Oncostatin M in the anti-inflammatory response.

Oncostatin M (OM) is a pleiotropic cytokine of the interleukin 6 family, whose in vivo properties and physiological function remain in dispute and poorly defined. These in vivo studies strongly suggest that OM is anabolic, promoting wound healing and bone formation, and anti-inflammatory. In models of inflammation OM is produced late in the cytokine response and protects from lipopolysaccharide (LPS)-induced toxicities, promoting the re-establishment of homoeostasis by cooperating with proinflammatory cytokines and acute phase molecules to alter and attenuate the inflammatory response. Administration of OM inhibited bacterial LPS-induced production of tumour necrosis factor alpha and septic lethality in a dose dependent manner. Consistent with these findings, in animal models of chronic inflammatory disease OM potently suppressed inflammation and tissue destruction in murine models of rheumatoid arthritis and multiple sclerosis. T cell function and antibody production were not impaired by OM treatment. Taken together, these data indicate that the activities of this cytokine in vivo are anti-inflammatory without concordant immunosuppression.

CD34+ selected bone marrow grafts are radioprotective and establish mixed chimerism in dogs given high dose total body irradiation.

BACKGROUND: Canine stem cell transplantation models have provided important preclinical information for human clinical studies. The recent cloning of cDNA for canine CD34 and the production of monoclonal antibodies that recognize canine CD34 have been the basis for the development of techniques for the large-scale enrichment of canine hematopoietic progenitor cells. In this study, we evaluated the in vivo functional properties of canine bone marrow CD34+ cells after a myeloablative conditioning regimen. METHODS: After 920 cGy total body irradiation, three dogs received infusion of autologous CD34+ selected cells from the marrow, three dogs CD34+ depleted autologous marrow cells, and two dogs received CD34+ autologous marrow cells that were immunomagnetically selected and then further purified by cell sorting. In addition, four dogs received allogeneic marrow enriched for CD34+ cells from dog leukocyte antigen-identical littermates to investigate long-term repopulating function of CD34+ cells. Chimerism studies were performed using polymerase chain reaction to detect highly polymorphic microsatellite markers. RESULTS: In three recipients of autologous marrow enriched for CD34+ cells to between 29% and 70% (1.6 x 10(6) to 3.4x10(6) CD34+ cells/kg), prompt and full hematopoietic recovery occurred, whereas in three dogs that received marrow depleted of CD34+ cells (1 x 10(7) cells/kg), no hematopoietic recovery was achieved. In two dogs that received highly purified CD34+ cells (purity: 98% and 96%, 0.79x10(6) to 0.547x 10(6) CD34+ cells/kg), delayed but full hematopoietic recovery was seen. Three of four allograft recipients of 1.75x10(6) to 6.8x10(6) CD34+ cells/kg engrafted and showed full hematopoietic recovery, whereas one dog rejected the graft. The three long-term survivors showed stable mixed hematopoietic chimerism with predominantly donor hematopoiesis. CONCLUSION: Transplantation of canine CD34+ cells after lethal total body irradiation provides radioprotection and gives rise to long-term hematopoietic reconstitution. Stable donor/host mixed chimerism was observed in allograft recipients most likely as a result of T-cell depletion of the grafts. Our findings suggest a future role for canine preclinical transplant studies involving in vitro manipulation of hematopoietic pro.

Regulation of inflammatory responses by oncostatin M.

Oncostatin M (OM) is a pleiotropic cytokine produced late in the activation cycle of T cells and macrophages. In vitro it shares properties with related proteins of the IL-6 family of cytokines; however, its in vivo properties and physiological function are as yet ill defined. We show that administration of OM inhibited bacterial LPS-induced production of TNF-alpha and lethality in a dose-dependent manner. Consistent with these findings, OM potently suppressed inflammation and tissue destruction in murine models of rheumatoid arthritis and multiple sclerosis. T cell function and Ab production were not impaired by OM treatment. Taken together these data indicate the activities of this cytokine in vivo are antiinflammatory without concordant immunosuppression.

