Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The cystic fibrosis (CF) transmembrane conductance regulator (CFTR) protein is a plasma membrane anion channel and a regulator of other ion channels that is mainly expressed by exocrine epithelia (1). In CF, the most frequent recessive genetic disease among Caucasians, mutations in the cftr gene lead to partial or total loss of CFTR function, which, in turn, is thought to impair transepithelial chloride and bicarbonate transports directly as well as sodium and water transports indirectly. Abnormal ion and water transports can easily account for symptoms of CF such as the high NaCl content in sweat and the thick, dehydrated digestive secretions clogging the ileum and bile ducts. However, the lung manifestation of CF, responsible for more than 90% of the morbidity and mortality among patients, does not lend itself to such simple explanations (2). CF airway disease associates the presence of mucus plugs in the airway lumen with long-lasting bacterial infections (mainly with Pseudomonas aeruginosa) and persisting local and systemic inflammation. One puzzling feature of CF airway disease is that it only starts after birth, despite strong CFTR expression in the prenatal lung. By contrast, meconium ileus and atresia of vas deferens readily occur during gestation (3). Histopathologic surveys of prenatal CF lungs did not reveal any difference compared with normal lungs, except for slight morphologic changes of submucosal glands (4, 5).

Recently, clinical studies on infants with CF revealed very early signs of inflammation, including a high proinflammatory chemokine interleukin (IL)-8 content and a high neutrophil count in the lung intraluminal fluid, even in the absence of detectable infection (6). This led to the suggestion that inflammatory processes may be primarily altered in CF lungs. Whether inflammation arises independently from infection is still unknown (9), as is the exact cascade of events underlying the early phase of CF airway disease. This current lack of knowledge is partly explained by the difficulty of extensive studies in young infants with CF and by the absence of relevant model systems. For example, mouse models created by targeted mutations in the murine cftr gene typically exhibit severe digestive impairment but barely any lung disease (10), although some show spontaneous leukocyte accumulation in the lung interstitium (11, 12).

In the past decade, models of human development and disease based on tissue engraftment in xenotolerant hosts, e.g., severe combined immunodeficient (SCID) mice, have flourished. In particular, human fetal hematopoietic grafts in SCID mice have allowed investigators to explore important steps of blood cell development (13). We described previously how human fetal airways can also be developed and maintained in the long term in the SCID host (14). Complete histologic maturation of the surface epithelium and submucosal glands is reached past a few weeks in the host, with no difference between CF and non-CF grafts. Mature CF grafts display expected anomalies in ion transports, e.g., the absence of cyclic adenosine monophosphate-dependent Cl transport and increased relative amiloride-sensitive Na transport and adenosine triphosphate-dependent Cl transport. Remarkably, all features of mature grafts tested so far were independent of the age at implantation and the duration of engraftment, which emphasizes the reliability of this model.

In the present study, we used this unique model of naive and mature CF and non-CF human airways to address three important questions. First, what is the inflammatory status of the CF tracheal mucosa before any infection? Second, inasmuch as mature CF airway grafts display no gross histologic sign of disease before infection, will an experimental challenge with P. aeruginosa be sufficient to provoke a pathologic outbreak compared with matched non-CF controls? Finally, how do the processes of inflammation and infection as seen in this model match with what we know of early CF airway disease in vivo?

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Human Fetal Airway Grafts in SCID Mice

Experiments on human tissues and live animals were approved by CNRS Ethics Committee for Life Sciences. Fetal tracheas were obtained from miscarriages and medical abortions, in compliance with the current French legislation. In the case of diagnosed or suspected infection, tissues were discarded. All manipulations were performed under sterile conditions, i.e., under a laminar flow hood and using sterile instrumentation and sterile phosphate-buffered saline (PBS) supplemented with antibiotics (penicillin-streptomycin 1%; GIBCO, Cergy Pontoise, France). SCID mice were bred in our pathogen-free facility and anesthetized by intraperitoneal injection of 0.4 ml Hypnomidate (ethomidate; Janssen-Cilag, Issy les Moulineaux, France) before any manipulation.

