APOE from astrocytes restores Alzheimer's Aβ-pathology and DAM-like responses in APOE deficient microglia

APOE om as ocy es es o es Alzheime ’s Aβ-pa hology and DAM-like esponses in APOE de icien
mic oglia
P ana P eman1,2, Daan Moecha s1,2, Em e Fe an3,4, Leen Wol s1,2, Lu ga de Se neels1,2, Disha Shah1,2,
Jochen Lamo e5, Su esh Poo a hingal1, An Snellinx1,2, Renzo Mancuso6,7, S i am Balusu1,2, Da id
Klene man3,4, Amaia M A anz8,9, Ma k Fie s1,2,+, Ba De S oope 1,2,10,+
1 VIB Cen e o B ain & Disease Resea ch, Leu en, Belgium.
2 Labo a o y o he Resea ch o Neu odegene a i e Diseases, Depa men o Neu osciences, Leu en
B ain Ins i u e (LBI), KU Leu en (Uni e si y o Leu en), Leu en, Belgium.
3 Yusu Hamied Depa men o Chemis y, Uni e si y o Camb idge, Camb idge, UK.
4 UK Demen ia Resea ch Ins i u e, Uni e si y o Camb idge, Camb idge, UK.
5 VIB FACS Expe ise Cen e , Cen e o Cance Biology, Leu en, Belgium.
6 Mic oglia and In lamma ion in Neu ological Diso de s (MIND) Lab, VIB-UAn we p, Cen e o
Molecula Neu ology, An we p, Belgium.
7 Depa men o Biomedical Sciences, Uni e si y o An we p, An we p, Belgium.
8 Labo a o y o Humanized Models o Disease, Achuca o Basque Cen e o Neu oscience, Leioa, Spain.
9 Ike basque Basque Founda ion o Science, Bilbao, Spain.
10 UK Demen ia Resea ch Ins i u e, Uni e si y College London, London, UK.
+ Co espondence o Ba De S oope : ba .des oope[email p o ec ed].uk o o da a analysis Ma k Fie s:
ma k. ie s@kuleu en.be
Final cha ac e coun (no spaces): 122,340
ABSTRACT
The majo gene ic isk ac o o Alzheime ’s disease (AD), APOE4, accele a es be a-amyloid (Aβ)
plaque o ma ion, bu whe he his is caused by APOE exp essed in mic oglia o as ocy es is deba ed.
We exp ess he e he human APOE iso o ms in as ocy es in an Apoe-de icien AD mouse model. This
is no only su icien o es o e he amyloid plaque pa hology bu also induces he cha ac e is ic
ansc ip ional pa hological esponses in Apoe-de icien mic oglia su ounding he plaques. We ind
ha bo h APOE4 and he p o ec i e APOE2 om as ocy es inc ease ib illa plaque deposi ion, bu
di e en ially a ec soluble Aβ agg ega es. Mic oglia and as ocy es show speci ic al e a ions in
unc ion o APOE geno ype exp essed in as ocy es. Ou expe imen s indica e a cen al ole o he
as ocy es in APOE media ed amyloid plaque pa hology and in he induc ion o associa ed mic oglia
esponses.
Keywo ds: Alzheime ’s disease/As ocy es/Mic oglia/APOE/β-amyloid pa hology
INTRODUCTION
The APOE4 a ian o apolipop o ein E was iden i ied 30 yea s ago as he majo gene ic isk ac o o
Alzheime ’s Disease (AD) a he popula ion le el (Co de e al., 1993; S i ma e e al., 1993; Benne
e al., 2009) while he APOE2 a ian was he i s iden i ied p o ec i e allele agains AD (Co de e al.,
1994; Nagy e al., 1995). Apolipop o ein E is a lipid ca ie and as such in ol ed di ec ly o indi ec ly in
a spec um o pa hophysiological p ocesses. In AD, i s speci ic ole in p omo ing β-amyloid (Aβ) plaque
pa hology is, howe e , well es ablished (Schmechel e al., 1993; Kok e al., 2009; Fleishe e al., 2013;
Mu phy e al., 2013). While e ec s on clea ance o Aβ (Deane e al., 2008; Cas ellano e al., 2011;
Ve ghese e al., 2013) ha e been desc ibed, he majo ole is di ec p omo ion o Aβ ib illiza ion (Ma
e al., 1994; Wisniewski e al., 1994; Cas ano e al., 1995) as also suppo ed by in i o expe imen s
(Huynh e al., 2017; Liu, Zhao, e al., 2017). This in e p e a ion is suppo ed by gene ic s udies in
humans. The APOE4 allele is associa ed wi h inc eased amyloid deposi ion (Pol ikoski e al., 1995;
Benne e al., 2009; Hanson e al., 2013; Gonneaud e al., 2016), possibly explaining he ea ly onse
o disease in homozygous APOE4 ca ie s (Blacke e al., 1997; Ma ins e al., 2005; Sando e al., 2008;
Powell e al., 2021). On he o he hand, AD mouse models show limi ed amyloid plaque pa hology in
he absence o Apoe (Bales e al., 1999; Hol zman, Fagan, e al., 2000; Ul ich e al., 2018). APOE4-
a ge ed eplacemen mice, exp essing he human APOE4 allele ins ead o he mouse Apoe, c ossed
wi h di e en AD mouse models such as PDAPP, APP/ PS1ΔE9 and 5xFAD, accele a ed plaque
o ma ion compa ed o APOE3- a ge ed eplacemen mice (Hol zman, Bales, e al., 2000; Fagan e al.,
2002; Bales e al., 2009; Cas ellano e al., 2011; Youmans e al., 2012; Rod iguez e al., 2014). This has
also been obse ed wi h i al- ec o media ed o e -exp ession o APOE4 iso o ms in AD mouse b ains
(Doda e al., 2005; Hud y e al., 2013; Zhao e al., 2016).
Logically, i is assumed ha he p o ec i e e ec o APOE2 is he opposi e o APOE4. The p o ec i e
e ec is e lec ed in a low p e alence o APOE2 alleles in AD pa ien coho s (Talbo e al., 1994; Wes ,
William Rebeck and Hyman, 1994; Reiman e al., 2020) and educ ion o Aβ plaque load in APOE2
ca ie s (Nagy e al., 1995; Pol ikoski e al., 1995; Chiang e al., 2010; Kan a ci e al., 2012; Se ano-
Pozo e al., 2015; G o he e al., 2017). O he s udies sugges ha APOE2 p o ec ion may be a he le el
o au pa hology o e en a he le el o delay in cogni i e de ici s, wi hou educ ion in Aβ plaque
o ma ion (Be lau e al., 2007, 2009). Simila ly, in mouse models, he e ec o APOE2 on Aβ plaques
emains con o e sial wi h some s udies con i ming dec eased amyloid deposi ion compa ed o APOE3
(Fagan e al., 2002; Cas ellano e al., 2011; Hud y e al., 2013; Zhao e al., 2016), while o he s show
simila o e en inc eased deposi s (Doda e al., 2005; Bales e al., 2009; Youmans e al., 2012;
Rod iguez e al., 2014). Besides, in many s udies he APOE2 geno ype was no included when epo ing
APOE4 iso o m-speci ic e ec s on amyloid plaque o ma ion (F ye e al., 2005; Kim e al., 2011; Bien-
Ly e al., 2012; Liu, Zhao, e al., 2017; Mahan e al., 2022).
The cellula sou ce ha p oduces he pa hologically ele an APOE is ano he poin o deba e. The
p o ein is exp essed mainly in he li e . Howe e , APOE is no ans e ed o e he blood b ain ba ie
(Lane-Dono an e al., 2016; Huynh e al., 2019) sugges ing ha as ocy es a e he main con ibu o o
APOE in he b ain (Boyles e al., 1985; Zhang e al., 2014). Unde s ess condi ions APOE migh be
exp essed by neu ons (Xu e al., 2006), and unde hese condi ions APOE4 impai s synap ic unc ion
and agg a a es au pa hology (And ews-Zwilling e al., 2010; Bu ini e al., 2010; Kno e le e al., 2014;
Kou sodend is e al., 2023). Neu onal APOE, howe e , does no seem o be associa ed speci ically wi h
amyloid pa hology. In con as , mic oglia can signi ican ly up egula e APOE exp ession as pa o hei
DAM (Disease Associa ed Mic oglia) esponse o he amyloid plaques (Ke en-Shaul e al., 2017;
K asemann e al., 2017; Sala F ige io e al., 2019; Chen e al., 2020). Howe e , econs i u ion o APOE
exp ession in mic oglia only in an Apoe knockou backg ound (Mancuso e al., 2024) o mic oglia-
speci ic knockou o APOE had no e ec on amyloid plaque o ma ion o mic oglial associa ion wi h
plaques (Henning ield e al., 2022). This sugges s ha APOE up egula ion in mic oglia, while an
impo an cha ac e is ic o DAM cell s a e induc ion, is no di ec ly in ol ed in he ini ial plaque
deposi ion, which as we discussed, seems o be he main AD causing pa hological e ec o APOE4.
Thus, i appea s ha in he e y ea ly phases o AD, when Aβ is belie ed o seed small soluble
agg ega es, and no DAM cell s a e is ye induced in he mic oglia, as ocy es se e as he main sou ce
o APOE in he b ain. APOE exp ession speci ically unde he GFAP-p omo e modi ied plaque
deposi ion in an iso o m-speci ic manne (Hol zman, Bales, e al., 2000; Fagan e al., 2002; Liu, Zhao,
e al., 2017), while as ocy e speci ic knockou o APOE3/APOE4 led o educ ion o Aβ plaque
o ma ion (Mahan e al., 2022), in line wi h his hypo hesis. The e o e, we in es iga ed whe he i is
indeed su icien o exp ess APOE in as ocy es o induce Aβ plaque o ma ion, and mo e impo an ly,
whe he ha is su icien o induce he DAM mic oglia esponse. In e es ingly, in ou expe imen al
pa adigm, APOE2 exp ession only in as ocy es did no educe he amyloid plaque o ma ion compa ed
o APOE3, e en when APOE4 had signi ican ly highe le els han bo h. We p esen ansc ip omic
changes in as ocy es exp essing he di e en APOE iso o ms, ha sugges possible impai men o
p o eos asis and au ophagy in APOE4 as ocy es. We p o ide e idence ha mic oglia lacking
endogenous APOE exp ession, can be s i ed in hei esponse o amyloid pa hology by APOE sec e ed
by as ocy es. Mo eo e , we show ha he APOE e ec o as ocy es on amyloid plaques is la gely cell
au onomous, as in he absence o mic oglia, amyloid plaques a e s ill gene a ed.
RESULTS
Exp ession o APOE in as ocy es is su icien o es o e ib illa amyloid plaque pa hology
In o de o s udy he ole o as ocy e-de i ed APOE on amyloid plaque pa hology, we exp essed he
di e en human APOE iso o ms ( u he e e ed o as APOE2, APOE3 and APOE4), along wi h epo e
mChe y p o ein, speci ically in as ocy es in an Apoe-de icien AD mouse model (AppNL-G-F x Apoe-/-),
using ecombinan AAV2/8 pa icles (injec ed ICV a pos na al day 2/3, see Ma e ials and Me hods o
de ails) and he as ocy e speci ic p omo e ALDH1L1 (Fig. 1A). AAV pa icles exp essing only mChe y
p o ein we e used as con ols ( u he e e ed o as APOEKO). A 2 mon hs o age, we obse ed
mChe y exp ession in he co ical egions, along wi h limi ed sp ead in he subco ical a ea (Fig. S1A).
A 6 mon hs o age, we used immuno luo escence o con i m mChe y exp ession in ALDH1L1+
as ocy es (Fig. S1B). The ALDH1L1 p omo e is speci ically exp essed in as ocy es in mouse b ain
(Cahoy e al., 2008; Zhang e al., 2014) and has been ex ensi ely used o a ge gene exp ession in an
as ocy e-speci ic se ing (Cla ke e al., 2018; Michalo icz e al., 2019; Hasel e al., 2021; Endo e al.,
2022; Mahan e al., 2022). Using low cy ome y, we ound ha a ound 14% o ACSA2+ as ocy es
exp essed mChe y (Fig. S1Cii) and a ound 90% o he mChe y-posi i e cells we e posi i e o
as ocy e ma ke ACSA2 (Fig. S1Ciii). I is likely ha emaining mChe y-posi i e cells a e a
subpopula ion o as ocy es ha a e ACSA2-nega i e. We do no exclude ha a small numbe o non-
as ocy ic cells a e ansduced by he i al pa icles. Howe e , we con i med by immuno luo escence
he lack o mChe y exp ession in NeuN+ neu ons, Apc+ oligodend ocy es and Iba1+ mic oglia (Fig.
S1D). We also con i med APOE exp ession in mChe y+ cells (Fig. S1E), demons a ing ha we
econs i u ed human APOE exp ession in as ocy es.
We cha ac e ized he amyloid pa hology in he ansduced mice a 6 mon hs o age. AppNL-G-F mice
exhibi ed a high load o amyloid plaque deposi s in co ical egions as obse ed by X-34 s aining (Fig.
S2B). This was la gely absen when hey we e c ossed wi h Apoe-/- mice (Figs. 1B, S2A-B). Exp ession
o human APOE iso o ms in as ocy es es o ed he ib illa Aβ plaque deposi ions in he AAV-
ansduced co ical egions (Figs. 1B, S2A). APOE4 and, ema kably, also APOE2, showed signi ican ly
inc eased load o ib illa plaques compa ed o APOE3 econs i u ed animals (Fig. 1C). We u he
quan i ied mo phological di e ences, namely size and compac ness (Fig. S3A), o indi idual Aβ plaques
be ween he g oups (see Ma e ial and Me hods o de ails). The APOE3 g oup showed a signi ican
inc ease in he p opo ion o medium sized plaques in he ange o 20-40 µm diame e , while APOE2
and APOE4 bo h showed inc eased abundance o la ge plaques (abo e 40 µm diame e ) (Fig. S3B).
Plaques in he APOE3 g oup showed signi ican ly inc eased compac ness ac oss all size anges (Fig.
S3Ci-iii). The e o e, ib illa plaques o med in he p esence o APOE3 end o be smalle and mo e
compac mo phologically compa ed o plaques o med in he p esence o APOE2 and APOE4.
We quan i ied human APOE exp ession in homogenized b ain issues using a human APOE speci ic
ELISA (Fig. 1D) and qPCR (Fig. S1F) and ound no signi ican o e all di e ences in le els o exp ession
be ween APOE2, APOE3 and APOE4 g oups (Figs. 1D, S1F). In e es ingly, we obse ed ha APOE
sec e ed om as ocy es colocalized wi h X-34 posi i e ib illa amyloid plaques (Fig. S2C). The e o e,
we wonde ed whe he each APOE iso o m could di e en ially a ec plaque o ma ion wi hin each
expe imen al g oup. We used ELISA o quan i y he plaque associa ed guanidine soluble Aβ42 le els in
he di e en samples as a quan i a i e ead ou o amyloid deposi ion. Linea eg ession analysis o
examine he ela ionship be ween APOE le els and guanidine soluble Aβ42 le els (Fig. 1E), showed a
ela i ely s ong posi i e associa ion be ween APOE4 and Aβ42 (R2 = 0.61, slope = 0.23, p = 0.007).
