1
Ca eine in ake exe s dual genome-wide e ec s on hippocampal me abolism
1
and lea ning-dependen ansc ip ion
2
3
Isabel Pai a1†, Luc ezia Cellai2,3†, Céline Me iaux2,3†, Lau anne Poncele 4†, Ouada Nebie2,3, Jean-
4
Michel Saliou5, Anne-Sophie-Lacos e5, An hony Papegaey2,3, He é D obecq6, S éphanie Le G as7,
5
Ma ion Schneide 8, Enas M. Malik8, Ch is a E. Mülle 8, Emilie Fai e2,3, Ke in Ca alho2,3, Vic o ia
6
Gomez-Mu cia2,3, Didie Vieau2,3, B yan Thi oux2,3, Sabiha Edda kaoui2,3, Thibaud Lebou ie 2,3,9,
7
Es elle Schuelle 1, Lau a Tzeplae 1, I is G gu ina1, Jona han Seguin1, Jona han S aube 4, Luisa V.
8
Lopes10, Luc Buée2,3, Valé ie Buée-Sche e 2,3, Rod igo A. Cunha11,12, Rima Ai -Belkacem4‡, Nicolas
9
Se gean 2,3‡, Jean-Sébas ien Annico e13,14‡, Anne-Lau ence Bou illie 1‡*, Da id Blum2,3‡*
10
11
12
† Equal con ibu ions
13
‡ Equal con ibu ions
14
15
1. Uni e si y o S asbou g, CNRS, UMR7364 - Labo a oi e de Neu oscience Cogni i es e
16
Adap a i es (LNCA), F-67000 S asbou g, F ance.
17
2. Uni e si y o Lille, Inse m, CHU Lille, UMR-S1172 LilNCog - Lille Neu oscience & Cogni ion, Lille,
18
F ance.
19
3. Alzheime and Tauopa hies, LabEx DISTALZ, F ance.
20
4. ImaBio ech SAS, Pa c Eu asan é, F-59120 Loos, F ance.
21
5. Uni . Lille, CNRS, Inse m, CHU Lille, Ins i u Pas eu de Lille, UAR CNRS 2014 - US Inse m 41 -
22
PLBS, F-59000 Lille, F ance
23
6. CIIL - Cen e d’In ec ion e d’Immuni é de Lille (CIIL) - INSERM U1019 - UMR 9017
24
7. Uni . S asbou g, CNRS UMR7104, Inse m U1258 - GenomEas Pla o m – IGBMC - Ins i u de
25
Géné ique e de Biologie Moléculai e e Cellulai e, F-67404 Illki ch, F ance.
26
8. Pha maCen e Bonn, Pha maceu ical Ins i u e, Pha maceu ical & Medicinal Chemis y, Uni e si y
27
o Bonn, D-53121 Bonn, Ge many.
28
9. CHU Lille, Memo y Clinic, Lille F ance.
29
10. Ins i u o de Medicina Molecula , Faculdade de Medicina de Lisboa, Uni e sidade de Lisboa,
30
Lisbon, Po ugal.
31
11. CNC - Cen e o Neu oscience and Cell Biology, Uni e si y o Coimb a, 3004-504 Coimb a,
32
Po ugal.
33
12. Facul y o Medicine, Uni e si y o Coimb a, 3004-504 Coimb a, Po ugal.
34
13. Uni . Lille, INSERM, CNRS, CHU Lille, Ins i u Pas eu de Lille, Inse m U1283 / CNRS UMR8199
35
- EGID, 59000 Lille, F ance.
36
14. Uni . Lille, INSERM, CHU Lille, Ins i u Pas eu de Lille, U1167 – RID-AGE-Fac eu s de isque
37
e dé e minan s moléculai es des maladies liées au ieillissemen , 59000 Lille, F ance.
38
39
40
# Co espondence o:
41
Da id Blum, Inse m UMR-S1172, “Alzheime & Tauopa hies”, Place de Ve dun, 59045, Lille Cedex,
42
F ance. O cid Numbe : 0000-0001-5691-431X. Tel: +33320298850, Fax: +33320538562.
43
da id.blum@inse m.
44
Anne-Lau ence Bou illie , Labo a oi e de Neu oscience Cogni i es e Adap a i es (LNCA),
45
UMR7364 Cn s Unis a, 67000 S asbou g, F ance. O cid Numbe : 0000-0002-2317-928.0
46
lau e[email p o ec ed]
47
48
49
Con lic o in e es . The au ho s ha e decla ed ha no con lic o in e es exis s.
50
2
Abs ac
51
52
Ca eine is he mos consumed psychoac i e subs ance wo ldwide. S ikingly, molecula pa hways
53
engaged by i s egula consump ion emain unclea . We he ein add essed he mechanisms
54
associa ed wi h habi ual (ch onic) ca eine consump ion in he mouse hippocampus using un a ge ed
55
o hogonal-omics echniques. Ou esul s e ealed ha ch onic ca eine exe s conce ed pleio opic
56
e ec s in he hippocampus, a he epigenomic, p o eomic and me abolomic le els. Ca eine lowe s
57
me abolic- ela ed p ocesses in he bulk issue, while i induces neu onal-speci ic epigene ic changes
58
a synap ic ansmission/plas ici y- ela ed genes and inc eased expe ience-d i en ansc ip ional
59
ac i i y. Al oge he , hese indings sugges ha egula ca eine in ake imp o es he signal- o-noise
60
a io du ing in o ma ion encoding, in pa h ough a ine- uning o me abolic genes while boos ing he
61
salience o in o ma ion p ocessing du ing lea ning in neu onal ci cui s.
62
63
64
65
66
3
In oduc ion
67
68
Ca eine is he mos consumed psychoac i e subs ance wo ldwide (abou 80% o he popula ion) ia
69
die a y in ake om co ee, ea and soda be e ages. I s popula i y de i es om i s abili y o enhance
70
well-being and some cen al- ela ed unc ions such as a en ion and ale ness (1). La ge
71
epidemiological s udies poin ou an in e se associa ion be ween co ee/ca eine consump ion and
72
all-cause mo ali y (2–4). In gene al, he impac o ca eine on human heal h ollows an in e ed bell-
73
shaped dose- esponse cu e wi h bene i s obse able a doses o 200-400 mg pe day, ha can be
74
ecapi ula ed by 0.3 g/L p.o. in oden s.
75
Compelling epidemiological and expe imen al e idence suppo ha habi ual/ch onic ca eine
76
consump ion no malizes synap ic plas ici y and cogni i e decline in al e ed allos a ic si ua ions such
77
as ageing, Alzheime ’s disease o o he neu o-psychia ic condi ions (5–7). A mo e limi ed numbe
78
o s udies howe e also suppo ha , independen ly o i s abili y o a o a ousal and a en ion,
79
ca eine may exhibi cogni i e-enhancing p ope ies. A e being ewa ded wi h ca eine, honeybees
80
a e able o emembe a p e iously lea ned lo al scen (8). Also, acu e ca eine adminis a ion in a s
81
can enhance memo y es pe o mance (9, 10). In Humans, ca eine in ake immedia ely ollowing
82
lea ning imp o es disc imina ion pe o mance 24 hou s la e (11). These esul s a e in line wi h
83
obse a ions suppo ing he abili y o ca eine o modula e hippocampal/co ical exci abili y in
84
homeos a ic condi ions. Indeed, ca eine ea men in hippocampal slices enhances basal synap ic
85
ansmission (12–14) and modula es long- e m po en ia ion (LTP) in oden s’ hippocampus (12, 15,
86
16) and sha p wa e ipple complexes, ha a e p oposed o unde lie memo y consolida ion (17).
87
Ca eine also con ols neu onal exci abili y and LTP-like e ec s in he human co ex (18, 19). Mos
88
o hese s udies howe e ely on acu e adminis a ions wi h limi ed ele ance owa ds
89
habi ual/ch onic consump ion.
90
Despi e ca eine’s popula i y, b ain molecula changes associa ed wi h i s ch onic in ake emain ill-
91
de ined. Ca eine is known o essen ially in e e e wi h he adenosine gic sys em whe e i ac s as an
92
4
an agonis (20). Howe e , adap i e downs eam pa hways engaged by habi ual/ch onic ca eine
93
consump ion ha e been la gely o e looked. In he p esen s udy, we used a combina ion o unbiased
94
o hogonal-omics echniques o analyze he epigenome, ansc ip ome, p o eome and me abolome
95
o he mouse hippocampus in o de o unco e he molecula pa hways impac ed by ch onic ca eine
96
consump ion in neu onal p ocessing du ing lea ning.
