2.1.

Tissue Preparation

Animals were available from other studies, supplied in accordance with protocols approved by the Animal Care Committee at the University. Tissue samples were harvested immediately after death from 27 male rats (25 Wistar, 2 Sprague Dawley), weight range 35 to $560\mathrm{g}$ (aged 15 to 115 days); animals were euthanized by either carbon dioxide inhalation or intraperitoneal injection of pentobarbital $(80\mathrm{mg}∕\mathrm{kg})$ . Three or four tendon samples were dissected from the base region of each tail along with two samples of skin (one sample for measuring inner skin and one for outer skin). Tissue samples were stored in saline at $4°\mathrm{C}$ until required for up to $6\mathrm{h}$ prior to spectroscopic measurements followed by fixation in 10% formalin for $24\mathrm{h}$ for subsequent histological studies.

2.2.

Age-Weight Correlation

The weights of all animals in the experiments were known and for a few animals the actual age at sacrifice was also known. Tabulated data on animal ages and weights from the supplier (Charles River Company) were used to explore empirical relations. Age versus weight had a sigmoidal relationship fit to $\mathrm{Age}=459∕[1+(w∕w_0{)}^{-2.6}]$ with $R^2=0.998$ and a low standard error (SEE of $\pm 4.4\mathrm{day}$ ), where age is measured in weeks, $w$ is weight in grams, and $w_0$ is an empirical constant, not necessarily newborn weight. We chose to use weight for all graphical representations since weights were known for all animals.

2.3.

Laser-Induced Fluorescence Instrumentation

The method for obtaining fluorescence spectra was developed for earlier research.^{23, 24} Fluorescence emission was triggered by UV excitation using an Argon laser (Innova 90, Coherent Laser Products, Palo Alto, California) operating at $357\mathrm{nm}$ with a chopped wave output (0.2-ms exposure time; 1.0-ms delay) using a Uniblitz shutter producing an emission energy of 0.5 to $1.5\mathrm{mJ}∕\mathrm{pulse}$ (Vincent Associates, Rochester, New York) as previously described.²⁵ By using a low laser power well below that at which photobleaching had been demonstrated in earlier pilot studies we avoided this complication. The laser beam was focused onto a 6F (French) laser angioplasty ring catheter (CeramOptec, Bonn, Germany). Spectra were collected with the tip of the catheter in perpendicular contact with the tissue surface, positioned using a micrometer stage with three degrees of freedom. The excitation radiation and collection of emission spectra occur through the same probe as illustrated in Fig. 1 . Spectra were recorded with an Optical Multichannel Analyzer (Model 1462, EG&G PARC, Toronto, Ontario). Prior to each data recording session the fluorescence emission intensity (FEI) spectra were calibrated with a mercury lamp.

Fig. 1

Instrument capture setup for detection of emission spectra following UV excitation. Positioning of the fiber optic, in the $x$ , $y$ , and $z$ directions, is achieved with micrometer controls. The fiber optic is placed perpendicular to tissue surface for optimal emission collection. Far right is a cross section of the fiber optic illustrating the concentric ring of silica fibers used to excite and collect light.

Spectra were obtained from animal samples of tail tendon and skin (dermal and epidermal surfaces) and separately from standard preparations of ECM proteins (collagen types I, and III and elastin) described in greater detail in the subsequent section. Ten spectra were recorded for each sampling site after background subtraction, and stored digitally for postprocessing. The system has a diminishing sensitivity with tissue depth, to a maximum penetration of $0.2\mathrm{mm}$ , with a sampling area of $50\times 50\mathrm{\mu }\mathrm{m}$ . Tissue sample thickness of skin and tendon increased with age with values of 1 and $2\mathrm{mm}$ , respectively, for the oldest animal group.

2.4.

