1

Glutamine metabolism controls stem cell fate reversibility and long-term maintenance in the hair

follicle

Christine S. Kim1,2#, Xiaolei Ding3#, Kira Allmeroth1,2§, Leah C. Biggs4,5,6§, Olivia I. Kolenc7,

Nina L’Hoest1,2, Carlos Andrés Chacón-Martínez1,2, Christian Edlich-Muth1, Patrick

Giavalisco1, Kyle P. Quinn7, Martin S. Denzel1,2, Sabine A. Eming2,3,8,9* and Sara A.

Wickström1,2,4,5,6*

1Max Planck Institute for Biology of Ageing, Cologne, Germany 2 Cluster of Excellence Cellular Stress Responses in Ageing-associated Diseases (CECAD), University of Cologne, Cologne, Germany 3Department of Dermatology, University of Cologne, Cologne, Germany 4Helsinki Institute of Life Science, Biomedicum Helsinki, University of Helsinki, Helsinki, Finland 5Wihuri Research Institute, Biomedicum Helsinki, University of Helsinki, Helsinki, Finland 6 Stem Cells and Metabolism Research Program, Faculty of Medicine, University of Helsinki, Helsinki, Finland 7Department of Biomedical Engineering, University of Arkansas, Fayetteville, AR, USA 8Center for Molecular Medicine Cologne (CMMC), University of Cologne, Cologne, Germany 9Institute of Zoology, Developmental Biology Unit, University of Cologne, Cologne, Germany

# These authors contributed equally § These authors contributed equally

*To whom correspondence should be addressed: sara.wickstrom@helsinki.fi

sabine.eming@uni-koeln.de

Lead contact: sara.wickstrom@helsinki.fi

Manuscript

2

Summary

Stem cells reside in specialized niches that are critical for their function. Upon activation hair

follicle stem cells (HFSCs) exit their niche to generate the outer root sheath (ORS), but a

subset of ORS progeny returns to the niche to resume a SC state. Mechanisms of this fate

reversibility are unclear. We show that the ability of ORS cells to return to the SC state

requires suppression of a metabolic switch from glycolysis to oxidative phosphorylation and

glutamine metabolism that occurs during early HFSC lineage progression. HFSC fate

reversibility and glutamine metabolism are regulated by the mammalian target of rapamycin

complex 2 (mTORC2)-Akt signaling axis within the niche. Deletion of mTORC2 results in a

failure to re-establish the HFSC niche, defective hair follicle regeneration, and compromised

long-term maintenance of HFSCs. These findings highlight the importance of spatiotemporal

control of SC metabolic states in organ homeostasis.

Introduction

A defining property of stem cells (SCs) is their capacity for self-renewal and multilineage

differentiation. Cell-intrinsic networks cooperate with signals from the microenvironment to

fine-tune these two key properties of SCs to maintain tissue homeostasis (Blanpain and

Fuchs, 2014; Chacón-Martínez et al., 2018). Hair follicles are one of the few tissue

compartments that undergo natural regeneration in mammals. Hair follicles regenerate

through cyclical bouts of rest (telogen), growth (anagen), and degeneration (catagen). At the

start of the hair cycle, quiescent HFSCs residing in the so-called bulge niche are activated

to give rise to progenitor cells (also known as transit amplifying cells) which subsequently

differentiate into outer root sheath (ORS) cells to supply the cells needed for hair follicle

down-growth. These ORS cells have exited the bulge niche and lost the expression of the

key bulge HFSC marker CD34, but still express SOX9 and continue to proliferate during

early- to mid-anagen (Hsu et al., 2011). After the completion of the growth cycle, a subset

of ORS cells returns to the bulge niche, resume quiescence and CD34 expression, and

remain in this state until the next hair cycle (Hsu et al., 2011; Lay et al., 2016). The

establishment of this new quiescent HFSC state is associated with the generation of a new

bulge niche adjacent to the old one (Gonzales and Fuchs, 2017; Muller-Rover et al., 2001).

Thus, HFSC activity is tightly controlled in time and space, making the hair follicle an ideal

system to investigate how SC quiescence, activation, and differentiation are regulated, and

what are the factors that determine whether a progenitor/transit-amplifying cell will return to

the quiescent SC state or fully commits to terminal differentiation.

3

Differentiation of embryonic and hematopoietic SCs associates with a change in the

carbohydrate metabolic program, characterized by a transition from glycolytic metabolism

to oxidative phosphorylation (OXPHOS) (Ito and Ito, 2016; Shyh-Chang et al., 2013).

Generation of energy by anaerobic glycolysis is inefficient, yielding only 2 ATP molecules

per glucose molecule in contrast to the 36 generated by OXPHOS, but superoxide anions

are produced in this process, generating mitochondrial reactive oxygen species (ROS).

Excessive production of ROS can abrogate various SC properties including quiescence and

self-renewal, which is a major reason why many SCs prefer glycolysis as their energy source

(Ito and Ito, 2016).

In addition to ATP production, SCs utilize glucose-derived carbons for lipid synthesis and

for production of hexosamines, ribose, glycerol, serine and glycine. Excess glucose-derived

carbons can be secreted as lactate. Differentiation, on the other hand, is generally

associated with proliferation and increased metabolic requirements, explaining the need for

OXPHOS at this stage to maximize ATP production (Kobayashi and Suda, 2012; Miyamoto

et al., 2007). To increase the flux of the cycle for biosynthesis, differentiating SCs and cancer

cells have also been shown to utilize additional carbon sources, such as glutamine, which

can be converted into both Acetyl-CoA and TCA cycle intermediates (Jin et al., 2015;

Oburoglu et al., 2014). In contrast to the role of metabolic changes in SC differentiation and

cancer, the role of metabolism in SC fate reversibility and reprogramming is largely unknown.

An important master regulator of metabolism is the mammalian target of rapamycin (mTOR)

pathway. mTOR is a serine/threonine protein kinase that integrates growth factor receptor

signaling with cell growth, proliferation and survival through two distinct multi-protein

complexes, referred to as mTOR complex 1 and 2 (mTORC1 and mTORC2). Both

complexes contain mTOR, mLST8 and Deptor. Binding of mTOR to Raptor defines the

mTORC1 complex, whereas Rictor binding to mTOR generates the mTORC2 complex.

mTORC1 regulates protein translation and anabolic metabolism in response to amino acid

and nutrient levels and is an important regulator of cell growth (Albert and Hall, 2015;

Eltschinger and Loewith, 2016; Laplante and Sabatini, 2012). mTORC2 acts downstream of

growth factor signaling to activate Akt, Pkcα, and Sgk1 (Sarbassov et al., 2005), but its

functions in SC regulation and metabolism are less well understood.

4

Here we address the molecular mechanisms that define the tipping point between SC fate

reversibility and commitment to terminal differentiation. We show that HFSC lineage

progression towards the ORS is controlled by a metabolic switch to increased OXPHOS and

glutaminolysis, and the ability of ORS progenitors to return to the SC state depends on their

ability to suppress this metabolic switch. The metabolic reprogramming is regulated by

TORC2-Akt signaling-mediated suppression of glutaminase expression and glutamine

metabolism. Consequently, deletion of mTORC2 in the epidermis results in loss of HFSC

fate reversibility, defective hair follicle regeneration, and impaired long-term maintenance of

HFSCs. Collectively, our work demonstrates the remarkable metabolic flexibility of HFSCs,

which is required to maintain a stable SC population throughout the lifespan of mice.

Results

HFSCs and early progenitor cells have distinct metabolic profiles

To gain insight into the molecular mechanisms that govern HFSC fate, we utilized an ex vivo

skin organoid culture system (3C culture) that specifically promotes growth of CD34-positive

and CD34-negative cells, which based on transcriptome and marker expression represent

a mixture of HFSCs, hair follicle ORS progenitors and inner bulge cells (Chacón-Martínez

et al., 2016) (Fig. 1A; Supplementary Fig. S1A). Both ORS progenitors and the inner layer

of bulge cells (CD34-negative, Keratin 6-positive) represent HFSC progeny and act as niche

cells for bulge HFSCs (Hsu et al., 2014). Specifically, most of these CD34-negative cells in

3C cultures expressed the HFSC and ORS progenitor marker and master transcriptional

regulator Sox9, and ORS identity gene Barx2, whereas interfollicular epidermis markers

(Joost et al., 2020) were not enriched (Supplementary Fig. S1A, B). A subset of these CD34-

negative cells was also positive for the inner bulge marker Keratin 6 and the hair follicle-

specific marker Keratin 16 (Supplementary Fig. S1A) (Hsu et al., 2011; Nowak et al., 2008).

As Keratin 6 and 16 can also be induced upon wound healing (Joost et al., 2018), we

confirmed that no enrichment of other wound healing marker genes was observed in the

CD34-negative population (Supplementary Fig. S1C), indicating that Keratin 16 and 6

represented HF identity markers of the CD34-negative population. As expected based on

the HFSC-like transcriptional profile, the cultured and subsequently fluorescence-activated

cell sorting (FACS)-purified CD34+/6 integrin+ cells displayed higher proliferative potential

in a colony-forming assay than the cultured CD34-/6 integrin+ cells (Supplementary Fig.

S1D).

5

We have previously shown that these cultured CD34+/6 integrin+ HSFCs and CD34-/6

integrin+ ORS progenitors (from here on referred to as 3C-HFSCs and 3C-ORS progenitors,

to distinguish them from in vivo HFSC and ORS cells) co-exist in a 50:50 population

equilibrium driven by bidirectional, dynamic interconversion of the two states, thus providing

a perfect model system to study SC fate reversibility (Chacón-Martínez et al., 2016). By

analyzing the transcriptomes of FACS-purified 3C-CD34+/6 integrin+ and 3C-CD34-/6

integrin+ cells (Chacón-Martínez et al., 2016), we identified genes involved in mitochondrial

ATP synthesis and OXPHOS to be most significantly overrepresented in the CD34-negative

3C-ORS progenitor population (Fig. 1A, B). To investigate if these differential gene

expression patterns would reflect differences in the levels of mitochondrial respiration in the

two populations, we quantified the amounts of mitochondria in CD34-positive HFSCs and

CD34-negative progenitors using transgenic mice expressing a fluorescent marker for

mitochondria (mito-YFP; (Sterky et al., 2011), as well as by analyzing expression levels of

genes involved in mitochondrial biogenesis (Scarpulla et al., 2012). FACS analyses of both

freshly isolated in vivo HFSCs and 3C-HFSCs showed significantly less fluorescently

labeled mitochondria as well as lower expression levels of mitochondrial biogenesis genes

in HFSCs compared to progenitors (Fig. 1 C; Supplementary Fig. S1E). To establish if the

higher amounts of mitochondria in CD34-negative progenitors would reflect their increased

OXPHOS activity, we performed Seahorse metabolic flux measurements of OXPHOS and

glycolysis. In agreement with their greater mitochondrial mass, the flux analyses revealed

that 3C-progenitors had significantly higher rates of basal respiration as well as higher

respiratory capacity compared to 3C-HFSCs (Fig. 1D, E). In contrast, no major differences

were observed in glycolytic activity (Fig. 1F).

The activity of the TCA cycle is not only required for electron transport chain (ETC) function,

but it can also be used to provide carbon for biosynthetic reactions. As OXPHOS but not

aerobic glycolysis was increased in progenitors, we sought to analyze the status of the TCA

cycle by quantifying key metabolic intermediates using liquid chromatography mass

spectrometry (LC-MS). The analyses showed that metabolites from TCA cycle were

increased in in vivo CD34-/6+ progenitors (Fig. 1G). In contrast, metabolic intermediates

of glycolysis were not altered (Supplementary Fig 1F). Collectively these data indicate that

the progenitor state is associated with increased levels of OXPHOS and TCA cycle activity.

Anagen entry is associated with increased oxidation in upper hair follicle cells in vivo

6

Given that cell dissociation and FACS may cause metabolic changes in a metabolite-,

compartment-, or cell type-specific manner (Binek et al., 2019; Llufrio et al., 2018), we aimed

to directly quantify metabolic changes in HFSCs and their progeny during their activation

and subsequent lineage progression. For this, we utilized label-free multiphoton microscopy

that can be used to quantify autofluorescence from NADH and FAD and second harmonic

generation (SHG) from fibrillar collagen (Jones et al., 2018; Kolenc and Quinn, 2019). The

quantum yield of NADH and FAD autofluorescence is an order of magnitude lower than

typical exogenous fluorophores used to stain tissues. However, the relatively weak natural

fluorescence of NADH and FAD can be efficiently measured through multiphoton

microscopy to assess relative changes in metabolic activity (Kolenc and Quinn, 2019). An

optical redox ratio of [FAD/(NADH+FAD)] was computed from the NADH and FAD

fluorescence intensities within the HFSC bulge and ORS area in C57BL/6J mice at the

transition between postnatal day (P) 21 (telogen resting stage) and P23 (early anagen

growth phase). Decreases in the redox ratio of cells or tissues have been attributed to

increased glycolysis, whereas increases are associated with increased glutaminolysis and

OXPHOS (Georgakoudi and Quinn, 2012; Kolenc and Quinn, 2019; Varone et al., 2014). A

higher optical redox ratio was detected within the upper hair follicle bulge/ORS area in the

early anagen stage (P23) follicles relative to telogen stage (P21) hair follicles (Fig. 2A, B).

This significant increase in optical redox ratio over two days is indicative of increased

oxidation. The in vivo optical measurements during the phase of HFSC

activation/differentiation were consistent with the metabolic flux profiles of FACS-purified

progenitors compared to HFSCs (Fig. 1G) and were further corroborated by a measured

increase in the LC/MS-derived redox ratios of FAD/(NADH+FAD), ATP/ADP, ATP/AMP and

SAM/SAH in freshly isolated cells (Supplementary Fig. S2).

Low oxygen tension promotes the HFSC state

To understand if the metabolic switch towards increased OXPHOS upon anagen induction

would control HFSC fate, we probed if limiting pO2 would influence HFSC activation (i.e. the

switch from a quiescent to an actively cycling state) or lineage progression. To this end, we

performed 3C cultures in ambient air (20% O2) or in reduced oxygen (2% O2). FACS

analyses revealed that 14 d of 3C culture in 2% O2 significantly increased the proportion of

3C-CD34+/6+ cells within the cultures (Fig. 3A). qRT-PCR analysis of key HFSC fate

master regulators and identity genes showed that expression of these genes was

upregulated in the total cell population under reduced oxygen, whereas the expected

enrichment of these transcripts in FACS purified 3C-CD34-positive cells compared to 3C-

7

CD34-negative cells remained unchanged, confirming that the low pO2 -triggered increase

reflected an increase in 3C-HFSCs within the organoids (Fig. 3B, Supplementary Fig. S3A).

To test if the increased proportion of HFSCs in the cultures was caused by a pO2-dependent

activation of 3C-HFSCs, or alternatively a selective growth advantage of 3C-HFSCs

compared to progenitors in low pO2, we analyzed the growth and apoptosis rates of both

populations in 20% and 2% O2 but observed no differences (Supplementary Fig. S3B).

Absence of differential proliferation/survival suggested that 2% O2 did not control HFSC

activation, but rather could either prevent HFSC lineage progression to ORS or to promote

progenitor reversal to the HFSC state. To test this hypothesis, we performed lineage tracing

experiments where we FACS-purified CD34+/6+ 3C-HFSCs expressing GFP (LifeAct-GFP

(Riedl et al., 2010) with CD34-/6+ progenitors lacking GFP expression, mixed them 1:1 to

ensure comparable population composition of the cultures, and traced the conversion of 3C-

HFSCs to progenitors and vice versa (Fig. 3C). FACS analyses after 14 d of tracing revealed

that 2% O2 both enhanced return of progenitors to the HFSC state as well as reduced HFSC

lineage progression (Fig. 3D, Supplementary Fig. S3C).

