OUP user menu

The regulation of trophoblast migration across endothelial cells by low shear stress: consequences for vascular remodelling in pregnancy

Joanna L. James, Judith E. Cartwright, Guy S. Whitley, Darrel R. Greenhill, Andreas Hoppe
DOI: http://dx.doi.org/10.1093/cvr/cvr276 152-161 First published online: 19 October 2011


Aims In early human pregnancy placental trophoblasts migrate along uterine spiral arteries (SAs) and remodel these vessels into wide-bore conduits in a process essential for successful pregnancy. Until 10–12 weeks gestation trophoblasts plug spiral arteries, resulting in slow, high-resistance blood flow. This work examined the consequences of these low shear stress conditions on trophoblast migration, adhesion molecule expression, and attraction to chemotactic factors.

Methods and results Trophoblasts were cultured on fibronectin or human endothelial cells for 6–12 h under 0.5–6 dyne/cm2 shear stress using the BioFlux200 system, and imaged by time-lapse microscopy. Computer-based imaging algorithms were developed to automatically quantify migration. Chemotaxis assays were run using parallel flow. Trophoblasts cultured on fibronectin or endothelial cells did not undergo directional migration in 0.5 and 2 dyne/cm2 cultures; however, in 4 and 6 dyne/cm2 trophoblasts migrated with the direction of flow (n= 4, P< 0.001). Shear stresses did not affect the speed of trophoblast migration, or adhesion molecule expression (E-selectin, α4, β1, and αvβ3 integrin). Trophoblasts cultured on endothelial cells migrated into media containing interleukin-8, macrophage chemoattractant protein-1, or Regulated-upon-Activation-Normal-T-cell-Expressed-and-Secreted (RANTES) (n= 5, P< 0.05).

Conclusions Shear stress increases trophoblast migration in the direction of flow, challenging the idea that trophoblasts migrate down spiral arteries retrograde to flow. This suggests that low shear stresses generated by trophoblast plugging of spiral arteries in the first trimester may favour arterial remodelling by preventing the migration with flow seen at higher shear stresses, allowing trophoblasts to migrate down the arteries in response to alternate stimuli such as uterine or endothelial cell-derived chemotactic factors.

  • Trophoblast
  • Endothelial Cell
  • Migration
  • Adhesion molecules
  • Shear stress

1. Introduction

The placenta is a foetal organ that forms a crucial link between the mother and foetus throughout pregnancy, forming both a physical connection to the mother's uterus, and providing the site of all nutrient and gas exchange. In the first trimester (12 weeks) of pregnancy specialized placental cells termed trophoblasts grow out from the placental chorionic villi and invade into the decidua and its blood vessels [the spiral arteries (SA)] as far as the myometrial segments. These trophoblast migrate along the luminal surfaces of the SA and remodel these vessels by interdigitating between the endothelial cells (ECs) and replacing the endothelial lining and most of the musculoelastic tissue in the vessel walls such that a high-flow, low-resistance circulation is established to allow a constant and maximal flow of blood to the placenta and foetus as pregnancy progresses (Figure 1).1,2 These events are essential for pregnancy, and their importance is exemplified when they are insufficient, resulting in complications such as the hypertensive disorder pre-eclampsia and intrauterine growth restriction (IUGR).3

Figure 1

Schematic diagram demonstrating the process of spiral artery remodelling. During pregnancy the SA are invaded by placental trophoblasts and remodelled from their tightly coiled structure (left-hand side) to wide bore conduits where trophoblasts have completely replaced the endothelial layer and removed the musculoelastic tissue such that the artery is independent of maternal vasoconstriction (right-hand side). For the first 10–12 weeks of pregnancy trophoblasts also plug these SA (demonstrated in the central artery) preventing the flow of maternal blood to the placental surface.

