The tumor suppressor LKB1 regulates starvation-induced autophagy under systemic metabolic stress

Autophagy is an evolutionarily conserved process that degrades cellular components to restore energy homeostasis under limited nutrient conditions. How this starvation-induced autophagy is regulated at the whole-body level is not fully understood. Here, we show that the tumor suppressor Lkb1, which activates the key energy sensor AMPK, also regulates starvation-induced autophagy at the organismal level. Lkb1-deficient zebrafish larvae fail to activate autophagy in response to nutrient restriction upon yolk termination, shown by reduced levels of the autophagy-activating proteins Atg5, Lc3-II and Becn1, and aberrant accumulation of the cargo receptor and autophagy substrate p62. We demonstrate that the autophagy defect in lkb1 mutants can be partially rescued by inhibiting mTOR signaling but not by inhibiting the PI3K pathway. Interestingly, mTOR-independent activation of autophagy restores degradation of the aberrantly accumulated p62 in lkb1 mutants and prolongs their survival. Our data uncover a novel critical role for Lkb1 in regulating starvation-induced autophagy at the organismal level, providing mechanistic insight into metabolic adaptation during development.


Results
lkb1 mutants fail to activate autophagy under nutrient limitation. To determine if autophagy initiation and maintenance is affected in the lkb1 mutants we previously generated 37 , we analyzed wt and lkb1 larvae between 5-7 dpf during the metabolic transition following yolk depletion. lkb1 mutants are indistinguishable from wt larvae up to day 5-6 dpf 37 (while there is still yolk). We chose this time window also because the morphological lkb1 phenotype of flattened intestine and darkened liver is apparent at 7 dpf 37 and the majority of lkb1 mutants die at 8 dpf ( Supplementary Fig. S1). Autophagic activity is commonly monitored by accumulation of the membrane-bound form of MAP1LC3B (microtubule-associated proteins 1 A/1B light chain 3B, Atg8 in yeast; Lc3B in zebrafish), Lc3-II, which is a ubiquitin-like protein 39 that localizes in autophagosomal membranes upon induction of autophagy. To enable visualization of Lc3-II accumulation in autophagosomes, we first blocked the fusion of autophagosomes with lysosomes by treating the larvae with 2.5 μM chloroquine 40 for 14 hours (h) before analysis.
We found that Lc3-II protein levels were lower in lkb1 mutant larvae compared to their wt siblings after yolk depletion at 6 and 7 dpf (Fig. 1A). Note that the Lc3 antibody in zebrafish recognizes predominantly the cleaved Lc3B-II form, which still accurately reflects autophagic activity 41 (Supplementary Fig. S2A), and we were only able to detect a faint signal for Lc3B-I in zebrafish lysates (Fig. 1A). To verify this result, we used an alternative marker of autophagy by analyzing the expression of Atg5-containing protein complexes during development. We found that while the expression of the common ~56 KD complex was unaffected, the ~47 KD complex, which is indicative of autophagy induction 31 was undetectable in the lkb1 mutants (Fig. 1A). Finally, we also assessed the levels of Beclin 1 (Becn1), a protein involved in autophagosome nucleation 30,42 and also commonly used as an autophagy indicator. Becn1 expression was also strongly reduced in the lkb1 mutants ( Supplementary Fig. S2B). While Lc3-II and Becn1 levels progressively increased in wt larvae between 5-7 dpf, indicating upregulation of autophagy upon yolk termination, lkb1 mutants did not show such an increase, suggesting they fail to activate autophagy under nutrient limiting conditions (Fig. 1A, and Supplementary Fig. S2B).
To examine the spatial distribution of autophagy, we performed immunofluorescence analysis with antibodies against Lc3B on transverse liver and intestine sections of lkb1 and wt larvae at 7 dpf. Lc3B expression was strongly reduced in lkb1 intestines and livers as compared to their wt counterparts (Fig. 1B-E and Supplementary  Fig. S2E,F), confirming and supporting the immunoblotting results. Furthermore, immunohistochemistry (IHC) against Becn1 on transverse sections of wt and lkb1 mutants showed markedly reduced Becn1 staining in the lkb1 mutants compared to their wt counterparts ( Supplementary Fig. S2C,D).
