Keratinocyte growth factor-induced proliferation of rat airway epithelium is restricted to Clara cells in vivo

Keratinocyte growth factor (KGF) is a member of the fibroblast growth factor (FGF) family that specifically binds to the KGF receptor, the IIIb splice variant of FGF receptor 2. As FGF receptor 2IIIb is mainly expressed in epithelium, KGF specifically stimulates epithelial cells to proliferate 1, 2. In the lung, KGF (also known as FGF-7) is a potent mitogen of alveolar epithelial type II cells in vitro 3, 4 and in vivo 5–8. Kinetics of the incorporation of 5-bromo-2′-deoxyuridine (BrdU) into alveolar cells shows that hyperplasia of type II cells peaks at about 2–3 days after treatment with human recombinant (rHu) KGF (rHuKGF) given via the airways 5. As was comprehensively reviewed by Ware and Matthay 2, pre-treatment with rHuKGF protects the lung against a variety of lung injuries e.g. acid instillation 9, α-naphthylthiourea 10, hyperoxia 11, hydrostatic pulmonary oedema 12 and bleomycin 13. The mechanisms that are thought to contribute to the protective effect of rHuKGF include enhancement of alveolar repair, upregulation of alveolar epithelial fluid transport 14, 15, modulation of the alveolar coagulation/fibrinolysis system 11, and inhibition of alveolar epithelial cell apoptosis 16–18. However, very little is known about the potential contribution of the conducting airway epithelium to rHuKGF-related mechanisms of lung protection 19–21.

Although, in their initial paper, Ulich et al. 5 demonstrated that rHuKGF induced airway epithelial cell proliferation in rat lungs in vivo, little is known about which airway cells respond and their physiological alterations. rHuKGF induces proliferation of bronchial epithelial cells in primary culture and bronchial cell lines such as Calu-3, BEAS-2B and 4MBR-5 cells 22, 23. Polymerase chain reaction and northern blot analyses demonstrated the presence of transcripts of the KGF receptor in foetal and adult rat lung 24, as well as in primary cultures of normal human bronchial epithelial cells and in the bronchial epithelial cell line BEAS-2B 23. Transgenic mice that express human KGF under the control of the Clara cell secretory protein 10 kD (CC10) promoter upon doxycycline administration exhibited marked transient airway epithelial hyperplasia with thickening of the columnar epithelium and frequent transitions to a pseudostratified epithelium 8. Yano et al. 21 indicated that after instillation of rHuKGF in vivo, the bronchiolar epithelium responds with altered protein expression, i.e. an increase in surfactant protein D (SP-D), and a decrease in CC10. Nothing is known, however, about changes of the membrane-bound adhesion molecule CD44v6, which has been implicated in cell-cell and cell-matrix interactions as well as in airway epithelial repair 25, 26. Since CD44v6 is present in pulmonary stem cells of epithelial origin but not in cells of neuroendocrine lineage 27, CD44v6 may serve as an additional marker to evaluate the origin of the cells proliferating in response to KGF.

KGF was shown to alter gene expression in primary human airway epithelial cells in vitro 28. In particular, KGF decreased transcripts for many interferon-induced genes, which suggests that it may dampen the response of epithelial cells to inflammatory mediators. However, the airway epithelium is composed of a number of different cell types displaying distinctive structural and functional characteristics that may vary from species to species, e.g. the greater number of Clara cells in the larger airways of the lungs of rodents as compared with humans 29. While these previous studies document cellular changes in response to human KGF, it remains unclear which type(s) of airway epithelial cells proliferate(s) in response to rHuKGF in vivo and how their phenotype is altered.

Therefore, the purpose of this study was to investigate the proliferative response of specific airway epithelial cell types to human KGF. The effect of human KGF was studied using two in vivo models of transiently induced pulmonary epithelial cell proliferation. First, rHuKGF-protein was instilled via the left main bronchus to induce short-term hyperplasia, which resolves within 1 week 5, 7, 13. In a second model, the human KGF gene was introduced by instillation of a replication-deficient human type 5 adenovirus vector expressing human KGF to induce epithelial cell hyperplasia, which may persist for up to 3 weeks 30.

