Abstract
Genotype plays a central role in the comprehensive management of
pheochromocytomas and paragangliomas, highlighting the critical need
for specific in vivo genetic models. Yet, animal models fall short of
fully recapitulating the biological complexity of these tumours. We
generated first-generation loss-of-function zebrafish models for sdhb,
a canonical PPGL-associated gene, using CRISPR/Cas9. Sdhb-CRISPants
exhibit increased heart rates, reduced swimming activity and premature
death. In whole fish extracts, normetanephrine (NM), metanephrine (MN),
and dopamine (DA) levels were about three times higher in sdhb
CRISPants than in control larvae. In the bathing medium, NM and MN were
also significantly elevated, along with 3-MT. Complementary metabolic
and transcriptomic profiling revealed that sdhb CRISPants exhibit a
clear signature of Complex II dysfunction and upregulation of genes
involved in the hypoxia response, angiogenesis, stress response, and
glycolysis. Our work validates the relevance of CRISPant zebrafish
models to study the pathogenicity of PPGL-causing genetic variants in
vivo.
Subject terms: Genomic engineering, Functional genomics, Mutagenesis
Introduction
Pheochromocytomas (PHEOs) and paragangliomas (PGLs) (PPGLs) are rare
tumors originating from tissues that arise from the embryonic neural
crest^[52]1. PHEOs are located in the adrenal medulla, and PGLs are
along the sympathetic and parasympathetic ganglia chains^[53]2. These
tumors can be non-secretory (normal plasma and/or urinary
catecholamines) or secretory (plasma and/or urinary catecholamine
levels above the upper limit of normal)^[54]3. PPGLs are the most
hereditary tumors in adults^[55]4 with predisposing germline pathogenic
variants found in around 40% of cases^[56]5. Over 20 susceptibility
genes are described to date, divided into three clusters. Tumors in the
pseudohypoxic cluster 1 are associated with a noradrenergic biochemical
phenotype and necessitate close monitoring due to the risk of
metastasis and recurrence^[57]6. Tumors in the kinase signaling-related
cluster 2 exhibit an adrenergic phenotype characterized by significant
rises in metanephrine levels (MN)^[58]7. Currently, the clinical
characteristics of patients with Wnt signaling-related cluster 3 tumors
is less known, but aggressive behavior appears probable^[59]8,[60]9.
With the recent advancements in genetics within the field of PPGLs,
genotype has become central to the medical and clinical management of
these tumors^[61]5,[62]8. Firstly, the genotype can impact prognosis
and follow-up. Pathogenic variants in specific susceptibility genes,
like SDHB (Succinate Dehydrogenase Complex Iron Sulfur Subunit B), are
linked to a poorer prognosis with an increased risk of metastatic or
recurrent disease^[63]10,[64]11. The recurrence rate for PPGLs is
~6.5–17.4%, varying according to the implicated gene^[65]12–[66]14.
Expert consensus recommends that follow-up intensity should be adjusted
based on genotype^[67]4,[68]5,[69]8. Secondly, identifying a germline
pathogenic variant enables the presymptomatic screening of family
members of an index case^[70]15. Presymptomatic detection enables the
longitudinal monitoring of patients over time, allowing for the
identification of disease earlier and reducing the mortality and
morbidity associated with PPGLs, particularly in the context of mutated
SDHB PPGLs^[71]16. However, when germline and somatic mutations are
considered, approximately 30% of PPGLs remain without an identified
genetic cause^[72]8. This gap highlights the need to identify novel
genetic variants, which may be uncovered using next-generation
sequencing approaches. However, as new PPGL-associated genetic variants
are identified, their prevalence is low, and their pathogenicity
remains elusive. Indeed, the functional characterization of putative
genetic variants associated with PPGL has been limited by the lack of
relevant in vivo models combining quantifiable clinically relevant
phenotypes (i.e., measurable catecholamine hypersecretion), speed and
ease of use (e.g. genetic accessibility and speed of
development)^[73]17. Therefore, there is a need to establish a
streamlined process for assessing the pathogenicity of new PPGL
susceptibility genes.
SDHB pathogenic variants are among the most frequently identified
germline variants in PPGLs and confer an aggressive phenotype with a
higher incidence of metastatic disease (up to 46.6%)^[74]18,[75]19. The
SDHB gene encodes a subunit of the succinate dehydrogenase (SDH)
enzyme, also known as Complex II of the mitochondrial electron
transport chain^[76]20. This enzyme is responsible for converting
succinate into fumarate within the Krebs cycle^[77]8. This gene is
included in “cluster 1 A” of PPGL susceptibility genes, which, when
mutated, disrupts the Krebs cycle and leads to impaired mitochondrial
oxidative phosphorylation, succinate accumulation, and HIF-α
stabilization^[78]8. In SDHB-related tumorigenesis, a heterozygous
germline mutation is followed by a somatic second hit, such as loss of
heterozygosity, leading to biallelic inactivation^[79]21. This results
in accelerated protein degradation, reducing the levels of mutant SDHB
protein and leading to insufficient function^[80]22.