To further confirm that activation of autophagy is impaired in lkb1 mutants, we monitored expression levels of the p62 protein (also known as sequestosome 1, (SQSTM1). p62 is an adaptor protein that targets ubiquitinated proteins or organelles that bind to it for selective autophagy 43 . Accumulation of p62 has been observed in mouse AMPK-deficient fibroblasts 44 , and is associated with liver toxicity in autophagy-deficient mouse liver 45 . p62 itself is also an autophagy substrate, thus accumulation of p62 levels is a marker for impaired autophagy 46 . In agreement with our model that lkb1 mutants have impaired autophagy, Western blot analysis of p62 levels at 5, 6 and 7 dpf showed progressive accumulation of p62 specifically in the lkb1 mutants (Fig. 1A). IHC performed on transverse sections of lkb1 intestine and liver confirmed a marked accumulation of p62 in lkb1 larvae, whereas wt siblings were devoid of staining (Fig. 1F,G).
Collectively, these findings demonstrate that whole-body autophagy is impaired in lkb1 mutants during the feeding-fasting transition in zebrafish, which could contribute to their premature death.

Abrogation of autophagy further decreases survival of lkb1 mutants.
To investigate the effect of inhibiting autophagy on lkb1 larvae survival prior to yolk depletion, we blocked autophagosome formation using an antisense morpholino oligonucleotide (MO), that targets the translational start-site of atg5 mRNA 28, 47 , atg5MO. We confirmed that injection of atg5MO abolishes Atg5 protein expression ( Supplementary Fig. S3). Atg5-knockdown led to a reduction in Lc3-II levels compared to the negative control in both wt and lkb1 mutants at 4 dpf, before yolk depletion, confirming autophagy suppression upon atg5MO injection ( Fig. 2A). While all un-injected wt and lkb1 larvae were alive at 4 dpf, a significant number of atg5MO-injected embryos were found dead at that time point. Genotyping all larvae at 4 dpf revealed that 75% of atg5MO-injected lkb1 larvae had died compared to only 25% of atg5MO-injected wt larvae (Fig. 2B). Thus, lkb1 mutants, which fail to induce autophagy at the metabolic transition, are also more sensitive to autophagy inhibition at earlier embryonic stages.
The autophagy defect in lkb1 mutants can be ameliorated by mTOR-dependent and -independent mechanisms. We next investigated the mechanism behind the impaired autophagy observed in the lkb1 mutants. Signaling through mTOR is known to inhibit autophagy, and we and others have previously reported that mTOR activity is high in wt larvae between 2 and 5 dpf and is downregulated at later stages of larval development 37, 47, 48 . This suggests that at the time of yolk depletion, mTOR activity is switched off, enabling the activation of autophagy. It has also been shown in mice that suppression of mTOR activity at birth enables activation of autophagy 49 . We hypothesized that mTOR inactivation was defective in the absence of Lkb1. Therefore, we first assessed the status of mTOR signaling in lkb1 mutants at the metabolic transition (6 dpf) by analyzing phosphorylation of the mTOR-substrate ribosomal protein S6 (RS6) by Western blot. Total RS6 levels were almost undetectable in wt larvae at this stage, consistent with our previous report 37 . However, both total and phospho-RS6 levels were high in lkb1 mutants (Fig. 3A), indicating active mTOR signaling, which was inhibited by rapamycin treatment. In comparison, in rapamycin-treated 6 dpf wt larvae, we observed increased phospho-RS6 expression (Fig. 3A). This could be explained by a known developmental delay caused by chronic mTOR inhibition during development 50 . Consistent with this, rapamycin-treated wt larvae retained significant amounts of yolk at 7 dpf, demonstrating a delay in larval development ( Supplementary Fig. S4C).
