: D.I.Pritchard, The Boots Institute, School of Pharmaceutical Sciences, University of Nottingham, Nottingham NG7 2RD, UK
Calreticulin was recently identified as a hookworm Necator americanus allergen, implying secretion, and contact with cells of the immune system, or significant worm attrition in the tissues of the host. As human calreticulin has been shown to bind to and neutralize the haemolytic activity of the complement component C1q, and to be putatively involved in integrin-mediated intracellular signalling events in platelets, it was of interest to determine whether a calreticulin from a successful nematode parasite of humans, with known immune modulatory and antihaemostatic properties, exhibited a capacity to interfere with complement activation and to interact with integrin domains associated with cell signalling in platelets and other leucocytes. We can now report that recombinant calreticulin failed to demonstrate significant calcium binding capacity, which is a hallmark of calreticulins in general and may indicate inappropriate folding following expression in a prokaryote. Nevertheless, recombinant calreticulin retained sufficient molecular architecture to bind to, and inhibit the haemolytic capacity of, human C1q. Furthermore, recombinant calreticulin reacted in surface plasmon resonance analysis (SPR) with peptides corresponding to cytoplasmic signalling domains of the integrins αIIb and α5, in a calcium independent manner. SPR was also used to ratify the specificity of a polyclonal antibody to hookworm calreticulin, which was then used to assess the stage specificity of expression of the native molecule (in comparison with reverse transcriptase-polymerase chain reaction), to indicate its apparent secretion, and to purify native calreticulin from worm extracts by affinity chromatography. This development will allow the functional tests described above to be repeated for native calreticulin, to ascertain its role in the host–parasite relationship.
Calreticulin is known as a highly conserved protein, present in every cell of higher organisms, except erythrocytes and yeast. First identified as a calcium binding protein located to the endoplasmic reticulum (ER), calreticulin has subsequently been implicated in multiple cell functions (1). In addition to playing an important role in regulating intracellular calcium concentration, calreticulin acts as an intracellular chaperone for glycoproteins, using well defined lectin domains to engage carbohydrate side chains. Furthermore, calreticulin can also contribute to the regulation of cell adhesion and gene expression (2,3). It is contained in granules of cytotoxic T-cells and released by neutrophils during inflammation, possibly modulating immune effector mechanisms in the process (4). Significantly in this respect, human calreticulin inhibits C1q-dependent complement activity in vitro (5).
The discovery of this multidomained and biologically multifunctional protein as a hookworm allergen has led to speculation as to its biological function in the host–parasite relationship (6). As calreticulin also demonstrates anticoagulant activity, by binding to blood clotting factors IX, X and prothrombin, it might represent one of the components with antihaemostatic activity found in secretory products from blood feeding parasites (7). Furthermore, the chaperone function of parasite calreticulin may also aid in the correct folding and export of important glycosylated secreted proteins, such as superoxide dismutase (SOD), capable of scavenging reactive oxygen intermediates (8). The presence of calreticulin in the penetration gland cells of schistosome cercariae suggests a regulatory influence on calcium-dependent proteases involved in skin penetration and parasite migration (9).
In the present study, we focus on the potential ability of a hookworm calreticulin to interfere with a key component of the human complement system, C1q, and its ability to react directly with intracellular signalling domains of adhesion molecules considered important to leucocyte and platelet function. As calreticulin is unlikely to be expressed in sufficient quantities, or at all, in native worm secretions to allow purification and bioassay, recombinant calreticulin has been over expressed under optimized conditions in a bacterial system, then purified for biological assay. The data obtained indicate that recombinant hookworm calreticulin binds to, and inhibits the biological function of, human C1q, and binds specifically to the cytoplasmic signalling domains of a number of integrins.
This leads us to believe that the calreticulin allergen from Necator americanus has an anti-inflammatory and antihaemostatic function. However, we have to date merely proven that hookworm calreticulin has the potential to fulfil these functions; the experimental platform established here will now allow us to determine whether native N. americanus calreticulin is truly a secretory molecule with an ability to access the intracellular milieu.