Material and methods

Results

Both endobronchial instillation of rHuKGF-protein and Ad5-HuKGF resulted in a remarkable increase in proliferating airway epithelial cells (fig. 1⇓). The proliferative response could be demonstrated by immunohistochemical detection of BrdU incorporated into the nuclei during S-phase as well as by immunostaining for Ki67, a nuclear proliferation marker expressed throughout the cell cycle (fig. 2⇓). The fraction of airway epithelial surface covered by BrdU-positive cells was highest at day 1 after rHuKGF instillation and gradually declined to zero at day 7 (fig. 3⇓). After adenoviral transfer of the human KGF-gene, the degree of proliferation on day 2 was similar to instillation of rHuKGF-protein (compare figs 3 and 4⇓⇓) whereas the levels of BrdU incorporation were higher at days 3 and 7 after gene transfer than after protein instillation. The fraction of airway epithelial surface covered by Ki67-positive cells peaked at day 2 after rHuKGF instillation and declined to control levels at day 7 (fig. 3⇓). After gene transfer, Ki67 gradually declined and only reached control levels at day 21 (compare figs 3 and 4⇓⇓). While BrdU-incorporation was always markedly higher in lungs exposed to rHuKGF, individual lungs instilled with PBS also demonstrated considerable immunoreactivity for Ki67. Consequently, the difference between rHuKGF- and PBS-treated animals in the fraction of Ki67 immunoreactive cells did not reach statistical significance, which is in clear contrast to the fraction of BrdU-positive cells.

Clara cells and pulmonary neuroendocrine epithelial cells (PNEC) are potential airway epithelial progenitor cells, particularly in the distal bronchioles 32, 33. To identify which one of these potential progenitor cells responded to KGF-treatment in vivo, immunoreactivities for CC10 and calcitonin gene-related peptide (CGRP) (fig. 1⇑), which are specific markers of Clara cells and PNECs, respectively, were examined. Further, double immunostainings were performed using antibodies against CC10 and CGRP in conjunction with staining for BrdU (figs 5 and 6⇓⇓). Using a monoclonal antibody to CGRP, no alterations in PNECs were observed throughout the experimental period (fig. 1⇑). Since only a few PNECs were observed per section, the labelling index for BrdU was determined from double immunostainings (table 2⇓). The authors observed that only three out of a total of 165 cell profiles of PNECs showed nuclear labelling for BrdU, observed in six lungs treated with rHuKGF-protein, while the total number of BrdU-positive cell profiles amounted to 4,085. In contrast, double immunostaining for BrdU and CC10 revealed that Clara cells constituted the majority of the proliferating population of distal airway cells following rHuKGF treatment in vivo (figs 5 and 6⇓⇓).

The phenotypic alterations associated with KGF-induced airway epithelial proliferation were defined based on changes in the immunoreactivity for CC10 and SP-D, respectively (fig. 7⇓), which are both well known secretary products of Clara cells 34. There were also changes in the expression pattern of the membrane-bound adhesion molecule CD44v6, which has been implicated in cell-cell and cell-matrix interactions and airway epithelial repair 25, 26, and which is not expressed in cells of the neuroendocrine system 27. Changes in CC10 and SP-D were estimated by means of stereology. The intensity of CC10 staining was considerably decreased after human KGF-treatment (p=0.001), whereas SP-D staining increased (p<0.001), as judged by estimating the airway surface fraction covered by CC10- and SP-D-positive cells, respectively (fig. 3⇑). Notably, the alterations in CC10-immunoreactivity were confined to a subpopulation of Clara cells that exhibited very weak staining, which was frequently restricted to the apical portion of the cytoplasm (fig. 7⇓). Until day 3 after instillation with rHuKGF, the fraction of these weakly staining cells increased (fig. 3d⇑) at the expense of Clara cells that strongly stained for CC10 (fig. 3c⇑). These alterations were highly significant in comparison to PBS-instilled lungs.

The immunoreactivity for CD44v6 was markedly increased in lungs that received rHuKGF-protein instillation or human KGF-gene transfer (fig. 7⇑). The alteration in cell shape from a cuboidal to a slightly columnar bronchiolar epithelium was particularly apparent as expression of CD44v6 increased at the lateral cell membranes. This effect was already seen at day 1 after instillation of human KGF protein. It was most prominent at days 2 and 3 after rHuKGF instillation, and persisted up to day 14 in lungs subjected to human KGF gene transfer. Again, the alterations of the airway epithelium had resolved at day 7 after rHuKGF-instillation and by day 21 after gene transfer. There was no effect on airway brush and alveolar epithelial type III cells identified by a monoclonal antibody against cytokeratin 18 (not shown).