A total of 13 CF (gestational age: 18.9 ± 1.2 wk, range: 10-24; engraftment age: 12.8 ± 1.4 wk, range: 8-22) and 22 non-CF (gestational age: 22.0 ± 1.4 wk, range: 14-36; engraftment age: 14.2 ± 1.2 wk, range: 8-25) histologically mature airway grafts from five CF fetuses (four F508/F508 and one F508/1717.1G A) and eight non-CF fetuses (four Down's syndromes, two spontaneous abortions, one anemia, and one cot death), respectively, were prepared for this study. As previously described (15, 16), tracheas were dissected into two to five cylindrical pieces and implanted subcutaneously in the flanks of 6- to 8-wk-old C.B.17 scid/scid (SCID) mice (one graft per flank). Connective tissue membranes lined internally with neoformed mature airway epithelium occluded both ends of the grafts (17), whose lumen was filled with liquid. The histologic maturation of the surface epithelium and submucosal glands was undistinguishable between CF and non-CF grafts (14). The whole surface epithelium was pseudostratified with few mucus-secreting cells and numerous ciliated cells, and submucosal glands included both mucous and serous cells (15, 16).

Collection of Intraluminal Airway Liquid and Measurement of the IL-8 Content

Six CF and eight non-CF airway liquid (AL) samples were collected for this part of the study. The method used to collect AL ensured minimal evaporation and preserved sample sterility, as previously described in full detail (16). In brief, the membrane occluding the terminal part of the graft was incised in a nonvascularized area. AL was collected with a positive displacement micropipette (Hirschmann Laborgerate, Mannheim, Germany) and frozen at 20°C until use. The IL-8 content of AL was determined using standard sandwich enzyme-linked immunosorbent assay procedure, as detailed elsewhere (18).

Preparation of Human Airway Grafts for Immunohistochemistry and Electron Microscopy

For immunohistochemistry, graft portions were embedded in cryoprotective medium OCT (Tissue-Tek, Miles, Inc., Elkhart, IN), frozen in liquid nitrogen, and cut into serial 5-µm frozen sections. Sections were postfixed for 5 min using either a PBS-10% formaldehyde solution at room temperature or 100% methanol (Merck, Darmstadt, Germany) at 20°C.

For scanning electron microscopy (SEM) and transmission electron microscopy (TEM), graft portions were fixed overnight with 2.5% glutaraldehyde in 0.15 M PBS, postfixed in 2% OsO₄, and quickly dehydrated in ethanol gradient. Specimens were finally embedded in Agar 100 resin (Agar Scientific, Orsay, France). Ultrathin sections were obtained on an ultramicrotome (Ultracut E; Leica, Rueil Malmaison, France) and specimens were observed using a transmission electron microscope (Hitachi H300; Elexience, Verrière le Buissen, France) at 75 kV, after counterstaining with lead citrate and uranyl acetate. For SEM, fixed samples were critical point-dried and coated with 15 nm gold-palladium particles. SEM was carried out using a Philips XL30 electron microscope (Philips, Limeil Brèvannes, France) operating at 10 kV.

Immunohistochemical Analysis of the Grafts before and after Infection

Detection of IL-8-positive cells within the grafts was performed using a mouse antihuman IL-8 monoclonal antibody (mAb) (1/50; Biosource, Camarillo, CA). Resident human leukocytes were identified with mouse mAb anti-CD45 (1/50 pan-leukocyte; Becton Dickinson, le Pont de Claix, France). Further characterization was performed using mouse mAbs antitryptase (1/100; Dako, Trappes, France) for mast cells, anti-CD68 (1/100; Dako) for macrophages, and anti-CD3 (1/50; Dako) for T lymphocytes. Murine leukocytes were identified with rat mAbs anti-Ly5 (pan-leukocyte, 1/200), anti-Gr1 (neutrophils, 1/200), and anti-Mac1 (macrophages, 1/100), all from PharMingen (le Pont de Claix, France). P. aeruginosa was identified using rabbit antiserum anti- P. aeruginosa PAO1 strain (P. aeruginosa wild-type strain, 1/25; Diagnostic Pasteur, Marne la Coquetto, France). Immunofluorescence was performed using corresponding biotinylated secondary antibodies and streptavidin-conjugated fluorescein isothiocyanate (1/600; Dako) or streptavidin-conjugated Cy3 (1/ 600; Dupont-NEN, Boston, MA), yielding green and red fluorescence, respectively. Nonspecific staining was prevented by preincubation with appropriate sera. Observations were made using an epifluorescence Nikon Axiophot microscope equipped with an Hg lamp (Nikon, Champigny sur Marne, France).

