Introduction

Abstract Introduction References

Airway inflammation is an important feature of asthma and nasal polyps. Airway inflammation in patients with asthma and nasal polyps is characterized by the increased presence of several cell types, including mast cells, lymphocytes, and, in particular, eosinophils (1). Mast cells and eosinophils found in the nasal mucosa of patients with allergic rhinitis can vary in number in nasal epithelial scrapings in relation to disease severity, but in polyp scrapings a high number of metachromatic cells and eosinophils is invariably found, regardless of allergic or nonallergic status.

Recently, several findings have highlighted the role of inflammatory-cell progenitors in contributing to airway inflammation in asthma. One of these is the presence of circulating colony forming units (cfus) for eosinophils/ basophils in increased numbers in the blood of allergic individuals (2, 3); these numbers fluctuate in relation to exacerbation and resolution of clinical asthma (2), and are increased after allergen inhalation in association with the development of airway hyperresponsiveness (4). Increases in peripheral blood and bone marrow inflammatory-cell progenitors in response to allergen inhalation challenge presumably contribute to the accumulation of mature inflammatory cells in the airways of asthmatic subjects, through enhanced production and release of these progenitors by the bone marrow (5).

CD34 is a cell surface, sialomucin-like glycoprotein that is expressed on hemopoietic progenitor cells, normal vascular endothelium, and fibroblasts. CD34 is expressed most strongly on primitive hemopoietic cells, and is progressively lost as cells differentiate (8); small numbers of these primitive, CD34⁺ hemopoietic progenitors are found in normal human peripheral blood (9). Eosinophil/basophil progenitors circulate in human peripheral blood as a subpopulation of progenitors, identifiable as cells that form tight "eosinophil-type" colonies in methylcellulose, or as CD34⁺ cells responsive to interleukin (IL)-5 or which bear IL-5 receptors (6, 7, 10). The development of a sustained tissue eosinophilia may thus be hypothesized to occur via an increase in the production and delivery of these progenitor cells that circulate in the peripheral blood, especially of atopic individuals (2, 11); eosinophil/basophil progenitors identified by colony assays have also been isolated from nasal polyp tissues (12). Hemopoietic colony-stimulating activity for eosinophil/basophil cfus can be derived from epithelial cells or fibroblasts of nasal polyps (11, 12); these structural cells of the polyp can produce a variety of hemopoietic cytokines, including stem cell factor (13), which can stimulate CD34⁺ cells to evolve to mast cells (13, 14). Indeed, numbers of mast cell progenitors also fluctuate in response to viral (15) or parasitic (16) stimuli associated with mast cell hyperplasia in rodents.

We hypothesized that CD34⁺ cell progenitors could be localized within the inflamed airway, and specifically within nasal polyps. To examine this, we performed immunostaining for CD34 in tissue from nasal polyps. Because glucocorticoids are known to diminish the number of mature eosinophils at sites of allergic disease in the mucosa, we also investigated the effect of the topical corticosteroid budesonide on CD34 immunoreactivity in nasal polyps, using a monoclonal antibody (Ab) directed against a major epitope of CD34.

	Materials and Methods

Subjects

Nasal polyps were obtained from 21 patients who underwent surgery for nasal obstruction. Nine patients (eight males and one female, aged 49.4 ± 21.1 yr [mean ± SD]) had not taken systemic or topical corticosteroids during the 1 mo before surgery. Twelve patients (10 males and two females, aged 48.3 ± 16.4 yr) had received the topical nasal corticosteroid budesonide at 200 to 400 µg/d during the preceding month. There were no differences between the two groups in severity of presenting symptoms prior to therapy. All patients were referred to the Ear, Nose, Throat Clinic at McMaster Hospital for polypectomy, which was done under topical or general anesthesia. No patients had taken antibiotics during the month before surgery. The studies done on the patients conformed to guidelines for human experimentation and were approved by the Ethics Committee of Chedoke-McMaster and Hamilton Civic Hospitals; when appropriate, subjects provided written informed consent.

