Salmonella Typhimurium is one of the main pathogens compromising porcine and human health as well as food safety, because it is a prevailing source of foodborne infections due to contaminated pork. A prominent problem in the management of this bacteriosis is the number of subclinically infected carrier pigs. As very little is known concerning the mechanisms allowing Salmonella to persist in pigs, the objective of this study was to develop an immunohistochemical approach for the detection of salmonellae in tissue of pigs experimentally infected with Salmonella Typhimurium. Samples were obtained from a challenge trial in which piglets of the German Landrace were intragastrically infected with Salmonella enterica serovar Typhimurium DT104 (1.4-2.1×1010 CFU). Piglets were sacrificed on days 2 and 28 post infection. Tissue samples of jejunum, ileum, colon, ileocecal mesenteric lymph nodes (Lnn. ileocolici), and tonsils (Tonsilla veli palatini) were fixed in Zamboni’s fixative and paraffin-embedded. Different immunohistochemical staining protocols were evaluated. Salmonella was detected in varying amounts in the tissues. Brown iron-containing pigments in the lymph nodes interfered with the identification of Salmonella if DAB was used as a staining reagent. Detergents like Triton X-100 or Saponin enhanced the sensitivity. It seems advisable not to use a detection system with brown staining for bacteria in an experimental setup involving intestinal damage including haemorrhage. The use of detergents appears to result in a higher sensitivity in the immunohistochemical detection of salmonellae.

Salmonella is an important pathogen threatening porcine and human health as well as food safety. Amid the most pervasive sources of foodborne diseases, Salmonella enterica serovar Typhimurium (S. Typhimurium) is especially strongly linked with human disease caused by the consumption of contaminated pork.1-3 In pigs, an S. Typhimurium infection triggers clinical symptoms with enterocolitis, often followed or accompanied by sub-clinical infections of silent carrier animals that can function as a reservoir, infect other animals and transmit the pathogen to the food chain.4-6 Salmonella infects cells lining the epithelial layer of the small and large intestine such as M-cells, absorptive enterocytes or goblet cells and may cross this barrier via different mechanisms to invade the lamina propria.7-9 After reaching the lamina propria of the intestinal mucosa, Salmonella is mainly taken up by macrophages in which they then replicate in a protected intracellular niche and which may also transmit the bacteria to other organs.10,11 Salmonellae harbour a sophisticated arsenal of mechanisms to survive and replicate in the host. Although bacterial persistence is a key phase of a pathogen’s life cycle and represents an opportunity for disease control, very little is known about how the pathogen survives for long periods of time in the mammalian host in the presence of immunosurveillance.12 In order to study the largely unknown mechanisms used by Salmonella to persist in pigs5 and particularly to trace Salmonella’s route through the body, it is a great challenge to reliably mark and track salmonellae in histological sections of different organs and tissues. As part of a big research consortium, one aim of our working group was the demonstration of S. Typhimurium in paraffin embedded tissues from experimentally infected pigs. Since immunohistochemistry represents a suitable approach to do this,13 we evaluated several protocols for applicability which yielded very heterogeneous results. The refined immunohistochemical protocol is presented. Further information concerning other aspects of the same experiment may be found elsewhere.14,15

The samples for this study were obtained during a Salmonella challenge trial already described.14 In short, samples were obtained from a probiotic feeding trial in which piglets of the German Landrace were intragastrically challenged with Salmonella enterica serovar Typhimurium DT104 (1.4-2.1×1010 CFU). Piglets from each group were sacrificed on days 2 and 28 post infection (DPI). The animals were euthanized by an overdose of pentobarbiturates (Narcoren, Merial GmbH, Germany) under general azaperone (Stresnil, Janssen Animal Health, Neuss, Germany) -ketamine (10% ketamine, Bremer Pharma GmbH, Warburg, Germany) anaesthesia. Samples of mid-jejunum, ileum, colon ascendens, ileocecal mesenteric lymph nodes (Lnn. ileocolici), and tonsils (Tonsilla veli palatini) were taken within 15 min after sacrifice and treated as already described.14,16 All samples were rinsed in ice-cooled Ringer solution. Intestinal samples were cut open on the mesenterial side, trimmed to squares and pinned on cork pieces with the mucosal side facing upwards. The tissues were fixed for 26h in Zamboni’s fixation solution and rinsed in PBS, dehydrated in a graded series of ethanol, embedded in paraffin, cut to 5 µm thin sections, mounted on HistoBond® slides (Paul Marienfeld GmbH & Co. KG, Lauda-Königshofen, Germany), dewaxed in xylene and rehydrated in a decreasing series of ethanol. Experimental approval had been given by the local authority / Regional Office for Health and Social Affairs Berlin (Landesamt für Gesundheit und Soziales, Berlin ID: G0348/09).

Different immunohistochemical staining protocols with a monoclonal mouse anti-Salmonella Typhimurium antibody (Mouse-anti-Salmonella Typhimurium, monoclonal, Maus-IgG, 5mg/mL, Clone 8C11C, Acris Antibodies GmbH, Herford, Germany) were evaluated. Two protocols are presented here for comparison (Table 1).