Protein Reference and Tissue Spectra

We used ECM protein standards from Sigma Chemical to establish separate reference spectra for each of collagen type I (rat tail tendon C-7661, acid soluble powder, Sigma Chemical, St. Louis, Missouri), collagen type III (calf skin C-3511, acid soluble), and elastin (bovine neck ligament E-6527, saline soluble powder). The proteins were placed in saline to provide a moist contact with the fiber optic catheter tip and to mimic the hydrated state of proteins in tissue. For tissue analysis, measurements were acquired in vitro from rat tail tendon and skin samples (epidermal and dermal surface). The 27 animals were divided into three weight classes for detection of gross fluorescence changes and based on natural weight clusters. Groups A, B, and C having 6, 6, and 15 rats respectively (presented in Table 1 with results) with mean weights and standard errors (SEs) of $70\pm 14$ , $182\pm 21\mathrm{g}$ , and $510\pm 11\mathrm{g}$ . The animals were also divided into seven groups with uneven $n$ ( $n=2$ , 2, 4, 2, 2, 11, 4 in order of increasing weight) based on the data clusters to obtain a regression fit as a function of weight.

2.5.

Analysis of Spectral Data

Statistical methods for spectral data included analysis of variance (ANOVA) and multiple regression, as previously described.^{21, 23, 26} Postprocessing began with calculating the average intensity at each frequency for each sampling site.²² Mean values of maximum FEI were plotted as a function of weight, where the error bars are standard error. The mean spectra were then smoothed with a 15-point moving average (software written in MATLAB 6.5) and normalized to peak height by dividing fluorescence intensity at all wavelengths by the maximum intensity of the smoothed spectra (similar methods incorporated by Marcu ²⁷). The wavelength edge below $390\mathrm{nm}$ was not analyzed because these data included components from the incident excitation beam.

The mean normalized FEI over the range of 390 to $600\mathrm{nm}$ was analyzed with a one-way ANOVA using statistical software (SPSS version 11.5, Chicago, Illinois). A value of $p<0.05$ was our reference for statistical significance. Least significant difference (LSD) post hoc analysis was used to evaluate all possible pairwise comparisons between the three weight groups. Subsequent analysis of normalized average tendon spectra for the three weight groups was performed using stepwise linear regression; we used the extracted protein signature spectra (collagen types I and III and elastin) as reference functions. Regression analysis was used to calculate the relative weightings of each reference spectra that, when combined, closely approximated the measured tendon spectra (shown later in Fig. 6).

2.6.

Histological Preparation

Samples of fixed tissue for each weight range from rat tail tendon and skin were processed for histology with paraffin embedding, sectioned at $5\mathrm{\mu }\mathrm{m}$ and stained with Movat’s pentachrome stain for histological analysis²⁸ (two sections per slide). We used a Zeiss Axioskop microscope equipped with a color video camera (Sony Power HAD 3CCD) and Northern Eclipse program (Empix Imaging Inc., Mississauga, Ontario) for point counting. This was done to calculate volume fractions of elastin and collagen in tendon and skin as a reference for fluorescence spectroscopy measurements. For polarized light, measurements were made on unstained sections cut parallel to the surface of the tendon and from the inner and outer surfaces of the flattened skin samples (i.e., surface of dermis and epidermis).

2.7.

Birefringence Measurements

The methods for measuring the strength of birefringence are well established by other researchers⁷ and have been incorporated into previous studies by the authors.^{10, 29} Measurements were made on a Nikon Optiphot-Pol polarizing microscope (Nikon, Melville, New York) using a Sénarmont compensator, and a narrow-band interference filter that produces near monochromatic light $(\lambda =546\mathrm{nm})$ . Birefringent materials such as collagen have two indices of refraction, $n_o$ and $n_e$ , associated with the “ordinary” and “extraordinary” rays of the resolved incident light. When linearly polarized light passes through the birefringent element of tissue, it is resolved into two rays. These pass through the tissue at different velocities, resulting in a phase difference, or retardation, of one ray relative to the other. The Sénarmont compensator is a $\lambda ∕4$ plate inserted between the specimen and the analyzer, which retards one of the emerging rays by a quarter wavelength, converting the elliptically polarized light emerging from the birefringent tissue back to linearly polarized light at a rotated angle. This angle $\theta$ is proportional to the phase retardation or birefringence, and is measured by rotating the analyzer until extinction of the light is obtained. The birefringence $B$ is calculated from $B$ $\left(\mathrm{nm}\right)=(\lambda ∕180)\theta$ , where $\lambda$ is the wavelength of the incoming light. To determine the relationship between birefringence and maturity, the animals were separated by weight into seven groups. From each of the samples of tail tendon and inner and outer skin, 10 measurements of collagen birefringence were made from different fiber regions.

