INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The conducting-airway epithelium is lined by a pseudostratified epithelium, which is a complex epithelium composed of ciliated, secretory, and basal cells (1, 2). Epithelial cells contain keratin filaments, which belong to a family of more than 20 different cytoskeletal proteins (3); different cell types express different pairs of keratin proteins (4). In the respiratory epithelium, basal cells are the only cell type that are firmly attached to the basement membrane through hemidesmosomes (5). Keratins 5 and 14 are preferentially expressed in basal cells (6) and may form a resilient filament network (6) attached to hemidesmosomes. Basal cells are ubiquitous in the human conducting-airway epithelium, although the numbers of these cells are reduced in the peripheral airways (1, 9). The number of basal cells is related to the height of the columnar epithelium (5), underscoring the role of these cells in anchoring the pseudostratified respiratory epithelium to the basement membrane.

In the epidermis, keratins 5 and 14 are reliable markers of proliferative basal cells (10). In analogy to the situation in the skin, the basal cell has been suggested by many researchers to be the stem cell in the tracheobronchial epithelium (11). In the bronchiolar epithelium, where basal cells are sparse or absent, nonciliated columnar cells (i.e., the Clara cell) are thought to be the stem cells (16). However, the concept that the basal cell is the stem cell of the respiratory epithelium of the major airways has been challenged by several studies. Basal cells are the last cell type to appear in fetal lung development (17). Work performed with animal models has shown that the nonciliated columnar cell is the predominant proliferative cell type in the response to chemical injury (5). Because the composition of epithelial cell populations varies markedly between different species (17), application of concepts obtained from experimental models to the human lung has to be done with great caution.

Intermediate cells form a poorly defined layer located just above the basal cells, which they resemble electron microscopically (1, 18). In addition to the suprabasal location of their nuclei, intermediate cells have been defined light microscopically by the absence of morphologic characteristics of ciliated, neuroendocrine, and secretory cell types (11, 13, 18, 19). Intermediate cells meeting this definition have been shown to contribute to cell renewal in rodent conducting-airway epithelium (11, 13, 19). Although Donnelly and colleagues (13) tend to regard intermediate cells as constituting a legitimate cell compartment composed of differentiating cells derived from basal cells, other authors hesitate to attribute a role of this category to cell renewal (19) for lack of positive identification criteria for this cell type. However, if intermediate cells are defined not only by morphologic criteria but also by a positive immunoreactivity of keratins 5 and 14 shared with basal cells, then these "parabasal" cells could represent a reasonably demarcated and thus quantifiable cell compartment.

The proliferation fraction of the human airway epithelium has been the subject of two recent studies (20, 21). According to these reports, less than 1% of airway epithelial cells are proliferating at any time; cycling cells are predominantly located in the proximity of the basement membrane. These findings suggest that the basal cell represents the major cell type of the proliferative compartment in the human respiratory epithelium, but detailed quantification relating to this issue is lacking.

The aims of this study were to quantify in normal human airway epithelium: the number of basal and parabasal cells as defined by cell-nucleus position and the expression of keratins 5 and 14; the overall proliferation compartment; and the contribution of basal and parabasal cells to the proliferation compartment. To this end, human lungs, obtained by autopsy and considered by rigorous criteria to be normal, were fixed by tracheal infusion of formalin, and cross-sections from central to peripheral airways were stained sequentially with an antibody to the proliferation-associated protein Ki-67 (MIB-1) and to the antikeratin antibody 34E12 as a marker for (para)basal cells.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Material

In De Wever Hospital, Heerlen, and Maastricht University Hospital, Maastricht, The Netherlands, perfusion fixation of lungs was performed if deceased patients were available for autopsy within 6 h after death. Perfusion fixation was done in lungs of 275 cases. Seven samples were taken, from the trachea down to peripheral lung tissue, and were routinely processed. A standard hematoxylin and eosin (H&E) stain was used for histologic assessment. In order to obtain a more or less normal subject population, all cases with primary or secondary pulmonary diseases were excluded. After these selection procedures, a total pathologic score (TPS) was established for each slide according to Berend and colleagues (22). Four variables: inflammatory cellular infiltrate, pigmentation, fibrosis, and muscle hypertrophy, were scored in small airways with an internal diameter of 2 mm or less, using both H&E and elastica-von Gieson-stained sections. The four variables were graded from 0 (= normal) to 3 (= severely abnormal) for each airway, after which an average score was calculated for each variable for each slide. Slides of patient material were independently assessed by two observers (J.E.B. and F.B.J.M.T.). The calculated Cohen's kappa score (23) for these two observers was +0.77 (95% CI: 0.52 to 1.02). From the average scores of the slides, a TPS score was calculated as follows: the average score for each variable was expressed as a percentage of the maximum score. Next, the four percentages were added. The maximum theoretical TPS was 400.

