SEARCH

SEARCH BY CITATION

Keywords:

  • tight junction organization;
  • image analysis;
  • fluorescence microscopy;
  • quantification

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS AND DISCUSSION
  5. CONCLUSIONS
  6. LITERATURE CITED

The spatial organization of proteins in a cell population or in tissues is an important parameter to study the functionality of biological specimens. In this article, we have focused on tight junctions which form network-like features in immunofluorescence microscopy images. Usually, the organization or disorganization of tight junctions is noticed qualitatively. The aim of this article is to present a simple method to quantify the organization level of tight junction network using image analysis with a dedicated macro developed with Image J software. The method has been validated with simulated images displaying regular decrease of network organization. Then, the macro has been applied to immunofluorescence microscopy images of cells in culture and of tissue sections. © 2012 International Society for Advancement of Cytometry

Today, research in the field of animal cells mainly focuses on biochemical aspects such as gene expression and signal transduction. However, it is also of interest to develop methods to study the spatial organization of proteins at a single cell level or at a cell population level. Among the proteins implemented by cells, those involved in tight junctions, particularly in epithelial cells, play a key role in the maintenance of the protective functions of epithelia. Tight junctions, or zonula occludens, are the closely associated areas of two cells whose membranes join together forming a virtually impermeable barrier to fluid. In electron microscopy ultrathin-section, tight junctions are visualized as a zone in which adjacent plasma membranes are closely apposed, and circumscribe the cell like a belt (1). Just beneath the plasma membrane of tight junctions, PDZ domain-containing cytoplasmic proteins directly interact with tight junction-associated integral membrane proteins and form cytoplasmic plaques (2). Three structurally related proteins, zonula occludens-1 (ZO-1), ZO-2, and ZO-3, are included in these cytoplasmic plaques. ZO-1 was the first reported tight junction-associated molecule. Because tight junctions are very dynamic structures that assemble, grow, reorganize, and disassemble during physiological or pathological events, immunolocalization of ZO-1 protein is frequently used to assess their spatial distribution. In a given cell, this protein exhibits a continuous distribution along the contacts with neighboring cells, indicating the formation of a junctional belt.

The association of the belt coming from neighboring cells forms a typical network that can be easily visualized by conventional fluorescence microscopy. This network appears progressively during epithelial differentiation or can be disrupted in the presence of external agents that may alter the epithelial barrier. It is, therefore, of interest to determine quantitative parameters to characterize this network. Quantitative analysis of protein network in cells has been mainly applied to the study of the cytoskeleton dynamics (3–5). Computerized image analysis coupled to the determination of different parameters has also been suggested for studying and quantifying the combined process of tumor invasion and vascular network formed during angiogenesis (6). The aim of this work has been to develop a simple method to quantify the variations in tight junction protein spatial organization on immunofluorescent images recorded with a conventional microscope on cells in culture or on tissue sections.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS AND DISCUSSION
  5. CONCLUSIONS
  6. LITERATURE CITED

Preparation of Bacterial Supernatant

Bacterial supernatant was prepared by growing Pseudomonas aeruginosa (PAO1 strain) in Trypticase Soy Broth (TSB) medium to stationary phase at 37°C for 48 h under mild agitation. Supernatants of 5 × 109 colony-forming units (cfu/ml) were obtained by centrifugation at 3,000 rpm for 15 min at 4°C and filtration through a 0.2-μm filter (Pall Gelman Science, Ann Arbor, MI). The bacterial supernatant was used after dilution (1:10) in RPMI 1640 supplemented culture medium.

Culture of Human Airway Epithelial Cells

The use of human tissues was authorized by the bioethical law 94–654 of the Public Health Code of France, with a written consent from the patients. The human airway tissue samples used here were nasal polyps obtained from patients undergoing nasal polypectomy. The tissue was then digested with 0.1% pronase E (Sigma Aldrich, St. Quentin Fallavier, France) for 12 h at 4°C, and the dissociated epithelial cells were seeded on glass coverslips coated with type I collagen (2.5 mg/ml in 0.016 mM acetic acid). The cells were grown to confluence in RPMI 1640 supplemented culture medium and incubated for 30 min with 10% PAO1 supernatant. We analyzed the protective effect of salmeterol on airway epithelial barrier junction by preincubating the airway cells with salmeterol (2 × 10−7 M) for 16 h and analyzing the ZO1 expression after incubation with PAO1 supernatant (30 min). For immunolocalization of ZO1, cells were fixed in methanol for 10 min at −20°C.

Mouse Model of Lung Injury

Animal studies were conducted according to protocols approved by the Institutional Ethical Committee for animal experiments from our institute. C57BL/6 mice (Harlan Sprague Dawley Inc., Gannat, France) received an intra-tracheal injection of a single dose of porcine pancreatic elastase (PPE) or PBS to study acute lung injury biomarkers. Mice were killed at 7 days postinjection. Frozen tissue was used for immunofluorescent labeling of ZO-1. Cryosections were used for immunolocalization of ZO-1.