Characterization of monoclonal antibodies that recognize canine CD34.

Using a polyclonal antiserum against canine CD34, we previously found that CD34 is expressed on canine bone marrow progenitor cells in a manner analogous to that found in humans. To further characterize CD34+ cells and to facilitate preclinical canine stem cell transplant studies, monoclonal antibodies (MoAbs) were raised to CD34. A panel of 10 MoAbs was generated that reacted with recombinant CD34 and with CD34+ cell lines and failed to react with CD34- cell lines. Binding properties of five purified MoAbs were determined by BIAcore analysis and flow cytometric staining, and several MoAbs showed high affinity for CD34. Two antibodies, 1H6 and 2E9, were further characterized, and in flow cytometry studies typically 1% to 3% of stained bone marrow cells were CD34+. Purified CD34+ bone marrow cells were 1.8- to 55-fold enriched for colony-forming unit-granulocyte-macrophage and for long-term culture initiating cells as compared with bone marrow mononuclear cells, whereas CD34- cells were depleted of progenitors. Three autologous transplants were performed with CD34+ cell fractions enriched by immunomagnetic separation. After marrow ablative total body irradiation (920 cGy), prompt hematopoietic recovery was seen with transplanted cell doses of

Liver LDL receptor mRNA expression is decreased in human ApoB/CETP double transgenic mice and is regulated by diet as well as the cytokine oncostatin M.

We have investigated liver LDL receptor mRNA expression in nontransgenic, human cholesteryl ester transfer protein (CETP) transgenic, and human apolipoprotein (Apo) B/CETP double transgenic mice fed a normal chow diet and a high fat, high cholesterol diet (HFHC). Three weeks of HFHC feeding increased total serum cholesterol 1.5-fold in the nontransgenic, 3.1-fold in the CETP transgenic, and 3.4-fold in the ApoB/CETP double transgenic mice. To examine the liver LDL receptor mRNA expression among the different groups of mice fed the normal diet or fed the HFHC diet, we developed a quantitative reverse-transcribed polymerase chain reaction assay in which the LDL receptor mRNA level was normalized with the beta-actin mRNA. The results show that on the normal chow diet, the LDL receptor mRNA expression levels were lower in the ApoB/CETP mice than in the nontransgenic mice and the human CETP transgenic mice. Liver LDL receptor gene expression was lower in all groups of mice fed the HFHC diet, with the lowest level of expression in the ApoB/CETP mice. Similar results were obtained by Northern blot analysis. In addition, we have previously shown that the cytokine oncostatin M (OM) increases LDL receptor gene expression in HepG2 cells. In this study, we used the ApoB/CETP mice as the model system to examine the in vivo activity of OM on liver LDL receptor gene expression. Our data show that OM increased the level of liver LDL receptor mRNA up to 80% to 90% when the animals were fed the HFHC diet. The results from these studies demonstrate that the expression of the liver LDL receptor in the ApoB/CETP mice is suppressed compared with nontransgenic mice and that the expression of the hepatic LDL receptor gene in these mice is subjected to the normal cholesterol feedback regulation. In addition, LDL receptor gene expression in these mice is also inducible by a positive regulator.

Oncostatin M-specific receptor mediates inhibition of breast cancer cell growth and down-regulation of the c-myc proto-oncogene.

