ATMIN is a transcriptional regulator of both lung morphogenesis and ciliogenesis

Initially identified in DNA damage repair, ATM-interactor (ATMIN) further functions as a transcriptional regulator of lung morphogenesis. Here we analyse three mouse mutants, Atmingpg6/gpg6, AtminH210Q/H210Q and Dynll1GT/GT, revealing how ATMIN and its transcriptional target dynein light chain LC8-type 1 (DYNLL1) are required for normal lung morphogenesis and ciliogenesis. Expression screening of ciliogenic genes confirmed Dynll1 to be controlled by ATMIN and further revealed moderately altered expression of known intraflagellar transport (IFT) protein-encoding loci in Atmin mutant embryos. Significantly, Dynll1GT/GT embryonic cilia exhibited shortening and bulging, highly similar to the characterised retrograde IFT phenotype of Dync2h1. Depletion of ATMIN or DYNLL1 in cultured cells recapitulated the in vivo ciliogenesis phenotypes and expression of DYNLL1 or the related DYNLL2 rescued the effects of loss of ATMIN, demonstrating that ATMIN primarily promotes ciliogenesis by regulating Dynll1 expression. Furthermore, DYNLL1 as well as DYNLL2 localised to cilia in puncta, consistent with IFT particles, and physically interacted with WDR34, a mammalian homologue of the Chlamydomonas cytoplasmic dynein 2 intermediate chain that also localised to the cilium. This study extends the established Atmin-Dynll1 relationship into a developmental and a ciliary context, uncovering a novel series of interactions between DYNLL1, WDR34 and ATMIN. This identifies potential novel components of cytoplasmic dynein 2 and furthermore provides fresh insights into the molecular pathogenesis of human skeletal ciliopathies.

Supp. Figure 2. ATMIN antibody detects multiple bands. (A) Western blot analysis of 11.5dpc embryo extracts shows the antibody to detect multiple band including a pair at ~110-120 kDA. (B) The lower of the two ATMIN bands is undetectable in gpg6 homozygotes. The Actin loading control is shown below.
Supp. Figure 3. DNA damage. Wildtype (A), Atmin gpg6/gpg6 (B) and wild type MMS treated (C, positive control) 12.5 dpc embryos were stained for 53BP1 localisation. Sections of the eye are shown. (D) Analysis of staining patterns between wildtype and Atmin gpg6/gpg6 embryos reveals no significant differences.
Supp. Figure 4. No cilia phenotype evident in gpg6 heterozygotes. (A) SEM image of gpg6/+ node, showing a similar phenotype to wild type embryos. (B) categorical analysis of cilia length from gpg6/+ embryos compared to wild type and gpg6/gpg6 embryos. No significant difference is evident between the wild type and gpg6/+ embryos.
Supp. Figure 5. Overexpression of Atmin in IMCD3 cells results in increased Dynll1 expression. n=3 repeats. qRT-PCR analysis revealed a 30-fold increase in Dynll1 expression when Atmin expression was induced by 10-fold. *** represents p<0.001 and error bars show standard deviation.
Supp. Figure 7. Dynll1 GT/GT mutant cilia demonstrate significant incidence of morphological abnormalities. SEM analysis of nodes from 2-4 somite stage wild type (A), Atmin gpg6/gpg6 (B) and Dynll1 GT/GT (C) embryos, revealing no obvious gross morphological differences at this magnification. Analysis of neural tube (D) and limb bud (E) cilia length from 11.5 dpc wild type (WT) and Dynll GT/GT embryos did not reveal statistically significant changes (p=0.183 and p=0.425 respectively). Three embryos were analysed per class. Analysis of cilia width (measured 0.5mm above the cell membrane) from the same samples in neural tube (F) and limb bud (G) cilia however, revealed significantly wider cilia in mutants (p=0.0166 and p= 0.173 respectively). (H) Categorical analysis of proportion of cilia falling into either "normal" or "bulging" classes for WT or mutant Dynll GT/GT nodes. Approximately half of the cilia in the mutant nodes presented the bulging phenotype.

Supp. Figure 8. IFT88 protein accumulates in Dynll /GT/GT mutant cilia.
Wild type, Dynll1/+ and Dynll1/Dynll1 cilia stained for the presence of IFT88 protein (green) and acetylated tubulin (red). Ten randomly selected cilia were imaged for each genotype, using identical imaging conditions and exposure. The IFT88 channel is shown in monochrome to the right of each merged image.
Supp. Figure 9. LC8 localises to nodal cilia. A maximum intensity projection of the node imaged in Supp. Movie. Top panel: anti-acetylated tubulin antibody, visualising cilia. Middle panel, anti LC8 antibody. Bottom panel shows combined image with acetylated tubulin in red and LC8 in green. A representative cilium (boxed) is shown magnified in the bottom right hand corner of each panel.
Supp. Figure 10. LC8 localises to primary cilia in NIH-3T3 cells. A cultured NIH-3T3 cell labelled for acetylated tubulin (top panel) and for LC8 (middle panel) revealing punctate LC8 staining in the cilium. A combined image with acetylated tubulin in red and LC8 in green is shown in the bottom panel.
Supp. Figure 12. LC8 expression in HEK293T cells. Myc-DYNLL1 and myc-DYNLL2 expressed in HEK293T cells, revealing endogenous protein at ~10kDa detected by anti-LC8 antibody and larger myc-tagged proteins identified by both anti-myc and anti-LC8 antibodies.
Movie 1. LC8 localises to nodal cilia. A 3 somite stage mouse embryonic node, stained for anti acetylated tubulin (red) and LC8 (green). A confocal image stack allowing all cilia within the node to be visualised, revealing that LC8 staining is evident in all nodal cilia.