lmmiinolo\iy and Cell Biology (1991) 69 261 -271

Low affinity binding of mouse immunoglobulin to human CD5+ B cells KATHRYN M. WESTON AND ROBERT L. RAISON Imnninohioh^y Unll. Univer.slly of Technology. Sydney. (Jorc Hill. A'5H', 2065 and Clinical Immunohfiy Research Centre. University of Sydney. NSW 2006 .Australia (Siibmiiied 23 May 1991. .deeepled 18 .July 1991.) Summary Polyreaetive immunoglobulin (lg) secreted by CD5-bearingB cells has the capacity to bind a broad range of self and foreign antigens. Flow cytometric analysis was used to detect low-altinity binding iif mouse lg molecules Io the surface of CD5-bearing B cells from patients with chronic lymphocytic leukaemia (CLL). Mouse lg isotypes G. A. and M. and IgG subclasses Ci|. Cna. and G3 bound to the B cell surface via a mechanism not involving the antigen-binding site of the mouse lg molecule. Fab and F(ab'): fragments of mouse lg associated with the CLL cells in a similar manner to intact lg. indicating that Fc receptor interactions were not involved. Dissociation of the mouse lg from the B cell surface by three washing steps distinguished this lowcr-aflinity binding from high-atfmity binding which occurs through the antigen-binding site of a mouse monoclonal antibody (MoAb) when it recognizes a specific ceil surface epitope. Blocking studies suggest that the low-atfinity binding occurred via surface IgM and surface igD (sIgM/sIgD) on the CD5-bearing B cells. The results are consistent with the expression of poiyreaclive ig on the surface of CD5-bearing B cclis in CLL.

INTRODUCI ION A subset of normal hutnan B celts has recently beendescribed which bears the T cell-associated antigen CD5 (1.2). CD5-positivc B cells constituteapproximately 20%ofthecirculatingBce!ls in normal peripheral blood (1,2). and 40-60%of B cells in human fetal spleen (3). They have also been reported to be present in normal spleen, lymph nodes and inflamed tonsils (4). Furthermore, the ncoplastic small lymphocyte population found in chronic lymphocytic leukaemia (CLL) isortheCD5 phenotype (5-7). Numbers of CD5-positi\ e B cells are also elevated in the peripheral blood of rheumatoid arthritis (RA) patients (8). CD5-positive B ceils from normal subjects and from RA patients were infected with Epstein-Barr virus (EBV) and transformed into antibody-seeretingeells(9). The monoclonal antibody (MoAb) seereted was found to be poly-

reactive, binding a number of antigens of different size and structure with varying attinitJes. The range of antigens bound included the self antigens single-stranded DNA (ssDNA), thyroglobulin and insulin, and foreign antigens, such as tetanus toxoid and bacterial polysaccharidc. In addition, polyreaetive immunoglobulin (lg) from both normal and RA patients has been shown to bind the Fc region of hutnan IgG. aithough with a lower affinity than conventional rheumatoid factor binding (9.10). Polyreaetive antibodies are primarily IgM antibodies although IgG and lg.\ polyreaetive antibodies of this type have been described. They are thought to represent the so-called 'natural antibodies" of serum which may function as a first line of defence against invading organisms, or to elear auto-antigens (11). While the structural basis for the observed polyreactivity is unclear, it is apparent that restricted variable domain of heavy chain

Correspondence: K. M. Weston.lmmunobiology Unit. University ofTcchnology. Sydney. WestbourneSt, Gore Hill. NSW 2065. Australia. .{hhrcviations used in iliis paper BSA, bovine serum albumin: CIRC. Clinical Immunology Research Centre; CLL. chronic lymphocytic leukaemia: EBV, Epstcm-Barr \irus: FiTC. tluoresccin isothiocyanate: Ig. immunoglobulin: K M A . kappa myeloma antigen: MoAb. monoclonal antibody. PBL. peripheral blood lymphocytes: PBS, phosphate buffered saline: PE. phycoerythrin: R.^. rheumatoid arthritis: sig. surface Ig: sIgD. surface IgD: sIGM. surface IgM: ssDN.A, singie-stranded DN.A: streptavidin-PE. strcptavidin-phycocrythrin: VH, variable domain of heavy chain: Vi_. variable domain of light chain.

