(1) Comparison of gene expression during the parasite life-cycle

A very good example of the use of microarrays in parasitology has been the monitoring of the transformation of Trypanosoma cruzi from the trypomastigote to the amastigote (Minning et al. 2003). This study utilized 4400 probes principally of partly sequenced genomic material and some material from an ORF library that was made from 200–400 bp size DNA. The library used a GFP (Green Fluorescent Protein) expression-based bacterial system for selection. The T. cruzi DNA fragments were inserted upstream from the GFP gene lacking a start codon. Thus, clones expressing GFP were selected for use on the microarray. Most of the differential expression was due to the up-regulation of 60 genes in the developing amastigote. Of these, 14 were found to have been characterized previously, 25 were novel, with the remainder being redundant. A subset of genes was validated using real-time PCR that correlated well with the microarray results. The authors aimed to find vaccine targets and identify amastigote-specific genes using this approach. While a relatively large amount of novel genes were identified, the nature of the microarray construction hampered some of the goals of the project. Probe material on the chip could have been more effective if amastigote cDNA had been utilized in an EST strategy study, to enable the maximal chance of finding stage-specific gene expression. No doubt limitations in sample size and the amount of mRNA that can be isolated from T. cruzi were a considerable problem, but since this microarray was probably produced in 2002, if not earlier, recently improved cDNA library construction kits may allow this research to be further developed. This in turn may produce a significantly improved microarray. Technological advances in RNA processing, including amplification kits (such as Ambion MessageAmp™ aRNA Kit) that can start with as little as 10 ng of mRNA and library construction kits (such as Invitrogen CloneMiner) starting with only 1–5 μg of mRNA. In addition, improved selection markers,result in larger insert sizes and more rapid library construction, from small amounts of material, a common limitation when working with parasites.

The strategy of using more specific cDNA probes was utilized in a later study by Baptista et al. (2004). This group constructed a relatively small cDNA microarray (665 genes; 730 sequences) to examine gene expression and genomic organization in different isolates of T. cruzi. Firstly a comparison of the genome between strains was carried out using genomic DNA that detected differential hybridization in 68 genes, of which a subset was validated by Southern blot analysis. The Southern analysis revealed variations in gene copy number and sequence differences that contributed to the hybridization differences evident in the microarray experiments. These researchers then tested cDNA for hybridizations that yielded information on gene expression between the two strains. They found that 84 probes were differentially expressed, of which a subset was validated by Northern blot analysis. Only 20% of the genes overall had variations in both expression profile and genomic copy number (and/or sequence variation), when these two groups of differentially hybridized genes were compared. This indicated that there was no correlation between the abundance of a particular gene within the genome and the expression of that gene as RNA, demonstrating a high degree of gene expression control and regulation.

(2) Comparative gene expression between related species and evolutionary considerations

There have been few investigations of inter-specific genomic variation using microarray technology generally, and, specifically, in parasitology. There are significant limitations in using a microarray platform with limited sequence homology or incomplete sequence, that result in low efficiency hybridizations.

The limited physical length of sequence available in short cDNA or oligonucleotide probe microarrays requires a high degree of homology with the RNA that is probed in order to obtain a hybridization signal. A low level of homology will result in a limited hybridization, which may be misinterpreted as a low gene expression level. However, in either situation, differential expression on the microarray will highlight the probability of a variation in sequence identify or gene expression, which can be further investigated by sequencing efforts. Two examples of successful interspecies analysis include cDNA-based microarrays from fish – Astatotilapia (Renn, Aubin-Horth and Hofmann, 2004) and Salmo (Rise et al. 2004). Gene expression between the different fish species was, as expected, more successful in the closer-related species, than with others further diverged. In both studies the principle of using a microarray platform to examine a wide range of species and present evolutionary and ecologically relevant data were demonstrated. Other microarray studies have examined different species of Drosophila (Watanbe et al. 2004) and interspecies variations in yeast (Gu et al. 2004). The large-scale examination of genes can identify small variations that occur between closely related species or strains, making microarray analysis an excellent method for taxonomic classification. The conclusions from these inter-species microarray studies can readily be adapted to parasitology. While, as discussed previously, a number of parasites are currently subject to major genome sequencing projects, a large number of species will not have such thorough examination, at least in the medium term. Microarray analysis between species may provide useful insights into evolutionary traits. A strategy to achieve this would be to use using data reduction techniques such as principle component analysis (Randall et al. 2003) or multidimensional scaling (Dugas et al. 2004). By simplifying the microarray data, broad relationships between species could then be demonstrated. As previously mentioned, extensive differential expression may either identify genes that have diverged between species or are even absent in one of the examined species. Such divergence or total absence of genes may reflect key evolutionary differences between species. The presentation of unique genes within a species may be indicative of relatively newer evolutionary adaptations to environmental pressures.

