Sialoglycans on lymphatic endothelial cells augment interactions with Siglec-1 (CD169) of lymph node macrophages.
Abstract
This is an open access article under the terms of the Creat ive Commo ns Attri butio nNonCo mmerc ialNoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is noncommercial and no modifications or adaptations are made. © 2021 The Authors. The FASEB Journal published by Wiley Periodicals LLC on behalf of Federation of American Societies for Experimental Biology. Marco D'Addio and Jasmin Frey contributed equally to this work. Abbreviations: Ackr4, atypical chemokine receptor 4; BEC, blood vascular endothelial cells; HEV, high endothelial venules; LEC, lymphatic endothelial cells; LN, lymph node(s); MCM, (lymph node) medullary cord macrophages; MSM, (lymph node) medullary sinus macrophages; PNAd, peripheral lymph node addressins; SCS, subcapsular sinus (of the lymph node); Siglec1, sialic acid binding Iglike lectin1, Sialoadhesin, CD169; SSM, (lymph node) subcapsular sinus macrophages. 1Institute of Pharmaceutical Sciences, ETH Zurich, Zurich, Switzerland 2Department of Surgery, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, Massachusetts, USA
FASEB J. 2021;35:e22017. | 1 of 20 https://doi.org/10.1096/fj.202100300R wileyonlinelibrary.com/journal/fsb2 This is an open access article under the terms of the Creat ive Commo ns Attri butio n- NonCo mmerc ial- NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non- commercial and no modifications or adaptations are made. © 2021 The Authors. The FASEB Journal published by Wiley Periodicals LLC on behalf of Federation of American Societies for Experimental Biology. Marco D'Addio and Jasmin Frey contributed equally to this work. Abbreviations: Ackr- 4, atypical chemokine receptor 4; BEC, blood vascular endothelial cells; HEV, high endothelial venules; LEC, lymphatic endothelial cells; LN, lymph node(s); MCM, (lymph node) medullary cord macrophages; MSM, (lymph node) medullary sinus macrophages; PNAd, peripheral lymph node addressins; SCS, subcapsular sinus (of the lymph node); Siglec- 1, sialic acid binding Ig- like lectin- 1, Sialoadhesin, CD169; SSM, (lymph node) subcapsular sinus macrophages. 1Institute of Pharmaceutical Sciences, ETH Zurich, Zurich, Switzerland 2Department of Surgery, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, Massachusetts, USA Correspondence Vivianne I. Otto, Institute of Pharmaceutical Sciences, ETH Zurich, Vladimir- Prelog- Weg 4, 8093 Zurich, Switzerland. Email: vivianne.otto@pharma.ethz.ch Present address Carlotta Tacconi, Department of Biosciences, Università delgli Studi di Milano, Via Celoria, 26 - Corpo B, 20133, Milano, Italy Funding information ETH Zurich foundation, Grant/Award Number: 2015- 48(2); OPO foundation Zurich K E Y W O R D S endothelial cells, glycocalyx, lymph nodes, macrophages, sialic acid- binding Ig- like lectin 1, sialic acids nline L ibrary on [13/02/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense 2 of 20 | D'ADDIO et al.
1 | INTRODUCTION
The mammalian vasculature consists of hierarchically organized blood and lymphatic vessels. Fluid leaking out of blood capillaries into the tissue is absorbed by open- ended lymphatic capillaries and transported back to the venous circulation. Moreover, blood and lymphatic capillaries are conduits for immune cells whose entry into and exit from the vasculature are tightly regulated.1 The luminal side of lymphatic and blood vessels is formed by a single cell- layered endothelium, which participates in the regulation of vascular permeability for fluids and transmigrating immune cells. Endothelial cells of different vascular structures and of different segments of the same vascular structure are functionally specialized and display characteristic molecular profiles.2,3 Lymphatic endothelial cells (LEC) differ from blood endothelial cells (BEC) in that they express the transcription factor Prospero- related homeobox- 1 (Prox- 1),4 the lymphatic endothelial hyaluronan receptor Lyve- 1,5 and podoplanin.6 The LEC in the different sinuses of the lymph node (LN) can be distinguished based on their expression of membrane glycoproteins that are important for their specialized functional roles.7,8 Many of the endothelial marker proteins are glycosylated.9 The expressed glycans can be directly recognized by glycan- binding proteins (lectins), and influence protein structure and functions. Thus, they have a major impact on molecular and cellular interactions. One of the bestcharacterized examples of a glycan- mediated cellular interaction in the vascular system is the binding of L- selectin on naïve lymphocytes to glycan epitopes on glycoproteins termed peripheral node addressins (PNAd),10,11 expressed by endothelial cells lining the high endothelial venules (HEV) in LN. This highly specific interaction mediated by the unique glycan motif sialylated LewisX and sulfated versions thereof and the cognate glycan- binding protein L- selectin is crucial for lymphocyte extravasation and homing to the LN. Early electron microscope studies revealed that the endothelial surface is coated with glycans.12,13 The endothelial glycocalyx is particularly rich in sialic acid, a monosaccharide present in terminal positions of glycans that are attached to either proteins or lipids.14,15 In addition to being adhesion ligands, the sulfate and sialic acid moieties of endothelial glycoproteins provide abundant negative charges to the surface of endothelial cells, which may serve to minimize the unspecific attachment of blood and immune cells by electrostatic repulsion. Little is known, however, about the overall glycosylation of endothelial cells and their functional roles. Thus, we were led to characterize and compare the glycans of endothelial cells derived from different vessel types and tissues, with a particular focus on unique glycosylation patterns that may be functionally relevant. Our studies reveal an important role for the unusually heavy sialylation of LEC glycans in promoting interactions with Siglec- 1+ (CD169+) macrophages in the LN subcapsular sinus.
2 | MATERIALS AND METHODS
2.1 | Human dermal endothelial cells
Primary human dermal microvascular LEC and donormatched BEC from foreskin16 were cultured on cell culture dishes precoated with PureCol® collagen solution type I (3 mg/ml, Advanced BioMatrix, Carlsbad, CA, USA) in endothelial cell growth basal medium (EBM, Lonza, Basel, Switzerland) supplemented with 20% FCS (Gibco/ Thermo Fisher Scientific, Waltham, MA, USA), 100 U/ml penicillin– streptomycin (Gibco/Thermo Fisher), 2 mM l- glutamine (Gibco/Thermo Fisher), 10 μg/ml hydrocortisone (Sigma- Aldrich, St. Louis, MI, USA) and 25 μg/ml cAMP (Sigma- Aldrich). For BEC, cAMP was replaced by 0.4% V/V Endothelial Cell Growth Supplement with Heparin (Promo Cell, Heidelberg, Germany). All cells were incubated at 37°C and 5% CO2. Before all experiments, cells were cultured in EBM medium supplemented with 20% FCS, 100 U/ml penicillin- streptomycin and 2 mM l- glutamine for 48 h. Their identity was confirmed by qPCR with LEC expressing high levels of PROX- 1, FLT4 (VEGFR3), and podoplanin and BEC expressing high levels of FLT1 (VEGFR1), but no PROX- 1, FLT4 or podoplanin.
2.2 | Isolation of N- glycans from human dermal endothelial cells
For the N- glycomic profile, N- glycans from human dermal LEC and donor- matched BEC were isolated and processed as described previously.17 In brief, cells were cultured in serum- free EBM medium for 4 h. Medium was removed and the cells were washed with PBS. To avoid any cleavage of surface proteins, cells were harvested by scraping and centrifuged. Cell pellets were sonicated and lysed in ice- cold lysis buffer (25 mM TRIS, 150 mM NaCl, 5 mM EDTA and 1% CHAPS (v/v), pH 7.4). Samples were then dialyzed 4 times against 4.5 liters of 50 mM ammonium bicarbonate (AMBIC, Sigma- Aldrich), pH 8.5 for 16 h at 4°C. Thereafter, samples were lyophilized and reduced in 1 mg/ml dithiothreitol (Sigma- Aldrich) in 0.6 M TRIS, pH 8.5 at 50°C for 2 h followed by carboxymethylation with 6 mg/ml iodoacetamide (Acros Organics, Fisher Scientific, Reinach, Switzerland) in 0.6 M TRIS, pH 8.5 at 15306860, 2021, 11, D ow nloaded from https://faseb.onlinelibrary.w iley.com /doi/10.1096/fj.202100300R , W iley O nline L ibrary on [13/02/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense | 3 of 20D'ADDIO et al. room temperature in the dark for 2 h. The carboxymethylation was stopped by dialysis against 4 times 4.5 liters of 50 mM AMBIC, pH 8.5 at 4°C for 12– 48 h. After dialysis, the reduced and carboxymethylated proteins were lyophilized. The subsequent tryptic digest was performed by adding 1 mg/ml trypsin (from bovine pancreas, TPCK treated, Sigma- Aldrich) dissolved in 50 mM AMBIC pH 8.5 at 37°C for 22 h. The (glyco- )peptides were then purified over a C18 Sep- Pak column (200 mg, Waters, Milford, MA, USA) equilibrated with 5% acetic acid. Loaded columns were washed with 20 ml of 5% acetic acid and eluted stepwise with 4 ml of 20%, 40% and 100% propan1- ol (v/v). The eluted fractions were pooled and concentrated by SpeedVac (Eppendorf Schweiz, Schoenenbuch, Switzerland) before lyophilization. To release N- glycans from the lyophilized glycopeptides, samples were digested with 5 μl peptide N- glycosidase (PNGase F, NEB, Ipswich, MA, USA; 500 U/μl) and incubated at 37°C for 24 h. After 12 h, additional 5 μl of PNGase F (500 U/μl) were added to each digestion mixture. The digestion was terminated by lyophilization. To separate the released N- glycans from the peptides, the dried digestion mixture was reconstituted in 5% acetic acid and loaded onto a C18 Sep- Pak column (200 mg) equilibrated with 5% acetic acid. The flow- through and the eluate of 5% acetic acid were collected, concentrated by SpeedVac (Eppendorf) and lyophilized to obtain the pure N- glycans.