97
98
5
Resul s
99
100
Mouse moni o ing and ca eine concen a ions. In ou expe imen al condi ions, nei he mo ali y
101
no signs o animal su e ing in ca eine- ea ed animals we e encoun e ed. A e age consump ion o
102
0.3 g/L ca eina ed wa e was 4.83 ± 0.15 mL/mouse/day esul ing in b ain ca eine concen a ions
103
o 3.6 ± 1.1 µM, co esponding o a mode a e in ake in Humans (20). Ca eine me aboli es
104
(pa axan hine, heob omine and heophylline) we e also de ec ed in he b ain o ea ed mice wi h
105
espec i e concen a ions o 1.9 ± 0.4 µM, 1.8 ± 0.3 µM, and 0.10 ± 0.03 µM (n=5).
106
107
Ch onic ca eine consump ion dec eases his one ace yla ion o me abolic- ela ed genes in
108
he hippocampus. We hypo hesized ha ch onic ca eine consump ion could a ec hippocampal
109
epigenome o mice. As ca eine is a psychos imulan , we ocused on wo ch oma in ma ks
110
associa ed wi h “ac i e ch oma in” and speci ic ansc ip ional s a es. His one H3 ace yla ion a lysine
111
27 (H3K27ac) is p e e en ially en iched a ac i e enhance s (21), also o ming la ge clus e s o
112
H3K27ac-en iched enhance s known as “supe -enhance s” on highly ansc ibed genes ha a e cell-
113
o issue-speci ic (22, 23). His one H3 lysines K9 and K14 (H3K9/K14ac), on which ace yla ion co-
114
occu s a many gene egula o y elemen s, allows o di e en ia e ac i e enhance s om inac i e ones
115
and hus ep esen s a dynamic ma k accoun ing o s imuli dependen ac i a ion (24). Locus speci ic
116
ace yla ion was e alua ed by ch oma in immunop ecipi a ion ollowed by sequencing (ChIP-seq)
117
expe imen s in do sal hippocampus o con ol (wa e ) and ca eine- ea ed mice. A o al o 2 biological
118
eplica es we e pe o med and P incipal Componen Analysis (PCA) o he wo his one ma ks was
119
gene a ed (Supplemen al Figu e 1A,B). Ch onic ca eine in ake signi ican ly dec eased he
120
ace yla ion o bo h his one ma ks a many genomic loci. H3K9/14ac was deple ed in 778 genomic
121
egions (768 genes) while only 3 we e a ely iden i ied as signi ican ly en iched in ca eine- ea ed
122
animals (FDR<1E-5) (Figu e 1A, Supplemen al Table 1). Gene on ology analysis using Genomic
123

6
Regions En ichmen o Anno a ions Tool (GREAT) e ealed ha hese ace yla ion-deple ed egions
124
we e associa ed wi h genes in ol ed in he egula ion o me abolic p ocesses (amide, lipids), mRNA
125
anspo , egula ion o ansla ion and dend i ic spine mo phogenesis and de elopmen (Figu e 1B).
126
A mo e obus e ec was obse ed in H3K27ac whose peaks we e ound dec eased in 2105 genomic
127
egions (1766 genes) and inc eased in only 4 genomic egions in ca eine s. con ol mice (FDR<1E-
128
5) (Figu e 1C, Supplemen al Table 2). Me abolic- ela ed pa hways, such as lipid ca abolic o amide
129
me abolic p ocesses we e among he dec eased peaks o bo h his one ma ks (Figu e 1B,D).
130
Addi ionally, H3K27ac-deple ed egions we e signi ican ly associa ed wi h myelin- ela ed p ocesses,
131
MAP kinase, nega i e egula ion o calcium-media ed signaling pa hways, as well as
132
he e och oma in o ganiza ion (Figu e 1D). We also pe o med Kyo o Encyclopedia o Genes and
133
Genomes (KEGG) pa hway analyses and iden i ied many p ocesses, some o which ela ed o
134
cAMP-, MAP kinase, Rap1-signaling pa hways and ci cadian en ainmen o bo h H3K9/14 and
135
H3K27ac deple ed egions (Figu e 1E). O no e, he KEGG pa hway da abase poin ed ou me abolic-
136
ela ed pa hways, such as “insulin signaling”, o genes deple ed in ace yla ion o bo h his one ma ks
137
(Figu e 1E) and “glucagon signaling pa hway” o hose associa ed wi h H3K9/14ac deple ed egions
138
(Figu e 1E, blue ba s). Those genes associa ed wi h insulin and glucagon signaling pa hways we e
139
ep esen ed by p o ein-p o ein in e ac ion ne wo k analysis (STRING), showing s ong
140
in e connec i i y (Figu e 1F, yellow and pink do s, espec i ely). As examples, genomic egion
141
ep esen a ion o he Insulin Recep o Subs a e 1 (I s1) gene, which is equi ed o insulin signaling
142
and ela ed spine ma u a ion and synap ic plas ici y (25) and he Glycogen Syn hase Kinase 3 Be a
143
(Gsk3b) gene a e shown (Figu e 1G), wi h signi ican ace yla ion deple ion o bo h ma ks in he
144
ca eine- ea ed g oup e sus con ol ( espec i ely le , H3K9/14ac, FDR=7.75E-05 and H3K27ac,
145
FDR=1.82E-12; igh , H3K9/14ac, FDR=2.58E-11 and H3K27ac, FDR=4.83E-05). O he egions,
146
such as hose associa ed wi h Dusp3, Psme3 and Mlh3 genes, did no exhibi such his one
147
ace yla ion changes upon ca eine ea men , a es ing o selec i i y o he ca eine e ec o bo h
148
his one ma ks (Supplemen al Figu e 1C). In addi ion, in eg a ed pa hway analysis (IPA) applied o
149
7
common ChIP-seq da a o bo h ma ks con i med ha me abolic pa hways, such as insulin o IGF-1
150
signaling, we e canonical pa hways down egula ed upon ca eine ea men (Supplemen al Table
151
3). Po en ial con ibu o s o he ca eine e ec s on he epigenome we e u he assessed using he
152
“ups eam egula o analysis” unc ion o IPA (Supplemen al Table 4). We iden i ied in he
153
ace yla ion-deple ed genes, TCF7L2 (T ansc ip ion ac o 7-like 2) as he mos signi ican ups eam
154
egula o inhibi ed upon ca eine consump ion o bo h ma ks. Fu he mo e, ADORA2A (A2AR) was
155
iden i ied as ano he ups eam egula o in he epigenomic da a, in s iking acco dance wi h he
156
p ima y abili y o ca eine o an agonize adenosine ecep o s (20). Al oge he , hese da a show ha
157
in he bulk hippocampus, ch onic ca eine ea men induces an o e all deace yla ion o wo ac i e
158
ansc ip ion ma ks, H3K27ac and H3K9/14ac, on genes ela ed o ansla ion, lipid and
159
glucose/insulin- ela ed me abolisms.
160
To assess whe he his his one ace yla ion deple ion exe s an e ec on gene ansc ip ion, we
161
pe o med RNA-sequencing (RNA-seq) o bo h wa e and ca eine- ea ed mice. Al hough di e en ial
162
exp ession analysis e ealed no s a is ically signi ican changes o gene exp ession be ween g oups
163
(Supplemen al Figu e 2A), ela i e quan i ica ion o he gene exp ession (z-sco e) co esponding
164
o all H3K27ac-deple ed loci showed an o e all dec ease in exp ession (Supplemen al Figu e 2B)
165
o e he same numbe o andomly chosen genes. Fu he mo e, we also checked by RT-qPCR (n=5-
166
6/g oup) exp ession le els o se e al genes chosen amongs he mos deple ed ones in H3K27ac
167
and obse ed a dec eased exp ession ollowing ch onic ca eine ea men (Supplemen al Figu e
168
2D, ed columns). Impo an ly, we ound ha some o hese genes, such as PBX Homeobox 1
169
(Pbx1), NAD Kinase 2 (Nadk2) and Spindle And Cen iole Associa ed P o ein 1 (Spice1), displayed
170
dec eased exp ession no only upon ch onic (2 weeks) bu also ollowing an acu e (24h) ca eine
171
ea men (Supplemen al Figu e 2D, g een columns). Howe e , he Cy och ome P450 Family 51
172
Sub amily A Membe 1 (Cyp51) gene, ha plays a cen al ole in choles e ol and lipid me abolisms,
173
showed dec eased exp ession solely upon ch onic ca eine ea men . Mo eo e , a pe sis en e ec
174
o ca eine on gene exp ession was obse ed o Pbx1 and Nadk2 genes, as hei exp ession
175
8
emained dec eased e en a e a 2-week ca eine wi hd awal ollowing ch onic adminis a ion
176
(Supplemen al Figu e 2D, blue columns).
