1

Changes in vascular extracellular matrix composition during decidual spiral arteriole

remodeling in early human pregnancy.

Samantha D Smith, PhD*1,2, Ruhul H. Choudhury, BSc*1,2, Patricia Matos, BSc1,2, James A

Horn BSc1,2, Stephen J Lye, PhD3, Caroline E Dunk, PhD3, John D Aplin, PhD1,2, Rebecca L 5

Jones, PhD1,2, Lynda K Harris, PhD1,2,4,5

1,2 Maternal and Fetal Health Research Centre (MFHRC), 1Institute of Human Development,

University of Manchester, 2St. Mary's Hospital, Central Manchester University Hospitals NHS

Foundation Trust, Manchester Academic Health Science Centre, Manchester, UK. 10

3Departments of Physiology and Obstetrics and Gynecology, Women’s and Infant’s Health,

Samuel Lunenfeld Research Institute, Mount Sinai Hospital, University of Toronto, Toronto,

Ontario, Canada.

4Manchester Pharmacy School, Stopford Building, Manchester, UK

5To whom all correspondence should be addressed. Maternal and Fetal Health Research Centre, 15

5th Floor (Research), St. Mary’s Hospital, Oxford Road, Manchester, M13 9WL, UK

Tel: 44 (0)161-276-6962

Fax: 44 (0)161-701-6971

lynda.k.harris@manchester.ac.uk

20

*SDS and RHC contributed equally to this publication.

Pages: 33 Tables: 2 Figures: 7

2

Running Title: ECM in remodeling spiral arterioles

Keywords (up to 5): extracellular matrix, extravillous trophoblast, placenta, spiral arteriole, 25

vascular remodeling

Author contributions: SDS, SJL, CED, JDA, RLJ and LKH designed the experiments; SDS,

RHC, PM and JAH performed the research; SDS, RHC, PM, JAH and LKH analyzed the data;

SDS and LKH wrote the manuscript. All authors discussed the results and commented on the 30

manuscript.

3

Abstract (250 words)

Uterine spiral arteriole (SA) remodeling in early pregnancy involves a coordinated series of

events including decidual immune cell recruitment, vascular cell disruption and loss, and 35

colonization by placental-derived extravillous trophoblast (EVT). During this process, decidual

SA are converted from narrow, muscular vessels into dilated channels lacking vasomotor

control. We hypothesized that this extensive alteration in SA architecture must require

significant reorganization and/or breakdown of the vascular extracellular matrix (ECM). First

trimester decidua basalis (30 specimens) was immunostained to identify spiral arterioles 40

undergoing trophoblast-independent and -dependent phases of remodeling. Serial sections were

then immunostained for a panel of ECM markers, to examine changes in vascular ECM during

the remodeling process. The initial stages of SA remodeling were characterized by loss of

laminin, elastin, fibrillin, collagen types III, IV and VI from the basement membrane, vascular

media and/or adventitia, and surrounding decidual stromal cells. Loss of ECM correlated with 45

disruption and disorganization of vascular smooth muscle cells, and the majority of changes

occurred prior to extensive colonization of the vessel wall by EVT. The final stages of SA

remodeling, characterized by the arrival of EVT, were associated with the increased mural

deposition of fibronectin and fibrinoid. This study provides the first detailed analysis of the

spatial and temporal loss of ECM from the walls of remodeling decidual SA in early 50

pregnancy.

55

4

Introduction

Remodeling of uterine decidual spiral arterioles (SA) is essential for a successful pregnancy.

During this process, SA are transformed from narrow, muscularized vessels into dilated

channels, that ensures delivery of sufficient blood to the materno-fetal interface to meet the

needs of the developing fetus (Pijnenborg et al., 2006b; Burton et al., 2009). Impaired 60

remodeling is associated with an increased risk of pre-eclampsia, fetal growth restriction, pre-

term labor and stillbirth (Kim et al., 2003; Ball et al., 2006; Brosens et al., 2011). The process

of remodeling is complex, but ultimately involves loss of vascular smooth muscle cells

(VSMC), which are replaced by a fibrinoid matrix containing non-contractile, invading

extravillous trophoblast cells (EVT) derived from the developing placenta (Ashton et al., 2005; 65

Harris et al., 2006). Although complete physiological transformation of the SA is dependent on

the actions of EVT, changes in vessel architecture are observed prior to invasion (Craven et al.,

1998; Smith et al., 2009); these have been termed trophoblast-independent remodeling events

and are maternally regulated (Crocker et al., 2005; Pijnenborg et al., 2006b).

70

Our previous work has shown that decidual SA remodeling can be separated into four distinct

stages (Smith et al., 2009), and that some of these events can be reproduced using a decidual

and placental co-culture model (Hazan et al., 2010). Unremodeled SA, classified as stage I, are

characterized by an intact endothelium, tightly organized layers of circumferential VSMC and

no evidence of immune cell or EVT infiltration. Stage II SA exhibit endothelial cell 75

hypertrophy and loss, coupled with extensive disruption and partial loss of VSMC; these

changes are coincident with infiltration of the vessel wall by uterine natural killer cells (uNK)

and decidual macrophages. Stage III SA contain fewer decidual leukocytes and the majority of

5

vascular cells have been lost. Endovascular EVT are evident and some have adhered to the

arterial wall. Stage IV SA exhibit extensive or complete loss of VSMC and endothelium, and 80

are often lined with a continuous layer of EVT, although some endothelium may be retained.

Arrival of endovascular EVT in the SA is associated with the deposition of fibrinoid, a fibrin-

related, acellular, extracellular material, within the vascular wall. Fibrinoid is believed to confer

structural integrity to the remodeled vessels and provide a scaffold for EVT colonisation (Aplin

and Foden, 1982). Fibrinoid is a recognized hallmark of remodeled SA (Wells et al., 1984); 85

however, the timing of its deposition in relation to EVT invasion is unclear.

To ensure a permanent state of vasodilatation is achieved, the ECM within the decidual SA

must be reorganized in parallel with the disruption and loss of vascular cells. This likely

mediated by local proteases; we and others have shown that invading uNK, macrophages, 90

VSMC and EVT are a potent source of matrix metalloproteinases (MMP) (Smith et al., 2009;

Harris et al., 2010; Anacker et al., 2011; Robson et al., 2012). MMP-2, -7, -9 and -12 have

been localized to EVT in and around remodeling vessels in vivo (Jones et al., 2006; Plaisier et

al., 2008; Smith et al., 2009; Harris et al., 2010), and infiltrating uNK and decidual

macrophages secrete MMP-7 and -9 (Plaisier et al., 2008; Naruse et al., 2009; Smith et al., 95

2009).

