Plasmodium falciparum infection remains a major health problem worldwide. Its complex life cycle has hampered standard methods for the study of pathogenesis. New approaches to elucidate parasite biology using whole genomic methods have provided insight into gene function, transcriptional regulation and stage specific biology 1234. Characterization of the in vivo biology of this pathogen, through adaptation of a whole genome approach, would provide insight into the host-parasite relationship, parasite virulence factors and inform new strategies for intervention. Genomic scale transcriptional profiling of P. falciparum during a natural infection is presented. Small amounts of parasite RNA, isolated from a few milliliters of a blood sample are found to be sufficient for whole genome transcriptional analysis. This data show that several genes are differentially expressed in vivo, indicating differences between the transcriptional program of 3D7 laboratory strain parasites growing in culture and naturally occurring infections in the human host.

Whole genome expression has been used in studies of bacterial pathogenesis to identify genes that are specifically transcribed under in vivo conditions 567. For example, genes involved in amino acid transport and metabolism are upregulated in Pasteurella multocida in vivo as compared to in vitro conditions 8. Similarly, analysis of P. falciparum gene expression patterns, particularly the subset of genes that are specifically expressed in the in vivo state may identify unique parasite biology when it resides in the host environment. Processes involving parasite metabolism, immune evasion and transmission may be altered in the highly specialized environment of the human host as compared to in vitro conditions. In addition, approximately 12% of P. falciparum's predicted genes have not been found to be expressed in any of the life cycle stages previously studied 9. Whole genomic analysis of the parasite in vivo may reveal the unique expression of such genes in vivo, providing additional targets for intervention.

To evaluate the integrity of the RNA transcripts from the in vivo isolated samples a denaturing RNA gel was carried out (Figure 1). The ribosomal bands are sharp with minimal RNA degradation. Despite buffy coat depletion there are human ribosomal bands present in addition to P. falciparum ribosomal bands. Human ribosomal bands are not seen on a denaturing RNA gel from in vitro cultivated 3D7 (data not shown). The most abundant transcript of human origin in the in vivo was haemoglobin RNA (Table 1). Human transcripts are also detected in 3D7 in vitro samples, but at a lower level of abundance.

The corresponding peripheral blood smears for the four in vivo samples contained only ring forms. Notably, only ring stages are present in the peripheral blood of P. falciparum infected patients; later stages are sequestered in the microvasculature. For this reason, the in vivo whole genome transcription data was compared to the in vitro chromosome 2 ring stage transcriptome. Three samples with parasitemias that were less than 0.3% and total volumes of up to 2.5 ml from Senegal were studied: this resulted in the detection of fewer transcripts than the sample obtained from a Nigerian patient who had parasitemia of 0.4% and underwent a larger blood draw. However, 50% of the top twenty five expressed transcripts in all four samples were shared (data not shown). Further analysis was performed on the Nigerian sample. Only one parasite line was detected in this sample through DNA genotyping of the K1, MAD20, RO33 alleles of msp1 and FC27 and IC1 alleles of msp2 using primers and methods previously reported 14. After total RNA was isolated, aliquots of 8 μg of total RNA were labeled using a modified Eberwine procedure 2. To maximize parasite transcript detection, 15 μg to 120 μg of cRNA were hybridized to the array and a quantitative expression level was calculated using the MOID algorithm for the Plasmodium genes on the array 2. Correlation coefficients comparing P. falciparum chromosome 2 expression levels utilizing 15 μg to 30 μg or 15 μg to 60 μg cRNA were 0.95 and 0.92, respectively. However, as cRNA concentration was increased to 120 μg, the background to noise ratio increased significantly, resulting in a decreased correlation coefficient (R = 0.72) (Figure 2a,2b and 2c).

The most abundant transcripts detected from the Nigerian in vivo sample are listed in Table 2. Genes in bold are uniquely expressed in the in vivo sample compared to 3D7 ring stage previously reported using the Kruskal-Wallis method 2. Notably, a number of genes encoding surface proteins such as rifins and SERA antigens appear overexpressed in vivo.

To confirm the accuracy of the results from the oligonucleotide array and to confirm the in vivo overexpression of a SERA antigen (PFB0355), the relative expression of five genes that had varying transcript abundance by array was carried out using real time PCR of cDNA generated from total RNA isolated from the Nigerian in vivo sample and a 3D7 in vitro ring stage sample. There is good correlation between the array results and those obtained by real time PCR (Figure 3). To compare abundance of PFB0355 cDNA between the in vivo and in vitro samples, the data is normalized to cDNA of PFB0120 to account for differences in starting parasite cDNA, secondary to human cDNA. The in vivo sample contained 0.15 ng cDNA of PFB0355c and 3D7 ring stage cDNA had 0.09 ng by real time PCR. When PFB0355c is normalized to PFB0120c, it was found to be ten fold overexpressed in vivo as compared to in vitro, consistent with the array results.

