Zebrafish raptor mutation inhibits the activity of mTORC1, inducing craniofacial defects due to autophagy-induced neural crest cell death

The mechanistic target of rapamycin (mTOR) pathway senses DNA damage, growth factor signaling, and the availability of nutrients like amino acids and glucose (Chantranupong et al., 2016; Feng et al., 2005; Gu et al., 2017; Gwinn et al., 2008; Inoki et al., 2002; Manning et al., 2002; Wolfson et al., 2016). It incorporates these signals to regulate cellular processes such as protein synthesis, cell growth, coordinated cell death and autophagy (Fingar et al., 2002; Hosoi et al., 1999; Hosokawa et al., 2009; Kim et al., 2011; Ma and Blenis, 2009). mTOR is a serine/threonine protein kinase that forms the catalytic subunit for two protein complexes called mTOR complex 1 and 2 (mTORC1 and mTORC2) (Sabatini et al., 1994). These two complexes can be identified by the presence of Raptor and Rictor, respectively. Both complexes receive signaling from growth factors, but only mTORC1 has been shown to sense amino acid availability, glucose, oxygen and cellular stressors such as DNA damage (Saxton and Sabatini, 2017). mTORC1 incorporates these upstream signals to control vital downstream mechanisms, including protein synthesis/translation, metabolism, cell growth, cell proliferation, cell survival and autophagy (Saxton and Sabatini, 2017).

Based on its cellular functions, the mTORC1 pathway is a key regulator of aging and cancer (Saxton and Sabatini, 2017). More recently, the role of mTORC1 in development has gained attention. In zebrafish, raptor is ubiquitously expressed from neurulation stage, 10 h post fertilization (hpf), until pharyngula stage, 24 hpf (Burkhalter et al., 2013). At 24 hpf, raptor is strongly expressed in the head, with some expression in the tail and digestive organs (Makky et al., 2007). This pattern of expression remains until 3 days post fertilization (dpf) and is reduced at 4 dpf (Makky et al., 2007). In zebrafish, mtor is involved in myelin sheath formation and retinal pigmented epithelium regeneration (Fedder-Semmes and Appel, 2021; Lu et al., 2022). Additionally, inhibitor studies have implicated mTOR function in development of the digestive tract and in fin and retina regeneration (Hirose et al., 2014; Zhang et al., 2020). However, the specific involvement of mTORC1 during zebrafish development is not known.

In mice, mTOR-deficient embryos die shortly after implantation (Guertin et al., 2006). Raptor mutants implant, but fail to expand or differentiate (Guertin et al., 2006). Owing to this early requirement of mTOR during development, conditional mouse mutants are required to study its role in embryogenesis. These studies have demonstrated the primary roles of mTORC1 in regulating cell size, differentiation and cell death. To study its role in chondrogenesis, mouse Prx1-Cre;mTOR^fl/fl and Prx1-Cre;Raptor^fl/fl models were created (Chen and Long, 2014). This study demonstrated that loss of mTORC1 signaling in chondrocytes produced shortened limbs, exencephaly and neonatal death (Chen and Long, 2014). Similarly, deletion of Mtor or raptor (Rptor) in the dental epithelia induced tooth malformation and cystogenesis via reduced cell size and proliferation (Nie et al., 2020).

Mouse Wnt1-Cre; mTOR^loxp/loxp, Wnt1-Cre; raptor^loxp/loxp, and Wnt1-Cre; rictor^loxp/loxp conditional mutants were used to study the role of mTORC1 and mTORC2 in neural crest derivatives (Nie et al., 2021, 2020, 2018). Wnt1-Cre; mTOR^loxp/loxp mice have malformations of neural crest derivatives, including a growth arrest in the facial primordia and orofacial clefts. These malformations are associated with excessive apoptosis and reduced proliferation of post-migratory neural crest cells. Interestingly, the Wnt1-Cre; raptor^loxp/loxp line phenocopied Wnt1-Cre; mTOR^loxp/loxp mice, whereas the Wnt1-Cre; rictor^loxp/loxp mice displayed only mild craniofacial malformations. Together, these data indicate that mTORC1 (rather than mTORC2) plays the primary role in neural crest survival and proliferation. Moreover, P53 (Trp53) haploinsufficiency attenuated the craniofacial phenotype in Wnt1-Cre; mTOR^loxp/loxp mice, suggesting that loss of mTORC1 induces neural crest cell death via activation of P53. This is supported by previous experiments in cell culture, which showed that the mTOR inhibitor rapamycin induces apoptosis via reduced expression of Mdm2, a negative regulator of P53 (Moumen et al., 2007). However, it remains unclear which function of mTORC1 leads to cell death.

One of the primary roles of mTORC1 is the regulation of autophagy. Autophagy is a self-degradative and cytoprotective process by which macromolecules and organelles are broken down into their component parts in response to cellular stress and nutrient deprivation. Three unique types of autophagy have been identified: macroautophagy, microautophagy and chaperone-mediated autophagy; the former is the best-studied form (hereafter referred to as autophagy). The cellular process of autophagy follows four sequential steps (initiation, elongation, maturation and degradation) and uses more than 30 proteins (Xie and Klionsky, 2007). All four of these steps are regulated by mTORC1 (Dossou and Basu, 2019).

