966

satisfactory serological response in young infants; a morecareful assessment of short and long term side-effects is

necessary and the duration of antibody responses and ofclinical protection will also have to be determined before thisvaccine can be advocated for immunisation of young infants.Another issue that needs to be addressed is whether the

vaccine, when given to young infants, alters the immuneresponse to subsequent measles vaccination at a later age.Several researchers have reported that children in theUnited States when first vaccinated at less than 10 months of

age with the Moraten strain of virus have had a depressed HIantibody response when revaccinated later.19,20 Our limiteddata pertaining to Gambian children given either E-Zvaccine or Schwarz vaccine at 18-20 wk of age do not

support this concept, for all these children respondedsuccessfully to revaccination.The E-Z vaccine holds promise that, when given in the

correct dose, it will successfully immunise young infantsagainst measles. If this proves true the way is open for moreefficient and effective immunisation campaigns in the

developing world. Will the goal of global measles

eradication2l then be translated from fantasy to fact?

We thank Mr S. K. Rahman for his skilled technical help and Mr B. Sarr,M. Daffeh, P. Cham, and Ms M. Jooffor help in the field. We also thank DrB. M. Greenwood and Dr R. Whitehead for their advice and encouragement.

Correspondence should be addressed to H. C. W., MRC Laboratories,Fajara, The Gambia, West Africa.

REFERENCES

1. Expanded Programme on Immunization Measles-spots that kill. EPI Update,Geneva: WHO, 1986.

2. Shahid NS, Clauquin P, Shaik K, Zimicki S. Long-term complications of measles inrural Bangladesh. J Trop Med Hyg 1983; 86: 77-80.

3. Morley D. Paediatric priorities in the developing world London Butterworths, 1973:220-22.

4. Foster A, Sommer A. Childhood blindness from corneal ulceration in Africa: causes,prevention, and treatment. Bull WHO 1986, 64: 619-23.

5. Aaby P, Bukh J, Lisse IM, Smits AJ. Measles mortality, state of nutrition, and familystructure. A community study from Guinea-Bissau. J Infect Dis 1983; 147:

693-701.

6. Aaby P, Bukh J, Lisse IM, da Silva MC. Measles mortality: Further communitystudies on the role of overcrowding and intensive exposure. Rev Infect Dis (inpress)

7. World Health Organisation. Weekly Epidemiol Rec 1979; 44: 337-39.8. Ministry of Health of Kenya and World Health Organization. Measles immunity in

the first year after birth and the optimum age for vaccination in Kenyan children.Bull WHO 1977; 55: 21-30.

9. Whittle HC, Rowland MGM, Mann GF, Lamb WH, Lewis RA. Immunisation of4-6 month old Gambian infants with Edmonston-Zagreb measles vaccine. Lancet1984; ii: 834-37.

10. Walsh JA. Selective primary health care: Strategies for control of disease in thedeveloping world. (IV) Measles Rev Infect Dis 1983; 5: 330-40.

11. Klein-Zabban ML, Foulon G, Gaudebout C, Badoual J, Adou A. Fréquence desrougeoles nosocomiales dans un centre de protection maternelle et infantile d’Abidjan. Bull WHO 1987; 65: 197-201.

12. Sinha WP. Measles in children under 6 months of age: an epidemiological study JTrop Paediatr 1981; 27: 120-22.

13. Loening WEK, Coovadia HM. Age in urban, peri-urban, and rural environmentsimplications for time of vaccination. Lancet 1983, ii: 324-26

14. Ikic D, Juzbasic M, Beck M, Hraber A, Cimbur-Scheriber T. Attenuation andcharacterization of Edmonston-Zagreb measles virus. Ann Immunol Hung 1972,16;175-81

15. Ikic DM. Edmonston-Zagreb strain of measles vaccine: epidemiologic evaluation inYugoslavia. Rev Infect Dis 1983, 5: 558-63.

16. Sabin AB, Arechiga AF, de Castro JF, et al. Successful immunization of children withand without maternal antibody by aerosolized measles vaccine. JAMA 1983; 249:2651-62.

17. Khanum S, Uddin N, Garelick H, Mann G, Tomkins A. Comparison ofEdmonston-Zagreb and Schwarz strains of measles vaccine given by aerosol orsubcutaneous injection. Lancet 1987, i: 150-53.

