nature immunology volume 15 number 3 mArCH 2014 215

6. Benhamron, s. et al. Eur. J. Immunol. doi:10.1002/eji.201343953 (18 November 2013).

7. Upton, J.P. et al. Science 338, 818–822 (2012).

8. Iqbal, J. et al. Cell Metab. 7, 445–455 (2008).9. Coelho, D.s. et al. Cell Rep. 5, 791–801

(2013).10. Iwakoshi, N.N., Pypaert, M. & Glimcher, L.H. J. Exp.

Med. 204, 2267–2275 (2007).11. Iborra, s. et al. J. Clin. Invest. 122, 1628–1643

(2012).12. subramanian, M. et al. J. Clin. Invest. (in the press).13. segura, E. & villadangos, J.A. Traffic 12, 1677–1685

(2011).14. Tardif, K.D., Mori, K., Kaufman, R.J. & siddiqui, A.

J. Biol. Chem. 279, 17158–17164 (2004).15. Han, D. et al. Cell 138, 562–575 (2009).

favor of Xbp1 mRNA splicing15 could have therapeutic benefit, including bolstering CD8α+ DC–mediated host defense based on the work of Osorio et al.2 presented here.

COMPETING FINANCIAL INTERESTSThe authors declare no competing financial interests.

1. Joffre, O.P., segura, E., savina, A. & Amigorena, s. Nat. Rev. Immunol. 12, 557–569 (2012).

2. Osorio, F. et al. Nat. Immunol. 15, 248–257 (2014).3. Hollien, J. & weissman, J.s. Science 313, 104–107

(2006).4. Ron, D. & Hubbard, s.R. Cell 132, 24–26 (2008).5. so, J.s. et al. Cell Metab. 16, 487–499 (2012).

investigating disease processes characterized by chronic, robust ER stress, such as certain types of neurodegenerative disease and advanced atherosclerosis. The type of hyperactivation of IRE-1α necessary for RIDD may be present in such settings. Moreover, it is possible that RIDD is acti-vated in cells infected with certain types of viruses, in which IRE-1α is activated but the transcriptional activity of XBP-1 is blocked14. If future studies do identify a role for RIDD in natural disease settings in vivo, innova-tive work on strategies that disable RIDD in

ve-cadherin phosphorylation decides: vascular permeability or diapedesisAdama Sidibé & Beat A Imhof

Phosphorylation of the adhesion molecule VE-cadherin at tyrosine residues modulates the opening of endothelial junctions during inflammatory reactions. The replacement of two distinct residues in VE-cadherin shows that Tyr685 regulates vascular permeability and Tyr731 regulates leukocyte diapedesis in vivo.

During an inflammatory response, leukocytes and fluids from the blood must reach the

affected tissue. Tight regulation of endothelial junctions of the vascular endothelium is a critical molecular mechanism for controlling such events. VE-cadherin is a vascular endothelial adhesion molecule, specifically and exclusively expressed by endothelial cells, that controls integrity of blood vessels. In this issue of Nature Immunology, Wessel et al. report constitutive phosphorylation of VE-cadherin at Tyr731 in vivo and demonstrate that dephosphorylation of that residue contributes to leukocyte extravasation, whereas mediator- induced phosphorylation of VE-cadherin at Tyr685 results in vascular permeability1.

VE-cadherin is a type I transmembrane pro-tein with a calcium-dependent adhesive function in vascular cell-cell contacts. It has an extracel-lular portion that engages in homophilic interac-tions in cis to form dimers. Those dimers interact in trans and stabilize adherens junctions between adjacent endothelial cells. The cytoplasmic tail of VE-cadherin is associated with the cell cytoskel-eton via proteins of the catenin family. Both the VE-cadherin homophilic interaction and its association with catenins have considerable importance for the integrity of the endothelium.

Post-translational modifications of VE-cadherin, such as tyrosine phosphorylation or cleavage, are reported to be involved in the induction of vascular permeability and leukocyte transmigra-tion. Several such studies were done in vitro and provided controversial data about the specific tyrosine residues involved—thus the urgent need for study of the in vivo relevance of VE-cadherin tyrosine phosphorylation. Instrumental for such final proof of different VE-cadherin func-tions in vivo are the new knock-in mice express-ing phosphorylation-deficient VE-cadherin mutants with substitution of phenylalanine for tyrosine at position 731 (Y731F) or 685 (Y685F) generated by Wessel et al.1.