Materials and methods
Necator americanus was maintained in syngeneic DSN hamsters (10). Two to 4-day-old neonate hamsters were infected percutaneously with 100 infective larvae and adult worms harvested from the small intestine 35 days postinfection. L4 stages of N. americanus were obtained 14–17 days post-infection using a modified Baermann technique (11). Infective L3 larvae were cultured from faecal material from infected hamsters by a previously described method (12). All life cycle stages were homogenized in phosphate-buffered saline (PBS), on ice using a Polytron (Kinematica, Lucerne, Switzerland). Insoluble material was removed by centrifugation at 13 000 g for 10 min. Protein content was determined using the BioRad protein assay (BioRad Laboratories, Hercules, CA, USA). Homogenates were stored at − 20°C until required.
Excretory/secretory (ES) products were collected from infective larvae as described previously (12). Briefly, freshly collected, ensheathed larvae were resuspended in water and exsheathed by bubbling carbon dioxide through the suspension for 2 h at room temperature. Exsheathed larvae were allowed to settle and then washed extensively with RPMI 1640 containing 100 IU/ml penicillin, 100 µg/ml streptomycin and 1% (w/v) amphotericin B under sterile conditions. Following this ‘sterilization period’, the larvae were cultured in RPMI 1640 for 72 h at 37°C, changing the culture medium every 24 h.
ES products were obtained from L4 and adult worms by washing the worms extensively in RPMI 1640 containing 100 IU/ml penicillin and 100 µg/ml streptomycin over a period of 2 h followed by further culture at 37°C in RPMI 1640 for 24 h. ES products obtained from infective larvae, L4 and adult worms were dialysed against distilled water, freeze-dried and stored at − 20°C.
All commercial antibodies were obtained from Sigma Chemical Co. (Poole, UK). Human post-infection plasma was collected during a field study in December 1996 in Papua New Guinea (6).
Production of a polyclonal antiserum to purified recombinant calreticulin
A New Zealand White Rabbit (Harlan Olac, Bicester, Oxford, UK) was injected subcutaneously with approximately 50 µg of purified recombinant calreticulin emulsified in Freund's complete adjuvant. On two subsequent occasions, 14 days apart, the rabbit was injected again with 50 µg of calreticulin in Freund's incomplete adjuvant. Finally, the rabbit was exsanguinated under terminal anaesthesia and the specificity of the plasma tested. Residual cross-reactivity was absorbed by gently mixing equal volumes of plasma and crude Escherischia coli lysate (3 mg/ml stock, Stratagene, La Jolla, CA, USA) at room temperature for 5 h. Immunoglobulin (Ig)G was prepared from this antiserum by affinity chromatography on Protein G-Sepharose (Pharmacia, Uppsala, Sweden) as described by the manufacturer.
Expression and purification of recombinant calreticulin
The plasmid pQE-CalδSig was used for the expression of soluble calreticulin. It encodes an N. americanus calreticulin cDNA deleted by polymerase chain reaction (PCR) of the N-terminal signal sequence and fused to 6x histidine tag to facilitate affinity purification. Plasmid pDS-SmE16 was used for expression of Schistosoma mansoni calmodulin (13), which served as a recombinant control protein in bioassay.
In order to optimize the expression of calreticulin a number of incubation conditions were monitored including IPTG concentration (1 mm to 4 mm), growth temperature (30–37°C), culture inoculum (1 : 6 to 1 : 10 dilutions from an overnight culture) and the time of expression. To investigate these parameters, samples of E. coli culture were taken every 60 min and the total cell protein analysed by Western blotting using human post-infection IgE. Bacterial growth was monitored by measuring optical density at 600 nm. Results of the optimization experiments are shown in Figure 1. Calreticulin expression was found to be optimal under the following conditions: transformed E. coli M15 (pREP4) were grown overnight at 37°C in LB broth containing kanamycin (30 µg/ml) and ampicillin (200 µg/ml). The culture was diluted 1 : 7 in fresh medium and, after 30 min, IPTG was added to a final concentration of 2·8 mm. After vigorous shaking for 3 h at 37°C, the culture was harvested by centrifugation.
