Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

SCIENCE TRANSLATIONAL MEDICINE | RESEARCH ARTICLE

ALZHEIMER’S DISEASE Copyright © 2022


The Authors, some
Reversal of synapse loss in Alzheimer mouse models by rights reserved;
exclusive licensee
targeting mGluR5 to prevent synaptic tagging by C1Q American Association
for the Advancement
of Science. No claim
Joshua Spurrier1†, LaShae Nicholson1†, Xiaotian T. Fang2†, Austin J. Stoner1†, Takuya Toyonaga2, to original U.S.
Daniel Holden2, Timothy R. Siegert3, William Laird1, Mary Alice Allnutt1, Marius Chiasseu1, Government Works
A. Harrison Brody1, Hideyuki Takahashi1, Sarah Helena Nies1,4, Azucena Pérez-Cañamás1,
Pragalath Sadasivam2, Supum Lee2, Songye Li2, Le Zhang1, Yiyun H. Huang2, Richard E. Carson2,
Zhengxin Cai2*, Stephen M. Strittmatter1*

Microglia-mediated synaptic loss contributes to the development of cognitive impairments in Alzheimer’s disease
(AD). However, the basis for this immune-mediated attack on synapses remains to be elucidated. Treatment with
the metabotropic glutamate receptor 5 (mGluR5) silent allosteric modulator (SAM), BMS-984923, prevents
-amyloid oligomer–induced aberrant synaptic signaling while preserving physiological glutamate response.
Here, we show that oral BMS-984923 effectively occupies brain mGluR5 sites visualized by [18F]FPEB positron
emission tomography (PET) at doses shown to be safe in rodents and nonhuman primates. In aged mouse models
of AD (APPswe/PS1E9 overexpressing transgenic and AppNL-G-F/hMapt double knock-in), SAM treatment fully
restored synaptic density as measured by [18F]SynVesT-1 PET for SV2A and by histology, and the therapeutic
benefit persisted after drug washout. Phospho-TAU accumulation in double knock-in mice was also reduced by
SAM treatment. Single-nuclei transcriptomics demonstrated that SAM treatment in both models normalized expression
patterns to a far greater extent in neurons than glia. Last, treatment prevented synaptic localization of the
complement component C1Q and synaptic engulfment in AD mice. Thus, selective modulation of mGluR5 reversed
neuronal gene expression changes to protect synapses from damage by microglial mediators in rodents.

Downloaded from https://www.science.org at Yale University on June 14, 2022


INTRODUCTION Synapse loss was initially documented ultrastructurally at autopsy,
Despite the millions of individuals afflicted, there remains no although it can now be tracked indirectly with fluorodeoxyglucose
disease-modifying therapy for Alzheimer’s disease (AD). Evidence positron emission tomography (FDG-PET) or directly by synaptic
of -amyloid (A) peptide accumulation is required for AD diagnosis vesicle glycoprotein 2A (SV2A) PET (10–12). PET tracers targeting
(1); however, other hallmarks are misfolded TAU (MAPT) accumu- SV2A, which is ubiquitously expressed in synaptic vesicles, permit
lation, phosphorylation and spreading, and microglial and astroglial quantification of synaptic density from noninvasive scans, facilitat-
reactivity (2, 3). Genetic evidence from dominant early-onset AD ing tracking of neurodegenerative disease progression and drug
mutations in amyloid precursor protein (APP) and presenilins, development. PET imaging with [11C]UCB-J previously demonstrated
from Down’s syndrome, and from protective APP alleles are all reduced hippocampal synaptic density of patients with MCI relative
consistent with a causative role for A (4). Biomarker studies found to unimpaired controls (10, 12, 13) and was also able to detect drug
A accumulation 2 decades before symptoms, suggesting an early rescue of synapse loss in a mouse amyloidogenic model [APPswe/
role for A in disease pathophysiology (5). However, the temporal PS1∆E9 (14); henceforth, APP/PS1] (15). The second-generation
disconnection between A accumulation and symptoms, as well as tracer, [18F]SynVesT-1, with longer half-life, higher resolution,
the clinical failure of multiple A-lowering therapies have expanded and greater potential for clinical translation, was able to discern
the focus of AD pathophysiology. Genetic studies implicate microglia differences in synaptic density between APP/PS1 mice and control
in AD risk and progression. For example, triggering receptor littermates (16).
expressed on myeloid cells 2 (TREM2) signaling has been shown to With regard to biochemical events at failing AD synapses, cellular
exert multiple effects on A-driven pathology (6). In addition, prion protein (PRPC) was identified as a high-affinity receptor for
classic complement cascade components have been directly impli- A oligomers (Ao) in an unbiased genome-wide expression clon-
cated in phagocytic synaptic removal in AD (7), although the basis ing screen (17). PRPC exhibits a unique selectivity for oligomeric A
of synaptic selectivity is not defined. (17–24). In amyloidogenic AD mouse models, Ao binding to PRPC
Crucial for clinical AD progression from presymptomatic to mild contributes to synaptic plasticity impairment, learning and memory
cognitive impairment (MCI) to dementia is the loss of synapses (8, 9). deficits, and synapse loss (17, 25–29). Fyn and Pyk2 (PTK2B) kinases
downstream of Ao and PRPC have been previously linked to TAU
(30–34) and implicated in AD risk (35). A screen for postsynaptic
1
Cellular Neuroscience, Neurodegeneration and Repair Program, Departments of transmembrane proteins linking Ao and PRPC to intraneuronal
Neurology and Neuroscience, Yale University School of Medicine, New Haven, CT
06510, USA. 2Yale PET Center, Department of Radiology and Biomedical Imaging,
signaling identified neuronal metabotropic glutamate receptor 5
Yale University School of Medicine, New Haven, CT 06510, USA. 3Allyx Therapeutics (mGluR5) (23, 36–45). Genetic loss and pharmacological inhibition
Inc., New Haven, CT 06513, USA. 4Graduate School of Cellular and Molecular studies have demonstrated that reduced mGluR5 activity alleviated
Neuroscience, University of Tübingen, Tübingen 72074, Germany. synaptic and memory deficits in multiple amyloidogenic AD mouse
*Corresponding author. Email: jason.cai@yale.edu (Z.C.); stephen.strittmatter@
yale.edu (S.M.S.) models (25,  36,  37,  40–42,  46–54). However, the molecular and
†These authors contributed equally to this work. cellular bases mediating the role of mGluR5 in synaptic loss in AD,

Spurrier et al., Sci. Transl. Med. 14, eabi8593 (2022) 1 June 2022 1 of 16
SCIENCE TRANSLATIONAL MEDICINE | RESEARCH ARTICLE

including neuroglial interactions and complement-driven synaptic To evaluate the translational potential of SAM, pharmacokinetic,
phagocytosis, remain unclear. metabolic, selectivity, and toxicological analyses were pursued.
Allosteric mGluR5 modulators have been subdivided into posi- Studies were conducted in mouse, rat, dog, and monkey to measure
tive (PAMs), negative (NAMs), and silent (SAMs). PAMs enhance the plasma kinetics of SAM after intravenous and oral administra-
and NAMs suppress glutamate-induced G protein–mediated Ca2+ tion. We also assessed the metabolic stability and profile of SAM,
mobilization and/or shift glutamate efficacy. Multiple NAMs re- and observed reduced stability in dog, which led us to choose monkey
duce both physiological glutamate signaling and Ao-PRP C–­ rather than dog as the second toxicology species (data file S1B).
dependent synaptic deficits. As a dose-limiting side effect, blockade Metabolite profiling of SAM identified unchanged parent compound
of glutamate at mGluR5 with a NAM impaired learning and memory and 38 tentative minor metabolites across human and preclinical
independently of AD (36, 55–60). We identified a SAM, BMS-984923, species; metabolic pathways observed in human were also observed
that does not alter basal or glutamate signaling (61–63) but does in investigated preclinical species, and no major metabolite com-
inhibit the PRPC-mGluR5 interaction to prevent pathological Ao prised more than 5% of total (data file S1B). Whereas plasma protein
signaling (39). This SAM has the potential for expanding the thera- binding of SAM was >99% in rat, monkey, and human (data file S1A),
peutic window for mGluR5 as a disease-modifying AD target. oral bioavailability was 50 to 90%, and mouse brain concentrations
Here, we use BMS-984923 to explore the connection between were about twice plasma concentrations (data file S1C). Safety evalu-
neuronal mGluR5 and glia-dependent synapse loss. We showed ation of SAM included a standard Good Laboratory Practice (GLP)
that altered expression of neuronal and glial genes modified neuro-­ mutagenic program, which demonstrated that SAM was not muta-
immune interaction in murine amyloidogenic (APP/PS1) and AD genic (data file S1D). In addition, rats tolerated single oral doses of
{homozygous double knock-in strain [APPNL-G-F/hMAPT (64, 65)]; SAM at 5, 15, and 45 mg/kg without relevant effects on neurobehav-
henceforth labeled dKI} models, resulting in reduced synaptic ioral function, as evaluated using a modified Irwin test, or on body
density. Treatment with SAM restored synaptic density. Comple- temperature (data file S1E). Repeat-dose toxicity studies of up to 7 days
ment tagging of synapses destined for removal is reported to attract in rat and monkey were used to set the dose for the GLP 28-day
microglia and astroglia for phagocytosis of synaptic fragments (66). repeat-dose studies in both species. Rats received 5, 15, or 45 mg kg−1
We showed here that SAM treatment prevented synaptic localiza- day−1, and monkeys received 10, 50, or 200 mg kg−1 day−1. From the

Downloaded from https://www.science.org at Yale University on June 14, 2022


tion of C1Q without altering total C1Q amount or overall gliosis, 28-day GLP studies, the no-observed-adverse effect level (NOAEL,
suggesting that targeting mGluR5 might be effective for producing data file S1E) doses were 15 and 200 mg kg−1 day−1 for rat and mon-
therapeutic effects in AD. key, respectively. The 28-day average maximum concentration and
area under the curve values at the NOAEL are 10,700 ng/ml and
191,000 ng hour/ml in rat and 15,100 ng/ml and 169,000 ng hour/ml
RESULTS in monkey. The average exposure over 24 hours in monkey at NOAEL
SAM occupies mGluR5 receptors at well-tolerated doses was more than 250-fold greater than the plasma IC50 measured from
Our previous data suggested that the mGluR5 SAM, BMS-984923, the receptor occupancy study using [18F]FPEB PET.
is able to reduce AD-related phenotypes in cell culture, brain slices, To further evaluate the specificity and selectivity of SAM, in vitro
and amyloidogenic transgenic (TG) mice (39). Here, to assess assays were completed to evaluate any potential interactions with
mGluR5 receptor occupancy, we completed [18F]FPEB PET in other targets. SAM was tested in a total of 508 binding and cell-
rhesus monkeys and in mice. In monkeys, single intravenous doses based functional assays in a CEREP panel using a single concentra-
of SAM (0.03 to 1.0 mg/kg) blocked mGluR5  in the brain in a tion at 10 M in duplicate (data file S1F). Overall, SAM showed
dose-dependent manner (Fig.  1,  A  and  B). Receptor occupancy no affinity/potency or a selectivity >300-fold for all targets, with
from Lassen plots ranged from 27% for the 0.03 mg/kg dose to the exception of PAR1 (protease-activated receptor 1) and PR
94.4 ± 0.4% for the 1.0 mg/kg dose (Fig. 1B). Plasma total SAM (progesterone receptor) for which there was a selectivity between
concentrations were measured throughout the course of the scan. 100- and 300-fold relative to the mGluR5 Ki of 0.6 nM. We assessed
The full dose-response curve revealed that SAM bound to mGluR5 the in vitro potential for SAM to be an inhibitor of the human trans-
with an ED50 (median effective dose) of 0.11 ± 0.02 mg/kg dose or porters P-glycoprotein (P-gp), organic ion transporting polypeptide
with an IC50 (median inhibitory concentration) of 24.8 ± 5.1 ng/ml 1B1 (OATP1B1), OATP1B3, and BSEP, as well as the potential of
total plasma concentration. Separate plasma protein binding studies SAM to be a substrate of the human transporter P-gp. The IC50’s for
indicated a free fraction of 0.1 to 0.5% in different species (data file each were all far above efficacious SAM concentrations, and SAM
S1A); therefore, the measured IC50 value equates to a free drug showed no evidence to be a P-gp substrate (data file S1G). Further-
concentration of about 0.3 nM, closely similar to the measured more, we investigated potential CYP inhibition and activation from
0.6 nM Ki (inhibition constant) for displacement of [3H]MPEPy for SAM parent and metabolite compounds and found that IC50’s were
in vitro mGluR5 binding (39). In mice, a single oral dose of SAM all far above efficacious SAM drug concentration (data file S1G).
(7.5 mg/kg) administered 2 hours before [18F]FPEB PET blocked SAM inhibited hERG (human ether-a-go-go related gene) in a GLP
brain mGluR5 with an average receptor occupancy of 98% across cell culture assay with an IC50 of 1.14 M, 1900-fold above efficacious
brain regions (Fig. 1C). To further assess occupancy of mGluR5, we exposure concentration (data file S1H). During an in vivo monkey car-
tested the ability of SAM pretreatment to prevent PAM-induced diovascular study at doses up to 200 mg/kg, the QT interval cor-
seizures in mice (39, 67). Doses at 3.75 mg/kg or above completely rected for heart rate was unaltered, and there was no observed effect
blocked mGluR5-PAM–induced seizures (Fig. 1D; ED50 = 1.5 ± on respiratory, central nervous system (CNS), or cardiovascular
0.8 mg/kg), whereas high doses of the chiral enantiomer of BMS- function. Thus, oral dosing with SAM occupied brain mGluR5 sites
984923 (SAM enantiomer, eSAM, at 30 mg/kg) had no effect on and was well tolerated at exposures more than 250-fold greater than
seizures, demonstrating stereoselective receptor binding (Fig. 1D). required to occupy 50% of the mGluR5 receptor sites.

