1.1 Coeliac disease

1. Introduction

1.1 Coeliac disease

Gluten sensitive enteropathy, commonly known as coeliac disease, is a permanent intolerance to ingested proteins found in wheat, barley and rye (Sollid 2000). It is a chronic condition with an inflammatory immune response to gluten or related proteins that centres on the proximal small intestine (Koning et al. 2005). In Western populations it is the most common inflammatory disease of the small intestine (Tjon, van Bergen & Koning 2010).

Coeliac disease is often described as an autoimmune disease as patients produce autoantibodies whilst the disease is in an active stage (Selimoglu & Karabiber 2010).

Despite this, it is not the autoantigen that drives disease progression, nor is it conclusively known what role the autoantibodies even have in the pathogenesis of coeliac disease (Lindfors, Mäki & Kaukinen 2010).

1.1.1 Symptoms of coeliac disease

Coeliac disease has varying symptoms, especially when debuting in adults. Cachexia, deficiency diseases due to malabsorption, gastrointestinal cramps, pain and diarrhoea are the most commonly reported by patients. Symptoms also vary with age, with children having the classical symptoms of diarrhoea, swollen abdomens and failure to thrive, whilst adults can also have symptoms such as dermatitis herpetiformis, anaemia, osteoporosis and ataxia (Tjon, van Bergen & Koning 2010; Volta & Villanacci 2011).

Microscopically, coeliac disease is characterized by histological changes of the affected tissue, with a loss of cellular differentiation and tissue organization (Volta &

Villanacci 2011). It results in villous atrophy, which gives a disease-typical presentation

of shortened or absent villi and crypt hyperplasia (Volta & Villanacci 2011). Patients with

untreated coeliac disease also have a compromised barrier function in their small intestine

(Caja et al. 2011). Under the microscope tissue biopsies from the small intestine appear smooth, lacking the undulating appearance of healthy tissue. Mucosal lesions appear in the duodenum and jejunum of patients, which negatively affect nutrient absorption (Trier 1991).

The duodenum and jejunum are infiltrated by a large lymphocyte population in the epithelium (intraepithelial lymphocytes, IELs) and lamina propria (Jabri & Sollid 2009).

An abnormal T-cell invasion is characteristic of coeliac disease and a rise in IELs with γδ T-cell receptors (TCRs) can be used as a diagnostic aid (Sollid & Lundin 2003).

1.1.2 Treatments

Classic coeliac disease has currently no cure other than a lifelong avoidance of gluten (Koning et al. 2005). Adherence to such a diet is effective, with disease specific antibodies decreasing and becoming negative after a year (Caja et al. 2011). Overt symptoms associated with the disease often recede and disappear altogether (Selimoglu & Karabiber 2010). Some patients do not respond to a gluten free diet and are diagnosed as having refractory sprue (Ryan & Kelleher 2000). Unlike coeliac disease, refractory sprue is a diagnosis of exclusion and is only given when symptoms cannot be explained by any other means.

1.2 Diagnosis

1.2.1 Small intestine biopsy

Before the advent of immune-based diagnostic methods, histochemical staining of affected mucosa of the small intestine was the gold standard and largely remains so now (Koning et al. 2005). Histochemical staining also allows for detection of IELs.

Problems with intestinal biopsy arise with incorrect technique: tissue must be

correctly orientated, representative of the diseased tissue as a whole and be taken from at

least four areas (Volta & Villanacci 2011). This is problematic when lesions are irregular and spaced out in the small intestine. Healthy tissue then lies next to damaged tissue and technique becomes even more important. Patient suffering must also be taken into consideration.

Tissue biopsies also have a rather narrow range of diagnosis. Patients with subclinical coeliac disease, atypical coeliac disease or silent coeliac disease all have minimal or no histological changes of the small intestine (Volta & Villanacci 2011).

Diagnosis with tissue biopsy alone would thus give potential false negatives due to the aforementioned appearance of “patchwork” lesions.

1.2.2 Serology

It was known already in the 1960s that coeliac disease patients produced antibody to wheat protein though the diagnostic value of this information was at first small. It was only with the discovery that patients were forming antibody to α-gliadin and these were unique to coeliac disease patients that it became useful (Leffler & Schuppen 2010; Barbato et al.

2011). In 1983 O’Farrelly et al proved that an enzyme linked immunosorbent assay (ELISA) with α-gliadin as the antigen allowed them to differentiate between adult sufferers of coeliac disease and non-sufferers. The discovery of the autoantibody anti-tTG resulted in a new range of diagnostic techniques, simpler and cheaper than biopsies.

Serological tests for coeliac disease enjoy ≥95% sensitivity and specificity, which puts it in league with diagnostic methods for infectious diseases (Leffler & Schuppen 2010). Anti-tTG autoantibodies also appear before changes to villus morphology (Kurpa et al. 2010). Importantly they only detect an ongoing reaction by the immune system to gluten-derived antigen; levels of autoantibody normalize with a gluten free diet (Leffler &

Schuppen 2010). This allows for testing of dietary compliance but can also give false

negatives.

1.2.3 Anti-tTG test specificity and sensitivity

Anti-tTG detection tests are used in two capacities: ELISAs, where patient serum is tested against tTG, and endomysial based tests where patient serum is tested against primate endomysial tissue (Leffler & Schuppen 2010). Of these two, endomysial antibody (EMA) testing is older, having been developed during the mid-1980s before the discovery that tTG was the autoantigen of coeliac disease. ELISA based tests using recombinant tTG is currently the most common due to both its accuracy and relative cost and easiness: EMA testing is immunofluorescence based and requires monkey oesophagus or human umbilical cord tissue (Leffler & Schuppen 2010). However, it does have a slightly higher specificity, with some studies approaching 100% specificity in adults and >89% in children (Rostrom et al. 2005). With respect to sensitivity, results are more varied, with numbers ranging from 75-98%, and no higher than that achieved with recombinant tTG based ELISAs (Rostrom et al. 2005; Leffler & Schuppen 2010). Age and smoking may affect EMA detection sensitivity negatively thus accounting for the varied results (Boger et al. 2007).

Sensitivity for both tTG ELISAs and EMA testing is lower when testing sera from patients with subclinical or silent coeliac disease where there is little lesion development (Rostrom et al. 2005). This makes screening for coeliac disease difficult as sensitivity is paramount (Boger et al. 2007).

1.3 Gluten: the driving force of coeliac disease

Coeliac disease requires a genetic component that makes the patient susceptible to coeliac

disease but is not enough in itself to develop it (Koning et al. 2005). A secondary

environmental factor is what initiates pathogenesis in patients and coeliac disease is quite

unique amongst autoimmune diseases in that said environmental factor is known: gliadin, a

peptide derived from gluten (Caja et al. 2011).

Gluten is the protein component of wheat, barley and rye and gives dough its elasticity (Green & Cellier 2007). It consists of numerous different gliadin and glutenin polypeptides and it is the alcohol-soluble gliadin fraction that is toxic to coeliac disease sufferers (Green & Cellier 2007; Sollid 2002). Gliadin can be subdivided into three groups, α, γ and ω, according to their sequence (Sollid 2000). Of these, α-gliadins are recognized most frequently by coeliac disease patients (Sollid 2000).