For approximately the first 10–12 weeks of gestation trophoblasts form plugs that occlude the SA, allowing only a maternal plasma filtrate to pass through, resulting in a high resistive index and decreased flow rate before 10–12 weeks of gestation.47 Furthermore, as the process of SA remodelling during this time involves the dedifferentiation or removal of the smooth muscle layer surrounding the SA,8 these arteries also lose the ability to dilate or constrict in order to regulate their shear stress. The trophoblast plugs dissipate between 10 and 12 weeks of gestation, allowing maternal blood to flow freely to the placenta, lowering the resistance and increasing the flow rate in the unplugged arteries.6,7,9

The extent of trophoblast invasion and SA remodelling and the complete occlusion of the SA by trophoblast plugs are unique to human pregnancy,10,11 Previous data have shown that flow-induced shear stress promoted the migration of macaque trophoblasts retrograde to the direction of flow, but that this migration was only affected by the flow rate in isolated cultures, and not in trophoblast–EC co-cultures.12,13 However, the effect of the changes in flow rate on the crucial process of SA remodelling has not been examined in humans.

Trophoblast migration down the SA is dependent on trophoblast–EC interactions. Both trophoblast migration and the ability of trophoblast to interact with and replace SA ECs involve adhesion molecules expressed by both cell types.1 However, the mechanisms by which trophoblast migration is regulated remain unclear. As trophoblasts differentiate and migrate away from the villous they undergo specific changes in adhesion molecule expression, down-regulating α6β4 integrin and E-cadherin, and up-regulating the expression of α5β1 and α1β1 integrin which promote trophoblast migration.14 In endovascular trophoblasts, further changes in adhesion molecule expression occur to facilitate adhesion to vascular ECs and interdigitation into the EC layer, including the up-regulation of α4β1 integrin [which binds to the vascular cell adhesion molecule (VCAM-1)], αvβ3 integrin, VE-cadherin, and E-selectin.1 Shear stress has been shown to regulate EC adhesion molecule expression, and some adhesion molecules such as integrins are capable of transducing haemodynamic forces into biochemical signals.15,16 Low shear stress and flow reversal promote leucocyte binding and transmigration by increasing expression of VCAM-1 and promoting expression of leucocyte chemoattractants such as MCP-1 and IL-8.17 The interdigitation of trophoblast into the EC layer shares some similarities with leucocyte transmigration, and thus it is possible that these factors also play a role in regulating trophoblast migration and interdigitation.18

It is often assumed that trophoblasts migrate down the SA retrograde to flow,19,20 but aside from the presence of trophoblasts in the arteries, little direct evidence exists to support this claim. The occlusion of the SA by trophoblast plugs in the first trimester means that blood flow through these vessels is likely to be very slow during this time, and our previous work has shown that this may create an environment favourable for the remodelling of the SA.21 Therefore, in this study we aimed to investigate the effect of low shear stress on trophoblast migration over EC monolayers, and determine the role of adhesion molecules and chemotactic factors in this process.

2. Methods

This investigation conformed with the principles outlined in the Declaration of Helsinki, and was approved by the Wandsworth Local Research Ethical Committee.

2.1 Isolation of human umbilical vein ECs

Umbilical cords were collected with informed consent from elective caesarian deliveries. The umbilical vein was cannulated and 15 mL of 3 mg/mL Dispase (Invitrogen, Paisley, UK) in PBS was used to isolate human umbilical vein endothelial cells (HUVECs) as previously described.21

2.2 Cell lines

The trophoblast cell line SGHPL-4, which was derived from first trimester extravillous trophoblasts,22 was used up to passage 25, and was cultured in Hams F10 (Sigma, Poole, UK) containing 10% FBS, 50 U/mL penicillin, 50 µg/mL streptomycin, 2 mM l-glutamine until use in shear stress cultures.

2.3 Shear stress cultures

Cells were cultured under shear stress using the BioFlux 200 flow system (Fluxion, San Francisco, USA). Channels of BioFlux plates (24 wells, 8 flow channels, or 48 wells, 24 flow channels) were coated with 20 µg/mL fibronectin (Sigma, Poole, UK), and incubated at 37°C for 45 min. Channels were flushed with HUVEC media for 5 min to remove excess fibronectin. HUVECs were labelled with 5 µM of MitoTracker Green (Molecular Probes, Eugene, USA) in a serum-free mix of RPMI:M199 media for 40 min, then washed in media twice for 10 min and trypsinized from the culture flask. HUVECs were seeded into flow channels in a concentrated droplet as previously described.21 HUVECs were incubated at 37°C for 4 h to form a monolayer, then SGHPL-4s were labelled with 5 µM of CellTracker Orange (Molecular Probes, Eugene, USA) and seeded onto HUVEC in the same manner as above. SGHPL-4s were allowed to adhere to the HUVEC layer for 3 h. Inlet wells were filled with pre-gassed HUVEC media, and flow was induced by tightly regulated air (containing 5% CO2) pressure in the inlet well.