In regard to the malaria parasites, information regarding life-cycle and inter-species studies are presented in the informative database ‘PlasmoDB – The Plasmodium Genome Resource’ [http://plasmodb.org/] (see Bahl et al. 2003 for review). In addition to presenting Plasmodium proteomics, primary sequence, and both cDNA and oligonucleotide microarray data, the web site also provides inter-correlations tools, such as plotting taxonomic divergence against sequence similarity, giving a representation of inter-species variation. Presently, the microarray data contained within the site is from P. falciparum and there are no reports of cross-species/strain hybridizations. The only comparisons made have been performed in silico from proteomic or sequence data. However, the framework and organization provided does present a suitable format for future microarray data that will arise from other species of Plasmodium that can be subjected to further analysis.

(2) Plasmodium microarray studies, leading the way in parasitology

Two landmark papers in 2003 described the construction of an oligonucleotide microarray for P. falciparum (Bozdech et al. 2003; Le Roch et al. 2003). Bozdech et al. (2003) developed software to aid in the design of 70mer oligonucleotide probes to 6000 Open Reading Frames from the parasite genome. The microarray was then used to analyse the gene expression of the trophozoite and schizont stages. As expected, major differences in the transcriptional profiles of these stages were detected. This study represents a good example where public domain sequences could be accessed, a design process followed, and a microarray produced by a non-commercial entity. The work also emphasized the advantage of in silico design of probes (as compared to using cDNA clones), so that by controlling GC content and other parameters, hybridization efficiencies can be enhanced. The work of Le Roch et al. (2003) again utilized an oligonucleotide design strategy for microarray construction but presented this on a much larger scale. In this study, a microarray (multiple chips) containing 260596 features from predicted coding regions (genomic) and 106630 probes from non-coding regions, was constructed. One of the distinct advantages of using probes designed from genomic sequences, as opposed to ESTs, is that the detection of gene expression is not limited to the life-cycle stages from which the ESTs were originally derived. This was an important consideration in this study since 9 different life-cycle stage of P. falciparum were examined using the microarray, and gene expression across the life-cycle was more effectively monitored with genome-derived probes. The authors were able to correlate and cluster genes based on similar expression levels with specific life-cycle stages to group genes with assumed similar functions. A major drawback of this study was the use of 25-mer oligonucleotides which may have had a deleterious effect on the specificity of the probes used in the hybridizations. The use of 60 mers has been reported to provide much higher specificity in hybridization for microarray analysis (Hughes et al. 2001; Summan et al. 2003).

Recent microarray studies in P. falciparum have become more focused. Specific cellular activities including cell cycle elements, overall gene regulation and factors involved in transcriptional control of gene expression, have all be examined in a recent publication by Gissot et al. (2004). PCR products ranging in size from 300–1500 bp were printed from 150 genes with homology to known proteins involved in signal transduction or the cell cycle. Differential expression of these prospective candidates was compared between a ‘normal’ clone of P. falciparum and a subclone with a defect in gametocytogenesis. When the two strains were examined during a gametocytogenesis time-course experiment, 114 genes consistently provided clear microarray data, of which 106 genes were identical in expression profile between the two clones. Eight genes were differentially expressed between the two strains during the time-course to differing extent. These experiments identified genes important in the dynamic events leading up to gametocytogenesis in P. falciparum.

Host responses to malaria were examined in an animal model, using P. cynomolgi bastianellii and rhesus monkeys (Ylostalo et al. 2005). A commercial oligonucleotide chip of the human genome (Affymetrix HG-U133A) was used to follow a time-course of infection, with RNA isolated from whole blood, before infection, during the initial liver phase, during peak parasitaemia and at the first/second relapses. A wealth of information was obtained but the most significant findings included a general down-regulation of genes during the initial liver phase of parasite infection, demonstrating a host response to the infection at the transcriptional level. In addition, ‘defensive-response’ genes were also identified by distinctive up-regulation. Some examples of these genes include α enolase (ENO1) which regulates inflammatory responses, and a NFE2L3 which regulates many erythroid specific genes.

(4) Investigation of schistosome transcriptomes with microarray technology

Highly informative microarray studies have been published for schistosomes (Hoffmann, Johnston & Dunne, 2002; Fitzpatrick et al. 2004). This group utilized cDNA microarrays consisting of ESTs from adult S. mansoni (576 features) and S. japonicum (457 features) to identify sex-specific gene expression; sex-associated genes were identified by microarray analysis and confirmed by RT-PCR. In the S. japonicum study (Fitzpatrick et al. 2004) functional studies (enzymatic assays and localization) were also presented to further demonstrate the gender-specific differences occurring at the transcriptional level. The report also presented some sex-specific gene differences that were evident between geographical isolates (Anhui and Zhejiang) of the Chinese strain of S. japonicum. More differential expression may have been detected in comparing geographical strains with more pronounced phenotypic difference such as the Philippine and Chinese strains of S. japonicum.