Although autophagy was historically viewed as a cytoprotective process, it is now known that autophagy can inhibit or activate cell death, depending on the cellular context. Upon starvation, loss of autophagy genes induces cell death in yeast, plants and human cell lines (Boya et al., 2005; Doelling et al., 2002; Tsukada and Ohsumi, 1993). In these cases, autophagy is a cytoprotective survival factor. Alternatively, autophagy-induced cell death was first shown in Drosophila development, during midgut removal (Denton et al., 2009). In this system, inhibition of caspases did not reduce midgut cell death; however, inhibition of autophagy genes reduced and delayed midgut cell death. Furthermore, mouse embryonic fibroblasts deficient in pro-apoptotic genes Bak (Bak1) and Bax undergo non-apoptotic cell death upon etoposide treatment (Shimizu et al., 2004). This cell death is associated with increased autophagosomes and is suppressed by the autophagy inhibitor 3-methyl adenine or by knockdown of the autophagy genes Atg5 or beclin 1 (Becn1) (Arakawa et al., 2017; Shimizu et al., 2004). Collectively, this provides strong evidence that impaired regulation of autophagy can induce or inhibit apoptosis, depending on the context. Understanding how mTOR-dependent autophagy regulates cell death in distinct tissues of the developing embryo will be important for understanding the role of this process in birth defects and developmental diseases.

Here, we characterize the role of mTORC1 during zebrafish craniofacial development. We generated a raptor mutant zebrafish line and demonstrate that proper craniofacial development requires mTORC1. Loss of raptor reduces the size of craniofacial elements via a reduction of individual cell size and increased apoptosis. We find that these defects are governed by an elevation of mTORC1-mediated autophagy. Mutants for the autophagy gene atg7 also have smaller than normal craniofacial skeletal elements, although they are larger than raptor single mutants. Zebrafish atg7;raptor double mutants have significantly larger skeletal elements than raptor single mutants. Collectively, our findings demonstrate that neural crest cell survival and subsequent craniofacial development require properly tuned autophagy.

To begin our analyses, we first validated the expression raptor in wild-type fish at 48 hpf (Fig. S1). Consistent with previous characterizations we find broad, likely ubiquitous, expression of raptor (Makky et al., 2007).

To test the function of raptor, we designed a gRNA targeting exon 7, out of 36 total exons (Fig. 1A). Sanger sequencing demonstrates that we induced a 4 bp deletion (Fig. 1B), inducing a predicted premature stop codon after residue 237, out of 1363 total amino acids (Fig. 1C) (http://zfin.org/ZDB-GENE-130530-685). We designate this allele raptor^au93, referred to herein as raptor mutants or raptor^−/−. The allele is predicted to be c.702_705delAGCT on the cDNA level and p.A231RfsX8 at the protein level. Loss of raptor leads to premature death beginning at 6 dpf, with no mutants alive by 10 dpf, in contrast to the 97.04% survival rate in their wild-type and heterozygous siblings (Fig. S2).

The nature of the au93 allele could result in nonsense-mediated decay of the transcript and subsequent loss of protein. However, qPCR using a primer set that amplifies across the mutation site revealed only a slight, and non-significant, reduction in transcript in mutants relative to wild types (Fig. 2A,B). To validate this finding, we designed qPCR primers that bind to the mutation site and selectively amplify the wild-type allele or the mutant allele (Fig. 2A). This strategy successfully distinguishes the alleles as there is virtually no amplification in mutants using the wild type-specific primer and vice versa (Fig. 2B). We conclude that there is no residual wild-type transcript in our mutant at 4 dpf and that the mutant allele induces minimal nonsense-mediated decay (Fig. 2B).

To assess the expression of Raptor protein in our mutants, we performed immunoblots with a N-terminal Raptor antibody (Cell Signaling, 2280) in raptor mutants and controls. At 4 dpf, no truncated protein was observed despite the presence of mRNA (see Fig. S3 for an immunoblot of a full membrane). Surprisingly, we observed that 20% of full-length Raptor protein levels remained in the mutant fish (Fig. 2C,D). We hypothesize that this Raptor protein could be the remnants of maternally deposited raptor transcript. RT-PCR from two-cell zebrafish embryos demonstrates that raptor is maternally deposited, as are all five canonical mTOR pathway genes that we measured (Fig. S4). We conclude that this raptor mutation is functionally a null mutation and that maternally deposited transcripts may be sufficient to continue early Raptor-dependent developmental events.

To validate that this mutant disrupts mTORC1 function, we measured total protein expression and phosphorylation of canonical mTORC1 targets: S6k, S6 and 4E-BP1. At 4 dpf, we observed total protein levels of mTORC1 targets S6k, S6 and 4E-BP1 in raptor mutants that are similar to those in wild-type and heterozygous fish (Fig. 3A,B). However, raptor mutant lysates had a 40% reduction in phosphorylation of each target, at known mTORC1-regulated phosphorylation sites (Fig. 3C,D). These data are consistent with those in mouse that showed a similar reduction in phosphorylation of these targets in Mtor or Rptor, but not Rictor, mutants (Chen et al., 2012; Nie et al., 2018). Thus, zebrafish raptor mutants, as expected, have reduced activation of mTORC1.

To determine the role of raptor in craniofacial development, we stained 4 dpf zebrafish with Alcian Blue and Alizarin Red, to label cartilage and bone, respectively. The craniofacial skeleton was largely intact in raptor mutants, with the exception of lost ceratobranchial cartilages (Fig. 4A-C). However, we observed a reduction in the size of the craniofacial skeleton in raptor mutants compared with wild types and heterozygotes (Fig. 4A-C′). To quantify this effect, we recorded six measures of skeletal elements. We measured the length and the width of the ethmoid plate as well as the length of the entire neurocranium (Fig. 4A-C) in wild types (n=32), heterozygotes (n=52) and mutants (n=26). We also measured the paired trabeculae, Meckel's cartilages and ceratohyals in the same three genotypes (n=64, 104 and 52, respectively).

We found highly statistically significant differences for all measures in mutants relative to wild types and heterozygotes (Fig. 4D-I,). Heterozygosity for raptor was well tolerated, with no differences between wild-type and heterozygous fish. Therefore, for simplicity we limit our comparisons here to wild-type and mutant fish (see Table S1 for complete statistical comparisons).

