How the Immune System Recognizes the Body

Copyright 1993 Scientific American, Inc.

Organisms have various mecha-nisms of distinguishing betweenthemselves and everything else.

Many plants, for example, have hardouter shells that not only protect themagainst invaders but also deÞne theplantsÕ outer limits. Yeasts have mat-ing-type genes that code for proteinsthat prevent mating between similarcells. Sponges have a collection of geneswhose products can be used to detectand repel alien colonies.

The human body has evolved one ofthe most elaborate mechanisms fordistinguishing invaders from itself. Thecells of the immune systemÑthe lym-phocytes, macrophages and othersÑmust learn to tolerate every tissue, everycell, every protein in the body. Theymust be able to distinguish the hemo-globin found in blood from the insulinsecreted by the pancreas from the vit-reous humor contained in the eye fromeverything else. They must manage torepel innumerable diÝerent kinds ofinvading organisms and yet not attackthe body.

Immunologists have always beenpreoccupied with the issue of how thebody learns to tolerate itself, but onlyduring the past decade or so have theydiscovered the details of what preventsthe all-important lymphocytesÑthe Tcells and the B cellsÑfrom attackingtheir host. Many immature lymphocyteshave the potential to respond to self-products and therefore pose a threat.The body tries to rid itself of all suchcells by resorting to several ingeniousprocesses. If an immune cell reacts to a self-product while it is developing

in the thymus or bone marrow, it isusually killed or inactivated. A maturelymphocyte will usually suÝer the samefate if it responds to a self-product and does not receive a second chemi-cal message. These basic strategies thatthe body uses to eliminate self-reactingcells have many variations becausethere are many diÝerent types of lym-phocytes and many diÝerent kinds ofself-products.

Despite the safeguards of the im-mune system, some self-reacting lym-phocytes are not inactivated or killed,and they can cause one of several ill-nesses known as autoimmune diseases.Immunologists are well aware that ifthey understood all the mechanisms fortolerance, they might be able to preventautoimmune diseases. Furthermore, suchinsights might help surgeons in theircontinual search for reagents that canprevent the immune system from re-jecting such transplanted tissues askidneys, hearts and lungs.

Higher vertebrates have manyways of detecting and destroy-ing invaders. Some of these are

relatively nonspeciÞc and depend onthe fact that groups of infectious or-ganisms make chemicals that are notproduced in large amounts by highervertebrates. For example, mammals candetect the presence of invading bacte-ria because bacteria produce peptidesthat begin with an unusual amino acidÑformyl methionineÑwhereas mammalsproduce only small amounts of suchpeptides. Indeed, in mammals, high con-centrations of peptides with formyl me-thionine attract white blood cells calledneutrophils, which then consume thebacteria producing the peptides. Simi-larly, mammals can detect some virusesbecause viruses produce much greaterquantities of double-strand RNA thanmammals do. Large amounts of dou-

ble-strand RNA provoke mammaliancells to produce proteins called inter-ferons, which in turn stimulate a seriesof reactions that help the host resistfurther viral infection.

Although these nonspeciÞc respons-es to chemicals made by bacteria andviruses are an absolutely crucial part ofthe immune system, vertebrates also re-quire mechanisms for identifying specif-ic invaders. The immune system mustbe able to recognize foreign productswhose chemistry is only slightly diÝer-ent from its own molecules.

The immune system has evolved threevery sophisticated methods of recogniz-ing foreign chemicals, or what are termedantigens. The basis of these mecha-nisms is the three kinds of so-called re-ceptor proteins found on lymphocytes.The Þrst method requires B cells thathave receptor proteins known as im-munoglobulins; the second relies on Tcells that have a receptor protein calledalpha-beta, and the third utilizes T cellsthat have the protein gamma-delta.

Many receptors are attached to thesurface of each lymphocyte, and theywill, in speciÞc circumstances, bind toantigens. Each receptor is made up oftwo diÝerent polypeptide chains; immu-noglobulins consist of so-called lightand heavy chains; the alpha-beta pro-tein is made of an alpha chain and abeta chain, whereas the gamma-delta

How the Immune SystemRecognizes the BodyThe human immune system has developed several

elegant processes that allow it to repel foreign invadersand yet not attack the body itself

by Philippa Marrack and John W. Kappler

PHILIPPA MARRACK and JOHN W.KAPPLER have pioneered new techniquesfor studying tolerance in immune cells.Since 1986 they have both been investi-gators at the National Jewish Center forImmunology and Respiratory Medicinein Denver, part of the Howard HughesMedical Institute Research Laboratories.Marrack and Kappler have been marriedfor 19 years.