To determine whether mTOR signaling mediates the inhibition of autophagy seen in the lkb1 mutants at 6 dpf, we examined whether rapamycin treatment could restore autophagy in these mutants. We treated wt and lkb1 embryos with rapamycin from 1 dpf onwards. We have previously reported that rapamycin-treated lkb1 larvae survive until 9 dpf, but still have a considerable amount of yolk, demonstrating a developmental delay 37 . Rapamycin-treatment resulted in elevated Lc3-II levels in both wt and lkb1 larvae, at 6 dpf ( Fig. 3B), indicating that autophagy in the lkb1 mutants is inhibited, at least in part, by mTOR signaling. The same was also observed when blocking autophagosome-lysosome fusion by chloroquine, which prevents degradation of autophagosome-associated Lc3, allowing monitoring of the autophagic flux 51 (Supplementary Fig. S4A,B). However, rapamycin-treatment led to only slight upregulation of Atg5 and complexed Atg5 in lkb1 mutants (Fig. 3C), and was not sufficient to decrease the marked p62 accumulation (Fig. 3B). These results suggest that while mTOR inhibition can at least partially restore autophagy in lkb1 mutant larvae, it cannot entirely alleviate the observed phenotype.  Rapamycin treatment leads to increased Lc3-II accumulation but does not increase Atg5 (complexes) nor restore p62 degradation in lkb1 mutants. (A) Representative Western blot analysis of Ribosomal protein S6 (RS6), Phospho-RS6 and Tubulin (loading control) in total protein lysates of wt and lkb1 trunks at 6 dpf that were treated with either 10 μM rapamycin from 24 hpf onwards, or with DMSO (negative control). Increased levels of RS6 and P-RS6 are observed in rapamycin-treated wt samples. Total RS6 levels did not change in lkb1 samples but P-RS6 decreased upon rapamycin treatment. (B) Representative Western blot analysis of p62, Lc3-II, and histone H3 (loading control). Larvae were treated with chloroquine (2.5 μM) for 14 h prior to processing. Rapamycin treatment leads to increased Lc3-II levels in both wt and lkb1 larvae, while p62 accumulation remained high in rapamycin-treated lkb1 mutants. (C) Representative Western blot analysis of Atg5 and Tubulin (loading control). Rapamycin treatment leads to an increase in the amount of the ~47 KD Atg5-containing complex in wt larvae and to a lesser extent in lkb1 mutants. Uncropped images of the blots are shown in Supplementary Fig. S9A-C.
We next analyzed the pro-survival PI3K pathway, which in response to external stimuli (growth factors, insulin) also suppresses autophagy, acting upstream of mTOR signaling 52 . To this end, we used the small molecule AR-12, an inhibitor of phosphoinositide-dependent kinase (PDK)-1, a component of the PI3K pathway 53 , which has been shown to activate autophagy in zebrafish 54 . Treatment of wt and lkb1 siblings with AR-12 from 1 dpf onwards, led to accumulation of Lc3-II protein levels in wt but not in lkb1 larvae at 6 dpf (Fig. 4A). Interestingly, AR-12 treatment resulted in upregulation of Atg5 expression and formation of Atg5/12 complexes in wt larvae indicating autophagy induction, but no changes in Atg5 expression were observed in lkb1 larvae (Fig. 4B). Furthermore, while p62 protein expression was diminished in wt larvae upon AR-12-mediated activation of autophagy, accumulation of p62 remained unchanged in AR-12-treated lkb1 larvae (Fig. 4A). This indicates that inhibition of the PI3K pathway fails to induce autophagy in mutant larvae. We next assessed phosphorylation of the mTOR-substrate RS6 upon AR-12 treatment in wt and lkb1 larvae at 6 dpf. RS6 and phosphorylated RS6 (P-RS6) were not detectable in wt larvae at 6 dpf (Fig. 4C), consistent with downregulation of mTOR activity at later developmental stages (here and refs 37, 47 and 48). AR-12 treatment did not affect the high protein levels of RS6 or P-RS6 seen in the lkb1 mutants, indicating that they are unsusceptible to PI3K pathway inhibition. In line with the lack of autophagy induction, AR-12 treatment did not enhance lkb1 survival, as no statistically significant differences were observed in the percentage of AR-12-treated lkb1 larvae alive at 9 dpf compared to DMSO-treated controls (Fig. 4D).
