Augmenting the antinociceptive effects of nicotinic acetylcholine receptor activity through lynx1 modulation

Neuronal nicotinic acetylcholine receptors (nAChRs) of the cholinergic system have been linked to antinociception, and therefore could be an alternative target for pain alleviation. nAChR activity has been shown to be regulated by the nicotinic modulator, lynx1, which forms stable complexes with nAChRs and has a negative allosteric action on their function. The objective in this study was to investigate the contribution of lynx1 to nicotine-mediated antinociception. Lynx1 contribution was investigated by mRNA expression analysis and electrophysiological responses to nicotine in the dorsal raphe nucleus (DRN), a part of the pain signaling pathway. In vivo antinociception was investigated in a test of nociception, the hot-plate analgesia assay with behavioral pharmacology. Lynx1/α4β2 nAChR interactions were investigated using molecular dynamics computational modeling. Nicotine evoked responses in serotonergic and GABAergic neurons in the DRN are augmented in slices lacking lynx1 (lynx1KO). The antinociceptive effect of nicotine and epibatidine is enhanced in lynx1KO mice and blocked by mecamylamine and DHβE. Computer simulations predict preferential binding affinity of lynx1 to the α:α interface that exists in the stoichiometry of the low sensitivity (α4)3(β2)2 nAChRs. Taken together, these data point to a role of lynx1 in mediating pain signaling in the DRN through preferential affinity to the low sensitivity α4β2 nAChRs. This study suggests that lynx1 is a possible alternative avenue for nociceptive modulation outside of opioid-based strategies.


Introduction
Pain is amongst the most common reasons for seeking medical treatment, however approximately 80 percent of pain sufferers worldwide do not have sufficient access to proper care [1][2]. The most widely used therapies for acute nociceptive pain involve agonists of the opioid receptors. Such agents, however, carry risk for major off-target effects [3], along with the high propensity for overdose, abuse and the development of tolerance [4], which is a rapidly PLOS

Model organism
Genetically modified female and male mice were used as the model organism for this study. Both wild type (wt) (C57BL/6J), lynx1KO and β2-GFP knockin mice were used. The mice tested were between the ages of 3-7 months (20-50g) for the nicotine-, epibatidine-and dihydro-β-erythroidine hydrobromide (DHβE) assays and 6-8 months (30-50g) for the mecamylamine (mec)-, locomotion-and temperature assays. Naïve mice were used for the nicotine-, epibatidine-and DHβE assays, and reused for the mec-, locomotion-and temperature assays with sufficient time in between assays (minimum 1 week). Adult naïve mice were used for PCR, immunohistochemical staining and western blot analysis. For the electrophysiological experiment 16-18 days old naïve mice were used. The lynx1KO and β2-GFP mice were transferred from the California Institute of Technology in Pasadena, CA, USA or bred at Lehigh University, PA, USA. Lynx1KO mice were generated using 129 ES cells [48] and backcrossed over 12 generations to the C57BL/6J background. β2-GFP knock-in mice were generated in the Lester laboratory, California Institute of Technology in Pasadena, CA, USA, and transferred to the Lehigh Central Animal Facility. Mice were kept on a 12:12 light-dark cycle and food and water were provided ad libitum. The mice were kept with up to 4 other mice and were ear punched for identification purposes. All procedures and protocols were approved and in compliance with Lehigh IACUC guidelines on the humane care and use of animals (IACUC protocol #147). All efforts were made to minimize animal suffering. Animal studies are reported in compliance with ARRIVE guidelines.

Genotyping
All mice were genotyped using a polymerase chain reaction assay. A mouse tail biopsy was obtained before and after experimentation for both pre-and post-hoc genotyping. The DNA was lysed overnight with Proteinase K and extracted using isopropanol or Qiagen© DNA Miniprep Kit (Qiagen, Hilden, Germany). The DNA was amplified through a standard polymerase chain reaction protocol. Each tissue sample was genotyped 2-3 times before a confirmation of the genotype was assigned.