SCIENTIFIC AMERICAN September 1993 81

T CELLS develop in the thymus, but ifone reacts to a protein made by its hostbefore it matures, it will die. The pro-cess eliminates many T cells that havethe potential to attack the body.


protein has, as you might guess, a gam-ma chain and a delta chain. Each chaincan vary in sequence from one cell toanother. For example, the alpha andbeta chains of any given T cell will al-most certainly diÝer from those of anyother T cell. The receptors of any givenT cell, therefore, will probably bind to a diÝerent set of materials than thoseof other T cells. Human beings have asmany as a million million T cells andconsequently have many diÝerent al-pha-beta molecules available to recog-nize foreign material.

Indeed, when one considers the vastnumber of alpha-beta receptors of Tcells as well as the many immunoglob-ulin molecules of B cells and the manygamma-delta receptors of T cells, it isno small wonder that none of the lym-phocytes recognizes products of itsown host. This remarkable phenomenonhas intrigued investigators for decades,and many theories have been put for-ward to explain how the human immunesystem learns to tolerate the cells ofthe body.

One of the Þrst ideas was that ani-mals simply cannot make lymphocytereceptors that are self-reactive. In partic-ular, humans might not have the genesnecessary to create alpha-beta recep-tors that could react with human pro-teins. Immunologists have known formany years that this explanation is notright. Nowadays we realize that be-

cause the composition and structure ofreceptors are determined somewhat at random, some receptors are likely to be able to bind to the chemicals oftheir host.

Randomness is introduced in at leasttwo ways. First, the receptors of lym-phocytes are made by random combi-nations of specialized gene segments.The alpha and beta chains of T cells,for example, are created by a randommixture of gene segments known as V-alpha, J-alpha, V-beta, D-beta and J-beta. Second, short, random segmentsof DNA are introduced into the assem-bling genes for the alpha-beta receptor.Thus, the organism has no absolutecontrol over the complete amino acidsequences of the receptors.

Control must be exerted in someother way, at some other stage. One of the Þrst researchers to test this hypothesis was Ray D. Owen of the Cal-ifornia Institute of Technology. In 1945he was studying the inheritance of bloodtypes in cattle. He found that twin cowsthat shared a placenta were very likelyto have the same blood type. This corre-lation was even seen in one case wherethe twins had diÝerent fathers. Owenconcluded that the correlation betweenblood types was a consequence of theexchange of lymphocytes and other bloodcells in utero. Furthermore, he suggest-ed that this early exchange preventedthe cows from rejecting each otherÕs

blood. Later Sir Peter B. Medawar, RupertE. Billingham and Leslie Brent of theNational Institute for Medical Researchin London showed that if blood cellswere transferred from an adult mouseinto an unrelated, newborn mouse, thenewborn could accept a skin graft fromthe adult later in its life. Hence, the in-troduction of blood cells at birth couldaÝect the ability of the individual to ac-cept not only blood but also skin.

Most important, this work and OwenÕsresearch led to the same broad conclu-sion: the immune system is not bornwith all the instructions needed to rec-ognize the products of its host ; ratherit learns what is self as it develops.

One of the Þrst ideas that ex-plained how the immune sys-tem learns tolerance to self was

put forward by Joshua Lederberg, nowat the Rockefeller University. In 1959he suggested that immature lympho-cytes may not react to antigens in thesame way that mature lymphocytes do.Usually if something binds to the im-munoglobulin of a mature B cell, thelymphocyte changes into an antibody-secreting cell ; if a molecule attaches to the receptors of a mature T cell, the lymphocyte becomes either a cyto-kine-secreting cell or a killer cell. Led-erberg postulated that if an antigenbinds to the receptor of an immaturecell, the cell might die instead of be-coming active.