Seven cases (six nonsmoking, one with unknown smoking status) with a TPS score of less than 100 remained for further study. In five cases the patients were female and in two male. The average age was 61 yr (range: 24 to 84 yr). The causes of death were myocardial infarction (three patients); cerebral infarction (one patient); saddle embolus of the lung (one patient); sudden death with aortic stenosis (one patient); and sick sinus syndrome (one patient).

Immunohistochemistry

Three-micrometer-thick tissue sections were placed on 3-amino propyltriethoxysilane-coated slides (Sigma Chemical Co., St. Louis, MO) and dried overnight in a convection oven (56° C). Slides were dewaxed in xylene, rehydrated in graded alcohols, and rinsed in distilled water. Subsequently, the slides were placed in plastic jars containing 500 ml of 10 mM citrate buffer in distilled water, pH 6. Two jars were placed in a 650W microwave oven, and irradiated to the boiling point. This was followed by 15 min of uninterrupted boiling, using full-power irradiation. After the jar was removed from the microwave oven, cooling to room temperature was achieved in about 20 min. Slides were incubated with the MIB-1 antibody (24) (Immunotech, Marseille, France; dilution 1:25) for 45 min at room temperature. Slides were washed thrice with phosphate-buffered saline (PBS), incubated with biotinylated rabbit antimouse antibody (Dako, Glostrup, Denmark; 1:500), and subsequently incubated with peroxidase-conjugated streptavidin label (Dako; 1:1,000). The slides were then incubated with diaminobenzidine (Sigma) for 10 min. Subsequently, slides were incubated with 0.0025% pepsin in 0.1 N HCl for 10 min at room temperature. After rinsing twice with PBS, the slides were incubated with the monoclonal antibody 34E12 (Dako; 1:400) for 45 min at room temperature. The 34E12 antibody is specific for high-molecular-weight keratin pair 1/10, which is present only in keratinized squamous epithelium, and for intermediate-molecular-weight keratin pair 5/14 (4, 25). After incubation with the 34E12 antibody, biotinylated rabbit-antimouse (Dako; 1:500) and alkaline phosphatase-conjugated streptavidin label (Dako; 1:200) were used as subsequent layers. After rinsing in PBS, slides were incubated with New Fuchsin (Biogenex, San Ramon, CA), and were then rinsed in PBS, counterstained in Mayer's hematoxylin, and placed in tap water for 15 min. Three drops of Imsol/Mount (Klinipath, Zevenaar, The Netherlands) were applied to each tissue section. Slides were then allowed to dry and mounted in Entellan (Merck, Darmstadt, Germany).

Negative control slides were stained with omission of the monoclonal antibodies MIB-1 and/or 34E12. As a positive control, a slide was used from a bronchial resection margin of a pneumonectomy specimen. A fixation delay study performed on a pneumonectomy specimen showed no impact of fixation delay of up to 12 h on either MIB-1 or 34E12 immunoreactivity.

