The Journal of Clinical Investigation | March 2002 | Volume 109 | Number 5 581

Degeneracy, as opposed to specificity, in immunotherapy

David A. HaflerCenter for Neurologic Disease, Brigham and Women’s Hospital and Harvard Medical School, Harvard Institutes of Medicine, Room 786, 77 Avenue Louis Pasteur, Boston, Massachusetts 02115, USA. Phone: (617) 525-5330; Fax: (617) 535-5333; E-mail: dhafler@rics.bwh.harvard.edu.

J. Clin. Invest. 109:581–584 (2002). DOI:10.1172/JCI200215198.

While it could be argued that Jenner,with his discovery of vaccination, ush-ered in the field of immunotherapy,manipulation of the human immunesystem to treat inflammatory diseasesis only now coming of age. The mech-anism behind immune specificity, thebasic premise of vaccination, waslargely solved with appreciation ofhow the MHC presents antigen, andwith crystallization of the MHC/T cellreceptor complex. However, as hasonly recently been appreciated, while agiven T cell receptor may ordinarilyexhibit exquisite specificity, binding tosome antigenic sequences can effect aconformational change to the recep-tor, allowing it to recognize MHC classII/peptide complexes in a highlydegenerate fashion.

This capacity may represent a funda-mental mechanism of autoimmune dis-ease, since it allows microbial antigensto trigger the T cell receptors of autore-active T cells. Indeed, as discussed below,the activation of these T cells probablyoccurs both indirectly, by means of theinnate immune system, and directly,through T cell receptor stimulation (1).

Fortunately, the T cell receptor’s degen-eracy may be key to therapies, withadvantages not seen with antigen-spe-cific approaches. A recent series of pub-lications offers new insights into howthis aspect of the cellular immuneresponse might be used to treat humanautoimmune disease. One well-estab-lished but still only incompletely under-stood reagent that may work throughsuch a mechanism is glatiramer acetate(GA) (Copaxone, registered trademarkof Teva Pharmaceutical Industries Ltd.),discussed by Karandikar et al. in thisissue of the JCI (2).

T cell receptor degeneracyHistorically, the ability to generateand study cloned T cells reactive toself antigens reinforced the image of Tcell receptor interactions as highlyspecific for individual antigens. Forexample, T cell clones isolated fromthe blood of patients with multiplesclerosis (MS) exhibit exquisite speci-ficity for the immunodominantp85–99 epitope of myelin basic pro-tein (MBP) (3). However, this speci-ficity is not absolute. Changing the T

cell receptor contact residue lysine atposition 93 to an arginine, or even justremoving a hydroxyl group by chang-ing a phenylalanine to a tyrosine atposition 91, can totally ablate T cellreactivity. Surprisingly, however, thelysine-to-arginine substitution alsoresults in a more degenerate patternof recognition by the T cell receptor,in that a tyrosine or other amino acidresidues can now be tolerated at posi-tions 91 or even 90 (4). Thus, while a Tcell receptor appears to be highly spe-cific in one situation, altering the pep-tide ligand can change the TCR con-formation to yield a higher degree ofT cell cross-reactivity.

Work using combinatorial chemistryto analyze another series of MBPp85–99–reactive T cell clones leads to asimilar conclusion. Hemmer et al. (5)identified a number of viral epitopesthat can trigger autoreactive T cellclones in a manner that would not bepredicted by simple algorithms. Indeed,one MBP-reactive T cell clone studiedby these authors recognized an epitopeof a different self protein entirely, themyelin oligodendrocyte glycoprotein.

CommentarySee related article, pages 641–649.

Figure 1Autoimmune mechanism of MS. Naive, myelin reactive T cells are activated by microbes that: 1) induce the innate immune system to provide costim-ulatory signals necessary for clonal expansion of naive CD4+ T cells and, 2) contain epitopes that are cross-reactive with self antigens. Activated CD4+

T cells cross the blood-brain barrier and recognize self antigen presented by microglia, local antigen-presenting cells (APC) that sample self antigens.The reactivation of autoreactive CD4+ cells leads to a cascade of events including recruitment of macrophages, antibody secretion, and eventual destruc-tion of myelin and axons.

Hence, a significant degree of function-al degeneracy exists in the recognitionof self antigens by T cells.