177
178
Impac o ch onic ca eine consump ion on hippocampal me abolome. Conside ing ha
179
ca eine dec eased his one ace yla ion o me abolic- ela ed genes, we u he assessed he impac
180
o he dec eased his one ace yla ion on he hippocampal me abolome. To do so, issue spa ial
181
dis ibu ion o molecules was isualized by MALDI (ma ix assis ed lase deso p ion ioniza ion) mass
182
spec ome y imaging analysis, acqui ed om he do sal hippocampus (B egma -1.7mm; Figu e 2A)
183
o wa e and ca eine- ea ed mice (n=6/g oup). PCA analysis was hen pe o med on he eco ded
184
mass spec ome y images om bo h mouse g oups (wa e and ca eine- ea ed), in o de o highligh
185
di e ences in hei hippocampal molecula dis ibu ion p o iles (Figu e 2B). This e ealed lipidomic
186
and me abolomic signa u es ela ed o ch onic ca eine in ake, esul ing in wo dis inc ly sepa a ed
187
clus e s. The iden i ica ion o me aboli es and lipids was based on he measu emen o hei m/z and
188
subsequen compa ison wi h di e en da abanks. In o al, 59% o he me abolome was assigned o
189
he biochemical class o me aboli es (27%) and lipids (32%) (Figu e 2C). The m/z alue o he
190
emaining 41% did no allow o a uni ocal assignmen o a speci ic biochemical class. Ul ima ely,
191
s a is ical analysis o he molecula da ase s e ealed ha ch onic ca eine consump ion induced a
192
majo dec ease in me aboli es and lipid le els (92% dec eased s. 8% inc eased; Figu e 2D). The
193
iden i ied species be ween wa e and ca eine g oups (p < 0.05), de ec ed in posi i e and nega i e
194
ioniza ion mode, a e lis ed in Supplemen al Table 5. Rela ed molecula images aken om
195
hippocampi o wa e and ca eine- ea ed mice, showing hei di e en le els and dis ibu ion a e
196
displayed in Figu e 2E.
197
198
P o eomic hippocampal signa u e associa ed wi h ch onic ca eine consump ion. To gain
199
insigh s in o he po en ial e ec o ch onic ca eine in ake a he p o ein le el, we pe o med mass
200
spec ome y p o eomic analysis o he bulk do sal hippocampus o wa e (con ol) and ch onic
201
9
ca eine- ea ed mice (n=3/g oup). Ca eine induced al e a ions o 179 p o eins, o which 49
202
displayed dec eased and 130 inc eased exp ession le els (Figu e 3A, Supplemen al Table 6). In
203
line wi h he wo p e ious da ase s (epigenomics and me abolomics), gene on ology and p o ein
204
ne wo k analysis e ealed ha dec eased p o eins we e again associa ed wi h pep ide and cellula
205
amide me abolic p ocesses as well as wi h mi ochond ia, wi h educ ion o NADH:Ubiquinone
206
Oxido educ ase Subuni A3 (NDUFA3) in ol ed in mi ochond ial espi a o y chain complex I
207
assembly, o Mi ochond ial Py u a e Ca ie 1 (MPC1) esponsible o anspo ing py u a e in o
208
mi ochond ia o o Long-Chain-Fa y-Acid-CoA Ligase 4 (ACSL4) in ol ed in lipid me abolism
209
(Figu e 3B). Toge he , hese h ee app oaches sugges a obus dec ease in me abolic p ocesses
210
induced by ch onic ca eine in ake in he bulk hippocampal issue. 35 ou o he 49 p o eins
211
dec eased by ca eine, including Insulin Deg ading Enzyme (IDE) and NDUFA3, we e e e sed by
212
ca eine wi hd awal. Only 14 p o eins, such as Insulin Like G ow h Fac o 2 Recep o (IGF2R)
213
emained dec eased ollowing ca eine wi hd awal (Supplemen al Table 6).
214
Gene On ology analysis o he inc eased p o eins e ealed h ee main p o ein clus e s: one ela ed
215
wi h RNA-binding and spliceosome, a second linked o au ophagosome and p o ein p ocessing o
216
endoplasmic e iculum, and a las one associa ed wi h glu ama e gic synapse and phospha ase
217
ac i i y. Conside ing ha ca eine induced exp ession o some synap ic p o eins, and con ols
218
glu ama e gic synap ic ansmission (e.g. (19), we u he assessed hei p edic ed ole in he
219
synap ic compa men using he Synap ic Gene On ologies and anno a ions (SynGO) (26). We
220
obse ed ha mos o he synap ic p o eins anno a ed we e ela ed o synap ic o ganiza ion and
221
signaling, mo e pa icula ly, o chemical synap ic ansmission, such as SH3 And Mul iple Anky in
222
Repea Domains 3 (SHANK3) ha encodes c i ical sca olding p o eins o glu ama e gic
223
neu o ansmission in he pos -synap ic densi ies (27), Synap opodin (SYNPO) a pa o he ac in
224
cy oskele on o pos synap ic densi ies (28) o CREB–Regula ed T ansc ip ion Coac i a o 1
225
(CRTC1) in ol ed in hippocampal plas ici y and memo y (29). O e all, p o eomic analysis e ealed
226
a dec ease in me abolism- ela ed p o eins, concomi an wi h an inc ease o neu onal/synapse-
227
16
in mi ochond ial ac i i y (e.g. NDUFA3 and MPC1). These ch onic changes we e o some poin
369
ela ed o acu e ca eine ea men as a ew genes we e simila ly impac ed ollowing a 24h and a 2-
370
week ca eine ea men , in line wi h Yu e al., 2009 (41), bu he main changes we e associa ed wi h
371
long- e m exposu e o ca eine, as ound o e.g. he Cyp51 gene, encoding a p o ein in ol ed in
372
choles e ol and lipid me abolism. In acco dance, we ound ha 14 o e 49 down egula ed
373
hippocampal p o eins we e s ill al e ed despi e 2 weeks o ca eine wi hd awal, indica ing a
374
pe sis ence o ch onic ca eine e ec s, as p e iously sugges ed (42). Among hese long-las ing
375
impac ed p o eins by ch onic ca eine in ake, we ound ACSL4 and GNA14, which a e in ol ed in
376
he cellula syn hesis o a y acids/lipids, o IGF2 ecep o and ITPR3, in ol ed in insulin-dependen
377
egula ions. Impo an ly, hese da a a e in line wi h and b ing molecula suppo o ecen unc ional
378
magne ic esonance imaging da a showing ha habi ual co ee d inke s exhibi dec eased b ain
379
unc ional connec i i y a es (43). As bulk hippocampal issue was in es iga ed, a ques ion lies in
380
unde s anding he cellula ypes unde lying such me abolic dec ease. Independen IPA analysis o
381
ou wo se s o epigenomic da a (ChIP-seq on bulk hippocampal issue and CUT&Tag-seq on
382
dissocia ed hippocampal cells, “all cells”) pa icula ly poin ed a h ee common ups eam egula o s:
383
TCF7L2 ( ansc ip ion ac o 7 like 2), MKNK1 (MAPK in e ac ing se ine/ h eonine kinase 1) and
384
NFASC (neu o ascin). In he mouse b ain, hese genes a e p edominan ly exp essed by non-
385
neu onal cells: TCF7L2 is p e e en ially exp essed by newly o med oligodend ocy es and
386
as ocy es, NFASC in newly o med oligodend ocy es, while MKNK1 is pa icula ly en iched in
387
mic oglia (see h ps://www.b ain naseq.o g/). IPA analysis o “all cells” CUT&Tag-seq da a u he
388
highligh ed he in ol emen o GLI1 and SOX2, ha a e bo h pa icula ly en iched in as ocy es.
389
These obse a ions s ongly suppo ha he basal/ es ing signa u es elici ed by ch onic ca eine
390
in ake may ely on non-neu onal, likely glial, esponses.
