The intestinal immune system · to distant mucosal sites andthen re-enter the lymph to recirculate.'7 Peripheral lymphoidtissues werenot examined in these experiments but other studies

Gut, 1989, 30, 1679-1685

The intestinal immune systemWILLIAM F DOE

(This article is one of a series linked with the Festschrift for Christopher Boot/i. See Gut Festschrift 1989; 30.)

The concept of localised intestinal immunity has itsorigins in Besredka's observation that oral immunisa-tion using killed salmonella organisms provided solidprotection against dysenteric infection irrespective ofthe titres of serum antibody.' But the clinical benefitswere variable and short lived and interest lapsed.Little further progress was made until Heremans'discovery of IgA and the marked predominance ofIgA-containing plasma cells in normal intestinalmucosa.2 The differences between the antigenicdeterminants of serum and secretory IgA led to theidentification of an additional polypeptide chain, thesecretory component (SC), which is essential to thesecretion and function of IgA antibody at mucosalsurfaces.'The importance of the mucosal immune system to

the pathogenesis of intestinal disease was swiftlyrecognised by Chris Booth. When I arrived as ahouseman at Hammersmith in 1969, several pio-neering immunological studies of coeliac disease hadbeen published. The clinical research ethos atHammersmith stimulated the study of human diseasenot only by applying basic scientific knowledge andtechniques but also by recognising that careful studyof human disease can also provide opportunities foradvancing our understanding of human biology. Irecall the outpatient clinic in 1969 when Chris handedme the referral letter for a new patient he had notseen: 'Fascinating, sounds like alpha chain disease.'

Several months later Heremans came to Hammer-smith to deliver a series of outstanding lectures on thesecretory immune system. Grand rounds that weekincluded the presentation of what was to be one of theearliest published cases of alpha chain disease.4

In the decades since these early discoveries, majoradvances in understanding the physiology of theintestinal immune response and the advent of mole-cular biology have helped to elucidate the inductionand regulation of the secretory antibody and cellularresponses at mucosal surfaces.

Gut associated lymphoid tissue (GALT)

The exposed surface of the intestinal mucosa is underconstant challenge by ingested foreign antigens inmicro-organisms, products of food digestion anddrugs. It is therefore not surprising that the intestines

contain the largest accumulation of lymphoid tissuesin the body' in the form of lymphoid aggregates inPeyer's patches and in the lamina propria (solitarylymphoid nodules) and as the scattered lymphocytepopulations found in the epithelium and in the laminapropria.Although nascent Peyer's patches are evident in

the newborn, the epithelium and lamina propria aredevoid of mononuclear cells. T lymphocytes mi-grating from the thymus rapidly populate the thymus- dependent areas of Peyer's patches and theepithelium but exposure to micro-organisms in thenormal environment is necessary to develop the Bcell population and their germinal follicles as shownby experiments conducted in germ free animals.6

Induction of the secretory immune response

Non-immune and immune mechanisms protect theprivileged environment of the lamina propria fromchallenge by foreign antigens. Gastric acidity, diges-tive proteases in the gastrointestinal tract, intestinalmobility, the commensal microflora and the mucouscoat or glycocolyx comprise some of the non-specificprotective barriers. The immune mechanisms mayoperate within the lumen of the gut, at the mucosalsurface or within the lamina propria. The inter-epithelial lymphocyte (IEL) population are pre-dominantly suppressor lymphocytes (Ts) in contrastwith the lamina propria and show evidence ofactivation whereas most of the lamina proprialymphocytes belong to the helper-inducer subset(Th). Class II MHC determinants which represent arestriction element in T cell dependent immuneresponses are expressed on normal small intestinalepithelial cells but not colon epithelial cells unless thecolon is inflamed. Expression of class II antigens ismodulated by lymphokine products of activated Tcells especially interferon y (IFNy). Bland andWarren7 reported that MHC class Il positive villousepithelial cells can present soluble antigen to primedT cells leading to antigen specific suppression.Although human colonic epithelial cells have beenreported to act as stimulators in autologous andallogeneic responses8 others have been unable tostimulate T cells in an allogeneic system using anMHC class II positive colon cancer cell line.'