Spurrier et al., Sci. Transl. Med. 14, eabi8593 (2022) 1 June 2022 2 of 16
SCIENCE TRANSLATIONAL MEDICINE | RESEARCH ARTICLE

A Rhesus - [18F]FPEB PET B Rhesus: [18F]FPEB receptor occupancy


1 month, followed by 1 month of drug
10.0 ED50: 0.11 + 0.02 mg/kg washout, which is more than 60 half-
9.0
100 IC50: 24.84 + 5.05 ng/ml
lives in mice. After 1 month of treatment,
MRI

8.0
vehicle-treated TG animals exhibited
75 significant (P < 0.05) reduction of both

Receptor occupancy (%)


7.0
the presynaptic marker SV2A and the
6.0
50
RO by plasma conc_M postsynaptic marker postsynaptic den-
Baseline

5.0
RO by plasma conc_F
sity protein 95 (PSD-95), in both den-
RO by dose_M
4.0 25 RO by dose_F
tate gyrus (DG) and CX, compared
3.0
Non-lin. fit by plasma conc. with the WT, whereas synaptic markers
Non-lin. fit by dose in SAM-treated APP/PS1 animals were
0
2.0
restored to WT amounts in all contexts
1.0 mg/kg

0.0 0.2 0.4 0.6 0.8 1.0 1.2


1.0 Dose (mg/kg)
(P  >  0.05), consistent with previously
0.0
SUVRCB 0 50 100 150 200 250 300
published results (fig. S2, A to E) (39).
Plasma conc (ng/ml)
This rescue persisted for at least a
month after treatment cessation, as
C Mouse - [18F]FPEB PET D Mouse: Blockade of PAM-induced seizure SAM-treated TG samples had signifi-
6.0 ED50: 1.5 + 0.8 mg/kg cantly higher (P < 0.01) synaptic density
MR template

1.0
(with both SV2A and PSD-95 in both DG
5.0
and CX) than vehicle-treated TG samples
0.8 after drug washout (Fig. 2, G to K, and
Normalized seizure score

4.0
fig. S2, F to H). Together, the data
0.6
showed a sustained disease-modifying
Vehicle

3.0
0.4
benefit of blocking the PRPC/mGluR5

Downloaded from https://www.science.org at Yale University on June 14, 2022


SAM
synaptic signaling pathway.
2.0
0.2 eSAM
mGluR5 SAM rescues synapse
density in dKI APPNL-G-F/
7.5 mg/kg

1.0 0.0
0.001 0.01 0.1 1 10
Oral dose (mg/kg) hMAPT mice
0.0
SUVRCB As APP mutant mice can have confound-
ing issues resulting from overexpres-
Fig. 1. SAM receptor occupancy of mGluR5. (A) Template MR images (top) and aligned PET images from a typical sion, competition with endogenous genes,
rhesus brain subject before (middle) and after intravenous dose of SAM (1 mg/kg, bottom), summed from 30 to 45 min
and transgenic promotors, we also tested a
after [18F]FPEB injection. Activity is expressed as SUVR, normalized using cerebellum. (B) Relationship between measured
SAM plasma concentration and mGluR5 receptor occupancy. Fit shown is two-parameter nonlinear mixed effect
homozygous APP NL-G-F/hMAPT dKI
(NLME) model. (C) Template MR images (top) and aligned averaged mouse brain PET images after oral dose of either strain (64, 65). We repeated the mGluR5
vehicle (middle) or SAM (7.5 mg/kg, bottom), summed from 45 to 60 min after injection of [18F]FPEB. Activity is ex- SAM treatment from 12 to 13 months
pressed as SUVR, normalized using cerebellum. (D) Mice were treated with increasing doses of SAM (0 to 7.5 mg/kg) of age in dKI mice and assessed SV2A
or SAM enantiomer (eSAM; 7.5 to 30 mg/kg) by oral gavage 90 min before receiving PAM (20 mg/kg) by intraperitoneal PET and synaptic immunohistology in
injection. Graphs depict the normalized severity of seizure activity (based on Racine scale) after PAM administration. different cohorts (fig. S1, D to G). Base-
line [18F]SynVesT-1 PET found decreased
synapse density in transgenic dKI mice
mGluR5 SAM rescues synapse density in TG compared to WT at 12 months of age (Fig. 3A), which was further
APPswe/PS1∆E9 mice confirmed by region of interest (ROI)–based analysis to be 19% lower
We sought to evaluate whether SAM rescue of preexisting synaptic hippocampal SUVR-1 in dKI mice compared to WT (Fig. 3B,
density deficit in aged AD mice was observed by repeated synaptic P < 0.01). In addition,1-month SAM treatment increased synaptic
PET imaging. Mice were aged to the point that synapse loss was density by 17% as detected by posttreatment [18F]SynVesT-1 PET
detectable using SV2A PET, a tool for monitoring synapse density compared to baseline PET in the same dKI mice (P < 0.05) (Fig. 3, C
with clinical utility in AD and other conditions (fig. S1, A to C and G). and D) to a density comparable to WT mice.
At 12 to 13 months of age, a comparison of baseline [18F]SynVesT-1 We sought to confirm the PET end point by immunohistological
PET found reduced [18F]SynVesT-1 binding in the hippocampus analysis. Both hippocampal and cortical sections of vehicle-treated
(HC) and cerebral cortex (CX) of APP/PS1 relative to wild-type dKI mice exhibited significantly (P < 0.05) reduced SV2A and PSD-
(WT) mice (Fig. 2, A to C; P < 0.0001). Rescans of the same mice 95 punctate staining relative to WT controls (Fig. 3, E to I, and fig.
after a 1-month treatment with the mGluR5 SAM showed a signifi- S3, A to C). In particular, 1-month treatment with SAM restored
cant (P < 0.0001) increase in hippocampal [18F]SynVesT-1 standard- synaptic markers to amounts equal to that of WT mice. To the
ized uptake value ratio minus 1 (SUVR-1) in APP/PS1 compared to extent that the amount of the pre- and postsynaptic markers reflects
WT mice (Fig. 2, D to F). synapse density, then the changes are expected to be similar in
We also tested whether SAM had a chronic disease–modifying individual animals. SV2A and PSD-95 immunoreactivity strongly
effect after treatment cessation by analyzing synaptic marker im- correlated with one another in both DG and CX from both APP/PS1
munostaining. Twelve-month-old APP/PS1 mice were treated for and dKI animal models (fig. S4, A to D; P < 0.0001). Furthermore,

Spurrier et al., Sci. Transl. Med. 14, eabi8593 (2022) 1 June 2022 3 of 16
SCIENCE TRANSLATIONAL MEDICINE | RESEARCH ARTICLE

Fig. 2. In vivo and in vitro assess- A Baseline SUVRBS-1 B C


ment of synaptic density in APP/ Avg. age = 56 + 5 weeks
Wild type vs. APP/PS1
1.0 ****
PS1 mice. (A) Averaged baseline 1.2 7.0
18 [18
F]SynVesT-1 PET
[ F]SynVesT-1 PET images for WT [ F]SynVesT-1 PET
18

(top) and APP/PS1 (bottom) mouse 1.0 6.0 0.75


brain. Activity is expressed as SUVR-1

Wild type

SUVRHC/BS -1
(normalized to brain stem). (B) Voxel-­ 0.8 5.0
wise analysis t value map compar- 0.5

ing APP/PS1 to WT mouse brain. 0.6 4.0


Color scale represents decrease in
0.25
APP/PS1 compared to WT. (C) ROI- 0.4 3.0

based comparison of hippocampal APP/PS1


[18F]SynVesT-1 SUVR-1 shows synapse 0.2 2.0
0.0
density in APP/PS1 compared to WT WT APP/PS1
0.0 1.0
mice. (D) Voxel-wise analysis t value
map comparing baseline and post- D APP/PS1-SAM E F
treatment synaptic density in APP/ Pre vs. Post
ns ns ns
**** 1.0 ****
1.0
PS1 mice. Color scale represents [ F]SynVesT-1 PET
18 7.0
0.9
increase in posttreatment compared 0.9
0.8
to pretreatment mice. (E) ROI-based 6.0
18 0.7
comparison of hippocampal [ F] 0.8

SUVRHC/BS -1

SUVRHC/BS -1
SynVesT-1 SUVR-1 from aged WT 5.0 0.6
0.7
and APP/PS1 animals at baseline (Pre) 0.5

and after SAM treatment (Post). 4.0 0.4


0.6
(F) Baseline (Pre) and posttreatment 0.3
3.0
(Post) hippocampal [18F]SynVesT-1 0.2
0.5
SUVR-1 from SAM-treated APP/PS1 0.1

Downloaded from https://www.science.org at Yale University on June 14, 2022


2.0
animals [seventh and eighth bars in 0.0 0.4
(E)] illustrates individual changes. 1.0
Pre Post Pre Post Pre Post Pre Post Pre Post
WT-veh WT-SAM APP/PS1 APP/PS1
(G) Representative images of SV2A veh SAM
immunofluorescent staining of DG
and CX of APP/PS1 mice after a G Post-washout SV2A staining in APP/PS1 H I
Dentate gyrus Cortex 10 10
month-long treatment washout ***ns **** ****ns ****
period. Scale bar, 20 m. Images from
8 8
SV2A staining in DG and CX of WT
DG SV2A+ area (%)

CX SV2A+ area (%)


animals and from PSD-95 in DG of
6 6
WT and APP/PS1 animals are shown
in fig. S2 (F to H). (H to K) Fractional
Vehicle

4 4
area of SV2A (H and I) or PSD-95
(J and K) immunoreactive puncta 2 2
from DG (H and J) or CX (I and K) of
post-washout animals treated with 0 0

vehicle or SAM. In (C), (E), (F), and 20 µm


WT-veh WT-SAM APP/PS1-veh APP/PS1-SAM
(H) to (K), each individual mouse is
shown as a single dot. In (C), data J K
6 6
are presented as means ± SEM from ** ns ** ** ns ***
n = 24 mice per group and compared
by unpaired t test. In (E), data are
CX PSD-95+ area (%)
DG PSD-95+ area (%)

presented as means ± SEM from n = 12 4 4

mice per group and compared by


SAM

repeated-measures two-way ANOVA


with Holm-Šídák’s multiple compari- 2 2

sons test. In (H) to (K), data are pre-


sented as means ± SEM from n = 14
to 24 mice per group and compared 0 0
by one-way ANOVA with Holm-Šídák’s
multiple comparisons test. **P < 0.01,
***P < 0.001, ****P < 0.0001. ns, not significant.

SV2A staining in DG was significantly correlated with hippocampal males. One recent experiment investigating sex differences in AD con-
SV2A PET SUVR-1 (fig. S4E; P < 0.0001), providing cross-validation cluded that the mGluR5 NAM, CTEP, blocked A pathophysiology
of the two methods. and improved cognitive declines selectively in male but not female
Multiple studies have provided clinical and preclinical evidence APP/PS1 mice (54). In contrast, our SV2A PET studies showed
supporting an increased incidence of AD in females compared to rescue of synaptic density loss in females by treatment with SAM in

Spurrier et al., Sci. Transl. Med. 14, eabi8593 (2022) 1 June 2022 4 of 16
SCIENCE TRANSLATIONAL MEDICINE | RESEARCH ARTICLE

Fig. 3. In vivo and in vitro as- A B C D


WT vs dKI dKI-SAM
sessment of synaptic density ** 1.0
0
Avg. age = 56 + 3 weeks 0
1.0 Pre vs. Post
*
in dKI mice. (A) Voxel-wise 18
[ F]SynVesT-1 PET 7.0 18
[ F]SynVesT-1 PET 7.0 0.9
9
analysis t value map comparing
synaptic density of dKI to WT 6.0
0.8
8
6.0
mice. Color scale represents de-

SUVRHC/BS -1
SUVRHC/BS -1
0.7
7
crease in dKI compared to WT. 5.0 5.0
0.6
6
(B) ROI-based comparison of 0.5
5

hippocampal [ 18 F]SynVesT-1 4.0 4.0 0.5


5
SUVR-1 in dKI mice compared to
0.4
4
WT (WT values are from APP/PS1 3.0 3.0
cohort presented in Fig. 2C). 3
0.3

(C) Voxel-wise analysis t value 2.0 0


0.0 2.0
2
0.2
map comparing synaptic densi- WT dKI Pre Post
ty pre- and posttreatment with 1.0 1.0
SAM. Color scale represents in-
crease in posttreatment com- E Dentate gyrus PSD-95 staining F G
WT dKI
pared to pretreatment mice. 5 **** ns
**** 4 ****ns ****
(D) Paired comparison of ROI-
based analysis of hippocampal 4

CX PSD-95+ area (%)


DG PSD-95+ area (%)
3
SUVR-1 between pretreatment
and posttreatment dKI mice. 3
2
(E) Representative images of
Vehicle

2
PSD-95 immunofluorescent stain-
ing of DG of WT and dKI mice 1
1
after treatment with vehicle or
SAM. Scale bar, 20 m. Images

Downloaded from https://www.science.org at Yale University on June 14, 2022


0 0
from SV2A staining of DG and 20 µm
WT-veh WT-SAM dKI-veh dKI-SAM
CX and from PSD-95 staining of
CX are shown in fig. S3 (A to C). H I
(F to I) Fractional area of PSD-95 10 * ns 0.063
6 ** ns *
(F and G) or SV2A (H and I) im-
munoreactive puncta from DG 8

CX SV2A+ area (%)


(F and H) or CX (G and I) of WT DG SV2A+ area (%)
4
and dKI animals treated with 6
SAM

vehicle or SAM. (J) Representa-


tive images of SV2A and PSD-95 4
2
immunofluorescent costaining
2
of CA1 of dKI mice after treat-
ment with vehicle or SAM. Scale
0 0
bar, 2 m. (K) Quantification of
synaptic loci (per cubic micrometer)
J CA1 SV2A/PSD-95 staining K
in CA1 of vehicle- or SAM-treated dKI-veh dKI-SAM 200
dKI animals. In (B), (D), (F) to (I), *
and (K), each individual mouse is
Synaptic loci (% dKI-veh)

shown as single dot. In (B), data 150

are presented as means ± SEM


from n = 24 (WT) or 10 (dKI) 100
mice per group and compared
by unpaired two-tailed t test. In
(D), data are compared by paired 50

two-tailed t test. In (F) to (I), data


are presented as means ± SEM 0
from n = 21 to 34 mice per group 2 µm dKI-veh
and compared by one-way ANOVA dKI-SAM
with Holm-Šídák’s multiple com-
parisons test. In (K), data are
presented as means ± SEM from n = 12 (dKI-veh) or 11 (dKI-SAM) mice per group and compared by unpaired t test. *P < 0.05, **P < 0.01, ****P < 0.0001. ns, not significant.

both the APP/PS1 and dKI models of AD. To further investigate a Materials and Methods). In the APP/PS1 cohorts (fig. S5A), the dKI
potential sex difference, we carried out a meta-analysis of our SV2A cohorts (fig. S5B), and in an analysis of the two models combined
and PSD-95 histology data segregated by sex (fig. S5, A to C). Stain- (fig. S5C), both males and females were fully rescued by SAM treat-
ing results from both synaptic markers in both DG and CX were ment with highly robust statistical significance (AD-veh versus
compiled into a synaptic score for each individual animal (see WT-veh: P < 0.01; AD-SAM versus WT-veh: P > 0.05).

Spurrier et al., Sci. Transl. Med. 14, eabi8593 (2022) 1 June 2022 5 of 16
SCIENCE TRANSLATIONAL MEDICINE | RESEARCH ARTICLE

To further confirm that these synaptic markers reflected synap- CX and in CA1, and this signal was decreased significantly (P < 0.01)
tic density, we assessed the juxtaposition of SV2A and PSD-95 in CX by 1-month SAM treatment (Fig. 4, E to G). pT217-TAU–
puncta in three dimensions using a modified SEQUIN method (68). positive area, although less than pS396-TAU, still showed accu-
For this purpose, we used super-resolution Airyscan microscopy in mulation (P  <  0.0001) in cerebral CX and in CA1, and was
confocal Z series of double-stained sections of the CA1 from vehicle- significantly decreased (P < 0.05) by SAM treatment in both
and SAM-treated dKI mice and scored synapses as pairs of puncta regions (Fig. 4, H and I). Thus, rescue of synapse loss by SAM in
stained with pre- and postsynaptic proteins within 250 nm of one the dKI model might, at least in part, be mediated by reduced pTAU
another. We observed that SAM treatment significantly (P < 0.05) accumulation; alternatively, reduced pTAU accumulation may be
increased the density of synaptic loci relative to vehicle-treated dKI secondary, in part, to synapse recovery with SAM.
(Fig. 3, J and K). Thus, in multiple strains with multiple measures,
blocking the mGluR5 pathway with SAM had a disease-modifying Single-nuclei transcriptomics of mGluR5 SAM–treated mice
effect to support recovery of synaptic density. reveals neuronal rescue
We considered the extent to which our primary outcome, To explore the molecular underpinnings linking SAM treatment
synapse density, matched with behavioral deficits in these mouse with synaptic recovery in an unbiased manner, we conducted deep
models. We have previously shown that recovery of synapse density single-nuclei transcriptomic analysis of amyloidogenic and AD
associated with behavioral benefit after treatment with SAM in models with and without mGluR5 SAM treatment (Figs. 5 and 6 and
APP/PS1 mice (39). SAM rescued spatial memory in the Morris figs. S7 to S13). We collected brain tissue from aged WT and APP/PS1
water maze, novel object memory, and passive avoidance learning. TG mice treated twice daily with either vehicle (V) or SAM (3.75 mg/kg)
We also showed that rescue of APP/PS1 synapse density by the (D) for 1 month from 12 to 13 months of age (WTV, WTD, TGV,
related mGluR5 NAM compound, MTEP, is accompanied by and TGD), as well as a separate cohort of similarly aged and treated
recovery of spatial memory and novel object memory (19, 27, 36). WT and dKI animals (WTV, WTD, dKIV, and dKID) (Fig. 5A).
Here, we measured memory performance in the dKI model. At Each group included four mice, and all were on a pure C57/Bl6J
3 months of age, the dKI mice exhibited no spatial memory deficit background. Cerebral CX and HC from each brain were dissected,
(fig. S6, A and B), but 12-month-old dKI mice treated with vehicle and the two regions were processed together for each mouse.

