A stereological perspective on placental morphology in normal and complicated pregnancies

Abstract

Stereology applied to randomly-generated thin sections allows minimally-biased and economical quantitation of the 3D structure of the placenta from molecular to whole-organ levels. With these sampling and estimation tools, it is possible to derive global quantities (tissue volumes, interface surface areas, tubule lengths and particle numbers), average values (e.g. mean cell size or membrane thickness), spatial relationships (e.g. between compartments and immunoprobes) and functional potential (e.g. diffusive conductance). This review indicates ways in which stereology has been used to interpret the morphology of human and murine placentas including the processes of villous growth, trophoblast differentiation, vascular morphogenesis and diffusive transport. In human placenta, global quantities have shown that villous maturation involves differential growth of fetal capillaries and increases in endothelial cell number. Villous trophoblast is a continuously renewing epithelium and, through much of gestation, exhibits a steady state between increasing numbers of nuclei in cytotrophoblast (CT) and syncytiotrophoblast (ST). The epithelium gradually becomes thinner because its surface expands at a faster rate than its volume. These changes help to ensure that placental diffusing capacity matches the growth in fetal mass. Comparable events occur in the murine placenta. Some of these processes are perturbed in complicated pregnancies: 1) fetoplacental vascular growth is compromised in pregnancies accompanied by maternal asthma, 2) changes in trophoblast turnover occur in pre-eclampsia and intrauterine growth restriction, and 3) uteroplacental vascular development is impoverished, but diffusive transport increases, in pregnant mice exposed to particulate urban air pollution. Finally, quantitative immunoelectron microscopy now permits more rigorous analysis of the spatial distributions of interesting molecules between subcellular compartments or shifts in distributions following experimental manipulation.

Introduction

Revealing the internal structure of the placenta often involves viewing sectional images on physical, optical or medical slices. Whilst slicing permits visualization of substructure (including localization of interesting molecules) at adequate resolution, it also causes loss of dimensional information, which, if ignored, has the potential to confuse interpretations of 3D spatial composition and organization. Fortunately, by randomizing the sampling of placentas, including the positions and directions of slices through them, the 3D properties of individual components (e.g. the average CT cell), sets of components (e.g. the terminal villi) or the average organ can be estimated efficiently and with minimal bias. Sampling methods, including those suitable for murine and human placentas, have been reviewed elsewhere (Coan et al. 2004; Mayhew, 2008; Veras et al. 2008).

Stereology (Howard & Reed, 2005; Mayhew, 2006a) combines random sampling with estimation tools to derive global quantities (total volumes, surfaces, lengths, numbers), average sizes (e.g. mean cell volume, mean layer thickness), spatial relationships (e.g. between immunogold markers and subcellular structures) or functional indices (e.g. diffusive conductance). Generating quantitative data in this way facilitates the interpretation of 3D functional morphology during development, complicated pregnancy and experimental manipulation.

The aim here is to emphasize the practical value of stereology in microscopical studies based on thin sections of human and animal placentas. Findings from recent studies illustrate how spatial quantities can be used to describe normal development of the placenta and the ways in which morphology is altered in complicated pregnancies. Topics include 1) fetoplacental vascular growth in pregnancies accompanied by maternal asthma, 2) trophoblast turnover in pre-eclampsia (PE) and intrauterine growth restriction (IUGR), 3) maternal vascular development and diffusive transport in placentas from mice exposed to polluted air, and 4) quantifying the spatial distributions of interesting molecules using high-resolution immunocytochemistry.

Morphometric descriptors of relevant processes

Growth and development

The growth of villi, fetoplacental vessels and maternal vascular spaces can be expressed in terms of volumes, surface areas, lengths or numbers. Volume is the measure of size in 3D and placental compartments are 3D spaces constructed from lower levels of organization (organelles, cells, extracellular matrices, tissues). The volumes of these spaces provide measures of their mass or bulk and so can be used not only to monitor growth but also to indicate the likely demands for nutrients and respiratory gases. Volume may also provide an index of supply, e.g. the volume of a vascular space is related to the transport of gases, nutrients or heat.

The surfaces of villi, capillaries and intervascular barriers act as interfaces between compartments and are involved in passive transport (e.g. diffusion of respiratory gases) and incorporate molecules active in transport and recognition (carriers, ion channels, receptors), cell-cell and cell-matrix adhesion (tight junctions, adhaerens junctions, desmosomes), communication (gap junctions) and metabolism (membrane-bound enzymes). Like volume, surface area can be employed to monitor growth.

Length, another measure of growth, describes the linear extent of arborizations (e.g. villi) and tubular structures (e.g. capillaries). The linear features may be straight or curved, continuous or discontinuous, branched or unbranched.

Particle number provides biologically useful information in two main contexts: (1) genesis, growth and transformation, and (2) communication and connectivity.

Genesis, growth and transformation

Growth may occur by proliferation (cell division), hypertrophy (cell growth) or accretion (interstitial or extracellular growth). For instance, an increase in placental size may depend on the number of cells, the sizes of a fixed set of cells or syncytium and the quantity of extracellular matrix.

Communication and connectivity

Vascular endothelial and smooth muscle cells communicate via gap-junctions. Where communication is due to molecular interactions, usually a marker or label will indicate a site of enzymic or other activity or the location of a specific protein (e.g. connexins). In immunoelectron microscopy, the markers are gold particles which are counted to monitor shifts in localization between study groups or to quantify labelling intensities of different compartments within a cell or syncytium.

The maturation status of villi might be expressed in terms of their calibre, which in human placenta declines as one passes along the topological sequence from stem villi to terminal villi. A convenient and simple way of estimating this is to divide total volume by total length. For linear structures, this equates to mean cross-sectional area and offers a less biased estimator of calibre than mean diameter (Kaufmann et al. 2004). Other useful measures of maturation are the vascularization or capillarization indices, which include volume, surface or length densities and capillary : villus surface or length ratios (Kaufmann et al. 2004).

