We showed VT binding in paediatric glomeruli, whereas, little or no binding was seen in the glomeruli of adult human renal sections [34].. Renal frozen section binding of VT1, VT2 and an
Trang 1Differential carbohydrate epitope recognition of globotriaosyl
ceramide by verotoxins and a monoclonal antibody
Role in human renal glomerular binding
Davin Chark1,2, Anita Nutikka1, Natasha Trusevych1,3, Julia Kuzmina1,3and Clifford Lingwood1,2,3
1 Research Institute, Division of Infection, Immunity, Injury and Repair, The Hospital for Sick Children, Ontario, Canada;
2
Department of Laboratory Medicine & Pathobiology and3Department of Biochemistry, University of Toronto, Canada
The role of renal expression of the glycosphingolipid
verotoxin receptor, globotriaosylceramide, in susceptibility
to verotoxin-induced hemolytic uremic syndrome is unclear
We show that a single glycosphingolipid can discriminate
multiple specific ligands Antibody detection of
globo-triaosylceramide in renal sections does not necessarily
pre-dict verotoxin binding The deoxyglobotriaosylceramide
binding profile for verotoxin 1, verotoxin 2 and monoclonal
anti-globotriaosylceramide are distinct
Anti-globotriaosyl-ceramide had greater dependency on the intact a-galactose
and reducing glucose of globotriaosylceramide than
vero-toxin 1, while verovero-toxin 2 was intermediate These ligands
differentially stained human kidney sections
Glomerulo-pathy is the primary verotoxin-associated pathology in
hemolytic uremic syndrome For most samples, verotoxin 1
immunostaining within adult glomeruli was observed (type
A) Some samples, however, lacked glomerular binding
(type B) Anti-globotriaosylceramide (and less effectively,
verotoxin 2) stained all glomeruli Verotoxin
1/anti-globo-triaosylceramide tubular staining was comparable Type B
glomerular/tubular globotriaosylceramide showed minor,
but significant, fatty acid compositional differences
Vero-toxin 1 type B glomerular binding became evident following
pretreatment with cold acetone, or methyl-b-cyclodextrin, used to deplete cholesterol Direct visualization, using fluo-rescein isothiocyanate-verotoxin 1B, showed paediatric, but no adult glomerular staining; this was confirmed by anti-fluorescein isothiocyanate immunostaining Acetone induced fluorescein isothiocyanate-verotoxin 1B glomerular staining in type A, but poorly in type B samples Compar-ison of fluorescein isothiocyanate-verotoxin 1B and native verotoxin 1B deoxyglobotriaosylceramide analogue binding showed an alteration in subspecificity These studies indicate
a marked heterogeneity of globotriaosylceramide expression within renal glomeruli and differential binding of vero-toxin 1/verovero-toxin 2/anti-globotriaosylceramide to the same glycosphingolipid Verotoxin 1 derivatization can induce subtle changes in globotriaosylceramide binding to signifi-cantly affect tissue binding Heterogeneity in glomerular globotriaosylceramide expression may play a significant (cholesterol-dependent?) role in determining renal pathology following verotoxemia
Keywords: Membrane glycosphingolipid receptor; lipid isoforms; carbohydrate presentation; hemolytic uremic syndrome; cholesterol
The glycosphingolipid (GSL) globotriaosylceramide
(Gala1–4Galb1–4glucosyl ceramide, Gb3), is the functional
receptor for the verotoxins (VTs, also termed Shiga toxins,
or Stx’s) produced by Escherichia coli [1] Gastrointestinal
infection with E coli producing such toxins can result in
hemorrhagic colitis which may progress to hemolytic uremic
syndrome (HUS), particularly in young children [2] Gb3
is also CD77, a differentiation marker of human germinal
centre B cells [3], the Pk blood group antigen [4] and a
marker of certain tumour cells [5–8], such that VT1 can be used as an antineoplastic agent [9,10] Afraction of Gb3
is found in cell surface cholesterol-enriched lipid micro-domains – rafts [11,12] This organization appears to regulate the intracellular routing of the verotoxin–Gb3 complex [13], such that protein synthesis can be induced, rather than inhibited, by VT for cells in which Gb3is not raft-associated As for all GSLs, heterogeneity of fatty acid, and to a lesser extent, of sphingoid base, generates a spectrum of lipid isoforms of Gb3 Both the lipid structure [14–17] and the local membrane phospholipid microenvi-ronment [18] impinge upon verotoxin–Gb3 binding In addition, there are variants of verotoxin, primarily VT1, VT2 and VT2c [19], which differentially bind to the carbohydrate moiety of Gb3, as determined by differential binding to deoxyGb3analogues [20] and Gb3lipid isoforms [14,16] These variants are also differentially involved in disease [21,22]
Gb3is also involved in the signal transduction of CD19 [23] and a2-interferon [24,25], due to N-terminal sequence similarity between the verotoxin B subunit and the
Correspondence to C Lingwood, Research Institute, Division of
Infection, Immunity, Injury and Repair, The Hospital for Sick
Children, Ontario M5G 1X8, Canada.
Fax: + 1 416 813 5993, Tel.: + 1 416 813 5998,
E-mail: cling@sickkids.ca
Abbreviations: AP, alkaline phosphatase; FITC, fluorescein
isothio-cyanate; Gb 3 , globotriaosylceramide; GSL, glycosphingolipid; NGS,
normal goat serum; SGC, sulfogalactosyl ceramide; TBS,
Tris-buffered saline; VT, verotoxin; VTEC, verotoxigenic E coli.
(Received 15 October 2003, accepted 24 November 2003)
Trang 2N-terminus of CD19 [26] and the a2-interferon receptor [27].
