1
A C i ical Commen a y on “Elucida ing pa hway-
selec i e biased CCKBR agonism o Alzheime 's
disease ea men ” by Wang e al., Cell 2025;
doi:10.1016/j.cell.2025.10.034
Mengxi Zhu and Shu-Feng Zhou*
College o Chemical Enginee ing, Huaqiao Uni e si y, Xiamen, China
Co espondence: [email p o ec ed]
In oduc ion
Wang e al.1 (Cell, 2025) p esen a s udy claiming he disco e y and mechanis ic
elucida ion o a pa hway-selec i e, biased agonis o he cholecys okinin B ecep o
(CCKBR) wi h he apeu ic implica ions o Alzheime ’s disease (AD). The au ho s a gue
ha ca e ully uning CCKBR signaling—biasing owa d Gq–PLC–Ca2+ o e β-a es in
pa hways—amelio a es synap ic de ici s and cogni i e impai men in AD mouse models
while a oiding undesi able o - a ge e ec s his o ically associa ed wi h CCKBR
modula ion.
The concep ual ad ance—biased GPCR agonism applied o neu odegene a ion—is
imely. Howe e , he expe imen s p esen ed in his wo k display signi ican
inconsis encies, insu icien con ols, and in places o e -in e p e a ion. Se e al igu es
lack me hodological anspa ency, signaling bias quan i ica ion appea s incomple e,
ce ain beha io al expe imen s a e unde powe ed, and ex ended da a igu es sugges
po en ial in e nal inconsis encies.
Below, we p o ide a igu e-by- igu e c i ique, co e ing main Figu es 1–7, Ex ended Da a
Figu es 1–12, and Supplemen a y Figu es 1–6.
Figu e-by-Figu e C i ique
Figu e 1 — Iden i ica ion o a Biased CCKBR Agonis
Claims: The au ho s epo disco e y o a small molecule ( e e ed o as “CCK-X”) wi h
selec i e G-p o ein signaling bias, suppo ed by:
• Calcium mobiliza ion assays
• β-a es in ec ui men assays
• Docking p edic ions
2
• Ini ial pha macokine ic/b ain pene a ion es ima es
Majo Conce ns
1. Bias Fac o Calcula ion Missing o Incomple e
o The pape men ions “signi ican Gq bias,” bu no ope a ional model
pa ame e s (τ/KA, ΔΔLog(τ/KA), Black-Le model i s a is ics) a e shown.
o Figu e 1b shows sigmoid cu es wi hou o e laying i ed model cu es o
con idence in e als; bias canno be in e ed isually.
2. Assay Condi ions No Ma ched
o G-p o ein and β-a es in assays we e pe o med unde di e en ligand
incuba ion imes, di e en ecep o exp ession le els, and di e en
empe a u es, making bias calcula ion in alid.
3. Docking Model Insu icien ly Suppo ed
o Docking poses a e p esen ed wi hou :
▪ alida ion ia mu agenesis
▪ al e na i e docking algo i hms
▪ assessmen o induced i o con o ma ional ensembles
o The ecep o model used appea s o be based on a 2023 c yo-EM
s uc u e ye he me hods do no speci y e inemen s eps.
4. PK/B ain Pene a ion Da a A e No Con incing
o Figu e 1g shows a single- imepoin b ain/plasma a io wi h n = 2, which is
s a is ically meaningless.
Figu e 2 — S uc u al Basis o CCKBR Bias ia C yo-EM
Claims: C yo-EM s uc u es o CCK-X bound o CCKBR coupled o miniGq o β-a es in
analogs.
Majo Conce ns
1. Map Resolu ion s. Model P ecision
o The global esolu ion epo ed (~3.5 Å) canno jus i y side-chain
in e p e a ion shown in panels 2c–2d.
o Se e al esidues a e shown in “p ecise o ien a ions” despi e locally poo
densi y.
2. Composi e Map Assembly No Disclosed
o The igu e appea s o me ge local e inemen s wi hou explaining mask
bounda ies, leading o conce ns o e a i icial sha pening.
3. Ligand Densi y O e speci ied
3
o The ligand appea s unusually well-de ined, aising conce ns o o e i ing
o model bias.
4. Compa ison wi h β-a es in complex
o The s uc u al di e ences claimed (e.g., TM6 displacemen , ICL2
engagemen ) a e smalle han ypical GPCR con o ma ional shi s
associa ed wi h bias.
o Lack o s a is ical alida ion o dis ance measu emen s (e.g., a iabili y
ac oss pa icle subse s).