HFSC lineage progression requires mitochondrial metabolism and glutaminolysis

To identify the molecular mechanisms by which low pO2 prevented HFSC lineage

progression and promoted progenitor return to the HFSC state, we examined the activation

of Hypoxia-inducible factor (Hif) pathway (Majmundar et al., 2010). To this end we subjected

3C organoid cultures to acute low pO2 (12 h) and analyzed Hif1 stabilization by western blot

and Hif1/2 target gene expression using qRT-PCR. As expected, Hif1 protein was

stabilized in 2% O2 (Supplementary Fig. S4A). Hif1/2 target gene analyses further

revealed that a subset of Hif targets (Vegf, Pgf, Glut1, Pai1) were mildly upregulated both in

3C-CD34-positive and 3C-CD34-negative cells, indicating that both populations respond to

low pO2 (Supplementary Fig. S4B). Intriguingly, however, the Hif1 target and regulator of

glycolysis Pdk2 was not induced, but it was expressed at higher levels in 3C-CD34-positive

cells compared to in 3C-CD34-negative cells, and the difference became slightly more

pronounced in low pO2. In addition, low pO2 suppressed expression of the Hif1/2 target

glutaminase (Gls), the amidohydrolase that generates glutamate from glutamine

(DeBerardinis and Cheng, 2010), specifically in 3C-CD34-positive cells (Supplementary Fig.

S4B), suggesting that the gene expression changes could not be attributed solely on Hif1

induction. Consistent with this notion, the stabilization of Hif1/2 using CoCl2 was not

8

sufficient to alter the population ratio of 3C-CD34-positive and 3C-CD34-negative cells

(Supplementary Fig. S4C).

We thus hypothesized that low pO2 could regulate HFSCs mainly through more complex,

possibly Hif-independent, effects on metabolism, and to test this we manipulated cellular

respiration pathways (Fig. 4A). To address if HFSC lineage progression required OXPHOS

and therefore the ETC, we blocked the ETC with two inhibitors, antimycin A and rotenone

(Boveris and Chance, 1973). Both inhibitors reduced the relative amount of CD34-negative

progenitors in the 3C cultures (Fig. 4B) without affecting the total numbers of live cells

(Supplementary Fig. S4D). To test if the HFSC state would depend on glucose oxidation we

interfered with pyruvate entry into the TCA cycle using dichloroacetate (DCA) and UK5099

(Hildyard et al., 2005; Stacpoole, 1989). Surprisingly, boosting glucose oxidation by DCA or

blocking it with UK5099 did not alter the 3C-HFSC-progenitor equilibrium (Fig. 4C,

Supplementary Fig. S4E). We further inhibited glycolysis using the non-metabolizable

glucose analogue 2-deoxyglucose (Wick et al., 1957) and by replacing glucose with

galactose that provides no net ATP generation through oxidation, forcing cells to rely

increasingly on OXPHOS for ATP production (Supplementary Fig. S4F, G). Strikingly, none

of these manipulations had a significant effect on the 3C-HFSC-progenitor equilibrium (Fig.

4C; Supplementary Fig. 4E-G). The lack of effect of blocking glycolysis on HFSC lineage

progression was unexpected, as previous studies had indicated that HFSC activation,

requires anaerobic glycolysis and lactate production (Flores et al., 2017). To assess the role

of anaerobic glycolysis and lactate on 3C-HFSC proliferation in our system, we inhibited

lactase dehydrogenase using FX11 (Flores et al., 2017), and observed that inhibiting lactate

production attenuated proliferation of both HFSCs and ORS progenitors (Supplementary Fig.

S4H), but did not affect the proportions of the two populations (Fig. 4D). This indicated that,

consistent with Flores et al., anaerobic glycolysis is required for HFSC activation, but it does

not influence HFSC to ORS lineage progression.

Glutamine has been shown to provide an alternative carbon source to glucose in cancer

cells and certain SCs (Jin et al., 2015; Oburoglu et al., 2014), and its increased consumption

is associated with increased optical redox ratio (Varone et al., 2014). As blocking glucose

entry to the TCA cycle had no major effect on HFSC fate, we asked if progenitor lineage

progression could instead require glutaminolysis. Supporting this idea, LC-MS analyses

showed higher levels of glutamine and glutamate in in vivo FACS-purified progenitors

9

compared to HFSCs (Supplementary Fig. S4I). Intriguingly, blocking glutaminase using

CB839 (Gross et al., 2014) resulted in an increase in the relative amounts of 3C-HFSCs in

organoid cultures, indicative of reduced lineage progression (Fig. 4E). Further, consistent

with the requirement for glutamate to fuel the TCA cycle to promote 3C-HFSC differentiation,

LC-MS analyses of 13C-labeled glutamine confirmed decreased entry of glutamine into the

TCA cycle in 3C-HFSCs under 2% O2 conditions where lineage progression was attenuated.

In contrast, no difference in entry of glycolytic metabolites was observed with 13C-glucose

tracing (Fig. 4F). The analyses further confirmed the pronounced differences in the

metabolic profiles of CD34-positive 3C-HFSCs and CD34-negative progenitors, with

progenitors displaying increased glutamine metabolism and absent metabolic response to

low pO2 (Fig. 4F). Importantly, the reduced lineage progression induced by the 2% O2 culture

conditions could be fully rescued by supplementing the TCA cycle with exogenous,

membrane permeable dimethyl--ketoglutarate, the direct downstream metabolite of

glutamate (Fig. 4G). Collectively, these data indicate that the HFSC state is metabolically

flexible and does not depend on glycolysis or TCA cycle. In contrast, HFSC lineage

progression to progenitors requires OXPHOS and glutamate entry into the TCA cycle.

mTORC2 controls HFSC metabolism and ORS cell fate reversibility through Akt

To identify the mechanisms that control HFSC metabolism and fate during lineage

progression, we returned to the RNAseq analyses to identify upstream regulators that could

explain the differential expression of metabolic genes in 3C-HFSCs and progenitors.

Interestingly, pathway analysis predicted Rictor, the defining component of mTORC2, as a

potential upstream regulator of the observed gene expression changes (Fig. 5A). To assess

the role of mTORC2 in HFSC lineage progression and fate reversibility, we isolated

epidermal cells from epidermis-specific Rictor knockout mice (RicEKO) (Ding et al., 2016) and

subjected them to 3C organoid culture. Intriguingly, RicEKO cells showed reduced 3C-HFSC

content in organoid cultures compared to control cells, and failed to respond to low pO2 with

an increase in 3C-HFSCs (Fig. 5B). The defect in HFSC fate dynamics was confirmed by

qRT-PCR analysis of key HFSC identity genes which were low in RicEKO organoids and were

not further upregulated in response to low pO2 (Supplementary Fig. 5A). Importantly, the

overall cell composition within the 3C cultures was comparable between RicEKO and controls,

both cultures consisting of Sox9-positive/CD34-positive 3C-HFSCs and Sox9-

positive/CD34-negative ORS progenitors (Supplementary Fig. 5B).

10

To assess if mTORC2 signaling was required to prevent excessive HFSC lineage

progression or to promote progenitor return to HFSC state, we FACS-purified HFSCs and

progenitor populations from control and RicEKO mice and placed them into 3C cultures

separately. Notably, RicEKO organoids showed comparable rates of 3C-HFSC lineage

progression to control cells, but the return of RicEKO progenitors to the CD34+/6+ HFSC

state was severely compromised (Fig. 5C; Supplementary Fig. S5C), indicating that

mTORC2 controls HFSC fate reversibility.

To address the molecular mechanism by which mTORC2 impacts HFSC fate reversibility

and the progenitor cell state, we analyzed the activation status of the key downstream target

of mTORC2, Akt-pS473 (Sarbassov et al., 2005), and observed its phosphorylation to be

significantly attenuated in RicEKO cells (Fig. 5D). The mTORC1 downstream target pS6 was

not altered, confirming the specific effect of Rictor deletion on mTORC2 activity (Fig. 5D).

Consistent with the reduced levels of Akt-pS473 in RicEKO cells, global inhibition of Akt in 3C

cultures phenocopied RicEKO cells in their inability to maintain a 50-50 population equilibrium

of 3C-HFSCs and progenitors, and to respond to low pO2 by promoting the HFSC state (Fig.

5E). Inhibition of Akt in Rictor-deficient cells did not induce an additive effect (Fig. 5E), further

confirming that Akt acts downstream of mTORC2. Interestingly, inhibition of mTORC1 using

rapamycin (Laplante and Sabatini, 2012; Loewith and Hall, 2011) did not affect 3C-HFSC

proportions (Supplementary Fig. S5D), indicating that the role of mTOR on HFSC to ORS

early fate reversibility is specific to mTORC2.

Finally, consistent with requirement for Akt downstream of hypoxia to promote the HFSC

state, we observed that 2% pO2 promoted Akt phosphorylation, particularly in the 3C-CD34-

negative progenitors (Fig. 5F). Hypoxia-dependent Akt activation was Rictor-dependent, as

shown by absence of pAkt induction in RicEKO cells in 2% pO2 (Fig. 5G). Further, activation

of Akt using SC79 (Jo et al., 2012) was sufficient to increase levels of 3C-HFSCs in 20% O2

conditions (Supplementary Fig. S5E). Collectively, the data indicate that the mTORC2/Akt

signaling axis is required downstream of low pO2 for the ability of progenitors to return to the

HFSC state.

mTORC2-Akt signaling axis regulates glutaminolysis and progenitor fate reversibility

As mTORC2-Akt signaling was required for the response of progenitors to low pO2, we next

sought to analyze the metabolic state of mTORC2-deficient cells. LC-MS quantification of

11

metabolites from cells in organoid cultures revealed that, as expected, lowering pO2 led to

accumulation of Acetyl-CoA and reduction of downstream TCA cycle metabolites (Fig. 6A).

This metabolic reprogramming depended on Akt activity, as Akt inhibition in low pO2 induced

an opposite effect of increased concentration of TCA metabolites (Fig. 6A). Furthermore,

whereas low pO2 led to reduced levels of glutamate, Akt inhibition increased the glutamate

concentration (Fig. 6B). Interestingly, inhibition of Akt had no substantial effects on

metabolites in high pO2 conditions (Fig. 6A, B).

To understand how low pO2 and Akt signaling could control the TCA cycle and in particular

glutamine metabolism, we analyzed the expression of Gls. Interestingly, whereas low pO2

suppressed Gls expression, inhibition of Akt counteracted this effect (Fig. 6C). The same

effect was observed in RicEKO cells (Fig. 6D).

Collectively these data suggested that low pO2 suppresses Gls expression in an mTORC2-

Akt-dependent manner to reduce glutamine hydroxylation to prevent the flux of glutamate

into the TCA cycle, and that this metabolic reprogramming is required to facilitate progenitor

return to the HFSC state. To challenge this prediction, we assessed if blocking glutamine-

to-glutamate hydroxylation would rescue the inability of RicEKO progenitors to return to the

HFSC state. Indeed, treating Rictor-deficient cells with the Gls inhibitor restored the ability

of RicEKO progenitor cells to convert into 3C-HFSCs (Fig. 6E). This effect on 3C-HFSC fate

was independent of cell proliferation or survival, as the Gls inhibitor had no major effects on

RicEKO progenitor proliferation or death (Supplementary Fig. S6A, B). Gls inhibition seemed

to mainly target 3C-ORS progenitors as 3C cultures from purified CD34-positive cells

showed a less pronounced increase in HFCS content upon Gls inhibition (Supplementary

Fig. S6C). Taken together these data indicate that 3C-HFSC lineage progression into the

progenitor state requires flux of glutamate into the TCA cycle to maintain the ETC, and that

this metabolic pathway needs to be suppressed by mTORC2-Akt-dependent

downregulation of Gls expression for progenitors to return to the HFSC state.

mTORC2 is required for HFSC fate reversibility and long-term maintenance in vivo

As the in vitro data so far assigned a critical role for mTORC2 signaling in metabolic

regulation and HFSC fate reversibility, we proceeded to assess the consequences of these

processes in vivo in the hair follicle (Fig. 7A). Newborn and early adult RicEKO mice are

characterized by a mild barrier defect and epidermal atrophy (Ding et al., 2016; Ding et al.,

2019), whereas hair follicle morphogenesis did not display obvious impairment in mutants

12

(Fig. 7B, C). In addition, careful inspection revealed overall normal hair follicle architecture

(Supplementary Fig. S7A). However, entry into the first postnatal telogen at P21 was

delayed in RicEKO mice (Fig. 7B, C). This was followed by later entry into anagen and again

a delayed entry into telogen in the following cycles (Fig. 7B, C), indicative of impaired HFSC

regulation.

More detailed analyses of the telogen hair follicles in RicEKO mice revealed striking absence

of the secondary bulge that is formed by the ORS progenitors when they return to

quiescence (Fig. 7D, E, Supplementary Fig. S7B) (Hsu et al., 2011). To assess whether the

absence of the second bulge was due to a club hair retention defect that can result in loss

of the first bulge (Higgins et al., 2009; Lay et al., 2016), we traced the fate of club hair through

a full hair cycle by staining the hair coat with a UV-visible dye. RicEKO showed comparable,

or even increased hair dye retention compared to controls, excluding a club hair retention

defect (Supplementary Fig. S7C). To assess if RicEKO were instead unable to generate a

new bulge, we performed lineage tracing analyses, where we labeled ORS cells by

administering BrdU in late anagen (P34 for control, P48 for RicEKO), when bulge HFSCs

have ceased proliferating (Hsu et al., 2011), and traced the fate of the label-retaining BrdU-

positive cells until the next telogen. Presence of BrdU-positive cells in the bulge niche

indicates that they are derived from ORS cells from previous anagen (Hsu et al., 2011; Lay

et al., 2016). As expected, no difference in initial BrdU label incorporation was observed

directly post BrdU administration in early anagen (Supplementary Fig. S7D), indicating

normal progenitor proliferation. Also, no defects in ORS identity, as marked by Barx2 and

Sox9, or abnormal apoptosis in catagen were observed (Supplementary Fig. S7E-G). In

contrast, when BrdU-labeled cells were traced to the following telogen phase, very few

labeled cells were found in RicEKO mouse bulges, whereas control mice showed significant

amounts of labeled cells in the new bulge (Fig. 7F). Importantly, and as expected, the old

bulge was negative for BrdU (Fig. 7F). This indicated that RicEKO ORS cells are unable to

form a new HFSC bulge niche.

To exclude that the hair follicle cycling defects and the lack of second bulge formation could

be indirect consequences of potential defects outside the hair follicle, we generated a HFSC-

specific deletion of Rictor using the bulge-specific Keratin19-Cre-ERT (Means et al., 2008).

As reported by others (Vagnozzi et al., 2015) Keratin 19-Cre-ERT-mediated Rictor deletion

was not complete (Supplementary Fig. S7H), but it was sufficient to trigger delayed entry

13

into telogen and subsequent compromised formation of the second bulge similar to the

Keratin 14-Cre deletion (Supplementary Fig. S7I-L), confirming that the phenotype of Rictor-

deficient hair follicles is HFSC-autonomous.

To test if the inability to form a second bulge was indeed due to the inability of ORS cells to

switch off glutaminolysis and the TCA cycle, we hypothesized that inhibition of Gls in RicEKO

mice should be sufficient to rescue this phenotype. We injected RicEKO mice intradermally

with the Gls inhibitor BPTES (Robinson et al., 2007) in mid-anagen and address timing of

telogen entry and bulge number. Intriguingly, inhibition of Gls enhanced formation of a

second bulge in RicEKO mice (Fig. 7G).

Consistent with the role of mTORC2-Akt in maintaining HFSC fate reversibility during HFSC

activation, Akt-pS473 levels were very low in telogen hair follicles, whereas a substantial

increase was observed in anagen specifically at the upper ORS region right below the bulge

HFSC niche (Fig. 7H). Further, and as predicted from the ability of low pO2 to trigger Akt-

pS473 in organoid CD34-/Sox9+ ORS progenitors, this increase in Akt phosphorylation in

anagen ORS coincided with pronounced hypoxia at the anagen-stage upper ORS

(Supplementary Fig. S7M). Akt activity remained high in catagen hair follicles, but it was

present only below and within the new and not the old bulge. As expected, Akt activity was

decreased in all hair follicle stages in RicEKO mice (Fig. 7H).

