Interleukin-1 is the initiator of Fallopian tube destruction during Chlamydia trachomatis infection

Authors


E-mail malenel@medmicro.au.dk; Tel. (+45) 89421746; Fax (+ 45) 86196128.

Summary

Chlamydia trachomatis infection is associated with severe Fallopian tube tissue damage leading to tubal infertility and ectopic pregnancy. To explore the molecular mechanisms behind infection an ex vivo model was established from human Fallopian tubes and examined by scanning electron microscopy and immunohistochemistry. Extensive tissue destruction affecting especially ciliated cells was observed in C. trachomatis infected human Fallopian tube organ culture. Interleukin-1 (IL-1) produced by epithelial cells was detected after infection. Addition of IL-1 receptor antagonist (IL-1RA) completely eliminated tissue destruction induced by C. trachomatis. The anti-inflammatory cytokine IL-10 reduced the damaging effect of C. trachomatis infection, however, to a lesser extent than IL-1RA. Furthermore, IL-1 was found to induce IL-8, a neutrophil attractant, using a signal transduction pathway involving p38 MAP kinase. Consequently, IL-1 has the potential to generate a cellular infiltrate at the site of infection in vivo. Blocking the IL-1 receptors by IL-1RA eliminated tissue destruction and cytokine production. Hence, these studies show the importance of IL-1 in initiating the tissue destruction observed in the Fallopian tube following C. trachomatis infection. Because leukocytes are absent in the ex vivo model, this study strongly indicates that IL-1 is the initial proinflammatory cytokine activated by C. trachomatis infection.

Introduction

Chlamydia trachomatis are obligate intracellular bacteria with a unique biphasic developmental cycle. C. trachomatis is the most common cause of sexually transmitted bacterial diseases in the Western World (Gerbase et al., 1998). In females, infection with the pathogen occurs in the epithelial cells of the endocervix. It ascends however, in some cases to the upper genital tract leading to salpingitis (Schachter, 1999). Approximately 70–80% of genital C. trachomatis infections are asymptomatic (Rahm et al., 1988). As a result, patients may not be treated for the infection but still develop severe sequelae such as tubal factor infertility (TFI) or ectopic pregnancy (Jones et al., 1982; Kosseim and Brunham, 1986).

The mechanisms leading to Fallopian tube pathology following infection remain to be fully determined. During early infection, a polymorphonuclear inflammatory response is produced in the superficial part of the infected epithelium, followed by subepithelial infiltration of cells of the immune system, e.g. lymphocytes, plasma cells and monocytes. Despite low infectivity, indicated by few chlamydial inclusions, the immune response against C. trachomatis is profound (Schachter, 1999). The intense human inflammatory response following infection is alleged to account for tissue destruction, fibrosis and scarring, ultimately leading to the severe sequelae observed after C. trachomatis infection.

Recently, focus has been directed towards cytokines to explain the severe tissue destruction observed following infection. Previous studies by Kinnunen and colleagues have shown an association between interleukin-10 (IL-10) polymorphism and TFI (Kinnunen et al., 2002). IL-10 is a pleiotropic modulator of the human immune response secreted by a wide variety of cells. It acts by suppressing the proinflammatory cytokine production as well as the antigen-presenting capacity of antigen presenting cells (de Waal Malefyt et al., 1991). Consequently, IL-10 is thought to function as a substantial suppressor of cellular immunity.

Chlamydia trachomatis has several factors (e.g. LPS, chlamydial HSP60 and CpG-DNA) possibly involved in the production of a number of cytokines through stimulation of Toll-like receptors (TLRs) (Ahmad-Nejad et al., 2002; Bulut et al., 2002; Erridge et al., 2004).