APOE2 ollows a simila end (R2 = 0.43, slope = 0.11, p = 0.077), while APOE3 le els showed no
associa ion (R2 = 0.006, slope = 0.005, p = 0.843). Thus, APOE4 has a s ong dose-dependen e ec on
amyloid ib illiza ion in his model o plaque deposi ion.
We wonde ed whe he he absence o ib illa amyloid plaques in he APOEKO g oup led o an inc ease
in soluble Aβ species in he mouse b ains. Using he single molecule pulldown (SiMPull) echnique
(Side is e al., 2021), we quan i ied he numbe o soluble Aβ agg ega es (Figs. 1F-G) elu ed by soaking
he b ain issue in a i icial CSF. APOEKO mouse b ains showed a signi ican inc ease in he numbe o
soluble Aβ agg ega es compa ed o o he expe imen al g oups (Fig. 1G), con i ming a shi owa ds
soluble species in he b ain pa enchyma in he absence o plaque deposi ion. In e es ingly, he APOE4
g oup which showed a high le el o ib illa plaque deposi ion, showed low le els o soluble Aβ
agg ega es (Fig. 1G). APOE2 was in e media y be ween APOEKO and APOE4. Mos ema kably, APOE3,
which showed lowe amoun s o ib illa plaque accumula ion han APOE4 (Fig. 1C), displayed low
le els o soluble agg ega es simila o APOE4 in his assay (Fig. 1G).
O e all, ou da a con i m and ex end he no ion ha APOE sec e ed om as ocy es is cen al in he
o ma ion o amyloid plaques in mice. While APOE4 accele a es his p ocess, APOE2 also shows a
simila end, especially when compa ed o APOE3, indica ing ha he APOE2 p o ec i e e ec is no
simply he opposi e o APOE4’s accele a ing e ec on plaque o ma ion in AD, a leas in he AD mouse
model used he e.
APOE exp ession modula es as ocy e cell-s a es in AppNL-G-F mouse b ain
Ou da a poin s o an impo an ole o as ocy es in amyloid plaque o ma ion and a dis inc ole o
he di e en APOE iso o ms in his p ocess a 6 mon hs o age. The e a e only limi ed ansc ip omic
s udies a ailable ha in es iga e he e ec o he human APOE a ian s on as ocy e pheno ypes in
i o (Se ano-Pozo e al., 2021; Tcw e al., 2022; Lee e al., 2023). We he e o e decided o analyze he
ansc ip ome p o iles o mChe y+/ACSA2+ as ocy es om mouse b ains a 6 mon hs o age using
d ople -based scRNAseq. A e il e ing ou non-as ocy e cells and low-quali y cells, we e ained
18,667 as ocy es ac oss he 4 expe imen al g oups. Clus e ing analysis e ealed 12 as ocy e sub-
popula ions (Fig. 2A) wi h unique gene signa u es (Fig. 2B, Table S1-S2), ag eeing wi h he high le el o
egional and unc ional he e ogenei y o his cell ype in he b ain. The clus e s di ided b oadly in o
wo g oups based on elencephalon and non- elencephalon iden i y, as p e iously desc ibed (Zeisel e
al., 2018) (Figs. 2A, B). Clus e s a1, a3 and a4 show a high exp ession o Ag , I ih3 and Slc6a11 which
a e ma ke s o non- elencephalon as ocy es. Clus e s a0, a2, a5, a6, a7, a8 and a9 show inc eased
exp ession o M ge8, Lhx2 and Ppp1 3g which a e ma ke s o elencephalon as ocy es (Fig. S4A).
Since we obse ed changes in plaque pa hology acco ding o APOE iso o ms in he co ical a eas, we
ocused on only he elencephalon as ocy es o u he analysis. Ou o hese, Clus e a2 was en iched
o homeos a ic genes o co ical as ocy es such as Epha5, Id3, Ca 2 (Mu ai and Pasquale, 2011;
Ba iuk e al., 2020; Thepa ambil e al., 2020). Sub en icula zone as ocy es we e en iched in Clus e
a5 (Slc38a1, Hopx, Zic1 and Zic4) (Miz ak e al., 2019; Hasel e al., 2021), hippocampal as ocy es in
Clus e a7 (Hopx, Hes5 and N 2 1) (Ha akeyama e al., 2004; Li e al., 2015; Zwei el e al., 2018;
Be acchi e al., 2020) and Clus e a9 (Hopx, G ap and Ig bp5) (Ye and D’E cole, 1998; Bushong e al.,
2002; Li e al., 2015), and s ia al as ocy es in Clus e a8 (C ym) (Chai e al., 2017; Olli ie e al., 2024).
We also iden i ied wo eac i e popula ions o as ocy es. Clus e a10 was mainly cha ac e ized by
down egula ion o homeos a ic as ocy e genes such as Slc1a2, Glul, Aqp4, and Gja1 (Fig. S4B), and
up egula ed G ap, Vim, S100a6 and Meg3 (Figs. 2B, S4B). Clus e a10 showed, in addi ion, inc eased
exp ession o plaque induced genes (PIGs) such as Cyba, Gpx4, Ig bp5, S100a6, Cd9, and G ap,
p e iously p oposed o be pa o he combined mic oglia-as ocy e esponse o amyloid plaques
(Chen e al., 2020) (Fig. S4C). Besides, Clus e a11 is cha ac e ized by up egula ion o in e e on genes
such as Cxcl10, I i 3, Iigp1 and I i m3, indica ing a cell-s a e o as ocy e associa ed o in lamma ion
(Hasel e al., 2021) (Fig. S4D).
Thus, ou APOE econs i u ion expe imen co e ed a wide ange o di e en as ocy e sub ypes and
cell s a es. O e all, di e en APOE iso o ms do no seem o ha e a majo e ec on he clus e s apa
om he APOEKO condi ion whe e Clus e a0 was s ongly en iched and Clus e a6 comple ely absen
(Fig. 2C). Func ional en ichmen (Gene On ology: Biological P ocess da abase) analysis o Clus e a6
showed en ichmen o e ms ela ed o nega i e egula ion o cellula (GO:0048523) and biological
(GO:0048519) p ocesses (Fig. 2D), d i en by genes such as Pde10a, Emd, N 4a2, and Ep s (Table S2).
Among he op ma ke s o Clus e a6 a e genes in ol ed in in lamma o y pa hways (Fos, Junb, N 4a2)
(Zenz e al., 2008; Es ada e al., 2020) and cell p oli e a ion and communica ion (B g2, Ccn1) (Jun and
Lau, 2011; Yunia i e al., 2019; Suzuki e al., 2021) (Fig. 2B, Table S1). Clus e a0 showed en ichmen
o e ms ela ed o s e ol (GO:0016126) and choles e ol (GO:0006695) biosyn he ic p ocesses (Fig.
2D), d i en by genes such as Msmo1, Insig1 and Hmgc (Table S2). To explo e he e ec o APOE
de iciency on as ocy e ansc ip omes u he , we pe o med di e en ial gene exp ession analysis o
compa e each expe imen al g oup agains he APOE3 g oup (Fig. 3A, Table S3). We ound ha he
highes numbe o DEGs we e in he APOEKO g oup- 375 UP and 321 DOWN genes, compa ed o
numbe o DEGs in APOE2 (p = 4.4e-64) o APOE4 (p = 4e-85). Top up egula ed genes in he APOEKO
as ocy es such as Ubb, Ubc and C sl (Fig. 3B, Table S3), a e in ol ed in p o ein deg ada ion, while
P gds (Kanekiyo e al., 2007; Kannaian e al., 2019) and Cs 3 (Sas e e al., 2004; Mi e al., 2007), a e
associa ed wi h inhibi ion o Aβ ib illiza ion, in line wi h absence o amyloid plaques in APOEKO mice.
O he up egula ed genes such as Nd g2 (Feng e al., 2022; Zhang e al., 2023), Gpx4 (Chen e al., 2020)
and Hmgb1 (Fan e al., 2016) a e known modula o s o as ocy e eac i i y.
We compa ed nex he di e en ial e ec s o APOE4 s. APOE3 and APOE2 s. APOE3 (Fig. 3D).
Compa ed o APOE3, bo h APOE4 and APOE2 signi ican ly up egula ed Pde10a, which was also highly
up egula ed in Clus e a6 (Fig. 2B) and showed minimal exp ession in APOEKO as ocy es (Fig. 3B, C).
Pde10a belongs o a amily o phosphodies e ases and hei inhibi ion has been sugges ed o modula e
AD and o he neu odegene a i e diso de s, po en ially ac i a ing CREB (cAMP esponse elemen -
binding)- dependen synap ic plas ici y and memo y o ma ion (O’Conno e al., 2004; Puzzo e al.,

2009; Ga cía-Os a e al., 2012), bu hese e ec s we e ela ed o exp ession o Pde10a in neu ons, no
as ocy es. Bo h APOE4 and APOE2 as ocy es also up egula ed eac i e ma ke s such as Nea 1 and
Cd9 (Chen e al., 2020; Habib e al., 2020; I win e al., 2023). APOE4 as ocy es speci ically up egula ed
se e al mi ochond ial genes (m -Nd1, m -Cy b, m -Rn 2). No ably, APOE4 also down egula ed se e al
genes in ol ed in au ophagy (Ac g1, Tax1bp1, Sqs m1, Lamp1, Lamp2) (Fig. 3D) and lipid me abolism
(Msmo1, Hmgcs1). On he o he hand, APOE2 as ocy es, simila o APOEKO, up egula ed genes
in ol ed in p o ein deg ada ion such as Ubb and Uba52, wi h down egula ion o lipid me abolism
ela ed genes (E nppl, Insig1) (Fig. 3D). Bo h APOE4 and APOE2 di e om APOE3 in he sense ha
APOE3 is cha ac e ized by up egula ed in e e on esponse genes such as I i27, Lgals3bp, I i 3 and
I i m3 (Fig. 3D). APOE3 as ocy es also up egula ed genes in ol ed in p o eos asis (Hspa8, Hsp90ab1,
Hspa5), Sesn3 ha educes eac i e oxygen species (ROS) in s ess induced s a e, genes egula ing
phospholipase enzymes/PI signaling (Acsl3, Pla2g7, Plcd4, P ex2), and synap ic plas ici y (Nna ). Thus,
di e en APOE iso o ms induce di e en ansc ip omic esponses in as ocy es in he AppNL-G-F mouse
model, wi h po en ial impai men o p o eos asis and au ophagy in APOE4-exp essing as ocy es.
We also explo ed he as ocy e ansc ip ome o exp ession le el and possible dys egula ion o
p e iously iden i ied AD isk genes (Bellenguez e al., 2022) (Fig. S4E). O e all, as ocy es exp essed
only a small numbe o hese isk genes. Ou o hese, Clu and Cox7c become highly up egula ed in he
absence o APOE, while Abca1 and So 1 ge down egula ed (Figs. 3B, S4E). Clu has been epo ed o
ha e he opposi e e ec o Apoe on amyloid pa hology, wi h o e exp ession o Clu in AD mice showing
dec eased plaque deposi ion (Chen e al., 2021). No ably, we saw dec eased exp ession o Fe m 2 and
C sb in APOE4 as ocy es (Fig. S4E), bu he di e ence was no signi ican compa ed o APOE3 g oup
(Table S3). A ecen s udy has shown ha down egula ion o Fe m 2 in as ocy es can lead o educ ion
o e i o y size and associa ed cogni i e de ici s (Endo e al., 2022). This sugges s ha APOE exp ession
in an amyloid en i onmen is linked o he modula ion o AD isk genes in as ocy es.
O e all, ou analysis indica es ha absence o exp ession o APOE has p o ound e ec s on as ocy e
ansc ip ional ea u es, indica ing he mul i unc ionali y o he p o ein. The di e ences be ween
APOE2, APOE3 and APOE4 a e less ou spoken, bu co e impo an unc ions such as in e e on
esponse genes, p o eos asis genes, PI signaling, lipid me abolism, and as salien obse a ion he
up egula ion o Pde10a.
Fib illa plaques induced by as ocy e-de i ed APOE elici a eac i e esponse in Apoe-de icien
mic oglia
Reac i e mic oglia s a es associa ed o amyloid plaques s ongly up egula e Apoe exp ession (Ke en-
Shaul e al., 2017; K asemann e al., 2017; Sala F ige io e al., 2019). We he e o e ques ioned whe he
APOE om as ocy es is necessa y and su icien o es o e he esponse o Apoe-de icien mic oglia
o amyloid plaques in ou model.
We used immuno luo escence o analyze he mo phological changes in mic oglia nea he as ocy e-
induced amyloid plaques. We indeed obse ed inc eased mic oglial clus e ing a ound he ib illa
plaque deposi s (Fig. 4A). To con i m his, we segmen ed indi idual mic oglia (Figs. S5A-B) and
classi ied hem based on hei p oximi y o amyloid plaques as “plaque-associa ed” (wi hin X34+
h esholded plaque a ea and up o 5 µm ou side a plaque’s edge) o “non-plaque-associa ed” (Figs.
S5C, D). Plaque-associa ed mic oglia clus e ed oge he wi h a signi ican ly lowe dis ance o
neighbou ing mic oglia, compa ed o he well- iled mic oglia p esen away om plaques (Fig. S5E).
Plaque-associa ed mic oglia also had signi ican ly lowe e i o y size (con ex a ea) (Fig. S5Fi) and o al
p ocess leng h (Fig. S5Fii), indica ing a less b anched, amoeboid mo phology, cha ac e is ic o eac i e
mic oglia. We con i med ha plaque-associa ed mic oglia s ained posi i e o phagocy ic ma ke Cd68
(Fig. 4B, S5G) and lysosomal p o ease C sd (Fig. 4C, S5H), cha ac e is ic o he DAM cell s a e (Ke en-
Shaul e al., 2017). Thus, Apoe-de icien mic oglia a e s ill able o adop a eac i e s a e in esponse o
ib illa plaques induced by as ocy e-de i ed APOE.
Nex , we wonde ed whe he he APOE iso o ms in luence mic oglial esponses, in e ms o numbe
and mo phological changes. We obse ed a signi ican ly highe numbe o mic oglia in he AAV-
ansduced egion o APOE4 compa ed o o he g oups (Fig. 4Di), bu eg ession analysis
demons a ed his was due o he s ong posi i e associa ion (R2 = 0.86, slope = 731.09, p = 3.7e-12) o
o al mic oglia numbe s wi h he le el o plaque load, independen o APOE iso o m (Fig. 4Dii).
In e es ingly, his inc ease in mic oglia popula ion wi h plaque load was d i en by inc eased numbe
o plaque-associa ed mic oglia (R2 = 0.97, slope = 973.13, p = 2e-16) while he non-plaque-associa ed
popula ion emained ela i ely s able despi e inc ease in plaque load (R2 = 0.5, slope = -242.04, p =
3.5e-05) (Fig. 4Dii). Addi ionally, he e we e no signi ican changes in mic oglial mo phology ( e i o y
size and o al p ocess leng h), o ei he non-plaque-associa ed (Fig. S5I) o plaque-associa ed
mic oglia (Fig. S5J), be ween he di e en expe imen al g oups. The e o e, a mo phological le el,
mic oglial esponses o ib illa plaques do no seem o depend on he iso o m o APOE p oduced by
as ocy es bu a e likely media ed di ec ly by he plaque load.
