RETRACTED: Endocytic recycling and vesicular transport systems mediate transcytosis of Leptospira interrogans across cell monolayer

Thank you for submitting your article "Endocytic recycling and vesicular transport systems mediate transcytosis of Leptospira interrogans across cell monolayer" for consideration by eLife. Your article has been reviewed by Gisela Storz as the Senior Editor and two reviewers, one of whom is, a Reviewing Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

The manuscript describes experiments addressing the question by which mechanism Leptospira interrogans, a bacterial pathogen of humans, is capable of undergoing transcytosis. Transcytosis of pathogenic bacteria (such as ListeriaListeria monocytogenes, Salmonella enterica, Escherichia coli or Streptococcus pneumoniae) is a crucial step in dissemination throughout a host during infection. Transcytosis of human endothelial cells by L. interrogans has previously been demonstrated in vitro but the nature of the pathway remained uninvestigated. Using a combination of immunocytochemistry, pharmacological treatments, and knock-downs the authors provide evidence that the pathogen is endocytosed by caveolae, stored in a Rab5-positive early endosomal compartment from which it is re-exocytosed involving a Rab11 positive recycling compartment, exocyst components and the exocytotic SNAREs VAMP2 and syntaxin1. Using classical transwell experiments the authors also show that endo- and exocytosis can occur at different cell poles, explaining how the bacterium is able to cross endothelial barriers.

Essential revisions:

1) Changes in fluorescence intensity. For many proteins, the fluorescence intensity after immunostaining appears to be massively upregulated after infection. However, it is not clear why this is the case, i.e. whether this is due to relocalization (e.g. recruitment from the cytoplasm to pathogen-containing vacuoles) or to an increase in expression levels. Moreover, fluorescence intensity is a tricky parameter since it is subject to photobleaching and requires that all samples to be compared are analysed in the same field of view.

To clarify this important issue, we are suggesting the following experiments:

a) show immunoblots of the proteins before and at different timepoints after infection. In our opinion, it is sufficient to do this for one or two of the cell lines. Also, include control proteins (not only actin but other proteins involved in membrane traffic, e.g. LAMP1) in order to clarify if the effects (if any) are selective.

b) In the fluorescence images (particularly Figure 3 and Figure 4) please show control stainings before infection, and/or very early time points. This is essential in order to understand the changes caused by the formation of the transcytotic and pathogen-containing endosomes.

2) In Figure 5, you normalize the number of exocytosed bacteria to the number of cells. However, the reduction may also be due to a reduced uptake in the knockdown or toxin-treated cells, and in this case the defect may not be in exocytosis but rather in uptake or even cell viability. It should be checked, at least for some of the conditions, whether uptake/infection is affected by the treatments. It appears that in Figure 5E-H this is only shown for cells that were not transfected with siRNAs or toxins.

3) Transcytosis is studied by adding bacteria to a completely confluent cell monolayer and the number of traversed bacteria is quantified. Any breach in the monolayer will lead to bacterial movements between mammalian cells. The authors have consequently determined the trans-epithelial(endothelial) electric resistance (TEER) as a sensitive means to uncover such breach. As documented in Figure 6C, the TEER was measured as about 200 ohm/cm2 in all experiments. This result is surprising in two respects: (i) The value is not very high, given that the device with media only will present about half that resistance (Barocchi et al., cited here), although some cell types may actually be confluent at this value; (ii) there is a vast variability in the precise TEER values which the literature indicates for different cell types reaching up to 4,000 ohm/cm2. It is striking that in this manuscript, six types of immortalized cells covering two different cell types and two different donor species have all the exact same TEER of around 200 ohm/cm2 (Figure 6C). If one therefore entertains the possibility that the bilayers were not really tightly sealed and that the motile bacteria would travel beyond the mammalian cell layer, one would expect that, after some time, 50% of the bacteria would be found in one Transwell compartment and 50% in the other, regardless to which side the bacteria were originally added. This is exactly what is observed here with all types of mammalian cells (Figure 6B). This result is also surprising in another way: Whichever receptor is used for the entry of L.i., it would be expected to be present at different concentrations on apical versus basolateral membrane domains and therefore equal rates of transcytosis in both directions (as described here) seems little likely even with sealed cell layers.

We understand that the data in Figure 7 support the notion that the layers are sealed since „transcytosis" is reduced when certain drugs or siRNAs are used. Still, it is of crucial importance to demonstrate the presence of a sealed monolayer which can, in addition to using TEER, simply be studied by testing the flux of FITC-dextran, horseradish peroxidase, inuin, radioactive sucrose or similar compounds between the Transwell compartments through assumptive sealed layers (there are numerous examples in the Transwell literature). Such control experiments are also important as a previous paper by Martinez et al. (cited by the authors) concluded that there was “significant disruption of endothelial cell layers” during infection with leptospira.

