Hutchins
et al
. eLife 2022;11:e63600. DOI: https://doi.org/10.7554/eLife.63600
1 of 23
RNA-
binding protein Elavl1/HuR is
required for maintenance of cranial neural
crest specification
Erica J Hutchins
1,2
*, Shashank Gandhi
3
, Jose Chacon
4
, Michael Piacentino
1
,
Marianne E Bronner
1
1
Division of Biology and Biological Engineering, California Institute of Technology,
Pasadena, United States;
2
Department of Cell and Tissue Biology, University of
California, San Francisco, San Francisco, United States;
3
The Miller Institute for Basic
Research in Science, University of California, Berkeley, Berkeley, United States;
4
Department of Biology, School of Math and Science, California State University
Northridge, Northridge, United States
Abstract
While neural crest development is known to be transcriptionally controlled via sequen-
tial activation of gene regulatory networks (GRNs), recent evidence increasingly implicates a role for
post-
transcriptional regulation in modulating the output of these regulatory circuits. Using available
single-
cell RNA-
sequencing datasets from avian embryos to identify potential post-
transcriptional
regulators, we found that
Elavl1
, which encodes for an RNA-
binding protein with roles in transcript
stability, was enriched in the premigratory cranial neural crest. Perturbation of Elavl1 resulted in
premature neural crest delamination from the neural tube as well as significant reduction in tran-
scripts associated with the neural crest specification GRN, phenotypes that are also observed with
downregulation of the canonical Wnt inhibitor
Draxin
. That
Draxin
is the primary target for stabiliza-
tion by Elavl1 during cranial neural crest specification was shown by RNA-
sequencing, RNA immu-
noprecipitation, RNA decay measurement, and proximity ligation assays, further supporting the idea
that the downregulation of neural crest specifier expression upon Elavl1 knockdown was largely due
to loss of
Draxin
. Importantly, exogenous
Draxin
rescued cranial neural crest specification defects
observed with Elavl1 knockdown. Thus, Elavl1 plays a critical a role in the maintenance of cranial
neural crest specification via
Draxin
mRNA stabilization. Together, these data highlight an important
intersection of post-
transcriptional regulation with modulation of the neural crest specification GRN.
Editor's evaluation
In this short report, Hutchins et al. reveal expression of the RNA-
binding protein (RBP) HuR in the
neural tube and cranial neural crest of chicken embryos. Knock-
down of HuR affects expression of
Axud1 and FoxD3 (both genes associated with neural crest specification) and of the Wnt antago-
nist Draxin previously shown by the authors to regulate neural crest specification and delamination.
The authors propose that HuR associates with Draxin mRNA and demonstrate that Draxin overex-
pression can rescue FoxD3 expression upon HuR knock down. The data is in line with the idea that
control of neural crest specification by HuR at least partially involves Draxin mRNA stabilization.
Introduction
Neural crest cells are an essential, multipotent cell population in the vertebrate embryo. During
development, these cells undergo coordinated induction, specification, and epithelial–mesenchymal
SHORT REPORT
*For correspondence:
erica.hutchins@ucsf.edu
Competing interest:
See page
16
Funding:
See page 16
Preprinted:
14 October 2020
Received:
30 September 2020
Accepted:
22 August 2022
Published:
03 October 2022
Reviewing Editor:
Elizabeth
Robertson, University of Oxford,
United Kingdom
Copyright Hutchins
et al
. This
article is distributed under the
terms of the Creative Commons
Attribution License, which
permits unrestricted use and
redistribution provided that the
original author and source are
credited.
Short report
Developmental Biology
Hutchins
et al
. eLife 2022;11:e63600. DOI: https://doi.org/10.7554/eLife.63600
2 of 23
transition (EMT) events to migrate and ultimately form a myriad of tissues, including craniofacial
structures, components of the peripheral nervous system, as well as many other derivatives (
Gandhi
and Bronner, 2018
). The transcriptional control of these events has been dissected and mapped
into modules of a feed-
forward gene regulatory network (GRN), which helps explain the detailed
sequence of events involved in neural crest development (
Martik and Bronner, 2017
;
Simões- Costa
and Bronner, 2015a
;
Williams et al., 2019
). Recently, in addition to transcriptional events, there has
been growing appreciation for the role that post-
transcriptional regulation plays in the establishment,
maintenance, and regulation of neural crest formation (
Bhattacharya et al., 2018
;
Cibi et al., 2019
;
Copeland and Simoes-
Costa, 2020
;
Forman et al., 2021
;
Sánchez-
Vásquez et al., 2019
;
Ward
et al., 2018
;
Weiner, 2018
).