Image Analysis and Measurements

An interactive image analysis system consisting of the Quantimet 570C (Leica, Wetzlar, Germany) and a microscope was used to investigate the specimens and controls. The epithelium of the slide-mounted specimens was assessed as follows: first, the internal diameter of the airway was measured, using a measuring rule at low power. Total epithelial thickness was assessed with the measuring rule at high power (×40 objective, numerical aperture = 0.65). Epithelial thickness was defined as the distance between the basement membrane and the luminal cell membrane excluding the cilia (26). Next, digitized video images were made of the airway epithelium. One video image (i.e., an epithelial segment) showed 25 to 50 airway epithelial cells on 140 µm basement membrane. Only epithelium with an orientation perpendicular to the basement membrane was evaluated. A line was drawn manually over the image of the basement membrane, using the computer mouse, and the luminal side of the epithelium was marked. The length of basement membrane was automatically determined. Subsequently, only nucleated epithelial cells were counted manually, and these were divided into six categories: total number of epithelial cells; number of basal and parabasal cells; number of MIB-1-immunoreactive cells; and number of MIB-1-positive basal and parabasal cells. The total number of epithelial cells was assessed by counting all nuclei located at the luminal site of the basement membrane. A cell was identified as a basal cell when the cytoplasm surrounding an airway cell nucleus was brightly immunostained with the monoclonal antibody 34E12 and the nucleus was adjacent to the basement membrane. A parabasal cell was defined as a cell with a brightly immunolabeled cytoplasm without cytoplasmic projections to the luminal surface, and with a nucleus positioned above the nuclei of basal cells but below the nuclei of luminal columnar cells. Occasionally, a ciliated or nonciliated columnar cell was 34E12 immunopositive. These cells were recognized as luminal cells by their position in the epithelium, and were not included in the parabasal-cell category. Epithelial-cell nuclei that stained unequivocally with diaminobenzidine were regarded as MIB-1-positive cells. MIB-1-positive (para)basal cells were calculated per individual airway as a percentage of the total number of MIB-1-positive cells. Reproducibility of the number of all airway epithelial cells, basal and parabasal cells, and MIB-1-positive cells was > 90%.

Data Handling and Statistics

During assessment of each airway, the number of cells within each of six categories (i.e., total number of airway epithelial cells, basal cells, parabasal cells, total number of MIB-1-positive cells, and basal and parabasal cells with MIB-1-positive nuclei), and the length of basement membrane, were determined in video images of epithelial segments. An epithelial segment corresponded to the amount of epithelium present in the ocular field of view. Data were aggregated until 10 successive epithelial segments were counted. These aggregated data (further referred to as a "data-batch") were recorded in a computer, with the internal airway diameter and epithelial height added manually. A data-batch contained 250 to 500 cells and approximately 1.4 mm of basement membrane. Assessment of the epithelium of an airway was terminated when a total of 30 epithelial segments were examined. For small airways with an internal diameter of less than 2 mm, only a few successive epithelial segments were present because of their small size; the data for several small airways were added for a particular slide when the difference in internal diameters was less than 0.2 mm. In this manner, 30 successive epithelial segments could be assessed in the small airways.

Cumulative percentages were calculated for each successive data-batch for each airway for five parameters: the percentages of basal cells, parabasal cells, MIB-1-positive cells, and basal and parabasal cells with MIB-1-positive nuclei. The airways were divided into four groups according to internal airway diameter:  4 mm,  2 mm but < 4 mm,  0.5 mm but < 2 mm, and < 0.5 mm. The data presented are the cumulative percentages calculated as follows: the mean cumulative percentage of a particular parameter was calculated per patient and per airway-diameter group. This resulted in 7 × 4 = 28 means (in total there were 92 assessed airways), as Table 1 shows for basal cells. The mean of a particular parameter was then obtained per airway diameter group by calculating the mean of the individual patient data for this airway-diameter group (column mean in Table 1). Also, the overall mean was obtained by calculating the mean of the 28 individual patient means. The SD was calculated with the nested analysis of variance (ANOVA) procedure (27); this approach not only detects variance between patient means, but also between the different observations for a particular patient and within an airway group. The SD was established as the total SD per airway-diameter group, but was also calculated separately for the data obtained among patients and within patients. For obtaining the SD of the overall mean, only the SD values calculated for different patient means were used, since the variances within patient data were large because of the different airway-diameter groups.

