Is pituitary gene therapy realistic?
Abstract
Current therapies for pituitary tumours are moderately successful in many cases but still suffer from significant limitations, with relatively poor long-term rates of endocrine cure from surgery, and long-term hypopituitarism after radiotherapy. Even in the case of the most readily treatable tumours, prolactinomas, medical therapy with dopamine agonists is limited by lack of response or side-effects in up to 10% of patients. This has led to increasing interest in the application of our knowledge of pituitary cell and molecular biology to evaluate the potential of gene therapy. Various vectors are available to facilitate gene delivery, and recombinant adenoviruses have been studied in detail because of their ability to transduce the postmitotic, nondividing cells of the pituitary gland. Various studies with reporter genes such as β-galactosidase have demonstrated high efficiency and long lasting expression of adenoviral transgenes in cultured pituitary cells in vitro. The feasibility of high level transgene expression has also been shown in vivo, but so far this requires stereotaxic intrapituitary injection to achieve adequate transduction. Ablation of pituitary cells has been demonstrated in cultured cell lines and in subcutaneous tumours in nude mice, though alternative animal models will be required to evaluate efficacy in more slowly proliferating tumours as found in man. Inflammatory responses have been documented in the pituitary gland as in other tissues, and this will require the evaluation of modified vectors to avoid significant adverse effects before human applications can be considered. In summary, gene therapy for pituitary disease is likely to be feasible in the future, but will require careful and extensive evaluation of efficacy and safety, using a variety of possible methods of gene delivery.
Introduction
The prospect of endocrine gene therapy has aroused great excitement in recent years as a way of applying our new knowledge of molecular biology to develop clearly targeted molecular or cellular treatment. ‘Gene therapy’ simply means therapy with nucleic acids instead of proteins, and has become possible because of improved methods for gene transfer in intact organisms. Gene transfer into cultured cell lines (transfection) has been established for many years, but a number of tools are now available that are much more efficient, particularly using genetically engineered (recombinant) viruses, and can be used in living tissues, not just cell culture. These new techniques of gene transfer in vivo have captured the imagination of clinical endocrinologists trying to improve tumour treatment and peptide hormone replacement therapy, but they also offer new approaches to the physiologist wishing to study or disrupt cell function without needing to create transgenic animals.
A number of recent reviews have introduced the general concepts behind pituitary or endocrine gene therapy and the vectors available (Davis, 1999; Stone et al., 1999,Kaiser, 2001), the reader is referred to these for more detailed information. Here, we will review all of the published studies of pituitary gene transfer in vitro and in vivo, and indicate what development and evaluation will be needed for gene therapy to be realistically applicable to pituitary disease in man.
Is pituitary disease a suitable target?
At first sight, it might appear unnecessary to consider gene therapy for the treatment of pituitary tumours. They are relatively benign and slow growing, hardly ever directly fatal, and are regularly treated with combinations of medical therapies that can normalize hormone hypersecretion. Clinical trials of gene therapy have so far addressed diseases for which there is no effective treatment, such as cancers or untreatable, fatal monogenic disease.
We have argued elsewhere, however, that treatment of pituitary disease is not as good as it might be, all of the current treatments have significant shortcomings (Davis et al., 1999). The results of traditional attempts at surgical extirpation have not improved significantly in the past 20 years, despite important technical advances such as intraoperative MR scanning and endoscopy, and still rely heavily on the expertise of the operator. Surgical therapy is curative in only 80–85% of patients with the smallest tumours in the very best series, and many series report much lower success rates (Soule et al., 1996; Lissett et al., 1998). Medical therapy is remarkably successful, but even dopamine agonist treatment for prolactinomas is ‘curative’ in only 85–90%, and commonly requires long-term treatment with resultant long-term side-effects (Bevan et al., 1992). Radiotherapy, probably including gamma knife treatment, is effective at preventing tumour recurrence, but causes progressive long-term hypopituitarism (Littley et al., 1991; Plowman, 1999). In addition to the shortcomings of conventional pituitary therapy, it is clear that at least some pituitary disease (for example acromegaly and hypopituitarism) is associated with excess mortality (Orme et al., 1998; Tomlinson et al. 2001). Thus, there is a case for developing better therapy for at least some patients with pituitary adenomas, and certainly for the rarer pituitary carcinomas, which have disappointing responses even to chemotherapy.