Downloaded from https://www.science.org at Yale University on June 14, 2022


for 1 month exhibited significantly impaired learning (fig. S6C; Nuclei isolated by sucrose density gradient centrifugation were pro-
P < 0.0001) and could not recall a hidden platform location during cessed for mRNA profiles with the 10X Genomics platform (~15,000
a probe trial (fig. S6D). In contrast, age-matched dKI mice treated per sample). In total, nearly 500,000 nuclei were sequenced to a
with SAM for 1 month spent significantly (P = 0.0059) more time in depth of ~20,000 reads each (fig. S7, A to D). The resulting DNA
the target quadrant than adjacent areas (fig. S6C). These memory sequences were processed in Seurat, and no confounding batch
phenotypes occurred without any effect of dKI genotype or SAM effects were observed (fig. S8, A and B). Integrated data were
treatment on the ability to swim to a visual platform (fig. S6E). DG clustered in Uniform Manifold Approximation and Projection (UMAP)
SV2A area was also positively correlated with spatial memory per- space (fig. S8C). Cell clusters were identified by canonical cell type
formance (fig. S4F; P = 0.048). Thus, the ability of SAM compound markers, with at least three for each cell type (fig. S8D). The expres-
treatment to restore synapse density in both APP/PS1 and dKI mice sion of one marker for each is shown (fig. S8G). Both models
correlated with recovery of spatial memory. exhibited similar numbers and proportions of each cell type (fig.
S8, E and F).
Phospho-TAU accumulation is reduced in dKI mice by We examined expression profiles for AD-associated and SAM-­
mGluR5 SAM regulated differentially expressed genes (DEGs) across cell types.
TAU hyperphosphorylation and neurofibrillary tangles are key All cell types showed substantial changes induced by APP/PS1 or
components of AD pathology and are thought to be driven by dKI compared to WT. For excitatory neurons (ExNs), a “volcano”
upstream A synaptopathology in human brain and to synergize with plot revealed the pattern of up- and down-regulated genes in TGV
A for further synapse loss (69–71). However, many mouse or dKIV group compared to the WTV group (Fig. 5, B and D).
amyloidogenic AD models (including APP/PS1) show little or no In contrast, many fewer DEGs were present in the comparison of
TAU pathology. The dKI strain inserts human MAPT into the SAM-treated TG or dKI samples with the WTV group (Fig. 5, C and E).
mouse Mapt locus and shows increases in certain phospho-TAU Indeed, 83 and 63% of AD-associated expression changes within
(pTAU) epitopes without tangles (65). Therefore, we assessed pTAU ExNs of APP/PS1 and dKI samples, respectively, were corrected by
pathology in the dKI strain and its response to SAM treatment. The SAM treatment (Fig. 5F). Expression patterns across the different
dKI mice were previously shown to have higher AT8 immuno- groups for each DEG revealed that SAM treatment restored both
reactivity in cortical brain tissue (65). Here, we confirmed by im- up-regulated and down-regulated genes (Fig.  5F). Similar rescue
munohistochemistry and biochemistry that AT8 amounts were was observed in inhibitory neurons (InNs) (Fig. 5G). The list of all
increased in cortical tissue from 12-month-old dKI animals SAM-corrected DEGs from ExNs and InNs and from either mouse
(Fig. 4, A to D; P < 0.0001). One-month SAM treatment significantly model was analyzed for pathway enrichment with ClueGO (74).
reduced AT8 in the CX (Fig. 4, C and D; P < 0.05). Differences in enrichment of networks composed of Gene Ontology
We evaluated two additional disease-associated pTAU markers (GO) cell compartment terms were detected (fig. S9, A to C).
in dKI brain sections. TAU phosphorylation at S396/S404 has been Specifically, neuronal SAM–corrected DEGs were strongly enriched
reported as an early event in AD progression (72), and the pT217- in GO terms related to neuronal synapses, with less prominently
TAU epitope has been documented as a specific plasma and populated networks related to ribosome, V–adenosine triphospha-
cerebrospinal fluid (CSF) marker of AD tauopathy (73). pS396-TAU–­ tase (ATPase), and glial projection. This analysis is consistent with
positive area was significantly (P < 0.0001) increased in dKI cerebral expected primary action of SAM and mGluR5 at neuronal synapses.

Spurrier et al., Sci. Transl. Med. 14, eabi8593 (2022) 1 June 2022 6 of 16
SCIENCE TRANSLATIONAL MEDICINE | RESEARCH ARTICLE

A Medial cortex AT8/NeuN staining B WT-veh WT-SAM dKI-veh dKI-SAM

WT dKI 75 kDa

50 kDa
AT8

50 kDa
Vehicle

37 kDa Actin

C D
0.20 ****ns ** 50 **** *
50 m
*
40

Medial cortex AT8+ area (%)


0.15

Cortical AT8/actin
30
0.10

20
SAM

0.05
10

0.00 0

Downloaded from https://www.science.org at Yale University on June 14, 2022


WT-veh WT-SAM dKI-veh dKI-SAM

E F H
Medial cortex pS396 staining
1.5 **** ** **** *
dKI
** 0.03
*
WT

Medial cortex pT217+ area (%)


Medial cortex pS396+ area (%)

1.0 0.02
Vehicle

0.5 0.01

0.0 0.00
100 m
G I
0.25 **** 0.086
0.025 **** *
** ****
0.20 0.020
CA1 pS396+ area (%)

CA1 pT217+ area (%)

0.15 0.015
SAM

0.10 0.010

0.05 0.005

0.00 0.000

Fig. 4. In vitro assessment of TAU pathophysiology in dKI mice. (A) Representative images of AT8 and NEUN immunofluorescent staining of medial CX of WT and dKI
mice after treatment with vehicle or SAM. Scale bar, 50 m. (B) Representative immunoblots of AT8 amounts in cortical tris-buffered saline (TBS) fraction of WT and dKI
animals treated with vehicle or SAM; actin was used as loading control. (C) Fractional area of AT8 immunoreactive puncta from medial CX of treated WT and dKI animals,
from micrographs as in (A). (D) Quantification of AT8 summed from TBS and Triton X-100 fractions, normalized to actin, from immunoblots as in (B). (E) Representative
images of pS396 immunofluorescent staining of medial CX of WT and dKI mice after treatment with vehicle or SAM. Scale bar, 100 m. (F and G) Fractional area of pS396
immunoreactive puncta from medial CX (F) or hippocampal CA1 (G) of WT and dKI animals treated with vehicle or SAM. (H and I) Fractional area of pT217 immunoreactive
puncta from medial CX (H) or hippocampal CA1 (I) of WT and dKI animals treated with vehicle or SAM. In (C) and (D) and (F) to (I), each individual mouse is shown as single
dot. Data are presented as means ± SEM from n = 7 to 20 mice per group and compared by one-way ANOVA with Holm-Šídák’s multiple comparisons test. *P < 0.05,
**P < 0.01, ****P < 0.0001. ns, not significant.

Spurrier et al., Sci. Transl. Med. 14, eabi8593 (2022) 1 June 2022 7 of 16
SCIENCE TRANSLATIONAL MEDICINE | RESEARCH ARTICLE

Fig. 5. snRNA-seq shows SAM-mediated A 4 WT veh 4 WT veh 12 months age 10X Genomics process & sequence
correction of expression changes in neu- 4 WT SAM 4 WT SAM Treat for 1 month 482,680 total nuclei sequences
4 APP/PS1 veh 4 dKI veh Cortex + hippocampus Mean UMI per nucleus ~1800
ronal and glia cell populations from 4 APP/PS1 SAM 4 dKI SAM Isolate nuclei Dataset (UMI x Nuc) = 8.7 × 10E08
amyloidogenic and AD mouse models.
(A) Cerebral CX and HC from WT, APP/PS1, B APP/PS1 excitatory neurons D dKI excitatory neurons
and dKI mice treated with vehicle or SAM 1000 Camk2n1
Hspa8 1000 Camk2n1Cst3 Hspa8

−Log10 (P value)
500 Ubb Tmsb4x Atp1b1

−Log10 (P value)
Itm2b Cd47 Calm1 Eef1a1 500 Actb
(n = 4 mice per group) from 12 to 13 months 100
Negr1
Aldoa Fth1 Ptgds 100 Calm1
Aldoa Fth1
Hsp90ab1

of age were fractionated to single nuclei and AD-veh Atp1b1


Hsp90ab1
Ubb Itm2b Pgam1 Slc25a4
Mdh1 Tubb2a Ptgds Arl6ip1 Negr1
vs Tmsb4x Mdh1
processed by 10X Genomics. (B to E) Volca- WT-veh
10
Slc2a13 Pgam1 Cst3 Slc25a4
Actb 10
Eef1a1

no plots of DEGs identified in APP/PS1 (B and 1


Arl6ip1
1
Tubb2a Cd47 Slc2a13

−0.6 −0.3 0.0 0.3 0.6 −0.6 −0.3 0.0 0.3 0.6
C) or dKI (D and E) ExN cell populations. Plots Tg-veh vs WT-veh (logFC) dKI-veh vs WT-veh (logFC)
show statistical significance (−log10, P value) C 1000 E 1000

−Log10 (P value)
Hsp90ab1 Camk2n1 Ubb Mdh1 Hspa8 Camk2n1
500
versus magnitude of gene expression chang-

−Log10 (P value)
Calm1 Atp1b1 Cd47 500 Eef1a1
Eef1a1 Tubb2a Slc25a4
100 Slc25a4 Mdh1 Itm2b
es (logFC) of AD-veh (B and D) or AD-SAM (C AD-SAM Aldoa Itm2b
Actb
Hspa8
Arl6ip1
100 Aldoa
Cd47
Calm1
Arl6ip1 Tmsb4x
Fth1 Ubb Pgam1
and E) when compared to WT-vehicle–treated vs
WT-veh
10 Slc2a13 Ptgds
Tubb2a
10 Cst3 Slc2a13
Hsp90ab1 Fth1
mice; vertical dashed line indicates 0.0 logFC, Cst3 Negr1 Pgam1
Tmsb4x Atp1b1
Ptgds Actb
1 1
and vertical solid lines indicate +0.1 logFC. A −0.6 −0.3 0.0 0.3 0.6 −0.6 −0.3 0.0 0.3 0.6
Tg-SAM vs WT-veh (logFC) dKI-SAM vs WT-veh (logFC)
selection of significant, AD-associated DEGs F G
is marked in red, with gene names identified APP/PS1 dKI APP/PS1 dKI
(B and D). DEGs are deemed significant if AD DEGs =127 AD DEGs = 284 AD DEGs = 102 AD DEGs = 274
Not cor. Not cor. Not cor. Not cor.
they exhibit an absolute logFC > 0.1 with SAM cor. 17% SAM cor. SAM cor.
7%
SAM cor.
P < 0.005 (Wilcoxon rank sum test). The 37% 45%
same AD-­associated DEGs are marked in 83% 93%
63% 55%
green (B and D) if their absolute logFC < 0.1.
For full DEG list from all cell types, refer to SAM- SAM-
data file S2. (F to I) Total number of AD-­ corrected P = 2.3e-50 P = 2.5e-73 corrected P = 4.2e-50 P = 4.5e-59
Excitatory neurons

Inhibitory neurons
P value P value
associated DEGs in ExNs (F), InNs (G), microglia

Downloaded from https://www.science.org at Yale University on June 14, 2022


WT-SAM

WT-SAM

WT-SAM

WT-SAM
dKI-SAM

dKI-SAM
(H), and reactive astrocytes (I) from APP/PS1
Tg-SAM

Tg-SAM
WT-veh

WT-veh

WT-veh

WT-veh
dKI-veh

dKI-veh
Tg-veh

Tg-veh
or dKI samples. Pie charts illustrate percent-
age of DEGs that are fully corrected by SAM
2

2
treatment (Fisher’s exact test); the size of the
Z-score scaled exp.

Z-score scaled exp.


1

1
chart is relative to total number of DEGs.
Heatmaps show single DEG expression with-
0

0
in each cell type. Analyses of remaining cell
−1

types are included in fig. S11. −1


−2

−2

H I
APP/PS1 dKI APP/PS1 dKI
AD DEGs = 108 AD DEGs = 390 AD DEGs = 260 AD DEGs = 310
Not cor. Not cor. Not cor. Not cor.
SAM cor. SAM cor. 7% SAM cor. SAM cor.
10%
35% 28%

90% 93% 65% 72%

SAM- SAM-
corrected corrected
SAM-mediated correction was
Reactive astrocytes

P = 7.5e-4 P = 9.4e-9 P = 5.4e-32 P = 4.7e-30


P value P value
predominantly neuronal, because SAM
Microglia

WT-SAM

WT-SAM

dKI-SAM
WT-SAM

WT-SAM

dKI-SAM

Tg-SAM
Tg-SAM

WT-veh

WT-veh

dKI-veh
WT-veh

WT-veh

dKI-veh

treatment failed to correct most DEGs


Tg-veh
Tg-veh

in each glia cell population analyzed


(Fig. 5, H and I, and fig. S10, A to C).
2

2
Z-score scaled exp.

Z-score scaled exp.

Increasing the threshold stringency


1

for differential expression analysis


0

had minimal effect on the propor-


−1

−1

tion of SAM-corrected DEGs across


−2

−2

cell types (fig. S11), highlighting not


only that SAM-mediated correc-
tions were primarily neuronal but
also that treatment corrected at least 73% of neuronal DEGs with in TGD and WTV, indicating that there is complete rescue. Points
the greatest fold-change difference, regardless of model. along the line y = 0 reflect an AD-related phenotype unaffected by
To visualize differential expression transcriptome-wide between SAM treatment. Although the linear regression of all genes showed
TGV, TGD, and WTV, the log fold change (logFC) for WTV com- a highly significant difference from slope  =  0  in every cell type
pared to TGV (AD effect, x axis) was plotted against that for TGD examined (P < 2.2 × 10–16), the effect was much more pronounced in
compared to TGV (SAM effect, y axis) (Fig. 6, A to I, and fig. S12, A neuronal subtypes. For APP/PS1 ExN, the slope was 0.70 and Pearson
to F). In this plot, points along the identity line of slope = 1 are equal R was 0.72, with close to full normalization by SAM (Fig. 6B).

