Comparison of histochemical methods for murine eosinophil detection in a RSV vaccine-enhanced inflammation model

Abstract

A comparative study of histochemical detection of eosinophils in fixed murine tissue is lacking. Five histochemical methods previously reported for eosinophil detection were quantitatively and qualitatively compared in an established murine RSV vaccine-enhanced inflammation model. Nonspecific neutrophil staining was evaluated in tissue sections of neutrophilic soft tissue lesions and bone marrow from respective animals. Eosinophils had granular red to orange-red cytoplasmic staining, depending on the method, whereas neutrophils had, when stained, a more homogenous cytoplasmic pattern. Nonspecific background staining of similar coloration was variably seen in arterial walls and erythrocytes. Astra Blue/Vital New Red, Congo Red, Luna, Modified Hematoxylin & Eosin, and Sirius Red techniques were all effective in detecting increased eosinophil recruitment compared to controls; however, differences in eosinophil quantification significantly varied between techniques. Astra Blue/Vital New Red had the best specificity for differentiating eosinophils and neutrophils, but had a reduced ability to enumerate eosinophils and was the most time intensive. The Luna stain had excessive non specific staining of tissues and a reduced enumeration of infiltrating eosinophils making it suboptimal. For multiple parameters such as eosinophil detection, specificity, and contrast with background tissues, the Sirius Red followed by Congo Red and Modified Hematoxylin & Eosin methods were useful, each with their own staining qualities.

INTRODUCTION

Respiratory diseases are a significant cause of morbidity and mortality worldwide (WHO, 2001). Asthma, eosinophilic pneumonia, allergic bronchopulmonary mycosis, and vaccine-enhanced RSV inflammation are representative of various pulmonary diseases characterized by eosinophilic inflammation and these conditions have been previously reviewed (Alberts, 2004; Castilow et al., 2007; Cottin et al., 2005; Fulginiti et al., 1969, Kim et al., 1969, Scott et al., 2006; Smith, 1992; Teran et al., 1999). Investigation of either pathogenesis or efficacy assessment of prophylactic/therapeutic treatment for these diseases require detectable end points in model animals such as pulmonary eosinophil recruitment (Broström et al., 2007; Lemiére et al., 2006; Menzies-Gow et al., 2007). Whereas bronchoalveolar lavage can be a useful tool in this assessment, it does not readily detect changes of eosinophil distribution in the lung and may not recognize eosinophil recruitment that has not entered into the alveolar air space (Castilow et al., 2008).

Inbred and genetically engineered murine models are commonly used to study eosinophilic respiratory diseases (Lai, 2008). Murine eosinophils are identified in routinely stained hematoxylin and eosin (H&E) tissue sections by an eosinophilic granular cytoplasm and a bi-lobed nucleus, whereas morphologic differentiation from neutrophils can be challenging for some investigators since neutrophils have overlapping morphologic characteristics (Percy et al., 2007). While the H&E method can be used to detect eosinophils in tissue (Arantes-Costa et al., 2008), investigators often report using a variety of histochemical methods to further identify and quantify these leukocytes in tissue sections. Review of the literature suggests there is little consensus as to which histochemical method is optimal and whether or not there are selective advantages in the use of specific methods. In this study, we assess five histochemical methods reported in the literature for their effectiveness for detection of eosinophils in fixed tissue sections.

MATERIALS AND METHODS

RSV vaccination model

All experimental animal protocols were approved by the University of Iowa Institutional Animal Care and Use Committee. The vaccine-enhanced respiratory syncytial virus (RSV) inflammation model is regulated through RSV protein vaccination in conjunction with a genetic background variation of mice (Castilow et al., 2008). Briefly, interferon (IFN)-γ competent (BALB/cAnNCr, National Cancer Institute, Bethesda, MD, n=7) and -deficient mice (BALB/c.129S7 (B6)-Ifng^tm1Ts/J, The Jackson Laboratory, Bar Harbor, ME, n=7) are vaccinated with RSV fusion (F) protein using a vaccinia virus (3 × 10⁶ PFU via scarification at the base of the tail). Three weeks later the mice were intranasally inoculated with 1-3 × 10⁶ PFU RSV (A2 strain). RSV infection promotes prominent eosinophilic pulmonary inflammation in IFN-γ deficient mice, whereas IFN-γ competent mice lack eosinophilic inflammation. (Castilow et al., 2008a). After seven days the mice were euthanized (cervical dislocation) and the lung tissues harvested.

