40
Bioconjugate Chem. 2002, 13, 40−46
An Anti-CD33 Antibody-Calicheamicin Conjugate for Treatment of Acute Myeloid Leukemia. Choice of Linker Philip R. Hamann,*,† Lois M. Hinman,† Carl F. Beyer,† Delores Lindh,† Janis Upeslacis,† David A. Flowers,‡ and Irwin Bernstein‡ Wyeth-Ayerst Research, 401 North Middletown Road, Pearl River, New York 10965 and Fred Hutchinson Cancer Research Center, 1124 Columbia Street, Seattle, Washington 98109. Received February 19, 2001; Revised Manuscript Received October 15, 2001
The anti-CD33 antibody, P67.6, has been chosen to target the potently cytotoxic calicheamicin antitumor antibiotics to acute myeloid leukemia (AML) due to the presence of CD33 on >80% of patient samples and its lack of expression outside the myeloid cell lineages, especially its lack of expression on pluripotent stem cells. Previous calicheamicin conjugates relied on the attachment of a hydrazide derivative to the oxidized carbohydrates that occur naturally on antibodies. This results in a “carbohydrate conjugate” capable of releasing active drug by hydrolysis of a hydrazone bond in the lysozomes where the pH is low. Conjugates have now been made that are formed by reacting a calicheamicin derivative containing an activated ester with the lysines of antibodies. This results in an “amide conjugate” that is stable to hydrolysis, leaving the disulfide that is present in all calicheamicin conjugates as the likely site of drug release from the conjugate. In this article, these two classes of calicheamicin-antibody conjugates are compared for potential use in AML with the anti-CD33 antibody P67.6. Conjugates of P67.6 are shown to require the site of hydrolytic release afforded by the carbohydrate conjugates in order to retain good potency and selectivity in vitro, in vivo, and ex vivo. The P67.6 carbohydrate conjugate of calicheamicin is selectively cytotoxic at 100d
8 1.4 2.43 1.71 -
a Relative affinity in a competitive assay versus radioiodinated P67.6. b Doses are given in cal equiv. The corresponding amount of antibody is on average 40-fold greater. c Ratio of the two IC50’s. d Concentration of antibody alone equivalent to that of either conjugate at the dose of cal equiv given in table. Actual concentration of antibody 4 µg/mL.
Figure 3. Time-course of the response of HL-60 xenograft tumors to amide and carbohydrate conjugates of P67.6. The doses are given in cal equiv. The corresponding amount of antibody is 40-fold greater. This experiment is representative of the general results seen with these conjugates.
negative Raji cells, and the IC50 values are given in cal equiv. The P67.6 carbohydrate conjugate was more than 7000-fold more potent than the P67.6 amide conjugate toward the antigen-positive cells. The specificity index of the P67.6 carbohydrate conjugate was greater that 100fold, while the P67.6 amide conjugate showed modest specificity for antigen-positive cells. Both types of conjugates retained good immunoaffinities. Also presented in Table 1 are results for several controls, which showed that P67.6 alone was not toxic to either cell type, that the admixture of P67.6 and NAc-gamma calicheamicin DMH was not different than NAc-gamma calicheamicin DMH (3) alone, and that the calicheamicin derivatives were less potent than the carbohydrate conjugate toward the HL-60 target cells. Comparative in Vivo Activity. Carbohydrate and amide conjugates of P67.6 were further compared in vivo for antitumor effects against HL-60 tumor xenografts implanted subcutaneously in nude mice. The results of one such test are presented in Figure 3. The two conjugates and NAc-gamma calicheamicin (2) were administered three times at the doses indicated spaced by 4 days starting on day seven post tumor implantation. In this test, no deaths were noted in the conjugate-treated groups during the 42-day duration of the study, while in the group treated with NAc-gamma calicheamicin (2), one
44 Bioconjugate Chem., Vol. 13, No. 1, 2002
Hamann et al.
Table 2. Comparison of the Ex Vivo Inhibition of Colony Formation. Single Experiment amide carbohydrate conjugate (7) conjugate (4) media drug control (5 and 3, respectively) CTM01 conjugate (control) P67.6 conjugate
14.5 ( 2.0 10.3 ( 3.2 10.5 ( 2.5 10.0 ( 4.4
14.5 ( 2.0 15.3 ( 3.6 14.5 ( 1.9 3.7a ( 1.4
a Significantly different from the P67.6 amide conjugate as well as from all other samples in the carbohydrate conjugate column at p < 0.02.