2.3 | Preparation of the CORA- reporter
The acetylation required to obtain Ac3GalNAc- α- Bn (CORA- reporter), was performed as described.18 In brief, freshly prepared pyridine: acetic anhydride (2:1; V/V) was added to benzyl 2- acetamido- 2- deoxy- α- d- galactop yranoside (Bn- α- GalNAc; Sigma- Aldrich, Cat#B4894). The mixture was stirred at 65°C for 1 h. After completion of the acetylation, the organic phase was evaporated by SpeedVac (Eppendorf) to obtain a sticky residue, resuspended in water and lyophilized overnight. DMSO was added to the dried product in order to make 50 mM stock solution, which was then stored at −20°C. The purity of the CORA- reporter was controlled by LC- MS.
2.4 | Isolation of Bn- O- glycans from human dermal endothelial cells
The Cellular O- glycome Reporter/Amplification (CORA) method was used as previously described.18 In brief, 12.5 × 104 cells were seeded in a 22 cm2 dish. After 2 days, the medium was replaced by fresh medium containing 50 μM CORA- reporter. After 3 days, the media was collected, centrifuged and the resulting supernatant was filtered over 10- kDa centrifugal filter (Amicon Ultra 10K, Merck) at 3500 rpm for 30 min. The Bn- O- glycanscontaining flow through was loaded onto a C18 Sep- Pak column (200 mg, Waters) equilibrated with 0.1% trifluoracetic acid (v/v). After several washes with 0.1% trifluoracetic acid (v/v), Bn- O- glycans were eluted with 50% (v/v) acetonitrile, 0.1% (v/v) trifluoracetic acid. The eluate was concentrated by SpeedVac (Eppendorf) and then lyophilized.
2.5 | Mass spectrometric analysis (MS)
Dried N- and O- glycans were permethylated using standard procedures.19 In brief, 1 ml of a slurry of crushed sodium hydroxide pellets with anhydrous dimethyl sulfoxide together with 0.5 ml iodomethane (Sigma- Aldrich) was added to the dried glycans. The suspension was vigorously mixed using a vortex and then shaken for 10– 30 min at room temperature (until the opaque suspension turned into a milky and viscous suspension). The reaction was quenched by the dropwise addition of water. The permethylated glycans were extracted into chloroform and the organic phase was washed four times with water. After the glycan extraction, chloroform was removed under a stream of nitrogen. The permethylated glycans were dissolved in 10 μl methanol, and 2 μl of dissolved sample was mixed with 2 μl of freshly prepared 2,5- dihydroxybenzoic acid matrix (20 mg/ml in 80% v/v aqueous methanol). Finally, 2 μl of the sample/matrix mixture was spotted onto a steel target plate and air- dried before analysis. MS acquisition was performed with a MALDI- TOF- MS Ultraflex- II TOF- TOF system (Bruker Daltonics, Billerica, MA, USA).
2.6 | Expression and production of
Siglec- 1- Fc fusion proteins CHO cells stably transduced to express either the WT or the R97A mutant of Siglec- 1- Fc20 (kindly provided by Paul R. Crocker, Division of Cell Signaling and Immunology, College of Life Sciences, University of Dundee, Dundee DD1 5EH, UK) were cultured as adherent cells in Glasgow minimum essential medium (GMEM, G5154 SigmaAldrich) supplemented with GS Supplement (58672C Sigma- Aldrich), 10% FCS, and 100 μM l- methionine sulfoximine (MSX, Sigma- Aldrich) and incubated at 37°C and 5% CO2. For protein production, cells were cultured in GMEM with 0.5% Hyclone FetalClone II Serum (GE Healthcare, Chicago, IL, USA) in a low- level IgG medium and incubated for 5– 7 days before harvest. Collected 15306860, 2021, 11, D ow nloaded from https://faseb.onlinelibrary.w iley.com /doi/10.1096/fj.202100300R , W iley O nline L ibrary on [13/02/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense 4 of 20 | D'ADDIO et al. medium was sterile filtered and then loaded onto a HiTrap Protein A column that had been equilibrated with 20 mM sodium phosphate, pH 7 (GE Healthcare). The fusion proteins were isocratically eluted with 100 mM citric acid, pH 3.5. Affinity chromatography was performed on an Äkta Start liquid chromatography system (GE Healthcare). Collected fractions were neutralized with 1 M TRIS- HCl buffer, pH 9, desalted and concentrated using 10- kDa centrifugal filters (Amicon Ultra- 4 10K, Merck Millipore, Burlington, MA, USA). Protein quality was assessed by Coomassie blue staining and Western blot using sheep polyclonal anti- mouse Siglec- 1 antibody (1:5,000, R&D systems, Minneapolis, MN, USA).
2.7 | Flow cytometry: Glyco- phenotyping and Siglec- 1 binding
Cells were trypsinized, harvested and washed twice with FACS buffer (1% FCS, 2 mM EDTA and 0.02% sodium azide in PBS). They were then incubated with biotinylated plant lectins (all from Vector Laboratories, Burlingame, CA, USA), namely MAL- I (20 μg/ml, Maackia amurensis lectin I), MAL- II (20 μg/ml, Maackia amurensis lectin II), SNA (10 μg/ml, Sambucus nigra agglutinin), LEL (5 μg/ml, Lycopersicon Esculentum (Tomato) lectin), PHA- L (10 μg/ml, Phaseolus Vulgaris Leucoagglutinin), or either WT or R97A mutant Siglec- 1- Fc (100 μg/ml in PBS) for 30 min on ice. Bound plant lectins and Siglec1- Fc were visualized with either fluorescein streptavidin (Vector Laboratories) or DyLight488 secondary goat anti- human IgG Fc (Invitrogen). The cell viability was assessed by Zombie NIR (1:500, BioLegend, San Diego, CA, USA). Nonspecific binding was determined using either fluorescein- streptavidin or DyLight488 secondary goat anti- human IgG Fc alone. Before acquisition, the cells were washed and resuspended in FACS buffer. For the enzymatic removal of sialic acids, the cells were incubated with 0.1 U/ml sialidase (α2- 3, - 6, - 8 neuraminidase from Vibrio cholerae, Roche, Basel, Switzerland) in fresh culture medium at 37°C for 1 h before staining. During sialidase treatment, the cells were resuspended at intervals of 15 min. Flow cytometric analysis was performed on a Beckman Coulter CytoFLEX flow cytometer. Data were analyzed using the FlowJo version 10.5.3 software.
2.8 | Mice
C57BL/6 WT mice, homozygous Siglec1W2QR97A knockin mice on the C57BL/6 background21 (kindly provided by Prof. Paul R. Crocker, Division of Cell Signaling and Immunology, College of Life Sciences, University of Dundee, Dundee DD1 5EH, UK) and heterozygous Ackr4+/GFP mice22 (kindly provided by Prof. Cornelia Halin, Institute of Pharmaceutical Sciences, ETH Zurich, Zurich, Switzerland) were bred in- house under specific pathogen- free conditions. All mice had food and water ad libitum and a 12- h light/dark cycle at constant temperature (22°C) and humidity (55%). In all experiments, animals were 8– 14 weeks old and littermates from both sexes were used when comparing between the different genotypes. All animal experiments were approved by the veterinary office of Kanton Zurich, Switzerland. Mice were, if necessary for the experimental procedure, anesthetized with isoflurane, and euthanized with a combination of Ketamine (12 mg) and Medetomidine (30 μg) injected i.p.
2.9 | Murine ear skin, lung, and lymph node endothelial cell flow cytometric analysis
Lungs and split ear halves of C57Bl/6N wildtype mice were minced using scissors and digested for 25 min at 37°C in basic medium (DMEM supplemented with 2% FCS and 1.2 mM CaCl2) containing either 1000 U/ml Collagenase I (Worthington) and 40 µg/ml DNAse I (Roche) (ear skin) or 0.4 mg/ml collagenase IV (Gibco Thermo Fisher) and 40 µg/ml DNAse I (lung) on a vertical shaker. The remaining ear fragments were smashed through a 70- µm cell strainer, washed with FACS buffer and pelleted by centrifugation. The remaining lung fragments were strained through a 70- µm cell strainer, washed with FACS buffer and pelleted by centrifugation. The lung single cells were then incubated with BD Pharm Lyse lysing solution (BD Biosciences) according to the manufacturer's instructions, strained through a 40- µm cell strainer, washed with FACS buffer and pelleted by centrifugation. Endothelial cells from skin- draining LN were isolated essentially as described.23 In brief, the LN were dissected and the capsule was broken using 25G injection needles. Subsequently, the tissue was digested in 0.8 mg/ml Dispase II (Roche), 0.2 mg/ml Collagenase I and 40 µg/ml DNAse I in DMEM for 20 min at 37°C. The supernatant was discarded and fresh digestion mix was added. The remaining LN fragments were incubated for 10 min at 37°C in the water bath. Fragments were mixed vigorously using a pipette and the supernatant was collected and stored on ice. The remaining tissue fragments were incubated with fresh digestion mix at 37°C and mixed every 5 min until all was digested. Finally, the supernatants from fraction two and three were pooled and used for analysis of LN EC. For the enzymatic removal of sialic acids before staining, single cells from ear skin, lung and LN were incubated with 0.1 U/ml sialidase from Vibrio cholerae in fresh culture medium at 15306860, 2021, 11, D ow nloaded from https://faseb.onlinelibrary.w iley.com /doi/10.1096/fj.202100300R , W iley O nline L ibrary on [13/02/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense | 5 of 20D'ADDIO et al. 37°C for 30 min, resuspending them at intervals of 10 min. Thereafter, the cells were incubated with biotinylated plant lectins (all from Vector Laboratories, Burlingame, CA, USA), namely MAL- I (20 µg/ml), MAL- II (20 µg/ml), SNA (20 µg/ml), or LEL (20 µg/ml), together with CD45— APC- Cy7 (1:400, 30- F11, BioLegend), CD31— APC (1:300, MEC 13.3, BD Biosciences), and Pdpn— PE- Cy7 (1:400, 8.1.1, eBioscience) antibodies for 30 min on ice. In the lung, LEC were identified using Thy1/CD90— FITC (1:200, 53- 2.1, eBioscience). After washing, bound plant lectins were visualized with streptavidin- PE (1:200, BioLegend) and cell viability was assessed with Zombie Aqua (1:500). Cells were washed once and resuspended in FACS buffer before analysis on a LSR Fortessa cell analyzer (BD, San Jose, CA, USA). FACS data were analyzed using the FlowJo version 10.5.3 software.