Quantitative PCR and RT-PCR
Tissue from the DRN was dissected and RNA was extracted using the RNeasy Mini kit (Qiagen, Hilden, Germany). Purified RNA was synthesized into cDNA using the qScript cDNA SuperMix (Quanta Bioscience, Beverly, MA, USA). Quantitative PCR was performed in triplicate using lynx1 TaqMan gene expression assay (ThermoFisher Scientific, Waltham, MA, USA) with GAPDH as reference gene. The expression of target genes was normalized to the expression of the reference gene [51]. PCR products were run on a 4% agarose gel. Furthermore, RT-PCR was also carried out using tryptophan hydroxylase 2 primers (FW: 5'CTGA AAGAGCGATCTGGCTTC3', Rev: 5'ATCTGGTTCCGGGGTGTAGA3') (Integrated DNA technologies, USA) to confirm localization of RNA to the DRN. PCR products were run on a 2% agarose gel.
incubating free-floating sections. Sections were stained with custom made anti-lynx1 monoclonal antibodies (1:1000, 1.7 ng/μl) [44]) o/n, and incubated with Alexa Fluor 568 conjugated anti-mouse secondary antibodies at 1:1000 dilution for 1 hour at RT (ThermoFisher Scientific, Waltham, MA, USA). Serotonergic neurons were identified using an anti-TPH2 (Tryptophan hydroxylase 2) antibody (1:500, ThermoFisher Scientific, Waltham, MA, USA) o/n, and incubated with Cy-2 conjugated donkey anti-rabbit secondary antibodies at 1:2000 dilution (Jackson Immunochemicals, West Grove, PA, USA) at RT for 1 hour. Sections were also incubated with rabbit polyclonal antibodies against lynx1 ( [39]. These antibodies were custom generated against a synthetic lynx1 peptide (TTRTYFTPYRMKVRKS) and previously tested by immunofluorescence and immunocytochemistry staining at 1:2000 dilution [39]. To confirm the specificity of staining of this antisera, affinity-purified anti-lynx1 antibodies were tested by Western blot of mouse brain extracts to confirm~9-12 kD band size, using antibodies purified on a nickel column to purify a bacterially produced lynx1-his protein, and subsequently tested on Western blots from lynx1KO mouse brain extracts. The staining in the lynx1 +/+ mouse brains demonstrated band at the expected band size, and no band was present in the lynx1KO -/brain samples [48]. DRN sections were incubated with anti-lynx1 antisera at 1:1000 dilution for 1 hour at RT. After washing, sections were incubated with Cy-2-conjugated donkey anti rabbit at 1:200 (Jackson Immunoresearch, West Grove, PA, USA) for 1 hour at RT. Sections were mounted on microscope slides and imaged using Nikon elements software on Nikon E1000 confocal microscope.

Dual labeling immunofluorescence
Adult C57bl6 mice were perfused as described above. Dissected brains were sectioned at 50μm on a freezing microtome. Serotonergic neurons were identified using an anti-TPH2 (Tryptophan hydroxylase 2) antibody (1:500, ThermoFisher Scientific, Waltham, MA, USA) o/n, and incubated with Cy-2 conjugated donkey anti-rabbit secondary antibodies at 1:2000 dilution (Jackson Immunochemicals, West Grove, PA, USA) at RT for 1 hour. Sections were co-stained with custom made anti-lynx1 monoclonal antibodies raised against mouse lynx1 protein purified from bacteria and refolded (Green Mountain Laboratories, mouse, 1:1000, 1.7 ngÁμl -1 ) [44] o/n, and incubated with Alexa Fluor 568 conjugated anti-mouse secondary antibodies at 1:1000 dilution for 1 hour at RT (ThermoFisher Scientific, Waltham, MA, USA). Sections were mounted and imaged using Nikon elements software on Nikon E1000 confocal microscope at 10, 20, 40 and 100x magnification. Sections were also stained singly with either anti-TPH2 or anti-lynx1 antibodies to establish staining pattern and control for bleed through staining. To further confirm the specificity of lynx1 staining, we stained side by side DRN sections with anti-lynx1 polyclonal and anti-TPH2 antibodies.

Co-immunoprecipitation (Co-IP)
β2-GFP knockin dorsal brain regions containing the DRN were dissected and immediately homogenized in the bullet blender tissue homogenizer (NextAdvance, Averill Park, NY, USA), with Co-IP buffer (50 mM Tris, 150 mM NaCl, 0.75% Triton-X 100, Pierce protease inhibitor cocktail). Dynabeads A (ThermoFisher Scientific, Waltham, MA, USA) were pre-incubated with 10μg rabbit anti-GFP primary antibodies (ThermoFisher Scientific,Waltham, MA, USA) and thoroughly washed. Brain homogenates at a concentration of 13 mgÁml -1 were incubated with the dynabeads-antibody complex for 3 days rotating at RT. After washing, GFP fusion protein complexes including interacting proteins were eluted and immediately prepared for western blot analysis.

