bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
Asymmetric robustness in the feedback control of the retinoic acid network response to
environmental disturbances
Madhur Parihar1,$, Liat Bendelac-Kapon2,$, Michal Gur2,$, Abha Belorkar1, Sirisha Achanta1,
Keren Kinberg2, Rajanikanth Vadigepalli1,*, Abraham Fainsod2,*
1Daniel Baugh Institute for Functional Genomics/Computational Biology, Department of
Pathology, Anatomy, and Cell Biology, Sidney Kimmel Medical College, Thomas Jefferson
University, Philadelphia, Pennsylvania, 19107 USA
2Department of Developmental Biology and Cancer Research, Institute for Medical Research
Israel-Canada, Faculty of Medicine, Hebrew University, Jerusalem 9112102, Israel
$equal contribution
*Corresponding authors: [email protected]
Review Codes for online datasets GEO SuperSeries GSE154408: cryfqygubnszryt
https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE154408
1 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
SUMMARY
Retinoic acid (RA) is a developmental signal whose perturbation is teratogenic. We show that
early embryos exhibit effective RA signaling robustness to physiological, non-teratogenic,
disturbances of this pathway. Transcriptomic analysis of transient physiological RA
manipulations during embryogenesis supported the robustness of RA signaling by identifying
mainly changes consistent with the progression of embryogenesis and not dramatic treatment-
induced changes. Transcriptomic pattern comparisons revealed that RA manipulation led to a
network-wide feedback regulation aimed at achieving signaling robustness and normalizing
RA levels. A trajectory analysis of target gene and RA network responses identified an
asymmetric robustness with a high sensitivity to reduced RA levels, and an activation threshold
to increased levels. Furthemore, high robustness to increased RA inversely correlated with a
low response to reduced RA. Biological replicates with similar robustness levels mounted
responses whose composition likely varies based on genetic polymorphisms to achieve similar
outcomes providing insights on the robustness mechanisms.
KEYWORDS
Embryo development; Retinoic acid; Xenopus embryo; time-series transcriptomics; temporal
gene expression pattern analysis; developmental trajectory analysis; autoregulatory feedback
control.
2 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
INTRODUCTION
Retinoic acid (RA) is one of the central regulatory signaling pathways active during
vertebrate embryogenesis as well as in adult tissue homeostasis regulating the transcription of
numerous downstream target genes (Campo-Paysaa et al., 2008; Clagett-Dame and DeLuca,
2002; le Maire and Bourguet, 2014; Marill et al., 2003; Mark et al., 2009; Metzler and Sandell,
2016). RA is synthesized from vitamin A (retinol) or other retinoids or carotenoids obtained
from the diet (Ghyselinck and Duester, 2019; Kedishvili, 2016). During embryogenesis,
changes in RA levels, signal timing or signal localization, result in severe developmental
malformations arising from both abnormally low and increased RA signaling. Excessive RA
signaling induces developmental malformations including brain defects, organ malformations
and additional anatomical anomalies (Clagett-Dame and Knutson, 2011; Collins and Mao,
1999; Cunningham and Duester, 2015; Marill et al., 2003; Mark et al., 2009; Shenefelt, 1972).
Syndromes linked to reduced RA signaling include vitamin A deficiency syndrome (VAD),
DiGeorge/VeloCardioFacial syndrome (DG/VCF), Fetal Alcohol Spectrum Disorder (FASD),
Congenital Heart Disease (CHD), neural tube defects, and multiple types of cancer (Coberly et
al., 1996; El Kares et al., 2010; Hartomo et al., 2015; Kim et al., 2005; Kot-Leibovich and
Fainsod, 2009; Pangilinan et al., 2014; See et al., 2008; Timoneda et al., 2018; Urbizu et al.,
2013). RA levels are tightly regulated at multiple levels throughout life to prevent aberrant
gene expression as a result of diet and other environmental changes. Discrete regulatory roles
of RA are usually separated temporally and spatially, taking place in different tissues,
embryonic regions, or cell types, requiring the fine-tuned regulation of the source, the level,
and the gene-regulatory response to this signal. This quantitative, spatial and temporal
regulation relies in part on the regulated expression and activity of RA biosynthetic and
metabolizing enzymes(Dobbs-McAuliffe et al., 2004; Duester et al., 2003; Hollemann et al.,
1998; Sakai et al., 2001).
3 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
RA biosynthesis involves two sequential oxidation steps: first, mainly alcohol
dehydrogenases (ADH) or short-chain dehydrogenase/reductases (SDR) oxidize vitamin A
(retinol, ROL) to retinaldehyde (RAL), followed by the retinaldehyde dehydrogenase
(RALDH) catalyzed oxidation of RAL to RA (Duester, 2008; Kedishvili, 2016; Parés et al.,
2008). RA availability is further affected by ROL, RAL, and RA binding proteins (Kono and
Arai, 2015; Napoli, 2017, 2016; Schroeder et al., 2008). ROL and RAL can also be produced
from retinyl ester stores or from ß-carotene from food sources (Blaner, 2019; O’Byrne and
Blaner, 2013). In vertebrate gastrula embryos, RA signaling is triggered by the activation of
raldh2 (aldh1a2) transcription whose protein product completes the last enzymatic step in RA
biosynthesis (Begemann et al., 2001; Chen et al., 2001; Grandel et al., 2002; Niederreither et
al., 1999). Then, RALDH2 expression and availability is the earliest rate-limiting step in RA
biosynthesis. During RA biosynthesis, substrate availability for the RALDH enzymes, RAL, is
controlled by members of the SDR, ADH and AKR families (Adams et al., 2014; Billings et
al., 2013; Feng et al., 2010; Porté et al., 2013; Shabtai et al., 2016; Shabtai and Fainsod, 2018).
Importantly, expression of many of the enzymes involved in RA biosynthesis is spatially
regulated, resulting in a gradient of RA activity peaking in the caudal end of the embryo (Dubey
et al., 2018; Niederreither et al., 1997; Schilling et al., 2016). Additional spatial and temporal
regulation of this signaling pathway is provided by regulated expression of other components,
including retinoic acid receptors (RAR and RXR) and retinoid-binding proteins (Cui et al.,
2003; Janesick et al., 2015; Lohnes et al., 1995; Mendelsohn et al., 1994; Xavier-Neto et al.,
2015). Besides the maternal nutritional status that can affect the levels of RA signaling in the
developing embryo, environmental exposure to chemicals such as alcoholic beverages
(ethanol) or other chemicals can affect the biosynthesis of RA or the status of this signaling
pathway (Paganelli et al., 2010; Shabtai et al., 2018). These observations point to the close
interaction of RA signaling and the environment and the necessity to adapt the RA signaling
4 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
network to nutritional changes and insults (e.g., ethanol). This adaptation and maintenance of
normal signaling levels under changing conditions is termed robustness (Eldar et al., 2004;
Nijhout et al., 2019). Taken together, RA metabolic and gene-regulatory components are under
feedback regulation by RA signaling which may provide robustness to the RA signal
homeostasis.
A deeper understanding of the RA signaling pathway during embryogenesis is required to
elucidate its multiple regulatory roles, and its regulation of the signaling robustness in the
presence of environmental disturbances. Very commonly, the RA pathway is studied by
increasing the levels of this signal from exogenous sources (Durston et al., 1989; Kessel, 1992;
Sive et al., 1990). Alternatively, loss-of-function studies take advantage of RAR inhibitors,
inverse agonists, inhibitors of RA biosynthesis, or degradation of the signal (Hollemann et al.,
1998; Janesick et al., 2014, 2013; Kot-Leibovich and Fainsod, 2009). In multiple RA loss-of-
function studies, the developmental malformations observed are milder than expected
suggesting the presence of a compensatory mechanism conferring robustness to perturbations
in RA signal (Blumberg et al., 1997; Hollemann et al., 1998; Janesick et al., 2014; Koide et al.,
2001; Shabtai et al., 2018; Sharpe and Goldstone, 1997). Paradoxical teratogenic outcomes
were observed in a number of RA manipulation studies, suggesting the activity of regulatory
feedback mechanisms (D’Aniello et al., 2013; D’Aniello and Waxman, 2015; Lee et al., 2012;
Rydeen et al., 2015). To further characterize the robustness of RA signaling during early
embryogenesis we employed clearly defined and transient RA manipulations that were
terminated during early gastrula, and the kinetics of recovery of the embryos was monitored
by RNAseq, qPCR and phenotypic analysis for several hours thereafter. These results
demonstrated a high robustness to physiological perturbations of the RA signal with relatively
small transcriptome-wide changes. Further transcriptomic analysis showed that components of
the RA metabolic and signaling network exhibited expression changes in a manner that is
5 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
dependent on the direction of RA manipulation, suggesting a mechanistic explanation for the
robustness observed. Multiple embryo clutches, i.e. biological repeats, were analyzed.
Comparative analysis of the biological repeats revealed differences between clutches in their
individual robustness response to enhanced or suppressed RA signaling. These results exposed
potential consequences of the underlying genetic differences between clutches as each one
originated from a single female, and most Xenopus laevis stocks in laboratories are outbred.
Our results suggest an asymmetric capacity for robust control of RA signal in the early embryo,
likely contributing to the human developmental defects that arise due to imbalance, most often,
a reduction) in Vitamin A levels during early development.
RESULTS
Physiological RA manipulation uncovers signaling robustness
To understand the regulatory role of RA signaling in early embryos, we routinely employ
two approaches to reduce the levels of this ligand (Fig. 1A). We either partially inhibit the
oxidation of retinaldehyde to RA with pharmacological RALDH inhibitors like 4-
diethylaminobenzaldehyde (DEAB) (Kot-Leibovich and Fainsod, 2009; Shabtai et al., 2016;
Shabtai and Fainsod, 2018), or we overexpress the RA hydroxylase, CYP26A1 (Hollemann et
al., 1998; Yelin et al., 2005), that renders this signal biologically inactive (Dobbs-McAuliffe et
al., 2004; Niederreither et al., 2002). Over a wide range of concentrations these treatments
result in clear, but unexpectedly mild, developmental defects for such a central regulatory
signaling pathway (compare Fig. 1B with Fig. 1C, D). An efficient approach to induce severe
phenotypes by RA signaling loss-of-function (LOF) is to combine knock-down treatments that
target the retinoid metabolism at different steps (Fig. 1A, E). These observations demonstrate
6 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
that RA metabolism and signaling exhibit strong robustness in the context of experimental
manipulation of RA levels.
To study the physiological robustness of retinoic acid signaling we focused on moderate
perturbations of this pathway within the range determined in X. laevis early embryos (about
100-150 nM all-trans RA) (Durston et al., 1989; Kraft et al., 1995; Kraft and Juchau, 1992;
Sive et al., 1990). To empirically determine the RA concentrations to use, we tested the effects
on gene expression of concentrations (1 nM - 1 µM) spanning the physiological range and
above (Fig. 2). Quantitative real-time PCR (qPCR) was performed on RNA samples of treated
embryos collected during early (st. 10.25) and late gastrula (st.12) (Nieuwkoop and Faber,
1967). We analyzed genes encoding enzymes involved in RA metabolism and known RA target
genes during early embryogenesis. Genes positively regulated by RA at both stages, hoxb1,
cyp26a1, and dhrs3 exhibited dose-dependent responses (Fig. 2A-C). The genes raldh2 and
rdh10, are down-regulated by most RA doses (Fig. 2D, F), while raldh3 (aldh1a3) is up-
regulated earlier on and is repressed later on by the treatment (Fig. 2E). Importantly, RA
concentrations as low as 10 nM had a significant effect on gene expression.
To specifically study the robustness of RA signaling in early Xenopus laevis embryos, we
established an experimental protocol where RA levels were transiently manipulated
pharmacologically within a physiological range. The perturbation was terminated by several
washings to remove the treatment and then the embryos were monitored post-treatment during
the recovery period. Embryos were treated with RA (10 nM) for 2 hours starting from late
blastula (st. 9.5) and washed during early gastrula (st. 10.25), about 2 hours after the treatment
was initiated. At different time points during the post-wash recovery period, samples were
collected to perform kinetic analysis of the changes in gene expression by qPCR (Fig. 3). To
monitor the changes in RA signaling levels by gene expression, two well-characterized RA-
regulated genes, hoxb1 and hoxb4, were studied. Expression analysis of both genes showed
7 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
that at the time of treatment washing (t0), both genes were upregulated when compared to
control sibling embryos (Fig. 3A). Two hours into the recovery, both RA target genes were
back to control expression levels when compared to stage-matched, untreated sibling embryos
from the same clutch. To begin understanding the robustness of RA signal and the dynamic
regulation of RA metabolism we studied the expression of dhrs3, which encodes an enzyme
that preferentially reduces retinaldehyde to retinol to attenuate RA biosynthesis (Feng et al.,
2010), cyp26a1, which targets RA for degradation, and raldh2 that produces RA from
retinaldehyde (Shabtai et al., 2016). As expected, both negative regulators of RA signaling,
dhrs3, and cyp26a1, are upregulated by the increased RA at t0 (Fig. 3B). Their return to normal
expression levels is only observed in the 4-hour samples in contrast to the hox genes that
returned to normal expression 2 hours earlier. As expected, raldh2 exhibited at t0 an RA-
promoted downregulation (Fig. 3B). It also took over 4 hours for this gene to return to normal
expression levels. These results indicate that the expression of genes encoding RA metabolic
enzymes, which themselves are regulated by RA (Fig. 2), is shifted to achieve normal RA levels
in the face of external perturbation to RA levels, even though these genes remain abnormally
expressed for a longer period.
A complementary study was performed by inhibiting RA biosynthesis taking advantage of
DEAB (Fig. 3C,D). All genes studied exhibited fluctuations during the recovery period. Also,
in this case, hoxd1 and hoxb1, key targets of RA signaling, reached almost normal levels at an
earlier stage than the genes encoding RA metabolic components (Fig. 3C). As expected, genes
encoding anabolic enzymes, e.g., raldh2, were upregulated, and catabolic components, e.g.,
cyp26a1, were downregulated (Fig. 3D).
The results of the kinetic analysis of the recovery from RA manipulation by qPCR
provided a novel and important support and insight into the robustness of RA signaling. We
observed that while RA downstream target genes reach normal expression levels relatively
8 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
quickly, expression of genes encoding RA metabolic enzymes is maintained at seemingly
abnormal levels for a longer period of time, such that their combined abnormal activity levels
likely result in almost normal RA signaling levels. Such expression changes suggest the
hypothesis that during early development RA signaling robustness is achieved, at least in part,
via feedback control of network components aimed at preventing a significant differential
system-wide gene expression response and the resulting teratogenic outcomes.
Non-teratogenic RA perturbations uncover signaling robustness at the transcriptomic scale
To gain a better understanding of the robustness of RA signaling and adaptation of the
metabolic network to disturbances, we performed a kinetic transcriptomic analysis. In an
attempt to optimize the kinetic study, parameters such as developmental stages and time
between samples were empirically tested. The experimental design involves treating embryos
for 2 hours from late blastula (st. 9.5) to early gastrula (st. 10.25) and collecting samples every
1.5 hours after terminating and washing the treatment (Fig. 4A). Embryos were treated with
RA (10 nM) or DEAB (50 µM). For each biological repeat, all treatments, controls, and time
points were collected from a single fertilization, from the eggs (clutch) of a single female. The
efficiency of the treatments and quality of the RNA samples was initially ascertained by qPCR
for changes in hox gene expression as a readout of the RA signal levels. Only those biological
repeats where both treatments exhibited the expected up-regulation (RA) or down-regulation
(DEAB) in hox gene expression were selected for RNA-seq.
We analyzed the time series transcriptomic data set for differential expressed genes
using a two-way ANOVA (t=0, 1.5, 3, 4.5 hours; treatments: RA, DEAB, control; n=6
biological replicates). Principal Component Analysis (PCA) of the gene expression variation
showed that all samples separated progressively along the first principal component (PC1)
9 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
which corresponds to the developmental stage (Fig. 4B). Surprisingly, RA manipulated
samples clustered with the control samples of the same developmental stage. Normal
transcriptomic changes as a result of progression through embryogenesis appear to be the
dominant variable distinguishing between the samples irrespective of treatment (Fig. 4B). The
second component (PC2) separated the sample groups at intermediate time points (1.5 and 3h)
from those at 0 and 4.5h time points, indicating a transient differential expression shift as the
next most dominant pattern in the data. The effect of the RA and DEAB treatments on the
transcriptome was not readily apparent in the next seven principal components. The the eighth
principal component (PC8) showed some separation of DEAB group from RA and Control
groups (Fig. 4C), whereas the tenth principal component (PC10) showed some separation of
the RA treatment group from Control and DEAB treatments (Fig. 4D). The top-ranked genes
along PC8 and PC10 showed distinct dynamic patterns across the treatments, whereas top-
ranked genes along PC1 and PC2 largely corresponded to in-common dynamic changes over
time (Fig. 4D; Supplemental Fig. S1). The patterns of both opposed RA manipulations, RA
and DEAB, closely resembled the control sample for the top-ranked genes along PC1 and PC2,
further supporting the normal developmental changes as the dominant pattern. The overall
magnitude of induced changes in gene expression in response to the RA or DEAB treatments
appears to be less than the normal transcriptome changes occurring during early developmental
stages, showing that RA signaling robustness possibly dampens the gene expression changes
otherwise induced by abnormal RA levels. Taken together, these results suggest that the RA
and DEAB treatments, applied at moderate physiologically relevant concentrations, do not alter
the transcriptome extensively in the whole embryo. These results are consistent with the
response of a robust system that functions to limit the gene expression changes in the majority
of the transcriptome.
10 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
Dynamic pattern analysis reveals the molecular genetic mechanisms underlying the RA
robustness response
In order to characterize the molecular transcriptomic mechanism for the robustness
response following the RA and DEAB treatments, we analyzed the clutch-averaged data using
an unbiased dynamic pattern analysis approach to categorize the gene expression profiles along
all possible discretized patterns (Kuttippurathu et al., 2016). Our analysis revealed a total of
4693 significantly differentially expressed genes with greater than 2-fold changes over time
(compared to t=0 within each treatment group; multiple testing corrected q-value < 0.05). For
each gene, a pattern based on the direction of up- or down-regulation above a 2-fold threshold
(three possibilities: up-regulation, down-regulation, or no change) at each of the three recovery
time points relative to early gastrula (t=0) was determined. This pattern analysis theoretically
should yield 3*3*3 = 27 possible discretized patterns. Such an exhaustive approach allows us
to enumerate the dominant as well as subtle patterns in the data and overcomes limitations of
conventional cluster analysis that is likely to miss or mask the smaller groups of genes with
distinct expression profiles over time. In our analysis, not all possible dynamic patterns had
representative genes. Out of the 27 possible dynamic patterns, only 10 patterns were exhibited
by genes in the transcriptome in at least one of the three experimental groups (RA, DEAB,
Control) (Fig. 5A). Of these, only 5 patterns were exhibited by a substantial number of genes
(>100) in at least one of the three experimental groups. These five patterns correspond to up-
or down-regulation at later time points (3h and 4.5h), and persistently induced or suppressed
expression at all three recovery time points, including progressive changes in gene expression
over time (Fig. 5A). At the 2-fold threshold, there were no genes that showed down-regulation
first and then up-regulation and vice versa. There were no significant differences in the gene
counts between RA or DEAB and control groups (two-tailed Z test, p > 0.05). Analysis of
statistically enriched pathways and processes (Gene Ontology analysis) in the control group
11 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
highlighted several functional annotations including development, cell cycle, morphogenesis,
RA biosynthesis, gastrulation, and others. (Fig. 5B; Supplemental Table S1). These expression
dynamics are consistent with the changes occurring in the early gastrula stage when the
endogenous RA system is activated. The patterns with relatively fewer genes (<100)
correspond to transient up- or down-regulation at 3h that is normalized by 4.5h, as well as
progressive down-regulation.
