Airway disease currently causes most of the morbidity and mortality in patients with cystic fibrosis (CF). However, understanding the pathogenesis of CF lung disease and developing novel therapeutic strategies have been hampered by the limitations of current models. Although the gene encoding the cystic fibrosis transmembrane conductance regulator (CFTR) has been targeted in mice, CF mice fail to develop lung or pancreatic disease like that in humans. In many respects, the anatomy, biochemistry, physiology, size, and genetics of pigs resemble those of humans. Thus pigs with a targeted CFTR gene might provide a good model for CF. Here, we review aspects of porcine airways and lung that are relevant to CF.
cystic fibrosis (CF) is a common, autosomal recessive disease caused by mutations in the gene encoding the cystic fibrosis transmembrane conductance regulator (CFTR) anion channel (53, 262). CF involves multiple organs, including the pancreas, sweat glands, vas deferens, intestine, and liver. However, airway disease is currently the source of most CF morbidity and mortality; airway inflammation and infection, especially with Pseudomonas aeruginosa, ultimately destroys the CF lung. Nineteen years after identification of the CFTR gene, controversies still surround the pathogenesis of airway disease. We lack answers to many questions, current treatments are inadequate, and CF remains a lethal disease.
A major impediment to understanding CF pathogenesis and to developing new treatments is the limitation of current animal models. Although CF mice have been produced, during their limited lifespan they do not develop airway disease resembling that typically found in humans (87, 88). Although there are many hypotheses, the reasons that CF mice fail to manifest the same airway disease as humans remains unknown. Frustrated by the inability to solve persistent research problems in CF and the inability to offer better treatments to patients in CF clinics, basic and clinical researchers have realized the need for a new animal model. The sheep (95, 265), monkey (268), ferret (141, 142, 220), and pig (191) have each been studied in an attempt to develop a new CF animal model. Each has advantages as a model.
Swine have become an important resource in biomedical research. They are excellent models for cardiovascular disease (247), obesity (66), diabetes (21, 136), alcoholism (258), hypertension (66), skin physiology (212), lipoprotein metabolism (81), intestinal function (59), nutrition (160), and injury and repair (271). The swine lung has become an excellent model for the normal human lung, for abnormalities in diseases, and for therapeutics. For example, porcine lungs have been used to study surfactant composition (215), surfactant function and therapy (101), lung development (82), lung transplantation, perfluorocarbon liquid ventilation (57), reperfusion injury (34), nitrous oxide effects on lung function (203), hyperoxia-induced and other toxin-induced lung injuries (89), pulmonary artery hypertension (24), endothelin and growth factor receptor biology (227), lung growth after lobectomy (125), bronchiolitis obliterans (2), effects of mechanical ventilatory modes (143), airway hyperresponsiveness (161), asthma (119), and many other diseases. Pigs have been employed to evaluate viral and nonviral vector-mediated gene transfer to the lung (48, 151, 162). The longevity of pigs (Table 1) also offers opportunities for investigating the pathogenesis of lung disease, the long-term therapeutic efficacy of treatments, and adverse effects of interventions that might only become apparent with time. The size of a porcine CF model would also allow testing of many interventions used in humans. Perhaps of greatest note, the similarity of pig and human organs has led to a large effort to develop them as a source of organs for xenotransplantation (44–46, 286). For many of these reasons, we chose to investigate the pig as a model for CF lung disease (191).
The purpose of this review is to provide information about aspects of porcine airways and lung that are relevant to CF. We focus on the lung because of its importance for CF. We hope that collecting and reviewing this information will help provide a resource and foundation for investigators interested in new models of CF.
REPRODUCTION AND GENETICS
Reproduction by Pigs
The reproductive characteristics of swine are favorable for their use as a CF model (Table 2). The gestation period for a domestic pig is ∼114 days and sexual maturity is reached in 6–8 mo with litter sizes ranging from 8 to 12 piglets. With the potential for nearly 3 deliveries per year, a single sow can produce 24–36 pigs per year. This relatively fast maturation rate and the large number of offspring generated from a single sow in 1 yr allow a colony to rapidly expand.
The porcine genome is comprised of 18 autosomes and 2 sex chromosomes and has a similar size and complexity to the human genome. Completion of the porcine genome sequencing project is expected in 2008, with public release and annotation following soon after (115). Numerous porcine genomic and expressed sequence tag (EST) sequences have been deposited in GenBank and other databases. Furthermore, extensive quantitative trait loci, linkage and physical maps, single nucleotide polymorphisms, and expression data are available from a variety of sources (39). Knowledge of the swine transcriptome has been advanced using tissue-specific and broad coverage microarrays from both commercial and academic sources (246).
Genetically Modified Pigs
Transgenic pigs were first described in the mid-1980s with the production of pigs that expressed various hormones, including human growth hormone (94). Since then, many genes have been expressed in pigs for agricultural and biomedical purposes (183). Two examples are the transgenic overexpression of a mutant rhodopsin gene to produce a porcine model of retinitis pigmentosa (180) and a transgenic pig expressing human decay accelerating factor (201), a complement inhibitor, made in an attempt to prevent complement-mediated hyperacute rejection following transplantation of porcine organs into humans.
Developing gene-targeted pigs has been more challenging. Mice with targeted modifications of their genome can be produced using homologous recombination in mouse embryonic stem (ES) cells, transferring the ES cells to blastocysts, producing chimeric mice in which the ES cells contribute to the germ line, and then breeding them to generate the desired gene-targeted mice. However, despite many attempts in other species, functional ES cells have been developed only for mice (36, 37). An alternate approach to generating gene-targeted animals became possible with the birth of Dolly, the first cloned sheep, in 1997 (266). In the ensuing 10 yr, somatic cell nuclear transfer (SCNT) has been used to clone numerous other animal species, including pigs (181). Production of the first gene-targeted pigs arose from efforts to develop porcine organs for xenotransplantation (49, 134). The target was the porcine α1,3-galactosyltransferase gene because the enzyme adds specific sugar groups to porcine cells causing them to be immunogenic and contribute to hyperacute rejection when transferred to primates.
Figure 1 shows a general strategy for generating a gene-targeted pig or other mammal. Cultured fetal fibroblasts have been successfully used for homologous recombination and as nuclear donors for pigs (49, 134), although other cells may also be suitable. Because classical homologous recombination in somatic cells is very inefficient (204, 265), investigators who targeted the α1,3-galactosyltransferase gene employed a promoter-trap strategy to improve gene targeting efficiency by decreasing the number of random integration events. This strategy requires that the target gene be expressed in the donor cell. The investigators then used the fibroblasts as nuclear donors for SCNT to produce pigs heterozygous for a null allele of the α1,3-galactosyltransferase gene. In addition to this porcine gene, other targeted mammalian (not including mouse) genes include the prion protein gene PRNP in cows and goats (188, 279), the immunoglobulin mu (IGHM) gene in cows (133), and the α1-procollagen gene in sheep (153).
We (191) recently reported the generation of CFTR-null and CFTR-ΔF508 heterozygote pigs. We used homologous recombination to disrupt exon 10 of one porcine CFTR allele in fibroblasts derived from a male fetus. Exon 10 encodes a critical region of nucleotide-binding domain 1 (NBD1), which is required for normal CFTR activity. The disruption was produced by inserting a neomycin-resistance cassette. The insertion also introduced a premature stop codon at amino acid position 508. These changes will prevent the production of functional CFTR. Using a similar strategy, we also deleted 3 bp to eliminate Phe508 within one porcine CFTR allele in fetal fibroblasts. CFTR-targeted fibroblasts were used as nuclear donors for SCNT. Figure 2 shows one litter comprising six male clones that were heterozygous for a disrupted CFTR gene.
PORCINE WILD-TYPE CFTR AND CFTR-ΔF508
Porcine and Human Wild-Type CFTR
Developing a porcine model of CF requires some knowledge of porcine CFTR (pCFTR). Both human (hCFTR) and pCFTR form anion channels that open following phosphorylation by cAMP-dependent protein kinase (175). The pCFTR single-channel conductance (6.7 pS) and open channel probability (0.39) are similar to those of human CFTR (8.3 pS and 0.39) measured in phosphorylated channels with 1 mM ATP. In addition, CFTR of both species localizes to the apical membrane when expressed in polarized, well-differentiated airway epithelia. The biosynthetic maturation of hCFTR and pCFTR can be monitored by following changes in the electrophoretic pattern of protein (40, 86). For both species, CFTR in the endoplasmic reticulum (ER) migrates as a core-glycosylated immature protein (band B), which matures in the Golgi complex to a fully glycosylated form (band C) that travels to the cell surface (Fig. 3).
