Exploiting the tumor microenvironment in the development of targeted cancer gene therapy.

The future success of cancer gene therapy is critically dependent upon the development of safe, practical and effective targeting strategies. In this study, we describe a novel and broadly applicable targeting approach in which the induction of apoptotic tumor cell death is linked to the differential expression within the tumor microenvironment of elevated levels of the pro-angiogenic cytokine vascular endothelial growth factor (VEGF). As VEGF is generally absent or produced at only low levels in most normal tissues, undesirable toxicity will not result even if the therapeutic gene in question is inadvertently expressed in non-targeted tissue sites. The basic approach makes use of a chimeric cell-surface protein in which the membrane-spanning and cytoplasmic 'death domain' of the pro-apoptotic protein Fas are fused in frame to the extracellular ligand-binding domain of the VEGF receptor Flk-1/KDR/VEGFR2 (Flk-1/Fas). The resultant chimeric Flk-1/Fas receptor was found to be stable and capable of inducing a rapid apoptotic response when expressed in tumor cells that produce endogenous VEGF. Importantly, in the absence of VEGF, transduced tumor cells remain viable although they can be triggered to die by the addition of recombinant VEGF. Given the key role played by VEGF in tumor development and progression, it is proposed that the Flk-1/Fas chimera may have great potential in the context of tumor cell-targeted cancer gene therapy.

Non-small cell lung cancer cyclooxygenase-2-dependent invasion is mediated by CD44.

Elevated tumor cyclooxygenase (COX-2) expression is associated with increased angiogenesis, tumor invasion, and suppression of host immunity. We have previously shown that genetic inhibition of tumor COX-2 expression reverses the immunosuppression induced by non-small cell lung cancer (NSCLC). To assess the impact of COX-2 expression in lung cancer invasiveness, NSCLC cell lines were transduced with a retroviral vector expressing the human COX-2 cDNA in the sense (COX-2-S) and antisense (COX-2-AS) orientations. COX-2-S clones expressed significantly more COX-2 protein, produced 10-fold more prostaglandin E(2), and demonstrated an enhanced invasive capacity compared with control vector-transduced or parental cells. CD44, the cell surface receptor for hyaluronate, was overexpressed in COX-2-S cells, and specific blockade of CD44 significantly decreased tumor cell invasion. In contrast, COX-2-AS clones had a very limited capacity for invasion and showed diminished expression of CD44. These findings suggest that a COX-2-mediated, CD44-dependent pathway is operative in NSCLC invasion. Because tumor COX-2 expression appears to have a multifaceted role in conferring the malignant phenotype, COX-2 may be an important target for gene or pharmacologic therapy in NSCLC.

Mechanisms mediating the effects of IL-3 gene expression on tumor growth.

IL-3 gene expression within tumors leads to host-cell infiltration, particularly by macrophages, slower tumor growth, and enhanced immunogenicity. Surprisingly, tumor-associated macrophages (TAMs) from within FSAN-JmIL3 tumors had decreased expression of TNF-alpha and iNOS. On short-term culture, TAMs from FSAN-JmIL3 tumors regained their capacity to produce TNF-alpha and NO, indicating that they were primed in vivo. In vitro experiments were unable to demonstrate differences between FSAN-JmIL3 and FSAN tumor cells in their ability to stimulate TNF-alpha production by TAMs. In the absence of evidence that TAM activation was responsible for the slower growth of FSAN-JmIL3 tumors, the response of tumor cells to these effector molecules was studied. TNF-alpha and NO were cytotoxic for FSAN-JmIL3 cells but growth stimulatory for FSAN. These tumor-related phenotypic changes may contribute as much if not more than functional changes in host infiltrating cells to the slower growth of FSAN-JmIL3 tumors in vivo.

Combining radiation therapy with interleukin-3 gene immunotherapy.

