Gradients in cellular metabolism regulate development of tonotopy along the chick cochlea

8 In vertebrates with elongated auditory organs, mechanosensory hair cells (HCs) are organised such 9 that complex sounds are broken down into their component frequencies along the basal-to-apical long 10 (tonotopic) axis. To generate the frequency-specific characteristics required at the appropriate 11 positions, nascent HCs must determine their tonotopic properties during development. This relies on 12 complex signalling between the developing HC and its local niche along the axis within the auditory 13 organ. Here we apply NAD(P)H fluorescence lifetime imaging (FLIM) and live imaging of mitochondria 14 to reveal metabolic gradients along the tonotopic axis of the developing chick cochlea. We further 15 show that re-shaping these gradients during development, by inhibiting cytosolic glucose metabolism, 16 alters normal Bmp7 and Chordin like-1 signalling leading to flattening of tonotopic patterning along 17 the axis. Our work supports a causal link between morphogen signalling and metabolic reprogramming 18 in specifying tonotopic morphologies in auditory HCs. 19 HC process allowing cells arising from a common progenitor population to differentiate with metabolic states. Extensive changes in cellular redox state occur during the first postnatal week of functional refinement in cells of the mouse cochlea 12,13 . These changes indicate local metabolic switches during the early phases of HC and SC functional maturation 12 . We report cell-specific differences in metabolic enzyme expression, NAD(P)H FLIM signatures and mitochondrial activity in the cochlear epithelium from as early as E14 in mouse and E9 in chick. These cell-specific differences 381 may reflect the developmental origins of metabolic coupling between HCs and SCs. Such cooperation is observed between neurons and astrocytes in the developing brain 68 . Neurons and astrocytes preferentially use different metabolic pathways under physiological conditions, due in part, to cell specific expression patterns of genes regulating energy metabolism 72 . Metabolic coupling is essential 385 for normal brain function and has also been linked with long-term function and survival of cochlear hair cells 69,70 . Long-term hair cell function is not possible without auxilliary input from supporting cell 387 neigbours. Understanding the developmental programs and specific factors that determine future HC- SC coupling within the cochlear niche is essential if we are to recapitulate such mechanisms and sustain long-term HC viability in regenerative or organoid-type models.


Introduction 20
Hearing in mammals and lower vertebrates relies on the function of mechanosensory hair cells (HCs) 21 and their associated glial-like supporting cells (SCs). In both mammals and birds, complex sounds are 22 separated into component frequencies by the anatomy of the cochlea, so that different frequencies 23 stimulate HCs located at different positions along the basal-to-apical long axis of the epithelium 24 (tonotopy). This phenomenon underlies our ability to differentiate between the high pitch of a 25 mosquito and the low rumbling of thunder. The specific factors regulating tonotopy remain, largely, 26 unclear. A better understanding of how specific factors regulate tonotopy and, therefore, the 27 formation of frequency-specific HC phenotypes would provide insights regarding both congenital 28 auditory defects and HC regeneration. In particular, high frequency HCs show increased vulnerability 29 to insults, including aging 1 , noise damage 2,3 and ototoxicity 4 . Therefore, information regarding the 30 pathways that specify this HC phenotype could identify potential methods to preserve these cells 31 and/or recover high frequency hearing loss. 32 Metabolism, encompassing the complex network of chemical reactions that sustain life, has emerged 33 as a key regulator of cell fate and differentiation 5 . Instructive roles for glucose metabolism and 34 reprogramming between glycolytic and oxidative pathways has been reported in various systems 35 during development including, for example, delamination and migration in embryonic neural crest 6 , 36 hair cells in the zebrafish inner ear 7 , specification of trophectoderm fate in the mouse embryo 8 and 37 cell fate decisions along the developing body axis 9-11 . As a complex tissue containing multiple cell 38 types, investigating how metabolism regulates HC formation within the cochlear niche requires 39 experimental approaches capable of interrogating metabolic pathways in live preparations, with single 40 cell resolution. The classic biochemical approaches from which our knowledge of metabolism has 41 formed involve the destructive extraction of metabolites from a sample for subsequent analysis. 