LederbergÕs hypothesisÑnow calledclonal deletion theoryÑprovides amechanism for eliminating lymphocytesthat react to self-products. The processworks in the following way. T and Bcells are produced constantly through-out the lifetime of the individual, eventhough the production of T cells mayslow down after adolescence. Regard-less of when they are produced, T andB cells will always develop surroundedby a sea of materials produced by thehost. Those immature cells whose re-ceptors recognize self-products are de-stroyed, according to the theory; conse-quently, only lymphocytes that are notself-reactive will develop to maturity.To be sure, the immature lymphocyteswill also die if they bind to a foreign an-tigen, but an immune response will becarried out by those lymphocytes thatmatured before the infection.

Some time after the clonal deletiontheory was proposed, researchers cameup with two other plausible explanationsfor why the immune system is tolerantto its host. One suggestion was that adeveloping lymphocyte might be inac-tivated permanently, instead of dying,when its receptors were engaged. ( Im-munologists describe an inactive lym-

82 SCIENTIFIC AMERICAN September 1993

IMMATURE T CELLS are killed when their receptors bind to peptides, as shown byexperiments with genetically engineered mice. The mice were designed to harborT cells whose receptors could recognize a peptide made only in male mice. In thefemale mice, T cells developed normally. In the male mice, T cells were absent be-cause the young T cells apparently bound to the peptide and died.


phocyte as Òanergic.Ó) The other hy-pothesis stated that self-reactive T andB cells might be kept at bay by lympho-cytes called suppressor cells.

For many years, researchers strug-gled to distinguish among these threehypotheses. Lymphocytes are clearlyvery good at recognizing foreign tissue.For example, a human being will rejectskin grafts from an unrelated personvery rapidly, whereas an individual willaccept skin tissue transplanted fromone part of the body to another. Like-wise, in culture dishes, lymphocytes failto be activated by other cells from theirhost but react violently to lymphocytesor cells from another individual. Yetthe outstanding issue remained: Doesthe immune system fail to respond toself-products because the potentiallyreactive lymphocytes are simply notthere or are inactive or are being sup-pressed by other cells?

To resolve this question, workers at-tempted to devise methods for identi-fying lymphocytes that recognized par-ticular antigens but did not necessarilyrespond by dividing. The developmentof these techniques turned out to be adaunting task. If the T and B cells of ananimal have never been exposed to aspeciÞc antigen, only a small fractionof the lymphocytes should have the po-tential to react to that antigen. The Òfre-quencyÓ of reacting cells, as immunol-

ogists say, is somewhere around one in a million. Indeed, the frequency is so low that it would be impossible todistinguish the few lymphocytes thatmight recognize self-products from themany that do not.

Nevertheless, researchers have recent-ly developed two experimental toolsthat circumvent this problem. The Þrstrequires a peculiar type of antigencalled a superantigen, whereas the sec-ond relies on genetically altered ani-mals known as transgenic mice. Boththe superantigen techniques and theexperiments with transgenic animalsdeserve to be described in some detail.

Whether a superantigen, or an-tigen, will bind to a lympho-cyte depends ultimately on the

composition and structure of the recep-tor. The alpha-beta receptor, for exam-ple, is a somewhat random assembly of such segments as V-alpha and V-beta.The receptor is designed mainly to rec-ognize foreign peptides, that is, anti-gens made by breaking down proteinsfrom invading organisms. The recep-tor, however, will bind only to a foreignpeptide that has already been attachedto one of the major histocompatibilitycomplex (MHC) proteinsÑspecializedmolecules found on the surface of or-dinary cells. As far as researchers cantell, all the variable segments of the al-

pha-beta receptor play a role in bindingthe MHC protein and its captured for-eign peptide. To recognize a speciÞc an-tigen, a T cell must have a receptor withexactly the right combination of vari-able segments.

Superantigens are a diÝerent story.Like their ordinary counterparts, super-antigens will attach to a particular typeof MHC molecule, but they will thenbind to a speciÞc V-beta segment of analpha-beta receptor, almost regardlessof the structure of the rest of the re-ceptor. Because the number of diÝerenttypes of V-beta segments is small com-pared with the number of diÝerent al-pha-beta receptors, many more T cellsare capable of recognizing a particularsuperantigen than are able to identify aspeciÞc antigen. In mice, for example,the number of different V-beta seg-ments totals about 20, and thus anyparticular V-beta can be found in aboutone in every 20 T cells that have alpha-beta receptors.