Bystander suppressionThe high frequency of activated,myelin-reactive T cells in the circula-tion and cerebrospinal fluid ofpatients with MS, an inflammatorydisease of the white matter in the CNS,is consistent with the hypothesis thatthe disease is initiated by a microbialinfection, as shown in Figure 1. In agenetically susceptible host, accordingto this model, infection by commonpathogens stimulates the innateimmune system, which activates theexpression of costimulatory moleculesby autoreactive T cells. At the sametime, the presence of microbial anti-gens that cross-react with self antigenstriggers these T cells and leads to theautoimmune destruction of myelinand neuronal axons.

The observation of epitope spread-ing in the experimental autoimmuneencephalomyelitis (EAE) model (6, 7)and the finding that self antigens elic-it diverse T cell receptor repertoireshave made approaches that target asingle, antigen-reactive T cell theoreti-cally untenable. In their place hasemerged the concept of bystander sup-pression: Autoreactive Th2 or Th3 Tcells, according to this model, are notlost quantitatively to negative selec-tion but persist and migrate to the

inflamed target organ, where they canbe reactivated in an antigen-specificmanner. However, cytokines producedby these cells then downregulateinflammation in the local milieu in anantigen-nonspecific mechanism (Fig-ure 2) (8), thus suppressing autoim-mune disease under most conditions.In this scenario, it should also benoted that high expression of costim-ulatory molecules in the sites ofinflammation facilitates the epitopespreading (7), and allows a greaterdegree of T cell receptor degeneracy bydriving clonal of potentially autoreac-tive T cells by lower affinityTCR/MHC peptide interactions (9).

Altered peptide ligandsIt was recognized over a decade agothat the strength of signal deliveredthrough the T cell receptor can deter-mine which cytokines are secreted bythe T cell (10). The cell apparentlymeasures affinity in part by timing theengagement between the T cell recep-tor and the peptide/MHC complex.With longer engagement, a total,rather than partial, T cell receptorcomplex has time to form, and theextent of ζ chain phosphorylationincreases correspondingly. Alteredpeptide ligand (APLs), which bindwith low affinity to the T cell receptor,weaken this signal. The ability of APLsto change the cytokine program of a Tcell from a Th1 to a Th2 response was

exploited first by Kuchroo andcoworkers as a therapy for autoim-mune disease (11). Using the murineEAE model of MS, these authorsshowed that APLs can activate IL-4secretion by both encephalitogenic Tcells and naive T cell clones that cross-react with self antigens.

Injection of APLs is of clear thera-peutic value in treating different mod-els of EAE (12), and autoreactivehuman T cell clones can also beinduced to secrete the antiinflamma-tory cytokines IL-4 and TGF-β after Tcell receptor engagement by APLs (13,14). However, as Ausubel et al. (15)noted, while APLs can induce Th2cytokine secretion of MBP-reactive Tcells isolated from the peripheral bloodT cells of MS patients, they can alsoinduce a heteroclictic response in somepatients, activating these MBP-reactiveT cells against the patient’s own tis-sues. These data provide a strongrationale for the therapeutic use ofAPLs in patients with autoimmune dis-ease. However, they also raise the issuethat in some instances, highly degener-ate T cell receptors can recognize APLsas self antigens.

A recently published phase II clinicaltrial testing an altered MBP p85–99peptide confirms both of these con-clusions. At the higher peptide dosagetested, two of seven MS patients devel-oped remarkably high frequencies ofMBP-reactive T cells, and these

582 The Journal of Clinical Investigation | March 2002 | Volume 109 | Number 5

Figure 2The CD4+ T cell model of action by glatiramer acetate (GA). GA induces strong MHC class II restricted proliferative responses by T cells. Daily injec-tions of GA induces a moderate loss of responsiveness to the antigen accompanied by a shift to a more Th2 type of CD4+ T cell. The surviving GA-reactive T cells show a greater degree of degeneracy, as measured by cross-reactive responses to combinatorial peptide libraries. Highly cross-reac-tive Th2 cells with degenerate T cell receptors migrate to the site CNS, recognize self antigens as weak agonists or “altered peptide ligands”. In response,they begin to secrete Th2/Th3 cytokines, and suppress inflammation by the mechanism of bystander suppression.

responses were associated with signif-icant increases in MRI-detectablelesions (16). In contrast, patients treat-ed with lower doses of the APL showedno such disease flare-ups and mayhave indeed exhibited some degree ofimmune deviation toward increases inIL-4 secretion of MBP-reactive T cells(17, 18). Thus, APLs represent a classicdouble-edged sword. In our outbredpopulation, given the high degree ofdegeneracy in the immune system, it isunclear whether it is possible to findAPLs of self peptides that pose no riskof cross-reactivity with self.