1679

on February 9, 2020 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.30.12.1679 on 1 D

ecember 1989. D

ownloaded from

William F Doe

Whether IEL require a second signal fromaccessory cells in the lamina propria remains to beanswered.Most studies of isolated GALT cells have been

performed using Peyer's patch cells. Whether thesereflect the function of all organised GALT is un-known. The dome of Peyer's patches is covered by aunique epithelium comprising cuboidal epithelialcells which express class II MHC antigens, very fewgoblet cells and specialised antigen-sampling cellscalled M (for microfold) cells."' M cells pinocytosesoluble antigens such as ferritin and horse-radishperoxidase or phagocytose particulate foreignantigens including viruses and whole bacteria, andtransport them intact across the epithelium to theunderlying lymphoreticular cells in the dome."Macrophages and dendritic cells are present in the

dome region, and MHC class II positive cells dis-playing dendritic cell morphology are also found inthe T and B cell zones. Dendritic cells found in cellsuspensions of Peyer's patches are fully able topresent antigen in vitro"' and those present in thelamina propria of the small and large intestine inmouse and man are also competent to presentantigens. Whether Peyer's patch macrophages alsofunction as antigen presenting cells, however, isuncertain, as lamina propria macrophages suppressantigen presentation by dendritic cells.'1

Peyer's patches contain a higher proportion of Bcells than peripheral nodes. While IgM-bearing Bcells predominate, there is significant enrichment forB cells displaying surface lgA and committed to IgAsynthesis consistent with the role ofGALT as a majorsite for the induction of IgA responses. Peyer'spatches are also greatly enriched for T cells of thehelper-inducer subset (Th) although suppressor-cytotoxic T cells (Ts) and the regulatory contra-suppressor T cells (Tcs) which appear to potentiateimmune responses to orally presented antigen, arealso found. The high degree of preferential localis-ation of B cells and IgA specific Th cells in Peyer'spatches is probably determined by the specificity ofbinding of these lymphocytes to the high endothelialvenule receptors present in postcapillary venules inPeyer's patches'4 which determine their represen-tation in Peyer's patches and ultimately regulate thenature of the immune response generated.Yet induction of mucosal immunity and the delivery

of the immune response is not entirely localised to theintestine. After antigen exposure, Th cells and B cellscommitted to specific IgA synthesis are rapidlygenerated in Peyer's patches and travel via themesenteric node into the thoracic duct to enter thecirculation before selectively migrating to mucosalsurfaces. Whether this inability to mount a completelylocalised immune response to antigen encountered at

a mucosal surface relates to separation of the T and Bcell zones in Peyer's patches from antigen presentingcells is not known. Mestecky and McGhee'` havespeculated that antigen presentation to lymphocytesdoes occur in GALT resulting in an 'initial induction'of immune responses and that the 'terminal inductivestimuli' may occur at distant mucosal sites after thesecommitted lymphocytes have migrated there.The migratory characteristics of IgA-committed

lymphoblasts and memory cells are quite distinct.Unlike memory lymphocytes, IgA-containinglymphoblasts do not recirculate but 'home' to theorgan of antigenic stimulation, secrete antibody,remain associated with the mucosal target tissue andprobably die within a few days.'` In rats primed andchallenged by intraintestinal injection of choleratoxin, specific antibody - containing lymphoblastswere found three to five days later in the thoracic ductlymph from where they entered the circulation andthen populated distant mucosal sites. The number oflymphoblasts in the thoracic duct greatly diminishedafter five days but the number of small lymphocytescapable of transferring specific immunity to naiveanimals (memory cells) peaked after two weeks andpersisted. The memory cells, which will produce T-and B-lymphoblasts when stimulated by specificantigen, enter the circulation from the lymph, 'home'to distant mucosal sites and then re-enter the lymphto recirculate.'7 Peripheral lymphoid tissues were notexamined in these experiments but other studiessuggest that memory lymphocytes from the thoracicduct also 'home' preferentially to mucosal lymphoidtissue.Lymphocyte 'homing' is not dependent on the

presence of antigen. IgA lymphoblasts migrate tofetal isografts of small intestine transplanted under-neath the kidney capsule of an adult mouse.'` Whenrats primed by intracolonic injection of cholera toxinwere challenged 14 days later by injecting antigeninto a surgically isolated loop of jejunum with anintact blood and lymph supply, blast cells appeared incomparable numbers in the stimulated intestinal loopand in the unstimulated jejunum. By cannulating anddraining the thoracic duct, the researchers confirmedthat antigen stimulation of memory cells in theisolated jejunal loop resulted in the production oflymphoblasts that entered the lymph and circulatedto populate the entire jejunum.'7 Although memorycells recirculate to populate the mucosal tissues,general mucosal dissemination of lymphoblasts com-mitted to IgA synthesis does not occur after secondarychallenge. When thoracic duct lymphoblasts fromanimals both primed and challenged by intracolonicinjection of cholera toxin were transferred into naiverecipients, large numbers of antibody specific plasmacells populated the colon but there were virtually

