WO1994004669A1 - Mice having cftr defect as model for cystic fibrosis - Google Patents

Abstract

A mammal other than man having a CF-related defect in at least one copy of C &cir& _f &cir& _t &cir& _r &cir& _.

Description

MICE HAVING CFTR DEFECT AS MODEL FOR CYSTIC FIBROSIS

The present invention relates to mice carrying a genetic defect and to their use in elucidating the causes, consequences therapy and prevention of cystic fibrosis in humans.

Cystic fibrosis is the most common fatal autosomal recessive disorder in Caucasian populations with approximately 1 in 2,500 affected by the disease and a carrier frequency of about 1 in 22. The disease is characterised by defective chloride transport in epithelial cells and excess mucus secretion. Neonatal intestinal blockage, or meconium ileus, is virtually diagnostic of the disease but is seen only in about 10 to 15% of cases. Pancreatic insufficiency and malabsorption in the gut are common features of the disease. Elevated sweat electrolytes and the absence of a characteristic chloride ion channel are also diagnostic. Males are typically infertile due to blockage or absence of the vas deferens. Chronic opportunistic lung infection, due to viscid mucus production, is the major cause of premature death. However, there is a broad spectrum of presentations and variable severity of clinical manifestations. Mildly affected patients may have only moderately elevated sweat chloride, minimal lung deficiency and be pancreatic sufficient.

Improved nutritional care, pancreatic enzyme supplementation, aggressive antibiotic treatment and physiotherapy have all combined to improve dramatically the life expectancy of CF patients. Heart lung transplant is a final option for adult patients, but this is still essentially a lethal disease and patients rarely survive beyond the third decade of life.

Until recently, the biochemical basis for the defect was unknown. Genetic linkage analysis assigned the disease locus to human chromosome 7. Application of positional cloning strategies and linkage disequilibrium studies pin- pointed the region of interest and led to the identification of the specific locus, transcript and primary mutation. The gene was named the cystic fibrosis transmembrane conductance regulator (CFTR) in light of the predicted role of the gene and from a comparison with other genes of known structure and function.

Approximately 70% of patients were found to carry a single mutation, 1F508, a 3bp deletion which removes a single phenylalanine residue. Subsequently over 170 independent CFTR mutations have been identified. The ΔF508 mutation is almost invariably associated with severe disease and pancreatic insufficiency. A small subset of missense mutations are associated with pancreatic sufficiency and are dominant to the ΔF508 mutation. Homo- zygous nonsense mutations have been reported both in patients with very mild or with severe clinical symptoms, and with and without pancreatic insufficiency. However, with the exception of pancreatic status, there is as yet no clear correlation between specific mutations and clinical phenotype. In general in humans, nonsense, frame-shift and splice-site mutations, together with a sub-set of missense mutations, including the major ΔF508 mutation, are pancreatic insufficient while individuals carrying one or two copies of a number of other missense mutations are pancreatic sufficient. Furthermore, patients with putative "null" mutations have been described as having either a very mild or severe phenotype.

Missense mutations indicate residues likely to be critical for normal function and these inferences have been largely borne out by ectopic expression studies; transfection of non-expressing cells in culture or cells from normally expressing tissues derived from cystic fibrosis patients with wild type CFTR cDNA, but not most mutant cDNA's, restores the characteristic chloride ion channel. Final proof that CFTR is the protein responsible for the characteristic electrophysiological effect has been provided by protein purification and reconstitution studies. Other possible causes of variability in phenotype are variation in genetic background, and the point during development of the exocrine tissues at and beyond which the physiological consequences of altered chloride transport precipitate pathological changes. The CFTR gene is present and conserved in all vertebrates examined. Molecular cloning of the mouse CFTR gene showed that it was highly conserved, notably for exon 10 which harbours the major ΔF508 mutation, and indeed all but 1 of 60 putative missense mutations in man occur at amino acid residues which are conserved in the mouse.

Limitations on clinical and in vitro studies inhibit further investigation of the disease in man. There is therefore an urgent need for an animal model of CF but there is no known natural model and no means of generating the spectrum of disease manifestations in animals has so far been forthcoming. A particular difficulty in this is that, as the defect is recessive, heterozygous animals do not display clinical signs of the disease and homozygous animals would usually die as neonates as a consequence of the disease.

Surprisingly the present inventors have established a mouse model by introducing a null mutation into the murine Cftr gene (i.e. the homologue of CFTR) in which homozygous animals display only a mild form of the disease and are therefore available for further investigation. The heterozygous animals appear quite healthy and are useful as breeding stock for production of the homozygous animals. Using these heterozygous animals it is also possible to produce heterozygous mice displaying more complex forms of CF by crossing with heterozygotes bearing other Cftr mutations or mutations in interacting genes or gene products. The present invention therefore provides a mammal other than man having a CF-related defect in at least one copy of Cftr.

CF-related defects may be, for instance, a null mutation (i.e which precludes protein expression), a ΔF508 deletion or any other missense or nonsense mutation of Cftr which results in clinical manifestations of CF in a mammal also bearing a null mutation in the second copy of Cftr. Preferably the mammal is a laboratory rodent, especially a mouse.

The animals of the invention have a CF-related defect in at least one copy of Cftr. Those which have such a defect in only one copy of Cftr are CF-carriers which are useful as breeding stock. Those which have a defect other than a null mutation or ΔF508 in each copy of Cftr are also useful as breeding stock and may show clinical manifestations of CF in which case they may also be useful as a model for the disease. Those animals which are homozygous for a null mutation or the ΔF508 mutation are useful as a model of CF and those which have one copy of Cftr with a null mutation or the ΔF508 mutation and one copy with another defect are useful as a model of complex CF. In particular, null-mutant homozygous mice display a phenotype having features of the human condition of cystic fibrosis which make them valuable for investigation of the biochemical basis and pathophysiology of the disease, interactions with other relevant genes, epigenetic and environmental factors, and the development and testing of therapeutic interventions. The mutation of the Cftr gene in mice may precipitate biochemical and physiological changes in the epithelium, or other cell or tissue types. which lead to pathological changes in, for instance, the gut, gonads, pancreas, liver and lung and thus model human disease pathology and abnormalities of the corresponding tissues and organs, albeit of different aetiology in roan. The null-mutant homozygous mouse therefore serves as a useful model for studying these processes and the possible effects of therapeutic agents and interventions.

Treatment of the mutant animals with drugs, alteration of the environment (including exposure to lung pathogens and alteration to the diet) , epigenetic and genetic factors (eg altering the genetic background and observing the synergistic effect of other, relevant transgenes, e.g. normal and mutant forms of the human (or other species) CFTR (or functionally/evolutionarily related genes) will modify the effects of the insertional disruption and consequently the pathology of the mouse in a manner which is not predictable, but may be useful.

The CF-related defects may be introduced by the technique of targeting an insertion into the Cftr gene in embryonal stem (ES) cells, returning the cells to blastocysts and implanting the blastocysts into pseudopregnant females to obtain chimaeric offspring which are chimaeric for the modified ES cells and host embryo. If the ES cells colonise the germ cells of chimaeric male mice then they can give rise to offspring which have a haploid inheritance from the ES cells and the modification will then be passed on to the next generation. Preferably the ES cells are derived from a male embryo since all male offspring will necessarily be derived from sperm bearing the insertion whereas when female ES cells are used only 50% of female offspring carry the insertion, complicating selection for further breeding and use as a model of CF. The chimaeric animals are screened and selected for ability to transmit the defect to offspring. This technique is well known in the art in general [Capecchi, Science. 24 . 1288-1292 (1989)] although it has not previously been applied to inducing defects in the Cftr gene and generation of null-mutated homozygous animals. Successful transgenic insertional disruption of the mouse Cftr gene in embryonic stem cells and generation of chimaeric mice having 1 copy of the disrupted gene in at least some cells is described in Dorin, J.R. et ai Transσenic Research 1, 101-105(1992) incorporated herein by reference.

The insertional strategy does result in a partial duplication of genomic sequences and it is at least theoretically possible for abnormal splicing to produce a normal transcript. Where this has been looked for in comparable experiments targeting the HPRT gene, no evidence for aberrant splicing has been found [Deng and Capercchis Mol.Cell Biol. 12.3365-3371 (1992)]. However, in another relevant study involving an insertion into the N-myc gene, a low level of aberrantly spliced normal transcript was delected [Moeus et al, Genes and Development 6 , 691-704 (1992) ] .

The present invention therefore provides, in one embodiment, a mouse having a CF-related transgenic insertional disruption of at least one copy of Cftr. Such mice typically have a heterologous DNA insert in exon 10 of Cftr. Preferably the insert comprises a DNA sequence containing stop codons in all possible reading frames and transcriptional stop signals.