Given that RNA-
binding proteins play an essential role in post-
transcriptional regulatory processes
(
Dassi, 2017
), we sought to broadly identify those with early roles in neural crest development. To
this end, we analyzed existing single-
cell RNA-
sequencing (scRNA-
seq) data (
Williams et al., 2019
)
from specification-
stage avian embryos to identify enriched RNA-
binding protein candidates. Using
this approach, we identified
Elavl1
as an enriched transcript in newly formed neural crest cells. Elavl1
is a nucleocytoplasmic shuttling protein from the ELAV (embryonic lethal abnormal vision) family of
RNA-
binding proteins, which have conserved roles in neural development (
Ma et al., 1996
;
Yao et al.,
1993
). It is a well-
established stabilizer of mRNA, a function often mediated via association with the
3
′
-untranslated region (3
′
-UTR) of its mRNA targets (
Abdelmohsen and Gorospe, 2010
;
Rothamel
et al., 2021
).
Elavl1 is essential for mammalian development and embryonic survival; Elavl1 null mouse
embryos exhibit lethality due to abnormal placental morphogenesis, and conditional epiblast-
null
embryos display a broad array of phenotypes, ranging from defects in ossification and cranio-
facial development to asplenia. Interestingly, despite the myriad of tissue systems affected by
Elavl1 knockout and relatively broad expression in wild-
type embryos, mechanistic insights suggest
Elavl1 acts on specific gene networks in a spatiotemporally controlled manner (
Katsanou et al.,
2009
). Thus, due to its complexity and specificity of function, much remains to be discovered with
eLife digest
As an embryo develops, different genetic programs become activated to give cell
populations a specific biological identity that will shape their fate. For instance, when certain sets of
genes get switched on, cells from the outermost layer of the embryo start to migrate to their final
destination within the body. There, these ‘neural crest cells’ will contribute to bones and cartilage in
the face, pigmented skin spots, and muscles or nerves in the gut.
When genes responsible for the neural crest identity are active, their instructions are copied into
an ‘RNA molecule’ which will then relay this information to protein-
building structures. How well the
RNA can pass on the message depends on how long it persists within the cell. Certain RNA-
binding
proteins can control this process, but it is unclear whether and how this regulation takes place in
neural crest cells. In their work, Hutchins et al. therefore focused on identifying RNA-
binding proteins
involved in neural crest identity.
Exploratory searches of genetic data from chick embryos revealed that, even before they started
to migrate, neural crest cells which have recently acquired their identity produced large amounts of
the RNA-
binding protein Elavl1. In addition, these cells did not behave normally when embryos were
deprived of the protein: they left the outer layer too soon and then switched off genes important for
their identity. Genetic studies of neural crest cells lacking Elavl1 revealed that this effect was due to
having lost the RNA molecule produced from the Draxin gene.
Introducing an additional source of Draxin into mutant embryos missing Elavl1 was enough to
restore normal neural crest behaviour. Further biochemical experiments then showed that the RNA
for Draxin decayed quickly in the absence of Elavl1. This suggests that the protein normally allows
Draxin’s RNA to persist long enough to pass on its message.
These results reveal a new mechanism controlling the identity and behaviour of the neural crest.
Since many cancers in adulthood arise from the descendants of neural crest cells, Hutchins et al. hope
that this knowledge could lead to improved therapies in the future.
Short report
Developmental Biology
Hutchins
et al
. eLife 2022;11:e63600. DOI: https://doi.org/10.7554/eLife.63600
3 of 23
respect to Elavl1’s essential roles and targets during embryonic development across tissue-
specific
contexts.
Here, we sought to determine the role of Elavl1 during cranial neural crest specification by taking
advantage of the chick embryo model, an amniote system in which it is possible to perturb Elavl1 func-
tion with precise spatiotemporal control via unilateral knockdown. The results show that perturbation
of Elavl1 led to premature neural crest delamination, as well as significant reduction in the expression
of genes within the neural crest specification GRN. We find that these effects were mediated by loss
of
Draxin
, a direct mRNA target for stabilization by Elavl1. Our data demonstrate a critical role for
Elavl1, and RNA-
binding protein-
mediated post-
transcriptional control, in the regulation of a critical
neural crest specification module.
Results
The RNA-binding protein Elavl1/HuR is expressed in cranial neural crest
Cranial neural crest cells are indispensable for proper craniofacial development (
van Limborgh et al.,
1983
;
Vega-
Lopez et al., 2018
). Whereas transcription factors have been well-
established critical
regulators of neural crest development and craniofacial morphogenesis reviewed in
Gou et al., 2015
,
growing evidence indicates an essential role for post-
transcriptional regulation in these processes
(
Cibi et al., 2019
;
Copeland and Simoes-
Costa, 2020
;
Dennison et al., 2021
;
Forman et al., 2021
).
To identify RNA-
binding proteins with potential roles in cranial neural crest specification, we analyzed
scRNA-
seq data for cranial neural crest isolated from avian embryos at the 5–6 somite stage (
Williams
et al., 2019
) we identified three distinct clusters (neural, premigratory [pNC], and delaminating/migra-
tory [mNC]) among which genes associated with the gene ontology (GO) term ‘binds to 3
′
-UTR’ were
differentially expressed (
Figure 1A–E
). To isolate potential positive regulators of neural crest speci-
fication, we then performed GO term analysis for ‘stabilizes RNA’ and ‘regulates translation’ among
the identified RNA-
binding proteins; only the
Elavl1
gene was associated with all three GO terms and
abundantly expressed in the isolated cranial neural crest cells (
Figure 1F, G
).
