http://informahealthcare.com/bmgISSN: 1040-9238 (print), 1549-7798 (electronic)

Editor: Michael M. CoxCrit Rev Biochem Mol Biol, Early Online: 1–19

! 2013 Informa Healthcare USA, Inc. DOI: 10.3109/10409238.2013.831024

REVIEW ARTICLE

GPI Transamidase and GPI anchored proteins: Oncogenes andbiomarkers for cancer

Dilani G. Gamage and Tamara L. Hendrickson

Department of Chemistry, Wayne State University, Detroit, USA

Abstract

Cancer is second only to heart disease as a cause of death in the US, with a further negativeeconomic impact on society. Over the past decade, details have emerged which suggest thatdifferent glycosylphosphatidylinositol (GPI)-anchored proteins are fundamentally involved in arange of cancers. This post-translational glycolipid modification is introduced into proteins viathe action of the enzyme GPI transamidase (GPI-T). In 2004, PIG-U, one of the subunits of GPI-T,was identified as an oncogene in bladder cancer, offering a direct connection between GPI-Tand cancer. GPI-T is a membrane-bound, multi-subunit enzyme that is poorly understood, dueto its structural complexity and membrane solubility. This review is divided into three sections.First, we describe our current understanding of GPI-T, including what is known about eachsubunit and their roles in the GPI-T reaction. Next, we review the literature connecting GPI-T todifferent cancers with an emphasis on the variations in GPI-T subunit over-expression. Finally,we discuss some of the GPI-anchored proteins known to be involved in cancer onset andprogression and that serve as potential biomarkers for disease-selective therapies. Given thatfunctions for only one of GPI-T’s subunits have been robustly assigned, the separation betweenhealthy and malignant GPI-T activity is poorly defined.

Keywords

Cancer, glycosylphosphatidylinositolanchor, GPI, GPI anchored proteins,post-translational protein modifications

History

Received 6 June 2013Revised 10 July 2013Accepted 30 July 2013Published online 27 August 2013

Introduction

Cancer is caused by uncontrolled abnormal cell growth.

According to statistics from the American Cancer Society, in

the United States nearly one in two men and one in three

women will develop a cancer in his or her life-time. Cancer is

also a public health threat worldwide. Geographic variations

in cancer types and prevalence can arise from differences in

regional lifestyle, genetics, diet and pollution, amongst other

factors (American Cancer Society, 2011).

In addition to the emotional toll of cancer on patients,

family and friends, statistics from the American Cancer

Society illustrate the devastating economic impact of cancer

worldwide, which stem from direct costs for medical care and

rehabilitation and indirect costs from morbidity and mortality

(American Cancer Society, 2011). In order to reduce cancer

mortality and improve each patient’s quality of life, it is

important to understand how different oncogenes and bio-

markers participate in cancer onset, progression and metas-

tasis. Several well established cancer biomarkers, including

the urokinase plasminogen-activated receptor (uPAR) and the

folate receptor, are C-terminally modified with a glycosylpho-

sphatidylinositol (GPI) anchor (Lacey et al., 1989; Ploug

et al., 1991). This glycolipid anchor is clearly essential for

proper translocation of these proteins; however, evidence

supporting any further functional involvement for the GPI

anchor, particularly with respect to tumor phenotypes, was

minimal. In 2004, the discovery of the GPI anchor biosyn-

thesis class U protein (PIG-U) as an oncogene in human

bladder cancer opened a new door to the possibility that the

enzyme involved in GPI anchoring, called GPI transamidase

or GPI-T, might itself be tumorigenic (Guo et al., 2004). PIG-

U is one of the five subunits that comprise the human GPI-T

although the function of PIG-U in this enzyme is unknown

(Hong et al., 2003). The reaction catalyzed by GPI-T and the

chemical structure of a typical human GPI anchor are shown

in Figure 1.

GPI membrane anchoring of proteins is an abundant

phenomenon that specifically tethers proteins to lipid bilayers.

Approximately 0.5% of all eukaryotic proteins are modified or

predicted to be modified by GPI-T to contain a GPI anchor

(Eisenhaber et al., 2001). GPI anchored proteins are almost

exclusively localized on the cell surface where they are

non-covalently associated with the plasma membrane via the

lipid portion of the anchor. GPI anchored proteins are engaged

in diverse processes like immune recognition, cellular com-

munication, signal transduction and embryogenesis (Ferguson,

1999; Fujita & Jigami, 2008; Fujita & Kinoshita, 2012;

Nozaki et al., 1999; Paulick & Bertozzi, 2008). Loss of GPI

anchoring is embryonically lethal to mammals and condition-

ally lethal to yeast (Benghezal et al., 1996; Leidich et al., 1994;

Address for correspondence: Tamara L. Hendrickson, Department ofChemistry, Wayne State University, 5101 Cass Avenue, Detroit, MI48202, USA. Tel: 313-577-6914. Fax: 313-577-8822. E-mail: Tamara.Hendrickson@chem.wayne.edu

Cri

tical

Rev

iew

s in

Bio

chem

istr

y an

d M

olec

ular

Bio

logy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f Z

ueri

ch Z

entr

um f

uer

Zah

n M

und

und

on 0

8/31

/13

For

pers

onal

use

onl

y.

Nozaki et al., 1999). Defects in GPI anchor biosynthesis can

cause diseases like paroxysmal nocturnal hemoglobinuria and

hyperphosphatasia mental retardation syndrome and a muta-

tion in the PIG-T subunit of GPI-T is connected to an

intellectual disorder (Brodsky & Hu, 2006; Krawitz et al.,

2013; Kvarnung et al., 2013). These different diseases

highlight the importance of GPI anchoring of proteins to

normal cell biology. This review focuses specifically on the

connections that link GPI membrane anchoring to abnormal

cell biology and cancer.

GPI-T is a complicated and poorly understood enzyme.

GPI-T is a membrane-bound, multi-subunit protein complex

found in the endoplasmic reticulum (ER). This enzyme

contains five known subunits, PIG-K, PIG-T, GPAA1, PIG-S,

PIG-U in humans (Benghezal et al., 1996; Hong et al., 2003;

Ohishi et al., 2001) (analogous to Gpi8, Gpi16, Gaa1, Gpi17

and Gab1 in yeast (Benghezal et al., 1996; Fraering et al.,

2001; Hamburger et al., 1995; Hong et al., 2003)). PIG-U, the

last subunit of GPI-T was identified more than a decade ago

and yet clear functional assignments for all but one of these

subunits have remained elusive. The exception is PIG-K

(Gpi8): this subunit comprises the catalytic machinery of the

enzyme (Benghezal et al., 1996; Meyer et al., 2000).

Understanding how changes in gene over-expression

participate in tumor onset or progression is difficult without

a clear picture of the enzyme itself in terms of its structure

and function.

The GPI anchor: a substrate for GPI-T

GPI anchors contain a common core structure that is

conserved across eukaryotes and contains an ethanolamine

phosphate, three mannoses, a glucosamine and a phosphati-

dylinositol group (Figure 1). However, tissue- and species-

specific core modifications and elaborations were identified

in GPI anchors from different sources (Conzelmann et al.,

1992, Englund, 1993; Ferguson et al., 1988; Fujita &

Kinoshita, 2010; McConville & Ferguson, 1993). The com-

plete biosynthetic pathway to produce the GPI anchor was

fully revealed by the early 2000’s (Figure 2). This pathway

requires more than 20 gene products, making GPI anchoring

of proteins one of the most complex and metabolically

expensive post-translational modifications (Eisenhaber et al.,

2003; Fujita & Kinoshita, 2012; Masterson et al., 1989). The

different enzymes involved in GPI biosynthesis have been

characterized to varying extents. Most are hydrophobic and

reside in the ER membrane (recently reviewed by (Fujita &

Kinoshita, 2012)).

Figure 1. The chemical structure of the GPI anchor and the reaction catalyzed by GPI-T. (A) Left: The basic chemical structure of a human GPI anchorfrom nucleated cells is shown as a representative example. Variations of this structure occur in other cell types (not shown). Right: A simplified cartoonrepresentation of this GPI anchor. (B) GPI-T displaces the C-terminal signal sequence with the GPI anchor, forming a new amide bond between the!-site carbonyl and the terminal amine of the phosphoethanolamine group on the third mannose in the GPI anchor. Figure 5 shows the mechanism ofthis reaction in greater detail. (see colour version of this figure at www.informahealthcare.com/bmg).

2 D. G. Gamage & T. L. Hendrickson Crit Rev Biochem Mol Biol, Early Online: 1–19

Cri

tical

Rev

iew

s in

Bio

chem

istr

y an

d M

olec

ular

Bio

logy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f Z

ueri

ch Z

entr

um f

uer

Zah

n M

und

und

on 0

8/31

/13

For

pers

onal

use

onl

y.

GPI anchors are present in minute amounts in human and

fungal cells and are also challenging to synthesize and purify.

Due to this complexity, small nucleophiles like hydrazine,

hydroxylamine and biotin hydrazide were identified as useful

GPI anchor mimics (Ramalingam et al., 1996). These GPI

anchor surrogates were used early on to characterize the

reaction catalyzed by GPI-T.

Because of limitations faced when isolating GPI anchors

from their natural environments, it has remained challenging

to understand the contributions made by different monosac-

charides or modifications. Syntheses of short series of GPI

anchor analogs have been reported (John & Hendrickson,

2010; Paulick et al., 2007). Synthesis of the full-length CD52

peptide, with its N-linked glycan and most of its GPI anchor,

was reported about 10 years ago and was recently followed up

by its synthesis with a complete GPI anchor (from the human

lymphocyte CD52 antigen) (Burgula et al., 2012; Shao et al.,

2004). These synthetic compounds (full-length anchors,

synthetic GPI anchored proteins, anchor mimics and anchor

analogs) can be used not only to better understand GPI-T but

also to investigate the functions of GPI-anchored proteins in

cells and for vaccine development. In fact, synthetic GPI

glycans were used in microarray studies to examine antitoxic

malaria responses and to develop carbohydrate-based vac-

cines to treat severe malaria (Kamena et al., 2008; Schofield

et al., 2002). Another recent study used an azide-labeled

N-acetylgalactosamine analog (GalNAz) to understand the

immobilization of GPI anchored proteins inside the cell as

well as to analyze the functional roles of branch modifications

(Vainauskas et al., 2012).

Defects in different GPI anchor biosynthetic steps cause

several types of inheritable and acquired diseases. For a

recent review see Almeida et al. (2009). To our knowledge,

defects in GPI anchor biosynthesis have not been directly

linked to cancer onset or propagation. However, patients with

paroxysmal nocturnal hemoglobinuria (PNH), a hemolytic

Figure 2. Biosynthetic pathway for GPI anchors in human nucleated cells. The GPI anchor is synthesized in the ER starting from phosphoinositol.At least 10 enzymes are needed for this pathway; these enzymes are summarized in Table 1. The first two steps of the synthesis take place on thecytoplasmic side of the ER and steps 4 through 10 occur on the luminal side. Later steps in this pathway can vary in different types of cells. (see colourversion of this figure at www.informahealthcare.com/bmg).

Table 1. Enzymes involved in GPI anchor biosynthesis in human nucleated cells. Additional modification reactions are known to occur in other typesof cells.

Step Enzyme Proteins involved References

1 GPI-GlcNAc transferase PIG-A, PIG-C, PIG-H, PIG-P,PIG-Q, PIG-Y, DPM2

Murakami et al. (2005);Watanabe et al. (1998, 2000);Tiede et al. (2000)

2 GlcNAc-PI de-N-acetylase PIG-L Watanabe et al. (1999)3 Flippase Unknown Vishwakarma & Menon (2005)4 Inositol acyltransferase PIG-W Murakami et al. (2003)5 a1-4 mannosyltransferase I PIG-M, PIG-X Ashida et al. (2005);

Maeda et al. (2001)6 a1-6 mannosyltransferase II PIG-V Kang et al. (2005)7 EtNP transferase I PIG-N Hong et al. (1999)8 a1-2 mannosyltransferase III PIG-B Takahashi et al. (1996)9 a1-2 mannosyltransferase IV PIG-Z Taron et al. (2004)

10 EtNP transferase III PIG-O, PIG-F Hong et al. (2000)

DOI: 10.3109/10409238.2013.831024 GPI Transamidase and GPI anchored proteins 3

Cri

tical

Rev

iew

s in

Bio

chem

istr

y an

d M

olec

ular

Bio

logy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f Z

ueri

ch Z

entr

um f

uer

Zah

n M

und

und

on 0

8/31

/13

For

pers

onal

use

onl

y.

disorder that results from a somatic mutation in PIGA

(Table 1), are at an increased risk of developing acute

leukemia (Brodsky & Hu, 2006).

Protein substrates for GPI-T

Proteins designated to be GPI anchored are ribosomally

synthesized as preproproteins and contain N-terminal signal

sequences targeting them for translocation into the ER.

Historically, it has been assumed that substrates for GPI-T

enter the ER via the secretory recognition particle (SRP)

(Caras et al., 1989). However, a recent report made the

compelling argument that these preproproteins are predom-

inantly delivered to the ER by an SRP-independent pathway

(Ast et al., 2013). This process relies on recognition of both

the N-terminal signal peptide and the C-terminal GPI-T-

specific signal sequence (described below). Once the

preproprotein is delivered to the ER, the N-terminal signal

sequence is cleaved by signal peptidase. The resultant

proprotein is recognized by GPI-T and the C-terminal GPI-T

signal sequence is displaced upon conversion to the mature

GPI-anchored protein (Figure 3). The use of an SRP-

independent pathway for translocation to the ER clearly

defines GPI membrane anchoring as a post-translational

protein modification.

