TINS-June 1986 271 Cellular immune reactivity within the CNS H. Wekerle, C. Llnington, H. Lassmann and R. Meyermann We have used autoaggressive rat T lymphocyte lines specific for defined protein components of peripheral or central myelin to study lymphocyte migration and antigen recognition within the nervous system. Our recent data suggest that the nervous system is constantly patrolled by low numbers of activated T cells. Such T cells penetrate the blood-brain barrier non-specifically and recognize 'their' antigen within the parenchyme on the surface of Ia-inducible glia cells, e.g. astrocytes or Schwann cells. The immune system ensures the body's integrity in the face of constant exposure to environmental and endo- genous pathogens. Lymphocytes per- manently patrol the body's tissues ~'2 and are exposed to a vast array of molecular structures. However, they will only recognize specific antigens and only when these are presented in the context of major histocompatibility complex (MHC) antigens on the membrane of specialized antigen pre- senting cells (APCs) 3. The majority of T lymphocytes (including most of the T helper cells) recognize antigen in the context of, or associated with class II MHC antigens (Ia determinants), whereas a second, smaller T cell subset, which includes most of the T killer cells, requires the presence of class I MHC antigens (Fig. 1). Almost all tissues contain strategic- ally distributed APCs that constitu- tively express MHC antigens and play a primary role in the initiation of an immune response 4"6. In addition, a number of cell types that are normally immunologically inert can be induced to act as APCs by certain stimuli, such as y-interferon 7-9. These inducible or facultative APCs seem to be essential regulatory elements in the local ampli- fication and termination of cellular immune responses. The concept of immune surveillance has been confirmed for almost all vertebrate tissues with the remarkable exception of the nervous system. Indeed, the nervous system has long been considered to be an immuno- logically privileged site, inaccessible to circulating lymphocytes 1°. A number of anatomical and functional obser- vations have been quoted to support this conceptl]-]3: (1) there is no complete lymphatic drainage of the nervous system; (2) the blood-brain barrier (BBB) provides a physical barrier to the entry of both pathogens and immune factors (cellular as well as humoral); (3) rejection of foreign tissue grafts into the CNS has often been major role in the initiation and subsequent regulation of intracerebral immune response. incomplete14; and (4) expression of MHC antigens within the CNS is minimal 15-17. In particular, dendritic cells expressing Ia antigens are not found in the CNS TM. The immune status of the CNS has been re-evaluated in order to learn more about the cellular autoimmune reactions that are suspected of playing a key role in the pathogenesis of multiple sclerosis 19. The availability of T cell lines specific for nervous system antigens 5, has paved the way to study directly T lymphocyte/CNS inter- actions. Such experiments have led to the surprising conclusions that the CNS is routinely surveyed by activated T lymphocytes that can cross the BBB and that astrocytes appear to play a Autoaggressive T lymphocyte lines The key to studying CNS/T-cell interactions has been the development of autoaggressive T cell lines and clones specific for defined nervous system antigens. Such T cells are capable of recognizing their appro- priate antigen within the nervous system and then triggering a cascade of events resulting in tissue specific inflammation. Two models of nervous system T-cell-mediated autoimmune disease have been established and rigorously characterized - these are experimental autoimmune encephalo- myelitis (EAE) and experimental auto- immune neuritis (EAN). In EAE, an inflammatory response directed pri- marily against the CNS is induced by inoculation of susceptible animals with homogenates or extracts of CNS myelin. Similarly, in EAN, auto- immune PNS inflammation follows Antigen ~" Fig. 1. Antigen recognition by T lymphocyte subsets. T lymphocytes recognize antigen only when it is presented on the membrane of suitable antigen presenting cells (APC). T cells of one major subset (CD4) are characterized by the human monoclonal marker T4 (mouse L3T4, rat W3/25) and contain the majority of the helper T lymphocytes; they recognize their antigen (0) only in association with class H histo- ' compatibility molecules ( la determinants) on the A PC. T cells of the complementary subset ( C Dg) defined by the marker T8 (mouse L yt 2, rat oxg), contain suppressor and killer cells and recognize antigen in the context of class I histocompatibility antigens. These histocompatibility antigens are polymorphic cell surface glycoproteins, Following recognition of the antigen, originally resting T cells are activated. Activation involves a drastic increase in cellular metabolism, induction of T cell mediators and receptors for growth factors, and rapid cell division, leading to the formation of large lymphoblasts. ~) 1986, Elsevier Science Publishers B.V., Amsterdam 0378 - 5912/86/$02.00