Activation of autophagy by an mTOR-independent pathway can be achieved using calpeptin, which inhibits the autophagy inhibitors calpain proteases 55 . Calpeptin treatment of lkb1 and wt embryos from 1 dpf onwards, in the presence or absence of chloroquine, enhanced Lc3-II levels in both wt and lkb1 larvae at 6 dpf ( Fig. 5A and Supplementary Fig. S5) without any effects on development or yolk absorption. Calpeptin treatment also resulted in upregulation of the amounts of Atg5 and complexed-Atg5 in both wt and lkb1 mutants (Fig. 5B). In contrast to rapamycin treatment, treatment with calpeptin restored p62 degradation in lkb1 mutants (Fig. 5A). Calpeptin treatment had no effect on RS6 phosphorylation in lkb1 mutants (Fig. 5C), consistent with calpeptin being an mTOR-independent autophagy activator 55 . Moreover, calpeptin-mediated activation of autophagy prolonged survival in 70% of the treated lkb1 larvae. Specifically, 17.5% of calpeptin-treated lkb1 larvae survived until 9 dpf, whereas only 1% of vehicle-treated lkb1 larvae were alive at this point (Fig. 5D). Therefore, we conclude that induction of mTOR-independent autophagy results in a more complete restoration of the lkb1 mutant phenotypes compared to that obtained upon inhibition of mTOR signaling. and Histone H3 and Tubulin (loading controls) in total protein lysates of wt and lkb1 trunks at 6 dpf. Larvae were treated with 1 μM AR-12 or DMSO (negative control) from 1 dpf, and with 2.5 μM chloroquine for 14 h prior to processing. AR-12 treatment leads to upregulation of Lc3-II levels in wt larvae but not in lkb1 mutants. p62 levels remain high in AR-12-treated lkb1 larvae. (B) AR-12 treatment leads to strong increase in the amounts of Atg5 and complexed Atg5 in wt larvae but not in lkb1 mutants. Uncropped images of the blots are shown in Supplementary  Fig. S10A,B. (C) Representative Western blot analysis of Ribosomal protein S6 (RS6), Phospho-RS6 and Tubulin (loading control). Protein levels of total RS6 and of P-RS6 were not affected by AR-12 treatment in lkb1 mutants. Uncropped images of the blots are shown in Supplementary Fig. S9A-C. (D) Graph depicting survival percentage of lkb1 larvae alive at 9 dpf. Embryos were treated with 1 μM AR-12 or DMSO from 1 dpf, collected at 9 dpf, and genotyped for the lkb1 gene. Data represent the means ± standard errors of the means (SEM) and are pooled from three independent experiments P value > 0.05, ns: not statistically significant.