Gluten proteins are rich in proline and glutamine, which make them resistant to digestion by gastric, pancreatic and intestinal brush border membrane proteases, as few mammalian proteases can hydrolyse amine bonds adjacent to proline residues (Jabri &

Sollid 2009; Green & Cellier 2007; Sollid 2002). As such, gliadin and glutenin often arrive as long fragments of 10-50 residues in the gut lumen, epitopes intact.

Normally such fragments are too large to cross the intestinal barrier; it was for a long time unclear exactly how coeliac disease started since the environmental antigen was theoretically separated from the immune system. However, studies have shown that gliadin itself is capable of activating cells of the innate immune system, such as monocytes, dendritic cells and macrophages (Jabri & Sollid 2009). These, in turn, promote CD4

and CD8

T-cell activity. An α-gliadin peptide was shown to be capable of upregulating of interleukin-15 (IL-15) production, a pro-inflammatory cytokine, by epithelial cells (Koning et al. 2005). IL-15, along with interferon-α and interferon-γ, is elevated in the intestinal epithelium and lamina propria of coeliac disease patients (Mention et al. 2003; Nilsen et al.

1998; Sabatino et al. 2007). Gliadin also appears to impair the gut’s barrier function, which would explain the escalation of the disease (Lammers et al. 2008). Figure 1.1 summarises these effects.

Just as gluten appears to be what initiates the inflammatory immune response, it is

also what drives coeliac disease; patients who remove gluten from their diet cease to have

any of the anti-gluten or autoantibody that characterize the disease and neither reappear whilst a gluten free diet is maintained (Leffler & Schuppen 2010). Intestinal lesions heal and patients will almost always experience a complete remission.

Figure 1.1: The effects of gliadin upon the immune system and intestinal barrier. By increasing the expression of IL-15 by epithelial cells and directly activating CD4⁺ T-cells and cells of the innate immune system e.g. macrophages and antigen presenting cells, gliadin activates intestinal epithelial lymphocytes.

Whilst gliadin’s effect on the innate immune system is important, it is the

recruitment of cellular and humoral arms of the adaptive immune system that causes

intestinal lesions. This requires the presentation of gliadin by antigen presenting cells

(APCs). In coeliac disease patients this has been shown to tie into two other factors of the

disease: genetic susceptibility and antigen modification.

1.4 Genetic susceptibility to coeliac disease

Epidemiological studies showed first degree biological relatives had a higher prevalence compared to the general population (Koning et al. 2005; Abadie et al. 2011). This strongly suggested there was a genetic component to the disease. Studies narrowed the field of suspects to the major histocompatibility (MHC) class II genes, a class of molecules found on dedicated APCs that bind extracellular antigen to present to CD4

T-cells. This discovery gave two things: firstly, it indicated which class of cells was involved in coeliac disease pathology; secondly, as gliadin-derived antigen bound MHC II rather ineffectively some form of antigen modification was occurring.

MHC II molecules bind peptide fragments 10-30 amino acids in length (Sollid &

Lundin 2003). Of those, only 9 amino acids end up in the peptide binding groove of the MHC II molecule. Of these 9, certain amino acids occupy anchor positions. Each anchor position has a preferred amino acid/s, a peptide binding motif, which promotes binding of the antigen to the MHCII peptide binding groove. The peptide binding motif depends on the MHC II allele an individual has.

Numbers vary between studies but generally speaking almost all coeliac disease sufferers have the same HLA-DQ alleles: >90% are HLA-DQ2

with the remaining patients possessing HLA-DQ8 (Tollefsen et al. 2006). Alleles encoding either HLA-DQ2 or HLA-DQ8 are necessary to develop coeliac disease, though possessing them is by itself insufficient to develop the disease (Green & Cellier 2007; Koning et al. 2005).

Furthermore, the necessity of having either HLA allele is not due to T-cells being restricted to HLA-DQ2/8 but the antigen being restricted to these alleles (Sollid 2000).

1.4.1 Peptide binding motifs

The discovery that coeliac disease sufferers shared a genetic profile was complicated by

the fact that both HLA-DQ2 and DQ8 prefer to bind proteins with negatively charged

amino acids in certain anchor positions (Sollid 2000). DQ8 shows a preference for negatively charged amino acids in positions P1, P4 and P9, as well as a preference for T- cell receptors with a negative charge to their CDR3β loop (Hovhannisyan et al. 2008).

DQ2 has 5 anchor residues in positions of P1, P4, P6, P7 and P9 (Tollefsen et al. 2006). P4 and P7 prefer aspartic acid or glutamic acid. P6 prefers aspartic acid, glutamic acid or proline. With the exception of proline, these are not amino acids common to gluten’s primary makeup.

This raised the question of how gliadin-derived peptide fragments could bind the peptide binding groove strongly enough to activate immune cells despite lacking the preferred peptide binding motif needed.

1.5 Antigen modification

At this point it is important to keep in mind what gliadin has had to overcome to create a pathological situation: firstly, it had to compromise the intestinal barrier and cross;

secondly it had to make the innate immune system react to it; thirdly, it must now recruit the adaptive immune system as well. To do this it must augment its ability to bind the peptide binding groove of an APC’s MHC II. Therefore, the gliadin fragments must undergo modification to attain the correct peptide binding motif for HLA-DQ2/8.

1.5.1 Transglutaminase

Tissue transglutaminase (tTG), also known as TG2, is a ubiquitous enzyme consisting of 687 amino acids (Fesus & Piacentini 2002). It has multiple functions in the body, the primary in vivo one being the crosslinking of lysine and glutamine in proteins (Fesus &

Piacentini 2002). Besides that, tTG’s multiple other duties can be grouped into 3 distinct

functions: (1) protein modification via amine incorporation, deamidation or acting as an

isopeptidase; (2) acting as a G protein in transmembrane signalling; (3) promoting cell

interaction with the extracellular matrix by mediating cell surface adhesion, integrin-

fibronectin (Fn) interactions and crosslinking the extracellular matrix’s proteins (Fesus &

Piacentini 2002).

tTG consists of 4 domains plus a Ca

binding site and its antipode, a GTP/GDP binding site (Fesus & Piacentini 2002). Binding of Ca

changes the enzyme structure and opens a channel that allows substrates access to the enzyme active site (compare open and closed configurations of tTG in Fig. 1.2). It is expressed continually by smooth muscle, meaning it is found in high levels in the digestive tract. Importantly for coeliac disease pathogenesis, in the gut tTG is located extracellularly underneath the gut epithelium (Sollid

& Lundin 2003).

Figure 1.2: 3-dimensional structure of tissue transglutaminase. tTG is shown in (A) closed, inactive conformation when bound by GDP; (B) open, active form when bound by Ca²⁺. Bound within the catalytic core is a gluten-like inhibitor. Shown are N-terminal β-sandwich (blue) and two C-terminal β-barrels (orange and red). The catalytic core is green (Pinkas et al. 2007).

Figure removed due to copyright.

Figure removed due to

copyright.

