Analysis of Porcine Kupffer Cell Recognition of Human Erythrocytes

Christopher Burlak II

Medical College of Ohio

2003 ii

DEDICATION

To my grandfather, who revealed to me as a child, how fulfilling creating can be.

My earliest memory of helping my grandfather was repairing the garage. I was 4 and my

task was to straighten bent nails with a hammer against the concrete. I was elated just to

be with him as I would miss the nail and painfully tap my finger. After I had finished my

can of nails I suppose I had passed the “test,” because through the years there had been an

endless list of projects that we had worked on together. Only now, after he has passed, I realize that the greatest thing we ever built was our relationship.

iii

ACKNOWLEDGMENTS

I would like to thank my wife, Stacey, for her understanding and patience. She has been my lighthouse amidst a treacherous shore.

Michael Rees, thank you. Without your support and advice, I surely would have faltered. Along the path of research you have guided me without leading me. I could not have come this far if it were not for you. Again, thank you.

I also would like to thank Lisa Twining for her support and friendship.

I owe many thanks to my Ph.D. committee Dr. David Aminoff, Dr. David

Dignam, Dr. Jerzy Jankun, and Dr. Julie Westerink. Whether consulting them individually or through a committee meeting, they have showed me respect and kindness.

I deeply appreciate that they have consistently been interested in my development as a person as well as a scientist.

There have been many people along the way who have given me guidance, reagents, support, and an ear…thank you all so much.

TABLE OF CONTENTS

Dedication...... ii

Acknowledgements...... iii

Table of Contents...... iv

Introduction...... 1

Manuscript 1: Carbohydrates borne on human glycophorin A are recognized by

porcine Kupffer cells...... 10

Manuscript 2: Terminal Sialic Acid Residues on Human Glycophorin A are

Recognized by Porcine Kupffer Cells...... 38

Manuscript 3: Porcine Annexin IV: a Lectin That Mediates Xenogeneic

Recognition of Human Erythrocytes ...... 66

Summary...... 92

Bibliography ...... 105

Abstract...... 130

INTRODUCTION

The goal of this project was to develop a means of extracorporeal liver support for

patients with fulminant hepatic failure to serve as a bridge to transplantation. It was

hypothesized that humoral and cellular components of human blood would injure the porcine liver and thus, human decay-accelerating factor (hDAF)-transgenic pig livers were expected to perform better than wild-type pig livers. In fact, it was found that human

blood caused relatively little damage to porcine livers (Rees et al., 2002a). The finding

was the opposite--that porcine livers injured the human blood. During 72 h of

extracorporeal perfusion nearly all of the human erythrocytes were lost from the perfused

blood (Figure 1) (Rees et al., 2002a). The mechanism behind this loss of human

erythrocytes became the major focus of the laboratory. It was initially thought that the

human erythrocytes were being destroyed by humoral mechanisms. The first barrier in

porcine to human xenografts is that of hyperacute rejection mediated primarily by natural antibodies with specificity to galactose α-1,3 galactose and complement (Galili et al.,

1988a,b; Cooper et al., 1993, 1994; Sandrin et al., 1993, 1994; Rees et al., 2002a). A disparity in complement regulation of human complement by porcine complement regulatory proteins exacerbate this species combination (Dalmasso et al., 1991; Cozzi et al., 1995). Several groups have produced pigs transgenic for human complement regulatory proteins (Fodor et al., 1994; Cozzi et al., 1995; Byrne et al., 1997; Cowan et al.,

2000; Costa et al., 2002). Just as humans destroy porcine cells by classical pathway-

Figure 1. Hematocrit During Extracorporeal Liver Perfusion

Figure 1. Hematocrit During Extracorporeal Liver Perfusion 100 90 80 e u

l 70 a 60 ng V ti r a t 50

t of S 40 n e c r

e 30

P Transgenic 20 Nontransgenic Alloperfusion 10 Human Blood/No Liver 0 0 1020304050607080 Time (hours) Transgenic, hDAF transgenic porcine livers perfused with human blood (n=5); Non-transgenic, wild-type porcine livers perfused with human blood (n=5); Alloperfusion, wild-type porcine livers perfused with wild-type allogeneic porcine blood (n=5); Human blood no liver, human blood perfused in the circuit in the absence of liver (n=1). Hematocrit is expressed as a percent of the starting hematocrit (transgenic versus nontransgenic, P=0.223; alloperfusion versus nontransgenic, P<0.012; alloperfusion versus transgenic, P<0.004).

mediated complement activation, so pigs might destroy human cells. It was proposed that porcine livers were producing porcine complement and antibodies, and together these porcine components were causing lysis of human erythrocytes that were not well protected by species-specific human complement regulatory proteins.

While porcine livers can mount a humoral graft versus host reaction, porcine antibody and complement do not cause classical pathway-mediated lysis of human

Figure 2. Human Erythrocytes are Bound and Phagocytosed by Porcine Kupffer Cells

a b

RBC Kupffer Hepatocyte Cell c Sinusoidal d Endothelium 2 µm

RBC

RBC Kupffer Cell RBC RBC binding RBCs

H&E, Perl’s stain and electron microscopy of a porcine liver perfused for 72 h with human blood. A-Hematoxylin and eosin stain. Arrows indicate hypertrophied Kupffer cells with multiple extracellular and intracellular phagocytosed erythrocytes (original magnification X 400). B-Perl’s stain. Arrows indicate hypertrophied Kupffer cells with prominent blue staining of intracellular iron breakdown products (original magnification X 400). C-Transmission electron micrograph of a porcine Kupffer cell (arrow) demonstrating erythrocytes (RBC) bound on the cell surface, (magnification, X 5000). D-Scanning electron micrograph showing the fenestrated liver sinusoidal endothelium with attached Kupffer cell (arrow) binding multiple erythrocytes (magnification X 1200).

erythrocytes during extracorporeal liver perfusion (Rees et al., 2002b, 2003b). Porcine

Kupffer cells in the liver recognize human erythrocytes and extract them from the recognition of human erythrocytes does not require opsonization by antibody or complement suggesting that porcine Kupffer cells have a receptor allowing them to recognize xenogeneic human erythrocytes (Rees 2003c).

There are similarities between these observations and the field of the recognition and extraction of senescent erythrocytes by the reticuloendothelial system. First, researchers in this field had already used the recognition of xenogeneic erythrocytes to help them understand the binding of senescent erythrocytes. Thus, there was already a precedent that macrophages were capable of recognizing xenogeneic erythrocytes. Yet despite this recognition in at least 16 publications, none of these previous investigators had used this recognition to study the role of macrophages in xenograft rejection (Lee et al.,

1968; Eibl 1971; Maruta et al., 1971; Elson et al., 1972; Jaroslow et al., 1973; Maruta

1973; Lalezari et al., 1974; Nomoto et al., 1974; Ouchi et al., 1976; Watanabe et al., 1980;

Rothlein et al., 1982; Vomel et al., 1985, 1988; Mohr et al., 1987; Crocker et al., 1988;

Bratosin et al., 1997). Uri Galili came to xenotransplantation from the field of senescent erythrocyte research. In his previous field, Galili was involved in a debate with other researchers as to the mechanism of extraction of senescent erythrocytes. This debate was reminiscent of the early days of immunology and the debate between Paul Ehrlich and Elie

Metchnikoff as to whether innate and acquired immunity to infection could be best explained by cellular or by humoral mechanisms (Silverstein 1989).

Humoralists (like Galili) believed that antibody provided the specificity of the

system, while the cellular components of the reticuloendothelial system simply non-

specifically phagocytosed erythrocytes marked for destruction by the carbohydrate-specific

antibodies (Galili 1988). Cellularists, including Schlepper-Schäfer and Kolb-Bachofen,

believed that carbohydrate-specific receptors on the surface of macrophages provided the

specificity and that opsonization was not necessary (Schlepper-Schafer et al., 1988).

David Aminoff took an intermediate position arguing that specificity resulted from both

humoral and cellular events, but argued as to what antigen was specifically recognized

(Aminoff 1988). Most agreed that the antigen being recognized was a carbohydrate, but

whether that carbohydrate was desialyated glycophorin exposing multiple galactose-β(1,3)

N-acetylgalactosamine or galactose α(1,3) galactose remained in dispute (Aminoff 1988;

Galili 1988). Thus it appears possible the receptor we discovered on porcine Kupffer cells

that recognized human erythrocytes is a carbohydrate-specific protein receptor (a lectin).

Lectins mediate carbohydrate binding by way of highly conserved regions within

the lectin polypeptide; this region is known as the carbohydrate-recognition domain

(CRD). In general, these regions have only moderate affinity for a given sugar, but lectins

generate high avidity by expressing multiple CRDs within a single lectin or by forming

oligomers of several subunits. In this way, many CRDs binding with moderate affinity

result in a high overall avidity. Until recently, most lectins (e.g., Concanavilin A) were

derived from plants or lower vertebrates (Janeway 1999; Watkins 2001; Kilpatrick 2002).

It is now clear that lectins are at least as abundant in mammals as in plant tissues and play

an important role in immune system function; lectins are found in all kingdoms of life and viruses(Vasta 1991; Vasta et al., 1994, 1996; Loris 2002).

The first to propose a role for carbohydrate-lectin interactions in the innate immune cells involved in xenograft rejection was Inverardi et al., (Inverardi et al., 1996; Inverardi et al., 1997). Three lines of evidence supported the proposal that human natural killer

(NK) cells recognize carbohydrate differences on porcine endothelium as a signal for binding and lysis. Inverardi’s group demonstrated that human NK cells binding to porcine endothelial cells was selectively inhibited by Melibiose (6-α-D-Galp-D-Glc) and α-D-

Mannose, that human natural antibody F(ab’)2 fragments (of which 1% bind to galactose

α-1,3 galactose) block the binding, and that human NK cells bind only to COS cells transfected with galactose α-1,3 galactose transferase and not vector only controls. In their discussion they suggest several different NK cell surface lectin receptors (NKRP-1, NKG-

2, and CD94) as possible candidates for mediating these effector functions (Inverardi et al.,

1997).

Others have proposed a role for lectin receptors in the recognition of xenografts, but in the context of NK cells (Bach et al., 1996; Candinas et al., 1996; Hancock et al.,

1997). Hancock et al., (1997) suggested that selective targeting of macrophage lectins with monoclonal antibodies, peptide inhibitors or carbohydrate inhibitors may provide an approach to blocking macrophage infiltration, activation and cytokine production during the development of delayed xenograft rejection. Itescu et al., have demonstrated that primate monocytes interact with galactose α-1,3 galactose on the surface of porcine

endothelial cells (Artrip et al., 1999; Kwiatkowski et al., 2000a,b). The transgenic expression of α-1,2 fucosyltransferase in porcine endothelial cells reduced both galactose

α-1,3 galactose cell surface expression and the susceptibility of such endothelial cells to human NK cell-mediated lysis and human monocyte adhesion (Artrip et al., 1999;

Kwiatkowski et al., 1999). Miyagawa et al., (1999) suggested a role for galactose α-1,3 galactose-dependent adhesion of human NK cells to porcine endothelial cells. Sheikh et al., (2000) proposed that galactose α-1,3 galactose may not be the only carbohydrate recognized. Holgersson et al., (2002) note that no galactose α-1,3 galactose binding receptor has been identified on human innate immune cells, suggesting that these findings should be viewed cautiously.

Pigs transgenic for complement regulatory proteins demonstrated that lectin receptors on innate immune cells play an important role in xenograft rejection. A new form of rejection has been identified that occurs 2-4 days after engraftment and is referred to as delayed xenograft rejection (DXR) or acute vascular rejection (AVR) (Bach et al.,

1996; Platt et al., 1998; Platt 2000). While this form of graft rejection remains undefined, it is likely that humoral immune responses, endothelial cell activation, platelet aggregation, disparities in coagulation regulation and innate cellular responses play a role (Bach et al.,

1996; Platt et al., 1998; Platt 2000). Although the immunological mechanisms underlying delayed xenograft rejection remain controversial, there is evidence that macrophages play an important role. Techniques that prevent the occurrence of hyperacute rejection have been employed to demonstrate the crucial role of macrophages in mediating delayed

xenograft rejection (Blakely et al., 1994; Fryer et al., 1994; Satake et al., 1994; Wallgren et

al., 1995; Lin et al., 1997; Fox et al.,1998; Itescu et al., 1998; Xu et al., 1998; Xia et al.,

2000). Goddard et al., demonstrated in a pig to baboon cardiac xenograft model that the

principal cell type found in rejecting grafts were macrophages (Goddard et al., 2002). The controversy is no longer whether or not macrophages are involved, but whether macrophages can recognize and destroy xenografts independently or if they require additional signals from other cellular or humoral sources.

This concept of innate cell-mediated lectin receptor recognition is consistent with recent discoveries showing that the innate immune system has specific receptors that recognize the “molecular signature” of microbial infections such as carbohydrates, bacterial cell wall structures, and double-stranded RNA (Fearon et al., 1996). Medzhitov

(2001) and Janeway (1989) have referred to these antigens as pathogen-associated molecular patterns (PAMP) and to the innate immune receptors recognizing these patterns as pattern-recognition receptors (PRR). The role of these receptors in immune responses is well recognized. It is important to explore the relationship of these receptors to the recognition of relevant xenotransplantation antigens.

The mouse macrophage galactose/N-acetylgalactosamine-specific lectin (MMGL) targets foreign or abnormal oligosaccharides. It requires Ca2+ for binding and is involved in the recognition of tumor cells by macrophages (Yamamoto et al., 1994; Sakamaki et al.,

1995; Ni et al., 1996). A human analog of MMGL has been identified (Ni et al., 1996;

Suzuki et al., 1996). The MMGL preferentially binds to highly branched N-glycans with

terminal galactose residues and clusters of truncated O-glycans, sugars that are

characteristic on the surface of tumor cells (Springer 1989; Ni et al., 1996). This receptor

is similar to the GalNAc receptor on rat Kupffer cells described by Mohr (1987). A rat

Kupffer cell GalNAc receptor cloned by Hoyle and Hill (1989) likely encodes the receptor

identified functionally by Mohr et al., as being involved with the rat Kupffer cell

recognition of xenogeneic erythrocytes (Hoyle et al., 1988). Interestingly, the GalNAc

receptor is expressed exclusively on resident liver macrophages (Haltiwanger et al., 1986;

Lehrman et al., 1986; Tiemeyer et al., 1992).

Our laboratory proposed to examine the very specific question of how porcine

Kupffer cells recognize human erythrocytes. In examining the significance of this study, we suggest two applications. First, an understanding of how porcine Kupffer cells recognize and phagocytose human erythrocytes may lead to a means to interfere with this recognition, and increase the likelihood of success of extracorporeal liver perfusion for supporting the life of a human patient in fulminant liver failure. A second application may be the uncovering of a relatively overlooked aspect of xenogeneic recognition. While there has been a great deal of focus on the recognition of xenogeneic carbohydrate differences by the humoral immune system, the innate cellular recognition of xenogeneic carbohydrates could play an equally important role in xenotransplantation.

MANUSCRIPT 1

Carbohydrates borne on human glycophorin A are recognized by porcine

Kupffer cells.

Abstract

BACKGROUND: We have previously shown that when porcine livers are perfused

with human blood, porcine Kupffer cells extract up to 3 units of human red blood cells

(RBC) over the course of a 72h perfusion. We have previously hypothesized that the

recognition event responsible for this interaction involves a lectin receptor on the surface

of the porcine Kupffer cell interacting with a carbohydrate epitope on the surface of the

human RBC.

METHODS: Treatments to disrupt the protein core of purified glycoproteins from

the surface of human RBC included: pronase, trypsin, 2-ME, and heating to 90°.

Alternatively, we have removed the carbohydrate residues from purified human RBC glycoproteins using glycosidases. RBC binding in the presence or absence of treated glycoproteins was quantified by 51Chromium-labeled RBC recognition by primary cultures of porcine Kupffer cells.

RESULTS: Human, but not porcine, RBC were bound by in vitro primary cultures of porcine Kupffer cells. The binding of human RBC was inhibited by pre-incubation of

porcine Kupffer cells with purified human erythrocyte glycoproteins (hEGP) from the membrane of human RBC. Pre-treatment of hEGP with pronase, trypsin, 2-ME or heating did not interfere with the ability of hEGP to inhibit the binding of human RBC to porcine

Kupffer cells. Deglycosylation of the purified hEGP completely disrupted the ability of hEGP to inhibit the binding of human RBC to porcine Kupffer cells.

CONCLUSIONS: We conclude that porcine Kupffer cells bind xenogeneic human

RBC by recognition of a carbohydrate epitope on the surface of human RBC. We

hypothesize that this binding is mediated by a porcine Kupffer cell lectin receptor.

Introduction

The importance of carbohydrate recognition by the humoral immune system in

xenogeneic hyperacute rejection is now well recognized (Galili 1993; Cooper et al., 1994).

In addition, discoveries over the last decade have made clear the importance of

carbohydrates and other molecular signatures of pathogens as ligands for specific receptors

utilized by the innate cellular immune system (Fearon et al., 1996). Proteins that recognize

carbohydrate ligands are known as lectins and the field of lectin research has demonstrated

the importance of these highly conserved proteins in both humoral and innate cellular

recognition (Goldstein 1980; Vasta et al., 1994; Loris 2002). Previous investigators have

suggested that carbohydrate recognition by lectin receptors expressed on the surface of

innate immune cells may play an as yet poorly recognized role in xenograft rejection

(Hancock et al., 1997; Artrip et al., 1999; Miyagawa et al., 1999). Our work in

extracorporeal liver perfusion demonstrated that porcine Kupffer cells recognize and

destroy human erythrocytes and led us to hypothesize a role for innate cellular lectin

receptors in the mechanism responsible for this xenogeneic immune response (Rees

2003c).

In contrast to the highly mutable genes of the adaptive immune system, macrophages appear to have an arsenal of receptors locked into their germline configuration (Stewart et al., 1989). It is these receptors that have attracted the interest of a small number of researchers in the field of xenotransplantation (Hancock et al., 1997;

Artrip et al., 1999; Miyagawa et al., 1999).

Our finding that porcine Kupffer cells specifically bind human but not porcine

erythrocytes led us to use this interaction in order to study the recognition of xenogeneic

targets by macrophage receptors (Rees et al., 2002b). The study of erythrocyte binding by

macrophages has already received significant attention in the study of how senescent

erythrocytes are recognized and extracted from the circulation (Aminoff 1988; Schlepper-

Schafer et al., 1988; Galili 1993). In fact, it was the study of this mechanism that led Galili to identify the importance of natural antibodies directed against galactose α-1,3 galactose prior to his recognition of the importance of these antibodies in the field of xenotransplantation (Galili 1993). While Galili focused on the importance of humoral recognition in the specific removal of effete erythrocytes, others in this field provided evidence that macrophage lectins provided the specificity observed (Aminoff 1988;

Schlepper-Schafer et al., 1988; Galili 1993). Without recognizing the potential overlap with the field of xenotransplantation, several investigators used the recognition of xenogeneic erythrocytes as a model system to better understand how senescent erythrocytes are eliminated (Rothlein et al., 1982; Mohr et al., 1987; Crocker et al., 1988;

Bratosin et al., 1997). These investigators have provided evidence that carbohydrates such

as N-acetylgalactosamine and lectins such as the GalNAc/Gal-particle receptor are

important in the recognition of xenogeneic erythrocytes by innate immune cells (Mohr et

al., 1987). Using strategies borrowed from the field of senescent erythrocyte research, we

have investigated the mechanism whereby porcine Kupffer cells recognize and bind human

erythrocytes.

We present evidence that porcine Kupffer cells bind human erythrocytes by recognition of carbohydrate, rather than protein antigens. We suggest that human glycophorin A is one of the glycoproteins bearing the carbohydrate responsible for recognition of the human erythrocyte. The prominence of lectins as innate germline receptors, the recognition of galactose α-1,3 galactose during xenotransplantation, lectin recognition of xenogeneic erythrocytes and the evidence put forth here, lead us to propose that porcine Kupffer cells recognize human erythrocytes via a lectin/carbohydrate interaction.

Materials and Methods

Isolation of Porcine Kupffer Cells

Large white pigs (15 - 20 kg) were pre-medicated with a combination of Ketamine

(22-33 mg/kg) (Abbott Laboratories, Abbot Park, IL) and Xylazine (2 mg/kg) (AnaSed,

Lloyd Laboratories, Shenandoah, IL) IM. Medicated pigs were anesthetized with

Isoflurane (1-1.5%). The left lobe of the liver was excised and infused with ice cold

collagenase (0.05%)(Sigma, St. Louis, MO) Using sterile technique, approximately one

third of the liver lobe was sliced into 0.5 – 1 cm thick slices, mixed and infused with fresh

collagenase and incubated for 45 min at 37oC with gentle agitation. Liver slices were minced in ice cold phosphate buffered saline (Oxoid Inc. PBS tablets, Ogdensburg, NY) and then passed through 500, 212, and 106 micron sieves (CSC Scientific Inc, Fairfax,

VA).

The resulting suspension was supplemented with 10% fetal calf serum (Invitrogen,

Grand Island, NY), and allowed to sediment in 250 ml conical tubes on ice for 30 min.

The resulting supernatant was centrifuged at 600 g for 15 min. The supernatant was discarded and the pellet was resuspended in PBS + 10% FCS for a second 30 min sedimentation. The supernatant was centrifuged at 600 g for 15 min. The pellets then were combined, washed once with PBS + 10% FCS and brought up to a final volume of

250 ml in PBS + 10% FCS.

The resulting suspension was layered over a metrizamide solution [33.8g Sodium

metrizamide (Sigma, St. Louis, MO), 110ml of distilled water, 2ml of 1M Hepes solution

(Sigma, St. Louis, MO), 0.036g CaCl22H2O (BDH 100704Y), 0.1g KCL (Sigma, St.