In the neurocranium, the length (P<3.8×10⁻⁶) and width (P<1.3×10⁻⁵) of the ethmoid plate (Fig. 4D) was smaller in raptor mutants (mean±s.e.m.=149.94±4.12 µm and 152.75±3.36 µm, respectively) than in wild types (178.97±3.89 µm and 174.11±2.83 µm, respectively). Trabeculae length (Fig. 4F) was significantly reduced (P<1×10⁻⁸) in mutants (164.38±2.33 µm) relative to wild types (181.08±2.22 µm). Consistent with these findings, the length of the entire neurocranium (Fig. 4G) was also reduced (P<1×10⁻⁷) in raptor mutants (581.29±7.37 µm) compared with wild types (mean=647.03±7.24 µm).

In the viscerocranium (Fig. 4A′-C′), we measured Meckel's cartilage and ceratohyal length (Fig. 4H,I). We found each element to be substantially smaller in mutants (121.57±2.01 µm and 164.47±2.69 µm, respectively) than in wild types (152.08±2.19 µm and 206.41±2.68 µm, respectively). These differences were highly statistically significant (P<1×10⁻⁸, for both measures).

In addition to the skeletal defects, the eyes of raptor mutants appeared small (Fig. 4C). We measured the width of the eye to quantify eye size (Fig. 4J). Here again, we found statistically significant differences between mutants relative to either heterozygotes or wild types, but no difference between wild types and heterozygotes. We limit our discussion to the difference between mutants and wild types (P<1×10⁻⁸). The raptor mutant eyes averaged 293.77±3.90 µm compared with wild types with an average of 324.16±3.86 µm. To determine whether the effect on head structures was due to a general reduction in size, we measured total body length and found no statistically significant differences between groups (Fig. 4K, Fig. S5). We also normalized the skeletal measures to the overall body length. We found that all of the craniofacial elements remained significantly smaller when accounting for body length (Fig. S6).

Given the strong linear relationship between body length and age (Cubbage and Mabee, 1996), we also analyzed overall length at 6 dpf and still found no differences (P>0.95) between mutant (3.83± 0.029 mm, n=7) and wild-type fish (3.82±0.024 mm, n=9). Combining these data, we find that although there are no differences by genotype at either 4 or 6 dpf, regardless of genotype, all 6 dpf fish were significantly larger than the 4 dpf (Table S1). Furthermore, the overall growth of the fish over these 2 days was nearly identical, with mutant and wild-type fish growing 1.01 and 1.1 mm, respectively. These findings demonstrate that the reduction in the craniofacial skeleton is not due to a generalized developmental delay.

To further characterize the developmental trajectory of the craniofacial skeleton, we analyzed earlier time points to assess differentiation of these neural crest-derived structures. At 30 hpf, cranial neural crest cells are labeled by dlx2a and chondrogenic condensations are labeled with barx1. Additionally, the expression of nkx3.2, which is necessary for jaw joint formation, is only being initiated at this time. We found that the expression of these three markers was indistinguishable between mutant and wild-type embryos (Fig. S7). Furthermore, at the onset of chondrogenesis, 52 hpf, mutant and wild-type embryos appear identical (Fig. S8).

To determine whether there were generalized defects in the mutant zebrafish, we quantified the length of the notochord at 6 dpf and the number of melanocytes in the trunk at 30 hpf. We found no significant difference in the notochord length between wild types (730.05±3.8 μm) and mutants (725.66±9.93 μm, P=0.7345), and no significant difference in the number of melanocytes between wild types (18.13±1.52) and mutants (17.91±1.46, P=0.9187) (Table S1). Collectively, these data demonstrate that there are no generalized defects or developmental delay to the mutant fish, indicating that mTORC1 signaling is particularly important for maintaining the proper size of the craniofacial skeleton.

The reduction in cartilage size could be due to a reduction in the size of individual chondrocytes and/or a loss of cells. We examined these possibilities using the trabecula, because it is made up of a single sheet of cells that can be accurately measured and counted. We quantified cell size by calculating circumference and area of five randomly picked cells from each trabecula. We observed a significant reduction in the area and circumference of trabecula cells in the raptor mutants compared with the wild-type fish (Fig. 5A,B, representative images in Fig. 5D,E, P=0.0000013 and P=0.0138, respectively). The area of wild-type fish cells averaged 77.2±2.30 μm/cell (n=95) and their circumference averaged 39.5±0.78 μm/cell (n=95), while the area of mutant cells averaged 61.15±2.06 μm/cell (n=73) and their circumference averaged 36.67±0.80 μm/cell (n=75). This finding is consistent with the known role of mTORC1 in regulating cell size.

To determine whether there were differences in the total number of chondrocytes, we counted the number of cells within the trabeculae. The trabeculae are defined as the neural crest-derived bilateral cartilages connecting the ethmoid plate to the mesoderm-derived posterior neurocranium, based on fate map analyses (McCarthy et al., 2016). We observed a significant reduction in the total number of cells in the trabecula of mutants (Fig. 5C, representative images in Fig. 5F,G, P=0.0038). Wild-type fish averaged 40.11±0.83 cells/trabecula (n=19), whereas mutants averaged only 35.3±1.29 (n=18). This reduction in the total number of cells could indicate that there is impairment in the regulation of neural crest cell death, as described in mouse models (Nie et al., 2018).