42 Probing metabolism in this way, although valuable, means that the spatial organisation of complex 43 tissues is lost. We have previously exploited the intrinsic fluorescence of reduced nicotinamide 44 adenine dinucleotide (NADH) and its phosphorylated analog NADPH to investigate the metabolic state 45 in HCs and SCs cells of mature cochlear preparations during aging and ototoxicity 12,13 . Both NADH and 46 NADPH serve as reporters of metabolism because they function as redox regulators, transporting 47 reducing equivalents between the enzyme-catalysed reactions of cellular metabolism. 48 Reprogramming of metabolic pathways in tissues causes changes in the time-resolved properties of 49 the NAD(P)H fluorescence signal 14 . Fluorescence lifetime imaging (FLIM) of cellular NAD(P)H therefore 50 provides a non-invasive and label-free method with which to probe metabolism at the single cell level 51 throughout cochlear development 15 . 52 Using previously defined morphological markers of tonotopy as read-outs for unique HC phenotype 53 16,17 we here investigate a role for metabolism in specifying proximal (high frequency) verses distal 54 (low frequency) HCs. We first identify a gradient in NADPH-linked metabolism along the tonotopic axis 55 of the developing chick cochlea. This gradient is then further confirmed through analysis of metabolic 56 mRNAs and the expression patterns of metabolic effector proteins along the tonotopic axis. We 57 functionally test the role of these gradients by modulating metabolic flux through different branches 58 of glucose metabolism at key stages in HC formation. We find that glucose flux specifically through 59 the pentose phosphate pathway (PPP) instructs tonotopic HC morphologies, by regulating the graded 60 expression of Bmp7 and its antagonist Chordin-like-1, known determinants of tonotopic identity 18 . 61 We also characterise metabolic differences and the timing of this divergence in nascent HCs and SCs. 62 Our findings highlight a novel role for cytosolic glucose metabolism in specifying HC positional identity 63 and provide the first characterisation of unique metabolic states associated with the emergence of HC 64 and SC lineages.
phenotype 19 . Based on this hypothesis, the changes in bound we observe here would indicate dynamic 90 shifts between glycolytic and oxidative pathways throughout BP development ( Figure 1J). 91 92 As HCs and SCs are not easily identified at stages before E9 (Figure 1 D, E), the NAD(P)H FLIM data 93 were initially analysed from the whole epithelium along the proximal to distal axis rather than from 94 specific cell types. However, at E14 and E16 we were able to specifically examine bound and bound in 95 HCs and SCs (Figure 2A, B). We found that both parameters differed significantly between cell types 96 at all positions along the proximal-to-distal axis ( Figure 2C, D, E). Additionally, the axial gradient in 97 bound (NADPH/NADH ratio) ( Figure 2C) in SCs but not HCs is consistent with our results from whole 98 epithelial measurements at E6 and E14 ( Figure 1E-J), showing the same metabolic differences as a 99 function of tonotopic position along the cochlea. 100 101 Differential expression of mRNAs regulating NADPH production along the tonotopic axis. 102 Having identified a gradient in the cellular balance of NADPH/NADH along the BP at E6, we sought to 103 investigate the metabolic pathways underlying the differences in bound and bound. By exploiting our 104 previously generated bulk RNA-seq and Affymetrix microarray data sets 18 we measured differential 105 expression of metabolic mRNAs along the BP at E6.5 and E14. In these experiments, BPs had been 106 separated, for bulk RNA-seq and Affymetrix microarray analysis, into proximal, middle, and distal 107 thirds prior to isolation of mRNA 18 . Microarray data were analysed to identify transcripts with 108 expression levels at least two-fold higher in the proximal compared to distal half of the BP. We 109 identified multiple genes with differential expression between proximal and distal regions that are 110 known to regulate NADPH-producing pathways ( Figure 3). The largest contributor of cytosolic NADPH 111 is the oxidative branch of the pentose phosphate pathway (PPP) 23 . NADPH can also be generated as 112 a result of mitochondrial tricarboxylic acid (TCA) cycle activity, where levels are maintained through 113 the activity of multiple enzymes 23,24 . Although the specific mRNAs expressed at E6.5 and E14 differ, 114 at both stages metabolic genes showing the highest differential expression between proximal and 115 distal regions were those associated either with diversion of glucose into the PPP or towards aerobic 116 respiration in mitochondria ( Figure 3A, B). Together, these expression data reflect differences in 117 NADPH-linked metabolism along the BP. Higher expression of genes regulating NADPH production in 118 the proximal BP ( Figure 3) is also consistent with the higher bound reported for this region (Figure 1). 