2.4 Determination of cell migration by timelapse microscopy

In order to determine the direction and speed of migration of SGHPL-4s across HUVEC monolayers, channels of 24-well Bioflux plates containing either co-cultures of HUVECs and SGHPL-4s, HUVECs only or SGHPL-4s only were run under shear stresses of 0.5, 2, 4, or 6 dyne/cm2 for 7 h. Cells were visualized using an Olympus IX71 inverted phase contrast microscope with a motorized stage and cooled CCD camera (Hamamatsu model C4742-95) enclosed in a humidified chamber at 37°C. Phase images were captured at 10 min intervals, with red and green fluorescent images captured every 30 min using Volocity software (PerkinElmer, Cambridge, UK). A minimum of four sequences were captured from each flow channel, and cultures were repeated at least three times at each flow rate. In total, 132 sequences were analysed, equating to ∼2640 individual cells (based on an estimated average of 20 cells/sequence).

2.5 Algorithm development and image analysis

An automated mechanism for quantifying SGHPL-4 motility was developed to analyse the large number of sequences in this study (132 sequences in total with ∼85 frames each). Our new imaging approach was informed by the work of Siegert et. al.,23 who developed a gradient method to quantify cell and tissue flow on a per pixel level. The underlying principle is based on the calculation of a velocity vector field from the spatio-temporal distribution of brightness patterns which is commonly referred to as ‘optical flow’.24 Since we are interested in the movement of all cells as a population, it is no longer necessary to distinguish between individual cells but rather treat cells, either single or clustered, as brightness patterns that move through the image (Figure 2). As the degree of CellTrackerOrange labelling of SGHPL-4s was very variable, rather than obtaining the motion vectors from the fluorescent signal itself, motion vectors were calculated from the movement of cellular regions (Figure 2C) using the following steps: First, images were pre-processed with a 5 × 5 mean filter to reduce noise. Second, binary images of cellular regions were obtained by calculating a suitable threshold automatically from multi-scale histogram analysis. Normalized histograms over decreasing scales were obtained as previously described25 and the mean intensity of the histograms over all scales was determined as a suitable threshold value. Regions with intensities greater than the calculated threshold were classified as cellular regions (white) (Figure 2C), and areas of small isolated debris were removed. The velocity Vn at each pixel position classified as belonging to a cellular region is calculated over the whole image sequence using the following approach. First, the binary images were smoothed with a large Gaussian filter (sigma = 15) in order to provide a smooth surface. Then, motion vectors (shown overlaid in Figure 2D) were calculated for each pixel position within cellular regions according to the Horn and Schunck method.26 A new parameter termed directional bias (defined as the normalized distribution of movement directions, weighted by their speeds, from cellular regions in the whole image sequence) was used to determine whether cells were moving retrograde to the flow direction. The directional bias is calculated for each angle α as follows.

Figure 2

Image processing stages involved in migration and speed determination. (A) Red fluorescent channel of SGHPL-4s in co-culture in the channel of a 24-well Bioflux plate. (B) Segmented cellular regions obtained by multi-scale histogram analysis were subsequently convolved with a large Gaussian kernel to provide a smooth motion surface. (C) Images were processed by optical flow analysis to calculate the motion vector field (black arrows), here overlaid with the original fluorescent SGHPL-4 image at selected region of interest (white square).

Embedded Image (1)

where N denotes the total number of pixels belonging to all cellular regions in the whole image sequence, |Vn| denotes the speed and ∠Vn denotes the movement direction at each pixel location in the cellular regions.

The directional bias is a numeric value with the range (0–1) whereby values <0.5 indicate cell movements retrograde to the flow direction and values >0.5 indicate cell movements in the flow direction. To indicate whether the movement of SGHPL-4 cells is affected by the shear flow, we calculated separately an average of the directional bias values for movements to the right (90°–270°) (retrograde to flow, R) and to the left (0°–90° and 270°–360°) (with flow, W). A significant difference between R and W values would indicate a bias induced by shear stress.