Finally, although the young RicEKO mice did not show a reduction in hair follicle density or

HFSCs (Supplementary Fig. S7N, O), aging of RicEKO mice was accompanied by hair loss

(Fig. 7I), reduced hair follicle density (Fig. 7I) and reduction in HFSCs (Fig. 7J). Collectively,

these data indicate that mTORC2-Akt signaling is required for timely ORS progenitor return

to the HFSC state within the hypoxic upper ORS to establish the new bulge SC niche. The

failure to do so results in age-induced reduction of HFSCs, leading to reduction in hair

follicles and thinning of the hair coat.

Discussion

Changes in metabolic pathways actively influence SC differentiation in multiple tissues (Ito

and Ito, 2016; Shyh-Chang et al., 2013). Yet, little is known about the tenability of this

metabolic reprogramming and its role in SC fate reversibility. Our current data demonstrate

that the transition of HFSCs to the ORS progenitor state involves activation of OXPHOS and

14

flux of glutamine into the TCA cycle. This metabolic switch to OXPHOS/glutamine

metabolism needs to be suppressed to allow progenitors return to the bulge SC state at the

end of the anagen growth phase, identifying a previously unknown role for metabolic

flexibility in SC fate decisions. We further unravel a role for localized TORC2-Akt signaling,

through repressing Gls expression and thus glutamine metabolism within the bulge niche

region, for metabolic reprogramming required for ORS progenitor-mediated generation of

the new bulge niche. Consequently, the failure to locally activate mTORC2-Akt signaling

results in loss of SC fate reversibility and the ability of HFSCs to regenerate their bulge niche

after the hair cycle, leading to HFSC exhaustion and hair loss.

Various SCs reside in hypoxic niches and hypoxia has consequently been shown to be

important for the maintenance of hematopoietic, mesenchymal and neural SCs (Ito and Ito,

2016). Also the hair follicle niche is mildly hypoxic; whereas the epidermis has a partial O2

pressure (pO2) between 0.2-8% and the lower dermis reaches 10%, the pO2 of the hair

follicle is only 0.1-0.8 % (Bedogni et al., 2005; Evans and Naylor, 1967; Peyssonnaux et al.,

2008). Our data further identifies the anagen upper ORS as particularly hypoxic, which could

be a result from the expansion of the hypoxic hair follicle area during anagen. It was recently

shown that the anaerobic glycolysis, and in particular lactate production through lactate

dehydrogenase is critical for HFSC activation to proliferate at the onset of the anagen growth

phase (Flores et al., 2017, Miranda et al., 2018), further consistent with the hypoxic state of

the HFSC niche (Rezvani et al., 2011). In agreement with this, we observe that low pO2

conditions promote maintenance of the HFSC state, whereas HFSC differentiation is

associated with increased OXPHOS. Consistent with the Flores et al (2017) study, we

observe that anaerobic glycolysis and lactate are required for HFSC and ORS progenitor

proliferation to initiate the hair follicle anagen growth phase. However, glycolysis does not

regulate the lineage progression of HFSCs to ORS progenitors or the return of ORS cells to

the HFSC state. In fact, these cell states show similar levels of glycolysis, whereas Flores

et al (2017) showed that HFSCs show increased levels of glycolytic metabolism when

compared to the total epidermis (including terminally differentiated cells). This indicates that

both HFSCs and the early ORS progenitor state cells likely have relatively high levels of

glycolysis compared to the rest of the epidermis, and that this is critical for their proliferation,

whereas the role of OXPHOS in lineage progression is independent from proliferation and

primarily regulates the progenitor-to-HFSC conversion.

15

Interestingly, due to the lack of change in glycolysis, the increase in TCA cycle flux is

facilitated by increased glutamine metabolism. Glutamine can contribute to essentially every

core metabolic task of proliferating cells. It participates in bioenergetics, supports cell

defences against oxidative stress and can also replace glucose in the production of

macromolecules (DeBerardinis and Cheng, 2010). We observe that the increase in

glutaminolysis is regulated by the mTORC2-Akt signalling axis which, like hypoxia,

suppressed the expression of Gls. Our observation is consistent with previous cell culture

studies showing increased glutamine consumption in Rictor-deficient keratinocytes

(Tassone et al., 2017). Notably, it has previously been shown that cancer cell proliferation

under hypoxic conditions relies on mitochondrial glutamine metabolism (Sun and Denko,

2014). Thus, it is feasible to hypothesize that within the hypoxic niche, whereas lactate is

required for the early activation and proliferation of HFSCs, glutamine could provide the

necessary carbon source to facilitate efficient differentiation of HFSC immediate progeny,

similarly to what has very recently been reported in chondrocytes (Stegen et al., 2019).

Consequently, at the end of the anagen growth phase returning to the hypoxic niche, ORS

progenitors would need to reprogram their metabolism by locally activating mTORC2-Akt

signalling to suppress Gls expression facilitating the switch to back to the quiescent HFSC

state. This is consistent with the observed pattern of Akt activation that occurs specifically

within the upper ORS directly adjacent to the bulge niche and is temporally restricted to the

anagen-catagen phase. Interestingly, the HFSC-ORS lineage progression was rapamycin

insensitive. This is consistent with previous studies showing that mTORC1 activity is

restricted to the inner root sheath of anagen hair follicles (Castilho et al., 2009). Thus,

whereas mTORC2 regulates early HFSC fate determination, mTORC1 activity is critical for

hair follicle differentiation (Castilho et al., 2009).

In summary, the present study demonstrates the functional importance of precise

spatiotemporal control of metabolic flexibility to facilitate SC fate reversibility that is required

for the maintenance of a stable SC population for the lifetime of the organism.

Limitations of Study

One limitation of our study is the lack of methodology to trace and quantify metabolic fluxes

in vivo in the cycling hair follicle. The organoid model for hair follicle stem cells used here

recapitulates key early features but not all the steps of hair follicle differentiation. In addition,

16

a genetic model directly targeting glutamine metabolism would further clarify the role of this

pathway in hair follicle stem cell fate.

Author contributions

SAW conceived and supervised the study. CSK designed and performed most of the

experiments and analyzed data. XD and SAE developed the mouse models and

conceptualized, performed and analyzed most of the in vivo experiments. KA, NLH, LCB,

and CAC-M designed and performed cell culture experiments. OK and KPQ designed,

performed and analyzed the multiphoton redox ratio imaging experiments. CE and PG

designed and performed the metabolite analyses. MD conceived experiments and analyzed

data. SAW wrote the paper. All authors commented and edited the manuscript.

Acknowledgements

We thank Yekaterina A. Miroshnikova for discussions and help with experiments, Anu M.

Luoto for technical assistance, Anu Suomalainen and Carien Niessen for critical reading of

the manuscript, Nils-Göran Larsson (Karolinska Institute, Sweden) for Mito-YFP mice,

Roland Wedlich-Söldner (University of Münster, Germany) for LifeAct mice, Guoqiang Gu

(Vanderbilt University, USA) for Keratin 19-Cre ERT mice, Michael Hall and Markus Rüegg

(Biozentrum, University of Basel, Switzerland) for Rictor flox mice, and Catherin Niemann

and Anna Geueke (University of Cologne, Germany) for support with the Ric-K19-CreERT

mouse model. We thank the FACS & Imaging and Comparative Biology core facilities at MPI

for Biology of Ageing for support. This work was supported by the Max Planck Society (to

SAW), the Deutsche Forschungsgemeinschaft (DFG): (project number 73111208 - SFB 829

to MD, SAE, and SAW), the Cluster of Excellence CECAD (to MD, SAE, and SAW); the

Research Unit FOR2599, Research Unit FOR2240, Center for Molecular Medicine and the

DEBRA International Foundation (all to SAE); Helsinki Institute of Life Science, Wihuri

Research Institute, Jane and Aatos Erkko Foundation, European Research Council (ERC)

under the European Union's Horizon 2020 research and innovation programme (grant

agreement 770877) (all to SAW).

Declaration of interests

S.A.W. and C.A.C.M disclose that they are inventors of a patent application

PCT/EP2016/073675 (WO 2017/060240 (A1)) for the 3C Culture method of epidermal stem

cells. All other authors have no competing interests.

17

Main Figure Titles and Legends

Fig. 1. HFSCs and early progenitor cells have distinct metabolic profiles

(A) Schematic of the pilosebaceous unit and cell surface markers used to identify specific

sub-populations: interfollicular epidermis (IFE; Sca1+/CD34-/6 integrin+; in blue), outer

bulge hair follicle stem cells (HFSC; Sca1-/CD34+/6 integrin+; in orange) and outer root

sheath (ORS; Sca1-/CD34-/6 integrin+; in magenta). Epidermal cells were isolated and

placed in 3C organoid culture, after which 3C-CD34+/6 integrin+ and 3C-CD34-/6+ cells

were purified and analyzed.

(B) GO-term analyses from RNAseq data from FACS-purified 3C-CD34+/- cells.

(C) Normalized mean intensity of mito-YFP mice from freshly isolated (left) and 3C cultured

HFSCs and progenitors (n=4-6 mice; *P < 0.05, **P< 0.01, Mann-Whitney).

(D) Representative quantifications of real-time cellular oxygen consumption rates (ORC) (left)

and extracellular acidification rates (ECAR) (right) of FACS-purified 3C-CD34+/- cells.

(E) Quantification of basal respiration (left) and spare respiratory capacity (right) in FACS-

purified 3C-CD34+/- cells (n=3 mice; *P< 0.05, Ratio paired t-test).

(F) Quantification of glycolysis level, glycolysis capacity, and glycolysis reserve in FACS-

purified 3C-CD34+/- cells (n=3 mice; ns=not significant).

(G) Relative amount of key TCA cycle metabolites quantified by LC-MS from freshly isolated

CD34+/-; 6 integrin+ cells, normalized to CD34+ cells (mean SD; n=3-4 mice; *P< 0.05,

Mann-Whitney).

Fig. 2. Anagen entry is associated with increased oxidation in upper hair follicle cells

in vivo

(A) Schematic illustration of the pilosebaceous unit containing the interfollicular epidermis

(IFE), sebaceous gland (SG) and the hair follicle with the outer bulge HFSCs (in orange),

inner bulge niche cells (in yellow) and hair germ/outer root sheath (ORS) located below the

SG and above the dermal papilla (DP). HS=hair shaft; ROI=region of interest used for

measurements.

(B) Quantification of redox ratios from HFSC bulge and ORS areas (dashed lines in A)

(minimum to maximum box and whiskers plot; ***P=0.0135 Student’s t-test ;n=9 follicles

from 2 mice (anagen), n=20 follicles from 2 mice (telogen).

(C) Representative in vivo images of hair follicles. Top row: Two photon excited

autofluorescence from NADH (green) and FAD (blue) and collagen SHG (red) to identify

18

and segment the HFSC bulge area. Bottom row: Optical redox ratio maps of

FAD/(NADH+FAD) autofluorescence. Dashed line shows HFSC bulge. Scale bar 50 µm.

Fig. 3. Low oxygen tension promotes the 3C-HFSCs state

(A) FACS analysis of 3C-CD34+ cells (%) after 14 d of culture in 20% or 2% O2 (n=4

mice/condition; *P< 0.05, Mann Whitney).

(B) qRT-PCR analysis of key HFSC signature genes from 3C cultures in 20% or 2% O2

(mean SD, n=3 mice/condition; *P< 0.05, Mann-Whitney).

(C) Workflow of the lineage tracing experiments. GFP+ CD34+/6+ cells and GFP- CD34-

/6+ cells were FACS-purified, mixed 1:1 and 3C cultured under 20% or 2% O2 for 14 days.

(D) FACS analyses of GFP – cells (left panel) and GFP + cells (right panel) shows increased

conversion of GFP- cells into CD34+ and decreased conversion of GFP+ cells into CD34-

(mean SD n=3 mice/condition; *P< 0.05, Mann Whitney).

Fig. 4. HFSC differentiation requires mitochondrial metabolism and glutaminolysis

(A) Schematics summarizing flux of metabolites into the TCA cycle and the targets of the

inhibitors used.

(B) FACS analyses of HFSC content from 10-day 3C cultured cells treated with 1 nM

Antimycin A (AA) or 10 nM Rotenone (Rot) for 2 days (n= 5 mice (AA), 3 mice (Rot); *P<

0.05, **P< 0.01, Mann Whitney).

(C) FACS analyses of HFSC content from 3C cultured cells treated with 1 µM

dichloroacetate (DCA) (n= 3 mice/condition).

(D) FACS analyses of HFSC content from 3C cultured cells treated with 50 µM FX11 for 2

days (n= 3 mice).

(E) FACS analyses of HFSC content from 10-day 3C cultured cells treated with 20 nM

CB839 glutaminase inhibitor for 2 days (n=4 mice; *P< 0.05, Mann Whitney).

(F) LC/MS quantification of unlabeled (M0) and 13C-labeled (M4/M5) metabolite ratios from

12-day 20% or 2% O2 cultured cells treated with 13C-labeled glucose (pyruvate) or 13C-

glutamine (all other metabolites) for 1 h prior to sample collection and subsequent FACS-

purification of CD34+/- cells. (meanSD; n = 4 mice/condition; *P<0.05, **P<0.001,

***P<0.0007, ****P< 0.0001; ANOVA/Fisher’s LSD).

19

(G) FACS analyses of HFSC content from 10-day 3C cultured cells supplemented with 200

µM dimethyl-α-ketoglutarate for 2 days (n=6 mice (20% O2)/4 mice (2% O2); *P=0.049,

Kruskal-Wallis/Dunn’s; ns=not significant).

Fig. 5. mTORC2 controls HFSC metabolism and fate reversibility through Akt.

(A) Ingenuity pathway analysis from RNAseq data predicts Rictor as potential upstream

regulator of gene expression changes in CD34-/6+ cells.

(B) FACS analyses of 3C-HFSC content from cells isolated from Rictor-deficient mice

(RicEKO) and their control littermates (CTR) cultured in 3C under 20% or 2% O2 for 14 days

(n=6 mice/genotype; ***P= 0.0002, ****P< 0.0001, ANOVA/Dunnett’s).

(C) FACS analyses of HFSC content from FACS-purified CD34+/- cells from RicEKO and

CTR mice cultured for 3 days under 20% O2 (n=4 mice (CD34+)/3 mice (CD34-); *P< 0.05,

Mann-Whitney test).

(D) Representative western blots of pAkt and pS6K from RicEKO and CTR keratinocytes.

Quantification represents pAkt/total Akt ratio (n=2 mice/genotype from 4/4 analyzed). (E)

FACS analyses of HFSC content from 12-day 3C cultured RicEKO and CTR cells treated with

10 µM Akt inhibitor for 2 days (n=6 mice/genotype; ***P< 0.0007, ****P< 0.0001, all statistical

comparisons are to CTR 20% O2 and only significant differences are indicated; Sidak’s

multiple comparisons).

(F) Representative images and quantification of pAkt in 3C organoids grown in 20% or 2%

O2. Note increased pAkt-S473 in response low p02, particularly in 3C-CD34- cells (95%

confidence interval box-and-whiskers; n>17 organoids/condition pooled across 4 mice;

**P=0.0098, ***P=0.0005, ANOVA/Fisher’s; scale bars 70 µm).

(G) Representative western blots of pAkt from RicEKO and CTR keratinocytes exposed to 5h

of 20% or 2% O2. Quantification represents pAkt/total Akt ratio (n=2 mice/genotype from 4/4

analyzed).

Fig. 6. mTORC2-Akt signaling axis regulates glutaminolysis and progenitor fate

reversibility

(A, B) Cells were cultured in 3C under 20% or 2% O2 for 12 d after which 10µM Akt inhibitor

was added. 2 days after application of inhibitor key TCA metabolites (A) and glutamate (B)

were quantified by LC-MS (n = 3 mice/condition; * P< 0.05, Student’s t-test).

