Engineering the carbohydrate binding site of Epa1p from Candida...

1

© The Author 2014. Published by Oxford University Press. All rights reserved. For permissions, please e‐mail: [email protected]

Engineering the carbohydrate binding site of Epa1p from Candida

glabrata: generation of adhesin mutants with different carbohydrate

specificity

Francesco S. Ielasi1, Tom Verhaeghe2, Tom Desmet2, Ronnie G. Willaert1

1Department of Bioengineering Sciences, Structural Biology Research Center (SBRC); Vrije

Universiteit Brussel; 1050 Brussels, Belgium

2Department of Biochemical and Microbial Technology, Centre for Industrial Biotechnology and

Biocatalysis, Ghent University, 9000 Ghent, Belgium

Corresponding author: Ronnie G. Willaert, Department of Bioengineering Sciences, Structural Biology

Brussels (SBRC); Vrije Universiteit Brussel, Brussels, Belgium; Tel: +32 2 629 1846, Fax: +32 2 629

1963; E-mail: [email protected].

Glycobiology Advance Access published July 21, 2014 by guest on July 21, 2014

http://glycob.oxfordjournals.org/D

ownloaded from

http://glycob.oxfordjournals.org/

2

Abstract

The N-terminal domain of the Epa1p adhesin from Candida glabrata (N-Epa1p) is a calcium-

dependent lectin, which confers the opportunistic yeast the ability to adhere to human epithelial cells.

This lectin domain is able to interact with galactosides and, more precisely, with glycan molecules

containing the Gal-1,3-GalNAc disaccharide, also known as the T-antigen. Based on the

crystallographic structure of the N-Epa1p domain and the role of the variable loop CBL2 in glycan

binding, saturation mutagenesis on some residues of the CBL2 loop was used to increase the binding

affinity of N-Epa1p for fibronectin, which was selected as a model of a human glycoprotein. Two

adhesin mutants, E227A and Y228W, with improved binding features were obtained. More

importantly, a glycan array screening revealed that single point mutations in the CBL2 could produce

significant changes in the carbohydrate specificity of the protein. In particular, lectin molecules were

generated with a high affinity for sulfated glycans, which may find an application as molecular probes

for the identification of 6-sulfogalactose containing glycans and glycoconjugates.

Keywords: Candida glabrata / Epa1 / epithelial adhesin / glycan array / saturation mutagenesis.

by guest on July 21, 2014http://glycob.oxfordjournals.org/

Dow

nloaded from


3

Introduction

The yeast Candida glabrata is a commensal member of the human microbiome. It is especially

localized in the mucosae of different organs, where it doesn’t represent a threat for individuals in

healthy conditions. However, it may become a problem in patients whose immune system has been

compromised, including HIV seropositive patients or organ transplant acceptors (Fidel et al., 1999;

Pelroth et al., 2007). In combination with the more virulent C. albicans, it can cause localized

infections, such as oropharyngeal candidiasis, esophagitis, vulvovaginitis and urinary tract infections,

or in the worst cases even systemic candidiasis. C. glabrata and other non-albicans Candida species

are considered nowadays as emerging opportunistic organisms, as they represent the etiological

agents of an increasing number of fungal infections. This depends especially on the resistance of

these yeasts to several antimicrobial agents (Miceli et al., 2011).

Adherence of C. glabrata to human epithelial cells mainly relies on Epa (epithelial adhesins)

proteins, especially on Epa1p, Epa6p and Epa7p (Cormack et al., 1999; Frieman et al., 2002;

Castano et al., 2005), which are endowed with a calcium-dependent and lactose-sensitive lectin

functionality (Cormack et al., 1999). Epa1p and Epa7p can also mediate adherence to endothelial

cells (Zupancic et al., 2008), while Epa1 can adhere to macrophages and peripheral blood

mononuclear cells (Kuhn and Vyas, 2012). The Epa proteins are glycosylphosphatidylinositol-

anchored cell wall proteins (GPI-CWP), characterized by a well-defined modular structure (Frieman et

al., 2002). The N-terminal domain confers the protein the ability to adhere to the host cells, while the

central serine/threonine-rich region has the structural role of extending the ‘’sticky’’ domain outwards

of the cell wall. The C-terminus is modified with a GPI function and keeps the protein attached to the

cell wall.

The structure and function of Epa proteins has been thoroughly studied not only at the cellular

level, but also at the molecular level. An in-depth binding characterization of the N-terminal domain of

Epa proteins (N-Epa1p, N-Epa6p, N-Epa7p) has been performed by means of glycan array analysis

(Zupancic et al., 2008; Maestre-Reyna et al., 2012). N-Epa1p can recognize galactose-containing

glycans with a specificity for -1,3- and -1,4-linked galactose moieties, but it shows preference for

glycan structures containing the core 1 structure of mucin-type O-glycans, also known as the T-

antigen (Gal-1,3-GalNAc). The 3D structure of N-Epa1p has been solved by X-ray crystallography

(Ielasi et al., 2012). The adhesin is able to recognize carbohydrate ligands with micromolar affinity via


Dow

nloaded from


4

a PA14 domain (Rigden et al., 2004). This domain has a PA14-like -sandwich topology and contains

a calcium-dependent carbohydrate binding pocket. The CBL2 loop, a variable loop whose amino acid

side chains regulate the specificity and the promiscuity of the binding pocket (Ielasi et al., 2012;

Maestre-Reyna et al., 2012), is important for the interaction with disaccharides and larger glycan

determinants.

Fibronectin (FN) is a large glycoprotein assembled from two 250 kDa subunits and

characterized by three different types of sequence repeats (Buck and Horwitz, 1987; Pankov and

Yamada, 2002). It contains different binding sites for other molecules of the extracellular matrix

(ECM), including proteins like fibrin, collagen and integrins, as well as glycosaminoglycans such as

heparin. Several forms of this protein (at least 20) are present in the human body. They are all

encoded by the same gene, and based on their solubility, they can be classified either as plasma FN

or as cellular FN. The former is a soluble form of the glycoprotein; it is present in blood plasma and

involved in wound healing and blood clotting by interacting with platelets and fibrin (Lenselink, 2013).

Fibronectin glycosylation mainly consists of complex-type N-glycans with non-reducing terminal

lactosamine moieties. The presence of additional sialic acid or fucose residues is typical for

respectively the soluble plasmatic and cellular form of fibronectin (Takasaki et al., 1980; Wang et al.,

1990; Tajiri et al., 2005). Mucin-type O-glycosylation was also found on plasma fibronectin,

particularly in the forms of T-antigen and sialyl-T-antigen (Tajiri et al., 2005).