Electrophysiology
Animals, aged 16-18 days, were split into 2 groups according to their genotype. The mice were deeply anaesthetized with isoflurane. Frontal midbrain slices at 300 μm were made with a Vibratome (Vibratome Co., St Louis, MO, USA), and the slices were incubated in an oxygenated sucrose solution at 35˚C for 1 hour. The recordings were perfused with Ringer solution containing the following ingredients (in mM): NaCl 128, KCl 2.5, NaH2PO4 1.25, CaCl2 2, MgSO4 1, NaHCO3 26, and dextrose 10, pH 7.4. Whole-cell patch clamp recordings were conducted in DRN slices visualized through an upright microscope (Scientifica, UK) equipped with infrared differential interference contrast optics (IR-DIC). Resistance of the recording pipette (1.2 mm borosilicate glass) was 7-9 MO. Tips of the recording pipettes were filled with a potassium gluconate-based intracellular solution: potassium gluconate 120, KCl 6, ATP-Mg 4, Na2GTP 0.3, EGTA 0.1, and Hepes 10, pH 7.3, 310 mosmolÁL -1 . Cell type identification using the action potential (AP) shoulder was performed under current clamp recording conditions to discriminate serotonergic neurons (with shoulder) and GABAergic neurons (without a shoulder). The cell was then switched into voltage clamp mode and held at -70 mV for the assessment of nicotine-evoked peak responses. Nicotine was dissolved in ACSF for fast-application to the neuronal cell body using a Picospritzer (20 p.s.i., 200 ms) to pressure-eject the nicotine solution (Parker Hannifin corporation). The peak value of response was measured and statistically analyzed among different groups.

Western blot analyses
Samples were denatured in 1x sample buffer (ThermoFisher Scientific, Waltham, MA, USA) at 95˚C and run on a 15% SDS-PAGE gel. Gels were transferred onto activated PVDF membrane via semi-dry transfer. The membrane was blocked with 5% milk/0.05% Tween-PBS for 1 hour at 4˚C followed by an overnight incubation at 4˚C in T-15 anti-lynx1 (1:1000, Santa Cruz Biotechnology, Dallas, TX, USA). After thorough washing, the membrane was incubated with conjugated donkey anti-goat (Abcam, Cambridge, MA, USA) at 1:10,000 for 1 hour at 4˚C. Membranes were incubated in ECL (ThermoFisher Scientific,Waltham, MA, USA) and exposed to film. Actin controls were run using mouse anti-actin (Abcam, Cambridge, MA, USA) at 1:1000 dilution, and goat anti-mouse (Life Technologies, Carlsbad, CA, USA) at 1:40,000 dilution.

Hot-plate assay
The hot-plate test was used to measure response latencies according to a method previously described [53]. The hot-plate (Columbus Instruments, Columbus, OH, USA) was set at a temperature of 55˚C, and the latency to lick the hind-paw or jump off the surface was evaluated as an index of nociception. Immediate signs of temperature sensation such as fluttering of the feet or licking of the front paws were also noted, however the mice were not removed from the hot-plate until jumping or hind-paw licking. A cutoff time of 60 seconds was determined [11].

