As little as five years ago, the idea that it would be possible to generate an ovine model of cystic fibrosis (CF) was frequently met with humour and disbelief. Fortunately for those of us attempting to produce a large animal model of this debilitating disease, the generation of cloned sheep (1 ,2 ) changed all of that. The case must now be made for the production of an ovine model for CF. The justification may be applied to large animal models of many other human genetic diseases. Recent data showing that the cloning technology is applicable to other species, such as the cow (3 ), provide further support for the feasibility of this approach.
The use of animal models of human disease has been fundamental to elucidating the pathological mechanisms of many disorders. A number of well characterized, naturally occurring animal models of human genetic diseases exist. In addition, the relative ease and efficiency of the mouse `gene knockout' technology has transformed the possibilities of generating animal models of specific monogenic disorders. However, mouse knockout models vary in their similarity to human disease. A number of strains of CF mice have been generated that carry different mutations (4 -13 ), but none of them have cystic fibrosis, a disease primarily characterised by recurrent lung infection and pancreatic destruction. CF mice have normal pancreatic function and little airway disease in a limited life span. Instead, the majority of strains die from intestinal obstruction. There is no doubt that the CF mice have been extremely useful in addressing a number of questions relating to CF disease mechanisms; however, their limitations are now restricting progress.
It is not surprising that CF mice do not have the same disease as humans with CF since they show differences in anatomy and physiology that are pertinent to lung and pancreatic function. Some aspects of these species differences were apparent before the CF mice were generated, others have been identified through physiological and molecular analysis of CF mice.
The precise mechanisms of the evolution of CF lung disease in humans remain controversial though submucosal glands may play a central role. The anatomy of the human and mouse lungs are divergent, particularly with respect to submucosal glands. Mice have very few of these glands and they are restricted to the tracheal submucosa. In humans abundant submucosal glands are associated with trachea and bronchi. In human fetal lung the cystic fibrosis transmembrane conductance regulator gene (CFTR) mRNA is seen throughout the respiratory epithelium (14 ). Postnatally human CFTR protein is expressed at very low levels in the surface epithelium of the airways and only at higher levels in the serous portion of submucosal glands (15 -17 ). Lung pathology in CF is associated with failure to clear mucous secretions from the airways that encourages the growth of certain pathogenic bacteria. Though a number of mucins (MUC2,4 and 5AC) are produced in the airway epithelium, expression of one mucin, MUC5B, is restricted to the mucous cells of the submucosal gland (18 ). Though other mechanisms may be defective in the CF lung disease process, including aspects of innate immunity (19 ,20 ), the differences in submucosal gland distribution may be crucial to the outcome of CF in mouse and human. There may also be significant differences in the array of ion channels that are expressed in the human and rodent airway epithelium.
There is no pathology in CF mouse pancreas. In contrast, complete pancreatic destruction, following obstruction of the pancreatic ducts in utero, is seen in CF humans. Analysis of the electrophysiology of the mouse pancreas has shown that the spectrum of chloride ion channels expressed in the rodent pancreas is different from that seen in human (21 ). The CFTR chloride ion channel is not expressed at high levels in the mouse pancreas, in contrast to humans where the pancreas is the site of most abundant CFTR expression (15 ). Further, the mouse pancreatic ducts express high levels of a calcium-activated chloride ion channel that might partially compensate for the presence of a mutant CFTR channel (21 ).
At the molecular level, we have shown that the ovine CFTR cDNA shows a high degree of conservation at the DNA coding and predicted polypeptide levels with human CFTR, much greater than the similarity between mouse and human CFTR (22 ). The sheep and human CFTR protein show 90.8% identity and 95.3% similarity when conservative amino acid differences are considered, compared with figures of 77.7 and 88.7%, respectively, between human and mouse. We have shown that the tissue specific patterns of expression of the ovine CFTR gene are very similar to those seen in humans. Further, the developmental expression of CFTR in the sheep is equivalent to that observed in humans (22 ).
At the physiological level, the sheep lung epithelium has been the subject of intense investigation for many years, both during development and postnatally (23 -26 ). There is now a substantial body of data that shows close anatomical, functional and electrophysiological similarities between the ovine and human lung. Being a herbivore, the sheep has a digestive system that diverges from that seen in humans, but CFTR expression levels in the ovine pancreas equate to those seen in humans (22 ). Disruption of the ovine CFTR gene would be predicted to have a fundamental effect on ovine pancreatic physiology. At least for the two organ systems for which the CF mouse is an inadequate model of CF disease, the CF sheep is likely to be much more useful.
Treatment of CF lung disease is an important goal of current CF research programmes. A major investment in gene therapy is underway. Sheep are useful to evaluate the techniques of gene delivery, followed by bronchoscopy to investigate the efficacy and safety of gene transfer procedures. The advantages of an ovine CF model are also applicable to the assessment of pharmacological approaches to correct CF lung disease. An ovine model of CF would increase the speed at which gene therapy and other clinical trials could be carried out. Some investigations which could not be carried out in humans would become feasible. A further important advantage of sheep is their longevity in comparison with rodents, which would enable assessment of long term therapeutic procedures. One additional aspect of an ovine model of CF that currently receives relatively little attention, but may well become crucial at some stage in the future, relates to the early development of the disease. It is known that CF pathology commences in utero (27 ), so intra-uterine intervention may be called for to treat this disease effectively. For example, the obstruction of CF pancreatic ducts by deposits of secreted material commences in the mid-trimester of human gestation and by term the pancreas is structurally and functionally destroyed (28 ). Availability of an ovine model of CF would enable the investigation of this and other aspects of the early disease process which are impossible to study in humans.
