Loss of stemness factor Mex3a leads to deficits in cell surface protein trafficking and translation, cilia structure, and planar cell polarity in mature neurons.

A. Immunofluorescence for Mex3a (magenta) and Calmegin (green) in Mex3a WT (left) and cKO (right) main olfactory epithelium, three-week-old littermates. Representative image from at least 5 biological replicates. DAPI in blue. Scale bar 25 µM. B. Representative immunofluorescence image for OR protein (pooled antibodies against C7, M71, and P2) in white. Staining in cilia highlighted with magenta triangles. Scale bar 25 µM. C. Violin plot quantifying OR cell body protein signal intensity per OR positive cell from Mex3a WT (left) and Mex3a cKO (right) immunofluorescence experiment described in (B). n = 1101 WT and 1278 Mex3a cKO OR positive cells from three biological replicates of four-week-old littermates. Statistics, Wilcoxon rank sum test. D. Representative Airyscan image of immunofluorescence for Arl13b in ciliary basal bodies. Scale bar, 2 µM. Magenta arrows highlight examples of basal bodies at origin of cilium projections. E. Quantification of Arl13b normalized integrated intensity within each dendritic knob using radial profile analysis. Radius of each knob normalized to one micron. n = 81 Mex3a WT dendritic knobs (blue) and 161 cKO dendritic knobs (yellow). Statistics, Fisher’s Z (1925), z = 19.313, p-value = 0.0000. F. Representative scanning electron microscopy image of OSN dendritic knob with cilium projections in Mex3a WT (left) and cKO (right). Scale bar, 1µM. G. Visual representation of basal body locations for each cilium projection over the dendritic knob for Mex3a WT (left) and cKO (right). All knob radii were normalized to 1 µM. H. Quantification of cilium projection distances from the center of the dendritic knob. n = 41 Mex3a WT dendritic knobs (blue) and 72 cKO dendritic knobs (yellow). Statistics, Wilcoxon rank sum test. I. Representative immunofluorescence image of MOE tissue Tgn46 in magenta, Mor28 OR in green, DAPI in blue. Scale bar 25 µM. J. Violin plot quantifying polarity of OSNs in Mex3a WT (left) and cKO (right). Polarity is quantified by measuring the angle between a line through the apex of Tgn46 and OR staining and the basal/apical axis of the epithelium where that cell resides (see diagram). n= 408 cells from WT, 328 from Mex3a cKO, 5 biological replicates each genotype at 4 weeks old. Statistics Wilcoxon rank sum test. K. Representative spinning disk confocal deconvolved SoRA live cell images of a HEK293T cell transiently transfected with mStayGold-UtrCH (green) and mCherry or Mex3a-mCherry (magenta). Scale bar, 10 µM. For all statistical tests in this figure: P<0.05 = *, P<0.01=**, P<0.001=***.

Mex3a knockout leads to altered differentiation outcomes in the olfactory epithelium

A. Heatmap generated with Morpheus Broad software of normalized protein group quantities for 6026 proteins (y-axis) measured by Mass Spectrometry across 7 samples (x-axis). Proteins were quantified from three replicates of OMPiresGFP+ sorted cells from Mex3a WT littermate controls, and four replicates of OMPiresGFP+ sorted cells from Mex3a cKO mice. B. Gene Ontology Analysis from 321 proteins found to be depleted from Mex3a cKO mature OSNs (LFC < -0.58). C. Gene Ontology Analysis from 165 proteins found at higher levels in Mex3a cKO mature OSNs (LFC > 0.58). D. STRING representation of proteins which exhibit a LFC (Mex3a cKO – WT) < -0.58, Student’s t-test p-value < 0.05, and raw protein counts in all 7 samples to ensure sufficient coverage. E. STRING representation of proteins that are increased in Mex3a cKO samples (LFC > 0.58), Student’s t-test p-value < 0.05, and raw protein counts in all 7 samples. Proteins with thick borders are markers of sustentacular cells as determined by scRNA-Seq from MOE (data from8).