In regards to coeliac disease, tTG is important for two reasons, of which one is its role in antigen modification. In an HLA-DQ2

or HLA-DQ8

individual, modification of gliadin fragments by tTG creates immunogenic epitopes that trigger coeliac disease (Anderson et al. 2000). Specifically, tTG deamidates glutamine to glutamic acid, thus allowing gliadin to achieve the peptide binding motif required to bind HLA-DQ2/8, (Arentz-Hansen et al. 2000; Molberg et al. 1998). Figure 1.3 shows this antigen modification for HLA-DQ2. The rate of this reaction increases when pH is lower than 7.3 (Sollid 2002). As gluten contains little variation in its amino acid content, its structure is repetitious. This increases the likelihood that deamidation by tTG will produce the correct peptide binding motif.

Figure 1.3: Modification of gluten by tTG allows gliadin derived antigen to achieve the correct binding motif for HLA-DQ2. By changing neutral amino acids (blue spheres) into negative amino acids (yellow spheres) in positions P4, P6 and P7, gliadin-derived antigen (orange) can bind the peptide binding groove of HLA-DQ2 (pink) tighter, which allows for activation of CD4⁺ T-cells via the TCR (green).

However, a question is raised as to how tTG is activated, as it is not constitutively active in the gut mucosa (Siegel et al. 2008). Low Ca

levels in the cytosol can explain the inactivity of intracellular tTG but not extracellular tTG which exists in a Ca

abundant environment. Tissue injury, brought on by inflammation and/or viral infections, can activate tTG (Siegel et al. 2008; Jabri & Sollid 2009). Tissue injury can also release intracellular tTG. Along with the disruption of Ca

homeostasis caused by cell lysis, this will result in activated tTG in the gut epithelium. It is unknown if gluten in itself can cause sufficient tissue injury to activate tTG.

1.6 Lymphocyte recruitment

Thanks to the antigen modification of gliadin fragments by tTG, the adaptive immune system can now come into play. As mentioned earlier, the discovery that almost all coeliac disease patients shared the same HLA-DQ allele strongly indicated T-cells must play a vital part in pathogenesis. It is now known that T-cell invasion and subsequent reaction to gluten in the small intestine causes the gut lesions characteristic of coeliac disease (Abadie et al. 2011; Sollid & Lundin 2003). A rise in IELs consisting of CD4

T cells with γδ TCRs and CD8

T cells with αβ TCRS in the lamina propria and epithelium is a characteristic of coeliac disease (Gianfrani, Auricchio & Troncone 2005).

Furthermore, as mentioned earlier, α-gliadin is capable of inducing IL-15 production,

a pro-inflammatory cytokine that also acts as a growth factor for IELs (Gianfrani,

Auricchio & Troncone 2005). IL-15 also favours TCR independent activation of IELs

(Mention et al. 2003; Meresse et al. 2006). Gluten an also bind TCRs with a negative

CDRβ and activate CD4

T-cells directly (Jabri & Sollid 2009). This activation would be

small in scale but enough to produce a tTG activating inflammation, which would in turn

result in antigen modification and large scale T-cell activation.

1.6.1 CD4

T-cells

Disrupted homeostasis of IELs is characteristic of the disease (Mention et al. 2003).

Gliadin specific, HLA-DQ2 restricted CD4

T-cells are found in coeliac disease patients’

guts but not healthy individuals (Molberg et al. 1997). By deamidating glutamine to glutamic acid, gliadin derived antigen fragments can now bind the MHC II peptide binding groove much tighter. Thus antigen modification by tTG enables activation of gliadin reactive CD4

T-cells by APCs (fig. 1.4) (Molberg et al. 1998). The T-cells in question react to a number of different epitopes on gliadin, with some epitopes more common amongst coeliac disease sufferers than others (Sollid & Lundin 2003). In general, α-gliadin derived epitopes are more common than γ- or ω-gliadin or glutenin derived ones (Sollid 2000).

Polarization of CD4

T-cells depends on many factors, amongst them the cytokine milieu, the APC and the antigen itself (Nilsen et al. 1998). The gliadin-specific CD4

T- cells found in coeliac disease patients predominantly possess a Th1/Th0 cytokine profile and produce pro-inflammatory cytokines, amongst them interferon-γ and IL-21 (Gianfrani, Auricchio & Troncone 2005).

Figure 1.4: From gluten to activation of CD4⁺ T-cells. Gluten is broken down to large fragments of gliadin and glutenin fragments. Gliadin is further modified by tissue transglutaminase to allow gliadin to achieve the optimal peptide binding motif for the HLA-DQ2/8 binding groove. This allows activation of CD4⁺ T-cells.

As studies continue it appears CD4

T-cells orchestrate the immune response that ultimately creates the gut lesions and autoantibodies of coeliac disease (Sollid & Lundin 2003). CD4

T-cells have the ability to activate the humoral and cytotoxic cells of the adaptive immune system and their activation can have grave consequences. Particularly their ability to produce interferon-γ seems to be important, indicating that the immune processes triggered and/or affected by the interferon-γ are of particular interest.

1.6.2 CD8

T-cells

The intestinal epithelium naturally contains CD8

T-cells whose function is to eliminate infected epithelial cells. These lymphocytes proliferate early in coeliac disease and their presence correlates with the tissue destruction (Jabri & Sollid 2009). As CD8

T-cells have such destructive powers their “license to kill” is heavily regulated via numerous TCR- MHC I and co-stimulatory molecule interactions.

Initially CD8

T-cells were largely ignored as no gluten specific CD8

lymphocytes could be found in coeliac disease patients (Meresse et al. 2006). However, the inflammatory environment CD4

T-cells create via their cytokine production, along with other factors such as IL-15, allows for the activation of cytotoxic CD8

T-cells independent of their TCR and MHC class I (Meresse et al. 2006; Meresse et al. 2004).

Especially IL-15 is thought to have a vital role in the direct activation of cytolytic activity regardless of TCR specificity (Meresse et al. 2004).

One of the results of this activity is the release of intracellular tTG and disruption of Ca

homeostasis. This may play a role in antigen modification.

1.6.3 B-lymphocytes and coeliac disease antibodies

Having recruited T-cells we now turn to B-cells. Extracellular deposits of IgA can be

found in the proximal small intestine mucosa of coeliac disease patients (Caja et al. 2011).

These are located in the basement membrane, crypt epithelium and blood vessel walls of the mucosa.

Coeliac disease sufferers have two disease-specific antibodies: anti-gliadin antibody (AGA) and anti-tTG autoantibody (Sollid & Khosla 2005). AGA are antibody against the tTG modified gliadin peptides and are a logical result of the immune reactions in play during coeliac disease (O'Farrelly et al. 1983): one of the more important functions of CD4

T-cells is to augment the humoral immune response by expediting plasma cell proliferation. They also appear early, before the development of lesions (Kurpa et al.

2010). AGA can be both IgG and IgA class (Leffler & Schuppen 2010).

The antibodies produced during coeliac disease are thought to be involved in the tissue remodelling, though the exact mechanism is still lacking details. Studies have shown that coeliac disease antibodies can induce cell proliferation, inhibit differentiation, and negatively affect gut barrier function and the vascular system (Caja et al. 2011). Which of these are caused by AGA and which by autoantibody is unknown.