Glatiramer acetateAn alternative approach to the use of asingle APL is the administration ofpeptide mixtures that contain manydifferent antigen specificities. Ran-dom copolymers that contain aminoacids commonly used as MHCanchors and T cell receptor contactresidues have been proposed as possi-ble “universal APLs.” GA is a randomsequence polypeptide consisting offour amino acids — alanine (A), lysine(K), glutamate (E), and tyrosine (Y), ata molar A/K/E/Y ratio of 4.5:3.6:1.5:1— and with an average length of40–100 amino acids (19). Directlylabeled GA binds efficiently to differ-ent murine H-2 I-A molecules, as wellas to their human counterparts, theMHC class II DR molecules, but itdoes not recognize MHC class II DQor MHC class I molecules in vitro (20).

In phase III clinical trials, GA, subcu-taneously administered to patientswith relapsing-remitting MS, decreas-es the rate of exacerbations and pre-vents the appearance of new lesionsdetectable by MRI (21). This repre-sents perhaps the first successful useof an agent that ameliorates autoim-mune disease by altering signalsthrough the TCR.

A “universal antigen” containingmultiple epitopes would be expectedto induce proliferation in vitro innaive T cells from the circulation, dueto its expected high degree of cross-reactivity with other peptide antigens.Indeed, GA induces strong MHC classII DR–restricted proliferative respons-es in T cells isolated from MS patientsor from healthy controls (22). In mostpatients, daily injection with GAcauses a striking loss of responsive-ness to this polymer antigen, accom-panied by greater secretion of IL-5and IL-13 by CD4+ T cells, indicatinga shift toward a Th2 response(22–26). The paper by Karandikar andcoworkers (2) in this issue of the JCIconfirms that this treatment causes ashift by CD4+ T cells toward a Th2 orTh3 phenotype, as judged byincreased levels of TGF-β mRNA andcell-associated IL-4. In addition, thesurviving GA-reactive T cells exhibit ahigh degree of degeneracy, as meas-ured by their ability to cross-reactwith a large variety of peptides repre-sented in a combinatorial library (23).

Thus, in vivo administration of GAinduces highly cross-reactive CD4+ Tcells that are immune-deviated tosecrete Th2 cytokines (Figure 2). Wehave proposed that GA-inducedmigration of highly cross-reactive Th2(and perhaps Th3) cells to sites ofinflammation allows their highlydegenerate T cell receptors to contactself antigens, which they recognize asweak agonists, much like APLs. TheseT cells then apparently secrete sup-pressive, Th2/Th3 cytokines, thusrestricting local inflammation (22).We concluded that, through this formof bystander suppression, GA caneffect an immune deviation to a Th2response and may prove useful in avariety of autoimmune disorders.

CD8+ T cellsIn their present report (2), Karandikarand coworkers have found that MHCclass I–restricted CD8+ T cells inuntreated MS patients respond weak-ly to GA. Treatment with GA stimu-lates these responses, restoring themto levels observed in healthy individu-als. Because GA therapy significantlyincreases IFN-γ expression by CD8 Tcells, the authors propose that thesecells regain an ability to suppressmyelin-reactive Th1 cells (Figure 3).