The geometry of the zigzag patterns of collagen is well documented in the literature. We investigated age-related changes in the geometry of fibers of RTT by making measurement of the crimp angle, $\theta$ and of the crimp period or wavelength, $2{\ell }_o$ .^{30, 31} A representative sample of slides from the three weight groups was measured with two sampling areas per slide. Since the plane of the zigzag is not necessarily the same as the plane of sectioning, we also made measurements of the three dimensional (3-D) orientation of collagen fibers using polarized light. The measurement of the 3-D orientation of collagen fibers using a well-established technique is described elsewhere.³² In brief, the alignment of collagen fibers was measured using a Zeiss polarized light microscope with a rotating universal stage attachment. The universal stage enables the slide to be tipped and rotated in oblique planes. Thus, the two angles that define the orientation of a collagen fiber in 3-D space can be measured—the azimuthal angle (in the plane of the stage) and the elevation angle (out of the plane of stage). The crimp angle and period for the three weight groups were analyzed using ANOVA.

3.1.

Micrographs

The micrographs in Fig. 2 , showing a tendon cross section stained with Movat’s pentachrome and a polarizing micrograph of a longitudinal section stained with the picrosirius red⁸ (for micrograph contrast enhancement), demonstrate the dimensions and general morphology of the tissue. The cross section shows the relatively unique feature of RTT having collagen bundles surrounded by a thin sheath of elastin. The micrograph of Fig. 2a and two other similar micrographs at different RTT stages of maturation were used to calculate the proportion of tail tendon that is elastin used for later comparisons with the calculations from fluorescence analysis. The percentage of elastin, $\sim 10\%$ of the tendon volume, was unchanged among the three weight groups.

Fig. 2

(a) Micrograph of tendon cross section with Movat pentachrome stain where nuclei and elastin are stained black and fibrous tissue is yellow, illustrating tissue organization, scale $\mathrm{bar}=0.2\mathrm{mm}$ ; and (b) circularly polarized light micrograph of a longitudinally sectioned tendon stained with picrosirius red, scale $\mathrm{bar}=0.2\mathrm{mm}$ . The schematic, lower right, illustrates the zigzag pattern of the collagen fibers with crimp angle $\left(\theta \right)$ and crimp period $\left(2{\ell }_o\right)$ .

The variable fiber wave pattern appears as a zigzag, as shown by circularly polarized light [Fig. 2b]. In the planar direction, the collagen fibers had a mean amplitude angle and angular deviation of $32\pm 8\mathrm{deg}$ (to the long axis) and no significant differences between the three RTT age groups. However, the crimp period increased with weight from $48.6\pm 9.8\mathrm{\mu }\mathrm{m}$ ( $\sim 35\text{-}\mathrm{g}$ animal) to a maximum of $150.9\pm 19.5\mathrm{\mu }\mathrm{m}$ ( $\sim 580\text{-}\mathrm{g}$ animal) and did show a significant change between the immature and mature groups $p<0.01$ (ANOVA). The zigzag of the fibers defines a local plane in three dimensions, which may be quite different from the sectioning plane. These measurements are made from the projection of the zigzag crimp in the sectioning plane; hence, the crimp angles will be underestimated when the plane of the zigzag is substantially different from the sectioning plane. Three-dimensional measurements of the angular orientation of collagen fibers out of the sectioning plane showed that there were distinct alignment differences among adjacent fiber bundles and along individual fibers. These ranged to a maximum of $40\mathrm{deg}$ crimp angle out of the section plane. Our measured crimp angles are greater than those described by Diamant ³⁰ but are within the range of values described by Gathercole and Keller.³¹ The crimp period or wavelength varies as a function of weight and was found to increase significantly during maturation, being within the published range.³⁰ Results of the crimp period or wavelength are independent of the plane of the zigzag.

3.2.

Birefringence Measurements

Blinded sets of birefringence measurements from longitudinally cut tissue sections of tail tendon and inner and outer skin samples from each animal were obtained by two microscopists. The data were grouped into the same seven weight groups as were used for the fluorescence study. Results for tendon and skin are shown in Fig. 3 . Inner versus outer skin birefringence data were found to be highly correlated $(y=0.96x+0.19,R^2=0.95)$ and were therefore combined for weight relationships. The strength of birefringence $B$ increased with animal weight $W$ and reached a plateau for both skin and tendon, described by $B=a(1-e^{-bW})+c$ , where $R^2=0.95$ and $R^2=0.90$ , respectively. Birefringence values were higher for tendon compared to skin, an expected finding due to the dominance of type I collagen and larger fiber size.