The results shown in Fig. 5 highlight two aspects of the effect of perturbations in the
RA signal. First, at the 2-fold threshold, there are no distinctive dynamic patterns of gene
expression changes that occur only in response to RA or DEAB treatment conditions. Second,
the number of genes per pattern is of a similar order of magnitude between the two treatments
and control. These results suggest that, at physiologically relevant concentrations employed
here, RA and DEAB treatments modulate the transcriptional and post-transcriptional regulators
at a scale resembling the changes in controls, albeit with differences in potential targets. For
instance, only a subset of genes that showed late down-regulation (first row in Fig. 5A) were
common across all the three groups (772 genes, Fig. 5C). Several genes showed similar late
down-regulation between the treatments or between one of the treatments and the Control
group (Fig. 5C). In addition, many genes showed a late down-regulation pattern only in one of
the treatment groups or in the Control (Fig. 5C). While the Venn diagram analysis (Fig. 5C)
revealed extensive overlap between the RA and DEAB treatments and Control samples within
a given differential regulation pattern, the overlap across patterns is not immediately clear from
this analysis.
In order to exhaustively compare the RA and DEAB treatment groups for their effect
on gene expression relative to the unperturbed temporal pattern observed in the control
samples, we adapted a recently developed unbiased approach named COMPACT for analyzing
time-series differential transcriptomic profiles across multiple experimental conditions
12 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
(Kuttippurathu et al., 2016). In this scheme, we started with the statistically significant genes
within each treatment group (two-way ANOVA; q<0.05), and constructed discrete patterns
based on differential gene expression between RA and control groups (81 theoretically possible
patterns, with up-, down- and no-regulation at t=0, 1.5, 3 and 4.5 h). In parallel, we constructed
similar discrete patterns for the DEAB versus control groups. Only the patterns with at least
one gene in either comparison were included in the subsequent analysis, yielding 32 distinct
patterns (Fig. 6A). Finally, we intersected the two distributions to create a matrix of
comparative patterns where each element corresponds to a distinct pair of patterns
corresponding to RA vs. control and DEAB vs. control (Fig. 6B). The discrete patterns were
based on a 1.3-fold average difference between RA (or DEAB) and control groups at each of
the four time points. The effect size threshold was chosen at a lower level than 2-fold as the
differential expression analysis revealed that the RA/DEAB perturbations were leading to a
smaller magnitude of changes at each time point, as compared to the larger changes occurring
normally over time, additional suggestive evidence of robustness.
Analysis of the COMPACT matrix (Fig. 6B, Extended Data 1) demonstrates that the
majority of the genes sensitive to the RA signal manipulation exhibited sensitivity to only one
direction of the RA perturbation. A total of 193 genes showed a response only to the addition
of exogenous RA relative to Control (middle row), whereas 224 genes were only responsive to
inhibition of RA production by DEAB (middle column). Of the genes that showed responses
to both RA and DEAB perturbations, 88 genes showed up-regulation or down-regulation
irrespective of RA or DEAB treatment (quadrants a and d in Fig. 6B). A set of 48 genes showed
opposite transcriptional outcomes in response to exogenous RA increase versus inhibition of
RA production (quadrants b and c in Fig. 6B). Interestingly, the set of genes that showed up-
regulation by exogenous RA addition and down-regulation by DEAB (quadrant b) contained
several genes involved in the biosynthesis and metabolism of RA (Fig. 6C).
13 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
We mapped the expression changes shown in Fig. 6B to the RA signaling and metabolic
network (Fig. 6D,E). Notably, genes encoding proteins involved in suppressing RA levels
(cyp26a1, dhrs3) were up-regulated and the genes encoding for proteins involved in RA
production (aldh1a2, aldh1a3, and rdh10) were down-regulated in response to transient
increase in RA. These genes showed opposite regulation in response to exogenous inhibition
of RAL to RA oxidation by DEAB (Fig. 6E). As an independent evaluation, we analyzed the
differential expression time series data using another unbiased approach, weighted gene
correlation network analysis (WGCNA) (Langfelder and Horvath, 2008). Manual examination
of WGCNA indicated that one of the correlated gene expression modules (highest correlation
with treatment) contained a similar gene set as shown in Fig. 6C, supporting the findings from
our COMPACT analysis (green module in WGCNA results, Supplemental Table S2; Extended
Data 2). Taken together, our results support a mechanism aimed at maintaining homeostasis of
the RA signaling levels to counter the effects of exogenous perturbations, nutritional and
environmental, and prevent teratogenic effects by a transcriptional feedback control system.
Clutch-wise heterogeneity demonstrates multiple alternative mechanisms to recover from RA
perturbations
The analysis above describes RA as a robust signaling pathway capable of responding to
environmental disturbances. Our results show that an important aspect of the regulation of
robustness involves changes in RA network components, and identified the main RA network
components responding during early gastrulation (Fig. 6). Interestingly, the PCA analysis
suggested clutch-specific heterogeneity in gene expression changes where all samples from the
same clutch tended to cluster together but the different clutches separated slightly from each
other (Fig. 4B,C; Supplemental Fig. S2). For these reasons, we analyzed the transcriptional
changes in the RA metabolic network within each clutch. We sought to determine whether the
14 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
clutch-to-clutch variation observed in the RNAseq data is evident even if we utilize a different
technical approach to measure the changes in gene expression in a separate set of manipulated
embryos. Hence, we generated an additional set of samples to study the dynamic expression
changes in RA metabolic network genes using high-throughput real-time PCR (HT-qPCR;
Fluidigm Biomark). This separate cohort of six additional clutches (biological repeats labeled
G-L) treated with RA or DEAB following the same experimental design (Fig. 4A) used for
RNAseq study of the original six clutches (A-F). We designed PCR primers for multiple
members of the RA metabolic network and a number of Hox genes as targets of the RA signal
(Supplemental Table S2). Our results show that the heterogeneity between clutches observed
in the HT-qPCR data resembles the clutch-to-clutch variation evident in the RNAseq data
(Supplemental Figs. S2 and S3).
We ordered the 12 biological repeats (clutches) according to the earliest time at which
hoxa1 returned to the baseline levels (Fig. 7). This ordering intermingled the clutches of the
RNAseq and HT-qPCR analysis. Our analysis identified high variability in the response to the
RA and DEAB treatments among the individual clutches. Clutches with significant up-
regulation of hoxa1 expression as a result of RA addition showed limited down-regulation in
expression in response to inhibition of RA biosynthesis (DEAB treatment; clutches C, L, J, K,
H, I in Fig. 7). The opposite response was also observed where DEAB induced a strong
response on hoxa1 expression while the RA treatment induced a mild to very weak response
(clutches E, D, F, G, A, B in Fig. 7). Examining the clutch-to-clutch variation in the extent of
changes in gene expression revealed a heterogenous correspondence between hoxa1 and RA
metabolic network genes. The suppressors of RA signaling, e.g., cyp26a1 and dhrs3, showed
significantly altered expression in clutches (C, L, J, K, H) with larger deviations in hoxa1
expression and a strong and extended response to the addition of RA. In contrast, the extent of
15 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
differential expression of the RA producers (e.g., aldh1a2 and rdh10) after RA manipulation
was relatively mild and did not fully align with the deviation in hoxa1 expression (Fig. 7). Such
a heterogeneous correlation between extent of feedback regulation and clutch-to-clutch
variation in hoxa1 expression was observed across the RA metabolic network (Supplementary
Fig. S4).
Asymmetry of robustness in response to increase versus decrease in RA signal
We sought to quantitatively rank the robustness of each clutch in an integrated manner
based on the extent of shift in multiple hox genes and the complementary feedback regulatory
response of the RA metabolic network genes. To this end, we pursued a trajectory-based
approach in which the temporal evolution of RA or DEAB treatment groups were compared to
their controls based on the regulatory outcome of target gene expression and RA network
feedback regulatory gene expression. In this analysis, the samples of all biological repeats
(clutches) at all time points were projected onto the first three principal components based on
the expression of hox genes (hoxa1, hoxa3, hoxb1, hoxb4, hoxd4) or of RA metabolic network
genes. A principal curve was fit to the projected data, representing the trajectory in which the
system evolves over time for all clutches combined. Each clutch was visualized separately
along the principal trajectory, allowing us to compare the temporal evolution of deviations
(distances) between treated samples and controls (select clutches shown in Fig. 8A and 8B; all
clutches included in Supplemental Fig. S5). As a multi-gene measure of robustness of each
clutch, the net absolute distance between the treated samples (RA or DEAB), and controls at
all time points, was computed along the principal curve in the hox gene or the RA network
trajectory maps. Our results revealed a wide range of clutch-to-clutch variability in robustness
as assessed by the integrated deviation from the control trajectory of multiple hox genes (Fig.
8C; Supplemental Fig. S5). The distance between the treatments and controls decreased over
16 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
time in nearly all the clutches with significant differences in the time taken to close the gap
(Fig. 8C). Clutches with low robustness to RA addition showed larger distances between RA-
treated samples and controls along the hox gene trajectory at any given time point (e.g., Fig.
8A, clutch C), compared to robust clutches with decreasing distances between RA-treated
samples and controls over time (e.g., Fig. 8A, clutch E). By contrast, DEAB treatment resulted
in an opposite pattern of robustness to that of RA addition (Fig. 8B). For example, clutch C
showed the widest deviation from control samples in the hox gene trajectory, whereas clutch E
showed least deviation (Fig. 8B top row and Fig. 8C). Clutch J exhibited intermediate
robustness responses to both the increase and the reduction of RA levels (Fig. 8A,B). The anti-
correlated pattern of robustness to increase versus decrease in RA levels suggests an
asymmetric tradeoff in the early embryo in the ability to counter changes in RA levels (Fig.
8E). The hox responses to decreased and increased RA levels revealed that all biological repeats
aligned along a diagonal and covered the whole range of responses (Fig. 8E). This result
suggested the establishment of an RA gradient among the clutches.
In contrast, similar analysis of the RA network component changes showed a scattered
pattern of the biological repeats with no clear trend. The feedback regulatory response based
on the RA metabolic network trajectory map (Fig. 8A and 8B bottom rows; Supplemental Fig.
S5) was variable across clutches, with a characteristic pattern of reduced deviation over time
in several clutches (Fig. 8D). There was no significant correlation (or anti-correlation) in the
net shift in feedback regulatory action between RA addition versus reduction (Fig. 8F).
Interestingly, the scattering distribution depended on the direction of the RA manipulation. The
range of clutch distribution along the x- axis in Fig. 8F suggests an immediate response to any
reduction in RA levels, and an upper limit to this response that points to a constrained ability
to mount a corrective feedback regulatory action for large reductions. In contrast, the clutch
distribution along y-axis suggests that the response to increased RA requires a minimal change
17 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
to be activated and it exhibits an upper threshold (Fig. 8F). These observations suggest a high
sensitivity to RA reduction but a lower sensitivity to slight RA increase.
An efficiency-efficacy matrix organizes the clutch variability in robustness and feedback
regulation
We examined if the asymmetric robustness to direction of RA change in correlates to the
extent of feedback regulatory action (Fig. 9). We formulated a “efficiency-efficacy matrix”
(Fig. 9A). The sections of this matrix delineate distinct possibilities based on all permutations
of the extent of robustness and the phenotypic outcome, i.e., the net shift in the hox gene
trajectory, versus extent of the feedback regulatory response, i.e., the net shift in RA metabolic
network trajectory. In this scheme, a high robustness (low net shift along the hox gene
trajectory) may be achieved by a mild or a strong response in the RA metabolic network gene
expression, yielding efficient and effective closed-loop feedback control scenarios,
respectively (lower quadrants in Fig. 8A). Similarly, the cases of low robustness (high net shift
along the hox gene trajectory) may be characterized by mild or strong gene expression changes
in the RA metabolic network components, corresponding to inefficient or ineffective closed-
loop feedback control, respectively (upper quadrants in Fig. 9A). In response to an increase in
RA levels, half of the clutches were distributed within the efficient and effective zone of the
robustness efficiency matrix (Fig. 9B; blue letters). Five clutches exhibited a substantial shift
in hox and RA network expression in response to increased RA, suggesting an inability to
correct the RA change, i.e. ineffective response. One clutch was located in an inefficient zone
as it exhibited a substantial shift in hox gene expression along with a limited extent of gene
expression shift in RA metabolic network components. By contrast, the distribution of clutches
in response to reduction in RA levels (DEAB treatment) distributed between the efficient and
inefficient quadrants, characterized by limited gene expression shift in the RA metabolic
18 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
network while yielding a wide range of robustness in hox gene expression (Fig. 9B; green
letters).
We examined if the clutches with similar levels of robustness and overall feedback
regulation use similar strategies for feedback regulation, i.e., whether they show similarity in
the genes that are differentially regulated in response to RA manipulation. Interestingly, we
found that the identity of differentially regulated genes in the RA metabolic network can vary
between clutches co-localized in the efficiency-efficacy matrix (Figs. 6E and 9C). For example,
in response to addition of RA, clutches A, E and F showed similar differential regulation of
dhrs3, aldh1a2, and cyp26a1, but diverged in the regulation of other genes such as, stra6,
sdr16c5, adhfe1, rdh13, aldh1a3, and cyp26c1 (Fig. 9C). Similarly, in response to DEAB
treatment, clutches A, B and D showed downregulation of dhrs3 and cyp26a1, and upregulation
of rbp1, suggesting feedback aimed at reducing activity of biochemical processes leading to
reduction in RA levels, and increasing import of retinol. However, there was much variability
across these clutches in the differential expression of aldh1a2 and aldh1a3, i.e., the feedback
control aimed at increasing the production of RA (Figs. 6E and 9D). Taken together, the results
described in Figs. 8 and 9 provide strong evidence that the early embryo is differentially
capable of responding to increase versus decrease in RA levels, with a distinct range of
feedback regulatory responses elicited between the two scenarios.
DISCUSSION
Robustness of the retinoic acid signaling pathway
Signaling pathway robustness is a central characteristic of all regulatory networks to ensure
signaling consistency and reliability. Variations in signaling levels can arise as a result of
changing environmental conditions or genetic polymorphisms that can affect the expression or
19 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
activity level of network components. We describe a systems biology approach to study the
robustness of RA metabolism and signaling based on transient manipulation of ligand levels.
Gastrula stage Xenopus laevis embryos have been reported to contain RA levels, around the
100 nM - 150 nM range (Durston et al., 1989; Kraft et al., 1995; Kraft and Juchau, 1992). In
contrast, due to the clear and severe developmental defects induced, Xenopus embryos, and
many other systems, are commonly treated with RA concentrations in the 1 µM to 10 µM range
to perturb this signaling pathway (Sive et al., 1990; Taira et al., 1994). To focus our robustness
study of RA signaling to the physiological range, we showed that by increasing the RA levels
by as little as about 10% (10 nM) we could consistently induce expression changes in RA-
regulated genes. Interestingly, embryo treatments with RA concentrations in the physiological
range exhibit very slight developmental malformations suggesting the induction of
compensatory mechanisms to control morphogen signaling and prevent abnormal gene
expression (Hollemann et al., 1998; Reijntjes et al., 2005; Sive et al., 1990). We obtained
similar results when we reduced the levels of RA by either blocking the biosynthesis (DEAB
treatment) or by targeting this ligand for degradation (CYP26A1 overexpression). These
observations are supported by multiple loss-of-function studies describing mild developmental
malformations induced by RA signaling reduction (Blumberg et al., 1997; Hollemann et al.,
1998; Janesick et al., 2014; Koide et al., 2001; Kot-Leibovich and Fainsod, 2009; Shabtai et
al., 2018; Sharpe and Goldstone, 1997; Shukrun et al., 2019). Therefore, increased or decreased
RA levels in the physiological range result in mild developmental defects, suggesting the
activation of compensatory mechanisms. We show that one approach to overcome the
robustness of RA signaling is to interfere with the metabolic/signaling network at two different
steps as in the case of DEAB together with CYP26A1. A possible explanation for this outcome
is that manipulating the RA network at more than one point hampers its ability to efficiently
elicit a feedback regulatory response.
20 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
RA signaling robustness was further analyzed taking advantage of transient RA
manipulations and the temporal kinetic, transcriptome-wide (RNAseq) analysis of the
restoration of normal gene expression patterns. The large expression differences observed
corresponded to normal transcriptome changes resulting from the progression through
embryogenesis. The close clustering of the RA-manipulated (increased or decreased RA) and
control samples at each time point further supports the robustness of this signaling pathway.
Alternatively, the treatments were inefficient, a possibility we could rule out by qPCR
screening of all RNA samples for Hox expression changes prior to sequencing and subsequent
computational analysis of the RNAseq data for individual gene expression shifts. Therefore,
the PCA analysis suggests the activation of an efficient robustness response of the RA network
to maintain normal, non-teratogenic, target gene expression levels during early gastrulation.
The RA robustness response emerges from autoregulatory changes in the RA metabolic
network
Our results clearly support the robustness of RA signaling when manipulated within the
physiological range. Then, how is this robustness achieved? What is the mechanism activated
to achieve RA robustness? Is the RA metabolic network modified to bring about this
robustness? The qPCR analysis already provided insights on the RA robustness mechanism.
RA target genes affected by the treatment revealed abnormal expression levels at the end of the
manipulation (t=0), and for some target genes (hox), we observed a speedy return to normal
expression levels already after 1.5 hours from the end of the treatment. Genes encoding
components of the RA metabolic network also exhibited abnormal expression levels at t=0.