Little is known about the specific ECM composition of decidual SA; however, healthy arterial

walls contain collagens, laminin and elastin, and proteoglycans such as decorin, versican and

perlecan (Jacob, 2003). Collagen types I, III and V are observed in large banded fibrils 100

throughout the intima, media and adventitia, whereas collagen type IV is a principal component

6

of the basal lamina to which endothelial cells and VSMC adhere (Shekhonin et al., 1985).

Laminin is also a major protein in basal lamina in the decidua and decidual SA (Charpin et al.,

1985; Wewer et al., 1985; Church et al., 1996), forming a cross-linked structure to which other

membrane and ECM molecules (such as collagen type IV) can bind. Other constituents of the 105

decidual vascular ECM in humans and rodents include collagen type VI, heparan sulphate

proteoglycan, fibrillin and fibronectin (Aplin et al., 1988; Earl et al., 1990; Mulholland et al.,

1992; Dziadek et al., 1995); many of these components can be synthesized by resident vascular

cells (Harris and Aplin, 2007). Unlike the myometrial spiral arteries, which exhibit a well

formed elastic lamina together with concentric mural layers of elastin (Khong et al., 2003), the 110

decidual SA are not believed to contain elastin in the form of a true elastic lamina.

We hypothesized that decidual leukocytes act in conjunction with invading EVT to orchestrate

ECM remodeling within the decidual SA, facilitating the process of vascular transformation. In

the current study, we sought to characterize the ECM composition of intact decidual SA and 115

document changes in ECM architecture in relation to VSMC loss and EVT infiltration.

7

Materials and methods

Tissue collection

Placental and decidual tissue (n = 30) were obtained from women undergoing elective surgical 120

termination of pregnancy at St. Mary’s Hospital, Manchester for social reasons in the first

trimester (6-12 weeks gestation). Written informed consent was obtained from all patients and

ethical approval was obtained from Central Manchester Research Ethics Committee

(03/CM/031) and the North West Research Ethics Committee (08/H1010/28;

08/H1010/55(+5)). 125

Tissue processing and fixation

Decidua and placenta were separated following gross morphological examination and were

randomly sampled. Multiple biopsies of decidua basalis (10-20mm3) were washed in Tris-

buffered saline (TBS) and fixed in 10% neutral buffered formalin at 4°C for 24h, prior to 130

routine histological processing and embedding in paraffin wax.

Immunohistochemistry for vascular and ECM markers

Serial sections of wax-embedded decidua basalis (5µm) were deparaffinized in Histoclear and

alcohol, rehydrated and washed in dH2O. To facilitate antigen unmasking, slides were 135

submerged in sodium citrate buffer (275ml; 0.01M containing 0.05% (v/v) Tween 20, pH 6.0;

room temperature), the container was covered with plastic film to prevent evaporation and

microwaved for 2 X 5 min. Slides were left to cool for 20 min, then washed in cold dH2O for 5

min. Sections to be immunostained for collagen type III were treated differently: slides were

incubated with 1% (w/v) trypsin (Type II-S from porcine pancreas; 1000-2000 units/mg; Sigma 140

8

Aldrich) in 10mM calcium chloride for 20 min at 37°C, then washed in cold dH2O for 5 min.

The staining pattern of every antibody was tested with and without antigen retrieval; antigen

retrieval was only used where necessary, or if it did not affect the staining pattern and was

required for another antibody on a serial section. Endogenous peroxide activity was blocked by

incubating tissue sections in 3% (v/v) hydrogen peroxide for 10 min. Tissue sections were 145

washed three times in 0.05M Tris buffered saline (TBS) and incubated with non-immune block

(10% goat serum (Sigma Aldrich, UK), 2% human serum (in house) and 0.1% Tween-20 in

TBS) to inhibit non-specific antibody binding. Primary antibodies, diluted to the required

concentration in TBS (detailed in Table 1), were applied to the tissue sections, which were

incubated overnight at 4oC in a humidity chamber. Antibody concentrations were optimized 150

prior to characterization of the tissue samples. Negative controls (non-immune or isotype

control IgG, as appropriate) were also included; concentrations were matched to the

corresponding primary antibodies. Slides were washed (3 X 5 min; TBS) and the secondary

antibodies, diluted in TBS (Table 1), were applied for 30 min at room temperature. Slides were

washed (3 X 5 min; TBS) and incubated with avidin peroxidase (5µg/ml in TBS; Sigma 155

Aldrich) for 30 min at room temperature, then further washed in TBS and incubated for 1-5

min with 0.05% (w/v) DAB and 0.015% (v/v) hydrogen peroxide (Sigma Aldrich). All tissue

sections stained with the same antibody were incubated with DAB for the same length of time,

so that comparisons could be made between individual samples. Finally, slides were washed in

dH2O, counterstained with Harris’ hematoxylin (Sigma Aldrich), dehydrated in alcohol and 160

Histoclear, and mounted in DPX (Sigma Aldrich).

9

Periodic Acid Schiff reaction

The Periodic Acid Schiff (PAS) reaction was used to examine fibrinoid deposition in the

decidua; the reaction was performed using a commercially available kit (Sigma Aldrich). 165

Sections of decidua basalis were dewaxed in Histoclear, rehydrated and washed in dH2O.

Periodic acid (1g/dl) was applied to sections for 5 min, which were then rinsed with dH2O, and

incubated with Schiffs reagent for 10 min. After further dH2O washes, sections were

counterstained with Harris’ hematoxylin, dehydrated in alcohol and Histoclear and mounted in

DPX. 170

Microscopy and image analysis

A Leitz Dialux 22 microscope was used in conjunction with a QI Cam Fast 1394 camera and

Image ProPlus 6.0 imaging software; exposure times were matched at image capture. To select

representative spiral arteriole sections for display in each figure, stained slides and captured 175

images were reviewed and discussed by a minimum of 4 authors. Semi-quantitative assessment

of DAB staining intensity was performed by two independent observers: immunostained

samples were assigned a score of − (no staining), + mild staining, ++ moderate staining or +++

(intense staining). Slides were stained in large batches to minimize inter-experimental

variation; comparisons were only made between tissues stained in the same run. 180

10

Results

Changes in decidual vascular ECM during SA remodeling

Initial immunohistochemical analysis of archived decidual tissue was performed using 185

antibodies to smooth muscle actin and HLA-G to identify samples containing remodeling spiral

arterioles. More detailed analysis identified samples of decidua basalis from 30 pregnancies

that each contained examples of unremodeled vessels, vessels colonised by CD45-positive

leukocytes and vessels colonised by HLA-G-positive endovascular trophoblast, as previously

described (Smith et al., 2009). Individual arterioles did not exist in disparate stages of 190

remodeling; the majority of individual tissue sections contained multiple profiles of individual