It should be noted that the phenotype of mice carrying the insertional disruption may be variable due to the potential instability of the event. Aberrant splicing and exon skipping may result in a low level of functional gene product in varying numbers of cells at different times. A perfect somatic reversion of the insertion event would result in the reconstitution of the normal gene structure and permit normal gene transcription and translational to function protein in all clonal derivatives. If the original insert contained homologous DNA engineered to include a CF-related defect and such an excision event leaves the homologous DNA in position, a CF- defective Cftr gene will be formed. This may occur spontaneously or may be induced by conventional techniques [in an adaptation of the "hit-and-run" strategy of Hasty e al. Nature. 350. 243-246 (1991)] and will produce animals with a CF-related defect but no heterologous DNA. Such a technique may be adopted to introduce mutations such as ΔF508. An imperfect excision event may result in a missense or different form of nonsense mutation, the effect of which may be reduce, accentuate or leave unchanged the phenotypic effect of the original insertion. The consequences of any such event would likewise be expressed in all descendent cells. It may be necessary to screen and select mice with stable mutations for use in model systems. Quantitative and qualitative measurement of the level of mouse cftr expression and correlation with the pathophysiology will be valuable for this and for comparison with the human condition in prognosis, treatment and therapy. The present invention also relates to

(l) the use of heterozygous animals of the invention for breeding homozygous animals.

(2) the use of homozygous animals of the invention as model for CF; and

(3) a method for assessing the potential benefit of an agent for treating or preventing CF in humans comprising the use of an animal according to the invention. The invention will now be illustrated with reference to the figures of the accompanying drawings in which Fig.l shows the insertional vector and targeting scheme used in Examples 1 and 2.

S_UBSTITUTE sneer Fig. 2 shows tissue sections stained according to expression of transferred reporter genes. Fig. 3 shows southern blots of amplified mRNA. Fig. 4 shows graphs of nasal PD following treatment with forskolin and amiloride and histograms of the PD changes. Fig. 5 shows graphs of tracheal PD in vitro following treatment with forskolin and amiloride and histograms of the PD changes.

Fig. 6 shows histograms of PD change in various tissues following treatment with forskolin.

In each of the histograms of Figs. 4,5 and 6, stippled areas represent results for cf/cf animals, shaded areas represent results for treated (TR) animals and hatched areas represent results for +/+ animals.

Figure Legends

Figure 2

Localisation of β-galactosidase expression in the murine trachea, bronchi and alveoli: Blue staining is visible in the epithelia of the trachea of mice nebulised

with pCMV-0^u (a and d) but not in control animals exposed to pBR322 (b and e). Haematoxylin and eosin staining of an adjacent section to a and d showing no

evidence of inflammation (c and f). (a, b and c, 320x; d, e and f, 800x). Representative staining of the bronchi (g, 320x and h, 800x) and alveoli (i, 800x) are also shown. Figure 3

Detection of human CFTR RNA in cβcf mice. Samples from the RNA-PCR

PCNA and human CFTR multiplex reactions were run on agarose gels after 25

cycles of amplification. Hybridisation of the PCNA specific oligonucleotide to the

Southern blot and the human F77?-specifιc internal oligonucleotide are shown. Mice 9A, 9B, 9C, 9D, 9F, 9H and 81 received human CE77? cDNAs to both the respiratory and gastrointestinal tracts. Mice 9G and 7B were untreated and 7E

received pCMV- 3 to both respiratory and gastrointestinal tract. Comparison of the

treated animals showed that the ratio of mouse PCNA to human CFTR PCR

product was variable between animals.

Figure A

The response of the in vivo nasal PD to superfusion with amiloride (A) (iOO μM) in HEPES-Krebs and forskolin (F) (10 μM) in low chloride (11 mM) HEPES-

Krebs (a) Representative response from an untreated cflcf mouse (maximum baseline: -20.3 mV), (b) representative response from a treated cflcf mouse (maximum baseline: -21.9 V), (c) response of the treated cβcf mouse showing largest degree of correction (maximum baseline -21.0 mV) (d) representative

response from a +/+ mouse (maximum baseline -17.0 mV). (e) shows the mean change in PD in cflcf mice ( n= 10) treated cβcf mice. (TR, n=6) and

+/+ mice ( n=6). (cflcf v +/+ p=0.001, cβcf s TR p=0.08, +/+ v TR p=0.13). Error bars indicate SEM and PD refers to absolute changes. The

examples shown were chosen to represent, as closely as possible, the mean

forskolin responses of the indicated groups. Figure 5

The in vitro responses of the trachea to addition of amiloride (100 μM, mucosal, panels a and b), forskolin (10 μM, mucosal, panels c and d) and A23187 (10 μM, mucosal, panels e and f) in cβcf( n = 11 , except for amiloride response n = 19),

treated cflcf ( n=7) and +/+ ( n= 13, except for amiloride response n=22) mice. The cflcf mice include untreated animals as well as those exposed to control

DNA (pCMV3), since no differences were seen between the two groups. (Values

for the control DNA group (n=7) were: amiloride PD: -44.2% (7.8), Isc: -30.1 % (11.8); forskolin PD: 0.5 mV (0.1), Isc: 7.3 μA cm'² (5.0); A23187 PD: 1.5 mV (0.9), Isc: 39.5 μA cπr² (18.5)). Error bars indicate SEM and PD refers to

absolute changes. Statistical comparison refers to Isc^: +/+ v cflcf. amiloride

p <0.001 , forskolin p <0.01, A23187 p=0.61; +/+ v treated (TR): amiloride p=0.16, forskolin p=0.09, A23187 p=0.71; cflcf v TR: amiloride p=0.07, forskolin p=0.09, A23187 p=0.53. Baseline values were: +/+ PD: -2.8 mV (0.4), Isc,,: 55.5 μA cm^'2 (9.6), G: 22.4 mS cm'² (1.9); cβc ^"PD: -2.4 mV (0.5),

Isc,,: 57.3 μA cm ² (8.9), G: 18.2 S cπr² (2.5); TR PD: -2.7 mV (0.8), Isc,,: 51.3 μA cm^'2 (6.9), G: 17.6 mS cm^'2 (2.1). The percentage reduction of baseline produced by amiloride was PD: cflcf ^'43.7% (5.4); TR 54.9% (7.7); +/+ 63.0% (5.0), Isc,,: cflcf 36.0 (7.5); TR 51.5 (11.3); +/+ 64.3 (5.8). (g) representative trace from an untreated cβcf mouse (baseline parameters prior to pretreatment: PD

-1.2 V, Isc_eς 30.1 μA cm^'2, G 24.2 mS cm^"2), (h) representative trace from a

treated cβcf mouse (baseline: PD -0.6 mV, Isc^ 12.7 μA cm^'2, G 21.4 mS cm'²),

(i) trace from the treated cflcf mouse showing greatest degree of correction (baseline: PD: -2.0 mV, Isc_^,: 34.0 μA cm^"2, G: 17.4 mS cm^'2), (j) representative trace from a +/ + mouse (baseline: PD: -3.3 V, Isc_e<1: 100.0 μA cm"², G: 30.2 mS cm^"2). F = Forskolin and A23 = A23187. The examples shown were chosen to represent, as closely as possible, the mean forskolin responses of the indicated groups.

Figure 6

The mean responses of jejunum (panels a and b), ileum (panels c and d) and rectum (panel e) to forskolin (10 μM). cflcf (jejunum n=24, ileum π =7,

rectum n = 14). treated (TR) (n = 8 all tissues). +/+ (jejunum n=22, ileum n=9, rectum n=9). Error bars indicate SEM. All statistical comparisons refer to Isc measurements. +/+ v cflcf. jejunum p<0.0001, ileum p=0.01 , rectum p=0.0001; +/+ v TR: jejunum p< 0.001, ileum p<0.05, rectum p<0.01; cflcf ^"v TR: jejunum p=0.78, ileum p=0.15, rectum p=0.24.

The invention will be illustrated by the following Examples- Example 1

Transgenic mice have been established following injection of correctly targeted ES cells into host blastocysts and the creation of male mice chimaeric for host and targeted ES cells (Dorin, J.R. et a^.. , loc. cit. ) . Male chimaeras showing germline transmission of the transgene have been identified. Male and female offspring each carrying one copy of the planned alteration to the CFTR gene, i.e heterozygotes, and having no obvious abnormalities, were crossed and the progeny studied. The transgene shows apparently normal Mendelian segregation, with approximately one quarter of the offspring having both copies of the CFTR gene disrupted by the insertional gene targeting event, as shown by DNA analysis.

Although some of these homozygous mutant mice died perinatally, there was not a substantial excess of deaths over those observed for the other genotypes (i.e. homozygous wild type and heterozygote) . Surprisingly, most of these mutant mice survived beyond weaning. However, post-mortem analysis of those which died, and of mice which survived but were killed for examination, showed significant and variable pathological changes in the gut, lung, salivary glands and vas deferens of males, and demonstrate changes in vivo in potential difference in the epithelium of the respiratory and gastrointestinal tracts, consistent with a defect in chloride ion transport of the epithelium, and paralleling those which characterise patients mildly to moderately affected by cystic fibrosis.

Example 2 Introduction

The mouse cystic fibrosis transmembrane conductance regulator gene was disrupted in embryonal stem cells using an insertional gene targeted vector, derived chimaeras which transmit the disruption to heterozygous offspring were selected and the homozygous mutant offspring obtained from heterozygous crosses displayed a spectrum of viability and intestinal pathology which parallels the severe form of the human condition of cystic fibrosis. Heterozygotes (for the Cftr insertion) sometimes also display at least some aspects of CF pathology reflecting a complex aetiology relying on other (non-engineered) mutations either in the second copy of Cftr or in cooperating genes or gene products and are useful in investigating and rescuing such mutations. The insertional mouse mutant provides a valuable model system to investigate the pathophysiology of severe cystic fibrosis and may facilitate the development and testing of therapies relevant to the treatment of patients.