2.10 | Murine lymph node lymphatic endothelial cell isolation and culture
Primary murine lymph- node LEC were isolated as previously described.24 In brief, skin- draining LN were digested with 0.25 mg/ml Liberase DH (Roche) and 200 U/ml DNAse I (Roche) in RPMI medium (Gibco/Thermo Fisher) at 37°C for 1 h. The LN single cell suspension was filtered and seeded in MEM- α medium (Gibco) supplemented with 100 U/ml penicillin/streptomycin, 10% FCS and 2 mM l- glutamine, into cell culture dishes pre- coated with 10 μg/ml PureCol® collagen solution type I and 10 μg/ ml fibronectin (Merck Millipore). At confluency, cells were detached by using Accutase (Biological Industries, Beit HaEmek, Israel) and endothelial cells were positively selected with CD31- conjugated microbeads (Miltenyi Biotec. Bergisch Gladbach, Germany). LEC were used at passage four for experiments and cellular identity was assessed based on their expression of CD31, podoplanin, VEGFR- 3, CD36, and CD44 by flow cytometry (data not shown). To prevent synthesis of complex type N- glycans, cells were cultured in the presence of 10 μM of the mannosidase I inhibitor kifunensine (Sigma Aldrich) for 72 h at 37°C and 5% CO2 prior to flow cytometric analysis.
2.11 | Lymph node immunofluorescence
Skin- draining LN from WT or Ackr4+/GFP mice were harvested and immediately fixed in 2% paraformaldehyde in PBS at 4°C for 4 h. After fixation and before embedding in optimum cutting temperature compound (OCT, TissueTek, Sysmex Suisse, Horgen, Switzerland), they were placed into a sucrose gradient increasing from 10% to 15% and finally 30% sucrose in PBS at 4°C for at least 2 h per step. Otherwise, LN were embedded directly in OCT and snap frozen in liquid nitrogen. Frozen sections of 7 μm thickness were cut and fixed with acetone (−20°C) and 80% methanol (4°C). For the detection of Ackr4- GFP, LN sections from Ackr4+/GFP mice were fixed in 4% paraformaldehyde in PBS for 15 min, washed with PBS and permeabilized with 0.3% Triton X- 100 for 15 min. The tissue sections were then blocked with Carbo- Free Blocking solution (Vector Laboratories), followed by a second blocking step using the Streptavidin/Biotin Blocking Kit (Vector Laboratories) according to the manufacturer's instructions. For the removal of sialic acids, the LN sections were incubated with 0.1 U/ml sialidase at 37°C in a dark and humid chamber for 2 h before fixation. Primary antibodies and lectins were used as follows: polyclonal chicken anti- GFP (1:100, Aves Labs, Tigard OR, USA) for Ackr4- GFP, rabbit anti- mouse Lyve- 1 (1:300, AngioBio, San Diego, CA, USA), rat anti- mouse/ human PNAd (1:200, MECA- 79, BioLegend), biotinylated MAL- I (1:100, Maackia amurensis I, Vector Laboratories), biotinylated MAL- II (1:20, Maackia amurensis I, Vector Laboratories), biotinylated PNA (1:200, Arachis hypogaea, Sigma Aldrich). After washing, bound antibodies and lectins were visualized with Alexa Fluor 488 and Alexa Fluor 647 secondary polyclonal antibodies (1:200, all Invitrogen, Carlsbad, CA), namely goat anti- chicken (A- 11039), donkey anti- rabbit (A- 21206), donkey anti- rat (A- 21208), donkey anti- rabbit (A- 31573), and streptavidin conjugated Alexa Fluor 594 (1:200, S- 32356). Slides were counterstained with Hoechst 33342 and mounted with Mowiol mounting medium. Images were acquired on a Zeiss Axioskop2 mot plus microscope and confocal Zeiss LSM 880 upright microscope with a Zeiss AxioImager.Z2.
2.12 | RNA in situ hybridization (RNAscope)
For RNA in situ hybridization, the LN were fixed in fresh 10% neutral buffered formalin at room temperature for 20 h immediately after tissue dissection. The fixed LN were then dehydrated using a standard ethanol series followed by xylene and embedded in paraffin. To probe for the lymphatic marker Prox1 and different sialyltransferases on 5 μm LN sections, the RNAscope Multiplex Fluorescent Reagent Kit v2 (Advanced Cell Diagnostics, Newark, CA, USA) with the RNAScope probes Prox1 (Mm- Prox1- C2 targeting bp 590- 1769 of NM_008937.2), St3gal1 (Mm- St3Gal1- C1, targeting bp 1293- 2243 of NM_009177.4), St3gal4 (Mm- St3Gal4- C1, targeting bp 228- 1139 of NM_009178.4), and St3gal6 (Mm- St3gal6- C1, targeting bp 439- 1849 of NM_001357433.1) were used. The RNAScope kit was used according to the manufacturer's instructions. The following pretreatment conditions were 15306860, 2021, 11, D ow nloaded from https://faseb.onlinelibrary.w iley.com /doi/10.1096/fj.202100300R , W iley O nline L ibrary on [13/02/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense 6 of 20 | D'ADDIO et al. used: target retrieval using a steamer for 15 min under gentle boiling and a protease treatment at 40°C for 15 min. The probes were visualized by Opal 570 reagent and Opal 620 reagent (both 1:1500, Perkin Elmer, Waltham, MA, USA). All slides were counterstained with Hoechst 33342 (Invitrogen) and mounted with ProLong Gold Antifade Mountant (Invitrogen). Images were acquired on a confocal Zeiss LSM 880 upright with a Zeiss AxioImager.Z2.
2.13 | Analysis of LN LEC scRNA- Seq data
Data were downloaded from ArrayExpress under the accession number E- MTAB- 8414 and analyzed using Seurat v2.3.425 as previously described.7 Briefly, unsupervised clustering was performed and visualized with Uniform Manifold Approximation and Projection (UMAP).26 The expression patterns of selected markers were plotted using the “FeaturePlot” function in Seurat.
2.14 | Lymph node macrophage flow cytometric analysis
Single mouse LN were minced using scissors and digested in 0.4 mg/ml collagenase IV (Gibco Thermo Fisher) and 40 μg/ml DNAse I (Roche) in DMEM supplemented with 2% FCS and 1.2 mM CaCl2 at 37°C on a rotating wheel for 25 min. The remaining LN fragments were disaggregated by pipetting up and down 99 times using an automated multichannel pipette (Eppendorf Xplorer plus). To maintain the single cell suspension, 7.5 μl of 0.5 M EDTA were added and the pipetting cycles were repeated. Medium was added to neutralize the digestion mixture. Before staining, the single cell suspension was strained through a 40 μm cell strainer, washed with FACS buffer and pelleted by centrifugation. The total cells per LN were divided into two fractions and each of them was transferred to a round bottom 96- well FACS plate. The cells were incubated with CD16/32 (1:50, BioLegend) in PBS on ice for 20 min to block non- specific antibody binding. After blocking, the cells were incubated for 30 min on ice with the primary antibodies (all from BioLegend, if not mentioned otherwise). One of the two fractions per LN was incubated with CD11b— BV605 (1:100, M1/70), F4/80— AF647 (1:100, CI:A3- 1, Bio- Rad), Siglec- 1 (CD169)— PE- Cy7 (1:100, 3D6.112), CD80— BV650 (1:50, 16- 10A1), and viability dye Zombie Aqua (1:500) in PBS. CD3— FITC (1:200, 17A2), CD19— FITC (1:200, 6D5) and LY6- G— FITC (1:400, 1A8, BD Biosciences, Allschwil, Switzerland) antibodies were additionally used to dump out leukocytes as recommended in Ref. [27]. To analyze the proliferative status of macrophages, intracellular staining was performed with Ki67— eFluor450 (1:200, SoIA15, eBioscience, San Diego, CA, USA) using an intracellular staining kit (72- 5775, eBioscience) according to the manufacturer's instructions. The second cell fraction was incubated with CD3— PE- Cy7 (1:200, 145- 2C11), CD4— PE (1:400, RM4- 5, BD Biosciences), CD8— APC- Cy7 (1:200, 53- 6.7), B220— APC (1:200, RA3- 6B2), CD19— FITC and Zombie Aqua. Cells were washed once and resuspended in FACS buffer before analysis on a LSRFortessa cell analyzer (BD, San Jose, CA, USA). FACS data were analyzed using the FlowJo version 10.5.3 software.