Table 3. Comparison of the Ex Vivo Inhibition of Colony Formation. Composite Results sample P67.6 amide conjugate (7) P67.6 carbohydrate conjugate (4)
no. with specific inhibition no. of patients 25-59% 60-89% g90% 12 12
4 1
2 4
0 4
of five animals had died at the 75 µg/kg dose and all were dead at the 150 µg/kg dose (data not shown). This experiment showed that the P67.6 carbohydrate conjugate was significantly more active than the P67.6 amide conjugate at all three doses tested. The P67.6 carbohydrate conjugate produced nine tumor-free survivors out of 10 treated animals in the 300 and 150 µg/kg × 3 treatment groups, whereas no tumor-free animals were seen with the P67.6 amide conjugate at any dose. Comparative ex Vivo Activity. Testing of the P67.6 amide and carbohydrate conjugates was carried out on pediatric AML patient samples. In one experiment, eight pediatric AML samples were tested with these two conjugates as well as control conjugates made with the nonbinding antibody CTM01 and also appropriate unconjugated drug controls. The data, shown in Table 2, indicate that at 2 ng/mL cal equiv only the targeting antibody conjugate made with a hydrolyzable linker is capable of inhibiting colony formation in the majority of samples, as indicated by the averages. Although there were individual patient samples that were inhibited by the amide conjugate, the average of all eight patients showed no statistical difference from media alone. In a total of three experiments, run with various combinations of the available samples, 12 different pediatric AML samples were tested with both the P67.6 amide and carbohydrate conjugates. The results, summarized in Table 3, show that at 2 ng/mL cal equiv there was some growth inhibition of AML colony-forming cells using the amide conjugate, while substantially more growth inhibition was seen using the carbohydrate conjugate. DISCUSSION
Antibody-targeted chemotherapy has been studied for many years by numerous academic and pharmaceutical research groups. Aside from the requirement for an antibody that targets a pertinent, internalizing antigen, there appears to be two major criteria that need to be satisfied for activity. The first of these is that the antigen must be capable of carrying enough of the drug into the cell to deliver a toxic dose. Three factors come into play here: the antigenic density and turnover rate, the molar loading of drug on the antibody, and the inherent potency of the drug. The antigenic density and turnover rate is difficult to alter in patients, while the drug loading is difficult to increase significantly beyond a certain level. The potency of the drug, however, can vary by many orders of magnitude, depending on the choice an investigator makes (5, 18). The calicheamicins are one of the
most potent classes of natural products, and this allows calicheamicin conjugates to target a relatively broad range of antigens, including those with relatively low expression, such as CD33 (19). The second requirement for antibody-targeted chemotherapy is that a mechanism must be incorporated into the conjugate design to allow for drug release after internalization into the target cells. Two of the most common sites of drug release are disulfide linkages and hydrazones. The former allows for reductive release of the drug, presumably through interaction with glutathione, while the latter allows for hydrolytic release in the acidic lysozomes where many antibody/antigen complexes end up after internalization. Conjugates of calicheamicin, by the nature of their structure and mechanism of DNA damage, always have a disulfide present that can serve as a site of release of the calicheamicin from the antibody. Indeed, increasing the stability of this disulfide leads to a marked improvement in the therapeutic index of the conjugates (8). However, the desire to have site-selective conjugation led to the initial use of the procedure developed by Cytogen that involves the oxidation of the naturally occurring carbohydrates in the hinge region of the antibody and reacting the resultant aldehydes with nucleophiles, usually hydrazides. The carbohydrate conjugates that are formed are capable of releasing active drug by hydrolysis in the lysozomes where the pH is low (16, 17). A new class of calicheamicin conjugate has now been made that is based on the reaction of an activated ester derivative with the lysines on the antibody. The resultant amide conjugate does not contain a hydrazone bond and is therefore stable to hydrolysis under physiological conditions, but still contains a disulfide bond. These conjugates can therefore be used to answer a basic question about the requirements for efficient release of a cytotoxic agent such as the calicheamicins from various antibodies. The results in this article show that for the anti-CD33 antibody P67.6, the amide conjugates do not exhibit potent, selective cytotoxicity while the carbohydrate conjugates do. Thus, the disulfide bond does not appear to be an efficient site of release with this antibody. Although carbohydrate conjugates were initially made with calicheamicin because the use of the carbohydrates precludes reactions near the antigen-binding site, the amide conjugate of P67.6 retains virtually all of its binding affinity when an amide conjugate is produced. Indeed, only one out of almost two dozen antibodies has lost any significant immunoaffinity when a calicheamicin amide conjugate was made (data not shown). In vitro, the P67.6 amide conjugate is about 100-fold less potent than the calicheamicin reference drugs. This is indicative of unproductive metabolism due to insufficient release of an active calicheamicin derivative intracellularly. On the other hand, the P67.6 carbohydrate conjugate is at least 100-fold more cytotoxic than the calicheamicin reference compounds. This perhaps surprising result may be due to the efficient internalization, release, and utilization of the calicheamicin once carried into the cell by the antibody. The significant cytotoxicity of the P67.6 carbohydrate conjugate against the CD33-negative Raji cells during long-term exposure is presumed to be due to the result of the slow extracellular release and subsequent uptake of the NAc-gamma calicheamicin DMH (6) from the conjugate. In vivo, the P67.6 amide conjugate does show significant inhibition of tumor growth of the HL-60 tumor in nude mice but was incapable of producing any long-term, tumor-free survivors, even at the highest doses tested.
