Ana María Porras Hernándeza,
Maria Tenje‡
*a and
Maria Antfolk‡*bc
aDept. of Materials Science and Engineering, Science for Life Laboratory, Uppsala University, Uppsala, Sweden. E-mail: maria.tenje@angstrom.uu.se
bDept. of Biomedical Engineering, Lund University, Lund, Sweden. E-mail: maria.antfolk@bme.lth.se
cBiotech Research and Innovation Centre, University of Copenhagen, Copenhagen, Denmark
First published on 21st October 2022
Left-right asymmetry is a conserved property in nature and observed in the human body, a property known as cell chirality. Cell chirality is often studied using micropatterned in vitro models. However, micropattern geometry and size often varies across different studies, making it challenging to compare results. Here, we utilized micropatterned RGD-peptide lines on hyaluronic acid hydrogels to investigate the effect of the micropattern width on the exhibited cell chirality bias of brain microvascular endothelial cells. Overall, this cell type exhibited a negative chirality bias on micropatterned lines ranging from 10 μm to 400 μm in width, where the negative bias was most pronounced on the 100 μm wide lines. We also observed that this exhibited chirality bias varied across the line width. This work serves as a guide to determine optimal micropattern width for further investigations on cell chirality bias and its prominence in e.g., disease states or upon exposure to toxic substances.
Cell chirality has been observed in vivo, but only in invertebrates.5 However, in vitro both vertebrates and human cell chirality has been observed, demonstrating an inherent cell chirality under specific culture conditions.6–8 A pioneering study by Wan et al. determined that cell chirality is cell type specific, when they were the first to observe that different cell types or disease states express different left-right orientation biases.9 This study has since been extended and this chirality bias has now been reported on various cell types (Table 1).
Cell type | Bias | Species | Tissue | Phenotype | Cell origin | Pattern type | Pattern width (μm) | Ref |
---|---|---|---|---|---|---|---|---|
Human umbilical vein endothelial cells (HUVEC) | Negative | Human | Umbilical vein | Endothelial cells | Primary | Ring | 200, 150, 125 | 9, 13 and 23, respectively |
Human brain microvascular endothelial cells | Negative | Human | Brain capillary | Endothelial cells | Primary | Ring | 150 | 18 |
Human skeletal muscle cells (hSkMC) | Positive | Human | Skeletal muscle | Myoblast | Cell line | Ring | 200 | 9 |
C2C12 | Positive | Mouse | Skeletal muscle | Myoblast | Cell line | Ring | 200 | 9 |
C2C12 | Negative | Mouse | Skeletal muscle | Myoblast | Cell line | Line | 200 | 15 and 16 |
Rat cardiac fibroblast | Negative | Rat | Heart | Cardiac fibroblast | Primary | Ring | 200 | 9 |
Human adipose stem cells (hASC) | Negative | Human | Adipose | Stem cells | Primary | Ring | 200 | 9 |
Human mesenchymal cells (hMSC) | Negative | Human | Bone marrow | Stem cells | Primary | Ring | 200 | 9 |
Human primary skin fibroblast | Negative | Human | Skin | Fibroblast | Primary | Ring | 200 | 9 |
Human skin fibroblast line | Negative | Human | Skin | Fibroblast | Cell line | Ring | 200 | 9 |
Human skin cancer fibroblast line | Positive | Human | Skin | Fibroblast | Cell line | Ring | 200 | 9 |
NIH/3T3 | Negative | Mouse | Embryo | Fibroblast | Cell line | Ring | 200 | 9 |
MDCK | Positive | Dog | Kidney | Epithelial | Cell line | Ring | 235 | 19 |
MDCK | Positive | Dog | Kidney | Epithelial | Cell line | Line | 200 | 14 |
MC3T3-E1 | Negative | Mouse | Calvaria | Osteoblast | Cell line | Ring | 200 | 9 |
Micropatterned substrates have been used to study cell chirality in vitro, where cells have been cultured on well-defined boundaries in the micrometre scale to elucidate the mechanisms of cell chirality development.7–10 Micropatterned substrates are a suitable technology for such studies, as they provide a simple mean to achieve well-defined edges that are the required cue for the cells to display their chirality bias.9,11 Furthermore, the method can be used to produce a large number of microscale patterns to enable high-throughput studies with high statistical power.