Louis, MO), pH adjusted to 7.6 with NaOH (Sigma, St. Louis, MO), final density 1.08 –

1.09 units and 0.2µm filter sterilized]. Twenty-five ml of cell suspension was layered over

13 ml of metrizamide solution and centrifuged at 3007 g for 45 min at 4oC. The cells at the interface were carefully removed, suspended in Hank’s Balanced Salt Solution (HBSS,

Invitrogen, Grand Island, NY), supplemented with 10 % FCS and centrifuged at 469 g for

7 min. The pellets were combined and sequentially washed with HBSS with 10 % FCS until the supernatant contained only rare cellular debris. The pellet then was washed once with CMRL media [CMRL – 1066 medium, (Invitrogen, Grand Island, NY) with 100 U/ml of Penicillin – Streptomycin (Invitrogen, Grand Island, NY) and 2 mM L-Glutamine

(Invitrogen, Grand Island, NY)] supplemented with 10 % FCS and the resulting pellet was resuspended in 250ml CMRL media. The cell suspension was distributed equally to five

2 o 150 cm tissue culture flasks. These flasks were placed in a humidified, CO2, 37 C incubator overnight to allow the macrophages to adhere. After 24 h, the CMRL media was replaced to remove the non-adherent cells. We generally found that adherent cells remained viable for at least 1 mo.

Rosetting Assay

Kupffer cells, scraped from 150 cm2 culture flasks, were washed in CMRL media

(without FCS) three times and pelleted at 600 g to remove FCS and cell debris. The volume of the final pellet was adjusted to 1x106 cells per ml. Porcine Kupffer cells were

allowed to adhere to 13mm round cover glasses in a humidified, CO2, 37°C incubator for 2 hours in CMRL media (without FCS). The media then was removed and replaced with

250µl of fresh CMRL media (without FCS) or a previously prepared sample.

Samples to be tested were brought to the appropriate concentrations in 250 µl

CMRL media (without FCS) and incubated with Kupffer cells for 15 minutes. After the 15

min incubation, 500 µl of a 0.01 % solution of either 51Cr-labeled human or porcine erythrocytes (described below) was added to each well. Erythrocytes and Kupffer cells were incubated for 2 h to allow for the formation of erythrocyte rosettes with the Kupffer cells. Erythrocytes were isolated and labeled with 51Chromium as previously described

(Morrissey et al., 1992).

After the 2 h rosetting incubation, these coverglasses were gently washed in CMRL

media to remove unbound erythrocytes and then were placed in scintillation vials to be

counted on a gamma counter (Beckman Gamma 5500b, Beckman Instruments Inc.,

Fullerton, CA). Each assay was performed a minimum of three times in triplicate and the

results presented as the mean value for each experiment with standard error.

Isolation and Analysis of Purified Glycophorin

Human and porcine erythrocyte membrane fragments were prepared according to

Dodge (1963) (Dodge et al., 1963). Human and porcine membrane glycoproteins were isolated from the fragments as described by Marchesi V.T. et al., (Marchesi et al., 1971;

Marchesi et al., 1972).

Human and porcine erythrocyte glycoprotein (EGP) samples were tested for protein and carbohydrate content using the bicinchoninic acid (BCA) method (Pierce Inc.,

Rockford IL) and the phenol/sulfuric acid detection method, respectively.(Moriyasu et al.,

1982) Samples were subjected to 4-12% gradient SDS-PAGE (NuPAGE precast gel system, Invitrogen Inc., Grand Island, NY) and transferred to PVDF membranes for

Western blotting. Whole human erythrocyte membranes solubilized in 0.5% sodium dodecylsulfate (SDS) were used as a control for both positive and negative binding.

Membranes were blotted with either anti-human glycophorin A monoclonal antibody

(1:2500) (DAKO Inc. Carpinteria, CA) or anti-Band-3/glycophorin B monoclonal antibody

(1:2500) (Sigma Inc., St. Louis, MO). A goat anti-mouse secondary antibody (1:10,000)

(Sigma, St. Louis, MO ) conjugated to horseradish peroxidase was used to detect binding when reacted with diamino- benzidine (DAB) reagent (Biorad Inc., Hercules, CA). Gels were stained with either GelCode Blue (Pierce Inc., Rockford, IL) for protein or the

Glycoprotein Staining Kit (Pierce Inc., Rockford, IL) for carbohydrates. Manufacturer’s instructions were observed for both electrophoresis and Western blotting.

Immunohistochemistry

Kupffer cells were isolated and prepared as described in the rosetting assay above.

After adherence to 13 mm round coverglasses, Kupffer cells were blocked with BSA

(1mg/ml) in CMRL media for 2 h prior to the addition of samples. Human EGP

(500ug/ml) in CMRL media with BSA (1mg/ml) was incubated with Kupffer cells for 15

min prior to washing. The isotype control, mouse IgG1 (1:2500), for the primary antibody

was purchased from Dako (Carpinteria, CA). Anti-mouse IgG conjugated to HRP (Sigma

Inc., St Louis, MO.) was used as the secondary antibody and this was developed with

Vector Blue stain (Vector Labs, Inc., Burlingame, CA ).

Trypsin or Pronase Degradation

Human EGP was treated with trypsin (200 U) (Sigma Inc. St. Louis, MO) or

pronase (100 U) (Sigma Inc. St. Louis, MO) for 1 h at 37°C and then heated to 95°C for 5 min to inactivate the enzyme. Samples were diluted to 500 µg protein/ml CMRL media

(without FCS) for the rosetting assay. Twenty µl of each sample was saved for electrophoresis and Western blotting.

Heat and 2-ME Treatment

Human EGP was dissolved in CMRL media (without FCS)(1mg/ml) and subjected to either 95°C for 5 min, 8 mM β-mercaptoethanol, or both. Samples were diluted to 500

µg protein/ml CMRL media for the rosetting assay.

Enzymatic Deglycosylation of Human EGP

Based on the known glycosylation pattern of human glycophorin A (the most

prevalent protein band from the isolated human erythrocyte membrane proteins), the

following enzymes were obtained to completely cleave all the predicted carbohydrate

moieties: PNGaseF, sialidase A, beta 1-4 galactosidase, β N-acetylhexosaminidase, endo-

O-glycosidase and hexosaminidase (Prozyme) (Schenkel-Brunner 2000). The enzymes

were used to deglycosylate 2 mg of human EGP for each experiment performed in

triplicate. Sialidase A (0.1 units), 100 units of PNGaseF, 0.06 units of β 1-4 galactosidase,

16.6 units of β N-acetylhexosaminidase, and 0.025 units of endo-O-glycosidase were

added to 2 mg human EGP in NaHPO3 (pH 6) buffer (supplied with enzymes) to completely deglycosylate human EGP. As a control, 2 mg of human EGP was incubated under the same conditions without enzymes. Both the treated and the sham-treated samples were incubated at 37°C for 4 d then heated to 95°C for 5 min to inactivate the enzyme. Treated and sham treated human EGP were centrifuged in a YM-10 centricon concentrator (Millipore Inc., Bedford, MA) seven times to remove the reaction buffer. The reaction buffer was replaced with CMRL media used in the rosetting assay during the last step in the sample preparation. Deglycosylation was examined by SDS-PAGE followed by

GelCode Blue protein detection and carbohydrate detection using the Glycoprotein

Staining kit. Western blot analysis was performed with an anti-human glycophorin A

monoclonal antibody. Samples were diluted to 500 µg protein/ml CMRL media (without

FCS) for the rosetting assay.

Inhibition of Rosetting with EDTA and EGTA

Porcine Kupffer cells were incubated with either Ethylenediaminetetraacetic acid

(EDTA)(25mM) or Ethylenebis (oxyethylenenitrilo) tetraacetic acid (EGTA)(25mM) solubilized in CMRL media for 15 min prior to the addition of human or porcine erythrocytes labeled with 51Chromium and suspended in CMRL media containing 25mM

EDTA or EGTA. Samples were treated following the methods of the rosetting assay

described above.

Results

Isolation and Analysis of Human and Porcine EGP

To examine their potential carbohydrate ligands we isolated the membrane

glycoproteins from human and porcine erythrocytes. Protein and carbohydrate staining

after SDS-PAGE analysis of the isolated glycoproteins demonstrated that the predominant

band of the hEGP was 62 kDa (Figure 1a and b) and heavily glycosylated. Anti-human

glycophorin A Western blot analysis revealed that the sample was primarily composed of

the human glycophorin A dimer at 62 kDa and trimer at 93 kDa (Figure 1c). Anti-human

Band 3/glycophorin B Western blot analysis revealed no binding in the human EGP

sample lanes (Figure 1c). In contrast, the positive control lane of whole human

erythrocytes showed positive binding at the molecular weights corresponding with band 3,

glycophorin B and heterodimers of glycophorin B with other glycophorins (Figure 1c)

(Cochet et al., 2001). These data suggest that human glycophorin A was the predominant

glycoprotein present in the isolated human EGP.

A dose dependent curve was generated when a reciprocal dilution of human EGP

(500µg to 0.06125 µg protein/ml CMRL media) was tested in the rosetting assay (Figure

2). Human EGP (500 µg protein/ml CMRL media) but not porcine EGP (500 µg

protein/ml CMRL media) inhibited the rosetting assay compared to rosetting without EGP

(Figure 2). All further use of human EGP in the rosetting assay was used at the

concentration that caused 100% inhibition (500 µg protein/ml) so as to give the greatest

Figure 1. Analysis of Human and Porcine Erythrocyte Glycoproteins (EGP).

A B C Schiff’s base Coomassie staining Western Blotting staining α-Band3 α-hGA 191kDa

39 28 19 12 3 4 12341212

Human and porcine EGP isolated by LIS were subjected to SDS-PAGE analysis. Glycoproteins were reduced with 2-mercaptoethanol and heated for 5 minutes at 95°C. Gels were either stained blue for protein (A) with GelCode blue (Pierce) or magenta for glycoproteins (B) with periodic acid/Schiff’s base reagent. The human erythrocyte glycoproteins isolated previously and SDS-solubilized human erythrocyte membrane fragments (as a control lane) were subjected to SDS-PAGE and transferred to a PVDF membrane (C). The blot was probed with anti-glycophorin A or anti-band 3/glycophorin B monoclonal antibodies and detected with secondary antibody conjugated to HRP. The numbers on the bottom row correspond to the lane directly above containing human erythrocyte membrane fragments (1), LIS-isolated human EGP, (2), porcine erythrocyte membrane fragments (3), or LIS-isolated porcine EGP (4). Identification of glycophorin A corresponds to bands previously identified by western blot analysis.

Figure 2. Analysis of Human and Porcine EGP in the Rosetting Assay.

120

98 100 100

86 79 80

60 52 % Binding

20 20

8 -1 3 0 0.5 0.5 0.25 0.125 0.0625 0.0312 0.0156 0 0 (Concentration in mg/ml)

Pig Human EGP None EGP

Human rbc Pig rbc

Porcine Kupffer cells were incubated with either porcine or human erythrocytes radiolabeled with 51Chromium for 2 hours followed by gamma radiation detection. Kupffer cells that were incubated with human erythrocytes were pre-incubated with LIS isolated human erythrocyte glycoproteins at the indicated concentrations for 15 minutes prior to incubation with human erythrocytes radiolabeled with 51Chromium. Pre- incubation with LIS isolated porcine erythrocyte glycoproteins demonstrated no inhibition of human erythrocyte binding by porcine Kupffer cells.

amount of resolution when examining diminution of inhibition by various alterations of the isolated glycoproteins.

Immunohistochemistry

Kupffer cells incubated with human EGP were shown to specifically recognize human glycophorin A as indicated by blue staining (Figure 3). Kupffer cells incubated with human EGP and the isotype control antibody, human EGP without primary antibody, human EGP without secondary antibody and Kupffer cells not incubated with human EGP but treated with primary and secondary antibody were negative (Figure 3 and data not

shown). Kupffer cells not incubated with human EGP showed no blue staining (Figure 3).

The data presented in Figures 1-3 suggest that one of the carbohydrate moieties of

glycophorin A is the epitope that is recognized by porcine Kupffer cells when binding

human erythrocytes.

Heat and 2-ME Inhibition

To examine the importance of secondary and tertiary protein structures we treated

human EGP or CMRL media alone with heat (95°C for 15 min) and/or β-mercaptoethanol

(15mM) prior to incubation with Kupffer cells in the rosetting assay. These treatments had no effect on the activity of human EGP (Figure 4). Sham treatment (media without glycoproteins) had no effect on binding. These data suggest that neither secondary nor

tertiary protein structures are important in the ligand that is bound by porcine Kupffer cells.

Figure 3. Immunohistochemical Analysis of Porcine Kupffer Cells Ability to Bind

Human Glycophorin A.

100X 1000X

KC+GA+ Isotype control +2°

KC+No GA+ α-GA+2°

KC+GA+ α-GA+2°

Human erythrocyte glycoproteins were isolated as described previously and incubated (500ug/ml) with porcine Kupffer cells (in media with 2mg/ml BSA) in culture for 15 minutes followed by gentle washing to remove non-adherent glycoproteins and then fixed with acetone/methanol (1:1). Fixed Kupffer cells were subjected to immunohisotochemical staining with anti-human glycophorin A mouse monoclonal IgG antibodies or mouse monoclonal IgG antibodies directed towards various goat epitopes.

Figure 4. Removal of Secondary and Tertiary Conformations of Human EGP.

120

97 102 100 100

% binding 40

3 4 3 -1 -2 0 hGP hGP 90 media 90 hGP + 8mM media + 8mM hGP 90 me dia 90 Human rbc Porcine rbc degrees C degrees C 2-ME 2-ME degrees C + degrees C + 8mM 2-ME 8mM 2-ME

-20 Human glycoproteins were heated to 95°C for 5 minutes and/or treated with 15mM beta-mercaptoethanol. Identical treatments of media samples were used as controls. Samples were incubated with porcine Kupffer cells in culture for 15 minutes prior to the addition of 0.1% human erythrocytes labeled with Chromium51 in media. Human erythrocytes and porcine Kupffer cells were incubated in co-culture for 2 hours followed by gentle washing to remove non-adherent erythrocytes. Rosetting was verified by light microscopy and quantitated by gamma-detection.

Trypsin or Pronase Degradation

These findings, however, do not rule out refolded or linear protein epitopes as the

ligand of interest. Trypsin or pronase cleavage of human EGP was complete as revealed

by SDS-PAGE analysis and anti-human glycophorin A Western blotting (Figure 5a and b).

Trypsin or pronase treatment caused a large shift in migration of human glycophorins as observed by SDS-PAGE (Figure 5a and b). In addition, the anti-human glycophorin A

Western blot revealed that trypsin or pronase treatment destroyed the epitope recognized by the anti-human glycophorin A monoclonal antibody as no band was detected in contrast to the human erythrocyte control lane (Figure 5a and b). Despite obvious deterioration of the protein core, neither trypsin nor pronase impaired the ability of the treated glycoproteins to inhibit the binding of human erythrocytes by porcine Kupffer cells. These data suggest that protein fragmentation of the human erythrocyte glycoproteins does not impact their activity.

Deglycosylation of Human EGP

SDS-PAGE analysis of human EGP treated with the deglycosidase cocktail revealed an increase in migration of human glycophorin A (Figure 6). Anti-human glycophorin A western blotting revealed that the epitope recognized by the monoclonal antibody remained intact (Figure 6) unlike treatment with trypsin or pronase. When the deglycosylated human EGP were tested for their inhibitory potential in the rosetting assay,

deglycosylated human EGP had no ability to inhibit Kupffer cell binding of human erythrocytes (Figure 6). These data support our hypothesis that a carbohydrate—and not

Figure 5. Analysis of Trypsin and Pronase Degraded Human EGP in the Rosetting

Assay.

120

A Coomassie Western Blot: 108 α-hGA 100 100 185kDa

98 80

nding 60 i

49 % B 40 38

28 20 10 17 4 2 0 Sham trypsin Trypsin digest Trypsin control human rbc pig rbc 12 34 5 digest 200U (no hGP)

120 B Coomassie Western Blot: α-hGA 102 100 185kDa 100

98 80

49 binding % 38 40 28 17 20 20

8 12 34 5 4

0 Sham pronase Pronase digest Pronase control human rbc pig rbc digest (no hGP)

Human erythrocyte glycoproteins (hGA) were treated with trypsin (A) or pronase (B) (each with sham controls). Treated glycoproteins were subjected to SDS-PAGE and transferred to PVDF membranes. The samples for section A are represented in the following order for SDS-PAGE and Western blotting: sham-treated hGA (1), trypsin- treated hGA (2), human erythrocyte fragments (3), sham-treated hGA (4), trypsin-treated hGA (5). The samples for section B are represented in the following order for SDS-PAGE and western blotting: sham-treated hGA (1), pronase-treated hGA (2), human erythrocyte fragments (3), sham-treated hGA (4), pronase-treated hGA (5),Gels were stained blue for

protein with GelCode blue (Pierce). The western blots were probed with anti-glycophorin A monoclonal antibodies and detected with secondary antibody conjugated to HRP. Treated and untreated glycoprotein as well as heat-inactivated trypsin and pronase control samples were incubated with porcine Kupffer cells in culture for 15 min prior to the addition of 1% human erythrocytes labeled with Cr51 in media. Human erythrocytes and porcine Kupffer cells were incubated in co-culture for 2 h followed by gentle washing to remove non-adherent erythrocytes. Rosetting was verified by light microscopy and quantitated by gamma-detection. Each graph represents the sum of three experiments performed in triplicate.

Figure 6. Analysis of Deglycosylated Human EGP in the Rosetting Assay.

140 Western Blot: Shiff’s base

α d -hGA staining d e e t t a a

P 104 P c c

120 yl yl G b G b s s E r E r o o e e h h yc yc l m l m hol hol g g e

e 100 W W hEGP hEGP D 100 Sha D 185kDa Sha

98kDA

g 80 62kDa Indin

B 49kDa 60 % 38kDa

40 28kDa 17kDa

20 8 9 3

0 hGP Sham deglycosylation deglycosylation Human rbc Porcine rbc

Human erythrocyte glycoproteins (hGA) were treated with a cocktail of deglycosylating enzymes (PNGaseF, sialidase A, beta 1-4 galactosidase, β N- acetylhexosaminidase, hexosaminidase and endo-O-glycosidase (Prozyme) ) or sham controls. Treated glycoproteins were subjected to SDS-PAGE and transferred to a PVDF membrane. Gels were stained magenta for carbohydrate with PAS (Pierce). The Western blot was probed with anti-glycophorin A monoclonal antibodies and detected with secondary antibody conjugated to HRP (inset). Treated and untreated glycoprotein samples were incubated with porcine Kupffer cells in culture for 15 min prior to the addition of 0.1% human erythrocytes labeled with Cr51 in media. Human erythrocytes and porcine Kupffer cells were incubated in co-culture for 2 h followed by gentle washing to remove non-adherent erythrocytes. Binding was verified by light microscopy and quantitated by gamma-detection. Each graph represents the sum of three experiments performed in triplicate.

a protein epitope—is the target antigen on human erythrocytes recognized by porcine

Kupffer cells.

EDTA/EGTA Inhibition

The divalent cation chelators, EGTA or EDTA (25mM) inhibited the rosetting assay to 36% and 39% binding, respectively. These findings suggest that porcine Kupffer cell recognition of human erythrocytes is calcium dependent and may involve a calcium dependent lectin.

Discussion

We have isolated a subset of human erythrocyte glycoproteins that appears to be composed primarily of glycophorin A; these glycoproteins have been shown to inhibit the binding of human erythrocytes to porcine Kupffer cells. The identification of such an inhibitory glycoprotein provided the opportunity to evaluate whether it was the carbohydrate or protein portion of the molecule that was responsible for its inhibitory potential. Disruption of the protein portion of these glycoproteins had no effect on inhibitory activity, while cleaving the carbohydrate moieties completely disrupted the ability of the glycoproteins to inhibit the binding of human erythrocytes to porcine Kupffer cells. The binding of porcine Kupffer cells to human erythrocytes was inhibited by calcium chelators suggesting that calcium-dependent lectin receptors may be involved in

the recognition process. Thus, porcine Kupffer cells appear to bind human erythrocytes by way of lectin receptors that recognize a xenogeneic carbohydrate.

Lectins are at least as abundant in mammals as in plant tissues; lectins can be found in all kingdoms of life ranging from viruses through bacteria, sponges to snails and plants to animals(Vasta et al., 1994; Loris 2002). Lectins have been divided into the families, C-, S-, P-, I-, L- and R-type, based on the structure of their carbohydrate- recognition domain (CRD) with significant primary structural homology (other than species homologues) and other characteristics (Dodd et al., 2001; Kilpatrick 2002). For example, the C-type lectins are so named because they require Ca2+ for carbohydrate- binding activity, a characteristic consistent with the calcium-dependence of the receptor- ligand interaction presented in this report (Weis et al., 1998).

The first group to propose a role for carbohydrate-lectin interactions in the innate immune cells involved in xenograft rejection was Inverardi et al., (Inverardi et al., 1997).

These authors presented three lines of evidence to support this theory: 1) human NK cells binding to porcine endothelial cells were selectively inhibited by Melibiose (6-α-D-Galp-

D-Glc) and α-D-Mannose; 2) human natural antibody F(ab’)2 fragments (of which 1% bind to galactose α-1,3 galactose) blocked this binding; and 3) human NK cells bound only to COS cells transfected with galactose α-1,3 galactose transferase and not vector only controls. In their discussion they suggested several different NK cell surface lectin receptors (NKRP-1, NKG-2, and CD94) as possible candidates for mediating these effector functions (Inverardi et al., 1997).