Given that Elavl1 knockout mice often display defects in craniofacial structures (
Katsanou et al.,
2009
), and the transcript appeared enriched in premigratory cranial neural crest cells (
Figure 1E
),
we hypothesized a potential role for Elavl1 during cranial neural crest specification. To test this possi-
bility, we first examined the expression pattern of Elavl1 in the developing chick embryo. Early in
neurulation, when the neural plate border (NPB) is established within the rising neural folds, Elavl1
expression was detected in the anterior open neural tube and closing neural folds surrounding the
anterior neuropore but absent from Pax7-
expressing NPB cells (
Figure 2A
). As the neural tube closed,
when neural crest specification is complete, Elavl1 expression became enriched throughout the neural
tube and overlapped with Pax7 expression in premigratory cranial neural crest cells (
Figure 2B–D
).
Following cranial neural crest EMT, Elavl1 remained expressed in the migratory neural crest cells, as
well as throughout the brain and neural tube (
Figure 2E–G
). Thus, Elavl1 is expressed in specified,
premigratory cranial neural crest cells following establishment of the NPB and is retained during the
onset of EMT and in early migrating cranial neural crest cells.
Elavl1 downregulation alters cranial neural crest specification and
delamination
To determine what, if any, role Elavl1 has in cranial neural crest specification, we perturbed Elavl1
function in the early embryo using a translation-
blocking antisense morpholino oligo (MO). We elec-
troporated control or Elavl1 MOs bilaterally into gastrula stage chick embryos and analyzed neural
crest specification using quantitative fluorescent hybridization chain reaction (HCR) to measure
expression of markers of specified neural crest at HH9 (
Figure 3A, B
). Given Elavl1’s association with
Wnt signaling (
Kim et al., 2015
) and the essential roles Wnt signaling plays during early neural crest
development (
Milet and Monsoro-
Burq, 2012
;
Rabadán et al., 2016
;
Simões-
Costa and Bronner,
2015a
;
Steventon and Mayor, 2012
;
Wu et al., 2003
;
Yanfeng et al., 2003
), we focused on the Wnt
effector
Axud1
, its target and neural crest specifier
FoxD3
, and the Wnt antagonist
Draxin
(
Hutchins
and Bronner, 2018
;
Hutchins and Bronner, 2019
;
Simões-
Costa et al., 2015b
). Following Elavl1
knockdown (
Figure 3—figure supplement 1
; 61.4 ± 0.9% of the control side, p < 0.001, paired
t
- test,
n
= 5 embryos, 15 sections), we observed significant reduction in the levels of
Axud1
(
Figure 3C
; 73.7
Short report
Developmental Biology
Hutchins
et al
. eLife 2022;11:e63600. DOI: https://doi.org/10.7554/eLife.63600
4 of 23
Figure 1.
RNA-
binding proteins are differentially expressed in premigratory and migratory cranial neural crest. (
A
) Schematic of early chick cranial
neural crest cells at premigratory stages (HH8+/9−) in intact heads and cross-
section expressing Citrine fluorescent protein under control of the
FoxD3 NC1 enhancer used by Williams et al. to sort cranial neural crest cells for single-
cell RNA-
sequencing (scRNA-
seq;
Williams et al., 2019
). (
B,
C
) Dimensionality reduction using Uniform Manifold Approximation and Projection (UMAP) on published scRNA-
seq data from
Williams et al., 2019
Figure 1 continued on next page
Short report
Developmental Biology
Hutchins
et al
. eLife 2022;11:e63600. DOI: https://doi.org/10.7554/eLife.63600
5 of 23
± 3.1% of the control side, p = 0.03, Wilcoxon signed-
rank test,
n
= 6 embryos),
FoxD3
(
Figure 3D
;
61.3 ± 5.5% of the control side, p = 0.002, Wilcoxon signed-
rank test,
n
= 10 embryos), and
Draxin
(
Figure 3E
; 66.6 ± 1.8% of the control side, p = 0.03, Wilcoxon signed-
rank test,
n
= 6 embryos)
transcripts compared to contralateral control sides. To parse whether this was a specific effect or
a broad defect in neural crest development, we also examined additional neural crest genes and
found no significant difference in expression between control or Elavl1 knockdown sides for
Pax7
(
Figure 3F
; 87.4 ± 3.1% of the control side, p = 0.1, Wilcoxon signed-
rank test,
n
= 5 embryos) or
Tfap2b
(
Figure 3G
; 98.2 ± 7.0% of the control side, p = 0.6, Wilcoxon signed-
rank test,
n
= 4 embryos),
indicating a specific defect in a subset of genes required for cranial neural crest specification.
