Background

The development of a whole organism body plan is a challenging and intricate process that requires a perfect coordination of instructive signals modulating the expression of developmental genes at specific space and time. These instructive signals are genetically encoded by sets of regulatory elements (RE) with stage-specific functions (1,2).  During developmental transitions, these RE repertoires modulate the dynamics of transcription (transcription initiation, maintenance over time and decommissioning) by becoming active in specific developing structures during delimited time windows. During these complex and continuous processes, gene expression robustness often relies on regulatory landscapes that combine sets of enhancers with very sharp and restricted temporal and spatial expression patterns together with enhancer sets presenting broader and overlapping expression domains. Unraveling the mechanisms behind this dynamism in enhancer deployment has been usually approached focusing on the spatial dimension. However, although inherent to it, the definition of temporally distinct enhancer repertoires remains less clear.

In this work developed by Rouco et al. the team focuses on characterizing how different enhancer repertoires switch on, maintain, and switch off gene transcription in vivo along a continuous developmental period: the development of the limb bud.

Summary

Shox2^trac mouse: an in vivo reporter system for the temporal dissection of the regulatory program controlling Shox2 expression

Shox2 gene is stably expressed from the early limb bud stage up to the formation of proximal limb derived structures, playing an important role in the formation of humerus and femur, among others. After examination of H3K27ac and promoter Capture-C profiles of the Shox2 TAD (a 1.1Mb TAD comprising 13 early, 3 common and 4 late putative enhancers active in E10.5 and/or E13.5 whole limbs) the authors hypothesize that its stable expression might be sustained by a combinatorial usage of stage specific enhancers.

To test this hypothesis, the authors develop a strategy to follow the changes in Shox2 gene expression during limb development by generating transgenic mice with a specific reporter construct.

How does it work?

Shox2^trac mouse harbors two different genetically engineered loci:

• Shox2 locus

This locus contains an insertion of a regulatory sensor cassette located 1kb upstream of the Shox2 transcriptional start site. This cassette is constituted by a minimal β-globin promoter, a mCherry reporter open reading frame (ORF) followed by a destabilized PEST sequence, a P2A self-cleavage sequence (allows the production of mCherry and CRE as two different proteins) and the CRE recombinase ORF. Here, mCherry works as the most immediate reporter of Shox2 expression. Since it has a short lifespan (thanks to the PEST sequence added downstream), the red signal will only be present when the reporter is actively turned on.

• Rosa26 locus

The Rosa26 locus contains a cassette with a splice acceptor followed by a floxed STOP signal and the EYFP ORF. Therefore, anytime the CRE recombinase is produced at the Shox2 locus cassette, it will cut permanently right on floxed sites, removing the STOP and allowing EYFP to be expressed continuously. Here, EYFP acts as a “footprint”, indicating that Shox2 either was or it is being expressed at that moment.

This elegant system allows to clearly identify cells at different stages of transcription*:

• Transcription maintenance: cells undergoing transcription maintenance will present both dmCherry and EYFP signal.
• Transcription decommissioning: cells having undergone transcription decommissioning present only EYFP signal, indicating that Shox2 was expressed, and the CRE that allows EYFP signal was produced, but Shox2 expression (and subsequently, dmCherry) cannot be detected anymore.
• Inactive cells: characterized by the absence of any reporter signal.

The fact that each of the two different transcriptional status comes with a specific signal makes of this regulatory sensor cassette a perfect tool for identifying the cells´ transcriptional status (with respect to Shox2 expression) in further single cell analyses. In addition, it allows to FACS-sort the cells based on their dmCherry and EYFP signals.

Single-cell insights into Shox2 transcriptional dynamics trace early decommissioned cells in the distal limb

Microscopic examination of Shox2 embryos confirmed that the system was reliable, as the reporter signals recapitulated the expected expression dynamics of Shox2. Next step was to translate the red (dmCherry) and yellow (EYFP) signals observed under the microscope to single-cell data. To do so, the team investigated Shox2 transcriptional dynamics in mesenchymal hindlimb cells during continuous developmental stages by performing scRNA-seq of Shox2^trac embryos at E10.5, E11.5, E12.5, and E13.5.

After merging the expression data of the four timepoints under study, manual clustering annotation was performed based on the expression of specific cell identity markers. 15 total mesenchymal clusters classified cells in a range of progressing identities, ranging from progenitors to more differentiated identities. RNA velocity analysis, which allows to define directions of cell identity transitions based on a ratio between unspliced and spliced mRNA, was performed to track the cell identities that the progenitors were giving rise to.