Gene Targeting and Germline Transmission The construction of the insertional vector plV3.5H and successful targeting of the Cftr gene in embryonal stem cells is described in Dorin, J.R. e_t_ a∑. , loc. cit. and is illustrated in Fig.l in which the neomycin sequence is indicated by dark hatching and labelled neo. Restriction sites are labelled as follows:

The position of the 1.2H probe is indicated as are the 6 and 5kb hybrising fragments mentioned below. Exons 9 and 10 are shown by solid blocks. Introns are indicated by light hatching, the plasmid sequence by stippling.

Briefly, a 3.5kb Hind III fragment of DNA isogenetic with the E14 embryonal stem cell line and which encompasses part of intron 9 and extends into exon 10 was cloned into the pMClneopolyA vector. This insertional vector was linearised at the Asp718 site, electroporated into the E14 embryonal stem cell line and G418 resistant clones screened by Southern blotting for the presence of a novel, diagnostic fragment using a flanking Cftr genomic probe, 1.2H, not present in the vector. [DNAs were digested with Xba I + Sal I, transferred to nylon membranes and probed with a 1.2 b Hind III fragment of exon 10 external to the insertional vector, as described in Dorin, J.R. et al. , loc cit. ] A 6kb hybridising fragment represents the endogenous fragment while the hybridising fragment at 5kb is indicative of a correctly targeted event.

Correctly targeted clones were obtained at a frequency of about 2% of G418 resistant clones. Clones [which retained a modal chromosome number of 40 and the ability to differentiate jLn vitro1 were injected into C57 Bl/6 donor blastocysts which were implanted into pseudopregnant female mice to obtain chimaeric offspring. Thirteen (ten male) of 24 offspring were chimaeric as judged by coat colour markers, and 8 males were mated [to MF1 (Olac) females] to test for germline transmission. Six transmitting males were identified whose offspring had grey coat colour (genotype +/P. c/c^ch) , indicative of germline transmission. These were genotyped by Southern blotting using the 1.2H probe. Approximately half (54 out of 113) of the mice were heterozygous for the wild-type 6kb Xba I fragment and the novel 5kb Xba I-Sal I fragments and were designated Cftr^τ°^H(neo'^m)1HGU. The cf/+ mice (heterozygous for the insertional mutation) were phenotypically normal. Heterozygotes were mated and the offspring genotyped and studied. Production of Viable Homozygous Mutant (cf/cf) Mice

One male (derived from chimaera 42) was mated to three females (all derived from chimaera 48) and produced 24 offspring, of which 8 were wild type (+/+) , 10 were heteroxygous (+/cf) and 6 were homozygous mutant (cf/cf) . A single mouse died perinatally, was histologically normal, and genotyped as being cf/+. All other mice survived beyond weaning and cf/cf mice showed no overt failure to thrive over the study period (up to 30 days post-partum) . Electrophysiology

If targeted disruption of the Cftr gene effectively abolishes gene function electrophysiological changes consistent with the absence of the CFTR chloride channel would be expected in all cf/cf mice, irrespective of disease status. A blind trial was carried out on ten mice, comprising one control mouse plus nine offspring selected from the three litters in the study group. These nine coded mice were also examined for pathological changes (see below) . Gastrointestinal and respiratory tract potential differences (PDs) were determined iri vivo by an established method [Knowles et al, New Eng.J. Med.. 303. 1489-1495 (1981) ] modified for the mouse. The pooled results are given in Table 1.

Tabie 1 : Potential Difference Measurements in Gastrointestinal and Respiratory Tract.

Gastrointestinal and respiratory tract potential difference (PD) measurements were previously established in a set of MF1 mice (data not shown, results summarised below under Method). For this experiment, baseline measurements were first confirmed on a normal control mouse. Nine experimental mice were sampled from three cf/ + cf / ♦ litters and analysed in a blind trial.

For statistical analysis of significance, the results from individuals identified by DNA analysis as cf/cf were pooled and compared with the combined cfi + and + / + genotypes. Means and (in brackets) standard errors of the means (SEM) are shown, and significant differences between the two classes are asterisked (* « p < 0.05, ** _* p < 0.01). FZ= forskolin plus zardavenne A upwards arrow indicates an increase towards more negative values, and conversely for a downwards arrow

Method:

Bucca! PD was used to validate the integrity ol the circuit prior to further measurements Nasal and airway PDs were recorded following a distal tracheostomy Tracheal PD was recorded from the proximal trachea 'Peripheral airways' rep'esent the most peripheral part of the airways reached by the exploring electrode of 0.5mm diameter. 'Bronchial' measurements were taken at the mid-pomt between the tracheal and peripheral airway recordings Caecal and colonic PD was measured under direct vision following opening of the abdominal cavity. The composition of the bathing solution (mM) was NaCI, 140. KCI. 6, MgCl2, 1 ; CaCl2* 2, Glucose, 10; HEPES, 10; pH 7.4 in all experiments except in low chloride solution where NaCI was replaced by equimolar sodium gluconate (measured chloride was 11mM). Measured values were not corrected for junctional potentials. It is unlikely that the paraceliular shunt is affected in the tracheal epithelium of cystic fibrosis patients.⁷ Forskolin (1μM) and zardaverine (100μM) were dissolved in ethanol and DMSO respectively and were perfused in the presence of amiloride (100μM). The combination of ethanol and DMSO has been shown not to significantly alter murine caecal and tracheal PD. The measured murine nasal PD profile mimics that of human nasal epithelium and data given are obtained from the maximum measured values. _ 20 _

Comparison of the data was made using the Mann- Whitney U-test and all values are expressed as the mean± SEM, for convenience. * ■ p < 0.05, •* - p < 0.01. Previously recorded nasal PDs measured using the above technique in MF1 mice showed a mean of -15.1 mV (SEM ±1.7, n«9) and rectal PDs of -16.7mV (SEM ±1.3, n«7). For comparison, human nasal PD values are: Mean of normals, -19mV (range -2 to -36mV, n«145); mean of cystic fibrosis patients, -46mV (range -33 to -77mV, n«60). The murine tracheal PD profile also mimicked that in man. Comparable values in unaffected subjects are : trachea, -16mV (SEM ±2); bronchial airway, - 11 mV (SEM ±1 ), and peripheral airway, -6mV (SEM ±1 ), n=23.

Table 1 Measurement of Potential Differences in Respiratory and

Gastrointestinal Tract

ι o Tissue c / c (n=4) cfl + , *+ / -+ (n=5

Buccal - 13.8 (±1.4 - 14.2 (±0.7)

Rectal - 5.5 (±1.2 - 16.2 (±2.6)

Rectal FZ T 1.2 (±0.4 T 6.1 (±1.3)

Caecum/Colon baseline - 3.7 (±0.6 - 10.4 (±0.9) 15 Caecum/Colon FZ i 0.3 (±0.2 T 1.6 (±0.5)

Nasal - 27.3 (±1.9 - 19.0 (±1.4)

Trachea low Cl" T 1.0 (±0.6 T 4.8 (±0.6)

Upper Trachea - 15.8 (±1.7 - 12.3 (±2.3)

Bronchial - 7.0 (±1.9 - 7.5 (±0.6) 20 Peripheral - 4.3 (±1.9 - 3.5 (±0.6) All four cf/cf mice were unequivocally identified. Baseline rectal PD in the homozygous mutant animals was significantly lower than the combined heterozygous/wild type group. Colonic/caecal PD was similarly significantly reduced in homozygous mutants.

Following perfusion of each of the above regions with a combination of forskolin (an activator of adenylate cyclase) and zardaverine (a phosphodiesterase inhibitor) , in the presence of amiloride (an inhibitor of sodium transport) , a significant reduction in the resultant hyperpolarization was seen in all homozygous mutant mice. Nasal PD was significantly increased in the homozygous mutant mice, but no difference was discernible throughout the lower airways below the larynx. Perfusion of the trachea with a low chloride solution showed a significant reduction in the resultant hyperpolarization in each of the cf/cf animals. Heterozygotes (range -10 to -11 mV) demonstrated rectal PDs intermediate between cf/cf (range - 4 to -9 mV) and +/+ (range -18 to -24 mV) mice, as well as either a reduced response to forskolin and zardaverine in the rectum, or to low chloride perfusion in the trachea, neither of which overlapped with the cf/cf values. In summary, all nine animals were studied without prior knowledge of genotype and all were correctly identified.

Pathology

Histological examination was conducted on formalin fixed tissue from the three litters described above. The histology of the wild type and heterozygous mouse is completely normal and indistinguishable. By contrast, the mutant cf/cf mouse shows pathological changes consistent with those seen in cystic fibrosis. There was mild colon dilation with abnormal mucus accumulation and a resultant distension of epithelial cells, most prominent within the crypts. The gonads were normal, but in one cf/cf male increased mucin was also seen in the vas deferens. One mouse, at thirty days post-partum, showed mild focal pulmonary atelectasis consistent with presymptomative lung disease, and mild dilation of salivary ducts. There was no overt histological abnormality in the pancreas and this is consistent with the average body weight gain of the mice. In two out of six cf/cf mice examined, no overt histopathological abnormalities were detected, consistent with a variable penetrance, as seen in man. However, these two mutant mice were readily identified by PD measurements (see above) .