Accumulation of p62 is an important regulator of autophagy in lkb1 mutants. Aberrant p62 accumulation appeared as a hallmark of impaired autophagy in lkb1 mutants, and strongly correlated with survival. While p62 is primarily thought of as a receptor delivering cargo proteins to autophagosomes for degradation, it has also been implicated in enhancing mTOR activity 56 , thereby regulating autophagy as well. Loss of p62 function led to increased autophagy in mammalian cells and in C. elegans 56 . We thus set out to determine whether reducing p62 levels in larvae would affect autophagy and survival. To this end, we injected a sqstm1/p62 MO, targeting splicing of sqstm1/p62 mRNA 54 , into 1-2-cell stage embryos. RT-PCR confirmed that the sqstm1/p62 MO blocked sqstm1/p62 mRNA splicing until at least 5 dpf (Supplementary Fig. S6). Western blot analysis of 6 dpf larvae showed decreased p62 expression compared to un-injected controls in both wt and lkb1 lysates (Fig. 6A). This was coupled with increased Lc3-II protein levels, suggestive of autophagy induction. Knockdown of p62 significantly prolonged lkb1 survival up to 9 dpf: Approximately 70% of sqstm1/p62 MO-injected lkb1 mutants survived to 9 dpf, whereas less than 5% of un-injected lkb1 larvae were alive at this time-point (Fig. 6B). Thus, depleting p62 is sufficient to activate impaired autophagy in lkb1 mutants and extend survival.

Discussion
Organisms adapt their metabolism in response to nutrient limitation to restore energy homeostasis and ensure survival. Here, we identify a novel link between metabolic adaptation during development and induction and maintenance of autophagy, mediated by the tumor suppressor Lkb1. Specifically, we use metabolically compromised Lkb1-deficient zebrafish larvae to show that Lkb1 is crucial in the induction of autophagy in response to the metabolic challenge accompanying depletion of the maternal nutrient supply. Our data therefore reveal an essential function for Lkb1 in controlling starvation-induced autophagy at the organismal level in vertebrates.
Overall autophagy levels in lkb1 mutants are lower compared to those of wt siblings: while expression of autophagy-related proteins is progressively upregulated following yolk depletion in wt larvae, induction of autophagy in lkb1 mutants is strongly attenuated. Importantly, we demonstrate that genetic and chemical manipulation of autophagy levels significantly impacts lkb1 larvae survival: inducing autophagy by mTOR-dependent andindependent mechanisms prolongs survival, and suppressing autophagy by Atg5 depletion leads to premature death selectively of the mutants. The increased susceptibility of lkb1 larvae to Atg5 depletion during development  Supplementary Fig. S11A-C. (D) Graph depicting survival percentage of lkb1 mutants alive at 9 dpf. Embryos were treated with 50 μM calpeptin or DMSO from 1 dpf, collected at 9 dpf, and genotyped for the lkb1 gene. 17,5% out of a total 25% (70% of calpeptin-treated lkb1 larvae) are alive at 9 dpf. Only 1% of DMSO-treated lkb1 larvae are alive at 9 dpf. Data depicted in (D) represent the means ± standard errors of the means (SEM) and are pooled from three independent experiments (n = 100/ experiment). **P value < 0.05. occurred even while the yolk is not yet consumed, suggesting that even though the larvae do not show a morphological phenotype at this embryonic stage, the loss of Lkb1 appears to sensitize them to additional stress. This stress may be specifically autophagy inhibition, or related to alternative mechanisms, as autophagy-independent functions have been reported for several of the autophagy-related genes 57, 58 , including Atg5 59 .
Various mechanisms, including mTOR and PI3K signaling, as well as calpains, are known to regulate autophagy 19 , and likely interact at multiple levels. Indeed, our results, together with published work, indicate that all these influence the energy-sensing defect we observe in lkb1 mutants. We show that activating autophagy by calpeptin, which inhibits the action of the general autophagy inhibitors calpains 55 , led to robust upregulation of Atg5 expression and restored degradation of p62 in lkb1 mutants. Thus, calpeptin fully rescued the autophagy defect of the lkb1 larvae and prolonged their survival. In contrast, while the mTOR-inhibitor rapamycin increased Lc3-II accumulation in lkb1 larvae, autophagy was not completely restored since p62 still accumulated. This may be due to the high mTOR activity in the mutants that could not be fully blocked by rapamycin treatment under these experimental conditions. In addition, although rapamycin treatment also prolonged lkb1 survival, we believe this was likely due to a generalized growth delay, evidenced by the presence of a considerable amount of yolk at 7 dpf (this study and refs 37 and 50), rather than due to partial restoration of autophagy. A developmental delay caused by rapamycin is further supported by the persistence of RS6 expression in rapamycin-treated wt larvae at 7 dpf when mTOR would normally be suppressed (mTOR signaling is suppressed in wt larvae upon yolk depletion at 5-6 dpf 37, 47, 48 ).