The pituitary gland is an attractive target for development of new ‘molecular’ therapies for another important reason, in that the molecular and cell biology of the pituitary gland is now well understood. Much has been learned recently about the molecular biology of pituitary cell type differentiation and gene expression (Dasen & Rosenfeld 1999). Although the molecular pathogenesis of tumours is still not well understood, important progress has been made in identifying some potential mechanisms which may indicate possible targets for future therapy (Farrell & Clayton, 2000; Heaney & Melmed, 2000). In addition, there is a variety of readily available cell systems and in vivo models for studying pituitary physiology.
It therefore seems timely to start exploring the potential for therapeutic gene transfer in pituitary tumours, though safety issues will be paramount in considering novel transgenic treatments for patients whose clinical prognosis is already very good in most cases. In the foreseeable future, gene therapy is unlikely to be considered for small pituitary microadenomas, but may plausibly find a role for large aggressive tumours or for locally invasive postsurgical residual disease that is resistant to existing medical treatment. For the occasional patients with pituitary carcinomas too, conventional therapy and chemotherapy have limited efficacy, and effective gene therapy would be a welcome hope.
Gene transfer into pituitary cells: recombinant adenovirus as a suitable vector
A number of issues should be considered in choosing a suitable technique for gene transfer into the pituitary. These include: the efficiency of gene transfer, the capacity of the vector for integrated transgenes, the potential cytotoxicity and immunogenicity of the vector, and whether or not the vector DNA becomes integrated into the host genome. In principle, naked plasmid DNA can be used for gene transfer (Horton et al., 1999), and prolonged local delivery of transgene products can be achieved using ‘gene-activated matrices’ (Bonadio et al., 1999). Efficiency can be increased in some situations using polyplex or lipoplex DNA conjugates, for example neuronal cells have been successfully transfected using cholera toxin β-chain-poly(D-lysine)/β-galactosidase (Barrett et al., 1999) but the efficiency is still relatively low. Somatic cells can be engineered ex vivo and re-injected into the patient in order to re-express deficient proteins (Bailey et al., 1999), but this is not appropriate for ablation strategies. A series of viral vectors have therefore been explored as much more efficient tools (for a detailed review see Stone et al., 1999).
Recombinant adenoviruses have so far appeared the most appropriate choice for applications to pituitary cells, particularly because of their high efficiency in infecting nondividing cells (Wilson, 1996). The adenoviral genome has been studied for many years, and is a linear double-stranded DNA molecule of about 36kbp. The genes include four ‘early’ genes and five ‘late’ genes (named from their expression patterns during viral replication), which encode a series of proteins required for transcription, host cell growth, DNA replication and viral particle assembly, among other functions. Recombinant adenoviruses can be engineered by deleting critical genes that are required for replication, usually most of the E1A and E1B regions and the E3 region, such that the virus is able to replicate only if the E1A proteins are provided by a host cell. The human embryonic kidney 293 cell line was originally produced by adenoviral infection and this is used to provide the missing E1A protein for propagation of recombinant adenoviruses in the laboratory. Adenoviruses have sufficient cloning capacity to allow the incorporation of regulatable elements such that transgene expression can be switched on or off, for example by the tetracycline-regulatable system (Harding et al., 1998). Newer generations of adenovirus are now being studied in which much more extensive deletions have been made. These ‘gutless’ adenoviruses have a much larger cloning capacity, and may be less immunogenic in some circumstances (Parks et al., 1996; Thomas et al. 2000).