For the identification of the gold-brown pigments encountered in lymph nodes, the very same histological samples employed for immunohistochemistry were used. For the demonstration of iron, Berlin blue method (trivalent iron) and Turnbull blue method modified according to Quincke (bi- and trivalent iron) were applied.17 Lipofuscin was detected according to Hueck and Pearse.18 To validate the staining protocols, liver samples of goat, sheep, cow or rat, available in our institute from stock, were used as positive controls. The liver was chosen, since a number of different pigments may be seen as an incidental finding within hepatocytes and Kupffer cells, amongst them lipofuscin and hemosiderin.19

S. Typhimurium was detected immunohistochemically in varying amounts depending on the time post-infection, tissue localization and the protocol used. The positively stained objects appeared as roundish to longish particles with a diameter of ~2 µm. With protocol 1, single bacteria were observed lying freely between cells of tissues as well as intracellularly, often in cells with the morphology of macrophages (Figure 1). Salmonellae were abundant in the tunica mucosa of ileum and colon 2 DPI, where several bacteria appeared to group in clusters. The ileal domes were particularly frequented by the pathogen. S. Typhimurium was found in lymph nodes, albeit in low numbers, and was not detectable in tonsils. In addition to the immunohistochemically labelled salmonellae, spots of brown pigment with the same size as the bacteria were visible in sections of lymph nodes. They also appeared in the control sections (Figure 2C). As the brown iron-containing pigments (see below) in the lymph nodes interfered with the identification of Salmonella if DAB was used as a staining reagent, HistoGreen was used instead, which labels the targeted bacteria in a bright green-blue color (Figure 2 D,E). After implementing a permeabilization step with detergents like Triton X-100 or Saponin (protocol 2), we found that the sensitivity was considerably enhanced. More staining signals were visible and we also found the bright green-blue color easier to recognize (Figure 3). In addition to a higher amount of Salmonella, which could now be detected in ileal and colonic tissues, it was now frequently possible to show the presence of bacteria in lymph nodes and tonsils. In lymph nodes and tonsils the staining signals were dispersed throughout the tissue and the bacteria appeared not to be grouped in clusters (Figure 3D). In the positive controls derived from cultured S. Typhimurium, the staining intensity between single bacteria varied notably. Whereas approximately 1% of the bacteria exhibited a strong positive reaction and circa 5-10% exhibited a moderate staining signal, most cells showed very weak or almost invisible staining grades. The strongly stained bacteria were the biggest ones. They were rodshaped and the staining signal was situated at the perimeter of the cells. The moderately stained cells appeared to be a little smaller and were also rod shaped. Bacteria with weak or nearly no staining appeared to be the smallest ones and exhibited a more roundish shape. A similar phenomenon could be observed in the tissue samples, in which the staining intensity differed between single bacteria in the gut lumen and inside the tissue (Figure 3C).

In the above mentioned samples of lymph nodes, particularly in their medulla, spots of brown pigment were visible, which had the same size as the expected immunohistochemical reaction product for S. Typhimurium (Figure 2 A,C). Iron deposits were positively demonstrated using Turnbull blue and Berlin blue (Figure 2B). Lipofuscin reaction was negative. Positive staining reactions for lipofuscin as well as iron were detectable in the positive controls.

Histochemistry, especially immunohistochemistry, is a suitable approach to investigate the exact localization of a pathogen in situ. It enables the researcher to correlate its occurrence to e.g. pathologic lesions or other pathogens.20,21 In the present study, S. Typhimurium was detected immunohistochemically in different porcine tissues of the intestine and in the tonsils. Frequently, the pathogen was spotted within cells, often in those which morphologically resembled macrophages. The staining results concerning bacterial morphology and distribution were principally comparable to descriptions found in the literature.6,8,22-24 In the sections of lymph nodes including control sections (IgG - and buffer control), spots of brown pigment of approximately the same size as the expected immunohistochemical reaction product for S. Typhimurium were visible. Since these pigment granules were of similar size and color as the labelled bacteria, it was necessary to distinguish them from the microorganisms. The pigment granules were found to contain iron, presumably representing hemosiderin, which can be a result of mucosal haemorrhages.25,26 Consequently, the detection system was changed to a green color to solve this problem, although it has to be noted that DAB gives a more crisp staining result compared to HistoGreen. It seems advisable not to use a detection system with brown staining in experimental setups involving intestinal damage including haemorrhage.