Talon resin (Clontech, Basingstoke, UK), which contains cobalt as a ligand for 6x histidine tagged proteins, was used for purification of recombinant N. americanus calreticulin and S. mansoni calmodulin. Cells were re-suspended in 1 : 10 culture volume of sonication buffer (0·1% Triton-X100, 10 mm Tris/HCl, 10 mm NaH2PO4, 100 mm NaCl, pH 8·0). Lysozyme was added to a final concentration of 0·5 mg/ml and the suspension incubated at room temperature for 30 min. The cells were sonicated on ice at 200 W with 10 s bursts with intervals of 20 s for 10 cycles. All subsequent steps of the purification were carried out at 4°C. The lysate was centrifuged for 10 min at 13 000 g to separate soluble proteins from the insoluble debris. The supernatant was added to Talon resin (approximately 1 : 400 of culture volume) and gently shaken for 3 h. The matrix was collected by centrifugation for 2 min at 5000 g, washed twice with ten bed volumes of sonication buffer for 15 min each with gentle shaking. The matrix was then transferred to a 5 ml glass chromatography column and washed twice with three bed volumes of sonication buffer. The protein was eluted by adding four volumes of elution buffer (0·1% Triton-X100, 10 mm Tris/HCl, 10 mm NaH2PO4, 100 mm NaCl, 100 mm EDTA pH 8·0) and collecting 0·5 ml fractions. Protein eluting from the column was analysed by SDS-PAGE.
Finally, protein eluting from Talon resin was separated using preparative native PAGE (12%). By using prestained standards (Sigma SDS-7B), a strip of gel corresponding to the molecular mass of calreticulin was cut out and the electroeluted using a BioRad mini electroeluter according to the manufacturers instructions. Purity of electroeluted calreticulin was assessed by 12% SDS-PAGE and Western blots probed with human post-infection IgE.
Affinity purification of native N. americanus calreticulin
IgG prepared from the polyclonal rabbit anti-calreticulin was bound to Affiprep Hz Hydrazide support (BioRad). This matrix binds the Fc region of IgG allowing the correct orientation of the IgG for optimal protein binding (14). N. americanus homogenate prepared in 0·1 m sodium phosphate buffer, pH 7·4 was applied to a 5-ml of prepared matrix (pre-equilibrated with 0·1 m sodium phosphate buffer, pH 7·4). The column was washed sequentially with two column volumes of 0·5 m NaCl, 0·1 m sodium phosphate buffer, pH 7·4 and five column volumes of 0·1 m NaCl, 0·1 m sodium phosphate buffer, pH 7·4. Bound protein was eluted with 0·1 m sodium citrate pH 4. 800 µl of eluant was collected into 200 µl of 0·5 m Tris, pH 8·5 to neutralize the acidity of the elution buffer and preserve activity.
Calcium binding assays
Calcium binding was assessed by observing an electrophoretic band shift in the presence of calcium ions or EDTA following SDS-PAGE. A 12% gel was poured containing 0·1% SDS (15% for calmodulin, a control calcium binding protein). 5 µg of N. americanus calreticulin or S. mansoni calmodulin were preincubated for 10 min with either 10 mm CaCl2 or 5 mm EDTA. Native sample buffer (0·125 m Tris pH 6·5, 10% glycerol, 0·005% bromophenol blue) was added and the proteins electrophoresed on their respective gels in the presence of native running buffer (25 mm TrisCl pH 8·8, 250 mm glycine). Gels were stained with Coomassie Brilliant Blue R250.