20

(C) qRT-PCR analysis of Gls expression from 14-day 3C cultured cells under 20% or 2% O2

with or without Akt inhibitor (n=3 mice/condition; *** P< 0.001, ****P< 0.0001, ANOVA/Holm-

Sidak).

(D) qRT-PCR analysis of Gls expression from 14-day 3C cultured RicEKO and control (CTR)

cells under 20% or 2% O2 (n=3 mice/condition; *P< 0.05, ***P< 0.001, ANOVA/Holm-Sidak).

(E) FACS-purified CD34- cells from RicEKO and CTR mice cultured in 3C for 3 days with or

without 20 nM CB839 glutaminase inhibitor after which CD34+ cells (%) were analyzed via

FACS (n=3 mice/condition; *P< 0.05, **P< 0.01, Ratio paired t-test).

Fig. 7. mTORC2 is required for HFSC fate reversibility and long-term maintenance in

vivo

(A) Schematic illustration of hair follicle (HF) cycling in mice. HFs undergo cycles of telogen

(rest), anagen (regeneration), and catagen (degeneration) throughout the lifetime of the

animal. Each HF generates a new bulge and new club hair with every anagen.

(B) H/E stainings of control (CTR) and RicEKO skin at various hair cycle stages. Scale bars

200 µm.

(C) Quantification of hair follicle (HF) cycling (mean SEM; n=3-4 mice/genotype/timepoint).

(D, E) Representative H&E staining images of CTR and RicEKO mice in 2nd telogen (D) and

quantification (E) of hair follicles with more than two bulges (red arrows) (n = 5

mice/genotype; **P< 0.01, Mann Whitney, scale bars 100 µm).

(F) Experimental timeline, representative immunofluorescence images and quantification of

lineage tracing experiments where proliferative ORS cells were labeled with a BrdU pulse in

late anagen after which BrdU labeled cells were traced during the following telogen phase.

(n = 4 mice/genotype; *P< 0.05, Mann Whitney; scale bars 100 µm).

(G) Representative H&E staining images and quantification of telogen bulge number of

RicEKO mice treated with BPTES to inhibit glutaminolysis during anagen (n = 4 mice/condition;

*P< 0.05, Mann Whitney; scale bars 100 µm).

(H) Representative immunofluorescence images and quantification of telogen, anagen and

catagen hair follicles shows no pAkt (magenta) in telogen hair follicle bulges (CD34; cyan),

whereas anagen whereas anagen and catagen upper ORS shows high activity in CTR. pAkt

is decreased in telogen, anagen, and catagen skin of RicEKO. Unspecific immunoreactivity

of the hair shaft is marked with an asterisk (n = 4 mice/genotype/time point; *P< 0.05, Mann

Whitney; scale bars 100 µm).

21

(I) CTR and RicEKO mice at 24 months age. Red arrow points to areas of sparse hair (right).

Quantifications of hair follicle (HF) density of 20 months-old mice (left) (n= 4 mice/genotype,

*P= 0.0286, Mann Whitney).

(J) Representative hair follicle bulges and quantification of CD34+ cells from 20 months-old

mice (n= 5 mice/genotype, *P=0.027 Student’s t-test; scale bars 100 µm).

STAR METHODS

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will

be fulfilled by the Lead Contact, Sara A. Wickström (sara.wickstrom@helsinki.fi).

Materials Availability

This study did not generate new unique reagents.

Data and Code Availability

Datasets and source data are available from the corresponding author upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Primary cells and cell lines

Epidermal cells were isolated from telogen-stage (P65-69) C56BL/6 mice, LifeAct-GFP

(Riedl et al., 2010), or Rictor-deficient mice (Ding et al., 2016) and cultured in 3C conditions

(Chacón-Martínez et al., 2016). For this, epidermal single cell suspensions were generated

by incubating skin pieces in 0.8% trypsin for 50 min. Cells were embedded in a 1:1 mixture

of growth factor-reduced Matrigel and 3C medium (MEM Spinner’s modification (Sigma), 5

µg/mL insulin (Sigma), 10 ng/mL EGF (Sigma), 10 µg/mL transferrin (Sigma), 10 µM

phosphoethanolamine (Sigma), 10 µM ethanolamine (Sigma), 0.36 µg/mL hydrocortisone

(Calbiochem), 2mM glutamine (Gibco), 100 U/mL penicillin and 100 µg/mL streptomycin

(Gibco), and 10% chelated fetal calf serum (Gibco), 5 µM Y27632, 20 ng/mL mouse

recombinant VEGF, 20 ng/mL human recombinant FGF-2 (all from Miltenyi Biotec)). Cells

were cultured in 5% CO2 at 37 C unless otherwise indicated. For hypoxia experiments cells

were cultured in a hypoxia chamber (Hypoxic Glove Box, Cat# 8375250, CoyLabs) in 2%

O2 at 37 C. No cell lines were used in this study.

Mouse models

22

Wild type male C57Bl6/J mice, as well as Mito-YFP (Sterky et al., 2011) and LifeAct (Riedl

et al., 2010) mice, have been described earlier and were used for cell isolation and for 3C-

HFSC organoids. Rictor was deleted in the epidermis as previously described (Ding et al.,

2016). Briefly, mice carrying a loxP-flanked Rictor allele were crossed with a transgenic

mouse line expressing Cre recombinase under the control of the human Keratin 14 promoter

(Bentzinger et al., 2008; Hafner et al., 2004), leading to epidermis-specific deletion of rictor

(RicEKO) in the entire epidermis. Rictor LoxP/wt-Krt14Cre+, Rictor LoxP/LoxP and Rictor

LoxP/wt mice were phenotypically indistinguishable from each other and were all used as

controls. To generate a deletion specifically in bulge HFSCs, mice carrying a loxP-flanked

Rictor allele were crossed with a mouse carrying a Cre recombinase fused to a mouse

estrogen receptor ligand binding domain that is inserted upstream of the Keratin 19 (Krt19)

gene (K19-CreERT) (Means et al., 2008). To activate K19-CreERT, mice were fed with

tamoxifen diet (TAM400/CreER, TD.55125, Envigo) for 3 weeks starting from P26, the

beginning of anagen. Tissue samples for histology were collected at various time points

indicated. Gender-matched littermates were used in all experiments. For 3C organoid

cultures, tissues were harvested when both RicEKO and CTR mice were in telogen stage

(P65-69).

Animals were housed and maintained according to FELASA guidelines in the animal facility

of the Max Planck Institute for Biology of Ageing, facility of the Department of Pharmacology,

University of Cologne, Germany and University of Helsinki. All experiments were approved

by local authorities (permit numbers 2017.A082, 2014.A067 and ESAVI-6357-2020).

METHOD DETAILS

3C organoid culture

Epidermal cells were isolated from male telogen-stage C56BL/6 mice, LifeAct-GFP (Riedl

et al., 2010), or Rictor-deficient mice (Ding et al., 2016) and cultured in 3C conditions as

described earlier (Chacón-Martínez et al., 2016). Briefly, epidermal single cell suspensions

were generated by incubating skin pieces in 0.8% trypsin for 50 min. Cells were embedded

in a 1:1 mixture of growth factor-reduced Matrigel and 3C medium (MEM Spinner’s

modification (Sigma), 5 µg/mL insulin (Sigma), 10 ng/mL EGF (Sigma), 10 µg/mL transferrin

(Sigma), 10 µM phosphoethanolamine (Sigma), 10 µM ethanolamine (Sigma), 0.36 µg/mL

hydrocortisone (Calbiochem), 2 mM glutamine (Gibco), 100 U/mL penicillin and 100 µg/mL

streptomycin (Gibco), and 10% chelated fetal calf serum (Gibco), 5 µM Y27632, 20 ng/mL

mouse recombinant VEGF, 20 ng/mL human recombinant FGF-2 (all from Miltenyi Biotec)).

23

Where indicated Akt inhibitor (1L6-Hydromethyl-chiro-inositol-2(R)-2-O-methyl-3-O-

octadecyl-sn-glycerocarbonate) (10µM, Merck), CB839 (20 nM, Cayman Chemical), CoCl2

(100 µM, Sigma), Sodium dichloroacetate (DCA, 1µM, Sigma), UK5099 (Cayman Chemical),

Antimycin A (1 nM, Sigma), Rotenone (10nM, Sigma), SC 79 (5 µg/ml, Tocris) and dimethyl-

alpha-ketoglutaric acid (200 µM, Sigma) were added. 2-Deoxy-D-glucose (2-DG, Agilent)

was used at various concentrations noted in the figure legend.

Cells were cultured in 5% CO2 at 37 C unless otherwise indicated. For hypoxia experiments

cells were culture in a hypoxia chamber (Hypoxic Glove Box, Cat# 8375250, CoyLabs) in 2%

O2 at 37 C.

Flow cytometry

For freshly isolated epidermal cells, single cell suspensions were generated as above. For

3C cultured cells, matrigel droplets were degraded for 8 min at 37°C in 0.5% Trypsin, 0.5

mM EDTA in PBS. Trypsin was neutralized using FACS buffer (2mM EDTA, 2% FBS in

PBS).

Cells were centrifuged for 5 min at 2000 rpm at 4°C, resuspended in FACS buffer and

stained for cell surface markers, CD34-eFluor 660 (Thermo Fisher Scientific, Cat # 50-0341-

82, 1:100) and ITGA6-PE/Cy7 (Biolegend, Cat #313622, 1:1000) on ice. 15 min before the

analysis, 7AAD (Thermo) was added to each sample. Cells were analyzed using

FACSCANTO II (BD Biosciences) or sorted using FACSAria Illu Svea (BD Biosciences)

and/or FACSAria Fusion (BD Biosciences). Sorted cells were collected into 15-mL conical

tubes with cell culture medium at 4ºC, centrifuged, re-suspended in medium and cultured in

3C.

Pathway analyses

Gene ontology (GO) term analyses of the published RNAseq dataset were carried out using

Panther. The pathway analysis was carried out using Ingenuity Pathway Analysis (IPA;

QIAGEN)

In vivo multiphoton autofluorescence imaging

In vivo multiphoton imaging experiments were approved and performed according to the

University of Arkansas IACUC (Protocol #18084). C57BL/6J male mouse aged 21 days and

23 days were anesthetized using isoflurane (2-5% induction, 1-3% maintenance. The dorsal

24

side of the ears were depilated from the base of the ear to the tip. The ear was then gently

mounted to a custom 3D printed platform using double sided tape (Li et al., 2012). Individual

hair follicles were located and imaged using a Bruker Ultima Investigator multiphoton

microscope (Middleton, Wisconsin) equipped with a Titanium:Sapphire laser (Spectra-

Physics; Santa Clara, California). Images were acquired at a resolution of 1024x1024 with

12-bit depth via a 20x, 1.0 NA water-immersion objective (Olympus; Tokyo, Japan) at 2.5x

digital zoom. Autofluorescence emission was collected through a 680 nm low pass filter

(Chroma, ET680sp-2p) and series of dichroic mirrors that separated emission into four

GaAsP photomultiplier tubes (PMTs) (Hamamatsu; H10770PB-40). NADH fluorescence

isolated at 755nm excitation using a PMT equipped with a 460(±20) nm filter, and FAD

fluorescence was isolated at 855nm excitation using a PMT equipped with a 525(±25) nm

filter. Collagen second harmonic generation (SHG) signal was collected at both excitation

wavelengths using a PMT equipped with a 430nm short pass filter. Hair follicles were easily

recognized due to the intensely autofluorescent hair shafts surrounded by dermal collagen

SHG signal. An overlay of the autofluorescence and SHG images was generated for each

image field and adjusted for contrast and brightness (Fig. 2C, top row) to identify the HFSC

bulge area. The HFSC bulge area was identified by the cellular morphology surrounding

the hair shaft below the strongly autofluorescent sebaceous glands. This bulge area,

excluding pixels containing the hair shaft, was segmented (Fig. 2C, bottom row, dashed

lines) using a manual tracing function written in MATLAB.

Image fluorescence intensities were normalized as previously described to account for

differences in laser power and PMT voltage (Jones et al., 2018; Quinn et al., 2013). The

pixels within the traced HFSC bulge area were then used to compute the average redox

ratio of [FAD/(NADH+FAD)] from the normalized NADH and FAD intensities in 29 different

hair follicles (n=9 anagen, n=20 telogen). Visualization of the redox ratio was achieved by

mapping values corresponding to the redox ratio of each pixel to a jet color map in MATLAB.

Differences in the average optical redox ratio between anagen and telogen bulge regions

were determined using an unpaired t-test.

EdU and Annexin V incorporation

For analysis of cell proliferation, cells were cultured for 12-13 days before EdU (10µM)

incubation for 24 h. Single cell suspensions were incubated with Fixable Viability Dye eFluor

506 (Thermo) for 25 min on ice, washed once with FACS buffer, incubated with antibodies

against cell surface markers as described above and subsequently fixed with 4% PFA. Cells

25

were washed once, permeabilized with Click-iT saponin-based permeabilization buffer for

15 min at room temperature in the dark and washed. The EdU reaction cocktail was

prepared following the manufacturer’s protocol for Click-iT Plus EdU Alexa Fluor 488 Flow

Cytometry Assay Kit (Thermo), incubated for 15 min at room temperature and subjected to

flow cytometry analysis.

For analysis of apoptosis using Annexin V staining, single cell suspensions were incubated

with PE-tagged anti-Annexin V (1:25; Biolegend) for 15min at room temperature. Cells were

washed and 7AAD was incubated for 5min at room temperature prior to flow cytometry

analysis.

Seahorse metabolic flux measurements

Cells were FACS sorted into CD34+ITGA6+ and CD34-ITGA6+ cells and seeded in 3C

conditions on Seahorse XF96 V3 PS cell Culture Microplates (Agilent) and allowed to

recover for 4d. Seahorse measurements were performed following the manufacturer’s

protocol from Agilent Seahorse XF Cell Mito Stress Test Kit and XF Glycolysis Stress Test

Kit. In short, a day before the experiment, calibrant solution was added to XFe96 sensor

cartridges and incubated overnight under 37 ºC without CO2. The day of the experiment,

cultured cells were washed twice and incubated for 45 min at 37 ºC without CO2 with

Seahorse assay media adjusted to pH 7.4 containing 10 mM Glucose and 2 mM L-Glutamine

for mito stress protocol and 2 mM L-Glutamine only for glycolysis stress protocol. For

OXPHOS measurements, 20 µM oligomycin, 20 µM FCCP, and 0.5 µM rotenone/antimycin

A were injected into port A, B, and C respectively. For glycolysis analysis, 100 mM glucose,

10 µM oligomycin, 500 µM 2-DG were injected into port A, B, and C respectively. After the

experiment, cells were lysed in ice-cold RIPA buffer, boiled at 65ºC for 10min, after which

dsDNA content was measured with a Qubit fluorometer (Thermo) according to the

manufacturer’s instructions. DNA concentrations were used to normalize Seahorse

measurements to cell numbers.

qRT-PCR

RNA was isolated using RNeasy Plus Mini Kit or RNeasy Plus Micro Kit (Qiagen). 500 ng of

RNA was subjected to reverse transcription using SuperScript IV VILO Master Mix (Thermo)

following the manufacturer’s protocol. PCR was performed with CFX384 Touch Real Time

PCR Detection System (Bio-Rad) using the DyNAmo Color Flash SYBR Green Mix

(Thermo). Gene expression was quantified using the Ct method using normalization to

26

S18 or Actb. For comparisons of freshly isolated and cultured HFSCs, normalization was

performed using the ERCC RNA Spike-in Mix (Invitrogen). The primer sequences are

provided in Table S1.

Metabolite extraction for LC-MS

Cells were either freshly FACS purified from skin or 3C cultured for 12-14 days after which

they were FACS purified or directly pelleted by centrifugation and snap frozen in liquid

nitrogen. For 13C glutamine/glucose flux analyses, on the day of sample collection 3C-

cultured cells were washed once with PBS, incubated with DMEM with 10% chelated FBS,

and without glucose and glutamine and phenol red (Gibco) supplemented with U-13C6-D-

Glucose (Cambridge Isotope Laboratories) and L-Glutamine (Sigma) or 13C5-L-Glutamine

(Cambridge Isotope Laboratories) and D-Glucose (Agilent) for 3 h under culturing conditions.