Computer modeling of lynx1-AChR complexes
A low-resolution structure model of α4:α4/lynx1 complex was first generated using a template-based approach. The structure of α4 subunit and that of lynx1 were taken from the Protein Data Bank (PDB) entries 5kxi [55] and 2l03 [45], respectively. Each target structure (α4 subunit or lynx1) was searched against all known proteins in the PDB to identify the PDB proteins (templates) having similar global structures. The global structure similarity between the target and template structures was measured by TM-align [56]. The PDB protein searches identified a PDB entry, 4hqp (α7 nicotinic receptor chimera in complex with α-bungarotoxin, Figure  The α4 subunit structure has a closed-in conformation of loop C. In addition, the structure of lynx1 has more structured loop I, compared to other members of lynx family (e.g., CD59, candoxin, erabutoxin-a) [45]. These structural features result in a number of bad contacts in the initial low-resolution α4:α4/lynx1 complex model. The model was refined through minimizing the bad contacts. We sampled thermally acceptable 100 conformations for α4 loop C and lynx1 loop I using Backrup application in Rosetta [57], generating 100 × 100 combinations of α4:α4/lynx1 complexes with different loop conformations. For each complex, lynx1 was moved outwards from α4:α4 by 2 Å using an interval of 1 Å along a vector defining the geometric centers of two protein binding surfaces to find a complex structure with the minimum number of bad contacts. The refined α4:α4/lynx1 complex model was subjected to 150-ns molecular dynamics (MD) simulations to obtain a high-resolution complex model. During the simulations, weak positional restraints were applied to the backbone heavy atoms of α4:α4, aiming at maintaining the pentameric orientation of α4:α4. For lynx1, we applied or removed the positional restraints every 10-ns to prevent dissociation of lynx1 from α4:α4. No restraints were applied to loop B, loop C, and β8-β9 loop in α4:α4 and loop I in lynx1, allowing their conformational flexibility, as they might involve protein-protein interactions. After the 150-ns simulations, we removed all the positional restraints to lynx1 and performed additional 100-ns simulations for equilibration. All MD simulations were prepared through CHARMM-GUI [58][59] and performed with explicit water molecules and 150 mM KCl for physiological salt concentration using CHARMM36 force filed [60] and NAMD software [61] at 303.15 K. The final snapshot was energy-minimized and used to describe atomic level interactions between α4:α4 and lynx1 in the complex model.
To generate a β2:β2/lynx1 complex model, we used the same template-based modeling approach based on PDB: 4hqp and simulation protocol as α4:α4/lynx1. In the case of β2:β2/ lynx1, however, we did not perform the additional 100-ns equilibration MD simulations. For a structure comparison of α4:α4/lynx1 with β2:β2/lynx1, lynx1 structure in the final α4:α4/lynx1 complex model was copied onto β2:β2 to generate the β2:β2/lynx1 complex model.

Data presentation and statistical analyses
Hot-plate, locomotion and body temperature data were presented as mean ±S.E.M. Data were analyzed using either the Student T-test or the two-way ANOVA test with a Tukey post-hoc test. All assumptions for the Student's T-test and two-way ANOVA were considered. A pvalue less than 0.05 was considered statistically significant. All statistical tests were performed in IBM SPSS Statistics 24 (IBM Corporation, Armonk, NY, USA). The data and statistical analysis comply with the recommendations on experimental design and analysis in pharmacology.

