Extracellular matrix rigidity modulates human cervical smooth muscle contractility - new insights into premature cervical failure and spontaneous preterm birth

Abstract

Spontaneous preterm birth (sPTB), a major cause of infant morbidity and mortality, must involve premature cervical softening/dilation for a preterm vaginal delivery to occur. Yet, the mechanism behind premature cervical softening/dilation in humans remains unclear. We previously reported the non-pregnant human cervix contains considerably more cervical smooth muscle cells (CSMC) than historically appreciated and the CSMC organization resembles a sphincter. We hypothesize that premature cervical dilation leading to sPTB may be due to 1) an inherent CSMC contractility defect resulting in sphincter failure and/or 2) altered cervical extracellular matrix (ECM) rigidity which influences CSMC contractility. To test these hypotheses, we utilized immunohistochemistry to confirm this CSMC phenotype persists in the human pregnant cervix and then assessed in vitro arrays of contractility (F:G actin ratios, PDMS pillar arrays) using primary CSMC from pregnant women with and without premature cervical failure (PCF). We show CSMC from pregnant women with PCF do not have an inherent CSMC contractility defect but that CSMC exhibit decreased contractility when exposed to soft ECM. Given this finding, we used UPLC-ESI-MS/MS to evaluate collagen crosslink profiles in cervical tissue from non-pregnant women with and without PCF and found women with PCF have decreased collagen crosslink maturity ratios, which correlates to softer cervical tissue. These findings suggest having soft cervical ECM may lead to decreased CSMC contractile tone and a predisposition to sphincter laxity that contributes to sPTB. Further studies are needed to explore the interaction between cervical ECM properties and CSMC cellular behavior when investigating the pathophysiology of sPTB.

Introduction

Preterm birth (PTB) still complicates about 1 in 10 pregnancies in the U.S. [1,2]. Although various etiologies exist, the final common pathway to a premature vaginal delivery must involve premature remodeling, softening and dilation of the cervix [3]. Despite decades of research, how the human cervix remodels and dilates prematurely leading to premature cervical failure and sPTB remains unclear largely because human cervical tissue is challenging to obtain during pregnancy. As a result, effective therapeutic interventions to prevent premature cervical failure and sPTB remain limited.

The prevailing paradigm of human cervical tissue structure, established by Danforth in the 1940s, stated the cervix is a homogenous, collagenous structure with minimal cellular content [4,5]. As a result, researchers have focused on how cervical collagen remodels during pregnancy into a compliant cervix that allows for cervical dilation and delivery of a fetus [6–16]. Recently, Yoshida et al reported that the mechanical strength of cervical extracellular matrix (ECM) relies on the type and degree of collagen crosslinking in its collagen network. Specifically, they measured immature (divalent) collagen crosslinks (hydroxylysinonorleucine [HLNL], dihydroxylysinonorleucine; [DHLNL]), mature (trivalent) collagen crosslinks (deoxypyridinoline [DPD], pyridinoline [PYD]) and mechanical properties of cervical tissue at various time points in a mouse pregnancy and found that as the cervix softens, collagen crosslink maturity ratios (ratio of mature [PYD+DPD] to immature crosslinks [HLNL+DHLNL]) decrease [11]. This established an association between cervical collagen crosslink maturity ratios and cervical tissue stiffness.

In 2010, Oxlund et al suggested cervical tissue may be more heterogeneous than previously reported as they found that the density of smooth muscle in the lower half of the cervix increased as one moved from the external os upwards towards the internal os [17]. In 2016, we reported that that area of the internal os (upper aspect of the cervix where the uterus meets the cervix, Fig. 1) in the non-pregnant cervix contains a significant amount of contractile smooth muscle, which is circumferentially oriented around the endocervical canal [18]. These findings led us to question if a sphincter may exist in the upper aspect of the cervix and if premature cervical failure (characterized by funneling or dilation of the upper aspect of the cervix on transvaginal sonogram; Fig 1) may represent sphincter failure [19,20].

Fig. 1.

Transvaginal sonogram image of a normal cervix (top image) that is long and closed at both the internal and external os. The bottom sonogram image shows a cervix that has started to fail prematurely characterized by dilation/funneling of the upper cervix resulting in a short cervix.

Interestingly, the idea of a sphincter at the internal os is not new. In the early 1900s, Aschoff reported that a sphincter may exist at the internal os. However, at that time, the internal os was defined to be located in the isthmus of the uterus [21]. In 1942, Kearns published a study that included analysis of intact uterine specimens (which included the cervix) and noted that the histological border between the cervix and isthmus is not well defined. However, he did note that an internal uterine sphincter of circular smooth muscle existed in the isthmus [22]. Danforth’s seminal article on cervical tissue architecture in the 1940s used this definition of the internal os (which was located in the isthmus of the uterus) and thus did not include this area in his study of the cervix and might explain why he reported the cellular content in the cervix was minimal [4]. Subsequently, in the 1970s, Ferenczy suggested the isthmus and the internal os of the cervix are actually the same structure [23]. Over the next two decades, for reasons that are unclear, the concept of a sphincter in the cervix appeared to be lost to clinicians and researchers who continued to regard the cervix as a mostly collagenous structure with minimal cellular content. The internal os continued to be included as part of the cervix and has been used as a landmark when measuring cervical length by transvaginal ultrasound [19,20]. In the 1990s, the possibility of a contractile body in the human cervix reemerged as studies showed that the human cervix has the ability to contract independently from the uterus [24–31]. Our study in 2016 thus re-discovered that a significant amount of contractile smooth muscle exists in the upper aspect of the cervix in the area of the internal os, which we defined as part of the cervix [18]. More recently, MRI studies have also shown that an “occlusive structure” exists at the level of the internal os of the cervix [32].

