The Left-Right Side-Specific Neuroendocrine Signaling from Injured Brain: An Organizational Principle

Abstract A neurological dogma is that the contralateral effects of brain injury are set through crossed descending neural tracts. We have recently identified a novel topographic neuroendocrine system (T-NES) that operates via a humoral pathway and mediates the left-right side-specific effects of unilateral brain lesions. In rats with completely transected thoracic spinal cords, unilateral injury to the sensorimotor cortex produced contralateral hindlimb flexion, a proxy for neurological deficit. Here, we investigated in acute experiments whether T-NES consists of left and right counterparts and whether they differ in neural and molecular mechanisms. We demonstrated that left- and right-sided hormonal signaling is differentially blocked by the δ-, κ- and µ-opioid antagonists. Left and right neurohormonal signaling differed in targeting the afferent spinal mechanisms. Bilateral deafferentation of the lumbar spinal cord abolished the hormone-mediated effects of the left-brain injury but not the right-sided lesion. The sympathetic nervous system was ruled out as a brain-to-spinal cord-signaling pathway since hindlimb responses were induced in rats with cervical spinal cord transections that were rostral to the preganglionic sympathetic neurons. Analysis of gene–gene co-expression patterns identified the left- and right-side-specific gene co-expression networks that were coordinated via the humoral pathway across the hypothalamus and lumbar spinal cord. The coordination was ipsilateral and disrupted by brain injury. These findings suggest that T-NES is bipartite and that its left and right counterparts contribute to contralateral neurological deficits through distinct neural mechanisms, and may enable ipsilateral regulation of molecular and neural processes across distant neural areas along the neuraxis.


Introduction
][9][10][11] After a unilateral injury to the hindlimb sensorimotor cortex, animals exhibit hindlimb postural asymmetry (HL-PA) with flexion of the limb contr alater al to the lesion (ie, contralesional), and asymmetry of reflexes with gr eater acti vity on the contra-vs.2][13][14] A cause of the contr alater al effects of br ain lesions has been consider ed as solel y neur oanatomical-based on the decussation of the descending neural pathways. 6 , 15 , 16merging evidence indicates that, in addition to neural mechanisms, the contralateral effects of a unilateral brain injury (UBI) are mediated through the humoral pathwa y b y neurohormones that produce either the left-or right-side-specific effects. 6 , 12-14This humoral signaling was identified in animals whose descending neur al tr acts w er e disa b led by complete transection of the spinal cord that was then followed by a brain lesion.Strikingly, rats with transected thoracic spinal cords and unilateral injuries of the hindlimb sensorimotor cortex developed contr alater al hindlimb flexion, asymmetry in hindlimb withdraw al r eflexes, and asymmetric changes in gene expression patterns in the lumbar spinal cord.Left-side brain injur y r esulted in right hindlimb flexion, while injury of the right hemisphere induced the left hindlimb flexion.Hypophysectomy abolished these effects, whereas serum from animals with UBI injected into rats with intact brains induced HL-PA in those animals.Arg-v asopr essin and β-endorphin wer e identified as molecules that mediate the effects of the left-sided brain injury.They ar e pr oduced in the hypothalamic-pituitar y system and evoke HL-PA with right hindlimb flexion in animals with intact brain. 12Thus, the left-right side-specific neuroendocrine signals ma y b ypass descending neur al tr acts and con ve y information on the side of brain injury.These neurohormones, released from the hypothalamus or pituitary gland, could be a part of a general mechanism that spans the nervous system, or even the entire body, and thus ena b les differ ential neur oendocrine contr ol of the left and right body sides.From a clinical standpoint, this phenomenon may contribute to asymmetric neurological deficits secondary to stroke and traumatic brain injury and may be pharmacolog ically targ eted by neurohormonal antagonists.How this topographic left-right side-specific neuroendocrine system (T-NES) is organized and functions is still an enigma.Thr ee sta ges ma y be en visaged: the encoding of signals from the left and right hemispheres into side-specific neurohormones in the hypothalamus and pituitary gland, the release of these neurohormones into the blood, and the decoding of these hormonal messages into left-right-sided responses in the spinal cord and peripheral nervous system. 6 , 12Differential encoding of top-down signaling from two anatomically symmetric hemispher es r equir es bipartite , later alized, and hemisphere (side) specific organization of the T-NES.
We reasoned that the T-NES consists of two counterparts that differ entiall y pr ocess and conv ey the left and right sidespecific messages, that their activities are balanced in intact rats, and that this balance may be perturbed by a UBI.These two parts may be mirror-symmetric in their structure , (e g, in cell type composition and connectivity), or they may differ in their internal ar c hitectur e and exploit differ ent endocrine, neural, and molecular mechanisms to produce symmetric physiological outcomes.
In this study, we addressed these hypotheses with an aim to c har acterize the T-NES counterparts and to r ev eal their lateralized features.The HL-PA, a proxy for neurological deficit with binary left-or right-sided outcomes including directional asymmetr y in postur e and motor functions, 11-14 , 17 w as used to c har acterize and compare the left and right T-NES counterparts in acute experiments.To analyze the T-NES, neural pathways between the brain and lumbar spinal cord were disa b led by complete spinal cord transection at the cervical level.These tr ansections w er e r ostr al to the thor acic pre ganglionic sympathetic neurons and allowed us to examine the paravertebral sympathetic chain of ganglia as the signaling pathway from the brain to the lumbar spinal cord.The left or right hindlimb sensorimotor cortex w as injur ed to evoke signaling through the left or right T-NES counterparts in order to separatel y anal yze their featur es.The cortex w as injur ed by a b lation in order to restrict the injured area to the hindlimb sensorimotor cortex and to examine specific changes in lumbar spinal circuits and hindlimb motor functions.In this biologically relev ant acute injur y model pathological factors that may interfere with the T-NES functions (eg, neuroinflammation, widespread dama ge to neur ons, axons, and b lood v essels) that ar e pr oduced by traumatic brain injury and stroke 18 , 19 could be largely excluded.
Analysis of signaling from the sensorimotor cortex injured on the left-or right-side demonstrated that the T-NES is binary and functionally asymmetric; the left-or right-side T-NES counterparts differ entl y target the contr alater al afferent systems controlling hindlimb functions.Experiments with opioid receptor antagonists confirmed that the T-NES is bipartite and that the messages from the left and the right hemispheres are differ entiall y contr olled thr ough the δ-, κ-, and μ-r ece ptors.Analysis of gene expression suggested that UBI affects the hypothalam us and pituitar y gland, that the side-specific molecular processes are coordinated between the hypothalamus and the lumbar spinal cord by the T-NES, and that this coordination is ipsilateral and impaired by UBI.

Animals
Adult male Sprague Da wle y rats (Taconic, Denmark) weighing 190-410 g were used in the study.The animals r ecei v ed food and water ad libitum and were kept in a 12-h day-night cycle (light on from 10:00 pm to 10:00 am ) at a constant environmental temperature of 21 • C (humidity: 65%) and randomly assigned to their r especti v e experimental gr oups.Appr ov al for animal experiments was obtained from the Malm ö/Lund Ethical Committee on Animal Experiments (No. M7-16).Experiments were performed from 9:00 am to 8:00 pm .After the experiments were completed, the animals were given a lethal dose of pentobarbital.

Spinal Cord Transection
The animals were anesthetized with sodium pentobarbital anesthesia (intraperitoneal, I.P.; 40 mg/kg body weight, as an initial dose and then 6 mg/kg ev er y hour).Core temperature of the animals w as contr olled using a feedback-r egulated heating system.
The experimental design included rats with UBI, which was preceded by a complete spinal cord transection.Anaesthetized animals were mounted onto the stereotaxic fr ame , and the skin of the back was incised along the midline at the level of the superior thoracic vertebrae.After the back m uscles wer e r etracted to the sides, a laminectomy was performed at the C6 and C7 vertebr ae .A 3-4-mm spinal cord segment between the two vertebrae was dissected and removed. 12The completeness of the transection was confirmed by (i) inspecting the cord during the operation to ensure that no spared fibers bridged the transection site and that the rostral and caudal stumps of the spinal cord were completel y r etracted; and (ii) examining the spinal cord in all animals after termination of the experiment.

Brain Surgery
The head of the rats mounted onto the stereotaxic frame was fixed in a position in which the bregma and lambda were located at the same horizontal level.After local injection of lidocaine (Xylocaine, 3.5 mg/mL) with adrenaline (2.2 μg/mL), the scalp was cut open, and a piece of the parietal bone located 0.5-4.0mm posterior to the bregma and 1.8-3.8mm lateral to the midline 20 w as r emov ed.The part of the cer e bral corte x located below the opening that includes the hind-limb r e pr esentation ar ea of the sensorimotor cortex was aspirated with a glass pipette (tip diameter 0.5 mm) connected to an electrical suction machine (Craft Duo-Vec Suction Unit, Rocket Medical Plc, UK).Care was taken to av oid dama ging the white matter below the cortex.After the a b lation, b leeding w as stopped with a piece of Spongostone, and the bone opening w as cov er ed with a piece of Tis-suDura (Baxter, Germany).For sham operations, animals underwent the same anesthesia and surgical pr ocedur es, but the cortex was not ablated.
After completion of all surgical pr ocedur es, the wound was closed with 3-0 suture (AgnTho's, Sweden), and the rat was kept under an infrared radiation lamp to maintain appropriate body temperature during monitoring of postural asymmetry and during stretching force analysis.

Dorsal Rhizotomy
A bilateral dorsal rhizotomy was performed in rats with complete transection of the cervical spinal cords 3 h after the UBI.After laminectomy from the T11 to L3 vertebral level, the dura was opened, and the dorsal roots were cut bilaterally from the L1 to S2 spinal levels with a pair of fine scissors as close to their exit as possible from the spinal column so that the spinal cord was not damaged.After each cut, the dorsal rootlets were flipped to make sure that the rhizotomy was complete.2][23][24] The HL-PA and stretc hing resistance w ere anal yzed befor e UBI, 3 h after UBI befor e rhizotomy, and 0.5 h after rhizotomy.

Histological Analysis of Brain Injury
Localization and size of cortical lesions wer e anal yzed in rats with left side ( n = 5) and right side ( n = 5) UBI 3-5.5 h after the injury.After perfusion with 4% par aformaldehyde , the br ain was r emov ed and postfixed in the same fixati v e ov ernight.Then the brain was soaked in phosphate-buffered saline with 30% sucrose for 48 h, dissected into blocks, which were then sliced into 50 μm sections with a fr eezing micr otome.Ev er y fourth section w as stained with toluidine (Nissl stain), and all the stained sections across the lesion site w ere photogr aphed, and the rostrocaudal r especti v e mediolateral extension as well as lesion v olume wer e calculated.