GPI-T recognizes and cleaves the C-terminal signal

sequence of the proprotein at the !-site, forming a new

amide bond between the !-site carbonyl and the appropriate

amine on the GPI anchor. The !-site is so named because it

becomes the C-terminal residue of the mature GPI-anchored

protein. This residue is immediately followed by the !þ 1

and !þ 2 residues (and so forth towards the C-terminus);

the remainder of this sequence is composed of a hydrophilic

spacer and a hydrophobic peptide (Eisenhaber et al., 1998;

Moran & Caras, 1991). Several studies have analyzed the

identity of the !-site amino acid and GPI-T’s ability to

tolerate substitutions. Most of this work relied on a protein

construct called preprominiPLAP, a minimalistic version of

human placental alkaline phosphatase. PreprominiPLAP

promoted significant advances in the field because it

contained a poly-Met sequence suitable for metabolic labeling

with 35S-Met and it was significantly smaller than native

PLAP so that the different processing intermediates could be

resolved by gel electrophoresis (namely, the preproprotein, the

proprotein, and the GPI-anchored protein, as well as a

truncated hydrolytic product) (Kodukula et al., 1991, Varma

& Hendrickson, 2010). Analysis of preprominiPLAP mutants

revealed that alanine, cysteine, glycine, asparagine and serine

are good !-site candidates for human GPI-T. Similar results

were obtained using human decay accelerating factor (hDAF),

another GPI-anchored protein; in this case, aspartate was

identified as a good !-site but cysteine was not (Micanovic

et al., 1990, Moran et al., 1991). In general, the !-site residue

should be a small, hydrophilic amino acid (Kodukula et al.,

1991). The first web server to predict the presence and

identity of ! sites in protein sequences was put forward in

1999, with a false positive prediction rate of only 0.3%

(Eisenhaber et al., 1999).

The !þ 1 position is typically small but can be any amino

acid other than proline. The requirements for !þ 2 are much

more stringent (Gerber et al., 1992, Kodukula et al., 1991).

This position is almost always alanine, glycine or serine

(Gerber et al., 1992; Kodukula et al., 1991). Because the ! to

!þ 2 positions tend to be small amino acids, this region has

Figure 3. Cartoon schematics of a protein substrate for GPI-T and its processing intermediates. The preproprotein (A), which is destined for GPIanchoring, has an N-terminal signal sequence that is cleaved by signal peptidase to produce the proprotein (B). The GPI-T signal sequence on theC-terminus of this protein contains a hydrophilic region followed by a hydrophobic region. The signal sequence is cleaved by GPI-T between the ! and!þ 1 amino acids to attach the GPI anchor, producing the mature, GPI anchored protein (C). (Refer to Figure 1 for the symbols used to designate theGPI anchor.). (see colour version of this figure at www.informahealthcare.com/bmg).

4 D. G. Gamage & T. L. Hendrickson Crit Rev Biochem Mol Biol, Early Online: 1–19

Cri

tical

Rev

iew

s in

Bio

chem

istr

y an

d M

olec

ular

Bio

logy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f Z

ueri

ch Z

entr

um f

uer

Zah

n M

und

und

on 0

8/31

/13

For

pers

onal

use

onl

y.

been referred to as the Small Amino acid Domain (SAD)

(Eisenhaber et al., 1998; Gerber et al., 1992).

The !þ 2 residue is followed by the remainder of the GPI-T

signal sequence. This peptide is typically between 18 and 32

amino acids long and ends at the C-terminus of the protein. It

can be broken down into two sections, an 8–12 amino acid

spacer sequence that is predominantly hydrophilic, followed by

a 15–20 amino acid hydrophobic sequence (Caras et al., 1989;

Eisenhaber et al., 1998; Moran & Caras, 1991). Remarkably,

the GPI-T C-terminal signal sequence does not contain a

consensus motif. In fact, in one report, completely artificial

signal sequences (e.g. Ser3-Thr8-Leu14) were appended onto

the C-terminus of CD46 and were shown to be viable, enabling

GPI anchoring in vivo (Coyne et al., 1993). Consequently,

recognition of this sequence by GPI-T is analogous to

recognition of the N-terminal secretory signal sequence by

signal peptidase, more than it is to the methods used by other

co- and post-translational modification enzymes to select their

substrates. Recent findings suggest that the hydrophobicity of

the hydrophobic region needs to be marginal compared to

type II transmembrane anchors (Galian et al., 2012). Similarly,

the hydrophilic spacer lacks a consensus sequence but the

relative hydrophilicity and the length of the peptide play

important roles (Coyne et al., 1993; Moran & Caras, 1991).

Amino acids N-terminal to the !-site are required but without

sequence or size specificity (Caras et al., 1989; De Groot et al.,

2005; Lu et al., 1994; Moran & Caras, 1991).

The smallest known GPI anchored protein is the CD52 or

Campath-1 antigen. In humans, the full-length CD52 gene

encodes a 61 amino acid protein that begins with an

N-terminal signal peptide, which is 24 amino acids in

length. The C-terminal GPI-T signal sequence begins with

the !-site at Ser36 and proceeds to the C-terminus (Ser61).

Thus, the proprotein is 37 amino acids long and the fully

mature, GPI-anchored protein contains only 12 amino acids

(Xia et al., 1993).

Despite the simplicity of the rules that define the GPI-T

signal sequence, some studies have suggested that GPI-T

shows species specificity for its protein substrates (Moran &

Caras, 1994; Morissette et al., 2012; Takos et al., 2000).

General and species-specific prediction algorithms have been

developed and have revolutionized the ability of researchers

to predict not only GPI anchoring but also the identity of the

one or two most likely !-sites (Eisenhaber et al., 1999;

Fankhauser & Maser, 2005; Pierleoni et al., 2008; Poisson

et al., 2007). One recent GPI-T signal peptide prediction tool

demonstrates high accuracy using a position-specific scoring

matrix (PSSM) (Mukai et al., 2010, 2013). One thing that

these in silico analyses suggest is the possibility that GPI-T

recognizes and processes more than one !-site in a single

peptide, leading to subtle heterogeneity during maturation. (In

other words, a protein substrate with two putative ! sites

might be processed at both positions so that a mixture is

produced where the anchor can be attached at either !-site.)

To our knowledge, the experimental identification of pro-

cessing at more than one !-site has not yet been observed.

However, very few efforts at genome-wide characterizations

of anchored proteins, particularly with !-site validation, have

been reported, so this possibility cannot be rejected at the

present time.

The GPI transamidase complex

Five GPI-T subunits have been identified so far, with

homologs in eukaryotes ranging from yeast to humans; all

five subunits are essential for the attachment of GPI anchors

to proteins. As mentioned above, these subunits are called

PIG-K, PIG-T, GPAA1, PIG-S and PIG-U in humans

(Benghezal et al., 1996; Hong et al., 2003; Inoue et al.,

1999; Ohishi et al., 2001) and are analogous to Gpi8, Gpi16,

Gaa1, Gpi17 and Gab1 in yeast, respectively (Benghezal

et al., 1996; Fraering et al., 2001; Hamburger et al., 1995;

Hong et al., 2003; Ohishi et al., 2001). In Trypanosoma

brucei, PIG-U and PIG-T are replaced by TTA1 and TTA2,

two unrelated subunits (Nagamune et al., 2003). Table 2

summarizes the sizes of these different subunits as well as

their predicted number of transmembrane domains and

glycosylation sites for orthologs from humans, yeast and

T. brucei. For the remainder of this article, we will use the

names of the human GPI-T subunits unless specifically

talking about an experiment conducted with GPI-T from other

species.

Homologs of PIG-K, GPAA1 and PIG-T are conserved

across eukaryotes. In yeast, these core subunits can be purified

together as a complex (Fraering et al., 2001). In contrast, in

humans, all five subunits can be isolated together (Hong et al.,

2003). Based on mutagenic analyses and its similarity to

caspases, the PIG-K subunit was identified as the catalytic

active site and is the best characterized of the five known

subunits (Meyer et al., 2000; Meitzler et al., 2007; Toh et al.,

2011). Possible roles for some of the remaining subunits have

been proposed and are discussed individually below. The

hydrophobicity of the subunits, the complexity of the GPI-T

enzyme, and poor expression levels of the different subunits

have contributed to the lack of progress in further character-

ization of this enzyme. Another drawback has been the lack of

a high-throughput assay for GPI-T. Nearly all methods to assay

this enzyme’s activity are both cumbersome and qualitative.

Significant data is accumulating that supports the hypoth-

esis that the GPI-T complex contains more than one copy of

some or all of its subunits. In particular, native PAGE analysis

of the pure, heterotrimeric GPI-T complex from yeast

revealed that this complex resolves into two assemblies with

molecular weights of �430 and �650 kD (Fraering et al.,

2001). Given the molecular weights of the individual

subunits, a complex of only �185 kD is predicted in yeast.

All three of these GPI-T subunits (Gpi8, Gaa1 and Gpi16)

contain probable glycosylation sites (Table 2), but it seems

unlikely that glycans could account for an increase in MW of

the 4450 kD needed to explain the 650 kD complex. Thus,

Conzelmann and colleagues proposed higher order oligomer-

ization for GPI-T (Fraering et al., 2001). The human GPI-T

complex (from HeLA cells) has a velocity sedimentation

value of 17.7S, also consistent with a globular complex with a

mass of �460 kD (Vainauskas et al., 2002). In this work,

GPAA1 was also observed to interact with a- and b-tubulin;

thus, the possibility that tubulin is the source of increase in the

molecular weight of GPI-T in humans cannot be ruled out.

(Tubulin was not apparent in the yeast GPI-T complex

analyzed by native gel.) Additionally, PIG-K, the active site

subunit of GPI-T, has sequence and putative structural

DOI: 10.3109/10409238.2013.831024 GPI Transamidase and GPI anchored proteins 5

Cri

tical

Rev

iew

s in

Bio

chem

istr

y an

d M

olec

ular

Bio

logy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f Z

ueri

ch Z

entr

um f

uer

Zah

n M

und

und

on 0

8/31

/13

For

pers

onal

use

onl

y.

similarity to caspases. The soluble domain of Gpi8

(the PIG-K ortholog from yeast) partly assembles into a

homodimer when heterologously expressed in Escherichia

coli, analogous to caspase dimerization (Meitzler et al.,

2007). Gpi16 is attached to Gpi8 by a known disulfide bond

(Ohishi et al., 2003) (see below); thus, by analogy, the

hypothesis that the Gpi8 homodimer is symmetrically

modified by two Gpi16 subunits is intuitive. Gpi8 dimeriza-

tion has recently been called into question (discussed further

in the next section) (Toh et al., 2011). Understanding

the stoichiometry and organization of GPI-T is going to be

crucial to understanding its function. Additional research is

needed in this area.

The next sections summarize what is known about the

structures and functions of the individual subunits of GPI-

T. The possible functional roles for each subunit, as they

are currently understood, are discussed individually here

and are summarized for human GPI-T in Table 3. Their

possible roles in cancer will be discussed later in this

review.

The PIG-K (Gpi8) subunit

PIG-K is the catalytic subunit of GPI-T. This �47 kD subunit

nominally belongs to the C13 cysteine protease family

(Benghezal et al., 1996; Meyer et al., 2000). PIG-K has a

large soluble domain, oriented to the luminal side of the ER,

and a single C-terminal transmembrane region (Figure 4A)

(Benghezal et al., 1996). The soluble domain has sequence

similarity to caspases, a family of cysteine proteases that

regulate cell death (Meyer et al., 2000). Analysis of conserved

His and Cys residues indicated that His164 and Cys206 are

the catalytic residues in human PIG-K (His157 and Cys199 in

yeast) (Meyer et al., 2000; Ohishi et al., 2000). By analogy to

cysteine proteases, the histidine presumably deprotonates the

cysteine, which nucleophilically attacks the amide bond

between the ! and !þ 1 residues, creating a thioester

intermediate, which is subsequently converted to a new

amide with the GPI anchor (Figure 5) (Zacks & Garg, 2006).

A Rosetta-predicted structure of the soluble domain of

yeast Gpi8 was built based on putative structural homology

Table 2. Features of GPI-T from humans, S. cerevisiae and T. brucei. Specific references are provided for publications where a given TM domain orglycosylation site was predicted or experimentally examined. Asterisks (*) indicate glycosylation sites or TM regions that were only predicted in theUniProt database (The UniProt Consortium, 2012).

Subunit Size (kD) Putative glycosylation sites Transmembrane regions References

Human GPI-TPIG-K 45.3 – One* The UniProt Consortium (2012)GPAA1 67.6 Two: N203, N517 Eight Hiroi et al. (1998)PIG-S 61.7 Two: N267*, N370* Two* The UniProt Consortium (2012)PIG-T 65.7 Three: N164, N291*, N327* One* Chen et al. (2009)PIG-U 50.1 – Nine* The UniProt Consortium (2012)

S. cerevisiaeGpi8 47.4 Three: N23, y N256*, N346* One* Benghezal et al. (1996)Gaa1 69.2 Two: N87, N383*z Six Hamburger et al. (1995)Gpi17 60.8 Five: N100*, N170*, N228*, N247*, N299* Two* The UniProt Consortium (2012)Gpi16 68.8 Two: N28, y N184* One Fraering et al. (2001)Gab1 44.7 – Eight * The UniProt Consortium (2012)

T. bruceiTbGpi8 36.7 One: N25 No Kang et al. (2002)TbGaa1 51.2 – Six Nagamune et al. (2003)TbGpi16 75.8 – One Nagamune et al. (2003)TTA1 41.9 Two: N79, N259 Two Nagamune et al. (2003)TTA2 45.6 – Six Nagamune et al. (2003)

yIn yeast, both Gpi8 and Gpi16 contain reasonable N-linked glycosylation sites within or immediately adjacent to their N-terminal signal peptides.These sites are listed here for completeness but they have not been characterized; they may not be glycosylated or may have been cleaved from theprotein during N-terminal processing.zN383 is predicted as a glycosylation site using UniProt. However, Hamburger et al. reported results that argue that this site is not glycosylated.