391
Concomi an wi h his de-ace yla ion p ocess obse ed in he bulk hippocampus, we showed ha
392
ch onic ca eine was able o induce a neu on-au onomous epigenomic esponse using bo h ac i e
393
(H3K27ac) and ep essi e (H3K27me3) ma ks: ace yla ion o H3K27 was en iched while i s i-
394

17
me hyla ion was deple ed a genes ela ed o memb ane po en ial, po assium ion egula ion and
395
lea ning and memo y p ocesses. This sugges s ha he o e all ch onic ca eine e ec posi i ely
396
egula es neu onal ac i i y and synap ic ansmission. P o eomic s udies suppo ed his a gumen as
397
a se ies o iden i ied up egula ed p o eins we e ela ed o he glu ama e gic synapse. I is in e es ing
398
o no e ha 73 ou o 130 up egula ed p o eins -some o hem ela ed o he synapse- emained
399
ele a ed e en a e a 2-weeks ca eine wi hd awal, e ealing a long-las ing impac o ch onic ca eine
400
in ake on neu ons. In eg a ion o epigenomic and p o eomic da a pa icula ly poin ed owa ds
401
CRTC1, known o ac as a coincidence senso o calcium and cAMP signals in neu ons igge ing a
402
ansc ip ional esponse in ol ed in la e-phase LTP main enance a hippocampal synapses (44). We
403
u he obse ed ha ch onic ca eine in ake impac s he lea ning/ aining-induced ansc ip ome by
404
signi ican ly enhancing he numbe o di e en ially egula ed genes. In eg a ion o he lea ning-
405
induced genes wi h epigenomic da a iden i ied a g oup o 121 genes ela ed o me abolic p ocesses
406
ha , besides being o e -ac i a ed in ca eine- ea ed mice in lea ning condi ions, we e also de-
407
ace yla ed wi h dec eased o e all exp ession in es ing condi ions (z-sco e). This sugges s ha he
408
es ing-s a e e ec o ca eine in non-neu onal/glial cells migh be a p e- equisi e o he obus
409
ac i a ion o me abolic pa hways hen imp o ing quali y and p ecision o lea ning-associa ed
410
p ocesses, in line wi h i s cogni i e enhancing unc ion.
411
Thus, a majo o e all conclusion o he p esen s udy is he abili y o egula ca eine in ake o exe
412
a long- e m e ec on neu onal ac i i y/plas ici y in he adul b ain, h ough conce ed ac ions on he
413
epigenome, ansc ip ome, p o eome and me abolome, ul ima ely lowe ing me abolic- ela ed
414
p ocesses; and o simul aneously inely uning ac i i y-dependen egula ions o a mo e e icien
415
esponse o expe ience. In o he wo ds, in non-neu onal cells ca eine dec eases -omic ac i i ies
416
unde basal condi ions and imp o es he signal- o-noise a io du ing in o ma ion encoding in b ain
417
ci cui s, hus con ibu ing o bols e he salience o in o ma ion in b ain ci cui s. Rema kably, his dual
418
and opposi e impac o ca eine unde es ing condi ions and upon b ain ac i a ion is in line wi h
419
human b ain imaging s udies: unde basal condi ions ca eine inc eased b ain en opy (45) and
420
18
dec eased unc ional connec i i y (46), whe eas i inc eases BOLD ac i a ion in he on opola and
421
cingula e co ex in a e bal wo king memo y ask (47) e lec ing an inc eased p ocessing po en ial.
422
Addi ionally, neu ophysiological s udies on he pu a i e a ge s o ca eine - adenosine ecep o s –
423
a e in line wi h his dual ole o ca eine, as shown by he opposi e e ec s o A2AR o enhance
424
glu ama e elease con as ing wi h he A1R-media ed inhibi ion o basal synap ic ansmission (48),
425
which is also con olled by A2AR (49). Finally, ou da a also show ha he ampli ude o he
426
ansc ip omic e ec s o ca eine was a g ea e when neu onal ne wo ks we e ac i a ed du ing he
427
lea ning p ocess a he han in basal condi ions, as no ed by o he s when s udying he impac o
428
ca eine on gene exp ession in he basal ganglia (50). This migh pa icula ly ela e o a “p iming” o
429
neu onal ac i i y which would a o he ise o ac i i y-dependen esponse, as i has been sugges ed
430
o he mechanism o ac ion o HDAC inhibi o s (51). How ca eine coo dina es hese epigenomic
431
esponses in he di e en cell ypes is an in e es ing ques ion ha we a e cu en ly pu suing.
432
Finally, he p esen s udy highligh s he molecula impac o ca eine in he homeos a ic b ain, ha
433
will dese e u he in es iga ions, namely ega ding he di e en ial mechanisms ope a ing a he
434
cell-speci ic le el o modula e physiological b ain ac i i y in es ing and ac i i y se ings. Ou da a
435
ha e addi ional a - eaching implica ions. While i is ecognized ha ca eine exhibi s no malizing
436
p ope ies in models o synap ic dys unc ion, as in Alzheime ’s disease (52–54), he cell-speci ic
437
molecula mechanisms emains o be unco e ed. In he opposi e side o he allos a ic b ain spec um
438
(55), ca eine has been sugges ed o impac synap ic a e in b ain de elopmen (56, 57) bu he
439
in ol emen o neu onal s. non-neu onal mechanisms emains ill-de ined. I is he e o e pa icula ly
440
ele an and impo an o add ess, a a la ge scale, he in eg a ed ac ions o ca eine in neu onal s.
441
non-neu onal cells in he imma u e, homeos a ic and ageing b ain.
442
443
19
Ma e ials and Me hods
444
445
Animals. Male C57Bl6/J mice (Cha les Ri e Labo a o ies, F ance) we e housed in a pa hogen- ee
446
acili y (Uni e si y o Lille, F ance). Mice we e 5-6 pe cage (GM500, Tecniplas ) and main ained
447
unde con olled housing condi ions o empe a u e (22°C) and ligh (12-hou ligh /da k cycle), wi h
448
ad libi um access o ood and wa e .
449
450
Ca eine ea men . Two- h ee-mon hs-old mice we e andomly assigned o he wo ollowing
451
expe imen al g oups: wa e (con ol) and ca eine. Ca eine solu ions we e kep in da k bo les hus
452
p o ec ed om ligh and changed weekly. T ea men s a ed a 8-9 weeks o age and las ed o wo
453
weeks. The ch onic ca eine ea men in mice has been se in o de o mimic he usual dose ange
454
o ca eine consump ion in Humans. The selec ed ca eine dose o 0.3 g/L p.o., adminis e ed h ough
455
d inking wa e a 0.3 g/L, has been p e iously shown o p o ide a signi ican bene i in
456
neu odegene a i e con ex s (54, 58, 59). Rega ding he compa ison o ca eine exposu e o 2
457
weeks s. 24 hou s s. ca eine emo al, we p oceed as ollows: 6 animals we e kep unde wa e
458
and o he 6 animals we e ea ed wi h ca eine o 2 weeks and e u ned o wa e o 2 addi ional
459
weeks (ca eine wi hd awal g oup). When he la e g oup o animals e u ned o wa e , an addi ional
460
g oup ha was unde wa e o 2 weeks was hen ea ed wi h ca eine. A las g oup was kep unde
461
wa e o 2 weeks and ea ed wi h ca eine o only 24 hou s. All animals we e hen sac i iced he
462
same day, he do sal hippocampus was sampled and s o ed as indica ed below and used o
463
p o eomics and RT-qPCR analysis.
464
465
Quan i a i e de e mina ion o ca eine and me aboli es in b ain samples. B ain issues om
466
wa e and ca eine g oups we e used o assess concen a ions o ca eine and i s me aboli es
467
(pa axan hine, heob omine and heophylline). Samples we e weighed and 1 mL o 1% o mic acid
468
(FA) solu ion was added o each sample. To de e mine he eco e y a e, con ol samples we e
469
20
spiked wi h a mix u e o ca eine, pa axan hine, heob omine and heophylline (10 µM each). The
470
issues we e lysed using 7 mm s ainless s eel beads and Tissue Lyse LT (Qiagen) o 8 min a 50
471
s okes/minu e, hen ea ed wi h an ul asonic ba h o 5 minu es and subsequen ly cen i uged o
472
15 minu es a 23000xg and 4°C. The supe na an s we e ans e ed o Amicon® Ul a 2 ml 3K
473
cen i ugal il e uni s (Me ck). The emaining pelle s we e subjec ed o he same p o ocol o issue
474
dis up ion and cen i uga ion using 1 mL o acidi ied wa e (FA 1%). Amicon® il e s con aining he
475
combined supe na an s om he wo- old ex ac ion p ocess we e cen i uged o 140 minu es a
476
7500xg and 23°C. Fil a es we e used o liquid ch oma og aphy-mass spec ome y analysis.