In o de o explo e he changes in mic oglial eac i i y a he ansc ip omic le el, we pe o med
d ople -based scRNAseq on CD11b+/CD45-low mic oglia isola ed a 6 mon hs o age om he mouse
b ains exp essing di e en APOE iso o ms in as ocy es. A e il e ing ou non-mic oglial cells and low-
quali y cells, we e ained 18,569 Apoe-de icien mic oglia o e he ou expe imen al g oups (APOEKO,
APOE2, APOE3 and APOE4). Unbiased clus e ing analysis led o he iden i ica ion o 7 sub-popula ions
o mic oglia (Figs. S6A-B, Table S4-S5). We did no ind majo shi s in cell s a es as obse ed in o he
expe imen al pa adigms ((Ke en-Shaul e al., 2017; Sala F ige io e al., 2019) p obably because he AAV
ec o es o es amyloid plaque o ma ion only in he a ea o ansduc ion (Fig. S2A). I is indeed
ema kable ha he e ec s o APOE exp ession in as ocy es on amyloid plaque o ma ion is limi ed
o hose b ain a ea whe e he APOE is exp essed, despi e APOE being a sec e ed p o ein. As we isola ed
mic oglia om he whole b ain, including he non- ansduced egions, only a ac ion o he o al pool
o mic oglia is exposed o he amyloid plaques. Howe e , we obse ed in e es ing shi s in gene
signa u es by pe o ming di e en ial gene exp ession analysis (Fig. S7A, Table S6). Among he
up egula ed genes in he APOEKO g oup we e se e al genes in ol ed in he cy okine esponse s a e
(Ccl3, Ccl4, Ccl12, Il1a, Gp 84, Jun) (Fig. S7B). Thus, APOEKO, which shows absence o ib illa plaques
and inc eased soluble Aβ agg ega es, is associa ed wi h mic oglia ending o adop a cy okine esponse
s a e. In e es ingly, among he di e en se s o DEGs, he highes in e sec ion was obse ed be ween
APOEKO and APOE4 (p = 4.2e-78), whe e 52 genes we e commonly up egula ed in he wo g oups (Fig.
S7A, S7C, Table S6) including signi ican o e lap (p = 2.8e-11) o a small subse o plaque induced genes
(PIGs) (C1qb, C1qc, Cd63, Cd9, Cs 3, Fce 1g, Hexb, Ly86) (Chen e al., 2020) (Fig. S7C, Table S6). Thei
up egula ion in he APOEKO g oup which lack ib illa plaques, sugges s ha induc ion o hese plaque-
associa ed mic oglial ma ke s possibly p ecedes ib illa plaque o ma ion in amyloid models. We
compa ed also he di e en ially exp essed genes in APOE4 s. APOE3 and APOE2 s. APOE3 o
unde s and he di e en ial e ec o he disease p omo ing and he disease p o ec i e allele e sus
APOE3 ha is conside ed neu al (Fig. S7D). We obse ed ha compa ed o APOE2, APOE4 had a
signi ican up egula ion o se e al DAM ma ke genes such as Cd74, Cs 7, Cd63, Cd9, Ccl6, complemen
cascade genes (C1qa, C1qb, C1qc) and he p o-in lamma o y gene N kbia (Sousa e al., 2018). This was
accompanied by down egula ion o homeos a ic genes (Cx3c 1, P2 y13, Cc 5, Siglech). Also, Me k
which is necessa y o de ec ion and engul men o amyloid plaques by mic oglia (Huang e al., 2021),
and Inpp5d, a known AD isk ac o which has a ole in limi ing plaque o ma ion (Cas anio e al.,
2023), a e down egula ed in APOE4. In e es ingly, compa ed o APOE2 and APOE4, APOE3 up egula ed
se e al ma ke s o in e e on esponse in mic oglia (I i m3, I i 3, I i27l2a, Lgals3bp, S a 1) (Figs. S7D).
APOE3 also showed an up egula ion o an igen p esen a ion genes (H2-K1, H2-D1, H2-Q7).
O e all, ou da a show ha Apoe-de icien mic oglia a e able o eac o ib illa plaques moun ing a
DAM-like esponse when APOE is p o ided ia as ocy es.
As ocy e-de i ed APOE is su icien o ib illa amyloid plaque pa hology
Ou da a sugges wo possibili ies: ei he APOE om he as ocy es ac s di ec ly on amyloid plaque
o ma ion and hen mic oglia espond o he plaques o med, o APOE di ec ly modi ies mic oglia
esponses and hus hey sculp he amyloid plaques wi h he help o he APOE deli e ed by he
as ocy es.
To add ess his, we deple ed mic oglia using PLX3397 ea men a 2 mon hs o age be o e he ib illa
plaques become isible and con inued ea men un il he mice we e 6 mon hs o age (Fig. 5A). Despi e
s ong deple ion o mic oglia (pe cen age educ ions o APOE2: 83%, APOE3: 77%, APOE4: 87%) unde
hose condi ions (Figs. 5B, S8A), ib illa plaque deposi ion in APOE exp essing co ical egions
emained subs an ial, wi h only APOE4 g oup showing s a is ically signi ican dec ease in plaque load
upon deple ion (Fig. 5C). As deple ion o mic oglia was pe o med a 2 mon hs and ini ial amyloid
plaque seeding migh ha e occu ed al eady be o e ha ime poin , we canno ule ou he possibili y
ha mic oglia played a ole in he APOE induced ini ial seeding o he plaques o ha he emaining
mic oglia a e su icien o induce plaque seeds.
Since we ha e ea lie seen ha APOE le els also in luence Aβ plaque deposi ion (Fig. 1E), we used
mul iple eg ession analysis o accoun o he con ounding e ec o APOE le els while compa ing
guanidine-soluble Aβ42 le els be ween PLX3397- ea ed and un ea ed animals in each expe imen al
g oup (Fig. 5D). I is no able ha especially in he APOE3 mice, he APOE le els a e inc eased in he
mic oglia deple ed b ains (Fig. S8Bii). This inc ease in APOE le els was p esen in APOE4 mice as well
(Fig. S8Biii), wi h APOE2 mice showing a simila non-signi ican end (Fig. S8Bi). In e es ingly, o he
APOE3 g oup, like abo e (Fig. 1E), we ound no signi ican associa ion be ween Aβ42 le els and APOE
le els and he e was no signi ican shi in mean Aβ42 le els (β = -28.53, SE = 38.31, p = 0.4708) wi h
PLX- ea men . On he o he hand, bo h he APOE2 g oup (β = -148.2, SE = 32.22, p = 0.0013) and he
APOE4 g oup (β = -357.1, SE = 60.64, p = 0.0002) showed a s a is ically signi ican dec ease in mean
Aβ42 le els wi h PLX- ea men compa ed o he non- ea ed g oup (Fig. 5D). The e o e, his sugges s
ha while APOE om as ocy es is su icien o plaque deposi ion a 6mon hs o age in his model,
mic oglia play a ole in modula ing Aβ pa hology in APOE2 and APOE4 exp essing mice.
DISCUSSION
We in es iga ed he ole o APOE in he in e play o as ocy es, mic oglia and amyloid plaques. The
s udy con i ms ha APOE de i ed om as ocy es is su icien o induce ib illa Aβ plaques. I was
unexpec ed ha he p o ec i e APOE2 iso o m had simila , albei less p onounced, e ec s as
pa hogenic APOE4 on amyloid plaque o ma ion in ou expe imen s. Thus, i s p o ec i e ole in AD is
likely downs eam o he amyloid plaque o ma ion, a ec ing he cellula phase o he disease, i.e. he
esponse o he cells in he b ain o amyloid plaques (De S oope and Ka an, 2016).
P e ious s udies ha e shown ha knocking ou mouse Apoe leads o a signi ican educ ion o ib illa
plaque o ma ion in di e en AD mouse models (Bales e al., 1999; Hol zman, Fagan, e al., 2000; Ul ich
e al., 2018). We con i m he e in he AppNL-G-F mouse ha despi e he p esence o s ong agg ega ion
p omo ing mu a ions in his pa icula Aβ sequence, Apoe dele ion almos annihila es he gene a ion
o ib illa amyloid plaques, while inc easing he soluble Aβ agg ega e species (Fig. 1). These species
induce a cy okine esponse (Mancuso e al., 2019) in he Apoe-de icien mic oglia. In e es ingly, Apoe-
de icien as ocy es up egula e genes associa ed wi h inhibi ion o Aβ ib illiza ion (P gds and Cs 3)
and egula ion o p o ein deg ada ion (Ubb, Ubc and C sl) (Wegiel e al., 2000; Apel , Ach and Schliebs,
2003; Liu, Hu, e al., 2017; Da is e al., 2021). O e all, loss o APOE exp ession has clea ly mul iple
e ec s on he as ocy e-amyloid plaque-mic oglia in e ac ions, in acco dance wi h he cen al ole o
APOE in his impo an aspec o AD pa hogenesis. I has been epo ed ha educ ion o ib illa
amyloid plaques ia as ocy ic-APOE knockou (Xiong e al., 2023), bu no mic oglial-APOE knockou
(Henning ield e al., 2022), esul s in a shi owa ds o ma ion o ce eb al amyloid angiopa hy (CAA).
Howe e , we did no in es iga e he o ma ion o CAA in ou model and u u e s udies will be needed
o unde s and his aspec o cell-speci ic APOE on plaque pa hology.
Res o a ion o APOE exp ession in as ocy es alone is su icien o induce o ma ion o ib illa plaques
and mic oglia eac i i y, ex ending indings om p e ious s udies (Hol zman, Bales, e al., 2000; Liu,
Zhao, e al., 2017; Mahan e al., 2022). In gene al expe imen s sugges he o de o APOE4 > APOE3 >
APOE2 o e ec s on amyloid deposi ion (Fagan e al., 2002; Cas ellano e al., 2011; Hud y e al., 2013;
Zhao e al., 2016), and he e o e ou obse a ion ha APOE2 is mo e simila o APOE4 han APOE3 in
ha ega d in ou model, is somewha puzzling. Ne e heless o he s udies ha e also epo ed ha
APOE2 exp ession is associa ed wi h simila o e en highe plaque load han APOE3 (Doda e al.,
2005; Bales e al., 2009; Youmans e al., 2012; Rod iguez e al., 2014). In e es ingly, ha ing an APOE2
allele (Be lau e al., 2007, 2009) o he a e iso o m APOE3Ch (Ch is chu ch a ian ) (A boleda-
Velasquez e al., 2019), shields pa ien s agains demen ia e en in he p esence o high amyloid plaque
load. The p o ec i e e ec s o APOE3Ch downs eam o he amyloid plaque o ma ion (on which a
ela i e modes e ec was obse ed) we e also seen in an AD amyloid mouse model whe e
o e exp ession had s onge e ec in educing au seeding and p opaga ion along wi h neu onal
dys ophy a ound plaques upon injec ion o AD- au b ain ex ac (Chen e al., 2023).
No consensus exis s whe he and how addi ional pa hophysiological unc ions o APOE2 and APOE4
con ibu e o he speci ic isk o AD. A majo p oblem o cla i y he ole o APOE in Alzheime ’s disease
speci ically and o de ine a he apeu ic hypo hesis o suppo a ansla ional app oach, is he
pleio opic unc ion o APOE. Indeed, APOE4 has been implica ed in he p oin lamma o y s a e in
mic oglia (B own e al., 2002; Col on, B own and Vi ek, 2005; Vi ek, B own and Col on, 2009; Lin e al.,
2018; Lan anco e al., 2021; Se ano-Pozo e al., 2021) and as ocy es (A naud e al., 2022);
endolysosomal dys unc ion (Ji e al., 2002; Nu iel e al., 2017; P asad and Rao, 2018; Xian e al., 2018;
Pohlkamp e al., 2021); impai ed au ophagy (Simono i ch e al., 2016, 2019; Pa con e al., 2018);
inducing and p omo ing sp eading o au pa hology, likely media ed by APOE ecep o s (B ech e al.,
2004; Shi e al., 2017, 2021; Rauch e al., 2020; Wang e al., 2021); neu odegene a ion (Agos a e al.,
2009; Susan o e al., 2015; Lup on e al., 2016); main aining in eg i y o pe icy es and blood b ain
ba ie (Bell e al., 2012; Mon agne e al., 2020, 2021); and impai ed myelina ion in oligodend ocy es
(Blancha d e al., 2022). We sugges ha he APOE2 p o ec i e e ec is la gely d i en by e ec s on he
cellula eac ions downs eam o he amyloid plaque o ma ion. We canno exclude howe e ha he
APOE2 p omo ing e ec on amyloid plaques as seen in he cu en wo k is model speci ic, and ha
exp essing APOE2 also in o he cell ypes migh induce amyloid plaque educing e ec s. I has also
been shown ha APOE2 exp ession leads o inc eased APOE le els in mouse models and human b ain
(Riddell e al., 2008; Bales e al., 2009; Coneje o-Goldbe g e al., 2014), which can in luence amyloid
pa hology. This is no ecapi ula ed in ou model due o inhe en limi a ion o AAV-media ed
o e exp ession.
APOE3 exp ession in as ocy es in ou expe imen s esul s in wo phenomena: (1) he es o a ion o
amyloid plaque o ma ion a 6 mon hs and (2) he dec ease in soluble Aβ agg ega es in hese animals.
A i s glance, inc eased ib iliza ion migh explain dec eased le els o soluble Aβ agg ega es, bu he
APOE2 allele causes mo e amyloid plaques han APOE3 in ou model, while i also lea es mo e soluble
Aβ agg ega es in he ex acellula milieu (Figs. 1C and 1G). In addi ion, APOE2, bu no APOE3, shows
a dose dependen ela ionship be ween le el o exp ession and amyloid agg ega ion (Fig. 1E), as
APOE4 does. The esul s hus indica e ha he ela ionship be ween APOE geno ypes and Aβ plaque
o ma ion is no simple. Fo he in e p e a ion o he da a, i is impo an o ake in o accoun ha
APOE is sec e ed he e only by as ocy es bu is aken up ia LDL ecep o s in o he endosomal
compa men s o all cell ypes in he b ain, including he mic oglia and as ocy es. Aβ, when
endocy osed, becomes concen a ed in he acidic endosomal compa men s, which p omo es amyloid
ib il o ma ion and hen sec e ion (Knaue e al., 1992; Chung e al., 1999; Hu e al., 2009). The e o e,
up ake o Aβ bound o APOE migh cause inc eased concen a ions o Aβ in hese acidic compa men s.
