Immunoloiiy und Cell liiolo>>y

(>9. 41 1-422

Preparation and characterization of a chimeric CD 19 monoclonal antibody H. ZOLA. P. J. MACARDLE. T, BRADFORD. H. WEEDON. H. YASUP AND Y. KUROSAWAi Di'parinwni ql Clinical Immunology. I-'Undcts Medical Ccnin: Bedford Park. Somli .lu.slralia. and for Comprehensive Medical Science. I-'iijita Health L'niversiiy. foyoake. Aichi. Japan. (Submitted 28 November 199]. Acecpiai for publication 28 November 1991.) Summary FMC"63 is an lgCi2a mouse monoclonal antibody belonging to the CDI9 cluster, CDI9 aniibodics react witlui 95kDa protein expressed by cells of ihe B lymphoc\tc hneage. from pre-B cells to mature B l\niphocytcs. C D l ^ antibodies have heen suggested as candidates for immunological attack on leukaemic and lyniphoma cells of the B lineage because ihe antigen is restricted lo ihc B lineage, Wiih the polcntial use of FMC63 in immunolberapy in mind, we ha\e produced a minisc-human chimera in which the genes coding for the \'DJ region of Ihe hea\y chain and the VJ region of the liglu chain derive from the FMC63 n-iousc hybridoma. while the C region genes code for human lg(il. The genes have been transfectcd liack into a mouse myeloma line, which secretes low levels ol' immunoglobulin, {Ig). This Ig was puriticd and biolinylaied in order to determine the speci licily of the an! i body. The chimeric antibody has a read ion proltle concordant with llie original F M ( 6 3 antibodv. bul has the properties of a human IgCil. including iho ability to fi,\ human complement. However, the antJbod) is not cytoloxic in vilro in the presence of complement or cells capable of mediating antibody-dependeni cellular cytoloxicity. Possible reasons for this and ways of using the antibody are discussed.

INTRODLICITON The use of monoclonal antibodies in viva for tumour imaging and therapy has shown promising results in animal model studies and in human trials (1.2). but a number of limitations remain to be overcome. One of the major limitations is the immunogenicity of mouse monoclonal antibodies in man (3-5). e\'en in situations where the patient is profoundly iinniunosuppressed (6), The use of human monoclonal antibodies provides a possible solution. and at least one sueeessful clinical study has been reported (7). However, severe technical limitations in producing human monoclonal antibodies remain, and few have reached a stage where they can be considered for patient use. The alternative approach is to use murine monoclonal antibodies, engineered (at the protein level or at the gene level) to reduce immunogenicity. Boulianne ci al. (8) and Morrison (7 al.

(9) first eonstrueted ehimeric antibodies against haptens. TNI* and phosphor\lcholine, Lui ei al. (10). Hale ; gene, have been described elsewhere (21). SP2/0, a non-lg-seereting mouse myeloma, was used as the host for transfection (29). Introduction of constructed plasmid DNA into myeloma cells was carried out by eleetroporation (30). Growth of hybridoma, purification and biotinylation of monoclonal antibody The myeloma line eontaining the translccted ehimeric genes (referred to as FMC63Hu) was cultured in HT tiiedium. Supernatant was eolIccted 3 times eaeh week and pooled. The human

4L-!

CHIMERirCDI9MONOrLONAL.\NTIBODV

!g content of these pools, as determined by enzyme-linked immunoassay (ELISA: sec below) varied from l-5i.ig/inL. Forpurilication, 800 mL pooled supernatant was passed through a I mL bed of protein-G Sepharose(Pharmacia. Uppsala) at a flow rate of approximately 50 mL/h. in a coldroom. Unbound protein was washed oft" with 20 mmol/L phosphate buller (pH 7.0). and the bound Ig was then eluted using 0.1 mol/L glycine(pH 2.5) in 0.15 mol/L NaCl and collected into a vessel containing 1 mol/L Tris (pH 7.6) to minimize exposure at low pH. Human Ig was assayed by ELISA in all the fractions. Purity was assessed by eleetrophoresis of a concentrated sample on agarose gel. staining with Coomassie Blue and scanning the protein bands to determine the percentage of total protein constituting the Ig band. Technical details of these methods have been deseribed (\9). The purified material was eoneentratcd to 2 mg protein/niL and biotinylated as described by Kenncy et al. (31) using biotin N-suceininiide ester (NHS-D-biotin. Calbiochem. San Diego. CA). Immunofluorescence and cytotoxicity Unless stated otherwise, monoclonal antibodies were used as undiluted culture supernatants, with Ig eonecntrations of 1-5 ng/mL. Preliminary titration showed that sueh concentrations were saturating (i.e. doubling the antibody eoneentration gave only a minor increase in staining intensity). Reactivity of mouse monoclonal antibody with eells was tested by indirect inimunotluoreseence using lluoreseein-conjugated sheep antimouselg. Reactivity of human antibody was tested similarly using ant ihuman Ig. or where the target cells expressed Ig, using biotinylated monoelonal antibody and streptavidin conjugated to nuoreseein or phycoerythrin. Fluoresccnee was measured by flow eytometry using a F.'\CS Analyzer instrument (Becton Dickinson, C.\)with standard settings. Cytotoxieity was assayed by trypan blucexelusion and by the fluorcscein diacetate method (32). A number of variations in conditions were examined, as described in the results section. Horse antihunian lymphoeyte serum was used as a cytotoxie eontrol, and the target eells were NALM/6 and KM3 pre-B or R.AJl Burkitt lymphoma eell line cells. Other assays Human Ig in FMC63Hu preparations were assayed by ELISA using routine procedures. The