We measured TUNEL-positive cell death in first and second pharyngeal arch cranial neural crest cells to determine whether cell death contributes to the craniofacial defects in raptor mutants. At 24 hpf, we observed no differences in the level of cell death between groups, and neural crest cells appeared to populate the pharyngeal arches normally (Fig. S9). At 36 hpf, we observed a slight, non-significant, increase in heterozygous fish with 2.9±0.36 TUNEL-positive cells (n=10) relative to wild types with 1.57±0.57 TUNEL-positive cells (n=7; Fig. 6A,B,D). The raptor mutants averaged 9.6±2.41 TUNEL-positive cells (n=5), a significant increase in cell death relative to both wild types and heterozygotes (Fig. 6CD, P=0.0006 and P=0.0019, respectively). This increase in neural crest cell death at 36 hpf is consistent with the reduced number of neural crest-derived chondrocytes in Fig. 5. Furthermore, this increase in cell death may also explain the smaller craniofacial elements observed in Fig. 4.

To directly test the involvement of cell death in the craniofacial defects in raptor mutants, we treated zebrafish with a caspase inhibitor previously demonstrated to be effective in zebrafish (McCarthy et al., 2013). We quantified viscerocranial elements to determine the involvement of apoptosis in the raptor mutant phenotype. We found that caspase inhibition significantly restored the number of ceratobranchial cartilages, which number five per side in wild types, and improved the size of Meckel's cartilage. Untreated raptor mutants had an average of 2.65±0.13 ceratobranchial cartilages per side (n=14) (Fig. 7). This number was significantly elevated to an average of 4.42±0.19 ceratobranchial cartilages in inhibitor-treated mutants (P<0.0001, n=12). The size of Meckel's cartilage was also significantly improved (P=0.0189) in inhibitor-treated mutants (113±2.35 μm, n=12) relative to untreated mutants (104.9±2.17 μm, n=12, Fig. 7). The ceratohyal cartilage was larger, but not significantly so, in inhibitor-treated fish relative to untreated fish (177.98±4.45 and 172.54±4.06, respectively; n=12 for both groups). This partial rescue may be related to the inhibitor not fully blocking cell death but is also likely due to the impacts of mTORC1 on cell size (see Fig. 5). Collectively, these results demonstrate that elevated apoptosis contributes to the craniofacial defects in raptor mutants.

One of the primary regulators of autophagy is mTORC1 and excess autophagy can lead to cell death. Because mTORC1 inhibition is known to activate autophagy, we predicted that autophagy would be elevated in raptor mutant fish. We measured autophagy by identifying overlapping puncta in transgenic CMV:EGFP-lc3 zebrafish with LysoTracker Red stain (Fig. 8A-C″) (He and Klionsky, 2010; Mathai et al., 2017; Moss et al., 2020). As a control, we measured autophagy in 36 hpf rapamycin-exposed wild-type embryos. These embryos averaged 16±1.30 dual puncta in the first two pharyngeal arches (n=5), nearly a fourfold increase compared with vehicle, which averaged 4.4±1.08 overlapping puncta (n=5, P=0.00027, Fig. 8D), validating the approach. Similarly, raptor mutants average about twice the number of overlapping puncta in the first two pharyngeal arches, compared with wild-type and heterozygous siblings (Fig. 8A,B,C,E). Wild-type fish averaged 4±0.80 overlapping puncta (n=5), heterozygotes averaged 5.2±0.95 (n=5), and mutants averaged 9.8±0.91 (n=5). This elevation in mutants was significant relative to both heterozygotes (P=0.017) and wild types (P=0.0039).

To validate our transgenic findings, we characterized the ratio of Lc3-II/Lc3-I as a measure of autophagy. At 4 dpf, when only 20% of Raptor protein is present, we observed a significant 2.4±0.41-fold increased Lc3-II/Lc3-I ratio in raptor mutants (n=5), relative to wild types (Fig. 9A,B, s.e.m.=0.26, n=5, P=0.035). At 36 hpf, when cell death is elevated in raptor mutants, we similarly observe a significant 3.92±1.04-fold increased Lc3-II/Lc3-I ratio in raptor mutants (n=5) compared with wild-type (s.e.m.=0.32, n=5) and heterozygous fish (Fig. 9C,D, P=0.033). This demonstrates that autophagy is elevated at a time when neural crest cell death is occurring in zebrafish raptor mutants.

Elevations in the Lc3-II/Lc3-I ratio can be due to either an elevation in the rate of autophagy (autophagic flux) or a blockage of autophagosome maturation. To determine whether autophagic flux is elevated in raptor mutants, we employed chloroquine, a small molecule autophagy inhibitor (Boya et al., 2005; Lum et al., 2005). Chloroquine inhibits autophagic flux by disrupting the fusion between the autophagosome and the lysosome, and is known to have additional cellular effects, including its role as an antimalaria medication via preventing polymerization of a toxic heme (Mauthe et al., 2018; Sullivan et al., 1996). Chloroquine should increase total Lc3-II levels if flux is elevated (Zhang et al., 2013). Indeed, we observed a 1.69±0.23-fold increase in wild-type fish (no treatment: n=3; chloroquine-treated: s.e.m.=0.16, n=3; P=0.34), a 1.75±0.16-fold in heterozygous fish (no treatment: n=3; chloroquine-treated: s.e.m.=0.74, n=3; 0.27), and a 2.90±0.28-fold in mutants (no treatment: n=3; chloroquine-treated s.e.m.=0.71, n=3, 0.032) (Fig. S10). This increase in Lc3-II in the presence of the autophagy inhibitor chloroquine supports our hypothesis that the elevated autophagy observed in raptor mutants is indeed autophagic cargo/substrate flux.

We performed immunoblot analyses on 36 hpf mutants to verify a reduction of Raptor protein at the time point when we observed elevated autophagy and cell death. Indeed, we found that raptor mutants (s.e.m.=0.10, n=5) have roughly a 40% reduction in Raptor protein compared with wild-type fish (Fig. 8E,F, s.e.m.=0.14, n=5, P=0.074). Here too, we do not observe any truncated protein, suggesting that the Raptor protein is derived only from maternally provided transcripts. We conclude that reductions in Raptor protein led to elevated autophagy and cell death, which are associated with later craniofacial defects.