Interestingly, their return to normal transcript levels was delayed beyond the time required for
the RA targets to reach normal expression. These observations suggest a scenario where the
RA metabolic network through an RA-dependent feedback regulatory mechanism is altered in
21 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
an attempt to restore normal signaling and target gene expression levels. Multiple reports have
described the regulation of select individual RA network component expression by RA. Most
of these studies deal with RA treatments resulting in the up-regulation of enzymes involved in
the suppression or reduction of RA signaling like CYP26A1, ADHFe1, and DHRS3, and down-
regulation of RA producers (anabolic enzymes) like RALDH2 and RDH10 (Chen et al., 2001;
Dobbs-McAuliffe et al., 2004; Fujii et al., 1997; Hollemann et al., 1998; Kam et al., 2013;
Sandell et al., 2012; Shabtai et al., 2017; Sonneveld et al., 1998; Strate et al., 2009).
To understand the full extent of kinetic response to transient RA manipulation, we pursued
a transcriptome-wide analysis of the recovery kinetics. Out of the 27 possible kinetic patterns
consisting of upregulation, downregulation or no change following treatment and washing,
only 10 were represented in any of the treatment or control samples, and only five patterns
were exhibited by a substantial number of genes (n>100). These observations suggest that
among the RA-regulated genes, direct or indirect, there is a limited number of possible
regulatory outcomes either in an attempt to normalize RA levels, i.e. robustness, or as target
genes. Although this classification is qualitative and there might be differences in intensity, it
suggests a limited repertoire of RA regulatory responses. To further understand the regulation
of RA targets we performed a comparative analysis of the patterns observed during increased
and decreased RA levels using our unbiased approach, COMPACT (Kuttippurathu et al.,
2016). This analysis revealed that most genes (75.40%) responding to RA manipulation and
recovery exhibited a response to either increased or decreased RA levels. Only 48 genes out of
553 (8.67%) exhibited reciprocal responses to both, increased and decreased RA levels.
Interestingly, many of these genes are involved in RA metabolism or signaling, in agreement
with their function in maintaining non-teratogenic RA levels.
Our results provide an integrative network-wide view that incorporates the concerted
feedback regulation of multiple pathway components to achieve normal signaling under
22 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
changing environmental conditions. The temporal discrepancy in the return to normalcy
between targets and network components and the network-wide changes in expression patterns
suggests a response taking place at multiple levels. The initial response to changes in RA levels
will most probably be mediated by the actual enzymes and factors already present in the cell.
In parallel, the same change in RA signaling will initiate a transcriptional response which in
the case of increased RA will involve not only the up-regulation and eventual enhancement of
the enzymatic RA suppressor activity, i.e. DHRS3, ADHFe1, and CYP26A1, but importantly,
a complementary down-regulation and activity reduction in RA producing enzymes (RALDH2
and RDH10). This transcriptional response will necessarily exhibit a slight delay until fully
functional as the network component genes will have to undergo transcription, followed by
translation and post-translational modifications, when required.
We observed that the expression of some RA network components exhibits oscillatory
behavior close to control expression levels, probably as a result of the fine-tuning of the RA
signal. This fine-tuning of the RA signal levels coupled to the inherent delay in the
transcriptional response could transiently result in the inversion of the overall signaling
direction, and an oscillatory transcriptional behavior. In a few instances such paradoxical
observations of RA signaling outcome following RA manipulation, i.e. overcompensation,
have been reported (D’Aniello et al., 2013; Lee et al., 2012; Rydeen et al., 2015). These
observations identify a very dynamic feedback regulatory network continuously fine-tuning
itself in response to perturbations.
Asymmetric response to increased and decreased RA levels
Our study provided new insights on the network responses to increased and decreased RA
levels. One important outcome from our study is the observation that the responses to increased
and decreased RA are not inverse but rather governed by different regulatory rules. These
observations were reached by performing a trajectory-based analysis in an attempt to rank the
23 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
RA robustness based on the response of hox genes as RA targets, and the extent of
transcriptomic changes of the components of the RA network. We derive the following primary
conclusions regarding this asymmetry of RA signal robustness:
First, the response to reduced RA signaling is activated after very slight reduction, while the
response to increased RA is only activated above a threshold. Also, the response to reduced
RA reaches a lower upper threshold than the response to increased RA. The alignment of the
hox responses along a diagonal suggested the establishment of an RA gradient among the
clutches. In contrast, the network responses showed no clear trend across clutches. Notably,
the network responses show different thresholds depending on the direction of the RA
manipulation. These observations suggest a high sensitivity to RA reduction but a lower
sensitivity to slight RA increase, i.e., an asymmetric capacity to mount responses dependent on
the direction of RA changes.
Second, very efficient responses to reduced RA are accompanied by weak responses to
increased RA, within the same clutch. The inverse situation was also observed suggesting an
asymmetric response to RA manipulations. The alignment of the robustness responses, hox
changes, fitted a diagonal with a negative slope. This negative slope suggested that while a
clutch might very efficiently deal with increased RA, i.e. high robustness, the same clutch
struggles to compensate for a reduction in RA, i.e. low robustness. The inverse situation, and
also more “balanced” clutches were observed. These observations suggest that not only
sensitivity response thresholds are involved, but also the robustness capacity to increased and
decreased RA are interconnected in an inverse fashion. These observations led us to formulate
an efficiency-efficacy matrix to link the robustness response of the RA network to the outcome
of their activity as detected by the hox response.
Third, different clutches with similar robustness levels to increased or reduced RA levels
can mount robustness responses by incorporating different components of the RA network. The
24 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
transcriptomic and HT-qPCR analyses allowed us to perform a detailed determination of the
network components comprising the robustness response, i.e. the mechanism of the robustness.
Clutches in proximity in the trajectory-based analysis or in the efficiency-efficacy matrix
mount responses by changing different sets of genes. The RA network includes multiple
components, i.e. enzymes, whose biochemical activity overlaps or is very similar. The RA
producing genes, aldh1a1, aldh1a2, and aldh1a3, also known as raldh1, raldh2, and raldh3
respectively are one example. All three enzymes oxidize retinaldehyde to produce RA
(Cunningham and Duester, 2015; Kedishvili, 2016, 2013; Shabtai et al., 2016). Some of the
differences between them apparently include enzymatic efficiencies and their expression
patterns including location, timing, and intensity (Blentic et al., 2003; Chen et al., 2001; Lupo
et al., 2005; Romand et al., 2004; Shabtai et al., 2018). A similar situation can be argued for
enzymes with RA degrading or biosynthetic suppressing activity like CYP26A1, B1, and C1,
or DHRS3, ADHFe1, RDH13 and others (Belyaeva et al., 2017, 2008; Hollemann et al., 1998;
Shabtai et al., 2017; Sonneveld et al., 1998). Some of these enzymes can reduce or prevent the
production of retinaldehyde while others make the RA biologically inactive targeting it for
degradation. Irrespective of the biochemical mechanism, these enzymes function to prevent
excessive RA signaling. These observations suggest that in the RA network where redundant
enzymes and factors performing similar activities are common, a robustness response can be
established using different components to reach the same level of robustness.
Fourth, genetic polymorphisms probably explain in part the different responses between
clutches to increased or decreased RA levels, and the metabolic and regulatory composition of
such responses. In our experiments, we utilized commercially available Xenopus laevis
laboratory stocks which are outbred and exhibit extensive genetic variability (Savova et al.,
2017). Therefore, we propose that one of the main differences between clutches in their
response to RA manipulation is probably the result of genetic variability. Formally, technical
25 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
issues could somehow account for some of the clutch-to-clutch variability uncovered in the
PCA analysis. To further support the genetic basis of the RA response, we analyzed six
additional clutchesby high-throughput real-time PCR. This qPCR analysis showed that the
new clutches exhibit overlapping responses to RA manipulation to the clutches analyzed by
RNAseq. These observations further supported the involvement of genetic variability. We
mined the Savova et al. (2017) data identifying multiple polymorphisms in RA network
components (Supplemental Table S4). All these observations support the potential contribution
of genetic variability in the differential robustness, intensity and response composition,
observed between embryo clutches.
Retinoic acid signaling changes due to environmental changes and disease risk
RA levels are tightly regulated throughout life at multiple levels to prevent aberrant
signaling and as a consequence, abnormal gene expression as a result of diet and other
environmental insults (Blaner, 2019; Blaner et al., 2016; Ghyselinck and Duester, 2019;
Kedishvili, 2016; Coberly et al., 1996; Lie et al., 2019; Paganelli et al., 2010; Shabtai et al.,
2018). During early embryogenesis, when RA biosynthesis initiates and it is restricted to a
limited repertoire of metabolic enzymes, this regulation is particularly important and
susceptible to changes.
For several decades, retinoic acid signaling has been the focus of intensive study due to
the severe teratogenic effects of abnormal levels and its involvement in the regulation of
numerous embryonic processes, oncogenes, and other signaling pathways. While the exposure
of embryos or cells to RA induces dramatic developmental changes and malformations,
treatment of the same embryos with retinol, the RA precursor, requires much higher
concentrations (~100X) to induce similar defects (Durston et al., 1989). Several models could
26 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
explain the different teratogenic potential of the precursor (retinol) or its final product (RA). In
both instances, members of the CYP26 family of cytochrome P450 enzymes should partially
neutralize the RA added or produced (Dobbs-McAuliffe et al., 2004; Duester et al., 2003;
Hollemann et al., 1998; Sakai et al., 2001). The main difference then becomes the fact that
retinol has to go through RA biosynthesis while RA is already the final active signaling ligand
(Duester, 2008; Kedishvili, 2016; Parés et al., 2008). The oxidation of retinol to retinaldehyde
is a reversible reaction which can reduce substrate availability for the ALDH enzymes. On the
other hand, treatment with RA can only be countered through inactivation by CYP26 enzymes.
Therefore, the reduced teratogenic efficacy of the retinol treatment could be the result of
reduced conversion to RA as part of the feedback regulation of this network, i.e. robustness,
while RA treatment is very restricted in its robustness response.
We obtained evidence of the RA signaling robustness employing several experimental
approaches. In all instances, we observed a fast response to the RA manipulation such that at
the transcriptome level, the treated samples were not significantly different from controls. All
our experiments were performed by either partially inhibiting the endogenous levels of RA, or
by slightly increasing (about 10% increase) the physiological content of RA in the embryo.
Under these conditions, the RA robustness of the embryo efficiently regulates and normalizes
the transcriptome as a whole via feedback. Based on the comparative analysis of the 12 clutches
(genetic backgrounds), we can suggest that robustness efficiency will have a threshold beyond
which it will become ineffective in restoring normal RA signaling. Our clutch analysis suggests
that this threshold might be strongly dependent on genetic polymorphisms affecting enzymatic
activity or gene expression parameters. In support, a threshold or toxicological tipping point
for RA signaling was recently described in a cell-based model (Saili et al., 2019). The clutch
analysis also showed that the network response, i.e. the genes actually up- or down-regulated,
is also dependent on genetic variability like promoter polymorphisms. Then, the
27 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
developmental malformations arising from environmental insults on RA signaling largely
depend on genetic polymorphisms which will determine the efficiency and threshold of the
response and the actual network components comprising such a response.
MATERIALS AND METHODS
Embryo culture
Xenopus laevis frogs were purchased from Xenopus I or NASCO (Dexter, MI or Fort
Atkinson, WI). Experiments were performed after approval and under the supervision of the
Institutional Animal Care and Use Committee (IACUC) of the Hebrew University (Ethics
approval no. MD-17-15281-3). Embryos were obtained by in vitro fertilization, incubated in
0.1% MBSH and staged according to Nieuwkoop and Faber (Nieuwkoop and Faber, 1967).
Embryo Treatments
all-trans Retinoic acid (RA), Dimethyl sulfoxide (DMSO), and 4-
Diethylaminobenzaldehyde (DEAB), were purchased from Sigma-Aldrich (St. Louis,
Missouri). Stock solutions of RA, and DEAB, were prepared in DMSO. Two-hour treatments
of 10 nM RA, or 50 µM DEAB, were initiated during late blastula (st. 9.5) and terminated at
early gastrula (st. 10.25) by three changes of 0.1% MBSH and further incubation in fresh 0.1%
MBSH for the desired time.
Total RNA purification from embryos and cDNA preparation
For each sample, 5-10 staged embryos were collected and stored at -80°C. RNA
purification was performed using the Bio-Rad Aurum Total RNA Mini Kit (according to the
manufacturer's instructions). RNA samples were used for cDNA synthesis using the Bio-Rad
iScriptTM Reverse Transcription Supermix for RT-qPCR kit (according to the manufacturer's
instructions).
28 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
Expression analysis
Quantitative real-time RT-PCR (qPCR) was performed using the Bio-Rad CFX384
thermal cycler and the iTaq Universal SYBR Green Supermix (Bio-Rad). All samples were
processed in triplicate and analyzed as described previously(Livak and Schmittgen, 2001). All
experiments were repeated with six different embryo batches. qPCR primers used are listed in
(Supplemental Table S2).
RNASeq data analysis
Sequencing was performed at the Thomas Jefferson University Genomics Core using
Illumina HiSeq 4000. Reads were mapped to the genome using the Xenopus laevis 9.1 genome
with STAR alignment and a modified BLAST and FASTQC in the NGS pipeline (STAR
average mapped length: 142.34). Annotation of the mapped sequences using verse identified
31535 genes. Raw counts were further filtered for non-zero variance across all samples
resulting in 31440 scaffold IDs.
High throughput qPCR
cDNA samples were directly processed for reverse transcriptase reaction using
SuperScript VILO Master Mix (Thermo Fisher Scientific, Waltham, MA), followed by real-
time PCR for targeted amplification and detection using the Evagreen intercalated dye-based
approach to detect the PCR-amplified product.
Intron-spanning PCR primers were designed for every assay using Primer3 and BLAST
for 24 genes from Retinoic Acid metabolism and target pathway (Supplemental Table S2). The
standard BioMark protocol was used to preamplify cDNA samples for 22 cycles using TaqMan
PreAmp Master Mix as per the manufacturer’s protocol (Applied Biosystems, Foster City, CA,
USA). qPCR reactions were performed using 96.96 BioMark Dynamic Arrays (Fluidigm,
South San Francisco, CA, USA) enabling quantitative measurement of multiple mRNAs and
29 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
samples under identical reaction conditions. Each run consisted of 30 amplification cycles (15
s at 95°C, 5 s at 70°C, 60 s at 60°C). Ct values were calculated by the Real-Time PCR Analysis
Software (Fluidigm). Samples were run in triplicate for the 24 genes. The primers in the first
pre-amplification group selectively bind for the long isoforms of the genes. In the second pre-
amplification group, primers were selected for the short isoforms only. Third pre-amplification
group binds to all the 24 genes. In order to remove the technical variability caused due to long
and short isoform of the genes, Ct value for each gene in a sample was selected as the median
value of the three pre-amplification runs. Relative gene expression was determined by the ΔΔCt
method. Gapdh was used as a housekeeping gene for normalization of the data.
Data normalization and annotation
Data analysis on raw genes count was performed using the R statistical analysis tool
version 3.6.0 on a 64-bit Windows platform. For the RNA-seq data, the raw genes counts were
first converted into log2-transformed values using the “regularized log (rlog)” transformation
from the DESeq2 package, which minimizes differences between samples for rows with small
counts (Love et al., 2014). The gene expression data was then normalized across samples
against the experimental variation and batch effects using COMBAT method in R using a non-
parametric adjustment (Johnson et al., 2007). Following batch correction, the gene list was
filtered for a minimum expression threshold to remove genes with normalized count less than
5 across all 72 samples. The expression data for the remaining genes was normalized using
quantile normalization. RNA-Seq transcript/array probes IDs were transformed to Official
Gene Symbol using merged list from 3 sources: the Xenopus laevis scaffold-gene mapping,
DAVID Bioinformatics Resource 6.8 (Huang et al., 2009) or AnnotationData package
“org.Xl.eg.db” maintained by Bioconductor (Carlson, 2017). The original scaffold IDs were
retained along with the Official Gene Symbols for cross-reference purposes.
Differential gene expression analysis
30 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
The normalized data was analyzed using an Empirical Bayes Statistical (eBayes) model that
considered the following two variables and their interactions: (1) Treatment (Control, RA,
DEAB) and (2) Time post treatment-washout (t = 0, 1.5, 3, 4.5h). Differentially expressed
genes were identified based on statistically significant effects of Time, Treatment or an
interaction between these two factors. P-values were adjusted for multiple testing using
topTable from limma (Ritchie et al., 2015) package in R (q ≤ 0.05). The significance-filtered
differential gene expression data was used in an established Principal Component Analysis
(PCA) approach using the prcomp function implemented in R. The samples were annotated
based on a combination of treatment and time, yielding 12 distinct sample groups. For each of
the selected PC, expression of 100 top positively-loaded and 100 top negatively-loaded genes
was visualized using a heat map.
Dynamic pattern analysis and COMPACT analysis
First, for each time point for each of the treatment conditions, the gene expression data for
all six clutches (A,B,C,D,E,F) was averaged. Within treatment groups RA, DEAB, Control, the
gene expression data at time points t=1.5, 3, 4.5h was normalized by subtracting the
corresponding ‘t=0h’ group. This average differential gene expression data was then
discretized to three levels (+1, 0, −1) based on a fold-change threshold [±2 (up, no or down-
regulation)]. Within the three treatment groups, this discretization yielded a dynamic response
pattern vector for each gene, encoded by one of 27 (3 levels^3 time-points) possible ordered
sets. Counts of genes in each treatment group that follow each of the 27 * 27 (=729) possibilities
were compared. Functional enrichment analysis was performed for geneset in various dynamic
pattern vectors in the Control conditions, using functions enrichGO and simplify from the R
package “clusterProfiler” (Yu et al., 2012).
COMPACT analysis of RA and DEAB after normalizing to Control
31 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
First, for each time point for each of the treatment conditions, the gene expression data for
all six clutches (A,B,C,D,E,F) was averaged. Within treatment groups RA and DEAB, the gene
expression data were normalized to Control at each time point by subtracting the expression of
the Control group at the corresponding time point. This Control-normalized differential gene
expression data for both treatment groups was then discretized to three levels (+1, 0, −1) based
on a fold-change threshold [±log2(1.3) (up, no or down- regulation)]. This discretization
yielded a dynamic response pattern vector for each gene, encoded by one of 81 (3 levels^4
time-points) possible ordered sets. Subsequently, RA and DEAB groups were compared to
count the number of genes corresponding to each of the 81 * 81 (=6561) possibilities; to create
a 81 × 81 matrix representing the comparative dynamic response pattern counts (COMPACT)
(Kuttippurathu et al., 2016). For a given COMPACT matrix of comparative conditions RA(vs.
Control) and DEAB (vs. Control), the element at the ith row and jth column of the matrix
contains the number of genes that show an ‘i’th pattern in DEAB and ‘j’th pattern in RA. For
a coarse-grained version of the detailed 81x81 COMPACT, pattern-vector counts for each
treatment group were further aggregated based on the first time-point, yielding 9 groups of
pattern vectors per treatment group. The pair of treatment group (RA and DEAB) was then
compared to count the number of genes corresponding to each of the 9 * 9 (=81) possibilities.
