Pregnancy-Induced Dynamical and Structural Changes of Reproductive Tract Collagen

Abstract

The tissues and organs of the female reproductive tract and pelvic floor undergo significant remodeling and alterations to allow for fetal growth and birth. In this work, we report on a study of the alterations of murine reproductive tract collagen resulting from pregnancy and parturition by spectrophotometry, histology, and ¹³C, ²H nuclear magnetic resonance (NMR). Four different cohorts of rats were investigated that included virgin, multiparous, two- and fourteen-day postpartum primiparous rats. ¹³C CPMAS NMR revealed small chemical shift differences across the cohorts. The measured H-C internuclear correlation times indicated differences in dynamics of some motifs. However, the dynamics of the major amino acids, e.g., Gly, remained unaltered with respect to parity. ²H NMR relaxation measurements revealed an additional water reservoir in the postpartum and multiparous cohorts pointing to redistribution of water due to pregnancy and/or parturition. Spectrophotometric measurements indicated that the collagen content in virgin rats was highest. Histological analysis of the upper vaginal wall indicated a signature of collagen fiber dissociation with smooth muscle and a change in the density of collagen fibers in multiparous rats.

Introduction

Pregnancy-induced alterations to the tissues and organs of the female pelvic floor and reproductive tract allow for fetal growth and facilitate spontaneous vaginal birth. For such change to occur, the extracellular matrix (ECM) of these structures, composed of collagen, elastic fibers, and smooth muscle, may undergo significant remodeling. Collagen, the most abundant ECM protein in the human body, plays a vital role in providing structure and tensile strength to tissues of the female pelvic floor (i.e., pelvic floor connective tissue) and reproductive tract. Studies of the reproductive tract in humans and rodents have demonstrated that collagen content of the vagina, cervix, and uterus is drastically altered during gestation, only to return to prepregnancy levels after postpartum involution (1, 2, 3). Additionally, collagen and other ECM proteins remodeled in the reproductive tract during pregnancy and parturition have been associated with altered tissue biomechanics that typically results in an increase in tissue distensibility, which returns to the non-pregnant state postpartum (3, 4, 5, 6).

The influence of multiparity and aging on the process of remodeling is less clear. Rundgren (3) studied the influence of multiparity and age on the collagen content and biomechanics of the reproductive tract by comparing old (22–23 months of age) multiparous versus old nulliparous rats, and young virgin versus old virgin rats. He concluded that age had a greater influence on the physical properties of the uterine horns and cervix, while multiparity had greater influence in the vaginal wall. As multiparity and aging are major risk factors for the development of pelvic floor disorders such as pelvic organ prolapse and stress urinary incontinence, it is thought that a better understanding of these processes may provide greater insight into the pathogenesis of these conditions (7, 8, 9).

In light of the known alterations to collagen and the biomechanical properties of the reproductive tract, we applied nuclear magnetic resonance (NMR) methods to study the effects of parturition on collagen. Collagen is a fibrous protein and does not form crystals, but the x-ray structures of collagen like proteins and peptides have been reported by Bella et al. (10). To characterize the structural and dynamical properties of collagen at molecular level, NMR is an ideal tool as ¹³C chemical shifts can be useful for conformational characterization of the secondary structure, and NMR relaxation times may be applied to measure dynamical characteristics (11, 12). The signature molecular structure of collagen consists of 3 α-polypeptide chains bound in a triple helix configuration, stabilized by interchain hydrogen bonds. These chains are composed of a series of a triplet amino-acid patterns: glycine-X-Y, where X and Y can be any amino acid (but are frequently proline and hydroxyproline, respectively) (13). Collagen exists as either homotrimers or heterotrimers. There have been at least 28 different collagen types described. Collagen I, III, and V have been found in the pelvic floor connective tissue where I and III are most abundant. Collagen I fibers provide the majority of tissue resistance to tension. Collagen III confers greater flexibility and distension to tissue, while collagen V appears to be of minor importance as it forms small fibers of very low tensile strength. The copolymerization of collagen I with III and V forms fibrils that influence the biomechanics of tissue (14). It is thought that the higher the collagen I/III ratio in the ECM, the greater its strength. These fibrillar collagens are thought to be the principal determinants of strength in the vaginal wall. Changes in ratios, structure, and/or dynamics of these collagens likely contribute to changes in vaginal tissue behavior (15).

A well-resolved ¹³C NMR spectrum of collagen-like peptides and collagen fibrils from bovine tendons has been obtained, allowing for structural characterization (16, 17, 18). In previous NMR studies, ¹³C NMR chemical shift measurements were used to understand the flexibility of the collagen triple helix, which is sensitive to the ring conformation of the collagen imino residues. In addition, recent ¹³C solid state NMR was employed to follow the molecular disorder of collagen in articular cartilage, due to alkaptonuria (19). Dawson and Wray (20) also applied ³¹P NMR to study rat uteri; however, to date, no other study has made use of ¹³C NMR to study the effect of parity on the molecular structure of collagen in the reproductive tract.

In this study, we report on an experimental investigation of the effects of parity on the structure and dynamics of upper vaginal wall collagen derived from Sprague-Dawley rats using NMR spectroscopic and relaxation techniques. ¹³C cross polarization under magic-angle spinning (CPMAS) NMR was employed to probe structural changes. ¹³C relaxation experiments were also performed to measure correlation times of internuclear ¹³C-¹H motions to probe dynamical fluctuations of the spectroscopically resolvable moieties across virgin and multiparous rats, as well as primiparous rats at 2 and 14 days postpartum. ²H T₁-T₂ experiments based on an inverse Laplace transform (ILT) (21) have been performed to characterize the distribution and dynamics of water in the upper vaginal wall tissue. This powerful experimental technique has been previously applied on various systems such as potato tissue (22), cement paste (23), bovine nuchal ligament and aortic elastin (24), spider silk (25), cortical bone (26) and in water-saturated sedimentary rock (27). Lastly, the collagen content in the upper vaginal wall of each of the samples studied was measured by spectrophotometry and histology.