Other viral vectors that may be suitable candidates for pituitary gene therapy include adeno-associated virus (AAV) and lentivirus. AAV is a parvovirus that can integrate into the host genome, but can also infect nondividing cells (Xiao et al., 1999), however, its applications may be limited by its small cloning capacity. The HIV-based lentivirus also transduces postmitotic cells and has sufficient capacity to allow inducible transgene constructs to be accommodated (Kafri et al. 2000; Lever, 2000). This may prove to be a good alternative to adenovirus in the future.
Evaluation of gene therapy in vitro: pituitary cell culture
Preliminary studies of adenoviruses as tools for gene transfer have naturally taken place in cultured pituitary cells. Investigators have utilized both rat pituitary cell lines (for example mammosomatotrophic GH3 cells and corticotrophic AtT20 cells) and primary cultures of normal rat pituitary glands. Cell lines have proved remarkably robust and reproducible, while primary cultures are essential for comparisons of the different endocrine cell types in the mixed population of the normal pituitary.
The first reported study of adenoviral infection of pituitary cells described the expression of tyrosine hydroxylase in primary cultures of human lactotroph adenoma cells using a recombinant adenovirus (Freese et al., 1996). This achieved increased production of dopamine and reduced the prolactin secretion by the prolactinoma cells.
A detailed study by Castro and colleagues reported the characteristics of adenoviral infection of each of the cell types in normal pituitary cell cultures (Castro et al., 1997). The β-galactosidase reporter gene was expressed under the control of either the cytomegalovirus (CMV) or the Rous sarcoma virus (RSV) promoters within the recombinant adenovirus genome, and was efficiently expressed in normal endocrine cells for over two weeks. Neill and colleagues confirmed the efficacy of protein transduction in normal rat gonadotroph cells, using adenoviral gene transfer in vitro to over-express G-protein coupled receptor kinases and decrease GnRH-stimulated LH secretion (Neill et al., 1999).
Evaluation of pituitary gene therapy in vivo: animal models
Normal pituitary gland in vivo
The traditional first choice for animal studies is the rat, and initial studies have used direct stereotactic or transauricular injection into the rat pituitary. Marker gene expression studies have confirmed that highly efficient gene transfer can be achieved by recombinant adenoviruses into all endocrine cell types of the normal pituitary gland in vivo, similar to earlier results in vitro (Windeatt et al. 2000). Intracarotid injection would be an attractive alternative, but unfortunately the one study of this route demonstrated very low levels of β-galactosidase expression compared to direct intrapituitary injection (Lee et al. 2000), implying that future gene therapy applications might require intratumoral injection at the time of pituitary surgery to achieve sufficiently high transgene expression.
There are limitations to the use of rodents for evaluation, consequently we have recently evaluated adenoviral expression of marker genes in the normal sheep pituitary gland. The sheep is attractive as the pituitary gland is large and similar in configuration to the human pituitary gland, allowing a realistic estimate of the extent of regional expression of adenoviral transgenes after direct injection. The animal is large enough to take serial blood samples for longitudinal evaluation of endocrine function, and there are abundant normative data concerning normal patterns of hormone secretion. Studies of adenoviral β-galactosidase marker gene expression have revealed that substantial regions of this large pituitary gland can be effectively transduced by stereotaxic transcranial injection of adenoviruses, and that tight cell-type specificity can be achieved using a pituitary hormone gene promoter. The intrapituitary injection of virus caused no apparent disruption of normal pituitary secretory function over the following seven days (Davis et al. 2001).
In vivo models of pituitary tumours
The architecture of the normal pituitary, with its extensive network of sinusoids and fenestrated capillaries differs markedly from that of adenomas, so that patterns of virus dissemination may be different. Hence it will be important to study real pituitary tumours in vivo, though there are a number of drawbacks to existing model systems that must be borne in mind in interpreting the results.