If elevated autophagy causes craniofacial defects in raptor mutants, then chloroquine should also restore the size of the craniofacial skeleton in mutants. Zebrafish embryos were exposed to chloroquine from 6 hpf to 4 dpf, stained for cartilage and bone, and imaged. In contrast to unexposed raptor mutants (Fig. 4), chloroquine-exposed mutants closely resemble wild types and heterozygotes (Fig. 10A-C′). To quantify this effect, we compared linear measures of unexposed (from Fig. 4) and chloroquine-exposed (Fig. 10D-K) fish. A full list of statistical differences can be found in Table S1; for clarity, we focus our analysis on chloroquine-exposed and unexposed raptor mutants. All measures except ethmoid plate width, ceratohyal length and eye width were significantly larger in the chloroquine-exposed raptor mutants compared with unexposed mutants (Fig. 10D-K, see Table S1 for full statistics). Ethmoid plate length and width, as well as neurocranium length were restored to be statistically indistinguishable from untreated wild types, although the other measures remained smaller. Collectively, these data demonstrate that mTORC1 inhibition reduces the size of craniofacial elements partially through increased cell death due to the overactivation of autophagy.

To provide additional support for our inhibitor findings, we used a genetic approach to inhibit autophagy. The atg7^sa1476 mutant allele has been shown to reduce autophagy in zebrafish and is hereafter referred to as atg7^−/− (Mawed et al., 2022; Siddiqi et al., 2019). We examined raptor and atg7 wild-type, single-mutant and double-mutant embryos for craniofacial skeleton defects at 4 dpf. Unlike chloroquine exposure, loss of atg7 did cause a reduction in the craniofacial skeleton, although not as profound as loss of raptor alone (Fig. 11A-D′). Furthermore, the phenotype in the double mutant closely resembled that of the atg7 single mutant (Fig. 11B,B′,D,D′).

To quantify this effect, we compared linear measurements across groups (Fig. 11E-L). As before (Fig. 4), all measures except body length were significantly smaller in raptor single mutants relative to wild types (Table S1, Fig. 11E-L; for clarity, we only show the statistical comparison of raptor single mutants with raptor;atg7 double mutants in the figure). The atg7 mutant phenotype consisted of significant reductions, relative to wild type, in all measures except ethmoid plate width and body length (Table S1, Fig. 11E-L). For all measures except body length, raptor single mutants were significantly smaller than atg7 single mutants (Table S1, Fig. 11E-L).

If autophagy causes the craniofacial defects in raptor mutants, we expect the double mutant phenotype to be less severe than the raptor single mutant phenotype. Consistent with our model, the size of all craniofacial measurements except ceratohyal length was significantly increased in raptor^−/− ;atg7^−/− fish, relative to raptor^−/− fish (Table S1, Fig. 11E-J). Additionally, eye size was also improved in the double mutant (Table S1, Fig. 11K). This is consistent with the model of craniofacial defects in raptor mutants being due to excessive autophagy. Collectively, these results suggest that a balance of autophagy plays a crucial role in development, as inhibition or elevation of autophagy disrupts craniofacial morphology.

Both cell death and autophagy are regulated by raptor. We have demonstrated that inhibition of either of these cellular processes partially rescues raptor mutation-induced craniofacial defects (Figs 7, 10 and 11). To determine whether elevated autophagy is the underlying cause of cell death, we analyzed chloroquine-treated fish. We quantified neural crest cell death via TUNEL in the fli1:EGFP transgenic line (Fig. 12A-C). In contrast to untreated raptor mutants (Fig. 6), mutants exposed to chloroquine beginning at the onset of gastrulation had no statistical difference in cranial neural crest cell death (Fig. 12D, wild type: n=7, mean±s.e.m.=1.86±0.55; heterozygote: n=12, 2.58±0.72; mutant: n=8, 4.0±0.47). These findings indicate that loss of mTORC1 signaling induces cell death via an increase in autophagy.

In this study, we have explored the role of mTORC1 signaling in craniofacial development. We have demonstrated that loss of the mTORC1 gene raptor severely reduces the size of the zebrafish craniofacial skeleton. This finding is in contrast to previous studies suggesting that mTOR inhibition, via rapamycin treatment, does not alter craniofacial development (Makky et al., 2007). This difference is unlikely due to chemical versus genetic attenuation of the pathway, as our previous analyses using rapamycin also showed defects to the craniofacial skeleton (McCarthy et al., 2013). Instead, this difference is likely due to the dose of rapamycin used across studies. Here, we show that this size impairment is due to a reduction of chondrocyte cell size and neural crest cell death. In addition, we have shown that this cell death is, at least in part, due to increased autophagy. Finally, we show that chemical and genetic inhibition of autophagy partially rescues the defects in raptor mutants.

Rptor mutant mice implant at embryonic day (E) 6.5 but die shortly thereafter, before E8.5-e9.5 (Guertin et al., 2006). Here, we show that raptor mutant zebrafish begin to die off at 6 dpf, with no mutants alive by 10 dpf. We believe it is unlikely that Raptor is dispensable for early embryogenesis in zebrafish but not mouse. We postulate that this increase in viability is due to maternally provided raptor mRNA (Harvey et al., 2013), although germline transplantation would be required to definitively demonstrate this. Others have shown, and we have confirmed here, that raptor and other mTOR pathway genes are maternally provided in zebrafish (Harvey et al., 2013). Although Rptor is also maternally provided in mouse (Zhang et al., 2022), clearance of maternal mRNAs occurs very early in mouse, by the two-cell stage. In contrast, maternally provided mRNAs are largely degraded in zebrafish between 4 and 6 hpf (Zhao et al., 2017), when there are more than 1000 cells.