Along the same lines, we strongly suggest a second, simple control experiment which seems required based on the fact that all siRNA and drug treatments lead to the exactly same inhibition of transcytosis over time (Figure 7) and given the above concerns with sealed host cell monolayers: If transcytosis is inhibited by all of these compounds, then this should be reflected in a higher number of intracellular bacteria over time in the presence of such treatments versus untreated cells. Thus, the number of bacteria (or, at least, increase in bacterial fluorescence) at 24 hours of infection in mock-treated and infected host cell samples should be determined.

4) This comment related to the exocytotic SNAREs, which may not necessarily require additional experiments: A surprising finding is that in these cells, transcytotic release is mediated by the set of SNAREs also responsible for exocytosis of synaptic vesicles. First, Syntaxin 1 is almost exclusively expressed in neurons and neuroendocrine cells, and the expression levels in non-neuronal cells are barely detectable by immunocytochemistry. Second, the predominant localization of syntaxin 1 is at the plasma membrane (although it is also present on intracellular vesicles). In Figure 4, the only syntaxin staining that can be discerned is identical with Lep-vesicles. Are you sure the staining is specific? Did you try different antibodies? The western blots shown in the supplement are very clean and show strong expression of both syntaxin 1 and VAMP2 that is surprising. However, it appears that the knockdowns have all worked, supporting that the antibodies recognize the correct proteins, and the results obtained with the botulinum toxins are proving your point. Another control (just to be sure) would be to do cross-stainings for the SNAREs after BoNT transfection: Does BoNT/D affect syntaxin 1 stainings, and BoNT/C1 VAMP2 stainings? One would expect that this is not the case.

Essential revisions:

1) Changes in fluorescence intensity. For many proteins, the fluorescence intensity after immunostaining appears to be massively upregulated after infection. However, it is not clear why this is the case, i.e. whether this is due to relocalization (e.g. recruitment from the cytoplasm to pathogen-containing vacuoles) or to an increase in expression levels. Moreover, fluorescence intensity is a tricky parameter since it is subject to photobleaching and requires that all samples to be compared are analysed in the same field of view.

To clarify this important issue, we are suggesting the following experiments:

a) show immunoblots of the proteins before and at different timepoints after infection. In our opinion, it is sufficient to do this for one or two of the cell lines. Also, include control proteins (not only actin but other proteins involved in membrane traffic, e.g. LAMP1) in order to clarify if the effects (if any) are selective.

b) In the fluorescence images (particularly Figure 3 and Figure 4) please show control stainings before infection, and/or very early time points. This is essential in order to understand the changes caused by the formation of the transcytotic and pathogen-containing endosomes.

At first, we thank the reviewers for reading our manuscript and providing their comments. According to the reviewers’ opinion, we used all the tested cells for detection of Rab5/11, Sec3/15, VAMP2, syntaxin1 and LAMP1 expression during infection. The Western Blot assays showed that all the proteins had no significant increase of expression during infection and the results were shown in the Figure 2B of revised manuscript. We also searched the previous reports about expression of these proteins, but the references are rare. Four recent references showed that the Rab5 during infection of primary human endothelial cells with Kaposi’s sarcoma-associated herpesvirus (Kumar et al., 2016), the Rab5 during infection of canine kidney epithelial cells with enteropathogenic E. coli (Pedersen et al., 2017), the Rab11 during infection of human airway epithelial cells with influenza A virus (de Castro-Martin et al., 2017), and the syntaxin1 and SNAP25 during infection of Trichoplusiani cells with nucleopolyhedrovirus (Guo et al., 2017) had no increase of expression. Only an earlier reference showed that the Exo70 during infection of human hepatoma cells with Dengue virus had an increased expression (Chen et al., 2011). In addition, the confocal microscopic figures about the early stage (1, 2 or 4 hours) during infection and before infection in the original Figure 2, Figure 3 and Figure 4 were added in the revised manuscript as the reviewers’ request (revised Figure 3, Figure 4 and Figure 5, and revised Figure 3—figure supplement 1, Figure 4—figure supplement 1 and Figure 5—figure supplement 1).

2) In Figure 5, you normalize the number of exocytosed bacteria to the number of cells. However, the reduction may also be due to a reduced uptake in the knockdown or toxin-treated cells, and in this case the defect may not be in exocytosis but rather in uptake or even cell viability. It should be checked, at least for some of the conditions, whether uptake/infection is affected by the treatments. It appears that in Figure 5E-H this is only shown for cells that were not transfected with siRNAs or toxins.