Verotoxin B subunit and monoclonal anti-Gb3can induce
apoptosis [28,29], particularly in lymphoid cells [30] Tissue
screening with monoclonal anti-Gb3indicates a wider Gb3
distribution [31,32] than inferred from pathogenesis of HUS
or VT tissue targeting and pathology in animal models [33]
We have shown previously the expression of Gb3within
the renal glomerulus, as monitored by the binding of
fluorescein isothiocyanate (FITC)-labelled VT1, correlates
with the age-related incidence of HUS following
verotoxi-genic E coli (VTEC) infection [34] Ninety per cent of HUS
cases occur in children under 3 years of age We showed VT
binding in paediatric glomeruli, whereas, little or no binding
was seen in the glomeruli of adult human renal sections
[34] The distribution of Gb3 was thus implicated in the
epidemiology and hence, etiology, of VT-induced disease
[35] However, this differential renal glomerular Gb3
expression has recently been questioned At 4C, VT1
binds similarly to both adult and paediatric glomeruli [36]
Anti-Pk is used to type red cells, yet the binding of VT
to human red cells is only observed at 4C and is not
significant under physiological temperature conditions [37]
We therefore questioned whether anti-Gb3 and VT bind
Gb3in the same manner and whether any differences could
shed light on the Gb3expression in the human kidney in
relation to VT-induced disease Our present studies validate
our previous age-related binding [34], but show a marked
difference in Gb3recognition according to ligand This work
adds a new, clinically relevant, dimension to the
lipid-mediated heterogeneity of Gb3 recognition, which may
provide a precedent for other GSL receptor functions
Materials and methods
VT1, VT1B and VT2 were purified as described [38–40] Rat
mAb anti-Gb3(38.13) hybridoma was a generous gift from
J Wiels (Institut Gustave Roussy, Villejuif, France)
Culture supernatant was affinity purified using Gb3-celite
[41] SULF1 mAb anti-sulfogalactosyl ceramide (SGC)
culture supernatant [42] was kindly provided by P Fredman
(Department Clinical Neuroscience, University of
Gote-borg, Sweden) Rabbit anti-VT2e was supplied by C Gyles
(Department Microbiology, University of Guelph, Ontario)
Polyclonal rabbit anti-VT1B 6869 Ig was prepared in our
laboratory Biotin-conjugated goat anti-(rat IgM) serum
and biotin-conjugated goat anti-rabbit Ig were from
Jack-son Immunoresearch Laboratories (West Grove, PA,
USA) Horseradish peroxidase (HRP)-conjugated goat
anti-rabbit and HRP-conjugated goat anti-mouse Igs were
purchased from Sigma StreptABComplex/alkaline
phos-phatase (AP) were from Dako (Carpinteria, CA, USA)
Avidin/Biotin blocker was from Vector (Burlingame, CA,
USA) True Blue Peroxidase substrate and HistoMark
Red were from Kirkegaard & Perry Laboratories (KPL,
Gaithersburg, MD, USA)
Receptor enzyme linked immunosorbant assay (RELISA)
The wells of a 96-well Evergreen microtitre plate (DiaMed
Laboratory Supplies Inc Mississauga, ON, CA) were
incubated with 150 lL aliquots of 5% (w/v) BSAin
10 mM NaCl/P (pH 7.2) for 2 h at room temperature,
washed three times with ddH2O and allowed to dry completely Ten microliters of Gb3 (100 lgÆmL)1) in an 85% ethanolic solution was then added in duplicate to each well and dried overnight After blocking with 1% BSA/ NaCl/Pi(150 lL per well) for 1 h, wells were successively treated with 50 lL each of serially diluted VT1, VT2, or mAb 38.13 in 1% BSA/NaCl/Pi; corresponding antibodies (rabbit anti-VT1B 6869, rabbit anti-VT2e, or biotin-conju-gated goat anti-rat IgM, respectively, all at 1 : 2000 in 1% BSA/NaCl/Pi); and finally the corresponding horseradish peroxidase-conjugated goat anti-rabbit or HRP-conjugated streptavidin, both at 1 : 2000 in 1% BSA/NaCl/Pi) Each step consisted of a 1 h incubation (100 lL per well) followed by two washings with 1% BSA/NaCl/Pi A fter one final wash with NaCl/Pi, freshly prepared 2,2¢-azino-bis-3-ethylbenzthiazoline-6-sulfonic acid solution (ABTS; Sigma – 0.5 mgÆmL)1ABTS, 0.3 lLÆmL)130% hydrogen peroxide in 0.08Mcitrate/0.1Mphosphate buffer, pH 4.0) was added (100 lL per well) to the wells After a 40 min incubation at room temperature, absorbence at 405 nm in each well was measured with an ELISAplate reader (Dynatech Laboratories) The assay was also performed using VT1, VT2 or mAb 38.13 (each at 1 lgÆmL)1) on serially diluted concentrations of Gb3
Tissue preparation Human renal cortical tissue was harvested 24 h post mortem Grossly normal appearing tissue was excised, embedded in Tissue Tek OCT Compound (Sakura Finetek, Torrence, CA, USA), and snap-frozen in liquid nitrogen Frozen tissue was then sectioned (6 lm) and stored at )70 C Sections to be stained were dried overnight at room temperature and all subsequent steps were performed in a humid chamber
Staining with FITC-conjugated VT1B Sections were blocked with 1% normal goat serum (NGS) (Jackson Immunoresearch Laboratories, Westgrove, PA, USA) in 50 mM Tris-buffered saline (TBS) pH 7.4 for
20 min at room temperature NGS (1% in TBS) was used as the diluent for all toxins and antibodies Sections were stained with FITC-labelled VT1B, prepared as described [43] at 5 lgÆmL)1for 1 h at room temperature, extensively washed with TBS and fixed for 20 min in 4% paraformal-dehyde/NaCl/Pi After washing, sections were treated with