Cells were then washed with PBS, trypsinized, neutralized with ice-cold DMEM with 10%

chelated FBS, FACS sorted, and snap frozen in liquid nitrogen.

Metabolite extraction from each cell pellet was performed using 1 mL of a mixture of

40:40:20 [v:v:v] of pre-chilled (-20°C) acetonitrile:methanol:water (OptimaTM LC/MS grade,

Thermo Fisher Scientific). The samples were subsequently vortexed until the cell pellets

were fully suspended, before incubating them on an orbital mixer at 4°C for 30 min at 1500

rpm. For further disintegration, samples were sonicated for 10 min in an ice cooled bath

sonicator (VWR, Germany) before centrifuging them for 10 min at 21100x g and 4°C. The

metabolite-containing supernatant was collected in fresh tubes and concentrated to dryness

in a Speed Vac concentrator (Eppendorf). The protein-containing pellets were also collected

and used for protein quantification (BCA Protein Assay Kit, Thermo Fisher Scientific).

Anion-Exchange Chromatography Mass Spectrometry

Extracted metabolites were re-suspended in 100 µl of water (OptimaTM LC/MS grade,

Thermo Fisher Scientific,). 80 µl were used for the analysis of anionic metabolites from the

TCA cycle and glycolysis, while 20 µL were used for the analysis of amino acids (see below,

LC-MS analysis of amino acids).

Anions were analysed using a Dionex ionchromatography system (ICS 5000, Thermo Fisher

Scientific) essentially as described previously (Schwaiger et al., 2017). In brief, 10 µL of

polar metabolite extract were injected in full loop mode using an overfill factor of 3, onto a

Dionex IonPac AS11-HC column (2 mm × 250 mm, 4 μm particle size, Thermo Scientific)

equipped with a Dionex IonPac AG11-HC guard column (2 mm × 50 mm, 4 μm, Thermo

27

Scientific). The column temperature was held at 30°C, while the auto sampler was set to

6°C. A potassium hydroxide gradient was generated by the eluent generator using a

potassium hydroxide cartridge that was supplied with deionized water. The metabolite

separation was carried at a flow rate of 380 µL/min, applying the following gradient. 0-3 min,

10 mM KOH; 3-12 min, 10−50 mM KOH; 12-19 min, 50-100 mM KOH, 19-21 min, 100 mM

KOH, 21-21.1 min, 100-10 mM KOH. The column was re-equilibrated at 10 mM for 8 min.

For the targeted analysis of pool sizes the eluting metabolites were detected in negative ion

mode using ESI MRM (multi reaction monitoring) on a Xevo TQ (Waters) triple quadrupole

mass spectrometer (MS). For the analysis of isotope ratios the samples were measured on

a Q-Exactive HF high resolution MS (Thermo Fisher Scientific). The settings on the Xevo

TQ were the following: capillary voltage 1.5 kV, desolvation temperature 550°C, desolvation

gas flow 800 L/h, collision cell gas flow 0.15 mL/min. Two MRM transitions (one quantitative

and one confirmative MRM transition). The quantitative MRM transitions were: Citrate m/z

precursor mass (M-H+) 191, fragment mass (M-H+) m/z 111 : cone voltage 18V, collision

energy 10V. Isocitrate m/z precursor mass (M-H+) 191 fragment mass (M-H+) m/z 72: cone

voltage 18V, collision energy 12V. α-ketoglutarate m/z precursor mass (M-H+) 145 fragment

mass (M-H+) m/z 56: cone voltage 18V, collision energy 12V. Succinate m/z precursor mass

(M-H+) 117 fragment mass (M-H+) m/z 55: cone voltage 20V, collision energy 14V. Fumarate

m/z precursor mass (M-H+) 115 fragment mass (M-H+) m/z 27: cone voltage 22V, collision

energy of 8V. Malate m/z precursor mass (M-H+) 132 fragment mass (M-H+) m/z 71: cone

voltage 22V, collision energy of 16V.

Data analysis and peak integration for this datatype was performed using the TargetLynx

Software (Waters).

The high resolution Q-Exactive HF MS was operating in negative ionization mode monitoring

the mass range m/z 50-750. The heated ESI source settings of the mass spectrometer were:

Spray voltage 3.2 kV, capillary temperature was set to 275°C, sheath gas flow 60 AU and

aux gas flow 20 AU at a temperature of 300°C. The S-lens was set to a value of 60. Targeted

data analysis was performed as described in the section for the tracing analysis of amino

acids (see below).

LC-MS analysis of cellular pool sizes and 13C fluxes of amino acids

For amino acid analysis, the benzoylchlorid derivatization method (Wong et al., 2016) was

used. In brief: 20 µl of the aqueous sample, were mixed with 10 µl of 100 mM sodium

carbonate (Sigma) followed by the addition of 10 µl 2% benzoylchloride (Sigma) in

28

acetonitrile (VWR). Samples were then analyzed using an Acquity iClass UPLC (Waters)

connected to a Q-Exactive HF (Thermo Fisher Scientific) operating.

For each analysis 2 µl of the derivatized sample were injected onto a 100 x 1.0 mm HSS T3

column, which is packed with 1.8 µm particles (Waters). The flow rate was 100 µL/min and

the buffer system consisted of buffer A (10 mM ammonium formate, 0.15% formic acid in

water) and buffer B (acetonitrile). The gradient was: 0% B at 0 min; 0-15% B 0-0.1 min; 15-

17% B 0.1-0.5 min; 17-55% B 0.5-14 min, 55-70% B 14-14.5 min; 70-100% B 14.5-18 min;

100% B 18-19 min; 100-0% B 19-19.1 min, 19.1-28 min 0% B. The mass spectrometer was

operating in positive ionization mode monitoring the mass range m/z 50-750. The heated

ESI source settings of the mass spectrometer were: Spray voltage 3.5kV, capillary

temperature 275°C, sheath gas flow 40 AU and aux gas flow 20 AU at a temperature of

300°C. The S-lens was set to 60.

Data analysis of isotope ratios was performed using the TraceFinder software (Version 4.2,

Thermo Fisher Scientific). Identity of each compound was validated by authentic reference

compounds, which were analyzed independently. For the isotope enrichment analysis the

area of the extracted ion chromatogram (XIC) of each isotope [M + H]+ were determined

with a mass accuracy (<5 ppm) before calculating the proportions of each detected isotope

towards the sum of all isotopes of the corresponding compound. These proportions are

displayed as percent values for each isotope.

LC-MS analysis of cellular pools of acetyl and succinyl CoA

For the analysis of acetyl and succinyl CoA the extracted metabolites were re-suspended in

100 µl of UPLC grade acetonitrile:water (80:20 [v:v], OptimaTM LC-MS-grade, Thermo Fisher

Scientific). The samples were analyzed on an Acquity iClass UPLC (Waters), using a

SeQuant ZIC-pHILIC 5 µm polymer 150 x 2.1 mm column (Merck) connected to a Xevo TQ-

S (Waters) triple quadrupole mass spectrometer.

3 µl of re-suspended metabolite extract was injected onto the column and separated using

a flow rate of 350 µl/min of buffer A (20mM ammonium carbonate, 0.1% ammonium

hydroxide ) and buffer B (acetonitrile) using the following gradient: 0-1 min 20-70% A; 1-3

min 70-80% A; 3-4 min 80-98%A; 4-4.5 min 98% A; 4.5-5 min 98-20% A. The column is re-

equilibrated at 20% A for additional 5 min.

The eluting metabolites were detected in negative ion mode using ESI MRM (multi reaction

monitoring) applying the following settings: capillary voltage 1.5 kV, desolvation temperature

550°C, desolvation gas flow rate 800 L/h, collision cell gas flow 0.15 mL/min. The following

29

MRM transitions were used for relative compound quantification of acetyl CoA m/z precursor

mass (M+H+) 810, fragment mass (M+H+) m/z 303: cone voltage of 98V, collision energy

28V, while succinyl CoA was analysed using a precursor mass (M+H+) m/z 868, fragment

mass (M+H+) m/z 361: cone voltage of 2V, collision energy 34V. Data analysis and peak

integration were performed using the TargetLynx Software (Waters).

Gas chromatography mass spectrometry

Similar to the above described 13C flux analysis of glycolysis and TCA cycle intermediates

using ion chromatography coupled to high resolution MS, high resolution GC-MS (Q-

Exactive GC, Thermo Fisher Scientific) was employed.

For this purpose, dried metabolite pellets required a two-step derivatization using

methoxyamine (methoxyamine hydrochloride, Sigma) and N-Methyl-N-trimethylsilyl-

trifluoracetamide (MSTFA, Macherey-Nagel) before performing the GC-MS analysis.

In brief: Dried samples were re-suspended in 5 µL of a freshly prepared (20 mg/mL) solution

of methoxyamine in pyridine (Sigma). The re-suspended pellets were incubated for 90 min

at 40°C on an orbital shaker (VWR) at 1500 rpm. After adding additional 45 µL of MSTFA,

the samples were incubated for additional 30 min at 40°C and agitation at 1500 rpm. At the

end of the derivatisation the samples were centrifuged for 10 min at 21100x g and 40 µL of

the clear supernatant were transferred to fresh autosampler vials with conical glass inserts.

For the GC-MS analysis 1 µL of each sample was injected using a PAL autosample system

(Thermo Fisher Scientifc) using a Split/Splitless (SSL) injector at 300 C in splitless mode.

The carrier gas flow (helium) was set to 2 ml/min using a 30m DB-35MS capillary column

(0.250 mm diameter and 0.25 µm film thickness, Agilent). The GC temperature program was:

2 min at 85°C, followed by a 15°C per min ramp to 330°C. At the end of the gradient the

temperature was held for additional 6 min at 330°C. The transfer line and source

temperature are both set to 280°C. The filament, which was operating at 70 V, was switched

on 2 min after the sample was injected. During the whole gradient period the MS was

operated in full scan mode covering a m/z range between 70 and 800 at a scan speed of 20

Hertz.

Similar to the analysis of the isotope enrichment analysis in the amino acids, isotope

enrichment analysis in glycolysis and TCA cycle metabolites were determined using GC-MS

(Q-Exactive GC-Orbitrap, Thermo Fisher Scientific). For this purpose, metabolites were

derivatized using a two-step procedure starting with an methoxyamination (methoxyamine

30

hydrochloride, Sigma) followed by a trimethyl-silylation using N-Methyl-N-trimethylsilyl-

trifluoracetamid (MSTFA, Macherey-Nagel).

In brief: Dried samples were re-suspended in 5 µL of a freshly prepared (20 mg/mL) solution

of methoxyamine in pyridine (Sigma) to perform the methoxyamination. These samples were

then incubated for 90 min at 40°C on an orbital shaker (VWR) at 1500 rpm. In the second

step additional 45 µL of MSTFA were added and the samples were incubated for additional

30 min at 40°C and 1500 rpm. At the end of the derivatisation the samples were centrifuged

for 10 min at 21100x g and 40 µL of the clear supernatant was transferred to fresh auto

sampler vials with conical glass inserts (Chromatographie Zubehoer Trott). For the GC-MS

analysis 1 µL of each sample was injected using a PAL autosample system (Thermo Fisher

Scientifc) using a Split/Splitless (SSL) injector at 300 °C in splitless mode. The carrier gas

flow (helium) was set to 2 ml/min using a 30m DB-35MS capillary column (0.250 mm

diameter and 0.25 µm film thickness, Agilent). The GC temperature program was: 2 min at

85 °C, followed by a 15 °C per min ramp to 330 °C. At the end of the gradient the temperature

is held for additional 6 min at 330 °C. The transfer line and source temperature are both set

to 280 °C. The filament, which was operating at 70 V, was switched on 2 min after the sample

was injected. During the whole gradient period the MS was operated in full scan mode

covering a mass range m/z 70 and 800 with a scan speed of 20 Hertz.

For the data analysis peak areas of extracted ion chromatograms of each isotope of

compound-specific fragments [M - e-]+ were determined using the TraceFinder software

(Version 4.2, Thermo Fisher Scientific) with a mass accuracy (<5 ppm). Subsequently

proportions of each detected isotope towards the sum of all isotopes of the corresponding

compound-specific fragment were determined. These proportions are given as percent

values for each isotope.

Details on the compound-specific fragments of the analysed compounds: 3-phosphoglyceric

acid was analysed from a three carbon-containing fragment with the chemical formula

C14H36O7PSi4 and the m/z 459.12702. Phosphenolpyruvic acid was analysed from a three

carbon-containing fragment (C11H26O6PSi3) with a m/z of 369.07693. Pyruvic acid was

analysed from a three carbon-containing fragment (C6H12NO3Si) and the m/z of 174.05809.

Lactic acid was analysed from a three carbon-containing fragment (C8H19O3Si2) and a m/z

of 219.08672. Citric acid was analysed from a five carbon-containing fragment

(C11H21O4Si2) and a m/z of 273.09729. Alpha-ketoglutaric acid was analysed from a five

carbon-containing fragment (C8H12NO3Si) and a m/z of 198.0580963. Succinic acid was

analysed from a four carbon-containing fragment (C9H19O4Si2) and a m/z of 247.08164.

Fumaric acid was analysed from a four carbon-containing fragment (C9H17O4Si2) and a

31

m/z of 247.08164. Malic acid was analysed from a four carbon-containing fragment

(C9H17O4Si2) and a m/z of 247.08164. The retention time and therefore identity of each

compound was validated by authentic reference compounds which were analysed

independently.

H&E staining

The upper back skin of mice was dissected fixed in ice-cold 4% PFA for 1h on ice,

dehydrated with 50% ethanol overnight, and processed using ExcelsiorTM AS Tissue

Processor (Thermo), embedded in paraffin and sectioned. Subsequently, paraffin was

removed and H&E staining was performed using automated Gemini slide stainer (Thermo).

Stained tissues were mounted on No. 1 rectangular coverslips with Cytoseal XYL mounting

medium (Thermo). Images of H&E stained tissue samples were obtained using Nikon

Eclipse Ci (Nikon) using Nikon NIS Elements D software and a 20x dry lens.

Western blots

Cells were lysed in radio immunoprecipitation assay (RIPA) buffer (150 mM NaCl, 50 mM

Tris-HCl, 5mM EDTA, 0.1% SDS, 1% sodium deoxycholate and 1% Triton X-100,

supplemented with proteinase inhibitor protease and phosphatase inhibitor tablets (Roche).

Protein concentration was determined by Micro BCA Protein Assay Kit (Thermo Scientific)

after which 20 μg protein was subjected to SDS–PAGE. Separated polypeptides were

transferred onto PVDF membranes. After blocking (5% non-fat milk in TBS-Tween buffer),

membranes were incubated with primary antibodies. Rictor (1:1000, cat #2114), Akt (1:1000,

cat #9272), Akt-pS473 (1:1000, cat #9271), S6-pS240/244 (1:1000, cat #5364) were all from

Cell Signaling and used at 1:1000 dilution. β-actin (1:2000) was from Sigma (A2066) and

Hif1 from Novus Biologicals (1:1000, cat #NB100-479). After incubation with horseradish

peroxidase-conjugated secondary antibodies (DAKO), the blot was developed with ECL

substrate (Pierce) and exposed on X-ray film (Amersham Biosciences) or BioRad Chemidoc

Imaging System.