Thus, in he absence o APOE, clea ance o Aβ is lowe ed, esul ing in inc eased soluble Aβ agg ega es
in he medium as demons a ed wi h he APOEKO. When APOE3 is p esen , Aβ pep ides a e aken up
by as ocy es and mic oglia, esul ing in a dec ease o soluble Aβ (Figs 1F-G). This p ocess is ecep o
we e calcula ed o quan i y hei mo phological changes (Peng e al., 2010; Xiao and Peng, 2013). Fo
coun ing numbe o mic oglia associa ed o each plaque, 3d z-s acks we e la ened o i s pos
mic oglia segmen a ion. X-34-posi i e plaques we e h esholded and segmen ed as desc ibed abo e.
Any mic oglia whose cen oid was loca ed wi hin X-34-posi i e h esholded plaque a ea and up o 5
µm ou side a plaque’s edge was conside ed as plaque-associa ed mic oglia. Fo each mic oglia, he
mean dis ance o i s h ee nea es neighbou s we e also calcula ed as a measu e o clus e ing.
FACS isola ion o mic oglia and as ocy es
Fo sample collec ion, 6 mon hs old mice we e anaes he ized using CO2 chambe and pe used wi h
ice-cold 1x DPBS (Gibco, #14190-144). Each mouse b ain, excluding ce ebellum and ol ac o y bulbs
we e dissec ed and placed in FACS bu e (1x DPBS, 2% FCS and 2 mM EDTA) + 5 μM Ac inomycin D
(Ac D, Sigma, #A1410-5MG) o ansc ip omics. B ains we e mechanically and enzyma ically
dissocia ed using Mil enyi neu al issue dissocia ion ki P (Mil enyi, #130-092-628) supplemen ed wi h
5 μM Ac D. Nex , samples we e passed h ough a 70 μm s aine (BD2 Falcon), washed in 10 ml o ice-
cold FACS bu e + 5 μM Ac D and spun a 300 c o 15 minu es a 4 °C. No e ha 5 μM Ac D was kep
du ing collec ion and enzyma ic dissocia ion o he issue o p ese e ansc ip ional s a e and p e en
a i icial induc ion s ess- esponse genes du ing issue dissocia ion (Pe occhi e al., 2007; Ma sh e al.,
2022). Ac D was emo ed om he myelin emo al s ep o p e en oxici y de i ed om long- e m
exposu e. Following dissocia ion, myelin was emo ed by esuspending pelle ed cells in o 30% iso onic
Pe coll (GE Heal hca e, #17-5445-02) and cen i uging a 300 c o 15 minu es a 4 °C. Accumula ing
laye s o myelin and cellula deb is we e disca ded and Fc ecep o s we e blocked in FcR blocking
solu ion (1:10, Mil enyi, #130-092-575) in cold FACS bu e o 10 minu es a 4°C. Nex , cells we e
washed in 5 ml o FACS bu e and pelle ed cells spli in o wo suspensions (one each o as ocy e and
mic oglia so ing), and incuba ed wi h he ollowing an ibodies: o mic oglia- PE-CD11b (Mil enyi,
#130-113-806, 1:50), BV421-CD45 (BD, #563890, 1:500); o as ocy es- APC-ACSA2 (Mil enyi, #130-
116-142, 1:200), BV421-CD11b (BD, #562605, 1:500), PE-O4 (Mil enyi, #130-117-357, 1:50); in cold
FACS bu e o 30 minu es a 4 °C. To alSeq™-A cell hashing an ibodies (1:500, Biolegend) and iabili y
dye (eFluo 780, The mo Fishe Scien i ic, #65-0865-14) we e also added o bo h ac ions du ing he
incuba ion s ep. A e incuba ion, cells we e washed, and he pelle was esuspended in 500 μl o FACS
bu e and passed h ough 35 μm s aine p io so ing. Fo isola ing mic oglia, cell suspension was
loaded in o he inpu chambe o MACSQuan Ty o Ca idge and cells we e so ed on MACSQuan
Ty o Cell So e a 4 °C, based on CD11b-posi i e and CD45-low popula ions. Fo isola ing AAV-
ansduced as ocy es, cells we e so ed on A ia Fusion, based on nega i e ga ing o O4 and CD11b
popula ions, ollowed by posi i e ga ing o double-posi i e ACSA2 and mChe y popula ions. FACS
da a was analysed using FCS exp ess so wa e.
Single cell lib a y p epa a ion and sequencing
Fo single-cell RNA sequencing mic oglia and as ocy es, 20,000-30,000 cells o each ype om each
mouse we e so ed as desc ibed abo e and dilu ed o a inal concen a ion o 1000 cells/µl. All he
samples we e indi idually hashed using To alSeq™-A cell hashing an ibodies, 3000 cells/animal we e
pooled and loaded on o he Ch omium Nex GEM Chip G (PN #2000177). The DNA lib a y p epa a ions
we e gene a ed ollowing manu ac u e s’ ins uc ions (CG000204 Ch omium Nex GEM Single Cell 3’
Reagen Ki s 3.1). In pa allel he hash ag oligo lib a ies we e p epa ed acco ding o manu ac u e s’
ins uc ions (BioLegend - To alSeq™-A An ibodies and Cell Hashing wi h 10x Single Cell 3’ Reagen Ki
3 3.1 P o ocol) using 16cycles o he index PCR. F om he 4 expe imen al g oups (APOEKO, APOE2,
AOE3, and APOE4), a o al o 8 lib a ies (APOEKO_as ocy es, APOE2_as ocy es, APOE3_as ocy es,
APOE4_as ocy es, APOEKO_mic oglia, APOE2_mic oglia, APOE3_mic oglia, APOE4_mic oglia),
con aining 22 biological eplica es (n(APOEKO)= 2, n(APOE2)= 3, n(APOE3)=3, n(APOE4)= 3), we e
sequenced (se ice p o ided by BGI Genomics, China) a ge ing 90% mRNA and 10% hash ag oligo
lib a y (50,000 eads/cell) on a DNBSEQ-G400 (MGI Tech) pla o m wi h he ecommended ead
leng hs by 10X Genomics wo k low.

Analysis o single-cell RNA sequencing da ase s
Alignmen and analysis pipeline
Raw BCL iles we e aligned o a modi ied GRCm39 mouse genome, which was appended wi h he
human APOE gene. We used Bed ools 2.27.1 o ex ac he human APOE sequence and a housand
base bu e om GRCh38 (Quinlan and Hall, 2010). This was hen appended o he mu ine GRCm39
genome in an a i icial ch omosome. The esul ing genome was hen il e ed using he mkg unc ion
om Cell ange (6.1.2), in acco dance wi h 10x guidelines (Zheng e al., 2017) and subsequen ly
p ocessed in o a unc ional e e ence using cell ange mk e . Finally, each o he eigh scRNAseq
lib a ies we e aligned o a cus om genome using Cell ange ’s coun unc ion which ma ches each
ansc ip o i s co esponding gene using he STAR aligne (Dobin e al., 2013). Raw coun ma ices
we e impo ed in R ( 4.3.3) o da a analysis. Da ase s we e analysed using R packages: Seu a (4.4)
(Hao e al., 2021), sca e (1.30.1) (McCa hy e al., 2017), clus ee (0.5.1) (Zappia and Oshlack, 2018),
and Supe Exac Tes (1.1) (Wang, Zhao and Zhang, 2015). Visualiza ions we e done using unc ions om
Seu a , Supe Exac Tes o ggplo 2 (3.5).
Quali y con ol o cells and samples
In his o e exp ession model, since APOE is unde a cons i u i e p omo o , and hence no a pa o he
no mal ansc ip ional egula o y ne wo ks, we excluded human APOE gene exp ession om bo h
as ocy e and mic oglia da ase s p io o he ollowing s eps.
Fo each lib a y o mic oglia, low-le el quali y con ol was done by i s il e ing ou cells wi h < 200
genes de ec ed and genes exp essed in less han 3 cells, om he aw coun s ma ix. Subsequen
il e ing in ol ed h ee s eps. S ep 1: Cell hashing in o ma ion was added o each lib a y using
Seu a ::HTODemux() unc ion o assign each cell as ei he single s, double s o nega i es. Only single s
we e e ained, as double s and single s had abno mally high and low numbe o eads, espec i ely,
indica i e o poo -quali y cells. S ep 2: Lib a ies om he 4 expe imen al g oups we e me ged using
Seu a ::me ge() unc ion, o ha e in o al 23,495 cells (Fig. S9A). Cell clus e ing was pe o med using
he pipeline desc ibed below. Cell ype anno a ion was done using seu a ::AddModuleSco e() o assign
cell sco es based on p e iously iden i ied signa u es o di e en b ain cell ypes (Zeisel e al., 2018).
~9% o cells had non-mic oglial iden i y wi h hal o hose ha ing mac ophage iden i y. Con amina ing
cell ypes we e iden i ied (Figs. S9A, B) as mac ophages (mic oglia QC Clus e 4), endo helial cells
(mic oglia QC Clus e 5), mix o as ocy e and mic oglial cells (mic oglia QC Clus e 6), monocy es
(mic oglia QC Clus e s 7 and 9), and oligodend ocy es (mic oglia QC Clus e 10). Ou o he es ~10%
sepa a ed ou as a clus e (mic oglia QC Clus e 2) wi h low numbe o eads and gene coun (Fig. S9C).
S ep 3: A e il e ing ou hese clus e s, each lib a y was spli in o indi idual ones and quali y con ol
was e ined wi h an addi ional s ep using sca e ::isOu lie () unc ion. He e cells wi h numbe o eads
o genes o pe cen age o mi ochond ial genes (%mi o), ou side n median absolu e de ia ion (n( eads)
= +/- 4, n(genes) = - 4, n(%mi o) = + 5) om lib a y median, we e emo ed. Finally, 18,569 mic oglia
we e e ained o analysis as good-quali y cells.
Fo as ocy es, he same s eps as desc ibed abo e wi h ew modi ica ions we e execu ed. S ep 1: Cell
hashing did no wo k e icien ly o as ocy es, as a good p opo ion o cells iden i ied as nega i es had
ead coun s in he same ange as Single cells. Thus, we used he o he wo s eps o il e ou low-
quali y cells. S ep 2: Lib a ies om he 4 expe imen al g oups we e me ged o ha e in o al 32,982
cells (Fig. S9D). A e cell ype anno a ion as desc ibed abo e, ~7% o cells had non-as ocy e iden i y.
Con amina ing cell ypes we e iden i ied (Figs. S9D, E) as ependymal cells (as ocy e QC Clus e 5),
mu al cells (as ocy e QC Clus e 7), mic oglia (as ocy e QC Clus e 9), endo helial cells (as ocy e QC
Clus e 10) and a popula ion o unde e mined iden i y (as ocy e QC Clus e 8). Ou o he es , ~31%
sepa a ed ou as clus e s (as ocy e QC Clus e s 1 and 6) wi h low numbe o eads and gene coun
(Fig. S9F). S ep 3: A e il e ing ou hese clus e s, each lib a ies we e spli in o indi idual ones using
Seu a ::Spli Objec () and quali y con ol was e ined wi h an addi ional s ep using sca e ::isOu lie ()
unc ion. He e cells wi h numbe o eads o genes o pe cen age o mi ochond ial genes (%mi o),
ou side n median absolu e de ia ion (n( eads) = +/- 2.5, n(genes) = - 2.5, n(%mi o) = + 5) om lib a y
median, we e emo ed. Finally,18,667 as ocy es we e e ained o analysis as good-quali y cells.
Lib a y In eg a ion and Cell Clus e ing
To emo e ba ch e ec s be ween lib a ies while cell clus e ing, he Seu a Objec was spli in o
indi idual lib a ies using Seu a ::Spli Objec (). Each lib a ies we e indi idually no malized
(Seu a ::No malizeDa a) and a iable genes iden i ied (Seu a ::FindVa iableFea u es). Genes ha a e
epea edly a iable ac oss lib a ies we e selec ed o in eg a ion (Seu a :: Selec In eg a ionFea u es()).
In eg a ion ancho s we e iden i ied (Seu a ::FindIn eg a ionAncho s()) and ba ch co ec ion was
pe o med (Seu a ::In eg a eDa a()). A e scaling and cen e ing genes (Seu a ::ScaleDa a()) in he
in eg a ed da ase , P incipal Componen Analysis (PCA) was pe o med. n dimensions om he PCA (
n(as ocy es) = 50; n(mic oglia) = 40) we e selec ed o iden i ying clus e s (Seu a ::FindNeighbo s()
and Seu a ::FindClus e s)). To a oid o e -clus e ing o unde -clus e ing, mul iple esolu ions we e used
o he FindClus e () unc ion and clus ee::clus ee() was used o selec ing an op imum esolu ion
(as ocy es = 0.7; mic oglia = 0.4). Fo isualiza ion o clus e s, dimensionali y educ ion by Uni o m
Mani old App oxima ion and P ojec ion (UMAP), was pe o med wi h he Seu a ::RunUMAP()
unc ion. Clus e ma ke s o as ocy es and mic oglia we e iden i ied using Seu a ::FindAllMa ke s()
unc ion which calcula es di e en ial exp ession o each clus e agains he es . As ocy e clus e s
we e sco ed using seu a ::AddModuleSco e(), o ma ke s om p e iously published genese s o
as ocy e cell s a es (Zamanian e al., 2012; Hasel e al., 2021), plaque-induced genes (PIGs) (Chen e
al., 2020) and AD isk genes (Bellenguez e al., 2022).
Di e en ial Exp ession
Fo bo h as ocy es ( elencephalon sub ype) and mic oglia, di e en ial exp ession was pe o med
be ween he 4 expe imen al g oups (each o APOEKO, APOE2 and APOE4 compa ed o APOE3), using
Seu a ::FindMa ke s(). Genes exp essed in a leas 5% o cells we e calcula ed. Fo calcula ing p- alues,
MAST was used which is a GLM- amewo k ha ea s cellula de ec ion a e as a co a ia e. Since he e
we e di e ences in he gende a io be ween he g oups, gende iden i y was eg essed ou du ing
di e en ial exp ession using la en . a s pa ame e . Pos -analysis, genes wi h hei adjus ed P alue <
0.05 (pos -hoc, Bon e oni co ec ion) and | log( old-change) | > 0.2 we e conside ed as signi ican .
Supe Exac Tes (Wang, Zhao and Zhang, 2015) was used o calcula e s a is ical signi icance (Tables S3,
S6) and isualize o e lap o di e en ially exp essed genes while compa ing each g oup wi h APOE3.
Size o backg ound gene popula ion (n) o Supe Exac Tes was collec ed om o al numbe o genes
in he Seu a objec o as ocy es and mic oglia lib a y (n(as ocy es)=25321, n(mic oglia)=23819).
Volcano plo s and quad an plo s using ggplo 2::ggplo () and do plo s using Seu a ::Do Plo () we e
used o isualize di e en ially exp essed genes be ween he g oups.
Gene On ology (GO) En ichmen Analysis
Fo as ocy es and mic oglia, clus e ma ke s iden i ied, as men ioned abo e, was used o pe o m
unc ional en ichmen analysis. This was done using g:P o ile web in e ace
(h ps://bii .cs.u .ee/gp o ile /gos ), whe e he clus e ma ke s we e que ied agains GO biological
p ocess da abase, wi h g:SCS mul iple es ing co ec ion me hod applying signi icance h eshold o
0.05 (Kolbe g e al., 2023). Ba plo s using ggplo 2::ggplo () was used o isualize he op 10
signi ican ly en iched e ms o as ocy e clus e s a0 and a6.
