Gradients of glucose metabolism regulate morphogen signalling required for specifying tonotopic organisation in the chicken cochlea

  1. James DB O'Sullivan
  2. Thomas S Blacker
  3. Claire Scott
  4. Weise Chang
  5. Mohi Ahmed
  6. Val Yianni
  7. Zoe F Mann  Is a corresponding author
  1. Centre for Craniofacial and Regenerative Biology, Faculty of Dentistry Oral and Craniofacial Sciences, King's College London, United Kingdom
  2. Research Department of Structural and Molecular Biology, University College London, United Kingdom
  3. National Institute on Deafness and Other Communication Disorders, National Institutes of Health, United States

Abstract

In vertebrates with elongated auditory organs, mechanosensory hair cells (HCs) are organised such that complex sounds are broken down into their component frequencies along a proximal-to-distal long (tonotopic) axis. Acquisition of unique morphologies at the appropriate position along the chick cochlea, the basilar papilla, requires that nascent HCs determine their tonotopic positions during development. The complex signalling within the auditory organ between a developing HC and its local niche along the cochlea is poorly understood. Using a combination of live imaging and NAD(P)H fluorescence lifetime imaging microscopy, we reveal that there is a gradient in the cellular balance between glycolysis and the pentose phosphate pathway in developing HCs along the tonotopic axis. Perturbing this balance by inhibiting different branches of cytosolic glucose catabolism disrupts developmental morphogen signalling and abolishes the normal tonotopic gradient in HC morphology. These findings highlight a causal link between graded morphogen signalling and metabolic reprogramming in specifying the tonotopic identity of developing HCs.

Editor's evaluation

Morphogens such as Sonic hedgehog and Bone morphogenetic proteins (BMP) are known to establish the tonotopic organization of the cochlea. In this paper, the authors demonstrated for the first time that differential glucose metabolism along the basilar papilla (chicken cochlea) regulates the gradient of BMP7 signaling required for establishing tonotopy of the cochlea.

https://doi.org/10.7554/eLife.86233.sa0

Introduction

Hearing relies upon the life-long function of mechanosensory hair cells (HCs) and their associated glial-like supporting cells (SCs) within the cochlea. In both mammals and birds, different frequencies stimulate HCs located at different positions along the basal-to-apical long axis of the auditory epithelium to separate complex sounds into their spectral components. This phenomenon, known as tonotopy, underlies our ability to differentiate between the high pitch of a mosquito and the low rumbling of thunder. The specific factors regulating the development of tonotopy remain largely unclear. As high-frequency HCs show increased vulnerability to insults, including ageing (Gordon-Salant, 2005), noise damage (Fettiplace and Nam, 2019; Wu et al., 2020), and ototoxicity (Forge and Richardson, 1993), awareness of the mechanisms underlying the formation of frequency-specific HC properties is crucial to understanding both acquired auditory defects, HC repair and regeneration. Enhanced knowledge of the pathways that drive specification of HC phenotypes at different frequency positions could identify novel strategies to preserve and restore high-frequency hearing loss.

Metabolism, encompassing the complex network of chemical reactions that sustain life (summarised in Figure 1), has emerged as a key regulator of cell fate and differentiation (Ito and Ito, 2016). Reciprocity between metabolic networks and the epigenome has been extensively studied in models of cancer cell biology and tumourigenesis (Kinnaird et al., 2016). Here, chromatin modifying enzymes (involved in histone acetylation and methylation) that drive cell fate switches rely upon metabolic intermediates as cofactors or substrates, highlighting a link between cell metabolism and transcriptional regulation (Campbell and Wellen, 2018). Reprogramming between glycolytic and oxidative pathways has also been reported in developing tissues, including migratory neural crest cells (Bhattacharya et al., 2020), the zebrafish otic vesicle (Kantarci et al., 2020), trophectoderm in the mouse embryo (Chi et al., 2020), and the presomitic mesoderm (Oginuma et al., 2017; Oginuma et al., 2020; Bulusu et al., 2017; Miyazawa et al., 2022). Nevertheless, a regulatory role for metabolism has not been explored in the context of cell fate and patterning in the developing inner ear epithelia. This is, in part, because the classic biochemical approaches from which our knowledge of metabolism has formed involve the destructive extraction of metabolites from a sample. Probing metabolism in this manner, although valuable, means that any spatial organisation of metabolic pathways in complex tissues is lost. As the cochlea contains multiple cell types, investigating the role of metabolism in the regulation of their development requires experimental approaches capable of interrogating metabolic pathways in live preparations with single-cell resolution.

Pathways of glucose catabolism regulating cellular NADPH/NADH.

(pink mitochondrial OXPHOS) – Glucose metabolism in mitochondria. Following its conversion from glucose during glycolysis, pyruvate is transported into the mitochondria via the mitochondrial pyruvate carrier (MPC) and enters the tricarboxylic acid (TCA) cycle. Its sequential oxidation provides reducing equivalents in the form of NADH to the electron transport chain (ETC), driving ATP production by oxidative phosphorylation (OXPHOS). (yellow glycolysis) – Cytosolic glucose flux via the main branch of glycolysis. In this process, one molecule of glucose is anaerobically converted into two molecules of pyruvate to yield two molecules of ATP. Lactate dehydrogenase (LDH) acts to maintain the pool of NAD+ necessary for glycolysis to take place by oxidising NADH upon the reduction of pyruvate to lactate. (green pentose phosphate pathway) – Cytosolic glucose flux into the oxidative branch of the pentose phosphate pathway (PPP). Running parallel to glycolysis, the PPP branches off at glucose 6-phosphate (G6P) generating NADPH and ribose 5-phosphate (R5P). PPP shuttles carbons back into the main glycolytic pathway at glyceraldehyde 3-phosphate and fructose 1,6-bisphosphate. Different pathways of glucose flux can be targeted for pharmacological intervention. Inhibitors for various metabolic branch points are indicated in red (UK5099, YZ9, 2-DOG, 6-AN, Shikonin).

We have previously demonstrated that fluorescence lifetime imaging microscopy (FLIM) provides a label-free method to identify metabolic differences between inner ear cell types (Blacker et al., 2014). By spatially resolving differences in the fluorescence decay of the reduced redox cofactors nicotinamide adenine dinucleotide (NADH) and its phosphorylated analogue NADPH (Figure 1) we can extract information about the metabolic state of a cell (Blacker et al., 2014). Here, we apply this technique to investigate a role for metabolism in specifying morphological properties of proximal (high-frequency) verses distal (low-frequency) HCs in the chick cochlea. The HC phenotypes associated with different tonotopic positions are defined using previously characterised morphometrics as read-outs (Tilney et al., 1992; Goodyear and Richardson, 1997). By applying NAD(P)H FLIM in different regions of the developing basilar papilla (BP), we identify a gradient in NADPH-linked glucose metabolism along the tonotopic axis. The NAD(P)H gradient did not originate from a tonotopic switch between glycolytic and oxidative pathways or from differences in cellular glucose uptake. We find that the metabolic gradient along the developing BP originates instead from tonotopic differences in the catabolic fate of cytosolic glucose once it has entered the cell. By modulating the flux of glucose through specific cytosolic branches, we systematically interrogated its role in specifying tonotopic properties of developing HCs. We found that the cellular balance of glucose entering the pentose phosphate pathway (PPP) and the main branch of glycolysis (Figure 1) instructs tonotopic HC morphology by regulating the graded expression of Bone morphogenetic protein 7 (Bmp7) and its antagonist Chordin-like-1 (Chdl1), known determinants of tonotopic identity (Mann et al., 2014). This work highlights a novel role for cytosolic glucose metabolism in specifying HC positional identity at the morphological level providing the first evidence of a link between metabolism and morphogen signalling in the developing inner ear.

Results

NAD(P)H FLIM reveals differences in the cellular balance between NADPH and NADH along the tonotopic axis of the developing BP

NAD and NADP are metabolic cofactors responsible for ferrying reducing equivalents between intracellular redox reactions throughout the cellular metabolic network (Figure 1). The two molecules are fluorescent in their reduced (electron-carrying) forms NADH and NADPH, a feature that is lost upon oxidation to NAD+ and NADP+. The spectral properties of NADH and NADPH are identical, meaning that their combined signal emitted from living cells is labelled as NAD(P)H. FLIM of NAD(P)H has shown significant promise for identifying changes in the metabolic pathways active at a given location in living cells.

NAD(P)H FLIM typically resolves two fluorescence lifetime components in live cells and tissues one with a duration of around 0.4 ns coming from freely diffusing NAD(P)H (τfree) and the other of 2 ns or more from the pool of NAD(P)H that is bound to enzymes and cofactors (τbound) (Skala et al., 2007; Yu and Heikal, 2009; Figure 2B). Changes in the duration of τbound indicate switching in the enzyme families that are bound to the overall NAD(P)H population. NAD(P)H FLIM can therefore report changes in metabolic state in live cells during physiological processes. We used NAD(P)H FLIM to monitor metabolism along the proximal-to-distal (tonotopic) axis of the BP during development (Figure 2E–H). The gradient observed in τbound throughout BP development, specifically at E6 and E9, is consistent with our previous work, where we showed that graded morphogen signalling along the BP establishes HC positional identity between E6 and E8 (Mann et al., 2014; Thiede et al., 2014). Around E6, when cells in the BP begin acquiring their positional identity (Mann et al., 2014; Thiede et al., 2014), we observed a significant difference in τbound along the tonotopic axis (Figure 2D–H). The proximal-to-distal gradient in τbound (Figure 2D) was also evident at E9, when a majority of cells are post mitotic (Katayama and Corwin, 1989), and at E14, when HCs functionally resemble those in a mature BP (Figure 2E, F). Changes in τbound report shifts in the cellular balance between NADPH and NADH (Blacker et al., 2014; Blacker et al., 2013; Gafni and Brand, 1976), which reflects glucose catabolism in distinct branches of glycolysis (Figure 2H). These data therefore suggest alterations in the balance between NADH- and NADPH-linked glucose metabolism along the tonotopic axis of the developing BP (Figure 2H).

A proximal-to-distal metabolic gradient in the developing chick cochlea.

(A) Two-photon fluorescence image showing NAD(P)H in a live basilar papilla (BP) explant at E14 and the origin of inherent fluorescence at the nicotinamide ring. (B) NAD(P)H fluorescence lifetime imaging microscopy (FLIM) resolves two components corresponding to freely diffusing (shorter lifetime, τfree) and enzyme bound (longer lifetime, τbound). Changes in τbound imply changes in the specific enzymes to which NAD(P)H is binding and, therefore, the metabolic state of the cell. The proportion of the total NAD(P)H population that is bound to enzymes, labelled αbound, determines the relative contribution of the two species immediately after excitation. (C–F) FLIM images of the bound NAD(P)H fluorescence lifetime signal τbound in the proximal and distal BP regions at E6 and E14. White asterisks indicate the hair cells (HCs). Higher magnification images highlight the differences in τbound between proximal and distal HCs at E14 (arrowheads). (G) Quantification of τbound during development shows a shift from NADPH to NADH producing pathways. Line graphs highlight differences in τbound between proximal (black) and distal (grey) BP regions throughout development. Scale bars = 50 μm. Data are mean ± standard error of the mean (SEM); E6: n = 6, E9: n = 4, E14: n = 6, and E16: n = 5 biological replicates. *p < 0.05, ***p < 0.001 two-way analysis of variance (ANOVA). (H) Schematic of the chick BP, indicating the proximal and distal regions. Proposed gradient in cellular NADPH/NADH and thus glucose flux along the developing BP. Bottom schematic depicts interpretation of the τbound lifetime signal reported by NAD(P)H FLIM along the proximal-to-distal axis. The gradient in τbound duration reflects differences in fate of glucose catabolism. Short lifetimes (orange) indicate NADH production and therefore glucose flux through the main glycolytic pathway. Longer lifetimes (blue) indicate NADPH production and glucose catabolism in the pentose phosphate pathway (PPP). Differences in the τbound lifetime duration thereby confer differences in the catabolic fate of glucose.

Live imaging of mitochondrial metabolism and glucose uptake along the tonotopic axis of the developing BP

The redox states of the cellular NAD(P) pools are highly interrelated with the balance between cytosolic glycolysis and mitochondrial oxidative phosphorylation (OXPHOS) (Russell et al., 2022; Ying, 2008). We therefore tested whether the gradient in cellular NADPH/NADH reflected a progressive shift from glycolytic to mitochondrial OXPHOS, differences in cellular glucose uptake, or in the catabolic fate of glucose by conducting live imaging in BP explants using a range of metabolic indicators. To assess mitochondrial metabolism, we used the potentiometric fluorescent probe tetramethyl-rhodamine-methyl-ester (TMRM), a cell permeable dye that reports mitochondrial membrane potential (ΔΨmt) in living cells (Duchen et al., 2003). TMRM reports glycolytically derived pyruvate oxidation in the mitochondrial tricarboxylic acid (TCA) cycle and the activity of the mitochondrial electron transport chain (Figure 1). 2-(N-(7-Nitrobenz-2-oxa-1,3-diazol-4-yl)Amino)-2-Deoxyglucose (2-NBDG) is a fluorescent glucose analogue that when transported into cells via glucose (GLUT) transporters provides an estimate of cellular glucose uptake (Yamada et al., 2007). Thus, the 2-NBDG fluorescence measured in a given cell after a defined period of loading reflects the rate of glucose uptake by that cell (Yamada et al., 2007).

Explants were dual loaded with 350 nM TMRM and 1 mM 2-NBDG and both fluorescence signals were analysed from single cells between E7 and E16. TMRM fluorescence revealed no significant difference in ΔΨmt between proximal and distal regions at any developmental stage (Figure 3A, B). Analysis of TMRM fluorescence revealed a consistently higher ΔΨmt in fully differentiated HCs compared to SCs from E14 onwards (Figure 3C). To rule out whether the higher TMRM fluorescence occurred due to differences in dye uptake via the HC transduction channel, explants were dual loaded with the permeant mechanoelectrical transduction (MET) channel blocker FM1-43 (Gale et al., 2001; Figure 3—figure supplement 1). These findings indicate no significant difference in mitochondrial activity along the tonotopic axis throughout development, suggesting that the gradient in NADPH/NADH reported by τbound (Figure 2C–G) does not arise from variations in the balance between cytosolic and mitochondrial ATP production, as often observed in development (Bhattacharya et al., 2021) but from differences in the specific route utilised for the processing of glucose in the cytosol. This was supported by measurements of 2-NDBG fluorescence in the same cells (Figure 3—figure supplement 2A). These analyses revealed no differences in glucose uptake along the tonotopic axis at any developmental stage or between cell types (Figure 3—figure supplement 2B, C), suggesting differences in the fate rather than overall flux of glucose underpin the gradient in τbound (i.e., NADPH/NADH).

Figure 3 with 5 supplements see all
Live imaging of mitochondrial metabolism in hair cells (HCs) and supporting cells (SCs) at different positions along the tonotopic axis.

(A) Mitochondrial membrane potential measured using tetramethyl-rhodamine-methyl-ester (TMRM) in single z-planes from image stacks in the proximal and distal regions of live basilar papilla (BP) explants. (B) TMRM fluorescence indicates a significant increase in mitochondrial activity between E7 and E9, followed by significant decrease between E9 and E14. (C) Differences in mitochondrial activity (TMRM fluorescence) between HCs and SCs along the tonotopic axis at E14. Data are mean ± standard error of the mean (SEM). **p > 0.01, ***p < 0.001 for proximal and distal regions two-way analysis of variance (ANOVA). E7: n = 6, E9: n = 5, E14: n = 6, E16: n = 6 biological replicates. HCs versus SCs n = 6 biological replicates *p > 0.05 two-way ANOVA. Scale bar = 40 μm.

Tonotopic expression of metabolic mRNAs along the proximal-to-distal axis of the developing BP

To further probe the biochemical basis for the gradient in τbound we exploited existing transcriptional data sets generated from proximal and distal regions of the developing BP (Mann et al., 2014). Prior to mRNA isolation for bulk RNA-seq and Affymetrix microarray analysis, BPs were separated into proximal, middle, and distal thirds. Data were then analysed for differential expression of metabolic mRNAs involved in NADPH regulation and cytosolic glucose flux at E6.5 and E14 (Figure 3—figure supplement 3A; Mann et al., 2014). From the combined data sets, we identified multiple genes involved in NADPH-linked glucose metabolism with differential expression along the tonotopic axis (Figure 3—figure supplement 3A). Expression of these genes was verified using RNA scope and Immunohistochemistry. The only enzyme linked with cytosolic glucose flux and cellular NADPH/NADH showing a consistent differential expression throughout development, using all three validation methods, was pyruvate kinase M2 (Pkm2) (Figure 3—figure supplement 3A, Figure 4—figure supplements 13). No probe controls were used to validate the labelling (Figure 3—figure supplement 5).

Pkm2 protein is expressed tonotopically along the developing BP

Pyruvate kinase M2 (Pkm2) is a unique splice isoform of the enzyme pyruvate kinase (PK) and catalyses the final rate-limiting step in glycolysis. Pkm2 regulates the activity of metabolic enzymes in the upper branch of glycolysis by acting as a gatekeeper and diverting glucose flux towards pyruvate production or into the PPP (Grüning et al., 2011). Given this regulatory role in the catabolic fate of cytosolic glucose (Grüning et al., 2011), and the fact that increased Pkm2 activity is linked with higher cellular NADPH (Yang and Lu, 2015), we hypothesised that in correlation with the τbound gradient, Pkm2 expression would be higher in the proximal region. Consistent with the observed gradient in NAD(P)H (Figure 2) and in Pkm2 mRNA levels (Figure 3—figure supplement 3A, Figure 4—figure supplements 13), we show higher Pkm2 protein expression in HCs but not SCs at the proximal compared to distal end of the BP between E9 and E14 (Figure 4).

Figure 4 with 3 supplements see all
The metabolic gatekeeper Pkm2 is expressed in a tonotopic gradient during basilar papilla (BP) development.

(A) BP whole-mounts labelled for the metabolic enzyme Pkm2 in proximal and distal regions throughout development (E9, E10, and E14). Images show Pkm2 expression at the level of the hair cell (HC) nuclei. Epithelial z-position was determined using Phalloidin and Calbindin staining within the same preparation (images not shown). (B–D) Quantification of Pkm2 fluorescence intensity in proximal and distal BP regions at E9, E10, and E14. HC and supporting cell (SC) regions of interest (ROIs) were determined using Phalloidin and Calbindin staining within the same preparation. Data are mean ± standard error of the mean (SEM). E9: n = 9, E10: n = 6, E14: n = 13 independent biological replicates. *p = <0.05, **p = <0.01, ***p = <0.001, ****p = <0.0001 two-way analysis of variance (ANOVA). Scale bars are 20 μm.

Higher intracellular pH in the proximal BP favours Pkm2 activity associated with an increased NADPH/NADH ratio and glucose flux into the PPP

The metabolic function of Pkm2 is determined by whether the enzyme exists as a tetramer rather than a dimer. In its dimeric form, Pkm2 functions as a metabolic switch, diverting glucose towards the PPP for biosynthesis or towards pyruvate for energy production (Nandi et al., 2020). Allosteric modifications regulating the ratio between the tetrameric and dimeric forms of Pkm2 are driven by factors in the surrounding environment including intermediate metabolites and pH (Nandi et al., 2020; Zhang et al., 2019). Given the pH-dependent nature of PKM2 allostery and that the main rate-limiting enzymes driving PPP-linked glucose metabolism display optimal activity at alkaline cytosolic pH (Alfarouk et al., 2020), we next investigated differences in intracellular pH (pHi) along the tonotopic axis using the indicator pHrodo Red. When using this probe, low pHrodo Red fluorescence reflects an alkaline pH and high fluorescence a more acidic pH.