Calcium binding was assessed in parallel using a method previously described (15). E. coli (M15) lysate, recombinant N. americanus calreticulin, bovine serum albumin, bovine calmodulin, adult N. americanus homogenate and ES products were resolved by native PAGE and transferred onto nitrocellulose. The filter was incubated in 60 mm KCl, 5 mm MgCl2, 10 mm imidazole, pH 6·8 for 1 h with several changes of buffer. The filter was then incubated for 10 min in the above buffer supplemented with 45Ca2+ (1 mCi/l). Finally the filter was washed with distilled water for 5 min and dried at room temperature. The filter was exposed to X-ray film for 2 days prior to development.
Measurement of calreticulin binding to C1q by ELISA
Microtitre plates were coated overnight at 4°C with 50 µl PBS containing 0–0·3 µg human C1q or 0–0·17 µg of purified recombinant N. americanus calreticulin. Following three washes with 0·05% Tween 20/PBS, the wells were blocked with 200 µl of 1% nonfat dried milk powder/PBS for 1 h at 37°C. To assess protein binding, the wells were incubated with either 0·5 µg recombinant N. americanus calreticulin or 0·125 µg C1q in 50 µl PBS for 2 h at 37°C. The plates were washed a further three times with 0·05% Tween 20/PBS and either 50 µl of rabbit anti-N. americanus calreticulin (1 : 400) or goat anti-human C1q (1 : 100) serum was added to each well and the plate incubated overnight at 4°C. The plates were washed as before and binding was detected by probing with either goat anti-rabbit IgG or rabbit antigoat IgG conjugated to alkaline phosphatase (1 : 5000) for 2 h at 37°C. Plates were developed by the addition of 200 µl of p-nitrophenylphosphate (5 mg/ml) and the absorbance measured at 405 nm. Polyclonal goat anti-human C1q and rabbit anti-N. americanus calreticulin sera were used to confirm dose-dependent binding of human C1q and N. americanus calreticulin to the wells (not shown).
C1q-dependent haemolytic assay
Human C1q was purified as described previously (5). C1q-deficient serum (Sigma) was diluted 1 : 20 in DGVB++[isotonic Veronal buffered saline containing 0·1 mm CaCl2, 0·5 mm MgCl2, 0·1% (w/v) gelatin and 1% (w/v) glucose]. 100 µl of diluted serum was incubated with 0–4 µg of human C1q at 37°C for 30 min. 100 µl (108 cells/ml) of sheep red blood cells (SRBC) sensitized with rabbit anti-SRBC IgG were added to each tube and incubated for a further 30 min at 37°C. Intact cells were pelleted by centrifugation and the absorbance at 405 nm of 100 µl of the supernatant measured to monitor haemoglobin release.
Under these conditions, 1 µg of C1q was found to cause approximately 40% haemolysis. This amount of C1q was incubated with 0–3 µg of recombinant N. americanus calreticulin in 100 µl of C1q-deficient serum (diluted 1 : 20 in DGVB++) for 30 min at 37°C. 100 µl of sensitized sheep red blood cells (108 cells/ml) were added to each sample, followed by a further incubation at 37°C for 30 min. After centrifugation, the amount of released haemoglobin was determined by measuring the absorbance at 405 nm of 100 µl of the resulting supernatant. Total haemolysis (100%) was measured by lysing 100 µl of sensitized sheep red blood cells with water. Normal human serum was used as positive control, and C1q-dependent haemolytic activity was expressed as a percentage of the total haemolysis. Background activity/spontaneous haemolysis were defined as 0% and determined by incubating 100 µl sensitized sheep red blood cells with diluted C1q-deficient serum without the addition of C1q or recombinant calreticulin. Recombinant S. mansoni calmodulin was used as a recombinant calcium-binding control protein, at the same concentrations as calreticulin in all assays.
