R

Ap

Ma

b

a

ARAA

KTpMCE

1i

(uIrcslaietrTf

a

MAf

(

0h

Molecular Immunology 55 (2013) 117– 119

Contents lists available at SciVerse ScienceDirect

Molecular Immunology

jo u rn al hom epa ge: www.elsev ier .com/ locate /mol imm

eview

re membrane proteins favored over cytosolic proteins in TAP-independentrocessing pathways?�

argarita Del Vala,∗,1, Silvia Lázaroa,1, Manuel Ramosb, Luis C. Antóna

Centro de Biología Molecular Severo Ochoa, CSIC/Universidad Autónoma de Madrid, Madrid, SpainCentro Nacional de Microbiología, Instituto de Salud Carlos III, Majadahonda (Madrid), Spain

r t i c l e i n f o

rticle history:eceived 1 October 2012ccepted 8 October 2012vailable online 24 November 2012

a b s t r a c t

Recognition of infected or altered cells by CD8+ cytotoxic T lymphocytes is mediated by direct interactionof their T-cell receptor with peptides presented by MHC class I molecules. Peptides are transferred forassembly with newly synthesized MHC molecules by the transporters associated with antigen processing(TAP). Yet, a fraction of described epitopes are presented independently of TAP. Current belief is that

eywords:ransporters associated with antigenrocessingajor histocompatibility complex

D8+ cytotoxic T lymphocyte

most of them derive from membrane proteins, mostly from their signal sequences, and are processed byvesicular proteases. A thorough review of the published data may challenge some of these views.

© 2012 Elsevier Ltd. All rights reserved.

pitope

. Are membrane proteins favored over cytosolic proteinsn TAP-independent processing pathways?

Ever since transporters associated with antigen processingTAP) were identified over 20 years ago, there has been a reg-lar description of individual epitopes presented by MHC class

molecules and recognized by CD8+ T lymphocytes that do notequire TAP. Most of them are derived from viruses or from tumorells. Most of them derive from proteins that are targeted to theecretory route (78%, all shaded areas in Fig. 1A and B) and areocated in their membrane or luminal domains. It is thus gener-lly assumed that for MHC class I epitopes to bypass TAP, it is

mportant that the source protein bypasses TAP by entering thendoplasmic reticulum (ER). However, when large-scale mass spec-rometry analyses of MHC class I or class I-like ligands have beeneported, the vast majority of the ligands detected in uninfectedAP-deficient cells derive from cytosolic proteins (including a fewrom cytosolic parts of membrane proteins) (72%, white areas in

Abbreviations: ER, endoplasmic reticulum; TAP, transporters associated withntigen processing.� This article belongs to Special Issue on Antigen Processing and Presentation.∗ Corresponding author at: Unidad de Inmunología Viral, Centro de Biologíaolecular Severo Ochoa (CSIC-UAM), Nicolás Cabrera 1, Campus Universidad

utónoma de Madrid, Cantoblanco, 28049 Madrid, Spain. Tel.: +34 911 964 460;ax: +34 911 964 420.

E-mail addresses: mdval@cbm.uam.es (M. Del Val), silvialaz@cbm.uam.esS. Lázaro), mbuylla@isciii.es (M. Ramos), lanton@cbm.uam.es (L.C. Antón).

1 Both authors contributed equally to this work.

161-5890/$ – see front matter © 2012 Elsevier Ltd. All rights reserved.ttp://dx.doi.org/10.1016/j.molimm.2012.10.018

Fig. 1A and B) (reviewed in Del Val et al., 2011). The situation isfully reversed. There is no clear explanation for this discrepancy.

A possibility is that there is a basic difference in TAP-independent antigen processing between uninfected and virus-infected cells. For example, virus-infected cells that are veryactively expressing viral proteins or cells transfected and express-ing a large amount of a given protein may be under conditions ofenhanced folding and degradative stress (Schroder, 2008), unlikeuninfected cells. Overall distribution of membrane vs. cytosolic pro-teins is otherwise similar for cells and for large viruses and wouldnot account for this difference.