Discussion

The mutation here induced in the mouse is intended to create a "null" phenotype by introduction of plasmid sequences containing translation stop signals in all reading frames, plus a neomycin resistance gene which contains strong transcriptional stop signals. No residual normal transcript has been detected by Northern blot analysis or RNA PCR analysis.

The experiments provide evidence that mice homozygous for an insertional disruption of the CFTR gene mimic important features of cystic fibrosis, including intestinal econium and ductal blockage, excess mucin in the vas deferens, altered pathology in the ling and salivary glands and altered chloride ion transport in the respiratory and gastro intestinal tracts.

In the large intestine, cystic fibrosis patients demonstrate a reduced rectal PD as well as the characteristic physiological defect of impaired chloride secretion, in response to elevation of cAMP. It has previously been shown in vitro that murine caecum responds to forskolin and zardaverine, in the presence of amiloride, with an increase in PD and equivalent short circuit current, which can be largely inhibited by bumetanide. Therefore, this response is very likely to represent chloride secretion. The large intestines of the four cf/cf mice in this study thus appear to demonstrate two of the electrophysiological abnormalities of the gastrointestinal tract of cystic fibrosis patients. An elevated nasal PD is recognised as one of the hallmarks of cystic fibrosis and is used in the clinical diagnosis of patients. Human nasal epithelium predominantly absorbs sodium in the resting state and the raised PD in cystic fibrosis relates to an increase in this absorption. It has previously been shown that murine tracheae display ion transport properties similar to those characteristic of humans. Although technical considerations precluded the use of nasal amiloride in this study, the significantly increased nasal PD in the cf/cf mice argues strongly for an increased sodium absorption in the nasal mucosa of these animals. The lower airway PD of a limited number of cystic fibrosis patients has been shown to be increased. Interestingly, this is not consistently observed and was not seen in this study. Nevertheless, the tracheae of all the cf/cf animals tested demonstrated the characteristically reduced permeability of these tissues. It was possible to distinguish heterozygotes on the basis of a minimum of two of the above criteria.

At this stage, there is no evidence for abnormal pancreatic histology, possibly due to the ameliorating effect of the large volume of acinar fluid seen in the mouse, but not in man. Thus, the ductal blockage which typifies most cystic fibrosis patients might be expected to be less severe in mice, or to develop later. Differences between the pulmonary anatomy of mouse and man has called into question whether a Cftr mutant mouse would mimic the human lung disease. However, the one cf/cf mutant mouse described here which exhibited clear signs of lung damage is significant and consistent with the age of onset of lung disease in human patients. It will be of interest to monitor the effect of aging on lung disease, as well as the consequences of exposure to lung pathogens such as Pseudomonas aeruσinosa. In the ucoid form, P.aeruσinosa is peculiar to cystic fibrosis and accelerates lung damage. The presence of mucin in the vas deferens of one male examined also mimics the pathology in cystic fibrosis and thus this mouse model may be valuable for studying male infertility.

In summary, the mutant cf/cf mice display the well established electrophysiological defect associated with cystic fibrosis and a relatively mild pathology, as judged by histology at the time of examination (10-30 days post- partum) . A long term study of survival and pathology is in progress.

The age of onset and severity of cystic fibrosis is extremely variable. At one end of the spectrum, neonates die of meconium ileus, while at the other the disease is not diagnosed until late adolescence, or beyond. However, with the exception of pancreatic status, there is as yet no clear correlation between specific mutation and clinical phenotype. Rather, the weight of evidence favours the existence of additional genetic factors influencing the expression of the disease. In this regard, a mouse model has an additional important experimental advantage in that the genetic background can be carefully controlled and modified. Such experiments will clarify the relationship between genotype and phenotype and allow investigation of contributory genetic and epigenetic factors. Ashkenazi Jews and Georgians homozygous for 'null' mutations have been reported with severe phenotypes, whereas in populations of mixed ethnic origin comparable mutations were associated with mild disease. The mice described here, which are the result of outcrossing strain 129 (E14 genotype) to an MF1 outbred genetic background, display a mild phenotype. Experiments are in progress to determine the phenotypic effect of crossing the insertional mutation onto various inbred backgrounds.

The ability to distinguish heterozygotes from wild type and mutant mice by rectal potential difference measurements argues that there can only be minimal functional protein. It will be valuable to explore in vivo the pharmacological possibilities for increasing chloride ion channel activity through overstimulation of any residual protein, or via alternative ion channel pathways. It is now also possible to modulate the level of functional CFTR by crossing these mice with other transgenic mice carrying missense mutations.

The insertional disruption of the mouse CFTR gene produces a phenotype which mirrors important aspects of the human condition. These mutant mice are therefore valuable in reaching a fuller understanding of the epigenetic and pathophysiological development of the disease. Future studies may facilitate the development of new procedures, interventions and therapies relevant to the management and treatment of the disease in patients, for instance somatic gene therapy (e.g. using ad-^»novirus or other vectors for introducing transgenes into lung tissue) . Summary

Cystic fibrosis (CF) is caused by mutations in the cystic Fibrosis trans embrane conductance regulator (CFTR) gene. CF is an appropriate disease for treatment by somatic gene therapy as the major affected organ,

5 the lung, is relatively accessible to gene delivery. With the development of

transgenic mouse models for this disease, it is now possible and important to test gene transfer protocols for efficacy prior to human trials. We report

gene transfer to the Edinburgh insertional mutant mouse (cflcf), delivering

CFTR cDNA-liposome complexes into the airways by the clinically relevant

o method of nebulisation. Correction was measured both in vivo in the nasal epithelium, important from the perspective of future clinical assessment, as well as in vitro in the trachea. We show full restoration of cAMP related chloride responses in some animals and demonstrate, in the same tissues,

human CFTR cDNA expression. Overall, a range of correction was seen

-1 5 with restoration of approximately 50% of the deficit between wild type mice

and untreated cf/cf controls. Mean values for treated animals were not significantly different to untreated controls (nose p=0.08, trachea p=0.09). The levels of human mRNA also varied, both within and between animals.

Baseline tracheal ion transport parameters and calcium mediated chloride

20 responses are not different in the cf/cf compared to wild type mice and

were not altered by treatment. These findings differ from the recent report of instillation of CFTR cDNA-liposome complexes into the trachea of the Cambridge mouse, and extend the use of liposome mediated gene transfer to a clinical context. We also report modest correction in the intestinal tract

2 following direct instillation and provide initial encouraging safety data for both the respiratory and intestinal tract following the Hposome mediated gene delivery. The non-viral nature and potentially lower immunogenicity of DNA-liposomes are intrinsic advantages which suggest that this may be a therapeutic alternative to recombinant human CFTR adenoviruses. However, several difficulties identified in the present study may be relevant to the therapeutic use of DNA-liposome complexes in man.

Introduction

Cystic fibrosis (CF) is inherited as an autosomal recessive disorder and is

characterised by reduced cAMP-regulated chloride secretion in the epithelial cells

of the respiratory and gastrointestinal tract and sweat glands'. The principal morbidity and mortality results from pulmonary disease with a current median

survival approaching 30 years². Following cloning of the cystic fibrosis

transmembrane conductance regulator (CFTR) gene³, genetic complementation of

the ion transport defect has been demonstrated in vitro*'⁵. The development of

mouse models for CF⁶ provides a system for the in vivo assessment of new

treatments, including gene therapy, prior to testing in man.

Delivery of gene transfer agents to the human airways is likely to be by

nebulisation. In man the nasal epithelium demonstrates the characteristic ion transport defect⁹ and is likely to play an important role in monitoring the efficacy of clinical trials. We have, therefore, studied whether this non invasive method of

delivering human CFTR cDNA-liposome complexes is capable of correcting the

CF ion transport defect in both nasal and tracheal epithelium of the Edinburgh CF

transgenic mouse, generated by insertional mutagenesis (cfl ¹⁰. Validation of our

nebuliser system with methylene blue, amiloride and the reporter gene β- galactosidase (/3-gal) is presented, together with assessment of toxicity of DNA-

cationic liposome complexes. Using this technique we have introduced human

CFTR cDNA into the airways of cflcf mice and demonstrate, in the same animals,

expression of human CFTR mRNA and correction of the characteristic cAMP

related chloride defect. In addition, the effect of topical application of DNA-

liposomes to the intestinal tract of these mice was tested and modest correction observed.

Recently, Hyde et al reported instillation of human CFTR cDNA-liposome

complexes into the Cambridge mutant mouse", and unexpectedly showed correction of a range of ion transport abnormalities over and above the

characteristic cAMP mediated chloride defect. Our study addresses these and other important issues arising from this study, extends these findings and assesses the use of CFTR cDNA-liposome mediated gene transfer in the context of a pre-

clinical trial. The implications of this study for the development of an effective clinical protocol are discussed.

