Human/mouse chimeric antibodies show low reactivity with human anti-murine antibodies (HAMA).

Human anti-murine antibody (HAMA) response is a serious problem in the repeated infusion of murine monoclonal antibodies (MoAbs). HAMA positive sera were obtained from seven patients with colorectal cancer, pancreas cancer, malignant melanoma or myocardial infarction who had previously received radiolabelled MoAbs. The nature of HAMA was analysed using size exclusion high performance liquid chromatography (HPLC) after incubating with radiolabelled MoAbs including IgG, Fab or human/mouse chimeric Abs. Immune complexes composed of HAMA and MoAbs were formed. The percentage of radioactivity with a high molecular weight was related to HAMA levels determined by enzyme linked immunosorbent assay. Most radioactivity present in immune complex shifted to the antibody fraction after the addition of normal murine serum. All of seven sera were reactive with all four murine IgGs and this suggests that HAMA in these patients recognised the constant region of MoAbs. In one patient, HAMA was considered to recognise the variable region and to be anti-idiotypic. There was no significant binding with human/mouse chimeric Abs in any HAMA positive serum, although five out of seven patients were reactive with murine MoAb Fab, indicating that HAMA was composed of Abs responsive to the CH1 or CL region of murine IgG. These results suggest that (1) HAMA was composed of Ab responsive to Fc portion and/or CH1 or CL region of murine IgG, and (2) human/mouse chimeric Abs look promising in the repeated infusion of MoAb in HAMA positive patients.

Most monoclonal antibodies (MoAbs) used in the diagnosis and therapy of cancer or acute myocardial infarction are derived from mice, and the development of human antimurine monoclonal antibodies (HAMA) in patients who received murine MoAbs intravenously continues to limit severely their repeated application. Circulating HAMA may form high molecular weight complexes with the injected MoAbs, resulting in rapid blood clearance and reduced tumour targeting and may cause serious sickness (Shroff et al., 1985;Shawler et al., 1985;Courtenay-Luck et al., 1986;Sakahara et al., 1989). However the development of HAMA differs among patients and the form of the MoAb; that is, whether it is whole IgG, Fab or F(ab')2 fragments (Brown et al., 1988).
Recently human/mouse chimeric MoAbs have been efficiently engineered by ligating the heavy chain enhancer element to chimeric light and heavy chain genes (Morrison et al., 1984;Nishimura et al., 1987). The antigenicity of these human/mouse chimeric Abs are expected to be reduced when intravenously infused.
We investigated the properties of circulating HAMA in the sera of patients who were administered whole IgG, F(ab')2 or Fab fragment or murine MoAbs, using size exclusion high performance liquid chromatography (HPLC) analysis. We present the heterogeneous nature of HAMA and the potential of human/mouse anti-tumour chimeric monoclonal antibodies in these HAMA positive patients.

Monoclonal antibodies
The MoAb designated ZCE-025 is a murine IgGI recognising CEA and was provided by Hybritech Inc. (San Diego, CA) through Teijin Ltd. (Tokyo, Japan) in a purified DTPAcoupled form. MoAb 96.5 is a murine IgG2a recognising melanoma cell surface antigen p97. NL-1 is a murine IgG2a antibody directed against common acute lymphocytic leukaemia antigen (CALLA) identical with CDI0 (Letarte et al., 1988;Monod et al., 1989). Human/mouse chimeric NL-1 (cNL-1) antibody was obtained by ligating the human heavychain enhancer element to the chimeric heavyand lightchain genes as previously reported (Nishimura et al., 1987;Saga et al., 1990). The Fab fragment of RIIDIO, an antimyosin antibody (Khaw et al., 1984;Khaw et al., 1987), was provided by Centocor (Malvern, PA) through Daiichi Radioisotope Laboratories Ltd. (Tokyo, Japan) in an "'In-labelled form. SF-25 antibody, generated by fusing myeloma cells and spleen cells of mice immunised with human hepatocellular carcinoma cells, is an IgG, antibody reactive with the 125 kilodalton antigen on the cell surface of some hepatocellular carcinoma and colon cancers (Takahashi et al., 1988). This antibody and its human/mouse chimeric counterpart were provided by Centocor through Toray Industries, Inc. (Tokyo, Japan).