The goal of this study was to explore immunological strategies to increase local and systemic tumor control in patients receiving radiation therapy. In previous studies, interleukin-3 (IL-3) gene expression within murine tumors was shown to increase their response to irradiation through immune mechanisms. In this study, the efficacy of systemically administered IL-3 gene-transduced irradiated tumor cell vaccines was tested for their ability to augment radiation responses against established immunogenic (FSAR) and nonimmunogenic (FSAN) tumors. Vaccines of irradiated FSAR/FSAN or FSAN-JmIL-3/FSAR-JmIL-3 cells were given intraperitoneally just before and after local irradiation of parental tumors with diameters of 8 mm, as well as in two booster doses. The IL-3 gene-transduced tumor cell vaccines were more effective than the parental vaccines at delaying tumor growth after irradiation, although no complete cures resulted. Responses were largely specific to the tumor type, indicating that tumor-specific immunity was enhanced by IL-3 vaccine administration. When the experiment was repeated in the C3H/HeJ mice, which are deficient in tumor necrosis factor-alpha production, the vaccines were still effective, but less so than in C3H/HeN mice. Systemic IL-3 vaccine treatment increased intratumoral levels of intercellular adhesion molecule-1, Mac-1, EB22/5.3, tumor necrosis factor-alpha, and IL-1 mRNA in irradiated tumors, indicating that cellular infiltration was part of the response. The study demonstrates that local radiation therapy can enhance the efficacy of genetically altered vaccine-based immunotherapy for cancer by decreasing tumor burden. At the same time, tumor cell vaccines may improve the cure rate of local radiation therapy by eliminating residual cancer cells. Although less effective than intratumoral gene expression, administration of IL-3 gene-transduced tumor cell vaccines is clinically a more feasible strategy that may be useful in situations in which the tumor load is small.

Photodynamic therapy-mediated immune response against subcutaneous mouse tumors.

The curative ability of photodynamic therapy (PDT) is severely compromised if treated tumors are growing in immunodeficient hosts. Reconstitution of severe combined immunodeficient (scid) mice with splenocytes from naive immunologically intact BALB/c mice did not improve the response to Photofrin-based PDT of EMT6 tumors growing in these animals. In contrast, adoptive transfer of BALB/c splenocytes containing EMT6 tumor-sensitized immune cells had a dramatic effect on tumor regrowth after PDT. For instance, full restoration of the curative effect of PDT was achieved with scid mice that received splenocytes from BALB/c donors that were cured of EMT6 tumors by PDT 5 weeks before adoptive transfer. Splenocytes obtained from donors cured of EMT6 tumors using X-rays were much less effective. Selective in vitro depletion of specific T-cell populations from engrafting splenocytes indicated that CTLs are the main immune effector cells responsible for conferring the curative outcome to PDT in this experimental model, whereas helper T lymphocytes play a supportive role. The immune specificity of these T-cell populations was demonstrated by the absence of cross-reactivity between the EMT6 and Meth-A tumor models (mismatch between tumors growing in splenocyte donors and recipients). The immunocompetent BALB/c mice that received adoptively transferred splenocytes containing PDT-generated, tumor-sensitized immune cells also benefited from the improved outcome of PDT of tumors they were bearing. This was demonstrated not only with the fairly immunogenic EMT6 tumor model but also with weakly immunogenic Line 1 carcinomas. The results of this study indicate that PDT is a highly effective means of generating tumor-sensitized immune cells that can be recovered from lymphoid sites distant to the treated tumor at protracted time intervals after PDT, which asserts their immune memory character. It is also shown that the treatment of tumors by PDT creates the conditions necessary for converting the inactive adoptively transferred pre-effector, tumor-sensitized immune cells into fully functional antitumor effector cells. An additional finding of this study is the evidence of NK cell activation in PDT-treated Meth-A sarcomas.

Alternatively spliced CD44 isoforms containing exon v10 promote cellular adhesion through the recognition of chondroitin sulfate-modified CD44.

Correlations have been noted between the expression of certain alternatively spliced CD44 isoforms and the metastatic propensity of various histologically distinct tumor cell types. The precise mechanism by which particular CD44 isoforms contribute to the metastatic process is, however, unclear. In the present study we demonstrate that CD44R2, a CD44 isoform highly expressed on activated and transformed hemopoietic cells, can recognize and bind a common determinant present on CD44H and CD44R1. Importantly, CD44H lacked this activity. Pretreatment of TIL1 cells expressing CD44H or CD44R1 with chondroitinase ABC inhibited adhesion to CD44R2, suggesting that the unique inserted region present within the CD44R2 molecule, encoded by exon v10, mediates cell adhesion by potentiating the recognition of chondroitin sulfate moieties presented in association with other CD44 molecules. These data help explain the differential involvement of v10-containing CD44 isoforms in tumor metastasis.