Methodology

Animals. Mice were studied at 4-10 weeks of age (weight 22-35 g). Wild-type (MF1, Oxford Laboratory Animals, Oxford, UK), and c 7c animals⁶, were housed

in a temperature-controlled (21°C) room with food (Special Diet Services, Witham, Essex, UK) and water freely available.

Administration methods. The nebuliser system consisted of a standard nebuliser

with or without the addition of a baffle system (System-22 Optimist^®, Medic-Aid

Ltd, Pagham, UK). The median mass diameter of the aerosol generated was

measured using a laser diffraction system (Malvem Mastersizer). The aerosol was generated with 95% O-₂ / 5% CO₂ and passed into a small continuously warmed tube (volume 100 ml) in which the animal was placed. A total volume of 3 ml was used for each nebulisation. Animals were exposed to the nebuliser individually

(methylene blue, amiloride and 3-gal) or in pairs (CFTR treatment), where a second animal was placed distal to the first in the nebulisation tube (volume 200 ml). In this case, the position of the mice was reversed for the treatment of the second animal, and was used to increase the total dose received. Instillation into the murine intestinal tract was through a 1.3 mm plastic tube advanced slowly through the rectum to a distance of 3 cm, with instillation of up to 2 ml -over

approximately 1 minute. No signs of distress were seen up to 21 days following lung and intestinal treatment.

Methylene Blue Delivery. Initial assessment of aerosol delivery to the murine tracheo-bronchial tree was performed using methylene blue dye (David Bull Laboratories, Mulgrave, Australia). Animals were nebulised at 6 or 12 l.min"¹, with or without the Optimist* attachment. Immediately following nebulisation

animals were anaesthetised with 17 g/kg midazolam (Roche, Welwyn Garden City, UK) and 1 mg/kg fentanyl / 33 mg/kg fluanisone (Janssen Animal Health, Oxford, UK) and perfused with phosphate buffered saline (PBS) through both the

inferior vena cava and right ventricle of the heart. The nose, trachea and lungs were dissected to allow direct visualisation of the deposited methylene blue (n =9) .

Assessment of intestinal delivery was also performed using methylene blue (n=6).

Delivery of Amiloride. Animals were nebulised as described above, with either

amiloride (1 mM, n=6) or diluent (HEPES-Krebs (mM): Na⁺ 140, CY 152, K⁺

6, Ca²⁺ 2, Mg²⁺ 1, Glucose 10, HEPES 10 pH 7.4, n=6). Following nebulisation, animals were anaesthetised (as above) and the tracheal and bronchial

PD measured in vivo (with the investigator blinded to the treatment received) using the system previously described⁶. Briefly, the exploring electrode consisted of a fine plastic tube (outer diameter (OD) 0.4 mm) perfused with HEPES-Krebs and connected to a high impedance voltmeter by a calomel electrode, with values

referenced to a subcutaneous electrode. The offset of the electrodes was recorded prior to measurements and suitable adjustments made to recorded values. The exploring electrode was introduced through an opening fashioned in the laryngeal cartilage and advanced distally. Tracheal PD was measured 5 mm distal to the laryngeal cartilage / tracheal junction, and at 15 and 20 mm distal to this point for the bronchial measurements. A mean bronchial PD was calculated for each animal,

which was used for statistical analysis. It is important to note that measurements could only be taken approximately 10 minutes after completion of nebulisation

because of the time necessary for anaesthesia and preparation of the laryngotomy. This is likely to lead to an underestimate of the drug's effect (the t,^ of amiloride is 10 minutes in sheep²⁵).

/5-galactosidase gene delivery and detection. Gene delivery was assessed using

the plasmid pCMV-3 (Clontech Laboratories Inc., California, USA), which contains the /3-galactosidase gene under the control of the human cytomegalovirus immediate early promoter and enhancer¹⁴. The DNA (prepared by standard methods²⁶) in HEPES buffered saline was complexed with the cationic liposome DOTAP (N-[l-(2,3-Dioleoyloxy) propyl]-N,N,N-trimethyl-ammoniummethyl- sulphate; Boehringer Mannheim UK, Lewes, UK) at a ratio by weight of 1:6, with

a final concentration of DOTAP of 0.5 mg/ml. The plasmid pBR322 was used as control DNA.

Animals were nebulised with 1 mg of DNA in a total volume of 12 ml of plasmid DNA-liposome complex, given in four nebuhsations over a total of 40 minutes. Fifty per cent of the DNA-liposome complex was delivered with the Optimist* attachment, and 50% without. Four days following nebulisation, animals were anaesthetised, perfused with heparinised PBS as above, and then with ice cold 2% formaldehyde / 0.02% glutaraldehyde. The trachea was also filled with _ 34 _

this fixing solution. The trachea and lungs were dissected en bloc and immersed in fixing solution at 4°C for 2 hours, permeabilised using 0.02% Nonidet-P40

(BDH, Poole, UK) and then stained by both infusion and immersion in staining

solution [5 mM K₃Fe(CN)₆, 5 mM K_<Fe(CN)₆.3H₂O, 2 mM MgCl₂, 0.01 % Sodium deoxycholate (BDH, Poole, UK) and 0.02% Nonidet-P40 in PBS containing 1 mg/ml X-gal (5-bromo-4-chloro-3-indolyl-S-D-galactopyranoside, Pharmacia Biotech Ltd, Milton Keynes, UK) dissolved in N-N-dimethylformamide (Sigma, Poole, UK)] for 3-5 hours at 30°C. The tissue was transferred through three changes of 70% ethanol before further dehydration and embedding in

paraffin wax. Four micron step sections through the entire trachea and lungs were counterstained with nuclear fast red and examined by bright field microscopy. Background endogenous S-gal activity was detected as a blue reaction product predominantly in thyroid follicular epithelium, as foci in cortical thymus and in mucous acini of the laryngeo-tracheal submucosal glands of animals. Only those

animals showing positive staining in the thyroid or thymus were used for quantitative analysis (pBR322 n=2, DOTAP n = l, pCMV/S n=2). Up to 6 sections from each of these animals were scored semi-quantitatively for the amount (score 1-4 relating to <25%, 25-50%, 50-75% and 75-100% of cells stained respectively²⁷) of positive staining in tracheo-bronchial epithelia. Adjacent sections

were also stained with haematoxylin and eosin and examined for evidence of inflammation or loss of surface epithelium. A similar experimental procedure was used for DOTMA/DOPEN-[l-(2,3-dioleyloxy)propyl]N,N,N-trimethylammonium chloride (Lipofectin: Gibco/BRL, Paisley, UK). CFTR gene delivery. Transfection of the CF7R cDNA was mediated using the cationic liposome DC-Chol/DOPE, (3/S[N-N\N'-dimethyl amino ethane - carbamoyl] cholesterol/dioleoyl phosphatidyl ethanolamine). The reagent DC-Choi was synthesised and liposomes produced using a modification of a previously described protocol²⁸. DC-Choi was synthesised in dry tetrahydrofuran, the product column purified on silica gel using chloroform: ethanol (4:1) as the eluent, and then re-crystalised from ethyl acetate in 80% yield; liposomes were prepared in

80 ml batches using DC-Choi and DOPE (Sigma, Poole, UK). Two stages of

sonication were employed to form liposomes with an average diameter of 400-500

nm, as estimated using a photocorrelation laser spectrometer (Zeta-sizer 3). All plasmid DNA was prepared using standard methods²⁶ and complexed with the DC- Chol/DOPE liposomes at a ratio of 1:5 (wt/wt). The final concentrations of liposome (in water) were 0.5 mg/rnl (airway) and 1 mg/ml (intestine). The CFTR cDNA expression plasmids used were: (1) pSPC-CFTR (p3.7 SP-C hCFTR (936C) t-intron poly A) which has been previously described²⁹ and shown to produce human CFTR expression in vivo. (2) pCMV-CFTR, which was constructed by replacement of a Nde II Pst I restriction fragment containing the SP- C promoter in pSPC-hCFTR (936C) t intron poly A² with a Not I/Nhe 1 restriction fragment containing the human CMV immediate early promoter/enhancer from ptgCMVHyTK. (3) pSV-CFTR which was constructed by insertion of the CFTR cDΝA from pBQ4.7⁵ (a gift of L.C. Tsui and J. Rommens) into the Pst I site of pSVK3 (Pharmacia Biotech Ltd., Poole, UK)

placing the CFTR cDΝA under transcriptional control of the SV40 early promoter. Control plasmids used were pCMV-/3, p(pro⁺) TAG Νeo³⁰ and pBluescript (Stratagene Ltd., Cambridge, UK).

The treatment protocol used was; all treated animals (n=8): lung - 1 mg

pCMV-CFTR, 1 mg pSV-CFTR and 1 mg pSPC-CFTR; intestine - 200 μg

pCMV-CFTR, 200 μg pSV-CFTR; control treated animals: lung - 1 mg pCMV-0 (n=5) or pCMV-3 (1 mg), p(pro⁺) TAG Neo (1 mg) and pBluescript (1 mg)

(n=2); intestine - 400 μg pCMV-/S (n=5) or 200 μg pCMV-0 and 200 μg p(pro⁺)

TAG Neo (n=2), untreated: cflcf (n= 19), +/+ (n=22). For statistical analyses

control treated and untreated cflcf mice were pooled since no differences in

electrophysiological measurements were detected. Additional details of the control treated animals are shown in the legend to Fig. ⁶ • Animals were tested at 16-24 hours (CFTR) and 16-68 hours (controls) after treatment.

