SEARCH

SEARCH BY CITATION

Keywords:

  • Enzyme-based method;
  • image analysis;
  • immunohistochemistry;
  • light microscopy;
  • quantification;
  • sampling paradigm

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Despite the advent of ever newer microscopic techniques for the study of the distribution of macromolecules in biological tissues, the enzyme-based immunohistochemical (IHC) methods are still used widely and routinely. However, the acquisition of reliable conclusions from the pattern of the reaction products of IHC procedures is hindered by the regular need for subjective judgments, in view of frequent inconsistencies in staining intensity from section to section or in repeated experiments. Consequently, when numerical comparisons are required, light microscopic morphological descriptions are commonly supplemented with analytical data (e.g. from Western blot analyses); however, these cannot be directly associated with accurate structural information and can easily be contaminated with data from outside the region of interest. Alternatively, to eliminate the more or less subjective evaluation of the results of IHC staining, procedures should be developed that correct for the variability of staining through the use of objective criteria. This paper describes a simple procedure, based on digital image analysis methods and the use of an internal reference area on the analyzed sections, that reduces the operator input and hence subjectivity, and makes the relative changes in IHC staining intensity in different experiments comparable. The reference area is situated at a position of the section that is not affected by the experimental treatment, or a disease condition, and that can therefore be used to specify the baseline of the IHC staining. Another source of staining variability is the internal heterogeneity of the object to be characterized, which means that identical fields can never be analyzed. To compensate for this variability, details are given of a systematic random sampling paradigm, which provides simple numerical data describing the extent and strength of IHC staining throughout the entire volume to be characterized. In this integrated approach, the figures are derived by pooling relative IHC staining intensities from all sections of the series from a particular animal. The procedure (1) eliminates the problem arising from the personal assessment of the significance of the IHC staining intensity, (2) does not depend on the precise dissection of the tissue on a gross scale and (3) considerably reduces the consequences of limited, arbitrary sampling of the region of interest for IHC analysis. The quantification procedure is illustrated by data from an experiment in which inflammatory reactions in the murine spinal cord, measured as microglial activation, were followed by IHC after the lesion of the sciatic nerve.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

There is a continuous need for accurate comparisons of the distributions of macromolecules in experimental biology or in medical diagnostics, and recurrent attempts are made to achieve the objective light microscopic characterization of the intensity and distribution of reaction products in immunohistochemically (IHC) stained sections. The primary obstacle during the quantification of staining is the frequent non-reproducibility of the pattern, strength and contrast of staining in repetitive investigations. This can originate from variations in the recording conditions (e.g. from uncontrolled, illumination-induced fading in fluorescent microscopy [Kirkeby & Thomsen, 2005], or from differences in the alignment and aperture setting in transmitted light microscopy), which can be at least partially corrected or compensated. However, even with identical microscopic conditions, certain variations still arise from unavoidable differences in the replication of the staining protocols in repeated experiments. Although this manuscript focuses on the problems associated with enzyme-based techniques, some of the difficulties are common in both enzyme- and fluorescence-based methods. To overcome the difficulties of quantification, individualized classification criteria are often introduced. The methods applied may involve an entirely subjective approach, based on the observer's estimation of staining intensity, stained cell density and cell morphology (Colburn et al, 1997; He et al., 1997). Alternatively, a semi-quantitative analysis may be applied, with the measurement of greyscale intensity, an operator-dependent thresholding method being used for determination of the percentage of the total area covered by stained cells (Blackbeard et al., 2007). Automated analysis of corrected immunoreactive areas in selected sections from the region of interest is also possible, likewise based on a subjective, visual selection of the threshold for image segmentation (Stuesse et al., 2000).

The other major cause of problems during quantification is the biological variability. The main source of such experimental error is the interindividual variability, which is responsible for about 70% of the observed experimental variance (Howard & Reed, 1998). Further variance stems from the hierarchical nature of microscopic investigations, that is, from the frequently neglected differences between blocks and sections: in other words, from the way the object is sampled. The contribution of this factor to the overall variance is in the range of 20–25% (Howard & Reed, 1998). Generally speaking, such statistical variability may be decreased by increasing the number of observations (Gundersen & Østerby, 1981). The first problem, however, is associated at least in part, with observer subjectivity. This type of error (bias) does not affect the precision of the measurements, but since the real values sought experimentally are normally not known, the systematic deviations from these values cannot be estimated (see e.g. West, 1999).