We also performed immunostaining for Pax7 in cross-
section to assess if Elavl1 knockdown altered
neural crest cell number or dorsal neural tube morphology. Interestingly, the total number of Pax7 +
cells was unaffected with Elavl1 knockdown (101.3 ± 4.5% of the control side, p = 0.6, one-
sample
Wilcoxon signed-
rank test,
n
= 4 embryos, 11 sections); however, we found a significant increase in
the number of Pax7 + cells that delaminated from the neural tube (139.4 ± 8.4% of the control side,
p = 0.002, one-
sample Wilcoxon signed-
rank test), and concomitant decrease in the number of Pax7
+ cells retained within the dorsal neural tube (66.0 ± 5.3% of the control side, p = 0.001, one-
sample
Wilcoxon signed-
rank test:
Figure 3H, I
). To determine what impact premature delamination might
have on cranial neural crest migration, we also performed whole mount immunostaining for Pax7
at HH9 + and found significant decrease in cranial neural crest emigration away from the midline
(
Figure 3—figure supplement 2
; 63.6 ± 4.1% of the control side, p < 0.001, Wilcoxon matched-
pairs
signed- rank test,
n
= 5 embryos, 5 measurements per embryo averaged). Taken together, these data
suggest that Elavl1 is required during early cranial neural crest development to regulate specification
and prevent premature delamination.
Draxin
is the primary target of Elavl1 during cranial neural crest
specification
To parse the mechanism of Elavl1 function during cranial neural crest specification, we sought to
more broadly identify the RNA targets of Elavl1 using bulk RNA-
sequencing (RNA-
seq) and differen-
tial gene expression analysis (
Figure 4A
). Given that Elavl1 is known to bind to and stabilize its RNA
targets via 3
′
-UTR interaction (
Chen et al., 2002
;
Dormoy-
Raclet et al., 2007
;
Katsanou et al., 2009
;
Rothamel et al., 2021
;
Shi et al., 2020
), we expected expression of potential targets to be reduced
with Elavl1 knockdown. Among the differentially expressed genes we identified (
Figure 4—figure
supplement 1
), 12 were significantly downregulated, with four having established roles in neural
crest development and which we validated using HCR—
Axud1
,
Draxin
,
BMP4
, and
Msx1
(
Figure 4B
,
Figure 4—figure supplement 1
). We also identified several canonical neural crest genes that were
unaffected with Elavl1 knockdown (e.g
.
,
Pax7
,
Tfap2b
,
Snai2
,
Sox9
, and
Zeb2
), consistent with HCR
data (
Figure 3
,
Figure 4—figure supplement 1
). Together these data suggest Elavl1 does not broadly
bind and stabilize the transcripts of neural crest genes, rather it targets specific RNAs to drive cranial
neural crest specification.
Notably,
FoxD3
failed to meet our stringency cutoff during RNA-
seq analysis due to low expres-
sion at the examined stages, and is likely downregulated with Elavl1 knockdown (
Figure 3D
) due to
indirect effects from loss of
Axud1
(
Simões-
Costa et al., 2015b
) and therefore unlikely to be a
bona
fide
target of Elavl1. To determine whether Elavl1 directly or indirectly interacts with
Axud1
,
Draxin
,
BMP4
, and
Msx1
mRNAs, we first measured the rate of mRNA decay with actinomycin D treatment
to assess the stability of these RNAs, and
Pax7
as a nontarget, with or without Elavl1 knockdown
identified three distinct clusters neural, premigratory (pNC), and delaminating/migratory (mNC). Expression of marker genes (
Sox3
and
Cdh2
for neural,
Pax7
and
Snai2
for premigratory cranial neural crest, and
Sox10
and
Ets1
for early migratory cranial neural crest) was used to label the three subclusters.
(
D
) The majority of the FoxD3-
NC1
+
cells were also positive for the expression of transcription factors
Pax7
and
Snai2
, which label the dorsal neural tube
and premigratory/delaminating neural crest cells. These triple positive NC1
+
/Snai2
+
/Pax7
+
were further processed for gene ontology analysis to identify
post-
transcriptional regulators. (
E
) A strip-
plot showing expression and abundance of a subset of genes that are associated with the gene ontology
term ‘binds to 3
′
-UTR’. (
F
) A three-
way Venn diagram shows overlap between genes associated with the gene ontology terms ‘binds RNA’, ‘regulates
translation’, and ‘stabilizes RNA’. Only three genes,
Elavl1
,
Dazl
, and
Igf2bp1
, were associated with all three. (
G
) Feature plots showing the expression
distribution of the three genes identified in (
F
). Only
Elavl1
is abundant among all NC1
+
/Pax7
+
/Snai2
+
cells.
Figure 1 continued
Short report
Developmental Biology
Hutchins
et al
. eLife 2022;11:e63600. DOI: https://doi.org/10.7554/eLife.63600
6 of 23
Figure 2.
The RNA-
binding protein Elavl1 is expressed in premigratory and migratory cranial neural crest.