Results
We chose the DRN to begin our investigations because it has been shown to mediate nociceptive signals and express high levels of nAChRs which when activated contribute to antinociception [10]. To determine whether the DRN express lynx1, the mRNA expression of lynx1 and TPH2 in the DRN was analysed using quantitative PCR (relative expression lynx1/GAPDH: 0.0024) and Reverse Transcriptase PCR (RT-PCR) ( Fig 1A). We chose TPH2 as a marker of the DRN, since it is an enzyme involved in the synthesis of serotonin, and is highly enriched in the DRN [62][63]. Although this establishes the expression of lynx1 in the DRN, we sought an independent confirmation using another methodology. We also performed immunohistochemical staining using previously published anti-lynx1 antibodies. Immunofluorescence staining using a polyclonal antibody specifically shown to label lynx1 protein [39,48,50], showed expression of lynx1 in ventral regions of the DRN as compared to dorsal regions of the DRN (Fig 1B and 1C) or outside the DRN ( Figure C in S1 Fig). Side by side comparative staining of anti-lynx1 monoclonal [44], and anti-TPH2 polyclonal antibody (Fig 1D and 1E), showed similar regions of staining of TPH2 and lynx1 in the region of the DRN. Co-labeling studies with these same antibodies showed areas of dual labeling of lynx1 and TPH2, as well as lynx1 staining which were not labeled with TPH2 at 10x (Fig 1F), 20x (Fig 1G), and 40x ( Fig  1H) magnification. This indicates that lynx1 is expressed in both serotonergic and non-serotonergic neurons in the DRN.
We next were interested to understand if lynx1 modulates nAChRs in the DRN, using lynx1KO slices to assess nicotine-evoked response properties (Fig 2A). There are two major cell types in the DRN, serotonergic and GABAergic, which can be discriminated by the action potential shape [65], with the serotonergic neuron having a shoulder in the falling phase of the the DRN, as a coronal plane of section (C) Expression of lynx1 protein (green) in the dorsal raphe nucleus, using anti-lynx1 pAb immunofluorescence staining [39,64]and donkey anti-rabbit Cy2 secondary antibody, imaged at 4x magnification. (D and E) Side by side labeling of lynx1 (anti-lynx1 mAb, Alexa red), and TPH2 (anti-TPH2 pAb, Cy2, green), 10x magnification, scale bar = 200 μm. (F) Dual labeling immunofluorescence staining using anti-lynx1 mAb (red) and anti-TPH2 pAb (green), merge (yellow), 20x magnification, scale bar = 100 μm. (G) Dual labeling immunofluorescence staining using anti-lynx1 mAb (red) and anti-TPH2 pAb (green), merge (yellow), 40x magnification, scale bar = 50 μm.
https://doi.org/10.1371/journal.pone.0199643.g001 action potential. We obtained whole-cell measurements from these two neuronal cell types in the DRN, the neurons with an AP shoulder being serotonergic-like neurons (serotonergic) and those without an AP shoulder being GABAergic-like (GABAergic) neurons, using these parameters to discriminate them from one another (Fig 2B) [65]. We measured acute responses to fast application of pressure-ejected nicotine (200 μM) in neurons of brain slices in the DRN from wt and lynx1KO. We found that there were differences in the peak response between wt and lynx1KO neurons (Fig 2C), with higher peak nicotine response in neurons from lynx1KO vs wt slices (GABAergic neurons: wt 47.8 ± 16.1 pA, n = 9 (mice N = 3) vs. lynx1KO 170 ± 38.2 pA, n = 17 (mice N = 6); p = 0.008, student t test. Serotonergic neurons: wt 61.0 ± 13.2 pA, n = 12 (mice N = 4) vs. lynx1KO 164 ± 34.6 pA, n = 19 (mice N = 6); p = 0.01, student t test), in response to fast application of nicotine between the two genotypes and in both cell types (Fig 2C). These data demonstrate a modulatory action of lynx1 on nicotinic receptors in one of the brain regions involved in nociceptive signaling.
To address this possibility, we sought to understand the effect of lynx1 in our genetic mice using a behavioral test of acute thermal nociception. Lynx1KO mice have been shown to be more sensitive to the effects of nicotine in electrophysiological, motor learning assays [48] and visual evoked responses in the visual cortex [50] and therefore we predicted that the lynx1KO mice would be augmented in either baseline nociceptive responses and/or sensitivity to the antinociceptive effects of nicotine. We conducted thermal assays in wt and lynx1KO mice using a hot-plate paradigm. The sensitivity to the heat from the hot-plate was assessed by measuring the response latency to react (secs) under saline and nicotine conditions. No differences in baseline (saline) nociception between wt and lynx1KO mice were found (mean difference 0.8 sec, p = 0.899, two-way ANOVA. n = 8, 8; wt, lynx1KO, respectively. Cohen's D 0.13). Both genotypes, however, showed a dose-dependent increase in time spent on the hot-plate after nicotine administration (Fig 3A, Figure A in S2 Fig). Lynx1KO mice showed more sensitivity to nicotine as compared to wt mice at lower doses (0.5 mgÁkg -1 nicotine; mean difference 13.4 sec, p = 0.034, two-way ANOVA, n = 8, 9, Cohen's D 1.37. 1.0 mgÁkg -1 nicotine; mean difference 12.2 sec, p = 0.034, two-way ANOVA, n = 8, 14, Cohen's D 1.09). These data suggest that genetic removal of lynx1 in mice increased the antinociceptive actions of nicotine. We saw no differences due to sex (Fig 3B).
