Efficacy of chronic ultrasound neurostimulation on behaviors and distributed brain metabolism in depressive-like mice

Major depression is one of the main factors contributing to the Global Burden of Disease. Current treatment strategies (e.g., antidepressants and neurostimulation techniques) of major depression show some limitations including inaccuracy and invasiveness. Ultrasound neurostimulation (USNS) has been recently introduced as a physical non-invasive method for brain tissue stimulation and has gained increasing interest. In this study, we sought to evaluate the efficacy of transcranial USNS in an unpredictable chronic mild stress (UCMS) mouse model. The results show that transcranial USNS of the infralimbic cortex reduced anxiety-related behaviors as well as some, but not all, depression-related parameters. [18F]-FDG microPET imaging and brain metabolomic analyses showed that USNS triggered the activation of targeted brain region in addition to brain areas at a distance from the targeted zone, alleviating anxiety and depression-related behaviors induced by the UCMS regimen. Transcranial ultrasound neurostimulation show therapeutic potential in some aspects of major depression.


Introduction
second most prevalent cause of illness-induced disability (1), which makes this disorder one of the 34 main contributors to the Global Burden of Disease (2). It is generally treated with chronic 35 antidepressants (ADs), which consist of drugs increasing monoaminergic neurotransmission such 36 as selective serotonin reuptake inhibitors (SSRIs). However, nearly 65% of the patients do not 37 respond to this first-line therapy, and it is established that 30-50% of patients are resistant to AD 38 compounds (3,4), which means they do not show remission after treatment with several ADs: this 39 condition is conventionally referred to as treatment resistant depression (TRD). Furthermore, MD 40 often presents itself with a high and disabling comorbidity to anxiety disorders, where SSRIs show 41 a less reliable spectrum of therapeutic efficacy (5). 42 In the past decade, progress in the treatment of TRD has involved neurostimulation, which consists 43 in activating/inhibiting the cerebral networks whose functioning is modified in MD. Regions of 44 interest include the dorsolateral prefrontal cortex (dlPFC), the subgenual part of the anterior 45 cingulate cortex (sgACC), the nucleus accumbens or the lateral habenula (6). While the dlPFC is 46 a cortical area that can be targeted using neurostimulation methods such as repeated transcranial 47 The functional focality and effects of acute repeated USNS was evaluated in eight mice under 1% 126 isoflurane. From single-pulse USNS data and previous studies (16), the pattern of acute USNS was 127 constructed as a 10-min session of 60 USNS pulses (1 pulse: 400 kPa, 160-msec) repeated at a rate 128 of 0.1 Hz and delivered over the prefrontal cortex under 1% isoflurane anesthesia. To  The immunolabelling of reactive c-Fos neurons revealed that acute USNS elicited neural activation 136 in prefrontal regions. The activity index (i.e. the effect size of acute USNS) was mildly increased 137 in M1 (0.96), motor cortex M2 (1.34), cingulate cortex (Cg, 1.02) and prelimbic cortex (PrL, 0.52), 138 despite showing no statistical difference with Sham mice ( Figure 2B). Otherwise, the activity index 139 was further increased in IL (2.67) and multiple analyses revealed that acute USNS elicited 140 significant c-Fos activity along antero-posterior axis of the structure (F(1,3)=14.6, p=0.032). At 141 distance from prefrontal structures, acute USNS did not evoke specific c-Fos activity in the 142 olfactory areas (OL) nor in the auditory cortex (Ad; Figure 2C). In connected subcortical regions, 143 acute USNS induced significant c-Fos activity in subfields of the dorsal/ventral hippocampus 144 (Table II). 145

160
Chronic repeated USNS treatment alleviates anxiety-related behaviors 161 The effects of chronic repeated USNS treatment was evaluated in the main cohort of fifty-three 162 mice subjected to the UCMS regimen from day 0 to day 35. At day 7, UCMS mice were 163 significantly different from non-stressed mice in terms of coat state deterioration, a standard 164 measure of UCMS onset and evolution (30) (Supplementary C); from this timepoint, UCMS mice 165 were semi-randomly distributed into the four following treatment/control groups. From day 29 to 166 day 33, chronic USNS (i.e. 5 sessions of acute USNS) was applied to fifteen mice while fourteen 167 mice followed a sham condition ("Sham"). From day 38 to day 43, mice were tested for depressive-168 like and anxiety-related behaviors in several paradigms. The effects of chronic USNS were 169 assessed against a classic AD compound (fluoxetine, "Flx"), administered chronically through 170 drinking water from day 7 on and controlled (vehicle, "Veh"; Figure 3A). To provide ground 171 values, a group of twelve non-stressed, untreated mice ("Naive") were processed at the same age 172 through the same behavioral paradigms. 173 Overall, both treatments modified behaviors to a different extent. The ability for nest-building, a 174 daily-living measure classically affected by chronic stress (31, 32), was modified by treatments (F 175 (3, 49)=7.9, p=0.0002); chronic USNS mice built the nest faster than Sham mice (p=0.0005), 176 whereas Flx mice did not perform over Veh mice ( Figure 3B). Mice were then tested for 177 depressive-like behaviors in a reward-maze paradigm, built of three successive chambers with a 178 palatable biscuit laid in the center of the furthest (30). The relative consumption of the reward, 179 depending on the latency to reach said reward, was modified by treatments (F (3, 49)=5.