SA as they spiraled through the tissue, where the proximal end was remodeled and the distal

end was not. Thus, within a single tissue section it was possible to observe profiles of

individual, dilated, trophoblast-lined SA in the proximal decidua (identified by remnants of

placental villi attached to the decidual surface), which could be traced into the distal decidua. 195

The profiles of these distal SA segments exhibited VSMC disruption in the absence of EVT,

and deeper still exhibited intact VSMC layers; thus all stages of remodeling could be observed

within profiles of a single SA. Tracing individual SA through the placental bed as progressive

remodeling occurred ensured that any remodeling vessel that was imaged and analyzed was a

distal portion of an intact SA rather than a vein or a gland; veins were excluded by their 200

morphology (thin layers of VSMCs, intact flattened endothelium, dilated lumens). Every

patient sample contained a minimum of two arterioles in all stages of remodeling. Using these

carefully characterized samples, immunostaining for collagen types III, IV, VI, laminin,

fibronectin, fibrillin and elastin was performed on serial sections, together with markers of

EVT and VSMC, to examine temporal changes in vascular ECM during the remodeling 205

11

process. The reported changes in ECM composition in remodeling SA were consistent across

all samples and gestational ages examined. There was no correlation between the gestational

age of the sample and the proportion of arterioles in each stage of remodeling, as previously

observed (Smith et al., 2009).

210

Collagen type III

Intense pericellular staining of collagen type III was observed in the decidual stroma (Fig 1A;

a). In unremodeled stage I SA (Fig 1C), collagen type III was detected in a thick band (b)

within the adventitia, but was absent from medial VSMC, endothelial cells and the endothelial

basement membrane. In stage II and III remodeling arterioles (Fig 1D, E), which display 215

disruption, disorganization and loss of VSMC, adventitial expression was reduced and fibers

appeared disrupted (c). Pericellular staining of adjacent decidual stromal cells was also reduced

in some areas. In stage IV remodeling arterioles (Fig 1F), which had been heavily colonized by

endovascular EVT, collagen type III immunoreactivity was markedly diminished or absent

from the vessel wall, but was observed in the basement membrane of SA (d) and in the decidual 220

stroma. Decidual stromal cell expression was also reduced in areas of interstitial EVT

infiltration (Fig 1E; e).

Collagen type IV

Intense pericellular staining for collagen type IV was detected around the periphery of decidual 225

stromal cells (Fig 2A; a). In unremodeled stage I SA (Fig 2C), collagen fibers were detected in

tightly bound, organized layers (b). Intense staining was detected in the basement membrane

(BM) underlying the vascular endothelium, and transverse and longitudinal layers of collagen

12

type IV were evident surrounding medial VSMC (Fig 2C). In stage II and III remodeling

arterioles (Fig 2D, E), collagen type IV expression within the vessel wall was weak or absent, 230

especially in areas of VSMC loss. Collagen type IV was still present in the basement membrane

in stage II, but in reduced quantities and gaps (c) were evident (Fig 2D). In stage IV

remodeling arterioles (Fig 2F), collagen type IV expression in the arterial wall was minimal.

Weak staining extending into the stroma was observed; however, pericellular decidual staining

was retained. 235

Collagen type VI

Areas of intense pericellular staining for collagen type VI were observed around the periphery

of a proportion of decidual stromal cells (Fig 3A; a). Although collagen type VI was evident in

the adventitia and outermost VSMC of unremodeled SA (b), and some areas of basement 240

membrane (BM) were weakly immunopositive, it was absent from the inner layers of medial

VSMC (Fig 3C, D; c). Collagen type VI was not observed in VSMC of remodeling arterioles;

however, some endovascular EVT within stage VI SA were immunopositive (Fig 3F; d).

Laminin 245

Pericellular laminin staining of varying intensity was detected in decidual stromal cells (Fig

4A, B; a). Within stage I unremodeled SA (Fig 4D), laminin was detected in the basement

membrane (BM) and weak staining was noted between layers of VSMC (b). In stages II- IV

(Fig 4E-G), laminin staining was very weak or absent in the vessel wall. Laminin was also

absent from endovascular EVT (Fig 4G). 250

13

Fibronectin

Pericellular fibronectin staining was observed in the decidual stroma (Fig 5A; a). In stage I

unremodeled SA (Fig 5C), fibronectin staining was observed in the outmost layers of medial

VSMC (b). In stage II and III remodeling SA (Fig 5D, E) weak, diffuse fibronectin 255

immunoreactivity was observed in some areas of the arterial wall. Interstitial (c) and

endovascular (d) EVT were immunopositive for fibronectin (Fig 5E) and in stage III and IV

remodeled arterioles (Fig 5F), the majority of luminal EVT were also positive for fibronectin

(e).

260

Fibrillin

In stage I unremodeled SA (Fig 6A), fibrillin was expressed in the adventitia and medial

VSMC (a), although discrete layers were not observed. Moderate levels of fibrillin staining

were observed throughout the decidual stroma (Fig 6A, D; b). Fibrillin expression or

localization was not significantly changed in the walls of stage II, III (not shown) or IV 265

remodeling SA (Fig 6B); however, interstitial and endovascular EVT exhibited pericellular

fibrillin expression (Fig 6B, c).

Elastin

Whilst organized, concentric layers of elastin were absent from unremodeled SA, discrete areas 270

of elastin immunoreactivity were observed within the inner layers of VSMC and areas

corresponding to the internal elastic lamina (Fig 6E; d). Decidual cells were also weakly

immunopositive for elastin (e). Following the onset of remodeling, the amount of elastin within

the vascular wall was reduced. In stages II, III and IV, weak elastin immunoreactivity was

14

observed within VSMC, EVT and decidual stromal cells and in some parts of the endothelial 275

basement membrane (Fig 6G; BM), but distinct fibers were no longer seen.

Fibrinoid

Fibrinoid deposition within the decidua was examined using PAS staining. PAS detects

glycogen, elastic fibril-associated glycoprotein and other highly glycosylated ECM as intense 280

pink staining of acellular material. The basement membrane (BM) of stage I SA, which is rich

in glycosylated molecules, was detected as organized bands of PAS-positive material which

encompassed the inner layers of medial VSMC (Fig 7A). PAS also labeled the decidual stromal

cell basement membrane (Fig 7A, B; a). Stage II and III remodeling SA, which exhibit

disruption and loss of VSMC layers (Fig 7B’), stained weakly with PAS (Fig 7B; c). PAS-285

positive fibrinoid present around transformed vessels is thicker and lacks a fibrillar appearance

(c); stage IV SA, which are characterized by extensive endovascular EVT infiltration (Fig

7C’’), were surrounded by areas of diffuse pink staining, indicative of fibrinoid deposition (Fig

7C). Weakly positive intraluminal and intravascular EVT were observed within these vessels