Materials and Methods

Preparation of tissues for NMR

The animal protocol for this study was approved by the Institutional Animal Care and Use Committee at the Albert Einstein College of Medicine. Sixteen Sprague-Dawley rats (four per group) were used in total. These included virgin rats (11–14 weeks of age), multiparous rats (9–15 months old), two-day postpartum primiparous rats (para-1 (2 days postpartum)), and 14-day postpartum primiparous rats (para-1 (14 days postpartum)) after spontaneous vaginal delivery that were not nursing. All animals were sacrificed in a CO₂ chamber and their reproductive tracts harvested. Multiparous rats were sacrificed at least two weeks after their final parturition. A ∼1-cm portion of upper vaginal tissue was excised and used for our NMR studies. Lower vaginal tissue and cervix were discarded. Two rats in each group were used.

To remove fat from the sample, the following procedure was performed (28): each specimen was washed with saline solution three times. Then, the tissue was kept in 100% ethanol for 7–10 min and the procedure was repeated twice. The tissue was then placed in a (50:50) mixture of ethanol and diethyl ether (99%) for 7–10 min (two times). Finally, the tissue was placed in diethyl ether for 10 min (2 times). All samples were defatted using the same procedure and to keep samples hydrated, all the samples were stored in distilled water at −20°C. NMR experiments were performed on water-hydrated tissue. Furthermore, careful attention was taken to check the sample mass before and after NMR experiments to avoid dehydration during the experiments.

¹³C NMR experimental parameters

¹³C NMR experiments were performed at a magnetic field strength of 17.61 T using an Avance spectrometer (Bruker, Billerica, MA). Each sample was packed in a 4-mm center-packing rotor with an insert to prevent the loss of water as well as to center the sample with respect to the RF coil. All the experiments were performed at (300 ± 1) K. All pulse sequences used in this study are shown in Fig. S1 in the Supporting Material. In all experiments, the ¹³C π/2 pulse was 5 μs and the spinning frequency was (10 kHz ± 10 Hz). In the CP experiments, the contact time was set to 0.5 ms and recycle delay was 6 s. The spectra were acquired by accumulating 10,000 scans using 80 kHz TPPM decoupling (29). ¹³C T_1ρ were measured at two different spin-lock RF field strengths (25.00 and 50.00 kHz). The spin-locking time interval used for the low field varied from 50 to 750 μs and for the high-field measurements, the value was varied from 100 to 3000 μs. For all ¹³C T₁ measurements, the delay time τ was varied from 1 to 5000 ms in 10 steps. Analysis of data was performed using MATLAB (The MathWorks, Natick, MA) and matNMR (30) with a Gaussian multiplication broadening factor of 100 Hz for CP spectra and 150 Hz for relaxation measurements. ¹³C NMR spectra presented in this work were referenced to adamantane (with Tetramethylsilane (TMS) at 0 ppm). The mass of hydrated samples used in the NMR experiments for virgin, multiparous, para-1(2 day postpartum), and para-1(14 day postpartum) was roughly (41.8 ± 0.1) mg, (37.9 ± 0.1) mg, (54.4 ± 0.1) mg, and (53.9 ± 0.1) mg, respectively.

²H NMR relaxation

All the ²H T₁-T₂ experiments were performed on a magnetic field strength of 4.70 T using a Unity system (Varian, Palo Alto, CA) with a liquids double-resonance NMR probe (Varian) at 25°C. A portion of samples used in ¹³C NMR experiments was used for ²H NMR experiments. Samples were soaked in D₂O for ∼48 h before experiments. For all the experiments, the ²H π/2 pulse was 24 μs and the interpulse spacing in the CPMG train was 700 μs. For the two-dimensional (2D) T₁-T₂ measurements, the recovery time was logarithmically incremented from 1 ms to 10 s in 100 steps to measure T₁, and stroboscopic detection of 6000 echo peaks was used for the T₂ relaxation time measurement. A 2D ILT algorithm described elsewhere was applied to analyze the ²H T₁-T₂ relaxation data (21).

Histology and microscopy

A small sample of the upper vaginal wall was placed overnight in 100 mL of PBS (phosphate-buffered saline) with 30 g sucrose. The sample was then placed in CryoGel (Ref-02694-AB; Leica Biosystems, Fisher Scientific (Guilford, CT) and SPI Supplies, Structure Probe (West Chester, PA)), sectioned on a cryostat (CM1850; Leica Biosystems) at 10 μm and stained using the elastic stain kit (REF HT25A-1KT; Sigma-Aldrich, St. Louis, MO) following a previously reported protocol (5). The slides were placed in a solution containing 20 mL hematoxylin solution, 3 mL of ferric chloride solution, 8 mL of Weigert iodine solution, and 5 mL of deionized water for 10 min. Following this step, the slides were then placed in a working ferric chloride solution consisting of 3 mL ferric chloride and 37 mL of deionized water for 2 min. The slides were then gently rinsed with water, placed in 100% ethanol for a few seconds, and again gently rinsed with deionized water. In the last step of the staining procedure, the slides were placed in Van Gieson’s solution for 1.5 min, followed by a gentle rinse with 100% ethanol, and placed in xylene for a few seconds. Coverslips were mounted using Eukitt quick-hardening mounting medium (REF 03989; Sigma-Aldrich) and the slides were left to dry overnight. The slides were photographed using a DC4-156-S digital microscope (National Optical & Scientific Instruments, Schertz, TX) at a magnification of 10×. Approximately 50 images were taken from each sample from different regions of the upper vaginal wall. The photographs were then analyzed by an image processing program developed in MATLAB. All the images for collagen fiber analysis were taken using 100× objective with immersion oil on a BX41 microscope (Olympus, Melville, NY).

Collagen content quantification

Upper vaginal wall tissue was harvested, defatted, and lyophilized for 24 h, as described above. For the collagen quantification, 3.4–3.5 mg of lyophilized tissue from each sample was placed in 300 μL of 6 M HCl solution in hydrolysis tubes under nitrogen after evacuation. For hydrolysis, the samples were kept in an oil bath at 110°C for 24 h. After hydrolysis, tubes were opened and allowed to dry at 110°C. The samples were resuspended in 2 mL of PBS for 1 h at 60°C. Each sample was centrifuged and the supernatant was used for collagen quantification after twofold dilution using 4 M HCl. Collagen standards were prepared in the range of 0–300 μg/mL from the 1200-μg/mL stock solution provided by QuickZyme Biosciences (Leiden, South Holland, the Netherlands). All of the samples and standard dilution procedures were adapted from a standard assay collagen kit and previously reported protocol (31). The standard and samples were read in a 96-well plate reader at wavelength 584–590 nm (Exciting, 584 nm; Emitting, 590 nm) using a FLUOstar OPTIMA microplate reader (BMG Labtech, Thermo Fisher Scientific, Waltham, MA).

