Male CBA/J mice were obtained from The Jackson Laboratory and were maintained at the American Association for Accreditation of Laboratory Animal Care, approved animal facility to the Centers for Disease Control and Prevention (CDC, Atlanta, USA) in accordance with institutional guidelines and federal regulations. The mice were infected by subcutaneous injection of 45 S. mansoni cercariae and liver samples were collected for uninfected control mice (n=5) and mice with 6 week (n=10), 8 week (n=10), 12 week (n=10) and 20 week (n=10) infections within the same time period and snap frozen at −80 °C. The 6, 8, and 12 week infected animals were randomly selected while the 20 week infected animals were selected based on their classification as having MSS or HSS (n=5 per group) upon necropsy [5]. Splenomegaly was expressed as percent spleen to body weight ratio (%SBW) and hepatomegaly expressed as percent liver to body weight ratio (%LBW). Two mice from each of the 8 and 12 week infection groups had noticeably higher %SBWs than other mice in the groups and thus were classified as pre-HSS (circled icons, Fig. 1A) and hence the infected mice from HSS (n=5), MSS (n=5), 12 week (n=8), 8 week (n=8) and mice with pre-HSS (n=4) were compared to uninfected control mice (n=5). All experiments performed were approved by Institutional Animal Care and Use Committee of the CDC as well as the The Victoria University of Wellington Animal Ethics Committee.

Bio-Rad Protein Assay (Bio-Rad Laboratories, Life Science Group, USA) was used for mouse liver lysate total protein measurements (as per the manufacturer’s guidelines).

The liver lysate was subjected to 2D-DIGE using the Minimal CyDye Kit (GE Healthcare Bio-Sciences AB, Uppsala, Sweden) as per the manufacturer’s guidelines. The experimental design is described in Table S1 of supplemental material. The sequentially scanned 2D-DIGE gel images were analysed using DeCyder^™ 2-D (v 6.5, GE Healthcare). Data normalisation and statistical analyses were carried out using DeCyder software. The Biological Variation Analysis (BVA) module was used to match 69 gel images for pI ranges 4–7 and 6–11 each. The BVA compared the protein average volume ratios between study groups that changed by greater than 2-fold and one-way analysis of variance p value (1-ANOVA) ≤0.01 and passed the False Discovery Rate (FDR) feature; these spots were called ‘protein spots of interest’. FDR feature eliminates false positive results, which are 1% of the significantly changed (1-ANOVA≤0.01) protein spots [9]. The Extended Data Analysis module was used for multivariate analyses which included the Principle Component Analysis (PCA) and Hierarchical Cluster Analysis (HCA) features as done previously [5]. We analysed the data in accordance for a target power of 0.8 [10]. The protein spot volume data for individual spots was extracted using the XML Toolbox module as described previously [11] to evaluate the mean and standard deviation measurements provided in Table S2 of the supplemental material.

Protein spots of interest were excised from preparative gels for tryptic digestion using an Ettan^™ Digester (GE Healthcare). Tryptic digests were analysed using cyano-4-hydroxycinnamic acid (CHCA; Sigma-Aldrich) matrix and MALDI-TOF/TOF 5800 (Applied Biosystems, Foster City, CA) mass spectrometer operated in positive ion mode and continuous stage motion. Each spot was externally calibrated using TOF/TOF Calibration Mixture (Applied Biosystems), containing des-Arg¹-Bradykinin (m/z 905.05), Glu¹-Fibrinopeptide B (m/z 1571.61), angiotensin I (m/z 1297.51) and fragments of the adrenocorticotropic hormone 1–17 (m/z 2094.46), 18–39 (m/z 2466.72) and 7–36 (m/z 3660.19). MS spectra acquired over a mass range of 800–3500 m/z with a total of 400 laser shots/spectrum for each spot. A maximum of 21 precursors with S/N over 50 were selected for fragmentation. The MS/MS spectra were collected using air as the collision gas and were calibrated using angiotensin I fragmentation. For protein identification, all MS/MS data was submitted to Protein Pilot 3.0 software (Applied Biosystems) using MASCOT (v2.2.04, Matrix Sciences Ltd., UK) with IPI Mouse protein database v3.73 and search parameters as described previously [5]. The search results can be viewed at PRIDE (http://www.ebi.ac.uk/pride/) with accession numbers: 19316–19324.