(Fig 7C; d). 290

15

Discussion

This study aimed to characterize the changes in vascular ECM composition during remodeling

of the decidual SA in early pregnancy. Here we confirm previous reports that collagen type IV 295

and laminin are major components of unremodeled vessels (Charpin et al., 1985; Wewer et al.,

1985; Aplin et al., 1988), and demonstrate that they are localized to VSMC and endothelial

basement membranes. We show that collagen types III and VI are present within the adventitia

of intact decidual SA, fibrillin is expressed by medial VSMC, and elastin and fibronectin are

observed in the medial ECM. 300

It has long been recognized that the initial stages of decidual SA remodeling involve severe

disruption and disorganization of the VSMC layers, which is accompanied by endothelial

swelling and basophilia (Craven et al., 1998; Kam et al., 1999); we have previously shown that

these changes are coincident with infiltration of decidual leukocytes (Smith et al., 2009). Here 305

we demonstrate that these alterations in vascular cell morphology are accompanied by

significant reorganization and/or degradation of the vascular ECM. VSMC within SA are

encapsulated within a basement membrane containing collagen type IV and laminin, as seen in

unremodeled stage I vessels. Fibronectin, fibrillin and small elastic fibrils are also evident as

pericellular deposits in some areas of VSMC at stage I. In normally functioning arterioles, cells 310

and ECM act as a unit for the transduction of mechanical forces (Martinez-Lemus, 2012). As a

major component of the decidual vascular ECM, collagen type IV degradation is likely to be

required to facilitate disruption of medial VSMC layers. Indeed, in the early stages of

remodeling where vascular cell layers are disrupted, collagen type IV fibers become

disorganized but remain abundant. We speculate that this initial disruption and ‘loosening’ of 315

16

the ECM, probably induced by local release of proteases and cytokines (Robson et al., 2012),

facilitates colonization by decidual leukocytes (Smith et al., 2009), which may have the

capacity to effect further ECM degradation within the vascular wall (Robson et al., 2012).

Collagen type VI is one of the first ECM components to be lost from stage II remodeling SA. 320

In earlier work we suggested that perivascular collagen type VI plays a role in integrating blood

vessels into the surrounding decidual stroma (Mulholland et al., 1992; Mylona et al., 1995). As

it is localized to the adventitia and outermost layers of VSMC, the first point of contact for

invading decidual leukocytes and EVT, it is the first layer of ECM that must be breached to

allow access to the underlying vascular cells. Similarly, expression of collagen type III and 325

laminin were greatly reduced in stage II, concomitant with evidence of VSMC disruption.

In stage III arterioles, which exhibit major disruption and loss of VSMC and endothelium,

collagen type IV distribution appeared much weaker, and in some areas was visible as

cytoplasmic reactivity in migrating, disrupted VSMC, suggesting degradation of the previously 330

highly organized fibers. In stage IV arterioles, expression of collagen type IV was in the

vascular wall was very low. The fate of the collagen type IV fibers between stages II and IV

remains unknown; dispersed fibers within the stroma may be phagocytosed by macrophages,

which are still present in and around the SA wall in the later stages of remodeling (Smith et al.,

2009). Alternatively, fragments could be further catabolized into smaller non-immunoreactive 335

peptides and may be cleared by the circulation or become incorporated into fibrinoid. Indeed,

the majority of vascular ECM in stage IV SA consisted of PAS-positive material; however,

varying levels of EVT-associated fibronectin, fibrillin, collagen type III and collagen type IV

17

were also evident. A qualitative summary of the progressive loss of ECM components from the

vascular wall during SA remodeling is given in Table 2. 340

Placental fibrinoid has been classified into two broad types: matrix-type fibrinoid and fibrin-

type fibrinoid, although both types are usually deposited together (Kaufmann et al., 1996).

Matrix-type fibrinoid, in which endovascular EVT are embedded, is secreted by endovascular

EVT and reportedly contains fragments of laminin, collagen type IV, heparan sulfate, 345

vitronectin and oncofetal fibronectin. In this study, we used PAS staining to identify the

presence of both forms of fibrinoid, and we have not differentiated between the two. As

observed in primates (Pijnenborg et al., 1996), we demonstrate that fibrinoid is only observed

in the walls of human SA in stage IV when EVT are present. This is consistent both with data

from our in vitro human tissue co-cultures (Hazan et al., 2010), where fibrinoid was not 350

detected (unpublished observations), and the work of others who have shown fibrinoid

deposition to be a late feature of remodeling (Brosens et al., 1967), and is associated with the

presence of interstitial EVT (Kam et al., 1999). Alternatively, the presence of plasma

components in interstitial fluid may be required for fibrin-type fibrinoid deposition. The exact

role of fibrinoid within remodeled SA is unproven but several functions have been proposed. It 355

supports the adhesion of trophoblasts (Aplin and Foden, 1982), and it is likely that during

destruction of the vascular wall and loss of the stability and support provided by the vascular

ECM, it is deposited to maintain structural integrity of the resulting channels. Secretions from

EVT embedded in the vascular wall probably contribute to fibrinoid, and its secretion has also

been detected in the path of interstitial EVT invasion (Pijnenborg et al., 2006a). Differences in 360

the composition of endovascular EVT- and interstitial EVT-derived fibrinoid have previously

18

been noted (Pijnenborg et al., 2006b); further work is still required to define its full range of

functions.

We have previously observed that localized disruption of the vessel wall corresponds with foci 365

of apoptotic vascular cells in remodeling SA (Smith et al., 2009; Hazan et al., 2010). Disruption

of ECM-adhesion molecule interactions that maintain VSMC in situ may also result in

cessation of vital cell survival signals, such as integrin-mediated signaling (Frisch and Francis,

1994). Loss of attachment of adherent cells to the ECM triggers a form of apoptotic cell death,

known as anoikis, which may represent an alternative mechanism for vascular cell loss during 370

active SA remodeling. In addition, healthy VSMC have been observed at a considerable

distance away from actively remodeling SA, suggesting that some VSMC may migrate into the

decidua following their release from the ECM. This could lead to dedifferentiation of VSMC

into myofibroblasts or pericytes and facilitate VSMC dispersal, as previously hypothesized

(Robson et al., 2009). 375

The majority of ECM constituents underwent significant catabolism in the early stages of SA

remodeling, prior to the arrival of endovascular EVT; this is the time when leukocytes infiltrate

the vascular wall (Smith et al., 2009). An increasing body of evidence suggests that decidual

uNK and macrophages actively participate in SA remodeling, but the specific effector functions 380

of these cells are still unclear. Soluble factors produced by uNK include angiopoietin-1,

angiopoietin-2, interferon-gamma and VEGF-C, (Lash et al., 2006; Lash et al., 2011); these

factors have been shown to induce VSMC disorganization and ECM breakdown in isolated

segments of human myometrial SA (Robson et al., 2012). Leukocytes are also a source of