Vascular permeability and leukocyte extrava-sation are very tightly controlled events that occur under physiological and pathophysiological con-ditions during inflammation. Both events can be mediated by transcellular routes (which pass through the endothelial cell body) and paracellu-lar routes (which pass through cell-cell junctions). Several studies have indicated that the controlled opening of endothelial junctions is necessary for leukocyte transmigration and the egress of solutes from bloodstream to underlying tissues at sites of inflammation. The paracellular route is regu-lated mainly by proteins of endothelial tight and adherens junctions. Delineating the mechanisms that affect the functions of such proteins during inflammation is of chief importance and would allow the development of new targeted therapies that not only limit edema formation during stroke

or myocardial infarction and leukocyte infiltra-tion in inflammatory diseases but also improve drug delivery in solid tumors in which intersti-tial pressure is substantially elevated because of enhanced vascular permeability.

Therefore, during the past decade, several studies have been aimed at understanding the effects of vascular permeability-increasing fac-tors, such as vascular endothelial growth factor (VEGF), tumor-necrosis factor (TNF), platelet-activating factor, histamine, bradykinin and thrombin, on leukocyte transmigration, vascu-lar permeability and the function of endothelial junctions. Several of those studies have reported a correlation between the phosphorylation of VE-cadherin tyrosine residues and a decrease in the stability of adherens junctions. Five tyrosine residues of the cytoplasmic tail of human VE-cadherin are reported to be phosphorylated in conditions of increasing permeability or leu-kocyte extravasation and are proposed to par-ticipate in endothelial junction destabilization. In Chinese hamster ovarian cells transfected to express mutant VE-cadherin with a simple phosphomimetic tyrosine-to-glutamate substi-tution or a phosphorylation-deficient tyrosine-to-phenylalanine substitution, phosphorylation of Tyr658 or Tyr731 dissociates p120-catenin and β-catenin from VE-cadherin, which leads to decreased endothelial barrier function2. That first in vitro study suggested specific roles for Tyr658 and Tyr731 in regulating the stability of endothe-lial junctions. Human umbilical vein endothelial

Adama Sidibé and Beat A. Imhof are in the

Department of Pathology and Immunology, Centre

Medical Universitaire, University of Geneva,

Geneva, Switzerland.

e-mail: beat.imhof@unige.ch

NE ws AND v IE wsnp

g©

2014

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.

216 volume 15 number 3 mArCH 2014 nature immunology

cells (HUVECs) expressing either Y658F or Y731F point-mutant VE-cadherin show less leukocyte transmigration3. Studies using com-mercially available phosphorylation-specific antibodies have shown that engagement of the intercellular cell adhesion molecule ICAM-1 after firm vascular adhesion of neutrophils induces phosphorylation of VE-cadherin at Tyr658 and/or Tyr731 (ref. 3). Studies of VE-cadherin tyrosine-to-phenylalanine point mutants have reported that phosphorylation of Tyr731, Tyr645 and Tyr733 is required for the ICAM-1-mediated transmigration of lymphocytes4.

In their present study, Wessel et al. investigate the relevance of phosphorylation of VE-cadherin at Tyr731 by generating knock-in mice that express a Y731F VE-cadherin mutant1, certainly the most pertinent way to elucidate the specific role of a single amino acid. They show that the commercial antibodies to phosphorylated Tyr731 used in previously published studies bind to the Y731F mutant VE-cadherin in a phosphorylation-nonspecific way. That raises serious questions about the interpretation of

previously published data acquired with such antibodies. The authors develop new antibodies specific for phosphorylated Tyr731 and find constitutive phosphorylation of VE-cadherin at Tyr731 in the lungs of wild-type mice and in untreated HUVECs in vitro. The transmigration of lymphocytes induces dephosphorylation of Tyr731 via the phosphatase SHP-2, followed by endocytosis of VE-cadherin dependent on the AP-2 adaptor complex. Accordingly, the Y731F knock-in mice develop less lymphocyte extra-vasation. Moreover, they do not exhibit any sub-stantial difference in VEGF-induced vascular permeability relative to that of wild-type mice. Furthermore, other permeability-inducing fac-tors, including TNF and hisamine, do not affect the constitutive phosphorylation of Tyr731. This suggests phosphorylation of Tyr731 as a mecha-nism that controls leukocyte transmigration.

