1254 Dispatch

Viral immunity: Interferons jog old T-cell memoriesDavid Gray

The mechanism by which life-long immunity to virusespersists is a matter of some controversy. Interferonsinduced by subsequent unrelated viral infections maypropagate existing memory T cells non-specifically.

Address: Department of Immunology, Royal Postgraduate MedicalSchool, Hammersmith Hospital, Du Cane Road, London W12 0NN,UK.

Current Biology 1996, Vol 6 No 10:1254–1255

© Current Biology Ltd ISSN 0960-9822

The immune system has a famously long memory of itsencounters with viruses. None of the Faroe Islanders wholived through a measles outbreak in 1781 became ill duringa second epidemic that affected most of the population 65years later [1]. The islands appear to have been measles-free during the intervening period. The immune memoryengendered by viruses can be life-long, in contrast to thateffected by more inert antigens. Immune memory gener-ated following vaccination with live, attenuated viruses ismuch longer-lived than that attainable with killed virusesor protein components.

There are many possible reasons for such selectivememory. One possibility is that it reflects the persistence oflive virus. A second is that the enormous antigen loadachieved by replicating virus could enhance both antigenstorage [2] and the magnitude of the response of specificlymphocytes. A third alternative is that live virus mightinduce qualitatively different responses in both the specificand innate arms of the immune system. In this last case, thetype of cytokine produced may be quite different followingvaccination with live or killed virus [3]. A recent study [4]shows that type I interferons (IFN I, consisting of IFN-aand IFN-b), which are produced in large amounts followingviral infection, can cause the proliferation of bystander Tcells and may be important during not only the generationbut also the maintenance of CD8 memory T cells.

The major effector arm of the immune response againstviruses consists of the cytotoxic, CD8-bearing T cells thatkill cells displaying viral peptides on their class I major his-tocompatibility (MHC) molecules. In mice infected with avirus, a large proportion of the CD8 T-cell pool becomesactivated and enters cell cycle. This is well documented byTough et al. [4], who found that ~75 % of CD8 T cellsincorporate bromodeoxyuridine, a measure of progressionthrough S phase of cell cycle, within seven days of infec-tion with lymphocytic choriomeningitis virus (LCMV).Focussing on the CD44hi CD8 T cells that represent acti-vated or memory cells, the labelling index is around 90 %.

Given that the frequency of LCMV-specific T cellsreaches a maximum of only 1 in 100, even at the peak ofthe response, most of this activation/proliferation must beantigen non-specific.

As virus-infected cells produce IFN I, Tough et al. [4]tested whether the proliferation of bystander T cells couldbe mimicked by a known inducer of IFN I, the syntheticdouble-stranded RNA, poly(IC). Within three days ofpoly(IC) injection, 70–80 % of CD8 T cells divided, andthis was inhibited by co-injection of an antibody thatblocks IFN I action. Furthermore, purified IFN-b causedthe proliferation of CD8 T cells. IFN I appears to act onlyon ‘pre-activated’ CD44hi T cells. Almost all of theincreased turnover of the CD8 pool is attributable to divi-sion of these CD44hi T cells; naive or resting CD44lo Tcells do not respond. IFN I-mediated T-cell activationdiffers from antigen-driven activation in two importantways. First, it provokes only one round of cell division andtherefore a two-fold increase in numbers; antigen-drivenresponses may involve more than ten divisions and often a10 000-fold increase in numbers of the specific clone. Andsecond, the IFN I activation is not full-blown, as markerstypical of antigen-activation, CD69 and CD25, remain low.

Following antigen-specific activation, T cells increaseexpression of the adhesion molecule CD44, a change thatremains stable for many months on what are presumablymemory T cells. The means by which these memory cellspersist remains controversial; one view is that individualmemory T cells have a very long lifespan, and another isthat memory cells require continuing contact with antigenstores for their survival (reviewed in [1,2]). Somewhere inbetween these extremes are theories invoking cytokinesas survival factors [5] or interactions with cross-reactiveantigens [6]. Can IFN I potentiate the long-term persis-tence of CD44hi CD8 memory cells? This was not directlytested by Tough et al. [4], although they did show thatIFN I influences the generation of memory cells; injectionof poly(IC) alone or during antigen priming enhances thesize of the CD44hi population of memory cells, and this isstable over a period of at least four weeks.

The implication of these observations is that, during anantigen-specific response, IFN I may tip the balancetowards memory-cell differentiation and away from theeffector cell pathway. To address the role of IFN I as a sur-vival factor for CD8 memory cells, it would be necessary toremove the IFN I signal, either by blocking IFN I withantibodies in mice in which a memory T-cell populationpre-existed, or by using gene ‘knockout’ mice deficient in

Dispatch 1255

IFN I or the IFN I receptor. Experiments in such micewould also establish if the function of IFN I was an obliga-tory one, or if other cytokines elicited by viral infection alsocontributed.