`
`Page 1 of 7
`
`YEDA EXHIBIT NO. 2074
`MYLAN PHARM. v YEDA
`IPR2015-00643
`
`

`
`272inoculation with PNS myelin in adju-vant. In both models the autoantigensinvolved have been isolated andcharacterized: myelin basic protein(MBP) and P2 protein in EAE andEAN, respectively. Both moleculeshave been sequenced20S21, and recentlythe gene for MBP has been cloned22.Both EAE and EAN are T-cell-mediated and populations of auto-antigen-ache (MBP or P2) T ceilstransfer the appropriate disease tonaive,syngeneic animals2sT”. Theselection and maintenance of perma-nent auto~tigen specific T cell linesand clones has become possible due torecent progress in the characterizationof immune ceil subsets using mono-clonal antibodies and the definition oftheir growth requirements=. Auto-aggressive T 41s can now be isolatedfrom animals with autoimmune dis-eases (i.e. EAE or EAN) andpropagated as pure, monospecific celllines. These cells will mediate the samedisease in naive recipients, and can besubsequently isolated again from theirsecondary hosts. The methods rely onthe sequential application of strongpositive selection pressures to selectthe antigen specific T cells from themultitude of irrelevant lymphocytes inthe immune system. In the initialselection step,
`with theappropriate antigen. Only antigen-specific T cells recognize the immuno-gen and respond by blast transform-ation and proliferation. The enlargedlymphoblasts may then be isolatedfrom the non-responding, irrelevantlymphocytes, by density gradientcentrifugation. They are further propa-gated in media containing the T cellgrowth factor. interleukin-2 (IL-2).This provides a second selection step,as activated T cells express high levelsof IL-2 receptors26 on their plasmamembrane and therefore proliferate inresponse to this lymphokine. On theother hand, resting lymphocytes ex-press little or no IL-2 receptor andeither die out or proliferate veryslowly. A stable, monospecific T linecan therefore be established by con-tinuously alternating antigen-specificactivation of the T cells (by stimulatingthem with antigen presented by suit-able APC) with IL-2 dependent propa-gation.All rat T cell lines isolated in our andother laboratories share two funda-mental characteristics: their membranephenotype and their mode of MHCrestriction. Using a panel of mono-clonal antibodies (mABs) defining T
`
`in-vivo
`in vitro
`primed lympho-cytes are incubated
`
`TINS-fune 1936
`
`ceil differentiation markers and subse-quent ~ytofluoromet~c analysis, it hasbeen shown that the T line cells bindmABs W3113 and W3125 but do notstain with 0X8. This defines them asmembers of the T subset which includesmost of the T helper celis, and whoseantigen recognition is restricted byclass II, la determinants of the MHC.Indeed, the Ia restriction of these lineswas directly demonstrated by screeningrecombinant and congenic APCs fortheir presenting capacity in T lineproliferation tests. In addition, mABsdirected against Ia epitopes were foundto block antigen presentation bysyngeneic APCs.Using MBP as the target antigen, thistechnique allows the selection andestablishment of highly encephalito-genie T cell lines that can induce alethal inflammatory response specificfor the CNS 4-5 days after anintravenous injection of 1 X 10’ acti-vated cells. In a similar manner, T celllines specific for PNS-derived P,protein will induce an antoimmunedisease which follows the same timecourse, but is completely restricted tothe PNS.Target specifkity of autoimmune Tlines -a property pe withinthe spe&k T IymphocytesThe exquisite target specificity ofautoimmune T cell lines has beenrevealed by comparing T lines recog-nizing MBP isolated from CNS myelinwith T lines specific for the P2 proteinisolated from bovine peripheral nerve
`
`Page 2 of 7
`
`YEDA EXHIBIT NO. 2074
`MYLAN PHARM. v YEDA
`IPR2015-00643
`
`

`
