News & Views | Published:


Border crossings

When inflammatory cells leave blood vessels to repair injured tissues, they are helped on their way by the endothelial cells lining the vessels. The invagination and expulsion of endothelial membrane may be the key.

Inflammation — described in antiquity to comprise 'rubor' (redness), 'calor' (heat), 'tumor' (swelling) and 'dolor' (pain) — is the first response of tissues to injury. Its purpose is to recruit inflammatory cells and other blood components to an affected area, where they repair the damage. To reach a tissue in need, the inflammatory cells must squeeze their way out through the walls of blood vessels, and the cells that form the vessel walls are an active partner in this process. But it is unclear how they accommodate and propel (or expel) migrating inflammatory cells. On page 748 of this issue, Mamdouh and colleagues1 reveal surprising details of the mechanism.

Inflammatory cells are together known as white blood cells or leukocytes, and include neutrophils, lymphocytes, monocytes and mast cells. Normally, fluids and solutes are exchanged between blood and tissues at the smallest of the blood vessels, the capillaries. But leukocytes instead emigrate from postcapillary venules, vessels that lead from capillaries into veins. The walls of venules have two layers: an inner, continuous lining of endothelial cells, and a surrounding mesh of support cells (pericytes). So leukocytes must crawl through both layers — in a process called diapedesis — in order to leave the blood at sites of injury. Details of how they do this, however, have been scant, as they leave little or no sign of their passage.

Some 30 years ago, it was shown that soluble proteins produced during inflammation cause endothelial cells to pull apart2, making the endothelial layer more permeable. The implication was that this could be the route that leukocytes and components of blood plasma take into inflamed tissues. Although it was not possible at the time to describe the precise anatomical path taken by the leukocytes, a previous analysis of tissue sections, showing the spatial relationship between the different cell types, suggested that leukocytes do squeeze out between adjacent endothelial cells3. Indeed, this is now well accepted. However, no obvious gaps in the blood-vessel lining are seen at sites of 'transendothelial' migration (transmigration)3. Moreover, there is a downside to a model in which leukocytes and plasma factors use the same exit strategy: there are likely to be situations in which blood vessels need to be permeable to one or the other, but not to both.

So several research groups have been looking at how the permeability of the endothelial layer is regulated. There is a rapidly growing list of adhesion molecules that occur preferentially at the borders between adjacent endothelial cells4,5,6; these border proteins include VE-cadherin, junctional adhesion molecules 1, 2 and 3, PECAM (also known as CD31) and CD99. Some of them maintain the integrity of the endothelial layer7; all are capable of binding with like molecules and have the potential to stick adjacent cells together. Interestingly, all except VE-cadherin and junctional adhesion molecule 1 are also found on various types of leukocyte.

Mamdouh et al.1 now build on these results, and suggest an unusual mechanism by which endothelial cells actively participate in leukocyte transmigration. Using cultured cells, and exploiting PECAM, VE-cadherin and CD99 as markers of endothelial-cell borders, the authors estimate that up to 30% of the membrane at these sites is invaginated in the absence of leukocytes. Border proteins move in and out of these invaginations, which are different from previously observed invaginations (such as recycling endosomes and caveolae), and may be a specialized property of endothelial cells. Mamdouh et al. also find that, when in contact with a transmigrating leukocyte, an endothelial cell transiently increases its surface area by disgorging its invaginated membrane where the two cells meet. This process seems to require, among other things, binding of PECAM on the endothelial cell to PECAM on the leukocyte.

These intriguing findings support the idea that, even as a leukocyte forces its way out between endothelial cells, it plugs the gap, using the same adhesion molecules that endothelial cells use to stick together. The transient increase in contact area between leukocytes and endothelial cells demonstrated by Mamdouh et al. could serve to maintain the seal, preventing plasma factors from leaking out during the one or two minutes it takes for transendothelial migration. In a further refinement, engagement of border proteins by a leukocyte (rather than an adjacent endothelial cell) might generate the responses in both cell types that are needed to propel the leukocyte through the junction (Fig. 1). It is also tempting to speculate that cycles of membrane expulsion and invagination, which would lengthen and shorten the borders between endothelial cells like a purse-string, could actually propel leukocytes through.

Figure 1: Model for how inflammatory cells reach tissues in need.

