For clues into how lamina neuron types are diversified, we integrated published single-cell RNA sequencing (scRNA-seq) datasets at developmental timepoints that span lamina development (Konstantinides et al., 2021; Kurmangaliyev et al., 2020; Özel et al., 2021; Figure 1—figure supplement 1A; see Materials and methods section). As published previously, on uniform manifold approximation and projection (UMAP) visualisations L1, L2, L3, and L4 neuronal clusters were closer to each other than they were to the L5 cluster (Özel et al., 2021). The LPC cluster was connected to differentiated neurons by two streams of cells: one leading to the L1-L4 clusters and the other leading to the L5 cluster (Figure 1D); such convergent streams are thought to represent intermediate states between progenitors and differentiated cells (Konstantinides et al., 2021; Özel et al., 2021).

First, we probed these data to identify additional early markers of cell identity for each of the lamina neuron types (beyond Slp2, Bsh, and Svp). Consistent with previous reports, scRNAseq analysis showed that L3 and L4 neurons expressed earmuff (erm) and apterous (ap), respectively (Tan et al., 2015; Ting et al., 2005). In addition, we found that reduced ocelli (rdo) was expressed at high levels in L2 neurons and at lower levels and more sporadically in L3 neurons, that L1 neurons expressed Vesicular glutamate transporter (VGlut) specifically, and that L5 and L3 neurons expressed high and low levels of defective proboscis extension response 8 (dpr8), respectively (Figure 1—figure supplement 1B–F). We used in situ hybridisation chain reaction (HCR) (Choi et al., 2018) to validate marker expression in vivo at 0 hr after puparium formation (0 hr, APF) (Figure 1—figure supplement 1G–K), thus providing additional markers for the differentiating lamina neuron types.

Next, we asked which transcripts distinguished L5 precursors from L1 to L4 precursors during development in the scRNA-seq data (see Materials and methods section). Surprisingly, comparing gene expression between the two streams connecting undifferentiated LPCs with L1-L4s and L5s, indicated that several well-established Hh signalling targets were expressed at higher levels in the L1-L4 stream compared with the L5 stream, suggesting that L5 precursors experience lower Hh signalling than other lamina precursors. These included the direct transcriptional target patched (ptc), as well as single-minded (sim), rhomboid (rho), and Zinc finger homeodomain 1 (zfh1) (Figure 1E–H; Albert et al., 2018; Biehs et al., 2010; Chen and Struhl, 1996; Umetsu et al., 2006). To validate these data in vivo, we examined the expression of a transcriptional reporter for ptc (ptc-lacZ) as a readout of Hh signalling activity (Chen and Struhl, 1996; Tabata and Kornberg, 1994; Figure 1I). We measured the mean fluorescence intensity (MFI) of β-galactosidase (β-Gal) in the lamina at six cell positions corresponding to each LPC position within a column (i.e. position 1 corresponded to the most distal cell [prospective L2] and position 6 corresponded to the most proximal cell [prospective L5] in columns; schematic in Figure 1J) from young to old columns (anterior to posterior axis). To do so, we generated average intensity projections (from 10 optical slices) obtained from the central lamina and then measured the fluorescence intensity profiles for each of the six cell positions from the youngest column to the oldest column (quantified in Figure 1J with summary statistics from a mixed effects linear model in Supplementary file 1; see Materials and methods for more detail). We also measured the MFI of β-Gal for each cell along the distal to proximal axis for the youngest column (Figure 1K). Consistent with the transcriptomic data where ptc expression was lowest in older maturing neurons (Figure 1E), ptc-lacZ expression eventually decreased along the young to old axis for each of the six (distal-proximal) cell positions (Figure 1I and J; N=10). Importantly, in young columns prior to neuronal differentiation, we observed a gradient of ptc-lacZ expression, which was highest in the distal lamina and decreased along the distal-proximal axis (Figure 1K; one-way ANOVA with Dunn’s multiple comparisons test, p<0.0001). Thus, we observed that the direct Hh signalling target, ptc, was graded along the distal to proximal length of the youngest column, indicating that there is a gradient of Hh signalling activity in lamina columns prior to neuronal differentiation.

