Customized Neurotization for Hindlimb Hemiplegia After a Hemorrhagic Stroke

Background Hemiplegia after hypertensive intracerebral hemorrhage(HICH) is prevalent,devasta-ting,and currently unable to intervene satisfactorily.Established success in neurotization spurs our interest in restoring skilled lower-limb movement.Here,we explored whether contralesional L4 transfer to ipsilesional L4 root,featuring simultaneous invigoration of exors and extensors,reanimated dexterous locomotion for the hemiplegic hindlimb after HICH. Methods We manufactured eligible rat models by injecting autologous vein blood into the left posterior limb of internal capsule (PLIC) and randomized thirty rats into three groups. We did not transect but expose bilateral L4 in group A (sham operation) after injection of saline,and infused equal blood followed by bilateral L4 transection and contralateral L4 transfer,in group B and group C,respectively.Behavioral,ultrastru-ctural,immunouorescent,and electromyographic assessments were applied to ecacy analysis.MRI localized the PLIC impact. Results We could see marked reduction in movements and capture a denervated potential in the right hindlimb of model rats (P (cid:0) 0.05).In the walking tests,group C exhibited an initially high slip,but,later, a larger increment in accuracy on the paretic side, compared to group B and group A(P (cid:0) 0.001).Also,at Week 17,rats gained ∼ 58.2% accuracy and actuated ankle motions in group C(P (cid:0) 0.05). In the retrograde tracing,at Week 9,uoro-gold labeled motoneurons had been numerously present in the left anterior horn of the spinal cord from the L4-to-L4 root.Histological and ultra-microstructural assessments certied ecient axons regeneration.Electrom-yography and pawprint analysis implied the denervated muscles were reliably reinnervated,and myodynamia improvement(p>0.05). Conclusions Contralateral L4 transfer

neurotization stays in early stage and yet need to be polished [23,24],its highlights have activated our aspiration for its extrapolation to a lower-limb paralysis after HICH,and further interrogation of maneuvers for a dexterous motion during neurotization.
Currently, there are few data on neurotization for lower extremities hemiplegia after HICH,only several types of nerve transfer,such as L6-to-L6,L4-to-L5,and L3-to-L4 in animals,as well as obturator nerve to femoral nerve in humans,are reported for repairing the paralyzed lower-limb after central neurological diseases [25][26][27][28].Disappointingly, only weak and rough motions were observed in the proximal major joints(i.e.,hip and knee), without activities in the distal joints(i.e.,ankle).One of main reasons consists in inadequate option of a source nerve.Anatomically, ne motor activities are principally accredited to a synergy between exors and extensors chie y composed of antero-posterior muscle groups(quadriceps femoris,semimembranosus,gastrocnemius,and tib-ialis anterior,etc.) in the lower extremity.Simultaneously,those muscles,coincidentally from the L4 root, drive distal joints activities.In brief,L4 root mainly bifurcates to compose both femoral and sciatic nerves and is theoretically understood as an optimal donor nerve,but need to be quali ed experimentally in humans or rodents even if the initial role of the L4 could be undertaken by other lumbar roots following transection.
As well-known,any practical technique will entail technological evolution in animal models preceding its extrapolation to patients.
Thus,in the study,we proposed the L4-to-L4 transfer prototype in a rat model,in which robust motions for the hip and knee joints and whole activities for the ankle were observed, and yet projected reliable data to researchers for positive reference.Besides, its feasibility and availability were con rmed by the behavioral,histological,and elect-romyographic evidence,as well as anatomic interpretations,also predicted a profound prospect.

Methods And Materials
Standard protocol approvals and registrations.

Animals
Thirty Sprague-Dawley male rats (250∼300 g) were used in this study,purchased from the Laboratory Animal Center at Nanjing Medical University,and housed three per cage in the controlled animal facility of Nanjing Medical University at 22 ± 1℃,humidity 60 ± 10% with water and food ad libitum. The animal care unit was sustained at a 12 h light-dark cycle, with lights on at 7:00 am. Animals were randomly assigned into three groups, each 10 rats. Group A was subjected to sham operation. Group B received bi-L4 transection, and Group C underwent L4-to-L4 root transfer following a hematoma-driven lesion in PLIC.
Noiseless room was kept for rat subjects.

