Patrizia Castellani

Selective targeting of tumoral vasculature: Comparison of different formats of an antibody (L19) to the ED-B domain of fibronectin

We recently demonstrated that a human recombinant scFv, L19, reacting with the ED‐B domain of fibronectin, a marker of angiogenesis, selectively targets tumoral vasculature in vivo. Using the variable regions of L19, we constructed and expressed a human “small immunoprotein” (SIP) and a complete human IgG1 and performed biodistribution studies in tumor‐bearing mice to compare the blood clearance rate, in vivo stability and performance in tumor targeting of the 3 L19 formats [dimeric scFv (scFv)2, SIP and IgG1]. The accumulation of the different antibody formats in the tumors studied was a consequence of the clearance rate and in vivo stability of the molecules. Using the SIP, the %ID/g in tumors was 2–5 times higher than that of the (scFv)2, reaching a maximum 4–6 hr after injection. By contrast, the accumulation of IgG1 in tumors constantly rose during the experiments. However, due to its slow clearance, the tumor‐blood ratio of the %ID/g after 144 hr was only about 3 compared to a ratio of 10 for the (scFv)2 and 70 for the SIP after the same period of time. The different in vivo behavior of these 3 completely human L19 formats could be exploited for different diagnostic and/or therapeutic purposes, depending on clinical needs and disease. Furthermore, the fact that ED‐B is 100% homologous in human and mouse, which ensures that L19 reacts equally well with the human and the murine antigen, should expedite the transfer of these reagents to clinical trials.

Phage antibodies with pan‐species recognition of the oncofoetal angiogenesis marker fibronectin ED‐B domain

Fibronectin (FN) exists in several polymorphic forms due to alternative splicing. The B‐FN isoform (with ED‐B domain inserted by splicing) is present in the stroma of foetal and neoplastic tissues and in adult and neoplastic blood vessels during angiogenesis but is undetectable in mature vessels. This isoform, therefore, represents a promising marker for angiogenesis, as already shown using the mouse monoclonal antibody (MAb) BC‐I directed against an epitope on human B‐FN. However, this MAb does not directly recognise the human ED‐B domain nor does it recognise B‐FN of other species; therefore, it cannot be used as a marker of angiogenesis in animal models. In principle, antibodies directed against the human ED‐B domain should provide pan‐species markers for angiogenesis as the sequence of this domain is highly conserved in different species (and identical in humans and mice). As it has proved difficult to obtain such antibodies by hybridoma technology, we used phage display technology. Here, we describe the isolation of human antibody fragments against the human ED‐B domain that bind to human, mouse and chicken B‐FN. As shown by immunohistochemistry, the antibody fragments stain human neoplastic tissues and the human, mouse and chicken neovasculature.

Differentiation between high- and low-grade astrocytoma using a human recombinant antibody to the extra domain-B of fibronectin.

Different fibronectin (FN) isoforms are generated by the alternative splicing of the primary FN transcript. We previously demonstrated that the isoform containing the extra domain B sequence of fibronectin (B-FN), a complete type-III-homology repeat, is a marker of angiogenesis that accumulates around neovasculature only during angiogenic processes. We produced a single-chain human recombinant antibody (scFv), L19, which reacts specifically with B-FN and selectively targets tumor vasculature in vivo. We used this scFv and an antibody against a pan-endothelial marker (Factor VIII) in a double-staining procedure on specimens of low- and high-grade astrocytomas to determine the percentage of B-FN-positive vessels, (denominating the resulting value angiogenic index [AI]). Compared to vascular density and proliferative activity (evaluated using antibodies to Factor VIII and Ki67, respectively), AI correlated better with tumor grade (1.6 +/- 2.6% and 92.0 +/- 8.7% of B-FN-positive vessels in low- and high-grade astrocytomas, respectively) and was a more precise diagnostic tool than either of the two conventional methods. In fact, discriminating analysis using these three parameters showed that only AI accurately classified 100% of the cases studied, compared to 64% and 89% correctly diagnosed by vascular density and of proliferating cells, respectively.