Analysis of HL-PA by the Hands-On and Hands-Off Methods
The HL-PA value and the side of the flexed limb were assessed as described elsewhere. 11 , 12 , 14Briefly, the measur ements wer e performed under pentobarbital anesthesia (40 mg/kg, i.p.).The level of anesthesia was c har acterized by a bar el y perce ptib le corneal reflex and a lack of overall muscle tone.The anesthetized rat was placed in the prone position on the 1-mm grid paper.
In the hands-on analysis, the hip and knee joints were straightened by gently pulling the hindlimbs backwards for 1 cm to reach the same level.Then, the hindlimbs were set free, and the magnitude of postural asymmetry (MPA) was measured in millimeters as the length of the projection of the line connecting symmetric hindlimb distal points (digits 2-4) on the longitudinal axis of the rat.The pr ocedur e w as r e peated 6 times in immediate succession.
In the hands-off method, silk threads were glued to the nails of the middle 3 toes of each hindlimb, and their other ends were tied to 1 of two hooks attached to the mov a b le platform that was operated by a micromanipulator constructed in the laborator y. 12 To r educe potential friction between the hindlimbs and the surface with changes in their position during stretching and after releasing them, the bench under the rat w as cov er ed with plastic sheet and the mov a b le platform was raised up to form a 10 • angle between the threads and the benc h surface .The limbs were adjusted to lie symmetrically, and stretching was performed over a distance of 1.5 cm at a rate of 2 cm/s.The thr eads then wer e r elaxed, the limbs wer e r eleased, and the r esulting HL-PA w as photographed.The pr ocedur e w as r e peated 6 times in succession, and the HL-PA values for a given rat were used in statistical analyses.
The limb that projected over a shorter distance from the trunk was considered to be flexed.The HL-PA was measured in mm with negati v e and positi v e HL-PA v alues that wer e assigned to rats with the left and right hindlimb flexion, r especti v el y.This measure, the postural asymmetry size (PAS), shows the HL-PA value and flexion side.The PAS does not show the proportion of the animals with asymmetry in each group, whether all or a small fraction of animals display the asymmetry; and cannot be used for analysis of rat groups with the similar number of left or right flexion.In the latter case, the HL-PA value would be a bout zer o.Ther efor e, the HL-PA w as also assessed by the MPA that shows absolute flexion size, and the pr oba bility of postural asymmetry ( P A ) that shows the proportion of animals exhibiting HL-PA at the imposed threshold ( > 1 mm).The MPA and P A do not show a flexion side.These 3 measures are obviously dependent; however, they are not redundant and for this reason, all ar e r equir ed for c har acterization of the HL-PA data and presentation.

Analysis of Hindlimb Resistance to Stretch
Stretc hing for ce w as anal yzed under pentobarbital anesthesia within 3-5 h after UBI using the micr omanipulator-contr olled force meter device constructed in the la borator y. 11 Two Mark-10 dig ital force gaug es (model M5-05, Mark-10 Corporation, USA) with a force resolution of 50 mg were fixed on a mov a b le platform operated by a micromanipulator.Three 3-0 silk threads were glued to the nails of the middle 3 toes of each hindlimb, and their other ends were hooked to 1 of 2 force gauges.The flexed leg of the rat in the prone position was manually stretched to the level of the extended leg; this position was taken as 0 mm point.Then both hindlimbs were stretched caudally, moving the platform by micromanipulator at a constant rate of 5 mm/s for 10 mm.No, or v er y little, trunk movement was observed with stretching for the first 10 mm; therefore, the data recorded for this distance were included in statistical analysis.The forces (in grams) detected by each of the 2 gauges were simultaneously recorded (100 Hz frequency) during stretc hing.F ive successive ramp-hold-r eturn str etches wer e performed as technical r e plicates.6][27] The r esistance anal yzed could hav e both neur ogenic and mechanical components, but their r especti v e contributions wer e not distinguished in the experimental design.The resistance was measured as the amount of mec hanical w ork W L and W R to stretch the left and right hindlimbs, where W was stretching for ce inte gr ated ov er the str etching distance interv al fr om 0 to 10 mm.
Doses and timeline for naloxone, 28 NTI, [29][30][31] BNI, [31][32][33] and FNA 30 r equir ed to b lock the r especti v e r ece ptors wer e r obustl y esta b lished in previous studies.The dose for naloxone was chosen to block all 3 subtypes of opioid r ece ptors.BNI and FNA exert long-lasting antagonistic effects that persist for at least 1 mo and ar e r ece ptor selecti v e fr om 24 h after administration.The anta gonists wer e purchased fr om Tocris (Minneapolis, MN).All test compounds were dissolved in saline.

Analysis of Gene Expression
Gene expr ession w as anal yzed in the pituitar y gland, and in the left and right halves of the hypothalamus and of the lumbar spinal cord.These tissues were collected 3 h after left UBI ( n = 12) or left sham surgery ( n = 11) that was performed in rats with complete transection of the spinal cord.The tissue samples were snap frozen and stored at −80 • C until assay.

Quantitative Real-Time PCR
Total RNA was purified by using the RNeasy Lipid Tissue Mini Kit (Qiagen, Valencia, CA, USA).RN A concentr ations w ere measured with Nanodrop (Nanodrop Technologies, Wilmington, DE, USA).RNA (500 ng) was reverse-transcribed to cDNA with the cDNA iScript Kit (Bio-Rad Laboratories, CA, USA) according to man ufactur er's pr otocol.cDN A samples w ere aliquoted and stored at -20 • C. cDN As w ere mixed with PrimePCR TM Probe assay and iTaq Uni v ersal Pr obes supermix (Bio-Rad) for qPCR with a CFX384 Touch TM Real-Time PCR Detection System (Bio-Rad Laboratories, CA, USA) according to man ufactur er's instructions.TagMan assay was performed in 384-well format with Ta gMan pr obes that ar e listed in Figur e 4 -figur e supplements S1 -S4 , S7 .
All pr ocedur es wer e conducted strictl y in accordance with the esta b lished guidelines for the qRCR based analysis of gene expression, consistent with the minimum information for publication of quantitati v e r eal-time PCR experiments guidelines. 34 , 35he raw qPCR data were obtained by the CFX Maestro TM Softw ar e for CFX384 Touch TM Real-Time PCR Detection System (Bio-Rad Laboratories, CA, USA).The mRNA levels of genes of interest were normalized to the geometric mean of expression levels of 2 r efer ence genes, Actb and Gapdh .GeNorm softw ar e w as used to analyze the gene expression stability ( M-value) of the 10 candidate r efer ence genes ( Actb, B2m, Gapdh, Gusb, Hprt, Pgk, Ppia, Rplpo13a, Tbp , and Tfrc ).The calculation of the M -value was based on the pairwise variation between 2 r efer ence genes.If the M-value was less than 1.5, it could be considered as a suita b le r efer ence gene.The smaller the M-v alue, the higher the stability of gene expression levels ( https://genorm.cmgg.be/and 36 ).The expr ession sta bility of candidate r efer ence genes w as computed for all sets of samples and identified Actb and Gapdh as the most sta b l y expr essed genes.For all 3 r egions anal yzed, the gene expression stability ( M-values) did not exceed 0.5.The optimal number of reference genes was determined by calculating pairwise variation ( V-value) by geNorm program.The V-value for Actb and Gapdh , the top r efer ence genes, w as 0.12 that did not exceed the 0.15 threshold demonstrating that analysis of these 2 genes is sufficient for normalization.
Genes selected as neur oplasticity-r elated wer e identified as suc h in sever al major studies.The selection of each gene from this set is justified by referring to these studies (see references below) and is not biased.To note, there is no established view on how to categorize genes as neuroplasticityrelated, and there are no lists of neuroplasticity-related genes consistent among studies.Thus, such a selection is arbitrary, and a set of selected genes could not be compr ehensi v e.The selected neur oplasticity-r elated genes wer e Arc , acti vityregulated cytoskeletal gene implicated in numerous plasticity paradigms; Bdnf , brain-deri v ed neur otr ophic factor r egulating synaptogenesis; cFos , a neuronal activity-dependent transcription factor; Dlg4 gene encoding PSD95 inv olv ed in AMPA r ece ptor-mediated synaptic plasticity and post NMDA r ece ptor acti v ation ev ents; Egr1 r e gulating tr anscription of growth factors, DN A damage , and isc hemia genes; Gap-43 coding for gr owth-associated pr otein Gap-43 that r egulates axonal gr owth and neur al netw ork formation; GluR1 and Grin2b coding for the glutamate ionotr opic r ece ptor AMPA Type Subunit 1 and NMDA r ece ptor subunit, r especti v el y, both inv olv ed in glutamate signaling and synaptic plasticity; Grin2a subunit of the glutamate r ece ptors that regulates formation of neural circuits and their plasticity; Homer-1 giving rise to homer scaffold protein 1, a component of glutamate signaling inv olv ed in nocice pti v e plasticity; Pcsk6 gene encoding pr opr otein conv ertase subtilisin/kexin type 6 inv olv ed in post-translational modification; Nfkbia (I-Kappa-B-Alpha) that inhibits NF-kappa-B/REL complexes regulating acti vity-de pendent inhibitor y and excitator y neur onal function; Syt4 (Synaptotagmin 4) playing a role in dendrite formation and synaptic growth and plasticity; and Tgfb1 that gives rise to transforming growth factor β1 regulating inflammation, expression of neuropeptides, and glutamate neurotoxicity [37][38][39][40][41][42][43][44][45][46][47][48][49][50][51][52] ( Figur e 4 -figur e supplement S4 ).
In the spinal cord, neur oplasticity-r elated genes ( Figure 4 figure supplement S4 ) along with the neur ope ptide and their r ece ptor genes ( Figur e 4 -figur e supplements S2 and S3 ) were analyzed besides the Avp, Avpr1b , Avpr2, Oxt , and Pomc genes that were expressed at low levels.

Sta tistical Anal ysis
Experimental data were processed and statistically analyzed after completion of the experiments by the statisticians who wer e not inv olv ed in their execution.No intermediate assessment was performed to avoid any bias in data acquisition.Experimenters were not blind because the signs of the asymmetr y wer e evident after brain injur y, and the UBI-animals with asymmetry exceeding 1.5 mm were selected for analysis of the antagonists.The asymmetry data were obtained by unbiased hand-off method and unbiased re gistr ation of stretc hing for ce .

Processing of Physiological Data
Ba yesian F ramework Predictors and outcomes scaled and centered before we fitted Bay esian re gression models via full Bayesian frame work b y calling Stan 2.21.7 (Stan Development Team 2022).RStan: the R interface to Stan.R package version 2.21.7 ( https://mc-stan.org/ ) from R 4.1.3(R Cor e Team 2022).R: A langua ge and envir onment for statistical computing.R Foundation for Statistical Computing, Vienna, Austria.URL ( https://www.R-project.org/ ) using the brms 2.18 53 interface.To reduce the influence of outliers, models used Student's t response distribution with identity link function unless explicitly stated otherwise.Models had no intercepts with indexing approach to predictors. 54Default priors were provided by the brms according to Stan recommendations. 55Interce pts, r esidual SD, and gr oup-lev el SD wer e determined \ fr om the weakl y informati v e prior Student t (3, 0, 10).The additional parameter ν of Student's distribution r e pr esenting the degr ees of freedom was obtained from the wide gamma prior gamma(2, 0.1).Gr oup-lev el effects wer e determined fr om the v er y weak informati v e prior normal (0, 10).Four MCMC chains of 40 000 iterations were simulated for each model, with a warm-up of 20 000 runs to ensure that effective sample size for each estimated parameter exceeded 10 000 56 producing stable estimates of 95% highest posterior density cr edib le interv als (HPD).MCMC dia gnostics were performed according to the Stan manual.P -values, adjusted using the m ulti v ariate t distribution with the same cov ariance structur e as the estimates, wer e pr oduced by fr equentist summary in emmeans 1.8.4-1 57 together with the medians of the posterior distribution and 95% HPD.The asymmetry and contrast between groups were defined as significant if the corresponding 95% HPD did not include zero and the adjusted P -v alue w as ≤.05.

Postural Asymmetry
The MPA w as inferr ed via Bay esian fr amew ork using Gaussian response distribution.The probability of HL-PA ( P A ) was inferred via Bayesian framework with Bernoulli response distribution and logit link function.

