The E3 ubiquitin ligase RNF220 maintains hindbrain Hox expression patterns through regulation of WDR5 stability

Reviewer #1 (Public Review):

The manuscript by Wang et al. investigates the role of Rnf220 in hindbrain development and Hox expression. The authors suggest that Rnf220 controls Hox expression in the hindbrain by regulating WDR5 levels. The authors combine in vivo experiments with experiments in P19 cells to demonstrate this mechanism. However, the in vivo data does not provide strong support for the claims the authors make and the role of Rnf in Hox maintenance and pons development is unclear.

Specific concerns with in vivo data:

A major issue throughout the paper is that Hox expression analysis is done exclusively through quantitative PCR, with values ranging from 2-fold to several thousand-fold upregulation, with no antibody validation for any Hox protein (presumably they are all upregulated).

In Figure 1, massive upregulation of most Hox genes in the brainstem is shown after e16.5 but the paper quickly focuses on analysis of PN nuclei. What are the other consequences of this broad upregulation of Hox genes in the brainstem? There is no discussion of the overall phenotype of the mice, the structure of the brainstem, the migration of neurons, etc. The very narrow focus on motor cortex projections to PN nuclei seems bizarre without broad characterization of the mice, and the brainstem in particular. There is only a mention of "severe motor deficits" from previous studies, but given the broad expression of Rnf220, the fact that is a global knockout, and the effects on spinal cord populations shown previously the justification for focusing on PN nuclei does not seem strong.

It is stated that cluster 7 in scRNA-seq corresponds to the PN nuclei. The modest effect shown on Hox3-5 expression in that data in Figure 1 is inconsistent with the larger effect shown in Figure 2.

Presumably, Hox genes are not the only targets of Rnf220 as shown in the microarray/RNA-sequencing data. There is no definitive evidence that any phenotypes observed (which are also not clear) are specifically due to Hox upregulation. The only assay the authors use to look at a Hox-dependent phenotype in the brainstem is the targeting of PN nuclei by motor cortex axons. This is only done in 2 animals and there are no details as to how the data was analyzed and quantified. The only 2 images shown are not convincing of a strong phenotype, they could be taken at slightly different levels or angles. At the very least, serial sections should be shown and the experiment repeated in more animals. There is also no discussion of how these phenotypes, if real, would relate to previous work by the Rijli group which showed very precise mechanisms of synaptic specificity in this system.

The temporal aspect of this regulation in vivo is not clear. The authors show some expression changes begin at e16.5 but are also present at 2 months. Is the presumed effect on neural circuits a result of developmental upregulation at late embryonic stages or does the continuous overexpression in adult mice have additional influence? Are any of the Hox genes upregulated normally expressed in the brainstem, or PN specifically, at 2 months? Why perform single-cell sequencing experiments at 2 months if this is thought to be mostly a developmental effect? Similarly, the significance of the upregulated WRD5 in the pons and pontine nuclei at 2 months in Figure 3 is not clear.

In Figure 3C the levels of RNF220 in wt and het don't seem to be that different.

Based on the single-cell experiments, and the PN nuclei focus, the rescue experiments are confusing. If the Rnf220 deletion has a sustained effect for up to 2 months, why do the injections in utero? If the focus is the PN nuclei why look at Hox9 expression and not Hox3-5 which are the only Hox genes upregulated in PN based on sc-sequencing? No rescue of behavior or any phenotype other than Hox expression by qPCR is shown and it is unclear whether upregulation of Hox9 paralogs leads to any defects in the first place. The switch to the Nes-cre driver is not explained. Also, it seems that wdr5 mRNA levels are not so relevant and protein levels should be shown instead (same for rescue experiments in P19 cells).

Other:
What is the relationship between Retinoic acid and WRD5? In Figure 3E there is no change in WRD5 levels without RA treatment in Rnf KO but an increase in expression with RA treatment and Rnf KO. However, the levels of WRD5 do not seem to change with RA treatment alone. Does Rnf220 only mediate WDR5 degradation in the presence of RA? This does not seem to be the case in experiments in 293 cells in Figure 4.

Why are the levels of Hox upregulation after RA treatment so different in Figure 5 and Figure Supplement 5?