Following this, the authors projected the signals coming from the different reporters of the sensor cassette over the identity clusters, finding a concordant distribution of the different transcriptional stages being present at more or less differentiated identities. For instance, Shox2 initiating cells (defined as Shox2+, in the absence of EYFP signal*) were primarily found in progenitor clusters. This approach brings the main finding of this section: proximal limb progenitor clusters retain Shox2 expression, whereas more distal clusters progressively silence the locus.

Maintaining Shox2 transcription relies on the temporal deployment of distinct enhancer repertoires

To examine whether Shox2 maintained expression involved the use of different sets of stage-specific enhancers. Maintaining cells (Shox2+, dmCherry-P2A-CRE +/-, EYFP+) where FACS-sorted at every tested time point in forelimbs and hindlimbs. Then, H3K27ac ChIP-Seq was performed. Following analyses identified a total of 34 H3K27ac-marked enhancers within the Shox2 TAD. The majority exhibited stage-specific activity: 26 enhancers functioned in early limb development (“early enhancers”), while 4 displayed consistent activity throughout all the period (“common enhancers”), and another 4 were restricted to later stages (“late enhancers”).

Changes in the 3D organization of the Shox2 locus coincide with temporal variations in the sets of active enhancers

To evaluate whether the dynamic changes in enhancer usage were associated to changes in the chromatin landscape, capture Hi-C maps were generated in mESC (representing the very initial point of the developmental trajectory) together with sorted cells (inactive and maintaining) at specific stages. A few examples proved that, early on in development (E11.5), the Shox2 locus interacts mostly with common and early enhancers, but at more advanced stages (E13.5), it interacts preferentially with late enhancers. This suggested that the 3D organization of the Shox2 locus contributes on the control of which enhancers are active at different times.

Functional validation of Shox2 early and late enhancers proves a degree of independency between early and late enhancers

To test the level of independency between the early and late enhancers, the authors proceeded to functionally validate subsets of each type of enhancers. First, they generated the Shox2^Δearly mouse, comprising a homozygous deletion of 8 out of 26 early enhancers in the Shox2^trac   background. Second, they generated the Shox2^Δlate mouse, harboring a homozygous deletion of 2 out of 4 late enhancers, 2 out of 4 common enhancers and 1 out of 26 early enhancers. For both models, fluorescent imaging, quantification of the proportion of inactive, maintaining and decommissioned cells as well as bulk RNA-seq were performed. Deleting early enhancers causes a sharp but temporary drop in Shox2 gene activity in early forelimb and hindlimb cells, coupled to a reduction in the proportion of maintaining cells at the earlier stages tested. Interestingly, the proportion of maintaining cells gets to catch up and reach control levels. Oppositely, deletion of late enhancers leads to a reduction in Shox2 gene activity only at later stages and, consequently, the number of decommissioned cells increases with respect to the Shox2 controls.

Shox2 decommissioning relies on enhancer-promoter disconnection

Same way as the enhancer activity and chromatin landscape dynamics of Shox2 maintaining cells was interrogated, this section focuses on the decommissioning process.  To explore this stage, C-HiC maps of FACS-sorted decommissioned cells from Shox2^trac E12.5 and E13.5 fore- and hindlimbs were generated, observing a reduction in enhancer-promoter interactions. Therefore, switching off the Shox2 gene involves a rapid breakdown in communication between the gene and its enhancers, followed by a decrease in the activity of those enhancers. Interestingly, some enhancers seem to remain active even after Shox2 is off.

Hoxd13 can induce Shox2 locus decommissioning

The rapid loss of connections between enhancers and the Shox2 gene, along with the quick decline in enhancer activity, suggests a targeted repression happening at the enhancer regions. This repression might involve specific cell-identity transcription factors binding to the DNA.

A major part of decommissioned cells is found in the distal part of the limb. This suggests that proteins controlling that region (distal limb TFs) are involved in turning off Shox2. In particular, HOXA/D13 proteins are known to repress Shox2 in the distal limb, as they bind to many of the Shox2 enhancers (3, 4).