The significance of decreased CD8+

T cell responsiveness to GA in patientswith MS is not clear, however. GA canbe viewed as a nonspecific TCR ago-nist, perhaps similar to an anti-CD3

The Journal of Clinical Investigation | March 2002 | Volume 109 | Number 5 583

Figure 3The CD8+ T cell model of action by GA. MHC class I restricted CD8+ T cell responses to GA, which are significantly lower at baseline in MS patientsthan in healthy controls, increase significantly following treatment, as seen in the greater number of IFN-γ–positive, GA responsive CD8+ T cells. GAtreatment is proposed to restore CD8+ suppressor function, which is dependent upon IFN-γ secretion. The mechanism for CD8+ suppression of theimmune response is unknown. Shown in the figure is a CD8+ T cell migrating to the CNS, regulating a CD4+ autoreactive T cell.

mAb, which would be expected to bindequally to all TCR complexes. Thus,the decreased response by CD8+ T cellsto GA may represent a general nonre-sponsiveness of MHC class I–restrict-ed CD8 T cells in patients with MS.This finding harks back to earlieranalyses of peripheral blood inpatients with MS, which also revealeda modest decrease in CD8+ T cell num-bers. Alternatively, GA may targetpathological subpopulations of CD8 Tcells, preventing destruction of the tar-get tissue. Indeed, we have recentlyfound evidence for the existence ofwhat appear to be pathologic CD8cells in the circulation of patients withMS. We examined the response ofhighly purified CD8+ T cells from MSpatients to TCR cross-linking in thepresence or absence of a costimulato-ry signal provided by anti-CD28. CD8+

T cells from MS patients, but not fromnormal subjects, secrete more lympho-toxin and TNF-α in response to T cellreceptor cross-linking (G. Buckle andD. Hafler, unpublished observations).The ability of these patients’ CD8+ Tcells to secrete high amounts of lym-photoxin, independent of costimula-tion, suggests that such cells could bedirectly involved in CNS damage inMS and might not act through a sepa-rate population of CD4+ Th1 cells, assuggested by Karandikar et al. (2).

ConclusionsHow can these changes in CD8+

responses be reconciled with what hasbeen demonstrated regarding GA’seffect on CD4+ T cells? To addressthis question, it will be essential todetermine how the CD8+ GA-reactiveT cells identified by Karandikar et al.(2) contribute to the disease processin EAE and MS. As the authors sug-gest, these cells may represent regula-tory T cells whose activity and abilityto secrete IFN-γ can be enhanced byGA treatment. Alternatively, they maybe activated, lymphotoxin-secretingcytotoxic T cells whose capacity tomediate demyelination is somehowinhibited by GA.

Perhaps different populations ofCD8+ cells in these individuals exhib-it different functions at the sametime. Indeed, the division of CD8+ Tcells into functionally discrete CD28+

and CD28– populations has beenknown for some time (27). It appearsthat GA may have more than onemechanism of action, inducing differ-ent effects in CD4+ and CD8+ T cellsthat work in separate but parallelimmune pathways. Clearly, furtherstudies are required on the effects ofGA on the various CD8+ T cell popu-lations in patients with MS.

AcknowledgmentsI would like to thank our laboratoryfor helpful comments reading thisCommentary. Disclosure: The authorhas received grant support from TevaNeuroscience.

1. Wucherpfennig, K.W., and Strominger, J.L. 1995.Molecular mimicry in T cell-mediated autoim-munity: viral peptides activate human T cellclones specific for myelin basic protein. Cell.80:695–705.

2. Karandikar, N.J., et al. Glatiramer acetate (Copax-one) therapy induces CD8+ T cell responses inpatients with multiple sclerosis. J. Clin. Invest.109:641–649. DOI:10.1172/JCI200215198.

3. Ota, K., et al. 1990. T-cell recognition of animmunodominant myelin basic protein epitopein multiple sclerosis. Nature. 346:183–187.

4. Ausubel, L.J., et al. 1996. Complementary muta-tions in an antigenic peptide allow for crossreac-tivity of autoreactive T-cell clones. Proc. Natl. Acad.Sci. USA. 93:15317–15322.

5. Hemmer, B., et al. 1997. Identification of highpotency microbial and self ligands for a humanautoreactive class II-restricted T cell clone. J. Exp.Med. 185:1651–1659.

6. Lehmann, P.V., et al. 1992. Spreading of T-cellautoimmunity to cryptic determinants of anautoantigen. Nature. 358:155–157.

7. Miller, S.D., et al. 1995. Blockade of CD28/B7-1interaction prevents epitope spreading and clini-cal relapses of murine EAE. Immunity. 3:739–745.

8. Chen, Y., et al. 1994. Regulatory T cell clonesinduced by oral tolerance: suppression ofautoimmune encephalomyelitis. Science.265:1237–1240.