RA network map and visualization
A schematic representation for the position and functioning of the genes involved in the RA
biosynthesis, metabolism, translocation and transcription was formulated from the literature.
For each time point for each treatment, expression value for each gene was mapped to the
corresponding label in the schematics using a color scale.
Gene correlation and clustering analysis - Gene expression data was analysed for both with
and without Control-normalization using Weighted Gene Coexpression Network Analysis
(WGCNA) (Langfelder and Horvath, 2008) to identify modules of genes with highly correlated
32 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
differential expression. We used a soft threshold value of 9 (8 for Control normalized data) to
identify the initial gene coexpression modules, followed by a dissimilarity threshold of 0.25 to
merge the initial modules into the final set of gene coexpression modules. Identified modules
were further correlated with the traits (batch, time, treatment).
Robustness score calculation and principal curve-based trajectory analysis
RNA-seq expression data of (clutches: A,B,C,D,E,F) and additional normalized qPCR
expression data (clutches: G,H,I,J,K,L) were independently genewise Z-score transformed.
Transformed data was then combined to result into 144 samples (12 clutches * 3 treatments *
4 timepoints). We further selected two subsets of genes: 1) “RA metabolism genes” [aldh1a2.L,
aldh1a3.L, crabp2.L, crabp2.S, cyp26a1.L, cyp26a1.S, cyp26c1.L, cyp26c1.S, dhrs3.L, rbp1.L,
rdh10.L, rdh10.S, rdh13.L, rdh14.L, sdr16c5.L, stra6.L]. This set represents the feedback
regulatory mechanism utilised in response to the RA levels perturbations. 2) “HOX genes”
[hoxa1.L, hoxa1.S, hoxa3.S, hoxb1.S, hoxb4.S, hoxd4.L] representing the phenotypic outcome
from the treatment. For each gene set, the PCA scores for the combined data was first calculated
using R function prcomp and then the scores from the first three principal components (PC1,
PC2 and PC3) are used to learn a 3-dimensional principal curve using the function
principal_curve from R package princurve. Two sets of points are specified for the ‘start’
parameter for this function, which determines the origin and direction of the curve: (1) The
centroid of the 0h-Control samples from all twelve clutches, and (2) centroid of all the
remaining 132 samples. For each geneset (RA metabolism genes and HOX genes), for each
clutch (A-L), for each treatment (RA or DEAB), a “net absolute expression shift” can be
calculated as the sum of distances along the principal curve, of the treatment samples from the
corresponding control samples:
+. ℎ$%#$(%) − +. ℎ$%()*%+)"(%) !. ℎ$%#$ = ' ()!( !" !" ) !" -(%(+. ℎ$%) %
33 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
#$(%) where, !. ℎ$%!" is the arc-distance from the beginning of the curve, for the “RA treatment samples” at time point “t”, for clutch “cl” for the principal curve learned for the geneset “Hox genes” (Returned as the parameter “lambda” from the principal curve function). The expression shift calculation allows ranking and sorting clutches along measures ‘HOX shift’ %+,-%.,*% %+,-%.,*% (!. ℎ$%!" ) and ‘RA Network shift’ (!. -01()$23!-!" ). A higher HOX shift for a clutch for a treatment indicates that the clutch is more robust to that particular treatment/perturbation. Furthermore, clutches were mapped onto the conceptual map across the spectrum of “HOX shift” and “RA Network shift” divided into hypothetical quadrants of Effective or Ineffective regulation, or Efficient or Inefficient regulation.
Availability of supporting data/additional files
The raw and normalized datasets for the RNAseq and HT-qPCR data are available online as
Gene Expression Omnibus datasets via SuperSeries GSE154408 containing RNAseq data:
GSE154399 and HT-qPCR data: GSE154407.
FUNDING
RV acknowledges financial support for the project from the National Institute of Biomedical Imaging and Bioengineering grant U01 EB023224 and from the Department of Pathology, Anatomy, and Cell Biology, Thomas Jefferson University. AF acknowledges financial support from the Israel Science Foundation (grant 668/17) and the Wolfson Family Chair in Genetics. RV and AF acknowledge the Pilot Funding grant from Thomas Jefferson University and The Hebrew University of Jerusalem collaborative research program.
ABBREVIATIONS USED
COMPACT: Comparative Matrix of Pattern Counts
DEAB: 4-diethylaminobenzaldehyde
PCA: Principal Component Analysis
PCR: Polymerase Chain Reaction
34 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
RA: retinoic acid
WGCNA: Weighted Gene Coexpression Network Analysis
AUTHOR CONTRIBUTIONS
R.V. and A.F. conceived and supervised the study and designed the experiments and analysis
methodology. L.B-K., M.G., K.K., and A.F. performed embryo experiments and real-time PCR
assessment and developed the figures. A.B. performed the initial analysis of the RNAseq data.
S.A. conducted the high-throughput PCR validation of RNAseq results. M.P. conducted the
analysis of transcriptomics and HT-qPCR data and performed the network and trajectory
analyses and developed the figures. M.P., R.V., and A.F. interpreted the results and drafted the
manuscript.
ETHICS DECLARATIONS
The authors declare no competing interests.
35 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
BIBLIOGRAPHY
Adams, M.K., Belyaeva, O.V., Wu, L., Kedishvili, N.Y., 2014. The retinaldehyde reductase activity of DHRS3 is reciprocally activated by retinol dehydrogenase 10 to control retinoid homeostasis. J. Biol. Chem. 289, 14868–14880. doi:10.1074/jbc.M114.552257 Begemann, G., Schilling, T.F., Rauch, G.J., Geisler, R., Ingham, P.W., 2001. The zebrafish neckless mutation reveals a requirement for raldh2 in mesodermal signals that pattern the hindbrain. Development 128, 3081–3094. Belyaeva, O.V., Adams, M.K., Wu, L., Kedishvili, N.Y., 2017. The antagonistically bifunctional retinoid oxidoreductase complex is required for maintenance of all-trans- retinoic acid homeostasis. J. Biol. Chem. 292, 5884–5897. doi:10.1074/jbc.M117.776914 Belyaeva, O.V., Korkina, O.V., Stetsenko, A.V., Kedishvili, N.Y., 2008. Human retinol dehydrogenase 13 (RDH13) is a mitochondrial short-chain dehydrogenase/reductase with a retinaldehyde reductase activity. FEBS J. 275, 138–147. doi:10.1111/j.1742- 4658.2007.06184.x Billings, S.E., Pierzchalski, K., Butler Tjaden, N.E., Pang, X.-Y., Trainor, P.A., Kane, M.A., Moise, A.R., 2013. The retinaldehyde reductase DHRS3 is essential for preventing the formation of excess retinoic acid during embryonic development. FASEB J. 27, 4877– 4889. doi:10.1096/fj.13-227967 Blaner, W.S., 2019. Vitamin A signaling and homeostasis in obesity, diabetes, and metabolic disorders. Pharmacol. Ther. 197, 153–178. doi:10.1016/j.pharmthera.2019.01.006 Blaner, W.S., Li, Y., Brun, P.-J., Yuen, J.J., Lee, S.-A., Clugston, R.D., 2016. Vitamin A absorption, storage and mobilization. Subcell Biochem 81, 95–125. doi:10.1007/978-94- 024-0945-1_4 Blentic, A., Gale, E., Maden, M., 2003. Retinoic acid signalling centres in the avian embryo identified by sites of expression of synthesising and catabolising enzymes. Dev. Dyn. 227, 114–127. doi:10.1002/dvdy.10292 Blumberg, B., Bolado, J., Moreno, T.A., Kintner, C., Evans, R.M., Papalopulu, N., 1997. An essential role for retinoid signaling in anteroposterior neural patterning. Development 124, 373–379. Boehm, S.L., Lundahl, K.R., Caldwell, J., Gilliam, D.M., 1997. Ethanol teratogenesis in the C57BL/6J, DBA/2J and A/J inbred mouse strains. Alcohol 14, 389–395. Borel, P., Desmarchelier, C., 2017. Genetic Variations Associated with Vitamin A Status and Vitamin A Bioavailability. Nutrients 9. doi:10.3390/nu9030246 Campo-Paysaa, F., Marlétaz, F., Laudet, V., Schubert, M., 2008. Retinoic acid signaling in development: tissue-specific functions and evolutionary origins. Genesis 46, 640–656. doi:10.1002/dvg.20444 Carlson, M., 2017. org.Xl.eg.db. Bioconductor. doi:10.18129/b9.bioc.org.xl.eg.db Chen, Y., Pollet, N., Niehrs, C., Pieler, T., 2001. Increased XRALDH2 activity has a posteriorizing effect on the central nervous system of Xenopus embryos. Mech. Dev. 101, 91–103. doi:10.1016/S0925-4773(00)00558-X Clagett-Dame, M., DeLuca, H.F., 2002. The role of vitamin A in mammalian reproduction and embryonic development. Annu. Rev. Nutr. 22, 347–381.
36 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
doi:10.1146/annurev.nutr.22.010402.102745E Clagett-Dame, M., Knutson, D., 2011. Vitamin A in reproduction and development. Nutrients 3, 385–428. doi:10.3390/nu3040385 Coberly, S., Lammer, E., Alashari, M., 1996. Retinoic acid embryopathy: case report and review of literature. Pediatric Pathology & Laboratory Medicine 16, 823–836. doi:10.1080/15513819609169308 Collins, M.D., Mao, G.E., 1999. Teratology of retinoids. Annu. Rev. Pharmacol. Toxicol. 39, 399–430. doi:10.1146/annurev.pharmtox.39.1.399 Cui, J., Michaille, J.-J., Jiang, W., Zile, M.H., 2003. Retinoid receptors and vitamin A deficiency: differential patterns of transcription during early avian development and the rapid induction of RARs by retinoic acid. Dev. Biol. 260, 496–511. Cunningham, T.J., Duester, G., 2015. Mechanisms of retinoic acid signalling and its roles in organ and limb development. Nat. Rev. Mol. Cell Biol. 16, 110–123. doi:10.1038/nrm3932 D’Aniello, E., Rydeen, A.B., Anderson, J.L., Mandal, A., Waxman, J.S., 2013. Depletion of retinoic acid receptors initiates a novel positive feedback mechanism that promotes teratogenic increases in retinoic acid. PLoS Genet. 9, e1003689. doi:10.1371/journal.pgen.1003689 D’Aniello, E., Waxman, J.S., 2015. Input overload: Contributions of retinoic acid signaling feedback mechanisms to heart development and teratogenesis. Dev. Dyn. 244, 513–523. doi:10.1002/dvdy.24232 Dobbs-McAuliffe, B., Zhao, Q., Linney, E., 2004. Feedback mechanisms regulate retinoic acid production and degradation in the zebrafish embryo. Mech. Dev. 121, 339–350. doi:10.1016/j.mod.2004.02.008 Dubey, A., Rose, R.E., Jones, D.R., Saint-Jeannet, J.-P., 2018. Generating retinoic acid gradients by local degradation during craniofacial development: One cell’s cue is another cell’s poison. Genesis 56. doi:10.1002/dvg.23091 Duester, G., 1991. A hypothetical mechanism for fetal alcohol syndrome involving ethanol inhibition of retinoic acid synthesis at the alcohol dehydrogenase step. Alcohol. Clin. Exp. Res. 15, 568–572. doi:10.1111/j.1530-0277.1991.tb00562.x Duester, G., 2008. Retinoic acid synthesis and signaling during early organogenesis. Cell 134, 921–931. doi:10.1016/j.cell.2008.09.002 Duester, G., Mic, F.A., Molotkov, A., 2003. Cytosolic retinoid dehydrogenases govern ubiquitous metabolism of retinol to retinaldehyde followed by tissue-specific metabolism to retinoic acid. Chem Biol Interact 143–144, 201–210. Durston, A.J., Timmermans, J.P., Hage, W.J., Hendriks, H.F., de Vries, N.J., Heideveld, M., Nieuwkoop, P.D., 1989. Retinoic acid causes an anteroposterior transformation in the developing central nervous system. Nature 340, 140–144. doi:10.1038/340140a0 Eberhart, J.K., Parnell, S.E., 2016. The genetics of fetal alcohol spectrum disorders. Alcohol. Clin. Exp. Res. 40, 1154–1165. doi:10.1111/acer.13066 Eldar, A., Shilo, B.-Z., Barkai, N., 2004. Elucidating mechanisms underlying robustness of morphogen gradients. Curr. Opin. Genet. Dev. 14, 435–439. doi:10.1016/j.gde.2004.06.009 El Kares, R., Manolescu, D.C., Lakhal-Chaieb, L., Montpetit, A., Zhang, Z., Bhat, P.V.,
37 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
Goodyer, P., 2010. A human ALDH1A2 gene variant is associated with increased newborn kidney size and serum retinoic acid. Kidney Int. 78, 96–102. doi:10.1038/ki.2010.101 Evans, B.J., Kwon, T., 2015. Molecular polymorphism and divergence of duplicated genes in tetraploid african clawed frogs (xenopus). Cytogenet. Genome Res. 145, 243–252. doi:10.1159/000431108 Feng, L., Hernandez, R.E., Waxman, J.S., Yelon, D., Moens, C.B., 2010. Dhrs3a regulates retinoic acid biosynthesis through a feedback inhibition mechanism. Dev. Biol. 338, 1– 14. doi:10.1016/j.ydbio.2009.10.029 Fujii, H., Sato, T., Kaneko, S., Gotoh, O., Fujii-Kuriyama, Y., Osawa, K., Kato, S., Hamada, H., 1997. Metabolic inactivation of retinoic acid by a novel P450 differentially expressed in developing mouse embryos. EMBO J. 16, 4163–4173. doi:10.1093/emboj/16.14.4163 Ghyselinck, N.B., Duester, G., 2019. Retinoic acid signaling pathways. Development 146. doi:10.1242/dev.167502 Grandel, H., Lun, K., Rauch, G.-J., Rhinn, M., Piotrowski, T., Houart, C., Sordino, P., Küchler, A.M., Schulte-Merker, S., Geisler, R., Holder, N., Wilson, S.W., Brand, M., 2002. Retinoic acid signalling in the zebrafish embryo is necessary during pre- segmentation stages to pattern the anterior-posterior axis of the CNS and to induce a pectoral fin bud. Development 129, 2851–2865. Green, R.F., Stoler, J.M., 2007. Alcohol dehydrogenase 1B genotype and fetal alcohol syndrome: a HuGE minireview. Am. J. Obstet. Gynecol. 197, 12–25. doi:10.1016/j.ajog.2007.02.028 Hartomo, T.B., Van Huyen Pham, T., Yamamoto, N., Hirase, S., Hasegawa, D., Kosaka, Y., Matsuo, M., Hayakawa, A., Takeshima, Y., Iijima, K., Nishio, H., Nishimura, N., 2015. Involvement of aldehyde dehydrogenase 1A2 in the regulation of cancer stem cell properties in neuroblastoma. Int. J. Oncol. 46, 1089–1098. doi:10.3892/ijo.2014.2801 Hollemann, T., Chen, Y., Grunz, H., Pieler, T., 1998. Regionalized metabolic activity establishes boundaries of retinoic acid signalling. EMBO J. 17, 7361–7372. doi:10.1093/emboj/17.24.7361 Huang, D.W., Sherman, B.T., Lempicki, R.A., 2009. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat. Protoc. 4, 44–57. doi:10.1038/nprot.2008.211 Janesick, A., Abbey, R., Chung, C., Liu, S., Taketani, M., Blumberg, B., 2013. ERF and ETV3L are retinoic acid-inducible repressors required for primary neurogenesis. Development 140, 3095–3106. doi:10.1242/dev.093716 Janesick, A., Nguyen, T.T.L., Aisaki, K., Igarashi, K., Kitajima, S., Chandraratna, R.A.S., Kanno, J., Blumberg, B., 2014. Active repression by RARγ signaling is required for vertebrate axial elongation. Development 141, 2260–2270. doi:10.1242/dev.103705 Janesick, A., Wu, S.C., Blumberg, B., 2015. Retinoic acid signaling and neuronal differentiation. Cell. Mol. Life Sci. 72, 1559–1576. doi:10.1007/s00018-014-1815-9 Johnson, W.E., Li, C., Rabinovic, A., 2007. Adjusting batch effects in microarray expression data using empirical Bayes methods. Biostatistics 8, 118–127. doi:10.1093/biostatistics/kxj037
38 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
Jones, K.L., Smith, D.W., Ulleland, C.N., Streissguth, P., 1973. Pattern of malformation in offspring of chronic alcoholic mothers. Lancet 1, 1267–1271. doi:10.1016/s0140- 6736(73)91291-9 Kam, R.K.T., Shi, W., Chan, S.O., Chen, Y., Xu, G., Lau, C.B.-S., Fung, K.P., Chan, W.Y., Zhao, H., 2013. Dhrs3 protein attenuates retinoic acid signaling and is required for early embryonic patterning. J. Biol. Chem. 288, 31477–31487. doi:10.1074/jbc.M113.514984 Kedishvili, N.Y., 2013. Enzymology of retinoic acid biosynthesis and degradation. J. Lipid Res. 54, 1744–1760. doi:10.1194/jlr.R037028 Kedishvili, N.Y., 2016. Retinoic acid synthesis and degradation. Subcell Biochem 81, 127– 161. doi:10.1007/978-94-024-0945-1_5 Kessel, M., 1992. Respecification of vertebral identities by retinoic acid. Development 115, 487–501. Kim, H., Lapointe, J., Kaygusuz, G., Ong, D.E., Li, C., van de Rijn, M., Brooks, J.D., Pollack, J.R., 2005. The retinoic acid synthesis gene ALDH1a2 is a candidate tumor suppressor in prostate cancer. Cancer Res. 65, 8118–8124. doi:10.1158/0008- 5472.CAN-04-4562 Koide, T., Downes, M., Chandraratna, R.A., Blumberg, B., Umesono, K., 2001. Active repression of RAR signaling is required for head formation. Genes Dev. 15, 2111–2121. doi:10.1101/gad.908801 Kono, N., Arai, H., 2015. Intracellular transport of fat-soluble vitamins A and E. Traffic 16, 19–34. doi:10.1111/tra.12231 Koren, G., 2012. The first description of fetal alcohol syndrome by French pediatrician Paul Lemoine. J. Popul. Ther. Clin. Pharmacol. 19, e223. Kot-Leibovich, H., Fainsod, A., 2009. Ethanol induces embryonic malformations by competing for retinaldehyde dehydrogenase activity during vertebrate gastrulation. Dis. Model. Mech. 2, 295–305. doi:10.1242/dmm.001420 Kraft, J.C., Juchau, M.R., 1992. Correlations between conceptal concentrations of all-trans- retinoic acid and dysmorphogenesis after microinjections of all-trans-retinoic acid, 13- cis-retinoic acid, all-trans-retinoyl-beta-glucuronide, or retinol in cultured whole rat embryos. Drug Metab. Dispos. 20, 218–225. Kraft, J.C., Kimelman, D., Juchau, M.R., 1995. Xenopus laevis: a model system for the study of embryonic retinoid metabolism. I. Embryonic metabolism of 9-cis- and all-trans- retinals and retinols to their corresponding acid forms. Drug Metab. Dispos. 23, 72–82. Kuttippurathu, L., Juskeviciute, E., Dippold, R.P., Hoek, J.B., Vadigepalli, R., 2016. A novel comparative pattern analysis approach identifies chronic alcohol mediated dysregulation of transcriptomic dynamics during liver regeneration. BMC Genomics 17, 260. doi:10.1186/s12864-016-2492-x Langfelder, P., Horvath, S., 2008. WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics 9, 559. doi:10.1186/1471-2105-9-559 Lee, L.M.Y., Leung, C.-Y., Tang, W.W.C., Choi, H.-L., Leung, Y.-C., McCaffery, P.J., Wang, C.-C., Woolf, A.S., Shum, A.S.W., 2012. A paradoxical teratogenic mechanism for retinoic acid. Proc Natl Acad Sci USA 109, 13668–13673. doi:10.1073/pnas.1200872109 Leibovich, A., Edri, T., Klein, S.L., Moody, S.A., Fainsod, A., 2020. Natural size variation
