Apoptosis assays with lymphoma cell lines: problems and pitfalls

Much attention has been focused on the manner in which tumour cells die after treatment with cytotoxic agents. The basic question is whether cells die via apoptosis or via direct damage from the toxic agent. Various assays have been used to make this distinction. However, we show herein that some of the widely used assays for apoptosis do not in fact distinguish between apoptosis and other forms of cell death. More specifically: (1) A sub-G₁ DNA content, identified by propidium iodide staining, does not distinguish between apoptotic and necrotic cells; (2) loss of mitochondrial membrane potential does not distinguish between apoptotic and necrotic cells, unless combined with an assay for an intact cell membrane; (3) subcellular fragments that arise from dead cells or from apoptotic bodies can interfere with some assays for apoptosis such as annexin V staining, as they may be close to the size of intact cells, making it difficult to decide where to set the size threshold; (4) irradiated cells display a large increase in nonspecific Ab binding. This may be partly due to an increase in cell size, but, regardless of the cause, it can lead to a mistaken conclusion that there is an increase in a particular antigen if appropriate control reagents are not tested; and (5) experiments utilising Ab crosslinking have neglected the role of cell aggregation, which can cause multiple problems including death from mechanical stress when cells are handled. Consideration of these factors will improve our ability to determine the mode of cell death.

Antibodies to CD20 and MHC class II antigen bound to B-lymphoma cells accumulate in shed cytoplasmic fragments

Antibodies (Abs) to CD20 and MHC class II antigen were found to exhibit a novel processing pathway after binding to the surface of RL B-lymphoma cells. The Abs were ‘excreted’ as a part of large cytoplasmic fragments. These fragments formed at cell–cell junctions, but gentle dispersal of the cells, to form a single-cell suspension of high viability, caused the release of most of the fragments. This process also occurred in Raji cells and in three other B-lymphoma cell lines (of seven tested). Six B-lymphoblastoid cell lines tested did not form these objects. Once they were recognised, the fragments could be identified in cell preparations by phase contrast microscopy or after staining with Wright's stain. They were induced by the binding of certain Abs, but not by most Abs bound to the cell surface. The mode of formation, detailed morphology and function of these cytoplasmic fragments remain to be determined. They are similar in many respects to the lymphoglandular bodies that have been described by pathologists for many years, which are characteristic of B-cell lymphoma, but which have not previously been described in cell lines. This type of Ab processing, if it occurs in patients, will have an impact on the therapeutic use of these Abs.

Therapy of disseminated B-cell lymphoma xenografts in severe combined immunodeficient mice with an anti-CD74 antibody conjugated with (111)indium, (67)gallium, or (90)yttrium

A radiolabeled antibody (Ab) to CD74 (the MHC class II invariant chain, Ii) was shown previously to effectively kill human B-lymphoma cells in vitro. Conjugates with both Auger electron and beta-particle emitters were able to kill cells, but the former displayed less nonspecific toxicity in the in vitro assay used. In this report, we have extended the studies to an in vivo model of tumor growth. The human B-cell lymphoma Raji was injected i.v. into severe combined immunodeficient mice, and radiolabeled Abs were injected at various times after tumor inoculation. The maximum tolerated dose (MTD), as well as lower doses, was tested. Tumor growth was monitored by hind-leg paralysis. With a 3-5-day interval before Ab injection, anti-CD74 conjugated to either (111)In or (67)Ga, at a dose of 240-350 microCi/mouse, produced a strong therapeutic effect, with greatly delayed tumor growth, and many of the treated mice were tumor free for >6 months. Control mice became paralyzed in 16-24 days, uniformly. Treatment at later time points (9-day interval) had little therapeutic effect. The MTD was required for optimal therapy. With the beta-particle emitter (90)Y, the MTD was much less, 25 microCi/mouse, and at this dose there was only a weak therapeutic effect. In conclusion, the data suggest that low-energy electrons are more effective than beta-particles in this model system. These results may be applicable to humans, particularly in the case of micrometastatic disease. This approach may also be effective with other Abs that accrete in large amounts.

Single-cell cytotoxicity with radiolabeled antibodies

Previous studies demonstrated the effective, antigen-specific killing of Raji B-lymphoma cells in vitro by radiolabeled anti-CD74, attributable largely to the high level of uptake, of approximately 10(7) antibody (Ab) molecules/cell/ day. This Ab is rapidly delivered to lysosomes for catabolism, so the radionuclide delivered accumulates primarily in lysosomes. In this study, we have tested Abs that bind to the same target cells in similar amounts, but remain primarily on the cell surface, to compare the potency of radioactivity delivered to the cell surface versus the cytoplasm. The Abs tested were anti-major histocompatibility complex class II and anti-CD20. 111In-labeled conjugates made with these two Abs killed cells very effectively and specifically, with 100% kill of sample of 5 x 10(5) cells. Because these Abs remain primarily on the cell surface, it would be predicted that residualizing radiolabels, which are trapped in lysosomes after Ab catabolism, would not be required, and this was observed, i.e., these two Abs were effective when labeled with either 125I or 131I, using conventional iodination, as well as with the residualizing label 111In-labeled DTPA. These results are in contrast to results obtained with anti-CD74, which required a residualizing radiolabel for effectiveness. The uptake of these radionuclides, in cpm/cell, was monitored, and this allowed estimation of the radiation dose delivered; the cytotoxicity observed was consistent with the estimated radiation dose delivered. To establish the generality of the results, we also demonstrated that 111In-labeled anti-CD74 effectively killed three other B-lymphoma cell lines, in addition to Raji and the adherent melanoma cell line SK-MEL-37. By using more potent radionuclides or conjugates of higher specific activity, this approach might be effective with other, lower density antigens.

Radionuclides linked to a CD74 antibody as therapeutic agents for B-cell lymphoma: comparison of Auger electron emitters with beta-particle emitters

We demonstrated previously that human B-cell lymphomas were effectively and specifically killed in vitro by an antibody to CD74 (LL1) linked to (111)In or other Auger electron emitters. This study was intended to more accurately compare the potency and specificity of 3Auger electron emitters, (111)In, 67Ga, and 125I, and to evaluate beta-particle emitters, 131I and 90Y. The unique property of LL1 is its high level of intracellular uptake.

METHODS

Raji B-lymphoma cells were incubated with serial dilutions of the radiolabeled Abs for 2 d and then monitored for cell growth by 2 assays: a cell counting assay and a clonogenic assay. The uptake of radioactivity per cell was monitored at various time points, and the radiation dose was calculated using published S values for radioactivity located in the cytoplasm. Both specific and nonspecific toxicity were evaluated.

RESULTS

The beta-particle emitters had considerably higher levels of nonspecific toxicity than the Auger electron emitters, but both 131I and 90Y, and particularly 131I, still had high levels of specificity. Both of these results were consistent with dosimetry calculations. Relative to the delivered disintegrations per cell, 131I and 67Ga were the most potent of the radionuclides tested, with 125I and (111)In being significantly weaker and 90Y being intermediate. The high potency of 67Ga, together with its low nonspecific toxicity, caused this radionuclide to have the highest specificity index.

CONCLUSION

When delivered by Ab LL1, both Auger electron and beta-particle emitters can produce specific and effective toxicity. The choice of the optimal radionuclide for therapy may depend on the ease and efficiency of labeling, the specific activity obtained, the nature of the tumor being targeted, and other factors, but the high specificity indices of the Auger electron emitters may be an advantage.

Limitations in the use of low pH extraction to distinguish internalized from cell surface-bound radiolabeled antibody

Internalization by cells of radiolabeled protein ligands bound to the cell surface is frequently analyzed by extraction of the cells with low pH buffers. This treatment supposedly strips the ligands from the cell surface, and remaining molecules are considered to be internalized. However, we show herein that: (1) low molecular weight catabolic products that are trapped within lysosomes (residualizing radiolabels) are efficiently extracted by low pH buffers, under the same conditions used to remove cell surface-bound material, and (2) low pH treatment lyses the majority of the cells, as shown with both a nonadherent and an adherent cell line, with the release of most of a (51)Cr label. Still, low pH extraction was effective at demonstrating Ab internalization, as has been demonstrated many times. These effects of low pH treatment may be attributed to the fixative properties of these buffers. Regardless of the mechanism, these data must be taken into consideration in interpreting the results of such experiments.