Given this history, we again hypothesize that the upper aspect of the cervix (where the cervix meets the uterus) may contain a sphincter and that sphincter failure resulting in premature cervical dilation may be due to several factors. First, it may represent an inherent contractility defect in the individual cervical smooth muscle cell (CSMC). Second, studies in airway and vascular biology have shown that extracellular matrix (ECM) rigidity can influence smooth muscle contractile behavior [33–35]. Specifically, airway smooth muscle cells exhibit enhanced contractility upon exposure rigid ECM compared to when exposed to softer ECM. Given that we know the cervix dramatically softens throughout pregnancy [6–8, 11, 12] we hypothesize that changes in ECM rigidity may influence contractility of the smooth muscle body in the upper aspect of the cervix.

To test these hypotheses, we first sought to use modern immunohistochemical techniques and confirm if the upper aspect of the pregnant human cervix (where the cervix meets the uterus) contains the same smooth muscle content and organization that we observed in the non-pregnant state. Next, we sought to determine if CSMC isolated from women with a history of premature cervical failure exhibit an inherent contractility defect and/or if the type of ECM protein or ECM rigidity influences CSMC contractility. Lastly, if an ECM-CSMC interaction exists, we sought to understand if women with premature cervical failure have softer ECM collagen crosslink profiles compared to women without a history of premature cervical failure.

Materials and Methods

Immunohistochemistry

Using an IRB-approved protocol, informed consent was obtained from healthy pregnant women greater than 18 years old at risk for cesarean hysterectomy due to abnormal placentation. Women with a history of prior cervical surgery (cone biopsy, loop electrosurgical excision procedure) were excluded. In our previous study of non-pregnant human cervical tissue, we amputated the cervix where the uterine arteries inserted into the cervix (which in our study correlated to the area of the internal os) [18]. As a pregnancy approaches full term in humans, the cervix is thought to be pulled up into uterus as the lower uterine segment develops and the cervix effaces. In this study, we sought to keep the location of analysis consistent with our previous study of non-pregnant cervical tissue. As such, after cesarean hysterectomy was performed in the third trimester, the cervix was immediately amputated from the uterus where the uterine arteries inserted into the cervix (as previously described [18]). Three mm axial slices were obtained from the level of the internal and external os and placed in 10% formalin. After 24h, the tissue was transferred to 70% ethanol and paraffin-embedded. Serial 5 μm sections of the tissue were stained for H&E, mature SMC markers (α-smooth muscle actin [SMA], smooth muscle heavy chain 22 [SM22], calponin) and contraction associated proteins (COX2, oxytocin receptor, connexin 43) using a previously published protocol [18]. The antibodies and dilutions were used as listed in Table 1 in Vink et al [18] for α-SMA, SM22, calponin, COX2 and oxytocin receptor. Fluorescent immunohistochemistry was used to assess α-SMA (Sigma, Catalog #C6198, 1:200 dilution) and connexin 43 (Abcam, Catalog #11370, 1:100 dilution) expression using a previously published protocol [18]. These markers were chosen to identify if the cells were mature SMC (that expressed contraction-associated proteins known to be involved in SMC contractility) rather than other cell types such as fibroblasts. For negative controls, primary antibody was omitted, and tissue was incubated in blocking solution overnight. CSMC organization within an entire slice of cervical tissue was appreciated by obtaining 10X and 20X images on a Nikon ECLIPSE E800 microscope (Nikon Inc., Melville, NY) and digitally reconstructing each slice with ZEN software (Zeiss). To approximate the volume fraction of CSMC in the tissue, we calculated the area fraction of CSMC on a tiled slice of cervical tissue from the internal and external os from each patient. Photoshop CS5 (Adobe Systems Inc., San Jose, CA) was used to count the number of SMA-positive pixels in each tiled image as previously described [18]. Specifically, the number of SMA-positive pixels was divided by the total number of pixels in each cervical slice to obtain the percent of positive SMA pixels.

Table 1:

Definitions of Study and Control Groups (A, B and C)