2.15 | Statistical analyses
All data are shown as mean values with standard deviations. The statistical significance of differences between mean values was calculated using a paired or unpaired, twotailed Student's t- test at a confidence level of 95% or 2- way Anova as indicated in the figure legends. Differences were considered statistically significant when p < .05, as marked by asterisks with (*) p < .05, (**) p < .01, (***) p < .001, and (****) p < .0001. All statistical analyses were performed using the GraphPad Prism version 8.2.1 software.
3 | RESULTS
3.1 | Human dermal lymphatic endothelial cells display more sialylated
N- glycans and more α2- 3- sialylated
O- glycans than blood vascular endothelial cells from the same individual
We first characterized the glycans present at the surface of human endothelial cells. We reasoned that glycans potentially involved in specific cellular interactions are most likely those that are (i) characteristic for a particular type of vascular endothelial cells and (ii) accessible for recognition and binding by extracellular ligands. We focused our initial studies on the glycans of human dermal LEC and BEC. To aid in profiling surface expression of glycans, we used a set of plant lectins with defined glycan specificities. The binding by Maackia amurensis lectin I (MAL- I) which recognizes α2- 3- sialylated N- glycans (and 3- O- sulfated Gal β1,4GlcNAc) or Sambucus nigra agglutinin (SNA) which recognizes α2- 6- sialylated N- glycans (and 6- O- sulfated galactose) was similar between LEC and BEC (Figure 1A– C). By contrast, binding of Maackia amurensis lectin II (MAL- II) which recognizes α2- 3- sialylated Gal- GalNAc (and 3- O- sulfated Gal- GalNAc) of O- glycans and glycolipids was more pronounced to LEC than to 15306860, 2021, 11, D ow nloaded from https://faseb.onlinelibrary.w iley.com /doi/10.1096/fj.202100300R , W iley O nline L ibrary on [13/02/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense | 7 of 20D'ADDIO et al. BEC (Figure 1D). Pretreatment of the cells with sialidase abolished MAL- II binding to both endothelial cell types (Figure 1E), demonstrating that binding was sialic aciddependent. SNA binding to LEC was partially reduced by sialidase treatment (Figure S1B), suggesting that LEC also carry α2- 6- sialylated glycans. However, a significant proportion of SNA binding to LEC as well as SNA binding to BEC and MAL- I binding to both cell types were sialidase- resistant, suggesting the presence of 3- O and 6- O- sialylated glycans along with 3- O and 6- O- sulfated glycans (Figures 1A and S1A,B). As LEC and BEC can be distinguished based on differential expression of certain membrane- bound glycoproteins, for example, Lyve- 1 and podoplanin, we questioned how these cells compare with regard to protein glycosylation. The N- glycans released from LEC and BEC were analyzed by MALDI- TOF MS (Figure 1F,G). In spite of somewhat similar putative compositions and implied structures, we observed differences in the relative abundances of many glycans. Sialylated N- glycans were more abundant on LEC than on BEC, whereas very large, non- sialylated glycans containing poly- N- acetyllactosamine (poly- LacNAc) were more abundant on BEC than on LEC (Figure S1D,E). This latter observation is in line with the more pronounced binding to BEC than to LEC of the Lycopersicon esculentum lectin (LEL), which recognizes poly- LacNAc (Figure S1C). To analyze the O- glycans, we used the metabolic approach of Cellular O- glycome Reporter/Amplification (CORA).18 When provided with the O- glycan precursor, LEC and BEC synthesized and secreted sialylated core- 1 and core- 2 O- glycans with di- sialylated core- 1 O- glycans prevailing (Figure S1F,G). The data suggest a quantitative difference in biosynthetic potential, as LEC produced more di- sialylated core- 1 O- glycans than BEC (Figure S1F,G). Taken together, these results demonstrate that there are differences between human dermal LEC and BEC in regard to their cell surface glycans. LEC carry more α2- 3- sialylated O- glycans/glycolipids as recognized by MAL- II and more α2- 6- sialylated glycans as recognized by SNA. Of the glycans attached to proteins, the very large N- glycans containing elongated poly- LacNAc chains are more abundant on BEC, whereas LEC produce more heavily sialylated N- glycans and core- 1 O- glycans. 3.2 | Murine dermal lymphatic endothelial cells display more α2- 3- sialylated and 3- O- sulfated O- glycans/ glycolipids than the pertaining blood vascular endothelial cells Subsequently, we analyzed the glycans on the cell surface of LEC and BEC isolated from murine ear skin. Both LEC and BEC were bound by MAL- I, suggesting that they carry glycans containing 3- O- sulfated Gal- GlcNAc (Figure S2A). Binding of MAL- II was more pronounced to LEC pointing to more α2- 3- sialylated and 3- O- sulfated Gal- GalNAc- containing O- glycans/glycolipids on LEC than BEC (Figure S2B). The proportion of MAL- II staining that resisted sialidase treatment, pointing to 3- O- sulfated Gal- GalNAc- containing glycans, was higher than on the human dermal LEC and BEC (compare Figure 1E). SNA binding to both murine dermal LEC and BEC was sialidase- resistant suggesting that it relied on 6- O- sulfated glycans (Figure S2C). Binding of LEL was similar to LEC and BEC, suggesting comparable amounts of elongated, poly- LacNAc- containing glycans (Figure S2D). Taken together, the higher abundance of negatively charged Gal- GalNAc- containing O- glycans/glycolipids on LEC than BEC is conserved between human and murine skin. Among these, the proportion of 3- O- sulfated GalGalNAc is higher on murine endothelial cells, whereas in human skin, the α2- 3- sialylated Gal- GalNAc- containing glycans predominate. The higher abundance of α2- 6- sialylated glycans on LEC and of elongated, poly- LacNAccontaining glycans on BEC observed for the human cells was not observed on the murine dermal endothelial cells. 3.3 | Lymphatic endothelial cells from murine lung and lymph nodes display more sialylated/sulfated and elongated, poly- LacNAc- containing glycans than the pertaining blood vascular endothelial cells We then questioned how the glycocalyces of LEC and BEC may compare in other murine organs, namely the lung and the LN. LEC isolated from both murine lungs and LN carried more 3- O sulfated and α2,3sialylated Gal- GlcNAc- containing structures recognized by MAL- I than the corresponding BEC (Figure S2E,I). Such a difference had not been observed in human or mouse skin. Similar to murine skin, the α2,3- sialylated and 3- O- sulfated Gal- GalNAc- containing O- glycans/ glycolipids detected by MAL- II were more abundant on LEC than on BEC from murine lung and LN (Figure S2F,J). As in murine skin, the 3- O- sulfated glycans predominated over the α2,3- sialylated O- glycans/ glycolipids, even though α2,3- sialylated structures were present in significant amounts as well. LEC displayed more α2,6- sialylated/6- O- sulfated glycans bound by SNA than BEC in both murine lung and LN with the α2,6- sialylated structures contributing a minor proportion of the SNA binding sites (Figure S2G,K). Lung and LN LEC displayed slightly more elongated poly- LacNAc chains than BEC (Figure S2H,L). Taken together, the 15306860, 2021, 11, D ow nloaded from https://faseb.onlinelibrary.w iley.com /doi/10.1096/fj.202100300R , W iley O nline L ibrary on [13/02/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense 8 of 20 | D'ADDIO et al. α2,3- sialylated and 3- O- sulfated Gal- GalNAc- containing O- glycans/glycolipids detected by MAL- II were more abundant on LEC than on BEC in all three murine organs analyzed. However, in murine lung and lymph nodes the LEC also carried more 3- O- sulfated and α2,3sialylated Gal- GlcNAc- containing glycans, more α2,6sialylated/6- O- sulfated glycans and more elongated poly- LacNAc chains than the pertaining BEC. 3, 41 7. 2 3, 05 6. 0 3, 86 7. 4 3, 60 4. 3 3, 50 5. 3 3, 59 1. 3 3, 77 7. 4 3, 23 0. 1 3, 14 4. 1 3, 24 3. 1 3, 67 9. 4 3, 31 8. 2 3, 95 4. 5 3,000 3,300 3,600 3,900 100 80 60 40 20 % In te ns ity Mass (m/z) Lor LEC(F) BEC 3, 41 7. 0 3, 05 5. 9 3, 86 7. 3 3, 60 4. 1 3, 50 5. 1 3, 59 1. 1 3, 77 7. 2 3, 23 0. 0 3, 14 3. 9 3, 24 3. 0 3, 67 9. 2 3, 31 8. 0 3, 95 4. 