Patients with CRIM1 deletions and pathogenic variants were ascertained by GeneMatcher (Sobreira et al., 2015). Phenotypic and genotypic details were collected in a spreadsheet. Consent was obtained for the publication of anonymous clinical details for each participant.

All animal experiments were performed under protocols approved by the Institute for Animal Care and Use Committee (IACUC) at the University of California, San Francisco, and the IACUC committee at Cincinnati Children’s Hospital Medical Center (IACUC #2022-0040). EKW strain zebrafish were maintained at 28°C on a 14-h (h) light/10-h dark cycle. We used a modification of the previous CRISPR methodology to target the single crim1 gene (ENSDART00000050534.6) in Danio rerio (Talbot and Amacher, 2014). We designed a small guide (sg)RNA targeting crim1 prior to the coding sequence for the von Willebrand factor type C domains, antistasin-like domains, and the cysteine-rich motor neuron 1 protein domain (Supplementary Table S3). We injected zebrafish eggs at the one-cell stage and used Sanger sequencing in batches of 10–20 larvae at 48 hpf to assess the success of the injections. F0 larvae with insertion/deletion (indel) variants were raised and genotyped at 6–8 weeks by fin clipping after inducing anesthesia by immersing the fish in 0.02% tricaine methanesulfonate (FINQUEL or Tricaine-S) until gill movement was slowed. DNA was extracted from the clipped tail fin and Sanger-sequenced to assess successful gene targeting (for genotyping primers, see Supplementary Table S4). We selected an F0 founder with a 2 base pair (bp) deletion, c.339_340delCT p.Leu112Leufs*3 that introduced a frameshift and predicted premature protein truncation and nonsense-mediated decay for the crim1 transcript (for Sanger sequencing traces, see Supplementary Figure S1). After outcrossing this F0 founder, F1 progeny were raised and genotyped using DNA obtained from fin-clipping, as mentioned above. Heterozygous F1 adults were in-crossed to obtain F2 zebrafish that were homozygous (HZ) for the 2 bp deletion in crim1. These HZ mutant larvae, henceforth referred to as crim1 ^−/− larvae, were viable. We raised and bred crim1 ^−/− larvae that appeared normal, without any eye defects, for our experiments. We also examined eye morphology in heterozygous crim1 ^+/− larvae. Our control fish were derived from the EKW strain or the NBT:DsRed strain that labels neurons under the neural ß-tubulin promoter and is not known to have any eye defects (Mazzolini et al., 2018).

We examined the crim1 ^−/− larvae, crim1 ^+/− larvae, and controls at 1–6 days post fertilization (dpf) for the ocular and lens findings previously observed in patients with MACOM and prior models of Crim1 and crim1 loss of function. We measured eye and lens size, head size, and body length compared to control larvae of the same age and developmental stage. Larvae with bent tails were measured along the body axis and the curved tail to capture the full length of the larva. We examined eye and lens morphology with hematoxylin and eosin (H&E) staining and immunohistochemistry (IHC) and used light microscopy to visualize the LE cells, LF cells, and retinal cell layers according to previously described methods (Krall et al., 2018; see Supplementary Table S4 for antibodies used and Supplementary Data for Method). Larvae photographed with differential interference contrast (DIC) microscopy at 3 dpf were treated with propylthiouracil after fertilization according to standard methods (Elsalini and Rohr, 2003). We used rabbit anti-phospho-histone H3 at 1–400 dilution (PH3; #9701S, Cell Signaling, Danvers, MA) with anti-rabbit Alexa Fluor 488 (Invitrogen, Grand Island, New York) for PH3 staining. We scored two sections per eye for PH3+ cells. At 48 hpf, we scored 7 control larvae, 5 crim1 ^+/− larvae, and 4 crim1 ^−/− larvae. At 72 hpf, we scored 9 control larvae, 4 crim1 ^+/− larvae, and 8 crim1 ^−/− larvae.