Cell surface expression and metabolism of major histocompatibility complex class II invariant chain (CD74) by diverse cell lines

We previously described the processing of antibodies to CD74 (the major histocompatibility complex class II-associated invariant chain, Ii), by B-cell lymphoma cell lines. These cells expressed relatively low levels of Ii on the surface, but the molecules were rapidly internalized and replaced by new molecules, so that approximately 8 x 10(6) antibody molecules per cell were taken up per day. We herein report the results of similar studies with other cell types, namely a melanoma, a colon carcinoma, a T-cell lymphoma and B-lymphoblastoid cell lines. The melanoma and the carcinoma were treated with interferon-gamma to induce high levels of the antigen. The T-cell lymphoma, HUT 78, was selected specifically because it was previously reported to lack cell surface Ii, while expressing the molecule intracellularly. However, HUT 78 displayed Ii on the cell surface, as did the other cell lines tested, and catabolism of the antibody was very fast on all of the cell lines. The capacity of four of the cell lines for cumulative antibody uptake was evaluated, using 'residualizing' radiolabels, which are trapped within the cell after catabolism of the antibody to which they were conjugated. A high level of uptake was observed in all cases, although there was significant variation between the cell lines. With melanoma SK-MEL-37, the total LL1 uptake in 24 hr was nearly 10(7) molecules per cell and the average turnover time for Ii on the cell surface was 4 min; with carcinoma HT-29, the total LL1 uptake in 24 hr was approximately 10(6) molecules per cell, and the average turnover time for Ii on the cell surface was 27 min. Based on the cell content of mature class II antigens (alphabeta), these data suggest that a large fraction, or all, of immature class II molecules (alphabetaIi) reach the cell surface before entering the peptide-loading compartment, independent of the particular cell type.

Targeting human cancer xenografts with monoclonal antibodies labeled using radioiodinated, diethylenetriaminepentaacetic acid-appended peptides

A new nonmetabolizable peptide approach to the production of residualizing radioiodine was evaluated in nude mice bearing xenografts of human lung adenocarcinoma (Calu-3) and B-cell lymphoma (Ramos). Monoclonal antibodies (MAbs) RS7 (anti-epithelial glycoprotein-1) and LL2 (anti-CD22) were radioiodinated using the thiol-reactive diethylenetriaminepentaacetic acid-D-peptide adducts IMP-R1 and IMP-R2. 125I-IMP-R1- and 125I-IMP-R2-labeled MAbs were compared to the MAbs iodinated by the conventional chloramine-T approach, (111)In, and 131I-dilactitoltyramine (DLT). In vivo biodistribution studies demonstrated a significant improvement in the tumor accretion of radiolabel using the 125I-IMP-R1 labeled MAbs compared with the conventionally iodinated antibodies. For example, at day 7, the percentage of injected dose per gram of tissue in Calu-3 was 7.9 +/- 4.1% and 18.1 +/- 7.9% (P < 0.05) for the conventional 131I- and 125I-IMP-R1-RS7, respectively, and tumor:nontumor ratios were 2.6-4.5-fold higher with the 125I-IMP-R1-RS7. It is estimated that 131I-IMP-R1-RS7 would deliver a dose to tumor (at the estimated maximum tolerated dose) 3.9 times greater than conventional 131I-labeled RS7, 1.4 times greater than 90Y-labeled RS7, and 0.7 times that of 131I-DLT-labeled RS7. Tumor accretion of 125I-IMP-R2-RS7 was also improved compared with conventionally iodinated antibody. However, this label also caused a large increase in kidney accretion. Similar improvements in tumor accretion and tumor:nontumor ratios were observed when 125I-IMP-R1-LL2 was used in the Ramos model. IMP-R1 offers a practical and useful residualizing radioiodine label because labeling efficiency is at least 10 times greater than that of the residualizing label DLT, without MAb aggregation. Structural modifications can be envisioned for further improvements in radioiodine incorporation, specific activity, and tumor dosimetry, and efforts along these lines are under way.

Processing of antibodies to the MHC class II antigen by B-cell lymphomas: release of Fab-like fragments into the medium

Lym-1, an anti-MHC class II Ab, displayed a unique processing pathway after binding to the surface of Raji B-lymphoma cells, in which Fab-like fragments were gradually released into the medium. The fragments had reduced interchain disulfide bonds. Fragmentation was markedly reduced by inhibitors of intracellular catabolism, namely ammonium chloride, chloroquine and leupeptin. The capacity of the process was high, and fragmentation of approximately 5x10(6) Ab molecules per cell per day was measured directly, in what can be considered to be a minimum estimate. Five other Abs to the MHC class II antigen were tested similarly on Raji and on three other B-cell lymphomas: none showed the same high level of fragmentation seen with Lym-1 binding to Raji, but significant fragmentation did occur with some of the Abs, particularly EDU-1 and L243. The level of fragmentation depended on the cell line as well as on the particular Ab. The other 5 Abs were all catabolized, to low molecular weight material, much more extensively than Lym-1. Part of the difference between Abs can probably be attributed to the fortuitous, preferential labeling of Lym-1 on the light chain, since the data suggest that the Fc fragment is fully degraded while the Fab-like fragment is released into the supernatant. This pathway of Ab processing is likely to be related to the physiology of the MHC class II antigen, which recycles into a mildly proteolytic intracellular compartment.

Enhancement of tumor-to-nontumor localization ratios by hepatocyte-directed blood clearance of antibodies labeled with certain residualizing radiolabels

To increase tumor-to-nontumor localization ratios of injected radiolabeled antibodies (Abs), several interrelated methods were used.

METHODS

The model systems used were two human carcinoma xenografts grown in nude mice, targeted by antibodies RS11 (antiepithelial glycoprotein-2) or MN-14 (anticarcinoembryonic antigen). The Abs were conjugated with biotin and 111In-benzyl diethylenetriamine pentaacetic acid, and, at various times after injection, were cleared by intraperitoneal injection of galactosylated streptavidin, which delivers the complexes to hepatocytes. The radiolabel used was selected because it is retained within tumors after catabolism of the Ab by the tumor cell but is quite rapidly excreted from hepatocytes into bile.

RESULTS

With blood clearance induced at 24 h, and dissection 5 h later, high tumor-to-nontumor ratios were attained. Depending on the model used, tumor-to-blood ratios were 16:1 to 31:1, and tumor-to-nontumor ratios for the kidney, lungs and bone were also high and greatly increased by the clearance regimen. Despite clearance into the liver, tumor-to-liver ratios remained >1, due to fairly rapid biliary excretion of the label. The absolute antibody uptake by the tumors was also high, because 24 h was allowed for the Ab to penetrate and bind to cells within the subcutaneous tumors.

CONCLUSION

The method described produced high tumor-to-nontumor ratios at 1 d after injection and may be advantageous for tumor imaging with antibodies. Radiation dosimetry calculations indicate that there is only a slight advantage with this approach for radioimmunotherapy.

Cytotoxicity with Auger electron-emitting radionuclides delivered by antibodies

We investigated the in vitro cytotoxic potential of Auger electron-emitting radionuclides delivered to the cytoplasm or, more specifically, to lysosomes, via antibodies. The antibody (Ab) used was LL1, which is specific for CD74, an epitope of the major histocompatibility complex (MHC) class II antigen invariant chain, Ii, present on the cell surface. It is taken up in large amounts, approximately 10(7) Ab molecules per cell per day, and delivered to lysosomes. The radioisotopes tested included (111)In, 99mTc and 125I. With sufficient specific activity, approximately 10 mCi/mg Ab, all of these isotopes were potent cytotoxic agents. 125I was active only if a "residualizing" form was used, meaning a form that is trapped within cells after catabolism of the Ab to which it was conjugated (conventional oxidative iodination produces a non-residualizing label). The conjugates of (111)In and 99mTc used are known to be residualizing. One hundred percent cell kill in vitro was obtained with (111)In and 125I, under conditions in which a non-reactive control Ab, conjugated in the same way, produced no significant toxicity. 99mTc was also potent and specific, but appeared somewhat less active than the other isotopes under the conditions evaluated. Although few Abs are accreted by cells at the same rate as LL1, it may be possible to use other Abs to deliver similar amounts of radioactivity, if Abs with higher specific activity can be produced. Such conjugated radioisotopes may be useful for attacking tumor cells in vivo, particularly for single cells or micrometastases.

Labeling of monoclonal antibodies with diethylenetriaminepentaacetic acid-appended radioiodinated peptides containing D-amino acids

The optimal use of radioiodinated internalizing monoclonal antibodies (mAbs) for radioimmunotherapy necessitates the development of practical methods for increasing the level of retention of 131I in the tumor. Lysosomally trapped ("residualizing") iodine radiolabels that have been previously designed are based mostly on carbohydrate-tyramine adducts, but these methods have drawbacks of low overall yields and/or high levels of mAb aggregation. We have developed a method using thiol-reactive diethylenetriaminepentaacetic acid (DTPA)-peptide adducts wherein the peptides are assembled with one or more D-amino acids, including D-tyrosine. Two such substrates, R-Gly-D-Tyr-D-Lys[1-(p-thiocarbonylaminobenzyl)DTPA], referred to as IMP-R1, and [R-D-Ala-D-Tyr-D-Tyr-D-Lys]2(CA-DTPA), referred to as IMP-R2, wherein R is 4-(N-maleimidomethyl)cyclohexane-1-carbonyl, were synthesized by preparing functional group-protected peptides on a solid phase, selectively derivatizing the lysine side chain with 1-(p-isothiocyanatobenzyl)DTPA or DTPA dianhydride (CA-DTPA), deprotecting other functional groups, and finally derivatizing the peptide's N-terminus so it contained a maleimide group. Radioiodinations of the peptides followed by conjugations to disulfide-reduced mAbs, carried out as a one-vial procedure, resulted in 32-89% overall yields, at specific activities of 1.8-11. 1 mCi/mg, with less than 2% aggregation. Two internalizing mAbs, LL2 (anti-CD 22 B-cell lymphoma mAb) and RS7 (an anti-adenocarcinoma mAb which targets EGP-1 antigen), labeled with this procedure exhibited a 2-3-fold better cellular retention in Ramos and Calu-3 tumor cell lines, in vitro, respectively, compared to the same mAbs radioiodinated with the chloramine-T method. The rationale for the new approach, syntheses, radiochemistry and in vitro data are presented.