Stretc hing F or ce
The amount of mechanical work W L and W R to stretch the left and right hindlimbs, r especti v el y, w as computed by inte gr ating the smoothed stretching force measurements over stretching distance from 0 to 10 mm using loess smoothing computed by loess function from R package stats with parameters span = 0.4 and family = "symmetric."Asymmetry was assessed both as the left/right asymmetry index AI L/R = log 2 ( W L/ W R ), the contra-and ipsilesional asymmetry index AI C/I = log 2 ( W C/ W I ), and as the difference in work between left and right hindlimbs W L-R = ( W L -W R ) and between contra-and ipsilesional hindlimbs W C-I = ( W C − W I ).The AI and the difference in W were inferred via Bayesian fr amew ork by fitting linear m ultilev el models that included operation type (left UBI vs .right UBI vs .sham) as the factor of interest.

Molecular Analysis Expression Levels
The Lilliefors and Levene's tests re vealed de viations from normality and differences in the variances between the rat groups, r especti v el y, for the expression levels and the asymmetry index of several genes in each rat group.The mRNA lev els wer e compar ed se paratel y for the pituitar y gland and left and right halv es of the hypothalamus between UBI or sham surgery groups using Mann-Whitney test followed by Bonferr oni corr ection for a number of tests ( n = 17 and 56, r especti v el y).FC w as computed as a ratio of median expression levels in the UBI to sham groups.
The asymmetry index (AI L/R = median[log 2 L/R)], where L and R were gene expression levels in the left and right halves of hypothalamus or spinal cord, r especti v el y), w as computed for each gene in each ar ea ( Figur es 4 and 5 ), and compared between UBI and sham surgery groups using Mann-Whitney test followed by a Bonferroni correction for multiple tests ( n = 28 and 20 for the hypothalamus and spinal cord, respectively).Because no significant differences between UBI and sham surgery groups wer e r ev ealed, the gr oups wer e combined, and the pooled data were used for analysis of lateralization of gene expression.Onesample version of non-parametric Wilcoxon signed-rank test was applied to compare the AI L/R with zero, followed by Bonferr oni m ultiple testing corr ection (28 genes for hypothalamus).Data for the spinal cord wer e acquir ed and analyzed in our previous study. 12near model fitting and analysis were performed in R using lm, summary.lm, and confint commands.Differ ences wer e considered to be significant if the P -value corrected for multiple testing ( P adjusted ) did not exceed .05.

Gene-Gene Co-Expression Patterns
In the hypothalamus and spinal cord se paratel y, genes were categorized into 2 groups that were defined as the left (LdN; AI L/R > 0) and right (RdN; AI L/R < 0) dominant gene coexpression netw orks ( F igure 4 -figure supplement S9 ; Figure 5 figure supplement S1 ).Three categorization variants were used in the following correlation analysis.Genes were assigned into 2 groups: (i) by their median AI L/R in the combined UBI and sham surger y gr oup (v ariant 1); and (ii) by their median AI L/R in the sham surger y gr oup onl y (v ariant 2); and (iii) by their mean AI L/R in the combined UBI and sham surgery group (variant 3).Three v ariants wer e se paratel y applied for anal ysis of corr elation patterns in each the hypothalamus and spinal cord, and between these areas ( source data : the EXCEL source data files "Ta b le III-S6 23 05 10.xlsx").In the hypothalamus, all genes showed stable patterns between the LdN and RdN in 3 categorization variants besides Mor that wobbled.In the spinal cord, 5 LdN genes and 8 RdN genes showed sta b le patterns across the 3 variants, while 7 genes wobbled between the sides.The significant contrast: the P -v alue w as ≤ .05for (i) all 3 v ariants after corr ection, or for (ii) any 2 of them, while for the third variant it was < .05 and < .10before and after the correction, respectively.
The corr elation structur e (g ene-g ene co-expression pattern) for each area, side, between the sides, and across the areas was examined using the Spearman's rank correlation coefficient calculated for all gene pairs.The pattern of interactions between genes was c har acterized by the coordination strength (magnitude of correlations or the absolute value of the correlation coefficient av era ged acr oss pairwise corr elations) and the pr oportion of positi v e corr elations.
Robust and unbiased P-v alues wer e determined in the absence of distributional assumptions by permutation testing.A perm utation pr ocedur e w as employ ed to c har acterize the distribution of each statistical test under the null hypothesis of non-r e plication and non-pr eserv ation.Perm utation test 58 with R = 10 6 bootstrap r e plicates implemented in the R/boot packa ge w as used to analyze the data.To generate null distribution, we permuted the data across (i) rat identification numbers (IDs) , (ii) Treatment (UBI and sham surgery) , (iii) Module (left and right) within each individual rat; and (iv) CNS area (hypothalamus and spinal cord) within each module.
The R/boot.pval package was used to compute the P -value. 59P -v alues wer e adjusted using Benjamini-Hochberg family wise multiple test correction that was applied separately for all tasks (source data: the EXCEL source data files "Ta b le III-S6 23 05 10.xlsx").Three sets were designed to compare the coordination strength, and the other 3 sets to compare the proportion of positi v e corr elations.Two sets for both the hypothalamus and spinal cord were constructed for analysis of intra-area correlations in the strength and the proportion, and 2 sets for analysis of the inter-ar ea corr elations between the hypothalam us and spinal cord in the strength and the proportion.Each set included comparisons between UBI and sham surgery groups.
Task 2 .Pairwise g ene-g ene intra-modular correlations internal for each the left and right gene networks were compared between the left and right modules lm and rm, r especti v el y (lmLdN-lmLdN vs. rmLdN-rmLdN; lmRdN-lmRdN vs. rmRdN-rmRdN correlations) in each area separately to assess differences in the intra-modular coordination for each gene network between left and right modules ( Figures 4 and 5 ).
Task 3 .Pairwise g ene-g ene inter -modular correlations internal for each the left and right gene netw orks, w er e compar ed between these networks with each other (lmLdN-rmLdN vs. lmRdN-rmRdN correlations) in each area, to assess differences in the inter-modular coordination between the networks ( Figure 4  In each of 7 tasks, all correlation patterns were compared between the UBI and sham surgery groups.

T he Br ain Injury-Induced HL-PA in Ra ts With Transected Cervical Spinal Cord
The asymmetric effects of UBI on hindlimb posture and reflexes are mediated through the descending neural tracts and the humoral pathwa y b y the T-NES. 6 , 121][62] To explore this possibility, hindlimb responses to UBI wer e anal yzed in rats with a complete spinal cord transection performed at the level rostral to the preganglionic sympathetic neurons.A 3-4-mm segment of the C6-C7 spinal cord was excised, the hindlimb r e pr esentation ar ea of the sensorimotor cortex was then unilaterally ablated by suction, and HL-PA was analyzed.

Analysis of HL-PA
HL-PA w as anal yzed befor e (r eferr ed to as pr e) and 3 h after UBI or sham surgery (post) ( Figure 1 A and B) using both the handson and hands-off methods of hindlimb stretching followed by photographic and/or visual recording of asymmetry in animals under pentobarbital anesthesia. 11 , 12Data from these 2 methods correlate well with each other ( Figur e 1 -figur e supplement S2 ).The HL-PA data presented in Figure 1 and throughout the paper are for the hands-off assay.HL-PA was c har acterized by (i) the size of postural asymmetry (PAS) in mm, (ii) the MPA in mm, and (iii) the pr oba bility of developing HL-PA ( P A ).In contrast to the MPA, the PAS shows the direction of the asymmetry; negative and positi v e PAS v alues ar e assigned to left and right hindlimb flexion, r especti v el y.In the P A calculations, rats with MPA > 1 mm were defined as asymmetric; the 1 mm MPA was the 94th per centile in r ats before UBI or sham surgery and after sham surgery.
Figure 1 D and E shows the middle value of each group (median), a 95% highest posterior density cr edib le interv al (95% HPD), and the pr oba bility distribution of the middle value (posterior distribution) based on Bayesian r egr ession (see Glossary).95% HPD is an interval within which an unobserved parameter value falls with a 95% probability.It is analogous to confidence interv als.Animals wer e defined as significantl y asymmetric if 95% HPD in a group did not include zero and adjusted P -values were < .05.
Following UBI, but not sham surgery, rats with transected cervical spinal cords exhibited robust HL-PA with high statistical significance for both PAS and P A ( Figure 1 D and E).Hindlimb responses to brain injury were developed on the contralesional side; thus, left UBI induced right hindlimb flexion, and right UBI induced left hindlimb flexion.The PAS and P A in the UBI rats with cervical spinal cord tr ansection w ere similar to those pr eviousl y r e ported for UBI animals with intact and transected thoracic spinal cords. 12 , 13Figure 1 F and G shows the contrasts between animal groups for PAS and P A , denoted as PAS and P A , r especti v el y, along with the statistical significance for these contrasts (see Glossary).Each contrast is a difference between the median of the UBI group and the median of the sham surgery group.For example, the contrast PAS for "Post: L-UBI-Sh" is the median of the left UBI group minus the median of the sham surger y gr oup for PAS measur ed 3 h after brain surger y (post), and its P -value = 1e-10 after multiple correction.The PAS and P A of left and right UBI rats were significantly greater than those of sham surgery rats 3 h after brain surgery.Both P A and MPA did not differ between the left and right UBI groups, respectively.

Hindlimb Stretching Resistance
Next, we examined the effects of UBI in rats with transected cervical spinal cords on the biomechanical properties of the contraand ipsilesional hindlimbs ( Figure 1 C, H-J).Hindlimb passi v e m usculo-articular r esistance to str etch w as assessed in anesthetized rats before and 3 h after UBI or sham surgery.Leftright asymmetry in resistance was assessed as (i) the difference in work between the left and right hindlimbs as , where W L and W R were the work applied to stretch the left and right hindlimbs, r especti v el y ( Figur e 1 I); and (ii) the left-right asymmetry index for work as AI L/R = log 2 ( W L / W R ) ( Figur e 1-figur e supplement S3 ).Both the W L-R and the AI L/R wer e anal yzed because the y ma y de pend differ entl y on the stretc hing distance .The contr alesional-ipsilesional asymmetry Re pr esentati v e traces of the stretching force recorded from the left and right hindlimbs of rats with transected cervical spinal cords before and 3 h after UBI or sham surgery are shown in Figure 1 H.The force r equir ed to str etch the hindlimbs increased with the degree of stretch.No significant differences wer e observ ed in str etc hing for ce betw een the hindlimbs in r ats anal yzed befor e sham surger y and UBI, and in rats after sham surger y ( Figur e 1 H and I).The differ ence in the work r equir ed to stretch the left and right hindlimbs was strong and statistically significant 3 h after both left and right UBI, while no asymmetr y w as evident after sham surger y ( Figur e 1 I; Figur e 1figure supplement S3B ).The contrasts between both the left and right UBI groups and the sham surgery group in W L-R ( Figure 1 J) and AI L/R ( Figure 1-figure supplement S3C ) were robust and statistically significant.Both left and right UBI increased resistance to stretching of the contralesional hindlimb compared to the ipsilesional hindlimb.The stretching resistance's W C-I strongly correlated with MPA ( Figure 1 K).
We concluded that the hindlimb asymmetries induced by UBI were not mediated by the sympathetic nervous system or descending neur al tr acts.Instead, the humor al pathwa y ma y be the sole option for left-right side-specific signaling from the injured brain to hindlimb muscles in these experiments.