1680

on February 9, 2020 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.30.12.1679 on 1 D

ecember 1989. D

ownloaded from

The intestinal immune system

none found in the jejunum. These studies suggestthat the lymphoblasts can distinguish between highendothelial cell receptors in the colon and jejunumindicating that the colon receptors have their own'homing' specificity. The presence of antigen, how-ever, profoundly affects the magnitude and persist-ence of the response due to antigen-driven clonalexpansion of antibody forming cells and their precur-sors.I'

Intestinal fluid immunoglobulins

A specialised form of IgA, secretory IgA, representsthe predominant antibody isotype in intestinal fluidwhere the relative concentrations of the majorimmunoglobulin isotypes mirror their relativedistribution in the antibody-containing cells of theintestinal mucosa.' Unlike IgA in the serum andcerebrospinal fluid which largely comprises 7S mono-mers, secretory IgA is an 11S dimer made up of twoIgA monomers joined by a covalently linked peptidenamed J chain.' Both IgA and IgM plasma cellssynthesise J chain which greatly enhances the bindingof an additional polypeptide, secretory component(SC), thereby completing the assembly of secretoryIgA and IgM molecules. The details of secretory IgAstructure have recently been reviewed." J chain is anacidic peptide of 137 residues which may be involvedin the intracellular polymerisation of IgA. Onemolecule of J chain joins two IgA monomers whereasthree or more J chain molecules are involved informing the IgM pentamer within the plasma cellbefore secretion.

Secretory component (SC) is a glycopeptidemolecule synthesised by epithelial cells in secretoryglandular tissue and mucosal surfaces. In somemammalian species, but not man, hepatocytes alsosynthesise and display SC. Secretory componentexists in three molecular forms; as a membraneprotein expressed on the outer surface of epithelialcells, it acts as a receptor for polymeric immuno-globulins. Secretory component forms part ofsecretory IgA and IgM molecules in mucosal fluidswhere it is also found as a free glycoprotein. Secretorycomponent interacts with a specific binding site onthe Fc region of SIgA and IgM, stabilises theirquaternary structures, and increases their resistanceto proteolytic digestion.' Structural studies of humanSC reveal a characteristic composition, comprising549 to 558 amino acids and a high carbohydratecontent (20%). Twenty cysteine residues create 10intrachain disulphide bridges in a single polypeptidechain that has five regions of remarkable homology.The fifth homology region contains an additionallabile disulphide bond that is involved in the for-mation of a disulphide bridge linking SC to the two

chains of one of the monomers in dimeric IgA. Thehomology regions of SC are structurally related to thevariable domain of immunoglobulin light chains andprobably have a similar tertiary structure. Since thesecondary structure of J chain also shows immuno-globulin like folding, the interactions of a, ,u and Jchains with SC may be based on the complementariesof their domain-like structures.2" Both the additionalpeptides SC and J chain bind covalently to the Fcregion of IgA and IgM.The constant domains of the a chains of the two

subclasses of IgA, IgA, and IgA2 are very similar.The C terminus of the a chain resembles that of the ,uchain and extends further than the C termini of the y,£ and o chains of IgG, IgE and IgD strongly suggest-ing that this extension is implicated in the ability ofIgA and IgM to form polymers."The predominant structural differences between

IgA, and lgA2 occur in the hinge region where IgA,has a unique sequence of 13 amino acids that aresusceptible to cleavage by highly specific bacterialIgA proteases. The corresponding sequence is lackingin IgA2 molecules which are therefore resistant tonearly all known bacterial proteases with the excep-tion of a clostridial IgA protease that cleaves apeptide bond in the hinge region of both IgA1 andIgA2 of the A2m(1) allotype.2'

Plasma cells producing 1OS IgA are found in allmucosal tissues especially in the intestinal laminapropria whereas plasma cells in the bone marrow,spleen and peripheral lymph nodes secrete mono-meric (7S) lgA. There is also a selective distributionof cells producing the IgAI and IgA2 subclasses. IgA1plasma cells form the majority in all tissues apartfrom the colon but there is marked variation in theratio of lgA, to IgA2 plasma cells between lymphoidtissues from different sites. That IgA2 is the prepon-derant subclass in the colon may relate to localstimulation by endotoxins from commensal gramnegative organisms which mainly induce an IgA2subclass antibody response resistant to bacterialproteases."