Tissues

Whole lung tissue was collected, and fixed (10% neutral-buffered formalin) under vacuum then routinely processed and paraffin embedded. The lungs were not inflated with fixative as mucus assessment and scoring is commonly required in the model. Serial sections (4 μm) were cut onto glass slides (Superfrost/Plus, Fisher); each slide was allowed to dry in a 37° C incubator overnight and then baked at 60° C for 30 minutes. The tissues were deparaffinized through xylenes, rehydrated in graded alcohols and rinsed in distilled water before the staining protocols were performed, followed by dehydration in graded alcohols, cleared in xylene and cover-slipped. All histochemical methods were followed according to their original citations unless specifically stated and in which we were able to optimize the method in preliminary studies.

Congo Red Protocol

The Congo Red protocol was slightly modified from the previous description (Friend et al., 2000) to optimize eosinophil detection. The tissues were immersed in Gill's Double Strength Hematoxylin (Polysciences, Cat # NC9480878) for five minutes (modified from original 30 second immersion) before being rinsed in running tap water. Sections were then stained with pre-filtered 0.5% Congo Red (pH 8-9, Aldrich) for fifteen minutes (extended from the original three to six minute immersion) and rinsed with deionized water.

Luna Protocol

The Luna protocol was performed as previously described with slight modifications (Hirasawa et al., 2007). The sections were immersed in working Hematoxylin-Biebrich (Sigma, Cat # H-3136 and Acros, CI 26905, respectively) scarlet solution (five minutes), then dipped (∼8x) in 1% acid alcohol and rinsed in tap water. Sections were then dipped (∼5x) in lithium carbonate solution until sections turned blue and washed in running tap water (two minutes).

Astra Blue/Vital New Red Protocol (AB/VNR)

This stain protocol was followed as previously described (Duffy et al., 1993). The sections were immersed in Astra Blue (Merck) for thirty minutes at room temperature before being rinsed in running tap water then immersed in Vital New Red (Pfaltz and Bauer; CI 253800) for thirty minutes at room temperature then rinsed in running tap water. The sections were counterstained in Harris hematoxylin (Surgipath, Cat # 01560) for five seconds then rinsed in running tap water and blued in ammonia.

Sirius Red Stain Protocol

For the Sirius Red stain protocol, we followed the online modification of the original method published by Llewellyn, including elimination of the sodium chloride step (Llewellyn, 1970, http://stainsfile.info/StainsFile/stain/amyloid/siriusllew.htm). Sections were placed in Harris hematoxylin (Surgipath, Cat # 01560, two minutes) and rinsed in running tap water followed by a rinse in 100% ethanol. The sections were immersed in an alkaline (pH 8-9) Sirius red solution (Sigma, CI 35780, two hours) and rinsed in running tap water

Modified Hematoxylin and Eosin (H&E) Protocol

Rather than utilize conventional H&E staining, we chose to use a Modified H&E so as to minimize background tissue eosin staining. The sections were immersed in Harris hematoxylin (Thermo Scientific, Cat # 72704, five minutes) then rinsed in tap water. Sections were subsequently placed in 0.5% acid alcohol (five seconds), rinsed in tap water, blued in 1% ammonia, rinsed again in tap water and rinsed a final time in deionized water. Sections were immersed in an alkaline eosin solution (pH 8.4, twenty seconds) which aided in the reduction of nonspecific “background” eosinophilic staining for enhanced contrast to detect eosinophils (Kiernan 2006). Standard H&E method was performed as described above except for the eosin, which was at a routine pH of 4.4.