Those important findings suggest a previously unknown mechanism by which dephospho-rylation of Tyr731 leads to the internalization VE-cadherin dependent on SHP-2 and AP-2 and thus modulates leukocyte transmigration. It will

be important to elucidate the upstream events that lead to those events during leukocyte emi-gration. Although it is important, the inhibition of leukocyte extravasation is not complete in the Y731F knock-in mice, which indicates that other tyrosine residues of VE-cadherin might com-pensate for the absence of Tyr731. Candidates for this might be the previously reported Tyr658 or Tyr733. Studies of knock-in mice in which all those tyrosine residues are replaced with pheny-lalanine would be informative, but generating such mice would be challenging. Alternatively, independent mechanisms for the phosphoryla-tion of VE-cadherin may also exist.

The intriguing work by Wessel et al.1 indi-cates an unexpected role for SHP-2 in the desta-bilization of endothelial junctions. SHP-2 is classically known and considered as a junction- stabilization phosphatase that dephospho-rylates the VE-cadherin–catenin complex to maintain the integrity of endothelial junctions. Activation of SHP-2 is needed for the recovery of endothelial junctions and restoration of bar-rier integrity after stimulation with thrombin5. Wessel et al. now suggest that SHP-2 may have ‘dual faces’ with opposing functions that would change in a time-dependent manner during leukocyte diapedesis1. In the early phase, leu-kocyte emigration could induce activation of SHP-2, leading to dephosphorylation of VE-cadherin at Tyr731 and destabilization of adherens junctions, while during later phases, SHP-2 would dephosphorylate the catenins and restore barrier integrity. Further investiga-tion is needed to confirm such a dual function for SHP-2 during leukocyte extravasation.

Moreover, tyrosine phosphorylation is the result of a balanced action between phosphatase activities and kinase activities. Surprisingly, in the resting phase of the endothelium, Tyr731 of VE-cadherin remains constitutively phospho-rylated despite the presence of several protein tyrosine phosphatases (PTPs) other than SHP-2 in the vicinity of the VE-cadherin–catenin com-plex1. One PTP of particular importance that is expressed exclusively in endothelial cells and is constitutively associated with VE-cadherin is VE-PTP, a vascular endothelium–specific PTP. Its dissociation from VE-cadherin has been shown to be involved in leukocyte extravasation6. However, Wessel et al. now show that VE-PTP does not have any effect on the phosphorylation of VE-cadherin at Tyr731 (ref. 1), which demon-strates that leukocyte transmigration relies on at least two different molecular mechanisms that involve PTPs and VE-cadherin.

The tyrosine phosphorylation of VE-cadherin has also been correlated with an increase in vas-cular permeability upon stimulation with VEGF in vitro and in vivo. The direct stimulation of the effect of VE-cadherin phosphorylation

Figure 1 Mechanisms for the in vivo control of leukocyte diapedesis and vascular permeability by phosphorylated vE-cadherin. A switch in the phosphorylation of vE-cadherin upon challenge with leukocytes or soluble factors specifies the final outcome. Leukocytes induce the activation of sHP-2, which leads to the dephosphorylation of the phosphorylated Tyr731 residue (p-Y731) of vE-cadherin. Dephosphorylated Tyr731 associates with the AP-2 complex and elicits endocytosis of vE-cadherin. Conversely, permeability-inducing factors such as vEGF, TNF or histamine induce src-mediated phosphorylation of vE-cadherin at Tyr685 (p-Y685), which leads to vascular permeability. Cleavage of vE-cadherin dependent on the metalloproteinase ADAM-10 is induced by both processes, but this correlates only with permeability and transmigration and awaits proof obtained by studies of mutant vE-cadherin with an altered proteolytic binding site.