262

K, M, WESTON . \ N D R. L. RAISON

family usage and a lack of somatic mutation may coincide with this phenomenon. Two VH families. VijIII and V||IV. were found to predominate when eight human natural auto-antibody cDNA clones were sequenced (12). Two of the VnllI genes sequenced were essentially identical to previously published germline genes. CD5positiveB cells from CLL patients are characterized by the co-expression of surface IgM (sIgM) and surface IgD (sIgD) and by the use of a restricted repertoire ofgermiine genes for both the H (13) and L( 14) chains. Thus, the polyspecificity of the Ig expressed by these cells may reflect the use ofgermiine V genes that encode large, multi-reactive binding sites (II). So far. the demonstration of polyreactivity has been confined to Ig secreted by CD5-bearing B cells following transformation with EBV. In this paper, we demonstrate an apparent low-affinity interaction of mouse Ig with human IgM and IgD on the surface of CD5-positivc B cells from patients with CLL. The binding was not through the Fc region of the mouse Ig molecule, nor was it a direct result of speeific binding via the antigen-binding site of the mouse Ig. Further, the binding was not restricted to a particular mouse Ig isotype. Our findings extend the range of antigens interacting with polyreactive IgM and raise the possibility that this phenomenon may cause false positive results when immunofluorescence and flow cytometry are used for the analysis of the surface antigens of CLL B cells.

MATERIALS AND METHODS Peripheral blood lymphocytes and human cell lines Peripheral blood lymphocytes (PBL) from a CLL patient were obtained after leucapheresis at the Royal Prince Alfred Hospital. Camperdown. NSW. Red blood cells were lysed by incubation of 10 mL packed cells with 10 mL'Tris/NH4Cl (17 mmol/L Tris. 0.14 mol/LNHaCl.pH 7.2) at 37T for 5 min. followed by washing in RPMI 1640 and centrifugation and then a further Tris/NH4CI treatment for 1 min at 37T. The lymphocyteswerethenwashed with RPMI 1640 and either stored in liquid nitrogen or used immediately for surface immunofluorescence studies. Normal human lymphocytes were isolated from normal peripheral blood by density gradient centrifugation (800,i,', 15 min) over 'Lymphoprep" (Nycomed. Norway). The cells at the interface were collected, washed three times with RPMI and used immediately for immuno-

fluorescence staining. Human cell lines were maintained in in vino culture in RPMI 1640 supplemented with fetal bovine serum (10%). HEPE.S (4.76 g/L). sodium bicarbonate (8.5 mg/L). penicillin (100 iu/niL) and streptomycin (0.1 mg/mL). at 37T, In an atmosphere of 5% CO:. Antibodies The mouse MoAb used in this study were either commercially obtained or prepared by investigators at the Clinical Immunology Research Centre(CIRC). University of Sydney. K-1-21 is an IgGl MoAb with specificity for the kappa myeloma antigen (KMA) on the surface of human myeloma and lymphoma cells. The human ceil line. HMy2. bears this antigen. K-l-21 also binds to free but not heavy chain associated kappa light chains (15). K-l-21 was purified from ascites by affinity chromatography on a column of human-free kappa light chains bound toSepharose CL4B. Biotin-labelled K-l-21 was prepared according to establish protocols (16), L7 is an lgG| mouse MoAb with specificity for an antigen from Mycnhaclerittm bovi\ (17). L7 and the other MoAb used to determine binding of dirterent Ig isotypes were kindly supplied by Dr K. Z. Walker. CIRC. University of Sydney'. The culture supernatant from hybridomas producing these MoAb was used for immunolluorescence studies. Commercially available MoAb specific for human CD5 (igG2a). CD3(lgGi). and CDl')(IgG|) were obtained from Becton Dickinson. Mountain View. CA. USA. as was the streptavidin-phycoerythrin complex (streptavidin-PE). Polyclonal sheep and donkey antisera, unlabelled or labelled with fluorescein isothiocyanate (FITC) were obtained from .Silenus Laboratories. Hawthorn. Vic., Australia. Unless otherwise stated, all antisera were titrated to determine the optimum concentrations for use In tlow cytometry. Surface immunofluorescence \0^ PBL or 10-^ HMy2 eells were suspended in 0.5 mL cold phosphate buffered saline/bovine serum albumin (HBS/BSA; 0.1% BSA. 0.1% NaNi in 0.14 mol/L PSA eontaining 0.85% NaCl, pH 7.2). The cells were centrifuged in 1.5 mL plastic centrifuge tubes for 4-6 s at room temperature in an Eppendori 5415C centrifuge at 14 000 r/niin. After the supernatant was aspirated and the first antibody (25 |iL) was added, the pellet was mixed gently to resuspendthe cells and incubated at 4^^ for 30 min. with gentle