Transforming growth factor‐β regulates the splicing pattern of fibronectin messenger RNA precursor

Fibronectin (FN) polymorphism is caused by alternative splicing patterns in at least three regions (ED‐A, ED‐B and IIICS) of the primary transcript of a single gene. Using monoclonal antibodies, we previously demonstrated that transforming growth factor‐β (TGF‐β) preferentially increases the accumulation of the FN isoforms containing the ED‐A sequence in cultured normal human fibroblasts [Balza et al., (1988) FEBS Lett. 228, 42‐44]. To determine the basis of this effect, we have examined through S1 nuclease analysis, the levels of ED‐A‐ and ED‐B‐containing mRNAs in cultured normal human skin flbroblasts before and after TGF‐β treatment. These experiments have shown that TGF‐β increases the relative amount of m‐RNA for ED‐A‐ and ED‐B‐containing FN isoforms. These data demonstrate that a growth factor may regulate the splicing pattern of a pre‐mRNA.

Molecular imaging of atherosclerotic plaques using a human antibody against the extra-domain B of fibronectin

Current imaging modalities of human atherosclerosis, such as angiography, ultrasound, and computed tomography, visualize plaque morphology. However, methods that provide insight into plaque biology using molecular tools are still insufficient. The extra-domain B (ED-B) is inserted into the fibronectin molecule by alternative splicing during angiogenesis and tissue remodeling but is virtually undetectable in normal adult tissues. Angiogenesis and tissue repair are also hallmarks of advanced plaques. For imaging atherosclerotic plaques, the human antibody L19 (specific against ED-B) and a negative control antibody were labeled with radioiodine or infrared fluorophores and injected intravenously into atherosclerotic apolipoprotein E–null (ApoE−/−) or normal wild-type mice. Aortas isolated 4 hours, 24 hours, and 3 days after injection exhibited a selective and stable uptake of L19 when using radiographic or fluorescent imaging. L19 binding was confined to the plaques as assessed by fat staining. Comparisons between fat staining and autoradiographies 24 hours after 125I-labeled L19 revealed a significant correlation (r=0.89; P<0.0001). Minimal antibody uptake was observed in normal vessels from wild-type mice receiving the L19 antibody and in atherosclerotic vessels from ApoE−/− mice receiving the negative control antibody. Immunohistochemical studies revealed increased expression of ED-B not only in murine but also in human plaques, in which it was found predominantly around vasa vasorum and plaque matrix. In summary, we demonstrate selective targeting of atheromas in mice using the human antibody to the ED-B domain of fibronectin. Thus, our findings may set the stage for antibody-based molecular imaging of atherosclerotic plaques in the intact organism.

Identification of a Glioblastoma-Associated Tenascin-C Isoform by a High Affinity Recombinant Antibody

Tenascin-C exists in several polymorphic isoforms due to alternative splicing of nine fibronectin-like type III repeats. Large Tenascin-C isoforms are present in almost all normal adult tissues but are upregulated in fetal, regenerating, and neoplastic tissues. Here, we report a human antibody fragment, TN11, derived from a phage library with high affinity for the spliced repeat C and demonstrate that this repeat is undetectable in normal adult tissues, barely detectable or undetectable in breast, lung and gastric carcinomas, meningioma, and low grade astrocytoma, but extremely abundant in high grade astrocytoma (grade III and glioblastoma), especially around vascular structures and proliferating cells. The antibody appears to have potential for development of a therapeutic agent for patients with high grade astrocytoma.

DAMPs and inflammatory processes: the role of redox in the different outcomes

Inflammation is deeply entangled with redox modulation. Triggering of PRRs on inflammatory cells induces ROS generation. As a consequence, activated cells mount antioxidant responses to counteract the possible harmful effects of oxidation. Therefore, when repair is completed, homeostasis is restored. Here, we describe some recent results showing that an exuberant antioxidant response to pro–oxidant inflammatory stimuli modifies not only the intra– but also the extracellular redox and contributes to the outcome of the inflammatory process. In particular, the role of redox modulation in IL–1β secretion, in B lymphocyte differentiation to plasma cells, and in tumor progression will be discussed, and the potential consequences of extracellular redox alterations on DAMP activity will be considered.