Next, we blocked nAChRs using a non-selective channel blocker of nAChRs, mecamylamine, in the same hot-plate assay. We injected nicotine (1 mgÁkg -1 ) 10 minutes after mecamylamine (2.5 mgÁkg -1 ) injection. We found that blocking nAChRs with mecamylamine was able to inhibit the augmentation in nicotine-mediated antinociception observed in both wt and lynx1KO mice (Fig 3C) to normal levels of responsiveness similar to saline injected controls (nicotine-treated lynx1KO vs. nicotine+mec treated lynx1KO mice: mean difference 21.1 sec, p = 0.0001, Student's T-test, nicotine-treated lynx1KO vs. mec treated lynx1KO mice: mean difference 26.15 sec, p<0.00001, Student's T-test, nicotine-treated wt vs. nicotine+mec treated wt mice, mean difference 15.4 sec, p = 0.002, saline treated wt vs nicotine-treated wt, mean difference 12.89, p = 0.014, saline-treated KO vs saline-treated KO nicotine, mean difference 25.0 sec, p<0.00001). All mice were examined for tissue damage post experiments, and no sign of tissue damage or discomfort were observed. These data suggest that that the augmented antinociceptive effect of nicotine in the lynx1KO mice is mediated through nAChRs, and that binding of lynx1 to this receptor modulates its antinociceptive properties.
Mice were tested on the hot-plate 15 minutes after injection. Nicotine-mediated antinociception is augmented in lynx1KO mice at nicotine concentrations 0.5 mgÁkg -1 and 1.0mgÁkg -1 compared to wt mice. Each data point presented as mean ± SEM. Ã P<0.05 compared to wt controls at corresponding concentrations of nicotine. wt: wild type, KO: lynx1 knockout. (B) The effect of nicotine on antinociception in male and female lynx1KO mice. Antinociceptive Lynx1 antinociception modulation responses in male and female lynx1KO mice after I.P. injections of 0.5 mgÁkg -1 nicotine (n = 8 male, n = 8 female, p = 0.844, student's t-test) using the hot-plate assay. Mice were tested on the hot-plate 15 minutes after injection. There are no antinociceptive difference between sexes. Each data point presented as mean ± SEM. (C) Antinociceptive responses in wt and lynx1KO mice after I.P. injection of the general nAChR blocker, mecamylamine (2.5 mgÁkg -1 ) and nicotine (1mgÁkg -1 ) (nicotine treated lynx1KO mice (n = 14) vs. nicotine+mecamylamine treated lynx1KO mice (n = 13): p = 0.003, Student's T-test and nicotine treated wt mice (n = 8) vs. nic-otine+mecamylamine treated wt mice (n = 16): p<0.001, Student's T-test) using the hot-plate assay. Injections of mecamylamine block the antinociceptive effect of nicotine in both genotypes. Data indicates that lynx1 operates through nAChRs. Data presented as mean ± SEM time. Ã P<0.05 nicotine injected animals compared to nicotine+mecamylamine injected animals. wt: wild type (C57bl6, lynx1 +/+ and β2-GFP), KO: lynx1 knockout. We performed statistical analysis between sex, and no significant differences were seen between female and male mice (Fig 3B).
To confirm the interaction of lynx1 with α4β2 nAChRs, we performed co-immunoprecipitation experiments using β2-GFP mice [71] to pull out native nAChR complexes from the brain. We used anti-GFP antibodies to pull-down nAChRs and associated protein, and lynx1 as an associated part of the complex was probed using anti-lynx1 antibodies and Western blot analyses. We could detect lynx1 present at enriched levels in the pull-down lane, whereas the flow-through lane was blank ( Figure D in S1 Fig), suggesting that lynx1 forms a stable complex with β2 Ã -containing nAChRs. This in support of recent reports of lynx1 in aiding the assembly and stoichiometry of α4β2 receptors [47]. In that report, there was an apparent shifting to the low sensitivity stoichiometry of (α4) 3 (β2) 2 by a preferential binding affinity of lynx1 to the α4: α4 interface (Fig 5C). In order to gain insight into this idea, we employed computational modeling to address stoichiometry-specific interactions.
We performed simulations of lynx1 binding to the interfaces of the unique interfaces of the α4β2 receptor stoichiometries: α4:α4 vs. β2:β2. The overall architecture of lynx1 at the α4:α4 interface is shown in Fig 6. From comparisons between α4:α4 and β2:β2 nAChRs structures in complex with lynx1, we found that more favorable interactions exist between the α4:α4 interface and lynx1 (Fig 7). In the α4:α4/lynx1 complex model (Fig 7A and 7C), lynx1 D55 forms three hydrogen bonds with two residues (S170 and H169) in the α4 subunit, while two hydrogen bonds in α4 subunit are due to substitution of E165 in the β2 subunit for S170 in the α4 subunit. Loop C of the α4 nAChR subunit closely contacts with loop I of lynx1, forming hydrogen bonds and hydrophobic interactions (α4 Y197 with lynx1 V6 and M18). Lynx1 R31 also interacts with nAChRs loop C through a hydrogen bond. On the contrary, loop C of the β2 subunit (Fig 7B and 7D) is shorter, making it difficult to generate favorable interactions with lynx1 loop I. In particular, G154 in the α4 subunit is replaced by R149 at the equivalent position in the β2 subunit, implying that the presence of this positive charged residue could prevent lynx1 R38 from approaching and interacting with Y95, W151, and Y196. In the α4:α4/lynx1 complex model, R38 appears to form hydrogen bonds with Y100 and W156, and cation-π interaction with Y204 [72]. Therefore, our modeling and simulations support the experimental results of preferential binding affinity of lynx1 to the α4:α4 interface, as compared to the β:β interface.
More favorable interactions exist between the α4:α4 interface and lynx1; lynx1 D55 forms three hydrogen bonds with two residues (S170 and H169) in the α4 subunit, while two hydrogen bonds in the α4 subunit are due to substitutions of E165 in the β2 subunit for S170 in the α4 subunit. Loop C of the α4 subunit closely contacts with loop I of lynx1, forming hydrogen bonds and hydrophobic interactions (α4 Y197 with lynx1 V6 and M18). Lynx1 R31 also interacts with loop C through a hydrogen bond. On the contrary, loop C of the β2 subunit is shorter, making it difficult to generate favorable interactions with lynx1 loop I. The lynx1 protein and receptor is colored yellow and green/cyan, respectively. Labels for α4:β2 and lynx1 are colored blue and red, respectively. Potential hydrogen binding are represented as dashed red lines.

