Sex-specific role of sensory neuron LKB1 on metabolic stress-induced mechanical hypersensitivity and mitochondrial respiration

Abstract

Pain disorders induce metabolic stress in peripheral sensory neurons by reducing mitochondrial output, shifting cellular metabolism, and altering energy use. These processes implicate neuronal metabolism as an avenue for creating novel therapeutics. Liver kinase B1 (LKB1) mediates the cellular response to metabolic stress by inducing the 5′-adenosine monophosphate activated kinase (AMPK) pathway. The LKB1-AMPK pathway increases energy-producing processes, including mitochondrial output. These processes inhibit pain by directly or indirectly restoring energetic balance within a cell. Although the LKB1-AMPK pathway has been linked to pain relief, it is not yet known which cell is responsible for this property, as well any direct ties to cellular metabolism. To elucidate this, we developed a genetic mouse model where LKB1 is selectively removed from Nav1.8+ pain sensory neurons and metabolically stressed them by fasting for 24 h. We found females, but not males, had neuron-specific, LKB1-dependent restoration of metabolic stress-induced mitochondrial metabolism. This was reflected in mechanical hypersensitivity, where the absence of LKB1 led to hypersensitivity in female, but not male, animals. This discrepancy suggests a sex- and cell-specific contribution to LKB1-dependent fasting-induced mechanical hypersensitivity. Although our data represent a potential role for LKB1 in anti-pain pathways in a metabolic-specific manner, more must be done to investigate these sex differences.

INTRODUCTION

Chronic pain is an ongoing worldwide crisis, with existing treatments failing to address underlying pathways (1, 2). It is thus crucial to research basic cellular mechanisms to develop better therapeutics for pain. Various studies have examined the effects of whole body metabolism on pain; for example, caloric excess enhances and caloric restriction reduces pain states (3). Importantly, whole body metabolism is interconnected with cell metabolism, including the cellular response to metabolic stress induced by fasting, or caloric restriction for long periods. Investigating cellular metabolism represents an attractive avenue for treating pain. In particular, processes along the 5'-adenosine monophosphate-activated kinase (AMPK) pathway represent appropriate cellular targets to treat pain conditions.

Endogenously, fasting activates AMPK in a liver kinase B1 (LKB1)-dependent manner (4). Physiologically, this leads to an increase in catabolic (energy-producing) processes and a decrease in anabolic (energy-consuming) processes. A primary anabolic process AMPK inhibits is protein production, including the ion channels responsible for a hyperactive phenotype in pain states (5). Fasting itself can also lead to the reduction of proteins, including markers of inflammation, associated with hypersensitivity (6–9). Furthermore, downstream AMPK pathways can induce genes that increase mitochondrial activity, alleviating the mitochondrial dysfunction in dorsal root ganglion (DRG) neurons associated with painful neuropathy (10, 11). However, it is unclear how a single metabolic insult can influence cellular metabolism and subsequent effects on pain. This includes a metabolic challenge that we do not believe would normally cause pain-inducing pain in the absence of LKB1 activity.

Although LKB1 has been shown to contribute to axonal development (12, 13), little is known about its direct contributions to energetic regulation in neurons. We hypothesize that the LKB1 pathway in sensory neurons is responsible for directing cellular energy use from a high-energy to a lower-energy state using more fat and protein catabolism. Animals lacking LKB1 in peripheral pain-sensing neurons (Nav1.8^LKB1) will not be able to appropriately respond to metabolic stress induced by a 24 h fast and will experience enhanced pain-like behaviors, measured by hypersensitivity to a mechanical force (14), over their littermate counterparts with normal LKB1 expression (LKB1⁺). Because the AMPK pathway is responsible for affecting energetic balance at the transcriptional and translational levels and has little to no activity in the absence of LKB1 (15), we predict there will not be an immediate whole body hypersensitivity after fasting. Rather, LKB1⁺ animals will respond to metabolic stress via pathways that Nav1.8^LKB1 animals cannot appropriately activate; without these pathways, hypersensitivity will develop. Monitoring animals’ whole body metabolism by measuring weight distinguished effects were due to LBK1 removal in pain-sensing neurons, as opposed to a subpopulation of Nav1.8+ viscera-sensing neurons (16). Initially, activation of these pathways can be predicted by their mitochondrial response. We expect a cellular bioenergetic response to fasting, which we will measure via a Mitostress assay, where LKB1⁺ neurons are able to produce more energy after fasting than their Nav1.8^LKB1 counterparts. Ultimately, our approach of genetic, behavioral, and biochemical experiments elucidates how manipulating the metabolism of an individual cell type can affect hypersensitivity.

MATERIALS AND METHODS

Animals

Animals were group-housed (3–5/cage) in a pathogen-free, temperature- and humidity-controlled facility in shoebox cages with corncob bedding. Standard laboratory diet chow and water were available ad libitum. Animals were on a 12-h light/dark cycle, with dark cycle from 18:00–06:00. Mice with lox-p sites flanking exons 3–6 of the Stk11 gene (LKB1^fl/fl) (FVB:129S6) were ordered from Frederick National Laboratory for Cancer Research (Strain No. 01XN2; RRID:IMSR_NCIMR:01XN2) and crossed with transgenic mice expressing cre recombinase under the Scn10 promotor (Nav1.8cre) (C57/BL6), originally obtained from Professor John Wood (University College London) but commercially available from Infrafronteir (EMMA ID 04582; RRID:IMSR_EM:04582) (17). Our breeding strategy led to conditional knockout animals with one copy of the Nav1.8-cre allele and two copies of the LKB1 floxed gene, referred to as Nav1.8^LKB1. Their littermate controls that normally express LKB1 have two copies of the LKB1 floxed genes without the cre, referred to as LKB1⁺. Animals were weaned at 21–28 days, at which time a tail biopsy was taken. DNA from this biopsy was extracted, and animals were genotyped via PCR reaction and validated loxP-flanked sequences were amplified by PCR. Animal weights were recorded weekly during the light cycle, starting immediately after animals were weaned. All experiments used littermate controls. Both sexes of animal were used in each experiment. Mice were monitored for gross abnormalities (e.g., anophthalmia); if gross abnormalities were found, mice were excluded from the study (<1%). All protocols and experiments were approved by the University of Texas at Dallas Institutional Animal Care and Use Committee, protocol nos. 16-07 (breeding) and 17-10 (experiment). Experiments were performed in accordance with Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines and the National Institutes of Health Guide for the Care and Use of Laboratory Animals (NIH Publications No. 8023, revised 1978).

Genetic validation.

Animals (8–16 wk old) were deeply anesthetized with isoflurane and euthanized by cervical dislocation. Spinal columns were extracted, bisected, and lumbar dorsal root ganglia (L4–6) were extracted from both sides of the spinal column and frozen in liquid nitrogen. Total RNA was extracted using RNA Stat60 (Teltest). cDNA was synthesized using the high-capacity cDNA Kit (Applied Biosystems). TaqMan Primers for Stk11 (LKB1) (ID: Mm00488474_g1) and 18 s (ID: Hs99999901_s1) were purchased from Applied Biosystems. The mRNA contents were normalized to 18 s mRNA levels. All assays were performed using an ABI Prism 7900HT sequence detection system (Applied Biosystems). The relative amounts of all mRNAs were calculated using the ΔΔCT assay to express fold change from LKB1/WT group. Experimenters were blinded to both sex and genotype of animals during data collection.