Less is known about the clearance of the protein products of these maternal mRNAs, which could conceivably last much longer. We first observe neural crest cell defects in 36 hpf raptor mutants. At this time, we see a 40% reduction of Raptor protein, relative to wild types. Surprisingly, we found, relative to wild types, 20% of the Raptor protein still remains in raptor mutants up to at least 4 dpf. Many zebrafish genes have a paralog, due to the additional whole-genome duplication in the teleost fish lineage. However, there is no evidence for a raptor paralog in zebrafish that could be a source of this protein, although this does not rule out the possibility. Alternatively, it is possible that Raptor is a highly stable protein, especially when it is bound within mTORC1, enabling it to persist in zebrafish until 4 dpf. These data are supported by a recent study that found Raptor protein remained in a Raptor mutant hippocampal culture line until 18 days post-deletion in vitro, at least 4 days longer than similar analyses with Tsc1 (Karalis et al., 2022). Together, these data indicate that Raptor may be a highly stable protein, especially when it is bound within mTORC1. We propose that this, combined with the relatively rapid development of zebrafish, allows raptor mutant zebrafish to survive to later developmental times. Our findings highlight the importance of taking the potential for residual maternally provided gene function into consideration when analyzing mutant phenotypes. Our findings also have implications for the interpretation of conditional mutants, as Raptor protein could persist well after recombination.

Previous work has implicated the mTORC1 pathway in craniofacial development. Roberts Syndrome is characterized by a variety of defects, including reduced limb and head size, and is caused by mutation of ESCO2, which appears to interact with the mTORC1 pathway. Human cell lines derived from individuals with Roberts Syndrome display reduced mTORC1 signaling (Xu et al., 2013). Zebrafish esco2 morphants and mutants have profound defects more severe than we see in raptor mutants. However, although esco2 appears to be maternally provided (Percival et al., 2015; Xu et al., 2013) its protein stability is not known, which makes direct comparisons with raptor difficult (Fishman et al., 2023 preprint). Craniofacial defects in esco2 morphants are alleviated by treatment with L-leucine, which activates mTORC1 (Xu et al., 2013). Similarly, disruption of the mTORC1 pathway associates with a zebrafish model of Fetal Alcohol Syndrome and L-leucine treatment also improves these facial phenotypes (McCarthy et al., 2013).

Similar to mouse conditional knockouts, we did not observe gross disruption to the pharyngeal arches early in development. However, as we note above, this should not be construed to indicate that mTORC1 function is dispensable for early neural crest cell development. Although inhibitor studies could potentially address the early function of mTORC1 in neural crest development, such studies are fraught with potential off-target effects. Genetically encoded protein degron-based systems continue to be developed (Aksenova et al., 2022). Such systems in which proteins can be conditionally degraded would provide much needed insight into early functions of Raptor in development and developing tissues.

In both mouse and zebrafish, Raptor function is necessary for proper facial growth following neural crest cell migration into the pharyngeal arches. In both species, elevated levels of neural crest cell death are observed within the pharyngeal arches (Nie et al., 2018). In mouse, growth and fusion of the facial prominences is disrupted (Nie et al., 2018). This is similar to the overall reduction in craniofacial elements that we observe in zebrafish. Furthermore, we show that this reduction of Raptor decreases the size of chondrocytes. Cell size was not examined in the neural crest conditional Raptor mutant. However, our finding is consistent with reduced chondrocyte size in the humerus of Prx1 conditional Raptor mutants (Chen and Long, 2014).

The effects of attenuating mTORC1 function are likely to be context dependent, however. For example, treatment with rapamycin reduces craniosynostosis caused by expression of constitutively active Bmpr1a (Kramer et al., 2018). Additionally, we found that atg7 mutants, in which autophagy is reduced, have craniofacial defects. These findings suggest that the overall activity of mTORC1 must be balanced to maintain normal craniofacial development.

Raptor is broadly, if not ubiquitously, expressed and is essential for functions required in all cells. Therefore, it seems odd that mutant phenotypes appear largely restricted to the craniofacial skeleton and, at least parts of, the CNS (eyes). However, similar effects are observed in mutants disrupting RNA biogenesis or ribosomopathies. Indeed, the phenotypes we observe in our raptor mutants closely parallel those in polr1c and polr1d zebrafish mutants (Noack Watt et al., 2016). Similarly, exposure to environmental toxicants, such as alcohol, would be expected to have global effects on all cells, yet cranial neural crest cells and neurons appear most sensitive to these exposures (Cartwright and Smith, 1995; Dunty et al., 2001; Kotch and Sulik, 1992). It is likely that these cell types have higher metabolic demands and, as such, are more sensitive to these types of perturbations.

Our study indicates that regulation of autophagy by mTORC1 is essential for proper neural crest cell survival. The importance of autophagy during animal development appears mixed. In mice, autophagosomes appear at the one- to four-cell stage and autophagy-deficient Atg5 oocytes fertilized with Atg5 null sperm fail to develop beyond the 8-cell stage (Tsukamoto et al., 2008). Moreover, homozygous mutation of the autophagy gene Beclin1 (Becn1) induces developmental delay by E7.5 and embryonic lethality between E7.5 and E8.5 (Yue et al., 2003). However, mouse mutants in many other autophagy genes survive to neonatal ages (Bialik et al., 2018). Moreover, autophagy is often viewed as a protective response to environmental stimuli. Indeed, it was not until the mid to late 2000s that clear evidence of autophagy-induced cell death emerged (Levine and Klionsky, 2004).