Effects of Bila ter al Deafferenta tion of Lumbar Spinal Segments on HL-PA Induced by the Left UBI or Right UBI
HL-PA Formation HL-PA ma y de v elop in r esponse to acti v ation of spinal r eflexes or changes in spinal circuits. 11We next sought to determine whether affer ent somatosensor y input is r equir ed for persistence of HL-PA induced by the UBI through a humoral pathway.Left or right UBI was performed in rats with transected cervical spinal cords, and the effects of bilateral rhizotomy of the dorsal r oots fr om the L1 to S2 lev els on HL-PA wer e anal yzed ( Figur e 2 ).In rats with left UBI, the PAS and P A were markedly reduced, 3.0and 3.5-fold, r especti v el y, after rhizotomy ( Figure 2 B and C).Contr ariwise , in the right-side UBI rats, the PAS and P A demonstrated only small, approximately 1.3-fold, decreases after rhizotomy.No signs of asymmetr y wer e r ev ealed in the sham surgery rats after rhizotomy.
Contrast in both the PAS and P A w as str ong and highl y significant between the UBI groups and the sham surgery group before the rhizotomy and between the right UBI and sham surgery group after rhizotomy ( Figure 2 D and E).The left UBI minus sham surgery contrast was negligible in rats analyzed after rhizotomy.The r elati v e impact of rhizotomy on the effects of left and right UBI w as anal yzed as contrast of contrasts ( Figure 2 F and G).Eac h contr ast of contr asts is a difference between the contrast befor e rhizotomy min us the contr ast after rhizotom y.For example , the contr ast of contr asts MPA for "Post-Rhz: L-UBI-Sh" ( Figure 2 F) is the contrast MPA "Post: L-UBI-Sh" measured before rhizotomy ( Figure 2 D) minus the contrast MPA "Rzh: L-UBI-Sh" measured after rhizotomy ( Figure 2 D), and P -value = 1e-10 for this MPA after multiple correction.The contrast of contrasts was high and significant when comparing the left UBI group to the sham surgery group, whereas it was much smaller when the right UBI rats were compared with sham rats.Most interesting, contrast of contrasts was high and significant for comparison of the right UBI group to the left UBI gr oup ( Figur e 2 F and G).

Stretching Resistance Analysis
The stretching resistance of the contra-and ipsilesional hindlimbs in rats with transected cervical spinal cords was anal yzed befor e and after the bilateral rhizotomy that was performed 3 h after UBI or sham surgery ( Figure 2 H-K; Figure 2 figure supplement S1 ).In rats with left UBI but not with right UBI, rhizotomy abolished the differences in resistance between the hindlimbs ( Figure 2 I; Figure 2 -figure supplement S1B ).No rhizotom y effects w ere evident in r ats with sham surger y.Wher eas contrast in both the W L-R and AI L/R between UBI groups and sham surgery group was strong and highly significant before rhizotom y ( F igur e 2 J), no differ ences wer e evident between left UBI group and sham surgery group after rhizotom y ( F igure 2 J; Figure 2 -figure supplement S1C ).Contr asts betw een the right UBI group and the sham surgery group remained strong and significant after rhizotomy.
Impact of rhizotomy (contrast: before vs. after rhizotomy) on the effects of left and right UBI (contrast: UBI vs. sham surgery) w as compar ed as contrast of contrasts in both the W CI and AI CI ( Figure 2 K; Figur e 2 -figur e supplement S1D ).The contrast of contrasts for the left UBI group was high and significant, while that of the right UBI group was noticeably smaller.Furthermore, the effects of left and right UBI in both the W C-I and AI C-I were differ entl y affected by rhizotomy as evident from the analysis of contr ast of contr asts ( F igure 2 K; F igur e 2 -figur e supplement S1 ).
Thus, the HL-PA and str etching r esistance data corr elate with each other and demonstrate that the effects of the left and right UBI mediated through the humoral pathway are differentiall y sensiti v e to bilateral deaffer entation.The effects of the left UBI but not the right-side UBI appar entl y de pend on affer ent somatosensory input.These results align with previous findings that in rats maintained for 3 d with intact spinal cords after UBI, bilateral lumbar dorsal rhizotomy eliminated HL-PA after the left but not the right lesion. 11The experiments were performed under isoflurane anesthesia.Hence, the sidedness of the effects of deafferentation does not depend on whether the spinal cord is transected before or after brain injury and on the type of anesthesia (ie, isoflurane or pentobarbital).

Effect of Opioid Antagonists on Hindlimb Asymmetry in Posture and Stretching Resistance Induced by Left and Right UBI
We pr eviousl y demonstrated that the selecti v e opioid anta gonists inhibited the formation of HL-PA after UBI in rats with intact spinal cords and that the effects of the antagonists were left-right side-specific. 13These findings suggest that the opioid system controls signaling from the injured area to spinal motoneur ons thr ough descending neur al tr acts and that the anta gonists interfer e with this pr ocess.Howev er, the asymmetric effects of UBI on hindlimb posture also were blocked by naloxone, a non-selecti v e opioid anta gonist, in rats with transected spinal cords indicating the inv olv ement of the opioid r ece ptors in neur oendocrine signaling fr om the injur ed brain through the bloodstream. 12Here, we examined whether the opioid system controls UBI effects mediated by the humoral pathway and determined whether this control is side-and r ece ptorsubtype-specific.The effects of the selecti v e μ-, δ-, and κ-opioid anta gonists β-funaltr examine (FNA), naltrindole (NTI), and norbinaltorphimine (BNI), r especti v el y, and naloxone on the asymmetry in hindlimb posture and stretching resistance in rats with completely transected cervical spinal cords were compared between the left and right UBI gr oups ( Figur e 3 ; Figur e 3 -figur e supplement S1 ).1][32][33] The r ats w er e administer ed with these anta gonists 24 h befor e UBI ( Figur e 3 A).NTI and naloxone wer e administer ed 3 h after UBI to rats that exhibited HL-PA with the MPA > 1.5 mm.HL-PA was then analyzed 1 h later ( Figure 3 B).
In the left UBI gr oup, ther e w as a significant reduction in MPA by 3.2-fold induced by NTI and 2.5-fold induced by BNI, while the FNA effects were not pronounced ( Figure 3 C and D).In contrast, in the right-side UBI group administration of FNA, but not of NTI or BNI, resulted in substantial MPA reduction (2.3fold).Data were analyzed for the MPA in order to compare the effects of opioid antagonists on HL-PA after left UBI and right UBI, which was not feasible with the PAS.Naloxone inhibited the effects of both the left-and right-side injuries (2.2-and 3.7-fold, r especti v el y).The effects of the antagonists were significantly different between the left and right UBI groups ( Figure 3 E).NTI and BNI pr efer entiall y inhibited formation of the right hindlimb flexion, whereas, in contr ast, FN A inhibited flexion on the left side.
Administration of BNI and NTI markedly decreased the contra-ipsilesional hindlimb asymmetry in stretching resistance W C-I after the left UBI, while their effects were minor in rats with the right UBI ( Figure 3 F-L).The FNA effects were less pronounced ( Figure 3 G, I, and K).Naloxone substantially reduced the W C-I in rats either with the left or right UBI ( Figure 3 H, J, and  L).Analysis of contrast of contrasts r ev ealed significant differences between the left and right UBI groups in the effects of BNI and NTI ( Figure 3 K and L).The effect of FNA on the W C-I after the right UBI slightly exceeds that after the left side injury ( Figure 3 K).
Thus, the HL-PA and stretching resistance data are in general a gr eement; the effects of the left and right UBI were differently inhibited by the opioid antagonists suggesting that the left and right T-NES counterparts are differentially controlled by the opioid r ece ptor subtypes.

Gene Expression and Co-Expression Patterns
Signals from the injured hemisphere may be encoded into leftright side-specific hormonal messages in the hypothalamicpituitary system that then are released into the blood. 12These messag es targ et the lumbar spinal cord and may produce lateralized changes in gene expression.As a result, the UBI-induced changes in the balance between the left and right gene expression patterns may be coordinated across the neuroendocrine and motor regions.The asymmetries w er e anal yzed after the transection befor e the UBI (pr e), and then 3 h after the UBI (Ant1).The control group Ctrl1 consisted of rats with UBI that were not treated with drugs.(B) Experimental design 2. The asymmetries were assessed 3 h after UBI (Post).The rats with MPA greater than 1.5 mm were selected for further analysis and treated with NTI or Nal (Ant2) or used as controls after saline treatment, or left untreated (Ctrl2).Asymmetries were analyzed 1 h later.Group design and the number of rats in the groups are given in Figure 3  The pr er equisite for the left-right side-specific encoding of neur ohormonal messa ges may be an asymmetrical organization of hypothalamic neur osecr etor y circuits, including their gene expr ession pr ofiles, and the side-specific r esponsi v eness of these circuits to a unilateral impact.We pr eviousl y r e ported that the "decoding" lumbar spinal cord is c har acterized by asymmetric gene expression patterns and that the UBI produced the ipsi-contralesional side-specific changes in gene expression and g ene-g ene co-expression in rats with complete spinal cord transection. 12 , 13 , 63ere , w e examined if UBI targets the hypothalamus and pituitary gland as the "encoding" areas by analysis of gene expression; i.e., if UBI affects gene expression in the ipsi-or contralesional hypothalamus; and if hypothalamic expression of neurohormonal and neuroplasticity-related genes is lateralized.To r ev eal r egulator y humor al inter actions betw een the "encoding" hypothalamus and "decoding" spinal cord, we then c har acterized g ene-g ene co-expression patterns between these regions along with perturbations in these patterns produced by UBI in rats with complete spinal cord transection.Tissue samples were collected 3 h after left UBI or left sham surgery that was performed in rats with completely transected thoracic spinal cord.Expression data for the lumbar spinal cord of these rats were taken from our previous study. 12The left UBI was used because its effects were stronger than those of the right-side injury in rats with intact spinal cords 11 , 13 and because it produced changes in left-right coordination of expression of the neuropeptide and neuroplasticity-related genes in the lumbar spinal cord of rats with transected spinal cords. 12e Hypothalamic-Pituitary System We reasoned that the "encoding" system that mediates the neuroendocrine UBI effects in the h ypothalamus in volves genes of the releasing and inhibitory hormones ( Crh , Ghrh , Gnrh1 , Sst , and Trh ), neur ope ptides and their r ece ptors genes ( Avp , Avpr1a , Nts , P enk , Pd yn , P omc , Oprm1 , Opr k1 , Oprd1 , and Oxt ; Figure 4 -figure supplement S1 -S3 ), along with neuroplasticity-related genes coding for regulators of axonal sprouting, synapse formation, neur onal survi v al, and neur oinflammation ( Arc , Bdnf , Dlg4 , Homer-1 , Gap43 , Syt4 , and Tgfb1 ), transcriptional regulators of synaptic plasticity ( cFos , Egr1 , and Nfkbia ), and essential components of the glutamate system critical for neuroplasticity ( GluR1 , Grin2a , and Grin2b ) ( Figur e 4 -figur e supplement S4 ).These genes were revealed as neuroplasticity-related in several studies each (for details of gene selection, see the "Materials and methods" section).
Expression of the Avp [fold change (FC) = 3.46x], Gap43 (FC = 1.18x), and Nts (FC = 1.32x) genes was significantly affected by the left UBI in the left hypothalam us ( Figur e 4 A-C).Also, the UBI effects were nominally significant for the Crh (FC = 1.76x),Sst (FC = 1.59x),Bdnf (FC = 1.25x),Syt4 (FC = 1.22x),Pomc (FC = 2.49x), and Ghrh (FC = 1.63x) genes ( Figur e 4 -figur e supplement S5 ).The expression levels of these genes were lower in the left hypothalamus in the UBI group compared to sham surgery group.In the right hypothalamus, expression of these genes w as decr eased but the UBI effects were not significant.Nonetheless, the UBI-induced changes were consistent between the left and right sides in their magnitude.Pearson and Spearman's rank correlation coefficients between log-scaled FCs induced by the left UBI in the left and the right hypothalamus were equal to 0.79 ( P = 5.4 × 10 −7 ) and 0.48 ( P = .010),r especti v el y ( Figur e 4 -figur e supplement S6 ).Fitting the data with a linear model with an arbitrary intercept (logFC right ≈ a logFC left + b ) resulted in estimates of a = 0.64 (95% CI [0.45, 0.83]) that was significantly smaller than 1.The estimate for b was close to zero (-0.02; 95% CI [ −0.12, 0.08]).Consistently, absolute values of the FC significantly differed between the sides (Wilcoxon signed rank test: P = .018).Thus, the effects of left UBI were significantl y gr eater in the left hypothalamus than in the right hypothalamus.
No significant differences in the AI L/R = log 2 [L/R] (where L and R denote expr ession lev els in the left and right hypothalam us, r especti v el y) wer e identified between UBI and sham groups, and these groups were combined for analysis of lateralization.Comparison of the AI L/R with zero identified 3 genes ( Pomc , P = .009;Trh , P = .014;and Ghrh , P = .014)with higher expression in the left hypothalamus while other 3 genes ( Grin2b , P = .002;GluR1 , P = .010;and Arc , P = .010)showed higher expression on the right side ( Figure 4 E).Lateralization was nominally significant for the Avp , Oprk1 , Syt4 , and Oxt genes that showed higher expression in the left hypothalamus, and for the Grin2a , Homer , Nts , Erg1 , Pcsk6 , Penk , Gnrh1 , and Dlg4 genes that demonstrated higher expression on the right side.
Thus, in the hypothalamus expression of a subset of the neur ohormonal, neur ope ptide, and neur oplasticity-associated genes was lateralized and affected by UBI on the ipsilesional side.Among genes responded to UBI in the neuroendocrine system were A vp , A vpr1b , and Pomc that give rise to Arg-vasopressin, the v asopr essin r ece ptor V1B and β-endorphin that, as we have demonstrated, mediate the effects of the left UBI on HL-PA through humoral pathway. 12