The present manuscript provides an integrated procedure that attempts to correct for both types of variability. The method consists of two modules: compensation for the changes in staining intensity at the section level, and the application of systematic sampling at the object level. The irregularity of staining in repeated experiments is taken into account by determining the staining intensity in reference regions in each section that are presumed to be unaffected by the experimental treatment. The density of IHC staining (or the percentage area covered by stained profiles) in the area of interest (AOI) will be related to that in the reference area (AOI′) in the same section. To determine the positivity (significance) of the immune reaction, objective criteria are used: the standard deviation (SD) and the background level of the staining intensity are determined for the AOI′ and a segmentation threshold (background + 2 × SD) is then applied to the whole section. Regions with staining intensity exceeding this limit are accepted as significantly stained profiles. To compensate for the non-homogeneous spatial distribution of the stained structures within the volume to be characterized, a robust systematic random sampling is applied (Gundersen & Jensen, 1987). A specimen containing the entire region of interest is dissected and sectioned equidistantly, with a starting position set at random. The parameters characterizing the IHC staining in each section, related to the AOI′, are then pooled to arrive at simple numerical data expressing the IHC staining of the object of interest.

To illustrate the procedure step by step, an experimental model of nerve injury is used. Nerve transection (axotomy), which is a well-established model of neuronal injury (Koliatsos & Price, 1996), leads to the activation of microglial cells at the site of injury in the central nervous system (Kreutzberg, 1996; Obál et al., 2001). This type of cellular response can be attenuated by treatment with minocycline, a member of the tetracycline class of antibiotics (Zemke & Majid, 2004). In our example, the microglial response in the ventrolateral horn of the murine spinal cord to axotomy of the sciatic nerve is compared with that after minocycline treatment. An enzyme-based IHC method is used to detect the microglia, and the proposed integrated evaluation/sampling procedure is applied to demonstrate the effect of treatment.

Material and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Animals

B6/SJL hybrid mice were bred in the Biological Research Center (Szeged, Hungary) from breeding pairs originally obtained from Charles River Laboratories (Budapest, Hungary). Twelve-week-old male mice were used for this study. All animal experiments were performed in accordance with the institutional guidelines for the use and care of animals and with Hungarian governmental laws relating to animal protection (protocol no. XVI./03876/001/2006). The animals were housed (at most four animals per cage) at room temperature under a 12 : 12 h light/dark cycle, with free access to food and water.

Surgical procedure and pharmacological treatment

Surgery was performed under Avertin (tribromoethanol; Fluka, Buchs, Switzerland) anaesthesia (240 mg kg−1 body weight in a 0.3 mL volume, administered intraperitoneally). The left sciatic nerve was transected in the mid-thigh, and a 3–4-mm portion was removed to prevent regeneration. The incision was then sutured, and the animals were returned to their cages after they had recovered from the narcosis. One group of operated mice were injected intraperitoneally once a day with minocycline (an inhibitor of microglial activation; Sigma, Budapest, Hungary), dissolved in sterile saline solution (40 mg kg−1 body weight in a 0.1 mL volume), first applied 1 h prior to the surgery. The animals in both groups (operated + minocycline-treated, or operated only) were sacrificed on day 7 after the sciatic nerve transection.