Carcinoembryonic antigen as a target for radioimmunotherapy of human medullary thyroid carcinoma: antibody processing, targeting, and experimental therapy with 131I and 90Y labeled MAbs

The poor prognosis of patients with advanced medullary thyroid carcinoma (MTC) has prompted a search for new treatment modalities. In this report we explore the characteristics of carcinoembyronic antigen (CEA) as a target for radioimmunotherapy (RAIT) of MTC, with respect to antibody processing, targeting, and experimental therapy. In vitro studies showed a high level of CEA expression on the cell surface of the MTC cell line TT. MAbs bound to the cell were predominantly retained for several days, although there was also a significant level of internalization and catabolism. Immunohistology of frozen sections of tumor xenografts demonstrated that approximately half of the cells were darkly stained, however, some cells expressed little or no CEA. In biodistribution studies in nude mice bearing TT tumors, the mean percent injected dose per gram of tumor observed at three days post injection (time of maximum uptake) of 125I-MN-14 was 19.7%. When the MAb was labeled with 88Y, a residualizing label, a much higher accretion, 50.5%, was observed at the time of maximum uptake (7 days). Significant anti-tumor effects were seen at the maximum tolerated doses of 131I- and 90Y-MN-14, compared with relatively rapid tumor growth in untreated animals or those treated with the same dose of control MAbs. Importantly, it was observed that 90Y-MN-14 yielded significantly improved therapeutic efficacy in comparison to 131I-MN-14, which may have important implications for design and conduct of future clinical trials for the treatment of MTC.

Intracellular processing of 99Tcm-antibody conjugates

The catabolism of 99Tcm-antibody conjugates after internalization by B-cell lymphomas was investigated, using antibody LL1, an antibody to the MHC class II invariant chain which is internalized and catabolized very rapidly. Intact IgG antibodies were labelled with 99Tcm after mild reduction. The 99Tcm label was strongly retained within cells, similar to 'residualizing' labels such as 111In-diethylenetriamine pentaacetate (111In-DTPA), but different from a conventional iodine label. Unlike 111In-DTPA, 99Tcm was not retained in a low molecular weight form, but instead was found to be bound to a large number of different cellular proteins, and was retained in the cytoplasm rather than in lysosomes. Therefore, this form of 99Tcm represents a new paradigm of intracellular retention of a radiolabel.

Rapid blood clearance of mouse IgG2a and human IgG1 in many nude and nu/+ mouse strains is due to low IgG2a serum concentrations

We reported previously that the blood clearance of injected mouse IgG2a was extremely rapid in many strains of nude and nu/+ mice. In an attempt to determine the cause of this phenomenon, the levels of endogenous IgG2a in the blood of these mice was assayed. It was found that the serum level of IgG2a was extremely low in many of these mice, below 50 microg/ml, which is 20-100 times lower than the expected normal value. Great heterogeneity between individual mice was observed in their blood level of IgG2a, and there was an excellent correlation between low blood IgG2a levels and rapid clearance of injected IgG2a. Thus, the blood IgG2a levels are so low that a novel, previously undescribed effect occurs, namely the rapid clearance of small amounts of injected IgG2a. The clearance is due primarily to binding sites in the spleen and liver. The low level of endogenous IgG2a is not due to the lack of a thymus, since it occurs in nu/+ as well as nude mice, but can probably be attributed to the very clean environment in which these mice are raised. In assays of sera from approximately 50 mouse strains, low IgG2a levels were found in all nude colonies and also in some normal mouse strains. Some nude mice displayed relatively normal IgG2a clearance rates despite having low levels of endogenous IgG2a. In repeated bleedings of individual mice, IgG2a levels were found to fluctuate greatly. A similar clearance effect was observed with a human IgG1 Ab injected into mice. This rapid clearance of injected IgG, of certain subclasses, represents a practical problem for many experiments in which antibodies are used for diagnosis or therapy, and several methods of circumventing the problem are discussed.

Processing of antibodies bound to B-cell lymphomas and lymphoblastoid cell lines

BACKGROUND

Previous experiments demonstrated that some human B-cell lymphoma cell lines were unusual in that antibodies bound to the cell surface dissociated at high levels. This did not occur with non-B-cell hematologic tumors or with carcinomas. In this study, additional B-cell lymphoma and lymphoblastoid (Epstein-Barr virus-transformed) cell lines were tested.

METHODS

The antibodies selected for most experiments, MA103 and anti-CD45, react with relatively high avidity to the cell surface. Antibodies to CD19, CD20, and CD22 also were tested on certain cell lines. The antibodies were labeled with 125I. After binding to the surface of viable cells, unbound antibody was washed away, and the fate of the bound antibody was investigated for 2-3 days.

RESULTS

Of the eight B-cell lymphomas tested, three had high levels of dissociation, two had low levels of dissociation, and three had intermediate levels of dissociation. The six lymphoblastoid cell lines had only slightly elevated levels of dissociation, relative to non-B cell lines. Sublines of Raji and Ramos cells were identified that varied greatly in the level of antibody dissociation. The level of dissociation from lymphomas was correlated with the tendency of the cell lines to cluster, with single cells displaying less dissociation than clustered cells. However, some exceptions to this correlation were noted. Cell lines such as Ramos, which showed little dissociation of anti-CD20, displayed relatively rapid catabolism of this antibody.

CONCLUSIONS

The level of antibody dissociation as well as the rate of antibody catabolism will affect the results of radioimmunotherapy strongly because these factors affect the time interval for which the cells are in contact with the radioisotope. Different B-cell lines display markedly different levels of dissociation. There is some evidence suggesting that antibody dissociation is high with fresh human tumor cells, but further investigation of this point is required.

Manipulation of blood clearance to optimize delivery of residualizing label-antibody conjugates to tumor cells in vivo

We have attempted to improve the therapeutic index of radioimmunotherapy by manipulating the blood clearance rate and the catabolism of the radiolabel. The general strategy is to allow the antibody (Ab) to circulate in the blood for 2-3 days, then to clear it rapidly by a method that delivers the Ab to hepatocytes. In addition, the radiolabel selected has two key properties: it is a residualizing label (which is lysosomally trapped after catabolism), so it is retained well by tumor cells, but is excreted rapidly by hepatocytes into bile.

METHODS

In initial experiments, three residualizing radiolabels were tested for their rate of excretion after specific delivery in vivo to either hepatocytes, via galactosylated Ab, or Kupffer cells, via immune complexes. A label showing rapid biliary excretion only after delivery to hepatocytes, (111)In-benzyl-diethylenetriamine tetraacetic acid, was then used for radioimmunodetection in a protocol of delayed rapid blood clearance in which clearance was by hepatocytes. This was achieved by using galactosylated Ab, combined with temporary inhibition of the asialo-glycoprotein receptor on hepatocytes. Ab RS11 and the lung adenocarcinoma Calu-3 xenograft in nude mice were used. Control experiments were performed with a conventional 125I label and with 125I-dilactitol-tyramine.

RESULTS

Indium-benzyl-diethylenetriamine tetraacetic acid was identified as a label that was excreted more rapidly from hepatocytes than from Kupffer cells, by biliary excretion. Using this radiolabel with delayed rapid blood clearance, very high tumor/blood ratios were obtained, 166:1 at day 3, but tumor/normal tissue ratios for other tissues were not as high. There appeared to be some uptake of the radiolabel by all normal tissues tested, including the lungs and muscle. Dosimetry calculations suggested that the therapeutic index was no better than with a simple Ab injection.

CONCLUSION

Antibody catabolism can be directed towards either hepatocytes or Kupffer cells, and this difference can strongly affect the excretion rate of radiolabels, since only hepatocytes can excrete degradation products into bile. Processing will also depend on the particular radiolabel. These factors are particularly important for protocols involving delayed rapid blood clearance, since liver uptake is so rapid. The methods described should stimulate other approaches of manipulating Ab blood clearance and radiolabel catabolism to achieve improved therapeutic results.