Eosinophil quantification and statistics

Pulmonary eosinophils were quantified as described in a previous study (Castilow et al., 2008b). Briefly, sections of lung were examined by a pathologist blinded from the study and ten random foci were selected from low power (avoiding major airways/vessels), eosinophils counted per viewing field (600x magnification) and averaged for each lung. Eosinophil count data was not normally distributed therefore a natural log transformation was applied to normalize the data distribution prior to analysis. Assessment of eosinophil detection for each stain was performed with a linear mixed model analysis for repeated measures. This model was chosen for its ability to compare means since the different staining methods were performed on each tissue specimen. The mixed model analysis was used to account for the correlation from within the same tissue specimens as well as between groups, including the potential inter-stain relationships for eosinophil scoring and staining significance. A Tukey-Kramer post hoc test was performed to confirm statistical significance. Statistical significance was defined as p<0.05.

Qualitative assessment of eosinophil staining

Qualitative assessments of staining parameters were performed by a pathologist. Pulmonary eosinophils were examined for features such as eosinophil staining, contrast with background tissues, and lack of nonspecific neutrophil staining. Nonspecific murine neutrophil staining was assessed by examination of neutrophilic cervical abscesses (i.e. botryomycosis) and bone marrow from two BALB/cNCr mice. Neutrophil staining was also detected to a similar extent a mouse hepatitis virus (MHV) pneumonia model (n=3) with neutrophilic inflammation; however images were best represented from the bone marrow specimens so the MHV data is not shown.

RESULTS

Routine H&E can be an effective method for eosinophil detection. However, for some investigators, the “overlapping” staining pattern and morphology of eosinophils versus neutrophils can be problematic. In addition, eosin staining of background tissues can lower the visual detection of the target (eosinophils granules) and lower the optical contrast between cellular target and background (Figure 1) which can increase time required for examination and cause eye strain when examining numerous slide sections (personal experience of authors).

FIGURE 1.

Eosinophils (arrowhead) and neutrophils (arrow) from IFN-γ deficient murine lung, and bone marrow from mice with cervical abscess; hematoxylin and eosin method. Bar = 15 μm.

In this study we assessed five histochemical stains previously reported for eosinophil detection (Figure 2). In tissue sections, eosinophils typically had a fine to moderately-sized, distinctly granular cytoplasmic staining that varied in coloration and contrast with surrounding tissues depending on the histochemical method. Whereas coloration differences were obvious between methods, there were no dramatic differences in eosinophil staining intensity during examination. Some background tissues, including arterial walls and erythrocytes, often had similar staining coloration to eosinophil granules (Table 1). Whereas this did not seem to prevent visual detection of eosinophils, it did slow the efficiency and speed of examination because of the lowered contrast between eosinophils and background tissue. The Luna method had the most noticeable background staining that was patchy and included the nucleus/cytoplasm of various epithelia and stroma. In preliminary work, this background staining could be reduced by decreasing incubation times of the Hematoxylin-Biebrich scarlet solution; however, this further reduced the detectability and intensity of eosinophil staining. For most methods, using freshly prepared reagents for each batch was crucial for optimal consistency as staining pattern differences can occur between batches.

FIGURE 2.

Eosinophils (arrowhead) and neutrophils (arrow) from IFN-γ deficient murine lung, and bone marrow from mice with cervical abscess; Astra Blue/Vital New Red (AB/VNR), Congo Red, Luna, Modified Hematoxylin & Eosin (Mod H&E), and Sirius Red methods. Bar = 15 μm

Table 1.

Qualitative comparison of staining by histochemical methods used for eosinophil detection

Stain Method	Eosinophil cytoplasmic granules	Neutrophil cytoplasmic granules	Other Pulmonary Tissue Staining	Suggested Binding Mechanism
Astra Blue/ Vital New Red	Red granular	Absent to rare diffuse cytoplasmic staining	Arterial walls (weak), erythrocytes	Cationic colorant 1
Modified H&E	Eosinophilic granular	Absent to mild- moderate diffuse cytoplasmic staining	Erythrocytes, arterial walls	Stains structures with proteins rich in arginine and lysine 2
Congo Red	Orange-red granular	Absent to mild diffuse cytoplasmic staining	Arterial walls	Hydrogen bonding of azo/amine groups to hydroxyl radicals of eosinophils 3
Sirius Red	Red granular	Absent to mild diffuse cytoplasmic staining	Arterial walls, some interstitial collagen	Cationic colorant 1
Luna	Orange-red granular	Negative to moderate to strong orange-red diffuse cytoplasmic and nuclear staining; often in a patchy fashion	Erythrocytes, Arterial walls	Cationic colorant 1

¹Kiernan, J. (1999)

²Kiernan, J. (2006)

³Linda L. Vacca. (1985).

