In this study we provide evidence that replicative senescence causes disrupted cell-cell contacts between endothelial cells (HUVECs). These disrupted cell-cell contacts are a result of alterations in the distribution of adherens junction proteins and decreased expression of tight junction proteins. These alterations are not only observed between senescent cells, but also at cell-cell contacts between non-senescent and senescent cells and along the entire periphery of non-senescent cells lining a senescent cell. The alterations in distribution and expression of junction proteins coincides with increased permeability for small (450 Da) and large (44 kDa) tracer molecules, indicating that the barrier function is compromised. Monolayers of co-cultures of non-senescent and senescent HUVECs also showed increased permeability for both tracer molecules. In addition, the expression of cPLA2α, that has been shown to play a role in the maintenance of the integrity of the endothelial barrier , was severely down-regulated in senescent cells. This down-regulation of cPLA2α might play a role in the decreased barrier function of senescent monolayers.
Senescent endothelial cells have been identified at sites of atherosclerotic lesions [10, 11] and have been proposed to contribute to the onset and/or the progression of atherosclerosis [11, 20, 21]. However, a causal relationship between senescence and the development of atherosclerosis has not yet been proven and, to our knowledge, the effect of replicative senescence on endothelial cell-cell contacts has not been studied before. Using non-replicative senescence models (either over-expression or silencing of specific proteins) did indicate altered barrier function of an endothelial monolayer [22, 23]. Venkatesh et al.  observed that the activation of the notch signaling pathway in endothelial cells results in a senescence-like phenotype and is associated with increased permeability of the monolayer caused by altered VE-cadherin and beta-catenin expression and localization. Huang et al. showed that in human brain endothelial cells, silencing of TERT or inhibition of TERT activity, leads to senescent features and affects the junctional complexes, which is accompanied by reduced expression of tight junction proteins, including ZO-1. However, no functional measurements were done is this study. In the present study we did not observe down-regulation of ZO-1, but we did observe an alteration in distribution of ZO-1.
We have recently shown that Golgi-localized cPLA2α is involved in Golgi-to-plasma membrane trafficking of junction proteins . The down-regulation of cPLA2α in senescent cells, as described in this paper, could therefore disrupt the trafficking of junction proteins to the plasma membrane and thereby altering the distribution and expression of these proteins. Taddei et al. reported that in endothelial cells clustering of VE-cadherin at the cell-cell junctions is necessary to induce transcription of claudin-5. We hypothesize that disrupted trafficking of VE-cadherin to the plasma membrane of senescent cells, at least partly due to the decreased expression of cPLA2α, causes a decrease in VE-cadherin signaling in surrounding young cells. This subsequently leads to decreased expression of claudin-5 in these young cells.
Senescent cells are known to secrete factors that can affect the structure and function of neighboring cells . However, the fact that the effect of senescent cells on junction morphology is only observed in cells that are in contact with senescent cells suggests that, in our setup, secreted factors do not play a role. This is supported by our finding that pre-conditioned medium from senescent cells did not effect on the distribution pattern of claudin-5 and VE-cadherin in non-senescent cells (data not shown).
We hypothesize that, in vivo, the presence of senescent endothelial cells exerts the same effect on endothelial barrier function as in the current in vitro model. Accumulation of low density lipoprotein (LDL) in the intima can be, at least partially, attributed to disrupted endothelial junctions  and is seen as one of the initial steps in atherogenesis [18, 19], especially at bifurcations where blood flow is disturbed. Since disturbed flow increases endothelial cell turnover in vivo [32, 33], it is to be expected that endothelial cells at these locations senesce faster. Indeed, senescent cells have been identified in vivo in atherosclerotic tissue [10, 11]. Once the atherosclerotic process starts, the extensive oxidation of LDL in the intima could lead to mitochondrial damage and ROS production in endothelial cells , further promoting cellular senescence directly or indirectly by an increased cell death and thus increased cell turnover. Mathematical modeling of the dynamics of endothelial cell damage, repair and telomere shortening suggests that in humans, at an age of 65 years, approximately 2-5% of the vascular endothelial cells are senescent .
The presence of senescent endothelial cells might affect the atherosclerotic process in several ways. First of all, the compromised junctional complexes and the subsequent increase in vascular permeability due to the presence of senescent cells in vivo might increase the transport of LDL over the endothelium, as described above. Moreover, endothelial cell senescence is accompanied by an increase in adhesive properties towards macrophages [7, 16]. This, in combination with the described decrease in tight junction protein expression upon senescence, might aid infiltration of the macrophages into the vessel wall, especially since claudin-5 is thought to be one of the most important junction proteins in permeability control .