Published ahead of print on May 5, 2008, doi:10.1165/rcmb.2007-0385OC
© 2008 American Thoracic Society DOI: 10.1165/rcmb.2007-0385OC Validation of Nasal Potential Difference Measurements in Gut-Corrected CF Knockout Mice1 Department of Gene Therapy, Faculty of Medicine at the National Heart and Lung Institute, Imperial College London, London, United Kingdom; and The UK Cystic Fibrosis Gene Therapy Consortium Correspondence and requests for reprints should be addressed to Dr. Uta Griesenbach, Department of Gene Therapy, Imperial College London, Manresa Road, London SW3 6LR, UK. E-mail: u.griesenbach{at}imperial.ac.uk
Attempts at correcting the nasal potential difference (PD) in cystic fibrosis (CF) mice have long been used in preclinical gene and small molecule therapy development. However, in general, CF mice suffer from intestinal disease, are runted, and have high mortality rates; they are therefore difficult to work with, especially if large numbers are required. Because of this, large-scale PD studies in CF mice have not been performed. Working with CF mice has become substantially easier after the generation of the gut-corrected CF-knockout mouse. Fatty acid–binding promoter (FABp)-mediated expression of CFTR in the gut, but not the airways, prevents the intestinal disease of the CF knockout mouse. This model has given us the unique opportunity to systematically study PDs in large numbers of CF mice. The nose, but not the lungs, of these animals mimic the bioelectric defect seen in humans. We have therefore assessed the bioelectrics of the respiratory epithelium comparing FABp-CF and wild-type mice. The large body of data gathered in CF and wild-type mice allowed us, for the first time, to establish power calculations that should inform sample sizes required in gene and small molecule therapy development. In addition, we address the important issues of intra-animal variability as well as intra- and inter-operator variability for scoring the traces, and the effect of age and sex on nasal PD in CF mice. These data should allow a more informed use of CF animals in future studies.
Key Words: cystic fibrosis • nasal potential difference • CF mice • gut-corrected
Cystic fibrosis (CF) is a lethal, monogenic disease (reviewed in Ref. 1). New therapies, including gene therapy, and small molecule pharmacology are currently being developed (reviewed in Refs. 2, 3). To aid pre-clinical drug development and to better understand disease pathophysiology, several CF knockout mouse models have been developed. In these mice the cystic fibrosis transmembrane conductance regulator (Cftr) gene, which encodes an epithelial chloride channel, is either completely knocked out (nulls) or naturally occurring mutations such as It is well documented that the altered sodium and chloride transport in CF airway epithelial cells leads to changes in potential difference (PD) across the apical membrane of epithelial cells. The baseline PD, predominantly a readout of sodium absorption, and the response to the sodium channel inhibitor amiloride is increased in CF, whereas the response to perfusion with a low chloride solution, which induces chloride flux via CFTR, is reduced or absent.
To assess the effects of gene and small molecule therapy on the respiratory tract, it is useful to use a strain that has very low, or absent, residual CFTR function such as the UNC-null (7), G551D (8), or Importantly, Zhou and coworkers have generated a double-transgenic CF knockout mouse, which is based on the UNC-null mouse (CFTRtm1Unc), but also carries a human CFTR (hCFTR) transgene under the control of the rat intestinal fatty acid–binding protein gene promoter (FABp) (11). These mice express human CFTR in the gut and, consequently, do not develop intestinal disease. In contrast to other CF knockout mice, homozygote FABp-CF male and female mice breed, avoiding the need for genotyping and culling of unwanted heterozygote and wild-type littermates. Data on nasal PD measurements in FABp-CF mice are limited. Steagall and colleagues (12) reported that FABp-CF mice have a response to low chloride perfusion similar to that of CF-null mice (CFTRtm1Unc), but n numbers were low (four per group). In addition, it was unclear at what position in the mouse nose the PD was measured. This is relevant because the relative distribution of respiratory and olfactory epithelium varies with distance from the snout (13). In general, the respiratory epithelium predominates in the proximal nose (< 3 mm from the tip) and the olfactory epithelium predominates farther back (> 5 mm). In the context of assay development, it is important to ensure that PD measurements are taken in a region predominantly composed of respiratory cells, because these, rather than the olfactory epithelium, are the likely target cells for CF lung therapy. Here, we have assessed nasal PD in FABp-CF mice in the nose and compared values to those of wild-type mice. FABp-CF mice were maintained on a mixed genetic background and, therefore, genetically matched wild-type mice were not available. To ensure that differences in PD were Cftr specific and not strain dependent, we therefore, first compared PD in C57Bl/6 congenic G551D CF mice and C57Bl/6 wild-type mice, and subsequently compared PD in C57Bl/6 congenic G551D and FABp-CF mice. The generated mean values and standard errors were used to carry out power calculations that may aid future study design. We also compared PD in males and females, as well as in young (< 3 mo) and older (> 5 mo) mice, to select a suitable sex and age range for the pre-clinical screening. We quantified intra-animal variability on repeat measurements, as well as the intra- and inter-operator variability for scoring PD traces. PD measurements are unreliable in damaged epithelium and we also assessed if pre-treatment with diluents before PD measurement affected the reading. This is relevant because gene as well as small molecule therapy will require administration of DNA and drugs before the assessment of efficacy using PD measurements. These studies should provide useful data to allow optimization of novel therapy assessment.