255
Discussion 256 Numerous brain structures are known actors in the pathophysiology of MD, yet the growing need 257 to act therapeutically on these regions remains only partially answered by current ADs and 258 neurostimulation techniques. The present study used repeated mechanical US waves to non-259 invasively target the IL/sgACC in a mouse model of MD. Chronic repeated USNS treatment 260 impacted various behavioral endpoints induced by the UCMS regimen, which was associated to 261 short-term changes in brain metabolic activity on the site of stimulation (prefrontal cortex and its 262 close surroundings) but also at distant, connected limbic regions, such as the striatum, the dorsal 263 hippocampus and the raphe nucleus. Measures of well-being (nest-building) and anxiety-related 264 behaviors (open-field task) were ameliorated by chronic USNS, while items relative to anhedonia 265 and reward-seeking were not readily modified compared to classic SSRI treatment (fluoxetine). 266 Metabolites were modified at long-term in cortical regions on target site and in subcortical regions 267 in the amygdala and the hippocampus, involving glutamate pathways that might correlate to 268 longer-term changes in brain plasticity. 269 The ability of US waves to reliably stimulate cortical regions through the cranium has been 270 reproduced in the current study with single-pulse USNS of the primary motor cortex M1. to the motor success at low dose anesthesia (26), the quantitative analysis of MEPs showed that 273 the strength of the stimulation and its reliability over multiple stimulations might depend on other 274 factors than acoustic pressure alone. Consistent rising of motor success with acoustic pressure 275 points toward an activation of glutamatergic neurons, but the inconsistency of MEP amplitudes 276 with higher pressures (>400 kPa) could reflect the recruitment of different neural populations 277 depending on US parameters, even though desensitization biases were controlled with a 10-sec 278 spacing of US pulses (16). Such variations upon intensity/pulse length were observed for different 279 neurostimulation approaches (i.e. rTMS) and might apply to USNS (33, 34). Thus, the 280 determination of optimal US parameters in terms of pressure/pulse length appears pivotal to induce 281 reliable brain stimulation with US waves. Furthermore, the deeper anesthesia used during the MEP 282 procedure (1% isoflurane) could have modified the quality of motor signals at pressures above 400 283 kPa and so despite higher motor success rates at 0.1% isoflurane; the concentrations used in most 284 studies for MEP analysis (<0.25 %) were too low to fit those of the chronic USNS paradigm (1%), 285 where a semi-awoken state could impart anxiety biases. We thus reckoned that the current findings 286 at 1% isoflurane could be more readily transposed to the treatment condition of chronic USNS. 287 Concurrently, motor responses could be diminished by slow-moving cortical spreading 288 depolarization (14), occurring for lastly tested pressures (500 kPa), which further underpin the 289 need to finely tune US pulses in terms of intrinsic parameters and sustained effects (35). 290 When applied for one unique session, acute USNS was able to evoke neural activity in the IL of 291 mice with little spatial inaccuracy at stimulation site: surrounding brain regions such as the PrL or 292 the Cg, M2 and M1 were not readily affected by the stimulation in comparison to sham-treated 293 mice. On the other hand, distant hippocampus regions were affected by acute USNS. Given the 294 geometric properties of the US beam, the results suggest that, similarly to standard M1 stimulation, 295 a threshold pressure appears for patterns of repeated stimuli, under which the activation of a brain 296 region was not seen on immediate c-Fos labelling; the lack of c-Fos activation in other prefrontal 297 regions suggests that the spatial resolution of acute USNS is not directly correlated to its functional 298 resolution. The narrower range of effective acoustic pressures could be responsible for the specific 299 stimulation of the IL (1-mm wide at 400 kPa). Furthermore, the auditory cortex was not affected 300 by acute USNS, which further supports a functional targeting of the IL (36, 37). Because c-Fos 301 labelling does not discriminate between glutamatergic and GABAergic neurons, the results show 302 that brain regions might react differently to acute USNS. As pulsed neurostimulation is described 303 to preferentially act on axons rather than somas (38), the reactivity of a brain region to the 304 application of USNS might depend on the orientation of the stimulation and the frequency of pulse 305 repetition as it has been shown for rTMS (39), highlighting the discrepancy between geometric 306 and functional focality when applying repeated, sustained stimulation. 307 When applied chronically for 5 days, behavioral measures were modified by repeated USNS. As 308 an indicator of well-being in rodents, the onset of nest-building was reduced by chronic USNS, 309 while it was not affected by fluoxetine (similar effects of the drugs were previously reported (31)). Likewise, glutamate pathways were similarly modified in all studied regions, and more glutamate levels were reported to increase in the serum and in frontal regions of MD patients (46, 363 47). In rodents, glutamate is implicated in the expression of depressive-like and anxiety-related 364 behaviors (48). Furthermore, glutamate pathways in the hippocampus, but not the prefrontal 365 cortex, might be pivotal to antidepressant response in SSRI-treated mice (49). The current findings 366 thus show that the effects of chronic USNS on glutamate pathways could be associated to a 367 therapeutic response similarly to classic fluoxetine treatment, although both treatments did not 368 modify the same behaviors in the UCMS model. In addition, our results showed that other 369 hippocampal functions, such as neurogenesis, were not modified by chronic USNS, which further 370 supports that the behavioral effects were mainly observed on anxiety and not directly on anhedonic 371 features. In the present study, few metabolic modifications were found in the Cg of stimulated 372 mice, even though this structure was in the propagation path of the US beam during stimulation. We should emphasize that microPET imaging was performed at day 36 while behavioral tests were 379 carried out from day 38 to day 43, and metabolomic analyses performed after brain harvest at day 380 43. Further investigation of the effects of chronic USNS treatment on the brain is required to 381 explore potential effects and therapeutic outcomes. Stimulation of the M1 highlighted the need to 382 identify a precise threshold that does not exceed or fall short the efficient intensity. The 383 metabolomic analyses suggested that USNS did not induce excitotoxic effects on the target as invasiveness. In this study, we evaluated the potential of US neurostimulation (USNS) in an 396 unpredictable chronic mild stress (UCMS) model. In comparison to pharmacological treatment 397 (fluoxetine), the results showed that selected US application on the prefrontal cortex counteracts 398 behavioral modifications induced by the UCMS regimen and decreases anxiety-related behaviors 399 to a greater extent than classic fluoxetine treatment. Next to these effects, chronic repeated USNS 400 treatment triggered the activation of various brain regions including regions at distance from the 401 targeted zone as confirmed by microPET imaging and metabolomic analyses. These results 402 demonstrate the potential of USNS as a therapeutic tool for major depression. 403

