Review T -cell immune responses in the brain and their ...

Brain Research Reviews 42 (2003) 97–122www.elsevier.com/ locate/brainresrev

Review

T -cell immune responses in the brain and their relevance for cerebralmalignancies

a , a b,c a*Paul R. Walker , Thomas Calzascia , Nicolas de Tribolet , Pierre-Yves DietrichaLaboratory of Tumour Immunology, Division of Oncology, Geneva University Hospital, 24 rue Micheli-du-Crest, 1211 Geneva 14, Switzerland

bDepartment of Neurosurgery, Lausanne University Hospital, Laussane, SwitzerlandcDepartment of Neurosurgery, Geneva University Hospital, Geneva, Switzerland

Accepted 12 December 2002

Abstract

In order that cellular immune responses afford protection without risk to sensitive normal tissue, they must be adapted to individualtissues of the body. Nowhere is this more critical than for the brain, where various passive and active mechanisms maintain a state ofimmune privilege that can limit high magnitude immune responses. Nevertheless, it is now clear that immune responses are induced toantigens in the brain, including those expressed by cerebral malignancies. We discuss hypotheses of how this can occur, although detailssuch as which antigen presenting cells are involved remain to be clarified. Antitumor responses induced spontaneously are insufficient toeradicate malignant astrocytomas; many studies suggest that this can be explained by a combination of low level immune responseinduction and tumor mediated immunosuppression. A clinical objective currently pursued is to use immunotherapy to ameliorateantitumour immunity. This will necessitate a high level immune response to ensure sufficient effector cells reach the tumor bed, focusedcytotoxicity to eradicate malignant cells with little collateral damage to critical normal cells, and minimal inflammation. To achieve theseaims, priority should be given to identifying more target antigens in astrocytoma and defining those cells present in the brain parenchymathat are essential to maintain antitumour effector function without exacerbating inflammation. If we are armed with better understandingof immune interactions with brain tumor cells, we can realistically envisage that immunotherapy will one day offer hope to patients withcurrently untreatable neoplastic diseases of the CNS. 2003 Elsevier Science B.V. All rights reserved.

Theme: Disorders of the nervous system

Topic: Neuro-oncology

Keywords: Neuro-immune interactions; Immune privilege; Malignant astrocytoma; Brain infiltrating lymphocytes; T cells; Immunosuppression;Immunotherapy

Contents

1 . Initiation of immune responses: key mechanisms and those applicable to the brain....................................................................................... 981 .1. Immune recognition of antigens ....................................................................................................................................................... 981 .2. Modified immune responses in the ‘immune privileged’ brain ............................................................................................................ 991 .3. Initiation of primary immune responses ............................................................................................................................................ 1001 .4. Antigen capture............................................................................................................................................................................... 1011 .5. Antigen capture in the brain and initiation of CNS immune responses ................................................................................................. 102

1 .5.1. Hypothesis A......................................................................................................................................................................... 1021 .5.2. Hypothesis B......................................................................................................................................................................... 102

Abbreviations: APC, antigen presenting cell; BBB, blood–brain barrier; CDR, complementarity determining region; CNS, central nervous system; CSF,cerebrospinal fluid; CTL, cytotoxic T lymphocyte; DC, dendritic cell; EAE, experimental autoimmune encephalomyelitis; FasL, Fas ligand; ISF, interstitialtissue fluid; MHC, major histocompatibility complex; TCR, T-cell receptor; TGF, transforming growth factor

*Corresponding author. Tel.:141-22-372-9854; fax:141-22-372-9858.E-mail address: [email protected](P.R. Walker).

0165-0173/03/$ – see front matter 2003 Elsevier Science B.V. All rights reserved.doi:10.1016/S0165-0173(03)00141-3

mailto:[email protected]

98 P.R. Walker et al. / Brain Research Reviews 42 (2003) 97–122

1 .5.3. Hypothesis C......................................................................................................................................................................... 1042 . Effector stages of T-cell immune responses ............................................................................................................................................... 104

2 .1. Entry of primed T cells into the CNS................................................................................................................................................ 1042 .2. Molecules involved in transmigration ............................................................................................................................................... 105

1 12 .3. Antigen-specific restimulation in the brain: are specialized APCs necessary for CD8 as well as CD4 T cells?.................................... 1052 .4. APC candidates in the brain ............................................................................................................................................................. 107

3 . T-cell immune responses to brain tumors in clinical and experimental situations .......................................................................................... 1083 .1. T-cell immune response against human malignant astrocytoma........................................................................................................... 1083 .2. Immune escape of astrocytomas ....................................................................................................................................................... 109

3 .2.1. Immunosuppression by soluble factors .................................................................................................................................... 1103 .2.2. Immunosuppression by cell-mediated interactions .................................................................................................................... 110

3 .3. Immunotherapy—from rodent models to the clinic ............................................................................................................................ 1103 .3.1. Adoptive immunotherapy ....................................................................................................................................................... 1113 .3.2. Active vaccination ................................................................................................................................................................. 1123 .3.3. Immunomodulation with cytokines ......................................................................................................................................... 1133 .3.4. Compatibility and synergy of antitumor immune responses with other therapies ......................................................................... 114

4 . Concluding remarks................................................................................................................................................................................. 114Acknowledgements ...................................................................................................................................................................................... 115References................................................................................................................................................................................................... 115

1 . Initiation of immune responses: key mechanisms microglial cells, the macrophages of the brain. However,and those applicable to the brain there are severe limitations to immune recognition by

germ-line encoded receptors. Microbial pathogens evolve1 .1. Immune recognition of antigens and mutate at a very high rate, generating an extraordinary

number of variants that would soon exhaust the capacity ofThe role of the immune system is to maintain the the host genome to encode a sufficient number of specific

functional integrity of the host in the face of biological receptors to detect all microbial pathogens. Thus, althoughthreats arriving from the exterior (pathogenic microbes), or details of new receptors are regularly being reported, theirfrom within (neoplastic changes). It is clear that the number is probably in the region of a few hundred.necessity for such a defense system is organism-wide, at Furthermore, innate immune responses are essentially theleast for non-expendable, non-self renewing tissues and same at each re-exposure to infection; there is no specificorgans such as the brain. The only caveat that may be immunological memory.raised is if there is an efficient non-immunological defense A complementary solution to immune recognitionmechanism. For the CNS, the physical protection afforded evolved in the jawed vertebrates: lymphocyte-based adap-by the skull and a certain physical isolation provided by tive immune responses. A large number of T and Bthe endothelial /cellular barriers of the brain may raise the lymphocytes are generated in each individual, each bearingthreshold for the necessity of immune intervention. How- clonally distributed antigen receptors that are structurallyever, the substantial list of cerebral malignancies and related and are assembled following similar mechanisms ofinfections with pathogenic neurotropic viruses illustrates gene rearrangement. For both T and B cells, tolerance tothat the CNS is far from totally protected. self antigens is maintained by the elimination or regulation

The original conceptual framework for understanding of autoreactive cells at different stages of their existencethe operation of the immune system was that it functioned [175]. B cells can directly recognize free, native antigen,by the discrimination of self from non-self. We are whereas T-cell antigen recognition is more complex,apparently ‘hard-wired’ with an innate immune system that requiring an antigen presenting cell (APC). Within anydetects certain basic patterns that are highly conserved given individual of a species, APCs display a selection ofamongst pathogenic organisms, and affords some immedi- major histocompatibility complex (MHC) molecules thatate protection against infection by many microbes. The signal to the developing T cells a complex molecularinnate immune system includes complement components signature corresponding to ‘self’, which normally stimu-found in plasma, but in the interstitial tissue fluid (ISF) lates no immune response, but which serves as a referenceand cerebrospinal fluid (CSF) of the CNS, concentrations point for the same lymphocytes when later there isare very much lower. Cell-mediated innate immune re- exposure to ‘non-self’ in the form of inevitable antigenicsponses include phagocytes bearing toll like receptors challenge by microbes. MHC molecules fulfil this role by(TLR) that recognize conserved microbial-associated prod- sampling degraded proteins present within a cell anducts such as lipoteichoic acid (on Gram-positive bacteria), presenting peptides derived therefrom at the cell surface tomannans (on fungi) and lipopolysaccharide (on Gram- T cells [88]. In the case of MHC class I molecules, whichnegative bacteria) [152]. In the CNS, the resident cells are expressed by most somatic cells (although notably,capable of mediating cellular natural immunity are the constitutive MHC expression by many cells of the brain is

P.R. Walker et al. / Brain Research Reviews 42 (2003) 97–122 99

very low), these peptides are generally 8–10 amino acids lymphocytes. In contrast, the proteins that furnish thein length and are contained in a groove in the MHC peptides that are assembled with MHC class II moleculesmolecule that is closed at each extremity. The resulting typically arise from the extracellular milieu, for exampleMHC/peptide complex serves as the ligand for the T-cell bacterial antigens from strains that do not parasitize thereceptor (TCR) of CD8 T cells. For MHC class II intracellular compartment [131]. Despite this apparentmolecules, expression is normally limited to hematopoietic dichotomy in the handling of endogenous and exogenouscells and thymic epithelial cells, although class II may be antigens, it has become apparent in recent years that thereinduced on other cell types by interferon-g. The groove of are many exceptions to the general rule. For example,MHC class II molecules permits binding of somewhat apoptotic cells can be phagocytosed by dendritic cellslonger peptides than for MHC class I (most are 13–17 (DCs), leading to presentation of antigens derived there-amino acids), since the extremities of the peptide are not from on MHC class I molecules, a mechanism known asanchored. The MHC class II /peptide complex is the ligand cross-presentation [6].

1for the TCR of CD4 T cells.Because only few cells with a given fine specificity are 1 .2. Modified immune responses in the ‘ immune

present in an individual before antigen exposure, to privileged’ brainachieve sufficient specific cells for meaningful biologicalfunction, lymphocytes must be appropriately stimulated to Certain observations of immune responses in the CNSundergo clonal expansion. This takes a few days (during have led to the designation of the brain parenchyma as anwhich time the innate immune response may be containing immune privileged site. This terminology arises fromthe pathogen), but the result is an army of effector cells transplantation experiments in which there is extendedexpressing antigen receptors of identical specificity that survival of tissues transplanted to the CNS, compared withmay be able to eliminate the pathogen. A proportion of this their survival in other sites [151,28]. In fact, rather than anexpanded clone of lymphocytes is also committed to aberration, it can be postulated that the risks associatedbecoming memory cells, available for rapid response to with allowing a full-blown immune response to developfuture exposure to the same antigen, even many years later. are too severe to permit in such a critical site as the brain.In contrast, most of the effector cells are eliminated by Uncontrolled inflammation would rapidly lead to a severeapoptosis after elimination of the primary source of antigen augmentation of intracranial pressure due to the volume[237,227]. limitations within the confines of the skull. This in turn

A unique, genomically economical solution permits the would compromise neurologic function and augment deathadaptive immune system to generate a sufficiently large of critical cells that have rather limited capacity fornumber of TCRs to bind virtually any conceivable antigen, renewal. To limit the possibilities of such a scenario, thewithout dedicating an unreasonable proportion of its brain is relatively well protected from physical injury bygenome to the task. This is achieved by somatic re- the thick skull and a degree of cushioning afforded by thearrangement of a limited number of gene segments, a CSF. Moreover, passive entry of many pathogenic organ-process also employed in the assembly of the B-cell isms present in the blood is limited by the presence of thereceptor for antigen, the immunoglobulin molecule. All tight junctions present between the endothelial cells of theTCRs are dimers, comprised mainly of ana and ab chain, cerebral vasculature, the so-called blood–brain barrierbut a minority population of T cells exists that expresses a (BBB) [41,198]. Apart from tight junctions with extremelyg and ad chain. Onlyab TCR T cells will be discussed in high electrical resistance that limit paracellular diffusion,detail here, these cells recognize short peptides bound to the BBB has many other special features not present inMHC molecules as described above, whereasgd T cells microvessels from other organs. The endothelial cells lackmay have other ligands than classical MHC/peptide pinocytotic vesicles that limit transcellular flux, theycomplexes. The matureab TCR is non-covalently bound express distinct cytoplasmic enzymes, and they possessto a cluster of invariant proteins, termed CD3, that are polarized specific transport systems for import of metabo-responsible for signal transduction. In addition, the TCR is lites and export of toxic waste from the brain. Whilst theassociated with either a CD4 or a CD8 co-receptor, endothelial cell barrier is the ‘front-line’ of the BBB, itsdictating MHC class II or MHC class I restriction, integrity is maintained by other cells that intimatelyrespectively. contact its abluminal surface. Astrocytes are particularly

Peptides that bind to MHC class I and class II molecules implicated, they almost totally surround the vessel withare generated by different pathways. For MHC class I their foot processes [119], but pericytes and perivascularbinding peptides, they are most commonly considered to cells may also be important. The overall consequence ofbe of endogenous origin, i.e. self proteins and those the presence of the BBB is that access to the brainencoded by viral genes or by genes mutated or over- parenchyma is limited for many components of the im-expressed following neoplastic transformation. Thus, MHC mune system. Thus, concentrations of immunoglobulinsclass I expressing infected cells or tumor cells may and complement components are significantly lower than

1potentially be detected and killed by CD8 cytotoxic T those found in plasma. Immune cell trafficking, whilst not


excluded, is more limited than for other sites [105]. 1 .3. Initiation of primary immune responsesFurthermore, a lack of organized lymphatic drainage andlow MHC expression in the brain parenchyma are factors There have been enormous advances over the past fewthat contribute to the regulation of immune responses, but years in our understanding of how T-cell immune re-not for their abrogation. Such considerations should lead sponses are initiated in vivo. This necessitates subtleus to understand immune privilege in the CNS as a relative choreography involving several cell types and molecules,rather than an absolute term. In fact the choice of with interactions occurring at different anatomical sites.terminology to describe the nature of CNS immune These fundamental aspects need to be discussed before theresponses should certainly not be fixed—in terms of particularities of the CNS are considered.detecting spontaneous immune responses, we are now The sooner that T lymphocytes are alerted to themore adept at assessing them than we used to be. And as presence of an antigen, the better the protection they canfar as immunization is concerned, we now understand afford the host. Most antigens derived from pathogens ormuch more about the basic rules of immune responses, and from newly formed malignancies will be present onlywe can induce responses under conditions where this was locally at early stages. The challenge to the immunenot previously possible. Thus, we are currently at an system is how to orchestrate an encounter between a rareexciting moment for neuroimmunology. Firstly we can specific T cell (calculated to be around 1/100 000 naive Tincorporate many of the enormous conceptual advances in cells for certain immune responses: [124]) and an appro-general immunology to better understand immune re- priately presented antigen. Considering the T cells in thesponses in the CNS. And secondly, we can hope to profit first instance, before stimulation by antigen (i.e. naive Tfrom recently developed procedures and technology to cells) they are in constant recirculation between blood andmanipulate immune responses if we make appropriate secondary lymphoid organs (lymph nodes, tonsils, Peyer’smodifications for the particularities of the brain. patches and the spleen). Naive T cells are not generally

An anatomical consideration must be made for the found in healthy non-lymphoid tissues and are essentiallydiscussions that will be developed in this review. The absent from normal brain, as detected by immunohisto-‘brain’ and the ‘CNS’ are clearly not a homogenous chemistry [102]. More recent observations have challengedanatomical unit. The choroid plexus, the ventricles, the this assumption, although the results are still consistentmeninges and the CSF have more direct interactions with with a preferential trafficking of activated T cells throughthe systemic immune system than the brain parenchyma, the CNS [39]. In the inflamed brain, there may be BBBarguably the ‘most privileged’ part of the CNS. One disruption as a result of the relaxation of tight junctionsconsequence is that in the former sites, immune responses and upregulation of adhesion molecules and chemokinesare similar to those found elsewhere in the body, whereas [49,122]. This does not result in unrestricted leakage ofthe parenchyma may be considered to have rather subtle, blood cells into the brain (for example, erythrocytes arelow level interactions with the systemic immune system. rarely found in the immune infiltrate). However, naive

1Immune responses are the result of complex interactions antigen inexperienced CD4 T cells are detected in theinvolving many cell types in specialized microenviron- brain parenchyma under these conditions [133]. Thements. Nevertheless, many advances in our understanding possibility that such naive T cells may be present in theof immunological concepts have evolved from somewhat brain emphasizes the importance of understanding antigenreductionist studies on individual cell lineages and mole- presentation in the CNS, but to date our understanding ofcules. The discussions that follow are limited to T-cell naive T-cell activation is best understood in the secondary

1immune responses, with a particular slant towards CD8 T lymphoid organs, particularly the lymph nodes and Peyer’scells. This is principally due to the demonstrations in many patches, into which T cells enter via the high endothelialanimal models and now, in various tumor immuno- venules. These specialized vessels express adhesion mole-

1therapies under clinical trial, that CD8 T cells are the cules termed addressins that engage withL-selectin andprincipal antitumor effector cell. Indeed, responses that are integrin molecules expressed by naive T cells. It is withincell-mediated rather than humoral responses may be of the specialized architecture of lymphoid organs that naivegreater significance for immune effector function in the T cells have the opportunity to encounter their specificintact CNS, because of the possibility of T cells to infiltrate antigen on an APC (typically a DC). Encounters between