N383 lies between the second and third transmembrane domains of Gaa1 and is presumably inaccessible to the glycosylation machinery.

Table 3. Proposed functional roles for the five subunits of human GPI-T.

Subunit Possible roles or functions References

PIG-K Similarities to caspases and other cysteine proteasesContains all or some of the enzyme’s catalytic machineryAttached to PIG-T by a disulfide bond

Benghezal et al. (1996);Meyer et al. (2000);Ohishi et al. (2003)

GPAA1 May contain part of the active site and be involved in peptide bindingand/or recognition

Hiroi et al., (1998)

PIG-S Essential for thioester intermediate formation between PIG-K and the protein substrate Ohishi et al. (2001);Zhu et al. (2005)

PIG-T Essential for thioester intermediate formation between PIG-K and the protein substrateAttached to PIG-K by a disulfide bond

Ohishi et al. (2001);Ohishi et al. (2003)

PIG-U Loosely associated with the rest of the GPI-T complexWeak similarity to fatty acid elongasesPossibly involved in lipid recognition or binding

Vainauskas & Menon (2004);Hong et al. (2003)

6 D. G. Gamage & T. L. Hendrickson Crit Rev Biochem Mol Biol, Early Online: 1–19

Cri

tical

Rev

iew

s in

Bio

chem

istr

y an

d M

olec

ular

Bio

logy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f Z

ueri

ch Z

entr

um f

uer

Zah

n M

und

und

on 0

8/31

/13

For

pers

onal

use

onl

y.

between caspases and Gpi8 (Figure 4B) (Meitzler et al.,

2007). This model positions the backbones of the two

catalytic residues of Gpi8 (His157 and Cys199) in similar

locations and orientations as their counterparts in caspases.

Caspases are active as homodimers, leading to the hypothesis

that Gpi8 also assembles into a homodimer, an

oligomerization step that may be essential for enzyme

activity. Dimerization of the soluble domain of Gpi8 was

observed by native PAGE and by mass spectrometry.

Furthermore, dimerization was disrupted by the introduction

of mutations at positions corresponding to the face of caspase

dimerization. Significant Gpi8 monomer was also observed in

Figure 4. The PIG-K subunit. (A) PIG-K has a single soluble domain (�340 amino acids) and one transmembrane domain. Human PIG-K is notglycosylated however there is one site of N-glycosylation in yeast Gpi8. The catalytic cysteine (Cys206 in humans) is noted. PIG-K is connected toPIG-T via a single disulfide bond (not shown). (B) The soluble domain of PIG-K has putative sequence and structural homology with caspases(Meitzler et al., 2007; Meyer et al., 2000). The structure of caspase-1 from Spodoptera frugiperda (PDB: 1M72, gray (green in the figure availableonline)) is overlayed onto a Rosetta model of S. cerevisiae Gpi8 (black (magenta in the figure available online)) (Meitzler et al., 2007). (C) Thesequences for portions of the active site of human PIG-K (NP_005473), S. cerevisiae Gpi8 (NP_010618), human caspase-14 (NP_036246), and S.frugiperda caspase-1 (AAC47442) were aligned using Clustal W. Conserved residues are colored in dark gray (blue in the figure available online),including the histidine and cysteine that form the catalytic dyad for each enzyme. Residues highlighted in light gray (magenta in the figure availableonline) indicate positions that show similarity in at least three of the four sequences. (D) A closeup of the catalytic dyads in S. frugiperda caspase-1(gray (green in the figure available online)) and the model of S. cerevisiae Gpi8 (black (magenta in the figure available online)) from panel B. Theactive site cysteine in caspase-1 is shown alkylated by an irreversible inhibitor. The model of Gpi8 places the His/Cys catalytic dyad within hydrogenbonding distance. (see colour version of this figure at www.informahealthcare.com/bmg).

Figure 5. Proposed mechanism for GPI-T mediated protein transamidation. (see colour version of this figure at www.informahealthcare.com/bmg).

DOI: 10.3109/10409238.2013.831024 GPI Transamidase and GPI anchored proteins 7

Cri

tical

Rev

iew

s in

Bio

chem

istr

y an

d M

olec

ular

Bio

logy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f Z

ueri

ch Z

entr

um f

uer

Zah

n M

und

und

on 0

8/31

/13

For

pers

onal

use

onl

y.

this work, leading to the proposal that Gpi8 dimer reflects the

native oligomerization state and the monomer represents

poorly folded or misfolded Gpi8 as a consequence of

heterologous expression in E. coli. Toh et al. recently

reported a very similar isolation of the soluble domain of

Gpi8 (Toh et al., 2011). However, they used size exclusion

chromatography (SEC) to isolate only the monomeric form of

Gpi8. It is unclear whether or not they examined their

preparations for dimer and it is probable that the Gpi8 dimer

was lost during SEC purification. As discussed above, clear

evidence from two other research groups also support the

hypothesis that GPI-T assembles into a higher order oligomer

in yeast (Fraering et al., 2001) and in humans (Vainauskas

et al., 2002). Consequently, the preponderance of available

evidence argues that GPI-T assembles into a higher order

oligomer. However, additional characterization of this

enzyme’s stoichiometry is clearly mandated.

The GPAA1 (Gaa1) subunit

GPAA1 was the first subunit identified in the GPI-T complex

with a size of 67 kD (Hamburger et al., 1995). It has a single

N-terminal TM domain, a soluble domain, and five

C-terminal TM domains (Figure 6) (Hamburger et al.,

1995). GPAA1 shares 25% sequence identity and 57%

similarity with yeast Gaa1 (Hiroi et al., 1998). It assembles

into a stable complex with Gpi16 and Gpi8 in yeast (Fraering

et al., 2001). In human cells, GPAA1 associates with PIG-K,

PIG-T, PIG-S and PIG-U and is essential for transamidase

activity (Ohishi et al., 2000, 2001; Vainauskas et al., 2002).

A portion of the soluble domain of yeast Gaa1 was

characterized recently using small angle X-ray scattering

(SAXS) providing a low resolution map of a fragment

(residues 70–247) of this domain (Saw et al., 2013).

While its exact function is unknown, evidence suggests

that GPAA1 recognizes and stabilizes the C-terminal signal

sequence of the peptide substrate through a conserved Pro609

in the last transmembrane helix (Chen et al., 2003;

Vainauskas et al., 2002; Vainauskas & Menon, 2004).

Additionally, photo cross-linking studies also support the

hypothesis that GPAA1 interacts with protein substrates for

GPI-T. GPI-T from GPAA1 knockout mouse cells were still

capable of generating the thioester intermediate between Gpi8

and a substrate protein, but this intermediate was not

processed to the mature, GPI-anchored protein (Ohishi

et al., 2000). Combined, these observations are consistent

with GPAA1 containing part of the active site (in addition to

Gpi8) and/or a substrate recognition domain.

The PIG-T (Gpi16) subunit

PIG-T is a 69 kD protein with a large N-terminal hydrophilic

region and a C-terminal transmembrane domain (Figure 7)

(Ohishi et al., 2001). A mutation in PIG-T is connected to a

recessive intellectual disability syndrome, which, to our

knowledge, is the only known GPI-T defect associated with a

disease other than cancer (Kvarnung et al., 2013). In yeast,

Gpi16 is co-purified along with GST-Gpi8 and Gaa1; whereas,

in human all five subunits co-purify as a complex with GST-

Flag-PIG-K (Fraering et al., 2001; Hong et al., 2003). Even

though the exact function of this subunit is not clear, PIG-T is

essential for formation of the carbonyl intermediate between

Gpi8 and the protein substrate during transamidation (Figure 5)

(Ohishi et al., 2001). Some evidence suggests that PIG-T

stabilizes PIG-K and GPAA1: When PIG-T was knocked out,

reduced expression levels of the other GPI-T subunits were

observed (Ohishi et al., 2001). PIG-T makes a functionally

relevant disulfide bond with PIG-K through two conserved

cysteine residues, Cys92 (in PIG-K) and Cys182 (in PIG-T),

making it the only subunit covalently linked to the catalytic

subunit (Ohishi et al., 2003). This linkage is not essential for

the formation of the GPI-T complex, but is important for GPI-T

activity (Ohishi et al., 2003).

The PIG-S (Gpi17) subunit

PIG-S is a 61 kD protein with a large soluble domain in

between two transmembrane regions (Figure 8) (Ohishi et al.,

Figure 7. PIG-T has one N-terminal soluble domain (�506 amino acids)and a single C-terminal transmembrane domain. PIG-T is connected toPIG-K via a disulfide bond (not shown). PIG-T has three N-linkedglycosylation sites in its soluble domain. (see colour version of thisfigure at www.informahealthcare.com/bmg).

Figure 6. GPAA1 has one N-terminal transmembrane domain, a singlesoluble domain (�323 amino acids) and seven C-terminal transmem-brane domains (The UniProt Consortium, 2012). Two N-linkedglycosylation sites are found in GPAA1 at Asn203 and Asn517.(see colour version of this figure at www.informahealthcare.com/bmg).

8 D. G. Gamage & T. L. Hendrickson Crit Rev Biochem Mol Biol, Early Online: 1–19

Cri

tical

Rev

iew

s in

Bio

chem

istr

y an

d M

olec

ular

Bio

logy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f Z

ueri

ch Z

entr

um f

uer

Zah

n M

und

und

on 0

8/31

/13

For

pers

onal

use

onl

y.

2001). In yeast, Gpi17 is essential for GPI-T activity, but it

does not stably interact with the core GPI-T subunits (Gpi8,

Gpi16 and Gaa1) and its exact function is unknown

(Zhu et al., 2005). As observed with PIG-T, knockout of the

PIG-S gene eliminated formation of the thioester intermediate

between PIG-K and the proprotein substrate (Ohishi et al.,

2001). PIG-S is one of the subunits replaced by TTA1 and

TTA2 in T. brucei (Figure 8).

The PIG-U (Gab 1) subunit

PIG-U was the fifth (and presumably final) subunit identified

as a component of GPI-T (Figure 8) (Hong et al., 2003). This

38 kD protein is highly hydrophobic and has between eight

and 10 transmembrane regions. Deletion of this gene inhibits

the formation of cell surface GPI-anchored proteins (Hong

et al., 2003). Vainauskas and Menon suggested that PIG-U is

more loosely associated with the GPI-T complex than any of

the other subunits, based on differential immunoprecipitation

patterns with digitonin versus nonidet-solubilized microsomes

(Vainauskas & Menon, 2004). Although its contribution to the

GPI-T complex is unknown, PIG-U does show weak sequence

similarity with fatty acid elongases suggesting that it may be

involved in recognition of the lipid portion of the GPI anchor

(Hong et al., 2003). PIG-U was the first GPI-T subunit found

in human cancer (Guo et al., 2004). PIG-U is also replaced by

TTA1 and TTA2 in the T. brucei GPI-T (Figure 8).

The TTA1 & TTA2 subunits

Two of the five human GPI-T subunits (PIG-S and PIG-U) are

not conserved across all eukaryotes. In trypanosomes, these

two subunits are replaced by Trypanosomatid Transamidase

1 (TTA1) and Trypanosomatid Transamidase 2 (TTA2)

(Figure 8) (Nagamune et al., 2003). TTA1 has two trans-

membrane helices, one at each terminus. The intervening

hydrophilic soluble region is predicted to face the luminal

side of the ER and contain two N-glycosylation sites. On the

other hand, TTA2 contains multiple transmembrane domains

and two small, soluble domains TTA1 and TTA2 do not share

sequence homology with any mammalian, yeast, plant, insect

or nematode GPI-T subunits; however, orthologs are present

in Leishmania major. TTA1 and TTA2 are linked to each

other through a disulfide linkage and knockout of either of

these subunits inhibits the transfer of the GPI anchor onto its

protein substrates.

The relevance of TTA1 and TTA2 to a review of GPI-T and

cancer is not immediately obvious. However, these two

trypanosomal subunits have replaced PIG-S and PIG-U, the

same two subunits that do not co-purify as part of a robust

complex from Saccharomyces cerevisiae (Fraering et al.,

2001). Even though we do not yet understand the impact of

these observations, they do suggest that the roles of PIG-S and

PIG-U in human GPI-T may be peripheral compared to those

of PIG-K, PIG-T and GPAA1. In other words, they hint that

Figure 8. Subunits PIG-S and PIG-U are found in human GPI-T (A) and are replaced by TTA1 and TTA2 in T. brucei (B). PIG-S and TTA1 are notrelated in sequence but have topological similarities and both contain two putative glycosylation sites. PIG-U and TTA2 both have two small solubledomains and several transmembrane domains however they are topological dissimilar and do not share sequence similarity. (see colour version of thisfigure at www.informahealthcare.com/bmg).

DOI: 10.3109/10409238.2013.831024 GPI Transamidase and GPI anchored proteins 9

Cri

tical

Rev

iew

s in

Bio

chem

istr

y an

d M

olec

ular

Bio

logy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f Z

ueri

ch Z

entr

um f

uer

Zah

n M

und

und

on 0

8/31

/13

For

pers

onal

use

onl

y.

