Specific pathogen free, 8-10 week old female BALB/c mice (Harlan, UK) were maintained in individually ventilated cages on diet 41b ad lib in a 12 h:12 h light-dark cycle. All experiments were carried out in accordance with the animals (Scientific Procedures) Act 1986, and were approved by the UK Home Office inspectorate and institutional review committee.

Pcc clone AS was originally isolated from thicket rats (Thamnomys rutilans) and was cloned by serial dilution and passage 36 . Parasites were recovered from frozen blood stabilates by passage through donor mice. Experimental parasite inoculations were prepared from donor mice by diluting blood in calf serum solution (50% heat-inactivated foetal calf serum, 50% Ringer's solution [27 mM KCl, 27 mM CaCl2, 0.15 M NaCl, 20 units heparin per mouse]). Each mouse received 0.1 ml of inoculum intraperitoneally (i.p) corresponding to an infective dose of 1 × 10⁶ or 1 × 10⁵ parasitized red blood cells (RBC), depending on the experiment. An inoculum of naïve RBC was given as a control for erythrocyte proteins.

The filarial nematode Ls was maintained by cyclical passage between gerbils (Meriones unguiculatus) and mites (Ornithonyssus bacoti) as described previously 37 . Infection was initiated by subcutaneous (s.c) injection of 25 infective (L3) larvae. For co-infection experiments in which the influence of malaria on chronic nematode infection was addressed, 1 × 10⁶ Pcc parasitized RBC were introduced i.p on Day 60 of an established Ls infection and mice were sacrificed on day 20 post-Pcc infection, as described previously 21 . Whole blood was collected from the brachial artery and serum recovered after clotting at room temperature.

Nb worms were maintained by serial passage through Sprague-Dawley rats. L3 larvae were obtained by culturing the faeces of infected rats at 26°C for a minimum of 5 days 38 . For acute nematode-malaria co-infection, infection was initiated by s.c injection of 200 infective (L3) larvae on the same day that Pcc was introduced by inoculation i.p of 1 × 10⁵ parasitized RBC. Mice were sacrificed on Day 20 post-infection under terminal anaesthesia. Whole blood was collected from the brachial artery and was separated using Sera Sieve (Hughes & Hughes Ltd).

Two malaria antigens were used in this study: a recombinant protein and a crude antigen homogenate prepared from parasitized erythrocytes. The recombinant Merozoite Surface Protein-1₁₉ (MSP-1₁₉) was originally sequenced, cloned and expressed from Pcc AS clone, as described previously 39 . In brief, the MSP-1₁₉ nucleotide sequence was inserted into Pichia pastoris vector pIC9K and protein expression carried out in Pichia pastoris strain SMD1169. This antigen was used in ELISA at a concentration of 1 μg/ml.

The crude malaria homogenate - lysed Pcc parasitized red blood cell extract (pRBC) - was prepared from whole blood of mice with a parasitemia in excess of 20%. Mice were bled by cardiac puncture with a heparinised syringe and blood stored at -80°C prior to 3 rounds of freeze-thaw to lyse the parasitized red blood cells. The lysed cells were sonicated, on ice, twice for 30 sec at 10 Amp and centrifuged at 16060 g for 10 min. The supernatant was stored at -80°C. Similarly, a naïve red blood cell extract (nRBC) was prepared as a control for RBC proteins; responses to this antigen amongst infected mice were indistinguishable from naïve (data not shown). In the Ls experiments this antigen was used in ELISA at 0.5 μg/ml and in the Nb experiments at 5 μg/ml.

Ls and Nb extracts (LsA and NbA) were prepared by homogenisation of adult nematodes in PBS. The somatic extracts were centrifuged at 1000 g for 20 mins and the pellet discarded. The extract was stored at -20°C. LsA was used in ELISA at 0.5 μg/ml and NbA at 5 μg/ml.