Explants were dual loaded with the pHi indicator pHrodo Red (Figure 5) and the live probe SIR-actin to distinguish HCs from SCs (Figure 5—figure supplements 1 and 2). We identified opposing proximal-to-distal gradients in pHi in HCs and SCs along the tonotopic axis, using pHrodo Red, which reported a more alkaline pHi in HCs at the proximal compared to distal end of the organ (Figure 5). The higher pHi in the proximal region reflects a metabolic phenotype consistent with higher PPP activity and dimeric Pkm2. Overall, the higher pH and Pkm2 expression levels and the possible dimeric confirmation are consistent metabolically with a longer τbound (NAD(P)H lifetime). To investigate whether the proximal-to-distal gradient in pH was maintained at later developmental stages, we also quantified the pHrodo Red signal in HCs and SCs at E14. At later developmental stages, we find the pH gradients to be reversed (Figure 5—figure supplement 3). As tonotopic patterning and positional identity are specified between E6 and E7.5 (Mann et al., 2014), the gradient at E14 is unlikely to impact the gradient in HC morphology.

Figure 5 with 3 supplements see all
Intracellular pH varies as a function of frequency position during early basilar papilla (BP) development.

(A) Intracellular pH, reported by pHrodo Red fluorescence intensity, in the proximal and distal BP regions at E9. High fluorescence indicates acidic pH and low fluorescence a more alkaline pH. (B) Mean pHrodo Red fluorescence in measured from hair cells (HCs) and supporting cells (SCs) in proximal and distal frequency BP regions. Note the proximal-to-distal gradient in intracellular pH. Data are mean ± standard error of the mean (SEM) from 11 independent biological replicates for HCs and 12 independent biological replicates for SCs. *p = <0.05, **p = <0.01 two-way analysis of variance (ANOVA). Scale bar is 10 μm.

Cytosolic glucose metabolism is necessary for tonotopic patterning in the chick BP

Having identified tonotopic gradients in NAD(P)H τbound and Pkm2 expression, we investigated a functional role for metabolism in tonotopic patterning by systematically inhibiting glucose flux into different metabolic pathways (Figure 1). First, we blocked the entirety of cytosolic glucose metabolism using 2-deoxy-D-glucose (2-DOG), an inhibitor of the enzyme hexokinase (Barban and Schulze, 1961), which occurs upstream of the branching of PPP and glycolysis. BP explants were established at E6.5 and maintained for 7 days in vitro (DIV) to the equivalent of E13.5, in control medium or that containing 2 mM 2-DOG supplemented with 5 mM sodium pyruvate (NaP), ensuring adequate substrate supply to the TCA cycle. In a normal BP, proximal HCs have larger luminal surface areas and cell bodies and are more sparsely populated compared to those in the distal region (Tilney et al., 1992; Goodyear and Richardson, 1997). These morphological gradients are recapitulated in BP explant cultures during development (Mann et al., 2014). Here, these metrics were determined by measuring differences in the HC lumenal surface area, the size of HC nuclei and the HC density within defined regions of interest (ROIs) (100 × 100 μm2) along the length of the organ. Lumenal surface area was measured using Phalloidin staining at the cuticular plate and nuclear size with DAPI (Figure 6A).

Figure 6 with 3 supplements see all
Blocking cytosolic glucose metabolism at key stages of cochlear development induces distal-like phenotypes in the proximal basilar papilla (BP).

(A, B) Maximum z-projections of BP explants showing Phalloidin and DAPI staining in the proximal and distal regions. Explants were maintained from E6.5 for 7 days in vitro (equivalent to E13.5) in either control medium or medium supplemented with 2 mM 2-deoxy-D-glucose (2-DOG) + 5 mM sodium pyruvate (NaP). Phalloidin staining depicts differences in hair cell (HC) morphology between proximal and distal regions and DAPI indicates the gradient in HC size. (C) HC lumenal surface area measured in 2500 μm2 regions of interest (ROIs) in the proximal (black bars) and distal (grey bars) BP regions for all culture conditions. In controls, mean lumenal surface decreases progressively from the proximal-to-distal region. This gradient is abolished if glucose catabolism is blocked with 2-DOG between E6.5 and E13.5. 2-DOG caused a significant decrease in HC size in the proximal but not distal region. 2-DOG treatments were reduced to 24 or 48 hr to identify the developmental time window during which glycolysis takes effect. Following wash-out of 2-DOG after 24 hr, explants developed with normal HC positional identity. Explants treated with 2-DOG for 48 hr showed no recovery of positional identity following wash-out indicated by the flattening of HC morphology along the BP. (D) Quantification of HC nuclei area in the same 2500 μm2 ROI areas. Treatment with 2-DOG induced similar, yet less pronounced effects to those seen at the HC cuticular plate. Data are mean ± standard error of the mean (SEM). *p < 0.05, ***p < 0.001 two-way analysis of variance (ANOVA). Controls; n = 6; controls LSA; n = 3; 2-DOG, n = 5; 24 2-DOG, n = 3; and 48 hr 2-DOG, n = 3 biological replicates. Red stars indicate two-way ANOVA tests between proximal control and proximal 2-DOG and distal control and distal 2-DOG conditions. To ensure adequate substrate supply to the tricarboxylic acid (TCA) cycle, 2-DOG-treated explants were supplemented with NaP. G6P – glucose 6-phosphate, F6P – fructose 6-phosphate, F16BP – fructose 1,6-bisphosphate, 2-DOG – 2-deoxyglucose. Scale bars are 20 μm.

In control cultures, HCs developed with the normal tonotopic morphologies (lumenal surface area, nuclear size and gross bundle morphology) (Figure 6A, C, D, Figure 6—figure supplement 1). In contrast, when glucose catabolism was blocked between E6.5 and E13.5 equivalent, tonotopic patterning was abolished. This was indicated by a uniformly more distal-like HC phenotype along the BP (Figure 6A, B). In addition to changes in HC morphology, treatment with 2-DOG caused a significant increase in HC density in the proximal but not distal BP region (Figure 6—figure supplement 2A, C) again consistent with loss of tonotopic patterning along the organ.

Changes in glucose metabolism have been linked with reduced cellular proliferation (Zhao et al., 2019). We therefore investigated the effects of 2-DOG on proliferation in developing BP explants. We hypothesised that because the majority of cells in the BP are postmitotic by E10 (Katayama and Corwin, 1989), adding 5-ethynyl-2′-deoxyuridine (EdU) to cultures in the presence and absence of 2-DOG for 48 hr between E8 and E10 would capture any 2-DOG-dependent differences in proliferative capacity. We observed a consistent reduction in proliferation throughout the whole explant when glucose metabolism was blocked with 2-DOG (Figure 6—figure supplement 2B, D). Increased proliferation is therefore unlikely to account for the higher cell density observed in the proximal region following the inhibition of glycolysis. Further studies are needed to determine the specific mechanisms underlying this frequency-specific increase in HC density.

As shown in our previous work, reciprocal morphogen gradients of Bmp7 and Chdl1 establish HC positional identity at the morphological level along the developing BP between E6.5 and E8 (Mann et al., 2014). To determine whether cytosolic glucose metabolism acts during this same developmental window, we blocked hexokinase activity for defined periods during BP development using 2-DOG. Explants were established at E6.5 and treated for either 24 or 48 hr followed by wash out with control medium.

These treatments correspond to the developmental window (E6.5–E8) described previously for refinement of tonotopic morphologies in developing HCs along the proximal-to-distal axis (Mann et al., 2014). The gradient in HC morphology developed normally in BPs treated with 2-DOG for 24 hr but was absent in those treated for 48 hr (Figure 6C–D, Figure 7A & C and D). These results suggest that glucose metabolism acts within the same developmental time window as Bmp7 and Chdl1 to set up tonotopic patterning along the BP. These findings are also consistent metabolically with the proximal-to-distal gradient in NADPH/NADH (τbound) observed at E6, E9, and E14 (Figure 2G).

Figure 7 with 2 supplements see all
Glucose flux through the pentose phosphate pathway modulates hair cell development and positional identity along the tonotopic axis of the basilar papilla (BP).

(A, B) Maximal z-projections of BP explants cultured from E6.5 for 7 days in vitro with control medium or medium containing 2 μM 6-aminonicotinamide (6-AN). Images show the epithelial surface in proximal and distal BP regions stained with Phalloidin. (C) Treatment of explants with 6-AN, a specific blocker of flux through the pentose phosphate pathway, caused a significant reduction in proximal compared to distal hair cell lumenal surface area. (D, E) Quantification of hair cell luminal surface area, nuclear area and cell density in defined 2500 μm2 regions of interest (ROIs) from the proximal (black bars) and distal (grey bars) BP regions. Data are mean ± standard error of the mean (SEM). **p = <0.01, ***p = <0.001 two-way analysis of variance (ANOVA). Nuclei: controls n = 6, 48 hr n = 3, 7 days n = 5; LAS: controls n = 5, 48 hr n = 3, 7 days n = 5; cell density: controls n = 5, 7 days n = 4. Independent biological replicates. Red stars indicate statistical significance for proximal and distal regions when compared between control and 6-AN conditions. Scale bars are 10 μm. G6P – glucose 6-phosphate, F6P – fructose 6-phosphate, F16BP – fructose 1,6-bisphosphate.

To further confirm a role for cytosolic glucose metabolism in establishing HC positional identity, we employed a second method of modulating the pathway independently of hexokinase activity (Pascale et al., 2019). The rate of cytosolic glycolysis can be lowered indirectly by raising cytosolic levels of the metabolite S-adenosyl methionine (SAM). Consistent with 2-DOG treatment, explants incubated with SAM between E6.5 and E13.5 lacked correct tonotopic patterning indicated by the flattening of HC morphologies along the tonotopic axis (Figure 6—figure supplement 3).

Flux through the PPP is important for tonotopic patterning along the BP

Studies in other systems have linked both the PPP and TCA cycle with cell fate decisions during development and differentiation (Chi et al., 2020; Arnold et al., 2022). We therefore sought to further dissect the metabolic signalling network during specification of tonotopy in the developing BP. To investigate a role for PPP-linked glucose metabolism, BP explants were established as described, and treated with 50 mM 6-aminonicotinamide (6-AN) between E6.5 and E13.5 equivalent. Treatment with 6-AN inhibits the rate-limiting PPP enzyme glucose 6-phosphate dehydrogenase (Figure 1). By comparison with control cultures, inhibition of PPP metabolism caused a significant decrease in HC size within 2500 μm2 areas measured in the proximal BP (Figure 7, Figure 7—figure supplement 1).

To determine whether this effect was specific to glucose flux through the PPP we also blocked phosphofructokinase (PFK), a rate-limiting enzyme further down in the glycolytic pathway, using 1 mM YZ9 (Figure 7—figure supplement 2). Blocking PFK activity inhibits the glycolytic cascade involved in pyruvate production but does not change the activity of G6PD in the PPP (Chi et al., 2020). YZ9 treatment resulted in a reduction in HC size, especially in the proximal region leading to a reduction in the HC gradient when analysed using pairwise comparisons (Sidak’s multiple comparisons, p = 0.17). However, YZ9 was unique among our metabolic inhibitor treatments in that it did not produce a significant interaction term in our two-way analysis of variance (ANOVA) (Figure 7—figure supplement 2C and Source data 1 and Source data 2). 2-DOG, SAM, and 6-AN treatments conversely all produced significant interaction terms after treatment, indicating a reduction in the normal proximal–distal gradient in cell size similar to that observed after Chdl1 and Bmp7 treatments (Source data 1 and Source data 2). Therefore, whilst blocking glycolysis and the PPP elicited significant changes in the cell size, disrupting the downstream branches of glycolysis while leaving PPP flux intact had no effect (Figure 7—figure supplements 1 and 2 and Source data 1 and Source data 2). Consistent with a higher NADPH/NADH ratio these findings suggest that tonotopic patterning is regulated by glucose flux in the upper branch of glycolysis, rather than by enzymes in the lower branch of the pathway.

Combined activity in PPP and Pkm2 pathways is required for specification of HC positional identity and patterning in the developing BP sensory epithelium

Having identified graded differences in Pkm2 expression (Figure 3—figure supplement 3, Figure 4), we investigated whether the enzyme is required for specification of HC morphologies during development. Although both Pfk and Pkm2 regulate glycolysis, the two enzymes regulate different stages of the pathway with respect to glucose flux and the re-entry of PPP products. Inhibition of Pkm2 would cause build-up of both G3p and F6p, whereas inhibition of PFK would only lead to accumulation of F6P. We therefore hypothesised that Pkm2 inhibition would have a greater impact PPP activity and tonotopic patterning.

Pkm2 activity was blocked during development from E6.5 to E13.5 equivalent using the pharmacological inhibitor shikonin (Figure 8). Treatment of explants with shikonin during HC formation abolished the normal tonotopic gradient in HC lumenal surface area (Figure 8A, C), where HC cuticular plate circumference was significantly increased in the distal but not proximal BP region compared to controls. Absence of Pkm2 activity did not alter HC nuclear area (Figure 8D), suggesting that the enzyme could play a more important role regulating HC morphology and patterning at the epithelial surface.

Glucose catabolism in pentose phosphate pathway (PPP) and Pkm2 pathways is necessary for specifying correct hair cell (HC) morphology and patterning along the tonotopic axis of the developing basilar papilla (BP).

(A) Maximal z-projections showing Phalloidin staining at the epithelial surface in proximal and distal BP regions of control and shikonin and shikonin + 6-aminonicotinamide (6-AN)-treated explants. (B) Schematic illustrating the metabolic enzymes targeted by shikonin and 6-AN. (C, D) Quantification of HC lumenal surface area and nuclear are in control and shikonin BP explants established at E6.5 and maintained for 7 days in vitro. (E, F) Quantification of HC lumenal surface area and nuclear are in control and shikonin + 6 AN explants treated from E6.5 for 7 days in vitro. HC circumference, nuclear area, and HC density were quantified in 2500 μm2 regions of interest (ROIs) in proximal and distal regions. Data are mean ± standard error of the mean (SEM). *p = <0.05, **p = <0.01, ***p = <0.001 two-way analysis of variance (ANOVA). Controls: n = 5 for LSA, n = 6 for nuclear area, shikonin: n = 4, shikonin + 6-AN: n = 6. Scale bars are 20 μm.

2-DOG blocks the activity of all pathways in cytosolic glycolysis, a network that encompasses both Pkm2 and PPP-linked glucose catabolism. Having probed Pkm2 and PPP metabolism independently we tested whether blocking both pathways during development was additive and could mimic the inhibitory effects of 2-DOG on HC positional identity (Figure 6). We find that blocking both pathways during development abolished the normal gradient in HC morphology along the tonotopic axis, thus mimicking the effects seen when inhibiting glycolysis using 2-DOG (Figure 8A, E, F). These data therefore suggest that metabolic activity in both PPP and Pkm2 pathways is important for establishing HC positional identity along the developing BP.

Pyruvate-mediated OXPHOS in mitochondria maintains progression of HC development but does not regulate tonotopic patterning

In the developing BP, live imaging of mitochondrial activity using TMRM revealed no apparent difference in OXPHOS along the tonotopic axis. To determine whether mitochondrial metabolism influences tonotopic patterning during development, we blocked uptake of glycolytically derived pyruvate into mitochondria by inhibiting the mitochondrial pyruvate carrier with UK5099 (Zhong et al., 2015; Figure 9B). HCs in explants treated with UK5099 between E6.5 and E13.5 developed with abnormal morphologies at all positions along the BP and displayed either immature stereociliary bundles or lacked them completely (Figure 9A, C, red arrowheads). To determine whether this effect was due to a global developmental arrest, explants were also established at E8, by which time tonotopy is specified (Mann et al., 2014) but bundles are not yet developed, and maintained for 7 DIV to the equivalent of E15. Compared to controls, HCs of explants treated with UK5099 were smaller and displayed abnormal bundle morphologies at all positions along the BP (Figure 9C, Figure 9—figure supplement 1). The role that mitochondria play in shaping HC morphologies and functional properties at different frequency positions is at present unclear and will require further investigation. However, our findings suggest that pyruvate-mediated OXPHOS plays a more significant role in maintaining the overall progression of development rather than regulating the patterning along the tonotopic axis.

Figure 9 with 1 supplement see all
Mitochondrial OXPHOS is necessary for the normal developmental progression of hair cells (HCs) but not positional identity.

(A) HC morphology at the surface of the basilar papilla (BP) epithelium in explants stained with Phalloidin. Cultures were established at E6.5 and maintained for 7 days in vitro in control medium or that supplemented with the mitochondrial inhibitor UK5099. (B) Blocking pyruvate uptake into mitochondria with UK5099 disrupts normal tricarboxylic acid (TCA) cycle activity and thus mitochondrial OXPHOS by blocking pyruvate uptake via the mitochondrial pyruvate carrier (MPC1/MPC2). (C) Blocking mitochondrial OXPHOS from E6.5 to E13.5 equivalent caused developmental abnormalities in HCs along the BP including reduced HC size and immature stereocilial bundles (red arrowheads in A) in both proximal and distal regions compared to controls. To test whether mitochondrial OXPHOS is required for overall developmental progression, cultures were also established at E8 and treated with 1 μM UK5099 for 7 days in vitro to the developmental equivalent of E15. HCs showed no apparent recovery of normal morphology following UK5099 treatment from E8 compared to E6.5. Data are mean ± standard error of the mean (SEM). ***p < 0.001, two-way analysis of variance (ANOVA). Controls: n = 8, UK5099 E6.5: n = 4, E8: n = 4.

Glucose metabolism regulates expression of Bmp7 and Chdl1 along the tonotopic axis

In many developing systems, gradients of one or more morphogen act to regulate cell fate, growth and patterning along a given axis (Towers et al., 2012; Averbukh et al., 2014). In the chick cochlea, reciprocal gradients of Bmp7 and its antagonist Chdl1 play key roles in determining HC positional identity. As disruption of both the normal gradient in Bmp7 (Mann et al., 2014) and glucose metabolism induce similar defects in morphological patterning (Figure 6, Figure 10A–D), we investigated the possibility of a causal interaction between the metabolic and the morphogen signalling networks in the developing BP.

Figure 10 with 2 supplements see all
A tonotopic gradient in NAD(P)H producing glucose metabolism specifies hair cell (HC) positional identity along the basilar papilla (BP) by regulating gradients of Bmp7 and Chdl1.

