n e w s a n d v i e w s

732 volume 15 | number 7 | june 2009 nature medicine

That central memory T cells are generated after Gsk-3β inhibition in vivo raises the ques-tion of whether this pathway normally oper-ates during memory formation and whether it occurs in a specific environment or niche. Wnt signaling is active in the bone marrow to restrain the differentiation of hematopoi-etic stem cells, but it is unknown whether memory T cells also use this pathway for their generation. Finally, these results lend them-selves to the ongoing debate on the lineage relationship between effectors and memory cells by showing that acquisition of cytotoxic function need not occur for cells to develop into TSCM cells, at least not under conditions of Gsk-3β inhibition. Whether these findings are a faithful replication of a physiological pathway or a beneficial outcome of strategic pharmacology will be decided by the stud-ies that will undoubtedly be initiated by this report.

1. Herodotus. Histories (de Selincourt, A., trans.) edn. 11 (Penguin, London, 1972).

2. Gattinoni, L. et al. Nat. Med. 15, 808–813 (2009).3. Berke, G. Annu. Rev. Immunol. 12, 735–773 (1994).4. Surh, C.D. & Sprent, J. Immunity 29, 848–862

(2008).5. Croce, J.C. & McClay, D.R. Methods Mol. Biol. 469,

3–18 (2008).6. Mosimann, C., Hausmann, G. & Basler, K. Nat. Rev.

Mol. Cell Biol. 10, 276–286 (2009).7. Malhotra, S. & Kincade, P.W. Cell Stem Cell 4, 27–36

(2009).8. Stemberger, C. et al. Semin. Immunol. 21, 62–68

(2009).

ing to potent and sustained cytotoxic T cell responses at the next exposure of antigen.

Taken together, the results of Gattinoni et al.2 outline a new approach to optimize CD8+ T cell responses, and these findings have the potential for clinical application for vaccines and adop-tive immunotherapy. Generating TSCM cells means that fewer cells may be required for adoptive immunotherapy, making a complex technique less cumbersome and potentially more widely used. Similarly, vaccines supple-mented with Gsk-3β inhibitors could produce a more durable immunity against infectious agents—one that isn’t susceptible to terminal differentiation and the ‘exhaustion’ of T cells that weakens immunity over time.

Much work remains before the physiology and molecular regulation of this phenom-enon is fully understood. Only a subset of the total responders become TSCM cells, leading Gattinoni et al.2 to speculate that asymmetric inheritance of signaling molecules during cell division may be involved in the acquisition of this cell fate. The direct visualization of these events may reveal the answer to this question. Furthermore, although the authors assume that inhibiting Gsk-3β induces activation of the Wnt–β-catenin pathway2, other signaling and downstream events could play a part in the generation of the T memory stem cells. A genetic approach might confirm the molecular pathways that are active in this system.

Wnt–β-catenin pathway, such as an inhibi-tor of glycogen synthase kinase-3β (Gsk-3β), which represses Wnt signaling2. Activation of the Wnt pathway inhibited both the prolif-eration of the treated cells and their ability to kill target cells. What’s more, the percentage of cells expressing markers of TCM cells and stem cells increased.

After adoptive transfer and antigenic stimu-lation, the treated cells rapidly differentiated into effector CD8+ T cells, producing cyto-kines and undergoing substantial prolifera-tion. Moreover, small numbers of transferred T cells could induce the destruction of tumors derived from B16 melanoma cells better than inoculation with either TEM or TCM cells. Importantly, some cells continued to express TCM markers after multiple rounds of divi-sion, indicating they had retained their ‘stem-ness’ and thereby earning the designation of T memory stem cells (TSCM cells)8.

If the investigators included Gsk-3β inhibi-tors during vaccination of mice in vivo, there was an increase in the percentage of CD8+ T cells with markers of the TCM cells2, although this cannot be unambiguously ascribed to a direct effect on CD8+ T cells themselves rather than another cell type. If true, how-ever, this observation suggests that activating the Wnt–β-catenin pathway could induce the formation of T memory stem cells during the initial priming step of immunization, lead-

Targeting lymphotoxin depletes pathogenic T cellsAnna M Hansen & Rachel R Caspi

A monoclonal antibody directed against lymphotoxin-a (LT-a) expressed by pathogenic T cells can prompt the clearance of these cells from the body (pages 766–773). The findings bring us one step closer to targeting only the cell populations that cause harm in autoimmune diseases while leaving beneficial arms of the immune system largely intact.