Angiogenesis

Angiogenic changes may be quantified effectively by estimating nett growth (total capillary volume, surface, length or endothelial cell number) and indices of villous capillarization and capillary calibre (Kaufmann et al. 2004). Again, the latter can be estimated as mean cross-sectional area. To assess the contributions made by branching, estimates are required of the numerical density of branch sites and this information can be combined with length data to estimate segment lengths (Gambino et al. 2002). However, microvascular arrangements and branching patterns can be analysed directly in 3D by scanning electron microscopy or confocal microscopy (e.g. Jirkovska et al. 2002).

Villous trophoblast recruitment and loss

Trophoblast presents a number of different compartments which reflect its phases of differentiation (Mayhew, 2001). The volumes of these compartments (including CT and ST) can be estimated, together with their surface areas or complements of nuclei. It is possible also to subdivide the ST compartment to account for areas of thinning (vasculosyncytial membranes) or thickening (syncytial knotting) or loss of epithelial integrity due to damage (Mayhew & Barker, 2001). By determining relevant ratios (numerical or surface), the epithelial steady state between CT and ST, and the association between trophoblast compartments and perivillous fibrin-type fibrinoid, can be evaluated.

Passive diffusion

As it grows, the fetus requires more oxygen and nutrients from the mother. Oxygen crosses the placenta by passive diffusion as do certain nutrients and other gases. For the intervascular barrier, diffusive transport is partly determined by its physical dimensions, including its surface areas and effective diffusion distances. The total volume of a barrier, divided by surface area, can be used to express an arithmetic mean thickness, e.g. the mean depth of the trophoblastic epithelium. Barrier thickness is inversely proportional to diffusive conductance but, in this context, it is preferable to monitor harmonic rather than arithmetic mean thickness because harmonic means deal more effectively with barriers of varying local thicknesses.

The capacity of the human placenta for passive diffusion is expressed as a total conductance in cm³ min⁻¹ kPa⁻¹ (Mayhew et al. 1990, 1993a,b; Ansari et al. 2003). In the human placenta, oxygen dissociates from haemoglobin in maternal erythrocytes and crosses maternal plasma, trophoblast, fetal endothelium and plasma before binding to haemoglobin in fetal erythrocytes. Consequently, at least six tissue compartments offer resistances to diffusion and these can be summed to derive a total resistance, the reciprocal of which is total conductance. Serial resistances can be calculated from vascular space volumes and oxygen–haemoglobin chemical reaction rates, surface areas, harmonic mean thicknesses and permeability coefficients.

The intervascular barrier of the murine placenta differs in composition from that of the human placenta and estimates of its diffusive conductance have been confined to the intervascular tissue rather than the placenta as a whole (Coan et al. 2004; Veras et al. 2008). This intervascular barrier comprises a layer of CT, two layers of ST and a layer of fetal vascular endothelium.

In reality, several factors (e.g. vascular shunts and local perfusion : diffusion ratio inequalities) reduce the effectiveness of the placenta for passive diffusion. Consequently, this approach provides the maximal diffusive conductances achievable under optimal conditions. Nevertheless, estimates have real comparative worth. Moreover, mass-specific conductances can be calculated by relating total conductances to fetal weight (Mayhew et al. 1993a,b; Coan et al. 2004).

Spatial shifts in molecular localization

Molecules detected by high-resolution immunocytochemistry localize in different ultrastructural compartments which may be volumes (e.g. nucleus, mitochondria, cytosol) or surfaces (e.g. plasma membrane, cristae membranes). Their presence in different compartments can be assessed by counting immunogold particles. The conventional way of evaluating intensity of labelling has been to relate numbers of gold particles to compartment profile areas or trace lengths, an index known as labelling density, LD (Griffiths, 1993). Recently, more efficient ways of estimating LD have been devised and these can be used to express preferential labelling of compartments as a relative labelling index. This provides a measure of the degree to which a compartment is labelled in comparison with random labelling (when all compartments are expected to show the same LD). By combining these approaches with appropriate statistical tests, it is now possible to compare labelling patterns between compartments in a cell or to follow shifts in patterns between groups of cells (Mayhew & Lucocq, 2008b). Between-group comparisons rely simply on counts of gold particles associated with compartments and do not require estimates of compartment sizes.

The following examples illustrate the use of these measures to study the placenta.

Morphological features of normal placental development

Human placenta

Villous trees are established in two phases (Benirschke & Kaufmann, 2000; Kaufmann et al. 2004): an early phase provides the main branches (stem and immature intermediate villi) and, starting around mid-gestation, a later phase establishes a multitude of fine peripheral branches (mature intermediate and terminal villi). Terminal villi have an extensive surface (>10 m² at term) and small calibre (40–100 µm) and are crucially important for transplacental exchanges. Their trophoblast comprises inner proliferative CT cells and an outer differentiating ST which surround a mesodermal stroma containing fetoplacental capillaries. In the third trimester, the main tissue layers intervening between the maternal and fetoplacental circulations of these villi are the ST, scattered CT cells, epithelial basal laminae and variable amounts of stroma, and vascular endothelium. Local thinning of trophoblast, coupled with peripheralization of capillaries, reduces effective diffusion distances by creating vasculosyncytial membranes (Benirschke & Kaufmann, 2000). The incompleteness of the CT layer later in gestation contributes to this thinning and explains why the intervascular barrier is described as haemomonochorial. In fact, CT cells maintain contact via contiguous processes which radiate from the cell body and seem to form a functional continuum (Jones et al. 2008).