capture antibody was rabbit antihunian Ig. This was followed by the test sample, and the detection reagent was alkaline phosphatase conjugated sheep anti-human IgCi (Silenus Laboratories. Melbourne, Vic.) Mouse Ig was determined in a similar ELISA assay. Total protein was measured by the method of Bradford (33). RESULTS Engineering of the Chimeric antibody Active VH and V^^ genes were isolated from hybridoma FMC63 as follows. In southern hybridization of EcoRI-digested FMC63. DNA. JM probe identified two rearranged bands at 6.2 and 5.2 kb in addition to a 6.4 kb band that was also observed in P3.653 DNA (Fig. I). Sinee JH probe identified two bands at 8.5 and '*.5 kb in the BaniHI-digested FMC63 DNA in addition to the P3-derived 6.2 kb band, we judged that two copies of chromosome 12 other than the P3derived chromosome 12 were retained in the hybridoma. We eloned these three EeoRl bands in X.WES vector, and named clones FMC63H3.2, FMC63-H6.2 and FMC63-H6.4 lor the 5.2, 6.2 and 6.4 kb bands, respectively (Fig. I). Southern hybridization of the isolated clones with DH probes indicated that FMC63-H5.2 contains an abortive DH-JH structure (Fig. 1). Northern hybridization of FMC63 RN.A with FMC63-H6.2asa probe eteaily indicated that it contains an aetive VH gene expressed in this eell (Fig. 1). In southern hybridization of Hindllldigested FMC63 DNA. JK probe identified two bands at 3.8 and 5.5 kb (Fig. 2), The 5.5 kb band was observed in both FMC63 and P3 DNA. Sinee Southern hybridization of other restriction enzyme-digested FMC63 DNA with J», probe indieated that only one chromosome 6 was retained in this hybridoma other than P3derived chromosome 6 (data not shown), we cloned the 3.8 kb band in XTSK and named it FMC'63-K3.8.

Northern

hybridization

of

FMC63 RNA with F M C 6 3 - K 3 . 8 as a probe indieated that it contains an aetive V^ gene expressed in this eell (Fig. 2). The 6.2 kb EeoRI fragment of FMC63-H6.2 and the 3.8 kb Hindlll fragment of FMC63K,3.8 were inserted into the EeoRI siteof plasmid pSV2-HGlgpt and the Hindlll site of plasmid pSV2-HCKn™ (28), respectively (Fig. 3). The clones that contain V and C genes in the same orientation were selected by restriction mapping. Thus. P S V 2 - H G V F M C and pSV2-

414

H. Z O L . A / . / . I / .

a 12

3 4

HS 1 Idcmitianian ofjin active V H gene in liyhridoma FMr63. (a) .Southern hvbridi-^ation ofgcnoniic DNA vMlh JH piobc. DNA used: P.1.653, lane, i,.^: FMC63. lanes 2,4. Rcstrict.on eniymes liscd: EeoRI, lanes i 2' rniniHI. lane 3,4. (h). Southern hybndizaiion of isolated clones wiih JH probe. FMCft3-H5 2 lane I- r M C 6 3 H(v-4, lane 2: FMC6.VH6.2, lane 3. (e) Southern hybridization of isolated elones wiili D H probe FMrft3-H5 2 r w ' v l ' , .V. ,^-'""^'^- '''"'-• -• FMC-6-VHf,,:. lane 3. (d) Northern hybridization of total FMC-63 R N \ wi'th l-MC63-H6.2asa probe. ? Mg of total RNA was applied. Positions of ribosomal RNA are marked

HCK VFMC were constructed. Both plasmid DNAs were introduced into myeloma SP2/0 eells by eleetroporation. Seleetion conditions were as described previously (29). On subsequent culture, Ig levels were around 1 Mg^'mL, und the line was recloned. Two subelones were seleeted on the basis of ELLSA assay as the best produeers. .M'ter eontinued culture, the FMC63Hu.l subelone was used tor subsequent work. Yields from this subline ranged from 1-9 |ig/mL in ordinary tissue culture.