Representative epifluorescence images of wild-
type HH8− (
A
), HH9 (
B–D
), and HH9+ (
E–G
) chick embryos, in
whole mount (
A, B, E
) and cross-
section (
C, D; F, G
) immunostained for Elavl1 (cyan) and Pax7 (magenta). Dashed
white line (
B, E
) indicates level of cross-
section (
C, D; F, G
), respectively; dotted white lines outline regions of
Figure 2 continued on next page
Short report
Developmental Biology
Hutchins
et al
. eLife 2022;11:e63600. DOI: https://doi.org/10.7554/eLife.63600
7 of 23
(
Figure 4C
); if Elavl1 is required for transcript stability, target RNAs should decay at a faster rate with
loss of Elavl1 compared to control. Interestingly, among all the transcripts tested, only
Draxin
had a
significant reduction in transcript stability (p = 0.001, Mann–Whitney test;
Figure 4D
), suggesting that
Draxin
is the primary target bound and stabilized by Elavl1 during cranial neural crest specification.
To test this hypothesis, we searched the 3
′
-UTR sequences of
Draxin
,
Axud1
,
BMP4
,
Msx1
,
Pax7
, and
Tfap2b
mRNAs for putative Elavl1-
binding sites. The
Draxin
3
′
-UTR contained four high probability-
binding sites, whereas the other 3
′
-UTRs contained only one (
BMP4
,
Msx1
) or none (
Axud1
,
Pax7
,
Tfap2b;
Figure 4—figure supplement 2
). Given that Elavl1 contains three RNA recognition motifs
(RRMs) that cooperate for RNA recognition (
Pabis et al., 2019
), it is possible that multiple contacts
are required for Elavl1 to bind and stabilize RNAs in vivo, and by extension is unlikely for Elavl1 to bind
RNAs with only a single putative binding site. Thus, we hypothesize that Elavl1 specifically targets and
stabilizes
Draxin
in cranial neural crest through multiple contact sites within its 3
′
-UTR.
To confirm that
Draxin
mRNA is bound by Elavl1 in vivo, we first performed an RNA immunoprecip-
itation (RIP) followed by quantitative reverse transcription-
PCR (qRT-
PCR) to pull down endogenous
Elavl1 ribonucleoprotein (RNP) complexes. To this end, we incubated lysate generated from wild-
type
HH9 embryonic heads with magnetic beads coated with either Elavl1 antibody or a rabbit IgG nonspe-
cific control antibody, then eluted bound RNA and performed qRT-
PCR, comparing immunoprecipi-
tated RNAs (‘IP’) with RNAs extracted from a fraction of the input lysate (‘Input’). We expected that
RNAs bound by Elavl1 would be enriched in the IP compared to the Input, whereas nonspecifically
associated RNAs, that is nontargets, would not (
Figure 4E
). We found that
Pax7
and
FoxD3
mRNAs
(nontargets) were neither enriched in the IP, nor significantly different from each other (p = 0.89, one-
way analysis of variance (ANOVA) with Tukey’s post hoc test); however,
Draxin
mRNA was significantly
enriched with Elavl1 IP compared to
Pax7
and
FoxD3
(p < 0.001, one-
way ANOVA with Tukey’s post
hoc test;
Figure 4F
), suggesting that Elavl1 specifically associated with endogenous
Draxin
mRNA.
However, this assay could not distinguish whether
Draxin
was bound directly or indirectly, or where
within the transcript it might be associating with Elavl1.
To test whether
Draxin
was directly bound by Elavl1 within the 3
′
-UTR, we performed a proximity
ligation assay (PLA), wherein in situ fluorescent signal can be detected as puncta only when two
proteins are in close proximity (<40 nm), indicating a direct interaction in vivo. Taking advantage of
the MS2-
MCP reporter system (
Tutucci et al., 2018
), we electroporated a construct encoding a GFP-
tagged MS2 bacteriophage coat protein (MCP-
GFP) alone (‘Control’) or in combination with (‘Exper
-
imental’) a construct containing a Luciferase coding region, MS2 stem loops (bound by MCP when
transcribed), and the endogenous
Draxin
3
′
-untranslated region (MS2-
Draxin
3
′
-UTR) and performed
PLA with antibodies against Elavl1 and GFP (
Figure 4G
). We observed significantly more PLA puncta
with expression of MS2-
Draxin
3
′
-UTR (
Figure 4H–I
), indicating a specific and direct interaction
between Elavl1 and the
Draxin
3
′
-UTR.
Elavl1 maintains cranial neural crest specification via
Draxin
mRNA
stabilization
Our data suggest that Elavl1 specifically binds and stabilizes
Draxin
mRNA as its primary target during
cranial neural crest specification. To determine if defects in cranial neural crest specification with Elavl1
knockdown were indirect, and due to
Draxin
downregulation, we examined
FoxD3
,
Axud1
,
Msx1
,
and
BMP4
expression in Draxin knockdown embryos. We electroporated control and Draxin MO
(
Hutchins and Bronner, 2018
;
Hutchins and Bronner, 2019
) bilaterally, and performed HCR at HH9.