The autophagy receptor and substrate p62 aberrantly accumulates in lkb1 mutants indicating deficient autophagy. p62 is also a regulator of autophagy, as it participates in a feed-forward loop in which p62 enhances mTOR activity resulting in reduced autophagy, in turn leading to higher p62 levels in mice 60 . Here we also show that depletion of p62 in lkb1 larvae leads to activation of autophagy and prolonged survival. This implies that as the amount of p62 decreases due to autophagosomal clearance, its effect on mTOR activity is also reduced, and thus autophagy can be maintained. Furthermore, the aberrant accumulation of p62 in lkb1 larvae may in itself contribute to their premature lethality, as it has been shown that increased levels of p62 in autophagy-deficient mouse livers cause hepatotoxicity (reviewed in ref. 60). Further supporting our hypothesis, in apoptosis-impaired tumor cells with deficient autophagy, p62 accumulation triggers a positive feedback loop for the generation of reactive oxygen species (ROS) leading to enhanced genomic instability and tumorigenesis 61 .
PI3K signaling is a nutrient-sensing pathway that is also implicated in starvation-induced autophagy. Inhibition of the PI3K pathway activated autophagy in wt larvae, but not in lkb1 mutants, and did not prolong their survival. This is consistent with our previous findings that PI3K signaling is compromised in lkb1 mutants 37 . We postulate that defective PI3K signaling may contribute to the autophagy defect seen in these mutants. While AMPK is considered a major regulator of metabolism and has an important role in induction of autophagy under energetic stress 23, 44 , it is not overtly activated in wt larvae at 7 dpf 37 ; in agreement with these data, studies in mice have also reported that 24 hours of fasting did not lead to significant AMPK activation 62,63 . Thus, the autophagy defect we describe in lkb1 mutants is unlikely to be solely attributable to impaired AMPK signaling, and deregulation of additional pathways, such as PI3K signaling and AMPK/mTOR-independent pathways may also be involved. Hence, nutrient-sensing pathways (like the PI3K pathway) and energy-sensing pathways (like the AMPK pathway) are likely in close cross-talk with each other, not only through their convergence on mTOR signaling but also through different, mTOR-independent mechanisms.
Together, our data indicate that Lkb1 plays an important role in the regulation of autophagy at the whole-organism level, and confirm that autophagy is critical for survival during the metabolic transition in Figure 6. p62 knock-down extends lkb1 mutants' survival. (A) Representative Western blot analysis of p62, Lc3-II, and histone H3 (loading control) in total protein lysates of wt and lkb1 trunks at 6 dpf that were either injected with an sqstm1MO at the one-cell stage or controls. The larvae were treated with 2.5 μM chloroquine for 14 h prior to processing. p62 knock-down leads to increased Lc3-II levels in both wt and lkb1 larvae. p62 expression in lkb1 larvae is reduced upon p62 knock-down. Uncropped images of the blots are shown in Supplementary Fig. S12. (B) Graph depicting survival percentage of lkb1 larvae alive at 9 dpf. Embryos were injected with 0,5 mM sqstm1MO at the one cell stage, collected at 9 dpf, and genotyped for the lkb1 gene. Data represent the means ± standard errors of the means (SEM) and are pooled from three independent experiments. ***P value ≤ 0.0001. development. Since defects in autophagy are implicated in a plethora of diseases, a better understanding of the upstream regulatory pathways could provide new insights into their pathophysiology.

Materials and Methods
Zebrafish strains and Screening Methods. Zebrafish were handled in compliance with the local animal welfare regulations and were maintained according to standard protocols (zfin.org). Their culture was approved by the local animal welfare committee (DEC) of the University of Leiden and all protocols adhered to the international guidelines specified by the EU Animal Protection Directive 2010/63/EU. Genotype analysis for lkb1 mutants embryos was performed as previously described 37 .