For BrdU labeling, 48 hpf and 72 hpf larvae were incubated in egg water containing 10 mM 5-bromo-2′-deoxyuridine (BrdU)/15% dimethyl sulfoxide (DMSO) solution (Sigma-Aldrich, St. Louis, MO) on ice for 30 min. Post-incubation, larvae were rinsed with fresh egg water and then placed at 28°C for 1 h (Janssens et al., 2013). BrdU-labeled larvae were euthanized and fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS). BrdU-labeled sections were treated with 2N hydrochloric acid (HCl)/0.5% Triton/PBS for 60 min prior to incubation with mouse anti-BrdU (#B2531, 1-400, Sigma-Aldrich), followed by conjugation with anti-mouse Alexa 488 (Invitrogen, Carlsbad, CA). At least two sections per eye and both eyes were counted manually for BrdU + cells using ImageJ software.

We performed ‘bulk’ RNA-seq experiments in dissected eyes from in-crossed crim1 ^−/− larvae and controls at 72 hpf according to the previously described methods (Krall et al., 2018). RNA quality and quantity were reviewed using a bioanalyzer (2100 Bioanalyzer, Agilent), and libraries were prepared with an Ovation^® RNA-Seq System V2 Kit (NuGEN). With this method, total RNA was reverse transcribed to synthesize first-strand cDNA using a combination of random hexamers and a poly-T chimeric primer (Krall et al., 2018). The RNA template was then partially degraded by heating, and the second-strand cDNA was synthesized using DNA polymerase. The double-stranded DNA was then amplified using single primer isothermal amplification (SPIA). SPIA is a linear cDNA amplification process in which RNase H degrades RNA in DNA/RNA heteroduplex at the 5′-end of the double-stranded DNA, after which the SPIA primer binds to the cDNA and the polymerase starts replication at the 3′-end of the primer by displacement of the existing forward strand (Krall et al., 2018). Random hexamers were then used to linearly amplify the second-strand cDNA. Finally, libraries from the SPIA-amplified cDNA were prepared using the Ultralow DR Library Kit (NuGEN). The RNA-seq libraries were analyzed for quality and primer dimers using a bioanalyzer and quantified by quantitative polymerase chain reaction (qPCR) prior to sequencing (KAPA Library Quantification Kit). High-throughput sequencing was conducted using single-end 50 lane(s) on a HiSeq 4000 instrument (Illumina).

The quality of raw single-end reads of the RNA-seq dataset was tested by FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/) to evaluate the per-base sequence quality, quality scores, sequence length distribution, and overrepresented adapter/primer sequences. After quality control with FastQC, data trimming and clipping were performed using BBDuk and aligned against the zebrafish reference genome (GRCz11) using STAR. Alignment results were then assessed to check the quality with Qualimap. RSEM was used to quantify genes. Finally, DESeq2 was used to conduct a downstream analysis for differentially expressed genes. Statistically significant differentially expressed genes (DEGs) with log2 (fold change) greater than ±0.5, an adjusted p-value ≤0.05, and the mean fragments per kilobase of exon per million fragments mapped (FPKM) ≥1 in all datasets were identified by comparing crim1 ^−/− mutants with control datasets using an in-house Python script. Principal component analysis (PCA) was performed on global gene expression to determine the relatedness between the individual replicates of wildtype and mutant datasets.

We designed gene-specific, reverse transcriptase quantitative polymerase chain reaction (RT-qPCR) primers flanking at least one intron for the top-ranking up- and down-dysregulated genes in the bulk RNA-seq experiments (for primers, see Supplementary Table S5). Each RT-qPCR was run in triplicate with melt curve analysis showing a single peak. All experiments were performed on at least three biological replicates of control and crim1 ^−/− larvae, following the previously described methods (Krall et al., 2018). We generated a time course of crim1 expression from 24 to 72 hpf. Eyes were dissected from zebrafish larvae into cold, 1x PBS at 72 hpf. A total of 40 pairs of eyes were pooled in a 1.5-mL microfuge tube, spun, and snap-frozen at −80°C after PBS removal. Total RNA was extracted using the PureLink RNA Mini Kit (Invitrogen, Carlsbad, CA). RNA was treated with Ambion™ DNase I (RNase-free, Invitrogen, Carlsbad, CA) to remove impurities. cDNA was synthesized from 1.5 µg of total RNA with oligo(dT)₂₀ priming using SuperScript III Reverse Transcriptase (Invitrogen, Carlsbad, CA). Reactions were run on an ABI Prism 7500 Sequence Detection System, and data were analyzed according to the ΔΔCt method using eef1a1l1 (NM_131263.1) as an internal control according to previously described methods (Wu et al., 2015).

Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) staining was performed on cryosections of controls and crim ^−/− larvae from 48 to 72 hpf, according to the manufacturer’s protocol (1168479510, Roche, Indianapolis, IN). We examined a minimum of four larvae from each experimental group and analyzed 2–4 non-consecutive mid-eye sections per eye at each timepoint. For apoptotic cleaved caspase-3, we stained cryosections from controls and crim ^−/− larvae at 24, 30, 48, and 72 hpf. A minimum of four larvae from each experimental group were analyzed using 2–4 non-consecutive sections from the middle of each eye for each timepoint. Sections were stained and counted using the count tool in ImageJ software. The number of TUNEL-positive or cleaved caspase-3-positive cells was divided by the total number of nuclei labeled with 4′,6-diamidino-2phenylindole (DAPI) and plotted against age. After calculating the mean ± standard error (SE), groups with statistically significant differences compared to controls were determined using a Student’s t-test.

Fifty whole larvae were collected from controls and crim1 ^−/− larvae at 12 and 24 hpf. Additionally, 100 whole eyes were collected from controls and crim1 ^−/− larvae at 48 and 72 hpf. These eye tissues were rapidly frozen on dry ice, then thawed, and mechanically homogenized. The homogenates were centrifuged at 8,000 relative centrifugal force (rcf) at 4°C. The resulting supernatants containing soluble proteins were carefully collected. These supernatants were then used for quantifying C1q levels using an enzyme-linked immunosorbent assay (ELISA), following the protocols provided by the manufacturer (MBS023066, MyBioSource, San Diego, CA). Optical density at 450 nm was read using a Bio-Tek Microplate Reader System with Gen5 software (Bio-Tek, Santa Clara, CA).

Fifty heads were collected from control and crim1 ^−/− larvae at 72 hpf. Whole heads were homogenized, solubilized, and washed (Michel and Miller, 2023). Supernatants were pre-cleared, and the protein lysates were measured using a Direct Detect Spectrometer system (DDHW00010, Millipore Sigma, Burlington, MA, United States). Lysates were incubated with recombinant anti-Crim1 (AB5699, 1:200, Sigma-Aldrich, St. Louis, MO) and anti-β1-integrin (Itgβ1; clone MB1.2, MAB1997, 1:50, Sigma-Aldrich, St. Louis, MO) at 2 µg of antibody overnight at 4°C. After washing, the supernatants were used for Western blot analysis according to standard methods. We loaded 20 µL of 40 µg of immunoprecipitants and 20 µL of 4 µg of input using 4%–20% Mini-PROTEAN^® TGX Stain-Free™ Protein Gels (Bio-Rad, Hercules, CA).

Statistical analysis was performed using t-tests according to standard methods. We used a significance level of P-value < 0.05.

The clinical and genetic data from patients with CRIM1 variants predicting haploinsufficiency are summarized in Table 1 and presented in full in Supplementary Table S6. We ascertained a 48-year-old female with bilateral cornea plana and microcornea, bilateral cataracts (right eye, dense nuclear and partially subluxated; left eye, surgically removed intraocular lens, aphakic), bilateral iris colobomas, and a coloboma of the retina in the left eye, with bilateral retinal detachments. Due to significant corneal band keratopathy and dense media opacity in the right eye, examination of the optic nerve and posterior pole of the retina could not to be performed. Her aphakic left eye distance visual acuity was 20/80 with a +7.00 + 2.00 × 81 refraction. Her right eye had no light perception. Other findings included mild, low-frequency sensorineural hearing loss that normalized in the mid frequencies and sloped to a mild-to-moderate loss at high frequencies bilaterally. A renal ultrasound (US) scan showed mild, focal caliectasis/hydronephrosis of the right kidney at the mid to lower pole of the collecting system and diffuse mild left hydronephrosis. The patient and her father had a heterozygous, 39,942 bp deletion encompassing exons 14 to 17 and the 3′ UTR region of CRIM1. The deletion also included exons 6 to 8 of Fasciculation and Elongation protein, Zygin/zeta-2 (FEZ2; OMIM #62486), which has not yet been associated with a phenotype in humans.

