Check the article by D. Bouis et al, Angiogenesis 4, 91-102, 2001. They review pretty much of the different immortalized cell lines. We used EA.hy 926 quite often, they are fine and you can buy them from ATCC. Passaging of HUVECs...never did this more than 5-10 times at the max, otherwise cells behaved differently every time with longer passage ##, stopped growing, gave inconsistent responses. FCS may affect your particular measurements as well, did you try omitting it (or reducing it to, say, 1% a day before you start assays)?

HUVECs can be used to p5 but they do show some variability donor to donor, and they don't always react similarly to other primary human endothelial cells. Endothelial cell lines may give you more repeatable results, but they will be harder to sell to reviewers. Primary human aortic, coronary artery, or microvasculature endothelial cells, depending on your disease model, are ideal, and you can generally get some basic patient information (sex, age, cause of death) to reduce variability.

if you isolate HUVEC yourself, the lack of reproducibility in the response might be due to donor variation. Since we started using commercially available HUVEC cells pooled from several donors (Lonza), our experimental variation has decreased. Immortalized cells respond often differently than primary cells since most probably some signaling pathways are changed. You should then compare primary vs cell line in the context of your studies.

As said above, You can try pooled (10 donor) HUVECs, which may reduce line to line variability. Also P5 is probably the max they should be used.

I would agree with all of the earlier responses. Pooled HUVEC are definitely more consistent performers than single donor, but HUVEC are also unusual endothelial cells and reviewers often balk at studies performed or proposed using HUVEC alone. Primary cells are arguably more clinically or physiologically relevant to most disease models, but at least in our hands they are often even more difficult to produce consistent results with than HUVECs. We tend to use pooled HUVEC to work out basic experimental conditions and then try to reproduce the response in a primary cell model. And do not underestimate the effect of serum. Fresh serum at the start of an experiment can have huge effects on the cells; on the other hand many ECs will begin to undergo apoptosis if kept in less than 1% serum for more than a few hours.

I would use HUVEC for the reasons described above by others. Please try the same passage of cells (e.g. 3rd or 4th passage) everytime. I agree with others that P5 should be the maximum. Whatever passage you use, please keep doing your exp at the same passage everytime. Other critical factors are confluency of the cells, please try 85-90% confluency for your experiments. I use the following protocol: Let the P3 cells become confluent in a 10-cm plate by keeping them fed with fresh EGM2 medium on alternate days, split them to 1:8 ratio around 5 PM, let them grow for 48 hours, Starve them in 0.5% serum plus EBM2 medium overnight and next morning treat them around 7 AM with or without VEGF. If your expoeriments do not require starvation (as in the case with TNFa usually), I would split a confluent P3 HUVEC 10-cm plate in the morning (around 8 AM) to 1:8 ratio, then let them grow for 48 hrs, and perform experiments around 8-9 AM. We always use HUVEC from the same commercial supplier. Please feel free to email me with further questions: [email protected]

Primary HUVEC should not be propagated beyond passage 5 (cell isolation being passage 0). If you buy in your primary cells, check at which passage they arrive, as it varies between companies. As mentioned by others, pooling cells from various donors reduces but does not remove variability. If you buy in single-donor HAEC, for example, cells from young, male donors tend to be less variable. Also, avoid cells from donors which smoked.

Serum-starving your cells for 6h before an experiment should also help to reduce variability.

In addition to the above comments I could add a few extra points:

1) Minimize the number of HUVEC "division" (not just passages), i.e. seed cells at higher density.

2) Do experiments at p1-2 (if you can). If the cells "work" - freeze them in several aliquots. That will help you to eliminate the "donor" factor (which inevitably always will give you quite a few surprises that you will not be able to figure them out...) and will provide with enough of cells to finish the experiment. Don't go any further that p5-6 (even if morphologically HUVEC will look very nice). With division/passage number HUVEC tend to loose some functions, including the ability to induce oxidative stress etc.

3) Compare not only the levels of "kinin"-dependent suppression of TNF-a-induced responses, but also the magnitude/levels of TNF-a-induced (or any other "positive" control agent-induced) activation of HUVEC. If the HUVEC are "normal" - the levels of TNF-a-dependent induction of ICAM-1 will be similar from donor to donor and from experiment to experiment.

Good luck!