Several strategies have been used so far to engineer carbohydrate-binding properties of

lectins, including random mutagenesis (Yabe et al., 2006), site-directed mutagenesis (Salomonsson

et al., 2010) or the combination of these methods (Hu et al., 2012). An additional semi-random

technique is saturation mutagenesis, which allows generating all possible mutations for a specific site

and has been successfully used to change the carbohydrate specificity of structurally characterized

proteins (Imamura et al., 2011; Hu et al., 2013).

In this work, we use saturation mutagenesis in combination with an ELISA-based detection

method to engineer the carbohydrate binding properties of N-Epa1p. First, we found that fibronectin,

purified from human plasma, is recognized by N-Epa1p with submicromolar affinity and that its

binding with the adhesin domain is lactose sensitive. Next, fibronectin was selected as a model ligand

for N-Epa1p, to assess the suitability of a saturation mutagenesis method for adhesin engineering.

Specifically, we aimed to increase the N-Epa1p affinity for glycosylated fibronectin. Two positions in


Dow

nloaded from


5

the N-Epa1p variable loop CBL2, namely E227 and Y228 (Figure 2A), were selected as targets for

mutagenesis, since they are critical for the interaction with large glycan structures (Maestre-Reyna et

al., 2012). The binding properties of the selected hits from the mutant screening were characterized

quantitatively by surface plasmon resonance, and qualitatively by glycan array screening. It was found

that specific mutations in the CBL2 loop not only increased the affinity for the glycoprotein but also

modified the carbohydrate specificity, specifically towards the recognition of sulfated glycan moieties.

Results

Binding of wild type N-Epa1p to fibronectin

Surface Plasmon Resonance (SPR) was used to measure the affinity of N-Epa1p for fibronectin, that

was immobilized onto the surface of a sensor chip (Figure 1A). The equilibrium dissociation constant

(KD) value was estimated at 911 nM. In order to verify that the observed interaction was specifically

mediated by galactose-containing glycans attached to fibronectin, binding inhibition experiments were

performed by applying solutions with a constant N-Epa1p concentration and increasing

concentrations of lactose or glucose to the sensor chips (Figure 1B). The binding of N-Epa1p to

fibronectin could be blocked by lactose in a concentration-dependent manner, but was not affected by

the presence of glucose.

Optimization of binding detection

A detection system based on an ELLA setup was chosen to detect binding of mutants of the Epa1p

lectin domain to fibronectin. This system is based on the spectrophotometric identification of the lectin

– anti-His antibody – anti-mouse-alkaline phophatase (AP) antibody complexes bound to the

immobilized glycoprotein ligand. Prior to the mutant screenings, the conditions for the assays had to

be optimized, especially in terms of the amount of glycoprotein ligand immobilized in the microtiter

plate. Other elements also had to be checked, such as the absence of any non-specific binding event

taking place in parallel to the N-Epa1p - fibronectin interaction and the reproducibility of the detection.

A preliminary assay was performed by using a 96-well plate with different amounts of

immobilized fibronectin. Glycoprotein solutions of 10, 5, 2.5, 1 and 0 g/ml and 2% w/v BSA were

used for ligand coating and blocking, respectively. The plate was incubated with cell lysate samples,

coming from a wild-type-only culture plate, and lysis buffer as a negative control. The increase in


Dow

nloaded from


6

absorbance values over time showed linearity in a time range of 1 h and a ligand concentration-

dependent trend. This revealed that the slope, and thus binding onto fibronectin, is correlated with the

amount of fibronectin present on the bottom of the wells. The extremely low absorbance values in

wells coated only with BSA, allowed to rule out any non-specific binding of N-Epa1p to the wells, while

the low signals detected for the wells coated with fibronectin and incubated just with lysis buffer

excluded any non-specific interaction between the antibodies used for the detection and the

immobilized glycoprotein. Plotting the slopes versus the respective fibronectin concentration in

solution didn’t show an inflection at the highest concentration. From this, it was possible to deduce

with sufficient confidence that no binding saturation of the wild-type N-Epa1p was reached with the

maximum amount of immobilized fibronectin. Thus, it was chosen to use a 10 g/ml protein solution

for well coating in the following assays, which allows obtaining appreciable absorbance values during

binding detection.

A second preliminary assay was conducted to determine the variability of the detection

method. Wells coated with the same amount of fibronectin were incubated with cell lysate samples

coming from a wild-type culture plate. Binding was detected in the same way as described before,

and the slopes were calculated over 1 h of absorbance data. The coefficient of variability (CV) of 38

slope values from the same number of wells, was calculated to be 22.3%. This variability can be

considered as reasonably low, considering that crude cell lysate (from different bacterial colonies) and

unpurified N-Epa1p samples were used, and slight differences among protein expression yields may

exist.

Detection of mutants with improved ligand binding affinity

The optimized ELLA method was used to identify N-Epa1p mutants with higher binding affinities for

fibronectin. Libraries of variants with E227 and Y228 single-point mutations were screened (Figure

2B). From the E227 library, only 6 clones satisfied our selection criterion (Avg+2xSD), while other

clones showed similar or lower binding compared to the wild-type N-Epa1p. Sequencing of the hits

revealed three times the mutant E227G and twice the mutant E227A. The top hit turned out to be a

wild-type construct, which must be a false-positive result since no mutations could be found in the

complete sequenced region (between the T7 promoter and the T7 terminator of the plasmid).


Dow

nloaded from


7

The Y228 library generated more than double the amount of hits, and among the 14 clones

satisfying the selection criteria, the 8 highest-ranking were subjected to further investigation.

Sequencing detected 7 times the Y228W mutant (present in the 5 highest positions in the ranking)

and once the Y228A mutant. Generally, 60% of the constructs showed activity similar to or lower than

the wild-type construct.

Binding characterization of the N-Epa1p mutants

Interactions of the identified N-Epa1p mutants with fibronectin were further investigated by means of

SPR. This characterization was essential to rule out the possibility that the hits, obtained from the

initial ELLA screening, are simply related to different protein yields from the bacterial expression

system. Solutions containing different concentrations of the mutant proteins were injected onto sensor

chips functionalized with the glycoprotein ligand. Response values at equilibrium were extracted from

sensorgrams (Figure 1C) and used to calculate equilibrium dissociation constants. Analysis of N-

Epa1p variants revealed a substantially unchanged affinity for the Y228A hit (1.15 M), a decreased

affinity for the E227G hit (1.72 M) and an increased affinity for the E227A and Y228W hits (317 nM

and 545 nM, respectively).