Tissue preparation and sampling

Under terminal anaesthesia, the animals were perfused transcardially with 30-mL phosphate-buffered saline (PBS; 10 mm) followed by 30 mL 3% paraformaldehyde (PFA; Sigma) in 10 mm PBS. The lumbar segments of the spinal cords throughout the entire region involving the expected inflammatory reaction (approximately 10–12 mm in length) were removed and post-fixed overnight in 3% PFA (in 10 mm PBS) at 4°C (Fig. 1(A)). Tissue blocks were then cryoprotected in 30% sucrose (Sigma) in 10 mm PBS for 3 days at 4°C, and embedded in OCT medium (Tissue-Tek, Zoeterwoude, the Netherlands) for transverse sectioning. Transverse sections at a thickness of 30 μm were cut from the spinal cord along its longitudinal axis on a cryostat (Kryostat 1720; Leitz, Wetzlarr, Germany) in such a way that the starting positions were selected at random: the first N sections (N selected at random between 1 and 10) were cut and discarded; then, systematically, each section at a distance of 300 μm was harvested, until the segment of spinal cord had been sectioned through (Figs 1(B) & (C)). Sections were collected in 10 mm PBS in a 24-well tissue culture plate (one section in each well) and stored at 4°C until processed further.

image

Figure 1. Dissection and sampling of tissue for immunohistochemical (IHC) characterization. (A) A piece of tissue (with a characteristic length of 10–12 mm) from the lumbar portion of the spinal cord was dissected, post-fixed and embedded for sectioning. The dissection was made according to gross anatomical landmarks in such a way that the specimen encapsulated the entire region with the expected inflammatory reaction (red-shaded area). It is not necessary to know the exact location of the region of interest within the dissected tissue. (B) The dissected piece of the spinal cord was systematically sampled by transverse sectioning along the axis of the spinal cord. The starting position for sampling was randomly selected by determining the first location of the specimen where the entire cross-section of the spinal cord could be recognized (onset point; section no. 0); then, before collection of the sections, a random number of sections (between 1 and 10) with 30-μm thickness were cut and discarded. Thus, the sampling started with section no. 1 located at an arbitrary distance xμm (1 < x < 300 μm) from the border of the tissue. Next, a systematic set of sections was cut at 300 μm distances until the whole spinal cord sample was sectioned through. (C) A set of equidistantly spaced sections was prepared from each spinal cord for IHC staining and quantification, which means that each section in the series represented a 300-μm transverse slice of the spinal cord (15 slices in the figure). The region of interest within the sections was determined by means of microscopic landmarks in the sections. Since the size (length) of the dissected tissue was intentionally larger than that of the region with the expected inflammation (red-shaded area), the very first and the very last sections in the series may not display an increased number of microglial cells in the region of interest.

Download figure to PowerPoint

Immunohistochemical staining of sections

To detect microglial cells with an IHC procedure, free-floating sections were stained according to the avidin–biotin technique (Cuello, 1993). Sections were rinsed first with three changes in PBS, which was followed by a 30-min blockade of endogenous peroxidase activity with 0.3% hydrogen peroxide in 10 mm PBS containing 0.2% Triton X-100 (TPBS; Sigma). Then, after washing (three changes in PBS), to reduce the non-specific staining, sections were incubated in 2% normal goat serum (Vector, Burlingame, CA, USA) in TPBS for 1 h. Microglial cells were labelled with anti-CD11b antibody (1 : 500 in TPBS; rat-anti-mouse antibody, Serotec, Oxford, UK) during incubation at 4°C overnight. After washing in PBS (three changes), sections were incubated in biotin-labelled secondary antibody for 1 h (1 : 800 in TPBS, goat anti-rat IgG, Vector), followed by washing in PBS (three changes), and incubation in the avidin–biotin complex (1 : 1600 in PBS, Vector) for 1 h, at room temperature. Finally, after washing in PBS, the reaction was visualized by incubation in diaminobenzidine (DAB, 5% in PBS; Sigma) for 15 min. Stained sections were thoroughly rinsed in PBS, mounted on silane-coated glass slides, dehydrated in graded series of ethanol, processed through xylane (Molar, Budapest, Hungary) and cover-slipped with Entellan (Merck, Darmstadt, Germany).