Porcine and Human CFTR-ΔF508
The most common CF-causing mutation deletes phenylalanine at position 508 (ΔF508) (53, 262). This mutation disrupts processing of the protein so that nearly all hCFTR-ΔF508 is retained in the ER and degraded. As a result, very little of the band B form can be detected. Unlike hCFTR-ΔF508, pCFTR-ΔF508 produces some band C protein (Fig. 3), suggesting that a fraction escapes ER retention and traffics to the Golgi complex (175). Consistent with these findings, some pCFTR-ΔF508 can be detected at the apical membrane of airway epithelia, whereas hCFTR-ΔF508 is expressed diffusely throughout the cell. In addition to the processing defect, the ΔF508 mutation reduces the open state probability of hCFTR to 27% and pCFTR to 46% of the corresponding wild-type channels. The net effect of CFTR processing, localization, and function can be assessed by measuring transepithelial current in CF airway epithelia. hCFTR-ΔF508 generated only 6–7% of the Cl− current of wild-type hCFTR. In contrast, pCFTR-ΔF508 generated 25% of the wild-type pCFTR.
Thus the processing defect of hCFTR-ΔF508 is partially attenuated in pCFTR-ΔF508, and structural differences between human and porcine CFTR are likely responsible (140, 175). The pCFTR amino acid sequence is nearly 93% identical to that of hCFTR, with the regions of greatest sequence difference being at the NH2- and COOH-terminal portions of NBD1 (139). Determining the effects of differences in these regions may lead to a better understanding of the functional interaction of domains within CFTR and, more importantly, could drive the search for drugs to target both the processing and the channel defect. In addition, generation of a CFTR-ΔF508 pig could provide a model in which to learn how the mutation alters protein synthesis and function and give the field a model to evaluate therapeutic strategies for correcting CFTR-ΔF508 defects.
Anatomy and Morphology of the Porcine Lung
The porcine lung has two lobes on the left side and four lobes on the right side (65, 166). The lobes on the left side are designated as the left cranial lobe and left caudal lobe. The left cranial lobe is further subdivided into cranial and caudal components that are demarcated by the cardiac notch. The four lobes on the right side are designated as right cranial lobe, right middle lobe, right caudal lobe, and right accessory lobe. The right cranial lobe ventilation is unique in that it is connected directly to the trachea by the tracheal bronchus proximal to the tracheal bifurcation. The comparable volumes of porcine and human lungs are thought to be advantageous for the study of respiratory function (120).
Porcine and human lungs are characterized by well-developed lobularity with each lobule demarcated by collagenous interlobular septa that are relatively thick compared with some other species (155). The interlobular collagenous network is incomplete in the human lung, which allows for collateral ventilation between lobules (248). In the porcine lung, the interlobular septa are functionally “complete” inhibiting collateral ventilation, but collateral ventilation within the lobule may still occur (248, 274). The pleura of both human and porcine lungs are relatively thick and collagenous and have a vascular supply originating from the bronchial arteries that eventually form anastomoses with circulation of the pulmonary artery (155, 248). The vascular circulation of both humans and pigs have a common pattern of organization and development; however, the maturation rate in the porcine lung is rapid and reflective of the differential rates of overall body growth. For example, the porcine neonate may take only 1 wk to double birth weight whereas an infant may take 6 mo (98, 187).
The general anatomic distribution of the porcine airways recapitulates that of the human with some minor exceptions. Cartilage surrounding porcine airways is more widespread in distribution than in humans, extending farther down in the tracheobronchial tree (98). In perinatal pigs, dissection of airways in the left caudal lobe to within 5 mm of the pleura consistently yielded 19 generations of bronchi (187). Yet, the pig has only 3 generations of bronchioli after the last cartilage plate before reaching the lobule; this compares with ∼10 generations in humans (98). Other minor differences include longer porcine terminal bronchioles than in humans as well as shorter and less well-defined porcine respiratory bronchioles (98, 155).
In neonatal pigs, total lung volume is generally less than in human infants whereas alveolar development is generally more advanced (52, 98). At birth, porcine respiratory bronchioles are typically connected to paired alveolar ducts whereas human respiratory bronchioles are often connected to two or three saccules (primordial alveolar ducts), consistent with their differential maturation (98). Furthermore, the average length of all lung tissue distal to the terminal bronchiolus (i.e., the acinus) is ∼5 mm in the pig and ∼11 mm in the human infant. Postnatally, porcine alveoli undergo a rapid multiplication phase in the first 2–4 wk of life that is comparable with the alveolar multiplication seen during the first 3 yr in human infants (52, 98). Whereas the morphological development of pulmonary architecture in the pig is comparable with humans, its rate of maturation is rapid.
Porcine lungs have cellular lineages and composition that are comparable with human lungs (18, 120, 149, 159, 270). In the development of the porcine lung (gestation ∼114 days), Baskerville (18) observed differentiation of bronchial epithelia into ciliated cells, goblet cells, and basal cells by 80 days gestational age without morphological detection of submucosal glands (SMGs). By 92 days gestational age, the primordial SMGs were first observed as early downgrowths into the lamina propria with rare extension into the smooth muscle layer (18). Porcine SMGs continue to develop in maturation and number through gestation and into the postnatal period (Fig. 4), which is comparable with gestational and postnatal developmental regulation observed for human SMGs (18, 219). With maturation, the bronchial-bronchiolar junction becomes a major transition zone in the airway mucosa as there is an abrupt decrease in SMGs in smaller airways (6). In the porcine lung, this transition zone is macroscopically detectable as airways with an outer diameter of ∼1 mm. Thus selection of larger proximal airways (e.g., 2–4 mm outer diameter) are often used for isolation and assessment of SMGs in the pig (5, 106, 109, 242). Porcine SMGs have been classified using video microscopy into three broad subtypes based on the morphology of the terminal collecting ducts. These three subtypes have been described as antral, linear, or convoluted ducts (9, 107). Antral ducts are the most prominent and contain cilia that exhibit functional activity. All subtypes appear to be responsive to acetylcholine, but differences in extent of secretory activity between the SMG subtypes have not been discerned (107).
Transepithelial Electrolyte Transport by Porcine Airway Epithelia
Defective epithelial electrolyte transport is a hallmark of CF. Abnormal ion transport was first discovered in the sweat gland, where measurements of defective Cl− transport continue to be a diagnostic mainstay. The sweat gland has also taught us much about the mechanisms of transport in CF (184). However, lung disease is currently the major source of CF morbidity and mortality, and therefore substantial attention has focused on airway epithelia. Yet, the contribution of defective airway epithelial electrolyte transport to the pathogenesis of disease remains controversial. In addition, there is a critical lack of information about the human CF lung during the postnatal period, precisely when loss of CFTR may be initiating disease. This section will review what is known about electrolyte transport in porcine airway epithelia.
Electrolyte Transport Across Nasal Epithelia
Measurement of transepithelial voltage (Vt) across nasal epithelia in vivo is a frequently performed procedure for assessing CFTR function in humans and mice (217). Nasal Vt measurements often serve as a supplemental diagnostic test for CF in humans. Nasal Vt is also used to evaluate the response to interventions, including potential pharmaceuticals. Thus assaying nasal Vt might be a useful procedure for assessing a CF pig.
Salinas et al. (197) measured an average nasal Vt of approximately −15 mV in the pig. Amiloride reduced Vt, suggesting the presence of epithelial Na+ channels (ENaC). Subsequent perfusion with a solution containing a low Cl− concentration and then with forskolin caused Vt to hyperpolarize, indicating the presence of a constitutive Cl− conductance (GCl) that became larger when cAMP levels increased. These measurements qualitatively and quantitatively resemble those in humans. In contrast to its effect on hCFTR, the inhibitor CFTRinh-172 failed to reduce nasal Vt in the pig (197) and was not very effective at inhibiting GCl in cultured porcine tracheal epithelia (146, 234). As indicated below, GlyH-101 may be a more effective inhibitor of pCFTR.
Electrolyte and Liquid Transport by Tracheal Epithelia
The size of pigs has allowed investigators to excise the trachea and large airways to measure salt and liquid movement across native airway epithelia. Freshly excised tissue can provide a significant advantage because it allows experimental manipulation yet may closely resemble in vivo conditions.