Interleukin-1 is a multifunctional interleukin produced by epithelial cells (reviewed by Dinarello, 1996) in response to various stimuli, e.g. bacteria. The IL-1 family comprises two agonists, IL-1α and IL-1β and one antagonist, IL-1 receptor antagonist (IL-1RA). IL-1α exists in a cell-associated and a secreted form, whereas IL-1β exists only in the latter and is thought to be involved in systemic inflammation. IL-1RA binds specifically to the IL-1 receptors in a competitive manner and hereby acts as a regulator of the IL-1 response; consequently, the effect of IL-1 can be investigated by blocking IL-1 receptors using recombinant IL-1RA. Binding of IL-1 to the receptor initiates a pathway in the human cell potentially leading to activation of several mitogen activated protein (MAP) kinases, e.g. Jun N-terminal kinase (JNK), p38 and extracellular signal-regulated kinase (ERK). Of these p38 is considered an important kinase involved in the transcription of several cytokines and adhesion molecules and is activated by signalling through a number of receptors, e.g. IL-1 receptors, TLRs and tumour necrosis factor (TNF) receptors (Saklatvala, 2004).

Interleukin-1 is involved in host inflammatory responses and in addition to TNF-α it is a potent inducer of IL-8. Even though IL-1 is not chemotactic towards neutrophils, the interleukin is capable of inducing other cytokines as well as adhesion molecules through a cascade effect, leading to an influx of immunological cells. Of the induced cytokines, IL-8 is considered the main neutrophil attractant and is produced by a diversity of cell types, e.g. epithelial cells, endothelial cells and monocytes. Consequently, the local response against the pathogen has the ability to attract cells of the immune system.

In the present study, a human Fallopian tube organ culture (FTOC) was applied as model allowing for investigation of the local immune response to C. trachomatis with regard to tissue destruction and cytokine response. Presence of interleukins following C. trachomatis infection was detected using immunohistochemistry (IHC) and morphological changes in the tubal epithelium in response to infection were investigated by scanning electron microscopy (SEM).

Results

Fallopian tubes from four women undergoing hysterectomy for benign gynaecological diseases were obtained and an FTOC was established to monitor the pathological effect of C. trachomatis infection (Baczynska et al., 2007). Results obtained from all patients were similar.

Fallopian tube tissue obtained from each patient was incubated in media supplemented with antibiotics for 24 h or for 5 days and examined by SEM (Baczynska et al., 2007). Control samples from all patients revealed that the Fallopian tube tissues appeared healthy and undamaged after 24 h as wells as after 5 days of incubation. The mucosal surface of the controls was intact with many ciliated cells in addition to secretory cells covered with microvilli. The cilia were morphologically normal, e.g. they did not adhere to each other nor were they swollen (Fig. 1A). Furthermore, IHC staining showed that the uninfected control samples were negative for C. trachomatis, IL-1α and IL-8 (Fig. 1B–D).

Figure 1.

Uninfected control sample examined by SEM and IHC.
A. The mucosal surface of the Fallopian tube appears healthy and undisrupted following 5 days of incubation. Many ciliated cells in addition to secretory cell covered with microvilli can be observed. The cilia are long and morphologically normal.
B. No C. trachomatis infected cells can be observed.
C and D. The cells are negative for IL-1α and IL-8 respectively. Black bar represents 20 μm.

Effect of C. trachomatis infection on tissue morphology and cytokine response in the FTOC

To investigate the morphological consequence of C. trachomatis infection, the FTOCs were infected with C. trachomatis serovar D and observed by SEM. Infection with the pathogen resulted in severe tissue damage affecting especially ciliated cells and to a lesser extent secretory cells (Fig. 2A). Many of the ciliated cells appeared dead and were dissociated from the epithelial layer. A number of secretory cells were ruptured. To determine whether lysis of infected cells was the direct cause of tissue destruction, C. trachomatis infected cells were visualized in the FTOC using IHC staining of paraffin embedded samples. C. trachomatis inclusions were localized in the epithelial layer of the Fallopian tube (Fig. 2B). Staining of multiple sections were necessary to identify C. trachomatis infected cells. Hence, only in few cells C. trachomatis inclusions were seen in IHC, disproportional to the severe tissue destruction observed by SEM, indicating that tissue morphology following infection is not entirely due to a mechanical destruction by host cell lysis following infection. To further investigate the effect of C. trachomatis infection, induced IL-1α and IL-8 were visualized by IHC. IL-1α staining was located on the surface of the tissue, revealing that the interleukin was induced by C. trachomatis and present in a membrane bound form (Fig. 2C). IL-8-positive cells were located in submucosal mononuclear cells (Fig. 2D).