Our findings demonstrate that activation of autophagy via mutation of raptor reduces the size of craniofacial elements. Mutation of atg7, which is required for autophagosome formation, also led to reduced size of craniofacial elements in our model. This corroborates experiments in mice where they conditionally mutated essential autophagy genes Atg5 or Fip200 (Rb1cc1) via an Osterix:Cre line that primarily targets osteoblasts, osteocytes and hypertrophic chondrocytes (Thomas et al., 2019). Mutants had reduced mass of neural crest-derived and mesoderm-derived bones of the cranial vault and cranial base. Moreover, a SNP in autophagy gene HIF1A was shown to associate with an increased risk of non-syndromic cleft lip with or without cleft palate (Lou et al., 2020). Together, these results suggest that disturbance of autophagy in either direction impairs craniofacial morphology.

Surprisingly, inhibition of autophagy via chloroquine exposure had no obvious effect on craniofacial morphology in wild-type fish. This apparently conflicts with our data demonstrating that atg7 mutants have reduced craniofacial element size. This discrepancy could be explained by the fact that we began chloroquine exposure at 6 hpf, which coincides with the onset of gastrulation. Moreover, it is likely that our dose of chloroquine did not inhibit autophagy to the same degree as atg7 mutation as this gene is required for autophagosome formation.

Our data show that mTORC1 inhibition induces autophagy and increases the number of TUNEL-positive neural crest cells. Chemical inhibition of autophagy reduced the number of TUNEL-positive cells in the raptor mutants compared with wild-type or heterozygous fish. Our results strongly support a model whereby mTORC1 inhibition induces autophagy that, in turn, induces neural crest cell death leading to craniofacial defects. Combining transgenics that label apoptotic cells (van Ham et al., 2010) with transgenics designed to measure autophagic flux (Kaizuka et al., 2016) would be useful to define the level of disruption to autophagic flux that leads to apoptosis. Such analyses could help define the level of autophagy that is tolerated by specific cell types.

Collectively, our results add substantially to our understanding of autophagy during development. Our work establishes a direct link between the role of mTORC1 in regulating autophagy to its crucial function in craniofacial development. Furthermore, our findings demonstrate that the regulation of autophagy by mTORC1 is upstream of its role in mediating cell death. In summary, we demonstrate that the levels of autophagy and mTORC1 activity must be finely balanced for proper craniofacial development.

Zebrafish (Danio rerio) husbandry and embryo collection were conducted in accordance with the recommendations in The Zebrafish Book (Westerfield, 2007) and the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All zebrafish were housed at the University of Texas at Austin under IACUC-approved conditions. Developmental staging of embryos was determined as previously characterized (Kimmel et al., 1995). All zebrafish were maintained on an AB-derived genetic background.

We used the ZiFiT Targeter (http://zifit.partners.org/ZiFiT/) to identify gRNA binding sites for raptor (rptor). With these target sites, we made raptor gRNA using the MEGAscript T7 Kit (Thermo Fischer Scientific, AM1333). The oligo used to generate the gRNA is aattaatacgactcactataGGCAGAGGCATCTGAGCTAGgttttagagctagaaatagc (capitalized nucleotides code for the gRNA). Embryos were injected with a 2 nl bolus of a cocktail containing: 500 ng/μl Cas9 protein (IDT) and 250 ng/μl raptor gRNA in water and Phenol Red (to visualize the injection). A F0 fish found to be carrying a germline 4 bp mutation in raptor was backcrossed to AB to establish the raptor^au93 mutant line.

DNA was extracted from zebrafish tails or whole embryos and genotyped for raptor using high-resolution melt analysis (Samarut et al., 2016). Primers for high-resolution melt analyses were forward (GGCAGCAATAAACCCAAACC) and reverse (TTCATCGAGGGAGGCAGA). The resulting product is 57 bp in the wild-type allele and 53 bp in the mutant allele. Primers for PCR were forward (CAGCAACAGTAGCAACAGTAACATCAACAGC) and reverse (TCAGCAGGGAGGTCAGGATTC). The resulting product in the wild-type allele is 295 bp and 291 in the mutant allele.

The zebrafish mutant line atg7^sa1476 was purchased from the Zebrafish International Resource Center. Genotyping for atg7^sa1476 was performed using forward (TGCTGTTTGAGTACTGTTCAGA) and reverse (CAAGGTATGGTGCTGCTGTA) primers followed by digest using the restriction enzyme AFLII (New England Biolabs, R0520S), resulting in 363 bp in the wild-type allele and 218/145 bp in the mutant allele. Transgenic lines TG(fli1:EGFP)^y1 (Lawson and Weinstein, 2002) and Tg(CMV:EGFP-map1lc3b) ^zf115 (generously provided by Dr Jeff Bronstein, UCLA, USA) are referred to as fli1:EGFP and CMV:EGFP-lc3 for simplicity.

Total RNA was extracted using the illustra RNAspin mini RNA isolation kit (GE Healthcare, 25-0500-71). Embryos and larvae were homogenized with a motorized pestle (Argos, A0001). Samples were purified with an RNA Clean & Concentrate kit (Zymo, R1014). A NanoDrop microvolume spectrophotometer (Thermo Fischer Scientific) was used to determine the concentration of each sample. We generated cDNA using SuperScript First-Strand Synthesis System for RT-PCR (Thermo Fischer Scientific, 11904018). To amplify both alleles, we used the forward (GGCAGCAATAAACCCAAACC) and reverse (GCTTTGCACTCTTCTGCATAC) primers; to selectively amplify the wild-type allele we used the forward (ACCACCCACTAGCTCAGAT) and reverse (GCTTTGCACTCTTCTGCATAC) primers; and to selectively amplify the mutant allele we used the forward (CCACCCACTCAGATGCC) and reverse (GCTTTGCACTCTTCTGCATAC) primers. Maternally provided gene products (Fig. S3) were amplified using the following primers: castor1 forward (AAATTACACTGTTGTCCTCGAC), castor1 reverse (TGGAAGCATTGCCATTCG), depdc5 forward (TCACCAAACAGCGCATGA), depdc5 reverse (CCAGTGAGGCAGGTTGTAAT), raptor forward (GGCAGCAATAAACCCAAACC), raptor reverse (GCTTTGCACTCTTCTGCATAC), sesn2 forward (CAGAAGGACGTGCTGATCATA), sesn2 reverse (CTCCATGATAAGGGCAGTGG), tsc2 forward (CTTCTCCAACTTCTCTGCTCTG), tsc2 reverse (GGTCACTAACTTATTCCCAACCA), wdr24 forward (TCTTCACCGAACACAAGCG) and wdr24 reverse (GCACTTCATGAATCCGTCCT).