Semiquantitative Analysis of Leukocyte Localization in Airway Grafts before Infection

Because among genotype-matched fetal tracheal grafts the absolute number per type of inflammatory cells varied from one graft to the other (see RESULTS), we chose to study the distribution of inflammatory cells within or beneath the lamina propria. To evaluate the distribution of inflammatory cells within or beneath the lamina propria of CF and non-CF grafts, we performed a semiquantitative analysis on six CF and five non-CF fetal tracheal grafts (from five CF and five non-CF independent fetal tracheas, respectively). A total of 7 mm of basal lamina was screened on at least six different sections from each graft, after staining with a hematoxylin and eosin solution kit (Shandon, UK). Sections were observed at a constant magnification of ×40 under a Zeiss Axiophot microscope using eyepieces equipped with a graduated rule and a grid (Zeiss, le Pecq, France). Results were expressed as percentages of total inflammatory cells present either within the lamina propria (at a distance less than or equal to 100 µm) or beneath the lamina propria (at a distance ranging between 100 and 200 µm).

Experimental Challenge with P. aeruginosa

After biopsy and AL collection, airway grafts were dissected into hemicylindar portions of airway mucosa (of about 1 cm²). Graft portions were incubated at 37°C, 5% CO₂ for 1, 3, or 6 h after mucosal challenge with 50 µl of either RPMI medium or 10⁸ colony-forming units/ml P. aeruginosa suspension of the PAO1 strain, which had been cultured overnight at 37°C in trypticase soy broth under mild agitation and resuspended in RPMI medium after centrifugation. Graft portions were harvested at the time of initial biopsy and after 1, 3, or 6 h of contact with RPMI (control medium) or P. aeruginosa. The migration of human and murine leukocytes through the basal lamina of CF and non-CF grafts was investigated by immunohistochemistry using anti-CD45 and anti-Ly5 antibodies, respectively, as well as by TEM. Epithelial integrity, remodeling, and P. aeruginosa adherence to epithelial cells were assessed by immunohistochemistry, SEM, and TEM.

Data Analysis

Values are presented as means ± standard error. Data analysis was performed using STATISTICA software (Statsoft, Tulsa, OK) and Student's t test as appropriate. The percentages of total inflammatory cells present within and beneath the lamina propria in CF versus non-CF airway grafts were compared using a ² test. P < 0.05 was considered statistically significant.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Increased IL-8 Content in the Intraluminal AL of CF Grafts before Infection

AL from CF and non-CF grafts was collected and its IL-8 content measured (Figure 1). An 8-fold increase was found in the IL-8 content of CF AL compared with non-CF (10.1 ± 2.2 and 1.2 ± 0.6 ng/ml, respectively; P < 0.05). Within CF and non-CF groups, the IL-8 content was statistically independent of time variables (age at implantation and duration of engraftment). CF and non-CF groups were identical for the age at implantation and the duration of engraftment. We did not detect any difference in immunohistochemical distribution of IL-8 in CF versus non-CF grafts: submucosal glands, some surface mucous cells, and AL itself stained positively for IL-8, at the exclusion of any non-epithelial cell subset (not shown).

View larger version (12K): [in this window] [in a new window]   Figure 1.   IL-8 content of the intraluminal AL from CF and non-CF grafts. The IL-8 content was measured in the AL from six CF and eight non-CF grafts with similar mean time variables (age at implantation and duration of engraftment, P > 0.1 for both). Values obtained for CF AL, 10.1 ± 2.2 ng/ml, and for non-CF AL, 1.2 ± 0.6 ng/ml, were significantly different (P < 0.05). Within CF and non-CF groups, IL-8 values were statistically independent from time variables (P > 0.1 for both).

Abnormal Localization of Human and Murine Leukocytes in CF Grafts before Infection

We observed that both human and murine leukocytes coexist in the mesenchyme of the human airway grafts. Human neutrophils and other short-lived hematopoietic subsets (red blood cells, platelets) are not maintained in the human airway grafts. Resident human leukocytes typically include mast cells, macrophages, and, less frequently, T cells, which continuously colonize the mucosa, from 7, 10, and 12 wk of gestation, respectively. In the airway grafts, the amount of resident human leukocytes thus depends on the age at implantation. Murine leukocytes, which migrate into the human mucosa after implantation, mainly include neutrophils and macrophages. We observed that CF and non-CF tissues are colonized similarly, both before (human leukocytes) and after (murine leukocytes) implantation. A striking difference was observed in the localization pattern of leukocytes, depending on the genotype. In CF airway grafts, focal subepithelial clusters of both human (Figure 2B) and murine leukocytes (Figure 2D) were consistently identified. These subepithelial clusters were absent from non-CF grafts, where both human (Figure 2A) and murine (Figure 2C) leukocytes were mostly found deep in the mesenchyme. A semiquantitative analysis (Figure 2E) showed that the differential pattern of leukocyte localization in CF versus non-CF grafts was highly significant (P < 0.001) and statistically independent of time variables within each group. Time variables did not significantly differ between CF and non-CF groups.