Studies of Feasibility of Reducing the Required Number of Epithelial Cells per Individual Airway for Reliable Estimation of Cell Proportions

A study was done of the feasibility of reducing the number of epithelial cells required for establishing reliable estimates of the proportion of (para)basal cells and MIB-1-positive cells in airway epithelium. A standard error (SE) of cumulative proportion of each data-batch, using the formula SE_prop = sqrt(p[1  p]/n), was calculated, in which p = cumulative proportion of a cell type [MIB-1-positive or (para)basal cells] and n = cumulative number of all airway epithelial cells (28). When the SE_prop of all three cell types (i.e., basal cells, parabasal cells, and MIB-1-positive cells) was < 0.025, the cumulative data for these cell types and the length of basement membrane obtained with the data-batches up to that point for that particular airway were used for subsequent analysis. With SE_prop = 0.025 used as a cutoff point, the corresponding 95% CI is p ± 0.05. The extent of the 95% CI is 0.10, irrespective of the magnitude of the proportion. Using the SE_prop = 0.025 cutoff point, 39,422 epithelial cells and 150 mm of basement membrane (263 cells/mm basement membrane) were used for data analysis, representing 39% of the original data set. A mean of 429 epithelial cells per airway were taken into account. The numbers of basal cells, parabasal cells, MIB-1-positive cells, MIB-1-positive basal and parabasal cells calculated after using the SE_prop method were compared with the original data presented in Table 2. The difference in the numbers of basal cells, parabasal cells, and MIB-1-positive cells was less than 3.5%; the difference in the number of MIB-1-positive basal cells and parabasal cells was less than 10%. Therefore, use of the calculated SE of proportion seems to be validated in the quantitation of (para)basal cells and MIB-1-positive cells in human conducting airway epithelium.

Reduction of Number of Individual Airways to be Evaluated per Patient and per Airway-diameter Category

After application of the SE_prop formula to reduce the number of epithelial cells required per individual airway, we conducted a subsequent feasibility study in order to establish a reasonable number for the airways that had to be sampled to adequately reflect the population mean. Using the SDs obtained with the nested ANOVA procedure, the number of airways required per patient and per airway category was calculated at a 90% confidence level.

For the largest airways, the extent of the 90% CI was chosen to be 24% for basal cells, 11% for parabasal cells, and 1.35% for MIB-1-positive cells. These CIs were arbitrarily chosen to correspond to the 90% CIs of the population means as calculated from the original data set (presented in Table 2). The 90% confidence limits were reached after assessment of 400 epithelial cells in each of two individual airways in this largest-airway-diameter category. For the other three airway-diameter categories, the outcome of the statistical procedures was similar (i.e., assessment of two airways per patient and per category).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In total, 101,800 cells and 375 mm of basement membrane were counted; the number of assessed airways and epithelial cells per airway-diameter group are presented in Table 2. The mean number of airway epithelial cells expressed per millimeter of basement membrane was 338 cells/mm in the airways with the largest diameter ( 4 mm), and 173 cells/mm in the smallest airways (< 0.5 mm). Per airway, a mean of 1,107 epithelial cells were assessed.

Basal and Parabasal Cells

A total number of 25,075 basal cells (Figure 1) were counted. Table 1 shows the number of basal cells expressed as the mean percentage of airway epithelial cells per patient and per airway-diameter group. In airways with a diameter of 2 mm or more, 30 ± 7% (mean ± SD) of epithelial cells were basal cells. Basal cells were present in the smallest airways, with a diameter < 0.5 mm, but the percentage basal cells (6 ± 4%) was considerably lower than in the larger airways. Comparing the number of basal cells expressed per millimeter of basement membrane with the total height of the airway epithelium revealed a strong correlation (r = 0.83; p < 0.001).

View larger version (162K): [in this window] [in a new window]   Figure 1.   (a) Bronchus with 34E12-immunostained basal cells and parabasal cells (dark gray cytoplasm). Black, MIB-1-positive nuclei are seen in two basal cells (arrowheads) and one parabasal cell (arrow). (b) Bronchus with immunostained basal cells (dark gray cytoplasm), one having an MIB-1-positive nucleus (arrow). In addition, a single parabasal cell is present with an MIB-1-positive nucleus (arrowhead ). (c) Bronchiole with two basal cells in the left-hand corner (arrowheads). A basal cell and a ciliated cell, both with MIB-1-positive nuclei, are present in the middle of the photograph (arrow). (d ) Bronchiole with a single basal cell (arrow). All photomicrographs, magnification: ×1,000.