Cell chirality offers an attractive comparative property of cells across cell studies due to its binary-switch nature.9,12 However, comparison between different studies is often complicated by the use of different micropattern geometries (e.g., lines or rings) and different pattern widths.9,13–18 Typically, pattern widths between 100 and 300 μm have previously been used. Previous studies have reported that equivalent chirality bias was observed on both geometries, however no quantitative data was provided. Likewise, it has also been observed that there is no cell chirality bias exhibited on circle or square patterned geometries.9
From an analytical point of view, the determination of cell chirality based on alignment bias is more accessible on micropatterned lines, compared to rings. The analysis of cell orientation on ring micropatterns requires custom-made image processing algorithms in MATLAB or Python,9,19 as the x-axis changes depending on the ring location. On the other hand, cell orientation on micropatterned lines is straightforward as the x-axis is constant, and it can be performed in readily available image analysis software, such as CellProfiler™.20
Endothelial cells readily orient on micropatterned surfaces as well as in vivo where this exhibit promotes an atheroprotective phenotype.21,22 These cells have also been seen to exhibit a chirality bias. Previous studies have reported that endothelial cells exhibit a negative chirality bias on ring micropatterns. This was observed in human brain microvascular endothelial cells and in human umbilical vein endothelial cells (HUVECs) on ring patterns 150 μm or 200 μm wide, respectively.9,18 However, it is still largely unclear under which boundary conditions cells exhibit their chirality bias and at which pattern width this feature is most pronounced. Especially, micropatterns of widths less than 100 μm have been completely overlooked, even though blood vessels range from a couple of micrometers to centimeters in diameter, making their circumference in the range of 10 μm at their smallest.
In this paper, we used mouse brain microvascular endothelial cells to understand if these cells exhibit a cell chirality bias on micropatterned lines and investigate the effect of the different line widths on the cell chirality bias expressed. To find the optimal micropattern width where the cells exhibited the most pronounced chirality bias, we used RGD-peptide micropatterned hyaluronic acid hydrogels with lines of various widths (10–400 μm). Furthermore, we investigated the local expression of cell chirality across the width of the lines. When studying the effects of e.g., disease states or toxic substances on cell chirality, where small changes, such as randomization or a reduced exhibited cell chirality bias might be the result, these small changes may go undetected on suboptimal micropattern sizes. We envision that our results can serve as a guide for further investigations on cell chirality.
Cell chirality can be defined as the biased cell orientation with respect to the micropattern direction, where we define the cells oriented between Θ = −90° and 0° as exhibiting a negative chirality bias and the cells orientated between Θ = 0° and 90° as exhibiting a positive chirality bias (Fig. 1B). To visualize the line patterns, we used a fluorescently conjugate RGD peptide. The cells were visualized staining the nucleus, zonula occludens 1 (ZO-1), and F-actin. To investigate the cell orientation with respect to the micropattern we used CellProfiler™, an open source image analysis software.20 Here, we inputted confocal maximum intensity projections and calculated the orientation of the cell body, nucleus, and actin fibres compared to the micropattern (Fig. 1C).
CellProfiler™ calculates the orientation of the cell by fitting the segmented cell to an ellipse. The orientation then corresponds to the angle between the x-axis (determined by the RGD pattern direction) and the major axis of the ellipse, giving values between ±90° (ref. 20) (Fig. 1B). A cell with its major axis positioned to the right will have a negative orientation (−90° to 0°) and a cell with its major axis positioned to the left, will have a positive orientation (0° to 90°) (Fig. 1B).