Human anti-murine antibody (HAMA) determination by enzyme linked immunosorbent assay (ELISA) The Fab fragment of anti-myosin Ab or a mixture of murine monoclonal antibodies (whole IgG) was immobilised in 96well polystyrene plates (Costar, MA, USA) by coating a well with 5 tLg of Ab in glycine buffer pH 8.5 containing 5 mM EDTA at 4°C for 20 h. The plates were washed with 0.15 M phosphate buffered saline (PBS) containing 0.05% Tween (PBS-Tween), incubated with PBS containing 0.5% bovine serum albumin (BSA) at 37°C for an hour, and again washed with PBS-Tween. Serially diluted serum samples were added and incubated at 37°C for 1 h. After washing the plates with PBS-Tween, 50 LIl of goat anti-human IgG coupled with horseradish peroxidase was added to each well and incubated at 37°C for 1 h. The plates were washed with PBS-Tween and 200 tLI of o-phenylenediamine in citrate buffer was added and incubated at room temperature for 30 min. The optical density (O.D.) at a wavelength of 492 nm was recorded. The background O.D. was less than 0.05 (Fab, IgG). An O.D. of less than 0.3 (Fab) or 0.6 (IgG), which were the mean plus one standard deviation of a normal group respectively, was considered HAMA negative. When the O.D. was more than 0.3 or 0.6, an absorption test was performed. In short, serum was diluted in geometric progression with PBS containing murine MoAbs, and serum was also diluted with PBS as control. HAMA was determined in these aliquots by the above assay. Absorption rate was calculated as: When the O.D. was more than 0.3 (Fab) or 0.6 (IgG) and the absorption rate was more than 50%, the aliquot was considered HAMA positive. The results are expressed as dilution titers of the most diluted aliquots that showed positive HAMA tests.
Size exclusion high performance liquid chromatography (HPLC) analysis One hundred jil of serum sample was incubated for 16 h at 4°C with 50 ng of radiolabelled MoAbs in a total volume of 400pl in 0.15 M PBS containing 0.25% BSA. They were applied to HPLC system with a TSKgel G3000SW column (7.5 mm x 60 cm) (Tosoh, Tokyo, Japan) equilibrated in 0.1 M phosphate buffer. Protein was detected at the absorbance of 280 nm. An outline detector Model 170 (Beckman Instruments Inc., Berkley, CA, USA) was connected to monitor the radioactivity of each fraction. The immune complex formation was detected as radioactivity fraction of a molecular weight larger than the injected material. A com-puter program that analysed chromatographic curves was used to determine the percentage of radioactivity found in complexes. Inhibition studies were performed to ascertain whether this immune complex formation could be blocked by adding normal ICR or BALB/c murine serum or unlabelled corresponding MoAbs. After patient sera (100 l), radiolabelled MoAb (50 ng 501l-') and 250 tl of normal murine serum or unlabelled MoAb (50 fig) were mixed together in a total volume of 400 gl, and incubated for 16 h, then similarly analysed by HPLC.

Patients
HAMA positive sera were obtained from seven patients who had received radiolabelled murine MoAbs (Table I). These included three patients with colorectal cancer who received 42 mg of "'In-labelled ZCE-025 (Abdel-Nabi et al., 1987), whose serum CEA levels were all within normal limits of 2.5 ng ml-', two patients with malignant melanoma who were injected with 20mg of 96.5 and/or 20 mg of ZME-018 (Koizumi et al., 1988), one pancreatic cancer patient who was administered with 2 mg of '3'I-labelled F(ab')2 fragment of MoAb mixture recognising CA19-9 and CEA (Chatal et al., 1984), and one patient with acute myocardial infarction who received 0.5 mg of "'In-labelled Fab fragment of anti-myosin Ab Rl lDIO (Khaw et al., 1987). Serum samples were separated from peripheral blood obtained before and after infusion with radiolabelled MoAbs and stored at -20°C until use. HAMA of these seven cases determined by ELISA were all negative before MoAb administration but were positive at 2 to 4 weeks after infusion.