Secretory IgA function

The protective role of secretory IgA against a varietyof foreign antigens, including food antigens bacteriaviruses and toxins, is well established. S-lgA blocksthe access of potentially allergenic molecules derivedfrom food or drugs. S-IgA mediated responses atmucosal surfaces to locally encountered antigens arereadily detected but prolonged topical exposure toprotein or particulate antigens is required to induceserum antibodies.`2 Antibodies of all three majorisotypes specific for food and microbial antigens are

1681

on February 9, 2020 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.30.12.1679 on 1 D

ecember 1989. D

ownloaded from

William F Doe

detectable in the sera of normal individuals. They aremostly IgA, class and very high serum titres arefound in IgA deficient patients. S-IgA may also blockreaginic sensitivity reactions at the surface of theintestinal epithelium as reported for nasal epithelium.Because some dietary antigen is clearly absorbed bynormal subjects, the importance of S-IgA antibodymay lie in reducing the amount of antigen that gainsaccess to the lamina propria.

S-IgA can neutralise biologically active antigens,including viruses and toxins and enzymes, preventadherence of bacterial pathogens to epithelial sur-faces and enhance the antibacterial efficiency ofother immune effector systems. The effectiveness ofS-IgA as a neutralising antibody against viruses isshown by the responses to the Sabin oral livepoliovirus vaccine where protection correlates withlevels of secretory antibody.` The four antigencombining sites of S-lgA also allow it to functionefficiently as an agglutinin and the capacity of S-IgAto protect by inhibiting bacterial adherence to themucosal epithelium is exemplified by S-lgA anti-bodies to Vibrio cholera. There is also evidence thatIgA has bacteriocidal potential in cooperation withcomplement and lysozyme and S-lgA can opsonisebacteria as outlined below.

Unlike IgM or IgG, however, secretory IgA doesnot recruit the powerful mediator systems that areactivated in systemic immune responses. In partic-ular, sIgA-antigen complexes do not activate theclassical or alternative pathways of complement.'4Chemically aggregated IgA in high concentrationdoes activate the alternative pathway but the bio-logical relevance of the finding is questionable.IgA and IgA-containing immune complexes are

capable of interacting with neutrophils, mononuclearphagocytes, epithelial cells lining mucous mem-branes, T, B and NK cells. The Fc receptors for Cot2domain of polymeric and monomeric IgA and forsIgA of both subclasses that are displayed by sub-populations of neutrophils and mononuclearphagocytes increase in number when these cells areexposed to high levels of IgA. These findings mayexplain the enhanced phagocytic activity of mucosalneutrophils compared to those from the circulation.'`The functional significance of Fco receptors onphagocytes is uncertain. Earlier reports suggestedthat IgA had little capacity as an opsonin, but recentstudies indicate that particles coated with IgA arereadily phagocytosed by neutrophils from the oralcavity." Neutrophils may also clear IgA complexesfrom the blood. Macrophages in colostrum and milkcontain large amounts of S-IgA suggesting that theyprovide the means for transporting S-IgA into thelower gastrointestinal tract of breastfed infants.There is also evidence that IgA can 'arm' neutrophils

and Th cells to provide cell mediated antibacterialactivity.2'The secretory component (SC) displayed on the

basolateral membrane of mucosal epithelial cells actsas a receptor for polymeric IgA and IgM by bindingto a J chain-related site. The SC-polymeric IgAcomplex is internalised by the epithelial cell andtransported as a vesicle to its apical surface where theSC-IgA complex is released.' Although humanhepatocytes do not express surface SC they do havesurface membrane receptors specific for asialoglyco-proteins called hepatic binding proteins that bindhuman polymeric IgA, and to a lesser extent mono-meric lgAl. lgA2 does not inhibit this interaction.The human liver rapidly clears S-IgAl from thecirculation indicating that the hepatic binding proteinacts as a receptor for S-IgA, uptake.29 Unlike thosespecies that have hepatocytes bearing SC, humans donot transport large quantities of circulating IgAacross the hepatocyte and into the bile. In humanbile, only about 50% of the polymeric IgA is derivedfrom the plasma.