In addition, we obtained nasal polyp tissue by biopsy from one patient both before and after treatment with corticosteroid. The patient's nasal passages were decongested with 0.5% phenylephrine (Neosynephrine), and were then sprayed with 4% xylocaine solution to anesthetize the nasal mucosa. Percutaneous 2-3-mm biopsy sections of nasal polyps were obtained with 5-mm Thru-Cut (Smith & Nephew, Memphis, TN) biopsy forceps. This procedure was done before and 1 wk after treatment with tapering doses (24 mg initially, decreasing by 8 mg daily to discontinuation) of methylprednisolone given orally. This study was approved by the Human Subjects Committee of the Washington University School of Medicine.

Immunohistochemical Localization of CD34 in Nasal Polyps

To examine cellular localization of CD34 within nasal polyps in relation to therapy, nasal polyp tissue sections of 5-mm thickness were fixed in formalin for 3 h. The tissues were embedded in paraffin, and serial sections (10 µm) were cut and placed onto slides coated with 3-aminopropyltrietoxysilane (APTEX) (Sigma Chemical Co., St. Louis, MO). Sections were stained through a modification of the avidin-biotin complex (ABC) method (Histostatin-SP kit; Zymed Laboratories Inc., San Francisco, CA), as described previously (17). Sections stained for CD34 were trypsinized before staining for 30 min. Negative controls without primary Ab were processed in parallel. Sections were also stained for von Willebrand factor (VWF), a marker widely used for identifying endothelium in routinely processed tissue (18). To demonstrate VWF on vascular endothelia, a purified mouse monoclonal antihuman VWF Ab (PharMingen, Mississauga, ON, Canada) was used at 1:63 dilution. Mouse monoclonal antihuman CD34 Ab (QBEND/10; Novocastra Laboratories Ltd., Newcastle, UK), the isotype of which was immunoglobulin (Ig)G₁, was used at a dilution of 1:100 to detect CD34 protein in nasal polyps. Biotinylated goat antimouse Ab (Sigma) was used as a second Ab at a dilution of 1:200. Peroxidase-conjugated avidin was used at a dilution of 1:500. The QBEND/10 monoclonal Ab for CD34, the monoclonal antihuman VWF Ab, biotinylated goat antimouse IgG, streptavidin peroxidase, and the chromogenic substrate 3-amino-9-ethylcarbazole were sequentially applied to the sections. Mouse IgG₁ myeloma protein (Sigma) was used as a negative control. After immunostaining, the slides were counterstained with Mayer's hematoxylin (Sigma).

Paraffin-embedded nasal tissues, prepared from biopsy samples fixed in Carnoy's solution, were stained with chromotrope 2R (Sigma) for 10 min to enumerate eosinophils. The number of eosinophils was averaged from a total of 10 fields in lamina propria.

Quantification of Immunohistochemical Staining

All tissue samples were coded and sections were counted blindly with a conventional microscope. Quantification of staining in the paraffin-embedded nasal polyp samples was done in a region extending 0.5 mm in depth from the lower border of the basal membrane, using a calibrated eyepiece (×10) and an objective lens with a magnification of ×40 for each tissue (×400 magnification). The number of stained endothelia was determined in an area of 0.5 mm² of this region. In the cytocentrifuged isolated mononuclear cell (MNC) samples, percent positivity for CD34 staining was evaluated as a percentage given by the numbers of the positive cells divided by the numbers of MNC at ×400 magnification. The mean numbers of labeled cells were calculated from a total of 10 high power fields (hpf) in this area.