Development of a streptavidin-anti-carcinoembryonic antigen antibody, radiolabeled biotin pretargeting method for radioimmunotherapy of colorectal cancer. Studies in a human colon cancer xenograft model

Pretargeting methodologies can produce high tumor:blood ratios, but their role in cancer radioimmunotherapy (RAIT) is uncertain. A pretargeting method was developed using a streptavidin (StAv) conjugate of MN-14 IgG, an anti-carcinoembryonic antigen (CEA) murine monoclonal antibody (mab) as the primary targeting agent, an anti-idiotype antibody (WI2 IgG) as a clearing agent, and DTPA- or DOTA-conjugated biotin as the radiolabeled targeting agent. A variety of reagents and conditions were examined to optimize this method. At 3 h, 111In-DTPA-peptide-biotin tumor uptake was 3.9 +/- 0.8% per gram and tumor:blood ratios were > 11:1. By 24 h, this ratio was 178:1, but tumor accretion declined in accordance with the gradual loss of StAv-MN-14 from the tumor. Tissue retention was highest in the liver and kidneys, but their tumor:organ ratios were > 2:1. Dosimetry predicted that radiolabeled MN-14 alone would deliver higher tumor doses than this pretargeting method. Increasing the specific activity and using DOTA-biotin in place of DTPA increased tumor uptake nearly 2-fold, but analysis of StAv-MN-14's biotin-binding capacity indicated over 90% of the initial biotin-binding sites were blocked within 24 h. Animals fed a biotin-deficient diet had 2-fold higher 111In-DOTA-biotin uptake in the tumor, but higher uptake also was observed in all normal tissues. Although exceptionally adept at achieving high tumor:blood ratios rapidly, the tumor uptake of radiolabeled biotin with this pretargeting method is significantly (p < 0.0001) lower than that with a radiolabeled antibody. Endogenous biotin and enhanced liver and kidney uptake may limit the application of this method to RAIT, especially when evaluating the method in animals, but with strategies to overcome these limitations, this pretargeting method could be an effective therapeutic alternative.

The advantage of residualizing radiolabels for targeting B-cell lymphomas with a radiolabeled anti-CD22 monoclonal antibody

CD22 antibodies (Abs) bound to B-cell lymphomas are known to be internalized and catabolized rapidly. Therefore, it would be expected that use of CD22 as a target for radioimmunotherapy should be enhanced by the use of "residualizing" radiolabels, which are trapped within the cell after catabolism of the Ab to which they had been conjugated. Our study was intended to evaluate this hypothesis using Ab LL2. In initial experiments, we found that LL2 binding was strongly temperature dependent, with approximately 15-fold greater binding at 37 degrees C than at 0 degrees C. A series of experiments suggested that this difference is due to a conformational change in the antigen at low temperature, so that the LL2 epitope is partially blocked. In vitro, residualizing labels-including 125I-dilactitol tyramine and 111In-DTPA-were retained by cells much longer than a conventional iodine label. In vivo, residualizing labels also showed a marked advantage in terms of uptake by Ramos B-cell lymphoma xenografts in nude mice. However, the absolute Ab uptake by xenografts was quite low, in comparison with results obtained with many carcinoma xenografts, which appears to be due in part to vascular properties of the B-cell lymphoma xenografts.

Advantage of residualizing radiolabels for an internalizing antibody against the B-cell lymphoma antigen, CD22

LL2 is an anti-CD22 pan-B-cell monoclonal antibody which, when radiolabeled, has a high sensitivity for detecting B-cell, non-Hodgkin's lymphoma (NHL), as well as an antitumor efficacy in therapeutic applications. The aim of this study was to determine whether intracellularly retained radiolabels have an advantage in the diagnosis and therapy of lymphoma with LL2. In vitro studies showed that iodinated LL2 is intracellularly catabolized, with a rapid release of the radioiodine from the cell. In contrast, residualizing radiolabels, such as radioactive metals, are retained intracellularly for substantially longer. In vivo studies were performed using LL2-labeled with radioiodine by a non-residualizing (chloramine-T) or a residualizing method (dilactitol-tyramine, DLT), or with a radioactive metal (111In). The biodistribution of a mixture of 125I (non-residualizing chloramine-T compared to residualizing DLT), 111In-labeled LL2 murine IgG2a or its fragments [F(ab')2, Fab'], as well as its humanized, CDR-grafted form, was studied in nude mice bearing the RL human B-cell NHL cell line. Radiation doses were calculated from the biodistribution data according to the Medical International Radiation Dose scheme to assess the potential advantage for therapeutic applications. At all assay times, tumor uptake was higher with the residualizing labels (i.e., 111In and DLT-125I) than with the non-residualizing iodine label. For example, tumor/blood ratios of 111In-labeled IgG were 3.2-, 3.5- and 2.8-fold higher than for non-residualizing iodinated IgG on days 3, 7 and 14, respectively. Similar results were obtained for DLT-labeled IgG and fragments with residualized radiolabels. Tumor/organ ratios also were higher with residualizing labels. No significant differences in tumor, blood and organ uptake were observed between murine and humanized LL2. The conventionally iodinated anti-CD20 antibody, 1F5, had tumor uptake values comparable to those of iodinated LL2, the uptake of both antibodies being strongly dependent on tumor size. These data suggest that, with internalizing antibodies such as LL2, labeling with intracellularly retained isotopes has an advantage over released ones, which justifies further clinical trials with residualizing 111In-labeled LL2 for diagnosis, and residualizing 131I and 90Y labels for therapy.

The Oka blood group antigen is a marker for the M6 leukocyte activation antigen, the human homolog of OX-47 antigen, basigin and neurothelin, an immunoglobulin superfamily molecule that is widely expressed in human cells and tissues

The high-frequency blood group antigen Ok(a) is carried on a red cell membrane glycoprotein (gp) of 35-69 kDa that is widely distributed on malignant cells of different origins. Immunostaining of hemopoietic cells and a range of normal human tissues demonstrated a wide distribution of the Ok(a) gp that appears to be nonlineage-restricted, although certain tissues show differentiation-related expression. Ok(a) gp was purified from red cell membranes by immunoaffinity chromatography using mAb A103 and amino acid sequence analysis was performed. The N-terminal 30 amino acids are identical to the predicted sequence of M6 leukocyte activation antigen (M6), a member of the Ig superfamily (IgSF) with two IgSF domains. There are homologs in rat (MRC OX-47 or CE9), in mouse (basigin or gp42), and in chicken (HT7 or neurothelin). The molecular basis of the Ok(a) mutation was established by sequencing M6 cDNA derived from normal and Ok(a-) EBV-transformed B cell lines. A point mutation in the translated portion of M6 cDNA, G331AG-->AAG gives rise to a predicted E92-->K amino acid change in the first Ig-like domain of the Ok(a-) form of the protein. Transfection of mouse NS-0 cells with normal or Ok(a-) cDNA confirmed the identity of the protein and only the Ok(a-) transfectants failed to react with monoclonal anti-Ok(a) Ab.

Binding parameters of antibodies reacting with multivalent antigens: functional affinity or pseudo-affinity

For antibodies binding bivalently to multivalent antigens, including the cell surface, it has been common to determine the 'functional affinity', even though it is clear that the equations describing true affinity only apply to monovalent interactions. We here summarize evidence that 'functional affinities' are not valid, because they do not accurately describe the interactions that occur. A basic discrepancy is that dissociation, in practice, is generally not a first-order reaction. A more useful approach to compare antibodies is to determine the kinetics of dissociation over a long time period, until near-complete dissociation has occurred.

Advantage of a residualizing iodine radiolabel for radioimmunotherapy of xenografts of human non-small-cell carcinoma of the lung

Attachment of 131I to MAbs through adducts such as dilactitol-tyramine (DLT), which remain lysosomally trapped after catabolism of the labeled MAb, can greatly increase the residence time of the radiolabel at the tumor site. Our previous studies demonstrated a marked increase in accretion of 131I in a human lung-cancer xenograft model, using 131I-DLT in comparison to 131I linked to MAb by the conventional chloramine-T methodology.

METHODS

In this study, biodistribution experiments were performed to evaluate the effect of protein dose on the accretion of 131I-DLT-labeled MAb RS11 in tumor and nontumor tissues, and in vivo radioimmunotherapy experiments compared the effect of single injections of 131I-DLT-labeled MAb RS11 to conventional 131I-labeled RS11 and an untreated control group.

RESULTS

Dosimetry calculations based on the biodistribution data indicate only small changes in absorbed dose-to-tumor and nontumor tissues with increasing protein dose up to 100 micrograms, with a predicted absorbed dose to tumor of from 21,000 to 25,000 cGy/mCi. A single dose of 100 microCi of 131I-DLT-RS11 was found to cause tumor regression. At 7 wk postinjection of the radiolabeled MAbs, tumor volume in 73% of the animals administered 131I-DLT-labeled RS11 remained smaller than at the time of MAb injection. This is compared to 14% of the tumors in the conventionally labeled 131I-RS11 group and none in the untreated group.

CONCLUSION

The use of the residualizing radiolabel DLT provides a therapeutic advantage in comparison to conventional 131I-labeled RS11.

