Uptake of Tropheryma whipplei by Intestinal Epithelia

Background: Tropheryma whipplei (TW) can cause different pathologies, e.g., Whipple’s disease and transient gastroenteritis. The mechanism by which the bacteria pass the intestinal epithelial barrier, and the mechanism of TW-induced gastroenteritis are currently unknown. Methods: Using ex vivo disease models comprising human duodenal mucosa exposed to TW in Ussing chambers, various intestinal epithelial cell (IEC) cultures exposed to TW and a macrophage/IEC coculture model served to characterize endocytic uptake mechanisms and barrier function. Results: TW exposed ex vivo to human small intestinal mucosae is capable of autonomously entering IECs, thereby invading the mucosa. Using dominant-negative mutants, TW uptake was shown to be dynamin- and caveolin-dependent but independent of clathrin-mediated endocytosis. Complementary inhibitor experiments suggested a role for the activation of the Ras/Rac1 pathway and actin polymerization. TW-invaded IECs underwent apoptosis, thereby causing an epithelial barrier defect, and were subsequently subject to phagocytosis by macrophages. Conclusions: TW enters epithelia via an actin-, dynamin-, caveolin-, and Ras-Rac1-dependent endocytosis mechanism and consecutively causes IEC apoptosis primarily in IECs invaded by multiple TW bacteria. This results in a barrier leak. Moreover, we propose that TW-packed IECs can be subject to phagocytic uptake by macrophages, thereby opening a potential entry point of TW into intestinal macrophages.


Introduction
The Gram-positive rod Tropheryma whipplei (TW) can cause two different pathologies that affect the gastrointestinal tract. The first, classical Whipple's disease (WD), is a rare, systemic, chronic infectious disease that occurs in only a minority of patients exposed to TW and depends on a defect of the host's innate immune system that is thus far only partially understood [1][2][3][4][5][6]. The second is a self-limiting gastrointestinal infection that makes this bacterium more prevalent in human disease than previously appreciated [7][8][9][10][11][12][13].

Establishment of Epithelial Models for Translocation of T. whipplei
It is well-known that in WD, TW is found in high numbers within small intestinal LP macrophages. This can be demonstrated either by indirect immunofluorescence using an antibody recognizing TW or by PAS staining of the duodenal mucosa of patients with established WD (Figure 1A,B). To determine whether this is secondary to transepithelial bacterial passage and to further elucidate the underlying mechanisms as well as the effects exerted on the intestinal barrier, we sought to establish models that could be used to improve our understanding of epithelial TW transmigration. First, we mounted duodenal mucosa from endoscopic biopsies of patients with a healthy duodenum to miniaturized Ussing chambers as previously described and exposed the small intestine by adding a high infectious dose of TW (10 8 bacteria per chamber) to the luminal Ussing compartment for up to 6 h [21]. Although the time of exposure was short compared to the presumed pathophysiology of WD, in which TW might be present close to the small intestinal mucosa for much longer periods of time, microscopy revealed sporadic TW rods by immunostaining as well as PAS staining ( Figure 1C-F). The number of stained rods was considerably lower in the samples exposed to TW ex vivo (Ussing chamber) compared to the mucosa of patients with established WD ( Figure 1C-F vs. Figure 1A,B). This was an expected finding, given the relatively short period of TW exposure in the Ussing experiment compared to the length of exposure in an individual infected with TW. Interestingly, the rods were localized to LPantigen-presenting cells (HLA-DR-positive) that were either CD68-positive macrophages or DC-SIGN-, CD11c-, and S100-positive dendritic cells ( Figure 1E,F). Second, we established a cell-culture-based model by apical infection of IEC lines with TW ( Figure 1G-J). Beginning 3 to 6 h after infection and increasing over time (maximum length of experiment 24 h), immunostained TW rods were identified within IECs. Of note, the pattern of infection was heterogeneous, with clusters of TW-positive IECs found in between a large number of TW-uninfected IECs. IECs infected with a high bacterial load were frequently subject to apical cell shedding ( Figure 1H). uninfected IECs. IECs infected with a high bacterial load were frequently subject to apical cell shedding ( Figure 1H). healthy duodenal mucosa that were luminally exposed ex vivo to TW in Ussing chambers for 3 h. TW was detected within LP (arrows in (C,D)). Counterstaining for epithelia ((E)-cadherin, white) and antigen-presenting LP cells (HLA-DR, red). PAS-positive material within LP cells (arrows, (D)). TW (green) within the epithelial layer (arrow), within HLA-DR-positive cells (asterisk), and intercellularly within the LP (arrowhead). (F): detail of (E). (G-J): IF-stained sections with single confocal XY-planes of Caco-2 (G,H) and IPEC monolayers ((I), detailed in (J)) that were apically exposed to TW for 24 h. Red: Cytoskeletal F-actin. TW was found intra-and subepithelially (F) and accumulated in rounded cells that were shed to the apical compartment (H). Size bars, 20 µm (A-F), 10 µm (G-J).