19

MMP, which are likely mediators of ECM degradation in decidua and the arterial wall. Indeed, 385

inhibition of MMP action in the mouse uterus causes major disruption of decidualization

(Alexander et al., 1996). In other vascular beds, and indeed in cycling non-pregnant

endometrium, leukocytes are believed to be major effectors of tissue remodeling, through

production of MMP-3, -7 and -9 (Salamonsen and Lathbury, 2000; Kaitu'u et al., 2005), and

inflammatory mediators such as TGF-β and chemokines (Salamonsen and Lathbury, 2000). 390

EVT are also believed to participate in the degradation of vascular ECM, by producing MMP

capable of cleaving ECM components in the vascular wall (Isaka et al., 2003; Jones et al.,

2006). EVT are known to express a diverse range of MMP in the first trimester of pregnancy

(Anacker et al., 2011), and MMP-2 and -9 are differentially expressed by interstitial and

endovascular EVT (Jones et al., 2006), suggesting they have different functions that may relate 395

to their distinct environments. VSMC resident within remodeling SA have also been identified

as a source of MMP and may contribute to local ECM remodeling (Harris et al., 2010).

In summary, this study has provided a detailed analysis of the temporal loss of ECM from the

walls of remodeling decidual SA in human pregnancy and highlighted that the majority of 400

ECM breakdown occurs as an early event in the remodeling timeline. Whilst this observational

study cannot directly correlate leukocyte infiltration into stage II remodeling SA with ECM

breakdown, our findings suggest that a significant proportion of ECM reorganization occurs

during the trophoblast-independent phase of SA remodeling. We also speculate that the

impaired endovascular EVT invasion and incomplete SA remodeling observed in pre-eclampsia 405

could be ascribed to a reduction in ECM turnover, perhaps due to insufficient number or

activity of decidual uNK, macrophages or EVT. Alternatively, an inherent resistance of

20

maternal vascular ECM to degradation may be the cause, in addition to other defects in

trophoblast invasion; however, more detailed investigations will be required to further elucidate

this complex series of remodeling events. 410

21

Acknowledgements

We thank the research midwives at St. Mary’s Hospital, Manchester for their invaluable help

with subject recruitment and tissue collection. This work was funded by Tommy’s the Baby 415

Charity, The MRC and The Canadian Institute of Health Research. The MFHRC is supported

by Tommy’s, an Action Research Endowment Fund, the Manchester BRC and the Greater

Manchester CLRN. LKH is a BBSRC David Phillips Research Fellow.

420

22

References

Alexander C.M., Hansell E.J., Behrendtsen O., Flannery M.L., Kishnani N.S., Hawkes S.P. and

Werb Z. (1996). Expression and function of matrix metalloproteinases and their inhibitors at

the maternal-embryonic boundary during mouse embryo implantation. Development. 122,

1723-1736. 425

Anacker J., Segerer S.E., Hagemann C., Feix S., Kapp M., Bausch R. and Kammerer U. (2011).

Human decidua and invasive trophoblasts are rich sources of nearly all human matrix

metalloproteinases. Mol. Hum. Reprod. 17, 637-652.

Aplin J.D. and Foden L.J. (1982). A cell spreading factor, abundant in human placenta,

contains fibronectin and fibrinogen. J. Cell Sci. 58, 287-302. 430

Aplin J.D., Charlton A.K. and Ayad S. (1988). An immunohistochemical study of human

endometrial extracellular matrix during the menstrual cycle and first trimester of pregnancy.

Cell Tissue Res. 253, 231-240.

Ashton S.V., Whitley G.S., Dash P.R., Wareing M., Crocker I.P., Baker P.N. and Cartwright

J.E. (2005). Uterine spiral artery remodeling involves endothelial apoptosis induced by 435

extravillous trophoblasts through Fas/FasL interactions. Arterioscler. Thromb. Vasc. Biol. 25,

102-108.

Ball E., Bulmer J.N., Ayis S., Lyall F. and Robson S.C. (2006). Late sporadic miscarriage is

associated with abnormalities in spiral artery transformation and trophoblast invasion. J. Pathol.

208, 535-542. 440

Brosens I., Robertson W.B. and Dixon H.G. (1967). The physiological response of the vessels

of the placental bed to normal pregnancy. J. Pathol. Bacteriol. 93, 569-579.

23

Brosens I., Pijnenborg R., Vercruysse L. and Romero R. (2011). The "Great Obstetrical

Syndromes" are associated with disorders of deep placentation. Am. J. Obstet. Gynecol. 204,

193-201. 445

Burton G.J., Woods A.W., Jauniaux E. and Kingdom J.C. (2009). Rheological and

physiological consequences of conversion of the maternal spiral arteries for uteroplacental

blood flow during human pregnancy. Placenta. 30, 473-482.

Charpin C., Kopp F., Pourreau-Schneider N., Lissitzky J.C., Lavaut M.N., Martin P.M. and

Toga M. (1985). Laminin distribution in human decidua and immature placenta. An 450

immunoelectron microscopic study (avidin-biotin-peroxidase complex method). Am. J. Obstet.

Gynecol. 151, 822-826.

Church H.J., Vicovac L.M., Williams J.D., Hey N.A. and Aplin J.D. (1996). Laminins 2 and 4

are expressed by human decidual cells. Lab. Invest. 74, 21-32.

Craven C.M., Morgan T. and Ward K. (1998). Decidual spiral artery remodelling begins before 455

cellular interaction with cytotrophoblasts. Placenta. 19, 241-252.

Crocker I.P., Wareing M., Ferris G.R., Jones C.J., Cartwright J.E., Baker P.N. and Aplin J.D.

(2005). The effect of vascular origin, oxygen, and tumour necrosis factor alpha on trophoblast

invasion of maternal arteries in vitro. J. Pathol. 206, 476-485.

Dziadek M., Darling P., Zhang R.Z., Pan T.C., Tillet E., Timpl R. and Chu M.L. (1995). 460

Expression of collagen alpha 1(VI), alpha 2(VI), and alpha 3(VI) chains in the pregnant mouse

uterus. Biol. Reprod. 52, 885-894.

Earl U., Estlin C. and Bulmer J.N. (1990). Fibronectin and laminin in the early human placenta.

Placenta. 11, 223-231.

24

Frisch S.M. and Francis H. (1994). Disruption of epithelial cell-matrix interactions induces 465

apoptosis. J. Cell Biol. 124, 619-626.

Harris L.K., Keogh R.J., Wareing M., Baker P.N., Cartwright J.E., Aplin J.D. and Whitley G.S.