We sought to determine if graded Hh pathway activity in the lamina arises from a concentration gradient of Hh ligand. We analysed the distribution of Hh protein tagged with superfolder Green Fluorescent Protein (Hh::GFP) expressed under the control of endogenous enhancers of the hh locus and inserted into the genome using a bacterial artificial chromosome in a single copy (Chen et al., 2017). This transgene was previously reported to rescue a hh amorph, indicating that it is a functional hh allele (Chen et al., 2017). We first used immunohistochemistry (IHC) to detect Hh::GFP and observed higher levels in younger photoreceptor cell bodies compared to older photoreceptor cell bodies in the eye disc, consistent with previous reports (Figure 2A, A'; Huang and Kunes, 1996). We also observed Hh::GFP in photoreceptor axons in the optic stalk but observed very little signal in photoreceptor axons in the lamina (Figure 2A, A'; N=18). Secreted proteins can be particularly sensitive to fixation and washes during IHC, therefore to avoid any confounding artefacts potentially caused by IHC, we visualised Hh::GFP live in cultured brain explants (Bostock et al., 2020; Figure 2B). We measured Hh::GFP MFI as a function of distal-proximal distance for the youngest lamina column (see Materials and methods section) and observed a gradient polarised from high to low from the distal to proximal ends of columns (Figure 2B–C; N=10). These measurements do not distinguish between intracellular and extracellular Hh::GFP, but instead, represent total Hh::GFP. Nonetheless, these data indicate that Hh ligand is present in a distal to proximal gradient in the lamina, consistent with the gradient in Hh pathway activity along lamina columns and suggest that ligand availability is responsible for the activity gradient.

Hh signalling is known to trigger the early steps of lamina development, including promoting early lamina marker expressions such as dac and sim, as well as promoting terminal LPC divisions and column assembly (Huang and Kunes, 1998; Huang and Kunes, 1996; Sugie et al., 2010; Umetsu et al., 2006). However, no role for Hh in lamina neuron specification or differentiation has been identified thus far. In the vertebrate neural tube, a gradient of Shh signalling specifies ventral progenitor identities (Placzek and Briscoe, 2018), therefore, we wondered whether the graded Hh pathway activity we observed could specify neuronal identity in LPCs; in particular, we hypothesised that high Hh signalling levels in LPCs specify distal neuron identity (i.e. L2s) and low Hh signalling levels specify proximal neuron identity (i.e. L5s).

To test this, we asked whether increasing Hh signalling cell autonomously would specify LPCs to differentiate with distal lamina neuron identity (i.e. L2). We generated LPCs with a gain of function in Hh pathway activity by inducing positively labelled clones (Lee and Luo, 2001) that were homozygous mutants for the negative Hh pathway regulator ptc (Chen and Struhl, 1996; Kalderon, 2005). The ptc mutant clones never contained L5s (Slp2 and Bsh co-expressing cells), L4s (Bsh-only expressing cells), or L1s (Svp and Slp2 co-expressing cells) but instead only contained neurons singly positive for Slp2, indicative of L2/L3 identity (Figure 3A–B; 17/17 clones from 15 optic lobes; fully penetrant phenotype). To determine the precise identity of these neurons (L2 or L3), we examined the expression of rdo and erm by HCR in ptc mutant clones in the proximal lamina and found mixed rdo and erm expression (Figure 3C–D; N=15/15 clones), which indicated that both L2 and L3 neurons were present in the clones. Thus, increasing Hh signalling maximally in the lamina resulted in distal neuron specification (mixed L2 and L3) at the expense of L1, L4, and L5 specifications.