Group	Inclusion Criteria	Used in the following experiments
Study Group A (n=21 patients)	Pregnant women over 18 years of age carrying singleton gestations with a history of premature cervical remodeling leading to sPTB < 32 weeks undergoing history-indicated transvaginal cerclage placement between 11–17 weeks	F:G actin ratio assay (cell lines from each of the 21 patients included)* PDMS pillar arrays (cell lines from 6 of the 21 patients included)*
Control Group A (n=13 patients)	Pregnant women over 18 years old carrying singleton pregnancies without a history of premature cervical shortening/failure or sPTB undergoing voluntary termination of pregnancy between 11–17 weeks.	F:G actin ratio assay (cell lines from 13 patients included)* PDMS pillar arrays (cell lines from 6 of the 13 patients included)*
Study Group B (n=9 patients)	Pregnant women over 18 years of age carrying singleton gestations with a history of premature cervical remodeling leading to sPTB < 32 weeks undergoing physical exam or ultrasound-indicated transvaginal cerclage placement between 17−23+6 weeks.	F:G actin ratio assay (cell lines from 9 patients included)*
Control Group B (n=7 patients)	Pregnant women over 18 years old carrying singleton pregnancies without a history of premature cervical shortening/failure or sPTB undergoing voluntary termination of pregnancy between 17–23+6 weeks.	F:G actin ratio assay (cell lines from 7 patients included)*
Study Group C (n=10 patients)	Non-pregnant women over 18 years old undergoing open abdominal cerclage placement due to a history of prior premature cervical failure resulting in second trimester loss or sPTB < 32 weeks.	Collagen crosslink study (n=10 patients; one biopsy per patient)*
Control Group C (n=10 patients)	Non-pregnant, premenopausal women between 1850 years old undergoing total hysterectomy for benign indications without a history of premature cervical failure or sPTB.	Collagen crosslink study (n=10 patients; one biopsy per patient)*

^*Please refer to Supeplementary Table S3 for patients’ demographic data

Primary CSMC Contractility Studies

Cervical tissue biopsies (one biopsy per patient) were collected using IRB-approved protocols from healthy pregnant women (greater than 18 years old) carrying singleton pregnancies with a history of sPTB < 32 weeks due to asymptomatic premature cervical shortening or dilation who were undergoing history-indicated cerclage between 11–17 weeks (Study Group A) or physical exam/ultrasound-indicated transvaginal cerclage placement between 17–23+6 weeks (Study Group B) (Table 1). Control groups consisted of healthy women (greater than 18 years old) carrying singleton pregnancies with no history of sPTB or premature cervical remodeling who were undergoing voluntary termination of pregnancy between 11–17 weeks (Control Group A) or 17–23+6 weeks gestation (Control Group B) (Table 1). Women with a history of prior cervical surgery (cone biopsy, loop electrosurgical excision procedure) were excluded. Full thickness core cervical tissue biopsies (approximately 0.5–1cm long x 1–2mm wide) were obtained at 12 o’clock on the anterior lip of the upper aspect of the cervix. The biopsies were obtained as perpendicular as possible to the endocervical canal (starting from the outside [vaginal side] of the cervix). Biopsies were obtained using a Miltex biopsy device (Miltex, #33–31) from women in Study Groups A and B after the vagina was cleaned with betadine solution. Biopsies were obtained in Control Groups A and B after the vagina was cleaned with betadine solution and before laminaria insertion using an Achieve Automatic Biopsy Device (Merit Medical, South Jordan, UT). Biopsies were immediately placed in cold Smooth Muscle Growth Medium-2 (SmBm-2) with manufacturer’s recommended additives (Lonza) and promptly transported to the lab on ice for CSMC isolation.

SMC Isolation and Characterization

Primary CSMC cultures from each patient were established by enzymatic dissociation of fresh cervical tissue biopsies using the Worthington Papain tissue dissociation kit (Worthington Biochemical Company). Briefly, under sterile conditions, cervical biopsies were minced then enzymatically dissociated using papain and collagenase. After ovamucoid/albumin separation, isolated CSMC cells were seeded into a 6-well tissue culture plate. CSMC were incubated in SmBm-2 with manufacturer’s recommended additives (Lonza) and an antibiotic/antifungal agent (5X anti-anti, Gibco) overnight then changed to SmBm-2 media without the antibiotic/antifungal agent on day 1 after isolation. To maintain primary muscle phenotype, serial plating was performed, and experiments were restricted to less than 6 passages in culture.

Primary CMSC lines were stained for mature smooth muscle cell markers: α- SMA, SM22, and desmin [18, 36, 37]. These markers were chosen to identify if the cells were mature smooth muscle cell rather than other cell types such as fibroblasts. CSMC were seeded on rat tail type I collagen (dilution 1:100; Corning, Corning, NY)-coated 8-well Falcon Chambered Cell Culture slides and incubated with SmBm-2 media with manufacturer’s recommended additives (Lonza). Once confluent, cells were fixed on ice in 4% paraformaldehyde (PFA) for 15 minutes, permeabolized with 0.1% Triton X-100 in 1X PBS, and incubated with blocking solution (2% bovine serum albumin [BSA], 3% donkey serum in 1X PBS) for 1 hour at room temperature (RT) before overnight incubation with primary antibody (see Supplemental Table S1) at 4°C. Slides were washed with 1X PBS, incubated with fluorescently labeled secondary antibodies (see Supplemental Table S1) for 30 minutes at RT and then washed in 1X PBS and coverslipped with Vectashield with DAPI mounting media (Vector). Images were captured with an Olympus IX83 microscope (Olympus) and edited with Adobe Photoshop. Percent positive staining was calculated using ImageJ software.