ELISA was used to measure antigen-specific IgG antibodies in the serum of nematode-infected, Pcc-infected or co-infected mice. In the Pcc-Ls study, sera were added in a serial dilution 1/100 - 1/400 and a dilution was then chosen whereby all samples fell in the linear range of the curve; for IgG1, a dilution of 1/200 and for IgG2a 1/100. For the subsequent Pcc-Nb study, serum samples were added in a serial dilution 1/50 - 1/819200. Antibody titres were calculated as the reciprocal of the greatest dilution at which optical density (O.D) was greater than the mean plus 3 standard deviations of the O.D values observed for control mouse sera at 1/200 dilution.

Antibody responses to MSP-1₁₉, pRBC, NbA or LsA were determined for IgG isotypes IgG1, IgG2a, and IgG3. 96 well maxisorp immunoplates (Nunc) were coated at 4°C overnight with either recombinant or crude antigens at the concentrations indicated (see Antigens section) in 0.06 M carbonate buffer (0.04 M NaHCO₃, 0.02 M NaCO₃, pH9.6) in a final volume of 50 μl per well. Non-specific binding was blocked with 5% FCS in carbonate buffer (200 μl/well) for 2 hours at 37°C. Wells were washed three times in Tris buffered saline with 0.1% Tween (TBST) after each step. Serum samples were added in serial dilutions as indicated using TBST as a diluent, in a final volume of 75 μl per well and incubated for 2 hours at 37°C. Isotype specific detection antibodies were diluted in TBST in a final volume of 50 μl per well. For IgG1, HRP conjugated goat anti-mouse IgG1 (Southern Biotech 1070-05) was used at 1/6000, HRP conjugated goat anti-mouse IgG2a (Southern Biotech 1080-05) at 1/4000 and HRP conjugated goat anti-mouse IgG3 (Southern Biotech 1100-05) was used at 1/1000. Plates were incubated for 1 hour at 37°C. An additional wash in distilled water was carried out before developing with ABTS peroxide substrate (Insight Biotechnology), 100 μl per well, at room temperature for 20 minutes. O.D was read at 405 nm using a spectrophotometer.

Polyclonal IgE levels were determined by sandwich ELISA. 96 well maxisorp immunoplates (Nunc) were coated overnight at 4°C with 100 μl of IgE capture antibody (2 μg/ml; clone R35-72 Pharmingen) diluted in carbonate buffer. Plates were blocked with 5% non-fat skimmed milk in carbonate buffer for 2 hr at 37°C. Plates were washed 5 × in TBST before addition of sera at 1/10 and 1/20 dilutions in a final volume of 50 μl/well and left overnight at 4°C. For the standard curve two- fold serial dilutions of purified mouse IgE, κ monoclonal isotype standard (Pharmingen) were used. After 5 washes in TBST, 100 μl of biotinylated detection antibody (2 μg/ml; clone R35-118 Pharmingen) diluted in TBST with 5% FCS was added and plates left at 37°C for 90 mins. Plates were washed 5 × in TBST prior to incubation with ExtrAvidin peroxidase (SIGMA), diluted 1:8000 in TBST with 0.5% FCS, for 30 mins at 37°C. After a final wash in distilled water, plates were developed with 100 μl TMB microwell peroxidase substrate system (Insight Biotechnology Ltd) and read at 650 nm.

In order to determine the extent to which carbohydrate or protein moieties contributed to the antibody response the antigens were pre-treated with periodate. Antigen-specific IgG1, IgG2a and IgG3 antibodies were measured in response to antigens treated with periodate. The ELISA was carried out as detailed for the Nb co-infection experiment with untreated antigens but the following additional steps were included after blocking with 5% FCS: carbonate buffer, prior to sample addition. TBST wash (×3) was followed by the addition of 10 mM sodium (meta) periodate diluted in 50 mM sodium acetate in a final volume of 100 μl/well. Plates were incubated at 37°C for 1 hour and then washed in 50 mM sodium acetate. To stop the activity of periodate, 100 μl of 50 mM sodium borohydride solution was added to each well.