Immunohistology

Animals (8–16 wk old) were deeply anesthetized with isoflurane and euthanized by cervical dislocation. Spinal columns were extracted, bisected, and lumbar dorsal root ganglia (L4-6) were extracted from both sides of the spinal column and postfixed in ice-cold 4% paraformaldehyde (PFA) for 1 h before being transferred to 30% sucrose overnight. Tissues were then frozen in optimal cutting temperature (OCT) and cut into 10-μm sections on a cryostat (Leica).

Tissue sections were fixed in 4% PFA in PBS for 1 h at room temperature. After three washes with PBS, cells were permeabilized and blocked with buffer (2% normal goat serum, 1% bovine serum albumin, 0.05% Triton-X, 0.05% Tween-20, and 0.05% sodium azide in PBS) overnight at 4°C and incubated with primary antibody solution (2% normal goat serum, 0.05% Triton-X, 0.05% Tween-20, and 0.05% sodium azide in PBS; 1:1,000 calcium chloride was added for IB4 staining) at 4°C overnight (see Table 1 for details on primary antibodies used). After primary antibody incubation, unbound antibodies were removed via washing once in PBS; tissue sections were then incubated with Alexa Fluor (Invitrogen; goat-anti-mouse IgG2a 488 Cat. No. A21131, RRID:AB_2535771; goat anti-mouse IgG1 647 Cat. No. A21240, RRID:AB_2535809) conjugated secondary antibody solution (0.05% Tween-20 in PBS) for 2 h at room temperature. In some instances, tdTomato protein was expressed in Nav1.8cre⁺ tissues; no differences in Nav1.8 antibody and tdTomato were seen, and this protein was used for quantification. Unbound secondary antibodies were removed by washing three times with PBS, and cells were incubated in DAPI (1:5,000 in PBS) for 5 min at room temperature. Excess DAPI was washed off three times with PBS. Slides were mounted with Gelvatol, which was allowed to cure overnight at 4°C before sealing with clear nail polish (Sally Hansen). At least 24 h after sealing, images were taken on a Zeiss Axiobserver 7 epifluorescent microscope (see Table 2 for details). Z-stacks were created in ZEN (Zeiss, version 3.1; RRID:SCR_013672) to determine the profile of neurons in the DRG. The areas of neurons positive for each marker were quantified automatically using CellSens Dimensions software (Olympus, version 1.18; RRID:SCR_014551) and were used to generate histograms. Experimenters were blinded to sex and genotype of animals during data collection.

Table 1.

Antibodies

Antibody	Vendor	Cat. No.	Identifier	Raised in	Working Concentration	Validation
Nav1.8	Neuromab	73-166	RRID:AB_10672261	Mouse (IgG2a)	1:1,000	(21)
NeuN	EMD Millipore Corp.	MAB377	RRID:AB_2298772	Mouse (IgG1)	1:500	(22)
IB4	Invitrogen	I21411	RRID:AB_2314662	N/A	1:500	(23)
DAPI	Sigma	D9542-10mg	N/A	N/A	1:5,000	(24)

All primary antibody solutions were made in buffer; DAPI solution was made in 1× PBS.

Table 2.

Parameters used to image IHC

Antibody	Wavelength, nm	Exposure Time, ms
Nav1.8	488	920
Nav1.8 Tomato	555	68
NeuN	647	420
IB4	488*	340
DAPI	408*	30

^*Alexa fluor conjugated to a secondary antibody was not needed. IHC, immunohistology

Affective Behaviors

Animals were exposed to the following tests on separate days. All experiments used adult animals (2–4 mo) of both sexes. Whenever possible, the same animals were used for all behaviors. Experimenters were blind to genotype of animals. Animals were allowed to acclimate to a behavior room for 30–60 min before testing. White noise was playing during all experiments. Except where noted, experiments were conducted between 07:00 and 09:30, videos were recorded by a camera suspended ∼1 m above the floor and scored by AnyMaze software (Stoelting, Co., version 5.2), and apparatuses were cleaned with 70% ethanol or Rescue chemical between animals. In the elevated plus maze (EPM) and open field (OF) assays, the experimenter was obscured from animals’ view by the use of opaque black curtains hung ∼1 m above the floor. Experimenters were blinded to genotype of animals during behavioral recordings and both genotype and sex of animals during data analysis. If an animal was excluded from one test for whatever reason, it was removed from all tests (<1%).

Elevated plus maze.

Individual animals were placed in the center of an elevated plus maze consisting of two open and two closed arms (500 mm) and allowed to explore the apparatus for 360 s, starting once the experimenter was out of view of the camera. Only behaviors in the open or closed arms of the apparatus were considered to be directly related to anxiety-like behaviors; as by definition, animals in the center zone were neither in the open nor closed zones (18), these behaviors were not scored. Animals were excluded from analysis if they fell off the apparatus at any time, were noted as not moving for over 50% of the experiment or stayed in the center of the apparatus for over 50% of the experiment (>180 s) (<1%).

For fasting studies, food was removed 24 h before testing. In brief, experimenters could not be blinded to fasted state of animals during behavioral recordings; however, during data analysis, experimenters were blinded to fasted status in addition to sex and genotype.

Open field.

Individual animals were placed in the center of a 600 mm × 600 mm open field box and allowed to explore for 360 s, starting once the experimenter was out of the camera’s field of view. Animals were excluded from analysis if it was determined they were not moving over 50% of the time in the apparatus (>180 s) (<1%).

Forced swim test.

A modified version of the forced swim test (FST) (19) was used. In brief, individual animals were placed in a ∼25-cm deep by 20-cm diameter clear Plexiglas cylinder ∼2/3 full of 23°C–25°C water and allowed to swim for 360 s before being dried with a paper towel and returned to their home cage, which was kept warm via an overhead heat lamp and a heating pad placed underneath the cage. If an animal was immobile (no movement from any paw or tail) for the full first 60 s of the test, it was removed from the apparatus and excluded from study (<1%). Animals housed in the same cage were allowed to use the same Plexiglas cylinder (no more than 5 animals/cylinder, with separate sexes being tested in different cylinders). Fecal matter was removed between mice. Additional cleaning of apparatuses was not performed during the experiment. Videos were recorded by a laptop webcam pointed at the side of the FST apparatus. These videos were then hand-scored by an experimenter blind to the animal’s genotype to determine the time immobile. Animals that were immobile for over 60 s were excluded from study (<1%).

Rotarod.

Animals were allowed three 5-min trials on the rotarod apparatus (IITC), with a 15- to 30-min rest between trials. Rarely, trials were deemed invalid (e.g., a lighter animal not appropriately causing the apparatus platform to drop, thus causing an unclear representation of when their trial ended). In these cases, animals were subjected to a fourth trial to take the place of trial 3. Animals that needed more than one retrial were excluded from the data set (<1%). Experiments were conducted from 19:30 to 22:00, using red light to correspond to the animals’ dark cycle.

Fasting Studies

In studies where fasted animals were used, animals were fasted 22–26 h before the start of behavior. Cages of animals (3–5 animals) were randomly assigned to a “fasted” (allowed access to water but no food for 24 h) or “fed” (allowed food and water ad libitum) condition. Food was immediately returned to animals following behavioral testing.