Spurrier et al., Sci. Transl. Med. 14, eabi8593 (2022) 1 June 2022 8 of 16
SCIENCE TRANSLATIONAL MEDICINE | RESEARCH ARTICLE

A The regression of expression profiles for DEGs in InNs showed a


AD-SAM vs AD-veh (LogFC)
0.6 Slope = 1
Full rescue APP / PS1 DEG
Corrected by SAM slope nearly as close to 1. In these neuronal populations, most DEGs
0.3
Slope = 0 dKI DEG were corrected by SAM [colored pink (APP/PS1) or green (dKI)]
No rescue Corrected by SAM
0.0 and largely fell along the y = x identity line. In contrast, although the
DEG not corrected
by SAM slopes in microglia and astrocytes were also significantly different
−0.3
Transcripts below
from 0 (P < 2.2 × 10−16), they were substantially shallower than in
−0.6 DEG cutoff neurons, because most DEGs failed to be corrected (black dots)
−0.6 −0.3 0.0 0.3 0.6 (Fig. 6, B to I, and fig. S12, A to F).
WT−veh vs AD−veh (LogFC)
Both the APP/PS1 and dKI models showed a substantial (P < 2.3 ×
B APP / PS1
C dKI 10−50) rescue of expression patterns by SAM, in correlation with
0.6
dKI-SAM vs. dKI-veh (LogFC)
0.6 synapse rescue by immunohistology and PET imaging (Figs. 2 and
Tg-SAM vs. Tg-veh (LogFC)

3). We compared the AD-related DEGs of all major cell types of the
Excitatory neurons

0.3 0.3
two models and found significant overlap in each context (P < 4.8 ×
0.0 0.0 10−4) (fig. S13). In ExN, the overlap of DEGs contained 46 genes
−0.3 −0.3
(fig. S10, A and B). Of these 46 core DEGs, most were expression-­
R =0.72 R =0.44 normalized by SAM in both models; 36 were rescued in APP/PS1,
P <2.2e-16 P <2.2e-16
−0.6 −0.6 and 31 in dKI (fig. S13B). Thus, single-nuclei RNA sequencing
−0.6 −0.3 0.0 0.3 0.6 −0.6 −0.3 0.0 0.3 0.6
WT−veh vs. Tg−veh (LogFC) WT−veh vs. dKI−veh (LogFC) (snRNA-seq) profiling methods revealed robust expression changes
D E across multiple cell types and pronounced global rescue by SAM
treatment, but the rescue was far more complete in neurons than
dKI-SAM vs. dKI-veh (LogFC)
Tg-SAM vs. Tg-veh (LogFC)

0.6 0.6
in glia.
Inhibitory neurons

0.3 0.3
Research into the role of glial cells in AD progression has identi-
0.0 0.0 fied expression signatures of disease-associated microglia (DAM)
(75) and astrocytes (DAA) (76) in the 5XFAD TG amyloidogenic
−0.3 −0.3

Downloaded from https://www.science.org at Yale University on June 14, 2022


R =0.61 R =0.46 mouse model, each with unique transcriptomic profiles. Our data
−0.6 P <2.2e-16 −0.6 P <2.2e-16 here demonstrated conservation of a subset of these characteristic
−0.6 −0.3 0.0 0.3 0.6 −0.6 −0.3 0.0 0.3 0.6 DEGs in both cell types (fig. S13, G to I). Although microglia from
WT−veh vs. Tg−veh (LogFC) WT−veh vs. dKI−veh (LogFC)
dKI animals had a higher number of DEGs relative to WT controls
F G
than did APP/PS1 (and thus greater overlap with DAMs), both mouse
dKI-SAM vs. dKI-veh (LogFC)

0.6 0.6
Tg-SAM vs. Tg-veh (LogFC)

models had a similar number of overlapping DEGs with DAA.


0.3 0.3
Microglia

0.0 0.0 C1Q localization to PSD-95 puncta is prevented by mGluR5


blockade, and synaptic engulfment is reduced
−0.3 −0.3
R =0.43 R =0.34
Although our single-nuclei transcriptomic data revealed minimal
−0.6 P <2.2e-16 −0.6 P <2.2e-16 glial normalization by SAM, activation of glial cells and C1Q synaptic
−0.6 −0.3 0.0 0.3 0.6 −0.6 −0.3 0.0 0.3 0.6 tagging have been strongly implicated in AD synapse loss (7, 77, 78).
WT−veh vs. Tg−veh (LogFC) WT−veh vs. dKI−veh (LogFC)
We hypothesized that gliosis and complement release may still
H I occur in the presence of SAM but that innate immune mediators
dKI-SAM vs. dKI-veh (LogFC)

0.6 0.6
Tg-SAM vs. Tg-veh (LogFC)

might fail to target the expression-normalized neuronal synaptic


Reactive astrocytes

0.3 0.3 elements. Specifically, the synapse-rescuing effect of SAM treat-


0.0 0.0
ment might relate to a prevention of complement tagging of synapses
rather than an upstream reduction of microgliosis or complement
−0.3 −0.3 production. First, we examined whether the synapse-rescuing
R =0.37 R =0.11
−0.6 P <2.2e-16 −0.6 P <2.2e-16 mGluR5 SAM normalized gliosis, C1Q amounts, or colocalization
−0.6 −0.3 0.0 0.3 0.6 −0.6 −0.3 0.0 0.3 0.6 with PSD-95. We have previously examined astrocyte, microglia,
WT−veh vs. Tg−veh (LogFC) WT−veh vs. dKI−veh (LogFC)
and A area in the hippocampi of treated and nontreated APP/PS1
Fig. 6. Transcriptome-wide correction by SAM treatment in neuronal and glial mice and found that SAM treatment had no effect on gliosis or
cell types. (A) Cell type–specific comparison of AD-associated DEGs and SAM-­ plaque load (39,  79,  80). Here, we obtained similar results for
corrected DEGs in APP/PS1 and dKI samples. LogFC between WT-veh and either SAM-treated dKI (fig. S14, A to F). IBA1 (a microglial marker; fig.
APP/PS1-veh or dKI-veh (AD effect) is plotted along the x axis. LogFC between S14, A and B), GFAP (an astrocyte marker; fig. S14, C and D), and
vehicle- and SAM-treated APP/PS1 or dKI samples (SAM effect) is plotted along the thioflavin S (amyloid plaques; fig. S14, E and F) staining were
y axis. Black points represent genes with logFC > 0.1 and P < 0.005. Colored points significantly increased (P < 0.01, P < 0.05, and P < 0.0001, respec-
(pink = APP/PS1, green = dKI) represent DEGs that were corrected by SAM treat-
tively) in 12-month-old vehicle-treated dKI animals, relative to
ment. Points along the identity line (x = y) represent genes with equivalent differ-
WT, and none was affected by SAM treatment. To assess total C1Q
ential expression between AD-SAM and WT-veh, relative to AD-veh, indicating
complete rescue by SAM. Points along the line “y = 0” reflect genes unaffected by
protein, we completed C1Q immunoblots in the dKI and APP/PS1
SAM. The regression line (Pearson’s correlations) represents transcriptome-wide samples with and without SAM treatment. Total C1Q in dKI HC sam-
effects of SAM treatment. Expression pattern for different cell types is provided for ples was substantially increased in the vehicle-treated samples rela-
ExNs (B and C), InNs (D and E), microglia (F and G), and reactive astrocytes (H and tive to WT, and this was not altered by the SAM treatment (dKI;
I). Analyses of remaining cell types are included in fig. S12. Fig. 7, A and B; P < 0.0001).

Spurrier et al., Sci. Transl. Med. 14, eabi8593 (2022) 1 June 2022 9 of 16
SCIENCE TRANSLATIONAL MEDICINE | RESEARCH ARTICLE

A B 10.0 **** ns
WT-veh WT-SAM dKI-veh dKI-SAM
****
37 kDa 8.0
C1q WT-veh
25 kDa
6.0

C1q / actin
WT-SAM

50 kDa 4.0 dKI-veh


Actin
37 kDa dKI-SAM
2.0

0.0
C C1q / PSD-95 staining in dentate gyrus D
WT dKI
0.15
****ns **

area overlapping C1q


DG PSD-95+ fractional
WT-veh
0.10

WT-SAM
Vehicle

dKI-veh
0.05
dKI-SAM

5 µm
0.00

Downloaded from https://www.science.org at Yale University on June 14, 2022


E
0.06 ** ns *
area overlapping C1q
DG PSD-95+ fractional

WT-veh
0.04
WT-SAM
SAM

APP/PS1-veh
0.02
APP/PS1-SAM

0.00

F SV2a / GFAP staining in CA1 G


dKI-veh dKI-SAM
0.4
**** *
****
0.3 WT-veh
CA1 SV2a+ area (%)
engulfed in GFAP

WT-SAM
0.2
dKI-veh

dKI-SAM
0.1

5 µm 0.0

Fig. 7. Synaptic localization of C1Q in amyloidogenic and AD mouse models. (A) Total C1Q in HC from WT and dKI animals treated with vehicle or SAM were mea-
sured by Western blotting; actin was used as loading control. (B) Quantification of immunoblot in (A). C1Q intensity was normalized to actin. Data are means ± SEM
from n = 6 mice per group and compared by one-way ANOVA with Holm-Šídák’s multiple comparisons test. (C) Representative images of C1Q and PSD-95 immunofluo-
rescent staining of DG of WT and dKI mice after treatment with vehicle or SAM imaged with Airyscan super-resolution. Scale bar, 5 m. Yellow arrows highlight C1Q/
PSD-95 colocalization. (D and E) Fractional area of PSD-95 immunoreactive puncta overlapping C1Q in DG of dKI (D) or APP/PS1 (E) animals treated with vehicle or
SAM. Each individual mouse is shown as a single dot. Data are presented after measurement by Mander’s coefficient as means ± SEM from n = 6 to 10 mice per group and
compared by one-way ANOVA with Holm-Šídák’s multiple comparisons test. (F) Representative orthogonal views from different planes (x/y, x/z, or y/z) of SV2A and GFAP
immunofluorescent staining of CA1 of dKI mice after treatment with vehicle or SAM. Scale bar, 5 m. (G) Fractional area of SV2A immunoreactive puncta engulfed in GFAP
in CA1 of WT or dKI animals treated with vehicle or SAM. Each individual mouse is shown as a single dot. Data are presented as means ± SEM from n = 8 to 10 mice per
group and compared by one-way ANOVA with Holm-Šídák’s multiple comparisons test. *P < 0.05, **P < 0.01, ****P < 0.0001.

Spurrier et al., Sci. Transl. Med. 14, eabi8593 (2022) 1 June 2022 10 of 16
SCIENCE TRANSLATIONAL MEDICINE | RESEARCH ARTICLE

To assess innate immune tagging of synapses by complement, mGluR5 SAM (BMS-984923) blocked this AD process, while
we analyzed DG sections by super-resolution Airyscan microscopy preserving physiological glutamate signaling, and had a broad safety
with antibodies against C1Q and PSD-95 (Fig. 7C). At 12 months margin in toxicology studies. Thus, SAM-mediated correction of
old, the fraction of PSD-95 overlapping with C1Q was significantly neuronal expression blocked C1Q synaptic localization and pre-
increased in both dKI (Fig. 7D; P < 0.0001) and APP/PS1 (Fig. 7E; vented synaptic engulfment to preserve synapses in rodent models.
P < 0.01) vehicle-treated mice, consistent with published reports in Synapse loss is a robust biomarker of AD progression that
other models (7, 81, 82). In contrast, animals treated for 1 month correlates well with cognitive decline (8–12). Development of a non-
with SAM exhibited PSD-95/C1Q colocalization that was reduced invasive in vivo imaging approach to quantify synaptic density
relative to vehicle-treated animals (P < 0.05) and was indistinguishableis essential for early diagnosis of AD. SV2A, a ubiquitously and
from WT. C1Q overlap with the presynaptic marker SYNAPSIN homogenously expressed isoform of SV2, is found in synapses
1/2 was also significantly elevated (P < 0.05) in DG of vehicle-treated throughout the brain. Expanding on recent studies (16, 84, 85), we
dKI mice and restored to WT amount in SAM-treated animals used longitudinal PET imaging with [18F]SynVesT-1 to first demon-
(fig. S15, A and B). This finding is consistent with the hypothesis strate decreases in hippocampal synaptic density of two separate
that SAM normalization of synaptic signaling and neuronal gene mouse models of amyloidogenesis and AD and then to show
expression in these amyloidogenic and AD models eliminates syn- pharmacological rescue. These findings were confirmed using
aptic tagging by complement, despite persistent glial activation and immunohistochemical staining of SV2A. The successful application
C1Q production. of [18F]SynVesT-1 in preclinical evaluation of treatment advances
After synaptic tagging by C1Q, both astrocytes and microglia are its utility as a tool for development of therapeutic agents of AD and
thought to participate in synaptic removal by phagocytic mecha- other neurodegenerative diseases, and supports its development as
nisms. One study highlighted a greater role for astrocytes in clear- a tool for diagnosis and assessment of AD disease progression.
ing synaptic debris and microglia for somatic debris (83). To assess Pharmacological rescue of synaptic density in aged APP/PS1
astroglial engulfment of synaptic elements, brain samples from and dKI mice revealed that new synapses are being formed in these
12- to 13-month-old WT and dKI animals were costained with animals. It is documented by repeated in vivo two-photon imaging
SV2A and GFAP after 1 month of treatment with vehicle or SAM that aged WT mice (86) and aged APP/PS1 mice (87) continue to

Downloaded from https://www.science.org at Yale University on June 14, 2022


(Fig. 7, F and G). SAM-treated dKI samples exhibited significantly form dendritic spines and stabilize new synapses. Because new
(P  <  0.05) lower SV2A signal contained within GFAP-positive dendritic spines are continually gained and stabilized throughout
astrocytes in CA1, relative to dKI-vehicle samples. Although SAM-­ aging, it is potentially possible to restore overall synaptic density
treated dKI samples did exhibit reductions in astrocytic engulfment by blocking the mechanisms underlying accelerated synaptic loss
of SV2A, the engulfment was still significantly higher than WT without an effect on synaptogenesis per se. On the basis of the acute
samples (Fig. 7G; P < 0.0001). Together, SAM blockade of mGluR5-­ effect of Ao to drive dendritic spine loss and the ability of SAM
dependent and AD-selective synaptic signaling normalizes neuronal treatment to prevent this acute loss (19,  36,  39), we favor the
gene expression and is associated with reductions of both C1Q hypothesis that synaptogenesis is not altered by SAM but rather
synaptic tagging and histologic evidence of synaptic engulfment. that increased density over time is due to a shift in the balance
of gain and loss. The observation that SAM treatment of WT
animals did not result in greater synaptic density is also consistent
DISCUSSION with a specific action to limit AD-related synaptic loss and thereby
In the current study, we showed that synapse loss can be tracked permit a net increase of synapse density over 1 month in the
using longitudinal PET imaging in rodents. Using this method, we disease model.
found that synapse loss was mGluR5-dependent in both a MAPT In addition to synapse loss, microscopic pathological hallmarks
gene KI AD mouse model and a TG amyloidogenic mouse model. of AD in humans include A plaques and neurofibrillary tangles
The mGluR5 intervention was initiated after synapse density had composed of hyperphosphorylated TAU (2,  3). APP/PS1 mice
declined at 12 months of age, and SAM treatment was able to re- recapitulate amyloid phenotypes but do not develop TAU tangles.
verse the deficit in reimaged mice. This disease-modifying benefit To assess SAM-mediated rescue of pathological TAU phenotypes,
persisted for 1 month after washout and was coupled with reduced we previously treated 3×Tg mice (39). These mice express the
pTAU accumulation by dKI mice expressing human TAU. Mecha- human MAPT (P301L) transgene in addition to the APPswe transgene
nistically, single-nuclei transcriptomic profiling revealed normaliza- and a Psen1 transgene with the familial AD (fAD) M146V mutation
tion of neuronal signatures, but little change in glial cells. However, (88). SAM treatment of aged 3×Tg mice reduced soluble p(S199/
this does not imply that glial responses are separable from AD-­ S202)-TAU and insoluble TAU. To extend this finding to a setting
related synapse loss. In contrast, the mGluR5-mediated neuronal without overexpression or FTLD-mutant TAU, we used the homo-
changes eliminated the AD-dependent localization of C1Q to zygous dKI model APP NL-G-F /hMAPT. We confirmed pTAU
synaptic sites and reduced the AD-associated synaptic fragment accumulation and found that pTAU phenotypes were decreased by
engulfment by astrocytes. Our data support the hypothesis that SAM treatment. We have previously shown that SAM treatment
the synapse-depleting neuroglial interaction in these amyloidogenic prevents aberrant activation of proline-rich tyrosine kinase 2 (Pyk2,
and AD models is prevented by mGluR5 blockade–dependent nor- PTK2B) (39), a kinase linked to AD risk and reported to phosphorylate
malization of neuronal gene expression. However, we cannot rule TAU at Y18 (89). In the amyloidogenic and AD models studied
out additional effects of SAM treatment on nonneuronal cells that here, SAM treatment lowered A-triggered TAU phosphorylation,
were not detected by extensive expression profiling or immuno- at least in part by normalization of downstream Pyk2. SAM-­
histology but contributed to synaptic density restoration. Of trans- mediated reduction of pTAU accumulation may contribute to
lational relevance, a subnanomolar potent and orally available restoration of synaptic density.