Anti-CD33 Calicheamicin Conjugate: Linker Choice
In contrast, the P67.6 carbohydrate conjugate was capable of eradicating the tumors routinely at the two highest doses tested. Similar conjugates of other antibodies which do not bind to HL-60 cells show no significant effects on HL-60 xenografts, and conjugates of P67.6 do not show significant effect on xenograft tumors that do not express the CD33 antigen (data not shown). The unconjugated calicheamicin derivatives that have been tested as controls in this and similar experiments show at best modest antitumor effects at the MTD. Ex vivo experiments were done to examine the response of a number of AML biopsies, thereby giving reassurance that the results with the HL-60 cells were not dependent on some unusual property of that cell line. A comparison of the activity of the two P67.6 conjugates in the ex vivo experiments clearly shows that the carbohydrate conjugate is superior and is capable of inhibiting colony formation in a significant percentage of patient samples. It should be noted that 100% inhibition is not necessarily desirable in this assay, as the formation of normal colonies from CD33-negative precursors is possible. The absolute predictive value of this assay is questionable, because only about 25% of AML patient samples grow colonies under these conditions. However, the results clearly show a difference for the two classes of conjugate and also indicate the possibility of selective cytotoxicity toward leukemic cells in at least a subset of patients. The consistent results from these three test systems, in vitro, in vivo, and ex vivo, clearly show the need for a site of hydrolytic release in the linker between the calicheamicin and the P67.6 antibody even though a disulfide is present. Preliminary experiments indicated that reduction of the hydrazone bond in the carbohydrate conjugate of P67.6 with cyanoborohydride led to a conjugate that resembled the amide conjugate in its properties (data not shown). This would tend to preclude any special effect of attaching the calicheamicin to a carbohydrate reside being responsible for the observed activity. It is unclear at this point why a site of hydrolytic release is necessary. HL-60 cells are relatively small and rich in lysozomes. Processed fragments of antibody appear in the cell supernatant quickly after treatment with radiolabeled antibody (20). Perhaps the calicheamicin is externalized with such fragments when only the disulfide bond is available for calicheamicin release, but when a hydrazone bond is present, the calicheamicin is released from the antibody before these fragments leave the cell. This is consistent with the known internalization of P67.6 into lysozomes (20) and the acid nature of this subcellular compartment (16, 17). Initial experiments with tritium-labeled calicheamicin derivatives to explore this issue have been unsuccessful because of the unavailability of tritiated reagents with high enough specific activity to follow the low levels of the very potent calicheamicin derivatives in cells. It should be noted that although both disulfides and hydrazones are viewed as acceptable sites for cleaving chemoimmunoconjugates to release an active drug, there is no prior publication that examines the comparison of a drug conjugate made with a disulfide with that made with a hydrazone. For calicheamicin conjugates the results with P67.6 have been shown not to be general. Indeed, for the antiMUC1 antibody, CTM01 (21), a hydrolyzable bond had no benefit, and the amide conjugate was equivalent or superior in all preclinical models that were examined,
Bioconjugate Chem., Vol. 13, No. 1, 2002 45