In previous studies, the current authors have shown that apoptosis was involved in the resolution of alveolar epithelial hyperplasia 7. To address whether this was also true for the resolution of airway epithelial cell proliferation, double staining with an anti-cytokeratin antibody (MNF-116), a marker of epithelial cells, was performed in combination with the TUNEL assay 31. However, only single apoptotic airway epithelial cells were seen in human KGF-treated lungs (fig. 8⇓). Occasionally, macrophages containing apoptotic bodies were seen in the subepithelial interstitial tissue. However, there was no sign of an increased incidence of apoptotic airway epithelial cells in comparison with untreated lungs or lungs that had received PBS.

Discussion

This study investigated the in vivo response of the airway epithelium to a potent epithelial mitogen, KGF. The effect of KGF was examined in two in vivo rat models, i.e. endobronchial instillation of the recombinant human KGF-protein 13 and Ad5-HuKGF gene transfer 30, respectively. It has been demonstrated, by quantitative stereology in conjunction with immunohistochemistry, that either treatment resulted in transient proliferation as well as transient phenotypic alterations of the airway epithelium. By means of double immunolabelling experiments, only one cell type of the distal airway epithelium was shown to proliferate in response to KGF, the Clara cell.

Although previous studies have demonstrated that KGF is a mitogen of airway epithelial cells in vitro and in vivo 5, 6, 8, 21–23, nothing was known about the precise type of cell that responded to human KGF. In this study, it has been established by means of double immunofluorescence staining, that Clara cells are the major cell type of the bronchiolar epithelium that proliferate in response to human KGF-treatment. There were no signs of any significant proliferation of PNECs, another potential progenitor cell type of the airway epithelium, particularly in response to large-scale Clara cell loss 33, 35.

Although PNECs constitute a very small cell population of the airways and are characterised by a low rate of proliferation 36, PNECs may proliferate considerably under certain conditions. Experimentally induced hyperplasia of PNECs has been observed to occur in response to various toxic gases or carcinogenic stimuli 37. Depending on the specific stimulus, the proliferative effect may be specifically restricted to a certain airway region 38, 39. PNEC proliferation was also seen in a number of non-neoplastic human lung diseases 37. A marked increase in the number of PNECs was observed in neonatal bronchopulmonary dysplasia 40 as well as in human neonates with congenital diaphragmatic hernia (CDH) 41. In an experimental model of CDH, lung malformation and its rescue was reported to be associated with decreased levels of KGF protein and messenger ribonucleic acid (mRNA) 42. This is corroborated by studies performed in foetal mouse lung explants, which indicated that incubation with rHuKGF inhibits rather thanstimulates the proliferation of PNECs 37.

Hong et al. 43 demonstrated, in a very elegant transgenic approach, that when Clara cells were ablated, CGRP-immunoreactive PNECs were unable to repopulate the Clara cell-depleted distal airways. In addition, Borthwick et al. 44 reported that the potential stem cell population of the trachea, which was identified as BrdU label-retaining cells after oxygen saturation-induced injury, did not express the PNEC marker CGRP 44. As most neuroendocrine tumours and the cells of the neuroendocrine system do not express CD44v6 27, the increased expression of this membrane-bound adhesion molecule in response to rHuKGF further indicates that the responsive cells are not PNECs.

The proliferative potential of airway progenitor cells may depend on at least two factors not addressed in many of the previous studies: 1) the airway level examined and 2) the extent of injury produced. The pattern of epithelial repair was shown to vary widely depending on the degree of injury to the airways 45, 46. In the current study, airway level was carefully defined to include terminal bronchioles, the last conducting airway before the gas exchange region, and the extent of injury is not an issue here as human KGF was administered to cells in steady state.

As expected, the kinetics of cell proliferative and phenotypic changes differed between protein instillation and gene transfer in bronchiolar versus alveolar epithelial cells. Airway epithelial proliferation, assessed by immunohistochemistry for Ki67 as well as for BrdU, reached highest levels at day 1–2 after both types of KGF treatment. Thus, proliferation of Clara cells peaked at a slightly earlier time-point than the proliferation of alveolar epithelial type II cells (day 2–3) reported in previous studies 5, 7, 21, 47. This was even true for proliferation of the airway epithelium after adenoviral KGF gene transfer, where hyperplasia of type II cells also peaked at day 2–3 30 (unpublished data). These data indicate that Clara cells may act more rapidly upon stimulation of repair mechanisms than alveolar epithelial type II cells.