Spurrier et al., Sci. Transl. Med. 14, eabi8593 (2022) 1 June 2022 11 of 16
SCIENCE TRANSLATIONAL MEDICINE | RESEARCH ARTICLE

Previous studies support sex-dependent differences in AD blocked synapse loss. Consistent with these findings, we show here
incidence. Using APP/PS1 mice, a recent study found that negative that preventing localization of C1Q to synapses by SAM treatment
allosteric modulation of mGluR5 selectively improved cognitive is sufficient to prevent synaptic engulfment. snRNA-seq analysis
decline in male but not female mice. However, our SV2A PET studies demonstrated that SAM treatment had the strongest effect in neu-
and extensive histological data found no difference between sexes in rons, with minimal alterations to expression changes in glial cells.
multiple brain regions using the same APP/PS1 background or the Specifically, neuronal SAM–corrected DEGs are enriched in GO
dKI model. One explanation for the difference in sex dependency terms related to synapses. Biochemical and histological analysis
between these studies relates to the use of a SAM in the current showed that C1Q, IBA1, and GFAP amounts were not altered by
study at full receptor occupancy, in contrast to a NAM with atten- SAM treatment, supporting the transcriptomic data. Despite the
dant toxicity and dose limitation in the previous work. Pharmaco- persistence of gliosis and plaque density in SAM-treated samples,
kinetic and receptor occupancy were not monitored in the CTEP we showed elimination of the synaptic localization of C1Q and a
study (54). Nonetheless, it is clear that both male and female AD reduction of synaptic engulfment markers in astrocytes. The
mice of different strains benefit from SAM treatment target- observed reduction of synaptic engulfment by GFAP-positive cells
ing mGluR5. may be sufficient for synaptic recovery by itself, although it is
The list of genes altered in the AD mouse neurons and corrected also possible that SAM treatment reduces engulfment by other glial
by SAM overlaps strongly with data from human AD expression populations not examined here, including microglia and other
profiling. Nineteen of the 46 SAM-corrected ExN DEGs common GFAP-negative phagocytic cells. Both multiple epidermal growth
to both models are human dorsolateral prefrontal CX ExN DEGs factor–like domains 10 (Megf10), a C1Q receptor, and MerTk
(90). A number of these genes have also been reported as AD contribute to astrocytic clearance of neuronal debris and synap-
biomarkers. Aldolase, fructose-bisphosphate A (ALDOA) function tic elimination in developing and adult CNS (77, 105). However,
is down-regulated in AD hippocampal tissue (91) and is a CSF no expression changes in these receptors were detected in neu-
biomarker discriminating between AD and non-AD cognitive im- rons in the current analysis. Among the genes differentially
pairment (92). Calmodulin (CALM1) is part of an intersecting sub- expressed in both amyloidogenic and AD models and rescued by
network from four consensus modules of genome-wide association SAM in both ExNs and InNs is CD47, a protein shown to inhibit

Downloaded from https://www.science.org at Yale University on June 14, 2022


studies (GWAS) for CSF t-TAU/A1–42 ratio in the Alzheimer’s complement-­dependent synaptic pruning (106). Thus, normal-
Disease Neuroimaging Initiative (ADNI) (93). CYSC (CST3) cor- ization of the phagocytosis-inhibiting signal CD47  in neurons
relates with CSF concentrations of A42 and TAU (94) and is re- by SAM may contribute to synaptic protection. Although CD47 is
duced in AD CSF (95). Prostaglandin D2 synthase/-trace (PTGDS) normalized by SAM treatment in both models, the direction of dif-
is also reduced in AD CSF (95) and is part of network module with ferential expression in the two models is opposite (up-regulated
a positive correlation to AD in frontal CX tissue (96). The ATPase in APP/PS1 neurons, down-regulated in dKI neurons), perhaps
subunit ATP1B1 is included in a network module with a negative due to a later disease stage for APP/PS1 versus dKI. Regardless,
correlation to AD phenotypes (96). Malate dehydrogenase 1 (MDH1) rescuing the altered neuronal expression profile prevented
is elevated in AD CSF (97) and is a hub in gene networks contribut- synaptic tagging, effectively blocking complement-dependent
ing to AD (98). Heat shock protein 90 alpha family class B member 1 synaptic loss.
(HSP90AB1), a molecular chaperone, is a subnetwork node within an Our study has limitations. Although we used two different
“amyloid neuropathies” network of AD (51). Integral membrane strains, the data are derived from murine models that do not fully
protein 2B (ITM2B) regulates glutamate release and synaptic func- recapitulate human AD. In particular, the TAU phenotypes and cell
tion (99), whereas dominant ITM2B mutations cause familial British death in these models are limited compared to the human condi-
and Danish dementias, diseases with amyloid angiopathy, and neuro- tion. Human AD also has a substantially slower progression and
fibrillary tangles (100). BRI2 (encoded by ITM2B) binds APP and much longer course than any murine model. The mouse strains are
inhibits A42 aggregation (100). BRI2 is also increased in early-stage inbred, and mice were housed in conditions relatively free of
AD HC, and deposits are associated with A plaques (101). CAMK2N1 pathogens. Thus, the translation of these studies to human requires
is a potent and specific inhibitor of CAMKII in the postsynaptic den- detailed clinical investigation.
sity. CAMKII is downstream of the PRPC-mGluR5 signaling com- Although our snRNA-seq data were supported by histological
plex aberrantly activated by Ao (38). Dysregulation of hippocampal evidence, most identified DEGs were not validated. Additional
CAMKII has been reported in AD (102). experiments are necessary to confirm the transcriptional changes
Despite synapse recovery by SAM treatment, microglial and astro- and investigate whether these changes manifest at the protein level.
cyte gene expression and histologic changes in the amyloidogenic Such experiments will also shed light on the mechanistic under-
and AD models were altered by SAM to a minimal extent. Although pinnings of synaptic engulfment, including specific C1Q receptors
there was a relative absence of glial changes, we cannot rule out and the responsible cell types.
direct effects of SAM on glial action as a potential contributing Although SAM rescue is mediated by mGluR5, we cannot rule
factor. However, even if SAM directly acts on glial cells, most out contributions of other A cell surface receptors to neurodegen-
AD-related transcriptional changes were observed selectively in eration. This could potentially explain the failure of SAM treatment
neurons. This is consistent with glial activation being driven by to fully rescue the observed increases in pTAU in the dKI model.
amyloid accumulation rather than synaptic damage (103). Subsequent Despite this, synaptic density was restored to WT levels in dKI
to glial activation in control AD mice, C1Q production tags synapses mice, so these contributions must not be essential.
for engulfment by both astrocytes and microglia (7, 104). C1Q and In conclusion, SAM is a potent and selective inhibitor of the
C3 were found to be up-regulated and colocalized to synapses in Ao-induced mGluR5 complex. Toxicological studies showed
AD models, whereas inhibition of classic complement components safety and tolerability at doses more than 250-fold above the

Spurrier et al., Sci. Transl. Med. 14, eabi8593 (2022) 1 June 2022 12 of 16
SCIENCE TRANSLATIONAL MEDICINE | RESEARCH ARTICLE

dose required for 50% receptor occupancy in primate brain and for SUPPLEMENTARY MATERIALS
rescue of pathophysiological TAU, C1Q synaptic localization, synap- www.science.org/doi/10.1126/scitranslmed.abi8593
Materials
tic engulfment, and synaptic depletion in mouse amyloidogenic and
Figs. S1 to S15
AD models. This work further reinforces that targeting an Ao-­ Data files S1 to S4
induced mGluR5 complex can modify the course of AD and sup- MDAR Reproducibility Checklist
ports the advancement of SAM intervention into human testing References (108–120)
(ClinicalTrials.gov, NCT04805983). SAM treatment was shown to View/request a protocol for this paper from Bio-protocol.
have a disease-modifying benefit, because synaptic density is main-
tained even after a month of drug washout. Because gliosis and REFERENCES AND NOTES
plaque density are unchanged by SAM treatment, SAM interven- 1. C. R. Jack Jr., D. A. Bennett, K. Blennow, M. C. Carrillo, B. Dunn, S. B. Haeberlein, D. M. Holtzman,
W. Jagust, F. Jessen, J. Karlawish, E. Liu, J. L. Molinuevo, T. Montine, C. Phelps, K. P. Rankin,
tion was still effective even after plaque accumulation. As such,
C. C. Rowe, P. Scheltens, E. Siemers, H. M. Snyder, R. Sperling, NIA‐AA Research Framework:
there is potential for synergy with pharmacological treatments that Toward a biological definition of Alzheimer's disease. Alzheimers Dement. 14, 535–562 (2018).
suppress gliosis and/or target A or TAU. Future work should 2. J. M. Long, D. M. Holtzman, Alzheimer disease: An update on pathobiology and treatment
explore the molecular details of C1Q synaptic recruitment, the strategies. Cell 179, 312–339 (2019).
mechanisms of synaptic engulfment, and the therapeutic develop- 3. B. De Strooper, E. Karran, The cellular phase of Alzheimer's disease. Cell 164, 603–615
(2016).
ment of the SAM compound. 4. A. A. Pimenova, T. Raj, A. M. Goate, Untangling genetic risk for Alzheimer's disease. Biol.
Psychiatry 83, 300–310 (2018).
5. E. McDade, G. Wang, B. A. Gordon, J. Hassenstab, T. L. S. Benzinger, V. Buckles,
MATERIALS AND METHODS A. M. Fagan, D. M. Holtzman, N. J. Cairns, A. M. Goate, D. S. Marcus, J. C. Morris, K. Paumier,
C. Xiong, R. Allegri, S. B. Berman, W. Klunk, J. Noble, J. Ringman, B. Ghetti, M. Farlow,
Study design
R. A. Sperling, J. Chhatwal, S. Salloway, N. R. Graff-Radford, P. R. Schofield, C. Masters,
The objective of this study was to investigate the ability of an M. N. Rossor, N. C. Fox, J. Levin, M. Jucker, R. J. Bateman; Dominantly Inherited Alzheimer
mGluR5 SAM (BMS-984923) to restore synaptic density and the Network, Longitudinal cognitive and biomarker changes in dominantly inherited
underlying molecular mechanisms. For nonclinical pharmacology, Alzheimer disease. Neurology 91, e1295–e1306 (2018).
drug metabolism and pharmacokinetics (DMPK), and toxicology 6. A. Deczkowska, A. Weiner, I. Amit, The physiology, pathology, and potential therapeutic
applications of the TREM2 signaling pathway. Cell 181, 1207–1217 (2020).

Downloaded from https://www.science.org at Yale University on June 14, 2022


experiments, multiple cell lines and animal models were used (Sup- 7. S. Hong, V. F. Beja-Glasser, B. M. Nfonoyim, A. Frouin, S. Li, S. Ramakrishnan, K. M. Merry,
plementary Materials). To assess synaptic density, rescue, and its Q. Shi, A. Rosenthal, B. A. Barres, C. A. Lemere, D. J. Selkoe, B. Stevens, Complement
mechanisms, we treated amyloidogenic AD mice and subjected them and microglia mediate early synapse loss in Alzheimer mouse models. Science 352,
to PET imaging and behavioral assays; after sacrifice, tissue was 712–716 (2016).
8. S. W. Scheff, S. T. DeKosky, D. A. Price, Quantitative assessment of cortical synaptic
used for immunohistochemistry, biochemistry, and snRNA-seq.
density in Alzheimer's disease. Neurobiol. Aging 11, 29–37 (1990).
Research methods and results are reported in compliance with 9. S. W. Scheff, D. A. Price, F. A. Schmitt, E. J. Mufson, Hippocampal synaptic loss in early
the Animal Research: Reporting of In Vivo Experiments (ARRIVE) Alzheimer's disease and mild cognitive impairment. Neurobiol. Aging 27, 1372–1384
guidelines. Sample sizes were chosen on the basis of previous re- (2006).
sults to enable adequate statistical power. Mice were random- 10. A. P. Mecca, M. K. Chen, R. S. O'Dell, M. Naganawa, T. Toyonaga, T. A. Godek, J. E. Harris,
H. H. Bartlett, W. Zhao, N. B. Nabulsi, B. C. V. Wyk, P. Varma, A. F. T. Arnsten, Y. Huang,
ized to treatment groups using the standard = RAND() function R. E. Carson, C. H. van Dyck, In vivo measurement of widespread synaptic loss
in Microsoft Excel, and groups were matched for age and gender in Alzheimer's disease with SV2A PET. Alzheimers Dement. 16, 974–982 (2020).
(fig. S1). Details about sample number for each experiment are 11. M. Colom-Cadena, T. Spires-Jones, H. Zetterberg, K. Blennow, A. Caggiano, S. T. DeKosky,
outlined in fig. S1G and in the relevant figure legend. For experi- H. Fillit, J. E. Harrison, L. S. Schneider, P. Scheltens, W. de Haan, M. Grundman,
C. H. van Dyck, N. J. Izzo, S. M. Catalano; Synaptic Health Endpoints Working Group, The
mental groups with n < 20, individual data values are provided in
clinical promise of biomarkers of synapse damage or loss in Alzheimer’s disease.
data file S3. All experiments were conducted in a blinded fashion Alzheimers Res. Ther. 12, 21 (2020).
with respect to genotype and treatment. Outliers were identified in 12. M. K. Chen, A. P. Mecca, M. Naganawa, S. J. Finnema, T. Toyonaga, S. F. Lin, S. Najafzadeh,
PRISM by ROUT and removed. For specific experiments, see rele- J. Ropchan, Y. Lu, J. W. McDonald, H. R. Michalak, N. B. Nabulsi, A. F. T. Arnsten, Y. Huang,
vant Materials and Methods section for exclusion criteria, outcome R. E. Carson, C. H. van Dyck, Assessing synaptic density in Alzheimer disease with synaptic
vesicle glycoprotein 2A positron emission tomographic imaging. JAMA Neurol. 75,
measures, statistical methods, and experimental procedures. Re-
1215–1224 (2018).
agents and experimental animals are summarized in data file S4. 13. N. B. Nabulsi, J. Mercier, D. Holden, S. Carre, S. Najafzadeh, M. C. Vandergeten, S. F. Lin,
A. Deo, N. Price, M. Wood, T. Lara-Jaime, F. Montel, M. Laruelle, R. E. Carson, J. Hannestad,
Statistics Y. Huang, Synthesis and preclinical evaluation of 11C-UCB-J as a PET tracer for imaging
All results are presented as means ± SEM, unless otherwise the synaptic vesicle glycoprotein 2A in the brain. J. Nucl. Med. 57, 777–784 (2016).
14. J. L. Jankowsky, D. J. Fadale, J. Anderson, G. M. Xu, V. Gonzales, N. A. Jenkins,
specified. GraphPad Prism 9 software was used for statistical analy- N. G. Copeland, M. K. Lee, L. H. Younkin, S. L. Wagner, S. G. Younkin, D. R. Borchelt, Mutant
sis. Data were analyzed using unpaired t tests and one-way, presenilins specifically elevate the levels of the 42 residue beta-amyloid peptide in vivo:
two-way, or three-way analysis of variance (ANOVA), followed Evidence for augmentation of a 42-specific gamma secretase. Hum. Mol. Genet. 13,
by post hoc Tukey’s multiple comparisons test, Dunnett’s multiple 159–170 (2004).
15. T. Toyonaga, L. M. Smith, S. J. Finnema, J. D. Gallezot, M. Naganawa, J. Bini, T. Mulnix,
comparisons test, Holm-Šídák’s multiple comparisons test, or
Z. Cai, J. Ropchan, Y. Huang, S. M. Strittmatter, R. E. Carson, In vivo synaptic density
Fisher’s least significant difference post hoc pairwise compari- imaging with 11C-UCB-J detects treatment effects of saracatinib in a mouse model
sons test as specified in the figure legends. Only two-sided tests of Alzheimer disease. J. Nucl. Med. 60, 1780–1786 (2019).
were used, and all data analyzed met the assumption for the 16. P. Sadasivam, X. T. Fang, T. Toyonaga, S. Lee, Y. Xu, M. Q. Zheng, J. Spurrier, Y. Huang,
specific statistical test that was performed. Probability levels S. M. Strittmatter, R. E. Carson, Z. Cai, Quantification of SV2A binding in rodent brain
using [18F]SynVesT-1 and PET imaging. Mol. Imaging Biol. 23, 372–381 (2021).
of P < 0.05 were considered statistically significant. Statistical 17. J. Lauren, D. A. Gimbel, H. B. Nygaard, J. W. Gilbert, S. M. Strittmatter, Cellular prion
tests for gene expression in single-nuclei data were carried out in protein mediates impairment of synaptic plasticity by amyloid-beta oligomers. Nature
Seurat V3.6 (107). 457, 1128–1132 (2009).