Surface plasmon resonance analysis
Experiments were conducted with a Biacore X (Biacore AB, Uppsala, Sweden). For comparison, purified recombinant human and N. americanus calreticulin was diluted to 100 µg/ml in 10 mm sodium acetate buffer, pH 3·5, and immobilized on research grade carboxymethyl-dextran (CM5) sensor chips previously activated using N-ethyl-N′-(dimethylaminopropyl)carbodiimide hydrochloride and N-hydroxysuccinimide. After coupling, remaining reactive groups on the sensor chip were inactivated using 1 m ethanolamine-HCl pH 8·5. Binding assays were performed at 25°C, using as running buffer 50 mm Tris-HCl pH 7·4, 150 mm NaCl, 0·005% Tween 20, and as free analyte synthetic peptides corresponding to the integrin cytoplasmic domain of either wild type αIIb subunit, αIIb with a R995A substitution (αIIb R995A), αIIb deleted of the 989KVGFFKR995 juxta-membrane motif (αIIb N996-Q1008), wild-type α2, α5, or αv subunit (Neosystem, Strasbourg, France). The specificity of the rabbit anticalreticulin antiserum was checked after a pre-incubation with 5 mg/ml soluble carboxymethyl-dextran to reduce non–specific interaction. Samples were injected over the calreticulin-coated biosensor at 5 µl/min, and the binding of the analyte was monitored in real time as a variation in the SPR angle expressed in arbitrary resonance units (RU, 1000 RU correspond to approximately 1 ng of bound protein/mm2). Following each cycle of analysis, the sensor chips were regenerated with short pulses of 10 mm glycine pH 2·0 leaving the surface intact for additional interaction studies. The data were analysed using Biaevaluation 3·0 software (Biacore) after subtraction of the background signal monitored on a reference cell made of ethanolamine-substituted dextran.
Expression pattern of calreticulin
10 µg of larval, L4 and adult homogenates and ES products were resolved by SDS-PAGE (12%) under reducing conditions (16) and transferred onto nitrocellulose (17). Western blots were blocked with 5% nonfat dried milk powder in Tris buffered saline (TBS) for 1 h and incubated overnight at 4°C with rabbit anti N. americanus calreticulin (1 : 400 in blocker). After washing with TBS/0·05% Tween 20, the blots were incubated with alkaline phosphatase conjugated goat anti rabbit IgG (Sigma, 1 : 1000). Antibody binding was visualized by incubation with 5-bromo-4-chloro-3-indolyl phosphate/nitroblue tetrazolium.
RNA was extracted from L3, L4 (17 days post-infection) and 35-day-old adult N. americanus using the RNeasyTM Kit (Qiagen Inc., Santa Clarita, CA, USA), according to manufacturer's instructions. The RNA was resuspended in RNase free sterile water and the quality and quantity of RNA was estimated by agarose gel electrophoresis.
Separate reverse transcription reactions were performed on L3, L4 or adult stage N. americanus total RNA (approximately 5 µg) with oligo-dT (17) primer using M-MLV reverse transcriptase (Promega, Madison, WI, USA) according to the manufacturer's instructions. The resulting single stranded cDNA samples were used as templates for PCR reactions using the following primers: CalF (5′-TGACATTTTCAGCTACAAGGG-3′ nucleotides 447–467) and CalR (5′-TATAGTTCCTCGTGACCCTC-3′ nucleotides 1235–1216) (5). PCR was carried out using Taq polymerase for 30 cycles of 95°C for 1 min; 58°C for 1 min; 72°C for 2min; with a final extension of 10 min at 72°C. Appropriate controls included PCR on stage-specific RNA which had not been subjected to reverse transcription, PCR with no template (water only) and positive PCR controls using the cloned calreticulin gene as template to verify PCR conditions. Amplified products were size-fractionated on a 1% (w/v) agarose gel and viewed under ultraviolet light.