Mechanism of Transepithelial Passage of T. whipplei
To obtain a better understanding of the mechanism TW utilizes to pass the epithelial barrier, we aimed to inhibit the uptake of TW by IECs. Since actin facilitates all types of endocytosis, we applied the inhibitors of actin polymerization latrunculin and cytochalasin D, which reduced TW uptake by 59% and 50%, respectively (Figures 2A and S1). Accordingly, aggregations of F-actin were visualized around internalized TW ( Figure 2B). Moreover, genistein, an inhibitor of tyrosine kinases that was previously shown to inhibit caveolin-dependent endocytosis did not reduce bacterial uptake (Figures 2A and S1) [22]. In a next step, we aimed to specifically block distinct endocytic pathways that may contribute to TW uptake via the expression of dominant-negative mutants. The expression of the GTPase-deficient dynamin mutant K44A substantially reduced TW uptake by IECs, suggesting a prominent role for the large GTPase dynamin in this process ( Figures 2C and S2). In contrast, the expression of a dominant-negative Eps mutant (GFP-ED95/295-Eps) that targets clathrin-dependent endocytosis did not result in significant changes in TW uptake ( Figures 2C and S2) [22,23]. Subcellular fractionation by ultracentrifugation revealed TW was localized to caveolin-1-and caveolin-2-positive fractions ( Figure S3). Furthermore, TW was found within fractions that were positive for markers of early endosomal (rab5), late endosomal (rab7), and lysosomal (cathepsin) compartments ( Figure S3).
Next, we transfected two distinct C-terminal caveolin fragments in IECs: D82-caveolin lacks the capacity to activate Ras, and K135-caveolin is a strong inhibitor of caveolin-based endocytosis ( Figure S4) [24]. Interestingly, only the expression of D82-caveolin was sufficient to block TW endocytosis, arguing for a role of a Ras GTPase in the endocytic uptake of TW and against a direct uptake by caveolae as found for SV40 endocytosis, as this would be inhibited by the expression of K135-caveolin ( Figure 2D) [22]. Furthermore, the immunostaining of endogenous IEC caveolin-1 ( Figure 2E) and caveolin-2, as well as live cell experiments visualizing the TW uptake process of IECs expressing caveolin-mRFP, revealed only rare colocalization events of caveolin and TW ( Figure 2E and videos S1 and S2). These colocalizations were mostly found in relatively large perinuclear organelles and never in membrane-associated caveolin foci, suggesting that the initial uptake did not occur through classical caveolae. Since caveolin-based uptake has previously been shown to be associated with the activity of small GTPases, cells were pretreated with the Ras inhibitor FTS and the Rac inhibitor EHT-1864, both of which proved to be potent inhibitors of TW uptake, further supporting a role for small GTPases in the endocytic uptake of TW ( Figures 2F and S5) [25]. Cytoskeletal reorganization and actin-driven protrusions, facilitated by the activation of the Ras/Rac1-pathway, are mediated by PI3K [26,27], whose inhibition with LY294002 decreased TW uptake. Of note, the caveolin-dependent TW uptake by IECs might be subject to positive feedback, as caveolin-1 and caveolin-2 protein levels were found to be upregulated after 24 h of IEC exposition to TW (Figures 2G and S6). When IECs exposed to TW were examined by transmission electron microscopy (TEM), intact TW rods were identified within TW-containing vacuoles that did not fulfil the TEM criteria for caveolae ( Figure 3A). Of note, actin polymerization and membrane ruffling occurred at the site of TW entry ( Figure 3A). the TEM criteria for caveolae ( Figure 3A). Of note, actin polymerization and membrane ruffling occurred at the site of TW entry ( Figure 3A).
To examine the occurrence of endocytic TW uptake in patients with active WD, the duodenal mucosa of WD patients was double-stained for TW and endocytic markers. Here, TW revealed a robust colocalization with the late endosomal and lysosomal markers Lamp-1 and cathepsin ( Figure 3B,C). Of note, endocytic TW uptake was restricted to viable bacteria ( Figures S7 and S8).  respective wild-type constructs. (E): Immunofluorescence/confocal imaging of TW (green) and endogenous caveolin (red) of Caco-2 cells exposed to TW. (F): Inhibition of TW uptake in Caco-2 cells by inhibitors of small GTPases: FTS (20 µM) and EHT-1864 (5 µM; n = 16). (G): Quantification of Western blot of total lysates for caveolin-1 and caveolin-2 after exposure of Caco-2 cells to TW (n = 5-6). (A,C,D,F,G): Results are expressed as single values with median. Dispersion is expressed as minimum to maximum. Significance levels were corrected for multiple comparisons. * p < 0.05, ** p < 0.01, *** p < 0.001. Size bars, 20 µm.