Other investigators have proposed a role for lectin receptors in the recognition of xenogeneic epitopes, though most have been in the context of NK cells (Hancock 1997).

Hancock et al., have suggested that selective targeting of macrophage lectin with monoclonal antibodies, peptide inhibitors or carbohydrate inhibitors may provide a new and specific approach to blocking macrophage infiltration, activation and subsequent cytokine production during the development of delayed xenograft rejection (Hancock

1997). Studies from Itescu’s laboratory have demonstrated that both primate and human monocytes interact with galactose α-1,3 galactose on the surface of porcine endothelial cells (Artrip et al., 1999). In addition, this group found that the transgenic expression of α-

1,2 fucosyltransferase in porcine endothelial cells reduced both galactose α-1,3 galactose cell surface expression and the susceptibility of such endothelial cells to human NK cell- mediated lysis and human monocyte adhesion (Artrip et al., 1999). Like Itescu and

Inverardi, Miyagawa et al., suggested a role for galactose α-1,3 galactose-dependent adhesion of human NK cells to porcine endothelial cells (Miyagawa et al., 1999). In contrast, Sheikh et al., warned that galactose α-1,3 galactose may not be the only carbohydrate recognized (Sheikh et al., 2000). It is necessary to point out that no galactose

α-1,3 galactose binding receptor has yet been identified on human innate immune cells.

The predominant component of the human EGP was glycophorin A(hGA), the second most abundant protein on human erythrocytes with approximately 1 million molecules per erythrocyte (Dahr et al., 1986; Gardner et al., 1989). Because hGA was the predominant glycoprotein isolated and immunohistochemistry confirmed that porcine

Kupffer cells specifically bound hGA, it seems likely that the target carbohydrate ligand is borne on hGA. Human GA bears a single asparagine within its extracellular domain, which gives rise to a complex bi-antennary N-linked oligosaccharide characterized by a 1

α-fucose residue linked at the C-6 position of the proximal N-acetylglucosamine residue, 1 terminal β-N-acetylglucosamine residue at the C-4 position of the α-mannosyl residue of the core, and terminated with α2-6 linked N-acetyl neuraminic acid (Schenkel-Brunner

2000). This unique glycoprotein also bears 15 O-linked oligosaccharides that have been found as tri-, tetra-, penta-, and hexasaccharides varying in composition and linkage within amino acids 1-50 (Schenkel-Brunner 2000). Interestingly, destruction of the protein portion of hGA had no effect on its inhibitory ability (Figure 5), yet destruction of the carbohydrate domains described above destroyed the inhibitory activity of hGA (Figure 6).

Compositional analysis of the saccharides found on hGA revealed that 55.5% were neutral sugars (glucose, galactose, mannose, and L-fucose), 27.8% were hexosamines (N- acetylglucosamine and N-acetylgalactosamine), and 16.7% N-acetylneuraminic acid

(Rosenberg et al., 1968). The most structurally prominent oligosaccharide on hGA is

Neu5Ac α2-3 or α2-6 Gal β1-3 GalNAc and these same three monosaccharides also occur at the highest frequency (Schenkel-Brunner 2000). The composition of carbohydrates involved in the glycosylation of glycophorin A will serve as an important guide in continuing efforts to elucidate the exact target recognized by porcine Kupffer cells.

It is of interest to note earlier work that investigated the macrophage lectins recognizing these sugars. Previous authors exploring the binding of human and sheep

erythrocytes by rat Kupffer cells revealed that terminal GalNAc and Gal saccharides were

the epitopes being recognized (Mohr et al., 1987). It was later revealed that a lectin, the

GalNAc particle receptor, exists on rat Kupffer cells and specifically recognizes GalNAc

and Gal saccharides (Kolb-Bachofen 1991). Sialoadhesin has been characterized as a non-

phagocytic, sialic acid-dependent sheep erythrocyte receptor (Crocker et al., 1988).

Similar to our observations that porcine Kupffer cells but not porcine peripheral blood

monocytes bind human erythrocytes (unpublished observations), sialoadhesin was found to

be expressed on resident macrophages but not on peripheral blood monocytes (Crocker et

al., 1988). Both of these receptors, the GalNAc particle receptor and sialoadhesin, are

likely candidates for the receptor responsible for recognition of the dominant saccharides

present on glycophorin A.

We have shown here that porcine Kupffer cells recognize human erythrocytes

during in vitro co-culture via a calcium-dependent and carbohydrate-dependent

mechanism. The oligosaccharide epitope recognized, however, remains to be revealed.

Future efforts to expose the composition and linkage of this epitope may lead to a molecule

capable of binding the porcine Kupffer cell lectin and inhibiting erythrocyte clearance

during extracorporeal porcine liver perfusion. The porcine model described is not only a

means to analyze the binding of porcine Kupffer cells to human erythrocytes, but also may reveal a mechanism of macrophage recognition of xenogeneic epitopes in general.

MANUSCRIPT 2

Terminal Sialic Acid Residues on Human Glycophorin A are Recognized

by Porcine Kupffer Cells

Abstract

BACKGROUND: We have previously shown that recognition of human

erythrocytes by porcine Kupffer cells is mediated by a carbohydrate dependent mechanism

and the ligand recognized on the human erythrocyte is borne on the glycophorin A

glycoprotein. We have continued this work by exploring the possible ligands that exist on

human glycophorin A and testing their ability to inhibit erythrocyte rosette formation.

METHODS: Human erythrocytes were tested for both ABO and MN specificity

and used as targets in a 51Chromium quantitative erythrocyte rosette assay.

Monosaccharides known to be borne on human glycophorin A (hGA) or not present on hGA were tested for their ability to inhibit the rosetting assay, in addition to neuraminyl lactoses, bovine and porcine submaxillary mucins (BSM and PSM, respectively) and hyaluronic acid (HA). PNGaseF or sialidase A were used to cleave carbohydrates from hGA that were then subjected to SDS-PAGE and tested for inhibitory potential in the rosetting assay. Human erythrocytes were also treated with sialidase and tested in the rosetting assay.

RESULTS: Porcine Kupffer cell recognition of human erythrocytes was insensitive to differences in blood groups A, B, O, or MN. However, human erythrocyte recognition was disrupted by the monosaccharide, N-acetylneuraminic acid at 30 mM and the trisaccharide mixture, neuraminyl lactoses at 30mM by 25% and 30%, respectively. A dilution of BSM but not PSM inhibited the rosetting assay by 17% (.2mg/ml), 33%

(1mg/ml), and 53% (2mg/ml). The same dilution of HA had no effect on rosetting. Human glycophorin A was treated with PNGaseF to cleave N-linked oligosaccharides and cleavage was confirmed by increased mobility when analyzed by SDS-PAGE. Cleavage of the N-linked oligosaccharides of hGA did not impair its ability to inhibit the rosetting assay when compared with sham-treated hGA. In contrast, sialidase A cleavage of sialic acid from hGA completely abrogated its ability to inhibit the rosetting assay. Treatment of whole human erythrocytes with sialidase A likewise prevented recognition by porcine

Kupffer cells.

CONCLUSIONS: We conclude that terminal sialic acid on oligosaccharides of

hGA is a target recognized by porcine Kupffer cells. This evidence suggests a potential

role for a sialic acid receptor in innate cellular recognition of xenogeneic epitopes.

Introduction

We have previously demonstrated that porcine Kupffer cells bind human erythrocytes and that this binding can be inhibited by glycoproteins purified from the surface of human erythrocytes (Burlak II 2004a). Because the majority of these isolated glycoproteins were identified as human glycophorin A (hGA), the present work examines whether a carbohydrate present on hGA is the epitope recognized by porcine Kupffer cells.

Glycophorin A is an intrinsic membrane protein with a molecular mass of 37kD that is the second most prevalent protein expressed on human erythrocytes with approximately 1 million molecules per erythrocyte (Dahr et al., 1986; Gardner et al., 1989). Approximately

67% of all of the sialic acid on the human erythrocyte is borne on hGA. This is not surprising since hGA is roughly 60% carbohydrate and represents 1.6% of the total erythrocyte protein mass (Bretscher 1971; Dahr et al., 1976). Human glycophorin A is composed of 131 amino acids with 15 O-linked oligosaccharides attached to either serine or threonine within positions 1-50 and 1 N-linked oligosaccharide on the asparagine of position 26 (Tomita et al., 1975; Dahr et al., 1976; Tomita et al., 1978; Pisano et al.,

1993).

O-linked oligosaccharides on hGA occur in many forms and are typically either tri-

, tetra-, penta-, or hexasaccharides varying in composition and linkage (Thomas et al.,

1969; Lisowska et al., 1980; Adamany et al., 1983; Fukuda et al., 1987). The O-linked oligosaccharides have been found to bear blood group antigens such as the ABO, MN, and

Cad types (Dahr et al., 1976; Takasaki et al., 1978; Blanchard et al., 1983). With the

exception of the MN blood group that occurs on peptide positions 2-4, the ABO and Cad blood groups and other saccharides bound to either serine or threonine appear to vary in position. The N-linked oligosaccharide is a bi-antennary oligosaccharide with a branched mannose core. The structure is unique; however, it has only one terminal β-N- acetylglucosamine residue at the C-4 position of the α-mannosyl residue of the core portion and one α-fucose residue linked at the C-6 position of the proximal N- acetylglucosamine (GlcNAc) residue (Yoshima et al., 1980). Interestingly, structural features such as blood group antigens and substitutions are likely targets in our search for a macrophage lectin-binding epitope.

Different functions have been ascribed to the varied oligosaccharides found on the surface of erythrocytes. Rat Kupffer cells have been shown to bind aging rat erythrocytes through lectin recognition of a terminal galactose (Vaysse et al., 1986). This syngeneic binding occurs as sialic acid is cleaved from the surface of aging erythrocytes and galactose molecules become exposed. Rat Kupffer cells recognize the aged rat erythrocytes through the galactose particle receptor (Schlepper-Schafer et al., 1988). Using xenogeneic erythrocytes, rat Kupffer cells have been shown to recognize both sheep and human erythrocytes independent of antibody and complement opsonization (Mohr et al.,

1987). Human erythrocyte binding to rat Kupffer cells was inhibited by N- acetylgalactosamine (GalNAc) and lactose (Lac) suggesting that the GalNAc / galactose

(Gal) particle receptor was responsible. Sheep erythrocytes were best inhibited by a combination of heat aggregated IgG, gangliosides, and fucose (Fuc). However,

neuraminidase-treated sheep erythrocytes could be completely inhibited from binding when treated with saccharides specific for the GalNAc / Gal particle receptor. This evidence has set a precedent for the involvement of carbohydrates in xenogeneic recognition.

We have previously shown that human erythrocytes are recognized by porcine

Kupffer cells during a xenoperfusion event (Rees et al., 2002b). Porcine Kupffer cells recognized the human erythrocytes independent of antibody and complement (Rees 2003c) and via a carbohydrate- dependent mechanism (Burlak II 2004a). This study explores the nature of the carbohydrate recognition of hGA by porcine Kupffer cells. We provide evidence that while the N-linked oligosaccharide, human ABO blood groups, hyaluronic acid and various monosaccharides are unrelated to binding, a terminal sialic acid of possibly either α2-3 or α2-6 linkage is responsible for porcine Kupffer cell recognition of human erythrocytes.

Materials and Methods

Isolation of Porcine Kupffer Cells

Porcine Kupffer cells were isolated as reported by Rees (2003) (Rees 2003c).

Rosetting Assay and Isolation of hGA

Rosetting assay was performed in accordance with the established protocol (Burlak

II 2004a).

Glycophorin A Isolation

Isolation of GA was performed according to the established protocol (Burlak II

2004a).

ABO Blood Group Specificity

Human blood was drawn into lithium heparin tubes and tested for blood type by agglutination with monoclonal antibodies for a specific blood type (Gamma

Biological, Houston, TX). Blood typed samples were centrifuged and the plasma and buffy coat were removed. The erythrocytes of each blood type were labeled separately with 51Chromium (Morrissey et al., 1992). 51Chromium-labeled human erythrocytes were used in the rosetting assay as we have previously described Burlak (2004) (Burlak II

2004a). Human blood type AB was excluded because it contains blood type A and blood type B oligosaccharides as well as the H antigen.

Monosaccharides in the Rosetting Assay

N-acetylneuraminic acid (Neu5Ac), Gal, Fuc, GalNAc, GlcNAc, mannose (Man) and arabinose (Ara) were used at the described concentrations. Each monosaccharide was solubilized in with CMRL media (CMRL – 1066 medium, (Invitrogen, Grand Island, NY) with 100 U/ml of Penicillin – Streptomycin (Invitrogen, Grand Island, NY) and 2 mM L-

Glutamine (Invitrogen, Grand Island, NY)) and the pH was adjusted to pH 7.0 as necessary. The monosaccharide solutions were incubated with porcine Kupffer cells for 15 min prior to the addition of 51Chromium-labeled human erythrocytes. The monosaccharides at each concentration were tested in the rosetting assay three times in triplicate.

Neuraminyl Lactoses in the Rosetting Assay

Neuraminyl lactoses (V-Labs Inc., Covington, LA) were solubilized in CMRL media. Neuraminyl lactoses were tested in the rosetting assay three times in triplicate.

Enzymatic Deglycosylation by PNGaseF or Sialidase A

Sialidase A and PNGaseF (Prozyme Inc., San Leandro, CA) were used according to the manufacturer’s instructions. Either 100 units of PNGaseF or 0.1 units of sialidase A were incubated with 2mg of hGA for 4 d. A “sham” treated hGA sample, without enzyme, was incubated under the same conditions. Enzyme treated, sham, and untreated hGA samples were analyzed by SDS-PAGE to verify a change in migration derived from

removal of N-linked oligosaccarhides or desialation. Removal of sialic acid from human

erythrocytes was confirmed by measuring released sialic acid (data not shown).

Hyaluronic Acid in the Rosetting Assay

Hyaluronic acid (Sigma Inc., St. Louis, MO) at concentrations of 0.2, 1.0, and 2.0

mg/ml was prepared in CMRL media and incubated with porcine Kupffer cells for 15

minutes prior to the addition of 51Chromium-labeled human erythrocytes in the rosetting assay.

Salivary mucins in the rosetting assay

Porcine submaxillary mucin (PSM) was isolated from porcine submaxillary glands freshly excised and immediately chilled. The fat and connective tissue were removed, the gland was cut into small pieces, and macerated in a Waring blender with 3 volumes of water. After centrifugation the supernatant was added to 3 volumes of 95% ethanol with

stirring. Flocculation was facilitated by adding a few drops of a saturated solution of

potassium acetate until a visible precipitate forms. The precipitate was collected by

centrifugation and allowed to dry after washing with ice cold acetone (Aminoff et al.,

1970). The sialic acid content was determined by both bovine submaxillary mucin (BSM)

and PSM using the orcinol method and detected by absorbance at 572nm. Bovine

submaxillary mucin was obtained from Sigma Inc. (St. Louis, MO). Mucins were

incubated at the concentrations of 0.2, 1.0, and 2.0 mg/ml with porcine Kupffer cells for 15 min prior to the addition of chromium51 labeled human erythrocytes.

PNGaseF tTreated hGA

PNGaseF (100 units)(Prozyme Inc. San Leandro, CA) was incubated with 2mg of hGA for 4 d in accordance with the manufacturer’s instructions. A “sham” treated hGA sample, without enzyme, was incubated under the same conditions. Enzyme treated, sham, and untreated hGA samples were analyzed by SDS-PAGE to verify a change in migration derived from removal of N-linked oligosaccarhides.

Results

ABO blood group specificity

Having previously established that carbohydrate epitopes can be responsible for

Kupffer cell binding of human erythrocytes, we wanted to explore the likelihood of blood

group oligosaccharides as potential xenoantigens. 51Chromium-labeled human erythrocytes of blood types A, B, and O were used in a quantitative rosetting assay that demonstrated no significant difference in the recognition of erythrocytes bearing the blood types A and B, and O.

Monosaccharides in the Rosetting Assay

Monosaccharides used in the rosetting assay were chosen based on the known glycosylation pattern of hGA so that carbohydrates both present and absent were evaluated. Neu5Ac, Gal, Fuc, GalNAc, GlcNAc, and Man are monosaccharides known to occur in N- and O-linked oligosaccharides of hGA, while Ara has not been found. Gal,

Fuc, GalNAc, GlcNAc, Man, and Ara at concentrations up to 240 mM had no effect on rosetting (Figure 1). Neu5Ac, however, revealed dose-dependent inhibition beginning at 5 mM and reaching 75% inhibition at 240 mM (Figure 1).

Neuraminyl Lactoses in the Rosetting Assay

Having revealed that Neu5Ac significantly inhibited the rosetting assay, we wished to explore the effects of Neu5Ac linked to a disaccharide substrate as it is found on human

glycophorin A. Neuraminyl lactoses prepared from bovine milk are approximately 80%

Neu5Ac linked α-2,3 to Lac and 20% Neu5Ac linked α-2,6 to Lac with no more than 5% of the total composition being unbound Lac. From 0 mM to 30 mM, neuraminyl lactoses showed dose-dependent inhibition reaching a maximum of 30% inhibition at 30 mM

(Figure 2).

Hyaluronic Acid in the Rosetting Assay

It has been suggested that the effect of sialic acid on binding is due to the large overall negative charge of the Neu5Ac (Varki 1997; Wasylnka et al., 2001). Hyaluronic acid is composed of a peptide backbone that contains numerous substituted uronic acids.

Concentrations of 0.2, 1.0, and 2.0 mg/ml hyaluronic acid were unable to inhibit the recognition of human erythrocytes by porcine Kupffer cells (Figure 3).

Sialidase A Treated hGA and Human Erythrocytes

To address the role of Neu5Ac as a terminating saccharide of human glycophorin

A, we used the enzyme sialidase A to cleave sialic acid residues. Sialidase A-treated hGA analyzed by SDS-PAGE showed an increased mobility when compared to sham-

Figure 1. Monosaccharide Inhibition of Human Erythrocyte Binding to Porcine

Kupffer Cells.

140

120

100 GalNAc GlcNAc Gal Fuc 80 Man Ara NANA

% Binding 60

0 0mM 1mM 5mM 15mM 30mM 60mM 120mM 240mM pig rbc

Monosaccharides were prepared in CMRL media and the pH of each solution was adjusted to pH 7.0 if necessary. Each monosaccharide was incubated with porcine Kupffer cells for 15 min prior to incubation with 1% human erythrocytes labeled with 51Cr in media. Rosetting was verified by light microscopy and quantitated by gamma-detection. Binding is expressed as a percentage of the binding of human erythrocytes to porcine Kupffer cells in the absence of monosaccharides. Each bar graph represents the average of three experiments performed in triplicate. The error bars express standard error.

Figure 2. Neuraminyl Lactose Inhibition of Human Erythrocyte Binding to Porcine

Kupffer Cells.

120

100 101 100

80 70

60 % Binding

0 0mM 10mM 15mM 30mM pig rbc

Neuraminyl lactoses were solubilized in CMRL media at the concentrations shown along the x-axis. Porcine Kupffer cells were incubated with each concentration of neuraminyl lactoses for 15 min prior to the addition of a 1% solution of 51Cr-labeled human erythrocytes. Rosetting was verified by light microscopy and quantitated by gamma-detection. Samples were tested three times in triplicate. The error bars express standard error. Pig erythrocyte binding in the absence of neuraminyl lactoses served as a negative control.

Figure 3. Hyaluronic Acid Inhibition of Human Erythrocyte Binding to Porcine

Kupffer Cells.

120

99 100 95 100 100

80 ng

ndi 60 % Bi

0 .2mg/ml 1mg/ml 2mg/ml hrbc prbc

Hyaluronic acid isolated from cock’s comb was solubilized in CMRL media at the concentrations shown along the x-axis. The samples were incubated with the Kupffer cells for 15 min prior to the addition of a 1% solution of 51Cr-labeled human erythrocytes. Rosetting was verified by light microscopy and quantitated by gamma-detection. Samples were tested three times in triplicate. The error bars express standard error.

treated hGA (data not shown). Sialidase treatment was found to completely remove the

ability of hGA to inhibit rosetting when it was compared to the ability of sham-treated

hGA to inhibit rosetting (Figure 4a). Similarly, sialidase A-treated human erythrocytes

were not bound by porcine Kupffer cells as compared to untreated human erythrocytes

(Figure 4b). Sialidase A-treated intact human erythrocytes were incubated with the sialic

acid α−2,3-specific lectin Mackia ameurensis (1mg/ml) for 2 h and evaluated for agglutination. Sialidase A- treated cells did not agglutinate while untreated erythrocytes agglutinated, suggesting that treatment with sialidase A removed sialic acid from the surface of intact human erythrocytes (data not shown).

Salivary Mucins in the Rosetting Assay

Within the saliva of mammals are glycoproteins secreted from the submaxillary glands that have been found to bear sialic acids of varying substitutions depending on the species of origin. PSM which are heavily substituted with Neu5Gc in contrast to BSM which bear up to 12% Neu5Ac, were tested in the rosetting assay. At concentrations of

0.2, 1.0, and 2.0 mg/ml PSM had no effect on rosetting, while BSM revealed maximal inhibition of 53% at 2.0 mg/ml (Figure 5).

PNGaseF Treated hGA

Removal of the single N-linked oligosaccharide on each hGA glycoprotein had no effect on the ability of porcine Kupffer cells to bind human erythrocytes. When analyzed

by SDS-PAGE, PNGaseF treated hGA migrated at approximately 55kDA, while sham treated hGA migrated at 62kDa as expected.

Figure 4. Evaluation of Desialated Human EGP to Act as an Inhibitor of Human

Erythrocyte Binding to Porcine Kupffer Cells (A) and of Desialated Human

Erythrocytes to be Bound by Porcine Kupffer Cells (B).