Putative RNA targets of Mex3a uncovered by in vivo HyperTRIBE

A. Disordered domain prediction for Mex3a protein, as determined using PrDOS server115 B. Live cell imaging of HEK293T cells transiently transfected with GFP-Sec61b and Mex3a-mCherry. Images acquired with high resolution SoRA microscope and deconvolved using FIJI plugin Microvolution. Scale bar, 2 µM. C. Design of transgenic allele used to find RNA targets of Mex3a in vivo in the olfactory epithelium. The mouse Adar deaminase domain with a HyperTRIBE point mutation (E957Q) was cloned downstream of the Mex3a gene, followed by a V5 tag, t2a and mCherry. The construct is induced by a tetO with minimal CMV promoter, driven by gg8-tTA or OMP-tTA in the MOE. After sorting mCherry positive and negative cells from the MOE, RNA-Seq is conducted and A→I mutations (read out as G after sequencing) are used to infer RNA targets. D. Scatterplot comparing the number of editing sites per transcript on the y-axis (set to 100 for each for comparison) to expression level (TPM) of that transcript on the x-axis. mRNAs shown in red were found in all 6 mCherry+ RNA-Seq libraries from six different tetOMex3a-HyperTRIBE transgenic founders. While 1-2 of these mRNAs were observed in each HyperTRIBE-only sample, they were edited to a lesser extent and were not the same mRNAs across all 3 of the HyperTRIBE-only samples. E. Gene Ontology analysis of putative HyperTRIBE RNA targets of Mex3a after driving the construct in mice with the gg8-tTA driver. Targets were given a weighted score based on number of editing sites and how frequently the same RNA was edited across replicates. The RNAs with scores from 0.85 to 1 were considered putative targets. F. ECDF graph comparing the translational efficiency scores in Ngn1-GFP sorted cells8 of HyperTRIBE targets compared to non-HyperTRIBE RNAs. Statistics: Two-sample Kolmogorov-Smirnov test. For all statistical tests in this figure: P<0.05 = *, P<0.01=**, P<0.001=***.

Mex3a Ubiquitin targets identified in MOE are implicated in trafficking and translation

A. STRING representation of putative ubiquitin targets of Mex3a. Ubiquitinated peptides were enriched from Mex3a WT and cKO MOE and proteins with significantly higher ubiquitination in the WT context were considered putative targets of Mex3a. Size of circle relates to log2 ratio of enrichment in WT compared to Mex3a cKO, and opacity of circle reflects Q value. Proteins are circled with blue (cytoskeleton), pink (mitochondria), or magenta (endomembrane) rings to represent cellular localization. STRING interactions depicted as weighted lines with more confident interactions shown as thicker lines. B. Representative live cell images from cultured OSNs, Mex3a WT and cKO; Ngn1-GFP. Magenta arrows point towards examples of vesicles measured in analysis. C. Quantification of number of vesicles per cell, n = 22 WT cells, 25 Mex3a cKO cells, two biological replicates each genotype. Statistics: Wilcoxon rank sum test. D. Experimental design to test which ubiquitin linkage is conferred onto putative ubiquitin targets Serbp1 and Rps7. Each HA tagged ubiquitin construct harbors lysine→arginine mutations to allow only one ubiquitin-linkage, except for the WT-ubiquitin construct which has all lysines intact. HEK293T cells are transiently transfected, and mEGFP construct is immunopurified with GFP-magnetic beads. Input and immunopurified samples are run by SDS page for each condition, and Western blot is performed for HA and GFP (Figures 4E and S4A-C). E. Western blot for HA after immunopurifying mEGFP tagged constructs (Rps7, left, Serbp1, right). Molecular weight of ladder markers shown on left Western. See Figure S4A-C for additional controls. For all statistical tests in this figure: P<0.05 = *, P<0.01=**, P<0.001=***.

Mex3a levels are associated with changes in Serbp1 and phosphorylated eEF2 levels in ribosomes