Most important, because so many Tcells respond to a speciÞc superanti-gen, researchers can observe the reac-tion. To do so, they Þrst obtain an anti-body that can bind to the V-beta targetof a superantigen. The antibody is thentagged with a molecule that ßuorescesunder ultraviolet light. Hence, the ßuo-rescent antibody will attach to T cellsthat respond to the superantigen, and


DEATH OF YOUNG T CELLS that bind to self-proteins helps toprevent the immune system from attacking the body. T cellsare exposed to most self-products as they develop in the thy-mus. Some self-proteins are made in the thymus, whereas

others are carried there, from organs such as the kidney, bytraveling cells. Those young T cells whose receptors bind toself-products die. Because some self-products never reachthe thymus, some self-reactive cells reach maturity.

PROTEIN

PEPTIDEMHC

CARRIER CELL

CARRIER CELL IMMATURE T CELL

THYMUS CELL

NO BINDING

T CELL DIES

T CELL RESPONSIVETO SELF-PRODUCTS

T CELL DIES

T CELL RESPONSIVETO FOREIGN ANTIGENS

NO BINDING

IMMATURE T CELL

IMMATURE T CELL

IMMATURE T CELL

KIDNEY THYMUS


workers can identify the cells using amicroscope or a ßuorescent-activatedcell sorter.

To test this technique, investigatorsÞrst used a superantigen produced bythe mouse mammary tumor virus. Micebecome infected with the virus throughthe milk of their mothers. The virus in-vades the lymphocytes of mice by man-ufacturing a superantigen and stimulat-ing the lymphocyte. The mouse mam-mary tumor virus, like the virus thatcauses AIDS, is a retrovirus. Such virus-es contain genes made of RNA, but theyreproduce by making DNA copies oftheir RNA. This DNA is inserted intothe DNA of infected cells. The viralDNA then gives rise to viral RNA andproteinsÑmaterials that assemble toform new infectious viruses.

Occasionally, retroviruses infect thecells that produce sperm or eggs. If thathappens, the virus may become part ofthe DNA of the oÝspring and may ceaseto be an infectious organism. In fact,nearly all mice have one or more mam-mary tumor viruses integrated into theirDNA. These viral integrants then gener-ate proteins that are, for all intents andpurposes, self-products. The viral pro-teins are made constantly throughoutthe life span of the animal, just as gen-uine self-proteins are.

We and our colleagues at the Nation-al Jewish Center for Immunology andRespiratory Medicine in Denver usedthe superantigens produced by theseviral integrants to test how the immunesystem responds to self-products, since,as far as the mouse is concerned, theproteins made by the integrant are self-products. In 1988 we and several other

research teams began to examine the ef-fects of the superantigen made by MTV-7, a strain of mammary tumor virus thathas naturally integrated into the DNAof some mice. This superantigen reactswith certain V-beta segments on the re-ceptors of mouse T cells. SpeciÞcally,the superantigen binds to the segmentsknown as V-beta 6, V-beta 7, V-beta 8.1and V-beta 9.

Working with Uwe Staerz, then at theBasel Institute for Immunology, we fo-cused on the eÝects of the MTV-7 super-antigen on V-beta 8.1. We found that inmice whose DNA is free of MTV-7, asmuch as 8 percent of T cells have V-beta8.1 as part of their receptors. On theother hand, mice whose DNA includesMTV-7 did not harbor any mature T

cells with V-beta 8.1. At the same time,a Swiss research teamÑincluding H.Robson MacDonald of the Ludwig Insti-tute for Cancer Research in Lausanneand Rolf M. Zinkernagel and Hans Hen-gartner of the University of ZurichÑre-ported that T cells bearing V-beta 6 arealso missing in mice whose DNA hasMTV-7. More recently Edward Palmerand his colleagues at the National Jew-ish Center obtained similar results forV-beta 9, and we discovered the samefor V-beta 7.

All these experiments showed thatthe superantigen made by integratedMTV-7 somehow leads to the disappear-ance of T cells that can react with thesuperantigen. So, in this case, T cellsthat can recognize self-products areneither inactivated nor suppressed byother cells; the T cells are simply notthere and must have died at somestage in their development.