Light microscopic image recording

IHC-stained sections were examined in an Olympus Vanox-T AH-2 light microscope (Olympus, Tokyo, Japan). The alignment of the microscope was checked at the beginning of each microscopic session with special attention to the coaxial position of the lenses, the apertures and the recording CCD camera. The same magnification (objective lens) was used to record all images belonging in a particular series. To avoid any slight variability in reproducing aperture settings, both the field limiting and contrast apertures were kept at the fully open position during (digital) photography. Images were recorded with a Spot RT CCD camera (Diagnostic Instruments, Sterling Heights, MI, USA) in colour mode, using full (1600 × 1200 bit) resolution at 8-bit depth for each (RGB) colour component. The automatic exposure option of the camera was used if all the image components of interest in a section could fit into a single field of view. When more than one grabbed image was necessary from a section, the first snapshot was made with automatic camera control, and the following photographs were taken with the same exposure settings as the first. At the beginning of each microscopic session, a flatfield image was taken and used through the session to correct for uneven illumination. The white balance was adjusted for each individual slide. Recorded images were stored in colour mode in uncompressed files with tagged image file format.

Image analytical procedures in single sections

Digital images were analyzed by using the built-in functions of the Image-Pro-Plus image analysis software (Media Cybernetics, Silver Spring, MD, USA) running under the Windows XP operation system on a PC. (Most of the used or equivalent operations are also available in other free or commercial software.)

Although the original colour (digital) pictures were used for easier orientation and to specify the AOI in the sections, all operations were performed on images previously converted to 8-bit greyscale. The evaluation steps included (1) specification of the AOI and AOI′, (2) calculation of the background level and variance of the staining intensity on the 0–255 greyscale, (3) segmentation of the significantly stained pixels and (4) calculation of the percentile area of the significantly stained profiles and/or their average density in the AOI, relative to the similar values in the AOI′. A simplified version of such calculations to estimate relative changes in IHC staining has already been tested in another model of nerve injury (Obál et al., 2006).

Specification of the AOI and the corresponding AOI′ The AOI in a single section could be determined by using characteristic anatomical landmarks, such as, in our case, the border of the grey and white matter, the position of the central channel and the perimeter of the pools of motor neurons (Fig. 2(A)). In complex structures, detailed atlases should be consulted, but in most cases the specific regions can be easily marked. In our study, a general histological stain (cresyl violet, which stains Nissl substance and cell bodies) was used in one set of sections to prove that correct orientation was achieved in the IHC-stained sections with no counterstaining, or without the regular need for reference material (Fig. 2(B)). When the region to be analyzed was successfully identified in a section, it was outlined with the pointer device and marked as an (irregular) AOI. For reference purposes, any indifferent region of the section would be suitable; however, exploiting the symmetric organization of the nervous system, we propose that an identical area on the contralateral side of the section should be used. To define this area, the AOI on the operated side was mirrored about the symmetry axis of the section; then, to correct for occasional slight distortions of the sections, it was manually positioned to the corresponding location on the non-operated side (Fig. 2(B)).

image

Figure 2. Specification of the reference area and the area to be analyzed. (A) A section containing the whole cross-sectional area of the spinal cord of a non-operated animal was stained with cresyl violet, which reveals the distribution of neuronal and glial cells. The border of the grey and white matter is readily recognized (dotted line), as are the positions of the central channel (CC) and the pools of symmetrically positioned large motor neurons at ventral (VP and VP’) and ventrolateral (VLP and VLP’) locations (as outlined, left and right). (B) On individual microglia (CD11b) staining, the major anatomical landmarks in the section remain identifiable (border of the grey and white matter, central channel). The sciatic nerve cut induced microglial activation at the dorsal horn (DH), and in the ventral + ventrolateral pools of motor neurons. The border of the ventro-lateral pool was outlined (area of interest, AOI), the AOI was then mirrored to the symmetry axis (SA) of the section to define the reference area (AOI′). The position of AOI′ was manually corrected, if this was required owing to slight asymmetry of the section.