Epithelial Barrier in T. whipplei Infection
Next, we aimed to elucidate the effect of TW on intestinal barrier function. The transmural electrical resistance (R) was analyzed by mounting healthy human duodenal mucosa on Ussing chambers. Exposure to TW caused a moderate (12%), but significant, reduction in R, which was not found in parallel experiments using E. coli K12 as control bacteria (noninvasive and nonpathogenic bacterium) or when heat-inactivated TW were used instead (Figures 4A,D and S9 and E. coli K12. To identify the mechanism underlying To examine the occurrence of endocytic TW uptake in patients with active WD, the duodenal mucosa of WD patients was double-stained for TW and endocytic markers. Here, TW revealed a robust colocalization with the late endosomal and lysosomal markers Lamp-1 and cathepsin ( Figure 3B,C). Of note, endocytic TW uptake was restricted to viable bacteria ( Figures S7 and S8).

Epithelial Barrier in T. whipplei Infection
Next, we aimed to elucidate the effect of TW on intestinal barrier function. The transmural electrical resistance (R) was analyzed by mounting healthy human duodenal mucosa on Ussing chambers. Exposure to TW caused a moderate (12%), but significant, reduction in R, which was not found in parallel experiments using E. coli K12 as control bacteria (noninvasive and nonpathogenic bacterium) or when heat-inactivated TW were used instead (Figures 4A,D and S9 and E. coli K12. To identify the mechanism underlying TW-associated barrier leaks, we exploited the IEC cell culture model. Importantly, all IEC lines tested showed reduced transepithelial resistance (TER) values upon apical exposure to TW ( Figure 4B). Most of the cell lines revealed a scatter-like pattern of TW uptake. However, HT-29/B6 cells, known to secrete large amounts of apical mucus, revealed patches of numerous apical TW rods [28]. To determine whether IEC-secreted mucus might contribute to the capacity of TW to adhere to IECs, we biochemically removed the mucus using the secretolytically active drug N-acetylcysteine (NAC). Treatment with NAC removed the characteristic TW patches from the HT-29/B6 monolayer ( Figure 4C,E).
Although NAC pretreatment was insufficient to rescue the TER decrease in HT-29/B6 cells, it ameliorated the barrier defect found in human intestinal mucosae mounted on Ussing chambers ( Figure 4B,F).
lines tested showed reduced transepithelial resistance (TER) values upon apical exposure to TW ( Figure 4B). Most of the cell lines revealed a scatter-like pattern of TW uptake. However, HT-29/B6 cells, known to secrete large amounts of apical mucus, revealed patches of numerous apical TW rods [28]. To determine whether IEC-secreted mucus might contribute to the capacity of TW to adhere to IECs, we biochemically removed the mucus using the secretolytically active drug N-acetylcysteine (NAC). Treatment with NAC removed the characteristic TW patches from the HT-29/B6 monolayer ( Figure 4C,E). Although NAC pretreatment was insufficient to rescue the TER decrease in HT-29/B6 cells, it ameliorated the barrier defect found in human intestinal mucosae mounted on Ussing chambers (

Epithelial Apoptosis Due to T. whipplei Exposure
Next, we hypothesized that the TW-triggered apoptosis of IECs might contribute to the barrier defect described above. We thus quantified activated caspase-3 in epithelia exposed to TW and found an 11-fold induction, which was abrogated when cells were pretreated with an apoptosis inhibitor ( Figure 5A,D). Interestingly, proliferation was decreased upon TW exposure ( Figure 5B,D). The inhibition of apoptosis ameliorated the barrier defect. However, a substantial portion of the barrier defect appeared to be unrelated to apoptosis since some barrier defect remained although the inhibition of apoptosis appeared to be complete ( Figure 5A,C). Interestingly, confocal microscopy as well as TEM revealed that IECs with a high load of TW were particularly prone to undergoing apoptosis-associated cell death (Figures 1H and 5E). Furthermore, TW uptake was more frequently found in apoptotic IECs compared to nonapoptotic cells ( Figure S10). Thus, these results suggested that TW-packed IECs in particular undergo apoptosis, thereby contributing to an epithelial barrier defect.
Next, we hypothesized that the TW-triggered apoptosis of IECs might contribute to the barrier defect described above. We thus quantified activated caspase-3 in epithelia exposed to TW and found an 11-fold induction, which was abrogated when cells were pretreated with an apoptosis inhibitor ( Figure 5A,D). Interestingly, proliferation was decreased upon TW exposure ( Figure 5B,D). The inhibition of apoptosis ameliorated the barrier defect. However, a substantial portion of the barrier defect appeared to be unrelated to apoptosis since some barrier defect remained although the inhibition of apoptosis appeared to be complete ( Figure 5A,C). Interestingly, confocal microscopy as well as TEM revealed that IECs with a high load of TW were particularly prone to undergoing apoptosis-associated cell death ( Figures 1H and 5E). Furthermore, TW uptake was more frequently found in apoptotic IECs compared to nonapoptotic cells ( Figure S10). Thus, these results suggested that TW-packed IECs in particular undergo apoptosis, thereby contributing to an epithelial barrier defect.