S a is ical analysis
Numbe o samples included in each expe imen is included in he esul s sec ion/ igu e legends.
Sample size es ima ion was based on pilo expe imen s. All s a is ical compa isons we e pe o med in
G aphpad P ism ( 10.0.2) o R ( 4.3.3). S a is ical es used o each expe imen is epo ed in he
igu e legends. Fo analysis whe e indi idual plaques o mic oglia we e plo ed as da a poin s, mixed
e ec s models we e used o accoun o he a iabili y be ween indi idual mice and Q-Q plo s we e
used o assess he no mali y o he esiduals, using he lme4 package o mixed e ec s models and he
“es ima ed ma ginal means” package in R. Fo compa ison o ASM alues in Figs. 2 and 5, he linea
mixed-e ec s model was used wi h mouse ID and diame e o plaques as con ounding ac o s. ASM
esponse a iable was also log- ans o med o imp o e he no mali y o he esiduals. Di e en ial
abundance o plaque-associa ed mic oglia a ound indi idual plaques in Fig. 4 was es ed using binomial
gene alized linea mixed-e ec s model wi h mouse ID as con ounding ac o . Mo phological
di e ences be ween plaque-associa ed mic oglia and non-plaque-associa ed mic oglia in Fig S3 was
es ed using linea mixed-e ec s model wi h mouse ID as con ounding ac o . Di e ences in o al
numbe o di e en ially exp essed genes be ween g oups (compa ed o APOE3) was es ed using
Fishe ’s exac es wi h Bon e oni co ec ion o p- alues. En ichmen o plaque-induced genes (PIGs)
(Chen e al., 2020) in in e sec ion o di e en ially exp essed genes in mic oglia da ase was es ed
using hype geome ic es wi h Bon e oni co ec ion o p- alues.
DATA AVAILABILITY
Raw da a used o his esea ch a icle is a ailable upon eques o he co esponding au ho s. Single
cell RNA sequencing da a gene a ed in his s udy a e a ailable a Gene Exp ession Omnibus (GEO)
da abase wi h accession numbe GSE252454.
ACKNOWLEDGMENTS
We hank Ve onique Hend ickx and Ambe Claes o help wi h he mouse colonies and D ies T’Syen,
Ka ien Ho é and Ka leen C aessae s o echnical assis ance. Mouse expe imen s we e suppo ed by
CBD Mouse Expe ise Uni (KU Leu en and VIB). Con ocal mic oscopy was pe o med in he VIB Bio
Imaging Co e (LiMoNe). Pa o he igu e schema ics was c ea ed wi h BioRende (bio ende .com).
FUNDING SOURCE
This p ojec ecei ed unding om he Eu opean Resea ch Council (ERC) unde he Eu opean Union’s
Ho izon 2020 Resea ch and Inno a ion P og amme (g an ag eemen no. ERC-834682 CELLPHASE_AD).
This wo k was also suppo ed by unding om UKRI and he Medical Resea ch Council (MR/Y014847/1)
ia he Demen ia Resea ch Ins i u e. Fu he suppo was gi en by he Flande s Ins i u e o
Bio echnology (VIB zw), a Me husalem g an om KU Leu en and he Flemish Go e nmen , he Fonds
oo We enschappelijk Onde zoek, KU Leu en, The Queen Elisabe h Medical Founda ion o
Neu osciences, he Opening he Fu u e campaign o he Leu en Uni e si ai Fonds, The Belgian
Alzheime Resea ch Founda ion (SAO-FRA) and he Alzheime ’s Associa ion USA. B.D.S. holds he Bax-
Vanlu elen Chai o Alzheime ’s Disease. D.K. is unded by UK Demen ia Resea ch Ins i u e, which
ecei es i s unding om DRI L d. unded by he UK Medical Resea ch Council and he Royal Socie y.
A.M.A. is unded by he Minis e io de Ciencia e Inno ación unde g an no.
MCIN/AEI/10.13039/501100011033 (PID2021-125443OB-100 also by FEDER Una mane a de hace
Eu opa and RYC2020-029494-I by FSE in ie e en u u u o), he Alzheime ’s Associa ion (AARG-21-
850389), and he Basque Go e nmen (PIBA-2020-1-0030).
AUTHOR CONTRIBUTIONS
P.P., A.M.A. and B.D.S. concei ed and designed he s udy and planned he expe imen s. M.F. checked
all he bioin o ma ics da a and hei in e p e a ions. P.P. pe o med expe imen s and analyzed da a
oge he wi h B.D.S. and A.M.A.. D.M. p o ided bioin o ma ic suppo and analyzed da a oge he wi h
M.F. E.F. pe o med SimPull expe imen s, unde he supe ision o D.K.. L.W., J.L., and S.P., assis ed
wi h single cell RNA-sequencing expe imen s. L.S. gene a ed he mouse lines. D.S. assis ed wi h
gene a ion o AAV ec o s. A.S. assis ed wi h immunohis ochemis y expe imen s. R.M. and S.B.
assis ed wi h designing and execu ion o mic oglia deple ion s udies. All au ho s discussed he esul s
and commen ed on he manusc ip
CONFLICT OF INTEREST
B.D.S. is o has been a consul an o Eli Lilly, Biogen, Janssen Pha maceu ica, Eisai, AbbVie and o he
companies. B.D.S is also a scien i ic ounde o Augus ine The apeu ics and a scien i ic ounde and
s ockholde o Muna The apeu ics.
FIGURE LEGENDS
Main Figu es
Figu e 1. Exp ession o APOE in as ocy es is su icien o es o e ib illa Aβ plaque
pa hology. (A) AAV2/8 ec o was used o exp ess human APOE unde he ALDH1L1 p omo e . AAV
pa icles we e injec ed ia ICV in AppNL-G-F x Apoe-/- neona al mice on pos na al days wo o h ee
(P2/P3). Samples we e collec ed o analysis a 6 mon hs o age. (B) IF images o AAV- ansduced
co ical egions (see Fig. S2A) a 6 mon hs o age. mChe y ( ed) shows he ansduced as ocy es. X-
34 (whi e) shows he ib illa plaque deposi s. Scale ba : 100 µm. (C) Ba plo s showing (i) numbe o
X-34+ plaques pe ield o iew (FOV, 20x magni ica ion) and (ii) a ea o X-34+ plaques (as ac ion o
o al a ea in FOV, 20x magni ica ion), in he AAV- ansduced co ical egion. Da a poin s show mean
alue o 3 ields o iew pe mouse (n= 6-10 mice pe g oup). (D) Ba plo showing o al APOE le els in
b ain homogena es measu ed by MSD-ELISA. Da a poin s show mean alue o 2 echnical eplica es
pe mouse (n= 8-10 mice pe g oup). ns = non-signi ican . (E) Sca e plo showing linea eg ession
be ween o al APOE le els (x-axis) and co esponding guanidine-soluble Aβ42 le els (y-axis), measu ed
by MSD-ELISA. Da a poin s show mean alue o 2 echnical eplica es pe mouse (n= 8-10 mice pe
g oup). (F) TIRF images o mab 6E10-posi i e Aβ agg ega es (whi e) cap u ed om soaked b ain
ac ion using SimPull echnique. Scale ba : 10 µm (G) Quan i ica ion o mAb 6E10-posi i e Aβ
agg ega es. Da a poin s show mean alue o 3 echnical eplica es pe mouse (n= 4 mice pe g oup, 9
ield o iews pe echnical eplica e). Colou legend o expe imen al g oups in (D), (E) and (G) a e
indica ed in (C). S a is ical es s: Da a p esen ed as mean ± SEM (C), (D), and (G). One-way ANOVA and
Tukey’s mul iple compa ison es (C), (D), and (G). Signi icance shown o pai wise compa isons *p <
0.05; **p < 0.01; ***p < 0.001, ****p < 0.0001
Figu e 2. APOE exp ession and as ocy e cell-s a es in AppNL-G-F mouse b ain. (A) UMAP plo
showing18,667 as ocy es (mChe y+/Acsa2+) so ed om 6 mon hs old mouse b ains om he ou
expe imen al g oups (n= 2 o 3 mice pe g oup). Di e en sub-popula ions iden i ied ha e been
assigned Clus e numbe s. Do ed lines sepa a e he clus e s wi h elencephalon as ocy e signa u es.
(B) Do plo showing he op 10 di e en ially exp essed genes in each clus e . Colou scale indica es
no malized exp ession le el, scaled pe gene (z-sco e). Do size indica es pe cen age o cells, in each
clus e , exp essing he gene. (C) Ba plo showing he p opo ion o as ocy es om each expe imen al
g oup p esen in he di e en clus e s. (D) Ba plo showing op 10 signi ican ly en iched GO Biological
p ocess e ms o Clus e s a0 and a6, o de ed acco ding o -log10(adjus ed p- alue). S a is ical es s:
g:SCS mul iple es ing co ec ion me hod (Kolbe g e al., 2023) in (D) applying signi icance h eshold o
0.05.
Figu e 3. APOE iso o ms modula e as ocy e ansc ip ome in AppNL-G-F mouse b ain. (A) Upse plo
showing numbe o di e en ially exp essed genes (UP o DOWN) in elencephalon as ocy es om
each expe imen al g oup, compa ed o APOE3 g oup. Ba plo shows he numbe o di e en ially
exp essed genes (DEGs) common be ween di e en DE analyses (o e lapping se s indica ed by black
do s in column below he ba s). Ba s a e colo ed acco ding o s a is ical signi icance o in e sec ion o
DEGs. (B) Volcano plo showing di e en ially exp essed genes be ween APOEKO elencephalon
as ocy es and APOE3 exp essing elencephalon as ocy es. Da apoin s o signi ican genes a e
colou ed (Red o UP and blue o DOWN). Signi icance assigned based on | Log2(Fold Change) | > 0.2
and adjus ed p- alue <0.05. (C) Violin plo showing no malized gene exp ession o Pde10a in
elencephalon as ocy es om he expe imen al g oups. (D) Quad an plo compa ing di e en ial
exp ession o genes in elencephalon as ocy es in APOE2 s APOE3 (along x-axis) and in APOE4 s
APOE3 (along y-axis). Colou s in legend key indica e signi icance o genes up- o down egula edin
APOE2 o APOE4 o bo h. Signi icance o di e en ially exp essed genes based on | Log2(Fold Change)
| > 0.2 and adjus ed p- alue < 0.05). Pea son’s co ela ion, R= 0.5. S a is ical es s: MAST di e en ial
exp ession es in (A), (B) and (D), p- alues we e adjus ed wi h Bon e oni co ec ion based on he o al
numbe o genes in he da ase . Supe Exac Tes (Wang, Zhao and Zhang, 2015) in (A).

Figu e 4. Apoe-de icien mic oglia moun eac i e esponses o ib illa amyloid plaques. (A) IF
images o AAV ansduced co ical egions (see Fig. S2A), a 6 mon hs o age. mChe y ( ed) shows he
ansduced as ocy es. X-34 s aining (whi e) shows he ib illa plaque deposi s. Co-s aining wi h an i-
Iba1 an ibody shows mic oglial cells (g een). Scale ba : 100 µm. (B) IF images o AAV-APOE4 ansduced
co ical egions (see Fig. S2A) a 6 mon hs o age. X-34 s aining (whi e) shows he ib illa plaque
deposi s. Co-s aining wi h an i-Iba1 an ibody shows mic oglial cells (g een) and an i-Cd68 an ibody
shows phagocy ic s uc u es ( ed) inside mic oglia su ounding plaques. Scale ba : 100 µm. (C) IF
images o AAV-APOE4 ansduced co ical egions a 6 mon hs o age. X-34 s aining (whi e) shows he
ib illa plaque deposi s. Co-s aining wi h an i-Iba1 an ibody shows mic oglia (g een) and an i-C sD
an ibody shows lysosomal s uc u es ( ed) inside clus e ed mic oglia su ounding plaques. Scale ba :
100 µm. (D) (i) Ba plo showing o al numbe o mic oglia pe ield o iew (FOV, 40x magni ica ion)
in AAV ansduced egions. Da a poin s show mean alue o 3 ields o iew pe mouse (n= 6-7 mice
pe g oup). (ii) Sca e plo showing linea eg ession be ween a ea o X-34+ s aining (as ac ion o
o al a ea in FOV) and o al numbe o mic oglia (solid iangle), o numbe o plaque-associa ed
mic oglia (solid ci cle), o numbe o non-plaque-associa ed mic oglia (hollow ci cle), o each
expe imen al g oup. Da a poin s show mean alue 3 ields o iew pe mouse (n= 6-7 mice pe g oup).
Shape legends o di e en ca ego ies o mic oglia in (i) and (ii) a e indica ed. One-way ANOVA and
Tukey’s mul iple compa ison es in (Di). Signi icance shown o pai wise compa isons **p < 0.01; ***p
< 0.001, ****p < 0.0001
Figu e 5. As ocy e-de i ed APOE ini ia es, and mic oglia modi y, ib illa Aβ plaque pa hology. (A)
Expe imen al design illus a ing PLX3397 ea men o mic oglial deple ion in AppNL-G-F x Apoe-/- mice.
T ea men s a ed a 2 mon hs o age by mixing PLX3397 (600mg/kg) in chow and samples we e
collec ed o analysis a 6 mon hs o age. (B) IF images o AAV ansduced co ical egions (see Fig. S2A)
a 6 mon hs o age om con ol ( op ow) and PLX3397- ea ed (bo om ow) g oups. mChe y ( ed)
shows he ansduced as ocy es. X-34 s aining (whi e) shows he ib illa plaque deposi s in mouse
b ain co ex. Co-s aining wi h an i-Iba1 an ibody shows mic oglial cells (g een). Scale ba : 100 µm. (C)
Ba plo s showing numbe o X-34+ plaques pe ield o iew (FOV, 20x magni ica ion) and a ea o X-
34+ plaques (as ac ion o o al a ea in FOV, 20x magni ica ion), in he AAV- ansduced co ical egion
o (i) APOE2, (ii) APOE3 and (iii) APOE4 g oups and hei espec i e PLX3397 ea ed g oups. Da a
poin s show mean alue o 3 ields o iew pe mouse (n= 4-10 mice pe g oup). Da a poin s o con ol
non- ea ed g oups a e he same as shown in Fig. 1C (D) Sca e plo showing linea eg ession
be ween o al APOE le els (x-axis) and co esponding guanidine-soluble Aβ42 le els (y-axis), measu ed
by MSD-ELISA o PLX3397- ea ed and con ol mice om (i) APOE2, (ii) APOE3 and (iii) APOE4 g oups.
Da a poin s show mean alue o each mouse (n= 4-10 mice pe g oup. 2 echnical eplica es pe
mouse). Da a poin s o con ol non- ea ed g oups a e he same as shown in Fig. 1E. S a is ical es s:
Da a ep esen ed as mean ± SEM in (C). Unpai ed - es in (C). p- alues a e indica ed.