1brain parenchyma. As far as the focus on CD8 T cells is naive T cells and mature DCs in the paracortical area areconcerned, these cells have the potential for the most facilitated by chemokine gradients [97,165,48], and by aexquisite fine specificity and a mechanism of action that is specialized extracellular matrix that encourages T-cellgenerally mediated by direct cell contact, without neces- migration and serial sampling of DCs [98]. At thissarily involving bystander killing of antigen-negative moment, there are two possible outcomes. If there is notargets. Furthermore, for direct cell-mediated effector encounter with specific antigen, T cells will leave via thefunction, there are a significantly wider number of po- efferent lymph to re-enter the blood at the thoracic duct.tential target cells in the brain that express MHC class I Alternatively, antigen specific activation occurs.molecules than those that express MHC class II molecules. Naive T-cell activation has particularly stringent require-


ments. If only the minimum ligand for antigen recognition IL-4 secretion [81], whilst 4-1BB ligand may preferentially1is detected, i.e. the correct peptide and MHC combination, stimulate CD8 T cells and interferon-g secretion

this will induce signaling in naive T cells that results in [130,222].anergy or apoptosis. A second co-stimulatory signal isrequired to elicit full effector function of the T cell. The 1 .4. Antigen capturebest defined costimulation pathway is that mediated byCD80 and CD86 (B7.1 and B7.2) that interact with CD28 Since the initial source of an antigen challenge will onlyon T cells, however the same B7 molecules can also give rarely be directly in lymphoid organs, the transport andrise to negative signals by binding to CTLA-4 [223]. More localization of these antigens must be considered. Therecently other costimulatory molecules and receptors have spleen is a somewhat special case, since its architecturebeen identified, which may give different qualitative or enables a direct sampling of blood-born antigens that canquantitative outcomes to T-cell activation [1,111]. Based then be taken up directly by resident DCs. For otheron in vitro and in vivo data, mature DCs appear to be lymphoid tissue, antigenic material draining from tissuesideally equipped to prime naive T cells (Fig. 1). They accumulates in the lymph and is transported to the lymphsecrete chemokines (MIP-3b and 6Ckine) to attract naive nodes via the afferent lymphatics. However the route thatT cells [48,66,206] which are then encouraged to form is now becoming more clearly defined is the activeclusters with DCs through interactions between adhesion transportation of antigenic material by DCs [25]. Certainmolecules—integrins (b1 and b2) and immunoglobulin DCs are resident in normal tissues such as the Langerhans’superfamily members (CD2, CD50, CD54 and CD58) cells of the skin. Other DCs or their precursors may be[29,101]. These interactions may occur in specialized recruited once an inflammatory response occurs. In theirmembrane domains referred to as ‘rafts’, which lead to the resting, or so-called ‘immature’ form, DCs continuouslyformation of an ‘immunological synapse’ between the T sample the local environment by endocytosing micro-cell and the APC [95]. DCs then provide strong antigen- organisms, dead cells and cellular debris. However, know-specific stimulation by virtue of a high density of MHC- ing the recirculation patterns of naive T cells, the likeli-peptide complexes at their surface, 10- to 100-fold higher hood of them stimulating a naive T cell of the rightthan on B cells and monocytes [114]. Mature DCs are specificity directly in peripheral tissue is rare. If theequipped to deliver a second signal to T cells by expres- immature DCs receive signals that may be interpreted assion of a range of co-stimulatory molecules, of which ‘danger signals’, they undergo maturation, a series ofCD86 is probably the most important. However, OX40 events that triggers various phenotypic and functionalligand and 4-1BB ligand are also expressed, with the changes that encourage migration via the afferent lymph to

1former being particularly important for CD4 T cells and the T-cell areas of lymph nodes. Maturation signals

1 1Fig. 1. Activation of naive T cells by dendritic cells. Antigen-specific activation of CD4 or CD8 T cells occurs optimally in secondary lymphoid tissuewhere naive T cells traffic and follow chemokine gradients to interact with dendritic cells. Efficient presentation of MHC class I /peptide complexes to

1 1CD8 T cells or MHC class II /peptide complexes to CD4 T cells is achieved by mature dendritic cells that express high levels of MHC molecules,adhesion molecules and costimulatory molecules.


include pathogen-associated molecular patterns on the undergo a unique differentiation program depending uponsurface of microorganisms, and endogenous factors such as the degree of infiltration into the brain parenchyma, thatinterferon-g released from virally infected cells and heat results in the phenotype of a perivascular or parenchymalshock proteins liberated as a result of necrotic cell death. microglial cell. Indeed, the functional capacities of mi-Maturation not only allows the DC to migrate to lymphoid croglia residing deep in the parenchyma are restricted, withorgans, it also diminishes the phagocytic and endocytic much gene expression turned off [181]. Whether alter-capacities of the DC, whilst upregulating its costimulatory native differentiation pathways exist in the CNS is un-capacity. The final stages of DC activation are only known. However, in vitro studies of cells derived from

1completed after interaction with CD4 T cells, in which new-born mouse brain and cultured with different cyto-1activation signals work in both directions, with the CD4 kines including GM-CSF, M-CSF, and in some cases in

cell activating the DC via CD40–CD40L interactions as co-culture with astrocytes, demonstrated that cells withwell as it becoming activated itself. The mature ‘con- certain similarities to DCs could be cultured [11,209,188],ditioned’ DC is then in an appropriate activation state to but only limited functional and phenotypic analyses were

1prime CD8 T cells [137,194,214,32]. carried out. In vivo, DC like cells have been described inthe brain parenchyma of mice suffering from experimental

1 .5. Antigen capture in the brain and initiation of CNS autoimmune encephalomyelitis (EAE) or from thoseimmune responses parasitized byToxoplasma gondii [234,80]. In the latter

case, functional studies were also carried out in whichIn the light of the usual recirculation patterns of naive T brain-derived DCs efficiently triggered antigen-specific and

cells and the optimal microenvironment for T-cell priming primary allogeneic T-cell responses and secreted IL-12.in secondary lymphoid organs, it can be reasonablyproposed that the initiation of T-cell immune responses to 1 .5.2. Hypothesis Bbrain-derived antigens will occur in such sites. However, An alternative possibility for T-cell priming is thatthe highly particular cellular composition of the normal APCs do not transport antigen to lymphoid organs, butbrain parenchyma requires consideration of hypotheses that there is drainage of soluble or particulate antigenic materi-take into account these features (Fig. 2). al. This has been demonstrated in several studies, princi-

pally using radiolabeled materials (reviewed by Cserr and1 .5.1. Hypothesis A Knopf [57]). Antigens present in the brain parenchyma are

Here, we can consider that T-cell priming will occur as predicted to drain with the flow of ISF. This is produced asin other sites, by uptake of antigen by immature DCs, then a filtrate of plasma across the capillary endothelium, itsubsequent DC maturation and migration to lymph nodes. bathes the cells of the brain parenchyma, picking upHowever, there is an absence of resident DCs in the antigen material, then finds the route of least resistance—normal brain parenchyma, but in an ongoing inflammatory probably the perivascular spaces, to eventually drain eitherresponse, DCs have been detected [234], although their into the subarachnoid space with the CSF, then to venousmaturation status was not assessed. If these cells derive blood via arachnoid granulations, or to lymphatics viafrom DC precursors, recruited only after substantial in- arachnoid sheaths of certain cranial nerves and spinalflammatory responses have been initiated and sustained nerve roots to finally reach the cervical lymph nodes. Insolely by innate immune mechanisms, it would be ex- support of the proposed drainage patterns of potentiallypected that T-cell responses would be significantly de- antigenic material with the ISF, high molecular weightlayed. There is evidence for and against such kinetics. radiolabeled substances infused into the brain parenchymaDelayed immune responses in the brain compared with preferentially accumulate in the deep cervical lymphother sites were reported in early transplant studies nodes. In contrast, lower molecular weight substances are[151,28], although the details of the mechanism respon- lost to the blood, presumably due to less restricted passagesible for eventual graft rejection were not studied. In through vascular endothelium. Similar experiments, withcontrast, more recent studies with transplantable tumors similar conclusions, were also carried out using transfer of[243,91,132] and neurotropic viruses [182,197] showed whole cells (lymphocytes, erythrocytes and macrophages)very rapid kinetics. However, with all studies reliant upon in some early studies [57,169]. However, we currentlyintracranial implantation of cells or viruses, the absolute understand cell migration to be very dependent on adhe-integrity of normal brain structure cannot be totally sion molecule expression and chemokine gradients andguaranteed. Antigenic material may thus potentially leak to since many such factors have only recently been defined,the periphery, bypassing normal CNS antigen handling these earlier studies are difficult to interpret. Functionalmechanisms. The other unknown factor for recruitment of data supporting the importance of the cervical lymph nodesDC precursors such as monocytes to an inflamed brain is in CNS responses have shown that their removal di-whether differentiation to an immature DC can occur in the minishes B- and T-cell immune responses to antigensbrain parenchyma. Indeed, monocytes recruited to the expressed in the brain [183,100]. A cautionary remarkbrain at various stages of fetal and adult life are thought to must be made regarding generalizations about antigen


Fig. 2. Hypothetical mechanisms of T-cell immune response induction to primary cerebral malignancies. (A) Sampling of tumor-derived antigens bydendritic cells (DCs), recruited to the tumor site after initiation of an inflammatory response (see Sections 1.4 and 1.5). After phagocytosis of antigenicmaterial, DCs may differentiate to a mature phenotype under the influence of local factors expressed by stressed cells, such as heat shock proteins. MatureDCs migrate to secondary lymphoid tissue via pathways that are still undefined for the brain, where they activate naive T cells to differentiate and undergoclonal expansion. (B) Tumor debris and antigens derived therefrom drain with the ISF and potentially the CSF (see Section 1.5) to reach secondarylymphoid tissue wherein antigenic material is phagocytosed by resident APCs. If these APCs are also induced to express costimulatory molecules,activation and clonal expansion of naive T cells can occur. In the absence of costimulation, tolerance may ensue. (C) Antigen capture by a brain residentAPC such as a microglial cell (as opposed to a DC), followed by migration to secondary lymphoid tissue by a currently undefined pathway. Since bothpositive and negative outcomes of interactions with T cells have been reported for microglia (see Section 2.4), the consequences in the secondary lymphoidtissue may include tolerance induction as well as T-cell activation.


drainage pathways from the brain parenchyma. This con- the CNS are not mutually exclusive, and each may eachcerns the models and techniques used. Injecting with apply in different situations. ‘Classical’ immune responseprecision into defined CNS sites in small laboratory induction, by DC uptake of antigen (Hypothesis A) is mostanimals is technically difficult. Techniques pioneered by likely to occur in the context of acute inflammatory states,Cserr are able to limit BBB damage and involve the such as in infection or trauma induced either experimental-infusion of very small volumes in order to minimize ly /surgically, or by injury. Efficient immune responsetechnical artifact [100], but it is doubtful whether such induction in the absence of local APC involvement (Hy-precision is reproducibly obtained by all researchers in the pothesis B) is improbable unless there is a total absence offield. Another consideration is that the degree of communi- signals induced by stress or infection, thus this is likely tocation between ISF and CSF in different species may not occur under experimental conditions in which antigenicbe identical, with more separation between these compart- substances are infused intracerebrally. We consider that thements in man than in rats [247]. Overall, these data offer final hypothesis, invoking unique brain APCs (Hypothesisconvincing evidence for a degree of antigen drainage from C) particularly merits further investigation, since it is thebrain parenchyma, but certain difficulties remain. The first resident APCs that have the first opportunity for CNSis that if parenchymal antigens follow similar drainage antigen uptake. As we elucidate the behavior of such cells,patterns to the CSF, why does antigen implanted directly it should be possible to determine whether this hypothesisin the CSF induce much better responses than that im- is totally independent of a DC hypothesis, or whether localplanted in the parenchyma [132]? One explanation may be APCs readily differentiate into a DC-like cell, or whetherthat this is a reflection of the quantity of fluids draining DC precursors are efficiently recruited as a more efficientfrom the brain, with 90% being CSF and only 10% ISF in second line APC.the rat [57]. The other consideration is that whilst proteinand other antigens may be readily taken up by phagocyticAPCs, these cells require signals (for example, mediated2 . Effector stages of T-cell immune responsesby pathogen expressed molecules or host expressed stressproteins) to induce maturation to a state where they can Whilst the mechanisms responsible for the induction ofactivate naive T cells. It is far from clear whether such spontaneous T-cell immune responses in the CNS arestimuli will always accompany efflux of potentially an- hypothetical, there are more direct data concerning thetigenic material from the brain. The important consequence effector phase of the response. T cells infiltrating the brainis that an immature DC that phagocytoses antigenic parenchyma can be directly visualized in both clinical andmaterial draining from the brain may not adequately experimental situations and in some cases, sufficient cellsexpress costimulatory molecules and may thereby tolerize can be isolated for ex vivo functional tests.rather than activate naive T cells.

2 .1. Entry of primed T cells into the CNS1 .5.3. Hypothesis C

A further mechanism by which T cells may be primed to Both in vivo and in vitro data have contributed to ourbrain-derived antigens is that the antigen is transported to current understanding of how T cells pass the BBB to enterthe lymph node by a cell other than a DC. To date there is the brain parenchyma. The earliest published experimentsno direct evidence in support of this hypothesis, but there detailing this T-cell traffic were those of Wekerle et al.have been certain tantalizing observations. The first was in [246], who showed that adoptively transferred T cellsa transplant model, in which rat brain allografts were entered the brain, regardless of their specificity, as long asimplanted into the brain parenchyma. A small number of they were activated. Later studies by Hickey et al. [106]

1macrophage like cells expressing donor MHC were found showed that after in vitro activation with mitogens, CD41in host lymph node and spleen [42]. In a recent study [185] and CD8 T cells rapidly entered the brain parenchyma,

in a totally syngeneic system, 2 h after intracerebral with CNS concentration peaking at 9–12 h after intraven-1injection of antigen there was co-localization of the antigen ous transfer. CD4 T cells reactive with a CNS expressed

1with Mac1 cells (staining macrophages/microglial cells) antigen (myelin basic protein) were selectively retained in1 1in the brain parenchyma. After 4 h antigen , Mac1 cells the brain parenchyma, whereas cells of other specificities

were detectable in the cervical lymph node. However, this exited within 1 to 2 days. Similar findings were reported1result would be compatible with drainage of either free for CD8 T cells specific for a viral antigen expressed in

antigen or APCs. Whilst of great interest, neither of these the brain [116]. Thus, immunosurveillance by activated Tstudies directly established APC-mediated transport of cells would appear to follow similar rules for the brain asantigen from brain parenchyma to lymphoid organs and for other sites [45], except that T-cell entry to other sitessubsequent activation of naive T cells. It is clear that much may be significantly more efficient. For example, in a rattechnical ingenuity will be necessary to directly address model, adoptively transferred activated T cells entered thesuch questions, particularly in small laboratory animals. brain parenchyma much less efficiently than for other sites

These hypotheses for initiation of immune responses in (six times less than in muscle and more than 140 times less


than that found in liver, for the same weight of tissue) ‘pioneer’ T cells, entering a non-inflamed CNS, or whether[105]. an inflammatory response is under way

Although there are no brain specific molecular addres- [116,50,239,43,12]. Thus, early innate immune responsessins that have been unequivocally defined to date, there are (which are very difficult to adequately model in EAEclearly certain key interactions that facilitate T-cell ex- systems or after intracerebral implantation) will certainlytravasation to the CNS. However, it should be appreciated influence the subsequent recruitment of specific T cells by

1that many of the conclusions are based on CD4 T cells, the induction of adhesion molecule expression. In chronicprincipally in EAE models, that may not necessarily reflect conditions such as in the case of malignant astrocytoma,

1the situation for CD8 T cells. It is also important to note the tumor cells themselves can influence leukocyte infiltra-that T-cell entry to the CNS can occur even in the absence tion by secretion of chemotactic factors or factors regulat-of an inflammatory locus, although chemokines liberated ing adhesion molecule expression [61,213,240]. A chal-from an inflammatory site will presumably attract more lenge of future research in neuroimmunology is thateffector cells than normal tissue. current advances in ‘mainstream’ immunology in defining

1subsets of CD4 T cells such as Th1 and Th2 [157] and12 .2. Molecules involved in transmigration regulatory populations [189], CD8 T-cell Tc1 or Tc2

subsets [158] as well as populations of memory or effectorLymphocyte transmigration across the BBB broadly subsets based on chemokine receptor expression [205,138]

proceeds according to the multistep model of homing are investigated in the context of CNS immune responses.[228,264]. Tethering is the first step, in which activated Tcells make contact with endothelial cells principally with 2 .3. Antigen-specific restimulation in the brain: are

1 1the tips of their microvillous surface protrusions, forming specialized APCs necessary for CD8 as well as CD4temporary, low affinity bonds that rapidly dissociate at the T cells?cell’s upstream end, but which are replaced by new bondsformed downstream. This binding is sufficient to reduce Upon T-cell entry into the perivascular space, T cellsthe velocity of the T cells to a rolling motion along the will be ready to pursue a variety of careers depending uponvessel wall. The bonds are principally between selectin the stimuli they receive there. The minimum requirementsmolecules and oligosaccharides, with P-selectin on the for antigen recognition are MHC class I expression for

1 1endothelial cell being particularly implicated for the initial CD8 T cells and MHC class II expression for CD4 T1tethering of CD4 T cells to non-inflamed brain vascula- cells, together with the appropriate antigenic peptide. This

ture [50], although this has been contested by other groups may be adequate for eliciting certain biological functions[239]. Higher affinity interactions are necessary to fully (cytolytic activity, cytokine release) from an effector T cellarrest cells to permit diapedesis into the extracellular with a high affinity TCR, but other factors such asmatrix, these are mediated by integrins expressed by the T expression of adhesion molecules, costimulatory moleculescell (e.g. LFA-1,a4b1, a4b7) interacting with cell adhe- and cytokines can make antigen presentation more effi-sion molecules (CAMs) on the endothelial cell (e.g. cient. The question of brain APCs at the effector stages ofICAM-1, VCAM-1, MAdCAM-1). This interaction may immune responses is generally thought of as being par-

1require T-cell integrin activation by chemokines immobil- ticularly critical for CD4 T cells, because MHC Class IIized on endothelial cells interacting with receptors on T expression is less widely expressed than MHC class I, evencells, transmitting signals through G proteins [239]. Sig- after exposure to inflammatory cytokines. Furthermore,

1naling to the brain endothelial cell may also be important, CD4 T cells are frequently not the ultimate effector cellparticularly through ICAM-1. This has been proposed to in an immune response, they may be working throughstimulate a reorganization of the endothelial cell actin macrophages or B cells for example, which do not requirecytoskeleton, leading to either pore formation or disaggre- MHC expression to exert their effector functions. Whilst

1gation of tight junctions, thereby facilitating T lymphocyte this may occasionally be the case for CD8 T cells [255],diapedesis [3]. However, modifications of the multistep most fully differentiated classical CTLs are thought toparadigm of transendothelial migration have been post- directly kill any target expressing adequate peptide/MHCulated for the CNS, in whicha4-integrin may be able to class I complexes at the surface, for example a virallymediate binding to VCAM-1 and possibly MAdCAM-1 infected cell or a tumor cell. The question therefore arises

1under physiological flow (thus permitting tethering, al- as to whether CD8 T cells have any requirement forthough a rolling step may not be obligatory) as well as additional contacts with other APCs. In the case ofunder static conditions [239,72]. The complexities and antitumor immune responses we have recently demon-subtleties of T-cell /endothelial cell interactions are an strated that tumor-specific, brain infiltrating CTLs doactive area of current research and undoubtedly involve interact with brain APCs (Calzascia et al., manuscript inmany molecular interactions, not all of which have been preparation). The functional consequences of antitumorfully characterized. Furthermore, the importance of differ- CTLs interacting not only with their tumor cell targets, butent interactions will vary according to whether T cells are also with local APCs have not yet been directly de-


termined, but data from different systems suggest that a natively, there could be an amplification of the immunevariety of outcomes are possible (Fig. 3). Immune re- response, for one or several of the following reasons.