Experimental layout 405
Eighty-three male BALB/cByJRj mice were obtained from Janvier Labs (Le Genest-Saint-Isle, 406 France), aged 9 weeks (29±1.5 g) at the beginning of the experiments. Animals were housed in 407 standard condition (12:12 light-dark cycle, room temperature 22±2° C, free access to food and 408 water). Thirty mice were used for setting-up optimal US neurostimulation (USNS) parameters with 409 motor response measurements, determined with: 1) electrode-free video recordings of the targeted 410 limb (n=20) and 2) electromyography (n=10). The remaining cohort (n=53) was divided in four 411 distinct treatment groups (Veh, Flx, Sham, chronic USNS) that underwent 35 days of the 412 unpredictable chronic mild stress regimen (UCMS) to induce depressive-like behaviors. Then, 413 from day 7 on, mice from the Veh (vehicle) and the Flx (fluoxetine) groups were chronically 414 treated through drinking water respectively with water alone (n=12) or 15 mg/kg fluoxetine 415 (n=12). Every day from day 29 to day 33, mice from the chronic USNS and the Sham groups were 416 either treated with repeated USNS under 1% isoflurane gaseous anesthesia (n=15) or solely 417 anesthetized (n=14). The whole cohort underwent behavioral tasks from day 38 to day 43 to test 418  Unpredictable chronic mild stress 487 The main cohort of fifty-three mice followed the UCMS regimen from day 0 to day 35 ( Figure  488 3A), as described previously (54). Briefly, mice were isolated in 24×11×12 cm cages without 489 environmental enrichment and submitted to daily random socio-environmental stressors including 490 exposition to another mouse bedding, removal of sawdust, contention and light-dark cycle 491