18. Mann GF, Allison LMC, Compeland JA, Agostini CFM, Zuckerman AJ. Asimplified plaque assay system for measles virus. J Biol Standard 1980; 8: 219-25.

19. Wilkins J, Wehrle PF. Additional evidence against measles vaccine administration toinfants less than 12 months of age: altered immune response following active/passive immunization. J Pediatr 1979; 94: 865-69.

20. Linnemann CC, Dine MS, Roselle GA, Asky PA. Measles immunity afterre-vaccination: results m children vaccinated before 10 months of age. Pediatrics1983; 67: 332-35

21 Hopkins DR, Hinman AR, Koplan JP, Lane JM. The case for global measleseradication. Lancet 1982; i: 1396-98.

GALACTOSYLATION OF IgG ASSOCIATEDOLIGOSACCHARIDES: REDUCTION INPATIENTS WITH ADULT AND JUVENILEONSET RHEUMATOID ARTHRITIS ANDRELATION TO DISEASE ACTIVITY

RAJ B. PAREKH1DAVID A. ISENBERG2BARBARA M. ANSELL3

IVAN M. ROITT2RAYMOND A. DWEK1

THOMAS W. RADEMACHER1

Glycobiology Unit, Department of Biochemistry, University ofOxford;1 Departments of Immunology and Rheumatology Research,University College and Middlesex School of Medicine, London;2

Department of Rheumatology, Northwick Park Hospital, Harrow3

Summary The prevalence of agalactosyl N-linkedoligosaccharides on serum IgG was

determined for patients with juvenile onset and with adultrheumatoid arthritis. A. significant difference in the

prevalence of these structures from age matched controlswas found in both types of arthritis. In patients with adultonset rheumatoid arthritis, the results showed a strongcorrelation between the prevalence of IgG-associatedagalactosyl oligosaccharides and disease activity. Acorrelation between disease activity and agalactosylstructures was also seen in a retrospective analysis of serialIgG samples from patients with juvenile onset disease. Thefinding that childhood onset arthritis and adult rheumatoidarthritis share a defect of glycosylation of serum IgGsuggests that there may be a greater similarity between thesetwo varieties of rheumatoid arthritis than has been hithertoconsidered. The observation that the incidence of

agalactosyl oligosaccharides on IgG fluctuates with diseaseactivity provides indirect evidence for a seminal role for thischange of glycosylation in the inflammatory process which,in rheumatoid arthritis, is focused on the synovial tissuesand results in bone erosions and joint destruction.

Introduction

VARIOUS immunological indices have been found to beabnormal in rheumatoid arthritis, but the aetiology of thisdisease is not established:1 both infection and heredity arebelieved to play a part. Juvenile arthritis is characterised byan onset of arthritis under the age of 16 yr, and its differential

diagnosis requires the active exclusion of separate diseasessuch as infection, congenital anomalies of the musculo-skeletal system, and haematological disorders. Juvenilearthritis is currently subclassified into three types on thebasis of the pattern of illness in the first three months.2

"Systemic onset" involves a high swinging fever associatedwith an atypical rash, and in many cases with

lymphadenopathy, hepatosplenomegaly, and occasionallypericarditis: at this stage there may be no joint symptoms (ormerely arthralgia), but polyarthritis subsequently develops.A "polyarthritic onset" is recognised when five or morejoints are involved from the outset. A "pauciarticular onset"is described when the arthritis is confined to four or fewer

joints: this third subgroup, the largest, is further subdividedinto those with a young age of onset (under 9 yr, usually girls,and often associated with chronic iridocyclitis), and thoseolder (usually boys, who carry HLA B27, and whosesymptoms relate more to spondyloarthritis).A previous study of patients with adult onset rheumatoid

arthritis demonstrated characteristic changes in the N-

glycosylation of serum IgG,3.4 with an elevated relativeincidence of agalactosyl N-linked oligosaccharides, bothouter arms terminating in N-acetylglucosamine. We now

967

Fig I-Percentage prevalence of agalactosyl monosaccharidesequences, G(0), from the IgG of patients with juvenile onset andadult onset rheumatoid arthritis related to age.

All juvenile onset patients had active disease and were separated as to themode of onset: systemic x ; polyarticular 0; pauciarticular .

Data from individual adult onset patients is separated into those with activedisease 0; and those without W.The solid line depicts the regression function for normal subjects (n = 111).

The hatched lines depict the regression functions for 2 SD from the mean (forages with n > 3).