Ballard and colleagues (8) removed trachea and mainstem bronchi from 3- to 6-wk-old pigs (lumen diameter 7.9 mm) and studied them in Ussing chambers. They reported Vt values of −9.7 and −4.0 mV in trachea and mainstem bronchi, respectively. The corresponding short-circuit currents (Isc) were 83 and 42 μA/cm2. Adding amiloride to inhibit apical ENaC reduced Isc by ∼41% and transepithelial electrical conductance (Gt) by ∼18%. Subsequent addition of bumetanide to the basolateral surface to inhibit the sodium, potassium, chloride cotransporter (NKCC) reduced the remaining current by ∼60%. Removing Cl− from the apical surface caused a large hyperpolarization, indicating the presence of a substantial GCl. In agreement with most mechanistic models of airway epithelia ion transport, these observations suggest that ENaC and a large GCl, probably CFTR, provide the two major electrically conductive pathways across the apical membrane of larger airways.
Porcine airway epithelia exhibit relatively high water permeability, although the specific aquaporins involved remain unidentified (47). Porcine trachea spontaneously absorbs liquid (47). Adding luminal amiloride or removing luminal Na+ eliminated absorption, indicating that active Na+ transport drives absorption. The pathway for counterion absorption is less clear. Removing luminal Cl− reduced liquid absorption 60%, but the anion channel inhibitors diphenylamine-2-carboxylate (DPC) and 5-nitro-2-(3-phenylpropylamino)benzoic acid (NPPB) had no significant effect. Although these inhibitors are neither very specific nor potent, the results suggested that the pathway for Cl− absorption might not be through the cell but rather through the paracellular pathway (47).
These studies provide an important starting point for understanding transepithelial electrolyte and liquid movement in wild-type pigs and for comparison with airways from CF pigs when they become available. It will also be interesting to learn how these properties change in wild-type and CF pigs as development progresses from the fetus, to the newborn, and to the adult.
Electrolyte Transport by Differentiated Cultures of Tracheal Epithelia
The most commonly used method for exploring mechanisms of electrolyte transport by non-CF and CF airway epithelia has been to harvest airway epithelial cells from trachea and large bronchi and place them in tissue culture. When the cells are grown on a permeable filter support at the air-liquid interface, they differentiate into an epithelium that retains many of the properties of native airway epithelia (123).
Liu et al. (146) found that primary cultures of human and porcine airway epithelia show quantitatively similar electrolyte transport. Both show an amiloride-inhibited current and a cAMP-induced Cl− current that was inhibited by basolateral bumetanide. The functional effect of apical DIDS, which can nonspecifically inhibit non-CFTR Cl− channels (as well as numerous other proteins), was also similar. One difference became apparent when Liu et al. (146) applied Cl− channel blockers. Glibenclamide, NPPB, CFTRinh-172, and GlyH-101 reduced Cl− secretory current by 70–90% in human airway epithelia. Whereas NPPB and GlyH-101 showed a quantitatively similar effect in porcine airway epithelia, glibenclamide (10% inhibition) and CFTRinh-172 (56% inhibition) were both less effective. This dissimilarity may reflect differences between the structures of human and porcine CFTR.
Electrolyte Transport by Small Airways
Whether CF airway disease begins in small vs. large airways continues to be debated. Better knowledge in this area could improve our understanding of pathogenesis and help with therapeutic strategies. However, studies of very small airways in humans have not been possible because of the inability of obtaining and successfully studying these tiny airways in tissue removed during human lung transplant. Moreover, there are no reports of studies in CF mice, likely because of the extremely small size of their bronchioles and differences between the structure of their lungs and those of humans. Even in larger animals, inaccessibility of the small airways, technical difficulties, and the risk of injuring the tissue have severely curtailed studies. However, some data have provided insight into electrolyte transport in porcine airways of ∼1 mm (8, 10, 259).
Ballard et al. (5, 8, 10) dissected and perfused small airways. With symmetrical solutions on both sides of the epithelia, they found a Vt of −4.3 mV in large bronchi (1.7 mm lumen diameter), −4.5 mV in small bronchi (0.7 mm), and −1.5 mV in bronchioles (0.25 mm). Adding apical amiloride inhibited Vt in all airway regions, suggesting that ENaC-dependent Na+ transport was the predominant active, electrogenic transport process from the trachea to the bronchioles. Probably by hyperpolarizing the apical membrane, amiloride also induced a secretory Cl− current (detected as a reduction in Vt on bumetanide addition) in all airway regions except the smallest bronchioles, where only a minor effect was apparent. However, Cl− substitution experiments suggested a prominent GCl throughout. No evidence for HCO3− secretion was detected under basal conditions (108). As in proximal airways, the β-adrenergic agonist isoproterenol, ATP, and UTP stimulated Cl− transport, and acetylcholine stimulated both Cl− and HCO3− transport (105, 108).
To further evaluate these very small airways, Wang and colleagues (259) developed a preparation that avoided traumatizing porcine bronchioles ∼1 mm in diameter. As in the studies by Ballard, their data revealed the presence of amiloride-sensitive Na+ channels. When they replaced Cl− with gluconate in the luminal solution, Vt dramatically hyperpolarized to values as large as −90 mV (Fig. 5). This result indicates that constitutive GCl is much larger than any other conductance in this epithelium. The voltage changes were reduced by CFTR channel inhibitors, showed an anion selectivity similar to CFTR, and were increased by elevating cellular cAMP levels.
These studies teach us several things about small porcine airways. First, ENaC-dependent Na+ transport is the major active transepithelial transport process. Thus these airways are probably absorptive, at least under basal conditions. Second, GCl dominates the small porcine airways, and CFTR is very likely responsible. Third, under basal conditions, GCl was constitutively active, although agents that elevate intracellular cAMP could increase it further. Fourth, the cellular pathway may play a greater role in determining Gt than the paracellular pathway.
What do these findings mean about net transepithelial salt (and hence liquid) transport in the small airways? Although we cannot be certain, the data suggest that absorption would be driven by ENaC, and Cl− would follow passively. The data also suggest that Cl− likely flows through the cellular pathway, perhaps through CFTR. Although Cl− could accompany Na+ by flowing through the paracellular pathway, that pathway would have to be highly Cl−-selective to generate such large Cl− diffusion voltages, and there is no precedent for such shunts. Interestingly, these findings suggest that electrolyte transport in the small airways shares substantial similarity to that in the sweat gland duct (184). One region that may differ is the distal bronchi, where Ballard et al. (5, 8) found that amiloride inhibited Vt 73% vs. a bumetanide-inhibited Isc of 59%. Whether salt absorption or secretion prevails there remains uncertain. Hopefully, the availability of a CF pig may allow investigators to address some of the persistent questions about small airways and how CF affects their function.
Electrolyte Transport in Alveolar Epithelia
CFTR also plays an important role in controlling liquid movement across alveolar epithelia. β-Adrenergic agonists and agents that elevate cellular levels of cAMP increase alveolar liquid clearance under both basal and pathological conditions (71, 152, 164). Na+ absorption through amiloride-sensitive ENaC and through nonamiloride-sensitive pathways provides the driving force for liquid absorption (75, 169). Apical CFTR channels provide a transcellular pathway through which Cl− can accompany Na+. In CF mice and when CFTR was inhibited, cAMP-stimulated liquid clearance was prevented (71, 72, 152, 164), and when CFTR was overexpressed clearance increased (164). Additional studies have shown CFTR transcripts, protein, and channel function in alveolar type II epithelial cells (25, 72). Thus, in alveoli, CFTR provides an important pathway for Cl− absorption. As previously noted (152), this function is similar to that in the sweat gland duct (184). Although evidence supporting a role for CFTR in alveolar liquid clearance has come from studies in several species, we are aware of no studies that have tested these mechanisms in porcine lung. However, a porcine model might provide new opportunities to better learn how CFTR contributes to liquid homeostasis in the distal lung.
SMGs are composed of a series of interconnecting tubules and ducts localized in the interstitium that lies beneath the surface epithelium of cartilaginous airways. The most distal regions of the network are comprised of serous acini and tubules (Fig. 6). Secretory products move vectorially from the distal serous tubules through mucous tubules and accumulate in collecting ducts. At the proximal end of each SMG, collecting and ciliated ducts connect the glandular tubules to the airway lumen (158). Each of the spatially distinct SMG regions contains specialized cell types that control the content and viscosity of its secretory products as well as the timing of their expulsion in response to airway irritation and infection (157, 165, 239, 263). CFTR expression is tightly regulated in many cell types in the lung, and it is presently unclear how CFTR function at distinct anatomic sites contributes to disease progression in CF (114). SMGs are one of these unique anatomic sites that express high levels of CFTR (69, 112, 240) and have been hypothesized to play an important role in the pathogenesis of CF lung disease (196, 257, 269, 287).