Internalization and catabolism of radiolabelled antibodies to the MHC class-II invariant chain by B-cell lymphomas

The fate of antibody (Ab) LL1, which reacts with the invariant chain (Ii) subunit of the immature MHC class-II antigen (CD74) after binding to the surface of B-cell lymphomas was investigated. This Ab was internalized and catabolized very rapidly, much faster than other Abs that are considered to be rapidly internalized, such as CD19, CD22 and anti-(transferrin receptor). Such internalization did not depend on Ab cross-linking. The capacity of this uptake process was determined in long-term experiments by increasing the Ab concentration: in 1 day, approx. 8 x 10(5) Ab molecules per cell were catabolized. This analysis was facilitated by the use of radiolabels that are trapped within cells after catabolism of the Abs to which they were conjugated. If the Ab is a reliable marker for the Ii antigen, which is likely, we can conclude that Ii directed to the cell surface appears to be sufficient, indeed more than sufficient, to account for the cell content of mature class-II molecules.

Antibody penetration of tumor GS-7 xenografts in nude mice: a model for mucinous adenocarcinoma of the colon

A new cell line derived from a human adenocarcinoma of the colon, GS-7, was propagated as a s.c. tumor in nude mice. This tumor histologically is a mucinous adenocarcinoma (also designated mucoid or colloid) with characteristic large mucin pools that are not lined by an epithelial layer but may contain scattered, randomly distributed cancer cells. Ten to 20% of human colorectal adenocarcinomas are of this histological type, but rapidly growing xenografts with this histology have been rarely used experimentally. This tumor, therefore, constitutes a useful model for similar human tumors. The mucin pools contain large amounts of carcinoembryonic antigen and tumor-associated glycoprotein 72, and the cells express epithelial glycoprotein 2 on their surface. The ability of antibodies injected i.v. to penetrate this tumor was investigated, using both biotinylated and radioiodinated antibodies (Abs). The results demonstrate that Abs can effectively penetrate the mucin pools, and that large amounts of Ab can localize there. This tumor type may have advantages as a target for certain forms of experimental immunotherapy.

Processing of antibodies bound to B-cell lymphomas and other hematological malignancies

In an attempt to recognize general patterns in the processing of Abs (antibodies) bound to the surface of tumor cells, eight monoclonal antibodies were tested on 10 hematological malignancies of various histological types. The results were compared with previous findings obtained with carcinomas, melanomas, and gliomas, using some of the same antibodies. The data demonstrated that some B-cell lymphomas appear to be unusual in that Abs were unable to bind to them irreversibly; except for those Abs that were rapidly internalized, none of the Abs tested was able to bind irreversibly to the B-cell lymphomas Raji or RL. In contrast, most Abs bound irreversibly to the tumors of other histological types, including other hematological tumors. Irreversible Ab binding to B-cell lymphomas was achieved by cross-linking the Abs on the cell surface. Such differences between cell lines may be due to differences in the supramolecular structure of the surface membrane, which affect the frequency or stability of bivalent Ab binding. The Ab binding interaction could not be described in terms of "functional affinity." These results may lead to improvements in the use of Abs for tumor immunotherapy and for other purposes.

Rabbit complement lyses tumor cells without massive C3 deposition

These experiments were performed to determine why rabbit complement lyses tumor cells very efficiently, while not having particularly strong activity in hemolytic assays or in any other complement assay. The target cells used were human tumor cells coated with three different mouse IgG(2a) monoclonal antibodies, and complement from 5 mammalian species were tested. In antibody titration experiments, rabbit complement was found to lyse target cells at a relatively low antibody concentration, insufficient to allow lysis by complement of other species. Since this result was still observed after absorption of rabbit serum with target cells, the potency of rabbit complement cannot be attributed to the presence of natural antibodies. We then assayed C3 deposition on target cells, using two types of (125)I-labeled anti-C3 Abs to measure C3 deposition: goat antibodies specific for C3 of the human, guinea pig, rabbit, rat or mouse, and chicken antibodies to human C3 which cross-react with C3 of other mammals. Unexpectedly, complement of the human, rat, guinea pig, and BUB mouse deposited large amounts of C3 on the surface of target cells, while rabbit complement deposited 100-1,000 fold less. We discuss the possible reasons that C3 deposition does not correlate with cytotoxicity, and may indeed be inversely related. These data indicate that there is a fundamental difference in the complement cascade between rabbits and the other species tested. The potent lytic activity of rabbit complement is likely to be related to this difference, although the mechanism is not yet understood.

On the validity of "functional affinity" determination for antibodies binding to cell surface antigens or other polyvalent antigens

A widely used method of comparing different antibodies (Abs) is to determine their "functional affinities." This value is supposedly a constant that reflects the basic binding interaction between antibody and antigen and, if it is determined under standard conditions, allows the comparison of Abs used by different laboratories. However, I present here both theoretical and experimental evidence that, for Abs binding bivalently, functional affinity determinations seem to be invalid. Experimental data were obtained with erythrocyte targets, to eliminate interference due to the internalization or catabolism of bound Ab, but similar results have been obtained with tumor target cells. Several fundamental discrepancies between the theoretical expectations and the actual values obtained were demonstrated, all of which can be attributed to the effects of bivalent binding. One discrepancy is that the functional affinity determined did not equal the ratio of the rate constants for association and dissociation. A second is that the functional affinity was not a constant but, rather, depended on trivial experimental conditions, such as the volume of incubation. In addition, it has been recognized previously that functional affinity is affected strongly by the structure of a multivalent antigen and, in particular, by any change that makes it more or less likely that an Ab will bind bivalently. The value of other methods of comparing the avidity of different Abs is discussed. In many situations, Ab binding can be considered predominantly irreversible for practical purposes.

Lysine-directed radioiodination of proteins with a cyanuric chloride derivative of aminofluorescein

Protein radioiodination via iodo-5-([4,6-dichlorotria- zin-2-yl]amino)fluorescein(DTAF), a cyanuric chloride derivative of aminofluorescein, was characterized. Commercially available DTAF was iodinated by the Iodogen reaction and then conjugated to IgC antibodies in a 4-h incubation in borate buffer, pH 9.0. With low amounts of protein, 10 micron, molar ratios of iodine:IgC of nearly 1:1 were obtained. With 25 micron protein, which was used routinely, 15-20% efficiency of (125)I incorporation was obtained. Polylysine was labeled efficiently, and labeling was inhibited by ethanolamine, which is consistent with previous data indicating that conjugation is to amino in the protein. A low level of aggregates, primarily dimers, was generated. This procedure is a simple, inexpensive method to test the advantage of lysine-directed protein iodination, with proteins for which oxidative iodination is unsuitable. Abs labeled with DTAF retained strong antigen-binding activity, and, in the case of one Ab, were much more active than the chloramine T-labeled Ab. Catabolic products of iodo-DTAF conjugates, produced after internalization and degradation of Ab bound to the cell surface, were retained within some but not all human carcinoma cell lines.

Effects of radiolabeling monoclonal antibodies with a residualizing iodine radiolabel on the accretion of radioisotope in tumors

The effect of using a "residualizing" iodine radiolabel, dilactitol-iodotyramine, for radioimmunolocalization of antibodies to tumors was investigated. This tracer is designed to be lysosomally trapped after catabolism of the labeled antibody. mAbs RS7 and RS11 were used for in vivo and in vitro studies on the uptake and retention of radioisotope into tumor cells. Both are murine IgG1 mAbs with pancarcinoma reactivity, which react with integral membrane glycoproteins. mAb RS7 has been shown to be relatively rapidly catabolized by the antigen-bearing cell line Calu-3, whereas RS11 is catabolized more slowly in the same cells. An 111In- or 88Y-p-isothiocyanatobenzyl-diethylenetriamine pentaacetic acid conjugate was also tested because these radiometals are known to be lysosomally trapped, and iodination via chloramine T was used to provide a baseline. In vitro, a substantial increase in retention of the label by cells was observed when the dilactitol-tyramine DLT- or 111In-labeled mAbs were used, and the improvement gained by the use of these residualizing labels was greater with the use of the rapidly catabolized mAb (RS7) than it was with the more slowly catabolized mAb (RS11). In biodistribution studies in nude mice bearing Calu-3 tumor xenografts, a dramatic improvement in the tumor accretion of the radiolabel was seen with the use of the 131I-labeled DLT- or 88Y-labeled mAbs. For example, at day 7 the percentage of injected dose/g in the tumor was 5.54 +/- 1.47% (SD), 38.06 +/- 8.04%, and 43.18 +/- 19.50% for the conventionally iodinated, DLT- and 88Y-labeled RS7, respectively. Dosimetry calculations performed on the biodistribution data predict increases of approximately 8- and 4-fold in the absorbed dose to tumor with the use of 131I-labeled DLT- and 90Y-labeled mAbs, respectively, compared to the conventional 131I. In contrast to in vitro findings, these results were similar for both RS7 and RS11, suggesting that the use of DLT may be advantageous for most of the mAbs binding to the cell surface, including antibodies that are catabolized relatively slowly. The advantage of 131I-labeled DLT over 90Y is due to the longer physical half-life of the 131I.