477
Samples we e sepa a ed by using a Dionex Ul iMa e 3000 HPLC sys em wi h an in eg a ed a iable
478
wa eleng h de ec o , se a 280 nm, and equipped wi h a C18 column (EC Nucleodu ® C18 G a i y
479
column, 2 mm ID x 50 mm, 3 µm, Mache ey & Nagel). Samples (5 µL) we e injec ed a low a e o
480
300 µL/minu es. A sol en g adien was un om 90% A (wa e con aining 0.2% FA and 2 mM
481
ammonium ace a e) and 10% B (me hanol con aining 2 mM ammonium ace a e) o 50% A and 50%
482
B o e 10 minu es.
483
The elua e was analyzed wi h a coupled mass spec ome e ESI-mic OTOF-Q (B uke Dal onics).
484
Da a we e acqui ed in posi i e ull scan MS mode wi h a scan ange m/z 50-1000. Iden i ica ion and
485
quan i ica ion o he xan hine de i a i es we e pe o med using Da a Analysis so wa e (B uke
486
Dal onics). The limi o de ec ion was 5 nM o ca eine and 10 nM o i s me aboli es (pa axan hine,
487
heob omine and heophylline).
488
489
Lea ning ac i a ion in he Mo is wa e maze. An A lan is Mo is Wa e Maze (MWM) ank was
490
placed in a oom wi h se e al isual ex a-maze cues. Wa e opaci ied wi h powde ed chalk (Blanc
491
de meudon) was main ained a a empe a u e o 21°C. Mice om wa e (con ol) and ca eine g oups
492
we e habi ua ed o he se -up o wo consecu i e days (habi ua ion 1 and 2). Du ing habi ua ion 1,
493
mice we e allowed o disco e he pool illed wi h 5 cm heigh o wa e and a isible pla o m du ing
494
60 seconds. Du ing habi ua ion 2, mice we e allowed o swim in he pool illed wi h wa e in absence
495
21
o he pla o m o 60 seconds. The ollowing 3 days (acquisi ion day 1–3), mice we e ained o
496
localize he pla o m hidden unde nea h he opaci ied wa e using he spa ial cues p esen in he
497
oom. In each acquisi ion day, mice pe o med ou ials each o 60 seconds maximal du a ion. Each
498
ial was e mina ed when he mouse eached he pla o m o a e he 60 seconds. Mice ailing o
499
ind he pla o m we e gen ly guided o he pla o m and allowed o s ay o 8–10 seconds. Du ing he
500
aining days, mice we e subjec ed o MWM in a andom o de , so ha hey we e es ed a di e en
501
imes o he day. All MWM e alua ions o ca eine- o wa e - ea ed mice we e pe o med by
502
expe imen e blind o mouse ea men s.
503
504
Sac i ice and b ain issue p epa a ion. Fo ansc ip omic analysis, mice om Lea ning g oup we e
505
killed by ce ical disloca ion, one hou a e he las aining sec ion, while mice om he Home cage
506
g oup we e killed a he same ime. F eshly dissec ed issues we e immedia ely ozen in liquid
507
ni ogen and kep a -80°C un il RNA ex ac ion. Simila sac i ice p ocedu es we e used o animals
508
used o p o eomic and RTqPCR analyses. Fo molecula MALDI imaging expe imen s, mice we e
509
deeply anes he ized wi h sodium pen oba bi al (50 mg/kg, i.p.), and hen ansca dially pe used wi h
510
cold NaCl (0.9%). B ains we e collec ed, ozen on d y ice and s o ed a -80°C un il use.
511
512
RNA-seq analysis. To al RNA was ex ac ed om do sal hippocampal issues using TRIzol eagen
513
(In i ogen) (n=4/g oup). F eshly dissec ed issue was chopped, homogenized in 300 μL o TRIzol
514
eagen , and ozen (20 minu es a -80°C), ollowed by 3-minu es cen i uga ion a 14000xg be o e
515
chlo o o m/isoamyl ex ac ion. The supe na an was used o p ecipi a e RNA wi h isop opanol and
516
RNase- ee glycogen (30 minu es a 4°C). The pelle was washed once wi h 70% e hanol and
517
esuspended in Milli-Q wa e . A new RNA p ecipi a ion was pe o med wi h 100% e hanol and 3 M
518
sodium ace a e (o e nigh a -20°C). A e wo u he 70% e hanol washes, he pelle was ai -d ied
519
and esuspended in 30 μL nuclease- ee Milli-Q wa e , hea ed 6 minu es a 50°C, and RNA
520
quan i ica ion was pe o med. RNA-seq lib a ies (n=4/g oup) we e gene a ed om 500 ng o o al
521

22
RNA using Illumina® T uSeq® S anded mRNA Lib a y P ep Ki 2. B ie ly, ollowing pu i ica ion wi h
522
poly-T oligo a ached magne ic beads, he mRNA was agmen ed using di alen ca ions a 94°C o
523
2 minu es. The clea ed RNA agmen s we e copied in o i s -s and cDNA using e e se
524
ansc ip ase and andom p ime s. S and speci ici y was achie ed by eplacing dTTP wi h dUTP
525
du ing he second-s and cDNA syn hesis by DNA Polyme ase I and RNase H. Following he addi ion
526
o a single “A” base and he subsequen liga ion o he adap e on double-s anded cDNA agmen s,
527
he p oduc s we e pu i ied and en iched wi h PCR [30 s a 98°C; (10 seconds a 98°C, 30 seconds
528
a 60°C, 30 seconds a 72°C) × 12 cycles; 5 minu es a 72°C] o c ea e he cDNA lib a y. Su plus
529
PCR p ime s we e u he emo ed by pu i ica ion using AMPu e XP beads (Beckman Coul e ), and
530
he inal cDNA lib a ies we e checked o quali y and quan i ied using capilla y elec opho esis.
531
Sequencing was pe o med on he Illumina® Genome Hiseq4000 as single-end 50 base eads
532
ollowing Illumina’s ins uc ions. Reads we e mapped on o he mm10 assembly o Mus musculus
533
genome using STAR 2.5.3a (60) and he Bow ie 2 aligne 2.2.8 (61). Only uniquely aligned eads
534
we e kep o u he analyses. Quan i ica ion o gene exp ession was pe o med using HTSeq-coun
535
0.6.1p1 (62) and gene anno a ions om Ensembl elease 90 and “union” mode. Read coun s we e
536
no malized ac oss lib a ies wi h he me hod p oposed by Ande e al. (2010) (63). Compa isons o
537
in e es we e pe o med using he es o di e en ial exp ession p oposed by Lo e (64) and
538
implemen ed in he DESeq2 Bioconduc o lib a y ( 1.16.1). Resul ing p- alues we e adjus ed o
539
mul iple es ing using he Benjamini and Hochbe g me hod (65).
540
541
Ch oma in Immunop ecipi a ion (ChIP). F eshly dissec ed issue was chopped by a azo blade
542
and apidly incuba ed in 1.5 mL phospha e-bu e ed saline (PBS) con aining 1% o maldehyde o
543
10 minu es a oom empe a u e. To s op ixa ion, glycine was added (0.125 M inal concen a ion).
544
Do sal hippocampi om 4 mice we e pooled pe sample and wo biological eplica es pe condi ion
545
we e used o he ChIP-seq. Tissue samples we e hen p ocessed as desc ibed in Cha e jee e al.
546
(34) and sonica ed using he Diagenode Bio up o (30 seconds ON-30 seconds OFF a High Powe
547
23
x 35 cycles). Sonica ed ch oma in was cen i uged 10 minu es a 14000xg, he supe na an collec ed
548
and dilu ed 1:10 in ChIP dilu ion bu e (0.01% SDS, 1.1% T i on X-100, 1.2 mM EDTA, 16.7 mM
549
T is-Cl, pH 8.1, 167 mM NaCl). A ac ion o he supe na an (50 µL – 10%) om each sample was
550
sa ed be o e immune-p ecipi a ion o ‘ o al inpu ch oma in’. Supe na an s we e incuba ed o e nigh
551
(4°C) wi h 1/1000 p ima y an ibodies agains H3K9/14ac (Diagenode #C15410200) and H3K27ac
552
(Abcam #ab4729), ollowed by p o ein A Dynabeads (In i ogen) o 2 hou s a oom empe a u e.