PIG-K, PIG-T and GPAA1 may constitute the catalytic core of

GPI-T for all species. While this hypothesis remains specu-

lative, it is likely that these observations will ultimately

contribute to our understanding of human GPI-T function and

the roles of these subunits in cancer.

GPI-T and cancer

The amplification of oncogenes contributes to human car-

cinogenesis (Bishop, 1991). Chromosomes 8q and 20q are

frequently amplified in many cancers including breast,

bladder, ovarian and endometrioid carcinomas (Guan et al.,

1996; Kallioniemi et al., 1994, 1995; Pere et al., 1998;

Wilting et al., 2006). Out of the five GPI-T subunits, the genes

encoding PIG-U, PIG-T and GPAA1 are localized in the

20q11, 20q13 and 8q24 chromosome regions, respectively,

positions that are considered hotspots for most cancers

(Nagpal et al., 2008). The genes encoding PIG-K and PIG-

S are located at 1p31 and 17p13, respectively (Nagpal et al.,

2008). Simple localization of a gene within an oncogenic

amplicon is insufficient to identify an oncogene. Amplicons

contain multiple genes, not all of which have increased copy

numbers in the corresponding tumors nor are over-expressed

to a significant degree. However, known oncogenic amplicons

make useful starting points to identify new oncogenes and to

better understand tumorigenesis.

The first hint for the importance of GPI-T in cancer was

reported by Trink and colleagues in 2004, with their discovery

of the first oncogenic GPI-T subunit, PIG-U, in bladder cancer

(Guo et al., 2004). With this finding, the possibility of over-

expression of other GPI-T subunits in different cancer types

came into the picture. Another critical study showed that

breast cancer cells have significantly elevated levels of cell

surface GPI-anchored proteins that are more typical of

mesenchymal stem cells than of healthy breast tissue (Zhao

et al., 2012). This finding is consistent with over-expression

of one or more GPI-T subunits leading to up-regulation of

GPI-T catalytic activity as a mechanism for tumor initiation or

invasion. This section will discuss our current understanding

of the over-expression of each GPI-T subunit in different

cancer types, at both the mRNA and protein levels, and their

importance as oncogenes or biomarkers. It is clear that a

number of different downstream events can be activated or

regulated by over-expression of different GPI-T subunits.

PIG-U and cancer

PIG-U was the fifth subunit identified in the GPI-T complex, a

hydrophobic protein that is essential for GPI-T activity (Hong

et al., 2003). Building on the discovery of germline transloca-

tion of the 20q11 chromosomal region in uroepithelial cancer

(Schoenberg et al., 1996), the CDC91L1 (PIG-U) gene was

characterized for its role in bladder cancer development

(Guo et al., 2004). This gene lies adjacent to the germline

translocation site. Over-expression of PIG-U in mice induced

tumorigenesis, providing strong evidence that this subunit

acts as an oncogene (Guo et al., 2004). Furthermore, forced

over-expression of PIG-U in cell culture induced an increase

in cell growth rate and enhanced over-expression of pro-

teins known to be GPI anchored. Of particular interest was

the observed over-expression of urokinase plasminogen

activated receptor (uPAR) (Guo et al., 2004). This GPI-

anchored protein is a well-characterized oncogene for most

cancers (Andres et al., 2012; Mekkawy et al., 2009). Increased

STAT-3 phosphorylation was also observed as a downstream

effect of uPAR over-expression (Guo et al., 2004), suggesting

that tumorigenicity arises from perturbations in JAK/stat cell

signaling (Figure 9). In total, this report suggests that

over-expression of PIG-U increases GPI-T activity and

anchoring of substrate proteins although the mechanism by

which activity is increased remains unknown, particularly

since PIG-U is not the catalytic subunit of GPI-T.

A subsequent study concluded that CDC91L1 is rarely

overexpressed in urothelial cell carcinomas (where 2.4% over-

expression of CDC91L1 mRNA was observed (Schultz et al.,

2005) compared to 430% CDC91L1 amplification in cell

lines and primary bladder tumors (Guo et al., 2004)). Finally,

a third group assessed a larger data set of bladder urothelial

cell carcinoma. In this study, CDC91L1 mRNA was

over-expressed in 30.1% of tumors compared to healthy

cells. PIG-U protein over-expression occurred in 75.3% of

tumor samples (Shen et al., 2008). These differences in over-

expression levels of both mRNA and protein presumably arise

from different factors such as tumor stage, age and gender of

the patient, or other environmental factors that remain poorly

understood.

Expression patterns for all five GPI-T subunits were

analyzed in 19 different cancer types and compared to healthy

tissues from the same organ and the same patient using

microarray technology (Nagpal et al., 2008). Basal level

expression of different subunits varied in different types of

healthy tissue (Nagpal et al., 2008). PIG-U mRNA was over-

expressed in 60% of colon and ovarian cancer samples versus

healthy tissue (Nagpal et al., 2008). In lymph node tumors,

PIG-U protein was expressed at moderate to low levels in 90%

of malignant tissues, but was not detectable in the corres-

ponding healthy tissues. Also a significant increase of PIG-U

protein production was observed in both ovarian and breast

cancer cells and over-expression occurred in 60% of large cell

lung carcinoma cells (Nagpal et al., 2008). PIG-U was over-

expressed in 42% of breast cancer cells (Wu et al., 2006), as

well as in prostate cancer (Nagpal et al., 2008).

PIG-T and cancer

The PIGT gene is also positioned in a chromosomal hot spot

(chromosomal region 20q13.12). With the discovery of PIGU

as an oncogene, the possibility of other GPI-T subunits as

oncogenes, including PIG-T, became relevant (Guo et al.,

2004; Nagpal et al., 2008). Over-expression of PIG-T was first

found in human breast cancer. An increase in PIG-T

expression correlated with downstream over-expression and

phosphorylation of paxillin (Wu et al., 2006), a known cell

invasion-related and tumorigenic protein (Conway et al.,

2006; Huang et al., 2004).

In the same microarray report discussed for PIG-U, PIG-T

mRNA over-expression was observed in 60% of uterine, 50%

of thyroid and melanoma, and 30% of breast cancer samples

compared to healthy tissues (Nagpal et al., 2008). Significant

PIG-T protein over-expression was observed in colon, thyroid,

lung and pancreatic cancers; over-expression at lower levels

10 D. G. Gamage & T. L. Hendrickson Crit Rev Biochem Mol Biol, Early Online: 1–19

Cri

tical

Rev

iew

s in

Bio

chem

istr

y an

d M

olec

ular

Bio

logy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f Z

ueri

ch Z

entr

um f

uer

Zah

n M

und

und

on 0

8/31

/13

For

pers

onal

use

onl

y.

also occurred in both squamous cell carcinoma and lung

adenocarcinoma cells (Nagpal et al., 2008).

A combination of mass spectrometry, separate siRNA

inhibition of PIG-T and GPAA1 expression, and separate

over-expression of each of these subunits led to the identi-

fication of 19 GPI-anchored proteins that are specifically

expressed in breast cancer cells and are either poorly

expressed or not expressed in healthy breast tissue (Zhao

et al., 2012). Eighteen of these biomarkers are present in

mesenchymal stem cells, suggesting that all or some of these

proteins facilitate dedifferentiation of breast cancer cells.

Furthermore, reduction of either PIG-T or GPAA1 levels by

siRNA reduced expression levels of the FOXC2 transcription

factor by 80%. Over-expression (by viral infection) of either

subunit increased expression of FOXC2 at both the mRNA

and protein levels (Zhao et al., 2012). The authors posited that

over-expression of either PIG-T or GPAA1 affects signal

transduction pathways (presumably by increased expression

Figure 9. Proposed mechanisms for how GPI-T may participate in cancer: (A) Overexpression of uPAR, to upregulate the JAK/STAT phosphorylationpathway; (B) Activation of paxillin phosphorylation; (C) Upregulation of FOXC2 expression and downstream signaling. The solid lines representpathways that have direct experimental support. The dashed lines represent pathways that likely involve intermediate steps that are currentlyuncharacterized.

DOI: 10.3109/10409238.2013.831024 GPI Transamidase and GPI anchored proteins 11

Cri

tical

Rev

iew

s in

Bio

chem

istr

y an

d M

olec

ular

Bio

logy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f Z

ueri

ch Z

entr

um f

uer

Zah

n M

und

und

on 0

8/31

/13

For

pers

onal

use

onl

y.

of a GPI-anchored cell surface receptor) that leads to

increased FOXC2 expression (Figure 9). FOXC2 is over-

expressed in breast and colon cancers and is involved in

mitochondrial biogenesis and increased cell metabolism

(Lidell et al., 2011; Zhao et al., 2012).

Cigarette smoke extract (CSE) induces cancer formation in

head, neck, bladder and breast cells (Yiping et al., 2009). CSE

also induces over-expression of three GPI-T subunits: PIG-T,

PIG-U and GPAA1.

GPAA1 and cancer

Frequent amplification in chromosomal region 8q24 in

different cancer types makes this region a chromosomal

hotspot (Balsara et al., 1997; Sonoda et al., 1997; Tsuchiya

et al., 2002; Yokota et al., 1999). The GPAA1 gene is in the

8q24.3 region and thus, like PIG-U, is a possible oncogene

(Hiroi et al., 1998; Inoue et al., 1999).

GPAA1 mRNA levels were increased 69% in head

and neck squamous carcinoma and 40% in uterine cancer

cells (Jiang et al., 2007; Nagpal et al., 2008). Significant

over-expression of GPAA1 protein was observed in ovary

and thyroid cells, along with �40% over-expression in

prostate cancer and 10%–20% in lung adenocarcinoma cases

(Nagpal et al., 2008). PCR-array profiling of 20 pairs of

liver tissues (healthy versus tumor samples) identified 117

genes with different expression levels, only seven of which

were amplified in hepatocellular carcinoma and both

hepatitis B virus positive carcinoma and hepatitis C virus

positive carcinoma (Kurokawa et al., 2004). GPAA1 was

one of these seven genes. Amplification was observed at

both the mRNA (75%) and protein levels (90%) (Ho et al.,

2006).

When GPAA1 was over-expressed in breast cancer cells,

levels of phosphorylated paxillin also increased, thereby

activating Brk-mediated phosphorylation and promoting cell

invasion that is linked to tumor metastasis (Figure 9) (Chen

et al., 2004; Conway et al., 2006; Wu et al., 2006). Along

with PIG-U and PIG-T, GPAA1 was over-expressed in the

presence of CSE, which led to an initiation of paxillin

phosphorylation in head, neck, bladder and breast cancers

(Yiping et al., 2009). GPAA1 over-expression led to its

association with the epidermal growth factor receptor (EGFR)

and EGFR phosphorylation in the presence of epidermal

growth factor (Yiping et al., 2009). As a consequence, PIG-T

and PIG-U were phosphorylated by EGFR. It was proposed

that this GPAA1–EGFR interaction and phosphorylation leads

to phosphorylation of paxillin to induce cancer initiation

(Yiping et al., 2009). GPAA1 over-expression was observed in

a variety of different cancers that did not correlate with over-

expression of other GPI-T subunits. The connection between

GPAA1 and EGFR may explain this divergence (Yiping et al.,

2009). Elevated levels of GPAA1 also increased FOXC2

protein levels (Lidell et al., 2011; Zhao et al., 2012).

PIG-K and cancer

Human PIG-K is the catalytic subunit of the GPI-T complex

(Benghezal et al., 1996; Meyer et al., 2000). Compared to

other subunits, PIG-K resides on a different chromosome

(1p31.1), in a region that is frequently lost in various human

cancers (Rooney et al., 1999). In breast cancer, PIG-K was

over-expressed in both ovarian (64%) and uterine (67%)

cancers (Nagpal et al., 2008). However, PIG-K was down-

regulated 50% in both bladder and hepatocellular carcinoma

cells and 40% in colon carcinoma cells, based on mRNA

levels; similar down-regulation was observed at the protein

level (40%, 100% and 40%, respectively) (Nagpal et al.,

2008). In order to understand the reason for diminished PIG-

K expression, all 10 PIG-K exons were examined in samples

Figure 9. Continued.

12 D. G. Gamage & T. L. Hendrickson Crit Rev Biochem Mol Biol, Early Online: 1–19

Cri

tical

Rev

iew

s in

Bio

chem

istr

y an

d M

olec

ular

Bio

logy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f Z

ueri

ch Z

entr

um f

uer

Zah

n M

und

und

on 0

8/31

/13

For

pers

onal

use

onl

y.

from 45 different colorectal cancer patients. A single

nucleotide polymorphism at position rs1048575 (outside the

coding region), which changed C/C to either G/C or G/G, was

identified that varied with race (Dasgupta et al., 2012).

In contrast, PIG-K was undetectable in normal lymph node

tissues but accumulated in 65% of lymph node cancer samples

(Nagpal et al., 2008).

Given the close connections between GPI-T and cancer, in

general, it is perhaps surprising that PIG-K, the catalytic

subunit of GPI-T, is more likely to be down-regulated than up-

regulated in many cancers. In yeast, depletion of Gpi8 causes

changes in actin morphology (depletion of Gab1, but none of

the other GPI-T subunit, showed similar effects) (Grimme

et al., 2004). These changes offer one possible scenario for

how reduced Gpi8 expression might lead to downstream

tumorigenesis without increasing GPI-T activity and anchor-

ing of oncogenic GPI-anchored proteins.