Figure 2.

C. trachomatis infected tissue examined by SEM and IHC.
A. A severe destruction of especially the ciliated cells can be observed. Many of the ciliated cells are disassociated from the epithelial layer and some of the secretory cells are ruptured.
B. A few C. trachomatis infected cells can be recognized in the epithelial layer (indicated by arrows).
C. IL-1α can be seen bound to the epithelial layer.
D. IL-8 staining is located in mononuclear cells in the submucosa (indicated by arrow). Black bar represents 20 μm.

Interleukin-1 receptor antagonist eliminates the tissue destruction observed after C. trachomatis infection

To establish whether the observed degenerative effect of C. trachomatis infection on the tissue was a result of epithelial IL-1 production in response to infection, IL-1RA was added to the FTOC 2 h prior to infection and kept in the medium for the duration of the experiments. Five days post infection the tissue was examined using SEM. The addition of IL-1RA eliminated entirely the destructive effect of C. trachomatis (Fig. 3A). Despite infection with C. trachomatis the appearance of the tissue was equivalent to the uninfected controls; no deciliation was observed and the cilia in the sample appeared healthy. Additionally, the secretory cells were intact and covered with microvilli. IHC was applied to determine the effect of IL-1RA on the secreted cytokine pattern. A similar level of C. trachomatis infected cells were detected after addition of IL-1RA, indicating that the lack of tissue destruction observed in SEM is not due to unsuccessful infection of the tubes (Fig. 3B). In contrast to the results obtained from the C. trachomatis infected samples, IL-1α staining was weak and consequently, the blocking with IL-1RA was successful (Fig. 3C). There were no IL-8-positive cells present. Thus, IL-1 production was essential for IL-8 secretion in theC. trachomatis infected FTOC (Fig. 3D).

Figure 3.

Effect of IL-1RA on C. trachomatis infected tissue examined by SEM and IHC.
A. IL-1RA was added to the samples 2 h prior to infection. Despite presence of microorganisms, the epithelial layer resembles that of the uninfected control sample. The ciliated- and secretory cells appear morphologically normal showing that addition of IL-1RA has eliminated the degenerative effect of C. trachomatis entirely.
B. C. trachomatis infected cells can be detected.
C. Modest staining of IL-1α.
D. IL-1RA has reduced IL-8 staining completely. Black bar represents 20 μm.

Determination of the direct effect of IL-1α on the morphology of uninfected tissue and on the cytokine response

It was investigated whether addition of IL-1α to the FTOC was capable of promoting tissue damage since addition of IL-1RA reduced tissue damage by C. trachomatis completely. Therefore, recombinant IL-1α was added to the medium of an uninfected sample for the duration of the experiment. This resulted in a severe destruction of especially the ciliated cells (Fig. 4A). Moreover, a large number of cells were dissociated from the epithelial layer resembling the destruction caused by C. trachomatis. As expected all samples were negative for C. trachomatis (Fig. 4B) and positive for IL-1α staining (Fig. 4C) using IHC. Additionally, the experiment revealed that IL-1α itself is capable of inducing IL-8 (Fig. 4D).

Figure 4.

Effect of IL-1 on uninfected tissue examined by SEM and IHC.
A. The morphology of the tissue resembles a C. trachomatis infected sample. The cytokine has a pronounced cytotoxic effect on especially the ciliated cells.
B. No C. trachomatis infected cells can be observed.
C. IL-1α can be detected bound to the epithelial layer.
D. Many IL-8-positive cells can be recognized. Black bar represents 20 μm.