Creation of an internal capsule insult
Based on the published neural tracing results [29,30],we created a hematoma-caused lesioning in the posteromedial area of PLIC.Rats were anesthetized with10% chloral hydrate (400 mg/Kg body weight) intraperitoneally with body temperature at 37℃±0.5 using a heating pad.Brie y,after restrained in a stereotactic frame(RWD-68025,Shengzheng Biotechnology Co.,Ltd.),a rat was made a scalp incision along the midline followed by hemostasis using ophthalmobipolar, drilled a small hole in the skull.We gradually inserted a hydraulic microinjector,perpendicularly,into the target area (AP -3.1 from bregma; ML ± 3.3 from the midline;DV -7.8; n = 30)( Fig. 1A-B),slowly injected ∼180 µl blood or saline into the target site for 15 min at a rate of 12 µl/min,with a 30G Hamilton syringe connected to an UltraMicroPump (WPI,Sarasota,FL,USA).The sham group received saline injection following by only exposure of bi-L4.Post-surgery,rats were transported to a recovery chamber with ketoprofen (2 mg/kg,i.m.) for analgesia for three consecutive days.Seven days post-injection,we used a T2-weighted MRI image to indicate location of the investigated area (Fig. 1C).Both behavioral and electromyog-raphic tests supported an adjudication of a rat model.

Choice of a donor nerve
In a preliminary experiment,twenty-four naive rats (weighed 250-300 g) were classi ed into two groups,each 12 rats.Unilateral L4 root was excised sharply in one group,with-out transection of all other lumbar roots,while in another group, the counterpart was retained with other lumbar roots excised completely (i.e.,L2,L3,L5,and L6).Postopera-tively, rats received an intramuscular injection of penicillin (80,000U per day) for three days,application of lidocaine ointment to the skin incision for ve days,underwent observations and measurements in a blinded fashion.Seven days later,behavioral and electromyographic tests con rmed eligibility of L4 root as a donor target.

Behavioral tests
We conducted the beam and ladder rung walking tests as well as footprint analysis as per the established protocols at baseline,9,13,and17 weeks,respectively [31]. Fifteen days before the baseline evaluations, rats were trained on a per-procedure basis. A schematic illustration of the corresponding protocol is exhibited in Fig. 1D.
In the walking-beam test, animals (n = 10/group) were trained to cross a horizontal beam prior to surgery. A single run was deemed as satisfaction if the animal walked through the beam incessantly at a constant gait. Nine satisfying runs in each animal were used to calculate their performance in a blinded manner by an experienced reviewer. The number of times that an animal slipped off the ledge with its affected hindlimb was recorded and then normalized for the sum of steps taken. Slips onto the ledge were rated as a full slip (given a score of 1) and a half slip (given a score of 0.5) was scored if the hindlimb touched a side of beam. A slipping ratio was designated as the times of slip per the sum of right hindlimb steps.
Likely,in the walking-ladder test,we trained rats (n = 10/group) to travel across a horizontally-placed ladder from a neutral cage to their home cage in advance.Traveling across the ladder at a uniform velocity was described as a favorable status,and nine quali ed runs per animal were collected for their performance calculations. A slipping rate was evaluated as the number of slip per the sum of right hindlimb steps.
In the footprint analysis, we enticed rats to walk across the gangway straightly for three times for acclimatization.Subsequently,rats were tested in a gangway coated by aligned white papers,with a black cage at the end,following the right forepaw and handsaw were rendered blue and red ink respectively.The eligible data were gathered for e cacy analysis of intervention.
We calculated open eld walking in conformity with the Basso,Beattie and Bresnahan (BBB) rating scale.we left individual rat subjects in an open eld (90 × 150 cm), observed for ve minutes,and rated hindlimb activities from 0 to 21.