(A) Phalloidin staining at the surface of BP explants in the proximal and distal regions. Explants were established at E6.5 and incubated for 7 days in vitro in control medium or medium containing 2-deoxy-D-glucose (2-DOG) + sodium pyruvate (NaP) or Bmp7. (B, C) Treatment with 2-DOG- or Bmp7-induced HC morphologies consistent with a more distal phenotype in the proximal BP. HC lumenal surface area was determined using Phalloidin staining at the cuticular plate in 2500 μm2 areas. (D) Treatment with Bmp7 between E6.5 and E13.5 equivalent results in increased HC density in the proximal BP region. HC density was counted in proximal and distal BP regions using defines regions of interest (ROIs) of 10,000 μm2. (E) Treatment of explant cultures with 2-DOG + NaP from E6.5 for 72 hr in vitro disrupts the normal tonotopic expression of Bmp7 and its antagonist Chdl1. Images show in situ hybridisation for Bmp7 and Chdl1 in BP whole-mounts treated with 2-DOG + NaP from E6.5 for 72 hr in vitro. Images are representative of 6 biological replicates. (2-DOG) Controls: n = 6, 2-DOG: n = 6. Data mean ± standard error of the mean (SEM). **p < 0.01 two-way analysis of variance (ANOVA). (Bmp7) Controls: n = 11, Bmp7: n = 10. Data mean ± SEM. **p < 0.01 two-way ANOVA. Scale bars (A) control scale bar is 20 μm, DOG and Bmp7 are 50 μm. Scale bars for in situ data (E) are 10 μm. (F) Schematic of the chick BP, showing the graded differences in HC size and density along the tonotopic axis. The opposing gradients in Bmp7 activity and in cellular NAD(P)H/NADH (glycolysis) are indicated. Red boxes indicate regions of measurement for HC lumenal surface areas and cell density.

To investigate the regulatory effects of cytosolic glucose metabolism on the expression gradients of Bmp7 and Chdl1, explants were established at E6.5 and maintained for 72 hr in vitro (equivalent of E9.5) in control medium or that containing 2-DOG + NaP. Whole-mount in situ performed on explant cultures showed that disrupted glucose metabolism altered the normal expression of Bmp7 and Chd1along the BP (Figure 10E). Following treatment with 2-DOG, Bmp7 expression appeared to increase along the entire BP while Chdl1 showed a reciprocal decrease along the length of the organ. It is challenging to predict how this global change in expression levels would impact the activity of each morphogen along the tonotopic axis, but it does support our hypothesis that there is a causal interaction between glycolytic and Bmp7-Chdl1 networks. The precise nature of this interaction requires further investigation. We speculate that the increased Bmp7 and reduced Chdl1 expression in the proximal region (Figure 10E, Figure 10—figure supplement 1), in response to perturbed glucose flux (by treating with 6-AN), would induce expansion of distal-like HC morphologies into the proximal region. Blocking mitochondrial-linked glucose catabolism with UK5099 did not alter the expression of Bmp7 (Figure 10—figure supplement 2).

Treatment with Chdl1 restores normal tonotopic patterning when glycolysis is blocked during development

Modulating the reciprocal gradients of Bmp7 and Chdl1 along the proximal-to-distal axis alters tonotopic patterning in nascent HCs (Mann et al., 2014). We further show that the normal gradients of Bmp7 and Chdl1 are disrupted along the BP when glycolysis or PPP activity are blocked during development (Figure 10E, Figure 10—figure supplement 1). As treatment with 2-DOG causes a global increase in Bmp7 and decrease in Chdl1 expression (Figure 10E), we therefore hypothesised that treatment of explants with 2-DOG in the presence of Chdl1 protein might restore tonotopic patterning when glycolysis is blocked. Analysis of explants treated between E6.5 and E13.5 equivalent with 2-DOG + 0.4 μg/ml Chdl1 showed a partial rescue of HC morphologies (lumenal surface area and nuclear area) along the proximal-to-distal axis (Figure 11, Figure 11—figure supplement 1, and Source data 1 and Source data 2). Whilst the precise length and number of stereocilia could not be accurately quantified using Phalloidin staining in these explants, the overall bundle morphology also appeared consistent with that reported previously for proximal and distal BP regions (Thiede et al., 2014; Tilney and Saunders, 1983; Son et al., 2015).

Figure 11 with 1 supplement see all
Chdl1 restores tonotopic patterning along the basilar papilla (BP) when cytosolic glycolysis is blocked with 2-deoxy-D-glucose (2-DOG).

(A) Maximum z-projections of the epithelial surface in the proximal and distal regions of Phalloidin-stained BPs. Explants were established at E6.5 and maintained for 7 days in vitro in either control medium or that containing 2-DOG and Chdl1. Phalloidin staining indicates differences in hair cell (HC) lumenal surface area and gross bundle morphology between proximal and distal regions. The tonotopic gradient in HC lumenal surface was restored when explants were treated with 2-DOG in the presence of Chdl1. (B) The gradient in HC nuclear area was maintained along the proximal-to-distal axis when explants were treated with 2-DOG in the presence of Chdl1. Effects of Chdl1 were only apparent at concentrations of 0.35 μg/ml or above. (C) HC lumenal surface area measured from 2500 μm2 regions of interest (ROIs) in proximal (black) and distal (grey) BP regions. Effects of different Chdl1 concentrations on the HC luminal surface are indicated. Data are mean ± standard error of the mean (SEM). ** p < 0.01, *** p < 0.001 two-way analysis of variance (ANOVA). Dose–response data – Controls: n = 6, Chdl1 0.25 n = 4, 0.35 n = 4, 0.4 n = 4. Chdl1 + 2-DOG: Controls n = 6, Chdl1 + 2-DOG n = 4 biological replicates for HC lumenal surface areas and n = 3 for nuclear area. Scale bars are 10 μm.

Taken together, our findings suggest that a distinct metabolic state coupled with a specific morphogen level can regulate HC morphology at different positions along the tonotopic axis during development. These data also provide further evidence indicating a causal interaction between metabolic and morphogen signalling networks during development. Ascertaining a role for cytosolic glucose metabolism in specifying proximal verses distal HC fate, specifically related to frequency tuning, would require a detailed analysis of HC physiological properties. Future work should therefore determine whether altering metabolism affects the developmental acquisition of not only HC morphology, but also the intrinsic electrophysiological properties and firing characteristics documented for HCs at different tonotopic positions (Fuchs and Evans, 1990; Fuchs et al., 1988; Fettiplace and Fuchs, 1999).

Discussion

Generating new HCs that recapitulate the features of those in a healthy cochlea requires a detailed knowledge of the cell biology driving their formation. As high-frequency HCs are more vulnerable to insult, there is also a need to understand differences in the specific factors and signalling pathways that drive the identity of distinct HC subtypes. Over recent years, metabolism has emerged as a key driver of cell fate and function across various biological systems and cellular contexts (Ghosh-Choudhary et al., 2020; Tarazona and Pourquié, 2020). Taking both qualitative and quantitative approaches, we identify regional differences in metabolism along the frequency axis of the developing chick cochlea and explore a role for causal signalling between graded morphogens and glucose metabolism in establishing HC positional identity. We identify a tonotopic gradient in cellular NADPH, originating from differences in glucose flux between high- and low-frequency HCs. Tonotopic differences in the catabolic fate of glucose in glycolysis or the PPP modulates Bmp7 and Chdl-1 signalling along the developing BP. This study provides the first evidence supporting a role for crosstalk between metabolism and morphogen gradients in the developing auditory system, building on our current understanding of cell fate specification.

NAD(P)H FLIM reveals a gradient in metabolism along the tonotopic axis of the developing chick cochlea

Using NAD(P)H FLIM, we uncovered a proximal-to-distal gradient in cellular NADPH resulting from tonotopic differences in the fate of cytosolic glucose. The biochemical basis for this gradient was further investigated by exploring differences in mitochondrial activity, pHi, and the expression of metabolic enzymes along the tonotopic axis. Collectively, these analyses indicate that differences in the fate of cytosolic glucose, (Grüning et al., 2011; Yang and Lu, 2015; Alfarouk et al., 2020) rather than between glycolytic and oxidative metabolic states underpin the increased τbound lifetime and higher NADPH/NADH ratio reported here for high-frequency HCs. The higher expression of PKM2 and the more alkaline pH identified in proximal HCs are also consistent with a higher cellular NADPH/NADH ratio (Nandi et al., 2020; Zhang et al., 2019).

PPP metabolism, HC size, and positional identity

Cell geometry and size contribute to overall tissue architecture during development and are important for long-term function of the cochlea in vertebrates. Cell size, cell membrane composition, and metabolic rate are tightly correlated (Hulbert and Else, 1999). PPP-derived NADPH is utilised extensively in proliferating cells during development, where it regulates cell cycle progression, differentiation, and growth (Vander Heiden et al., 2009; Vizán et al., 2009). Increased glucose metabolism has been reported in regenerating (Love et al., 2014; Patel et al., 2022) and developing systems (Chi et al., 2020; Oginuma et al., 2017; Bulusu et al., 2017; Miyazawa et al., 2022) where it plays an important role in regulating cell fate, behaviour, and shape. However, the chosen path of glucose catabolism is context dependent and differs across processes and between tissues. PPP metabolism regulates cell division and proliferation through its ability to generate lipid and nucleotide precursors (Stincone et al., 2015). Increased glucose flux into the PPP but not the main branch of glycolysis was also recently shown to regulate the balance between proliferation and cell death in the regenerating limb (Patel et al., 2022). In the BP, HCs exit the cell cycle in three progressive waves following a centre-to-periphery progression beginning at E5. During this process, both HCs and SCs are added in apposition, to the edges of the band of post-mitotic cells that preceded them (Katayama and Corwin, 1989). HC differentiation then begins in the distal portion of the BP at around E6 and extends proximally along the cochlea expanding across the width of the epithelium (Goodyear and Richardson, 1997; Cotanche and Sulik, 1984; Bartolami et al., 1991). These graded differences in HC size along the organ are an essential requirement for correct auditory coding (Fettiplace and Fuchs, 1999). As larger cell size is correlated with increased G6PDH activity and thus more glucose flux into the PPP (Vizán et al., 2009), the higher activity in the proximal BP region may underlie the graded differences in HC size that arise during development. It could be argued that the smaller cell size induced following metabolic perturbation in the proximal BP is a result of impaired differentiation. However, given that distal HCs maintain their small size in the mature organ, this morphological change observed in the proximal region following treatment with 2-DOG and 6-AN is consistent with a distal-like phenotype. Without a detailed characterisation of distal and proximal cell growth from E6 through to adult stages, this puzzle is challenging to resolve.

PPP metabolism is also closely linked with de novo synthesis of lipids and cholesterol, which form an integral part of cell membranes (Sykes et al., 1986). Functional interactions between ion channel complexes in the membrane and the local lipid environment have been described previously in mammalian (Hirono et al., 2004; Gianoli et al., 2017) and avian HCs (Purcell et al., 2011). Frequency tuning in the BP relies on the intrinsic electrical properties of the HCs themselves, where graded differences in the number and kinetics of voltage-gated calcium channels and calcium-sensitive (BK type) potassium channels underlie the ability of HCs to resonate electrically in response to sound (Fettiplace and Fuchs, 1999). By regulating cholesterol in the HC membrane, the graded PPP activity observed here may also regulate aspects of HC electrical tuning.

A causal link between metabolism and morphogen signalling during development sets up HC positional identity

Morphogen signalling gradients have well defined roles in directing cell identity along developing axes, where cells determine their fate as a function of morphogen concentrations at different positions along them (Towers et al., 2012; Dessaud et al., 2008; Sagner and Briscoe, 2019). We showed previously that reciprocal gradients of Bmp7 and Chdl1 establish HC positional identity along the developing BP (Mann et al., 2014). The gradient of Bmp7 is established by Sonic hedgehog (Shh) signalling emanating from ventral midline structures, including the notochord and floor plate (Son et al., 2015). Here, we identify a gradient in glucose metabolism that regulates the morphology of developing HCs along the tonotopic axis through a causal interaction with the Bmp7–Chdl1 network. Disrupting this gradient using 2-DOG, SAM, or 6-AN mimics the effects on HC morphology reported previously for altered Shh (Son et al., 2015) and Bmp7 signalling (Mann et al., 2014), which induced distal-like HC phenotypes in the proximal BP. Furthermore, the effects of impaired glycolysis could be partially rescued in explants treated with 2-DOG in the presence of Chdl1. Our findings thus indicate a complex and causal interplay between Bmp7 and Chdl1 morphogen gradients and glucose metabolism in the specification of HC tonotopic identity. Metabolic gradients are also known to regulate elongation of the body axis and somite patterning (Oginuma et al., 2017; Miyazawa et al., 2022). By establishing a gradient in intracellular pH, glycolysis drives graded Wnt signalling and specifies mesodermal versus neuronal cell fate along the developing body axis (Oginuma et al., 2017; Oginuma et al., 2020; Bulusu et al., 2017). The specific role of the pHi gradient along the developing BP is unclear, however given the importance of the Wnt signalling pathway in both cochlear development and regeneration and repair, it will be important to investigate crosstalk between the two in the context of HC formation and tonotopic identity.

In conclusion, our findings indicate a causal link between PPP activity and graded morphogen signalling in specifying HC morphology along the tonotopic axis during development. However, a detailed physiological analysis is required to accurately confirm whether these morphological changes reflect a switch between proximal and distal HC fate. Future work should determine how altering the fate of glucose affects the morphological and functional development of the stereociliary bundle, the intrinsic electrophysiological properties and the firing characteristics of HCs at different tonotopic positions. Untangling further the interactions between the components of the Shh, Bmp7, and metabolic signalling networks will advance our understanding of how HCs acquire the unique morphologies necessary for auditory coding. From what we understand about frequency selectivity in vertebrates (Gleich et al., 2005), recapitulation of tonotopy will require that any gradient, and its associated signalling networks, scale correctly in different inner ear sensory patches and across species with varying head size and cochlear lengths. Understanding how the mechanical constraints associated with growth and patterning in different sense organs modulate these networks will advance our understanding of how to drive formation of specific HC phenotypes in inner ear organoid models.

Materials and methods

Embryo care and procedures

Request a detailed protocol

Fertilised White Leghorn chicken (Gallus gallus domesticus) eggs (Henry Stewart & Co Ltd, UK) were incubated at 37.5°C in an automatic rocking incubator (Brinsea) until use at specific developmental stages between embryonic day 6 (E6) and E16. Embryos were removed from their eggs, staged according to Hamburger and Hamilton (1951) and subsequently decapitated. All embryos were terminated prior to hatching at E21. All procedures were performed in accordance with United Kingdom legislation outlined in the Animals (Scientific Procedures) Act 1986.

Preparation of BP explants for live imaging studies

Request a detailed protocol

BPs were collected from chick embryos between E7 and E16, and explants were established at E13 to E16 in accord with United Kingdom legislation outlined in the Animals (Scientific Procedures) Act 1986. Explants were placed onto Millicell cell culture inserts (Millipore ) and maintained overnight at 37°C in medium 199 Earl’s salts (M199) (Gibco, Invitrogen) containing 2% foetal bovine serum and 5 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) buffer (Life Technologies). For live imaging experiments, cultures were transferred to glass-bottom 50 mm MatTek dishes and held in place using custom-made tissue harps (Scientifica). Cultures were maintained in L-15 medium at room temperature throughout the experiment.

BP culture

Request a detailed protocol

BPs were isolated from embryos incubated for between 6 (E6.0) and 8 (E8.0) days and maintained in chilled Leibovitz’s L-15 media (Gibco, Invitrogen). Papillae were dissected as described previously (Jacques et al., 2014) and cultured nerve-side-down on Millicell cell culture inserts (Millipore ). Cell culture inserts were placed into 35-mm culture dishes containing 1.5 ml of 199 Earl’s salts (M199) medium (Gibco, Invitrogen) supplemented with 5 mM HEPES buffer and 2% foetal bovine serum. Papillae were maintained in M199 medium plus vehicle (control media) for up to 7 DIV until the equivalent of E13.5. For all treatments, a minimum of four samples were analysed. The following factors were applied to experimental BPs in culture at the specified concentrations: 2-deoxyglucose (2-DOG) 2 mM (Sigma), NaP 5 mM (Sigma), 6-AN 2 μM (Sigma), S-(5′-adenosyl)-L-methionine chloride dihydrochloride (SAM) 50 μM (Sigma), YZ9 1 μM (Sigma), Shikonin 1 μM (Sigma), Bmp7 recombinant protein 0.4 μg/ml (R&D Systems 5666-BP-010/CF), and Chordin like-1 recombinant protein 0.4 μg/ml (R&D Systems 1808-NR-050/CF). For 2-DOG wash-out experiments, cultures were treated for 24 or 48 hr followed by wash out with control medium for the remainder of the experiment up to 7 days. For paired controls, medium was also changed at 24 and 48 hr in culture. At the conclusion of each experiment (7 DIV), cultures were fixed in 4% paraformaldehyde (PFA) for 20 min at room temperature, washed thoroughly three times with 0.1 M phosphate buffered saline (Invitrogen) and processed for immunohistochemistry.

Fluorescence lifetime imaging

Request a detailed protocol

NAD(P)H FLIM was performed on an upright LSM 510 microscope (Carl Zeiss) with a 1.0 NA ×40 water-dipping objective using a 650 nm short-pass dichroic and 460 ± 25 nm emission filter. Two-photon excitation was provided by a Chameleon (Coherent) Ti:sapphire laser tuned to 720 nm, with on-sample powers kept below 10 mW. Photon emission events were registered by an external detector (HPM-100, Becker & Hickl) attached to a commercial time-correlated single photon counting electronics module (SPC-830, Becker & Hickl) contained on a PCI board in a desktop computer. Scanning was performed continuously for 2 min with a pixel dwell time of 1.6 µs. Cell type (HC vs. SC) and z-position within the epithelium was determined prior to FLIM analysis using the mitochondrially targeted fluorescent dye TMRM. The dye was added to the recording medium, at a final concentration of 350 nM, 45 min before imaging. TMRM fluorescence was collected using a 610 ± 30 nm emission filter. Excitation was provided at the same wavelength as NAD(P)H to avoid possible chromatic aberration. The 585 ± 15 nm emission spectrum of TMRM ensured its fluorescence did not interfere with acquisition of the NAD(P)H FLIM images.

FLIM data analysis

Request a detailed protocol

Following 5 × 5 binning of photon counts at each pixel, fluorescence decay curves of the form (Blacker et al., 2014),

I(t)=Z+I0([1αbound]et/τfree+αboundet/τbound)

were fit to the FLIM images using iterative reconvolution in SPCImage (Becker & Hickl), where Z allows for time-uncorrelated background noise. Matrices of the fit parameters τfree, αbound, and τbound and the total photons counted were at each pixel, were exported and analysed for HCs and SCs, and proximal and distal BP regions, using SPCImage and ImageJ software packages.

2-NBDG, TMRM live imaging

Request a detailed protocol

BP were isolated from E7, E9, E14, and E16 chick embryos in chilled L-15 mediumand subsequently incubated in 1 mM solution of 2-NBDG (N13195, Thermo Fisher Scientific) in L-15 medium at room temperature for 1 hr. The medium was then replaced with a fresh solution of 1 mM 2-NBDG and 350 nm TMRM (T668, Thermo Fisher Scientific) in L-15 and incubated for a further hour at room temperature. Thereafter, the BPs were washed several times with fresh medium containing 350 nM TMRM and mounted in a 3.5-mm glass bottom MatTek dish. 3D image stacks with an optical thickness of 1 μm were captured using a Leica SP5 confocal microscope with an HCX PL APO ×63/1.3 GLYC CORR CS (21°C) objective.

Measurement of pHi using pHrodo Red

Request a detailed protocol

BP was dissected in cold L-15 media and incubated for 1 hr at room temperature with 5 µM pHrodo Red Intracellular pH Indicator (Invitrogen P35372) and 1 nM SiR-actin (SpiroChrome SC001) in L-15 medium. Samples were subsequently mounted in Mattek (50 mm) dishes and held in place using custom-made imaging grids. Explants were imaged using an inverted ZEISS LSM980 confocal microscope using a ×63 objective and digital zoom of ×1.8. Z-intervals were kept consistent at 0.4 µm across all developmental stages.