1.7 From inflammatory disorder to autoimmune disease

One characteristic of coeliac disease is the production of autoantibody against the patient’s

own tTG, turning it from a chronic inflammatory disorder into an autoimmune one

(Schuppan et al. 1998; Dieterich et al. 1997). Exactly how this breakdown of self-tolerance

occurs is unknown. One theory is that anti-tTG is produced as a side effect of

internalization of gliadin by phagocytic APCs whilst modifying the peptides (Lindfors,

Mäki & Kaukinen 2010; Sollid & Lundin 2003). This hapten carrier complex model has

the advantage of explaining why tTG deamidates gliadin instead of incorporating amine

groups: deamidation occurs more frequently below pH 7.3 such as the first stage of

endocytosis (Sollid & Lundin 2003; Sollid 2002).

The difference between coeliac disease and other autoimmune diseases is that autoantibody production is tied to the presence of gluten: patients on a gluten free diet are eventually free of both AGA and anti-tTG (Lindfors, Mäki & Kaukinen 2010). It is worth noting the strong ties between gluten and coeliac disease pathogenesis (fig. 1.5). This is due to the effect gluten has on CD4

T-cells. The removal of gluten sees a halt to disease progression as it causes a cessation of CD4

T-cell activity. Without helper T-cells, antibody production by B-cells ceases as well.

Figure 1.5: From gluten to antibody production. B-cells require helper T-cells for activation. CD4⁺ T-cells are in turn activated by tTG-modified gliadin-presenting APCs. Removal of gluten from the diet removes the

“engine” that drives the immune response causing a cessation of disease progression and regression of pathological tissue changes.

It is also unknown what role anti-tTG autoantibodies have in pathogenesis, if any. Studies are hampered by the ubiquity of the enzyme in question. It is thought they may be responsible for the breakdown in angiogenesis (Caja et al. 2010), intestinal crypt cell differentiation (Halttunen & Mäki 1999) and epithelial cell proliferation (Barone et al.

Gastroenterology). Antibodies produced against other variants of transglutaminase, such as

TG6, may also result in some of the rarer symptoms of coeliac disease e.g. ataxia

(Lindfors, Mäki & Kaukinen 2010). Anti-tTG autoantibodies also interfere with enzyme activity (Byrne et al. 2010).

1.8 A modified ELISA

Earlier on it was stated that due to the excellent correlation between anti-tTG and coeliac disease anti-tTG is used in screening more often than AGA (Leffler & Schuppen 2010).

Since it is preferable that coeliac disease patients be diagnosed before the appearance of intestinal lesions, anti-tTG based tests should be as sensitive as possible.

tTG exposed to the extracellular environment will bind Fn with high affinity and virtually instantaneously (Jeong et al. 1995; Turner & Lorand 1989). The interaction is Ca

independent and non-covalent. This instantaneous complexing is important to prevent unwanted enzymatic activity when cellular damage releases intracellular tTG (Turner &

Lorand 1989).

Using this, a modified ELISA was developed wherein, instead of tTG being bound directly to the ELISA plate, it was instead attached to Fn (Teesalu et al. 2009). His-tagged recombinant tTG and purified human Fn was used. It was hoped the ELISA would thereby achieved the same orientation of the tTG as found in EMA testing by mirroring the in vivo conformation found extracellularly. More epitopes would be displayed and this would make the test more sensitive whilst retaining the cost effectiveness of ELISA testing.

1.9 The aim of this project The aims of this project were two-fold:

1) to determine whether recombinant Glutathione S-transferase-tagged (GST-

tagged) tTG (as opposed to His-tagged tTG as used by Teesalu et. al.) would bind

Fn. The Fn binding region of tTG is still in question though studies strongly suggest

it is located at the start of the N-terminal region (Jeong et al. 1995). The first seven

residues are thought to be the most important. GST tags are larger than His tags and could potentially block the Fn-binding region;

2) to compare the modified ELISA with the conventional tTG assay. Specifically,

was the modified ELISA more sensitive? As stated earlier, sensitivity is important

when screening populations. It is also a future hope to develop a method sensitive

enough to retire the need for tissue biopsies, both from a cost point of view as well as

for the benefit of the patient.

2. Materials and Methods

2.1 Affi-Gel 15

0.1.1 Coupling of fibronectin to Affi-Gel 15

200µL cold Fn, at a concentration of 0.5µg/µL, was added to 800µL cold 50mM 4-(2- hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), pH 7.5, and kept on ice. Affi-gel slurry was vortexed for uniform suspension of beads and 1mL was transferred to a 12mL tube. Gel was washed with 3mL cold distilled H

O and then centrifuged at 3000rpm for 2min at 4°C. Supernatant was removed with a glass Pasteur pipette and discarded. 950µL of the Fn-HEPES solution was added to the gel slurry and allowed to incubate for 4hrs at 4°C with agitation. Gel slurry was then centrifuged at 3000rpm for 2min. Supernatant was frozen for later comparison to Fn-HEPES solution on SDS-PAGE gels. 100µL blocking buffer (1M ethanolamine) was added and incubated for 1hr at 4°C with agitation. After incubation 1mL HEPES was added and the solution centrifuged at 7000rpm for 3min. The supernatant was discarded and washing step repeated. Gel slurry was then resuspended in 500µL and stored at 4°C.

0.1.2 Modified ELISA to confirm attachment to Affi-Gel 15

Affi-gel beads were blocked with blocking buffer for 1hr at 4°C. 50µL blocked and 50µL

Fn-coupled beads were each mixed with 100µL rabbit anti-Fn antibody (Dako) diluted

1:1000 in phosphate buffered saline solution with tween (137mM NaCl, 2.7mM KCl, 8mM

Na

HPO

, 1.46mM KH

HPO

, 0.2% v/v Tween-20, pH 7.2) (PBS/Tween), vortexed, and

then incubated 30min at room temperature with occasional inversion. Following

incubation, beads were washed with 3x1mL PBS/Tween. Beads were then incubated with

100µL polyclonal swine anti-rabbit Ig horse radish peroxidase (HRP)-conjugated antibody

(Dako), diluted 1:1000 in PBS/Tween, for 30min at room temperature with occasional

inversion and then washed with 4x1mL PBS/Tween. Substrate was made up by dissolving 3,3’,5,5’-Tetramethylbenzidine (TMB) (Sigma Aldrich) in 1mL dimethylsulfoxide and then adding 9mL phosphate citrate buffer with sodium perborate (Sigma Aldrich). 100µL substrate was added to each of the bead solutions and then incubated for 3min. Stop solution was 50µL H

SO

. Beads were then centrifuged 1000rpm for 2min and supernatant read at 450nm.

0.1.3 Coupling of tissue transglutaminase to fibronectin

0.5µg/µL GST-tagged tTG was diluted down to 0.2µg/µL with tTG coating buffer (50mM Tris, 150mM NaCl, 5mM CaCl

, pH 7.5). 50µL of the diluted tTG was added to an equal volume of beads and incubated at room temperature for 1.5hrs. Bead slurry was then centrifuged and supernatant removed. Both 0.2µg/µL tTG solution and post incubation supernatant were analysed with SDS-PAGE to determine if tTG had bound the Fn on the beads.

0.2 12% SDS-PAGE gel

Resolving gel was made of 4mL 30% acrylamide, 2.5mL 1.5M Tris (pH 8.8), 100µL 10%

SDS, 100µL 10% (NH

)

S

O

, 4µL tetramethylethylenediamine (TEMED) and 3.3mL distilled H

O. Stacking gel was made of 500µL 30% acrylamide, 379µL 1.5M Tris (pH 8.8), 30µL 10% SDS, 30µL 10% (NH

)

S

O

, 4µL TEMED and 2mL distilled H

O.