553
A e se e al washes (low sal , high sal , LiCl and TE bu e s), he esul ing DNA-p o ein complexes
554
we e elu ed in 300 µL elu ion bu e (1% SDS, 0.1 M NaHCO3). The c osslinking was e e sed
555
(o e nigh a 65°C) and he DNA was subsequen ly pu i ied wi h RNAse (30 minu es a 37°C) and
556
p o einase K (2 hou s a 45°C). DNA om he immunop ecipi a ed and inpu samples was isola ed
557
using Diagenode Mic oChIP DiaPu e columns wi h 20 µL nuclease- ee milliQ wa e in low binding
558
ubes. ChIP samples we e u he pu i ied a he Genomeas Pla o m using Agencou AMPu e XP
559
beads (Beckman Coul e ) and quan i ied using Qubi (In i ogen).
560
561
ChIP-seq lib a ies and sequencing. ChIP-seq lib a ies we e p epa ed om 2-10 ng o double-
562
s anded pu i ied DNA using he Mic oPlex Lib a y P epa a ion ki 2 (C05010014, Diagenode s.a.,
563
Se aing, Belgium), acco ding o manu ac u e 's ins uc ions. DNA was i s epai ed and yielded
564
molecules wi h blun ends. Nex , s em-loop adap o s wi h blocked 5’ ends we e liga ed o he 5’ end
565
o he genomic DNA (gDNA), lea ing a nick a he 3’ end. The adap o s canno liga e o each o he
566
and do no ha e single-s and ails hus non-speci ic backg ound is a oided. In he inal s ep, he 3’
567
ends o he gDNA we e ex ended o comple e lib a y syn hesis and Illumina compa ible indexes we e
568
added h ough a PCR ampli ica ion (4+7 cycles). Ampli ied lib a ies we e pu i ied and size-selec ed
569
using Agencou AMPu e XP beads (Beckman Coul e ) o emo e uninco po a ed p ime s and o he
570
eagen s. P io o analyses, DNA lib a ies we e checked o quali y and quan i ied using a 2100
571
Bioanalyze (Agilen ). The lib a ies we e loaded in he lowcell a 8 pM concen a ion, and clus e s
572
24
we e gene a ed using he Cbo and sequenced using he Illumina HiSeq 4000 echnology as single-
573
end 50 base eads ollowing Illumina’s ins uc ions. Image analysis and base calling we e pe o med
574
using RTA and CASAVA.
575
576
ChIP-seq analyses. Sequenced eads we e mapped o he Mus musculus genome assembly mm10
577
using Bow ie 1.0.0 wi h he ollowing pa ame e s «-m1-s a a-bes -y-l40». Sam ools me ge 1.3.1
578
(66) was used o combine biological eplica es by condi ion. Then, BED ools in e sec 2.26.0 (67)
579
was used o emo e eads loca ed wi hin ENCODE blacklis ed egions. SICER (SICER-d .sh) 1.1
580
(68) was used o de ec di e en ially bound egions on he pools o biological eplica es using he
581
ollowing pa ame e s: «Species: mm10, E ec i e genome size as a ac ion o e e ence genome:
582
0.74, Th eshold o edundancy allowed o ea ed eads: 1, Th eshold o edundancy allowed o
583
WT eads: 1, Window size: 200 bps, F agmen size: 200 bps, Gap size: 600 bps, FDR o
584
iden i ica ion o en iched islands: 1E-2, FDR o iden i ica ion o signi ican changes: 1E-2. Finally,
585
di e en ially bound egions we e anno a ed wi h espec o he closes gene using Home
586
anno a ePeaks.pl 4.11.1 (69). An FDR o 1E-5 was used in di e en ial analyses (ca eine s.
587
con ol).
588
589
Neu onal and all cells isola ion. Neu onal and all cells suspensions we e ob ained om mouse
590
hippocampus ch onically ea ed wi h ca eine o wa e (con ol). Fo ha , we used Neu al Tissue
591
Dissocia ion (Mil enyi, #130-092-628) and Neu on Isola ion Ki s (Mil enyi, #130-115-389), ollowing
592
manu ac u e 's ins uc ions wi h some adap a ions. B ie ly, wo mouse hippocampi we e pooled pe
593
sample and ha es ed in a p e-hea ed bu e solu ion con aining papain. This was ollowed by se ies
594
o manual mechanical dissocia ions, using scisso s and i e polished Pas eu pipe es o descending
595
diame e , and incuba ions a 37°C unde slow o a ion. The solu ion was hen il e ed (50 µm) and
596
cen i uged (10 minu es, 300xg, a oom empe a u e) and myelin was emo ed using Myelin
597
Remo al Beads II ki (Mil enyi, #130-096-733), incuba ing o 15 minu es a 4°C, cen i uging (10
598
25
minu es, 300xg a 4°C) and il e ing he sample h ough MS columns (Mil enyi, #130-042-201) placed
599
in MiniMACS™ Sepa a o (Mil enyi, #130-042-102) o collec he myelin deple ed low- h ough, ee
600
o cell deb is. The ‘all cells’ suspension was collec ed a his poin and coun ed using he TC20
601
Au oma ed Cell Coun e (Bio-Rad, #1450102) o ob ain a o al o 70,000 cells pe sample. Wi h he
602
emaining o he samples, we p oceeded wi h neu onal isola ion acco ding o manu ac u e 's
603
ins uc ions, inally deple ing he samples h ough MS columns o collec he low- h ough en iched
604
in neu ons. The samples we e coun ed and 70,000 cells pe sample we e aken o CUT&Tag
605
expe imen s.
606
607
Clea age Unde Ta ge s and Tagmen a ion (CUT&Tag). Ha ing isola ed all cells and neu onal
608
popula ions we p oceeded wi h CUT&Tag me hod o assess hei genome-wide H3K27ac and
609
H3K27me3 ch oma in s a e. The p o ocol was adap ed om ha desc ibed by Kaya-Oku e al., 2019
610
(31) The me hod is based on digi onin-induced cell pe meabiliza ion (Sigma, #300410-250MG) and
611
concana alin A-coa ed magne ic beads (Cell signaling, #93569S) immobiliza ion. This is ollowed by
612
o e -nigh incuba ion a 4°C wi h p ima y an ibodies agains H3K27ac (Abcam, #ab4729) and
613
H3K27me3 (Diagenode, #C15410195), ollowed by 1 hou incuba ion wi h he seconda y an ibody
614
(An ibodies online, #ABIN101961). The loaded-Tn5 is hen added (Diagenode, #C01070001) and
615
he clea ed DNA is ex ac ed using MinElu e PCR Pu i ica ion Ki (Quiagen, # 28004). Lib a y
616
p epa a ion was conduc ed using Nex e a p ime s (Illumina, #FC-131-2001) and pos -PCR clean-up
617
using SPRI bead slu y (Beckmann Coul e , #B23317). Concen a ion o he collec ed DNA was
618
achie ed by Qubi (In i ogen, #Q32851). Two biological eplica es we e used pe g oup and Rabbi
619
IgG (Diagenode #C15410206) was used as con ol.
620
621
CUT&Tag analyses. Reads (pai ed-end) we e mapped o Mus musculus genome (assembly mm10)
622
using Bow ie2 (61) 2.2.8 wi h de aul pa ame e s excep o “–end- o-end- e y-sensi i e-no-mixed
623
–no-disco dan -I10-X700”. P io o peak calling, eads wi h mapping quali y below 30 we e emo ed
624
32
Re e ences
763
764
1. Smi h A. E ec s o ca eine on human beha io .. Food Chem Toxicol. 2002;40(9):1243–1255.
765
2. Kim Y, Je Y, Gio annucci E. Co ee consump ion and all-cause and cause-speci ic mo ali y: a
766
me a-analysis by po en ial modi ie s.. Eu J Epidemiol. 2019;34(8):731–752.
767
3. Lo ield E e al. Associa ion o Co ee D inking Wi h Mo ali y by Gene ic Va ia ion in Ca eine
768
Me abolism: Findings F om he UK Biobank. JAMA In e n Med. 2018;178(8):1086–1097.
769
4. F eedman ND, Pa k Y, Abne CC, Hollenbeck AR, Sinha R. Associa ion o co ee d inking wi h
770
o al and cause-speci ic mo ali y. N Engl J Med. 2012;366(20):1891–1904.
771
5. Fla en V e al. F om epidemiology o pa hophysiology: wha abou ca eine in Alzheime ’s disease?
772
Biochem Soc T ans. 2014;42(2):587–592.
773
6. Cunha RA. How does adenosine con ol neu onal dys unc ion and neu odegene a ion?. J
774
Neu ochem. 2016;139(6):1019–1055.
775
7. Cellai L e al. The Adenosine gic Signaling: A Complex bu P omising The apeu ic Ta ge o
776
Alzheime ’s Disease. F on Neu osci. 2018;12:520.