A second family of two generations comprised a 12-year-old male with myopia and his father. The boy had an increased axial length and vision loss from optic atrophy in both eyes, with temporal field loss in the right eye and left total inferior temporal field loss. The optic nerves were cupped with 3+ optic atrophy of the neuro-retinal rim in the right eye. The left optic disc was normal, with pallor at the temporal neuro-retinal rim. Optical coherence tomography (OCT) showed a severely thin retinal nerve fiber layer in his right eye and a mildly thin retinal nerve fiber layer in his left eye. Other medical findings comprised febrile seizures at 2 years of age, testicular torsion treated with orchidopexy, an arachnoid cyst, and a Chiari malformation that reportedly resolved on subsequent magnetic resonance imaging (MRI) of the brain. His milestones and development were normal, except for requiring 6 months of speech therapy at the age of 2 years. His father was myopic and wore glasses from 10 years of age and was diagnosed with corneal pannus, cataract at a young age, and visual field loss. The father’s examination showed a similar optic nerve appearance to his son and bilateral, diffuse optic atrophy with cupping. Both father and son had a single-nucleotide deletion in CRIM1, c.2701delC p.Leu901Cysfs*22 (NM_016441.3). This variant affects exon 15 out of 17 exons in the longest isoform of CRIM1, which is the only known coding transcript. The variant is not present in the 1000 Genomes Project but was found in one individual in gnomAD v4.1 (1/1,424,618 alleles for an allele frequency of 7.02e⁻⁷; no homozygous allele) and predicts premature protein truncation and nonsense-mediated decay (MutationTaster; Schwarz et al., 2010). The nucleotide variant also increases an acceptor splice site (wildtype 0.21 mutant 0.27; exon gtgg/CCCA intron; MutationTaster; Schwarz et al., 2010).

We in-crossed male and female zebrafish that were HZ for the 2bp deletion in crim1. At 48 hpf, the crim1 ^−/− larvae showed increased numbers of malformations and shortened body lengths compared to controls (Figure 1A). Curved and shortened body axes and shortened or deformed tails were observed in 81.3% of crim1 ^−/− larvae compared to 3.2% of controls (Figure 1A). At 24 hpf, there were no obvious external differences between the controls, heterozygous crim1 ^+/− larvae, and crim1 ^−/− larvae (Figure 1B). However, by 48 hpf, the crim1 ^−/− mutants demonstrated tail deformities (Figure 1C, panel iii), and at 72 hpf, the crim1 ^−/− mutants had curved tails similar to prior MO models of crim1 loss of function (Figure 1C, panel vi; Kinna et al., 2006). Heterozygotes appeared similar to controls at 48 hpf (Figure 1C, panel ii) and 72 hpf (Figure 1C, panel v), and for this reason, we did not quantify the malformations or study the heterozygotes further. Approximately 18% of the HZ crim1 ^−/− larvae retained a normal external appearance and ocular appearance (Figure 1A).

We measured the longest axis for eye diameter, head diameter, and body length in the crim1 ^−/− larvae and controls (see Figures 2A–C for measurements obtained). We found a significant decrease in the eye size in the crim1 ^−/− larvae at 48 hpf and 72 hpf (Figure 2D; P-value for 48 hpf = 0.003 and P-value for 72 hpf = 0.002). This decrease in the eye size resembles microphthalmia observed in the mouse models of Crim1 loss of function (Supplementary Table S1; Chiu et al., 2012; Fan et al., 2014; Zhang et al., 2016). In contrast, the ratio of eye diameter to body length was significantly increased in the crim1 ^−/− larvae at 48 hpf and 72 hpf (P-value for 48 hpf = 1.71e⁻⁰⁶ and P-value for 72 hpf = 2.65e⁻⁰⁵), consistent with a greater reduction in the body length than in the eye size (Figure 2E). Finally, the ratio of eye diameter to head diameter was increased in the crim1 ^−/− larvae compared to controls at 48 hpf but not at 24 hpf or 72 hpf (Figure 2F).