50 mMammonium chloride for 10 min, washed, mounted with fluorescent mounting media (Dako, Carpinteria, CA, USA), and observed under a Polyvar fluorescent micro-scope under incident UV illumination
Immunoperoxidase detection of FITC-VT1B Sections were blocked with endogenous peroxidase blocker (1 mM sodium azide, 1 UÆmL)1 glucose oxidase, 10 mM
glucose) at 37C for 1 h After extensive rinses with TBS, sections were blocked with 1% NGS/TBS for 20 min at room temperature Serial sections were stained with FITC– VT1B (1 lgÆmL)1) for 30 min, washed with TBS and incubated with either biotin-conjugated rabbit anti-FITC (1 : 1000) or rabbit anti-VT1B 6869 (1 : 1000) for 30 min
Trang 3After washing and a 30 min incubation with
HRP-conjugated streptavidin (1 : 500) or HRP-HRP-conjugated goat
anti-rabbit Ig (1 : 500), respectively, FITC–VT1B binding
was chromagenically visualized with True Blue Peroxidase
substrate incubated at room temperature for 5 min
Sections were immersed in water for 5 min, dehydrated
through an ethanolic series, cleared with xylene, and
mounted with Permount Negative control slides were
tre-ated without FITC–VT1B The sensitivity of the
immuno-peroxidase detection allowed for the use of 1 lgÆmL)1
FITC–VT1B instead of the 5 lgÆmL)1used for visualization
by fluorescence
Immunostaining with VT1, VT1B or VT2
The immunoperoxidase staining procedure for VT1 (or
VT1B) was performed as previously [44] but with
modifi-cations Sections were first blocked for endogenous
peroxi-dase as above, and blocked for 20 min with 1% (v/v)
NGS/TBS Sections were then successively incubated with
1 lgÆmL)1 VT1 (or VT1B), rabbit anti-VT1B 6869
(1 : 1000), and HRP-conjugated goat anti-rabbit Ig
(1 : 500) Each step consisted of a 30 min incubation
followed by extensive washing with TBS VT1 (or VT1B)
binding was visualized with True Blue then washed,
dehydrated cleared, and mounted as above
Temperature-dependent binding of VT1 was performed in the same
manner, except all binding steps were performed at either
4C or 37 C Sections to be stained with VT2 were blocked
with Avidin/Biotin blocker followed by 1% (v/v) NGS/TBS
as above, and successively treated with 10 lgÆmL)1VT2,
rabbit VT2e (1 : 1000), biotin-conjugated goat
anti-rabbit (1 : 1000), and StreptABComplex/AP Each step
consisted of a 30 min incubation followed by extensive
washing with TBS VT2 binding was visualized with
HistoMark Red AP substrate Sections were washed,
dehydrated, cleared, and mounted as above Control slides
were treated identically but not exposed to the toxins
Double immunostaining with VT1 and mAb 38.13
Renal sections to be double immunostained were treated
with endogenous peroxidase blocker, Avidin/Biotin blocker
and 1% (v/v) NGS/TBS as above, with TBS washing
between steps After a 30 min incubation with VT1 and
mAb 38.13 together (1 lgÆmL)1and 5 lgÆmL)1,
respect-ively), sections were successively treated with
biotin-conju-gated goat anti-rat IgM (1 : 1000), and StreptABComplex/
AP, in the same manner as above, and mAb 38.13 binding
was visualized with HistoMark Red substrate After
wash-ing with water and 20 min blockwash-ing with 1% (v/v) NGS/
TBS, sections were incubated with rabbit anti-VT1B 6869
(1 : 1000) and HRP-conjugated goat anti-rabbit Ig
(1 : 500), each step consisting of 30 min followed by TBS
washing VT1 binding was visualized with True Blue
Peroxidase substrate and slides were washed, dehydrated,
cleared, and mounted as above Serial tissue sections were
also stained with VT1 and mAb 38.13 independently
Negative control slides were treated without VT1 or with
a rat IgM isotype control for mAb 38.13 Using this
combination of substrates, True Blue Peroxidase must be
developed last as the blue product is soluble in TBS
Epitope unmasking treatments Frozen sections were either incubated with 1.8 UÆmL)1 neuraminidase from Clostridium perfringens (Sigma) in 0.1M acetate buffer (pH 4.7) for 3 h at 37C, 0.125% trypsin (Sigma) in NaCl/Pifor 30 min at room temperature, acetone for 5 min at 4C or 10 mMmethyl-b-cyclodextrin for 45 min at 37C After extensive washing with TBS, sections were blocked with endogenous peroxidase blocker and 1% (v/v) NGS/TBS as above, and successively treated with 1 lgÆmL)1 VT1, rabbit anti-VTB 6869, and HRP-conjugated goat anti-rabbit Ig in the same manner as described VT1 binding was visualized with True Blue Peroxidase substrate Direct binding of FITC–VT1B was also assayed as above after acetone treatment
Renal sulfatide staining Renal sections were treated with endogenous peroxidase blocker and 1% (v/v) NGS/TBS as above, and stained with mouse anti-SGC SULF1 (2 lgÆmL)1) followed by incuba-tion with HRP-conjugated goat anti-mouse Ig (1 : 500) Each step consisted of a 30 min incubation followed by TBS washing SULF1 binding was visualized with True Blue Peroxidase substrate
Thin layer chromatography overlay of deoxyGb3
analogues Deoxy derivatives of a synthetic Gb3analogue containing globotriaose in anomeric linkage to a bis-C16-alkyl sulfone aglycone were synthesized as described [45,46] Glycolipids (5 lg) were resolved on Sil G UV plastic-backed silica TLC plates (Machery-Nagel) with chloroform/methanol/water (65 : 25 : 4, v/v/v), dried, and one plate was stained with 0.5% (w/v) orcinol in 3M H2SO4 The remaining plates were blocked with 0.6% gelatin in water at 37C overnight After washing extensively with water, plates were incubated with VT1 (0.3 lgÆmL)1), VT1B (0.3 lgÆmL)1), VT2 (3 lgÆmL)1), undiluted mAb 38.13 culture supernatant, or FITC–VT1B (0.3 lgÆmL)1) for 2 h at room temperature (all dilutions in TBS) After three washings with TBS, plates were incubated with the corresponding antibody (rabbit anti-VTB 6869, rabbit anti-VT2e, biotin-conjugated goat anti-rat IgM, or biotin-conjugated rabbit anti-FITC) at