3 100 80 60 40 20 % In te ns ity Mass (m/z) 3,000 3,300 3,600 3,900 (G) (A) MAL-II R α2-3 R 3S R 3S MAL-I R 3S R α2-3 SNA R α2-6 R 6S R α2-6 R 6S Galactose (Gal) N-Acetylgalactosamine (GalNAc) Sialic acid (Sia) N-Acetylglucosamine (GlcNAc) - +Sialidase (E) - + LEC BEC 0.0 0.5 1.0 gM FI (x 1 05 ) MAL-II ** * (D) MAL-II LE C BE C 0 1 2 3 gM FI (x 1 05 ) * (B) MAL-I LE C BE C 0 1 2 3 gM FI (x 1 05 ) ns (C) SNA LE C BE C 0 1 2 3 gM FI (x 1 05 ) ns 15306860, 2021, 11, D ow nloaded from https://faseb.onlinelibrary.w iley.com /doi/10.1096/fj.202100300R , W iley O nline L ibrary on [13/02/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense | 9 of 20D'ADDIO et al. 3.4 | Alpha2,3- sialylated glycans are present at particularly high densities on primary mouse LEC and the outer edge of the mouse LN The particular glycosylation pattern on murine LN LEC compared to BEC, and the notion that the LN is particularly rich in different types of functionally distinct lymphatic and blood vascular endothelia, prompted us to focus our further studies on this organ. Differing glycosylation of LN LEC and BEC surfaces may be functionally important in guiding/controlling vessel- specific cell- cell recognition and immune cell transmigration processes, such as the glycan- mediated homing of lymphocytes through the HEV.28,29 We thus set out to search for vascular niches within the mouse LN that— analogous to the HEV— display unique glycan epitopes. We analyzed the glycans present on primary LN LEC. Similar to the LEC isolated ex vivo from murine LN, they were stained by MAL- I, MAL- II and SNA (Figure 2A– C); F I G U R E 2 Alpha2,3- sialylated glycans are present at particularly high densities on primary mouse LEC and the outer edge of the mouse lymph node. (A– C) Binding of MAL- I, MAL- II, and SNA to primary lymph node LEC was assessed by flow cytometry before (−) and after (+) treating the LEC with sialidase. Cells stained with fluoresceinstreptavidin alone served as the negative control (black line). Data are from one representative experiment out of five. (D) Immunofluorescence of murine LN sections stained with MAL- I, MAL- II, SNA, and PNA (peanut agglutinin) before or after sialidase treatment. All scale bars are 100 μm F I G U R E 1 Human dermal lymphatic endothelial cells display more sialylated N- glycans and more α2- 3- sialylated O- glycans than blood vascular endothelial cells from the same individual. (A) The glycan structures bound by the plant lectins MAL- I and MAL- II (Maackia amurensis lectin I and II) as well as SNA (Sambucus nigra agglutinin) are shown in symbolic representation. MAL- I preferentially binds N- linked or core 2 O- linked glycans containing the trisaccharide Siaα2- 3Galβ1- 4GlcNAc but also the corresponding 3- O- sulfated Galβ14GlcNAc. MAL- II binds O- linked glycans containing the trisaccharide Siaα2- 3Galβ1- 3- GalNAc, and the corresponding 3- O- sulfated structure. SNA binds sialic acid in α2- 6- linkage to either Gal or GalNAc. It also recognizes the cognate 6- O- sulfated glycans. (B- D) Binding of these lectins to matched human dermal LEC and BEC (lymphatic and blood vascular endothelial cells) from three individual donors was assessed by flow cytometry. Binding is given as ΔgMFI (difference between the geometric mean fluorescence intensity of the stained sample and the negative staining control) (E) Binding of MAL- II was measured before and after treating LEC and BEC with sialidase (n = 3). All graphs (B– E) show data from one representative experiment out of four and dots represent the result obtained for the cells of one human donor. For statistical analysis, a paired, two- tailed Student's t- test was applied. p < .05 (*); p < .001 (**); not significant (ns). N- glycans released from LEC (F) and BEC (G) of one human donor were analyzed by MALDI- TOF MS. The medial molecular range between m/z 3000 and 4000 is shown. Putative structures corresponding to the masses detected are represented next to the corresponding mass peaks. Three mass peaks assigned to tri- antennary N- glycans are highlighted with green color to facilitate semiquantitative comparisons. Whereas the heights of the peaks at 3056 m/z corresponding to the monosialylated glycan are roughly comparable between the mass spectra (F) and (G), those at 3417.0 and 3777.2 corresponding to the di- and tri- sialylated glycan, respectively, are clearly lower in (G) (red arrows). Data were obtained from permethylated N- glycans of the 50% MeCN fraction and all molecular ions are present in sodiated form (M+Na)+ 15306860, 2021, 11, D ow nloaded from https://faseb.onlinelibrary.w iley.com /doi/10.1096/fj.202100300R , W iley O nline L ibrary on [13/02/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense 10 of 20 | D'ADDIO et al. MAL- II binding was abolished upon sialidase treatment (contrasting with the MAL- II staining of the ex vivo isolated LN LEC that was only partially affected by sialidase treatment), whereas MAL- I and SNA interactions were only weakly affected (Figure 2B; as in the ex vivo isolated LN LEC). Upon pretreatment of LN LEC with the mannosidase- I inhibitor kifunensine (Figure S3A), which blocks synthesis of complex- type N- glycans, staining by MAL- I and SNA was strongly reduced as expected, whereas there was little alteration in MAL- II staining, also as expected (Figure S3B). These results suggest that a large proportion of the glycan epitopes recognized by MAL- I and SNA are present on N- glycans, whereas those recognized by MAL- II are present on O- glycans and glycolipids. Taken together we observed that mouse LN- LEC display O- glycans/glycolipids that are heavily α2- 3- sialylated, whereas their Nglycans are mostly 3- O and 6- O- sulfated but only scarcely α2- 3- and α2- 6- sialylated. When probing sections of mouse LN with MAL- I, MAL- II, and SNA, we observed that only the outer edge of the LN that contains the SCS was strongly stained (Figure 2D). Arachis hypogaea agglutinin (peanut agglutinin, PNA), which binds non- sialylated core 1 O- glycans (Galβ1- 3GalNAcα1- Ser/Thr) did not generate a signal. Upon sialidase treatment, MAL- II- staining but not MAL- I or SNA staining was greatly reduced, similar to what we had observed in the FACS analysis of ex vivo isolated LN LEC. Following sialidase treatment, PNA binding was enhanced at the outer border of the LN and also in the LN parenchyma. Together these data reveal that the outer edge of the LN comprising the SCS as well as LEC isolated from mouse LN display dense glycan coats that are likely to contain 3- O- and 6- O- sulfated N- glycans as well as abundant α2- 3- sialylated O- glycans/glycolipids. 3.5 | α2- 3- Sialylated glycans are displayed at uniquely high densities on the lymphatic endothelia forming the floor of the LN SCS We questioned whether the cells at the outer border of the LN that carry the types of glycans recognized by MAL- I, MAL- II, and SNA, might include LEC. The LEC in the SCS are subdivided into ceiling and floor LEC. The former can be recognized based on their expression of the atypical chemokine receptor 4 (Ackr4), but not Lyve- 1, whereas the latter typically express Lyve- 1, but not Ackr4.30 Interestingly, MAL- I and MAL- II stained the Lyve- 1+ LEC of the SCS, but not the Ackr4+ LEC (Figure 3A,B, upper panels). The colocalization with Lyve- 1 was limited to the SCS, as other lymphatic endothelial structures including the medullary sinus were not stained by MAL- I or MAL- II (Figure 3A,B, lower panels). However, there were also non- endothelial cell clusters within the B cell follicles, in particular just below the SCS, that were stained by MAL- II (Figure 3B). We then examined the HEV for expression of 6- O- sulfated GlcNAc- containing PNAd glycans (Figure 3C,D). The PNAd+ structures were not stained by MAL- I or MAL- II, even though some MAL- I+ cells were found in close vicinity to the PNAd+ cells. The predominant terminal glycan motif in mouse HEV is likely sulfated (and potentially also α2- 3- sialylated) Lewis X, which is recognized by the anti- PNAd antibody, but not by MAL- I or MAL- II. Conversely, the floor LEC of the SCS appears to lack sulfated glycans in sufficient amounts to be detected by the anti- PNAd antibody. Taken together, these findings show that in addition to the unique glycocalyx of the HEV, there are also other, very distinct glycocalyces that are characteristic for specific lymphatic endothelia within the mouse LN. Namely, α2- 3- sialylated (and putatively 3- O sulfated) glycans are present at uniquely high densities on the LEC forming the floor, but not those forming the ceiling of the SCS or other lymphatic sinuses. 3.6 | The spatial distribution of sialylated glycans in the mouse LN is not determined by a corresponding expression of sialyltransferases We questioned whether the distinctly high abundances of α2- 3- linked sialic acid on SCS floor LEC may be due to particularly high expression of the cognate sialyltransferases (ST). Thus, we analyzed cells for expression of St3gal1, St3gal4, and St3gal6 by in situ hybridization F I G U R E 3 Alpha 2- 3- Sialylated glycans are displayed at uniquely high densities on the lymphatic endothelia forming the floor of the lymph node subcapsular sinus. Fluorescent stainings of mouse lymph nodes with MAL- I (A, C) or MAL- II (B, D) for 3- O- sulfated and α2- 3- sialylated N- and O- glycans/glycolipids (magenta) and with an antibody against Lyve- 1 (Lymphatic vessel endothelial hyaluronan receptor- 1; green) for lymphatic endothelia. (A, B) Immunostaining for Ackr4 (Atypical chemokine receptor 4; yellow) visualizes the Lyve- 1- negative ceiling lymphatic endothelium of the SCS. Arrows and the arrow head in B point to Lyve- 1+ lymphatic endothelia of the medullary and cortical sinus, respectively, that are not stained by MAL- II. The asterisk marks the location of a B cell follicle. Additional immunofluorescent staining for PNAd (Peripheral lymph node vascular addressin; yellow) on HEV. Scale bars are 50 and 100 μm in the upper and lower panels, respectively 15306860, 2021, 11, D ow nloaded from https://faseb.onlinelibrary.w iley.com /doi/10.1096/fj.202100300R , W iley O nline L ibrary on [13/02/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense | 11 of 20D'ADDIO et al. 15306860, 2021, 11, D ow nloaded from https://faseb.onlinelibrary.w iley.com /doi/10.1096/fj.202100300R , W iley O nline L ibrary on [13/02/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense 12 of 20 | D'ADDIO et al. using RNAscope.31 St3gal1 mainly catalyzes the transfer of sialic acid to the Galβ1- 3GalNAc of O- glycans, whereas St3gal4 and St3gal6 mediate the transfer of sialic acid to the terminal Galβ1- 4GlcNAc of N- glycans.32 To localize the LEC, we probed for expression of Prox1. All three ST were expressed not only in LEC but throughout the LN (Figure 4). Expression levels are not particularly high in the SCS St3gal1 is abundantly expressed in the B cell follicle and St3Gal4 and St3Gal6 expression is particularly high in structures that may represent the HEV. We analyzed for ST expression in the four LEC clusters discovered by unsupervised clustering based on deep RNA sequencing at single- cell resolution. We observed that cells expressing St3Gal1, St3Gal2, St3Gal4, and St3Gal6 are not more numerous among the SCS floor LEC than the SCS ceiling LEC or the medullary sinus LEC (Figure S4). These results indicate that the distinct sialylation of SCS floor LEC is not due to a particularly high expression of α2- 3- sialyltransferases. It seems likely to result from a cell- specific expression of glycoproteins that display α2- 3- sialylated glycans in dense clusters. 