Gene-Gene Co-Expression within and between the Hypothalamus and Spinal Cord
Analysis of g ene-g ene co-expr ession patterns uncov er r egulatory inter actions betw een tissues and br ain ar eas. 11 , 64-67Her e, we evaluated if such patterns are coordinated in a lateral fashion across the hypothalamus and spinal cord, and if this coordination is mediated through humoral pathway and affected by left UBI.To take into account the lateralization factor, we se paratel y anal yzed genes with higher expr ession either on the left-or right-side of the hypothalamus and the spinal cord, and defined them as the left dominant network (LdN; AI L/R > 0) or right dominant network (RdN; AI L/R < 0) genes in eac h re g ion.We examined if g ene-g ene co-expression patterns differ between these networks and for both network between the sides in the hypothalamus and spinal cord; if the patterns are coordinated between these regions; and if the coordination is ipsi-or contr alater al and perturbed by a unilateral brain lesion.Gene-gene co-expr ession w as anal yzed by pairwise Spearman correlations.The coordination strength and the directions (signs) of interactions in the g ene-g ene co-expression patterns were assessed as mean of the absolute value of the correlation coefficient Rho and the proportion of positive corr elations, r especti v el y.All samples were dissected from the same rats with transected spinal cord that also had the left UBI or left sham surgery.Genes of the opioid and vasopressin systems were included because of their neur ope ptide pr oducts ar e inv olv ed in asymmetric spinal responses to brain injury.The set of neur oplasticity-r elated genes w as the same in the hypothalamus and the spinal cord (described in the "Materials and methods" section).

Categorization of Genes into the Left and Right Networks
Genes were categorized into the LdN and RdN in the hypothalamus and spinal cord separately.To avoid a bias, the categorization was performed in 3 variants: The LdN and RdN genes were defined by (1) their median AI L/R in the combined sham surgery and UBI group; (2) their median AI L/R in sham surgery group only; and (3) their mean AI L/R in the combined sham surgery and UBI gr oup ( Figur es 4 E and 5 A; Figure 4 -figure supplement S9 ; Figure 5 -figure supplement S1 ).A P -value between correlation patterns was determined by the permutation test with Benjamini-Hochberg family wise multiple test correction.The contrast was defined as significant using a stringent criterion: i.e., if the Pv alue w as ≤.05 for (i) all 3 v ariants after corr ection, or for (ii) any 2 of them while for the third variant it was < .05 and < .10before and after the corr ection, r especti v el y.Left-side gene expression was defined as left module (lm) of LdN and RdN networks, whereas right-side expression as their right module (rm).

Correlation Patterns in the Hypothalamus and Spinal Cord
We compared the LdN and RdN in their coordination strength and the proportion of positi v e corr elations for intra-modular correlations (LdN-LdN , RdN-RdN , and LdN-RdN) for both left and right modules se paratel y; these intra-modular correlations between left and right modules; and inter-modular correlations between the networks (ie, lmLdN-rmLdN and lmRdN-rmRdN, r especti v el y) ( Figur es 4 and 5 ; Figure 4 -Supplements S10 and S11 ; Figure 5 -Supplement S2 ).These comparisons were performed for both the sham surgery and the UBI groups separately, and between them.
In the hypothalamus, the permutation test revealed significant differences in the coordination strength between internal LdN and RdN correlations (LdN-LdN > RdN-RdN) in the rm of sham surgery rats, and between the left and right modules for both of the networks (lm < rm) in the UBI rats ( Figure 4 F).
The proportion of positive correlations in the hypothalamus r obustl y and significantl y differ ed between internal LdN and RdN correlations, and between them and mixed (LdN-RdN) corr elations ( Figur e 4 G).The differ ences wer e r ev ealed at all comparisons in both modules in sham surger y gr oup ( n = 6) and at most of them ( n = 5) in the UBI group.The pattern of differences was the same across the modules and animal group: the proportion for LdN-LdN correlations was significantly higher than that for RdN-RdN, and for both of them was higher than that for LdN-RdN.Furthermore, the proportion was significantly larger in the inter-modular LdN (lmLdN-rmLdN) corr elations compar ed to the inter-modular RdN (lmRdN-rmRdN) corr elations ( Figur e 4 -figure supplement S11 ).
In the spinal cord, the coordination strength was significantly higher in LdN compared to a mixed pattern (LdN-RdN) in the lm of sham surgery r ats ( F igure 5 B).The pattern of differences in the proportion of positive correlations was generally the same as that in the hypothalamus, however less contrasts were significant ( Figure 5 C).The proportion in LdN was larger than that in RdN and mixed LdN-RdN pattern in both sham surgery and UBI gr oups.Furthermor e, the pr oportion w as larger in the intermodular LdN correlations in the sham surgery group vs. UBI gr oup ( Figur e 5 -figur e supplement S2 ).
In summary, the LdN and RdN wer e markedl y and significantl y differ ent fr om each other in the coordination str ength and the proportion of positive correlations both in the hypothalamus and spinal cord.The UBI produced contrasting effects on the left and right hypothalamic modules in the coordination strength that were higher for both gene networks in the rm compared to the lm.For significant differences, both the coordination and the proportion were higher for the LdN compared to the RdN.Strikingly, correlations were largely positive within eac h netw ork and mostly ne gati v e between the networks, suggesting positi v e r egulator y inter actions among the genes in eac h network and negati v e r egulations between the networks.These differ ences ar e clearl y seen on heatmaps ( Figur es 4 H and I and 5 D and E).

Coordination of Gene Expression between the Hypothalamus and Spinal Cord
We ne xt e xamined if LdN and RdN are coordinated between the hypothalamus and the lumbar spinal cord through the humoral pathway, and if this crosstalk is perturbed by UBI ( Figure 6 , Figure 6 -figure supplement S1 ).We first analyzed the ipsilateral correlations internal for each LdN and RdN, between the left modules of the hypothalamus and spinal cord; and se paratel y between their right modules ( Figure 6 ).The LdN coordination strength in sham surger y gr oup w as asymmetric with higher level in the rm while the asymmetry was diminished after the UBI ( Figure 6 B).The proportion of positive correlations for both LdN and RdN in the sham surgery group was strongly asymmetric: It was much higher on the left vs .right side for the LdN, and, to the contrary, on the right vs .left side for the RdN.At the same time, the proportion and coordination str ength wer e quite similar between LdN patterns in the left modules and RdN patterns in the right modules, and vice versa between RdN patterns in the left modules and LdN patterns in the right modules ( Figure 6 B and C).
Left UBI impaired the hypothalamic-spinal cord crosstalk ( Figure 6 B and D).Notably, only the LdN patterns were significantly affected.UBI resulted in a strong decrease of the coordination strength and elevation of the proportion of positive correlations in the LdN patterns on the contralesional, right side.Contr ariwise , this proportion was strongly decreased on the ipsilesional, left side.Furthermore, in the UBI group, the proportion was higher for the right-side LdN pattern than for the left-side LdN pattern.
Analysis of the contr alater al (diagonal) correlations between the left hypothalamus and right spinal cord and between the right hypothalamus and left spinal cord for each the LdN and RdN did not r ev eal significant patterns and UBI effects ( Figure 6 -figure supplement S1 ).
In conclusion, the robust side-specific ipsilateral patterns in the coordination of gene expression between the hypothalamus and spinal cord that differed between LdN and RdN and strong perturbations in these patterns by the unilateral a b lation injury in animals with transected spinal cords were revealed.The findings suggest a functional link between these 2 regions that is the ipsilateral, left-right side-specific, and mediated by the endocrine signaling.

Discussion
The Left-Right Side-Specific Humoral Signaling from the Brain to the Spinal Cord: An Alternati v e to Neural Pathways Our earlier study showing that HL-PA and asymmetry in hindlimb reflexes occurred in rats with completely transected spinal cords was the basis for the hypothesis that the contralateral effects of unilateral brain lesions are mediated through humor al pathway. 6 , 12How ever, signaling from the brain to the lumbar spinal cord through the par avertebr al sympathetic chain was not excluded in this study because the spinal transection was performed at the T2-T3 level and the neural connections between the brain and the superior preganglionic neurons were left intact.Activity of the sympathetic preganglionic neurons located in the upper thoracic and lumbar segments is coordinated at a supraspinal (medullary) level 68 and muscle sympathetic nerve activity is controlled by central commands. 69Furthermore, the spinal somato-sympathetic nerve reflexes may contribute to the maintenance of muscle contractile force both before and after spinal cord transection. 60In the present study, the supraspinal part of the central nervous system was fully disconnected from the preganglionic spinal neurons by spinal cord transection at the C6-C7 level that w as r ostral to the thor acic pre ganglionic sympathetic neurons.Despite the complete transection, UBI still induced asymmetric hindlimb responses.Thus, the mechanisms of the brain injury-induced HL-PA formation mediated through both the par avertebr al sympathetic ganglia and descending neural pathways were ruled out.These experiments presented unambiguous proof of the left-right side-specific endocrine signaling in rats with transected spinal cords.