777
8. W igh GA e al. Ca eine in Flo al Nec a Enhances a Pollina o 's Memo y o Rewa d. Science.
778
2013;339(6124):1202– 1204.
779
9. Ma ques S, Ba alha VL, Lopes LV, Ou ei o TF. Modula ing Alzheime ’s disease h ough ca eine:
780
a pu a i e link o epigene ics.. J Alzheime s Dis. 2011;24 Suppl 2:161–171.
781
10. Angelucci MEM, Cesá io C, Hi oi RH, Rosalen PL, Da Cunha C. E ec s o ca eine on lea ning
782
and memo y in a s es ed in he Mo is wa e maze. B azilian J Med Biol Res. 2002;35(10):1201–
783
1208.
784
11. Bo o a D e al. Pos -s udy ca eine adminis a ion enhances memo y consolida ion in humans.
785
Na Neu osci. 2014;17(2):201–203.
786
12. Lopes JP, Pliásso a A, Cunha RA. The physiological e ec s o ca eine on synap ic ansmission
787
and plas ici y in he mouse hippocampus selec i ely depend on adenosine A(1) and A(2A) ecep o s.
788
Biochem Pha macol. 2019;166:313–321.
789
13. Cos enla AR, Cunha RA, de Mendonça A. Ca eine, adenosine ecep o s, and synap ic plas ici y.
790
J Alzheime s Dis. 2010;20 Suppl 1:S25-34.
791
14. Simons SB, Ca uana DA, Zhao M, Dudek SM. Ca eine-induced synap ic po en ia ion in
792
hippocampal CA2 neu ons. Na Neu osci. 2011;15(1):23–25.
793
15. Lao-Pe eg ín C e al. Ca eine-media ed BDNF elease egula es long- e m synap ic plas ici y
794
h ough ac i a ion o IRS2 signaling. Addic Biol. 2017;22(6):1706–1718.
795

33
16. Blaise JH, Pa k JE, Bellas NJ, Gi chell TM, Phan V. Ca eine consump ion dis up s hippocampal
796
long- e m po en ia ion in eely beha ing a s. Physiol Rep. 2018;6(5):e13632.
797
17. Wa anabe Y, Ikegaya Y. Ca eine Inc eases Hippocampal Sha p Wa es in Vi o. Biol Pha m Bull.
798
2017;40(7):1111–1115.
799
18. Hanajima R e al. E ec o ca eine on long- e m po en ia ion-like e ec s induced by quad ipulse
800
ansc anial magne ic s imula ion. Exp B ain Res. 2019;237(3):647–651.
801
19. Ke kho s A e al. Ca eine Con ols Glu ama e gic Synap ic T ansmission and Py amidal Neu on
802
Exci abili y in Human Neoco ex. F on Pha macol. 2017;8:899.
803
20. F edholm BB, Bä ig K, Holmén J, Nehlig A, Z a au EE. Ac ions o ca eine in he b ain wi h
804
special e e ence o ac o s ha con ibu e o i s widesp ead use. Pha macol Re . 1999;51(1):83–
805
133.
806
21. Hein zman ND e al. Dis inc and p edic i e ch oma in signa u es o ansc ip ional p omo e s and
807
enhance s in he human genome. Na Gene . 2007;39(3):311–318.
808
22. Hnisz D e al. Supe -enhance s in he con ol o cell iden i y and disease. Cell 2013;155(4):934–
809
947.
810
23. J. PSC e al. Ch oma in s e ch enhance s a es d i e cell-speci ic gene egula ion and ha bo
811
human disease isk a ian s. P oc Na l Acad Sci. 2013;110(44):17921–17926.
812
24. Ka modiya K, K ebs AR, Oulad-Abdelghani M, Kimu a H, To a L. H3K9 and H3K14 ace yla ion
813
co-occu a many gene egula o y elemen s, while H3K14ac ma ks a subse o inac i e inducible
814
p omo e s in mouse emb yonic s em cells. BMC Genomics 2012;13:424.
815
25. Sánchez-Sa asúa S e al. IRS1 exp ession in hippocampus is age-dependen and is equi ed o
816
ma u e spine main enance and neu i ogenesis. Mol Cell Neu osci. 2022;118:103693.
817
26. Koopmans F e al. SynGO: An E idence-Based, Expe -Cu a ed Knowledge Base o he
818
Synapse. Neu on 2019;103(2):217-234.e4.
819
27. Sheng M, Kim E. The Shank amily o sca old p o eins. J Cell Sci. 2000;113(11):1851–1856.
820
28. Mundel P e al. Synap opodin: an ac in-associa ed p o ein in elencephalic dend i es and enal
821
podocy es. J Cell Biol. 1997;139(1):193–204.
822
29. Pa a-Damas A e al. CRTC1 Func ion Du ing Memo y Encoding Is Dis up ed in
823
Neu odegene a ion. Biol Psychia y 2017;81(2):111–123.
824
30. Wang L, Wal e P. Msp1/ATAD1 in P o ein Quali y Con ol and Regula ion o Synap ic Ac i i ies.
825
Annu. Re . Cell De Biol. 2020;36(1):141–164.
826
31. Kaya-Oku HS e al. CUT&Tag o e icien epigenomic p o iling o small samples and single cells.
827
Na Commun. 2019;10(1):1930.
828
32. Han S e al. Regula ion o dend i ic spines, spa ial memo y, and emb yonic de elopmen by he
829
34
TANC amily o PSD-95-in e ac ing p o eins. J Neu osci. 2010;30(45):15102–15112.
830
33. Pa a-Damas A, Rubió-Fe a ons L, Shen J, Sau a CA. CRTC1 media es p e e en ial
831
ansc ip ion a neu onal ac i i y- egula ed CRE/TATA p omo e s. Sci Rep. 2017;7(1):18004.
832
34. Cha e jee S e al. Reins a ing plas ici y and memo y in a auopa hy mouse model wi h an
833
ace yl ans e ase ac i a o . EMBO Mol Med. 2018;10(11).
834
35. Ma ínez G e al. Regula ion o Memo y Fo ma ion by he T ansc ip ion Fac o XBP1. Cell Rep.
835
2016;14(6):1382–1394.
836
36. Cao L e al. VEGF links hippocampal ac i i y wi h neu ogenesis, lea ning and memo y. Na
837
Gene . 2004;36(8):827–835.
838
37. Mews P e al. Ace yl-CoA syn he ase egula es his one ace yla ion and hippocampal memo y.
839
Na u e 2017;546(7658):381–386.
840
38. Jacobson KA, on Lubi z DK, Daly JW, F edholm BB. Adenosine ecep o ligands: di e ences
841
wi h acu e e sus ch onic ea men . T ends Pha macol Sci. 1996;17(3):108–113.
842
39. Fe é S. An upda e on he mechanisms o he psychos imulan e ec s o ca eine. J Neu ochem.
843
2008;105(4):1067–1079.
844
40. Doepke C e al. Ca eine: F iend o Foe? Annu Re Food Sci Technol. 2016;7:117–137.
845
41. Yu L e al. Unco e ing mul iple molecula a ge s o ca eine using a d ug a ge alida ion
846
s a egy combining A2A ecep o knockou mice wi h mic oa ay p o iling. Physiol Genomics
847
2009;37(3):199–210.
848
42. S enningsson P, Nomikos GG, F edholm BB. The s imula o y ac ion and he de elopmen o
849
ole ance o ca eine is associa ed wi h al e a ions in gene exp ession in speci ic b ain egions. J
850
Neu osci. 1999;19(10):4011–4022.
851
43. Magalhães R e al. Habi ual co ee d inke s display a dis inc pa e n o b ain unc ional
852
connec i i y. Mol Psychia y 2021;26(11):6589–6598.
853
44. A. KK e al. TORC1 is a calcium- and cAMP-sensi i e coincidence de ec o in ol ed in
854
hippocampal long- e m synap ic plas ici y. P oc Na l Acad Sci. 2007;104(11):4700–4705.
855
45. Chang D e al. Ca eine Caused a Widesp ead Inc ease o Res ing B ain En opy. Sci Rep.
856
2018;8(1):2700.
857
46. Tal O e al. Ca eine-Induced Global Reduc ions in Res ing-S a e BOLD Connec i i y Re lec
858
Widesp ead Dec eases in MEG Connec i i y. F on Hum Neu osci. 2013;7:63.
859
47. Koppels ae e F e al. Does ca eine modula e e bal wo king memo y p ocesses? An MRI
860
s udy. Neu oimage 2008;39(1):492–499.
