We here report expression patterns of transcripts containing HERV-W gag
sequences in human tissues. Transcripts containing HERV sequences in general, along with other repetitive regions of the genome are generally not prioritized in human genomics studies. Due to the difficulty in finding specific primers and probe sequences, cross-reactions between different elements would render regular PCR or array based techniques ineffective for distinguishing individual members within a HERV-family. Tm-analysis has the advantage of detecting variations in the sequence of an amplified region, predicting the least number of different sequences required to account for a set of Tm data. It can, however, never predict more than the minimum
number of different sequences as different sequences can have indistinguishable Tms [29
]. Expression pattern variations between such sequences can therefore remain undetected. However, with Tm-analysis it is possible to screen samples for differences in expression patterns requiring sequencing of only a limited number of products within a Tm category of interest. These can subsequently be differentially analyzed by more specific assays, as exemplified in [28
] or can be sequenced selectively based on Tm. The approach used allows a higher resolution of repetitive sequence analysis without the expense of large scale sequencing. The evaluation of cost versus gain in information of the method used here as compared to the high throughput sequencing data produced in next generation sequencing technologies is a subject for future studies.
The current findings illustrate part of the hidden complexity of the human transcriptome. We can conclude that expression of HERV-W gag sequences varies between human tissues systematically and consequently in a non-random fashion. Tissue type and cell composition appear to be a larger determinant of the expression profile of such transcripts than the individual from which the tissues were obtained. For instance the expression profile of HERV-W gag containing transcripts in testis, a pooled sample of, according to the manufacturer, 53 individuals contained only 3 categories of 13 possible represented (an unlikely finding if expression was random).
Neighbor-joining of the Pearson correlation coefficients resulted in a dendrogram which groups brain regions and gonads together and tissues rich in cells of the immune system together. This dendrogram resembles similar constructs built from coding transcript expression data [31
], indicating that transcripts containing HERV-W gag
elements vary between tissues similarly to such transcripts. Due to the limited number of data points the tree structure is not stable and is hence only an indication of similarities. Previous studies on transcription in different tissues, including different brain regions, suggest that functional specialization is reflected at the level of transcription [32
]. Based on the clustering of tissues in Figure , it appears as if this extends also to the transcription of HERV-W elements.
We observed that the degree of differences in expression patterns of HERV-W gag
elements between individuals was not constant across tissues and cells. Expression patterns of coding transcripts are known to vary in whole blood samples depending on age, gender, time of day and health status [35
], despite this whole blood exhibited the most homogenous expression pattern of the tissues investigated here. Interestingly, expression patterns in brain tissue exhibited the largest variation across individuals of all tissues examined. Since spleen samples obtained from the same individuals, displayed far less variation, these findings cannot be attributed to post-mortem
or nucleic acid purification artifacts. Indeed, Franz and coworkers [36
] reported that death caused exaggerated homogeneity in expression profiles of coding transcripts in human brain. Human fibroblasts lines from different individuals were maintained for 3-5 passages under identical conditions, yet differences between individuals remained, regardless of treatment. Thus, in addition to environmental cues, genetic or epigenetic components [37
] appear to contribute to the HERV-W gag
Whether this tissue specific transcription is a consequence of specific mechanisms or reflects transcriptional leakage from transcription of coding regions remains to be established. Indeed, putative promoter structures (i.e. long terminal repeat regions) in the HERV-W and other families are active in human cells [20
]. Recent studies indicate that tens of thousands of human transcripts are initiated at retroviral promoters [38
]. Furthermore, Gogvadze and coworkers [39
] recently reported functions of HERV-K transcripts in regulation of gene expression.