119 120 Tonotopic expression of metabolic effector proteins. 121 Modulation of metabolic pathways is driven predominantly through post-translational modifications 122 of metabolic effector proteins 25 . Isocitrate dehydrogenase 3 (IDH3) and Lactate dehydrogenase (LDH) 123 have well-defined roles in regulating the switch between glycolytic and oxidative metabolism during 124 Warburg-like reprogramming 25-28 . They also regulate metabolic flux in the mitochondrial TCA cycle 125 and cytosolic glycolysis respectively ( Figure 4A). Therefore, we analysed the expression of IDH3 126 subunit A and LDH subunit A along the tonotopic axis of the developing BP. There was no difference 127 in IDH3A expression between proximal and distal regions at early stages. However, between E14 and 128 E16, the expression gradient of IDH3A reversed along the tonotopic axis ( Figure 4B, C, supplementary 129 Figure 1A). Downregulation of IDH3A has been linked specifically with a switch from oxidative to 130 glycolytic pathways by reducing glucose flux in the mitochondrial TCA cycle 28 . The distal-to-proximal 131 gradient in IDH3a expression at E14 ( Figure 4B, D, supplementary Figure 1A) is consistent with a higher 132 cytosolic glucose flux in the low frequency region of the BP. Alongside the higher NADPH/NADH ratio 133 reported for the proximal BP ( Figure 1) these findings are consistent with greater PPP activity in the 134 low frequency region and progressively more oxidative metabolic phenotypes towards the distal 135 region. The reversal in the IDH3A gradient at E16 suggesting a shift in metabolic function along the 136 tonotopic axis as the system matures. Studies in other systems have shown that IDH3A regulates 137 4 synaptic transmission by controlling vesicle fusion and recycling. Downregulation of IDH3A expression 138 also phenocopies loss of the synaptic Ca 2+ sensor synaptotagmin1 29 . The opposing gradient in IDH3A 139 expression at E16 ( Figure 4C, D, supplementary Figure 1A) may therefore coincide with the functional 140 refinement of hair cell synapses along the BP at this time. No difference in LDHA expression was seen  141  between proximal and distal regions at any developmental stage. Instead, LDHA expression switched  142 from SCs to HCs between E7 and E10 and was restricted to SCs again between E10 and E14 143 (supplementary Figure 2). The mechanisms regulating these cell-specific changes in LDHA expression 144 are not known. The timing of these switches does however overlap with important stages in HC 145 development. It will therefore be of interest to investigate potential crosstalk between LDHA-linked 146 metabolism and the developmental pathways that instruct HC and SC specification. 147 148 A gradient in glucose uptake is established along the tonotopic axis during BP development.

149
To further characterise differences in glucose metabolism along the tonotopic axis, we used the 150 fluorescent D-glucose derivative 2-[N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)-amino]-2-deoxy-D-glucose 151 (2-NBDG) 30 and the potentiometric fluorescent dye tetramethyl-rhodamine-methyl-ester (TMRM). 152 The cellular fluorescence after incubation with 2-NBDG for 2 hours was used as an indicator of glucose 153 uptake. TMRM was used to report mitochondrial membrane potential (mt), itself a read-out of 154 glycolytically-derived pyruvate oxidation in the mitochondrial tricarboxylic acid (TCA) cycle. Explants 155 were dual-loaded with TMRM and 2-NBDG and fluorescence from both dyes was analysed from single 156 cells in live BP explants between E7 and E16 ( Figure 4E-J, supplementary Figure 1B, C). We observed 157 no tonotopic difference in glucose uptake during early development or between HC and SC subtypes. 158 A significant gradient in glucose uptake was evident however along the proximal-to-distal axis at E14 159 ( Figure 4F). There was no significant difference in mt between proximal and distal end at any 160 developmental stage, indicating uniform mitochondrial activity along the BP. Despite no differences 161 along the tonotopic axis, mt did change significantly (p < 0.05 Student's t-test) with developmental 162 age ( Figure 4H-J, supplementary Figure 1C) reflecting a reprogramming of mitochondrial metabolism 163 during HC formation at a stage when cells are becoming post mitotic 31 . The higher mt observed in 164 HCs compared to SCs from E9 may reflect a unique metabolic state that develops in association with 165 commitment to different cell lineages ( Figure 4J). These findings also confirm that the gradient in 166 NADPH/NADH ( Figure 1D-I) along the BP originates specifically from tonotopic differences in cytosolic 167 and not mitochondrial metabolism. 