Next, we sought to disrupt Hh signalling cell autonomously in the lamina and assess the effects on lamina neuron specification. When we blocked Hh pathway activity completely in clones mutant for smoothened (smo), we did not recover any Dac-expressing LPCs (Figure 3—figure supplement 1A; N=15/15 clones; fully penetrant phenotype), consistent with previous reports (Huang and Kunes, 1998). Therefore, to reduce Hh pathway activity partially, we used a lamina-specific driver (R27G05-Gal4), which drives expression in Dac-expressing LPCs exiting the posterior lamina furrow (Figure 3—figure supplement 1B), to express a repressor form of the transcriptional effector of Hh signalling, Cubitus interruptus (Ci; referred to as Ci^rep) (Busson and Pret, 2007). We used a temperature-sensitive Gal80 (Gal80^ts) to restrict Gal4 activity temporally by raising crosses at the permissive temperature (18°C) for Gal80^ts and shifting to the restrictive temperature (29°C) to allow Gal4 activity from the third larval instar (see Materials and methods section). Expressing Ci^rep in LPCs led to a 2.8-fold increase in the proportion of Bsh and Slp2 double-positive L5s relative to controls, while the proportion of other lamina neuron subtypes was reduced relative to controls (Figure 3E, F and H–L and Figure 3—figure supplement 1C–F; one-way ANOVA with Dunn’s multiple comparisons test; p-values <0.001; N=10). Moreover, whereas in control laminas L5s were only observed as a single row in the proximal lamina, expressing Ci^rep caused Bsh and Slp2 double-positive L5s to be distributed along the entire distal-proximal lengths of columns (Figure 3F, N=10). Similarly, knocking down Ci in the lamina by RNA interference (Ci^RNAi) (Figure 3G; N=9) resulted in ectopic L5s (2.4-fold increase relative to controls, one-way ANOVA with Dunn’s multiple comparisons test; p<0.0001) and a reduction in the proportion of L2-L3s and L4 lamina neurons relative to controls (Figure 3H–L, one-way ANOVA with Dunn’s multiple comparisons test; p^L2-L3<0.0001 and p^L4<0.0001; see Figure 3—figure supplement 1C–F for raw values). Thus, decreasing Hh signalling autonomously within LPCs was sufficient to induce the most proximal cell identity (L5s), throughout the distal-proximal lengths of columns and at the expense of other neuronal types. In summary, disrupting Hh signalling activity cell autonomously to either high or low extremes specified LPCs with either the most distal or proximal lamina neuron identities, respectively.

The data presented thus far are consistent with Hh acting as a morphogen to specify cell identities in the lamina, that is, in a concentration-dependent manner; however, our cell-autonomous manipulations only determined the impact of experiencing either the high or low extremes of pathway activity on lamina neuron identity specification. Therefore, we attempted to tune Hh signalling to intermediate levels by using the Gal4/UAS system to express Ci^rep in the lamina as before under two temperature conditions (in the presence of Gal80^ts): 29°C for a strong inhibition of Hh signalling (as above) and 25°C for a milder inhibition. We then evaluated the distribution of lamina neuron types using rdo, erm, VGlut, ap, dpr8, and bsh. Strong inhibition of Hh signalling (lamina^ts >Ci^rep at 29°C) led to a 2.2-fold increase relative to controls in the proportion of the lamina that co-expressed dpr8 and bsh (i.e. L5s) at the expense of other markers (Figure 4A–F and J–N; N^dpr8+bsh=8; one-way ANOVA with Dunn’s multiple comparisons test; p^dpr8+bsh<0.0001). As before, dpr8 and bsh co-expressing cells (L5s) were no longer restricted to a single row in the proximal lamina but were distributed along the entire distal-proximal lengths of columns (Figure 4C and F). In contrast, milder inhibition of Hh signalling (lamina^ts >Ci^rep at 25°C) led to a 2.5-fold increase relative to controls in the proportion of the lamina that expressed VGlut (i.e. L1s) and a 1.9-fold increase relative to controls in the proportion of the lamina that expressed ap (i.e. L4s), at the expense of rdo expressing (erm-negative) L2s (Figure 4A–C– and J–N; one-way ANOVA with Dunn’s multiple comparisons test; p^VGlut<0.0001; p^ap<0.01). We obtained similar results when we evaluated the distribution of neuron types with Slp2, Svp, and Bsh (Figure 4—figure supplement 1). Thus, intermediate neuron identities (L1s and L4s) were favoured when we tuned Hh signalling to intermediate levels.