During behavioral recordings, experimenters were blinded to the genotype of animals, however, could not be blinded to sex. As fasted animals met the definition of pathological thinness (loss of 10%–15% of baseline body weight), experimenters could not be blinded to fasted status during behavioral recordings. During data analysis, experimenters were blinded to sex, genotype, and fasted status of animals. Animals were excluded from analysis if during the study they were determined to be mechanically sensitive before fasting (see Mechanical hypersensitivity for additional details) or developed any gross health concerns (including not regaining weight or losing weight after food was returned) (<1%).

Mechanical hypersensitivity.

Testing with Von Frey filaments (Stoelting, Cat. No. 58011) was performed to test animal’s mechanical hypersensitivity. Animals were acclimated to Plexiglas boxes with mesh flooring for 2–7 days before the start of testing. Animal weights were taken immediately before animals were put into boxes. Animals that did not baseline (i.e., responded to a weight below 0.8 g) 2 days in a row were excluded from the experiment. Fasting began immediately after the second baseline (day 0), and mechanical hypersensitivity was recorded for 5 days afterward (days 1–5). Food was returned to animals after Von Frey testing on day 1. Experiments were performed at the end of the animal’s light cycles (14:00–18:00). To determine hypersensitivity at specific timepoints, the up-down method (14) was used to calculate withdrawal thresholds; lower withdrawal thresholds indicate increased hypersensitivity. Withdrawal thresholds were used to calculate area under the curve (AUC), indicating summative changes during a time range; less AUC indicates increased hypersensitivity. For AUC data, timepoints were divided immediately after fasting (“fast,” days 1–2) and after that food was returned and animals were allowed to feed ad libitum (“resolution,” days 2–5).

Seahorse.

After fasting was completed, dorsal root ganglia (DRGs) were aseptically removed and cultured. In brief, animals were deeply anesthetized with 100% isoflurane and euthanized by cervical dislocation. DRGs were extracted and kept in ice-cold Hank’s balanced salt solution (HBSS; Fisher Scientific, Cat. No. 14–170-112) before being enzymatically digested with 1 mg/mL collagenase A (Sigma, Cat. No. 10103586001; made in HBSS) for 20 min at 32°C followed by 1 mg/mL collagenase D (Sigma, Cat. No. 1188866001; 10% papain, Sigma, Cat. No. 10108014001) in HBSS) for 20 min at 32°C. Trypsin inhibitor (1 mg/mL) (Sigma, Cat. No. 10109886001; made in DMEM supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin) was used to stop enzymatic digestion. Cells were then triturated and strained through a 70-μM cell strainer before being plated at 8 × 10⁴ cells on eight-well Seahorse XFp plates (Agilent, Cat. No. 103025-100) coated in poly-d-lysine and laminin in serum-free media (DMEM supplemented with 1% penicillin-streptomycin). Seahorse experiments were performed on a Seahorse XFp Analyzer (Agilent, Cat. No. 7802 A).

Cells were allowed to rest for 4–6 h before undergoing a MitoStress assay (Agilent, Cat. No. 103010-100) with minor modifications from manufacturer protocol. In brief, serum-free medium was replaced with Seahorse DMEM supplemented with 100 μM pyruvate, 100 μM glycine, and 1 μM glucose; cells were allowed to calibrate for 1 h. After calibration, cells were treated with 1.5 μM oligomycin, followed by 1.5 μM carbonyl cyanide-4 (trifluoromethoxy) phenylhydrazone, followed by 0.5 μM rotenone/AA; oxygen consumption rates (OCAR) and extracellular acidification rates (ECAR) were monitored.

After the MitoStress assay, cells were fixed in 4% PFA for 30 min at room temperature; nuclei were then stained with DAPI. Plates were imaged with a Zeiss fluorescent microscope (details in Table 2); DAPI was used to determine cell count for normalization (ImageJ, NIH; RRID:SCR_003070). Initial analysis of Seahorse data was performed in WAVE desktop software (Agilent, version 2.6.1; RRID:SCR_014526). Experimenters were blinded to sex, genotype, and fasted conditions during Seahorse experiments.

Statistics

All statistics were reported by GraphPad/Prism software (GraphPad Software, Inc., version 8.0.0; RRID:SCR_002798). Statistical tests were chosen based on the number of variables considered; one variable (genotype only) used Student’s t test, two variables (genotype and fasted state) used two-way ANOVA, and three variables (genotype, sex, and fasted state) used three-way ANOVA. Post hoc tests were performed when a significant interaction was seen. Tukey’s post hoc test was used when two effects were seen; Sidak’s or Bonferroni’s post hoc tests were used with one effect. Outliers were removed via Grubbs’ test. All data are presented as means ± SE. Details of individual statistical tests are included in figure legends or results section. Statistical significance was defined as P ≤ 0.05.

RESULTS

Removal of LKB1 from Nav1.8+ Neurons Does Not Adversely Affect Development

Gene expression revealed LKB1 was removed from lumbar DRGs (Fig. 1A). Animals were born at expected Mendelian ratios, exhibited no defects prohibiting their maturation, and were fertile. Animals exhibited typical lifespan growth independent of genotype (Fig. 1B) (Table 3). DRGs developed appropriately in both genotypes (Table 4) (total neurons Fig. 1, Cand D, Nav1.8+ small diameter nociceptors Fig. 1, Eand F, and IB4+ medium diameter nociceptors Fig. 1, Gand H).

Figure 1.

LKB1 deletion in peripheral sensory neurons does not have adverse developmental consequences. A: LKB1 was removed from Nav1.8-containing peripheral sensory neurons via the cre-lox system and verified via qPCR analysis, n = 5 or 63 DRGs from 5 to 6 animals; **P < 0.01). B: animals were weighed from time of weaning (4 wk) to time of experiment (15 wk). C–H: to assess neuron development, lumbar dorsal root ganglias (DRGs; levels L4–6) were taken from animals of both sexes and stained for all neurons (NeuN, C and D), small and medium diameter pain-sensing neurons containing Nav1.8 (E and F), and small and medium diameter pain-sensing nonpeptidergic neurons marked by isolectin B4 (IB4, G and H). In representative immunohistology images (D: NeuN; F: Nav1.8; H: IB4), n = 1–3 DRG sections from 2 or 3 animals, scalebar: 50 μm. Rotarod data for total distance traveled (I), maximum speed (J), and latency until fall (K) (n = 11–17 animals, *P < 0.05, **P < 0.01). L: forced swim test data. No significant differences were detected in forced swim test (unpaired t test; n = 13–21 animals). Data are presented as means ± SE. LKB1, liver kinase B1.

Table 3.

Mixed-effects ANOVA results for weight across lifespan

Source of Variation	F (DFn, DFd)	P Value	Significant
Male
Age	F (11, 379) = 279.4	<0.0001	Yes
Genotype	F (1, 60) = 0.8535	= 0.3593	No
Female
Age	F (1.518, 38.09) = 99.78	<0.0001	Yes
Genotype	F (1, 72) = 0.05912	= 0.8086	No

Age, but not genotype, was a significant factor in animal weight.

Table 4.

Two-way ANOVA results for neuronal subpopulation histograms

Neuronal Population	Source of Variation	F (DFn, DFd)	P Value	Significant
Male
NeuN	Size	F (10) = 5.060	=0.0086	Yes
	Genotype	F (1, 10) = 1.340e-033	>0.9999	No
Nav1.8	Size	F (9) = 8.726	=0.0017	Yes
	Genotype	F (1, 9) = 2.259e-033	>0.9999	No
IB4	Size	F (7) = 10.04	=0.0035	Yes
	Genotype	F (1, 7) = 1.405e-031	>0.9999	No
Female
NeuN	Size	F (8) = 35.20	<0.0001	Yes
	Genotype	F (1, 8) = 0.000	>0.9999	No
Nav1.8	Size	F (9) = 8.757	=0.0017	Yes
	Genotype	F (1, 9) = 8.620e-033	>0.9999	No
IB4	Size	F (6) = 4.658	=0.0416	Yes
	Genotype	F (1, 6) = 2.730e-032	>0.9999	No

Animal genotype did not contribute to differences seen in neuronal populations.