Immunohistochemistry

Request a detailed protocol

Inner ear tissue was collected at various developmental stages, fixed for 20 min to 1 hr in in 0.1 M phosphate-buffered saline (PBS) containing 4% PFA, and processed for whole-mounts immunohistochemistry. The BP was then fine dissected and permeabilised in PBS containing 0.3% Triton for 30 min before immunostaining using standard procedures (Mann et al., 2014). Samples were stained with primary antibodies for PKM2 1:100 (Cell Signalling 4053T). Antibody staining was visualised using secondary antibodies conjugated to either Alexa 488 or Alexa 546 (Invitrogen). Cultures were incubated with all secondary antibodies for 1 hr at room temperature 1:1000, washed thoroughly in 0.1 M PBS. Samples were then incubated for an additional hour with either Alexa Fluor-conjugated Phalloidin-546 1:250 (Invitrogen) to label filamentous actin and DAPI 1:1000 to label cell nuclei. Tissue samples were mounted using Prolong Gold antifade reagent (Invitrogen). 3D image stacks of mounted samples were captured using a Leica SP5 confocal microscope with an HCX PL APO ×63/1.3 GLYC CORR CS (21°C) objective.

EdU staining

Request a detailed protocol

Control or 2-DOG-treated cultures were incubated for 48 hr in 10 μM EdU from E8 to E10. Cultures were subsequently fixed for 15 min in 4% PFA at room temperature and then washed in 0.1 M PBS. Explants were then processed for EdU staining following the Click-iT EdU 488 protocol (Thermo Fisher Scientific).

Image analysis

Request a detailed protocol

Analysis of z-stacks from immunohistochemistry stains as well as 2-NBDG, TMRM, and pHrodo Red live imaging experiments was carried out using the Fiji distribution of ImageJ. For each sample, a z-plane 2 μm beneath the surface of the epithelium was selected using Phalloidin or SiR-actin labelling for further analysis. For each of these selected z-planes, a 100 μm × 100 μm ROI was chosen containing intact tissue in which all HCs were optimally orientated for analysis. Mean fluorescence intensity of the tissue was measured for HCs and SCs from within defined 100 μm × 100 μm ROIs at E7, E9, and E10 timepoints. At E14 and E16, HCs and SCs were manually segmented. At younger stages, when HCs and SCs were not easily identified, fluorescence intensity was measured from within the whole epithelium. HC labels were dilated by 3 μm, which provided selections which included both HCs and their surrounding SCs. By subtracting the HC segmentation from the dilated label, we were thus able to measure the fluorescence intensity of whole HCs separately from their surrounding support cells in the 2-NBDG data. A similar approach was adopted when measuring TMRM fluorescence intensity at E14 and E16. However, we noticed that signal was concentrated around the HC periphery. To ensure that the fluorescence intensity best reflected only the mitochondria and was not reduced by the low fluorescence from the centre of each HC, we measured mean fluorescence intensity only up to 2 μm from the cell membrane. Likewise, for TMRM data at E7 and E9, mitochondria were segmented using Fiji’s auto-local thresholding (Niblack) prior to intensity measurements, to avoid a biased estimate of fluorescence intensity due to empty space surrounding each mitochondrion.

Analysis of HC morphology

Request a detailed protocol

Data were analysed offline using ImageJ software. HC lumenal surface area and cell size were used as indices for HC morphology along the tonotopic axis. To determine the HC density, the lumenal surfaces of HCs and cell size, cultures were labelled with Phalloidin and DAPI. Then, the number of HCs in 50 μm × 50 μm ROI (2500  μm2 total area) located in the proximal and distal BP regions were determined. Proximal and distal regions were determined based on a calculation of the entire length of the BP or explant. In addition, counting ROIs were placed in the mid-region of the BP along the neural to abneural axis to avoid any confounding measurements due to radial differences between tall and short HCs. For each sample, HCs were counted in four separate ROIs for each position along the BP. Lumenal surface areas were determined by measuring the circumference of individual HCs at the level of the cuticular plate. Nuclear size was determined using the DAPI signal.

Statistical testing and analyses

Request a detailed protocol

All data were assessed for normality prior to application of statistical tests, with a threshold of p < 0.05 used for determining significance. When comparing between proximal and distal regions within the same tissue explant, paired t-tests with unequal variance were used. This statistical approach was chosen given that measurements were made from different regions within the same sample and were therefore not independent from each other. Comparisons made between different developmental stages were assumed independent from one another and thus here, independent t-tests and two-way ANOVAs were used.

In situ hybridisation

Request a detailed protocol

Inner ear tissue was dissected and fixed in 4% PFA overnight at 4°C. Tissue was subsequently washed three times for 30 min in 0.1 M PBS, dehydrated in ascending methanol series (25–100%) and stored at −20°C until use. Immediately before the in situ protocol, tissue was rehydrated in a descending methanol series (100–25%). Antisense digoxigenin-labelled RNA probes for Bmp7 were kindly provided by Doris Wu (NIDCD, NIH). Chd-l1 was synthesised as described previously (Mann et al., 2014). In situ hybridisation was performed at 68°C following the protocol as described previously (Wu and Oh, 1996).

RNAscope

Request a detailed protocol

Gene-specific probes and the RNAscope Fluorescent Multiplex Reagent Kit (320850) were ordered form Advanced Cell Diagnostics. BP was collected from E8 to E10 chick embryos, fixed overnight in 4% PFA, and subsequently cryopreserved through a sucrose gradient (5%, 10%, 15%, 20%, and 30%). Samples were embedded in cryomolds using Tissue-Tek O.C.T compound and sectioned on a cryostat at 10–12  μm thickness. RNAscope hybridisation protocol was carried out based on the manufacturer’s (ACD) suggestions. All fluorescent images were obtained on a Zeiss LSM900 confocal microscope.

RNA-seq analysis

Request a detailed protocol

For bulk RNA-seq analysis, all genes with a Log2 p-value >1 were considered significantly expressed in the distal BP region and all genes with a Log2 <1 significantly expressed in the proximal BP region. Statistical significance levels were calculated by one-way ANOVA. For a gene to be considered ‘differential’, at least one region of the BP (proximal, middle, or distal) was required to be ≥0.5 RPKM. A fold change of ≥2 was imposed for the comparison between distal and proximal regions. A final requirement was that middle region samples had to exhibit RPKM values mid-way between proximal and distal regions to selectively capture transcripts with a gradient between the two ends. Bulk Affymetrix data were analysed for differentially expressed mRNAs encoding metabolic effector proteins that regulate cellular NADPH levels. Microarray signals were normalised using the RMA algorithm. The mRNAs expressed at significantly different levels in distal versus proximal BP were selected based on ANOVA analysis using the Partek Genomics Suite software package (Partek, St. Charles, MO, USA). *p < 0.05. For detailed description of analysis and protocols please refer to Mann et al., 2014.

Data availability

All data and source data are available in manuscript and supporting files.

References

    1. Barban S
    2. Schulze HO
    (1961)
    The effects of 2-deoxyglucose on the growth and metabolism of cultured human cells
    The Journal of Biological Chemistry 236:1887–1890.

Decision letter

  1. Doris K Wu
    Reviewing Editor; National Institutes of Health, United States
  2. Kathryn Song Eng Cheah
    Senior Editor; University of Hong Kong, Hong Kong
  3. Doris K Wu
    Reviewer; National Institutes of Health, United States
  4. Jinwoong Bok
    Reviewer; Yonsei University, Republic of Korea
  5. Katie Kindt
    Reviewer; National Institutes of Health, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting the paper "Gradients in cellular metabolism regulate development of tonotopy along the cochlea" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, including Doris Wu as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by a Senior Editor. The following individuals involved in the review of your submission have agreed to reveal their identity: Jinwoong Bok (Reviewer #2); Katie Kindt (Reviewer #3).

Comments to the Authors:

We are sorry to say that, after consultation with the reviewers, we have decided that this work will not be considered further for publication by eLife.

Specifically, the causal relationship between the NADPH/NAD gradient and tonotopy has many outstanding issues that require further clarification. Strategies employed to inhibit cellular metabolism in this study could have many secondary effects that may not be related to the establishment of tonotopy. Furthermore, the mechanisms underlying the reported NADPH/NADH gradient in the developing BP and whether this gradient was truly affected by the experimental manipulations remain unclear.

Reviewer #1 (Recommendations for the authors):

This manuscript investigated the distribution of several metabolic indicators during basilar papilla (BP) development, which included the use of FLIM and live imaging of mitochondria. The authors showed a decreasing NADPH:NADH ratio from the proximal to basal region of the BP and provided evidence this differential ratio underlies the tonotopic organization of the organ. When glycolysis or more specifically the pentose phosphate pathway was blocked in vitro (presumably resulting in the reduction of NADPH production and NADPH:NADH ratio), hair cells in the proximal region of the BP adapted distal hair cell morphologies. Bmp7 expression, which is normally low, and Chordin expression, which is normally high in the proximal region of the developing papilla were concomitantly decreased when glycolysis was inhibited with 2-DOG treatments. Since metabolism could affect many cellular pathways, it is important to discern the specificity of blocking glycolysis in affecting tonotopy. First, is NADPH:NADH ratio indeed disrupted with 2-DOG treatments? Was FLIM conducted on specimens treated with 2-DOG? Could NADPH:NADH ratio be raised in the distal region and results in proximal hair cell morphologies? Tonotopy is a graded phenomenon along the cochlear duct but the proximal and apical regions of the BP but not the middle region were primarily compared in the study. The upregulation of Bmp7 and the downregulation of Chordin as a result of 2-DOG treatment did not seem graded but the control samples also did not show a clear gradient of expression of these genes. This could be confounded by the culture conditions as the graded expression pattern of Bmp7 and chordin in the normal BP published previously are clear. Therefore, cellular analyses of the middle region of the BP will help to substantiate the hypothesis.

More specific comments are listed as follows:

1) Lines 109-112. It is a bit puzzling that the proximal-distal gradient of NADPH/NADH ratio (tbound), which was proposed to be important for establishing the tonotopic gradient, was only observed at E6 and E14 but not at E9. What could be the explanation?

2) In Figure 5, blocking glycolysis resulted in hair cells of the proximal region adapting morphology of distal hair cells is an interesting result. However, is the proximal hair cell phenotype indeed caused by the disruption of tonotopy, or the distal hair cell pattern is the default and further differentiation into proximal hair cell morphologies require more glycolysis/NADPH? There seems to be a concomitant increase in hair cell number in the proximal region after 2-DOG treatment, which suggests other cellular events could be occurring and account for the hair cell morphological phenotype.

3) There is a lot of speculation without clear referencing on the role of developmental changes that were not found to show a gradient along the P-D axis e.g. Idh3a expression pattern going from apex to base gradient at E14 but switch to base to apex at E16, and LDHA expression switched between hair cells and supporting cells. It's a bit distracting and confusing to the main story.

4) In Figure 6, blocking the pentose phosphate pathway with 6-AN seems to affect the luminal size of distal HCs as well when compared to controls. In the distal region, there seems to be an increase in the density of the HCs. Thus, results in Figure 8 suppl 1 showing the lack of change in EdU uptake in the presence of 2-DOG is important to address this concern as well as to comments listed in #3. However, the images shown in Figure 8 suppl 1 were not convincing and seem to show there was an increase in proliferation in the proximal region of the treated samples when compared to control. Results in the distal region seemed variable. Providing quantification of these results would help to strengthen the conclusion that proliferation was not involved.

5) The lack of gradient differences in the mouse data, though interesting on its own, does not add to the manuscript.

Clarifications to improve the manuscript are listed as follows:

1. Figure 1E-H, Highlight the region in the low-power image where the inset was taken. Describe in the text the interpretations of the color differences in insets of G and H.

2. The term basilar papilla should be defined in the Introduction rather than just used in the title of Figure 3 and Materials and methods.

3. Line 345-346, cite reference indicating hair cell synapse maturation in BP at E16.

4. Figure 3B and C, it is not clear if the proximal to basal comparison of Aldh1a3 level was reversed between E6.5 and E14, or if the representation of the ratio was different between the two ages. I could have missed it, but I didn't see the description in the text in Figure 3C.

5. The differential expression of Ldha along the BP in Figure 3 was further investigated but not Aldh1a3, which was also differentially expressed at E6.5 and E14.

6. Figure 4E, F, it looks like 2NBDG uptake was variable among cells in both proximal and basal regions with some super bright labels in the proximal region. In any case, Figure 4E stated there was no difference in 2NBDG uptake between proximal and basal regions, but Figure 4F showed a difference at E14. What is the sample size for E14 in Figure 4F and the statistics applied? What is the take-home message here for readers?

7. Lines 391-393, there was no proximal to the basal gradient of NADPH/NADH at E9 as shown in Figure 1. So, it is hard to draw a conclusion about cytosolic versus mitochondria metabolism.

8. Line 540, would blocking PFK increase the PPP pathway?

9. Indicate which region of the cochlear duct was taken for images shown in Figure 9A-D.

10. Discussion (line 768-770). I don't think there is enough existing evidence to state that cell cycle exit in the BP progresses from distal to proximal regions. The data from Corwin's lab (cited reference) suggest a neural to the abneural spread of cell cycle exit, unlike their counterpart in the mouse inner ear.

Reviewer #2 (Recommendations for the authors):

This study investigated whether a proximal-to-distal gradient of glucose metabolism exists in the developing basilar papilla (BP) and whether the metabolic gradient plays a role in instructing the positional identity of auditory hair cells along the tonotopic axis. Several assays were applied to address the presence of metabolic gradient, including NAD(P)H FLIM, re-analysis of RNA-seq and microarray datasets, IDH3A immunofluorescence, and 2-NBDG/TMRM fluorescence. Using FLIM was an excellent choice as this method provides a spatial relationship between the metabolic state and tonotopic features of hair cells along the developing BP. To address the involvement of glucose metabolism in tonotopic patterning, several pharmacological interventions were applied to inhibit specific metabolic pathways, which revealed that cytosolic glycose metabolism, but not mitochondrial metabolism plays a role in tonotopic patterning. Furthermore, the signaling gradients known to direct tonotopic patterning are disrupted by inhibition of glucose metabolism. This study links morphogen gradients with metabolic states along the developing auditory organ, providing novel insights into how tonotopy patterning is regulated. However, several experimental results were inconsistent with each other in supporting the hypothesis that the NADPH/NADH gradient and glycolysis are directly involved in establishing the tonotopy in the BP. Furthermore, the mere measurement of the surface area of hair cells is not a sufficient criterion to determine tonotopy. Other features such as hair bundle height and the number of stereocilia per hair cell should be included.

Specific comments:

1) One of the major limitations of this study is that only one feature, the surface area, was used to investigate the tonotopy. Several tonotopic features of hair cell morphology have been reported, including hair cell surface area (used in this study), hair bundle height, stereocilia number, and hair cell length. It was shown that hair cell surface area is not much different between proximal and distal regions in the early stage, and as BP matures, the tonotopic differences in the surface area become obvious because the surface area in the proximal region is increased to a greater extent than the distal region (Tilney et al., 1986, Dev Biol). Thus, the smaller surface area in the proximal region of BPs treated with 2-DOG, SAM, or 6-AN compared to controls may be due to defective developmental progression induced by abnormal glucose metabolism. Therefore, tonotopic changes in hair morphology should be confirmed by analyzing other tonotopic features.

2) The author argued for an expression gradient of NADPH/NADH regulators along the tonotopic axis based on previously generated RNA-seq and microarray datasets (Nat Comm, 2014). However, while LDHA mRNA levels are ~4-fold higher in the distal region than in the proximal region in the E6.5 RNA-seq data (Figure 3B), LDHA fluorescent intensity does not differ significantly between the proximal and distal regions at E7 (Figure 4 Supplement 2B). Therefore, it is essential to validate the expression gradients of selected genes from RNA-seq and microarray datasets using more reliable quantitative methods such as qRT-PCR or Western blot.

3) Based on the 2-NBDG and TMRM fluorescence results (Figure 4), the authors concluded that the gradient of NADPH/NADH originates specifically from tonotopic differences in cytosolic metabolism (line 392). However, there is a temporal gap between the two gradients. Tonotopic differences in NADPH/NADH are already significant at E6 (Figure 1), whereas the tonotopic difference in cytosolic glucose uptake is not significant until E14 (Figure 4). This temporal discrepancy makes it difficult to conclude that NADPH/NADH gradient (E6) originates from the glucose uptake gradient (E14). Similarly, it is difficult to link a causal relationship of tonotopic differences between surface area and glucose uptake because while glucose uptake is not significantly different until E14 (Figure 4), the surface area can be changed by blocking glycolysis at E6-E8 (Figure 5). To argue that metabolic gradients play a role in tonotopic patterning, the tonotopic gradient of NADPH/NADH, its regulators, and glucose uptake should be clearly correlated spatially as well as temporally in the developing BP.

4) Changes in Bmp7 and Chdl1 gradients by 2-DOG treatment (Figure 8) provide a molecular mechanism by which glucose metabolism regulates the tonotopic patterning of BP. However, 2-DOG has been shown to influence several cellular processes other than glycolysis, such as angiogenesis and autophagy. Therefore, it is crucial to confirm that the changes in Bmp7 and Chdl1 gradients by 2-DOG are due to inhibition of glucose metabolism but not due to other cellular effects. This will be confirmed if SAM and 6-AN, but not YZ9 and UK5099, which induced surface area changes similar to 2-DOG, also cause similar changes in Bmp7 and Chdl1 gradients.

1) Specify the model system in the abstract since birds and mammals appear to have different mechanisms for tonotopic patterning.

2) Figure 1I., τbound values were consistently higher in the proximal than distal regions from E6 to E16, although statistical analysis indicates significant differences at E6 and E14 but not at E9 and E16. Non-significance at E9 and E16 may be due to the small number of n, which made statistical power not strong enough to draw statistical significance. To give the readers a better idea of how variable τbound values are during development, it is recommended that the graphs on the left are presented as a boxplot with individual data points. Also, provide error bars and statistical analysis for the graphs on the right.

3) Figure 2C and D. Similar to the point above, it is recommended to present the graphs as a box plot with individual data points. Does "n = 10 biological replicates" mean 10 different BP samples? If so, indicate how many hair cells and supporting cells per BP sample were used to calculate τbound values?

4) Line 323. There are several instances where the authors refer to "other systems". It will be better to specify the system the authors refer to so that the readers get an idea of how the "other systems" are related to the auditory system. e.g., "studies in other systems" in lines 458, 505, and 685.

5) Figure 8. Why are the surface areas of control BPs so different between B (~40 μm2 in proximal) and C (~20 μm2 in proximal)?

6) Figure 8 Supplement 1. It is difficult to conclude based on the images whether 2-DOG treatment affects proliferation in the developing BP. Counting EdU+ cells will help to conclude whether or not cell proliferation is changed by inhibiting glucose metabolism.

7) Figure 9. The results from the mouse organ of Corti are interesting but do not appear to be relevant to the rest of the manuscript, since differences in τbound values were compared only between HCs and SCs, but not along the tonotopic axis. It is recommended to either present the τbound values between the proximal and distal regions of the organ of Corti or save them for further study.