Several indirect lines of evidence suggest that SMGs may be important in CF lung disease: 1) SMGs exhibit a defined early pathophysiology (duct dilatation) in the CF lung, starting at neonatal stages (20, 51, 109, 174, 219); 2) CFTR expression is highest in SMGs of the cartilaginous airways (69, 112, 240); and 3) glandular serous cells secrete high levels of antibacterial proteins and express high levels of CFTR (16, 17, 69).
More direct evidence for the importance of CFTR in airway SMG function has also emerged from studies of human CF tissue. For example, CF SMGs exhibit a lack of secretion in response to agonists [vasoactive intestinal peptide (VIP) and forskolin] that elevate intracellular cAMP and activate CFTR (121, 122). Additionally, the mucous secreted by CF SMGs is more viscous than that secreted by non-CF counterparts (113). It is presently unclear whether the pH of SMG secretions is altered in CF. Although no difference in glandular pH has been detected when comparing CF and non-CF adults (113), recent studies suggest that hyperacidity of glandular secretions occurs in pediatric CF patients (216, 267).
Despite the identification of CFTR-dependent glandular defects in CF, animal models that are suitable for conclusively determining whether SMGs are, in fact, involved in CF pathogenesis have been lacking. Mice, which have been used for many studies of CF, do not possess SMGs distal to the trachea. The porcine lung, however, is very similar to that of human with respect to airway cell biology and the abundance of SMGs throughout the cartilaginous airways. SMGs in porcine and human airway tissue also appear to use the same pathways to control and regulate secretion (42).
Numerous studies have now evaluated the electrophysiological basis of SMG secretion in proximal airways of the pig and human. These studies have suggested that the mechanisms of glandular secretion are quite similar, if not identical, in the two species. Such studies have defined several important features of glandular secretion. For example, in porcine airways, forskolin- and VIP-induced liquid secretion by SMGs is facilitated by bumetanide-sensitive Cl− secretion and DIDS-sensitive HCO3− secretion (6, 12, 42). These results suggested either a Na+/HCO3− cotransporter or a Na+-dependent anion exchanger in the basolateral membrane may facilitate the transepithelial HCO3− transport required for glandular secretions. Additionally, studies by this same group have shown that acetylcholine-induced liquid secretion by porcine SMGs is also driven by both Cl− and HCO3− (11, 241). The relationships between CFTR function and these glandular secretion pathways have been more recently characterized in CF and non-CF tracheobronchial tissues. These studies demonstrated that CF SMGs fail to respond to forskolin and VIP but exhibit normal secretory rates in response to carbachol (an acetylcholine analog) (122, 267). Similar defects have also recently been documented in tracheal SMGs from the CF mouse (104).
Glandular secretions are potentially facilitated through three regions of the SMG (gland collecting ducts, serous cells, and mucous cells). Recent studies evaluating intact porcine glands suggest that the majority of cAMP-induced glandular fluid secretions originate from the serous cells and that the gland ducts likely play a role in modifying the composition of these secretions as opposed to facilitating bulk flow of fluid (275). In addition to fluid and mucins, SMGs secrete many bactericidal proteins into the airways. For example, SMGs have been shown to produce lactoperoxidase (LPO; Ref. 264), LL-37 (14, 15, 129), β-defensin-1 (15), lysozyme (50, 58, 131, 239, 260), and lactoferrin (50, 185), all of which are thought to bolster airway immunity against bacterial infection. The majority of these antibacterial products are produced by serous cells of the SMGs.
Mucociliary transport is an important component of the pulmonary host defense system. This process removes inhaled and aspirated pathogens and particulates. As in other forms of bronchiectasis, mucociliary transport can be defective in advanced CF. However, although mucociliary clearance has been measured on several occasions in CF, a comprehensive review by Robinson and Bye (189) concluded “when mucociliary clearance has been directly measured in CF patients, the literature is inconclusive as to the outcome.” This lack of a clear difference between CF and non-CF patients occurred even though studies were done in adults with established disease; in these patients, the effects of chronic inflammation and infection are superimposed on whatever pathogenic defect initiated the process. It has been difficult to detect mucociliary transport defects before the development of significant lung disease in young children and neonates for ethical and technical reasons, yet this is the time most relevant to CF pathogenesis.
Mucociliary transport can be measured using both ex vivo and in vivo techniques. Using freshly excised porcine tracheas, Ballard and coinvestigators have performed extensive studies with an emphasis on fluid transport, SMG function, and mucociliary transport. They showed that inhibition of anion and liquid secretion by bumetanide and dimethylamiloride decreased mucociliary transport, whereas acetylcholine enhanced this process (7, 13). Additionally, they observed that treating the luminal surface of fluid-depleted tracheas with aqueous solutions and surface-active substances restored mucociliary transport, with surface-active substances having a greater effect (7). These findings have led to the conclusion that anion and liquid secretion is required for mucociliary transport in the trachea and are enhanced with application of surface-active substances. Using a similar model, formaldehyde exposure has been shown to reversibly impair ciliary activity in porcine trachea. Following endotracheal intubation or tracheostomy, movement of Cardiogreen dye along ex vivo tracheal preparations is impaired (1a, 97). Figure 7A shows an ex vivo trachea mounted in an experimental setup based on the above model. Particle transport along the trachea is observed over time.
Since ciliary movement is required for effective mucociliary transport, ciliary beat frequency (CBF) measurements are important for studying this process. CBF is typically quantified in vitro (61, 97, 118), but Svartengren et al. (226) have also developed an in vivo method in pigs to measure CBF with laser light scattering microscopy and bronchoscopy, measuring a CBF of 5 Hz.
In vivo mucociliary transport measurements (in humans and animals) are usually performed with one of two methods. The first is whole lung clearance of radioactive material and involves inhalation of radioactive aerosols (usually containing 99mTc-labeled albumin, Fe2O3, or polystyrene) followed by measures of time-dependent clearance with a gamma camera (189). Data are plotted as retention of radioactivity vs. time for the entire lung and/or specific regions. This technique has been performed in humans but not previously in pigs. Therefore, to test clearance in pigs, we administered aerosols of 99mTc-sulfur colloid and measured clearance over time. Figure 7B shows an example; baseline clearance rates were in the range previously reported for normal sheep (194) and those described for humans (60).
The second method of measuring mucociliary clearance in vivo involves placement of a radiopaque particle or droplet in the distal trachea and subsequently tracking proximal particle movement over time (tracheal mucus velocity or transport). Eckerbom et al. (67) studied the effects of a heat and moisture exchanger on mucociliary transport rates in pigs by placing a droplet of 99mTc-marked macroaggregate albumin on the trachea and tracked movement over time with a gamma camera. They found a decrease, although not statistically significant, in transport rates in the control group (4.9 mm/min) compared with the heat and moisture exchanger group (8.3 mm/min) suggesting that adequate airway hydration is important for normal mucociliary transport (67). We modified previously described methods to track the movement of radiopaque Teflon particles with noninvasive computed tomography (CT) imaging and computerized airway reconstruction algorithms. This method allows for three-dimensional (3-D) imaging of the lung and particle tracking (velocity measurements) in individual airways. Figure 7, C and D, demonstrates typical images that can be obtained from this method.