In addition to these initial validation experiments, we confirmed deletion of LKB1 from peripheral sensory neurons did not affect gross motor or affective behaviors. Males of either genotype behaved comparably on Rotarod (Fig. 1, I–K) (Table 5). Female Nav1.8^LKB1 exhibited some improvements in gross motor coordination; however, these improvements were limited to the second trial (Fig. 1, I–K) (Table 6). Animals spent similar amounts of time immobile in the forced swim test (Fig. 1L) (Table 7). To confirm no effects of genotype on gross motor or exploratory behaviors, we elected to test animals in the open field (Fig. 2). Animals exhibited no differences in spontaneous locomotion, although a preference for the outer zone of the apparatus developed (Table 8).

Table 5.

Two-way repeated measures ANOVA for male rotarod

Parameter	Source of Variation	F (DFn, DFd)	P Value	Significant
Speed	Interaction	F (2, 58) = 2.427	=0.0972	No
	Trial	F (2, 58) = 12.83	<0.0001	Yes
	Genotype	F (1, 29) = 0.1399	=0.7111	No
Time	Interaction	F (2, 58) = 3.347	=0.0421	Yes
	Trial	F (2, 58) = 13.64	<0.0001	Yes
	Genotype	F (1, 29) = 0.1841	=0.6710	No
Distance traveled	Interaction	F (2, 58) = 4.188	=0.0200	Yes
	Trial	F (1.830, 53.06) = 10.25	=0.0003	Yes
	Genotype	F (1, 29) = 0.09145	=0.7645	No

Trial was determined to cause significant differences in all parameters. Although genotype and trial were identified as causing differences in both time and distance traveled, post hoc tests did not reveal genotype-dependent differences.

Table 6.

Mixed-effects ANOVA for female rotarod

Source of Variation	F (DFn, DFd)	P Value	Geisser–Greenhouse’s Epsilon	Significant
Speed
Trial × Genotype	F (2, 55) = 0.6319	=0.5354		No
Trial	F (1.854, 50.99) = 4.863	=0.0134	0.9271	Yes
Genotype	F (1, 28) = 10.07	=0.0036		Yes
Time
Trial × Genotype	F (2, 55) = 1.022	=0.3665		No
Trial	F (1.830, 50.32) = 5.629	=0.0076	0.9148	Yes
Genotype	F (1, 28) = 10.49	=0.0031		Yes
Distance traveled
Trial × Genotype	F (2, 55) = 2.287	=0.1112		No
Trial	F (1.859, 51.11) = 6.172	=0.0048	0.9293	Yes
Genotype	F (1, 28) = 13.40	=0.0010		Yes

Both trial and genotype were determined to separately influence all parameters. Post hoc tests revealed Nav1.8^LKB1 females performed better than LKB1+ females during the second trial only. LKB1, liver kinase B1.

Table 7.

Unpaired t test results for time immobile in forced swim test

Sex	t, DF	P Value	Significant
Male	2.018, 28	= 0.0533	No
Female	1.136, 32	= 0.2664	No

Animals of different genotypes did not exhibit significantly different behaviors.

Figure 2.

Removal of LKB1 from peripheral sensory neurons does not alter exploratory or locomotor behaviors. A and B: genotype did not affect the number of times an animal entered into either zone (A and B), the total amount of time animals spent in either zone (C and D), the distance animal traveled (E and F), time spent mobile (G and H), or immobile (I and J) in either zone of an open field apparatus. A preference developed for the inner zone of the apparatus (C–J) in all animals. n = 13–21 animals ****P < 0.0001. Data are presented as means ± SE. LKB1, liver kinase B1.

Table 8.

Two-way ANOVA results for open field

Parameter	Source of Variation	F (DFn, DFd)	P Value	Significant
Male
Number of entries	Interaction	F (1, 64) = 0.0006919	=0.9791	No
	Zone	F (1, 64) = 0.02343	=0.8788	No
	Genotype	F (1, 64) = 0.06407	=0.8010	No
Time in zone	Interaction	F (1, 64) = 1.029	=0.3142	No
	Zone	F (1, 64) = 2702	<0.0001	Yes
	Genotype	F (1, 64) = 0.0008396	=0.9770	No
Distance traveled in zone	Interaction	F (2, 96) = 0.8826	=0.4170	No
	Zone	F (2, 96) = 63.54	<0.0001	Yes
	Genotype	F (1, 96) = 3.497	=0.0645	No
Time mobile in zone	Interaction	F (1, 64) = 2.164	=0.1461	No
	Zone	F (1, 64) = 2276	<0.0001	Yes
	Genotype	F (1, 64) = 0.1691	=0.6822	No
Time immobile in	Interaction	F (1, 64) = 1.703	=0.1966	No
zone	Zone	F (1, 64) = 125.6	<0.0001	Yes
	Genotype	F (1, 64) = 1.576	=0.2139	No
Female
Number of entries	Interaction	F (1, 60) = 1.056e-006	=0.9992	No
	Zone	F (1, 60) = 0.04098	=0.8403	No
	Genotype	F (1, 60) = 3.510	=0.0659	No
Time in zone	Interaction	F (1, 60) = 4.708	=0.0340	Yes
	Zone	F (1, 60) = 2503	<0.0001	Yes
	Genotype	F (1, 60) = 8.496e-006	=0.9977	No
Distance traveled in	Interaction	F (2, 90) = 0.2053	=0.8148	No
zone	Zone	F (2, 90) = 152.2	<0.0001	Yes
	Genotype	F (1, 90) = 3.196	=0.0772	No
Time mobile in zone	Interaction	F (1, 60) = 1.132	=0.2915	No
	Zone	F (1, 60) = 1923	<0.0001	Yes
	Genotype	F (1, 60) = 0.2044	=0.6528	No
Time immobile in	Interaction	F (1, 60) = 3.376	=0.0711	No
zone	Zone	F (1, 60) = 57.86	<0.0001	Yes
	Genotype	F (1, 60) = 3.086	=0.0841	No

Both males and females developed a preference for the inner zone of the apparatus; neither sex had any genotype-dependent differences.

LKB1 in Nav1.8+ Contributes to Sensitization in A Sex-Dependent Fashion

After establishing that LKB1 deletion did not result in gross deficiencies, we tested mechanical hypersensitivity after a metabolic stressor. A 24-h fast resulted in all animals losing a significant amount of weight which was regained once food was returned (Fig. 3A, male, Fig. 3B, female) (Table 9).

Figure 3.