1 : 1000 for 1 h, followed, after washing, by HRP-conju-gated goat anti-rabbit or HRP-conjuHRP-conju-gated streptavidin as appropriate at 1 : 1000 for 1 h Finally, plates were washed extensively with TBS and developed with 4-chloro-1-naph-thol peroxidase substrate
Renal glomeruli purification Renal cortical tissue was obtained at autopsy from a 68 year-old Type B phenotype kidney was dissected and stored at )20 C Tissue was then thawed, minced with a razor blade into a paste-like consistency, and pushed through a 50 mesh/
230 lm stainless steel tissue sieve screen (Bellco Glass, Inc Vineland, NJ, USA) The filtrate containing intact glomeruli was washed through a 150 mesh/94 lm screen with NaCl/
Pi Glomerular cores were washed off the screen and collected separately from the final filtrate (tubular fraction)
Trang 4Purity of glomerular and tubular fractions was verified
microscopically Tissue retained by the 50 mesh sieve was
designated the glomeruli-depleted fraction Each fraction
was centrifuged at 2000 g for 3 min and pellets were
extracted in chloroform:methanol (2 : 1) overnight The
lower phase of a Folch partition was dried down and
saponified in 0.1MNaOH/100% methanol overnight After
neutralizing with HCl and desalting, lower phase lipids were
separated on TLC plates VT1 and mAb 38.13 overlays were
performed as described above
Mass spectrometry
The glomerular and tubular glycolipid fractions purified
from the type B kidney above were subject to mass
spectrometry to determine the ceramide heterogeneity
The lower phase GSL extracts (above) were saponified
with 1M NaOH in methanol, neutralized with HCl,
partitioned against water and washed twice GSLs were
isolated from the lower phase by elution from a silica
column in acetone/methanol (9 : 1, v/v) This fraction was
then passed through a DEAE column in methanol to
remove acidic GSLs Approximately 10 pmole Gb3 was
analyzed without further separation in the Mass
Spectro-metry Laboratory at the University of Toronto Saturated
2,5-dihydroxybenzoic acid in methanol was used as the
matrix solution The sample was dissolved in 20 lL
methanol and 1 lL was spotted on the sample target, and
then 1 lL of the saturated matrix solution loaded into the
mass spectrometer The MALDI MS was acquired in
DE-reflection, positive mode on an Applied Biosystems
Voyager-DE STR MALDI-TOF mass spectrometer
equipped with a 377 nm laser The accelerating voltage
was set at 20 KV, grid voltage at 94%, guide wire at 0.05%,
extraction delay time of 175 nsec and low mass gate at
800 Da The mass spectra were externally calibrated with
the molecular mass of a mixture of standard peptides
Results
Comparison of ligand/Gb3binding by RELISA
The binding of VT1, VT2, and mAb anti-Gb3 (38.13) to
Gb3 was assessed using a RELISAwhere binding was
determined as a function of the ligand concentration
(Fig 1A) and the immobilized Gb3concentration (Fig 1B)
While binding parameters cannot be calculated using this
assay, all three ligands showed similar dose-dependent,
saturable binding to Gb3
Binding to deoxyGb3analogues shows different hydroxyl
dependence for VT1and anti-Gb3
To examine the carbohydrate recognition epitopes of Gb3
for VT1, VT2, and mAb anti-Gb3, binding to a series of
synthetic monodeoxyGb3 analogues was assayed by TLC
overlay (Fig 2) The hydroxyl groups of the trisaccharide
moiety were each removed in turn Amarked difference
between the binding profile of the antibody and VT1
was clearly seen, while VT2, although more weakly,
bound many of the same deoxyGb3 analogues as the
mAb anti-Gb Deoxy substitutions within the terminal
a-galactose were less tolerated by mAb anti-Gb3than VT1, most notably at the 3¢ and 4¢ deoxy positions Similarly, deoxy substitutions in the b-glucose, proximal to the ceramide lipid, were more adverse for mAb anti-Gb3 Substitutions at the 2 or 6, but not 3-deoxy positions, were well tolerated by VT1, whereas 3-deoxyglucose but not 2 or
6 substitution allowed mAb anti-Gb3binding VT2 binding was sensitive to any glucose substitution and, like mAb anti-Gb3, little residual binding after any a-galactose hydroxyl substitution was seen With exception of the 3 position, hydroxyl substitutions within the b-galactose were not tolerated by all three ligands In general, all hydroxyls within the trisaccharide were required for full VT2 binding,
as binding to the deoxy analogues were much weaker than
to the parent Gb3bisalkyl analogue Also of note, VT2, but not VT1, required the presence of the hydroxyl at the 6-glucose position
Renal frozen section binding of VT1, VT2 and antiGb3
are not equivalent The binding of these three ligands to adult human renal tissue was then compared by overlay of frozen serial sections VT1, VT2 and mAb anti-Gb staining were
Fig 1 Binding of VT1, mAb anti-Gb 3 and VT2 to Gb 3 The binding of VT1 (d), mAb 38.13 (j), and VT2 (m) to immobilized Gb 3 as a function of (A) ligand concentration (at 1 lg Gb 3 per well) (B) Gb 3
concentration (at 1 lg ligand per mL), as determined by RELISA Similar saturable Gb 3 binding is seen for all three ligands.