On the basis of combined culture and serologic results, over 97% of patients with CF are infected with P. aeruginosa (35). P. aeruginosa acquisition is associated with pulmonary function decline and early mortality. Whereas Haemophilus influenzae and Staphylococcus aureus sometimes precede P. aeruginosa infection, their prevalence declines over time whereas that of P. aeruginosa increases. Wild-type pigs are not ordinarily susceptible to spontaneous infection with the classic CF pathogens. Endogenous infections have been extensively studied in pigs because of their commercial importance and because production and housing methods can expose them to respiratory pathogens. The prevention and treatment of bacterial airway infections represent a major effort by commercial pig production operations. Pigs are routinely vaccinated and treated with antibiotics, however, airway colonization and infection can still occur (74). Both Gram-positive and Gram-negative bacteria can cause respiratory illnesses in pigs; some of the more common naturally occurring organisms include Mycoplasma hyopneumoniae (26, 148, 171, 176), H. parasuis (148, 253), Streptococcus suis (102, 148), Actinobacillus pleuropneumoniae (148), and Pasteurella multocida (92, 93). However, under normal circumstances, people without CF are also not susceptible to such microorganisms. Some recent evidence suggests that, like humans, pigs can carry S. aureus and that S. aureus, including methicillin-resistant S. aureus (MRSA), can be passed between the two species. S. aureus is the most common bacteria isolated in the air of confined animal feeding operations in Ohio (85). MRSA colonization of animals has also been reported in pig farms in France (3), The Netherlands (103), and Canada (126). In a recent report, MRSA was isolated in 81% of the pig farms, and in 37% of the farms the farmer and his relatives also carried the same strain of MRSA that colonized the pigs (55).
Porcine models have been used to evaluate experimental pneumonias following bacterial inoculation with human pathogens. For example, after 4 days of mechanical ventilation, most pigs developed a spontaneous pneumonia with organisms that include P. multocida, S. suis, S. aureus, and P. aeruginosa (150). The histological, bacteriological, and pathogenic aspects of the pneumonia resembled early onset ventilator-associated pneumonia in humans. Figure 8 shows results after bronchoscopic instillation of a relatively large volume (2 ml) and number (108 colony-forming units) of P. aeruginosa strain O1 (PAO1) into the left lower lobe of pigs. Three hours after instillation, bronchoalveolar lavage (BAL) showed increased numbers of cells with >50% neutrophils (Fig. 8, A and B). A chest CT showed air space disease, and histological examination showed inflammation consistent with a pneumonia (Fig. 8, C and D). These data suggest that delivering large amounts of P. aeruginosa to the distal airway results in acute infection rather than the bacterial colonization of airways expected in the CF lung.
Viruses that infect pigs include porcine influenza virus, porcine circovirus, porcine reproductive and respiratory syndrome virus, porcine parvovirus, vesicular stomatitis, and pseudorabies virus (29). The clinical manifestations are variable depending on the age of the pig, vaccination status, and herd immunity. Some of these viruses play an important role in the pathogenesis of bacterial infections (200, 230, 231). Pigs serve as a major reservoir of H1N1 and H3N2 influenza viruses and are frequently involved in interspecies transmission of influenza viruses (178). The major manifestations of these viruses include the following. Porcine circovirus: postweaning multisystemic wasting syndrome, porcine dermatitis, nephropathy syndrome, abortions, and stillbirths (205). Porcine reproductive and respiratory syndrome virus: pneumonia with the virus persisting in tissues and macrophages (138). Porcine parvovirus: postweaning multisystemic wasting syndrome (147). Porcine cytomegalovirus: rhinitis (202). Swine influenza virus: flu-like symptoms (272). Pseudorabies: pneumonia and predisposition to bacterial infections (168).
Sows produce a strong maternal immunity that persists in their offspring until 8–12 wk of age. As a result, the effects of the infection in weanlings are usually minimal. Pigs usually become subclinically colonized when still protected by maternal antibody and then stimulate their own immune response (22, 116, 171–173). Nasal colonization of wild-type pigs has been achieved by intranasal inoculation. Interestingly, only 20% of the pigs develop pneumonia (93). Nasal colonization with M. multocida and Bordetella bronchiseptica results in purulent rhinitis that with time results in atrophy of the nasal mucosa with remodeling of the turbinates and sinuses. Altered morphology and anatomy may result from bacterial colonization (78).
INNATE IMMUNITY AND INFLAMMATION
Similar to other mammals including humans, pigs have a full armamentarium of innate immune effectors in the respiratory tract to protect mucosal surfaces, elicit inflammatory responses, and orchestrate adaptive immune responses. Although some porcine host defense proteins are unique, most share compositional and functional similarities with their human counterparts (Table 3), particularly cytokines, chemokines, and chemokine receptors. When exposed to respiratory viral or bacterial pathogens, newborn and adult pigs produce a proinflammatory cytokine response and a chemokine response and modulate host defense peptide and protein expression in respiratory mucosa, respiratory secretions, and pulmonary macrophages. These responses are similar to those of naturally infected humans, isolated human airway epithelia, or pulmonary macrophages. Thus pigs make excellent models to study the pathogenesis of respiratory infection and inflammation. In addition, these features support the use of pigs to investigate therapeutic interventions for respiratory diseases.
Pulmonary Cells of Innate Immunity
The epithelial cells of the surface and SMGs play an important role as first responders and amplifiers of innate immunity. Although specific information for porcine respiratory epithelia is currently lacking, pigs are endowed with a full array of microbial pattern recognition receptors, including Toll-like receptors (210, 211, 237), peptidoglycan recognition proteins (199), and NOD-like receptors (236, 238).
In addition to epithelia, resident professional phagocytes including alveolar macrophages and polymorphonuclear leukocytes (PMNs) play important roles in porcine lung health. Similar to humans, the PMNs and alveolar macrophages of the newborn are relatively immunocompromised compared with the cells of older pigs (62). The alveolar macrophage is the most abundant cell type recovered by BAL in pigs, representing ≥94% of white blood cells (177). In contrast to humans, pigs possess a specialized pulmonary intravascular macrophage, resident in the pulmonary vasculature, that plays an important role in clearance of blood-borne pathogens (41). Porcine PMNs are an abundant source of a number of antimicrobial and immunomodulatory peptides and proteins (Ref. 281, see below).
Host Defense Polypeptides and Proteins
Pigs and humans share several classes of host defense peptides and proteins including lysozyme, lactoferrin, collectins (surfactant proteins A and D), palate lung nasal epithelial clone (PLUNC), and β-defensins (Table 3). The elastase inhibitors secretory leukocyte protease inhibitor (SLPI) and elafin also have orthologs in pigs (73, 77, 225). Whereas the human genome encodes a single cathelicidin gene (LL-37), pigs have several including protegrins (284), prophenins, porcine myeloid antimicrobial peptide (PMAP)-23, -36, and -37, and proline-arginine-rich 39-amino acid peptide (PR-39) expressed in their PMNs (32, 96, 281, 284). PMAP-23 is highly cationic with five arginines and two lysines (280, 281). Minimum inhibitory concentrations (MICs) range from 2 to 16 μM against Gram-negative bacteria (281), Gram-positive bacteria (281), and yeasts (137). PR-39 contains high concentrations of proline (45–49%) and arginine (24–29%) (23, 207). MICs range from 1 to 4 μM (23, 206). Different forms occur, and PR-39 isolated from porcine small intestine is slightly different in composition from that isolated from porcine neutrophils (207). A variant form of PR-39 called βPR-39 also exists (283). Prophenin has been identified in porcine leukocytes (96, 218, 283); it contains an abundance of proline and phenylalanine residues. Two forms exist, prophenin-1 (96) and prophenin-2 (283), that differ in their compositions. Prophenin-1 is substantially more active in vitro against Escherichia coli than against Listeria monocytogenes (96).
Similar to humans, pigs express an array of defensin peptides. No porcine α-defensins have been identified (281), but they express several β-defensin peptides. Porcine β-defensin-1 (pBD-1) is expressed in the cornified tip of the filiform papillae of the dorsal tongue and in the superficial squamous cell layers of the buccal mucosa (208, 282). The cDNA sequence of pBD-1 encodes a 64-amino acid prepropeptide, which contains the 42-residue natural form (208). The peptide is antimicrobial at 40 μg/ml, and activity is synergistic with other neutrophil-derived antimicrobial peptides such as PG-3 and PR-39 (208). pBD-1 mRNA is detected throughout the respiratory and digestive tracts and a variety of other tissues (Table 3). A recent bioinformatic analysis of the porcine EST database yielded 11 novel porcine β-defensin mRNAs (198), and this number is likely to increase as the porcine genome project is completed. Seven of these β-defensins, pBD-2, -3, -4, -114, -123, -125, and -129, demonstrated mRNA expression in lung tissue, but cell-specific localization and function remain to be determined.