Investigation of the used signal transduction pathway following C. trachomatis infection

Interleukin-1 binds to receptors on the host cell and initiates hereby a signal transduction pathway in the human cell involving MAP-kinases, e.g. p38, JNK and ERK. p38 is involved in the secretion of several cytokines. To determine whether p38 activation is involved in the observed tissue damage in C. trachomatis infected FTOCs, a p38 inhibitor was added to the samples 2 h prior to infection and kept in the medium for the duration of the experiment. When subjecting the samples to SEM, a moderate reduction of tissue destruction and number of dead cells were observed (Fig. 5A). These results indicate that p38 is involved in the destruction, but that it may not be the only signal transduction pathway activated by C. trachomatis. C. trachomatis infected cells could be detected in the epithelial layer (Fig. 5B). Inhibition of the p38 kinase resulted in a reduction of the number of IL-8-positive cells (Fig. 5D). The amount of IL-1α was reduced (Fig. 5C), but weak staining could still be observed, supporting that p38 plays a role in the signal transduction during the local chlamydial response, it is not, however, the only kinase involved.

Figure 5.

Effect of p38 inhibitor on C. trachomatis infected tissue examined by SEM and IHC.
A. p38 inhibitor was added to the samples 2 h prior to infection. A few dead cells can be observed but in general p38 inhibitor reduces tissue destruction following C. trachomatis infection. This result indicates that p38 plays a role in signal transduction following infection with the bacteria.
B. C. trachomatis-positive cells can be detected.
C. Some IL-1α staining can be detected.
D. No IL-8-positive cells can be seen after addition of p38 inhibitor. Black bar represents 20 μm.

The effect of IL-10 on tissue morphology and cytokine response following C. trachomatis infection

Because the immune response appears to be of major importance for the tissue destruction observed after a C. trachomatis infection, we wanted to investigate how a modulation of the immune response would affect the outcome of C. trachomatis infection. The pleiotropic modulator, IL-10 is suppressing cytokine secretion (de Waal Malefyt et al., 1991) and IL-10 polymorphism has previously been associated with development of TFI following C. trachomatis infection (Kinnunen et al., 2002). For that reason IL-10 was added to the tissue samples 2 h prior to infection and kept in the medium for the duration of the experiments. Five days post infection a reduction on tissue damage could be seen, although not as pronounced as after addition of IL-1RA (Fig. 6A). A reduction in the number of destroyed cells in comparison to the C. trachomatis infected samples could be detected. Although IL-10 clearly reduced tissue destruction following C. trachomatis infection, the tissue clearly showed damaged cells. The samples were therefore analysed using IHC. Again C. trachomatis infected cells was detected (Fig. 6B). IL-1α staining was negative (Fig. 6C) and only very few IL-8-positive cells were present (Fig. 6D).

Figure 6.

Effect of IL-10 on C. trachomatis infected tissue examined by SEM and IHC.
A. IL-10 addition resulted in a reduction of tissue destruction, although not as pronounced as after addition of IL-1RA. Only few dead cells could be observed.
B. C. trachomatis-positive cells can be detected.
C. Samples are negative for IL-1α.
D. No IL-8 staining can be detected.

UV-inactivated C. trachomatis has no effect on tissue morphology

To determine whether activating TLRs by chlamydial antigens like LPS could promote an epithelial interleukin response and thereby induce tissue destruction, UV-inactivated C. trachomatis was added to the tissue samples. Addition of inactivated C. trachomatis did not however, have any effect on the tissue morphology because the tissue appeared healthy and comparable with the uninfected control (Fig. 7). Thus, presence of C. trachomatis antigens in the absence of infection appears to be inadequate for inducing tissue destruction in this in vitro FTOC.

Figure 7.

Effect of UV-inactivated C. trachomatis on Fallopian tube tissue morphology. The epithelial layer resembles the uninfected control sample indicating that inactivated C. trachomatis is not sufficient to promote tissue damage.