Another possibility is that MHC class I ligands identified by massspectrometry might not be representative of real epitopes recog-nized by CD8+ T lymphocytes. Some doubts are cast as to whethercytosolic peptides might come from post-lysis binding to MHC.On the other hand, although state-of-the-art mass spectrometrytechniques have achieved in recent years unprecedented levels ofsensitivity, these are still below the sensitivity achieved with liv-ing CD8+ cytotoxic T lymphocytes in functional assays. Thus, whilemass spectrometry detects the most abundant peptides, functionalstudies of individual epitopes are driven by the interest of theepitope in protective immunity or by the availability of specificlymphocytes, which in turns depends on several technical as wellas physiological factors including the complexity of immunodom-inance.

Yet another possibility is that researchers in the field, includingourselves, might have been biased by the widely assumed precon-ceived idea that, in the absence of TAP, the cell machinery dealsmore easily with antigen processing of membrane proteins that

118 M. Del Val et al. / Molecular Immunology 55 (2013) 117– 119

6

126

5

31

6

3

10

15

LuminalTransm embraneSignal sequen ceCytosol

TAP-in depend entepitopes or liga nds: loca�on within source protein:

Individual epitopes

Mass spec ligands

A (n )

1

4

2

3

4

Proteasome No pr oteasomeIndividual epitopes

C (n)

22% 72%

19%

18%

22%

2%37% 9%

Mass spec ligands

Individual epitopes

B(%)

11%: HLA -A2

14%: HLA -A2

22%: EBV LMP2

Fig. 1. TAP-independent MHC class I ligands. Compilation of numbers (A) and % (B) of published TAP-independent MHC class I epitopes analyzed individually in functionalassays with specific CD8+ cytotoxic T lymphocytes, mostly in virus-infected or transfected cells (‘individual epitopes’) and of published MHC class I and class I-like ligandsdetected by mass spectrometry large-scale analyses of TAP-deficient uninfected cells (‘mass spec ligands’), broken down according to the epitope/ligand localization withinthe protein it derives from, as indicated. Accordingly, the three shaded sectors include all epitopes and ligands derived from membrane proteins (including membraneanchored, luminal and secreted proteins). The white sectors include epitopes and ligands derived from cytosolic proteins and a few derived from cytosolic parts of membraneproteins. Insets within sectors in B indicate the % of transmembrane-derived epitopes derived from a single protein, Epstein-Barr virus LMP2, and the % of signal-sequence-derived epitopes or ligands presented by a single MHC class I molecule, HLA-A2. (C) For ‘individual epitopes’, the numbers of epitopes that require proteasomes for antigenprocessing (‘proteasome’) or that are resistant to proteasome inhibitors (‘no proteasome’) are separately compiled. Numbers and percentages are indicated next to eachsector. See Supplemental information for reports included in the compilation.

mpsipmfdid

cascT

2s

ifpSHobeeps(a

ake it into the ER, where TAP is located, or into later vesicular com-artments. Indeed, this idea is understandable, as it just requiresuitable proteases in vesicular pathways, many of which are knownn cell biology. In contrast, in order to explain TAP-independentresentation of cytosolic proteins new forms of transport acrossembranes are needed, and these have not been described yet

or antigen processing pathways. This preconceived idea may haveriven the selection of epitopes for which the possibility of TAP-

ndependent presentation is raised and studied towards thoseerived from membrane proteins.

The large fraction of TAP-independent ligands derived fromytosolic proteins has an interesting implication: TAP-independentntigen processing pathways may contribute to sampling the cyto-ol to levels previously unrecognized, and thus more globallyontribute to immunosurveillance of altered cells by CD8+ cytotoxic

lymphocytes.

. Are TAP-independent epitopes mainly derived fromignal sequences?

Furthermore, it is generally accepted that most TAP-ndependent epitopes derive from signal sequences. This stemsrom the first observation of signal sequence derived peptidesresented by HLA-A2 in TAP-deficient cells (included in Fig. 1).ignal sequences have a hydrophobic portion, which fits with theLA-A2 preference for binding more hydrophobic peptides thanther MHC class I allotypes. Indeed, signal sequences presentedy HLA-A2 dominate the landscape of signal sequence derivedpitopes and ligands. From the five TAP-independent individualpitopes derived from signal sequences (Fig. 1A and B), 3 are

resented by HLA-A2, and from 31 signal-sequence derived masspectrometry detected TAP-independent MHC class I ligandsFig. 1A and B), as many as 24 are presented by HLA-A2. Thus,part from HLA-A2 and maybe HLA-B51, most TAP-independent