Cell Isolation and Methylcellulose Cultures

Nasal polyps were cut into pieces, which were incubated with stirring for two 60-min periods in 40 ml of McCoy's 5A medium (pH 7.2; 290 mOsm/kg) supplemented with 15% fetal calf serum (FCS), 1% penicillin-streptomycin, and 0.5 µM 2-mercaptoethanol (2-ME), containing 190 U/ ml of collagenase. The supernatants were filtered through a mesh stainless-steel sieve after a further 60-min incubation, and were pooled. The remaining tissue was incubated overnight at room temperature. The three harvested cell suspensions were centrifuged at 200 × g for 10 min, and were resuspended in supplemented McCoy's 5A medium. The layer of MNC was isolated from the cell suspensions by sedimentation on Percoll density gradients (1.077 g/ml) and was placed onto 100 × 200-mm tissue culture dishes (Falcon Plastics, Oxnard, CA) that had been coated with collagen (Collagen Biomaterials, Palo Alto, CA), and was incubated in supplemented McCoy's 5A medium for more than 8 h at 37°C in 5% CO₂. To confirm myeloid colony formation, nonadherent mononuclear cells (NAMNC) were cultured in methylcellulose as described (2, 3, 12). NAMNC were cultured for 9 d at 37°C in 5% CO₂ at 3 × 10⁵ cells/ml (1.5 × 10⁵ cells per well of 24-well tissue culture plates; Nunc, Roskilde, Denmark) in Iscove's modified Dulbecco's medium (GIBCO, Grand Island, NY) with 15% FCS, 1% penicillin-streptomycin, and 0.5 µM 2-ME in 0.9% methylcellulose, in the presence of 10 ng/ml of recombinant human granulocyte-macrophage colony-stimulating factor (rhGM-CSF). Cells in Day 7 granulocyte colonies of  40 cells were enumerated and assessed for morphologic properties through inverted light microscopy.

Isolated NAMNC were also spun to deposit them onto APTEX-coated slides, and these cytospin preparations were subjected to immunohistochemistry for CD34 expression.

Flow Cytometry and Gating Strategy

Samples of 2 × 10⁵ NAMNC from two nasal polyposis patients were simultaneously stained with anti-CD45 fluorescein isothiocyanate (FITC) (anti-HLE1) (Becton Dickinson, Mississauga, ON, Canada) and either anti-CD34 phycoerythrin (PE) (HPCA-2) (Becton-Dickinson) or isotype-matched negative control antibody for 30 min at 4°C. The cells were then washed with phosphate-buffered saline (PBS) plus 0.1% azide, fixed in PBS plus 1% paraformaldehyde, and refrigerated in the dark until ready for analysis.

To enumerate CD34⁺ progenitor cell numbers accurately in various samples, a multiparameter sequential gating strategy was used as described (7, 19). Briefly, the major population of lymphomononuclear cells was resolved on a dot plot of linear forward light scatter (FSC) versus linear side light scatter (SSC), and was gated into a region R1 to exclude contaminating events such as red blood cells, platelets, and cell debris (Figure 1A). CD45⁺ events were established on a dot plot of CD45 staining versus SSC, and were gated into a region designated R2 (Figure 1B), where they were analyzed for CD34 staining, with CD34⁺ cells being gated into a region designated R3 (Figure 1C). To determine the number of true CD34⁺ progenitor cells, events defined by R1, R2, and R3 were analyzed on a dot plot of CD45 staining versus SSC, where CD34⁺ blast cells clustered as defined into a region R4 (Figure 1D). These events fell into a discrete region, which represented lymphoblastoid cell characteristics (i.e., low to medium cell size and low granularity) (Figure 1E). Without changing any of the gates, analyses were done of the same sample stained with anti-CD45 FITC and PE-linked isotype control antibody (Figure 1F), or with anti-CD34 PE and FITC-linked isotype control antibody (data not shown).

View larger version (33K): [in this window] [in a new window]   Figure 1.   Multiparameter flow cytometric analysis for enumeration of CD34 progenitor cells in nasal polyp cell suspensions. Region R1 represents lymphomononuclear cells (A). Anti-CD45 FITC staining of cells in R1 is defined as R2 (B). CD34+ cells (R3) contained in R2 are scattered for CD34 staining versus SSC (C). The cells gated by R1, R2, and R3 are backscattered for CD45 staining versus SSC (R4) (D). Region R4 represents a lymphoblastic region in which all nasal polyp-derived CD34+ cells scatter (E). Analysis of the same sample stained with anti-CD45 FITC and PE-linked isotype control antibody was done with the same gating strategy (F).