`TINS-June 1986 273 myelin. In the rat, MBP is present in both CNS and PNS myelin, whereas P2 protein is localized exclusively in PNS myelin 27'2s. Interestingly, systematic neuropathological studies demon- strated that MBP-specific T lines home preferentially to the CNS29'3°; in this model only exceptionally have we observed inflammatory lesions in peri- pheral nerves remote from the spinal roots. Conversely, P2-specific T line cells are absolutely specific for PNS tissue. After a time lag of about 3 days following inoculation, peripheral nerves are rapidly infiltrated by immune-cells 31. This cellular invasion of the PNS is reflected by a complete loss of neural conduction that develops within a period of 18 h (Ref. 32). In contrast, CNS tissue adjacent to the spinal roots is left completely free of infiltrating cells 33. This remarkable capacity of the T lines to distinguish between closely related tissues is a function program- med within the reacting T lymphoeytes. Early experiments by Ben-Nun et al. have shown that activation is a prerequisiste for autoaggression by T cells; only activated, but not resting T line cells can mediate autoimmune disease (Ref. 29; H. Schltisener, PhD thesis 1984, Univ. of Wiirzburg). After activation, doses of MBP-specific T line cells as low as 104 cells / adult recipient can mediate clinically overt, though reversible EAE whilst lethal EAE is transferred by 106 or more cells. In contrast, resting, unactivated T line ceils are incapable of transferring EAE, even in doses exceeding 20 x 106 cells / adult recipient. Interestingly, the mode of T cell activation is not relevant for the induction of autoimmme disease. T cells activated by mitogens are just as autoaggressive as those activated by the appropriate auto- antigen 29'31. In addition, this indicates that co-transfer of antigen with the T line cells is irrelevant to the patho- genesis of these disease models. The ability of MBP-specific T lines to initiate CNS lesions in normal animals raises two fundamental questions: (1) what is the mechanism by which these encephalitogenic T cells leave the vasculature and enter the CNS, and (2) which cell type(s) can present MBP in the context of Ia antigen in order to further activate the T cells once they are within the CNS? It is apparent that at some point the T cells must cross the structural complex of endothelial cells, pericytes, astrocytic endfeet and associated basal lamina that forms that BBB (Fig. 2) in order to gain access to the CNS and that logically any of these structures could provide appropriate signals to T ceils. The blood-brain interface- barrier or avenue? In experimental models of auto- immune CNS disease, T lymphocytes are activated outside the CNS target tissue, either around the site of antigen inoculation in actively induced EAE or in vitro in the case of T cell transfer models. As the T ceils reach their target via the blood stream, the interface between the circulation and nervous system must play a role in determining the induction of autoimmune CNS disease. The vascular interface in the CNS is formed by a complex cellular system, which principally consists of concentric layers of endothelial cells, pericytes, basal membranes and tightly apposed astrocyte endfeet 34. These structures interact to form a physio- logical barrier which allows controlled exchange of cells and substances between blood and nervous system. Endothelia in the CNS differ from other vascular endothelia in several ways: they are tightly interconnected by an intricate network of tight junctionsaS'36; they are relatively de- void of pinocytotic vesicles and fene- strations; and they express specific enzyme and receptor markers 37'3s. Although the endothelial layer com- pletely surrounds the lumen and is primarily responsible for the barrier function, it should however be noted that the surrounding cells such as pericytes and astrocytes also play a role in the selective exchange of macro- molecules 39 (Fig. 3). Until recently it was not possible to determine the relevance of this physio- logical barrier to the pathogenesis of EAE. Is the BBB just a physical obstacle that has to be overcome before autoaggressive T cells enter the CNS? Or, could the BBB play a more active role in the process of CNS inflammation? The availability of encephalitogenic T cell lines has now. enabled us to approach this question directly. Indeed, our present studies have revealed an unexpected role for Fig. 3. Migration of lymphoid cells through CNS vascular wall. Ultrathin section, 3 days after i.v. injection of lO 6 L.BP cells. Bar: 1 ~m.