Human CFTR mRNA detection. For mRNA studies, peripheral lung (n=7),

rectum (n=6) and proximal colon (n=2) were frozen in liquid nitrogen immediately after completion of in vivo measurements and stored at -70°C. Tracheal tissue (n =5) was frozen following Ussing chamber measurements. Nasal tissue was not studied. Poly A+ mRNA was isolated, treated with 10 units RNase-

free DNase (Promega); pheno chloroform extracted, converted to cDNA (First strand cDNA synthesis kit, Pharmacia Biotech Ltd., Poole, UK) and subjected to amplification by PCR with Taq DNA polymerase (Perkins-Elmer Cetus, California, USA). To control for the amount of mRNA present in the reaction, the

sample was amplified using oligonucleotides specific for mouse PCNA

(proliferating cell nuclear antigen)³¹ (5'GGT TGG TAG TTG TCG CTG TA,

sense and 5'CAG GCT CAT TCA TCT CTA TGG, antisense). This produces a product of 720 bp encompassing exons 1 to 4 and hybridizes to the PCNA specific internal oligonucleotide PCNAI, (5'GAA CAG GAG TAC AGC CTG TGT AA). Human specific F7R oligonucleotides were used to assay for the presence of human CFTR mRNA. These oligonucleotides (HCFS 5'CCT GTC TCC TGG ACA GAA, sense) and (HCFA 5'GTC TTT CGG TGA ATG TTC TGA C, antisense) amplify a 336 bp human exon 13 product. Only the internal oligonucleotide HCF13I (5^* CAA ATG AAT GGC ATC GAA), which is

complementary to a region of human exon 13 absent from the murine cfir exon 13³² will hybridise to this amplified product. PCR reactions were carried out for

21 and 26 cycles with 55°C annealing for 45 sec, 72°C extension for 1 min 30 sec and 92°C denaturation for 25 sec. The PCR amplification products were evaluated

by agarose gel electrophoresis, followed by Southern blotting and hybridisation to the human CFTR specific (at 48°C) and then the PCNA sequence specific (at 55°C) internal oligonucleotides. A pool of mRNA was treated in an identical manner to the individual mRNA samples, but no reverse transcriptase was present in the cDNA reaction (-RT) and a reaction tube with no cDNA (-RNA) was included as a control for DNA plasmid contamination.

Electrophysiological techniques. Animals were anaesthetised and an opening in the laryngeal cartilage fashioned as noted above, together with a tracheotomy at the junction of the laryngeal cartilage and trachea. Buccal PD was used to validate the integrity of the circuit before further measurements were taken (+/+ -13.4

mV (0.6), cflcf -12.9 mV (0.5), treated cflcf -12.1 mV (0.5) p=NS all comparisons). For nasal measurements, the exploring electrode (OD 0.3 mm) was advanced into each nasal cavity in 1 mm increments, and the maximum stable PD (± 0.5 mV over 15 seconds) recorded. For drug responses, the exploring electrode was positioned 3 mm within the right nasal cavity and a perfusion

catheter (OD 0.25 mm) passed through the laryngeal opening to the junction of the two nasal passages, allowing retrograde perfusion (40 μl/min) of the nasal

epithelium. Drug responses were measured in the following order: amiloride (100 μM in HEPES-Krebs), forskolin (10 μM) and A23187 (10 μM, both in low chloride (11 mM) HEPES-Krebs with gluconate substitution). The junctional potential between the normal bathing solution and the low-chloride solution was ° ignored because of paired comparisons. It is unlikely that the paracellular shunt

^' is affected in CF airway epithelium³³. Rectal PD was measured using a tube (OD

0.4 mm) inserted to a distance of approximately 12 mm. Following the recording of stable values (± 0.2 mV over 15 seconds), drugs were administered as a 200 μl bolus over 15 seconds through a three-way connector. The sequence of drug 5 adminstration was amiloride (100 μM) followed by forskolin (10 μM), each dissolved in HEPES-Krebs.

Following in vivo measurements, animals were sacrificed, the trachea dissected and placed in Krebs Henseleit solution of composition (mM) Na⁺ 145.0;

Cl^' 126.0; K⁺ 5.9; Ca²⁺ 2.5; Mg²⁺ 1.2; HCO₃ ^" 26.0; PO₄ ^2" 1.2; SO₄ ^2" 1.2; glucose 0 5.6. The small intestine was placed in an equivalent solution, with the glucose replaced by equimolar mannitol. All tissues were transported to the laboratory at

4°C and mounted in Ussing chambers of aperture diameter 0.28 cm² (jejunum and ileum) or 0.03 cm² (trachea). Two to four chambers were obtained from the small intestine. Values were subsequently meaned to provide a single measurement per tissue for each animal. Both mucosal and serosal surfaces of the tissues were bathed in Krebs Henseleit solution at 37°C, pH 7.4, circulated by 95% O₂ / 5% CO₂, with the exception of the small intestine, where glucose was replaced with

equimolar mannitol on the mucosal surface. Studies were performed under open-

circuit conditions to maintain uniformity with in vivo measurements, but prior to and at the maximum response of each drug intervention, tissues were short- circuited until stable values were reached (approximately 5-10 seconds). Because

of the necessarily small aperture of the tracheal chamber and the consequent small

absolute Isc, we calculated Isc equivalent (Isc^,) from the passage of a 2 μA

current pulse. In the trachea, once stable values (gradient of trace <4%^M) were attained, tissues were pre-treated sequentially with amiloride (100 μM), 4,4'- dϋsothiocyanatostilbene^^'-disulfonic acid (DIDS, 100 μM), phloridzin (200 μM), and acetazolamide (100 μM). Subsequently, forskolin (10 μM) followed by A23187 (10 μM) was added. All drugs were added mucosally, except acetazolamide (serosally). In the small intestine, the sequence of additions was forskolin (10 μM, serosally), carbachol (1 mM, bilaterally), bumetanide (100 μM, serosally), glucose (5.5 mM, mucosally) and phloridzin (200 μM, mucosally). Recordings from all treated animals studied are included in the analyses, with the exception of nasal responses (n=2) and tracheal (n = l) in which no measurements could be obtained because of technical difficulties. This proportion is similar to that for +/+ mice and reflects the limitations imposed by animal size. All statistical comparisons were made using the Mann-Whitney U test. Result s Nebuliser system

The nebuliser system comprised a standard jet nebuliser with or without the

addition of a baffle. The particle size generated by this system ranged from 1.7 μm (at 12 Limn^-1 with baffle) to 3.7 μm (at 6 l.min"¹ without baffle). To assess

aerosol delivery to the murine tracheo-bronchial tree, methylene blue dye was

nebulised into wild-type (+/+) mice. Nasal deposition was qualitatively similar irrespective of the nebuliser setting. Deposition in the trachea appeared optimal

with a particle size of approximately 1.9 μm (12 l.min^"1 without baffle), whilst in the bronchial tree it was maximal with the smallest particles (1.7 μm). Therefore,

to maximise delivery to the entire airway, all subsequent aerosols were delivered using a combination of the nebuliser alone and the nebuliser-baffle system at 12 l.min'¹.

Nebulisation of amiloride The efficacy of this nebuliser system for drug delivery was assessed in +/+ mice by in vivo measurement of the change in tracheal and bronchial epithelial potential difference (PD) produced by the sodium channel blocker amiloride. Amiloride produced a 45% reduction in tracheal PD (diluent: -24.0 (2.1) V, amiloride: - 13.1 (3.0) mV, p<0.05) and a 39% reduction in bronchial PD (diluent: -4.6 (0.6) mV, amiloride: -2.8 (0.5) mV, p< 0.05) consistent with a sodium-absorbing

epithelium¹². Reporter gene transfer

DNA transfer was evaluated using the 0-gal reporter gene and two commercially available cationic liposomes, DOTAP and DOTMA/DOPE¹³ (Methodology). Following nebulisation of 1 mg of DNA (pCMV-S¹⁴) complexed with DOTAP, β- gal activity was evident in up to 40% of surface epithelial cells in the trachea and extra-pulmonary bronchi of +/+ mice (Fig.2a-h), compared to approximately 3%

in animals that received control DNA or liposome only. Staining was present in the airway surface epithelial cells and was variable with respect to intensity and distribution. Lung parenchyma showed focal positive staining, often at lung margins and apices, in alveolar cuboidal lining cells (possibly type II cells),

macrophages and interstitial cells (Fig. 2i). Similar results were obtained with

DOTMA/DOPE (data not shown).