Mouse Models Male and female wild-type C57Bl/6 mice and G551D CF knockout mice (8) were purchased from Charles River UK (Margate, Kent, UK). Breeding pairs of gut-corrected FABp-CF-knockout mouse (11) were purchased from Drs. M. Drumm and A. van Heeckeren (Case Western Reserve University, Cleveland, OH) and bred at Imperial College, London. Consistent with the known expression pattern of FABp, expression of human CFTR in the nose, while detectable, was 3 logs lower than in the gut (data not shown). All experiments were performed with approval of appropriate local Ethics Committees and according to Home Office regulations, and adhere to AJP guidelines.
PD Measurements in the Mouse Nose
Intra-Animal Variability
Variability of Operator Interpretation of the Bioelectric Data
Nasal Perfusion and Bolus Administration of Diluent Perfusion. Pre-perfusion PD and post–saline perfusion PDs were measured in the same animals. Pre-PD was measured at least 14 days before saline perfusion, to allow the epithelium to recover from possible damage induced during the PD measurements. Post-perfusion PD was performed at indicated times after perfusion. Saline (100 µl) was slowly (4 µl/min over 15 min) perfused onto the nasal epithelium using a thin (0.5 mm outer diameter) polythene catheter. Bolus. To assess of the effect of bolus administration of saline or HEPES-Krebs buffer, 100 µl were either administered as a single bolus to the nasal epithelium or as 4 x 25-µl aliquots (delivered in 15-min intervals). The animals were anesthetized as described above and the diluent was slowly dropped onto the nostrils using a standard Eppendorf pipette. The mice "sniffed" the liquid into the nose. Post-diluent PD was measured at indicated time points after bolus administration.
Statistical Analysis
Comparison of Nasal PD in Wild-Type, GD-B6, and FABp-CF-Knockout Mice Transepithelial nasal potential difference was measured in C57Bl/6 wild-type (WT-B6), C57Bl/6 congenic G551D (GD-B6), and FABp-CF mice. Baseline PD and amiloride responses were significantly (P < 0.005 and P < 0.01, respectively) increased in GD-B6 and FABp mice compared with WT-B6. As expected, the low chloride response was also significantly (P < 0.005) decreased in both strains of CF mice compared with WT-B6. There was no difference in nasal PD between G551D and FABp CF mice (Figure 1).
Baseline and low-chloride means and standard deviations of WT-B6 and FABp mice were used to carry out power calculations assuming a treatment effect of 10, 20, 30, 40, 50, 75, and 100% toward wild-type values (Table 1).
Electrical measurements in the mouse nose are technically difficult, even for experienced operators with small movements of the catheter markedly influencing the recording. We, therefore, assessed the percentage of recordings interpretable for each measurement. In wild-type mice (n = 17), between 90 and 100% of traces were measurable for baseline, amiloride, and low-chloride PD. In CF mice (n = 18), this percentage was similar for baseline PD and amiloride responses, but dropped to 80% for low-chloride values. This was likely due to the fact that an increase in chloride transport is easy to recognize in WT mice, whereas the precise point of change of the CF PD is harder to see if the trace is slightly noisy. Representative interpretable and noninterpretable FABp low-chloride responses are shown in Figures 1D–1F, respectively. Baseline, amiloride, and low-chloride responses can easily be determined in Figure 1D. Figure 1E, however, clearly highlights a limitation of the assay. Although baseline and amiloride responses can be quantified reliably, the low-chloride response wanders with no obvious pattern in a range of 5 mV. This problem is further enhanced in Figure 1F and leads to the trace not being interpretable for the low-chloride response.