Filamentous (F)-actin : Globular (G)-Actin Ratio Assays

To assess if primary CSMC from women with a history of premature cervical failure exhibit an inherent defect in filamentous actin formation compared to similar gestational-age controls without a history of premature cervical failure or sPTB, we performed F:G actin ratio assays (an established indirect measure of SMC contractility) [38, 39]. Immortalized HEK293T cells (ATCC) were used as a negative (non-contractile) control. For each cell line (HEK293T, Study Group A and B, Control Group A and B) 5000 cells were plated on rat tail collagen-coated (dilution 1:100; Corning, Corning, NY) 8-well Falcon Chambered Cell Culture slides and incubated overnight in SmBm-2 media with manufacturer’s recommended additives (Lonza). The following day, cells were washed (1X PBS three times) then incubated in serum and additive-free SmBm-2 media overnight. The following morning, cells were exposed to vehicle (serum and additive-free SmBm-2 media) or 1 μM oxytocin (Sigma) for 10 minutes and then fixed with 4% PFA in 1X PBS for 15 minutes. Cells were washed with 1X PBS, permeabolized with 0.1% Triton X-100 in 1X PBS for 5 minutes, blocked with 1% BSA/0.1% Triton X-100 in 1X PBS for 15 minutes and then simultaneously stained in the dark for 20min at RT with rhodamine-conjugated phalloidin (Molecular Probes R415; ThermoFisher Scientific), and Alexa 488–conjugated DNase I (Molecular Probes D12371; ThermoFisher Scientific) in 1% BSA in 1X PBS to identify F-actin and G-actin, respectively. Cells were washed with 1X PBS and coverslipped using Vectashield H-1200 (Vector). Fluorescent intensities were recorded within each microscopic field (24 10X images for each cell line treated with oxytocin and 24 10X images for each cell treated with no oxytocin) and quantified using Image J software (NIH). F:G actin ratios were calculated for each image (as previously described [38,39]) and an average F:G actin ratio value was generated for each cell line treatment group.

Polydimethylsiloxane (PDMS) Pillar Arrays

PDMS pillar arrays were performed to test if ECM rigidity or type of ECM protein influences CSMC contractile behavior [40–43]. In this array, cells are placed on top of carpet-like pillar arrays coated with an ECM protein. Using live time lapse imaging, cells are recorded as they pull on the pillar tops to contract. Contractile forces that are exerted by the cells are then calculated from the degree of pillar displacement. We used 1.8μm tall, 0.5μm wide pillars as previously described [41,42]. Briefly, PDMS (Sylgard 184, Dow Corning) was mixed thoroughly with its curing agent (10:1), degassed, poured over the silicon master, placed upside-down on a plasma-treated petri dish and cured at 70°C for 10–14 hours. The mold was then removed and the pillars were either UV treated for 2 hours to create hard pillars (rigidity of 40kPa, which resembles the stiffness of the human cervix in early pregnancy) or left without UV treatment to create soft pillars (rigidity of 2kPa resembles stiffness of the human pregnant cervix at term) [8, 9, 44]. To further determine if type of ECM protein influences CSMC contractility, pillars were then either coated with 10μg/mL of rat tail type I collagen (Corning, Corning, NY) diluted in ice cold 1X PBS (pH 4.5) or 50ug/mL of fibronectin diluted in 1X PBS. Pillars were subsequently incubated at 37°C for 1 hour, washed with 1X PBS and then SmBm-2 media containing manufacturer’s supplements. Primary CSMC from 6 women from Study Group A (women with a history of premature cervical remodeling leading to sPTB < 32 weeks) and 6 women from Control Group A were seeded at a density of 1X10⁵/pillar array and incubated at 37°C for 1 hour in SmBm-2 media.

Time lapse imaging of pillars was performed using an Orca-flash 2.8 camera (Hamamatsu) attached to an inverted microscope (Olympus IX-81). Pillars/cells were maintained at 37°C with a temperature isolation chamber running MicroManager software (UCSF). Images were recorded at 1 frame/second for 20 minutes using a 60x objective (1.4 NA oil immersion). Pillar displacements were calculated as previously [41,42] described using the ImageJ NanoTracking plug-in (National Institutes of Health). To account for stage drift, we subtracted the average displacement of pillars located remotely. Six cells from each participant in the study (n=6 patients; 36cells in the study group) and control group (n=6 patients; 36 cells in the control group) were analyzed. To calculate the average force exerted by each cell as it contracts across the pillars, the following equation was used: Force (pN) = pillar displacement (nm) x pillar bending stiffness (K in pN/nm). The bending stiffness (K) for UV treated hard pillars has been established as 60pN/nm and the bending stiffness for non-UV treated soft pillars is 3 pN/nm [44]. The average peak contractile force was then generated for all cells in the study and control groups on the following conditions: collagen-coated hard pillars, collagen-coated soft pillars, fibronectin-coated hard pillars, fibronectin-coated soft pillars.