Download figure to PowerPoint

Calculation of the background level and variance of the staining for segmentation.  The major anatomical landmarks were first identified in the original colour images of the sections, and the area where an increased immune reaction was expected on the operated side and the reference area on the contralateral side were then marked accordingly (Figs 3(A) & (B)). The colour image was next converted to the greyscale with a range of 0–255 (Figs 3(C) & (D)). To determine the background level, a large format low-pass filter (512 × 512 pixels) was first applied to the greyscale image (Figs 3(E) & (F)), which smoothed out the variability in the image owing to differences in IHC staining by substituting the grey value of each pixel with the average grey value of its neighbourhood with a dimension of 512 × 512 pixels. Although the image obtained approximated the grey level very well at image points outside the stained profiles (cf. Figs 3(D) & (F)), this background distribution is certainly contaminated with the signal from the heavily stained area, which is obvious from a comparison of Figs 3(C) and (E). Next, to correct for this effect, an adjusted (second) background level was calculated. To derive this number, the first background level within the AOI′ was used for thresholding: all the image points darker than this level (which, at this approximation, were regarded as the stained profiles) were excluded from the image, and an average grey value was calculated for the remainder of the pixels within the AOI′ (Figs 3(G) & (H)). This number was accepted as the final value representing the grey level of the background (BCKGND).

image

Figure 3. Step-by-step procedure for the segmentation of significantly stained profiles in IHC-stained sections. (A, B) According to the anatomical landmarks (shape of the section, border of the grey and the white matter [dotted line], etc.), the region where the increased staining is accepted is outlined on the operated side; then, symmetrically, the reference area is specified on colour images of the section (RGB, 8-bit depth). (C, D) The colour pictures are converted to greyscale images with a range of 0–255 (from black to white) and the AOI and AOI′ are next copied as overlay images. (E, F) Application of a large-scale low-pass filter (512 × 512 pixels) to the whole images resulted in approximated background images by smoothing (averaging) the details in the greyscale images. In this particular case, the filtering process resulted in an average greyscale (background) level of 233 in the reference area (AOI′). Comparison of the greyscale distribution within and outside the AOI on the operated side indicates how the recruitment of the stained cells (Fig. 3E) affects the local grey level of the background. (G, H) When the segmentation threshold was set to the calculated background (a grey value of 233 in the presented case), all the pixels with grey values lower than this limit (black ‘holes’ in the image) were excluded from the image, and a new average background level (238) was then calculated for the remainder of the pixels within the reference area (AOI′). With this new value for the background (BCKGND), a standard deviation (SD) of the greyscale distribution was calculated for the pixels darker than this level (i.e. grey values lower than this number) within the AOI′ in the greyscale images (Fig. 3D), and a cut-off value for the segmentation of the significantly stained profiles was calculated at BCKGND – 2 × SD. (I, J) With the cut-off value, the segmented profiles within the AOI and the AOI′ were outlined and superimposed on the original colour images for visual control of the results representing the significantly stained profiles.

Download figure to PowerPoint

The variance of the staining, that is, the variance of the grey level distribution of the images, in addition to the BCKGND value, was used as an aid to determine which image point should be regarded as significantly stained. To calculate the standard deviation (SD) of the distribution of staining intensity, all the image points within the AOI′ of the original greyscale images were taken into account. Finally, those pixels in the AOI and AOI′, whose grey values were lower than BCKGND – 2 × SD (i.e. significantly darker than the background), were dissected and accepted as significantly stained image points. For illustration and for visual control of the results, image areas with dissected pixels were outlined, and these profiles and the perimeters of the AOI and AOI′ were superimposed on the original colour images (Figs 3(I) & (J)).

Calculation of tissue-specific parameters from the relative changes in IHC staining in individual sections.  Two types of (non-independent) parameters were derived: (i) the percentile areas of the significantly stained profiles within the AOI and AOI′ and (ii) the intensities of average staining within these areas. The first parameter was readily obtained after segmentation of the greyscale images at the cut-off level, since this operation yields the total area of the AOI (equal to that of AOI′) and the sum of the areas of the segmented profiles in both regions. The numerical difference of these percentile areas results in a relative number describing how much larger the coverage of the cellular staining on the operated site is than that of an unperturbed, control area (Δ(area)).