To further test whether different thresholds of Hh ligand determine lamina neuron identities, we sought to manipulate Hh directly and to differing degrees of severity. We predicted that, if Hh was acting as a morphogen, progressively reducing Hh levels should result in progressively more proximal identities. To do so, we used animals harbouring a temperature-sensitive loss of function allele of hh (hh^ts2), which we raised at the permissive temperature (18°C) and then shifted to the restrictive temperature (29°C) for either 6 hr, 12 hr, 24 hr, 45 hr, or 72 hr before dissecting at the white prepupal stage (Figure 5A–G, Figure 5—figure supplement 1A-J). Consistent with the known role of Hh signalling in photoreceptor and early lamina development (Huang and Kunes, 1996), we recovered fewer and fewer photoreceptors and lamina neurons with increasing lengths of temperature shifts (Figure 5—figure supplement 1A-J), with no LPCs or lamina neurons recovered for the 45-hr and 72-hr temperature shifts (Figure 5F and G; Huang and Kunes, 1996). These results indicated that our perturbations effectively titrated the availability of functional Hh with longer periods at the restrictive temperature. Next, we used lamina neuron-type-specific markers to evaluate neuronal cell types under these conditions. We found that the proportion of L2-L3s (Slp2 single-positive neurons) decreased as the availability of functional Hh was titrated to lower and lower levels, whereas the proportion of L5s (Slp2 and Bsh double-positive neurons) increased (Figure 5A–E; quantified in Figure 5I–L; one-way ANOVA with Dunn’s multiple comparisons test; p-values <0.05). Importantly, the proportion of L1s (Slp2 and Svp double-positive neurons) increased only for the 6-hr temperature shift. Thus, L1 neurons, which normally occupy intermediate lamina column positions were favoured at the expense of distal neuron types at intermediate thresholds of Hh pathway activity (Figure 5C and J; one-way ANOVA with Dunn’s multiple comparisons test; p-values <0.05). Notably, under these conditions, L1s and L5s were no longer confined to their normal positions within lamina columns but were observed throughout (Figure 5A–D). Altogether our data establish that Hh acts as a morphogen to diversify lamina neuron types.

Others have reported previously that photoreceptors are the sole source of Hh to the optic lobes during lamina development (Huang and Kunes, 1996). We corroborated these results using an enhancer trap in the hh locus (hh-lacZ; hh^P30), which showed that β-Gal expression was restricted to photoreceptors in the eye disc and absent from the optic lobes, except for a few sparse cells near the central brain (Figure 6A). We reasoned that if photoreceptor-derived Hh indeed patterns lamina neuron identities, then disrupting Hh expression specifically in photoreceptors should in turn result in lamina neuron patterning defects. Hh secreted by developing photoreceptors is required to drive morphogenetic furrow progression and continued photoreceptor development in the eye disc (Greenwood and Struhl, 1999; Treisman, 2013). Consequently, removing hh expression completely in the eye disc has been shown to block both photoreceptor development (Greenwood and Struhl, 1999; Treisman, 2013) and lamina development, due to its early role in inducing LPCs (Huang and Kunes, 1998; Huang and Kunes, 1996). Therefore, we used GMR-Gal4, which is expressed after photoreceptor birth together with a previously validated RNAi line (Kim et al., 2014; Ni et al., 2011; Sahai-Hernandez and Nystul, 2013) to partially knock down Hh in photoreceptors, thus bypassing its early roles. Under these conditions, lamina size was not noticeably affected, indicative of a mild knockdown (Figure 6B and C). When we evaluated lamina neuron-type identities, we found that Slp2 and Svp expressing cells (L1s) increased by 1.2-fold relative to controls (Figure 6C; Mann-Whitney U test; p<0.001; N=10), consistent with more LPCs experiencing intermediate Hh signalling activity. Importantly, L1 neurons were no longer restricted to a single row at intermediate positions in columns but instead were recovered in more distal positions (Figure 6C). Thus, partially knocking down Hh expression in photoreceptors disrupted lamina neuron patterning in favour of more intermediate cell identities (L1s), consistent with mildly decreasing Hh signalling to more intermediate levels in the lamina.

Here, we uncovered a previously unknown role for Hh as a morphogen that specifies lamina neuron identities, thus revealing yet another layer of dependence of lamina development on photoreceptor-derived signals. Intriguingly, a mix of L2 and L3 identities were specified when Hh signalling was activated to maximum physiological levels (Figure 3D). This suggests that in addition to high Hh signalling levels, additional signals are required to further differentiate L2 from L3 identities. This is the first report of a morphogen patterning neuronal fates in Drosophila. It highlights the remarkable evolutionary conservation of this developmental strategy, which is widespread in vertebrates (Placzek and Briscoe, 2018; Sagner and Briscoe, 2019).