120 A 103 100 100

80 g n i 60 nd bi %

5 5 5

0 hGP Sham desialated Desialated Human rbc Porcine rbc

120 B 100 100 100

80 g n

60 ndi bi %

20 6 5

0 Sham desialated Desialated human Human rbc Porcine rbc human rbc rbc

A. Untreated EGP, Sham-treated EGP, and sialidase A-treated EGP were each solubilized in CMRL media and incubated with porcine Kupffer cells for 15 min prior to the addition of 51Cr-labeled human erythrocytes. B. Sham-treated and Sialadase A-treated human erythrocytes were incubated with porcine Kupffer cells for 2 h. Rosetting was verified by light microscopy and quantitated by gamma-detection. Samples were tested three times in triplicate. The error bars express standard error.

Figure 5. Bovine and Porcine Submaxillary Mucin Inhibition of Human Erythrocyte

Binding to Porcine Kupffer Cells.

120

101 96 100 100 95

66 PSM BSM ng

60 ndi i B

% 47

0 .2mg/ml 1mg/ml 2mg/ml Human rbc Pig rbc

Bovine submaxillary mucin (BSM) and porcine submaxillary mucin (PSM) were solubilized in CMRL media and incubated with porcine Kupffer cells for 15 min prior to the addition of a 1% solution of 51Cr-labeled human erythrocytes. Rosetting was verified by light microscopy and quantitated by gamma-detection. Samples were tested three times in triplicate. The error bars express standard error.

Figure 6. Evaluation of PNGaseF-Treated Human EGP to Inhibit Human

Erythrocyte Binding by Porcine Kupffer Cells.

120

100 100

80 ng

60 ndi Bi %

20 10 5 4

0 sham hGA PNGaseF treated human rbc pig rbc hGA

Sham-treated and PNGaseF-treated EGP were solubilized in CMRL media and incubated with porcine Kupffer cells for 15 min prior to the addition of a 1% solution of 51Cr-labeled human erythrocytes. Rosetting was verified by light microscopy and quantitated by gamma-detection. Samples were tested three times in triplicate. The error bars express standard error.

Discussion

We have previously demonstrated that a carbohydrate moiety is recognized on the surface of human erythrocytes by porcine Kupffer cells during in vitro co-culture (Burlak

II 2004a). Given this finding, we next wanted to explore the possible carbohydrate epitopes on human glycophorin A and assess their importance in the rosetting assay.

It is well known that ABO blood groups are immunologically recognized between individuals within a species by circulating antibodies. It has been theorized that the reason we have these antibodies directed towards other blood groups is that the offending blood group epitope may also exist on intestinal flora to which we are continuously exposed

(Galili et al., 1987). We tested A, B and O type human erythrocytes for their affinity to porcine Kupffer cells in the previously described rosetting assay. A, B and O erythrocytes were bound equally by porcine Kupffer cells. This suggests that differences in the GalNAc

(type A) or Gal (type B) substitution associated with that blood group is not involved in binding. However, the Fuc α-1,2 substitution at the blood group core that comprises the H antigen, also is present on blood groups A and B. Thus, blood group O cannot be ruled out as a potential candidate for recognition by porcine Kupffer cells.

While both N- and O-linked oligosaccharides contain many of the same saccharides, some of their composition and linkage is different. The sialic acid present in

N-linked oligosaccharides occurs in an α-2,6 linkage while O-linked oligosaccharides bear two additional linkages of sialic acid, α-2,3 and α-2,8 (Yoshima et al., 1980; Fukuda et al.,

1987). This seemingly small difference has been shown to be the determining factor in recognition by many lectins. Sialic acid binding Ig-like lectins, siglecs, as their name suggests, are a diverse group of membrane and secreted proteins in humans that recognize a variety of sialic acids terminally linked either α-2,3, α-2,6, or α-2,8 to oligosaccharide chains (Angata et al., 2002). We used the monosaccharides, Gal, Fuc, GalNAc, GlcNAc,

Man and Neu5Ac that are common to both N- and O-linked oligosaccharides in various linkages on hGA, in the rosetting assay. We also included the monosaccharide Ara, which is not found on human glycophorin A, as a carbohydrate control. While linkage is an important factor, at high concentrations monosaccharides have been shown to inhibit lectin carbohydrate interactions (Schlepper-Schafer et al., 1980; Schlepper-Schafer et al., 1981;

Mohr et al., 1987). Using a battery of mono and disaccharides, rat Kupffer cells have been shown to bind aging rat erythrocytes through protein recognition of a terminal Gal. This syngeneic binding occurs as sialic acid is cleaved from the surface of aging erythrocytes and Gal molecules are exposed (Vaysse et al., 1986). Rat Kupffer cells recognize the aged rat erythrocytes through the galactose particle receptor (Schlepper-Schafer et al., 1988).

Using xenogeneic erythrocytes, rat Kupffer cells have been shown to recognize both sheep and human erythrocytes independent of antibody and complement opsonization (Mohr et al., 1987). Human erythrocyte binding to rat Kupffer cells was inhibited by GalNAc and

Lac while sheep erythrocyte binding was best inhibited by an assortment of oligo- and monosaccharides (Mohr et al., 1987). The monosaccharides, Gal, Fuc, Ara, GalNAc,

GlcNAc, and Man, showed no ability to inhibit the rosetting assay. Neu5Ac, however,

inhibited the rosetting assay in a dose dependent manner (Figure 1). This result was not surprising considering the prominence of sialic acid as a terminal saccharide on human erythrocytes. Of the 40 acidic 9-carbon alpha-keto sugars, known as sialic acids, all are enzyme-catalyzed modifications of the most abundant form, Neu5Ac (Siebert et al., 1997).

It has been suggested that some or all of the affinity of sialic acid to a ligand is based on the saccharide’s overall negative charge, which is increased when in high concentration such as is found on human glycophorin A. Therefore, we addressed the possibility that human erythrocytes were being recognized based on their overall negative charge and not via a specific carbohydrate lectin interaction. We tested a dilution of hyaluronic acid, a negatively charged glycoprotein, in the rosetting assay. Unlike Neu5Ac, hylauronic acid had no effect on the ability of porcine Kupffer cells to bind human erythrocytes (Figure 3). This evidence suggests that porcine Kupffer cells specifically recognize terminal Neu5Ac on human erythrocytes in spite of their overall negative charge.

Thus far, we have examined the role of terminal sialic acid present on both N- and

O-linked oligosaccharides of human glycophorin A and total sialic acid from the surface of intact erythrocytes. Using the enzyme sialidase A, we removed sialic acid from both hGA and human erythrocytes. The sham-treated human erythrocyte glycophorin A inhibited the ability of porcine Kupffer cells to bind human erythrocytes. However, the sialidase A- treated hGA was no longer able to inhibit the rosetting assay (Figure 4). Similarly, sham- treated intact human erythrocytes were recognized by porcine erythrocytes as we have seen

previously. The sialidase A-treated intact human erythrocytes went completely unrecognized by porcine Kupffer cells in the rosetting assay. Taken together, these data further emphasize the importance of sialic acid in porcine Kupffer cell recognition.

To evaluate the role of sialic acid further, we also tested BSM and PSM mucin.

Through a point mutation, humans have lost the ability to produce the sialic acid called N- glycolylneuraminic acid (Neu5Gc) (Varki 2001). Neu5Gc differs from Neu5Ac in that the acetyl group on the 5th carbon of Neu5Ac has been additionally substituted with a hydroxyl group. This small difference has given rise to a significant immunological barrier between species. Human colon carcinoma cells have recently been shown to be immunologically recognized based on their expression of Neu5Gc (Malykh et al., 2001). Pigs, on the other hand, express both Neu5Ac and Neu5Gc in amounts specific to the tissue analyzed and stage of development (Hueso et al., 1985; Sato et al., 1998; Malykh et al., 2003).

The saliva of mammals is rich in glycoproteins called mucins that are heavily glycosylated. We isolated the submaxillary glands from an adult pig and extracted the mucins that have been shown to be heavily sialated with Neu5Gc and to a far lesser extent

Neu5Ac (Aminoff et al., 1970). We tested the isolated PSM and showed it to be approximately 16% sialic acid and 47% protein. We also obtained bovine submaxillary mucin that was composed of 12% sialic acid and 42% protein. While PSM had no effect on the ability of porcine Kupffer cells to bind human erythrocytes, BSM inhibited binding by 53% at 2mg/ml (Figure 5). This suggests that the difference between Neu5Ac and

Neu5Gc may explain the recognition of human erythrocytes by porcine Kupffer cells.

These data are supported by the finding that while human glycophorin A bears the majority of Neu5Ac found on the human erythrocyte, pig glycophorin A is predominantly sialated with Neu5Gc yet scantily sialated with Neu5Ac (Hueso et al., 1985; Sato et al., 1998;

Malykh et al., 2003).

The composition and structure of the single N-linked oligosaccharide on hGA has been characterized (Yoshima et al., 1980). The N-linked oligosaccharide is a bi-antennary complex oligosaccharide with a mannose core upon which are substituted lactose repeats terminated exclusively with α-2,6 linked Neu5Ac residues. Because the N-linked oligosaccharide on human glycophorin A has multiple epitopes and occurs only once, we chose to use the enzyme PNGaseF, which specifically removes N-linked oligosaccharides and leaves O-linked oligosaccharides unchanged, to treat human hGA. PNGaseF or sham treatment of hGA had little effect on the ability of hGA to inhibit the rosetting assay

(Figure 6). Both the PNGaseF- and sham- treated hGA were able to inhibit binding in the rosetting assay. This suggests that by removing the N-linked oligosaccharide, the epitope recognized by Kupffer cells was either unaffected or not diminished by an effective degree.

While we present evidence that sialic acid is responsible for porcine Kupffer cell recognition, the Neu5Ac linked α-2,6 to Lac on the N-linked oligosaccharide is either at an insignificant concentration to the total Neu5Ac on human glycophorin A or is not the correct linkage for recognition. Thus, the data presented herein raises the question of linkage as a decisive factor in recognition of Neu5Ac.

We have previously shown that opsonization by antibody or complement is not necessary for porcine Kupffer cells to bind human erythrocytes during in vitro co-culture

(Rees 2003c). It has been demonstrated that binding involves the direct recognition of a

target on human erythrocytes by a receptor on the surface of porcine Kupffer cells. The

latest evidence presented here suggests that the receptor necessary for recognizing human

erythrocytes recognizes sialic acid. Sialic acid binding Ig-like lectins, siglecs, which are

membrane bound proteins, contain two common structural motifs. The first structural

motif is a V-set Ig-like domain which is directly responsible for carbohydrate recognition

(Crocker et al., 2001b). The second domain has a varying number of C-2 set Ig-like

domains located at the N-terminus (May et al., 1998). The first siglec to be discovered was

sialoadhesin or siglec-1 (Crocker et al., 1986). While exploring the properties of murine

stromal tissue macrophages, Crocker and Gordon revealed that a lectin-like molecule on

the surface had hemagglutinating activity. Later studies showed the specificity of

sialoadhesin for sialic acid in vitro through the recognition of glycoconjugates (Crocker et

al., 1991). There are five reasons that porcine sialoadhesin appears to be a likely candidate

for human erythrocyte recognition. First, sialoadhesin has been shown to recognize

xenogeneic erythrocytes (Crocker et al., 1986). Second, sialoadhesin has been shown to bind but not phagocytose the erythrocytes, which is consistent with our observations

(Crocker et al., 1986). Third, sialoadhesin is differentially expressed on only a subpopulation of macrophages, which includes Kupffer cells (Angata et al., 2002). Fourth, the expression of sialoadhesin was shown to increase when cells were kept in culture

(Angata et al., 2002). This is consistent with our own observation that Kupffer cells rosette human erythrocytes most efficiently after 1 wk of incubation (unpublished observation). Fifth, sialoadhesin was not found to be expressed by peripheral blood monocytes (Hartnell et al., 2001). We have preliminary data suggesting that porcine peripheral blood monocytes do not recognize human erythrocytes. Yet porcine Kupffer cells readily bind human erythrocytes as we have shown (Rees 2003c).

Because of the relative prominence of sialic acids as terminal carbohydrates and their large structural variety, it seems obvious that they are involved in mediating and modulating various intercellular interactions (Kelm et al., 1997). Siglec-3, also known as

CD33 has been suggested to function as a negative signaling receptor by its ITIM-like structural motif on the cytoplasmic region (Crocker et al., 2001b). It has been proposed that when siglec-3 binds sialic acid, it sends a signal to the cell that it has bound self

(Ravetch et al., 2000). Together with the data revealing sialic acid differences between humans and pigs, it is possible that the absence of appropriate sialic acid-bearing ligands identifies a pathogen, thus distinguishing self from non-self (Muchmore et al., 1998; Varki

2001). It has been previously suggested that the lack of appropriate sialic acid epitopes may prevent inhibitory signaling via siglec receptors and lead to immune activation followed by xenograft rejection (Dorling et al., 2000; Ravetch et al., 2000; Forte et al.,

2001; Sharland et al., 2002).

In spite of the evidence presented above, that Neu5Ac is an inhibitor of porcine

Kupffer cell recognition of human erythrocytes, we still do not know the specific series of

saccharides and their linkage that compose the pertinent binding domain. Better

understanding of the specific inhibitory oligosaccharide will facilitate the development of

inhibitory molecules that could be used to prevent the loss of human erythrocytes during

extracorporeal porcine liver perfusion.

We have provided evidence that, of the various monosaccharides tested, sialic acid

inhibited porcine Kupffer cell binding of human erythrocytes. Neuraminyl lactoses, in a

mixture of α-2,3 and 2,6 linkages, inhibited rosetting. It appears that the overall negative

charge of the sialated hGA is not the characteristic responsible for binding based on the inability of hylauronic acid to inhibit the rosetting. Finally, when both hGA and intact human erythrocytes were treated with sialidase A, hGA lost the ability to inhibit the rosetting assay and treated human erythrocytes were no longer bound by porcine Kupffer cells. Thus, we have shown that the target epitope on the surface of human erythrocytes recognized by porcine Kupffer cells is a sialic acid containing oligosaccharide.

Manuscript 3

Porcine Annexin IV: a Lectin That Mediates Xenogeneic Recognition of

Human Erythrocytes.

Abstract

INTRODUCTION: We have previously shown that resident porcine liver

macrophages, Kupffer cells, recognize terminal sialic acid of the human erythrocyte glycoprotein, glycophorin A (Burlak II 2004a). We hypothesized that the receptor responsible for sialic acid recognition is a calcium dependent lectin on the surface of porcine Kupffer cells.

METHODS: We have used CNBr/sepharose affinity chromatography to retain proteins that specifically recognize the ligand, human glycophorin A, but not porcine glycophorin A or BSA conjugates. Mass spectrometry was used to identify protein bands from fractions analyzed by SDS-PAGE. To verify the presence of the identified protein on the surface of porcine Kupffer cells, an anti-human annexin IV monoclonal antibody was used as a probe for immunohistochemistry. The anti-human annexin IV monoclonal antibody was also used to assess the role of annexin IV in binding human erythrocytes.

Our readout method was the recognition of 51Chromium-labeled red blood cells (RBC) by

primary cultures of porcine Kupffer cells.

RESULTS: Three proteins were retained by the human glycophorin A column followed by sequencing and identification using mass spectrometry: bovine serum albumin

(62kDa), annexin I/II (42kDa), and annexin IV (35kDa). Annexin IV was found to be selectively expressed on the surface of porcine Kupffer cells as indicated by color development using immunohistochemistry while Kupffer cells probed with an isotype control had no color change. Furthermore, we were able to inhibit the recognition of human erythrocytes by porcine Kupffer cells with an increasing dilution of anti-annexin IV monoclonal antibody.

CONCLUSION: We conclude that the lectin, annexin IV, either in its entirety or in part, is the receptor expressed by porcine Kupffer cells that recognizes xenogeneic erythrocytes. It is known that pigs primarily express N-glycolylneuraminic acid while humans express N-acetylneuraminic acid. We hypothesize that this evolutionary difference between humans and pigs is the origin of this interaction and reveals a potential problem for pig to human xenotransplantation. We further propose that xenogeneic lectin- carbohydrate interactions such as this are not only demonstrated during xenoperfusion but may model a heretofore overlooked method of species differentiation.

Introduction

Lectins mediate carbohydrate binding by way of highly conserved regions within

the lectin polypeptide; this region is known as the carbohydrate-recognition domain

(CRD). In general, these regions have only moderate affinity for a given sugar, but lectins

generate high avidity by expressing multiple CRDs within a single lectin or by forming

oligomers of several subunits. In this way, many CRDs that bind with moderate affinity

result in a high overall avidity. Until recently, when immunologists heard the term

“lectin,” they thought of such plant lectins as Concanavilin A which is known for its ability

to non-specifically stimulate T-cell activation or eel agglutinins used to define the

carbohydrate nature of blood group antigens (Janeway 1999; Watkins 2001; Kilpatrick

2002). It is now clear that lectins are at least as abundant in mammals as in plant tissues

and play an important role in immune system function; lectins can be found in all

kingdoms of life ranging from viruses to bacteria, sponges to snails and plants to animals

(Vasta 1991; Vasta et al., 1994, 1996; Loris 2002).

Innate immune lectins have been previously suggested to be involved in xenograft rejection (Inverardi et al., 1996; Inverardi et al., 1997). Human natural killer (NK) cells have been shown to recognize carbohydrate differences on porcine endothelium as a signal for binding and lysis (Inverardi et al., 1996; Inverardi et al., 1997). NK cells were selectively inhibited from binding porcine endothelium by melibiose (6-α-D-Galp-D-Glc)

and α-D-Mannose. This same report also revealed that human natural antibody F(ab’)2

fragments (of which 1% bind to galactose α-1,3 galactose) block this binding.

Furthermore, they showed that human NK cells bind only to COS cells transfected with

galactose α-1,3 galactose transferase and not vector only controls. The NK cell surface

lectin receptors NKRP-1, NKG-2, and CD94 are possible candidates for mediating these

effector functions (Inverardi et al., 1997).

The evidence that lectin receptors on innate immune cells may play an important role in xenograft rejection has become clearer with the advent of pigs transgenic for complement regulatory proteins. Experiments with these animals has provided a window to view xenograft rejection mechanisms that lie beyond the first few minutes or hours of hyperacute rejection. A new form of rejection has been identified that occurs 2-4 days after engraftment and is variously referred to as delayed xenograft rejection or acute

vascular rejection (Bach et al., 1996; Platt et al., 1998). While this form of graft rejection

remains less well defined, it is likely that humoral immune responses, endothelial cell

activation, platelet aggregation, disparities in coagulation regulation and innate cellular

responses play a role (Bach et al., 1996; Platt et al., 1998).

Despite the fact that the immunological mechanisms underlying delayed xenograft

rejection remain controversial, there is mounting evidence that macrophages play an

important role. Using a variety of techniques that essentially prevent the occurrence of

hyperacute rejection, many investigators have demonstrated the crucial role of

macrophages in mediating delayed xenograft rejection (Blakely et al., 1994; Fryer et al.,

1994; Satake et al., 1994; Wallgren et al., 1995; Lin et al., 1997; Fox et al., 1998; Itescu et

al., 1998; Xu et al., 1998; Xia et al., 2000). It has been recently demonstrated in the clinically relevant pig to baboon cardiac xenograft model that the principal cell type found in rejecting grafts were macrophages (Goddard et al., 2002). Thus, the controversy no longer appears to be whether or not macrophages are involved in delayed xenograft rejection. Rather, the questions now center on whether macrophages can recognize and destroy xenografts independently or if they require additional signals from other cellular or humoral sources.

Support for the concept that macrophages can directly recognize xenogeneic epitopes via lectin receptors was suggested by Kwiatkowski et al., who showed that both primate and human monocytes interact with the oligosaccharide, galactose α-1,3 galactose,

on the surface of porcine endothelial cells (Kwiatkowski et al., 2000a; Kwiatkowski et al.,

2000b). In addition, this group found that the transgenic expression of α-1,2

fucosyltransferase in porcine endothelial cells reduced both galactose α-1,3 galactose cell

surface expression and the susceptibility of such endothelial cells to human NK cell-

mediated lysis and human monocyte adhesion (Artrip et al., 1999; Kwiatkowski et al.,

2000a). An interesting report from the field of senescent erythrocyte research described

using human erythrocytes as the source of xenoantigen to be recognized by rat

macrophages (Mohr et al., 1987). Subsequently, the rat Kupffer cell GalNAc receptor was

cloned and this protein likely represents the receptor that was functionally identified as

being involved with the rat Kupffer cell recognition of xenogeneic erythrocytes (Hoyle et al., 1988). Interestingly, the GalNAc receptor was found to be expressed only on resident

liver macrophages—Kupffer cells (Haltiwanger et al., 1986; Lehrman et al., 1986;

Tiemeyer et al., 1992).

We have previously reported that when porcine livers were perfused with human

blood to examine the efficacy of a normothermic extracorporeal liver perfusion system,

hematocrit levels decreased dramatically over 72 h (Rees et al., 2002b). Our first efforts to

understand the mechanism leading to the loss of erythrocytes during liver xenoperfusion revealed that antibodies and complement did not appear to be involved by way of classical pathway complement-mediated lysis (Rees 2003c). Further research revealed that the macrophages that lined the sinusoidal passages of the porcine liver, Kupffer cells, bound and phagocytosed the human erythrocytes (Rees 2003c). In vitro studies demonstrated that porcine Kupffer cells recognized human erythrocytes directly, in the absence of opsonization (Rees 2003c). We proposed that the mechanism responsible for this direct recognition was a lectin-carbohydrate interaction (Rees 2003c). Further studies revealed that sialic acid on the human glycophorin A glycoprotein was a ligand recognized by porcine Kupffer cells (Burlak II 2004a). This recognition was not the product of negative charge or other saccharides present on human glycophorin A (Burlak II 2004a). We now present evidence that porcine annexin IV is responsible, at least in part, for the xenogeneic recognition of human erythrocytes by porcine Kupffer cells.