Recently, Dona et al. described an SDHB mutated zebrafish model that
recapitulated some features of PPGLs, such as defects in energy
metabolism and swimming behavior, decreased mitochondrial complex 2
activity and succinate accumulation^[81]23. This was the first evidence
that zebrafish can be used as a relevant in vivo proxy for modeling
PPGL physiopathology. However, this study relied on a stable
CRISPR-engineered zebrafish strain, which requires at least 12–18
months to establish. This timeline does not align with the need for a
quick functional assessment of genetic variants associated with PPGLs
in vivo. Recently, the same team showed that SDHB ± adult fish (note
that homozygotes do not reach adulthood) showed only a mild phenotype
related to the clinical presentation of PPGLs^[82]24. This suggests
that while these models may enhance our general understanding of
underlying mechanisms, they are not well-suited for rapidly assessing
the pathogenicity of genetic variants of interest.
In this project, we provide a proof of concept for the rapid in vivo
functional characterization of PPGL loss-of-function genetic models.
These CRISPant models can be generated and analyzed within a week and
are, therefore, compatible with translational precision medicine
endeavors.
Results
Sdhb-CRISPant zebrafish depict robust SDHB loss-of-function and associated
phenotypes
To efficiently invalidate SDHB expression and function in zebrafish
embryos, we took advantage of a recently proven F0-knockout mutagenesis
approach using CRISPR/Cas9 to generate SDHB loss-of-function zebrafish
larvae (so-called CRISPants^[83]25). This approach enables the
assessment of phenotypic and biochemical assays in a mutant genetic
background of interest within days following microinjection (Fig.
[84]1A). We designed and microinjected one-cell-stage embryos with a
CRISPR cocktail targeting the SDHB gene at the exons 3, 4, and 5 levels
(Fig. [85]1B). The potency of each guide RNA (gRNA) was assessed by
High-Resolution Melting (HRM) analysis (Fig. [86]1C). The reduction in
sdhb expression was evaluated at 5 dpf at both the transcriptomic and
protein levels using RT-qPCR and western blot, respectively (Fig.
[87]1D–F). Notably, the control groups consisted of embryos
microinjected with mRNA encoding the Cas9 endonuclease, but without any
gRNAs; these are referred to as “Cas9” in our figures. To validate the
relevance of our sdhb-CRISPant model, we confirmed several phenotypes
previously described in a stable SDHB-mutant zebrafish strain^[88]23.
Particularly, we showed that sdhb-CRISPant larvae do not survive past
13 dpf ( < 50% survival rate reached at 8 dpf, Fig. [89]1G) and depict
a significantly reduced motor activity, especially during light
periods, during which Cas9 controls show increased swimming movements
(Fig. [90]1H). Finally, we also confirmed that the sdhb-CRISPant larvae
exhibit a significantly increased heart rate compared to Cas9 controls
(Fig. [91]1I, J). Altogether, these results confirm the potency of the
F0-CRISPant approach to rapidly and efficiently disrupt sdhb protein
levels and function in vivo.
Fig. 1. F0-injected sdhb-CRISPant zebrafish larvae display SDHB-associated
phenotypes.
[92]Fig. 1
[93]Open in a new tab
A Timeline representation of our CRISPant approach starting with the
microinjection of a gene-specific (i.e., sdhb) CRISPR/Cas9 mix in
single-cell embryos, followed by High-Resolution Melting (HRM)
genotyping from 1-day post-fertilization (dpf). Phenotypic and
biochemical readouts can be acquired only few days following injection.
B The genetic landscape of the zebrafish sdhb gene encompassing
targeted exons 3, 4 and 5. C The mutagenic power of the injected
CRISPR/Cas9 mix is validated by HRM comparing Cas9-injected embryos as
controls in which only mRNA encoding CAS9 was microinjected (no
gene-specific guide RNAs). D, E, F The level of sdhb expression is
reduced in sdhb CRISPants both at the transcriptomic and protein levels
as assessed by qRT-PCR (D) and western blot (E, F). (G) Survival of
sdhb CRISPants (n = 51) is compared to Cas9-injected controls (n = 29)
and uninjected larvae (n = 87) over 18 days. H The distance swam,
binned per 30 s, is recorded for sdhb CRISPants (n = 32) and
Cas9-injected controls (n = 32) during 60 min dark and 60 min light
periods. I, J Representative setup and results of heart rate
quantification in 5 dpf larvae (sdhb CRISPants, n = 13; Cas9-injected,
n = 12). I Left: larvae mounted laterally in 3% methylcellulose for
simultaneous video acquisition (~5 larvae per trial). Red boxes
indicate individual larvae analyzed. Right: magnified view of one larva
with the heart region outlined (dashed line), corresponding to the area
used for automatic beat detection by DanioScope™ software. No
anesthetic was used during acquisition. J Heart rate was quantified and
compared using an unpaired t-test with two-tailed p-value calculation:
** p-value < 0.01; **** p-value < 0.0001. Icons in (A) and (B) are
created in BioRender. Samarut, E. (2025)
[94]https://BioRender.com/xk8wiuf.
sdhb-CRISPant zebrafish larvae showed elevated secretion of free
catecholamines and MMN
With this new SDHB-CRISPant genetic model in our hands, we sought to
measure the levels of free catecholamines (e.g., dopamine,
norepinephrine and epinephrine) and free MN (e.g., 3-methoxytyramine or
3-MT, normetanephrine (NM) and MN) in larvae lysates and their bathing
medium compared to Cas9-controls (Fig. [95]2A, B). This approach is
physiologically relevant, as zebrafish larvae excrete catecholamines
and other metabolites into the surrounding medium via the pronephric
kidney and cloaca at this stage^[96]26,[97]27. We found that the levels
of NM and MN were three times higher in sdhb-CRISPants (NM: 3.69,
p < 0.0001, MN: 3.35, p < 0.0001) compared to control larvae.