Oxidative Host Defense System
A constitutive oxidative airway epithelial host defense system was recently described (76, 80, 195). This reactive oxygen species-producing system has three components: the airway epithelial proteins dual oxidase (Duox) 1 and 2, LPO secreted by SMGs, and a halide or pseudohalide ion (thiocyanate, SCN−) that is secreted by epithelia. Duox-mediated H2O2 production is sufficient to oxidize SCN− to hypothiocyanite (OSCN−) in the presence of LPO: SCN− + H2O2 (+ LPO) → OSCN−. OSCN− exerts its microbicidal activity extracellularly, in the airway surface liquid (ASL) environment. This activity appears to be impaired in CF due to a reduced halide permeability caused by the loss of CFTR function (43, 163, 179). Although this system has not yet been extensively studied in porcine airways, the three-component system appears to be functional in salivary glands and breast milk (186). Porcine gastrointestinal and thyroid epithelia express Duox2 (63, 68), and LPO is present in breast milk. Future studies will help define the importance of this potent antimicrobial system in the porcine airways.
Cytokines and Chemokines
Pigs produce a broad spectrum of cytokines and chemokines that are functional orthologs of human T helper type 1 (Th1) cytokines, Th2 cytokines, proinflammatory cytokines, chemokines, and regulators of T and natural killer cell activation and proliferation. Although the list of immunologic tools available for swine is less than for humans and mice, there are many useful immunologic reagents and commercial assays readily available. Th1 cytokines (IL2, IL12, and IFNγ), Th2 cytokines (IL3, IL4, IL5, and IL10), proinflammatory cytokines [IL1α, IL1β, IL6, granulocyte/macrophage colony-stimulating factor (GM-CSF), TNF-α, and IL12], chemokines [CXCL8/IL8 and CCL2/monocyte chemoattractant protein-1 (MCP-1)], and IL7, IL15, and IL18 have been reported in pigs, and many of these immunologic mediators, with their respective antibodies, are commercially available. Likewise, the cDNA of swine IL17 (PoIL17), has been isolated and recombinant PoIL17 has been expressed in E. coli (124). PoIL17 was biologically active and exerted similar effects to those of a human IL17. Multiplex flow cytometry using xMAP technology (117), ELISA (4), ELISPOT (4), and SearchLight Porcine Cytokine Arrays can be used to assess multiple cytokine expression profiles.
Porcine IL2 has 82.3% (70.7%), porcine IFNγ has 75.1% (60.2%), porcine IL12 (p40) has 86.3% (82.9%), and porcine IL7 has 85% (73%) nucleotide sequence (amino acid sequence) homologies with their human counterparts, respectively (170, 249). Porcine IL6 shares a significant homology with human IL6. Porcine chemokines have similar homologies. For example, a CXC chemokine, designated as chemokine cd00273 subgroup, has high homology with the small inducible cytokine B10 precursor (CXCL10) of humans (84%) and is defined as swine small inducible cytokine B10 precursor (145). Genes for six CC chemokine receptors (CCR1, CCR2, CCR3, CCR5, CCR9, and CCRL2) and two other chemokine receptors (CXCR6 and XCR1) are highly conserved among pigs and humans (209). Using pigs as models for CF will result in the rapid identification of new immunologic mediators and the development of novel methods for their detection as we seek to understand the etiology and pathophysiology of CF.
Porcine IgG antibodies against P. aeruginosa can be measured by ELISA using antigen absorbed onto polysorb plates and detected with peroxidase-conjugated goat anti-swine IgG heavy plus light chain antiserum. Serum IgG1 and IgG2 against P. aeruginosa can be measured by ELISA using antigen absorbed onto polysorb plates and detected with mouse anti-porcine IgG1 or IgG2 used as the primary antibody and peroxidase-conjugated goat anti-mouse IgG heavy plus light chain antiserum used as the secondary antibody.
Granulopoiesis in pigs is much less well-studied than in humans. Although the half-life of pig neutrophils is not known, it is likely similar to that of humans. Pigs express an ortholog of human GM-CSF (111). In addition, a cocktail of porcine IL3, stem cell factor, and GM-CSF stimulates the mobilization of porcine bone marrow (BM) progenitors in vivo in the pig and in baboons that underwent a conditioning regimen and porcine BM transplantation (132).
Genomics and large-scale expression profiling technologies have helped increase our understanding of inflammatory responses in the pig (64, 246, 285), however, little work has been done in the respiratory tract. Pigs are vulnerable to a number of respiratory Gram-positive and Gram-negative bacterial and viral infections (29, 213, 231). These diseases are well-characterized and include documented cellular, immunologic, and chemokine/cytokine responses to both spontaneous and experimental infections (229, 232, 233, 255, 256). The chemokine and cytokine profiles of bacterial and viral lung diseases of pigs are similar to humans (229). Pigs are susceptible to viral respiratory diseases, bacterial respiratory diseases, and polymicrobial respiratory diseases called porcine respiratory disease complex (PRDC). The latter complex can involve up to five viruses and eight bacteria (27, 28, 31, 214, 228). Pigs infected with porcine reproductive and respiratory syndrome virus (PRRSV) and/or M. hyopneumoniae had significantly increased levels of mRNA for many proinflammatory cytokines in pulmonary alveolar macrophages collected by BAL compared with those in uninfected control pigs (232). Increased levels of IL1β, IL8, IL10, and TNF-α proteins in BAL fluid, as measured by ELISA, confirmed the increased cytokine induction induced by the pathogens. IL10 and TGF play similar anti-inflammatory roles in the porcine lung as noted in humans (229).
MCP-1 and macrophage inflammatory protein-1 (MIP-1) induction was assessed in 1-day-old piglets inoculated intranasally with porcine circovirus 2 by RT-PCR. Maximum MCP-1 expression and MIP-1 expression were observed at 17 and 21 days postinoculation, respectively (128).
Pigs express an ortholog of IL8 that is associated with porcine bacterial and viral lung diseases (229, 232, 255). Porcine IL8 shares ∼75% amino acid similarity with human IL8 and exhibits transcriptionally regulated expression in response to LPS stimulation (83, 144). Macrophages contained low levels of IL8 mRNA, which increased ∼30-fold higher after exposure to 10 pg/ml bacterial LPS.
How the presence of porcine cytokines/chemokines, host defense polypeptides and proteins, or changes in their microenvironments would affect their activity and influence a CF pig pulmonary phenotype is unknown. It is interesting to note that ASL from porcine primary airway epithelial cell cultures shows a similar profile of loss of antimicrobial activity as the ionic strength of the test solution is increased (Fig. 9).
PORCINE LUNG IMAGING
Much of the current information about lung structure and function is derived from global measures of lung function, from static pathology studies performed in two dimensions, or from imaging studies with poor spatial resolution. There is little direct information about regional structure and associated physiology in the living lung. However, structural and functional imaging of the airways and lung parenchyma can now be used to evaluate the presence of disease to assess its activity and progression and to measure heterogeneity of lung structure and function. Newer lung imaging processes include macro-optical techniques, micro-optical techniques, and multi-row detector X-ray CT (MDCT). The optical techniques allow for rapid, relatively noninvasive image acquisition over large areas, such as the bronchoscopically visible airway tree, for quantification of airway health or very focused high-resolution imaging of targeted regions in a manner that is not destructive to the target tissue. MDCT image sets are acquired as slices that, when stacked, provide a 3-D contiguous data set that can be visually examined, and relevant structures such as airways can be identified and measured. The data sets consist of lung structures that have been detected through their interaction with X-rays, and recorded on a detector array, as a series of grayscale values. The smallest independent unit that makes up the image matrix in each slice in two dimensions is called a pixel; this has an x and a y dimension. The 3-D nature of these data sets can be captured because the image slice thickness (the z dimension) is the same as that in the plane pixel x and y dimensions. Adding the z dimension to a pixel creates an imaging unit known as a voxel. If the x, y, and z dimensions are the same, as can be the case with modern MDCT scanners, the voxel is a cube, leading to a smooth image when viewed from all directions. The capability of acquiring images of the lung with cubic voxels has allowed for increased understanding of lung structure.
The Airways: Virtual Bronchoscopy
A virtual bronchoscopy is a computer-generated visualization of a traditional bronchoscopy procedure with data acquired in 3-D using MDCT scanning; it provides a simulation of a bronchoscope traversing the pulmonary airways (73a, 155a, 155b, 205a). The structure of the pulmonary airways is extracted from MDCT volumetric data through automated airway tree segmentation. The segmentation involves a combination of adaptive region growing and a hybrid method utilizing both region growing and mathematical morphology. The segmented airway is then volume rendered, which produces a 3-D structurally accurate image of the inner pulmonary airway wall (243–245). The data, which can now be collected with isotropic voxels, can be visualized in simulated three dimensions. Since the voxel size is also known (each voxel has a uniform size with linear and volume dimensions), the image data can be interrogated for the grayscale changes that define anatomic borders as well as for size. Using specifically designed software, the airway tree can be extracted, and airway dimensions for each airway generation, including wall thickness, reported. Such measures can be compared on subsequent MDCT scans in the same animal or between different animals of the same age. To compare MDCT data, the scans must be acquired at controlled predetermined lung volumes, using identical scanning parameters. In this manner, the heterogeneity of the airways in the normative animal can be characterized and compared with any additional heterogeneity or homogenous changes that might be associated with a disease process. An example of the airways as seen in the normal porcine lung, using commercially available software, is illustrated in Fig. 10.