Immunofluorescence

Tissue biopsies were fixed (4% PFA), embedded in paraffin and sectioned. Sections were

de-paraffinized using a graded alcohol series, blocked in 10% normal goat serum, and

incubated with primary antibodies diluted in Dako Antibody Diluent over night at 4°C. Bound

primary antibody was detected by incubation with Alexa-Fluor 488- or Alexa Fluor 568-

32

conjugated antibodies (Invitrogen). Nuclei were counterstained with 4′, 6-diamidino-2-

phenylindole (DAPI, Invitrogen). After washing slides were mounted in Elvanol. The

following antibodies were used: Akt-pS473 (1:100; Cell Signaling cat #9271), CD34 (1:1000,

gift from Manuel Koch, University of Cologne), CD34 (1:100, Invitrogen cat #14-0341-82),

K16 (1:150, Abcam cat #Ab76414), K6 (1:500, Biolegend cat #905701), Sox9 (1:100, Santa

Cruz cat #sc-166505), and periostin (1:150, Santa Cruz cat #sc-398631). TUNEL staining

was performed with In Situ Cell Death Kit (Roche) according to manufacturer’s instructions.

All fluorescence images were collected by laser scanning confocal microscopy (SP8X; Leica)

with Leica Application Suite software (LAS X version 2.0.0.14332), using 40x, 63x or 100x

immersion objectives. Images where acquired at room temperature using sequential

scanning of frames of 1 µm thick confocal planes (pinhole 1) after which planes were

projected as a maximum intensity confocal stack. Images were collected with the same

settings for all samples within an experiment.

Epidermal whole mounts from back skin

Shaved back skin was dissected, after which adipose and connective tissue were removed

by scraping with a scalpel. Skin pieces were floated on thermolysine/PBS (Sigma-Aldrich)

at 37 °C (dermal side down) for 30-40 min. Epidermis was separated from dermis by peeling

and fixed in 4% PFA for 1.5 h at room temperature. After washing with PBS, epidermal tissue

was processed for staining as described above for immunofluorescence.

Image analyses

All images were analyzed using Fiji (Schindelin et al., 2012). To quantify intensity of hair

follicle immunofluorescence, a region of interest (ROI) was manually drawn around each

hair follicle of maximum intensity projected confocal stacks, after which mean grey value

within each ROI was quantified. BrdU- and Tunel-positive cells and hair follicle bulges were

counted manually. 3C images were analyzed by generating maximum intensity projections

of the acquired confocal stack. ROIs were manually drawn around individual organoids

based on DAPI staining. ROIs of interest were then scored for CD34 staining presence or

absence, and ultimately, mean gray value of the Akt-pS473 staining was determined for

each ROI.

BrdU injections and lineage trace

To identify the hair cycle stage, mouse back skin was shaved during predicted first telogen.

Skin color and morphology were monitored daily. BrdU (BD Pharmingen) was dissolved in

33

PBS and injected intraperitoneally at 50µg/g body weight in 150uL of volume twice at 6h

intervals to littermate mice during mid-anagen; control mice (~P34) and the hair follicle

cycling stage matched RicEKO mice (~P43) as described in Hsu et al. 2011. Back skin tissue

was collected at time points indicated and cryo embedded in OCT (Tissue-Tek). Tissue

sections were cut, fixed with ice-cold methanol, washed with PBS and permeabilized with

0.1% Triton-X-100/PBS for 5 min. DNA was denatured by soaking the samples in 2M HCl

for 30 min. Samples were then subjected to immunofluorescence analyses as described

above with anti-CD34 and anti-BrdU (1:100, BD Biosciences) antibodies.

Hair dye experiments

Mouse back skin was washed and dyed with glow-in-the-dark hair dye (Manic Panic, Hot

Pink) for 30 min during their first telogen hair cycle (~P21) as described in Lay et al 2016.

Hair samples were taken on P26, P39, P48, P53 and P65 and images were taken under an

epifluorescence microscope (Leica Sp8) using UV light.

BPTES injections

Anesthetized mice were shaved on their back skin during the mid-anagen phase (Control:

P34-36; RicEKO: P42-P45 ) and injected intradermally with BPTES (Cayman Chemical) at

12.5 mg/kg body weight in 10% DMSO in the total of 200 µL of PBS as described in (Xiang

et al., 2015) three times a week for 2 weeks or until the second telogen phase had been

reached. The back skin tissue was taken for histological analyses and processed as

described above for H/E stainings.

Hypoxyprobe injection and staining

Solid pimonidazole HCl (Hypoxyprobe-1; Hypoxyprobe Kit, Hypoxyprobe Inc) was

reconstituted in sterile saline and injected intraperitoneally at 60 mg/kg body weight at late

anagen (P38) or telogen (P75). Animals were sacrificed 1 hour post injection, and back skin

collected and fixed (4% PFA), embedded in paraffin and sectioned. Samples were

rehydrated through graded ethanol series and subjected to antigen retrieval in pH6 DAKO

reagent (Agilent Technologies, S2031). Tissue sections were blocked in 5% normal goat

serum/3% BSA/PBS for 1 hour. Sections were incubated with anti-pimonidazole mouse IgG1

monoclonal antibody (Mab1, 1:50, HypoxyprobeKit) in DAKO diluent at 4°C overnight.

Sections were washed with 1xPBS and then incubated with biotinylated secondary for 30

minutes at room temperature. Subsequently, sections were incubated in Vectastain Elite

34

ABC reagent for 30 minutes at room temperature and then washed with 1xPBS. Peroxidase

activity was visualized with Sigmafast 3,3’-Diaminobenzidine (D4168, Sigma). Slides were

then processed as for H&E staining.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical analyses were performed using GraphPad Prism software (GraphPad, version 8).

Statistical significance was determined by the specific tests indicated in the corresponding

figure legends. In all cases where a test for normally distributed data was used, normal

distribution was confirmed with the Kolmogorov–Smirnov test (α = 0.05). No methods were

used to predetermine sample size. All experiments presented in the manuscript were

repeated at least in 3 independent experiments/biological replicates. All P and n values as

well as statistical test used is reported in the figure legends. Mice were assigned to groups

randomly (drug treatments) or according to genotype. Analyses were not blinded, and no

datapoints were removed from the analyses.

References

Albert, V., and Hall, M.N. (2015). mTOR signaling in cellular and organismal energetics. Curr Opin Cell Biol 33, 55-66.

Bedogni, B., Welford, S.M., Cassarino, D.S., Nickoloff, B.J., Giaccia, A.J., and Powell, M.B. (2005). The hypoxic microenvironment of the skin contributes to Akt-mediated melanocyte transformation. Cancer Cell 8, 443-454.

Bentzinger, C.F., Romanino, K., Cloetta, D., Lin, S., Mascarenhas, J.B., Oliveri, F., Xia, J., Casanova, E., Costa, C.F., Brink, M., et al. (2008). Skeletal muscle-specific ablation of raptor, but not of rictor, causes metabolic changes and results in muscle dystrophy. Cell Metab 8, 411-424.

Binek, A., Rojo, D., Godzien, J., Ruperez, F.J., Nunez, V., Jorge, I., Ricote, M., Vazquez, J., and Barbas, C. (2019). Flow Cytometry Has a Significant Impact on the Cellular Metabolome. J Proteome Res 18, 169-181.

Blanpain, C., and Fuchs, E. (2014). Stem cell plasticity. Plasticity of epithelial stem cells in tissue regeneration. Science 344, 1242281.

Boveris, A., and Chance, B. (1973). The mitochondrial generation of hydrogen peroxide. General properties and effect of hyperbaric oxygen. Biochem J 134, 707-716.

Castilho, R.M., Squarize, C.H., Chodosh, L.A., Williams, B.O., and Gutkind, J.S. (2009). mTOR mediates Wnt-induced epidermal stem cell exhaustion and aging. Cell Stem Cell 5, 279-289.

35

Chacón-Martínez, C.A., Klose, M., Niemann, C., Glauche, I., and Wickström, S.A. (2016). Hair follicle stem cell cultures reveal self-organizing plasticity of stem cells and their progeny. EMBO J 36, 151-164.

Chacón-Martínez, C.A., Koester, J., and Wickström, S.A. (2018). Signaling in the stem cell niche: regulating cell fate, function and plasticity. Development 145.

DeBerardinis, R.J., and Cheng, T. (2010). Q's next: the diverse functions of glutamine in metabolism, cell biology and cancer. Oncogene 29, 313-324.

Ding, X., Bloch, W., Iden, S., Ruegg, M.A., Hall, M.N., Leptin, M., Partridge, L., and Eming, S.A. (2016). mTORC1 and mTORC2 regulate skin morphogenesis and epidermal barrier formation. Nat Commun 7, 13226.

Ding, X., Willenborg, S., Bloch, W., Wickström, S.A., Wagle, P., Brodesser, S., Roers, A., Jais, A., Brüning, J.C., Hall, M.N., Rüegg, M.A., Eming, S.A., (2020). Epidermal mTORC2 controls lipid synthesis and filaggrin processing in epidermal barrier formation. J Allergy Clin Immunol 145, 283-300.e8.

Eltschinger, S., and Loewith, R. (2016). TOR Complexes and the Maintenance of Cellular Homeostasis. Trends Cell Biol 26, 148-159.

Evans, N.T., and Naylor, P.F. (1967). The systemic oxygen supply to the surface of human skin. Respir Physiol 3, 21-37.

Flores, A., Schell, J., Krall, A.S., Jelinek, D., Miranda, M., Grigorian, M., Braas, D., White, A.C., Zhou, J.L., Graham, N.A., et al. (2017). Lactate dehydrogenase activity drives hair follicle stem cell activation. Nat Cell Biol 19, 1017-1026.

Georgakoudi, I., and Quinn, K.P. (2012). Optical imaging using endogenous contrast to assess metabolic state. Annu Rev Biomed Eng 14, 351-367.

Gonzales, K.A.U., and Fuchs, E. (2017). Skin and Its Regenerative Powers: An Alliance between Stem Cells and Their Niche. Dev Cell 43, 387-401.

Gross, M.I., Demo, S.D., Dennison, J.B., Chen, L., Chernov-Rogan, T., Goyal, B., Janes, J.R., Laidig, G.J., Lewis, E.R., Li, J., et al. (2014). Antitumor activity of the glutaminase inhibitor CB-839 in triple-negative breast cancer. Mol Cancer Ther 13, 890-901.

Hafner, M., Wenk, J., Nenci, A., Pasparakis, M., Scharffetter-Kochanek, K., Smyth, N., Peters, T., Kess, D., Holtkotter, O., Shephard, P., et al. (2004). Keratin 14 Cre transgenic mice authenticate keratin 14 as an oocyte-expressed protein. Genesis 38, 176-181.

Higgins, C.A., Westgate, G.E., and Jahoda, C.A. (2009). From telogen to exogen: mechanisms underlying formation and subsequent loss of the hair club fiber. J Invest Dermatol 129, 2100-2108.

Hildyard, J.C., Ammala, C., Dukes, I.D., Thomson, S.A., and Halestrap, A.P. (2005). Identification and characterisation of a new class of highly specific and potent inhibitors of the mitochondrial pyruvate carrier. Biochim Biophys Acta 1707, 221-230.

Hsu, Y.C., Li, L., and Fuchs, E. (2014). Emerging interactions between skin stem cells and their niches. Nat Med 20, 847-856.

Hsu, Y.C., Pasolli, H.A., and Fuchs, E. (2011). Dynamics between stem cells, niche, and progeny in the hair follicle. Cell 144, 92-105.

Ito, K., and Ito, K. (2016). Metabolism and the Control of Cell Fate Decisions and Stem Cell Renewal. Annu Rev Cell Dev Biol 32, 399-409.

36

Jin, L., Li, D., Alesi, G.N., Fan, J., Kang, H.B., Lu, Z., Boggon, T.J., Jin, P., Yi, H., Wright, E.R., et al. (2015). Glutamate dehydrogenase 1 signals through antioxidant glutathione peroxidase 1 to regulate redox homeostasis and tumor growth. Cancer Cell 27, 257-270.

Jo, H., Mondal, S., Tan, D., Nagata, E., Takizawa, S., Sharma, A.K., Hou, Q., Shanmugasundaram, K., Prasad, A., Tung, J.K., et al. (2012). Small molecule-induced cytosolic activation of protein kinase Akt rescues ischemia-elicited neuronal death. Proc Natl Acad Sci U S A 109, 10581-10586.

Jones, J.D., Ramser, H.E., Woessner, A.E., and Quinn, K.P. (2018). In vivo multiphoton microscopy detects longitudinal metabolic changes associated with delayed skin wound healing. Commun Biol 1, 198.

Joost, S., Annusver, K., Jacob, T., Sun, X., Dalessandri, T., Sivan, U., Sequeira, I., Sandberg, R., and Kasper, M. (2020). The Molecular Anatomy of Mouse Skin during Hair Growth and Rest. Cell Stem Cell 26, 441-457 e447.

Joost, S., Jacob, T., Sun, X., Annusver, K., La Manno, G., Sur, I., and Kasper, M. (2018). Single-Cell Transcriptomics of Traced Epidermal and Hair Follicle Stem Cells Reveals Rapid Adaptations during Wound Healing. Cell Rep 25, 585-597 e587.

Kobayashi, C.I., and Suda, T. (2012). Regulation of reactive oxygen species in stem cells and cancer stem cells. J Cell Physiol 227, 421-430.

Kolenc, O.I., and Quinn, K.P. (2019). Evaluating Cell Metabolism Through Autofluorescence Imaging of NAD(P)H and FAD. Antioxid Redox Signal 30, 875-889.

Laplante, M., and Sabatini, D.M. (2012). mTOR signaling in growth control and disease. Cell 149, 274-293.

Lay, K., Kume, T., and Fuchs, E. (2016). FOXC1 maintains the hair follicle stem cell niche and governs stem cell quiescence to preserve long-term tissue-regenerating potential. Proc Natl Acad Sci U S A 113, E1506-1515.

Li, J.L., Goh, C.C., Keeble, J.L., Qin, J.S., Roediger, B., Jain, R., Wang, Y., Chew, W.K., Weninger, W., and Ng, L.G. (2012). Intravital multiphoton imaging of immune responses in the mouse ear skin. Nat Protoc 7, 221-234.

Llufrio, E.M., Wang, L., Naser, F.J., and Patti, G.J. (2018). Sorting cells alters their redox state and cellular metabolome. Redox Biol 16, 381-387.

Loewith, R., and Hall, M.N. (2011). Target of rapamycin (TOR) in nutrient signaling and growth control. Genetics 189, 1177-1201.

Majmundar, A.J., Wong, W.J., and Simon, M.C. (2010). Hypoxia-inducible factors and the response to hypoxic stress. Mol Cell 40, 294-309.

Means, A.L., Xu, Y., Zhao, A., Ray, K.C., and Gu, G. (2008). A CK19(CreERT) knockin mouse line allows for conditional DNA recombination in epithelial cells in multiple endodermal organs. Genesis 46, 318-323.

Miranda, M., Christofk, H., Jones, L.D., and Lowry, W.E., (2018). Topical inhibition of the electron transport chain can stimulate the hair cycle. J Invest Dermatol 138, 968-972.

Miyamoto, K., Araki, K.Y., Naka, K., Arai, F., Takubo, K., Yamazaki, S., Matsuoka, S., Miyamoto, T., Ito, K., Ohmura, M., et al. (2007). Foxo3a is essential for maintenance of the hematopoietic stem cell pool. Cell Stem Cell 1, 101-112.

37

Muller-Rover, S., Handjiski, B., van der Veen, C., Eichmuller, S., Foitzik, K., McKay, I.A., Stenn, K.S., and Paus, R. (2001). A comprehensive guide for the accurate classification of murine hair follicles in distinct hair cycle stages. J Invest Dermatol 117, 3-15.

Nowak, J.A., Polak, L., Pasolli, H.A., and Fuchs, E. (2008). Hair follicle stem cells are specified and function in early skin morphogenesis. Cell Stem Cell 3, 33-43.

Oburoglu, L., Tardito, S., Fritz, V., de Barros, S.C., Merida, P., Craveiro, M., Mamede, J., Cretenet, G., Mongellaz, C., An, X., et al. (2014). Glucose and glutamine metabolism regulate human hematopoietic stem cell lineage specification. Cell Stem Cell 15, 169-184.

Peyssonnaux, C., Boutin, A.T., Zinkernagel, A.S., Datta, V., Nizet, V., and Johnson, R.S. (2008). Critical role of HIF-1alpha in keratinocyte defense against bacterial infection. J Invest Dermatol 128, 1964-1968.