1sponses may be downregulated, either because of Recruitment of CD8 T cells from the periphery may betolerogenic properties of the APC, or through the lysis of a enhanced by the presence of APCs in the brain paren-stimulatory APC by a fully differentiated CTL. Alter- chyma [51]. Furthermore, the subsequent penetration of T

Fig. 3. Effector stage of a T-cell-mediated response to a cerebral malignancy. T cells activated in the periphery extravasate through the intact or locallycompromised blood–brain barrier (BBB). Once in the perivascular space, there will be encounters with APCs such as microglial cells and dendritic cells(DCs) (see Section 2.3). (A) Interactions with APCs expressing costimulatory molecules may maintain viability and function of effector T-cell populations,and counteract immunosuppression mediated by endogenous or tumor-derived factors such as TGF-b. Synergistic co-operation between T-cell subsets

1 1(CD4 and CD8 ) may also be encouraged through T-cell clustering around an APC expressing both MHC class I and class II molecules. Efficientantitumor effector function can thereby result. (B) Interactions with local APCs that regulate immune function in order to limit neuroinflammation may alsorestrict antitumor function. T cells may be tolerized or retained in peritumoral regions and those that infiltrate the tumor matrix may succumb toimmunosuppression mediated by tumor expressed factors such as TGF-b and FasL, leading to tumor immune escape.


cells from the perivascular space into the parenchyma may functions such as phagocytosis and (in vitro) presentation1be dependent upon interactions with perivascular APCs, of antigen to CD4 T cells [105,10]. Such a process may

1although to date this has only been demonstrated for CD4 result in local cytokine release, influencing endothelial cellT cells [235]. The presence of costimulatory molecules by phenotype and may thereby regulate lymphocyte extrava-the APC may protect CTLs from activation induced cell sation.death and thus augment the magnitude of the response at Brain macrophages, generally referred to as microglialthe population level [112,140,37]. Moreover, the potent cells, are prime candidates for antigen presentation in thestimulatory capacity of a professional APC may enable brain [209,10,8,9]. Although many subpopulations exist inCTLs to fully differentiate or overcome inhibitory mi- different anatomical locations (meningeal macrophages,croenvironmental factors such as TGF-b [91,92]. It is also choroid plexus macrophages, epiplexus cells), two broadpossible that on the rare occasions when naive T cells can categories will be considered here: parenchymal microgliaenter the CNS, after BBB breakdown or inflammation and perivascular microglia [105,10]. Both populations are[133], a fully mature and activated professional APC of hematopoietic origin, but their entry and turnover in thewould be able to activate naive T cells. The benefits of brain occurs at different times and rates. The parenchymal

1APCs for CD8 T cells may also be indirect, with CTLs microglia populate the brain during fetal life and are1benefiting from cytokines (e.g. IL-2) released by CD4 T long-lived cells, with a turnover of only a few percent per

1cells clustered around a MHC class II APC, or other year [139]. Under non-pathological conditions, parenchym-cytokines (e.g. IL-12) liberated by an APC stimulated by al microglia constitutively express few classical macro-

1the CD4 T cell. The consequence may be local mi- phage markers (or express them at low levels) and arecroenvironmental conditions that protect against apoptosis poorly phagocytic. Perivascular microglia, bear moreand encourage a degree of intracerebral T-cell clonal resemblance to conventional macrophages, they are aexpansion [161]. Priorities of future research investigating dynamic, constitutively phagocytic population that are

1APCs for CD8 T cells include the demonstration of local regularly replenished, as originally observed in transplanta-APC function in different models or neuropathologies. tion studies in rats, in which the grafting of bone marrowThereafter, modulation of such functions can be most induced a new population of perivascular microglial cellsrationally attempted if the responsible cell type is defined of donor origin in the host [134,69]. When activated,(as discussed below). microglial cells possess certain characteristics of profes-

sional APCs, with expression of CD1a, MHC class I and2 .4. APC candidates in the brain class II molecules, as well as adhesion and costimulatory

molecules such as B7.1 (CD80), B7.2 (CD86), LFA3Whilst few cell types present in the normal brain (CD58), CD40 and ICAM-1 (CD54) [251,256,210]. In

constitutively express MHC molecules, once an inflamma- vitro, they can induce alloreactive T-cell responses andtory response is under way many resident or recruited cells stimulate T-cell lines to secrete cytokines and proliferate.are MHC positive and thus may contribute to antigen In vivo, B7 molecules are detected on reactive microglialpresentation. There are many technical difficulties in cells present in inflammatory multiple sclerosis lesionsisolating sufficient numbers of a given candidate brain [250] and in brains of mice suffering from Theiler’sAPC population to perform ex vivo analyses. Many virus-induced demyelinating disease [123]. Another inresearchers in the field have thus resorted to cultured cells, vivo observation was of donor T-cell-activated microglialoften from fetal or newborn animals to assess function. cell clusters in graft-versus-host disease, in which theWhilst these studies are informative, culture systems from microglial cell strongly expressed CD11b/c and MHCdifferent laboratories are rarely identical and none can class II [216]. Proliferation markers suggested that atotally match the complex interactions occurring in vivo. proportion of the microglia were cycling. Some authorsThe in vitro properties of the various cell types thus have suggested that microglial cells may have a role indefined are best interpreted as potential functions, which tolerance induction, or the termination of immune re-must eventually be confirmed in vivo. sponses [39,149,84,85]. Such effects were particularly

Following extravasation of T cells, the first potential associated with the activation state of the microglial cellencounters with APCs will be with endothelial cells and [149], and with their parenchymal rather than perivascularcapillary pericytes, on which MHC class II molecule and localization [85].adhesion molecule expression is inducible by interferon-g, A further non-professional APC candidate resident intumor necrosis factor-a, and IL-1 [75,74,179,23]. Endo- the brain is the astrocyte. As far as normal astrocytes arethelial cells may even be in a position to present antigen to concerned, MHC class I and II molecules are eitherT cells at the luminal surface, however since T cells undetectable or weakly expressed in vivo [136], whereasextravasate regardless of their antigen specificity [105], cultured astrocytes can express high level of MHC class Ithis is probably not of great physiological importance. and II molecules, particularly after incubation withPericytes are an interesting APC candidate, since in some interferon-g [63,245,148]. Cultured astrocytes can presentdifferentiation states they can acquire macrophage like foreign antigens to class I and class II restricted T cells


[63,245,226], but may be unable to trigger a complete infiltrative but do not generally metastasize outside theT-cell activation program [8,245]. Furthermore, for pre- brain. It may be assumed that at the earliest stages of

1sentation of antigens to CD4 T cells, they are proposed to neoplastic transformation, the first tumor cells are totallystimulate a Th2 T-cell immune response [1,10,8], rather contained within the normal architecture of the brain. Suchthan the Th1, pro-inflammatory response more readily primary brain tumors thus present a unique challenge forelicited by activated microglial cells. Cultured astrocytes CNS immunosurveillance, since unlike viral, bacterial orare considered to be the in vitro correlate of reactive parasitic infections, the systemic immune system will notastrocytes in vivo, readily induced after any form of brain have had the possibility of a direct contact with foreign ortrauma, possibly as a secondary consequence of microglial mutated self antigens. However, at later stages of tumorcell activation. However, in vivo induction of MHC progression there may be some destruction of BBBexpression in astrocytes is a controversial issue: certain integrity as well as substantial neoangiogenesis, resultingstudies suggest that MHC class I can be induced after viral in intratumoral vessels that do not bear the hallmarks ofinfection or exposure to IFN-g [55,221,120], whereas other the BBB.authors have suggested that this will only occur if there is Immunohistological and molecular analyses of humanneuronal degeneration [109,190]. malignant astrocytoma have shown that T-cell infiltration

Evidence from different organs (e.g. liver [143]; eye is a frequent occurrence [196,230,180], but it has only[78]) indicates that local APC are key players in directing occasionally been correlated with a favorable prognosisqualitative, site-specific aspects of immune responses. For [44]. However, in the absence of a definition of thethe brain, many or all of the candidates discussed above specificity and function of the infiltrating T cells, themay subtly influence local immune function. However, for significance of these findings is not clear. The first issueimmunomodulation in future clinical therapies, resident or that must be addressed is whether malignant astrocytomarecruited cells of hematopoietic origin, with professional cells are sufficiently antigenically distinct from normalAPC function, are the most enticing candidates to target tissue to be recognized by tumor-specific T cells. Theand manipulate. second is whether T cells activated by tumor expressed

antigens are capable of retaining effector function wheninfiltrating a tumor growing in the brain parenchyma.

3 . T-cell immune responses to brain tumors in It is apparent from recent literature that newly definedclinical and experimental situations antigens expressed by human tumors are regularly being

characterized [244,191]. However, few results to date haveThe principles of the induction of immune responses been published concerning astrocytoma antigens able to

have been considered as well as how the effector phase of spontaneously elicit an immune response. Most studiesCNS responses may occur. The final outcome depends on have assessed astrocytoma associated antigen expression atthe immune response that was induced and the efficiency the mRNA level [54,204,212] or by serology [53], ratherof the effector phase of the response. In the case of tumors than T-cell defined antigens. Many of the antigenslocated in the CNS that will be considered in this section, screened for in these studies are those that have beensimply observing whether a tumor is eliminated by an previously defined in melanoma, a reasonable startingantitumor immune response is not always very enlighten- point given the common neuroectodermal origin ofing. In order to understand whether and how immune melanocytes and astrocytes. The list of antigens potentiallyresponses directed towards brain tumors can operate, it is expressed by at least some astrocytomas includes severaltherefore necessary to look to systems in which it has been antigens that are not totally tumor-specific, but that havepossible to observe and analyze immune parameters. limited expression in normal somatic cells. The list of

putative antigens includes MAGE and GAGE family3 .1. T-cell immune response against human malignant members, tyrosinase, TRP-1, TRP-2, gp100, p97, SSX-1,astrocytoma SSX-2, SSX-4, SCP-1, and TS85. Expression of T-cell

epitopes derived from these antigens at the tumor cellAstrocytomas derive from neuroectodermal glial cells surface has been rarely studied. However, MAGE-1 an-

and are the most common primary brain neoplasms. The tigen expression is detected in some cultured astrocytomamalignant forms include anaplastic astrocytoma (grade III) cells and can be recognized by specific CTL, but thisand glioblastoma (grade IV); these are tumors that pro- antigen is not generally expressed by astrocytomas in vivogress rapidly and are almost invariably fatal. Indeed, [212,58], probably because of a different level of DNAdespite advances in the application of combined modality methylation induced by culture [59]. Further antigens thatcancer treatments [233], these have had little impact for are recognized by CTL in vitro, are SART1 , originally259

malignant astrocytoma patients, with a median survival identified in epithelial cancer cells, but now also shown torate of less than 12 months for glioblastomas [232]. From be expressed in malignant astrocytoma biopsies [113], andthe neuro-immunological standpoint, it is of interest that an epitope from the IL-13 receptora2 chain [173,60]. Thethese tumors arise and remain in the CNS, they are highly list of astrocytoma antigens confirmed to be capable of


stimulating T cells has remained limited for certain very servation of structural features of the critical TCR CDR3practical reasons. The classical approach to detecting region among the striking oligoclonal expansions that wetumor-specific T cells is the co-culture of patients’ T cells observed in astrocytoma is most adequately explained aswith putative antigen expressing autologous tumor cells, being the result of tumor antigen-driven clonal expansiontogether with appropriate T-cell growth factors such as of specific T cells. However, it is clear that the fullIL-2. This may allow preferential expansion of antigen significance of these cells will only be fully unraveledspecific T cells, which can subsequently be tested for when their specificity can be determined.specific, MHC-restricted recognition of the autologous The principal oligoclonal expansions of T cells with

1tumor cell lines, or antigens derived therefrom. If the homogeneous CDR3 length were mainly CD8 T cells, intumor infiltrating T cells are isolated, they may be pref- the patients that we examined. However, judging fromerentially enriched for tumor-specific cells, due to antigen- immunohistological and flow cytometric analysis, similar

1 1specific retention at the tumor site, as previously discussed. numbers of CD4 and CD8 T cells infiltrate the tumor.Astrocytoma infiltrating T lymphocytes can be isolated for This obviously raises the question as to the function and

1culture, although the yield is generally modest due to the specificity of the infiltrating CD4 T cells. Are theselimited biopsy size and a degree of infiltration that is less inflammatory cells recruited to the tumor site, but notintense than that found for certain other tumors such as specific for any locally expressed antigen? Are they in factmelanoma. The relatively limited number of cells that can specific, but with heterogeneous CDR3 expression? Is the

1be isolated necessitates significant in vitro expansion to number of CD4 T-cell epitopes relatively large, so1generate sufficient cells for testing. However, this problem retaining CD4 T cells of many specificities? Is the

1is compounded by the fact that astrocytoma infiltrating absence of dominant CD4 T-cell oligoclonal expansions alymphocytes (as for T cells in certain other diseases) reflection of an inadequate CD4 arm of the response, and

1exhibit reduced proliferative potential in culture and those thus a factor that may limit the efficacy of the CD8 cellscells that do grow in vitro show an important skewing that co-infiltrate? These are important questions, but onesduring the culture period [65,154,159] and may not that we cannot answer without more knowledge about the

1necessarily be representative of the starting population of specificity and function of the CD4 cells. Indeed, our1tumor infiltrating lymphocytes. understanding of CD4 T cells in tumor immunology is

1To circumvent such limitations, we have adopted a less advanced than for CD8 T cells, there are fewer MHC1novel molecular strategy to analyzea /b TCR T cells class II restricted tumor antigens that have been defined

1present in the immune infiltrate and in the blood of and the characteristics of the CD4 T-cell repertoiremalignant astrocytoma patients. As previously discussed, specific for these antigens are unknown.

1a /b TCR T cells recognize antigenic peptides presentedby MHC molecules using a heterodimer composed of ana 3 .2. Immune escape of astrocytomasand a b chain, the hypervariable complementarity de-termining region (CDR) 3 region of which carries the If optimal stimulation of the immune system does notprincipal antigenic specificity of a T cell. We have used occur when the tumor mass is small, elimination of a solidTCR spectratyping, a high resolution reverse-transcription brain tumor by immune effector cells is a formidablepolymerase chain reaction-based technique that determines challenge. Part of the problem may be the magnitude of theTCR b chain CDR3 length (five to 15 amino acids) within immune response that is induced: the immune infiltrationthe different V gene families. This is a powerful approach that is induced may be insufficient. But the tumor cellsto study TCR diversity in a blood or tissue RNA sample, may also resist immunological eradication by variousand thus to detect T-cell clonal expansions in vivo passive and active mechanisms of immune escape. Indeed,[150,176]. Indeed, the identification of recurrent CDR3 over the past three decades, many defects have beenregions in large T-cell populations generally indicates reported in the functional status of T cells in brain tumorantigen driven expansion of the corresponding T-cell patients, concerning circulating T cells as well as thoseclones. The analysis of a large panel of biopsies from infiltrating the tumor [67]. In vivo manifestations of thesemalignant astrocytoma patients showed that oligoclonal defects include delayed type hypersensitivity responses,expansions of T cells were present within various Vb low T-cell numbers in peripheral blood and low serumsubfamilies of all biopsies tested [180]. In contrast to this antibody titers. In vitro tests indicated poor responses tolocal reaction, a systemic immune response was not mitogens, poor cytotoxic function, defective expression ofgenerally detectable, since only exceptional oligoclonal high affinity IL-2 receptors and activation signaling defectsexpansions were detected in peripheral blood. It is there- [156]. Such phenomena may explain difficulties in cultur-fore unlikely that T cells clonally expanded in astrocytoma ing CTLs derived from astrocytoma infiltrating lympho-are the direct consequence of blood expansions, such as cytes, despite the addition of recombinant IL-2 [71,155].those observed in certain healthy people [73,215,164]. However, the real implications of these features cannot beThese data for brain tumors are similar to those found for fully assessed until it is known whether astrocytomaother types of cancers [147,77,220,47,46]. Overall, con- specific T-cell activity has been impaired. Nevertheless, in


1 1the absence of a direct confirmation of this, many poten- kill CD4 and also CD8 T-cell lines derived from thetially immunosuppressive soluble or cell associated mole- autologous tumor, specifically using the FasL–Fas path-cules have been characterized. way [241]. Other groups have independently confirmed

expression of FasL by astrocytoma [96,79,86] and, more-3 .2.1. Immunosuppression by soluble factors over, apoptotic T cells were observed in the proximity of

A list of potentially immunosuppressive molecules FasL expressing astrocytoma cells in vivo [64]. However,detected in astrocytomas or astrocytoma lines includes other than this indirect in vivo evidence of a role forprostaglandin E [67,56,83,211], gangliosides [249] and astrocytoma expression of FasL, the in vivo consequence2

IL-10 [167,153,107], all of which can demonstrate certain of FasL expression by tumors has been a very contro-immunosuppressive functions in vitro, but for which in versial issue. Indeed, tumor expression of FasL in murinevivo immunosuppression in astrocytoma patients remains models has been correlated either with enhanced tumorspeculative. However, one astrocytoma-derived factor that growth [99,16], or with enhanced tumor rejectionhas been explored in more depth, in vitro and in vivo, is [218,219,17], with augmented neutrophil recruitment as aTGF-b. Indeed, this cytokine was originally called gliob- likely mechanism. Microenvironmental factors will cer-lastoma cell-derived T-cell suppressor factor and was first tainly influence the consequences of FasL expression byidentified in the supernatant of a human glioblastoma cell tumor cells, for example, tumors co-expressing both FasLline that suppressed T-cell growth [56,82,35]. The im- and TGF-b may be particularly well adapted to combatmunosuppressive effects of TGF-b are multiple and com- CTL effector mechanisms (discussed in Ref. [241]), aplex. They include the inhibition of maturation and antigen principle subsequently confirmed in in vivo models inpresentation by DCs or other APCs, inhibition of T-cell which a FasL-positive colon carcinoma could escapeactivation and differentiation towards effector cells (either rejection if TGF-b was also present [52]. These data helpcytotoxic cells expressing perforin or Th1 or Th2 cells) to explain the role of FasL in this particular model (with[224,115,93,117]. Inhibiting TGF-b in vivo in experimen- possible analogy to the situation in the brain), but it istal tumor models has produced mixed results, probably likely that individual combinations of factors relevant tobecause the effects of this cytokine are not only on the different tumors or models are responsible for the diverseimmune system, but also on the tumor cell, which may be interpretations of these issues in the literatureprotected from Fas (CD95)-mediated apoptosis in some [241,242,193,168].circumstances [76,21]. Other in vivo studies using decorin, Recently, HLA-G has been proposed as an astrocytomaa natural inhibitor of TGF-b, suppressed the growth of C6 expressed molecule with immunosuppressive potentialrat astrocytoma in vivo [229], but whether this occurred [248]. This non-classical MHC class I molecule is ex-through TGF-b is unclear, because decorin is also im- pressed by a limited range of tissues, particularly themunostimulatory in a TGF-b2 independent fashion [160]. placenta, but also certain cancers. It is proposed toOverall, the immunoregulatory functions of TGF-b2 war- suppress NK and T-cell immune responses, but this isrant the attention it has received, but whether it will be controversial [40,22]. Regarding astrocytomas, a propor-feasible or advisable to inhibit this cytokine in brain tumor tion of astrocytoma cell lines and tumor biopsies expressed

