Bursts in Interferon Transcription and Enhanceosome Formation

A model of stochastic enhanceosome and preinitiation complex formation that incorporates transcriptional pulsing. Analytical and numerical results show good agreement with direct measurement of the expression of interferon beta gene in its natural chromatin environment in primary human cells. The results show that the distribution of mRNA across cells follows a power law with an exponent close to −1, and thus encompasses a range of variation much more extensive than a Gaussian. The model thus support the existence of transcriptional pulsing of an unmodified, intact gene regulated by a natural stimulus. (Power-Laws in Interferon-B mRNA Distribution in Virus-Infected Dendritic Cells, Hu, Jianzhong, et al., Biophys. J. (2009) 97:7,pp.1984)

The model without transcriptional pulsing is described in detail in Hu et al. It was based on the key experimental result that the level of IFNB1 induction in different cells in response to NDV infection was broad and dominated by intrinsic noise. Our modeling is focused on the power-law behavior due to infection from NDV and the mutant flu viruses. Since there are many potentially different sources of extrinsic noise, the robustness of the experimental results (with different individuals giving rise to similar power laws) provides the motivation for modeling only the intrinsic noise that arises from the stochastic fluctuations in the assembly of the enhanceosome.

Enhanceosome formation from the activated components was described as cooperative binding of components P1, P2, P3, and P4 to the IFNB1 promoter region. It is believed that the architectural protein HMGA1a binds to the promoter region facilitating the recruitment of the other components. Munshi et al. suggest that NFκB (p65) is detected initially at the promoter with IRF-1, ATF-2 is recruited later followed by the arrival of IRF-3, and finally IRF-7 that is synthesized in response to virus infection via the IFN autocrine loop. Although there is some evidence for two-phase kinetics with feedback, we focus on modeling the substantial induction of IFNB1 that is measured in the 9–12 h range after IRF-7 synthesis. We model the enhanceosome with four proteins, P1–P4, that may be taken to represent the architectural protein and the three transcription factors NFκB, IRF, and ATF-2.

For each gene, the reactions of the model for sequential cooperative binding of P1–P4 are given by


where for simplicity the rate constants are chosen equal in the first three reactions. Ds4 denotes the completed enhanceosome. We have allowed for a small rate for the last transcription factor allowing the enhanceosome to fall apart in contrast to the earlier model from Hu et al. This makes the entire model, with reactions from Eqs. (5) and (6), equivalent to a pulsing problem, which leads to power-law distributions with reasonable agreement for the timescales on which mRNA induction occurs.

Once the enhanceosome is completed on either of two chromosomes, there is a cascade of steps to assemble the preinitiation complex. The steps include histone acetylation, recruitment of the CBP-Pol II holoenzyme complex, SwI-SNF, and TATA-binding proteins. After this, the enhanceosome is in a transcribing state, where IFNB1 transcription takes place for some random time, before switching back to the nontranscribing state. This latter switching back-and-forth corresponds to transcriptional bursting and distinguishes the present extended model from the original one. We model the entire assembly of the preinitiation complex by a single step. The transcribing state is Ds4*, from which mRNA m is produced.

For the rate constants, we chose k1 = 1.132 × 10−7 s−1 in the Gillespie simulations with the actual rate obtained by multiplying the copy number of the transcription factors. The others are given by k2 = 0.002425 s−1, f = 1.5 × 10−4 s−1, and b = 3 × 10−4 s−1. We used a copy number of 12,000 for all the transcription factors. A transcription rate of 20 per min was used. The numbers for the assembly of the enhanceosome are approximately the same as in the earlier article (4), and were chosen so that there is rough agreement with experimental results on the time at which transcripts are first measured. We have included a small rate (smaller by a factor of 4) for the enhanceosome itself to break apart. We incorporated the fact that IFNB1 mRNA is stable for 2 h and degrades abruptly thereafter. For simplicity, we have used exponential degradation in the figures shown; we have verified that the nature of the degradation does not affect our results. The transcription and degradation rates determine the maximum number of mRNA produced in any cell. The rate of transcription given the decay rate that was measured experimentally was chosen to get agreement with the maximum number of transcripts observed. The transcription factors are assumed to be activated after 2 h. There is preliminary experimental evidence for this timescale. We have verified that including a small basal transcription rate from Ds (5% of the maximum rate) does not alter our conclusions.

Download the Fortran source code of the program.