Villous growth is influenced by fetoplacental angiogenesis which is also biphasic and involves changes in the number and dimensions of vessel segments: an initial phase of capillary branching is followed by one of greater non-branching angiogenesis (Benirschke & Kaufmann, 2000; Mayhew, 2002; Kaufmann et al. 2004). In a vessel segment, resistance to blood flow is proportional to length and inversely proportional to the square of cross-sectional area. For a set of segments, total resistance depends on whether the arrangement is parallel or serial. If parallel, total resistance is less than the partial resistance of any individual segment but, for a serial arrangement, it is equal to the sum of partial resistances. Consequently, parallel arrangements (produced by branching angiogenesis) tend to be better than serial arrangements (produced by non-branching angiogenesis) because they offer smaller overall impedance.

Branching angiogenesis occurs in two main ways (Djonov et al. 2003; Charnock-Jones et al. 2004). Sprouting creates side branches whilst intussusception involves creating two lumina by means of endothelial ingrowth. By contrast, non-branching angiogenesis involves elongation of existing vessel segments and this may be driven by either or both of two processes: endothelial cell proliferation and intercalation of endothelial progenitor cells (Charnock-Jones et al. 2004).

During the first trimester, large-calibre immature intermediate villi (with a complex capillary network surrounding larger central vessels) are common. In the third trimester, there is a preponderance of slender terminal villi containing one to two peripheral capillary loops (Leach et al. 2002). Phased changes in indices of villous capillarization, notably capillary : villus length ratios (Mayhew, 2002), are consistent with the notion that angiogenesis influences villous differentiation. Reductions in the mean calibres of fetal capillaries have been reported in some studies (Kaufmann et al. 2004).

Mouse placenta

The definitive placenta of the mouse is haemotrichorial and divisible into three main zones: the labyrinth, junctional zone and decidua basalis. The labyrinth lies closest to the chorionic plate and contains the irregularly shaped maternal vascular spaces and fetoplacental capillaries. Separating the two circulations is the intervascular barrier, which comprises a superficial layer of CT cells and two layers of ST. Recent studies from embryonic days E12.5 to E18.5 (Coan et al. 2004) have shown that the placenta reaches its maximum size by E16.5, with the labyrinth continuing to expand thereafter and at a faster rate than the other two zones. Within the labyrinth, the volumes and surfaces of the maternal vascular spaces expand rapidly until E16.5, but the growth of fetal capillaries continues from E12.5 to E18.5. The changes in fetal capillary volumes are associated with increases in total length and reductions in mean cross-sectional area.

Trophoblast volume increases between E12.5 and E16.5 and, this, together with the growth in maternal and fetal exchange surface areas, leads to a significant reduction in thickness of the intervascular barrier between E14.5 and E16.5 (Coan et al. 2004). This is mainly due to thinning of the CT and vascular endothelium components (Coan et al. 2005).

During human and murine pregnancies, oxygen diffusive conductances are matched to fetal weight despite the fact that tissue compartments experience different growth trajectories (Mayhew et al. 1993a; Coan et al. 2004; Mayhew, 2006b). In such circumstances, the regression line for a log-log plot of conductance against weight is expected to have a slope equal to 1. In fact, the regression line of log D_vm against log fetal weight (where D_vm signifies the conductance of the villous membrane), has a slope of 1.04 and this value is not significantly different from 1 (Mayhew, 2006b). Studies on murine placentas (Coan et al. 2004) also suggest that the intervascular conductance is commensurate with changes in fetal weight (slope 0.93). When combined with human findings, the results further suggest that mass-specific conductances may be similar across species, implying that different types of placenta may vary in absolute efficiency whilst still meeting the needs of their fetuses.

The effects of maternal asthma on villous growth and fetoplacental angiogenesis

Vascular patterns and villous development alter in various complicated pregnancies and are associated with differences in the expression of angiogenic growth factors and their receptors (Mayhew et al. 2004a). Changes can occur in the total and relative volumes, surface areas and lengths of villi and capillaries, as illustrated by recent findings on the effects of maternal asthma.

Asthma prevalence is increasing and adverse pregnancy outcomes include reduced birth weight, the origins of which are unresolved but might be attributable to asthma severity, drug treatment, placental vascular function or fetal hypoxia (Clifton et al. 2001; Bracken et al. 2003; Schatz et al. 2006). Drug treatment includes the use of glucocorticoids and, recently, we have examined the effects of asthma severity and glucocorticoid treatment on placental morphology (for details, see Mayhew et al. 2008).

Pregnant women (n = 60 asthmatics, n = 15 non-asthmatics) were recruited in the first trimester and measurements made of age, body weight, haematocrit, vital capacity and forced expiratory volume. Pregnancy complications other than asthma were excluded. Asthmatics were classified on the basis of asthma severity or glucocorticoid treatment. Here, findings are presented for four asthmatic groups: those not receiving steroids (n = 28) and those receiving antenatal glucocorticoids at low (n = 10), moderate (n = 15) or high (n = 7) levels of usage.

Immediately after delivery, umbilical cords were clamped close to their placental insertion. Birth and trimmed placental weights were recorded before full-depth columns of tissue were sampled systematically, diced and immersed in 10% phosphate-buffered formalin-saline. Tissue cubes were allowed to settle haphazardly in paraffin wax to randomize the positions and orientations of encounters between tissues and section planes. Microscopical fields on stained sections were selected by systematic uniform random (SUR) sampling and analysed stereologically to estimate the volumes of placental compartments and total lengths of villi and fetal capillaries. Mean cross-sectional areas of peripheral villi and capillaries, together with measures of villous capillarization (capillary volume densities and capillary : villus length ratios), were derived from global volumes and lengths. Groups were compared by analysis of variance and post-hoc testing (Mayhew et al. 2008).