Regulation of S-IgA immune responses

Many characteristics of the S-IgA system clearlydistinguish it from systemic immunity. GALT con-tains all the cells required for local responses and isconstantly exposed to antigenic stimulation and tobiological active bacterial components such asendotoxin, but a number of negative regulatoryfactors suppress the induction of a localised immuneresponse. There is selective priming of GALT B cellsleading to marked restriction of the isotype potentialof B cells to IgA expression.29 Endogenous gutendotoxin appears to enhance the host's resistance togram negative infections possibly by interacting withlymphoreticular cells in GALT resulting in inductionof innate host immunity to infection.The IgA response to antigens including carbo-

hydrate antigens is controlled by T cells. Peyer'spatch T cells stimulated by Conconavalin A help IgAexpression but suppress IgM and IgG in LPS-drivenB cell cultures whereas similarly treated splenic Tcells suppress all three isotypes.90 Oral administrationof soluble or particulate antigens induces Th cells inGALT which are specific for the IgA isotype andprobably account for the IgA isotype response thatresults from oral immunisation.Two separate modes of T cell regulation for the

IgA response apparently exist in GALT. T cells mayinduce B cells to switch isotypes (Tsw)3' by promotingthe switch of B cells bearing surface IgM (sIgM+) tocells expressing surface IgA (sIgA) - events that areinfluenced by factors released by activated T cellsincluding interleukins 4 and 5. A second mode of T

1682

on February 9, 2020 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.30.12.1679 on 1 D

ecember 1989. D

ownloaded from

The intestinal immune system

cell regulation of IgA responses may involve Th cellsbearing Fc receptors for IgA (Fcca) which preferen-tially interact with sIgA'B cells and differentiatethem into IgA-producing plasma cells.32 In mice, Fcctreceptors (FcaR) are also found on some Ts cells thatcan release the FcaR as an IgA-binding factor (IBFa)resulting in suppression of IgA synthesis in pokeweedmitogen driven spleen cell cultures. On exposure tolgA, activated FcctRT cells released IBFa andsuppressed IgA responses. Supernatants of FcaIRhybrids created by fusing Peyer's patch Th cells with aT cell lymphoma line, supported antigen-dependentIgA responses in cultures of normal Peyer's patch Bcells whereas FccR- hybrid supernatants did not.When IBFa was purified from the supernatants, highconcentrations suppressed T cell-dependent IgAresponses while low levels of IBFa enhanced IgAresponses.33 These results indicate that T cell selec-tion and induction of sIgA+ B cells involves FcaR andthat high concentrations of IBFct have a paradoxicalsuppressor effect. Whether these effects derive fromdifferent concentrations of the same IBFa moleculeor whether, as has been shown for IgE, there are twoclasses of IBF, one suppressor the other helper, thatdiffer in their carbohydrate moieties,34 has not beenestablished. Thus clones of T cells may produce bothsurface Fca receptors and a secretory form(s) ofIBFct which markedly influence the responses ofsIgA+ cells in a complex network to generate IgAspecific immunoregulation at exposed mucosal sitesinvolving selective homing mechanisms for redistri-bution of IgA-committed cells.

T cell networks and oral tolerance

Two separate immune responses follow oral admini-stration of antigen in large doses; S-IgA antibodiesare generated at mucosal surfaces and unresponsive-ness to the same antigen - that is, tolerance, developssystemically. The simultaneous generation of Th andTs cells specific for the same antigen implies anotherlevel of control which is apparently provided by thecontra-suppressor T cell (Tcs). Indeed Tcs, which areenriched along with Th cells in Peyer's patches, canabolish oral tolerance when adoptively transferred toan animal orally immunised by the same antigen. IgAisotype-specific Tcs cells have recently been reportedin Peyer's patches. These Tcs bind to immobilisedFcaR+ helper cells and, when adoptively transferred,also abolish oral tolerance.Although circulating immune complexes33 and

anti-idiotypic antibodies37 have also been proposed asmechanisms mediating oral tolerance, the evidencesuggests that a T cell suppression network plays themajor part in the downregulation of systemic immuneresponses.

Mucosal cellular responses

Studies of effector cell populations in isolated laminapropria mononuclear cells have identified markeddifferences in their relative frequency, functionalstate and phenotype in comparison to correspondingcell populations in the peripheral blood. Laminapropria lymphocytes show enhanced proliferationafter exposure to interleukin 2 (IL-2), a much higherproportion of IL-2 receptor-bearing cells in the Thand Ts subpopulations and an increased proportionof T cells expressing class II antigens of the majorhistocompatibility complex. Moreover, IL-2 receptormRNA is detectable in unstimulated lamina proprialymphocytes. But there is no evidence of IL-2 mRNAexpression. Peripheral blood lymphocytes do notexpress either mRNA.33 Mucosal T cells thereforeselectively express some of the genes associated withT cell activation resulting in 'partial' activation or'primed' mucosal T cells that may result from unique,possibly environmental activation signals in the intes-tine or altered regulation ofT cell function at mucosalsurfaces.