Quinn, K.P., Sridharan, G.V., Hayden, R.S., Kaplan, D.L., Lee, K., and Georgakoudi, I. (2013). Quantitative metabolic imaging using endogenous fluorescence to detect stem cell differentiation. Sci Rep 3, 3432.

Rezvani, H.R., Ali, N., Nissen, L.J., Harfouche, G., de Verneuil, H., Taieb, A., and Mazurier, F. (2011). HIF-1alpha in epidermis: oxygen sensing, cutaneous angiogenesis, cancer, and non-cancer disorders. J Invest Dermatol 131, 1793-1805.

Riedl, J., Flynn, K.C., Raducanu, A., Gartner, F., Beck, G., Bosl, M., Bradke, F., Massberg, S., Aszodi, A., Sixt, M., et al. (2010). Lifeact mice for studying F-actin dynamics. Nat Methods 7, 168-169.

Robinson, M.M., McBryant, S.J., Tsukamoto, T., Rojas, C., Ferraris, D.V., Hamilton, S.K., Hansen, J.C., and Curthoys, N.P. (2007). Novel mechanism of inhibition of rat kidney-type glutaminase by bis-2-(5-phenylacetamido-1,2,4-thiadiazol-2-yl)ethyl sulfide (BPTES). Biochem J 406, 407-414.

Sarbassov, D.D., Guertin, D.A., Ali, S.M., and Sabatini, D.M. (2005). Phosphorylation and regulation of Akt/PKB by the rictor-mTOR complex. Science 307, 1098-1101.

Scarpulla, R.C., Vega, R.B., and Kelly, D.P. (2012). Transcriptional integration of mitochondrial biogenesis. Trends Endocrinol Metab 23, 459-466.

Schwaiger, M., Rampler, E., Hermann, G., Miklos, W., Berger, W., and Koellensperger, G. (2017). Anion-Exchange Chromatography Coupled to High-Resolution Mass Spectrometry: A Powerful Tool for Merging Targeted and Non-targeted Metabolomics. Anal Chem 89, 7667-7674.

Shyh-Chang, N., Daley, G.Q., and Cantley, L.C. (2013). Stem cell metabolism in tissue development and aging. Development 140, 2535-2547.

Stacpoole, P.W. (1989). The pharmacology of dichloroacetate. Metabolism 38, 1124-1144.

Stegen, S., Laperre, K., Eelen, G., Rinaldi, G., Fraisl, P., Torrekens, S., Van Looveren, R., Loopmans, S., Bultynck, G., Vinckier, S., et al. (2019). HIF-1alpha metabolically controls collagen synthesis and modification in chondrocytes. Nature 565, 511-515.

Sterky, F.H., Lee, S., Wibom, R., Olson, L., and Larsson, N.G. (2011). Impaired mitochondrial transport and Parkin-independent degeneration of respiratory chain-deficient dopamine neurons in vivo. Proc Natl Acad Sci U S A 108, 12937-12942.

Sun, R.C., and Denko, N.C. (2014). Hypoxic regulation of glutamine metabolism through HIF1 and SIAH2 supports lipid synthesis that is necessary for tumor growth. Cell Metab 19, 285-292.

38

Tassone, B., Saoncella, S., Neri, F., Ala, U., Brusa, D., Magnuson, M.A., Provero, P., Oliviero, S., Riganti, C., and Calautti, E. (2017). Rictor/mTORC2 deficiency enhances keratinocyte stress tolerance via mitohormesis. Cell Death Differ 24, 731-746.

Vagnozzi, A.N., Reiter, J.F., and Wong, S.Y. (2015). Hair follicle and interfollicular epidermal stem cells make varying contributions to wound regeneration. Cell Cycle 14, 3408-3417.

Varone, A., Xylas, J., Quinn, K.P., Pouli, D., Sridharan, G., McLaughlin-Drubin, M.E., Alonzo, C., Lee, K., Munger, K., and Georgakoudi, I. (2014). Endogenous two-photon fluorescence imaging elucidates metabolic changes related to enhanced glycolysis and glutamine consumption in precancerous epithelial tissues. Cancer Res 74, 3067-3075.

Wick, A.N., Drury, D.R., Nakada, H.I., and Wolfe, J.B. (1957). Localization of the primary metabolic block produced by 2-deoxyglucose. J Biol Chem 224, 963-969.

Wong, J.M., Malec, P.A., Mabrouk, O.S., Ro, J., Dus, M., and Kennedy, R.T. (2016). Benzoyl chloride derivatization with liquid chromatography-mass spectrometry for targeted metabolomics of neurochemicals in biological samples. J Chromatogr A 1446, 78-90.

Xiang, Y., Stine, Z.E., Xia, J., Lu, Y., O'Connor, R.S., Altman, B.J., Hsieh, A.L., Gouw, A.M., Thomas, A.G., Gao, P., et al. (2015). Targeted inhibition of tumor-specific glutaminase diminishes cell-autonomous tumorigenesis. J Clin Invest 125, 2293-2306.

Figure 1 Click here to access/download;Figure;FIgure 1-FINAL.tif

Figure 2 Click here to access/download;Figure;Figure 2 -FINAL-01.tif

Figure 3 Click here to access/download;Figure;Figure 3 -FINAL-01.tif

Figure 4 Click here to access/download;Figure;Figure 4 -FINAL-01.tif

Figure 5 Click here to access/download;Figure;Figure 5-FINAL.tif

Figure 6 Click here to access/download;Figure;Figure 6 -FINAL.tif

Figure 7 Click here to access/download;Figure;Figure 7-FINAL-01.tif

KEY RESOURCES TABLE Kim et al.

REAGENT or RESOURCE SOURCE IDENTIFIER

Antibodies

Monoclonal rat anti-CD34-eFluor660

Thermo Fisher Scientific Cat.#50-0341-82

Monoclonal rat anti-CD49f-PE/Cyanine7

Biolegend Cat.#313622

7AAD Thermo Fisher Scientific Cat.#A1310

Fixable Viability Dye eFluor506 Thermo Fisher Scientific Cat.#65-0866-14

Annexin V-PE Biolegend Cat.#640908

Monoclonal rabbit anti-Rictor Cell Signaling Cat.#2114

Polyclonal rabbit anti-AKT Cell Signaling Cat.#9272

Polyclonal rabbit anti-AKT-pS473

Cell Signaling Cat.#9271

Monoclonal rabbit anti-S6-pS240/244

Cell Signaling Cat.#5364

Polyclonal rabbit anti-beta-actin Sigma-Aldrich Cat.#MAB1501 and A2066

Polyclonal rabbit anti-CD34 Manuel Koch, University of Cologne

N/A

Monoclonal rat Anti-CD34 Invitrogen Cat.#14-0341-82

Polyclonal rabbit anti-K16 Abcam Cat.#ab76414

Polyclonal rabbit anti-K6 Biolegend Cat.#905701

Monoclonal mouse anti-Sox9 Santa Cruz Cat.#sc-166505

Monoclonal mouse anti-Periostin Santa Cruz Cat.#sc-398631

Monoclonal mouse anti-BrdU BD Biosciences Cat.#555627

Polyclonal rabbit anti-Hif1 Novus Biologicals Cat #NB100-479

Chemicals, Peptides, and Recombinant Proteins

Insulin Sigma-Aldrich Cat.#11882

EGF Sigma-Aldrich Cat.#E9644

Transferrin Sigma-Aldrich Cat.#T5158

Phosphoethanolamine Sigma-Aldrich Cat.#P0503

Ethanolamine Sigma-Aldrich Cat.#E0135

Hydrocortisone Calbiochem Cat.#386698

L-Glutamine Thermo Fisher Scientific Cat.#A2916801

D-Glucose Sigma Cat.#80-99-7

Y27632 Miltenyi Biotec Cat.#130-104-169

Mouse recombinant VEGF Miltenyi Biotec Cat.#130-094-086

Human recombinant FGF-2 Miltenyi Biotec Cat.#130-093-838

AKT inhibitor Merck Cat.#124005

CB-839 Cayman Chemical Cat.#22038

Sodium dichloroacetate Sigma-Aldrich Cat.#347795

UK5099 Cayman Chemical Cat.#16980

Antimycin A Sigma-Aldrich Cat.#A8674

Rotenone Sigma-Aldrich Cat.#R8875

Alpha-ketoglutaric acid Sigma-Aldrich Cat.#K1128

U-13C6-D-Glucose Cambridge Isotope Laboratories Cat.#CLM-1822-H-0.1 13C5-L-Glutamine Cambridge Isotope Laboratories Cat.#CLM-1396-1

BrdU BD Pharmingen Cat.#550891

BPTES Cayman Chemical Cat.#19284

Glow-in-the-dark hair dye Manic Panic, Hot Pink N/A

H2O CHROMASOLV LC-MS Ultra

Honeywell Riedel-de Haen Cat.#14263

Acetonitrile OPTIMA LC/MS Fisher Scientific Cat.#A955

Methanol OPTIMA LC/MS Fisher Scientific Cat.#A456

Key Resource Table

Isopropanol OPTIMA LC/MS Fisher Scientific Cat.#A461

Formic acid Sigma-Aldrich Cat.#5438040100

Ammonium carbonate VWR Cat.#21217.260

Sodium carbonate BioXtra Sigma-Aldrich Cat.#S7795

Benzoyl chloride ACS 99% Sigma-Aldrich Cat.#259950

MSTFA Silylation Reagent Macherey-Nagel Cat.#701270.201

Methoxyamine hydrochloride Sigma-Aldrich Cat.#226904

U-13C15N Amino acid Mix Std Cambridge Isotope Laboratories Cat.#MSK-A2-1.2

ATP-13C10 sodium salt solution Sigma-Aldrich Cat.#710695

Citric acid-2,2,4,4-d4 Sigma-Aldrich Cat.#485438

Critical Commercial Assays

Click-iT Plus EdU Alexa Fluor488 Flow Cytometry Assay Kit

Thermo Fisher Scientific Cat.#C10633

Agilent Seahorse XF Cell Mito Stress Test Kit

Agilent Technologies Cat.#103015-100

Agilent Seahorse XF Glycolysis Stress Test Kit

Agilent Technologies Cat.#103020-100

Qubit dsDNA HS Assay Kit Thermo Fisher Scientific Cat.#Q32851

RNeasy Plus Mini Kit Qiagen Cat.#74104

RNeasy Plus Micro Kit Qiagen Cat.#74004

SuperScript IV VILO Master Mix Thermo Fisher Scientific Cat.#11755-250

DyNAmo Color Flash SYBR Green Mix

Thermo Fisher Scientific Cat.#F416

Pierce BCA Protein Assay Thermo Fisher Scientific Cat.#23227

HypoxyprobeKit Hypoxyprobe Inc Cat.# HP1-100Kit

ERCC Spike-in Mix Invitrogen Cat.# 4456740

Experimental Models: Organisms/Strains

Mouse: C57BL/6 J Jackson Laboratories JAX: 000664

Mouse: B6.RictorTm1loxP -Tg(Krt14-Cre)1Cgn/J

Bentzinger et al., 2008 Hafner et al., 2004

N/A

Mouse: B6.RictorTm1loxP -Krt19tm1(cre/ERT)Ggu/J

Bentzinger et al., 2008 Means et al., 2008

N/A

Mouse: B6-Tg(CAG-EGFP)1Osb/J

Riedl et al., 2010 N/A

Oligonucleotides

Primers for qPCR, see Table S1 This paper N/A

Software

Fiji Schindelin et al., 2012 http://imagej.net/Fiji

Graphpad Prism 8 GraphPad N/A

TraceFinder v.4.1 Thermo Fisher Scientific N/A

FreeStyle 1.6 Thermo Fisher Scientific N/A

Mass Lynx v4.1 SCN 950 Waters N/A

TargetLynx XS v 4.1 SCN 959 Waters N/A

Ingenuity Pathway Analysis Qiagen N/A

Panther Mi et al, 2019 Pantherdb.org

FlowJo Becton, Dickinson and Company Flowjo.com

Leica LAS X Leica N/A

BD FACSDivaTM Software Becton, Dickinson and Company N/A

Supplementary Figure Legends

Supplementary Fig. S1 (Related to Fig. 1) Characterization of in vivo and 3C organoid HFSCs

(A) Representative immunofluorescence images of 3C cell cultures grown for 12 days

and stained with antibodies against Periostin and Keratin 6 (top row), Periostin and

Keratin 16 (2nd row), CD34 and Sox9 (3rd row) or Periostin and CD34 (bottom row).

Scale bars 50 µm.

(B) Freshly isolated FACS-purified epidermal cells: CD34+/α6+ (HFSCs), CD34-

/α6+/Sca1- (HF progenitors), CD34-/α6+/Sca1+ (IFE progenitors) and 3C cultured,

FACS purified 3C-CD34+/α6+ and 3C-CD34-/α6+ cells were subjected for RT-qPCR

analysis of ORS (Sox9, Barx2) and IFE (Il1rs2, Ly6d) identity genes (mean ±SEM; n=4

mice/condition; *P< 0.05, ANOVA/Fisher’s).

(C) FACS purified 3C-CD34+/α6+ and 3C-CD34-/α6+ cells were subjected for RT-

qPCR analysis of a panel of wound-induced genes (mean ±SEM; n=4 mice/condition;

ns=not significant, Mann-Whitney).

(D) Representative images and quantification of colonies formed by FACS purified 3C-

CD34+/α6+ and 3C-CD34-/α6+ cells plated on feeder cells (n=6 cultures from 2 mice;

**P< 0.0025, Student’s t-test).

(E) Freshly isolated epidermal cells and 3C cultures grown for 14 days were FACS

sorted into pure populations of CD34+/α6+ and CD34-/α6+ for qPCR analysis of genes

involved in mitochondrial biogenesis (mean ±SD; n=4 mice/condition; *P< 0.05, Mann

Whitney).

(F) Relative amounts of key glycolysis metabolites quantified from CD34+/α6+ and

CD34-/α6+ cells by LC-MS, normalized to CD34- cells (mean ±SD; n=4 mice/condition).

Supplementary Fig. S2 (Related to Fig. 2) Redox metabolite ratios from HFSCs and progenitors

(A-C) LC-MS quantification of key redox metabolite ratios from FACS-purified CD34+/-

cells (mean ±SD; n=4 mice/condition; **P< 0.05, * P< 0.05, Mann Whitney).

Supplementary Fig. S3 (Related to Fig. 3) Effects of low pO2 on HFSCs

Supplemental Text and Figures

(A) qRT-PCR analysis of key HFSC signature genes from FACS purified CD34+/- cells

from 3C cultures in 20% or 2% O2 (mean ±SD, n=3 mice/condition; *p < 0.05, Mann-

Whitney, all statistical comparisons are to 20% O2 CD34-).

(B) Quantification of EdU incorporation and AnnexinV staining by flow cytometry. 3C

cultures in 20% or 2% O2 were incubated with 10µM EdU for 24h. (n=3-4 mice).

(C) Work flow and representative immunofluorescence images of 3C cultures stained

with antibodies against CD34 and visualized for GFP expression. GFP+/CD34+/α6+

and GFP-/CD34-/α6+ cells from LifeAct-GFP and wild type mice were FACS purified,

mixed in 1:1 ratio and cultured for 12 days under 20% or 2% O2 before fixation and

immunofluorescence analyses (n=3 mice/condition).

Supplementary Fig. S4 (Related to Fig. 4) Low pO2 and metabolism control HFSC state

(A) Representative western blot analysis of HIF1α from 3C organoids subjected to 12h

of 20% and 2% O2.

(B) RT-qPCR analysis of key HIF1α target genes from 3D-3C cultured cells subjected

to 12h of 20% and 2% O2 followed by FACS-purification of CD34+/- cells (mean ±SD,

n=4 mice/condition; *P< 0.05 **P<0.0097, ANOVA/Fisher’s).