Findings are summarized in Tables 1 and 2. Most of the differences associated with asthma severity were also detected when groups were classified according to glucocorticoid usage (Mayhew et al. 2008). We found significant between-group differences in placental composition, the main changes involving fetoplacental capillaries in peripheral villi. Compared with non-asthmatic controls, significant effects on villous maturity were detected, notably decreases in volumes of fetal capillaries and lengths of villi and fetal capillaries in the group with high steroid use. Capillary volumes were also reduced in the group with low steroid use. Smaller mean cross-sectional areas of villi and capillaries were found in the group not receiving glucocorticoid treatment (Table 2). In addition, there were group differences in capillary : villus length ratios, which were lower than controls in those with high steroid use. Changes in villus and capillary calibres were confined to asthmatics not receiving steroid treatment.

Table 1.

Maternal, neonatal and placental characteristics in non-asthmatic controls and in asthmatics grouped by glucocorticoid (GC) use. Values are group means (CVs expressed as % of means). Data based on Mayhew et al. (2008)

Variable	Controls (n = 15)	No GCs (n = 28)	Low GCs (n = 10)	Moderate GCs (n = 15)	High GCs (n = 7)
*‡Age, years	29.4 (15%)	24.3 (19%)§	29.5 (20%)	26.2 (20%)	26.6 (14%)
†Weight 36 weeks, kg	82.5 (13%)	88.7 (20%)	86.3 (28%)	85.0 (27%)	86.0 (38%)
Gestation, weeks	39.9 (3%)	40.2 (4%)	39.3 (3%)	39.8 (3%)	40.1 (3%)
Birth weight, kg	3.55 (13%)	3.65 (12%)	3.37 (17%)	3.42 (17%)	3.30 (17%)
Placental volume, mL	652 (21%)	609 (16%)	569 (19%)	615 (20%)	519 (11%)
HCTmat, %	36 (8%)	35 (8%)	36 (4%)	35 (8%)	36 (5%)
HCTf, %	46 (13%)	48 (15%)	46 (12%)	48 (15%)	50 (9%)

HCTmat and HCTf refer to maternal and fetal haematocrits, respectively. Results of 2-way ANOVA and post hoc testing:

^*significant group effect;

^†significant sex effect;

^‡significant interaction (group × sex) effect;

^§significantly different from non-asthmatic controls.

Table 2.

Morphometric indices of placental composition, villous capillarization and the mean cross-sectional areas of peripheral villi and capillaries in non-asthmatic controls and in asthmatics grouped by glucocorticoid (GC) use. Values are group means (CVs expressed as % of means). Data based on Mayhew et al. (2008)

Variable	Controls (n = 15)	No GCs (n = 28)	Low GCs (n = 10)	Moderate GCs (n = 15)	High GCs (n = 7)
Intervillous space, mL	213 (16%)	192 (18%)	183 (23%)	200 (23%)	163 (22%)
Stem villi, mL	71.4 (68%)	57.5 (41%)	72.3 (52%)	81.9 (71%)	66.0 (56%)
Peripheral villi, mL	326 (27%)	308 (20%)	268 (24%)	289 (30%)	241 (15%)
Trophoblast, mL	95.5 (29%)	89.3 (18%)	83.9 (26%)	85.2 (35%)	74.5 (14%)
Stroma, mL	184 (32%)	174 (25%)	154 (27%)	160 (31%)	143 (16%)
*Fetal capillaries, mL	46.9 (50%)	44.1 (42%)	30.5 (24%)‡	43.4 (43%)	23.7 (35%)‡§
Non-parenchyma, mL	41.5 (39%)	52.5 (43%)	44.9 (32%)	44.7 (48%)	49.8 (30%)
*†Peripheral villi, km	89.2 (23%)	103 (22%)	86.5 (33%)	86.6 (36%)	68.6 (19%)‡§
*Fetal capillaries, km	310 (37%)	382 (40%)	259 (39%)	359 (43%)	169 (44%)‡§
TS area villi, µm2	3700 (21%)	3020 (17%)‡	3360 (37%)	3530 (26%)	3560 (12%)¶
TS area capillary, µm2	150 (27%)	119 (26%)‡	126 (25%)	130 (41%)	149 (36%)
Capillaries, mL mL−1	0.147 (47%)	0.144 (33%)	0.116 (19%)	0.151 (33%)	0.098 (35%)
*Length ratio, km km−1	3.6 (38%)	3.7 (30%)	3.0 (26%)	4.3 (33%)	2.4 (31%)‡,§

Results of 2-way ANOVA and post-hoc testing:

^*significant group effect;

^†significant sex effect;

^‡significantly different from non-asthmatic controls;

^§significantly different from no GC and moderate GC groups;

^¶significantly different from no GC group.

Decreases in capillary volumes might be attributable to poor asthma control leading to fetal hypoxia, asthma severity, treatment by inhaled glucocorticoids, or other factors (Schatz et al. 2006). Because glucocorticoid status is related to, and not isolated from, asthma severity, resolution of these possibilities requires further studies. In some instances, the different origins of fetal hypoxia can be distinguished by studying patterns of villous development and changes in angiogenic growth factors (Kingdom & Kaufmann, 1997; Charnock-Jones et al. 2004; Mayhew et al. 2004a). While the morphological changes seen in asthmatic pregnancies superficially resemble postplacental hypoxia, we did not detect any increases in fetal haematocrits, although these have been reported previously (Littner et al. 2003).

The association between hypocapillarization of villi and steroid status raises the possibility that glucocorticoids compromise vascular morphogenesis. The effects may be mediated by hypoxia-inducible transcription factor-1 (HIF-1), vascular endothelial growth factor-A (VEGF) and other angiogenic factors. Administration of oral glucocorticoids to rat dams leads to reduced VEGF expression and fetoplacental capillary growth (Hewitt et al. 2006). At other tissue sites, glucocorticoids have been shown to impair HIF-1 function and affect angiogenesis via VEGF and angiopoietin-1 (Jośko et al. 2007; Kim et al. 2008; Wagner et al. 2008). Antenatal administration to rat and sheep dams also reduces fetal and placental weights and these effects may be mediated by decreased expression of genes involved in cell proliferation and trophoblast apoptosis (Kerzner et al. 2002; Baisden et al. 2007).