Anna M. Hansen and Rachel R. Caspi are in the

Laboratory of Immunology, National Eye Institute,

National Institutes of Health, Bethesda, Maryland,

USA.

e-mail: rcaspi@helix.nih.gov

Over the last ten years, the treatment options for individuals with autoimmune disorders have improved dramatically. A range of protein-based therapeutic agents have entered the market that specifically target the signaling pathways that cause conditions such as rheumatoid arthritis and multiple sclerosis. These therapies include agents that directly neutralize inflammatory cytokines, interfere with T cell activation or lead to the depletion of specific cell types known to

be involved in these diseases.Although the benefits of these therapies are

indisputable, they can come at a high price to the patient. In addition to targeting the pathological inflammation, these immunotherapeutic agents can also interfere with normal host defense against infection and malignancy1. Therefore, enhancing target specificity has been one of the primary goals of the field.

In this issue of Nature Medicine, Chiang et al.2 report the identification of an antibody that spe-cifically targets activated T cell subsets involved in autoimmunity and propose that this new approach may lead to a substantial improve-ment over current treatments.

We rely on our immune system to fight infec-

tion and to clear cells that have died or that have become malignant, but the same mechanisms that protect us can also bring harm. When the complex balance of regulation and responsive-ness breaks down, lymphocytes begin to attack self antigens, leading to autoimmune destruc-tion of the tissue. Consequently, therapeutic agents that target autoimmune cells tend to be detrimental to normal antimicrobial defense, because the same functions underlie both the normal and the pathologic immune processes. For example, tumor necrosis factor-a (TNF-a), which is a therapeutic target in rheumatoid arthritis and several other autoinflammatory diseases, is expressed by almost all cells of the immune system. TNF-a is essential in defending

©20

09 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

n e w s a n d v i e w s

nature medicine volume 15 | number 7 | june 2009 733

all activated T cells is also shared by T regulatory cells, whose role is to control immune responses, including those mediated by TH1 and TH17 cells6,7. Thus, treatment with CD25-specific antibodies has the potential to also deplete the beneficial T regulatory population.

The monoclonal antibody against LT-a developed by Chiang et al.2 may be able to home in on activated TH1 and TH17 while sparing other elements of the immune system. The expression of LT-a1β2 is limited largely to T cells, with low expression on B cells and natural killer cells that in this study appeared to be insufficient to trigger antibody-induced depletion. Regulatory T cells and TH2 cells were similarly unaffected by the antibody. Finally, by leaving the binding of LT-a1β2 to its recep-tor intact, this treatment lacks the deleterious effects on splenic architecture and germinal center formation of another LT-targeting agent, LTβR fusion protein, which was specifically developed to block the interaction between LT-a1β2 and its receptor and has shown mixed results in clinical trials2.

The selectivity and specificity of the LT-a– specific antibody for TH1 and TH17 cells, combined with its lack of deleterious effects on lymphoid tissue architecture, lowers the likelihood of serious side effects, suggesting the approach of Chiang et al. may be a promising future addition to the arsenal of biologics for treatment of debilitating autoinflammatory conditions.

1. Caspi, R.R. Nat. Rev. Immunol. 8, 970–976 (2008).2. Chiang, E. et al. Nat. Med. 15, 766–773 (2009).3. O’Garra, A. & Arai, N. Trends Cell Biol. 10, 542–550

(2000).4. Korn, T., Bettelli, E., Oukka, M. & Kuchroo, V.K. Annu.

Rev. Immunol. 27, 485–517 (2009).5. Abbas, A.R. Genes Immun. 6, 319–331 (2005).6. Sakaguchi, S., Yamaguchi, T., Nomura, T. & Ono, M.

Cell 133, 775–787 (2008).7. Shevach, E.M. Immunity 30, 636–645 (2009).

body against L-Ta that could stably bind LT-a1β2 expressed on the cell surface, leading to a process known as antibody-dependent cellular cytotoxic-ity. In this process, phagocytes such as macrophages recog-nize and eliminate antibody-coated cells (Fig. 1). Herein lies the novelty of this approach. Rather than simply neutral-izing the action of LT-a, this monoclonal antibody ‘tags’ LTa−1β2–expressing T cells, leading to their clearance by phagocytes. Although soluble LT-a3 was also neutralized, this interaction did not seem necessary for the therapeu-tic efficacy of the antibody. Mutations in the antibody that do not induce phagocytosis also rendered the antibody thera-peutically ineffective, despite binding LT-a3. Importantly, Chiang et al.2 found that this anti-body bound LT-a1β2 without interfering with its interaction with LTβR, leaving intact a major function of this cytokine in maintaining lym-phoid tissue structure and function.

