Fig. S4

mRNA was extracted from total retinal lysates from three independent tamoxifeninduced CKO and corn oil-injected control retinae. We performed qrtPCR analysis and found that one out of three CKO samples showed a significant reduction in Lats1  while all three showed a modest but significant reduction in Lats2 (A) . Importantly, in contrast to the Yap/Taz  CKOs (see Figure S3A), the Lats1/2  CKO retinae showed a significant increase in Cyclin D1  mRNA. This finding raised the possibility that loss of Lats1/2  in a subset of MGs results in more active YAP/TAZ, which drives an increase in Cyclin D1  and spontaneous MG cell cycle entry (further supported by Figure 3). Interestingly, the increases in Cyclin D3  levels were less significant suggesting that loss of Lats1/2  has a greater impact on Cyclin D1  expression (A) . Similar to the Yap/Taz  CKOs, the varying levels of mRNA reduction (Lats1/2 ) between mice likely reflect a combination of Cre mosaicism, variation in tamoxifen induction, and inefficiency in recombining four independent floxed loci in a given cell. Also, Lats1,  and Lats2  are expressed in retinal endothelial cells, which are expected to contribute to mRNA transcripts within total retinal lysate preparations (Neto et al., 2018; Sakabe et al., 2017). For qrtPCR, mRNA levels are relative to the corn oil control (set to 1) and depicted as fold change ±SEM (n = 3 independent pooled samples per group; Student’s t test). N = 3 per group. ONL = outer nuclear layer, INL = inner nuclear layer, GCL = ganglion cell layer. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001, ns = not significant. For visual resolution of the Lats1/2  CKO MGs, we generated Glast-CreERT2+/tg; Lats1flox/flox; Lats2flox/flox mice that carry the Rosa26R-nTnG+/tg Cre reporter (Prigge et al., 2013; Schmidt, 2013). We chose this approach because LATS1/2 antibodies proved ineffective for immunofluorescence. A subset of MGs in the Lats1/2  CKO retinae exhibit Cre activity indicated by patches of GFP expression from the ROSA26R-nTnG  Cre reporter (see green channel in B-D) . Within this population of MGs, we also observed a more intense CYCLIN D1 immunofluorescent signal as compared to corn oil-injected controls (see magenta channel in B-D and quantification in E) . It is important to note that the increase in CYCLIN D1 immunofluorescence frequently coincides with GFP (Cre reporter) expression (white arrows in D) . However, we also observed CYCLIN D1 upregulation in MGs that do not express GFP (blue arrows in D)  and GFP+ MGs that do not upregulate CYCLIN D1 (yellow arrow in D) . These data underscore the inefficiencies Cre-mediated recombination of 5 independent loci in a single cell and is the reason why we turned to the more straightforward Yap5SA  transgenic model to further investigate bypass of the Hippo pathway in MGs. N > 3 per sample. We next sought to determine whether MGs within the Lats1/2  CKO retinae also exhibited more nuclear YAP immunofluorescence, which would indicate loss of the Hippo pathway’s negative regulation of YAP. We were able to locate discrete regions of the CKO retinal INL that had MGs with more intense nuclear staining of YAP (compare F-G) . Importantly, this increase in fluorescence occurred in the same MGs that exhibited an increase in CYCLIN D1 (H-I, measurements were taken across the red line in H) . Therefore, these data suggest that the increase in CYCLIN D1 observed in MGs of the Lats1/2  CKOs is a consequence of deregulated YAP. These data also support our earlier conclusion (Figures 2, S3) that Cyclin D1  is transcriptionally regulated by YAP.  N > 3 per sample. EdU labeling of adult Lats1/2  CKO retinae showed EdU+/SOX2+ MGs in the INL of the central retina (arrowhead in J) . Panel J is an expanded view of the insets shown in Figure 3F. These results were further confirmed by SOX9 immunofluorescence. EdU+/SOX9+ MGs were observed throughout the retina including peripheral regions (arrowheads in K) . N > 3 per group. ONL = outer nuclear layer, INL = inner nuclear layer, GCL = ganglion cell layer.