Short-term infusion of glucocorticoids into the fetoplacental vasculature leads to arterial dilation, whilst the effects of intramuscular betamethasone on IUGR fetuses with absent or reversed flow in umbilical arteries include responses which increase the risk of perinatal mortality (Clifton et al. 2001, 2002; Simchen et al. 2004). In asthmatics, including those on high-dosage glucocorticoids, Doppler flow-velocity waveforms in umbilical arteries were reduced at 18 weeks of gestation but not at 30 weeks. These temporal differences are interesting because mid-gestation is critical in terms of villous growth, fetoplacental angiogenesis, and changes in oxygen tensions (Mayhew, 2002; Charnock-Jones et al. 2004; Kaufmann et al. 2004). Whilst inflammatory cytokines may also affect vascular morphogenesis as inhibitors or activators of angiogenesis (Clifton & Murphy, 2004; Romagnani et al. 2004; Salvucci et al. 2004), resolution of this possibility requires further investigation.

Human villous trophoblast and the effects of PE and IUGR

PE and IUGR are pregnancy complications responsible for significant perinatal morbidity and mortality. They may occur singly or together and both are associated with changes in placental morphology. Although poor implantation occurs in both types of pregnancy, stereological studies have shown that differences in placental morphology, particularly at the villous membrane, distinguish PE from IUGR (Teasdale, 1984, 1985, 1987; Egbor et al. 2006; Mayhew et al. 2007). The changes are associated with differences in diffusive conductances (Mayhew et al. 2007). In IUGR, with or without PE, total diffusive conductances are reduced in comparison with control or PE subjects.

Villous trophoblast exhibits phases of proliferation, recruitment, differentiation and loss (Huppertz et al. 1998; Mayhew et al. 1999). Following CT mitosis, some cells fuse into the overlying ST and it has been shown that, whilst total numbers of nuclei in CT and ST increase to term, ST : CT numbers are maintained at about 10 : 1 from at least 13 weeks of gestation to term (Simpson et al. 1992; Mayhew et al. 1994). Once CT cells fuse into the ST, nuclei follow a terminal differentiation pathway which culminates in apoptosis. A protracted pre-apoptotic phase is followed by a shorter execution phase (Mayhew et al. 1999; Benirschke & Kaufmann, 2000) and many apoptotic nuclei congregate in syncytial knots which are extruded into the intervillous space as part of normal epithelial turnover (Huppertz et al. 1998; Mayhew et al. 1999). The ST : CT ratio provides a convenient index of the epithelial steady state between CT proliferation and recruitment vs. ST differentiation and extrusion (Mayhew et al. 1999).

Perturbations of the ST : CT steady state occur in PE and IUGR. Apoptosis increases in PE and trophoblast fragments are shed into the maternal blood in greater numbers (Johansen et al. 1999; Ishihara et al. 2002). Cultured primary villous CT cells from cases of PE, IUGR and PE + IUGR differ in their secretory profiles and ability to form syncytia (Newhouse et al. 2007). Culturing explanted villi at different oxygen tensions has shown that hypoxia favours necrotic over apoptotic shedding (Huppertz et al. 2003) and can compromise fusion of CT into ST and completion of the late apoptotic phase within ST. The result is a shift from apoptosis to aponecrosis (a mixture of apoptosis and secondary necrosis), and aponecrotic shedding is a feature of PE (Huppertz & Kingdom, 2004; Huppertz & Herrler, 2005).

Published findings on villous growth and development and the turnover of trophoblastic epithelium indicate that morphological features distinguish term placentas in PE and IUGR (Table 3). Whilst the total volumes and surfaces of trophoblast in peripheral villi do not alter in PE, they decrease in IUGR and PE + IUGR. It is likely that the differences are attributable to changes in total numbers of nuclei (Teasdale, 1984, 1985, 1987), which are better descriptors of total proliferation and loss than proliferation and apoptosis indices (Mayhew et al. 2003b, 2007). The total complement of trophoblast nuclei is similar in PE and control placentas (Teasdale, 1985) but numbers decline in IUGR and PE + IUGR (Teasdale, 1984, 1987). The findings are consistent with observations on relative rates of trophoblast recruitment and loss: rates of differentiation and loss are accelerated in PE and IUGR (Smith et al. 1997; Axt et al. 1999; Johansen et al. 1999; Allaire et al. 2000; Erel et al. 2001; Ishihara et al. 2002) but rates of CT proliferation are greater in PE than in IUGR (Jones & Fox, 1980; Smith et al. 1998).

Table 3.

Summary of published findings on morphological changes in the villous membrane of peripheral villi found in cases of PE with and without IUGR. Changes are with respect to placentas from control (uncomplicated) pregnancies

Measures	Case Types	Natures of change	References	Comments
Surface area of capillaries	IUGR and PE + IUGR	Decrease	2,5,6,8,9,16,18,22,23, present	Some supposedly PE may be PE + IUGR
	PE	NS	4,16,18,22, present
Volume of trophoblast	IUGR and PE + IUGR	Decrease	2,5,6,8,18
	PE	NS	18
Surface of trophoblast	IUGR and PE + IUGR	Decrease	2,3,5,6,8,9,16,18,22,23, present	Some supposedly PE may be PE + IUGR
	PE	NS
Number of CT nuclei	IUGR and PE + IUGR	Decrease	2,5
	PE	NS	4
Number of ST nuclei	IUGR and PE + IUGR	Decrease	2,5
	PE	NS	4
Thickness of trophoblast	IUGR and PE + IUGR	Increase or NS	18	Arithmetic mean
Thickness of basal lamina	IUGR and PE + IUGR	Increase	1,10,17,20	Worse when IUGR is with ARED
Thickness of villous membrane	IUGR and PE + IUGR	NS or decrease	16,18	Arithmetic and harmonic means
	PE	NS	present	Arithmetic and harmonic mean
CT proliferation rate	IUGR and PE + IUGR	Increase or NS or decrease	1,7,10,12,14,17	Some IUGR with ARED. Nature of change influenced by method and way in which rate is expressed
	PE	Increase or NS	7,21
ST differentiation/loss rates	IUGR and PE + IUGR	Increase or NS	1,11,13,15,19,24,25,26	Methods and choice of rate vary
	PE	Increase	19,21,24,26,27	Increase may vary between early and late onset