1 1patients still remains uncertain. HLA-G protein and inhibition of CD4 and CD8 T-cellresponses was demonstrated in vitro, but this was only

3 .2.2. Immunosuppression by cell-mediated interactions tested after incubation of cell lines with high concen-The situation for cell-mediated immunosuppressive fac- trations of IFN-g (500 U/ml), or after gene transfer of

tors is similar to that found for soluble factors, with several HLA-G into glioma lines. A final candidate immuno-candidate molecules, but not all that have been thoroughly suppressive molecule is the CD70 transmembraneinvestigated in vivo. We investigated the possibility that glycoprotein, with expression by astrocytoma cell lines orglioma may utilize membrane-bound Fas ligand (FasL, in vivo now confirmed in two independent studiesCD95L) as a mechanism of immune escape. FasL belongs [252,104]. This molecule is also expressed on activated Tto the tumor necrosis factor family and is implicated in and B cells, with roles in regulating immune responses viaseveral biological functions through its interaction with Fas interaction with CD27, expressed on lymphoid cells [118].(CD95), a member of the tumor necrosis factor receptor / In vitro functional tests suggested a pro-apoptotic role ofnerve growth factor receptor family. FasL–Fas interaction tumor expressed CD70 when tested on PBMC targets; thisinduces the trimerization of Fas and a subsequent complex was augmented (correlating with expression levels ofcascade of intracellular events, potentially leading to CD70) when tumor cells were irradiated [252].apoptosis of Fas-positive cells, a mechanism central to

1immune homeostasis [162,146]. Indeed FasL astrocytoma3 .3. Immunotherapy—from rodent models to the cliniccells (cell lines and also astrocytoma cells tested ex vivo)can specifically and efficiently kill Fas-transfected P815 In view of the limitations in analyzing immune re-target cells, but not wild-type P815 [202]. Moreover, we sponses to spontaneous tumors arising in humans, it mayhave also shown that a human astrocytoma cell line can be hoped that appropriate experimental models will be


informative. The ideal model for a human malignant frequently inflammatory. Furthermore, chronic inflamma-astrocytoma will reproduce the key features of this neo- tion may provoke a release of sequestered autoantigens,plasm: poor immunogenicity, infiltrative but non-meta- leading to the initiation or perpetuation of autoimmunestatic tumor growth, expression of multiple mechanisms of disease. An indication of the possible risks of utilization ofimmune escape. Since such a model does not exist, we non-defined brain tumor vaccines came from early studiesmay have to exploit several different systems for their employing human glioma tissue to vaccinate differentapplicability to model certain aspects of the tumor–host species of experimental animals: this resulted in lethalrelationship. For the future, genetic models of ‘sponta- allergic encephalomyelitis [34]. Other unexpected unto-neous’ brain tumors may be informative; indeed, these ward effects arose in rodent trials of a gene therapymodels incorporate some of the genetic features and protocol for glioblastoma, in which there was a vectorheterogeneity typical of spontaneous human cancer [108]. induced intracerebral immune response that directly orHowever, such models have generally been designed to indirectly perpetuated reactive gliosis and demyelinationaddress genetic and pathologic issues and present a formid- [62]. However, even immunotherapeutic approaches thatable challenge for immunological analyses. For the mo- appear to be well defined can give unexpected results thatment, models used to investigate immune responses risk being uninterpretable if there is no possibility of aagainst cerebral malignancies have relied upon transplant- specific read-out for the antitumor immune responses. Thisable tumors, generally implanted into defined regions of was illustrated in a model in which tumor rejection ofthe brain under stereotaxic guidance. Overall, such models subcutaneous tumors was targeted towards a model antigenhave provided encouragement for future immunotherapy, expressed by gene transfer in rat 9L glioma cells. How-in that they have demonstrated that immune responses ever, the ultimate rejection of intracerebral tumor wasinduced in the periphery can mediate antitumor effects in mediated by immune effector cells of other specificitiesthe brain [15,18,192,207,20,171,94,4,142]. However, op- [142], that had presumably been induced by a process oftimization of any potential immunotherapy requires an epitope spreading, as seen in certain autoimmune diseasesassessment of the immune (or any other) mechanisms [135].responsible for tumor rejection. An adequate characteriza- The somewhat overwhelming list of potential problemstion of immune-mediated rejection cannot really be that may be encountered in any immunotherapy for brainachieved without knowledge of the antigenic targets of the tumors has not totally blocked the interest and implementa-antitumor response. Unfortunately, in certain rodent brain tion of clinical trials. This is understandable from a clinicaltumor models, characterization is sketchy and the systems perspective, given the lack of treatment options for patientsare not always totally syngeneic [91,236,33], although with malignant astrocytoma. It is also relevant for re-some recent studies have used better characterized tumor searchers involved with more fundamental aspects of themodels in the brains of syngeneic mice, such as B16/F10 CNS antitumor immune response, in order to stimulate andmelanoma [207,20], C3 sarcoma cells transfected with the guide the creation of better models. An overview of thehuman papilloma virus type 16 [171] and P815 cells categories of immunotherapy which may be considered fortransfected with the CW3 antigen [243]. treatment of brain tumors in a clinical setting is provided

With the findings that under ideal conditions in ex- in Table 1. To date, an objective analysis of the publishedperimental models, vigorous CTL-mediated immune rejec- data from clinical trials is extremely limited due to variabletion of a cerebral malignancy can be achieved, can such descriptions of patient groups or disease status, as well asresults be directly translated to the clinic? Perhaps the the sometimes premature reporting of survival data. Withprincipal limiting factor to date is the lack of defined tumor these limitations in mind, and with the absence of anyassociated antigens that are expressed by a sufficient obvious clinical breakthrough, immunotherapies will notproportion of tumor cells in vivo to serve as tumor be reviewed exhaustively, but rather for their futurerejection antigens. Immunotherapy approaches for tumors interest and potential.in sites other than those of the CNS, in particularmelanoma, have induced CTLs that react with differentia- 3 .3.1. Adoptive immunotherapytion antigens such as Melan-A [191]. Of course, such This approach offers the advantage of circumventingapproaches may induce a degree of destruction of normal deficits of antitumor effector cell populations in the host bytissue [177], which is considered acceptable up to a certain providing optimal conditions for the culture and amplifica-point, although the balance of these autoimmune effects tion of these cells in vitro, in the absence of tumor-derivedcompared with antitumor effects is under scrutiny immunosuppressive factors.[262,26,68]. For immunotherapy of brain tumors, the Most adoptive immunotherapy in glioma treatment tothreshold of acceptability of autoimmune reactions is lower date has been a non-specific therapy, exploiting thebecause of the indispensability of most normal tissue in the cytotoxic properties of lymphokine activated killer (LAK)brain and its limited capacity for self renewal. The related cells. Obtaining such cells is relatively simple comparedproblem is that of inflammation. High magnitude immune with other cellular therapies. It involves the culture ofresponses leading to efficacious antitumor activity are leukapheresis-derived autologous peripheral blood lympho-


Table 1Immunotherapies inducing cell-mediated effector function against malignant astrocytoma

Approach Advantages Drawbacks Monitoring

Non-specific adoptive immunotherapy. Efficient in vitro culture in the absence Lack of specificity; inflammation Main readout is clinical response

LAK cells with high-dose IL-2 of local immunosuppression

Specific adoptive immunotherapy. In vitro culture in the absence of local Sufficient in vivo sensitized cells Transferred cells can be tracked

T cells from tumor biopsy or immunosuppression; tumor specificity difficult to obtain

vaccinated patients can be measured before transfer

Non-defined vaccines. Whole tumor Identification of tumor antigens not Quantity of antigenic material needed Readouts based on ‘whole cells’

cell homogenate/eluate/RNA etc., with necessary; applicable to all HLA-types; to prepare vaccine; autoimmunity; require autologous cellular controls

cytokine or DC adjuvants broad immune response, limiting risk patient-specific vaccine preparation if —difficult for CNS

of tumor immune escape autologous cells used; vaccine composition

variable

Defined vaccines. Peptides corresponding Standardized preparation for a given Lack of defined antigens for human astrocytoma; All tests for specific T-cell function

to CTL epitopes with synthetic adjuvants, HLA-type; autologous tumor cells not specific for individual HLA types; immune possible, even without autologous

or pulsed onto DCs needed; degree of expression on tumor selection of tumor escape mutants control cells

and normal cells can be assessed

cytes with high concentrations of IL-2, this generates a vaccination of the patient with autologous tumor or tumor-polyclonal population of cytotoxic cells from cells of T derived antigens [254,184]. Since the patients are vacci-and natural killer (NK) origin that are able to lyse NK- nated at a site distant from the tumor, this may avoid theresistant targets, including tumor cells. However, the key impact of tumor associated immunosuppression. However,characteristic of LAK activity against autologous tumor whether any of the isolated clinical responses that werecells was not assessed in most trials. Strategies to maxi- reported correlated with tumor specific effector functionmize the potential cytotoxic function of LAK cells include cannot be assessed, due to absent or inadequate characteri-the concomitant administration of IL-2 and the direct zation. Since even minor variations in vaccination pro-transfer to the tumor cavity at the time of resection (for a cedures and subsequent in vitro T-cell culture significantlymore detailed review of these therapeutic approaches, see influence the function and quantity of antitumor effectorRefs. [263,260]). Although this approach to experimental cells, it is disappointing that we do not have the in-immunotherapy of brain tumors is one of the longest formation to rationally propose improvements to theseestablished and has yielded occasional isolated clinical techniques. There have been recent developments inresponses, it is still not possible to confirm any clinical immunotherapy for tumors in sites other than the CNS thatefficacy. Furthermore, there have been variable levels of aim to ameliorate engraftment and antitumor effect oftoxicity associated with LAK cell and IL-2 therapy [27], transferred cells by prior lymphodepletion of patients [68].mainly as a result of the cerebral edema induced by the This approach shows certain promise for the therapy oftreatment. When the critical and delicate nature of the CNS melanoma, which if confirmed, may be possible to adaptis taken into account, it is difficult to envisage how the for the treatment of malignant astrocytoma.therapeutic potential of a non-specific therapy reliant uponcytokines triggering intense inflammatory responses can be3 .3.2. Active vaccinationsafely applied to brain tumors. Cancer vaccines fall into two broad categories: whole

More sophisticated forms of adoptive immunotherapy tumor cell / tumor cell-derived vaccines and defined antigenthat exploit the specificity of the immune system have been vaccines. Non-defined or whole cell vaccines require noexplored. As we have discussed in Section 3.1, tumor prior definition of tumor antigens or MHC restrictioninfiltrating lymphocytes should be enriched for T cells with elements, many epitopes are presumably present in theclassical, MHC restricted specificity for tumor cells. A vaccine, lessening the considerable risk that tumor cellspilot study using expanded tumor infiltrating lymphocytes may mutate and escape a too finely focused immunehas been undertaken, but with no evidence that the response. However, assessing vaccine efficacy and cor-transferred cells were specifically cytotoxic for the au- relating any clinical responses with specific immunetologous tumor [186]. The size of tumor biopsies available responses is extremely difficult. Furthermore, as vaccinefrom CNS malignancies is extremely restricted and the few efficacy improves, there is a risk that autoimmune re-T cells derived from the tumor are difficult to expand in sponses to normal tissue antigens present in the vaccinevitro, these factors will inevitably limit any widespread preparations may also be induced [34,144]. As previouslyapplication of this approach. An interesting variant of argued elsewhere [170], rational improvement in non-adoptive immunotherapy is to attempt to increase the defined vaccines is difficult once such approaches arenumbers of T cells used for in vitro culture by a prior incorporated into clinical trials, even if there is the hope of


benefit for isolated patients. The alternative strategy of [225]. Critical differences in the experiments include theinducing immune responses to defined antigens is more species and strain of animal and corresponding tumor (ratstraightforward to interpret, but this requires that clinical and mouse), the use of DCs of different origin (DC celltrials fully integrate the collection of biological as well as lines, or spleen or bone marrow-derived precursors), theclinical data. The hurdle for defined vaccines for brain use of different sources of antigen (from RNA to wholetumors such as malignant astrocytoma is the paucity of cells) and different means of DC maturation. Finally, DCsantigen candidates able to be recognized by T cells (as were injected using different routes and frequencies, and atdiscussed in Section 3.1). Therefore, to date, clinical trials different stages relative to tumor implantation. Certainhave employed non-defined tumor-derived material for encouraging conclusions from these animal studies canvaccination. nevertheless be drawn, namely that no severe side effects

To induce immune responses against antigenic but were encountered, and that there is some impact ofpoorly immunogenic human tumors requires efficient immune effector cells on intracranial tumors. Treatment ofadjuvants that enhance and stimulate antigen presentation. malignant astrocytoma patients with DC-based therapies isThis can be achieved either by exploiting existing host now under way, and published results to date confirm thatantigen presentation pathways in vivo, or by implanting ex vaccines consisting of DCs pulsed with peptides elutedvivo generated DCs, or DC/ tumor cell hybrids as cellular from tumor cells [259], and DCs fused with tumor cellsadjuvants. The most common route to ameliorate endogen- [128], did not induce serious adverse effects. Evidently,ous DC function is to use recombinant cytokines, either efficacy can only be addressed in future randomized trials,secreted by gene transfer into tumor cells, or by the but if a higher priority is given to obtain adequateco-injection of a bystander secretor cell (reviewed in Ref. immunological data, the future development of this other-[178]). Key factors that have received much attention in wise exciting approach can be accelerated.this context are GM-CSF, IL-4, FLT3 ligand and CD40ligand, which may find future application in brain tumor 3 .3.3. Immunomodulation with cytokinesimmunotherapy, and certain of which are already being Experimental therapy of brain tumors has also beenexploited in clinical trials [172]. attempted by employing cytokines that may act at the

The ex vivo approach, of injecting cultured DCs, is of effector phase of the response. Indeed, there is a vastconsiderable interest because of the possibility of exactly palette of recombinant cytokines available that can becontrolling the parameters of the DCs that are injected. employed by direct injection, gene therapy or via the useRecent advances in understanding how DCs play a central of transduced secretor cells. Certain preclinical studiesrole in initiating, amplifying and regulating adaptive have suggested that CTL expansion and function againstimmune responses have provided much data on the ideal brain tumors may be enhanced by IL-2 [90,127], IL-7 [15],characteristic of an immunostimulatory DC [24]. The IL-12 [127,121] and IL-18 [126]. Effector functions notgeneralized method is to generate DCs ex vivo from bone exclusively mediated by CTLs may also be augmented bymarrow or peripheral blood precursors, load the DCs with IL-4 [30,89] and IL-10 [38,217]. The challenge is to knowtumor-derived antigens, then reinfuse into the animal or which of these reagents merit development for clinicalpatient. Despite widespread application of these ap- trial. Most of these cytokines have been tested in studiesproaches, the technology is difficult to standardize and using totally different tumors and experimental protocols,variation in antigen loading or the differentiation of at different doses, rendering comparisons of individualcultured DCs used for injection can not only alter the factors difficult to make. Assessing the potential interest ofmagnitude of immune responses that are subsequently a particular cytokine is particularly acute for brain tumorinduced, but can also switch the balance towards tolerance therapy because there is no predominant model, unlike theinduction [231]. Correlation of clinical efficacy with situation for melanoma where preclinical testing of im-vaccine formulation can therefore only be adequately made munotherapies frequently employs the poorly immuno-if DCs used for treatment are adequately characterized, genic B16 melanoma. Nevertheless, it should be acknowl-which, to date, has not always occurred. edged that certain groups have published useful data

The induction phase of antitumour vaccination in a comparing multiple therapeutic approaches in the sameperipheral site will be similar for most extracranial as well experimental brain tumor models [207,208,145]. The cur-as intracranial tumors. Therefore, the design of DC-based rent limitation for advancing brain tumor immunotherapybrain tumor therapies can profit from the wealth of is not the number of immunostimulatory molecules that arebiological and clinical expertise now accumulating from available, but rather our comprehension of the immuneother DC enhanced vaccines [231]. Indeed, the adjuvantici- consequences (rather than just immune rejection) in thety of DCs has been confirmed in various brain tumor CNS. For example, cytokines employed locally that inducemodels [14,103,129,253,257,258,141,5]. Study design is so excessive inflammation may be less applicable to intracra-totally different in these reports that meaningful com- nial than extracranial sites. Or cytokines that interact withparisons of efficacy are difficult to make, but many of the other cells that are unique to the CNS and that alsodetails of the outcomes have been previously reviewed participate in the immune dialogue, such as microglial


cells and astrocytes [7,201,200,199], should be thoroughly therapeutic molecule is TNF-related apoptosis-inducinginvestigated for their positive or negative regulatory ef- ligand (TRAIL) [19]. Although certain reports suggestedfects. Certain new approaches to immunotherapy that are some reactivity with normal cells of the brain [166], recenttailored for the CNS are particularly worthy of mention. developments have indicated routes by which its therapeu-One of the most interesting is based on the propensity of tic potential may be maximized by restoring apoptoticneural stem cells to track migrating glioma cells within the pathways in glioma cells using peptide drugs [87]. Usingbrain [2]. It was proposed that genetically manipulated such an approach, apoptosis and tumor regression wereneural stem cells could deliver therapeutic molecules with achieved in an intracranial human xenografted gliomaimmunomodulatory function to the site of an infiltrative model after TRAIL administration [87].tumor. To date, encouraging results have been obtainedwith the delivery of IL-4 and IL-12 to experimentalgliomas in rodents [31,70], but other molecules could also 4 . Concluding remarksbe exploited using the same approach.