168 169 Cytosolic glucose metabolism is necessary for tonotopic patterning in the chick cochlea. 170 Having identified differential NADPH-linked glucose metabolism along the proximal-to-distal axis, we 171 next investigated a functional role for this gradient in the establishment of tonotopy. phenotypes at different positions along the tonotopic axis (Figure 5 B, C). In control cultures, HCs 180 developed with the normal tonotopic gradient in morphologies (luminal surface area and density) 181 ( Figure 5A, C). In contrast, when glycolysis was blocked between E6.5 and E13.5 equivalent, tonotopic 182 patterning was not just abolished, HC phenotypes appeared flattened and uniformly more distal-like 183 along the BP. To further confirm a role for glycolysis in establishing positional identity, we employed 184 a second method to modulate glucose metabolism. Studies in other systems have shown that raising 185 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted April 12, 2022. ; https://doi.org/10.1101/2022.04.11.487851 doi: bioRxiv preprint cytosolic levels of the metabolite s-adenosyl methionine (SAM) blocks glycolysis independently of HK 186 activity 33 . As seen in treatments with 2-DOG, tonotopic patterning was flattened in explants cultured 187 with medium containing 50 M SAM (supplementary Figure 3). 188 189 A distal-to-proximal gradient of Bmp7 activity is known to establish tonotopy in the BP between E6 190 and E8 18 . To determine whether glucose metabolism acts during this same developmental window, 191 we treated explants with 2-DOG for defined periods during HC formation. BPs were established at E6.5 192 and treated with 2-DOG for either 24 or 48 hours followed by wash out with control medium. The 193 tonotopic gradient in HC size developed normally in BPs treated with 2-DOG for 24 hours followed by 194 wash-out but was absent in those treated for 48 hours ( Figure 5C). . We therefore sought to further dissect the glycolytic signalling network during specification of 203 tonotopy in the developing BP. To investigate a role for PPP-linked glucose metabolism, BP explants 204 were established as described, and treated with 50 M 6 Aminonicotinamide (6-AN) between E6.5 205 and E13.5 equivalent. Treatment with 6-AN inhibits the rate-limiting PPP enzyme glucose-6-phosphate 206 dehydrogenase (G6PD) ( Figure 6C). By comparison with control cultures, inhibition of PPP metabolism 207 caused a significant decrease in hair cell size within 2500 μm 2 areas measured in the proximal BP 208 ( Figure 6). To determine whether this effect was specific to glucose flux through the PPP we also 209 blocked phosphofrucktokinase (PFK), a rate limiting enzyme further down in the glycolytic pathway, 210 using 1M YZ9 (supplementary Figure 4). Blocking PFK activity inhibits the glycolytic cascade involved 211 in pyruvate production but does not change the activity of G6PD in the PPP 8 . Although we observed 212 a reduction in HC size in both BP regions, inhibition of PFK between E6.5 and E13.5 did not alter the 213 intrinsic tonotopic patterning in HCs along the axis of the epithelium (p = 0.01 Student's t-test proximal 214 verses distal HC size) (supplementary Figure 4). These findings suggest that tonotopic patterning and 215 specification of HC positional identity are regulated by metabolic glucose flux occurring upstream of 216 PFK. 217 Opposing gradients of glucose metabolism and mitochondrial OXPHOS have been described in 218 developing chick embryos, where they regulate elongation of the body axis and specification of cell 219 fates 9 . To investigate a role for the mitochondrial TCA cycle and thus OXPHOS along the tonotopic 220 axis, we blocked the mitochondrial pyruvate carrier using UK5099 (supplementary Figure 5B). 35 221 Preventing pyruvate uptake into mitochondria inhibits the TCA cycle and, as a result, impairs OXPHOS 222 (supplementary Figure 5B). HCs in explants treated with UK5099 between E6.5 and E13.5 did not 223 develop with normal gradients in HC size along the BP and had either immature stereocilial bundles 224 or lacked them completely in both proximal and distal regions (supplementary Figure 5A, C, red 225 arrows). To determine whether this effect was due to an overall arrest in HC development, explants 226 were established at E8.5, by which time positional identity is specified but bundles are not yet 227 developed and maintained for 7 days in vitro to the equivalent of E15.5. Compared to control explants, 228 those treated with UK5099 displayed no tonotopic gradient in cell size and HCs at all positions along 229 the BP showed immature bundle morphologies (supplementary Figure 5C). Together, these findings 230 suggest that mitochondrial TCA cycle activity and OXPHOS are necessary for the overall progression 231 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted April 12, 2022. ; https://doi.org/10.1101/2022.04.11.487851 doi: bioRxiv preprint of HC maturation. The role of these pathways in establishing tonotopy is at present unclear and 232 requires further investigation. 233