(C) FACS analyses of HFSC content from 10-day 3C cultured cells treated with 100

µM CoCl2 for 3 days (n= 4 mice/condition).

(D) Absolute numbers of viable cells from 3C cultures treated with ETC inhibitors

Antimycin A (AA) or Rotenone (Rot) (mean±SD; n=3 mice/condition).

(E) FACS analyses of HFSC content from 10-day 3C cultured cells treated with 350

µM UK5099 for 2 days (n= 4 mice/condition).

(F) FACS analyses of HFSC content from 10-day 3C cultured cells treated with 100

µM 2-DG for 2 days (n= 3 mice/condition).

(G) Quantification of HFSC content from 3C cultures where the culture medium was

supplemented with either 15 mM glucose or 15 mM galactose for 12 days (n=3 mice).

(H) Quantification of EdU incorporation from 10-day 3C cultured cells treated with FX11

for 2 days. EdU was added 12 h prior to sample collection (n = 4 mice/condition).

(I) LC/MS quantification of glutamate and glutamine concentrations from FACS-purified

CD34+/- cells (mean±SD; n = 3 mice/condition; *P< 0.05; Mann Whitney).

Supplementary Fig. S5 (Related to Fig. 5) TORC1 and TORC2 in HFSC state regulation

(A) RT-qPCR analyses of key HFSC signature genes from CTR and RictorEKO cells

cultured for 14 days in either 20% or 2% O2 (mean±SD; n=3 mice condition; *p < 0.02,

**p < 0.005, ns=not significant; ANOVA/Holm-Sidak).

(B) Representative immunofluorescence images of cells isolated from control (CTR)

and RictorEKO mice, 3C cultured for 12 days, and stained with antibodies against

Periostin and Keratin 6, Periostin and Keratin 16 or Periostin and Sox9. Scale bars 50

µm.

(C) Representative immunofluorescence images of FACS-purified CD34+/α6+ and

CD34-/α6- cells isolated from CTR and RictorEKO mice, 3C cultured for 7 days and

stained with antibodies against Periostin and Keratin 6. Scale bars 50 µm.

(D) FACS analyses of HFSC content from 10-day 3C cultured cells treated with

indicated concentrations of rapamycin for 2 days (n=3 mice/condition).

(E) FACS analyses of HFSC content from 10-day 3C cultured cells treated with SC79

for 3 days to activate Akt (n=4 mice/condition; P=0.052, Mann-Whitney).

Supplementary Fig. S6 (Related to Fig. 6) Glutamine metabolism controls HFSC state

(A) Quantification of EdU incorporation from 10-day 3C cultured control (CTR) and

RictorEKO cells treated with 20 nM CB839 glutaminase inhibitor for 2 days. EdU was

added 24 h prior to sample collection and analyzed by flow cytometry (n = 4 (CTR)/3

(RictorEKO) mice; *P<0.015, ANOVA/Holm-Sidak).

(B) Flow cytometry analyses of Annexin-positive apoptotic cells from 10-day 3C

cultured control (CTR) and RictorEKO cells treated with 20 nM CB839 for 2 days (n = 4

(CTR)/3 (RictorEKO) mice).

(C) FACS-purified CD34+ cells from RicEKO and CTR mice cultured in 3C for 3 days

with or without CB839 after which CD34+ cells (%) were analyzed via FACS (n=4

mice/condition). ns=not significant

Supplementary Fig. S7 (Related to Fig. 7) Rictor regulates HFSC and hair follicle cycling

(A) Representative images of H/E stained control (CTR) and RictorEKO skin tissue at

P69. Scale bars 200 µm.

(B) Representative images and quantification of back skin whole mounts stained for

CD34 (magenta) and DAPI (blue) to identify bulges. Note two bulges in CTR but not in

and RictorEKO hair follicles (white arrows; n=4 mice/genotype; *p=0.0286, Mann-

Whitney, scale bars 100 µm).

(C) Representative UV and brightfield images of back skin of CTR and RictorEKO mice

dyed with UV-light reactive hair dye during their first telogen hair cycle (P 21) and

imaged at indicated time points (n=3 mice/genotype).

(D) Quantification of BrdU-positive cells in ORS control (CTR) and RictorEKO mice (n=4

mice/genotype).

(E) Representative images of anagen hair follicles stained for Barx2 and CD34 to mark

the hair follicle bulge and ORS. Scale bars 100 µm.

(F, G) Representative images (F) and quantification (G) of Tunel staining to identify

apoptotic cells (white arrows) in catagen hair follicles shows no difference between

CTR RicEKO (n=3 mice/genotype; scale bars 100 µm).

(H) RT-qPCR analyses of Rictor gene expression in FACS-purified CD34+/a6+ and

CD34-/a6+ cells from CTR and Rictor-K19-CreERT mice (mean±SD; n = 4

mice/genotype; *P< 0.05; Ratio paired t-test).

(I, J) Representative images (I) and quantification (J) of hair cycle stage in back skin

of mice entering second postnatal telogen (n = 5 mice/genotype).

(K, L) Representative images of H/E staining (K) and quantification (L) of CTR and

Rictor-K19-CreERT hair follicles with 2 bulges (n = 4 mice; *P< 0.05; Mann Whitney;

scale bars 100 µm).

(M) Representative images of Hypoxyprobe DAB staining in telogen and anagen hair

follicles shows intense signal in bulge and upper ORS of anagen hair follicles (n=2

mice/stage; scale bars 100 µm).

(N) Quantification of hair follicle (HF) density of CTR and RicEKO mice at indicated time

points (n= 3-4 mice/time point).

(O) Quantification of CD34+ HFSCs in CTR and RicEKO mice at P52 (n= 7 mice (CTR)/6

mice (RicEKO)).

0

1

2

3

4

3PG Pyruvate Lactate Phosphoenol Pyruvate

A

TFAM PCG1a0

0.2

0.4

0.6

0.8

1.0

Fol

d C

hang

e

*

*

TFAM PGC1a

* *

Periostin K6 DAPI Merge

Periostin K16 DAPI Merge

CD34 Sox9 DAPI Merge

Periostin CD34 DAPI Merge

In vivo 3C cultureE

0

0.2

0.4

0.6

0.8

1.0

Fol

d C

hang

e

CD34- α6+

CD34+ α6+

FCD34- α6+

CD34+ α6+

Acetyl CoA

SUPPLEMENTARY FIGURE S1

Sox9 Barx2 Il1r2 Ly6d0

2

4

6

10

15

20

25

Fol

d ch

ange

CD34+/α6+ (HFSC)

CD34-/α6+/Sca1- (HF)

CD34-/α6+/Sca1+ (IFE)

3C-CD34+/α6+

3C-CD34-/α6+

ORS IFE

Tgln2 Anxa2 Thbs Tnc0

0.5

1.0

1.5

2.0

Fol

d ch

ange

3C-CD34+/α6+3C-CD34-/α6+

BC

*

*

D

3C-C

D34

+/α

6+3C

-CD

34-/α

6+

0

0.5

1.0

1.5

Colony size3C-CD34+/α6+3C-CD34-/α6+

0

5

10

15

20

25

Colony number

Col

ony

num

ber/

wel

l

Sur

face

are

a (m

m2 )

****

**

**

*

*

ns

ns

ns

ns

In vivo

Fol

d C

hang

e

CD34- CD34+0

0.02

0.04

0.06

0.08

FA

D/F

AD

+N

AD

H

**

A

CD34- CD34+

B C

CD34- α6+CD34+ α6+

AMP/ATP ATP/ADP0

0.5

1.0

1.5

Rat

io

SA

M/S

AH

Rat

io

0

20

40

60

*

*

*

SUPPLEMENTARY FIGURE S2

In vivo

0

20

40

60

80

100

EdU

+ c

ells

(%

)

0

2

4

6

8

10

Ann

exin

+ c

ells

(%

)

A B

20% O2 2% O2 20% O2 2% O2 20% O2 2% O220% O2 2% O2

0

20

40

60

80

100

EdU

+ C

D34

+ c

ells

(%

)

0

20

40

60

80

100

EdU

+ C

D34

- ce

lls (

%)

Dkk3 Id2 Nfatc1 Tcf30

2

4

6

8

10

Fol

d C

hang

e

C

20%

O2

2% O

2

FACS purification

GFP-CD34− α6+

Mix 1:1

3C culture 14 days

ImmunofluorescenceGFP / CD34

GFP+CD34+ α6+

GFP CD34 DAPI Merge

20% O2 / CD34- α6+

20% O2 / CD34+ α6+

2% O2 / CD34- α6+2% O2 / CD34+ α6+

*

*

*

**

*

**

SUPPLEMENTARY FIGURE S3

BA

Glutamate Glutamine0

1

2

3

4

5

Fol

d ch

ange

**

E

3C FX110

20

40

60

80

100

Ed

U+

cel

ls (

%)

3C-CD34+ α6+

3C-CD34- α6+

CD34+ α6+

CD34- α6+

3C+UK5099

3C-Ctrl0

20

40

60

80

100

CD

34+ɑ

6+ c

ells

(%

)

*

Hif1α

Actin

pO2:

C

*

D

Fol

d c

hang

e

G

3C AA Rot0

50000

100000

150000

200000

Live

cel

lsF

3C-Glucose

3C-Galactose

0

20

40

60

80

100

CD

34+ɑ

6+ c

ells

(%

)

3C+ 2-DG3C-Ctrl0

20

40

60

80

100

CD

34+ɑ

6+ c

ells

(%

)

H I

SUPPLEMENTARY FIGURE S4

VEGFPGF

GLUT1

PAI1

MM

P2

PDK2GLS

0

0.5

1.0

1.5

2.0

10

20

30

4020% O2 3C-CD34+20% O2 3C-CD34-

2% O2 3C-CD34+2% O2 3C-CD34-

20% 2%

3C-Crtl

3C-CoCl2

CD

34+ɑ

6+ c

ells

(%

)**

****

**

40

60

80

100

**

**

**

**

*

A 3

2

1

0

Fol

d ch

ange

CTR 20% O2

CTR 2% O2

RicEKO 20% O2

RicEKO 2% O2

Sox9 Nfac1 Dkk3

Periostin Keratin 6 DAPI Merge

Periostin Keratin 16 DAPI Merge

Periostin Sox9 DAPI Merge

CT

RR

icE

KO

CT

RR

icE

KO

CT

RR

icE

KO

B

Periostin Keratin 6 DAPI Merge

CT

RR

icE

KO

CT

RR

icE

KO

Pur

ified

CD

34- α

6+ c

ells

Pur

ified

CD

34+

α6+

cel

ls

C

D

CD

34+

α6+

cel

ls (

%)

0

20

40

60

80

3C 1 nM 10 nM

SUPPLEMENTARY FIGURE S5

3C+Rapamycin

40

60

80

100

CD

34+

α6+

cel

ls (

%)

3C 3C+SC79

p=0.0571E

CTRRicEKO

0

20

40

60

80

100

+ C

B83

9

EdU

+ c

ells

(%

) ns

*A B

CD34+ α6+ CD34- α6+ 0

5

10

15CTRRicEKO

Ann

exin

+ c

ells

(%

)

+ C

B83

9

CD34+ α6+ CD34- α6+

+ C

B83

9

+ C

B83

9

nsns

SUPPLEMENTARY FIGURE S6

C

20

40

60

80

100

CTR

RicEKO

CTR + CB839

RicEKO + CB839

ns

Day0 Day3

CD

34+

α6+

cel

ls (

%)

A

C D

CTR Telogen (P69) RICEKO Telogen (P69)

0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

Fol

d ch

ange

CTR

Ric-K19-CreERT

P26

P39

P48

P53

P65

UV Brightfield Merge UV Brightfield Merge

CTR RicEKO

p21 p32 p520

5

10

15

HF

/mm

CTR

M CD34- α6+CD34+ α6+

I

0

50

100

150catagen/anagen

telogen

CTR

Ric-K19-CreERT

Bac

k sk

in a

rea

(%)

CTR

Ric-K19

-Cre

CTR (P50) Ric-K19-CreERT (P50)

CTR Ric-K19-CreER0

20

40

60

80

100

*

((( )))K

CTR

0

10

20

CD

34+

cel

ls (

%)

O

RicEKO

RicEKO

*

0

20

40

60

80

100

HF

with

2 b

ulge

s (%

)

DA

PI C

D34

CTR Telogen (P62) RicEKO Telogen (P62)B

*

CTR RicEKO

0

5

10

15

20

Brd

U+

cells

in O

RS

CTR RicEKO

E

HF

with

2 b

ulge

s (%

)

CTR Catagen (P40) RicEKO Catagen (P52)

DA

PI T

UN

EL

0

5

10

15

TU

NE

L +

cel

ls /

low

er H

F

CTR RicEKO

P46

P46

F

CT

R A

nage

n (P

32)

Ric

EK

O A

nage

n (P

40)

BARX2 DAPISOX9

Bulge

ORS

ORSBulge

*

*

*

*

G

J L

H

SUPPLEMENTARY FIGURE S7

N

Neg

ativ

e C

TR

Hyp

oxyp

robe

(te

loge

n)

Hyp

oxyp

robe

(an

agen

)

ORS

Bulge

Supplementary Table S1 Primers for RT-qPCR (Related to STAR Methods) Gene Forward Sequence Reverse Sequence

S18 GATCCCAGACTGGTTCCTGA GTCTAGACCGTTGGCCAGAA

Actb TCAAGATCATTGCTCCTCCTG TACTTCTGCTTGCTGATCCAC

Gls CACACCCCGTCCCGGACTTTTTC TTAAGAGCAGCTCCCGTAGCATC

Sox9 AGGAAGCTGGCAGACCAGTA TCCACGAAGGGTCTCTTCTC

Dkk3 ATGCTATGCACCCGAGACAG GAACAGCAGGCCTCTTTGGA

Tcf3 CTCAGCAGCAAATCCAAGAGGCAGAG TGGGAAGACGCAGGGCTATCACAAG

Nfatc1 GGTGCTGTCTGGCCATAACT CCAGGGAATTTGGCTTGCAC

Id2 ATCCCCCAGAACAAGAAGGT TGTCCAGGTCTCTGGTGATG

TFAM AAGGATGATTCGGCTCAGG GGCTTTGAGACCTAACTGG

PCG1a AGCCGTGACCACTGACAACGAG GCTGCATGGTTCTGAGTGCTAAG

Rictor GAGAAAGCTGGGCCATCTGA AACCCGGCTGCTCTTACTTC

Vegfa ACGTCAGAGAGCAACATCAC CTGTCTTTCTTTGGTCTGCATTC

Glut1 CTTCCAGTATGTGGAGCAACT CTCGGGTGTCTTGTCACTTT

Pdk2 TACTTCCTGGACCGCTTCTA CAAAGATGAGGGTGTGTTGATTG

Pai1 CTCCACAGCCTTTGTCATCT ATTGTCTCTGTCGGGTTGTG

Mmp2 CAGGGAATGAGTACTGGGTCTATT ACTCCAGTTAAAGGCAGCATCTAC

Pgf TGCTGGTCATGAAGCTGTTC GGACACAGGACGGACTGAAT

Barx2 ACTCAGCTGCAAGTGAAGAC CTGTCCACCTTTAAGGACCATC

Ly6d ACCGTCACCTCAGTGGA CTCTCGTTGCATAGGTCAGTC

Il1r2 CTCCAGCAGTTCCCATAGTT GTGGGATGCGTTTCTGAATG

Anxa2 CAAGACCAAAGGAGTGGATGAG CTGGGCAGGTGTCTTCAATAG

Tagln2 AGAGGTGCAGCAGAAGATTG AAGCTTGCACAGAACCGT

Tnc TGGTCTTCTGATAGCCAACATC GTGTAACTTCTGGCACTCTCTC

Thbs1 CCTGGACTTGCTGTAGGTTATG GTAGGACTGGGTGACTTGTTTC

Ercc1 AGTAGACGGCCATAGCAGGA TTCGGGTCAACATCGAGCAA

Ercc2 AGCTGGCTCTTTAATGGGAGG CCTGCAATGGGTCTCCAACA