The Airways: Macro-Optical Bronchoscopy
The airway structure and mucosa can be assessed objectively through optical bronchoscopy using either fiber-optic or charged couple device (CCD) systems. Using these macro-optical bronchoscopes, three quantitative functional parameters can be assessed, mucosal color analysis, mucosal texture analysis, and airway fluorescence (192). To complement airway wall color change, airway wall texture can also be measured. Methods and results of the color assessment have been published (221, 224). Using hue and saturation as color measures, pixel-by-pixel real-time analysis of the bronchial mucosal color can easily be achieved and compared against a normative database to determine regions that are different from normal as well as regions that may change color over time. For instance, in acute inflammatory conditions, the airway wall becomes much redder, whereas chronic inflammation might lead to a paler mucosal color. High-quality CCD bronchoscopes are now available for large animal studies. These are purpose-built to allow for the extra length needed for the longer snout in intubated pigs or sheep. Such macro-optical assessments of the entire bronchoscopically visible airway mucosa may be a highly sensitive and early indicator of mucosal change from the normal state without having to undertake multiple and frequent small biopsy samples.
Mucosal blood flow has not been studied systematically in CF but is expected to be a factor in the phenotypic manifestations of the airway disease. An assessment of bronchial mucosal blood perfusion can be obtained by measuring the time intensity fluorescence in the airway mucosa bronchoscopically immediately coincident with an intravenous injection of fluorescein (223). In animal and human studies, the airway wall fluorescence signal appears at ∼20 s after an intravenous fluorescein bolus and peaks rapidly. Preliminary work in the porcine lung demonstrates that this technique works well and can define bronchial mucosal blood flow. Mucosal perfusion assessed in this manner might be a further indicator of airway mucosal health or disease. Additionally, MDCT virtual bronchoscopy and optical bronchoscopy image sets can be combined into one data set to allow ease of visualization of complex changes, including over time (222).
The Airway and Parenchyma: Micro-Optical Bronchoscopy
The use of novel micro-optical techniques has recently seen interest in the pulmonary field for their potential to translate into clinical in vivo biopsy systems. Methods include video-assisted microscopy (38, 56, 79), optical coherence tomography (OCT) (182, 278), and catheter-based confocal microscopy (CBCM) (127, 193). CBCM systems can be used for characterization of normal and diseased lung tissue in mouse models (167), in larger animals such as the pig, and in humans (235). These systems have the capability to noninvasively resolve structures down to the micron level using reflectance mode and down to the submicron level using fluorescence mode.
The CBCM are designed to fit through the auxiliary channel of a standard or ultrathin bronchoscope and therefore can be used to investigate lung tissue from the airway epithelium down to the alveoli via the subtending airways. Using appropriate fluorescent probes, identification of specific and nonspecific structures such as collagen, elastin, nerves, cell receptors, nuclei, and macrophages can be achieved. Figure 11 represents the porcine alveolar region imaged using the CBCM system with no external fluorophores applied. If safe external fluorophores, such as fluorescein, are given systemically, the alveolar wall detail is seen in more detail since the fluorescein is distributed throughout the lung water. A promising application for CF research in porcine models is the direct visualization of the airway epithelium using the CBCM system with the possible added potential to measure CBF in vivo. Using OCT imaging systems with improving resolution and image acquisition times, there is also the potential for measuring the ASL in vivo, an important feature that could be used to differentiate normal vs. diseased airway epithelium. Confocal microscopy has also been used for assessing sensory networks in porcine bronchial mucosa and formation of ganglia in the airways of fetal porcine lungs (135, 261). Recently, investigation of mice alveoli in the intact lung using confocal microscopy techniques with fluorescein staining has led to new understandings on alveolar mechanics otherwise not possible using traditional techniques (167).
MDCT IMAGING METHODS
There is increasing literature on the value of MDCT scan quantitative imaging in the lung (99, 100). Informative MDCT imaging requires that the pig be on a ventilator to control breathing and stop breathing at prescribed lung volumes for imaging the lung. A common protocol for porcine lung imaging includes 100 mAs, 120 kV, 1-mm collimation, an effective slice thickness of 1.3 mm, overlap of 0.65 mm, pitch of 1.2 mm, and a slice parameter mode of 32 × 0.6 mm, with 512 × 512 slice matrices. The lungs are typically imaged at 40% and 95% vital capacity. Software packages have been developed to automatically segment the lungs and the airways from the MDCT image data set and to categorize voxels within the lung fields according to their attenuation values (Hounsfield units). An example of airway segmentation in the porcine lung is shown in Fig. 10. Additionally, classification of the lung parenchyma into normal or diseased states can be based on the grayscale measures alone or on more involved pattern recognition algorithms. One example of pattern recognition is the adaptive multiple feature method, which recognizes different tissue patterns including normal lung parenchyma, ground glass, emphysema-like lung, and honeycomb patterns (250–252). With a combination of these visualization and measurement tools, the airways and the lung parenchyma can be assessed from the MDCT scan information. This methodology has now been adapted to the texture assessment of the lung in 3-D (276, 277). Such assessment can help answer the question as to the heterogeneity of lung changes in CF and whether the most severely affected airways also subtend the most severely affected parenchymal disease, an important yet unanswered question.
Porcine sinuses can also be evaluated using MDCT as has been performed in sheep (33). This procedure can provide measures of mucosal wall thickness and air space volume. Figure 12 shows an example of sinus MDCT images from the pig plus a 3-D rendering of the sinuses.
PULMONARY FUNCTION TESTS
Over time, almost all patients with CF experience progressively worsening airways obstruction secondary to chronic infection and accompanying recurrent pulmonary exacerbations. Spirometry provides an objective and convenient method of identifying declines in pulmonary lung function and responses to therapeutic interventions. Unfortunately, many of the techniques needed to perform standard pulmonary function tests (PFTs) require that the subject be cooperative and awake, making these tests difficult to perform in animals. Investigators, particularly in the areas of toxicology and asthma, have either adapted “traditional” PFT techniques to animals or have developed new techniques and protocols more easily performed in laboratory animals. In swine, several methods exist to determine lung function under pathological conditions or in response to treatment.
Esophageal Balloon-Pneumotachograph Technique
The esophageal balloon-pneumotachograph technique can measure respiratory rate, tidal volume, dynamic lung compliance, and total pulmonary resistance. This method has been successfully used to measure bronchial hyperresponsiveness and compliance in a porcine models of Sephadex-induced inflammation and following inoculation with M. hyopneumoniae. It has also been employed to investigate the effect of dietary fat consumption on pulmonary function (110, 154, 273). A limitation of this method is that sedation and/or anesthesia is needed for the required instrumentation.
Whole Body Barometric Plethysmography
Whole body barometric plethysmography (WBBP) allows measurements to be made on conscious, unrestrained animals. Sedation is not required, and the animals are not intubated. Animals are placed in airtight box, and pressure changes in the box, secondary to the respiratory cycle, are recorded. A parameter termed enhanced pause, Penh, is calculated and has been correlated with airway resistance. Despite extensive studies in mice with this method, controversy exists regarding Penh and whether it represents a valid measure of bronchoconstriction in mice (19). However, Halloy et al. (90, 91) have shown in swine that airway reactivity measurements made with WBBP correlate well with esophageal balloon-pneumotachograph measurements following acetylcholine and endotoxin challenge and have also tested WBBP after pulmonary challenge with P. multocida.
Impulse Oscillometry System
Klein et al. (129a, 130) have adapted the noninvasive impulse oscillometry system to swine. This technique is performed in nonintubated pigs that are either lightly sedated or trained. With the impulse oscillometry technique, airway impedance is measured in response to generated sound signals broadcast into the airways through a tight-fitting face mask. Flow and pressure signal analysis of test signals with a spectrum of frequencies allows for resistance measurement to be made in different portions of the airway tree. This technique has been successfully used for pulmonary function measurements in pigs after natural and experimental infection with Chlamydia suis (185a).