As with Elavl1 MO (
Figure 3
), we observed significant reduction in the levels of
FoxD3
(
Figure 5A
;
26.8 ± 3.2% of the control side, p < 0.02, Wilcoxon matched-
pairs signed rank,
n
= 7 embryos),
Axud1
(
Figure 5B
; 48.8 ± 5.1% of the control side, p < 0.01, paired
t
- test,
n
= 7 embryos),
Msx1
(
Figure 5C
;
49.7 ± 5.8% of the control side, p < 0.01, paired
t
- test,
n
= 4 embryos), and
BMP4
(
Figure 5D
; 63.6 ±
4.0% of the control side, p < 0.01, paired
t
- test,
n
= 4 embryos) transcripts compared to contralateral
control sides (
Figure 5E
).
premigratory and migratory neural crest as indicated. NF, neural folds; NT, neural tube; pNC, premigratory neural
crest; mNC, migratory neural crest. Scale bar, 50 μm.
Figure 2 continued
Short report
Developmental Biology
Hutchins
et al
. eLife 2022;11:e63600. DOI: https://doi.org/10.7554/eLife.63600
8 of 23
Figure 3.
Elavl1 knockdown alters cranial neural crest specification and delamination. (
A
) Schematic diagram illustrating experimental design. Gastrula
stage chick embryos were electroporated bilaterally with a standard control and translation-
blocking morpholino (MO) targeting
Elavl1
. Electroporated
embryos were subsequently processed for quantitative hybridization chain reaction (HCR) and analyzed in whole mount, comparing the knockdown
to the contralateral control side. (
B
) Quantitation of HCR processed embryos for control versus Elavl1 knockdown for cranial neural crest transcripts,
calculated as ratio of Elavl1 MO versus control MO integrated density. Representative confocal maximum intensity projection micrographs for
Axud1
(
n
= 6) (
C
),
FoxD3
(
n
= 10) (
D
),
Draxin
(
n
= 6) (
E
),
Pax7
(
n
= 5) (
F
), and
Tfap2b
(
n
= 4) (
G
) transcripts. Dotted white line indicates midline. MO, morpholino.
Scale bar, 50 μm. *p < 0.05, Wilcoxon signed-
rank test. (
H
) Representative apotome maximum intensity projection micrographs of cross-
sectioned
embryo bilaterally co-
electroporated with a fluorescent electroporation control construct (H2B-
RFP) and control MO (left) or Elavl1 MO (right)
immunostained for Pax7 (yellow). Nuclei were stained with DAPI (4
′
,6-
diamidino-
2- phenylindole)(blue). Dotted white line indicates midline. Dashed
white lines indicate limit of dorsal neural tube. Arrows indicate ‘neural tube’ Pax7 cells. Asterisks indicate ‘delaminated’ Pax7 cells. Scale bar, 20 μm. (
I
)
Figure 3 continued on next page
Short report
Developmental Biology
Hutchins
et al
. eLife 2022;11:e63600. DOI: https://doi.org/10.7554/eLife.63600
9 of 23
We next asked whether
Draxin
upregulation alone was sufficient to rescue the Elavl1 MO pheno-
type (
Figure 3B–E
). To this end, we co-
electroporated Elavl1 MO with a Draxin overexpression
construct (Draxin- FLAG;
Hutchins and Bronner, 2018
;
Hutchins and Bronner, 2019
), and assessed
neural crest specification with HCR. Indeed, exogenous
Draxin
was sufficient to significantly restore
FoxD3
(
Figure 5F
; 84.5 ± 9.6% of the control side,
n
= 6 embryos),
Axud1
(
Figure 5G
; 90.3 ± 5.0% of
the control side,
n
= 6 embryos),
Msx1
(
Figure 5H
; 91.0 ± 1.5% of the control side,
n
= 5 embryos),
and
BMP4
(
Figure 5I
; 96.5 ± 3.1% of the control side,
n
= 5 embryos) expression from Elavl1 knock-
down (p < 0.05, one-
tailed paired
t
-
test) to expression levels not significantly different from control
(
Figure 5J
; p > 0.09, Wilcoxon signed-
rank test). To determine if exogenous
Draxin
was sufficient to
rescue the premature delamination phenotype (
Figure 3H, I
) caused by Elavl1 knockdown, we also
performed immunostaining for Pax7 in cross-
section to assess neural crest cell number and dorsal
neural tube morphology. Indeed, co-
electroporation of Draxin-
FLAG with Elavl1 MO was able to
rescue the number of Pax7 + cells that delaminated and remained within the dorsal neural tube to
near-
control levels (
Figure 5L
; 91.6 ± 4.0% and 105.8 ± 12.0% of the control side, respectively,
n
=
3 embryos, 6 sections, p > 0.12, one-
sample Wilcoxon signed-
rank test). Taken together, these data
indicate that
Draxin
mRNA is the primary target of Elavl1 in premigratory cranial neural crest and is
stabilized via 3
′
-UTR interaction to maintain neural crest specification.
Discussion
Understanding of neural crest development has been greatly enhanced by the identification of key
transcriptional circuits that control its developmental progression. Recent studies suggest a critical
role for post-
transcriptional regulation in the refinement of the expression outputs of these GRNs.