Collagen Crosslink Analysis

To test whether women with a history of premature cervical failure have softer tissue in the baseline state prior to pregnancy, cervical tissue biopsies (one per patient) were obtained from non-pregnant women (greater than 18 years old) with a history of prior sPTB <32 weeks due to premature cervical shortening/dilation undergoing open abdominal cerclage placement (Study Group C) and non-pregnant, premenopausal women (18–50 years old) without a history of sPTB who were undergoing total hysterectomy for benign indications (Control Group C) using IRB-approved protocols (Table 1). In order to keep our area of analysis consistent with our 2016 study of non-pregnant human cervical tissue, cervical tissue biopsies were obtained from the anterior lip (at 12 o’clock) of the cervix at the level of the internal os (defined as area where the uterine arteries meet the cervix) using a 2mm Achieve Automatic Biopsy Device (Merit Medical). In women undergoing open abdominal cerclage placement, biopsies were obtained after dissection of the vesicouterine space to ensure the bladder was below the area of the uterine arteries. In women undergoing benign hysterectomy, biopsies were taken immediately after the uterus/cervix was removed from the patient. Samples were immediately frozen on dry ice then processed as previously described [18, 45]. Briefly, samples were lyophilized and dry weights for each sample were established. Sodium borohydride (NaBH4) was used to reduce the dehydrated samples before they were hydrolyzed in 12M HCL at 110°C for 18–24 hours. The hydrolysate was again lyophilized and suspended in heptaflurobutyric acid (HFBA) buffer. The molar content of hydroxyproline (HYP), mature (PYD, DPD) and immature (DHLNL, HLNL) collagen crosslinks were measured using UPLC-ESI-MS/MS. Total collagen content [mol] was calculated as a ratio of collagen to hydroxyproline of 7.14:1, and total collagen content [g] was calculated using a molecular weight = 131g/mol. Collagen concentration was determined by dividing total collagen content [mg] by tissue dry weight [mg]. Individual collagen crosslink densities were defined as crosslink content divided by collagen content on a mole per mole basis [PYD mol /collagen mol, DPD mol /collagen mol, DHLNL mol/collagen mol, HLNL mol/collagen mol]. Collagen crosslink maturity ratio is defined as total mature to total immature crosslinks [PYD+DPD mol /HLNL+DHLNL mol]. Patient demographics, differences in individual collagen crosslinks and maturity ratios were then compared between Study Group C and Control Group C.

Statistics

Statistical analyses were performed using GraphPad Prism 8 software (GraphPad Software Inc), SAS 9.4 and R. For F:G actin ratio assays, the F:G actin ratios between cells not exposed to oxytocin vs those exposed to oxytocin were compared using paired t-tests. For PDMS pillar array experiments, generalized estimated equations were used to analyze differences between the log-transformed average max contractile force of cells from the Study Group versus cells from the Control Group which addresses the fact that six cells from the same participant are correlated. For collagen crosslink studies, differences in individual crosslinks and maturity ratios were compared between Study and Control Groups using Student t-test. Student t-test was used to evaluate differences in patient demographic data and linear regression was performed to control for demographic differences where appropriate. P values of <0.05 were considered significant.

Results

Pregnant human cervical tissue contains a significant amount of smooth muscle at the internal os

Cervical tissue was collected from 10 pregnant women following cesarean hysterectomy. (See Supplemental Table S2 for demographic data). H&E slides were reviewed with a Pathologist who confirmed each cervical tissue sample did not exhibit evidence of placental invasion. Human cervical tissue obtained from the upper aspect of the cervix contains a significant amount of α-SMA positive CSMC (approximately 50–60% of the tissue) that appears slightly more disorganized than non-pregnant tissue. However, circumferential bands of CSMC were noted (Fig.2a, b). At the level of the external os, α-SMA positive CSMC comprised approximately 10–20% of the tissue (Fig. 3a, b). Some bundles of CSMC at the external os are circumferentially oriented around the endocervical canal but most CSMC are scattered in the stroma (Fig. 3b). α-SMA positive CSMC in pregnant human cervical tissue (at both the internal and external os) express mature smooth muscle markers (SM22, calponin) and contraction associated proteins (COX-2, oxytocin receptor, connexin 43) suggesting that these cells are indeed mature smooth muscle cells and not other cell types such as fibroblasts (Fig. 4,5).

Fig. 2.

This is a representative sample of a transverse slice of pregnant human cervical tissue obtained from the upper aspect of the cervix (area of the internal os). Tissue slices were stained with α-SMA. Fig. 2a: 10X images were obtained and tiled together to visualize and analyze the stained slice of tissue. In pregnant human cervical tissue, approximately 50–60% of the tissue in the upper aspect of the cervix is α-SMA positive cells. Fig 2b: 20X images were obtained and tiled together to analyze a section of the stained slice of tissue shown in Fig. 2a (boxed area). Although more disorganized than the non-pregnant cervix, the upper aspect of pregnant human cervix contains smooth muscle bundles that are circumferentially oriented around the endocervical canal (Fig. 2b arrows).

Fig. 3.

This is a representative sample of a transverse slice of pregnant human cervical tissue obtained from the area of the external os. Tissue slices were stained with α-SMA. Fig. 3a: 10X images were obtained and tiled together to visualize and analyze the stained slice of tissue. In pregnant human cervical tissue, approximately 10–20% of the tissue in the area of the external os contains α-SMA positive staining cells. Fig. 3b: 20X images were obtained and tiled together to analyze a section of the stained slice of tissue shown in Fig. 3a (boxed area). Although the external os contains some circumferential bundles of smooth muscle (Fig 3b arrows), a good portion of the tissue exhibits more dispersed smooth muscle bundles.

Fig. 4.

Colorimetric immunohistochemistry was used to stain pregnant human cervical tissue slices obtained from the upper aspect of the cervix (area of the internal os) and external os. Tissue slices were stained for mature smooth muscle markers (α-SMA, SM22, calponin) and contraction- associated proteins (oxytocin receptor, COX-2). The images in this figure are a representative sample which shows that α-SMA positive cells in the upper aspect of the cervix and the external os also stained positive for mature smooth muscle markers and contraction-associated proteins. This suggests these cells are mature smooth muscle cells and not other cell types such as fibroblasts. Images are 10X with the scale bar = 100μm.