Discussion

The present investigation demonstrated that in the absence of leukocytes C. trachomatis is still capable of promoting severe tissue damage to the Fallopian tube epithelium by inducing extensive destruction of especially the ciliated cells. The damage is disproportionately extensive compared with the relatively low number of cells infected, indicating that the severe tissue damage is not due to a mechanical destruction of the cells caused by host cell lysis following C. trachomatis infection. Furthermore, addition of IL-1RA to the FTOC was sufficient to inhibit the damaging effect of C. trachomatis on the epithelium. These results show that IL-1 production in response to infection has a toxic effect especially on the ciliated cells present in the human Fallopian tube, because IL-1RA can fully block the destructive effects of C. trachomatis infection.

Functional cilia are vital for successful pregnancy. In rats it has been shown that the cilia are capable of moving the ovum to the site of fertilization independent of muscular activity (Halbert et al., 1989). Additionally, women with immotile cilia or partially immotile cilia are subfertile (McComb et al., 1986; Halbert et al., 1997; Afzelius, 2004) clearly indicating the importance of functionally healthy cilia for successful pregnancy.

In the present study we showed that addition of IL-1 to FTOC resulted in destruction of the ciliated cells supporting our hypothesis that IL-1 induction could lead to scarring of the Fallopian tubes and TFI. The mechanism by which IL-1 causes damage to the cells of the Fallopian tube is still unsolved, however, IL-1 primes and triggers superoxide production and thus has the potential to induce tissue damage (reviewed by Brigelius-Flohe et al., 2004).

Because our studies were carried out in an ex vivo model, and therefore in the absence of leukocytes, it points out that the local response in the Fallopian tube, by IL-1, is sufficient to cause tissue destruction without leukocyte infiltration. Invasion of leukocytes is not necessary to initiate tissue destruction. This explains why many women do not develop symptoms of salpingitis yet still develop infertility following C. trachomatis infection.

During infection with C. trachomatis many cytokines are likely to be released. IL-1RA completely blocked the induction of the secondary cytokine, IL-8, as well as destruction of ciliated cells. This strongly supports the finding that in this ex vivo model IL-1 is the key initiator of tissue destruction and inflammation during C. trachomatis infection. The consequence of this may be the generation of a cellular infiltrate at the site of infection in vivo, possibly due to the induction of IL-8.

Interleukin-10 has profound anti-inflammatory abilities. Among these are inhibition of LPS-induced activation of MAP kinase p38 and reduction of IL-1 production and thereby reduced induction of other cytokines by IL-1 (de Waal Malefyt et al., 1991; Gesser et al., 1997; Rajasingh et al., 2006). Additionally, the IL-10 gene is highly polymorphic, and genetic variation among the population results in differential IL-10 production (Maurer et al., 2000; Reuss et al., 2002). Kinnunen et al. (2002) have recently shown an association between IL-10 promoter polymorphism (1082 AA genotype) and TFI. In addition, the 1082 AA genotype has been found to be associated with reduced IL-10 secretion (Öhman et al., 2006), supporting the protective role of IL-10 with regard to tissue destruction. Based on the present studies, IL-10 has been linked to the regulation of the local immune response to C. trachomatis, decreasing the tissue damage significantly and inhibiting IL-8 production. This adds to the possibility that some women will generate a weakened inflammatory response and further helps to explain why a substantial number of women are asymptomatic during an infection.

The p38 inhibitor did not reduce the tissue destruction completely. Thus, p38 is not the only kinase used by IL-1 in the induction of cytokines, suggesting that divergent pathways are triggered by C. trachomatis infection. This correlates with previous findings in monocytes where addition of p38 inhibitor merely lead to a decrease in IL-1α production in response to LPS and not a complete blockage (Rabehi et al., 2001).

We have further demonstrated that addition of UV-inactivated C. trachomatis had no effect on the morphology of Fallopian tube epithelium. This result indicates that an active C. trachomatis infection with protein synthesis is necessary for the induction of tissue damage in vitro. Nevertheless, it is still possible that C. trachomatis antigens may induce responses from innate immune cells and thus tissue damage at the site of infection in vivo.

Taken together, this study shows for the first time that IL-1 produced in response to C. trachomatis infection had a direct destructive effect on Fallopian tube epithelium affecting especially ciliated cells. These results are supported by the finding that IL-1 production induces pathways in the cells leading to production of the neutrophil attractant, IL-8. Additionally, blocking of this pathway at different levels minimizes or completely eliminates the destructive effect of C. trachomatis infection.