The attenuation law (Bouguer–Lambert–Beer law) states that

  • image

where I0 is the incident and It is the transmitted light intensity, M denotes the molar absorptivity (extinction coefficient), T is the section thickness and C is the local concentration of the absorbing species. Assuming homogeneous section thickness and similar extinction properties of the absorbing material in the AOI and AOI′, analogous equations can be written for the operated and control sides as:

  • image

Hence,

  • image

which means that the logarithm of the ratio of the light intensities (measured on a greyscale in this case) in the AOI and the AOI′ is proportional to the difference in concentration of the absorbing species in these areas, that is, the target of determination (Δ(concentration)).

Obtaining parameters which characterize the entire volume to be analyzed

In each equidistant section cut from the tissue, consistent individual parameters can be derived which are either proportional to the concentration of the IHC reaction products (Δ(concentration)), or to their dimensions (Δ(area)) on the operated side relative to the corresponding data on the control side. Summation of the corresponding data obtained from each section, furnished simple numerical data characterizing the dimension (ΣΔ(area)) or the extent of IHC staining (ΣΔ(concentration)) throughout the whole volume of the region of interest from a given animal (Figs 4(A) & (B)). The process of summation of data situated equidistantly along the third Cartesian axis is equivalent to a certain type of integration, which incorporates the third dimension; thus, the calculated numerical data should be regarded as having dimensions of volume and mass on an arbitrary scale.

image

Figure 4. Charts showing relative area (A) and density (B) data obtained from the analysis of individual sections belonging in the same systematic random series. The category axis gives the serial numbers of consecutive sections. Obviously, the exact location of the starting point of the systematic sampling is indifferent, since the difference between the operated and non-operated sides fluctuates around zero outside the volume affected by the experiment, and is therefore cancelled out on a statistical basis during summation of the data. (A) The percentile area covered by the IHC products in the reference area (AOI′) was subtracted from that in the region of interest (AOI) in each section and plotted. The sum of the data (taking into account their signs) is 159.54, which characterizes the extent of the IHC-stained structures within the volume to be analyzed. (B) The calculated numbers from each section [–ln(I(op)/I(co))], which are directly proportional to the differences in concentration of the IHC-stained structures, are plotted against the section number. The result of summation (0.67) characterizes the accumulation of IHC-stained material in the volume of interest.

Download figure to PowerPoint

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Six animals serving as controls underwent sciatic nerve transection, while four other mice received minocycline treatment in addition to the axotomy. The reducing effect of the minocycline treatment on the microglial activation in the spinal cord following axotomy was tested by comparing the levels of CD11b IHC staining in the ‘operated only’ and ‘operated + minocycline’ groups of animals with the methods presented earlier. The data in Table 1 reveal that for both the reduction in the volume occupied by the significantly stained cells (P= 0.022), and the reduction in the total density of staining (P= 0.026) in the ventrolateral pool of motor neurons, the effect of minocycline was significant (Student one-tailed t-test).

Table 1.  Effect of minocycline treatment on CD11b IHC staining after sciatic nerve cut.
AnimalsOperated onlyOperated + minocycline
Sum of relative stained area (%)Sum of ln(I(op)/I(co))Sum of relative stained area [%]Sum of ln(I(op)/I(co))
#1374.281.67375.791.47
#2394.031.83425.091.97
#3549.732.38417.561.92
#4642.553.19365.811.72
#5541.822.47
#6657.433.24
Mean526.642.46396.061.77
SEM49.010.6614.810.23

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Although snap freezing can reduce the tissue preparation time considerably, which may be critical if tissue bio-banking is associated with a rapid intra-operative diagnosis, perfusion or immersion fixation gives superior tissue preservation (Beltz & Burd, 1989); furthermore, even for diagnostic purposes, most of the tissue is still treated with formalin fixative (Steu et al., 2008). Admitting also that snap-frozen tissue can additionally offer exceptional possibility of the combination of stereological, histological and molecular analyses of adjacent sections from the same tissue (Schmitz et al., 2000), the chemical fixation preparation methods for IHC are still more widespread than the physical freezing procedures. For this reason, we followed the traditional perfusion fixation protocol for tissue preparation in the present study.