Immunolocalization of ZO-1

Cryofixed cell cultures or cryosections of lung tissues were saturated for 30 min with 3% bovine serum albumin in PBS. Samples were successively (after intermediate washes in PBS) incubated for 1 h with a mouse monoclonal antibody to ZO-1 (1:10; Zymed), a biotinylated-sheep anti-mouse antibody (1:50; Amersham, Aylesbury, UK), and an Alexa Fluor 488–conjugated streptavidin (1:100; Molecular Probes). After incubation with the different antibodies, samples were counterstained with Harris hematoxylin solution for 10 s, then mounted with Aquapolymount antifading solution (Polysciences, Warrington, PA) onto glass slides. Slides were observed under an Axiophot fluorescence microscope (Zeiss) at ×40 magnification.

Image Analysis

To quantify the tight junction organization or disorganization, we propose a simple method of image processing using ImageJ software (7). Tight junction images of a normal epithelium in fluorescence microscopy show a regular network with continuous peaks of maximum intensity coupled with low intensity background (Fig. 1). The first step of image processing consists in plotting the gray level profile along a line drawn by the user as shown in Figure 2A. From this profile, the user determines a threshold value to discard any background (Fig. 2B). This user operation is required even if a top-hat segmentation is applied on image before analysis. Indeed, although the images show clear network features, a weak background is always detected even after top-hat segmentation (data not shown). Then, the user defines a polygon which passes through the network. This polygon is then extended to successive larger polygons according to a step chosen by the user (Fig. 2C) (the polygon perimeter lengths depart from a few hundred to a few thousand pixels according to image size). Then, the software counts the number of intersections between the polygons and the intensity peaks corresponding to the protein network and calculates the ratio of the number of intensity peaks to the polygon perimeter calculated in pixels. This ratio is defined as the Tight Junction Organization Rate (TiJOR). To be independent of pixel size in analyzed images, all data (step length and polygon perimeter) are saved in pixels for calculations in this method. Therefore, this ImageJ macro can be used on images with different pixel sizes.

thumbnail image

Figure 1. Fluorescence microscopy image of tight junction immunolabeling in airway epithelial cells in culture. (Scale bar: 20 μm).

Download figure to PowerPoint

thumbnail image

Figure 2. Image processing using ImageJ: the first step consists in plotting the gray level profile along a line drawn by the user (A) to determine the threshold value of the background (B). Then, the user draws, through the network, a polygon which is then successively enlarged with a step chosen by the user (C).

Download figure to PowerPoint

To validate the analysis process, we used simulated images of a network which is progressively degraded according well-known characteristics: 20 to 70% of the network features are progressively withdrawn as shown in Figure 3.

thumbnail image

Figure 3. Simulated network used to test the software: image A shows the first polygon on the entire network. Images B to G show successively destroyed networks from A.

Download figure to PowerPoint

Statistical Analysis

All statistical analyses were performed with Statview (Abacus Concepts, Berkeley, CA, USA). Data are expressed as median and were compared by using the nonparametric Mann and Whitney test. Values of P < 0.05 were considered significant.

RESULTS AND DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS AND DISCUSSION
  5. CONCLUSIONS
  6. LITERATURE CITED

Our method was first applied to simulated images displaying a regular network and a progressive disappearance of this network. For each image, TiJOR was calculated over four polygons with enlarged step of 33 pixels. Then, images were thresholded and the percentage of pixel area of the network was calculated. Results presented in Figure 4 show the relationship between TiJOR and the network pixel area percentages. A high and significant correlation coefficient (r = 0.95, P <0.05 ) was observed between TiJOR and the network density. To go further in the validation of the technique, we used the image displayed in Figure 1 to verify the influence of the step width used to progressively extend the successive polygons. As shown in Figure 5A, we observed that TiJOR remained constant whatever the width of the step between the successive polygons. The geometry of the polygons also did not modify the value of TiJOR as demonstrated in Figure 5B. These data confirm the relevance of the method that we used for quantifying the network density.

thumbnail image

Figure 4. TiJOR versus bright pixel area percentages calculated on simulated networks. These two parameters are calculated using the first polygon defined in Figure 3 which is extended three times using enlarged step of 33 pixels. The mean and standard deviation are determined over these four polygons.

Download figure to PowerPoint

thumbnail image

Figure 5. TiJOR versus increasing step size between two polygons (A) and versus geometric shape of the polygon (B). TiJOR is calculated on the image represented in Figure 1.