MHC class I epitopes and ligands that derive from membraneproteins do not originate from signal sequences (Fig. 1A and B).Most derive from their luminal domains and a further few derivefrom transmembrane domains. This distribution (luminal moreabundant than signal sequence and transmembrane) for TAP-independent MHC class I ligands for non-HLA-A2 allotypes maymerely reflect the fact that luminal ectodomains constitute thelarger part of most membrane proteins. Membrane proteins withmultiple transmembrane regions and short cytosolic or luminalportions are the exception; they represent a minor fraction ofthe total but, interestingly, one of them, Epstein–Barr virus LMP2protein, accounts for all 6 transmembrane-region-derived epitopesidentified so far in functional assays (Fig. 1A and B).

It is also believed that the critical generation of the exact car-boxyl terminus of TAP-independent peptides is often performedby the signal peptidase protease itself. Again, data with HLA-A2are behind this generalization. HLA-A2 is the most abundant cau-casoid MHC class I allotype and has a particular propensity toeasily accommodate alanine in the carboxy termini of peptides.Alanine is a frequent carboxy-terminal residue in cleaved sig-nal sequences. Indeed, when TAP-independent mass spectrometrydetected ligands derived from signal sequences are analyzed, thecarboxy-termini of most (15) of those presented by HLA-A2 (24)coincide with the cleavage site of signal peptidase. However, somepresented by HLA-A2 (9) as well as all 7 presented by other MHCclass I and MHC class I-like molecules derive from internal parts ofthe signal sequences, requiring both amino- and carboxy-terminaltrimming (reviewed in Del Val et al., 2011). Relying on signal pep-tidase for generation of correct carboxyl termini in the ER wasalso supported by two facts: the accepted prominent role of pro-

teasomes in generating carboxyl termini in the cytosol, and thelack of evidence for significant carboxypeptidase activity in theER, until the recent description of the relevant role of angiotensin-converting enzyme (Shen et al., 2011).

Immu

3p

ottmebfbTtaTtpTTiTnptd

ba(

phmmsonbpmpoowd

pItdb

4

Iiwss

M. Del Val et al. / Molecular

. Are TAP-independent epitopes mainly processed byroteases other than the proteasome?

Finally, there is another widely extended assumption in the fieldf antigen processing, with a rationale that is similar to that behindhe preference for studying membrane proteins. Presuming that inhe absence of TAP there is no efficient route for degradation inter-

ediates to gain access to the ER lumen, it is assumed that, when anpitope is presented independently of TAP, it will also be processedy proteases other than the proteasome, presumably proteasesrom the secretory route. Both circumstances are recognized toe of rare occurrence, with the classical proteasome-dependent,AP-dependent antigen processing pathway accepted to be by farhe one employed by a cell to produce most of its MHC class I lig-nds. This idea has led to assume a fully classical pathway for anyAP-dependent epitope – sometimes even without testing for pro-easome involvement. It has also led to assume a fully alternative,roteasome-independent pathway, for TAP-independent epitopes.hus, for cytosolic proteins that are presented independently ofAP, even when they have access to the cytosolic proteasome, thenvolvement of the proteasome is rarely tested, assuming that ifAP is not involved, the proteasome will also not be involved. So far,o physical biochemical linkage has been demonstrated betweenroteasomes and TAP; therefore, we reason that there is no basiso assume that what bypasses TAP cannot come from proteasomeigestion.

Conversely, the possibility that TAP-dependent epitopes cane processed by proteases other than the proteasome has beenlready accepted in the field and gained momentum along the yearsreviewed in van Endert, 2011).

As pointed out above, the role of proteases other than theroteasome in antigen processing of TAP-independent peptidesas been analyzed almost exclusively for those that derive fromembrane proteins (Fig. 1C). Results confirm the hypothesis:ost tested TAP-independent epitopes are resistant to protea-

ome inhibitors (termed ‘no proteasome’ in Fig. 1C). For only 3f them, positive involvement of vesicular proteases such as sig-al peptidase, signal peptide peptidase, furin or cathepsins haseen demonstrated. Interestingly, the two TAP-independent butroteasome-dependent epitopes described derive from the trans-embrane domain of the thoroughly studied Epstein–Barr virus

olytopic membrane protein LMP2. We propose that we shouldpen our minds and test more often the possibility of pathwaysf antigen processing mediated by the cytosolic proteasome buthere the products gain access to MHC class I molecules indepen-ently of TAP.