PD Measurements Are Age- and Sex-Independent
Intra-Animal Variability To assess intra-animal variability after repeated PD measurements, FABp-CF mice underwent three measurements (Days 1, 14, and 28). Mean baseline PD and low-chloride response did not vary significantly at the different time points (Figure 4). However, the average %CV reached 21% and 41% for baseline and low-chloride measurements, respectively, indicating a higher variability in low-chloride responses compared with baseline PD.
Inter- and Intra-Operator Variability To assess inter-operator variability in scoring PD traces, three experienced operators were asked to score 22 FABp-CF traces for baseline and low-chloride values, and the mean %CV was calculated. Interestingly, the %CV was very low for scoring of baseline values (1.3%CV), indicating very high reproducibility between operators. However, for the low-chloride response the variability was much higher (36.0%CV). To assess intra-operator variability in scoring PD traces, a set of 30 traces was scored by two operators twice, approximately 3 months apart. Similar to inter-operator variability, the %CV was very low for baseline measurements (1.22 and 0.41%CV for operators 1 and 2, respectively), indicating very high reproducibility of the quantification, whereas quantification of the low-chloride response was more variable (14.8 and 22.9%CV for operators 1 and 2, respectively).
Effect of Pre-Treatment on Nasal PD
A time-lapse of 24 hours between administration and assessment of therapeutic effects is appropriate in the context of gene therapy, but may be too long for small molecule drug development. We therefore assessed whether bolus administration of saline to the nostril, which is rapidly "sniffed" and does not require insertion of a catheter, may be more appropriate. FABp mice received 100 µl saline as a bolus 1 hour before PD measurements; this significantly (P < 0.01) reduced baseline PDs (Figure 5D). We next assessed whether HEPES-Krebs (HK) buffer (HK, pH 7) might be better tolerated than saline (pH 5.4). There was no significant difference in mean baseline-PD between HK buffer–treated and untreated CF mice, although variability was higher in the former. In an attempt to further reduce diluent-induced toxicity, we split the 100-µl bolus into 4 x 25-µl aliquots, which were administered in 15-minute intervals over 60 minutes. Baseline PD was measured 30 minutes after the last administration and, importantly, was not different from that in untreated animals (Figure 5D). This indicates that small-volume bolus administrations might be suitable to administer small-molecule drugs, which require assessment of efficacy shortly after topical administration.
The generation of gut-corrected CF knockout mice (FABp mice) has allowed larger-scale studies in CF mice to be initiated. Here, we validated nasal PD measurements in this model as a prelude to their use in a core facility. The availability of large numbers of CF mice allowed us for the first time to study animal and operator variability, assess technical failure rates, and determine if sex and age affected nasal PD. Comparison between wild-type and CF PDs allowed us to carry out power calculations for predicted therapeutic effects ranging from 10% to 100% correction toward wild type, which is relevant to the design of future gene therapy or small-molecule drug-based CF mouse studies. As expected, baseline and amiloride responses were significantly increased in FABp when compared with wild-type mice, and the low-chloride response was significantly decreased. Similar to previous studies in mice and human (14–16) the low-chloride response was the optimal discriminator between wild-type and CF mice, whereas most overlap was noted in the amiloride response. Both baseline and amiloride PDs are mainly read-outs of transepithelial sodium transport, and we therefore focused on baseline PD for the purpose of this study. Our FABp-CF breeding colony was established in January 2002 with breeding pairs purchased from A. van Heeckeren and M. Drumm (Case Western Reserve University). At that time, only animals maintained on a mixed genetic background were available. We were aware that the absence of genetically matched wild-type littermates might affect the interpretation of our results. To address this point, we compared nasal PDs of FABp-CF mice, C57Bl/6 congenic G551D mice, and C57Bl/6 WT mice. Importantly, both strains of CF mice were significantly different from WT for all PD-related measurements, and there was no difference between FABp and G551D CF mice. From this we conclude that the mixed genetic background of the FABp mice did not affect interpretation of our data. Animal numbers in the G551D cohort were lower than in WT and FABp cohorts due to the restricted availability and poor health of these mice. The low-level expression of human CFTR in the nose of FABp mice did not appear to affect the nasal PD measurements compared with GD-B6 mice, which express no human CFTR.