Studies that have aimed at pituitary tumour ablation have so far used the conditional cytotoxin thymidine kinase (TK) derived from the herpes simplex virus HSV-1: this gene product acts by mono-phosphorylating nucleoside analogues such as ganciclovir, which are then further phosphorylated by celullar kinases to triphosphate metabolites and incorporated into replicating DNA, leading to apoptosis. For example, transplantable pituitary tumours may be passaged subcutaneously or under the kidney capsule in rodents. They grow rapidly, and growth rate can be monitored over a few weeks. Recent studies have shown that the growth of subcutaneous tumours grown in nude mice is dramatically reduced by adenoviral cytotoxic gene expression when the animals were fed ganciclovir (Lee et al., 1999). The problem with this approach and model, though, is that these tumours are probably too aggressive: the shrinkage that can be achieved is dramatic, but these tumours grow much more rapidly than human adenomas and are therefore likely to be especially susceptible to an ablation technique that relies on proliferation to achieve cell death. Human pituitary adenomas, by contrast, are relatively indolent with few mitotic figures, and may be more resistant to ablation by TK expression as it relies on a high rate of DNA replication to induce sufficient apoptotic cell death.
Hormonally induced pituitary hyperplasia (Guo et al., 1997) may be a better model. Adenoviral TK expression again has been used to reduce the lactotroph hyperplasia induced by oestrogen and sulpiride treatment (Southgate et al. 2000). A subsequent study using the same system found that the human prolactin promoter achieved cell-type specific TK expression in vivo, but was significantly less powerful than the CMV promoter and failed to induce significant reduction in induced hyperplasia (Windeatt et al. 2000). Although the response with the CMV-TK transgene was encouraging, the response probably depends – as in the nude mouse tumour model – on a higher rate of cell proliferation than is normally seen in human adenomas, and it remains uncertain whether TK will prove a suitable ablative gene. Thus the use of cytotoxins that require active cell division to induce cell death, such as TK, is unlikely to be effective for the majority of even the most aggressive adenomas and direct cytotoxins may therefore be more appropriate.
Further work with other tumour models may also help in choosing the best candidate ablation strategies, and the spontaneously arising tumours may be an appropriate but more challenging target. Indeed the first in vivo study of pituitary tumour ablation treatment was reported in tumour-prone retinoblastoma gene (Rb) +/− transgenic mice. These animals developed spontaneous melanotroph pituitary tumours whose growth was inhibited after injection with recombinant adenovirus expressing Rb cDNA (Riley et al., 1996), implying that restoration of defective tumour suppressor gene expression may be a viable future strategy.
Targeting gene therapy
Ideally, given that the pituitary gland comprises a range of intermingled cell types, a gene therapy strategy would aim to target a transgene to the specific cell type that constituted the adenoma. This would avoid the risk of collateral ablation of the various other neighbouring cell types and reduce the risk of hypopituitarism, which limits the value of other destructive therapies such as radiotherapy. Cell-type-specificity can be achieved by ‘transcriptional targeting’– by arranging that the expression of the transgene is controlled by pituitary hormone gene promoters, such as GH and prolactin. A series of studies in vitro and in vivo have demonstrated that cell-type-specificity can be achieved in this way (Lee et al., 1999; Southgate et al. 2000; Davis et al. 2001). However, in general the pituitary-specific promoters are less powerful than the commonly used viral promoters, and it remains to be seen whether this will be a significant limitation in achieving high levels of transgene expression.
Vectors can also be targeted to specific cell types by expression of a ligand that binds to specific receptors on the target cell surface: thus bacteriophage and adenovirus vectors can be engineered to targeting FGF receptors, amongst other relevant targets (Kassner et al.; 1999; Doukas et al., 1999; Gu et al., 1999). This has not yet been explored in the pituitary gland, but clearly has potential as at least some of the receptors that could be targeted are limited to few cell types, such as GnRH receptors on gonadotrophs or CRH receptors on corticotrophs.