General linear statistical models allowed us to frame and test questions such that we could determine whether differences in infection status and/or presence of carbohydrate antigen explained the observed variation in antibody responses. For more detailed explanation of the statistical methods employed see Grafen and Hails 40 . Infection status and treatment with periodate (or not) were included as categorical factors and their ability to predict antibody response was formally evaluated via Analysis of Variance (ANOVA). The serial dilution of sera in an ELISA produces ordinal data, which were log₁₀ transformed prior to analysis to ensure the data were approximately normally distributed, in accordance with the requirements of linear models. Analyses were carried out using the statistical package JMP 5.1 (SAS). The maximal model was fitted first and minimal models were obtained by sequentially removing non-significant terms (P-value > 0.05), beginning with interactions. Finally, whenever a factor was significant (P < 0.05), an All Pairs Tukey post-hoc test was carried out to identify which groups of mice differed significantly in antibody induction, with respect to infection status or periodate treatment.

We had previously used a model of Pcc-nematode co-infection in which Pcc is introduced into mice with pre-existing chronic Ls infection to investigate the dynamics of infection with regard to parasitological outcome and cytokine bias 21 . In the Pcc-Ls model the peak of malaria parasitemia was controlled by day 10 and resolved by day 14 post Pcc-infection. For this study, we envisaged that analysis of the antibody isotype response from these mice would provide a method to rapidly and quantitatively assess immune bias. We thus asked whether we could use antibody isotype ELISA for IgG2a and IgG1 to address whether nematode infection would skew the Th1 cell response to Pcc to a more Th2 cell biased response. Conversely, we wished to address whether the powerful Th1 cell response induced by malaria would have the capacity to alter an established IgG1 response to a nematode infection. As is common practice in many studies 41 42 43 44 45 46 47 48 , especially with large sample sizes as in this study, the ELISAs were performed with a fixed serum concentration derived from the linear point in a dilution series.

As expected from previous studies 49 50 51 , we were able to detect an IgG1 response against LsA in Ls mice. Co-infected (Pcc-Ls) mice also produced LsA-specific IgG1, but it was reduced in magnitude compared to the Ls mice (Fig 1Ai). Thus co-infection with Pcc appeared to down-regulate the anti-Ls specific IgG1 response. It also appeared that responses in Pcc-Ls mice were further biased toward a Th1 cell response through the induction of IgG2a to LsA (Fig 1Aii). Also as expected 32 52 , Pcc mice mounted highly biased Th1-cell responses as indicated by the predominance of Pcc (pRBC)-specific IgG2a over IgG1 (Fig 1B). Once again, Pcc-Ls mice appeared to alter this bias by increasing the amount of pRBC-specific IgG1 in comparison to the Pcc mice (Fig 1Bi).

At first glance, the results strongly suggested that the isotype and hence cytokine bias of each single-species infection was significantly impacted by co-infection. In support of this, polyclonal IgE was highest in Ls mice and absent in Pcc mice with Pcc-Ls mice exhibiting an intermediate level (Fig. 1C) although Pcc-Ls mice did not differ statistically from Ls mice in this polyclonal analysis. However, it was apparent that Pcc mice were exhibiting sizable antibody responses to LsA (X₁ in Fig 1Aii) and conversely Ls mice were exhibiting strong antibody responses to pRBC (X₂ in Fig 1Bi). Western blot analysis confirmed that these Ls-induced cross-reactive responses were directed against the parasite rather than the RBC (data not shown). We thus had to ask whether the shift away from LsA-specific IgG1 toward IgG2a responses in the Pcc-Ls mice was due to the influence of IFN-γ on the Ls-induced response or simply reflected the presence of cross-reactive Pcc-induced IgG2a responses to LsA. Conversely, was the apparent increase in pRBC-specific IgG1 responses in Pcc-Ls mice due to Ls-induced antibodies that cross-reacted with infected red blood cells? Because serum titres had not been determined in this study, we could not assess the relative strength of cross-reactive versus antigen-specific responses. Our observation that there was significant cross-reactivity at the sera dilution tested thus confounded our ability to interpret any changes in immunological bias during Pcc-Ls co-infection.