Longitudinal analysis of survival of lkb1 mutants. Larvae obtained from single matings of heterozygous lkb1 adults were analyzed over time. 48-95 larvae were genotyped on 6, 7, 8, 9, 10 and 11 dpf to assess the numbers of lkb1 mutants alive.
Immunohistochemistry and Immunofluorescence. For transverse sections, larvae were fixed in 40% ethanol, 5% acetic acid and 10% formalin for 3 h at room temperature followed by three washes in 70% ethanol before being dehydrated following serial washes in Histoclear and reducing ethanol concentrations. Larvae were then sectioned at 5 μm intervals using a Reichert-Jung 2050 microtome (Leica). Sections were deparaffinized and hydrated following by 20 min of antigen retrieval in sodium citrate buffer pH 6.0 at 100 °C. Sections were blocked in 5% BSA in PBS − 0.1% Tween-20 for 1 h at room temperature and incubated overnight with sheep anti-p62 (1:200, Abcam, #ab31545) and rabbit anti-BECN1 (1:150, Santa Cruz, #sc-11427). Endogenous peroxidase activity was blocked in 0.3% H 2 O 2 for 20 min at room temperature followed by incubation with rabbit anti-sheep antibody (1:800, Abcam, #ab6747) for 1 h at room temperature. Sections were incubated with 0.1 M imidazole prior to detection with 3,3′-diaminobenzidine (DAB) substrate and counterstaining with hematoxylin.
For immunofluorescence, larvae were fixed in 4% PFA overnight at 4 °C, embedded vertically in a 0.5% gelatin/30% albumin mixture and sectioned at 120 μm intervals using a VT1000S vibratome (Leica). Sections were transferred to the wells of a 24-well plate containing PBD (PBS + 0.1% Tween-20 and 0.5% Triton-X-100), which was then replaced with blocking solution (PBD + 1% BSA) for 1 h at RT. Sections were incubated with rabbit anti-LC3B (1:1000, Abcam, #ab51520) in blocking solution overnight at 4 °C. Sections were washed three times for 15 min in PBS-0.1% Tween-20 and incubated with secondary anti-rabbit 488 green fluorescent antibody (1:100, Thermofisher, #A11008) for 2 h at room temperature. Sections were then washed three times for 15 min in PBS-0.1% Tween-20 prior to be incubated with phalloidin-Alexa 588 (1:25, Thermofisher, #A12380) and DAPI (1:200, Thermofisher, #62248) for 30 min at room temperature in the dark and rinsed three times for 5 min with dH 2 O. Sections were then imaged using the Zeiss LSM5 Exciter confocal laser-scanning microscope. Equipment and settings. For immunohistochemistry, the sections were imaged on an upright compound Nikon Eclipse E800 microscope. The images were captured using a Nikon Digital Sight camera unit, equipped with a DS-Fi1 digital camera head and a DS-L2 camera controller. Pixel dimensions of the acquired images were W2584 X H1936 pixels, at 150 pixels/inch.
The images were processed using Photoshop CS6 software. The original images were scaled-down constraining proportions, and cropped to the area of interest. Adjustment of Image Levels was applied on whole images. Assembly of the composite figures and labeling was done on Illustrator CC2015.
Confocal images were obtained in a sequential manner using a Zeiss LSM5 Exciter Confocal Laser Scanning Microscope equipped with Argon (458, 488, 514 nm), and 405, 450 and 635 diode excitation lasers and a 40× water immersion objective (C-APOCHROMAT 40×/1.2 Water). Emission ranges were set at 420-480, 505-550 and 560-615 nm in separate channels to prevent bleeding. Images were obtained using the Leica application X software (Leica, Wetzlar, Germany) and post-acquisition data analysis was performed using ImageJ software.