3.7 | The α2- 3- sialylated glycans of mouse LN- LEC are bound by Siglec- 1 (CD169) The SCS floor LEC are interspersed with Siglec- 1+ (CD169+) macrophages and separate the lymph from the underlying LN parenchyma. The Siglec- 1+ SCS macrophages form an immunological barrier capturing particulate antigens from the incoming lymph33– 36 and limiting the systemic spread of pathogens.37,38 Siglec- 1 (also known as Sialoadhesin) is the largest member of the sialic acidbinding Ig- like lectin (Siglec) family containing 17 Ig- like domains. Its expression is restricted to some macrophage subsets residing in peripheral tissues and in secondary lymphoid organs such as the LN.21 Siglec- 1 lacks an intracellular signaling motif39 and is considered an adhesion and scavenging molecule. It preferentially binds α2- 3- sialylated glycans (Figure 5A).40,41 Due to the close association between Siglec- 1+ macrophages and SCS floor LEC in the LN, we questioned whether Siglec- 1 would bind the α2- 3- sialylated glycans displayed by LN LEC. We observed that recombinant mouse Siglec- 1- Fc containing the first three extracellular domains of Siglec- 1, which include the carbohydrate recognition domain (CRD),20 bound to murine LN LEC in a sialic acid- dependent manner (Figure 5B). The sialic acids relevant for binding were mostly present on O- glycans and possibly glycolipids, as only a minor proportion of Siglec- 1 binding was lost upon pretreatment of mouse LEC with kifunensine (Figure 5C). Siglec- 1 with the R97A mutation in its CRD20,42 exhibited only weak binding (Figure 5D). These findings show that Siglec- 1 binds the α2- 3- sialylated O- glycans/glycolipids of murine LN LEC and that binding requires that the CRD of Siglec- 1 is functional. 3.8 | The subcapsular sinus macrophages in the LN of mice whose Siglec- 1 lacks the ability to bind sialoglycans are reduced in numbers, are less proliferative, and less activated The binding of Siglec- 1- Fc to the sialylated glycans displayed by cultured LN LEC and the close vicinity of SCS floor LEC and Siglec- 1+ macrophages in the LN suggests that these two cell types may engage in functionally relevant, glycan- dependent interactions in vivo. To investigate such an interplay, we used mice with a non- functional Siglec- 1 CRD based on the W2QR97A double mutation.21 We assessed the numbers of macrophages contained in the LN of both, Siglec1W2QR97A and wild- type (WT) mice (for gating strategy see Figure S5A). The number of CD11b+ leukocytes was comparable in Siglec1W2QR97A and WT mice (Figures 6A and S5B). The percentages of B cells and T cells were comparable as well, even though we observed slightly, but significantly more CD8+ T cells, and a trend towards more T cells and less B cells in the Siglec1W2QR97A mouse LN (Figure S5C). When examining the fractions of CD11b+ cells belonging to the three LN macrophage subpopulations,43 we observed a striking, two- to threefold reduction in the SCS macrophages (SSM; Siglec- 1+, F4/80−; Figures 6D and S5B), no change of the medullary sinus macrophages (MSM; Siglec- 1+, F4/80+; Figures 6B and S5B), and a trend towards a slight increase of the medullary cord macrophages (MCM; Siglec- 1−, F4/80+; Figures 6C and S5B). We consistently observed reduced percentages of SSM in murine LN derived from different anatomical locations (Figure 6E). The residual SSM and the MSM in the LN of Siglec1W2QR97A mice comprised a significantly lower percentage of proliferating cells (based on expression of the nuclear antigen Ki67 that is only detectable in proliferating cells; Figures 6F,G and S6A) and of cells with an activated, proinflammatory phenotype (based on expression of the costimulatory molecule CD80 (B7- 1); Figures 6I,J and S6B). By contrast, we observed no such changes among the MCM (Figures 6H,K and S6A,B). These findings suggest that the ability of Siglec- 1 to bind to sialylated glycans influences the proliferative and proinflammatory phenotype of both types of Siglec- 1+ macrophages, the SSM and the MSM. The results indicate that engagement of Siglec- 1 is a key player in maintaining the Siglec- 1+ SSM population that resides in close vicinity to the SCS floor LEC. 15306860, 2021, 11, D ow nloaded from https://faseb.onlinelibrary.w iley.com /doi/10.1096/fj.202100300R , W iley O nline L ibrary on [13/02/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense | 13 of 20D'ADDIO et al. F I G U R E 4 The spatial distribution of sialylated glycans in the mouse lymph node is not determined by a corresponding expression of sialyltransferases. RNA in situ hybridization (RNAscope) for the sialyltransferases (A) St3gal1, (B) St3gal4, (C) St3gal6 (all in magenta) and the LEC- specific marker Prox1 (prospero- related homeobox- 1; green). The dashed line highlights the border between the subcapsular sinus and the B cell follicle (marked by the asterisk). Scale bars are 100 μm and those in the enlarged areas 50 μm 15306860, 2021, 11, D ow nloaded from https://faseb.onlinelibrary.w iley.com /doi/10.1096/fj.202100300R , W iley O nline L ibrary on [13/02/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense 14 of 20 | D'ADDIO et al.
4 | DISCUSSION
We have characterized the glycomes of endothelial cells derived from different vessel types and organs, looking for unique glycosylation patterns that may be functionally relevant. With regard to protein glycosylation, we found that primary human dermal LEC and BEC synthesize fucosylated N- glycans, but LEC appear to carry higher proportions of the sialylated structures, whereas BEC express more of the very large, elongated N- glycans containing poly- LacNAc repeats (Figures 1F,G, S1C– E and S2). The preferred synthesis of elongated antennae may be a consequence of the lower sialylation in BEC, which competes less successfully with antenna elongation.44,45 LEC produce more sialic acids in α2- 3- linkage to Gal- GalNAc as present in disialylated core- 1 O- glycans (Figures 1A,D,E and S1F,G). The share of sialylated glycans in the α2- 6 sialylated and 6- O- sulfated glycans recognized by SNA is also higher in LEC (Figure S1B). MAL- I staining of LEC and BEC was not affected by sialidase treatment and is thus most likely due to 3- O- sulfated structures. Sulfated glycans could not be directly detected in mass spectrometry due to the protocol employed, which is based on permethylation of glycans to improve ionization efficiency and stabilize the otherwise labile sialic acids, but leads to the loss of sulfates.46 Our finding of substantial sialylation on the surface of vascular endothelial cells is in line with early reports on high sialic acid contents of vascular endothelia.14 Here, we show that LEC glycans are even more heavily sialylated than BEC glycans. From a functional point of view, it is possible that LEC require even more negative charges to minimize non- specific cell– cell interactions via electrostatic repulsion as the lymph flow is much slower than the blood flow. Additionally, the abundant negative charges may serve to immobilize positively charged chemokines such as CCL21 that guide the migration of dendritic cells from the tissue into lymphatic capillaries and towards the LN.47,48 When analyzing endothelial cells isolated ex vivo from murine ear skin, we observed that negatively charged glycans were again more abundant on LEC compared to BEC. The glycans containing α2,3- sialylated and 3- O- sulfated Gal- GalNAc as recognized by MAL- II were more prominent on mouse skin LEC than on BEC (Figure S2B). However, on the murine endothelial cells, the 3- O- sulfated predominated over α2,3- sialylated Oglycans/glycolipids. Neither the fraction of α2,6- sialylated glycans observed in primary human LEC nor the higher abundance of elongated poly- LacNAc- containing glycans on human BEC were detected on the corresponding murine endothelial cells. F I G U R E 5 The α2- 3- sialylated glycans of mouse lymph node LEC are bound by Siglec- 1 (CD169). (A) The glycan structures bound by mouse Siglec- 1 as reported in the literature are shown in symbolic representation. Siglec- 1 preferentially binds glycans terminating in Siaα2- 3Galβ1- 4GlcNAc, but not the corresponding 3- O- sulfated Gal.40,53 However, also mono- and disialylated core- 1 O- glycans as well as Siaα2- 6Galβ1- 4GlcNAc are bound. (B) Binding of recombinant Siglec- 1- Fc to LEC isolated from murine lymph nodes was assessed by flow cytometry with or without treating the cells with sialidase and (C) with or without pretreating the cells with kifunensine. (D) Binding of WT and mutant R97A Siglec- 1- Fc to murine lymph node LEC was measured. The data shown derive from one representative experiment out of three (D)(C)(B) sialidaseuntreated kifunensineuntreated 0 20 40 60 80 100 % o f M ax 0 20 40 60 80 100 % o f M ax 102 103 104 105 106 107 Siglec-1 Fc 0 20 40 60 80 100 % o f M ax 102 103 104 105 106 107 Siglec-1 Fc 102 103 104 105 106 107 Siglec-1 Fc Siglec-1R97ASiglec-1WT (A) Siglec-1 Galactose N-Acetylgalactosamine Sialic acid N-Acetylglucosamine R R R R R β1-4 β1-3 α2-3 α2-6 β1-4 α2-3α2-3 α2-3 α2-6 β1-3β1-3 15306860, 2021, 11, D ow nloaded from