Samples were added to sample buffer in a minimum ratio of 1:1 sample:buffer and then heated at 95°C for 5min. Condensation was centrifuged down and samples loaded onto gel.

Molecular weight marker was a prestained protein marker from BioLabs. Gels were run at

35mA through stacking gel and then 30mA through resolving gel. Gels were stained for a

minimum of 2hrs, preferably overnight when practical, with 1% w/v Coomassie brilliant

blue in 5% glacial acetic acid and 50% methanol. Destaining was done with 10% glacial acetic acid and 12.5% isopropanol.

0.3 ELISA

0.3.1 Optimisation

Optimisation of ELISA was carried out in a series of steps. First Fn concentration was optimised by coating a Nunc Immunoplate with 100µL of 1.0, 0.5 or 0.25µg Fn diluted in either tTG coating buffer or carbonate/bicarbonate buffer (15mM Na

CO

, 35mM NaHCO

, pH 9.6) (table 2.1A). Following overnight incubated at 4°C, wells were blocked with 1% bovine serum albumin (BSA) in PBS/Tween for 1hr. Wells were manually washed with 3x200µL PBS/Tween. Primary rabbit anti-FN antibody (Sigma) was added at a dilution of 1:1000 in PBS/Tween for 1hr at room temperature. Wells were washed as previously. Secondary antibody, swine anti-rabbit Ig/HRP (Dako), was added at a dilution of 1:5000 in PBS/Tween for 1hr at room temperature. Wells were washed as previously.

Substrate was 100µL TMB (Sigma) added for 5min. Stop solution was 100µL H

SO

. Results were read at 450nm.

Secondly, primary and secondary antibody concentrations were optimised. Fn was diluted

in either or carbonate/bicarbonate buffer or tTG coating buffer. 100µL of each FN dilution

was pipetted into each well and allowed to incubate over night at 4°C. Wells were blocked

with 1% BSA as described earlier. Wells were washed as described earlier. Primary

antibody was added at dilutions of 1:1000, 1:5000 or 1:10,000 in PBS/Tween and

incubated for 1hr at room temperature. Wells were then washed as previously. Secondary

antibody was added at dilutions of 1:1000 or 1:5000 and incubated for 1hr at room

temperature. Wash was done as previously. 100µL substrate was added to each well and

left for 5min. Stop solution was 100µL 2M H

SO

. Plate was read at 450nm.

Thirdly, choice of diluting buffer for Fn was done by comparing OD

between tTG coating buffer and carbonate/bicarbonate buffer. This analysis was done in parallel with optimisation of antibody and Fn concentrations, using the results obtained.

Blank Blank Blank Blank Blank Blank

1µg 1µg 1µg 1µg 1µg 1µg

0.5µg 0.5µg 0.5µg 0.5µg 0.5µg 0.5µg

0.25µg 0.25µg 0.25µg 0.25µg 0.25µg 0.25µg

1° antibody 2° antibody 1° antibody 2° antibody

Blank 1:1000 Blank 1:5000

1:1000 1:1000 1:1000 1:5000

1:5000 1:1000 1:5000 1:5000

1:10,000 1:1000 1:10,000 1:5000

Table 2.1: ELISA optimisation. Schematic diagram of (A) Fn concentration optimisation. Pink cells represent dilutions done with coating buffer. Lavender cells represent dilutions done with carbonate/bicarbonate buffer. Fn was incubated overnight at room temperature. Primary antibody was diluted 1:1000 in PBS/Tween. Secondary antibody was diluted 1:5000 in PBS/Tween; (B) Primary and secondary antibody concentration optimisation. Antigen concentration was 1µg Fn/well incubated overnight at room temperature. Pink cells represent dilutions done with coating buffer. Lavender cells represent dilutions done with carbonate/bicarbonate buffer.

0.3.2 Standard curve

tTG was diluted with tTG coating buffer. A 96 well Nunc Immunoplate was coated with 0.3µg tTG/well at 4°C overnight. Wells were then auto-washed with 3x300µL PBS/Tween and blocked with 100µL 1% BSA for 1hr at room temperature. Doubling dilutions from 1:80 to 1:20480 of pooled anti-tTG positive patient sera in PBS/Tween and 3 patients

B

A

samples known to be strong positive, weak positive and negative for anti-tTG at 1:25 were then incubated for 1hr at room temperature. Each dilution was performed in triplicate.

Wells were then washed and incubated with secondary antibody (polyclonal rabbit anti- human IgA/HRP, Dako) at 1:1000 for 1hr at room temperature. Wash step was repeated before adding 100µL substrate. Length of reaction with substrate was decided by observing the colour of the blanks, with 4min selected as optimal. Stop solution was 100µL 2M H

SO

. Plates were read at 450nm.

A standard curve was created using GraphPad. Dilutions were given arbitrary units (AU) with the highest dilution as 100, the second highest 50, the next 25 etc. Patient samples were extrapolated from the standard curve. A dilution range of seven points from 1:320- 1:20480 was decided as optimum for the standard curve.

2.4 Comparison of anti-tTG binding for antigen coatings tTG, Fn-tTG and Fn

2.4.1 Optimisation

Initially ELISA was run by incubating 0.5µg Fn/well in columns 7-12 of a 96 well Nunc Immunoplate for 1hr at room temperature. The whole plate was then auto-washed with 3x300µL PBS/Tween. However, this was found to affect binding of tTG to plate wells.

Instead, only columns 7-12 were washed with PBS/Tween.

To see if PBS/Tween also affected binding of tTG to Fn columns 7-12 were coated with 0.5µg Fn/well for 1hr at room temperature. Columns 7-12 were then auto-washed with 3x300µL PBS. Columns 1-9 were then incubated with 0.3µg tTG/well overnight at 4°C.

ELISA was then run as described in 2.4.2.

2.4.2 Modified tTG ELISA

A 96 well Nunc Immunoplate was coated with 0.5µg Fn/well in columns 7-12 and incubated 1hr at room temperature. Columns 7-12 were then auto-washed with 3x300µL PBS/Tween using an automated plate washer. Columns 1-9 were then coated with 0.3µg tTG/well and incubated at 4°C overnight. The various levels of antigen layering are shown in figure 6. After washing, wells were blocked with 100µL of 1% BSA solution for 1hr at room temperature. Wash was repeated and columns 4-12 were then incubated at room temperature for 1hr with 100µL patient sera at a dilution of 1:25. Columns 1-3 contained the standard curve, which consisted of pooled anti-tTG positive patient sera in doubling dilutions of 1:320-1:20480. Wells were then washed and incubated for 1hr at room temperature with 100µL secondary antibody (rabbit anti-human IgA/HRP, Dako) diluted 1:1000 in PBS/Tween. After washing, 100µL substrate (TMB, Sigma) was added for 4min.

Stop solution was 100µL 2M H

SO

. Plates were read at 450nm.