Uptake of Apoptotic, T.-whipplei-Bearing Intestinal Epithelial Cells by Macrophages
Thus far, we have presented evidence that (i) TW is taken up by epithelial cells via a distinct endocytic process and (ii) TW-bearing IECs are likely to undergo apoptosis, thereby causing a barrier leak. Hypothetically, both processes could together promote the final translocation of TW into the subepithelial compartment. TW-containing IECs undergoing apoptosis could not only shed into the intestinal lumen but also be subject to phagocytosis by intestinal macrophages. Thus, we used a modified version of our previously published coculture model with Caco-2 cells and in vitro differentiated human macrophages ( Figure S11) [29]. Caco-2 Actin cells were seeded on filter transwells with macrophages attached to the well bottom. After allowing the Caco-2 Actin cells to form an epithelial layer, the cells were exposed to TW for 24 h. To elucidate whether TW together with IEC (apoptotic) material was taken up by subepithelial macrophages, confocal imaging was performed (Figures 5F and S12). Most macrophages were found to be positive for actin-GFP when seeded next to Caco-2 Actin cells, suggesting a regular uptake of IEC material by macrophages (Figures 5F and S5A). As a negative control, macrophages incubated with native Caco-2 cells did not reveal intracellular actin-GFP ( Figure S12B). However, the same macrophages that showed IEC remnants also contained TW material when exposed to TW. This result revealed the capacity of macrophages to take up TW and IECs simultaneously.

Discussion
WD and TW-induced gastroenteritis can develop in individuals exposed to TW after the uptake of this bacteria into the intestinal mucosa. Since the mechanisms that enable TW to enter the intestinal mucosa and the pathway TW utilizes to pass the IEC layer has not yet been characterized, we conducted a study using ex vivo human intestinal mucosa as well as genetically altered IECs to elucidate both the underlying mechanism of TW uptake and its impact on the epithelial barrier.
The mucosal barrier is a complex structure that shields the mucosa from the unlimited passage of luminal microorganisms and toxins into the system while regulating the uptake of water, electrolytes, and nutrients. It includes the mucus layer, which contains bicarbonate, mucins, and antimicrobial peptides, and the epithelial barrier, which is determined by the integrity of the IECs, the sealing of the paracellular space by a delicately regulated apical junctional complex, and the activity of the transcellular route, which is capable of transporting macromolecules as well as particles the size of microorganisms [30,31]. It is further regulated by a heterogeneous set of LP cells, including mucosal macrophages and dendritic cells that can, through their interaction with IECs, focally alter barrier function or even collect luminal antigen by extending transepithelial dendrites into the epithelium [29,32,33].

Transcellular Route of T. whipplei
Several pathogenic microorganisms, including Salmonella, Escherichia, Shigella, Yersinia, Campylobacter, Pseudomonas, and Vibrio species, have developed strategies to overcome this barrier, thereby invading the human organism. The preferred entry site along the gastrointestinal tract varies with the microorganism of interest, with Salmonella preferentially invading Peyer's patches in the ileal mucosa, Campylobacter invading small intestinal and colonic mucosa, and Shigella and Escherichia species being optimized to enter via the colonic mucosae [34][35][36]. In contrast to these pathogens, in gastrointestinal WD, TW is mostly found in small intestinal or duodenal macrophages or dendritic cells. However, infected macrophages can also be detected in the large intestinal mucosa, although in much lower numbers than in the small intestine [2]. Thus, we proposed IECs to be the primary entry site for TW bacteria, which was supported by the autonomous entry of TW in IECs in vitro, the immunostaining of duodenal tissue of WD patients, and the engulfment of TW by epithelial cells and intestinal macrophages when human duodena were exposed to TW ex vivo.
A further question relates to the entry route bacteria use to trespass the IEC layer: transcellular or paracellular. The immunostaining of human mucosae as well as the immunostaining and TEM of IEC cell lines revealed frequent intracellular TW but no paracellular TW. Thus, we concluded that TW translocated via endocytosis rather than via the facilitated permeabilization of apical junctions. This was further supported by the finding that we could block TW uptake by the inhibition of apical endocytosis. Similarly, a number of intestinal bacterial pathogens directly enter IECs, which subsequently pass the microbes on to a subepithelial compartment by either a basal secretion of bacteria or the phagocytosis of affected IECs through subepithelial mononuclear cells [35,37]. In contrast, for some bacteria, including Campylobacter and hemolysin-positive E. coli, the invasion of the intestinal mucosae was shown to occur via a paracellular pathway, by either a controlled entry through tight junctions or focal leaks [37][38][39]. Although we did not detect any paracellular TW, we cannot rule out this route completely, as the effective inhibition of IEC apoptosis only partially reversed the effect on the IECs' barrier function.