(2006). Invasive trophoblasts stimulate vascular smooth muscle cell apoptosis by a fas ligand-

dependent mechanism. Am. J. Pathol. 169, 1863-1874.

Harris L.K. and Aplin J.D. (2007). Vascular remodeling and extracellular matrix breakdown in 470

the uterine spiral arteries during pregnancy. Reprod. Sci. 14, 28-34.

Harris L.K., Smith S.D., Keogh R.J., Jones R.L., Baker P.N., Knofler M., Cartwright J.E.,

Whitley G.S. and Aplin J.D. (2010). Trophoblast- and Vascular Smooth Muscle Cell-Derived

MMP-12 Mediates Elastolysis during Uterine Spiral Artery Remodeling. Am. J. Pathol. 177,

2103-2115. 475

Hazan A.D., Smith S.D., Jones R.L., Whittle W., Lye S.J. and Dunk C.E. (2010). Vascular-

leukocyte interactions: mechanisms of human decidual spiral artery remodeling in vitro. Am. J.

Pathol. 177, 1017-1030.

Isaka K., Usuda S., Ito H., Sagawa Y., Nakamura H., Nishi H., Suzuki Y., Li Y.F. and

Takayama M. (2003). Expression and activity of matrix metalloproteinase 2 and 9 in human 480

trophoblasts. Placenta. 24, 53-64.

Jacob M.P. (2003). Extracellular matrix remodeling and matrix metalloproteinases in the

vascular wall during aging and in pathological conditions. Biomed. Pharmacother. 57, 195-202.

Jones R.L., Findlay J.K., Farnworth P.G., Robertson D.M., Wallace E. and Salamonsen L.A.

(2006). Activin A and inhibin A differentially regulate human uterine matrix 485

metalloproteinases: potential interactions during decidualization and trophoblast invasion.

Endocrinology. 147, 724-732.

25

Kaitu'u T.J., Shen J., Zhang J., Morison N.B. and Salamonsen L.A. (2005). Matrix

metalloproteinases in endometrial breakdown and repair: functional significance in a mouse

model. Biol. Reprod. 73, 672-680. 490

Kam E.P., Gardner L., Loke Y.W. and King A. (1999). The role of trophoblast in the

physiological change in decidual spiral arteries. Hum. Reprod. 14, 2131-2138.

Kaufmann P., Huppertz B. and Frank H.G. (1996). The fibrinoids of the human placenta:

origin, composition and functional relevance. Ann. Anat. 178, 485-501.

Khong T.Y., Adema E.D. and Erwich J.J. (2003). On an anatomical basis for the increase in 495

birth weight in second and subsequent born children. Placenta. 24, 348-353.

Kim Y.M., Bujold E., Chaiworapongsa T., Gomez R., Yoon B.H., Thaler H.T., Rotmensch S.

and Romero R. (2003). Failure of physiologic transformation of the spiral arteries in patients

with preterm labor and intact membranes. Am. J. Obstet. Gynecol. 189, 1063-1069.

Lash G.E., Schiessl B., Kirkley M., Innes B.A., Cooper A., Searle R.F., Robson S.C. and 500

Bulmer J.N. (2006). Expression of angiogenic growth factors by uterine natural killer cells

during early pregnancy. J. Leukoc. Biol. 80, 572-580.

Lash G.E., Naruse K., Robson A., Innes B.A., Searle R.F., Robson S.C. and Bulmer J.N.

(2011). Interaction between uterine natural killer cells and extravillous trophoblast cells: effect

on cytokine and angiogenic growth factor production. Hum. Reprod. 26, 2289-2295. 505

Martinez-Lemus L.A. (2012). The dynamic structure of arterioles. Basic Clin. Pharmacol.

Toxicol. 110, 5-11.

Mulholland J., Aplin J.D., Ayad S., Hong L. and Glasser S.R. (1992). Loss of collagen type VI

from rat endometrial stroma during decidualization. Biol. Reprod. 46, 1136-1143.

26

Mylona P., Kielty C.M., Hoyland J.A. and Aplin J.D. (1995). Expression of type VI collagen 510

mRNAs in human endometrium during the menstrual cycle and first trimester of pregnancy. J.

Reprod. Fertil. 103, 159-167.

Naruse K., Lash G.E., Innes B.A., Otun H.A., Searle R.F., Robson S.C. and Bulmer J.N.

(2009). Localization of matrix metalloproteinase (MMP)-2, MMP-9 and tissue inhibitors for

MMPs (TIMPs) in uterine natural killer cells in early human pregnancy. Hum. Reprod. 24, 553-515

561.

Pijnenborg R., D'Hooghe T., Vercruysse L. and Bambra C. (1996). Evaluation of trophoblast

invasion in placental bed biopsies of the baboon, with immunohistochemical localisation of

cytokeratin, fibronectin, and laminin. J. Med. Primatol. 25, 272-281.

Pijnenborg R., Ball E., Bulmer J.N., Hanssens M., Robson S.C. and Vercruysse L. (2006a). In 520

vivo analysis of trophoblast cell invasion in the human. Methods Mol. Med. 122, 11-44.

Pijnenborg R., Vercruysse L. and Hanssens M. (2006b). The uterine spiral arteries in human

pregnancy: facts and controversies. Placenta. 27, 939-958.

Plaisier M., Koolwijk P., Willems F., Helmerhorst F.M. and van Hinsbergh V.W. (2008).

Pericellular-acting proteases in human first trimester decidua. Mol. Hum. Reprod. 14, 41-51. 525

Robson A., Lash G.E., Innes B.A., Robson S.C. and Bulmer J.N. (2009). Potential Role of

VEGF-C, Ang2 and IFN-gamma in Spiral Artery Remodelling. Reproductive Sciences. 16,

110A-111A (146).

Robson A., Harris L.K., Innes B.A., Lash G.E., Aljunaidy M.M., Aplin J.D., Baker P.N.,

Robson S.C. and Bulmer J.N. (2012). Uterine natural killer cells initiate spiral artery 530

remodeling in human pregnancy. FASEB J. 26, 4876-4885.

27

Salamonsen L.A. and Lathbury L.J. (2000). Endometrial leukocytes and menstruation. Hum.

Reprod. Update. 6, 16-27.

Shekhonin B.V., Domogatsky S.P., Muzykantov V.R., Idelson G.L. and Rukosuev V.S. (1985).

Distribution of type I, III, IV and V collagen in normal and atherosclerotic human arterial wall: 535

immunomorphological characteristics. Coll. Relat. Res. 5, 355-368.

Smith S.D., Dunk C.E., Aplin J.D., Harris L.K. and Jones R.L. (2009). Evidence for immune

cell involvement in decidual spiral arteriole remodeling in early human pregnancy. Am. J.

Pathol. 174, 1959-1971.