The accuracy of the optical vector field-based approach was assessed by creating a synthetic image of a moving test object of a similar size to a SGHPL-4 cell (Figure 3A and B). For a movement in the direction of 225° of the object in Figure 3A, the mean direction from the motion vector field was computed as 224.1° with a standard deviation of 5.2°. For further validation, we compared the directional bias generated from tracking a total of 143 cells in 10 image sequences using our previously published semi-automatic tracking technique27 (Wsemi-auto) against our new automated motion vector field method (Wauto) (Figure 3C and D) and there was no significant difference (P > 0.05) between the two methods at lower (0.5–2 dynes/cm2) and higher flow rates (4–6 dynes/cm2).

Figure 3

Validation of the automated optical flow technique used to track cell migration. (A) A single object moving in the direction of 225° with superimposed motion field vectors (not to scale). The four values on the border denote the angular directions of the co-ordinate system. (B) Distribution of angular values (directions) for a movement in the direction of 225° of the object in (A) as determined by the optical flow technique. The mean direction is computed as 224.1° with a standard deviation of 5.2°. (C) Example trajectories of 10 semi-automatically tracked SGHPL-4 cells over a 48-h time period plotted from the same origin. This sequence expresses a clear bias for cells moving to the left with the direction of flow (resulting in a directional bias >0.5). (D) A comparison of the directional bias calculations of five different sequences of SGHPL-4s with flow rates ≥4 dyne/cm2 demonstrates no significant difference in directional bias [With (W) and Retrograde (R)] calculated from semi-automatically tracked cells or calculated by the new automatic optical flow approach (n= 5 sequences with at least 10 cells each, P> 0.05).

2.6 Determination of adhesion molecule expression in shear stress

Co-cultures of HUVECs and CellTracker Orange-labelled SGHPL-4s, HUVECs only and CellTracker Orange-labelled SGHPL-4s only were run in a 48-well BioFlux plate (Fluxion, San Francisco, USA) for 12 h at 2 or 6 dyne/cm2. Owing to limitations in the volume in the inlet and outlet wells, every 1½ h HUVEC media was transferred from the outlet to inlet well, and every 4½ h the inlet well was filled with fresh media. At the end of the 12-h period, channels were fixed by flushing with 4% paraformaldehyde in PBS at 2 dyne/cm2 for 15 min. By this time many of the SGHPL-4s had interdigitated into the HUVEC monolayer, in a manner similar to that which occurs in vivo. Channels were rinsed by flushing with PBS at 2 dyne/cm2 for 10 min, and stored at 4°C.

2.7 Immunocytochemistry of shear stress cultures

The expression of adhesion molecules on HUVECs and SGHPL-4s was determined by fluorescent immunocytochemistry. Non-specific binding was blocked with 10% normal goat serum in PBS, pH 7.4, containing 0.05% tween 20 (PBS-tween) which was introduced for 1 min at 3 dyne/cm to ensure fluid in the channel was completely replaced, and then run for 1 h at 1 dyne/cm2. Channels were washed with PBS-tween for 15 min at 2 dyne/cm2, then 150 µL of primary antibody [25 µg/mL E-selectin (R&D, Oxford, UK), 1 µg/mL αvβ3 integrin (R&D, Oxford, UK), 2 µg/mL α4 integrin (R&D, Oxford, UK), or 2 µg/mL β1 integrin (Santa Cruz Bioscience, USA)] was added to the inlet well and run through the channel at 3 dyne/cm2 for 1 min, then 1 dyne/cm2 for 1½ h. Channels were washed with PBS-tween as previously, and 150 µL of 1:500 anti-mouse Alexafluor-488 tagged secondary antibody (Invitrogen, Paisley, UK) was added in the same manner as the primary antibodies. Channels were then washed with PBS-tween for 15 min and a 1:250 dilution of Hoescht (Immunochemistry Technologies, Minneapolis, USA) in PBS was added at 3 dyne/cm2 for 1 min, then 1 dyne/cm2 for 15 min. At the time of staining, HUVECs and SGHPL-4s grown on fibronectin-coated chamber slides were stained with an irrelevant IgG1 antibody (2 µg/mL, Sigma, Poole, UK) as a negative control. Cells were observed on a Zeiss 510 meta confocal microscope running Confocal Microscopy Software version 3.2a (Zeiss, Hertfordshire, UK). Pinhole, laser power and gain settings were held constant for each primary antibody. Images captured were analysed blind and assessed semi-quantitatively by rating the adhesion molecule staining intensity in SGHPL-4s in each image on a scale of 0–3, where 0 = negative, 1 = weak, 2 = moderate, and 3 = strong. All cells and images were analysed by the same researcher on the same day in order to ensure consistency. All images from a minimum of three experiments were analysed, and an average score determined for each experiment.