Results and Discussion

¹³C CPMAS NMR studies of structural and dynamical changes in virgin, multiparous, and para-1—2- and 14-days postpartum Sprague-Dawley rat upper vaginal wall collagen

Collagen is a ubiquitous protein found in all multicellular organisms and is the most abundant component of extracellular matrix of many tissues in humans (32, 33). It is a major constituent of tendons, ligaments, and organic matrix in bone and dentin, and is present in the extracellular matrix of skin, arteries, and cartilage. Collagen is produced as long, thin molecules that align themselves into highly ordered one-dimensional (1D) fibrous arrays that are crosslinked to provide vertebrate tissues with high tensile strength (34, 35). Collagen has high mechanical strength but poor elasticity (36); it is 1091 times stiffer but 12 times less extensible than elastin (37). Its tensile strength is 0.12 GPa, which is 60 times stronger than that of elastin (37). At present, an accepted notion is that the role of collagen fibers in the tissue is to provide a scaffold for elastic fibers (36). Collagen is rich in glycine (33 ± 1.3)%, proline (11.8 ± 0.9)%, alanine (10.8 ± 0.9)%, hydroxyproline (9.1 ± 1.3)%, and glutamic acid (7.4 ± 10)%, which together constitute ∼72% of the protein (38).

Natural abundance ¹³C CPMAS NMR spectra were recorded and shown in Figs. S2 and S3 to investigate possible structural differences in the upper vaginal wall collagen samples studied (virgin, multiparous, para-1 (2 days postpartum), and para-1 (14 days postpartum)). Fig. S2, A–D, illustrates the aliphatic regions. Fig. S3, A–D, shows the carbonyl region of the spectra of the different samples studied. Among all the samples, CPMAS spectra were similar in terms of peak positions and overall appearance, and no significant chemical shift differences were observed. The chemical shift assignments of all the peaks are shown in Fig. 1 (CPMAS NMR spectrum of multiparous rat measured in this study) and the chemical shift values of all the samples are tabulated in Table 1. As the spectral features are very similar to previously reported collagen spectra (obtained from different materials such as collagen parchment, skin, bone, cartilage, tendons, and collagen-like peptides), the assignments of all the peaks were performed following previously published data reported in the literature (11, 16, 17, 18, 39, 40, 41, 42, 43).

Figure 1.

Cross-polarization ¹³C MAS NMR spectra of the multiparous rat tissue at 300 K. No peaks were resolvable from the range of 100–165 ppm. The spectrum was acquired by accumulating 10,000 scans at 10-kHz spinning speed at a magnetic field of 17.61 T. Chemical shift assignments are tabulated in Table 1.

Table 1.

Tabulated Results of the Chemical Shifts Observed at 300 K for the Virgin, Multiparous, para-1 Rat Upper Vaginal Wall Collagen

Figure Notation	Observed Chemical Shift				Assignment
Virgin	Multiparous	para-1 (2 Days Postpartum)	para-1 (14 Days Postpartum)
a	71.4	71.4	72.1	—	Cγ-Hyp
b	67.2	67.2	67.2	67.2	Cβ-Thr
c	62.0	62.0	62.4	62.0	Cβ-Ser
d	60.1, 59.5	60.1, —	60.6, 59.7	60.5, 59.2	Cα-Pro, Cα-Thr, Cα-Hyp, Cα-Phe
e	56.2	56.0	56.0	56.0	Cδ-Hyp,a Cα-Leu
f	54.4	54.6	54.8	54.3	Cα-Glu, Cα-Asp, Cα-Ser
g	52.6	—	—	52.6	Cα-Arg
h	49.8	49.8	49.8	50.3	Cα-Ala
i	48.7	48.0	48.5	48.9	Cδ-Pro
j	43.1	43.1	43.2	43.2	Cα-Gly
k	42.0	42.3	41.9	41.9	Cδ-Arg
l	40.2	40.2	40.4	40.4	Cβ-Leu
m	38.2	38.4	38.5	38.4	Cβ-Asp, Cβ-Leu, Cβ-Phe
n	37.4	37.2	37.2	36.9	Cβ-Hyp
o	34.2	34.4	34.5	34.5	Cβ-Ile
p	30.3	30.3	30.6	30.5	Cβ-Pro, Cβ-Lys, Cβ-Val
q	28.1	28.1	28.0	28.0	Cγ-Glu, Cβ-Arg, Cβ-Val
r	27.2	27.4	27.3	27.4	Cβ-Val
s	25.1	25.1	25.3	25.3	Cγ-Pro, Cβ-Glu, Cγ-Leu
t	23.3	23.3	23.5	23.4	Cγ-Arg, Cγ-Leu, Cδ-Leu
u	19.4	19.4	19.6	19.8	Cγ-Val, Cγ-Thr
v	17.1	17.3	17.3	17.4	Cβ-Ala
w	15.5	15.5	15.9	15.8	Cγ2-Ile
x	12.0	12.0	11.3	12.0	Cδ-Ile

^aThe C_δ Hyp chemical shift range has been reported to lie between 54.3 and 58.3 ppm (18, 40).