8) Line 853. While the Shh gradient is present along the auditory organ in both birds and mammals, the Bmp7 gradient is present only in birds but not in mammals (Son et al., 2014, PNAS).

9) It is clear that inhibition of glucose metabolism by 2-DOG, SAM, and 6-AN induces the proximal BPs to become more like distal BPs, such as decreased surface area and increased Bmp7 expression in the proximal region. Is it possible to conduct the opposite experiment, i.e., activation of glucose metabolism, which may cause distal BP to become like proximal BP?

Reviewer #3 (Recommendations for the authors):

In this manuscript, Blacker et al. examine cellular metabolism primarily in the chick auditory organ, the basilar papilla (BP). They use FLIM to visualize NAD(P)H in live tissue to study changes in glucose metabolism within the auditory epithelium over the course of development. They show that there are differences in metabolism between HCs and SCs. They also show that there are spatial differences in metabolism along the length of the BP at specific developmental time points. The authors bolster evidence for this metabolic gradient using existing RNAseq datasets that show differential expression of metabolic genes along the proximal-to-distal axis of the BP. They then use pharmacology to manipulate glucose metabolism (glycolysis, pentose phosphate pathway, and the TCA cycle). The authors show that altering glucose metabolism early in glycolysis or at the level of the pentose phosphate pathway results in a loss of tonotopy in the BP. In contrast, inhibiting glucose metabolism at the level of the TCA cycle impaired HC differentiation. Further, they show that inhibiting glycolysis alters morphogen gradients that are required for the emergence of tonotopy within the developing BP. Overall this work suggests that specific metabolic pathways may drive developmental changes in morphogen gradients and – perhaps – contribute to HC differentiation. This is an exciting advance for the hearing field, especially with respect to potential regenerative therapies.

Strengths of the paper:

1. The use of FLIM to study NAD(P)H fluorescence in living HCs and SCs across the auditory epithelium during the course of development.

2. The powerful use of pharmacology to probe the role of glucose metabolism in tonotopy in long-term cultures of BP explants.

3. Using multiple approaches for the study: live NAD(P)H imaging, RNAseq data, pharmacology, vital dyes, and immunohistochemistry.

4. Linking distinct metabolic pathways to developmental events is a novel and exciting area of research.

Weaknesses of paper:

1. The majority of the manuscript uses t-tests for comparisons. In many cases, a 2-way ANOVA is needed for comparisons. It is not clear that all the findings will hold up to more rigorous statistical analysis.

2. Substantial work is needed to enhance the clarity of the manuscript. There is a limited introduction to the different facets of glucose metabolism highlighted in this paper. Metabolism is a complex topic that a broad audience will not be intimately familiar with. In addition, the paper is rich with data. The authors discuss differences in metabolism along 3 variables (space, time, and cell type). Occasionally it is unclear what type of difference is being discussed and the trend of the difference.

3. Manipulations inhibiting glycolysis and oxidative phosphorylation over extended periods of time could have many secondary effects on development related to a lack of energy supply.

Overall, this is a very interesting line of research. Currently, the manuscript needs revision prior to publication.

1. Statistics. The majority of the manuscript uses t-tests, which are not appropriate when making multiple comparisons (ie: proximal vs. distal over multiple stages of development). Currently, it is not clear that all the findings will hold up to more rigorous statistical analysis. For example, Figure 1I, 1J, 2C, 2D, 4D, 4F, 4G, 4I, 4J, all of Figure 4 Supp. 1, Figure 4 Supp. 2B, 5C, Figure 5 Supp. 1C, 6D, Figure 6 Supp. 1B, 7C, 8B, 8C, 9E, and 9F should be analyzed using two-way ANOVA. It is not statistically sound to pick only specific comparisons to make.

Additionally, in some cases, there are stars above bars with no indication of what condition is being compared to (ie. Figure 9E). Figure 3A lacks statistical analysis altogether. Finally, there are several cases where the authors say that there is a "significant" difference but do not show that it is statistically significant (for example, the caption of Figure 4I says there is a significant decrease between E9 and E14, but there is no indication of statistical significance in the figure). Differences that are not statistically significant can be discussed as trends, but should not be called significant. Similarly, differences that are statistically significant should not be discussed as "no difference" (ex. Figure 4E/F caption but Figure 4F shows a statistically significant difference; Figure 4 Supp. 2).

2. Enhance the clarity of the manuscript. The manuscript is quite well written but currently, there is limited text in the introduction outlining the relevant parts of glucose metabolism highlighted in this paper. Or how NADPH and NADH feed into these pathways. In addition, the authors discuss differences in metabolism along 3 variables (space, time, and cell type), and occasionally it is not clear what type of difference is being discussed – what is increased or decreased and where. Schematics like those in 2E are helpful to the reader.

3. Manipulations inhibiting glycolysis and oxidative phosphorylation over extended periods of time could have many secondary effects on development. The authors should lighten claims of causation. At a minimum, there should be a discussion of possible secondary effects and the need for more targeted manipulations to draw a convincing causative link between cellular metabolism and the development of tonotopy within the BP.

4. The authors detect a proximal-to-distal gradient in tbound at E6 that "was still present at E14" and discuss this as a gradient "along the tonotopic axis between E6 and E14". However, their figure does not show a significant difference between the proximal and distal regions at E9. This could be related to a lack of statistical power (the n is lower for E9), but could alternatively represent a real fluctuation in the metabolic state over time. The authors should include a discussion of their E9 data, rather than glossing over it.

5. Although it may be beyond the scope of this paper, it would be nice to come full circle back to NAD(P)H imaging. Do the alteration to the PPP pathway lead to a loss of the NADPH/NADH gradient along the proximal-distal axis in the developing BP?

6. The authors make several claims regarding differences in TMRM. They should also consider that this cationic vital dye may enter HCs via mechanotransduction channels. This form of entry may explain why labeling is stronger in older HCs compared to SCs or younger HCs.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Crosstalk between glucose metabolism and Bmp7-Chordin-like 1 signalling specifies tonotopic identity in developing hair cells." for further consideration by eLife. Your revised article has been evaluated by Kathryn Cheah (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

Essential revisions:

While all three reviewers felt that the revised manuscript is a substantial improvement there are still some outstanding issues:

1) The Chordin rescue experiment requires additional controls and clarification.

2) Additional inconsistencies and quantification were requested by all three reviewers.

3) Comparing treatments of SAM or 6-AN to 2-DOG in the ability to affect Bmp7 signal, for example.

Reviewer #1 (Recommendations for the authors):

This revised manuscript is much improved and streamlined. The authors provided additional results showing that Chordin like-1, which is normally expressed in the proximal region of the BP could rescue the distalizing effect of 2-DOG in the proximal region, thus strengthening the link between metabolism and tonotopy. However, in this type of rescue experiment, 2-DOG and Chordin like-1 (Chdl-1) treatments should be included as controls within the same experiment, rather than using controls with no treatment alone. While readers could extrapolate 2-DOG effects from other figures, there was no description of Chdl-1 in dosage, condition, or results in the revised manuscript. Furthermore, the statement in lines 800-802 and results in Supplementary Table 1 seem to suggest that Chdl-1 treatment alone does not affect the tonotopic gradient by itself. Is that true? If so, this result is not consistent with previous studies.

Editorial suggestions to help improve the manuscript:

1) Line 28, what does flattening of hair cell morphology mean? Do you mean shorter hair bundles in the proximal versus the distal BP? If so, the images in this study are not clear enough for naïve readers to draw that conclusion.

2) Line 54, what does inner sense organs means? Are you referring to the inner ear or tissues deep in the body?

3) Figure 1 is very helpful but could use some revision: a) consider putting labels for the compartments at the top of the figure with a larger font. It may help readers to see the longitudinal lines represent mitochondrial membrane, which is not apparent, b) increase the font size of the pathways, c) no explanation for labels 1, 2, and 3 in the legend and I am not sure if they meant to illustrate something more than the color scheme, and d) add PKM2 to the metabolic flowchart.

4) Lines 293 and 295, a bit odd to split PKM2 results into two separate paragraphs.

5) Figure 2 a) Consider combining panels D and K, and expand B and C. Hard to read the label in C, b) I am still not clear why αbound level (dotted purple line) is in a steady state based on the purple and green curve and there is no explanation of αbound in the text, c) Panels E-H could be slightly bigger, d) move Scale bars=50um to the end of description for (E-H).

6) Line 205, probably better just to state τbound gradient not related to glucose uptake?

7) Line 236, the sentence is a bit awkward.

8) Line 292, define PKM2 when it first appeared in the text.

9) Line 339, Figure 3 or Figure 5?

10) Line 342, Figure without a number.

11) Lines 340-343, State in the text lower pHrodo Red signal meant higher pH.

12) No quantification of Edu-labeled cells in Figure 6 suppl 2.

13) Line 484, does SAM inhibit both glycolysis and PPP pathways similar to 2-DOG?

14) Typo in the last sentence of the legend for Figure 7 supplement 1.

15) Incomplete last sentence found in the legend for Figure 8.

16) Figure 9. The images in E are better in this revision. However, there is some inconsistency between the text and the figure legend. In the text, explants were established at E6.5 and incubated for 72 hours. In the legend, it was stated the in vitro incubation was for 7 days.

17) Figure 9. It may be better to put the number and statistics in the panels that they belong rather than at the end of the figure. Sample numbers need to be provided for panel E as well.

18) Supplemental Table 1. Are all p values in the left-hand columns controls? Confused by the statement in line 799, "control or inhibitor treatment groups". Or it should be controlled for inhibitor treatment groups? If so, the p values in the right-hand column are results from individual treatments. Then, Chdl-1 treatment did not affect the tonotopic differences observed in controls?

Reviewer #2 (Recommendations for the authors):

Figure 7 supplementary 2.

Based on the statistical analysis, the authors concluded that YZ9 treatment had no effect on HC morphology along the tonotopic axis. However, the graph in (B) shows that the luminal area of hair cells was dramatically reduced in the proximal but not in the distal region by YZ9 treatment, resulting in no significant difference between the proximal and distal regions. This suggests a loss of tonotopic identity of HC morphology by YZ9 treatment, similar to 2-DOG, SAM, or 6-AN. To support the authors' argument, it is important to confirm that YZ9 treatment did not induce tonotopic changes in HC morphology. Double-checking with other tonotopic parameters, such as HC nuclear area and cell number, will help to clarify this issue.

Figure 9.

(E) The Bmp7 expression domain in the control explant is peculiar in that it has a lining around the distal region, unlike the control image in Figure 9 supplement 1. Figure 3 supplement 4 shows that Bmp7 is normally expressed in the sensory domain.

Bmp7 in situ signals are generally stronger in 2-DOG-treated BP than in control, but there still seems to be a proximal-distal gradient with a generally stronger in situ signal. qRT-PCR with each half of the BPs, as done in their previous publication, will clearly show if the gradient is disrupted.

Figure 9 supplement 1.

The previous review questioned whether the changes in tonotopic identity caused by 2-DOG were due to the inhibition of glucose metabolism rather than other cellular effects. This question was crucial to support the authors' conclusion, as 2-DOG has been shown to affect several cellular processes other than glycolysis. The authors showed that blocking mitochondrial metabolism did not alter the Bmp7 gradient, but this doesn't solve the problem of the specificity of 2-DOG. It is worth investigating whether SAM and 6-AN, which induce luminal area changes similar to 2-DOG, also disrupt the same molecular pathways, such as the Bmp7 and Chld1 gradients.

Reviewer #3 (Recommendations for the authors):

This manuscript is substantially improved compared to the previous version. Thank you for all the hard work you put into responding to reviewer concerns. A revised manuscript would be welcomed by this reviewer.

– The RNAseq and microarray analyses are good for hypothesis generation, but the differential expression the authors detect often does not match the immunohistochemistry and RNA scope data they provide. It is especially confusing to say that there is no difference in LDHA and IDH3A protein expression across the BP, then suggest that differential RNA expression of these genes could drive the metabolic gradient the authors observe. Some of the variability in the data may be related to differences in animal age across experiments, but this variability makes it a bit challenging to interpret the data. In addition, not all the genes listed in Figure 3 Supp. 4A seems to reach statistical significance. The authors should consider framing Figure 3 Supp. 4A as a way to generate hypotheses that they then test with additional experiments, rather than claiming that they identified a whole set of differentially expressed genes when in fact they present contradictory data within the same manuscript. They should also clarify how they decided which metabolic RNAs to examine further.

– Although the authors used BMP7 as a positive control for RNA scope, there does not appear to be a proximal-to-distal difference in BMP7 fluorescence. In addition, the authors do not provide any negative controls to show a lack of fluorescence in the absence of gene-specific RNA scope probes. Without successful controls, the experimental data is hard to interpret. The authors should consider providing better controls.

– Line 297-298: "By controlling a metabolic feedback loop that regulates the switch between glycolysis and OXPHOS…" The authors previously showed data suggesting that differences in the NAD(P)H/NADH ratio are not due to changes in mitochondrial OXPHOS. Here, though, they suggest that PKM2 controls the amount of OXPHOS and is differentially expressed across the BP. How can we reconcile these two points?

– Figure 4, 5: How was the quantification in 4B-D and 5B performed? Are fluorescence values provided per HC/per SC? As HC size varies across the proximal-distal axis of the BP, it is not fair to quantify raw total fluorescence. The fluorescence should be normalized by the area occupied by each cell type.

– Figure 5: Could proximal-to-distal differences in pHrodo Red fluorescence result from the gradient of HC differentiation (i.e. more mature distal HCs, less mature proximal HCs) across the tonotopic axis at E9? In other words, do young HCs have a higher pHi that decreases as hair cells mature? This could explain the proximal-to-distal differences shown in Figure 5.

– Figure 6 Supp. 2 – To claim that there is no increased proliferation in response to 2-DOG, EdU+ cells should be quantified in control vs. 2-DOG cultures. It seems that the authors have only provided general cell counts (in C) and not EdU+ cell counts.

– Do the authors have any evidence that the various drug treatments are not simply blocking cell growth and differentiation? As mentioned by another reviewer, the distal-like phenotype could be a "default"; perhaps proximal HCs simply aren't maturing. It would be nice to show examples of cultures at 1DIV vs. 7 DIV to demonstrate how much cells grow during this time and how treated cultures at 7 DIV compare to immature 1 DIV cultures. (SAM treatment especially looks like it might impact hair bundle morphology in the proximal BP.)

– Figures 7 Supp. 2 and Figure 8 continue to rely on a single morphological measure of HC morphology as a readout of tonotopy. The authors should add additional quantification (cell density, nuclear area…), as has been provided for other manipulations.

– Figure 9 and Figure 9 Supp. 1: In the absence of quantification of the in situ hybridization data, please state how many examples show similar Bmp7 and Chdl1 gradients.

– To bolster the claim that 2-DOG treatment is specifically disrupting tonotopy and not just the differentiation of proximal HCs (as pointed out by Reviewer 1), the authors should quantify HC density in Figure 10. If HC density is also restored by providing Chdl-1 in the presence of 2-DOG, it will bolster the hypothesis that altered glucose metabolism has a specific effect on HC positional identity.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Crosstalk between glucose metabolism and morphogen signalling specifies tonotopic identity in developing hair cells" for further consideration by eLife. Your revised article has been evaluated by Kathryn Cheah (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there are some remaining editorial issues that need to be addressed, as outlined below:Reviewer #1:

The results of the revised manuscript are coming together nicely. To appropriately demonstrate the rescue effect of Chd1 on 2DOG requires the control, 2-DOG, and 2-DOG+Chd1 be all conducted within the same experiment. Although such an experiment was not provided, the authors have made a good faith effort in demonstrating a dose-response effect of chordin in reducing the differential differences between proximal and distal BP, and the revision is acceptable. Additionally, I have some editorial suggestions. These are merely suggestions, and the acceptance of the manuscript is not dependent on making these changes.

1) The experimental evidence suggests that the tonotopic morphogens are regulated by glucose metabolism. I don't see evidence showing that these morphogens regulate metabolism in return. Therefore, I question the choice of using the phrase "Cross talk" in the title.

2) The abstract, though clear, could be more concise and informative. I feel like it is important to summarize this 32-figure manuscript well!

3) Include PKm2 in the schematic diagram of Figure 1.

4) The first part of the Discussion reads more like an introduction.

Reviewer #2:

The revised manuscript is much improved with additional experimental data that strongly support the authors' conclusions. I appreciate the authors' efforts to make the manuscript in much better shape.

I have just one question. From the diagrams in Figure 7-supplement 2B and Figure 8B, it appears that PFK and PKM2 act in the same glycolysis pathway. However, inhibition of either resulted in different phenotypes. While inhibition of PFK by YZ9 did not affect tonotopic patterning, inhibition of PKM2 by shikonin disrupted tonotopic patterning, albeit at the epithelial level. Why did two inhibitors of the same pathway have different effects on tonotopic patterning?Reviewer #3:

The manuscript is greatly improved and nearly ready for publication.

There are a few typos and points to clarify that I would recommend to the authors. These recommendations should not require additional experiments.

List of recommendations for clarity:

Figure 2 Supplement 1 – this Figure is really helpful and we would recommend making it part of a main figure (perhaps you can condense Figure 2G, as these two plots show the same data, in favor of this helpful schematic). The clarity of the FLIM section is massively improved since the first version of the paper. The one link that is still hard to make is how the values of Tbound inform flux through the PPP vs. glycolysis. This figure beautifully illustrates exactly that. Also, consider adding PPP vs. glycolysis to the gradient in the Figure 2H schematic, and/or moving the sentence in line 109 to line 101 in the main text. Although all of this knowledge is obvious to those who think about metabolism all the time – these concepts can be confusing for someone not in the field.

We would recommend adding Pkm2 to Figure 1, as there is an entire figure focused on Pkm2.

Figure 3 Supp. 3E. Are you detecting statistically significant proximal-to-distal changes in Bmp7 here? It is not indicated in the quantification (no stars or "ns"). If the positive control is not showing a statistically significant change, it is hard to interpret the other data.

https://doi.org/10.7554/eLife.86233.sa1

Author response

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

Comments to the Authors:

We are sorry to say that, after consultation with the reviewers, we have decided that this work will not be considered further for publication by eLife.

Specifically, the causal relationship between the NADPH/NAD gradient and tonotopy has many outstanding issues that require further clarification. Strategies employed to inhibit cellular metabolism in this study could have many secondary effects that may not be related to the establishment of tonotopy. Furthermore, the mechanisms underlying the reported NADPH/NADH gradient in the developing BP and whether this gradient was truly affected by the experimental manipulations remain unclear.

In attempt to further characterise a role for metabolism in tonotopic patterning, we have included additional experiments showing a rescue of tonotopic patterning when explants are treated with medium containing both 2-DOG and Chordin like-1. We believe the observed effects when treating with 2-DOG and Chordin like-1 together (new data Figure 10) support a role for metabolism specifying tonotopy in the developing BP.

Reviewer #1 (Recommendations for the authors):

This manuscript investigated the distribution of several metabolic indicators during basilar papilla (BP) development, which included the use of FLIM and live imaging of mitochondria. The authors showed a decreasing NADPH:NADH ratio from the proximal to basal region of the BP and provided evidence this differential ratio underlies the tonotopic organization of the organ. When glycolysis or more specifically the pentose phosphate pathway was blocked in vitro (presumably resulting in the reduction of NADPH production and NADPH:NADH ratio), hair cells in the proximal region of the BP adapted distal hair cell morphologies. Bmp7 expression, which is normally low, and Chordin expression, which is normally high in the proximal region of the developing papilla were concomitantly decreased when glycolysis was inhibited with 2-DOG treatments. Since metabolism could affect many cellular pathways, it is important to discern the specificity of blocking glycolysis in affecting tonotopy.