Fasting induces hypersensitivity in a sex- and genotype-dependent manner. A and B: animals of the same genotype lost an equivalent amount of weight after a 24-h fast (****P < 0.0001; two-way repeated-measures ANOVA, Tukey’s test). C: fasted male animals developed an initial hypersensitive period independent of genotype; however, Nav1.8^LKB1 maintained impaired recovery from this hypersensitive period (*P < 0.05; **, ##P < 0.01) (two-way repeated measures ANOVA with Tukey’s test). D: fasted Nav1.8^LKB1 females developed hypersensitivity sooner than their LKB1⁺ counterparts (*P < 0.05; **P < 0.01) (two-way repeated-measures ANOVA with Tukey’s test). E: fasting itself produced sex- and genotype-dependent effects. Fasted LKB1+ males, Nav1.8^LKB1 males, and Nav1.8^LKB1 females all had less area under curve (AUC) after the fasting period compared with the recovery period of LKB1+ males (*), LKB1+ females ($), and Nav1.8^LKB1 females (#) who were fed during the entire experiment (*P < 0.05; **, $$P < 0.01, two-way ANOVA with Sidak’s multiple comparisons test). n = 7–11 animals. Downward arrow indicates when food was removed; upward arrow indicates when food was returned. For C and D: *Nav1.8^LKB1 fed vs. Nav1.8^LKB1 fasted; $LKB1⁺ fasted vs. Nav1.8^LKB1 fasted; #LKB1⁺ fed vs. Nav1.8^LKB1 fasted; % LKB1⁺ fed vs. LKB1⁺ fasted. Dotted line indicates average withdrawal thresholds of fed animals. Red background indicates “fast” period, green background indicates “recovery” period. Data are presented as means ± SE. LKB1, liver kinase B1.

Table 9.

Three-way ANOVA for animal weights after fasting

Source of Variation	F (DFn, DFd)	P Value	Significant
Male
Time	F (3.311, 115.9) = 83.85	<0.0001	Yes
Fed	F (1, 35) = 17.80	=0.0002	Yes
Genotype	F (1, 35) = 0.3962	=0.5331	No
Time × Fed	F (5, 175) = 74.34	<0.0001	Yes
Time × Genotype	F (5, 175) = 0.4945	=0.7801	No
Fed × Genotype	F (1, 35) = 0.1326	=0.7180	No
Time × Fed × Genotype	F (5, 175) = 0.9212	=0.4686	No
Female
Time	F (2.344, 75.00) = 56.83	<0.0001	Yes
Fed	F (1, 32) = 1.016	=0.3210	No
Genotype	F (1, 32) = 0.007315	=0.9324	No
Time × Fed	F (5, 160) = 59.73	<0.0001	Yes
Time × Genotype	F (5, 160) = 0.8777	=0.4975	No
Fed × Genotype	F (1, 32) = 0.006152	=0.9380	No
Time × Fed × Genotype	F (5, 160) = 0.7737	=0.5701	No

Matching for “time.” Fasting caused animals to lose weight independent of genotype.

Immediately after fasting, males had lower 50% withdrawal thresholds (Fig. 3C) (Table 10). Interestingly, on days 3–5 after fasting, only Nav1.8^LKB1 animals exhibited any hypersensitivity. Fasted Nav1.8^LKB1 females developed hypersensitivity on day 2 that persisted until the end of the study (Fig. 3D) (Table 10). Although fasted LKB1⁺ females eventually developed hypersensitivity, this did not occur until day 5.

Table 10.

Two-way ANOVA results for fasting and hypersensitivity (50% withdrawal thresholds)

Source of Variation	F (DFn, DFd)	P Value	Significant
Male
Genotype × Time	F (15, 170) = 1.283	=0.2175	No
Genotype	F (3, 34) = 9.976	<0.0001	Yes
Time	F (3.854, 131.0) = 5.006	=0.0010	Yes
Female
Genotype × Time	F (15, 155) = 1.025	=0.4329	No
Genotype	F (3, 31) = 14.21	<0.0001	Yes
Time	F (3.448, 106.9) = 0.7856	=0.5202	No

Although males had both a time- and genotype-sensitive hypersensitivity, female animals developed hypersensitivity in a genotype-dependent manner only.

Both fasted LKB1⁺ and Nav1.8^LKB1 males had less area under curve (AUC) than fed LKB1⁺ males, fed LKB1⁺ females, and fed Nav1.8^LKB1 females. Fasted Nav1.8^LKB1 females had less AUC than fed LKB1⁺ males, fed LKB1⁺ females, and fed Nav1.8^LKB1 females (Fig. 3E) (Table 11).

Table 11.

Two-way ANOVA for area under curve of fasting and refeeding

Source of Variation	F (DFn, DFd)	P Value	Significant
Interaction	F (7, 130) = 0.4607	=0.8613	No
Fast	F (1, 130) = 34.67	<0.0001	Yes
Group	F (7, 130) = 2.927	=0.0071	Yes

“Fast” denotes initial fast (days 1-2) vs. refeed period (days 2-5). “Group” denotes sex, genotype, and feeding status (e.g., LKB1+ male fasted). Fasting alone was shown to have an effect on animal hypersensitivity. LKB1, liver kinase B1.

LKB1 Affects Neuronal Metabolism Differently in Males and Females

Based on behavior data (Fig. 3, C–E), we determined that metabolic differences are most prevalent the day after fasting (day 1). To investigate how neurons use energy immediately after fast, DRGs were taken from fasted and fed animals, cultured, and subjected to an Agilent Seahorse MitoStress assay; a brief schematic of the experiment is represented in Fig. 4A. Male (Fig. 4B) and female (Fig. 4C) cell oxygen consumption rate (OCR) response was used to calculate basal respiration (Fig. 4D), maximal respiration (Fig. 4E), ATP production (Fig. 4F), nonmitochondrial ATP production (Fig. 4G), coupling efficiency (Fig. 4H), proton leak (Fig. 4I), and spare respiratory capacity (Fig. 4J). Table 12 details how parameters were calculated and what they represent. Table 13 lists ANOVA details.

Figure 4.

LKB1-dependent effects of fasting-induced metabolic challenge on neurons. A: outline of Agilent seahorse mitostress test and relevant calculated parameters. B: OCR trends in male animals. C: OCR trends in female animals. D: fasted Nav1.8^LKB1 males and females had lower basal respiration than fasted LKB1⁺ females. E: fasted Nav1.8^LKB1 males and females had lower maximal respiration than their LKB1⁺ female counterparts. F: ATP production was reduced in LKB1⁺ and Nav1.8^LKB1 males as well as Nav1.8^LKB1 females after fasting. This includes fasted LKB1⁺ females producing more ATP than their fed Nav1.8^LKB1 counterparts. G: nonmitochondrial ATP production was decreased in Nav1.8^LKB1 females after fasting. H: coupling efficiency was unchanged in males and females after fasting I: fasting increased proton leak in both female but not male LKB1⁺ animals. J: spare respiratory capacity was lower in Nav1.8^LKB1 females after fasting. *P < 0.05, **P < 0.01; Tukey’s multiple comparisons test. Data were normalized per 1,000 cells; n = 4–8 wells, with 8 × 10⁴ neurons from one animal per well. Data are presented as means ± SE. LKB1, liver kinase B1.

Table 12.