Airway epithelial proliferation returned to control levels by day 7 after instillation of rHuKGF protein and by day 21 after gene transfer, respectively. This compares well with the kinetics of restoration of the alveolar epithelium, which exhibited normal architecture after about 7 days following protein instillation 5, 7 and by day 21 following gene transfer 30 (unpublished data). Similarly, the marked airway epithelial hyperplasia seen in transgenic mice that conditionally express human KGF, under the control of the CC10 promoter during doxycycline administration, was reduced 7 days after cessation of doxycycline and was largely resolved after 14 days 8. It was found that 7 days after adenoviral transfer of the human KGF, airway epithelial proliferation was restricted to areas near foci of alveolar epithelial proliferation, which according to in situ hybridisation studies were most likely the result of KGF derived from solitary type II cells that express human KGF 30, and was completely resolved by day 21. Notably, KGF appeared to act in a negative feedback loop on the expression of its own receptor, which was downregulated at the transcription level by 2.3-fold in primary cultures of human airway epithelial cells exposed for 20 h to 50 ng KGF·mL⁻¹ 28. This may indicate that cells already stimulated by KGF may exhibit a weaker response to a second KGF stimulus.

Individual lungs instilled with PBS exhibited considerable immunoreactivity for Ki67, whereas BrdU incorporation was consistently higher in lungs exposed to rHuKGF. This finding indicates that the use of Ki67 as a proliferation marker may bear some inherent problem rather than that PBS is a mitogen of airway epithelial cells. A major problem in determining the proliferative or growth fraction of a given tissue is the definition of the nonproliferative state. Different techniques have been developed to assess cell proliferation 48. Standard procedures to determine the growth fraction are generally based on experimental protocols in which newly synthesised deoxyribonucleic acid (DNA) is labelled during the S phase, e.g. by offering BrdU to be incorporated into newly synthesised DNA. As has been pointed out by others 49–51, this approach may underestimate the growth fraction of cells. On the contrary, the nuclear protein Ki67 that is expressed throughout all phases of the cell cycle, i.e. from early G₁ through S, G₂, and M, labels all cells that have entered the cell cycle 51. Because only cells in G₀ are devoid of Ki67, this approach, however, may overestimate the growth fraction because all cells are labelled that have the potential to divide. In turn, these cells also have the choice to leave the cell cycle and become quiescent or terminally differentiated. Immunoreactivity for Ki-67 does not necessarily mean that a cell is unquestionably going to divide 51. The current data clearly indicate that rHuKGF induces airway epithelial cell division in the rat, as was demonstrated by the significant increase in cells that incorporated BrdU. On the contrary, the instillation of PBS alone may stimulate airway epithelial cells to enter the cell cycle (as evidenced by the immunoreactivity for Ki67), while additional events will be necessary to drive the cells towards mitosis. Therefore, Ki67 may not be the optimal choice to assess the mitotic activity of the airway epithelium in experiments that involve the instillation of fluid via the trachea.

The effect of human KGF on the airway epithelium was not limited to proliferation but included profound phenotypic alterations, which were the same irrespective of the mode of human KGF delivery to lung, i.e. protein instillation or gene transfer. The changes were most readily seen in the distal airways and were also transient. The bronchiolar epithelial cells, which are characterised by low cuboidal shape in control lungs, exhibited a more cylindrical shape after human KGF-treatment. The altered morphological appearance was accompanied by a marked increase in the expression of the membrane-bound adhesion molecule CD44v6 at the lateral cell membrane, an increase in cytoplasmic immunoreactive SP-D and a decreased cytoplasmic expression of immunoreactive CC10. While CD44v6 and CC10 reached control levels at day 7 after instillation of rHuKGF protein, the fraction of airway epithelial cells that stained for SP-D was still above control level. These data extend previous findings that the bronchiolar epithelium responded with an increase in SP-D, a transient decrease in CC10 after instillation of KGF protein in vivo, both at the level of protein and mRNA, whereas SP-A appeared unchanged 21. The data of Yano et al. 21 rely on the analysis of whole lung homogenates. Looking at alveolar type II cells isolated from KGF-treated lungs, however, the same study demonstrated that the levels of SP-D mRNA were decreased on a per cell basis 21. As appropriate techniques to isolate Clara cells at a sufficiently high purity and cell number are still lacking, quantitative morphometric labelling experiments are the only approach by which cell type-specific changes can be monitored. Using this approach, the current authors have demonstrated a shift from Clara cells that exhibit strong cytoplasmic immunoreactivity for CC10 to Clara cells that exhibit only very weak apical labelling for CC10. In addition, they have demonstrated that the fraction of cells which are strongly immunoreactive for CC10 decreased while weakly staining cells increased after gene transfer. Hence, these data suggest that the expression of CC10 is not homogeneously reduced in all Clara cells at a time whereas there is not a decrease but an increase in immunostaining for SP-D, which is a more uniform effect. In contrast, however, CC10 immunoreactivity appeared normal in transgenic mice that over express KGF after doxycycline administration 8.