Here, we identified and characterized Elavl1 as an RNA-
binding protein essential for the maintenance
of cranial neural crest specification via its stabilization of the Wnt antagonist
Draxin
. We found that
loss of Elavl1, and by extension its target,
Draxin
, interferes with output from multiple nodes of the
GRNs required for neural crest specification, including those driven by Wnt and BMP (
Hovland et al.,
2020
;
Simões-
Costa et al., 2015b
;
Tribulo et al., 2003
). Together, our data implicate Elavl1 as a point
of integration to coordinate signaling from parallel but independent GRNs.
At a mechanistic level, our study is consistent with previous work examining Elavl1 function in other
cellular contexts, with respect to its role as a stabilizing RNA-
binding protein via 3
′
-UTR interaction
(
Chen et al., 2002
;
Dormoy-
Raclet et al., 2007
;
Katsanou et al., 2009
;
Rothamel et al., 2021
;
Shi
et al., 2020
). In the context of embryonic development, our study aligns well with data from knockout
mouse consistent with an important role for Elavl1 in intersecting signaling cascades; interestingly, this
prior work similarly observed indirect downregulation of
BMP4
with loss of Elavl1, though an upstream
mediator remained unidentified (
Katsanou et al., 2009
). In neural crest specification, we identified a
single biologically relevant target of Elavl1, whereas prior high-
throughput studies found many RNAs
directly bound by Elavl1 (
Mukherjee et al., 2011
;
Rothamel et al., 2021
). Whether this is a feature
specific to neural crest is unclear, though the knockout mouse work suggests that during embryonic
development Elavl1 function and RNA targets are driven by spatiotemporal determinants (
Katsanou
et al., 2009
), which may be due RNP complex heterogeneity as a result of tissue-
specific expression
of other interacting RNA-
binding proteins. Taken in the context of these prior studies, we speculate
that despite broad tissue expression of Elavl1 across embryonic development, specificity in neural
crest is achieved through spatiotemporally regulated combinatorial expression of post-
transcriptional
regulators and RNA regulons (
Keene, 2007
;
Keene and Tenenbaum, 2002
).
It is important to note that, while Elavl1 expression persists in cranial neural crest during the initi-
ation of EMT and migration,
Draxin
must be rapidly downregulated for these processes to proceed
Quantification of the ratio of Pax7 + cells on Elavl1 MO (right) versus control MO (left) sides of cross-
sections. Data are from individual sections; sections
from same embryo are displayed in same color (
n
= 4 embryos, 11 sections). *p ≤ 0.002, one-
sample Wilcoxon signed-
rank test.
The online version of this article includes the following figure supplement(s) for figure 3:
Figure supplement 1.
Translation-
blocking morpholino suppresses Elavl1 expression.
Figure supplement 2.
Elavl1 knockdown inhibits cranial neural crest emigration.
Figure 3 continued
Short report
Developmental Biology
Hutchins
et al
. eLife 2022;11:e63600. DOI: https://doi.org/10.7554/eLife.63600
10 of 23
Figure 4.
Draxin
mRNA is the primary target of Elavl1 during cranial neural crest specification. (
A
) Schematic diagram illustrating experimental
design for RNA-
sequencing (RNA-
seq). Gastrula stage chick embryos were electroporated bilaterally with a standard control and translation-
blocking
morpholino (MO) targeting
Elavl1
. Dorsal neural folds were dissected from stage HH8+/9− embryos, pooled (
n
= 3), and processed for bulk RNA-
seq
(three biological replicates). (
B
) Volcano plot following differential expression analysis and filtering of RNA-
seq data. Of the 24 genes differentially
Figure 4 continued on next page
Short report
Developmental Biology
Hutchins
et al
. eLife 2022;11:e63600. DOI: https://doi.org/10.7554/eLife.63600
11 of 23
(
Hutchins and Bronner, 2018
;
Hutchins and Bronner, 2019
;
Hutchins et al., 2021
). Thus, we hypoth-
esize that Elavl1 becomes endogenously displaced from
Draxin
at the onset of EMT, though it is yet
unclear how this is achieved. RNA-
binding proteins are known to alter association with targets due to
post-
translational modifications such as phosphorylation or alternative RNA-
binding protein compe-
tition (
Dassi, 2017
;
García-
Mauriño et al., 2017
;
Liu et al., 2015
). Indeed, inhibition of serine–thre-
onine kinases has been shown in neural crest to increase cell–cell adhesions and negatively impact cell
migration (
Monier-
Gavelle and Duband, 1995
), suggesting kinase-
driven signaling pathway activa-
tion coincident with neural crest EMT; this is interesting given established roles for serine–threonine
phosphorylation in modulating Elavl1-
RNA-
binding activity and target selection (
Grammatikakis
et al., 2017
). Thus, we speculate that Elavl1 serine–threonine phosphorylation, either in combination
with or as an alternative to competitive RNA-
binding protein displacement, likely facilitates
Draxin
release and turnover at the onset of EMT.