Figure 2.1: Three antigen coatings. Three different antigen coatings were used for comparison of anti-tTG binding. Antigen coating 1 is representative of current ELISA antigen configurations wherein only tTG is used. It was also the antigen coating used for the standard curve. Antigen coating 2 uses the modified antigen configuration suggested by Teesalu et. al. whereby the ELISA wells are first incubated with Fn and then with tTG (Teesalu et al. 2009). tTG binds Fn with high affinity and almost instantaneously. This antigen coating theoretically reflects the in vivo configuration of tTG as found in the extracellular matrix and the one used with EMA-based diagnostic tests. Antigen coating 3 consists of solely Fn.

Standard curve was created using GraphPad as described in section 2.3.2. Patient samples were extrapolated from the standard curve and converted to AU as described previously.

0.5 Statistical analysis

All statistical analysis was carried out using Microsoft Excel and GraphPad.

0.5.1 Standard deviation, variability coefficient and correlation coefficient

Standard deviation (SD) and variability coefficient (CV) was calculated using Excel to determine pipetting accuracy of triplicates. For correlation studies, Pearson product- moment correlation coefficient (Pearson’s r) was calculated using GraphPad.

0.5.2 Inter assay variability

Patient serum #2 was chosen for calculation of inter-assay variability. Mean and SD were calculated for AU obtained for tTG, Fn-tTG and Fn coatings for each in-house ELISA run.

Inter-assay variability for each coating was then calculated using the following equation:

(SD

)

---x 100 (Mean

)

0.6 Patient sera

Ten patient sera from patients diagnosed with coeliac disease were obtained from St.

James’s Hospital, with ethical approval for coeliac disease research.

Patient serum was kept in aliquots of 25-40µL in -8°C and thawed on ice. Dilutions were

done with PBS/Tween. Diagnosis of anti-tTG autoantibody levels and EMA test results

came from St. James’s Hospital (table 2.2). Patient serological diagnosis was carried out

using Phadia’s EliA Celikey™ anti-tTG ELISA. Coeliac disease was confirmed by biopsy.

Negative patient sample came from a confirmed coeliac disease free volunteer.

Patient nr. Y.O.B. Gender EMA tTG

1 1962 F weak positive 3.2

2 1970 M positive 33.7

3 1972 F positive 4.2

4 1948 M weak positive 3.9

5 1949 F positive 11.1

6 1950 F weak positive 13.3

7 1942 F weak positive 4.3

8 1946 F positive 28.6

9 1956 M positive 53.8

10 1946 M positive 57.4

Table 2.2: Age, gender, anti-tTG autoantibody and EMA status of test sera. Data from St. James’s Hospital. Patients were assigned ID numbers separate from their specimen number. Coeliac disease was diagnosed by intestinal biopsy. For the standard curve, pooled sera from patients with high anti-tTG levels were used. Cut-off value for anti-tTG levels, as used at St. James’s hospital, Dublin, is 1.9.

3. Results

3.1 Fibronectin binds Affi-gel 15 beads

Before investigation of Fn and GST-tagged tTG interaction, it was first necessary to verify Fn had coupled to the Affi-gel 15 beads. Verification that Fn had coupled to the Affi-gel beads was to be done by comparing pre- and post-bead incubation Fn-HEPES solutions against Fn-coupled beads with SDS-PAGE. If Fn had bound the Affi-gel beads then post- incubation Fn-HEPES solution would have less Fn than the pre-incubation solution.

Furthermore, the Fn-coupled beads would also produce a 262 kDa band on the gel if coupling had been successful.

However, despite several attempts, gels were always blank except for the molecular weight marker and the positive control (pure Fn). For samples containing Fn-coupled beads this was thought to be due to Fn not decoupling from the beads despite pre-run treatment of samples. Stacking gel was retained and stained to see of Fn could be seen in the wells but these too remained unstained. The remaining possibility was that beads were unable to enter the narrow pipette tip used to load the gel. If the Fn was not decoupling from the beads prior to loading, this would effectively result in the wells being blank. Pre- and post-bead incubation supernatant were also measured the with a Nanodrop spectrophotometer. However, a minute amount of beads in the post-incubation supernatant interfered and no viable results could be generated for comparison.

Due to the various problems encountered, an ELISA was adapted to compare

"naked" beads blocked with ethanolamine and the Fn-coupled beads. This gave an OD

of 0.149 for "naked" beads and 0.607 for Fn-coupled beads. Along with the lack of protein

as visualized with SDS-Page gel in the post bead-incubation samples and the Affi-gel

manufacturer's guarantees for their product, it was concluded Fn had bound the Affi-gel

beads.

An attempt was also made to determine Fn concentration on the beads using a micro Bradford assay. It was hoped that a viable comparison between Fn-coupled beads, "naked"

beads blocked with ethanolamine and post bead-incubation supernatant could be made.

Due to the beads interfering with the spectrophotometer, no value was attained. However, as an interesting side effect, the Bradford reagent dyed the Affi-gel beads blue. Binding of Brilliant Blue G to protein causes a shift in maximum absorbance from 465 to 595nm, resulting in a blue colour. Under a microscope the well containing Fn-coupled beads contained far more blue spheres than the well with the post bead-incubation supernatant or the blocked beads. Whilst giving no numerical value to the amount of Fn coupled to the Affi-gel beads, it did imply the beads were coated with protein.

3.2 GST-tagged tissue transglutaminase binds fibronectin

To establish that GST-tagged tTG had maintained its in vivo ability to bind Fn with high affinity, Fn was attached to a solid support (Affi-gel beads) and then incubated with GST- tagged tTG. Supernatant of pre- and post-incubation tTG solutions were then examined for reduced levels of GST-tagged tTG.

By comparing the two supernatants as well as tTG incubated Fn-coupled beads using SDS-PAGE, it could be clearly seen that levels of a protein 103kDa in size had markedly decreased following incubation with the Fn-coupled beads (fig. 3.1). Gels showed dark bands in the correct region (tTG has a size of 78kDa and the GST-tag is 26kDa) for the pre-incubation tTG solution and a markedly fainter band in the same region post- incubation.

Samples of the tTG incubated Fn-beads also gave faint bands in the aforementioned

region meaning tTG had bound the Fn-coupled beads. This strongly indicates tTG was

bound to Fn and not coupled to the Affi-gel beads themselves: as noted earlier, Fn did not

become uncoupled from beads even following heating to 95°C.

Figure 3.1: SDS-PAGE verifies binding of tTG to Fn. Post-incubation supernatant (lane 2) has a far fainter band at 103kDa than pre-incubation tTG solution (lane 1). Lanes 4 and 5 contained 10uL Fn-tTG coupled beads and show bands in the same region as other two samples.

3.3 ELISA optimisation

Optimisation of ELISA was done by comparing two different coating buffers for Fn incubation and different dilutions of secondary antibody and Fn.

Literature cites different coating buffers for Fn and tTG. This was deemed

cumbersome when performing the modified ELISA. The purpose of this step was therefore

to see if the tTG coating buffer could be used for Fn coating as well without a loss of

performance. For choice of Fn coating buffer, Fn diluted with carbonate/bicarbonate buffer

was compared to Fn diluted with tTG coating buffer. An ELISA was then run with 0.5µg

Fn/well and anti-Fn monoclonal antibody as the primary antibody. Comparison of

carbonate/bicarbonate buffer versus tTG coating buffer showed little difference in regards

to Fn binding to ELISA wells (fig. 3.2). Fold changes (carbonate/bicarbonate buffer

divided by tTG coating buffer) for each antibody dilution ranged between 0.9-1.3. Whilst

Fn diluted with carbonate/bicarbonate buffer did have slightly higher OD

values, it was

not significant enough to warrant usage of two different buffers (carbonate/bicarbonate

buffer for Fn and coating buffer for tTG) when running the modified ELISA. tTG coating buffer was therefore used to dilute both Fn and tTG.