Immunohistochemistry is a powerful tool to demonstrate microorganisms in tissue samples;27 however, an important question can be raised concerning its detection limit. As microorganisms are at the limit of light microscopical detection and results may vary from slide to slide because of heterogeneous distribution of the bacteria in the tissue, histology is not the method of choice for routine diagnosis and quantification of bacterial infections in tissues.22 One method routinely used to detect and quantify microbes is microbial plate counting. Interestingly, an organ-specific difference between results of our immunohistochemical labelling of S. Typhimurium and microbial plate counting done in the same trial by Kreuzer et al. could be found.2 2 DPI levels around 103 CFU/g tissue could be quantified via plate counting in the tonsils, jejunum and lymph nodes, whereas S. Typhimurium was hardly detectable in these organs via immunohistochemistry using protocol 1. In contrast, immunohistochemistry of S. Typhimurium in ileal and colonic tissue 2 DPI was reliably possible. Plate counting done for these organs resulted in higher levels, namely between 104 - 106 CFU/g tissue. We therefore conclude that the number of CFUs was under the immunohistochemical detection limit in tonsils, jejunum and lymph nodes with protocol 1. In protocol 1 we applied only a low concentration (0.05%) of a mild membrane solubiliser (Tween 20) for a short time (5 min). In protocol 2 we applied a higher concentration (0.1%) of a mild detergent (Saponin) or a harsh detergent (Triton X-100) for a longer time (20 min). This modification enhanced the immunohistochemical sensitivity enormously and S. Typhimurium was also detectable in the previously negative organs. The observation of sensitivity enhancement due to detergents stands in contrast to the statement that antigen retrieval is in general not required for the demonstration of bacteria in fixed tissues.28 For example Searle, et al. used a permeabilization step for the immunocytochemistry but not for the immunohistochemistry to detect Salmonella.29 In support of our findings, other studies also used detergents for histologic Salmonella demonstration, although for different applications, e.g., cryosections, thicker sections or immunocytochemistry.30-34 The rationale behind the use of detergents in immunocytochemistry and applications using thicker tissue sections or cryosections is to allow the antibody to reach the antigen if it is situated in a cell compartment shielded by a membrane, especially after aldehyde fixation.35 Detergents are surface-active molecules that self-associate and bind to hydrophobic surfaces in a concentration-dependent mode.36 For example for Saponin, it was demonstrated that through interaction with plasma membrane cholesterol, it makes cells permeable without major disruption of organelles, by literally opening pores in the plasma membrane when used in higher concentrations.37 Within mammalian cells, Salmonella inhabits a membrane-bound vacuole known as the Salmonella-containing vacuole but also colonizes the cytosol of cells.38 Therefore, we assume that a permeabilization step in the immunohistochemical protocol is necessary to access the bacteria situated in the cytosol as well as the ones in the membrane-surrounded vacuoles. Negative immunohistochemical results may otherwise be false negatives or the amount of detected bacteria artificially low.

To the best of our knowledge, there are very few published detection limits for the immunohistochemical identification of bacteria in histological samples (102 CFU g-1 tissue for mycobacteria in fish39). Based on our observations, we propose a detection limit of roughly 102-103 CFU per g tissue in our experimental setup. The detection limit may of course be different for e.g., different antigens, targeted bacteria and chosen staining protocols, as was also demonstrated in this study. Since some recent studies used enzymes in their staining protocols, this could be another option to enhance sensitivity.40-42 An additional reason for a varying immunohistochemical detection limit may be a potential change of surface structure of S. Typhimurium in different environments. The antibody used in this study was directed against heat-inactivated LPS from S. Typhimurium. As described in the Results section, staining signals from S. Typhimurium recovered directly out of the culture medium were heterogeneous and mostly weak, whereas those of tissue sections of the ileum and colon were strongly visible. We also noticed differences between the tissue resident bacteria and the ones in the intestinal lumen. The observed size differences between the variably stained bacteria could be attributed to the Quellung reaction.22 Quellung (German word for swelling) is the result of the combination of the polysaccharidal bacterial capsule antigens with the specific antibody, resulting in an apparent capsule swelling.43 For the fungus Cryptococcus neoformans, which is used as a system to study capsule reactions because it has a large polysaccharide capsule that is readily visible by light microscopy, it was shown that distinct capsular reactions depend on the antibody epitope specificity and the yeast serotype.44 Therefore different degrees of Quellung-reaction and resulting different detection sensitivities could also be possible in Salmonella-immunohistochemistry. It has been established that phase and antigenic variation lead to substantially altered heterogenic phenotypes of a clonal bacterial population. It has been shown that surface antigens in particular vary under differing conditions, even during the journey through the body, to avoid adverse immune reactions and establish long term persistence.45 Another reason for the stronger staining signals detected inside of the intestinal tissues might be the tendency of Salmonella to form microcolonies,46 thereby probably amplifying the antigen concentration in one spot. Additionally, the accumulation of dense material surrounding intracellular S. Typhimurium, supposedly originating from lysed bacterial products, was described in an transmission electron microscopical study.47 This material could also amplify the staining signal.

In conclusion, the use of detergents seems to be necessary for the proper immunohistochemical detection of Salmonella in paraffin embedded tissues and enhances the identification sensitivity. Additionally it is advisable not to use a detection system with brown staining for bacteria in an experimental setup involving intestinal damage including haemorrhage.