`
`Page 3 of 7
`
`YEDA EXHIBIT NO. 2074
`MYLAN PHARM. v YEDA
`IPR2015-00643
`
`

`
`274 TINS-June1986 the BBB in the initiation of inflamma- tory responses within the CNS. The approach taken was to follow histo- logically the sequence of events that occur following the direct application of activated, encephalitogenic T cells into the CNS. Activated MBP specific T cells (105 / rat) were injected intrathexally into naive Lewis rats thereby circumventing the BBB. Although control animals that received the same dose of cells via the normal i.v. route all developed both clinical and histological signs of EAE, no clinical signs of disease were noted in the group injected intra- thecally. Detailed morphological analysis of the events occurring in this latter group indicated that the intra- thecally injected T cells, rather than directly invading the CNS parenchyma, initially migrated out of the lepto- meningeal space via the leptomenin- geal vessels and so entered the blood stream. Only after having entered the circulation did some of the T line cells appear to cross the BBB to initiate small, clinically silent perivascular cuffs of inflammatory cells scattered throughout the CNS. It is at present an open question why these infiltrates failed to produce any clinical signs of EAE. It is possible that on their voyage through the meningeal space and blood the number of activated T cells was reduced and had become insufficient to cause overt clinical EAE. Conversely, the state of activation of the migrating cells may have declined to such a level as to be insufficient to cause overt neurological defects in the CNS parenchyme. Homing of autoimmune T cells- a signal of specificity of the endothelial surface? The observations that target specifi- city is a function of the T cells and that contact with an intact BBB is appar- ently required to mediate full EAE could imply that target recognition is signalled by the BBB. Indeed, several researchers have proposed the possi- bility that expression of tissue specific autoantigen occurs on the vascular surface of the endothelium 4°-42. In this case, it is implied that a circulating, activated T lymphocyte would recog- nize its autoantigen on the endothelial surface via the specific antigen receptor to trigger a local inflammatory re- sponse. If this is so, two postulates must be fulfilled: (1) the autoantigen must be expressed on normal endothelium (T cell mediated EAE is transferred to healthy animals); (2) the autoantigen must be expressed in an immunogenic state, i.e. in the context of Ia antigens (all encephalitogenic T lines are Ia- restricted). At the present time neither MBP nor Ia determinants have been convin- cingly demonstrated on the normal CNS endothelial inner surface. Al- though MBP-like molecules have been identified in normal serum 43, MBP itself has not been demonstrated on endothelial membranes of normal brains using immunocytochemical methods, although, in principle MBP can be transported from the CNS to the luminal surface of cerebral endothe- lial cells, after injection into the CNS parenchyme 44. Furthermore, the question of Ia expression on rodent brain endothelium is controversial despite the fact that Ia can be induced on human endothelium 45. A number of groups have claimed la expression on endothelia of normal and EAE rodent brains 46'47, So far, however, all these positive reports relied on frozen section preparations, which do not easily allow unequivocal resolution of the ex- ceedingly thin microvascular endo- thelia from surrounding pericytes or inflammatory cells. In fact, more recent studies employing immune electron microscopy with highly sensi- tive peroxidase and immunogold label- ling techniques strongly argue against Ia on brain endothelium. In these preparations, all the Ia immuno- reactivity was concentrated on the pericytes, perivascular mononuclear cells and infiltrating lymphocytes, whereas the luminal face of the endothelia was completely negative 48. These immunohistological results are relevant for the interpretation of in- vitro studies, which have suggested that mouse CNS microvascular cells can be induced by ConA supernatants to present antigen to primed syngeneic T Fig. 4. Meningeal infiltration 48 h after intrathecal injection of 105 L. B P cells. Many inflammatory cells adhere to or migrate through vessel wall (arrows). Plastic section, stained with toluidine blue, ×1000.
`
`Page 4 of 7
`
`YEDA EXHIBIT NO. 2074
`MYLAN PHARM. v YEDA
`IPR2015-00643
`
`

`