The reviewers’ comment is reasonable. We also found that the transfection of siRNAs and botulismotoxins could affect the viability of cells at a short period of time after treatment. We had performed a 24-hour pre-incubation for cell recover after transfection according to the manufacturer’s instruction of transfection kit but it was not stated in the original manuscript. In the revised manuscript, we added a statement about the 24-hours pre-incubation in the Materials and methods section of revised manuscript. According to the reviewer’s opinion, we added the experiments to detect the leptospiral uptake of siRNA- or botulismotoxin-treated cells by confocal microscopy. In addition, the leptospiral uptake in the anthrax toxin-treated cells was also detected. All the results showed that the treatment of siRNAs and toxins did not affect the leptospiral uptake and the data were shown in the revised manuscript (revised Figure 4H, Figure 5H, and revised Figure 1—figure supplement 2, Figure4—figure suplement 1D and Figure 5—figure supplement 1D).

3) Transcytosis is studied by adding bacteria to a completely confluent cell monolayer and the number of traversed bacteria is quantified. Any breach in the monolayer will lead to bacterial movements between mammalian cells. The authors have consequently determined the trans-epithelial(endothelial) electric resistance (TEER) as a sensitive means to uncover such breach. As documented in Figure 6C, the TEER was measured as about 200 ohm/cm2 in all experiments. This result is surprising in two respects: (i) The value is not very high, given that the device with media only will present about half that resistance (Barocchi et al., cited here), although some cell types may actually be confluent at this value; (ii) there is a vast variability in the precise TEER values which the literature indicates for different cell types reaching up to 4,000 ohm/cm2. It is striking that in this manuscript, six types of immortalized cells covering two different cell types and two different donor species have all the exact same TEER of around 200 ohm/cm2 (Figure 6C). If one therefore entertains the possibility that the bilayers were not really tightly sealed and that the motile bacteria would travel beyond the mammalian cell layer, one would expect that, after some time, 50% of the bacteria would be found in one Transwell compartment and 50% in the other, regardless to which side the bacteria were originally added. This is exactly what is observed here with all types of mammalian cells (Figure 6B). This result is also surprising in another way: Whichever receptor is used for the entry of L.i., it would be expected to be present at different concentrations on apical versus basolateral membrane domains and therefore equal rates of transcytosis in both directions (as described here) seems little likely even with sealed cell layers.

We understand that the data in Figure 7 support the notion that the layers are sealed since “transcytosis” is reduced when certain drugs or siRNAs are used. Still, it is of crucial importance to demonstrate the presence of a sealed monolayer which can, in addition to using TEER, simply be studied by testing the flux of FITC-dextran, horseradish peroxidase, inuin, radioactive sucrose or similar compounds between the Transwell compartments through assumptive sealed layers (there are numerous examples in the Transwell literature). Such control experiments are also important as a previous paper by Martinez et al. (cited by the authors) concluded that there was “significant disruption of endothelial cell layers” during infection with leptospira.

The cell monolayer integrity is an important basis for Transwell assay and TEER is a common indicator of cell monolayer integrity. We had read many references but the TEER values of cell monolayers from different cell types and species were distinctly various. Moreover, different cell media, cultural conditions and cell status also affect the TEER values of cell monolayer integrity. In the original manuscript, we showed a little different but not same TEER values (206-223Ω/cm²). Previous studies found that the TEER values of cell monolayer integrity were 200 Ω/cm² for human brain microvessel endothelial cells (Tuma and Hubbard, 2003), 200 and 240-280 Ω/cm² for canine renal epithelial cells (Barocchi et al., 2002; Yu et al., 2013), and 190-220 Ω/cm² for mouse fibroblasts (Alexander et al., 2018). On other hand, we know the reference (Martinez et al., 2010). This paper used the word “disrupt” (in fact, it should be “invasion”) but the authors stated that the infection of L. interrogans did not affect the cell monolayer integrity of human, canine and rat kidney epithelial cells. In our previous study, human renal tubular epithelial cell line (Hek293) and umbilical vein endothelial cell line (HUVEC) showed the TEER values of 211-217 Ω/cm² for cell monolayer integrity within a 24-hour infection of L. interrogans strain Lai (the same strain used in this study) but the TEER values were decreased after 24 hours during infection (Kassegne et al., 2014). In this study, all the time points of infection were within 24 hours. According to the reviewers’ request, we added the FITC-dextran assay to further determine the cell monolayer integrity in the revised manuscript. The results showed that all the six cell monolayers in this study are really integrity.