Each histochemical method was further assessed in its ability to detect quantitative differences in pulmonary eosinophil recruitment in the RSV vaccine-enhanced inflammation model. Each method was able to detect statistically significant increases in eosinophil infiltration in IFN-γ deficient (KO) as compared to IFN-γ competent (WT) mice as expected for the model (p<0.0001 for all methods, Figure 3). Interestingly, eosinophil detection in KO mice was dissimilar in extent between methods as the AB/VNR exhibited a significantly lower mean eosinophil count than Modified H&E (p<0.0001), Sirius Red (p<0.006), and Congo Red (p<0.003). The Luna method also yielded a lower mean eosinophil count than Modified H& (p<0.001), Sirius Red (p<0.01), and Congo Red (p<0.01).

FIGURE 3.

Comparison of histochemical methods to detect pulmonary eosinophils from IFN-γ deficient (KO) or IFN-γ competent (WT) mice in a vaccine-enhanced RSV inflammation model. * p<0.0001

An area of concern for investigators is the ability to readily differentiate tissue granulocytes as eosinophils and neutrophils. This is, in part, the underlying motivation for use of special stains in many published studies. An optimal histochemical method would specifically detect eosinophils while excluding neutrophils. To further assess this ability, we examined the various methods in bone marrow and soft tissues of two mice with chronic cervical botryomycosis and three mice infected with a strain of MHV (MHV data not shown) in which neutrophilic inflammation are predominant. In the bone marrow of these respective mice, eosinophils were detected by moderate to high intensity granular cytoplasmic staining. Eosinophil granule staining was equivalent to or slightly greater in the bone marrow compared to those seen in the lung of the RSV inflammation model. Neutrophils were morphologically detected based on slightly smaller size and typical nuclear morphology. Neutrophil cytoplasmic staining intensity ranged from absent to mild/moderate and, when present, cytoplasmic staining was homogenous, diffuse and less intense than that of eosinophils. The AB/VNR method had a distinct absence of neutrophil staining with only rare detection. (Figure 2 and Table 1). The Luna method had variable patchy neutrophil staining in bone marrow similar to the patchiness seen with background staining. The Modified H&E, Congo Red and Sirius Red all had a relative absence to mild amount of neutrophil staining, though the Mod H&E seemed to have slightly more neutrophil and background staining than the other two methods. Bone marrow eosinophil and neutrophil staining patterns was representative of that seen in the periphery of neutrophilic lesions. Neutrophil detection was typically confined to the periphery of the lesion and often in the vasculature or perivascular space. Once neutrophils were extravasated, there seemed to be rapid loss neutrophil staining (if present) which seemed to correspond with neutrophil degranulation as might be expected. Only the Luna method produced irregular to patchy staining in the degenerate neutrophils of the abscess which included nuclear and cytoplasmic staining.

Further considerations for histochemical staining of eosinophils are cost, technician labor time, and total stain time. For our laboratory, the total estimated material costs for one slide were comparable and ranged from $1.13 (Congo Red) to $2.59 (Luna). Technician labor time was similar and ranged from seven (Congo red) to fifteen (Luna) minutes with the exception of AB/VNR which was longer at about forty five minutes. The total method time from start to finish was similar at thirty-five to forty-five minutes with the exception of the AB/VNR which took two hours.

DISCUSSION

Detection and quantification of murine eosinophils in tissues are a major endpoint for investigators studying pathogenic mechanisms and in efficacy assessment of novel therapeutics in eosinophilic pulmonary diseases. Our study is topical and relevant as many of these techniques are commonly used (Vermaelen et al., 2003, Reis et al., 2001, and Hirasawa et al., 2007); however, documentation regarding their comparative usefulness for eosinophil detection in murine tissues is lacking.