Materials and Methods

Isolation of Kupffer Cells and Extract

Kupffer cells were isolated as described by Burlak et al., (Burlak II 2004a).

Primary cultures of Kupffer cells were scraped from culture plates using a plastic cell

scraper (Fisher Inc. Hanover Park, IL) and washed three times in CMRL media (CMRL –

1066 medium, (Invitrogen, Grand Island, NY) with 100 U/ml of Penicillin – Streptomycin

(Invitrogen, Grand Island, NY) and 2 mM L-Glutamine (Invitrogen, Grand Island, NY) to

remove cell debris and serum. The pellet of washed cells was suspended in 2ml of triton

lysis buffer containing 0.5% Triton X-100, 100mM NaCl, 50mM CaCl, 50mM Tris, 10%

glycerol (pH 7.5) and vortexed for 5 min at 4°C. The cell lysate then was centrifuged at

1000g for 10 min to remove nuclei and cell debris. The final supernatant was aliquoted and stored at -20°C.

Production of Affinity Column

Previously isolated human and porcine EGP and BSA (Sigma Inc., St Louis, MO) were each conjugated to activated cyanogen bromide sepharose 4B following the manufacturer’s instructions (Burlak II 2004a). One hundred fractions were collected at a flow rate of approximately 1ml / 3 min. Columns (1cm id X 21cm) were equilibrated with loading buffer (triton lysis buffer) followed by elution buffer (0.5% Triton X-100, 100mM

NaCl, 25mM EDTA, 50mM Tris (pH 7.5)). Five hundred microliters of a 1mg / ml

solution of Kupffer cell lysate was added to each column and 50 fractions were collected.

Elution buffer was added and 50 fractions were collected. Fractions were tested for protein by the BCA protein determination method (Pierce Inc., Rockford, IL).

Mass Spectrometry and Sequencing of Porcine Annexin IV

Fractions from the affinity column that tested positive for protein were analyzed by

SDS-PAGE and stained for protein with GelCode Blue (Pierce Inc., Rockford, IL). Protein bands eluted from the hGA affinity column were cut from the gel and digested with sequencing-grade, modified trypsin (Promega, Madison, WI) overnight at 37°C. The peptides were extracted with 60% acetonitrile: 0.1% TFA. The extract was concentrated down to ~ 15 µl using vacufuge. Two microliters of the sample were separated on a reverse phase column (75 µm id X 5 cm X 15 µm Aquasil C18 Picofrit column, New

Objectives). The eluted (peptides) were directly introduced into an ion-trap mass spectrometer (LCQ-Deca XP Plus, Thermo Finnigan, San Jose, CA) equipped with nano- spray source. The mass spectrometer was operated on a double play mode where the instrument was set to acquire a full MS scan (400-2000 m/z) and a collision induced dissociation (CID) spectrum on the most abundant ion from the full MS scan. The CID spectra were either manually interpreted or searched against an appropriate non-redundant database using TurboSEQUEST. Mass spectroscopy was performed by the Proteomics

Core Laboratory at the Medical College of Ohio.

Anti-annexin IV Monoclonal Antibody Immunohistochemistry

Kupffer cells were isolated and prepared as described previously (Burlak II 2004a).

After adherence to 13 mm round coverglasses, Kupffer cells were blocked with BSA (1mg

/ ml) in CMRL media for 2 h prior to fixation. The isotype control, anti-human

glycophorin A mouse IgG monoclonal antibody (1:2500) (Dako, Carpinteria, CA), and the

anti-human annexin IV mouse IgG monoclonal antibody (1:2500) (BD Biosciences

Transduction Laboratories, Lexington, Kentucky) were incubated with the fixed Kupffer cells for 1 h prior to three washes with tris buffered saline pH 7 plus Tween (TBST). Anti- mouse IgG conjugated to horseradish peroxidase (HRP) (1:10,000) (Sigma Inc., St Louis,

MO), the secondary antibody, was incubated with the Kupffer cells for 1 h prior to five washes with TBST for 15 min each. Final detection and color development was performed with Vector Blue (Vector Labs, Inc., Burlingame, CA) according to the manufacturer’s instructions. As a control, fixed Kupffer cells were incubated without primary or secondary antibody and also developed with the Vector Blue kit.

Anti-annexin IV Monoclonal Antibody Inhibition of the Rosetting Assay

The rosetting assay was performed as previously described by our laboratory

(Burlak II 2004a). Porcine Kupffer cells were blocked with mouse IgG (1:1000) for 1 h and removed by pipette. The anti-human annexin IV mouse IgG monoclonal antibody or the isotype control mouse IgG antibody was immediately added to the Kupffer cells at the described concentrations for 15 min prior to the addition of human erythrocytes (Figure 4).

After 2 h of incubation with the 51Chromium-labeled human erythrocytes, the coverglasses were gently washed to remove non-adherent erythrocytes and placed in scintillation vials for gamma detection.

Results

Affinity Chromatography

Using a well established method of affinity chromatography, we sought to retain

proteins specific for human glycophorin A (hGA) yet having no specificity for porcine

erythrocyte glycoproteins or BSA. Each affinity column showed a similar loading protein

profile but different elution profiles (Figure 1). All three columns retained small amounts

of BSA, yet only the hGA affinity column retained other proteins. Loading and eluting

fractions were compared by SDS-PAGE and silver staining, which revealed that loading

fractions were similar. Prior to elution, the three fractions revealed no observable protein.

However, the first three elution fractions revealed that the hGA gave rise to proteins at

42kDa and 35kDa in addition to a 62kDa BSA band (Figure 1).

Mass Spectrometry and Sequencing of Porcine Annexin IV

To identify the bands detected by silver staining we subjected coomassie stained

bands from SDS-PAGE to sequence analysis by mass spectrometry. The 35kDa band gave

rise to eight peptides after trypsin digestion. Amino acid sequences from all eight

fragments corresponded to porcine annexin IV protein (Table I). The 42kDa band gave

rise to nine peptides. After sequencing, four peptides corresponded to annexin I, while five

peptides corresponded to annexin II. The 62kDa band gave rise to three peptides that,

when sequenced, corresponded to BSA.

Figure 1. Human Glycophorin A-Conjugated Affinity Column Screening of Porcine

Kupffer Cell Lysates.

0.4000

0.3500 185kDa 0.3000 A 98 n ei

t 0.2500 o

r 62 0.2000 l p

/m 0.1500 49 g m 0.1000 38 0.0500 28 0.0000 17 1 7 13 19 25 31 37 43 49 55 61 67 73 79 85 91 97

0.4500 0.4000 185 0.3500 B 98 0.3000 ein t o

r 0.2500 62 l p 0.2000 49 /m

g 0.1500 m 38 0.1000 0.0500 28 0.0000 17 1 7 13 19 25 31 37 43 49 55 61 67 73 79 85 91 97

0.4500

0.4000 185 0.3500 C 98 n i

e 0.3000 t

o 62 r 0.2500

l p 0.2000 49 /m

g 0.1500

m 38 0.1000 0.0500 28 0.0000 17 1 8 15 22 29 36 43 50 57 64 71 78 85 92 99

Activated CNBr sepharose was used to prepare human glycophorin A- (A), porcine glycophorin A- (B), and BSA- (C) conjugated affinity columns. Porcine Kupffer cell lysate (500 µl of a 1mg/ml solution) was added to each column and fifty 1ml fractions were collected. Elution buffer was added and 50 additional 1ml fractions were collected. The fractions were plotted based on the protein concentration of each fraction on the left. Fractions 11, 12, 66, 67, 68, 69, 70, and 71 for each column were subjected to SDS-PAGE analysis followed by silver staining on the right.

Table I. Peptide and Amino Acid Sequence Data Obtained by Mass Spectrometry.

Acc. No. Protein Name Theoretical Observe Peptide Sequence

Mass d Mass

P19619 Annexin I (Lipocortin I) 605.34 605.62 299-303 IMVSR

(Calpactin II) (Chromobindin 9) 829.50 829.35 251-257 VLDLELK

(P35) (Phospholipase A2 908.44 908.42 205-212 ALYEAGER

inhibitory protein) 1733.78 1734.20 189-204 SEDLAINDDLADTDAR

P04272 Annexin II (Lipocortin II) 2064.98 2064.96 179-196 RAEDGSVIDYELIDQDA

(Calpactin I heavy chain) 859.52 859.51 81-88 R

(Chromobindin 8) (P36) (Protein 880.44 880.33 197-204 KELASALK

I) (Placental anticoagulant protein 1222.59 1222.41 105-115 DLYDAGVK

IV) (PAP-IV) (Bovine) 1094.52 1094.35 69-77 TPAQYDASELK

qDIAFAYQR

P08132 Annexin A4 (Annexin IV) 832.44 832.50 219-225 DIEQSIK

(Lipocortin IV) (Endonexin I) 831.44 831.34 63-69 DLLDDLK

847.41 847.59 135-141 SLEDDIR

1383.70 1383.28 124-134 INQTYQLQYGR

1174.60 1174.61 260-270 GLGTDDNTLIR

1570.81 1570.72 203-214 NHLLHVFDEYKR

1692.88 1692.56 29-44 GLGTDEDAIISVLAYR

1666.85 1666.22 226-241 SETSGSFEDALLAIVK

P02769 Serum albumin precursor 1305.71 1305.96 402-412 HLVDEPQNLIK

(Bovine) 1639.93 1640.20 437-451 KVPQVSTPTLVEVSR

886.41 886.13 131-138 DDSPDLPK

Fractions 69, 70, and 71 containing proteins eluted from the human glycophorin A affinity column (figure 1A) were concentrated and subjected to SDS-PAGE analysis followed by protein staining and excision from the gel. Excised bands were digested with

trypsin and extracted with 60% acetonitrile :0.1% TFA and concentrated to ~ 15 µl. Two

µl of the sample was separated on a reverse phase column (75 um id X 5 cm X 15 um

Aquasil C18 Picofrit column, New Objectives). Eluent (peptides) were directly introduced

into ion-trap mass spectrometer (LCQ-Deca XP Plus, Finnigan) equipped with nano-spray

source. Mass spectrometer was operated on a double play mode where the instrument was

set to acquire a full MS scan (400-2000 m/z) and a collision induced dissociation (CID) spectrum on the most abundant ion from the full MS scan. The CID spectra were either manually interpreted or searched against an appropriate non-redundant database using

TurboSEQUEST. The accession number for the homologous sequence is listed in the first column followed by the protein most current name in the second column (other names may be listed). The third and fourth columns contain the mass predicted by the sequence and the actual mass determined by mass spectrometry. The fifth and sixth columns contain the amino acid position and sequence of the observed peptide.

Western Blot

Western blot analysis of the fractions from the human glycophorin A/sepharose column using anti-human annexin IV monoclonal antibody revealed that the two loading fractions and the three eluted fractions blotted contained annexin IV. The three fractions collected prior to elution did not contain annexin IV when analyzed by Western blot

(Figure 2).

Immunohistochemistry

While it is significant to find that annexin IV is retained by the human GA column and is found in a Kupffer cell membrane lysate, it is necessary for porcine annexin IV to be expressed on the surface of porcine Kupffer cells to substantiate our findings. Porcine

Kupffer cells probed with anti-human annexin IV monoclonal antibody directed towards the N-terminus of human annexin IV revealed that porcine annexin IV is expressed on the surface of the cells (Figure 3). Interestingly, there were other cells present that resembled

Kupffer cells in size and shape that did not stain positive for annexin IV (Figure 3).

Porcine Kupffer cells incubated with anti-human glycophorin A mouse IgG or not incubated with a primary or secondary antibody did not show a color change.

Anti-human Annexin IV in the Rosetting Assay

Most monocytes express receptors that bind the Fc region of antibodies. When we incubated porcine Kupffer cells with a dilution of mouse IgG we found that this had no effect on the ability of the porcine Kupffer cells to bind human erythrocytes. When

Figure 2. Western Blot Analysis of Porcine Kupffer Cell Proteins Eluted From a

Human Glycophorin A Affinity Column.

185kDa

98kDa

62kDa

49kDa

38kDa

28kDa 17kDa 1 23456789

Samples from the human glycophorin A affinity column were subjected to western blot analysis using the anti-human annexin IV monoclonal antibody. Lanes 1 and 2 represent a 20 µl sample from fractions 11 and 12 (Figure 1A). Lanes 3, 4, and 5 represent a 20 µl sample from fractions 66, 67, and 68 (Figure 1A). Lanes 6, 7, and 8 represent a 20 µl sample from fractions 69,70, and 71. Lane 9 represents a 5 µl sample of human endothelial cell lysate known to contain annexin IV. Lane 10 contains 5 µl of protein ladder.

Figure 3. Immunhistochemical Staining for the Presence of Annexin IV on Porcine

Kupffer Cells.

100X 1000X

KC+ Isotype control +2°

KC+ developer

KC+α-AnxIV +2°

Porcine Kupffer cells (in media with 2 mg/ml BSA) were fixed with acetone/methanol (1:1). Fixed Kupffer cells were subjected to immunohisotochemical staining with anti-human annexin IV mouse monoclonal IgG antibodies or anti-human glycophorin A mouse monoclonal IgG antibodies (isotype control). Secondary antibody was a goat anti-mouse IgG1. Porcine Kupffer cells incubated with or with out antibody were developed with alkaline phosphatase blue staining reagent (Vector Blue).

we used the same dilutions of mouse IgG anti-human annexin IV, we found a dose- dependent inhibition of rosetting that began at 1:1,000 and reached maximal inhibition of

55% at a 1:10 dilution (Figure 4).

Discussion

Annexins are a family of structurally related proteins that exhibit calcium- dependent binding of phospholipids (Raynal et al., 1994). Each of the 14 annexins have a highly conserved C-terminal region that consists of four 70 amino acid imperfect repeated units (or eight for annexin VI) (Geisow 1986). Each unit contains a highly conserved 17 amino acid domain called the endonexin-fold (Geisow 1986). These domains also contain a conserved amino acid residue that putatively binds calcium (Sohma et al., 1999). In the absence of phospholipid, the affinity for calcium is relatively low (25-1000 µm). In the presence of a specific phospholipid, however, the affinity for calcium can increase 100- fold (Glenney 1986; Pigault et al., 1990). These values, however, are based on the binding of multiple annexins. Due to their highly hydrophobic C-terminus, annexins precipitate in solution making it difficult to assess the effect of individual domains on the collective avidity for calcium and specific targets (Nelson et al., 1995). Analysis of the minor sequence variation of the C-terminus repeats revealed that annexins have been well conserved (Morgan et al., 1997a; Morgan et al., 1997b). In contrast, it has recently been theorized that the repeated units of the C-terminus of annexins may have different specificities for phospholipids (Sohma et al., 2001). Thus far, it appears that calcium binding to the endonexin-fold may be involved in determining the type of ligand bound. In our efforts to characterize the binding of human erythrocytes

Figure 4. Anti-Human Annexin IV mAb Inhibition of Human Erythrocyte Binding to

Porcine Kupffer Cells.

120

102 100 99 98 100 100 100

78 80

Mouse IgG Anti-AnxIV

60 53

% Binding 45

0 1 to 10,000 1 to 1000 1 to 100 1 to 10 Human rbc pig rbc

The anti-human annexin IV monoclonal antibody was prepared in the dilutions shown along the x-axis. Porcine Kupffer cells were incubated with each concentration of anti-human annexin IV for one hour prior to the addition of a 1% solution of 51Cr-labeled human erythrocytes. The isotype control at the same dilutions was tested simultaneously. Rosetting was verified by light microscopy and quantitated by gamma-detection. Samples were tested three times in triplicate. The error bars express standard error. by porcine Kupffer cells, we revealed that the binding was sensitive to both EDTA and

EGTA (25mM) (Burlak II 2004b). In light of this data, the calcium dependence of annexin

IV is consistent with its purposed role in binding human erythrocytes.

In contrast to the C-terminal domain, the sequence and length of the N-terminal domain of annexins are highly variable. The N-terminus or “tail” can be only a few amino acids as seen with annexin V or 160 amino acids as seen with annexin VII. The N-termini of annexins seem to have greater similarity between species than between individual annexins (Hauptmann et al., 1989). The N-termini are considered to be the regulatory portion of annexins as they have been shown to contain sites for phosphorylation, proteolysis, and interactions with proteins and carbohydrates (Gerke 1990; Kojima et al.,

1992; Jensen et al., 1996; Diaz et al., 2000; Satoh et al., 2000). Recently, the N-terminal domain of annexins IV, V, and VI was found to contain a carbohydrate-binding site that varied in carbohydrate specificity (Ishitsuka et al., 1998). These authors showed that annexin IV bound GP-2, a glycoprotein involved in apical sorting and secretion of zymogens, by its terminal sialic acids with α 2-3 linkages in the presence of calcium (Tsuji et al., 1981). This finding supports our recent data revealing evidence that the monosaccharide, N-acetylneuraminic acid, and the trisaccharides, neuraminyl lactoses

(~80% α2-3 and ~20% α2- 6 linked to lactose mixture), inhibited porcine Kupffer cell recognition of human erythrocytes) (Burlak II 2004b). These data are significant in light of our finding that other monosaccharides found on human glycophorin A had no effect on rosetting at the same concentration (Burlak II 2004b).

Affinity columns have been created from a variety of substrates to absorb proteins, glycoproteins, carbohydrates, DNA, and mRNA. All of these affinity columns are similar in that they require the receptor ligand interaction to have a large dissociation constant

(Kd) under a given set of conditions. Lectin affinity columns have been widely used to

screen complex mixtures of proteins and glycoproteins from a cell homogenate

(Gangopadhyay et al., 1996; Ishitsuka et al., 1998). The Kd of an individual carbohydrate

recognition domain (CRD) of a lectin and the mirrored carbohydrate ligand is usually

lower than the analogous protein/protein interaction (Monsigny et al., 2000). The strength

of lectin-carbohydrate interactions occurs due to multiple CRDs on the same or individual

lectin(s) binding to a cluster of ligands creating high avidity. Because we have previously

shown that human glycophorin A remains bound to porcine Kupffer cells under a variety

of conditions, we used the mild detergent, lithium diiodosalicylate (LIS), to isolate human

glycophorin A and subsequently create a CNBr/sepharose affinity column as described in

the methods. LIS-isolated human glycophorin A covalently bound to the CNBr/sepharose

matrix absorbed four proteins which were detected upon elution by silver staining and

BCA protein measurement (Figure 1A). LIS-isolated porcine glycophorin A and BSA

columns absorbed BSA but no other proteins when examined by silver staining (Figure 1B

and C). The proteins eluted from the human glycophorin A affinity column were revealed

to be BSA (~62kDa), annexin I and II (~42kDa), and annexin IV(~35kDa) by mass

spectrometry (approximate molecular weights were determined by SDS-PAGE). Both the

porcine glycophorin A column and BSA column gave rise to BSA upon elution. This suggests that BSA is non-specifically recognized by the CNBr/sepharose matrix while other proteins are not retarded or bound.

Annexins possess a characteristic expression in different tissues. They also

maintain an intracellular distribution suggesting that annexins are involved in the

physiological functions specific to a tissue. The expression of annexins has been explored

to varying depths. Annexin I has been found in leukocytes of peripheral blood, tissue

macrophages, T-lymphocytes and in certain epithelial cells (respiratory and urinary system,

superficial cells of non-keratinised squamous epithelium) (Dreier et al., 1998). Despite

their high homology, annexin II has been found in endothelial cells, myoepithelial cells

and epithelial cells of the respiratory and urinary systems (Dreier et al., 1998). Annexin IV

was almost exclusively found in epithelial cells. Annexins I, II, and IV were found in

epithelia of the upper respiratory system, Bowman's capsule, urothelial cells, mesothelial

cells, peripheral nerves, the choroid plexus, ependymal cells and pia mater and arachnoid

of meninges (Dreier et al., 1998). Using immunohistochemistry, we revealed that porcine

Kupffer cells expressed annexin IV (Figure 3). These data are in contrast to previous

findings that annexin IV was focused in epithelial cells. Due to the carbohydrate binding

properties of annexin IV (Neu5Ac α2-3 –R), its expression pattern between species may

vary significantly. Through a point mutation in the CMP-sialic acid hydroxylase gene,

humans have lost the ability to produce the sialic acid called N-glycolylneuraminic acid

(Neu5Gc) (Brinkman-Van der Linden et al., 2000; Varki 2001). N-glycolylneuraminic acid differs from Neu5Ac in that the acetyl group on the 5th carbon of N-acetylneuraminic

acid has been additionally substituted with a hydroxyl group. This small difference has

given rise to a possible immunologic barrier between species. Our closest relative, the

chimpanzee, has approximately 99% genetic identity to our genome yet predominantly

sialates oligosaccharides with Neu5Gc (Sarich et al., 1967; King et al., 1975; Muchmore et

al., 1998). Thus, at some point following divergence from a common ancestor, humans

have potentially developed unique roles for sialic acid, specifically Neu5Ac. This is

demonstrated by the recent finding that a strain of human colon carcinoma cells was

immunologically recognized only after expression of Neu5Gc (Muchmore et al., 1998;

Varki 2001). Of importance to the field of xenotransplantation, pigs have been found to

express both Neu5Ac and Neu5Gc, though Neu5Gc is the predominant form of sialic acid

found (Hueso et al., 1985; Sato et al., 1998; Malykh et al., 2003). The fact that Neu5Gc is

so prevalent in pigs and is not routinely expressed in humans makes it plausible that the

function annexin IV on porcine monocytes differs significantly from that observed in

humans.