Supplemen a y Figu es
Figu e S1. AAV-media ed ansduc ion enables exp ession o APOE speci ically in as ocy es in i o.
(A) IF image o AAV ansduced mouse b ain co onal sec ions a 2 mon hs o age showing dis ibu ion
o ansduced as ocy es (mChe y in ed) in co ex and subco ical a eas (in AppNL-G-F x Rag2-/- x Apoe-
/- mice. Scale ba : 500 µm. (B) IF image o AAV ansduced co ical egion (see Fig. S2A) o 6 mon hs
old AppNL-G-F x Apoe-/- mouse b ain, showing mChe y ( ed) co-localizing wi h as ocy e ma ke Aldh1l1
(g een). Image below shows a zoomed (5x) iew o me ged image indica ed by he inse box o whi e
dashed-lines. Scale ba : 100 µm. (C) Con ou plo om low cy ome y analysis showing (i) densi y
dis ibu ion o mChe y+ and Acsa2+ as ocy es, spli o show (ii) p opo ion o mChe y+ cells in o al
Acsa2+ cells, and (iii) p opo ion o Acsa2+ as ocy es in o al mChe y+ cells, in 6 mon hs old AppNL-G-
F x Apoe-/- mouse b ain. Colou scale indica es coun densi y o each con ou bin. (D) IF images o AAV
ansduced co ical egion (see Fig. S2A) o 6 mon hs old AppNL-G-F x Apoe-/- mouse b ain. mChe y ( ed)
shows he ansduced cells. Co-s aining wi h an i-Iba1 an ibody showing mic oglia (cyan) (No e ha
his is he same image as S1B and co-s aining was done wi h Aldh1l1); wi h an i-Apc an ibody showing
oligodend ocy es cells (g een); wi h an i-NeuN an ibody showing neu ons (whi e) (No e ha Apc and
NeuN panels a e shown om he same image as hese ma ke s we e co-s ained in he same slide).
Images below show zoomed in (5x) iews indica ed by he inse box o whi e dashed-lines. Scale ba :
100 µm. (E) IF images o AAV ansduced co ical egion (see Fig. S2A) o 6 mon hs old AppNL-G-F x Rag2-
/- x Apoe-/- mouse b ain, showing APOE3 (g een) co-localizing wi h mChe y+ ( ed). Scale ba : 100 µm.
(F) Ba plo showing human APOE mRNA le els a 6 mon hs o age in mouse b ain homogena es
measu ed by semiquan i a i e eal- ime PCR. Da a poin s show mean alue o 3 echnical eplica es
pe mouse (n= 3 mice pe g oup). ns = non-signi ican . S a is ical es s: Da a p esen ed as mean ± SEM
in (F). One-way ANOVA and Tukey’s mul iple compa ison es (F). Signi icance shown o pai wise
compa isons: ****p < 0.0001
Figu e S2. As ocy e-de i ed APOE leads o ib illa Aβ plaque o ma ion in i o. (A) IF images
showing AAV ansduced ( ed) egions in he co ex and he associa ed dis ibu ion o X-34+ ib illa
plaque deposi s (whi e) in 6 mon hs old AppNL-G-F x Rag2-/- x Apoe-/- mice. Scale ba : 500 µm (B) IF images
showing dis ibu ion o X-34+ ib illa plaque deposi s (whi e) and Iba1+ mic oglia clus e ing (g een) in
co ical egion in 6 mon hs old AppNL-G-F x Rag2-/- mice ( op panel). Absence o ib illa plaque deposi s
and clus e ed mic oglia in co ical egion o 6 mon hs old AppNL-G-F x Rag2-/- x Apoe-/- mouse b ain
(bo om panel). Scale ba : 500 µm. (C) IF images o AAV ansduced co ical egion o 6 mon hs old
AppNL-G-F x Apoe-/- mouse b ain, showing APOE3 (g een) co-localizing wi h mChe y+ ( ed) as ocy es
and X-34+ (whi e) ib illa plaque deposi s. Scale ba : 100 µm.
Figu e S3. As ocy e-de i ed APOE modula es he size and compac ness o ib illa Aβ plaques. (A)
Composi e ile showing indi idual X-34+ ib illa plaques, wi h inc easing plaque size (measu ed in
diame e ) along he x-axis and inc easing compac ness (measu ed in Angula Second Momen (ASM)
alue) along he y-axis. Highe ASM alues co ela e wi h highe compac ness. (B) S acked ba plo
showing p opo ion o X-34+ ib illa plaques om APOE2, APOE3 and APOE4 g oups in di e en size
ca ego ies (diame e in µm) along he x-axis. Lowe limi o plaque size se o 5 µm in diame e . Size
ca ego ies used: small (5-20 µm diame e ), medium (20-40 µm diame e ) and la ge (abo e 40 µm
diame e ). (C) Box plo showing ASM alues as an es ima e o plaque compac ness in APOE2, APOE3
and APOE4 g oups. Size ca ego ies used: (i) small (5-20 µm diame e ), (ii) medium (20-40 µm diame e )
and (iii) la ge (abo e 40 µm diame e ). Da a poin s show alue o indi idual plaques (n= 8-10 mice pe
g oup. 3 FOV pe mouse). S a is ical es s: Da a p esen ed as median and in e qua ile ange ±  alues
wi hin 1.5 imes he in e qua ile ange (C). Linea mixed e ec s model wi h Tukey’s HSD es in (C)
done a mouse sample le el. Signi icance shown o pai wise compa isons **p < 0.01; ***p < 0.001,
****p < 0.0001
Figu e S4. APOE exp ession modula es as ocy e cell-s a es in AppNL-G-F mouse b ain. (A) UMAP plo
(same as Figu e 6a), showing exp ession o ma ke s o non- elencephalon (Ag , Slc6a11, Slac6a9) and
elencephalon (M ge8, Lhx2, Ppp1 3g) as ocy es. Colou scale indica es no malized gene exp ession.
(B) Violin plo showing no malized module sco e in each clus e , o ma ke gene se om p e iously
iden i ied eac i e as ocy e cell-s a es. JLZ_Pan- Pan- eac i e as ocy es, JLZ_A1- A1 as ocy es,
JLZ_A2- A2 as ocy es om (Zamanian e al., 2012), PH_Clus 4- Clus e 4, PH_Clus - Clus e 8 om
(Hasel e al., 2021). (C) Violin plo showing no malized gene exp ession in each clus e , o homeos a ic
as ocy e ma ke s (Slc1a2, Glul, Aqp4, Gja1) and eac i e as ocy e ma ke s (G ap, Vim, S100a6, Meg3)
(D) Do plo showing exp ession o p e iously iden i ied Plaque Induced Genes (PIGs) (Chen e al.,
2020) in each clus e . Colou scale indica es no malized exp ession le el, scaled pe gene (z-sco e). Do
size indica es pe cen age o cells, in each g oup, exp essing he gene. (E) Do plo showing exp ession
o p e iously iden i ied AD isk genes, spli by expe imen al g oups. Colou scale indica es no malized
exp ession le el, scaled pe gene (z-sco e). Do size indica es pe cen age o cells, in each g oup,
exp essing he gene. Squa es a ound do s ma k s a is ically signi ican genes (exp essed in mo e han
25% o cells in each g oup and wi h adjus ed p- alue < 0.05) based on di e en ial exp ession agains
APOE3 g oup. G een squa es indica e genes wi h | Log2(Fold Change) | > 0.2. Red squa es indica e
genes wi h | Log2(Fold Change) | < 0.2. S a is ical es s: MAST di e en ial exp ession es in (E), p-
alues we e adjus ed wi h Bon e oni co ec ion based on he o al numbe o genes in he da ase .
Figu e S5. Apoe-de icien mic oglia moun eac i e esponses o ib illa amyloid plaques. (A)
Rep esen a ion o h esholding ( i s s ep) and segmen a ion o mic oglia cell bodies, based on an i-
Iba1 immunos aining, using a combina ion o Cellpose (second s ep: op panel) and Labki (second
s ep: bo om panel) algo i hms. See Ma e ials and Me hods o de ails. (B) Rep esen a ion o
h esholding ( i s s ep) and acing (second s ep) o p ima y and seconda y p ocesses o segmen ed
mic oglial cell. Sum o he leng h o p ima y and seconda y p ocess calcula ed as “ o al p ocess leng h”.
A ea o bounding box (in whi e) calcula ed as “con ex a ea” o e i o y size. (C) Numbe o mic oglia
segmen ed and quan i ied om AAV ansduced co ical egions (40x magni ica ion), a 6 mon hs o
age, om he expe imen al g oups, u he classi ied as plaque-associa ed (wi hin X34+ h esholded
plaque a ea and up o 5 µm ou side a plaque’s edge) o non-plaque-associa ed. (n= 6-7 mice pe g oup.
3 ields o iew pe mouse). (D) Ba plo showing p opo ion o o al numbe o mic oglia om all he
expe imen al g oups (Non-plaque-associa ed mic oglia = 0.74; Plaque-associa ed mic oglia = 0.26). (E)
Box plo showing mean dis ance be ween each mic oglia o i s nea es h ee neighbou s, o mic oglial
cells classi ied as plaque-associa ed o non-plaque-associa ed. Da a poin s show indi idual mic oglia
om all expe imen al g oups. (F) Box plo s showing (i) e i o y size (con ex a ea) occupied by, and
(ii) o al p ocess leng h o , mic oglia classi ied as plaque-associa ed o non-plaque-associa ed. Da a
poin s show indi idual mic oglia om all expe imen al g oups. Tex u e legends o mic oglia
classi ica ion in (D), (E) and (F) a e indica ed. (G) IF images o AAV-APOE2 and AAV-APOE3 ansduced
co ical egions (see Fig. S2A) a 6 mon hs o age. X-34 s aining (whi e) shows he ib illa plaque
deposi s. Co-s aining wi h an i-Iba1 an ibody shows mic oglial cells (g een) and an i-Cd68 an ibody
shows phagocy ic s uc u es ( ed) inside mic oglia su ounding plaques. Scale ba : 100 µm. (H) IF
images o AAV-APOE2 and AAV-APOE3 ansduced co ical egions (see Fig. S2A) a 6 mon hs o age.
X-34 s aining (whi e) shows he ib illa plaque deposi s. Co-s aining wi h an i-Iba1 an ibody shows
mic oglia (g een) and an i-C sD an ibody shows lysosomal s uc u es ( ed) inside clus e ed mic oglia
su ounding plaques. Scale ba : 100 µm. (I) Box plo s showing (i) e i o y size (con ex a ea) occupied
by, and (ii) o al p ocess leng hs o , non-plaque-associa ed mic oglial cells compa ed be ween he
expe imen al g oups. Da a poin s show indi idual mic oglia om each expe imen al g oup. (J) Box
plo s showing (i) e i o y size (con ex a ea) occupied by, and (ii) o al p ocess leng hs o , plaque-
associa ed mic oglial cells compa ed be ween expe imen al g oups. Da a poin s show indi idual
mic oglia om each expe imen al g oup. Colou s legends o expe imen al g oups in (G) and (H) a e
indica ed. S a is ical es s: Da a ep esen ed as median and in e qua ile ange ±  alues wi hin 1.5
imes he in e qua ile ange in (D), (E), (F), (I) and (J). Linea mixed e ec s model in (C-E). Linea mixed
e ec s model wi h Tukey’s HSD es in (F), (I) and (J). Signi icance shown o pai wise compa isons
****p < 0.0001
Figu e S6. As ocy e-de i ed APOE in luences he exp ession o mic oglial cell s a e ma ke s. (A)
UMAP plo showing 18,569 APOE-de icien mic oglia (Cd11b+/Cd45-low) so ed om 6 mon hs old
mouse b ains om he ou expe imen al g oups (n= 2 o 3 mice pe g oup). Di e en sub-popula ions
iden i ied h ough unbiased clus e ing ha e been assigned clus e numbe s. (B) Do plo showing he
op 10 di e en ially exp essed genes in each clus e . Colou scale indica es no malized exp ession
le el, scaled pe gene (z-sco e). Do size indica es pe cen age o cells, in each clus e , exp essing he
gene. (C) Ba plo showing he p opo ion o mic oglia om each expe imen al g oup p esen in he
di e en clus e s.
Figu e S7. As ocy e-de i ed APOE modula es gene exp ession in mic oglia. (A) Upse plo showing
numbe o di e en ially exp essed genes (UP o DOWN) in mic oglia om each expe imen al g oup,
compa ed o APOE3 g oup. Ba plo shows he numbe o di e en ially exp essed genes (DEGs)
common be ween di e en DE analyses (o e lapping se s indica ed by black do s in column below he
ba s). Ba s a e colo ed acco ding o s a is ical signi icance o in e sec ion o DEGs. (B) Volcano plo
showing di e en ially exp essed genes be ween APOEKO elencephalon as ocy es and APOE3
exp essing elencephalon as ocy es. Da apoin s o signi ican genes a e colou ed (Red o UP and
blue o DOWN). Signi icance assigned based on | Log2(Fold Change) | > 0.2 and adjus ed p- alue
<0.05. Genes discussed in ex a e indica ed in ed. (C) Do plo showing exp ession o he 52 genes
(Figu e 2A) commonly up egula ed in APOEKO and APOE4 mic oglia, spli by expe imen al g oups.
Colou scale indica es no malized exp ession le el, scaled pe gene (z-sco e). Do size indica es
pe cen age o cells, in each g oup, exp essing he gene. Plaque-induced genes (PIGs) (Chen e al., 2020)
a e colou ed in ed on y-axis. (D) Quad an plo compa ing di e en ial exp ession o genes in mic oglia
in APOE2 s APOE3 (along x-axis) and in APOE4 s APOE3 (along y-axis). Colou s in legend key indica e
s a is ical signi icance o genes up- o down egula ed inAPOE2 o APOE4 o bo h. . Signi icance o
di e en ially exp essed genes based on | Log2(Fold Change) | > 0.2 and adjus ed p- alue < 0.05).
Pea son’s co ela ion, R= 0.52. S a is ical es s: MAST di e en ial exp ession es in (B), (C) and (D), p-
alues we e adjus ed wi h Bon e oni co ec ion based on he o al numbe o genes in he da ase .
Figu e S8. APOE le els and mic oglia deple ion. (A) Box plo s showing a ea o Iba1+ cells (as ac ion
o o al a ea in FOV) in AAV ansduced co ical egions a 6 mon hs o age, om con ol and PLX3397-
ea ed (shaded) g oups in (i) APOE2, (ii) APOE3, (iii) APOE4. Pe cen age educ ions a e: APOE2: 83%,
APOE3: 77%, APOE4: 87%. Da a poin s show mean alue o 3 FOV pe mouse (n= 4-10 mice pe g oup).