In Figures 4B+C which lanes are input and which are IP? There is no quantitation of Figure 4D, from the blot it does look that there is a reduction in the last 2 columns as well. The band in the WT flag lane seems to have a bubble. Need to quantitate band intensities. Same for E, the effect does not seem to be completely reversed with MG132.

Reviewer #2 (Public Review):

Wang, Liu, et al. identified Rnf220 and Wdr5 as novel regulators of Hox gene expression during pons development. Phenotypic characterization of Rnf220 deficient mice with single-cell transcriptomics, qRT-PCR, and axonal tracing methods show that Rnf220 knockdown causes de-repression of Hox gene expression at multiple stages of pons development to regulate the final formation of the pontine nuclei neural circuit. Additionally, they also perform exhaustive expression analysis of multiple genes in the Hox family cluster to identify specific gene groups that are targeted by Rnf220. Furthermore, they also demonstrate that Rnf220 modulates Hox gene expression by directly binding to Wdr5, thus targeting it for ubiquitination and subsequent degradation. To elucidate the molecular mechanism of this interaction, they perform detailed immunoprecipitation assays and identify the precise Wdr5 amino acid residues that are targeted by Rnf220. Intriguingly, they show that inhibition of Wdr5 in Rnf220 deficient mice reverses the de-repression of Hox gene expression suggesting the direct involvement of Rnf220-Wdr5 interaction in modulating Hox gene expression during pons development. These data highlight the role of a new form of Hox gene regulation via the ubiquitination of epigenetic modulator Wdr5.

The conclusions of this paper are mostly supported by the data provided, but the downstream molecular and tissue-level effects of Wdr5 knockdown/inhibition need to be further characterized to establish its definitive role in pons development.

(1) Figure 1E shows that Rnf220 knockdown alone could not induce an increase in Hox expression without RA, which indicates that Rnf220 might endogenously upregulate Retinoic acid signaling. The authors should test if RA signaling is downstream of Rnf220 by looking at differences in the expression of Retinaldehyde dehydrogenase genes (as a proxy for RA synthesis) upon Rnf220 knockdown.

(2) In Figure 2C-D further explanation is required to describe what criteria were used to segment the tissue into Rostral, middle, and caudal regions. Additionally, it is unclear whether the observed change in axonal projection pattern is caused due to physical deformation and rearrangement of the entire Pons tissue or due to disruption of Hox3-5 expression levels. Labeling of the tissue with DAPI or brightfield image to show the structural differences and similarities between the brain regions of WT and Rnf220 +/- will be helpful.

(3) Line 192-195. These roles of PcG and trxG complexes are inconsistent with their initial descriptions in the text - lines 73-74.

(4) In Figure 4D, the band in the gel seems unclear and erased. Please provide a different one. These data show that neither Rnf220 nor wdr5 directly regulates Hox gene expressions. The effect of double knockdown in the presence of RA suggests that they work together to suppress Hox gene expression via a different downstream target. This point should be addressed in the text and discussion section of the paper. example for the same data which shows a full band with lower intensity.

(5) In Figure 4G the authors could provide some form of quantitation for changes in ubiquitination levels to make it easier for the reader. They should also describe the experimental procedures and conditions used for each of the pull-down and ubiquitination assays in greater detail in the methods section.

(6) Figure 5 shows that neither Rnf220 nor wdr5 directly regulate Hox gene expressions. The effect of double knockdown in the presence of RA suggests that they work together to suppress Hox gene expression via a different downstream target. This point should be addressed in the text and discussion section of the paper.

(7) In Figure 6, while the reversal of changes in Hox gene expression upon concurrent Rnf220; Wdr5 inhibition highlights the importance of Wdr5 in this regulatory process, the mechanistic role of wdr5 and its functional consequences are unclear. To answer these questions, the authors need to: (i) Assay for activated and repressive epigenetic modifications upon double knockdown of Rnf220 and Wdr5 similar to that shown in Figure 4- supplement 1. This will reveal if wdr5 functions according to its intended role as part of the TrxG complex. (ii) The authors need to assay for changes in axon projection patterns in the double knockdown condition to see if Wdr5 inhibition rescues the neural circuit defects in Rnf220 +/- mice.