View larger version (11K): [in this window] [in a new window]   View larger version (61K): [in this window] [in a new window]   Figure 2.   Localization of leukocytes in CF and non-CF airway grafts. Representative cross-sections of a non-CF (A and C) and a CF (B and D) graft stained with anti-CD45 (human leukocytes: A and B) and anti-Ly5 (murine leukocytes: C and D), respectively. Before infection, both human and murine leukocytes (arrows) accumulate under the basal lamina and occasionally within the surface epithelium (e) in CF grafts, whereas they are mostly found into the mesenchyme of non-CF grafts. Bar: 25 µm in A, C, and D and 50 µm in B, which is shown at lower magnification to illustrate the presence of several subepithelial clusters of human leukocytes in CF grafts at rest. (E) Most inflammatory cells (bars represent mean values) identified in CF grafts are located in the lamina propria (within 100 µm from the basal lamina), whereas in non-CF grafts, most inflammatory cells are located deeper in the mesenchyme, at a distance from the basal lamina exceeding 100 µm (P < 0.001). The leukocyte localization patterns in CF and non-CF groups were statistically independent of time variables. Mean time variables were similar in CF and non-CF groups.

Exacerbated Response to P. aeruginosa Primoinfection in CF Airway Grafts

Despite significant inflammatory imbalance, we found that naive CF airway grafts remained histologically unaffected. Our next step was thus to expose naive CF (n = 7) and non-CF (n = 11) grafts to mucosal challenge with P. aeruginosa (or medium as a control) for 1, 3, and 6 h. At 1 h after infection, in both CF and non-CF grafts, bacteria mainly adhered to luminal mucus (Figures 3A and 3B), while the integrity of the epithelium was preserved. In non-CF grafts, the pseudostratified organization was maintained after 3 and even 6 h of bacterial challenge (Figures 3C and 3E). By contrast, severe exfoliation of epithelial cells readily occurred in CF grafts at 3 h, culminating in an intense shedding at 6 h (Figures 3D and 3F). Bacteria adhered to exfoliated cells, infiltrated through the CF epithelium, and reached basal cells (Figures 4A and 4C), the basal lamina (Figure 4B), and the underlying mesenchyme (Figure 4D). Exfoliation of the CF epithelium as soon as 3 h after infection was concomitant with massive transepithelial migration of human and murine leukocytes (Figures 5B-5D), whereas in the non-CF mucosa, leukocytes were still mostly located under the basal lamina (Figure 5A). In non-CF grafts, events of transepithelial leukocyte migration, epithelial exfoliation, and bacterial penetration into the lamina propria were less extensive and only detected by 6 h after infection. No change in leukocyte distribution or epithelial architecture was detected in CF and non-CF grafts challenged with control medium.

View larger version (114K): [in this window] [in a new window]   Figure 3.   Successive steps of mucosal colonization by P. aeruginosa. Bacteria are indicated by arrowheads. At 1 h after infection, bacteria are mainly found trapped by mucus in residual AL (A, bar = 25 µm) and between cilia (B, bar = 2.5 µm). At 3 h, the non-CF surface epithelium is intact (C) while the CF surface epithelium (e) has lost its junctionality, allowing bacteria to enter intercellular spaces (D) and to gain access to basal cells and to the underlying the basal lamina (bl). At 3 h, SEM analysis shows bacteria adhering to residual AL on top of the still intact non-CF surface epithelium (E), while the CF surface epithelium (F) exhibits large areas of exfoliation, with bacteria drowning into intercellular spaces (original magnification: ×2,313).

View larger version (86K): [in this window] [in a new window]   Figure 4.   Increased mucosal damage in CF airway grafts after P. aeruginosa infection. Representative pictures were taken at 3 h after infection on portions of CF grafts, using immunofluorescent detection of P. aeruginosa (A and C: bar = 10 µm) and TEM (B and D: bars = 5 and 2.5 µm, respectively). Bacteria are indicated by arrowheads. At this intermediate timepoint, bacteria adhere to desquamated surface epithelial cells (A) and to basal cells (B and C); entire portions of surface epithelium (e) are exfoliating or totally detached in CF grafts, allowing bacteria to progress toward the basal lamina (bl) and in the mesenchyme (D).