The novel cloning technology will enable the production of human genetic disease models in a variety of different animal species. The relevance of the model to the disease will depend on the particular site of disease pathology, the anatomical and physiological similarity of the animal and humans and the degree of genetic divergence. In the case of CF one would propose, on the basis of phylogenetic comparisons of the CFTR locus (29 ), that monkeys, cows, sheep, pigs and possibly rabbits might be useful. Practical and ethical considerations lead us to propose a rank list with sheep and cows at the top, though projects to generate a CF monkey and a CF sheep are also underway (Wine, Renard and Denamur, personal communication).
Two main routes are being investigated to evaluate this. One is to look for naturally occurring CFTR mutants, existing in heterozygote form in sheep populations and then to breed from them. This approach, which is being pioneered by Tebbutt and colleagues in New Zealand (30 ) is clearly labour intensive and high risk, since if there were no heterozygote advantage for the mutation, the gene had a standard mutation rate and it was lethal in homozygotes, the number of DNA samples that would have to be screened to find one mutation might be in the order of 10 000. Since the CFTR gene has 27 exons, an exon-specific rapid DNA based screening method such as SSCP might be impractical. However, given the element of luck involved in this approach, it might still be successful and a potential CFTR mutation has already been identified (30 ).
Another approach is to use gene targeting strategies in the sheep and this route may be quite rapid. The resources for generating ovine CFTR targeting constructs are available (22 ). There will be hurdles to overcome: the efficiency of achieving homologous recombination in the fetal sheep cells that will be used for the cloning experiments; the longevity of these cells in culture and their ability to maintain pluripotency at G0 of the cell cycle may be a limiting factor.
The question of which CFTR mutant to generate is answered more readily. Since 70% of northern European mutations in the CFTR gene involve the deletion of a phenylalanine residue at amino acid 508 (n F508), a n F508 sheep would be useful. This mutation affects CFTR processing, and a small amount of normal CFTR protein does reach the cell surface to function as a chloride ion channel, thus exploitation of a n F508 sheep model for testing therapeutic strategies would be valuable. Five classes of mutation have been described in the CFTR gene based on how they affect transcription, translation and processing of CFTR mRNA or in what way they disable the function of the CFTR protein as a cAMP-activated chloride ion channel (31 ,32 ). n F508 is a class II mutation; an additional useful ovine model would be of a class I null mutation, resulting in the complete abolition of CFTR protein production. Given the remarkable flexibility in the splicing machinery of the CFTR gene (33 ) a stop mutation in a 5' exon would be preferable, such as have already been described for three strains of CF mouse (7 ,8 ,13 ).
The genetic background into which the CFTR mutation is introduced may be relevant, though at this stage it is unclear whether, as in the case of knockout CF mice (13 ), this will affect the phenotype. Most genetic information on the ovine CFTR locus has been derived from Romney sheep while the sheep cloned to date were derived from Finn Dorset or Poll Dorset cells; however, none of these strains of sheep are homogenous.
The ethics of the new cloning technology remain controversial and will not be discussed here. The main point to be considered in the context of a cloned CF sheep is the rationale for generating a sick animal, assuming here that the CF sheep will exhibit pathology similar to the human disease. Herein lies the main justification. Despite the fact that the projected average life-span of a baby born today with CF is in the region of 30-40 years, many babies and children with severe disease still die at a young age. CF requires life-long and intensive treatment (including daily physiotherapy routines, frequent courses of antibiotics and pancreatic enzyme supplements) resulting in a major reduction in quality of life. An ovine model of CF would enable the direct testing of therapeutic agents. There is no doubt that this would accelerate the development of efficient treatment protocols. The CF market is generally regarded as too small to be of economic importance and hence to warrant major investment by pharmaceutical companies. However, the availability of a relevant, robust and inexpensive animal model might be an inducement for establishing further industrial/academic collaborations.
As to the CF sheep themselves, since the animals would be maintained as heterozygotes, and the carrier state (at least in humans) is not associated with any phenotype, these animals would be perfectly normal. CF is an autosomal recessive disease so breeding carrier rams and ewes would produce CF homozygotes at a frequency of 1 in 4 offspring and a relatively small number of animals would have CF disease. Until CF sheep are produced we cannot predict the precise phenotype and the degree of distress caused to the animals. It is probable that symptoms might be alleviated by therapeutic approaches commonly used on humans. If other routes were available that would yield efficient treatment for CF disease in a short time frame and not involve animal models these would obviously be the avenues of choice. However, efficient CF gene therapy is still a distant prospect and so we must take advantage of all technological advances to attempt to treat this severe and often lethal disease.
In the current climate of research support this question is particularly relevant. In comparison with laboratory mice, sheep are expensive to generate (one sheep costing about the same as 20 mice) and maintain (the weekly cost being about six times that for a mouse). Ideally this project should be carried out in New Zealand, where the cost of sheep is substantially lower. However, the practical value of an ovine model of CF would well justify the expenditure. The project is ideally suited to an integrated programme of collaboration, bringing together expertise from different research groups in all relevant areas from molecular genetics through CF treatment to animal husbandry. It would be a flagship for the application of the new cloning technology and its uses in the investigation, treatment and potential cure of human disease.
Human Molecular Genetics
Pages
Introduction
Why Did The CF Mice Not Live Up To Expectation?
Why Should CF Sheep Be Any Better?
Other Advantages Of A CF Sheep
Would Any Other Animals Provide Better CF Models?
Will It Be Possible To Generate A CF Sheep?
Should We Generate A Ovine Model Of CF?
Can We Afford To Generate And Maintain An Ovine Model Of CF?
References
REFERENCES
This page is maintained by OUP admin. Last updated Sat Oct 18 13:29:50 BST 1997. Part of the OUP Journals World Wide Web service.
Copyright
Oxford University Press, 1997