Human oncostatin M (OM) is a M(r) 28,000 glycoprotein that has been shown to regulate cell proliferation and differentiation. The biological activities of OM can be mediated by two different heterodimeric receptor complexes, the leukemia inhibitory factor (LIF)/OM shared receptor and the OM-specific receptor. In this study, we have examined the growth-regulatory effect of OM on 10 breast cancer cell lines derived from human tumors. The cellular proliferation of seven of these breast cancer cell lines was inhibited by OM. The three cell lines that did not respond to OM treatment lacked the expression of OM receptors. The growth-inhibitory activity of OM is examined further in the H3922 breast cancer cell line, which expresses the high-affinity OM receptor at a relatively higher level. We found that the cellular proliferation of H3922 cells was induced strongly by extrogenous epidermal growth factor (EGF), EGF-like factor, and basic fibroblast growth factor. The proliferative activities of these growth factors can be abolished totally by cotreatment of H3922 cells with OM. Treatment of H3922 cells with OM for 24 h did not block EGF binding or the induction of EGF receptor tyrosine phosphorylation. This finding suggests that OM interferes with the mitogenic signal at steps distal to the EGF receptor. Examination of proto-oncogene expression demonstrated that OM down-regulates the c-myc gene in H3922 cells. The biological effects reported herein are not shared by the OM-related cytokines interleukin 6 or LIF, as demonstrated by the inability of these proteins to inhibit cell growth or modulate c-myc gene expression in breast cancer cells. Additionally, the high-affinity binding of labeled OM cannot be displaced by LIF. Together, these data suggest that OM is a growth inhibitor for breast cancer cells. The inhibitory activity is mediated predominantly through the OM-specific receptor, and activation of this receptor abrogates growth factor stimulation and down-regulates the c-myc proto-oncogene.

Regulation of an extrathymic T-cell development pathway by oncostatin M.

Most of the T lymphocytes that populate the immune system develop in the thymus before its involution during late adolescence. Therefore, subsequent losses in T cells caused by HIV infection, chemotherapy or age-related factors can greatly diminish immune responses to new antigenic challenge. Here we report the discovery of a thymus-independent pathway of T-cell development that may provide help for T-cell immunodeficiency. We show that expression of an oncostatin M transgene in the early T lineage stimulates a dramatic accumulation of immature and mature T cells in lymph nodes. A functional thymus is not required for this effect as reconstitution of nu/nu mice with transgenic bone marrow stimulated a 500-fold increase in Thy-1+ lymph node cells and restored immune responsiveness to allogeneic mouse melanoma cells. This lymphopoietic pathway is not unique to transgenic mice because administration of oncostatin M protein produced a similar response in non-transgenic mice. These results identify a new pathway of T-cell development and a potential treatment for T-cell immunodeficiency with oncostatin M.

Canine CD34: cloning of the cDNA and evaluation of an antiserum to recombinant protein.

Increasingly, enriched populations of hematopoietic progenitors are used in experimental and clinical transplantation studies. The separation of progenitors is based on the expression of CD34, a marker preferentially expressed on progenitor cells. The dog model has been important for preclinical transplant studies, because it has proven predictive for outcomes in human hematopoietic stem cell transplantation. To identify and isolate canine hematopoietic progenitors, we have cloned a cDNA encoding a CD34 homologue from a canine myelomonocytic leukemia cell line, ML2. The CD34 homologue cDNA predicts an amino acid sequence that is highly conserved with human and murine CD34 in the cytoplasmic domain, transmembrane domain, and C-terminal end of the extracellular domain, but shows considerable divergence from these sequences at the amino-terminal end of the protein. In Western blotting studies, canine CD34 homologue (caCD34) appears to be a heavily and variably glycosylated protein with a molecular weight of approximately 100 kD and shows some tissue-specific differences in protein mass. To evaluate the expression of caCD34 protein, the extracellular domain of caCD34 was expressed as an Ig fusion protein and used as an immunogen to generate a rabbit polyclonal antiserum. The antiserum reacted against the fusion protein, against vascular endothelium, and with three leukemic cell lines. Approximately 1% of canine bone marrow cells stained brightly with antibodies to caCD34 and this population was 25- to 50-fold enriched for colony-forming units-granulocyte-macrophage as compared to unfractionated marrow mononuclear cells. These findings suggest that the canine CD34 homologue is expressed on bone marrow progenitor cells and, thus, that this molecule should be a valuable marker for identifying and isolating canine hematopoietic progenitors for experimental hematopoiesis and stem cell transplantation.