Expression and purification of soluble, recombinant calreticulin from N. americanus
The conditions required for the over expression of recombinant calreticulin were optimized as described in the materials and methods and yields assessed by Western blotting with post-infection IgE (Figure 1a, pre-optimization and Figure 1b post-optimization). The bands developed represent the yields of protein obtained from 2·4 ml (pre-optimization) and 1 ml (post-optimization) of bacterial culture. Pure recombinant protein was obtained by metal affinity chromatography using Talon resin and subsequent electroelution (Figure 1c). The protein showed an apparent molecular mass of 56 kDa on SDS/PAGE (Figure 1c), despite a predicted molecular mass of 47·5 kDa. This anomalous mobility is already known from other homologues, e.g. human or rabbit calreticulin (5). The purified recombinant protein predictably reacted with human IgE from patients from Papua New Guinea infected with N. americanus in Western blot analysis (Figure 1d). This serum had been used previously to identify and isolate calreticulin as a hookworm allergen, and showed no cross-reactivity with crude E. coli lysate when probing with IgE (6). Purified recombinant calreticulin also seems to exist in a dimeric/multimeric form, as indicated by the presence of a band of higher molecular mass in a native PAGE (Figure 1e) and associated IgE-based Western blot (Figure 1f).
Native N. americanus calreticulin was purified from N. americanus homogenate as described in the material and methods using an IgG preparation of a polyclonal rabbit antiserum raised to recombinant N. americanus calreticulin. The specificity of the antibody was further ratified by SPR analysis. The purified protein resolved on SDS-PAGE at approximately 56 kDa (Figure 1g) and was recognized on Western blots by human postinfection IgE (Figure 1h).
N. americanus calreticulin does not bind calcium
Recombinant N. americanus calreticulin (Figure 2a) and S. mansoni calmodulin (Figure 2b) were assessed for their ability to bind calcium using band shift electrophoresis as described in the materials and methods. S. mansoni calmodulin demonstrated a band shift in the presence of calcium (Figure 2b) while no calcium binding was observed with N. americanus calreticulin. Similarly, recombinant N. americanus calreticulin failed to bind 45Ca2+ (Figure 2c).
Recombinant N. americanus calreticulin binds to human C1q and inhibits C1q-dependent haemolysis
Despite an apparent inability of the recombinant hookworm calreticulin to bind calcium, it was conclusively demonstrated that the hookworm protein bound to C1q immobilized in the wells of a microtitre plate under physiological conditions and in a dose-dependent manner (Figure 3a). In a complementary assay, binding also occurred when calreticulin was immobilized in the plate (Figure 3b). However, these results only suggest that calreticulin may interfere with C1q function. Therefore, the ability of recombinant calreticulin to interfere with C1q-mediated haemolysis was assessed.
To study the impact of calreticulin binding to C1q on biological function, a haemolysis assay was established, using a method previously described (5). In brief, to obtain a correlation between haemolysis and complement activity, different amounts of human C1q were added back to C1q-deficient serum and the percentage of haemolysis determined. 5 µg/ml C1q was found to cause the requisite 40% haemolysis (Figure 3c). In this range, inhibitory and activating effects on C1q function can be measured equally. Using this assay, 0–3 µg of N. americanus calreticulin decreased the haemolytic activity of human C1q from 40% to 0%, respectively (Figure 3d). Recombinant S. mansoni calmodulin used as a control calcium binding protein showed no inhibition of haemolysis under the same conditions.
Surface plasmon resonance analysis of N. americanus calreticulin
Recombinant N. americanus(Figure 4a) calreticulin interacted weakly with αIIb and α5 cytoplasmic tail peptides of integrin α subunits but not with α2 or αv. This interaction was shown to be dose-dependent and in the case of N. americanus calreticulin Ca2+-independent (Figure 4b). Mutation or deletion within the highly conserved KVGFFKR membrane-proximal sequence of αIIb abolished the binding of N. americanus calreticulin (Figure 4c).
The specificity of the polyclonal rabbit anti-N. americanus calreticulin antiserum
As a prelude to further investigate the properties of the native protein, purified recombinant calreticulin was used to raise a polyclonal anti serum in a NZW rabbit. On Western blots, following absorption with E. coli lysate, the rabbit antiserum recognized recombinant N. americanus calreticulin (Figure 5a, lane 5). No cross-reactivity was observed with recombinant human calreticulin L, P, and N-domains [fused to maltose binding protein (5)] or full-length rabbit calreticulin (Figure 5a, lanes 1–4, respectively). Similarly, rabbit antiserum raised against recombinant human calreticulin recognized recombinant human L, P, N-domains (Figure 5b, lanes 6–8) but failed to cross-react with full-length rabbit calreticulin or N. americanus calreticulin (Figure 5b, lanes 9–10, respectively).