We used cryosections with H&E staining and light microscopy to examine eye and lens morphology in the crim1 ^−/− larvae compared to controls at 3, 4, and 5 dpf (Figure 3). We also used differential interference contrast (DIC) microscopy to examine lens morphology in the crim1 ^−/− mutant larvae compared to controls at 3 dpf (Figure 4). The lenses from crim1 ^−/− mutant larvae appeared smaller in size compared to controls (Figure 3, panels A and E), and this difference in eye size persisted at 4 dpf and 5 dpf (Figure 3, panels B, C, and D and F, G, and H). The lens morphology also appeared to be irregular in crim1 ^−/− mutant larvae compared with the typical spherical shape of control lenses (Figure 3, panel F compared to panel B and panel H compared to panel D). Examining the lenses with DIC microscopy showed that controls had a regular appearance of the lens in 8/8 larvae (Figure 4, panels A and B) compared to 3/14 crim1 ^−/− mutant larvae that had an irregular, roughened appearance of the lens (Figure 4, panels D-F). However, some of the crim1 ^−/− mutant larvae retained a normal appearance of the lens (Figure 4, panel C). The quantification of lens diameter for 16 lenses in controls and crim1 ^−/− mutant larvae showed that crim1 ^−/− larvae had significantly smaller lenses than controls (Figure 4, panel G, showing the method of measurement, and panel H, showing lens diameters; P-value = 0.036).

We then examined various ocular cell types for differences between the crim1 ^−/− mutant larvae and controls. Staining with the zl-1 antibody that binds to differentiating LF cells (Greiling et al., 2010) showed irregular and amorphous-appearing lenses in the crim1 ^−/− larvae compared to controls (Figure 5, panels A and B). Staining of the corneal epithelium with the E-cadherin antibody that binds to the corneal epithelium was similar between controls and crim1 ^−/− larvae (Figure 5, panels C and D). Retinal lamination was grossly normal when examined using the zn-5 antibody that binds to amacrine cells and retinal ganglion cells in the retinal ganglion cell layer (Figure 5, panels E and F), but there was a significant increase in thickness of the retinal ganglion cell layer in the crim1 ^−/− larvae compared to controls (Supplementary Figure S2; P-value = 0.00034). There was no obvious difference between controls and crim1 ^−/− mutant larvae using the zpr-1 antibody that binds to photoreceptor cells (Figure 4, panels G and H). We did not observe colobomas in the crim1 ^−/− larvae.

We reasoned that the smaller eye size in the crim1 ^−/− mutant larvae could be due to reduced cell proliferation. Staining with PH3 did not show any difference between crim1 ^+/− heterozygous larvae and controls and between crim1 ^−/− mutant larvae compared and controls (Figure 6, panels A–F; Figure 7A). There was a significant reduction in staining in crim1 ^−/− mutant larvae compared to controls at 48 hpf (P = 0.013; Figure 6 panels G–I; Figure 7B), consistent with reduced cellular proliferation, but this difference was not statistically significant for heterozygous crim ^+/− larvae compared to controls at 48 hpf or between crim1 ^−/− larvae compared to controls at 72 hpf.

A three-dimensional PCA plot confirmed the separation of gene expression in the crim1 ^−/− samples compared to the control samples (Figure 8A). There were 223 genes that were significantly dysregulated (adjusted P-value < 0.05), with 109 upregulated genes and 114 downregulated genes (Supplementary Table S7). A heatmap showing the separation of the DEGs in control and crim1 ^−/− larvae is shown in Figure 8C, and a volcano plot showing the distribution of dysregulated genes is shown in Figure 8B. The top 16 downregulated and upregulated DEGs and transcripts from the bulk RNA-seq experiments comparing gene expression from crim1 ^−/− larval eyes with those of controls at 72 hpf are shown in Table 2. crim1 expression was downregulated in the crim1 ^−/− larvae compared to controls with an adjusted P-value of 1.72e⁻¹¹ (Table 2), consistent with the predicted decrease in crim1 expression after CRISPR targeting of this gene. This result was confirmed with RT-qPCR at 24, 48, and 72 hpf (see Figure 9A; P-value = 0.04; Supplementary Figure S3). Chloride intracellular channel 4 (clic4) was the most significantly downregulated gene, with an adjusted P-value of 2.77e⁻⁴¹ (Table 2). RT-qPCR also confirmed reduced expression for this gene in RNA from crim1 ^−/− mutant eyes compared to controls (P-value = 0.02; Figure 9A). IHC revealed mildly reduced expression of clic4 in the lens at the same timepoint (Supplementary Figure S4), but expression of this gene was still detectable in the crim1 ^−/− mutant lenses.