Immunoblotting was performed according to Philipp et al. (2008). In brief, embryos were dechorionated, euthanized and deyolked, and heads were separated from tails. The tails were used for genotyping. The heads were lysed in RIPA buffer (Sigma-Aldrich, R0278) containing 10% 2-mercaptoethanol (Gibco, 31350010) and 1% inhibitor cocktail (Sigma-Aldrich, P5726; Sigma-Aldrich, P2714). Heads of the same genotype were pooled and homogenized twice, for 30 s each time, with a hand pestle mortar mixer (Argos, A0001). Samples were centrifuged at 15,000 g for 10 min at 4°C. Protein concentration was determined using a BSA standard (Sigma-Aldrich, A4503), Protein assay dye reagent concentrate (Bio-Rad, 5000006) and Cytation 5 fluorescent plate reader (Agilent). Immunoblotting was performed using the primary antibodies against Raptor (Cell Signaling, 2280), p70 S6K (Cell Signaling, 2708), Phospho-p70 S6K (Cell Signaling, 2905), rpS6 (Cell Signaling, 2217), Phospho-rpS6 (Cell Signaling, 4857), 4E-BP1 (Cell Signaling, 9644), Phospho-4E-BP1 (Cell Signaling, 2805), Lc3B (Novus Biologicals 600-1384) and HRP secondary antibody (Cell Signaling, 7074), all diluted 1:1000 in TBST (20 mM Tris 7.5 pH, 150 mM NaCl, 0.1% Tween-20). The Gapdh loading control membranes were stripped and stained using anti-S6 and anti-p-S6. ImageJ/FIJI was used to process images and quantify immunoblots.

At 4 dpf, zebrafish embryos were stained with Alcian Blue and Alizarin Red to detect cartilage and bone, respectively (Walker and Kimmel, 2007). Full body images were taken on an Olympus szx7 stereomicroscope. Images of zebrafish heads were taken with a Zeiss Axio Imager-A1 microscope and edited using Gimp (Version 2.10). Linear measurements of craniofacial elements were carried out using ImageJ. All measurements were performed blinded to genotype and treatment. Eye diameter was measured as the widest distance between two points for each eye, from a ventral view. For body length normalization, facial element size was divided by body length to derive a ratio for each individual zebrafish. Melanocyte quantification was carried out using lateral images of a 70×350 μm region immediately above the yolk extension in 30 hpf embryos. A t-test was performed on data counts of wild type (n=15) and mutant (n=11). Trabeculae cells were randomly chosen for area and circumference measurements using a random number generator (random.org). Fluorescent embryos were imaged on a Zeiss LSM 980 confocal using Zen software and analyzed using ImageJ/FIJI. Graphs were produced with RStudio, Graphpad Prism 9.5.1 and Microsoft PowerPoint.

Zebrafish embryos were fixed and prepared for terminal deoxynucleotidyl transferase dUTP nick end-labeling (TUNEL) staining following published protocols (Wilfinger et al., 2013). The staining was performed using the TUNEL Assay Apoptosis Detection Kit, CF594 (Biotium, 30064) according to the manufacturer's instructions.

Autophagy was measured via transgenics as previously described (He and Klionsky, 2010). Briefly, CMV:EGFP-lc3 embryos were treated with 10 μM LysoTracker Red DND-99 (Thermo Fischer Scientific, L7528) for 45 min and then were fixed overnight. Autophagy was measured by counting overlapping puncta from CMV:EGFP-lc3 and Lysotracker red. The oral ectoderm and pharyngeal pouches were used to define the extent of the pharyngeal arches.

Stock solutions of Chloroquine diphosphate salt [Sigma-Aldrich, C6628 (CAS Number: 50-63-5; PubChem Substance ID: 24278090)] were prepared in DMSO and stored at 100 mM at −20°C until use. Zebrafish embryos were exposed to 1 mM from 6 hpf to 4 dpf. Caspase 3 inhibitor 2 (EMD Millipore, 264155-250UG) was prepared as previously described and used at 25 μM (McCarthy et al., 2013).

One-way ANOVA with Tukey's HSD was used to compare between genotypes for qPCR, immunoblots, linear measurements, TUNEL stains and fluorescent autophagy measurements. A paired t-test was used to compare area, circumference and cell count between wild type and mutants (Fig. 5A-C).

Fluorescent whole-mount in situ hybridization was performed according to a Molecular Instruments Protocol based on (Choi et al., 2018). All probes were designed by Molecular Instruments.

We thank all of the fish facility technicians, especially Nika Sarraf and Adriana van Zaten, whose expert animal husbandry was essential for this work. We also thank Shuga Sun for research and technical support. We are grateful to the undergraduate researchers Lucas Kaiser and Thomas Le for their work. Finally, we thank Josh Everson for editing the manuscript.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.202216.reviewer-comments.pdf