Annexin IV has been called by a variety of other names such as endonexin I,

protein II, chromobindin 4, 32.5K calcinectrin, placental anticoagulant protein II, placental

protein PP4-X, 35-beta-calcimedine, and lipocortin IV (Crumpton et al., 1990). It seems

that this protein with numerous names has an equally numerous number of in vitro and in

vivo activities. Secreted annexin IV has been reported to have anticoagulant activity (Tait

et al., 1988), while membrane bound annexin IV plays a role in exocytosis (Zaks et al.,

1991). Intracellular annexin IV has been reported to be involved in the regulation of both

the cytoskeleton and chloride ion channels (Kaetzel et al., 1994). Although annexin IV has recently been described to bind glycosaminoglycans terminated with sialic acid its function

for sialic acid recognition has not been elucidated (Ishitsuka et al., 1998). Recent evidence

suggests that annexin IV may have endogenous ligands in the exocrine pancreas (Tsujii-

Hayashi et al., 2002). These authors showed that annexin IV binds terminal Neu5Ac with

α-2,3 linkages of GP-2 in the presence of calcium, which plays a role in apical sorting and

secretion of zymogens (Tsujii-Hayashi et al., 2002). To directly assess the role of porcine annexin IV in porcine Kupffer cell binding of human erythrocyte recognition, we incubated porcine Kupffer cells with anti-human annexin IV monoclonal antibody at the dilutions shown (Figure 4). At each dilution, a mouse IgG was tested to control for non- specific inhibition. Porcine Kupffer cells were inhibited from binding human erythrocytes beginning at a 1:1000 dilution (22%) and reaching maximal inhibition at a 1:10 dilution

(55%). The ability to inhibit this interaction with monoclonal antibodies further supports

our observation that annexin IV bound specifically to the hGA affinity column. These data

together with the immunohistochemical data suggests that recognition of extracellular

α−2,3 linked Neu5Ac by porcine Kupffer cells may pertain to a unique immunological role in porcine innate cellular immunity.

(2001) and Janeway (1989) have referred to these antigens as pathogen-associated molecular patterns and to the innate immune receptors recognizing these patterns as

pattern-recognition receptors (Janeway 1989; Medzhitov 2001). The important role of these receptors in immune responses is becoming increasingly well recognized and, thus, it seems timely to explore the relationship of these receptors to the recognition of relevant xenotransplantation antigens.

SUMMARY

In an attempt to treat patients with liver failure with a form of “liver dialysis” it was observed that human erythrocytes were removed from the circulation over the course of 72 hours (Rees et al., 2002a). Without an appreciable increase in free hemoglobin suggesting red cell lysis, there was not an obvious explanation for the mechanism of erythrocyte loss.

Light and electron microscopy of fixed slides prepared from the perfused livers revealed that macrophages that line the sinusoidal passages of the liver, Kupffer cells, were binding and phagocytosing the human erythrocytes at the rate of one unit of blood per day of perfusion (Rees et al., 2002b). This finding was later observed in vitro when isolated porcine Kupffer cells were demonstrated to bind human erythrocytes but not porcine erythrocytes (Rees 2003c).

The hypothesis that porcine Kupffer cells bind human erythrocytes via a lectin- carbohydrate interaction was derived from two important observations. First, opsonization was not required for porcine Kupffer cells to bind human erythrocytes, suggesting that direct recognition occurred (Rees 2003c). Second, rat Kupffer cell binding of xenogeneic

erythrocytes was mediated by a lectin-carbohydrate interaction (Mohr et al., 1987). This

hypothesis was a novel concept of recognition within the field of xenotransplantation given

its focus on innate cellular recognition, but was consistent with the finding that

carbohydrate recognition was crucial in humorally-mediated hyperacute rejection (Galili

1998). Thus, our hypothesis was that porcine Kupffer cells recognize human erythrocytes

via a lectin carbohydrate interaction.

Within the manuscripts of this dissertation we used the rosetting assay described

previously to explore porcine Kupffer cell recognition of human erythrocytes. Both EDTA

and EGTA were found to inhibit the binding of human erythrocytes. This suggested that

the interaction was calcium dependent. Fragmented human erythrocytes were found to

inhibit porcine Kupffer cell recognition. Glycoproteins were isolated from both human

and porcine erythrocyte membrane fragments and protein, carbohydrate, and western blot

analysis revealed that the samples were predominantly composed of human or porcine

glycophorin A, respectively. A dilution of the isolated human erythrocyte glycoproteins

(EGP) in the rosetting assay were found to create a dose-dependent inhibition. The porcine

EGP, however, had no effect at the highest concentration tested. These data suggested that the human EGP sample contained a molecule that was capable of inhibiting the rosetting assay.

The human glycophorin A band from numerous SDS-PAGE gels was eluted and tested in the rosetting assay. From 40 SDS-PAGE gels one milligram of protein was obtained. The “SDS-PAGE-purified” human glycophorin A inhibited the rosetting assay to the same degree as the original human EGP sample. While this finding did not exclude an inhibitory molecule that migrated at the same rate when analyzed by SDS-PAGE, it did suggest that the ligand recognized by porcine Kupffer cells could be present on human glycophorin A.

Based on this information our hypothesis was modified to porcine Kupffer cells recognize a carbohydrate on human glycophorin A on the surface of human erythrocytes.

In an attempt to disprove this hypothesis, the human EGP sample was treated to remove secondary and tertiary conformations, to break disulfide bonds, and to degrade the protein using trypsin and pronase. These treatments were designed to test the influence of the protein portions of human EGPs for their role in inhibition. When tested in the rosetting assay, these four samples maintained their ability to inhibit porcine Kupffer cell recognition of human erythrocytes to the same degree as the sham treated samples. This finding suggested that if the binding was not related to a protein component of human glycophorin A.

To examine the potential for human glycophorin A to be recognized by porcine

Kupffer cells human EGP followed by a probe of anti-human glycophorin A monoclonal antibody were incubated with Kupffer cells. Cells positive for binding human glycophorin

A bacame blue in color, while all of the control cells remained unstained. These findings confirmed that human glycophorin A was directly recognized by porcine Kupffer cells.

The carbohydrate portion of human glycophorin A was degraded using a cocktail of enzymes specific for the oligosaccharide structures that have been previously proposed.

After verifying deglycosylation by PAS staining of SDS-PAGE and Western blotting analysis, the sample was tested in the rosetting assay. The deglycosylated human EGP sample had no ability to inhibit porcine Kupffer cell recognition of human erythrocytes, yet the sham treated human EGP retained 100% of its ability to inhibit the assay.

Based on this finding, the saccharides that have been shown to comprise the oligosaccharides of human glycophorin A were tested in the rosetting assay for their inhibitory potential. Each saccharide was tested using serial dilution samples, including a saccharide known not to exist within the oligosaccharides of human glycophorin A. As shown in Manuscript 1, Figure 1 within this dissertation, only N-acetylneuraminic acid was able to inhibit the rosetting assay. N-acetylneuraminic acid linked to lactose was also tested in the rosetting assay and found to inhibit comparably with the monosaccharide, N- acetylneuraminic acid. To confirm these findings a more targeted deglycosylation of the human EGP sample was necessary to elucidate the saccharides involved. Sialidase A was used to cleave terminal N-acetylneuraminic acid residues from oligosaccharides in the human EGP sample as well as whole erythrocytes. Sialidase A-treated human EGP was similar to the deglycosylated human EGP sample in that it’s ability to inhibit the rosetting assay was completely abolished. Intact human erythrocytes treated with sialidase A were unable to be recognized by the porcine Kupffer cells while sham treated intact human erythrocytes formed human erythrocyte rosettes comparable to the rosettes observed with the untreated human erythrocytes.

It has been previously suggested that some of the attractive force of heavily sialated glycoproteins is their overall negative charge (Varki 1997; Wasylnka et al., 2001). To address this issue, hyaluronic acid, which is not sialated but nonetheless bears a large negative charge was tested in the rosetting assay. Hyaluronic acid had no effect on the rosetting assay thus suggesting that porcine Kupffer cells recognize human erythrocytes by

their terminal N-acetylneuraminic acids and not by the overall negative charge that is generated.

While the sialic acid that exists on human erythrocytes is terminal N- acetylneuraminic acid, porcine erythrocytes are predominantly sialated with N- glycolylneuraminic acid. Mucins secreted in mammalian saliva have been shown to bear oligosaccharides terminated with sialic acids (Hueso et al., 1985; Sato et al., 1998; Malykh et al., 2003). While it has been shown that bovine submaxillary mucin (BSM) and Porcine submaxillary mucin (PSM) are similar in sialic acid content, others have suggested that

BSM is predominantly terminated with N-acetylneuraminic acid while PSM is predominantly terminated with N-glycolylneuraminic acid (Sherblom et al., 1983; Higa et al., 1985; Savage et al., 1986). Testing a dilution of both glycoproteins in the rosetting assay revealed that while porcine submaxillary mucin had no effect on the rosetting assay, bovine submaxillary mucin successfully inhibited the rosetting assay. This does more than suggest that N-acetylneuraminic acid is responsible for recognition. This evidence suggests that the inability of humans to terminate oligosaccharides with N- glycolylneuraminic acid may be the defining characteristic that induced their recognition by porcine Kupffer cells during xenoperfusion.

The way in which a saccharide is joined or linked to another saccharide may be important to the recognition by porcine Kupffer cells as well. The N-linked oligosaccharide on the surface of human glycophorin A bears only terminal N- acetylneuraminic acid linked α-2,6 to the oligosaccharide. The N-linked oligosaccharides

were specifically removed with the enzyme PNGaseF. The inhibition of rosetting by human glycophorin A deglycosylated in this way was comparable to that of the sham treated sample, suggesting that either terminal N-acetylneuraminic acid linked α-2,6 is not the correct linkage or that the removal of this structure did not substantially change the overall concentration of the pertinent linkage on the human glycophorin A glycoprotein.

This evidence supported the presence of an inhibitor within the human EGP sample that was likely to be terminal N-acetylneuraminic acid. The human EGP sample was used as a ligand cross-linked to an affinity column to extract the porcine Kupffer cell receptor from a membrane lysate. Three affinity columns were prepared; a porcine Kupffer cell lysate was aliquoted into three samples and passed over the human EGP, porcine EGP, and bovine serum albumin affinity columns. Only the human EGP column released two bands as revealed by silver and coomassie staining of SDS-PAGE of column fractions. Mass spectrometric analysis of the eluted proteins after tryptic digestion by tryptic peptide length and sequence. High homology matches were found for annexin I, II, and IV within the two protein bands eluted (in addition to BSA). Of these three, annexin IV has recently been shown to have a calcium-dependent carbohydrate binding domain specific for N- acetylneuraminic acid specifically linked α-2,3 to an oligosaccharide (Kojima et al., 1996).

Annexin I and II were not explored as part of this dissertation.

Thus the hypothesis became: Porcine annexin IV on the surface of porcine Kupffer cells recognizes terminal N-acetylneuraminic acid residues linked to oligosaccharides attached to human glycophorin A on the human erythrocyte membrane. Anti-human

annexin IV monoclonal antibody was used for both immunohistochemistry and testing for inhibition in the rosetting assay. When Kupffer cells were incubated with anti-annexin IV as shown in Manuscript 3, Figure 4 of this dissertation, a strong development of blue color was observed indicating the presence of integral or surface bound annexin IV. The control cells, including an isotype control, revealed no color development. Using this antibody to probe a western blot composed of samples from the loading and eluting fractions of the human EGP affinity column, the annexin IV band developed in the fractions containing the loading peak, yet absent from the loading lanes just prior to elution. The three elution fractions that were examined by western blot were found to be positive for annexin IV.

Finally, when a dilution of anti-annexin IV monoclonal antibody was incubated with porcine Kupffer cells prior to incubation with human erythrocytes, a dose-dependent inhibitory response was revealed. Porcine Kupffer cells incubated with the same dilutions of a mouse IgG isotype control had no effect on rosetting. Therefore, congruent with the above hypothesis, this dissertation suggests that annexin IV on the surface of porcine

Kupffer cells recognizes N-acetylneuraminic acid linked to an oligosaccharide borne on human glycophorin A.

Implications

The direct benefit of this body of work lies in the possibility of preventing the loss of human erythrocytes during extracorporeal porcine liver perfusion to support a human

patient with fulminant liver failure. We now have an understanding of how porcine

Kupffer cells bind human erythrocytes and can develop molecules to interfere with this

interaction. A classic method of inhibition is to use antibodies specific for an epitope to block recognition. An alternative to antibody production against annexin IV would be the synthesis of a neoglycoprotein, a core protein either chemically or enzymatically substituted with oligosaccharides. An albumin core substituted with N-acetylneuraminic acid of the appropriate linkage to an underlying oligosaccharide could be produced. Both an antibody directed against annexin IV and a neoglycoprotein could be tested during extracorporeal liver perfusion to determine if such an approach would prevent the loss of erythrocytes observed. The anti-porcine annexin IV antibody would have to be screened so as not to bind to human annexin IV and both strategies would have to be tested for safety in humans. Nonetheless, successful inhibition by either of these molecules within the extracorporeal perfusion model would suggest a potentially clinically relevant

molecule.

An alternative to using an inhibitor would be to develop an annexin IV-knockout

pig. Provided that the knockout mutation was not lethal or resulting in an unacceptable

product, the resulting porcine Kupffer cells could be tested in vitro for their ability to bind human erythrocytes. If binding were not seen, such livers could be studied during extracorporeal xenoperfusion to see if the problem of loss of human erythrocytes had been eliminated.

100

Macrophage tolerance has typically been thought of as using macrophages isolated from the donor to present self antigen to a local population of recipient T-cells, thereby inducing T-cell tolerance through clonal deletion and/or anergy (Lambolez et al., 2002).

An interesting idea inspired by this dissertation is the concept of inducing tolerance of the innate immune system, specifically, macrophages. Tolerance within the innate immune system is not a new idea. It has been shown that the B1 subset of B-cells produces low- affinity, polyspecific antibodies against the blood group antigens A and B (Wuttke et al.,

1997). B1-cells also comprise the majority of neonatal B cells and are thought be T-cell independent thus part of the innate immune system (Wuttke et al., 1997). It has been suggested that neonatal exposure to blood group carbohydrates that coincidentally match bacterial carbohydrates tolerize the B1-cells to these carbohydrate antigens. Blood group carbohydrates not expressed neonatally then are responded to later on the surface of bacteria leading to the well known presence of anti-blood group antibodies in humans

(Wuttke et al., 1997). Thus, there does appear to be a mechanism for tolerizing carbohydrate-specific antibodies that are expressed in germline configuration and, as such, have innate immune system characteristics. Is it possible that germline macrophage lectin receptors might also be tolerized by neonatal exposure to the appropriate carbohydrate ligands? Identification of the exact target oligosaccharide recognized by the porcine annexin IV lectin would provide a means to test this hypothesis if the target antigen could be expressed in the neonatal pig.

101

Another potential gene knockout would be toward the enzyme responsible for

adding the hydroxyl subgroup to the acetyl on the 5th carbon of N-acetylneuraminic acid.

Humans have lost the ability to create this commonly expressed enzyme and therefore

cannot produce N-glycolylneuraminic acid (Varki 2001). As a result, humans sialate

oligosaccharides almost exclusively with N-acetylneuraminic acid. The repercussions of

this alteration in nature was obviously not lethal and allowed humans to develop the capacity to mount an immune response against N-glycolylneuraminic acid. This is evident

in recent publications that found that humans produce antibodies directed against N-

glycolylneuraminic acid when expressed on tumors or xenografts (Malykh et al., 2001).

Such a glycolyl-knockout pig may express N-acetylneuraminic acid on its erythrocytes as

the pig still has the enzymes required to do this. Such a pig may tolerize its macrophages

to N-acetylneuraminic acid borne on erythrocytes by exposing macrophages to large

amounts of this carbohydrate neonatally. The expression of annexin IV in these knockout

pigs could then be compared with wild-type pigs to see if the presence of N-

acetylneuraminic acid on porcine erythrocytes induced a differential expression of this

receptor in the respective pig strain.

As porcine Kupffer cells recognize human epitopes, human macrophages may

express a receptor that recognizes porcine epitopes in a similar manner. Thus, a receptor

of this nature could play a role xenograft rejection. To date, a lectin with specificity

toward N-glycolylneuraminic acid has not been revealed. Considering the current research

exploring the development of antibodies directed toward N-glycolylneuraminic acid

102

expressed on tumors and the relevance of this information for the field of xenotransplantation, it seems prudent to explore the possibility of germline receptors specific for N-glycolylneuraminic acid in humans. This concept is consistent with the observations of other investigators who have found that human innate immune cells can bind to porcine target cells via carbohydrates (Inverardi et al., 1996, 1997; Artrip et al.,

1997, 1999; Kwiatkowski et al., 1999, 2000 a,b; Miyagawa et al., 1999; ).

The evidence for lectin recognition of carbohydrate epitopes in innate immunity has been well documented. Many of these interactions are based on the recognition of a bacterial or fungal pathogen. In this way, humans have receptors capable of targeting an epitope that is characteristic of a group of pathogens. This concept has been thoroughly discussed by Janeway et al., using the terms pattern recognition receptors (PPR) and pathogen associated molecular patterns (PAMP) (Janeway et al., 2002). In the context of xenotransplantation, however, PPRs have an as yet undetermined significance in rejection based on carbohydrate difference between species. As mentioned above, the gene necessary to produce the enzyme that catalyzes the hydroxylation of the N-acetyl group of

N-acetylneuraminic acid to create N-glycolylneuraminic acid has undergone a point mutation in recent history so recent that only humans and a few species of apes have the mutation. Pigs, however, produce N-glycolylneuraminic acid in a tissue specific manner.

Thus, I propose a new hypothesis: differences in glycosylation between species that are coincidentally shared with PAMPs allow recognition of xenogeneic targets by PPRs and contribute to the rejection of xenografts. The quantitative model that we have developed in

103

which porcine Kupffer cells bind human erythrocytes could be manipulated for use as a screening tool for the recognition of xenoantigens of various cells.

In addition to the direct recognition of a xenoantigen, a recent report has presented evidence that the lack of a self marker can influence binding. The newly developed macrophage receptor, SIRPα, has been shown to be involved in macrophage mediated erythrophagocytosis (Oldenborg et al., 2000). Previously, erythrocytes were not known to bear self molecules such as HLA antigens. Recent discoveries have made it clear that other receptor-ligand pairs can serve a role in self-recognition such as SIRPα/CD47 or

CD33/Sialic acid. It was found that SIRPα recognition of CD47 on the surface of erythrocytes prevented the erythrophagocytosis of healthy erythrocytes (Oldenborg et al.,

2000). Thus, if CD47/SIRPα generates a negative signal it seems logical that another receptor would be necessary to create a positive signal for erythrophagocytosis. Therefore, phagocytosis would only occur in the absence of SIRPα or in a disproportionate stimulation of a positive signal. This discovery puts forth the possibility that recognition of human erythrocytes by porcine Kupffer cells is not based solely on a difference in sialic acid type but partly on a lack of a porcine CD47-like self signal on human erythrocytes that would be bound by a SIRPα-like receptor on the porcine Kupffer cells. In this way, the absence of the SIRPa/CD47 signal to the porcine Kupffer cell would be a permissive signal to the Kupffer cell announcing that destruction of the sialic acid bearing cell bound by annexin IV was an appropriate activity.

104

Interestingly, it appears that sialic acid also has been found to be involved in negative signal transduction. CD33 (Siglec-3) and related siglecs, bear an ITIM-like motif in the cytoplasmic region of these proteins which has been associated with inhibitory signaling (Crocker et al., 2001a). The authors suggested that the CD33-related siglecs exert an inhibitory effect bound to the appropriate sialic acid-bearing oligosaccharide, thus preventing activation when a cell bears the marker of self (Crocker et al., 2001a). CD33- related siglecs also have been shown to transmit inhibitory signals when binding self sialic acids (Ravetch et al., 2000). It is worth noting that sialic acids are rarely expressed in lower species which is consistent with the theory that they appeared relatively late in evolution. Indeed, recent genomic databases from Caenorhabditis elegans and Drosophila melanogaster failed to uncover obvious sequences encoding known enzymes of the sialic acid biosynthetic pathway (Angata et al., 2000). Thus, it is a reasonable hypothesis that the absence of appropriate sialic acid-bearing ligands identifies a pathogen whereas appropriate sialic acid expression distinguishes self. Just as the absence of human MHC

Class I molecules allows for the activation of NK cells against porcine targets, so the lack of appropriate sialic acid epitopes may prevent inhibitory signaling via siglec receptors and lead to immune activation with subsequent xenograft rejection (Dorling et al., 2000;

Ravetch et al., 2000; Forte et al., 2001; Sharland et al., 2002).

105

BIBLIOGRAPHY

Adamany, A. M.; Blumenfeld, O. O.; Sabo, B. and McCreary, J. 1983 A carbohydrate

structural variant of MM glycoprotein (glycophorin A). J. Biol. Chem., 258(19):

11537-45.

Aminoff, D. 1988 The role of sialoglycoconjugates in the aging and sequestration of red

cells from circulation. Blood Cells, 14(1): 229-57.

Aminoff, D. and Morrow, M. P. 1970 Effect of sialidase on blood group specificity of

hog submaxillary glycoproteins. FEBS Lett., 8(6): 353-358.

Angata, T. and Varki, A. 2000 Cloning, characterization, and phylogenetic analysis of

siglec-9, a new member of the CD33-related group of siglecs. Evidence for co-

evolution with sialic acid synthesis pathways. J. Biol. Chem. 275(29): 22127-35.