This preconception has also impacted the field of cross-resentation, that is, presentation of external antigens by MHC class

molecules in professional antigen presenting cells. Yet, recently,he dissociation between proteasomes and TAP has been clearlyescribed, by identifying a route where proteasomes are involvedut TAP is not (Merzougui et al., 2011).

. Future perspectives

A review of the published data on TAP-independent MHC class epitopes and ligands reveals strikingly different results depend-

ng on the approaches used. Data gathered from functional assays

ith individual epitopes hint at membrane proteins as the mainource of TAP-independent epitopes, while large-scale high sen-itivity mass spectrometry studies of MHC class I or class I-like

nology 55 (2013) 117– 119 119

ligands disclose a major contribution of cytosolic proteins to therepertoire of TAP-independent ligands. At this stage, we advocatefor global functional analyses of large collections of epitopes to gaininsight into the nature of TAP-independent presentation to CD8+ Tlymphocytes. Indeed, this is now amenable to study for viruses,as catalogs of epitopes have now been published for large viruses,including epitopes derived from viral source proteins located in alltypes of cellular compartments. Similar functional studies shouldalso be feasible with the large catalogs of abundant cellular MHCclass I ligands identified by mass spectrometry and would dispeldoubts on the origin of cytosolic peptides.

In addition, further work on identifying routes that allow theprocessing intermediates to access vesicular lumen as well as pro-teases involved in antigen processing of TAP-independent epitopesis clearly needed. The fact that peptides derived from luminaland transmembrane regions, and even from signal sequences ofmembrane proteins do require both amino- and carboxy-terminaltrimming should encourage the search for and the functionaldemonstration of the role of endo- or carboxy-proteases ingenerating the correct carboxyl end of individual epitopes. For TAP-independent epitopes, it may seem that the proteasome may loseits leading role as generator of carboxyl termini. However, moreattention should be paid to its role in TAP-independent presen-tation of cytosolic epitopes. In addition, the data indicating thatmany of these epitopes derived from membrane proteins are resis-tant to proteasome inhibitors should be interpreted with caution;while suggesting that the proteasome is not a major player in theirprocessing, these data do not fully discard a minor role for it, whichmerits exploration. In summary, future work on identifying pro-teases involved in TAP-independent presentation should not onlybe concentrated on vesicular proteases.

Acknowledgements

We thank Salvador Iborra, Daniel López and Dolores Jaraque-mada for fruitful discussions and Noel García-Medel for proteomedata. Work in the laboratory is supported by Spanish Ministerio deCiencia e Innovación and by Red Temática de Investigación Cooper-ativa en SIDA from Instituto de Salud Carlos III. S.L. was supportedby Formación de Personal Investigador program from Ministeriode Ciencia e Innovación. The authors declare that they have noconflicting financial interests.

Appendix A. Supplementary data

Supplementary data associated with this article can be found,in the online version, at http://dx.doi.org/10.1016/j.molimm.2012.10.018.

References

Del Val, M., Iborra, S., Ramos, M., Lázaro, S., 2011. Generation of MHC class I ligandsin the secretory and vesicular pathways. Cellular and Molecular Life Sciences 68,1543–1552.

Merzougui, N., Kratzer, R., Saveanu, L., van Endert, P., 2011. A proteasome-dependent, TAP-independent pathway for cross-presentation of phagocytosedantigen. EMBO Reports 12, 1257–1264.

Schroder, M., 2008. Endoplasmic reticulum stress responses. Cellular and MolecularLife Sciences 65, 862–894.

Shen, X.Z., Billet, S., Lin, C., Okwan-Duodu, D., Chen, X., Lukacher, A.E., Bernstein,K.E., 2011. The carboxypeptidase ACE shapes the MHC class I peptide repertoire.Nature Immunology 12, 1078–1085.

van Endert, P., 2011. Post-proteasomal and proteasome-independent generation ofMHC class I ligands. Cellular and Molecular Life Sciences 68, 1553–1567.