In order to address the utility of antibody isotype responses further, we embarked on co-infection experiments with Pcc and the nematode Nb. Because Pc-Nb is an acute model, whereby the nematode is cleared by day 7 and the peak of malaria parasitemia is controlled by day 10, the antibody data could be collected after only 20 days of co-infection, a more practical time frame than the 80 days required for the Pcc-Ls experiments. This also allowed us to address whether our observations of antibody cross-reactivity were a more general feature of Pcc-nematode infection. Given the apparent cross-reactivity observed at a fixed dilution of sera in the Pcc-Ls ELISA we used endpoint titres derived from a serial dilution (1:50 - 1:819200) in the Pcc-Nb assays to address whether this readout would overcome cross-reactivity problems. To determine if the specificity of the assay could be improved with the use of recombinant antigens we also included the malaria protein, MSP-1₁₉ 39 not available to us for the Ls studies.

The antibody responses we observed on Day 20 of Pcc-Nb co-infection (Fig 2) paralleled those we had seen in the Pcc-Ls experiments at Day 80. For example, as seen in Fig 2Aiii, Nb mice made IgG1 biased responses against NbA, and responses in Pcc-Nb mice were intermediate between Nb and Pcc mice. In addition, Pcc mice mounted a strong MSP-1₁₉-specific IgG2a response that was reduced in Pcc-Nb mice (Fig 2Bi). As before, levels of polyclonal IgE in Pcc-Nb mice were intermediate (data not shown). We again observed cross-reactivity, whereby Nb mice mounted detectable IgG1 and IgG2a responses to both recombinant and crude malaria antigens (indicated by X_1&2 in Fig 2A and X_4&5 in Fig 2B, respectively). The magnitude of the Nb-induced IgG2a cross-reactive response is particularly striking with titres against crude and recombinant malaria antigens reaching 2500 and 200 respectively. Similarly, Pcc mice mounted responses to NbA (X₃ in Fig 2A and X₆ in Fig 2B). It is important to note that these titres although low are markedly greater than background responses (mean plus 3 standard deviations of serum responses from control mice), which are represented as zero on the y-axis. The immune bias that is apparent in serum antibody isotype responses is fully supported by cytokine responses in the lymph nodes of Pcc-Nb infected mice as we have recently described 53 . Of interest, no cross-reactivity was observed at the T-cell level.

The analysis of both Pcc-Ls and Pcc-Nb co-infection indicates that the issue of cross-reactivity is a factor investigators are likely to routinely encounter. Determining the qualitative and quantitative aspects of the cross-reacting antibody responses are not only important for the practical analysis of immune deviation but could be of real biological relevance during co-infection.

As expected, antibody responses were biased, in terms of isotype, by infection status. The bias in isotype due to a particular infection (Th2 associated IgG1 induced during Nb infection, for example) was extended to non-specific antigens, as seen in the IgG1 response of Nb mice to both MSP-1₁₉ and pRBC (X₁ and X₂ in Fig 2Ai + 2Aii). However, Pcc-specific IgG2a titres in Pcc mice were significantly higher than the cross-reactive response induced in Nb mice (Fig 2Bi). Thus, although Nb mice made cross-reactive IgG2a responses, these were no longer detectable with increasing dilution of sera (Fig 2Bi). In this case capitalising on the differences in strength of antigen-specific and cross-reactive responses clarified interpretation of immune bias in co-infected mice. Similarly, IgG1 responses to NbA were significantly higher in Nb mice than the cross-reactive response induced by Pcc mice (Fig 2Aiii). However, titre of Pcc-induced cross-reactive IgG2a to NbA did not differ significantly from Nb mice (X₆ in Fig 2Biii).

We can conclude from this analysis that cross-reactive IgG1 responses to NbA were only detectable at dilutions less than 1:100 and thus higher dilutions may be used to avoid cross-reactivity when assessing antigen-specific antibody isotype profiles for the purpose of interpreting immune bias. However, increasing sera dilution did not always overcome the cross-reactivity observed, as IgG2a responses to NbA in Pcc mice were still observed at 1:2500. This cross-reactivity warrants further investigation, as it is likely to be important biologically. Indeed even cross-reactive responses detectable only at high serum concentrations may still have functional relevance in vivo.