Seahorse parameter calculations

Parameter	Formula	Measures
Basal respiration	(Last OCR measurement before Oligo) − (Minimum OCR after Rot/AA)	Oxygen consumption used to meet cellular ATP demand and resulting from proton leak
Maximal respiration	(Maximum OCR after FCCP) − (Minimum OCR after Rot/AA)	Maximum rate of respiration, caused by mimicking an “energetic challenge” and causing rapid oxidation of substrates
ATP production	(Last OCR measurement before Oligo) − Minimum OCR after Oligo)	ATP required to meet energetic needs of cell
Nonmitochondrial ATP production	Minimum OCR after Rot/AA	ATP produced by enzymes and other cellular processes
Coupling efficiency	(ATP production rate/Basal respiration rate) ×100	Percentage of ATP production related to ATP synthase vs. proton leak
Proton leak	(Minimum OCR after Oligo) − (Nonmitochondrial ATP production)	Basal respiration not coupled to ATP production
Spare respiratory capacity	(Maximal respiration) − (Basal Respiration)	Measurement of the cell’s ability to respond to energetic demand
Spare respiratory capacity (%)	(Maximal Respiration/Basal Respiration) ×100	How closely cell is acting at theoretical maximum

Adapted from Aglient seahorse MitoStress test user manual. FCCP, carbonyl cyanide-4 (trifluoromethoxy) phenylhydrazone; OCR, oxygen consumption rate; oligo, oligomycin, rot/AA, rotenone/antimycin A.

Table 13.

Three-way ANOVA results for Seahorse MitoStress assay

Parameter	Source of Variation	F (DFn, DFd)	P Value	Significant
Non mitochondrial ATP	Fed	F (1, 35) = 0.8700	=0.3574	No
	Sex	F (1, 35) = 0.006163	=0.9379	No
	Genotype	F (1, 35) = 3.099	=0.0871	No
	Fed × Sex	F (1, 35) = 1.565	=0.2193	No
	Fed × Genotype	F (1, 35) = 7.496	=0.0097	Yes
	Sex × Genotype	F (1, 35) = 1.211	=0.2786	No
	Fed × Sex x Genotype	F (1, 35) = 0.4225	=0.5199	No
Basal respiration	Fed	F (1, 35) = 0.01978	=0.8890	No
	Sex	F (1, 35) = 1.334	=0.2559	No
	Genotype	F (1, 35) = 8.574	=0.0060	Yes
	Fed × Sex	F (1, 35) = 2.642	=0.1130	No
	Fed × Genotype	F (1, 35) = 7.290	=0.0106	Yes
	Sex × Genotype	F (1, 35) = 2.494	=0.1233	No
	Fed × Sex × Genotype	F (1, 35) = 1.872	=0.1800	No
Maximal respiration	Fed	F (1, 35) = 0.4832	=0.4915	No
	Sex	F (1, 35) = 0.5232	=0.4743	No
	Genotype	F (1, 35) = 9.936	=0.0033	Yes
	Fed × Sex	F (1, 35) = 0.5743	=0.4536	No
	Fed × Genotype	F (1, 35) = 10.23	=0.0029	Yes
	Sex × Genotype	F (1, 35) = 0.1111	=0.7409	No
	Fed × Sex × Genotype	F (1, 35) = 1.285	=0.2646	No
Proton leak	Fed	F (1, 35) = 0.09471	=0.7601	No
	Sex	F (1, 35) = 0.7618	=0.3887	No
	Genotype	F (1, 35) = 8.976	=0.0050	Yes
	Fed × Sex	F (1, 35) = 0.03636	=0.8499	No
	Fed × Genotype	F (1, 35) = 6.961	=0.0123	Yes
	Sex × Genotype	F (1, 35) = 0.1819	=0.6724	No
	Fed × Sex × Genotype	F (1, 35) = 1.586	=0.2163	No
ATP production	Fed	F (1, 35) = 0.002449	=0.9608	No
	Sex	F (1, 35) = 1.212	=0.2785	No
	Genotype	F (1, 35) = 6.061	=0.0189	Yes
	Fed × Sex	F (1, 35) = 4.599	=0.0390	Yes
	Fed × Genotype	F (1, 35) = 5.393	=0.0262	Yes
	Sex × Genotype	F (1, 35) = 4.773	=0.0357	Yes
	Fed × Sex × Genotype	F (1, 35) = 1.462	=0.2347	No
Spare respiratory capacity	Fed	F (1, 35) = 0.6943	=0.4104	No
	Sex	F (1, 35) = 0.2227	=0.6399	No
	Genotype	F (1, 35) = 8.301	=0.0067	Yes
	Fed × Sex	F (1, 35) = 0.1100	=0.7421	No
	Fed × Genotype	F (1, 35) = 9.179	=0.0046	Yes
	Sex × Genotype	F (1, 35) = 0.03709	=0.8484	No
	Fed × Sex × Genotype	F (1, 35) = 0.8334	=0.3675	No
Spare respiratory capacity %	Fed	F (1, 35) = 0.004663	=0.9459	No
	Sex	F (1, 35) = 0.01693	=0.8972	No
	Genotype	F (1, 35) = 9.694	=0.0037	Yes
	Fed × Sex	F (1, 35) = 0.4667	=0.4990	No
	Fed × Genotype	F (1, 35) = 3.753	=0.0608	No
	Sex × Genotype	F (1, 35) = 2.389	=0.1312	No
	Fed × Sex × Genotype	F (1, 35) = 0.006347	=0.9370	No
Coupling efficiency	Fed	F (1, 35) = 0.05192	=0.8211	No
	Sex	F (1, 35) = 0.001024	=0.9747	No
	Genotype	F (1, 35) = 0.04104	=0.8406	No
	Fed × Sex	F (1, 35) = 3.314	=0.0772	No
	Fed × Genotype	F (1, 35) = 0.1336	=0.7169	No
	Sex × Genotype	F (1, 35) = 2.684	=0.1103	No
	Fed × Sex × Genotype	F (1, 35) = 1.265	=0.2684	No

Genotype and fasting were both determined to have significant effects on mitochondrial function.

Interestingly, an LKB1-dependent effect was seen in females but not males. Fasted Nav1.8^LKB1 females had lower basal (Fig. 4D) and maximal (Fig. 4E) respiration than LKB1⁺ females. Furthermore, fasting decreased ATP production in males of both genotypes and Nav1.8^LKB1 females (Fig. 4F). This includes decreased nonmitochondrial ATP production in Nav1.8^LKB1 females (Fig. 4G). No significant differences in coupling efficiency were seen (Fig. 4H). Although this suggests that animals were able to produce appropriate ATP to mitigate effects of proton leak, it should be noted that fasted Nav1.8^LKB1 females show less proton leak than LKB1⁺ females (Fig. 4I). Furthermore, fasted Nav1.8^LKB1 animals of either sex had lower spare respiratory capacity than LKB1⁺ females (Fig. 4J). These data suggest LKB1 is necessary to regulate neuronal energy use after metabolic stress in females, but not males.

Neither Fasting Nor Deletion of LKB1 in Nav1.8+ Neurons Adversely Affect Anxiety-Like Behavior

To ensure fasting only affected hypersensitivity, we elected to expose animals to the elevated plus maze (EPM) to rule out anxiety-like behaviors as a confound. Animals that enter open arms more, including spending more time and traveling more in this zone, are considered to exhibit less anxiety-like behaviors. We found animals spent similar amounts of time and traveled similar distances in both zones of the apparatus independent of fasted status (Fig. 5) (Table 14). Thus, it is likely our previous results were specific to the effects of fasting on nociception rather than altered affective behaviors.

Figure 5.