Interestingly, 3-MT displayed similar levels between both groups.
Dopamine (DA) levels were also elevated in sdhb larvae compared to Cas9
(2.9, p < 0.0001). However, norepinephrine (NE) and epinephrine (E)
were lower in sdhb mutants compared to Cas9 control larvae (0.81,
p = 0.017; 0.51, p < 0.0001) (Fig. [98]2C). When we checked the level
of these hormones in the fish bathing medium, we also observed that the
level of NM, MN, and to a lower extent 3-MT were significantly
increased (NM: 4.32, p < 0.0001; MN: 9.4, p < 0.0001 3-MT: 1.5,
p = 0.0076) compared to the bathing medium of Cas9-controls (Fig.
[99]2D). There was no statistical difference for DA and NE in the
bathing medium of sdhb-CRISPants compared to controls. Of note, the
levels of epinephrine were at or below the lower detection limit in
most samples for the former and in some samples for the latter (n.d in
Fig. [100]2D). Altogether, our data show that sdhb-CRISPant larvae
display hormonal hypersecretion, which is relevant to sdhb-associated
PPGLs.
Fig. 2. Catecholamine and metanephrine levels are affected in PPGL CRISPant
zebrafish models.
[101]Fig. 2
[102]Open in a new tab
A Individual embryos are microinjected at the one-cell stage and larvae
are pooled at 4 dpf overnight. Catecholamines and metanephrines levels
were assessed on day 5 from whole larvae lysates and bathing medium. B
Metabolic path of catecholamines and metanephrine synthesis. Enzymes
are italicized (DBH: Dopamine beta-hydroxylase, PNMT:
Phenylethanolamine N-Methyltransferase, COMT:
Catechol-O-methyltransferase). C,D Catecholamines and metanephrines
levels in sdhb CRISPants (n ≥ 6 pools of 10 larvae) compared to
Cas9-injected controls (n ≥ 6 pools of 10 larvae) (whole larvae, (C)
and bathing medium (D)). E,F Catecholamines and metanephrines levels in
nf1 CRISPants (whole larvae, (E) and bathing medium (F). Of note, the
levels of NE were below the detection limit in Cas9-control samples,
and a subjective basal value was used to calculate the NE fold-change
increase in nf1-CRISPant larvae. Unpaired t-test with two-tailed
p-value calculation: * p-value < 0.05; ** p-value < 0.01; ***
p-value < 0.001; **** p-value < 0.0001. Icons are created in BioRender.
Samarut, E. (2025) [103]https://BioRender.com/xk8wiuf.
Catecholamine and MN levels are reliable biochemical readouts of different
genetic PPGLs
We first wanted to validate the specificity of this hormonal
hypersecretion by showing that it cannot be attributed to a nonspecific
general physiological distress (i.e., premature death or weakened
physiology). To do so, we showed that catecholamine and MN levels are
not increased in an independent genetic model. We used the
scn1lab-mutant genetic model, an ortholog of the human SCN1A gene,
whose loss of function is associated with epileptic encephalopathy and
is characterized by premature death and general physiological distress
in zebrafish^[104]28. In these mutants (scn1lab^-/-), we noted no
increase in the levels of free catecholamines and MN, either in larval
lysates or in the ambient bath. Instead, the levels of these hormones
were significantly reduced in the mutants compared to their siblings
(Fig. [105]S1A). More importantly, this confirms that the hormonal
hypersecretion observed in our sdhb-CRISPant model is specific and
cannot be attributed to general physiological distress.
We also wanted to validate the specificity of these significant changes
by showing that they cannot be attributed to a nonspecific general
metabolic distress caused by the disturbance of the Krebs cycle, as is
the case for mutations in PPGL cluster 1^[106]20. To do so, we
generated F0-CRISPRant zebrafish targeting another PPGL canonical gene
(NF1). This gene belongs to a different genetic cluster (cluster 2)
associated with PPGL pathogenesis, which does not affect the integrity
of the Krebs cycle but rather kinase signaling^[107]8. The zebrafish’s
genome encompasses two paralogous genes (nf1a and nf1b) orthologous to
the human NF1 gene. Thus, we designed and validated six specific gRNAs
(three against each paralog) targeting these genes in zebrafish (Fig.
[108]S1B). Notably, we found a significant increase in the level of all
free catecholamines (DA: 1.95, p = 0.0245; NE:1.47, p < 0.0001; E:
1.54, p < 0.0001) and all free MN (3-MT: 2.94, p < 0.0001; NM 3.86,
p < 0.0001; MN 4.4, p < 0.0001) in nf1-CRISPant larvae extracts
compared to Cas9-controls (Fig. [109]2E). Importantly, it is worth
noting that the elevation of norepinephrine and epinephrine in
nf1-CRISPant compared to sdhb-CRISPant (Fig. [110]S1C, D) is relevant
to an adrenergic phenotype, which is a hallmark of PPGL cluster 2
mutations^[111]8. Consistently, the level of secreted catecholamines
and MN in the bathing medium of nf1-CRISPants was also found elevated
compared to Cas9-controls (Note that the levels of DA and E were below
the detection threshold in this assay, Fig. [112]2F and [113]S1D).
Altogether, these data confirm that the biochemical quantification of
free catecholamines and MN from whole larvae lysate and bathing medium
is a specific readout of PPGL.