Because its primary target during specification is downregulated while Elavl1 expression persists,
this suggests there are likely additional targets and roles for Elavl1 beyond
Draxin
stabilization. Elavl1
has been shown in other contexts to stabilize Snail1 (
Dong et al., 2007
) and matrix metalloprotease-
9
(MMP- 9) (
Yuan et al., 2011
), factors with well-
established roles in neural crest EMT (
Cano et al., 2000
;
Kalev-
Altman et al., 2020
;
Monsonego-
Ornan et al., 2012
;
Strobl-
Mazzulla and Bronner, 2012
;
Taneyhill et al., 2007
). Given that the neural crest specification GRNs is proceeded by activation of
an EMT GRN also coinciding with Elavl1 expression, we speculate that Elavl1 likely intersects with
additional GRNs following specification. It is possible one or more of the downregulated genes we
identified through RNA-
seq that were not important for specification may yet be targets of Elavl1,
with functions later in neural crest development. To fully understand how Elavl1 exerts control over
key developmental processes, these possibilities will need to be explored.
expressed following Elavl1 knockdown (12 upregulated, 12 downregulated), four genes (
Draxin
,
Axud1
,
Msx1
,
BMP4
) have established roles in neural
crest development and were significantly downregulated. (
C
) Schematic diagram illustrating experimental design for RNA stability assay. Gastrula stage
chick embryos were electroporated bilaterally with control or translation-
blocking morpholino (MO) targeting
Elavl1
. Dorsal neural folds were dissected
from stage HH8+/9− embryos; left neural folds were used as the 0 hr time point, whereas right neural folds were treated with actinomycin D for 30 min,
2 hr, or 4 hr prior to total RNA extraction and quantitative reverse transcription-
PCR (qRT-
PCR) to measure RNA decay. (
D
) Transcript stability plots show
Draxin
mRNA stability is significantly reduced (*p = 0.001, Mann–Whitney test) with Elavl1 knockdown (blue) compared to control (gray), whereas other
neural crest mRNAs (
Axud1
,
Msx1
,
BMP4
,
Pax7
) are not (
ns
, nonsignificant, p > 0.37, Mann–Whitney test). (
E
) Schematic illustrating experimental design
of RNA-
binding protein/RNA co-
immunoprecipitation (RIP) to test RNA association with Elavl1 in vivo for neural crest targets. Lysates generated from
HH9 heads were incubated with antibody-
coated beads for Elavl1 or a nonspecific IgG to co-
immunoprecipitate protein with bound RNAs. In qRT-
PCR,
specifically bound RNAs would be more abundant and reach threshold before RNAs that were nonspecific, and therefore would have smaller
C
T
values.
C
T
, threshold cycle. (
F
) Fold enrichment of RNAs eluted from RIP (
n
= 16 embryos), quantified by qRT-
PCR, performed in triplicate.
ns
, nonsignificant,
p = 0.89, one-
way analysis of variance (ANOVA) with Tukey’s post hoc test. *p < 0.001, one-
way ANOVA with Tukey’s post hoc test. Error bars, standard
error of the mean (SEM). (
G
) Schematic diagram illustrating experimental design for proximity ligation assay (PLA). Gastrula stage chick embryos were
electroporated bilaterally with a construct expressing a nuclear localized, GFP-
tagged MS2 bacteriophage coat protein (MCP-
GFP) alone (left) or in
combination with a construct containing a Luciferase coding region, MS2 stem loops (which are bound by MCP when transcribed), and the endogenous
Draxin
3
′
-untranslated region (MS2-
Draxin
3
′
UTR) (right). Following fixation and cross-
sectioning at HH9, tissues were incubated with primary
antibodies made in goat and rabbit that recognized GFP and Elavl1, respectively. Secondary antibodies against goat and rabbit IgG were labeled with
complementary oligonucleotides that generate a fluorescent signal due to rolling circle amplification only when in close proximity (<40 nm). Thus,
fluorescence signal (magenta) would indicate in vivo interaction between MCP-
GFP and endogenous Elavl1. (
H
) Representative confocal maximum
intensity projection micrograph of dorsal neural folds from cross-
sectioned HH9 embryo bilaterally electroporated with MCP-
GFP (green) alone
(‘control’, left) or with MS2-
Draxin
3
′
-UTR (‘experimental’, right), processed for PLA (magenta) as illustrated in panel (
G
), and stained for DAPI (blue).
Boxes in (
H
) indicate zoomed-
in areas in (
H’’
) and (
H’’’
). Scale bar, 5 μm. (
I
) Quantitation of total number of PLA puncta per section for (
n
= 3 embryos, 2
sections/embryo). *p = 0.016, two-
tailed Wilcoxon matched-
pairs signed-
rank test.
The online version of this article includes the following figure supplement(s) for figure 4:
Figure supplement 1.
RNA-
sequencing (RNA-
seq) identified four neural crest genes specifically downregulated with Elavl1 knockdown.
Figure supplement 2.
Putative Elavl1-
binding sites within the
Draxin
3
′
-untranslated region (UTR) predicts a direct interaction.
Figure 4 continued