Leukocytediapedesis

Vascular

permeability

p p

Diapedesis switcher Permeability switcher

Diapedesis Permeability

Fluid and soluteegress

Extravasatedleukocyte

Pericyte

Endothelial cells

Vessel lumen

Endothelialcells

Leukocytes

Vessel lumen

TNF VEGF

Others

Cytoplasm Cytoplasm

Leukocytediapedesis

Vascular

permeability

PTP Kinase

ActiveSHP-2

ActiveSrc

AP-2

AP-2

?

?

Y731

VE-PTP VE-PTP

Y685p-Y731

Y685

p-Y685

p-Y731

Csk

ADAM-10 ADAM-10

VE-cadherin VE-cadherin

VEGF,TNF,

histamine

Y685 Y731 Y685 Y731

Y65

8 Y733

Y64

5

Y65

8 Y733

Y64

5

Leukocytes

Basementmembrane

NE ws AND v IE wsnp

g©

2014

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.

nature immunology volume 15 number 3 mArCH 2014 217

(TCR) makes more contacts with the self major histocompatibility complex (MHC) molecules than it does with the fragmented foreign peptide embedded within and pre-sented by the MHC. Moreover, even during an immune response, the vast majority of MHC molecules are loaded with self peptides rather than foreign peptides. Although the recogni-tion of complexes of self peptide and MHC (self peptide–MHC) was once thought to be relevant for T cells only during their develop-ment in the thymus, it is now clear that such interactions continuously affect mature T cells in the periphery. Hence, avidity of the TCR for self peptide–MHC is needed to sustain the survival of naive T cells under physiological conditions and to drive homeostatic prolif-eration under lymphopenic conditions1. More intriguingly, there is mounting evidence that the avidity of the TCR for self peptide–MHC

Unlike other cell types of the immune sys-tem, T cells have evolved to recognize

foreign antigens only in the context of self. The antigen-binding T cell antigen receptor

proposed to induce tyrosine phosphorylation of VE-cadherin and cleavage of its extracellular domain, which increasingly seems relevant to various inflammatory diseases12. It would then be very interesting to investigate the potential link between phosphorylation of VE-cadherin at Tyr685 and shedding of the protein ectodo-main. Wessel et al.1 have set a milestone on the way to full understanding of all the mechanisms that influence the functions of VE-cadherin in leukocyte transmigration and vascular perme-ability as two differently regulated events.

COMPETING FINANCIAL INTERESTSThe authors declare no competing financial interests.

1. Florian wessel, M.w. Nat. Immunol. 15, 223–230 (2014).

2. Potter, M.D., Barbero, s. & Cheresh, D.A. J. Biol. Chem. 280, 31906–31912 (2005).

3. Allingham, M.J., van Buul, J.D. & Burridge, K. J. Immunol. 179, 4053–4064 (2007).

4. Turowski, P. et al. J. Cell Sci. 121, 29–37 (2008).5. Timmerman, I. et al. Mol. Biol. Cell 23, 4212–4225

(2012).6. Nottebaum, A.F. et al. J. Exp. Med. 205, 2929–2945

(2008).7. Adam, A.P., sharenko, A.L., Pumiglia, K. & vincent,

P.A. J. Biol. Chem. 285, 7045–7055 (2010).8. wallez, Y. et al. Oncogene 26, 1067–1077 (2007).9. Monaghan-Benson, E. & Burridge, K. J. Biol. Chem.

284, 25602–25611 (2009).10. Orsenigo, F. et al. Nat. Commun. 3, 1208 (2012).11. Baumeister, U. et al. EMBO J. 24, 1686–1695

(2005).12. sidibé, A. et al. Arthritis Rheum. 64, 77–87 (2012).

phosphorylated at Tyr685 but instead depends on signaling by external permeability factors such as VEGF, TNF or histamine.