Sdhb-CRISPant zebrafish larvae display broad metabolic perturbations beyond
Krebs metabolites
We next sought to dig into the metabolic perturbations caused by sdhb
loss-of-function, particularly among Krebs cycle metabolites and
related metabolic pathways (Fig. [114]3A). In whole-larvae lysates, we
found a significant increase in the levels of several Krebs cycle
intermediates in sdhb-CRISPrant compared to Cas9 Controls (Fig.
[115]3E–J). Notably, the levels of acetyl-CoA (1.586 p = 0.002; Fig.
[116]3E), citrate-isocitrate (sdhb: 2.402 p = 0.0008; Fig. [117]3F) and
succinate (sdhb: 21.78 p < 0.0001; Fig. [118]3H) were higher in sdhb
CRISPants than in Cas9-control larvae. This is consistent with the
impaired enzymatic activity of SDHB, which converts succinate to
fumarate (Fig. [119]3A). We also looked the levels of glycolysis
intermediates and showed that while the levels of glycerol-3-phosphate
(Gro3P) (1.652, p < 0.0001; Fig. [120]3B) and lactate (4.421
p < 0.0001; Fig. [121]3D) were significantly increased in sdhb
CRISPants, the level of DHAP (dihydroxyacetone phosphate) was
significantly reduced (0.71, p = 0.0112; Fig. [122]3C). We quantified
the levels of various dinucleotides and nucleotides central to
bioenergetics (Fig. [123]3A, K–P). We found that sdhb-CRISPant larvae
displayed higher levels of ATP (2.560, p < 0.0001; Fig. [124]3M), GTP
(1.996, p = 0.0014; Fig. [125]3N) and NAD^+ (1.498, p = 0.0013; Fig.
[126]3O) but lower levels of NADH (0.536, p = 0.0013) compared to
Cas9-control larvae. This suggests significant reprogramming of
mitochondrial energy production. Finally, we quantified the levels of
several amino acids and other metabolites linked to Krebs cycle and
redox regulation (Fig. [127]3A, Q–V). Interestingly, we noted a
significant increase in the levels of leucine (2.287 p < 0.000 1; Fig.
[128]3Q) and arginine (1.610 p = 0.0007; Fig. [129]3S) in sdhb-CRISPant
larvae compared to Cas9-controls. Conversely, the levels of aspartate
(0.2722 p < 0.0001; Fig. [130]3R), glutamate (0.8111 p < 0.0001; Fig.
[131]3U) and oxidized glutathione (sdhb: 0.7011 p < 0.0001; Fig.
[132]3V) were significantly lower in sdhb CRISPants than in
Cas9-control larvae.
Fig. 3. sdhb CRISPants depict broad metabolic pertubations.
[133]Fig. 3
[134]Open in a new tab
A Schematic of Krebs cycle intermediates (in blue) and other
metabolites from related pathways (in black) which have been assayed
from whole 5 dpf larvae lysates. The changes observed in sdhb CRISPants
are indicated with arrowheads (green: increase; red: reduction). B–V
Individual metabolite quantified from whole sdhb-CRISPant larvae (pools
of 10; n = 9 pools of 10) to Cas9-injected control larvae (n = 10 pools
of 10). W Oxygen consumption rate (OCR) was measured in whole
Cas9-injected control and sdhb-CRISPant larvae at 2 and 5 days
post-fertilization (dpf) using a Seahorse XFe96 analyzer. Unpaired
t-test with two-tailed p-value calculation: * p-value < 0.05; **
p-value < 0.01; *** p-value < 0.001; **** p-value < 0.0001.
Given that SDHB encodes a core subunit of mitochondrial complex II,
which links the Krebs cycle to the electron transport chain (ETC), we
investigated whether its disruption impairs mitochondrial respiration.
To this end, we measured the basal oxygen consumption rate (OCR) in
sdhb CRISPant zebrafish embryos and larvae using a Seahorse XFe96
analyzer. Surprisingly, OCR was not significantly different from
control embryos at either 2 or 5 dpf (Fig. [135]3W). This unexpected
preservation of mitochondrial respiration despite sdhb deficiency
suggests the activation of compensatory metabolic mechanisms to sustain
oxidative phosphorylation. Altogether, while these data confirm the
expected changes in the levels of glycolytic and Krebs cycle
intermediates associated with the impairment of SDHB activity (i.e.,
succinate and lactate increase), they also broaden the scope of the
metabolic perturbations caused by sdhb loss-of-function and indicate
that basal mitochondrial respiration is unaffected.
Sdhb loss-of-function leads to broad transcriptomic changes in zebrafish
sdhb-CRISPants
Finally, we sought to profile the transcriptional changes caused by
sdhb loss-of-function in vivo. To do so, we extracted RNAs from whole
5-dpf larvae and sequenced their transcriptome using Next-Generation
Sequencing (Fig. [136]4A). Upon differential gene expression (DEG)
analysis, we identified 443 upregulated genes and 326 downregulated
genes (filtered by a 5% adjusted p-value and log2 Fold Change
[MATH: ≥1 :MATH]
) in sdhb CRISPants compared to Cas9-controls (Fig. [137]4B and
Supplementary Data [138]1). We built a heat-map plotting the more
significant negative DEG correlations between Cas9-control and
sdhb-CRISPant samples. This map highlights sdhb as strongly
underexpressed in the mutant situation (Fig. [139]4C). Conversely,
hypoxia response genes (hif1al and igfbp1a), angiogenesis gene
(angptl4), stress response genes (fkbp5 and hsd11b2) and glycolysis
gene (pfkfb4b) were all upregulated in the mutant condition (Fig.