Results
Almost all radioactivity of radiolabelled MoAb when chromatographed alone or after incubating with normal human serum, remained associated with the IgG (Figure la) or Fab (Figure lb) peak reflecting the molecular weight of the injected material. With HAMA positive sera, examined using five radiolabelled murine IgG; ZCE-025, 96.5, NL-1,  or Fab fragment of RI 1D10, the radioactivity was associated with an entity having a higher molecular weight than original IgG (Figure Ic) or Fab peaks, forming immune complexes composed of HAMA-antibody. The percentage of immune complex formation as determined by HPLC was significantly correlated with HAMA titers determined by the ELISA assay (Table I).  after incubation with normal human serum. c, "'In-labelled ZCE-025 after incubation with HAMA positive serum (case 4), the radioactivity was eluted in a higher molecular weight peak (Vo;void volume) than the original IgG peak. d, "'In-labelled ZCE-025 after incubation with HAMA positive serum (case 4) and normal murine serum. Addition of normal murine serum displaced almost all the radioactivity to the original IgG peak.
This immune complex formation could be inhibited by incubating with excess normal murine serum and most radioactivity was displaced to the parental IgG ( Figure Id) or Fab peak. In case 2 with colorectal cancer who had previously received 42 mg of ZCE-025 Ab, and case 5 with malignant melanoma who had previously received 20 mg of 96.5 Ab, 18% and 28% of the radioactivity was still associated with the immune complexes, respectively, even after incubation with excess normal ICR or BALB/c murine serum. However, the immune complex formation could be completely absorbed by incubating with 50 pg of unlabelled ZCE-025 Ab, but not by other anti-tumour MoAbs, indicating the presence of anti-idiotype Ab in the serum of case 2. (Case 5 was not determined due to the lack of unlabelled 96.5 Ab).
Fab fragments of murine MoAbs and chimeric human/ mouse antibodies demonstrated different reactivities. Cases 2 and 3, who received whole IgG, failed to form immune complexes with RlID0 Fab as determined by HPLC analysis and ELISA assay, whereas cases 4 and 7 who were infused with F(ab')2 and Fab fragments, respectively, were reactive with "'In-labelled Fab fragment as well as with all whole IgGs examined. After incubating chimeric NL-1 and SF-25 antibodies with HAMA positive serum, minimal formation of the higher molecular weight species was detectable, which was hard to displace by adding excess normal human (data not shown) or normal murine sera (Figure 2). Similar small immune complex formation was seen in normal human serum (Table I).

Discussion
Murine MoAb infusion in humans should induce a HAMA response, which may be a key obstacle to the repeated infusion of MoAbs (Reynolds et al., 1989;Dillman, 1990). To evaluate HAMA, we used ELISA and size exclusion HPLC analysis and significant relation was observed between titers assessed by the two methods. HPLC provides a simple method to evaluate changes in molecular size as would occur with the formation of radiolabelled MoAb and HAMA (Reynolds et al., 1987). In our HPLC system, the 50 ng of radiolabelled MoAbs in 100 lI of HAMA positive serum is Retention time (min) Figure 2 Size exclusion HPLC chromatograms of '251I-labelled NL-1 and cNL-1 after incubation with HAMA positive serum (case 5). a,b, 125I-labelled NL-1 showed a higher molecular weight peak (a). Addition of normal murine serum displaced most of the radioactivity to the original IgG peak, but some of the radioactivity remained in the higher molecular weight peak (b). c,d, '25I-labelled cNL-l formed a smaller peak than the original NL-1 (c), which was slightly absorbed after incubation with normal murine serum (d). equivalent to 1.5 mg of MoAbs in 3,000 ml of plasma volume. If MoAbs exceed this quantity, the complex formation of MoAbs and HAMA decreases. Since more than 1.5 mg of MoAb is infused in most of the clinical immunoscintigraphies, our HPLC assay estimates sufficiently the degree of the in vivo complex formation. However, it is difficult to distinguish anti-idiotypic antibodies from circulating antigens by incubating sera with MoAbs and analysing the complex formation using HPLC. Binding of "'Inlabelled ZCE-025 to the serum of case 2 with colorectal cancer was not wholly blocked by excess normal murine serum or by murine MoAbs, although excess unlabelled ZCE-025 inhibited the binding completely. The serum CEA concentrations in case 2 were within upper normal limits of less than 2.5 ng ml-' (serum CEA levels exceeding 100 ng ml-' were detectable by our HPLC system, data not shown), and a part of HAMA in this patient seemed to be reactive with the idiotype of ZCE-025 antibody (Shawler et al., 1985). Fab or F(ab')2 fragments are expected to reduce the rate of developing HAMA after the injection of murine Abs. Brown et al. reported the absence of HAMA response in 663 patients who were given 0.5 mg of "'In-labelled anti-myosin Fab fragments (Brown et al., 1988). However, in the multicenter studies performed in Japan, five out of 406 (1.2%) patients developed HAMA after a single infusion of antimyosin Fab, determined using a similar ELISA assay (Kawai et al., 1990). A positive HAMA response to the IgG and Fab was also found in case 7 with acute myocardial infarction using HPLC analysis. The difference in the incidence of HAMA in patients with heart diseases between the USA and Japan remains to be explained.
