonant irradiation isan important factor that might a#ect measure-ments in vivo. The present experiments, how-ever, where photobleaching was negligible,show that a part of the tissue cross-linked

chromophores associated to triple helix fibrils,undergo permanent damage and are thuscapable of reflecting irreversible phase tran-sitions during tissue thermal denaturation.

In the current work, experiments designedto register irreversible changes to type I colla-gen from Achilles bovine tendon, showed thatboth SHG and fluorescence (optical non-invasive methods unlike colorimetry) havesuccessfully demonstrated thermally inducedphase transitions around the temperature of64�C.

This result can prove invaluable to laserablative surgery, where the condition andpermanent conformational changes to the tis-sues surrounding the ablation area can beprobed in vivo. Another field where the cur-rent work can be applied to, is laser tissuewelding where it is important to know whenphase transition has been achieved, so thatfurther irradiation and consequently thermaldamage can be avoided.

REFERENCES

1. Fine S, Hansen WP. Optical Second harmonic gener-ation in biological systems. Appl Opt 1971;10:2350–3.

2. Freund I, Deutch M, Sprecher A. Connective tissuepolarity. Optical second-harmonic microscopy,crossed-beam summation, and small-angle scatteringin rat-tail tendon. Biophys J 1986;50:693–712.

Fig. 5. Differential curves of the data in Fig. 2 with inversed sign so that maximum corresponds to biggest rate of irreversiblethermal destruction of triple helix population in the case of SHG and cross-link chromophores in the case of fluorescence at thethree emission wavelengths.


3. Guo Y, Ho PP, Tirksliunas A, Liu F, Alfano RR.Optical harmonic generation from animal tissues bythe use of picosecond and femtosecond laser pulses.Appl Opt 1996;35:6810–13.

4. Theodossiou T, Georgiou E, Hovhannisyan V, Yova D.Visual observation of infrared laser speckle patternsat half their fundamental wavelength. Lasers Med Sci2001;16:34–9.

5. Yova D, Theodossiou T, Hovhannisyan V. Photo-induced e#ect of hypericin on collagen and tissue withhigh collagen content. SPIE Proc 1999;3565:174–80.

6. Hovanessian V, Lalayan A. Second harmonic gener-ation in biofiber-containing tissues. Proc Int Conf OnLasers ’96, Portland, OR, 1996:107–9.

7. Georgiou E, Theodossiou T, Hovhannisyan V,Politopoulos K, Rapti GS, Yova D. Second and thirdoptical harmonic generation in type I collagen, bynanosecond laser irradiation, over a broad spectralregion. Opt Comm 2000;176:253–60.

8. Fraser RDB, Macrae TP. The crystalline structure ofcollagen fibrils in tendon. J Mol Biol 1979;127:129–33.

9. Burshtein EA. Autoluminescence of Proteins (Natureand Applications). Moscow: VINITI AN SSSR (series‘Biofizika’, in Russia), 1977.

10. Uchiyama A, Ohishi T, Takahashi M, Kushida K,Inoue T, Fuijeand M, et al. Fluorophores from aginghuman cartilage. J Biochem 1991;110:714–18.

11. Volkov AS, Kumekov SE, Syrgaliev EO, ChernyshovSV. Photoluminescence and anti-Stokes emission of

native collagen in visible range of spectrum. Biofizika1991;36:770–3.

12. Fujimoto D, Akiba K, Nakamura N. Isolation andcharacterization of a fluorescent material in bovineAchilles tendon collagen. Biochem Biophys ResCommun 1977;76:1124–9.

13. Menter JM, Williamson JD, Carlyle K, Moore CL,Willis I. Photochemistry of type I acid soluble calf skincollagen. Dependence of excitation wavelength.Photochem Photobiol 1995;62(3):402–8.

14. Yova D, Hovhannisyan V, Theodossiou T. Photochemi-cal e#ects and hypericin photosensitized processes incollagen. J Biomed Opt 2001;6:52–57.

15. Miles CA, Burjanadez TV, Bailey AJ. The kinetics ofthe thermal denaturation of collagen in unrestrainedrat tail tendon determined by di#erential scanningcalorimetry. J Mol Biol 1995;245:437–46.

16. Tiktopulo EI, Kajava AV. Denaturation of type Icollagen fibrils is an endothermic process accompaniedby a noticeable change in the partial heat capacity.Biochemistry 1998;37:8147–52.

17. Shnyrov VL, Lubsandirzhieva VC, Zhadan GG,Permyakov EA. Multi-stage nature of the thermaldenaturation process in collagen. Biochem Int1992;26:211–17.

18. Rochdi A, Foucat L, Renou JP. E#ect of thermaldenaturation on water collagen interactions: NMRrelaxation and di#erential scanning calorimetryanalysis. Biopolymers 1999;50:690–6.

Paper received 2 February 2001;accepted after revision 7 May 2001.


thermally induced irreversible conformational changes in collagen probed by optical second harmonic...

Documents