https://faseb.onlinelibrary.w iley.com /doi/10.1096/fj.202100300R , W iley O nline L ibrary on [13/02/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense | 15 of 20D'ADDIO et al. FIGURE 6 The subcapsular sinus macrophages in the lymph nodes of mice whose Siglec- 1 lacks the ability to bind sialoglycans are reduced in numbers, are less proliferative, and less activated. FACS analysis showing the number (#) of CD11b+ live singlet LN macrophages (A) and the percentages of medullary sinus macrophages (MSM; B), medullary cord macrophages (MCM; C), and subcapsular sinus macrophages (SSM; D) from one auricular LN of Siglec- 1W2QR97A (n = 6) and WT mice (n = 7). The data derive from one representative experiment out of three (Figure S5B). The percentage of SSM among the CD11b+ cells contained in brachial, axillary, inguinal and popliteal lymph nodes of Siglec- 1W2QR97A and WT mice was assessed. The data derive from one additional experiment. Each dot or square represents the result obtained from one mouse (E). The phenotypes of the SSM (F, I), the MSM (G, J), and the MCM (H, K) in the auricular lymph nodes of Siglec- 1W2QR97A and WT mice were assessed with respect to proliferation (F, G, H; expression of the nuclear protein Ki67) and inflammatory activation (I, J, K; expression of the costimulatory molecule and CD28 ligand CD80). Data shown are from one representative experiment out of three (Figure S6A,B). Statistical analysis was performed using an unpaired two- tailed Student's t- test. not significant (ns); p < .05 (*), p < .01 (**), p < .001 (***), and p < .0001 (****) (F) 0 20 40 60 80 100 % p ro lif er at in g S S M W T Si gle c1 W 2Q R9 7A ** (G) % C D 80 + of S S M W T 0 20 40 60 80 100 % p ro lif er at in g M S M *** (I) (J) W T 0 20 40 60 80 100 *** % C D 80 + of M S M Si gle c1 W 2Q R9 7A Si gle c1 W 2Q R9 7A (H) W T 20 40 60 80 100 % p ro lif er at in g M C M ns Si gle c1 W 2Q R9 7A 0 20 40 60 80 100 W T ** Si gle c1 W 2Q R9 7A Si gle c1 W 2Q R9 7A (K) W T 0 20 40 60 80 100 ns % C D 80 + of M C M Si gle c1 W 2Q R9 7A + 0 (A) (B) (C) (D) (E) ns 0 10,000 20,000 30,000 # C D 11 b+ W T 0 25 50 % M S M o f C D 11 b+ ns W T 0 25 50 % M C M o f C D 11 b+ ns W T 0 25 50 % S S M o f C D 11 b+ **** br ac hia l ax illa ry ing uin al po pli te al 0 25 50 WT Siglec-1W2QR97A W T % S S M o f C D 11 b+ Si gle c1 W 2Q R9 7A Si gle c1 W 2Q R9 7A Si gle c1 W 2Q R9 7A * ns ** ns 15306860, 2021, 11, D ow nloaded from https://faseb.onlinelibrary.w iley.com /doi/10.1096/fj.202100300R , W iley O nline L ibrary on [13/02/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense 16 of 20 | D'ADDIO et al. The higher abundance of glycans containing α2,3sialylated and 3- O- sulfated Gal- GalNAc as recognized by MAL- II was consistently observed between LEC and BEC isolated from mouse skin, lung and lymph nodes (Figure S2B,F). Lung and LN LEC additionally expressed higher amounts of glycans recognized by MAL- I comprising a small proportion of α2,3- sialylated structures in addition to 3- O- sulfated structures. The lung BEC represented the vast majority (96%) of endothelial cells in our sample and consisted of two populations differing in their abundance of 3- O- sulfated glycans. That there are molecularly diverse BEC subpopulations in murine and human lungs (four in murine and eight in human lungs) was recently reported in two independent studies based on single cell RNA sequencing.49,50 These subpopulations were attributed to different vessel types (eg, arteries, veins, capillaries) and to different locations/functions within the lung (eg blood vessels in human alveoli and airways). That humans have more BEC subpopulations in their lungs than mice, was attributed to the evolution of additional cellular functions allowing for a 6000- fold increase in lung size and a 30- fold prolonged live span.50 In contrast to the lung, where BEC were the predominant endothelial cells, the LEC made up for 53% of the LN endothelial cells in our sample. These numbers may reflect the organotypic anatomies and functions. Lungs rely on a dense blood vascular network allowing for efficient oxygen uptake from the inhaled air. LN are specialized to favor encounters between antigen- presenting cells and lymphocytes. They comprise a multitude of highly specialized vascular structures that regulate transmigration and trafficking of immune cells under normal and inflammatory conditions. The abundant negative charges on LN vessels may reflect the great importance of controlling cell- cell interactions with immune cells and to immobilize and present chemoattractants and growth factors that guide the migration and differentiation of immune cells to and within the various lymphatic sinus and blood vascular beds of the LN. Our analyses of glycans in cultured, primary LEC derived from mouse LN revealed that they have similar glycosylation features as observed for the LEC directly isolated from mouse lymph nodes. (Figures 2A– C and S2I– K). However, they were much more uniform and do not appear to comprise LEC subpopulations that differ in glycosylation. This may be due to the fact that in culture, the LEC share the same microenvironment, both with regard to mechanical forces and growth factors/nutrients. This suggests that in vivo, the cellular environment of the LEC in the different sinus systems of the lymph node may provide microenvironmental cues that trigger and maintain the differentiation of specialized LEC subtypes differing in glycosylation. Analyses of entire sections of mouse LN then indicated that there are indeed substantial differences in the glycomes of different subtypes of lymphatic endothelia. We observed a distinctive clustering of α2- 3- sialylated and likely 3- O- and 6- O- sulfated glycans at the outer rim of the LN, where the SCS is located (Figure 2D). Upon closer examination, the α2- 3- sialylated Gal- GalNAc recognized by MAL- II as well as the 3- O- sulfated/α2- 3- sialylated Nglycans recognized by MAL- I (compare Figure 1A) were found to be concentrated on the Lyve1+ LEC that form the floor of the SCS overlying the B cell follicles (Figure 3A,B). The ceiling of the SCS and the medullary sinus were not stained. With regard to their glycomes, the results suggest that there are at least two subtypes of lymphatic endothelia within the LN. Two recent studies provided gene expression maps showing the degree of relatedness between four to six distinct clusters of mouse LN LEC.7,8 The molecular phenotypes of SCS floor LEC, SCS ceiling LEC, and medullary sinus LEC differed markedly (Figure S4). SCS floor LEC and SCS ceiling LEC represented the two most distant cell clusters in nearest neighbor alignments8 and unsupervised clustering,7 in spite of their close anatomical locations. The phenotypic differences included differential expression of specific cell surface glycoproteins. Lyve- 1, MAdCAM1 (Mucosal addressin cell adhesion molecule 1), CD44, integrin- α2B (Itga2b), and the membraneassociated Glycam- 1 were expressed by floor LEC, Ackr4 by ceiling LEC, and Lyve- 1, Mrc1 (Macrophage mannose receptor 1) and MARCO (Macrophage Receptor with Collagenous structure) by medullary sinus LEC. It can be speculated that Glycam- 1 contributes to the densely clustered, sialylated O- glycans detected by MAL- II on the SCS floor LEC, whereas MAdCAM1, CD44, and ItgA2b may present 3- O- sulfated/α2- 3- sialylated glycan epitopes on N- glycans as recognized by MAL- I. Lyve- 1, Mrc1, and MARCO may not present such glycan epitopes in sufficient densities for detection by MAL- I on the medullary sinus LEC. Interestingly, MadCAM1 and Glycam- 1 are among those endothelial glycoproteins of the HEV that express the L- selectin ligands sialyl Lewis X and 6- O- sulfo sialyl Lewis X (sialic acid α2- 3Galβ1- 4(Fucα1- 3(SO3- 6)) GlcNAc) collectively termed PNAd.9– 11,51 It appears that their glycosylation differs between the HEV and SCS floor LEC. HEV, but not the SCS floor LEC, were stained by the MECA- 79 antibody against PNAd. Thus, even though Glycam- 1 and MAdCAM- 1 are expressed in both locations, they only carry the L- selectin ligands in the HEV. The SCS floor LEC were stained by MAL- I, but the HEV were not. This may be explained by a predominance of sialylated glycan epitopes that carry α1- 3- linked fucose on the internal GlcNAc in glycans from the HEV. This fucose 15306860, 2021, 11, D ow nloaded from https://faseb.onlinelibrary.w iley.com /doi/10.1096/fj.202100300R , W iley O nline L ibrary on [13/02/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense | 17 of 20D'ADDIO et al. residue does not hinder binding of the PNAd antibody to 6- O- sulfo sialyl Lewis X on extended core- 1 O- glycans,52 but prevents binding of MAL- I to the underlying α2- 3- sialylated Gal- GlcNAc. Along with differences in the glycoproteins expressed, differing expression levels of sialyltransferases may account for the spatially restricted presence of α2- 3- sialylated glycans on SCS LEC. To our surprise, we found that the RNA expression of the sialyltransferases St3gal1, St3gal4, and St3gal6 was present not only in the Prox1+ LEC, but also in various other LN cells, probably including some B cells in the B cell follicles (Figure 4). Single cell RNAseq data of LN LEC subsets did not reveal differential expression of these sialyltransferases in fLEC, cLEC, medullary LEC nor Ptx3+ LEC (Figure S4; Ref. [7]). The expression level of α2- 3- sialyltransferases, therefore, does not appear to be critical factor in determining the amount of α2- 3- sialylated glycans that accumulate at the surface of the LEC forming the SCS floor and which are recognized by the plant lectins. It is likely that a key factor is the expression of membrane