Spurrier et al., Sci. Transl. Med. 14, eabi8593 (2022) 1 June 2022 13 of 16
SCIENCE TRANSLATIONAL MEDICINE | RESEARCH ARTICLE

18. M. A. Kostylev, A. C. Kaufman, H. B. Nygaard, P. Patel, L. T. Haas, E. C. Gunther, Cohorts for Heart and Aging Research in Genomic Epidemiology, S. Moebus, P. Mecocci,
A. Vortmeyer, S. M. Strittmatter, Prion-protein-interacting Amyloid- oligomers of high M. Del Zompo, W. Maier, H. Hampel, A. Pilotto, M. Bullido, F. Panza, P. Caffarra, B. Nacmias,
molecular weight are tightly correlated with memory impairment in multiple Alzheimer J. R. Gilbert, M. Mayhaus, L. Lannefelt, H. Hakonarson, S. Pichler, M. M. Carrasquillo,
mouse models. J. Biol. Chem. 290, 17415–17438 (2015). M. Ingelsson, D. Beekly, V. Alvarez, F. Zou, O. Valladares, S. G. Younkin, E. Coto,
19. J. W. Um, H. B. Nygaard, J. K. Heiss, M. A. Kostylev, M. Stagi, A. Vortmeyer, T. Wisniewski, K. L. Hamilton-Nelson, W. Gu, C. Razquin, P. Pastor, I. Mateo, M. J. Owen, K. M. Faber,
E. C. Gunther, S. M. Strittmatter, Alzheimer amyloid- oligomer bound to postsynaptic P. V. Jonsson, O. Combarros, M. C. O'Donovan, L. B. Cantwell, H. Soininen, D. Blacker,
prion protein activates Fyn to impair neurons. Nat. Neurosci. 15, 1227–1235 (2012). S. Mead, T. H. Mosley Jr., D. A. Bennett, T. B. Harris, L. Fratiglioni, C. Holmes, R. F. de Bruijn,
20. L. M. Smith, S. M. Strittmatter, Binding sites for amyloid- oligomers and synaptic toxicity. P. Passmore, T. J. Montine, K. Bettens, J. I. Rotter, A. Brice, K. Morgan, T. M. Foroud,
Cold Spring Harb. Perspect. Med. 7, a024075 (2017). W. A. Kukull, D. Hannequin, J. F. Powell, M. A. Nalls, K. Ritchie, K. L. Lunetta, J. S. Kauwe,
21. M. A. Kostylev, M. D. Tuttle, S. Lee, L. E. Klein, H. Takahashi, T. O. Cox, E. C. Gunther, E. Boerwinkle, M. Riemenschneider, M. Boada, M. Hiltuenen, E. R. Martin, R. Schmidt,
K. W. Zilm, S. M. Strittmatter, Liquid and hydrogel phases of PrPC linked to conformation D. Rujescu, L. S. Wang, J. F. Dartigues, R. Mayeux, C. Tzourio, A. Hofman, M. M. Nothen,
shifts and triggered by Alzheimer's Amyloid- oligomers. Mol. Cell 72, 426–443.e12 C. Graff, B. M. Psaty, L. Jones, J. L. Haines, P. A. Holmans, M. Lathrop, M. A. Pericak-Vance,
(2018). L. J. Launer, L. A. Farrer, C. M. van Duijn, C. Van Broeckhoven, V. Moskvina, S. Seshadri,
22. S. Chen, S. P. Yadav, W. K. Surewicz, Interaction between human prion protein J. Williams, G. D. Schellenberg, P. Amouyel, Meta-analysis of 74,046 individuals identifies
and amyloid-beta (Abeta) oligomers: Role of N-terminal residues. J. Biol. Chem. 285, 11 new susceptibility loci for Alzheimer's disease. Nat. Genet. 45, 1452–1458 (2013).
26377–26383 (2010). 36. J. W. Um, A. C. Kaufman, M. Kostylev, J. K. Heiss, M. Stagi, H. Takahashi, M. E. Kerrisk,
23. L. T. Haas, M. A. Kostylev, S. M. Strittmatter, Therapeutic molecules and endogenous A. Vortmeyer, T. Wisniewski, A. J. Koleske, E. C. Gunther, H. B. Nygaard, S. M. Strittmatter,
ligands regulate the interaction between brain cellular prion protein (PrPC) Metabotropic glutamate receptor 5 is a coreceptor for Alzheimer A oligomer bound
and metabotropic glutamate receptor 5 (mGluR5). J. Biol. Chem. 289, 28460–28477 to cellular prion protein. Neuron 79, 887–902 (2013).
(2014). 37. F. H. Beraldo, V. G. Ostapchenko, F. A. Caetano, A. L. Guimaraes, G. D. Ferretti, N. Daude,
24. B. R. Fluharty, E. Biasini, M. Stravalaci, A. Sclip, L. Diomede, C. Balducci, P. La Vitola, L. Bertram, K. O. Nogueira, J. L. Silva, D. Westaway, N. R. Cashman, V. R. Martins,
M. Messa, L. Colombo, G. Forloni, T. Borsello, M. Gobbi, D. A. Harris, An N-terminal V. F. Prado, M. A. Prado, Regulation of amyloid  oligomer binding to neurons
fragment of the prion protein binds to Amyloid- oligomers and inhibits their and neurotoxicity by the prion protein-mGluR5 complex. J. Biol. Chem. 291, 21945–21955
neurotoxicity in vivo. J. Biol. Chem. 288, 7857–7866 (2013). (2016).
25. L. T. Haas, S. V. Salazar, M. A. Kostylev, J. W. Um, A. C. Kaufman, S. M. Strittmatter, 38. L. T. Haas, S. M. Strittmatter, Oligomers of amyloid  prevent physiological activation
Metabotropic glutamate receptor 5 couples cellular prion protein to intracellular of the cellular prion protein-metabotropic glutamate receptor 5 complex by glutamate
signalling in Alzheimer's disease. Brain 139, 526–546 (2016). in Alzheimer disease. J. Biol. Chem. 291, 17112–17121 (2016).
26. S. V. Salazar, C. Gallardo, A. C. Kaufman, C. S. Herber, L. T. Haas, S. Robinson, J. C. Manson, 39. L. T. Haas, S. V. Salazar, L. M. Smith, H. R. Zhao, T. O. Cox, C. S. Herber, A. P. Degnan,
M. K. Lee, S. M. Strittmatter, Conditional deletion of Prnp rescues behavioral and synaptic A. Balakrishnan, J. E. Macor, C. F. Albright, S. M. Strittmatter, Silent allosteric modulation

Downloaded from https://www.science.org at Yale University on June 14, 2022


deficits after disease onset in transgenic Alzheimer's disease. J. Neurosci. 37, 9207–9221 of mGluR5 maintains glutamate signaling while rescuing Alzheimer's mouse phenotypes.
(2017). Cell Rep. 20, 76–88 (2017).
27. D. A. Gimbel, H. B. Nygaard, E. E. Coffey, E. C. Gunther, J. Lauren, Z. A. Gimbel, 40. A. Hamilton, M. Vasefi, C. Vander Tuin, R. J. McQuaid, H. Anisman, S. S. Ferguson, Chronic
S. M. Strittmatter, Memory impairment in transgenic Alzheimer mice requires cellular pharmacological mGluR5 inhibition prevents cognitive impairment and reduces
prion protein. J. Neurosci. 30, 6367–6374 (2010). pathogenesis in an Alzheimer disease mouse model. Cell Rep. 15, 1859–1865 (2016).
28. G. T. Corbett, Z. Wang, W. Hong, M. Colom-Cadena, J. Rose, M. Liao, A. Asfaw, T. C. Hall, 41. A. Hamilton, J. L. Esseltine, R. A. DeVries, S. P. Cregan, S. S. Ferguson, Metabotropic
L. Ding, A. DeSousa, M. P. Frosch, J. Collinge, D. A. Harris, M. S. Perkinton, T. L. Spires-Jones, glutamate receptor 5 knockout reduces cognitive impairment and pathogenesis
T. L. Young-Pearse, A. Billinton, D. M. Walsh, PrP is a central player in toxicity mediated by in a mouse model of Alzheimer's disease. Mol. Brain 7, 40 (2014).
soluble aggregates of neurodegeneration-causing proteins. Acta Neuropathol. 139, 42. N.-W. Hu, A. J. Nicoll, D. Zhang, A. J. Mably, T. O'Malley, S. A. Purro, C. Terry, J. Collinge,
503–526 (2020). D. M. Walsh, M. J. Rowan, mGlu5 receptors and cellular prion protein mediate
29. S. A. Purro, A. J. Nicoll, J. Collinge, Prion protein as a toxic acceptor of Amyloid- amyloid--facilitated synaptic long-term depression in vivo. Nat. Commun. 5, 3374 (2014).
oligomers. Biol. Psychiatry 83, 358–368 (2018). 43. D. Zhang, Y. Qi, I. Klyubin, T. Ondrejcak, C. J. Sarell, A. C. Cuello, J. Collinge, M. J. Rowan,
30. G. Lee, S. T. Newman, D. L. Gard, H. Band, G. Panchamoorthy, Tau interacts with src-family Targeting glutamatergic and cellular prion protein mechanisms of amyloid -mediated
non-receptor tyrosine kinases. J. Cell Sci. 111(Pt. 21), 3167–3177 (1998). persistent synaptic plasticity disruption: Longitudinal studies. Neuropharmacology 121,
31. L. M. Ittner, Y. D. Ke, F. Delerue, M. Bi, A. Gladbach, J. van Eersel, H. Wolfing, B. C. Chieng, 231–246 (2017).
M. J. Christie, I. A. Napier, A. Eckert, M. Staufenbiel, E. Hardeman, J. Gotz, Dendritic 44. S. Lee, S. V. Salazar, T. O. Cox, S. M. Strittmatter, Pyk2 signaling through Graf1 and RhoA
function of tau mediates amyloid-beta toxicity in Alzheimer's disease mouse models. Cell GTPase is required for amyloid- oligomer-triggered synapse loss. J. Neurosci. 39,
142, 387–397 (2010). 1910–1929 (2019).
32. E. D. Roberson, B. Halabisky, J. W. Yoo, J. Yao, J. Chin, F. Yan, T. Wu, P. Hamto, N. Devidze, 45. S. V. Salazar, T. O. Cox, S. Lee, A. H. Brody, A. S. Chyung, L. T. Haas, S. M. Strittmatter,
G. Q. Yu, J. J. Palop, J. L. Noebels, L. Mucke, Amyloid-/Fyn-Induced synaptic, network, Alzheimer's disease risk factor Pyk2 mediates amyloid--induced synaptic dysfunction
and cognitive impairments depend on tau levels in multiple mouse models and loss. J. Neurosci. 39, 758–772 (2019).
of Alzheimer's disease. J. Neurosci. 31, 700–711 (2011). 46. Q. Wang, D. M. Walsh, M. J. Rowan, D. J. Selkoe, R. Anwyl, Block of long-term potentiation
33. M. Larson, M. A. Sherman, F. Amar, M. Nuvolone, J. A. Schneider, D. A. Bennett, A. Aguzzi, by naturally secreted and synthetic amyloid beta-peptide in hippocampal slices is
S. E. Lesne, The complex PrPc-Fyn couples human oligomeric A with pathological tau mediated via activation of the kinases c-Jun N-terminal kinase, cyclin-dependent kinase
changes in Alzheimer's disease. J. Neurosci. 32, 16857–16871 (2012). 5, and p38 mitogen-activated protein kinase as well as metabotropic glutamate receptor
34. A. C. Kaufman, S. V. Salazar, L. T. Haas, J. Yang, M. A. Kostylev, A. T. Jeng, S. A. Robinson, type 5. J. Neurosci. 24, 3370–3378 (2004).
E. C. Gunther, C. H. van Dyck, H. B. Nygaard, S. M. Strittmatter, Fyn inhibition rescues 47. M. Renner, P. N. Lacor, P. T. Velasco, J. Xu, A. Contractor, W. L. Klein, A. Triller, Deleterious
established memory and synapse loss in Alzheimer mice. Ann. Neurol. 77, 953–971 (2015). effects of amyloid beta oligomers acting as an extracellular scaffold for mGluR5. Neuron
35. J. C. Lambert, C. A. Ibrahim-Verbaas, D. Harold, A. C. Naj, R. Sims, C. Bellenguez, 66, 739–754 (2010).
A. L. DeStafano, J. C. Bis, G. W. Beecham, B. Grenier-Boley, G. Russo, T. A. Thorton-Wells, 48. H. Zhang, L. Wu, E. Pchitskaya, O. Zakharova, T. Saito, T. Saido, I. Bezprozvanny, Neuronal
N. Jones, A. V. Smith, V. Chouraki, C. Thomas, M. A. Ikram, D. Zelenika, B. N. Vardarajan, store-operated calcium entry and mushroom spine loss in amyloid precursor protein
Y. Kamatani, C. F. Lin, A. Gerrish, H. Schmidt, B. Kunkle, M. L. Dunstan, A. Ruiz, knock-in mouse model of Alzheimer's disease. J. Neurosci. 35, 13275–13286 (2015).
M. T. Bihoreau, S. H. Choi, C. Reitz, F. Pasquier, C. Cruchaga, D. Craig, N. Amin, C. Berr, 49. F. Raka, A. R. Di Sebastiano, S. C. Kulhawy, F. M. Ribeiro, C. M. Godin, F. A. Caetano,
O. L. Lopez, P. L. De Jager, V. Deramecourt, J. A. Johnston, D. Evans, S. Lovestone, S. Angers, S. S. Ferguson, Ca2+/Calmodulin-dependent protein kinase II interacts
L. Letenneur, F. J. Moron, D. C. Rubinsztein, G. Eiriksdottir, K. Sleegers, A. M. Goate, with group I metabotropic glutamate and facilitates receptor endocytosis and ERK1/2
N. Fievet, M. W. Huentelman, M. Gill, K. Brown, M. I. Kamboh, L. Keller, P. Barberger-Gateau, signaling: Role of -amyloid. Mol. Brain 8, 21 (2015).
B. McGuiness, E. B. Larson, R. Green, A. J. Myers, C. Dufouil, S. Todd, D. Wallon, S. Love, 50. C. R. Overk, A. Cartier, G. Shaked, E. Rockenstein, K. Ubhi, B. Spencer, D. L. Price, C. Patrick,
E. Rogaeva, J. Gallacher, P. S. George-Hyslop, J. Clarimon, A. Lleo, A. Bayer, D. W. Tsuang, P. Desplats, E. Masliah, Hippocampal neuronal cells that accumulate -synuclein
L. Yu, M. Tsolaki, P. Bossu, G. Spalletta, P. Proitsi, J. Collinge, S. Sorbi, F. Sanchez-Garcia, fragments are more vulnerable to A oligomer toxicity via mGluR5–implications
N. C. Fox, J. Hardy, M. C. D. Naranjo, P. Bosco, R. Clarke, C. Brayne, D. Galimberti, for dementia with Lewy bodies. Mol. Neurodegener. 9, 18 (2014).
M. Mancuso, F. Matthews; European Alzheimer's Disease Initiative (EADI); Genetic and 51. Y. He, M. Wei, Y. Wu, H. Qin, W. Li, X. Ma, J. Cheng, J. Ren, Y. Shen, Z. Chen, B. Sun,
Environmental Risk in Alzheimer's Disease; Alzheimer's Disease Genetic Consortium; F. D. Huang, Y. Shen, Y. D. Zhou, Amyloid  oligomers suppress excitatory transmitter

