Hi Utku, I have worked with endos before, but not in flow conditions.

My first observation is that these endos are overconfluent. Maybe starting with a 50% confluent cell density will help.

Endos that are crowded are linked together to form a solid impermeable monolayer through cadherins and CD31.

Confluent endos don't really have a lot of space to modify their cell-to-cell connections and realign themselves.

Secondly, how can you tell if they are aligning with or against the flow?

Hi Steingrimur,

Thanks for your feedbacks. Yes, these ones are a bit crowded. I tried half concentration (here they are 10 M cells/ml and I did the same experiment with 5M cells/ml as well). However, I couldn't get consistent results with that neither. The next think that will be doing is going to be seeding the cells at even lower concentrations and letting them incubate couple days inside the microchannels as they reach confluency. Then, I going to start the flow.

For the exact direction of alignment, I fix and stain them for nucleus and Golgi. Under uniform shear stress the nucleus is dragged downstream and I measure the angle between the line that passes through nucleus and Golgi and the direction of fluid flow for each cell.

Dear Utku,

Try to check if there are any fluid dyamics problems. Here what you can try/check:

1. Post the geometry of your device so that we can evaluate if some seconday flows may be present

2. Use tracer partices to confirm the flow velocity in your cell.

3. Microchannels made of elastic materials under high pressure stretch in the radial direction, which may be imply a stress in the direction perpendicular to the flow, which may explain the cell aligment. You need to reduce the constrictions to the flow downstream to the region of interest to reduce the total pressure.

Hi Utku, keep us informed. This is interesting to me.

Hi, even 100% confluent ECs (cobblestone) should align parallel under flow conditions. Taking subconfluent cells is not physiological because these cells need cell-cell contacts. How long do you perform your experiments? It could take some time for the alignment (12 hrs) takes place. What is the difference between Exp.1 and Exp. 2 in your photo ... different locations? Could be that, e.g. at bifurcations, you have different (=non laminar) flow conditions.

Hi Andreas, why do say that that static 100% confluent ECs should align parallel to a suddenly induced flow?

I have worked with many types of primary ECs and you never let them go to 100% confluency because many become senescent at that density.

In vivo, ECs are almost always exposed to flow whether confluent or not.

Although I have to admit that this pure EC system and flow is very artificial since there are no SMCs present to support the ECs. It could be a model for leaky tumor capillaries consisting of only ECs.

Hi, I stated 100% because it was suggested to start with 50% confluency. ECs do not align to flow or shear stress in a short time, it wil take some time. I may also be dependent on the amount of shear stress. I worked for quite some time with self-isolated primary HUVEC (P0) which are exposed to laminar shear using a cone-plate viscosimeter in a range of 10-30 dyn/cm2. We also observed this behaviour in ibidi µ slides after exposure to 10 dyn/cm2. You have to take them 100% confluent because some genes are expressed in dependence of the cell density/cell-cell contact like, e.g. caveolin-1 which is known to be important for the caveolae formation in these cells.

In our hands, 100% confluent cells do not become senescent, they are stable for some days. How do you define "senescent"? 100% confluence is their natural state. Of course, every cell culture model is some kind of artificial because other vascular cells like SMCs are missing, even ECs are sitting on the membrane elastica interna and produce theit own matrix. Becaues in vivo ECs are continously exposed to shear stress, flow is a kind of "sense of home" for these cells.

Hi everyone,

Thanks for your discussion.

@Joao Miranda:

1- I ran experiments with multiple geometries to understand whether this inconsistency is caused by the unpredicted flow profiles. I did simulations on all of them and I haven't seen any secondary flow. Some of the geometries are attached. The first one is from another publication (geometry attached) and it has 4 channels with different shear stress magnitudes in each. In this article, they showed bEnd.3 cell line oriented parallel to the flow in this microchannel design. I, again, have seen HUVECs and HPAECs orienting perpendicular to the direction of the flow in these channels. I also tried PPFC-like channels made of PDMS (geometry attached, dimensions are in mm) but I again HUVECs did not orient parallel to the flow at the shear stress magnitude of 25 dyne/cm2.

2- Tracer particle is a good idea, though I measure the flow rate before and after the each experiment by measuring the media accumulated within a defined period of time from the outlets. But it can give some idea about any unexpected disturbed flow in the channels, if it exists.