1There is now compelling evidence that CD8 T-cell3 .3.4. Compatibility and synergy of antitumor immune immune responses occur in the CNS and that at least in

1responses with other therapies experimental models, CD8 T cells can mediate protectiveEfficacious treatment of highly malignant brain tumors roles such as the elimination of tumor cells. However,

is unlikely to be achieved by single modality therapies. It spontaneous antitumor immune responses in humans areis therefore useful to consider the compatibility and inefficient, yet the need for new therapeutic approaches insynergy of other therapies for CNS neoplastic disease with brain tumors such as malignant astrocytoma is particularlyimmunotherapy. Whilst chemotherapy and radiotherapy are urgent. We thus need to understand how antitumor re-poorly efficacious for malignant astrocytoma [233,110], sponses can be safely amplified, taking into account thethey do induce a degree of tumor cell death and thus may fact that the relationship between CNS malignancies andprovide a potentially immunostimulatory in vivo source of the immune system is likely to proceed along differenttumor antigen, provided the immune system has not been lines than in other sites. Indeed, the tumor–host relation-impaired by these treatments. Future protocol design ship is complex: many factors influence the final outcomeshould thus look to ways of appropriate staging of the of tumor growth, arrest or rejection, with the growth ofdifferent treatment elements to improve compatibility. aggressive brain tumors indicating that the balance is inImmune responses in gene therapies are a complex issue. favor of the tumor. We can speculate that this is a result ofOn the positive side, they may be at least partly respon- a series of circumstances that are prominent in the de-sible for so-called ‘bystander effects’, of gene therapy, velopment of human astrocytoma, but which are absent inwhereby a greater number of tumor cells are eliminated many experimental tumors.than are actually transduced by the therapeutic gene Spontaneous human tumors inefficiently prime the im-[163,238]. However, immune responses directed towards mune system, resulting in a late response which is betterviral vectors, particularly early generation adenovirus adapted to minimize inflammation rather than to stimulatevectors, limit repeat utilization of the vector and may also potent CTL activity. Furthermore, at early stages of tumorinduce chronic neuroinflammation [62]. A particularly growth, there is little signaling of ‘danger’ until the tumorpromising compatibility may be found between antian- is well advanced. This is in contrast to certain experimentalgiogenic therapies and immunotherapy [125,187]. Indeed, models in which the immune system may be alerted by theseveral of the cytokines already discussed as immuno- trauma associated with the surgical implantation of tumorstimulatory agents (including interferon-g, IL-4 and IL-12) cells.may also directly or indirectly inhibit tumor angiogenesis. The microenvironment of a large tumor mass that

Other therapies for brain tumors aim to exploit apoptosis initially develops with little immune hindrance may be-pathways by engaging death receptors in malignant cells come predominantly immunosuppressive. The low-level[36]. Fas ligand was one of the first molecules to be antitumor response that is eventually induced is thenproposed and gave some encouraging results in an ex- unable to mediate efficient effector function in the hostileperimental gene therapy of rodent glioma [13]. However, tumor microenvironment in the brain. In contrast, the rapidFas is expressed by so many different cell types, including response induced against certain experimental tumorscells of the immune system [162], normal astrocytes [200] (which in most cases do not liberate high levels ofand hepatocytes [195], that toxicity and immunosuppres- immunosuppressive factors such as TGF-b) is able tosion may limit its clinical application. Even the effects of eliminate such tumors at an earlier stage, when theylocal administration of Fas ligand within the CNS are present a more manageable target for CTL attack.difficult to predict because of the potential of this molecule The putative tumor antigens present in human as-to manifest either pro- or anti-inflammatory effects at trocytoma may be expressed only at a low level, or theydifferent stages of ongoing immune and inflammatory may be inefficiently processed or presented on MHC classresponses [174,261,203]. A more promising and less toxic I molecules. Furthermore, since many of the tumor an-


and B cells in naive CD41 T cell priming and Th1/Th2 celltigens may be overexpressed antigens also present onrestimulation, Eur. J. Immunol. 29 (1999) 2705–2714.normal tissues, the host T cells may be tolerant, or of low

[10] F. Aloisi, F. Ria, L. Adorini, Regulation of T-cell responses by CNSavidity. Future studies with model brain tumor antigens in antigen-presenting cells: different roles for microglia and astrocytes,which there is also some expression on normal host cells Immunol. Today 21 (2000) 141–147.

[11] F. Aloisi, R. De Simone, S. Columba-Cabezas, G. Penna, L.will be of great interest in mimicking this clinical reality.Adorini, Functional maturation of adult mouse resting microglia intoIt may not be necessary to rectify all of the inadequaciesan APC is promoted by granulocyte-macrophage colony-stimulating

of the spontaneous immune response to malignant brain factor and interaction with Th1 cells, J. Immunol. 164 (2000)tumors to achieve some clinical benefit. However, with an 1705–1712.

[12] R. Alon, Encephalitogenic lymphoblast recruitment to resting CNSunderstanding of the basic rules for T-cell immune re-microvasculature: a natural immunosurveillance mechanism?, J.sponses in the brain, rational proposals can be made toClin. Invest. 108 (2001) 517–519.identify key areas in which to modify the neuroim-

[13] B.B. Ambar, K. Frei, U. Malipiero, A.E. Morelli, M.G. Castro, P.R.munological balance. This should be undertaken cautious- Lowenstein, A. Fontana, Treatment of experimental glioma byly, since the particularities of spontaneous immune re- administration of adenoviral vectors expressing Fas ligand, Hum.

Gene Ther. 10 (1999) 1641–1648.sponses in the CNS have presumably evolved to minimize[14] H. Aoki, M. Mizuno, A. Natsume, T. Tsugawa, K. Tsujimura, T.the risks associated with high level inflammatory immune

Takahashi, J. Yoshida, Dendritic cells pulsed with tumor extract-responses. Successful immunotherapy for cerebral malig- cationic liposome complex increase the induction of cytotoxic Tnancies must find this narrow path between immune- lymphocytes in mouse brain tumor, Cancer Immunol. Immunother.

50 (2001) 463–468.mediated protection and immune-mediated pathology.[15] T. Aoki, K. Tashiro, S. Miyatake, T. Kinashi, T. Nakano, Y. Oda, H.

Kikuchi, T. Honjo, Expression of murine interleukin 7 in a murineglioma cell line results in reduced tumorigenicity in vivo, Proc. Natl.Acad. Sci. USA 89 (1992) 3850–3854.A cknowledgements

[16] H. Arai, S.Y. Chan, D.K. Bishop, G.J. Nabel, Inhibition of thealloantibody response by CD95 ligand, Nat. Med. 3 (1997) 843–

The work in our laboratory is partly supported by grants 848.from OncoSuisse, the Association for International Cancer [17] H. Arai, D. Gordon, E.G. Nabel, G.J. Nabel, Gene transfer of Fas

ligand induces tumor regression in vivo, Proc. Natl. Acad. Sci. USAResearch, the Fondation Lionel Perrier and the Ligue94 (1997) 13862–13867.Genevoise Contre le Cancer.

[18] A. Asai, Y. Miyagi, H. Hashimoto, S.H. Lee, K. Mishima, A.Sugiyama, H. Tanaka, T. Mochizuki, T. Yasuda, Y. Kuchino,Modulation of tumor immunogenicity of rat glioma cells by s-mycexpression: eradication of rat gliomas in vivo, Cell Growth Differ. 5R eferences(1994) 1153–1158.

[19] A. Ashkenazi, R.C. Pai, S. Fong, S. Leung, D.A. Lawrence, S.A.[1] A.K. Abbas, A.H. Sharpe, T-cell stimulation: an abundance of B7s, Marsters, C. Blackie, L. Chang, A.E. McMurtrey, A. Hebert, L.

Nat. Med. 5 (1999) 1345–1346. DeForge, I.L. Koumenis, D. Lewis, L. Harris, J. Bussiere, H.[2] K.S. Aboody, A. Brown, N.G. Rainov, K.A. Bower, S. Liu, W. Yang, Koeppen, Z. Shahrokh, R.H. Schwall, Safety and antitumor activity

J.E. Small, U. Herrlinger,V. Ourednik, P.M. Black, X.O. Breakefield, of recombinant soluble Apo2 ligand, J. Clin. Invest. 104 (1999)E.Y. Snyder, Neural stem cells display extensive tropism for 155–162.pathology in adult brain: evidence from intracranial gliomas, Proc. [20] D.M. Ashley, J.H. Sampson, G.E. Archer, S.K. Batra, D.D. Bigner,Natl. Acad. Sci. USA 97 (2000) 12846–12851. L.P. Hale, A genetically modified allogeneic cellular vaccine gener-

[3] P. Adamson, S. Etienne, P.O. Couraud, V. Calder, J. Greenwood, ates MHC class I-restricted cytotoxic responses against tumor-Lymphocyte migration through brain endothelial cell monolayers associated antigens and protects against CNS tumors in vivo, J.involves signaling through endothelial ICAM-1 via a rho-dependent Neuroimmunol. 78 (1997) 34–46.pathway, J. Immunol. 162 (1999) 2964–2973. [21] D.M. Ashley, F.M. Kong, D.D. Bigner, L.P. Hale, Endogenous

[4] Y. Akasaki, T. Kikuchi, S. Homma, T. Abe, D. Kofe, T. Ohno, expression of transforming growth factor beta1 inhibits growth andAntitumor effect of immunizations with fusions of dendritic and tumorigenicity and enhances Fas-mediated apoptosis in a murineglioma cells in a mouse brain tumor model, J. Immunother. 24 high-grade glioma model, Cancer Res. 58 (1998) 302–309.(2001) 106–113. [22] D. Bainbridge, S. Ellis, P. Le Bouteiller, I. Sargent, HLA-G remains

[5] Y. Akasaki, T. Kikuchi, S. Homma, T. Abe, D. Kofe, T. Ohno, a mystery, Trends Immunol. 22 (2001) 548–552.Antitumor effect of immunizations with fusions of dendritic and [23] R. Balabanov, T. Beaumont, P. Dore-Duffy, Role of central nervousglioma cells in a mouse brain tumor model, J. Immunother. 24 system microvascular pericytes in activation of antigen-primed(2001) 106–113. splenic T-lymphocytes, J. Neurosci. Res. 55 (1999) 578–587.

[6] M.L. Albert, B. Sauter, N. Bhardwaj, Dendritic cells acquire antigen [24] J. Banchereau, B. Schuler-Thurner, A.K. Palucka, G. Schuler,from apoptotic cells and induce class-I restricted CTLs, Nature 392 Dendritic cells as vectors for therapy, Cell 106 (2001) 271–274.(1998) 86–89. [25] J. Banchereau, F. Briere, C. Caux, J. Davoust, S. Lebecque, Y.J. Liu,

[7] F. Aloisi, The role of microglia and astrocytes in CNS immune B. Pulendran, K. Palucka, Immunobiology of dendritic cells, Annu.surveillance and immunopathology, Adv. Exp. Med. Biol. 468 Rev. Immunol. 18 (2000) 767–811.(1999) 123–133. [26] J. Banchereau, A.K. Palucka, M. Dhodapkar, S. Burkeholder, N.

[8] F. Aloisi, F. Ria, G. Penna, L. Adorini, Microglia are more efficient Taquet, A. Rolland, S. Taquet, S. Coquery, K.M. Wittkowski, N.than astrocytes in antigen processing and Th1 but not Th2 cell Bhardwaj, L. Pineiro, R. Steinman, J. Fay, Immune and clinical

1activation, J. Immunol. 160 (1998) 4671–4680. responses in patients with metastatic melanoma to CD34[9] F. Aloisi, F. Ria, S. Columba-Cabezas, H. Hess, G. Penna, L. progenitor-derived dendritic cell vaccine, Cancer Res. 61 (2001)

Adorini, Relative efficiency of microglia, astrocytes, dendritic cells 6451–6458.


[27] D. Barba, S.C. Saris, C. Holder, S.A. Rosenberg, E.H. Oldfield, Leridant, T. Hercend, J. Even, P. Kourilsky, F. Triebel, Evidence forIntratumoral LAK cell and interleukin-2 therapy of human gliomas, T-cell clonal expansion in a patient with squamous cell carcinoma ofJ. Neurosurg. 70 (1989) 175–182. the head and neck, Cancer Res. 54 (1994) 1292–1297.

[28] C.F. Barker, R.E. Billingham, Immunologically privileged sites, [48] J.J. Campbell, J. Hedrick, A. Zlotnik, M.A. Siani, D.A. Thompson,Adv. Immunol. 25 (1977) 1–54. E.C. Butcher, Chemokines and the arrest of lymphocytes rolling

under flow conditions, Science 279 (1998) 381–384.[29] D. Bell, J.W. Young, J. Banchereau, Dendritic cells, Adv. Immunol.72 (1999) 255–324. [49] B. Cannella, A.H. Cross, C.S. Raine, Adhesion-related molecules in

the central nervous system. Upregulation correlates with inflamma-[30] S. Benedetti, M.G. Bruzzone, B. Pollo, F. Di Meco, L. Magrassi, B.tory cell influx during relapsing experimental autoimmune en-Pirola, N. Cirenei, M.P. Colombo, G. Finocchiaro, Eradication of ratcephalomyelitis, Lab. Invest. 65 (1991) 23–31.malignant gliomas by retroviral-mediated, in vivo delivery of the

interleukin 4 gene, Cancer Res. 59 (1999) 645–652. [50] M.D. Carrithers, I. Visintin, S.J. Kang, C.A.J. Janeway, Differentialadhesion molecule requirements for immune surveillance and in-[31] S. Benedetti, B. Pirola, B. Pollo, L. Magrassi, M.G. Bruzzone, D.flammatory recruitment, Brain 123 (2000) 1092–1101.Rigamonti, R. Galli, S. Selleri, F. Di Meco, C. De Fraja, A.Vescovi,

E. Cattaneo, G. Finocchiaro, Gene therapy of experimental brain [51] M.J. Carson, C.R. Reilly, J.G. Sutcliffe, D. Lo, Disproportionate1tumors using neural progenitor cells, Nat. Med. 6 (2000) 447–450. recruitment of CD8 T cells into the central nervous system by

professional antigen-presenting cells, Am. J. Pathol. 154 (1999)[32] S.R.M. Bennett, F.R. Carbone, F. Karamelis, R.A. Flavell, J.F.A.P.481–494.Miller, W.R. Heath, Help for cytotoxic-T-cell responses is mediated

by CD40 signalling, Nature 393 (1998) 478–480. [52] J.J. Chen, Y. Sun, G.J. Nabel, Regulation of the proinflammatoryeffects of Fas ligand (CD95L), Science 282 (1998) 1714–1717.[33] A.S. Beutler, M.S. Banck, A. Aguzzi, D. Wedekind, H.J. Hedrich,

Curing rat glioblastoma: immunotherapy or graft rejection?, Science [53] Y.T. Chen, M.J. Scanlan, U. Sahin, O. Tureci, A.O. Gure, S. Tsang,276 (1997) 20–21. B. Williamson, E. Stockert, M. Pfreundschuh, L.J. Old, A testicular

antigen aberrantly expressed in human cancers detected by autolog-[34] D.D. Bigner, O.M. Pitts, C.J. Wikstrand, Induction of lethal ex-ous antibody screening, Proc. Natl. Acad. Sci. USA 94 (1997)perimental allergic encephalomyelitis in nonhuman primates and1914–1918.guinea pigs with human glioblastoma multiforme tissue, J. Neuro-

surg. 55 (1981) 32–42. [54] D.D. Chi, R.E. Merchant, R. Rand, A.J. Conrad, D. Garrison, R.Turner, D.L. Morton, D.S. Hoon, Molecular detection of tumor-[35] S. Bodmer, K. Strommer, K. Frei, C. Siepl, N. de Tribolet, I. Heid,associated antigens shared by human cutaneous melanomas andA. Fontana, Immunosuppression and transforming growth factor-b

gliomas, Am. J. Pathol. 150 (1997) 2143–2152.in glioblastoma. Preferential production of transforming growthfactor-b , J. Immunol. 143 (1989) 3222–3229. [55] A.Y. Christian, M. Barna, Z. Bi, C.S. Reiss, Host immune response2

¨[36] O. Bogler, M. Weller, Apoptosis in gliomas, and its role in their to vesicular stomatitis virus infection of the central nervous systemcurrent and future treatment, Front. Biosci. 7 (2002) e339–e353. in C57BL/6 mice, Viral Immunol. 9 (1996) 195–205.

[37] L.H. Boise, A.J. Minn, P.J. Noel, C.H. June, M.A. Accavitti, T. [56] W.T. Couldwell, V.W. Yong, P. Dore-Duffy, M.S. Freedman, J.P.Lindsten, C.B. Thompson, CD28 costimulation can promote T cell Antel, Production of soluble autocrine inhibitory factors by humansurvival by enhancing the expression of bcl-xl, Immunity 3 (1995) glioma cell lines, J. Neurol. Sci. 110 (1992) 178–185.87–98. [57] H.F. Cserr, P.M. Knopf, Cervical lymphatics, the blood–brain

[38] A.A. Book, K.E. Fielding, N. Kundu, M.A. Wilson, A.M. Fulton, J. barrier and the immunoreactivity of the brain: a new view, Immunol.Laterra, IL-10 gene transfer to intracranial 9L glioma: tumor Today 13 (1992) 507–512.inhibition and cooperation with IL-2, J. Neuroimmunol. 92 (1998) [58] C. De Smet, C. Lurquin, P. Van der Bruggen, E. De Plaen, F.50–59. Brasseur, T. Boon, Sequence and expression pattern of the human

[39] T. Brabb, P. von Dassow, N. Ordonez, B. Schnabel, B. Duke, J. MAGE2 gene, Immunogenetics 39 (1994) 121–129.Goverman, In situ tolerance within the central nervous system as a [59] C. De Smet, S.J. Courtois, I. Faraoni, C. Lurquin, J.P. Szikora, O.mechanism for preventing autoimmunity, J. Exp. Med. 192 (2000) De Backer, T. Boon, Involvement of two Ets binding sites in the871–880. transcriptional activation of the MAGE1 gene, Immunogenetics 42

[40] V.M. Braud, D.S. Allan, A.J. McMichael, Functions of nonclassical (1995) 282–290.MHC and non-MHC-encoded class I molecules, Curr. Opin. Im- [60] W. Debinski, D.M. Gibo, Molecular expression analysis of restric-munol. 11 (1999) 100–108. tive receptor for interleukin 13, a brain tumor-associated cancer /

[41] M.W. Brightman, T.S. Reese, Junctions between intimately apposed testis antigen, Mol. Med. 6 (2000) 440–449.cell membranes in the vertebrate brain, J. Cell Biol. 40 (1969) [61] I. Desbaillets, M. Tada, N. de Tribolet, A.C. Diserens, M.F. Hamou,648–677. E.G. Van Meir, Human astrocytomas and glioblastomas express

[42] R.D. Broadwell, B.J. Baker, P.S. Ebert, W.F. Hickey, Allografts of monocyte chemoattractant protein-1 (MCP-1) in vivo and in vitro,CNS tissue possess a blood–brain barrier: III. Neuropathological, Int. J. Cancer 58 (1994) 240–247.methodological, and immunological considerations, Microsc. Res. [62] R.A. Dewey, G. Morrissey, C.M. Cowsill, D. Stone, F. Bolognani,Tech. 27 (1994) 471–494. N.J. Dodd, T.D. Southgate, D. Klatzmann, H. Lassmann, M.G.