NS indicates no significant change; ARED is absent or reversed end-diastolic flow in umbilical arteries. References: 1. Jones & Fox (1980); 2. Teasdale (1984); 3. Boyd & Scott (1985); 4. Teasdale (1985); 5. Teasdale (1987); 6. Pivalizza et al. (1990); 7. Arnholdt et al. (1991); 8. Jackson et al. (1995); 9. Burton et al. (1996); 10. Macara et al. (1996); 11. Smith et al. (1997); 12. Smith et al. (1998); 13. Erel et al. (2001); 14. Chen et al. (2002); 15. Ishihara et al. (2002); 16. Ansari et al. (2003); 17. Madazli et al. (2003); 18. Mayhew et al. (2003a); 19. Austgulen et al. (2004); 20. Battistelli et al. (2004); 21. Huppertz & Kingdom (2004); 22. Mayhew et al. (2004c); 23. Egbor et al. (2006); 24. Goswami et al. (2006); 25. Axt et al. (1999); 26. Allaire et al. (2000); 27. Johansen et al. (1999).

The comparisons in Fig. 1 are based on the morphometric literature summarized in Table 3 together with findings on trophoblast loss in terms of syncytial fragments (whether apoptotic or aponecrotic) and microparticles (MPs, either ST microvillous fragments or assorted cell-free debris). Figure 1 incorporates the following important elements: 1) the surface area of peripheral villi is maintained in PE but declines in IUGR and PE + IUGR; 2) trophoblast growth is determined mainly by increased numbers of nuclei and the total number per placenta is maintained in PE but declines in PE and PE + IUGR; 3) trophoblast thickness is determined by spatiotemporal changes in the steady state and contributes to villous membrane thickness which is affected also by the proximity of subjacent capillaries (related to their calibre, degree of peripheralization and extent of villous capillarization); 4) CT proliferation and ST apoptosis rates are not absolute measures and their impacts depend heavily on the total complements of CT and ST nuclei; 5) in uncomplicated pregnancy, terminal differentiation involves apoptosis but minimal necrotic damage or MP loss. Consequently, trophoblast extruded into the maternal intervillous space normally comprises membrane-bound syncytial fragments containing multiple apoptotic or pre-apoptotic nuclei; 6) PE involves accelerated turnover and failure to complete apoptosis so that ST shows increased aponecrosis and releases greater numbers of syncytial fragments and MPs.

Fig. 1.

Villous membrane morphology in PE and IUGR. In each schematic, height represents the arithmetic thickness of the membrane (from luminal aspect of vascular endothelium to maternal aspect of villous trophoblast). Width is proportional to total villous surface area. CT cells lie on basal lamina and some post-mitotic cells are recruited (small arrows) into ST where terminal differentiation occurs. Turnover involves loss (large arrows) of membrane-bound ST fragments (SFs) and microparticles (MPs). Membrane thickness depends on the steady state between recruitment and loss and on the proximity of fetal capillaries to overlying trophoblast. In uncomplicated term pregnancies, ST differentiation leads to apoptosis, SFs contain multiple apoptotic nuclei and there is minimal loss of MPs. In PE, proliferation and loss increase and are associated with aponecrosis. Surface areas and thicknesses are maintained by the steady state. In IUGR and PE + IUGR, surface areas decline and overall thickness is preserved. However, in PE + IUGR, MP loss is greater than in PE or IUGR alone and rates of loss in IUGR are comparable to those in controls. See also Huppertz & Herrler (2005) and Goswami et al. (2006).

It is likely that steady states alter when total villous surfaces are reduced, as in IUGR and PE + IUGR (Fig. 1). For given rates of trophoblast recruitment and loss, and a given steady state between them, greater absolutes amount of shedding would be expected from larger surfaces and volumes. In PE, trophoblast shedding is enhanced and associated with aponecrosis, MPs and membrane-bound fragments (Huppertz & Kingdom, 2004; Goswami et al. 2006). Losses are matched by recruitment because trophoblast volumes, surfaces and numbers are maintained. However, in IUGR, there are reduced volumes and surfaces and shedding levels. If the arithmetic mean thickness of trophoblast increases (Mayhew et al. 2003a), this might be due to greater recruitment vs. loss or shedding of smaller fragments or recruitment of more or larger CT cells (Mayhew et al. 2007). Whatever the mechanism, a constant arithmetic mean thickness of the villous membrane overall (Mayhew et al. 2007) suggests greater peripheralization of capillaries within villi. In PE + IUGR, trophoblast and villous membrane thicknesses are preserved despite reduced villous surfaces.

Goswami et al. (2006) monitored MP concentrations in blood from maternal peripheral veins in age-matched controls and cases of PE and IUGR. Compared with controls, values in early-onset PE were almost 160% higher and those in all cases of PE (early- and late-onset) over 120% higher. However, values in IUGR did not alter significantly. Taking into account the reported changes in trophoblast surface areas, and other things being equal, we can deduce that IUGR produces MPs in amounts roughly commensurate with the loss in surface area (i.e. the rate of production per unit of surface is essentially ‘normal’). In contrast, the reduced villous surface in early-onset PE (equivalent to PE + IUGR) produces higher concentrations of MPs and so represents a disproportionately greater rate of loss than either late-onset PE or IUGR.