L4 Nerve Roots Surgery Procedures
Two weeks after creation of rat hemiplegia models,we performed nerve root transfer in group C, conducted bi-L4 transection in group B,and carried out the only exposure of bi-L4 in group A.After anesthetized with 10% chloral hydrate intraperitoneally(0.4 ml /100 g),laid supine,shaved,and xed on the miniature operation table,animals were made a median incision 3∼4 cm long longitudinally in the abdomen,which centered on the L4 and paralleled the anterior superior iliac spine.Subsequently,bi-L4 nerve roots were seen under the operating microscope(SZ61,Olympus).In Group C,the left L4 root (intact side) was traced and transected as distally to intervertebral foramen as possible, while the right L4 root was severed as proximally as possible. The proximal stump of the left L4 nerve root was transferred to the distal stump of the right L4 root using 10 − 0 Prolene sutures,with the right proximal stump secured to the ambient psoas major. In Group B,the stumps of bi-L4 roots were xed with the muscle to avoid neural reconn-ection ( Fig. 1E-F). An absorbable hemostatic sponge was carefully stuffed around the operative eld before strict skin closure.

T2-Weighted MRI for localization
Seven days after blood injection,T2-weighted MRI showed the PLIC lesion on a Bruker Biospec 7-T MRI system. Animals were anesthetized with 5% and maintained with 2% halothane (in 30% O2:70% N2O, vol/vol), and were then intubated and mechanically ventilated at 65 beats/min.A second T2-weighted image set was gained throughout the lesion.
Electrophysiological evaluation at a set period Electromyographic examination Five days following the left PLIC insult, we conducted surface electromyograms for the identi cation of a rat model.Mildly anesthetized animals with 10% chloral hydrate (0.15 ml/100 g),we shaved the bilateral lower-limbs and placed a reference electrode and recording electrode on the lateral or medial thigh of the hindlimb detected, respectively. Rats were stimulated at an increasing intensity of current from 0.5 mA to1mA, with self-adhesive electrodes in place.Based on neuroanatomic association in lower extremities, we performed serial electromyography with a concentric needle in the anterior-posterior muscle groups 10 days following the severance of a lumbar root,recorded myokymic potentials and positive sharp waves to locate muscles from the L4, and achieved according data under the condition of the only excision of L4 or only retention of L4,respectively.At 9,13,and 17 weeks post-transfer,by acupuncture electromyogram, we tested the L4-supplied muscles in the case of early neurogenic or MUAP(compo-und muscular action potential,MUAP) changes,besides the denervated potentials.Add-itionally,we excluded positive sharp waves without myokymic potentials as a conf-ounding variable, assessed neurogenic or myogenic alterations of MUAPs,and nally detected a reinnervated potential in the investigated muscles at a scheduled time.
H-re ex recording in the gastrocnemius As such,after the aforementioned preparations,we directed a pair of receiving electrodes into the gastrocnemius in the right hindlimb and placed a stimulating electrode transcu-taneously behind the medial malleolus.All the electrodes linked with a four-channel electrophysiology instrument(Galileo NT LineKey, Italy).We stimulated the reinnerv-ated nerve using a single pulse (2 ms, 5 Hz) at an initial current of 0.1 mA, with 0.1 mA increment until reaching a maximum current,and determined latencies for H-wave and M-wave.
Immuno uorescence for regenerated nerves Rats were injected an overdose of 10% chloral hydrate intraperitoneally and transcardially perfused with 600 ml saline and 400 ml 4% paraformaldehyde solution.The regrowing section of a regenerative nerve,5 mm away from the coaptation site,was resected rostrally and caudally, immerged to 4% paraformaldehyde overnight at 4℃, followed by 30% sucrose in 0.1M phosphate buffer overnight at 4℃,and were sliced longitudinally at a thickness of 20 µm with a manual rotary microtome(LEICA CM1905).Every fourth section was chosen for immuno uorescence.Three chosen sections per segment were retained and processed on a per-procedure basis.Next,we incubated the specimens with microtubeassociated protein-2(MAP-2,Abcam) diluted at 1:250 in primary antibody dilution buffer,laid them on a reciprocal shaker running overnight at 4℃.The primary antibody was skipped in the negative control.Following MAP-2 incubation,carefully washed in 0.01 M PBS,and incubated with Alexa Fluor-693 (1:250, Sigma) at 37℃ for 30 min, tissue sections were added an anti-quenching reagent and quickly photographed under a uorescence microscope(LEICA DM2500).
Retrograde tracing of the motoneurons to the quadriceps femoris At week 9,13 and 17 post-transfer, we surgically exposed the quadriceps femoris on the paretic side in parallel with severance of other lumbar roots(i.e.,L2,L3,L5 and L6), infused 1 µl 4% uoro-gold(UE-F4040) into the muscle at 3 points in the treated rats,and nally closed the incision after saline wash.In a naïve rat,we infused the equal dose into the counterpart in the intact hindlimb for a control.Seven days later,we sectioned the lumbar spinal cord from the L4-to-L4 root at 20 µm thickness,taking pictures under confocal microscopy(Zeiss LSM880 with NLO & Airyscan).Soma and partial dendritic arbors tagged by uoro-gold were enrolled in this study.Under 10 objective lens,at 50 µm separation photograph,we captured the labeled motoneurons at 480 nm wave-length.z-stack was used for reconstruction of optical sections.
Ultrastructural assessment for the regrowing nerve To illustrate the reinnervated dynamics,the regrowing roots,3 mm away from the coapta-tion site,were transected rostrally and caudally for ultrastructural evaluation at a prescribed timepoint.Generally,nerve sections obtained were xed in 2.5% glutaral-dehyde, dehydrated in both graded ethanol and propanone,permeated with resin.Next, the specimens examined were subjected to longitudinal and transverse ultrathin slices, then stained with 3% uranyl acetate-lead citrate.Ultrastructures for the regenerative axons were observed under FEI (Tecnai Spirit Biotwin,USA),and captured at a various resolution.G-ratio is described as an axon diameter divided by a ber one,directly representing the size of a regenerative axon.We repetitively calculated it and deemed it as a convictive parameter re ecting axons remyelination.