Glucose metabolism regulates tonotopic expression of Bmp7 and Chdl-1 234
In many developing systems, gradients of one or more morphogen act to regulate cell fate, growth 235 and patterning along a given axis 36,37 . In the chick cochlea, reciprocal gradients of Bmp7 and its 236 antagonist Chdl-1 play key roles in determining positional identity. As disruption of both the normal 237 gradient in Bmp7 activity 18 and glucose metabolism induce similar effects on tonotopic patterning 238 ( Figure 7A-C), we sought to investigate potential crosstalk between metabolism and known 239 developmental signalling pathways that set up tonotopy. We investigated the regulatory effects of 240 glucose metabolism on the expression gradients of Bmp7 and Chdl-1 along the developing BP. Explants 241 were established at E6.5 and maintained for 72 hours in vitro (equivalent of E9.5) in control medium 242 or that containing 2mM 2-DOG + 5mM sodium pyruvate. Disrupting glucose metabolism along the BP 243 altered the normal expression gradients in both Bmp7 and Chdl-1 ( Figure 7E). Specifically, following 244 treatment with 2-DOG, the Bmp7 expression increased into the proximal BP region and expression of 245 Chdl-1 decreased along the entire length of the organ. Equalised Bmp7 expression following treatment 246 with 2-DOG caused expansion of HC phenotypes that more closely resembled distal differentiation 247 into the proximal BP. Taken together, our findings indicate a causal link between glycolysis and Bmp7 248 signalling in specifying HC identity along the tonotopic axis. 249 250 Changes in glucose metabolism do not alter proliferation in the BP 251 Changes in glucose metabolism have been linked with reduced cellular proliferation in other systems 252 38 . To determine whether the change in HC morphology observed in 2-DOG treated explants was a 253 result of altered proliferation, the mitotic tracer EdU was added to cultures with 2-DOG for 72 hours. 254 We observed no differences in proliferation in proximal or distal regions following treatment with 2-255 DOG + 5mM NaP (supplementary Figure 6).

257
A role for NADPH-linked metabolism in the developing mouse cochlea 258 Unlike in the avian auditory system, HCs in the mammalian cochlea cannot spontaneously regenerate 259 after damage 39 . Although we can generate new HCs using organoid models that recapitulate aspects 260 of early inner development, these cells are not functionally viable long-term and do not display all 261 features of mature HCs 40 . Identifying novel factors and signalling pathways that regulate the 262 specification and formation of HCs and SCs in the mammalian auditory system is therefore of 263 significant interest. Metabolism has not been explored in developing HCs and SCs of the mammalian 264 cochlea. We do know, however, that in the adult cochlea HCs and SCs have distinct metabolic states 265 linked to their function 12 . We therefore characterised the metabolic states of HCs and SCs in the 266 mammalian cochlea between E14 and P0, the developmental window for cell fate specification and 267 functional refinement. As observed in the chick cochlea, there was no difference in the NAD(P)H FLIM 268 signal between HCs and SCs at early developmental stages. At later time points however, there was a 269 significant (p < 0.05, Student's t-test) increase in bound and bound between E14 and E16 indicating a To investigate the biochemistry underlying metabolic differences between HC and SC sub-types, we 275 interrogated existing single cell datasets derived across 3 developmental time points from the mouse 276 cochlea 42 . HCs were identified using expression of Pou4f3 and lack of Sox2 42,43 . Conversely, SCs were 277 identified using expression of Sox2 42,44 . We then performed a comparative analysis using the Seurat 278 (v4) package, in R 45,46 . As discussed, an increased bound reflects a higher NADPH/NADH ratio, 279 regulated mainly by the PPP activity in the cytosol. We therefore investigated expression of genes 280 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted April 12, 2022. ; https://doi.org/10.1101/2022.04.11.487851 doi: bioRxiv preprint encoding enzymes that regulate PPP metabolism. We found that PPP pathway regulatory enzymes are 281 more highly expressed in SCs compared to HCs at all developmental time points (supplementary Figure  282 7G). We observed developmental changes in the expression of regulatory enzymes of NADPH 283 metabolism. Increased expression in cytosolic Nadk, (generates NADP + from NAD), in HCs between 284 E14 and E16 (supplementary Figure 7G) likely