BINDING OF MOUSE IG TO CD5 + B CELLS

263

agitation after 15 min. Following incubation with the lirst antibody, the cells were washed once or three times, depending on the experiment, by addition ofO.5 mLcold PBS/BSAand subsequent mi.xing and ccntrit'ugation (4 s at room temperature). After aspirating the supernatant from the final wash, the second antibody (20-25 pL of fluoreseein isothioeyanate [FITC]eonjugated antibody) was added and incubated at 4''C for 30 min as before. The eells were washed three times with cold PBS/BSA and tixed with 0.5 mL \% paratormaldehyde in 0.14 mol/L phosphate buflcr containing 0.85% NaCl. pH 7.2. before analysis byflowcytometry. In the experiments using the streptavidin-biotin system, the ecIls were incubated with biotinylated-K-1-21 (25^1). washed once with eold PBS/BSA. incubated with sheep anti-mouse Ig antiserum (20 [iL). washed three times then incubated with strcptavidin-PE ( 2 0 | J L ) . washed three times, fixed as above and analysed by flow eytometry. To determine the proportion of CD5-bearing B cells in the peripheral blood of the CLL patient, a double staining technique was performed. Cells were ineubated with FITC-antihuman Ig (25 |iL). followed by PE-anti-CD5 (25 pL). after which they were fixed and analysed by flow eytometry. All ineubations were carried out at 4°C for 30 min with three washes of PBS/BSA alter each incubation. Cells expressing both CD5 and surface Ig (slg) appear in the upper right quadrant of the profile while single staining cells appear in the upper left quadrant (phycoerythrin [PE] labelled cells) or the lower right quadrant (FITC labelled eells). Unstained cells appear in the lower left quadrant. To determine the proportion of CD5bearing cells binding mouse Ig via the lowaltinity mechanism, cells were incubated with mouse Ig(K-l-21; 25 jiLof lOOpg/mL). washed once, then incubated with FITC-sheep antimouse

were analysed on a smgle-laser FACS 440 (Beeton Dickinson. Mountain View, CA. USA). Routinely. IO-* ceils were analysed per sample. A marker was set on the control histogram with 3% or less ofthe cells being to the right of this channel marker. The percentage of positive cells was then determined by subtracting cells to the right ofthis marker in theeontrol histogram from the cells to the right ofthis marker in the test histogram. FITC and PE were excited by using the 488 nm line of an argon ion laser (200 mW). Fluoreseent emissions for FITC and PE were selectively collected by using a 535+ 15 nm bandpass filter or a 575 ± 25 nm bandpass filter respectively. Surface staining was measured on a logarithmic scale. Within eacfi experiment the laser power, photomultiplier tube voltage, scatter, and tluorescence gains were kept constant. For the double-staining experiments, the cells were analysed on a FACSean tlow cytometer (Becton Dickinson). .\ total of lO** cells were analysed per sample and fluorescence was measured on a logarithmic scale. The eompensation was set on the instrument so that single labelled cells appeared in difl"erent quadrants of the resulting profile. The .v-axis showed the intensity of green fluorescence (FITC) and the y axis showed the intensity of red tluorescence (PE). FITC and PE were exeited with a wavelength of 488 nm and emissions were collected at 511 and 543 nm respectively.