Figure 3.2: Optimisation of ELISA. Fn was diluted in either tTG coating buffer (lavender bars) or carbonate/bicarbonate buffer (green bars). Primary antibody was anti-Fn antibody diluted 1:1000, 1:5000 or 1:10,000. Secondary antibody was diluted either 1:1000 or 1:5000. Mean OD450 for blanks shown as dotted line.

The optimal concentration of secondary antibody was decided on as 1:1000. This

gave OD

values two to three times greater than those of the 1:5000 dilutions, even with

low levels of primary antibody (fig. 3.2). It was also greater than that of the blanks, giving

a larger margin between values for patient sera and potential background noise (which

varied in OD

from 0.07-0.23 during in-house ELISA runs).

Initially Fn dilution was also to be optimised but the ELISA failed. Due to time constraints Fn dilution was therefore chosen as 0.5µg Fn/well as this gave useable OD

values when coupled with the chosen primary and secondary antibody dilutions.

From previous studies and due to limited availability of GST-tagged tTG, 0.3µg tTG/well was used as the tTG dilution (Byrne et al. 2007).

3.4 ELISA standard curve

For the different ELISA runs to be compared to one another, a standard curve was generated using doubling dilutions of pooled patient sera with high levels of anti-tTG autoantibodies. Using the standard curve, OD

values obtained from spectrophotometric measurement of ELISA wells could be converted to AU from 1.56-100. OD

measurements of patient sera that fell within the range of the standard curve could be converted to AU.

Using initial doubling dilutions of 1:40 to 1:10240, the optimal range for the standard curve was found to be 1:320 to 1:20480 (fig. 3.3). Serum from patient #1 (weak positive for EMA and 3.2 in anti-tTG autoantibody levels, the lowest of patient samples) and patient #10 (strong positive for EMA and 57.4 in anti-tTG autoantibody levels, the highest of patient samples) could both be extrapolated from the standard curve. The negative control also fell within the range of the standard curve.

For patient sera, dilution was chosen at 1:25. This dilution gave values within range

of the standard curve for both patient #1 and #10. However, whilst running the ELISA

later, patients #8 and #9 both had values outside the standard curve. These two patient

samples were therefore diluted 1:50.

Figure 3.3: Standard curve was used to extrapolate arbitrary units for patient anti-tTG autoantibody levels.

Doubling dilution of pooled positive patient sera from 1:320 to 1:20480 was used. Dilutions were then given arbitrary units with 1:320 equalling 100 and 1:20480 equalling 1.56. OD450 values used were means of triplicates with background levels (blanks) removed.

3.5 Coupling of tissue transglutaminase to fibronectin does not markedly increase anti-tTG binding

A two level antigen coating, tTG bound to Fn, was compared to the current ELISA setup

of tTG only, to see if there was an improvement in anti-tTG binding. A Fn only antigen

coating was included as well. This was to screen for antibody binding directed at the

something other than tTG in the Fn-tTG coating as coeliac disease patients have antibodies

directed at antigens other than gliadin and tTG (Krupickova et al. 1999).

Figure 3.4: Line diagram of AU values obtained for three different antigen coatings. A trend of general decrease in anti-tTG binding in the Fn-tTG coating compared to the tTG coating can be clearly seen with the exception of two patients. Serum from patients #8 and #9 were diluted 1:50. All others were diluted 1:25.

Figure 3.5: Fold change analysis of Fn-tTG coating compared to tTG alone. Values were calculated by dividing AU values for Fn-tTG antigen coating with AU values for tTG antigen coating. A value of 1 meant the two different antigen coatings were equivalent. Values <1 meant the Fn-tTG antigen coating was inferior to the tTG coating. Values >1 meant the Fn-tTG antigen coating was superior. Mean fold change is purple.

It was hoped that by binding tTG to Fn, anti-tTG activity would increase due to the autoantigen being in the in vivo configuration. However, results showed little to no improvement of anti-tTG binding for the Fn-tTG coating compared to the tTG coating.

Each ELISA run showed a distinct pattern of anti-tTG activity actually decreasing

following the coupling of tTG to Fn. This pattern becomes very distinct in figure 3.4 where

patient sera values are plotted against the three antigen coatings. Only patients #2 and #7

show an increase in anti-tTG binding. The level of decrease varied from patient to patient

but was overall far more common a trend than an increase.

To properly see quantitative changes in AU values from tTG antigen coating to Fn- tTG antigen coating, the fold change was plotted (fig. 3.5). Fold change was calculated by dividing the AU calculated for Fn-tTG coating with the AU for tTG coating. Only patients

#2 and #7 showed an increase in AU, though small: 12.4% and 3.7% respectively. The remaining patients did not achieve a reading equal to the tTG antigen coating, instead experienced a decrease in anti-tTG activity. Mean fold change for all patient samples (n=10) was 0.829.

3.6 In-house ELISA performance

Patient data regarding EMA and anti-tTG autoantibody levels were available from St.

James’s Hospital, Dublin. These were compared to results from own ELISA runs to measure faithfulness (table 3.1)

.

patient
 St.
James’s
Hospital
 In‐house
results
(AU)



 EMA
 tTG
 tTG
 Fn‐tTG
 Fn


1
 weak positive 3.2 1.3 1.2 0.8

2
 ^{positive 33.7 70.2 78.9 6.1}

3
 ^{positive 4.2 1.3 0.9 0.6}

4
 weak positive 3.9 18.6 16.3 3.7

5
 positive 11.1 31.8 29.9 5.5

6
 weak positive 13.3 1.4 1.2 0.5 7
 weak positive 4.3 21.5 22.3 3.9

8
 positive 28.6 171.4* 57* 7.4*

9
 positive 53.8 153.4* 101.2* 36.6*

10
 ^{positive 57.4 1.5 1.4 0.9}

Table 3.1: Results of in-house ELISA compared to St. James’s Hospital diagnostic results. Values taken for the different antigen coatings are collated from three different ELISA runs and are in arbitrary units. (*) denotes patient sera diluted 1:50 – these AU values have been doubled. All others were diluted 1:25.

Figure 3.6: Correlation study between data from St. James’s Hospital and in-house ELISA showed a moderately positive correlation between the commercial ELISA and the in-house ELISA despite both having the same antigen coating of tTG only. Pearson’s r= 0.5. Number of samples n=10.

Antigen coating Inter-assay variability (%)

tTG ‐


Fn-tTG 94.6%


Fn 81.0%


Table 3.2: Inter-assay variability for antigen coatings. AU of patient #2 from all in-house ELISA runs (n=3) were used to calculate mean and standard deviation. Standard deviation was then divided with mean.

Data for tTG antigen coating not available.

A correlation study between the tTG values from St. James’s Hospital and the in- house tTG data gave a Pearson’s r=0.5 (fig. 3.6). This meant correlation between the commercial tTG ELISA and the in-house tTG ELISA was moderately positive.