The fate of antibodies and their radiolabels bound to tumor cells in vitro: the effect of cross-linking at the cell surface and of anti-idiotype antibodies

In order to obtain rapid blood clearance of circulating antibodies (Ab) at a desired time, cross-linking reagents such as second Ab are often employed. Such reagents will generally bind to Ab located at the tumor site as well as free Ab, and we therefore investigated whether the cross-linking of Ab bound to the surface of tumor cells affects the processing of those Ab. Cross-linking was induced in various ways: a polyclonal second Ab [rabbit anti-(mouse IgG)], a monoclonal rat anti-(mouse IgG constant region) Ab, and streptavidin used in conjunction with a biotinylated first Ab. Processing was followed for 3 days, to allow nearly all of the bound Ab to reach its ultimate fate. Results depended strongly on the particular first Ab used. Two basic effects were observed. First, the second Ab efficiently prevented the early dissociation of intact Ab from the cell; once the second Ab bound, there was virtually no dissociation of the primary Ab bound to the cells. For most Ab, where only a small proportion of bound Ab dissociated intact, this effect was relatively small. However, for an unusual Ab, where the majority dissociated intact (L6) the effect of a second Ab in prolonging Ab retention by the cell was dramatic. Second, cross-linking sometimes resulted in markedly accelerated internalization and degradation of the bound Ab, coupled with the release of degradation products into the medium. This process resulted in much shorter retention of the radioisotope by the cell. If a "residualizing" radiolabel was used, 125I-dilactitoltyramine, which is probably trapped within lysosomes after Ab catabolism, the effect of the second Ab in accelerating loss from the cell was largely prevented. We also tested anti-idiotype Ab as cross-linking reagents. In addition to testing anti-idiotype Ab known to react with the cell-bound primary Ab, we also tested anti-idiotype Ab not expected to bind to cell-bound Ab, initially as a negative control. Unexpectedly, all anti-idiotype Ab tested induced rapid release of the primary Ab from the cell. This effect was similar to the effect of a large excess of unlabeled Ab, and we attribute it to the blocking of the free binding site of a "wobbling" Ab, which prevents its rebinding to a second antigen molecule. We conclude that the use of selected anti-idiotype Ab to clear circulating Ab, while not reacting with cell-bound Ab, must be done cautiously.(ABSTRACT TRUNCATED AT 400 WORDS)

The processing and fate of antibodies and their radiolabels bound to the surface of tumor cells in vitro: a comparison of nine radiolabels

Processing radiolabeled degradation products is the key factor affecting retention of antibodies within the cell. In this study, we have analyzed the processing of antibodies labeled in nine different ways.

METHODS

Antibodies were labeled with three different radioisotopes and seven different forms of 125I. Eight of the radiolabels (except 188Re) were conjugated to the same antibody, MA103, and tested on the renal carcinoma cell line SK-RC-18 and/or the ovarian carcinoma cell line SK-OV-6. Rhenium conjugation utilized the antibody RS7, the target cell line ME180 and three of the other radiolabels were also tested with this antibody-target cell combination for comparison.

RESULTS

Iodine conjugated to antibodies by conventional methods was rapidly released from the cell after antibody catabolism. In contrast, iodinated moieties, such as dilactitol-tyramine and inulin-tyramine were retained within cells four to five times longer.

CONCLUSIONS

The use of radiolabels that are trapped within cells after antibody catabolism can potentially increase the dose of radiation delivered to the tumor, from the same amount of radioactivity deposited by a factor of four or five. The prolonged retention of 111In relative to 125I is not due to deiodination of iodine conjugates, but rather to intracellular retention of catabolic products containing 111In, perhaps within lysosomes.

Localization of complement-fixing cytotoxic monoclonal antibodies to subcutaneous tumors and liver metastases in a syngeneic immunocompetent rat colon carcinoma model

To study the immune effects of complement-fixing cytotoxic monoclonal antibodies (mAbs) in a syngeneic immunocompetent animal model, mouse mAbs reactive with the transplantable rat colon carcinoma K12/TRb were generated. This system was used in part because rats have a complement system superior to that of mice. Seven murine IgG mAbs that reacted strongly with cell surface determinants of the K12/TRb rat colon carcinoma cell line were produced by immunizing MRL/Mp-1pr/1pr autoimmune mice, known to produce an increased amount of complement-fixing IgG2a and IgG3 immune cytotoxic antibodies, with K12/TRb cells. These mAbs were screened for their specificity of reaction, and two of these mAbs were extensively tested for their ability to lyse cells in vitro and localize to K12/TRb tumors in syngeneic BD IX rats. IgG2a mAbs 27-3 and 61-5 were able to mediate both complement and lymphocyte cytotoxicity in vitro and localize to subcutaneous tumors and liver metastases in this immunocompetent rat model.

Processing of antibody-radioisotope conjugates after binding to the surface of tumor cells

BACKGROUND

Previous experiments indicated that most antibodies binding to cell surface antigens are internalized gradually and degraded within lysosomes, with a half-life of degradation of approximately 1 day, for most antibodies. The research discussed in this article extended our studies to eight additional antibodies reacting with six different antigens, including three antigens anchored in the membrane by glycosyl-phosphatidylinositol. The authors also tested antibodies labeled with 111indium, as well as 125iodine, to determine whether different radiolabels would be processed differently.

METHODS

Antibodies were radiolabeled with 125I or with 111In bound to benzyl-DTPA. After binding to the surface of tumor cells in vitro, excess antibody was washed away, and the fate of the radiolabel was investigated over periods of 3-7 days. Radiolabel released into the supernatant or retained by the cells was analyzed to determine whether it was still on intact antibody.

RESULTS

In 13 of the 15 antibodies that were tested, a similar pattern of irreversible binding and gradual catabolism was observed. Iodine conjugated to antibodies was released rapidly from the cell after antibody catabolism. In contrast, the 111In was retained within the cell much longer than 125I, with the rate of degradation and release into the medium being at least fivefold slower. More than 50% of the bound 111In was still present on the cells after 7 days. Biochemical analysis of the retained 111In extracted cells after 4-6 days demonstrated that it was no longer associated with antibodies and was in a low molecular weight form, probably still associated with the chelator benzyl-DTPA.

CONCLUSIONS

Different radiolabels are processed by tumor cells differently, after catabolism of the antibody to which they originally were conjugated. The data suggest that the prolonged retention of 111In, relative to that of 125I, is due not to deiodination of iodine conjugates, but rather to intracellular retention of catabolic products containing 111In, perhaps within lysosomes. The use of radioisotopes that are retained within cells after antibody internalization and degradation may improve both radioimmunodetection and radioimmunotherapy of cancer.

Internalization of an intact doxorubicin immunoconjugate

An immunoconjugate between doxorubicin and anti-(carcinoembryonic antigen) (CEA) was prepared by using aminodextran (M(r) = 40,000) as the intermediate carrier, and the carbohydrate moiety of the antibody as the linking site. The resulting immunoconjugate was subjected to an in vitro evaluation for the internalization on the target cells (Lo Vo), and compared to that of unconjugated antibody, as well as the cellular uptake of unconjugated doxorubicin. The internalization was evaluated microscopically by following the translocation of the red fluorescence of doxorubicin and the green fluorescence of the fluorescein-isothiocyanate-labeled goat anti-(mouse Ig) antibody, which visualizes the location of the primary mouse antibody. Anti-CEA monoclonal antibody (NP-4) was found to internalize into Lo Vo cells. The immunoconjugate made with this antibody was similarly internalized, and the doxorubicin was found to distribute with the primary antibody. The cell surface and cytoplasm were the major compartments of their distribution. These results indicate that the drug molecules were indeed delivered into the cells by the antibody as an intact conjugate. Unconjugated doxorubicin, on the contrary, was quickly absorbed by the cells and concentrated in the nucleus within 30 min, and never showed a distribution in the cytoplasm or cell membrane as in the nucleus by this procedure. The intermediate drug conjugate, doxorubicin-dextran, did not show internalization. The internalization of NP-4 antibody (or the doxorubicin conjugate) was also confirmed by studying the intracellular catabolism of the cell-bound antibody (or conjugate). The release of the degraded antibody by the cells, as differentiated by trichloroacetic acid precipitation techniques, was considered an indication of internalization. Lysosomes were involved in the degradation, since the process was markedly inhibited in the presence of the lysosomal enzyme inhibitor, ammonium chloride.