Nonetheless, it is not obvious how the Hh gradient is established in the lamina, such that it is polarised from high to low along the distal-proximal lengths of lamina columns, as photoreceptor axons themselves span this full distance. Axonal trafficking of Hh was shown to be regulated by an evolutionarily conserved motif in its cleaved C-terminal domain (Chu et al., 2006; Daniele et al., 2017), disruption of which affected neuronal diversity in the lamina specifically (Chu et al., 2006). This indicates that axonal trafficking may play a role in establishing the Hh gradient; however, more work is required to uncover the exact molecular mechanisms involved. It is unexpected that neurons, here photoreceptors, define the neuronal diversity of their target field, and raises the possibility that neurons in other contexts could also be patterning their target fields during development through morphogen gradients. Indeed, many vertebrate neuronal types express Shh (Dakubo et al., 2003; Farmer et al., 2016; Garcia et al., 2010; Gonzalez-Reyes et al., 2012; Harwell et al., 2012; Petralia et al., 2011; Wallace, 1999); however, their ability to generate instructive gradients to pattern distant neural fields has not been determined thus far.

D. melanogaster strains and crosses were reared on a standard cornmeal medium and raised at 25 or 29°C or shifted from 18 to 29°C for genotypes as indicated in Supplementary file 2.

We used the following mutant and transgenic flies in combination or recombined in this study (see Supplementary file 2 for more details; {} enclose individual genotypes, separated by commas):

Canton S, {ptc-lacZ/TM6B} (a gift from D Kalderon), {ywhsflp¹²², tub-gal4, UAS-GFP; FRT40A, Gal80/Cyo} (a gift from F Schweisguth), {ey-Gal80; sp/Cyo;} (BDSC: 35822), {;Gal80ts; TM2/TM6B} (BDSC: 7108), {w¹¹¹⁸;; R27G05-Gal4} (BDSC: 48073), {;;UAS-CD8::GFP} (BDSC: 32187), {y¹ sc^* v¹ sev²¹; Ci-RNAi} (BDSC: 64928), {UAS-Ci⁷⁶} (a gift from J Treisman), {y¹; FRT42D, ptc^S2/CyO} (BDSC: 6332), {ywhsflp¹²², tub-gal4, UAS-RFP; FRT42D, Gal80/Cyo} (a gift from G Struhl), {FRT40A, smo³ /CyO} (a gift from C Boekel), {yw¹¹¹⁸; hh::sfGFP} (BDSC: 86271), {w¹¹¹⁸; hh^ts2 e^s/TM6B} (BDSC: 1684), {;;ry⁵⁰⁶, hh^P30} (BL5530), {w^*; GMR-Gal4;} (BL1104), {;;UAS-hh-RNAi} (BL32489).

We generated smo³ and ptc^S2 MARCM (Figure 3—figure supplement 1A and Figure 3I and J, respectively) clones by heat-shocking larvae 2 days after egg laying (AEL) at 37°C for 90 min. To generate one wild-type MARCM clone per lamina (Figure 1—figure supplement 1B, C), we heat-shocked larvae (1 day AEL) for 60 min at 37°C. All MARCM crosses were raised at 25°C until dissection at 0–5 hr APF.

We dissected eye-optic lobe complexes from early pupae (0–5 hr APF) in 1× phosphate-buffered saline (PBS), fixed in 4% formaldehyde for 20 min, blocked in 5% normal donkey serum, and incubated in primary antibodies diluted in the block for two nights at 4°C. Samples were then washed in 1× PBS with 0.5% TritonX (PBSTx), incubated in secondary antibodies diluted in the block, washed in PBSTx, and mounted in SlowFade (Life Technologies).

We used the following primary antibodies in this study: mouse anti-Dac^2-3 (1:20, Developmental Studies Hybridoma Bank; DSHB), rat anti-Elav (1:100, DSHB), mouse anti-Elav (1:20, DSHB), mouse anti-Svp (1:50, DSHB), rabbit anti-Slp2 (1:100; a gift from C Desplan), rabbit-Bsh (1:500; a gift from C Desplan), mouse anti β-Gal (1:500; Promega #Z3781), chicken anti-GFP (1:400; EMD Millipore), mouse anti-Sim (1:20; a gift from T Tabata), chicken anti-RFP (1:500; Rockland #600-901-379s), rabbit-anti-GFP (1:500; Thermofisher #A6455), AlexaFluor405 conjugated goat anti-HRP (1:100; Jackson Immunolabs), AlexaFluor405-, Cy3-, or AlexaFluor647-conjugated goat anti-HRP (1:200; Jackson Immunolabs). Secondary antibodies were obtained from Jackson Immunolabs or Invitrogen and used at 1:800. Images were acquired using Zeiss 800 and 880 confocal microscopes with 40× objectives.