especially against resistant cell lines.2,3 The combined results for these two antibodies indicate that one optimal design of conjugate does not exist for all antibodies. Each antibody must be examined separately, and a variety of different types of conjugates must be tried in order to individually optimize a specific delivery system. The contrasting results obtained with the anti-P67.6 and antiMUC1 antibody conjugates probably depend more on the physiology of the target cells and the antigen that is targeted than on any specific properties of the antibodies or cytotoxic agent. In conclusion, a calicheamicin conjugate of the antiCD33 antibody is a potent and selective agent targeted toward CD33-positive cells in cell culture, in xenograft experiments, and in ex vivo models of AML, as long as a site of hydrolytic release is present. Thus, the P67.6 carbohydrate conjugate of calicheamicin was envisioned to be an effective new agent for the treatment of AML and other CD33-positive diseases. The P67.6 antibody has been humanized and, as the result of an unexpected problem, the carbohydrate conjugate has been abandoned and replaced by a conjugate containing a bifunctional linker that attaches to lysines on the antibody, but that allows for formation of the hydrazone that is needed for activity (22). This improved conjugate, referred to as CMA-676 and assigned the generic name of gemtuzumab ozogamicin, has successfully completed initial clinical trials (23, 24) and has been approved under the name Mylotarg by the FDA as the first antibody-targeted chemotherapeutic agent. ACKNOWLEDGMENT
We would like to thank Irwin Hollander for helpful discussion, technical help, and editorial assistance, George Ellestad for helpful scientific discussions, and Martha Godwin and J. Philip James for technical support. LITERATURE CITED (1) Lee, M. D., Dunne, T. M., Chang, C. C., Morton, G. O., and Borders, D. B. (1987) Calicheamicins, a novel family of antitumor antibiotics, 1: Chemistry and partial characterization of γ1I. J. Am. Chem. Soc. 109, 3464-3466. (2) Myers, A. G., Cohen, S. B., and Kwon, B. M. (1994) A study of the reaction of Calicheamicin γ1Ι with glutathione in the presence of double-stranded DNA. J. Am. Chem. Soc. 116, 1255-1271. (3) Ellestad, G. A., Ding, W.-D., Zein, N., and Townsend, C. A. (1995) DNA-cleaving properties of calicheamicin γ1I. Enediyne Antibiotics as Antitumor Agents (D. B. Borders, and T. W. Doyle, Eds.) pp 137-160, Marcel Dekker, New York. (4) Lee, M. D., Ellestad, G. A., and Borders, D. B. (1991) Calicheamicins: discovery, structure, chemistry, and interaction with DNA. Acc. Chem. Res. 24, 235-43. (5) Liu, C., and Chri, R. V. J. (1997) The development of antibody delivery systems to target cancer with highly potent maytansinoids. Exp. Opin. Invest. Drugs 6, 169-172. (6) Chari, R. V. J., Jackel, K. A., Bourret, L. A., Derr, S. M., Tadayoni, B. M., Mattocks, K. M., Shah, S. A., Liu, C., Blaettler, W. A., and Goldmacher, V. S. (1995) Enhancement of the selectivity and antitumor activity of a CC-1065 analogue through immunoconjugate formation. Cancer Res. 55, 4079-4084. (7) Hinman, L. M., Hamann, P. R., Wallace, R., Menendez, A. T., Durr, F. E., and Upeslacis, J. (1993) Preparation and characterization of monoclonal antibody conjugates of the calicheamicins: a novel and potent family of antitumor antibiotics. Cancer Res. 53, 3336-3342. 3 Hamann, P. R., Hinman, L. M., Beyer, C. F., Lindh, D., and Mountain, A. A calicheamicin conjugate with a fully humanized anti-MUC1 antibody shows potent antitumor effects in breast and ovarian tumor xenografts. To be submitted for publication.
