Whipple’s disease (WD) is a rare disease with an approximate prevalence of less than three cases per million people investigated in the general European population (data from Italy) (1, 2). It is caused by chronic systemic infection with the gram-positive bacterium Tropheryma whipplei (T. whipplei), which belongs to the group of Actinobacteria, one of the largest subgroups in the Bacteria domain. Related infectious organisms include Mycobacterium and Corynebacterium species (3). T. whipplei are non-motile and rod-shaped, about 1.4 µm in length and surrounded by a trilaminar cell wall containing glycans at the external membrane (4). Humans represent the only known host of T. whipplei and, most commonly, middle-aged white men are affected (mean patient age is 55 years, 85% of patients are male). Patients with classic WD usually exhibit a wide range of non-specific gastrointestinal symptoms (e.g., diarrhea, abdominal pain, and malabsorption), ascites, weight loss, anemia, lymphadenopathy, and fever (1, 5, 6). WD is a multi-systemic disease with manifestation in various extra-intestinal organs including the central nervous system (CNS) (10–50% of WD patients), the heart (endocarditis), joints (isolated arthritis, spondylodiscitis), the eye (uveitis), and lung (pneumonia) (7).

Notably, patients can also be asymptomatic carriers of T. whipplei, with up to 11% of the general European population testing positive in feces (data from Switzerland) (8, 9). T. whipplei is endemic in sewer workers and hyper-endemic in rural Senegal, as well as in families of patients and carriers (10, 11). Exposure to contaminated soil has been described as a possible way of infection with T. whipplei (1, 10, 12, 13); however, the most common infection transmission route is via the fecal-oral route. Close contact to carriers (1, 10, 11, 14) poor hygiene and living conditions (14), and the absence of toilets (11) are considered risk factors for infection with T. whipplei.

Inherited or acquired immunodeficiency is suspected to be required for the development of classic WD (15–17). The characteristic duodenal histology, with marked infiltration of the intestinal mucosa by macrophages, which can engulf T. whipplei but are incompetent to fully degrade them, can be explained by the lack of excessive local inflammation and an alternation in the phagocytic cell activation toward the phenotype of M2 macrophages (18).

Until the introduction of antibiotic treatment in the 1950s, WD was considered fatal due to progressive cachexia or involvement of the CNS (1, 6, 18–20). Presently, WD is managed by an antibiotic regimen consisting of an initiation therapy with ceftriaxone followed by retention therapy with co-trimoxazole orally over the course of 1 year (1, 19, 20).

Using EM, T. whipplei was and can be successfully visualized in various stages of degradation in the extracellular spaces and in macrophages of WD patients, for example in duodenal biopsies. However, due to its time consuming analysis, EM is no longer recommended as a diagnostic technique for WD (16, 21, 26).

Tropheryma whipplei can be cultured by the centrifugation-shell vial technique and a human fibroblast cell line (human erythroleukemia cell line); however, this approach is a time consuming method due to the generation time of 18 days of T. whipplei. Microbial culture has successfully been applied as method for the detection of T. whipplei from synovial fluids and feces of patients with WD (3, 22, 27, 28).

With the development of a T. whipplei specific polyclonal antibody, immunohistochemistry has become an important tool in the diagnostic algorithm of WD (29, 30). Immunohistochemistry represents a specific method for diagnosing WD even in patients with atypical manifestations and when PAS stains lacks specificity (16, 17, 24, 30).

Corresponding to PAS staining, anti-T. whipplei immunohistochemistry remains positive during the course of antibiotic treatment. The degraded remains of T. whipplei persist in the affected macrophages, and because the antibody links to epitopes in the wall of the bacteria, macrophages stain positive during and after treatment (1, 29, 30).

However, T. whipplei specific polyclonal antibody has to be produced in macrophages of rabbits and is not commercially available.

Tropheryma whipplei-specific quantitative polymerase chain reaction (qPCR) is a reliable and specific method for diagnosing infections with T. whipplei, if confirmed by sequencing of multiple T. whipplei target genes to avoid false positive results. A singular positive result from gastrointestinal tissue may be the result of a simple colonization with T. whipplei. An actual WD should be suspected if saliva and feces specimens are both tested positive via qPCR and clinical symptoms are also present (31, 32). qPCR has also been described as a supportive diagnostic tool for the analysis of synovial fluid in suspected seronegative rheumatic diseases (33).