[43] S. Brocke, C. Piercy, L. Steinman, I.L. Weissman, T. Veromaa, Castro, P.R. Lowenstein, Chronic brain inflammation and persistentAntibodies to CD44 and integrin alpha4, but notL-selectin, prevent herpes simplex virus 1 thymidine kinase expression in survivors ofcentral nervous system inflammation and experimental en- syngeneic glioma treated by adenovirus-mediated gene therapy:cephalomyelitis by blocking secondary leukocyte recruitment, Proc. implications for clinical trials, Nat. Med. 5 (1999) 1256–1263.Natl. Acad. Sci. USA 96 (1999) 6896–6901. [63] S. Dhib-Jalbut, C.V. Kufta, M. Flerlage, N. Shimojo, H.F. McFar-

[44] W.H. Brooks, W.R. Markesbery, G.D. Gupta, T.L. Roszman, Rela- land, Adult human glial cells can present target antigens to HLA-tionship of lymphocyte invasion and survival of brain tumor restricted cytotoxic T-cells, J. Neuroimmunol. 29 (1990) 203–211.patients, Ann. Neurol. 4 (1978) 219–224. [64] V.V. Didenko, H.N. Ngo, C. Minchew, D.S. Baskin, Apoptosis of T

[45] E.C. Butcher, L.J. Picker, Lymphocyte homing and homeostasis, lymphocytes invading glioblastomas multiforme: a possible tumorScience 272 (1996) 60–66. defense mechanism, J. Neurosurg. 96 (2002) 580–584.

[46] A. Caignard, M. Guillard, C. Gaudin, B. Escudier, F. Triebel, P.Y. [65] P.-Y. Dietrich, P.R. Walker, V. Schnuriger, P. Saas, M. Guillard, C.Dietrich, In situ demonstration of renal cell carcinoma specific Gaudin, A. Caignard, T cell repertoire analysis reveals significantT-cell clones, Int. J. Cancer 66 (1996) 564–570. repertoire selection during in vitro lymphocyte culture, Int. Im-

[47] A. Caignard, P.Y. Dietrich, V. Morand, A. Lim, C. Pannetier, A.M. munol. 9 (1997) 1073–1083.


[66] M.C. Dieu, B. Vanbervliet, A. Vicari, J.M. Bridon, E. Oldham, S. [83] A. Fontana, F. Kristensen, R. Dubs, D. Gemsa, E. Weber, ProductionAit-Yahia, F. Briere, A. Zlotnik, S. Lebecque, C. Caux, Selective of prostaglandin E and an interleukin-1 like factor by culturedrecruitment of immature and mature dendritic cells by distinct astrocytes and C6 glioma cells, J. Immunol. 129 (1982) 2413–2419.chemokines expressed in different anatomic sites, J. Exp. Med. 188 [84] A.L. Ford, A.L. Goodsall, W.F. Hickey, J.D. Sedgwick, Normal(1998) 373–386. adult ramified microglia separated from other central nervous system

[67] A.R. Dix, W.H. Brooks, T.L. Roszman, L.A. Morford, Immune macrophages by flow cytometric sorting. Phenotypic differencesdefects observed in patients with primary malignant brain tumors, J. defined and direct ex vivo antigen presentation to myelin basicNeuroimmunol. 100 (1999) 216–232. protein-reactive CD41 T cells compared, J. Immunol. 154 (1995)

4309–4321.[68] M.E. Dudley, J.R. Wunderlich, P.F. Robbins, J.C. Yang, P. Hwu, D.J.Schwartzentruber, S.L. Topalian, R. Sherry, N.P. Restifo, A.M. [85] A.L. Ford, E. Foulcher, F.A. Lemckert, J.D. Sedgwick, MicrogliaHubicki, M.R. Robinson, M. Raffeld, P. Duray, C.A. Seipp, L. induce CD4 T lymphocyte final effector function and death, J. Exp.Rogers-Freezer, K.E. Morton, S.A. Mavroukakis, D.E. White, S.A. Med. 184 (1996) 1737–1745.Rosenberg, Cancer regression and autoimmunity in patients after [86] B. Frankel, S.L. Longo, T.C. Ryken, Human astrocytomas co-clonal repopulation with antitumor lymphocytes, Science 298 (2002) expressing Fas and Fas ligand also produce TGFb2 and Bcl-2, J.850–854. Neurooncol. 44 (1999) 205–212.

[69] M.A. Eglitis, E. Mezey, Hematopoietic cells differentiate into both [87] S. Fulda, W. Wick, M. Weller, K.M. Debatin, Smac agonists sensitizemicroglia and macroglia in the brains of adult mice, Proc. Natl. for Apo2L/T, Nat. Med. 8 (2002) 808–815.Acad. Sci. USA 94 (1997) 4080–4085. [88] K.C. Garcia, L. Teyton, I.A. Wilson, Structural basis of T cell

[70] M. Ehtesham, P. Kabos, A. Kabosova, T. Neuman, K.L. Black, J.S. recognition, Annu. Rev. Immunol 17 (1999) 369–397.Yu, The use of interleukin 12-secreting neural stem cells for the [89] K.M. Giezeman-Smits, H. Okada, C.S. Brissette-Storkus, L.A. Villa,treatment of intracranial glioma, Cancer Res. 62 (2002) 5657–5663. J. Attanucci, M.T. Lotze, I.F. Pollack, M.E. Bozik, W.H. Chambers,

[71] L.H. Elliott, W.H. Brooks, T.L. Roszman, Activation of immuno- Cytokine gene therapy of gliomas: induction of reactive CD41 Tregulatory lymphocytes obtained from patients with malignant cells by interleukin-4-transfected 9L gliosarcoma is essential forgliomas, J. Neurosurg. 67 (1987) 231–236. protective immunity, Cancer Res. 60 (2000) 2449–2457.

[72] B. Engelhardt, The role of alpha 4-integrin in T lymphocyte [90] R.P. Glick, T. Lichtor, T.S. Kim, S. Ilangovan, E.P. Cohen,migration into the inflamed and noninflamed central nervous system, Fibroblasts genetically engineered to secrete cytokines suppressCurr. Top. Microbiol. Immunol. 231 (1998) 51–64. tumor growth and induce antitumor immunity to a murine glioma in

[73] J. Even, A. Lim, I. Puisieux, L. Ferradini, P.Y. Dietrich, A. Toubert, vivo, Neurosurgery 36 (1995) 548–555.T. Hercend, F. Triebel, C. Pannetier, P. Kourilsky, T-cell repertoires [91] L.B. Gordon, S.C. Nolan, H.F. Cserr, P.M. Knopf, C.J. Harling-in healthy and diseased human tissues analysed by T-cell receptor Berg, Growth of P511 mastocytoma cells in BALB/c mouse brainbeta-chain CDR3 size determination: evidence for oligoclonal elicits CTL response without tumour elimination, J. Immunol. 159expansions in tumours and inflammatory diseases, Res. Immunol. (1997) 2399–2408.146 (1995) 65–80. [92] L.B. Gordon, S.C. Nolan, B.R. Ksander, P.M. Knopf, C.J. Harling-

[74] Z. Fabry, C.S. Raine, M.N. Hart, Nervous tissue as an immune Berg, Normal cerebrospinal fluid suppresses the in vitro develop-compartment: the dialect of the immune response in the CNS, ment of cytotoxic T cells: role of the brain microenvironment inImmunol. Today 15 (1994) 218–224. CNS immune regulation, J. Neuroimmunol. 88 (1998) 77–84.

[75] Z. Fabry, M.M. Waldschmidt, D. Hendrickson, J. Keiner, L. Love- [93] L. Gorelik, R.A. Flavell, Transforming growth factor-b in T-cellHoman, F. Takei, M.N. Hart, Adhesion molecules on murine brain biology, Nat. Rev. Immunol. 2 (2002) 46–53.microvascular endothelial cells: expression and regulation of ICAM- [94] M.R. Graf, M.R. Jadus, J.C. Hiserodt, H.T. Wepsic, G.A. Granger,1 and Lgp 55, J. Neuroimmunol. 36 (1992) 1–11. Development of systemic immunity to glioblastoma multiforme

[76] H. Fakhrai, O. Dorigo, D.L. Shawler, H. Lin, D. Mercola, K.L. using tumor cells genetically engineered to express the membrane-Black, I. Royston, R.E. Sobol, Eradication of established intracranial associated isoform of macrophage colony-stimulating factor, J.rat gliomas by transforming growth factorb antisense gene therapy, Immunol. 163 (1999) 5544–5551.Proc. Natl. Acad. Sci. USA 93 (1996) 2909–2914. [95] A. Grakoui, S.K. Bromley, C. Sumen, M.M. Davis, A.S. Shaw, P.M.

[77] F. Farace, F. Orlanducci, P.Y. Dietrich, C. Gaudin, E. Angevin, M.H. Allen, M.L. Dustin, The immunological synapse: a molecularCourtier, C. Bayle, T. Hercend, F. Triebel, T cell repertoire in machine controlling T cell activation, Science 285 (1999) 221–227.patients with B chronic lymphocytic leukemia. Evidence for multi- [96] C. Gratas, Y. Tohma, E.G. Van Meir, M. Klein, M. Tenan, N. Ishii,ple in vivo T cell clonal expansions, J. Immunol. 153 (1994) O. Tachibana, P. Kleihues, H. Ohgaki, Fas ligand expression in4281–4290. glioblastoma cell lines and primary astrocytic brain tumors, Brain

[78] D.E. Faunce, J. Stein-Streilein, NKT cell-derived RANTES recruits Pathol. 7 (1997) 863–869.APCs and CD81 T cells to the spleen during the generation of [97] M.D. Gunn, K. Tangemann, C. Tam, J.G. Cyster, S.D. Rosen, L.T.regulatory T cells in tolerance, J. Immunol. 169 (2002) 31–38. Williams, A chemokine expressed in lymphoid high endothelial

[79] N. Favre, B. Bonnotte, N. Droin, A. Fromentin, E. Solary, F. Martin, venules promotes the adhesion and chemotaxis of naive T lympho-Fas (CD95) ligand expression by tumor cell variants can be cytes, Proc. Natl. Acad. Sci. USA 95 (1998) 258–263.unrelated to their capacity to induce tolerance or immune rejection, [98] M. Gunzer, A. Schafer, S. Borgmann, S. Grabbe, K.S. Zanker, E.B.Int. J. Cancer 82 (1999) 359–367. Brocker, E. Kampgen, P. Friedl, Antigen presentation in extracellu-

[80] H.G. Fischer, U. Bonifas, G. Reichmann, Phenotype and functions lar matrix: interactions of T cells with dendritic cells are dynamic,of brain dendritic cells emerging during chronic infection of mice short lived, and sequential, Immunity 13 (2000) 323–332.

¨with toxoplasma gondii, J. Immunol. 164 (2000) 4826–4834. [99] M. Hahne, D. Rimoldi, M. Schroter, P. Romero, M. Schreier, L.[81] S. Flynn, K.M. Toellner, C. Raykundalia, M. Goodall, P. Lane, CD4 French, P. Schneider, T. Bornand, A. Fontana, D. Lienard, J.-C.

T cell cytokine differentiation: the B cell activation molecule, OX40 Cerottini, J. Tschopp, Melanoma cell expression of Fas (Apo-1/ligand, instructs CD4 T cells to express interleukin 4 and upregu- CD95) ligand: Implications for tumor immune escape, Science 274lates expression of the chemokine receptor, Blr-1, J. Exp. Med. 188 (1996) 1363–1366.(1998) 297–304. [100] C. Harling-Berg, P.M. Knopf, J. Merriam, H.F. Cserr, Role of

[82] A. Fontana, H. Hengartner, N. de Tribolet, E. Weber, Glioblastoma cervical lymph nodes in the systemic humoral immune response tocells release interleukin 1 and factors inhibiting interleukin 2- human serum albumin microinfused into rat cerebrospinal fluid, J.mediated effects, J. Immunol. 132 (1984) 1837–1844. Neuroimmunol. 25 (1989) 185–193.


[101] D.N. Hart, Dendritic cells: unique leukocyte populations which [121] T. Joki, T. Kikuchi, Y. Akasaki, S. Saitoh, T. Abe, T. Ohno,control the primary immune response, Blood 90 (1997) 3245– Induction of effective antitumor immunity in a mouse brain tumor3287. model using B7-1 (CD80) and intercellular adhesive molecule 1

[102] S.L. Hauser, A.K. Bhan, F.H. Gilles, C.J. Hoban, E.L. Reinherz, (ICAM-1; CD54) transfection and recombinant interleukin 12, Int.S.F. Schlossman, H.L. Weiner, Immunohistochemical staining of J. Cancer 82 (1999) 714–720.human brain with monoclonal antibodies that identify lymphocytes, [122] W.J. Karpus, N.W. Lukacs, B.L. McRae, R.M. Strieter, S.L.monocytes, and the Ia antigen, J. Neuroimmunol. 5 (1983) 197– Kunkel, S.D. Miller, An important role for the chemokine macro-205. phage inflammatory protein-1 alpha in the pathogenesis of the T

[103] A.B. Heimberger, L.E. Crotty, G.E. Archer, R.E. McLendon, A. cell-mediated autoimmune disease, experimental autoimmune en-Friedman, G. Dranoff, D.D. Bigner, J.H. Sampson, Bone marrow- cephalomyelitis, J. Immunol. 155 (1995) 5003–5010.derived dendritic cells pulsed with tumor homogenate induce [123] Y. Katz-Levy, K.L. Neville, A.M. Girvin, C.L. Vanderlugt, J.G.immunity against syngeneic intracerebral glioma, J. Neuroim- Pope, L.J. Tan, S.D. Miller, Endogenous presentation of selfmunol. 103 (2000) 16–25. myelin epitopes by CNS-resident APCs in Theiler’s virus-infected

[104] J. Held-Feindt, R. Mentlein, CD70/CD27 ligand, a member of the mice, J. Clin. Invest. 104 (1999) 599–610.TNF family, is expressed in human brain tumors, Int. J. Cancer 98 [124] J. Kaye, S.M. Hedrick, Analysis of specificity for antigen, Mls, and(2002) 352–356.

allogenic MHC by transfer of T-cell receptora- andb-chain genes,[105] W.F. Hickey, Leukocyte traffic in the central nervous system: the

Nature 336 (1988) 580–583.participants and their roles, Semin. Immunol. 11 (1999) 125–137.

[125] R. Kerbel, J. Folkman, Clinical translation of angiogenesis in-[106] W.F. Hickey, B.L. Hsu, H. Kimura, T-lymphocyte entry into the

hibitors, Nat. Rev. Cancer 2 (2002) 727–739.central nervous system, J. Neurosci. Res. 28 (1991) 254–260.[126] T. Kikuchi, Y. Akasaki, T. Joki, T. Abe, M. Kurimoto, T. Ohno,[107] M. Hishii, T. Nitta, H. Ishida, M. Ebato, A. Kurosu, H. Yagita, K.

Antitumor activity of interleukin-18 on mouse glioma cells, J.Sato, K. Okumura, Human glioma-derived Interleukin-10 inhibitsImmunother. 23 (2000) 184–189.antitumor immune responses in vitro, Neurosurgery 37 (1995)

[127] T. Kikuchi, T. Joki, S. Saitoh, Y. Hata, T. Abe, N. Kato, A.1160–1167.Kobayashi, T. Miyazaki, T. Ohno, Anti-tumor activity of[108] E.C. Holland, Gliomagenesis: genetic alterations and mouseinterleukin-2-producing tumor cells and recombinant interleukin 12models, Nat. Rev. Genet. 2 (2001) 120–129.against mouse glioma cells located in the central nervous system,[109] M.S. Horwitz, C.F. Evans, F.G. Klier, M.B. Oldstone, Detailed inInt. J. Cancer 80 (1999) 425–430.vivo analysis of interferon-gamma induced major histocompatibili-

[128] T. Kikuchi, Y. Akasaki, M. Irie, S. Homma, T. Abe, T. Ohno,ty complex expression in the central nervous system: astrocytes failResults of a phase I clinical trial of vaccination of glioma patientsto express major histocompatibility complex class I and II mole-with fusions of dendritic and glioma cells, Cancer Immunol.cules, Lab. Invest. 79 (1999) 235–242.Immunother. 50 (2001) 337–344.[110] P. Hosli, A.P. Sappino, N. de Tribolet, P.Y. Dietrich, Malignant

glioma: should chemotherapy be overthrown by experimental [129] T. Kikuchi, Y. Akasaki, T. Abe, T. Ohno, Intratumoral injection oftreatments?, Ann. Oncol. 9 (1998) 589–600. dendritic and irradiated glioma cells induces anti-tumor effects in a

[111] A.A. Hurwitz, E.D. Kwon, A. van Elsas, Costimulatory wars: the mouse brain tumor model, Cancer Immunol. Immunother. 51tumor menace, Curr. Opin. Immunol. 12 (2000) 589–596. (2002) 424–430.

[112] G. Iezzi, K. Karjalainen, A. Lanzavecchia, The duration of [130] Y.J. Kim, S.H. Kim, P. Mantel, B.S. Kwon, Human 4-1BBantigenic stimulation determines the fate of naive and effector T regulates CD28 co-stimulation to promote Th1 cell responses, Eur.cells, Immunity 8 (1998) 89–95. J. Immunol. 28 (1998) 881–890.