These findings support the notion of differences between PE and IUGR, re-emphasizing the need to monitor fetal growth in PE and to resolve pure PE from pure IUGR and from PE + IUGR. They also suggest differences between IUGR and PE + IUGR and these may be related to stimulated release of MPs associated with endoplasmic reticulum stress (Yung et al. 2008).

Placental diffusive conductance and the effects of urban air pollution

The mass-specific conductance response seen in normal human and murine pregnancies may not obtain in all pregnancies as shown by an experimental model of the effects of air pollution on murine placenta. Air pollutants may be man-made, biological, industrial or geological and the main pollutants affecting human health are carbon monoxide, nitrogen dioxide, sulphur dioxide, ozone, lead, hydrocarbons and particulate matter (PM).

The impact of air pollutants on human health and pregnancy outcomes is of growing concern and is associated with prematurity and IUGR (Bobak, 2000; Djemek et al. 2000; Rogers & Dunlop, 2000; Lee et al. 2003; Šrám et al. 2005; Wang & Pinkerton, 2007). Even moderate levels of air pollution can affect the reproductive health of mice (Mohallem et al. 2005; Lichtenfels et al. 2007) and, recently, we have shown that mice exposed to particulate air pollution not only produce smaller offspring but also show changes in placental functional morphology (Veras et al. 2008).

Two groups of second-generation Balb C mice were mated inside paired exposure chambers (see Veras et al. 2008) which produced differences in levels of particulate matter by filtering air sampled close to a busy traffic intersection. In one chamber, the air was not filtered (group NF) and in its partner chamber, it was filtered (group F). Animals were kept at ambient conditions of temperature, humidity and air pressure and were exposed to the same levels of gaseous pollution. In group F chambers, serial filters removed larger particles so that only particulate matter smaller than 2.5 µm in diameter (PM_2.5) remained. After exposure, all pregnant females were killed on E18 and fetuses and placentas were removed and weighed.

Placentas were immersion-fixed in buffered 4% formalin and sampled using a multistage SUR design (Mayhew, 2006a, 2008; Veras et al. 2008). Each organ was cut into slices orthogonal to the chorionic plate to generate sets of vertical sections for estimating exchange surface areas and tissue volumes. One set of slices was embedded in paraffin wax (to estimate volumes of different zones) and the other was embedded in glycolmethacrylate resin (to analyse volumes and surfaces within the labyrinth). Other quantities (vessel calibres, total oxygen diffusive conductance and mass-specific conductance of the intervascular barrier) were derived from the primary measures. Here, statistical comparisons between F and NF groups are drawn using the Mann–Whitney U-test.

Findings are summarized in Tables 4 and 5. In the NF group, fetal weight declined by about 18% and total litter weight by about 31%. There were no significant differences in total placental volume or in the volumes of the decidua basalis, chorionic plate, junctional zone or labyrinth (Table 4). However, within the labyrinth, maternal blood volume was about 34% smaller. Apparent differences in the volumes of fetal capillaries and trophoblast were not significant. In group F, the exchange surface areas of the maternal blood spaces and fetal capillaries were 19 cm² and 17 cm², respectively (Table 5). Although the roughly 52% increase in capillary surface area was significant, there was no change in maternal surface area. Differences in vascular volumes and surfaces were accompanied by changes in the apparent mean diameter of maternal blood spaces (from 33 µm to 23 µm) but not in mean diameters of fetal capillaries (Table 5).

Table 4.

Litter sizes, fetal weights and placental volumes in filtered and non-filtered groups of mice. Values are group means (CVs expressed as % of means). Data based on Veras et al. (2008)

Variable	Filtered (n = 6)	Non-filtered (n = 6)	P value
Fetal weight, g	0.846 (17%)	0.693 (16%)	< 0.05
Litter weight, g	6.64 (27%)	4.61 (37%)	< 0.05
Placenta volume, mm3	95.4 (14%)	97.1 (19%)	NS
Decidua volume, mm3	16.6 (55%)	21.0 (62%)	NS
Chorionic volume, mm3	9.3 (26%)	11.5 (34%)	NS
Junctional zone volume, mm3	20.1 (13%)	20.5 (23%)	NS
Labyrinth volume, mm3	49.4 (12%)	44.1 (23%)	NS
Trophoblast volume, mm3	25.4 (18%)	24.0 (19%)	NS
Maternal blood volume, mm3	15.7 (12%)	10.3 (43%)	< 0.05
Fetal capillary volume, mm3	8.3 (27%)	9.6 (33%)	NS

Table 5.

Exchange surface areas (cm²), vessel calibres (equivalent diameter, µm), barrier thicknesses (µm) and total and mass-specific oxygen diffusive conductances (cm³ min⁻¹ kPa⁻¹ and cm³ min⁻¹ kPa⁻¹ g⁻¹) of placentas in filtered and non-filtered groups. Values are group means (CVs expressed as % of means). Data based on Veras et al. (2008)

Variable	Filtered (n = 6)	Non-filtered (n = 6)	P value
Maternal blood surface	19.0 (13%)	17.8 (17%)	NS
Fetal capillary surface	17.1 (8%)	25.9 (21%)	< 0.01
Maternal space diameter	33 (13%)	23 (29%)	< 0.01
Fetal capillary diameter	19 (25%)	15 (32%)	NS
Diffusive conductance	0.0022 (11%)	0.0037 (40%)	< 0.05
Mass-specific conductance	0.0027 (27%)	0.0054 (41%)	< 0.05

In group F, the total oxygen diffusive conductance of the intervascular barrier amounted to 0.0022 cm³ min⁻¹ kPa⁻¹and this value increased significantly in the NF group. Moreover, mass-specific conductances more than doubled in the NF group (0.0054 vs. 0.0027 cm³ min⁻¹ kPa⁻¹ g⁻¹).