Identification and characterization of CD44RC, a novel alternatively spliced soluble CD44 isoform that can potentiate the hyaluronan binding activity of cell surface CD44.

Soluble CD44 proteins generated by proteolytic cleavage or aberrant intron retention have been shown to antagonize the ligand binding activity of the corresponding cell surface receptor, inducing apoptosis and inhibiting tumor growth. Interestingly, such findings appear to contradict recent studies demonstrating a correlation between the presence of high levels of soluble CD44 in the serum of cancer patients and poor prognosis. In the present study, we report the cloning of a novel, naturally occurring, differentially expressed, soluble CD44 isoform, designated CD44RC, which, in contrast to previously described soluble CD44 proteins, can dramatically enhance the hyaluronan binding activity of cell surface CD44. Sequence analysis suggests that CD44RC is generated by an alternative splicing event in which the 3' end of CD44 exon 2 is spliced into an internal splice acceptor site present within exon 18, altering reading frame and giving rise to a soluble protein with a unique COOH terminus. Functional studies suggest that CD44RC enhances hyaluronan binding by adhering to chondroitin sulfate side-chains attached to cell surface CD44, generating a multivalent complex with increased avidity for hyaluronan.

Correlation of CD44 expression with proliferative activity of normal human breast epithelial cells in culture.

A number of studies have shown that certain variant isoforms of CD44 are overexpressed in human breast cancer, suggesting their use as indicators of the presence of malignant cells. We now show that CD44 isoform mRNA and protein expression is upregulated in normal human breast epithelial cells (HBEC) when these cells are stimulated to proliferate in culture. Reverse transcription-PCR analysis of cultured normal HBEC revealed complex patterns of CD44 mRNA expression that were indistinguishable from patterns previously shown to be characteristic of tissue samples containing malignant HBEC. CD44v6-expressing cells were identified in cultures generated from FACS-purified populations of either normal luminal (CALLA-MUC-1+) or myoepithelial (CALLA+MUC-1-) cells, even though immunohistochemical analysis of normal breast tissue sections confirmed CD44v6 expression to be limited to the myoepithelium in vivo. Increased expression of both CD44v mRNA and protein in cultured populations of normal HBEC was shown to correlate positively with the proportion of cells that were proliferating (Ki-67+) independent of cell density. These results indicate that activation of CD44 variant isoform expression in HBEC occurs as a normal response to factors that stimulate their proliferation and suggests caution in the use of this marker to identify malignant cells.

Hyaluronic acid capsule modulates M protein-mediated adherence and acts as a ligand for attachment of group A Streptococcus to CD44 on human keratinocytes.

We used wild-type and isogenic mutant strains of group A Streptococcus (GAS) that expressed M protein, capsule, or both to study the function of M protein and the hyaluronic acid capsular polysaccharide in attachment of GAS to human keratinocytes. Types 6 and 24, but not type 18, M protein were found to mediate attachment of GAS to soft palate or skin keratinocytes, but this interaction was prevented by the hyaluronic acid capsule on highly encapsulated, or mucoid, strains. Monoclonal antibody to CD44, the principal hyaluronic acid-binding receptor on keratinocytes, inhibited attachment of both highly encapsulated and poorly encapsulated wild type strains of GAS, but not the attachment of acapsular mutants. Transfection of K562 cells with cDNA encoding human CD44 conferred the capacity to bind each of six wild-type strains of GAS, but not to bind acapsular mutants. Because, in contrast to other potential adhesins, the group A streptococcal capsule is both highly conserved and surface-exposed, it may serve as a universal adhesin for attachment of diverse strains of GAS to keratinocytes of the pharyngeal mucosa and the skin.

Effects of IL-3 gene expression on tumor response to irradiation in vitro and in vivo.