Angata, T. and Varki, A. 2002 Chemical diversity in the sialic acids and related alpha-keto

acids: an evolutionary perspective. Chem. Rev., 102(2): 439-69.

Artrip, J. H.; Kwiatkowski, P.; Michler, R. E. and Itescu, S. 1997 Immunologic barriers

in pig-to-primate cardiac xenotransplantation. Ann.Transplant, 2(1): 5-11.

Artrip, J. H.; Kwiatkowski, P.; Michler, R. E.; Wang, S. F.; Tugulea, S.; Ankersmit, J.;

Chisholm, L.; McKenzie, I. F.; Sandrin, M. S. and Itescu, S. 1999 Target cell

susceptibility to lysis by human natural killer cells is augmented by alpha(1,3)-

galactosyltransferase and reduced by alpha(1, 2)-fucosyltransferase. J. Biol. Chem.,

274(16): 10717-22.

106

Bach, F. H.; Winkler, H.; Ferran, C.; Hancock, W. W. and Robson, S. C. 1996 Delayed

xenograft rejection. Immunol. Today, 17(8): 379-84.

Blakely, M. L.; Van der Werf, W. J.; Berndt, M. C.; Dalmasso, A. P.; Bach, F. H. and

Hancock, W. W. 1994 Activation of intragraft endothelial and mononuclear cells

during discordant xenograft rejection. Transplantation 58(10): 1059-66.

Blanchard, D.; Cartron, J. P.; Fournet, B.; Montreuil, J.; van Halbeek, H. and Vliegenthart,

J. F. 1983 Primary structure of the oligosaccharide determinant of blood group Cad

specificity. J Biol Chem 258(12): 7691-5.

Bratosin, D.; Mazurier, J.; Slomianny, C.; Aminoff, D. and Montreuil, J. 1997 Molecular

mechanisms of erythrophagocytosis: flow cytometric quantitation of in vitro

erythrocyte phagocytosis by macrophages. Cytometry 30(5): 269-74.

Bretscher, M. S. 1971 A major protein which spans the human erythrocyte membrane. J

Mol Biol 59(2): 351-7.

Brinkman-Van der Linden, E. C.; Sjoberg, E. R.; Juneja, L. R.; Crocker, P. R.; Varki, N.

and Varki, A. 2000 Loss of N-glycolylneuraminic acid in human evolution.

Implications for sialic acid recognition by siglecs. J Biol Chem 275(12): 8633-40.

Burlak II, C. B.; Twining, L. M. and Rees, M. A. 2004a Carbohydrates borne on human

glycophorin A are recognized by porcine Kupffer cells. [Submitted].

Transplantation.

107

Burlak II, C. B.; Twining, L. M. and Rees, M. A. 2004b Terminal sialic acid residues on

human glycophorin A are recognized by porcine Kupffer cells. [Manuscript in

preparation].

Byrne, G. W.; McCurry, K. R.; Martin, M. J.; McClellan, S. M.; Platt, J. L. and Logan, J.

S. 1997 Transgenic pigs expressing human CD59 and decay-accelerating factor

produce an intrinsic barrier to complement-mediated damage. Transplantation

63(1): 149-55.

Candinas, D.; Belliveau, S.; Koyamada, N.; Miyatake, T.; Hechenleitner, P.; Mark, W.;

Bach, F. H. and Hancock, W. W. 1996 T cell independence of macrophage and

natural killer cell infiltration, cytokine production, and endothelial activation during

delayed xenograft rejection. Transplantation 62(12): 1920-7.

Cochet, S.; Volet, G.; Cartron, J. P. and Bertrand, O. 2001 New procedures for

glycophorin A purification with high yield and high purity. J Chromatogr B

Biomed Sci Appl 750(1): 109-19.

Cooper, D. K.; Good, A. H.; Koren, E.; Oriol, R.; Malcolm, A. J.; Ippolito, R. M.;

Neethling, F. A.; Ye, Y.; Romano, E. and Zuhdi, N. 1993 Identification of alpha-

galactosyl and other carbohydrate epitopes that are bound by human anti-pig

antibodies: relevance to discordant xenografting in man. Transpl. Immunol., 1(3):

198-205.

Cooper, D. K.; Koren, E. and Oriol, R. 1994 Oligosaccharides and discordant

xenotransplantation. Immunol Rev 141: 31-58.

108

Costa, C.; Zhao, L.; Burton, W. V.; Rosas, C.; Bondioli, K. R.; Williams, B. L.; Hoagland,

T. A.; Dalmasso, A. P. and Fodor, W. L. 2002 Transgenic pigs designed to express

human CD59 and H-transferase to avoid humoral xenograft rejection.

Xenotransplantation, 9(1): 45-57.

Cowan, P. J.; Aminian, A.; Barlow, H.; Brown, A. A.; Chen, C. G.; Fisicaro, N.; Francis,

D. M.; Goodman, D. J.; Han, W.; Kurek, M.; Nottle, M. B.; Pearse, M. J.;

Salvaris, E.; Shinkel, T. A.; Stainsby, G. V.; Stewart, A. B. and d'Apice, A. J. 2000

Renal xenografts from triple-transgenic pigs are not hyperacutely rejected but cause

coagulopathy in non-immunosuppressed baboons. Transplantation 69(12): 2504-

15.

Cozzi, E. and White, D. J. 1995 The generation of transgenic pigs as potential organ

donors for humans. Nat. Med., 1(9): 964-6.

Crocker, P. R. and Gordon, S. 1986 Properties and distribution of a lectin-like

hemagglutinin differentially expressed by murine stromal tissue macrophages. J

Exp. Med., 164(6): 1862-75.

Crocker, P. R.; Hill, M. and Gordon, S. 1988 Regulation of a murine macrophage

haemagglutinin (sheep erythrocyte receptor) by a species-restricted serum factor.

Immunology, 65(4): 515-22.

Crocker, P. R.; Kelm, S.; Dubois, C.; Martin, B.; McWilliam, A. S.; Shotton, D. M.;

Paulson, J. C. and Gordon, S. 1991 Purification and properties of sialoadhesin, a

sialic acid-binding receptor of murine tissue macrophages. Embo. J., 10(7): 1661-9.

109

Crocker, P. R. and Varki, A. 2001a Siglecs in the immune system. Immunology, 103(2):

137-45.

Crocker, P. R. and Varki, A. 2001b Siglecs, sialic acids and innate immunity. Trends

Immunol., 22(6): 337-42.

Crumpton, M. J. and Dedman, J. R. 1990 Protein terminology tangle. Nature, 345(6272):

212.

Dahr, W.; Uhlenbruck, G.; Janssen, E. and Schmalisch, R. 1976 Heterogeneity of human

cell membrane sialoglycoproteins. Blut., 32(3): 171-84.

Dahr, W.; Wilkinson, S.; Issitt, P. D.; Beyreuther, K.; Hummel, M. and Morel, P. 1986

High frequency antigens of human erythrocyte membrane sialoglycoproteins, III.

Studies on the EnaFR, Wrb and Wra antigens. Biol. Chem. Hoppe Seyler, 367(10):

1033-45.

Dalmasso, A. P.; Vercellotti, G. M.; Platt, J. L. and Bach, F. H. 1991 Inhibition of

complement-mediated endothelial cell cytotoxicity by decay-accelerating factor.

Potential for prevention of xenograft hyperacute rejection. Transplantation, 52(3):

530-3.

Diaz, A.; Ibarguren, S.; Breijo, M.; Willis, A. C. and Sim, R. B. 2000 Host-derived

annexin II at the host-parasite interface of the Echinococcus granulosus hydatid

cyst. Mol. Biochem. Parasitol., 110(1): 171-6.

110

Dodd, R. B. and Drickamer, K. 2001 Lectin-like proteins in model organisms:

implications for evolution of carbohydrate-binding activity. Glycobiology, 11(5):

71R-9R.

Dodge, J. T.; Mitchell, C. and Hanahan, D. J. 1963 The preparation and chemical

characteristics of hemoglobin-free ghosts of human erythrocytes. Arch. Biochem.

Biophys., 100: 119-30.

Dorling, A.; Monk, N. and Lechler, R. 2000 HLA-G inhibits the transendothelial cell

migration of human NK cells: a strategy for inhibiting xenograft rejection.

Transplant Proc., 32(5): 938.

Dreier, R.; Schmid, K. W.; Gerke, V. and Riehemann, K. 1998 Differential expression of

annexins I, II and IV in human tissues: an immunohistochemical study. Histochem.

Cell Biol., 110(2): 137-48.

Eibl, M. M. 1971 Rosette formation by sensitized human erythrocytes on various

heterologous macrophages. Clin. Exp. Immunol., 8(3): 435-9.

Elson, C. J.; Allan, D.; Elson, J. and Duffus, W. H. 1972 The relationship between the

morphology of rosette-forming cells and their mode of rosette formation.

Immunology, 22(2): 291-300.

Fearon, D. T. and Locksley, R. M. 1996 The instructive role of innate immunity in the

acquired immune response. Science, 272(5258): 50-3.

Fodor, W. L.; Williams, B. L.; Matis, L. A.; Madri, J. A.; Rollins, S. A.; Knight, J. W.;

Velander, W. and Squinto, S. P. 1994 Expression of a functional human

111

complement inhibitor in a transgenic pig as a model for the prevention of

xenogeneic hyperacute organ rejection. Proc, Natl. Acad. Sci. USA, 91(23): 11153-

Forte, P.; Pazmany, L.; Matter-Reissmann, U. B.; Stussi, G.; Schneider, M. K. and

Seebach, J. D. 2001 HLA-G inhibits rolling adhesion of activated human NK cells

on porcine endothelial cells. J. Immunol., 167(10): 6002-8.

Fox, A.; Koulmanda, M.; Mandel, T. E.; van Rooijen, N. and Harrison, L. C. 1998

Evidence that macrophages are required for T-cell infiltration and rejection of fetal

pig pancreas xenografts in nonobese diabetic mice. Transplantation, 66(11): 1407-

16.

Fryer, J. P.; Leventhal, J. R.; Dalmasso, A. P.; Chen, S.; Simone, P. A.; Jessurun, J.; Sun,

L. H.; Reinsmoen, N. L. and Matas, A. J. 1994 Cellular rejection in discordant

xenografts when hyperacute rejection is prevented: analysis using adoptive and

passive transfer. Transpl. Immunol., 2(2): 87-93.

Fukuda, K.; Honma, K.; Manabe, H.; Utsumi, H. and Hamada, A. 1987 Alkali-labile

oligosaccharide units of a sialoglycoprotein from rabbit erythrocyte membranes.

Biochim. Biophys. Acta, 926(2): 132-8.

Galili, U. 1988 The natural anti-Gal antibody, the B-like antigen, and human red cell

aging. Blood Cells,14(1): 205-28.

112

Galili, U. 1993 Interaction of the natural anti-Gal antibody with alpha-galactosyl epitopes:

a major obstacle for xenotransplantation in humans. Immunol. Today, 14(10): 480-

Galili, U. 1998 Anti-gal antibody prevents xenotransplantation. Science & Medicine 5: 32.

Galili, U.; Buehler, J.; Shohet, S. B. and Macher, B. A. 1987 The human natural anti-Gal

IgG. III. The subtlety of immune tolerance in man as demonstrated by

crossreactivity between natural anti-Gal and anti-B antibodies. J Exp Med 165(3):

693-704.

Galili, U.; Mandrell, R. E.; Hamadeh, R. M.; Shohet, S. B. and Griffiss, J. M. 1988a

Interaction between human natural anti-alpha-galactosyl immunoglobulin G and

bacteria of the human flora. Infect. Immun., 56(7): 1730-7.

Galili, U.; Shohet, S. B.; Kobrin, E.; Stults, C. L. and Macher, B. A. 1988b Man, apes, and

Old World monkeys differ from other mammals in the expression of alpha-

galactosyl epitopes on nucleated cells. J. Biol. Chem., 263(33): 17755-62.

Gangopadhyay, A.; Lazure, D. A.; Kelly, T. M. and Thomas, P. 1996 Purification and

analysis of an 80-kDa carcinoembryonic antigen-binding protein from Kupffer

cells. Arch. Biochem. Biophys., 328(1): 151-7.

Gardner, B.; Parsons, S. F.; Merry, A. H. and Anstee, D. J. 1989 Epitopes on

sialoglycoprotein alpha: evidence for heterogeneity in the molecule. Immunology,

68(2): 283-9.

113

Geisow, M. J. 1986 Common domain structure of Ca2+ and lipid-binding proteins. FEBS

Lett., 203(1): 99-103.

Gerke, V. 1990 Structural and functional characterization of protein I (p36(2)p11(2)) and

II (p32)-calcium/phospholipid binding proteins with homologies to lipocortin I.

Adv. Exp. Med. Biol., 269: 135-8.

Glenney, J. 1986 Phospholipid-dependent Ca2+ binding by the 36-kDa tyrosine kinase

substrate (calpactin) and its 33-kDa core. J. Bio.l Chem., 261(16): 7247-52.

Goddard, M. J.; Dunning, J.; Horsley, J.; Atkinson, C. Pino-Chavez, G. and Wallwork, J.

2002 Histopathology of cardiac xenograft rejection in the pig-to-baboon model. J.

Heart Lung Transplant, 21(4): 474-84.

Goldstein, I. J.;Hughes, R.C.; Monsigny, M.; Osawa, T. and Sharon, N. 1980 What should

a lectin be called? [Letter]. Nature, 285: 66.

Haltiwanger, R. S.; Lehrman, M. A.; Eckhardt, A. E. and Hill, R. L. 1986 The distribution

and localization of the fucose-binding lectin in rat tissues and the identification of a

high affinity form of the mannose/N-acetylglucosamine-binding lectin in rat liver.

J. Biol. Chem., 261(16): 7433-9.

Hancock, W. W. 1997 Delayed xenograft rejection. World J. Surg., 21(9): 917-23.

Hancock, W. W.; Miyatake, T.; Koyamada, N.; Kut, J. P.; Soares, M.; Russell, M. E.;

Bach, F. H. and Sayegh, M. H. 1997 Effects of leflunomide and deoxyspergualin

in the guinea pig-->rat cardiac model of delayed xenograft rejection: suppression of

114

B cell and C-C chemokine responses but not induction of macrophage lectin.

Transplantation, 64(5): 696-704.

Hartnell, A.; Steel, J.; Turley, H.; Jones, M.; Jackson, D. G. and Crocker, P. R. 2001

Characterization of human sialoadhesin, a sialic acid binding receptor expressed by

resident and inflammatory macrophage populations. Blood, 97(1): 288-96.

Hauptmann, R.; Maurer-Fogy, I.; Krystek, E.; Bodo, G.; Andree, H. and Reutelingsperger,

C. P. 1989 Vascular anticoagulant beta: a novel human Ca2+/phospholipid

binding protein that inhibits coagulation and phospholipase A2 activity. Its

molecular cloning, expression and comparison with VAC-alpha. Eur. J. Biochem.,

185(1): 63-71.

Higa, H. H. and Paulson, J. C. 1985 Sialylation of glycoprotein oligosaccharides with N-

acetyl-, N-glycolyl-, and N-O-diacetylneuraminic acids. J. Biol. Chem., 260(15):

8838-49.

Holgersson, J.; Ehrnfelt, C.; Hauzenberger, E. and Serrander, L. 2002 Leukocyte

endothelial cell interactions in pig to human organ xenograft rejection. Vet.

Immunol. Immunopathol., 87(3-4): 407-15.

Hoyle, G. W. and Hill, R. L. 1988 Molecular cloning and sequencing of a cDNA for a

carbohydrate binding receptor unique to rat Kupffer cells. J. Biol. Chem., 263(16):

7487-92.

115

Hueso, P.; Reglero, A.; Rodrigo, M.; Cabezas, J. A.; Reuter, G. and Schauer, R. 1985

Isolation and characterization of gangliosides from pig lymphocytes. Biol. Chem.

Hoppe Seyler, 366(2): 167-71.

Inverardi, L.; Clissi, B.; Stolzer, A. L.; Bender, J. R. and Pardi, R. 1996 Overlapping

recognition of xenogeneic carbohydrate ligands by human natural killer

lymphocytes and natural antibodies. Transplant Proc., 28(2): 552.

Inverardi, L.; Clissi, B.; Stolzer, A. L.; Bender, J. R.; Sandrin, M. S. and Pardi, R. 1997

Human natural killer lymphocytes directly recognize evolutionarily conserved

oligosaccharide ligands expressed by xenogeneic tissues. Transplantation, 63(9):

1318-30.

Ishitsuka, R.; Kojima, K.; Utsumi, H.; Ogawa, H. and Matsumoto, I. 1998

Glycosaminoglycan binding properties of annexin IV, V, and VI. J. Biol. Chem.,

273(16): 9935-41.

Itescu, S.; Kwiatkowski, P.; Artrip, J. H.; Wang, S. F.; Ankersmit, J.; Minanov, O. P. and

Michler, R. E. 1998 Role of natural killer cells, macrophages, and accessory

molecule interactions in the rejection of pig-to-primate xenografts beyond the

hyperacute period. Hum. Immunol., 59(5): 275-86.

Janeway, C. A., Jr. 1989 Approaching the asymptote? Evolution and revolution in

immunology. Cold Spring Harb. Symp. Quant. Biol., 54 Pt 1: 1-13.

Janeway, C. A., Jr. and Medzhitov, R. 2002 Innate immune recognition. Annu. Rev.

Immunol., 20: 197-216.

116

Janeway, C. A., Jr.;Travers, P.; Walport, M. and Capra, D.J. 1999 In: Immunobiology,

The Immune System in Health and Disease., Elsevier Science/Garland Publishing.

Jaroslow, B. N. and Larrick, J. W. 1973 Clearance of foreign red cells from the blood of

aging mice. Mech. Aging Dev., 2(1): 23-32.

Jensen, P. H.; Jensen, T. G.; Laug, W. E.; Hager, H.; Gliemann, J. and Pepinsky, B. 1996

The exon 3 encoded sequence of the intracellular serine proteinase inhibitor

plasminogen activator inhibitor 2 is a protein binding domain. J. Biol. Chem.,

271(43): 26892-9.

Kaetzel, M. A.; Chan, H. C.; Dubinsky, W. P.; Dedman, J. R. and Nelson, D. J. 1994 A

role for annexin IV in epithelial cell function. Inhibition of calcium-activated

chloride conductance. J. Biol. Chem., 269(7): 5297-302.

Kelm, S. and Schauer, R. 1997 Sialic acids in molecular and cellular interactions. Int.

Rev. Cytol., 175: 137-240.

Kilpatrick, D. C. 2002 Animal lectins: a historical introduction and overview. Biochim.

Biophys. Acta, 1572(2-3): 187-97.

King, M. C. and Wilson, A. C. 1975 Evolution at two levels in humans and chimpanzees.

Science, 188(4184): 107-16.

Kojima, K.; Ogawa, H. K.; Seno, N.; Yamamoto, K.; Irimura, T.; Osawa, T. and

Matsumoto, I. 1992 Carbohydrate-binding proteins in bovine kidney have

consensus amino acid sequences of annexin family proteins. J. Biol. Chem.,

267(29): 20536-9.

117

Kojima, K.; Yamamoto, K.; Irimura, T.; Osawa, T.; Ogawa, H. and Matsumoto, I. 1996

Characterization of carbohydrate-binding protein p33/41: relation with annexin IV,

molecular basis of the doublet forms (p33 and p41), and modulation of the

carbohydrate binding activity by phospholipids. J. Biol. Chem., 271(13): 7679-85.

Kolb-Bachofen, V. 1991 A membrane-bound form of the acute-phase protein C-reactive

protein is the galactose-specific particle receptor on rat liver macrophages.

Pathobiology, 59(4): 272-5.

Kwiatkowski, P.; Artrip, J. H.; John, R.; Edwards, N. M.; Wang, S. F.; Michler, R. E. and

Itescu, S. 1999 Induction of swine major histocompatibility complex class I

molecules on porcine endothelium by tumor necrosis factor-alpha reduces lysis by

human natural killer cells. Transplantation, 67(2): 211-8.

Kwiatkowski, P.; Artrip, J. H.; Michler, R. E.; Wang, S. F.; McKenzie, I. F.; Sandrin, M.

S. and Itescu, S. 2000a Human monocytes bind to ligands on porcine endothelium

containing the gal alpha (1,3)-Gal antigen. Transplant Proc., 32(5): 931.

Kwiatkowski, P.; Artrip, J. H.; Wang, S. F.; Michler, R. E. and Itescu, S. 2000b Human

monocytes bind to two distinct carbohydrate capping structures on porcine

endothelium. Transplant Proc., 32(5): 927.

Lalezari, P.; Nehlsen, S. L.; Sinha, S. B.; Stemerman, M. B. and Veith, F. J. 1974

Spontaneous macrophage-erythrocyte rosette formation. A species-specific

phenomenon. Immunology, 27(3): 457-62.

118

Lambolez, F.; Jooss, K.; Vasseur, F. and Sarukhan, A. 2002 Tolerance induction to self

antigens by peripheral dendritic cells. Eur. J. Immunol., 32(9): 2588-97.

Lee, C. L.; Zandrew, F.; Gernand, L. and Davidsohn, I. 1968 Erythrophagocytosis in

vitro. I. Specificity and sensitivity. Vox. Sang., 15(3): 203-8.

Lehrman, M. A. and Hill, R. L. 1986 The binding of fucose-containing glycoproteins by

hepatic lectins. Purification of a fucose-binding lectin from rat liver. J. Biol. Chem.,

261(16): 7419-25.

Lin, Y.; Vandeputte, M. and Waer, M. 1997 Contribution of activated macrophages to the

process of delayed xenograft rejection. Transplantation, 64(12): 1677-83.