Discussion
In this study we discovered that lynx1 is expressed in at least one region of the nociception pathway in the CNS, and that mice without the lynx1 gene demonstrated more nicotine-mediated antinociception. Peak nicotine-evoked responses were larger in lynx1KO neurons in the DRN. Furthermore, the data suggest that a possible mechanism of antinociception in the lynx1KO is mediated through the α4β2 nAChR subtype. The data are congruent with the hypothesis that nAChRs of lynx1KO mice are more sensitive to the effects of nicotine than wtanimals, resulting in an improvement in cholinergic-mediated behavior; in this case antinociception.
Serotonergic neurons in the DRN are well established components in nociception pathways and our data demonstrate the presence of lynx1 in DRN neurons (Fig 1) [28,[73][74][75]. Hypersensitive nAChR receptors in both serotonergic and GABAergic neurons caused by lynx1 removal could shift the excitatory-inhibitory balance towards antinociception. This does not exclude the role of other brain regions, however, as lynx1 is widely expressed in the CNS [18]. Also, our data do not exclude a role of peripheral sites expressing nAChRs in the nociceptive behavior. Lynx1 has not been detected in high levels in the periphery, but the effects of lynx1 outside the CNS awaits further testing. Although our data demonstrate that lynx1KO mice are more sensitive to the effects of nicotine in the hot-plate assay than wt mice, we do not exclude effects of endogenous acetylcholine acting in antinociception in lynx1KO mice. Since we do not see an effect of a non-nicotinic drug (e.g ibuprofen), ( Figure C   endogenous acetylcholine would likely be subtle, indicating another example of elevated cholinergic tone in these mice [18,50].
Although the precise mechanism by which lynx1 influences antinociception is as yet unknown, lynx1 has been shown to have global and multiple effects on nAChR function [44,[47][48][49], so there are candidate mechanisms. The role of specific nAChRs in the DRN and nociception has been demonstrated [36]. Our data are in line with reports about the involvement of α4β2 nAChRs in antinociception [10,11,[76][77], but does not exclude the involvement of other nAChR subtypes [20,23,24,78]. Lynx1 has also been shown to act on α5α3β4 nAChRs [49], but an involvement at other α4β2 Ã nAChRs (e.g., α5α4β2, α6α4β2, etc. ) has yet to be tested. In this model, we have not included the role of α7 nAChRs, which has been shown to mediate nociceptive signaling and to bind to lynx1 proteins [22,44]. Thus, further research into the specific subtypes involved is needed. The biophysical mechanism of lynx1 action has been hypothesized to be due to interactions at the interface of subunits of the nAChR pentamer. Lynx1 has also been shown to have global allosteric actions on nAChR function. This includes, not only orthosteric effects on agonist sensitivity, but also includes effects on desensitization kinetics, recovery from desensitization [44,48] and nAChR receptor closed times [49]. The antinociceptive actions of lynx1 removal could be explained by a lynx1-directed shift in the stoichiometry of the α4β2 nAChRs from a HS to a LS [47] (Fig 4B). The HS nAChR, adopting the (α4) 2 (β2) 3 stoichiometry, exhibits higher sensitivity and slower desensitization, whereas the LS nAChR, adopting the (α4) 3 (β2) 2 stoichiometry, exhibits lower sensitivity to agonist, faster desensitization and higher Ca 2+ permeability [69,79] In the lynx1KO mice, we expect a higher relative expression of the HS stoichiometry, which is supported by our data which show an enhanced sensitivity to nicotine in lynx1KO mice as compared to wt mice. The computer simulation data further support the HS stoichiometry by (Figs 6 and 7) suggesting a preferential binding of lynx1 to the α4:α4 interface of the α4β2 nAChRs. We also saw that the β2 Ã -selective nAChR antagonist DHβE was able to block nicotine's antinociceptive effect in lynx1KO mice, which further suggests that lynx1 is able to modulate the α4β2 nAChR subtype, demonstrating the β2 Ã nAChRs involvement in antinociception but not locomotion or thermal regulation [80] (Fig 4).
Untreated acute pain can influence the development of chronic pain and can have deleterious effects on quality of life and account for billions in hospital costs [81,82]. Thus, management of acute pain is essential for the patient and the public health in both the short and long term [83][84]. Our data addresses acute pain more so than chronic or neuropathic pain, but the relationship of lynx1 to plasticity suggests that other types of pain (neuropathic, peripheral, chronic, etc.) might be altered in our genetic model. A recent paper reported that epibatidine's antinociceptive efficacy declined over time in a mouse model of chronic pain and that no changes in the number and affinity of α4β2 nAChRs occurred [85], suggesting the possibility that lynx1 may modulate behavioral tolerance to nicotine. However, this is speculative and further investigation is needed.
Summary: nAChRs have been implicated in nociceptive processing and has been explored as a therapeutic avenue for the alleviation of nociception. The modulator lynx1 can regulate nicotinic activity through its ability to bind and regulate nAChRs. We found a nicotine-mediated antinociception involving α4β2 nAChRs, which is augmented in lynx1KO mice.