PIG-S and cancer

The PIG-S gene resides in chromosomal region 17p13.2, a

region lost in certain cancers (Rooney et al., 1999). However,

PIG-S is over-expressed in breast cancer tissues compared to

healthy tissues (Nagpal et al., 2008). A significant

over-expression of PIG-S was seen in thyroid cancer samples

and mRNA levels of PIG-S were amplified 60% in lung,

40% in ovarian and liver, and 50% in thyroid cancers (Nagpal

et al., 2008).

How does GPI-T subunit over-expression lead tocancer?

Over-expression of each GPI-T subunit has been observed in

one or more cancer types with different frequencies and

different patterns. For example, in breast cancer samples,

PIG-T, PIG-U and GPAA1 are commonly over-expressed

compared to healthy tissue. In ovarian tumor samples, PIG-T,

PIG-K and GPAA1 were over-expressed. In some colon

cancer samples, PIG-T is over-expressed, but PIG-K expres-

sion is suppressed (Nagpal et al., 2008). Table 4 highlights

some examples of the distinct patterns of subunit expression

observed in different tumor types and samples.

The five GPI-T genes have the hallmarks of oncogenes,

tumor biomarkers and potential targets for the development of

new chemotherapeutics. However, a great deal remains to be

understood before GPI-T subunits can be used to detect or treat

cancer. For example, how does over-expression of one subunit

induce tumorigenesis? The work described above has led to

proposals for different mechanisms (Figure 9). First, and most

logically, over-expression of a GPI-T subunit can lead to

increased GPI-T activity and increased presentation of GPI-

anchored proteins on the surface of cancer cells. The obser-

vation that PIG-U over-expression increased uPAR cell surface

presentation and Jak/STAT cell signaling supports this hypoth-

esis (Guo et al., 2004; Mekkawy et al., 2009). So does the fact

that GPAA1 over-expression in breast cancer correlated with

increased cell surface presentation of 18 GPI anchored proteins

involved in cell dedifferentiation (Zhao et al., 2012). Second,

over-expression of GPI-T subunits can cause perturbations in

cell signaling and transcription that facilitate tumor growth.

Evidence is accumulating supporting roles for GPI-T in

modulating paxillin phosphorylation and over-expression of

the FOXC2 transcription factor (Yiping et al., 2009; Zhao et al.,

2012). Neither paxillin nor FOXC2 is GPI anchored so the

subunit over-expression mechanism that leads to these per-

turbations is not clear. The apparent recruitment of EGFR to the

GPI-T complex upon GPAA1 over-expression is perhaps the

most intriguing observation (Yiping et al., 2009). This

association led to phosphorylation of EGFR, PIG-T, and PIG-

U by EGFR tyrosine kinase. It is not known how these

phosphorylation events impact GPI-T activity; however,

expression of these different subunits and recruitment of

EGFR correlated with paxillin phosphorylation (Yiping et al.,

2009).

It is clear that changes in GPI-T subunit overexpression

impact the concentrations of different cell surface GPI-

anchored proteins and modulate signal transduction. The

specific perturbations in GPI-T that lead to these conse-

quences remain poorly understood. For example, is EGFR

involved in the dedifferentiation of breast cancer after GPAA1

over-expression? Or do the GPAA1/EGFR interactions induce

tumorigenesis via a mechanism that is different from GPAA1-

induced dedifferentiation? And, at a more basic level, how

Table 4. Examples of reported changes in GPI-T subunit expression (compared to healthy cells of the same tissue).

Cancer type PIG-U PIG-T PIG-K GPAA1 PIG-S References

Bladder cancer """ ## Nagpal et al. (2008); Shen et al. (2008)Breast cancer "" "" " " Nagpal et al. (2008); Wu et al. (2006)Colon cancer " ## Nagpal et al. (2008)Head and neck squamous carcinoma """* Jiang et al. (2007)Hepatocellular carcinoma ### ""* Nagpal et al. (2008)Hepatitis positive hepatocellular carcinoma """ Ho et al. (2006)Lymph node cancer " """ Nagpal et al. (2008)Lung carcinoma """ " " """* Nagpal et al. (2008)Melanoma ""* Nagpal et al. (2008)Ovarian cancer " """ " ""* Nagpal et al. (2008)Pancreas cancer " Nagpal et al. (2008)Prostate cancer " "" Nagpal et al. (2008)Squamous cell carcinoma " Nagpal et al. (2008)Thyroid cancer ""* " ""* Nagpal et al. (2008)Uterine cancer """* """ ""* Nagpal et al. (2008)

Symbols are as follows: """ indicates450% overexpression; "" indicates 20%–50% overexpression; " indicates 0%–20% overexpression; ### indicates450% down-regulation; ## indicates 20%–50% down-regulation; * indicates data obtained from mRNA levels (all other data reflect characterizationof protein expression levels).

DOI: 10.3109/10409238.2013.831024 GPI Transamidase and GPI anchored proteins 13

Cri

tical

Rev

iew

s in

Bio

chem

istr

y an

d M

olec

ular

Bio

logy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f Z

ueri

ch Z

entr

um f

uer

Zah

n M

und

und

on 0

8/31

/13

For

pers

onal

use

onl

y.

does over-expression of each subunit impact GPI-T activity? It

is easy to hypothesize that GPI-T activity is up-regulated in all

cases, but this hypothesis is contraindicated by the fact that

PIG-K, the catalytic subunit, is actually down-regulated in

many tumors.

With respect to tumorigenesis and GPI-T, it is clear that

we are still looking at only the tip of the iceberg. Further

cell biology studies are needed, in addition to a careful

assessment of the impact of subunit over-expression on GPI-T

activity.

Functions of GPI-anchored proteins relevant totumorigenesis

GPI-anchored proteins participate in diverse functions

including immune responses, embryogenesis, fertilization,

cell wall biosynthesis, signal transduction and others

(Hazenbos et al., 2004; Kawagoe et al., 1996; Pittet &

Conzelmann, 2007; Ueda et al., 2007).The medical relevance

of GPI-anchored proteins is clear because specific GPI

anchored proteins are crucial for tumor growth and invasion

as well as other diseases like Creutzfeldt-Jakob disease,

bovine spongiform encephalopathy, and African sleeping

sickness (Ferguson, 1999; Guo et al., 2004; Nakatsura et al.,

2003; Prusiner; 1991).

The following section discusses the physiological func-

tions of a few GPI-anchored proteins in cancer along with

their potential relevance as oncogenes, biomarkers and

therapeutic targets. This section does not represent a

comprehensive list of GPI-anchored proteins in cancer.

Instead, it is meant to highlight the different ways that GPI-

anchored proteins are known to participate in tumorigenesis

and cancer progression. The importance of these proteins in

cancer also implicates GPI-T, or its individual subunits, as

targets for the development of new chemotherapies. The

challenge with targeting GPI-T for cancer treatments, how-

ever, is that any suitable drug would likely have to access the

ER to be effective. Thus, GPI-T may prove to be a difficult

drug target, but changes in the expression of GPI-anchored

proteins in cancers, like those described below, offer an

additional set of potential targets.

Urokinase plasminogen activated receptor

Urokinase plasminogen activated receptor (uPAR) belongs to

the urokinase plasminogen activating system (uPAS), which

also includes the urokinase plasminogen activator (uPA), and

two serine proteinase inhibitors, plasminogen activator

inhibitor-1 (PAI-1) and -2 (PAI-2). uPAR is a �60 kD

glycoprotein that is GPI anchored (Aceto et al., 1998). It has

three domains, D1, D2 and D3, linked together by conserved

disulfide bonds (Mekkawy et al., 2009). In healthy tissues,

uPAR is expressed at moderate levels; strong expression is

seen in tissues undergoing extensive remodeling (Solberg

et al., 2001). uPAR regulates extracellular proteolysis by

binding to uPA and activating plasminogen-generating plasma

(Ellis et al., 1992). Because of the above properties, uPAR is

over-expressed in almost all cancer types; upregulation

of uPAR causes downstream changes in a number of

different cell signaling pathways (some of which are

described below) (Almasi et al., 2011; Andres et al., 2012;

Asuthkar et al., 2012; Bujanda et al., 2013; Park et al., 2011;

Waldron et al., 2012).

Different expression levels of components of the uPA

system act as biomarkers for different cancers and these

receptors and enzymes can serve as therapeutic targets

(Ossowski & Reich, 1983; Rea et al., 2013). The most

effective way to use this system is by inhibiting uPA using

small molecule inhibitors or by interfering with the uPA/

uPAR interaction. Small molecules such as 3-amidinopheny-

lalanine negatively affect the uPA system and thereby limit

the invasiveness of head and neck carcinoma cells, and

cervical and breast cancer cell lines. Soluble uPAR inhibits

cell proliferation in ovarian cancer (Ertongur et al., 2004;

Wilhelm et al., 1994). Catalytically inactive uPA fragments

and peptide constructs that can be used as antagonists or

toxins were also useful in treating uPAR-activated cancer

cells (Jankun, 1992; Ploug et al., 2001).

Glypican-3

Glypicans are GPI-anchored heparin sulfate proteoglycans

that regulate the activity of heparin binding growth factor

(Filmus & Selleck, 2001). So far, six glypicans have been

identified in mammals, all of similar size (60–70 kD) (Filmus

& Selleck, 2001). Mutations that takes place in Glypican-3

can cause loss-of-function, leading to Simpson–Golabi–

Behmel syndrome, a rare X-chromosome-linked overgrowth

defect (Pilia et al., 1996). The expression levels of glypicans

differ in growth stage and tissue-specific manners, however,

expression predominates during development and in develop-

mental morphogenesis (Li et al., 1997; Litwack et al., 1998).

The ability of glypicans to regulate growth and survival

indicates their relevance in tumor progression. The first

relationship between cancer and glypicans was seen in human

pancreatic cancer, where glypican-1 was over-expressed

(Kleeff et al., 1998). Glypican-3 is over-expressed in

hepatocellular carcinoma and in the clear cell carcinoma of

ovaries (Nakatsura et al., 2003; Umezu et al., 2010).

However, down-regulation of glypican-3 was observed in

breast, lung and ovarian cancer cells (De Rienzo et al., 2000;

Kim et al., 2003; Yun-Yan et al., 2001). The high expression

levels of Glypican-3 in both hepatocellular and ovarian clear

cell carcinoma have led to the evaluation of this protein as a

therapeutic target using cell- and antibody-based immu-

notherapies with some promising results (Ho & Kim, 2011;

Ishiguro et al., 2008; Nakatsura et al., 2004). Its differential

over-expression in different cancers suggests that glypicans

can be used as biomarkers using immunohistochemistry and

they may be suitable as therapeutic targets.

Folate binding receptor

Two folate binding receptors (FR) are GPI-anchored glyco-

peptides that have high affinity to folic acid (Kd� 1 pM)

(Vastag, 2000). Four different isoforms of FRs are known (a,

b, g and �), however, only the a and b isoforms are GPI

anchored (Shen et al., 1995; Yan & Ratnam, 1995). FR-a is

the most widely studied isomer, which has limited expression

14 D. G. Gamage & T. L. Hendrickson Crit Rev Biochem Mol Biol, Early Online: 1–19

Cri

tical

Rev

iew

s in

Bio

chem

istr

y an

d M

olec

ular

Bio

logy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f Z

ueri

ch Z

entr

um f

uer

Zah

n M

und

und

on 0

8/31

/13

For

pers

onal

use

onl

y.

levels in normal tissues but is over-expressed in a variety of

cancer cell types including ovarian, lung, breast and others

(Cagle et al., 2012; Leung et al., 2013; O’Shannessy et al.,

2012). Due to its high affinity for folic acid, the FR/folate

interaction has been used in radiopharmacology, chemother-

apy, and magnetic resonance imaging (Goren et al., 2000;

Konda et al., 2001; Leamon et al., 2002).

During the last two decades, folate-based radio-conjugates

have been developed to use in PET and SPECT imaging

and tested in clinical trials in patients with folate receptor

positive solid tumors (Fisher et al., 2008; Siegel et al., 2003).

Several radioisotopes have been used for PET imaging,

including fluorine-18, gallium-68, terbium-152 and scan-

dium-44 (Fani et al., 2012; Fischer et al., 2012; Koumarianou

et al., 2012; Muller et al., 2012). The EC90 vaccine has been

used in folate immune therapy and is in phase I studies for

patients with renal cell cancer (Amato et al., 2013). Several

folate receptor targets have been synthesized to use in

chemotherapeutics. For example, folate conjugate EC145 is

in phase I clinical studies for patients with refractory tumors

(Ak & Sanl|er, 2012; LoRusso et al., 2012). Over-expression

of folate receptor a in lung cancer (72% in adenocarcinomas

and 51% in squamous cell carcinomas) indicates the import-

ance of the folate receptor as a therapeutic target and the need

for more investigation of this GPI-anchored protein (Cagle

et al., 2012; Wood, 2012).

Prostasin

Prostasin is a serine protease highly expressed in the kidneys,

prostate and lungs (Chen et al., 2001). It is a GPI-anchored

protein that acts as channel activating protease-1 (CAP-1) and

activates epithelial sodium channels, which maintain the salt

and fluid balance in the kidneys (Rossier, 2004; Schild &

Kellenberger, 2001). Prostasin is down-regulated in gastric

and prostate cancer cells but it is over-expressed in pancreatic,

breast and oral cancer cells (Sakashita et al., 2008; Takahashi

et al., 2003) Recently prostasin has been identified as a

potential tumor marker for early stages of ovarian cancer

(Costa et al., 2009). Even though the role of prostasin in

cancer cells is not well understood, the use of prostasin

inhibitors such as protease nexin–1 (PN-1) have been

investigated (Selzer-Plon et al., 2009).