To test this, the researchers induced the ectopic expression of HOXD13 in the Shox2 active proximal part of the limb by using the Ulnaless construction on Shox2^trac background. As expected, this led to more decommissioned cells and fewer maintaining cells in the proximal region, without affecting inactive cells. This suggests that HOXD13 can indeed repress Shox2 by binding and inactivating its early-acting enhancers.

Key findings

• Shox2 gene expression relies on different sets of enhancers at different stages of limb development (early vs late).
• Removing either early or late-acting enhancers only has a temporary effect on Shox2 expression, suggesting these sets work independently.
• The 3D structure of the Shox2 locus changes along with enhancer activity, and these connections weaken rapidly when the gene is turned off.
• The gene Hoxd13 can actively turn off Shox2 expression, possibly by repressing some of its enhancers.

Why I chose this preprint?

This work by Guillaume Andrey lab focuses on exploring the complexity of enhancer deployment processes from a temporal perspective. The selected approach is a great example of unravelling a complex biological question by leveraging and combining the strengths of different techniques (in vivo reporter assays + bulk and single cell RNA-Seq) for orthogonal validation of the tested hypotheses.

Experimentally, I particularly enjoyed the part of the approach that allowed to track decommissioned cells, a strategy that can be further exploited for digging into the mechanisms of repression initiation and maintenance during extended time periods.

Another great motivation to write this Prelight comes from the fact that the Junior Researchers Group at IBTTEC elected Dra. Raquel Rouco (first author of this work) as an invited speaker to our External Seminar Series 2024. Next week, she will share this elegant story with IBBTEC members from different research backgrounds.  This Prelight aims to be a first glimpse into this exciting approach with great potential to be extrapolated to different research fields.

 

(*)Transcription initiation could not be captured as mCherry+/EYFP- due to technical reasons.

 

Questions to the authors

1. Regarding the 3D contact analysis in mESC vs Shox2 inactive cells, you report contacts between Shox2 and three of its early enhancers in Shox2 inactive cells. What do you think these contacts reflect? Did you check the H3K27ac profile at these enhancers in inactive cells? Could they reflect sites of active repression of Shox2, favouring the specification into different cell identities?

 

2. In the Shox2^Δearly forelimb, you detect a decrease in the proportion of maintaining cells at E10.5 that gradually returns to control levels at later stages. An interpretation of this result could be that, Shox2^Δearly  progenitors are compensating  the partial loss of early enhancers, ultimately resulting in similar numbers of maintaining cells, with less early enhancers acting. Could you elaborate on the possible mechanism/s behind this “catching up”?

 

3. Even though Shox2 decomissioned cells contribute to the formation of posterior digits, after removal of early enhancers and loss of decommissioned cells at the distal limb, digits 4 and 5 are still formed and apparently normal. Based on this, how do you explain the role/contribution of early Shox2 expression in future posterior digit formation?

 

4. As a more general aspect, this regulatory tracking strategy can be useful in other research fields outside of developmental biology. Could you provide some examples of its applications?

 

 

References

(1) Robson, M.I., Ringel, A.R., and Mundlos, S. (2019). Regulatory Landscaping: How Enhancer-Promoter Communication Is Sculpted in 3D. Mol Cell 74, 1110-1122.

(2) O. Symmons, V. V. Uslu, T. Tsujimura, S. Ruf, S. Nassari, W. Schwarzer, L. Ettwiller, F. Spitz, Functional and topological characteristics of mammalian regulatory domains. Genome Res. 24, 390–400 (2014). 10.1101/gr.163519.113

(3) Beccari, L., Yakushiji-Kaminatsui, N., Woltering, J.M., Necsulea, A., Lonfat, N., Rodriguez-Carballo, E., Mascrez, B., Yamamoto, S., Kuroiwa, A., and Duboule, D. (2016). A role for HOX13 proteins in the regulatory switch between TADs at the HoxD locus. Genes Dev 30, 1172-1186.

(4) Sheth, R., Barozzi, I., Langlais, D., Osterwalder, M., Nemec, S., Carlson, H.L., Stadler, H.S., Visel, A.,Drouin, J., and Kmita, M. (2016). Distal Limb Patterning Requires Modulation of cis-Regulatory Activities by HOX13. Cell Rep 17, 2913-2926.

Tags: crispr-cas9, enhancer, forelimb, hindlimb, limb development, mouse, mus musculus, reporter assay, scrna-seq, transcription dynamics

doi: https://doi.org/10.1242/prelights.37111

Read preprint (No Ratings Yet)