Liver sinusoidal endothelial cells contribute to the uptake and degradation of entero bacterial viruses

The liver is constantly exposed to dietary antigens, viruses, and bacterial products with inflammatory potential. For decades cellular uptake of virus has been studied in connection with infection, while the few studies designed to look into clearance mechanisms focused mainly on the role of macrophages. In recent years, attention has been directed towards the liver sinusoidal endothelial cells (LSECs), which play a central role in liver innate immunity by their ability to scavenge pathogen- and damage-associated molecular patterns. Every day our bodies are exposed to billions of gut-derived pathogens which must be efficiently removed from the circulation to prevent inflammatory and/or immune reactions in other vascular beds. Here, we have used GFP-labelled Enterobacteria phage T4 (GFP-T4-phage) as a model virus to study the viral scavenging function and metabolism in LSECs. The uptake of GFP-T4-phages was followed in real-time using deconvolution microscopy, and LSEC identity confirmed by visualization of fenestrae using structured illumination microscopy. By combining these imaging modalities with quantitative uptake and inhibition studies of radiolabelled GFP-T4-phages, we demonstrate that the bacteriophages are effectively degraded in the lysosomal compartment. Due to their high ability to take up and degrade circulating bacteriophages the LSECs may act as a primary anti-viral defence mechanism.


Results and Discussion
Viruses are quickly (minutes) and extensively (>90%) eliminated by the liver, with LSECs in particular being the primary site of uptake, leaving only a small fraction of circulating virus to infect the body 9,10,16,17 . However, little is known about what these nanoparticles undergo once they enter the scavenging LSECs. Here we investigated the uptake of bacteriophages by primary cultures of rat LSECs, focusing on the clearance ability of the LSECs rather than viral infection. T4 bacteriophages were used as a model virus, which we genetically engineered to express the green fluorescent protein (GFP) in the capsid 19 in order to allow live cell imaging of interaction of the phages with the cells. The integrity of the phages was confirmed by negative staining and transmission electron microscopy showing that the head of the phage was attached to the contractile tail (Fig. 1). Importantly, the phages were not aggregated, but found as single particles, a critical prerequisite for their recognition by LSECs, and not by the Kupffer cells which engulf larger complexes (>200 nm) 22,23 . Freshly isolated rat LSECs in culture were pulsed for 15 min with a low concentration of Alexa Fluor-647-formaldehyde treated bovine serum albumin (AF647-FSA) (5 µg/ml), non-attached ligand washed off, and the cells further incubated for another 1.5 h to functionally mark the late endosomal and lysosomal compartments 24 . The cells were then challenged with GFP-T4-phages and imaged in real time using deconvolution microscopy (DV), with 5-min intervals for 60 min from the time the phages were detected intracellularly ( Fig. 2 and Supplementary Video 1). Due to low GFP-fluorescence intensity per phage particle, the DV was unable to resolve individual phage particles. As a consequence, the phages could not be detected during the initial phase of uptake (not shown). It was only after 20-25 min, when phages had clustered in the endosomal compartment, that the accumulated fluorescence signal was sufficient to resolve and thus visualize the phages. Few GFP-T4-phages were found colocalized in the same vesicles as AF647-FSA at 25 min post-incubation ( Fig. 2A,B), after which the phages displayed a gradual accumulation in the same compartments as the AF647-FSA ( Fig. 2A-C). These compartments are late endosomes/lysosomes (Fig. 3). These findings are in line with previously reported endocytosis of FITC-labelled ligands in rat and pig LSECs, where shortly after internalization and during the first 20 min, the ligands were found mostly in early endosomes, and some (approximately 23% in rat) in late endosomes, and by 2 h all were transferred to late endosomes [25][26][27] .
At the end of the incubation period, the plasma membrane of the live cell culture was labelled with CellMask Green (CMG), and a subset of the same cells that were studied using DV were then imaged at super resolution using structured illumination microscopy (SIM). This allowed us to confirm that the GFP-T4-phage containing cells were indeed LSECs due to the expression of fenestrations (Fig. 2D,E), the signature morphological characteristic of LSECs 28 . The fenestrations are small transmembrane pores with approximate diameters of 50 to 150 nm which provide open channels between the sinusoidal blood and the subendothelial space, facilitating the transfer of substrates between the blood and the parenchymal hepatocytes.