Thrombocytopoietic properties of oncostatin M.

Oncostatin M (OM) is a 28-kD glycoprotein that exhibits a panoply of biologic effects. Based on histologic observations of increased splenic megakaryocytes in nude mice implanted with an OM-secreting cell line, the thrombocytopoietic properties of OM in mice were investigated in culture and in vivo. Alone, OM did not induce megakaryocytic colony formation, but in combination with murine interleukin-3 (IL-3), OM markedly enhanced colony formation. The effects of OM on colony formation were similar to those of IL-6. OM alone augmented acetylcholinesterase in short-term marrow cultures. In normal mice, the administration of OM augmented platelet counts without increasing other circulating blood cell counts. The increment in counts exceeded that observed with IL-6. The kinetics of the OM response suggested that maximal increases in platelets occurred 3 days after the cessation of OM administration, irrespective of the duration of administration. In irradiated mice, OM administration accelerated platelet recovery and prevented the decrease in red blood cells observed in irradiated control animals. The data show that OM behaves as a megakaryocytic maturation factor in vitro and augments platelet production in vivo. Based on these animal data, OM may have potential clinical utility as a thrombocytopoietic agent.

Induction and reversal of long-lived specific unresponsiveness to a T-dependent antigen following CTLA4Ig treatment.

We have examined the ability of CTLA4Ig to induce long-lasting Ag-specific tolerance to a T-dependent Ag. Treatment of mice with murine CTLA4Ig following immunization with SRBC induced immune unresponsiveness to SRBC that was only reversible upon repeated Ag exposure. This unresponsiveness was Ag specific, lasted > 90 days, was observed in thymectomized mice, and was maintained even during challenge with another strong T-dependent Ag. A third challenge with SRBC, however, led to an Ab response. Splenic T cells and B cells from unresponsive mice were functional upon transfer to irradiated hosts, indicating that they had been neither depleted nor rendered permanently "tolerant" by CTLA4Ig. The mechanism of immunosuppression by CTLA4Ig was investigated by measuring cytokine transcripts in spleens of immunized mice. CTLA4Ig treatment following primary immunization blocked the induction of IL-2 transcripts in splenic T cells and IL-4 transcripts in both T cells and non-B, non-T cells. Splenocytes from CTLA4Ig-treated, SRBC-unresponsive mice showed altered induction of IL-2 and IL-4 transcripts, but T cells nonetheless became primed for IL-4 mRNA synthesis despite the lack of a measurable Ab response. Anti-IL-4 mAb and human CTLA4Ig were synergistic in their ability to induce unresponsiveness, indicating that incomplete suppression of IL-4 production by CTLA4Ig limited its effectiveness.

In vitro and in vivo activities of a doxorubicin prodrug in combination with monoclonal antibody beta-lactamase conjugates.

A cephalosporin derivative of doxorubicin (C-Dox) was evaluated as a prodrug for activation by mAb conjugates of the beta-lactamase from Enterobacter cloacae P99 (beta L; EC 3.5.2.6). The conjugates consisted of beta L and the F(ab') fragments of either of the mAbs L6, P1.17, or 96.5. L6 binds to antigens on a variety of carcinomas, including the two lung adenocarcinoma cell lines H2981 and H2987 used in this study. 96.5 binds to the melanoma-associated antigen p97, and P1.17 was used for the control conjugate. C-Dox was found to be less cytotoxic to three different tumor cell lines in vitro compared to the parent drug doxorubicin (Dox). Immunospecific activation took place when the cells were pretreated with beta L conjugates that could bind to antigens on the tumor cells. In vivo toxicity and pharmacokinetics studies in athymic female nu/nu mice revealed that C-Dox was at least 7-fold less toxic than Dox (on a molar basis), despite the fact that a > or = 320-fold greater area-under-the-curve (blood concentration versus time) of C-Dox compared to Dox was obtained 0-2 h after administration of the two agents. Pharmacokinetic studies at maximum tolerated doses in mice bearing xenografts of either H2981 or H2987 revealed that the intratumoral levels of Dox after treatment with L6-beta L in combination with C-Dox were higher than were obtained by either systemic treatment with Dox or a combination of P1.17-beta L and C-Dox. This finding suggested that the conversion of C-Dox to Dox was tumor specific and dependent on the presence of the targeted antigen. Furthermore, the best antitumor activity against both H2981 and H2987 tumors was obtained by treatment with L6-beta L and C-Dox compared to P1.17-beta L and C-Dox or Dox alone. Thus, higher levels of Dox corresponded to greater therapeutic effects in both of the tumor models studied.