perturbations. 492
Treatments 493 Pharmacological treatments (15 mg/kg fluoxetine) were given from day 7 on to achieve chronic 494 administration at week 5 during analyses. Short delays were described before (30) and allow to 495 observe positive effects of the drug in the experimental timespan. At week 1 of UCMS, the coat 496 state, a standard measure of general deterioration due to UCMS, was significantly deteriorated 497 compared to non-stressed mice and UCMS mice were semi-randomly divided into the four 498 treatment groups (Veh, Flx, Sham, chronic USNS; Supplementary C). The chronic repeated USNS 499 treatment was given during the fifth week of the UCMS (stressors: 08:00 -12:00 AM, treatment: 500 starting at 02:00 PM). To perform chronic USNS, another function generator (Agilent, City, CA, 501 USA), called external trigger (Figure 2A), was used to control the repetition rate of the US stimulus 502 defined in the motor cortex stimulation procedure. A 5 Vpp square wave, set to 0.1 Hz, was used 503 to trigger the main function generator, set for 80,000 cycles (160-msec pulse) of a sine wave (500 504 kHz). The pattern was repeated for 10 min (60 US stimuli) every 10 sec (0.1 Hz); higher repetition 505 frequencies were reported to blunt cortical response (16). This constituted one treatment session 506 of chronic USNS (or "acute USNS"); one session was applied every day for 5 consecutive days 507 from day 29 to day 33 to constitute the chronic protocol of chronic USNS (total of 300 US stimuli 508 over 60 min). Through stereotaxic framing, the collimator was positioned along the (IL), the rodent's equivalent of the sgACC connectivity-wise (17). 511

Behavioral Measures 513
Behaviors were assessed during the dark phase of the light-dark cycle, i.e. the active, awoken phase 514 for the animals. The effects of USNS were assessed against those of the fluoxetine treatment from 515 day 38 to day 43 with 1) the nest-building test (55), 2) the reward-maze test (30) and 3) the open-516 field task. 517 1. For the nest-building task, mice were moved on day 37 to Makrolon type III cages, then at 07:30 518 am of day 38, a square pressed cotton (5 cm²) was placed on the sawdust. After 5 hours in standard 519 condition, the state of the nest, built in cotton, was scored on a predefined scale (55). The mice 520 were then put back in their smaller home cage.  were visually checked for misregistration. Each summed image was also used for statistical 562 analysis. The regions of interest (ROI) atlas of Mirrione in Paxinos coordinates were merged to 563 create a whole brain mask (WBM). To normalize the [ 18 F]-FDG uptake, tissue activity was divided 564 by the whole brain activity, calculated as the average activity in the WBM. Prior to statistical 565 analysis, the WBM was applied over all PET scans to exclude extracerebral regions. The signals 566 extracted using the ROIs on the Z-score maps were considered for further analysis when 567 representing at least 50 contiguous voxels for a statistical threshold set at p<0.05. tissue by two successive extractions, after homogenization, with a mixture of methanol/water (1/1, 572 0.75 mL). After centrifugation, the supernatant was collected, and the solvent evaporated by means 573 of a speedvac. The dry residues were finally taken up in 150 µL of MeOH/H2O (1/1). 10 µL 574 extracts of each sample were pooled to obtain a mixture used as a quality control. Finally, 20 µL 575 were used for LC-HRMS analysis. Fifteen quality control (QC) samples were injected to 576 equilibrate the chromatographic system before each analyses batch. The running order of samples 577 was randomized, and QCs were analyzed every 10 samples. The autosampler temperature 578 (Ultimate WPS-3000 UHPLC system, Dionex, Germany) was set at 4° C and the injection volume 579 for each sample was 5 µL. 580 where Omax is the highest output (set at 10) and Omin the barest (set at 0), I0 is the input half-612 maximal value and dI the slope. 613 On the evaluation of the most effective pressure step above the motor threshold, a repeated measure 614 1-way analysis of variance (ANOVA) was performed with corrected multiple comparisons (post-615 hoc Tukey). Coefficient of variation (relative standard variation, RSD) were computed for each 616 pressure step in order to characterize high-varying distributions. 617 On the study of acute repeated USNS and c-Fos densities, repeated measures 2-way ANOVAs 618 were performed accounting for stimulation (acute USNS/Sham) and bregma with corrected 619 multiple comparisons (post-hoc Tukey). The activity index of acute USNS, that is, the relative 620 considered significant when p<0.05 631 On the study of metabolomics, a first univariate statistical Mann-Withney analysis (XLSTAT) was 632 performed to select metabolites whose expression was significantly different between chronic 633 USNS mice and Sham mice. Next, metabolites whose expression ratios (fold change: FC) between 634 chronic USNS and Sham were greater than 1.25 or less than 0.75 were selected. The pathway 635 enrichment analysis was conducted by the free web software Metaboanalyst (60) to map mouse 636 metabolic pathways corresponding to metabolites selected prior to analysis. The pathway plots 637 were based on the Kyoto Encyclopedia of Genes and Genomes (KEGG) database, and the National 638 Center for Biotechnology Information (NCBI) database was searched to define gene functions. 639 Only the metabolic pathways for which the FDR (false discovery rate) corrected statistics is 640 significant were retained for discussion.