2.8 Chemotaxis assays

The migration of SGHPL-4s cultured on HUVECs towards IL-8 (10, 50, or 100 ng/mL, Peprotech, London, UK), MCP-1 (10, 50, or 100 ng/mL, Peprotech, London, UK), or RANTES (1, 5 or 10 ng/mL, Peprotech, London, UK) was determined using a Bioflux 24-well plate in dual flow mode. Well A was filled with 2 mL of cytokine-containing HUVEC media while Well B was filled with 2 mL of the standard HUVEC media, then 1 dyne/cm2 of shear stress was applied to each well simultaneously, generating 2 dyne/cm2 of shear stress in the channel being imaged. As the Bioflux system generates laminar flow in the channel, the two media feeding from A and B travel in parallel, although at lower shear stresses limited mixing occurs over the centerline generating a chemotactic gradient. Plates were run for 12 h using the timelapse microscopy equipment previously described, with phase contrast images captured every 15 min and red fluorescent images captured every 30 min. The attraction of SGHPL-4s towards or away from the cytokine containing media was quantified by dividing the channel horizontally into two, tracking the migration of SGHPL-4s manually using Volocity software, and recording which half of the channel they ended the experiment in relative to their origin. For each channel these data are expressed as the net migration of SGHPL-4s into the cytokine containing media, calculated by the formula: (Percentage of SGHPL-4 that migrated into the cytokine containing media) − (Percentage of SGHPL-4 that migrated out of the cytokine containing media). Experiments were repeated five times. At least three sequences were analysed in each experiment so that a total of 40 cells were analysed from each half of the channel per experiment (80 cells in total).

2.9 Statistical analysis

Data was analysed statistically using GraphPad Prism v5.0 (GraphPad, La Jolla, CA, USA). An ANOVA followed by a Bonferroni post-test was used to determine statistical differences in directional bias and speed of migration. Semi-quantitative scoring of adhesion molecule immunohistochemistry images were analysed using an ANOVA followed by a Kruskal–Wallis and Dunn's multiple comparison test. A repeated measures ANOVA followed by a Dunnet's post-test was used to determine statistical differences between individual sets of chemotaxis data. Statistical significance was assumed at P< 0.05.

3. Results

3.1 Shear stress affects the direction, but not speed, of trophoblast migration across EC monolayers

In order to determine how low levels of shear stress affected the direction of trophoblast migration across EC layers, SGHPL-4s were either cultured alone or on confluent layers of HUVECs and then run under shear stresses from 0.5 to 6 dyne/cm2 for 12 h, imaged by time-lapse microscopy, and the directional migration of SGHPL-4s was measured by computer algorithm. In 0.5 or 2 dyne/cm2 cultures, SGHPL-4s cultured alone or on HUVEC monolayers did not demonstrate any significant directional bias of migration (n= 3, P > 0.05) (Figure 4). However, under 4 and 6 dyne/cm2 of shear stress a significant difference in the directional bias of cell migration in the direction of flow was observed both in SGHPL-4s cultured alone (n= 4, P< 0.001) and SGHPL-4s cultured on HUVECs (n= 4, P< 0.001) (Figure 4). No significant differences in the speed of SGHPL-4 migration were observed between any of the shear stresses examined when SGHPL-4s were cultured on HUVEC monolayers (Figure 5). When SGHPL-4s were cultured on fibronectin, no significant difference was observed between any of the shear stresses employed with the exception of the comparison between 2 and 6 dyne/cm2. Here, a statistically significant increase in speed in 6 dyne/cm2 cultures in comparison with 2 dyne/cm2 cultures was observed (P< 0.05, Figure 5), but as there was no significant difference in speed between 0.5 or 4 dyne/cm2 cultures and 6 dyne/cm2 cultures it is unclear how much biological significance this finding holds.