FSA is a frequently used and highly specific endocytic ligand for the LSEC scavenger receptors stabilin-1 and stabilin-2 29,30 , and we chose to use AF647-conjugated FSA here as a late endosome/lysosomal marker for live cell imaging. Previous studies using unconjugated and fluorescently conjugated FSA have shown that the ligand is avidly endocytosed by LSECs in vivo and in vitro, and transported to the lysosomes where it starts to be degraded within 10 min of cellular uptake 31,32 . When conjugated with fluorescent adducts (i.e. FITC or TRITC), or with gold particles, FSA follows the same route of uptake and degradation as unconjugated FSA, except that the non-degradable adducts remain trapped intralysosomally in LSECs for at least 24 h after uptake [25][26][27] . AF647 adduct acts in a similar way when bound to carrier ligands (i.e. albumin or dextran), as shown previously using other cell types 33,34 . Of note, in contrast to the present study, all previous studies assessing intracellular localization of endocytosed ligands in LSECs were performed by fixing the cells after various times of incubation with the ligands (pulse-chase studies). Currently, only two types of markers have been used in live cells to specifically label lysosomes for time lapse imaging. These are (i) LysoTracker probes, which are weak basic amines that accumulate in the acidic lumen of lysosomes 35 , and (ii) fluorescently labelled dextran and gold-labelled bovine serum albumin, both ligands that enter the cells via endocytosis and accumulate in the late endosomes/lysosomes 36,37 . In our setup, we found that LysoTracker ® Red DND-99 (LT) used at the recommended concentration of 50-75 nM was susceptible to rapid photobleaching (data not shown). To bypass this limitation, we increased five-fold the LT concentration, allowing us to capture approximately 5-6 images of the lysosomal dye before photobleaching. However, high concentrations of LT and prolonged live cell imaging are known to affect the lysosomal pH that www.nature.com/scientificreports www.nature.com/scientificreports/ will sometimes radically disturb the intensity and spectral characteristics of a fluorophore 35,38 . In contrast, FSA, the formaldehyde-modified BSA, is a non-toxic and frequently used marker for LSECs, that is rapidly taken up by these cells at high specificity and efficiency 39 . Furthermore, FSA is easily tagged with fluorescent dyes. On this basis we incubated LSECs with a low concentration of AF647-FSA (5 µg/ml) for only 15 min, after which the non-bound ligand was washed off and the cells imaged. Given the very high endocytic efficiency of LSECs, visible uptake of this marker ligand is achieved at low ligand dose and short incubation period, avoiding non-specific uptake in all cells other than LSECs. In the present work this was a prerequisite for optimal imaging, avoiding overloading of the cells with fluorescence. AF647-FSA gradually accumulated in the lysosomes and remained colocalized with the LT up to 20 h of live imaging (Fig. 3). To further confirm that AF647-FSA can serve as a functional marker for different endocytic compartments, we used monensin as a tool to stop the internalization of the FSA at the early endosomes. Monensin is an ionophore that exchanges protons for Na + and K + thus perturbing gradients across cellular membranes, causing multiple effects on the intracellular transport of a variety of ligands, e.g. acidification of early endosomes, inhibition of receptor recycling back to the cell surface, inhibition of receptor-ligand dissociation and further transport of ligands to late endosomes/lysosomes 40 . In the presence of monensin, the AF647-FSA was arrested in fluorescent ring-like structures indicative of early endosomes 25 (Fig. 3). No colocalization of the ligand with the lysosomes was further observed. This is consistent with previous reports of uptake of other ligands by LSECs in the presence of monensin 24,41,42 . With these experiments, we demonstrated that AF647-FSA used in live cell imaging could serve three purposes: i) as a specific functional www.nature.com/scientificreports www.nature.com/scientificreports/ marker for endocytically active LSECs, ii) as a marker for endosomal/lysosomal integrity, and iii) as a non-toxic and highly photostable marker for time-lapse imaging in live LSECs.