Fasting does not induce anxiety-like behaviors. Zones of elevated plus maze are defined as open and closed arms. A: males had no differences in zone entries. B: fasted Nav1.8^LKB1 females entered the closed arms of the apparatus less frequently than their fed counterparts. C: males had no differences in time spent in zone. D: females had no differences in time spent in zone. E: males of all exhibited a preference for the closed arms of the apparatus, indicated by distance traveled in zone (P < 0.0001: ****LKB1⁺ fed; ####Nav1.8^LKB1 fed; $$$$LKB1⁺ fast; %%%%Nav1.8^LKB1 fast). F: Fasted LKB1+ females did not exhibit a zone preference (P < 0.0001: ****LKB1⁺ fed; ####Nav1.8^LKB1 fed; %%%%, Nav1.8^LKB1 fast). Zone entries and time in zone three-way ANOVA with Sidak’s test, distance traveled in zone three-way ANOVA with Tukey’s multiple comparisons test. n = 8–20 individual animals. Data are presented as means ± SE. LKB1, liver kinase B1.

Table 14.

Three-way ANOVA results for elevated plus maze

Parameter	Source of Variation	F (DFn, DFd)	P Value	Significant
Male
Number of entries	Zone	F (1, 63) = 3.017	=0.0873	No
	Fed	F (1, 63) = 7.018	=0.0102	Yes
	Genotype	F (1, 63) = 0.5820	=0.4484	No
	Zone × Fed	F (1, 63) = 0.6611	=0.4192	No
	Zone × Genotype	F (1, 63) = 1.980	=0.1643	No
	Fed × Genotype	F (1, 63) = 0.2822	=0.5971	No
	Zone × Fed × Genotype	F (1, 63) = 3.163	=0.0801	No
Time in zone	Zone	F (1, 63) = 1.597	=0.2110	No
	Fed	F (1, 63) = 10.26	=0.0021	Yes
	Genotype	F (1, 63) = 0.006224	=0.9374	No
	Zone × Fed	F (1, 63) = 0.04719	=0.8287	No
	Zone × Genotype	F (1, 63) = 0.07462	=0.7856	No
	Fed × Genotype	F (1, 63) = 0.6644	=0.4181	No
	Zone × Fed × Genotype	F (1, 63) = 2.515	=0.1178	No
Distance traveled in zone	Zone	F (1.567, 97.18) = 292.9	<0.0001	Yes
	Fed	F (1, 62) = 3.921	=0.0521	No
	Genotype	F (1, 62) = 0.008870	=0.9253	No
	Zone × Fed	F (2, 124) = 1.269	=0.2846	No
	Zone × Genotype	F (2, 124) = 0.6986	=0.4992	No
	Fed × Genotype	F (1, 62) = 0.5526	=0.4601	No
	Zone × Fed × Genotype	F (2, 124) = 2.288	=0.1058	No
Female
Number of entries	Zone	F (1, 57) = 13.94	=0.0004	Yes
	Fed	F (1, 57) = 14.29	=0.0004	Yes
	Genotype	F (1, 57) = 2.239	=0.1401	No
	Zone × Fed	F (1, 57) = 0.3872	=0.5362	No
	Zone × Genotype	F (1, 57) = 0.1264	=0.7235	No
	Fed × Genotype	F (1, 57) = 0.07546	=0.7845	No
	Zone × Fed × Genotype	F (1, 57) = 0.8113	=0.3715	No
Time in zone	Zone	F (1, 57) = 7.686	=0.0075	Yes
	Fed	F (1, 57) = 0.04269	=0.8370	No
	Genotype	F (1, 57) = 0.09304	=0.7615	No
	Zone × Fed	F (1, 57) = 0.4405	=0.5096	No
	Zone ×Genotype	F (1, 57) = 0.7282	=0.3971	No
	Fed × Genotype	F (1, 57) = 0.1728	=0.6792	No
	Zone × Fed × Genotype	F (1, 57) = 1.243	=0.2696	No
Distance traveled in zone	Zone	F (1.626, 92.71) = 228.8	<0.0001	Yes
	Fed	F (1, 57) = 16.56	=0.0001	Yes
	Genotype	F (1, 57) = 5.870	=0.0186	Yes
	Zone × Fed	F (2, 114) = 8.537	=0.0004	Yes
	Zone × Genotype	F (2, 114) = 1.951	=0.1469	No
	Fed × Genotype	F (1, 57) = 0.1672	=0.6842	No
	Zone × Fed × Genotype	F (2, 114) = 0.6425	=0.5279	No

Male animals developed a slight preference for closed arms of the apparatus, but no genotype or fasting-dependent differences were noted. Initial effects of both fasting and genotype were seen in female animals; however, post hoc tests revealed that fasted LKB1+ females did not develop a zone preference. LKB1, liver kinase B1.

DISCUSSION

Although males and females are established to have whole body metabolic differences, it is still unclear how cell metabolism drives whole body metabolism, especially in the context of pain. Although previous studies have shown that caloric restriction possesses pain-relieving qualities, it is not known which cells are directly responsible for these effects. This includes any sexual dimorphisms in cell metabolism and antinociceptive metabolic processes. The present study examines the role of sensory neuron-specific LKB1, and how an otherwise innocuous metabolic challenge can induce nociceptive behaviors. Interestingly, a sexually dimorphic effect was seen in both whole body behavioral and cell-specific metabolic data, although the results are somewhat conflicting. Most importantly, our animals did not have metabolic differences at baseline (Fig. 1B) or after a 24 h fast (Fig. 3, A and B), despite reports of Nav1.8 afferents from the vagus nerve innervating abdominopelvic viscera (16). This suggests the differences we saw in cell metabolism (Fig. 4) contributed only to animal’s hypersensitive responses (Fig. 3, C–E).

LKB1 in Peripheral Sensory Neurons Does Not Adversely Affect Animal Development or Motor Behaviors

Removal of LKB1 from peripheral sensory neurons was not associated with lethality. This is in keeping with previous studies that showed removal of LKB1 from one cell/tissue type does not necessarily lead to detrimental effects on whole organism development, but removal of LKB1 from the whole body is lethal (20). Although other literature has shown deletion of LKB1 adversely affects axonal development and neuronal polarization (12), our data suggest that neuronal cell bodies develop appropriately in the absence of LKB1 (Fig. 1, C–H). Animals of the same sex but different genotypes weighed the same throughout their lifespan (Fig. 1B). This is consistent with previous literature showing that deletion of neuronal LKB1 does not affect weight or energy homeostasis in male or female mice (21). Most importantly, animals lost and regained a similar amount of weight after a 24-h fast (Fig. 3, A and B). This suggest that while neuronal LKB1 may play roles in other whole body metabolic processes, animals are still able to maintain and regulate their food intake and energy expenditure normally. Thus, our behavioral results were not due purely to different whole body metabolism in different genotypes.

In addition to unaltered whole body metabolism, animals have otherwise comparable depressive-like behaviors (Fig. 1L), gross locomotor behaviors (Fig. 1, I–K), and spontaneous locomotion and exploratory behaviors (Fig. 2). These data suggest no off-target effects of LKB1 deletion in Nav1.8-containing tissues (e.g., altered exploratory behaviors relating to altered food-seeking behaviors). This also suggests that these animals lack motor impairments seen in other models, such as LKB1 deletion from Schwann cells (22) and adipocytes (23). However, it is important to note that motor impairments seen in other models come at later ages than we studied; as such, there may be an interaction between peripheral LKB1 activity and motor impairments in aged animals.

Taken together, these data suggest that deletion of LKB1 allows for appropriate development of cells, neuronal ganglia, and the entire animal in the absence of metabolic stress; it does not address how these animals respond to a metabolic challenge. Our next set of experiments sought to determine if a metabolic challenge was required to see behavioral and cellular differences.