To escape the blood and reach injured tissues, inflammatory cells (leukocytes) must cross the blood-vessel wall. a, Leukocytes move out of the blood flow and roll along the layer of endothelial cells that lines blood vessels; they then stop, and move to the junctions or borders between endothelial cells. These cells are held together in part through the self-binding protein PECAM on their surface (green); PECAM also occurs on leukocytes. Mamdouh et al.1 show that PECAM is also located in membrane invaginations (red) in endothelial cells. b, When PECAM molecules on the endothelial-cell surface and on the leukocyte bind, the membrane invaginations are disgorged. This delivers stored PECAM to interact with leukocyte PECAM, perhaps generating signals that are required to move the leukocyte through the border. It may also plug the gap in the endothelial layer, preventing other blood components from leaking out. Curved arrows represent signalling events.

The ability to rapidly mobilize a supply of membrane could also explain a previously puzzling alternative means of transmigration. Early studies8 had concluded that neutrophils move between endothelial cells but that lymphocytes pass straight through them, actually penetrating the cell body. Subsequent work, however, showed that although lymphocytes do not pass through endothelial cells9, neutrophils can do so, albeit in very specific experimental conditions10.

The new results do not resolve this discrepancy, but do suggest a possible mechanism for 'transcellular' movement. Mamdouh et al. propose that a leukocyte could be engulfed by a single endothelial cell if invaginated membrane is discharged not onto the surface adjoining another endothelial cell, but instead onto the surface of the blood-vessel interior. (Indeed, some border proteins can become relocated to this surface, particularly during inflammation.) How exactly this might enable the leukocyte to pass through the endothelial cell is unclear, but it might exploit known cellular mechanisms for taking up and secreting molecules. Such a transcellular route could be particularly important in certain organs or for certain types of leukocyte. For example, in the central nervous system it would maintain the integrity of the endothelial layers that constitute the blood–brain barrier. Similarly, lymphocytes and monocytes continually survey the body for pathogens by moving from the bloodstream into tissues, and it might be advantageous for this to occur independently of changes in the permeability of the endothelium.

Although PECAM is not always involved in leukocyte transmigration11, Mamdouh and colleagues' model1 will help in identifying the molecular events that regulate the endothelial lining during inflammation. Previous studies from the same group4 provided compelling evidence that monocytes need PECAM to begin their penetration of endothelial layers, and CD99 to complete it. Other border proteins, such as the junctional adhesion molecules, might substitute for PECAM or CD99 during the transmigration of different types of leukocyte, depending on the tissue in need. It seems likely, then, that transendothelial migration involves a multi-step cascade of adhesive interactions that are coordinated in space and time, rather like the process by which leukocytes move out of the fast-flowing bloodstream and onto the inner endothelial surface before transmigration12.


  1. 1

    Mamdouh, Z., Chen, X., Pierini, L. M., Maxfield, F. R. & Muller, W. A. Nature 421, 748–753 (2003).

  2. 2

    Majno, G., Shea, S. M. & Leventhal, M. J. Cell Biol. 42, 647–672 (1969).

  3. 3

    Marchesi, V. T. Q. J. Exp. Physiol. 46, 115–118 (1961).

  4. 4

    Schenkel, A. R., Mamdouh, Z., Chen, X., Liebman, R. M. & Muller, W. A. Nature Immunol. 3, 143–150 (2002).

  5. 5

    Johnson-Leger, C., Aurrand-Lions, M. & Imhof, B. A. J. Cell Sci. 113, 921–933 (2000).

  6. 6

    Luscinskas, F. W., Ma, S., Nusrat, A., Parkos, C. A. & Shaw, S. K. Semin. Immunol. 14, 105–113 (2002).

  7. 7

    Corada, M. et al. Proc. Natl Acad. Sci. USA 96, 9815–9820 (1999).

  8. 8

    Marchesi, V. T. & Gowans, J. L. Proc. R. Soc. B 159, 283–290 (1964).

  9. 9

    Schoefl, G. I. J. Exp. Med. 136, 568–588 (1972).

  10. 10

    Feng, D., Nagy, J. A., Pyne, K., Dvorak, H. F. & Dvorak, A. M. J. Exp. Med. 187, 903–915 (1998).

  11. 11

    Duncan, G. S. et al. J. Immunol. 162, 3022–3030 (1999).

  12. 12

    Springer, T. A. Cell 76, 301–314 (1994).

Download references

Author information

Correspondence to Ann Ager.

Rights and permissions

Reprints and Permissions

About this article

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.