Depletion of activated TH1 and TH17 cells by this monoclonal antibody against LT-a was found to be effective at reducing autoimmune disease symptoms in mouse models of multiple sclerosis (experimental autoimmune encepha-lomyelitis) and rheumatoid arthritis (collagen-induced arthritis). The treatment decreased the expression of proinflammatory cytokines and relieved other symptoms of these conditions, even when administered after the onset of dis-ease. Although disease amelioration was par-tial, it was similar to that achieved in the mouse model of rheumatoid arthritis by treatment with anti–TNF-a, currently a clinical standard of care for the condition2.

Whether targeting LT-a will prove to be a bet-ter “mouse trap” for the treatment of autoim-mune diseases than currently available options remains to be determined and depends not only on efficacy but also on side effects. Regimens for therapeutic targeting of T cells are already in use clinically or are in advanced stages of develop-ment1. Although it is clear that these regimens are effective therapies, some agents can lead to general T cell depletion without distinguish-ing between activated and nonactivated T cells, leaving the patient markedly immunocompro-mised.

Other, more specific agents that attempt to target only activated T cells may inadvertently deplete cells with similar surface markers that have protective, regulatory functions. For exam-ple, the IL-2 receptor (CD25) that is expressed by

against infection by stimulating phagocytosis of bacteria and production of antimicrobial pep-tides. It is not possible to neutralize the TNF-a that inflames a rheumatoid joint without also neutralizing the TNF-a that protects from tuberculosis, and consequently, tuberculosis poses a serious threat to the health of patients receiving anti–TNF-a therapy.

After microbial or other antigenic stim-uli, a common precursor T cell gives rise to T helper cells known as T helper type 1 (TH1; interferon-γ–producing), TH17 (interleukin-17 (IL-17)-producing) and TH2 (IL-4–, IL-5– and IL-13–producing) cells. Each of these T helper cell subsets has a distinct role in fighting particu-lar types of infections3,4. However, each can also be involved in immune responses that damage healthy tissue. TH2 cells participate in allergies and asthma, whereas TH1 and TH17 cells are commonly involved in autoimmune disorders.

Chiang et al.2 set out to identify a specific marker of T helper cells in their activated state. To accomplish this needle-in-a-haystack task, they performed a genome-wide search using the immune response in silico (IRIS) database5. This database was developed from a vast microarray study and features only genes shown to be spe-cifically expressed in immune cells.

From this search, they uncovered an unex-pected candidate in the somewhat enigmatic cytokine lymphotoxin-a (LT-a)2. LT-a was first characterized as a secreted homotrimer (LT-a3) but has since been found as a membrane-bound heterotrimer with LT-β, predominantly as the molecule LT-a1β2 (Fig. 1). Secreted LT-a3 binds the TNF receptors TNFRI and TNFRII and shares many of the functions of TNF-a. LT-a1β2 binds its own unique receptor, LT-β receptor (LTβR). In addition to overlapping functions between LT-a3 and TNF-a, the cre-ation of LTβR-deficient mice revealed a crucial, nonredundant role for this cytokine in the for-mation of secondary lymphoid tissues, such as the spleen and lymph nodes, and of germinal centers where B cells proliferate and mature after activation.

On the basis of their IRIS search, Chiang et al.2 examined the expression of LT-a1β2 on the various T cell subsets. They confirmed that TH1 and TH17 cells express high amounts of LT-a1β2 on their surface, which was maintained as the cells produced their hallmark inflamma-tory cytokines IFN-γ and IL-17. Expression LT-a1β2 on TH2 cells was lower than on TH1 and TH17 cells and was rapidly lost after activa-tion. This observation indicated that it might be possible to selectively target activated TH1 and TH17 cells, which are thought to have the most substantial role in autoimmunity, while sparing TH2 cells.

Chiang et al.2 developed a monoclonal anti-

TNFRITNFRII

LTβR

LT-a3

Anti–LT-a

FcRT cell

LT-a1β2

Activatedphagocyte

Lymphoid andtissue cells

Figure 1 Schematic of binding of LT-a–specific antibody (anti–LT-a). The monoclonal antibody identified by Chiang et al.2 blocks the interaction between LT-a3 and TNFRI and TNFRII but not the binding of LT-a1β2 to LTβR. FcR, Fc receptor.

Kat

ie V

icar

i

©20

09 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.