Transcriptional promiscuity in testes

Numerous genes are expressed preferentially in spermatids, the haploid germ cells of the testis. Although some have defined, spermatid-specific roles, the expression and function of the products of most of these genes are not restricted to spermatids [1]. Examples include transcription factors that have defined functions elsewhere, kinases, metabolic enzymes, and so on. Recently, another group of proteins has been added to this list — the components of the basal RNA polymerase II (pol II) transcription machinery [2]. Recent reports suggest that the pol II machinery pre-assembles into a holoenzyme complex [3,4]. The levels of the components of this complex coordinately increase 30to 100-fold during the early haploid stages of spermatogenesis ([2] and unpublished observations). Thus, early spermatid nuclei may have much higher concentrations of holoenzyme than do somatic cells. In this light, it is interesting to consider what would be the consequences of increasing holoenzyme levels by two orders of magnitude. All models of transcriptional regulation to date include the assumption that the basal transcription machinery is constant and limiting; differences in the activity of a promoter between cell types have been presumed to result solely from differences in the ‘attractiveness’ of that promoter for a fixed concentration of pol II machinery. Promoter attractiveness can be regulated by altering DNA methylation, chromatin structure or transcription factor assemblages [5–7]. But changes in the concentration of holoenzyme might also alter the rate of transcription initiation at a promoter. If the relative activities of the three promoters depicted in Figure 1a are plotted as a function of holoenzyme concentrations, each promoter gives a different activity curve (Fig. 1c). Thus, changes in holoenzyme concentration can affect the activities of various promoters differentially. For example, raising the holoenzyme concentration from 1.0 to 100 arbitrary units activates promoter C negligibly, whereas promoter B, which is a weak promoter at a holoenzyme concentration of 1.0, is now activated to the point of being nearly as strong as promoter C. Interestingly, at elevated holoenzyme concentrations, some DNA sequences that do not normally act as promoters — for example, promoter A in the figure — may now promote very strong transcription. Is there any experimental evidence to support such a model? A recent paper [4] on mammalian RNA polymerase II holoenzyme showed that promoters which require transcription factors for maximal activity in vitro using crude nuclear extracts (which have relatively low holoenzyme concentrations) were highly active in the absence of such factors when supplied with a high concentration of purified holoenzyme. The addition of transcription factors did not further activate transcription, suggesting that at high holoenzyme concentrations, weak and strong promoters are similarly active. In other studies [3], purified yeast holoenzyme was shown to be responsive to activators. Figure 1c shows how, depending on holoenzyme concentrations, promoters might exhibit great differences in their response to activators. What are the implications of this model for spermatid-specific gene expression? The model suggests that many genes which are expressed in spermatids might not require spermatid-specific transcription factors for their activation. Rather, early spermatids, by having elevated holoenzyme concentrations, may provide a permissive environment for transcription initiation. Thus, poor promoters in particular should have elevated activity in early spermatids (Fig. 1c). The model is consistent with the large number of genes that are expressed in the testis, and with the observation that testis-specific expression frequently involves both up-regulation of existing promoters and recruitment of additional promoters [1]. In contrast, genes for abundant spermatid-specific proteins have additional regulatory mechanisms to ensure that the difference in expression between spermatids and other cell types is orders of magnitude greater still [1,8]. The model proposed here does not suggest that genes which are expressed in spermatids, including those with known functions elsewhere, are expressed simply ‘by 768 Current Biology 1996, Vol 6 No 7