Values of G(O) for patients with either juvenile onset or adult onsetrheumatoid and active disease were significantly different from age-matchedcontrols (p < 0-01) by analysis of covariance. G(O) values for patients withinactive disease were not significantly different from the G(O) values obtainedfor normal subjects.

report the prevalence of such structures on serum IgG inpatients with juvenile arthritis, and the association betweenthe prevalence of such structures and disease activity in bothjuvenile and adult onset cases.

Patients and Methods

All of the adult onset cases (n = 53; 19 male, 34 female) met theAmerican Rheumatism Association’s criteria for classic or definitedisease.5 The clinical activity in these adult onset cases was

independently assessed according to a previously published index.6

Fig 2-Percentage prevalence of agalactosyl monosaccharide

sequences from IgG of patients with adult onset rheumatoidarthritis related to clinical score.

Clinical score: 1 = inactive/mild, n = 3; 2 = moderately active, n = 7;3 = severely active, n = 4.

Fig 3-A) Retrospective analysis of G(0) in patients with juvenileonset arthritis.

7 were in remission at the time of the second serum sample. Mode of onsetwas systemic in 6, x ; polyarticular in 3, 0; and pauciarticular in 1, <S). Most

patients progressed to symmetrical polyarthntis (n = 8); progression in 2patients was limited to the pauciarticular type of arthritis. The solid curve linerepresents the age-dependent function for the normal subjects between ages1-50 yr." (Separate set of patients to those reported in fig 1.)

B) Retrospective analysis of G(0) in a patient presenting withsystemic onset juvenile arthritis.

Progression to symmetrical polyarthritis, before inactive disease ( . andeventual remission (A ).

Active disease was associated with active synovitis, and was usuallyaccompanied by a raised erythrocyte sedimentation rate (ESR). Forthis study, we defined remission as a lack of clinical activityassociated with a normal ESR and requiring no drug therapy for 2yr.Serum IgG from 27 patients with juvenile onset arthritis was

studied (9 male, 18 female). 14 patients had presented with systemiconset; 9 patients with pauciarticular arthritis, which in 5 cases wasassociated with chronic iridocyclitis; and 4 patients with

polyarticular onset. 22 patients progressed to symmetricalpolyarthritis; 5 remained with pauciarticular disease, 3 of whom hadan asymmetric presentation. The age of onset was between 6 mo and11 yr. Synovitis, with or without a raised ESR, was required for

968

definition of active disease. Inactivity was defined as no evidence ofactive synovitis and a normal ESR, although joint limitation maystill have been present. Remission was defined as for subjects withadult onset arthritis. Medications included penicillamine, gold,steroids, and chlorambucil.IgG was purified from fresh or stored (&mdash; 20&deg;C, 1-15 yr) serum

samples as reported previously.’ The N-linked oligosaccharideswere released from the IgG samples using anhydrous hydrazine,purified and 3H labelled by reduction as reported previously.’ Thepercentage incidence of N-linked monosaccharide sequences whichcontained no terminal galactose residues&mdash;G(0)&mdash;was determinedwith an exoglycosidase mixture.8 Linear regression analysis wasperformed on the G(0) vs age data for normal subjects and for eachof the disease groups. In each case, juveniles (age 1 to 19 yr) andadults (age 20 yr and over) were analysed separately. Differencesbetween the regression lines for the diseased groups and the

corresponding normal patients were assessed by analysis ofcovariance. Probabilities are reported for the null hypothesis thatthe normal and disease data lie on a coincidental line. All statisticswere calculated with the Minitab programme running on an IBMPC/AT.9

Results

G(0) is age-related, so the values of G(0) for patients witharthritis are presented as a function of age. The resultsshown in fig 1 confirm the original observation that patientswith adult onset rheumatoid arthritis with active disease arecharacterised by an increase in IgG-associated agalactosylmonosaccharide sequences (ie, terminating with residuesother than galactose, predominantly with N-

acetylglucosamine). No differences between the G(0) valuesof normal subjects and adult rheumatoid patients withinactive disease were found. Fig 1 also shows the G(0) valuesfor patients with juvenile onset arthritis: clearly, the G(0)values in patients with active juvenile onset and adultrheumatoid arthritis are similar in magnitude. Fig 2 showsthe relation between G(0) and clinical score for patients withadult onset rheumatoid arthritis: some of the patients withactive disease (score 2 and 3) had low ESR values, despitethe presence of active synovitis. The G(0) values seem to beindependent of the mode of onset in the juvenile patients.Fig 3A shows the G(0) values found at the time of diseaseonset and at a subsequent time of inactivity (or remission)for juvenile onset cases: the G(0) values obtained at the timeof disease inactivity are within the normal range for this agegroup. To further evaluate the relation of G(0) to diseaseprogression, serial IgG samples were analysed from a singleindividual (fig 3B). The G(0) values vary progressively, andagain there appears to be a correlation between this variableand disease activity, with a dramatic range of variation inG(0) during the course of disease progression and remissionin this patient.

