Permissive and instructive Hox codes govern limb positioning

Reviewer #1 (Public review):

Summary:

This study investigates the role of Hox genes in determining the position of the forelimb bud through experimental loss- and gain-of-function approaches in chicken embryos. The loss-of-function experiments involved expressing dominant-negative versions of specific Hox genes in the limb bud to assess their necessity for limb formation. Gain-of-function experiments entailed expressing full-length Hox genes anterior to the limb field in the lateral mesoderm. The results were evaluated by analyzing the expression of genes involved in limb development, such as Fgf8, Fgf10, Shh, and Tbx5, the latter specifically marking the forelimb.

The findings indicate that introducing dominant-negative forms of Hoxa4, Hoxa5, Hoxa6, and Hoxa7 into the forelimb field reduces bud size and downregulates certain limb markers. Conversely, introducing active versions of these genes rostral to the normal forelimb position shows that Hox4 and Hox5 have no effect, whereas Hox6 and Hox7 extend the forelimb anteriorly or create a small bulge rostral to the forelimb. The authors conclude that Hox4 and Hox5 provide permissive cues for forelimb formation throughout the neck region, with the final forelimb position determined by the instructive cues of Hox6/7 in the lateral plate mesoderm.

Strengths:

The authors endeavor to address the longstanding question of what determines limb position, particularly that of the forelimb, in the vertebrate embryo.

Weaknesses:

In my opinion, the study is preliminary and requires additional controls and explanations for conflicting results observed in mice:

(1) The activity of the dominant negatives lacks appropriate controls. This is crucial given that mouse mutants for PG5, PG6, PG7, and three of the four PG4 genes show no major effects on limb induction or growth. Understanding these discrepancies is essential.

(2) The authors mention redundancies in Hox activity, consistent with numerous previous reports. However, they only use single dominant-negative versions of each Hox paralog gene individually. If Hox4 and Hox5 functions are redundant, experiments should include simultaneous dominant negatives for both groups.

(3) The main conclusion that Hox4 and Hox5 provide permissive cues on which Hox6/7 induce the forelimb is not sufficiently supported by the data. An experiment expressing simultaneous dnHox4/5 and Hox6/7 is needed. If the hypothesis is correct, this should block Hox6/7's capacity to expand the limb bud or generate an extra bulge.

(4) The identity of the extra bulge or extended limb bud is unclear. The only marker supporting its identity as a forelimb is Tbx5, while other typical limb development markers are absent. Tbx5 is also expressed in other regions besides the forelimb, and its presence does not guarantee forelimb identity. For instance, snakes express Tbx5 in the lateral mesoderm along much of their body axis.

(5) It is important to analyze the skeletons of all embryos to assess the effect of reduced limb buds upon dnHox expression and determine whether extra skeletal elements develop from the extended bud or ectopic bulge.

Reviewer #2 (Public review):

Summary:

This manuscript investigates the role of Hox genes in the specification of forelimb position. The central conclusions are that Hox paralogy group (PG) 6/7 genes are both necessary and sufficient to induce forelimb buds. In addition, the authors argue that HoxPG4/5 genes are necessary, but, by contrast to Hox PG6/7 genes, Hox PG4/5 genes are not sufficient to induce forelimb budding. To test the roles of Hox4-7 genes in limb development, the authors use both gain-of-function (GOF) and loss-of-function (LOF) approaches in chick embryos.

In LOF experiments, they produced dominant negative forms of Hoxa4, Hoxa5, Hoxa6, and Hoxa7, which lack the DNA-binding domain, and they electroporated these constructs into the prospective wing field of the lateral plate mesoderm (LPM) in pre-limb bud stage (HH12) chick embryos. All 4 constructs resulted in down-regulation of Tbx5 (an early marker of forelimb development), and of its target gene, Fgf10, which is required for the initiation of limb budding, in the lateral plate mesoderm. The dominant negative experiments also caused down-regulation of Fgf8 in the overlying limb ectoderm and a marked reduction in the size of the early wing bud. Based on the LOF results, the authors conclude that each of the Hoxa4-7 genes is required for the specification of the forelimb field and for the establishment of the Fgf10-Fgf8 feedback loop in wing bud mesenchyme and overlying epithelium.

The authors then use a GOF strategy to investigate whether the same genes are sufficient to induce forelimb budding. They test this hypothesis using the neck, a region that is known to be incompetent to form limbs in response to Fgf signaling. Overexpression of full-length Hoxa6 and Hoxa7 in the neck region caused ectopic expression of Tbx5 in the neck region, which fits with "posteriorization" of cells at neck level, as Tbx5 typically marks the forelimb and flank (interlimb) region of the lateral plate mesoderm. Consistent with a posterior transformation of positional identity (neck to forelimb), overexpression of Hoxa6 or Hoxa7 leads to activation of Fgf10 expression and development of an ectopic forelimb bud from (or extension of the normal forelimb bud into) the neck region). By contrast, overexpression of either Hoxa4 or Hoxa5 in the neck region is not sufficient to induce ectopic forelimb budding. Curiously, the ectopic forelimb buds do not express Fgf8 in the overlying ectoderm or develop beyond the bud stage. The latter finding is consistent with previous work showing that neck ectoderm is not competent to support outgrowth of transplanted limb bud mesenchyme. The authors investigate the mechanistic basis of this early arrest of outgrowth by comparing the transcriptomes of ectopic limb buds, normal forelimb buds, and normal neck cells.

The RNA sequencing analysis shows that while some limb development genes (e.g., Lmx1b, Hoxa9, Hoxd9, Hoxa10, Hoxd10) are activated in the ectopic limb bud, other key components of the circuit (e.g., Shh, Fgf8, Hox12/13 paralogs) are not established, leading them to conclude that failure of neck ectoderm to form an AER underlies the arrested outgrowth of ectopic limb buds.