SCLC-cluster-2 antibodies detect the pancarcinoma/epithelial glycoprotein EGP-2

Analysis of the antibodies submitted to the 3 International Workshops on Small Cell Lung Cancer Antigens has resulted in the identification of 15 clusters of antibody reactivity. One of these clusters, named SCLC cluster 2, is characterized by reactivity against an epithelium-associated 38 kDa membrane glycoprotein. SCLC cluster 2, and a number of other antibodies with reportedly similar reactivities, were shown to recognize a protein encoded by the GA733-2 gene, whereas the newly defined SCLC cluster 13 antibodies react with the GA733-1 gene product. We propose to call the antigen detected by SCLC-cluster-2 antibodies "epithelial glycoprotein 2" (EGP-2), and the epithelium-associated glycoprotein recognized by antibodies clustered in SCLC cluster 13 (see elsewhere in this volume) "epithelial glycoprotein 1" (EGP-1). A short overview of the characteristics of both proteins and the applications of anti-EGP-2 antibodies is presented.

Characterization of cluster 13: the epithelial/carcinoma antigen recognized by MAb RS7

Cluster 13 was defined by 2 independently derived murine monoclonal antibodies (MAbs), RS7 (IgG1) and MR54 (IgG2a), which were raised against human squamous-cell carcinoma of the lung and a human ovarian-carcinoma cell line, respectively. Immunologic and biochemical evidence demonstrated that RS7 and MR54, as well as 2 additional MAbs, MR6 (IgG2a) and MR23 (IgG1), generated in the same fusion as MR54, recognize the same antigen, a 46- to 48-kDa glycoprotein. Evaluation of the expression of antigen on the surface of tumor cell lines, Western blotting analyses, competitive binding studies, and double-determinant ELISA assays, support this conclusion. Two distinct epitopes are defined by these MAbs. In order to further characterize this antigen, amino-acid-sequence analyses were performed on peptides derived from antigen purified by affinity chromatography with MAb RS7. The sequence data obtained from 2 peptides, which were independently generated by CNBr cleavage and trypsin digestion respectively indicated identity to GA733-1. The GA733-1 genomic DNA sequence predicted a type-1 membrane protein of 35 kDa, with 4 potential N-linked glycosylation sites. The GA733-1 protein product has not been identified previously, and MAbs to this tumor-associated antigen were not previously known.

Biochemical and immunological characterization of the human carcinoma-associated antigen MH 99/KS 1/4

We have characterized the 38-kDa transformation-associated membrane glycoprotein MH 99, whose expression is highly elevated in many epithelial malignancies. A spontaneous cleavage of MH 99 into a 32- and a 6-kDa chain in some carcinoma cell lines was recently shown. Sequence homologies to nidogen, a matrix-adhesion molecule, support the suggestion of a receptor-like function. In this study, we characterized biochemical and immunogenic aspects of MH 99. Transformed epithelial cell lines which do not spontaneously cleave MH 99 were exposed to 8 proteases with distinct specificities. Each of the enzymes produced similar specific fragmentation into chains of about 30 to 32 and 6 kDa, indicating a characteristic cleavage site of MH 99. The fragments were not distinguishable from those in carcinoma cells showing spontaneous cleavage of MH 99. The specific fragmentation depends on the localization in intact membranes and is not shared by other membrane proteins. N-glycosylation of MH 99 of about 4 kDa was exclusively found on the 32-kDa fragment. Characterization of antigenic epitopes was performed using 16 different monoclonal antibodies (MAbs). Only 2 independent determinants were found. One is located on the 32-kDa chain and is recognized only by the MM 104 MAb. The other 15 antibodies bind to a dominant epitope on the 6-kDa fragment which can be divided into 3 overlapping sub-epitopes. The unique features of MH 99 indicate that its immunogenic epitopes are mainly located at its 6-kDa chain, and support the suggestion of a transformation-associated cell-surface receptor which might be proteolytically regulated.

Re-evaluation of the concept of functional affinity as applied to bivalent antibody binding to cell surface antigens

Although it is recognized that the bivalent binding of an antibody to a multivalent antigen cannot be characterized by true affinity, "functional affinity" is frequently determined. "Functional affinity" is calculated by the same methods used for true affinity, and is presumed to have similar biological significance. The data presented herein demonstrate, for IgG antibodies binding to cell surface antigens, that "functional affinity" does not adequately describe the interactions that occur. First, the dissociation rate was not first-order, but rather reflected heterogeneity of bound Ab. A minority of bound Ab (probably monovalently bound) dissociated relatively rapidly, while the majority dissociated very slowly, and, for most purposes, should be considered irreversibly bound. Although irreversible binding is incompatible with a state of equilibrium, the dissociation rate was greatly increased in the presence of free Ab, which may explain why equilibrium is reached in a Scatchard-type experiment. In addition, "functional affinity" was found to vary depending on trivial experimental conditions, such as the volume of incubation. Thus, antibodies do not possess a single "functional affinity", but rather a spectrum of "functional affinities", which are therefore of little value in comparing different antibodies and cannot be used to predict the extent of binding under other experimental conditions. Both of these discrepancies can be attributed to the effect of bivalent Ab binding. We conclude that "functional affinity" determination is inappropriate.

The specificity of alternative complement pathway-mediated lysis of erythrocytes: a survey of complement and target cells from 25 species

Sera from 20 species of mammals were tested for their ability to lyse erythrocytes from 18 species of mammals and birds by the alternative complement pathway. Erythrocytes were not lysed by homologous complement, with one minor exception, but all erythrocytes tested were lysed by at least one complement source, and all sera tested except that of the horse lysed at least one type of erythrocyte. Control experiments indicated that lysis was via the alternative complement pathway and that antibodies were not involved. Complement from the various species could be ranked from most active to least active, and erythrocytes could be ranked from most susceptible to least susceptible. There was an inverse correlation between complement activity and erythrocyte susceptibility. The ranking of the orders of placental mammals, from strongest to weakest complement, was carnivore > artiodactyl (ruminants and swine) > primate = armadillo > rodent > rabbit > horse. Opossum serum had activity that placed it in the centre of this range. Ferret complement, the most potent tested, lysed all erythrocytes tested except for homologous erythrocytes, with APCH50 titres as high as 4000. Although the overall reactivity pattern was clear, there were several striking exceptions. For example, the only complement source which lysed ferret erythrocytes was sera of the mouse. The amount of sialic acid present on erythrocytes of 14 mammals was determined, and was, in general, directly correlated with resistance to alternative complement pathway lysis, although there were prominent exceptions to this correlation, involving erythrocytes of the horse, burro and human. All 20 types of complement were also tested for their ability to lyse antibody-coated human tumour cells, under conditions in which both the classical and alternative complement pathways were functional. The data obtained suggest that alternative pathway activation is, in some cases, a major factor determining the effectiveness of a particular complement source in the lysis of xenogeneic tumour cells.

Analysis of high complement levels in Mus hortulanus and BUB mice

BUB/BnJ mice were previously identified as having exceptionally potent complement activity, relative to common mouse strains, in the lysis of antibody-coated human tumor cells. We describe herein our investigation into the molecular and genetic basis for this difference between mouse strains, and also our results with wild mice and mouse strains recently derived from the wild, to determine whether low complement levels are characteristic of wild mice. BUB complement was compared with complement from BALB/c and C57BL/6 mice. BUB mice had higher levels of most individual classical pathway components, except for C1, than the other two strains, but the difference was generally only 2-3-fold, so insufficient to fully explain the difference observed with tumor target cells. CH50 titers on antibody-coated sheep erythrocytes also demonstrated only a 2-4-fold difference. However, CH50 titers on antibody-coated human erythrocyte target cells demonstrated a difference similar in magnitude to that seen with human tumor targets. These results suggest that the difference between mouse strains depends partly on the use of human, rather than sheep, target cells. In an assay for alternative complement pathway activity using neuraminidase-treated human erythrocytes as targets, complements of BALB/c and BUB mice were similar in activity, suggesting that the difference between mouse strains is manifested in the early steps of complement activation. Analysis of F1 and backcross mice suggested that the difference in complement level between BUB and BALB/c or C57BL/6 mice is controlled by semi-dominant genes, and cannot be attributed to a single gene. Wild mice and mice recently derived from the wild generally had low complement levels, similar to most laboratory mice. However, three strains of aboriginal mice, including Mus hortulanus (spicilegus) and Mus spretus, had complement levels higher than that of BUB mice, and as high as sera from the rabbit or rat, which are the most potent known complement sources for the lysis of human tumor cells. In comparison with BUB mouse sera, M. hortulanus sera had at least four-fold higher levels of C3, C6, C8 and C9, and some or all of these differences may explain its higher total complement activity. In the lysis of antibody-coated human erythrocytes, M. hortulanus serum was more potent than any other complement source tested, including sera of the guinea pig, rat, rabbit or human. These strains may be useful in investigating the role of complement in various pathological processes, and in investigating the genetic regulation of the complement system.