Trang 5compared using double-label immunohistochemistry, where
HRP, staining blue, was used to detect VT1 binding, and
AP, staining red, was used to detect either VT2 or anti-Gb3
binding Immunolabelling resulted in a purple stain where
VT1 and anti-Gb3colocalized In the majority of samples,
the staining of anti-Gb3 and VT1 corresponded In some
cases, however, tubular staining by anti-Gb3 was distinct
from that of VT1 and this varied from sample-to-sample
More significantly, renal glomerular staining by anti-Gb3vs
VT1 could be quite distinct In all adult samples studied
(renal tissue from four autopsies ages 38, 46, 68 and 73 years
with no renal pathology), mAb anti-Gb3staining within the
glomerulus was observed (Fig 3Ac,g), while VT1 staining
within the glomerulus was present in three samples both in
single (Fig 3Aa) and double (Fig 3Ab) staining with
anti-Gb3 (designated type Aphenotype) In the other sample
(from a 68 year-old; designated type B phenotype), VT1
staining within the glomerulus was clearly absent both by
single immunostaining (Fig 3Ae) and double
immuno-staining with VT1 and anti-Gb3(Fig 3Af) of serial sections
Tubular staining was similar to that observed in type A
samples MAb anti-Gb3 and VT1 staining were, for the
most part, coincident However, some tubules stained by
anti-Gb3were not bound by VT1 (e.g some pink tubules
are seen in Fig 3Af) VT2 glomerular staining was seen in
all samples, although glomerular staining in the type B
sample was somewhat less (compare Fig 3Ad with h)
Ligand binding was examined routinely at room
tempera-ture However, for type B sections, the effect of temperature
on VT1 glomerular binding was determined At 37C, no
binding within the glomerulus was observed (Fig 3Aa), as seen at room temperature At 4C, VT1 glomerular binding was detected in a serial section (Fig 3Bb) Tubular VT1 staining was also more distinct and discriminatory at 37C (and at room temperature) compared with 4C
Type B glomerular and tubular Gb3are similar and bound by VT1
Renal glomeruli were isolated from this sample to determine whether the Gb3content was in any way unusual GSL was isolated from the residual tubular fraction also The glycolipid fraction from the glomeruli and tubules were subjected to VT1 and mAb anti-Gb3 TLC overlay (Fig 4A) No differences in the Gb3 species detected by such binding, as compared to the renal glomerular and glomerular-depleted fractions, were observed Mass spectro-metric analysis (Fig 4B) showed both the glomerular and tubular Gb3to be comprised primarily of C16, C22, C24 and C24:1 fatty acids C18 and C20 fatty acid Gb3 were detected in the tubular but not the glomerular fraction Hydroxylated C20 fatty acid was detected in glomerular but not tubular Gb3
Differential FITC–VT1B and VT1B renal section staining Amajor fraction of the adult renal samples used in this study were positive for VT1 glomerular staining (type A) using the indirect immunoperoxidase staining procedure This contrasts with our previous report in which no adult glomeruli were labelled by direct binding of FITC-conju-gated VT1 [34] We therefore compared the renal staining
of FITC-labelled VT1B with unmodified VT1B staining detected by the immunoperoxidase procedure (Fig 5) Adult renal medulla showed coincident FITC–VT1B (Fig 5A) and immunoperoxidase VT1B labelling (Fig 5B)
of the same tubules Some tubules were more reactive with VT1B than FITC–VT1B FITC–VT1B labelling within the renal cortex (Fig 5C,E) validated our previous results [34] that FITC-labelled toxin does not stain any adult renal glomeruli In a paediatric sample, FITC–VT1B staining of glomeruli (and some tubules) was evident (Fig 5G) as we had reported previously [34] Pediatric glomerular and tubular staining by VT1 by immunoperoxidase detection was also clear (Fig 5H) In type B adult renal sections (Figs 5E,F), immunoperoxidase anti-VT1 labelling of VT1B-treated serial sections confirmed the lack of glomer-ular staining However, in type Asamples, while FITC– VT1B glomerular staining was negative (Fig 5C), the immunoperoxidase staining of VT1B-treated serial sections showed renal glomeruli bound native VT1B (Fig 5D) To verify the distribution of FITC–VT1B in these renal frozen sections, we used an indirect peroxidase system using anti-FITC as the primary antibody (Fig 5I,K) This indirect immunoperoxidase assay, confirmed that the FITC-labelled toxin was not found within any adult renal glomeruli but restricted to the tubules (Fig 5I,K), as detected previously
by monitoring tissue section fluorescence directly (Fig 5A,C,E) However, when FITC-conjugated VT1B-treated type Aserial sections were visualized using the anti-VT1 peroxidase system, additional glomerular VT1B staining could be observed (Fig 5J), indicating the presence of
Fig 2 Comparison of VT1, mAb anti-Gb 3 and VT2 binding to
monodeoxyGb 3 analogues by TLC overlay DeoxyGb 3 analogues were
separated by TLC and visualized by (A) orcinol chemical detection.
Immunodetection of ligand binding was detected by TLC overlay for
(B) VT1, (C) mAb anti-Gb 3 or (D) VT2 Gb 3 analogues tested
for binding: lanes (1) parent Gb 3 bisalkyl analogue; (2) 2¢¢-deoxy;
(3) 3¢¢-deoxy; (4) 4¢¢-deoxy; (5) 6¢¢-deoxy; (6) 2¢-deoxy; (7) 3¢-deoxy;
(8) 6¢-deoxy; (9) 2-deoxy; (10) 3-deoxy and (11) 6-deoxy analogues.
Adistinct hydroxyl requirement is seen for each Gb 3 ligand.
Trang 6residual unlabelled VT1B within the FITC–VT1B
prepar-ation, binding to the type Aglomeruli Such anti-VT1
glomerular labelling was not seen for FITC–VT1B treated
type B sections (Fig 5L), and glomeruli were similarly
anti-FITC negative (Fig 5K)
FITC-conjugation alters VT1B deoxyGb3binding
Based on these results, we suspected that modification of
the VT1B by FITC-conjugation might have subtly altered
the ability of VT1B to bind Gb3 We therefore compared the
binding of FITC-labelled VT1B and native VT1B to the
series of deoxyGb3 analogues by TLC overlay Separate
binding assays were visualized using either the anti-FITC
peroxidase or the anti-VT peroxidase detection system
(Fig 6) Conjugation of VT1B with FITC significantly
affects the deoxyGb3analogue binding profile as monitored with anti-FITC (Fig 6C) when compared to the native VT1B-subunit (Fig 6A) Specifically, binding to the 6-deoxyglucose analogue is much reduced The binding of residual unconjugated toxin within the FITC–VT1B pre-paration was detected with anti-VT1 Fig 6B
Basis for lack of VT1B renal type B glomerular staining The basis of the differential VT1 and anti-Gb3binding of the type B renal sample was examined Cell surface GSLs may be cryptic [47] via masking by adjacent proteins or sialated glycoconjugates [48] Pretreatment of type B renal sections with either trypsin or neuraminidase/sialidase had
no effect on VT1 glomerular binding (Fig 7A,B); however, strong VT1 binding was observed after cold acetone
Fig 3 Comparison of VT1, mAb anti-Gb 3 and VT2 staining of adult renal frozen sections (A) Serial renal sections from the sample from the 46 year-old (a–d) and the 68 year-year-old (e–h) were stained with VT1 alone (a,e), VT1 and mAb 38.13 together (b,f), mAb 38.13 alone (c,g), or VT2 alone (d,h)
at room temperature In single immunostained sections: (a,e) VT1 (blue) (c,g) mAb anti-Gb 3 (red) (d,h) VT2 (red); colocalization of ligands stains purple in VT1/38.13 double immunostained sections (b,f) The sample from the 46 year-old is representative of type Aphenotype (VT1 positive glomeruli) and that from the 68 year-old of type B phenotype (VT1-negative glomeruli) MAb anti-Gb 3 stains all glomeruli (B) VT1 staining of a type B section at 37 C (a) is compared with staining of a serial section stained at 4 C (b) VT1 glomerular staining is seen at 4 C but is absent at