The commonly experienced difficulties inherent in the interpretation of IHC-stained sections are mainly due to unavoidable variations arising from two main sources: the variability of the staining at the section level and the diversity of the structure at the tissue level. In consequence, subjective input from the operator is necessary at critical points in the evaluation, which influences the results to an incalculable degree. The approach presented here for the evaluation of IHC-stained specimen, exemplified by an experimental application, attempts to reduce these uncertainties.

Procedures at the section level

Operator input.  In the presented method, the input of the operator is reduced to one unavoidable step: specification of the boundary of the studied AOI. The use of an identical AOI′ in the same section for control purposes ensures that changes relative to an indifferent area are calculated in each case even if, by chance, the specification of the AOI is not absolutely precise. If the AOI is to some extent smaller than the area delineated by the boundary of the anatomical region to be analyzed, the area will be under-represented, but the calculated relative difference between the AOI and AOI′ will not be considerably affected. If the AOI is somewhat larger than the area to be analyzed, the measured difference from the control region will be only slightly ‘diluted’ with data from an unaffected area, that is, the method will merely lose some sensitivity. The major landmarks in the sections, however, allow easy and reliable delineation of the area AOI (see Fig. 2). Accordingly, all such inconsistencies would cause relatively minor changes in the results.

Determination of the background staining. In the present experiments, no counterstaining was performed, and there was therefore no need for spectral (colour) aided segmentation of the IHC-stained profiles from the counterstained background, which would have required complex operations (van der Laak et al., 2000; Brey et al., 2003; Pham et al., 2007). Thus, instead of the selection of a specific wavelength (or colour) for analysis, or decoupling of the intensity data from the red–green–blue colour model, all images were converted to greyscale images represented only by 8-bit data at each image point, and the image analysis procedures were performed in a single channel with simplified calculations.

In each section, the intensity of the background staining (determined in an indifferent reference area) is used to identify image points with a level of staining significantly above that of the surroundings both in the AOI′ and within the AOI. Thus, the procedure ensures that the variability of staining in repeated experiments or the heterogeneity of the staining from section to section in a given series will be compensated. The procedure, which automatically calculates the background intensity on the basis of the distribution of the staining intensity without the aid of the operator, consists of two steps. In the first step, an average background image is calculated by applying a large-format low-pass filter to the digital image of the section, which yields a greyscale value at each image point with the average for its 512 × 512 pixel neighbourhood (Figs 3(E) & (F)). Since the great majority of the image points show no evidence of staining (Figs 3(C) & (D)), the result of this filter approximates closely, though not perfectly to the true background distribution (Figs 3(E) & (F)). After low-pass filtering, an average grey value (approximate background level) can be calculated within the AOI′ (233; Fig. 3(F)). The deviation of this number from the true (unknown) background is obviously greater in those regions where stained profiles is notably accumulate (cf. Figs 3(C), (E) and (D), (F)). To correct for this ‘contamination’ of the background distribution from the stained cells, as a second approximation, all image points with grey values lower than the calculated first background value (233) are excluded from the calculation, and a new average background value (BCKGND = 238) is determined for the remainder of the pixels within the AOI′. This procedure of successive approximation of the true average background within the AOI′ should in principle be stopped when the difference between the results of successive iterations is smaller than a specified limit. In the presented experimental example, even in the third step, the change in the calculated average background greyscale value was less than one digit on the 0–255 scale, and accordingly no further iterations were performed.

Segmentation of stained profiles and calculation of parameters characterizing the stained structures. After determination of the BCKGND and the SD values of the greyscale distribution in the AOI′, a cut-off level is determined to segment those image points that are considered significantly stained, that is, pixels with grey values significantly different from that of the background. This seemingly arbitrary decision is supported by statistical principles if the deviation of the cut-off level from the BCKGND value is set in units of SD, which determines the confidence of the results. In biological applications, this offset is normally set to a value in the range 2 × SD to 3 × SD implicating a probability between 67 and 99% for the segmented image to reflect the distribution of the authentic stained profiles. In our application example, the level BCKGND – 2 × SD was consistently applied.