Here, the chemical shift assignments of the major amino acids (Gly, Pro, Ala, and Glu) present in collagen were taken into account, as these constitute 72% of the total amino acids. In the spectra, the C_γ Hyp peaks appear at 71.4 ppm for the multiparous rats and the position is fairly constant for virgin and para-1 (2 days postpartum) cohorts (Fig. S2, A–C). However, this peak is not visible (or masked by noise) in the para-1 (14 days postpartum) sample (Fig. S2 D). It is to be noted that all the experiments were performed on collagen samples that had undergone trauma during pregnancy and parturition. As a result, changes in dynamics of side-chain moieties may be expected (see ¹³C T₁), which, in turn, may alter the ¹³C signal intensity, as the experiments were performed using cross polarization. However, additional experiments were performed at higher magnetic field values (21.10 T) to resolve the Hyp signal in para-1 (14- and 2-days) samples. This signal is present in both samples (data not shown) and chemical shift reported was the same as that observed at 17.61 T. The region from ∼54 to 60 ppm belongs to several amino acids including two major constituents, e.g., C_α-Pro, C_α-Hyp. In this region, no major changes were observed. The C_α Gly peak, appearing at 43 ppm, is well resolved in the case of virgin and multiparous rats. However, in the case of para-1 (2 days postpartum) and para-1 (14 days postpartum), the C_α Gly signal is not well resolved, which may have resulted from a change in collagen content, and/or dynamics discussed further below (see Table 5).

Table 5.

Collagen Content of Rat Upper Vaginal Wall for the Virgin, Multiparous, and para-1 Cohorts Determined by Spectrophotometry

Sample	Total Collagen (%)
Virgin (n = 2)	83.6 ± 1.87
Multiparous (n = 2)	68.3 ± 0.61
para-1 (2 days) (n = 3)	63.1 ± 0.32
para-1 (14 days) (n = 3)	54.1 ± 0.31

n is the number of biological replicas used for this study, and the percentages shown are normalized to the dry sample weight before hydrolysis. Collagen content was measured per unit dry sample mass.

The tissues used for all NMR experiments also contained elastin in weak concentration (∼1.5% by weight) (44). Previous NMR studies on hydrated elastin showed that at room temperature the ¹³C signal intensities obtained by ¹H-¹³C cross polarization are relatively weak, and only a few peaks are resolvable (45, 46). The inefficiency of the cross-polarization method for elastin arises from its highly mobile nature in a hydrated state. The contribution of the ¹³C signal in our collagen ¹H-¹³C CPMAS spectra from elastin is therefore very small. Lastly, we note that in the preparation of the samples, the tissues were exposed to ethanol briefly. Previous studies of the exposure of ethanol in the ¹H NMR spectrum of collagen inside hard tissue (e.g., cortical bone) have been noted (47), and one may rightly inquire if the ethanol treatment could alter the water content of our samples. Our spectra do not appear as broad and unresolved as those of dehydrated collagen (see Fig. 1 of Rai and Sinha (48)). In contrast our spectra, which were acquired using a specialized rotor with a center packing system to prevent dehydration during spinning, appear well resolved and similar to the case of hydrated collagen (see Fig. 1 of Rai and Sinha (48)).

Molecular motions may be generally described in terms of autocorrelation functions. The spectral density function, ${\mathcal{J}}_n\left(\omega \right)$ is the Fourier transform of the autocorrelation function, which is defined as

$${\mathcal{J}}_n\left(\omega \right)=\frac{1}{\sqrt{2\pi }}{\int }_0^{\infty }{\mathcal{G}}_n\left(\tau \right)e^{-i\omega \tau }d\tau n=\mathrm{0,1,2},\dots .$$ (1)

Assuming a single correlation time, the spectral density function may be written as

$${\mathcal{J}}_n\left(\omega \right)=\frac{2}{5}\left(1-y^2\right)\frac{{\tau }_c}{1+{\left({\tau }_c\omega \right)}^2},$$ (2)

where y is the order parameter that describes the amplitude of motion and ${\tau }_c$ is the correlation time of the ¹H-¹³C internuclear vectors.

The spin lattice relaxation time in a rotating frame for two spins I, S in terms of spectral density functions is given by (49)

$$\frac{1}{T_{1\rho }^{IS}}={\mu }_S^2\left\{\begin{array}{l}\frac{1}{2}\left[{\mathcal{J}}_0\left({\omega }_I-{\omega }_S\right)+3{\mathcal{J}}_1\left({\omega }_I\right)+6{\mathcal{J}}_2\left({\omega }_I+{\omega }_S\right)\right]\\ +3{\mathcal{J}}_1\left({\omega }_S\right)+\frac{1}{3}\left[{\mathcal{J}}_0\left({\omega }_e-2{\omega }_R\right)+{\mathcal{J}}_0\left({\omega }_e+2{\omega }_R\right)\right]\\ +\frac{2}{3}\left[{\mathcal{J}}_0\left({\omega }_e-{\omega }_R\right)+{\mathcal{J}}_0\left({\omega }_e+{\omega }_R\right)\right]\end{array}\right\}.$$ (3)

The spin lattice relaxation time $T_1^{IS}$ is given by

$$\frac{1}{T_1^{IS}}={\mu }_S^2\left[{\mathcal{J}}_0\left({\omega }_I-{\omega }_S\right)+3{\mathcal{J}}_1\left({\omega }_I\right)+6{\mathcal{J}}_2\left({\omega }_I+{\omega }_S\right)\right].$$ (4)

In the above expressions, ${\omega }_e$ is the frequency of the spin-locking field (applied to ¹³C nuclei); ${\omega }_R$ is the spinning frequency of the rotor, ${\mu }_s^2=\left({\gamma }_I^2{\gamma }_S^2{\hslash }^2/4r^6\right)$; ${\gamma }_I$ and ${\gamma }_S$ are the gyromagnetic ratios for I and S spins; and ${\omega }_I$ and ${\omega }_S$ are the Larmor frequencies of the ¹H and ¹³C spins, respectively. The dynamical characteristics of the system can be described by measuring the average rotational correlation times of the carbon-proton internuclear vectors. The ¹³C T_1ρ relaxation time is often in the range of μs to ms, whereas the ¹³C T₁ is much larger, and may be several seconds long. As the relaxation time T_1ρ has a different dependence on the spectral density functions than T₁, it is sensitive to different timescales of motion. We note that the spin-spin contribution from the thermal coupling between Zeeman carbon and proton dipolar reservoirs in the observed T_1ρ values is usually not negligible. These contributions may result in an altered relaxation pathway and influence the measured relaxation times because the dipolar contribution depends on the ratio between spin-lock and local dipolar field, which is readily suppressed by either setting the high spin-lock frequency or by implementing high spinning speeds (50, 51, 52). In this study, the measurements were performed at 10-kHz spinning to avoid possible denaturation of collagen as a result of heating. Lastly, we note that the above equations assume a simple form of the spectral density with only one correlation time. Other models, such as the model-free approach, make use of two or more correlation times to characterize dynamics (53). We note that determining the exact form of the spectral density, or the constant ${\mu }_s^2$ and the order parameter y, requires several measurements (e.g., at different magnetic fields, for example) and is beyond the scope of this work. With the above assumptions, we qualitatively discuss differences in the correlation times, determined by the ratio of two relaxation times from the above equations.