(B) Ba plo s showing o al APOE le els in b ain homogena e om con ol and PLX3397- ea ed
(shaded) g oups in (i) APOE2, (ii) APOE3, (iii) APOE4, measu ed by MSD-ELISA. Da a poin s show mean
alue o 2 echnical eplica es pe mouse (n= 4-10 mice pe g oup). S a is ical es s: Da a ep esen ed
as median and in e qua ile ange ±  alues wi hin 1.5 imes he in e qua ile ange in (A); mean ± SEM
in (B). Non-pa ame ic Wilcoxon es in (a). Unpai ed - es in (b). *p < 0.05; **p < 0.01; ***p < 0.001
Figu e S9: Quali y con ol o mic oglia and as ocy e single cell lib a ies. (A) UMAP plo showing
23,495 APOE-de icien mic oglia (Cd11b+/Cd45-low) so ed om 6 mon hs old mouse b ains om he
ou expe imen al g oups (n= 2 o 3 mice pe g oup). (B) Do plo showing he op 10 di e en ially
exp essed genes in each QC clus e o mic oglia lib a y. Colou scale indica es no malized exp ession
le el, scaled by gene (z-sco e). Do size indica es pe cen age o cells, in each clus e , exp essing he
gene. Mic oglia QC Clus e s 0, 1, 2, 3 and 8 ha e mic oglial iden i y. Con amina ing cell ypes iden i ied
a e mac ophages (Mic oglia QC Clus e 4), endo helial cells (Mic oglia QC Clus e 5), mix o as ocy e
and mic oglial cells (Mic oglia QC Clus e 6), monocy es (Mic oglia QC Clus e s 7 and 9) and
oligodend ocy es (QC Clus e 10). (C) Violin plo showing (i) nUMI ( o al coun o RNA ansc ip s
cap u ed), (ii) nGene ( o al numbe o genes), (iii) mi o (pe cen age o mi ochond ial genes), (i ) ibo
(pe cen age o ibosomal genes), in each mic oglial QC clus e . (D) UMAP plo showing 32,982
as ocy es (mChe y+/Acsa2+) so ed om 6 mon hs old mouse b ains om he ou expe imen al
g oups (n= 2 o 3 mice pe g oup). (E) Do plo showing he op 10 di e en ially exp essed genes in
each QC clus e o as ocy e lib a y. Colou scale indica es no malized exp ession le el, scaled by gene
(z-sco e). Do size indica es pe cen age o cells, in each clus e , exp essing he gene. As ocy e QC
Clus e s 0, 1, 2, 3, 4 and 6 ha e as ocy e iden i y. Con amina ing cell ypes iden i ied a e ependymal
cells (As ocy e QC Clus e 5), mu al cells (As ocy e QC Clus e 7), mic oglia (As ocy e QC Clus e 9),
endo helial cells (As ocy e QC Clus e 10) and a popula ion o unde e mined iden i y (As ocy e QC
Clus e 8). (F) Violin plo showing (i) nUMI ( o al coun o RNA ansc ip s cap u ed), (ii) nGene ( o al
numbe o genes), (iii) mi o (pe cen age o mi ochond ial genes), (i ) ibo (pe cen age o ibosomal
genes), in each as ocy e QC clus e .
Table S1. Lis o ma ke s iden i ied o as ocy e clus e s om single cell ansc ip omic analysis.
Table S2. Lis o GO:Biological P ocess en iched e ms o as ocy e clus e s om unc ional
en ichmen analysis.
Table S3. Lis o di e en ially exp essed genes (DEGs) (agains APOE3) in as ocy es om each
expe imen al g oup, wi h de ails o s a is ical es o o e lap o DEGs be ween di e en g oups.
Table S4. Lis o ma ke s iden i ied o mic oglia clus e s om single cell ansc ip omic analysis.
Ma sh, S.E. e al. (2022) ‘Dissec ion o a i ac ual and con ounding glial signa u es by single-cell sequencing o
mouse and human b ain’, Na u e Neu oscience, 25(3), pp. 306–316. A ailable a :
h ps://doi.o g/10.1038/s41593-022-01022-8.
Ma ins, C. a. R. e al. (2005) ‘APOE alleles p edic he a e o cogni i e decline in Alzheime disease: A
nonlinea model’, Neu ology, 65(12), pp. 1888–1893. A ailable a :
h ps://doi.o g/10.1212/01.wnl.0000188871.74093.12.
McCa hy, D.J. e al. (2017) ‘Sca e : p e-p ocessing, quali y con ol, no maliza ion and isualiza ion o single-cell
RNA-seq da a in R’, Bioin o ma ics, 33(8), pp. 1179–1186. A ailable a :
h ps://doi.o g/10.1093/bioin o ma ics/b w777.
Mi, W. e al. (2007) ‘Cys a in C inhibi s amyloid-be a deposi ion in Alzheime ’s disease mouse models’, Na u e
Gene ics, 39(12), pp. 1440–1442. A ailable a : h ps://doi.o g/10.1038/ng.2007.29.
Michalo icz, L.T. e al. (2019) ‘As ocy e-speci ic ansc ip ome analysis using he ALDH1L1 bacTRAP mouse
e eals no el bioma ke s o as ogliosis in esponse o neu o oxici y’, Jou nal o Neu ochemis y,
150(4), pp. 420–440. A ailable a : h ps://doi.o g/10.1111/jnc.14800.
Miz ak, D. e al. (2019) ‘Single-Cell Analysis o Regional Di e ences in Adul V-SVZ Neu al S em Cell Lineages’,
Cell Repo s, 26(2), pp. 394-406.e5. A ailable a : h ps://doi.o g/10.1016/j.cel ep.2018.12.044.
Mon agne, A. e al. (2020) ‘APOE4 leads o blood–b ain ba ie dys unc ion p edic ing cogni i e decline’,
Na u e, 581(7806), pp. 71–76. A ailable a : h ps://doi.o g/10.1038/s41586-020-2247-3.
Mon agne, A. e al. (2021) ‘APOE4 accele a es ad anced-s age ascula and neu odegene a i e diso de in old
Alzheime ’s mice ia cyclophilin A independen ly o amyloid-β’, Na u e Aging, 1(6), pp. 506–520.
A ailable a : h ps://doi.o g/10.1038/s43587-021-00073-z.
Mu ai, K.K. and Pasquale, E.B. (2011) ‘Eph ecep o s and eph ins in neu on–as ocy e communica ion a
synapses’, Glia, 59(11), pp. 1567–1578. A ailable a : h ps://doi.o g/10.1002/glia.21226.
Mu phy, K.R. e al. (2013) ‘Mapping he e ec s o ApoE4, age and cogni i e s a us on 18F- lo be api PET
measu ed egional co ical pa e ns o be a-amyloid densi y and g ow h’, Neu oImage, 78, pp. 474–
480. A ailable a : h ps://doi.o g/10.1016/j.neu oimage.2013.04.048.
Nagy, Z.S. e al. (1995) ‘In luence o he apolipop o ein E geno ype on amyloid deposi ion and neu o ib illa y
angle o ma ion in Alzheime ’s disease’, Neu oscience, 69(3), pp. 757–761. A ailable a :
h ps://doi.o g/10.1016/0306-4522(95)00331-C.
Nasla sky, M.S. e al. (2022) ‘Global and local ances y modula e APOE associa ion wi h Alzheime ’s
neu opa hology and cogni i e ou comes in an admixed sample’, Molecula Psychia y, 27(11), pp.
4800–4808. A ailable a : h ps://doi.o g/10.1038/s41380-022-01729-x.
Nu iel, T. e al. (2017) ‘The Endosomal-Lysosomal Pa hway Is Dys egula ed by APOE4 Exp ession in Vi o’,
F on ie s in Neu oscience, 11, p. 702. A ailable a : h ps://doi.o g/10.3389/ nins.2017.00702.
O’Conno , V. e al. (2004) ‘Di e en ial Ampli ica ion o In on-con aining T ansc ip s Re eals Long Te m
Po en ia ion-associa ed Up- egula ion o Speci ic Pde10A Phosphodies e ase Splice Va ian s *’,
Jou nal o Biological Chemis y, 279(16), pp. 15841–15849. A ailable a :
h ps://doi.o g/10.1074/jbc.M312500200.
Olli ie , M. e al. (2024) ‘C ym-posi i e s ia al as ocy es ga e pe se e a i e beha iou ’, Na u e, 627(8003), pp.
358–366. A ailable a : h ps://doi.o g/10.1038/s41586-024-07138-0.
Pa con, P.A. e al. (2018) ‘Apolipop o ein E4 inhibi s au ophagy gene p oduc s h ough di ec , speci ic binding
o CLEAR mo i s’, Alzheime ’s & Demen ia: The Jou nal o he Alzheime ’s Associa ion, 14(2), pp. 230–
242. A ailable a : h ps://doi.o g/10.1016/j.jalz.2017.07.754.
Peng, H. e al. (2010) ‘V3D enables eal- ime 3D isualiza ion and quan i a i e analysis o la ge-scale biological
image da a se s’, Na u e Bio echnology, 28(4), pp. 348–353. A ailable a :
h ps://doi.o g/10.1038/nb .1612.
Pe occhi, F. e al. (2007) ‘An isense a i ac s in ansc ip ome mic oa ay expe imen s a e esol ed by
ac inomycin D’, Nucleic Acids Resea ch, 35(19), p. e128. A ailable a :
h ps://doi.o g/10.1093/na /gkm683.
Pohlkamp, T. e al. (2021) ‘NHE6 deple ion co ec s ApoE4-media ed synap ic impai men s and educes
amyloid plaque load’, eLi e, 10, p. e72034. A ailable a : h ps://doi.o g/10.7554/eLi e.72034.
Pol ikoski, T. e al. (1995) ‘Apolipop o ein E, demen ia, and co ical deposi ion o be a-amyloid p o ein’, The
New England Jou nal o Medicine, 333(19), pp. 1242–1247. A ailable a :
h ps://doi.o g/10.1056/NEJM199511093331902.
Powell, D.S. e al. (2021) ‘The Rela ionship o APOE ε4, Race, and Sex on he Age o Onse and Risk o
Demen ia’, F on ie s in Neu ology, 12. A ailable a :
h ps://www. on ie sin.o g/a icles/10.3389/ neu .2021.735036 (Accessed: 29 Augus 2023).

P asad, H. and Rao, R. (2018) ‘Amyloid clea ance de ec in ApoE4 as ocy es is e e sed by epigene ic
co ec ion o endosomal pH’, P oceedings o he Na ional Academy o Sciences o he Uni ed S a es o
Ame ica, 115(28), pp. E6640–E6649. A ailable a : h ps://doi.o g/10.1073/pnas.1801612115.
Puzzo, D. e al. (2009) ‘Phosphodies e ase 5 inhibi ion imp o es synap ic unc ion, memo y, and amyloid-be a
load in an Alzheime ’s disease mouse model’, The Jou nal o Neu oscience: The O icial Jou nal o he
Socie y o Neu oscience, 29(25), pp. 8075–8086. A ailable a :
h ps://doi.o g/10.1523/JNEUROSCI.0864-09.2009.
Quinlan, A.R. and Hall, I.M. (2010) ‘BEDTools: a lexible sui e o u ili ies o compa ing genomic ea u es’,
Bioin o ma ics, 26(6), pp. 841–842. A ailable a : h ps://doi.o g/10.1093/bioin o ma ics/b q033.
Rauch, J.N. e al. (2020) ‘LRP1 is a mas e egula o o au up ake and sp ead’, Na u e, 580(7803), pp. 381–385.
A ailable a : h ps://doi.o g/10.1038/s41586-020-2156-5.
Reiman, E.M. e al. (2020) ‘Excep ionally low likelihood o Alzheime ’s demen ia in APOE2 homozygo es om a
5,000-pe son neu opa hological s udy’, Na u e Communica ions, 11(1), p. 667. A ailable a :
h ps://doi.o g/10.1038/s41467-019-14279-8.
Riddell, D.R. e al. (2008) ‘Impac o Apolipop o ein E (ApoE) Polymo phism on B ain ApoE Le els’, The Jou nal
o Neu oscience, 28(45), pp. 11445–11453. A ailable a : h ps://doi.o g/10.1523/JNEUROSCI.1972-
08.2008.
Rod iguez, G.A. e al. (2014) ‘Human APOE4 inc eases mic oglia eac i i y a Aβ plaques in a mouse model o
Aβ deposi ion’, Jou nal o Neu oin lamma ion, 11, p. 111. A ailable a : h ps://doi.o g/10.1186/1742-
2094-11-111.
Sai o, T. e al. (2014) ‘Single App knock-in mouse models o Alzheime ’s disease’, Na u e Neu oscience, 17(5),
pp. 661–663. A ailable a : h ps://doi.o g/10.1038/nn.3697.
Sala F ige io, C. e al. (2019) ‘The Majo Risk Fac o s o Alzheime ’s Disease: Age, Sex, and Genes Modula e he
Mic oglia Response o Aβ Plaques’, Cell Repo s, 27(4), pp. 1293-1306.e6. A ailable a :
h ps://doi.o g/10.1016/j.cel ep.2019.03.099.
Sando, S.B. e al. (2008) ‘APOE ε4 lowe s age a onse and is a high isk ac o o Alzheime ’s disease; A case
con ol s udy om cen al No way’, BMC Neu ology, 8(1), p. 9. A ailable a :
h ps://doi.o g/10.1186/1471-2377-8-9.
Sas e, M. e al. (2004) ‘Binding o cys a in C o Alzheime ’s amyloid be a inhibi s in i o amyloid ib il
o ma ion’, Neu obiology o Aging, 25(8), pp. 1033–1043. A ailable a :
h ps://doi.o g/10.1016/j.neu obiolaging.2003.11.006.
Schindelin, J. e al. (2012) ‘Fiji: an open-sou ce pla o m o biological-image analysis’, Na u e Me hods, 9(7),
pp. 676–682. A ailable a : h ps://doi.o g/10.1038/nme h.2019.
Schmechel, D.E. e al. (1993) ‘Inc eased amyloid be a-pep ide deposi ion in ce eb al co ex as a consequence o
apolipop o ein E geno ype in la e-onse Alzheime disease.’, P oceedings o he Na ional Academy o
Sciences, 90(20), pp. 9649–9653. A ailable a : h ps://doi.o g/10.1073/pnas.90.20.9649.
Se ano-Pozo, A. e al. (2015) ‘APOEε2 is associa ed wi h milde clinical and pa hological Alzheime disease’,
Annals o Neu ology, 77(6), pp. 917–929. A ailable a : h ps://doi.o g/10.1002/ana.24369.
Se ano-Pozo, A. e al. (2021) ‘E ec o APOE alleles on he glial ansc ip ome in no mal aging and Alzheime ’s
disease’, Na u e aging, 1(10), pp. 919–931. A ailable a : h ps://doi.o g/10.1038/s43587-021-00123-6.
Shi, Y. e al. (2017) ‘ApoE4 ma kedly exace ba es au-media ed neu odegene a ion in a mouse model o
auopa hy’, Na u e, 549(7673), pp. 523–527. A ailable a : h ps://doi.o g/10.1038/na u e24016.
Shi, Y. e al. (2021) ‘O e exp essing low-densi y lipop o ein ecep o educes au-associa ed
neu odegene a ion in ela ion o apoE-linked mechanisms’, Neu on, 109(15), pp. 2413-2426.e7.
A ailable a : h ps://doi.o g/10.1016/j.neu on.2021.05.034.