`TINS-June1986 275 cell populations 49. Another, though circumstantial, argument against 'sim- ple' immune recognition of MBP/ Ia complexes on endothelium is the fact that only activated, but not resting T cells home to the target tissue. Resting cells efficiently recognize their antigen whenever presented on an Ia + APC. Nevertheless, even when injected in high numbers, they never infiltrate the CNS and cause disease (Ref. 29; H. Schliisener, PhD thesis 1984, Univ. of Wtirzburg). All these lines of evidence strongly argue against the possibility that MPB is expressed on thenormal endothelial surface in the appropriate immuno- logical context to act as the target epitope for the antigen receptor of circulating MBP specific T cells. How then do these cells infiltrate the CNS to initiate a target specific response? Recently data from two distinct experi- mental approaches have provided a partial answer to this question. They may implicate an essential, but passive role for CNS endothelium in the induction of EAE. Two reports have documented that activated, but non- resting T lymphocytes have the capacity to non-specifically attach to endothelial cells and subsequently penetrate the endothelial barrier s°'51 (Fig. 4). Interestingly, in this case, activated T cells behave like activated macrophages and metastasizing tumor cells s2. The observation that antigen specific rodent T cells interact with xenogeneic bovine aortic endothelia rules out the possibility that either immune specific recognition, or a CNS specific mechanism are involved in this interaction 52. It appears that during their activation, T lymphoblasts ac- quire a set of enzymes and receptors, among them endoglycosidases, which help them to degrade basal membrane constituents, and possibly to attach to the endothelial surface and to open tight junctions 51. It could therefore be envisaged that endothelia are unspecifically pene- trated by activated T blasts. This implies, however, two essential predic- tions: (1) not only CNS-specific T blasts, but any activated T cell, irrespective of its actual antigen specificity should be expected to cross the BBB; and (2) in EAE the signal for tissue specificity should be found beyond the BBB. Very recently, in our laboratory, we labelled in-vitro MBP and, as a control, ovalbumin-specific T blasts with 14C- thymidine and followed their migration patterns after i.v. injection. We found Fig. 5. Antigen presentation by astrocytes. Aggregation of MBP-specific T line cells around syngeneic astrocytes in the presence of M B P. Astrocytes identified by staining of the G FA P-containing cytoplasmic filaments; cell nuclei stained with propidium iodide.( Courtesy of W. Fierz.) by autoradiography that both specific as well as non-specific T cell blasts invaded the CNS within 2-6 h of the injection (R. Meyermann et al., un- published observations). The number of ovalbumin-specific blasts in the CNS then decreased over the next few days, whereas MBP-blasts (CNS specific) showed a biphasic immigration pat- tern. Three days after the initial wave of MBP specific cells appearing in the CNS there was a second wave of infiltration, which coincided with the non-specific inflammation and peri- vascular cuff formation typical of acute EAE 53. These data are in line with observations obtained in actively in- duced EAE (C. H. Brunner, K. Vass and H. Lassmann, unpublished obser- vations). Also in this situation we observed a slight increase of T lymphocytes within the CNS during the preclinical stage of disease, accom- panied by Ia expression of local cells in the CNS parenchyme. This early stage is then followed by a second wave of inflammatory cells, entering the CNS tissue at the time of onset of neurological signs. Since immune reactions and T cell activation occur throughout the life of an individual, T cells should therefore be found in any normal CNS tissue. Indeed, using lymphocyte differenti- ation markers and immunocytochemi- cal techniques, low but consistent numbers of both major T cell subsets are demonstrable in normal human or rodent brains 54'55. It is of obvious interest to determine the state of activation and the migration pattern of these T cells infiltrating the human CNS. This evidence therefore depicts a situation in which the CNS is constantly patrolled by activated T cells that non- specifically cross the BBB. Astrocytes-the facultative antigen presenter cells of the CNS? In the CNS of normal rodents only few cells constitutively expressing Ia antigens are found 48'55. These cells are mainly located in the leptomeninges (the pia-arachnoideal membranes). Extremely rare Ia + cells are located in the perivascular connective tissue space of the CNS (in average 1 cell mm -2 of CNS area), and none are found within the CNS parenchyme. As outlined above, the interface between blood circulation CNS paren- chyme is formed by a complex assembly of concentrically arranged cellular elements. The vascular endo- thelial monolayer forming the actual physical barrier is surrounded by pericytes, and beyond a basal mem- brane by a tight network of astrocytic endfeet 56. Interestingly, data from cell culture experiments suggest that astro- cytes can influence endothelial differ- entiation and induce certain enzymes on endothelial monolayer cells 57. Astrocyte processes are surrounded by highly specialized membranes with a high density of orthogonal particle arrays, commonly seen on cells in- volved in ion transport 5s and, indeed, ion exchange has been reported to be more active in peripheral astrocyte membranes than in perinuclear areas 59. Recently, the work of Fontana et al.