Mechanism of Endocytic Uptake
The next question arising is: What is the molecular mechanism underlying TW uptake? Our study suggests a mechanism dependent on the actin cytoskeleton, since latrunculin and cytochalasin D were both able to reduce TW uptake, as well as on the large GTPase dynamin, since the expression of a dominant-negative dynamin mutant reduced TW uptake. Dynamin is known to pinch off vesicles from the cell membrane in a GTP-dependent manner via its mechanoenzyme activity [40]. Of note, it has been shown that dynamin is upstream of rab5, which is required for early endosome formation [41].The dynamin dependency of bacterial uptake by IECs has previously been shown for Listeria monocytogenes and Salmonella spp. [42,43]. Furthermore, based on the results of our experiments using the GFP-ED95/295-Eps mutant, we propose that clathrin-coated vesicle formation does not play any role in TW uptake. Instead, TW endocytosis might be affected by caveolin function, not caveolae as such, which is supported by the analysis of TW uptake after the expression of two different caveolin fragments. TW endocytosis was not affected by the K135-caveolin fragment known to strongly affect caveolin-mediated endocytosis (which is not surprising, since TEM imaging showed no apical surface caveolae and TW is far too large for endocytosis by a single caveola) but was impaired by the expression of the longer D82-caveolin fragment, which selectively impairs the activation of membrane-associated Ras [24]. Thus, the distinct function of caveolin might be to activate signaling via small GTPases, which then act on the actin cytoskeleton [44,45]. Further evidence for this interpretation results from experiments using two inhibitors of small GTPases (FTS and EHT-1864) that reduced TW endocytosis to a significant extent and from the experiments using actin inhibitors. Previously published specificity data of these GTPase inhibitors suggested that Rac1 rather than Ras GTPase was actually inhibited [46,47]. Interestingly, it has been shown only recently that Rac1 activity must be tightly controlled during the process of macropinocytosis and that this process is preceded by the activation of caveolin-1 [48,49]. Moreover, the activation of the small GTPase Rac1 by caveolin has previously been shown to play a role in the invasion of Salmonella typhimurium in a HeLa cell model [25]. Therefore, our data in combination with previously published data suggest a mechanism that includes the activation of Ras and Rac1 (and, subsequently, dynamin), with caveolin acting as a scaffold in the IEC membrane ( Figure 6).
Within macrophages, TW creates a niche by manipulating steps of phagosomelysosome maturation, thereby facilitating its replication in an acidic environment, which is critical for its prolonged persistence in host cells during WD [50][51][52]. We suggest a similar mechanism in IECs, since several proteins derived from the endolysosomal pathway are enriched in the TW-containing vacuole. Our data suggest that the passage of TW through IECs seems to be related to phagocytosis or macropinocytosis.
The focus of the current study was to identify host-related mechanisms that allow TW to pass the human intestinal barrier. One further exciting question not answered by this study relates to virulence factors TW might use to trigger its own uptake by the IEC layer. The sequencing of the TW genome has uncovered two genes, TWT_701 and TWT_189, that were grouped as putative type-II/type-IV pathway secretion systems (T4SS) by a conserved domain analysis [53]. Notably, TW has a reduced genome and type-II/type-IV secretion systems are complex and large protein complexes [53]. Type-IV secretion systems have been identified in Helicobacter pylori, Neisseria meningitides, and Pseudomonas aeruginosa and consist of pili through which bacteria insert proteins into the host cell. Interestingly, it was shown at a molecular level that Neisseria passed the blood-brain barrier by inducing a T4SSdependent switch in endothelial cell polarity [54]. Similarly, the passage of Pseudomonas aeruginosa through an IEC layer is facilitated by type-IV pili [55]. Thus, the two TWT genes might be an interesting starting point for future research on pathogen-related factors facilitating the TW penetration of the intestinal epithelium. The current assumption is that TW enters the host via the small intestinal mucosa, where it is found in mucosal macrophages. Mucosal inflammation results in mucosal flattening and malabsorption of nutrients. Disease progression can involve multiple organs, including joints (arthritis), the heart (endocarditis), or central nervous system (cerebral Whipple). Interestingly, TW is also a causative pathogen of transient gastroenteritis, as it was found in 15% of children with gastroenteritis. Thus, it not only accounts for severe pathologies but is also a prevalent pathogen. Nevertheless, the mechanism TW uses to invade and damage the small intestinal mucosa is incompletely understood. In the present work, we used model systems for Whipple's disease to reveal the initial uptake mechanism of TW by intestinal epithelial cells (IECs). Specifically, IECs were shown to endocytose TW by a mechanism involving actin polymerization, the large GTPase dynamin, caveolin, and the small GTPases Ras/Rac1. Although caveolin was involved, it functioned as a signaling molecule inducing endocytosis by large vesicular structures rather than as the structural component of caveolae. Furthermore, heavily TW-invaded IECs underwent apoptosis, thereby causing an epithelial barrier defect (as a putative mechanism of acute gastroenteritis due to apoptosis-induced leak flux diarrhea). Consecutively, apoptotic IECs were shown to be subject to uptake by lamina propria macrophages and dendritic cells. Predisposed hosts with insufficient bacterial clearance mechanisms due to persistent exposure, specific immunological, and genetic alterations will develop Whipple's disease.
Within macrophages, TW creates a niche by manipulating steps of phagosome-lysosome maturation, thereby facilitating its replication in an acidic environment, which is critical for its prolonged persistence in host cells during WD [50][51][52]. We suggest a similar mechanism in IECs, since several proteins derived from the endolysosomal pathway are enriched in the TW-containing vacuole. Our data suggest that the passage of TW through IECs seems to be related to phagocytosis or macropinocytosis.
The focus of the current study was to identify host-related mechanisms that allow TW to pass the human intestinal barrier. One further exciting question not answered by this study relates to virulence factors TW might use to trigger its own uptake by the IEC layer. The current assumption is that TW enters the host via the small intestinal mucosa, where it is found in mucosal macrophages. Mucosal inflammation results in mucosal flattening and malabsorption of nutrients. Disease progression can involve multiple organs, including joints (arthritis), the heart (endocarditis), or central nervous system (cerebral Whipple). Interestingly, TW is also a causative pathogen of transient gastroenteritis, as it was found in 15% of children with gastroenteritis. Thus, it not only accounts for severe pathologies but is also a prevalent pathogen. Nevertheless, the mechanism TW uses to invade and damage the small intestinal mucosa is incompletely understood. In the present work, we used model systems for Whipple's disease to reveal the initial uptake mechanism of TW by intestinal epithelial cells (IECs). Specifically, IECs were shown to endocytose TW by a mechanism involving actin polymerization, the large GTPase dynamin, caveolin, and the small GTPases Ras/Rac1. Although caveolin was involved, it functioned as a signaling molecule inducing endocytosis by large vesicular structures rather than as the structural component of caveolae. Furthermore, heavily TW-invaded IECs underwent apoptosis, thereby causing an epithelial barrier defect (as a putative mechanism of acute gastroenteritis due to apoptosis-induced leak flux diarrhea). Consecutively, apoptotic IECs were shown to be subject to uptake by lamina propria macrophages and dendritic cells. Predisposed hosts with insufficient bacterial clearance mechanisms due to persistent exposure, specific immunological, and genetic alterations will develop Whipple's disease.