39 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
among embryos leads to the corresponding scaling in gene expression. Dev. Biol. 462, 165–179. doi:10.1016/j.ydbio.2020.03.014 Lemoine, P., Harousseau, H., Borteyru, J.P., Menuet, J.C., 1968. Les enfants de parents alcooliques: Anomalies observees a propos de 127 cas. Ouest méd 8, 476–482. le Maire, A., Bourguet, W., 2014. Retinoic acid receptors: structural basis for coregulator interaction and exchange. Subcell Biochem 70, 37–54. doi:10.1007/978-94-017-9050- 5_3 Lie, K.K., Meier, S., Sørhus, E., Edvardsen, R.B., Karlsen, Ø., Olsvik, P.A., 2019. Offshore crude oil disrupts retinoid signaling and eye development in larval atlantic haddock. Front. Mar. Sci. 6. doi:10.3389/fmars.2019.00368 Livak, K.J., Schmittgen, T.D., 2001. Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods 25, 402–408. doi:10.1006/meth.2001.1262 Lohnes, D., Mark, M., Mendelsohn, C., Dollé, P., Decimo, D., LeMeur, M., Dierich, A., Gorry, P., Chambon, P., 1995. Developmental roles of the retinoic acid receptors. J. Steroid Biochem. Mol. Biol. 53, 475–486. Loucks, E., Carvan, M.J., 2004. Strain-dependent effects of developmental ethanol exposure in zebrafish. Neurotoxicol. Teratol. 26, 745–755. doi:10.1016/j.ntt.2004.06.017 Love, M.I., Huber, W., Anders, S., 2014. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550. doi:10.1186/s13059-014-0550-8 Lupo, G., Liu, Y., Qiu, R., Chandraratna, R.A.S., Barsacchi, G., He, R.-Q., Harris, W.A., 2005. Dorsoventral patterning of the Xenopus eye: a collaboration of Retinoid, Hedgehog and FGF receptor signaling. Development 132, 1737–1748. doi:10.1242/dev.01726 Marill, J., Idres, N., Capron, C.C., Nguyen, E., Chabot, G.G., 2003. Retinoic acid metabolism and mechanism of action: a review. Curr Drug Metab 4, 1–10. Mark, M., Ghyselinck, N.B., Chambon, P., 2009. Function of retinoic acid receptors during embryonic development. Nucl. Recept. Signal. 7, e002. doi:10.1621/nrs.07002 Mendelsohn, C., Lohnes, D., Décimo, D., Lufkin, T., LeMeur, M., Chambon, P., Mark, M., 1994. Function of the retinoic acid receptors (RARs) during development (II). Multiple abnormalities at various stages of organogenesis in RAR double mutants. Development 120, 2749–2771. Metzler, M.A., Sandell, L.L., 2016. Enzymatic metabolism of vitamin A in developing vertebrate embryos. Nutrients 8, pii: E812. doi:10.3390/nu8120812 Michiue, T., Yamamoto, T., Yasuoka, Y., Goto, T., Ikeda, T., Nagura, K., Nakayama, T., Taira, M., Kinoshita, T., 2017. High variability of expression profiles of homeologous genes for Wnt, Hh, Notch, and Hippo signaling pathways in Xenopus laevis. Dev. Biol. 426, 270–290. doi:10.1016/j.ydbio.2016.12.006 Napoli, J.L., 2016. Functions of Intracellular Retinoid Binding-Proteins. Subcell Biochem 81, 21–76. doi:10.1007/978-94-024-0945-1_2 Napoli, J.L., 2017. Cellular retinoid binding-proteins, CRBP, CRABP, FABP5: Effects on retinoid metabolism, function and related diseases. Pharmacol. Ther. 173, 19–33. doi:10.1016/j.pharmthera.2017.01.004
40 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
Niederreither, K., Abu-Abed, S., Schuhbaur, B., Petkovich, M., Chambon, P., Dollé, P., 2002. Genetic evidence that oxidative derivatives of retinoic acid are not involved in retinoid signaling during mouse development. Nat. Genet. 31, 84–88. doi:10.1038/ng876 Niederreither, K., McCaffery, P., Dräger, U.C., Chambon, P., Dollé, P., 1997. Restricted expression and retinoic acid-induced downregulation of the retinaldehyde dehydrogenase type 2 (RALDH-2) gene during mouse development. Mech. Dev. 62, 67–78. doi:10.1016/S0925-4773(96)00653-3 Niederreither, K., Subbarayan, V., Dollé, P., Chambon, P., 1999. Embryonic retinoic acid synthesis is essential for early mouse post-implantation development. Nat. Genet. 21, 444–448. doi:10.1038/7788 Nieuwkoop, P.D., Faber, J., 1967. Normal table of Xenopus laevis (Daudin): A systematical and chronological survey of the development from the fertilized egg till the end of metamorphosis. North-Holland Publishing Company, Amsterdam. Nijhout, H.F., Best, J.A., Reed, M.C., 2019. Systems biology of robustness and homeostatic mechanisms. Wiley Interdiscip. Rev. Syst. Biol. Med. 11, e1440. doi:10.1002/wsbm.1440 O’Byrne, S.M., Blaner, W.S., 2013. Retinol and retinyl esters: biochemistry and physiology. J. Lipid Res. 54, 1731–1743. doi:10.1194/jlr.R037648 Paganelli, A., Gnazzo, V., Acosta, H., López, S.L., Carrasco, A.E., 2010. Glyphosate-based herbicides produce teratogenic effects on vertebrates by impairing retinoic acid signaling. Chem. Res. Toxicol. 23, 1586–1595. doi:10.1021/tx1001749 Pangilinan, F., Molloy, A.M., Mills, J.L., Troendle, J.F., Parle-McDermott, A., Kay, D.M., Browne, M.L., McGrath, E.C., Abaan, H.O., Sutton, M., Kirke, P.N., Caggana, M., Shane, B., Scott, J.M., Brody, L.C., 2014. Replication and exploratory analysis of 24 candidate risk polymorphisms for neural tube defects. BMC Med. Genet. 15, 102. doi:10.1186/s12881-014-0102-9 Parés, X., Farrés, J., Kedishvili, N., Duester, G., 2008. Medium- and short-chain dehydrogenase/reductase gene and protein families : Medium-chain and short-chain dehydrogenases/reductases in retinoid metabolism. Cell. Mol. Life Sci. 65, 3936–3949. doi:10.1007/s00018-008-8591-3 Porté, S., Xavier Ruiz, F., Giménez, J., Molist, I., Alvarez, S., Domínguez, M., Alvarez, R., de Lera, A.R., Parés, X., Farrés, J., 2013. Aldo-keto reductases in retinoid metabolism: search for substrate specificity and inhibitor selectivity. Chem. Biol. Interact. 202, 186– 194. doi:10.1016/j.cbi.2012.11.014 Pullarkat, R.K., 1991. Hypothesis: prenatal ethanol-induced birth defects and retinoic acid. Alcohol. Clin. Exp. Res. 15, 565–567. doi:10.1111/j.1530-0277.1991.tb00561.x Reijntjes, S., Blentic, A., Gale, E., Maden, M., 2005. The control of morphogen signalling: regulation of the synthesis and catabolism of retinoic acid in the developing embryo. Dev. Biol. 285, 224–237. doi:10.1016/j.ydbio.2005.06.019 Ritchie, M.E., Phipson, B., Wu, D., Hu, Y., Law, C.W., Shi, W., Smyth, G.K., 2015. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 43, e47. doi:10.1093/nar/gkv007 Romand, R., Niederreither, K., Abu-Abed, S., Petkovich, M., Fraulob, V., Hashino, E., Dollé, P., 2004. Complementary expression patterns of retinoid acid-synthesizing and -
41 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
metabolizing enzymes in pre-natal mouse inner ear structures. Gene Expr. Patterns 4, 123–133. doi:10.1016/j.modgep.2003.09.006 Rydeen, A., Voisin, N., D’Aniello, E., Ravisankar, P., Devignes, C.-S., Waxman, J.S., 2015. Excessive feedback of Cyp26a1 promotes cell non-autonomous loss of retinoic acid signaling. Dev. Biol. 405, 47–55. doi:10.1016/j.ydbio.2015.06.008 Saili, K.S., Antonijevic, T., Zurlinden, T.J., Shah, I., Deisenroth, C., Knudsen, T.B., 2019. Molecular characterization of a toxicological tipping point during human stem cell differentiation. Reprod. Toxicol. 91, 1–13. doi:10.1016/j.reprotox.2019.10.001 Sakai, Y., Meno, C., Fujii, H., Nishino, J., Shiratori, H., Saijoh, Y., Rossant, J., Hamada, H., 2001. The retinoic acid-inactivating enzyme CYP26 is essential for establishing an uneven distribution of retinoic acid along the anterio-posterior axis within the mouse embryo. Genes Dev. 15, 213–225. doi:10.1101/gad.851501 Sandell, L.L., Lynn, M.L., Inman, K.E., McDowell, W., Trainor, P.A., 2012. RDH10 oxidation of Vitamin A is a critical control step in synthesis of retinoic acid during mouse embryogenesis. PLoS ONE 7, e30698. doi:10.1371/journal.pone.0030698 Savova, V., Pearl, E.J., Boke, E., Nag, A., Adzhubei, I., Horb, M.E., Peshkin, L., 2017. Transcriptomic insights into genetic diversity of protein-coding genes in X. laevis. Dev. Biol. 424, 181–188. doi:10.1016/j.ydbio.2017.02.019 Schilling, T.F., Sosnik, J., Nie, Q., 2016. Visualizing retinoic acid morphogen gradients. Methods Cell Biol. 133, 139–163. doi:10.1016/bs.mcb.2016.03.003 Schroeder, F., Petrescu, A.D., Huang, H., Atshaves, B.P., McIntosh, A.L., Martin, G.G., Hostetler, H.A., Vespa, A., Landrock, D., Landrock, K.K., Payne, H.R., Kier, A.B., 2008. Role of fatty acid binding proteins and long chain fatty acids in modulating nuclear receptors and gene transcription. Lipids 43, 1–17. doi:10.1007/s11745-007- 3111-z See, A.W.-M., Kaiser, M.E., White, J.C., Clagett-Dame, M., 2008. A nutritional model of late embryonic vitamin A deficiency produces defects in organogenesis at a high penetrance and reveals new roles for the vitamin in skeletal development. Dev. Biol. 316, 171–190. doi:10.1016/j.ydbio.2007.10.018 Shabtai, Y., Bendelac, L., Jubran, H., Hirschberg, J., Fainsod, A., 2018. Acetaldehyde inhibits retinoic acid biosynthesis to mediate alcohol teratogenicity. Sci. Rep. 8, 347. doi:10.1038/s41598-017-18719-7 Shabtai, Y., Fainsod, A., 2018. Competition between ethanol clearance and retinoic acid biosynthesis in the induction of fetal alcohol syndrome. Biochem. Cell Biol. 96, 148– 160. doi:10.1139/bcb-2017-0132 Shabtai, Y., Jubran, H., Nassar, T., Hirschberg, J., Fainsod, A., 2016. Kinetic characterization and regulation of the human retinaldehyde dehydrogenase 2 enzyme during production of retinoic acid. Biochem. J. 473, 1423–1431. doi:10.1042/BCJ20160101 Shabtai, Y., Shukrun, N., Fainsod, A., 2017. ADHFe1: a novel enzyme involved in retinoic acid-dependent Hox activation. Int. J. Dev. Biol. 61, 303–310. doi:10.1387/ijdb.160252af Sharpe, C.R., Goldstone, K., 1997. Retinoid receptors promote primary neurogenesis in Xenopus. Development 124, 515–523. Shenefelt, R.E., 1972. Gross congenital malformations. Animal model: treatment of various
42 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
species with a large dose of vitamin A at known stages in pregnancy. Am. J. Pathol. 66, 589–592. Shukrun, N., Shabtai, Y., Pillemer, G., Fainsod, A., 2019. Retinoic acid signaling reduction recapitulates the effects of alcohol on embryo size. Genesis 57, e23284. doi:10.1002/dvg.23284 Sive, H.L., Draper, B.W., Harland, R.M., Weintraub, H., 1990. Identification of a retinoic acid-sensitive period during primary axis formation in Xenopus laevis. Genes Dev. 4, 932–942. doi:10.1101/gad.4.6.932 Sonneveld, E., van den Brink, C.E., van der Leede, B.M., Schulkes, R.K., Petkovich, M., van der Burg, B., van der Saag, P.T., 1998. Human retinoic acid (RA) 4-hydroxylase (CYP26) is highly specific for all-trans-RA and can be induced through RA receptors in human breast and colon carcinoma cells. Cell Growth Differ. 9, 629–637. Strate, I., Min, T.H., Iliev, D., Pera, E.M., 2009. Retinol dehydrogenase 10 is a feedback regulator of retinoic acid signalling during axis formation and patterning of the central nervous system. Development 136, 461–472. doi:10.1242/dev.024901 Suzuki, A., Yoshida, H., van Heeringen, S.J., Takebayashi-Suzuki, K., Veenstra, G.J.C., Taira, M., 2017. Genomic organization and modulation of gene expression of the TGF-β and FGF pathways in the allotetraploid frog Xenopus laevis. Dev. Biol. 426, 336–359. doi:10.1016/j.ydbio.2016.09.016 Taira, M., Otani, H., Jamrich, M., Dawid, I.B., 1994. Expression of the LIM class homeobox gene Xlim-1 in pronephros and CNS cell lineages of Xenopus embryos is affected by retinoic acid and exogastrulation. Development 120, 1525–1536. Timoneda, J., Rodríguez-Fernández, L., Zaragozá, R., Marín, M.P., Cabezuelo, M.T., Torres, L., Viña, J.R., Barber, T., 2018. Vitamin A deficiency and the lung. Nutrients 10. doi:10.3390/nu10091132 Urbizu, A., Toma, C., Poca, M.A., Sahuquillo, J., Cuenca-León, E., Cormand, B., Macaya, A., 2013. Chiari malformation type I: a case-control association study of 58 developmental genes. PLoS ONE 8, e57241. doi:10.1371/journal.pone.0057241 Xavier-Neto, J., Sousa Costa, Â.M., Figueira, A.C.M., Caiaffa, C.D., Amaral, F.N. do, Peres, L.M.C., da Silva, B.S.P., Santos, L.N., Moise, A.R., Castillo, H.A., 2015. Signaling through retinoic acid receptors in cardiac development: Doing the right things at the right times. Biochim. Biophys. Acta 1849, 94–111. doi:10.1016/j.bbagrm.2014.08.003 Yelin, R., Schyr, R.B.-H., Kot, H., Zins, S., Frumkin, A., Pillemer, G., Fainsod, A., 2005. Ethanol exposure affects gene expression in the embryonic organizer and reduces retinoic acid levels. Dev. Biol. 279, 193–204. doi:10.1016/j.ydbio.2004.12.014 Yu, G., Wang, L.-G., Han, Y., He, Q.-Y., 2012. clusterProfiler: an R package for comparing biological themes among gene clusters. OMICS 16, 284–287. doi:10.1089/omi.2011.0118 Zhong, G., Ortiz, D., Zelter, A., Nath, A., Isoherranen, N., 2018. CYP26C1 Is a Hydroxylase of Multiple Active Retinoids and Interacts with Cellular Retinoic Acid Binding Proteins. Mol. Pharmacol. 93, 489–503. doi:10.1124/mol.117.111039
43 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
FIGURE LEGENDS:
Figure 1. Phenotypic robustness of the retinoic acid metabolic pathway. Retinoic acid levels were manipulated in Xenopus laevis embryos. Inhibition of RALDH activity with DEAB or CYP26A1 overexpression to render retinoic acid inactive were utilized to reduce levels of this signal. (A) Schematic diagram of the retinoic acid metabolic pathway and the steps affected. (B) Control embryo at st. 27. (C) Embryo injected with capped RNA (0.8 ng) encoding the CYP26A1 enzyme. A lineage tracer (ß-galactosidase RNA) was included to ensure a dorsal injection. (D) Embryo treated with DEAB (50 µM) from the midblastula transition until st. 27. (E) Embryo treated with DEAB and injected with cyp26a1 mRNA.
Figure 2. Retinoic acid manipulation in the physiological range. Embryos were treated with increasing concentrations of all-trans RA from 1 nM to 1 µM. Treatments were initiated at the midblastula transition (st. 8.5) and RNA samples were collected at early (st. 10.25) and late (st. 12) gastrula. The response of the RA metabolic and target genes was studied by qPCR. (A) hoxb1 (B) cyp26a1 (C) dhrs3 (D) raldh2 (aldh1a2) (E) raldh3 (aldh1a3) (F) rdh10. Statistical significance (Student’s t-test) was calculated compared to the expression level in the control group. *, p<0.05; **, p<0.01; ***, p<0.001; ns, not significant.
Figure 3. Kinetics of the recovery from RA manipulation. Embryos were transiently treated with either 10 nM RA (A,B), or 50 µM DEAB (C,D). Treatments were initiated during late blastula (st. 9.5), and by early gastrula (st. 10.25) the treatments were washed. RNA samples were collected at different time points during the recovery period. The response of RA target genes and genes encoding RA metabolic enzymes was studied by qPCR. Statistical significance (Student’s t-test) was calculated compared to the expression level at the end of the treatment
(t0). *, p<0.05; **, p<0.01; ***, p<0.001; ****, p<0.0001; ns, not significant.