On the face of it, the idea that the survival of memory cellsspecific for one antigen is potentiated by subsequentinfections with quite different viruses is an attractive one,as it provides a mechanism whereby memory cells can bekept in a state of readiness if the external conditionsdemand it. A number of groups have reported thatmemory cells specific for one virus can be reactivated bysubsequent challenge with viruses that are supposedlyunrelated [7]. Most convincingly, a paper two years agofrom Selin et al. [8] documented the reactivation ofLCMV-specific memory cytotoxic T cells during subse-quent infection with either vaccinia or Pichinde virus. Thereactivation was not based on cross-reactive recognition ofa linear peptide, as an LCMV-derived nucleoprotein (NP)peptide, but not the equivalent, partially homologousregion of Pichinde virus NP, activated the LCMV-specificmemory cells. In spite of this, the authors [8] favouredcross-reaction as the explanation their data. The resultsnow provided by Tough et al. [4] suggest that the reactiva-tion of memory cells by non-homologous viruses is due totheir elicitation of IFN I and possibly other cytokines.

Unfortunately this interpretation is complicated by a morerecent paper, in which Selin et al. [7] report that the reacti-vation of memory cells during infection with unrelatedviruses is associated with a fall, not an increase, in the fre-quency of cytotoxic T cells reacting to the original virus.Thus, the number of LCMV-reactive T cells at steadystate decreases each time there is a subsequent infectionwith vaccinia or Pichinde virus. Although these results areclearly statistically significant, one is left wondering if a2–4-fold reduction in frequency of LCMV-specific T cellsis biologically significant as far as the memory response isconcerned. It is possible that this represents a homeostasisof the CD8 T-cell pool, purging itself of cells to makeroom for the expansion of new clones.

Which cells would be deleted in such a homeostaticprocess? In view of the action of IFN I to cause cell divisionin CD8 T cells with an activated phenotype, prime candi-dates might be those memory cells that revert to a partiallyresting/naive phenotype (CD45RBhi/L-selectinhi). It wouldbe interesting to know if CD45RBhi/CD62Lhi/CD44hi ‘par-tially resting’ memory cells are more or less responsive toIFN I than their CD45RBlo/CD62Llo/CD44hi counterparts.The action of IFN I during an infection might be to top upthe pool of ‘non-resting’ memory cells, in the face ofdecrease in the total number of memory cells for a particu-lar antigen caused by the homeostasis mechanism. Theaction of specific or even cross-reactive antigen in memorymaintenance would clearly be more potent [2].

Interestingly, IFN I is reported to have little activating orproliferative effect on CD4 T cells. However, more CD4T cells divide during a viral infection than can beaccounted for by an antigen-specific response [4]. If acytokine is involved in the maintenance of CD4 T cells, itis different to that used by CD8 T cells and is likely to bedistinct from nerve growth factor, which may be importantfor memory B-cell survival [9]. Many questions remain,but these observations open up the possibility that thelifespan of memory cells is regulated by the cytokinemilieu in which they find themselves, which in turn isdirectly influenced by the external environment.

AcknowledgementsThanks go to Hans Stauss and Gitta Stockinger for their very helpful com-ments.

References1. Ahmed R, Gray D: Immunological memory and protective

immunity: understanding their relation. Science 1996, 272:54–60.2. Gray D: Immunological memory. Annu Rev Immunol 1993,

11:49–77.3. Nguyen L, Knipe DM, Finberg RW: Mechanism of virus-induced Ig

subclass shifts. J Immunol 1994, 152:478–484.4. Tough DF, Borrow P, Sprent J: Induction of bystander T cell

proliferation by viruses and type I interferon in vivo. Science 1996,272:1947–1950.

5. Unutmaz D, Pileri P, Abrignani S: Antigen-independent activation ofnaive and memory resting T cells by a cytokine combination. J ExpMed 1994, 180:1159–1164.

6. Beverley PCL: Is T cell memory maintained by cross-reactivestimulation? Immunol Today 1990, 11:203–205.

7. Selin LK, Vergilis K, Welsh RM, Nahill SR: Reduction in otherwiseremarkably stable virus-specific cytotoxic T lymphocyte memoryby heterologous virus infection. J Exp Med 1996, 183:2489–2499.

8. Selin L, Nahill SR, Welsh RM: Cross-reactivities in memorycytotoxic T lymphocyte recognition of heterologous viruses. J ExpMed 1994, 179:1933–1943.

9. Torcia M, Bracci-Laudiera L, Lucibella M, Nencioni L, Labardi D,Rubartelli A, Cozzolino F, Aloe L, Garaci E: Nerve growth factor is anautocrine survival factor for memory B lymphocytes. Cell 1996,85:345–356.