Conclusion

GPI-anchored proteins play vital roles in different cancers and

correlate to changes in GPI-T expression. Even though the

importance of GPI-anchored proteins in cancer is well

established, the importance of GPI-T came into the picture

only in 2004 with the discovery of PIG-U as an oncogene in

bladder cancer (Guo et al., 2004). Since this discovery, several

interesting findings have shown that the expression levels of

different GPI-T subunits are highly variable in different cancer

types and between patients. One key question that needs to be

addressed is how the underexpression of PIG-K, the catalytic

subunit, affects GPI-T function in a way that promotes tumors.

A detailed knowledge of GPI-T structure and function is

needed in order to understand the role of this enzyme in cancer.

Also GPI-anchored proteins have a variety of functions apart

from their oncological roles upon over-expression. Different

GPI-anchored proteins are over-expressed in different cancer

cells, making them potential biomarkers and therapeutic

targets. In many ways, this field is still in its infancy, with

many interesting questions yet to be examined.

Acknowledgements

The authors thank Sandamali Ekanayaka and Travis J. Ness

for their critical comments on the manuscript.

Declaration of interest

The authors report no declarations of interest.

References

Aceto J, Kieber-Emmons T, Cines DB. (1998). Carboxy-terminalprocessing of the urokinase receptor: implications for substraterecognition and glycosylphosphatidylinositol anchor addition.Biochemistry 38:992–1001.

Ak G, Sanl|er SH. (2012). Synthesis of folate receptor-targeted anddoxorubicin coupled chemotherapeutic nanoconjugate and researchinto its medical application. Prep Biochem Biotech 42:551–63.

Almasi CE, Brasso K, Iversen P, et al. (2011). Prognostic and predictivevalue of intact and cleaved forms of the urokinase plasminogenactivator receptor in metastatic prostate cancer. Prostate 71:899–907.

Almeida A, Layton M, Karadimitris A. (2009). Inherited glycosylpho-sphatidyl inositol deficiency: a treatable CDG. Biochim Biophys Acta1792:874–80.

Amato RJ, Shetty A, Lu Y, et al. (2013). A phase I study of folateimmune therapy (EC90 vaccine administered with GPI-0100 adjuvantfollowed by EC17) in patients with renal cell carcinoma.J Immunother 36:268–75.

American Cancer Society. (2011). Global cancer facts & figures. 2nd ed.Atlanta: American Cancer Society. Available from: http://www.cancer.org/acs/groups/content/@epidemiologysurveilance/documents/document/acspc-027766.pdf [last accessed 31 May 2013].

Andres SA, Edwards AB, Wittliff JL. (2012). Expression of urokinase-type plasminogen activator (uPA), its receptor (uPAR), and inhibitor(PAI-1) in human breast carcinomas and their clinical relevance.J Clin Lab Anal 26:93–103.

Ashida H, Hong Y, Murakami Y, et al. (2005). Mammalian PIG-X andyeast Pbn1p are the essential components of glycosylphosphatidyli-nositol-mannosyltransferase I. Mol Biol of Cell 16:1439–48.

Ast T, Cohen G, Schuldiner M. (2013). A network of cytosolic factorstargets SRP-independent proteins to the endoplasmic reticulum.Cell 152:1134–45.

Asuthkar S, Gondi CS, Nalla AK, et al. (2012). Urokinase-typeplasminogen activator receptor (uPAR)-mediated regulation ofWNT/b-catenin signaling is enhanced in irradiated medulloblastomacells. J Biol Chem 287:20576–89.

Balsara BR, Sonoda G, Du Manoir S, et al. (1997). Comparativegenomic hybridization analysis detects frequent, often high-level,overrepresentation of DNA sequences at 3q, 5p, 7p, and 8q in humannon-small cell lung carcinomas. Cancer Res 57:2116–20.

Benghezal M, Benachour A, Rusconi S, et al. (1996). Yeast Gpi8p isessential for GPI anchor attachment onto proteins. EMBO J 15:6575–83.

Bishop JM. (1991). Molecular themes in oncogenesis. Cell 64:235–48.Brodsky RA, Hu R. (2006). PIG-A mutations in paroxysmal nocturnal

hemoglobinuria and in normal hematopoiesis. Leuk Lymphoma 47:1215–21.

Bujanda L, Sarasqueta C, Cosme A, et al. (2013). Evaluation of alpha 1-antitrypsin and the levels of mRNA expression of matrix metallopro-teinase 7, urokinase type plasminogen activator receptor and COX-2for the diagnosis of colorectal cancer. PLoS ONE 8:e51810.

Burgula S, Swarts BM, Guo Z. (2012). Total synthesis of a glycosylpho-sphatidylinositol anchor of the human lymphocyte CD52 antigen.Chem – Eur J 18:1194–201.

Cagle PT, Zhai QJ, Murphy L, Low PS. (2012). Folate receptor inadenocarcinoma and squamous cell carcinoma of the lung: potential

DOI: 10.3109/10409238.2013.831024 GPI Transamidase and GPI anchored proteins 15

Cri

tical

Rev

iew

s in

Bio

chem

istr

y an

d M

olec

ular

Bio

logy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f Z

ueri

ch Z

entr

um f

uer

Zah

n M

und

und

on 0

8/31

/13

For

pers

onal

use

onl

y.

target for folate-linked therapeutic agents. Arch Pathol Lab Med 137:241–4.

Caras IW, Weddell GN, Williams SR. (1989). Analysis of the signal forattachment of a glycophospholipid membrane anchor. J Cell Biol 108:1387–96.

Chen H-Y, Shen C-H, Tsai Y-T, et al. (2004). Brk activates rac1 andpromotes cell migration and invasion by phosphorylating paxillin.Mol Cell Biol 24:10558–72.

Chen L-M, Skinner ML, Kauffman SW, et al. (2001). Prostasin is aglycosylphosphatidylinositol-anchored active serine protease. J BiolChem 276:21434–42.

Chen R, Anderson V, Hiroi Y, Medof ME. (2003). Proprotein interactionwith the GPI transamidase. J Cell Biochem 88:1025–37.

Chen R, Jiang X, Sun D, Han G, et al. (2009). Glycoproteomics analysisof human liver tissue by combination of multiple enzyme digestionand hydrazide chemistry. J Proteome Res 8:651–61.

Conway WC, Van Der Voort Van Zyp J, Thamilselvan V, et al. (2006).Paxillin modulates squamous cancer cell adhesion and is important inpressure-augmented adhesion. J Cell Biochem 98:1507–16.

Conzelmann A, Puoti A, Lester RL, Desponds C. (1992). Twodifferent types of lipid moieties are present in glycophosphoinositol-anchored membrane proteins of Saccharomyces cerevisiae. EMBO J11:457–66.

Costa FP, Batista Junior EL, Zelmanowicz A, et al. (2009). Prostasin, apotential tumor marker in ovarian cancer: a pilot study. Clinics 64:641–4.

Coyne KE, Crisci A, Lublin DM. (1993). Construction of syntheticsignals for glycosyl-phosphatidylinositol anchor attachment. Analysisof amino acid sequence requirements for anchoring. J Biol Chem 268:6689–93.

Dasgupta S, Pal P, Mukhopadhyay ND, et al. (2012). A single nucleotidepolymorphism in the human PIGK gene associates with low PIGKexpression in colorectal cancer patients. Int J Oncol 41:1405–10.

De Groot PWJ, Ram AF, Klis FM. (2005). Features and functions ofcovalently linked proteins in fungal cell walls. Fungal Genet Biol 42:657–75.

De Rienzo A, Ferriola PC, Filmus J, et al. (2000). Expression of GCP3,an X-linked recessive overgrowth gene, is silenced in malignantmesothelioma. Oncogene 19:410–16.

Eisenhaber B, Bork P, Eisenhaber F. (1998). Sequence properties ofGPI-anchored proteins near the omega-site: constraints for thepolypeptide binding site of the putative transamidase. Protein Eng11:1155–61.

Eisenhaber B, Bork P, Eisenhaber F. (1999). Prediction of potential GPI-modification sites in proprotein sequences. J Mol Biol 292:741–58.

Eisenhaber B, Bork P, Eisenhaber F. (2001). Post-translational GPIlipid anchor modification of proteins in kingdoms of life: analysis ofprotein sequence data from complete genomes. Protein Eng 14:17–25.

Eisenhaber B, Maurer-Stroh S, Novatchkova M, et al. (2003). Enzymesand auxiliary factors for GPI lipid anchor biosynthesis and post-translational transfer to proteins. Bioessays 25:367–85.

Ellis V, Pyke C, Eriksen J, et al. (1992). The urokinase receptor:involvement in cell surface proteolysis and cancer invasion. Ann NYAcad Sci 667:13–31.

Englund PT. (1993). The structure and biosynthesis of glycosylphosphatidylinositol protein anchors. Annu Rev Biochem 62:121–38.

Ertongur S, Lang S, Mack B, et al. (2004). Inhibition of the invasioncapacity of carcinoma cells by WX-UK1, a novel synthetic inhibitor ofthe urokinase-type plasminogen activator system. Int J Cancer 110:815–24.

Fani M, Tamma M-L, Nicolas GP, et al. (2012). In vivo imaging of folatereceptor positive tumor xenografts using novel 68Ga-NODAGA-folateconjugates. Mol Pharm 9:1136–45.

Fankhauser N, Maser P. (2005). Identification of GPI anchor attachmentsignals by a kohonen self-organizing map. Bioinformatics 21:1846–52.

Ferguson MA. (1999). The structure, biosynthesis and functions ofglycosylphosphatidylinositol anchors, and the contributions of tryp-anosome research. J Cell Sci 112:2799–809.

Ferguson M, Homans S, Dwek R, Rademacher T. (1988). Glycosyl-phosphatidylinositol moiety that anchors Trypanosoma brucei variantsurface glycoprotein to the membrane. Science 239:753–9.

Filmus J, Selleck SB. (2001). Glypicans: proteoglycans with a surprise.J Clin Invest 108:497–501.

Fischer CR, Muller C, Reber J, et al. (2012). [18F]fluoro-deoxy-glucosefolate: a novel PET radiotracer with improved in vivo properties forfolate receptor targeting. Bioconjugate Chem 23:805–13.

Fisher RE, Siegel BA, Edell SL, et al. (2008). Exploratory study of99mTc-EC20 imaging for identifying patients with folate receptor–positive solid tumors. J Nucl Med 49:899–906.

Fraering P, Imhof I, Meyer U, et al. (2001). The GPI transamidasecomplex of Saccharomyces cerevisiae contains Gaa1p, Gpi8p, andGpi16p. Mol Biol Cell 12:3295–306.

Fujita M, Jigami Y. (2008). Lipid remodeling of GPI-anchored proteinsand its function. Biochim Biophys Acta – Gen Subj 1780:410–20.

Fujita M, Kinoshita T. (2010). Structural remodeling of GPI anchorsduring biosynthesis and after attachment to proteins. FEBS Lett 584:1670–7.

Fujita M, Kinoshita T. (2012). GPI-anchor remodeling: potentialfunctions of GPI anchors in intracellular trafficking and membranedynamics. Biochimic Biophys Acta – Mol Cell Biol Lipids 1821:1050–8.

Galian C, Bjorkholm P, Bulleid N, Von Heijne G. (2012). Efficientglycosylphosphatidylinositol (GPI) modification of membrane pro-teins requires a c-terminal anchoring signal of marginal hydrophobi-city. J Biol Chem 287:16399–409.

Gerber LD, Kodukula K, Udenfriend S. (1992). Phosphatidylinositolglycan (PI-G) anchored membrane proteins. Amino acid requirementsadjacent to the site of cleavage and pi-g attachment in the COOH-terminal signal peptide. J Biol Chem 267:12168–73.

Goren D, Horowitz AT, Tzemach D, et al. (2000). Nuclear delivery ofdoxorubicin via folate-targeted liposomes with bypass of multidrug-resistance efflux pump. Clin Cancer Res 6:1949–57.

Grimme SJ, Gao X-D, Martin PS, et al. (2004). Deficiencies in theendoplasmic reticulum (ER)-membrane protein Gab1p perturb trans-fer of glycosylphosphatidylinositol to proteins and cause perinuclearER-associated actin bar formation. Mol Biol Cell 15:2758–70.

Guan X-Y, Xu J, Anzick SL, et al. (1996). Hybrid selection of transcribedsequences from microdissected DNA: isolation of genes within anamplified region at 20q11–q13.2 in breast cancer. Cancer Res 56:3446–50.

Guo Z, Linn JF, Wu G, et al. (2004). Cdc91l1 (PIG-U) is a newlydiscovered oncogene in human bladder cancer. Nat Med 10:374–81.

Hamburger D, Egerton M, Riezman H. (1995). Yeast Gaa1p is requiredfor attachment of a completed GPI anchor onto proteins. J Cell Biol129:629–39.

Hazenbos WLW, Clausen BE, Takeda J, Kinoshita T. (2004). GPI-anchordeficiency in myeloid cells causes impaired FcgR effector functions.Blood 104:2825–31.

Hiroi Y, Komuro I, Chen R, et al. (1998). Molecular cloning of humanhomolog of yeast Gaa1 which is required for attachment ofglycosylphosphatidylinositols to proteins. FEBS Lett 421:252–8.

Ho JC, Cheung ST, Patil M, et al. (2006). Increased expression ofglycosyl-phosphatidylinositol anchor attachment protein 1 (Gpaa1) isassociated with gene amplification in hepatocellular carcinoma. Int JCancer 119:1330–7.

