Neural responses prior to licking onset in the striatal matrix compartment in mice

47 48 Licking is a continuous tongue thrust observed during drinking in rodents and humans 49 and is often studied as an essential tongue movement for feeding and swallowing. The 50 striatum, a component of the basal ganglia, plays a critical role in licking onset; 51 however, it is unclear how the two compartments of the striatum—the matrix and 52 striosomes—contribute to the control of licking onset. In this study, we used male and 53 female transgenic mice that selectively expressed Cre recombinase in matrix or 54 striosome neurons and subjected them to operant conditioning based on licking of a 55 spout, during which neuronal activity in both compartments was measured using fiber 56 photometry. Only matrix neurons showed responses prior to licking onset. In addition, 57 the matrix neural response before licking onset was larger when mice licked a spout 58 ipsilateral to the recording hemisphere of the brain than that observed when licking the 59 contralateral spout. This response was observed similarly in mice conditioned to receive 60 a reward regularly and those conditioned to receive a reward randomly, suggesting that 61 the response was unrelated to whether the reward was predictable or unpredictable. 62 Matrix neural activity was negatively correlated with the number of licks during the 63 water intake behavior following the first lick. These findings suggest that matrix 64 neurons are involved in the preparatory process for licking onset as well as in the 65 regulation of licking frequency during water intake. 66 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 4 Significant Statement 67 68 This study demonstrated that during the expression of operant conditioning behaviors 69 based on licking, striatal matrix neurons showed responses prior to licking onset. 70 Additionally, these responses were larger when the mouse licked the spout ipsilateral to 71 the brain hemisphere undergoing recording than when a spout in the contralateral 72 direction was licked. This result was also true for mice conditioned using either regular 73 or random reward conditions. Additionally, the number of licks during water ingestion 74 behavior following the initial lick was negatively correlated with matrix neuron activity. 75 These changes in matrix neuron activity are suggested to be involved in the preparatory 76 process for licking onset, independent of reward prediction, and in the regulation of 77 licking frequency during drinking. 78 79 80 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 5

81 82 Tongue movements are integral to the processes of eating and swallowing, 83 particularly for the formation and transportation of the bolus (Hiiemae and Palmer, 1999). 84 An appropriate model for studying tongue movements is licking behavior, which consists 85 of a continuous tongue thrusting motion. Licking behavior is generated by the central 86 pattern generators in the brainstem (Travers et al., 1997) , which are regulated by top -87 down signals from the basal ganglia (Deniau and Chevalier, 1992; Redgrave et al., 1992; 88 Shammah-Lagnado et al., 1992; Rossi et al., 2016; Toda et al., 2017) . The striatum is a 89 major cortical input site of the basal ganglia. Previous studies in rodents demonstrated 90 that their licking movements were impaired by dopamine deficiency in the striatum 91 (Skitek et al., 1999; Ciucci et al., 2011; Chen et al., 2019) , and that stimulation of direct 92 and indirect pathway striatal neurons initiated and suppressed licking, respectively 93 (Bakhurin et al., 2020) . The striatum consists of two neurochemically and anatomically 94 distinct compartments : the matrix, which is rich in calbindin (Dong et al., 2025) and 95 receives inputs from the sensorimotor and associative cortices, and the striosomes (also 96 known as patches), which are rich in µ -opioid receptors and prodynorphin (Cui et al., 97 2014), receive input from the limbic cortex , and monosynaptically project to midbrain 98 dopaminergic neurons (Gerfen, 1984, 1989; Jiménez -Castellanos and Graybiel, 1989; 99 Eblen and Graybiel, 1995; Kincaid and Wilson, 1996). Although previous in vivo calcium 100 imaging studies have shown that striosomal neural activity correlates with the number of 101 licks during reward intake in classical conditioning (Bloem et al., 2017; Yoshizawa et al., 102 2018), it remains unclear which compartment plays a more dominant role in the control 103 of licking movements. 104 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 6 Therefore, in the present study, we selectively recorded striatal neural activities 105 from the matrix and stri osome compartments of mice during left and right licking 106 movements and then analyzed licking-related neural activities. Matrix neurons responded 107 before the onset of licking, and the responses were larger when mice licked the waterspout 108 ipsilateral to the brain hemisphere being recorded than when licking the spout on the 109 contralateral side. This activity was not affected by reward prediction. Our findings 110 suggest that matrix neurons a re more dominant than striosomal neurons in controlling 111 tongue movements. 112 113 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 7

114 115 Animals 116 The Hokkaido University Animal Use Committee approved this study. Male and 117 female Calb1-IRES-Cre (129S -Calb1tm2.1(cre)Hze/J, The Jackson Laboratory Cat# 118 028532; four male mice, one female mouse; 8–10 weeks old) and Pdyn-IRES-Cre mice 119 (129S-Pdyn(tm1.1(Cre)/Mjkr)/LowlJ, The Jackson Laboratory Cat# 027958; three male 120 mice, two female mice; 8 –10 weeks old ) were housed individually under a 12/12 h 121 light/dark cycle (lights on at 7 A.M.; off at 7 P.M.). Experiments were performed during 122 the light phase. Water intake was restricted to 1–2 mL/day for 2 days before and during 123 the experiments. Food was provided ab libitum for the entire period. 124 125 Surgery 126 Mice were anesthetized with isoflurane (1.0%–4.0%) and placed in a stereotaxic 127 frame. The skull was exposed, and a hole was drilled in the skull. For fiber photometry 128 recordings, AA V5.CAG.Flex.GCaMP6f.WPRE.SV40 (left hemisphere: five mice, right 129 hemisphere: five mice, 100835-AA V5, Addgene, Watertown, MA, USA) was injected 130 into the dorsomedial striatum (DMS) (AP: +0.5, ML: 1.75, DV: 2.85 mm from the brain 131 surface, volume: 400 nL ) using a microsyringe pump (Legato100, Kd Scientific, 132 Holliston, MA, USA). After adeno-associated virus (AA V) injection, an optical probe 133 (diameter: 400 μm, length: 5.0 mm, R-FOC-BL400C-50NA, RWD, Guangdong, China) 134 was implanted 200 μm above the AA V injection coordinates (AP: +0.5, ML: 1.75, DV: 135 2.65 mm from the brain surface). The optical fiber was then fixed with adhesive dental 136 cement (Super Bond, Sun Medical, Shiga, Japan). A head plate (CF-10, Narishige, Tokyo, 137 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 8 Japan) was fixed with pink dental cement (Unifast 2, GC, Tokyo, Japan). Analgesics and 138 antibiotics were applied postoperatively as required (meloxicam, 1 mg/kg s.c.; 0.1% 139 gentamicin ointment, ad usum externum). 