Spurrier et al., Sci. Transl. Med. 14, eabi8593 (2022) 1 June 2022 14 of 16
SCIENCE TRANSLATIONAL MEDICINE | RESEARCH ARTICLE

release via presynaptic depletion of phosphatidylinositol-4,5-bisphosphate. Nat. 71. M. A. Busche, B. T. Hyman, Synergy between amyloid- and tau in Alzheimer's disease.
Commun. 10, 1193 (2019). Nat. Neurosci. 23, 1183–1193 (2020).
52. K. S. Abd-Elrahman, A. Hamilton, A. Albaker, S. S. G. Ferguson, mGluR5 contribution 72. S. Mondragon-Rodriguez, G. Perry, J. Luna-Munoz, M. C. Acevedo-Aquino, S. Williams,
to neuropathology in Alzheimer mice is disease stage-dependent. ACS Pharmacol. Phosphorylation of tau protein at sites Ser(396-404) is one of the earliest events
Transl. Sci. 3, 334–344 (2020). in Alzheimer's disease and Down syndrome. Neuropathol. Appl. Neurobiol. 40, 121–135
53. S. F. Kazim, S. C. Chuang, W. Zhao, R. K. Wong, R. Bianchi, K. Iqbal, Early-onset network (2014).
hyperexcitability in presymptomatic Alzheimer's disease transgenic mice is suppressed 73. S. Palmqvist, S. Janelidze, Y. T. Quiroz, H. Zetterberg, F. Lopera, E. Stomrud, Y. Su, Y. Chen,
by passive immunization with anti-human APP/A antibody and by mGluR5 blockade. G. E. Serrano, A. Leuzy, N. Mattsson-Carlgren, O. Strandberg, R. Smith, A. Villegas,
Front. Aging Neurosci. 9, 71 (2017). D. Sepulveda-Falla, X. Chai, N. K. Proctor, T. G. Beach, K. Blennow, J. L. Dage, E. M. Reiman,
54. K. S. Abd-Elrahman, A. Albaker, J. M. de Souza, F. M. Ribeiro, M. G. Schlossmacher, O. Hansson, Discriminative accuracy of plasma phospho-tau217 for Alzheimer disease vs
M. Tiberi, A. Hamilton, S. S. G. Ferguson, A oligomers induce pathophysiological mGluR5 other neurodegenerative disorders. JAMA 324, 772–781 (2020).
signaling in Alzheimer's disease model mice in a sex-selective manner. Sci. Signal. 13, 74. G. Bindea, B. Mlecnik, H. Hackl, P. Charoentong, M. Tosolini, A. Kirilovsky, W. H. Fridman,
eabd2494 (2020). F. Pages, Z. Trajanoski, J. Galon, ClueGO: A Cytoscape plug-in to decipher functionally
55. R. H. Porter, G. Jaeschke, W. Spooren, T. M. Ballard, B. Buttelmann, S. Kolczewski, grouped gene ontology and pathway annotation networks. Bioinformatics 25,
J. U. Peters, E. Prinssen, J. Wichmann, E. Vieira, A. Muhlemann, S. Gatti, V. Mutel, 1091–1093 (2009).
P. Malherbe, Fenobam: A clinically validated nonbenzodiazepine anxiolytic is a potent, 75. H. Keren-Shaul, A. Spinrad, A. Weiner, O. Matcovitch-Natan, R. Dvir-Szternfeld,
selective, and noncompetitive mGlu5 receptor antagonist with inverse agonist activity. T. K. Ulland, E. David, K. Baruch, D. Lara-Astaiso, B. Toth, S. Itzkovitz, M. Colonna,
J. Pharmacol. Exp. Ther. 315, 711–721 (2005). M. Schwartz, I. Amit, A unique microglia type associated with restricting development
56. A. L. Rodriguez, M. D. Grier, C. K. Jones, E. J. Herman, A. S. Kane, R. L. Smith, R. Williams, of Alzheimer's disease. Cell 169, 1276–1290.e17 (2017).
Y. Zhou, J. E. Marlo, E. L. Days, T. N. Blatt, S. Jadhav, U. N. Menon, P. N. Vinson, J. M. Rook, 76. N. Habib, C. McCabe, S. Medina, M. Varshavsky, D. Kitsberg, R. Dvir-Szternfeld, G. Green,
S. R. Stauffer, C. M. Niswender, C. W. Lindsley, C. D. Weaver, P. J. Conn, Discovery of novel D. Dionne, L. Nguyen, J. L. Marshall, F. Chen, F. Zhang, T. Kaplan, A. Regev, M. Schwartz,
allosteric modulators of metabotropic glutamate receptor subtype 5 reveals chemical Disease-associated astrocytes in Alzheimer's disease and aging. Nat. Neurosci. 23,
and functional diversity and in vivo activity in rat behavioral models of anxiolytic 701–706 (2020).
and antipsychotic activity. Mol. Pharmacol. 78, 1105–1123 (2010). 77. T. Iram, Z. Ramirez-Ortiz, M. H. Byrne, U. A. Coleman, N. D. Kingery, T. K. Means, D. Frenkel,
57. U. C. Campbell, K. Lalwani, L. Hernandez, G. G. Kinney, P. J. Conn, L. J. Bristow, The J. El Khoury, Megf10 is a receptor for C1Q that mediates clearance of apoptotic cells by
mGluR5 antagonist 2-methyl-6-(phenylethynyl)-pyridine (MPEP) potentiates astrocytes. J. Neurosci. 36, 5185–5192 (2016).
PCP-induced cognitive deficits in rats. Psychopharmacology 175, 310–318 (2004). 78. C. M. Henstridge, M. Tzioras, R. C. Paolicelli, Glial contribution to excitatory and inhibitory
58. K. Abou Farha, R. Bruggeman, C. Balje-Volkers, Metabotropic glutamate receptor 5 synapse loss in neurodegeneration. Front. Cell. Neurosci. 13, 63 (2019).
negative modulation in phase I clinical trial: Potential impact of circadian rhythm 79. T. O. Cox, E. C. Gunther, A. H. Brody, M. T. Chiasseu, A. Stoner, L. M. Smith, L. T. Haas,

Downloaded from https://www.science.org at Yale University on June 14, 2022


on the neuropsychiatric adverse reactions-do hallucinations matter? ISRN Psychiatry J. Hammersley, G. Rees, B. Dosanjh, M. Groves, M. Gardener, C. Dobson, T. Vaughan,
2014, 652750 (2014). I. Chessell, A. Billinton, S. M. Strittmatter, Anti-PrPC antibody rescues cognition
59. Y. M. Lu, Z. Jia, C. Janus, J. T. Henderson, R. Gerlai, J. M. Wojtowicz, J. C. Roder, Mice and synapses in transgenic alzheimer mice. Ann. Clin. Transl. Neurol. 6, 554–574 (2019).
lacking metabotropic glutamate receptor 5 show impaired learning and reduced CA1 80. E. C. Gunther, L. M. Smith, M. A. Kostylev, T. O. Cox, A. C. Kaufman, S. Lee, E. Folta-Stogniew,
long-term potentiation (LTP) but normal CA3 LTP. J. Neurosci. 17, 5196–5205 (1997). G. D. Maynard, J. W. Um, M. Stagi, J. K. Heiss, A. Stoner, G. P. Noble, H. Takahashi,
60. J. Xu, Y. Zhu, A. Contractor, S. F. Heinemann, mGluR5 has a critical role in inhibitory L. T. Haas, J. S. Schneekloth, J. Merkel, C. Teran, Z. K. Naderi, S. Supattapone,
learning. J. Neurosci. 29, 3676–3684 (2009). S. M. Strittmatter, Rescue of transgenic Alzheimer's pathophysiology by polymeric
61. K. J. Gregory, E. N. Dong, J. Meiler, P. J. Conn, Allosteric modulation of metabotropic cellular prion protein antagonists. Cell Rep. 26, 145–158.e8 (2019).
glutamate receptors: Structural insights and therapeutic potential. Neuropharmacology 81. H. Lui, J. Zhang, S. R. Makinson, M. K. Cahill, K. W. Kelley, H. Y. Huang, Y. Shang,
60, 66–81 (2011). M. C. Oldham, L. H. Martens, F. Gao, G. Coppola, S. A. Sloan, C. L. Hsieh, C. C. Kim,
62. K. J. Gregory, C. Malosh, M. Turlington, R. Morrison, P. Vinson, J. S. Daniels, C. Jones, E. H. Bigio, S. Weintraub, M. M. Mesulam, R. Rademakers, I. R. Mackenzie, W. W. Seeley,
C. M. Niswender, P. J. Conn, C. W. Lindsley, S. R. Stauffer, in Probe Reports from the NIH A. Karydas, B. L. Miller, B. Borroni, R. Ghidoni, R. V. Farese Jr., J. T. Paz, B. A. Barres,
Molecular Libraries Program (National Center for Biotechnology Information, 2010). E. J. Huang, Progranulin deficiency promotes circuit-specific synaptic pruning by
63. K. J. Gregory, M. J. Noetzel, J. M. Rook, P. N. Vinson, S. R. Stauffer, A. L. Rodriguez, microglia via complement activation. Cell 165, 921–935 (2016).
K. A. Emmitte, Y. Zhou, A. C. Chun, A. S. Felts, B. A. Chauder, C. W. Lindsley, 82. B. Dejanovic, M. A. Huntley, A. De Maziere, W. J. Meilandt, T. Wu, K. Srinivasan, Z. Jiang,
C. M. Niswender, P. J. Conn, Investigating metabotropic glutamate receptor 5 allosteric V. Gandham, B. A. Friedman, H. Ngu, O. Foreman, R. A. D. Carano, B. Chih, J. Klumperman,
modulator cooperativity, affinity, and agonism: Enriching structure-function studies C. Bakalarski, J. E. Hanson, M. Sheng, Changes in the synaptic proteome in tauopathy
and structure-activity relationships. Mol. Pharmacol. 82, 860–875 (2012). and rescue of Tau-induced synapse loss by C1q antibodies. Neuron 100, 1322–1336.e7
64. T. Saito, Y. Matsuba, N. Mihira, J. Takano, P. Nilsson, S. Itohara, N. Iwata, T. C. Saido, Single (2018).
App knock-in mouse models of Alzheimer's disease. Nat. Neurosci. 17, 661–663 (2014). 83. E. C. Damisah, R. A. Hill, A. Rai, F. Chen, C. V. Rothlin, S. Ghosh, J. Grutzendler, Astrocytes
65. T. Saito, N. Mihira, Y. Matsuba, H. Sasaguri, S. Hashimoto, S. Narasimhan, B. Zhang, and microglia play orchestrated roles and respect phagocytic territories during neuronal
S. Murayama, M. Higuchi, V. M. Y. Lee, J. Q. Trojanowski, T. C. Saido, Humanization corpse removal in vivo. Sci. Adv. 6, eaba3239 (2020).
of the entire murine Mapt gene provides a murine model of pathological human tau 84. M. Naganawa, S. Li, N. Nabulsi, S. Henry, M. Q. Zheng, R. Pracitto, Z. Cai, H. Gao,
propagation. J. Biol. Chem. 294, 12754–12765 (2019). M. Kapinos, D. Labaree, D. Matuskey, Y. Huang, R. E. Carson, First-in-human evaluation
66. B. Stevens, N. J. Allen, L. E. Vazquez, G. R. Howell, K. S. Christopherson, N. Nouri, of 18F-SynVesT-1, a radioligand for PET imaging of synaptic vesicle glycoprotein 2A.
K. D. Micheva, A. K. Mehalow, A. D. Huberman, B. Stafford, A. Sher, A. M. Litke, J. Nucl. Med. 62, 561–567 (2021).
J. D. Lambris, S. J. Smith, S. W. John, B. A. Barres, The classical complement cascade 85. S. Li, Z. Cai, X. Wu, D. Holden, R. Pracitto, M. Kapinos, H. Gao, D. Labaree, N. Nabulsi,
mediates CNS synapse elimination. Cell 131, 1164–1178 (2007). R. E. Carson, Y. Huang, Synthesis and in vivo evaluation of a novel PET radiotracer for
67. F. Yang, L. B. Snyder, A. Balakrishnan, J. M. Brown, D. V. Sivarao, A. Easton, A. Fernandes, imaging of synaptic vesicle glycoprotein 2A (SV2A) in nonhuman primates. ACS Chem. Nerosci.
M. Gulianello, U. M. Hanumegowda, H. Huang, Y. Huang, K. M. Jones, Y. W. Li, 10, 1544–1554 (2019).
M. Matchett, G. Mattson, R. Miller, K. S. Santone, A. Senapati, E. E. Shields, F. J. Simutis, 86. R. Mostany, J. E. Anstey, K. L. Crump, B. Maco, G. Knott, C. Portera-Cailliau, Altered
R. Westphal, V. J. Whiterock, J. J. Bronson, J. E. Macor, A. P. Degnan, Discovery synaptic dynamics during normal brain aging. J. Neurosci. 33, 4094–4104 (2013).
and preclinical evaluation of BMS-955829, a potent positive allosteric modulator 87. J. K. Heiss, J. Barrett, Z. Yu, L. T. Haas, M. A. Kostylev, S. M. Strittmatter, Early activation
of mGluR5. ACS Med. Chem. Lett. 7, 289–293 (2016). of experience-independent dendritic spine turnover in a mouse model of Alzheimer's
68. A. D. Sauerbeck, M. Gangolli, S. J. Reitz, M. H. Salyards, S. H. Kim, C. Hemingway, disease. Cereb. Cortex 27, 3660–3674 (2017).
M. Gratuze, T. Makkapati, M. Kerschensteiner, D. M. Holtzman, D. L. Brody, T. T. Kummer, 88. S. Oddo, A. Caccamo, J. D. Shepherd, M. P. Murphy, T. E. Golde, R. Kayed, R. Metherate,
SEQUIN multiscale imaging of mammalian central synapses reveals loss of synaptic M. P. Mattson, Y. Akbari, F. M. LaFerla, Triple-transgenic model of Alzheimer's disease
connectivity resulting from diffuse traumatic brain injury. Neuron 107, 257–273.e5 (2020). with plaques and Tangles. Neuron 39, 409–421 (2003).
69. L. Mucke, D. J. Selkoe, Neurotoxicity of amyloid -protein: Synaptic and network 89. C. Li, J. Gotz, Pyk2 is a novel Tau tyrosine kinase that is regulated by the tyrosine kinase
dysfunction. Cold Spring Harb. Perspect. Med. 2, a006338 (2012). Fyn. J. Alzheimers Dis. 64, 205–221 (2018).
70. S. Tu, S. Okamoto, S. A. Lipton, H. Xu, Oligomeric A-induced synaptic dysfunction 90. H. Mathys, J. Davila-Velderrain, Z. Peng, F. Gao, S. Mohammadi, J. Z. Young, M. Menon,
in Alzheimer's disease. Mol. Neurodegener. 9, 48 (2014). L. He, F. Abdurrob, X. Jiang, A. J. Martorell, R. M. Ransohoff, B. P. Hafler, D. A. Bennett,