Whereas the above methods represent more traditional approaches for measuring airway obstruction, recent studies suggest that high-resolution chest CT imaging (HRCT) and detection of air trapping is a more sensitive indicator of early airways obstruction in CF patients (30, 54, 84, 190). Air trapping represents air that is unable to be expired from the lung. On expiratory HRCT images, air trapping appears as low attenuation areas. We have begun to explore this methodology in pigs. Figure 13A shows CT images at total lung capacity and functional residual capacity following intrapulmonary delivery of methacholine. The accompanying histogram (Fig. 13B) details the changes in pulmonary parenchymal density following methacholine challenge. Finally, with an increasing number of studies using hyperpolarized helium and magnetic resonance imaging (MRI) scanning to detect airflow limitations and pulmonary parenchymal abnormalities, MRI may also prove to be useful in a swine model of CF to assess pulmonary function (70, 156, 254).
OTHER ORGAN SYSTEMS INVOLVED IN CF
CF is a multisystem disease. As with the lung, lack of a suitable animal model has limited understanding of disease pathogenesis in other organs. This is an important problem because disease in other organs can cause patients significant morbidity and is occasionally lethal. Thus an animal model that develops disease like that in humans might offer opportunities for understanding, preventing, and treating CF-related disease in many organs. Although a description of the physiology, anatomy, histology, and biochemistry of other porcine organs is beyond the scope of this review, here we briefly mention some of the other manifestations of CF where the availability of a new animal model might provide investigators with an opportunity to impact the disease. For further information about CF manifestations in other organs, please refer to Ref. 262.
Within the first day or two of life, meconium ileus affects 10–20% of patients who have mutations associated with severe disease. The cause remains uncertain, but it has been speculated that pancreatic failure or abnormal intestinal electrolyte transport cause the intestinal obstruction of meconium ileus and the secondary manifestations such as microcolon. Although noninvasive procedures can relieve the obstruction in some patients, surgery is still frequently required. Once meconium ileus has been corrected, patients have a clinical course similar to that of other patients with severe disease. Some patients also develop a small bowel obstruction termed (distal intestinal obstruction syndrome, DIOS). DIOS can occur repeatedly in patients with CF. CF mice do not develop meconium ileus, however, they do display different intestinal abnormalities. The porcine intestine has frequently been used as a model in studies of nutrition and electrolyte transport and may provide new insight into the pathogenesis of meconium ileus and DIOS.
Nearly all patients with severe mutations (for example, the ΔF508 mutation) develop exocrine pancreatic insufficiency. The pathological and clinical findings led Dorothea Anderson to originally name the disease “cystic fibrosis of the pancreas.” Supplementing patients with porcine pancreatic enzymes and fat-soluble vitamins has reduced the morbidity and mortality associated with CF pancreatic disease, however, it remains a significant clinical problem. A widely held hypothesis is that plugging of the pancreatic ducts causes the organ destruction. However, there are many persistent questions about pancreatic development, pathogenesis, inflammation, and destruction in CF. The reason that mice do not develop CF pancreatic disease remains unknown.
With extensive exocrine pancreas destruction, the islets of Langerhans are also sometimes destroyed, and patients can develop diabetes mellitus. Of note, their similarity to human islets has led to attempts to develop xenotransplantation of porcine islets to humans with diabetes mellitus.
Patients with CF can develop hepatic disease that often remains clinically silent until the manifestations of portal hypertension emerge. The disease appears to primarily affect the biliary ducts, where CFTR is expressed. It has been hypothesized that loss of Cl− and HCO3− transport alters the hydration and composition of bile leading to plugging of portal biliary ductules. However, much remains to be learned about the pathogenesis, and it is possible that studies of a CF animal model could suggest novel therapeutic approaches. Of note, the porcine liver shows many similarities to the human liver, hence the attempts to develop xenotransplantation of porcine liver to humans. CF mice do not develop liver disease like that in humans with CF.
Congenital bilateral absence of the vas deferens causes infertility in ∼97% of males with CF. The anatomic changes appear to occur after the structures have initially formed, although the pathogenesis remains uncertain. The testes are usually normal in size.
Elevated concentrations of Cl− and Na+ in eccrine sweat are present at birth and throughout life in patients with CF, and measurement of sweat Cl− concentration is a routine diagnostic procedure. Pigs do not rely on sweating for evaporative cooling, although they have a few eccrine sweat glands on the nose. Thus a CF pig may not be a particularly good model for eccrine sweat gland dysfunction, although studying the few porcine eccrine sweat glands might provide a comparison with human sweat glands.
Patients with CF also have other disease manifestations, including osteoporosis, reduced fertility in females with CF, episodic arthropathy, and an increased risk of cancer. A CF pig might prove to be a useful model for these conditions.
This review indicates that porcine lungs share many anatomic, histological, biochemical, and physiological features with human lungs. The literature suggests that the function of airway epithelia and SMGs in large part match those of human, and inflammation and infection in many ways parallel responses in humans. Although there are differences between the human and porcine lung, porcine lungs are much more like those of humans than are mouse lungs. Thus we are hopeful that a CF pig will develop lung disease that mimics that in humans with CF. Without question, there will be differences between the two. But if the CF porcine lung progresses at least part way toward the disease in humans, it will open new opportunities to probe pathogenesis and develop novel therapeutic strategies.
A CF pig should also provide unique opportunities, including the chance to focus on early postnatal and young pigs. There is a critical lack of information about the human CF lung during development and the early postnatal period. Yet, this is precisely when loss of CFTR may initiate disease. In addition, it has been difficult to define the role of CFTR in many airway processes, because once the disease is underway, chronic inflammation and infection cloud interpretations and prevent a distinction between pathogenic mechanisms and downstream consequences of airway disease. A CF pig may also provide the opportunity to study the small airways. Although this area is hypothesized to be important in pathogenesis, no data exist for the human lung. A CF pig could also become an invaluable resource for testing therapies. At the present time, we cannot ask whether a potential treatment prevents lung disease in a mouse, because they do not develop disease that mimics that in patients. Finally, there is a renewed emphasis on preventive strategies now that genetic screening is increasingly identifying newborns with the disease; findings from a CF pig model could inform the field about new strategies for their care.
What if a CF pig turns out to have a clinical phenotype like that of the CF mouse with little pulmonary disease? We think this unlikely based on earlier studies of the normal porcine lung. However, even if pigs and mice have the same pulmonary phenotype, the results should be highly informative and help direct the field. To illustrate this point, consider two of many potential examples. 1) SMGs are hypothesized to be central to CF pathogenesis. One hypothesis states that CF mice lack typical CF lung disease because they have only a few SMGs whereas glands are abundant in the cartilaginous airways of humans. Thus, if the CF pig has no pulmonary phenotype, it might shift emphasis away from SMGs as an underlying factor. 2) It has been proposed that activity of an alternative Cl− channel replaces the function of CFTR in mouse airway epithelia, thereby protecting CF mouse airways from disease. If we find that pigs have no pulmonary phenotype, and have alternative Cl− channel function like that in humans, it would indicate that the alternative Cl− channels thought to be important in preventing disease in the mouse are probably not an important protective factor. Conversely, if CF pigs do not develop lung disease and have alternative Cl− channel function like that in mice, it will suggest they compensate for the loss of CFTR to prevent lung disease. This conclusion would make them an important therapeutic target. It would also suggest the CF porcine airway as an excellent model in which to better understand their biochemical, molecular, and functional properties.
Thus comparing results from CF pigs with the extensive data already obtained in humans with CF and CF mice should teach us much about how loss of CFTR causes lung disease, no matter what the phenotype of a CF pig.
We thank the National Heart, Lung, and Blood Institute (Grant HL-51670), the National Institute of Diabetes and Digestive and Kidney Diseases (Grant DK-54759), the National Center for Research Resources (Grant RR-13438), Food for the 21st Century, the Cystic Fibrosis Foundation, and the Howard Hughes Medical Institute (HHMI). C. S. Rogers was supported by National Institutes of Health Training Grant HL-07638. D. A. Stoltz is a Parker B. Francis Fellow. L. S. Ostedgaard was supported by the Cystic Fibrosis Foundation. M. J. Welsh is an Investigator of the HHMI.
- Copyright © 2008 the American Physiological Society