Barrier Defect Caused by T. whipplei
Another question to address was the integrity of the small intestinal mucosa and the IEC layer following contact with and entry of TW. By analyzing the ex vivo Whipple model and by conducting TER studies on various IEC cell lines, we presented evidence that TW impaired epithelial barrier function via its potential to induce IEC apoptosis. Notably, this might also explain our observation that inhibition of endocytic pathways not completely prevented the transepithelial passage of TW. It has been shown that TW also induces apoptosis in macrophages via caspase-3 [56]. Naturally, a reduction of barrier function has been found for other pathogens, including enteropathogenic E. coli (EPEC), Campylobacter, and Yersinia enterocolitica [57][58][59]. The barrier defect described in our ex vivo model occurred after a short exposure (hours) of intestinal epithelia to TW and thus might not reflect the mechanism of diarrhea found in WD [60,61]. However, since the finding suggests that the barrier defect might result in a leak flux, it might play an important role in TW-induced gastroenteritis [7]. Leak flux diarrhea has previously been identified as a causative mechanism in various enteropathies, including celiac disease, lambliasis, norovirus infection, and HIV enteropathy [62]. Interestingly, we found some evidence for the mucus layer having a role in facilitating TW adherence to the epithelial layer and a consequent reduction in barrier function. Similarly, Campylobacter jejuni was shown to adhere to intestinal mucus and even propagate within the mucus, as well as to impair barrier function [63]. This highlights that the ability of intestinal pathogens to interact with the mucus layer is an independent determinant for pathogen virulence.
Is the modulation of barrier function also related to the TW uptake mechanism discussed above? It has previously been proposed that pathogens entering the intestine might invade the mucosa primarily by entering IECs, which then become apoptotic and are phagocytosed by LP monocytic cells. This might be of specific relevance to WD, since its hallmark is the occurrence of mucosal macrophages harboring TW. Applying TW on the coculture model that we had previously established suggests that this might in fact hold true for TW uptake as well, since both epithelial and TW components were found to be enriched in the subepithelial macrophages [29]. However, further studies will be necessary to further characterize this specific aspect.
The transmission of TW via the fecal-oral route is likely. However, the ability of TW and the mechanism by which the bacteria pass the intestinal epithelial barrier are currently unknown. Hence, we used static models (with inherent limitations) to study selected aspects of TW infection.
In summary, we propose a transepithelial uptake mechanism for TW into the small intestinal mucosa that includes dynamin-and caveolin-dependent, but not clathrin-dependent, endocytic mechanisms. TW-harboring IECs might undergo apoptosis, which then leads to uptake by LP macrophages and a defective epithelial barrier ( Figure 6). The latter presumably contributes to TW-associated diarrhea by a leak flux mechanism.