Strengths:

This study provides the first evidence that altering the Hox code in neck lateral plate mesoderm (LPM) is sufficient to induce ectopic development of forelimb buds at the neck level. For more than 30 years, developmental biologists have speculated and provided indirect evidence that Hox genes are involved in the specification of forelimb position, but to my knowledge, no study has shown that altering Hox gene expression alone can induce limb development outside of the normal limb field. The finding that Hox6/7 paralogs are sufficient for forelimb bud development, whereas Hox4/5 paralogs are not, suggests that specification of forelimb identity requires instructive signaling that is a specific property of Hox6/7 paralogs. The GOF experiments significantly extend the knowledge of limb specification beyond that which has come from Hox gene manipulations in mice.

Weaknesses:

(1) By contrast to the GOF experiments that induce ectopic limb budding, the LOF experiments, which use dominant negative forms of Hoxa4, Hoxa5, Hoxa6, and Hoxa7, are more challenging to interpret due to the absence of data on the specificity of the dominant negative constructs. Absent such controls, one cannot be certain that effects on limb development are due to disruption of the specific Hox proteins that are being targeted.

(2) A test of their central hypothesis regarding the necessity and sufficiency of the Hox genes under investigation would be to co-transfect the neck with full-length Hoxa6/a7 AND the dnHoxA4/a5. If their hypothesis is correct, then the dn constructs should block the limb-inducing ability of Hoxa6/a7 overexpression (again, validation of specificity of the DN constructs is important here).

(3) The paper could be strengthened by providing some additional data, which should already exist in their RNA-Seq dataset, such as supplementary material that shows the actual gene expression data that are represented in the Venn diagram, heatmap, and GO analysis in Figure 3.

(4) The results of these experiments in chick embryos are rather unexpected based on previous knockout experiments in mice, and this needs to be discussed.

Author response:

Reviewer #1:

Weaknesses:

(1) The activity of the dominant negatives lacks appropriate controls. This is crucial given that mouse mutants for PG5, PG6, PG7, and three of the four PG4 genes show no major effects on limb induction or growth. Understanding these discrepancies is essential.

Given the importance of the Loss of Function (LOF) experiments, we will provide additional evidence for the validity of the dominant-negative strategy and constructs used.

(2) The authors mention redundancies in Hox activity, consistent with numerous previous reports. However, they only use single dominant-negative versions of each Hox paralog gene individually. If Hox4 and Hox5 functions are redundant, experiments should include simultaneous dominant negatives for both groups.

To clarify redundancies in Hox activity, we will test whether simultaneous expression of dominant-negative forms of more than one Hox genes induces a stronger effect compared to the expression of a single dominant-negatives Hox genes.

(3) The main conclusion that Hox4 and Hox5 provide permissive cues on which Hox6/7 induce the forelimb is not sufficiently supported by the data. An experiment expressing simultaneous dnHox4/5 and Hox6/7 is needed. If the hypothesis is correct, this should block Hox6/7's capacity to expand the limb bud or generate an extra bulge.

We agree that this is an excellent additional experiment to corroborate our conclusion and will perform this experiment in our revision.

(4) The identity of the extra bulge or extended limb bud is unclear. The only marker supporting its identity as a forelimb is Tbx5, while other typical limb development markers are absent. Tbx5 is also expressed in other regions besides the forelimb, and its presence does not guarantee forelimb identity. For instance, snakes express Tbx5 in the lateral mesoderm along much of their body axis.

To date, Tbx5 is the best marker for the forelimb. While it is true that the Tbx5 expression is broader than the limb field, this occurs only at early stages before forelimb bud formation. We will work towards a further definition of this extra bulge.

(5) It is important to analyze the skeletons of all embryos to assess the effect of reduced limb buds upon dnHox expression and determine whether extra skeletal elements develop from the extended bud or ectopic bulge.

We have analysed the cartilage structure of operated embryos with GOF experiments and found no skeletal elements within the ectopic wing bud in the neck. Additionally, in our revision, we can further analyse the wing skeleton of operated embryos with LOF experiments, which would provide more detailed assessments of the impact of dominant-negative Hox genes on wing bud formation.

Reviewer #2:

Weaknesses

(1) By contrast to the GOF experiments that induce ectopic limb budding, the LOF experiments, which use dominant negative forms of Hoxa4, Hoxa5, Hoxa6, and Hoxa7, are more challenging to interpret due to the absence of data on the specificity of the dominant negative constructs. Absent such controls, one cannot be certain that effects on limb development are due to disruption of the specific Hox proteins that are being targeted.

We will revise our manuscript to clarify the specificity of the dominant-negative strategy used.

(2) A test of their central hypothesis regarding the necessity and sufficiency of the Hox genes under investigation would be to co-transfect the neck with full-length Hoxa6/a7 AND the dnHoxA4/a5. If their hypothesis is correct, then the dn constructs should block the limb-inducing ability of Hoxa6/a7 overexpression (again, validation of specificity of the DN constructs is important here).

This is an excellent idea and we will implement the experiment in our revision.

(3) The paper could be strengthened by providing some additional data, which should already exist in their RNA-Seq dataset, such as supplementary material that shows the actual gene expression data that are represented in the Venn diagram, heatmap, and GO analysis in Figure 3.

We will incorporate this suggestion and include additional data from our RNA-seq analysis.

(4) The results of these experiments in chick embryos are rather unexpected based on previous knockout experiments in mice, and this needs to be discussed.

In our revision, we will appropriately expand the discussion on the discrepancies observed between knockout mouse models and our chick embryo experiments.