Ho M, Kim H. (2011). Glypican-3: a new target for cancer immuno-therapy. Eur J Cancer 47:333–8.

Hong Y, Maeda Y, Watanabe R, et al. (1999). Pig-n, a mammalianhomologue of yeast Mcd4p, is involved in transferring phosphoetha-nolamine to the first mannose of the glycosylphosphatidylinositol.J Biol Chem 274:35099–106.

Hong Y, Maeda Y, Watanabe R, et al. (2000). Requirement of PIG-F andPIG-O for transferring phosphoethanolamine to the third mannose inglycosylphosphatidylinositol. J Biol Chem 275:20911–19.

Hong Y, Ohishi K, Kang JY, et al. (2003). Human PIG-U and yeastcdc91p are the fifth subunit of GPI transamidase that attaches GPI-anchors to proteins. Mol Biol of Cell 14:1780–9.

Huang C, Jacobson K, Schaller MD. (2004). A role for JNK-paxillinsignaling in cell migration. Cell Cycle 3:3–5.

Inoue N, Ohishi K, Endo Y, et al. (1999). Human and mouse Gpaa1(glycosylphosphatidylinositol anchor attachment 1) genes: genomicstructures, chromosome loci and the presence of a minor class intron.Cytogenet Genome Res 84:199–205.

Ishiguro T, Sugimoto M, Kinoshita Y, et al. (2008). Anti–glypican 3antibody as a potential antitumor agent for human liver cancer. CancerRes 68:9832–8.

Jankun J. (1992). Antitumor activity of the type 1 plasminogen activatorinhibitor and cytotoxic conjugate in vitro. Cancer Res 52:5829–32.

16 D. G. Gamage & T. L. Hendrickson Crit Rev Biochem Mol Biol, Early Online: 1–19

Cri

tical

Rev

iew

s in

Bio

chem

istr

y an

d M

olec

ular

Bio

logy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f Z

ueri

ch Z

entr

um f

uer

Zah

n M

und

und

on 0

8/31

/13

For

pers

onal

use

onl

y.

Jiang W-W, Zahurak M, Zhou Z-T, et al. (2007). Alterations of GPItransamidase subunits in head and neck squamous carcinoma. MolCancer 6:74.

John F, Hendrickson TL. (2010). Synthesis of truncated analogues forstudying the process of glycosyl phosphatidylinositol modification.Org Lett 12:2080–3.

Kallioniemi A, Kallioniemi O-P, Citro G, et al. (1995). Identification ofgains and losses of DNA sequences in primary bladder cancer bycomparative genomic hybridization. Genes Chromosomes Cancer 12:213–19.

Kallioniemi A, Kallioniemi OP, Piper J, et al. (1994). Detection andmapping of amplified DNA sequences in breast cancer bycomparative genomic hybridization. Proc Natl Acad Sci USA 91:2156–60.

Kamena F, Tamborrini M, Liu X, et al. (2008). Synthetic GPI array tostudy antitoxic malaria response. Nat Chem Biol 4:238–40.

Kang JY, Hong Y, Ashida H, et al. (2005). PIG-V involved in transferringthe second mannose in glycosylphosphatidylinositol. J Biol Chem280:9489–97.

Kang X, Szallies A, Rawer M, et al. (2002). GPI anchor transamidase ofTrypanosoma brucei: in vitro assay of the recombinant protein andVSG anchor exchange. J Cell Sci 115:2529–39.

Kawagoe K, Kitamura D, Okabe M, et al. (1996).Glycosylphosphatidylinositol-anchor-deficient mice: implications forclonal dominance of mutant cells in paroxysmal nocturnal hemo-globinuria. Blood 87:3600–6.

Kim H, Xu G-L, Borczuk AC, et al. (2003). The heparan sulfateproteoglycan GPC3 is a potential lung tumor suppressor. Am J RespirCell Mol Biol 29:694–701.

Kleeff J, Ishiwata T, Kumbasar A, et al. (1998). The cell-surface heparansulfate proteoglycan glypican-1 regulates growth factor action inpancreatic carcinoma cells and is overexpressed in human pancreaticcancer. J Clin Invest 102:1662–73.

Kodukula K, Micanovic R, Gerber L, et al. (1991). Biosynthesis ofphosphatidylinositol glycan-anchored membrane proteins. Design of asimple protein substrate to characterize the enzyme that cleaves theCOOH-terminal signal peptide. J Biol Chem 266:4464–70.

Konda SD, Aref M, Wang S, et al. (2001). Specific targeting of folate–dendrimer MRI contrast agents to the high affinity folate receptorexpressed in ovarian tumor xenografts. Magn Reson Mater Phys, BiolMed 12:104–13.

Koumarianou E, Loktionova NS, Fellner M, et al. (2012). 44Sc-DOTA-BN[2-14]NH2 in comparison to 68Ga-DOTA-BN[2-14]NH2in pre-clinical investigation. Is 44Sc a potential radionuclide for PET?Appl Radiat Isot 70:2669–76.

Krawitz Peter m, Murakami Y, Rieb A, et al. (2013). PGAP2 mutations,affecting the GPI-anchor-synthesis pathway, cause hyperphosphatasiawith mental retardation syndrome. Am J Hum Genet 92:584–9.

Kurokawa Y, Matoba R, Nakamori S, et al. (2004). PCR-array geneexpression profiling of hepatocellular carcinoma. J Exp Clin CancerRes 23:135–41.

Kvarnung M, Nilsson D, Lindstrand A, et al. (2013). A novel intellectualdisability syndrome caused by GPI anchor deficiency due tohomozygous mutations in PIGT. J Med Genet 50:521–8.

Lacey SW, Sanders JM, Rothberg KG, et al. (1989). ComplementaryDNA for the folate binding protein correctly predicts anchoring tothe membrane by glycosyl-phosphatidylinositol. J Clin Invest 84:715–20.

Leamon CP, Parker MA, Vlahov IR, et al. (2002). Synthesis andbiological evaluation of EC20: a new folate-derived, 99mTc-basedradiopharmaceutical. Bioconjugate Chem 13:1200–10.

Leidich SD, Drapp DA, Orlean P. (1994). A conditionally lethal yeastmutant blocked at the first step in glycosyl phosphatidylinositolanchor synthesis. J Biol Chem 269:10193–6.

Leung F, Dimitromanolakis A, Kobayashi H, et al. (2013). Folate-receptor 1 (FOLR1) protein is elevated in the serum of ovarian cancerpatients. Clin Biochem [Epub ahead of print]. doi:10.1016/j.clinbiochem.2013.03.010.

Li M, Choo B, Wong Z-M, et al. (1997). Expression of OCI-5/glypican 3during intestinal morphogenesis: regulation by cell shape in intestinalepithelial cells. Exp Cell Res 235:3–12.

Lidell ME, Seifert EL, Westergren R, et al. (2011). The adipocyte-expressed forkhead transcription factor Foxc2 regulates metabolismthrough altered mitochondrial function. Diabetes 60:427–35.

Litwack ED, Ivins JK, Kumbasar A, et al. (1998). Expression of theheparan sulfate proteoglycan glypican-1 in the developing rodent. DevDyn 211:72–87.

Lorusso PM, Edelman MJ, Bever SL, et al. (2012). Phase I study offolate conjugate EC145 (vintafolide) in patients with refractory solidtumors. J Clin Oncol 30:4011–16.

Lu CF, Kurjan J, Lipke PN. (1994). A pathway for cell wall anchorage ofSaccharomyces cerevisiae alpha-agglutinin. Mol Cell Biol 14:4825–33.

Maeda Y, Watanabe R, Harris CL, et al. (2001). PIG-M transfers the firstmannose to glycosylphosphatidylinositol on the lumenal side of theER. EMBO J 20:250–61.

Masterson WJ, Doering TL, Hart GW, Englund PT. (1989). A novelpathway for glycan assembly: biosynthesis of the glycosyl-phospha-tidylinositol anchor of the trypanosome variant surface glycoprotein.Cell 56:793–800.

McConville MJ, Ferguson MA. (1993). The structure, biosynthesis andfunction of glycosylated phosphatidylinositols in the parasitic proto-zoa and higher eukaryotes. Biochem J 294:305–24.

Meitzler JL, Gray JJ, Hendrickson TL. (2007). Truncation of the caspase-related subunit (Gpi8p) of Saccharomyces cerevisiae GPI transami-dase: dimerization revealed. Arch Biochem Biophys 462:83–93.

Mekkawy AH, Morris DL, Pourgholami MH. (2009). Urokinaseplasminogen activator system as a potential target for cancer therapy.Future Oncol 5:1487–99.

Meyer U, Benghezal M, Imhof I, Conzelmann A. (2000). Active sitedetermination of Gpi8p, a caspase-related enzyme required forglycosylphosphatidylinositol anchor addition to proteins.Biochemistry 39:3461–71.

Micanovic R, Kodukula K, Gerber LD, Udenfriend S. (1990). Selectivityat the cleavage/attachment site of phosphatidylinositol-glycananchored membrane proteins is enzymatically determined. Proc NatlAcad Sci 87:7939–43.

Moran P, Caras IW. (1991). A nonfunctional sequence converted to asignal for glycophosphatidylinositol membrane anchor attachment.J Cell Biol 115:329–36.

Moran P, Caras IW. (1994). Requirements for glycosylphosphatidylino-sitol attachment are similar but not identical in mammalian cells andparasitic protozoa. J Cell Biol 125:333–43.

Moran P, Raab H, Kohr WJ, Caras IW. (1991). Glycophospholipidmembrane anchor attachment. Molecular analysis of the cleavage/attachment site. J Biol Chem 266:1250–7.

Morissette R, Varma Y, Hendrickson TL. (2012). Defining theboundaries of species specificity for the Saccharomyces cerevisiaeglycosylphosphatidylinositol transamidase using a quantitative in vivoassay. Biosci Rep 32:577–86.

Mukai Y, Ikeda M, Tanaka H, et al. (2013). Discrimination ofmammalian GPI-anchored proteins by hydropathy and amino acidpropensities. Biosci Biotechnol Biochem 77:526–33.

Mukai Y, Sasaki T, Oura O, Ikeda M. (2010). Identification ofmammalian GPI-anchored proteins based on amino acid propensitiesin their GPI attachment signal sequences. Trends Inform Sci Comp106–10. doi:10.1109/TISC.2010.5714617.

Muller C, Zhernosekov K, Koster U, et al. (2012). A unique matchedquadruplet of terbium radioisotopes for PET and SPECT and for a-and b�-radionuclide therapy: an in vivo proof-of-concept study with anew receptor-targeted folate derivative. J Nucl Med 53:1951–9.

Murakami Y, Siripanyapinyo U, Hong Y, et al. (2003). PIG-W is criticalfor inositol acylation but not for flipping of glycosylphosphatidylino-sitol-anchor. Mol Biol Cell 14:4285–95.

Murakami Y, Siripanyaphinyo U, Hong Y, et al. (2005). The initialenzyme for glycosylphosphatidylinositol biosynthesis requires PIG-Y,a seventh component. Mol Biol Cell 16:5236–46.

Nagamune K, Ohishi K, Ashida H, et al. (2003). GPI transamidase ofTrypanosoma brucei has two previously uncharacterized (trypanoso-matid transamidase 1 and 2) and three common subunits. Proc NatlAcad Sci 100:10682–7.

Nagpal JK, Dasgupta S, Jadallah S, et al. (2008). Profiling the expressionpattern of GPI transamidase complex subunits in human cancer. ModPathol 21:979–91.

Nakatsura T, Komori H, Kubo T, et al. (2004). Mouse homologue of anovel human oncofetal antigen, glypican-3, evokes T-cell–mediatedtumor rejection without autoimmune reactions in mice. Clin CancerRes 10:8630–40.

DOI: 10.3109/10409238.2013.831024 GPI Transamidase and GPI anchored proteins 17

Cri

tical

Rev

iew

s in

Bio

chem

istr

y an

d M

olec

ular

Bio

logy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f Z

ueri

ch Z

entr

um f

uer

Zah

n M

und

und

on 0

8/31

/13

For

pers

onal

use

onl

y.

Nakatsura T, Yoshitake Y, Senju S, et al. (2003). Glypican-3,overexpressed specifically in human hepatocellular carcinoma, is anovel tumor marker. Biochem Biophys Res Commun 306:16–25.

Nozaki M, Ohishi K, Yamada N, et al. (1999). Developmentalabnormalities of glycosylphosphatidylinositol-anchor-deficientembryos revealed by Cre/IoxP system. Lab Invest 79:293–9.

O’Shannessy D, Somers E, Maltzman J, et al. (2012). Folate receptoralpha (FRA) expression in breast cancer: identification of a newmolecular subtype and association with triple negative disease.SpringerPlus 1:22.

Ohishi K, Inoue N, Kinoshita T. (2001). PIG-S and PIG-T, essential forGPI anchor attachment to proteins, form a complex with Gaa1 andGpi8. EMBO J 20:4088–98.

Ohishi K, Inoue N, Maeda Y, et al. (2000). Gaa1p and Gpi8pare components of a glycosylphosphatidylinositol (GPI) transamidasethat mediates attachment of GPI to proteins. Mol Biol Cell 11:1523–33.

Ohishi K, Nagamune K, Maeda Y, Kinoshita T. (2003). Two subunits ofglycosylphosphatidylinositol transamidase, GPI8 and PIG-T, form afunctionally important intermolecular disulfide bridge. J Biol Chem278:13959–67.