Figure 4

Directional migration of SGHPL-4s in low shear stress. Normalized distribution of the direction of cell migration weighted by speed (in degrees) as determined by optical flow vector analysis of all cellular regions in cultures of SGHPL-4s cultured on either confluent HUVEC monolayers [(A), n= 75 sequences, ∼1500 cells] or fibronectin [(B), n= 57 sequences, ∼1140 cells] under shear stresses from 0.5 to 6 dyne/cm2. Directions between 90 and 270° are retrograde to the flow direction [retrograde to flow (R), as indicated by two dashed lines on the graph], whereas directions between 270–360 and 0–90° were in the direction of flow [with flow (W)]. Differences in directional bias in SGHPL-4 cultured on HUVEC layers (C) or fibronectin (D) were determined by comparing the average directional bias of cells migrating with flow (W) to those migrating retrograde to flow (R). No significant directional bias was observed in cultures in 0.5 or 2 dyne/cm2 of shear stress. In 4 and 6 dyne/cm2 of shear stress both SGHPL-4s cultured on HUVEC layers (C) and fibronectin (D) significantly migrated in the direction of flow (n= 4, P < 0.001, ***). Experiments were repeated three times at 0.5 and 2 dyne/cm2 and four times at 4 and 6 dyne/cm2 at each shear stress, with a minimum of 60 cells quantified in each experiment.

Figure 5

Speed of SGHPL-4 migration in low shear stress. Histogram demonstrating the speed of SGHPL-4 migration alone or in co-culture with HUVECs. No significant differences in speed were observed when SGHPL-4s were cultured on HUVECs. When SGHPL-4s were cultured on fibronectin, a significant difference in speed was observed between 2 and 6 dyne/cm2 cultures only (P < 0.05). Experiments were repeated three times at 0.5 and 2 dyne/cm2 and four times at 4 and 6 dyne/cm2 at each shear stress, with a minimum of 60 cells quantified in each experiment.

3.2 Low shear stress does not affect the expression of key adhesion molecules involved in trophoblast–EC interactions

In order to investigate whether the differences in SGHPL-4 migration across HUVECs with increased shear stress were a result of alterations in adhesion molecule expression or distribution, SGHPL-4s were cultured alone or on confluent HUVEC layers under 2 or 6 dyne/cm2 of shear stress, and the expression of key adhesion molecules identified in trophoblast–EC interactions (E-selectin, α4 integrin, β1 integrin, and αvβ3 integrin) were determined by immunohistochemistry. No discernable difference in the cellular localization of any of the adhesion molecules was observed. Semi-quantitative analysis of adhesion molecule expression did not reveal any difference in the level of expression between cultures under either 2 or 6 dyn/cm2 of shear stress or when SGHPL-4s were cultured separately or together in HUVEC co-cultures (Figure 6).

Figure 6

Adhesion molecule expression under low shear stress. Histograms demonstrating the results of semi-quantitative analysis of SGHPL-4s in co-culture with HUVECs, and SGHPL-4s alone following immunohistochemistry with antibodies reactive with α4 integrin (A), αvβ3 integrin (B), β1 integrin (C), and E-selectin (D). No significant differences in SGHPL-4 staining intensity was evident between 2 dyne/cm2 and 6 dyne/cm2 cultures, regardless of whether SGHPL-4s were cultured alone or co-cultured with HUVECs. Experiments were repeated three times.

3.3 Trophoblasts migrate towards EC chemoattractants

In order to investigate alternative stimuli that may drive the migration of trophoblasts down the SA during pregnancy, we examined the effect of cytokines known to be derived from ECs. SGHPL-4s were cultured on confluent HUVEC layers under 2 dyne/cm2 of shear stress with cytokine-containing media flowing along half of the channel, and control HUVEC media flowing along the other half, and the net migration of cells into the cytokine containing region of the channel was calculated. SGHPL-4 demonstrated significant net migration into media containing IL-8 (50 and 10 ng/mL, n= 6, P< 0.05), MCP-1 (50 ng/mL only, n= 5, P< 0.05), and RANTES (10 ng/mL only, n= 5, P< 0.05) (Figure 7).