Glycan array screening of the N-Epa1p mutants

We subjected the four hit mutants to a semi-quantitative glycan array screening in order to determine

if a single-point mutation can produce significant changes in carbohydrate specificities of N-Epa1p.

We found indeed some major changes in ligand specificity for some of the mutants, tested at 200

g/ml on the glycan arrays (Figure S2 and Table S1). Particularly, the E227A mutant could bind

strongly all lactose and lactosamine structures with a sulfated hydroxyl on galactose C6 ([6-SO3-

]Gal-1,4-Glc/GlcNAc, glycan n. 42-45, 297), and among these, preference was found for a second

sulfation on glucose/glucosamine 6-OH (glycan n. 45 and 297). While the mutation retained the

affinity for glycans containing Gal-1,3–GalNAc and Gal-1,3–GlcNAc moieties, it changed in regards

to the recognition of the sulfated structures, which became the highest-affinity carbohydrate ligands

among all the carbohydrates present in the glycan array. A completely new ligand has also been

found, i.e. a lactosamine with a phosphate group on the galactose 6-OH ([6-PO4-]Gal-1,4-GlcNAc,

glycan n. 518), and it binds to N-Epa1 with a higher specificity than lactosamine.


Dow

nloaded from


8

The only major variation in the glycan array profile of Y228W, compared to the wild type

adhesin, is the decreased binding to glycans containing a -1,4-linked galactose (glycan n. 151-172)

and to complex N-glycans and related branches (glycan n. 351-575). On the other hand, a change in

specificity similar to that observed for E227A, was observed for the E227G variant. Although the

interaction with glycans is significantly diminished, this N-Epa1p variant is still able to recognize

lactose and lactosamine with a sulfate group on both C6 hydroxyl groups or with a single sulfate on

the glucose ring (glycan n. 45, 155, 156), but not the analogues ones with only one sulfate on the

galactopyranose ring. Next best binders include structures with a -1,3-linked galactose, and the

mutant is not able anymore to recognize highly-branched glycan structures. This results in a N-Epa1p

variant characterized by lower affinity for carbohydrates, but also with a very narrow specificity. The

variant Y228A shows also increased binding to the disaccharides containing two sulfated hydroxyls

on the C6, but it recognizes mostly Gal-1,3-containing substructures and, unlike the wild type,

preferentially interacts with the T-antigen epimer Gal-1,3-GlcNAc, especially if a sulfate group on the

reducing GlcNAc ring is present (glycan n. 444 and 510).

A more detailed analysis of N-Epa1p mutant specificities was performed by the interpretation

of glycan array screening results obtained with 20 g/ml protein samples (Figure 3A and Figure S3).

For the wild-type construct and the mutants, the binding of different linear glycan determinants,

normalized to the binding of the T-antigen, was evaluated, (Figure 3B). A similar binding analysis,

relative to the best binder of wild-type N-Epa1p, has been already performed for subtype-switched N-

Epa1p variants (Maestre-Reyna et al., 2012) and gave good indications of the changes in substrate

specificities. In our case, a purely qualitative comparison of the mutant binding to simple glycan

determinants with the binding to Gal-1,3-GlcNAc allowed to understand how the single mutations

influenced the ligand preference for each hit mutant.

The array results show a significantly reduced number of high affinity ligands for N-Epa1p

mutants at this concentration, and, especially in the case of E227A, E227G and Y228A, the

generation of very sharp carbohydrate affinities. Sulfated lactose and lactosamine derivatives have

already been indicated as the best ligands for E227A and E227G variants. For the first variant, one

sulfation on the galactose residue is already enough to have significant binding, while for the second

variant, sulfation on the glucose moiety is required for the interaction with the carbohydrate. In both

cases, a second sulfate group results in a higher affinity. The Y228A mutant partially follows the trend


Dow

nloaded from


9

of E227G, but preferentially recognizes Gal-1,3-[6-SO3-]GlcNAc and [6-SO3

-]Gal-1,3-[6-SO3-

]GlcNAc, with no real distinction between the two disaccharides.

Concerning the glycan series containing a Gal-1,3 moiety (Figure S4), one difference arises

for the Gal-1,3-GalNAcstructure containing the reducing GalNAc anomer (glycan n. 139-141). In

this case, all mutants show a reduced binding specificity and preference for the anomer in contrast

to the relative promiscuity of the wild-type adhesin. From the 20 g/ml data analysis, the increased

affinity of the Y228A mutant for Gal-1,3-GlcNAc is evident, and, even more striking is the preference

for the two Gal-1,3-Gal - containing structures, which are the best ligands for this N-Epa1 variant. A

general decrease in specificity is observed for some linear Gal-1,4 series (Figure S4), with an

exception for Y228A, relatively to lactose and dimers of lactosamine linked by a 1,6-glycosidic

linkage. Additionally, the same variant of N-Epa1p cannot distinguish between the T-antigen and

Gal-1,4-GlcNAc.

Molecular docking of sulfated disaccharides to N-Epa1 variant structures

An attempt to explain the structural basis of the increased specificity of N-Epa1p mutants for sulfated

disaccharides was made by performing computational docking simulations with the N-Epa1p

structures (wild type, E227A and Y228A) and models of the sulfated disaccharides [6-SO3-]Gal-1,3-

[6-SO3-]GlcNAc and [6-SO3

-]Gal-1,4-[6-SO3-]GlcNAc. Different docking conformations were scored

according to their free energies of binding (the lower, the better), and visually inspected afterwards.

All simulations, independently of the ligand or the protein mutation, yielded a first solution in which the

sulfated galactose moiety coordinates the calcium divalent ion with the 3-OH and 4-OH groups. This

is consistent with the structural data currently available (Ielasi et al., 2012; Maestre-Reyna et al.,

2012) (Figure 4). Lower-ranking docking solutions were found, which predicted binding of the sulfated

carbohydrate via direct interaction between the sulfate group on the galactose moiety and calcium.

However, we did not consider these solutions in our binding hypotheses, due to the rather non-

specific character of these interactions and the significant difference, in terms of computed binding

energies, with the first solutions.