LKB1 Modulates Nociception and Mitochondrial Metabolism in a Sex-Dependent Manner After A 24-h Fast

A 24-h fast induced a sexually dimorphic, as well as genotype-dependent, effect on mechanical hypersensitivity. Males experienced an initial period of mechanical hypersensitivity after a 24-h fast independent of genotype, with Nav1.8^LKB1 animals remaining sensitive (Fig. 3C). Fasted female Nav1.8^LKB1 animals experienced mechanical hypersensitivity earlier than their LKB1⁺ counterparts (Fig. 3D). No differences were seen in weight loss or regain between genotypes, suggesting this effect is due to differences in neuronal metabolism rather than a whole body reaction to the fast (Fig. 3, A and B). Furthermore, anxiety-like behaviors were not affected by fasting (Fig. 5), suggesting that any nociceptive effects of fasting were not due to altered cortical behaviors but were specific to changes in peripheral sensory neurons.

A further sexual dimorphism was seen when neuronal metabolism was tested (Fig. 4). The absence of LKB1 decreased metabolic activity in female neurons after a fast, including decreasing basal (Fig. 4D) and maximal (Fig. 4E) respiration, spare respiratory capacity (Fig. 4J), nonmitochondrial ATP production (Fig. 4G), and proton leak (Fig. 4I). These data suggest that in female nociceptors, LKB1 is necessary to cope with metabolic stress. As Nav1.8^LKB1 females exhibited hypersensitivity after food was returned (Fig. 3D), this impaired metabolic response may have “primed” nociceptors to result in inappropriate energy usage. Male neurons performed similarly independent of fast or genotype, suggesting other processes were responsible for the hypersensitivity seen in male Nav1.8^LKB1 animals (Fig. 3C).

Implications of Cellular LKB1 Activity in Metabolism and Nociception

Our data represent a sexually dimorphic role for neuronal metabolism in the development of mechanical hypersensitivity. Although females experienced an initial LKB1-dependent metabolic response to metabolic stress, fasting still led to mechanical hypersensitivity, albeit later than in the absence of LKB1 (Fig. 3D). Males experienced both an initial development of mechanical hypersensitivity independent of LKB1, followed by an LKB1-dependent hypersensitive period (Fig. 3C) but few LKB1-dependent changes in cellular metabolism (Fig. 4). Our findings suggest neuronal metabolism may drive hypersensitivity in females, similar to other studies linking hypersensitivity to neurons in females but not males (24). Thus, it is imperative to consider 1) how whole body metabolism may have played a role in our findings and 2) what other cells may be responsible for the male nociceptive response to metabolic stress.

Male and female animals have been shown to exhibit different whole body metabolic and endocrine responses to metabolic stress induced by an intermittent fasting regiment, with the fasting period lasting 24 h (25). Overall, whereas females have increased corticosterone and lower plasma ghrelin, males have decreased leptin and increased adiponectin (25). Interestingly, in adipocytes, a sexually dimorphic effect of LKB1 activity, wherein androgens inhibit its expression and estrogens facilitate its expression, has been shown (26). This suggests that LKB1 itself is crucial in sexually dimorphic cellular metabolism. Indeed, our fasted female animals exhibited improved metabolic responses in response to fasting, whereas males did not. This may be due to the intrinsic effects of gonadal hormones on LKB1 expression. Furthermore, the protective aspect seen may be responsible for fasted Nav1.8^LKB1 females developing hypersensitivity earlier than their LKB1⁺ counterparts (Fig. 2D). Without LKB1 expression in nociceptors, animals would not experience its metabolically protective effects and thus not be able to appropriately cope with energy demand.

Although males experienced hypersensitivity not reflected by neuronal mitochondrial activity, a different cell type may be responsible for the initial hypersensitive period in male animals. While its effects on neuronal metabolism would not be present immediately after the fast, it could affect downstream LKB1-dependent pathways in neurons, explaining later genotype-dependent hypersensitivity. As fasting induces different responses in different tissues, our results potentially may be due to different cell types in males and females (27). Interestingly, human adipose exposed to an in vivo fasting regime (10 days with water, multivitamin, 20 mEq potassium chloride, and 200 mg allopurinol but no food consumption) has been shown to acquire a proinflammatory profile (28, 29), although it is important to note that this fasting regime was for significantly longer than the period of fasting we tested, mice have a higher basal metabolic rate than humans (30), and the results of this study were not shown to be LKB1 pathway-specific.

The primary downstream pathway of LKB1 is AMPK, which is known to cause various antiinflammatory effects in macrophages (28). It is not clear if fasting affects LKB1 activity in neurons and macrophages equally. If neurons are not appropriately able to respond to fasting in an LKB1-dependent manner, their communication with macrophages may be interrupted. Nevertheless, these differing responses in different cell types suggest nociception in response to a long fast is not mediated by neurons alone.

Although our results suggest the LKB1 pathway is crucial to understanding sexual dimorphism in metabolism and pain, more must be done to fully elucidate how cell metabolism can contribute to chronic pain. Several other metabolic pathways have been shown to interact with fasting and chronic pain. For example, fasting inhibits the mTOR pathway, downstream of the AMPK pathway, which has been shown to be upregulated during chronic pain (5). Overall, AMPK activity has been linked with inhibiting the cell-growth signals associated with the mTOR pathway and may partially represent how fasting-induced metabolic stress is able to alleviate pain, including but not limited to a disruption of protein upregulation associated with chronic pain (5, 9, 31). Interestingly, in male animals, the AMPK pathway has been seen to interact with the cannabinoid pathway during periods of caloric restriction (32). While this still does not explain the original period of hypersensitivity, an inability of LKB1 to activate AMPK and subsequent cannabinoid activity could explain the results we see in male animals. However, it is important to note we did not directly study the results of AMPK activity, although removal of LKB1 is sufficient to ablate AMPK activity (15). There are also several other pathways that LKB1 could impact, although each play similar roles to ultimately relieve energetic demand (33). Other metabolic pathways that can alleviate energy demand, such as peroxisome proliferator-activated receptors (PPARs) also may be attractive targets for future therapeutics (34). However, it is not clear which cell is ultimately responsible for the pain-relieving qualities of the PPAR pathways, especially as different isoforms can induce catabolic and anabolic pathways. Although our study lays precedence in connecting the role of nociceptor metabolism to whole body antinociception, there are still many metabolic pathways in a variety of cells that may explain our data.

Perspectives and Significance

Clinically, women are overrepresented as chronic pain patients. Existing literature suggests sexual dimorphism on a cellular level is responsible for this overrepresentation, with females having different mechanisms of pain development and maintenance. Despite this, the exact cell responsible for male versus female pain remains elusive. Existing literature suggests chronic pain causes metabolic crisis in pain-sensing neurons independent of sex. Our data further suggest that male and female DRG neurons respond to metabolic crisis differently, with an initial dimorphism in response to metabolic crisis is partially responsible for the disparity between sexes. Nevertheless, additional work to fully understand both how neuronal metabolism contributes to sexually dimorphic pain and its cell-specific corelates is needed.

GRANTS

This study was supported by National Institutes of Health Grant K22NS096030 (to M. D. Burton), The Rita Allen Foundation (to M. D. Burton), and The University of Texas Rising STARS program research support grant (to M. D. Burton).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

K.M.G. and M.D.B. conceived and designed research; performed experiments; analyzed data; interpreted results of experiments; prepared figures; drafted manuscript; edited and revised manuscript; and approved final version of manuscript.