Statistical Analysis
Differences among groups were calculated using repeated-measures ANOVA, with post-hoc LSD pairwise comparisons applied as appropriate. All data were expressed as the mean ± S.E.M., with SPSS 21.0 (SPSS Inc.) and image J (1.52p) used for data analysis. The statistical signi cance value was set at a p value of 0.05 or lower.

Data availability
Requests for anonymized data will be reviewed by the corresponding author.

Results
Identi cation of a right hindlimb paralysis Seven days after injury to the dorsomedial area of PLIC,animals had an onset of pronouncedly reduced activities in the right lower extremity,anterogradely galloped in circles centering on the right hindlimb.In the walking tasks,there was > 85% slip rate in the right hindlimb,as opposed to in the left lowerlimb(P<0.05).In footprint test, the fore-hindpaw prints were non-overlapping for at least 6 months, with a certain spacing (19.2 ± 0.7 mm)in between (P>0.05).Additionally,denervated MUAPs with a lengthened duration and shortened amplitude,were detected in the hemiplegic hindlimb,by contrast to in the left hindlimb,further corroborating a truth that fore-hindpaw print was separated.The results implied an establishment of a rat model (Fig. 2).
Availability of L4 nerve root as a donor nerve Ten days after the only transection of L4,with acupuncture electromyography,rats were markedly detected myokymic potentials and positive sharp waves in the quadriceps femoris,moderately in both semimembranosus and gastrocnemius, mildly in the tibialis anterior,and hardly in the biceps femoris,maximally scored 20 in BBB score scale.Inst-ead, in the context of retention of L4 root but excision of other lumbar roots, rats were captured normal potentials in the corresponding muscles,particularly in the quadriceps femoris,semimembranosus,gastrocnemius,and tibialis anterior,maximally rated 13 of BBB score.We could conclude from BBB score that under the only retention of the L4 nerve root, the hindlimb function might recover to ∼70% of a native state within 2 weeks,as compared to under the only excision of the L4 nerve root,and from the electromyography that the L4dominating muscles could actuate exion and extension for major joints in the hindlimb,especially for the ankle.These results demonstrated eligibility of L4 as a source nerve (Fig. 3).