Tables 2 and 3 give the results of the ¹³C T_1ρ relaxation times measured at two different locking fields for different samples used in this study. From Table 2 it is clear that some of the peaks such as, C_γ-Val, C_γ-Thr, C_γ-Glu, C_β-Arg, and C_β-Val showed significant changes in their correlation times (τ_c). For virgin rats the measured correlation time for the C_γ-Val or C_γ-Thr is (5.57 ± 1.77) μs and this value is smaller (0.91 ± 0.16) μs in multiparous rats. Similar trends are observed for C_γ-Glu, C_β-Arg, or C_β-Val; here the correlation time reduces from (4.57 ± 1.05) μs (virgin rats) to (1.87 ± 0.09) μs (multiparous rats), respectively. This measurement indicates that these moieties appear to be in a more mobile environment in the multiparous rats. The correlation time of (4.07 ± 0.81) μs was measured for the C_γ-Val or C_γ-Thr in para-1 (2 days postpartum) rats (Table 3), which is close, but somewhat smaller than the value observed in virgin rats. This observation indicates that after parturition (2 days postpartum), C_γ-Val or C_γ-Thr resides in a less mobile state compared to virgin rats. However, it was not possible to measure τ_c for these peaks in the para-1 (14 days postpartum) rats. A different scenario is observed for the C_γ-Glu, C_β-Arg, and C_β-Val, where the correlation time observed in para-1 (2 days postpartum) and para-1 (14 days postpartum) rats is close to the value observed in multiparous rats. For major amino acids, e.g., Gly and Hyp, no marked changes in the correlation time are observed. Changes are also observed in the correlation time for C_γ-Pro, C_β-Glu, or C_γ-Leu (2 days: 2.48 ± 0.41 μs, 14 days: 1.37 ± 0.43 μs) and C_α-Glu, C_α-Asp, or C_α-Ser (2 days: 5.22 ± 1.95 μs, 14 days: 2.47 ± 0.71 μs). The measured para-1 (2 days postpartum) correlation times appear to overlap slightly with the measurements in multiparous rats C_γ-Pro, C_β-Glu, or C_γ-Leu (2.75 ± 0.39) μs and C_α-Glu, C_α-Asp, or C_α-Ser (4.08 $\pm$ 0.71) μs. These observations suggest that remodeling of the ECM resulting from pregnancy and/or parturition alters the dynamical characteristics of specific collagen side-chain groups. The data also suggests that the value for para-1 (14 days postpartum) after parturition return somewhat to those observed in virgin rats C_α-Glu, C_α-Asp, and C_α-Ser (3.21 $\pm$ 0.95) μs. Additional changes may have resulted from alterations leading to structural heterogeneity resulting in overlap of peaks or a redistribution of water in the extracellular matrix, discussed further below, which may not be completely reversible. Recently, we have observed marked changes in the dynamics of elastin after in vitro exposure to glucose by ¹³C NMR relaxation measurements (54). Glucose exposure causes elastin to stiffen (as measured by tensile testing) and larger H-C correlation times were observed in comparison to samples in water only. Moreover, we observed evidence of glucose interacting with elastin, as a cross-polarized glucose signal was observed in our ¹³C NMR spectra. In our collagen spectra, the broad signal in the range of 70–80 ppm may result from sugars in the tissue (e.g., due to glycosylation). Thus, the effects of glucose on the samples studied may also be a contributor to the alterations observed in the correlation times, and may require further investigation.

Table 2.

Tabulated ¹³C T_1ρ Relaxation Times for Virgin and Multiparous Rat Upper Vaginal Wall Collagen Samples at 300 K

Assignment	Virgin			Multiparous
T1ρ (ms) (50.00 kHz)	T1ρ (ms) (25.00 kHz)	τc (μs)	T1ρ (ms) (50.00 kHz)	T1ρ (ms) (25.00 kHz)	τc (μs)
Cγ2-Ile	5.37 ± 0.79	3.24 ± 0.30	3.73 ± 1.56	—	—	—
Cβ-Ala	5.04 ± 0.65	3.42 ± 0.31	3.01 ± 1.35	—	—	—
Cγ-Val, Cγ-Thr	6.50 ± 0.98	2.98 ± 0.24	5.57 ± 1.77	5.35 ± 0.41	5.05 ± 0.49	0.91 ± 0.16
Cγ-Pro, Cβ-Glu, Cγ-Leu	2.20 ± 0.10	1.44 ± 0.07	3.23 ± 0.58	2.20 ± 0.06	1.56 ± 0.06	2.75 ± 0.39
Cγ-Glu, Cβ-Arg, Cβ-Val	2.40 ± 0.19	1.27 ± 0.07	4.57 ± 1.05	1.90 ± 0.09	1.56 ± 0.10	1.87 ± 0.09
Cβ-Pro, Cβ-Val, Cβ-Lys	2.13 ± 0.16	1.99 ± 0.06	1.01 ± 0.70	2.00 ± 0.09	1.47 ± 0.12	2.53 ± 0.78
Cβ-Ile	4.40 ± 0.90	2.35 ± 0.13	4.51 ± 1.74	—	—	—
Cβ-Leu	3.32 ± 0.39	2.61 ± 0.13	2.14 ± 0.96	2.23 ± 0.14	1.70 ± 0.13	2.32 ± 0.84
Cα-Gly	1.77 ± 0.16	1.19 ± 0.12	3.07 ± 1.20	1.05 ± 0.05	0.92 ± 0.09	1.46 ± 0.91
Cα-Glu, Cα-Asp, Cα-Ser	2.04 ± 0.16	1.34 ± 0.10	3.21 ± 0.95	2.26 ± 0.10	1.29 ± 0.08	4.08 ± 0.71
Cα-Pro, Cα-Hyp, Cα-Thr, Cα-Phe	1.99 ± 0.15	1.12 ± 0.09	4.17 ± 1.06	2.31 ± 0.10	1.18 ± 0.08	4.82 ± 0.79
Cγ-Hyp	2.07 ± 0.26	1.31 ± 0.05	3.43 ± 1.00	2.31 ± 0.21	1.42 ± 0.17	3.61 ± 1.40

Correlation times were determined using Eqs. 2 and 3 as described in the text. Two different locking fields are shown in parentheses.