[140]4C). These expression changes of relevant DEGs (i.e. hif1al) were
confirmed by RTqPCR on independent biological samples (Fig. [141]4D–I).
We also found that other subunits of the SDH complex, enzymes of the
Krebs cycle or mitochondrial biogenesis markers (i.e. PGC1-alpha)
showed no significant transcriptional changes (Fig. [142]S2A).
Moreover, we examined the differential expression of genes involved in
the mitochondrial electron transport chain (ETC) and found that only
sdhb met our differential expression threshold (log2FC > 1) in sdhb
CRISPants. Four additional ETC-related genes showed mild but
statistically significant regulation: cox4i2 and atp1b1a were
moderately upregulated, while sdhaf2 and ndufaf3 were slightly
downregulated (Supplementary Data [143]2). These modest changes may
reflect localized or adaptive responses to the loss of SDHB. However,
the vast majority of ETC components, including structural subunits of
complexes I to V, remained transcriptionally unchanged, which is
consistent with the preserved oxygen consumption rate (OCR) observed in
sdhb-deficient larvae (Fig. [144]3W).
Fig. 4. Transcriptomic changes in sdhb CRISPants.
[145]Fig. 4
[146]Open in a new tab
A Experimental timeline from one-cell stage embryo microinjection to
whole larvae mRNA extraction at 5 dpf for RNA-sequencing. B Volcano
plot with differentially expressed genes (DEGs) in red (filtered with a
value < 0.05 and an absolute log2FoldChange > 1). The full list of DEGs
is available in Supplementary Data [147]1. C Heatmap representing the
top 20 DEGs between sdhb-CRISPant samples (sdhb_1, sdhb_2, sdhb_3) and
Cas9 controls (cas9_1, cas9_2, cas9_3). D–I qRT-PCR validation of
changes in the expression of relevant genes between sdhb-CRISPant
(n = 6 pools of 10 larvae) and Cas9-injected control (n = 6 pools of 10
larvae) at 5 dpf. J Lollipop plot of top 15 significantly enriched KEGG
pathways clustered from sdhb vs cas9 DEGs. The complete list of
enriched KEGG pathways is available in Supplementary Data [148]2. Icons
in panel A are created in BioRender. Samarut, E. (2025)
[149]https://BioRender.com/xk8wiuf.
Finally, using DAVID resources^[150]29, we conducted a KEGG pathway
enrichment analysis with the DEG list, which included genes that met
the criteria of p value ≤ 0.05 and log2 Fold Change ≥ 1, encompassing
both upregulated and downregulated genes (Fig. [151]4J). Our analysis
revealed several significantly enriched KEGG pathways in our
sdhb-CRISPant samples, DNA replication, cell cycle and Metabolic
Pathways (Fig. [152]4J and Supplementary Data [153]3). These data are
consistent with the function of the SDHB gene in both the mitochondrial
respiratory chain and several main metabolic processes via the Krebs
cycle.
Discussion
Genotype constitutes a pivotal factor in the comprehensive management
of PPGLs, emphasizing the need to establish genetically engineered
biological in vivo models to elucidate underlying pathogenic mechanisms
and therapeutic targets. Although it models biallelic sdhb loss, our in
vivo CRISPant design reflects the fully inactivated state found in
tumor tissue^[154]21. In this study, we present the first published
instances of elevated catecholamines/MN in a sdhb mutant zebrafish
model.
Our sdhb CRISPant zebrafish model faithfully recapitulates key features
of SDHB-related PPGLs and offers a rapid and scalable in vivo system
for functional assessment. This is evidenced by the strong elevation of
NM and MN in both whole fish extracts and bathing media, as well as
increased dopamine (DA) levels. The excess of NM and MN reached
3–4-fold in tissue and 5–9-fold in the medium. These biochemical
alterations are in line with previous descriptions in sdhb-KO zebrafish
models^[155]23 and provide a phenotypic indication of disease-related
catecholamine overproduction. Notably, heart rate was also elevated,
suggesting a functional effect of catecholamine excess.
The specificity of the catecholamine profile further validates the
relevance of our model. When targeting nf1, a PPGL-related gene from a
different molecular cluster (Kinase signaling cluster 2 for nf1 vs.
pseudohypoxic signaling cluster 1 for sdhb), larvae exhibited an
adrenergic profile characterized by elevated NE/E levels^[156]8, while
this was absent in sdhb CRISPants. This divergence supports the
genotype–phenotype specificity of catecholamine dosage in our in vivo
models and reinforces the notion that systemic gene disruption in
CRISPants reflects localized endocrine effects, particularly at the
level of the interrenal gland.
Importantly, this catecholamine elevation cannot be attributed to
stress, general weakness, or shortened lifespan. Using the scn1lab
mutant zebrafish model of Dravet syndrome^[157]28 as a disease control,
we observed that these mutants displayed reduced levels of
catecholamines/MN despite severe physiological impairment. This
strengthens the argument that the catecholamine signature in sdhb
CRISPants is specific and not a secondary effect of systemic illness.