Download figure to PowerPoint

The method was then applied to biological samples. We have previously investigated the protective effect of salmeterol, a long-acting β2-adrenergic agonist, on P. aeruginosa (PAO1) supernatant–induced alteration of the epithelial junctional barrier (8). When airway epithelial cell cultures were exposed to PA01 supernatants, we demonstrated by Western Blotting that apical expression of ZO-1 protein was maintained in salmeterol-pretreated cell cultures, whereas it disappeared after PAO1 supernatant cell exposure in cultures not pretreated with salmeterol. The immunofluorescence images of ZO-1 obtained in this previous study were used to apply the present method of network quantification. As shown in Figure 6A in control cell cultures, a homogeneous immunolabeling of ZO-1 was observed, whereas exposition of cells to PAO1 supernatant decreased the expression of ZO1 (Fig. 6B). Pretreatment of cells with salmeterol prevented decrease of ZO1 labeling induced by PAO1 supernatant (Fig. 6C). The quantitative data obtained from these images presented in Figure 6D, demonstrate a significant decrease of TiJOR in cells treated with PAO1 supernatant and the annulation of this decrease when cells were pretreated with salmeterol.

thumbnail image

Figure 6. Effect of PA01 supernatant on TiJOR. Immunolabeling of ZO1 in (A) control airway epithelial cells. (B) Airway epithelial cells incubated with PA01 supernatant. (C) Airway epithelial cells preincubated with salmeterol. Measurement of TiJOR according to the three experimental conditions (D). (Scale bar: 20 μm)

Download figure to PowerPoint

We further tested our method on lung tissue sections of mice having received intrapulmonary elastase or PBS. In PBS-injected mice, ZO-1 adopts a uniform distribution in lung tissues (Fig. 7A) whereas a decreased and discontinuous ZO-1 labeling expression was observed in elastase-injected mice (Fig. 7B). The data obtained by measurement of TiJOR presented in Figure 7C, show a significant decrease of TiJOR correlated with the elastase-induced ZO1 disorganization.

thumbnail image

Figure 7. Effect of elastase injection on TiJOR. Immunolabeling of ZO1 on lung cryosections of C57BL/6 control mice (A). Cryosections of elastase treated C57/BL/6 mice (B). Measurement of TiJOR according to the two experimental conditions (C). (Scale bar: 20 μm)

Download figure to PowerPoint

In comparison with another robust fractal method like the box count method used by “Fractal dimension and lacunarity” plugin of ImageJ, our TiJOR method is slightly less accurate and slightly faster. However, the main advantage of our method is that it can be easily used by biologists without any specific knowledge of image analysis techniques, without image pretreatment, leading to a rapid learning and user-friendly method.

CONCLUSIONS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS AND DISCUSSION
  5. CONCLUSIONS
  6. LITERATURE CITED

This study introduces a very simple but powerful method for quantifying the organization of a network-like immunolabeling using ImageJ software. The quantification process that we propose is mainly applicable to cell cultures or tissue exhibiting membranous proteins present in tight junctions or adherent junctions. Our method is particularly suitable to analyze the effect of agents able to alter or to protect protein networks.

LITERATURE CITED

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS AND DISCUSSION
  5. CONCLUSIONS
  6. LITERATURE CITED
  • 1
    Furuse M.Molecular basis of the core structure of tight junctions.Cold Spring Harb Perspect Biol 2010;2:a002907. Review.
  • 2
    González-Mariscal L,Betanzos A,Avila-Flores A.MAGUK proteins: Structure and role in the tight junction.Semin Cell Dev Biol 2000;11: 315324.
  • 3
    Beil M,Braxmeier H,Fleischer F,Schmidt V,Walther P.Quantitative analysis of keratin filament networks in scanning electron microscopy images of cancer cells.J Microsc 2005;220 (Pt 2):8495.
  • 4
    Portet S,Vassy J,Beil M,Millot G,Hebbache A,Rigaut JP,Schoëvaërt D.Quantitative analysis of cytokeratin network topology in the MCF7 cell line.Cytometry 1999;35:203213.
  • 5
    Weichsel J,Herold N,Lehmann MJ,Kräusslich HG,Schwarz US.A quantitative measure for alterations in the actin cytoskeleton investigated with automated high-throughput microscopy.Cytometry A 2010;77A:5363.
  • 6
    Blacher S,Jost M,Melen-Lamalle L,Lund LR,Romer J,Foidart JM,Noël A.Quantification of in vivo tumor invasion and vascularization by computerized image analysis.Microvasc Res 2008;75:16978.
  • 7
    Abramoff MD,Magalhaes PJ,Ram SJ.Image Processing with ImageJ.Biophoton Int 2004;11:3642.
  • 8
    Coraux C,Kileztky C,Polette M,Hinnrasky J,Zahm JM,Devillier P,De Bentzmann S,Puchelle E.Airway epithelial integrity is protected by a long-actin beta2-adrenergic receptor agonist.Am J Respir Cell Mol Biol 2004;30: 605612.