Spurrier et al., Sci. Transl. Med. 14, eabi8593 (2022) 1 June 2022 15 of 16
SCIENCE TRANSLATIONAL MEDICINE | RESEARCH ARTICLE

M. Kellis, L. H. Tsai, Single-cell transcriptomic analysis of Alzheimer's disease. Nature 570, allosteric modulators of mGluR5: Defining molecular switches to create
332–337 (2019). a pharmacological tool box. Bioorg. Med. Chem. Lett. 26, 4165–4169 (2016).
91. W. M. Brooks, P. J. Lynch, C. C. Ingle, A. Hatton, P. C. Emson, R. L. Faull, M. P. Starkey, Gene 09.
1 G. M. Sheldrick, A short history of SHELX. Acta Crystallogr. A 64, 112–122 (2008).
expression profiles of metabolic enzyme transcripts in Alzheimer's disease. Brain Res. 110. A. A. Bachmanov, D. R. Reed, G. K. Beauchamp, M. G. Tordoff, Food intake, water intake,
1127, 127–135 (2007). and drinking spout side preference of 28 mouse strains. Behav. Genet. 32, 435–443 (2002).
92. M. Zhou, R. U. Haque, E. B. Dammer, D. M. Duong, L. Ping, E. C. B. Johnson, J. J. Lah, 111. K. Lim, D. Labaree, S. Li, Y. Huang, Preparation of the metabotropic glutamate receptor 5
A. I. Levey, N. T. Seyfried, Targeted mass spectrometry to quantify brain-derived (mGluR5) PET tracer [(18)F]FPEB for human use: An automated radiosynthesis and a novel
cerebrospinal fluid biomarkers in Alzheimer's disease. Clin. Proteomics 17, 19 (2020). one-pot synthesis of its radiolabeling precursor. Appl. Radiat. Isot. 94, 349–354 (2014).
93. W. Cong, X. Meng, J. Li, Q. Zhang, F. Chen, W. Liu, Y. Wang, S. Cheng, X. Yao, J. Yan, S. Kim, 112. C. M. Sandiego, N. Nabulsi, S. F. Lin, D. Labaree, S. Najafzadeh, Y. Huang, K. Cosgrove,
A. J. Saykin, H. Liang, L. Shen; Alzheimer’s Disease Neuroimaging Initiative, Genome-wide R. E. Carson, Studies of the metabotropic glutamate receptor 5 radioligand [11C]ABP688
network-based pathway analysis of CSF t-tau/A1-42 ratio in the ADNI cohort. BMC with N-acetylcysteine challenge in rhesus monkeys. Synapse 67, 489–501 (2013).
Genomics 18, 421 (2017). 113. R. B. Innis, V. J. Cunningham, J. Delforge, M. Fujita, A. Gjedde, R. N. Gunn, J. Holden,
94. J. Sundelof, J. Sundstrom, O. Hansson, M. Eriksdotter-Jonhagen, V. Giedraitis, A. Larsson, S. Houle, S. C. Huang, M. Ichise, H. Iida, H. Ito, Y. Kimura, R. A. Koeppe, G. M. Knudsen,
M. Degerman-Gunnarsson, M. Ingelsson, L. Minthon, K. Blennow, L. Kilander, H. Basun, J. Knuuti, A. A. Lammertsma, M. Laruelle, J. Logan, R. P. Maguire, M. A. Mintun,
L. Lannfelt, Cystatin C levels are positively correlated with both Abeta42 and tau levels E. D. Morris, R. Parsey, J. C. Price, M. Slifstein, V. Sossi, T. Suhara, J. R. Votaw, D. F. Wong,
in cerebrospinal fluid in persons with Alzheimer's disease, mild cognitive impairment, R. E. Carson, Consensus nomenclature for in vivo imaging of reversibly binding
and healthy controls. J. Alzheimers Dis. 21, 471–478 (2010). radioligands. J. Cereb. Blood Flow Metab. 27, 1533–1539 (2007).
95. S. F. Hansson, U. Andreasson, M. Wall, I. Skoog, N. Andreasen, A. Wallin, H. Zetterberg, 114. V. J. Cunningham, E. A. Rabiner, M. Slifstein, M. Laruelle, R. N. Gunn, Measuring drug
K. Blennow, Reduced levels of amyloid-beta-binding proteins in cerebrospinal fluid occupancy in the absence of a reference region: The Lassen plot re-visited. J. Cereb. Blood
from Alzheimer's disease patients. J. Alzheimers Dis. 16, 389–397 (2009). Flow Metab. 30, 46–50 (2010).
96. Q. Zhang, C. Ma, M. Gearing, P. G. Wang, L. S. Chin, L. Li, Integrated proteomics 115. Y. Ma, D. Smith, P. R. Hof, B. Foerster, S. Hamilton, S. J. Blackband, M. Yu, H. Benveniste,
and network analysis identifies protein hubs and network alterations in Alzheimer's In vivo 3D digital atlas database of the adult C57BL/6J mouse brain by magnetic
disease. Acta Neuropathol. Commun. 6, 19 (2018). resonance microscopy. Front. Neuroanat. 2, 1 (2008).
97. L. Higginbotham, L. Ping, E. B. Dammer, D. M. Duong, M. Zhou, M. Gearing, C. Hurst, 116. J. Verhaeghe, D. Bertoglio, L. Kosten, D. Thomae, M. Verhoye, A. Van Der Linden,
J. D. Glass, S. A. Factor, E. C. B. Johnson, I. Hajjar, J. J. Lah, A. I. Levey, N. T. Seyfried, L. Wyffels, S. Stroobants, J. Wityak, C. Dominguez, L. Mrzljak, S. Staelens, Noninvasive
Integrated proteomics reveals brain-based cerebrospinal fluid biomarkers relative quantification of [11C]ABP688 PET imaging in mice versus an input function
in asymptomatic and symptomatic Alzheimer's disease. Sci. Adv. 6, (2020). measured over an arteriovenous shunt. Front. Neurol. 9, 516 (2018).
98. U. V. Mahajan, V. R. Varma, M. E. Griswold, C. T. Blackshear, Y. An, A. M. Oommen, 117. R. J. Racine, Modification of seizure activity by electrical stimulation: II. Motor seizure.
S. Varma, J. C. Troncoso, O. Pletnikova, R. O'Brien, T. J. Hohman, C. Legido-Quigley, Electroencephalogr. Clin. Neurophysiol. 32, 281–294 (1972).
118. S. Bolte, F. P. Cordelieres, A guided tour into subcellular colocalization analysis in light

Downloaded from https://www.science.org at Yale University on June 14, 2022


M. Thambisetty, Dysregulation of multiple metabolic networks related to brain
transmethylation and polyamine pathways in Alzheimer disease: A targeted microscopy. J. Microsc. 224, 213–232 (2006).
metabolomic and transcriptomic study. PLOS Med. 17, e1003012 (2020). 119. C. S. McGinnis, L. M. Murrow, Z. J. Gartner, DoubletFinder: Doublet detection in single-cell
99. W. Yao, T. Yin, M. D. Tambini, L. D'Adamio, The familial dementia gene ITM2b/BRI2 RNA sequencing data using artificial nearest neighbors. Cell Syst. 8, 329–337.e4 (2019).
facilitates glutamate transmission via both presynaptic and postsynaptic mechanisms. 120. H.  Wickham, in Use R! (Springer International Publishing, 2016).
Sci. Rep. 9, 4862 (2019).
100. M. Del Campo, C. E. Teunissen, Role of BRI2 in dementia. J. Alzheimers Dis. 40, 481–494 Acknowledgments: We thank T. Saito and T. C. Saido for APPNL-G-F/hMAPT mice. We thank
(2014). K. DeLuca and the Yale PET Center staff for their expert technical support. Funding: This
101. M. Del Campo, J. J. Hoozemans, L. L. Dekkers, A. J. Rozemuller, C. Korth, A. Muller-Schiffmann, work was supported by research grants from the NIH to Z.C. (R01AG069921), to S.M.S.
P. Scheltens, M. A. Blankenstein, C. R. Jimenez, R. Veerhuis, C. E. Teunissen, BRI2-BRICHOS (U01AG058608, R01AG034924, R01AG066165, RF1AG070926, and R56AG074015), and to
is increased in human amyloid plaques in early stages of Alzheimer's disease. Neurobiol. A.H.B. (F31AG066483, T32NS007224, and T32NS041228), and by the Thome Memorial
Aging 35, 1596–1604 (2014). Foundation and the Falk Medical Research Trust to S.M.S. Author contributions: J.S. designed
102. L. C. Reese, F. Laezza, R. Woltjer, G. Taglialatela, Dysregulated phosphorylation experiments, wrote the first draft of the manuscript, and completed the mouse dosing, tissue
of Ca2+/calmodulin-dependent protein kinase II- in the hippocampus of subjects handling, and snRNA-seq. L.N. and L.Z. processed the snRNA-seq data. A.J.S., W.L., M.A.A., and
with mild cognitive impairment and Alzheimer's disease. J. Neurochem. 119, M.C. completed histological assessments of brain. J.S. and M.C. completed mouse behavior.
791–804 (2011). X.T.F., T.T., D.H., P.S., S. Lee, and S. Li completed SynVesT-1 PET and mouse FPEB studies. D.H.
103. T. Wyss-Coray, J. Rogers, Inflammation in Alzheimer disease-a brief review of the basic completed monkey FPEB PET studies. T.R.S. supervised toxicological studies. A.H.B., H.T., S.H.N.,
science and clinical literature. Cold Spring Harb. Perspect. Med. 2, a006346 (2012). and A.P.C. contributed to histology and biochemistry. Y.H.H., R.E.C., and Z.C. supervised the
104. T. Wu, B. Dejanovic, V. D. Gandham, A. Gogineni, R. Edmonds, S. Schauer, K. Srinivasan, PET studies, whereas S.M.S. designed the study, wrote the paper, and supervised all aspects.
M. A. Huntley, Y. Wang, T. M. Wang, M. Hedehus, K. H. Barck, M. Stark, H. Ngu, O. Foreman, Competing interests: T.R.S. is an employee, and S.M.S. is founder, equity holder, director, and
W. J. Meilandt, J. Elstrott, M. C. Chang, D. V. Hansen, R. A. D. Carano, M. Sheng, paid consultant of Allyx Therapeutics, which seeks to develop mGluR5 compounds for AD and
J. E. Hanson, Complement C3 is activated in human AD brain and is required holds a license for BMS-984923. Y.H.H., Z.C., S. Li, and R.E.C. are inventors on the patent
for neurodegeneration in mouse models of amyloidosis and tauopathy. Cell Rep. 28, application (PCT/US2018/018388) submitted by Yale University that covers the production
2111–2123.e6 (2019). and use of the radioligand [18F]SynVesT-1 (formerly referred to as [18F]SDM-8). T.R.S. and S.M.S.
105. W. S. Chung, L. E. Clarke, G. X. Wang, B. K. Stafford, A. Sher, C. Chakraborty, J. Joung, are inventors on a planned patent application to be submitted jointly by Yale University and
L. C. Foo, A. Thompson, C. Chen, S. J. Smith, B. A. Barres, Astrocytes mediate synapse Allyx Therapeutics that covers the production and use of BMS-984923 for neurodegenerative
elimination through MEGF10 and MERTK pathways. Nature 504, 394–400 (2013). disease. S.M.S. also consults for ReNetX Bio and Proteowise on unrelated topics. Data and
106. E. K. Lehrman, D. K. Wilton, E. Y. Litvina, C. A. Welsh, S. T. Chang, A. Frouin, A. J. Walker, materials availability: All data associated with this study are present in the paper or the
M. D. Heller, H. Umemori, C. Chen, B. Stevens, CD47 protects synapses from excess Supplementary Materials. The snRNA-seq data are available from NCBI GEO database,
microglia-mediated pruning during development. Neuron 100, 120–134.e6 (2018). accession number GSE171095.
107. A. Butler, P. Hoffman, P. Smibert, E. Papalexi, R. Satija, Integrating single-cell
transcriptomic data across different conditions, technologies, and species. Nat. Submitted 5 April 2021
Biotechnol. 36, 411–420 (2018). Resubmitted 6 December 2021
108. H. Huang, A. P. Degnan, A. Balakrishnan, A. Easton, M. Gulianello, Y. Huang, M. Matchett, Accepted 25 March 2022
G. Mattson, R. Miller, K. S. Santone, A. Senapati, E. E. Shields, D. V. Sivarao, L. B. Snyder, Published 1 June 2022
R. Westphal, V. J. Whiterock, F. Yang, J. J. Bronson, J. E. Macor, Oxazolidinone-based 10.1126/scitranslmed.abi8593

Spurrier et al., Sci. Transl. Med. 14, eabi8593 (2022) 1 June 2022 16 of 16
Reversal of synapse loss in Alzheimer mouse models by targeting mGluR5 to
prevent synaptic tagging by C1Q
Joshua SpurrierLaShae NicholsonXiaotian T. FangAustin J. StonerTakuya ToyonagaDaniel HoldenTimothy R.
SiegertWilliam LairdMary Alice AllnuttMarius ChiasseuA. Harrison BrodyHideyuki TakahashiSarah Helena NiesAzucena
Pérez-CañamásPragalath SadasivamSupum LeeSongye LiLe ZhangYiyun H. HuangRichard E. CarsonZhengxin
CaiStephen M. Strittmatter

Sci. Transl. Med., 14 (647), eabi8593. • DOI: 10.1126/scitranslmed.abi8593

A silent modulator for AD


The loss of synapses observed in patients with Alzheimer’s disease (AD) contributes to the progressive cognitive
impairments. Although microglia activity has been shown to participate in synaptic loss, the mechanisms mediating
the process are not completely elucidated. Now, Spurrier et al. evaluated the role of metabotropic glutamate receptor
5 (mGluR5) on synaptic loss and showed that oral administration of an mGluR5 silent allosteric modulator (SAM)
restored synaptic density and reduced phosphorylated TAU accumulation in mouse models of AD. Mechanistically,
the treatment reversed gene expression changes induced in AD mice and prevented synaptic localization of the

Downloaded from https://www.science.org at Yale University on June 14, 2022


complement protein C1Q. The results suggest that SAMs targeting mGluR5 could be an effective approach for limiting
AD-related synaptic loss.

View the article online


https://www.science.org/doi/10.1126/scitranslmed.abi8593
Permissions
https://www.science.org/help/reprints-and-permissions

Use of this article is subject to the Terms of service

Science Translational Medicine (ISSN ) is published by the American Association for the Advancement of Science. 1200 New York Avenue
NW, Washington, DC 20005. The title Science Translational Medicine is a registered trademark of AAAS.
Copyright © 2022 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of Science. No claim
to original U.S. Government Works

View publication stats

You might also like