Discussion

The N-glycosylation defect of serum IgG (namelydecreased outer-arm galactosylation), first described

amongst adult onset patients with rheumatoid arthritis, isshown by this study to be present almost uniformly injuvenile onset cases, and is detectable irrespective of themode of onset in this younger age group. The defect wasrelated to disease activity both in serial studies amongst thejuvenile onset patients and amongst adult onset patientswhose clinical activity was independently assessed. Theseobservations raise questions both about the importance ofthe N-glycosylation defect in patients with rheumatoidarthritis, and about the relation between the adult andjuvenile modes of onset.

Studies with a large number of healthy subjects haveshown that the number of oligosaccharide chains on serumIgG whose outer arms lack galactose and terminate

predominantly in N-acetylglucosamine is age-related.8 Inparticular, there is significantly more agalactosyl IgGbetween the ages 1-15 and 40-70 yr than in adolescence and

young adult life. This observation makes it essential to relate

every result that is obtained to age. When this is done, noabnormality is discernible amongst individuals withosteoarthritis (as was originally reported)4, but both theadult and juvenile onset patients with active rheumatoidarthritis remain well outside the normal limits. An analysisof over 300 sera from patients with approximately 25different autoimmune rheumatic diseases, seronegativearthropathies, and infectious diseases showed that the defectin galactosylation is virtually confined to rheumatoidarthritis, tuberculosis, and Crohn’s disease (RademacherTW et al, unpublished). The low galactose values are not aconsequence of chronic excessive IgG synthesis, sincenormal glycosylation was seen in primary systemic lupuserythematosus (SLE) and leprosy--diseases with

persistently raised IgG levels far in excess of those seen inrheumatoid arthritis. Furthermore, the defect does not ariseas a simple acute phase reaction, since many of the SLEpatients with a normal oligosaccharide arrangement of theirIgG had an ESR greater than 70 mm/hr. The defect is notsimply linked to HLA DR status, since the high prevalenceof HLA DR4 in adult onset patients with rheumatoidarthritis is not found in most juvenile onset cases,10 or intuberculosis (de Vries RBP, personal communication).The lack of terminal galactose residues in the Fc portion

of IgG molecules could contribute to the autoantigenicity ofthis region.4 Possible mechanisms for this include exposureof particular aminoacids in the Fc region which are usually"concealed" by the overlying oligosaccharide structure;and shortened oligosaccharides terminating in N-

acetylglucosamine acting as an epitope. Autosensitisation tothe Fc portion of IgG remains an important concept in adultonset rheumatoid disease, and approximately 70% of thesepatients are IgM rheumatoid factor positive; most of theothers have IgG and, to a lesser extent, IgA rheumatoidfactors. Although conventional agglutination tests forrheumatoid factors are usually negative in juvenile onsetcases, IgG rheumatoid factors are commonly detected bysolid phase assays.11,12 However, the juvenile onset caseshave several clinical features which distinguish them frompatients with adult onset disease. For example, majorsystemic disease in the absence of overt synovitis is very rarein adult onset rheumatoid arthritis, unlike the juvenile onsetcases. The pattern of joint involvement also differs betweenthe childhood and adult onset cases. In adults, rheumatoidarthritis is principally a symmetrical disease of the smalljoints of the feet and hands, especially the proximalinterphalangeal and metacarpophalangeal joints; childrenmore often have knee and distal or proximal interphalangealarthropathy, but rarely metacarpophalangeal involvement.