Lisowska, E.; Duk, M. and Dahr, W. 1980 Comparison of alkal-labile oligosaccharide

chains of M and N blood-group glycopeptides from human erythrocyte membrane.

Carbohydr. Res., 79(1): 103-13.

Loris, R. 2002 Principles of structures of animal and plant lectins. Biochim. Biophys.

Acta, 1572(2-3): 198-208.

Malykh, Y. N.; King, T. P.; Logan, E.; Kelly, D.; Schauer, R. and Shaw, L. 2003

Regulation of N-glycolylneuraminic acid biosynthesis in developing pig small

intestine. Biochem. J., 370(Pt 2): 601-7.

Malykh, Y. N.; Schauer, R. and Shaw, L. 2001 N-Glycolylneuraminic acid in human

tumours. Biochimie., 83(7): 623-34.

Marchesi, V. T. and Andrews, E. P. 1971 Glycoproteins: isolation from cellmembranes

with lithium diiodosalicylate. Science, 174(15): 1247-8.

119

Marchesi, V. T.; Tillack, T. W.; Jackson, R. L.; Segrest, J. P. and Scott, R. E. 1972

Chemical characterization and surface orientation of the major glycoprotein of the

human erythrocyte membrane. Proc. Natl. Acad. Sci. USA, 69(6): 1445-9.

Maruta, H. 1973 [Self-recognition in erythrophagocytosis (author's transl)].

Tanpakushitsu Kakusan Koso, 18(5): 639-47.

Maruta, H. and Mizuno, D. 1971 Selective recognition of various erythrocytes in

endocytosis by mouse peritoneal macrophages. Nat. New Biol., 234(51): 246-8.

May, A. P.; Robinson, R. C.; Vinson, M.; Crocker, P. R. and Jones, E. Y. 1998 Crystal

structure of the N-terminal domain of sialoadhesin in complex with 3' sialyllactose

at 1.85 A resolution. Mol. Cell., 1(5): 719-28.

Medzhitov, R. 2001 Toll-like receptors and innate immunity. Nat. Rev. Immunol., 1(2):

135-45.

Miyagawa, S.; Nakai, R.; Yamada, M.; Tanemura, M.; Ikeda, Y.; Taniguchi, N. and

Shirakura, R. 1999 Regulation of natural killer cell-mediated swine endothelial

cell lysis through genetic remodeling of a glycoantigen. J. Biochem. (Tokyo),

126(6): 1067-73.

Mohr, M.; Kolb, H.; Kolb-Bachofen, V. and Schlepper-Schafer, J. 1987 Recognition of

xenogeneic erythrocytes: the GalNAc/Gal-particle receptor of rat liver

macrophages mediates or participates in recognition. Biol. Cell, 60(3): 217-24.

Monsigny, M.; Mayer, R. and Roche, A. C. 2000 Sugar-lectin interactions: sugar clusters,

lectin multivalency and avidity. Carbohydr. Lett., 4(1): 35-52.

120

Morgan, R. O. and Fernandez, M. P. 1997a Annexin gene structures and molecular

evolutionary genetics. Cell Mol. Life Sci., 53(6): 508-15.

Morgan, R. O. and Pilar Fernandez, M. 1997b Distinct annexin subfamilies in plants and

protists diverged prior to animal annexins and from a common ancestor. J. Mol.

Evol., 44(2): 178-88.

Moriyasu, M.; Ito, A. and Omura, T. 1982 Purification and properties of arylsulfatase C

from rat liver microsomes. J. Biochem. (Tokyo), 92(4): 1189-95.

Morrissey, G. J.; Gravelle, D. R.; Lo, J. and Powe, J. E. 1992 Chromium-51 labeling of

sheep red blood cells. Lab. Anim. Sci., 42(1): 70-2.

Muchmore, E. A.; Diaz, S. and Varki, A. 1998 A structural difference between the cell

surfaces of humans and the great apes. Am. J. Phys. Anthropol., 107(2): 187-98.

Nelson, M. R. and Creutz, C. E. 1995 Combinatorial mutagenesis of the four domains of

annexin IV: effects on chromaffin granule binding and aggregating activities.

Biochemistry, 34(9): 3121-32.

Ni, Y. and Tizard, I. 1996 Lectin-carbohydrate interaction in the immune system. Vet.

Immunol. Immunopathol., 55(1-3): 205-23.

Nomoto, K.; Tsuda, S.; Minami, K. and Takeya, K. 1974 Immune response against

hamster erythrocytes in the low-responder mouse strains. IX. Phagocytic activity of

peritoneal macrophages of the high- and low-responder mouse strains. Jpn. J.

Microbiol., 18(3): 203-10.

121

Oldenborg, P. A.; Zheleznyak, A.; Fang, Y. F.; Lagenaur, C. F.; Gresham, H. D. and

Lindberg, F. P. 2000 Role of CD47 as a marker of self on red blood cells. Science,

288(5473): 2051-4.

Ouchi, E.; Seiji, K.; Sato, J.; Nomura, N. and Watabe, S. 1976 Erythrocyte rosette forming

activity of macrophage. Tohoku. J. Exp. Med., 118 Suppl: 199-205.

Pigault, C.; Follenius-Wund, A.; Lux, B. and Gerard, D. 1990 A fluorescence

spectroscopy study of the calpactin I complex and its subunits p11 and p36:

calcium-dependent conformation changes. Biochim. Biophys. Acta, 1037(1): 106-

14.

Pisano, A.; Redmond, J. W.; Williams, K. L. and Gooley, A. A. 1993 Glycosylation sites

identified by solid-phase Edman degradation: O-linked glycosylation motifs on

human glycophorin A. Glycobiology, 3(5): 429-35.

Platt, J. L. 2000 Acute vascular rejection. Transplant Proc., 32(5): 839-40.

Platt, J. L.; Lin, S. S. and McGregor, C. G. 1998 Acute vascular rejection.

Xenotransplantation, 5(3): 169-75.

Ravetch, J. V. and Lanier, L. L. 2000 Immune inhibitory receptors. Science, 290(5489):

84-9.

Raynal, P. and Pollard, H. B. 1994 Annexins: the problem of assessing the biological role

for a gene family of multifunctional calcium- and phospholipid-binding proteins.

Biochim. Biophys. Acta, 1197(1): 63-93.

122

Rees, M. A. 2003a Immunological effects of hepatic xenoperfusion. In: Clinical

Medicine and Clinical Veterinary Medicine. Cambridge, University of Cambridge:

458.

Rees, M. A.; Butler, A. J.; Chavez-Cartaya, G.; Wight, D. G.; Casey, N. D.; Alexander, G.;

Khuder, S. A.; White, D. J. and Friend, P. J. 2002a Prolonged function of

extracorporeal hDAF transgenic pig livers perfused with human blood.

Transplantation, 73(8): 1194-202.

Rees, M. A.; Butler, A. J.; Davies, H. F.; Bolton, E.; Wight, D. G.; Skepper, J.; White, D.

J. and Friend, P. J. 2002b Porcine livers perfused with human blood mount a

graft-versus-"host" reaction. Transplantation, 73(9): 1460-7.

Rees, M. A., Butler, A.J., Brons, G.M., Negus, M.C., Skepper, J.N., Friend, P.J. 2003b

Human complement regulatory mechanisms are able to inhibit complement-

mediated lysis of human erythrocytes by porcine complement. [Submitted].

Transplantation.

Rees, M. A., Butler, A.J., Negus, M.C., Davies, H.F.S., Friend, P.J. 2003c Evidence of

macrophage receptors capable of direct recognition of xenogeneic epitopes without

opsonization.[Submitted]. Transplantation.

Rosenberg, S. A. and Guidotti, G. 1968 The protein of human erythrocyte membranes. I.

Preparation, solubilization, and partial characterization. J. Biol. Chem., 243(8):

1985-92.

123

Rothlein, R. and Kim, Y. B. 1982 Porcine alveolar macrophages discriminate between self

and nonself in lectin-mediated cellular cytotoxicity. Cell Immunol., 68(2): 368-76.

Sakamaki, T.; Imai, Y. and Irimura, T. 1995 Enhancement in accessibility to

macrophages by modification of mucin-type carbohydrate chains on a tumor cell

line: role of a C-type lectin of macrophages. J. Leukoc. Biol., 57(3): 407-14.

Sandrin, M. S. and McKenzie, I. F. 1994 Gal alpha (1,3)Gal, the major xenoantigen(s)

recognised in pigs by human natural antibodies. Immunol. Rev., 141: 169-90.

Sandrin, M. S.; Vaughan, H. A.; Dabkowski, P. L. and McKenzie, I. F. 1993 Anti-pig

IgM antibodies in human serum react predominantly with Gal(alpha 1-3)Gal

epitopes. Proc. Natl. Acad. Sci. USA, 90(23): 11391-5.

Sarich, V. M. and Wilson, A. C. 1967 Rates of albumin evolution in primates. Proc. Natl.

Acad. Sci. USA, 58(1): 142-8.

Satake, M.; Korsgren, O.; Ridderstad, A.; Karlsson-Parra, A.; Wallgren, A. C. and Moller,

E. 1994 Immunological characteristics of islet cell xenotransplantation in humans

and rodents. Immunol. Rev., 141: 191-211.

Sato, C.; Kitajima, K.; Inoue, S. and Inoue, Y. 1998 Identification of oligo-N-

glycolylneuraminic acid residues in mammal-derived glycoproteins by a newly

developed immunochemical reagent and biochemical methods. J. Biol. Chem.,

273(5): 2575-82.

124

Satoh, A.; Hazuki, M.; Kojima, K.; Hirabayashi, J. and Matsumoto, I. 2000 Ligand-

binding properties of annexin from Caenorhabditis elegans (annexin XVI, Nex-1).

J. Biochem. (Tokyo), 128(3): 377-81.

Savage, A. V.; Koppen, P. L.; Schiphorst, W. E.; Trippelvitz, L. A.; Van Halbeek, H.;

Vliegenthart, J. F. and Van den Eijnden, D. H. 1986 Porcine submaxillary mucin

contains alpha 2----3- and alpha 2----6-linked N-acetyl- and N-glycolylneuraminic

acid. Eur. J. Biochem., 160(1): 123-9.

Schenkel-Brunner, H. 2000 In: Human Blood Groups: Chemical and Biochemical Basis

of Antigen Specificity, Springer-Verlag, Wien.

Schlepper-Schafer, J.; Friedrich, E. and Kolb, H. 1981 Galactosyl specific receptor on

liver cells: binding site for tumor cells. Eur. J. Cell Biol., 25(1): 95-102.

Schlepper-Schafer, J. and Kolb-Bachofen, V. 1988 Red cell aging results in a change of

cell surface carbohydrate epitopes allowing for recognition by galactose-specific

receptors of rat liver macrophages. Blood Cells, 14(1): 259-74.

Schlepper-Schafer, J.; Kolb-Bachofen, V. and Kolb, H. 1980 Analysis of lectin-

dependent recognition of desialylated erythrocytes by Kupffer cells. Biochem. J.

186(3): 827-31.

Sharland, A.; Patel, A.; Lee, J. H.; Cestra, A. E.; Saidman, S. and Waneck, G. L. 2002

Genetically modified HLA class I molecules able to inhibit human NK cells

without provoking alloreactive CD8+ CTLs. J. Immunol., 168(7): 3266-74.

125

Sheikh, S.; Parhar, R.; Kwaasi, A.; Collison, K.; Yacoub, M.; Stern, D. and Al-Mohanna,

F. 2000 Alpha-gal-independent dual recognition and activation of xenogeneic

endothelial cells and human naive natural killer cells. Transplantation, 70(6): 917-

28.

Sherblom, A. P. and Bourassa, C. R. 1983 Specificity of submaxillary gland

sialyltransferases. Biochim. Biophys. Acta, 761(1): 94-102.

Siebert, H. C.; Andre, S.; Reuter, G.; Kaptein, R.; Vliegenthart, J. F. and Gabius, H. J.

1997 Comparison between intact and desialylated human serum amyloid P

component by laser photo CIDNP (chemically induced dynamic nuclear

polarization) technique: an indication for a conformational impact of sialic acid.

Glycoconj. J., 14(8): 945-9.

Silverstein, A. M. 1989 In: A history of Immunology. San Diego, Academic Press, Inc.

Sohma, H.; Creutz, C. E.; Gasa, S.; Ohkawa, H.; Akino, T. and Kuroki, Y. 2001

Differential lipid specificities of the repeated domains of annexin IV. Biochim.

Biophys. Acta, 1546(1): 205-15.

Sohma, H.; Creutz, C. E.; Saitoh, M.; Sano, H.; Kuroki, Y.; Voelker, D. R. and Akino, T.

1999 Characterization of the Ca2+-dependent binding of annexin IV to surfactant

protein A. Biochem. J., 341 ( Pt 1): 203-9.

Springer, G. F. 1989 Tn epitope (N-acetyl-D-galactosamine alpha-O-serine/threonine)

density in primary breast carcinoma: a functional predictor of aggressiveness. Mol.

Immunol., 26(1): 1-5.

126

Stewart, J. and Weir, D. M. 1989 Carbohydrates as recognition molecules in macrophage

activities. J. Clin. Lab. Immunol., 28(3): 103-8.

Suzuki, N.; Yamamoto, K.; Toyoshima, S.; Osawa, T. and Irimura, T. 1996 Molecular

cloning and expression of cDNA encoding human macrophage C-type lectin. Its

unique carbohydrate binding specificity for Tn antigen. J. Immunol., 156(1): 128-

35.

Tait, J. F.; Sakata, M.; McMullen, B. A.; Miao, C. H.; Funakoshi, T.; Hendrickson, L. E.

and Fujikawa, K. 1988 Placental anticoagulant proteins: isolation and comparative

characterization four members of the lipocortin family. Biochemistry, 27(17):

6268-76.

Takasaki, S.; Yamashita, K. and Kobata, A. 1978 The sugar chain structures of ABO

blood group active glycoproteins obtained from human erythrocyte membrane. J.

Biol. Chem., 253(17): 6086-91.

Thomas, D. B. and Winzler, R. J. 1969 Structural studies on human erythrocyte

glycoproteins. Alkali-labile oligosaccharides. J. Biol. Chem., 244(21): 5943-6.

Tiemeyer, M.; Brandley, B. K.; Ishihara, M.; Swiedler, S. J.; Greene, J.; Hoyle, G. W. and

Hill, R. L. 1992 The binding specificity of normal and variant rat Kupffer cell

(lectin) receptors expressed in COS cells. J. Biol. Chem., 267(17): 12252-7.

Tomita, M.; Furthmayr, H. and Marchesi, V. T. 1978 Primary structure of human

erythrocyte glycophorin A. Isolation and characterization of peptides and complete

amino acid sequence. Biochemistry, 17(22): 4756-70.

127

Tomita, M. and Marchesi, V. T. 1975 Amino-acid sequence and oligosaccharide

attachment sites of human erythrocyte glycophorin. Proc. Natl. Acad. Sci. USA,

72(8): 2964-8.

Tsuji, T.; Irimura, T. and Osawa, T. 1981 The carbohydrate moiety of band 3 glycoprotein

of human erythrocyte membranes. Structures of lower molecular weight

oligosaccharides. J. Biol. Chem., 256(20): 10497-502.

Tsujii-Hayashi, Y.; Kitahara, M.; Yamagaki, T.; Kojima-Aikawa, K. and Matsumoto, I.

2002 A potential endogenous ligand of annexin IV in the exocrine pancreas.

Carbohydrate structure of GP-2, a glycosylphosphatidylinositol-anchored

glycoprotein of zymogen granule membranes. J. Biol. Chem., 277(49): 47493-9.

Varki, A. 1997 Sialic acids as ligands in recognition phenomena. Faseb. J., 11(4): 248-55.

Varki, A. 2001 Loss of N-glycolylneuraminic acid in humans: Mechanisms,

consequences, and implications for hominid evolution. Am. J. Phys. Anthropol.,

Suppl 33: 54-69.

Vasta, G. R. 1991 In: The multiple biological roles of invertibrate lectins: their

participation in nonself recognition mechanisms. Boca Raton, CRC Press.

Vasta, G. R. and Ahmed, H. 1996 Animal lectins as cell surface receptors: current status

for invertebrate species. Prog. Mol. Subcell Biol., 17: 158-82.

Vasta, G. R.; Ahmed, H.; Fink, N. E.; Elola, M. T.; Marsh, A. G.; Snowden, A. and Odom,

E. W. 1994 Animal lectins as self/non-self recognition molecules. Biochemical

128

and genetic approaches to understanding their biological roles and evolution. Ann.

NY Acad. Sci., 712: 55-73.

Vaysse, J.; Gattegno, L.; Bladier, D. and Aminoff, D. 1986 Adhesion and

erythrophagocytosis of human senescent erythrocytes by autologous monocytes

and their inhibition by beta-galactosyl derivatives. Proc. Natl. Acad. Sci. USA,

83(5): 1339-43.

Vomel, T.; Hager, K. and Platt, D. 1988 Clearance of heterologous, homologous and

damaged homologous erythrocytes by the isolated perfused rat liver. Vet. Immunol.

Immunopathol., 18(4): 361-8.

Vomel, T.; Muller, N. and Platt, D. 1985 Phagocytosis of erythrocyte membranes (white

ghosts) from different species by the isolated perfused rat liver. Vet. Immunol.

Immunopathol., 9(3): 273-82.

Wallgren, A. C.; Karlsson-Parra, A. and Korsgren, O. 1995 The main infiltrating cell in

xenograft rejection is a CD4+ macrophage and not a T lymphocyte.

Transplantation, 60(6): 594-601.

Wasylnka, J. A.; Simmer, M. I. and Moore, M. M. 2001 Differences in sialic acid density

in pathogenic and non-pathogenic Aspergillus species. Microbiology, 147(Pt 4):

869-77.

Watanabe, M. and Russell, A. S. 1980 Increased phagocytosis of sheep red cells by

macrophages from New Zealand mice. J. Reticuloendothel. Soc., 27(5): 455-9.

129

Watkins, W. M. 2001 The ABO blood group system: historical background. Transfus.

Med., 11(4): 243-65.

Weis, W. I.; Taylor, M. E. and Drickamer, K. 1998 The C-type lectin superfamily in the

immune system. Immunol. Rev., 163: 19-34.

Wuttke, N. J.; Macardle, P. J. and Zola, H. 1997 Blood group antibodies are made by

CD5+ and by CD5- B cells. Immunol. Cell Biol., 75(5): 478-83.

Xia, G.; Ji, P.; Rutgeerts, O. and Waer, M. 2000 Natural killer cell- and macrophage

mediated discordant guinea pig-->rat xenograft rejection in the absence of

complement, xenoantibody and T cell immunity. Transplantation, 70(1): 86-93.

Xu, H.; Gundry, S. R.; Hancock, W. W.; Matsumiya, G.; Zuppan, C. W.; Morimoto, T.;

Slater, J. and Bailey, L. L. 1998 Prolonged discordant xenograft survival and

delayed xenograft rejection in a pig-to-baboon orthotopic cardiac xenograft model.

J. Thorac. Cardiovasc. Surg., 115(6): 1342-9.

Yamamoto, K.; Ishida, C.; Shinohara, Y.; Hasegawa, Y.; Konami, Y.; Osawa, T. and

Irimura, T. 1994 Interaction of immobilized recombinant mouse C-type

macrophage lectin with glycopeptides and oligosaccharides. Biochemistry, 33(26):

8159-66.

Yoshima, H.; Furthmayr, H. and Kobata, A. 1980 Structures of the asparagine-linked

sugar chains of glycophorin A. J. Biol. Chem., 255(20): 9713-8.

Zaks, W. J. and Creutz, C. E. 1991 Ca(2+)-dependent annexin self-association on

membrane surfaces. Biochemistry, 30(40): 9607-15.

130

ABSTRACT

This dissertation examines the interaction of porcine liver macrophages, Kupffer

cells, and human erythrocytes. Previous research in our laboratory revealed that when

porcine livers were perfused with human blood, porcine Kupffer cells bound and

phagocytosed circulating human but not porcine erythrocytes. This binding was

independent of antibody or complement opsonization, suggesting direct recognition of

human erythrocytes by porcine Kupffer cells. There is substantial evidence that Kupffer

cells from other species have germline lectin receptors that specifically recognize

xenogeneic carbohydrate epitopes. Thus, our initial hypothesis was: porcine Kupffer cells

recognize human erythrocytes via a lectin-carbohydrate interaction.

We used a quantitative assay in which 51Cr-labeled erythrocytes bound to porcine

Kupffer cells were detected. This assay allowed for the manipulation of the conditions under which the cells were incubated in order to elucidate the underlying mechanism of binding. Various techniques were used to isolate proteins and glycoproteins, such as

detergent solubilization, gel chromatography, and affinity chromatography. The latest

advances in mass spectrometry were utilized to reveal protein identity through tandem MS

sequencing.

We provide evidence that N-acetylneuraminic acid, borne on human glycophorin

A, is the ligand being recognized by porcine Kupffer cells. This interaction is calcium

dependent and can be inhibited with the N-acetylneuraminic acid monosaccharide or

neuraminyl lacotses bearing α-2,3 and α-2,6 linkage. We also have evidence to suggest

131

that the lectin annexin IV on the surface of porcine Kupffer cells is responsible at least in part, for the recognition of human erythrocytes.

These findings support our initial hypothesis that a lectin-carbohydrate interaction is responsible for the recognition of xenogeneic human erythrocytes by porcine Kupffer cells. We propose that these data could be used to improve the performance of extracorporeal xenoperfusion. More importantly, these data support a new hypothesis that innate immune cellular recognition of xenogeneic carbohydrates by lectin receptors plays an important role in xenograft rejection.