Ig (20 |JL) to stabilize the binding ofthe mouse Ig. After three washing steps unoeeupied binding sites on the bound FITC-sheep anti-mouse ig were blocked by the addition of excess mouse Ig and the cells were incubated with PE-anti-CD5 (25 |iL). The eells were washed three times with PBS/BSA.fixedand analysed byflowcytometry. All incubations were carried out at 4°C for 30 min as described earlier.

To block the slgM. 10^ PBL from the CLL patient were incubated with sheep anti-human IgM antiserum for 45 min at 4°C. An aliquot of the cells was further incubated with FITCdonkey anti-sheep antiserum and examined by flow cytometry to observe that binding of the antiserum to the cell surface had taken place. The cells were washed three times and assessed for low-affinity binding of mouse Ig as described. To block the slgD. cells were incubated with polyclonal anti-human IgD antiserum for 90 min. at 4°C, washed three limes and assessed for low-affinity binding of mouse Ig as described.

flow cytometric analysis With the exception of the double-staining experiments, stained single cell suspensions

Preparation of F(ab')2 and Fab fragments

F(ab')2 and Fab fragments of K-l-21 and L7 were obtained by hydrolysis of purified K-l-21 or L7 with either pepsin or papain according to established protocols (!6). Blocking of sIgM and s/gD

264

K. M. WESTON AND R. L. RAISON

RESULTS Binding of mouse Ig to human CDS+ B cells Uinding of a range of mouse MoAb lo lymphocytes from a patient with CLL was demonstrated by indirect immunonuorcscence using FiTC-labelled sheep anti-mouse Ig and the single wash protocol outlined previously (Fig. I). Mouse lg isolypes G. M. and A, and IgG subclasses Gj. Gia. and G3 were bound by 8090% of the lymphocytes present in the peripheral blood of the CLL patient. The Mo.-Vb used have specificity for mammalian or nonmammalian antigens which arc not normally present on the surface of CD5-positive B cells, that is. anti-mouse Ly2.2 (IgM). anti-rat/mouse erythrocytcs (IgA). anti-keyhoic limpet haemocyanin (lgG|). and anti-.V. /c/j/ucdgG^a, IgG3). Double-staining of the CLL lymphocytes with anti-CD5 and a mouse igGi MoAb. K-1-21. shows that the cells binding the mouse Ig were CD5+ (Fig. 2a) and comprised approximately 94% of the total lymphocyte population. This corresponds to the proportion of CD5+ B

o

lymphocytes in the peripheral blood of this patient (93%: Fig. 2b). The proportion orCD5 + cells in the lymphocyte population was 100% as expected since both T and malignant B lymphocytes e.xpress CDS in CLL. The CD5-bearing B cells from eight of eight CLL patients tested were found to bind mouse Ig in the same manner.

Binding of mouse Ig to CLL B cells involves a low-affinity interaction The high-affinity binding of the mouse IgG| MoAb. K-1-21, to a unique surface antigen was compared to the binding of this antibody to the surface of CD5 positive B cells from a CLL patient (Table 1). K-1 -21 recognizes the KMA on the surface of the human myeloma cell line, HMy2(l5). Bindingof K-1-21'to the surface of HMy2 cells was demonstrable whether the cells were washed once or three times after incubation with K-1-21. The number of positively stained cells, and the intensity of fluorescence was identical in both cases. Because of the

PBS

IgM

igG3

2.5%

89%

86%

K A A IgGl

lgG2a

IgA

84%

81%

53%

A A A Log fluorescence intensity

Fig. 1. Binding of mouse Ig isotypes to CLL B ceils. 10^ PBL from a t LL patient were incubated with culture supernatants from hybridomas producing Mo.Ab of the isoiypc specified. Cells were washed once and incubated with FITC-shcep anti-mouse Ig before analysis by flow cytometry. Background fluorescence of cells incubated wiih i*BS followed by the second antibody is shown in the top left histogram. The scale on the .v axis comprises five logarithmic cycles.

