Genome-wide expression profiling offers new perspectives for studying host-pathogen interactions to decipher, at the transcriptional level, how host cells react to infection and how pathogens adapt to their host's microenvironment. In the present study, we took advantage of SAGE to analyze the transcriptomes of both the infected MΦ and the intracellular parasite Leishmania using a one-step approach. Our working hypothesis was that, having extracted the bulk of mRNA molecules from a co-culture of parasites and infected MΦs, it would be possible to separate, in the resulting SAGE library, the respective contributions of each organism to the mixed collection of tags. The proportion of ambiguous gene signatures was found to be lower than 1.5%, confirming the validity of this approach. Such unambiguous tag species identification would be more difficult to reach using alternative high-throughput transcriptomic methods, such as microarrays, due to the difficulties in assessing the extent of cross-hybridization between the human and the parasite transcripts.
Separating the contribution of both organisms in an infected MΦsAGE library raised no technical problems and could be performed on a desktop computer using the functions of a commercial database management system (MS-Access). To distinguish tags according to their origin, we considered that merging all publicly available leukocyte libraries would generate a set of tags that are representative of human transcripts. Despite this extended coverage, it is clear that the deconvolution of both transcriptomes could not be complete, since unmatched tags that could not be ascribed to either of the two species (H. sapiens or L. major) may either correspond to very specific human transcripts expressed only in Leishmania-infected MΦs and never generated elsewhere or may reveal stage-specific parasite transcripts strictly specific of the intracellular stage. This problem was pointed out in a recent study , suggesting that the human genome might actually contain twice as many transcribed regions as currently annotated. Moreover, the ENCODE project consortium highlighted the number and complexity of the RNA transcripts generated comparatively to the small number of protein-coding genes (≈ 21,000) currently annotated on the human genome [33, 34].
SAGE was used as a quantitative approach, to evaluate the expression levels of mRNAs and to calculate the respective amount of material from human or parasite origin. With the reasonable assumption that mRNAs originated only from living cells, our data demonstrated the importance (49%) of the parasitic load in infected cells.
In spite of this heavy parasitic burden, a salient feature emerged from multivariate statistics: that parasite infection has, at the global level, an apparent marginal impact (only 2.4% of the transcripts were found modulated) on the expression profile of infected MΦs. Thus, the mRNA profile of infected MΦs contrasted with that of monocytes exposed to LPS because it revealed many fewer alterations in gene expression.
However, although Leishmania parasites do not seem to induce dramatic changes in the transcriptional remodeling program of MΦs, a closer analysis detected physiologically significant alterations in gene transcription. Despite their discreetness, these alterations could harmfully weaken macrophages' microbicidal defense task and homing properties. Indeed, our analysis showed that several MΦ antiparasitic pathways were altered at the level of mRNA expression upon infection by L. major parasites. In particular, we were able to show that several members of the S100A family, among others, are up- or down-regulated by infection. This is in contrast to a previous study using microarray technology that reported almost stable signals between non-infected MΦs as compared to L. major-infected MΦs for this gene family . Other differences in the expression levels of several chemokine family members were observed between the two studies, except for CXCL3 and IL8 transcripts, which were strongly up-regulated.
Whether the discrepancies between the two approaches reflect differences in the experimental protocols used by the two studies (e.g., cell-parasite incubation time, parasite strains or human genetic variability) or are attributable to differences in the sensitivity of the two techniques to accurately quantitate the mRNA of expressed genes is unclear. It is noteworthy that our results concerning the IFNγ pathway, are in agreement with those obtained recently by Dogra et al. in THP1-infected cells . Indeed, transcripts of STAT-1, a key actor of this pathway, were drastically down-regulated at 24 h after infection, though there is no external activation by IFNγ. In addition, we found that several IFNγ-inducible chemokines (CXCL9 and CXCL10) were down-modulated. Since key proteins belonging to this pathway are also inhibited upon L. major infection (K Ben-Aissa, Personal communication), such effects render the MΦ refractory to any potential activation by IFNγ and obviously favor parasite survival. Other genes, among those involved in antigen presentation and implicated in the stabilization and the recycling of classical MHC class II and in the binding and the capture of antigens were also down-modulated by L. major, as reported by Chaussabel et al. and Dogra et al. [10, 11].
Our results also show that several genes encoding pro-inflammatory mediators were up-regulated, while other family members were down-modulated. This indicates that Leishmania have a remarkable capaCity to specifically inhibits the transcription of several molecules associated with pro-inflammatory responses. It is notable that this peculiarity of L. major infection does not completely fit - in contrast to other pathogens (i.e., Mycobacterium tuberculosis, Listeria monocytogenes, Escherichia coli, Bordetella pertussis, Candida albicans, etc.) - with the so called "common host-transcriptional response" , stressing the particularity of this parasite. This is probably a survival mechanism whereby the parasites can inhibit a harmful inflammatory reaction in order to slip silently into the MΦ and successfully establish inside the host.
In addition to the analysis of the MΦ transcriptome, in the last 5 years, several studies have focused on the parasitic transcriptome taking advantage of the availability of L. major genome sequence . Although this genome (35 Mb distributed across 36 chromosomes of varying lengths i.e., 0,3 to 2,8 Mb, and coding for roughly 8,370 manually annotated protein-coding genes), was declared to be finished in 2005 , only 2,191 L. major ESTs originating from cDNA libraries of various sources, such as promastigote or amastigote full length cDNA libraries, are reported on NCBI.
Hence, our tag-to-gene mapping for parasite transcripts was rather encouraging, compared to the number of sequenced tags. Indeed, we were able to list up to 900 tags expressed in at least two copies in the metacyclic promastigote stage but totally absent from the intracellular amastigote stage, generating useful data for better data mining. In addition, among the tags common to L. major promastigote and MΦ-infected libraries, 19% (697/3,666) were differentially expressed. This led us to estimate (without taking into account the tags that were specifically intracellular and present only in the infected MΦ library) the transcripts differentially expressed, between the two parasitic stages, to roughly 1,600 tags, representing approximately 20% of transcripts if one considers the 8,370 annotated Leishmania genes registered in the databases.
This figure is several-fold higher than those reported from a variety of Leishmania species (i.e., L. major [13–15, 18], L. donovani , L. infantum  and L. mexicana ), which clearly show limited differences using microarrays (ranging from 0.2 to 5% of total genes) in stage-specific gene expression between the promastigote and amastigote life stages. These studies also show that the vast majority of genes are constitutively expressed [18, 20, 37]. One should note that these studies analyzed the amastigote transcripts, either using amastigote parasites derived from BALB/c lesions or axenic amastigotes obtained in vitro, whereas our study used the amastigotes derived from human MΦs.
However, while analyzing the functional significance of gene expression in Leishmania, we should consider that it is mainly regulated at the post-transcriptional level. As highlighted by Cohen-Freue et al. , the alteration in mRNA levels of regulated genes in Leishmania does not necessarily correlate with subsequent protein abundance. The functional significance is better manifested at the protein level, which is regulated by mechanisms such as stage-specific translational control, RNA stability, processing events and post-translational modifications. Nonetheless and despite these limitations, transcriptomic approaches for Leishmania could mainly help to better annotate its genome and to study the stability and translational regulation of its transcripts.