Fig. 5.

Fluorescent immunohistochemistry was used to stain for the contraction-associated protein, connexin 43, in order to determine if this protein (along with the other contraction-associated proteins shown in Fig. 4) were present on the α-SMA positive cells in the area of the upper cervix and the external os. The images in this figure are a representative sample which shows that α-SMA positive cells at the upper aspect of the cervix and the external os also stained positive for connexin 43. This finding, along with the findings in Fig. 4, suggest the α-SMA positive cells are mature smooth muscle cells that express proteins involved in smooth muscle cell contractility and not other cell types. Images are 40X with the scale bar = 20μm.

ECM rigidity influences primary human CSMC contractility

One cervical tissue biopsy was obtained from each woman in Study Group A (n=21 women), Control Group A (n=13 women), Study Group B (n=9 women) and Control Group B (n=7 women) Demographic data is listed in Supplemental Table S3. Similar to previous studies, no complications were noted after obtaining cervical biopsies during pregnancy, even in patients with a short cervix [46]. Immunocytochemistry studies of primary CSMC (isolated from the biopsy from each patient) demonstrates over 98% of the primary CSMC cells express mature smooth muscle cell markers (α-SMA, SM22, desmin; Fig. 6) confirming that these cells were not other cell types such as fibroblasts.

Fig. 6.

Primary CSMC were isolated from Control Groups A and B and Study Groups A and B then immunofluorescently stained for smooth muscle markers (α-SMA, desmin, SM22). Images in the figure are representative images of the primary CSMC lines which demonstrates over 98% of the isolated cells express smooth muscle markers suggesting they are indeed smooth muscle cells and not other cell types. Scale bar = 100μm.

F:G actin ratio assays

F:G actin ratio assays were conducted on the same primary CSMC lines that underwent mature smooth muscle marker staining described above. 1μM oxytocin exposure significantly increased F:G actin ratios (indicating increased contractility) in CSMC from in Control Groups A, B and Study Groups A, B (Fig. 7). Oxytocin exposure did not increase F:G actin ratios in non-contractile HEK293T cells (Fig. 7). This suggests that primary human CSMC are contractile and CSMC from women with a history of premature cervical failure do not exhibit an inherent contractility defect at the level of the actin cytoskeleton (when plated on glass slides).

Fig. 7.

CSMC were isolated from the cervical tissue biopsy obtained from each pregnant woman in Control Group A, Study Group A, Control Group B and Study Group B. Primary CSMC were treated with serum-free media or serum-free media containing a contractile agent (1μM oxytocin) for 10 min and then assessed by F:G actin ratio assays. HEK293T cells were used as non-contractile controls. HEK293T cells did not exhibit increased F:G actin ratios after oxytocin exposure (No oxytocin vs Plus oxytocin: 1.77 ± 0.64 vs 1.82 ± 0.54, P=0.8). Each group of CSMC exhibited significantly increased F:G actin ratios after oxytocin exposure suggesting increased contractility at the level of the actin cytoskeleton (No oxytocin vs Plus oxytocin: Control Group A 3.7 ± 2.3 vs 7.0 ± 4.0, P<0.0001; Study Group A 3.6 ± 1.9 vs 6.6 ± 3.3, P<0.0001; Control Group B 3.3 ± 2.1 vs 6.2 ± 3.3, P=0.005; Study Group B: 4.1 ± 3.1 vs 7.2 ± 3.6, P=0.006). All values reported are the mean ± SEM.

PDMS Pillar Arrays

A subset of primary CSMC from 6 women in Control Group A and 6 women in Study Group A (that underwent staining and F:G actin assays as outlined above) were used for this experiment. Demographic data are listed in Supplemental Table S3 and S4. We found that ECM rigidity influences CSMC contractile behavior. Specifically, CSMC seeded on collagen or fibronectin-coated hard pillars (simulates the rigidity of human cervix in early gestation) contracted with significantly more force when compared to CSMC that were seeded on soft pillars (simulates the rigidity of the cervix at term) (Fig. 8a, b). On soft, pillars, the type of ECM protein appeared to influence CSMC contractility. CSMC from Control Group A contracted more forcefully on collagen-coated pillars compared to fibronectin-coated pillars (Fig. 8c). This trend was also noted in CSMC from Study Group A (Fig. 8c) and in CSMC from both Control Group A and Study Group A on hard pillars (Fig. 8d) but the P-values did not reach statistical significance.

Fig. 8.

Primary CSMC from women in Control Group A and women in Study Group A were placed on soft or hard PDMS pillars, which were coated with either collagen or fibronectin. Six cells from each of the 6 participants in the Study Group (n= total 36 cells) and 6 cells from each of the 6 participant in the Control Group (n=total 36 cells) were analyzed. On both collagen and fibronectin-coated pillars, primary CSMC from both patient groups exerted significantly more contractile force on hard compared to soft pillars (Collagen-coated hard vs soft pillars: Control Group A 8.2 ± 0.1 vs 5.5 ± 0.1 pN, P<0.0001; Study Group A 8.4 + 0.1 vs 5.5 ± 0.1 pN, P<0.0001; fibronectin coated hard vs soft pillars: Control Group A 8.1 ± 0.1 vs 5.2 ± 0.1 pN, P<0.0001; Study Group A 8.2 ± 0.1 vs 5.3 ± 0.1 P<0.0001; Fig. 8a, b). On soft pillars, CSMC from Control Group A contracted with more force on collagen-coated fibers compared to fibronectin-coated pillars (5.5 ± 0.1 vs 5.2 ± 0.1 respectively, P=0.02; Fig. 8c). Although CSMC from Study Group A showed similar results on soft pillars, the difference was not statistically significant. (Fig. 8c) On hard pillars coated with either collagen or fibronectin, the trend where CSMC contracted with more force on collagen-coated compared to fibronectin-coated pillars was also seen but the differences did not meet statistical significance. (Fig. 8d) All values reported are the mean ± SEM.