Figure 7

SGHPL-4 migration into cytokine containing media. Histograms showing net migration of SGHPL-4s into media containing IL-8 (A), MCP-1 (B), or RANTES (C) in 2 dyne/cm2 shear stress cultures using a Bioflux200 chemotaxis assay. A net migration of 0 indicates that cells did not exhibit a bias towards either half (cytokine containing or control) of the flow channel, whereas net migration towards the cytokine containing half of the channel is represented by a positive integer. Migration that was significantly different to control levels is indicated by * (P < 0.05). Experiments were repeated five times.

4. Discussion

Despite the dogma that trophoblasts are stimulated to migrate retrograde to flow down the SA, little direct evidence exists to support this claim. Furthermore, the occlusion of the SA by trophoblast plugs for the majority of the first trimester would create a low flow environment during the time of maximal invasion, bringing into question whether retrograde migration induced by flow would be required during this time. The findings in this study present evidence of a potential scenario for trophoblast migration after arterial plugging by demonstrating that under very low shear stress conditions trophoblast do not undergo directional migration, and as shear stress increases the number of trophoblasts are stimulated to migrate in the direction of flow.

In this paper, we describe an automated approach to quantify cell migration by determining the motion of cellular regions on the basis of optical flow analysis. Previously used algorithms were found suitable for calculating motion vectors in more homogeneous objects.26 However, the novel automatic approach developed in this work calculates a motion vector field from segmented cellular regions by optical flow analysis, which overcomes problems of variability in cell labelling. We have shown through the validation that the optical flow-based technique produces a motion vector field with an accuracy of ±5.2° standard deviation. This has allowed us to track the migration of a large number of cells automatically over this study (132 sequences, containing ∼2640 individual cells).

The finding that trophoblasts cultured on HUVEC monolayers do not undergo directional migration under very low shear stresses (0.5 and 2 dyne/cm2) and migrate in the direction of flow when shear stress is increased (4 and 6 dyne/cm2) has several potential implications for SA remodelling. First, this suggests that increased shear stress would be detrimental to SA remodelling as it would stimulate migration of trophoblasts in the opposite direction to invasion, and potentially result in shallow invasion down the arteries. This supports our previous hypothesis that low shear stress in plugged first trimester SA favours SA remodelling.21 Second, it remains unclear why trophoblasts do not migrate further than the inner third of the myometrium, and it is interesting to speculate that if trophoblasts do not migrate retrograde to flow, the increased flow of blood through the uterine vasculature as pregnancy progresses may act to limit trophoblast invasion and/or facilitate the re-endothelialization of the uterine vessels towards the end of pregnancy. Finally, these results raise the question as to the relative contributions of interstitial and endovascular trophoblasts to SA colonization and remodelling.

Our findings contrast with previous observations that macaque trophoblasts migrated retrograde to flow at 15 dyne/cm2 only in isolated cultures, and not when trophoblasts were co-cultured with ECs.12,13 However, 15 dyne/cm2 of shear stress is much greater than would be expected in a plugged first trimester SA in the human, and in our study using much lower shear stresses, we observed no differences between the behaviour of SGHPL-4s cultured on HUVECs or on fibronectin. Furthermore, retrograde migration was not observed in any of the shear stresses employed. These results suggest that the migration of trophoblasts in the direction of flow is an intrinsic feature of these cells and not largely dependent on interactions between trophoblasts and ECs.

Adhesion molecules represent a key method of interaction between trophoblasts and ECs that play important roles in migration and cell signalling.1,14 Previous work using higher rates of shear stress (15 dyne/cm2) demonstrated the redistribution of the EC intercellular adhesion molecule-1 (ICAM-1) both in EC only cultures, and when co-cultured with trophoblasts.28 In contrast, in our work using much lower shear stresses, no redistribution was observed in four of the key adhesion molecules expressed on trophoblasts that have been identified to be important in facilitating trophoblast–EC interactions. This may be due to the lower shear stresses employed in this model, which were selected as an attempt to model levels of shear stress in plugged SA in the first trimester. Alternatively, it is possible that other adhesion molecules not examined such as VE-cadherin or protocadherin-12 may play a role in regulating trophoblast migration under low shear stress.1,29 Furthermore, EC adhesion molecule expression, activation and morphology is also known to be regulated by the frequency of pulsatile flow.30 No data have been published on the response of trophoblasts to pulsatile flow, but investigation of the effect of pulsatile flow on trophoblast migration or adhesion molecule expression in EC co-cultures would be of great interest to examine in the future. Finally, it is important to note that it was only possible to analyse this data semi-quantitatively by immunohistochemistry as it is not possible to remove only cells in the channels exposed to the correct level of shear stress from the system to quantify adhesion molecule expression by western blotting or PCR, thus it is possible that small differences in expression may exist that could have been detectable by these methodologies.