corresponds to the higher anabolic activity associated 285 with differentiation along the cochlea at this time and thus the greater requirement for PPP-derived 286 NADPH 47 . Additionally at the same time points, we detected a pronounced upregulation in Pgls, a 287 major regulator of the PPP and of ribose-5-phosphate production 48 . Heightened availability of this 288 enzyme would provide the necessary biosynthetic precursors for nucleic acid synthesis during cell 289 differentiation 48 . We also investigated differences in the expression of genes encoding regulators of 290 mitochondrial NADPH production (supplementary Figure 7H). We observed that expression of 291 mitochondrial NADPH-producing enzymes is similar between E14 and E16 in both HCs and SCs but 292 becomes restricted to HCs by P1. This likely reflects progressive commitment along the HC lineage, 293 which we know is associated with enhanced mitochondrial activity 42 . 294 295 Discussion 296 To successfully address hearing loss and balance defects, we must be able to generate new, 297 functionally viable HCs following damage, aging or ototoxic insult. As certain types of HC are more 298 vulnerable to insult than others, it is essential we understand the specific factors that specify different 299 HC subtypes such as high and low frequency cells. Generating new HCs that recapitulate features of 300 those in a healthy cochlea requires a detailed knowledge their formation. Additionally, we must 301 resolve how they navigate the path of maturation and survival within their surrounding niche along 302 the cochlea. Previous studies have focused largely on understanding the role of transcription factors 303 in HC formation and how manipulation of these in the adult inner ear can be exploited to promote 304 regeneration 43,49-52 . Taking a novel approach, we here explored the coordinated regulation of complex 305 signalling between developmental pathways and metabolism in differentiating HCs and SCs of the 306 inner ear. 307 308 Using FLIM imaging of cellular NAD(P)H, we observed a proximal-to-distal gradient in glucose 309 metabolism in the developing BP resulting primarily from tonotopic differences in PPP activity. 310 Analysis of bulk RNA-seq data also identified differentially expressed mRNAs encoding enzymes that 311 divert glucose away from glycolysis and towards the PPP (GOT2 53 , PKM2 54 , PFKFB4 55,56 ) along the 312 tonotopic axis of the developing BP. High levels of NADPH generated through PPP-linked metabolism 313 have been reported in proliferating cells and during differentiation in a number of systems 57 . Graded 314 differences in cell size along the cochlea are important or correct for tonotopy and hearing. Cell size 315 is determined by two opposing processes, growth and division 58 and PPP activity is known to regulate 316 the signalling pathways that coordinate these 59 . The mechanisms linking cell growth, cell cycle 317 progression and cell size regulation are poorly understood in intact tissues such as the cochlea. In the 318 BP, cell cycle exit is initiated in the distal region at embryonic day 5 (E5). Terminal mitosis then 319 progresses in a wave from distal-to-proximal regions and across the neural-to-abneural axis until E9, 320 when the majority of cells are postmitotic 31 . As the critical size for cell cycle progression is determined 321 by environmental factors, one hypothesis is that by differentially regulating growth along the BP and 322 thus timing of cell cycle exit, the gradient in glycolysis could instruct tonotopic differences in HC cell 323 size. Classically, by diverting glucose towards the PPP, nascent cells would be exposed to lower levels 324 of mitochondrially-derived reactive oxygen species (ROS) thus protecting newly synthesised DNA from 325 ROS-induced damage during cell division. Enhanced PPP activity and increased NADPH levels would 326 therefore poise cells for the increased anabolic activity associated with differentiation 57 . At E9, we 327 also see a significant increase in mitochondrial metabolism reported by a more polarised membrane 328 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is therefore be a consequence of increased Shh activity, which would lead to increased Bmp7 signalling. 