140 141 Behavioral tasks 142 The heads and bodies of mice were restricted using a head plate and a metal tube, 143 respectively, and spouts were placed on both the left and right sides of their mouths (Fig. 144 1A). Licks were detected by interruptions of an infrared beam placed in front of the water 145 tube. In the ipsilateral block, water was delivered from the spout on the side ipsilateral to 146 the recording hemisphere, whereas in the contralateral block , it was delivered from the 147 contralateral side. Each trial began by lighting a light-emitting diode (LED) (Fig. 1B). 148 When mice spontaneously licked the spout on the appropriate side of the block, a drop of 149 5% sucrose water (4 µL) was immediately presented. At the end of a trial, the LED was 150 turned off, followed by a 10±3 s inter-trial-interval. A daily session consisted of a 20-min 151 ipsilateral block and a 20 -min contralateral block . An additional experiment was 152 performed to measure the effect of reward prediction on neural activity, which consisted 153 of a 40-min session. When mice spontaneously licked the waterspout on the ipsilateral 154 side, a drop of 5% sucrose water was alternately delivered or not delivered after a 0.5 s 155 delay. The other behavioral task components were similar to those performed in the first 156 experiment. 157 158 Fiber photometry 159 The fiber photometry system consisted of 465 and 405 nm excitation channels 160 that were used to obtain a calcium-dependent signal and a calcium-independent isosbestic 161 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 9 signal, respectively. Fluorescence from GCaMP6f and isosbestic fluorescence were 162 directed with dichroic mirrors (iFMC6_IE(400 -410)_E1(460-490)_F1(500-163 540)_E2(555-570)_F2*(580-680)_S, Doric) and were acquired using a photodetector. 164 The signals were passed through a 10× amplifier and were sampled at 1 kHz using a data 165 acquisition system (Power1401, Cambridge Electronic Design, Cambridge, UK). The 166 acquired photometry signals were processed using custom-written MATLAB code 167 (MATLAB R2018a, MathWorks, Natick, MA, USA). The detailed protocol was 168 described in our recent paper (Yoshizawa and Funahashi, 2025). 169 170 Immunohistochemistry 171 After all experiments were completed, the mice were deeply anesthetized with 172 pentobarbital sodium and then perfused with 4% paraformaldehyde. Brains were 173 carefully removed so that the optical fibers would not cause tissue damage, post-fixed in 174 4% paraformaldehyde at 4 °C overnight, and then transferred to a 30% sucrose/0.1M 175 phosphate buffer solution at 4 °C until the brains sank to the bottom. Coronal sections 176 including the striatum were cut at a thickness of 50 μm on a freezing microtome (REM-177 710; Yamato, Saitama, Japan). Free -floating sections were washed four times in 178 phosphate-buffered saline (PBS) for 15 min and placed in blocking buffer containing 10% 179 normal donkey serum (017 -000-121, Jackson ImmunoRese arch Laboratories, West 180 Grove, PA, USA) and 0.1% Triton X-100 in PBS for 1 h at room temperature. The sections 181 were simultaneously incubated in chicken anti -GFP primary antibody (GFP-1010, Aves 182 Labs, Davis, CA, USA) diluted 1:500 in blocking buffer overnight at 4°C. Afterward, 183 sections were washed four times for 15 min in PBS. The sections were then incubated in 184 donkey anti -chicken Alexa Fluor 488 secondary antibody (703-545-155, Jackson 185 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 10 ImmunoResearch Laboratories) diluted 1:500 in blocking buffer for 2 h at room 186 temperature. The sections were washed four times for 15 min in PBS, mounted on glass 187 slides, and coverslipped with VECTASHIELD Mounting Medium with DAPI (Vector 188 Laboratories, Newark, CA, USA). A fluorescence microscope (Eclipse Ci -L, Nikon, 189 Tokyo, Japan) was used to inspect the stained tissue, and images were obtained using 190 NIS-Elements software (NIS-Elements D, Nikon). 191 192 Experimental design and statistical analysis 193 The analyses include 7164 behavioral and neural trials recorded over a total of 194 57 sessions with 10 mice. We used appropriate statistical tests when applicable, i.e., 195 paired or unpaired t -tests and Pearson correlation analysis with or without Bonferroni’s 196 multiple comparison s tests. Differences were considered statistically significant when 197 p<0.05. Details are described in the Results section. 198 199 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 11

200 201 Licking-related neural activities in the matrix compartment during operant 202 conditioning 203 204 Head-fixed mice performed an operant conditioning task (Fig. 1A, B). 205 Transgenic mice (Calb1 -IRES-Cre) selectively expressing Cre in their matrix neurons 206 (Evans et al., 2020) were employed to record licking -related neural activit y from the 207 matrix. After Cre-dependent expression of the genetic calcium sensor GCaMP6f (Chen 208 et al., 2013) was induced by AA V injection into the DMS (Fig. 1C), fiber photometry 209 recordings were performed during the task . In all five Calb1 -IRES-Cre mice, we 210 confirmed that GCaMP6f-expressing neurons were located at the tip of the optical fiber 211 (Fig. 1D). Figure 2(A, B) shows representative licking behavior and GCaMP fluorescence 212 recorded from the matrix of the left hemisphere. When the mouse licked the ipsilateral 213 (left) and contralateral (right) spouts, the fluorescence increased before the onset of the 214 first lick after the LED was illuminated. This increased fluorescence continued during the 215 water ingestion licks. The average fluorescence during the pre-licking period (−1.0 to 0 s 216 before the onset of the first lick) was significantly larger than that during the baseline 217 period ( −2.0 to −1.0 s before the onset of the first lick) in both blocks (ipsi lateral: 218 −0.46±0.054, base line and 0.044±0.086, pre -licking, p=1.2e −07; cont ralateral: 219 −0.39±0.061, base line and 0.28±0.10, pre -licking, p=1.6e −07, paired t -test, all 220 fluorescence was measured using z-scores, Fig. 2C). There was no significant difference 221 between the ipsilateral and contralateral blocks regarding the average fluorescence during 222 the pre-licking and licking (0 to 1.5 s after the onset of the first lick) periods (pre-licking: 223 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 12 0.044±0.086, ipsi lateral and 0.28±0.10, cont ralateral, p=0.084; licking: 0.52±0.045, 224 ipsilateral and 0.38±0.081, cont ralateral, p=0.12, unpaired t -test, Fig. 2D, E). In the 225 ipsilateral block, there was no significant correlation between the number of licks and the 226 average fluorescence during the licking period (r= −0.17, p=0.11, Pearson correlation 227 analysis, Fig. 2 F), whereas in the contralateral block, there was a significant negative 228 correlation (r=−0.52, p=1.2e−05). 