Cellular chirality bias, and conversely cell orientation, has previously been measured in different ways e.g., by studying the cell body orientation, nucleus orientation, or the orientation of other organelles in relation to each other.15,16,18 Here, we measured both the cell body (as depicted by ZO-1 staining in the cell membrane) and the nucleus (as depicted by Hoechst staining) orientation and compared that to the micropatterned line direction. From our previous study investigating the alignment of brain microvascular endothelial cells, we had not found any statistically significant differences between the alignment of the cell body and the nucleus, on a population level, and these two measurement are essentially interchangeable.21 However, we still wanted to make sure this was true when it came to the chirality bias as well. Hence, we investigated the chirality bias of both the cell body and the nucleus and found that there were no statistically significant differences between these two measurements on the same line width, even though it was noted that the cell body chirality measurements displayed a more pronounce trend as the lines’ width increased (Fig. 2A).
By looking at the percentage of cells exhibiting a negative or positive chirality bias on the different micropatterned line widths it could be seen that the cells exhibited their most pronounced bias at the 100 μm wide lines for both the cell body and the nucleus as compared to the cells seeded on the non-patterned hydrogels (p value ≤ 0.001) (Fig. 2A, ESI Tables 1 and 2†). It could also be observed that going from the 10 μm to 100 μm wide lines, the exhibited chirality bias increased more and more. At the 200 μm wide line the chirality bias seemed to be decreasing again, while still being exhibited. On the other hand, cells on the 300 μm and 400 μm wide lines did not seem to be exhibiting any chirality bias compared to cells seeded on the non-patterned hydrogels, and we did not find any statistically significant difference between theses either.
Finding the most pronounced chirality bias on the 100 μm wide lines is also supported when looking at the mean orientation of the cells. Not only are more cells negatively oriented on these lines but the mean orientation away from 0° is also highest here, with a mean orientation angle compared to the line direction of −9.19° (Fig. 2B, ESI Tables 3 and 4†). When the line width is decreased to 10 μm, the cells’ mean orientation angle is instead approaching 0° (0.14°). When the line width is instead approaching infinity from the cell’s perspective, on the non-patterned hydrogels, the mean orientation angle even becomes positive (1.73°). Statistically the mean orientation of the cells on the 100 μm wide lines are significantly different from the cell mean orientations on all but the 75 μm and 200 μm wide lines.
For the nuclei orientations, this trend is not as evident and even reverts, as the orientation away from 0° instead increases, going from the 25 μm line and approaching the 10 μm wide line (ESI Tables 5 and 6†). Even so, the nuclei orientation still confirms that brain microvascular endothelial cells on the 100 μm wide lines exhibit the most pronounce chirality bias, and the nucleus mean orientation on the 100 μm wide lines also showed a statistically significant difference compared the nucleus mean orientation on all other lines except the 75 μm and 200 μm wide lines, as seen for the cell mean orientations as well. In addition, there was no statistically significant difference between the cell and the nucleus mean orientation on these lines (p = 0.547).
While the cells on the 10 μm wide lines are strongly oriented with the line direction, they experience less freedom to move, both their cell body and their nucleus, and exhibit their chirality bias. This is likely due to these cells receiving simultaneous orientation cues from touching both appositional boundaries. Hence, these cells cannot exhibit their fully pronounce chirality profile.
The cells on the wider lines, as well as the non-patterned hydrogels, on the other hand, don’t receive enough orientation cues to exhibit any pronounced chirality bias. On these wider lines or areas, only the cells close to the line boundaries receive directional cues, but these don’t reach the cells in the line centre. The directional cues from the line boundary are propagated inwards to neighbouring cells. However, the directional information is usually only propagated a few cells inwards from the boundary, and the propagation length is highly cell type dependent.14 The 100 μm wide lines here provide just the right balance between directional cues and freedom to move for the brain endothelial microvascular cells to allow them to exhibit their most pronounced chirality bias. These results give important implications as a suboptimal pattern width, leading to suboptimal exhibition of chirality bias, might have detrimental effects on further experimental results. Small changes in cell chirality may go undetected if suboptimal micropattern sizes are used. This might be detrimental when studying the effects of e.g., disease states or toxic substances on cell chirality, where such small changes, such as randomization or a reduced exhibited cell chirality bias might be the result.