355 356 It will be of interest to investigate a role for glucose metabolism in regulating the Shh-Bmp7-Chdl-1 357 signalling network along the proximal-to-distal axis in the chick cochlea and Shh signalling along the 358 basal-to-apical axis in the mouse. Studies in other systems have also identified a role for Shh in 359 regulating glucose metabolism through modulating activity of glycolytic enzymes, 64 most significantly, 360 members of the bifunctional 6-phosphofructo-2-kinase/fructose-2, 6-bisphosphatase (PFKFB 1-4) 361 family [64][65][66] . The activity of these enzymes is bi-directional, highly regulated at the post-translational 362 level and cell and/or tissue-specific 56 . We show PFKFB4 expression in a proximal-to-distal gradient 363 along the BP at E6.5. A regulatory signalling network between graded Shh, Bmp7 and PFKFB (1-4)-364 regulated glycolysis has not been explored in any developing tissue but would provide further insight 365 as to how metabolism might fine tune HC phenotypes along the tonotopic axis. Untangling the precise 366 interactions between components of the Sonic Hedgehog-Bmp7-Retinoic acid and glycolytic signalling 367 networks will further our knowledge of how hair cells are tuned during development. From what we 368 understand about frequency selectivity in vertebrates 67 , recapitulation of tonotopy will also require 369 that any gradients and the signalling networks they instruct, scale correctly in different inner ear 370 sensory patches and across species with varying head size and cochlear lengths. It will be important 371 to understand how the molecular mechanisms that link metabolic and developmental pathways 372 during HC scale within different inner ear sensory patches and across species. 373 374 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted April 12, 2022. ; https://doi.org/10.1101/2022.04.11.487851 doi: bioRxiv preprint Developmental programming of energy balance within HC and SC lineages is an important 375 physiological process allowing cells arising from a common progenitor population to differentiate with 376 unique metabolic states. Extensive changes in cellular redox state occur during the first postnatal week 377 of functional refinement in cells of the mouse cochlea 12,13 . These changes indicate local metabolic 378 switches during the early phases of HC and SC functional maturation 12 . We report cell-specific 379 differences in metabolic enzyme expression, NAD(P)H FLIM signatures and mitochondrial activity in 380 the cochlear epithelium from as early as E14 in mouse and E9 in chick. These cell-specific differences 381 may reflect the developmental origins of metabolic coupling between HCs and SCs. Such cooperation 382 is observed between neurons and astrocytes in the developing brain 68 . Neurons and astrocytes 383 preferentially use different metabolic pathways under physiological conditions, due in part, to cell 384 specific expression patterns of genes regulating energy metabolism 72 . Metabolic coupling is essential 385 for normal brain function and has also been linked with long-term function and survival of cochlear 386 hair cells 69 and E16, HCs were manually segmented. HC labels were dilated by 3 μm, which provided selections 496 which included both hair cells and their surrounding SCs. By subtracting the hair cell segmentation 497 from the dilated label, we were thus able to measure the fluorescence intensity of whole HCs 498 separately from their surrounding support cells in the 2-NBDG and LDHA data. A similar approach was 499 adopted when measuring TMRM and IDH3A fluorescence intensity at E14 and E16. However, we 500 noticed that signal was concentrated around the HC periphery. In order to ensure that the 501 fluorescence intensity best reflected only the mitochondria and was not reduced by the low 502 fluorescence from the centre of each HC, we measured mean fluorescence intensity only up to 2 μm 503 from the cell membrane. Likewise, for TMRM and IDH3A data at E7 and E9, mitochondria were 504 segmented using Fiji's auto-local thresholding (Niblack) prior to intensity measurements, to avoid a 505 biased estimate of fluorescence intensity due to empty space surrounding each mitochondrion. 506

Analysis of hair cell morphology 507
Data were analysed offline using image J software. Hair cell luminal surface area and cell size were 508 used as indices for HC morphology along the tonotopic axis. To determine the hair cell density, the 509 luminal surfaces of hair cells and cell size, cultures were labelled with phalloidin and DAPI. Then, the 510 number of hair cells in 50 μm × 50 μm regions of interest (2,500 μm 2 total area) located in the proximal 511 and distal BP regions were determined. Proximal and distal regions were determined based a 512 calculation of the entire length of the BP or explant. In addition, counting ROIs were placed in the mid-513 region of the BP along the neural to abneural axis to avoid any confounding measurements due to 514 radial differences between tall and short hair cells. For each sample, hair cells were counted in four 515 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is  (1 − bound ) free + bound free ). A proximal-to-distal gradient is evident in τbound but not αbound 797 throughout BP development. White asterisks indicate the HCs. Higher magnification images 798 highlight the differences in mean lifetime between proximal and distal HCs at E14.