229 230 Licking-related neural activit y in the striosome compartment during operant 231 conditioning 232 233 Transgenic mice (Pdyn-IRES-Cre) selectively expressing Cre in their striosomal 234 neurons (Evans et al., 2020; Xiao et al., 2020; Yoshizawa and Funahashi, 2025) were 235 employed to record licking -related neural activity from the striosomes . After Cre -236 dependent expression of GCaMP6f was induced by AA V injection into the DMS, fiber 237 photometry recordings were performed during the task (Fig. 1E). We confirmed that 238 GCaMP6f-expressing neurons were located at the tip of the optical fiber in all five Pdyn-239 IRES-Cre mice (Fig. 1 F). Fig ure 3(A, B) shows representative licking behavior and 240 GCaMP fluorescence recorded from striosomes in the right hemisphere. In contrast to the 241 matrix, increased fluorescence during the pre-licking period was not observed in either 242 the ipsilateral or contralateral blocks (ipsi lateral: −0.067±0.085, base line and 243 −0.059±0.093, pre -licking, p=0.84; cont ralateral: −0.061±0.047, base line and 244 −0.027±0.047, pre -licking, p=0.53, paired t -test, Fig. 3C). The fluorescence peaked 245 during the licking period. There was no significant difference between the ipsilateral and 246 contralateral block s regarding the average fluorescence during the licking period 247 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 13 (ipsilateral: 0.69±0.072, contralateral: 0.80±0.053, p=0.23, unpaired t -test, Fig. 3D). In 248 both the ipsilateral and contralateral blocks, a significant positive correlation was 249 observed between the number of licks and the average fluorescence during the licking 250 period (ipsi lateral: r=0.42, p=0.0032; cont ralateral: r=0.39, p=6.2e -05, Pearson 251 correlation analysis, Fig. 3E). 252 253 Comparison of licking -related neural activities between the matrix and striosome 254 compartments 255 256 To quantitatively examine differences in licking-related neural activities between 257 the matrix and striosomes, we first averaged the GCaMP fluorescence of all five Calb1-258 IRES-Cre mice (Fig. 4 A). The average fluorescence during the pre-licking period was 259 significantly greater than that during the baseline period in both the ipsilateral and 260 contralateral blocks (ipsilateral: −0.36±0.017, base line and 0.30±0.026, pre -licking, 261 p=7.8e−126, n=1318 trials ; contralateral: −0.31±0.018, baseline and 0.058±0.021, pre-262 licking, p=5.0e−83, n=1401 trials, paired t-test, Fig. 4B) and was significantly larger in 263 the ipsilateral block than in the contralateral block (ipsilateral: 0.30±0.026, contralateral: 264 0.058±0.021, p=9.6e−13, unpaired t -test, Fig. 4C). The fluorescence during the licking 265 period was also significantly larger in the ipsilateral block than in the contralateral block 266 (ipsilateral: 0.69±0.018, contralateral: 0.46±0.019, p=1.9e−17, unpaired t-test, Fig. 4D). 267 Although the correlation coefficient between the number of licks and the average 268 fluorescence during the licking period was not different between the ipsilateral and 269 contralateral blocks (ipsilateral: r= −0.070±0.060, contralateral: r= −0.14±0.064, p=0.49, 270 n=20 sessions, paired t -test, Fig. 4E), the correlation coefficient was significantly 271 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 14 negative in the contralateral block (p=0.043, paired t-test). 272 Next, we averaged the GCaMP fluorescence of all five Pdyn-IRES-Cre mice (Fig. 273 4F). In both the ipsilateral and contralateral blocks, the average fluorescence was not 274 significantly different between the baseline and pre -licking period (ipsilateral: 275 −0.061±0.013, baseline and −0.063±0.012, pre -licking, p=0. 85, n= 1289 trials; 276 contralateral: −0.0555±0.013, baseline and −0.036±0.013, pre-licking, p=0.12, n=1222 277 trials, paired t -test, Fig. 4 G). There was no significant difference in the average 278 fluorescence of the pre -licking period between the blocks (ipsilateral: −0.063±0.012, 279 contralateral: −0.036±0.013, p=0.14, unpaired t-test, Fig. 4H), whereas during the licking 280 period, the fluorescence was significantly larger in the contralateral block than in the 281 ipsilateral block (ipsilateral: 0.31±0.014, contralateral: 0.37±0.014, p=0.0017, Fig. 4I) . 282 The correlation coefficient between the number of licks and the average fluorescence 283 during the licking period was not significantly different between the blocks (ipsilateral: 284 r=0.10±0.049, contralateral: r=0.58±0.048, p=0.47, n=20 sessions, paired t-test, Fig. 4J). 285 The correlation coefficients were not significantly different from zero in either block 286 (ipsilateral: p=0.056, contralateral: p=0.24, paired t-test). 287 288 Effects of reward prediction on the licking-related activity of matrix neurons 289 290 The striatum plays a critical role not only in motor control but also in reward 291 prediction (Samejima et al., 2005; Ito and Doya, 2009; Kim et al., 2009; Ito and Doya, 292 2015; Yoshizawa et al., 2018, 2023). To clarify whether matrix neuron responses prior to 293 licking onset reflected motor-related or reward-predictive neural activity, we recorded 294 matrix neural activity during another operant conditioning experiment in which mice 295 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 15 alternately received a reward and no reward after licking the spout on the ipsilateral side 296 (Fig. 5A). This side was used because the pre-licking response was larger in the ipsilateral 297 block than in the contralateral block. In addition, a 0.5 s delay was inserted before the 298 reward presentation to test whether the reward presentation influenced neural activity 299 during the licking period. 300 In each trial, we measured the reaction time (RT) from when the LED was 301 illuminated to the onset of the first lick because under the reward alternation scheme, the 302 RT in the reward trial became shorter than that in the no -reward trial when subjects 303 successfully predicted the upcoming reward (Isomura et al., 2013) . The RT was shorter 304 or longer following subsequent no-reward or reward trials, respectively (post-no-reward 305 trials: p=1.5e−05, post-reward trials: p=0.00054, paired t -test, Fig. 5 B). Figure 5(C, D) 306 shows representative licking behaviors and GCaMP fluorescence recorded from neurons 307 in the matrix compartment , respectively. The fluorescence was significantly increased 308 during the pre -licking period compared with that during the baseline period in both 309 reward and no-reward trials (reward trial: −0.032±0.076, baseline and 0.0042±0.078, pre-310 licking, p= 6.9e−06; no-reward trial: −0.38±0.065, baseline and −0.041±0.079, pre -311 licking, p=4.9e−08, paired t-test, Fig. 5E). The average fluorescence was not significantly 312 different between reward and no -reward trials during the pre-licking and delay periods 313 (pre-licking: 0.0042±0.078, reward trial and −0.041±0.079, no -reward trial, p=0.68; 314 delay: 1.6±0.085, reward trial and 1.6±0.093, no-reward trial, p=0.99, unpaired t-test, Fig. 315 5F, G). During the licking period, the fluorescence was significantly larger in the reward 316 trial than in the no-reward trial (0.62±0.061, reward trial and 0.82±0.075, no-reward trial, 317 p=0.036, Fig. 5H). 318 To quantitatively examine the difference in the GCaMP signal between reward 319 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 16 and no-reward trials, we averaged the trial -by-trial signal of the session s in which mice 320 successfully predicted the reward and no -reward trials (17 sessions including five mice, 321 Fig. 6A). In both reward and no-reward trials, the average fluorescence was significantly 322 larger during the pre -licking period than during the baseline period (reward trial: 323 −0.15±0.022, baseline and 0.097±0.022, pre-licking, p=1.3e−34, n=966 trials; no-reward 324 trial: −0.32±0.021, baseline and 0.10±0.025, pre-licking, p=5.7e−77, n=968 trials, paired 325 t-test, Fig. 6B). During the pre-licking and delay period s, the average fluorescence was 326 not significantly different between the reward and no -reward trials (pre -licking: 327 0.097±0.022, reward trial and 0.10±0.025, no-reward trial, p=0.82; delay: 1.08±0.027, 328 reward trial and 1.2±0.028, no-reward trial, p=0.053, unpaired t-test, Fig. 6C, D). During 329 the licking period, the average fluorescence was significantly greater in the no -reward 330 trial than in the reward trial ( 0.37±0.021, reward trial and 0.50±0.023, no-reward trial, 331 p=2.5e−05, Fig. 6E). These results indicate that increased fluorescence in the pre-licking 332 period reflected motor-related neural activity. 333 334 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 17