37 C Tubular staining is more distinct and discriminatory at 37 C Magnification, ·22.
Trang 7treatment (Fig 7C) As cold acetone may extract steroids,
we pretreated kidney sections with methyl-b-cyclodextrin to
deplete cholesterol more selectively [49] This induced
punctate VT1 glomerular staining (Fig 7D) in the type B
sample As cold acetone pretreatment was able to induce
renal glomerular VT1 binding in type B samples, the same
procedure was performed to examine whether the binding
of FITC-VT1B, which did not bind to any adult glomeruli,
could be similarly induced FITC–VT1B glomerular
bind-ing was induced by acetone treatment of type Asamples but
less significantly in type B (Fig 7E–H)
SGC is highly expressed in the kidney [50] and we have
described SGC as a deceptor which when present, can
inhibit VT1 binding to Gb3/cholesterol lipid domains
in vitro[51] We therefore stained serial sections with VT1 and mAb anti-SGC (Fig 7I–L) No glomerular anti-SGC staining was seen, but much of the anti-SGC staining was found for structures not stained by VT1 Similarly, VT1 staining was largely exclusive of anti-SGC binding This was apparent for all sections but was particular clear in type B samples
Discussion
Cell membrane GSL receptor recognition is a more complex process than the recognition of protein or glycoprotein receptors by their appropriate ligands Glycolipids are dynamic structures, both in terms of their lateral mobility and their organization within the plasma membrane The identification of cholesterol and sphingolipid enriched, detergent resistant plasma membrane lipid microdomains, which serve as foci for transmembrane signal transduction [52], has provided a strong impetus to consider GSLs as more than structural bilayer components [53] The hetero-geneity within the lipid moiety of GSLs generates a series of isoforms for each carbohydrate and this heterogeneity may determine or modulate their organization within rafts [54] and subsequent intracellular trafficking pathways [13,55] In the case of VT1–Gb3 binding, although the binding specificity is defined solely by amino acid/carbohydrate contacts [20,56] the lipid-free carbohydrate has a barely detectable binding affinity [15,17] Indeed, the lipid-free globotriaose oligosaccharide and Gb3 glycolipid bind in separate sites within the VT1B subunit pentamer [57] In addition, the fatty acid content of Gb3can markedly affect
VT binding and this effect is different for the different forms
of VT [14,16] Hydroxylation of specific Gb3 fatty acid isoforms for example, preferentially enhances VT2 binding, correlating the high renal hydroxylated fatty acid-Gb3 content [58] and VT2 susceptibility of mice [59] Gb3fatty acid heterogeneity also promotes VT binding [60] In a lipid bilayer, Gb3containing fatty acids shorter than C16 were not bound Increasing the chain length up to C22 increased VT1 binding while that of VT2c was preferential for C18 [14] Such effects can be modulated by the phospholipid acyl chain length within the GSL containing membrane micro-environment [18] Thus, the hydrophobic component and membrane play a central role in VT/Gb3 recognition We have proposed that this effect is mediated by an H-bond network within the interface region at the serine moiety of the GSL [61] to restrict sugar conformation and solvation Molecular modelling predicted two Gb3 binding sites within each monomer of the B subunit pentamer [20] with
site 1 being predominant, while cocrystallization studies identified three Gb3sites, with site 2 dominating [56] Our present studies are the first to identify substitutions within the glucose moiety of Gb3 that affect VT binding, and as such, are more consistent with site 1, rather than site 2,
Gb3 occupancy Moreover, the major difference between VT1 and VT2 binding was the lack of VT2 binding to the 6-deoxyglucose analogue, which is consistent with a predic-ted H-bond from this hydroxyl to VT2, but not VT1, when
Gb3 is docked in site 1 [20] The presence of multiple binding sites predicts that receptor multivalency plays a significant role in determining Gb binding avidity [62] but
Fig 4 Characterization of type B glomerular Gb 3 (A) Glycolipids
from purified renal fractions from a type B (VT1 glomerular-negative)
sample were separated by TLC and visualized by: (a) orcinol chemical
detection; (b) VT1 overlay; (c) mAb anti-Gb 3 overlay; lanes: (1)
gly-colipid standards at 1 lg each; (2) tubular fraction; (3) glomerular
fraction and (4) glomerular-depleted fraction Each lane represents
neutral lipids standardized for Gb 3 content Glycolipid standards are,
from top of the plate, glucosylceramide, galactosylceramide,
lactosyl-ceramide, Gb 3 , Gb 4 , Forssman glycolipid (B) MA LDITOF MS (a)
Tubular Gb 3 , (b) glomerular Gb 3 from type B kidney Only the Gb 3
mass range is shown The masses highlighted are the sodium adducts of
Gb 3 containing the indicated fatty acid.