Intestinal administration

To assess the potential distribution of drugs and DNA when directly instilled into

the lumen of the intestinal tract via the rectum, methylene blue dye was again used as a marker. Administration of 2 ml resulted in uniform distribution of dye in the rectum, colon, caecum, and approximately 3 cm into the distal ileum, but not into

the jejunum. - Δ 2 -

Effect of the liposome DNA complexes on epithelial morphology and ion

transport properties

The airways and lung parenchyma were examined histologically in untreated mice,

and mice which had received DNA complexed with DOTAP, or DOTAP alone. The groups were indistinguishable, with one animal from each (including an untreated animal) showing a mild to moderate lymphomononuclear infiltrate

restricted to the alveolar interstitium, together with apparent hypertrophy of non-

ciliated bronchiolar (Clara) cells. There was no evidence of epithelial necrosis and

no other structural changes were noted in any animal. Similarly, no evidence of histological damage was seen after the administration of DOTMA/DOPE.

The effect of nebulised and directly instilled DNA-DOTAP complexes on

the in vivo and in vitro electrophysiological properties of +/+ mice, are shown in Table 2 .

TABL E 2 Comparison of the electrophysiological properties of +/+ mice and +/+ mice

exposed to DNA-liposome complexes. No significant effect of exposure to DNA complexed with the cationic liposome DOTAP was seen on the potential difference

(PD), short circuit current (Isc or Isc_^, for trachea) or conductance (G) whether

administered by nebulisation or direct instillation, in any of the tissues tested.

Values are mean (SEM). TABLE 2

- aa -

Epithelial damage may be detected by a decrease in the transepithelial

PD, short circuit current (Isc) or by an increase in tissue conductance (G). No

significant adverse effect of treatment was seen on these parameters, suggesting that cellular and paracellular pathways were not markedly altered. The lack of

consistent change in either histology or bioelectricai parameters, suggests that the

DNA-liposome complexes tested have no major deleterious effect on either the

respiratory or intestinal epithelium following single application.

CFTR treatment

Following validation of the nebuliser and direct instillation techniques, we

introduced human CF77? cDNA constructs (Methodology) by nebulisation into the airways and by direct instillation into the intestinal tract of cflcf mice⁶. We chose to complex the CFTR cDNA constructs with the liposome DC-Chol/DOPE since

it has received approval for use in man¹⁵ and because in vitro transfection studies

demonstrated at least equal efficiency to that achieved with DOTAP or

DOTMA/DOPE (data not shown).

4/04669

- 45 -

The results of mRNA-PCR analysis (from the same animals used for the electrophysiological studies) are shown in Fig. 3. Human CFTR mRNA expression

was detected in all animals tested following treatment. The level of transgene

expression was variable within an animal (trachea, peripheral lung, rectum) and

between animals. Qualitatively, similar results were obtained on repeat processing

of the RNA samples. Human CFTR mRNA was not detected in the proximal colon

of either of the treated animals tested (data not shown).

The in vivo response of the nasal epithelium to forskolin (which, in the presence of amiloride, increases chloride efflux principally through changes in

cAMP) is shown in Fig. .- Treatment resulted in a two-fold increase in the mean forskolin response, with restoration of 50% of the deficit between cflcf ^'and +/+ mice, although the mean values of the two groups were not significantly different. Five of 6 treated animals showed nasal forskolin responses greater than the mean

of the cflcf group. No effect of treatment was seen on baseline maximum nasal PD

(cflcf. -20.4 mV (SEM 0.9), treatment: -20.5 V (2.0), +/+: -16.2 mV^'(1.2), p<0.01 cflcf v +/+), or on the amiloride sensitive component of the nasal PD:

(cflcf. 52% (SEM 5), treatment: 59% (5), +/+: 46% (6), p=NS all

comparisons). Responses to the calcium ionophore A23187 (which causes chloride secretion through increasing intracellular Ca ⁺) were not altered by treatment

(cflcf. 2.6 mV (0.5), treatment: 1.9 V (0.9), +/+: 4.3 mV (0.9)).

Consequently, the ratio of cAMP:Ca²⁺ related responses increased from 0.9 in

cflcf animals to 2.7 in treated cflcf animals, in comparison to 1.9 in +/+ mice.

The in vitro responses of the tracheal epithelium are shown in Fig. 5 .

Following characterisation of the components of the forskolin response in both

+/+ and cflcf mice (EWFWA et al, manuscript submitted), we pretreated tissues

with amiloride, DIDS, phloridzin and acetazolamide (to reduce sodium transport, non-CFTR mediated chloride secretion, sodium-glucose cotransport and

bicarbonate secretion respectively) in order to minimise the contribution of non- CFTR components of the forskolin response. The response of untreated cflcf animals to both amiloride and DIDS is altered compared to +/+ mice; treatment

with human CFTR cDNAs substantially restored both amiloride (Figs. 5 a and b)

and DIDS components (PD: cflcf. -0.6 V (0.2), TR: -0.4 mV (0.1), +/+: -0.1

mV (0.2); Isc^: cflcf. -20.9 μA/cm² (3.7), TR: -6.2 μA/cm² (2.7), +/+: + 1.8 A-A/cm² (0.9); cflc v +/+ p<0.01, cflcf \ TR p<0.05) towards +/+ levels. The phloridzin and acetazolamide responses are not different in +/+ and cflcf

animals and were not altered by treatment (data not shown). As in the nasal

epithelium, forskolin responses were increased two-fold by treatment with

restoration of 50% of the deficit. Five of 7 treated animals showed tracheal forskolin responses greater than the mean of the cflcf group, although as^" in the nose, the group means were not significantly different. Responses to A23187 were not substantially altered in treated animals and consequently the ratio of cAMP to

Ca²⁺ mediated responses increased in the trachea (cflcf. 0.16, treated: 0.25, +/+:

0.47). We conclude that treatment produced complete correction of cAMP-related

chloride secretion in the upper and lower airways of some animals (for example

Figs. 4 c and 5k) and partially correction in others (for example Figs. 4b and 5j),

in accordance with the demonstrated presence of human CFTR mRNA.

The forskolin responses of the jejunum and distal ileum (in vitro) and

rectum (in vivo) of control and treated animals are shown in Fig.6 . The jejunum

serves as a region of untreated tissue for each data set and provides an indication

of the inherent variability of the bioelectric responses between animals. Forskolin

responses were unchanged by treatment in the jejunum, but were increased in both

the distal ileum and rectum, resulting in restoration of approximately 20% of the

deficit between cflcf and +/+ animals in both the ileum and rectum. Treatment

also increased the response to carbachol (which produces chloride secretion

through Ca²⁺ mediated pathways) in the ileum (restoration of approximately 25%),

but, as expected, did not change the jejunal response. Stimulation of Na⁺-glucose

cotransport by glucose and its inhibition by phloridzin, which do not differ in the

small intestine of cflcf and +/+ mice, were not altered by treatment.

Discussion

Cystic fibrosis is a suitable disease for treatment by somatic gene therapy as the

principal cellular defect is known, the gene responsible cloned and the major

affected organ, the lung, is relatively accessible to gene delivery. Expression of 94/04669

_ 48 _

transgenes using recombinant adenoviral or liposome mediated transfer has been

demonstrated following direct instillation or nebulisation into normal laboratory animals^16'19. Human trials for treating the lung defect in CF using recombinant

CFTR adenovirus have recently received approval. However, a number of issues,

including the efficacy of complementation by CFTR cDNA and the safety of different gene transfer systems need to be addressed and resolved. The development of genetic models of CF in the mouse has provided such an

opportunity.

Recently, Hyde et al reported human CFTR cDNA-liposome delivery to the ^• Cambridge CF mutant mice". Several features of that study were unexpected, and

differ both from our results and from relevant data in CF patients. In the

Cambridge mice, tracheal baseline bioelectric properties and responses to

amiloride, forskolin and A23187 are all markedly reduced, In contrast, the

characteristic pattern in CF patients includes normal20, or increased9 baseline bioelectric properties, increased amiloride and reduced cAMP mediated responses and relatively well preserved calcium linked chloride secretion²¹. This is also the

pattern of responses seen in the Edinburgh mutant mouse with the exception of the

amiloride effect. Thus, prior to treatment cflcf znά +1+ animals in our study

differed principally in the reduced cAMP mediated chloride secretion. The severe runting and high perinatal mortality characteristic of the Cambridge model make both age and weight-related matching to wild-type controls difficult. The effect of

these two parameters on airway ion transport parameters is unknown. Because

runting is not a feature of the Edinburgh model, mutant animals can be matched for weight and age with their wild-type litter mates.

Gene delivery to the Cambridge mutant mouse was by direct distal tracheal

instillation, deposited at the carina. In contrast, in this study gene delivery was by nebulisation to the entire respiratory tract. Validation with both amiloride and the /3-gal reporter gene suggests that the described nebuliser system can deliver either of these forms of treatment to the entire murine airways. This procedure has

several practical advantages over direct instillation, all relevant to the development

of an effective, non-invasive clinical protocol. DNA is delivered to the entire respiratory tract so that electrophysiological correction can be measured in nasal

epithelium in vivo, of relevance to future trials in man. The risk of physical injury

to the trachea is minimised, deposition is 1'kely to be less focal and is less

dependent on mucociliary clearance for distribution. Finally, delivery can be

repeated easily and with minimal discomfort. This is directly relevant to clinical trials since unless stem cells can be stably transfected, repeated treatment will be needed in man at intervals dependent on the life span of the lung epithelium.