The Bipartite T-NES: Intrinsic Neurohormonal and Neural Asymmetry
The UBI-induced signaling is binary, either left-or right-sided.This could determine the bipartite structure of the T-NES that by encoding and decoding hormonal messages, may selecti v el y pr opa gate the effects of either left or right brain lesion ( Figure 7 A and B).The bipartite structure is supported by the findings that the left and right side-specific T-NES functions are differentially affected by selective opioid antagonists.The δ-antagonist NTI and the κ-antagonist BNI both inhibited HL-PA after left, but not after right UBI, whereas the μ-antagonist FNA interfered with the effects of right but not left UBI ( Figure 7 A and B).Of note, the opioid anta gonists differ entiall y b locked the effects of left and right UBI in rats with intact spinal cords. 13The signaling from the injured brain that causes the asymmetric hindlimb response in that study may be mediated by both neural and endocrine mechanisms.The pr eferr ed side of inhibition by δand μ-anta gonists w as the same in rats with intact 13 and completely transected (this study) spinal cords.However, the side affected by the κ-antagonist was different between these groups of animals.Thus, opioid mechanisms thr ough differ ent r ece ptor subtypes may allow flexibility of outcome based on synergy or antagonism of neural and endocrine pathways.
Translation of the T-NES humoral messages into the leftright side-specific hindlimb responses may occur in lumbar neural circuits or at peripheral nerve endings.We previously demonstrated that opioid peptides, synthetic opioids, and Arg-v asopr essin may induce HL-PA after their intravenous or intr athecal administr ation into r ats with intact br ain but with completely transected spinal cords. 12 , 14 , 70-73The striking finding was that the side of the flexed limb depended on the compound administered.The μ/ δ-opioid agonist Met-enkephalin, and the κ-opioid agonists dynorphin, bremazocine, and U-50,488 all induced flexion of the left hindlimb.In contrast, the δ-a gonist Leu-enke phalin, β-endorphin, and Arg-v asopr essin cause the right limb to flex. 12 , 71 , 72 , 74After brain injury, these neurohormones may be released from the endocrine glands into the b loodstr eam and induce side-specific effects thr ough their r ece ptors later alized in the spinal cord or the peripher al afferent and motor nerve terminals.In the cervical and lumbar spinal cord, the expression of the opioid r ece ptors is lateralized to the left, and the proportions of their subtypes and their coexpression patterns differ between the left and right sides. 13 , 63cti v ation of lateralized r ece ptors by endogenous neurohormones may r e pr esent a decoding mechanism in the bipartite T-NES.
Bilateral deafferentation of the lumbar segments in animals with transected cervical spinal cords did not interfere with HL-PA caused by the right UBI, suggesting that spinal reflexes are not inv olv ed, while the asymmetr y may persist due to acti v ation of motoneurons or changes in neuromuscular system.In contr ast, bilater al lumbar rhizotom y abolished HL-PA formed after the left UBI, suggesting that its maintenance r equir es an afferent input and depends on spinal reflexes.Thus, the left and right T-NES counterparts may act through different neural mechanisms and ena b le ov erall mirr or-symmetric outcomes that is flexion of either right and left hindlimb, r especti v el y ( Figur e 7 A and B).
The asymmetric spinal processing of the effects of UBI is in a gr eement with other findings regarding spinal cord asymmetries. 11 , 63 , 75-81Three-quarters of cervical spinal cords are asymmetric with a larger right side. 80Mono-and polysynaptic spinal reflexes showed rightward lateralization. 11 , 77-79Lateralized signals from an injured brain that target spinal circuits may be clinically relevant.Patients with stroke and cerebral palsy often do not relax their m uscles-they ar e tonically constricted without any voluntary command.3][84] This form of muscle overactivity may have a central mechanism 83 , 85 that does not depend on afferent input in contrast to spasticity based on exacerbated reflex excitability. 86In this respect, HL-PA developed in rats after the right-side injury may be mechanistically similar with spastic dystonia and could model this human neuropathology.These animal findings suggest that spastic dystonia may affect mor e fr equentl y the left than right lower limb in clinical cases.

La ter alized Crosstalk between the Hypothalamus and Spinal Cord
We pr eviousl y r e ported that gene expr ession patterns in the spinal cord are lateralized and affected by left UBI through a humoral pathway with clear differences between the contraand ipsilesional sides. 11 , 12 , 63Here , w e show ed that expression of the neurohormonal and neuroplasticity-related genes was also different between the left and right sides in the hypothalamus and affected by a unilateral cortical lesion in this area.Left UBI decreased the expression levels of a subset of these genes in the ipsilesional hypothalamus with no significant changes on the right side.In the pituitary gland, left UBI resulted in elevation of expression of Avpr1b that encodes the Arg-vasopressin V1B r ece ptor.Arg-v asopr essin may induce HL-PA with flexion of the right limb in rats with intact brain and mediate the effects of left UBI on the hindlimb posture. 12These effects wer e b locked by SSR-149415, the selecti v e anta gonist of the V1B r ece ptor that is mainl y expr essed in the anterior pituitary. 87It was hypothesized that Arg-vasopressin acting through the V1B r ece ptor on the pituitary corticotropes stimulates the release of the pr oopiomelanocortin-deri v ed β-endorphin that pr oduces HL-PA with the right hindlimb flexion. 12Changes in V1B r ece ptor expression suggest plasticity in the Arg-vasopressin system that signals from the injured brain.

The Left-Right Side-Specific Gene Co-expression Networks
Analysis of g ene-g ene co-expression patterns sugg ests that the neurohormonal, neuropeptide, and neuroplasticity-related genes form the left and right gene co-expression networks in the hypothalamus and spinal cord.It also suggests that these netw orks are later ally coordinated across these 2 regions and that UBI perturbs this coordination.In both regions, pairwise g ene-g ene correlations internal to the LdN and RdN were generall y positi v e, while those between the networks were mostly negati v e, suggesting an anta gonistic r elation between the 2 networks.The hypothalamus-spinal cord correlation patterns wer e strikingl y differ ent between the LdN and RdN.In the pr oportion of positi v e corr elations, the ipsilateral hypothalam usspinal cord co-expression pattern for the LdN genes displayed marked left-right asymmetry ( Figure 7 C).Similarly, the pattern of ipsilateral inter-ar ea corr elations for the RdN was also asymmetric, but the direction was opposite to that of LdN.At the same time, the patterns were almost mirror-symmetric between the LdN on the left side and the RdN on the right side, and between the RdN on the left side and the LdN on the right side ( Figure 7 C).These nearly perfect in their structure ensembles can be defined as "allo-symmetric."In contrast to the ipsilateral patterns, the diagonal (contralateral) inter-area correlations were similar between LdNs and RdNs and remained unaffected by UBI.Asymmetry of both LdN and RdN and "allo-symmetry" betw een them w ere c har acteristics of the contr ol gr oup and wer e impair ed by the unilater al br ain lesion.For example , the direction of the left-right LdN asymmetry was reversed after the left UBI.

Functional Implications
Formation of the LdN and RdN by neurohormonal, neuropeptide, and neur oplasticity-r elated genes, along with the asymmetry and "allo-symmetry" of their patterns were revealed in animals with completely transected spinal cords, suggesting that they wer e esta b lished by the T-NES and remodeled after TBI through the humoral pathway ( Figure 7 C and D).The functional role of the LdN and RdN may be to amplify the inher entl y weak lateralized effects of individual neurohormones and to strengthen the left and right side-specific regulations by these molecules.This mechanism could operate within and between the left and right halv es of CNS ar eas, and acr oss CNS r egions and their left and right sides along the neuraxis.
The LdN Avp and Pomc genes gi v e rise to Arg-v asopr essin and β-endorphin that induce right hindlimb flexion in animals with intact brain and may mediate the effects of the left-side UBI. 12 , 74 In contrast, Met-enkephalin and dynorphin deri v ed fr om P enk and Pd yn , constituents of the RdN, produce flexion of the left hindlimb. 14 , 70-72The hypothalamic neurohormones oxytocin and TRH, whose genes are constituents of the LdN, and gonadotr opin-r eleasing hormone , tr anscribed from the RdN Ghrh1 gene, produce lateralized responses and may regulate lateralized brain functions.The oxytocin r ece ptors mediate the effects of this pe ptide r eleased fr om the hypothalam us on pup r etriev al behavior thr ough acti v ation of the auditor y cortex on the left but not right side. 88][98][99] An intriguing possibility is that neurohormones and neur ope ptides with asymmetric actions may be organized into the left-and right-sided functional networks that control the entire left and right hemispher es, r especti v el y.These networks may differ between the hemispheres, whereas their inte gr al activities may be balanced between the left and right sides.1][102][103] The binary, the left-sided, or right-sided responses to peptide administration were assessed in rats with intact brain but completely transected spinal cords.Intr athecal administr ation of the extr acts r esulted in dev elopment of HL-PA.Remarka b l y, the direction of the asymmetry depended on whether an extract was prepared from the left or right hemisphere.The "left" extract induced flexion of left hindlimb, while the right hindlimb was flexed after administration of peptides from the right hemisphere.No asymmetry was formed after administration of the total peptide pool prepar ed fr om the whole brain.Thus, peptides with side-specific actions w ere later alized in the br ain, and the inte gr al activity of the "left" and "right" peptide factors was balanced between the hemispheres.Biochemical analysis demonstrated that factors inducing HL-PA were multiple short peptides that remain to be identified. 103

Limitations
T-NES was identified in acute experiments lasting 3-6 h after UBI in anesthetized male rats with completely transected spinal cords.Hence, a role for this phenomenon in the persistent biological and pathophysiological pr ocesses r equir es further investigation.This can be addressed by analyzing the protracted effects of unilateral TBI or stroke on contr alater al postur al and motor deficits in subchronic experiments with awake animals whose spinal cords are completely transected to disable neural pathways.The methodology may consist of a combined beha vioral, electroph ysiological, and biomechanical assessment of hindlimb function while performing body-weight-supported ste pping.Additionall y, it is necessar y to ev aluate the efficacy of T-NES in female animals, considering their different endocrine status.
HL-PA, a proxy for neurological deficits, enabled the discovery and c har acterization of T-NES.The model is binary, featuring 2 qualitati v el y distinct r esponses that ar e generated on either the left-or right-side.HL-PA can model human neurological deficits, such as hemiparesis, hemiplegia, and spastic dystonia secondary to TBI and stroke.It allows for reliable and quick testing of multiple hypotheses and is relatively easy to perform.On the other hand, HL-PA cannot be analyzed in awake animals, and knowledge of its mechanisms is limited.The neural pathways that may mediate signals from the injured brain region to the hypothalamic-pituitary system, the coding and decoding molecular mechanisms, and "left" and "right" hormones released by this system, as well as the afferent and central HL-PA mechanisms, still r equir e exploration.
In the molecular part of this study, the categorization of genes into LdN and RdN was based on the direction of their lateralization, but not on statistical significance or range of asymmetry.The selection of neuroplasticity-related genes was arbitrary, and the selected set was not comprehensive.Nevertheless, r obust differ ences between the LdN and the RdN in their intraand inter-ar ea corr elation patterns wer e uncov er ed, suggesting that a substantial part of these genes was correctly assigned to a r especti v e network.The anal yzed set of neur ohormonal and neur oplasticity-r elated genes, although r elati v el y small, allowed us to identify the LdN and RdN.Further transcriptome-wide analysis could reveal a complete structure of the left-rightspecific gene expression networks.

Conclusions
In addition to descending neur al tr acts, the contr alater al effects of brain injury may also be mediated by the endocrine system through the humoral pathway. 12A third potential signaling pathw ay thr ough the parav ertebral sympathetic chain of ganglia has been ruled out in the present study by the results obtained from rats with transected cervical spinal cords.Here, we uncovered the organizational principle of T-NES; in particular, its bipartite structure and functional and molecular asymmetries.The left and right T-NES counterparts mediate the effects of left and right brain injury, respectively, and enable overall mirrorsymmetric functional responses (eg, right and left hindlimb flexion).However, the neural and neurohormonal mechanisms underlying these responses are different.The maintenance of the left T-NES effects, but not those of the right T-NES r equir ed afferent input.Activation of motoneurons but not hindlimb reflexes may underlie the effects of the right UBI.Antagonists of δ-, κ-, and μ-opioid r ece ptors differ entiall y inhibited neur ohormonal signaling from the left and right hemispheres.Thus, endogenous opioid peptides may convey the "left" and "right" T-NES signals via the humoral pathway or differ entiall y contr ol their processing in the hypothalamus or spinal cord.
Analysis of g ene-g ene co-expression patterns revealed leftright side-specific gene co-expression networks and their ipsilateral coordination across the hypothalamus and spinal cord.The ipsilater al inter actions differed betw een the left and right gene networks on each side of the body and for both networks between the sides of the body.Left UBI perturbed these patterns by affecting the LdN.The findings suggest the side-specific ipsilateral endocrine crosstalk between the hypothalamus and lumbar spinal cord that coordinates molecular processes between these regions and is reorganized in the response to a unilateral brain lesion.
5][106][107] Lateralized processes may be regulated by the side-specific neurohormonal mechanisms that operate on either the left or right side. 6 , 14 , 63 , 76 , 88 , 89 , 91 , 108-113Our findings suggest a more general role for the lateralized neuroendocrine system.A fundamental feature of the bilaterian body is its symmetrical organization and function, which r equir es r obust contr ol of the balance between left-and right-sided processes.We hypothesize that the bipartite T-NES is part of this left-right-specific control mechanism.T-NES may be based on the lateralized neurohormonal networks and may act locally (eg, within brain and spinal cord areas) or along the neuraxis by signaling from the left and right hemispheres to the ipsilateral or contr alater al side of the body.A unilater al br ain lesion could shift this balance to the left or right, depending on the side of injury, and thereby disrupt leftright side-specific neurohormonal control, leading to asymmetric functional impairments.From a clinical point of view, it is essential to evaluate the contribution of neural and endocrine pathways to protracted neurological deficits after TBI and stroke, including hemiparesis and hemiplegia, and to develop pharmacological means to r estor e the impair ed neur ohormonal balance.