To address this issue, we have used multiple inhibitors that target different branches of glucose flux. The same phenotype was produced in response to all inhibitors except YZ9. From a comparison of the statistical results between all inhibitor experiments (Supplementary table 1), we believe that the observed changes in cell morphology are a result of dysregulated metabolism rather than off-target effects.

First, is NADPH:NADH ratio indeed disrupted with 2-DOG treatments? Was FLIM conducted on specimens treated with 2-DOG?

For FLIM measurements, treated explants need to be transferred from culture dishes to FLIM imaging dishes. This causes additional mechanical and osmotic stress, which may impact metabolism and introduce variation to NAD(P)H FLIM measurements. In addition, there are gross morphological changes during long-term culture, and given that fluorescent markers cannot be used in combination with FLIM it is not feasible to identify proximal and distal ends.

Could NADPH:NADH ratio be raised in the distal region and results in proximal hair cell morphologies?

The reviewers raise an excellent point, and whilst this would be a valuable experiment, we believe it is beyond the scope of the current study. There are no pharmacological agents available to specifically target the NADPH/NADH ratio in vitro. This could potentially be done using an in ovo electroporation approach to knock out NAD kinase in different regions of the BP. However, as it is not possible to control the region-specific uptake of constructs during in ovo electroporation, it would not be possible to target NADPH/NADH pools in proximal and distal BP regions specifically.

Tonotopy is a graded phenomenon along the cochlear duct but the proximal and apical regions of the BP but not the middle region were primarily compared in the study. The upregulation of Bmp7 and the downregulation of Chordin as a result of 2-DOG treatment did not seem graded but the control samples also did not show a clear gradient of expression of these genes. This could be confounded by the culture conditions as the graded expression pattern of Bmp7 and chordin in the normal BP published previously are clear. Therefore, cellular analyses of the middle region of the BP will help to substantiate the hypothesis.

New images from repeated in situs have been added for clarity in Figure 9 were re-ran to more clearly show the change in gradient following treatment. See Figure 9 panel E. In addition, we have included in situ data for Bmp7 expression following treatment with the OXPHOS inhibitor UK5099 (Figure 9 Supplement 1).

More specific comments are listed as follows:

1) Lines 109-112. It is a bit puzzling that the proximal-distal gradient of NADPH/NADH ratio (tbound), which was proposed to be important for establishing the tonotopic gradient, was only observed at E6 and E14 but not at E9. What could be the explanation?

Additional n numbers have been added to this data set, and all data have been re-analysed for statistical significance using 2-way ANOVA testing. Following new analysis, the gradient in tbound (NADPH/NADH) is significant at E6, E9 and E14 (see revised Figure 2).

2) In Figure 5, blocking glycolysis resulted in hair cells of the proximal region adapting morphology of distal hair cells is an interesting result. However, is the proximal hair cell phenotype indeed caused by the disruption of tonotopy, or the distal hair cell pattern is the default and further differentiation into proximal hair cell morphologies require more glycolysis/NADPH?

This is interesting point! The distal pattern may indeed be the default. It may also be the altered PPP metabolism interferes with normal cell cycle progression and differentiation. Cell cycle exit and cell size, growth are tightly coupled with metabolism, in particular PPP metabolism and pH (see new data added in Figure 4). Altering these gradients along the BP during development with 2-DOG, 6AN or SAM may therefore interfere with the normal patterning in cell size and number. We have included this argument as a key point in the discussion (Line 813).

There seems to be a concomitant increase in hair cell number in the proximal region after 2-DOG treatment, which suggests other cellular events could be occurring and account for the hair cell morphological phenotype.

We have re-done the EdU experiments and quantified cell density in the proximal and distal regions in control and treated explants. Overall blocking glycolysis consistently reduces EdU staining. Cultures were set up at E8 and maintained in culture for 48h covering the active phase of mitotic cell division. The increase in cell density in the proximal but not distal region is consistent to what we saw with Bmp7 and based on new data (Figure 6 Supplement 2), does not seem to be driven by an increase in proliferation.

3) There is a lot of speculation without clear referencing on the role of developmental changes that were not found to show a gradient along the P-D axis e.g. Idh3a expression pattern going from apex to base gradient at E14 but switch to base to apex at E16, and LDHA expression switched between hair cells and supporting cells. It's a bit distracting and confusing to the main story.

We thank the reviewer for this comment and fully agree that the previous way in which the data were presented was unclear. This section of the manuscript has now been significantly re-written for clarity and new results have been included to support our overarching hypothesis (Figures 4 and 5).

We refer the reviewer to Line 206 section beginning “Expression of Lactate and isocitrate dehydrogenases in the developing BP…”.

4) In Figure 6, blocking the pentose phosphate pathway with 6-AN seems to affect the luminal size of distal HCs as well when compared to controls. In the distal region, there seems to be an increase in the density of the HCs. Thus, results in Figure 8 suppl 1 showing the lack of change in EdU uptake in the presence of 2-DOG is important to address this concern as well as to comments listed in #3. However, the images shown in Figure 8 suppl 1 were not convincing and seem to show there was an increase in proliferation in the proximal region of the treated samples when compared to control. Results in the distal region seemed variable. Providing quantification of these results would help to strengthen the conclusion that proliferation was not involved.

We refer the reviewer to new data included in Figure 6 Supplement 2 and quantification for changes in proliferation and cell density.

We have also included further analysis for HC nuclear area in addition to luminal surface area measurements. Higher magnification images have also been included in the Supplementary data to highlight changes in HC luminal surface areas (Figure 6 Supplement 1 and Figure 7 Supplement 1).

5) The lack of gradient differences in the mouse data, though interesting on its own, does not add to the manuscript.

We agree that these data distract from the main focus of the study and have therefore removed it from the manuscript.

Clarifications to improve the manuscript are listed as follows:

1. Figure 1E-H, Highlight the region in the low-power image where the inset was taken. Describe in the text the interpretations of the color differences in insets of G and H.

This Figure has now been significantly revised (see new Figure 2).

2. The term basilar papilla should be defined in the Introduction rather than just used in the title of Figure 3 and Materials and methods.

This definition has been included in both the abstract and introduction.

3. Line 345-346, cite reference indicating hair cell synapse maturation in BP at E16.

This discussion point is no longer included in the manuscript.

4. Figure 3B and C, it is not clear if the proximal to basal comparison of Aldh1a3 level was reversed between E6.5 and E14, or if the representation of the ratio was different between the two ages. I could have missed it, but I didn't see the description in the text in Figure 3C.

Aldh3 expression data has been removed. See updated Figure 3 Supplement 4.

5. The differential expression of Ldha along the BP in Figure 3 was further investigated but not Aldh1a3, which was also differentially expressed at E6.5 and E14.

This section of the paper has now been significantly revised and Aldh3 expression data are no longer included.

6. Figure 4E, F, it looks like 2NBDG uptake was variable among cells in both proximal and basal regions with some super bright labels in the proximal region. In any case, Figure 4E stated there was no difference in 2NBDG uptake between proximal and basal regions, but Figure 4F showed a difference at E14. What is the sample size for E14 in Figure 4F and the statistics applied? What is the take-home message here for readers?

This is now addressed in the Results section (line 200 and Figure 3 Supplement 2). Sample size has been added to the legend for Figure 3 Supplement 2, n = 5.

To provide further clarity and hopefully address the reviewer’s concern, 2-NBDG provides an indirect measure of intracellular glucose metabolism. The dye provides a measure of glucose uptake across the plasma membrane, but subsequent to its phosphorylation by hexokinase (HK) 2-NBDG does not provide a measure of whether glucose is catabolised in glycolysis, or via the PPP. It is therefore not clear how the 2-NBDG signal should be interpreted relative to the FLIM gradient in bound. Further studies are needed to untangle the how the two assays and their combined signals are best interpreted as read-outs of metabolic phenotype.

7. Lines 391-393, there was no proximal to the basal gradient of NADPH/NADH at E9 as shown in Figure 1. So, it is hard to draw a conclusion about cytosolic versus mitochondria metabolism.

Upon increasing the sample size (n = 4) the gradient in bound is now significant at both E6 and E9. See revised Figure 2.

8. Line 540, would blocking PFK increase the PPP pathway?

Blocking PFK should not affect flux into the PPP, as this step is glucose catabolism is regulated by activity of G6PDH. The PPP can however feed metabolites back into the main branch of glycolysis from the non-oxidative branch, so may contribute to activity in the lower branch of glycolysis.

9. Indicate which region of the cochlear duct was taken for images shown in Figure 9A-D.

Added to Figure 9.

10. Discussion (line 768-770). I don't think there is enough existing evidence to state that cell cycle exit in the BP progresses from distal to proximal regions. The data from Corwin's lab (cited reference) suggest a neural to the abneural spread of cell cycle exit, unlike their counterpart in the mouse inner ear.

This section has now been re-written: See paragraph beginning line 814.

Reviewer #2 (Recommendations for the authors):

This study investigated whether a proximal-to-distal gradient of glucose metabolism exists in the developing basilar papilla (BP) and whether the metabolic gradient plays a role in instructing the positional identity of auditory hair cells along the tonotopic axis. Several assays were applied to address the presence of metabolic gradient, including NAD(P)H FLIM, re-analysis of RNA-seq and microarray datasets, IDH3A immunofluorescence, and 2-NBDG/TMRM fluorescence. Using FLIM was an excellent choice as this method provides a spatial relationship between the metabolic state and tonotopic features of hair cells along the developing BP. To address the involvement of glucose metabolism in tonotopic patterning, several pharmacological interventions were applied to inhibit specific metabolic pathways, which revealed that cytosolic glycose metabolism, but not mitochondrial metabolism plays a role in tonotopic patterning. Furthermore, the signaling gradients known to direct tonotopic patterning are disrupted by inhibition of glucose metabolism. This study links morphogen gradients with metabolic states along the developing auditory organ, providing novel insights into how tonotopy patterning is regulated. However, several experimental results were inconsistent with each other in supporting the hypothesis that the NADPH/NADH gradient and glycolysis are directly involved in establishing the tonotopy in the BP. Furthermore, the mere measurement of the surface area of hair cells is not a sufficient criterion to determine tonotopy. Other features such as hair bundle height and the number of stereocilia per hair cell should be included.

It is very challenging to quantify bundle morphology accurately in long-term explants (cultured for more than 7 days). We believe accurate bundle measurements would need to be done using an in ovo electroporation model.

Specific comments:

1) One of the major limitations of this study is that only one feature, the surface area, was used to investigate the tonotopy. Several tonotopic features of hair cell morphology have been reported, including hair cell surface area (used in this study), hair bundle height, stereocilia number, and hair cell length. It was shown that hair cell surface area is not much different between proximal and distal regions in the early stage, and as BP matures, the tonotopic differences in the surface area become obvious because the surface area in the proximal region is increased to a greater extent than the distal region (Tilney et al., 1986, Dev Biol). Thus, the smaller surface area in the proximal region of BPs treated with 2-DOG, SAM, or 6-AN compared to controls may be due to defective developmental progression induced by abnormal glucose metabolism. Therefore, tonotopic changes in hair morphology should be confirmed by analyzing other tonotopic features.

Additional analysis of HC nuclear area and cell density in proximal and distal regions have been included.

2) The author argued for an expression gradient of NADPH/NADH regulators along the tonotopic axis based on previously generated RNA-seq and microarray datasets (Nat Comm, 2014). However, while LDHA mRNA levels are ~4-fold higher in the distal region than in the proximal region in the E6.5 RNA-seq data (Figure 3B), LDHA fluorescent intensity does not differ significantly between the proximal and distal regions at E7 (Figure 4 Supplement 2B). Therefore, it is essential to validate the expression gradients of selected genes from RNA-seq and microarray datasets using more reliable quantitative methods such as qRT-PCR or Western blot.

We thank the reviewer for raising this important point. For this specific study, we chose to use RNA-scope or immunohistochemistry to look at expression of metabolic enzymes along the BP as these techniques preserve the spatial resolution needed to look at metabolism on a cell-by-cell basis. Using western blot and qPCR analysis, whilst informative, would not maintain the spatial and single cell resolution within the tissue. A further caveat is that metabolic reprograming is challenging to detect at the transcriptional level, as it is regulated at the protein level.

We have included additional analysis for GOT2, LDHB and PKM2 (see Figure 3 Supplement 4 and Figure 4).

3) Based on the 2-NBDG and TMRM fluorescence results (Figure 4), the authors concluded that the gradient of NADPH/NADH originates specifically from tonotopic differences in cytosolic metabolism (line 392). However, there is a temporal gap between the two gradients. Tonotopic differences in NADPH/NADH are already significant at E6 (Figure 1), whereas the tonotopic difference in cytosolic glucose uptake is not significant until E14 (Figure 4). This temporal discrepancy makes it difficult to conclude that NADPH/NADH gradient (E6) originates from the glucose uptake gradient (E14). Similarly, it is difficult to link a causal relationship of tonotopic differences between surface area and glucose uptake because while glucose uptake is not significantly different until E14 (Figure 4), the surface area can be changed by blocking glycolysis at E6-E8 (Figure 5). To argue that metabolic gradients play a role in tonotopic patterning, the tonotopic gradient of NADPH/NADH, its regulators, and glucose uptake should be clearly correlated spatially as well as temporally in the developing BP.

The temporal gradients will not necessarily match as they report different steps of the glucose catabolism pathway. 2NBDG reports uptake of glucose into the cell via GLUT transporters. NADPH/NADH reports differences in the path of glucose catabolism once it has been phosphorylated by HK. Because 2-NBDG cannot by phosphorylated or metabolised by HK, it does not reflect the different paths by which glucose is metabolised. We hope this clarifies the reviewer’s concern and have re-written this in the Results section: see section beginning Line193 – “Live imaging of mitochondrial metabolism and glucose uptake along the tonotopic axis of the developing BP…”.

…it difficult to conclude that NADPH/NADH gradient (E6) originates from the glucose uptake gradient…

We thank the reviewer for raising this question and have made attempts to clarify this in the text. The gradient in NADPH reported by tbound is not a reflection of glucose uptake but of glucose flux into the PPP or NADPH production in mitochondria. We also acknowledge that tbound can of course report a mixture of these two metabolic processes. We have now included a schematic (see Figure 1) illustrating the different metabolic branches and pathways of glucose catabolism.

4) Changes in Bmp7 and Chdl1 gradients by 2-DOG treatment (Figure 8) provide a molecular mechanism by which glucose metabolism regulates the tonotopic patterning of BP. However, 2-DOG has been shown to influence several cellular processes other than glycolysis, such as angiogenesis and autophagy. Therefore, it is crucial to confirm that the changes in Bmp7 and Chdl1 gradients by 2-DOG are due to inhibition of glucose metabolism but not due to other cellular effects.

In attempt to address this issue we chose to use a combination of different inhibitors. As inhibitors of glycolysis (2-DOG, SAM) and of the PPP (6-AN) elicited the same reduction in luminal SA and nuclear area, we conclude that it unlikely for all inhibitors to elicit the same off target effects on cell morphology.

This will be confirmed if SAM and 6-AN, but not YZ9 and UK5099, which induced surface area changes similar to 2-DOG, also cause similar changes in Bmp7 and Chdl1 gradients.

In situ for Bmp7 following UK5099 treatment has been included. We did not see any notable change in Bmp7 in whole mount in situ after treatment with UK5099 for 7 days in vitro (see Figure 9 Supplement 1). We have also included new data showing the effects of 2-DOG treatment in the presence of Chordin-like 1 (Figure 10).

1) Specify the model system in the abstract since birds and mammals appear to have different mechanisms for tonotopic patterning.

This has been specified and the definition/abbreviation added to abstract.

2) Figure 1I., τbound values were consistently higher in the proximal than distal regions from E6 to E16, although statistical analysis indicates significant differences at E6 and E14 but not at E9 and E16. Non-significance at E9 and E16 may be due to the small number of n, which made statistical power not strong enough to draw statistical significance. To give the readers a better idea of how variable τbound values are during development, it is recommended that the graphs on the left are presented as a boxplot with individual data points. Also, provide error bars and statistical analysis for the graphs on the right.

Additional replicates have been added and all data are now significant at E6, E9 and E14 but not E16. However, by E16 HCs are fully differentiated so it is assumed that tonotopic patterning would be established at this stage.

3) Figure 2C and D. Similar to the point above, it is recommended to present the graphs as a box plot with individual data points. Does "n = 10 biological replicates" mean 10 different BP samples? If so, indicate how many hair cells and supporting cells per BP sample were used to calculate τbound values?

Cell numbers are now indicated, and box plots have been used throughout to show individual data points.

4) Line 323. There are several instances where the authors refer to "other systems". It will be better to specify the system the authors refer to so that the readers get an idea of how the "other systems" are related to the auditory system. e.g., "studies in other systems" in lines 458, 505, and 685.

The text has been significantly re-written. This statement is no longer included in the article.

5) Figure 8. Why are the surface areas of control BPs so different between B (~40 μm2 in proximal) and C (~20 μm2 in proximal)?

In Figure 9B measured HC liminal SAs in 2500 µm2 areas. Figure 9C measure HC luminal SAs in 10000 µm2 SAs.

6) Figure 8 Supplement 1. It is difficult to conclude based on the images whether 2-DOG treatment affects proliferation in the developing BP. Counting EdU+ cells will help to conclude whether or not cell proliferation is changed by inhibiting glucose metabolism.

Additional EdU experiments have been added and cell numbers have been quantified in proximal and distal regions. See Figure 6 Supplement 2.

7) Figure 9. The results from the mouse organ of Corti are interesting but do not appear to be relevant to the rest of the manuscript, since differences in τbound values were compared only between HCs and SCs, but not along the tonotopic axis. It is recommended to either present the τbound values between the proximal and distal regions of the organ of Corti or save them for further study.

This figure has now been removed from the paper.

8) Line 853. While the Shh gradient is present along the auditory organ in both birds and mammals, the Bmp7 gradient is present only in birds but not in mammals (Son et al., 2014, PNAS).

We thank the reviewer for drawing attention to this error. Reference of these data in the text has been corrected. See line 844.

9) It is clear that inhibition of glucose metabolism by 2-DOG, SAM, and 6-AN induces the proximal BPs to become more like distal BPs, such as decreased surface area and increased Bmp7 expression in the proximal region. Is it possible to conduct the opposite experiment, i.e., activation of glucose metabolism, which may cause distal BP to become like proximal BP?

There are currently no pharmacological activators of glycolysis of PPP flux. The only way to activate glucose catabolism in different branches would be through the over expression or knock-down of specific rate-limiting enzymes using in ovo electroporation. Whilst important experiments to address, we believe these go beyond the scope of the current study.