[113] T. Imaizumi, T. Kuramoto, K. Matsunaga, S. Shichijo, S. Yutani, [131] J. Klein, A. Sato, The HLA system. First of two parts, N. Engl. J.M. Shigemori, K. Oizumi, K. Itoh, Expression of the tumor- Med. 343 (2000) 702–709.rejection antigen SART1 in brain tumors, Int. J. Cancer 83 (1999) [132] P.M. Knopf, C.J. Harling-Berg, H.F. Cserr, D. Basu, E.J. Sirulnick,760–764. S.C. Nolan, J.T. Park, G. Keir, E.J. Thompson, W.F. Hickey,

[114] K. Inaba, M. Pack, M. Inaba, H. Sakuta, F. Isdell, R.M. Steinman, Antigen-dependent intrathecal antibody synthesis in the normal ratHigh levels of a major histocompatibility complex II-self peptide brain: tissue entry and local retention of antigen-specific B cells, J.complex on dendritic cells from the T cell areas of lymph nodes, J. Immunol. 161 (1998) 692–701.Exp. Med. 186 (1997) 665–672. [133] A.M. Krakowski, T. Owens, Naive T lymphocytes traffic to

[115] T.H. Inge, K.M. McCoy, B.M. Susskind, S.K. Barrett, G. Zhao, inflamed central nervous system, but require antigen recognitionH.D. Bear, Immunomodulatory effects of transforming growth for activation, Eur. J. Immunol. 30 (2000) 1002–1009.factor-beta on T lymphocytes. Induction of CD8 expression in the [134] W. Krivit, J.H. Sung, E.G. Shapiro, L. Lockman, Microglia: theCTLL-2 cell line and in normal thymocytes, J. Immunol. 148 effector cell for reconstitution of the central nervous system(1992) 3847–3856. following bone marrow transplantation for lysosomal and perox-

[116] D.N. Irani, D.E. Griffin, Regulation of lymphocyte homing into the isomal storage diseases, Cell Transplant. 4 (1995) 385–392.brain during viral encephalitis at various stages of infection, J. [135] V. Kumar, Determinant spreading during experimental autoimmuneImmunol. 156 (1996) 3850–3857. encephalomyelitis: is it potentiating, protecting or participating in

[117] P. Jachimczak, U. Bogdahn, J. Schneider, C. Behl, J. Meix- the disease?, Immunol. Rev. 164 (1998) 73–80.ensberger, R. Apfel, R. Dorries, K.H. Schlingensiepen, W. Brysch, [136] L.A. Lampson, W.F. Hickey, Monoclonal antibody analysis ofThe effect of transforming growth factor-b2-specific phosphoro- MHC expression in human brain biopsies: tissue ranging fromthioate-anti-sense oligodeoxynucleotides in reversing cellular im- ‘histologically normal’ to that showing different levels of glialmunosuppression in malignant glioma, J. Neurosurg. 78 (1993) tumor involvement, J. Immunol. 136 (1986) 4054–4062.944–951. [137] A. Lanzavecchia, Immunology. Licence to kill, Nature 393 (1998)

[118] S. Jacquot, CD27/CD70 interactions regulate T dependent B cell 413–414.differentiation, Immunol. Res. 21 (2000) 23–30. [138] A. Lanzavecchia, F. Sallusto, Dynamics of T lymphocyte re-

[119] R.C. Janzer, M.C. Raff, Astrocytes induce blood–brain barrier sponses: intermediates, effectors, and memory cells, Science 290properties in endothelial cells, Nature 325 (1987) 253–257. (2000) 92–97.

[120] K.W. Jarosinski, P.T. Massa, Interferon regulatory factor-1 is [139] L.J. Lawson,V.H. Perry, S. Gordon, Turnover of resident microgliarequired for interferon-gamma-induced MHC class I genes in in the normal adult mouse brain, Neuroscience 48 (1992) 405–415.astrocytes, J. Neuroimmunol. 122 (2002) 74–84. [140] M. Lenardo, F.K.M. Chan, F. Hornung, H. McFarland, R. Siegel, J.


Wang, L. Zheng, Mature T lymphocyte apoptosis-immune regula- [157] T.R. Mosmann, T lymphocyte subsets, cytokines, and effectortion in a dynamic and unpredictable antigenic environment, Annu. functions, Ann. NY Acad. Sci. 664 (1992) 89–92.Rev. Immunol. 17 (1999) 221–253. [158] T.R. Mosmann, L. Li, S. Sad, Functions of CD8 T-cell subsets

[141] L.M. Liau, K.L. Black, R.M. Prins, S.N. Sykes, P.L. DiPatre, T.F. secreting different cytokine patterns, Semin. Immunol. 9 (1997)Cloughesy, D.P. Becker, J.M. Bronstein, Treatment of intracranial 87–92.gliomas with bone marrow-derived dendritic cells pulsed with [159] P.A. Moss, S.L. Rowland-Jones, P.M. Frodsham, S. McAdam, P.tumor antigens, J. Neurosurg. 90 (1999) 1115–1124. Giangrande, A.J. McMichael, J.I. Bell, Persistent high frequency of

[142] L.M. Liau, E.R. Jensen, T.J. Kremen, S.K. Odesa, S.N. Sykes, human immunodeficiency virus-specific cytotoxic T cells inM.C. Soung, J.F. Miller, J.M. Bronstein, Tumor immunity within peripheral blood of infected donors, Proc. Natl. Acad. Sci. USA 92the central nervous system stimulated by recombinant Listeria (1995) 5773–5777.monocytogenes vaccination, Cancer Res. 62 (2002) 2287–2293. ¨[160] C. Munz, U. Naumann, C. Grimmel, H.G. Rammensee, M. Weller,

[143] A. Limmer, J. Ohl, C. Kurts, H.G. Ljunggren, Y. Reiss, M. TGF-b-independent induction of immunogenicity by decorin geneGroettrup, F. Momburg, B. Arnold, P.A. Knolle, Efficient pre- transfer in human malignant glioma cells, Eur. J. Immunol. 29

1sentation of exogenous antigen by liver endothelial cells to CD8 (1999) 1032–1040.T cells results in antigen-specific T-cell tolerance, Nat. Med. 6 [161] P. Musette, J.F. Bureau, G. Gachelin, P. Kourilsky, M. Brahic, T(2000) 1348–1354. lymphocyte repertoire in Theiler’s virus encephalomyelitis: the

[144] B. Ludewig, A.F. Ochsenbein, B. Odermatt, D. Paulin, H. Hengar- nonspecific infiltration of the central nervous system of infectedtner, R.M. Zinkernagel, Immunotherapy with dendritic cells di- SJL/J mice is associated with a selective local T cell expansion,rected against tumor antigens shared with normal host cells results Eur. J. Immunol. 25 (1995) 1589–1593.in severe autoimmune disease, J. Exp. Med. 191 (2000) 795–804. [162] S. Nagata, P. Golstein, The Fas death factor, Science 267 (1995)

[145] K. Lumniczky, S. Desaknai, L. Mangel, B. Szende, H. Hamada, 1449–1456.E.J. Hidvegi, G. Safrany, Local tumor irradiation augments the [163] D. Nanda, M.J. Driesse, P.A. Sillevis Smitt, Clinical trials ofantitumor effect of cytokine-producing autologous cancer cell adenoviral-mediated suicide gene therapy of malignant gliomas,vaccines in a murine glioma model, Cancer Gene Ther. 9 (2002) Prog. Brain Res. 132 (2001) 699–710.

1 144–52. [164] Y.N. Naumov, E.N. Naumova, J. Gorski, CD4 and CD8[146] D.H. Lynch, F. Ramsdell, M.R. Alderson, Fas and FasL in the circulating alpha/beta T-cell repertoires are equally complex and

homeostatic regulation of immune responses, Immunol. Today 16 are characterized by different levels of steady-state TCR expres-(1995) 569–574. sion, Hum. Immunol. 48 (1996) 52–62.

[147] A. Mackensen, G. Carcelain, S. Viel, M.C. Raynal, H. Michalaki, [165] V.N. Ngo, H.L. Tang, J.G. Cyster, Epstein–Barr virus-inducedF. Triebel, J. Bosq, T. Hercend, Direct evidence to support the molecule 1 ligand chemokine is expressed by dendritic cells inimmunosurveillance concept in a human regressive melanoma, J. lymphoid tissues and strongly attracts naive T cells and activated BClin. Invest. 93 (1994) 1397–1402. cells, J. Exp. Med. 188 (1998) 181–191.

[148] P.T. Massa, K. Ozato, D.E. McFarlin, Cell type-specific regulation [166] R. Nitsch, I. Bechmann, R.A. Deisz, D. Haas, T.N. Lehmann, U.of major histocompatibility complex (MHC) class I gene expres- Wendling, F. Zipp, Human brain-cell death induced by tumour-sion in astrocytes, oligodendrocytes, and neurons, Glia 8 (1993) necrosis-factor-related apoptosis-inducing ligand (TRAIL), Lancet201–207. 356 (2000) 827–828.

[149] M.K. Matyszak, S. Denis-Donini, S. Citterio, R. Longhi, F. [167] T. Nitta, M. Hishii, K. Sato, K. Okumura, Selective expression ofGranucci, P. Ricciardi-Castagnoli, Microglia induce myelin basic interleukin-10 gene within glioblastoma multiforme, Brain Res.protein-specific T cell anergy or T cell activation, according to 649 (1994) 122–128.their state of activation, Eur. J. Immunol. 29 (1999) 3063–3076. [168] J. O’Connell, Fas ligand and the fate of antitumour cytotoxic T

[150] M.G. McHeyzer-Williams, J.D. Altman, M.M. Davis, Tracking lymphocytes, Immunology 105 (2002) 263–266.antigen-specific helper T cell responses, Curr. Opin. Immunol. 8 [169] M. Oehmichen, H. Wietholter, H. Gruninger, M. Gencic, Destruc-(1996) 278–284. tion of intracerebrally applied red blood cells in cervical lymph

[151] P.B. Medewar, Immunity to homologous grafted skin. III. The fate nodes. Experimental investigations, Forensic Sci. Int. 21 (1983)of skin homografts transplanted to the brain, to subcutaneous tissue 43–57.and to the anterior chamber of the eye, Br. J. Exp. Pathol. 29 [170] R. Offringa, S.H. van der Burg, F. Ossendorp, R.E. Toes, C.J.(1948) 58–69. Melief, Design and evaluation of antigen-specific vaccination

[152] R. Medzhitov, C.J. Janeway, Innate immunity, N. Engl. J. Med. strategies against cancer, Curr. Opin. Immunol. 12 (2000) 576–343 (2000) 338–344. 582.

[153] A. Merlo, A. Juretic, M. Zuber, L. Filgueira, U. Luscher, V. [171] H. Okada, H. Tahara, M.R. Shurin, J. Attanucci, K.M. Giezeman-Caetano, J. Ulrich, O. Gratzl, M. Heberer, G. Spagnoli, Cytokine Smits, W.K. Fellows, M.T. Lotze, W.H. Chambers, M.E. Bozik,gene expression in primary brain tumours, metastases and mening- Bone marrow-derived dendritic cells pulsed with a tumor-specificiomas suggests specific transcription patterns, Eur. J. Cancer 29A peptide elicit effective anti-tumor immunity against intracranial(1993) 2118–2125. neoplasms, Int. J. Cancer 78 (1998) 196–201.

[154] S. Miescher, T.L. Whiteside, S. Carrel, V. Von Fliedner, Functional [172] H. Okada, I.F. Pollack, M.T. Lotze, L.D. Lunsford, D. Kondziolka,properties of tumor-infiltrating and blood lymphocytes in patients F. Lieberman, D. Schiff, J. Attanucci, H. Edington, W. Chambers,with solid tumors: effects of tumor cells and their supernatants on P. Robbins, J. Baar, D. Kinzler, T. Whiteside, E. Elder, Geneproliferative responses of lymphocytes, J. Immunol. 136 (1986) therapy of malignant gliomas: a phase I study of IL-4-HSV-TK1899–1907. gene-modified autologous tumor to elicit an immune response,

[155] S. Miyatake, H. Kikuchi, K. Iwasaki, J. Yamashita, Z.Y. Li, Y. Hum. Gene Ther. 11 (2000) 637–653.Namba, M. Hanaoka, Specific cytotoxic activity of T lymphocyte [173] F. Okano, W.J. Storkus, W.H. Chambers, I.F. Pollack, H. Okada,clones derived from a patient with gliosarcoma, J. Neurosurg. 69 Identification of a novel HLA-A*0201-restricted, cytotoxic T(1988) 751–759. lymphocyte epitope in a human glioma-associated antigen, inter-

[156] L.A. Morford, L.H. Elliott, S.L. Carlson, W.H. Brooks, T.L. leukin 13 receptor alpha2 chain, Clin. Cancer Res. 8 (2002)Roszman, T cell receptor-mediated signaling is defective in T cells 2851–2855.obtained from patients with primary intracranial tumors, J. Im- [174] Y. Okuda, S. Sakoda, H. Fujimura, S. Nagata, T. Yanagihara, C.C.munol. 159 (1997) 4415–4425. Bernard, Intrathecal administration of neutralizing antibody against


1Fas ligand suppresses the progression of experimental autoimmune can be a temporal bridge between a CD4 T-helper and a T-killerencephalomyelitis, Biochem. Biophys. Res. Commun. 275 (2000) cell, Nature 393 (1998) 474–478.164–168. [195] I. Rodriguez, K. Matsuura, K. Khatib, J.C. Reed, S. Nagata, P.

[175] B.A. Osborne, Apoptosis and maintenance of homeostasis in the Vassalli, A bcl-2 transgene expressed in hepatocytes protects miceimmune system, Curr. Opin. Immunol. 8 (1996) 245–254. from fulminant liver destruction but not from rapid death induced

by anti-fas antibody injection, J. Exp. Med. 183 (1996) 1031–[176] C. Pannetier, J. Even, P. Kourilsky, T-cell repertoire diversity and1036.clonal expansions in normal and clinical samples, Immunol. Today

16 (1995) 176–181. [196] M.L. Rossi, J.T. Hughes, M.M. Esiri, H.B. Coakham, D. Brownell,Immunohistological study of mononuclear cell infiltrate in malig-[177] D.M. Pardoll, Inducing autoimmune disease to treat cancer, Proc.nant gliomas, Acta Neuropathol. 74 (1987) 269–277.Natl. Acad. Sci. USA 96 (1999) 5340–5342.

[197] J.F. Rowell, D.E. Griffin, The inflammatory response to nonfatal[178] D.M. Pardoll, Spinning molecular immunology into successfulSindbis virus infection of the nervous system is more severe in SJLimmunotherapy, Nat. Rev. Immunol. 2 (2002) 227–238.than in BALB/c mice and is associated with low levels of IL-4[179] W.M. Pardridge, J. Yang, J. Buciak, W.W. Tourtellotte, HumanmRNA and high levels of IL-10-producing CD41 T cells, J.brain microvascular DR-antigen, J. Neurosci. Res. 23 (1989) 337–Immunol. 162 (1999) 1624–1632.341.

[198] L.L. Rubin, J.M. Staddon, The cell biology of the blood–brain[180] G. Perrin, V. Schnuriger, A.-L. Quiquerez, P. Saas, C. Pannetier, N.barrier, Annu. Rev. Neurosci. 22 (1999) 11–28.de Tribolet, J.M. Tiercy, J.P. Aubry, P.-Y. Dietrich, P.R. Walker,

[199] P. Saas, J. Boucraut, P.R. Walker, A.L. Quiquerez, M. Billot, S.Astrocytoma infiltrating lymphocytes include major T cell clonalDesplat-Jego, Y. Chicheportiche, P.Y. Dietrich, TWEAK stimulationexpansions confined to the CD8 subset, Int. Immunol. 11 (1999)of astrocytes and the proinflammatory consequences, Glia 321337–1349.(2000) 102–107.[181] V.H. Perry, M.K. Matyszak, S. Fearn, Altered antigen expression of

[200] P. Saas, J. Boucraut, A.-L. Quiquerez, V. Schnuriger, G. Perrin, S.microglia in the aged rodent CNS, Glia 7 (1993) 60–67.Desplat-Jego, D. Bernard, P.R. Walker, P.-Y. Dietrich, CD95 (Fas/[182] L. Pewe, S.B. Heard, C. Bergmann, M.O. Dailey, S. Perlman,Apo-1) as a receptor governing astrocyte apoptotic or inflammatorySelection of CTL escape mutants in mice infected with a neuro-responses: a key role in brain inflammation?, J. Immunol. 162tropic coronavirus: quantitative estimate of TCR diversity in the(1999) 2326–2333.infected central nervous system, J. Immunol. 163 (1999) 6106–

[201] P. Saas, P.R. Walker, A.-L. Quiquerez, D. Chalmers, J.F. Arrighi,6113.´A. Lienard, J. Boucraut, P.-Y. Dietrich, A self-defence mechanism[183] M.J. Phillips, M. Needham, R.O. Weller, Role of cervical lymph

of astrocytes against Fas-mediated death involving interleukin-8nodes in autoimmune encephalomyelitis in the Lewis rat, J. Pathol.and CXCR2, Neuroreport 13 (2002) 1921–1924.182 (1997) 457–464.