Asymmetry index
AI, the asymmetry index.AI L/R , the left / right asymmetry index computed as log 2 (L/R), where L and R are values for the left and right sides.The AI L/R w as anal yzed both for the r esistance to str etch ( W ) and for the gene expression levels.AI C/I , the contralesional/ipsilesional asymmetry index computed as log 2 (C/I), where C and I are values for the contralesional and ipsilesional sides.It was used in the analysis of W .
Brain injury UBI, a unilateral brain injury.L-UBI, a left unilateral brain injury.R-UBI, a right unilateral brain injury.

CNS areas
HPT, the hypothalamus.Pit, the pituitary gland.SpC, the spinal cord.

Correla tion anal ysis
Coordination str ength, ma gnitude of corr elations (a bsolute value of the correlation coefficients) averaged across pairwise correlations.
PAS, the postural asymmetry size in mm, with the direction of the asymmetry shown as negative values for left hindlimb flexion and positi v e v alues for right hindlimb flexion.
MPA, the magnitude of postural asymmetry size in mm.P A , the pr oba bility to develop HL-PA with the MPA > 1 mm.Musculo-articular resistance to stretching W , the work in gm × mm for passi v e hindlimb musculoarticular resistance to stretching.
W L-R , the difference in the work applied to stretch the left (L) and right (R) hindlimbs.
W L/R , the left/right asymmetry index computed as log 2 ( W L / W R ), where W L and W R are W values for the left and right hindlimbs.
W C-I , the difference in the work applied to stretch the contralesional (C) and ipsilesional (I) hindlimbs.

Molecular analysis
LdN, the left dominant gene co-expression network consisting of genes with the asymmetry index AI L/R > 0.
RdN, the right dominant gene co-expression network consisting of genes with the asymmetry index AI L/R < 0. lm, the left module constituted by gene transcripts of the left half of the hypothalamus or the spinal cord.
rm, the right module constituted by gene transcripts of the right half of the hypothalamus or the spinal cord.
Significant asymmetry, 95% HPD does not include zero and adjusted P -value < .05.
Contrast, the median of one group minus the median of another group.Denoted as .

Figure 1 .
Figure 1.Asymmetry in hindlimb posture (HL-PA) and stretching resistance (SR) induced by the unilateral ablation of the hindlimb r e pr esentation ar ea of sensorimotor cortex (UBI) in rats with completely transected cervical spinal cords.(A) Experimental design.The spinal cord was transected at the C6-7 level and then followed by either a left UBI (L-UBI; n = 10), right UBI (R-UBI; n = 12), or sham surgery (Sh; n = 7).The asymmetr y lev els wer e measur ed befor e (pr e) and 3 h after (post) UBI or sham surgery.(B) The UBI-induced HL-PA was manifested as flexion of the left or right hindlimb.(C) The str etching r esistance w as anal yzed as the amount of mechanical work W r equir ed to str etch a hindlimb, calculated as the inte gr al of stretc hing for ce over a distance of 0-10 mm.The force w as measur ed using a micr omanipulatorcontrolled force meter consisting of 2 digital force gauges fixed on a mov a b le platform.(D) The size of postural asymmetry (PAS) measured in millimeters (mm), and (E) the pr oba bility to dev elop HL-PA ( P A ) a bov e a 1 mm thr eshold (indicated in D by v ertical dotted lines).Negati v e and positi v e PAS v alues ar e assigned to rats with left and right hindlimb flexion, r especti v el y. (F and G) The contrasts in PAS and P A between the UBI groups and the sham surgery group are denoted as PAS and P A , r especti v el y, and wer e computed for the pr e and post time points se paratel y. (H) Re pr esentati v e traces of the str etc hing for ce r ecorded fr om the left and right hindlimbs before the UBI and sham surgery and 3 h after these surgeries.(I) Differences in stretching force between the left and right hindlimbs W L-R in gm × mm analyzed 3 h after UBI or sham surgery.(J) The contrast between animal groups in W L-R denoted as W L-R 3 h after UBI or sham surgery.(K) Pearson correlation between the magnitude of postural asymmetry (MPA) and the difference in work between the hindlimbs contralateral and ipsilateral to UBI and expressed as W C-I in gm × mm.The data pr esented ar e for left and right UBI groups analyzed 3 h after brain surgery.The median (represented as circles), 95% HPD (lines), and posterior density (distribution) from Bayesian regression are used to plot the PAS, P A , W L-R , and contrasts.Asymmetry and contrasts among the groups were deemed significant, with a 95% HPD not encompassing zero and adjusted P -values of ≤.05.Adjusted P -values are presented numerically on the plots.The PAS and W L-R values for individual rats are indicated by crosses in D and I. Source data: The EXCEL source data file "masterfile-210807.xlsx"and source data folder "/HL-PA/data/SF/."

Figure 2 .
Figure 2. Effects of bilateral deafferentation of lumbar spinal cord on HL-PA and hindlimb asymmetry in stretching resistance (SR) induced by left UBI (L-UBI) and right UBI (R-UBI) in rats with completely transected cervical spinal cords.(A) Experimental design.The spinal cord was transected at the C6-7 level, followed by either a left UBI ( n = 8), right UBI ( n = 11), or sham surgery (Sh; n = 7).The asymmetries were analyzed 3 h after UBI or sham surgery (post) and in the same rats after bilateral rhizotomy performed from the L1 to S2 spinal levels (Rhz).(B) The HL-PA size (PAS) in millimeters (mm) and (C) the pr oba bility to dev elop HL-PA ( P A ) a bov e the 1 mm threshold (denoted in B by dotted vertical lines).(D and E) The contrasts in PAS and P A between the UBI groups and the sham surgery group are denoted as PAS and P A , r especti v el y, and computed for the post and Rhz time points se paratel y. (F and G) The effects of rhizotomy on differences in the MPA (or P A ) between L-UBI, R-UBI, and sham surgery (Sh) analyzed as contrast of contrasts between (i) L-UBI and sham surgery: MPA (or P A ) = [(L-UBI Post -Sh Post ) -(L-UBI Rhz -Sh Rhz )]; (ii) R-UBI and sham surgery: MPA (or P A ) = [(R-UBI Post -Sh Post ) -(R-UBI Rhz -Sh Rhz )]; and (iii) R-UBI and L-UBI: MPA (or P A ) = [(R-UBI Post -L-UBI Post ) -(R-UBI Rhz -L-UBI Rhz )].(H) Re pr esentati v e traces of the str etc hing for ce r ecorded fr om the left and right hindlimbs of rats with UBI or sham surgery after rhizotomy.(I) Differences in stretc hing for ce betw een the left and right hindlimbs W L-R in gm × mm in rats with UBI or sham surgery after rhizotom y. (J) The contr asts in W L-R betw een the UBI and sham surgery groups denoted as W L-R in rats after rhizotomy.(K) The effects of rhizotomy on differences between contralesional hindlimb and ipsilesional hindlimb W C-I = ( W Contra -W Ipsi ) analyzed as contrast of contrasts between (i) L-UBI and sham surgery: W C-I = [(L-UBI Post -Sh Post ) -(L-UBI Rhz -Sh Rhz )]; (ii) R-UBI and sham surgery: W C-I = [(R-UBI Post -Sh Post ) -(R-UBI Rhz -Sh Rhz )]; and (iii) R-UBI and L-UBI: W C-I = [(R-UBI Post -L-UBI Post ) -(R-UBI Rhz -L-UBI Rhz )].The PAS, P A , MPA, W L-R , W C-I , and contrasts are plotted as median (circles), 95% HPD (lines), and posterior density (distribution) from Bayesian r egr ession.Asymmetr y and contrasts among the groups were deemed significant, with a 95% HPD not encompassing zero and adjusted P -values of ≤.05.Adjusted P-v alues ar e pr esented n umericall y on the plots.Crosses in (B) and (I) denote the PAS and W L-R values for individual rats, respectively.Source data: The EXCEL source data file "masterfile-210807.xlsx"and source data folder "/HL-PA/data/SF/."