Reviewer #3 (Recommendations for the authors):

In this manuscript, Blacker et al. examine cellular metabolism primarily in the chick auditory organ, the basilar papilla (BP). They use FLIM to visualize NAD(P)H in live tissue to study changes in glucose metabolism within the auditory epithelium over the course of development. They show that there are differences in metabolism between HCs and SCs. They also show that there are spatial differences in metabolism along the length of the BP at specific developmental time points. The authors bolster evidence for this metabolic gradient using existing RNAseq datasets that show differential expression of metabolic genes along the proximal-to-distal axis of the BP. They then use pharmacology to manipulate glucose metabolism (glycolysis, pentose phosphate pathway, and the TCA cycle). The authors show that altering glucose metabolism early in glycolysis or at the level of the pentose phosphate pathway results in a loss of tonotopy in the BP. In contrast, inhibiting glucose metabolism at the level of the TCA cycle impaired HC differentiation. Further, they show that inhibiting glycolysis alters morphogen gradients that are required for the emergence of tonotopy within the developing BP. Overall this work suggests that specific metabolic pathways may drive developmental changes in morphogen gradients and – perhaps – contribute to HC differentiation. This is an exciting advance for the hearing field, especially with respect to potential regenerative therapies.

Strengths of the paper:

1. The use of FLIM to study NAD(P)H fluorescence in living HCs and SCs across the auditory epithelium during the course of development.

2. The powerful use of pharmacology to probe the role of glucose metabolism in tonotopy in long-term cultures of BP explants.

3. Using multiple approaches for the study: live NAD(P)H imaging, RNAseq data, pharmacology, vital dyes, and immunohistochemistry.

4. Linking distinct metabolic pathways to developmental events is a novel and exciting area of research.

Weaknesses of paper:

1. The majority of the manuscript uses t-tests for comparisons. In many cases, a 2-way ANOVA is needed for comparisons. It is not clear that all the findings will hold up to more rigorous statistical analysis.

2. Substantial work is needed to enhance the clarity of the manuscript. There is a limited introduction to the different facets of glucose metabolism highlighted in this paper. Metabolism is a complex topic that a broad audience will not be intimately familiar with. In addition, the paper is rich with data. The authors discuss differences in metabolism along 3 variables (space, time, and cell type). Occasionally it is unclear what type of difference is being discussed and the trend of the difference.

3. Manipulations inhibiting glycolysis and oxidative phosphorylation over extended periods of time could have many secondary effects on development related to a lack of energy supply.

Overall, this is a very interesting line of research. Currently, the manuscript needs revision prior to publication.

1. Statistics. The majority of the manuscript uses t-tests, which are not appropriate when making multiple comparisons (ie: proximal vs. distal over multiple stages of development). Currently, it is not clear that all the findings will hold up to more rigorous statistical analysis. For example, Figure 1I, 1J, 2C, 2D, 4D, 4F, 4G, 4I, 4J, all of Figure 4 Supp. 1, Figure 4 Supp. 2B, 5C, Figure 5 Supp. 1C, 6D, Figure 6 Supp. 1B, 7C, 8B, 8C, 9E, and 9F should be analyzed using two-way ANOVA. It is not statistically sound to pick only specific comparisons to make.

All statistical analysis has been re-done using 2-way ANOVA testing. We refer the reviewer to the data now included in table 1.

Additionally, in some cases, there are stars above bars with no indication of what condition is being compared to (ie. Figure 9E). Figure 3A lacks statistical analysis altogether. Finally, there are several cases where the authors say that there is a "significant" difference but do not show that it is statistically significant (for example, the caption of Figure 4I says there is a significant decrease between E9 and E14, but there is no indication of statistical significance in the figure). Differences that are not statistically significant can be discussed as trends, but should not be called significant. Similarly, differences that are statistically significant should not be discussed as "no difference" (ex. Figure 4E/F caption but Figure 4F shows a statistically significant difference; Figure 4 Supp. 2).

All concerns regarding significance and correct indication of this on figures have been addressed.

2. Enhance the clarity of the manuscript. The manuscript is quite well written but currently, there is limited text in the introduction outlining the relevant parts of glucose metabolism highlighted in this paper. Or how NADPH and NADH feed into these pathways. In addition, the authors discuss differences in metabolism along 3 variables (space, time, and cell type), and occasionally it is not clear what type of difference is being discussed – what is increased or decreased and where. Schematics like those in 2E are helpful to the reader.

Schematic indicating the different metabolic pathways and path of glucose catabolism has been added (see Figure 1).

3. Manipulations inhibiting glycolysis and oxidative phosphorylation over extended periods of time could have many secondary effects on development. The authors should lighten claims of causation. At a minimum, there should be a discussion of possible secondary effects and the need for more targeted manipulations to draw a convincing causative link between cellular metabolism and the development of tonotopy within the BP.

This section has now been significantly revised and schematics have been added to show cell-specific expression of metabolic enzymes and the metabolic processes they regulate.

4. The authors detect a proximal-to-distal gradient in tbound at E6 that "was still present at E14" and discuss this as a gradient "along the tonotopic axis between E6 and E14". However, their figure does not show a significant difference between the proximal and distal regions at E9. This could be related to a lack of statistical power (the n is lower for E9), but could alternatively represent a real fluctuation in the metabolic state over time. The authors should include a discussion of their E9 data, rather than glossing over it.

This concern has now been addressed in the discussion and the use of multiple inhibitors to target the same pathways was how we chose to address the concerns of off-target effects.

5. Although it may be beyond the scope of this paper, it would be nice to come full circle back to NAD(P)H imaging. Do the alteration to the PPP pathway lead to a loss of the NADPH/NADH gradient along the proximal-distal axis in the developing BP?

Following analysis using 2-way ANOVA the gradient in NADPH is significant at all stages from E6-E14.

6. The authors make several claims regarding differences in TMRM. They should also consider that this cationic vital dye may enter HCs via mechanotransduction channels. This form of entry may explain why labeling is stronger in older HCs compared to SCs or younger HCs.

This is important point and something we plan to address with future experiments. To do the FLIM imaging in acute explants we used morphological landmarks i.e otoconia of the utricle, saccule and Lagena at early stages and at E9 and E14 it is easy to differentiate between base and apex using the shape of the epithelium. This becomes significantly more difficult after 7 days in culture as the shape of the epithelium is slightly distorted. In the absence of cell-specific markers such as Phalloidin and calbindin. These experiments would require in ovo manipulation of metabolism.

[Editors’ note: what follows is the authors’ response to the second round of review.]

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

Essential revisions:

While all three reviewers felt that the revised manuscript is a substantial improvement there are still some outstanding issues:

1) The Chordin rescue experiment requires additional controls and clarification.

We fully agree with the reviewer regarding the necessity of this data and have thus included dose response experiments for Chordin like-1 in addition to media only controls and Chordin-like 1 treatments in the same experimental paradigm. Please see new data and analysis included in the revised Figure 10. All doses of Chordin like-1 are now clearly stated in Figure 11, in the main text (see line 348) and in the Methods section of the article (line 494).

We would also like to lessen the emphasis on any claim that metabolism is acting as a master regulator of HC tonotopy and instead highlight that a causal interaction between cytosolic glucose metabolism and Bmp7/Chdl-1 signalling is necessary for specifying HC size and regulating patterning of the mosaic at the epithelial surface (see revised paragraph beginning line 328).

2) Additional inconsistencies and quantification were requested by all three reviewers.

New data and additional quantification included in the manuscript as requested by the reviewers.

  • Additional quantification of hair cell density and nuclear area have been included for 2DOG, SAM, 6AN, YZ9 and shikonin.

  • Additional experiments have been included investigating a role for PKM2 in tonotopic patterning during development (see Figure 8).

  • As highlighted by all three reviewers, the previous expression data for IDH3A and LDHA did not fit logically within the study and caused confusion about which metabolic pathway was being probed. The data for IDH3A and LDHA are therefore no longer included in the current manuscript and the text has been revised accordingly.

  • Quantification of EdU positive cells has now been included (Figure 6 Supplement 2).

  • Cell density analysis has been included for Chdl1+2DOG and Chdl1 (0.4 g/mL) alone (new Figure 11 Supplement 1).

  • Dose response data have been added for Chdl1 as requested by the reviewers (Figure 11)

  • Further statistical tests have been performed on proximal vs proximal treated and distal vs distal treated explants (see revised Figure 6).

  • Further experiments have been conducted to investigate the pH gradient along the BP at more mature developmental stages. (see new Figure 5 Supplement 3).

  • Additional RNA scope analysis has been included for both BMP7 and PKM2 at E8, showing the tonotopic expression gradients for both genes (see new Figure 3 Supplement 4 and Figure 4 Supplement 1). New data showing the difference in PKM2 expression between proximal and distal regions has also been included at E10 (see Figure 4 Supplement 2) and E12 (see Figure 4 Supplement 3).

  • Negative RNA scope probe controls have been included for BMP7, PKM2 and SOX2 (see Figure 3 Supplement 5).

  • Additional in situ data have been included, showing BMP7 and CHDL1 expression following treatment with 6AN (see Figure 10 Supplement 1).

  • Data showing explants treated for 72h with UK5099 has been added as a Supplement to Figure 9 (new supplementary Fig?). As recommended by the reviewer ??, these data aim to support the claim that blocking mitochondrial OXPHOS causes significant a developmental delay (see Figure 9 Supplement 1).

3) Comparing treatments of SAM or 6-AN to 2-DOG in the ability to affect Bmp7 signal, for example.

Additional in situ experiments for BMP7 and CHDL-1 have been conducted with 6-AN (Figure 11 Supplement 1).

Reviewer #1 (Recommendations for the authors):

This revised manuscript is much improved and streamlined. The authors provided additional results showing that Chordin like-1, which is normally expressed in the proximal region of the BP could rescue the distalizing effect of 2-DOG in the proximal region, thus strengthening the link between metabolism and tonotopy. However, in this type of rescue experiment, 2-DOG and Chordin like-1 (Chdl-1) treatments should be included as controls within the same experiment, rather than using controls with no treatment alone. While readers could extrapolate 2-DOG effects from other figures, there was no description of Chdl-1 in dosage, condition, or results in the revised manuscript. Furthermore, the statement in lines 800-802 and results in Supplementary Table 1 seem to suggest that Chdl-1 treatment alone does not affect the tonotopic gradient by itself. Is that true? If so, this result is not consistent with previous studies.

Chdl-1 dose response experiments are now included in Figure 11 with doses clearly indicated.

Reference to the dose used for treatments (0.4 g/mL) is now stated in the main text of the Results (lines 348 and 494) and in the Methods. Treatments with Chdl-1 alone show disruption of the normal tonotopic gradient. These data are included in Figure 11 and highlighted additionally in the statistical data presented in Supplementary Tables 1 and 2.

Editorial suggestions to help improve the manuscript:

1) Line 28, what does flattening of hair cell morphology mean? Do you mean shorter hair bundles in the proximal versus the distal BP? If so, the images in this study are not clear enough for naïve readers to draw that conclusion.

This statement has been changed in the Abstract and in the main text to read:

“…abolishes tonotopic patterning and normalises the graded differences in hair cell morphology along the BP…” (Line 26).

The phrasing in Line 214 has also been changed for clarification.

2) Line 54, what does inner sense organs means? Are you referring to the inner ear or tissues deep in the body?

We have changed this to ‘developing inner ear epithelia’ (line 53).

  • Labels have been added to indicate the different cellular compartments.

  • Numbering has been removed for clarity.

  • Font size has been increased for all pathways.

  • Mitochondrial membrane has been more clearly labelled.

  • Site of shikonin (PKM2 inhibitor) activity added to schematic.

4) Lines 293 and 295, a bit odd to split PKM2 results into two separate paragraphs.

This section of the manuscript has been significantly re-written. The paragraphs for PKM2 are now combined and additional data has been included (see section beginning line 160).

We also refer the reviewer to the new data for PKM2 expression included in Figure 4 Supplements 1, 2 and 3.

5) Figure 2 a) Consider combining panels D and K, and expand B and C. Hard to read the label in C, b) I am still not clear why αbound level (dotted purple line) is in a steady state based on the purple and green curve and there is no explanation of αbound in the text, c) Panels E-H could be slightly bigger, d) move Scale bars=50um to the end of description for (E-H).

We thank the reviewer for their important comments. We have addressed all recommended edits and hope that the revised Figure 2 is now much clearer.

6) Line 205, probably better just to state τbound gradient not related to glucose uptake?

This statement has been reworded as follows:

The gradient in NADPH/NADH reported by τbound (Figure 2 D-I) does not arise from variations in the balance between cytosolic and mitochondrial ATP production, as often observed in development but from differences in the route of cytosolic glucose processing utilised. A new schematic has also been included for clarification of what τbound is showing (see Figure 2 Supplement 1).

7) Line 236, the sentence is a bit awkward.

Thank you and this sentence has been reworded.

8) Line 292, define PKM2 when it first appeared in the text.

Pyruvate kinase M2 defined in main text at first mention.

9) Line 339, Figure 3 or Figure 5?

Corrected to read Figure?

10) Line 342, Figure without a number.

This Figure/text has been removed.

11) Lines 340-343, State in the text lower pHrodo Red signal meant higher pH.

Thank you for this comment and we have now clearly stated the in both the legend for Figure 5 and in the main text.

12) No quantification of Edu-labeled cells in Figure 6 suppl 2.

We apologize for not including quantification, which has now been added.

13) Line 484, does SAM inhibit both glycolysis and PPP pathways similar to 2-DOG?

Beginning line 243 “To further confirm a role for cytosolic glucose metabolism in establishing HC positional identity, we employed a second method of modulating the pathway independently of HK inhibition. Cytosolic glycolysis can be blocked indirectly by increasing cytosolic levels of the metabolite s-adenosyl methionine (SAM).”

In response to the reviewer, SAM does not act in the same way as 2DOG to block cytosolic glycolysis. 2-DOG inhibits at the level of hexokinase, the first enzyme in the man glycolysis pathway and SAM affects glycolysis indirectly by altering the activity of metabolic enzymes in the pathway

(Pascale et a; 2019) doi:10.3390/medicina55060296

(Yu et al., 2019) doi:10.1016/j.molcel.2019.06.039. Epub 2019 Aug 13

We hope this is has been sufficiently clarified in the revised main text. See paragraph beginning line 242.

14) Typo in the last sentence of the legend for Figure 7 supplement 1.

Corrected.

15) Incomplete last sentence found in the legend for Figure 8.

Corrected.

16) Figure 9. The images in E are better in this revision. However, there is some inconsistency between the text and the figure legend. In the text, explants were established at E6.5 and incubated for 72 hours. In the legend, it was stated the in vitro incubation was for 7 days.

17) Figure 9. It may be better to put the number and statistics in the panels that they belong rather than at the end of the figure. Sample numbers need to be provided for panel E as well.

We thank the reviewer for identifying the error in the Figure legend. The legend for Figure 9 (now Figure 10) has been corrected to read:

“…treated with 2-DOG +NaP from E6.5 for 72 hours in vitro, as stated in the main text of the article.”

Legend order has been rearranged as suggested by the reviewer. Samples numbers have been added to panel E.

18) Supplemental Table 1. Are all p values in the left-hand columns controls? Confused by the statement in line 799, "control or inhibitor treatment groups". Or it should be controlled for inhibitor treatment groups? If so, the p values in the right-hand column are results from individual treatments. Then, Chdl-1 treatment did not affect the tonotopic differences observed in controls?

We refer the reviewer to the amended version of Supplementary Table 1. For clarification, data have been divided into statistical outcomes for luminal surface area and nuclear area. Please see new Supplementary tables 1 and 2.

Reviewer #2 (Recommendations for the authors):

Figure 7 supplementary 2.

Based on the statistical analysis, the authors concluded that YZ9 treatment had no effect on HC morphology along the tonotopic axis. However, the graph in (B) shows that the luminal area of hair cells was dramatically reduced in the proximal but not in the distal region by YZ9 treatment, resulting in no significant difference between the proximal and distal regions. This suggests a loss of tonotopic identity of HC morphology by YZ9 treatment, similar to 2-DOG, SAM, or 6-AN. To support the authors' argument, it is important to confirm that YZ9 treatment did not induce tonotopic changes in HC morphology. Double-checking with other tonotopic parameters, such as HC nuclear area and cell number, will help to clarify this issue.

We thank the reviewer for their comments and recommendations to clarify this issue. Additional quantification of HC number and nuclear area have been included for all inhibitor treatments and statistical data added to Supplementary Tables 1 and 2.

Figure 9.

(E) The Bmp7 expression domain in the control explant is peculiar in that it has a lining around the distal region, unlike the control image in Figure 9 supplement 1. Figure 3 supplement 4 shows that Bmp7 is normally expressed in the sensory domain.

Bmp7 in situ signals are generally stronger in 2-DOG-treated BP than in control, but there still seems to be a proximal-distal gradient with a generally stronger in situ signal. qRT-PCR with each half of the BPs, as done in their previous publication, will clearly show if the gradient is disrupted.

We thank the reviewer for raising this concern, and fully agree that from the in situ data alone it is challenging to make any assumption regarding the change in either gradient.

We would like to clarify that we are not proposing that glucose catabolism sets up either of the morphogen gradients. We believe the Bmp7 gradient is set up, as shown by Son et al. (2015) by the activity of Shh from the ventral midline structures.

The new in situ data included in Figure 10 Supplement 1 highlight the complicated regulation of BMP7 and CHDL1-1 expression by different metabolic pathways.

2DOG causes an increase in BMP7 and a decrease in CHDL-1 along the entire organ but, as highlighted by the reviewer the gradients of both morphogens appear maintained. We therefore feel that qRT-PCR analysis is unlikely to reveal different results regarding either expression gradient. 6AN reduces CHDL-1 in the proximal but not distal BP region and causes a subtle expansion of BMP7 towards the proximal end of the BP (see new Figure 10 Supplement 1). The reduced CHDL-1 combined with increased BMP7 following treatment with either 2DOG or 6AN would however be sufficient to induce more distal-like HC morphologies in the proximal region of the tissue (please see Figure 10 and Figure 10 Supplement 1).

We propose that the concentration of BMP7 and CHDL1 seen a HC at a given point along the BP, coupled with a unique metabolic phenotype regulates epithelial patterning and HC shape (HC nuclear area, LSA and density) along the developing BP. Without having performed a detailed physiological analysis, we would be reluctant to make any firm conclusion regarding the ‘HC fate’ or frequency tuning that accompanies the observed morphological changes. Understanding the precise mechanism linking cell shape with physiological function along the developing BP requires further investigation, which we feel is beyond the scope of the current study.

Figure 9 supplement 1.

The previous review questioned whether the changes in tonotopic identity caused by 2-DOG were due to the inhibition of glucose metabolism rather than other cellular effects. This question was crucial to support the authors' conclusion, as 2-DOG has been shown to affect several cellular processes other than glycolysis. The authors showed that blocking mitochondrial metabolism did not alter the Bmp7 gradient, but this doesn't solve the problem of the specificity of 2-DOG.

We agree with the reviewer that when using pharmacological inhibitors there are likely to be offtarget effects. Given the conserved morphological changes elicited across multiple inhibitors (see Supplementary Tables 1 and 2), we are confident however that the observed effects on cell morphology arise due to perturbation in glucose flux and not from off-target effects of the drug.

We agree that this is an important point to address and therefore offer the following response to address the concerns raised by the reviewer.

  • 2-DOG is a widely accepted inhibitor of hexokinase and thus glycolysis, so much so that it is the biochemical basis for PET scanning in cancer. Off-target effects are not widely reported. We offer the following citations for 2DOG in hope to satisfy any concerns raised by the reviewer:

  • 2-DOG can stimulate IGF1R signalling (https://doi.org/10.1074/jbc.M109.005280), thereby activating Akt by inducing its phosphorylation (https://doi.org/10.1158/1535-7163.MCT-070559). However, Akt would be unable to activate glycolysis under these conditions as hexokinase would remain inhibited.