Baseline and low-chloride responses were identical in male and female mice, as well as in young (< 3 mo) and older (> 5 mo) FABp mice. In contrast, a study by Colledge and coworkers reported that short circuit current forskolin responses in the trachea of
To the best of our knowledge this study has, for the first time, allowed sample size calculations for a range of anticipated treatment effects. These may provide useful guidelines for the design of future studies aimed at increasing chloride transport or decreasing sodium absorption. The power calculations indicate that large animal numbers are required to detect small changes in baseline PD and low-chloride Nasal PD measurements may, therefore, have limited value when assessing chloride channel modulators with anticipated small treatment effects. Large n numbers (see power calculations) may to a degree overcome this limitation, but may not always be practical. Importantly, n numbers predicted by the sample size calculations are far higher than numbers traditionally included in gene therapy and small drug-based pre-clinical research, which may lead to false-negative interpretation of low-level CFTR function. This is particularly relevant when considering that there is good evidence that even low levels of residual CFTR function may ameliorate CF lung disease. Thus, approximately 5 to 10% of wild-type cells can correct chloride transport on a CF background (1, 17). In addition, Dorin and colleagues showed that approximately 5% of the normal level of Cftr gene expression per cell results in a disproportionately larger correction of the chloride ion transport defect (50% of normal) and complete rescue of the intestinal disease (100% survival) (18). Most importantly, patients with "mild" CF mutations such as R117H who have some residual CFTR function (19) generally have less severe lung disease. Sample size calculations further support the notion that the low chloride response provides the optimal discrimination between WT and CF mice. For any given treatment effect, sample sizes were lower for the detection of significant changes in the low-chloride response. This may in part explain why gene transfer has shown significant partial correction of the chloride transport defect in mice (20) and human (21, 22), whereas reduction of sodium absorption has not yet been reported. Good quality bioelectric recordings are crucial for interpretation of treatment effects, but even in the hands of experienced operators approximately 20% of traces did not allow accurate scoring of the low-chloride responses. This was mainly due to movement of mice or the electrode during the procedure. Sample size calculations have to be adjusted for this failure rate to ensure adequate power. In addition to trace quality the accurate scoring of the responses is crucial. This study for the first time assessed intra- and inter-operator variability in scoring traces. Inter-operator variability for scoring baseline PD was remarkably low (1.3%CV), whereas the CV reached 36% for the low chloride response, despite using experienced operators (4 to 15 yr experience) and strict scoring guidelines. Assessment of intra-operator variability was similar with baseline PD having a markedly lower variability than low-chloride responses (0.8% and 18.9%, respectively). In addition, we also looked at inter-animal variability and determined that although the overall group mean for baseline and low-chloride responses did not change on repeat measurements over 28 days, the CV on repeat measurements was high (41%CV) for the low-chloride response but within a more acceptable range (21%CV) for the baseline PD. This variability is most likely due to physiologic changes in ion transport. We also assessed the effects of diluent perfusion and bolus administration to the nasal epithelium on PD. After perfusion with saline, baseline PD was reduced for at least 2 hours after dosing, but was unaffected if saline was perfused 24 hours before the PD measurement. Subtle epithelial damage caused by the catheter or the diluent, such as increased paracellular permeability, may explain these results. This suggests that after perfusion, PD measurements should not be performed within the first 24 hours. While we only assessed the effects of saline perfusion, which may not be predictive of other diluents, the data highlight that pilot experiments to assess the effects of diluent perfusion should always be performed. Finally, a 24-hour window between dosing and measurement may not be appropriate for short-acting drugs. We therefore attempted to reduce this time interval by administering the diluent as a bolus (no catheter) and were able to establish a suitable protocol, which will allow PD measurements 30 minutes after drug administration. In conclusion, the availability of large numbers of gut-corrected CF knockout mice has allowed us to accumulate a large number of nasal potential difference measurements in the mice. These data have allowed us to assess a number of potentially important variables relevant to this technique, as well as highlighting the potential for false-negative results related to the presented power calculations. While clearly cAMP-mediated stimulated chloride secretion would be the optimal assay, our experience suggests that this is not CFTR-specific in the murine lower airway. The response to low chloride will clearly recruit signals from any open nonselective channel, by providing a driving force for chloride secretion. This is of interest because in humans, this appears to allow good discrimination between CF and non-CF genotypes, suggesting that CFTR or CFTR-regulated anion channels are predominantly assessed.
The authors thank Drs. A. van Heeckeren and M. Drumm (Case Western Reserve, Cleveland, OH) for making FABp-CF breeding pairs available and Lucinda Somerton (Imperial College) for help with preparing the manuscript.
* These authors contributed equally to this work.  This work was funded by the Cystic Fibrosis Trust and the CF Trust Dr Benjamin Angel Fellowship (U.G.). Originally Published in Press as DOI: 10.1165/rcmb.2007-0385OC on May 5, 2008 Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form October 25, 2007 Accepted in final form March 31, 2008
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Top
Abstract
MATERIALS AND METHODS
F508 or G551D have been introduced (