46 Bioconjugate Chem., Vol. 13, No. 1, 2002 (8) Hinman, L. M., Hamann, P. R., and Upeslacis, J. (1995) Preparation of conjugates to monoclonal antibodies. Enediyne Antibiotics as Antitumor Agents (D. B. Borders, and T. W. Doyle, Eds.) pp 87-106, Marcel Dekker, New York. (9) Andrews, R. G., Singer, J. W., and Bernstein, I. D. (1989) Precursors of colony-forming cells in humans can be distinguished from colony-forming cells by expression of the CD33 and CD34 antigens and light scatter properties. J. Exp. Medicine 169, 1721-31. (10) Keating, M. J., Estey E., and Kantarjian H. (1993) Acute leukemia. Cancer, Principles and Practice of Oncology (V. T. DeVita, Jr., S. Hellman, and S. A. Rosenberg, Eds.) pp 193864, JB Lippincott Company, Philadelphia. (11) Appelbaum, F. R., Matthews, D. C., Eary, J. F., Badger, C. C., Kellogg, M., Press, O. W., Martin, P. J., Fisher, D. R., Nelp, W. B., Thomas, E. D., and Bernstein, I. D. (1992) The use of radiolabeled anti-CD33 antibody to augment marrow irradiation prior to marrow transplantation for acute myelogenous leukemia. Transplantation 54, 829-833. (12) van der Jagt, R. H. C., Badger, C. C., Appelbaum, F. R., Press, O. W., Matthews, D. C., Eary, J. F., Krohn, K. A., and Bernstein, I. D. (1992) Localization of radiolabeled antimyeloid antibodies in a human acute leukemia xenograft tumor model. Cancer Res. 52, 89-94. (13) Bernstein, I. D., Singer, J. W., Andrews, R. G., Keating, A., Powell, J. S., Bjornson, B. H., Cuttner, J., Najfeld, V., Reaman, G., Raskind, W., Sutton, D. M., and Fialkow, P. J. (1987) Treatment of acute myeloid leukemia cells in vitro with a monoclonal antibody recognizing a myeloid differentiation antigen allows normal progenitor cells to be expressed. J. Clin. Invest. 79, 1153-59. (14) Melchers, F. (1970) Biosynthesis of the carbohydrate portion of immunoglobulins. Biochem. J. 119, 765-772. (15) Rodwell, J. D., Alvarez, V. L., Lee, C., Lopes, A. D., Goers, J. W. F., King, H. D., Powsner, H. J., and McKearn, T. J. (1986) Site-specific covalent modification of monoclonal antibodies: in vitro and in vivo evaluations. Proc. Natl. Acad. Sci. U.S.A. 83, 2632-6. (16) Van Dyke, R. W., Root, K. V., Schreiber, J. H., and Wilson, J. M. (1992) Role of CFTR in lysosomal acidification. Biochem. Biophys. Res. Commun. 184, 300-5. (17) Ohkuma, S., and Poole, B. (1978) Fluorescence probe measurement of the intralysosomal pH in living cells and the perturbation of pH by various agents. Proc. Natl. Acad. Sci. U.S.A. 75, 3327-31.
Hamann et al. (18) Trail, P. A., Willner, D., Lasch, S. J., Henderson, A. J., Hofstead, S., Casazza, A. M., Firestone, R. A., Hellstro¨m, I., and Hellstro¨m, K. E. (1993) Cure of xenografted human carcinoma by Br96-doxorubicin immunoconjugates. Science 261, 212-215. (19) Xu, Y., and Scheinberg, D. A. (1995) Elimination of human leukemia by monoclonal antibodies in an athymic nude mouse leukemia model. Clin. Cancer Res. 1, 1179-87. (20) Press, O. W., Shan, D., Howell-Clark, J., Eary, J., Appelbaum, F. R., Matthews, D., King, D. K., Haines, A. M. R., Hamann, P., Hinman, L., Shochat, D., and Bernstein, I. D. (1996) Comparative metabolism and retention of iodine-125, yttrium-90, and indium-111 radioimmunoconjugates by cancer cells. Cancer Res. 56, 2123-9. (21) Aboud-Pirak, E., Sergent, T., Otte-Slachmuylder, C., Abarca, J., Trouet, A., and Schneider, Y.-J. (1988) Binding and endocytosis of a monoclonal antibody to a high molecular weight human milk fat globule membrane-associated antigen by cultured MCF-7 breast cancer cells. Cancer Res. 48, 31883196. (22) Hamann, P. R., Hinman, L. M., Hollander, I., Beyer, C. F., Lindh, D., Holcomb, R., Hallett, W., Upeslacis, J., Shochat, D., Mountain, A., Flowers, D. A., and Bernstein, I. (2002) Gemtuzumab ozogamicin, a potent and selective anti-CD33 antibody-calicheamicin conjugate for treatment of acute myeloid leukemia. Bioconjugate Chem. 13, 47-58. (23) Sievers, E. L., Appelbaum, F. R., Spielberger, R. T., Forman, S. J., Flowers, D., Smith, F. O., Shannon-Dorcy, K., Merger, M. S., and Bernstein, I. D. (1999) Selective ablation of acute myeloid leukemia using antibody-targeted chemotherapy: A phase I study of an anti-CD33 calicheamicin immunoconjugate. Blood 93, 3678-84. (24) Sievers, E. L., Larson, R. A., Estey, E., Stadtmauer, E. A, Roy, D.-C. C., Spielberger, R., Tarantolo, S., Berger, M. S., Eten, C., Manley, L., Bernstein, and Appelbaum, I. F. (1999) Preliminary results of the efficacy and safety of CMA-676 in Patients with AML in first relapse. Proceedings of The American Society of Clinical Oncology (ASCO) 18, 7a, abstr. 21, Lippencott, Williams, and Wilkins, Baltimore, MD.
BC0100206