Constructs
Human wild-type and K44A-dynamin1 were subcloned from pcDNA3 constructs into pEGFPN1 via the EcoR1 and BamH1 sites [64]. GFP-fused wild-type and GFP-ED95/295 mutant Eps constructs were a gift from Alexandre Benmerah (Université Paris Descartes, France) [23]. Caveolin1-mRFP and -GFP constructs were a gift from Lucas Pelkmans (University of Zürich, Switzerland) [22]. D82-and K135-caveolin constructs were generated in analogy to the study by Roy et al. using a PCR-based strategy and subcloned into an mRFPmodified pEGFPN1 vector via EcoR1 and Mfe1 sites ( Figure S1) [24]. When caveolin mRFP fragments were expressed together with wild-type caveolin-EGFP, distinct differences in cellular distribution and motility were found, as previously described (Video S3A,B) [24].

Bacterial Strains and Preparations
T. whipplei Twist Marseille (CNCM I-2202) was cultured in axenic medium as described previously and used as viable or heat-killed bacteria (100 • C, 90 min) [14,51,56]. Prior to infection, bacteria were resuspended in cell-line-specific medium.

Cell Culture and Transfection
The

Inhibition of Endocytic T. whipplei Uptake
For the inhibition of TW uptake, latrunculin A (0.2 µM), cytochalasin D (20 µM), and genistein (25 µM), as well as the Rac-GTPase inhibitor EHT 1864 (5 µM), the Ras-GTPase inhibitor FTS (20 µM), and the phosphoinositide 3-kinase (PI3K) inhibitor LY294002 (25 µM) were added to the apical compartment. For secretolysis, NAC (10 mM) was added to the apical compartment of HT-29/B6 cells. Cells transiently transfected with constructs containing dominant-negative endocytosis mutants coupled to EGFP were exposed to TW. The transient transfection of Caco-2 was performed using the Amaxa Nucleofector and Kit T (Lonza, Basel, Switzerland) according to the manufacturer's instructions. The uptake of TW by Caco-2 cells was assessed by flow cytometry and expressed as the mean fluorescence intensity normalized to respective controls ( Figure S13). Trypsinized and filtered cells were washed with phosphate-buffered saline containing 10 mM EDTA and 0.5% bovine serum albumin to remove TW attached to the cell surface, fixed in paraformaldehyde, and permeabilized with 0.5% Triton-X and saponin before staining. IEC cell suspension was analyzed by flow cytometry with rabbit-anti-TW (provided by Didier Raoult, Institut Hospitalo-universitaire Méditerranée Infection, Marseille, France; Aix Marseille Université, Institut de Recherche pour le Développement, Assistance Publique-Hôpitaux de Marseille, Microbes Evolution Phylogénie et Infections, Marseille, France), and visualized using the appropriate secondary donkey antirabbit antibody linked to Alexa488 or Alexa 647 (Life Technologies, 1:500). Data were acquired on a FACSCalibur or FACSCanto II (both BD Biosciences) and analyzed with FlowJo software (FlowJo, Ashland, OR) (for the gating strategy, see Figures S1, S2, S5, S8 and S10 in Supplementary Materials). The TW-specific antibody is validated and used for routine clinical diagnostics [15,65]. The monolayer experiments have revealed that IECs are infected infrequently together with a heterogeneous pattern of TW-infected cells after 24 h. Only few cells engulf TW and the amount of TW per cells varies markedly. This heterogeneity, together with the infrequency, explains that no clear distinct TW infected population can be separated.