Overall, both the cells, nuclei, and F-actin fibres are exhibiting a negative chirality bias in all five regions, however, this bias is more pronounced in region 3 located in the centre on the lines (Fig. 3B, ESI Tables 7–9†). On the opposite, the exhibited chirality is less pronounced in regions 1 and 5 located next to the line boundaries, where region 5 seems to have the lowest exhibited chirality bias. This is supported by the fact that both the cell and nucleus chirality bias was found to be statistically different from the centre region (3), while region 1 was not. The actin fibre chirality bias is also more pronounce in the centre regions (2–4) as these are statistically significantly different from the side regions (1 and 5). As there are more actin fibres than cells and nuclei smaller differences between these regions also becomes statistically significantly different as n increases. It can also be seen that both cells, nuclei, and F-actin fibres are exhibiting a negative orientation bias, where again this is more pronounced in the centre region 3 (Fig. 3C). In this centre region the mean orientation away from 0° of both the cells, nuclei, and F-actin fibres is also most pronounced and in line with the rest of the data (Fig. 3D, ESI Tables 10–15†). This is also supported by the presence of a statistically significant difference between the side regions (1 and 5) and the centre region (3) for the cell mean orientation, however, no statistically significant difference was visible on the mean nucleus orientation. We also observed a pronounced statistically significand differences in the actin fibre orientation between the different regions.
These differences can be explained by their distance from the outer boundary of region 1 and outer boundary of region 5. In region 3, in the centre of the line, the cells still receive directional cues from the cells closer to the boundaries, but they also have more freedom to move and exhibit their chirality bias. In region 1 the cells are restricted by the boundary on one side and free to move on the other. However, as they are preferably turning to the left, or towards the outer boundary of region 5, as they are negatively biased, they still have the freedom to move in this preferred direction and are thus only partly hindered to exhibit their preferred chirality bias. The cells in region 5, on the other hand, are restricted by the boundary on the other side. In this case, they are more hindered to exhibit their preferred chirality bias as they want to turn towards the region 5 outer boundary, but are hindered by the boundary.
Understanding the presence of differences along the line width is important when designing chirality experiments. Rather than analysing all cells, perhaps the cells along the outer boundaries should not be include in the analysis. Since these cells are more sterically hindered to exhibit their preferred chirality bias, by the presence of the boundary, subtill difference in experimental outcomes might be masked when including these cells in analysis, especially on the narrower lines where de facto a higher ratio of the cells will be closer to the boundaries. Choosing the optimal pattern width and including only the centre cells in the analysis may thus provide more physiologically relevant chirality experiment results.
We measured the mean intensity units of ZO-1 per cell in the different regions (1–5, Fig. 4A). It could be seen that the ZO-1 intensity per cell is higher in the three centre regions (2–4) than in the edge regions (Fig. 4B, ESI Table 16†). This is naturally so, because the cells in the centre regions have more cell–cell contact, while the cells on the boundaries don’t have any cells to form junctions with along the boundaries. As the cells along the boundaries are also more hindered to exhibit their chirality bias more mismatches might occur here with their neighbours on the centre side. Chirality mismatches, where neighbouring cells’ exhibited chirality is not matching, leads to poorer junction formation and thus less ZO-expression.
We also observed that the ZO-1 intensity per cell seems to be the highest in region 2, however, there was no statistically significant difference between the three centre regions (2–4). As region 2 is closer to the outer boundary of region 1 than to the outer boundary of region 5, they are less hindered to exhibit their preferred chirality bias and turn in the direction of the outer boundary of region 5, than the cells in regions closer to the outer boundary of region 1. This might result in less chirality mismatches and thus a higher ZO-1 expression in this region. However, this is not the region in where we observed the most pronounced chirality bias exhibition, as was seen in region 3. We hypothesize that even though the cells in region 3 exhibits a more pronounced chirality bias this region has more mismatches against the cells in region 4, leading to less ZO-1 expression in this region. As the cells in region 4 are situated closer to the outer boundary of region 5 they are more sterically hindered to turn towards this boundary and fully exhibit their preferred chirality bias. This might then create mismatches against the cells in region 3 which are freer to turn and exhibit their chirality bias.