Non-pregnant women with premature cervical failure may have softer cervical tissue

Collagen crosslink analysis

To determine if women with a history of premature cervical failure have softer cervical tissue prior to pregnancy, cervical tissue was collected from 10 women in Control Group C and 10 women in Study Group C. Demographic data are listed inSupplemental Table S3. Women in Control Group C were significantly older than women in Study Group C (44±3 vs 33±6 years; P=0.005; Supplemental Table S4). There was no significance difference in parity between the two groups (Supplemental Table S4). Following adjustment for patient’s age, cervical tissue from women in Study Group C exhibited a trend to lower mature (PYD, DPD) collagen crosslink density ratios and higher immature (HLNL) crosslink density ratio compared to women in Control Group C but these values did not reach statistical significance (Fig. 9a). Cervical tissue obtained from women in Study Group C exhibited significantly lower collagen concentration per dry weight of tissue (Fig. 9b) and collagen crosslink maturity ratio (Fig. 9c) compared to women in Control Group C. Given that prior studies have shown decreased collagen crosslink maturity ratios are directly correlated to softer cervical tissue mechanical properties [11], these results may suggest that women with a history of premature cervical failure leading to sPTB may have softer cervical tissue prior to pregnancy.

Fig. 9.

Cervical tissue biopsies were obtained (one biopsy per patient) from area of the internal os from non-pregnant women with a history of premature cervical failure leading to sPTB undergoing abdominal cerclage placement (Study Group C) and non-pregnant women without a history of sPTB undergoing total hysterectomy (Control Group C). Cervical tissue biopsies were evaluated for mature (PYD, DPD) and immature (DHLNL, HLNL) collagen crosslink densities, collagen concentration and collagen crosslink maturity ratios. Cervical tissue from women in Study Group C exhibited a trend to lower mature (PYD, DPD) collagen crosslink density ratios and higher immature (HLNL) crosslink density ratio compared to women in Control Group C but these values did not reach statistical significance. (Fig. 9a) Cervical tissue biopsies from the Study Group exhibited decreased collagen concentration and collagen crosslink maturity ratios compared to biopsies obtained from the Control Group (Collagen concentration: 25% ± 5% vs 40% ± 6% respectively, P=0.01; maturity ratio: 0.2 ± 0.2 vs 0.8 ± 0.2 respectively, P<0.0001; Fig. 9b, c). All values reported are the mean ± SEM.

Discussion

This study established several important findings which considerably expands our fundamental knowledge of human cervical tissue structure and function in pregnancy. First, we confirmed that the upper aspect of the cervix of the pregnant third trimester human cervix contains a significant amount of smooth muscle with areas that are circumferentially oriented around the endocervical canal. This finding differs from the prevailing paradigm established in the 1940–1960s, which reported that CSMC content decreases in pregnancy [4, 5, 47–49]. Our findings, however, are consistent with recent studies, which report persistence of functional smooth muscle throughout pregnancy in the cervix of rats, cows, and primates [50–52]. Our finding adds support to our hypothesis that the smooth muscle body in the upper aspect of the cervix may collectively function as a specialized sphincter whose function is to hold the fetus in the uterus in pregnancy. In addition, this functional smooth muscle body in the upper aspect of the cervix may also explain the regional differences that are noted when the cervix is examined clinically. Specifically, this paradigm may explain how the area of the internal os closes within minutes after delivery while the external os remains open/soft for weeks in the postpartum period. It may also explain why clinical exam of a multiparous cervix (women who have had multiple vaginal deliveries) commonly demonstrates an internal os that is tightly closed while the external os is loose and dilated. This concept is also supported by studies that were done in the 1950s which show that the pregnant cervix contracts independently from the uterus in response to contractile agonists and that the human cervix exhibits electromyographic activity during labor [24–31].

Our second goal was to determine if women with a history of premature cervical failure had inherent differences in actin-mediated CSMC contractility, which may explain why the upper aspect of the cervix dilates or fails prematurely in patients at risk for sPTB. When plated on glass slides, primary CSMC isolated from women with a history of premature cervical failure and sPTB do not appear to have an inherent contractility defect. However, when CSMC were seeded on soft or hard surfaces (pillar arrays) we found that ECM rigidity influences human CSMC contractility. Specifically, primary human CSMC exert more contractile force when exposed to hard surfaces and less force on soft surfaces. These findings agree with studies from various cell types in other organ systems, which illustrate that ECM rigidity can influence cell differentiation and behavior [33–35]. This finding adds cellular-level insight to the physiological changes we know occur in the cervix during pregnancy. Specifically, on clinical exam we observe the human cervix begins as a rigid, stiff structure in early pregnancy and dramatically softens towards the end of pregnancy in preparation for delivery [7, 8, 11, 12]. If CSMC contractile tone is enhanced when on rigid ECM, the rigid cervical ECM in early pregnancy may contribute to increased tone in the hypothesized sphincter that results in maintenance of cervical integrity early in pregnancy. Further, the finding that CSMC do not contract as well on softer matrix may begin to explain how a functional smooth muscle body in the upper aspect of the cervix decreases its inherent tone at the end of pregnancy due to an interaction with softer, remodeled cervical ECM.