Liver lysate proteins from uninfected control and 6 week, 8 week, 12 week, 20 week infected mice were transblotted onto Hybond^™-LFP membrane (GE Healthcare) and probed with antibodies for cytokeratin 18 and transferrin as described previously [5].

The DeCyder ‘2-fold change, 1-ANOVA≤0.01, with FDR’ results and the Swiss/UniProt accession numbers of the identified protein spots were evaluated using MetaCore^™ v6.7 (GeneGo, St. Joseph, MI, USA) specific for the liver organ and Mus musculus species. The MetaCore^™ computes a score for each possible network according to the input protein list. The score is based on the hypergeometric distribution and is calculated with the right-tailed Fisher’s Exact Test. The score is the negative logarithm to the base-10 of the Fisher’s Exact Test p value that indicates the probability of the input protein list in a given network being found together as a result of random chance. The analysis generates a list of significant networks based on the input protein list. Top scored significant networks with high p-values are noted.

GraphPad Prism (v4.0, GraphPad, San. Diego, CA, USA) was used for normalisation and statistical analysis.

Splenomegaly (%SBW) and hepatomegaly (%LBW) are marked features of chronic schistosomiasis and reflect spleen and liver pathology during the disease. After log transformation of %SBW to normalise the data, comparison of the %SBW between the six study groups showed HSS mice were statistically different (p≤0.0001) from the other five groups (Fig. 1A). %LBW was elevated in all five groups compared to uninfected control mice (Fig. 1B). Linear regression analyses demonstrated significant correlations between %LBW and %SBW at 6, 8, 12 and 20 weeks of infection (Fig. 1C) during schistosomiasis.

Using the liver lysates, 2D-DIGE was performed to determine how the proteomic signatures altered over the course of schistosome infection. For the two pI ranges, the number of ‘protein spots of interest’ was 299 and 136 respectively. The highest number of changes compared to uninfected mice was observed in pre-HSS mice with 85 (pI 4–7) and 111 (pI 6–11) protein spots that had greater than 2-fold change in average volume ratio that were statistically different (1-ANOVA≤0.01) among the 5 infected groups (taken together) compared to uninfected control mice, and were not eliminated upon application of the DeCyder software FDR function. Of these 196 protein spots, mass spectrometry identified 127 protein spots (Fig. 2, Table 1 and Table S2 of supplemental material). The remaining 69 protein spots were relatively low in abundance and were not identified. Importantly, the highest number of changes compared to uninfected mice was observed for pre-HSS mice, followed by 12 week infected mice, with a total of 110 spots in both pI ranges that had a greater than 2-fold change and 1-ANOVA≤0.01. Of these, 82 spots decreased while 28 protein spots increased in abundance (Fig. 1D).

To validate the protein spot data generated by 2D-DIGE, we assessed the protein abundance of two proteins (transferrin and cytokeratin) that showed distinct changes in abundance over the course of infection. Using linear regression analyses for cytokeratin 18 and transferrin 2D-DIGE spot volume data versus the western blot analysis band volume data, we found significant relationships between the two volumes (r²=0.2647, p<0.001 and r²=0.8494, p<0.001 respectively). Overall, this analysis indicated that similar results were generated from two independent experimental techniques involving the same samples (Figure S1 A, B of Supplementary material) thus supporting our use of spot volume data as a measure of protein abundance.