CTLA4Ig treatment ameliorates the lethality of murine graft-versus-host disease across major histocompatibility complex barriers.

Graft-versus-host disease (GVHD), a pathological condition associated with BMT, results from activation of donor T lymphocytes by host tissues. CD28 and CTLA-4 are structurally related T cell receptors for members of the B7 (CD80) gene family, which transmit important costimulatory signals for T cell activation in vitro and in vivo. Here we have investigated the effects of CTLA4Ig, a soluble form of CTLA-4, on lethal GVHD in a murine model. Lethal GVHD was induced by transfer of parent C57BL/6 bone marrow and spleen cells into lethally irradiated (C57BL/6 x DBA/2)F1 recipients. Short courses of treatment with CTLA4Ig did not block engraftment, but prolonged survival of BMT recipients even when administration was delayed for 6 days after transplantation. CTLA4Ig-treated survivors of GVHD maintained body weight and did not exhibit visible signs of GVHD. However, treatment regimens that maximally prolonged survival did not detectably prevent T cell-mediated hematological abnormalities associated with GVHD, including pancytopenia and abnormal cellular composition of the spleen. Our data thus show that the lethality of acute GVHD in this model system is more dependent upon CD28/CTLA-4 costimulation than are other GVHD-associated abnormalities, and can be blocked for an extended period by brief treatment with CTLA4Ig.

Selective activation of anticancer prodrugs by monoclonal antibody-enzyme conjugates.

Several recent reports have demonstrated that anticancer drugs can be generated site-selectively at solid tumors by monoclonal antibody-enzyme conjugates targeted to antigens on tumor cell surfaces. The first step in this drug targeting approach involves the delivery of the enzyme conjugate to a tumor cell population. After the conjugate has localized within the tumor and cleared from non-target tissues, a relatively non-cytotoxic drug precursor (prodrug) is administered. Upon contact with the targeted enzyme, the prodrug is converted into a toxic drug. Several examples are presented to illustrate this targeting strategy. Monoclonal antibody-beta-lactamase conjugates have been developed to activate a panel of anticancer prodrugs that are mechanistically dissimilar. The antitumor activities of the monoclonal antibody-beta-lactamase conjugate/prodrug combinations exceed those obtained by systemic drug administration, and are immunologically specific. In another example involving targeted cytosine deaminase for the generation of 5-fluorouracil, it is shown that as much as 17 times more drug can be delivered within a tumor compared to when 5-fluorouracil is administered alone. The method of using targeted enzymes for prodrug activation can be extended to include prodrugs that release very potent drugs, such as palytoxin, a marine natural product, and to treat cells that have the multidrug resistance phenotype. Some of the requirements for successful therapy with this approach for cancer therapy are discussed.