A total number of 3,774 parabasal cells were counted, predominantly in airways with a diameter of 2 mm or more (Table 2). Parabasal cells were absent in the smallest airways, with a diameter of < 0.5 mm. The combined number of basal and parabasal cells per millimeter of basement membrane was strongly correlated with the total height of the airway epithelium (r = 0.84; p < 0.001).

Proliferation Fraction

The proliferation fraction was defined by the immunoreactivity of airway-cell nuclei with the MIB-1 antibody (24). A total of 844 of 101,840 airway epithelial cells were immunoreactive. The proliferation fraction expressed as the percentage of MIB-1-positive epithelial cells was calculated per patient and per airway-diameter group (Table 2). The overall mean proliferation fraction was 0.87 ± 0.30%.

Of 25,075 basal cells, 423 cells, or 1.69% of the basal-cell compartment, were MIB-1 immunoreactive (Figure 1). On average, basal cells represented 48% of the proliferation fraction of the conducting-airway epithelium. In airways with a diameter of 0.5 mm or more, over 50% of the airway epithelial cells constituting the proliferation fraction were basal cells (Table 2, Figure 2). In the smallest airways (diameter < 0.5 mm), 30% of proliferative cells were basal cells. Of 3,774 parabasal cells, 198 cells, or 5.25% of the parabasal-cell compartment, were MIB-1 immunoreactive (Figure 1), constituting 15% of the proliferation fraction. In the largest airways (diameter  4 mm), 33% of proliferating epithelial cells were parabasal cells (Table 2, Figure 2). No significant correlation was present between the percentage of proliferating cells as either basal or parabasal cells and epithelial thickness (p = 0.2 and p = 0.06, respectively).

View larger version (31K): [in this window] [in a new window]   Figure 2.   The contribution of basal parabasal cells to the proliferation compartment per airway-diameter group.

Influence of Age on Number of (Para)basal and MIB-1-positive Cells

The patient group had a wide age range (i.e., 24 to 84 yr). No systematic differences were present in the number of basal, parabasal, or MIB-1-positive cells in the conducting-airway epithelium of patients of different ages.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The purpose of this study was to determine the distribution of basal and parabasal cells in normal human airway epithelium, using the antikeratin antibody 34E12, and to establish the proportion of proliferating airway epithelial cells that consists of basal and parabasal cells. This was the first systematic assessment of the distribution and proliferation of basal and parabasal cells in human lungs. The mean percentages of basal and parabasal cells were 23% and 3%, respectively. The overall proliferation compartment as defined by MIB-1 immunoreactivity was 0.87%; 48% of proliferating epithelial cells were basal cells and 15% were parabasal cells.

The number of basal cells expressed per millimeter of basement membrane was strongly correlated with the height of the respiratory epithelium (r = 0.83). This anatomic relationship corresponds to the findings of Evans and associates (26), who studied the airway epithelium in several species. The plausible role of basal cells in attachment to the basement membrane of complex epithelia in general has also been discussed by Purkis and colleagues (6). They hypothesized that keratins 5 and 14 may form a filamentous network that is tough and elastic. In our view, basal cells with a resilient cytoskeleton could play a useful supportive role in locations where the airways are relatively rigid. In the more delicate and elastic terminal and respiratory bronchioles, such tough supportive cells seem to be less needed.

Intermediate cells are a poorly defined cell population (1) located between basal cells and columnar cells. With electron microscopic techniques, morphologic similarities have been noted between intermediate cells and basal cells (1, 18), apart from the position of the cytoplasm relative to the basement membrane. Because of these similarities, intermediate cells could be named "parabasal" cells. On a light-microscopic level, intermediate cells have been defined negatively, as having nuclei positioned neither basally nor luminally, and with cytoplasm having neither cilia nor secretory granules (11, 13, 19). The intermediate-cell population defined in this way could comprise not only parabasal cells, but also secretory cells without readily discernible granules. In addition to using the morphologic criteria regarding cell-nucleus position and cell differentiation, we have defined the parabasal cell population by its immunoreactivity of the antikeratin antibody 34E12; the combined criteria efficiently eliminate both basal cells and columnar cells (Figure 1).