Despite the clinical differences between juvenile andadult onset rheumatoid arthritis, we have now shown thatthey share the same defect in respect of the glycosylation oftheir IgG and that there may be a greater similarity betweenthese two types of arthritis than was previously thought.The observation that the agalactosyl IgG levels fluctuatewith disease activity is also new: we are not yet certainwhether this fluctuation is relevant to the causation of the

inflammatory process in rheumatoid arthritis. The geneticand environmental factors leading to altered galactosylation

969

of the N-linked oligosaccharides of serum IgG in thesediseases are not known, but recent results indicate a

modulation of p-galactosyltransferase activity in B

lymphocytes: a specific 0-galactosyltransferase whichtransfers UDP-Gal to an asialo-agalacto IgG has beenreported to be present in B lymphocytes.13,14 The affinity ofthis enzyme for UDP-Gal in the B lymphocytes of patientswith rheumatoid arthritis is lower than in B lymphocytesfrom a control group, and the specific activity of thegalactosyltransferase towards asialo-agalacto IgG was foundto be reduced to 50-60% of control levels in adultrheumatoid patients. Further analysis of the mechanismunderlying the galactosylation defect may yield importantinsights into the pathogenesis of both disorders.

We thank Daryl Fernandes and David Ashford for help with the statisticalanalysis and Mrs P. M. Rudd for technical assistance. R. B. P., R. A. D., andT. W. R are members of the Oxford Glycobiology Unit, which is supportedby the Monsanto Company.

Correspondence should be addressed to T. W. R., Department ofBiochemistry, University of Oxford, South Parks Road, Oxford OXl 3QU.

REFERENCES

1. Morrow WJW, Isenberg DA. Autoimmune rheumatic disease. Oxford: BlackwellScientific, 1987: 161-67.

TUMOUR NECROSIS FACTOR AS ANAUTOCRINE TUMOUR GROWTH FACTORFOR CHRONIC B-CELL MALIGNANCIES

F. T. CORDINGLEYA. V. HOFFBRANDH. E. HESLOPM. TURNER

A. BIANCHI

J. E. REITTIEA. VYAKARNAMA. MEAGER

M. K. BRENNER

Department of Haematology, Royal Free Hospital, London;Department of Immunology, Sunley Research Institute, CharingCross Hospital, London, and National Institute of Biological

Standards and Control, South Mimms, Herts

Summary Recombinant tumour necrosis factor

(TNF) promotes survival and induces

proliferation in the tumour cells from two malignancies of Blymphocytes&mdash;hairy-cell leukaemia and B-chronic

lymphocytic leukaemia. Culture with TNF also inducesTNF mRNA and protein, so the cytokine may act as anautocrine tumour growth factor. These growth promotingeffects are antagonised by alpha but not by gammainterferon.

Introduction

TUMOUR necrosis factor (TNF) may induce lysis ofmalignant cells and cell-lines and regression of some animaltumours;l.2 this effect has led to investigation of thetherapeutic value of TNF in clinical oncology. Morerecently, however, TNF has been shown to act as a growthfactor for normal fibroblasts,3 T cells,4 and B cells. We nowshow that TNF can also act as a tumour growth factor andpromote the survival and proliferation of tumour cells in twotypes of B-cell malignancy-hairy-cell leukaemia (HCL)and B-chronic lymphocytic leukaemia (B-CLL).

Patients and Methods

- PanMH.&mdash;Ten patients, three with HCL and seven withB-CLL, were studied on one to four occasions. Diagnosis of HCLwas made on the basis of circulating cell morphology, phenotype

2. Ansell BM. Juvenile chronic arthritis. Curr Orthop 1986; 1: 81-89.3. Parekh RB, Dwek RA, Rademacher TW. Rheumatoid arthritis as a glycosylation

disorder. Br J Rheumatol (in press).4. Parekh RB, Dwek RA, Sutton BJ, et al Association of rheumatoid arthritis and

primary osteoarthritis with changes m the glycosylation pattern of total serum IgG.Nature 1985; 316: 452-57.

5. Ropes MW, Bennett GA, Cobb S, Jacobs R, Jessar RA. 1958 revision of diagnosticcriteria for rheumatoid arthritis. Bull Rheum Dis 1958; 9: 175-76.

6. Isenberg DA, Martin P, Hajirousou V, Todd-Pokropek A, Goldstone AH, SnaithML. Haematological re-assessment of rheumatoid arthritis. Br J Rheumatol 1986;25: 125-27.

7. Ashford D, Dwek RA, Welply JK, et al. The &bgr;1 &rarr;2-D-xylose and &agr;1 &rarr;3-L-fucosesubstituted N-linked oligosaccharides from Erythrina cristagalli lectin. Eur JBiochem 1987; 166: 311-20.