These estimates are comparable with those obtained by others (Coan et al. 2004; Rutland et al. 2005, 2007). Agreement is best for compartment volumes and vascular surfaces and any discrepancies are probably attributable to strain differences. However, larger discrepancies exist for values of arithmetic mean thickness, vessel calibres and diffusive conductances and it is likely that these are influenced additionally by differences in methods of estimation.

The mix of changes seen after exposure to non-filtered air is surprising in that some seem to be adaptive and others deleterious. Thus, the greater surface area of fetal capillaries, total diffusive conductance and mass-specific conductance of the intervascular barrier may be interpreted as adaptations aimed at maintaining or expanding oxygen and nutrient delivery to the fetus. The fact that fetal weight declines despite these adaptations implies that other factors exert more influential effects, notably on maternal blood space volumes and calibres. Changes at this site suggest compromised delivery of maternal blood to the placenta and an increase in resistance to its flow. Indeed, PM_2.5 has been shown to produce significant vasoconstriction in small arteries of the hearts and lungs of rats (Rivero et al. 2005). Amongst other factors which might influence birth outcomes are systemic alterations in haematocrit, erythrocyte deformability, blood viscosity and coagulability (Baskurt et al. 1990; Peters et al. 1997; Pekkanen et al. 2000; Rivero et al. 2005). Increases in such factors, exacerbating the effects of decreases in vessel calibres, could have marked effects on maternal blood rheology.

Quantitative immunoelectron microscopy and molecular localization

In immunoelectron microscopy, gold particles are used to label defined antigens in different intracellular compartments (organelles or membranes) and labelling patterns are compared by counting particles (Griffiths, 1993). The technique has been applied to localize interesting molecules within placentas but, surprisingly, there have been few attempts to quantify immunolocalization (see Mayhew & Desoye, 2004).

When quantifying the labelling intensity of compartments, LD is usually calculated and relates gold particle counts to the profile areas (golds µm⁻²) or trace lengths (golds µm⁻¹) of the sectional images of compartments. Recently, we developed simpler alternatives for counting immunogolds and statistically evaluating and comparing gold-labelling patterns. The methods address two basic questions: are some compartments preferentially labelled? (method 1), and do labelling patterns vary in different groups of cells? (method 2).

With method 1, numbers of gold particles lying on defined organelles or membranes are used to generate an observed frequency distribution. By randomly superimposing a lattice of test points on the same cell profiles, the numerical frequencies with which points overlie different organelles are determined and provided the expected distribution because random points hit compartments on sections with probabilities determined by profile areas. By replacing points with test lines, and counting sites of intersection with membrane traces, analogous procedures provide labelling distributions for different categories of membrane (Mayhew et al. 2002, 2003c). By treating membranes (surface-occupying compartments) as organelles (volume-occupying compartments), it is possible to deal with target proteins that translocate between membranes and organelles (Mayhew & Lucocq, 2008a,b).

Observed and expected distributions provide relative labelling indices (RLI = observed golds/expected golds) for each identified compartment and, for random labelling, the predicted value is RLI = 1. Observed and expected distributions are compared using Chi-squared analysis. If the observed distribution of gold particles proves to be non-random, RLI values and partial Chi-squared values identify compartments which are preferentially labelled. In fact, two criteria must be satisfied: first, RLI must be > 1 and, secondly, the partial Chi-squared must make a substantial contribution (at least 10%) to total Chi-squared. This approach may be used in tandem with LD values and has been applied to study gold particle distributions in various biological contexts (see Mayhew & Lucocq, 2008b).

With method 2, a simpler protocol is followed and raw gold counts in different groups of cells are compared directly by contingency table analysis (Mayhew & Desoye, 2004; Mayhew et al. 2004b). This method has been used to compare localization patterns of the glucose transporter, GLUT1, in cultured human trophoblast cells grown in euglycaemic (medium containing 5.5 mmol L⁻¹ glucose), hyperglycaemic (25 mmol L⁻¹ glucose) and osmotic control (19.5 mmol L⁻¹d-mannitol + 5.5 mmol L⁻¹ glucose) groups. Results (summarized in Table 6) showed a shift in labelling away from the plasma membrane and into the cell interior in cells exposed to hyperglycaemia and this shift seemed to be specific to glucose and not a response to osmotic stress (Mayhew & Desoye, 2004).

Table 6.

Immunolocalization of gold-labelled GLUT1 in three groups of human trophoblast cells. Values are observed (expected) numbers of gold particles in each compartment. For details, see Mayhew & Desoye (2004)

Compartments	Euglycaemia	Hyperglycaemia	Osmotic-control	Chi-squared values
Apical membranes	112 (101.3)	77 (101.3)	115 (101.3)	1.12, 5.84, 1.84
Basal membranes	34 (34.0)	33 (34.0)	35 (34.0)	0.00, 0.03, 0.03
Cell interior	54 (64.7)	90 (64.7)	50 (64.7)	1.76, 9.92, 3.33

The distributions are significantly different: total χ² = 23.9, df = 4, P < 0.001. The main sources of difference are a shift towards fewer gold particles on apical membranes and more at the cell interior (hyperglycaemia group) and fewer golds than expected at the cell interior (osmotic control group).

Concluding remarks

Design-based stereology has been used to interpret the morphology of human and murine placentas by providing precise and minimally biased estimates of functionally relevant structural quantities. Its sampling and estimation tools have been used to describe the processes of villous growth, trophoblast differentiation, vascular morphogenesis and diffusive transport in normal and compromised pregnancies. By way of illustration, changes in these processes during normal placental development have been summarized and it has been shown how fetoplacental vascular growth is compromised in cases of maternal asthma, that trophoblast turnover is affected differently in pre-eclampsia and intrauterine growth restriction, and that particulate urban air pollution affects maternal vasculature and diffusive transport in the murine placenta. Finally, new and efficient methods for quantitative immunoelectron microscopy have been emphasized because they now permit more rigorous analysis of the subcellular distributions of interesting molecules between compartments or in different experimental groups of cells or organs.