Wells M., Hsi B.L., Yeh C.J. and Faulk W.P. (1984). Spiral (uteroplacental) arteries of the 540

human placental bed show the presence of amniotic basement membrane antigens. Am. J.

Obstet. Gynecol. 150, 973-977.

Wewer U.M., Faber M., Liotta L.A. and Albrechtsen R. (1985). Immunochemical and

ultrastructural assessment of the nature of the pericellular basement membrane of human

decidual cells. Lab. Invest. 53, 624-633. 545

28

Figure legends

Figure 1. Serial sections showing immunolocalization of collagen type III in the decidua 550

basalis and spiral arterioles during remodeling in vivo. (A) Collagen (col) type III in the

decidual stroma. (B) Control IgG. (C) Col III expression in unremodeled (stage I) SA; (C’) α-

smooth muscle actin (α-SMA; VSMC); (C’’) HLA-G (EVT). (D) Col III expression in

remodeling (stage II) SA; (D’) α-SMA; (D’’) HLA-G. (E) Col III expression in remodeling

(stage III) SA; (E’) α-SMA; (E’’) HLA-G. (F) Col III expression in remodeled (stage IV) SA; 555

(F’) α-SMA; (F’’) HLA-G. Scale bar = 50µm. Inset = negative control. L = lumen; a =

pericellular col III staining; b = bands of col III; c = disrupted col III fibers; d =

immunoreactivity in the basement membrane; e = reduced col III expression in areas colonized

by interstitial EVT. Serial sections from n=20 individual decidual samples (6-12 weeks

gestation) and a minimum of n=24 vessel sections in each remodeling stage were assessed. 560

Representative images are shown.

Figure 2. Serial sections showing immunolocalization of collagen type IV in the decidua

basalis and spiral arterioles during remodeling in vivo. (A) Collagen (col) type IV in the

decidual stroma. (B) Control IgG. (C) Collagen (col) type IV expression in unremodeled (stage 565

I) SA; (C’) α-SMA (VSMC); (C’’) HLA-G (EVT). BM = basement membrane. (D) Col IV

expression in remodeling (stage II) SA; (D’) α-SMA; (D’’) HLA-G. (E) Col IV expression in

remodeling (stage III) SA; (E’) α-SMA; (E’’) HLA-G. (F) Col IV expression in remodeled

(stage IV) SA; (F’) α-SMA; (F’’) HLA-G. Scale bar = 50µm. Inset = negative control. L =

lumen; a = pericellular col IV staining; b = layers of col IV; BM = basement membrane; c = 570

gaps in the basement membrane. Serial sections from n=24 individual decidual samples (6-12

29

weeks gestation) and a minimum of n=32 vessel sections in each remodeling stage were

assessed. Representative images are shown.

Figure 3. Serial sections showing immunolocalization of collagen type VI in the decidua 575

basalis and spiral arterioles during remodeling in vivo. (A) Collagen (col) type VI in the

decidual stroma. (B) Control IgG. (C) Col VI expression in unremodeled (stage I) SA; (C’) α-

SMA (VSMC); (C’’) HLA-G (EVT). (D) Col VI expression in remodeling (stage II) SA; (D’)

α-SMA (VSMC); (D’’) HLA-G (EVT). (E) Col VI expression in remodeling (stage III) SA;

(E’) α-SMA (VSMC); (E’’) HLA-G (EVT). (F) Col VI expression in remodeled (stage IV) SA; 580

(F’) α-SMA; (F’’) HLA-G. Scale bar = 50µm. Inset = negative control. L = lumen; BM =

basement membrane; a =. pericellular col VI staining; b = adventitia/outer VSMC layers; c =

inner VSMC layers; d = endovascular EVT. Serial sections from n=24 individual decidual

samples (6-12 weeks gestation) and a minimum of n=32 vessel sections in each remodeling

stage were assessed. Representative images are shown. 585

Figure 4. Serial sections showing immunolocalization of laminin in the decidua basalis

and spiral arterioles during remodeling in vivo. (A, B) Laminin in the decidual stroma. (C)

Control IgG. (D) Laminin expression in unremodeled (stage I) SA; (D’) α-SMA (VSMC); (D’’)

HLA-G (EVT). (E) Laminin expression in remodeling (stage II) SA; (E’) α-SMA; (E’’) HLA-590

G. (F) Laminin expression in remodeling (stage III) SA; (F’) α-SMA; (F’’) HLA-G. (G)

Laminin expression in remodeled (stage IV) SA; (G’) α-SMA; (G’’) HLA-G. Scale bar =

50µm. Inset = negative control. L = lumen; BM = basement membrane; a = pericellular laminin

staining; b = layers of laminin. Serial sections from n=24 individual decidual samples (6-12

30

weeks gestation) and a minimum of n=32 vessel sections in each remodeling stage were 595

assessed. Representative images are shown.

Figure 5. Serial sections showing immunolocalization of fibronectin in the decidua basalis

and spiral arterioles during remodeling in vivo.

(A) Fibronectin in the decidual stroma. (B) Control IgG. (C) Fibronectin expression in 600

unremodeled (stage I) SA; (C’) α-SMA (VSMC); (C’’) HLA-G (EVT). (D) Fibronectin

expression in remodeling (stage II) SA; (D’) α-SMA; (D’’) HLA-G. (E) Fibronectin expression

in remodeling (stage III) SA; (E’) α-SMA; (E’’) HLA-G. (F) Fibronectin expression in

remodeled (stage IV) SA; (F’) α-SMA; (F’’) HLA-G. Scale bar = 50µm. Inset = negative

control. L = lumen; a = pericellular fibronectin staining; b = immunopositive VSMC; c = 605

interstitial EVT; d = endovascular EVT; e = luminal EVT. Serial sections from n=24 individual

decidual samples (6-12 weeks gestation) and a minimum of n=32 vessel sections in each

remodeling stage were assessed. Representative images are shown.