799
(I) Quantification of τbound during development shows a shift from NADPH to NADH producing 800 pathways. Line graphs highlight differences in τbound between proximal (black) and distal (grey) 801 BP regions throughout development.

802
(J) αbound, reflecting the cellular balance between free and enzyme-bound NAD(P)H, increases 803 significantly in both proximal (black) and distal (grey) BP regions between E6 and E14 and 804 decreases significantly between E14 and E16. Line graphs show differences in αbound in the 805 proximal (black) and distal (grey) regions. 806 The difference in mean NAD(P)H lifetime result from a shift in the cellular balance of 807 NADPH/NADH, and thus in utilisation of the metabolic pathways that produce these cofactors, 808 from high to low frequency regions. Scale bars = 50m. Data are mean  SEM; E6: n = 6, E9: 809 n= 4, E14: n = 7 and E16: n = 6 biological replicates. * p < 0.05 Student's t-test. 810 811 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted April 12, 2022. Log2 <1 are expressed significantly in the proximal BP region. Statistical significance levels 856 were calculated by one-way ANOVA. For a gene to be considered 'differential', at least one 857 region of the BP (proximal, middle or distal) was required to be ≥0.5 RPKM. A fold change of 858 ≥ 2 was imposed for the comparison between distal and proximal regions. A final requirement 859 was that middle samples had to exhibit RPKM values mid-way between proximal and distal 860 regions to selectively capture transcripts with a gradient between the two ends (REF  to E13.5) in either control medium or medium supplemented with 2mM 2-DOG + 5 mM Sodium 911 Pyruvate (NaP). Phalloidin staining depicts differences in hair cell morphology between 912 proximal and distal regions and DAPI indicates the gradient in hair cell size.

913
(C) Quantification of these differences. Hair cell luminal surface area was measured in 2500 914 m 2 areas in the proximal (black bars) and distal (grey bars) BP regions for all culture 915 conditions. In controls, mean hair cell luminal surface decreases progressively from the 916 proximal-to-distal region. This gradient is abolished if glycolysis is blocked with 2-DOG 917 between E6.5 and E13.5. 2-DOG caused a significant decrease in hair cell size in the proximal 918 but not distal region.  Explants were established at E6.5 and incubated for 7 days in vitro in control medium or 979 medium containing 2-DOG + NaP or Bmp7.

980
(B-C) Treatment with 2-DOG or Bmp7 induced hair cell morphologies consistent with a more 981 distal phenotype in the proximal BP. Hair cell luminal surface area was determined using 982 Phalloidin staining at the cuticular plate in 2500 2 m areas at both proximal and distal ends. (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted April 12, 2022. ; https://doi.org/10.1101/2022.04.11.487851 doi: bioRxiv preprint . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted April 12, 2022. switching between glycolytic and oxidative metabolism during cochlear development. Insets 1207 highlight differences in bound and thus metabolism between HCs (black bars) and SCs (grey 1208 bars) at E16 but not E14. White arrows indicate the SCs and white circles outline the HCs.

1209
(E) bound increases in both HCs (black bars) and SCs (grey bars) between E14 and E16 and 1210 decrease significantly in both cell types between E16 and P0. The change in bound indicates a 1211 shift in the cellular NADPH/NADH balance as a function of development. At E16, bound is 1212 significantly higher in SCs (grey bars) compared to HCs (black bars) showing metabolic 1213 divergence between cell types as a function of development. 1214 . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted April 12, 2022. ; https://doi.org/10.1101/2022.04.11.487851 doi: bioRxiv preprint (F) bound, reflects the proportion of cellular NAD(P)H that is bound to either enzymes or 1215 cofactors and the balance between glycolysis and OXPHOS. bound increases significantly in 1216 both hair cells and supporting cells between E14 and E16 but decreases significantly in 1217 supporting cells only between E16 and P0. This indicates a reprogramming of metabolic 1218 pathways both developmentally and between cell types. Data are mean  SEM. * p = < 0.05 1219 Student's paired t-test. E14: n = 3, E16: n = 5 and P0, n = 5 biological replicates. Scale bars 1220 are 50 m. . CC-BY 4.0 International license made available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprint this version posted April 12, 2022. ; https://doi.org/10.1101/2022.04.11.487851 doi: bioRxiv preprint