The fate of antibodies bound to the surface of tumor cells in vitro

The fate of monoclonal antibodies binding to the surface of human tumor cells in vitro was investigated. Seven antibodies, labeled with 125I, were tested on four cell lines, which included a melanoma and carcinomas of the ovary, kidney, and lung. The antibodies were selected only by the criterion that they not be rapidly internalized via coated pits, so that they would be representative of most antibodies reacting with cell surface antigens. After allowing binding during a 2-h incubation, unbound antibody was removed, and the release of intact or degraded antibody in the supernatant was monitored. The data demonstrate that most bound antibody was gradually degraded and released from the cell over a 2-3-day period, probably via internalization, while only a small fraction, less than 20% for most antibodies, appeared to dissociate intact. One exceptional antibody, MW207, dissociated largely intact. The release of intact antibody was virtually complete within 4 h, and radioactivity released after this time was predominantly in degraded form. These results demonstrate that antibody binding to the surface of viable cells must in general be considered irreversible, and hence the concept of affinity is not applicable. Since an Fab fragment of one of the antibodies dissociated rapidly, such irreversible binding appears to require bivalent attachment. Another conclusion of this study is that most antibodies binding to the cell surface are gradually internalized, which we suggest is due to the normal turnover of cell surface constituents via non-clathrin-dependent endocytosis. Several experimental approaches indicated that a large fraction of antibody retained by the cells, for at least 2 days after binding, was present at the cell surface.

Rapid blood clearance of immunoglobulin G2a and immunoglobulin G2b in nude mice

An extremely rapid blood clearance rate of murine IgG2a antibodies was found in all strains tested of outbred Swiss nu/nu mice, including mice from the major commercial suppliers. The clearance half-life was less than 5 h, in comparison to a 4-5-day half-life in BALB/c mice. Therefore, most of the IgG2a antibody injected i.v. in such mice is cleared before it can reach interstitial fluid, which interferes with immunotherapy and immunodetection experiments. Individual nude mice varied greatly in their IgG2a clearance rates, which hampered investigation of the phenomena. In our experience, approximately three-fourths of nude mice had a rapid or intermediate clearance rate, whereas the remainder had an approximately normal clearance rate. The clearance rate in nude mice was age-dependent, at least in some instances, in that a rapid clearance rate was observed at 2 months of age, whereas the same mice retested at 4 months of age had a normal clearance rate. Rapid clearance could be inhibited by increasing the dose injected: 100 micrograms/mouse resulted in a normal clearance rate, whereas 30 micrograms/mouse was insufficient to inhibit rapid clearance. The clearance rate of IgG2b antibodies was affected similarly to that of IgG2a, whereas the clearance rate of IgG1 and IgG3 was not affected. The Fc region of IgG2a was required in order for rapid clearance to occur. Biodistribution experiments demonstrated that rapid blood clearance was due, at least partially, to binding to the liver and spleen. To determine the genetic basis for rapid IgG2a clearance, approximately 20 inbred and outbred mouse strains were tested. Unexpectedly, nu/+ as well as nu/nu outbred Swiss mice displayed rapid clearance, whereas control +/+ mice did not, so this phenotype appears to be a dominant effect of the nu mutation. BALB/c nu/nu and nu/+ mice did not display rapid clearance, which may be due to expression of the Igh-1a gene, which codes for the IgG2a present in BALB/c mice and in the monoclonal antibodies used in these studies. In conclusion, this clearance effect must be considered in experiments involving murine IgG2a or IgG2b antibodies in outbred Swiss nude mice, except those in which high antibody doses of greater than 0.1 mg/mouse are used. One method of circumventing this problem is to increase the antibody dose injected; a better but more long-range method is to develop strains of outbred nude mice that do have this characteristic.

Galactose-conjugated antibodies in cancer therapy: properties and principles of action

Galactose conjugation of antibodies causes them to be recognized by the hepatic asialoglycoprotein receptor and therefore cleared very rapidly from the blood. In these investigations, some effector functions of galactose-conjugated antibodies were assayed, and several applications to experimental tumors in vivo were demonstrated. Galactose conjugation did not interfere with two antibody functions in addition to antigen binding, namely complement-mediated cytotoxicity and antibody-dependent cell-mediated cytotoxicity. This conjugation procedure was originally developed for its potential use in localized immunotherapy, such as i.p. Injection of galactose-antibody conjugates i.p. demonstrated, more conclusively than other methods that have been used, that the presence of ascites causes prolonged retention of antibody in the peritoneal cavity and that this effect is correlated with the volume of ascites present. In mice bearing i.p. tumor xenografts, i.p. injection of galactose-antibody conjugates resulted in high tumor/nontumor ratios at 28 h after antibody injection, with values of 40:1, 43:1, 77:1, and 11:1 for the blood, kidney, lung, and spleen, respectively, although the ratio was only 4:1 for the liver. Control experiments demonstrated that i.p. injection of unconjugated antibody or a galactose-conjugated nonreactive antibody produced much lower tumor/nontumor ratios. In investigations of possible systemic application of galactose-antibody conjugates, we found that injection of large amounts of an inhibitor that binds competitively to the hepatic receptor, asialo-bovine submaxillary mucin, can block clearance of galactose-conjugated antibodies for 2-3 days. In this way, high blood levels of antibody can be maintained for 2-3 days, thus allowing penetration and binding to solid tumors, followed by very rapid blood clearance. With this approach, using a human carcinoma growing s.c. in nude mice, high tumor/nontumor ratios were obtained 4 days after injection, with mean values of 43:1, 18:1, 17:1, and 15:1 for the blood, kidney, lung, and spleen, respectively, although the ratio for the liver was only 1.7:1. The blood level at this time was 0.04 +/- 0.02% (SD) of the injected dose/g, while the tumor level was 1.69 +/- 1.29% of the injected dose/g. In conclusion, galactose-conjugated antibodies appear to have diverse applications in regional or systemic immunotherapy.

Normal tissue reactivity of four anti-tumor monoclonal antibodies of clinical interest

Normal tissue reactivity on frozen sections was examined with 4 monoclonal antibodies (MAbs) that were reported previously to be negative or weakly reactive with normal tissues, and strongly reactive with some types of carcinoma. All 4 antibodies reacted strongly with certain normal epithelial cells. The antibodies tested include 2 antibodies to ovarian cancer, MOv18 and MOv19, one antibody to breast cancer, H23, and one antibody reactive with a range of carcinomas, B72.3. MOv18, MOv19 and H23 reacted with many normal glandular and ductal epithelial cells, while B72.3 reacted most strongly with secretions of the small intestine epithelium and the suprabasal squamous epithelial cells of the esophagus. Since the tissue distribution of MOv19 was very similar to that of another antibody described previously, MW207, these two antibodies were compared by competitive binding inhibition, and found to recognize the same epitope. Our data emphasize the importance of repeated, independent tests of antibody specificity.

Breast tumor radioimmunodetection with a 111In-labeled monoclonal antibody (MA5) against a mucin-like antigen

Monoclonal antibody MA5 recognizes a determinant displayed on high molecular weight antigens associated with secretory and malignant breast epithelial cells. MA5 reactivity with greater than 95% of primary and metastatic breast tumors, surface expression of the antigen, as well as its ability to localize within breast tumor xenografts prompted this initial study to determine the efficacy of MA5 to localize breast tumors by radioimmunoscanning. A total of 17 patients was monitored, each receiving 2 mg of purified MA5 labeled with 5 mCi of 111In. Some patients also received 3 or 18 mg of unlabeled carrier antibody (MA5); no serious allergic reactions were noted. Primary tumors, bone lesions, soft tissue recurrences, and lung metastases greater than 3 cm in diameter were detectable, whereas only one lesion (hilar node) less than 3 cm was localized. Significant antibody accumulation was noted in the liver and less significant uptake in the spleen and bone. The extensive fibrosis and poor vascularization of breast tumors may partly explain the limited sensitivity obtained thus far. The imaging results obtained with MA5 are compared with other antibodies which we show recognize the same antigens.

Patterns of antigen distribution in human carcinomas

Ten epithelial-specific monoclonal antibodies, including monoclonal antibodies to antigens that have been used extensively in immunodiagnosis and immunotherapy experiments, were tested for reactivity with 20 human carcinomas each of the colon, lung, and breast. The antibodies tested included B72.3, OC125, and antibodies to carcinoembryonic antigen, the 17-1A antigen, and the milk fat globule mucin antigen (epithelial membrane antigen). Striking differences in the pattern of antigen distribution were seen, with each antibody having a fairly consistent staining pattern, which was dependent on the tumor type. Two antibodies reacted with most or all tumor specimens and, when positive, reacted homogeneously with apparently every cell in the specimen. Other antibodies consistently produced a variegated staining pattern, typically with areas of positive cells surrounded by areas of negative tumor cells. A third pattern was strong localization to the luminal edge and/or secretions of glandular tumors; this pattern was seen primarily in colon carcinomas which have more well-developed glandular structures than breast or lung carcinomas. A correlation with biochemical properties of the antigens was evident, in that mucins or mucin-related antigens generally produced variegated staining of lung and breast carcinomas and luminal edge/secretion staining of colon carcinomas. Such differences in antigen distribution are likely to be a major factor in developing methods for immunodiagnosis and immunotherapy.