335 336 In the present study, we compared licking -related neural activity between the 337 matrix and the striosome compartments. The major findings are as follows: (1) a neural 338 response prior to licking onset was observed in the matrix but not in the striosome s; (2) 339 this pre-licking neural response was larger when mice licked the spout on the side 340 ipsilateral to the recording hemisphere than that observed when mice licked a spout on 341 the contralateral side ; (3) in the matrix, the neural response during the water ingestion 342 licks was negatively correlated with the number of licks, while in the striosomes, no 343 correlation was observed ; and (4) the pre-licking neural response in the matrix did not 344 reflect reward prediction. 345 Matrix neurons have been hypothesized to play a more dominant role in motor 346 control than striosome neurons because the matrix receives input from the motor cortices 347 (Graybiel and Matsushima, 2023) . In support of this hypothesis, chemogenetic 348 inactivation of matrix neurons impair ed the performance of a learned reach -to-grasp 349 ability (Lopez-Huerta et al., 2016) . A more recent study showed that matrix neurons 350 exhibit early activation at the onset of locomotion and that optogenetic activation of 351 matrix neurons promotes locomotion (Dong et al., 2025) . In the present study, an 352 increased neural response prior to licking onset was only observed in the matrix, 353 indicating the importance of the matrix in motor control of the tongue. 354 Electrophysiological studies reported that pyramidal neurons in the primary 355 motor cortex of monkeys (Tanji and Kurata, 1982) and medium spiny neurons in the 356 dorsolateral striatum (DLS) of rats (Isomura et al., 2013) showed firing activity 0.5 to 357 1.0 s before the onset of hand movements. Such firing is thought to be a preparatory 358 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 18 activity for movement initiation. The increased GCaMP fluorescence in the present study 359 also occurred from approximately 1.0 s before licking onset; therefore, there is a 360 possibility that the increase reflected firing activity in preparation for movement initiation. 361 Licking behavior is a continuous tongue thrusting motion caused by contraction 362 of the genioglossus muscle (Travers and Jackson, 1992) . The genioglossus muscle 363 comprises a matched pair of extrinsic muscle s of the tongue, originates from the mental 364 spine of the mandible , and terminates within the tongue. Bilateral contraction of the 365 genioglossus muscle causes forward protrusion of the tongue, whereas unilateral 366 contraction causes protrusion to the contralateral side (McClung and Goldberg, 2000; 367 Pittman and Bailey, 2009). The hypoglossal nerve innervates the ipsilateral genioglossus 368 muscle and receives bilateral supranuclear inputs (Ugolini, 1995; Zhang et al., 2014) . 369 Therefore, it seems paradoxical that patients with unilateral stroke describe tongue 370 deviation to the healthy side (Umapathi et al., 2000; Wei et al., 2012) . To explain this 371 phenomenon, previous studies have pointed to a bilateral asymmetry in the supranuclear 372 innervation of the hypoglossal nucleus (Lin and Barkhaus, 2009; Morecraft et al., 2014). 373 To our knowledge, the present study is the first report to demonstrate the asymmetry of 374 licking-related activity in the matrix. This result supports the bilateral asymmetry in the 375 supranuclear innervation of the hypoglossal nucleus. 376 The activity of matrix neurons during water ingestion licks was negatively 377 correlated with the number of licks (Fig. 2 F, 4E) and was greater in the no-reward trial 378 than in the reward trial (Fig. 5 H, 6E). The licking duration was also shorter in the no -379 reward trial than in the reward trial (Fig. 5 C). These results suggest that activation of 380 matrix neurons during water ingestion licks inhibits licking behavior. In contrast to the 381 matrix, striosome neuron activity during the water ingestion licks was not correlated with 382 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 19 the number of licks (Fig. 4J). A recent study reported that chemogenetic stimulation of 383 striosomal neurons inhibited contralateral rotation and the total distance traveled in the 384 task (Okunomiya et al., 2025). Therefore, striosomal neurons might play a different role 385 in licking and locomotion. 386 Many studies have demonstrated that striatal neurons contribute to the prediction 387 of future reward. For instance, electrophysiological and in vivo calcium imaging studies 388 have shown that value information of reinforcement learning is represented in neural 389 activity in the striatum of monkeys and rodents (Samejima et al., 2005; Ito and Doya, 390 2009, 2015; Yoshizawa et al., 2018). To clarify whether the matrix neuron response prior 391 to licking onset was motor -related activity or reward -predictive activity, we recorded 392 matrix neural activit y under the reward alternation paradigm , in which mice can easily 393 predict a reward or no reward (Isomura et al., 2013). Their reaction time after turning on 394 an LED was shorter in the reward trial than in the no-reward trial because they wanted to 395 receive a reward as soon as possible. This result indicates that the mice were able to 396 predict the upcoming reward. The pre-licking neural response in the matrix, however, was 397 not significantly different between the reward and no -reward trials, suggesting that the 398 pre-licking response did not reflect reward prediction, but rather preparation for licking 399 onset. Moreover, DLS firing activity prior to onset of hand movement has been reported 400 to be modulated by reward prediction (Isomura et al., 2013) . The licking-related neural 401 activity of the DLS matrix might be modulated by reward prediction. 402 403 404 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 20 Figure legends 405 406 Figure 1 | Measurement of GCaMP fluorescence in the matrix and striosomes 407 during operant conditioning. 408 A. Schematic illustration of the behavioral apparatus. The head and body of the mouse 409 were restricted by a metal frame and tube. The two spouts were placed to the left and 410 right of the mouth. Spout-licking behaviors were monitored using an infrared sensor. An 411 optical fiber was connected to the optical probe implanted in the dorsomedial striatum 412 (DMS) for fiber photometry of GCaMP6f. 413 B. Diagram of an operant conditioning task. Each trial began when a light-emitting 414 diode (LED) was illuminated. In the ipsilateral and contralateral blocks, a drop of 415 sucrose water was delivered immediately after the mice licked a waterspout placed on 416 the ipsilateral or contralateral side, respectively, of the fiber-implanted hemisphere. 