Hyde et al used in situ hybridisation to demonstrate delivery and indicate that

CFTR cDNA can be expressed in the airways of wild-type animals. In contrast, we demonstrate CFTR gene expression directly by RT-PCR on the same tissues from mutant animals on which electrophysiological responses were measured and show that correction was always accompanied by transgene expression. Following

instillation, full correction of the reduced baseline, amiloride, forskolin and

A23187 responses was seen in 4 of the 6 Cambridge animals, the lack of response in the remaining two being attributed to failure of delivery. It is presently unclear

why such a range of transport processes should change following treatment with

CFTR cDNA. No effect was seen in our study on baseline or calcium-mediated responses. However, the basic CF defect of reduced cAMP- ediated chloride secretion could be restored to +/+ levels. Overall, we observed a range of

complementation approximating a normal distribution. Nasal and tracheal restoration of cAMP-mediated chloride secretion were generally well matched

within individual animals.

The specificity of the forskolin response for CFTR mediated chloride secretion in murine trachea is not well defined. We have recently extensively studied the

components of this response (EWFWA et al, manuscript submitted) and suggest

that non CFTR related chloride secretion, as well as other ions may be involved.

We pretreated tracheal tissues with a number of inhibitors aimed at increasing the

specificity of the forskolin response for CFTR mediated chloride secretion. Under these conditions forskolin responses in treated animals demonstrated a consistent

shift towards +/+ values. Interestingly, the reduced amiloride response in the trachea of the Edinburgh mouse was also increased in the treated animals (as in the Cambridge mice), suggesting an important link between sodium and chloride

transport in both these models. Gene delivery was also assessed in our study in the intestinal tract by topical application. The normal level of CFTR expression is

higher in the gut than lung, as is epithelial cell turnover, both likely to contribute to the lower level of correction seen²². This study was designed as a precursor to human application and we believe several identified factors are relevant to such trials of gene therapy for CF in man. Firstly, as in the insertional mouse model electrophysiological end-points (such as

nasal PD²³ and its alteration by drugs) demonstrate a degree of overlap between the normally distributed values of CF and non CF subjects. Unless either large

numbers of patients are studied or gene transfer produces a consistent marked effect, statistical validation of correction, as in this study, may be difficult to achieve. Secondly, the range of correction seen in our study may relate to

variation in aerosol deposition, transfection efficiency or gene expression in vivo.

As in man, airway morphology will vary in individual mice resulting in variable

deposition. A recent study of reporter gene expression in murine airways delivered by nebulisation showed up to four fold differences in the amount of plasmid DNA deposited, but gene expression was more variable (up to 10 fold)¹⁹. Rate of

clearance, site of uptake and cell type specific differences in gene expression were suggested to be important. These factors are likely to be relevant to the variation

in mRNA expression and restoration of the ion transport defect seen in our study,

as well as to clinical trials in man. Thirdly, the airways of these cflcf mice⁶ show many similarities to the pathology found in CF subjects²⁴, but in the animals used in this study, bronchi were not filled with purulent secretions. Effective delivery

is, therefore, more likely than in the diseased human CF airway likely to be first

targeted in trials of gene therapy. Despite such an apparent advantage, complete correction was only achieved in some animals. Finally, the maximal CF7R cDNA dose which could be delivered was limited by practical considerations of liposome and vector production. Nebulisation in CF patients is likely to be more efficient 4/04669

- 52 - than in the mouse, but in view of the greater surface area of the human lung, large

quantities of DNA-liposome complexes will probably be required for effective

treatment.

In conclusion, we have demonstrated human CFTR mRNA expression in the respiratory tract of mutant mice following in vivo non-invasive gene transfer. This was accompanied by restoration of cAMP-mediated chloride secretion in some animals, with an overall two-fold increase in function. This suggests that non-

invasive liposome-mediated transfer of the CFTR cDNA to CF subjects in vivo

may be effective, but also indicates some of the difficulties which may be encountered in human trials. The use of a non-viral system which shows little or

no evidence of tissue damage, or potential to elicit an immune response may offer

significant advantages over currently proposed approaches based on recombinant adeno viruses.

References for Example 3

1. Quinton, P.M. Faseb J. 4, 2709-2717 (1990) .

2. FitzSimmons, S.C. J. Pediatr. 122, 1-9 (1993) .

3. Riordan, J.R. et al Science 245, 1066-1073 (1989) . 4. Rich, D.P. et al Nature 347, 358-363 (1990) .

5. Drumm, M. . et al Cell 62, 1227-1233 (1990) .

6. Dorifλ, J.R. et al Nature 359, 211-215 (1992) .

7. Snouwaert, J.N. et al Science 257, 1083-1088 (1992) .

8. Colledge, .H. et al Lancet 340, 680 (1992) . 9. Knowles, M. et al New Engl. J. Med. 305, 1489-1495 (1981) .

10. Oorin, J.R. et al Transgenic Research 1, 101-105 (1992).

11. Hyde, S.C. et al Nature 362, 250-255 (1993) .

12. Boucher, R.C. et al Sodium absorption in mammalian airways. In P.M. Quinton, J.R. Martinez, and U. Hopfer (Eds.), Fluid and electrolyte abnormalities in exocrine glands in cystic fibrosis. San Francisco:San Francisco Press, Inc., 1982, 271-287.

13. Feigner, P.L. et al Proc. Natl. Acad. Sci. USA 87, 7413- 7417 (1987) .

14. MacGregor. G.R. & Caskey, C.T. Nuc. Acids Res. 17, 2365

(1989) .

15. Nabel, G.J. et al Human Gene Therapy 3, 399-410 (1992).

16. Yoshimura, K. et al Nuc. Acids Res. 20, 3233-3240 (1992) 17. Rosenfeld, M.A. et al Cell 68, 143-155 (1992).

18. Hazinski. T.A. Seminars in Perinatology 16, 200-204

(1992) .

19. Stribling, R. et al Proc. Natl. Acad. Sci. USA 89, 11277- 11281 (1992) . 20. Alton, E.W.F.W. et al Thorax 47, 1010-1014 (1992) .

21. Welsh, M.J. Faseb J. 4, 2718-2725 (1990) .

22. Porteous, D.J. & Dorin, J.R. Hum. Mol. Gen. 2, 211-212 (1993) . 23. Alton, E.W.F.W. et al Eur. Respir. J. 3, 922-926 (1990) .

24. Geddes. D.M. and Graham, A. Cystic Fibrosis. In D.M. Mitchell (Ed.) Recent Advances in Respiratory Medicine. Churchill Livingstone, 1991. 203-227.

25. Mentz. W.M. et al Am. Rev. of Resp. Dis. 134, 938-943 (1986) .

26. Sambrook, J., Fritsch, E.F., and Maniatis, T. Molecular Cloning:A Laboratory Manual, New York:Cold Spring Harbor Laboratory Press, (1989) .

27. Feldman, D. et al (Am. J. Respir. Cell Mol. Biol. 5, 416- 423 (1991) .

28. Gao, X. & Huang. L. Biochem. Biophyε, Res. Comm. 179, 280-285 (1991) .

29. Whitsett, J. A. et al Nature Genetics 2, 13-20 (1992) .

30. Dorin, J.R. et al Science 243, 1357-1360 (1989) . 31. Yamaguchi, M. et al Nuc. Acids Res. 19, 2403-2408 (1991) .

32. Tata. F. et al Genomics 10, 301-307 (1991) .

33. Widdicombe, J.H. et al Proc. Natl. Acad. Sci. USA 82, 6167-6171 (1985) .

34. Smith, S.N. et al Clin. Sci. 82, 667-672 (1992) .

Claims

1. A mammal other than man having a CF-related defect in at least one copy of Cftr.

2. A mammal according to claim 1 which is a rodent.

3. A mammal according to claim 2 which is a mouse.

4. A mammal according to any one of claims 1 to 3 having at least one of a null-mutation a ΔF508 deletion, a missense mutation and a nonsense mutation, in at least one copy of Cftr.

5. A mammal according to any one of claims 1 to 4 having an insertional disruption in at least one copy of Cftr.

6. A mammal according to claim 5 wherein the insertional disruption is in exon 10 of Cftr.

7. A mammal according to any one of claims 4 to 6 wherein the insertional disruption comprises stop codons in all possible reading frames and transcriptional stop signals.

8. A mammal according to any one of claims 1 to 7 which is heterozygous for Cftr.

9. A mammal according to any one of claims 1 to 7 which is homozygous for Cftr.

10. A mammal according to any one of claims 1 to 7 having at least one of a null-mutation, a ΔF508 deletion, a missense mutation and a nonsense mutation in one copy of Cftr and a null mutation in the second copy of Cftr.

11. A process for producing a mammal according to any one of claims 1 to 10 comprising targeting an insertion into a Cftr gene in non-human mammalian embryonal stem ceils, returning the thus treated cells to blastocysts, implanting the blastocysts into pseudopregnant females, obtaining chimaeric offspring from such females and screening such offspring for the ability to transmit a CF- related defect to their progeny.

12. Use of a mammal according to any preceding claim in assessing the potential benefit of an agent for treating or preventing CF in humans.

13. Use of a heterozygous mammal according to claim 8 or claim 10 in breeding a homozygous mammal according to claim 9.