This data confirms that in vivo whole genomic expression can be performed despite the potential technical challenges of scarce RNA contained in a small blood volume sample and presence of abundant human RNA. Tri-Reagent BD was used to stabilize RNA and the samples were stored at -80°C before transport to the US. The denaturing gel suggests that the RNA remains intact using this reagent. Notably Kyes et al. reported that minus 80°C rather than 4°C or minus 20°C is the optimal temperature to store field sample RNA for detection of long transcripts such as var 15. Surprisingly, abundant human RNA was detected in the denaturing gel despite buffy coat depletion of the samples. In addition, the degree of hybridization to human probes on the oligonucleotide array used here was not seen in the previously studied in vitro sample 2. Haemoglobin is found to be the most abundantly expressed human transcript (Table 1) with considerably higher levels noted in the in vivo samples as compared to the 3D7 in vitro sample. Although human red cells are used for culturing in the in vitro system, reticulocytes which contribute to the haemoglobin detected are not abundant in in vitro samples. This is most likely due to the observation that reticulocyte levels display a 75% loss at 48 hours, when placed at 37°C, which is the condition of in vitro culture 16. In addition, other human RNA such as ribosomal RNA is more abundant in the in vivo samples and may be secondary to white cell contamination. Cross hybridization of human RNA to parasite probes may occur. However, Zhou et al have shown that human genomic DNA does not highly cross hybridize to the parasite probes on this custom array 17. The high specificity of this array is due to the nature of the highly AT rich parasite genome as compared to the human genome and the careful selection of parasite unique 25 mers probes 21819. Due to this high specificity, it is likely that buffy coat depletion is not necessary for analysis of in vivo parasite transcripts when using these probes.

The amount of blood volume necessary for comprehensive detection of transcription depends on level of parasitemia and method of microarray analysis. This analysis utilized the Affymetrix system which requires very little starting RNA as compared to other methods 20. Due to the presence of human RNA in the in vivo samples it was not possible to determine how much parasite RNA is required for whole genome analysis. The samples from Senegal were of low parasitemia and small volumes, whereas the 14 ml blood sample with 0.4% parasitemia was sufficient to detect a greater number of chromosome 2 transcripts. Four to five mls of packed blood in a patient with a greater than 2% parasitemia should provide sufficient material for whole genome analysis using these methods. The data demonstrates the reproducibility of the method by independent hybridizations and that maximal senstivity can be achieved with up to 60 μg of cRNA, using this array.

PFB0120w is the most abundant in vivo transcript in all samples encoded on chromosome 2. This gene is a member of a recently described gene family, etramps, expressed at early ring stage encoding a protein thought to be involved in erythrocyte remodeling; this was also the most highly expressed transcript in in vitro ring stage cultures 22122. The in vivo expressed genes from the Nigerian sample was compared to that of the in vitro 3D7 ring stage chromosome 2 transcriptome 2. As expected, there was a good correlation between the in vivo ring stage transcriptome and the 3D7 ring stage transcriptome. Most of the genes that are expressed in vivo are also expressed in vitro particularly those involved in cellular function and genes encoding hypothetical proteins (Table 2) 2. A number of differentially expressed genes involved in transmission and antigenic variation were identified. There is high transcription level of transmission blocking target antigen (PFB0405w) in the in vivo samples. Previously this gene has been demonstrated to be expressed and transcribed only in the sexual transmission stage 39. Since no transmissible forms were identified by microscopy this suggests that the in vivo samples may have more parasites that are undergoing or are committed to sexual development than is detectable by microscopy. Several genes encoding membrane proteins, including SERA and rifin genes, were also found to be differentially expressed. These genes are members of multigene family that encode surface proteins which are thought to be involved in immune evasion 2324. The observed increase in transcription for these genes in vivo could be due either to genetic or transcriptional modulation of the parasite's defense repertoire or geographic variation. Differences between the in vivo and in vitro expression of SERA antigens may be due to differences in the in vivo stage of development as this gene family is transcribed at later stages in the in vitro life cycle 2425. There was overall higher hybridization intensity of the in vitro samples due to higher parasite counts and a subset of genes were found to be overexpressed in vitro after normalization. This analysis focussed on genes overexpressed in vivo as these results would not be influenced by normalization algorithms. Overall, this data demonstrates that P. falciparum transcripts can be detected from in vivo samples, and that there are potentially important differences between trancription of in vivo samples and that of the 3D7 in vitro trancription profile.

In summary, this study provides evidence that whole genome gene expression in P. falciparum can be studied in vivo from a small blood sample of an infected patient. The in vivo sample however contains human RNA, whose quantity may vary from sample to sample and therefore differenes in parasite transcript level between samples must be reported relative to a reference transcript. Despite the abundance of human RNA the genomes are sufficiently different with resultant probes specificity. Predictably, there was a high correlation of in vivo expression with the in vitro ring stage 3D7 transcriptome 2. Importantly these data also suggest differences between in vivo and in vitro expression levels in genes typically found in transmissable forms and encoding variant surface proteins. Evaluation of trancription of genes specific for gametocyte development in specific patient populations may uncover the in vivo conditions that favor development of transmissable forms. Similarily, a whole genome analysis can comprehensively characterize expression of multigene families that encode variant surface proteins under in vivo conditions. Further exploration of the in vivo biology of P. falciparum using specific probes to all annotated genes will be undertaken to confirm and explore other important biological differences. This new approach will further the understanding of the host-pathogen interaction and may result in the development of new strategies to combat this disease.