Another approach, which should reduce HAMA But utilise murine MoAb is the generation of human/mouse chimeric MoAbs (Morrison et al., 1984;Brown et al., 1987). Genetic technology has made possible the exchange of the mouse constant region domains with those of human, to fabricate human/mouse chimeric antibodies, which retain the murine variable region but otherwise are human (Nishimura et al., 1987). By ligating the human heavy-chain enhancer element to chimeric lightand heavy-chain genes, human/mouse chimeric MoAbs were produced efficiently which reacted with a common acute lymphocytic leukaemia antigen (CALLA). Two human/mouse chimeric antibodies demonstrated little if any, reactivity with seven HAMA positive sera. The sera of cases 4 and 7 obtained after the infusion of F(ab')2 and Fab fragments, respectively, were reactive with both murine IgG and Fab, but not with the human/mouse chimeric monoclonal antibodies. It is very likely that the sera of the two patients and three out of five patients who received whole IgG and showed complex formation with RI lD0 Fab, reacted with the CHI and/or CL regions of the murine IgG. All seven HAMA positive sera bound to all radiolabelled murine MoAbs (IgG) and five out of seven patients were reactive with Fab of murine MoAb. In contrast, cNL-l or cSF-25 MoAbs were not reactive with these HAMA positive sera. In the clinical studies using the human/ mouse chimeric MoAbs, the antibody responses of the patients depend upon the infused chimeric MoAb, and the immunogenic potential of the chimeric MoAbs ranges wide (Meredith et al., 1991). The human/mouse chimeric MoAbs, especially those with low immunogenicity, are useful for repeated infusion of anti-tumour MoAbs in HAMA positive patients. Meanwhile, excess normal murine serum did not completely displace the binding of NL-1 or SF-25 MoAbs to HAMA positive serum and up to 15 to 24% of radioactivity was observed in a large molecular weight peak even after incubating with HAMA negative normal human serum. NL-1 recognises common acute lymphocytic leukaemia antigen (CALLA), which is identical with neutral endopeptidase (Letarte et al., 1988;Monod et al., 1989), and is clinically employed as an important cell surface marker for the diagnosis of human acute lymphocytic leukaemia. However, CALLA is not restricted to leukaemic cells and is also found on a variety of normal tissues (Losa et al., 1986). Complexes formed after adding normal murine serum or when incubated with normal human serum are most likely due to the binding of circulating antigens with radiolabelled MoAbs.
In summary, (1) HAMA of all seven patients formed complexes with all four murine IgGs, but not with two human/mouse chimeric antibodies, (2) five out of seven patients showed reactivity with Rl11DIO Fab in their sera, suggesting that they have HAMA recognising CHI or CL region, (3) one patient seemed to have anti-idiotypic antibody. HAMA showed variety in reactivity with murine MoAbs but human/mouse chimeric antibodies are promising even in most patients who became HAMA positive after repeated infusion with murine MoAbs.