8. Parekh RB, Isenberg DA, Roitt IM, Dwek RA, Rademacher TW. Age-specificgalactosylation of N-linked oligosaccharides of human serum IgG. J Exp Med (inpress).

9. Godfrey K. Simple linear regression in medical research. N Engl J Med 1985; 313:1629-36.

10. Miller ML, Glass DN. The major histocompatibility complex antigens in rheumatoidarthritis and juvenile arthritis. Bull Rheum Dis 1981; 31: 21-24.

11. Torrigiani G, Ansell BM, Chown EEA, Roitt IM. Raised IgG antiglobulin factors inStill’s disease. Ann Rheum Dis 1969; 28: 424-27.

12. Hay FC, Nineham LJ, Fletcher MR, Roitt IM. An improved method for the detectionof IgG and IgM rheumatoid factors in rheumatoid diseases. In: Jayson MIV, ed.Still’s disease-juvenile chronic polyarthritis. London: Academic Press, 1976:135-44.

13. Furukawa K, Matsuta K, Takeuchi F, et al. Alteration of a galactosyltransferase in Bcells of rheumatoid arthritis patients. From the IXth international symposium onglycoconjugates. Tourcoing (France): A. Lerouge, 1987: E56.

14. Axford JS, Mackenzie L, Lydyard PML, Hay FC, Isenberg DA, Roitt IM. ReducedB-cell galactosyltransferase activity in rheumatoid arthritis Lancet 1987; ii:

1486-88.

(with B-cell markers and a monoclonal antibody, FMC7, thatdetects a hairy-cell-associated marker), cytochemistry (positivity fortartrate-resistant acid phosphatase), and trephine or splenichistology, B-CLL was diagnosed by means of cell morphology,number, and phenotype (CD5 and CD20 positive). All patientswere selected for high circulating HCL or B-CLL cell numbers.Two of the three HCL patients had undergone splenectomy 5 and 6years previously. No patient was receiving therapy at the time ofstudy.

Recombinant cytokines.-Recombinant (r) cytokines, expressed inEscherichia coli, were the gift of Knoll-AG (BASF) (rTNF), Biogen(interferon -IFN and Kirby Warwick (rIFN-ot). Specificactivity of rTNF was 6-3 x 106 u/mg protein, of rIFN-y 3-3 x 106u/mg protein, and of rIFN-a 2-5 x 106 u/mg protein.

Cell preparation and culture.-Peripheral blood mononuclearcells were prepared by ’Ficoll’ sedimentation and then depleted ofmonocytes by adherence and of T cells by double-E rosetting. Cellsused in this study contained < 0-5 % T cells or monocytes; > 95 %of the mononuclear cell population from HCL patients and > 98 %from CLL patients were neoplastic cells. Viability counts werecarried out daily in 96-well microculture plates (Nunc). 180 )1 of cellsuspension at 106/ml in RPMI 1640 medium and 10% fetal calfserum selected for low mitogenicity were added to each wellfollowed by 10 III of either cytokines or RPMI 1640. At intervalsduring incubation at 37&deg;C in 5 % CO2, viability counts were carriedout with 0 25% ’Nigrosine’. All estimations were carried out intriplicate. Proliferation was estimated by pulsing plates prepared asabove with 1 pCi 3I-I-thymidine for 6 h at 37&deg;C and then harvestingDNA and measuring thymidine incorporation in a beta-scintillationcounter.

TNF receptor assay.-r TNF was iodinated with 1251 (AmershamRadiochemicals) by the iodogen method,6 producing a specificactivity of > 1000 fCi/ug of protein. TNF receptor assays werecarried out by the competitive displacement method7 and cell-bound rTNF-125I was separated from unbound TNF bycentrifugation through phthalate oil.a All estimations were carriedout in triplicate and analysed with the LIGAND computerprogram. 9

Enzyme-linked immunosorbent assay (ELISA).-Microtitreplates (Micro Elisa, Dynatech, Billingshurst) coated with rabbitanti-human IgM (DAKO, Weybridge) or goat anti-human IgGchain-specific antiserum (Sigma) were incubated with supernatantsfrom HCL or B-CLL cultures for 2 h; alkaline-phosphatase-conjugated goat anti-human antibodies specific for IgM or IgG, orfor anti-human kappa or lambda light chains (Sigma), were then