Beyond catecholamine secretion, our model also replicates other
SDHB-related clinical features. Sdhb CRISPants exhibited impaired
mobility, consistent with the hypotonia and muscle dysfunction reported
in germline biallelic SDHB-deficient patients^[158]30. This phenotype,
although non-tumoral, provides an additional functional endpoint
relevant to human disease and suggests broader systemic implications of
SDHB loss.
At the metabolic level, our findings point to significant mitochondrial
reprogramming in response to SDHB loss. As expected, succinate levels
were markedly elevated in sdhb larva^[159]31, a hallmark of SDHx
deficiency observed in patients and associated with malignancy. Our
analysis also revealed increased NAD + , reduced NADH, and elevated ATP
levels, indicating enhanced NADH oxidation, likely via compensatory
complex I activity^[160]32,[161]33. These adaptations were further
reflected by altered levels of lactate (increased), glutamate
(decreased), and aspartate (depleted), consistent with a shift toward
glycolytic metabolism and anaplerotic support via
glutaminolysis^[162]34. Despite this profound metabolic remodeling, OCR
measurements showed no significant decrease, suggesting that
mitochondrial respiration is preserved in vivo, likely through these
compensatory mechanisms. This was also observed in C. elegans sdhb-1
mutants, where basal oxygen consumption was comparable to controls.
However, these mutants failed to increase respiration in response to a
mitochondrial uncoupling agent (FCCP), indicating a loss of spare
respiratory capacity and revealing underlying mitochondrial
dysfunction^[163]35.
Although SDHB pathogenic variants are classically associated with
PPGLs, they and other PPGL-related genes are often broadly expressed
and implicated in fundamental metabolic pathways. Systemic phenotypes
observed in our model, including motility defects and metabolic
dysfunction, may thus reveal tissue-independent consequences of gene
loss and support the use of zebrafish for broader variant modeling.
Transcriptomic profiling supports this interpretation. While sdhb was
strongly downregulated, confirming CRISPR efficiency, most ETC-related
genes remained unchanged, including those of complexes I–V. Only
atp1b1a and cox4i2 were modestly upregulated. The latter, a
hypoxia-inducible isoform, is consistent with HIF pathway activation.
Furthermore, genes involved in glycolysis (pfkfb4b), amino acid
transport (slc1a4, slc3a2b), and pyruvate/lactate shuttling (slc16a6b)
were all upregulated, providing transcriptional evidence of metabolic
rewiring. We also observed induction of HIF target genes related to
hypoxia (hif1al, igfbp1a), angiogenesis (angptl4), and stress (fkbp5,
hsd11b2)^[164]36. These changes are in line with succinate-mediated
inhibition of prolyl hydroxylases and stabilization of HIF-α^[165]37.
This pattern is consistent with other SDHB-deficient cell and animal
models^[166]35,[167]38.
Altogether, the combination of succinate accumulation, HIF-α
stabilization, altered redox status, and transcriptional upregulation
of compensatory metabolic genes provides a coherent picture of energy
adaptation caused by sdhb loss-of-function. These features not only
validate the relevance of the zebrafish CRISPant model, but also
uncover new insights into the metabolic plasticity that may underlie
tumor cell survival in the context of SDHB loss.
Because CRISPant models result in systemic gene disruption, we cannot
formally attribute the observed phenotypes to specific tissues.
However, given the strong and selective increase in catecholamines and
MN observed in sdhb CRISPants, it is reasonable to assume that this
hormonal dysregulation originates from the interrenal gland, the
zebrafish homolog of the adrenal medulla. This interpretation is
consistent with the known function and expression pattern of sdhb in
endocrine tissues. Moreover, no tumors were observed in sdhb-CRISPant
larvae (Fig. [168]S2B, Fig. [169]2C), consistent with previous
reports^[170]17,[171]23,[172]24. This is likely due to the limited
lifespan of the model. Adrenal medullary hyperplasia may also occur in
the absence of gross tumors and still be associated with catecholamine
oversecretion^[173]39. While tissue-specific CRISPR knockouts would
help refine tissue-level interpretation, the speed and scalability of
the CRISPant approach enable rapid, large-scale functional screening.
Future development of mosaic or conditional two-hit models will be
important to more accurately mimic the clinical progression of
SDHB-related PPGLs.
In conclusion, our zebrafish CRISPant model of SDHB loss recapitulates
key features of human PPGLs, including elevated NM, succinate
accumulation, mitochondrial reprogramming, and a distinct
hypoxic/glycolytic transcriptional signature. This validates its
utility as a physiologically relevant model for functional validation
of PPGL genetic variants and opens new avenues for biomarker discovery
and therapeutic screening in a high-throughput vertebrate system. By
enabling rapid in vivo modeling of SDHx mutations, this platform could
support precision medicine approaches in genetic pheochromocytoma and
paraganglioma syndromes.
Methods
Zebrafish husbandry and euthanasia
Zebrafish (Danio rerio) larvae were maintained at 28 °C in E3 embryo
medium under a 12 h/12 h light-dark cycle, following standard
procedures described in The Zebrafish Book^[174]40. All experiments
were approved by the Institutional Animal Care Committee of the CRCHUM
(#F2PZ-85986) and conducted in compliance with the guidelines of the
Canadian Council on Animal Care. Larvae were monitored daily and
euthanized, when required, using an overdose of buffered tricaine
methanesulfonate (MS-222) at a final concentration of 0.3% (w/v),
followed by observation of cessation of heartbeat for at least 2 min
before disposal. This method is consistent with the recommended
practices for zebrafish euthanasia and was selected to minimize
distress and ensure rapid and effective loss of consciousness.