T cells begin as precursor cells. Dur-ing the fetal stage of animal develop-ment, the precursors originate in theyolk sac or liver, whereas young andadult animals spawn precursors in theirbone marrow. These cells migrate to thethymus where they start to build thegenes that contain the instructions formaking alpha chains, beta chains andother receptor-related proteins. Soonafter, as alpha-beta receptors start toappear in small quantities on the sur-face of the cells, the precursors be-come immature thymocytes. Such cellsthen go through a mysterious stage ofdevelopment referred to as positive se-lection. In this stage the immature thy-mocytes add an increasing number of

alpha-beta receptors to their surfaces.As we discovered, however, an im-

mature thymocyte will die in the thy-mus if it recognizes the superantigenproduced by integrated MTV-7. ManydiÝerent experiments on diÝerent ani-mals have now shown that superanti-gens cause the death of immature thy-mocytes about halfway through theirdevelopment.

Although these results oÝeredstrong support for the clonal de-

letion theory, immunologistswere forced to entertain the possibilitythat superantigens, which have so manyspecial properties, do not aÝect T cellsin the same way as normal antigens do.Harald von Boehmer and Michael Stein-metz and their collaborators at the Uni-versity of Basel and HoÝmannÐLa Rochetherefore approached the problem ofself-tolerance in a completely diÝerentway. They used transgenic mice, whichare produced by the injection of DNAinto fertilized mouse eggs. Such DNA isfrequently incorporated into the DNAof the developing embryo and is passedon to progeny.

Using this technique, von Boehmerand his colleagues created mice inwhich most of the T cells had the samealpha-beta receptor. To understand howthey accomplished this, one must knowthat as a precursor T cell develops in anordinary mouse, the cell actually buildsthe DNA that yields a particular alpha-beta receptor. Pieces of the DNA cod-ing for the receptor are contained with-in the DNA of all mouse cells, but onlyin developing T cells are the pieces rear-ranged into functional genes. Von Boeh-mer isolated DNA that coded for a spe-ciÞc alpha-beta receptor gene, whichhad already been rearranged. He theninjected mouse eggs with the rear-ranged DNA. As the mice developed,the rearranged DNA took precedenceover the unrearranged receptor genes.Hence, most T cells in the transgenicmice bore the receptor created by theinjected genes.

The receptor that von Boehmer choseto make was one that would bind anantigen present only in male mice, pro-vided that the antigen was accompa-nied by the MHC protein Db. In thosemice that made Db, he and his co-work-ers found, as expected, an undeniablediÝerence between the females andmales. In the female mice, many T cellshad the introduced receptor on theirsurfaces. In the male mice, however,cells bearing the receptor were almostcompletely missing; they had appar-ently been destroyed at an early stageof their development in the thymus.

These Þndings show that Lederberg


SUPERANTIGEN binds to only one partof a T cellÕs receptor, in this case the re-gion known as V-beta. Before the super-antigen attaches to the receptor, it mustbind to a major histocompatibility com-plex protein on the surface of a cell .


was right when he proposed the clonaldeletion theory. Immature lymphocytesgo through a stage when binding oftheir receptors causes them to die. Self-reactive cells are killed before they havea chance to proliferate and damage theirhost. The immune system does indeeduse clonal deletion to establish toler-ance to self.

Unfortunately, the clonal deletion the-ory does not address the problem ofhow the immune system learns to tol-erate self-products that the thymus ei-ther does not make or produces in ex-tremely small quantities. This concernapplies not only to proteins that arerelatively sequestered, such as thosemade in the brain or eye, but also toproteins that are made only in certainspecialized tissues.

In fact, many of these unusual self-antigens are transported to the thy-mus. Monocytes and B cells can takeup a protein in one part of the bodyand carry it to anotherÑto the thymusin particular. This kind of process ac-counts very well for how the immunesystem learns to tolerate many self-products that are not generated in thethymus.

Yet this scheme does not apply in allcases; for instance, it does not explainhow T cells learn to tolerate peptidesthat bind to class I MHC proteins. SuchMHC proteins attach only to peptidesthat are derived from proteins madewithin the cell itself. Monocytes and B

cells are thus incapable of transportingthe peptides of other cells to the thy-mus. Immature T cells in the thymus arenot exposed to some cytoplasmic pro-teins and so have no way of learning totolerate them. Some other mechanismmust be at work that either kills, inacti-vates or suppresses mature T cells.