Expression of a murine interleukin 3 gene in murine fibrosarcoma cells (FSA-JmIL-3) did not alter their survival after in vitro irradiation. However, FSA-JmIL-3 tumors established in vivo were much more sensitive to irradiation than was the parental tumor. Following 25 Gy of irradiation, parental fibrosarcoma tumors regrew after a growth delay of 10 days, but FSA-JmIL-3 tumors continued to regress. Examination of the cellular composition of tumors following irradiation revealed that, instead of tumor cell repopulation, the FSA-JmIL-3 tumors became heavily infiltrated with lymphocytes, indicating that the effect of irradiation was to allow the IL-3-elicited cellular immune response to infiltrate the tumors and mediate rejection. This study indicates that combining gene immunotherapy approaches with radiotherapy might increase the effectiveness of both, and it seems logical to pursue such treatment options.

Radiation response of connexin43-transfected cells in relation to the "contact effect".

Some cell lines grown for only two cell doublings as multicell spheroids develop a form of resistance to killing by ionizing radiation that has been called the "contact" effect. While our previous results have implicated a role for higher order chromatin structure in the contact effect, another possible explanation is the presence of intercellular gap junctions that might facilitate communication between cells grown as spheroids and thereby enhance the ability of cells to resist or recover from radiation damage. To examine the role of gap junctions in the contact effect, rat glioma C6 and mouse EMT6 cell lines were transfected with a gene encoding the gap junctional protein connexin43. While C6 glioma cells are deficient in gap junctional communication, cells from spheroids were nonetheless more resistant than monolayers to killing by ionizing radiation, and the contact effect was present to a similar extent in the three transfected clones. For mouse EMT6 cells, radiosensitivity was similar whether cells were grown as monolayers or spheroids. Transfection of EMT6 cells with connexin43 increased gap junctional communication but did not promote development of a contact effect. Tumor volume doubling time in SCID mice increased significantly for one transfected clone; however, doubling time in vitro was also increased relative to the EMT6 parent. We conclude that extensive gap junctional communication is not a requirement for the increased radiation resistance observed when some cell lines are grown as spheroids.

CD44 regulates phagocytosis of apoptotic neutrophil granulocytes, but not apoptotic lymphocytes, by human macrophages.

Phagocytosis of apoptotic neutrophil granulocytes by macrophages at inflammatory sites is an important determinant of the process by which inflammation resolves. We demonstrate that phagocytosis of apoptotic neutrophils, but not apoptotic lymphocytes, by human monocyte-derived macrophages is augmented rapidly following ligation of CD44 by bivalent Abs in vitro. Previously defined inhibitors of apoptotic cell recognition did not affect CD44-augmented phagocytosis of apoptotic neutrophils, suggesting that unique molecular recognition pathways are involved. These observations, together with the lack of effect of CD44 Abs upon macrophage phagocytosis of zymosan or Ig-opsonized erythrocytes, imply that CD44 may regulate the differential clearance of apoptotic leukocytes during evolution of inflammatory responses. This represents a novel role for CD44 in inflammation and tissue repair.

Role of macrophage-colony-stimulating factor in regulating the accumulation and phenotype of tumor-associated macrophages.

In order to better define the role played by tumor-cell-derived macrophage-colony-stimulating factor (M-CSF) in regulating the recruitment and phenotype of tumor-associated macrophages, Polyoma large T-transformed fibroblastoid cell lines, derived from M-CSF-deficient osteopetrotic op/op mice and their phenotypically normal op/+ littermate controls, were inoculated into SCID (severe combined immunodeficiency) recipients and both the proportion and phenotype of the macrophages present within the tumors generated were determined. The results obtained indicate that, although tumors derived from M-CSF-deficient and M-CSF-producing tumor cell inoculate contain a similar proportion of macrophages, the macrophages isolated from tumors lacking M-CSF appear morphologically less mature and express lower levels of interleukin 1 beta, tumor necrosis factor alpha and FcR gamma II mRNA. Taken together, these data suggest that, although M-CSF does not appear to play a critical role in determining the macrophage content of these tumors, it does play a role in modulating the phenotype, and potentially the functional activity of the macrophages present within the tumor microenvironment.

Diverse effects of anti-CD44 antibodies on the stromal cell-mediated support of normal but not leukaemic (CML) haemopoiesis in vitro.