Targeted Delivery of Tumor Necrosis Factor-α to Tumor Vessels Induces a Therapeutic T Cell–Mediated Immune Response that Protects the Host Against Syngeneic Tumors of Different Histologic Origin

Purpose: We sought to demonstrate that a single systemic administration of L19mTNFα (a fusion protein constituted by the scFv L19 specific for the oncofetal ED-B domain of fibronectin and tumor necrosis factor α, TNFα) in combination with melphalan induced complete and long-lasting tumor eradication in tumor-bearing mice and triggered the generation of a specific T cell–based immune response that protects the animals from a second tumor challenge, as well as from challenges with syngeneic tumor cells of different histologic origin. Experimental Design and Results: Treatment with L19mTNFα, in combination with melphalan, induced complete tumor regression in 83% of BALB/c mice with WEHI-164 fibrosarcoma and 33% of animals with C51 colon carcinoma. All cured mice rejected challenges with the same tumor cells and, in a very high percentage of animals, also rejected challenges with syngeneic tumor cells of different histologic origin. In adoptive immunity transfer experiments, the splenocytes from tumor-cured mice protected naive mice both from C51 colon carcinoma and from WEHI-164 fibrosarcoma. Similar results were also obtained in adoptive immunity transfer experiments using severely immunodepressed mice. Experiments using depleted splenocytes showed that T cells play a major role in tumor rejection. Conclusions: The results show that the selective targeting of mTNFα to the tumor enhances its immunostimulatory properties to the point of generating a therapeutic immune response against different histologically unrelated syngeneic tumors. These findings predicate treatment approaches for cancer patients based on the targeted delivery of TNFα to the tumor vasculature.

Lack of specificity of endoglin expression for tumor blood vessels

A monoclonal antibody (MAb; A11) has been raised following mouse immunization with cultured human microvascular endothelial cells. The MAb showed a strong positivity within tumor vessels in glioblastoma and breast carcinoma samples, and the distribution was consistent with antigen association with vascular endothelial cells. A purification procedure of the antigen was developed starting from DG‐RSV‐LT‐2, an immortalized human endothelial cell line. Molecular mass, N‐terminal sequence of the purified antigen and localization on endothelial cell surface allowed identification with human endoglin (CD105). Flow cytometry analysis of a group of normal and transformed cell lines showed that, besides endothelial cells and myelocytic leukemia cells already shown to be positive, fetal fibroblasts, choriocarcinoma, fibrosarcoma and rhabdomyosarcoma cell lines were also positive for this antigen. Immunohistochemic analysis of several normal adult tissues revealed a more extensive presence of the antigen in normal vessels compared to that described with previously characterized antibodies. In fact, even though the staining was weaker than in tumor tissues, all tissues were found to be positive, at least in microvessels, except for normal breast. Moreover, in some tissues (glands and reproductive tract) a positive reaction was observed in the stroma. Since endoglin has been proposed as a possible target for antiangiogenic therapy in tumor patients and our data demonstrate a sizable amount of endoglin in normal vessels and stroma, its clinical use should be carefully reevaluated.

The redox state of the lung cancer microenvironment depends on the levels of thioredoxin expressed by tumor cells and affects tumor progression and response to prooxidants

Here we report that human nonsmall cell lung carcinomas overexpress macrophage migration inhibitory factor (MIF) and thioredoxin (Trx), 2 oxidoreductases with cytokine function, and contain more abundant nonprotein thiols (glutathione and cysteine) than nonneoplastic lung tissues. Cell clones derived from the same lung carcinoma cell lines but expressing different levels of Trx and/or MIF displayed growth rates in vitro and in vivo correlating with Trx but not with MIF. Interestingly, the different clones generate extracellularly reduced nonprotein thiols, in amounts related to the Trx content and inhibited by inhibitors of Trx function. Each clone also showed distinct responses to the prooxidant compound arsenic trioxide. Cells with a strongly antioxidant and aggressive phenotype were more susceptible to the cytotoxic effect of the drug than cells expressing little Trx. The latter counteracted the oxidative stress by increasing Trx expression and thiol release. Together these results indicate that different human lung cancer cell lines have distinct redox properties defined by the levels of Trx and nonprotein thiols, the higher antioxidant phenotype correlating with the higher aggressiveness. Moreover, the redox phenotype dictates their response to prooxidant drugs and must be taken into account when therapeutic interventions with redox active substances are considered.