Behavioral assessment among three groups
In the footprint test, the right fore-hindpaw interval is summarized in Supplementary Table 1. Also,inbetween space in Group C signi cantly diminished over time till an overlapping status emerged at 17 weeks post-surgery, in contrast to group B(P < 0.001).In the walking-beam and -ladder tests,the overall slip and error rate for the bilateral hindlimbs were not statistically signi cant between group B and C,and attained 90% at baseline (P > 0.05). At 1 week, the slip and error ratio in group B drastically declined to 44% ∼56% as compared to group C.At 3 weeks, animals in Group A achieved ∼95% accuracy,compared with in group B and C(P 0.05).At week 5,in group B and C, growing behavioral improvement nevertheless started to emerge,but higher increment in group B.Nonetheless,from 9th week onwards,animals in group C displayed larger reduction in the slip and error rate than did in group B (P<0.05).Although increase in accuracy after 9 weeks,animals in these two groups failed to improve to the level in group A at all time points.The BBB score was signi cantly higher at 17 weeks than at 9 and 13 weeks(P<0.001).The data exhibited effective motor function regeneration for the hemiplegic hindlimb through L4 transfer (Fig. 4).
Electrophysiological evaluation for contralateral L4 transfer.
In group C,the reinnervated MUAPs could be elicited with concentric needles in the quadriceps femoris, semimembranosus, lateral gastrocnemius and tibialis anterior of the right hindlimb in 4 out of 10 rats(40%) at 9 weeks ,in 7 rats (70%)at 13 weeks,and in 9 rats(90%) at17 weeks,post-intervention.Also,the denervated MUAPs detected in the target muscles markedly decreased temporally.In H-re ex recording,the M/H-wave latency for the right gastrocnemius was remarkably longer at week 9 than at week 17 (P < 0.05) (Supplementary Table 2).The latency shortened over time till approximating to a normal level (P 0.05).Together,the results suggested that the L4-to-L4 root e ciently reinnervated the targeted endorgans (Fig. 5).
Fluorogold-labeled motoneurons in the ventral horn from the L4-to-L4 root.
In the ventral horn,we could see many uorogold-labeled motoneurons,despite largest labeled area existed in the control(P>0.05).At 9,13,and 17 weeks, tagged moto-neurons increased temporally,in contrast with the contralateral side (P < 0.05).These results exhibited that a high proportion of regenerated axons e caciously reinnervated the motor endplates (Fig. 6).
Immuno uorescence staining and ultrastructure for the reinnervated nerve root At the coaptation site,regrowing axons were remarkably observed, with mature ones tagged by MAP-2 accounting for 25.61 ± 5.18% at week 9,38.52 ± 3.64% at week 13,and 57.89 ± 6.25% at week17.Additionally,increasing reinnervation between the stumps occurred temporally till reaching nearly utter reinnervation at week 17(P>0.05) (Sup-plementary Fig. 1).Simultaneously,ultrastructure for the regenerative nerves exhibited a coexistence of myelinated and unmyelinated axons,further suggesting an uneven axon regeneration at the three timepoints (1,081 ± 469,12,572 ± 953,and 15,928 ± 1147,P<0.05).Over time, g-ratio for a regrowing axon declined,in parallel with an elevation of myelinogenesis. Additionally,more reinnervated axons occurred in a realigned and rearranged manner. The results demonstrated effective regeneration of nerve bers (Supplementary Fig. 2).