Docking of [6-SO3-]Gal-1,4-[6-SO3

-]GlcNAc to the wild type N-Epa1p yielded two highest-

ranking solutions, characterized by similar energies but flipped orientations of the GlcNAc ring. In

these two conformations, the sulfate group on the reducing carbohydrate is positioned close either to


Dow

nloaded from


10

K117 (Figure 4A) or R226. In the case of the E227A mutant, these two conformations are not

computed as energetically equivalent. The proximity with R226 of the sulphate-GlcNAc is predicted as

highly probable (Figure 4B), while significantly higher energies are attributed to all other solutions.

Similarly, the simulations with [6-SO3-]Gal-1,3-[6-SO3

-]GlcNAc, docked into the N-Epa1p

wild-type and Y228A binding sites, positioned the sulfate group on the reducing end close to R226. In

these cases, the orientation of the GlcNAc ring is different from the one determined for the GalNAc

ring in the N-Epa1p – T-antigen structure (Figure 2A) and the N-acetyl group is located closer to

residue 228. The orientations of the [6-SO3-]GlcNAc moiety, in the first docking solutions of both the

wild type (Figure 4C) and the mutant (Figure 4D) are identical, with similar predicted binding energies.

However, the carbohydrate is slightly tilted towards the R226 side chain in the wild type binding

pocket, likely generated by the repulsion between the tyrosine aromatic ring and the acetyl group on

the GlcNAc reducing end.

Discussion

We engineered the binding site of the N-terminal carbohydrate-binding domain of Epa1p in order to

improve its affinity towards fibronectin. Moreover, we investigated how the mutation of some key

amino acids in the binding pocket can affect the specificity of carbohydrate recognition by the

epithelial adhesin. Previously, the glycan specificity of a galectin from the mushroom Agrocybe

cylindracea (ACG) that was able to bind N-acetyl lactosamine and the T-antigen, has been modified

by saturation mutagenesis, targeting amino acids involved in glycan recognition. In a first case, the

mutation of E86 with an aspartate removed the affinity for Gal-1,4-GlcNAc and Gal-1,3-GalNAc

moieties, while the affinity for sialyl residues with an -2,3-glycosidic linkage was preserved (Imamura

et al., 2011). Also, the N46A mutant of the same lectin was characterized by increased affinity for the

GalNAc-1,3-Galdisaccharide and, consequently, for the blood group A and Forssman antigens (Hu

et al., 2013), while binding to other -galactosides was significantly decreased.

We used saturation mutagenesis to produce libraries of N-Epa1p variants, mutated in

positions E227 and Y228. These amino acids are part of the CBL2 variable loop, which is involved in

carbohydrate binding. Residues 227 and 228 in the N-Epa1p binding pocket can both contribute to

the degree of promiscuity in ligand binding (Maestre-Reyna et al., 2012). Structural and functional

analysis of Epa6 and Epa2 subtype-switched lectin domains revealed that the substitution of E227


Dow

nloaded from


11

and Y228 with less sterically hindering amino acids favors the accommodation of galactosides with

different glycosidic linkages, thus broadening glycan specificity. Moreover, it was suggested that

aromatic residues in position 228 are effective in packing interactions with the reducing ends of Gal-

1,3-Glc(NAc) or Gal-1,3-Gal(NAc) moieties.

Based on these considerations, we hypothesized that the higher affinity of the Y228W mutant

can be explained by an enhanced O-glycan binding, especially of -1,3-linked moieties, possibly

generated by the more extended aromatic -electron delocalization of the tryptophan side chain. On

the other hand, the higher affinity for fibronectin shown by the E227A mutant is less clear, but the

presence of 6-OH-sulfated moieties on fibronectin glycans may be the responsible factor. Covalent

modification of fibronectin with sulfated proteoglycans, such as chondroitin sulfate or dermatan

sulfate, has been described (Burtonwurster and Lust, 1993). Therefore, the N-Epa1p mutant could

possibly interact, more efficiently than the wild-type adhesin, with the terminal sulfated residues of

proteoglycan chains linked to fibronectin.

Overall, the main finding of our work is the dramatically increased binding, relative to the T-

antigen, to sulfated lactose and lactosamine disaccharides of the E227A and E227G mutants. This is

most likely caused by the absence of the glutamate side chain, which is supposed to eliminate the

electrostatic repulsion between the negatively charged carboxylic group and the sulfate groups

positioned on the ligands. Also, the E227G mutant specificity for a sulfate group on the reducing end

is possibly explained with the compensation of the missing E227 contribution in the disaccharide

binding with an interaction between the SO3- group and the near and positively charged R226. The

latter interaction doesn’t seem to stabilize the carbohydrate binding by the E227A mutant, whose

alanine may still hinder any interaction between the arginine side chain group and the sulfate function.

The Y228A mutant also showed enhanced specificity for sulfated disaccharides and, in

particular, a sulfate group on the reducing residue was found to be critical for binding of -1,3-linked

molecules. However, this is probably the result of a better accommodation of the sulfated

carbohydrates in the binding pocket due to the presence of a less sterically hindering alanine.

Molecular docking for this mutant suggested indeed an interaction between 6-SO3- groups on reducing

GlcNAc moieties and the positively charged R226. The steric repulsion between the carbohydrate N-

acetyl group and the side chain group of residue 228 would be sensibly reduced in the case of

Y228A, thus ligand binding would be favored, if compared to the wild-type adhesin. In this case,


Dow

nloaded from


12

docking experiments could only partially explain the difference in specificity between the native N-

Epa1p and the mutant adhesin.

Crystallization and X-ray structure determination of the N-Epa1p mutant complexes with

sulfated disaccharides, together with direct functional analysis of the protein – carbohydrate binding,

would clearly give further and interesting insights into these interactions. Unfortunately, further

experiments in this direction are limited due to the current unavailability of the mentioned sulfated

carbohydrate molecules in purified forms. Nonetheless, our results corroborate and extend recent

results where a similar improved affinity for sulfated galactosides, although less pronounced, was

found for the Epa2/Epa3-subtype switched N-Epa1p variants (Maestre-Reyna et al., 2012). Epa2p

has the sequence D227, N228, N229, instead of the wild-type Epa1p EYD sequence. Thus, the

removal of a negative charge and smaller side chains favor the presence of sulfate groups in the

binding site. The Epa3 variant has also an additional positive charge coming from a lysine in position

228 (GKD), but in the case of this variant binding to glycans is seriously impaired by a R226I

mutation.