`
`Page 5 of 7
`
`YEDA EXHIBIT NO. 2074
`MYLAN PHARM. v YEDA
`IPR2015-00643
`
`

`
`276 TINS- June 1986 has established that beyond their possible role in maintaining the struc- tural, metabolic and electrophysio- logical integrity of the brain paren- chyme, astrocytes can actively interact with cells from the immune system. This conclusion is based on the observation that astrocytes can be induced to secrete large amounts of both interleukin 1 and prostaglandins 6° whilst, on the other hand, soluble products released by activated T lymphocytes stimulate astrocyte pro- liferation 6a,62. These findings promp- ted us to test possible direct inter- actions of astrocytes with syngeneic T line cells, by culturing monolayers of neonatal rat astrocytes which con- tained more that 98% GFAP-positive cells with antigen-specific T line cells derived from the same rat strain. We found that astrocytes are able to activate T line cells in an immuno- specific manner (Fig. 5). Activation is antigen dose-dependent and is re- stricted by Ia-determinants 63'64. It was interesting that untreated astrocytes did not express any demonstrable surface Ia, but that they acquired high Ia densities during the process of antigen presentation. Other workers showed that Ia can be induced on astrocyte membranes by mediators like y-interferon 65'66. Detailed studies in- deed suggest that mediators released by activated T line cells can induce astrocytic Ia, which then enhances the astrocytes ability to present antigen to histocompatible T lymphocytes~. However, do these in-vitro obser- vations bear any relevance to the in- vivo situation? Low numbers of Ia + astrocytes have been reported in situ (although not yet proven at an ultrastructural level) in both multiple sclerosis and EAE 5s'6s but not as yet in normal CNS tissue. This sparse distri- bution of Ia + astrocytes, even in highly inflamed areas of the CNS, is not surprising. It can be explained by the low T cell astrocyte ratio in vivo as compared to the experimental situation in vitro. Furthermore, it may reflect the ability of astrocytes to down-modulate local immune reactions through secre- tion of mediators like prostaglandins ~j and apolipoprotein A. Hypothesis There are two key conclusions to be drawn from the data discussed above: (1) only activated, but not resting lymphocytes, can cross the BBB and patrol the CNS parenchyme; (2) they cooperate within the CNS primarily with astrocytes, which are facultative, inducible APC. We believe that these results should lead to re-evaluation of the immune status of the CNS, which contrary to traditional views; is not an immunological 'blind spot'. It seems to be under immune surveillance which, however, is adapted to meet the specialized requirements of CNS tissue. The CNS parenchyme is more vulnerable to damage than most other tissue. It is composed almost ex- clusively of a network of integrated and electrophysiologically active cellular elements with little connective tissue support. Moreover its regenerative capacity is very low. Any cellular immune reaction leads to local damage and subsequent scarring of the sur- rounding tissue. Such 'bystander damage' would be detrimental for CNS function and must therefore be reduced to a minimum. We speculate that this potential damage is minimized by reducing the number of lymphocytes entering the CNS and by focusing their responses to perivascular areas of the parenchyme. The fact that only activated T cells will cross the BBB greatly reduces the number of cells patrolling the CNS, as only a minority of circulating T cells are in a state of activation. These few activated immune cells represent the spectrum of current immune reactions and are therefore uniquely armed to protect the CNS from the potential metastatic spread of an initially topical pathogen. Astrocytes, by their anatomical distribution and functional properties as the primary APCs within the CNS parenchyme determine intrathecal immune reactivity in two fundamental ways. First, astrocytes are ubiquitous glial elements of the CNS. Their peculiar arrangement at the blood- brain interface is especially relevant considering their postulated function in clearing material from the tissue. Astrocytes may therefore concentrate antigenic determinants around the blood vessels, consequently, immune reactions would be focused to the same perivascular areas. In addition, toxic mediators and metabolites produced