Surface plasmon resonance analysis analysis also demonstrated the high specificity of the polyclonal antiserum raised to N. americanus calreticulin. A strong and stable binding was monitored when the antiserum was brought into contact with immobilized N. americanus calreticulin, whereas no interaction was noticed with human calreticulin-coated sensor chip (Figure 5c).
Expression pattern of N. americanus calreticulin
Rabbit polyclonal anti-N. americanus calreticulin serum was used in Western blot analysis to screen homogenates and ES products of L3, L4 and adult worms for the presence of calreticulin. No reactivity was seen in L3 homogenate or ES products (Figure 6a, lanes 1 and 2, respectively). Calreticulin was observed in L4 and adult homogenate (lanes 3 and 5). Reactivity was observed as a signal at approximately 85 kDa in the ES products of these two life cycle stages (lanes 4 and 6). Antibody affinity purified N. americanus calreticulin was used as a positive control (lane 7).
Reverse transcriptase (RT)-PCR analysis (Figure 6b) demonstrated the presence of calreticulin in L3, L4 and adult worms (lanes 6–8, respectively). Negative control reactions included PCR on stage-specific RNA not subjected to reverse transcription (lanes 2–4) and PCR with no template (lanes 5 and 9). The previously cloned calreticulin gene was used as a positive control template to verify PCR conditions (lane 10).
Necator americanus is exclusively a nematode parasite of humans (18). It is at each stage of its life cycle in intimate contact with the immune system, as it traverses the skin, migrates systemically to the lungs, then establishes a haematophagous existence at the mucosal surface of the small intestine. Infections are long-lived and recurrent in human populations (19,20), and are relatively asymptomatic, indicating a balanced host-parasite relationship probably involving the deployment of an integrated immune evasion strategy by the parasite. Research into the molecular basis of this strategy has begun to bear fruit, with the identification of a secreted metalloproteinase with a selectivity for the inflammatory chemokine eotaxin (21), a truly secreted and biochemically authentic superoxide dismutase (22), and the identification of a potential voltage-gated T cell potassium channel antagonists in EST data sets, which may be involved in killing activated human T cells by apoptosis (23). However, given the life style of N. americanus, with its apposition to the immune system, it is highly likely that the nematode will have evolved complementary strategies to interfere with pro-inflammatory pathways mediated by complement fixation, and cell migration and activation. In the latter context, it is important that N. americanus does not possess a neutrophil inhibitory factor (NIF, a CD11b antagonist) homologue per se (24), and may need an alternative strategy to interfere with integrin function.
The identification of a potentially biologically multifunctional calreticulin-like molecule in a cDNA expression library, selected immunologically using worm-specific IgE probes from infected patients (6) led to the hypothesis that hookworm calreticulin might be expressed as a component of a unified immune evasion strategy, given the documented ability of human calreticulin to interact with C1q and the cytoplasmic signalling domains of some integrins (2). The data presented herein highlight a strategy that can be used to over express and purify a biologically active recombinant calreticulin from a nematode parasite.
The most striking data illustrate the ability of N. americanus calreticulin to interfere in a functional sense with the human complement system, and to have the potential to interfere with cell signalling events in platelets and other leucocytes. The ability of hookworm calreticulin to interfere with complement function is not surprising. N. americanus calreticulin shares a homology of 66% with its human counterpart and furthermore the C1q binding sequences/motifs present in the human N- and P-domains seem well conserved in the hookworm molecule (6). C1q binds to the CH2 domain of IgG via the motif ExKxK, with possible replacement of glutamine by threonine or asparagine and of lysine by arginine (25). The amino acid sequence of N. americanus calreticulin comprises five out of the six predicted binding motifs of the human homologue, VQSKHKSDY (residues 37–45), FSYKGKN (144–150), KDIRCKDD (155–162) present in the N-domain, WDEREYIDD (214–222) and GEWKPRQ (248–255) in the P-domain (bold amino acids are charged or polar residues participating in the putative binding motif, amino acids homologous to human calreticulin are underlined). N. americanus calreticulin lacks the human C1q binding domain DEEKDKG, but contains an additional binding motif GEWKPRQ in the P-domain). In human calreticulin, PEPSCAN showed that KDIRCKDD bound and inhibited the haemolytic function of C1q most efficiently (26), a sequence which is faithfully reproduced in N. americanus calreticulin.