The final step of intracellular processing of ligands that have been taken up by LSECs is normally degradation in the endo-lysosomal compartment 39 . However, current fluorescence microscopy approaches lack the ability to probe the degradation of cargo within the cell. A previous study on peritoneal macrophages challenged with T2 phages, a close relative of T4 phages, attempted to detect degraded phages using transmission electron microscopy 43 . This "searching for the needle in the haystack" approach in the referred study did not succeed to show evidence of phage degradation. Other studies have used radioactive isotopes such as 51 Cr, 131 I, 35 S or 125 I to label T4 and M13 bacteriophages and polyoma JC and BK virus-like particles (VLPs) 9,10,44 . While these studies investigated the in vivo clearance mechanisms of these phages and viruses, showing that the liver is the major organ for uptake and degradation, they did not provide details on degradation at the cellular level. However, Simon-Santamaria et al. 10 found degradation products in the blood circulation after intravenous injection of 125 I-labeled JC polyoma VLPs, as well as uptake of non-labelled VLPs and intact virus in mouse LSECs in vitro, strongly suggesting that the LSECs were responsible for their degradation.
To investigate the intracellular processing of the bacteriophages in LSECs in the present study, we challenged the cells in vitro with trace amounts of 125 I-GFP-T4-phages. This enabled us to measure the rate of uptake and degradation over a time interval of 24 h. The functional integrity of LSECs was confirmed in these experiments by the ability of the cells to endocytose and degrade 125 I-FSA (not shown). We found that the LSECs rapidly internalized the 125 I-labeled GFP-T4 phages, supporting our imaging data presented in Fig. 2, and that increasing amounts of degradation products were released to the spent culture medium shortly after the phages had reached the late endosomes/lysosomes (Fig. 4A). Bacteriophages are relatively large viral particles, with a complex structure composed of hundreds of proteins, and highly stable under a variety of harsh environmental conditions. Their catabolism undoubtedly takes longer time than single proteins e.g. modified albumin and collagen alpha chains, known to be efficiently eliminated by LSECs 24,32 (Fig. S4), and is more dependent on the high concentration and specific activity of enzymes in the lysosomes, the terminal degradative compartment 24,45 . Nevertheless, during 24 h post incubation of LSECs with T4-phages, 27% and 16% of total added phages were internalized and degraded by the cells, respectively, demonstrating a substantial capacity of LSECs to take up and degrade the phages.
Using a variety of receptors (mannose receptor, scavenger receptors, and the endocytic Fc-gamma receptor IIb2), the LSECs can directly recognize and internalize pathogen associated molecular patterns, cellular debris and immune complexes, contributing thus to the liver's immune tolerance 46 . To gain further insights into the uptake mechanism of the T4-phages, we investigated whether the stabilin1/2, mannose receptor, or the FcγRIIb2 are involved in the uptake of T4-phages by incubating the cells with radiolabeled phages in the absence or presence of blocking concentrations of native, unlabeled FSA, RNaseB, and AGG, inhibitors for stabilin1/2, mannose receptor, and FcγRIIb2, respectively 31,47,48 . The results presented in Fig. 4B show a borderline significant After each time period, the supernatant from the cells and cell-free well was collected along with one 0.5 ml washing volume of PBS. Trichloroacetic acid (TCA) precipitation was then used to differentiate between free iodine = degraded phages (TCA soluble) (grey columns), and unbound, intact phages (TCA precipitable). Cell bound and internalized phages were quantified in the cell lysates, after solubilizing the cells in 1% SDS (white columns). The results were normalized by subtracting the amount of radioactivity corresponding to the non-specific binding and free 125 I in cell-free wells. Each experiment was performed in triplicates, on cells isolated from four animals (Total N = 12 cell cultures for each time point). Bars represent mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001 represent the statistical differences between the total endocytosis at 18 h and 24 h as compared to 4 h. (B) The specificity of uptake was studied by incubating the cells with 125 I-GFP-T4-phages in the absence (Control) or presence of blocking concentrations (0.1 mg/ml) of FSA, ribonuclease B (RNaseB) and aggregated gamma globulin (AGG), inhibitors for stabilin1/2, mannose receptor and FcγRIIb2, respectively. Cell association and degradation were assessed as above. The results are presented as relative uptake compared to the control which was set to 1. Each experiment was performed in triplicates, on cells isolated from 3 animals (Total N = 9 cell cultures for each time point). Bars represent mean ± SD. (2020) 10:898 | https://doi.org/10.1038/s41598-020-57652-0 www.nature.com/scientificreports www.nature.com/scientificreports/ (p = 0.0497) decrease of phage degradation in the presence of FSA as compared to control. Decreased degradation was also noted in the presence of AGG, although not significant (p = 0.17). Since these ligands had no inhibitory effects on the binding (rather the opposite in the presence of AGG), the results suggest that stabilin1/2 and FcγRIIb2 are probably not involved in the internalization of T4 phages in LSECs. The observed decrease in phage degradation is most probably due to FSA and AGG overloading the later stages of the endocytic pathway. No effect on the uptake and degradation was observed in the presence of RNase B either, suggesting that the mannose receptor is also not involved in the uptake T4 phages. Further studies are needed to investigate the mechanisms by which these cells recognize and internalize bacteriophages.