Sulfated O-glycans that contain a 6-sulfo galactose moiety, can be found throughout the body

and are sometimes associated with tumors. For instance, they have been found on the mucin MUC1,

which are produced by breast cancer cells (Seko et al., 2012). More generally, sulfomucins are found

in different tissues of the gastrointestinal and urogenital tracts of the human body (Nieuw Amerongen

et al., 1998). Sulfated lactosamine also represents the repetition unit of the glycosaminoglycan

keratan sulfate (Reitsma et al., 2007; Fundemburgh, 2000), which is associated to corneal cells and

articular cartilage glycosylation. Keratan sulfate is also synthesized in the brain, and its production is

increased in some forms of brain tumors, included astroctyoma and glioblastoma (Kato et al., 2008;

Hayatsu et al., 2008).

At present, there is however a lack of molecular probes able to specifically recognize glycans

containing the 6S-Gal moiety. The TJA-I lectin, from the plant Tricosanthes japonica, can bind

sulfated galactose (Yamashita et al., 1992), but its additional interactions with sialo- and asialo-Gal

moieties make this protein not suitable for the specific identification of 6S-Gal on biological substrates.

Directed evolution of a ricin-type lectin (EW29Ch), on the other hand, has been previously carried out

for the generation of a 6-sulfogalactose-specific molecule by a combination of random and site-

directed mutagenesis, and a glycoconjugate array detection system (Hu et al., 2012). The resulting


Dow

nloaded from


13

6S-Gal–specific mutant, used in combination with the EW29Ch template, led to the specific in vitro

identification of the sulfated glycoepitope overexpressed onto 6-O-Gal-sulfotransferase-transfected

CHO cells.

Lectins represent a precious tool for glycan profiling and diagnostics. Applications include the

generation of affinity chromatography stationary phases for the purification of human antibodies and

other plasma glycoproteins (Kabir 1998), the structural characterisation of sugars linked to

glycoproteins (Kaji et al., 2003), and the detection of subtle differences in protein glycosylation due to

various diseases (Satish and Surolia, 2001). Moreover, the generation of microarray devices can be

of outstanding importance not only for the high-throughput analysis of the glycosylation pattern on

single glycoproteins, but also to investigate the carbohydrate expression on bacteria or mammalian

cells for diagnostic purposes (Tateno et al., 2007; Gemeiner et al., 2009; Smith and Cummings,

2013). Tailoring of Epa adhesins, and other yeast adhesin for the recognition of target glycoproteins

or even specific glycan molecules could thus represent a novel source of proteins with lectin activity to

be used as probes in different applications.

In conclusion, the engineering of the CBL2 loop in the Epa1p N-terminal domain led to the

generation of two single-point mutants endowed with higher binding affinity for fibronectin. More

strikingly, some of the mutations produced an important change in specificity, dramatically restricting

the range of possible structures recognized by the adhesin lectin domain. We propose that this

method can be applied to increase the recognition of other glycoproteins by any Epa adhesin from C.

glabrata, as well as by other adhesins with a lectin-like activity, such as the Flo adhesins from S.

cerevisiae (Veelders et al., 2010; Ielasi et al., 2013). Furthermore, other strategies could be applied to

modulate protein specificity towards glycan molecules, for example the combination of the

characterized single-point mutations into double mutants, random mutagenesis or the modification of

other key amino acids in the binding pocket, such as R226 or D229.

Materials and methods

Preparation of DNA libraries

The gene encoding the N-terminal lectin domain of Epa1 (N-Epa1p), inserted into the pET-21b(+)

vector (Ielasi et al., 2012), was randomized at positions E227 and Y228 by site-saturation

mutagenesis. Libraries of mutants were generated using the forward primers Fw_E227NNS (5’-


Dow

nloaded from


14

TAGGTTATTTTATAATAACAGANNSTATGATGGTGCACTCAG-3’), Fw_Y228NNS (5’-

TAGGTTATTTTATAATAACAGAGAANNSGATGGTGCACTCAGTTTTAC-3’), the reverse primer

Rv_MP (5’-TACGATACGGGAGGGCTTAC-3’) and the Phusion High Fidelity DNA Polymerase (New

England Biolabs) in a modified megaprimer-whole plasmid protocol (Sanchis et al., 2008). A first PCR

amplification of a 1700 bp megaprimer was run, using final concentrations of 0.2 ng/µl for the plasmid

template, 0.25 M for the primers and 0.2 mM for the dNTP mixture (30 s at 98°C, followed by 30

cycles of 10 s at 98°C, 20 s at 50°C and 40 s at 72°C, with a final 5 min at 72°C). The amplified DNA

megaprimers mixtures were treated with DpnI to remove the plasmid template, and then used for a

whole plasmid PCR. This required 0.2 ng/µl of template, 4 ng/µl of megaprimers and 0.2 mM of dNTP

mixture (30 s at 98°C, followed by 25 cycles of 10 s at 98°C and 4 min at 72°C, with a final 5 min at

72°C). The purification of the plasmid libraries was performed with a QIAprep Miniprep kit (Qiagen).

The variability of mutated codons was checked by DNA sequencing.

Enzyme linked lectin assay (ELLA) screens

DNA libraries were used to transform electrocompetent Escherichia coli Origami 2 (DE3) cells. The

transformed cells were plated on agar solid medium supplemented with 100 g/ml ampicillin and

grown overnight at 37°C. Colonies were picked from cultures with the QPix II system (Genetix) and

used to inoculate 96-well plates filled with 175 µl LB and 100 g/ml ampicillin. For each library, two

plates were prepared in order to have sufficient codon variability. These were shaken overnight at

37°C, and then supplemented with glycerol to a final concentration of 20% v/v in order to allow

storage at -20°C.

For screening experiments, master plates were used to inoculate other LB-ampicillin 96-well

plates. After overnight growth at 37°C, cultures were induced for protein expression with isopropyl β-

D-1-thiogalactopyranoside (IPTG) to a final concentration of 50 M and shaken for 3 days at 12°C.