Table 3.

Tabulated ¹³C T_1ρ Relaxation Times at Two Different Locking Fields for para-1 Rat Upper Vaginal Wall Tissue Samples at 300 K

Assignment	para-1 (2 Days Postpartum)			para-1 (14 Days Postpartum)
T1ρ (ms) (50.00 kHz)	T1ρ (ms) (25.00 kHz)	τc [μs]	T1ρ (ms) (50.00 kHz)	T1ρ (ms) (25.00 kHz)	τc (μs)
Cγ2-Ile	6.49 ± 0.76	5.01 ± 0.51	2.24 ± 1.13	5.57 ± 0.31	3.63 ± 0.75	3.25 ± 1.85
Cγ-Val, Cγ-Thr	6.75 ± 0.55	3.86 ± 0.17	4.07 ± 0.81	—	—	—
Cγ Pro,Cβ-Glu,Cγ-Leu	2.44 ± 0.06	1.81 ± 0.08	2.48 ± 0.41	2.07 ± 0.06	1.84 ± 0.07	1.37 ± 0.43
Cγ-Glu, Cβ-Arg, Cβ-Val	1.97 ± 0.07	1.74 ± 0.09	1.41 ± 0.55	1.77 ± 0.07	1.36 ± 0.11	2.27 ± 0.72
Cβ-Leu	—	—	—	3.13 ± 0.08	2.48 ± 0.25	2.09 ± 0.77
Cα-Gly	1.56 ± 0.06	1.42 ± 0.28	1.20 ± 0.40	—	—	—
Cα-Glu, Cα-Asp, Cα-Ser	2.18 ± 0.23	1.05 ± 0.15	5.22 ± 1.95	2.10 ± 0.10	1.56 ± 0.11	2.47 ± 0.71
Cα-Pro, Cα-Hyp, Cα-Thr, Cα-Phe	2.07 ± 0.18	1.30 ± 0.16	3.48 ± 1.39	1.75 ± 0.04	1.13 ± 0.05	3.31 ± 0.42
Cγ-Hyp	1.59 ± 0.07	1.10 ± 0.06	2.89 ± 0.69	2.04 ± 0.14	1.81 ± 0.12	1.38 ± 0.87

Correlation times were determined using Eqs. 2 and 3 as described in the text. Two different locking fields are shown in parentheses.

The ¹³C spin lattice T₁ relaxation times for virgin and multiparous rats are presented in Table 4. Previous studies of ¹³C spin lattice relaxation times for the collagen fibrils and model peptides showed that different amino acids present in collagen exhibit different relaxation times (38). Referring to Table 4, small changes in T₁ values are observed for some of the moieties such as the C_α-Glu, C_α-Asp, or the C_α-Ser, C_δ-Hyp, or C_α-Leu peaks. The T₁ values appear slightly larger in multiparous rats in comparison to virgin rats for the C_α-Glu, C_α-Asp, or C_α-Ser peaks. However, for C_δ-Hyp or C_α-Leu peaks, the T₁ observed in the multiparous rat tissues (5.32 ± 0.41) μs appears smaller in comparison to the measured value in the virgin rat tissues (7.44 ± 1.11) μs. As we discuss below, we observed evidence of trauma to the tissue after pregnancy, as revealed by histology (Fig. 2). The changes in the relaxation times may arise from a redistribution of water due to this trauma from pregnancy, which may alter the dynamical characteristics of these particular side chains. For other moieties (Table 4), no significant differences in T₁ values were observed between the multiparous and virgin rats. T₁ values in human skin collagen at 300 MHz have been previously measured (38). Using our measured correlation times, we back-calculated the expected T₁ relaxation time at a 300 MHz Larmor frequency and found that T₁ values for the upper vaginal wall collagen were smaller than those observed in human skin. These differences may arise from advanced glycation end-products that are known to make collagen stiffer (55) and may alter the NMR relaxation times.

Table 4.

Tabulated ¹³C T₁ relaxation Times of Rat Upper Vaginal Wall Collagen for the Virgin and Multiparous Rats at 300 K

Assignment	Virgin T1 (s)	Multiparous T1 (s)
Cγ-Pro, Cβ-Glu, Cγ-Leu	1.52 ± 0.08	1.44 ± 0.05
Cβ-Pro, Cβ-Val, Cβ-Lys	1.40 ± 0.14	1.42 ± 0.08
Cβ-Leu	1.64 ± 0.19	1.31 ± 0.13
Cα-Gly	2.79 ± 0.35	2.34 ± 0.21
Cα-Glu, Cα-Asp, Cα-Ser	5.71 ± 0.64	6.63 ± 0.54
Cδ-Hyp, Cα-Leu	7.44 ± 1.11	5.32 ± 0.41
Cα-Pro, Cα-Hyp, Cα-Thr, Cα-Phe	7.07 ± 0.97	6.85 ± 0.53
Cγ-Hyp	4.49 ± 1.19	3.06 ± 0.34

Figure 2.

Representative histological image (at 100×) of the upper vaginal wall of the (A) multiparous and (B) virgin Sprague-Dawley rat sectioned at 10 μm. Collagen fibers are stained pink, smooth muscle is stained yellow, and elastic fibers are stained black. Histological images show possible collagen fiber disassociation with smooth muscle and a change in density of collagen fibers in multiparous and virgin rats. To see this figure in color, go online.