Along the same lines, we strongly suggest a second, simple control experiment which seems required based on the fact that all siRNA and drug treatments lead to the exactly same inhibition of transcytosis over time (Figure 7) and given the above concerns with sealed host cell monolayers: If transcytosis is inhibited by all of these compounds, then this should be reflected in a higher number of intracellular bacteria over time in the presence of such treatments versus untreated cells. Thus, the number of bacteria (or, at least, increase in bacterial fluorescence) at 24 hours of infection in mock-treated and infected host cell samples should be determined.

We thank the reviewers’ comment and suggestion. siRNAs are commonly used to knockdown the target genes. Anthrax toxin (lethal factor, edema factor and protective antigen) or botulismotoxin (BoNT/D-LC and BoNT/C-LC) have been used by many previous studies to inhibit the Rab11-Sec15 binding or cleave VAMP2 and SYN1 (Guichard et al., 2010; Rossetto et al., 2014). According to the reviewer’s suggestion, we added the immunofluorescence assay by confocal microscopy to detect the leptospires in the siRNA-, anthrax toxin or botulismotoxin toxin-treated cells after a 4-hour infection and then removal of the extracellular leptospires in the revised manuscript. The results were the same as the reviewer’s estimation: the leptospires in the siRNA/toxin-transfected cells at 24 hours of re-incubation after a 4-hour infection were more than those in the untreated cells (revised Figure 6E and 6F, and revised Figure 6—figure supplement 1B).

4) This comment related to the exocytotic SNAREs, which may not necessarily require additional experiments: A surprising finding is that in these cells, transcytotic release is mediated by the set of SNAREs also responsible for exocytosis of synaptic vesicles. First, Syntaxin 1 is almost exclusively expressed in neurons and neuroendocrine cells, and the expression levels in non-neuronal cells are barely detectable by immunocytochemistry. Second, the predominant localization of syntaxin 1 is at the plasma membrane (although it is also present on intracellular vesicles). In Figure 4, the only syntaxin staining that can be discerned is identical with Lep-vesicles. Are you sure the staining is specific? Did you try different antibodies? The western blots shown in the supplement are very clean and show strong expression of both syntaxin 1 and VAMP2 that is surprising. However, it appears that the knockdowns have all worked, supporting that the antibodies recognize the correct proteins, and the results obtained with the botulinum toxins are proving your point. Another control (just to be sure) would be to do cross-stainings for the SNAREs after BoNT transfection: Does BoNT/D affect syntaxin 1 stainings, and BoNT/C1 VAMP2 stainings? One would expect that this is not the case.

Syntaxin1 (SYN1) was firstly found in nerve cells. However, several previous reports showed that many other cells can express SYN1. For example, the rat alveolar epithelial type II cells could express SYN1 and SNAP-25 by Western Bolt assay (Zimmerman et al., 1999), the rat renal medullary collecting duct cells could express SYN1 by Western Bolt assay (Banerjee et al., 2001), rabbit gastric parietal cells could express SYN1/3 by immunofluorescence assay (Karvar et al., 2005), and rat islet cells could express SYN1 (Vieira et al., 2014). In our study, we found that SYN1 was expressed in human or mouse vascular endothelial cells, renal tubular epithelial cells and fibroblasts by immunofluorescence-based confocal microscopy and Western Bolt assay.

In the original Figure 4D, only a part of the SYN1 (blue color) were co-localized with Lep-vesicle-Sec15 (white color). The experiments had been repeated and the similar results were obtained. We are very careful to select antibodies because we know that the efficiency and specificity of antibodies from different companies are various. Usually, we use the antibodies from reputed international professional companies. In addition, in this study, all the antibodies including SYN1-IgG (Abcam) or VAMP2-IgG (Cell Signaling) were determined by preliminary experiments. We also used SYN1-IgG from Santa Cruz Co., but its efficiency was unsatisfactory.

In our experience, the membrane blocking agent is very important in Western Blot assay. In the Western Blot assays, we selected the skim milk agent from BD Co. for blocking. Usually, Western Blot assay is used for qualitative detection of expressed proteins. We are grateful to the review’s suggestion for increase of controls that will make our evidence more solid. Light chain of botulismotoxin D (BoNT/D-LC) or botulismotoxin C (BoNT/C-LC) has been reported as the VAMP2 or SYN1 cleaver (Rossetto et al., 2014). According to the reviewer’s request, we added the controls for the expression and staining of SYN1 in the BoNT/D-LC-transfected cells and VAMP2 in the BoNT/C-LC-trabnsfected cells in the Supplementary Information of revised manuscript. The results showed that the BoNT/D-LC or BoNT/C-LC transfection did not affect the expression and fluorescence staining of SYN1 or VAMP2.