Cells were harvested at 4500 rpm for 30 min, and frozen overnight at -80°C. For cell lysis, pellets

were resuspended in a buffer containing 1 mg/ml lysozyme, 0.05 % w/v bovine serum albumin (BSA),

50 mM Tris-HCl, 50 M NaCl, 10 mM CaCl2, 4 mM MgCl2 and 0.1 mM phenylmetylsulfonyl fluoride

(PMSF) as protease inhibitor, and incubated for 1 h at room temperature. Bacterial lysates were

incubated for 1 h at room temperature in MaxiSorp plates (Nunc), previously coated with fibronectin

(BD Biosciences, 10 g/ml protein in 100 mM NaHCO3 buffer pH 9.5, overnight incubation at 4°C,

followed by blocking with 2% w/v BSA in PBS). Afterwards, wells were incubated for 1 h with anti-His


Dow

nloaded from


15

tag mouse primary antibody (AbD Serotec) in PBS - BSA 0.2% w/v, followed by 1 h with anti-mouse –

alkaline phosphatase (AP) secondary antibody (Sigma) in the same buffer, and the AP substrate 2,4-

dinitrophenyl phosphate (DNPP) in 100 mM Tris-HCl pH 9.5, 5 mM MgCl2, 100 mM NaCl. Between

the incubation steps, the wells were washed 4 times with 0.05% Tween-20 in PBS. After the addition

of the AP substrate, the increase in absorbance at 405 nm was followed during 60 min, with one

measurement every minute (Biochrom Anthos Zenyth 200rt microplate reader).

To evaluate differences in binding among N-Epa1p mutants to fibronectin, kinetic plots were

constructed for all wells. The wells were ranked according to the slope values (which were assumed

as proportional to the amount of bound mutants) and these were compared to the average slope

obtained from a plate containing only wild type N-Epa1p. Mutants with a slope that exceeded the

average value (Avg) for twice the standard deviation (SD), were considered as hits and identified by

DNA sequencing.

Protein purification

Purification of N-Epa1p for binding assays was performed following the procedure already indicated

elsewhere (Ielasi et al., 2012).

Surface Plasmon Resonance (SPR)

The binding measurements were performed using the BIAcore 3000 system (GE Healthcare). The

increasing concentrations of N-Epa1p variants were injected over CM5 chips, on which fibronectin

was immobilized via an amine coupling method. A reference cell was coated with an equal amount of

BSA. The running buffer used was 10 mM CaCl2, 150 mM NaCl, 0.005% v/v Tween 20 and 10 mM

Hepes pH 7.4 (HBS). The injection was performed at 25°C using a flow rate of 20 μl/min for 2 min.

The dissociation was then monitored for 7 min. After the dissociation phase, the chip surface was

regenerated with 5 mM NaOH. Binding was determined by measuring the increase in resonance units

after subtraction of the background response obtained from the reference flow cell and the sample

containing only the buffer. The dissociation constants at the equilibrium state (KD) were determined

from binding experiments with a single set of concentrations and estimated using the following

steady-state affinity model with a 1:1 ligand-analyte ratio: Req = Rmax (KACA)/(KACA+1) where Req is the

response at the equilibrium state, KA is the association constant at the equilibrium state, CA is the


Dow

nloaded from


16

analyte concentration and Rmax is the maximum binding capacity for the particular experimental

condition. The results were analysed with the BIAevaluation software version 4.1 (GE Healthcare) and

with Prism 6 (GraphPad) software.

Glycan array analysis

N-Epa1 wild type and mutants were subjected to glycan array screening for binding to glycans printed

on a glass slide microarray (version 5.1) developed by the Consortium for Functional Glycomics (Blixt

et al., 2004) Screenings were performed at concentrations of 20 μg/ml and 200 μg/ml. The detection

of the proteins bound to the arrays was achieved by using the same anti-His tag primary and anti-

mouse-AP secondary antibodies as employed for the ELLA screenings. The average relative

fluorescence units were obtained for four replicates for each glycan. Error bars are based on the

standard error of the mean (SEM) for these replicates.

Molecular docking simulation

Generation of the models for the E227A and Y228A mutants and the [6-SO3-]Gal-1,3-[6-SO3

-]GlcNAc

and [6-SO3-]Gal-1,4-[6-SO3

-]GlcNAc ligands as well as the computational docking of the latter to the

N-Epa1 wild type (PDB code: 4A3X) (Ielasi et al., 2012) and mutant structures was carried out using

the molecular modeling program YASARA and the YASARA/WHATIF twinset (Vriend, 1990; Krieger

et al., 2002). N-Epa1 structures were prepared for docking by adjusting the pH to 7 ensure the correct

charge of the side chains and by optimizing the hydrogen network (Hooft et al., 1996) using the

YASARA/WHATIF twinset. A cubic search grid of 20 Å was defined to cover the N-Epa1binding site.

The flexibility of ligands was accounted for by allowing the rotation around flexible torsion angles.

Docking was subsequently performed with VINA (Trott and Olson, 2010) using the AMBER03 force

field for charge assignment (Duan et al., 2003) with default parameters, and all residues in the pocket

were kept fixed. The setup for docking was done with YASARA Structure (Krieger et al., 2002) and

the top hits out of hundred runs were selected for further analysis.

Funding

This work was supported by the Belgian Federal Science Policy Office (Belspo), the European Space

Agency (ESA) PRODEX program and the Research Council of the Vrije Universiteit Brussel.


Dow

nloaded from


17

Acknowledgements

We wish to acknowledge the Consortium for Functional Glycomics

(http://www.functionalglycomics.org) for the glycan analysis. FSI would like to acknowledge the

Agency for Innovation by Science and Technology (IWT, Belgium) for his PhD grant.

References

Blixt O, Head S, Mondala T, Scanlan C, Huflejt ME, Alvarez R, Bryan MC, Fazio F, Calarese D,

Stevens J, et al. 2004. Printed covalent glycan array for ligand profiling of diverse glycan binding

proteins. Proc Natl Acad Sci USA 101:17033–17038.

Buck CA, and Horwitz AF. 1987. Cell surface receptors for extracellular matrix molecules. Annu Rev

Cell Biol 3:179–205.

Burtonwurster N, and Lust G. 1993. Evidence for a glycosaminoglycan chain on a portion of articular

cartilage fibronectins. Arch Biochem Biophys 306:309–320.

Castaño I, Pan SJ, Zupancic M, Hennequin C, Dujon B, and Cormack BP. 2005. Telomere length

control and transcriptional regulation of subtelomeric adhesins in Candida glabrata. Mol Microbiol

55:1246–1258.

Cormack BP, Ghori N, and Falkow S. 1999. An adhesin of the yeast pathogen Candida glabrata

mediating adherence to human epithelial cells. Science 285:578–582.

Duan Y, Wu C, Chowdhury S., Lee MC, Xiong G, Zhang W, Yang R, Cieplak P, Luo R, Lee T, et al.

2003. A point-charge force field for molecular mechanics simulations of proteins based on

condensed-phase quantum mechanical calculations. J Comput Chem 24:1999–2012.