3- Again good point, there are couple of articles in the literature dealing with this problem. I had a chance to read them before having those experiments and I followed the design guidelines that they present in these articles (i.e. thick enough PDMS ceiling). I also bake the PDMS channels for like 1 hour at 90 C after plasma bonding which should further increase the stiffness of the PDMS.

@Andreas H. Wagner:

1- I agree. I am normally using them as confluent monolayer (I tested half concentration as well though, but inconsistency in cell orientation persisted). Although, I use confluent monolayer, there is a catch: I seed the cells into channels at high concentration so they become confluent immediately after they spread on the FN-coated substrate. I culture them like that and I start the flow the next day (We have shown the cell-cell junctions established through ZO-1 staining the next day). On the other hand, in the literature people culture cells in average 2-3 days before they start the flow. Do you think would this pre-static culture time is important? Maybe there are other cell-cell junction proteins that needs more time or cells need to lay down their own ECM for proper morphological response. To test if this is the case, I can seed the cells at a lower density and wait 2-3 days before they become confluent.

2- I ran all the experiments for 24 hours. I was thinking about having 48 or 72 hours experiments but this is not going to solve the problem as 24 seems more than enough based on the literature.

Hi Andreas, The primaries I have worked with become senescent at 100% confluency, which means that they don't grow if you try to split them at that density. I usually didn't use EC's past p5 and split them at 60-70% confluency.

Just curious, are you using HUV-EC-C CRL-1730 (ATCC) or CRL-4053?

@Utku: Could be that ECs are somewhat activated because of the seeding procedure. They need some time to attach and produce their own matrix. So if possible, 2-3 days time before starting the flow experiment is advisable.

@Steingrimur: I would not call it senescent when ECs stop growing at 100%. At this point they establish cell-cell contacts and change from the proliferating phenotype to the quiescent phenotype... the latter situation is the normal phenotype of an EC because in a vessel ECs are sitting at their place for quite a lifetime. We isolate our HUVEC from umbilical cors which we collect from hospitals. Never used cells from ATCC.

@Utku: We also tried such a complicatzed flow device like you have some time ago and made the experienec that it is quite difficult the seed the cells in a stisfactory manner and get a confluet monolayer in the curve shaped channels. They do not grow properly and seeding with high cell numbers and starting 24 hrs later is also artificial I am afraid. May be your observation is a product of chamber geometry and seeding-stressed ECs. But I have to admit that's difficult to solve such a problem. Good luck!

Hello,

....happy to find this interesting chat, but I am afraid I am a bit late to jump on a wagon.

I entirely agree with Andreas's comments with respect to EC culture conditions and alignment under flow conditions. I have a long-lasting experience in working with primary ECs under various conditions, including "flow". Based on my observations ECs in culture will always align with direction of the fluid flow! The degree of alignment will depend on both shear stress and duration of the experiment (i.e. time of perfusion).

Best of luck!

Gedas

Yes, that's what I was thinking Andres. Now, cells are growing inside the channels beginning from lower concentration and I am going to start the flow when they reach confluency.

In those compilated channels the curved regions are just to create 'fluidic resistance' to split the fluid into 4 chambers at different ratios. In this way, you can get wide range of shear stress in the separate compartments of the same microfluidic device. We actually image only the wide uniform shear stress regions at the straight chamber s at the end.

Try prewetting the microchamber before the Fn solution us added.

I had the same experience around 2 month ago. The cells aligned vertically to the flow. If I let the flow go for many days. Sometimes they align after 4 days. Did you solve the problem now?

I recently did flow on HUVECs at 40dyn/cm2 for 24 hours using IBIDI flow system and also observed perpendicular alignment relative to direction of flow. I think that the alignment of ECs is confluency dependent. Did you figure out what the problem was in your case?

Utku Sönmez I just found this thread, and have been experiencing this same phenomenon for the last several months. I have chased down the same leads Utku did, with no resolution to why they do what they do. I am also using custom devices, and so I have mapped flow patterns with beads, checked for flexing of the substrate due to pump using embedded beads, and tried various geometries, and a couple difference EC types.

The only thing that seems to work to get parallel alignment with flow is very wide (>400 um) flat channels, or if I pre-align the matrix with collagen biased towards the flow direction (happy to describe how if needed). Otherwise, the cells orient perpendicular, and quite robustly.