Figure 4. Retinoic acid signaling recovery in a kinetic analysis. (A) Schematic description of the experimental design to study the robustness of RA signaling and the transcriptomic changes involved in the maintenance of signaling robustness. (B) Principal Component Analysis of all six biological replicates revealed sample separation based on developmental stages. The samples grouped largely based on developmental time points following treatment, with the RA or DEAB groups clustering with the control samples. The t=0, 1.5, 3 and 4.5h
44 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
samples group were all distinctly separated by Principal Component 1. Principal Component 2 separated samples at intermediate time points (t=1.5 and 3h). The samples do not appear to separate based on experimental manipulation of RA levels for either PC1 or PC2. (C) Principal Component 8 revealed limited sample separation based on DEAB treatment (at t=3h) and RA addition (at t=1.5h). (D) Principal Component 10 revealed limited sample separation based on RA manipulation (t=1.5 and 3h). (E) Heatmap of gene expression of the top-100 positive and top-100 negative loadings corresponding to PC1, PC2, PC8 and PC10. A subset of the genes are highlighted based on their relevance to early developmental processes.
Figure 5. Comparison of the differential gene expression patterns across the two treatment groups compared to controls. (A) Genes were grouped into 27 discretized expression patterns based on three possible outcomes of expression (yellow: up-regulation, blue: down-regulation, grey: no change) at each of the three recovery time points compared to
the t0 sample. The numbers below each graph exemplify the discretized pattern as a numeric vector. For example, the pattern marked blue at only the last time point (first row) represents late down-regulation above a 2-fold change threshold. The counts next to the patterns indicate the number of genes that show the corresponding expression pattern in each of the two treatment groups and the controls. The pattern counts are based on a differential regulation threshold of 2-fold change compared to t=0. Only ten out of 27 dynamic patterns were exhibited by at least one gene and are included in the figure. The majority of genes showed differential regulation at 3 and 4.5h. Of note, no genes showed a biphasic response of up- and down- regulation within 4.5h. The progressive up-/down-regulation of genes constitute a dominant feature of the early developmental process, which in many instances represents a restoration of normal expression levels. (B) The Gene Ontology biological processes statistically enriched in the control group are indicated alongside the pattern counts. Details of statistical analysis results are available in Supplemental Table S1. (C) Venn diagrams to compare the overlap between the two treatments and control, illustrated for three differential gene expression patterns. The number of genes that show a similar differential expression pattern across two treatments and control is indicated in the middle of the Venn diagrams. The number of genes that showed similar differential expression patterns only in one or two experimental groups are also indicated in the corresponding overlapping regions of the Venn diagrams. Black, control; Orange, DEAB; Blue, RA.
45 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
Figure 6. Comparative pattern analysis to uncover the genes showing distinct changes in expression dynamics in response to opposing perturbations of the RA signal. (A) Genes were grouped into discretized expression patterns based on three possible outcomes of expression (yellow: up-regulation, blue: down-regulation, grey: no change) comparing either treatment versus control at each of the four recovery time points. For example, the pattern marked blue at all the time points (first column ) represents persistent down-regulation. The two counts below the patterns indicate the number of genes that show the corresponding expression pattern in RA vs. control and DEAB vs. control. Only 32 out of 81 theoretically possible (four time points, 3*3*3*3=81) dynamic patterns were exhibited by at least one gene and are included in the figure. (B) COMPACT matrix comparing gene expression changes due to RA and DEAB relative to control. Of the potential 81x81 patterns, only the subset of 32 x 32 patterns with non-zero number of genes in either perturbation group are shown. The gene counts were grouped within related patterns based on the time of initial up- or down-regulation. Extended Data 1 contains a version of the COMPACT matrix shown with the gene identifiers corresponding to the counts. (C) Dynamic expression patterns of genes showing opposite changes in response to RA and DEAB treatments. This set of genes contains several components of the RA metabolism and gene regulatory network. (D) RA metabolic network showing synthesis and degradation of RA as well as a few known transcriptional regulatory targets. (E) Mapping the differential expression data onto the RA network shown in panel D to highlight the differences in regulation of biosynthesis versus degradation of RA signal between RA and DEAB groups.
Figure 7. Clutch-wise differential expression dynamics of select RA network genes and targets. The clutches are ordered left to right based on the earliest time at which hoxa1 expression returned to the baseline levels. The data was combined from RNAseq (clutches A- F) and HT-qPCR (clutches G-L).
Figure 8. Trajectory analysis to compare the extent of multi-gene shift away from and return to control levels and rank clutch-wise robustness. (A) 3-dimensional principal curves for two separate gene sets: “RA pathway genes” and “HOX genes”, showing projections of the sample points on the curve. Black star indicates the beginning of the curve for the distance measurement along the path. (B) Principal curves with the projection of Control and treatment samples only for the clutch ranked “most shifted” or “least shifted”. Ranking of
46 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
clutches is based on the net (absolute) normalized shift of treatment samples from the corresponding Control sample for each time point. (C,D) A normalized expression shift profile calculated from the principal curve as the arc distance between the treatment and the corresponding control. Clutches are rank-ordered from lowest to highest net expression shift for HOX genes in the RA group. (E,F) Net (absolute) normalized shift of Hox genes (panel E) or RA metabolism genes (panel F) for each clutch over all time points, calculated for both RA and DEAB treatment. Clutches A-F data from RNA-seq , clutches G-L data from HT-qPCR.
Figure 9. Differential robustness to direction of change in RA level is related to the effectiveness of feedback regulatory action. (A) A schematic diagram of robustness efficiency matrix delineating the scenarios arising from all combinations of robustness and feedback response. (B) Distribution of clutches across the zones of the robustness efficiency matrix for RA and DEAB groups relative to the control. The letters indicate the distinct clutches. (C,D) Mapping the differential expression data onto the RA network shown in Fig. 6 to highlight the heterogeneity of differential regulation of RA network components amongst clutches closely situated in the robustness efficiency matrix (panel B). Clutches A,E, and F showed similar robustness efficiency in response to RA treatment (panel C), whereas clutches A, B, and D showed similar robustness efficiency in response to DEAB (panel D).
SUPPLEMENTAL INFORMATION:
Supplemental Figure S1. Genes affected by RA manipulation. Heatmap of gene expression of the top-100 positive and top-100 negative loadings corresponding to PC1, PC2, PC8 and PC10.
Supplemental Figure S2. Differential robustness to RA manipulation between embryo clutches. Principal Component Analysis of RNAseq data revealed heterogeneity across developmental stages and treatments. (A) Distribution of samples along PC1 and PC2 axes. (B) Distribution of samples along PC1 and PC10 axes. The letters represent the six distinct biological repeats, clutches A-F.
Supplemental Figure S3. HT-qPCR analysis of RA network components. Principal Component Analysis of HT-qPCR data revealed heterogeneity across developmental stages
47 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
and treatments. (A) Distribution of samples along PC1 and PC2 axes. (B) Distribution of samples along PC1 and PC10 axes. (C) Heatmap of time series differential expression of RA network genes.
Supplemental Figure S4. Clutch-wise heterogeneity in the response of RA network genes following RA manipulation. Clutch-wise differential expression dynamics of RA network genes. The clutches are ordered left to right according to the time taken for recovery of hoxa1.L.
Supplemental Figure S5. Trajectory analysis to compare the extent of multi-gene shift away from and return to control levels and rank clutch-wise robustness. Clutches are ordered left to right according to the extent of net deviation between control and RA treatment along the HOX gene trajectory. High deviation corresponds to low robustness and vice versa. (A) RA and control groups are highlighted along the trajectories. (B) DEAB and control groups are shown.
Supplemental Table S1. List of Gene Ontology annotations with corresponding genes and statistical significance corresponding to Figure 5.
Supplemental Table S2. Weighted Gene Correlation Network Analysis (WGCNA) of the time series RNAseq data. The green module is shown here.
Supplemental Table S3. List of HT-qPCR primers corresponding to the Hox genes and RA metabolic network.
Supplemental Table S4. RA network component polymorphisms between Xenopus strains. Based on Savova et al. (2017).
Extended Data 1. The Excel file contains the complete list of gene identifiers represented in the COMPACT matrix corresponding to Figure 6B.
Extended Data 2: The source Excel file contains all the gene expression modules from WGCNA.
48 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
Parihar et al. Figure 4 A E RA DEAB Control t=0 1.5 3 4.5h t=0 1.5 3 4.5h t=0 1.5 3 4.5h
B genes PC1 t=0 1.5 3 4.5h 50 50 0 0 PC2 score genes 50 -50 - PC2 PC2 -150-150 -100-100 --5050 0 50 100 150150 C PC1 score
40 t=0 1.5 3 4.5h 30 30 skida1.S hoxa1.L
20 hoxa1.S hoxa2.S skida1.L 10 10 cyp26a1.L dhrs3.L 0 PC8 score 10 -10 - PC8 genes PC8 -20
30 mxra5.S -30 - -150-150 -100-100 --5050 0 50 100 150150 cyp26a1.S D PC1 score mxra5.S cyp26a1.S
40 t=0 1.5 3 4.5h 30 30 20 10 10 aldh1a2.L
0 cyp26a1.L dhrs3.L 10 PC10 genes PC10 -10 - PC10 score hoxa2.S skida1.L -20 hoxa1.L 30 -30 - -150-150 -100-100 --5050 0 50 100 150150 PC1 score -2 0 2 (Relative Expression)
RA DEAB Control bioRxiv preprint doi: https://doi.org/10.1101/2020.07.15.203794; this version posted October 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.
Parihar et al. Figure 5
A B C t = 0 1.5 3 4.5h RA DEAB Control Enriched Processes (Control Patterns) 1074 1115 1175 regulation of insulin receptor signaling pathway / autophagy / fatty acid metabolic process / specification of symmetry 0 0 0 -1
peptide metabolic process / cell fate specification / 867 945 952 mesenchyme development / tissue migration 0 0 0 1
animal organ development / cell adhesion / regulation of cell 1013 975 937 migration / positive regulation of cardiocyte differentiation 0 0 1 1 842 811 753 microtubule nucleation / positive regulation of signal transduction / endomembrane system organization 0 0 -1 -1 animal organ development / positive regulation of 218 227 193 macromolecule biosynthetic process / Wnt signaling pathway / 0 1 1 1 retinol metabolic process 86 91 53 negative regulation of cell population proliferation / gastrulation
0 -1 -1 -1 DEAB 58 38 37 regulation of cellular component organization / mitotic centrosome separation 0 0 -1 0
23 27 18 myoblast differentiation 0 0 1 0
Control 4 4 2 0 1 1 0 1 1 1 RA 0 -1 -1 0 Parihar et al. Figure 6
A Dynamic expression patterns Down regulation at time t Control expression level at time t Up regulation at time t t=0h 1.5 3 4.5 RA vs. Control 9 34 66 59 31 58 42 30 DEAB vs. Control 28 22 85 136 16 5 46 22
Dynamic expression shift in response to B addition of RA C RA vs. Control DEAB vs. Control RA vs. Control t=0 1.5 3 4.5h t=0 1.5 3 4.5h t=0h abhd14a.S cmtm5.L 1.5 cyp26a1.S 3 cyp26c1.L 4.5 cyp26c1.S dhrs3.L dmrt2.L fetub.S gbx2.1.L gcnt2.L hnf1b.S hoxa1.L hoxa1.S hoxa2.S hoxa3.S hoxa5.S a b hoxb1.S hoxd1.L junb.L LOC100489456.L LOC108697667 LOC108699981 LOC108701709 LOC108701808 LOC108703560
DEAB vs. Control vs. DEAB LOC108707893 LOC108708049 LOC108708243 LOC108709115 LOC108709667 inhibition of RALDH by DEAB DEAB by RALDH of inhibition LOC108715248 LOC108718689 meis3.L
Dynamic expression shift in response to to response in shift expression Dynamic neb.L neurog3.S c d nfib.L pax6.S prph.L sema3f.L tdgf1.2.S Similar shifts in Shifted dynamics -0.75 0 0.75 a d RA and DEAB only with RA Relative log2 Expression Opposite shifts in Shifted dynamics b c ( vs Control) RA and DEAB only with DEAB D E RA vs. Control
degrad degrad degrad degrad
ROL ROL RAL RA ROL ROL RAL RA ROL ROL RAL RA ROL ROL RAL RA
RA RA RA RA RAR/RXR RAR/RXR RAR/RXR RAR/RXR RARE RARE RARE RARE Cytoplasm Cytoplasm Cytoplasm Cytoplasm Extracellular Extracellular DEAB vs.Extracellular Control Extracellular
degrad degrad degrad degrad
ROL ROL RAL RA ROL ROL RAL RA ROL ROL RAL RA ROL ROL RAL RA
RA RA RA RA RAR/RXR RAR/RXR RAR/RXR RAR/RXR RARE RARE RARE RARE Cytoplasm Cytoplasm Cytoplasm Cytoplasm