S2 Fig. The effect of nicotine and ibuprofen on antinociception and saline on locomotion.
A. Antinociceptive responses in wt and lynx1KO mice after I.P. injections of nicotine at concentrations of 0.25 mgÁkg -1 (n = 8 wt, 8 KO, p = 0.656 two-way ANOVA, cohen's D 0.40), 0.5 mgÁkg -1 (n = 8 wt, 18 KO, p = 0.122, two-way ANOVA, cohen's D 1.36), 1.0mgÁkg -1 (n = 8 wt, 14 KO, p = 0.032, two-way ANOVA, cohen's D 1.09) and 1.5mgÁkg -1 (8 wt, 8 KO. p = 0.657, two-way ANOVA, cohen's D = 0.13) using the hot-plate assay plotted in a semi-log format. Mice were tested on the hot-plate 15 minutes after injection. Nicotine-mediated antinociception is augmented in lynx1KO mice at nicotine concentrations 0.5 mgÁkg -1 and 1.0mgÁkg -1 compared to wt mice. Each data point presented as mean ± SEM. Ã P<0.05 compared to wt controls at corresponding concentrations of nicotine. wt: wild type, KO: lynx1 knockout. B. Effect of saline on locomotion in wt and lynx1KO mice after I.P. injections of saline (n = 8 wt, n = 8 KO, not significant). Locomotion were examined by scoring leg movements (seconds) in the time period 15-20 minutes post injection. The locomotor performance was binned into 5 minute time windows and showed no significant effect at any time window. Each data point presented as mean ± SEM. wt: wild type, KO: lynx1 knockout. C. Effect of ibuprofen sodium salt in wt and lynx1KO mice after I.P. injection, 20 mgÁkg -1 (n = 10 wt, n = 12 KO, not significant). Each data point presented as mean ± SEM. wt: wild type, KO: lynx1 knockout. (TIF)