417 After water delivery, the LED was turned off, followed by an inter-trial interval (ITI). 418 C. Schematic illustration of the measurement of GCaMP fluorescence from matrix 419 neurons. GCaMP6f was selectively expressed in matrix neurons via injection of 420 AA V5.CAG.Flex.GCaMP6f into the DMS of Calb1-IRES-Cre mice. An optical probe 421 was implanted in the DMS to measure the calcium-dependent fluorescence of GCaMP6f 422 excited by a 465 nm LED. 423 D. Histological image of Cre-dependent GCaMP6f-expressing neurons in the striatum 424 of a Calb1-IRES-Cre mouse. Scale bar: 500 µm. 425 E. Schematic illustration of the measurement of GCaMP fluorescence from striosome 426 neurons. GCaMP6f was selectively expressed in striosome neurons via injection of 427 AA V5.CAG.Flex.GCaMP6f into the DMS of Pdyn-IRES-Cre mice. 428 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 21 F. Histological image of Cre-dependent GCaMP6f-expressing neurons in the striatum of 429 a Pdyn-IRES-Cre mouse. GCaMP6f was mosaically expressed in the striatum. Scale 430 bar: 500 µm. 431 432 Figure 2 | Representative activity of matrix neurons during operant conditioning. 433 A. Spout-licking behavior in the ipsilateral and contralateral blocks, sorted by the first 434 lick after LED illumination. Black dots indicate the timing of spout-licking behaviors. 435 B. Average GCaMP fluorescence recorded from matrix neurons in the ipsilateral and 436 contralateral blocks. The fluorescence was recorded in the same session as the events in 437 A. Matrix neurons showed responses prior to the onset of the first lick in both blocks. 438 C. Average GCaMP fluorescence during the baseline (Base) and pre-licking (Pre) 439 periods. **: p0.05, 442 unpaired t-test. 443 E. Comparison of average GCaMP fluorescence during the licking period between the 444 ipsilateral and contralateral blocks. n.s.: p>0.05, unpaired t-test. 445 F. Correlation between the number of licks and average GCaMP fluorescence during the 446 licking period. The average fluorescence during the licking period is plotted against the 447 number of licks in the same period. Circles and the black line indicate the average 448 GCaMP fluorescence in each trial and the regression line, respectively. Pearson 449 correlation analysis. 450 451 Figure 3 | Representative activity of striosome neurons during operant 452 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 22 conditioning. 453 A. Spout-licking behavior in the ipsilateral and contralateral blocks, sorted by the first 454 lick after LED illumination. Black dots indicate the timing of spout-licking behaviors. 455 B. Average GCaMP fluorescence recorded from striosome neurons in the ipsilateral and 456 contralateral blocks. The fluorescence was recorded in the same session as the events in 457 A. Striosome neurons did not show responses prior to the onset of the first lick in either 458 block. 459 C. Average GCaMP fluorescence during the baseline (Base) and pre-licking (Pre) 460 period. n.s.: p≥0.05, paired t-test. 461 D. Comparison of average GCaMP fluorescence during the licking period between the 462 ipsilateral and contralateral blocks. n.s.: p>0.05, unpaired t-test. 463 E. Correlation between the number of licks and average GCaMP fluorescence during 464 the licking period. The average fluorescence during the licking period is plotted against 465 the number of licks during the same period. Circles and the black line indicate the 466 average GCaMP fluorescence in each trial and the regression line, respectively. Pearson 467 correlation analysis. 468 469 Figure 4 | Population analysis of licking-related neural activity in the matrix and 470 striosomes. 471 A. Matrix neural activity averaged over five mice performing an operant conditioning 472 task. GCaMP signals were aligned to the onset of the first lick after LED illumination. 473 Indigo and magenta lines indicate average activity in the ipsilateral and contralateral 474 blocks, respectively. Shaded areas indicate 95% confidence intervals. Yellow bins 475 indicate a significant difference in the z-score between blocks (p<0.01, unpaired t-test 476 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 23 followed by Bonferroni correction). 477 B. Comparison of matrix neural activity between the baseline (Base) and pre-licking 478 (Pre) periods. Indigo and magenta circles indicate the means of GCaMP signals in the 479 ipsilateral and contralateral blocks, respectively. **: p<0.01, paired t-test. 480 C, D. Comparison of matrix neural activity in the pre-licking (C) and licking (D) 481 periods between the ipsilateral and contralateral blocks. Mean ± SEM. **: p<0.01, 482 unpaired t-test. 483 E. Correlation coefficients between the number of licks and average GCaMP 484 fluorescence of each mouse from which matrix recordings were performed. Indigo and 485 magenta circles indicate average correlation coefficients of matrix recordings from mice 486 in the ipsilateral and contralateral blocks, respectively. *: p<0.05, n.s.: p≥0.05, paired t-487 test. 488 F. Striosome neural activity averaged over five mice performing the operant 489 conditioning task. 490 G. Comparison of striosome neural activity between the baseline (Base) and pre-licking 491 (Pre) periods. n.s.: p≥0.05, paired t-test. 492 H, I. Comparison of striosome neural activity during the pre-licking (H) and licking (I) 493 periods between the ipsilateral and contralateral blocks. Mean ± SEM. **: p<0.01, n.s.: 494 p≥0.05, unpaired t-test. 495 J. Correlation coefficients between the number of licks and the average GCaMP 496 fluorescence in the striosomes of each mouse. Indigo and magenta circles indicate the 497 average correlation coefficients of mice in which striosomes were recorded in the 498 ipsilateral and contralateral blocks, respectively. n.s.: p≥0.05, paired t-test. 499 500 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 24 Figure 5 | Effects of reward prediction on licking-related neural activity in the 501 matrix. 502 A. Diagram of an operant conditioning task designed to test effects of reward prediction 503 on licking-related neural activity in the matrix. Each trial began with LED illumination. 504 When the mouse spontaneously licked the waterspout placed on the ipsilateral side of 505 the optical probe-implanted hemisphere, a drop of sucrose water was alternately 506 delivered or not delivered after a 0.5 s delay period. At the end of each trial, the LED 507 was turned off, followed by a 10±3 s ITI. 508 B. Representative example of the reaction time (RT) in a single session. RTs were 509 shorter in the trials following no-reward (NoRwd) trials, whereas they were longer in 510 the trials following reward (Rwd) trials. Red and blue circles indicate the mean RTs in 511 NoRwd and Rwd trials, respectively. **: p<0.01, paired t-test. 512 C. Spout-licking behavior in reward and no-reward trials, sorted by the first lick after 513 LED illumination. Black dots indicate the timing of spout-licking behaviors. 