A recent addition to this work includes a report of the construction of a 7335 S. mansoni element microarray using a spotted oligonucleotide approach (Fitzpatrick et al. 2005). The oligo-probes were designed from 17329 S. mansoni EST sequences and represent approximately 50% of the estimated schistosome transcriptome. Again the focus of the study was to identify sex-biased genes, which resulted in the identification of 141 genes up-regulated in the female adult S. mansoni and 86 genes up-regulated in the male. Good correlations with previous data from this group were made, in addition to greatly extending the list of genes for analysis. Although the authors demonstrated strong reproducibility between their array experiments, as reflected in the R coefficient values, the confidence levels (or error modelling) for individual genes in individual microarray experiments were not presented. The authors further mentioned that all included data points were within 90% confidence limits in 3 of 5 experiments, but a higher confidence level would have provided more depth to their analysis and in the microarray in general.

Two almost simultaneous releases of EST data from S. japonicum (Hu et al. 2003) and S. mansoni (Verjovski-Almeida et al. 2003) have provided the platform for study of the schistosome transcriptome. A large proportion of the EST sequences generated still require characterization as only 45% of the S. mansoni and 65% of the S. japonicum ESTs showed similarity to sequences already in GenBank. The S. japonicum EST data were structured into 13131 clusters, including 8527 singletons of varying sizes and probably represent the majority of the transcriptome, estimated to contain approximately 15000 genes (Johnston et al. 1999). We have recently designed and utilized an oligonucleotide microarray based on the S. japonicum and S. mansoni EST data sets. The designed microarray contains 7055 S. japonicum and 12166 S. mansoni probes, effectively providing a much wider coverage of the schistosome transcriptome, which will facilitate in-depth comparative studies of the two species represented on the microarray.

(5) Microarray analysis of infected host tissues and the effects of parasitism

Another use for microarrays in parasitology is to examine the differential gene expression of host tissues in response to parasitic infections. A report by Sexton et al. (2004) described the use of a commercial oligonucleotide mouse array to monitor the effects of malaria infection. They identified hundreds of genes that were altered in response to the parasite. Modification in specific gene profiles that were related to glycolysis, immunology, and erythropoiesis were correlated with the underlying pathology of the infected host. While the study was comprehensive in terms of the microarray used, information of sampling numbers and whether tissues were pooled or not, was not provided. Significant variation in gene expression can occur between individuals, and this would be more pronounced with out-bred mice due to the wider genetic profile in the population under investigation. Fortunately, Sexton et al. (2004) used inbred C57BL/6 that would have given a more even genetic response. In pooling samples, subtle fluctuations in gene expression can be lost, or, alternatively, a single large variation in gene expression in one individual may present a misleading indication when the pooled group is examined. It is fully acknowledged that microarray experiments are expensive, but the study would have benefited if gene expression variation between individuals could have been determined. A compromise may have been to use a validation method (such as real-time PCR) to investigate individual RNA samples, in order to determine if any significant inter-individual variation occurred.

A similar study by Hoffmann et al. (2001) examined gene expression in schistosome-infected mouse liver using a cDNA microarray. They monitored various genes encoding immunological proteins in response to granuloma formation caused by S. mansoni infection. The unique pattern of gene regulation that was apparent, allowed a profile to be formulated that was able to differentiate a healthy host from an infected host by the gene expression within the liver.

A recent article by Diez-Tascon et al. (2005) has adapted a microarray resource to investigate the mechanism that leads to host immunity to parasitic infections. They utilized a bovine microarray to examine the transcriptional profile between genetically resistant and susceptible lambs infected with a mixture of naturally occurring nematode infections (including Haemonchus contortus, Trichostrongylus colubriformis and Ostertagia sp. although the precise composition of the initial infection was not clearly reported). The authors compared gene expression within the duodenum of the infected host, using a 10204-element cDNA microarray derived from a previously undescribed bovine library. It can only be assumed that a microarray derived from ovine material was not constructed since the bovine microarray was available and homology between the two species is high (average nucleotide homology 96%, Diez-Tascon et al. 2005) in the coding region. Differential expression in 126 genes was detected, with 112 of these having reduced expression in the resistant host breed compared to susceptible animals. These differentially expressed genes were subjected to lexical analysis (aligned into known pathways) and then, assigned Gene Ontology. Two distinct metabolic pathways were identified one of which was involved in immune response acquisition and the other in intestinal smooth muscle neogenesis.

Other studies that have investigated the host-parasite interplay include, monitoring the mosquito host of the malaria parasite, for susceptibility, using a 1200 cDNA microarray (based on Aedes aegypti ESTs), with 28 genes identified that were differentially expressed between susceptible and resistant mosquito strains (Chen et al. 2004b). In addition, the responses to Eimeria acervulina and Eimeria maxima infections were analysed in avian hosts using a 400 element cDNA microarray consisting of unique chicken genes, revealed both up and down-regulation of a number of genes (Min et al. 2003). Both studies demonstrate the utility of microarrays to address very specific and unique questions in parasitology.