Trang 8relating this to membrane Gb3glycolipid binding remains
unclear [57]
In the present study, we have shown five important new
findings: (i) VT1 binds deoxyGb3 analogues in a distinct
manner from VT2 and mAb anti-Gb3; (ii) all adult renal
glomeruli contain Gb3 as monitored by mAb anti-Gb3
binding; (iii) FITC-conjugation of VT1B alters the Gb3
binding such that adult renal glomerular Gb3is not bound;
(iv) unmodified VT1 (and VT1B) bind renal glomerular Gb3
within a significant fraction of adult samples and (v) VT1
(and VT1B) unreactive renal glomerular Gb3can be made accessible to FITC-VT1B/VT1 binding by cold acetone or methyl-b-cyclodextrin pretreatment Our studies also valid-ate our previous work [34] that FITC-VT1 glomerular binding is restricted to paediatric renal samples
Based on our present tissue staining results, we now propose two categories for Gb3expression within the adult renal glomerulus: (type A) Gb3 present and reactive with VT1, VT2, and mAb anti-Gb3; (type B) Gb3 present but only reactive with mAb anti-Gb, and to a lesser degree,
Fig 5 Comparison of VT1B and FITC–VT1B staining of renal frozen sections Medullar staining by (A) FITC–VT1B visualized by incident UV illumination and (B) immunoperoxidase detection of VT1B binding to the corresponding serial frozen section Arrows highlight tubules labelled by both procedures Arrowheads indicate tubules preferentially labelled by native VT1B Sections are from a type A sample Type A and B phenotype renal cortical staining by FITC–VT1B (C,E), respectively, and immunoperoxidase detection of VT1B binding to the corresponding serial frozen section (D,F) Type Aand B are both negative for FITC–VT1B glomerular binding Only Type Aexhibits glomerular staining by immunoper-oxidase detection of VT1B FITC–VT1B (G) and VT1 immunoperimmunoper-oxidase (H) staining of paediatric glomeruli Anti-FITC staining of an FITC– VT1B treated Type Asection (I) and anti-VT1 staining of the FITC–VT1B treated corresponding serial section (J) Anti-FITC immunoperoxidase staining of an FITC–VT1B treated of Type B section (K) and anti-VT1 staining of the FITC–VT1B treated corresponding serial section (L) No glomerular anti-FITC staining is seen but anti-VT1 detects VT1B glomerular binding (due to residual unlabelled VT1B in the FITC–VT1B sample) but only in type Asections Magnification, ·11.
Fig 6 Comparison of FITC–VT1B and VT1B binding to deoxyGb 3 analogues The effect of FITC-conjugation on VT1B/Gb 3 binding was assessed
by TLC overlay Lane (1) parent Gb 3 bisalkyl analogue; (2) 2¢-deoxy; (3) 3¢-deoxy; (4) 4¢-deoxy; (5) 6¢-deoxy; (6) 2¢-deoxy; (7) 3¢-deoxy; (8) 6¢-deoxy; (9) 2-deoxy; (10) 3-deoxy and (11) 6-deoxy analogues VT1B binding detected by anti-VT1 (A); FITC–VT1B binding detected by anti-VT1B (B) or anti-FITC (C) FITC-labelling of VT1B reduced binding to the 6 deoxyglucosyl analogue (lane 11).
Trang 9VT2 We have reported this VT1 binding phenotype
previously [44] The relative frequency of these categories
and their relationship to the incidence of HUS following
gastrointestinal VTEC infection remains to be determined
Studies along similar lines may also shed light on the
incidence of HUS in children
Our use of serial frozen section staining allows the direct
comparison of ligand binding and the effect of tissue
modification Our findings greatly expand the physiological
significance of lipid-based heterogeneity of Gb3 receptor
function, which now may impinge on the epidemiology of
HUS following gastrointestinal VTEC infection Our renal
section sample size is very small and therefore the variation
in renal glomerular Gb3 expression in relation to HUS
cannot be addressed in this study Nevertheless our studies
identify the phenotypic prototypes we would expect to
influence the incidence of VT-induced renal disease and
suggest the mechanism separating them The binding of
VT2 to adult renal glomeruli, not bound by VT1, is
consistent with the more common association of this variant
with renal disease [63–65]
Binding studies of VT1, VT2, mAb anti-Gb3, and FITC–
VT1B to deoxyGb3analogues show differential binding to
the sugar moiety and support our contention that these
ligands can preferentially bind different isoforms of Gb3
However, our TLC overlay binding data does not
com-pletely rationalize the differences in renal tissue staining
observed This binding assay, though demonstrating distinct
Gb binding subspecificity, is an ineffective mimic of cell
membrane Gb3 presentation [15] It is clear that the deoxyGb3 binding of mAb anti-Gb3 and VT2 is more restricted than that of VT1 While this is consistent with the similar glomerular staining by mAb anti-Gb3and VT2 but not VT1 in type B samples, overall, mAb anti-Gb3 renal glomerular staining is more, rather than less, widespread than VT1 Deoxy substitution at Glcb2OH or Glcb6OH is not tolerated by mAb anti-Gb3 Access to these hydroxyls could be restricted due to the close apposition of the ceramide moiety [66] maintained by intramolecular H-bonding of the ceramide N-H with the anomeric oxygen [67,68] The Glcb6OH may form an H-bond with 2-hydroxy fatty acid containing GSLs [69] within the interface of the membrane bilayer Thus, particularly when such H-bonds are formed, mAb anti-Gb3 (and VT2) binding may be restricted; in contrast VT1 binding could be unaffected In contrast, the Glcb3OH is involved in an intramolecular H-bond with the b-galactose [20] and the lack of VT1, as opposed to mAb anti-Gb3 binding to this analogue may indicate that restriction of rotation around this anomeric linkage is more important for VT1, as compared to mAb anti-Gb3 and VT2 binding Such considerations could be used to explain the differential VT1, as opposed to mAb anti-Gb3, renal glomerular staining by proposing that excess membrane cholesterol may restrict the formation of the Glcb3OH/Galb5O intramolecular H-bond In terms of FITC–VT1B binding, the major difference is reduced binding to the 6-deoxyglucose analogue as compared to VT1B (or VT1) The glucose residue of Gb is closely
Fig 7 Modulation of glomerular VT1 staining Frozen renal sections of type B phenotype were treated by the following procedures prior toVT1/ peroxidase immunostaining: (A) trypsin; (B) sialidase; (C) acetone and (D) methyl-b-cyclodextrin FITC–VT1B staining of type A(E,F) and type B (G,H) renal sections without (E,G) and after (F,H) acetone treatment Serial frozen sections of type A(I,J) and type B (K,L) phenotype were stained with VT1 (I,K) and mAb anti-SGC (J,L) Only VT1 glomerular staining was seen for type A samples In type B, neither VT1 nor anti-SGC glomerular staining was observed VT1 and anti-SGC staining are largely exclusive (Arrows show structures stained with VT1 but not anti-SGC, and arrowheads show those stained with anti-SGC but not VT1) Magnification, ·11.