Fidel PL Jr, Vazquez JA, and Sobel JD. 1999. Candida glabrata: review of epidemiology,

pathogenesis, and clinical disease with comparison to C. albicans. Clin Microbiol Rev 12:80–96.

Frieman MB, McCaffery JM, and Cormack BP. 2002. Modular domain structure in the Candida

glabrata adhesin Epa1p, a beta1,6 glucan-cross-linked cell wall protein. Mol Microbiol 46:479–492.

Funderburgh JL. 2000. Keratan sulfate: structure, biosynthesis, and function. Glycobiology 10:951–

958.

Gemeiner P, Mislovicová D, Tkác J, Svitel J, Pätoprstý V, Hrabárová E, Kogan G, and Kozár T. 2009.

Lectinomics II. A highway to biomedical/clinical diagnostics. Biotechnol Adv 27:1–15.


Dow

nloaded from


18

Hayatsu N, Ogasawara S, Kaneko MK, Kato Y, and Narimatsu H. 2008. Expression of highly sulfated

keratan sulfate synthesized in human glioblastoma cells. Biochem Biophys Res Commun 368:217–

222.

Hooft RW, Sander C, and Vriend G. 1996.. Positioning hydrogen atoms by optimizing hydrogen-bond

networks in protein structures. Proteins 26: 363–376.

Hu D, Tateno H, Kuno A, Yabe R, and Hirabayashi J. 2012. Directed evolution of lectins with sugar-

binding specificity for 6-sulfo-galactose. J Biol Chem 287:20313–20320.

Hu D, Tateno H, Sato T, Narimatsu H, and Hirabayashi J. 2013. Tailoring GalNAcα1-3Galβ-specific

lectins from a multi-specific fungal galectin: dramatic change of carbohydrate specificity by a single

amino-acid substitution. Biochem J 453:261–270.

Ielasi FS, Decanniere K, and Willaert RG. 2012. The epithelial adhesin 1 (Epa1p) from the human-

pathogenic yeast Candida glabrata: structural and functional study of the carbohydrate- binding

domain. Acta Crystallogr D Biol Crystallogr 68:210–217.

Ielasi FS, Goyal P, Sleutel M, Wohlkonig A, and Willaert RG. 2013. The mannose-specific lectin

domains of Flo1p from Saccharomyces cerevisiae and Lg-Flo1p from S. pastorianus: crystallization

and preliminary X-ray diffraction analysis of the adhesin-carbohydrate complexes. Acta Crystallogr

Sect F Struct Biol Cryst Commun 69: 779–782.

Imamura K, Takeuchi H, Yabe R, Tateno H, and Hirabayashi J. 2011. Engineering of the glycan-

binding specificity of Agrocybe cylindracea galectin towards α(2,3)-linked sialic acid by saturation

mutagenesis. J Biochem 150:545–552.

Kabir S. 1998. Jacalin: a jackfruit (Artocarpus heterophyllus) seed-derived lectin of versatile

applications in immunobiological research. J Immunol Methods 212:193–211.

Kaji H, Saito H, Yamauchi Y, Shinkawa T, Taoka M, Hirabayashi J, Kasai K, Takahashi N, and Isobe

T. 2003. Lectin affinity capture, isotope-coded tagging and mass spectrometry to identify N-linked

glycoproteins. Nat Biotechnol 21:667–672.

Kato Y, Hayatsu N, Kaneko MK, Ogasawara S, Hamano T, Takahashi S, Nishikawa R, Matsutani M,

Mishima K, and Narimatsu H. 2008.. Increased expression of highly sulfated keratan sulfate

synthesized in malignant astrocytic tumors. Biochem Biophys Res Commun 369:1041–1046.

Krieger E, Koraimann G, and Vriend G. 2002. Increasing the precision of comparative models with

YASARA NOVA--a self-parameterizing force field. Proteins 47:393–402.


Dow

nloaded from


19

Kuhn DM, and Vyas VK. 2012. The Candida glabrata adhesin Epa1p causes adhesion, phagocytosis,

and cytokine secretion by innate immune cells. FEMS Yeast Res 12:398–414.

Lenselink EA. 2013. Role of fibronectin in normal wound healing. Int Wound J 9999.

Maestre-Reyna M, Diderrich R, Veelders MS, Eulenburg G, Kalugin V, Brückner S, Keller P, Rupp S,

Mösch HU and Essen LO. 2012. Structural basis for promiscuity and specificity during Candida

glabrata invasion of host epithelia. Proc Natl Acad Sci USA 109:16864–16869.

Miceli MH, Díaz JA, and Lee SA. 2011. Emerging opportunistic yeast infections. Lancet Infect Dis

11:142–151.

Nieuw Amerongen AV, Bolscher JG, Bloemena E, and Veerman, EC. 1998. Sulfomucins in the

human body. Biol Chem 379:1–18.

Pankov R, and Yamada KM. 2002. Fibronectin at a glance. J Cell Sci 115:3861–3863.

Perlroth J, Choi B, and Spellberg B. 2007. Nosocomial fungal infections: epidemiology, diagnosis,

and treatment. Med Mycol 45:321–346.

Reitsma S, Slaaf DW, Vink H, van Zandvoort MAMJ., and oude Egbrink MGA. 2007. The endothelial

glycocalyx: composition, functions, and visualization. Pflugers Arch 454:345–359.

Rigden, D.J., Mello, LV, and Galperin, MY. 2004. The PA14 domain, a conserved all-beta domain in

bacterial toxins, enzymes, adhesins and signaling molecules. Trends Biochem. Sci. 29:335–339.

Salomonsson E, Carlsson MC, Osla V, Hendus-Altenburger R, Kahl-Knutson B, Oberg CT, Sundin A,

Nilsson R, Nordberg-Karlsson E, Nilsson UJ, et al. 2010. Mutational tuning of galectin-3 specificity

and biological function. J Biol Chem 285:35079–35091.

Sanchis J, Fernández L, Carballeira JD, Drone J, Gumulya Y, Höbenreich H, Kahakeaw D, Kille S,

Lohmer R, Peyralans JJP, et al. 2008. Improved PCR method for the creation of saturation

mutagenesis libraries in directed evolution: application to difficult-to-amplify templates. Appl Microbiol

Biotechnol 81:387–397.

Satish PR, and Surolia A. 2001. Exploiting lectin affinity chromatography in clinical diagnosis. J.

Biochem Biophys Methods 49, 625–640.