Proliferation of the respiratory epithelium has been studied predominantly in the large airways of laboratory animals. The basal cell is the dominant cell type in the proliferation fraction of normal airway epithelium when plastic tissue sections of rodent lungs are assessed shortly after [³H]thymidine pulse- labeling (11, 29). Nonciliated columnar cells (i.e., Clara cells) do contribute to the proliferation fraction of the rodent respiratory epithelium in the steady state, but to a lesser extent than do basal cells. The question remains whether basal cells are the stem cells of the mammalian conducting-airway epithelium.

The concept of stem cells is derived from studies performed on tissues with rapidly renewing cell populations, including blood and skin (30, 31). In these tissues, certain cells are thought to have an almost unlimited capacity for self renewal. These putative stem cells are considered to be slowly cycling (if at all) in a specific microenvironment (30), and to produce a transit amplifying cell population. This committed cell population comprises several generations of rapidly proliferating cells with a finite self-renewal capacity; as progressive differentiation of cell generations ensues, the self-renewal capacity eventually ceases (30). Direct identification of stem cells is virtually impossible, since specific markers for these cells are absent (30).

Examination of lung-tissue sections obtained from laboratory animals at various times after pulse-labeling with [³H]thymidine can be used to establish the temporal and spatial relationships of the dividing airway epithelial-cell population (12, 14, 32). Donnelly and associates (13) studied plastic tissue sections of rat tracheal epithelium up to 10 d after [³H]thymidine pulse-labeling. The fraction of [³H]thymidine-labeled basal cells peaked at 1 d after pulse-labeling, followed by a peak of intermediate cells at 2 d, and of both secretory cells and ciliated cells at 3 d. Donnelly and associates concluded that basal cells are the progenitor cells of the tracheal epithelium, and divide to yield intermediate cells, which in turn give rise to luminal epithelial cells (ciliated and nonciliated). Other researchers, using identical techniques, described a more complex picture of cell proliferation and differentiation in rats (11, 12) and hamsters (19). In these studies, basal cells appear to have been the most important progenitor cells in the conducting epithelium of the large airways in the steady state, with the capacity for self-renewal (19) suggesting stem-cell properties. Secretory-cell types (in particular Clara and serous cells) also contribute to cell renewal, and in fact appear to be the stem cells of the small airways, where basal cells are sparse or even absent (33). The complexity of airway epithelial-cell proliferation and differentiation becomes even more acute when rat epithelium is studied after exposure to tobacco smoke (12) or nitrogen dioxide (29). Secretory cells are the dominant cell type in the proliferative response to airway epithelial injury in these studies, which have led Evans and associates to conclude that the basal cell is not a primary progenitor cell at all, but plays an adhesive role only.

In humans, sequential tissue sampling after injection of a label is virtually impossible in the study of the conducting-airway epithelium in the steady state. Alternatively, immunohistochemical detection of naturally occurring cell-proliferation-associated nuclear proteins, such as Ki-67, can reliably identify proliferative cells (34, 35). These methods cannot reveal the temporal relationships of the dividing and differentiating cell populations, but do demonstrate those cells that are part of the proliferation compartment. When the number of the various epithelial-cell types of the conducting-airway epithelium are correlated with their respective shares of the proliferation compartment, insight may be gained into the role of these different cells in maintenance of the normal human airway epithelium. In the present study, basal cells were found to represent over 50% of the proliferation compartment of the conducting respiratory epithelium of airways with an internal diameter of 0.5 mm and more (Table 2 and Figure 2) whereas these cells were found to comprise up to 31% of the epithelial-cell population. Thus, 1.6% of the basal cells in these large airways contribute to the proliferation compartment of the tracheobronchial epithelium. If the relative percentage of a cell type that contributes to the proliferation compartment (i.e., the percentage contribution to the proliferation compartment of a cell type divided by the percentage of the same cell type contributing of the airway epithelium) is > 1%, this may be defined as "overrepresentation" of this cell type in the proliferation compartment; such overrepresentation might indicate an essential proliferative function. Overrepresentation of basal cells in the proliferation compartment was also conspicuous in the smallest airways (diameter < 0.5 mm), where the number of basal cells was 6%, whereas the share of these cells in the proliferative compartment was 29%; the relative percentage of basal cells contributing to the proliferation compartment was 4.8%. These findings indicate an important role of the basal cell in renewal of the human conducting-airway epithelium in the steady state, in addition to a possible adhesive function of basal cells. These two roles are not incompatible, but are instead a typical phenomenon in cells expressing keratins 5 and 14, as described in several tissues other than the lung. The expression of the "basal" type keratins 5 and 14 is characteristic of the compartment of proliferating cells in cornifying and noncornifying stratified squamous epithelia (7, 10). In these epithelia, basal cells are known to play a vital role in the adhesion of the epithelium to the basement membrane (6).