When conducting cell chirality studies is it important to optimize the pattern width to where the cells exhibit the most chirality bias. If too small patterns are used, the cells’ freedom to move is constrained, as the cells receive alignment information from both boundaries. This leads to that the cells cannot exhibit their fully pronounced chirality bias. When cells are seeded on too large patterns there is instead a decrease in exhibited chirality bias, as the transmission of the boundary alignment information cannot propagate between all cells across the whole pattern. At the optimal pattern width, the balance between the two boundaries is right, and the cells can exhibit their fully pronounced chirality bias, while still being aligned by the micropattern. In addition, based on the wide difference in chirality bias exhibition along the line width, cells along the boundaries, that are more sterically hindered to exhibit their preferred chirality bias might compromise a chirality-based analysis when small induced changes might be masked by their presence. Finding the optimal micropattern width and include only the cells in the centre of the micropattern in the analysis for chirality-based experiments might have a great impact on further experimental results.
This work serves as a guide to determine the optimal micropattern width and analysis area for further investigations where cell chirality bias might be lost due to substances causing cell dysfunction or disease.
Chirality studies consisted in seeding bEnd.3 cells (passage between 24 and 35) on the patterned hydrogels (lines 10, 25, 50, 75, 100, 200, 300 and 400 μm in width) at a seeding density of 20000 cells per cm2. The cells were cultured on the at 37 °C in 5% CO2 for 24 h until they were fixed and analysed.
The ZO-1 primary antibody was dissolved (1:100) in 50% blocking buffer in PBS and incubated overnight at 4 °C. The samples were then washed three times with Triton X-100 (0.3%). The goat anti-rabbit Alexa Fluor 568 antibody was dissolved in 50% blocking buffer in PBS and incubated for 2 h at room temperature (RT). The F-actin staining was performed using SPY620-actin and incubating for 1 h at RT followed by nuclei counterstaining with Hoechst 33342 (2000X) and incubating for 20 min at RT.
Cell alignment to the line micropattern was calculated using CellProfiler™,20 as previously described.21 Briefly, the nuclei were segmented and identified using the IdentifyPrimaryObject module on the nuclei image. Next, the cell was segmented using the nuclei as a seed in IdentifySecondaryOject on the ZO-1 image. Then, the cell and nuclei orientation were calculated using the MeasureObjectAreaShape module, where the objects are fitted into an ellipse. Then, the angle between the major axis of the ellipse and the x-axis, determined by direction of the underlying RGD line pattern, was calculated. The orientation values range between ±90°. An object whose major axis is oriented between −90° and 0° was deemed as negative, and an object that orientated between 0° and 90° was deemed positive (Fig. 1). The negative and positive directions are determined by the numpy coordinate space, where the coordinate origin is situated at the top left of the image.
To determine the cell chirality regionality data was obtained from analysing 12 images, collecting four images each from three different hydrogels patterned with the 100 μm wide lines which were divided into five separate 20 μm wide regions. The total number of cells analysed was around 900.
Statistical analysis of cell, nucleus, and actin orientation of brain microvascular cells in the different regions were assessed using a chi-square test.
One way analysis of variance (ANOVA) with a Games-Howell post hoc for multiple comparisons at a 95% confidence level was used to compare the mean ZO-1 expression per cell on the different line regions.
The significance is symbolized by non-significant or N.S. (p > 0.05), * (p ≤ 0.05), ** (p value ≤ 0.01), *** (p value ≤ 0.001).
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d2ra05434e |
‡ M. T. and M. A. jointly supervised this work. |
This journal is © The Royal Society of Chemistry 2022 |