Side-effects of adenoviral gene therapy
In view of the fact that pituitary disease is relatively benign in most cases, the potential adverse effects of gene therapy need to be considered extremely carefully. So far, only adenoviral vectors have been investigated in any detail. They are certainly effective vectors, but concerns have been raised about their safety, because of the immune response that can be generated. For several years it has been known that adenovirus injection into various tissues could result in an inflammatory response (for example, Byrnes et al., 1995; Chan et al., 1999), though this has often been modest and has been regarded as a problem only to the extent that it limited transgene expression. In 1999, the first death that was directly attributable to the viral vector was reported of a patient in a gene therapy trial (see Marshall, 2000; for an overview of the issues). In this case the virus had been injected into the circulation, but even so it was found at post-mortem to have disseminated surprisingly widely, and this may have contributed to the devastating multisystem failure (Marshall, 1999).
Inflammatory reactions within the pituitary gland in an otherwise healthy patient could be severe, and any local pituitary swelling could threaten the optic chiasm. Histological studies of the sheep pituitary gland after adenovirus injection do demonstrate significant histological evidence of an acute inflammatory reaction, with marked perivascular lymphocytic infiltration and tissue necrosis and fibrosis occurring within 7 days (Davis et al., in preparation). Studies in the rat have indicated that the inflammatory reaction peaked at 14 days and then subsided, but macrophage, T-cell and NK cell infiltration persisted for up to one month (Southgate et al. 2001). The pattern and severity of these responses may vary according to species and are probably dose-related, but the evidence so far raises significant concerns that adenoviruses may be too immunogenic to be suitable for clinical evaluation in the context of pituitary disease.
Prospects for pituitary gene therapy
There are significant issues to be addressed before gene therapy can be seen as a realistic clinical option for most forms of pituitary disease. However, there is a serious clinical problem that merits investment in developing new therapeutic strategies, and vectors are under trial for a number of other indications. These clinical trials, together with careful evaluation of suitable animal models of endocrine tumours, seem certain to develop important new opportunities for patients with aggressive pituitary disease. We should be prepared for relatively long lead times before we see clinical trials of pituitary gene therapy, but progress to date has been rapid. Systemic injection of vectors so far seems unlikely to generate high enough transgene expression in the pituitary, so at present it seems that direct surgical approaches will be needed, either into the body of a tumour or by injection into the bed of a tumour remnant, to minimize the dose of vector needed.
Gene therapy also seems likely to make an impact on understanding endocrine physiology: adenovirus vectors are highly efficient tools with which to transfer genes into primary pituitary cell cultures or the pituitary gland in vivo. Apart from studies of gene expression using marker genes such as luciferase or β-galactosidase, endocrine function can be manipulated by using toxin genes, blockade of oncogenes or over expression of tumour suppressor genes, or controlled expression of normal or mutant receptors and antiangiogenic factors, among many other potential applications. It will be essential to ensure that adenoviral transgenes are expressed for a sufficient duration – in general this appears not to be a problem if adequate amounts of cell death are achieved rapidly, but in some cases sustained – even lifelong – expression of transgenes would be required. For example, targeted expression of noncytotoxic genes such as tumour suppressor genes should in principle prevent tumour growth, but may not actually ablate a tumour, and therefore would need to be very prolonged to be therapeutically useful.
In summary, the revolution in molecular biology is now poised to make an impact on therapeutics and physiological investigation. Gene therapy is almost certain to become part of clinical practice in oncology in the next few years, though we should not expect an immediate revolution in endocrinology. A perspective is given by the gradual evolution of antibiotic therapy and chemotherapy over the past 50 years. The choice of vectors, of transgenes, of promoters and of delivery all require exploration, and the choice and efficacy will differ according to the tissue in question. Nonetheless, there is great enthusiasm to try to reap therapeutic benefits from the recent advances in pituitary cell and molecular biology, and we should expect a large number of exploratory studies over the coming five years (Kaiser, 2001). Advances in biology have allowed dramatic changes in the treatment of breast and prostate cancer. Treatment with dopamine agonists, somatostatin analogues and growth hormone antagonists has already offered the possibility of purely endocrine therapy for some selected patients. During the 21st century pituitary tumour treatment should be in a position to evolve much further.
Acknowledgements
We are grateful to the MRC and BBSRC for financial support.