Nevertheless, Tyr685 contributes to VEGF-induced vascular permeability in veins. The permeability factors induce phosphorylation of VE-cadherin at Tyr685 but not at Tyr731, which confirms that these two tyrosine residues are clearly involved in different processes1. Tyr731 controls leukocyte extravasation and Tyr685 controls vascular permeability8. It is conceiv-able that throughout evolution, some tyrosine residues of VE-cadherin have specialized in transducing permeability-inducing signals, whereas other tyrosine residues have specialized in the transendothelial migration of leukocytes (Fig. 1). Moreover, phosphorylated Tyr685 of VE-cadherin has been proposed as a docking site for Csk, an inhibitory kinase that phosphorylates and deactivates Src and thereby inhibits endothe-lial cell proliferation11. It would be an important milestone to determine whether phosphorylation of Tyr685 contributes to keeping the opening of endothelial junctions under control through neg-ative-feedback regulation of Src activity by Csk. Further research is needed to elucidate in full the in vivo mechanisms involved in the dephosphory-lation and phosphorylation of Tyr731 and Tyr685. Furthermore, several permeability-increasing factors, such as TNF and bradykinin, have been

on endothelial permeability has been ques-tioned7. Studies of VE-cadherin point mutants with tyrosine-to-phenylalanine substitution of different residues have shown that tyrosine phosphorylation of VE-cadherin is not sufficient to induce vascular permeability, whereas activity of the tyrosine kinase Src is indeed required7. Furthermore, the specific VE-cadherin tyrosine residues phosphorylated upon stimulation with VEGF has been a matter of debate. Tryptic pep-tide mapping, after stimulation of HUVECs with VEGF, has shown phosphorylation of VE-cadherin exclusively at Tyr685 by Src kinase8. Studies using the commercial available phos-phorylation-specific antibodies have reported that VEGF additionally induces phosphoryla-tion of Tyr658 and Tyr731 of VE-cadherin9. The lack of specificity of those commercial antibodies clearly now raises questions about all those findings. Another study using newly developed antibodies has reported constitutive phosphorylation of VE-cadherin at Tyr685 and Tyr658 in mouse veins and has proposed that these sites are required for vascular permeability in vivo10. Wessel et al. have also raised specific antibodies to VE-cadherin phosphorylated at Tyr685 and have confirmed their specificity for this particular amino acid in the knock-in mice1. With these new tools, they clearly dem-onstrate that VE-cadherin is not constitutively

Self-gratification yields not-so-naive T cellsChristopher E Martin & Charles D Surh

Naive T cells differentiate into memory subsets upon exposure to their cognate foreign antigen. However, before such encounters, some naive T cells are ‘imprinted’ through their interactions with self targets.

Christopher e. Martin is with the Division of

Developmental Immunology, La Jolla Institute for

Allergy and Immunology, La Jolla, California, USA,

and the Kellogg School of Science and Technology

Doctoral Program in Chemical and Biological

Sciences, The Scripps Research Institute, La Jolla,

California, USA. Charles D. Surh is with the Academy

of Immunology and Microbiology, Institute for Basic

Science, Pohang, Republic of Korea, the Division

of Developmental Immunology, La Jolla Institute for

Allergy and Immunology, La Jolla, California, USA,

and the Department of Integrative Biosciences and

Biotechnology, Pohang University of Science and

Technology, Pohang, Republic of Korea.

e-mail: csurh@ibs.re.kr

has an important role in regulating the responses of naive T cells to foreign targets2,3. In this issue of Nature Immunology, Persaud et al. use the most direct system implemented so far in a comparative analysis of the effect that the avidity of the TCR–self peptide–MHC interaction has on the response of naive T cells to their cognate foreign antigen4.

Two current hypotheses address the impor-tance of the avidity of the TCR for self molecules during the response to foreign antigens. First, a model of ‘coagonism’ has been proposed that is based on the observation that the recognition of noncognate self peptide–MHC by the TCR enhances the recognition of agonist foreign peptide–MHC complexes by naive T cells5,6. Agonist foreign peptide–MHC complexes are typically present at low density, and noncognate self peptide–MHC ligands are thought to recruit additional TCRs and the coreceptors CD4 or

NE ws AND v IE wsnp

g©

2014

Nat

ure

Am

eric

a, In

c. A

ll rig

hts

rese

rved

.