BINDING OF MOUSE IG TO CD5 + B CELLS

K-1-21

(-low

265

affinity')

Anti—human Ig LoQ lluorescsnce intensity Fi({. 2. Double staining of CLL lymphocytes with anti-CD5 and cither mouse lg( K-1-21) or anti-human Ig. lO^* PBL from a CLL patient were incubated with (a): mouse lg( lOOjig/mL K-1-21). FITC-shccpanti-mouse Ig. and !'E anti-CD5: or(h): FITC-anti-human Ig. and PE-anti-CD5. The .v axis represents tbe intensity of green lluorcsccnce (FITC) and the y axis represents the intensity of red fluorescence (PF). The percentage of cells in each quadrant was: (a) upper left 5.0"/(i. upper right 94.1 %. lower left 0.1 %. lower right 1.2%: (b) upper left 4.5%, upper right 94.3%, Un\er left 0.4%. lower right 0.5%.

necessity to assign a particular gate setting in the flow cytometric analysis, the percentage of cells positive was determined to be approximately 70%. However the unimodal nature of the fluorescence shift indicated that all HMy2 cells expressed the K M A antigen to some degree. K-1-21 also bound to CD-5 positive B cells from the CLL patient when one washing step was used between the primary and secondary antibody steps. However i f the cells were washed three times after incubation with K-1-21. the antibody was dissociated from the cell surface and the fluorescence was reduced to background levels

(Table I). The dissociation of mouse Ig with three washing steps indicated that the bindingof K-1-21 to the CD5-positive B cells is of a lower affinity than binding of K-I-21 to the K M A epitope on HMy2 cells. Each of the mouse Mo.Ab tested in Fig. 1 could be dissociated from the B cell surface by three washing steps, whereas antibodies with high-affinity for specific cell surface antigens were not dissociated by the more extensive washing protocol. High-affinity binding via the antigen-binding site, sueh as binding of a MoAb specific for CD3 to the CD3 antigen oti the T cell surface occurred in preference to the

266

K. M. WESTON AND R L. RAISON

lable I. Effect of one or three washing steps on the binding of K-1-21 to ihe surface of CLL B cells and HMv2 cells.

Primary antibody

No. of washes

K-1-21 K-1-21

Positive ('Ml)* HMy2 CLL 70.6 64.5

94.5 I.I

IO"- PBL from a CLL palicnt or IO> HMy2 cclis were incubated ulth K-l-21 (100 |.ig/mL). .^'^ter I or 3 washing steps. FITC-sheep anti-mouse Ig was added and the cells analysed by flow cytometry. •The background fluorescence of cells incubated with PBS followed by the second antibody ( < 3%) was subtracted to determine tbe percent age positive cells.

lower-affinity binding (Table 2). Thus, at a dilution of 5Mg/niL of an anti-CD3 Mo.^b. the mouse Ig bound only to the T cells in the CLL peripheral blood (4.6% of all cells). This result was the same whether one or three washing steps was used. However, if the Mo.A.b was present at a concentration exceeding that required to saturate the high-affinity binding sites, then binding via the lower-affinity mechanism was also observed. Thus, when the concentration of antiCD3 MoAb was increased to lOOjig/mL. andthe cells were washed once after incubation with antibody, the CLL B cells were also stained (71.5% cells stained). The binding to the B cells could be disrupted by three sequential washing steps so that cells incubated with anti-CD3 MoAb at 100 and washed three times showed only the

Table 2. Binding of anti-CD3 antibody to both T and B cells from a C L L patient under appropriate conditions of concenlration and washing protocol. Anti-CD3 antibody No. of washes

5 5 100 100

1

3 1 3

Positive 4.6 3.2 71.5

8.5

10'' PBL from a CLL patient were incubated witb anti-CD3 MoAb. washed, then incubated with FITC-sbeep anti-mouse Ig before analysis by flow Otometry. •Background fluorescence of cells incubated with I'BS followed by the second antibody {