Scientists have searched for amechanism by which mature T

cells could learn tolerance. SeveraldiÝerent experiments have now shownthat when mature T cells encounterself-products, they may either die orbecome inactive.

One such experiment was conductedby Jacques F.A.P. Miller and his col-leagues at the Walter and Eliza Hall In-stitute of Medical Research in Austra-lia. They worked with a gene for a classI MHC protein known as Kb. They intro-duced this gene into a mouse in such away that the gene was controlled by theinsulin gene. Hence, the mouse madeKb only in its cells that normally makeinsulin, that is, the beta cells of thepancreas. Because these cells are im-mobile, Kb could not Þnd its way to thethymus of these animals, and, not sur-prisingly, the thymocytes of these micecould bind to Kb. Mature T cells couldnot respond, however, unless they wereconfronted with Kb under very specialcircumstances. These results showedthat sometimes mature T cells that canrecognize self-antigens can survive in

animals but become anergic. Further-more, in other experiments Susan Webband her collaborators at the Scripps Re-search Institute in La Jolla, Calif., haveshown that under some conditions, ma-ture T cells die when exposed to self-antigen.

Immunologists do not know exactlywhat causes death rather than inactiva-tion of mature T cells. Perhaps the in-active cells are just intermediates ontheir way to the grave. In any case, theimportant outcome is that these cellscannot respond. Indeed, researchershave now gathered so many examplesof antigensÕ causing the death or inacti-vation of T cells that they are quitepuzzled about why and when antigensfrom invading organisms cause T cellsto become active. T cells with alpha-beta receptors seem to have been de-signed so that they will not usually beactivated when their receptors are en-gaged. The question is, therefore, howa mature T cell decides whether to di-vide, become inactive or die when itsreceptors react to something.

The issue was resolved in part by aclassic experiment performed some 30years ago by David W. Dresser, then atthe Medical Research Council Laborato-ries in England. At the time, investiga-tors were well aware that the immunesystem responds vigorously to aggregat-ed preparations of foreign protein or toprotein mixed with an adjuvant, suchas dead bacteria in mineral oil. Dress-


IMMUNE SYSTEM has a safety mechanism that prevents amature T cell from mounting an immune attack against itshost. Before a T cell can attack, it must receive two signals.The Þrst is the binding of an antigen to the T cellÕs receptor.

The second is typically the secretion and binding of a pro-tein, B7, for example (left). If a T cell is exposed to a self-pro-tein that is presented on a nonstimulatory cell, the T cell willdie or become inactive (right).


er found, however, that the immunesystem fails to respond to soluble for-eign proteins. In fact, once the immunesystem is exposed to a soluble foreignprotein, it will subsequently fail to re-act to any preparation of that protein.The system learns to tolerate solubleforeign proteins, at least in part, be-cause it eliminates the T cells that canrespond to such proteins, as was dis-covered in 1971 by Jacques M. Chillerand William O. Weigle of the ScrippsClinic and Roger Taylor of the MRCLaboratories.

Evidently, T cells can recognize theform of the antigen, although how theydo so was not clear until recently. Al-pha-beta receptors do not have a directway of detecting in what form the for-eign protein was introduced into thebody. As long as a peptide is attachedto an MHC protein, it can bind to an al-pha-beta receptor regardless of wheth-er the peptide came from a protein insolution or in an adjuvant. Somethingbesides the alpha-beta receptor mustbe giving the T cell information aboutthe form of the antigen.

In 1970 Peter A. Bretscher and Mel-vin Cohn of the Salk Institute for Bio-logical Studies suggested a solution tothe problem in an early form, and Þveyears later Kevin J. LaÝerty and Ali-stair J. Cunningham of the John CurtinSchool of Medical Research in Australia

reformulated the idea into its currentlyaccepted form. In immunologic terms,T cells need two signals to be stimulat-ed by an antigen: the Þrst signal comesfrom the binding of the alpha-beta re-ceptor, and the second is from some-thing else.