Endothelial Cell Cultures and Flow Cytometry

Endothelial cells were isolated from human umbilical cord veins after collagenase digestion, and were grown to confluence in six-well plates (35 mm; Costar, Cambridge, MA) as previously described (18, 20, 21). First-passage cell cultures were incubated in either dimethyl sulfoxide (10⁷ M) or budesonide (10⁷ M) (Astra Draco, Sweden) medium for 24 h (37°C, 5% CO₂/95% air). Half of the cells from each group were treated with tumor necrosis factor (TNF)- (100 ng/ml) (R&D Systems, Minneapolis, MN) for the final 4 h of culture. Cell suspensions were generated nonenzymatically, and direct immunofluorescence was done by incubating the endothelial cell suspensions with appropriate concentrations of anti-CD34PE (Clone QBEND-10; Immunotech, Marseille, France) or an irrelevant monoclonal Ab control as previously described (19). Cells were analyzed with an EPICS Profile flow cytometer (Coulter Corporation, Hialeah, FL). Human umbilical vein endothelial cells (HUVECs) were recovered for flow cytometry by treatment with Versene 1:5,000 solution for 5 min at 37°C (Biofluids Inc., Rockville, MD). A total of 80 to 90% of cells was recovered with this method.

Statistics

For statistical analysis, the Mann-Whitney rank sum test was used to assess differences between the numbers of CD34-immunoreactive cells and vessels and between the numbers of VWF-immunoreactive vessels in steroid-treated and steroid-untreated nasal polyps. Statistical significance was taken at P < 0.05.

	Results

Immunohistochemical Findings: CD34 and VWF Quantitation

A summary of the immunohistochemical findings with QBEND 10 (anti-CD34 monoclonal antibody) and anti-VWF Ab for the steroid-treated and -untreated subject groups is given in Table 1. In general, substantial immunolocalization of CD34 was observed in some mononuclear cells and endothelial cells in all nasal polyps (Figure 2A). Endothelial cells showed moderate to strong cytoplasmic staining for CD34. CD34 reactivity was found diffusely over the entire cytoplasm of CD34⁺ mononuclear cells. These CD34⁺ mononuclear cells had an elliptical shape and a single, round, centrally located nucleus resembling immature (high nucleus-to-cytoplasm ratio) hemopoietic cells in morphology (Figures 2A and 3A). VWF⁺ endothelial cells showed moderate to strong diffuse cytoplasmic immunostaining for VWF (Figure 2B); however, there were no apparent VWF⁺ mononuclear cells in any nasal polyp tissues. Furthermore, the percent positivity of CD34 staining among the isolated MNC was 0.5 ± 0.1% (mean ± SD), which was consistent with data on the frequency of CD34⁺ cells in human peripheral and cord blood samples (19) (Figure 3B).

View larger version (101K): [in this window] [in a new window]   Figure 2.   Immunohistochemical localization of CD34 in nasal polyps. (A) Diffuse immunoreactive CD34 protein in the cytoplasm of mononuclear cells in the lamina propria (arrow). (B) Endothelial cells show immunostaining for VWF protein with antihuman VWF Ab. (C) No immunostaining is observed with isotype control Ab (mouse IgG1). Scale bar = 7 µm (A), 9 µm (B), 24 µm (C).

View larger version (82K): [in this window] [in a new window]   Figure 3.   CD34 immunoreactivity of isolated MNC in nasal polyps. (a) CD34+ MNC have morphologic characteristics of immature myeloid cells. (b) CD34 immunoreactivity in polyp MNC is observed at a frequency expected of hemopoietic progenitors in human blood samples. (c) Negative control experiments show no staining. Scale bar = 6 µm (a), 24 µm (b), 9 µm (c ).