We have identified three non-cross-reacting anti-human CD44 monoclonal antibodies that have significant positive or negative (or no) effects on normal human haemopoiesis in the long-term culture (LTC) system. These effects manifested as increases or decreases in the number of LTC-initiating cells (LTC-IC), and the number of colony-forming cells (CFC) recovered from cultures in which either unseparated or highly purified CD34+ CD38- normal marrow cells were placed on pre-established normal marrow feeder layers in the presence or absence of each antibody. The effects seen were rapid and sustained, and dependent on the presence of a preformed feeder layer. Interestingly, the same anti-CD44 antibodies had no effect on the maintenance of leukaemic (Ph+) progenitors (from patients with chronic myeloid leukaemia) when these cells were cultured on preformed feeder layers established from normal marrow. CD44 appears to be part of a mechanism by which stromal elements can regulate primitive normal haemopoietic cells but not their leukaemic (Ph+) counterparts.

Molecular mechanisms regulating TNF-alpha production by tumor-associated macrophages.

The molecular mechanisms that regulate the production and/or functional activity of intratumoral tumor necrosis factor-alpha (TNF-alpha) remain poorly defined. To begin to address this issue we have examined the level of TNF-alpha mRNA and protein produced by macrophages present within immunogenic Fsa-R and non-immunogenic Fsa-N tumors grown in syngeneic Lps(d) C3H/HeJ and Lps(n) C3H/HeN mice. The results obtained indicate that macrophages isolated from tumors grown in Lps(d) C3H/HeJ mice express 5-10-fold less TNF-alpha than equivalent cells present in tumors grown in Lps(n) C3H/HeN mice. These data suggest that the mechanisms that operate within the tumor microenvironment to induce the production of TNF-alpha act, at least in part, via the same signal transduction pathway that is defective in Lps(d) C3H/HeJ mice. Interestingly, despite such differences in TNF-alpha production, tumors inoculated into C3H/HeJ and C3H/HeN mice grew at a similar rate and contained an almost identical proportion of macrophages. Moreover, tumor cells purified from tumors grown in C3H/HeJ and C3H/HeN mice produced similar quantities of the TNF-alpha-inducible cytokine GM-CSF. Thus, although differences in the level of TNF-alpha produced within tumors grown in C3H/HeN and C3H/HeJ mice are readily demonstrable, such differences appear to have little direct impact on the outcome of tumor growth.

Targeting gene therapy to cancer: a review.

In recent years the idea of using gene therapy as a modality in the treatment of diseases other than genetically inherited, monogenic disorders has taken root. This is particularly obvious in the field of oncology where currently more than 100 clinical trials have been approved worldwide. This report will summarize some of the exciting progress that has recently been made with respect to both targeting the delivery of potentially therapeutic genes to tumor sites and regulating their expression within the tumor microenvironment. In order to specifically target malignant cells while at the same time sparing normal tissue, cancer gene therapy will need to combine highly selective gene delivery with highly specific gene expression, specific gene product activity, and, possibly, specific drug activation. Although the efficient delivery of DNA to tumor sites remains a formidable task, progress has been made in recent years using both viral (retrovirus, adenovirus, adeno-associated virus) and nonviral (liposomes, gene gun, injection) methods. In this report emphasis will be placed on targeted rather than high-efficiency delivery, although those would need to be combined in the future for effective therapy. To date delivery has been targeted to tumor-specific and tissue-specific antigens, such as epithelial growth factor receptor, c-kit receptor, and folate receptor, and these will be described in some detail. To increase specificity and safety of gene therapy further, the expression of the therapeutic gene needs to be tightly controlled within the target tissue. Targeted gene expression has been analyzed using tissue-specific promoters (breast-, prostate-, and melanoma-specific promoters) and disease-specific promoters (carcinoembryonic antigen, HER-2/neu, Myc-Max response elements, DF3/MUC). Alternatively, expression could be regulated externally with the use of radiation-induced promoters or tetracycline-responsive elements. Another novel possibility that will be discussed is the regulation of therapeutic gene products by tumor-specific gene splicing. Gene expression could also be targeted at conditions specific to the tumor microenvironment, such as glucose deprivation and hypoxia. We have concentrated on hypoxia-targeted gene expression and this report will discuss our progress in detail. Chronic hypoxia occurs in tissue that is more than 100-200 microns away from a functional blood supply. In solid tumors hypoxia is widespread both because cancer cells are more prolific than the invading endothelial cells that make up the blood vessels and because the newly formed blood supply is disorganized. Measurements of oxygen partial pressure in patients' tumors showed a high percentage of severe hypoxia readings (less than 2.5 mmHg), readings not seen in normal tissue. This is a major problem in the treatment of cancer, because hypoxic cells are resistant to radiotherapy and often to chemotherapy. However, severe hypoxia is also a physiological condition specific to tumors, which makes it a potentially exploitable target. We have utilized hypoxia response elements (HRE) derived from the oxygen-regulated phosphoglycerate kinase gene to control gene expression in human tumor cells in vitro and in experimental tumors. The list of genes that have been considered for use in the treatment of cancer is extensive. It includes cytokines and costimulatory cell surface molecules intended to induce an effective systemic immune response against tumor antigens that would not otherwise develop. Other inventive strategies include the use of internally expressed antibodies to target oncogenic proteins (intrabodies) and the use of antisense technology (antisense oligonucleotides, antigenes, and ribozymes). This report will concentrate more on novel genes encoding prodrug activating enzymes, so-called suicide genes (Herpes simplex virus thymidine kinase, Escherichia coli nitroreductase, E. (ABSTRACT TRUNCATED)