Experimental procedures

Collection of tissue samples

The FTOC model was essentially performed as previously described (Baczynska et al., 2007). Briefly, tubal specimens from four non-pregnant, premenopausal women undergoing hysterectomy due to benign gynaecologic diseases were harvested. The surgery was planned in advance and performed after the menstrual period. Consequently, the epithelial layer was in the oestrogenic phase. Only anatomically normal Fallopian tubes were applied in the experiments. The project was approved by the local ethical committee (journal no: VF20050074). Immediately after surgery the Fallopian tubes were placed in Dulbecco Modified Eagle Medium (DMEM) mixed 1:1 with Ham's F12 medium (F12) (Invitrogen, Carlsbad, California, USA) enriched with 10% heat inactivated, sterile filtered fetal bovine serum (FBS), 10 μg ml−1 of gentamicin (Schering-Plough; Kenilworth; USA) and 0.25 μg ml−1 of amphotericin (Fungizone ®, Invitrogen).

The Fallopian tube samples were examined using a Leica Stereozoom (Leica Microsystems, Heerbrug, Switzerland) and cut into approximately 9 mm2 sections before incubation in DMEM : F12 medium with gentamicin and amphotericin overnight in sterile Petri dishes (NUNC, Roskilde, Denmark). All incubation was performed in a CO2 incubator at 35°C. After 24 h of incubation, one sample was fixed and prepared for SEM (Control Day 1).

Infection of tubal sections

The remaining tissue samples were divided into seven groups: (i) uninfected controls, (ii) C. trachomatis infected, (iii) C. trachomatis infected + IL-1RA, (iv) uninfected + IL-1, (v) C. trachomatis infected + p38 inhibitor, (vi) C. trachomatis infected + IL-10 and (vii) UV-inactivated C. trachomatis. The medium applied for these experiments was DMEM : F12 with gentamicin (8 μg ml−1).

The C. trachomatis infection of the samples were performed in 24-well plates (TBB, Trasadingen, Switzerland) and the FTOCs were infected with 4 × 104 IFU ml−1 of C. trachomatis serovar D (UW-3/CX) (corresponding to multiplicity of infection, moi = 0.8 in a monolayer of Hela cells) suspended in DMEM : F12 containing gentamicin and centrifuged for 15 min at 1100× g in a Beckmann GS-6R before incubation for 24 h.

To investigate the effect of IL-1, 10 μg ml−1 of IL-1RA (Anakinra, Kineret®, Amgen, Thousand Oaks, CA, USA) was added to the samples 2 h prior to infection. Additionally, we applied an uninfected sample that was placed in medium supplemented with 10 ng ml−1 of IL-1α (Sawai et al., 2005). Furthermore, the effect of IL-10 as well as p38 inhibitor was investigated and therefore IL-10 (100 ng ml−1) or p38 inhibitor, 4-(4-Fluorophenyl)-2-(4-methylsulfinylphenyl)-5-(4-pyridyl)1H-imidazole (SB203580, Calbiochem, La Jolla, CA, USA) 10 μg ml−1 was added to the samples 2 h prior to infection. To determine if inactivated C. trachomatis was able to promote tissue damage of the samples, an aliquot of C. trachomatis was inactivated using an UV-lamp for 20 min (model UVS-54, Mineralight, San Gabriel, CA, USA) and applied in the same concentration as live bacteria.

After incubation for 24 h the samples were relocated to Petri dishes (NUNC). The medium was changed every day and the applied cytokines were added in the previously described concentrations.

The samples for immunohistochemical staining were fixed on day 3 and for SEM on day 5.