[184] G.E. Plautz, G.H. Barnett, D.W. Miller, B.H. Cohen, R.A. Prayson, [202] P. Saas, P.R. Walker, M. Hahne, A.-L. Quiquerez, V. Schnuriger, G.J.C. Krauss, M. Luciano, D.B. Kangisser, S. Shu, Systemic T cell Perrin, L. French, E.G. Van Meir, N. de Tribolet, J. Tschopp, P.Y.adoptive immunotherapy of malignant gliomas, J. Neurosurg. 89 Dietrich, Fas Ligand expression by astrocytoma in vivo: maintain-(1998) 51. ing immune privilege in the brain?, J. Clin. Invest. 99 (1997)

[185] Z. Qing, D. Sewell, M. Sandor, Z. Fabry, Antigen-specific T cell 1173–1178.trafficking into the central nervous system, J. Neuroimmunol. 105 [203] K.A. Sabelko-Downes, A.H. Cross, J.H. russel, Dual role for Fas(2000) 169–178. ligand in the initiation of and recovery from experimental allergic

[186] K.B. Quattrocchi, C.H. Miller, S. Cush, S.A. Bernard, S.T. Dull, encephalomyelitis, J. Exp. Med. 189 (1999) 1195–1205.M. Smith, S. Gudeman, M.A. Varia, Pilot study of local autologous [204] U. Sahin, M. Koslowski, O. Tureci, T. Eberle, C. Zwick, B.tumor infiltrating lymphocytes for the treatment of recurrent Romeike, J.R. Moringlane, K. Schwechheimer, W. Feiden, M.malignant gliomas, J. Neurooncol. 45 (1999) 141–157. Pfreundschuh, Expression of cancer testis genes in human brain

[187] S. Rafii, D. Lyden, R. Benezra, K. Hattori, B. Heissig,Vascular and tumors, Clin. Cancer Res. 6 (2000) 3916–3922.haematopoietic stem cells: novel targets for anti-angiogenesis [205] F. Sallusto, A. Lanzavecchia, C.R. Mackay, Chemokines andtherapy?, Nat. Rev. Cancer 2 (2002) 826–835. chemokine receptors in T-cell priming and Th1/Th2-mediated

[188] F. Re, S.L. Belyanskaya, R.J. Riese, B. Cipriani, F.R. Fischer, F. responses, Immunol. Today 19 (1998) 568–574.Granucci, P. Ricciardi-Castagnoli, C. Brosnan, L.J. Stern, J.L. [206] F. Sallusto, P. Schaerli, P. Loetscher, C. Schaniel, D. Lenig, C.R.Strominger, L. Santambrogio, Granulocyte-macrophage colony- Mackay, S. Qin, A. Lanzavecchia, Rapid and coordinated switch instimulating factor induces an expression program in neonatal chemokine receptor expression during dendritic cell maturation,microglia that primes them for antigen presentation, J. Immunol. Eur. J. Immunol. 28 (1998) 2760–2769.169 (2002) 2264–2273. [207] J.H. Sampson, G.E. Archer, D.M. Ashley, H.E. Fuchs, L.P. Hale,

1[189] S. Read, F. Powrie, CD4 regulatory T cells, Curr. Opin. Immunol. G. Dranoff, D.D. Bigner, Subcutaneous vaccination with irradiated,113 (2001) 644–649. cytokine-producing tumors cells stimulates CD8 cell-mediated

[190] J.M. Redwine, M.J. Buchmeier, C.F. Evans, In vivo expression of immunity against tumors located in the ‘immunologicallymajor histocompatibility complex molecules on oligodendrocytes privileged’ central nervous system, Proc. Natl. Acad. Sci. USA 93and neurons during viral infection, Am. J. Pathol. 159 (2001) (1996) 10399–10404.1219–1224. [208] J.H. Sampson, D.M. Ashley, G.E. Archer, H.E. Fuchs, G. Dranoff,

[191] N. Renkvist, C. Castelli, P.F. Robbins, G. Parmiani, A listing of L.P. Hale, D.D. Bigner, Characterization of a spontaneous murinehuman tumor antigens recognized by T cells, Cancer Immunol. astrocytoma and abrogation of its tumorigenicity by cytokineImmunother. 50 (2001) 3–15. secretion, Neurosurgery 41 (1997) 1365–1373.

[192] M. Resnicoff, C. Sell, M. Rubini, D. Coppola, D. Ambrose, R. [209] L. Santambrogio, S.L. Belyanskaya, F.R. Fischer, B. Cipriani, C.F.Baserga, R. Rubin, Rat glioblastoma cells expressing an antisense Brosnan, P. Ricciardi-Castagnoli, L.J. Stern, J.L. Strominger, R.RNA to the insulin-like growth factor-1 (IGF-1) receptor are Riese, Developmental plasticity of CNS microglia, Proc. Natl.nontumorigenic and induce regression of wild-type tumors, Cancer Acad. Sci. USA 98 (2001) 6295–6300.Res. 54 (1994) 2218–2222. [210] J. Satoh, Y.B. Lee, S.U. Kim, T-cell costimulatory molecules B7-1

[193] N.P. Restifo, Not so fas: Re-evaluating the mechanisms of immune (CD80) and B7-2 (CD86) are expressed in human microglia butprivilege and tumor escape, Nat. Med. 6 (2000) 493–495. not in astrocytes in culture, Brain Res. 704 (1995) 92–96.

[194] J.P. Ridge, F. Di Rosa, P. Matzinger, A conditioned dendritic cell [211] Y. Sawamura, A.C. Diserens, N. de Tribolet, In vitro prostaglandin


E2 production by glioblastoma cells and its effect on interleukin-2 [230] D. Stavrou, A.P. Anzil, W. Weidenbach, H. Rodt, Immunofluores-cence study of lymphocytic infiltration in gliomas. Identification ofactivation of oncolytic lymphocytes, J. Neurooncol. 9 (1990) 125–T-lymphocytes, J. Neurol. Sci. 33 (1977) 275–282.130.

[231] R.M. Steinman, M. Pope, Exploiting dendritic cells to improve[212] D.L. Scarcella, C.W. Chow, M.F. Gonzales, C. Economou, F.vaccine efficacy, J. Clin. Invest. 109 (2002) 1519–1526.Brasseur, D.M. Ashley, Expression of MAGE and GAGE in

[232] L.A. Stewart, Chemotherapy in adult high-grade glioma: a sys-high-grade brain tumors: a potential target for specific immuno-tematic review and meta-analysis of individual patient data from 12therapy and diagnostic markers, Clin. Cancer Res. 5 (1999) 335–randomised trials, Lancet 359 (2002) 1011–1018.341.

[233] R. Stupp, P.Y. Dietrich, K.S. Ostermann, A. Pica, I. Maillard, P.[213] J. Schneider, F.M. Hofman, M.L. Apuzzo, D.R. Hinton, CytokinesMaeder, R. Meuli, R. Janzer, G. Pizzolato, R. Miralbell, F. Porchet,and immunoregulatory molecules in malignant glial neoplasms, J.L. Regli, N. de Tribolet, R.O. Mirimanoff, S. Leyvraz, PromisingNeurosurg. 77 (1992) 265–273.survival for patients with newly diagnosed glioblastoma mul-[214] S.P. Schoenberger, R.E.M. Toes, E.I.H. Van der Voort, R. Offringa,tiforme treated with concomitant radiation plus temozolomideC.J.M. Melief, T-cell help for cytotoxic T lymphocytes is mediatedfollowed by adjuvant temozolomide, J. Clin. Oncol. 20 (2002)by CD40-CD40L interactions, Nature 393 (1998) 480–483.1375–1382.

[215] R. Schwab, P. Szabo, J.S. Manavalan, M.E. Weksler, D.N. Posnett,[234] T. Suter, U. Malipiero, L. Otten, B. Ludewig, A. Muelethaler-1 1C. Pannetier, P. Kourilsky, J. Even, Expanded CD4 and CD8 T

Mottet, B. Mach, W. Reith, A. Fontana, Dendritic cells andcell clones in elderly humans, J. Immunol. 158 (1997) 4493–4499.

differential usage of the MHC class II transactivator promoters in[216] J.D. Sedgwick, A.L. Ford, E. Foulcher, R. Airriess, Central nervous the central nervous system in experimental autoimmune en-

system microglial cell activation and proliferation follows direct cephalitis, Eur. J. Immunol. 30 (2000) 794–802.interaction with tissue-infiltrating T cell blasts, J. Immunol. 160 [235] E.H. Tran, K. Hoekstra, N. Van Rooijen, C.D. Dijkstra, T. Owens,(1998) 5320–5330. Immune invasion of the central nervous system parenchyma and

[217] B.M. Segal, D.D. Glass, E.M. Shevach, Cutting Edge: IL-10- experimental allergic encephalomyelitis, but not leukocyte extrava-producing CD41 T cells mediate tumor rejection, J. Immunol. 168 sation from blood, are prevented in macrophage-depleted mice, J.(2002) 1–4. Immunol. 161 (1998) 3767–3775.

[218] K.-I. Seino, N. Kayagaki, K. Okumura, H. Yagita, Antitumor effect [236] J. Trojan, T.R. Johnson, S.D. Rudin, J. Ilan, M.L. Tykocinski,of locally produced CD95 ligand, Nat. Med. 3 (1997) 165–170. Treatment and prevention of rat glioblastoma by immunogenic C6

[219] K.-I. Seino, N. Kayagaki, N. Tsukada, K. Fukao, H. Yagita, K. cells expressing antisense insulin-like growth factor I RNA,Okumura, Transplantation of CD95 ligand-expressing grafts. In- Science 259 (1993) 94–97.fluence of transplantation site and difficulty in protecting allo- and [237] R.A. Tuma, E.G. Pamer, Homeostasis of naive, effector andxenografts, Transplantation 64 (1997) 1050–1054. memory CD8 T cells, Curr. Opin. Immunol. 14 (2002) 348–353.

[220] M. Sensi, G. Parmiani, Analysis of TCR usage in human tumors: a [238] P. Tunici, D. Gianni, G. Finocchiaro, Gene therapy of glioblas-new tool for assessing tumor-specific immune responses, Immunol. tomas: from suicide to homicide, Prog. Brain Res. 132 (2001)Today 16 (1995) 588–595. 711–719.

[221] P. Shrikant, E.N. Benveniste, The central nervous system as an [239] P. Vajkoczy, M. Laschinger, B. Engelhardt,a4-integrin-VCAM-1immunocompetent organ. Role of glial cells in antigen pre- binding mediates G protein-independent capture of en-sentation, J. Immunol. 157 (1996) 1819–1822. cephalitogenic T cell blasts to CNS white matter microvessels, J.

[222] W.W. Shuford, K. Klussman, D.D. Tritchler, D.T. Loo, J. Chalupny, Clin. Invest. 108 (2001) 557–565.A.W. Siadak, T.J. Brown, J. Emswiler, H. Raecho, C.P. Larsen, [240] E.G. Van Meir, Cytokines and tumors of the central nervousT.C. Pearson, J.A. Ledbetter, A. Aruffo, R.S. Mittler, 4-1BB system, Glia 15 (1995) 264–288.

1costimulatory signals preferentially induce CD8 T cell prolifer- [241] P.R. Walker, P. Saas, P.-Y. Dietrich, The role of Fas ligand (CD95L)ation and lead to the amplification in vivo of cytotoxic T cell in immune escape: the tumor cell strikes back, J. Immunol. 158responses, J. Exp. Med. 186 (1997) 47–55. (1997) 4521–4524.

[223] J.M. Slavik, J.E. Hutchcroft, B.E. Bierer, CD28/CTLA-4 and [242] P.R. Walker, P. Saas, P.Y. Dietrich, Tumor expression of Fas ligandCD80/CD86 families: signaling and function, Immunol. Res. 19 (CD95L) and the consequences, Curr. Opin. Immunol. 10 (1998)(1999) 1–24. 564–572.

[224] M.J. Smyth, S.L. Strobl, H.A. Young, J.R. Ortaldo, A.C. Ochoa, [243] P.R. Walker, T. Calzascia, V. Schnuriger, N. Scamuffa, P. Saas, N.Regulation of lymphokine-activated killer activity and pore-form- de Tribolet, P.Y. Dietrich, The brain parenchyma is permissive for

1ing protein gene expression in human peripheral blood CD8 T full antitumor CTL effector function, even in the absence of CD4 Tlymphocytes. Inhibition by transforming growth factor-b, J. Im- cells, J. Immunol. 165 (2000) 3128–3135.munol. 146 (1991) 3289–3297. [244] R.F. Wang, S.A. Rosenberg, Human tumor antigens for cancer

[225] A. Soling, N.G. Rainov, Dendritic cell therapy of primary brain vaccine development, Immunol. Rev. 170 (1999) 85–100.tumors, Mol. Med. 7 (2001) 659–667. [245] F. Weber, E. Meinl, F. Aloisi, C. Nevinny-Stickel, E. Albert, H.

[226] J.M. Soos, J. Morrow, T.A. Ashley, B.E. Szente, E.K. Bikoff, S.S. Wekerle, R. Hohlfeld, Human astrocytes are only partially compe-Zamvil, Astrocytes express elements of the class II endocytic tent antigen presenting cells. Possible implications for lesionpathway and process central nervous system autoantigen for development in multiple sclerosis, Brain 117 (1994) 59–69.presentation to encephalitogenic T cells, J. Immunol. 161 (1998) [246] H. Wekerle, C. Linington, H. Lassmann, R. Meyermann, Cellular5959–5966. immune reactivity within the CNS, Trends Neurosci. 9 (1986)

[227] J. Sprent, C.D. Surh, T cell memory, Annu. Rev. Immunol. 20 271–277.(2002) 551–579. [247] R.O. Weller, Pathology of cerebrospinal fluid and interstitial fluid

[228] T.A. Springer, Traffic signals for lymphocyte recirculation and of the CNS: significance for Alzheimer disease, prion disorders andleukocyte emigration: the multistep paradigm, Cell 76 (1994) multiple sclerosis, J. Neuropathol. Exp. Neurol. 57 (1998) 885–301–314. 894.

[229] M. Stander, U. Naumann, L. Dumitrescu, M. Heneka, P. Losch- [248] H. Wiendl, M. Mitsdoerffer, V. Hofmeister, J. Wischhusen, A.mann, E. Gulbins, J. Dichgans, M. Weller, Decorin gene transfer- Bornemann, R. Meyermann, E.H. Weiss, A. Melms, M. Weller, Amediated suppression of TGF-beta synthesis abrogates experimen- functional role of HLA-G expression in human gliomas: antal malignant glioma growth in vivo, Gene Ther. 5 (1998) 1187– alternative strategy of immune escape, J. Immunol. 168 (2002)1194. 4772–4780.


[249] C.J. Wikstrand, P. Fredman, L. Svennerholm, D.D. Bigner, De- models in mice by genetically modified dendritic cells pulsed withtection of glioma-associated gangliosides GM2, GD2, GD3, 39- Semliki forest virus-mediated complementary DNA, J. Neurosurg.isoLM1 39,69-isoLD1 in central nervous system tumors in vitro and 94 (2001) 474–481.in vivo using epitope-defined monoclonal antibodies, Prog. Brain [258] R. Yamanaka, S.A. Zullo, J. Ramsey, N. Yajima, N. Tsuchiya, R.Res. 101 (1994) 213–223. Tanaka, M. Blaese, K.G. Xanthopoulos, Marked enhancement of

[250] K. Williams, E. Ulvestad, J.P. Antel, B7/BB-1 antigen expression antitumor immune responses in mouse brain tumor models byon adult human microglia studied in vitro and in situ, Eur. J. genetically modified dendritic cells producing Semliki Forest virus-Immunol. 24 (1994) 3031–3037. mediated interleukin-12, J. Neurosurg. 97 (2002) 611–618.

[251] K. Williams, A. Bar-Or, E. Ulvestad, A. Olivier, J.P. Antel, V.W. [259] J.S. Yu, C.J. Wheeler, P.M. Zeltzer, H. Ying, D.N. Finger, P.K. Lee,Yong, Biology of adult human microglia in culture: comparisons W.H. Yong, F. Incardona, R.C. Thompson, M.S. Riedinger, W.with peripheral blood monocytes and astrocytes, J. Neuropathol. Zhang, R.M. Prins, K.L. Black, Vaccination of malignant gliomaExp. Neurol. 51 (1992) 538–549. patients with peptide-pulsed dendritic cells elicits systemic cytotox-

[252] J. Wischhusen, G. Jung, I. Radovanovic, C. Beier, J.P. Steinbach, icity and intracranial T-cell infiltration, Cancer Res. 61 (2001)A. Rimner, H. Huang, J.B. Schulz, H. Ohgaki, A. Aguzzi, H.G. 842–847.Rammensee, M. Weller, Identification of CD70-mediated apoptosis [260] P.M. Zeltzer, B. Moilanen, J.S. Yu, K.L. Black, Immunotherapy ofof immune effector cells as a novel immune escape pathway of malignant brain tumors in children and adults: from theoreticalhuman glioblastoma, Cancer Res. 62 (2002) 2592–2599. principles to clinical application, Child’s Nerv. Syst. 15 (1999)

[253] T.F. Witham, M.L. Erff, H. Okada, W.H. Chambers, I.F. Pollack, 514–528.7-Hydroxystaurosporine-induced apoptosis in 9L glioma cells [261] B. Zhu, L. Luo, Y. Chen, D.W. Paty, M.S. Cynader, Intrathecal Fasprovides an effective antigen source for dendritic cells and yields a ligand infusion strengthens immunoprivilege of central nervouspotent vaccine strategy in an intracranial glioma model, Neuro- system and suppresses experimental autoimmune en-surgery 50 (2002) 1327–1334. cephalomyelitis, J. Immunol. 169 (2002) 1561–1569.

[254] G.W. Wood, F.P. Holladay, T. Turner, Y.Y. Wang, M. Chiga, A pilot [262] A. van Elsas, R.P. Sutmuller, A.A. Hurwitz, J. Ziskin, J.Villasenor,study of autologous cancer cell vaccination and cellular immuno- J.P. Medema, W.W. Overwijk, N.P. Restifo, C.J. Melief, R. Offr-therapy using anti-CD3 stimulated lymphocytes in patients with inga, J.P. Allison, Elucidating the autoimmune and antitumorrecurrent grade III / IV astrocytoma, J. Neurooncol. 48 (2000) effector mechanisms of a treatment based on cytotoxic T lympho-113–120. cyte antigen-4 blockade in combination with a B16 melanoma

[255] G.F. Wu, A.A. Dandekar, L. Pewe, S. Perlman, CD4 and CD8 T vaccine. Comparison of prophylaxis and therapy, J. Exp. Med. 194cells have redundant but not identical roles in virus-induced (2001) 481–490.demyelination, J. Immunol. 165 (2000) 2278–2286. [263] C.M. van Herpen, P.H. De Mulder, Locoregional immunotherapy

[256] K.W. Wucherpfennig, Autoimmunity in the central nervous system: in cancer patients: review of clinical studies, Ann. Oncol. 11mechanisms of antigen presentation and recognition, Clin. Im- (2000) 1229–1239.munol. Immunopathol. 72 (1994) 293–306. [264] U.H. von Andrian, C.R. Mackay, T-cell function and migration.

[257] R. Yamanaka, S.A. Zullo, R. Tanaka, M. Blaese, K.G. Xanth- Two sides of the same coin, N. Engl. J. Med. 343 (2000) 1020–opoulos, Enhancement of antitumor immune response in glioma 1034.

Review T -cell immune responses in the brain and their ...

Documents

Transcript of Review T -cell immune responses in the brain and their ...