Mucosal cell mediated cytotoxicity unrestricted bythe MHC is also markedly distinct from that found inperipheral blood. Natural killer cells represent 2-3%of dispersed lamina propria cells, as identified by theNKH-1 monoclonal antibody, but are Leu 11`.Natural killer activity is very low or undetectable inunseparated lamina propria cells4`4 but can bediscerned in purified NKH-1+ cells. Mucosallymphokine activated killer (LAK) cells also expressa different phenotype to those in peripheral bloodbeing T11+, NKH-1-, Leu 11-, T4-, T3-, T11-, andT8- 33 41 and high levels of cytotoxicity to cell lines andto freshly isolated cells can be generated afterexposure of lamina propria lymphocytes to IL-2 inculture for 72h.4" 4' Non-specific cytolytic cellsactivated by lectins which are also present may wellbe a variant of the LAK cell. Although the numbersand cytotoxic activity of mucosal NK cells are verylimited, it remains possible that other NK cellcapabilities, including resisting viral infection, andimmunoregulation may be relevant to mucosalhomeostasis. Lymphokine activated killer cells lysesolid tumour cells in vitro but whether the conditionsexist to generate this population in vivo is unclear.Markers of T cell activation are markedly increasedin lamina propria lymphocytes compared to thosefrom peripheral blood. These include proliferativeresponses to IL-2, IL-2 receptor expression andfunction and expression of class II MHC antigens butthere is no enhancement of IL-2 synthesis. MucosalNK cells and other Fc-bearing cells including macro-phages may mediate antibody-dependent cytotoxicity(ADCC). Antibody-dependent cytotoxicity has been

1683

on February 9, 2020 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.30.12.1679 on 1 D

ecember 1989. D

ownloaded from

1684 William F Doe

shown in lamina propria cells but its importance tomucosal inflammation remains to be assessed.An indirect measure of in vitro primed cytotoxic T

cell activity that is not MHC restricted may beobtained using antibodies to the CD3 component ofthe human T cell receptor.4" High levels of cyto-toxicity were found in lamina propria cells fromhistologically normal colon possibly reflecting theexposure of the mucosa to consistent and diverseantigenic challenges. These effector cells display theCD2+, CD3+, CD8+, CD4-, CD16- and Leu7-phenotype and the pattern of target cell specificitymay be determined by the CD45 antigen which iscoupled to the trigger process for NK lysis of sometarget cells.43 This cell population may represent invitro primed cytotoxic T cells reactive against un-determined antigens including viruses.

Conclusion

Clinical research involving the detailed study ofindividual patients has made many important andoriginal contributions to understanding of thephysiology of the intestinal immune system in healthand disease. In this context, the study of individualpatients suffering from alpha chain disease (ctCD)represents a paradigm. The importance of diffuse,small intestinal lymphomas called Mediterraneanlymphomas (ML) as a cause of malabsorption in theMiddle East had been recognised for many years, butunderstanding was limited to descriptive accounts ofthe morphology of the cellular infiltrate as seen instained paraffin sections of intestinal tissue. Whenindividual patients in Paris' and at Hammersmith44'were investigated, it was discovered that the diffusecellular infiltrate represented a monoclonal expan-sion of benign appearing plasma cells whichsynthesised, and usually secreted, a marker proteincomprising a heavy chain fragment of IgA, devoid oflight chains. Analysis of the Fca fragment found inserum, urine and intestinal fluid revealed that itbound secretory component (SC), providing the firstevidence for the binding site of SC to the IgA dimer.Left untreated, axCD progresses to a frankly invasiveintestinal lymphoma. The discovery of ctCD, hasprovided a unique and important opportunity tostudy the natural history of a B cell lymphoma as wellas offering the prospect of early diagnosis by detectionof the marker protein and of cure.4"The molecular analysis of immune and inflam-

matory events at the mucosal surface in humanbeings has been greatly enhanced by the improvedaccess to tissue provided by developments in fibre-optics and in ultrasound, microassay systemsincluding HPLC, fluorimetry, monoclonal antibodiesand DNA probes, and the use of recombinant DNA

technology to engineer micro-organisms not only toproduce large quantities of previously unavailablenatural peptides in pure form but also to createrecombinant vaccines for oral immunisation againstintestinal infection and infestation. Besredka's visionof oral vaccines is now being realised.

Division of Clinical Sciences,John Curtin School of Clinical Research,Australian National University,Canberra.

References

I Besredka A. De la vaccination contra les ctats typhoidpar la voic buccale. Ann Inst Pasteur 1919; 33: 882.

2 Crabbc PA, Heremans JF. Etude immunohistochem-ique des plasmocytes de la muqucse intestinale humainenormale. Rev Fran( Etudes Clin biol 1966; 11: 484-92.