Side is, D.I. e al. (2021) ‘Soluble amyloid be a-con aining agg ega es a e p esen h oughou he b ain a ea ly
s ages o Alzheime ’s disease’, B ain Communica ions, 3(3), p. cab147. A ailable a :
h ps://doi.o g/10.1093/b aincomms/ cab147.
Simono i ch, S. e al. (2016) ‘Impai ed Au ophagy in APOE4 As ocy es’, Jou nal o Alzheime ’s Disease, 51(3),
pp. 915–927. A ailable a : h ps://doi.o g/10.3233/JAD-151101.
Simono i ch, S. e al. (2019) ‘The E ec s o APOE4 on Mi ochond ial Dynamics and P o eins in i o’, Jou nal o
Alzheime ’s disease: JAD, 70(3), pp. 861–875. A ailable a : h ps://doi.o g/10.3233/JAD-190074.
Sousa, C. e al. (2018) ‘Single‐cell ansc ip omics e eals dis inc in lamma ion‐induced mic oglia signa u es’,
EMBO epo s, 19(11), p. e46171. A ailable a : h ps://doi.o g/10.15252/emb .201846171.
Spangenbe g, E.E. e al. (2016) ‘Elimina ing mic oglia in Alzheime ’s mice p e en s neu onal loss wi hou
modula ing amyloid-β pa hology’, B ain: A Jou nal o Neu ology, 139(P 4), pp. 1265–1281. A ailable
a : h ps://doi.o g/10.1093/b ain/aww016.
S ephen, T.L. e al. (2019) ‘APOE geno ype and sex a ec mic oglial in e ac ions wi h plaques in Alzheime ’s
disease mice’, Ac a Neu opa hologica Communica ions, 7(1), p. 82. A ailable a :
h ps://doi.o g/10.1186/s40478-019-0729-z.
S inge , C. e al. (2021) ‘Cellpose: a gene alis algo i hm o cellula segmen a ion’, Na u e Me hods, 18(1), pp.
100–106. A ailable a : h ps://doi.o g/10.1038/s41592-020-01018-x.
S i ma e , W.J. e al. (1993) ‘Apolipop o ein E: high-a idi y binding o be a-amyloid and inc eased equency
o ype 4 allele in la e-onse amilial Alzheime disease.’, P oceedings o he Na ional Academy o
Sciences, 90(5), pp. 1977–1981. A ailable a : h ps://doi.o g/10.1073/pnas.90.5.1977.
Sun, J.-H. e al. (2015) ‘Gene ics o Vascula Demen ia: Sys ema ic Re iew and Me a-Analysis’, Jou nal o
Alzheime ’s disease: JAD, 46(3), pp. 611–629. A ailable a : h ps://doi.o g/10.3233/JAD-143102.
Susan o, T.A.K. e al. (2015) ‘Cogni ion, B ain A ophy, and Ce eb ospinal Fluid Bioma ke s Changes om
P eclinical o Demen ia S age o Alzheime ’s Disease and he In luence o Apolipop o ein E’, Jou nal o
Alzheime ’s Disease, 45(1), pp. 253–268. A ailable a : h ps://doi.o g/10.3233/JAD-142451.
Suzuki, K. e al. (2021) ‘Dele ion o B-cell ansloca ion gene 2 (BTG2) al e s he esponses o glial cells in whi e
ma e o ch onic ce eb al hypope usion’, Jou nal o Neu oin lamma ion, 18(1), p. 86. A ailable a :
h ps://doi.o g/10.1186/s12974-021-02135-w.
Talbo , C. e al. (1994) ‘P o ec ion agains Alzheime ’s disease wi h apoE ∈2’, The Lance , 343(8910), pp. 1432–
1433. A ailable a : h ps://doi.o g/10.1016/S0140-6736(94)92557-7.
Tcw, J. e al. (2022) ‘Choles e ol and ma isome pa hways dys egula ed in as ocy es and mic oglia’, Cell,
185(13), pp. 2213-2233.e25. A ailable a : h ps://doi.o g/10.1016/j.cell.2022.05.017.
Thepa ambil, S.M. e al. (2020) ‘As ocy es egula e b ain ex acellula pH ia a neu onal ac i i y-dependen
bica bona e shu le’, Na u e Communica ions, 11(1), p. 5073. A ailable a :
h ps://doi.o g/10.1038/s41467-020-18756-3.
Thomas, R. e al. (2016) ‘Epide mal g ow h ac o p e en s APOE4 and amyloid-be a-induced cogni i e and
ce eb o ascula de ici s in emale mice’, Ac a Neu opa hologica Communica ions, 4(1), p. 111.
A ailable a : h ps://doi.o g/10.1186/s40478-016-0387-3.
T opea, T.F. e al. (2018) ‘APOE, hough diso de , and SPARE-AD p edic cogni i e decline in es ablished
Pa kinson’s disease’, Mo emen Diso de s: O icial Jou nal o he Mo emen Diso de Socie y, 33(2),
pp. 289–297. A ailable a : h ps://doi.o g/10.1002/mds.27204.
Tsuang, D. e al. (2013) ‘APOE ε4 inc eases isk o demen ia in pu e synucleinopa hies’, JAMA neu ology, 70(2),
pp. 223–228. A ailable a : h ps://doi.o g/10.1001/jamaneu ol.2013.600.
Ul ich, J.D. e al. (2018) ‘ApoE acili a es he mic oglial esponse o amyloid plaque pa hology’, Jou nal o
Expe imen al Medicine, 215(4), pp. 1047–1058. A ailable a : h ps://doi.o g/10.1084/jem.20171265.
Van De Wal , S. e al. (2014) ‘sciki -image: image p ocessing in Py hon’, Pee J, 2, p. e453. A ailable a :
h ps://doi.o g/10.7717/pee j.453.
Ve ghese, P.B. e al. (2013) ‘ApoE in luences amyloid-β (Aβ) clea ance despi e minimal apoE/Aβ associa ion in
physiological condi ions’, P oceedings o he Na ional Academy o Sciences, 110(19), pp. E1807–E1816.
A ailable a : h ps://doi.o g/10.1073/pnas.1220484110.
Vi ek, M.P., B own, C.M. and Col on, C.A. (2009) ‘APOE geno ype-speci ic di e ences in he inna e immune
esponse’, Neu obiology o Aging, 30(9), pp. 1350–1360. A ailable a :
h ps://doi.o g/10.1016/j.neu obiolaging.2007.11.014.
Wah le, S.E. e al. (2005) ‘Dele ion o Abca1 inc eases Abe a deposi ion in he PDAPP ansgenic mouse model
o Alzheime disease’, The Jou nal o Biological Chemis y, 280(52), pp. 43236–43242. A ailable a :
h ps://doi.o g/10.1074/jbc.M508780200.
Wang, C. e al. (2021) ‘Selec i e emo al o as ocy ic APOE4 s ongly p o ec s agains au-media ed
neu odegene a ion and dec eases synap ic phagocy osis by mic oglia’, Neu on, 109(10), pp. 1657-
1674.e7. A ailable a : h ps://doi.o g/10.1016/j.neu on.2021.03.024.
Wang, M., Zhao, Y. and Zhang, B. (2015) ‘E icien Tes and Visualiza ion o Mul i-Se In e sec ions’, Scien i ic
Repo s, 5(1), p. 16923. A ailable a : h ps://doi.o g/10.1038/s ep16923.
Wegiel, J. e al. (2000) ‘Mic oglia cells a e he d i ing o ce in ib illa plaque o ma ion, whe eas as ocy es a e
a leading ac o in plague deg ada ion’, Ac a Neu opa hologica, 100(4), pp. 356–364. A ailable a :
h ps://doi.o g/10.1007/s004010000199.
Wennbe g, A.M. e al. (2018) ‘Associa ion o Apolipop o ein E ε4 Wi h T ansac i e Response DNA-Binding
P o ein 43’, JAMA neu ology, 75(11), pp. 1347–1354. A ailable a :
h ps://doi.o g/10.1001/jamaneu ol.2018.3139.
Wes , H.L., William Rebeck, G. and Hyman, B.T. (1994) ‘F equency o he apolipop o ein E ε2 allele is
diminished in spo adic Alzheime disease’, Neu oscience Le e s, 175(1), pp. 46–48. A ailable a :
h ps://doi.o g/10.1016/0304-3940(94)91074-X.
Wisniewski, T. e al. (1994) ‘Accele a ion o Alzheime ’s ib il o ma ion by apolipop o ein E in i o’, The
Ame ican Jou nal o Pa hology, 145(5), pp. 1030–1035.
Xia, Z. e al. (2024) ‘Co-agg ega ion wi h Apolipop o ein E modula es he unc ion o Amyloid-β in Alzheime ’s
disease’, Na u e Communica ions, 15, p. 4695. A ailable a : h ps://doi.o g/10.1038/s41467-024-
49028-z.
Xian, X. e al. (2018) ‘Re e sal o ApoE4-induced ecycling block as a no el p e en ion app oach o Alzheime ’s
disease’, eLi e, 7, p. e40048. A ailable a : h ps://doi.o g/10.7554/eLi e.40048.
Xiao, H. and Peng, H. (2013) ‘APP2: au oma ic acing o 3D neu on mo phology based on hie a chical p uning
o a g ay-weigh ed image dis ance- ee’, Bioin o ma ics, 29(11), pp. 1448–1454. A ailable a :
h ps://doi.o g/10.1093/bioin o ma ics/b 170.
Xiong, M. e al. (2021) ‘APOE immuno he apy educes ce eb al amyloid angiopa hy and amyloid plaques while
imp o ing ce eb o ascula unc ion’, Science ansla ional medicine, 13(581), p. eabd7522. A ailable
a : h ps://doi.o g/10.1126/sci anslmed.abd7522.
Xiong, M. e al. (2023) ‘As ocy ic APOE4 emo al con e s ce eb o ascula p o ec ion despi e inc eased
ce eb al amyloid angiopa hy’, Molecula Neu odegene a ion, 18(1), p. 17. A ailable a :
h ps://doi.o g/10.1186/s13024-023-00610-x.
Xu, Q. e al. (2006) ‘P o ile and egula ion o apolipop o ein E (ApoE) exp ession in he CNS in mice wi h
a ge ing o g een luo escen p o ein gene o he ApoE locus’, The Jou nal o Neu oscience: The
O icial Jou nal o he Socie y o Neu oscience, 26(19), pp. 4985–4994. A ailable a :
h ps://doi.o g/10.1523/JNEUROSCI.5476-05.2006.
Yang, H.-S. e al. (2018) ‘E alua ion o TDP-43 p o einopa hy and hippocampal scle osis in ela ion o APOE ε4
haplo ype s a us: a communi y-based coho s udy’, The Lance . Neu ology, 17(9), pp. 773–781.
A ailable a : h ps://doi.o g/10.1016/S1474-4422(18)30251-5.
Ye, P. and D’E cole, J. (1998) ‘Insulin-Like G ow h Fac o I (IGF-I) Regula es IGF Binding P o ein-5 Gene
Exp ession in he B ain*’, Endoc inology, 139(1), pp. 65–71. A ailable a :
h ps://doi.o g/10.1210/endo.139.1.5676.
Yin, Z., Rosenzweig, N., e al. (2023) ‘APOE4 impai s he mic oglial esponse in Alzheime ’s disease by inducing
TGFβ-media ed checkpoin s’, Na u e Immunology, 24(11), pp. 1839–1853. A ailable a :
h ps://doi.o g/10.1038/s41590-023-01627-6.
Yin, Z., He on, S., e al. (2023) ‘Iden i ica ion o a p o ec i e mic oglial s a e media ed by miR-155 and
in e e on-γ signaling in a mouse model o Alzheime ’s disease’, Na u e Neu oscience, 26(7), pp. 1196–
1207. A ailable a : h ps://doi.o g/10.1038/s41593-023-01355-y.
Youmans, K.L. e al. (2012) ‘APOE4-speci ic Changes in Aβ Accumula ion in a New T ansgenic Mouse Model o
Alzheime Disease*’, Jou nal o Biological Chemis y, 287(50), pp. 41774–41786. A ailable a :
h ps://doi.o g/10.1074/jbc.M112.407957.
Yunia i, L. e al. (2019) ‘Tumo supp esso s BTG1 and BTG2: Beyond g ow h con ol’, Jou nal o Cellula
Physiology, 234(5), pp. 5379–5389. A ailable a : h ps://doi.o g/10.1002/jcp.27407.
Zamanian, J.L. e al. (2012) ‘Genomic analysis o eac i e as ogliosis’, The Jou nal o Neu oscience: The O icial
Jou nal o he Socie y o Neu oscience, 32(18), pp. 6391–6410. A ailable a :
h ps://doi.o g/10.1523/JNEUROSCI.6221-11.2012.
Zappia, L. and Oshlack, A. (2018) ‘Clus e ing ees: a isualiza ion o e alua ing clus e ings a mul iple
esolu ions’, GigaScience, 7(7), p. giy083. A ailable a : h ps://doi.o g/10.1093/gigascience/giy083.
Zeisel, A. e al. (2018) ‘Molecula A chi ec u e o he Mouse Ne ous Sys em’, Cell, 174(4), pp. 999-1014.e22.
A ailable a : h ps://doi.o g/10.1016/j.cell.2018.06.021.
Zenz, R. e al. (2008) ‘Ac i a o p o ein 1 (Fos/Jun) unc ions in in lamma o y bone and skin disease’, A h i is
Resea ch & The apy, 10(1), p. 201. A ailable a : h ps://doi.o g/10.1186/a 2338.
Zhang, Q. e al. (2023) ‘NDRG2 egula es he o ma ion o eac i e as ocy e-de i ed p ogeni o cells ia No ch
signaling pa hway a e b ain auma ic inju y in a s’, F on ie s in Molecula Neu oscience, 16, p.
1149683. A ailable a : h ps://doi.o g/10.3389/ nmol.2023.1149683.
Zhang, Y. e al. (2014) ‘An RNA-sequencing ansc ip ome and splicing da abase o glia, neu ons, and ascula
cells o he ce eb al co ex’, The Jou nal o Neu oscience: The O icial Jou nal o he Socie y o
Neu oscience, 34(36), pp. 11929–11947. A ailable a : h ps://doi.o g/10.1523/JNEUROSCI.1860-
14.2014.
Zhao, L. e al. (2016) ‘In ace eb al adeno-associa ed i us gene deli e y o apolipop o ein E2 ma kedly educes
b ain amyloid pa hology in Alzheime ’s disease mouse models’, Neu obiology o Aging, 44, pp. 159–
172. A ailable a : h ps://doi.o g/10.1016/j.neu obiolaging.2016.04.020.
Zheng, G.X.Y. e al. (2017) ‘Massi ely pa allel digi al ansc ip ional p o iling o single cells’, Na u e
Communica ions, 8(1), p. 14049. A ailable a : h ps://doi.o g/10.1038/ncomms14049.
Zwei el, S. e al. (2018) ‘HOPX De ines He e ogenei y o Pos na al Sub en icula Zone Neu al S em Cells’, S em
Cell Repo s, 11(3), pp. 770–783. A ailable a : h ps://doi.o g/10.1016/j.s emc .2018.08.006.