Seko A, Ohkura T, Ideo H, and Yamashita K. 2012. Novel O-linked glycans containing 6’-sulfo-

Gal/GalNAc of MUC1 secreted from human breast cancer YMB-S cells: possible carbohydrate

epitopes of KL-6(MUC1) monoclonal antibody. Glycobiology 22:181–195.


Dow

nloaded from


20

Smith DF, and Cummings RD. 2013. Application of microarrays for deciphering the structure and

function of the human glycome. Mol Cell Proteomics 12:902–912.

Tajiri M, Yoshida S, and Wada Y. 2005. Differential analysis of site-specific glycans on plasma and

cellular fibronectins: application of a hydrophilic affinity method for glycopeptide enrichment.

Glycobiology 15:1332–1340.

Takasaki S, Yamashita K, Suzuki K, and Kobata A. 1980. Structural studies of the sugar chains of

cold-insoluble globulin isolated from human plasma. J Biochem 88:1587–1594.

Tateno H, Uchiyama N, Kuno A, Togayachi A, Sato T, Narimatsu H, and Hirabayashi J. 2007. A novel

strategy for mammalian cell surface glycome profiling using lectin microarray. Glycobiology 17: 1138–

1146.

Trott O, and Olson AJ. 2010. AutoDock Vina: improving the speed and accuracy of docking with a

new scoring function, efficient optimization, and multithreading. J Comput Chem 31:455–461.

Veelders M, Brückner S, Ott D, Unverzagt C, Mösch H-U, and Essen L-O. 2010. Structural basis of

flocculin-mediated social behavior in yeast. Proc Natl Acad Sci U.S.A. 107: 22511–22516.

Vriend G. 1990. WHAT IF: a molecular modeling and drug design program. J Mol Graph 8:52–56.

Wang YM, Hare TR, Won B, Stowell CP, Scanlin TF, Glick MC, Hård K, van Kuik JA, and

Vliegenthart, JF. 1990. Additional fucosyl residues on membrane glycoproteins but not a secreted

glycoprotein from cystic fibrosis fibroblasts. Clin Chim Acta 188:193–210.

Yabe R, Suzuki R, Kuno A., Fujimoto Z, Jigami Y, and Hirabayashi J. 2007. Tailoring a novel sialic

acid-binding lectin from a ricin-B chain-like galactose-binding protein by natural evolution-mimicry. J

Biochem 141:389–399.

Yamashita K, Umetsu K, Suzuki T, and Ohkura T. 1992. Purification and characterization of a Neu5Ac

alpha 2-->6Gal beta 1-->4GlcNAc and HSO3(-)-->6Gal beta 1-->GlcNAc specific lectin in tuberous

roots of Trichosanthes japonica. Biochemistry 31: 11647–11650.

Zupancic ML, Frieman M, Smith D, Alvarez RA, Cummings RD, and Cormack BP. 2008. Glycan

microarray analysis of Candida glabrata adhesin ligand specificity. Mol Microbiol 68:547–559.


Dow

nloaded from


21

Figure 1 – SPR analysis of the interaction between immobilized fibronectin and N-Epa1p

variants in solution. For all experiments, fibronectin was immobilized on CM5 chips up to densities

of ~700 RU. Two-fold serial dilutions of the N-Epa1p wild type and mutants solutions, plus buffer only,

were used in binding experiments for the determination of equilibrium constants. KD was calculated

using the highest point of the sensorgrams association phases (~Req) and a one binding site fitting

model. Standard error and R2 values, referring to the fitting procedure, are indicated in the panels

together with the KD values. (A) Wild type N-Epa1p binding sensorgrams (left panel) (concentration

range: 9.6 µM - 75 nM) and fitting curve (right panel). (B) Binding inhibition experiment with increasing

concentrations (0 M and 6 M to 1.5 mM) of lactose (Lac, middle panels) and glucose (Glc, lower

panels). (C) N-Epa1p E227A (9.1 M – 71 nM) and Y228W (7.8 M – 30 nM) binding sensorgrams

(left panels) and fitting curves (right panels).

Figure 2 – Screening of N-Epa1p mutant libraries. (A) The structure of N-Epa1p binding pocket, in

complex with the T-antigen (PDB code 4ASL, Maestre-Reyna et al., 2012) is shown. The mutant

libraries were generated by saturation mutagenesis of residues E227 and Y228, which are two key

residues in the N-Epa1p carbohydrate binding pocket. (B) The ranking of the library constructs was

based on their respective absorbance signal slopes. The red line in each subpanel indicates the slope

limit value, used as selection criterion. The colonies above the line were considered as hits and

chosen for DNA sequencing and further characterization. This ranking is related to one of the two

library well plates. No binding activity, above the slope limit value, was detected in the second plate.

Figure 3 – Glycan array binding profiles of N-Epa1p variants. (A) The glycan array profiles

showed here were obtained from 20 g/ml protein solutions. Graphical representations (CFG

symbols) of the highest-affinity glycan ligands on the array are reported on each chart. (B) Analysis of

N-Epa1p variants specificity related to the sulfate glycan series. The bar charts were obtained by

dividing in each glycan array set the average intensity values and the related standard deviations of

each glycan for the average intensity value measured for the T-antigen.

Figure 4 – Molecular docking results for N-Epa1p variants. The highest-ranked solutions for [6-

SO3-]Gal-1,4-[6-SO3

-]GlcNAc - N-Epa1p wild type (A), [6-SO3-]Gal-1,4-[6-SO3

-]GlcNAc – E227A (B),


Dow

nloaded from


22

[6-SO3-]Gal-1,3-[6-SO3

-]GlcNAc – wild type (C) and [6-SO3-]Gal-1,3-[6-SO3

-]GlcNAc – Y228A (D)

docking simulations are reported as stereo views of the four binding pockets. The calcium ion is

depicted as a green sphere in the three subpanels, while the mutated residues are indicated with bold

labels. The [6-SO3-]Gal-1,4-[6-SO3

-]GlcNAc ligand (panels A and B) is depicted in cyan, while the [6-

SO3-]Gal-1,3-[6-SO3

-]GlcNAc ligand is depicted in dark green (panel C and D).


Dow

nloaded from


23


Dow

nloaded from


24


Dow

nloaded from


25


Dow

nloaded from


26


Dow

nloaded from


Engineering the carbohydrate binding site of Epa1p from Candida...

Documents

Transcript of Engineering the carbohydrate binding site of Epa1p from Candida...