3 Tomasi TB. Structure and function of mucosal anti-bodies. Ann Rev, Med 1970; 21: 281-98.

4 Doc WF, Henry K, Hobbs JR. Dowling RH. Alphachain disease. QJ Med 1970; 39: 619.

5 Brandtzaeg P. Role of J chain and secretory componentin receptor-mediated glandular and hepatic transport ofimmunoglobulins in man. Scand J Immnunol 1985; 22:111-46.

6 Crabb6 PA, Nash DR, Bazin H, Eyssen H, HeremansJF. Immunohistochemical observations on lymphoidtissuc from conventional and germ-free mice. Lab Invest1970; 22: 448-57.

7 Bland PW, Warren LG. Antigen presentation byepithelial cells of the rat small intestine. I Kinetics,antigen specificity and blocking by anti la antisera.Immunology 1986; 58: 1-7.

8 Mayer L, Schlien R. Evidence for function of la mole-cules on gut epithelial cells in man. J Exp Med 1987: 166:147 1-83.

9 Sollid LM, Brandtzaeg P, Kvale D, etal. Teell epitheliuminteractions in relation to gut immunity. Monogr Allergy,1988; 24: 60-65.

10 Owen RL, Joncs AL. Epithelial cell specializationwithin human Peyer's patches: An ultrastructural studyof intestinal lymphoid follicies. Gastroenterology 1974;66: 189-203.

11 Owcn RL. Sequential uptake of horseradish peroxidaseby lymphoid follicle epithelium of Peyer's patches in thenormal unobstructed mousc intestinc: Anultrastructural study. Gastroenterology 1977; 72: 440-5.

12 Spalding DM, Koopman WJ, Eldridge JH, McGhee JR,Steinman RM. Accessory cells in murine Peycr's patch.I Identification and Enrichment of a functional dendriticcell. J Exp Med 1983; 157: 1646-59.

13 Pavli P, Hume DA, Doe WF. Isolation and enrichmentof human colonic antigen-presenting dendritic cells(DCs). Gastroenterology 1989: 5: A385.

14 Stevens SK, Weissman IL, Butcher EC. Differences inthe migration of B and T lymphocytes: organ-selectivelocalization in vivo and the rolc oflymphocyte-cndothelial cell recognition. J Immunol1982; 128: 844-5 1.

on February 9, 2020 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.30.12.1679 on 1 D

ecember 1989. D

ownloaded from

The intestinal immune system 1685

15 Mestecky J, McGhee JR. Immunoglobulin A (IgA):Molecular and cellular interactions involved in IgABiosynthesis and Immune response. Adv Immunol1987; 40: 153-245.

16 Husband A, Gowans JL. The origin and antigen-dependent distribution of IgA-containing cells in theintestine. J Exp Med 1978; 148: 1146-60.

17 Pierce NF, Cray WC Jr. Determinants of the local-ization, magnitude and duration of a specific mucosalIgA plasma cell response in enterically immunised rats. JImmunol 1982; 128: 1311-5.

18 Ferguson A, Parrott DMV. The effect of antigendeprivation on thymus dependent and thymus inde-pendent lymphocytes in the small intestine of the mouse.Clin Exp Immunol 1972; 17: 477-88.

19 Halpern MS, Koshland ME. Novel subunit in secretoryIgA. Nature (Lond) 1970; 228: 1276-8.

20 Zikan J, Novotny J, Trapane TL, et al. Secondarystructure of the immunoglobulin J chain. Proc Natl AcadSci USA 1985; 8: 5905-9.

21 Fujiyama Y, Kobayashi K, Senda S, et al. A novel IgAprotease from Clostridium sp. capable of cleaving IgAjand IgA2 A2m(1) allotype but not IgA2 A2m(2)allotype paraproteins. J Immunol 1985; 134: 573-6.

22 Heremans JF, Bazin H. Antibodies induced by localantigenic stimulation of mucosal surfaces. Ann NYAcadSci 1972; 190: 268-75.

23 Ogra PL, Karzon DT. Poliovirus antibody response inserum and nasal secretions following intranasal inocu-lation with inactivated poliovaccine. J Immunol 1969;102: 15-23.

24 Colten H, Bienenstock J. Lack of C3 activation throughclassical or alternate pathways by human secretory IgAanti-blood-group-A antibody. In: Mestecky J, LawtonAR, eds. The Immunoglobulin A system New York:Plenum, 1974; 45: 305.