In conclusion, we have demonstrated that freshly isolated and functionally intact LSECs efficiently take up and degrade T4 bacteriophages at high capacity, indicating that these cells act as a primary anti-viral defence system, adding to the role that these cells have in innate immunity. In vivo studies have reported that injected bacteriophages used in phage therapy are quickly removed from the circulation, thus resulting in lower efficiency of the therapy [49][50][51] . Administration of huge doses of phages is frequently required to saturate unwanted liver uptake, showing the great virus elimination capacity of the liver 9,17 . Our study showing that the LSECs take up and degrade phages may explain some of the challenges with using bacteriophages both in phage therapy and as gene delivery vehicles.

Isolation of liver sinusoidal endothelial cells (LSECs)
, and assessment of endocytosis and specificity of uptake. LSECs were isolated and purified from anesthetized rats as described 53 , and seeded at a density of 0.5 × 10 6 cells/cm 2 on fibronectin coated coverslip-bottom dishes (MatTek In Vitro Life Science Laboratories, Bratislava, Slovak Republic), or 24 well tissue culture plates (Sarstedt, Nümbrecht, Germany) in serum-free RPMI-1640. All experiments started 2-3 h after isolation and seeding. For quantitative studies of uptake and degradation, confluent cultures of freshly isolated LSECs established in 24-well culture dishes coated with fibronectin were incubated in 0.2 mL RPMI containing 3-4 × 10 4 cpm 125 I-GFP-T4-phages. Time-course endocytosis was performed by incubating the cells for 1, 2, 4, 8, 18, and 24 h. For each time point, 3 separate wells containing cells, and 1 cell-free well were used. At the end of each time point, the percent of degraded phages was measured by collecting the spent medium together with one wash volume of 0.5 mL PBS. TCA (0.75 mL, 20%) was added to precipitate intact phages. The amount of TCA-soluble radioactivity measured in the supernatant after centrifugation represented degraded phages. To determine the amount of cell bound and internalized phages, the cells were lysed in 2 × 0.5 mL of 0.1% sodium dodecyl sulfate (SDS) in PBS. The radioactivity was measured in all three fractions using a Cobra II, Auto-Gamma detector (Packard Instruments, Laborel, Oslo, Norway). The results were normalized by subtracting the amount of radioactivity corresponding to the non-specific binding and free 125 I in cell-free wells. The specificity of uptake was studied by incubating the cells with 125 I-GFP-T4-phages in the absence or presence of blocking concentrations (0.1 mg/ml) of FSA, ribonuclease B (RNaseB) or aggregated gamma globulin (AGG), inhibitors for stabilin1/2, mannose receptor and FcγRIIb2, respectively. fluorescence labelling and radioiodination. GFP-T4-phages in PBS were labelled with carrier-free Na 125 I, using Iodogen as described by the manufacturer (Pierce Chemicals), and separated from unbound 125 I on a PD-10 column. The resulting specific radioactivity was approximately 40 × 10 6 cpm per 10 6 PFU phages. (2020) 10:898 | https://doi.org/10.1038/s41598-020-57652-0 www.nature.com/scientificreports www.nature.com/scientificreports/ FSA was conjugated with Alexa Fluor-647 carboxylic acid succinimidyl ester according to the manufacturer's instructions (ThermoFisher Scientific). Free AF647 was separated from FSA using a Vivaspin 6, 10.000 kDa cutoff (ThermoFisher Scientific). The final concentration of the protein was measured with a NANODROP 2000 spectrophotometer (ThermoFisher Scientific).