Ossowski L, Reich E. (1983). Antibodies to plasminogen activatorinhibit human tumor metastasis. Cell 35:611–19.

Park JS, Park JH, Khoi PN, et al. (2011). MSP-induced RON activationupregulates uPAR expression and cell invasiveness via MAPK, AP-1and NF-�B signals in gastric cancer cells. Carcinogenesis 32:175–81.

Paulick MG, Bertozzi CR. (2008). The glycosylphosphatidylinositolanchor: a complex membrane-anchoring structure for proteins.Biochemistry 47:6991–7000.

Paulick MG, Wise AR, Forstner MB, et al. (2007). Synthetic analoguesof glycosylphosphatidylinositol-anchored proteins and their behaviorin supported lipid bilayers. J Am Chem Soc 129:11543–50.

Pere H, Tapper J, Wahlstrom T, et al. (1998). Distinct chromosomalimbalances in uterine serous and endometrioid carcinomas. CancerRes 58:892–5.

Pierleoni A, Martelli P, Casadio R. (2008). PredGPI: a GPI-anchorpredictor. BMC Bioinformatics 9:392.

Pilia G, Hughes-Benzie RM, Mackenzie A, et al. (1996). Mutations ingpc3, a glypican gene, cause the simpson-golabi-behmel overgrowthsyndrome. Nat Genet 12:241–7.

Pittet M, Conzelmann A. (2007). Biosynthesis and function of gpiproteins in the yeast Saccharomyces cerevisiae. Biochim BiophysActa – Mol Cell Biol Lipids 1771:405–20.

Ploug M, Østergaard S, Gardsvoll H, et al. (2001). Peptide-derivedantagonists of the urokinase receptor. Affinity maturation bycombinatorial chemistry, identification of functional epitopes, andinhibitory effect on cancer cell intravasation. Biochemistry 40:12157–68.

Ploug M, Rønne E, Behrendt N, et al. (1991). Cellular receptor forurokinase plasminogen activator. Carboxyl-terminal processing andmembrane anchoring by glycosyl-phosphatidylinositol. J Biol Chem266:1926–33.

Poisson G, Chauve C, Chen X, Bergeron A. (2007). Fraganchor: a large-scale predictor of glycosylphosphatidylinositol anchors in eukaryoteprotein sequences by qualitative scoring. Genomics ProteomicsBioinformatics 5:121–30.

Prusiner S. (1991). Molecular biology of prion diseases. Science 252:1515–22.

Ramalingam S, Maxwell SE, Medof ME, et al. (1996). COOH-terminalprocessing of nascent polypeptides by the glycosylphosphatidylinosi-tol transamidase in the presence of hydrazine is governed by the sameparameters as glycosylphosphatidylinositol addition. Proc Natl AcadSci 93:7528–33.

Rea VEA, Lavecchia A, Di Giovanni C, et al. (2013). Discovery of newsmall molecules targeting the vitronectin binding site of the urokinasereceptor that block cancer cell invasion. Mol Cancer Ther 12:1402–16.

Rooney PH, Murray GI, Stevenson DAJ, et al. (1999). Comparativegenomic hybridization and chromosomal instability in solid tumours.Br J Cancer 80:862–73.

Rossier BC. (2004). The epithelial sodium channel. Proc Am Thorac Soc1:4–9.

Sakashita K, Mimori K, Tanaka F, et al. (2008). Clinical significance oflow expression of prostasin mRNA in human gastric cancer. J SurgOncol 98:559–64.

Saw WG, Eisenhaber B, Eisenhaber F, Gruber G. (2013). Low resolutionstructure of the soluble domain GPAA1 (yGPAA170–247) of theglycosylphosphatidylinositol transamidase subunit GPAA1 fromSaccharomyces cerevisiae. Biosci Rep 33:361–9.

Schild L, Kellenberger S. (2001). Structure function relationships ofENaC and its role in sodium handling. In: Roach R, Wagner P, HackettP, eds. Hypoxia. New York, USA: Springer, 305–14.

Schoenberg M, Kiemeney L, Walsh PC, et al. (1996). Germlinetranslocation t(5;20)(p15;q11) and familial transitional cell carcin-oma. J Urol 155:1035–6.

Schofield L, Hewitt MC, Evans K, et al. (2002). Synthetic GPI as acandidate anti-toxic vaccine in a model of malaria. Nature 418:785–9.

Schultz IJ, Kiemeney LA, Witjes JA, et al. (2005). Cdc91L1 (PIG-U)mRNA expression in urothelial cell carcinomas. Int J Cancer 116:282–4.

Selzer-Plon J, Bornholdt J, Friis S, et al. (2009). Expression of prostasinand its inhibitors during colorectal cancer carcinogenesis. BMCCancer 9:201.

Shao N, Xue J, Guo Z. (2004). Chemical synthesis of a skeleton structureof sperm CD52 – a GPI-anchored glycopeptide. Angew Chem Int Ed43:1569–73.

Shen F, Wu M, Ross JF, et al. (1995). Folate receptor type. Gamma. Isprimarily a secretory protein due to lack of an efficient signal forglycosylphosphatidylinositol modification: protein characterizationand cell type specificity. Biochemistry 34:5660–5.

Shen Y-J, Ye D-W, Yao X-D, et al. (2008). Overexpression of cdc91L1(PIG-U) in bladder urothelial cell carcinoma: correlation with clinicalvariables and prognostic significance. BJU Int 101:113–19.

Siegel BA, Dehdashti F, Mutch DG, et al. (2003). Evaluation of 111In-DTPA-folate as a receptor-targeted diagnostic agent for ovariancancer: initial clinical results. J Nucl Med 44:700–7.

Solberg H, Ploug M, Høyer–Hansen G, et al. (2001). The murinereceptor for urokinase-type plasminogen activator is primarilyexpressed in tissues actively undergoing remodeling. J HistochemCytochem 49:237–46.

Sonoda G, Palazzo J, Du Manoir S, et al. (1997). Comparative genomichybridization detects frequent overrepresentation of chromosomalmaterial from 3q26, 8q24, and 20q13 in human ovarian carcinomas.Genes, Chromosomes Canc 20:320–8.

Takahashi M, Inoue N, Ohishi K, et al. (1996). PIG-B, a membraneprotein of the endoplasmic reticulum with a large lumenal domain, isinvolved in transferring the third mannose of the GPI anchor. EMBO J15:4254–61.

Takahashi S, Suzuki S, Inaguma S, et al. (2003). Down-regulatedexpression of prostasin in high-grade or hormone-refractory humanprostate cancers. Prostate 54:187–93.

Takos AM, Dry IB, Soole KL. (2000). Glycosyl-phosphatidylinositol-anchor addition signals are processed in Nicotiana tabacum. Plant J21:43–52.

Taron BW, Colussi PA, Wiedman JM, et al. (2004). Human Smp3p addsa fourth mannose to yeast and human glycosylphosphatidylinositolprecursors in vivo. J Biol Chem 279:36083–92.

The UniProt Consortium. (2012). Reorganizing the protein space atthe universal protein resource (uniprot). Nucleic Acids Res 40:D71–5.

Tiede A, Nischan C, Schubert J, Schmidt RE. (2000). Characterisation ofthe enzymatic complex for the first step in glycosylphosphatidylino-sitol biosynthesis. Intl J Biochem Cell Biol 32:339–50.

Toh YK, Kamariah N, Maurer-Stroh S, et al. (2011). Structural insightinto the glycosylphosphatidylinositol transamidase subunits PIG-Kand PIG-S from yeast. J Struct Biol 173:271–81.

Tsuchiya N, Kondo Y, Takahashi A, et al. (2002). Mapping and geneexpression profile of the minimally overrepresented 8q24 region inprostate cancer. Am J Pathol 160:1799–806.

Ueda Y, Yamaguchi R, Ikawa M, et al. (2007). PGAP1 knock-outmice show otocephaly and male infertility. J Biol Chem 282:30373–80.

Umezu T, Shibata K, Shimaoka M, et al. (2010). Gene silencing ofglypican-3 in clear cell carcinoma of the ovary renders it moresensitive to the apoptotic agent paclitaxel in vitro and in vivo. CancerSci 101:143–8.

Vainauskas S, Cortes LK, Taron CH. (2012). In vivo incorporation of anazide-labeled sugar analog to detect mammalian glycosylphosphati-dylinositol molecules isolated from the cell surface. Carbohydr Res362:62–9.

18 D. G. Gamage & T. L. Hendrickson Crit Rev Biochem Mol Biol, Early Online: 1–19

Cri

tical

Rev

iew

s in

Bio

chem

istr

y an

d M

olec

ular

Bio

logy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f Z

ueri

ch Z

entr

um f

uer

Zah

n M

und

und

on 0

8/31

/13

For

pers

onal

use

onl

y.

Vainauskas S, Maeda Y, Kurniawan H, et al. (2002). Structuralrequirements for the recruitment of GAA1 into a functionalglycosylphosphatidylinositol transamidase complex. J Biol Chem277:30535–42.

Vainauskas S, Menon AK. (2004). A conserved proline in the lasttransmembrane segment of GAA1 is required for glycosylpho-sphatidylinositol (GPI) recognition by gpi transamidase. J BiolChem 279:6540–5.

Varma Y, Hendrickson T. (2010). Methods to study GPI anchoring ofproteins. ChemBioChem 11:623–36.

Vastag BB. (2000). Folate gains momentum as a vehicle for drugdelivery. J Natl Cancer Inst 92:1800–1.

Vishwakarma RA, Menon AK. (2005). Flip-flop of glycosylpho-sphatidylinositols (GPI’s) across the ER. Chem Commun 4:453–5.

Waldron NN, Oh S, Vallera DA. (2012). Bispecific targeting of EGFRand uPAR in a mouse model of head and neck squamous cellcarcinoma. Oral Oncol 48:1202–7.

Watanabe R, Inoue N, Westfall B, et al. (1998). The first step ofglycosylphosphatidylinositol biosynthesis is mediated by a complex ofPIG-A, PIG-H, PIG-C and GPI1. EMBO J 17:877–85.

Watanabe R, Murakami Y, Marmor MD, et al. (2000). Initial enzyme forglycosylphosphatidylinositol biosynthesis requires PIG-P and isregulated by DPM2. EMBO J 19:4402–11.

Watanabe R, Ohishi K, Maeda Y, et al. (1999). Mammalian PIG-L andits yeast homologue Gpi12p are N-acetylglucosaminylphosphatidyli-nositol de-N-acetylases essential in glycosylphosphatidylinositol bio-synthesis. Biochem J 339:185–92.

Wilhelm O, Weidle U, Hohl S, et al. (1994). Recombinant solubleurokinase receptor as a scavenger for urokinase-type plasminogenactivator (uPA): inhibition of proliferation and invasion of humanovarian cancer cells. FEBS Lett 337:131–4.

Wilting SM, Snijders PJF, Meijer GA, et al. (2006). Increased gene copynumbers at chromosome 20q are frequent in both squamous cellcarcinomas and adenocarcinomas of the cervix. J Pathol 209:220–30.

Wood LD. (2012). Folate receptor alpha: a new tool in the diagnosis andtreatment of lung cancer. Oncotarget 3:668–9.

Wu G, Guo Z, Chatterjee A, et al. (2006). Overexpression ofglycosylphosphatidylinositol (GPI) transamidase subunits phosphati-dylinositol glycan class T and/or GPI anchor attachment 1 inducestumorigenesis and contributes to invasion in human breast cancer.Cancer Res 66:9829–36.

Xia M-Q, Hale G, Lifely MR, et al. (1993). Structure of the campath-1antigen, a glycosylphosphatidylinositol-anchored glycoprotein whichis an exceptionally good target for complement lysis. Biochem J 293:633–40.

Yan W, Ratnam M. (1995). Preferred sites of glycosylphosphatidylino-sitol modification in folate receptors and constraints in the primarystructure of the hydrophobic portion of the signal. Biochemistry 34:14594–600.

Yiping H, Jatin N, Barry T, Edward R. (2009). Tobacco smoke activatesprotein association and tyrosine phosphorylation of the GPI-transamidase complex subunits in human cancers. J Epith BiolPharm 2:14–22.

Yokota T, Yoshimoto M, Akiyama F, et al. (1999). Frequent multipli-cation of chromosomal region 8q24.1 associated with aggressivehistologic types of breast cancers. Cancer Lett 139:7–13.

Yun-Yan X, Virginia L, Jorge F. (2001). Glypican-3 expression issilenced in human breast cancer. Oncogene 20:7408–12.

Zacks MA, Garg N. (2006). Recent developments in the molecular,biochemical and functional characterization of Gpi8 and the GPI-anchoring mechanism. Mol Membr Biol 23:209–25.

Zhao P, Nairn AV, Hester S, et al. (2012). Proteomic identification ofglycosylphosphatidylinositol anchor-dependent membrane proteinselevated in breast carcinoma. J Biol Chem 287:25230–40.

Zhu Y, Fraering P, Vionnet C, Conzelmann A. (2005). Gpi17p does notstably interact with other subunits of glycosylphosphatidylinositoltransamidase in Saccharomyces cerevisiae. Biochim Biophys Acta1735:79–88.

DOI: 10.3109/10409238.2013.831024 GPI Transamidase and GPI anchored proteins 19

Cri

tical

Rev

iew

s in

Bio

chem

istr

y an

d M

olec

ular

Bio

logy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Uni

vers

ity o

f Z

ueri

ch Z

entr

um f

uer

Zah

n M

und

und

on 0

8/31

/13

For

pers

onal

use

onl

y.