were completely concordant with results lor FMC63 (murine), exeept that K562 showed a weak reactivity with the biotmylated antibody, which was also seen with biotin\latetl negative eontrol antibody and was thus interpreted as FcR-mediated uptake of aggregated antibody. PBL showed 3-12% positive cells, and the staining was speeifie lor Beelis (Fig, 5).Threeehronie lymphocytic leukaemia samples were tested, and these gave positive reaetions with both the murine and human

Specificity of the chimeric antibody

Cross-blockinf( of murine and humanized FMC6S and relative binding affinities

FMC"63Hu antibody was compared directt\ with FM('63 (murine) antibody for staining and speeifieity. Figure 4 shows comparable staining of NALM/6 eeil line. FMC63Hu stained the following eell lines: the B eell precursors NALM/6. NALM/I6.REH and KM3; the Bcells RAJI, JVM3, RAMOS. HRIK and HMY2, FMC63HU failed to stain the T eells .IM and MOLT/4, and the myelold K562, These results

CDI9-positivc cells from a human pre-B ALL eell line NALM/6 were treated with FMr63Hu or an antibody against a diHerent antigen, followed by a lluoreseein-eonjugated murine FMr63 (Fig. 4). Human F M r 6 3 inhibited binding of nuirinc FMC63. Controls in the same experiment showed that B4, a known (1)19 antibody, also inhibited binding of EMC63,

415

CH!MERirCDI9 MONOCLONAL ANTIBODY (a)

D

FMC63-H6.2 VKDHJH

1 2

En E

28

5.5-^ 183.8-i-

Ori

kb

SV40

polyA Ecogpt

(b) FMC63-K-3,8

rig. 2. Idcntiticaiion of an active V^-gene in hybridoma FMC63. (a) SoLilhcrn hybridization of Hindllldigcstcd genomie DNA with .IK probe: lane L P3; lane 2. FMC63. (b) Northern Inbridi/ation oftotal FMC63 RNA with FMC63-K3.8 as a probe.

while FMC16. against HLA class I. did not (Fig. 4), The relative affinities of the murine and chimeric antibody were compared by titrating Ihe abilily of each to block binding of lluorescein-conjugated FMC63 (murine). The conccntfation of FMr63Hti required to give 50% inhibition (measured in ternisofthe mean tluorescence intensity) was 400ng/mL. while the concentration of" murine FMC63 required to achieve the same level of inhibition was 950 ng/mL. Sinee the Ig concentrations were necessarily measured in different ELISA assays. the dirterence may not be significant, but the ehimeric antibody appears to be more cHective at binding, and it is reasonable to conclude that the binding site has not been damaged in the process of generating the chimera. Antigenicity of chimeric antibody The chimerie antibody was tested with a number of reagents directed against murine and human

Amp

polyA

Fig.3. Construction of plasmlds(a) pSV2-HGV|rMC and (b) pSV2-Ht\-VFMr. pSVI-HCilgpt and pSV2HCKnco have only one EeoRI and Hindltl site, respectively, upstream o\' human constant genes. (En) = enhancer; (E) = EcoRI: (B)=-namHI: (H)=HindlIl.

416

Ji. Z O L A /•/ I/.

fMC63Hu/FMC63-FITC

Fluorescence intensity

I-iR.4. ( omparative staining of human and mouse FMC6.3 wuh NALM/(\ and cross-blocking experinieni For naces a-d the hollow, histogram represents the negative control antibody, while the solid histogrants are: (a) i-MC 6.1 (mouse, delected with lluoreseeinated anli-mousc Ig): (b) B4 (the original CD I

Fluorescence intensity (green) Fig. 5. Reaetion of FMC63HU with blood lymphocytes, (a) Dual parameter analysis with CD20(Bccll marker). (b) Dual staining with CD3 (T eell marker). The results show concordance orFMC63Hii binding with CD20. and no staining orCD3 (T) cells by FMC63Hu.

The cells were incubated with antibody at 3TC for 30 min, after which the complement was added and the suspensions ineubated for a further 30 min at 37°C before addition of trypan blue. Human scrum was substituted for rabbit serum as a source of complement. With human serum, the rabbit antilymphocyte serum used as a control was only weakly eytotoxic, but a horse antihuman lymphocyte serum was ctlcctivc and was tised as a positive control. In this assay, a few dead cells were seen with FMC63Hu and the assay system wasehanged to the more quantitative fluorescein diacetate method in order to detect killing ofsrnall numbers of cells. Again, a number ofconditions were attempted, including the use of" different target cells (NALM/6. NALM/16 and RAJI). the addition of a wash after the antibody incubation and before addition of complement (to remove any complexes with anticomplementary activity), and the use of cold (4°C), warm (37°C) or intermediate (25°C) incubation temperatures. In ail of these exper-

iments, the horse antihuman lymphocyte serum was eytotoxic but FMC63Hu showed either vervminor (