Collagen content in upper vaginal wall tissues

During pregnancy, the remodeling of the vaginal wall and other pelvic floor connective tissues occurs throughout fetal development and involution (56). This remodeling alters the composition and concentration of the extracellular matrix; there is a rapid increase in the weight of the uterus as well as the size due to the deposition of collagen and elastin. In the case of rats, studies have shown that the increase in the weight of the uterus is six- to eight-fold compared to non-pregnant uteri (1, 34); in humans, it is 11-fold (36); and in ewe, it is 14-fold (57). There is a rapid formation of collagen in the uterus during pregnancy. However, after parturition, collagen content has been shown to reduce rapidly. Collagen formation is smallest on the placental side and maximal in the region of greatest uterine extension (36). Harkness and Harkness (1) showed that during the first 10 days of pregnancy, collagen content increases faster than the changes in weight and thereafter there is an observed decrease in the rate of increase in rat uteri. Soon after parturition, weight and collagen content of the uterus falls precipitously and reaches a value below non-pregnant controls. The increase in collagen content during pregnancy is fourfold per fetus (58, 59) in rats, whereas in humans there is a 6.8-fold increase (2). In rats, this increase has been shown to depend on various factors such as the number of fetuses in the uterus, the age, and the weight of the animal (1, 59, 60, 61). After parturition, the degradation of collagen is very rapid, and is complete within 3–4 days in rats (59), whereas in humans, it completes within three weeks (2).

The collagen content (per milligram lyophilized tissue of upper vaginal wall) for virgin, multiparous, para-1 (2 days postpartum), and para-1 (14 days postpartum) rats are presented in Table 5. Two biological replicas for the virgin and multiparous cohorts, and three for para-1 (2 days postpartum) and para-1 (14 days postpartum) cohorts were used for this study. Referring to Table 5, we found that the virgin rat tissues contained higher collagen content (83%) compared to the other three samples. This observation was expected, as after delivery collagen content has been reported to reduce below the non-pregnant value (1). It has been reported that collagen content in the reproductive tract increases rapidly at the 10th day of pregnancy, which reaches the highest level just before parturition and decreases precipitously after parturition that continues approximately up to 16 days (1). However, the rapid degradation of collagen that starts after parturition varies in different regions of the reproductive tract. It has been found that in the uterine horns this rapid degradation starts 12 h postpartum whereas the collagen content of the cervix does not begin to degrade until 24–48 h postpartum (59). According to observations made in the work by Harkness and Moralee (59), we would expect that the collagen content should be higher in para-1 (2 days postpartum) rats than in the virgin rats; this was not observed in our study (Table 5). It is to be noted that in our analysis, we have used the upper vaginal wall. Therefore, this region might have a different timescale for collagen degradation. Our study shows that the degradation is much faster and collagen content returns below the non-pregnant control value in two days. However, this degradation continues and the collagen content is even lower in para-1 (14 days postpartum) rats.

Additionally, in our measurements, the multiparous rats showed higher collagen content values compared to para-1 (2 days postpartum) and para-1 (14 days postpartum) cohorts. Several factors might affect the collagen content in the multiparous rats such as age, the number of fetuses in the uterus, parity, and weight (1, 59, 60, 61, 62). With increase in age and parity, collagen deposition increases, which might explain the higher values observed in the multiparous rats. Previously, most of the collagen content in rat uteri was reported based on the wet weight of the upper vaginal wall tissue, but in our study, we calculated the collagen content based on the dry weight of the tissue. Smith and Kaltreider (62) calculated the percentage of collagen in rat uteri based on the dry weight for the different ovarian cycles and found the highest collagen content in the metestrus phase 28.6%. In contrast, in the cervix of the rat uterus, the collagen content was reported to be 50% based on the sample dry weight (63). Thus, these results clearly indicate that the collagen content varies in different parts of the reproductive tract and may explain the differences between previously reported values and the numbers reported in this work.

Histological analysis

Representative histological images of the stained upper vaginal wall from multiparous and virgin rats at 100× magnification are shown in the parts of Fig. 2, A and B, respectively. In these figures, collagen fibers are stained pink, smooth muscle is stained yellow, and elastic fibers are stained black. A qualitative assessment of the images we studied (150 in total) revealed that collagen fibers are more sparsely distributed and disarrayed in multiparous rats compared to the virgin rats. Moreover, collagen fibers in multiparous rats appeared slightly thicker than those of virgin rats, but no quantitative measurements could be made. Most notably, the analysis of all the images we took indicated that there is a signature of collagen fiber dissociation with smooth muscles in multiparous rats but no evidence of collagen fiber fragmentation could be observed. These results are in agreement with the previous observations where the collagen fiber in cervix of pregnant rat appears to be disarrayed and disassociated with smooth muscles (64). Previously, we reported changes in the elastic fibers of virgin and multiparous rats; an increase in elastic fiber tortuosity and decrease in elastic fiber length was observed when comparing multiparous and virgin rat cohorts (5). We speculate that the changes in collagen density and thickness, as well as the observed dissociation of collagen fibers with smooth muscle in multiparous rats, may be an additional contributor to the decrease in stiffness of the tissue as observed in previous studies (4, 5, 65, 66). However, these changes appear small in comparison to changes observed in elastic fibers after pregnancy and parturition (5).

We also developed software (available upon request) to calculate the percentage of collagen content in the virgin and multiparous rats; we crudely estimated the values to 64.5 ± 11.1 and 60.3 ± 22.4, respectively, based on the pixel color. These values varied from the more precise spectrophotometric measurements discussed above (Table 5) and the differences between the two methods may arise from the threshold for distinguishing collagen from elastin and smooth muscle, or the particular sections we used from histology.

Dynamics and distribution of water in rat upper vaginal wall tissue samples

The dominant NMR relaxation pathway for the ²H nucleus is quadrupolar, because ²H-¹H and ²H-²H nuclear dipolar interactions are negligible. The T₁ and T₂ relaxation times are given by (67)

$$\frac{1}{T_1}=\frac{3{\pi }^2}{20}{C}_{Q_{\text{eff}}}^2\left[2J\left({\omega }_0\right)+8J\left(2{\omega }_0\right)\right],$$ (5)

$$\frac{1}{T_2}=\frac{3{\pi }^2}{20}{C}_{Q_{\text{eff}}}^2\left[3J\left(0\right)+5J\left({\omega }_0\right)+2J\left(2{\omega }_0\right)\right].$$ (6)

In the above expressions, $C_{Q_{\text{eff}}}=\left(e^2{q}_{\text{eff}}Q/h\right)\Gamma$ is the effective quadrupolar coupling constant, and Γ is a motional averaging parameter. The spectral density is defined by $J\left(m\omega \right)=\left({\tau }_c/1+{\left(m\omega {\tau }_c\right)}^2\right)$, where ${\tau }_c$ denotes the correlation time of the nuclear spin and surrounding electric field gradient, which is intramolecular in origin. Using the measured T₁ and T₂, the correlation times were calculated using Eqs. 5 and 6.