during an immune response will be readily cleared away via the blood circulation. Second, it is remarkable that within the CNS parenchyme initial antigen presentation appears to be performed solely by facultative APCs. *This was shown for astrocytes but may also be true for microglia. This capacity to present antigens, and the synthesis and expression of Ia antigens must be actively induced*. These properties will be promptly lost, when the local production of inducing factors is discontinued. Thus, faculta- tive APCs have the capacity to strictly limit a local immune response. The regulatory role of astrocytic APCs is further strengthened by their capacity to produce immunosuppressive medi- ators such as prostaglandins and apolipoprotein E. This hypothesis is by no means complete. Outstanding questions in- clude the importance of interactions between astrocytes and other CNS elements (such as endothelial and pericytic cells) and the possible contri- butions by other resident cells, such as microglia, to immune responses within the CNS. We hope that this concept will help clarify the mechanisms operating in chronic inflammatory CNS disease and, in addition provide an approach to study the development of interactions between the CNS and the immune system. Acknowledgements We thank all our colleagues who participated in different parts of this project, and are indebted to Carola Hammerich, who patiently typed this manuscript. The Clinical Research Unit for Multiple Sclerosis is supported by funds of the Hermann-und-Lilly-Schilling Stiftung. Selected references 1 Thomas, L. (1959) in Cellular and Humoral Aspects of the Hypersensitivity States (Lawrence, H. S., ed.). pp. 529--533, Harper & Row 2 Burnet, F. M. (1970) Prog. Exp. Tumor Res. 13, 1-17 3 Schwartz, R. N. (1984)Annu. Rev. lmmunol. 3, 237-262 4 Steinman, R. M. and Nussenzweig, M. C. (1980) Immunol. Rev. 53, 125-147 5 Streilein, J. W. andBergstresser, P. L. (1984) Transplant. 30, 319-323 6 Unanue, E. R., Belier, D., Liu, C. and Allen, P. M. (1984)J. lmmunol. 132, 1-5 7 Forsum, U. L., Klareskog, L. and Tonburd, R. (1979) Scand. J. lmmunol. 9, 343-349 8 Natali, P. G., De Martino, C., Quaranta, V., Nicotra, M.R., Frezza, F.. Tellegrino, M. A. and Serrone, S. (1981) Transplantation 3t, 7.5--81 9 Pober, J. S., Gimbrone, M. A. Jr, Cotran, R. S., Reiss, C. S., Burakoff, S. J., Siers, W. and Ault, K. A. (1983) J. Exp. Med. 157, 1339-1353 t0 Barker, C. F. and Billingham, R. E. (1977) Adv. lmmunol. 25, 1-54 11 Wikstrand, C. J. and Bigner, D. D. (198{l) Am. J. lath. 98, 517-568 12 Head, J. R. and Griffin, W. S. T. (1985) Froc. R. Soc. London Ser. B. 224, 375-387 13 Fontana, A. and Fierz, W. (1985) Springer
`
`Page 6 of 7
`
`YEDA EXHIBIT NO. 2074
`MYLAN PHARM. v YEDA
`IPR2015-00643
`
`

`
`TINS- June 1986 277 Semin. lmmuno. Pathol. 8, 97-110 14 Medawar, P. B. (1948) Br. J. Exp. Pathol. 29, 18--69 15 Schachner, M. and Sidman, R. L (1973) Brain Res. 60, 191-198 16 Pizarro, O., Hoecker, G., Rubinstein, P. and Ramos, A. (1961) Proc. Natl Acad. Sci. USA 47, 1900-1907 17 Wong, G. H. W., Bartlett, P. F., Clark- Lewis, I., McKimm-Breschkin, J.L. and Schrader, J. W. (1985) J. Neuroimmunol. 7, 255-278 18 Hart, D. N. J. and Fabre, J. W. (1981) J. Exp. Med. 154, 347-361 19 McFarlin, D. E. and McFarland, H. F. (1982) N. Engl. J. Med. 307, 1183-1188 and 1246- 1251 20 Martenson, R. E. (1984) in Experimental Allergic Encephalomyelitis: A Useful Model for Multiple Sclerosis (Alvord, E. C., Kies, M.W. and Suckling, A.S., eds), pp. 511-521, 21 Kitamura, K., Suzuki, M., Suzuki, A. and Uyemura, K. (1980) FEBS Lett. 115, 27-30 22 Roach, A., Boylan, K., Horvath, S., Prusiner, S. B. and Hood, L E. (1983) Cell 34, 799-806 23 Paterson, P. Y. (1960)J. Exp. Med. 111,119- 126 24 Hughes, R. A. C., Kadlubowski, M., Gray, I.A. and Leibowitz, S. (1981) J. Neurol. Neurosurg. Psychiatr. 44, 565-569 25 Ben-Nun, A., Wekerle, H. and Cohen, I. R. (1981) Eur. J. lmmmunoL 11, 195-199 26 Reske-Kunz, A., v. Steidern, D., ROde, E., Osawa, H. and Diamantstein, T. (1984) J. Immunol. 133, 1356-1361 27 Waehneldt, T. V. and Linington, C. (1980) in Neurological Mutations Affecting Myelination, INSERM Symposium No 14. (Baumann, N. ed.), pp. 389-412, INSERM 28 Trapp, B. D., Mclntyre, L.J., Quarles, R.H., Sternberger, N.H. and Webster, H. deF. (1979) Proc. NatlAcad. Sci. USA 76, 3552-3556 29 Naparstek, Y., Ben-Nun, A., Holoshitz, J., Reshef, T., Frenkel, A., Resonberg, M. and Cohen, I. R. (198