Synthesis and characterization of monoclonal antibody-beta-lactamase conjugates.

beta-Lactamase from Enterobacter cloacae (beta L) was conjugated to the Fab'2 fragment of the monoclonal antibody L6 through a thioether linkage. Although L6-Fab'2-beta L was capable of activating the antitumor prodrug, 7-(phenylacetamido)cephalosporin mustard, it was impaired in its ability to bind to antigens on the H2981 human lung adenocarcinoma cell line. As a result, studies were undertaken to prepare conjugates with preserved binding activities. L6-Fab'-beta L and a dimeric conjugate consisting of two individual L6-Fab' units linked to a single beta L molecule (dimeric L6-beta L) were prepared by linking L6-Fab'-SH to maleimide-substituted beta L. Analysis of these conjugates by SDS-PAGE indicated that the linkage involved heavy-chain thiol groups on L6 that are most likely in the hinge region and are therefore removed from the antigen binding site of the antibody. Cell binding studies revealed that the monovalent conjugate L6-Fab'-beta L bound as well as L6-Fab'. Dimeric L6-beta L displayed slightly less binding than L6-Fab'2, but bound substantially better than L6-Fab'2-beta L. Lower concentrations of dimeric L6-beta L compared to L6-Fab'2-beta L were required to convert the prodrug 7-(phenylacetamido)-cephalosporin mustard into the cytotoxic drug phenylenediamine mustard. Localization studies were performed in nude mice with H2981 subcutaneous tumor xenografts. At 96 h post conjugate treatment, there was no significant difference in tumor concentration between L6-Fab'2-beta L and dimeric L6-beta L.(ABSTRACT TRUNCATED AT 250 WORDS)

Intratumoral generation of 5-fluorouracil mediated by an antibody-cytosine deaminase conjugate in combination with 5-fluorocytosine.

Cytosine deaminase (CD) is a microbial enzyme that can convert the antifungal agent 5-fluorocytosine (5-FC) into the antitumor agent, 5-fluorouracil (5-FU). The enzyme was chemically conjugated to the L6 monoclonal antibody, forming a conjugate that bound to antigens on the H2981 lung adenocarcinoma. Detailed studies were undertaken to determine the extent to which L6-CD generated 5-FU in tumor-bearing mice. Very high tumor:blood ratios of L6-CD (42:1) in vivo were obtained by injecting the conjugate followed 24 h later by an antiidiotypic antibody that could bind to circulating L6-CD but not to L6-CD that was bound to H2981 cells. As a result, significantly more 5-FC could be administered (> 800 mg/kg) than 5-FU (90 mg/kg). L6-CD converted 5-FC into 5-FU such that the L6-CD/antiidiotypic monoclonal antibody/5-FC combination resulted in 17 times more intratumoral 5-FU compared to systemic 5-FU administration. The conversion was antigen dependent since much lower intratumoral 5-FU levels were obtained in H3719 tumors that failed to localize L6-CD. The conversion of 5-FC into 5-FU was low in blood, kidneys, and liver. This demonstrates that a major increase in intratumoral drug concentrations can be attained with an monoclonal antibody-enzyme conjugate in combination with an anticancer prodrug compared to systemic drug therapy.

Antitumor activities of a cephalosporin prodrug in combination with monoclonal antibody-beta-lactamase conjugates.

7-(Phenylacetamido)cephalosporin mustard (CM) and 7-(4-carboxybutanamido)cephalosporin mustard (CCM) were developed as anticancer prodrugs that could be activated site selectively by monoclonal antibody-beta-lactamase conjugates targeted to antigens present on tumor cell surfaces. Both CM and CCM were hydrolyzed by purified beta-lactamases from Escherichia coli (EC beta L), Bacillus cereus (BC beta L), and Enterobacter cloacae (ECl beta L). This resulted in the release of phenylenediamine mustard (PDM), a potent cytotoxic drug. The Km and kcat values of the reactions were determined, and it was found that ECl beta L effected the hydrolysis of CM and CCM more rapidly than the other enzymes. Conjugates of ECl beta L were prepared by reacting maleimide-substituted F(ab')2 fragments of the monoclonal antibodies L6 and P1.17 to ECl beta L that had been modified with sulfhydryl groups. In vitro experiments indicated that CCM (IC50 = 25-45 microM) was less toxic than PDM (IC50 = 1.5 microM) to H2981 lung adenocarcinoma cells (L6 antigen positive, P1.17 antigen negative) and that immunologically specific prodrug activation took place when the cells were treated with L6-ECl beta L. In vivo experiments in nude mice demonstrated that CCM was less toxic than CM, and that both prodrugs were much less toxic than PDM. Neither CCM nor PDM exerted antitumor activity on subcutaneous H2981 tumors in vivo. However, a significant antitumor effect was obtained in mice that received L6-ECl beta L 96 h prior to the administration of CCM. The effect was immunologically specific (P < 0.05), since a smaller degree of antitumor activity was obtained in mice that received the nonbinding control conjugate P1.17-ECl beta L prior to CCM.(ABSTRACT TRUNCATED AT 250 WORDS)