9. Anderson, D.E., et al. 1997. Weak peptide ago-nists reveal functional differences in B7-1 and B7-2 costimulation of human T cell clones. J.Immunol. 159:1669–1675.

10. Evavold, B.D., and Allen, P.M. 1991. Separation ofIL-4 production from Th cell proliferation by analtered T cell receptor ligand. Science.252:1308–1310.

11. Nicholson, L.B., et al. 1997. A T cell receptorantagonist peptide induces T cells that mediatebystander suppression and prevent autoim-

mune encephalomyelitis induced with multiplemyelin antigens. Proc. Natl. Acad. Sci. USA.94:9279–9284.

12. Brocke, S., et al. 1996. Treatment of experimentalencephalomyelitis with a peptide analogue ofmyelin basic protein. Nature. 379:343–346.

13. Windhagen, A., et al. 1995. Modulation ofcytokine patterns of human autoreactive T cellclones by a single amino acid substitution of theirpeptide ligand. Immunity. 2:373–380.

14. Ausubel, L.J., Krieger, J.I., and Hafler, D.A. 1997.Changes in cytokine secretion induced by alteredpeptide ligands of myelin basic protein peptide85-99. J. Immunol. 159:2502–2512.

15. Ausubel, L.J., Bieganowska, K.D., and Hafler, D.A.1999. Cross-reactivity of T-cell clones specific foraltered peptide ligands of myelin basic protein.Cell. Immunol. 193:99–107.

16. Bielekova, B., et al. 2000. Encephalitogenic poten-tial of the myelin basic protein peptide (aminoacids 83-99) in multiple sclerosis: results of aphase II clinical trial with an altered peptide lig-and. Nat. Med. 6:1167–1175.

17. Kappos, L., et al. 2000. Induction of a non-encephalitogenic type 2 T helper-cell autoim-mune response in multiple sclerosis after admin-istration of an altered peptide ligand in aplacebo-controlled, randomized phase II trial.The Altered Peptide Ligand in Relapsing MSStudy Group. Nat. Med. 6:1176–1182.

18. Crowe, P.D., et al. 2000. NBI-5788, an alteredMBP83-99 peptide, induces a T-helper 2-likeimmune response in multiple sclerosis patients.Ann. Neurol. 48:758–765.

19. Teitelbaum, D., et al. 1971. Suppression of exper-imental allergic encephalomyelitis by a syntheticpolypeptide. Eur. J. Immunol. 1:242–248.

20. Fridkis-Hareli, M. and Strominger, J.L. 1998.Promiscuous binding of synthetic copolymer 1 topurified HLA-DR molecules. J. Immunol.160:4386–4397.

21. Johnson, K.P., et al. 1995. Copolymer 1 reducesrelapse rate and improves disability in relapsing-remitting multiple sclerosis: results of a phase IIImulticenter, double-blind, placebo-controlledtrial. Neurology. 45:1268–1276.

22. Duda, P.W., et al. 2000. Human and murine CD4T cell reactivity to a complex antigen: recognitionof the synthetic random polypeptide glatirameracetate. J. Immunol. 165:7300–7307.

23. Duda, P.W., et al. 2000. Glatiramer acetate(Copaxone) induces degenerate, Th2-polarizedimmune responses in patients with multiple scle-rosis. J. Clin. Invest. 105:967–976.

24. Neuhaus, O., et al. 2000. Multiple sclerosis: com-parison of copolymer-1-reactive T cell lines fromtreated and untreated subjects reveals cytokineshift from T helper 1 to T helper 2 cells. Proc. Natl.Acad. Sci. USA. 97:7452–7457.

25. Gran, B., et al. 2000. Mechanisms ofimmunomodulation by glatiramer acetate. Neu-rology. 55:1704–1714.

26. Farina, C., et al. 2001. Treatment of multiple scle-rosis with Copaxone (COP): Elispot assay detectsCOP-induced interleukin-4 and interferon-gamma response in blood cells. Brain.124:705–719.

27. Turka, L.A., et al. 1990. CD28 is an inducible Tcell surface antigen that transduces a proliferativesignal in CD3+ mature thymocytes. J. Immunol.144:1646–1653.

584 The Journal of Clinical Investigation | March 2002 | Volume 109 | Number 5