A. Polysome profiling from Mex3a WT or cKO MOE tissue, two biological replicates for each genotype consisting of 10 pooled MOE for each replicate, 20 MOE for each genotype. Figure generated with QuAPPro116. B. Polysome to Monosome ratios for WT and Mex3a cKO samples. Area of monosome and polysomes measured using QuAPPro, and ratio quantified by dividing the area of polysomes by area of monosomes. Error bars, SEM. Statistics, Student’s t-test. Data from two biological replicates for each genotype. C. Representative Western blot for endogenous mouse eEF2 (phosphorylated at Thr56), Serbp1, and Rpl30 in fractions from Mex3a WT. Molecular weight of ladder markers shown applies to (C) and (D). D. same as (C) but for Mex3a cKO polysome fractions. E. Quantification of Western blot for Serbp1 and p-eEF2 presented as fold change of Mex3a cKO over WT, each fraction normalized to Rpl30 value from the same fraction. Data are plotted by fraction on the x-axis, with a thin dotted line where the monosome peak is observed by polysome profiling. Thick dotted line at y=1 to show positive (above the line) or negative (below the line) fold change compared to Mex3a WT. Data presented from two replicates each genotype. F. Volcano plots showing differentially abundant RNAs in late (fractions 35-42) versus early (factions 27-34) for Mex3a WT (top) and Mex3a cKO (bottom) identified using DESeq2, results calculated with lfcThreshold = 0.5849 for more stringent statistical test. log10 p-value is plotted on y-axis, but transcripts were considered significant only if adjusted p-value was < 0.05. G. Polysome profile of HEK293T cells transiently transfected with Serbp1-mEGFP and Mex3a-mCherry (dark green line) or Serbp1-mEGFP and mCherry (light blue line). Profiles were aligned at peak of monosome (dotted line). Figure generated with QuAPPro. Two replicates for each condition. Five 10cm dishes pooled per condition and replicate. H. Forty-two fractions were collected and RPL30 and Serbp1-mEGFP/SERBP1 levels were determined by Western blot for each fraction from the mCherry and Serbp1-mEGFP transfection. Note that bands at the expected sizes for human SERBP1 and mouse Serbp1-mEGFP are seen after blotting with the Serbp1 antibody. Molecular weight of ladder markers shown applies to (H) and (I). I. Representative Western blot for Serbp1 and RPL30 for polysome fractions from Mex3a-mCherry and Serbp1-mEGFP transfection. J. Fold change (Mex3a-mCherry transfection/mCherry transfection) of human SERBP1, mouse Serbp1 and p-eEF2 Western signals, normalized to RPL30. Thick dotted line at y=1 to show positive (above the line) or negative (below the line) fold change relative to mCherry control. Data presented from two replicates each of five pooled 10cm cell culture dishes per condition. K. Model depicting Mex3a conferring K27 ubiquitin on Serbp1 which promotes its recruitment to ribosomes and modulates hibernating ribosomes and/or translational initiation/elongation. For all statistical tests in this figure: P<0.05 = *, P<0.01=**, P<0.001=***.

Supporting Main Figure 1

A. Example representation of image analysis of OR positive cells’ localization in the MOE, same data as in Figure 1B and C. For each image, the apical and basal extent of the MOE was delineated in FIJI, then cells were assigned a position and partitioned into either the basal lower half (more immature/progenitor cell) or apical upper half (more mature OSN) of the epithelium. B. Violin plots quantifying OR mean fluorescence intensity per cell, for basal cells (left panels) or apical cells (right panels). C. Representative immunofluorescence image for Adcy3 (green), DAPI in blue, from greater than three biological replicates at 4-6 weeks of age. Scale bar, 30 µM. D. Violin plot quantifying fluorescence intensity of Adcy3 across MOE of three biological replicates, 4 weeks old. Entire section of MOE was scanned with W1-Yokogawa spinning disk confocal and 32 tiles of 160 µM x 160 µM were saved as regions of interest in FIJI. Mean fluorescence intensity in the green Adcy3 channel was calculated for each tile. Statistics, Wilcoxon rank sum test. E. Immunofluorescence for Adcy3, 100X magnification of staining in cilia. Images obtained with W1-Yokogawa spinning disk confocal. Representative image from two biological images at PN12. Scalebar, 20 µM. F. Immunofluorescence of Arl13b imaged with Zeiss LSM 800 confocal Airyscan microscope depicting dendritic knobs in apical, airway space above sustentacular and OSN nuclei counterstained with DAPI (blue). Scale bar, 10 µM. G. Fluorescence image of UtrCH-stayGold (green) and mCherry or Mex3a mCherry (magenta) in transiently transfected HEK293T cells. UtrCH is an actin binding protein described to mark filapodia and/or lamellipodia. Images taken with SoRA-W1 Yokogawa spinning disk confocal microscope. Scale bar, 10 µM. H. Endogenously labeled OMP-ires-GFP (top panels) or Ngn1-GFP (bottom panels) in green with Arl13b immunofluorescence in magenta. Mex3a WT on left and Mex3a cKO on right. Images taken with W1-Yokogawa spinning disk confocal microscope. Scale bar, 10 µM. For all statistical tests in this figure: P<0.05 = *, P<0.01=**, P<0.001=***.