To assess putative lamina neuron-type-specific marker genes in vivo, we designed HCR probes against slp2, svp, bsh, rdo, erm, VGlut, and dpr8. We designed 6–21 antisense probe pairs against each target gene, tiled along the annotated transcripts but excluding regions of strong sequence similarity to other transcripts, with the corresponding initiator sequences for amplifiers B3 and B5 (Choi et al., 2018). We purchased HCR probes as DNA oligos from Thermo Fisher (at 100 μM in water and frozen). All probe sequences are included as source data (see Supporting Zip Document 1).

Eye-optic lobe complexes were dissected, fixed, and permeabilised as above. Samples were incubated in probe hybridisation buffer at 37°C for 30 min before being incubated with probes at 0.01 µM at 37°C overnight. Samples were washed four times for 15 min at 37°C with probe wash buffer and then two times for 5 min with 5× saline-sodium citrate with 0.001% Tween 20 solution (to make a 20× SSCT solution in distilled H₂O, 58.44 g/mol sodium chloride, 294.10 g/mol sodium citrate, pH adjusted to 7 with 14 N hydrochloric acid, with 0.001% Tween 20). Samples were incubated in an amplification buffer for 10 min. Hairpins H1 and H2 for each probe were snap-cooled (hairpins were heated to 95°C for 90 s and cooled to room temperature for 30 min) separately to avoid hairpin oligomerisation. About 12 pmol of each hairpin was added to samples in an amplification buffer and incubated overnight at room temperature. Samples were washed for 10 min in SSCT and then incubated in darkness at room temperature with 1:15 dilution of DAPI (Sigma: D9542) for 90 min. Samples were washed in 1× PBS for 30 min and mounted as above.

To maximise temporal resolution during development as well as the number of cells analysed, we combined three publicly available scRNAseq datasets of optic lobes from the following developmental timepoints: wandering third instar larva, 0-hr APF, 12-hr APF, 15-hr APF, and 24-hr APF (NCBI GEO: GSE156455, GSE167266, GSE142789) (Konstantinides et al., 2022; Kurmangaliyev et al., 2020; Özel et al., 2021). We combined these datasets using the Seurat v.3 integration pipeline to remove batch effects between libraries (Stuart et al., 2019) as follows: using the default parameters in Seurat v4.0.1 we first normalised each dataset with the NormaliseData function. Next, we extracted the 2000 most variable features with the FindVariableFeatures function. We then integrated the data using the FindIntegrationAnchors and IntegrateData functions. Next, we clustered the integrated dataset using the following functions: ScaleData, RunPCA (using 150 principal components as in Konstantinides et al., 2021), FindNeighbours (80 dimensions), FindClusters (resolution = 5), and RunUMAP.

We annotated clusters corresponding to lamina cell types based on a combination of previous annotations from the source datasets (Özel et al., 2021) and known markers: dac, eyes absent (eya), tailless (tll), glial cells missing (gcm) for LPCs; svp and slp2 for L1s; slp2 for L2s; erm and slp2 for L3s; bsh and ap for L4s; bsh and slp2 for L5s (Chotard et al., 2005; Guillermin et al., 2015; Hasegawa et al., 2013; Huang and Kunes, 1996; Piñeiro et al., 2014; Tan et al., 2015).

To identify additional lamina-neuron-type markers, we used the FindMarkers (two-sided Wilcox rank-sum test) function with default parameters to identify positively expressed genes based on log-fold change in individual lamina neuron-type clusters relative to the other four lamina neuron clusters.

To analyse differentially expressed genes between the L1–L4 and the L5 convergent stream, we first used the CellSelector function to manually select the two streams as two individual clusters. Next, we used the FindMarkers (two-sided Wilcox rank-sum test) function with default parameters to identify positively or negatively expressed genes based on log-fold change. To visualise UMAPs and gene expression, we used the DimPlot and FeaturePlot functions.

We used Fiji-ImageJ (Schindelin et al., 2012) or Imaris (version x64-9.5.1) to process and quantify confocal images as described below. We used Adobe Photoshop and Adobe Illustrator software to prepare figures. We used GraphPad Prism8, JMP, and R (version 4.0.3) software to perform statistical tests. In all graphs, whiskers indicate the SD. N values are indicated on graphs.