Figure 6. Serial sections showing immunolocalization of fibrillin and elastin in the decidua 610

basalis and spiral arterioles during remodeling in vivo. (A) Fibrillin expression in

unremodeled (stage I) SA identified by (A’) α-SMA immunostaining. (B) Fibrillin expression

in remodeled (stage IV) SA identified by (B’) HLA-G immunostaining. (C) Control IgG. (D)

Fibrillin in the decidual stroma. (E) Elastin expression in unremodeled (stage I) SA identified

by (E’) α-SMA immunostaining. (F) Control IgG. (G) Elastin expression in remodeling (stage 615

III/IV) SA identified by (G’) α-SMA and (G’’) HLA-G immunostaining. Scale bar = 50µm. L =

lumen; a = VSMC; b = decidual stroma; c = pericellular fibrillin expression; d = internal elastic

31

lamina; e = immunopositive decidual cell; BM = basement membrane. Serial sections from

n=20 individual decidual samples (6-12 weeks gestation) and a minimum of n=24 vessel

sections in each remodeling stage were assessed. Representative images are shown. 620

Figure 7. Serial sections showing PAS staining in the decidua basalis and spiral arterioles

during remodeling in vivo. (A) PAS staining in unremodeled (stage I) SA; (A’) α-SMA

(VSMC); (A’’) HLA-G (EVT). (B) PAS staining in remodeling (stage II/III) SA; (B’) α-SMA;

(B’’) HLA-G. (C) PAS staining in in remodeled (stage IV) SA; (C’) α-SMA; (C’’) HLA-G. 625

Note endovascular EVT embedded within the fibrinoid (arrowheads). Scale bar = 50µm. L =

lumen; BM = basement membrane; a = pericellular PAS staining; b = reduced staining in areas

of VSMC loss; c = PAS-positive fibrinoid; d = EVT. Serial sections from n=20 individual

decidual samples (6-12 weeks gestation) and a minimum of n=24 vessel sections in each

remodeling stage were assessed. Representative images are shown. 630

32

Table 1. Details of antibodies used for immunohistochemistry

Antibody Source Working

concentration

Dilution

factor

Specificity

Human Leukocyte

Antigen–G (mouse)

Abcam, UK 2µg/ml 1:100 Extravillous

trophoblast

Anti-α-SMA (mouse) Dako, UK 0.18µg/ml 1:800 Vascular

smooth muscle

Anti-collagen type III

(mouse)

Abcam, UK 4.8µg/ml 1:500 Collagen type

III

Anti-collagen type IV

(mouse)

Sigma

Aldrich, UK

1µg/ml 1:200 Collagen type

IV

Anti-collagen type VI

(rabbit)

Sigma

Aldrich, UK

1µg/ml 1:500 Collagen type

VI

Anti-laminin (gamma

1; mouse)

Sigma

Aldrich, UK

4µg/ml 1:50 Laminin

Anti-fibronectin

(mouse)

Abcam, UK 1.8µg/ml 1:100 Fibronectin

Anti-elastin (rabbit) Sigma

Aldrich, UK

4µg/ml 1:250 Elastin

Anti-fibrillin (rabbit) Dr C Kielty,

University of

Manchester,

UK

4µg/ml 1:500 Fibrillin

Isotype control

mouse IgG

Sigma

Aldrich, UK

Matched to

each primary

antibody

- Negative

control

Non immune rabbit

IgG

Sigma

Aldrich, UK

Matched to

each primary

antibody

- Negative

control

Goat anti-mouse IgG Dako, UK 7µg/ml 1:200 Mouse IgG

Swine anti-rabbit IgG Dako, UK 4µg/ml 1:400 Goat IgG

635

33

Table 2. A semi-quantitative summary of changes in vascular extracellular matrix composition

observed during remodeling of decidual SA, as agreed by 2 independent assessors

Stage

Collagen

type III

Collagen

type IV

Collagen

type VI

Laminin Fibronectin Fibrillin Elastin Fibrinoid Intramural

trophoblast

Vascular

smooth

muscle

I +++ +++ + ++ - ++ ++ - - +++

II ++ + + - - ++ + - - ++

III ++ + + - ++ ++ + - ++ +

IV ++ + + - ++ ++ + ++ +++ -

640

Figure 1: Collagen type III

[C’] α-SMA[C] Col III, stage I [C’’] HLA-G

[D’] α-SMA[D] Col III, stage II [D’’] HLA-G

[F’] α-SMA[F] Col III, stage IV [F’’] HLA-G

[B] IgG[A] Col III

VSMC

EVT

VSMC

[E’] α-SMA[E] Col III, stage III [E’’] HLA-G

VSMC

EVT

L

L

L

L

L

L

b

b

c

c

c

d

dd

a

e

Figure 2: Collagen type IV

[B] IgG

α-SMA

[C’’] HLA-G[C’] α-SMA

[A] Col IV

[D’’] HLA-G[D’] α-SMA[D] Col IV, stage II

[F’’] HLA-G[F’] α-SMA[F] Col IV, stage IVEVT

VSMC

VSMC

VSMC[C] Col IV, stage I

BM

[E’’] HLA-G[E’] α-SMA[E] Col IV, stage III

VSMC

L

L

L

L

c

bb

c

BM

c

BM

a

EVT

Figure 3: Collagen type VI

[A] Col VI

[F’] α-SMA[F] Col VI, stage IV [F’’] HLA-G

[B] IgG

VSMC

EVT

EVT

L

[D’] α-SMA[D] Col VI, stage II [D’’] HLA-G

[C’] α-SMA[C] Col VI, stage I [C’’] HLA-G

VSMC

EVT

L

L

L

[E’] α-SMA[E] Col VI, stage III [E’’] HLA-G

EVT

VSMC

VSMC

a

b

bc

cb

b

d

BM

BM

Figure 4: Laminin

[D’] α-SMA[D] Laminin, stage I [D’’] HLA-G

[E’] α-SMA [E’’] HLA-G

[F] Laminin, stage III

[E] Laminin, stage II

[A] Laminin [C] IgG[B] Laminin

VSMC

VSMC

[G’] α-SMA [G’’] HLA-GEVT

[F’] α-SMA [F’’] HLA-GVSMC

EVT

[G] Laminin, stage IV

LL

L

L

L

BMb BM

b

VSMC

a

a

Figure 5: Fibronectin

[C’] α-SMA[C] FN, stage I [C’’] HLA-G

[E] FN, stage III

[F] FN, stage IV

[A] FN [B] IgG

VSMC

[E’] α-SMA [E’’] HLA-G

VSMC

EVT

[F’] α-SMA [F’’] HLA-G EVT

[D] FN, stage II [D’] α-SMA [D’’] HLA-G

VSMC

L

L

L

L

L

a

b

e

dc

Figure 6: Fibrillin/Elastin

Fibrillin

[A’] α-SMA[A] Fibrillin, stage I [C] IgG

[B’] HLA-G[B] Fibrillin, stage IV [D] Fibrillin

[E’] α-SMA[E] Elastin, stage I [F] l IgG

[G’] α-SMA[G] Elastin, stage III/IV [G’’] HLA-G

EVT

VSMC

VSMC

VSMC

EVT

L

L

L

L

L

b

a

c

c

b

e

d

dd

d

BM

Figure 7: Fibrinoid

[A’] α-SMA

VSMC

[B’] α-SMA

VSMC

[C’’] HLA-G

EVT

[A] PAS, stage I

L

[B] PAS, stage II/III

L

[C] PAS, stage IV

d

L

[B’’] HLA-G

[A’’] HLA-G

[C’] α-SMA

BM a

a

b

d

dd d

cVSMC

d