Crim1 recruits and co-localizes with β1-integrin at the plasma membrane and is involved in integrin recycling and β1-integrin-dependent cell adhesion (Argenzio et al., 2014). We, therefore, performed RT-qPCR to measure the expression of itgb1 encoding β1-integrin in the crim1 ^−/− mutant eyes compared to controls. We observed significant downregulation of itgb1 (P-value = 0.02; Figure 9A). We considered a model in which the reduced clic4 expression observed in the crim1 ^−/− larvae disrupts integrin-mediated cell interactions and investigated whether Crim1 can directly bind to Clic4. Immunoprecipitation showed a faint band at the expected size (29 kDa) in a lysate obtained from whole larval heads from controls at 72 hpf incubated with the recombinant Crim1 antibody and using the anti-Clic4 antibody for Western blotting (Supplementary Figure S5). However, the polyclonal nature of the Clic4 antibody resulted in non-specific bands, and our results do not demonstrate a definite interaction. We could not identify a suitable antibody for itgb1 to verify interactions between itgb1 and Clic4 or Crim1 (data not shown). There are multiple itgb1 homologs in zebrafish including integrin beta 1a (itgb1a), integrin beta 1b (itgb1b), integrin beta 1b.1 (itgb1b.1), and integrin beta 1b.2 (itgb1b.2; Mould et al., 2006; Wang et al., 2014). The Danio rerio genes itgb1a and itgb1b are closely homologous to integrin beta1 subunits present in other vertebrates, but itgb1b.1 and itgb1b.2 are unconventional beta1 subunits that have not yet been described in other vertebrate species (Wang et al., 2014).

Our bulk RNA-seq results also showed significant upregulation of fgf1b, with an adjusted P-value of 4.53e⁻¹² (Table 2). The upregulation of this gene was confirmed by RT-qPCR (P-value = 0.023; Figure 9B). IHC using an antibody for Fgf1 supported the increased expression of fgf1b in the crim1 ^−/− mutant lenses, with the distribution of signal throughout the lens rather than confined to the exterior of the lens (Supplementary Figure S6). However, it is also possible that this antibody would also detect fgf1a. Additional upregulated genes included serpinb1l2, with an adjusted P-value of 8.47e⁻⁴² (Table 1B). RT-qPCR confirmed significantly increased expression for serpinb1l2 (P-value = 0.02) in the crim1 ^−/− mutant eyes compared to controls but did not confirm significance for the other upregulated genes, such as gpib, bzw1a, and nampta (Figures 9B, C). We also examined c1q expression in our bulk RNA-seq data in light of results demonstrating the dysregulation of complement genes in HZ mice with a hypomorphic Crim1 allele compared to heterozygotes (Crim1 ^glcr11; GEO GSE62561 dataset; Zhang et al., 2023). c1q was significantly upregulated in our bulk RNA-seq data (P = 0.03; Supplementary Table S7). We, therefore, determined to verify this RNA-seq result using RT-qPCR (Figure 10) and an ELISA (Supplementary Figure S7). The RT-qPCR showed significant upregulation of c1qb (Figure 10; P = 0.018) and downregulation of c1qc (P = 0.003) and cd74b (P = 0.032). The expression of c1qa (P = 0.154) and cd74a (P = 0.391) was not significantly different. The ELISA results did not demonstrate significant dysregulation of the C1q protein in the crim1 ^−/− larvae (Supplementary Figure S7). For the Gene Ontology (GO) analysis (Supplementary Figure S8), a false discovery rate (FDR) P-value of <0.05 was used to determine significance. Only the category ‘molecular functions’ among the three categories of GO analysis (biological processes, cellular locations, and molecular functions) had significant results (FDR < 0.05). One of the significant results under this category was the ‘structural constituent of the eye lens’ (FDR was 1.27 E^-02).

Finally, we reasoned that the reduction in eye size could be associated with an increase in apoptosis in the lenses and eyes of the crim1 ^−/− larvae. Examination and quantification of apoptotic cells using cleaved caspase-3 by IHC at 24, 30, 48, and 72 hpf demonstrated a significant increase in cleaved caspase-3 cells at 24 and 72 hpf but not at 30 and 48 hpf in crim1 ^−/− larvae compared to controls (Supplementary Figures S9A, B). We also performed a TUNEL assay at 48 and 72 hpf and found no significant difference between the control and crim1 ^−/− larvae (data not shown).