View larger version (130K): [in this window] [in a new window]   Figure 5.   Transepithelial migration of leukocytes in infected CF and non-CF airway grafts. At 3 h after infection, CD45+ leukocytes (arrows) are massively transmigrating through surface epithelium (e) in a CF graft (B), whereas they are still mostly located under the basal lamina in a non-CF graft (A). Bar = 25 µm. TEM confirms that leukocytes cross the basal lamina (C) and are associated with epithelial exfoliation (ex) as they migrate toward the lumen in infected CF grafts (D). Bar = 3 µm.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

We previously observed (16) that AL from mature and naive CF and non-CF airway grafts exhibit similar viscosity (< 1 Pa.s), water content (> 97%), and ion composition. Here we present important complementary information with the finding that the IL-8 content of AL from CF grafts is specifically increased by about 8-fold compared with non-CF. This result is consistent with the increased IL-8 level found in primary cultures of CF airway epithelium (18) and in the bronchoalveolar lavage of CF infants and adults (6), although in that case neutrophils and macrophages may also contribute to IL-8 production. In our model, the IL-8 protein localized exclusively within submucosal glands and surface mucous cells. Therefore, the increased IL-8 content of CF AL likely relates to either a reduced turnover of the IL-8 molecule or an increased basal production by CF epithelial cells. Events involved in the selective upregulation of IL-8 in CF cells are poorly understood and may include a dysregulation of the transcriptional nuclear factor B-inhibitory complex, IB (21). Functionally, IL-8 is a potent chemoattractant not only for human neutrophils but also for murine neutrophils and other human leukocyte subsets. The enhanced recruitment of leukocytes that we observed in the lamina propria of CF grafts may thus, at least partly, be ascribed to the increased intraluminal IL-8 level. It may also relate to alternate factorse.g., IL-6; IL-1; IL-10; regulated on activation, normal T cells expressed and secreted; or IL-4 (20, 22, 23)which could not be investigated here due to the limited amounts of AL collected in the grafts.

Importantly, we found that both the molecular (IL-8) and cellular differences in the inflammatory balance of naive and mature CF and non-CF grafts did not depend on time variables, but exclusively on genotype. In the absence of any survey of inflammation in prenatal and early postnatal human CF airways, our observations represent the first conclusive evidence that the inflammatory balance is specifically altered in human CF airways before any other major AL alteration and most importantly, before the first infectious episode. Indeed, our study relies on human fetal tissues developed under pathogen-free conditions, as opposed to all clinical or experimental studies carried out so far, which were based on postnatal human CF airway tissues, biopsies, or cells having undergone previous infections. Our model, however, has some drawbacks, e.g., the limited availability of tissues and the absence of human neutrophils, although the latter may be partly compensated for by other leukocyte subsets (both human and murine).

When naive CF and non-CF airway grafts were further challenged with P. aeruginosa, we found that bacteria first adhered to the mucus present in the lumen, as previously reported in biopsies (24), but never to the pseudostratified epithelium. With the progressive loss of epithelial integrity, bacteria then adhered to exfoliated cells, to exposed basal cells, and to the basal lamina, which express high-affinity receptors for P. aeruginosa, e.g., the asialoGM1 (25), fibronectin, and the ₅₁ integrin (26, 27). Strikingly, the cascade of events after infectioni.e., loss of junctionality and bacterial penetration into the mucosawas greatly accelerated in CF grafts and correlated with the early and massive migration of subepithelial leukocytes to the lumen. Considering that inflammatory cells recruited to the airways can provoke epithelial cell detachment (28), increased IL-8 release in the lumen (29), and even increased bacterial adherence (30), we propose that the early and massive transepithelial migration of leukocytes in CF tracheal grafts is a major factor in the triggering of CF airway disease. Our results are compatible with the notion that significant disease is not seen in CF lungs before the first infectious outbreaks occur, i.e., after birth (2). Nevertheless, in several experiments using human fetal bronchiolar tissue, we observed massive murine neutrophil migration in the lumen of CF grafts and not in control grafts, before any infection (R. Tirouvanziam, unpublished data). This suggests that intrinsic differences may exist in the inflammatory status of distinct regions within CF airways, at least in this model.

Overall, this study adds to the growing body of evidence that inflammation occurs very early in CF airways, leading, when infection occurs, to the exacerbation of mucosal damage and paving the way for the relentless cycle of inflammation and infection that affects CF airways in vivo. In addition, our results support the use of integrated in vivo models, such as that of human CF and non-CF fetal airway grafts in SCID mice, to investigate the onset of CF airway disease, which still remains a major goal of CF research.