PCR analysis from cerebrospinal fluid is highly recommended in all WD cases, because asymptomatic T. whipplei infection of the CNS has been described in up to 40% of all patients with gastrointestinal manifestation of WD and represents a complication with a high mortality (1, 7, 17).

Bacterial species have previously been detected in situ by hybridizing fluorescent oligonucleotide probes (of 15–30 base pairs in length) targeted against specific sequences in either the 16S or 23S ribosomal RNA (rRNA) of bacteria (34–38). Fluorescence in situ hybridization (FISH) can be applied on cytological smears of fluids and formalin-fixed paraffin-embedded (FFPE) specimens alike. rRNA is abundant (~10⁵ copies per cell) and evenly distributed in the cytoplasm of bacteria, resulting in a positive staining of the whole bacterial cell by the respective FISH probe (34–40). Preamplification of specific rRNA sequences is not a prerequisite for this approach. Stained specimens can be analyzed by standard epifluorescence microscopy (34–40). The FISH probes can be individual designed and labeled, but a large amount of probes for diverse bacteria is also commercially available so far.

Fredricks et al. (5) investigated the localization of T. whipplei using FISH and confocal laser scanning microscopy in six intestinal and two lymph node specimens with WD and could describe the extracellular distribution of viable bacteria (5). Geissdorfer and colleagues (15) analyzed blood culture-negative endocarditis in 255 patients using microbiological culture, PCR, FISH, conventional histopathology, and immunohistochemistry: 16 cases tested positive for T. whipplei by PCR, while only one was positive by FISH (15). Audoly et al. (41) analyzed the impact of deglycosylation of a supposed T. whipplei biofilm on the discrepancies between diagnostic results in WD by FISH and immunofluorescence and hypothesized that surface glycoconjugates have a protective role for T. whipplei and should therefore be removed to allow efficient bacterial detection (41).

These studies, which show that T. whipplei can in principle be successfully detected and visualized by FISH, lead us to the question of whether FISH-based diagnosis can be a reliable tool in routine FFPE samples in cases of suspected WD. To this end, we used a specific probe for T. whipplei and established a protocol for FFPE routine specimens in our lab.

Immunohistochemical analysis was performed as previously described (30). Briefly, 3-µm thick sections were fixed to Super Frost slides (Thermo Scientific, USA), deparaffinized in xylene (2× 10 min) and rehydrated with graded ethanol (100, 85, 70, 50, 20%, each for 2 min), ending in distilled water. The slides were then incubated in 3% hydrogen peroxide for 10 min and non-specific antigen binding sites were blocked by treating the slides with Reagent 1 (ZytoChemPlus Horse Radish Peroxidase (HRP) Polymer Kit, Zytomed Systems, Germany) for 5 min. Afterward, the slides were incubated with a primary antibody for 1 h in a humidified chamber at room temperature. The primary antibody against T. whipplei, as described in “Whipple’s disease: Immunospecific and Quantitative Immunohistochemical Study of Intestinal Biopsy Specimens” by Lepidi et al. (30) and kindly provided by Dr. Didier Raoult (30), was used at a dilution of 1:500. Unbound primary antibody was removed via rinsing with washing buffer for 5 min, before Reagent 2 (ZytoChemPlus HRP Polymer Kit, Zytomed Systems, Germany) was used for signal enhancement for 20 min. After rinsing with washing buffer, slides were incubated with the secondary antibody (ZytoChemPlus HRP Polymer Kit, Zytomed Systems, Germany) for 30 min. The slides were rinsed again with washing buffer and the bound antibodies were detected with 3,3′-Diaminobenzidine (DAB, DAB Substrate Kit High Contrast, Zytomed Systems, Germany). Finally, sections were rinsed with distilled water for 15 min, cellular nuclei were stained for 2 min with haemalum, dehydrated in ascending Ethanol, and mounted with Eukitt media (Sigma-Aldrich, Germany).