Basal cells are not the only cell type immunolabled with antikeratin 5 and 14; parabasal cells are also present in normal human airway epithelium. With 7% of the epithelial cells in the largest airways (i.e., diameter  4 mm) consisting of parabasal cells, which represent 33% of the proliferation fraction (Table 2 and Figure 2). In airways with diameters of 2 to 0.5 mm, overrepresentation of parabasal cells is even more apparent: 1% of epithelial cells are parabasal cells, whereas the share of these cells in the cycling-cell compartment is 8%. Parabasal cells, having an intermediate position between basal and columnar cells, thus make a fair contribution to the proliferation fraction of normal conducting-airway epithelium in humans despite their relatively modest cell numbers. A high proliferative activity of intermediate cells has been noted in previous studies done with [³H]thymidine pulse-labelling in laboratory animals (11, 19, 29). Without immunohistochemical techniques, identification of airway epithelial cells with the light microscope is difficult (29), resulting in large differences in the definition of the intermediate cell. As a consequence, the putative origin of the intermediate cell has led to considerable controversy. Evans and associates defined the intermediate cell as an undifferentiated columnar cell which, in their opinion, is part of the secretory-cell compartment and not a direct descendant of the basal cell (29). Donnelly and colleagues (13) employed another definition of the intermediate cell, calling it a nonbasal, noncolumnar cell. In their study, the fraction of [³H]thymidine-labeled intermediate cells peaked at 1 d after the peak in labeled basal cells, followed by a subsequent increase in [³H]thymidine labeling of both secretory cells and ciliated cells. This pulse-chase experiment led Donnelly and associates to conclude that basal cells are the progenitor cells of the airway epithelium, with the intermediate cell as their direct descendants. In addition to using the criteria of Donnelly and colleagues for the intermediate cell, we defined the parabasal cell by its immunoreactivity of keratins 5 and 14. As basal cells are the only other epithelial cell type in normal human conducting airways to express these keratins, it is quite plausible to argue that the combined parabasal and basal-cell populations belong to a distinct cell compartment. Although the design of the present study prohibits the delineation of pathways of cell differentiation, it is quite conceivable that an interrelation exists between basal cells and parabasal cells, with one being the progenitor cell of the other. Although basal cells are overrepresented in the proliferation fraction of the epithelium of airways with a diameter of 0.5 mm and more, parabasal cells contribute even more the proliferative-cell population relative to their modest cell numbers. This difference in the relative contribution of basal and parabasal cells to the proliferation fraction fits with the hypothesis that basal cells are relatively slowly cycling "stem" cells, giving rise to parabasal cells, which represent a comparatively rapidly cycling, transit amplifying cell population.

Quantification procedures for epithelial cell types can be very laborious. We examined two statistical approaches, using the nested ANOVA procedure for the calculation of SD. The assessment of two airways per patient and per airway category, with a number of approximately 400 epithelial cells per airway, adequately reflected the population mean at a 90% confidence level. The use of these two simple guidelines very significantly reduces the required number of epithelial cells needed for the reliable estimation of epithelial-type cell proportions.

In summary, we conducted a systematic study of the distribution and proliferation of basal and parabasal cells in normal human conducting-airway epithelium. In the largest conducting airways (diameter  4 mm), the percentages of basal and parabasal cells were 31% and 7%, respectively; the contribution to the proliferation compartment was 51% for basal and 33% for parabasal cells. In the smallest airways (diameter < 0.5 mm), 6% of epithelial cells were basal cells, with a 30% contribution to the proliferation compartment, whereas parabasal cells were absent. The high fraction of basal and parabasal cells contributing to the proliferation compartment of normal human conducting-airway epithelium supports the theory that cells at or near the basement membrane are likely to be progenitor cells.