861
48. Cunha RA. Di e en cellula sou ces and di e en oles o adenosine: A1 ecep o -media ed
862
inhibi ion h ough as ocy ic-d i en olume ansmission and synapse- es ic ed A2A ecep o -
863
35
media ed acili a ion o plas ici y. Neu ochem In . 2008;52(1–2):65–72.
864
49. Lopes L V, Cunha RA, Ribei o JA. C oss alk be ween A(1) and A(2A) adenosine ecep o s in
865
he hippocampus and co ex o young adul and old a s. J Neu ophysiol. 1999;82(6):3196–3203.
866
50. Dassesse D, Leden C, Pa men ie M, Schi mann SN. Acu e and ch onic ca eine adminis a ion
867
di e en ially al e s s ia al gene exp ession in wild- ype and adenosine A(2A) ecep o -de icien
868
mice. Synapse 2001;42(2):63–76.
869
51. Bu ns AM, G ä J. Cogni i e epigene ic p iming: le e aging his one ace yla ion o memo y
870
amelio a ion. Cu Opin. Neu obiol. 2021;67:75–84.
871
52. Dua e JMN, Cunha RA, Ca alho RA. Adenosine A₁ ecep o s con ol he me abolic eco e y
872
a e hypoxia in a hippocampal slices. J Neu ochem. 2016;136(5):947–957.
873
53. Lau en C e al. Bene icial e ec s o ca eine in a ansgenic model o Alzheime ’s disease-like
874
au pa hology. Neu obiol Aging 2014;35(9):2079–2090.
875
54. A endash GW e al. Ca eine p o ec s Alzheime ’s mice agains cogni i e impai men and educes
876
b ain be a-amyloid p oduc ion. Neu oscience 2006;142(4):941–952.
877
55. Da id B, V. LL. S abilizing synapses. Science. 2021;374(6568):684–685.
878
56. G. SC e al. Adenosine Recep o An agonis s Including Ca eine Al e Fe al B ain De elopmen
879
in Mice. Sci T ansl Med. 2013;5(197):197 a104-197 a104.
880
57. Fe an G-C e al. Con e gence o adenosine and GABA signaling o synapse s abiliza ion du ing
881
de elopmen . Science. 2022;374(6568):eabk2055.
882
58. A endash GW e al. Ca eine e e ses cogni i e impai men and dec eases b ain amyloid-be a
883
le els in aged Alzheime ’s disease mice. J Alzheime s Dis. 2009;17(3):661–680.
884
59. Lau en C e al. Bene icial e ec s o ca eine in a ansgenic model o Alzheime ’s disease-like
885
au pa hology. Neu obiol Aging 2014;35(9):2079–2090.
886
60. Dobin A e al. STAR: ul a as uni e sal RNA-seq aligne . Bioin o ma ics 2013;29(1):15–21.
887
61. Langmead B, Salzbe g SL. Fas gapped- ead alignmen wi h Bow ie 2. Na Me hods
888
2012;9(4):357–359.
889
62. Ande s S, Pyl PT, Hube W. HTSeq--a Py hon amewo k o wo k wi h high- h oughpu
890
sequencing da a. Bioin o ma ics 2015;31(2):166–169.
891
63. Ande s S, Hube W. Di e en ial exp ession analysis o sequence coun da a. Genome Biol.
892
2010;11(10):R106.
893
64. Lo e MI, Hube W, Ande s S. Mode a ed es ima ion o old change and dispe sion o RNA-seq
894
da a wi h DESeq2. Genome Biol. 2014;15(12):550.
895
65. Benjamini Y, Hochbe g Y. Con olling he False Disco e y Ra e - a P ac ical and Powe ul
896
App oach o Mul iple Tes ing. J R S a Soc Se B-Me hodological. 1995;57(1):289–300.
897
36
66. Li H e al. The Sequence Alignmen /Map o ma and SAM ools. Bioin o ma ics
898
2009;25(16):2078–2079.
899
67. Quinlan AR, Hall IM. BEDTools: a lexible sui e o u ili ies o compa ing genomic ea u es.
900
Bioin o ma ics 2010;26(6):841–842.
901
68. Xu S, G ullon S, Ge K, Peng W. Spa ial clus e ing o iden i ica ion o ChIP-en iched egions
902
(SICER) o map egions o his one me hyla ion pa e ns in emb yonic s em cells. Me hods Mol Biol.
903
2014;1150:97–111.
904
69. Zang C e al. A clus e ing app oach o iden i ica ion o en iched domains om his one
905
modi ica ion ChIP-Seq da a. Bioin o ma ics 2009;25(15):1952–1958.
906
70. Amemiya HM, Kundaje A, Boyle AP. The ENCODE Blacklis : Iden i ica ion o P oblema ic
907
Regions o he Genome. Sci Rep. 2019;9(1):9354.
908
71. Heinz S e al. Simple combina ions o lineage-de e mining ansc ip ion ac o s p ime cis-
909
egula o y elemen s equi ed o mac ophage and B cell iden i ies. Mol Cell 2010;38(4):576–589.
910
911
37
912

38
Figu e 1. Hippocampal epigenomic al e a ions associa ed wi h ch onic ca eine consump ion.
913
(A) Volcano plo showing he di e en ial en iched genomic egions o H3K9/14ac (ChIP-seq) upon
914
ch onic ca eine ea men (778 dec eased and 3 inc eased peaks). Red do s ep esen he
915
signi ican di e en egions (FDR<1E-5). (B) Genomic Regions En ichmen o Anno a ions Tool
916
(GREAT) analysis showing he mos en iched biological p ocesses associa ed wi h he H3K9/14ac
917
dec eased peaks in ca eine- ea ed mice. Blue a ows poin owa ds me abolic p ocesses and
918
ansla ion ela ed e ms. (C) Volcano plo ep esen ing he di e en ially egula ed egions o
919
H3K27ac upon ch onic ca eine ea men (2105 dec eased and 4 inc eased peaks, wi h FDR<1E-
920
5). (D) GREAT analysis ep esen ing he mos common biological p ocesses associa ed wi h he
921
H3K27ac dec eased peaks in he ca eine g oup. Regula ion o me abolic p ocesses a e indica ed
922
by he blue a ows. (E) KEGG pa hway analyses o deple ed egions o bo h his one ma ks. Dashed
923
g ey line indica es he signi ican adjus ed p- alue <0.05. (F) Func ional p o ein-p o ein ne wo k
924
analysis (STRING) ep esen a ion o insulin and glucagon- ela ed genes ound dec eased in bo h
925
his one ace yla ion ma ks. (G) Rep esen a ion o he genomic egions (IGV) o he me abolic genes
926
I s1 and Gsk3b showing signi ican dec ease o H3K27ac and H3K9/14ac a e ca eine ea men
927
(I s1 H3K27ac FDR=1.82E-12; H3K9/14ac FDR=7.75E-05; Gsk3b H3K27ac FDR=4.83E-05;
928
H3K9/14ac FDR=2.58E-11). Two biological eplica es pe his one ma k we e used o ChIP-seq
929
expe imen s.
930
931
39
932
933
934
935
936
937
938
939
940
40
Figu e 2. Hippocampal me abolomic changes induced by ch onic ca eine consump ion.
941
Unsupe ised p incipal componen analysis (PCA) pe o med in he hippocampal egion o in e es
942
delimi a ed in yellow on he Nissl s aining o he b ain issue sec ion (A). Sco es om he
943
unsupe ised PCA in he hippocampus o Wa e - (in blue) and Ca eine- ea ed mice (in ed) a e
944
p esen ed in a plo whe e he di e ences be ween he molecula signa u es o he wo expe imen al
945
g oups clea ly eme ge (B). Pie cha s showing he dis ibu ion o he di e en classes o molecules
946
(C) and hei abundance changes (D) o m/z measu ed in posi i e o nega i e ioniza ion modes wi h
947
a signi ican quan i a i e di e ence a e he S uden ’s - es analysis in he hippocampus o Ca eine-
948
compa ed o Wa e - ea ed animals (N = 6/g oup). (E) Mass spec ome y images ob ained a a
949
spa ial esolu ion o 35 µm o m/z p esen ing a dec eased (g een) o inc eased (o ange) densi y in
950
he hippocampus o Ca eine- ea ed compa ed o Wa e - ea ed mice. The colo scale shows he
951
in ensi y o he m/z o in e es . Ce , ce amide; PC, phospha idylcholine; PI, phospha idylinosi ol; PS,
952
phospha idylse ine.
953
954
955
956
957
958
959
960
41
961
962
963
964
965
966
967
968
969