Recombinant N. americanus calreticulin also binds to αIIb and α5, but not to α2 or αv, integrin cytoplasmic domain peptides in a dose-dependent and calcium-independent manner. The interaction was transient, and in the case of αIIb dependent on conservation of the KVGFFKR motif within the αIIb cytoplasmic tail peptide. The sequence of the membrane-proximal domain is highly conserved among α integrin subunits (α2, KLGFFKR; α5, KLGFFKR; αIIb, KVGFFKR; αv, RMGFFKR) and this result indicates the high specificity of the sequence necessary to support calreticulin binding. This might suggest that hookworm calreticulin, should it gain access to the interior of the cell, would interfere with cell activation particularly as early indications are that the affinity of N. americanus calreticlin for αIIb and α5 is of the same order of magnitude as that recorded for human calreticulin (data not shown). The calcium independence of the biological events described would support the observation that N. americanus calreticulin does not, or does not need to, bind to calcium to exert biological function. Nevertheless, experiments will be repeated in due course using native hookworm calreticulin (Figure 1g), affinity purified from worm extracts using a mono-specific polyclonal antibody (ratified by surface plasmon resonance analysis) raised to the recombinant molecule. This antibody will also be used to probe leucocytes and platelets incubated in calreticulin, to determine whether the molecule is internalized.
This antibody was also used to probe Western blots of worm extracts from different life cycle stages, to investigate stage-specific expression and to probe for calreticulin in worm secretions. Preliminary data would indicate that calreticulin, as a molecular chaperone, is present in an as yet undefined molecular form in L3 homogenate and in the secretions of L4 and adult stages. This apparent appearance in secretions could explain the allergenicity of calreticulin in human infection, and suggests that this immune response is induced by actively secreted material, not by dead or dying worms. Nevertheless, these possibilities will be further investigated as this study progresses. IgE responses in infected populations do appear to impart a degree of resistance to N. americanus (27), and it is possible that this protection is linked to worm attrition in the tissues, and the subsequent release of calreticulin. The presence of mRNA for calreticulin in all life cycle stages presumably reflects the presence of message for ER-resident and not secreted calreticulin in the case of N. americanus larvae.
Immunoreactive secreted calreticulin resolves at 87 kDa on reduced Western blots, indicating as yet unresolved molecular complexity. If it is actually secreted, this anomalous migration would suggest that calreticulin is covalently complexed with a molecule cosecreted with and possibly chaperoned by calreticulin. The presence of a transglutaminase (unpublished data) in hookworm secretions could explain unusual covalent associations under conditions of laboratory culture. Experiments are in progress to investigate each of these possibilities. We have either a novel secreted calreticulin, an immunological indicator of worm attrition in infected patients, or both. Secretory forms of calreticulin have been identified in ticks and it is found on the cell surface in some instances (28–30).
In summary, we are beginning to accumulate molecular data that could explain the longevity of hookworm infections in humans, with the caveat that much of the data described has yet to be translated to an in vivo context. Nevertheless, significant strides have been made towards a full understanding of this intriguing host–parasite relationship, and the development of further molecular information about N. americanus will hasten this process.
The authors would like to thank Ralf Füllkrug, Ewald Beck for their assistance in the cloning and expression of N. americanus calreticulin. This work was supported by EC Biomed Grant No BMH4-CT98–3517.