Sections of 3 µm thickness were fixed on Super Frost slides (Thermo Scientific, USA) and air dried overnight. Sections were deparaffinized (60 min at 60°C, 2× 10 min 100% Xylene, and 5 min 100% ethanol) and air dried. For dissolving components, FISH probes were preheated to hybridization temperature and 10 µl of the indocarbocyanine (Cy3) labeled T. whipplei-specific probe (Biovisible, Netherlands) mixture was applied on each section and covered airtight with a coverslip. After hybridization at 52°C in a humid chamber for 18 h, sections were uncovered and washed at hybridization temperature in a wash-buffer (MCW 5, Biovisible, Netherlands) for 10 min. Sections were then rinsed with distilled water, air dried, and mounted with 4′,6-Diamidin-2-phenylindol (DAPI) Dura Tec (Zytomed Systems, Germany) and a coverslip.

The sequence of the T. whipplei-specific probe (TAT TGC AAC CCT CTG TAC CA) had previously been described by Geissdorfer et al. in Ref. (15). Compared to all 16S rRNA entries in the blast NCBI (National Centre for Biotechnology Information), the T. whipplei-specific probe showed 100% coverage to T. whipplei with a five nucleotide mismatch to the next similar organisms (Rhodanobacter strains, Dyella strains, Dokdonella strains, Tahibacter strains, and Frateuria strains) (42, 43).

In order to correlate conventional immunohistochemical analysis with FISH analysis, a combined immunofluorescence FISH analysis was performed.

First, FISH staining was carried out following our protocol described above. For the concomitant immunofluorescence staining, the primary antibody against T. whipplei was applied after rehydration with graded ethanol (100, 85, 70, 50, 20%, each for 2 min) finishing in distilled water, then incubating with 3% hydrogen peroxide for 10 min, and with Reagent 1 (ZytoChemPlus HRP Polymer Kit, Zytomed Systems, Germany) for 5 min. Subsequently, sections were washed and incubated with a cyanine (Cy2)-labeled secondary antibody (ImmunoResearch Laboratories Inc., USA) for 30 min. After washing, sections were air dried and mounted with DAPI Dura Tec (Zytomed Systems, Germany) and a coverslip.

Periodic acid–Schiff diastase staining was graded qualitatively, based on the morphological classification system described by von Herbay (16) and colleagues and semi quantitatively as described by Baisden et al. (16, 29).

The fluorescence slides were analyzed using an Olympus BX 51 epifluorescence microscope equipped with filters sensitive for DAPI (Excitation peak: 358 nm, DNA-emission peak: 461 nm and RNA-emission peak: 500 nm), Cy3 (Excitation peak: 550 nm, Emission peak: 570 nm), and Cy2 (Excitation peak: 492 nm, Emission peak: 510 nm) (Olympus corporation, Japan).

Photos were taken with a digital microscope (Keyence BZ9000 E, Osaka, Japan). Representative microphotographs were taken with filters for the detection of Cy3 (OP-66838 BZ TRITC Excitation wavelength: 540/25 Dichroic mirror wavelength: 565 Absorption wavelength: 605/55), Cy2 (OP-66836 BZ filter GFP-BP Excitation wavelength 470/40 Dichroic mirror wavelength: 495 Absorption wavelength: 535/50), and DAPI (OP-66834 BZ filter DAPI-BP Excitation wavelength 360/40 Dichroic mirror wavelength: 400 Absorption wavelength 460/50). Microphotographs were processed using the GNU Image Manipulation Program (GIMP, Version 2.8) on an IBM compatible PC running Windows 10.

Specimens of all 16 cases were morphologically classified using the system proposed by von Herbay et al. (16) using PAS diastase staining (16). In six biopsies, macrophages predominantly showed granular intensive PAS positive cytoplasm (type 1), eight biopsies showed a grossly granular and intensive PAS positive cytoplasm and a diffusely granular, slightly PAS positive cytoplasm in the background (type 2), and in five biopsies macrophages had a diffuse or finely granular, scarcely PAS positive cytoplasm (type 3) (Table 2).

Specimens of all 16 cases with WD were positive for T. whipplei in immunohistochemical analysis, using the polyclonal (rabbit) anti-T. whipplei antibody described by Lepidi et al. (30) (Table 2). Staining was predominantly intracellular and congruent with foamy PAS diastase positive macrophages. In these cells, the pattern of antibody labeling was similar to the pattern of PAS diastase staining (Figure 1).