We also found that non-pregnant women with a history of premature cervical failure resulting in sPTB have decreased collagen concentration and collagen crosslink maturity ratios (implying softer cervical tissue) compared to non-pregnant women without a history of premature cervical failure and sPTB. Our findings are similar to some prior studies that found non-pregnant women with a history of premature cervical remodeling exhibited lower collagen concentration compared to their non-pregnant controls [53–56]. Interestingly, Oxlund et al initially found a difference in collagen concentration between women with and without a history of cervical insufficiency but this difference was not statistically significant after controlling for parity and age [17]. Although it is difficult to explain the difference in results between Oxlund et al’s study and those that found a difference in collagen concentration between women with and without premature cervical failure, factors such as biopsy location and patient ethnicity/race may have played a role in the findings. Further studies are needed to investigate how these factors may influence cervical ECM properties. This is the first study, however, to evaluate collagen crosslink maturity ratios (which have been shown to be directly related to mechanical properties of cervical tissue) in these cohorts of non-pregnant women. [11]. Combining our crosslink maturity ratio data with our pillar array results, we postulate that women with a history of premature cervical failure may have softer cervical tissue prior to pregnancy which contributes to reduced CSMC tone throughout pregnancy and thus functionally a weaker sphincter. This in turn may begin to explain how and why a weakened sphincter fails in the mid-trimester as it is challenged by the increasing weight of the growing fetus.

Lastly, we found different ECM proteins can influence CSMC contractility. Normal CSMC appear to contract more forcefully when exposed to collagen-coated soft surfaces compared to fibronectin-coated soft surfaces. Given this, it is imperative that a modern proteomic approach is undertaken to determine which ECM proteins exist in the human cervix (at baseline), how ECM protein profiles change throughout pregnancy and how key ECM proteins influence CSMC contractility and mechanical stiffness of human cervical tissue. In addition, further studies are needed to understand how ECM protein profiles may differ in women who are destined for premature cervical failure leading to a midtrimester loss or sPTB. Several studies have conducted proteomic analyses of the mouse cervix in pregnancy and the postpartum state and would serve as an excellent foundation from which to perform a similar analysis on human cervical tissue [57,58].

This study has several limitations. First, although stereology is an alternative method of quantifying volume fraction of tissue components, it is not readily compatible with sections of the entire human cervix (over 1000 tiles); the time to acquire and stitch multiple sections to apply this method would be prohibitive. Therefore, we have used the area fraction of CSMC on a section as an approximation of the volume fraction of CSMC in the tissue. Second, it is our hypothesis that the possible sphincter in the cervix relaxes at the end of pregnancy to allow for delivery of the fetus at term. Interestingly, Ferland et al found that in rats, the contractility of the cervix in organ function baths may actually increase towards the pregnancy [51]. Although it is unclear if rats and humans share the same contractile behavior in the cervix during pregnancy, further studies are needed to evaluate the active contractile function of the cervix at the internal os as pregnancy progresses and after delivery using in vivo techniques. Third, further studies are also needed to evaluate whole human cervical tissue from patients without complete previas/abnormal placentation to confirm CSMC content, architecture and function is not influenced by placentation issues. Fourth, although no significant differences in F:G actin ratios and pillar array results were noted comparing CSMC from the Study Groups compared to their Controls, these findings may be limited by the small sample size when considering differences that may exist due to parity or ethnicity (both of which were beyond the scope of the present study). Lastly, in our collagen crosslink study, cervical tissue biopsies were immediately frozen after collection. Given the location of the biopsy, the outer layer of the cervix was likely disrupted due to the dissection of the vesicouterine space (prior to abdominal cerclage placement and hysterectomy) but the endocervical cellular layer remained intact. Although studies have not fully evaluated if patient age influences the thickness of the endocervical cellular layer at the level of the internal os, given that the patients in Control Group C were significantly older than the women in Study Group C, further studies are needed to ensure that age does not influence the endocervical cellular thickness at the internal os. Despite these limitations, we do note the strengths of this study. It is incredibly challenging to obtain human cervical tissue during pregnancy. Yet, we were able to analyze human cervical tissue samples from 10 pregnant women in the third trimester. In addition, this is the first study of its kind in its attempt and success in isolating primary CSMC from women with a history of premature cervical failure and similar gestational age controls.

In summary, our findings highlight the need to investigate the role of mechanobiology in human reproductive tissues. Specifically, studies are needed to understand not only how the ECM in the cervix changes in pregnancy but also how changes in cervical tissue ECM rigidity and protein profiles influence cellular behavior. These studies will be key to understanding how the cervix functions in normal pregnancy and how it malfunctions in pregnancies complicated by premature cervical failure. This knowledge may ultimately allow for discovery of novel pathways and targets that result in effective therapeutic options to prevent spontaneous preterm birth.