T. whipplei Translocation in a Caco-2/Macrophage Coculture Model
Monocytes were isolated from human peripheral blood using CD14 magnetic cell sorting (MACS, Miltenyi Biotech, Bergisch-Gladbach, Germany) and differentiated in vitro to macrophages by the addition of granulocyte-macrophage colony-stimulating factor (500 U/mL; Bayer, Berlin, Germany) to generate M1 macrophages, as previously published [29]. Hanging filter inserts (Nunc) were placed in 12-well plates with the lower filter side facing up. Caco-2 Actin cells (and for controls, nontransfected Caco-2 cells) were seeded on the lower filter side and were allowed to attach for 8 h (200,000 cells per filter insert, MEM with GlutaMax). Filter inserts were then transferred to a 24-well plate (now with the upper filter side facing up) to seed macrophages (1.5 × 10 5 cells per filter inserts) ( Figure S11). On culture day 14, TW was added to the apical compartment (MOI = 200) (= infection via the basolateral side). After 24 h of TW exposure, filters were washed and subjected to fixation (4% PFA, 15 min) and permeabilization (0.5% Triton X-100, 5 min). Cytospins of the cells of the macrophage compartment were prepared with a Cytospin 2 Cytofuge (Thermo Shandon, Frankfurt, Germany) at 100× g for 7 min, air-dried, fixed in 1% PFA for 60 min at 4 • C, and stored at −80 • C.
Semithin sections were stained with Richardson's solution. Ultrathin sections were stained with uranyl acetate and lead citrate and analyzed using a Zeiss EM 906 electron microscope.

Live Cell Microscopy
Caco-2 cells were grown on Lab-Teks (8-well-Lab-Tek™ II chambered coverglass, ThermoScientific) and transfected with either caveolin1-EGFP or D82-and K135-caveolin-mRFP fragments, as stated above. TW exposure was performed in a Hepes Ringer solution (pH 7.4, with the addition of 10 mM D-glucose) at an MOI of 100. TW was visualized via in vivo staining (BacLight stain-488, Molecular Probes). Live cell imaging was performed during the 12 h following TW exposure using a Zeiss LSM 780 equipped with a PM S1-Incubator controlled by a computer-assisted heating system (Incubation system S, Zeiss).

Analysis of Barrier Function of Small Intestinal Mucosae
Duodenal biopsy specimens of individuals without clinical signs of enteropathy were obtained during routine endoscopy from 8 patients (5 males, 3 females; age range 28-75 years, mean 54.7 years). Mucosae were mounted to the miniaturized Ussing chamber system as previously described [21,67]. Resistance values determined by a computerized automatic clamp device (Fiebig Hardware & Software, Berlin, Germany) were corrected for the resistance of the bathing fluid between the voltage-sensing electrodes and the empty filter. As a bathing solution, 10 mL of DMEM/F-12 Ham was used per chamber side. After 30 min of equilibration, bacteria (10 7 per chamber) were added to the mucosal compartment and resistance was monitored for up to 3 h. To reduce mucus levels, mucosae were pretreated with NAC (10 mM) for 15 min. Biopsies from patients with WD were obtained at the time of diagnosis.

Purification of the Intracellular Compartment Containing T. whipplei
For fractionation, Caco-2 cells that had been exposed to TW were lysed in 500 mM Na 2 CO 3 , Dounce homogenized (Wheaton Science, 20 strokes on ice), and homogenized further using an Ultra Turrax (3 × 10 s) [68]. A sucrose density gradient was established in a 12 mL Beckman ultracentrifugation tube by mixing 2 mL of the homogenate with 2 mL of 90% sucrose, followed by applying 4 mL 35% sucrose and then 4 mL 5% sucrose on top. Ultracentrifugation was performed at 54,000 rpm (20 h, 4 • C, Beckman's T70.1Ti-rotor), which is equivalent to a maximum relative centrifugal force of approx. 240,000× g. After centrifugation, 500 µL fractions were carefully collected to be later analyzed by Western blotting.

Statistical Analysis
Results were expressed as median, and the dispersion of data is expressed as first and third quartiles with minimum and maximum. All analyses were performed using GraphPad Prism version 9.3.1 software. After performing normality testing, single comparisons were assessed using either a two-tailed, unadjusted, unpaired Student's t-test or a Mann-Whitney test. Differences among groups were analyzed with an analysis of variance, followed by the Bonferroni adjusted t-test or Kruskal-Wallis test with Dunn's test to correct for multiple comparisons. p-values <0.05 were considered statistically significant.

Conclusions
TW enters IECs via an actin-, dynamin-, caveolin-and Ras-Rac1-dependent endocytosis mechanism and consecutively causes IEC apoptosis primarily in IECs invaded by multiple TW bacteria. This results in a barrier leak. Moreover, we propose that TW-packed IECs can be subject to phagocytic uptake by macrophages, thereby opening a potential entry point of TW into intestinal macrophages. Insufficient bacterial clearance in a predisposed host will cause WD and more frequently, transient gastroenteritis.