Antibody cross-reactivity in a broad range of systems can be attributed to reactivity with carbohydrate determinants 54 55 . Additionally, the IgG3 isotype is often associated with recognition of carbohydrates 56 and we observed cross-reactive IgG3 antibody responses during Pcc-Nb co-infection (Fig. 3). Nb mice mounted IgG3 responses to both MSP-1₁₉ and pRBC antigens, achieving titres of 200 and 3200 respectively (Fig 3Ci and 3Cii). Pcc mice mounted similar cross-reactive IgG3 responses to NbA (Fig 3Ciii). We thus chose to assess whether cross-reactivity in our Pcc-Nb co-infection system could be overcome by periodate treatment of the parasite antigens. Periodate oxidises carbohydrate to aldehydes, thus disrupting carbohydrate epitopes, which allowed us to distinguish if cross-reactive responses target the carbohydrate or protein moiety of an antigen. This could be of particular importance where detection of cross-reactive responses was not overcome by increasing serum dilution. In addition to clarifying the interpretation of shifts in immune bias, determining whether induction of specific isotype responses is driven by protein or carbohydrate recognition has important implications for vaccine design and diagnostic serology.

For the pRBC and MSP-1₁₉ antigens, periodate treatment did not significantly affect recognition by IgG1 antibodies (P ≥ 0.7). Periodate treatment of NbA however, significantly reduced anti-NbA IgG1 titres across all infection groups (Fig 3Aiii), suggesting common recognition of a carbohydrate moiety. In particular, the cross-reactive recognition of NbA, by IgG1 antibodies from Pcc mice, was ablated (X₇ in Fig 3Aiii).

Treatment of MSP-1₁₉ antigen with periodate resulted in a significant increase in IgG2a detected in sera from Pcc-Nb mice. Detection of IgG2a in singly-infected mice also followed this trend but was not statistically significant (Fig 3Bi). These results suggest recognition of a protein epitope on the recombinant antigen previously masked by glycosylation. Plasmodium species lack the glycosyltransferases required for any glycosylation other than attachment of GPI anchors 57 58 . However, inappropriate glycosylation of the recombinant protein can occur in the Pichia expression system 57 . The increase in protein-specific responses, following periodate treatment of the recombinant antigen (MSP-1₁₉), may thus reflect the response to the natural non-glycosylated parasite protein seen during infection. IgG2a recognition of periodate treated pRBC antigen was enhanced in sera from Pcc mice, though not in the Pcc-Nb or Nb mice (Fig 3Bii). This suggests the cross-reactive IgG2a response is not targeting carbohydrates, which could potentially be conserved amongst parasite antigens.

For NbA, it was only after treatment with periodate that differences amongst infection groups become apparent, whereby significantly greater amounts of (cross-reactive) IgG2a antibodies were detected in Pcc mice in comparison to Nb mice (X₈ in Fig 3Biii). Thus, the Pcc-induced cross-reactive IgG2a response to NbA (X₆ in Fig 2Biii) that was not lost with serial dilution was also maintained following periodate treatment (X₈ in Fig 3Biii). In contrast, the Nb-specific IgG2a response appears to target carbohydrate as periodate treatment reduced recognition of NbA in both Nb and Pcc-Nb mice (Fig 3Biii) although only statistically significant in the Pcc-Nb mice. This may indicate that atypical Th1-type responses to helminth antigens are driven by carbohydrate.

Disruption of carbohydrates via periodate treatment significantly affected the recognition of Pcc antigens by IgG3 antibodies. In particular, recognition of periodate treated pRBC was reduced in serum from all mice that had experienced Pcc infection; this was most evident in the Pcc-Nb mice (Fig 3Cii). The recognition of periodate treated MSP1-₁₉ was also significantly reduced in Pcc mice (Fig 3Ci). The more pronounced reduction in IgG3 response to the treated pRBC antigen may reflect the greater proportion of carbohydrate components in this crude antigen preparation in comparison to the single recombinant protein.