The task of identifying this secondsignal has preoccupied immunologistsfor the past decade. One clue came fromthe work of Ronald H. Schwartz and hiscollaborators at the National Institutesof Health [see ÒT Cell Anergy,Ó by Ron-ald H. Schwartz; SCIENTIFIC AMERICAN,August]. In 1987 his group demonstrat-ed that antigens bound to MHC pro-teins will not provoke cultured T cellsto divide if the cells that bear thoseMHC proteins are prepared in a certainway. Not only did the T cells fail to re-spond, but they were also unable to di-vide several days later when confrontedwith antigens bound to MHC proteinson live, unprepared cells. The preparedcells bearing the MHC proteins hadsomehow inactivated the T cells.

Later work has shown that the mech-anism of inactivation involves CD28, aprotein on the surface of T cells. WhenCD28 binds to a protein known as B7 orBB1Ñwhich resides on the surface of Bcells and macrophagesÑit delivers a sig-nal to the T cell. Normally, a T cell bindsan antigen and an MHC protein and aB7 at the same time. Consequently, the

T cell gets two signals: one through itsreceptor and another through CD28. If aT cell is confronted with an antigen ona cell that does not have functional B7,it will get the receptor signal withoutthe CD28 signal. The T cell is therebyinactivated. This Þnding lends strongsupport for the theory that T cells re-quire two signals to respond to an anti-gen. (It is worth emphasizing, however,that CD28-B7 is only one of many pos-sible secondary signals.)

Of course, the way the CD28-B7 signalis blocked in the laboratory is not theway it happens in the immune system.Researchers are still Þguring out whatcells can present antigens on MHC pro-teins but fail to deliver the C28-B7 sig-nal. The answer may be B cells, as sug-gested by such investigators as DavidC. Parker of the University of Massachu-setts at Worcester and Polly C. Matzing-er of the National Institute of Allergyand Infectious Diseases. Most B cells inanimals bear very little, if any, B7. Onlyafter the B cells have been stimulatedthemselves do they increase their pro-duction of B7 to measurable amounts.Therefore, if T cells encounter antigenon B cells, they may be inactivated be-cause they receive one signal withoutanother. At the present, this theory seemsplausible, but it has not been proved.Most likely, only certain specializedcellsÑmacrophages, dendritic cells andperhaps activated B cellsÑcan deliverboth the Þrst and second signals to Tcells, thereby activating them.

The body has many ways to dealwith a self-reactive T cell that hasan alpha-beta receptor. If the cell

is still developing in the thymus and itsreceptor binds to a self-product, it willdie. On the other hand, a mature cellwhose receptor binds to a self-productwill be inactivated or killed if it fails toreceive a second message, such as theCD28-B7 signal. Investigators are lesscertain about how B cells and T cellswith gamma-delta receptors respond toself-products.

The immune system seems to han-dle self-reactive B cells and their immu-noglobulin proteins in much the sameway as it takes care of self-reactive T cells with alpha-beta receptors. For example, in 1976 Norman R. Klinmanand his colleagues at the University ofPennsylvania and Ellen Vitetta and herco-workers at the University of TexasSouthwestern Medical Center at Dallasindependently showed that immature Bcells in tissue culture could be madetolerant to antigen much more easilythan mature B cells could. Later, SirGustav J. V. Nossal and his collabora-tors at the Hall Institute showed that


YOUNG B CELLS become inactive when they bind to something, as shown by ex-periments with transgenic mice. A group of mice was engineered to produce a pro-tein found in chicken: hen egg lysozyme (HEL). Other mice were designed to makeB cells with receptors that bind to HEL. The two groups of mice were mated, pro-ducing mice that produced HEL and B cells sensitive to HEL. B cells in these ani-mals could not be activated by reagents that usually stimulate such cells.


this phenomenon could involve B cellanergy as well as B cell death.

Recently researchers have been able todemonstrate this phenomenon in animalsrather than in tissue culture dishes, us-ing transgenic mice. For the experiments,they injected fertilized mouse eggs withfully rearranged, mature genes for a par-ticular immunoglobulin, thereby intro-ducing the genes into the mouse DNA.As B cells matured in the developingmice, the introduced genes caused al-most all the B cells to produce the spe-cific immunoglobulin on their surface.