The number of CD34-immunoreactive mononuclear cells was significantly higher in steroid-treated than in steroid-untreated nasal polyps (15.67 ± 2.08 cells/10 hpf versus 5.33 ± 1.36 cells/10 hpf, respectively, P = 0.002). The number of CD34⁺ endothelia was also significantly higher in steroid-treated nasal polyps than in steroid-untreated polyps (101.25 ± 6.24 vessels/0.5 mm² of lamina propria versus 57.22 ± 8.00 vessels/0.5 mm² of lamina propria, respectively, P = 0.0008). However, the number of VWF-immunoreactive vessels in steroid-treated and steroid- untreated nasal polyps showed no significant difference (62.33 ± 7.57 vessels/0.5 mm² of lamina propria versus 54.89 ± 10.78 vessels/0.5 mm² of lamina propria, respectively, P = 0.43). The intensity of CD34-immunoreactivity in nasal polyps of steroid-treated patients was not significantly different from those of steroid-untreated groups. Negative control experiments showed no staining with the respective isotype control antisera (Figures 2C and 3C). As a control for the effect of topical steroids on inflammatory cells in the polyps tested, we counted eosinophils, finding 8.3 ± 6.4 per hpf in untreated versus 2.6 ± 3.2 (mean ± SD) in steroid-treated polyps (P = 0.0001).

CD34 Immunostaining Before and After Systemic Corticosteroids

As shown in Figures 4A and 4B, there was more intense immunostaining for CD34, especially of mononuclear cells, after a course of systemic (oral) corticosteroid therapy. Quantitation revealed a doubling of CD34⁺ mononuclear cells after treatment (Table 2).

View larger version (165K): [in this window] [in a new window]   Figure 4.   CD34 staining in nasal polyps before and after treatment with systemic corticosteroids in one patient. (A) Immunolocalization of CD34 is observed on both MNC and endothelial cells in nasal polyp before corticosteroid treatment. (B) MNC show more intense staining for CD34 after corticosteroid treatment. Scale bar = 24 µm.

Nasal Polyp-Derived Colonies

Methylcellulose-based clonogenic assays were performed to confirm that nasal polyp MNC preparations in methylcellulose contained myeloid colony-forming cells (i.e., progenitors). Cultures of MNC isolated from nasal polyps yielded 5.8 ± 0.6 (mean ± SD) myeloid colonies per 10⁵ cells plated. A representative example is shown in Figure 5.

View larger version (128K): [in this window] [in a new window]   Figure 5.   Morphologic features of a nasal polyp-derived myeloid colony. Colonies were cultured from NAMNC in nasal polyp, in the presence of 10 ng/ml of rhGM-CSF.

FACS Analysis

True CD34⁺ progenitor cells were identified within the lymphomononuclear cell population (Figures 1A through 1E). They were defined as CD34⁺/CD45⁺ cells with low cell granularity (SSC) and a small to medium cell size (FSC) (Figure 1E): 0.26% (of 27,000) events represented true CD34⁺ progenitor cells. This percentage is equivalent to that reported for CD34⁺ cells in human peripheral blood samples (19). Percent staining of the background with the FITC-linked or PE-linked isotype control antibody in the identical gating regions to those shown in Figures 1A through 1E was 0.00% (data not shown) and 0.02% (Figure 1F), respectively.

Effects of Corticosteroids In Vitro on HUVEC Expression of CD34

To determine whether budesonide acts directly on the endothelium to induce CD34 expression, we cultured HUVECs (first passage) for 24 h with and without budesonide (10⁷ M), and performed flow cytometry with a PE-conjugated antibody to CD34 (n = 2; Figure 6). Half of the cells from each group were stimulated with TNF- (100 ng/ml) for the final 4 h of culture. Budesonide did not have any effect on the CD34 staining seen. TNF- stimulation resulted in little if any reduction in CD34 staining in both control- and budesonide-treated cultures.

View larger version (25K): [in this window] [in a new window]   Figure 6.   Effect of budesonide on CD34 expression by HUVEC in vitro. Although a slight increase in HUVEC CD34 staining was observed after in vitro budesonide treatment, the difference from control was not statistically significant. In addition, stimulation with TNF- (+) did not enhance CD34 expression in either control or budesonide-treated cultures. Results of two experiments, mean ± SD of replicate samples, are shown. Histogram was expressed by dividing the mean fluorescence intensity for anti-CD34 antibody by the mean fluorescence intensity for IgG isotype control.