Trang 10apposed to the interface between hydrophilic and lipophilic
components of the GSL As such, changes in the lipid
moiety or its membrane environment may more severely
impinge on this region to restrict FITC–VT1B binding to
explain the lack of adult renal glomerular binding
Our earlier studies showed that FITC-coupling at Lys53
of the VT1B subunit inhibited FITC-VT1B/Gb3 binding
[43] and that brief coupling times were necessary to avoid
this The modified Gb3 binding subspecificity we now
observe may be due to FITC-coupling to other lysine
residues within, or adjacent to, the Gb3binding site (e.g
Lys28) Brief FITC-coupling to VT1 has no effect on
cytotoxicity in vitro [43] and FITC–VT1B can compete with
native VT1 for Gb3binding [51]
These studies reveal a surprising complexity of tissue Gb3
receptor function Our finding that the same GSL can be
available for one specific ligand but not another,
demon-strates a new aspect to GSL receptor function This is
relevant to the recent report that VT1 and anti-Gb3induce
apoptosis via different signalling pathways, despite binding
to the same GSL [29] There are several other examples
where a single GSL is recognized by several ligands [61,69–
75] and differential recognition may also occur
The lack of VT1 (and VT1B) binding in glomeruli that
were mAb anti-Gb3 reactive (type B) is likely to be a
property of the membrane Gb3lipid environment as both
ligands were able to effectively bind purified glomerular Gb3
(from the same sample) by TLC overlay Mass spectrometry
of these samples showed a very similar fatty acid
compo-sition However there were differences in the minor species
[C18 and C20 for tubular and C20(OH) for glomerular Gb3]
which could have a bearing on VT1 reactivity Fatty acid
heterogeneity promotes VT1 binding [60] and hydroxylation
can preferentially increase VT2 binding [16] Our finding
that mAb anti-Gb3reactive glomerular structures become
VT1 reactive after cold acetone treatment is consistent with
a role of the membrane environment in Gb3 receptor
function While glycolipids and membrane phospholipids
are poorly, if at all, soluble in cold acetone,
immunolocal-ization of GSL after acetone treatment should nevertheless,
be interpreted with caution Solubility may be sufficient for
GSL diffusion However, as anti-Gb3reveals the presence of
Gb3 in the type B glomeruli, we favour an explanation
whereby Gb3presentation is altered by acetone to promote
VT1 binding Cholesterol is acetone soluble and might be
extracted from the glomerular membranes to achieve such
an effect This is the most likely interpretation, since
cholesterol depletion with methyl-b-cyclodextrin had a
similar effect to induce VT1 recognition, although this is
the first time methyl-b-cyclodextrin has been used to extract
tissue section cholesterol However, the selectivity of
methyl-b-cyclodextrin to deplete cholesterol has recently
been more rigorously investigated [76] The conditions used
in our study, while preferential for cholesterol, may also
extract some sphingomyelin, glycolipids and phosphatidyl
choline from cells Thus a role for cholesterol cannot be
inferred conclusively Cholesterol intercalates with
sphingo-lipids and cholesterol/Gb3enriched lipid microdomains play
a central role in determining VT1 cell sensitivity [13,77]
Although addition of cholesterol promotes VT1/Gb3
bind-ing in an in vitro lipid microdomain bindbind-ing assay we have
developed [51,78], excess cholesterol is inhibitory for VT1
(but not VT2) binding (C Lingwood & A Nutikka, unpublished observation) If glomerular membrane choles-terol was high, limited cholescholes-terol extraction might therefore selectively increase VT1 binding membrane Gb3 rafts to explain our observation The more aggregated VT1 staining seen after methyl-b-cyclodextrin treatment would be con-sistent with optimization of VT1/Gb3 raft binding Thus, potentially, the glomerular cholesterol content may be a risk factor for VT1 binding and hence, for the development of HUS This may also provide the basis of the age-related FITC–VT1B glomerular binding we have found
The lack of VT1 glomerular binding in type B samples is not due to SGC inhibition of VT1/Gb3 binding as these glomeruli lack SGC SGC is a major component of the renal GSL fraction [50] and may be involved in ion transport [79]
We have found the addition of SGC inhibits the binding of VT1 to Gb3/cholesterol microdomains prepared in vitro [51] The extensive expression of SGC in distal tubules could relate to their relative resistance to VT1 [80] VT1 and anti-SGC renal binding were found to be, for the most part, mutually exclusive This gross separation of Gb3and SGC implies functional distinction
Arecent report [36] has contested that there is no age-related difference in human glomerular Gb3 expression and reported that Gb3can be stained by VT1 or anti-Gb3
in all renal glomeruli In these studies, Gb3 was quant-itated by VT1/TLC overlay While suitable for compar-ison, this does not quantitate Gb3 in absolute terms Chemical detection or mass spectrometry is necessary to avoid bias due to preferential binding of VT1 to select
Gb3 isoforms Moreover, in this study, VT1 binding to renal sections at 4C was monitored, and 90% of the binding was reported lost at room temperature [36] The relevance of binding of VT1 to glomeruli (and to red blood cells [37]) at 4C only, to the pathophysiology of VT-induced disease is questionable It is possible that at
4C, the heterogenous Gb3 binding phenotypes we observe, are reduced to a common lower affinity mech-anism, universally present At 4C, the lateral mobility of
Gb3 will be less, reducing the entropic penalty on ligand binding Indeed, we found VT1 glomerular binding in type
B samples at 4C In addition, membrane lipid organ-ization will be altered below their phase transition temperatures [18] The lack of VT1 glomerular binding
in type B samples at room temperature, was also seen at
37C, validating our binding studies at room temperature
as physiologically relevant Tubular binding was also more restricted but more distinct at 37C as compared to 4 C This suggests heterogeneity of tubular VT1 sensitivity under physiological conditions
While the role of Gb3 in mediating VT cytotoxicity
in vitro[13,55,81] and in animal models [33,82,83] has been established, the role in human disease has yet to be defined Renal glomerulus-bound VT1 was found in paediatric, but not in adult cases of HUS [35] This is consistent with our original report [34] and present studies using FITC-VT1B The VT-dependent etiology of HUS in the elderly and the young may be different The reduced susceptibility of adults might be ascribed to a type B Gb3expression, but, though our numbers are small, type Aappears to predominate, implicating other factors Possibly renal Gb3expression can
be modulated Certainly renal Gb expression and VT1