Among the Þrst researchers to trythis technique were ChristopherC. Goodnow and his colleagues

at the University of Sydney. The work-ers created two groups of transgenicanimals. One contained the genes foran immunoglobulin that binds to theforeign protein hen egg lysozyme. Theother group contained a gene that in-structs cells to produce hen egg lyso-zyme. When mice from one group wereallowed to mate with mice from theother group, they produced oÝspringwhose DNA contained both types ofgene. Hence, the oÝspring had the abil-ity to make both hen egg lysozyme andthe immunoglobulin that binds to henegg lysozyme. The investigators foundthat B cells in the oÝspring were inacti-vated, conÞrming the results found intissue culture dishes.

A similar experiment was performedby David Nemazee, then at the Basel In-stitute for Immunology, and Kurt Buer-ki of Sandoz Pharma. They made trans-genic mice with a gene for an immu-noglobulin molecule that binds to theMHC protein Db. Some of the transgen-ic mice naturally produce Db in their

bone marrow. In those mice, the B

cells died.Why are the immature B cells some-

times inactivated and sometimes killedby contact with self-antigen? Investiga-tors have found that the form of theantigen determines the fate of the cells.Soluble antigens, such as hen egglysozyme, are more likely to inactivateimmature B cells that bind to them. Onthe other hand, cell-bound aggregatedantigens, such as Db, are more likely tokill immature B cells.

All in all, tolerance mechanisms for Bcells are very similar to those for alpha-beta T cells. Immature B cells die or areinactivated when their receptors bindto something. Whether tolerance canalso be imposed on mature B cells, as it can on mature T cells, remains to bedetermined.

In contrast to B cells, T cells thathave gamma-delta receptors are a mys-tery. In humans and mice, there areabout as many of these cells as B cellsor alpha-beta T cells, and the gamma-delta T cells are clearly very important.Yet scientists have only an incompletepicture of how these cells contribute to the immune system. Gamma-delta Tcells seem to respond to products thatare made by the host when, for exam-ple, tissue is abraded, overheated, ex-posed to dangerous metals or attackedby invading organisms. The receptorson these cells appear to have been de-signed particularly to bind to certaincomponents of self. If this is true, inves-tigators are then faced with the prob-lem of how these self-reactive cells arekept under control. At the moment,they have no idea.

A healthy immune system does notattack its own host. Unfortunately, the

immune system makes mistakes, and Tcells and B cells that can respond toself-antigens sometimes appear. Thesecells damage the body they occupy,leading to such diseases as rheumatoidarthritis, multiple sclerosis and lupus.At the moment, physicians treat thesediseases by waging a full-scale battleagainst the immune cells that causethem. Powerful anti-inßammatory drugsand chemicals that kill or slow downactivated T and B cells are used to keepthe autoimmune response in check.Sadly, these methods are not always ef-fective and in some cases have unwant-ed side eÝects. Immunologists hopethat if they continue to study how theimmune system learns to tolerate thebody, they will Þnd ways to improvethe treatment of autoimmune diseases.


FURTHER READING

ACTIVELY ACQUIRED TOLERANCE OF FOR-EIGN CELLS. R. E. Billingham, L. Brent andP. B. Medawar in Nature, Vol. 172, No.4379, pages 603Ð606; October 3, 1953.

T CELL TOLERANCE BY CLONAL ELIMINA-TION IN THE THYMUS. J. W. Kappler, N.Roehm and P. Marrack in Cell, Vol. 49,No. 2, pages 273Ð280; April 24, 1987.

A CELL CULTURE MODEL FOR T LYMPHO-CYTE CLONAL ANERGY. R. H. Schwartzin Science, Vol. 248, pages 1349Ð1356;June 15, 1990.

THE NEED FOR CENTRAL AND PERIPHERALTOLERANCE IN THE B CELL REPERTOIRE.C. C. Goodnow, S. Adelstein and A. Bas-ten in Science, Vol. 248, pages 1373Ð1379; June 15, 1990.

ABLATION OF ÒTOLERANCEÓ AND INDUC-TION OF DIABETES BY VIRUS INFECTIONIN VIRAL ANTIGEN TRANSGENIC MICE. P. S. Ohashi et al. in Cell, Vol. 645, No.2, pages 305Ð317; April 19, 1991.

QUINTUPLET COWS gave some of the Þrst evidence that theimmune system learns to tolerate self-products. The fetal cows

shared a single placenta in their motherÕs womb and exchangedblood. As a result, the calfs accepted blood from one another.


Documents

How the Immune System Recognizes the Body