In this RSV vaccine-enhanced inflammation model, all candidate methods were useful to detect the increased pulmonary eosinophil recruitment. Interestingly, there was significant variation in the absolute enumeration of eosinophils between methods even though these were serially sectioned through the same tissues. There are several possible explanations that could account for such variations.1) Each method has different suggested binding mechanisms (Table 1) for eosinophils which may have likely caused variations in uptake and affected detectability. 2) The presence of nonspecific staining in neutrophils could, in some instances, contribute to elevated eosinophil enumeration. The use of differences in staining pattern (e.g. distinct granules in eosinophils and diffuse homogenous staining in neutrophils) and morphology are useful to prevent this detection artifact. 3) “Increased” tissue staining (including both eosinophil granules and nonspecific background tissue) as seen with some techniques (e.g. Modified H&E) may also allow for enhanced staining of partially degranulated eosinophils which otherwise might be marginally stained (Daneshpouy, 2002). 4) Lastly, the level of visual contrast as well as the chromatic preferences by the examiner must be taken into account as this promotes a level of “user-friendliness” in tissue examination, a concept that can be appreciated by many pathologists. The authors of this study preferred the contrast offered by the red granular appearance of eosinophils on a lighter counter-stained background such as seen in the Sirius Red method versus the orange to red granular staining on a darker counter stain background used in the Congo Red and Modified H&E methods. Histochemical methods can sometimes be modified to adjust various parameters including eosinophil staining, background staining, and neutrophil staining so as to meet the preferences and comfort level of the investigator (Kiernan, 2006).

The resulting assessment of histochemical methods for eosinophil detection allows for some interpretative characterizations. The Luna method had generally higher amounts of patchy background staining of pulmonary epithelia and stroma. The cellular distribution of this background staining was similar to published images of Luna-stained murine lung (Angeli et al., 2008, Cotin et al., 2005; Hoenerhoff et al., 2006). This extra background staining may, in part, mask eosinophils from quick detection as suggested by the lower quantitation of pulmonary eosinophils. In addition, the Luna method also had a comparatively higher extent of neutrophil staining, but this did not evidently contribute to many false positives as it still had the lowest eosinophil count. These patchy nonspecific background and neutrophil staining suggest the Luna method should be excluded for eosinophil detection, especially in consideration of the other histochemical methods described in this study. The AB/VNR method showed good contrast between eosinophils and tissue background. Furthermore, it was very specific for eosinophils with near total absence of neutrophil staining, making it a viable option for investigators where specificity is essential or if the examiner has minimal histopathologic experience. In contrast, the AB/VNR method had reduced eosinophil enumeration ability and took moderately longer to complete (both direct labor and total technique time), potentially limiting its use. Sirius Red followed by Congo Red and Modified H&E proved to be our preferred histochemical methods for eosinophil detection. Each stain gave similar representative eosinophil numbers, had minimal neutrophil staining and had good contrast between eosinophils and background tissue for a “user-friendly” readability. The Sirius Red method was preferred because of the excellent tissue contrast for ease of eosinophil identification and cellular specificity which are both useful for investigators screening large numbers of tissues. Although the Congo Red did function well, it had slightly more neutrophil staining and the counterstain (as described from the published source) was darker than preferred, but the distinctive orange-red coloration of eosinophil granules were useful in eosinophil identification. The Modified H&E had a good detection of eosinophils in the lung; however, this method also had slightly more neutrophil and background staining than Sirius Red or Conge Red methods.

We have demonstrated that each of the selected histochemical methods reported in the literature are able to discern quantitative changes in the RSV vaccine-enhanced inflammation model. Even so, distinct differences were observed between methods in eosinophil enumeration and detection, along with nonspecific neutrophil or background tissue staining. This study suggests that Sirius Red, Congo Red and Modified H&E are all viable options for murine eosinophil detection, whereas the AB/VNR method might be used only for selective situations and the Luna method may be avoided. Furthermore, selection of a histochemical method should take into the examiner's histopathology experience as well as any chromatic preferences for a more productive and “ergonomic” visual examination.

ABBREVIATIONS PAGE

RSV

Respiratory Syncytial Virus

H&E

hematoxylin and eosin

Mod H&E

modified hematoxylin and eosin

AB/VNR

Astra Blue/New Vital Red

IFN

interferon

WT

wild type

KO

knock out

MHV

mouse hepatitis virus