	Discussion

In this study, we showed the presence of CD34-immunoreactive protein on MNC and vascular endothelial cells in the lamina propria of nasal polyps. In patients who had undergone 4 wk of topical corticosteroid treatment, the numbers of the CD34⁺ cells and vessels in nasal polyps were increased significantly over those in clinically similar patients with steroid-untreated polyps. In addition, in one patient whose polyp tissue was examined both before and after systemic (oral) corticosteroid therapy, CD34 immunostaining was more intense, with an increase in quantifiable CD34⁺ mononuclear cells. Because corticosteroid treatment may induce angiogenesis, we also performed immunostaining for VWF to compare the number of vessels in steroid-treated and steroid-untreated polyps: VWF reactivity was no different in steroid-treated than in steroid-untreated nasal polyps. Although CD34 immunoreactivity has been detected on osteoclasts, peripheral nerve sheath cells, and fibroblasts (18), CD34⁺ cells in polyps were different from these cells morphologically, resembling immature hemopoietic cells (22), as shown in Figures 2A and 3A. Furthermore, the presence of CD34⁺/CD45⁺ cells within the lymphomononuclear cell population isolated from nasal polyps and the derivation of myeloid colonies in methylcellulose from polyp NAMNC provide further evidence that polyp immunostainable CD34⁺ cells are true hemopoietic progenitor cells.

It is well known that the numbers of inflammatory cells such as eosinophils are increased in the airways in such chronic airway inflammatory conditions as bronchial asthma (23) and nasal polyps (24). More than one mechanism may contribute to persistent increases in airway inflammatory cells in these conditions. First, there is the possibility of enhanced bone marrow production and activation or prolonged survival of mature inflammatory cells, as has been demonstrated for tissue mast cells (25). Another possibility is that a process of increased infiltration and terminal differentiation of inflammatory-cell progenitors may occur from expanded progenitor populations in the blood. Indeed, increased numbers of inflammatory-cell (eosinophil/basophil) progenitors have been documented in the peripheral blood and bone marrow of atopic humans and those with asthma/airway hyperresponsiveness (2, 28), in the latter group in exacerbations of disease or in response to allergen inhalation (4). These progenitors, assayed as colony-forming cells, are present within nasal polyps (12), and may therefore serve as potential targets for hemopoietic differentiation factors which are also produced with nasal polyp tissues (11, 24). Thus, an increased population of circulating basophil/eosinophil progenitors, and the local presence of cytokines capable of stimulating their differentiation, especially in patients with atopy and nasal polyposis, provides circumstantial evidence that differentiation may occur locally within chronically inflamed tissues. Our finding of CD34⁺ mononuclear cells within nasal polyp tissue is consistent with this hypothesis. A complementary hypothesis is that steroids, although having no effect on cell trafficking, inhibit local (i.e., within-polyp) CD34⁺ cell differentiation; in this model, steroids act to block differentiation and prolong CD34 expression.

The expression of CD34 is not restricted to hemopoietic progenitor cells, but has also been detected by immunohistochemical analysis on vascular endothelium of normal and neoplastic tissues (29). It has also been shown that the relevant endothelial cells produce the same size protein (MW = 110 kD) and express identical CD34 messenger RNA (2.3 kb) to that expressed by hemopoietic cells, indicating that CD34 antibodies bind to the CD34 gene product on endothelial cells, and not to a cross-reactive epitope (30). Many antibodies have been developed to detect various epitopes of CD34 (22); among these, the anti-CD34 antibody QBEND 10 has been reported to label blood vessels in normal skin (35). This antibody has been applied to frozen and paraffin-embedded tissue sections for diagnosing vascular tumors, including Kaposi's sarcoma (33). Previously, studies of endothelial cells indicated that when freshly isolated, these cells are CD34⁺, but after a few passages they become CD34 (30). This finding suggests that the expression of CD34 protein by vascular endothelial cells is not constitutive but is instead regulated, possibly by cell contract, proliferation, or changes in the extracellular environment. Regulation of CD34 is also observed in hemopoiesis: the antigen is not expressed on hemopoietic cells as maturation progresses (22, 30).