514 D. Average GCaMP fluorescence recorded from matrix neurons in reward and no-515 reward trials. The fluorescence was recorded in the same session as the events in C. 516 Matrix neurons showed responses prior to the onset of the first lick in both reward and 517 no-reward trials. Blue and red lines indicate averaged fluorescence in reward and no-518 reward trials, respectively. Shaded areas indicate 95% confidence intervals. 519 E. Average GCaMP fluorescence during the baseline (Base) and pre-licking (Pre) 520 periods. Blue and red circles indicate the mean GCaMP signal in reward and no-reward 521 trials, respectively. **: p<0.01, paired t-test. 522 F–H. Comparison of matrix neural activity during the pre-licking (F), delay (G), and 523 licking (H) periods between reward and no-reward trials. Mean ± SEM. **: p<0.01, 524 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 25 n.s.: p≥0.05, unpaired t-test. 525 526 Figure 6 | Population analysis of the effects of reward prediction on licking-related 527 neural activity in the matrix. 528 A. Matrix neural activity averaged over five mice performing the reward alternation 529 operant conditioning task. GCaMP signals were aligned to the onset of the first lick 530 after LED illumination. Blue and red lines indicate averaged activity in reward and no-531 reward trials, respectively. Shaded areas indicate 95% confidence intervals. Yellow bins 532 indicate a significant difference in the z-score between blocks (p<0.01, unpaired t-test 533 followed by Bonferroni correction). 534 B. Comparison of matrix neural activity between the baseline (Base) and pre-licking 535 (Pre) periods. Blue and red circles indicate the mean GCaMP signals in reward and no-536 reward trials, respectively. **: p<0.01, paired t-test. 537 C–E. Comparison of matrix neural activity during the pre-licking (C), delay (D), and 538 licking (E) periods between reward and no-reward trials. Mean ± SEM. **: p<0.01, n.s.: 539 p≥0.05, unpaired t-test. 540 541 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 26

542 543 Bakhurin KI, Li X, Friedman AD, Lusk NA, Watson GD, Kim N, Yin HH (2020) 544 Opponent regulation of action performance and timing by striatonigral and 545 striatopallidal pathways. Elife 9. 546 Bloem B, Huda R, Sur M, Graybiel AM (2017) Two -photon imaging in mice shows 547 striosomes and matrix have overlapping but differential reinforcement -related 548 responses. Elife 6. 549 Chen K, Vincis R, Fontanini A (2019) Disruption of Cortical Dopaminergic Modulation 550 Impairs Preparatory Activity and Delays Licking Initiation. Cereb Cortex 551 29:1802-1815. 552 Chen TW, Wardill TJ, Sun Y , Pulver SR, Renninger SL, Baohan A, Schreiter ER, Kerr 553 RA, Orger MB, Jayaraman V , Looger LL, Svoboda K, Kim DS (2013) 554 Ultrasensitive fluorescent proteins for imaging neuronal activity. Nature 499:295-555 300. 556 Ciucci MR, Russell JA, Schaser AJ, Doll EJ, Vinney LM, Connor NP (2011) Tongue force 557 and timing deficits in a rat model of Parkinson disease. Behav Brain Res 222:315-558 320. 559 Cui Y , Ostlund SB, James AS, Park CS, Ge W, Roberts KW, Mittal N, Murphy NP, Cepeda 560 C, Kieffer BL, Levine MS, Jentsch JD, Walwyn WM, Sun YE, Evans CJ, 561 Maidment NT, Yang XW (2014) Targeted expression of μ-opioid receptors in a 562 subset of striatal direct -pathway neurons restores opiate reward. Nat Neurosci 563 17:254-261. 564 Deniau JM, Chevalier G (1992) The lamellar organization of the rat substantia nigra pars 565 reticulata: distribution of projection neurons. Neuroscience 46:361-377. 566 Dong J, Wang L, Sullivan BT, Sun L, Martinez Smith VM, Chang L, Ding J, Le W, Gerfen 567 CR, Cai H (2025) Molecularly distinct striatonigral neuron subtypes differentially 568 regulate locomotion. Nat Commun 16:2710. 569 Eblen F, Graybiel AM (1995) Highly restricted origin of prefrontal cortical inputs to 570 striosomes in the macaque monkey. J Neurosci 15:5999-6013. 571 Evans RC, Twedell EL, Zhu M, Ascencio J, Zhang R, Khaliq ZM (2020) Functional 572 Dissection of Basal Ganglia Inhibitory Inputs onto Substantia Nigra 573 Dopaminergic Neurons. Cell Rep 32:108156. 574 Gerfen CR (1984) The neostriatal mosaic: compartmentalization of corticostriatal input 575 and striatonigral output systems. Nature 311:461-464. 576 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 27 Gerfen CR (1989) The neostriatal mosaic: striatal patch-matrix organization is related to 577 cortical lamination. Science 246:385-388. 578 Graybiel AM, Matsushima A (2023) Striosomes and Matrisomes: Scaffolds for Dynamic 579 Coupling of V olition and Action. Annu Rev Neurosci 46:359-380. 580 Hiiemae KM, Palmer JB (1999) Food transport and bolus formation during complete 581 feeding sequences on foods of different initial consistency. Dysphagia 14:31-42. 582 Isomura Y , Takekawa T, Harukuni R, Handa T, Aizawa H, Takada M, Fukai T (2013) 583 Reward-modulated motor information in identified striatum neurons. J Neurosci 584 33:10209-10220. 585 Ito M, Doya K (2009) Validation of decision -making models and analysis of decision 586 variables in the rat basal ganglia. J Neurosci 29:9861-9874. 587 Ito M, Doya K (2015) Distinct neural representation in the dorsolateral, dorsomedial, and 588 ventral parts of the striatum during fixed - and free -choice tasks. J Neurosci 589 35:3499-3514. 590 Jiménez-Castellanos J, Graybiel AM (1989) Compartmental origins of striatal efferent 591 projections in the cat. Neuroscience 32:297-321. 592 Kim H, Sul JH, Huh N, Lee D, Jung MW (2009) Role of striatum in updating values of 593 chosen actions. J Neurosci 29:14701-14712. 594 Kincaid AE, Wilson CJ (1996) Corticostriatal innervation of the patch and matrix in the 595 rat neostriatum. J Comp Neurol 374:578-592. 596 Lin HC, Barkhaus PE (2009) Cranial nerve XII: the hypoglossal nerve. Semin Neurol 597 29:45-52. 598 Lopez-Huerta VG, Nakano Y , Bausenwein J, Jaidar O, Lazarus M, Cherassse Y , Garcia-599 Munoz M, Arbuthnott G (2016) The neostriatum: two entities, one structure? 600 Brain Struct Funct 221:1737-1749. 601 McClung JR, Goldberg SJ (2000) Functional anatomy of the hypoglossal innervated 602 muscles of the rat tongue: a model for elongation and protrusion of the 603 mammalian tongue. Anat Rec 260:378-386. 604 Morecraft RJ, Stilwell -Morecraft KS, Solon -Cline KM, Ge J, Darling WG (2014) 605 Cortical innervation of the hypoglossal nucleus in the non -human primate 606 (Macaca mulatta). J Comp Neurol 522:3456-3484. 607 Okunomiya T, Watanabe D, Banno H, Kondo T, Imamura K, Takahashi R, Inoue H (2025) 608 Striosome Circuitry Stimulation Inhibits Striatal Dopamine Release and 609 Locomotion. J Neurosci 45. 610 Pittman LJ, Bailey EF (2009) Genioglossus and intrinsic electromyographic activities in 611 impeded and unimpeded protrusion tasks. J Neurophysiol 101:276-282. 612 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 28 Redgrave P, Marrow L, Dean P (1992) Topographical organization of the nigrotectal 613 projection in rat: evidence for segregated channels. Neuroscience 50:571-595. 614 Rossi MA, Li HE, Lu D, Kim IH, Bartholomew RA, Gaidis E, Barter JW, Kim N, Cai 615 MT, Soderling SH, Yin HH (2016) A GABAergic nigrotectal pathway for 616 coordination of drinking behavior. Nat Neurosci 19:742-748. 617 Samejima K, Ueda Y , Doya K, Kimura M (2005) Representation of action-specific reward 618 values in the striatum. Science 310:1337-1340. 