Sample preparation for scanning electron microscopy

Tissue samples were prepared for SEM as previously described (Baczynska et al., 2007). Briefly, after completed incubation the tissue samples were washed gently in phosphate buffered saline (PBS) (pH 7.4) and 0.1 M sodium cacodylate buffer (SCB) (pH 7.2) before fixing in 2% glutaraldehyde (Sigma, St. Louis, USA) for 24 h at 5°C followed by a rinse in SCB. After this the samples were additionally post-fixed in 2% osmium tetroxide (OsO4) (Merck, Darmstadt, Germany) in SCB and washed in PBS before dehydration in ethanol-gradient. Finally specimens were subjected to critical point drying using CO2, placed on aluminium SEM stubs and coated with gold by a Pirani 10 Coater sputter (BOC Edwards, Crawley; UK). The samples were observed using SEM MaXim (CamScan, Waterbeach; UK) at 20–30 kV using secondary electron detector.

Antibodies for detection of IL-1, IL-8 and C. trachomatis

Monoclonal mouse anti-human antibodies against IL-1α (no 1190, Dainippon Pharmaceuticals, Japan and BAF200, R&D Systems, Minneapolis, USA) and against IL-8 (WS4, Dr Kouji Matsushima) were applied in concentrations of 1.2 μg ml−1, 10 μg ml−1 and 5 μg ml−1 respectively. In order to detect C. trachomatis an antibody against the C. trachomatis major outer-membrane protein (MOMP) was used. The monoclonal mouse anti-C. trachomatis MOMP (32.3) antibody was generated as described (Birkelund et al., 1988) and applied in a titre of 1:10. The antibodies were diluted in PBS supplemented with 10% normal goat serum (DakoCytomation, Glostrup, Denmark).

Fixing of tissue for IHC

The samples were washed once in PBS and fixed overnight in 4% formaldehyde at 5°C. Subsequently the samples were embedded in paraffin, cut into 5 μm sections (Microtome, HM360, MICROM, Walldorf, Germany), placed on Superfrost®Plus slides (Menzel-Glaser, Braunschweig, Germany) and dried at 37°C overnight.

Preparation of slides and immunohistochemical staining

The slides were deparaffinized twice in xylene for 5 min before rehydration of the tissues in ethanol gradient (99%-96%-70% for 2 × 5 min finishing with PBS for 5 min). The antigen was unmasked by boiling the slides for 20 min in TE-buffer (pH 9.0). Afterwards the tissue was blocked for endogen avidin and biotin (X0590, DakoCytomation) and for non-specific background staining using 10% goat serum (DakoCytomation). The slides were incubated for 30 min with the primary mouse anti-human antibody as mentioned above, followed by incubation for 30 min with the biotin–conjugated goat anti-mouse secondary antibody diluted 1:800 (115-065-146, Jackson ImmunoResearch, West Grove, PA, USA). Bound antibody was detected using StrepAB-Complex/AP (DakoCytomation) in accordance with the manufacturer's descriptions with an incubation time of 30 min. The samples were stained using Fuchin Substrate Chromogen System (DakoCytomation) for 6.5 min in accordance with the manufacturer's descriptions. Following each incubation step the slides were washed in PBS. The tissue was counterstained using Mayer's haematoxylin (Bie and Berntsen, Aarhus, Denmark) for 5 min and rinsed in tap-water.

The slides were fixed in Aquatex (Merck). Images of representative fields were acquired using an Olympus Provis AX70 microscope (Olympus, Lake Success, NY) equipped with a Leica DFC 320 Digital Camera system (Leica Camera AG, Solms Germany).

Acknowledgements

We are grateful to engineer Jacques Chevallier and to associate professor, PhD, Peter Funch for excellent technical assistance with SEM and critical point drying respectively. We wish to thank the medical personnel and surgeons at Horsens Hospital. We are also grateful to Lisbet W. Pedersen for linguistic revision of this article, and to MSc student Helene Jensen and PhD student Mette Drasbek for skilled laboratory assistance. This study was financially supported by ‘The Danish Medical Research Council’ (Grant no. 22-03-0245 and 271-05-0488), ‘Aarhus Universitets Forskningsfond’ and EU Grant no. EU FP6: NoE EPG LSHB-CT-2005-512061.

Ancillary