Ultrastructural analysis of umbilical artery and capillary vessels has shown that CD34 molecules are concentrated on membrane processes that interdigitate between adjacent endothelial cells (30). This observation, together with the biochemical feature of the CD34 molecule of being a highly negatively charged sialomucinlike glycoprotein, has led to the hypothesis that CD34 may play an antagonistic or inhibitory role in vascular endothelial cell adhesive function (31). Expression of CD34 at luminal junctions in vascular endothelia may therefore serve to modulate leukocyte infiltration into tissues and/or to increase the stringency requirements of ligand-receptor interactions that facilitate adhesion (31, 32).

The adhesion of leukocytes to vascular endothelia prior to their tissue infiltration appears to be regulated by various adhesion molecules on vascular endothelia, the expression of which can be modulated by various cytokines. In contrast to adhesion molecules such as E-selectin, intercellular adhesion molecule-1, and vascular cell adhesion molecule-1, the expression of CD34 is downregulated by cytokines such as IL-1, interferon-, TNF- in vitro and in vivo (29, 34). These data are compatible with the view that CD34 may have a negative modulating effect on the adhesive functions of endothelia, and that downregulation of this highly negatively charged molecule might be an important event in cytokine-mediated leukocyte adhesion and inflammation. It is therefore possible that the increased expression of CD34 on vessels in our steroid-treated nasal polyps might have been a reflection of inhibition of inflammatory cell/progenitor differentiation within, or of inflammatory cell infiltration into, polyp tissues.

That budesonide had essentially no effect on CD34 expression in the HUVEC model may have several explanations. First and foremost, it may indicate that steroid effects in vivo are in fact indirect. For example, it is conceivable that steroids inhibit TNF- release by other cell types found in polyps, such as eosinophils, and because TNF- is known to downregulate endothelial CD34 expression, its inhibition may have a net enhancing effect on CD34 expression. It is also possible that HUVECs provide a relatively steroid-unresponsive endothelial model that may or may not reflect the in vivo situation (21, 36).

In CD34-deficient, ovalbumin (OVA)-sensitized mice, eosinophil accumulation in the lung after OVA challenge is reduced (37), but no abnormalities are seen with lymphocyte or neutrophil recruitment. A role in eosinophil adhesion for a still not fully characterized, CD34-cross-reactive, 90-kD protein that may bind L-selectin has been postulated from these observations. Although these data suggest that CD34 is important in eosinophil infiltration, the role of CD34 in this process may involve a complex, multistep mechanism.

In humans, local administration of glucocorticoids decreases the number of mast cells in the skin (38), synovium (39), and nose (40), as well as the number of activated eosinophils in the airways in asthma and cases of nasal polyps (38). Several investigators have already shown that corticosteroids inhibit the production of cytokines by activated T lymphocytes and other structural cells (38), as well as inhibiting stem cell factor-induced differentiation of mast cells from progenitors in fetal liver in vitro (41). Our data show that an increase of CD34⁺ mononuclear cells occurs in steroid-treated nasal polyps. A possible mechanism for this increase in CD34⁺ cells in steroid-treated nasal polyps may be decreased production of hemopoietic cytokines and subsequent inhibition of the differentiation of mature cells from progenitors residing in nasal polyp tissue.

In conclusion, we have shown that CD34⁺ mononuclear cells and CD34⁺ endothelial cells are present in nasal polyps. Topical corticosteroid treatment is associated with increased numbers of all CD34⁺ cells, both mononuclear and endothelial. These findings are consistent with the presence of inflammatory-cell progenitors in nasal polyps. Furthermore, they suggest that topical corticosteroids suppress leukocyte recruitment by upregulating CD34 expression on vascular endothelium in nasal polyps.