Inter-assay variability was initially to be calculated using patient #2. Data was missing from one run for the tTG coating due to values being outside the range of the standard curve. Inter-assay variability for the Fn-tTG and Fn coating is shown in table 3.2 and was 94.6% for the Fn-tTG coating. Overall, AU values ranged from all values <1.6 in one run to outside the range of the standard curve in another. This was not due to faulty pipetting technique as CV for triplicates was between 0.2-25.3% with an average of 6.4%.

3.7 Tween does not noticeably affect tissue transglutaminase binding fibronectin

It was initially suspected the decrease in anti-tTG binding for the Fn-tTG coating compared to the tTG coating was due to Tween in the wash solution interfering with the attachment of tTG to Fn during the second step of antigen coating. An ELISA was run using patient sera with the five lowest anti-tTG levels using PBS to wash before the second antigen coating. Fold change analysis showed no marked increase in test results for 2/5 patients compared to washing with PBS/Tween (fig. 3.7). The exception was patient #5 who had over a 2-fold increase in AU. A decrease in test sensitivity was seen for 2/5 patients.

As Tween had such a negligible effect on Fn and tTG interaction, it was decided to

use 3x300µL PBS/Tween as the standard method for ELISA washes

Figure 3.7: comparison of fold change seen when washing with PBS/Tween or PBS only between antigen incubations. A marked increase in test sensitivity was only seen with patient #5. For all others, difference was of negligible diagnostic value.

.

3.8 Antibody binding of fibronectin was low for 90% of patients

The third antigen coating used in the in-house ELISA was Fn only. This was used to ensure values were due to tTG binding and not due to antibodies interacting with Fn

For nine samples patient sera showed far less antibody binding with Fn coating

compared to tTG coating (fig. 3.8). The exception to this was patient #9 who had AU=18.3

for Fn coating (fig. 3.8 circled). This was on par with patient #4 antibody activity for tTG

coating (AU=18.6) and only slightly less than patient #7 (AU=21.5). Antibody binding of

tTG coating for patient #9 was AU=76.7, the second highest value recorded for that

particular antigen coating.

Figure 3.8: comparison of antibody binding with Fn coating (green) to tTG coating (purple). Antibody binding is significantly less with Fn only. The exception to this is patient #9 where antibody activity was on level with 2/10 patients antibody binding with tTG only antigen coating (circled).

4. Discussion

Coeliac disease is a chronic form of enteropathy triggered and driven by dietary gluten (Trier 1991). What starts as an inflammatory immune reaction eventually becomes an autoimmune condition when the patient produces autoantibodies to the autoantigen tTG (Dieterich et al. 1997). Anti-tTG autoantibodies are highly disease specific and commonly used in diagnosis of coeliac disease (Caja et al. 2011). Final diagnosis is currently done with intestinal biopsy, a cumbersome method compared to serological tests (Koning et al.

2005). It is therefore hoped that a serological test could be made sensitive enough for screening and diagnosis. As of today, the two serological tests used are the immunofluorescence based assay for EMA and an ELISA using recombinant tTG (Caja et al. 2011). Both test anti-tTG levels.

In vivo extracellular tTG binds Fn with high affinity (Jeong et al. 1995; Turner &

Lorand 1989). A recent publication has suggested that this characteristic be used to create an ELISA with a two-tiered antigen coating: Fn would be coated to the plate and then His- tagged tTG could be coupled to Fn (Teesalu et al. 2009). This gave tTG a similar orientation to its in vivo one. Theoretically more tTG epitopes would be presented, thus increasing anti-tTG binding.

Bearing this in mind this project was aimed at answering two questions: firstly, would GST-tagged tTG bind Fn as the His-tagged tTG was shown to do in the Teesalu et.

al. publication?; secondly, would coupling of GST-tagged tTG to Fn cause an increase in

autoantibody binding?

The results of this project showed that despite the GST-tag being larger than the His-

tag used by Teesalu et. al. GST-tagged tTG was still capable of binding Fn. This was

proven using SDS-PAGE. By first attaching Fn to a solid support and then incubating it

with GST-tagged tTG the resulting Fn-tTG coupled beads had their non-covalent interactions disrupted by heating which allowed for visualisation of tTG on the gel.

The recombinant tTG ELISA used for coeliac disease diagnosis was then modified to allow for comparison of anti-tTG binding to three different antigen coatings; tTG, Fn-tTG and Fn. Results showed that coupling of tTG to Fn did not give an increase in anti-tTG binding for patient samples. In fact, 8/10 patients experienced a decrease in anti-tTG binding. With the exception of one patient sample, patient antibodies did not bind Fn.

4.1 GST-tagged tissue transglutaminase binds fibronectin

SDS-PAGE gels had been run with Fn-coupled beads with no results. This was concluded to be due to the strong covalent bond between Fn and the bead. This bond was not disrupted by heating so Fn never appeared in the loaded sample. This is in contrast to samples of the Fn-tTG beads which gave faint bands in the 103 kDa region. This strongly indicates tTG was bound to Fn and not coupled to the Affi-gel beads themselves: as noted earlier, Fn did not become uncoupled from beads even following heating to 95°C.

However, the bond between Fn and tTG is non-covalent and was destroyed by heating (Turner & Lorand 1989).

In vivo tTG binds Fn with high affinity. Whilst the results of this project showed the

GST-tag had not disrupted the Fn-binding region of tTG, it was not proven the GST-tag had not reduced the binding affinity. It was not within the scope of this project to investigate bond strength between GST-tagged tTG and Fn.

The decrease in anti-tTG binding seen with the Fn-tTG coating in comparison to the

tTG coating could be explained by a lower amount of tTG epitopes in the former. If GST-

tagged tTG binds Fn with lower affinity than that seen in vivo, a certain percentage of tTG

could disappear with each wash. Comparison of fold change analysis of patients with low

levels of anti-tTG showed no marked difference between washing with PBS versus

PBS/Tween during antigen incubations. This indicated GST-tagged tTG was not being displaced by Tween washes.

4.2 Coupling of tissue transglutaminase to fibronectin does not improve anti-tTG binding

ELISAs were run to compare the standard tTG coating to the two-tiered Fn-tTG coating using sera from patients diagnosed with coeliac disease. This showed no overall trend in improved anti-tTG binding for the Fn-tTG coating compared to the tTG coating, proving the coupling of tTG to Fn had no positive effect on antibody activity. On the contrary, data showed anti-tTG binding actually decreased followed coupling of tTG to Fn compared to the tTG coating.

The mechanism behind the decrease can be divided into three different possibilities:

1) A decrease in available epitopes 2) A decrease in antigen levels

3) Direct interference between antigen and antibody interaction

Each possible mechanism shall be treated in turn. Possible future studies to answer questions that arose during the course of this project shall also be discussed in a later subsection.

4.2.1 A decrease in available epitopes

The purpose behind the coupling of tTG to Fn was to increase the number of presented epitopes by mimicking the in vivo conformation of extracellular tTG. This took for granted that the amount of tTG in the Fn-tTG coating was equal to that found in the tTG coating.

However, Fn is part of the extracellular matrix which is a 3-dimensional structure. In

comparison an ELISA well is a far flatter, 2-dimensional structure. There is no guarantee