Swim test
At 5 days post-fertilization (dpf), larvae were transferred
individually into a 96-well plate and swim distance was recorded using
Basler GenIcam infrared camera and DanioVision lightproof recording
chamber with an infrared camera recording chamber (Noldus). Analysis
was performed using the Ethovision XT 13 software (Noldus) to quantify
the distance swam over time.
Zebrafish sdhb and nf1 CRISPant generation and scn1a genetic model
We designed gene-specific gRNA using the online tool CRISPRscan
([175]http://www.crisprscan.org/). Three gRNAs were used against 3
exons of interest encoding important protein functional domains. For
sdhb, we targeted exon 3 (GTCTGTTAGGTGTGGGCCGA), exon 4
(CAGCGTGTGTTGTGCTCCTC) and exon 5 (GGGCTCGATGGATTTATACT). For nf1, we
targeted both zebrafish nf1a and nf1b paralogs on exons 17
(TTGGGTACGACTGTGTGAAC), exon 28 (GTAGGTAAACGTGCCTAGTA) and exon 36
(CGACTGCTGACTGGCCTCAA) for nf1a and exon 20 (CGGTCCAGAAAGCGGCCGAG),
exon 25 (GCGAGGCATGTCCAGGCGAT) and exon 34 (GGGACCCAGGTAAGCGAGCA) for
nf1b. The expression of sdhb is enriched in the brain and myotome,
while nf1a and nf1b show broad expression, particularly in the head and
heart^[176]41,[177]42.
Synthetic gRNAs were ordered from Synthego (CA, USA) and Cas9 mRNA was
in vitro synthesized as described previously^[178]43. Zebrafish
wild-type embryos were microinjected with a 1 nL drop of a mix of
100 ng/μL of Cas9 mRNA and 30 or 100 ng/μL of exon-specific gRNAs into
one-cell stage embryos using a Picospritzer III pressure ejector.
Control-injected embryos were microinjected following the same process
but with Cas9 mRNA only (no gRNAs). As previously described, the
mutagenic score of each gRNA was confirmed by HRM^[179]44. A stable
genetic model of scn1lab (e.g. zebrafish ortholog of human SCN1A) with
a frameshifting mutation (ins+1) in exon 8 (leading to a premature stop
at position 288) was used as a model for Dravet syndrome a form of
epileptic encephalopathy^[180]45. The identification of homozygous
larvae was done via hyperpigmentation, as previously described^[181]45.
Their siblings, which included healthy heterozygous and wild-type
larvae, were taken as +/sib controls.
RT-qPCR and Western Blot
RT-qPCR was performed as previously described^[182]46. RNA was isolated
from three sets of 5–7 larvae using QIAGEN Rneasy ® Plus Mini Kit
according to the manufacturer’s protocol. 500 ng of total RNA was used
for cDNA synthesis using the SuperScript®Vilo™kit (Invitrogen). RT–qPCR
was run with SYBR Green Master Mix (Bioline) using a LightCycler®96
(Roche). Polr2d was used as the reference gene for ddCt differential
expression analysis. Primer sequences are available upon request. For
western blot, whole proteins were extracted from pools of seven 5 dpf
larvae using RIPA buffer supplemented with Protease Inhibitor Cocktail
and PhosSTOP (Millipore Sigma). 50 µg of total proteins were
immunoblotted using a monoclonal anti-SHDB antibody (1:1000, Abcam
#ab14714; 5% milk in PBS Tween 0.1%) or a monoclonal anti-GAPDH
antibody (1:500, Invitrogen MA5-44678; 5% BSA in PBS Tween 0.1%)).
Relative expression analysis was performed using ImageLab software
(Biorad).
Heart Rate Monitoring
5 dpf sdhb-mutant and Cas9-control zebrafish larvae were placed in a
6 cm plastic Petri dish. 3% methylcellulose was used to mount each
larva laterally in order to see the heartbeat. No anesthetic (e.g.,
tricaine) was used during the procedure to avoid any confounding
effects on cardiac activity. Video recordings of 1 min were captured
for each fish group using an iPhone 6 and an ocular holder. Heart rate
was quantified unbiasedly using the cardiology feature of the
DanioScope^TM software (Noldus).
Survival Assay
Groups of ~30 uninjected or injected larvae (either Cas9-control or
sdhb-mutant) were raised in a 500 ml beaker, and they were fed every
day from day 5. A mortality report was filled daily, and these values
were plotted in a Kaplan-Meier estimator, comparing mutant,
Cas9-controls and wild-type larvae.
Targeted metabolomics
Pools of ten 4 dpf larvae (sdhb-CRISPant or Cas9-controls) were placed
in 1.5 ml Eppendorf tubes with the lid open overnight at 28.5 °C. The
following day (e.g., 5 dpf), the bathing medium (1 ml) was harvested
and flash-frozen as well as dried larvae (pools of 10). Metabolites
from central carbon metabolism were measured from pools of ten larvae
as described previously^[183]47 with the following modification: larvae
pellets were extracted in ice-cold 80% methanol in water, 2 mM ammonium
acetate, pH 9.0. Catecholamines were measured using a method modified
from ref. ^[184]48. Briefly, samples of 10 larvae and 1 mL of bathing
water were freeze-dried before addition of 30 µL of 80% acetonitrile in
water, containing 50 nM of deuterated NM and norepinephrine as internal
standards (CDN Isotopes, Pointe-Claire, QC, Canada). Larvae samples
were sonicated in ice-cold water using a cup-horn sonicator (Q700,
Qsonica, Newtown, CT) at 150 watts for 2 min (cycles of 10 s on, 10 s
off). Both larvae and water samples were centrifuged 20,000 x g, 10 min
at 4 °C. Supernatants (20 µL) were derivatized and analyzed as
described in the original method. Detection of MN was achieved using
transition 406.2/104.9 [M (with 2 benzoyl groups) + H + ]. All
quantitative results were normalized to the Cas9-control group.