glycoproteins that are acceptors for copious, densely clustered sialic acids that define the heavily α2- 3- sialylated glycocalyx of SCS floor LEC. The SCS floor LEC are closely associated with Siglec- 1+ SSM. We found that recombinant Siglec- 1 bound to cultured LEC isolated from mouse LN. Binding depended on sialylated glycoproteins/glycolipids, whereas N- glycans provided only a minor fraction of the binding sites (Figure 5B,C). Siglec- 1 is known to bind α2- 3- sialylated (but not 3- O- sulfated) Gal linked β1- 3 to GalNAc or β1- 3/ β1- 4 to GlcNAc equally well.53– 55 In addition to α2- 3- sialylated glycans on the LEC, Siglec- 1 binding required a functionally intact CRD, as binding was abolished upon R97A mutation (Figure 5D). Mice with the double W2Q R97A mutation of Siglec- 1 displayed subtle changes in the relative numbers of the major immune cells in their LN compared to WT mice (Figure S5B,C). Percentages of B and T cells were similar, but there was a trend towards fewer B cells and more T cells and there were slightly more CD8+ T cells. The numbers of CD11b+ leukocytes were comparable (Figures 6A and S5B). Strikingly, however, the percentage of the SCS macrophages (SSM) among the CD11b+ leukocytes was reduced two- to three- fold (Figures 6D and S5B), whereas the medullary sinus macrophages (MSM) and the medullary cord macrophages (MCM) were the same. Similar changes in lymphocyte numbers have been found by others when exploring the mesenteric LN of mice that lacked Siglec- 1 expression by FACS.21,56 In that study, the number of SSM were however not found to be reduced in sections of inguinal LN stained with CR- Fc. CR- Fc is a chimeric protein composed of the cysteine- rich (CR) domain of the mannose receptor and the Fc domain of IgG and binds 4- O- sulfated GalNAc.57 CR- Fc binds some glycoforms of Siglec- 1 (but not others) and CR- Fc staining colocalizes with SSM.58 The differences in the specificities of the detection reagents and analysis used (CR- Fc and immunohistochemistry versus antibodies against CD11b, Siglec- 1, and F4/80 and FACS) in the type of LN analyzed (inguinal versus skin- draining LN), makes it difficult to compare the results on the number of SSM in LN of mice with a complete loss of Siglec- 1 expression to those with a loss of Siglec- 1 function. During development, the very first Siglec- 1+ macrophage population makes its appearance on embryonic day 18.5, closely associated to the floor LEC layer of the embryonic LN.59,60 Recruitment of the first Siglec- 1+ macrophages to the developing LN and the specification of SCS floor and ceiling LEC occur in parallel. This observation is well in line with the recent concept of resident macrophages depending on specific cellular niches that provide the cues for their differentiation and survival as tissue- specific macrophages with their set of specialized functions.60– 62 In the SCS of the adult LN, LEC and B cells provide trophic factors that are necessary for differentiation and maintenance of the Siglec- 1+ SSM. Colony- stimulating factor- 1 (CSF- 1) produced by LN LEC plays a major role for both, the SSM and MSM networks.60 Activation of RANK (receptor activator of nuclear factor kappa- B) in LEC by the RANK ligand (RANKL) expressed on marginal reticular cells (MRC; a type of stromal cells localized beneath the SCS and at the outer edge of the B cell follicles) is essential for the sinusoidal macrophage networks to form.63 Direct RANK signaling is required for maintenance of the SSM but not of the MSM. Follicular B cells were reported earlier to provide essential cues to the SSM.35,64,65 The influence of B cells on SSM relies on lymphotoxin- β (LTβ) signaling, as neutralization of this cytokine with LTβ receptor- Fc for four weeks led to an almost complete ablation of SSM and a more MSM- like, SIGN- R1+ phenotype of the macrophages remaining in the SCS. SIGN- R1 is a Ctype lectin that binds oligomannose N- glycans and Lewis type blood group antigens.66 The B cell lymphotoxin- SSM axis is independent of the LEC- CSF- 1- SSM axis, as CSF expression by LEC is unchanged in B cell- deficient mice.60 Taken together, the two unique components of the SSM niche on the floor of the LN SCS described in the literature are (i) direct RANK signaling in response to RANKL expressed on MRC and (ii) stimulation by lymphotoxin- β secreted by the follicular B cells. Here, we provide evidence for a third, unique component, the heavily α23- sialylated glycocalyx of SCS floor LEC. As the binding affinity of Siglec- 1 for its sialylated ligands is low (KD and IC50 values are in the range of 1 mM; Refs. [40,42]), but Siglec- 1 binding appeared strong, it is likely that ligands 15306860, 2021, 11, D ow nloaded from https://faseb.onlinelibrary.w iley.com /doi/10.1096/fj.202100300R , W iley O nline L ibrary on [13/02/2023]. See the T erm s and C onditions (https://onlinelibrary.w iley.com /term s-and-conditions) on W iley O nline L ibrary for rules of use; O A articles are governed by the applicable C reative C om m ons L icense 18 of 20 | D'ADDIO et al. for Siglec- 1 are present at high densities to promote multivalent binding and cell- cell interactions. The strikingly high density of α2- 3- sialylated glycans we observed in the floor LEC makes them likely binding partners. Siglec- 1 has a uniquely elongated molecular shape and lacks an intracellular signaling motif. Its function may thus be to anchor the macrophages in proximity of cells that carry α2- 3- sialylated glycans in particularly high amounts and densities. The macrophages may be confined in a specific position that is characterized by the unique LEC glycocalyx and allows them to receive signals from the locally produced and presented molecular signals, lymphotoxin- β and RANKL. We observed that the loss in the ability of Siglec- 1 to bind α2- 3- sialylated glycans led to a reduction in the proportions of proliferating macrophages and of macrophages with an activated, proinflammatory phenotype among both Siglec- 1+ macrophage subtypes, the SSM and MSM (Figures 6F,G,I,J and S6). This influence on the macrophage phenotype seems to occur through glycan recognition events that may also occur within the medullary sinus, possibly involving sialylated glycans whose low abundance, low density and/or fine structural features did not allow for detection by MAL- I or MAL- II. The reduction in numbers and activation of SSM observed may have functional consequences for the prevention of systemic spread of lymph- borne pathogens. The macrophages present in the subcapsular and medullary LN sinus are among the first immune cells that get into contact with viral and bacterial pathogens carried by the lymph into the draining lymph node.67 They capture viral particles within minutes after injection into the skin and prevent their systemic dissemination.38 SSM can relay and present antigens to naive lymphocytes. They can also transfer immune complexes and viral particles to the follicular B- cells.68 However, SSM seem to play more critical roles in innate immunity. If SSM are depleted, pathogens escape the draining LN and spread systemically, whereas adaptive immune responses still occur, even though with several hours of delay.67 In summary, the surface glycomes of endothelial cells derived from different vessel types differ especially in the abundance and density of α2- 3- siaylated/3- O- sulfated Oglycans/glycolipids and the highest densities are present on the LEC forming the floor of the murine LN SCS. We propose that this uniquely dense sialylation pattern is recognized by the Siglec- 1 of SSM providing anchorage within the cellular nice formed by the SCS floor LEC in close proximity to the MRC and follicular B cells. Thus, the glycosylation of lymphatic endothelial cells may be a key component of the SSM niche that contributes to the maintenance and functional differentiation of the SSM within the LN SCS.
ACKNOWLEDGMENTS
The authors thank Prof. Paul Crocker, University of Dundee for providing the CHO cells expressing the two versions of Siglec- 1- Fc as well as the Siglec1W2QR97A knock- in mice. Yuliang He, ETH Zurich, Institute of Pharmaceutical Sciences, Switzerland, is kindly acknowledged for performing the analysis of the single cell RNASeq data of LN LEC. This work was supported by the ETH foundation (project number 2015- 48(2)) and by OPO foundation, Zurich, Switzerland. Open Access Funding provided by Eidgenossische Technische Hochschule Zurich. DISCLOSURES The authors state explicitly that there are no conflicts of interest in connection with this study and manuscript. AUTHOR CONTRIBUTIONS Marco D'Addio designed experiments, acquired, analyzed, and interpreted data throughout the study, and drafted the manuscript. Jasmin Frey designed experiments, acquired, analyzed and interpreted data (FACS of cultured cells, cells isolated from mouse organs, and LN immune cells, qRT- PCR, IHC) to complete the study and revise the manuscript. Carlotta Tacconi conceived and designed experiments, and acquired FACS data (LN and immune cells). Catharina D. Commerford, Cornelia Halin, and Michael Detmar provided valuable material and intellectual input. Richard D. Cummings provided valuable intellectual input (glycosylation analysis using MS and plant lectins) and substantially revised the manuscript. Vivianne I. Otto conceived, planned, and supervised the study, and wrote the manuscript. All authors critically reviewed the article and approved the final manuscript. DATA AVAILABILITY STATEMENT All data that support the findings of this study are available from the corresponding author upon request. ORCID Vivianne I. Otto https://orcid.org/0000-0002-1157-0186