After segmentation, different parameters were derived to characterize the extent and strength of the IHC staining in the AOI relative to the AOI′. The difference between the numbers of pixels segmented in the AOI and the AOI′ (since these areas are identical) reflects the net increase in the stained cellular profiles due to the treatment. For easier comparison of the results obtained from different sections and different animals, these raw data were normalized to the total number of pixels within the AOI or the AOI′, and expressed as the percentile area difference (Δ(area)). The difference in staining intensity between the AOI and the AOI′ was determined directly by using the attenuation (Bouguer–Lambert–Beer) law, applied to the whole area of the AOI and the AOI′. Through measurement of the average (optical) density in these areas, a number corresponding to the difference in the percentile stained area was obtained (Δ(concentration)). The method is analogous to the classical procedures applied successfully in conventional (electron) microscopy to determine the local mass of a specimen by measuring the contrast of the micrographs (Halliday & Quinn, 1960; Zeitler & Bahr, 1965; Edie & Karlsson, 1977).

Procedures at the tissue level

Determination of the boundary of the AOI and the setting-up of the exact anatomical location along the direction perpendicular to the plane of the sections (unlike the same procedure within the plane of the section) is often hindered by the missing characteristic landmarks that can be recognized only in a set of consecutive sections (see Fig. 1). Thus, to overcome the need for serial sectioning in order to determine the exact anatomical location within the tissue, if probing of a tissue by means of microscopic sections is intended, arbitrary (random) sampling is often the choice. The paradigm, however, can lead to misleading results, due to the intra-individual variability of the tissue, an effect which is frequently neglected. However, as illustrated qualitatively in a composite figure assembled from the sections of our application example (Fig. 5), huge differences may be observed in the amount of staining, depending on the actual position of the section within the region to be analyzed. Quantitatively, as depicted in Fig. 4, fluctuations from section to section as large as 50–100% of the derived parameter values can be determined in the region of the tissue containing the expected reaction.

image

Figure 5. Composite images from sections of the spinal cord along its axis. View from the back of the animals (upper view). (A) Cresyl violet staining of the sections reveals the major structures within the tissue: the midline of the spinal cord and the large (motor) neurons on the operated and non-operated sides are clearly visible. Heterogeneity of the distribution of the large motor neurons along the axis of the spinal cord on both sides is evident. (B) CD11b IHC staining demonstrates a non-homogeneous distribution of activated microglial cells along the axis of the spinal cord on the operated side, which may correspond to the similar heterogeneous distribution of the injured motor neurons. Clearly, a large variation in the number of stained cells may be expected, depending on the actual positions of the sections within the region of the injury.

Download figure to PowerPoint

The proposed procedure of systematic sampling is not sensitive to the actual positions of the sections within the tissue, since the intra-individual variations average out during summation of the results obtained on the individual sections. The method is not sensitive to the actual starting point of the sampling either, as long as the dissected tissue contains the entire region of interest, since the differences neutralize each other during the summation process outside the affected volume, due to the statistical nature of the variations between the operated and the control side (see Fig. 4).

Application of the procedure

In our application example, as a positive control, the effect of an antibiotic (minocycline) on the local microglial reaction on the spinal cord was tested experimentally after lesion of the sciatic nerve. This known anti-inflammatory compound has been shown to improve regeneration and reduce microglial activation in neuronal lesions (Blackbeard et al., 2007; Shankaran et al., 2007; Henry et al., 2008; Mishra & Basu 2008). In accordance with the literature data, our method to quantify the relative staining intensity of IHC-stained microglial cells after axotomy, demonstrated a significant decrease in microglial activation after minocycline treatment (Table 1). The presented procedure is of general applicability when an unbiased estimation of relative changes in IHC staining intensity is required in specified anatomical regions of biological tissue.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

This work was supported by grants from the Hungarian Scientific Research Fund (OTKA-T 048718) and the Hungarian National Office for Research and Technology (GVOP-3.2.1. 2004–04-0052/3.0 and RET-DNT 08/04). The authors are grateful to Mr. Szabolcs Siklósi for his help with the artwork.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References