A lack of discernable differences in the level of adhesion molecule expression corresponds to the lack of difference in the speed of trophoblast migration that was observed between the vast majority of shear stresses examined, and to the observation that trophoblast migration does not appear to be a result of trophoblast–EC interactions.

If trophoblasts are not stimulated to migrate down the SA by blood flow, what alternate mechanisms may be responsible for arterial migration? We investigated the possibility that the absence of a directional influence on trophoblast migration may allow trophoblasts that invade into the arteries to be responsive to alternative chemotactic signals to stimulate their migration down the vessel lumen. In this work, we investigated three EC-derived cytokines (IL-8, MCP-1, and RANTES) that play roles in monocyte attraction and trafficking across the endothelium, a process that draws several parallels with trophoblast interdigitation into the arterial EC layer. In this study, we demonstrated that trophoblasts were attracted towards some concentrations of IL-8, MCP-1, and RANTES when cultured on HUVEC monolayers under low levels (2 dyne/cm2) of shear stress. RANTES has also previously been shown to increase the migration of macaque trophoblasts.31 Thus, these cytokines have the potential to attract trophoblasts down the SA and aid in remodelling of the uterine vasculature in the first trimester. While it would have been interesting to compare the results under both 2 and 6 dyne/cm2 conditions, this assay has been optimized for use as a chemotaxis assay at very low levels of shear stress, and as the amount of diffusion between the media in parallel flow is altered by the speed of the media flow it is not possible to compare levels of chemotaxis under different shear stresses without the experimental conditions also changing.

The production of chemoattractants by the endothelium of the SA may be increased by several factors in early pregnancy. First, HUVEC secretion of MCP-1 and IL-8 is increased under low shear stress conditions.32,33 Thus, the induction of these cytokines in ECs in low-flow plugged SA may play important roles in stimulating trophoblast migration down the SA, and facilitating their replacement of the arterial ECs. Second, the complex cytokine network in the first trimester decidua as a result of secretion by trophoblasts, uterine natural killer (uNK) cells and the endometrium itself will in turn affect the cytokines released by the arterial endothelium. TNFα (known to be released by uNK cells) increases EC IL-8 and MCP-1 expression,34 and incubation of uterine microvascular ECs with trophoblast conditioned media has also been reported to increase EC RANTES expression.28 Finally, it would also be of interest to examine how specific regulators of trophoblast migration known to be aberrant in pre-eclampsia such as Nodal or endocrine gland-derived vascular endothelial growth factor (EG-VEGF) may affect trophoblast migration under low shear stress.35,36

In conclusion, in this work we have demonstrated that trophoblasts are not stimulated to migrate retrograde to flow under low shear stress conditions, and that as shear stress increases the proportion of trophoblasts migrating retrograde to flow decreases. This has important consequences for the migration of trophoblasts down the uterine SA in early pregnancy, and the subsequent remodelling of these vessels, as it suggests that the low flow rates in SA plugged by trophoblasts may favour SA remodelling by allowing trophoblasts to migrate down the arteries in response to chemotactic factors produced by ECs lining the vessels.


This work was supported by the New Zealand Foundation of Science and Technology (SGRG0701). The BioFlux system employed in this research was purchased with the aid of grants from the Royal Society and St George's University of London.


We wish to thank the patients and staff of the fetal medicine unit St George's Hospital, London for the kind donation of primary tissue used in this work, in particular Professor Baskaran Thilaganathan and Dr Edward Dobransky. I also wish to thank Kinga Szewczyk of the St George's University of London Imaging Unit for her assistance with Confocal Microscopy.

Conflicts of interest: none declared.


View Abstract