Nonequivalence of Zfp423 premature termination codons in mice

Abstract

Genetic variants that introduce a premature termination codon (PTC) are often assumed equivalent and functionally null. Exceptions depend on the specific architectures of the affected mRNA and protein. Here we address phenotypic differences among early truncating variants of mouse Zfp423, whose phenotypes resemble Joubert Syndrome and Related Disorders (JSRD). We replicate quantitative differences previously seen between presumptive null PTC variants based on their position in the coding sequence. We show with reciprocal congenic strains that large phenotype differences between two PTC variants with the same predicted stop and reinitiation codons is due to the specific allele rather than different strain backgrounds, with no evidence for induced exon skipping. Differences in RNA structure, however, could influence translation rate across the affected exon. Using a reporter assay, we find differences in translational reinitiation between two deletion variants that corelate with predicted RNA structure rather than distance from the canonical initiation codon. These results confirm and extend earlier evidence for differences among Zfp423 PTC variants, identify parameters for translational reinitiation after an early termination codon, and reinforce caution in the null interpretation of early PTC variants.

Figure 1.. Unequal impacts of early vs. late premature termination codons among variants with protein levels below detection.

(A) Zfp423 gene structure showing constitutive exons 3 and 4. Adapted from UCSC genome browser. (B) Schematic of ZFP423 primary protein structure shows known binding domains (colored lines above) relative to C2H2 zinc fingers (lighter gray) above PTC variants, with relative lengths of shifted reading frames (pink bars), and exon 3 deletion allele. (C) Surface and coronal block face images with lines to illustrate the vermis width, cerebellum hemisphere anterior-posterior length (top); cortical thickness at 45°, 30°, and 15° and thickness of (cc) corpus callosum and (ac) anterior commissure (bottom). Adapted from (DESHPANDE et al. 2020). (D) Replotted data from Deshpande et al. showed less severe impacts of early compared with later PTC variants for body weight, cerebellum sagittal area at midline, and cortical area from dorsal view in the original data set. (E) Data from new paired samples replicate non-equivalent effects on weight and cortical area. Lines with numbers show comparisons with significant (p<0.05) differences between groups by Tukey’s honest significant differences post-hoc test after significant ANOVA for the three groups. Tests with p<0.0055 were considered to survive Bonferroni correction for the nine anatomical measures used.

Figure 2.. PTC variants early in the Zfp423 open reading frame.

(A) DNA sequences of three frame-shifting deletions near the 3’ end of exon 3. Single-letter amino acid translation shown above sense strand DNA sequence. Deleted bases indicated by thick black lines for each allele. Termination codons (underlined) and potential in-frame reinitiation codon (shaded box) within 5’ end of exon 4 are indicated, as are the predicted molecular weights of the potential 5’ translation products. Exon 3 splice donor and exon 4 splice acceptor sites are indicated with lower case. A frame-shifted arginine codon, AGA, that would be decoded by the n-Tr20_UCU tRNA (arrowhead) spans exons 3 and 4. (B) The C57BL/6J-specific modifier allele of n-Tr20_UCU does not have a major impact on D78Vfs*6 phenotype measures. (C) Extent of FVB (grey) and C57BL/6J (black) on chromosome 8 for reciprocal congenic strains. (D) FVB background does not make D78Vfs*6 more severe or comparable to H96Wfs*4 in brain measures. (E) Sample images from control littermate and exon 3 PTC variants on C57BL/6J (B6) and FVB/NJ congenic or coisogenic strain backgrounds. (F) H88Wfs*4 has lower survival frequency than D78Vfs*6 on either strain background, including complete lethality in C57BL/6J, comparable to other null alleles (ALCARAZ et al. 2011; ALCARAZ et al. 2020).

Figure 3.. Lack of exon skipping around exon 3 PTC variants.

(A) RT-PCR assay for exon 3 inclusion shows size variation due to deleted nucleotides but no evidence for exon skipping in comparison to an exon 3 deletion (Δ345). Similar results were obtained with two distinct primer sets. (B) Titration of Δ345 into a wild-type control shows RT-PCR detection sensitivity to ~0.5% exon skipping. Assay run using the same master mix, PCR plate, and gel as (A). Left and right dilution series from different RNA samples.

Figure 4.. RNA stem loop before the frame-shifted stop codon differentiates exon 3 alleles.

(A) Exon 3 deletion alleles relative to RNA sequence with a predicted stem loop. Vertical line, exon 3-exon 4 junction. Arrows indicate imperfect inverted repeat, with predicted paired bases shown in red (left arm) and cyan (right arm). (B) Predicted optimal structures of the mouse and identical human reference allele, H96Vfs*29 (Δ7), and H96Wfs*4 (Δ11), retaining the color coding from (A). (C) Predicted free energy (ΔG) for the optimal conformer (min.) and ensemble of conformer structures for each allele.

Figure 5.. Luciferase reporter assay.

(A) Dual luciferase reporter construct. EF1α promoter drives expression of Zfp423 5’ UTR and coding sequence through codon G128, fused to a Nano luciferase coding sequence (NanoLuc) followed by the Gapdh 3’ end with an additional polyadenylation signal. A Pgk1 promoter drives expression of an Nxf1 5’ UTR and Firefly luciferase coding sequence (Firefly Luc) followed by an SV40-T 3’UTR and polyadenylation signal. Schematics below illustrate the D78Vfs*6 (Δ59) and H96Vfs*4 (Δ11) variants, show the short altered reading frame (pink) and distance (line) between shifted stop codon and the next in-frame AUG at M126. (B) Schematic illustration of tested variants relative to the predicted stem loop in reference sequence. Right, computed ensemble ΔG (RNAfold) for just the sequence shown, excluding deletions. Deleted positions in light gray, base substitutions in red. (C) Relative response ratio (see methods) for the Nano luciferase reporter relative to the Firefly luciferase for each construct. Each dot represents a separate transfection.