Figure 3 .
Figure3.Effects of nor-binaltorphimine (BNI), β-funaltrexamine (FNA), and naltrindole (NTI), the selective κ-, μ-, and δ-opioid antagonists, respectively, and naloxone (Nal), the general opioid antagonist, on HL-PA and hindlimb asymmetry in stretching resistance (SR) induced by left UBI (L-UBI) and right UBI (R-UBI) in rats with completely transected cervical spinal cords.The spinal cord was transected at the C6-7 level, and then left or right UBI was performed.(A) Experimental design 1. BNI and FN A w er e administer ed 24 h befor e the tr ansection (Ant1).The asymmetries w er e anal yzed after the transection befor e the UBI (pr e), and then 3 h after the UBI (Ant1).The control group Ctrl1 consisted of rats with UBI that were not treated with drugs.(B) Experimental design 2. The asymmetries were assessed 3 h after UBI (Post).The rats with MPA greater than 1.5 mm were selected for further analysis and treated with NTI or Nal (Ant2) or used as controls after saline treatment, or left untreated (Ctrl2).Asymmetries were analyzed 1 h later.Group design and the number of rats in the groups are given in Figure3-figure supplement S1 .The direction of PAS in all animals in each group was the same; the left and right UBI rats exhibited positi v e and negati v e PAS v alues, r especti v el y. (C) The antagonist effects on the MPA.(D) Contrasts in the MPA between the r especti v e contr ol gr oups and the gr oups tr eated with anta gonists.Time points: 3 h after tr ansection for BNI and FN A (Ant1), and 4 h after transection for NTI and Nal, which was 1 h after their administration (Ant2).(E) Contrast of contrasts between the L-UBI and R-UBI groups in the effects of antagonists on the MP A: MP A = [(L-UBI Ant1 -Ctrl1) -(R-UBI Ant1 -Ctrl1)] for BNI and FNA; and MPA = [(L-UBI Ant2 -Ctrl2) -(R-UBI Ant2 -Ctrl2)] for NTI and Nal.The time points are the same as those in the (D).(F-H) Differences in stretching force between the contra-and ipsilesional hindlimbs W C-I in gm × mm.(I and J) Contrasts in the W C-I between the UBI groups treated with the antagonists and respective control groups.The time points are the same as those in (D).(K and L) Contrast of contrasts between the L-UBI and R-UBI groups in the effects of antagonists on the W C-I ; K: W C-I = [(L-UBI Ant1 -Pre1) -(R-UBI Ant1 -Pre1)] and L: W C-I = [(L-UBI Ant2 -Ctrl2) -(R-UBI Ant2 -Ctrl2)].Crosses denote the MPA and the W C-I values for individual rats.The median (represented as blac k cir cles), 95% HPD (black Figure3.Effects of nor-binaltorphimine (BNI), β-funaltrexamine (FNA), and naltrindole (NTI), the selective κ-, μ-, and δ-opioid antagonists, respectively, and naloxone (Nal), the general opioid antagonist, on HL-PA and hindlimb asymmetry in stretching resistance (SR) induced by left UBI (L-UBI) and right UBI (R-UBI) in rats with completely transected cervical spinal cords.The spinal cord was transected at the C6-7 level, and then left or right UBI was performed.(A) Experimental design 1. BNI and FN A w er e administer ed 24 h befor e the tr ansection (Ant1).The asymmetries w er e anal yzed after the transection befor e the UBI (pr e), and then 3 h after the UBI (Ant1).The control group Ctrl1 consisted of rats with UBI that were not treated with drugs.(B) Experimental design 2. The asymmetries were assessed 3 h after UBI (Post).The rats with MPA greater than 1.5 mm were selected for further analysis and treated with NTI or Nal (Ant2) or used as controls after saline treatment, or left untreated (Ctrl2).Asymmetries were analyzed 1 h later.Group design and the number of rats in the groups are given in Figure3-figure supplement S1 .The direction of PAS in all animals in each group was the same; the left and right UBI rats exhibited positi v e and negati v e PAS v alues, r especti v el y. (C) The antagonist effects on the MPA.(D) Contrasts in the MPA between the r especti v e contr ol gr oups and the gr oups tr eated with anta gonists.Time points: 3 h after tr ansection for BNI and FN A (Ant1), and 4 h after transection for NTI and Nal, which was 1 h after their administration (Ant2).(E) Contrast of contrasts between the L-UBI and R-UBI groups in the effects of antagonists on the MP A: MP A = [(L-UBI Ant1 -Ctrl1) -(R-UBI Ant1 -Ctrl1)] for BNI and FNA; and MPA = [(L-UBI Ant2 -Ctrl2) -(R-UBI Ant2 -Ctrl2)] for NTI and Nal.The time points are the same as those in the (D).(F-H) Differences in stretching force between the contra-and ipsilesional hindlimbs W C-I in gm × mm.(I and J) Contrasts in the W C-I between the UBI groups treated with the antagonists and respective control groups.The time points are the same as those in (D).(K and L) Contrast of contrasts between the L-UBI and R-UBI groups in the effects of antagonists on the W C-I ; K: W C-I = [(L-UBI Ant1 -Pre1) -(R-UBI Ant1 -Pre1)] and L: W C-I = [(L-UBI Ant2 -Ctrl2) -(R-UBI Ant2 -Ctrl2)].Crosses denote the MPA and the W C-I values for individual rats.The median (represented as blac k cir cles), 95% HPD (black lines), and posterior density (colored distribution) from Bayesian regression are used to plot the MPA, W C-I , and contrasts.Asymmetry and contrasts among the groups were deemed significant, with a 95% HPD not encompassing zero and adjusted P -values of ≤.05.Adjusted P -values are presented numerically on the plots.Source data: the EXCEL source data file "SDU-RDPA-Stat v2.xlsx" and source data folder/HL-PA-opioid-antagonists/data/SF/."

Figure 4 .
Figure 4.The UBI effects on gene expression patterns in the hypothalamus and pituitary gland.Analysis of the left (LdN) and right (RdN) dominant gene co-expression networks in the hypothalamus.(A-D) Gene expression levels in the left hypothalamus (HPT) and the pituitary gland (Pit) collected 3 h after left sham surgery (Sh: n = 11 rats) or left UBI ( n = 12 rats) that were performed in the spinalized rats.The expression levels in the log 2 scale and the Bonferroni adjusted P -values determined by Mann-Whitney test are shown.(E) The asymmetry index AI L/R = log 2 [L/R], where L and R are the median expression levels in the left and right hypothalamus, is shown for each gene analyzed.There were no differences in the AI L/R between sham surgery and UBI groups; therefore, they were combined ( n = 22) for statistical analysis.Wilcoxon signed-rank test followed by Bonferr oni m ultiple testing corr ection: * , P-adj < .05;* * , P-adj < .01;#, P ≤ .05(not adjusted).Boxes denote genes with AI L/R > 0 or AI L/R < 0 that were defined as the LdN and RdN genes.In (A-E), data are presented as boxplots with median and hinges r e pr esenting the first and third quartiles, and whiskers extending from the hinge to the highest/lowest value that lies within the 1.5 interquartile range of the hinge.(F and G) Patterns of intra-modular correlations internal for each the LdN (LdN-LdN) and RdN (RdN-RdN), and betw een the netw orks (LdN-RdN) are sho wn for the left and right modules of sham sur gery and left UBI groups.In (F), correlation strengths | Rho | (absolute value of each pairwise correlation) are presented as violin plots with white line indicating mean coordination strength.The proportion of positive correlations is shown as horizontal lines in (G).The 3 correlation patterns were compared within eac h (ie , left and right) module.Each pattern was compared between the modules and between UBI and sham surgery groups (for details, see Figure 4 -figure supplement S10A ).P -values were determined by permutation testing with Benjamini-Hochberg family wise multiple test correction.Significance for contrasts was determined by analysis of 3 AI L/R categorization variants ( Figure 4 -figure supplement S9 ); P -values are shown for the categorization variant with the median AI L/R of the combined sham surgery and UBI group.(H and I) Heatmaps for Spearman's rank coefficients for pairwise g ene-g ene correlations in the left-(lm) and right-(rm) modules of sham surgery and UBI groups.Source data: the EXCEL source data files "Hypoth SO UBI.xlsx; RD Hypophis Master file.xlsx;Ta b le III-S6 23 05 10.xlsx; raw groups.xlsx."

Figure 5 .
Figure 5. Gene co-expression patterns in the left and right lumbar spinal cord of the sham surgery and UBI rats.Analysis of the LdN and RdN gene networks.Expression of the neurohormonal and neuroplasticity-related genes was analyzed in the left and right halves of the spinal cord that were isolated 3 h after the left sham surgery ( n = 11) or left UBI ( n = 12) in spinalized rats.(A) Gene categorization using the AI L/R = log 2 [L/R], where L and R are the median expression levels in the left and right spinal cord into the LdN (AI L/R > 0) and RdN (AI L/R < 0).The median AI L/R is shown for the combined sham surgery and the left UBI group.There were no differences in the AI L/R between sham surgery and UBI gr oups; ther efor e , they w ere combined ( n = 22) for statistical analysis (for details, see Figur e 5 -figur e supplement S1 ).Wilcoxon signed-rank test followed by Bonferroni multiple testing correction: * , P-adj < .05;#, P ≤ .05(not adjusted).Data ar e pr esented as boxplots with median and hinges r e pr esenting the first and third quartiles, and whiskers extending from the hinge to the highest/lowest value that lies within the 1.5 interquartile range of the hinge.(B and C) Patterns of intra-modular correlations internal for each the LdN (LdN-LdN) and RdN (RdN-RdN), and between the networks (LdN-RdN).A violin plot for absolute Rho values of pairwise correlations, with horizontal line indicating overall coordination strength (defined as an av era ge of these absolute values) in (B) and the proportion of positive correlations in (C) for the left and right modules are shown for the sham surgery and UBI groups.The 3 correlation patterns were compared within each (ie, left and right) module.Each pattern was compared between the modules, and between UBI and sham surgery groups (for details, see Figur e 4 -figur e supplement S10B ).P -v alues wer e determined by perm utation testing with Benjamini-Hochberg family wise multiple test correction.Significance for contrasts was determined by analysis of 3 AI L/R categorization variants ( Figur e 5 -figur e supplement S1 ); P -v alues ar e shown for the categorization variant with the median AI L/R of the combined sham surgery and UBI group.(D and E) Heatmaps for Spearman's rank coefficients for pairwise g ene-g ene correlations in the left-(lm) and right-(rm) modules of sham surgery and UBI groups.Source data: the EXCEL source data file "SpinalC SO UBI Ctrl RD DD.xlsx; Ta b le III-S6 23 05 10.xlsx; raw groups.xlsx."

Figure 6 .
Figure 6.Ipsilateral coordination of the LdN and RdN between the hypothalamus and lumbar spinal cord.The effects of UBI.The experimental design and computation of LdN and RdN are described in Figures 4 and 5 .(A) Analyzed patterns of the ipsilateral pairwise g ene-g ene Spearman rank correlations between the hypothalamus (HPT) and spinal cord (SpC) on the left side ( α and β) and on the right side ( γ and δ).(B) The coordination strength and the proportion of positi v e corr elations for the correlation patterns depicted in (A).The correlation patterns were compared between the LdN and RdN ( α vs. β; γ vs. δ); each of them between the left and right modules ( α vs. γ ; β vs. δ), and all 4 patterns indi viduall y between UBI and sham surger y gr oups.P -v alues wer e determined by permutation testing with Benjamini-Hochberg famil y wise m ultiple test corr ection.Significance for contrasts was determined by analysis of 3 AI L/R categorization variants ( Figure 4 -figure supplement S9 ; Figure 5 -figure supplement S1 ); P -values are shown for the categorization variant with the median AI L/R of the combined sham surgery and UBI group.(C and D) Heatmaps for Spearman's rank coefficients for pairwise g ene-g ene correlations for the left-(lm) and right-(rm) modules in sham surgery and UBI groups.Source data: the EXCEL source data file "Hypoth SO UBI.xlsx; SpinalC SO UBI Ctrl RD DD.xlsx; Ta b le III-S6 23 05 10.xlsx; raw gr oups.xlsx."

Figure 7 .
Figure 7.A model of the bipartite asymmetric T-NES that ena b les the left-right side-specific signaling from the brain to the lumbar spinal cord (SpC) through the humoral pathway.The left (A) and right (B) T-NES counterparts mediate the contr alater al effects of the left-side and right-side brain injury.A unilateral brain lesion stim ulates the r elease of side-specific neur ohormones fr om the hypothalam us (HPT) and pituitar y gland (Pit) into the blood, they bind to neuronal receptors that are lateralized in the spinal cord 13 , 63 or peripheral neuronal endings, and induce contr alater al responses, e.g., hindlimb flexion.The δand κ-opioid r ece ptors may control signaling from the brain injured on the left side (left UBI), whereas the μ-opioid receptors may control signaling after the right UBI.The endogenous opioid peptides could differentially con ve y signals from the left and right hemispheres through the humoral pathway or control their processing in the hypothalamus or spinal cord.Although the left and right T-NES counterparts are not mirror symmetric to each other in their neural mechanisms, they produce overall symmetric functional responses, such as flexion of the right and left hindlimbs, respectively.The effects of the left T-NES but not right T-NES may r equir e an afferent input and depend on spinal reflexes.The right T-NES effects ma y de velop through activation of motoneurons or changes in the neuromuscular system.(C and D) The T-NESmediated ipsilateral crosstalk between the hypothalamic and lumbar spinal cord gene expression networks and its reorganization in response to a unilateral brain lesion.Ipsilateral interactions between the hypothalamus and spinal cord are depicted in each network as positi v e if the proportion of positive correlations is > 0.5 and negati v e if it is < 0.5.Arrows show the direction of changes in gene expression levels induced by the left UBI.The ipsilateral correlations significantly differ between the dominant left gene expression network and dominant right gene expression network on each body side and for both networks between the sides.Only interactions of the left networks wer e significantl y perturbed by the left UBI.The patterns of interactions were similar or almost mirror-symmetric (allo-symmetric) for the left networks on the left side and the right networks on the right side and for the right networks on the left side and the left networks on the right side.The diagonal (contr alater al) inter-re gional inter actions w ere not significant.