619 Shammah-Lagnado SJ, Costa MS, Ricardo JA (1992) Afferent connections of the 620 parvocellular reticular formation: a horseradish peroxidase study in the rat. 621 Neuroscience 50:403-425. 622 Skitek EB, Fowler SC, Tessel RE (1999) Effects of unilateral striatal dopamine depletion 623 on tongue force and rhythm during licking in rats. Behav Neurosci 113:567-573. 624 Tanji J, Kurata K (1982) Comparison of movement-related activity in two cortical motor 625 areas of primates. J Neurophysiol 48:633-653. 626 Toda K, Lusk NA, Watson GDR, Kim N, Lu D, Li HE, Meck WH, Yin HH (2017) 627 Nigrotectal Stimulation Stops Interval Timing in Mice. Curr Biol 27:3763 -628 3770.e3763. 629 Travers JB, Jackson LM (1992) Hypoglossal neural activity during licking and 630 swallowing in the awake rat. J Neurophysiol 67:1171-1184. 631 Travers JB, Dinardo LA, Karimnamazi H (1997) Motor and premotor mechanisms of 632 licking. Neurosci Biobehav Rev 21:631-647. 633 Ugolini G (1995) Specificity of rabies virus as a transneuronal tracer of motor networks: 634 transfer from hypoglossal motoneurons to connected second -order and higher 635 order central nervous system cell groups. J Comp Neurol 356:457-480. 636 Umapathi T, Venketasubramanian N, Leck KJ, Tan CB, Lee WL, Tjia H (2000) Tongue 637 deviation in acute ischaemic stroke: a study of supranuclear twelfth cranial nerve 638 palsy in 300 stroke patients. Cerebrovasc Dis 10:462-465. 639 Wei CC, Huang SW, Hsu SL, Chen HC, Chen JS, Liang H (2012) Analysis of using the 640 tongue deviation angle as a warning sign of a stroke. Biomed Eng Online 11:53. 641 Xiao X, Deng H, Furlan A, Yang T, Zhang X, Hwang GR, Tucciarone J, Wu P, He M, 642 Palaniswamy R, Ramakrishnan C, Ritola K, Hantman A, Deisseroth K, Osten P, 643 Huang ZJ, Li B (2020) A Genetically Defined Compartmentalized Striatal Direct 644 Pathway for Negative Reinforcement. Cell 183:211-227.e220. 645 Yoshizawa T, Funahashi M (2025) Dopamine release in striatal striosome compartments 646 in response to rewards and aversive outcomes during classical conditioning in 647 mice. Neurosci Res 212:61-68. 648 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 29 Yoshizawa T, Ito M, Doya K (2018) Reward -Predictive Neural Activities in Striatal 649 Striosome Compartments. eNeuro 5. 650 Yoshizawa T, Ito M, Doya K (2023) Neuronal Representation of a Working Memory -651 Based Decision Strategy in the Motor and Prefrontal Cortico-Basal Ganglia Loops. 652 eNeuro 10. 653 Zhang GH, Liu ZL, Zhang BJ, Geng WY , Song NN, Zhou W, Cao YX, Li SQ, Huang ZL, 654 Shen LL (2014) Orexin A activates hypoglossal motoneurons and enhances 655 genioglossus muscle activity in rats. Br J Pharmacol 171:4233-4246. 656 657 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint Fluorescence Probe 405 nm & 465 nm AAV5 -CAG- Flex-GCaMP6f DMS Fluorescence Probe 405 nm & 465 nm AAV5 -CAG- Flex-GCaMP6f DMS B C D E F LED On Ipsilateral lick Reward (sucrose water) LED Off ITI (10±3 s) Pdyn-IRES-Cre Probe GCaMP DAPI LED On Reward (sucrose water) LED Off ITI (10±3 s) Contralateral lick ・ Ipsilateral block ・ Contralateral block Figure 1 | Measurement of GCaMP fluorescence in the matrix and striosomes during operant conditioning. A. Schematic illustration of the behavioral apparatus. The head and body of the mouse were restricted by a metal frame and tube. The two spouts were placed to the left and right of the mouth. Spout -licking behaviors were monitored using an infrared sensor. An optical fiber was connected to the optical probe implanted in the dorsomedial striatum (DMS) for fiber photometry of GCaMP6f. B. Diagram of an operant conditioning task. Each trial began when a light-emitting diode (LED) was illuminated. In the ipsilateral and contralateral blocks, a drop of sucrose water was delivered immediately after the mice licked a waterspout placed on the ipsilateral or contralateral side, respectively, of the fiber-implanted hemisphere. After water delivery, the LED was turned off, followed by an inter-trial interval (ITI). C. Schematic illustration of the measurement of GCaMP fluorescence from matrix neurons. GCaMP6f was selectively expressed in matrix neurons via injection of AA V5.CAG.Flex.GCaMP6f into the DMS of Calb1-IRES-Cre mice. An optical probe was implanted in the DMS to measure the calcium -dependent fluorescence of GCaMP6f excited by a 465 nm LED. D. Histological image of Cre-dependent GCaMP6f- expressing neurons in the striatum of a Calb1 -IRES-Cre mouse. Scale bar: 500 µm. E. Schematic illustration of the measurement of GCaMP fluorescence from striosome neurons. GCaMP6f was selectively expressed in striosome neurons via injection of AA V5.CAG.Flex.GCaMP6f into the DMS of Pdyn-IRES-Cre mice. F. Histological image of Cre-dependent GCaMP6f-expressing neurons in the striatum of a Pdyn-IRES-Cre mouse. GCaMP6f was mosaically expressed in the striatum. Scale bar: 500 µm. A Calb1-IRES-Cre GCaMP DAPI Probe Optical probe Spout (Contralateral) Head plate Tongue Spout (Ipsilateral) .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint 250227_CB42 241127_PD89 A C D GCaMP signal [z -score] Ipsi Cont 1 86 1 61 Time from first lick [s] Time from first lick [s] #TrialGCaMP signal [z-score] Contralateral (Cont)Ipsilateral ( Ipsi) F #Lick r = -0.17 p = 0.11 n.s. ** ** GCaMP signal [z -score] Ipsilateral GCaMP signal [z -score] Pre-licking r = -0.52 p = 1.2e -05 Contralateral #Lick GCaMP signal [z -score] Licking n.s. GCaMP signal [z -score] E B Figure 2 | Representative activity of matrix neurons during operant conditioning. A. Spout-licking behavior in the ipsilateral and contralateral blocks, sorted by the first lick after LED illumination. Black dots indicate the timing of spout-licking behaviors. B. Average GCaMP fluorescence recorded from matrix neurons in the ipsilateral and contralateral blocks. The fluorescence was recorded in the same session as the events in A. Matrix neurons showed responses prior to the onset of the first lick in both blocks. C. Average GCaMP fluorescence during the baseline (Base) and pre-licking (Pre) periods. **: p0.05, unpaired t-test. E. Comparison of average GCaMP fluorescence during the licking period between the ipsilateral and contralateral blocks. n.s.: p>0.05, unpaired t-test. F. Correlation between the number of licks and average GCaMP fluorescence during the licking period. The average fluorescence during the licking period is plotted against the number of licks in the same period. Circles and the black line indicate the average GCaMP fluorescence in each trial and the regression line, respectively. Pearson correlation analysis. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint Ipsi Cont GCaMP signal [z -score] n.s. n.s.n.s. Ipsilateral Contralateral #TrialGCaMP signal [z-score] Time from first lick [s] Time from first lick [s] 11 9872 GCaMP signal [z -score] GCaMP signal [z -score] GCaMP signal [z -score] #Lick Ipsilateral Contralateral #Lick r = 0.42 p = 0.00032 r = 0.39 p = 6.2e -05 A C D ELicking n.s. n.s. B Figure 3 | Representative activity of striosome neurons during operant conditioning. A. Spout-licking behavior in the ipsilateral and contralateral blocks, sorted by the first lick after LED illumination. Black dots indicate the timing of spout-licking behaviors. B. Average GCaMP fluorescence recorded from striosome neurons in the ipsilateral and contralateral blocks. The fluorescence was recorded in the same session as the events in A. Striosome neurons did not show responses prior to the onset of the first lick in either block. C. Average GCaMP fluorescence during the baseline (Base) and pre-licking (Pre) period. n.s.: p≥0.05, paired t-test. D. Comparison of average GCaMP fluorescence during the licking period between the ipsilateral and contralateral blocks. n.s.: p>0.05, unpaired t-test. E. Correlation between the number of licks and average GCaMP fluorescence during the licking period. The average fluorescence during the licking period is plotted against the number of licks during the same period. Circles and the black line indicate the average GCaMP fluorescence in each trial and the regression line, respectively. Pearson correlation analysis. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint Time from first lick [s] GCaMP signal [z -score] Ipsilateral Contralateral ** ** Matrix GCaMP signal [z -score] GCaMP signal [z -score] GCaMP signal [z -score] Pre-licking LickingIpsi ContAveraged activityA B C D E Correlation coefficients Licking n.s. ** ** Time from first lick [s] GCaMP signal [z -score] Ipsilateral Contralateral n.s. ** Striosome GCaMP signal [z -score] GCaMP signal [z -score] GCaMP signal [z -score] Pre-licking LickingIpsi ContAveraged activityF G H I J Correlation coefficients Licking n.s. n.s. n.s. Figure 4 | Population analysis of licking-related neural activity in the matrix and striosomes. A. Matrix neural activity averaged over five mice performing an operant conditioning task. GCaMP signals were aligned to the onset of the first lick after LED illumination. Indigo and magenta lines indicate average activity in the ipsilateral and contralateral blocks, respectively. Shaded areas indicate 95% confidence intervals. Yellow bins indicate a significant difference in the z-score between blocks (p<0.01, unpaired t-test followed by Bonferroni correction). B. Comparison of matrix neural activity between the baseline (Base) and pre -licking (Pre) periods. Indigo and magenta circles indicate the means of GCaMP signals in the ipsilateral and contralateral blocks, respectively. **: p<0.01, paired t-test. C, D. Comparison of matrix neural activity in the pre-licking (C) and licking (D) periods between the ipsilateral and contralateral blocks. Mean ± SEM. **: p<0.01, unpaired t-test. E. Correlation coefficients between the number of licks and average GCaMP fluorescence of each mouse from which matrix recordings were performed. Indigo and magenta circles indicate average correlation coefficients of matrix recordings from mice in the ipsilateral and contralateral blocks, respectively. *: p<0.05, n.s.: p≥0.05, paired t-test. F. Striosome neural activity averaged over five mice performing the operant conditioning task. G. Comparison of striosome neural activity between the baseline (Base) and pre-licking (Pre) periods. n.s.: p≥0.05, paired t-test. H, I. Comparison of striosome neural activity during the pre-licking (H) and licking (I) periods between the ipsilateral and contralateral blocks. Mean ± SEM. **: p<0.01, n.s.: p≥0.05, unpaired t-test. J. Correlation coefficients between the number of licks and the average GCaMP fluorescence in the striosomes of each mouse. Indigo and magenta circles indicate the average correlation coefficients of mice in which striosomes were recorded in the ipsilateral and contralateral blocks, respectively. n.s.: p≥0.05, paired t-test. * .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint A B C GCaMP signal [z -score] GCaMP signal [z -score] #Trial #Trial 1 64 1 64 Reward trial ( Rwd) No-reward trial ( NoRwd ) Rwd NoRwd GCaMP signal [z -score] Pre-licking Delay Licking GCaMP signal [z -score] GCaMP signal [z -score] n.s. n.s. * E F G H ** ** Reaction time [s] ** ** Reward trial No-reward trial GCaMP signal [z -score] LED On (Trial start) Ipsilateral lick Reward (sucrose water) LED Off (Trial end) ITI (10 ±3 s) LED On (Trial start) Ipsilateral lick No reward (nothing) LED Off (Trial end) ITI (10 ±3 s) Delay (0.5 s) Delay (0.5 s) Reaction time Reaction time D Figure 5 | Effects of reward prediction on licking -related neural activity in the matrix. A. Diagram of an operant conditioning task designed to test effects of reward prediction on licking -related neural activity in the matrix. Each trial began with LED illumination. When the mouse spontaneously licked the waterspout placed on the ipsilateral side of the optical probe -implanted hemisphere, a drop of sucrose water was alternately delivered or not delivered after a 0.5 s delay period. At the end of each trial, the LED was turned off, followed by a 10±3 s ITI. B. Representative example of the reaction time (RT) in a single session. RTs were shorter in the trials following no-reward (NoRwd) trials, whereas they were longer in the trials following reward ( Rwd) trials. Red and blue circles indicate the mean RTs in NoRwd and Rwd trials, respectively. **: p<0.01, paired t -test. C. Spout-licking behavior in reward and no -reward trials, sorted by the first lick after LED illumination. Black dots indicate the timing of spout-licking behaviors. D. Average GCaMP fluorescence recorded from matrix neurons in reward and no - reward trials. The fluorescence was recorded in the same session as the events in C. Matrix neurons showed responses prior to the onset of the first lick in both reward and no -reward trials. Blue and red lines indicate averaged fluorescence in reward and no -reward trials, respectively. Shaded areas indicate 95% confidence intervals. E. Average GCaMP fluorescence during the baseline (Base) and pre -licking (Pre) periods. Blue and red circles indicate the mean GCaMP signal in reward and no -reward trials, respectively. **: p<0.01, paired t -test. F–H. Comparison of matrix neural activity during the pre -licking (F), delay (G), and licking (H) periods between reward and no -reward trials. Mean ± SEM. **: p<0.01, n.s.: p≥0.05, unpaired t-test. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint GCaMP signal [z -score] Reward No reward GCaMP signal [z -score] Rwd NoRwd Averaged activity A B Time from first lick [s] C D E GCaMP signal [z -score] GCaMP signal [z -score] GCaMP signal [z -score] Pre-licking Delay Licking n.s. n.s. ** ** ** Figure 6 | Population analysis of the effects of reward prediction on licking -related neural activity in the matrix. A. Matrix neural activity averaged over five mice performing the reward alternation operant conditioning task. GCaMP signals were aligned to the onset of the first lick after LED illumination. Blue and red lines indicate averaged activity in reward and no-reward trials, respectively. Shaded areas indicate 95% confidence intervals. Yellow bins indicate a significant difference in the z-score between blocks (p<0.01, unpaired t-test followed by Bonferroni correction). B. Comparison of matrix neural activity between the baseline (Base) and pre -licking (Pre) periods. Blue and red circles indicate the mean GCaMP signals in reward and no -reward trials, respectively. **: p<0.01, paired t-test. C–E. Comparison of matrix neural activity during the pre-licking (C), delay (D), and licking (E) periods between reward and no-reward trials. Mean ± SEM. **: p<0.01, n.s.: p≥0.05, unpaired t-test. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted May 28, 2025. ; https://doi.org/10.1101/2025.05.23.655747doi: bioRxiv preprint