To further assess the disease-form specific changes, we compared the expression pattern of proteins that changed significantly in pre-HSS, MSS, and HSS mice. We found that 13 protein spots in pre-HSS and 10 protein spots in HSS mice were specific for those 2 groups (Table 2). No protein spots were exclusive to MSS mice. These results indicated that the pre-HSS protein spots may be predictive of pathology in the mice that follow the severe disease course. Moreover, the abundance of protein spots (n=60) shared by the pre-HSS and HSS mice suggests that the process that promotes severe disease pathology has already started in the pre-HSS mice.

Multivariate analysis using principle components analysis (PCA) showed good separation of the study groups (each spot map represents the liver 2D-DIGE image for individual mice) for protein spots in pI range 4–7 (Fig. 3A). Uninfected control mice, 6 week infected mice, MSS and HSS mice demonstrated distinct clusters, with the 6 week infected mice clustered very closely to the uninfected controls. In contrast, while most 8 week and 12 week infected mice were clustered into distinct groups, the 8 week and 12 week mice that had been designated pre-HSS clustered with the HSS mice (Fig. 3A). Similar results were obtained for the pI 6–11 range PCA plot, but the clusters were less distinct because the proteins were more difficult to resolve on gels (data not shown). Overall, these results indicate that based upon proteomic signatures, the groups form distinct clusters, with the pre-HSS mice grouped with the HSS mice.

Unsupervised, hierarchical cluster analysis (HCA) segregated the study groups according to the individual disease status of each mouse based on proteins in the pI range 4–7 (Fig. 3B). The pre-HSS animals (wk0805, wk0809, wk1203 and wk1205) segregated with the HSS mice in the pI 4–7 dendrogram (Fig. 3B), supporting the conclusion that these mice were in the process of developing severe disease. The dendrogram for pI range 6–11 showed similar results but again the segregation of the groups was less clear due to the decreased resolution of those gels (data not shown). Taken together, this unsupervised HCA supports the findings of the supervised PCA analysis and indicates that the pre-HSS group is more similar to the HSS group than it is to other animals with the same length of infection.

To better understand how the changes we observed impacted liver pathology and processes, we used a web-based software tool, MetaCore^™ for pathway analysis of the DeCyder data. The MetaCore^™ pathway analysis showed the unique and common proteins between the different groups in our study. In the protein interaction networks, proteins (nodes) were pulled out from the original position to make visualisation, understanding and extracting information from the network easier. When the protein lists for 12 week infected mice, MSS, pre-HSS and HSS mice were compared, the analysis revealed 50 and 79 protein spots unique to HSS or pre-HSS mice, respectively in comparison to the MSS protein list (Fig. 4A, C). Further, 60 protein spots were shared by HSS and pre-HSS mice (Fig. 4B) and 51 protein spots were shared by 12 week infected and pre-HSS mice (Fig. 4D). These results again suggest that HSS and pre-HSS mice have a common protein set that relates to the development and pathology of schistosomiasis and reinforces the data presented in Table 1. MetaCore^™ generated a list of networks for HSS (15 networks), pre-HSS (21 networks), MSS (5 networks), 12 week (17 networks) and 8 week (3 networks) mice protein spots. Of these, the top rated networks revealed similarities for HSS and pre-HSS mice proteins further supporting the supposition that pre-HSS mice were developing severe disease (Fig. 5 A, B). Additionally, the protein lists for MSS and the remaining 8 and 12 week infected mice had similar protein networks, suggesting that these mice were following a disease progression course similar to MSS mice (Figure S2 of supplemental material). The first top rated network for MSS mice (Figure S2A) was similar to the top rated network for 8 week (Figure S2B). The second top rated network for MSS mice (Figure S2C) was similar to the top rated network for 12 week infected mice (Figure S2D). Finally, comparison of the pre-HSS and HSS pathways revealed several shared core molecules: ESR1 (Estrogen receptor 1), c-Myc, p53 and c-Jun all of which are signalling proteins indicating that changes in these signalling pathways may be involved in the development of severe schistosomiasis.