In all six patients with reported antimicrobial therapy, including three patients with longitudinal biopsies, immunohistochemical analysis for T. whipplei was positive. Congruent with reported results, the staining intensity declined after treatment, but nevertheless remained positive in all samples.

No specific positive signal for T. whipplei was detected in the control specimens.

To detect T. whipplei in the FFPE specimens, FISH was used on routine tissue sections of duodenal biopsies, lymph nodes, and brain tissue. These specimens showed an excellent morphological quality.

In nine cases, T. whipplei could successfully be visualized using the species-specific FISH probe. In the seven remaining cases, it was not possible to detect T. whipplei organisms (Figure 1; Table 2). As expected, all negative controls showed no signals with the species-specific FISH probe. In all seven cases having a diagnosed Helicobacter pylori infection, rod-shaped bacteria were detectable in the mucus of the gastric mucosa in the DAPI channel. These were negative for the T. whipplei-specific FISH probe, however.

In three cases of WD with an abundance of PAS diastase positive macrophages with a type 1 classification (according to von Herbay et al.) and positive immunostaining, the FISH probe visualized a prominent number of T. whipplei bacteria. In these patients, a specific antimicrobial therapy before sampling had not been reported (Figure 2).

In two other cases without antimicrobial therapy and a type 2 and 3 classification [according to von Herbay et al. (16)] with a slightly positive immunolabeling signal, bacteria could easily be identified using FISH but were not distributed uniformly. In one case without reported antimicrobial therapy, no bacteria were detectable with FISH. The 3 FISH-positive patients who underwent a previous specific antibiotic treatment all showed no bacteria in the initial biopsy. Although, after a period of 2 years, the macrophages showed an intensive, granular PAS positive cytoplasm and a diffuse, more finely granular PAS positive cytoplasm in the background type 2 and 3 [according to von Herbay et al. (16)]; this pattern likely indicates an intestinal remission, and only/very few bacteria could be detected by FISH in the subsequent biopsy.

In the other cases, no bacteria could be detected via FISH, and most of these were classified as advanced WD with macrophages of type 2 or 3 [according to von Herbay et al. (16)].

Tropheryma whipplei could be detected with FISH in 4 out of 10 cases with a specific antibiotic treatment, possibly indicating persistence or reoccurrence of active WD (Figure 3).

Using a combination of T. whipplei-specific FISH and immunofluorescence labeling, we could show that FISH and the specific antibody consistently label very similar areas in the tissues examined. However, some immunohistochemical signals were also found separate from FISH-positive bacteria. In contrast to FISH, antibody labeling was consistently found co-localizing with the PAS diastase-positive macrophages at the intracellular, and also at the extracellular compartments (Figure 4).

In order to verify our assumption that treated and inactive cases of WD are not detectable via FISH, a larger cohort of untreated cases of WD has to be investigated. Ideally, this would mean longitudinal studies of patients with WD before and after antibiotic treatment. By using multi-labeled FISH probes, it might be possible to increase signal intensity to allow detection of bacteria with a low content of rRNA and by this improving the overall sensitivity of the assay (47, 48).

Combination of immunohistological staining using species-specific antibodies and FISH probes could allow new insights into the cellular composition of ongoing infection and distribution of bacteria in their natural habitat (5, 41).

In our study, we could show in 12 patients from routine practice that FISH has the potential to diagnose active WD in FFPE specimens and enables the direct visualization and detection of individual vital bacteria in the infected tissue (5, 6, 34, 38, 49, 50). While sensitivity of FISH is not yet ideal, the commercially available FISH represents a valuable addition to the diagnostic kit and may be improved by the deglycolisation method proposed by Audoly et al. (41). FISH has value as an ancillary diagnostic tool, if used correctly. Our results show that FISH may be a helpful method to detect recurrence of disease after antibiotic treatment when the signal of PAS diastase and the specific immunohistochemistry lags behind the recurrence of disease, especially if the clinical course of the patient and antimicrobial treatment is considered.

Fluorescence in situ hybridization should not be used for a definite exclusion of WD but can visualize and resolve individual vital T. whipplei bacteria on formalin-fixed specimen of active WD. It therefore may be considered an indicator for the efficiency of antibiotic treatment or a potential indicator of a reactivation or reinfection in the course of WD treatment.