Fig. 3

Birefringence as a function of weight. Exponential fit described by tendon, $B=6.0(1-e^{-0.010W})+4.4$ $(R^2=0.90)$ , and skin, $B=7.8(1-e^{-0.010W})-0.78$ $(R^2=0.95)$ .

3.3.

Fluorescence Spectroscopy

The normalized and smoothed fluorescence emission spectra of purified extracted proteins commonly found in skin and tendon are shown in Fig. 4 . The FEI spectra for elastin and collagen types I and III have distinct spectral features. Elastin has a broad peak with its maximum ranging from 435 to $445\mathrm{nm}$ and a full width at half maximum (FWHM) of $110\mathrm{nm}$ . Type I collagen has a narrower peak with a maximum at $415\mathrm{nm}$ and a FWHM of $71\mathrm{nm}$ , and type III collagen peaks at $440\mathrm{nm}$ with FWHM of $95\mathrm{nm}$ . These characteristic spectral differences between the proteins enable separation using statistical approaches.

Fig. 4

Signature fluorescence emission spectra of the extracellular matrix proteins elastin and types I and III collagen in saline normalized to peak wavelength.

Smoothed and averaged emission spectra of tendon and inner and outer skin were obtained and grouped by animal weight and are plotted in Fig. 5 . Table 1 summarizes the numbers of animals in each group for analysis. The descending portion of the spectra is the most discriminating, and is therefore of importance for analysis. Significant differences in the one-way ANOVA $(p<0.05)$ were observed over the entire descending wavelength range for tendon, as illustrated in Fig. 5a, and specific ranges for inner skin (390 to 445 and 570 to $590\mathrm{nm}$ ) and outer skin (390 to 440 and 555 to $590\mathrm{nm}$ ) in Fig. 5b. Spectra for the inner skin were similar to those of tendon both in magnitude and peak wavelength.

Fig. 5

Emission spectra divided into three age groups (A, B, and C) for (a) rat tail tendon and (b) inner skin and outer skin.

The normalized spectral data for the three tendon weight groups were analyzed using stepwise regression in the 390- to $600\text{-}\mathrm{nm}$ range. Since collagen type I and elastin are the main components of tendon, these basis functions were input for all three weight groups. For tendon group A $(70\pm 14\mathrm{g})$ the variance is explained by collagen I and elastin with a high correlation of $R^2=0.96$ . Similarly for group B the variance is explained by collagen I and elastin with $R^2=0.98$ , and a sample fit is illustrated in Fig. 6 . For group C, the variance is explained by collagen types I and III and elastin with $R^2=0.98$ $(p<0.0001)$ . The morphometric analysis of the pentachrome-stained section of tendon correlates with the regression analysis. The maximum fluorescence intensity is a comparative measure of the molecular changes and increased cross-linking that occurs during maturation. Figure 7a displays the positive correlation between maximum emission intensity for rat tail tendon as a function of weight, using weighted regression analysis to incorporate the nonuniformity in error $(R^2=0.64)$ . For this part of the study the animals were divided into seven groups based on weight. The measurement process was repeated for RTT skin where there was a significant positive correlation for the epidermal surface $(R^2=0.65)$ ; however, for the dermal skin surface there was no correlation, as illustrated in Fig. 7b.

Fig. 6

Normalized spectrum of tendon group B represented by data points. A stepwise regression fit for tendon group B is represented by the solid line where $y=0.{144}_{\mathrm{collagen}1}+0.{887}_{\mathrm{elastin}}+0.003$ with $R^2=0.977$ .

Fig. 7

Graphs of maximum FEI as a function of weight for (a) tendon where the best fit line is ${\mathrm{FEI}}_{\max }=0.53w+453$ ( $R^2=0.64$ and $p=0.03$ ), and (b) dermal surface, ${\mathrm{FEI}}_{\max }=0.22w+497$ ( $R^2=0.09$ and $p=0.51$ ), and epidermal surface, ${\mathrm{FEI}}_{\max }=2.52w+445$ ( $R^2=0.65$ and $p=0.03$ ).