RNA-sequencing
Total RNA was extracted from three batches of 5 dpf sdhb-CRISPant and
Cas9-control larvae (10 larvae per sample). RIN greater than 9 was
confirmed for each sample using an Agilent 2100 Bioanalyzer System. The
NEBNext Ultra II directional RNA library prep kit for Illumina (New
Englands Biolabs Inc., Ipswich, MA, USA) was used to prepare mRNA
sequencing libraries, according to the manufacturer’s instructions.
Briefly, 1 ug of total RNA was purified using the NEBNext poly (A) mRNA
Magnetic Isolation module (New England Biolabs Inc., Ipswich, MA, USA)
and used as a template for cDNA synthesis by reverse transcriptase with
random primers. The specificity of the strand was obtained by replacing
dTTP with dUTP. This cDNA was subsequently converted to double-stranded
DNA and end-repaired. Ligation of adapters was followed by a
purification step with AxyPrep Mag PCR Clean-up kit (Axygen, Big Flats,
NY, USA), by excision of the strands containing the dumps and finally,
by a PCR enrichment step of nine cycles to incorporate specific indexed
adapters for the multiplexing. The quality of final amplified libraries
was examined with a DNA screentape D1000 on a TapeStation 2200, and the
quantification was done on the QuBit 3.0 fluorometer (ThermoFisher
Scientific, Canada). Subsequently, mRNA-seq libraries with unique dual
indexes were pooled together in equimolar ratio and sequenced for
paired-end 100 bp sequencing on an Illumina NovaSeq 6000 at the
Next-Generation Sequencing Platform, Genomics Center, CHU de
Québec-Université Laval Research Center, Québec City, Canada. The mean
coverage/sample was 25 M paired-end reads.
Differential gene expression analysis
Alignment was performed using STAR in two pass mode (v2.7.11a) against
the reference genome Danio rerio Ensembl version GRCz11.105. Read
counts were obtained using featureCounts (v2.0.6) prior to DESeq2
(v1.42.1) differential expression quantification using R version 4.3.3.
The Ensembl gene IDs were matched to gene names using biomaRt (v
2.58.2). Results (Supplementary Data [185]1) were filtered using an
absolute log2 Fold Change
[MATH: ≥ :MATH]
1 and p value of
[MATH: ≤ :MATH]
0.05 for further gene ontology analysis. Data visualization was
performed using several R packages. For PCA, the DEGs were transformed
using the regularized log function and used in the pcaplot function of
pcaExplorer (v1.0.2) using the top 1000 variables. Heatmaps were made
using pheatmap (v1.0.12) and represent the top 20 most differentially
expressed genes. The package ggplot2 (v 3.5.1) was used for volcano
plots.
Oxygen consumption rate (OCR)
The day before the assay, Seahorse XF 96-well sensor cartridges
(Agilent) were hydrated overnight at room temperature in the dark,
using 200 µL of Seahorse XF calibrant (Agilent) per well. On the day of
the assay, Cas9- or sdhb-injected zebrafish embryos were plated in a
Seahorse XF 96-well cell culture microplate (Agilent) in a total volume
of 200 µl of E3 medium.
The oxygen consumption rate (OCR) was measured during 25 cycles (mix:
0 min, wait: 1 min, measure: 2 min) using a Seahorse XFe96 analyzer
(Agilent). The OCR data were analyzed using Wave 2.6.1 software
(Agilent). The raw OCR data were first subjected to range-based
filtering. Values falling outside the physiological range ( < 30 or
>400 pmol/min) were excluded as outliers. For each larva, we then
computed the standard deviation (SD) of OCR measurements to evaluate
signal consistency. Larvae were included in the analysis if the
intra-individual SD was less than 50% of the mean OCR. For larvae
exceeding this threshold, individual data points were examined, and a
maximum of one outlier value was removed per larva. If this correction
brought the SD below 50% of the mean, the larva was retained;
otherwise, it was excluded from further analysis.
Statistical analysis
Graphpad Prism software (Version 9.5.1 for Mac, Graphpad Software,
[186]www.graphpad.com) was used to perform statistical analysis and
generate graphs. T-tests were performed for variables following a
normal distribution, while ANOVA was used for comparisons across
multiple groups. For survival analysis, a Kaplan-Meier curve was
generated, and statistical differences were assessed using a log-rank
test.
Supplementary information
[187]41525_2025_518_MOESM1_ESM.pdf^ (1.8MB, pdf)
Parisien-La Salle et al Supplementary Information
[188]Supplementary Data 1^ (54.9KB, xlsx)
[189]Supplementary Data 2^ (18.4KB, xlsx)
[190]Supplementary Data 3^ (16.6KB, xlsx)
Acknowledgements