The ²H 2D ILT T₁-T₂ relaxation methods were employed to investigate the dynamics and distribution of water in each of the upper vaginal tissues studied. However, we also performed 1D ²H NMR experiments though no spectroscopic information was obtainable (²H NMR spectrum of the ²H₂O hydrated multiparous rat tissues is shown in Fig. S4). Fig. 3 shows 2D ILT T₁-T₂ correlation maps of ²H₂O hydrated samples acquired at 25°C. In the 2D ILT map of the virgin rat tissue (Fig. 3 A), three water reservoirs (labeled a–c) apart from bulk (T₁ and T₂ value of ∼403 and 352 ms) water were resolved. The peak denoted bulk water and exhibits isotropic motion as T₁ ≈ T₂, whereas the other peaks (labeled a–d) exhibit unequal relaxation times and are indicative of anisotropic motion that may arise from the restrictions within the collagen matrix. Each peak corresponds to a different environment of water molecules in the samples indicating different dynamical characteristics of water. In the para-1 (2 days postpartum), para-1 (14 days postpartum), and multiparous cohorts, a fourth water environment (labeled d in Fig. 3) is observed. To perform these experiments, upper vaginal wall tissue was used; this experimental technique does not allow assigning the reservoirs of water to a particular structural motif of collagen, or association with collagen or elastin. Thus, all observable peaks may correspond to water present near collagen, elastin, or other ECM components.

Figure 3.

2D ILT map of the ²H T₁-T₂ results from (A) virgin, (B) multiparous, (C) para-1 (2 days postpartum), and (D) para-1 (14 days postpartum) rats. In all the samples except virgin rats, four peaks are well resolved (ignoring bulk water). This fourth peak labeled d may arise from changes in the extracellular matrix that may include changes in collagen, elastin, or both. The color bar to the right (in each figure) represents the signal intensity on logarithmic scale. The numerical values of the T₁ and T₂ and correlation times are provided in Table S1. To see this figure in color, go online.

From our histological analysis discussed above, collagen fibers appear to be dissociated with the smooth muscle, and changes in the collagen density or thickness are observed among virgin and multiparous rat cohorts. Thus, the appearance of water reservoir d in Fig. 3, B–D, may presumably arise from a redistribution of water or changes in collagen and/or elastin in the upper vaginal tissue/extracellular matrix after pregnancy and parturition. Lastly, the relaxation measurements and correlation times of peaks a–d for all the samples studied are presented in Table S1.

We further attempted to correlate the measured relaxation and correlation times with the previously reported values in bovine nuchal ligament elastin samples (24, 68) and commercially available rat tail collagen. The T₁-T₂ ILT map of ²H in ²H₂O hydrated elastin showed four water reservoirs including the bulk water (similar to our virgin rat samples) and were denoted α₁, α₂, β, and γ. In this study, we only considered peaks β and γ for comparison, as α₁ and α₂ correspond to the bulk water, and water between the elastin fibers, respectively (24, 68). The measured relaxation times for peak β (T₁ = 66 ms and T₂ = 19.9 ms) and γ (T₁ = 34 ms and T₂ = 5.3 ms) differ with the values observed for peaks b and c in this study for virgin rat sample. Additionally, T₁-T₂ experiments were also performed in pure rat tail collagen (data not shown) and the measured T₁, T₂, and τ_c values did not correlate with those observed in the tissues we used, which are composed of collagen and elastin. These observations indicate that water in the upper vaginal wall tissue is experiencing a different environment and exhibit different dynamics than water in pure collagen and in pure elastin samples. However, in this study no observable differences in the correlation times of peaks a–d across the samples studied based on pregnancy and parturition were observed. The measured relaxation times and correlation times across the samples were found to be similar within our experimental uncertainty, indicating that there are changes in the ECM that alter the redistribution of water, while preserving the structure of existing domains.

Conclusions

This work reports on a study of the structural and dynamical modifications of rat reproductive tract collagen with respect to parity. ¹³C CPMAS NMR spectra indicated no observable chemical shift differences despite the significant remodeling of vaginal wall and pelvic floor connective tissues during pregnancy and parturition. These results showed that the structure of collagen is preserved after pregnancy. However, some marked differences were observed in the dynamics of some amino acids. An increased mobility in some of the amino-acid side-chain moieties (e.g., C_γ-Val, C_γ-Glu, etc.) in the multiparous rats was observed in comparison to the virgin rats. For some of the amino acids such as the C_α-Glu, C_α-Ser, etc., the T₁ value was larger in the multiparous rat, in comparison to the virgin rat. However, we observed a decrease in T₁ value for the C_δ-Hyp or C_α-Leu peaks. Moreover, ²H T₁-T₂ measurements show the presence of four water reservoirs in the case of multiparous, para-1 (2 days postpartum), and para-1 (14 days postpartum) while only three reservoirs were observed in the virgin rat. The redistribution of the water in the extracellular matrix due to pregnancy and/or parturition may be responsible for the appearance of an additional water reservoir, which might have influenced the dynamics of some of the amino-acid side chains. Spectrophotometric measurements showed that the virgin rat had the highest percentage of collagen content among the samples. A signature of the collagen fiber dissociation, as well as possible changes in the density and thickness of the collagen fiber, were observed in multiparous and virgin rat samples. Although much smaller in concentration, elastin is also a part of extracellular matrix in the uterus that undergoes remodeling during parturition and various stages of pregnancy. This study highlights the consequences of pregnancy and parity on collagen structure, and a detailed study of the reproductive tract elastin is currently underway in our laboratory.

Author Contributions

B.D. and F.G. performed measurements and analysis of experimental data and contributed to writing the article; S.H. took histological images and contributed in writing; K.T.D. performed sample preparation and contributed to writing the article; and G.S.B. coordinated the article, performed measurements and analysis, and contributed to writing the article.

Editor: H. Jane Dyson.