Supporting Main Figure 2

A. UMAP projection of mature OSNs in scRNA-Seq experiment described in8, Mex3a WT on left, cKO on right. Cells are color-coded by their ER stress rank, which is determined by the chosen OR in each cell25 B. Volcano plot of differentially expressed genes at RNA level in high and low ER stress Mex3a WT mOSNs in the dataset presented in (A). Differentially expressed genes were used to calculate the high and low ER stress gene module scores quantified in (C). C. Violin plots quantifying the Module scores26 for “Low ER stress genes” in Low stress mOSNs on the left and “High ER stress genes” in High stress mOSNs on the right for both Mex3a WT and cKO mOSNs. Statistics, Wilcoxon rank sum test. D. Log normalized RNA expression levels for known markers of mOSNs in Mex3a WT and cKO mOSNs by scRNA-Seq. Seurat FindMarkers call, Wilcoxon rank sum test. E. Volcano plot showing differentially expressed genes (Mex3a cKO sustentacular cells / Mex3a WT sustentacular cells) found by scRNA-Seq of whole MOE, PN128. Genes found to be markers of mature OSNs (Seurat FindMarkers) are colored in red. F. Model depicting how Globose basal stem cells, which express high levels of Mex3a, differentiate into olfactory sensory neurons and sustentacular cells. In the absence of Mex3a, we observe changes in differentiation profiles where neurons exhibit higher levels of sustentacular markers, and sustentacular exhibit higher levels of neuronal markers. For all statistical tests in this figure: P<0.05 = *, P<0.01=**, P<0.001=***.

Supporting Main Figure 3

A. Experimental design for HyperTRIBE experiment. Negative control (yellow) tetO-HyperTRIBE-V5-t2a-mCherry does not express Mex3a, tetO-Mex3a-HyperTRIBE-V5-t2a-mCherry (green) does express Mex3a. To test the constructs, two MOE-specific drivers, OMP-tTA and gg8-tTA were used, which drive transgene expression in mature and immature neurons, respectively. For putative Mex3a targets, only the gg8-tTA driver was used, as Mex3a is expressed in immature neurons in normal development. B. Bar graph plotting transcript per million (TPM) of Adar gene (top) and Mex3a gene (bottom) from RNA-Seq libraries from mCherry+ (colored) or mCherry-(black) sorted cells from individual mice (three different founders for tetO-HyperTRIBE control (abbreviated HT_1 through _3) and six different founders for tetO-Mex3a-HyperTRIBE (abbreviated MexHT_1 through _6)). C. Bar graph depicting total number of editing sites (A→I, read out as G after sequencing) per RNA-Seq library of mCherry+ sorted cells. HyperTRIBE only transgenic founders in yellow, Mex3a-HyperTRIBE transgenic founders, green.

Supporting Main Figure 4

A. Control Western blots for results presented in Figure 4E. Left, input lysates taken before immunoprecipitation for transiently transfected HEK293T cells co-transfected with Mex3a and Rps7-mEGFP, HA is shown in magenta and GFP is shown in green. Right, Western blots after immunopurification of GFP constructs, shown in green, and HA-ubiquitin shown in magenta. B. same as (A) but for Serbp1-mEGFP. C. same as (A) and (B) but for mEGFP construct alone. Due to low expression of K33 ubiquitin construct for all conditions, we do not conclude whether targets are ubiquitinated with K33-linkage or not. All cells were cultured with proteasome inhibitor MG132 to improve K48-ubiquitin abundance.

Supporting Main Figure 5

A. QuAPPro was used to quantify the area of the monosome peak in each sample, and ratio relative to control was calculated for each experiment. While not significant, there is a trend for decreased monosome area in mouse Mex3a cKO MOE and increased monosome area in HEK293T cells transfected with Mex3a-mCherry B. Western blot in HEK293T cells transiently transfected with Mex3a-mCherry and Serbp1-mEGFP. Polysome profiling was performed and a subset of fractions around the monosome was used to blot for Mex3a and RPL30 by Western blot. Westerns for each protein run in parallel with same quantity of sample, images are from different Western blots. C. Second replicate of HEK293T polysome profiling experiment, profiles aligned at dotted vertical line and plotted with QuAPPro. D. Western blot in polysome fractions around the monosome for endogenous human p-eEF2 in HEK293T cells transfected with mCherry alone and Serbp1-mEGFP (top) or Mex3a-mCherry and Serbp1-mEGFP (bottom). Quantified in Figure 5J.