We use HUVEC only up to passage 1. Passaged cells were known to dedifferentiate. We isolate these cells ourselves after collecting umbilical cords from several hospitals. If HUVEC were passaged eNOS expression and NO formation are quickly going down in higher passaged cells. NO is an important factor for expression regulation of inflammatory genes. Thus, a NO breakdown will influence inflammatory gene expression. I strongly recommend against endothelial cell lines. They may look like endothelial cells but the classical physiology is missing. They are just like containers … For general questions I would recommend to pool cells from 4-5 umbilical cords, to avoid an influence of individual single nucleotide polymorphisms (SNPs).

We found that if HUVECs are used beyond passage 3, they loose responsiveness to TNFalpha-induced ICAM expression inhibition triggered induced by some vasoactive peptides. Anyone had seen such phenomenon in the lab? Thanks.

A few more observations:

1. HUVECs are better as they are more physiological and one can capture the donor-to-donor variation (hence, greater acceptance and applicability if an observation is noted in cells from different donors).

2. Always use HUVECs in passages 2-4. The point about using same passage number (as noted by a previous poster) is great.

3. You can always normalize your read-outs to that of untreated HUVECs in order to minimize variablity (widely done in case of western blots).

4. Use HUVECs only when they are confluent. Cell-cell contacts are necessary for normal signaling events.

5. We always keep HUVECs in low serum media (1% or less) overnight before experiments. It helps to quiesce them and synchronize their cell cycles.

Hope this helps.

We used HUVECs for a while to study eNOS and found after careful comparison that these cells contain considerably less eNOS protein and activity than HCAECs. The point being, you need to ensure the appropriate machinery for your model is indeed present and robust. Passaging is critical and again, carefully controlled experiments assessing the expression and specific endpoints of your experiments over several passages is critical. This can be your first figure in your manuscript.

I have a different way of looking at things from all the above contributors. In the early 1980's we were looking into Endothelium Derived Relaxing Factor (EDRF), now known to be Nitric Oxide. My job was to isolate primary endothelial cells from different species for experiments using a classic "cascade system" using strips of rabbit aortae.

I isolated endothelial cells from Human Umbilical veins, Porcine and Bovine thoracic aorta.

The first thing to mention is the number of cells isolated from the different tissues. Unsurprisingly the least amount was from Umbilical Veins (HUVEC's), next the Porcine and then the Bovine gave us the most cells. However the release of EDRF from the endothelial cells were different in all cases. We chose to use Porcine aortic endothelial cells (PAEC) because:-

Tissue was easily available on a daily basis.

The number of cells at Passage 1 available for experiments was huge i.e. between 1.5 and 4 x 10^7

The release of EDRF was immediate but not long lasting with addition with Bradykinin (1-100nM).

We were criticized heavily initially because we were not using Human cells and "what was the relevance"? However these particular experiments would have been impossible to do with HUVECS but our paper was one of the first to recognize EDRF as Nitric Oxide (Nature, Vol 327 N0. 6122, pp 524-526), 11th June 1987 Moncada et al).

Also we had to use the PAEC at Passage 1 because if the cells were further passaged the EDRF response was lost completely. We also looked at prostaglandin release profiles in these cells and they were found to change dramatically over a small number of passages. So logically any endothelial system that you want to look at will change dramatically over passage numbers

Todd gives the best advice i.e. to use an inducible cell line for eNOS.

We also use  "mature"  HUVEC 4-8 days (Day 8-12 culture) post confluence and plated on collagen I.  

Dear Prof. Rahleb,

I would use HUVECs up to passage 9 if cells are kept in highly specified media with all the essential growth factors (Endothelial Cell Growth Medium) can be purchased from Sigma. However, the immortalized cells are more stable and easy to grow.

Good luck!

HUVEC reflect a physiological state od endothelial cells, of course with specific tools/features (protein, signaling etc.) to respond to changing situations (e.g. stimulation with pro-inflammatory cytokines) which may differ from other endothelial cell types/lines. However, for many experimental questions HUVEC should be a first choise to study any effects which may exhibt a (patho)physiological relevance. Thus, I would propose to use HUVEC. However, to test whether your effects can be seen in endothelial cells per se, I would also propose to use other endothelial cell types, too (such as HMEC_1, HAoEC etc.) and to perform a comparative analysis of your expected effects/your experiments under pro-inflammatory condition (following my own experiences these reaction should be quite similar, e.g. see Circ Res. 2009 Mar 13;104(5):589-99. doi: 10.1161/CIRCRESAHA

and/or Oncol Rep. 2013 Jul;30(1):462-70 doi: 10.3892/or.2013.2413

and/or Circ J. 2011;75(5):1235-42

and/or Circ J. 2009 Sep;73(9):1746-52

and/or Biol Chem. 2008 Oct;389(10):1333-8. doi: 10.1515/BC.2008.152.

etc.).