Discussion
Given currently dissatis ed outcomes and large HICH populations, we need additional intervention to limb hemiplegia [32][33][34][35].Rationales behind the previous C7-to-C7 transfer for the paretic upper limb are beginning to be accepted and understood as scalability [36][37][38][39][40].Reinnervation of the denervated muscles,γ-circuit interruption,and establishing a new pathway remarkably contributed to a motor recovery post-transfer [41,42]. Contralateral lumbar neurotization engaging the similar mechanisms was considered available for a lower limb paresis after central neurological injury.Thus,in this project,we would rather extend the L4-to-L4 transfer to hindlimb hemiplegia than other transfer modalities in rats.
Why the L4 root was identi ed as an optimal source nerve,compared to other lumbar nerves? First,transecting the L4 root may not impair the lower extremity motion owing to compensation from neighbor roots,as shown by the BBB score.Second,biomechanic-ally,synergies between exors and extensors lead to a stable and re ned movement.An-atomically,the L4 is a sole root that simultaneously drives anteroposterior muscle groups,theoretically actuating robust exion and extension in a hindlimb.Finally,given the muscles innervated by the L4,including the quadriceps femoris,semimembranosus,gastrocnemius,and tibialis anterior,etc.,it was inferred that the L4 may not only power reliable proximal joints activities but also distal joints motions in the lower limb.The L4 as a donor nerve might represent a priority in functional recovery follo-wing nerve transfer.
In our study,we broke off the efferent pathway via damage to the posteromedial area of PLIC,to create a quali ed model rat,as indicated by electromyography and pawprint tests.Yet,only slight or no spasm was observed in a model rat.Besides,we presented the availability of the L4-to-L4 transfer for motor recovery. As well-known,a denervated muscle may be atrophic if a nerve ber fails to reinnervate at a certain time,i.e.,3 to 6 months.On the one hand,the results from H-re ex and retrograde tracing implied that the regenerative axons had strongly reinnervated the target muscles.on the other hand,based on a shortened latency for M/H-wave and g-ratio in ultrastructure temporally,we strongly speculated that nearly all regrowing bers could be myelinated and well-functioned.Finally,depending on an increasing regeneration between stumps, we suggested that larger numbers of regenerated bers reached the motor end-organ to generate better functional outcomes.Collectively,post-surgery,the axons could regen-erate into the target muscles to undertake due responsibility before their atrophy.
Here,we also demonstrated distinct observations from the L4-to-L4 transfer after a comparison of the other transfer modalities.Post-operation,in the distal major joints such as hip and knee joints,rats could conduct large-range,dexterous,and strong activi-ties.Additionally,in the ankle joint,full exion and extension were often seen at the early stage,as further implied in the walking-beam and walking-rung tests.As we know, central neurological injuries commonly cause motor control de cits in the whole leg,speci cally re ned ones.To achieve contiguous skilled locomotions,rats oblige to balance the weight support and true up limb placement rapidly.Hence, ne muscle,leg coordination,and ability to equilibrate weight-bearing stepping movements are required for a favorable performance in the walking tests.For that reason, movement accuracy and gait in the intact leg were provisionally negatively impacted as well.
Brie y, the extent to which action accuracy was improved and de cits in gait were rescued could be revealed by the skilled walking tasks,pawprint test,as well as acupuncture electromyography.Nevertheless,relatively long recovery time was taken after the contralateral L4 transfer.Also,on the intact hindlimb,a comparatively much loss of motor function emerged after transecting the L4,although it did not impair the locomotion.De-spite these shortcomings could not eclipse the neurotization application,we still hope that they will be overcome in the future by multidisciplinary evolutions.
Several limitations to this study remained to be solved in the future. Noteworthily,post-intervention,no signi cant motion was observed in digits, possibly due to a short exper-iment duration.The exact number of L4 nerve bers on the intact side innervating the contralateral muscles and the ratios to other lumbar nerve roots were yet not clari ed. As rats are quadrupeds, a lower-limb is more frequently implicated in the skilled motion in rats than in humans.Slight motor degradation tends to be more detectable in rats. However, the present study still extrapolates the concept to clinical practice.
In conclusion,contralateral L4 neurotization enables an effective motor recovery in the distal joint,intensi es performance in the proximal joints,and may be an optimal option for the hindlimb paralysis secondary to HICH.Further researches into the mechanisms underlying axonal regeneration and cortical plasticity in hindlimb recovery are inevitably desirable.       Also,in the walking tasks,no statistical signi cance in accuracy occurred at baseline between group B and group C.With time,a larger decline in slip rate for the affected side was observed in group C,as compared to in group B(P 0.05).Additionally,in group C,there was also a relatively high slip rate for the intact side owing to body imbalance.Later on,there was also improvement in accuracy. (E) Eventually,at week 17, rats in group C scored ~13 in the BBB scale score.These results implied a positive role of the intervention in motor regeneration.(BL=Baseline). Also,in the walking tasks,no statistical signi cance in accuracy occurred at baseline between group B and group C.With time,a larger decline in slip rate for the affected side was observed in group C,as compared to in group B(P 0.05).Additionally,in group C,there was also a relatively high slip rate for the intact side owing to body imbalance.Later on,there was also improvement in accuracy. (E) Eventually,at week 17, rats in group C scored ~13 in the BBB scale score.These results implied a positive role of the intervention in motor regeneration.(BL=Baseline).