imaging methods and analyses. Transmission electron microscopy (TEM). Glow discharged formvar-coated copper grids were placed on top of 5 µl droplets of GFP-T4-phages (stock solution of 10 8 PFU/ml in PBS was diluted 1/10 in 0.9% sodium chloride), for 5 min in a moist chamber. The grids were gently washed on droplets of double distilled water and incubated for 20 seconds on a droplet of freshly made 1% uranyl acetate in double distilled water. The uranyl acetate solution was partly removed by filter paper, and the grids air dried, with the remaining uranyl acetate solution faced up, before electron microscopy. The images were recorded in a JEOL JEM 1010 transmission electron microscope (TEM) (JEOL ltd, Tokyo, Japan), operating at 80 kV, and equipped with Morada digital CCD camera (Olympus, Tokyo, Japan). The analysis of the negatively stained GFP-T4-phages showed that the phages were intact and found as single particles (Fig. 1).
Deconvolution and structured illumination microscopy. After washing to remove non-adherent cells, cultures on coverslips were pre-incubated for 30 min with 0.25 μM LysoTracker in RPMI, then with 5 μg/mL AF647-FSA for 15 min, in the presence or absence of 10 μM monensin. The non-bound FSA was removed by washes with PBS, and the cells allowed to endocytose the FSA for another hour in culture medium containing LysoTracker with or without monensin. For live cell imaging, the LSECs were incubated with 0.3 mL RPMI containing 5 × 10 6 PFU GFP-T4-phages, and the uptake followed by deconvolution (DV) microscopy with 5 min intervals between acquisitions, over a period of 90 min. The experiments were terminated by live staining of the plasma membrane with CellMask Green (1:1000 in RPMI) and imaging using structured illumination microscopy (SIM). Live cell imaging was performed using a DeltaVision OMX V4 Blaze imaging system (GE Healthcare) equipped with a 60 × 1.42 NA oil-immersion objective (Olympus), 3 sCMOS cameras, a solid-state illumination source for widefield deconvolution imaging, and a 488 nm laser for structured illumination imaging. Additional widefield deconvolution imaging was performed on a DeltaVision Elite microscope (GE Healthcare) equipped with an identical 60 × 1.42 NA oil-immersion objective (Olympus), single sCMOS camera, and a solid-state illumination source. Widefield deconvolution images were acquired in 4-8 μm z-stacks with one image taken per 0.20-0.25 μm z-step. SIM image stacks of 2 μm were acquired with 15 raw images (five phases, three angles) acquired at each plane with a z-step size of 0.125 μm. During live imaging, the cells were kept under an incubator/control chamber supplied by GE Healthcare, calibrated to approximately 5% humidified CO 2 , and a temperature set at approximately 36 °C. This paper incorporates the results of imaging experiments from approximately 20 different cell isolations. Over the course of the study, experimental procedures were refined and/or had to be repeated/ optimized due to high fluorescence background, improvement of staining, quality of the primary cell cultures, etc. Typically, 5-10 fields of view were imaged for each experiment with 2 (for SIM) or 10 (for DV) cells per field of view. 1-6 imaging experiments were conducted for each cell isolation, including controls and photobleaching assessments ( Supplementary Fig. S3).
The experimental conditions are listed in Supplementary Table. Raw datasets were computationally reconstructed and colour alignment was performed using SoftWoRx software (GE Healthcare) for both deconvolution and structured illumination images. Colocalization was analysed in Volocity 6.3 (PerkinElmer) using the Costes threshold to compute the Pearson's correlation coefficient, for hand-selected image regions which only contained a cell (no cell-free background regions). For clarity of display, figure images were linearly adjusted for brightness and contrast using Fiji (https://fiji.sc, version 2.0.2.) 54 .
Statistical analysis. For all endocytosis experiments, the statistical analyses were performed using the Excel 2016 software (Microsoft, USA). Student's t test (two tailed, unpaired) was used for statistical analysis. All experiments were performed in triplicate, on cells isolated from 3 animals (Total N = 9 cell cultures for each time point and/ or inhibitor) and the results were presented as mean ± standard deviation. The significance was set at: non significant p > 0.05, significant *p < 0.05, **p < 0.01, and ***p < 0.001.