25 Fanger MW, Goldstine SN, Shen L. Cytofluorographicanalysis of receptors for IgA on human polymorpho-nuclear cells and monocytes and the correlation ofreceptor expression with phagocytosis. Mol Immunol1983; 20: 1019-27.

26 Tagliabue A, Boraschi D, Villa L, et al. IgA-dependentcell-mediated activity against enteropathogenic bacteria:distribution, specificity and characterisation of theeffector cells. J Immunol 1984; 133: 988-92.

27 Nagura H, Nakane PK, Brown WR. Translocation ofdimeric IgA through neoplastic colon cells in vitro. JImmunol 1979; 123: 2359-68.

28 Tomana M, Phillips JO, Kulhavy R, Mestecy J. Carbo-hydrate-mediated clearance of secretory IgA from thecirculation. Mol Immunol 1985; 22: 887-92.

29 Cebra J, Komisar JL, Schweitzer PA. CH isotype"switching" during normal B-lymphocyte development.Ann Rev Immunol 1984; 2: 493- 548.

30 Elson CO. Heck JA, Strober W. T cell regulation ofmurine IgA synthesis. J Exp Med 1979; 149: 632-43.

31 Kawanishi H, Saltzman L, Strober W. Mechanisms

regulating IgA class-specific immunoglobulin produc-tion in murine gut-associated lymphoid tissues 11.Terminal differentiation of postswitch sIgA-bearingPeyer's patch B cells. J Fxp Med 1983; 158: 649-69.

32 Kiyono H, Cooper MD, Kearney JF, et al. Isotypespecificity of helper T cell clones. Peyer's patch Th cellsprefcrentially collaborate with maturc IgA B cells forIgA responses. J Exp Med 1984; 159: 798-811.

33 Kiyono H, Mostellar-Barnum LM, Pitts AM,Williamson SI, Michalek SM, McGhee JR. Isotype-specific immunoregulation: IgA binding factors pro-duced by Fcu receptor positive T cell hybridomasregulate IgA responses. J Exp Med 1985; 161: 731-47.

34 Ishizaka K. Regulation of IgE synthesis. Ann RevImmunol 1984; 2: 159-82.

35 Suzuki 1, Kitamura J, Kiyono H, et al. Isotype-specificimmuno-regulation. Evidence for a distinct subset of Tcontrasuppressor cells for IgA responses in murinePeyer's patches. J Exp Med 1986; 164: 501-16.

36 Andre C, Heremans JF, Vaerman JP, Cambiasco CL. Amechanism for induction of immunological tolerance byantigen feeding: antigen-antibody complexes. J ExpMed 1975; 142: 1509-19.

37 Kagnoff MF. Effects of antigen-feeding on intestinaland systemic immune responses. III. Antigen-specificserum-mediated suppression of humoral antibodyresponses after antigen feeding. Cell Immunol 1978; 40:186-203.

38 Zeitz M, Greene WC, Peffer NJ, James SP. Lympho-cytes isolated from the intestinal lamina propria ofnormal nonhuman primates have increased expressionof genes associated with T cell activation. Gastroenter-ology 1988; 94: 647-55.

39 Shanahan F, Brogan M, Targan S. Human mucosalcytotoxic effector cells. Gastroenterology 1987; 92:1951-7.

40 Hogan PG, Hapel AJ, Doe WF. Lymphokine-activatedand natural killer cell activity in human intestinalmucosa. J Immunol 1985; 135: 1731-8.

41 Fiocchi C, Tubbs RR, Youngman KR. Human intestinalmucosa mononuclear cells exhibit lymphokine-activatedkiller cell activity. Gastroenterology 1985; 88: 625-37.

42 Phillips JH, Lanier LL. Lectin-dependent and anti CD-3induced cytotoxicity are preferentially mediated byperipheral blood cytotoxic T lymphocytes expressingLcu-7 antigen. J Immunol 1986; 136: 1579-85.

43 Shanahan F, Deem R, Nayersina R, Leman B, TarganS. Human mucosal T-cell cytotoxicity. Gastroenterology1988; 94: 960-7.

44 Seligman A, Hurez D, Mihaesco D, et al. Alpha chaindiscase; a new immunoglobulin abnormality. Science1968; 162: 1396-7.

45 Doe WF, Hobbs J, Henry K, et al. Five cases of alphachain disease. Gut 1972; 13: 947-57.

46 WHO Alpha chain disease and related small intestinallymphome. Revue Generale, Arch Franc Mal ApparDig 1976; 65: 591-607.

on February 9, 2020 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.30.12.1679 on 1 D

ecember 1989. D

ownloaded from