The role of host lymphoid populations in the response of mouse EMT6 tumor to photodynamic therapy.

Photodynamic therapy (PDT) treatment of murine EMT6 mammary sarcoma using Photofrin (10 mg/kg) and light (110 J/cm2) cured all these lesions growing in syngeneic BALB/c mice. In contrast, the same treatment produced initial ablation but no long-term cures of EMT6 tumors growing in either scid or nude mice, the immunodeficient strains sharing the same genetic background with BALB/c mice. No difference was detected in either the level of Photofrin accumulated per g of tumor tissue or the extent of tumor cell killing during the first 24 h after PDT of EMT6 tumors growing in BALB/c or scid mice. The assumption that the difference in tumor cures could be ascribed to the absence of functional lymphoid cells in scid and nude mice was supported by the results of experiments involving the adoptive T-cell transfer into scid mice or bone marrow transfer between BALB/c and scid mice. The adoptive transfer of splenic virgin T lymphocytes from BALB/c mice into scid mice performed 9 days before PDT of EMT6 tumors growing in the recipients was successful in delaying the recurrence of treated tumors. Adoptive transfer done immediately after PDT or 7 days after PDT had no obvious benefit. Even better improvement and a high cure rate of PDT-treated tumors was obtained with scid mice reconstituted with BALB/c bone marrow. In contrast, a marked drop in tumor cure rate was observed with BALB/c mice reconstituted with scid bone marrow. These results suggest that the activity of host lymphoid populations was essential for preventing the recurrence of EMT6 tumors following the PDT treatment used in this study. The contribution of PDT-induced immune reaction may, therefore, be of critical importance for the cure with at least some tumors.

Altered patterns of CD44 epitope expression in human chronic and acute myeloid leukemia.

Abnormal expression of different isoforms of CD44 has been found to characterize many types of malignant cells although data for human acute and chronic myeloid leukemia is limited. In this study, we have identified significant, albeit variable, increases in these diseases of the frequency of both light density and CD34+ cells expressing two particular CD44 epitopes, neither of which is commonly found on normal human marrow cells. One of these epitopes is unique to exon v10-containing isoforms of CD44. The other is located in the common region of CD44 and was previously revealed on T cells only after their activation. Interestingly, another T cell activation-associated epitope was found to be expressed on a high proportion of normal marrow cells including the CD34+ subset and this remained the case for most of the primary leukemic samples evaluated. As expected, >90% of cells in all primary normal and leukemic samples expressed high levels of CD44, as shown by their reactivity with an antibody specific for the CD44 hyaluronan-binding site. To begin investigating how expression of the CD44 epitopes seen more commonly on leukemic than on normal CD34+ cells may be modulated, and to identify potentially associated effects on the hyaluronan-binding ability of the CD44 expressed, the effect of phorbol ester treatment on these properties of CD44 were examined. For these studies, a panel of five different human leukemic cell lines that were found to exhibit different patterns of CD44 expression and function in the absence of phorbol ester were used. Both the level and the hyaluronan-binding properties of CD44 could be stimulated in some, but not all, of these leukemic cell lines. Taken together, our findings indicate that CD44 expression is perturbed in a variety of leukemic populations suggesting a possible relationship to some of the pathogenetic features they share.