Extracellular t=0 Extracellular 1.5 Extracellular 3 Extracellular 4.5h Parihar et al. Figure 7
C L J K H I E D F G A B 2 1 0 hoxa1.L -1
2 1 0
cyp26a1.L -1 2 1 score - dhrs3.L z 0
-1
1 0 -1 aldh1a2.L -2
2 1 0 -1
rdh10.S -2 RA vs. Control t=0 1.5 3 4.5h DEAB vs. Control Parihar et al. Figure 8
A Low Robustness to RA addition High C HOX genes (Phenotypic Outcome) C J E B E A K G J F I D H L C
0.2 0.0
distance -0.2 Normalized Outcome) HOX genes HOX (Phenotypic RA metabolism D (Feedback Regulatory Action)
0.2 0.0
distance -0.2 Normalized
RA metabolism RA RA vs. Control DEAB vs. Control t=0 3 1.5 4.5h (FeedbackRegulatory Action) B High Low E F Robustness to RA knock down Net absolute expression Net absolute expression shift C J E shift of HOX genes of RA metabolism genes 0.8 Outcome) HOX genes HOX (Phenotypic
0.4 RA vs. Control
Slope:R2: 0.42 -0.81 Slope:R2: 0.04 0.26
RA metabolism RA 2 2 0.0 RP-value:: 0.42, 0.0229p: 0.0229 RP-value:: 0.04, 0.5456p: 0.5456 0.0 0.4 0.8 0.0 0.4 0.8 (FeedbackRegulatory Action) RA DEAB Control 0 1.5 3 4.5h DEAB vs. Control Parihar et al. Figure 9 A B DEAB vs. Control RA vs. Control 0.8 max LOW robustness LOW robustness + + MILD feedback STRONG feedback response response = = Inefficient Ineffective regulation regulation 0.4 HIGH robustness HIGH robustness + + HOX Shift MILD feedback STRONG feedback HOX Shift response response = = Efficient Effective min regulation regulation 0.0 min max 0.0 0.4 0.8 RA Network Shift RA Network Shift C RA vs. Control At=0h Et=0h Ft=0h
degrad degrad degrad
CYP26A1 CYP26A1 CYP26A1 CYP26C1 CYP26C1 CYP26C1 SDR16C5 SDR16C5 SDR16C5 RDH10 RDH10 RDH10
ROL ROL RAL RA ROL ROL RAL RA ROL ROL RAL RA RBP1 ALDH1A2 RBP1 ALDH1A2 RBP1 ALDH1A2 STRA6 DHRS3 ALDH1A3 STRA6 DHRS3 ALDH1A3 STRA6 DHRS3 ALDH1A3 RDH13 RDH13 RDH13 RDH14 CRABP2 RDH14 CRABP2 RDH14 CRABP2 ADHFE1 ADHFE1 ADHFE1
RA RA RA RAR/RXR RAR/RXR RAR/RXR RARE RARE RARE
HOX genes HOX genes HOX genes
HOXA1 HOXB1 HOXD3 HOXA1 HOXB1 HOXD3 HOXA1 HOXB1 HOXD3 HOXA3 HOXB4 HOXD4 HOXA3 HOXB4 HOXD4 HOXA3 HOXB4 HOXD4
D Cytoplasm Cytoplasm DEAB vs. Control Cytoplasm Extracellular Extracellular Extracellular At=0h Bt=0h Dt=0h
degrad degrad degrad
CYP26A1 CYP26A1 CYP26A1 CYP26C1 CYP26C1 CYP26C1 SDR16C5 SDR16C5 SDR16C5 RDH10 RDH10 RDH10
ROL ROL RAL RA ROL ROL RAL RA ROL ROL RAL RA RBP1 ALDH1A2 RBP1 ALDH1A2 RBP1 ALDH1A2 STRA6 DHRS3 ALDH1A3 STRA6 DHRS3 ALDH1A3 STRA6 DHRS3 ALDH1A3 RDH13 RDH13 RDH13 RDH14 CRABP2 RDH14 CRABP2 RDH14 CRABP2 ADHFE1 ADHFE1 ADHFE1
RA RA RA RAR/RXR RAR/RXR RAR/RXR RARE RARE RARE
HOX genes HOX genes HOX genes
HOXA1 HOXB1 HOXD3 HOXA1 HOXB1 HOXD3 HOXA1 HOXB1 HOXD3 HOXA3 HOXB4 HOXD4 HOXA3 HOXB4 HOXD4 HOXA3 HOXB4 HOXD4
-0.75 0 0.75 Cytoplasm Cytoplasm Cytoplasm
Extracellular Extracellular Extracellular Relative log2 Expression ( vs Control) Parihar et al. Supplemental Figure S1
2 Clutch Clutch Time A Treatment B homez.L cdt1.L nsmaf.L zzef1.S fam46b.S 1 ythdc2.L trappc8.L ap4e1.S sorl1.L brwd3.L Xelaev18024710m.g dnajc14.L zyg11b.L glce.L LOC108719188 LOC100127794.L C ide.L rrm1.S gyg1.S eno1.L wdr1.L LOC108698233 xzar2 tmem169.L csnk1e.L LOC108698330 LOC108708693 atl2.L hist1h2aa.L gpi.S LOC108717688 LOC108708438 D cdk5r2.S haus3.L cntd2.L nde1.L plin2.S dand5.L 2 bix2.L mix1.S LOC108719961 lmnb3.S bzrap1.L Clutch LOC108695555 sdr42e1.L Clutch snx10.L stx11.S fnip1.L 0 E tmem57.S atp8a1.S LOC108705473 LOC108707181 rfc1.L wdr47.S LOC108704325 LOC108704537 MGC130623 fry.L per3.L LOC108709702 LOC108716896 A Time dapk1.L anapc7.L cdk5r2.L F LOC108705578 LOC108707803 pcmtd2.L plxnb3.S lrp8.L tdrd6.L LOC108709196 LOC108701404 LOC100137645 LOC108713875 mtmr10.L mapkapk2.S TreatmentLOC108712702 sin3b.L cbwd1.1.S ubox5.S B arvcf.L homez.L ccna1.L cdt1.L tpcn2.L nsmaf.L LOC108709750 klhdc10.L cep192.L zzef1.S fam46b.S brca2.L 1 ythdc2.L liph.S aldh2.L trappc8.L Time znf577.S ap4e1.S −1 sorl1.L wee1−a LOC108709286 pgk1.L brwd3.L LOC108714645 Xelaev18024710m.g LOC108713781 dnajc14.L zyg11b.L spdyc.L kiaa0355.S glce.L LOC108699883 LOC108719188 siva1.L LOC100127794.L C insrr.S ide.Larhgap5.L LOC108697788 rrm1.S ncam1.L gyg1.S crabp2.S eno1.L prpsap2.L 0 h wdr1.L rbm23.S LOC108698233 gli3.S xzar2 nhs.L meis2.L tmem169.L csnk1e.L pygm.L LOC108698330 qki.S LOC108708693 phip.L bmper.L cfh.L atl2.L hist1h2aa.L LOC108698469 gpi.S leng8.L ywhae.S LOC108717688 LOC108713018 LOC108708438 D cdk5r2.S sec61a1.S haus3.L sec61a1.L slc22a4.L 1.5 h slc30a8.L cntd2.L nxpe2.L nde1.L plin2.S LOC108711338 LOC108709441 dand5.L emp2.L bix2.L LOC108701328 emp2.S mix1.S LOC108719961LOC108711672 lmnb3.Sstag2.L c6orf62.L bzrap1.L LOC108718546 LOC108695555 −2 sdr42e1.Lsrsf11.L golm1.L snx10.L wwtr1.S stx11.S glipr2.L fam160b2.L fnip1.L 0 3 hE tmem57.S rpl7a.S atp8a1.S rpl7a.L LOC108705473 rps10.L LOC108716411 LOC108717090 LOC108707181 rfc1.L tmem39a.L wdr47.S caskin2.S fkbp1a.L LOC108704325 arfgap3.L LOC108704537 MGC130623 meox2.L LOC108696402 tubb.L fry.L bend3.S per3.L smug1.L LOC108709702 LOC108716896 arhgap36.L LOC108719919 dapk1.L 4.5 h ywhae.L anapc7.L rps20.L rpl29.L cdk5r2.L F LOC108705578LOC108715857 LOC108707803 rpl29.S rpl34 pcmtd2.L rpl17.S plxnb3.S lrp8.LLOC108713244 LOC108702869 tdrd6.L rps8.S LOC108709196 rps6.S rpl31.L LOC108701404 LOC100137645 rpl21.L LOC108713875 rps12.L mtmr10.L rpl28.S rps8.L rpl23a.L mapkapk2.S LOC108712702 rpl27a.L sin3b.L Xelaev18006566m.g pabpc4.L cbwd1.1.S LOC108700787 ubox5.S arvcf.L Xelaev18045492m.g Xelaev18030513m.g LOC108710952 ccna1.L rpl23.L tpcn2.L Treatment rps13.L LOC108709750 klhdc10.L slc35a2.L esf1.L cep192.L kif26a.L brca2.L kit.S liph.S aldh2.L wwtr1.L Time tdg.S znf577.S krt8.S −1 emc1.S wee1−a ipo5.L LOC108709286 pgk1.Lbtbd7.L kmt2e.S LOC108714645 phrf1.S LOC108713781 epas1.S Xelaev18035579m.g spdyc.L kiaa0355.S polr1a.L CTRL LOC108699883 rbm19.L siva1.L aim1.S LOC108702784 klf13.S insrr.S arhgap5.L LOC108709844 LOC108697788 LOC108708576 ncam1.L crabp2.S prpsap2.L 0 h rbm23.S gli3.S nhs.L meis2.L pygm.L DEAB qki.S phip.L bmper.L cfh.L golga5.S fam63b.S LOC108698469 lrrcc1.L leng8.L ywhae.S tmem87b.L LOC108707606 LOC108713018 LOC108695248 sec61a1.S LOC108711278 rbm24.S sec61a1.L slc22a4.Laptx.L 1.5 h slc30a8.L usp12.S naa10.S nxpe2.L LOC108699918 LOC108711338 Xelaev18005983m.g RA LOC108709441 tacc3.S emp2.L rsbn1l.L LOC108701328 MGC130644 enc1.S emp2.S LOC108711672 LOC108704916 stag2.L hivep1.S c6orf62.L LOC108716149 LOC108717044 LOC108718546 senp7.L −2 srsf11.L kcmf1.L golm1.L Xelaev18007772m.g zbtb8b.L wwtr1.S rcor1.L glipr2.L fam160b2.L LOC108716958 3 h tmem181.L sdf4.L rpl7a.S cog1.S rpl7a.L vsig10.L rps10.L LOC108716411 lhfpl2.S insr.L LOC108717090 LOC100158430 tmem39a.L zfpl1.L uso1.S caskin2.S fkbp1a.LLOC445863 PC10.loadings arfgap3.L uso1.L chd1.S meox2.L klhl12.S LOC108696402 tubb.Lgk.S gxylt1.L bend3.S znf516.S smug1.L mga.L znf292.S arhgap36.L LOC108719919 znf292.L 4.5 h ywhae.L bmp2.L rps20.L LOC108709771 lats1.S ift80.L rpl29.L LOC108715857 bbs9.L rpl29.S armc9.L zfand4.S rpl34 0.02 golga5.L rpl17.S LOC108713244 akap13.S znf654.L LOC108708073 LOC108702869 fam13a.L rps8.S sox7.L rps6.S rpl31.L trappc4.S vamp7.L rpl21.L pop4.L rps12.L adgrl2.S tmem259.L rpl28.S rps8.Lzc3h7a.L rpl23a.L rlf.L napg.L rpl27a.L LOC108706556 Xelaev18006566m.g pabpc4.Ldtl.S LOC443727 LOC108700787 elk4.L Xelaev18045492m.g kif7.L LOC108703181 Xelaev18030513m.g LOC108710952 ganab.L rpl23.L sprtn.S Treatment rps13.L cenpe.L csnk1d.S picalm.2.L slc35a2.L esf1.L exoc8.S kif26a.L LOC108697616 med10.S kit.S enox1.L wwtr1.L tdg.S lpcat1.S LOC108698716 evi5.L krt8.S saxo2.L emc1.S mtmr4.S ipo5.L btbd7.L LOC108704044 spin1.L kmt2e.S fam20b.L phrf1.S LOC108698103 LOC108715018 epas1.S Xelaev18035579m.gplk4.S polr1a.Lgigyf2.S CTRL clock.S rbm19.L LOC108709475 aim1.S LOC108702784rnf149.L tfcp2l1.S klf13.S kdm5b.S LOC108709844 adgrl2.L Xelaev18010420m.g LOC108708576 Xelaev18035048m.g −0.04 hes5.2.L dr1.S zhx3.L LOC108710165 hal.1.L wdr45b.S nck2.L LOC108706889 LOC108713989 t.L rnd1.L DEAB plekhn1.L Xelaev18030148m.g sap30.S LOC108698717 golga5.S LOC108697862 fam63b.S fam43a.S lrrcc1.L foxm1.S LOC108716190 tmem87b.L Xelaev18034297m.g LOC108707606 plcd1.L LOC108695248 LOC108708716 LOC108697277 LOC108711278 LOC108711818 rbm24.S PC8.loadings aptx.L Xelaev18038103m.g usp12.S Xelaev18003702m.g mig30.L sp1.S naa10.S LOC108718724 LOC108699918 Xelaev18005983m.g RA LOC108699808 dr1.L tacc3.S ythdf2.S rsbn1l.L fam122a.L MGC130644 fbxo28.L enc1.SLOC108716972 LOC108704916 ccdc51.S Xelaev18009254m.g hivep1.S ccnt2.S LOC108716149 LOC108717044ctdspl2.L senp7.Litgb6.L 0.04 mtmr12.L kcmf1.L LOC108699468 LOC100101336 Xelaev18007772m.g zbtb8b.L LOC108702529 rcor1.L ski.S LOC108716958 nacc1.S nacc1.L tmem181.L LOC108715362 sdf4.L Xelaev18047284m.g cog1.S LOC108717161 ccnj.L vsig10.L LOC108700819 lhfpl2.S insr.L Xelaev18045838m.g LOC100158430 clcf1.L e2f8.L nr1i2.L zfpl1.L frs2.S uso1.S LOC445863 c16orf52.L PC10.loadings ttbk1.S uso1.L arrb1 chd1.S LOC108712523 klhl12.S epha4.L gk.Sslc7a3.L gxylt1.L gata4.S cer1.S znf516.S paxbp1.S mga.L znf292.Srassf1.L znf292.Lcnrip1.L LOC108713288 bmp2.L LOC100049113 LOC108708592 LOC108709771 lats1.S LOC108703174 ift80.L LOC108709506 bbs9.L rad21.S sesn2.L MGC115496 armc9.L zfand4.S acsl3.L 0.02 golga5.L phldb3.L LOC108715821 akap13.S klf8.L znf654.L LOC108708073 Xelaev18043998m.g mfsd7.L tipin.L fam13a.L tmprss9.S sox7.L Xelaev18045468m.g trappc4.S vamp7.L rbmx.S sox15.L pop4.L lpar4.L adgrl2.S lhx5.S −0.04 fgfr4.S tmem259.L zc3h7a.LLOC108699714 LOC108719786 rlf.L LOC108719140 napg.L tmem79.S LOC108706556 dtl.Srbm5.S sebox.L LOC443727 sdc4.L elk4.L sp5l.S kif7.L LOC108703181 ganab.L sprtn.S cenpe.L csnk1d.S picalm.2.L exoc8.S LOC108697616 med10.S enox1.L lpcat1.S LOC108715854 LOC108698716 PC2.loadings LOC108705394 evi5.L frmd6.L saxo2.L frmd6.S gas6.L mtmr4.S atf6.S LOC108704044 spin1.L colgalt1.L fam20b.L LOC108717795 hmha1.S npdc1.1.S LOC108698103 LOC108709115 LOC108715018 plk4.S ccng2.S LOC108711059 gigyf2.S pvrl2.L clock.S entpd4.L LOC108709475 LOC108715914 rnf149.Lmdc13 tfcp2l1.S nktr.L 0.02 nfib.L kdm5b.S hoxa3.S adgrl2.L Xelaev18010420m.gXelaev18044735m.g Xelaev18035048m.g Xelaev18046453m.g −0.04 slc16a9.L hes5.2.L hyal2.L LOC108715248 dr1.S zhx3.L LOC108701808 LOC108710165 tmem79.L hal.1.L dusp6.L skida1.S wdr45b.S hoxa1.L nck2.L hoxa1.S LOC108706889 hoxa2.S Xelaev18044028m.g LOC108713989 azin2.S t.L rnd1.L Xelaev18030797m.g plekhn1.L Xelaev18030798m.g mxi1.L LOC443680 Xelaev18030148m.g cmtm5.L sap30.S LOC108698717 arid1b.S LOC108703560 LOC108697862 prph.L fam43a.S LOC108706603 foxm1.S atp6a1 LOC108716190skida1.L Xelaev18034297m.g cyp26a1.L dhrs3.L plcd1.L LOC108697667 LOC108708716 LOC108697277Xelaev18047114m.g LOC108711818 sox4.L PC8.loadings ppp1r3b.L Xelaev18038103m.g fbxo21.S sap30.L Xelaev18003702m.g mig30.L hic2.S sp1.S siah1.L LOC108718724 smad6.S LOC108715716 LOC108699808 med14.S dr1.L fzd5.S ythdf2.S rnf144b.L Xelaev18001106m.g fam122a.L coq2.L fbxo28.L LOC108716972 c5orf30.L ccdc51.S enc1.2.L LOC108707829 nodal.L Xelaev18009254m.g −0.02 sass6.S ccnt2.S ctdspl2.L tmem19.L eaf2.L itgb6.L 0.04 Xelaev18005471m.g mtmr12.L LOC108714468 LOC108699468 ccng1.S LOC100101336cxorf57.L LOC108702529 sgk223.S Xelaev18034358m.g ski.S slc30a1.S nacc1.S nacc1.Lfam43a.L LOC108715362 tmprss12.L pycrl.L Xelaev18047284m.g ndfip2.L klf5.S LOC108717161 ccnj.L slc22a23.L LOC108700819 e2f3.S Xelaev18045838m.g Xelaev18030809m.g arf5.L clcf1.L LOC108715815 e2f8.L LOC108713157 nr1i2.L casp6.L nenf.L frs2.S ehmt1.S c16orf52.L PC1.loadings ttbk1.S arid2.L arrb1 wdr5.S dusp1.L LOC108711453 LOC108712523 Xelaev18037947m.g epha4.L slc7a3.L LOC100158477.L ybx2.S gata4.S sgk223.L cer1.S Xelaev18007680m.g paxbp1.S LOC108712970 rassf1.Lstrn3.S cnrip1.L nudt4.S MGC99269LOC108713288 nt5m.L LOC100049113 LOC108708592gjc2.L LOC108703174 ulk1.L 0.005 ulk1.S LOC108709506 MGC115708 MGC114680 rad21.S sesn2.L LOC108698597 MGC115496 znrf1.L acsl3.L pex11b.L ndufb3.L LOC108719322 phldb3.L LOC108715821 lysmd3.L klf8.L med13.S Xelaev18018395m.g Xelaev18043998m.g arf6.2.L mfsd7.L tipin.L nup58.L Xelaev18022224m.g ddx5.S tmprss9.S ext1.L Xelaev18045468m.g klf2.L rbmx.S sox15.L LOC108695314 LOC108719186 lpar4.L bcl2l12.S lhx5.S −0.04 LOC108701475 tspo.L fgfr4.S LOC108699714suclg1.S LOC108719786 gtpbp2.S baiap2.LLOC108719140 LOC108697876 tmem79.S rbm5.Sfoxo3.L Xelaev18019676m.g sebox.L zmym4.L sdc4.L far1.L ubl7.S sp5l.S LOC108700103 csnk1g1.S bub3.S nom1.L ddx5.L tomm40.L tmem205.L ucp2.L ubtd1.S fam185a.L mturn.L st13.L idh2.S slc16a1.S pkn1.L stox1.S LOC108715854 ednra.S PC2.loadings LOC108705394 tmem150b.S −0.005 frmd6.L LOC108695976 LOC399435 frmd6.S col3a1.S gas6.L kiaa0895l.L atf6.S cass4.S rel.L colgalt1.L grin2b.L LOC108717795 hmha1.S morc3.2.L npdc1.1.S LOC108703005 LOC108711232 ipmk.L LOC108709115 MGC53199 ccng2.S LOC108711059 sh2d3c.L mxra5.S pvrl2.L Xelaev18025355m.g entpd4.L sh2d3c.S LOC108715914 mdc13 LOC108718251 mthfd2l.S 0.02 cyp26a1.S nktr.L gpr61.S nfib.L pdss1.S hoxa3.S Xelaev18044735m.gdusp5.L Xelaev18046453m.g LOC108711817 LOC108695508 slc16a9.L fam60a.L Xelaev18016707m.g hyal2.L LOC108715248 stox1.L LOC108701808 s1pr5.L tmem79.L hunk.L lzts1.L dusp6.L LOC495060 skida1.S hunk.S hoxa1.L lhx1.L dnah3.S hoxa1.S pdgfra.S hoxa2.S Xelaev18044028m.g efna1.L azin2.S pdgfra.L efnb2.S tmem150b.L Xelaev18030797m.g arl4a.S Xelaev18030798m.g mxi1.L irx3.L slc6a16.L LOC443680 mmp19.L cmtm5.L neurog3.S arid1b.S LOC108703560 phip.S prph.L LOC108706603 atp6a1 skida1.L cyp26a1.L dhrs3.L LOC108697667 Xelaev18047114m.g sox4.L ppp1r3b.L fbxo21.S sap30.LLOC108711232 sh2d3c.S hic2.S mxra5.S siah1.L Xelaev18025355m.g cyp26a1.S smad6.S LOC108715716 dusp5.L med14.S