Transcription factor interactions explain the context-dependent activity of CRX binding sites

Abstract

The effects of transcription factor binding sites (TFBSs) on the activity of a cis-regulatory element (CRE) depend on the local sequence context. In rod photoreceptors, binding sites for the transcription factor (TF) Cone-rod homeobox (CRX) occur in both enhancers and silencers, but the sequence context that determines whether CRX binding sites contribute to activation or repression of transcription is not understood. To investigate the context-dependent activity of CRX sites, we fit neural network-based models to the activities of synthetic CREs composed of photoreceptor TFBSs. The models revealed that CRX binding sites consistently make positive, independent contributions to CRE activity, while negative homotypic interactions between sites cause CREs composed of multiple CRX sites to function as silencers. The effects of negative homotypic interactions can be overcome by the presence of other TFBSs that either interact cooperatively with CRX sites or make independent positive contributions to activity. The context-dependent activity of CRX sites is thus determined by the balance between positive heterotypic interactions, independent contributions of TFBSs, and negative homotypic interactions. Our findings explain observed patterns of activity among genomic CRX-bound enhancers and silencers, and suggest that enhancers may require diverse TFBSs to overcome negative homotypic interactions between TFBSs.

Fig 1. Synthetic CREs sites reveal context-dependent effects of CRX and NRL sites.

(A) Design of synthetic CRE MPRA library reported in [13]. Combinations of CRX and NRL sites (up to four TFBSs) were cloned adjacent to either a Rho or a Hsp68 basal promoter. TFBSs could be in either forward or reverse orientation. (B) MPRA activity (y-axis) of CREs composed only of high affinity CRX sites (blue) is consistently lower than that of CREs with high affinity CRX sites and one NRL site (orange), relative to the Rho basal promoter. Sequential addition of high affinity CRX first activates, then represses the Rho basal promoter. Plot shows a subset of the data reported in [13].

Fig 2. A model of CRX and NRL-driven cis-regulatory activity in wild-type retina.

(A) The performance of different model architectures (measured as predictive information) fit to MPRA measurements of the CRX-NRL library in wild-type retina. Error bars indicate standard error. (B) The observed activity (y-axis) of test set sequences in wild-type retina compared to the latent phenotype (x-axis) inferred by the pairwise model. The non-linearity is the model mapping from latent phenotype to observed activity and intended to capture non-linear effects of the MPRA measurement process. (C) The observed activity (y-axis) of test set sequences in wild-type retina compared to the activity predicted by the pairwise model (x-axis). (D) Model parameters for additive and pairwise contributions of CRX and NRL sites and the Rho promoter to activity in wild-type retina, averaged across the four positions in synthetic CREs. For pairwise interactions, rows indicate the 5’ site and columns indicate the 3’ site. Forward and reverse orientation of the TFBS is indicated by (f) and (r). (E) Model parameters for additive and pairwise contributions of CRX and NRL sites to activity in Crx^-/- retina, averaged across positions and spacings.

Fig 3. A model of cis-regulatory activity driven by diverse TFBSs in wild-type mouse retina.

(A) Design of MPRA library of synthetic CREs with additional lineage-specific TFBSs. CREs contained five sites placed adjacent to the Rho basal promoter. Each CRE contained either three CRX sites and two sites for other TFs (3-1-1) or two CRX sites, two sites for another TF, and one site for a third TF (2-2-1). (B) Observed activity (y-axis) of test set sequences compared to the latent phenotype (x-axis) predicted by the pairwise model. (C) Model parameters representing additive and pairwise contributions of TFBSs averaged across positions. (D) Distribution of position-specific interactions with high-affinity CRX binding sites, broken down by partner TF. (E) MPRA activity of CREs with two (left) or three (right) CRX sites, grouped by TF identity. Each CRE contains sites for CRX and two other TFs. Activity is measured relative to the Rho basal promoter and basal activity is indicated by the dashed line.

Fig 4. Simplified balance model of context-dependent effects of binding sites for transcriptional activators.

(A) Simulated CRE activities calculated by Eq 1 for sequences with up to four TFBSs for CRX or NRL. Stepwise addition of sites for a single TF first increase then decrease activity. The first two columns show predicted expression of CREs with only CRX sites or with CRX sites plus one NRL site. Compare with the measured values in Fig 1B. (B) Model of CREs with five TFBSs shows how TF diversity reduces negative homotypic interactions and increases CRE activity. As CRX sites are replaced with sites for different TFs, TF diversity increases (x-axis) and the number of negative homotypic interactions decreases (orange crosses) and the overall CRE activity increases (blue squares). The total additive contribution of TFBSs (green circles) is equal to the total number of TFBSs and remains constant. (C) Simulated CRE activities calculated by Eq 1 (left panel) recapitulate trends in MPRA activity of genomic CREs (right panel). Increasing numbers of CRX sites first increase then decrease activity, but the presence of TFBSs for NRL, NEUROD1, RORB, or MAZ (‘Other TFBSs’) increases both predicted and measured CRE activity. Dashed lines indicate basal activity.