Until now, the functional consequences of these alterations induced by human KGF were not known. CC10-knockout mice exhibited an increased susceptibility to inhaled oxidant gases 52. Singh and Katyal 34 suggested that CC10 may function as a regulator of inflammation in the lung, a notion which is still a matter of debate 28, 52–55. CC10 was reported to be decreased in various conditions of inflammation 53, 56, 57, and deficiency in CC10 as a result of transgenic experiments or in humans with a chronic CC10 deficiency, e.g. in long-term cigarette smokers 58, 59, appears to be characterised by a tendency towards an exaggerated inflammatory response 53, 55, and may contribute to carcinogenesis 59. The potential of recombinant human CC10 as a therapeutic for inflammatory and fibrotic diseases is currently under investigation 55.

While in control rat lungs weak immunoreactivity for CD44v6 was observed in only a few airway epithelial cells, including bronchiolar Clara cells, CD44v6 was markedly increased after human KGF treatment. Notably, CD44v6 was markedly increased in human bronchial epithelial cell lines after mechanical damage in an in vitro model of repair 26. As CD44v6 was also seen to be markedly increased in hyperplastic alveolar epithelial type II cells after instillation of rHuKGF and was observed at the lateral membranes of type II/type I cell intermediates 7, the increase in CD44v6 most likely mirrors its functional involvement in cell migration, differentiation and/or altered cell-substrate adhesion in epithelial repair, as was also indicated by other studies 22, 26. In turn, loss of CD44v6 may result in detachment from the extracellular matrix with subsequent apoptosis 60.

The orchestration of the sequence of processes defined as “repair”, which includes proliferation, migration, differentiation and death of airway epithelial cells, is necessary to reconstitute the normal architecture of the airway epithelium after acute injury 46, 61. In turn, chronic lung diseases like asthma or chronic obstructive pulmonary disease are characterised, among other features, by profound alterations of the airway epithelial architecture resulting in a “remodelled” airway wall 62. The process of airway wall remodelling has been attributed to the combination of chronic repetitive injury on one hand and the ensuing repair process on the other. For example, exposure to environmental cigarette smoke prior to lung injury resulted in a delay in bronchiolar repair in mice 63.

As was comprehensively reviewed recently by Ware and Matthay 2, KGF has impressive protective effects against a wide variety of injurious stimuli when given as a pre-treatment in animal models. Various mechanisms are thought to contribute to this protective effect including enhancement of alveolar repair and inhibition of alveolar epithelial cell apoptosis 16–18, 64, and upregulation of alveolar epithelial ion and fluid transport 14, 65, which may be particularly important in acute lung injury models. KGF enhanced active ion transport across monolayers of isolated rat alveolar epithelial type II cells, primarily due to increased expression of Na⁺-K⁺-adenosine triphosphatase α₁-subunit 14. In the normal rat lung in vivo, intratracheal pre-treatment with KGF was reported to increase alveolar epithelial fluid transport primarily by its mitogenic effect on alveolar epithelial type II cells 65. Since Clara cells were reported to be involved in regulated ion and fluid transport in the distal airways 66, 67, future studies will be welcome that investigate the potential linkage between the induction of mitosis in Clara cells by KGF, as shown in the present study, and bronchiolar ion and fluid transport to reveal any airway-related contribution to the protective effect of KGF against pulmonary oedema in acute lung injury.

In this study, the effects observed were a result of the administration of KGF along the airway route. It would be interesting to learn from future studies if intravenous application of KGF induces similar changes in proliferation and phenotype of airway epithelial cells. Guo et al. 47 reported that intravenous administration of rHuKGF resulted in an equivalent proliferative response of rat airway epithelium as compared with intratracheal instillation, whereas the response of alveolar epithelial type II cells to intravenous administration was markedly lower. Notably, the same study reported correspondingly low responses of both alveolar and bronchial epithelium after intravenous versus intratracheal administration in mice.

In conclusion, the great potential of human keratinocyte growth factor to protect the lung against various injuries may not only rely on its effects on the alveolar epithelium but is likely to include its ability to stimulate processes implicated in airway epithelial repair, e.g. proliferation of Clara cells and increased expression of epithelial CD44v6. The contribution of other effects, e.g. on airway ion and fluid transport, awaits further elucidation.