Application of monoclonal antibodies against cytosine deaminase for the in vivo clearance of a cytosine deaminase immunoconjugate.

The selective delivery of anticancer drugs to tumors vs normal tissue using targeted antibody-enzyme conjugates for prodrug activation is limited by the amount of drug generated by blood-borne enzyme. Clearance of non-tumor-associated conjugate would increase the tumor/blood conjugate ratio, and enable larger amounts of prodrugs to be administered. A method for clearing the monoclonal antibody (mAb) conjugate L6-cytosine deaminase (L6-CD) was established by using an antibody raised against CD. The mAb 102-26 was obtained by immunizing BALB/C mice with CD conjugated to keyhole limpet hemocyanin. 102-26 was able to precipitate purified CD from solution as assessed by radioimmune precipitation and recognized CD in Western blot analyses. Similar studies were used to establish that 102-26 also recognized CD when conjugated to the L6 and 1F5 mAbs. Selective removal of L6-CD from the circulation of nude mice bearing H2981 human lung adenocarcinoma (L6-antigen positive) was achieved by injecting 102-26 24 h after L6-CD administration. High T/B ratios were obtained by clearance of a L6-CD (38:1 compared to 1.3:1 without clearance).

Prodrugs of doxorubicin and melphalan and their activation by a monoclonal antibody-penicillin-G amidase conjugate.

The syntheses and cytotoxic activities of substituted N-phenylacetamido derivatives of doxorubicin and melphalan are described. The derivatives were designed as prodrugs which could be activated in a site-specific manner by monoclonal antibody-penicillin-G amidase (mAb-PGA) conjugates. N-(Phenylacetamido)doxorubicin (2) and N-(phenylacetyl)melphalan (6) were found to be 10- and 20-fold less cytotoxic against H2981 lung adenocarcinoma cells than doxorubicin and melphalan, respectively. When incubated with PGA, the cytotoxicity of 2 and 6 increased and became equivalent to that of the corresponding drugs from which they were made. The poor solubility characteristics of 2 in aqueous solutions provided the basis for the development of the more soluble doxorubicin derivatives, N-(4-aminophenylacetyl)doxorubicin (3) and N-(4-phosphonooxy)phenylacetyl)-doxorubicin (4). In vitro cytotoxicity assays indicated that 3 and 4 were at least 1000-fold less toxic than doxorubicin against H2981 cells. PGA and the mAb conjugate L6-PGA were able to effect the activation of 3 and 6 on H2981 cells (L6-antigen positive). Hydrolysis of the phosphate group of 4 was required prior to activation with PGA or L6-PGA. This was achieved using alkaline phosphatase, or by exposing 4 to phosphatases present in cell culture medium. The activation of 3, 4, and 6 on H2981 cells by L6-PGA occurred in an immunologically specific manner, since activation could be blocked by saturating cell surface antigens with L6 prior to treatment with L6-PGA. These results demonstrate that 3, 4, and 6 are prodrugs that can be specifically activated to release clinically approved anticancer agents by a mAb-PGA conjugate.