  • 2-DOG can also activate autophagy pathways (https://doi.org/10.3390/ijms21010234) but this further is downstream of glycolysis, due to the resultant reduced ATP/AMP ratio.

  • It can inhibit N-glycosylation, leading to ER stress (https://doi.org/10.3390/ijms21010234). However, this results from the structural similarities between D-glucose and D-mannose, so if an alternative way to inhibit cytosolic glucose metabolism has the same effect, this offtarget effect (i.e. SAM, 6AN and shikonin) cannot be responsible for our observations.

It is worth investigating whether SAM and 6-AN, which induce luminal area changes similar to 2-DOG, also disrupt the same molecular pathways, such as the Bmp7 and Chld1 gradients.

New in situ data have been included for treatments with 6AN and 6AN + shikonin (see Figure 11 Supplement 1).

Reviewer #3 (Recommendations for the authors):

This manuscript is substantially improved compared to the previous version. Thank you for all the hard work you put into responding to reviewer concerns. A revised manuscript would be welcomed by this reviewer.

– The RNAseq and microarray analyses are good for hypothesis generation, but the differential expression the authors detect often does not match the immunohistochemistry and RNA scope data they provide. It is especially confusing to say that there is no difference in LDHA and IDH3A protein expression across the BP, then suggest that differential RNA expression of these genes could drive the metabolic gradient the authors observe.

We agree fully with the reviewer that the original data included in Figure 4 did not fit logically with the proposed hypotheses of the study. We have since removed these data, as the focus of the study is about differences in cytosolic glucose flux along the organ and not metabolic switching. We have included a new schematic (Figure 2 Supplement 1) to highlight to metabolic differences reported by FLIM analysis. As IDH3A and LDHA regulate switching between OXPHOs and glycolysis, we have revised the current version of the manuscript accordingly and focused more on the role and expression of PKM2, an enzyme that regulates both glucose flux into the PPP, and the NADPH/NADH ratio. We hope that this resolves the previous confusion.

Some of the variability in the data may be related to differences in animal age across experiments, but this variability makes it a bit challenging to interpret the data. In addition, not all the genes listed in Figure 3 Supp. 4A seems to reach statistical significance. The authors should consider framing Figure 3 Supp. 4A as a way to generate hypotheses that they then test with additional experiments, rather than claiming that they identified a whole set of differentially expressed genes when in fact they present contradictory data within the same manuscript. They should also clarify how they decided which metabolic RNAs to examine further.

We fully agree with the reviewer’s suggestion regarding the RNAseq and RNA scope data and have therefore conducted additional analysis and experiments investigating the role of the gate keeper enzyme PKM2 (Figure 4 Supplements 1-3, Figure 8, Figure 11 Supplement 1). PKM2 was chosen for further analysis, as it regulates glucose flux into the PPP or towards pyruvate and the NADPH/NADH ratio. This has also been clarified in the main text (see lines 157, 160 and 175).

– Although the authors used BMP7 as a positive control for RNA scope, there does not appear to be a proximal-to-distal difference in BMP7 fluorescence.

Thank you for this important observation. To address this point, new data showing serial crosssection of the BP at E8 has been included showing the distal-to-proximal gradient in Bmp7 (Figure 3 Supplement 4).

In addition, the authors do not provide any negative controls to show a lack of fluorescence in the absence of gene-specific RNA scope probes. Without successful controls, the experimental data is hard to interpret. The authors should consider providing better controls.

We concur with the reviewer and have now included RNA scope controls for all probes used (Figure 3 Supplement 5).

– Line 297-298: "By controlling a metabolic feedback loop that regulates the switch between glycolysis and OXPHOS…" The authors previously showed data suggesting that differences in the NAD(P)H/NADH ratio are not due to changes in mitochondrial OXPHOS. Here, though, they suggest that PKM2 controls the amount of OXPHOS and is differentially expressed across the BP. How can we reconcile these two points?

In the current study, we propose that the dimeric form of PKM2, regulated by pH might control glucose flux into the PPP or towards pyruvate. Given the low pH, high NADPH/NADH ratio and expression of PKM2, our data suggest activity consistent with glucose flux into the PPP rather than pyruvate production. This would agree with the lack of a TMRM gradient and thus mitochondrial OXPHOS along the BP.

Please see the revised paragraph beginning line 188.

– Figure 4, 5: How was the quantification in 4B-D and 5B performed? Are fluorescence values provided per HC/per SC? As HC size varies across the proximal-distal axis of the BP, it is not fair to quantify raw total fluorescence. The fluorescence should be normalized by the area occupied by each cell type.

IDH3A and LDHA immunofluorescence data are no longer included in the current version of the manuscript. Additional information on how the analysis was performed has however been included with Figure 5 (Figure 5 Supplements 1 and 2).

– Figure 5: Could proximal-to-distal differences in pHrodo Red fluorescence result from the gradient of HC differentiation (i.e. more mature distal HCs, less mature proximal HCs) across the tonotopic axis at E9? In other words, do young HCs have a higher pHi that decreases as hair cells mature? This could explain the proximal-to-distal differences shown in Figure 5.

We have included additional data investigating the pH gradient at E14 (see Figure 5 Supplement 3).

At later stages, as predicted by the reviewer, we do indeed find that the pH gradient is reversed. However, we are confident this does not affect specification of positional identity in nascent HCs, as tonotopic patterning is established between E6-E7 (Mann at al., 2014).

– Figure 6 Supp. 2 – To claim that there is no increased proliferation in response to 2-DOG, EdU+ cells should be quantified in control vs. 2-DOG cultures. It seems that the authors have only provided general cell counts (in C) and not EdU+ cell counts.

We fully agree with the reviewer and have now included new EdU quantification for this experiment (see Figure 6 Supplement 2).

– Do the authors have any evidence that the various drug treatments are not simply blocking cell growth and differentiation? As mentioned by another reviewer, the distal-like phenotype could be a "default"; perhaps proximal HCs simply aren't maturing. It would be nice to show examples of cultures at 1DIV vs. 7 DIV to demonstrate how much cells grow during this time and how treated cultures at 7 DIV compare to immature 1 DIV cultures. (SAM treatment especially looks like it might impact hair bundle morphology in the proximal BP.)

We thank the reviewer for raising this important point and, to address the concern, have included additional data showing the effect of UK5099 during short-term culture. We are confident that blocking mitochondrial OXPHOS causes developmental delay, but that perturbing glucose flux alters normal patterning. As indicated in Figure 9 Supplement 1, the bundle morphology at earlier stages in cultures (E9.5) is immature compared to cultures left to develop for the full 7 days in culture, even in the presence of cytosolic glucose flux inhibitors.

– Figures 7 Supp. 2 and Figure 8 continue to rely on a single morphological measure of HC morphology as a readout of tonotopy. The authors should add additional quantification (cell density, nuclear area…), as has been provided for other manipulations.

As recommended by the reviewer, additional quantification of cell density and nuclear area have been included for all inhibitor treatments. Please see revised Figure 6, Figure 6 Supplement 3, Figure 7, Figure 7 Supplement 2, Figure 8 and Figure 11.

– Figure 9 and Figure 9 Supp. 1: In the absence of quantification of the in situ hybridization data, please state how many examples show similar Bmp7 and Chdl1 gradients.

Information regarding the numbers of biological replicates have now been clearly stated in the legends for revised Figure 10 and Figure 10 Supplements 1 and 2.

– To bolster the claim that 2-DOG treatment is specifically disrupting tonotopy and not just the differentiation of proximal HCs (as pointed out by Reviewer 1), the authors should quantify HC density in Figure 10. If HC density is also restored by providing Chdl-1 in the presence of 2-DOG, it will bolster the hypothesis that altered glucose metabolism has a specific effect on HC positional identity.

We fully agree with the reviewer and have therefore included quantification of HC density in control, Chdl1+2DOG and Chdl1 0.4 g/mL treated explants. Please see Figure 11 Supplement 1.

[Editors’ note: what follows is the authors’ response to the third round of review.]

The manuscript has been improved but there are some remaining editorial issues that need to be addressed, as outlined below:

Reviewer #1:

The results of the revised manuscript are coming together nicely. To appropriately demonstrate the rescue effect of Chd1 on 2DOG requires the control, 2-DOG, and 2-DOG+Chd1 be all conducted within the same experiment. Although such an experiment was not provided, the authors have made a good faith effort in demonstrating a dose-response effect of chordin in reducing the differential differences between proximal and distal BP, and the revision is acceptable. Additionally, I have some editorial suggestions. These are merely suggestions, and the acceptance of the manuscript is not dependent on making these changes.

1) The experimental evidence suggests that the tonotopic morphogens are regulated by glucose metabolism. I don't see evidence showing that these morphogens regulate metabolism in return. Therefore, I question the choice of using the phrase "Cross talk" in the title.

We agree with the reviewer on this point and have therefore re-worded the title as follows: Causal gradients in glucose metabolism and morphogen signalling specify tonotopic identity in developing hair cells.

2) The abstract, though clear, could be more concise and informative. I feel like it is important to summarize this 32-figure manuscript well!

We thank the reviewer for this suggestion and agree regarding the importance of a concise, clear summary. We have therefore reworded the abstract and hope this more clearly outlines the take home findings of the study.

3) Include PKm2 in the schematic diagram of Figure 1.

Pkm2 (pyruvate kinase) has been added to the schematic in Figure 1.

4) The first part of the Discussion reads more like an introduction.

We appreciate this comment and have re-worded the opening paragraph of the discussion.

Reviewer #2:

The revised manuscript is much improved with additional experimental data that strongly support the authors' conclusions. I appreciate the authors' efforts to make the manuscript in much better shape.

We very much thank the reviewer for their favourable comments in response to the revised version of this article.

I have just one question. From the diagrams in Figure 7-supplement 2B and Figure 8B, it appears that PFK and PKM2 act in the same glycolysis pathway. However, inhibition of either resulted in different phenotypes. While inhibition of PFK by YZ9 did not affect tonotopic patterning, inhibition of PKM2 by shikonin disrupted tonotopic patterning, albeit at the epithelial level. Why did two inhibitors of the same pathway have different effects on tonotopic patterning?

This is an important point, and we thank the reviewer for drawing attention to it. The following explanation has been added to the main text for clarification – line 285. We would like to clarify that although YZ9 (Pfk) and shikonin (Pkm2) both inhibit parts of glycolysis, noting the two enzymes regulate different stages of the pathway with respect to glucose flux and the re-entry of PPP products. Inhibition of Pkm2 would cause build-up of both G3p and F6p, whereas inhibition of PFK would only lead to accumulation of F6P. From this it can be assumed that Pkm2 inhibition would have a greater impact on the activity of the PPP and therefore, given the gradient in NADPH, on tonotopic patterning.

Reviewer #3:

The manuscript is greatly improved and nearly ready for publication.

There are a few typos and points to clarify that I would recommend to the authors. These recommendations should not require additional experiments.

List of recommendations for clarity:

Figure 2 Supplement 1 – this Figure is really helpful and we would recommend making it part of a main figure (perhaps you can condense Figure 2G, as these two plots show the same data, in favor of this helpful schematic). The clarity of the FLIM section is massively improved since the first version of the paper. The one link that is still hard to make is how the values of Tbound inform flux through the PPP vs. glycolysis. This figure beautifully illustrates exactly that. Also, consider adding PPP vs. glycolysis to the gradient in the Figure 2H schematic, and/or moving the sentence in line 109 to line 101 in the main text. Although all of this knowledge is obvious to those who think about metabolism all the time – these concepts can be confusing for someone not in the field.

We thank the reviewer for these helpful suggestions and agree that these changes make Figure 2 significantly clearer. Please see revised Figure 2, which now contains the recommended edits. To make the description of our data in this section of the results clearer, we have moved the sentence from line 109 to line 101 as recommended by the reviewer.

We would recommend adding Pkm2 to Figure 1, as there is an entire figure focused on Pkm2.

We have now added pyruvate kinase (Pkm2) to the schematic in Figure 1.

Figure 3 Supp. 3E. Are you detecting statistically significant proximal-to-distal changes in Bmp7 here? It is not indicated in the quantification (no stars or "ns"). If the positive control is not showing a statistically significant change, it is hard to interpret the other data.

Although the distal-to-proximal gradient of Bmp7 was consistent in all 5 biological replicates, the high variation across samples meant that significance was not reached in the data. To address this issue, we therefore provided additional confirmation of the distal-to-proximal gradient using serial sections along the entire proximal-to-distal axis. The other mRNAs (Got2 and Ldhb) shown in Figure 3 supplement 3 were not significantly different along the BP and were therefore not pursued further in relation to the NADPH gradient along the BP.

https://doi.org/10.7554/eLife.86233.sa2

Article and author information

Author details

  1. James DB O'Sullivan

    Centre for Craniofacial and Regenerative Biology, Faculty of Dentistry Oral and Craniofacial Sciences, King's College London, London, United Kingdom
    Contribution
    Data curation, Formal analysis, Investigation, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
  2. Thomas S Blacker

    Research Department of Structural and Molecular Biology, University College London, London, United Kingdom
    Contribution
    Conceptualization, Methodology, Writing - original draft, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-8949-6238
  3. Claire Scott

    Centre for Craniofacial and Regenerative Biology, Faculty of Dentistry Oral and Craniofacial Sciences, King's College London, London, United Kingdom
    Contribution
    Data curation, Formal analysis, Investigation, Methodology
    Competing interests
    No competing interests declared
  4. Weise Chang

    National Institute on Deafness and Other Communication Disorders, National Institutes of Health, Bethesda, United States
    Contribution
    Data curation, Investigation, Methodology
    Competing interests
    No competing interests declared
  5. Mohi Ahmed

    Centre for Craniofacial and Regenerative Biology, Faculty of Dentistry Oral and Craniofacial Sciences, King's College London, London, United Kingdom
    Contribution
    Data curation, Investigation, Methodology
    Competing interests
    No competing interests declared
  6. Val Yianni

    Centre for Craniofacial and Regenerative Biology, Faculty of Dentistry Oral and Craniofacial Sciences, King's College London, London, United Kingdom
    Contribution
    Formal analysis, Methodology
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-9857-7577
  7. Zoe F Mann

    Centre for Craniofacial and Regenerative Biology, Faculty of Dentistry Oral and Craniofacial Sciences, King's College London, London, United Kingdom
    Contribution
    Conceptualization, Resources, Data curation, Formal analysis, Supervision, Funding acquisition, Investigation, Visualization, Methodology, Writing - original draft, Project administration, Writing – review and editing
    For correspondence
    zoe.mann@kcl.ac.uk
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4916-9574

Funding

Biotechnology and Biological Sciences Research Council (BB/V006371/1)

  • Zoe F Mann

Physiological Society

  • Zoe F Mann

King’s Prize Fellowship

  • Zoe F Mann

The funders had no role in study design, data collection, and interpretation, or the decision to submit the work for publication.

Acknowledgements

The authors wish to thank Dan Jagger, Matthew Kelley, Jeremy Green, Michael Duchen, Gyorgy Szabadkai, and Thomas Coate for providing important comments and critical discussion on the manuscript. This project was supported by funds from the King’s Prize Fellowship (King’s College London) (ZFM), The Physiological Society (Physiological research Grant) (ZFM), and the Biotechnology and Biological Sciences Research Council (BBSRC) grant BB/V006371/1 (ZFM).

Senior Editor

  1. Kathryn Song Eng Cheah, University of Hong Kong, Hong Kong

Reviewing Editor

  1. Doris K Wu, National Institutes of Health, United States

Reviewers

  1. Doris K Wu, National Institutes of Health, United States
  2. Jinwoong Bok, Yonsei University, Republic of Korea
  3. Katie Kindt, National Institutes of Health, United States

Version history

  1. Preprint posted: April 11, 2022 (view preprint)
  2. Received: January 17, 2023
  3. Accepted: August 3, 2023
  4. Accepted Manuscript published: August 4, 2023 (version 1)
  5. Version of Record published: August 14, 2023 (version 2)

Copyright

This is an open-access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.

Metrics

  • 651
    Page views
  • 119
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. James DB O'Sullivan
  2. Thomas S Blacker
  3. Claire Scott
  4. Weise Chang
  5. Mohi Ahmed
  6. Val Yianni
  7. Zoe F Mann
(2023)
Gradients of glucose metabolism regulate morphogen signalling required for specifying tonotopic organisation in the chicken cochlea
eLife 12:e86233.
https://doi.org/10.7554/eLife.86233

Further reading

    1. Cell Biology
    2. Microbiology and Infectious Disease
    Heledd Davies, Hugo Belda ... Moritz Treeck
    Tools and Resources

    Reverse genetics is key to understanding protein function, but the mechanistic connection between a gene of interest and the observed phenotype is not always clear. Here we describe the use of proximity labeling using TurboID and site-specific quantification of biotinylated peptides to measure changes to the local protein environment of selected targets upon perturbation. We apply this technique, which we call PerTurboID, to understand how the P. falciparum exported kinase, FIKK4.1, regulates the function of the major virulence factor of the malaria causing parasite, PfEMP1. We generated independent TurboID fusions of 2 proteins that are predicted substrates of FIKK4.1 in a FIKK4.1 conditional KO parasite line. Comparing the abundance of site-specific biotinylated peptides between wildtype and kinase deletion lines reveals the differential accessibility of proteins to biotinylation, indicating changes to localization, protein-protein interactions, or protein structure which are mediated by FIKK4.1 activity. We further show that FIKK4.1 is likely the only FIKK kinase that controls surface levels of PfEMP1, but not other surface antigens, on the infected red blood cell under standard culture conditions. We believe PerTurboID is broadly applicable to study the impact of genetic or environmental perturbation on a selected cellular niche.

    1. Cell Biology
    Bo Wang, Zheyong Liang ... Peijun Liu
    Research Article

    The primary cilium plays important roles in regulating cell differentiation, signal transduction, and tissue organization. Dysfunction of the primary cilium can lead to ciliopathies and cancer. The formation and organization of the primary cilium are highly associated with cell polarity proteins, such as the apical polarity protein CRB3. However, the molecular mechanisms by which CRB3 regulates ciliogenesis and the location of CRB3 remain unknown. Here, we show that CRB3, as a navigator, regulates vesicle trafficking in γ-tubulin ring complex (γTuRC) assembly during ciliogenesis and cilium-related Hh and Wnt signaling pathways in tumorigenesis. Crb3 knockout mice display severe defects of the primary cilium in the mammary ductal lumen and renal tubule, while mammary epithelial-specific Crb3 knockout mice exhibit the promotion of ductal epithelial hyperplasia and tumorigenesis. CRB3 is essential for lumen formation and ciliary assembly in the mammary epithelium. We demonstrate that CRB3 localizes to the basal body and that CRB3 trafficking is mediated by Rab11-positive endosomes. Significantly, CRB3 interacts with Rab11 to navigate GCP6/Rab11 trafficking vesicles to CEP290, resulting in intact γTuRC assembly. In addition, CRB3-depleted cells are unresponsive to the activation of the Hh signaling pathway, while CRB3 regulates the Wnt signaling pathway. Therefore, our studies reveal the molecular mechanisms by which CRB3 recognizes Rab11-positive endosomes to facilitate ciliogenesis and regulates cilium-related signaling pathways in tumorigenesis.