arg2.S fzd5.S LOC108711817 pdss1.S rnf144b.L Xelaev18001106m.g LOC108715914 Xelaev18034134m.g coq2.L zbtb47.L sh3rf1.S c5orf30.L LOC108695314 enc1.2.L LOC108707829 dusp11.L nodal.L LOC108698534 −0.02 LOC108718573 sass6.S dcp2.L ddx5.L tmem19.L eaf2.L iars.S kpna4.S Xelaev18005471m.g LOC108696549 LOC108714468 LOC398246 ccng1.S cxorf57.L ppp1r21.L rad1.L LOC100137634 sgk223.S Xelaev18034358m.g pmel.S slc30a1.L slc30a1.S fam43a.Lzdhhc16.S tmprss12.L usp25.L igsf9.L pycrl.L tsr3.L Xelaev18032363m.g ndfip2.L klf5.S farsb.S slc22a23.L c9orf69.L e2f3.S znf131.S LOC108711415 Xelaev18030809m.g arf5.L ccni.L irf6.2.L LOC108715815 oxr1.L osr1 LOC108713157 LOC108698560 casp6.L nenf.L Xelaev18038469m.g ehmt1.S LOC108707313 olfml2b.S PC1.loadings arid2.L dusp7.L LOC108704299 wdr5.S dusp1.L thsd7a.S phf13.L LOC108711453 LOC108716586 Xelaev18037947m.g b4galnt1.L LOC100158477.L ybx2.S neb.L slc9b2.S c12orf49.L sgk223.L Xelaev18007680m.g exo1.S LOC108699619 LOC108712970 strn3.Slpcat4.L lztfl1.Lnudt4.S actr3.L MGC99269 tk1.L fbxo5.S nt5m.L gjc2.L trappc2l.L ulk1.L cyb5b.L lmo7.L 0.005 ulk1.S cnot11.L MGC115708 MGC114680 LOC108706809 lin52.L LOC108698597 LOC108699601 LOC108709487 znrf1.L tmtc4.L pex11b.L ndufb3.L usp1−b LOC108719322 rnaseh2c.L Xelaev18032898m.g lysmd3.L tmem145.S LOC108719640 med13.S Xelaev18018395m.g vma21.L rhcg.L arf6.2.L LOC108702492 nup58.L LOC108717163 Xelaev18022224m.g ddx5.S cyb561.L LOC108715821 sall1.Sext1.L tmed8.L klf2.L LOC108712628 LOC108695314 LOC108719186LOC447061 LOC108705131 bcl2l12.S znf319.L LOC108701475 Xelaev18005310m.g chst2.S tspo.L suclg1.S march8.S gtpbp2.S fbxo5.L baiap2.L LOC108714253 Xelaev18024778m.g LOC108697876 LOC108702969 foxo3.L LOC108698217 Xelaev18019676m.g LOC108714191 LOC496402 zmym4.L dio3.L far1.L ubl7.S blvrb.L LOC108700103 rap2b.S march8.L LOC108695317 csnk1g1.S id3.S bub3.S nom1.L lpar2.L esr − 5.L ddx5.L aldh1a2.L tomm40.L LOC108701243 tmem205.L gatm.L ucp2.LMGC80198 LOC108715370 ubtd1.S Xelaev18023585m.g fam185a.L Xelaev18017910m.g mturn.L st13.LXelaev18017911m.g idh2.Srhou.L Xelaev18042589m.g slc16a1.S Xelaev18004616m.g npnt.S pkn1.L stox1.S LOC108701217 ednra.S LOC108714237 tmem150b.S LOC108715199 −0.005 LOC108702587 LOC108695976 arhgef40.L LOC399435 zdhhc6.S col3a1.S Xelaev18030091m.g gpa33.L kiaa0895l.L ocln.S cass4.S rel.L bsg.L grin2b.L kif3b.S LOC108706603 gpr61.S morc3.2.L atp6a1 LOC108703005 LOC108711232 cyp26a1.L dhrs3.L ipmk.L LOC108697667 MGC53199 tbc1d20.2.L sh2d3c.L ehmt1.S mxra5.SLOC108702899 Xelaev18025355m.g LOC108704345 Xelaev18034228m.g sh2d3c.S LOC108709662 LOC108718251 mthfd2l.Stfap2c.L LOC108702635 cyp26a1.S Xelaev18027819m.g gpr61.S ppil2.S wls.L pdss1.S dusp5.L traip.L LOC108711817 igsf9.S LOC108695508 usp36.S ttc23l.L fam60a.L Xelaev18000401m.g Xelaev18016707m.g LOC108704732 stox1.L dpf2.L myd88.S s1pr5.L dctn4.S hunk.L lzts1.L LOC108711590 LOC495060 smg7.S Xelaev18003190m.g LOC108699915 hunk.S aanat lhx1.L dnah3.S sh3glb2.L LOC108703372 pdgfra.S hyal2.L efna1.L tdgf1.2.S pdgfra.L LOC108715248 efnb2.SLOC108701808 tmem150b.L prph.L LOC108703560 arl4a.S hoxa2.S irx3.L slc6a16.Lskida1.L mmp19.Lhoxa1.L arid1b.S neurog3.S cmtm5.L Xelaev18002442m.g phip.S abhd14a.S Xelaev18034487m.g ranbp2.S LOC100158433.L unc5b.S pradc1.L axin1.L frmd6.L entpd4.L b4galt6.L LOC108719792 src.L pitpnm2.L Xelaev18003734m.g amd1.L vbp1.L LOC108711232 MGC53266 sh2d3c.S bin1.L mxra5.S cdk2.L alg5.L Xelaev18025355m.g cyp26a1.S LOC108707132 ptprt.L dusp5.L LOC108713843 znf414.L arg2.S LOC108698463 LOC108711817 pdss1.S Xelaev18031044m.g LOC108715914 rpia.L Xelaev18034134m.g zbtb47.L
PC10.loadings PC8.loadings PC2.loadings PC1.loadings sh3rf1.S LOC108695314 dusp11.L LOC108698534 LOC108718573 dcp2.L ddx5.L iars.S kpna4.S LOC108696549 LOC398246 ppp1r21.L rad1.L LOC100137634 pmel.S slc30a1.L zdhhc16.S usp25.L igsf9.L tsr3.L Xelaev18032363m.g farsb.S c9orf69.L znf131.S LOC108711415 ccni.L irf6.2.L oxr1.L osr1 LOC108698560 Xelaev18038469m.g LOC108707313 olfml2b.S dusp7.L LOC108704299 thsd7a.S phf13.L LOC108716586 b4galnt1.L neb.L slc9b2.S c12orf49.L exo1.S LOC108699619 lpcat4.L lztfl1.L actr3.L tk1.L fbxo5.S trappc2l.L cyb5b.L lmo7.L cnot11.L LOC108706809 lin52.L LOC108699601 LOC108709487 tmtc4.L usp1−b rnaseh2c.L Xelaev18032898m.g tmem145.S LOC108719640 vma21.L rhcg.L LOC108702492 LOC108717163 cyb561.L LOC108715821 sall1.S tmed8.L LOC108712628 LOC447061 LOC108705131 znf319.L Xelaev18005310m.g chst2.S march8.S fbxo5.L LOC108714253 Xelaev18024778m.g LOC108702969 LOC108698217 LOC108714191 LOC496402 dio3.L blvrb.L rap2b.S march8.L LOC108695317 id3.S lpar2.L esr−5.L aldh1a2.L LOC108701243 gatm.L MGC80198 LOC108715370 Xelaev18023585m.g Xelaev18017910m.g Xelaev18017911m.g rhou.L Xelaev18042589m.g Xelaev18004616m.g npnt.S LOC108701217 LOC108714237 LOC108715199 LOC108702587 arhgef40.L zdhhc6.S Xelaev18030091m.g gpa33.L ocln.S bsg.L kif3b.S LOC108706603 gpr61.S atp6a1 cyp26a1.L dhrs3.L LOC108697667 tbc1d20.2.L ehmt1.S LOC108702899 LOC108704345 Xelaev18034228m.g LOC108709662 tfap2c.L LOC108702635 Xelaev18027819m.g ppil2.S wls.L traip.L igsf9.S usp36.S ttc23l.L Xelaev18000401m.g LOC108704732 dpf2.L myd88.S dctn4.S LOC108711590 smg7.S Xelaev18003190m.g LOC108699915 aanat sh3glb2.L LOC108703372 hyal2.L tdgf1.2.S LOC108715248 LOC108701808 prph.L LOC108703560 hoxa2.S skida1.L hoxa1.L arid1b.S cmtm5.L Xelaev18002442m.g abhd14a.S Xelaev18034487m.g ranbp2.S LOC100158433.L unc5b.S pradc1.L axin1.L frmd6.L entpd4.L b4galt6.L LOC108719792 src.L pitpnm2.L Xelaev18003734m.g amd1.L vbp1.L MGC53266 bin1.L cdk2.L alg5.L LOC108707132 ptprt.L LOC108713843 znf414.L LOC108698463 Xelaev18031044m.g rpia.L PC10.loadings PC8.loadings PC2.loadings PC1.loadings Supplemental Supplemental al. et Parihar Figure S Figure B A 2 PC10 score PC2 score t=0h t=0h PC1 score PC1 PC1 score PC1 1.5 1.5 3 3 4.5 4.5 Parihar et al. Supplemental Figure S3
A C t=0 1.5 3 4.5h RA DEAB Control t=0 1.5 3 4.5h t=0 1.5 3 4.5h t=0 1.5 3 4.5h PC2 score
PC1 score B t=0 1.5 3 4.5h PC10 score
PC1 score -2 0 2 RA DEAB Control Relative Expression Parihar et al. Supplemental Figure S4 C L J K H I E D F G A B C L J K H I E D F G A B
aldh1a2.L rdh10.S
aldh1a3.L rdh13.L
crabp2.L rdh14.L
crabp2.S sdr16c5.L
cyp26a1.L stra6.L
cyp26a1.S hoxa1.L
cyp26c1.L hoxa1.S
cyp26c1.S hoxa3.S
dhrs3.L hoxb1.S
rbp1.L hoxb4.S
rdh10.L hoxd4.L RA vs. Control RA vs. Control t=0 1.5 3 4.5h DEAB vs. Control t=0 1.5 3 4.5h DEAB vs. Control Supplemental Supplemental al. et Parihar
RA metabolism HOX genes B RA metabolism HOX genes A (Feedback (Phenotypic (Feedback (Phenotypic Regulation) Outcome) Regulation) Outcome) Figure S Figure C 5 L H RA DEAB D High Low Robustness to RA knockdown Control Robustness to RA addition I 0 1.5 3 4.5h F High Low J G K A E B
Supplemental Table S1. List of Gene Ontology annotations with corresponding genes and statistical significance corresponding to Figure 5.
Control Total Pathway adjusted ID Description p value Pattern Genes Genes q value regulation of insulin receptor signaling GO:0046626 426 5 2.90E-06 3.99E-03 pathway GO:0006914 autophagy 426 14 7.65E-05 1.50E-02 GO:0006631 fatty acid metabolic process 426 11 1.45E-04 2.24E-02 0 0 0 -1 GO:0006096 glycolytic process 426 7 5.28E-04 3.80E-02 GO:0008354 germ cell migration 426 3 7.55E-04 4.02E-02 GO:0009799 specification of symmetry 426 5 1.04E-03 4.74E-02 GO:0010876 lipid localization 426 13 1.07E-03 4.74E-02 GO:0006518 peptide metabolic process 302 44 5.82E-13 2.24E-10 GO:0001708 cell fate specification 302 4 2.34E-03 2.11E-01 0 0 0 1 GO:0060485 mesenchyme development 302 7 3.95E-03 2.68E-01 GO:0090130 tissue migration 302 3 8.10E-03 4.42E-01 GO:0010631 epithelial cell migration 302 2 8.83E-03 4.42E-01 GO:0048513 animal organ development 320 35 2.34E-08 1.95E-05 GO:0007155 cell adhesion 320 26 3.21E-07 7.75E-05 0 0 1 1 GO:0030334 regulation of cell migration 320 5 6.53E-03 7.35E-02 GO:1905209 positive regulation of cardiocyte differentiation 320 2 9.88E-03 8.84E-02 GO:0007020 microtubule nucleation 250 3 3.01E-03 4.81E-01 GO:0009967 positive regulation of signal transduction 250 9 5.54E-03 4.81E-01 0 0 -1 -1 vesicle-mediated transport between endosomal GO:0098927 250 2 6.11E-03 4.81E-01 compartments GO:0010256 endomembrane system organization 250 7 6.13E-03 4.81E-01 GO:0048513 animal organ development 83 15 5.49E-07 1.25E-04 GO:0048732 gland development 83 5 1.42E-06 1.25E-04 positive regulation of macromolecule GO:0010557 83 11 3.29E-06 1.97E-04 biosynthetic process positive regulation of cellular biosynthetic GO:0031328 83 11 4.39E-06 1.97E-04 0 1 1 1 process GO:0016055 Wnt signaling pathway 83 8 2.71E-04 4.57E-03 cellular response to fibroblast growth factor GO:0044344 83 4 4.94E-04 6.21E-03 stimulus GO:0010817 regulation of hormone levels 83 4 1.92E-03 1.84E-02 GO:0042572 retinol metabolic process 83 2 2.43E-03 2.28E-02 negative regulation of cell population GO:0008285 15 3 7.57E-05 6.61E-03 0 -1 -1 -1 proliferation GO:0007369 gastrulation 15 2 5.21E-03 9.93E-02 GO:0007100 mitotic centrosome separation 14 2 1.87E-05 2.16E-03 GO:0140014 mitotic nuclear division 14 2 5.71E-03 8.00E-02 0 0 -1 0 GO:0051128 regulation of cellular component organization 14 3 8.47E-03 8.00E-02 GO:0031338 regulation of vesicle fusion 14 1 8.51E-03 8.00E-02 GO:0032418 lysosome localization 14 1 9.92E-03 8.00E-02 GO:0045663 positive regulation of myoblast differentiation 4 1 2.84E-03 1.71E-02 0 0 1 0 GO:0045445 myoblast differentiation 4 1 3.25E-03 1.71E-02
49
Supplemental Table S2. Weighted Gene Correlation Network Analysis (WGCNA) of the time series RNAseq data. The green module is shown here. The attached Extended Data 2 source Excel file contains all the gene expression modules from WGCNA.
Module Module RA vs. Control DEAB vs. Control Gene ID Gene Names Numbers Colors pattern pattern Xelaev18001038 LOC108703560 5 green 1_1_1_1 0_0_-1_-1 Xelaev18002824 sema3f.L 5 green 0_0_1_1 0_0_0_-1 Xelaev18006802 nfib.L 5 green 0_0_1_1 0_0_0_-1 Xelaev18007035 cmtm5.L 5 green 1_1_1_0 0_0_-1_-1 Xelaev18010181 tdgf1.2.S 5 green 0_1_1_0 0_0_0_-1 Xelaev18012533 LOC108708243 5 green 0_0_0_1 0_0_0_-1 Xelaev18013332 prph.L 5 green 1_1_1_1 0_0_0_-1 Xelaev18014991 hnf1b.S 5 green 0_1_1_0 0_0_-1_-1 Xelaev18015206 LOC108709115 5 green 0_0_1_0 0_0_0_-1 Xelaev18024402 pax6.S 5 green 0_0_1_0 0_0_0_-1 Xelaev18024482 LOC108715248 5 green 0_1_1_0 0_0_0_-1 Xelaev18025805 abhd14a.S 5 green 0_1_0_0 0_0_-1_0 Xelaev18030578 LOC108718689 5 green 0_1_1_0 0_0_-1_-1 Xelaev18030982 hoxa1.L 5 green 1_1_1_0 -1_-1_-1_-1 Xelaev18031638 gcnt2.L 5 green 0_0_0_1 0_0_-1_0 Xelaev18033067 hoxa1.S 5 green 1_1_1_0 -1_-1_-1_-1 Xelaev18033068 hoxa2.S 5 green 1_1_0_0 0_-1_-1_-1 Xelaev18033069 hoxa3.S 5 green 1_0_0_0 0_0_0_-1 Xelaev18033072 hoxa5.S 5 green 0_0_1_0 0_0_0_-1 Xelaev18034595 cyp26c1.L 5 green 0_1_1_1 0_-1_-1_-1 Xelaev18035730 dhrs3.L 5 green 1_1_1_0 -1_0_0_-1 Xelaev18036884 cyp26a1.S 5 green 1_1_1_0 -1_-1_0_0 Xelaev18036885 cyp26c1.S 5 green 1_1_1_1 0_0_-1_-1 Xelaev18037155 neurog3.S 5 green 0_1_0_0 0_0_0_-1 Xelaev18037556 LOC108697667 5 green 1_1_1_0 0_-1_0_0 Xelaev18038437 5 green 0_0_1_0 0_0_0_-1 Xelaev18039368 meis3.L 5 green 0_1_0_0 -1_-1_0_0 Xelaev18041724 LOC108699981 5 green 0_1_1_0 0_0_-1_-1 Xelaev18044027 LOC108701808 5 green 0_1_1_0 0_-1_0_0 Xelaev18044028 5 green 1_1_1_0 0_0_0_-1 Xelaev18044734 hoxd1.L 5 green 0_1_0_0 -1_0_0_0 Xelaev18045501 LOC100489456.L 5 green 0_0_1_0 0_0_-1_-1 Xelaev18045983 hoxb1.S 5 green 0_1_0_0 0_-1_-1_-1 Xelaev18047280 gbx2.1.L 5 green 0_1_1_1 0_-1_-1_-1
50
Supplemental Table S3. List of PCR primers corresponding to the hox genes and RA metabolic network for RT-qPCR and HT-qPCR analysis.
Gene Forward primer Reverse primer RT-qPCR dhrs3.L CAGGCGCAAGAAATCCTAAG CAAAGGCCACGTTACAGGAT aldh1a2.L ATGTTTGCCTGGAAGA GAGAGCAGTGAGCGGA cyp26a1.S CGATTCCTCAAGGTTTGGCTTCA ATTTAGCGGGTAGGTTGTCCACA hoxb1.L TTGCCCCAGTGCCAATGAC TCCCCCTCCAACAACAAACC hoxd1.S TTCTTGCGGGGATGTTTTAG CCGACTGGCATAAAGGAATG hoxb4.S CCAAGGATCTGTGCGTCAA GCAGGATGGAGGCGAACT gapdh GCTCCTCTCGCAAAGGTCAT GGGCCATCCACTGTCTTCTG HT-qPCR aldh1a2.L ATGTTTGCCTGGAAGATTGC GAGAGCAGTGAGCGGAGTCT cyp26c1.L AAACGGGTTCCTTTCTGTGT GCTTCGATTTACCCTACACTCTT hoxb1.S CCAACTTCACGACCAAACAA GTGGCTGCGATCTCTACTCTC hoxd4.L TTCCCTACCATCATTCCTTTC GAGTCATTATTTCCTGCTTTCTTT rbp1.L TGGAAATGCGAGTAGGAGATG GGGATGGTGGTTTATTGTGTG rdh13.L CAAGTGTCTACCTGGCTGTTG CCCAGAGTTTCCTTGCAGTT rdh14.L TGCCCGTACACAAAGACAGA GAGACCAAGGAGGTGGTGAG aldh1a3.L TAAAGCCCTGTCTGTTTCT CATACTCTCCAAGTTCCCTT crabp2.L AGCCACCCAAAGAAGACATAC CGATAAGAAACGAAAGCAGAAA crabp2.S TCAAAGGAGATGGACCCAAGA ATCAGCAGTCATGGTCAGGATAAG cyp26a1.L TCGAGGTTCGGCTTCATC CGGCACAATTCCACAACA cyp26a1.S CCGCTTTCTAACGCCACTT CACAATTCCACCACGAACAC cyp26c1.S AGCTCTGGTCCTTGAGATGG AGCCAATGCAGAGTTTCTCC dhrs3.L CAGGCGCAAGAAATCCTAAG CAAAGGCCACGTTACAGGAT hoxa1.L CCGCTCACTATATCCACCATTC TGGCAGGAGAACGACAAAC hoxa1.S AATTATGAGATGATGGAATGGTAAA TGACTGTAAACACCTAGTAAATGAGAG hoxa3.S AGCAGGGCAATGGAGTTT GGAGGGTCCACCAGAATTAG hoxb4.S CCAAGGATCTGTGCGTCAA GCAGGATGGAGGCGAACT hoxd3.S CTTCCCAGTCCACAATGAATAAA GCTCTTCTCCTCGCAGTTCTC rdh10.L CGTCTCTTTGCCCTGGAGTTT CACCATCTCCGCCGTCTC rdh10.S TTGCTTGGCCTGTAGAAGAGA TGCATGGCGAAATAGGAGTAG sdr16c5.L TTTGTGGTTCCTTCCCTCTC GTGCCATCAGTCTCCCTATACC stra6.L CCTGCTTACTTTCCTCCTGTTG GTGGGTGACATTAAAGATTGTAGAGA gapdh GCTCCTCTCGCAAAGGTCAT GGGCCATCCACTGTCTTCTG
51
Supplemental Table S4. RA network component polymorphisms between Xenopus strains1
Gene name Gene ID aa position aa1 aa2 aldh1a2 18020673m 240 P S cyp26c1 18034595m 388 F I rbp1 18027898m 26 T I rbp1 18027898m 149 E K rbp1 18027898m 27 H R rdh10 18033799m 133 H R rdh13 18026677m 14 C F sdr16c5 18032037m 35 A T stra6 18018036m 610 S N 1based on (Savova et al., 2017)
52