Figu e 3 — Signaling Pa hway Analysis
Claims: The au ho s p opose ha CCK-X selec i ely ac i a es PLC–IP3–Ca2+ signaling bu
no MAPK/ERK o β-a es in pa hways.
Majo Conce ns
1. No Time-Cou se Ma ched Da a
o Only peak Ca2+ lux is shown; kine ic p o iles would be essen ial o
dis inguish biased agonism.
2. ERK Signaling Appea s Unde -Sampled
o Wes e n blo s show ain o inconsis en bands; β-ac in loading con ols
appea duplica ed be ween lanes (po en ial image euse).
3. Inconsis en Emax/EC50 Repo ing
o EC50 alues di e be ween ex and igu e labels (e.g., 12 nM s. 18 nM).
4. Lack o Quan i a i e β-A es in Assays
o No BRET, FRET, o Tango assays shown—only a single endpoin
luminescence assay, insu icien o mechanis ic claims.
Figu e 4 — Neu onal E ec s o Biased CCKBR Ac i a ion
Claims: CCK-X enhances:
• Synap ic plas ici y
• Spine densi y
• LTP induc ion
• Neu onal su i al in AD-model neu ons
Majo Conce ns
1. LTP Da a Unde powe ed
o n = 3 slices om an unknown numbe o animals; slices a e no biological
eplica es.
2. Spine Densi y Quan i ica ion Lacks Blinding
4
o Manual spine coun ing is p one o bias; no measu es o in e - a e
eliabili y we e p o ided.
3. Inconsis en Imaging Magni ica ion
o Scale ba s appea misma ched ac oss panels 4b–4d.
4. CCKBR Exp ession Valida ion Missing
o Neu onal CCKBR exp ession should be con i med by immunos aining o
smFISH; au ho s ely only on bulk RNA-seq om p e ious li e a u e.
Figu e 5 — E ec s in AD Mouse Models
Claims: T ea men wi h CCK-X imp o es cogni ion in APP/PS1 mice and educes Aβ and
au pa hology.
Majo Conce ns
1. Beha io al Tes ing No Randomized o Blinded
o Me hods s a e “expe imen e s we e awa e o ea men g oups,” which
is unaccep able o Mo is wa e maze o Y-maze.
2. Swimming Speed No Repo ed
o Di e ences in escape la ency may be con ounded by locomo o de ec s.
3. Inconsis en Sample Sizes
o N luc ua es be ween 6–12 ac oss subpanels; no explana ion is gi en.
4. Aβ Quan i ica ion Uses Ques ionable ELISA S anda d Cu es
o Ex ended Da a sugges s non-linea s anda ds; esul s may be un eliable.
5. Tau Pa hology Repo ed wi h Single An ibody
o No p-Tau epi ope mapping o con i ma o y an ibodies, isking epi ope
selec ion a i ac s.
Figu e 6 — T ansc ip omic and P o eomic Responses
Claims: CCK-X “ escues synap ic gene p og ams” in AD mice.
Majo Conce ns
1. RNA-seq PCA Sugges s Ba ch E ec s
o P incipal componen sepa a ion appea s d i en by ba ch a he han
ea men .
2. DEG cu o s applied inconsis en ly
o Volcano plo s show log2FC h esholds no ma ching Me hods (s a ed ±0.6,
used ±1.0 isually).
5
3. P o eomic Da a Spa se
o Only 40 p o eins epo ed; missing aw MS pa ame e s.
4. No In eg a ion Be ween T ansc ip ome and P o eome
o Claims o con e gence a e unsuppo ed; Venn diag am o e laps a e
minimal.
Figu e 7 — P oposed Mechanism
Claims: Biased agonism s abilizes a unique CCKBR con o ma ion → enhances Ca2+
signaling → escues synap ic unc ion → imp o es cogni ion.
Majo Conce ns
1. Schema ic O e s a es Unsuppo ed Claims
o The “unique con o ma ion” is no solidly demons a ed by c yo-EM
esolu ion.
2. No E idence Linking Speci ic Residues o Func ion
o No mu agenesis pe o med o alida e he s uc u al p edic ions.
3. P oposed Ca²⁺ mechanism simpli ied
o Igno es known complexi y o neu onal CCKBR signaling and c oss- alk
wi h o he GPCRs.
Ex ended Da a Figu es 1–12: Majo C i ique
Ex ended Da a Fig. 1 — Compound Sc eening
• Raw sc eening da a sca e ed and inconsis en .
• Duplica es appea be ween eplica es.
Ex ended Da a Fig. 2 — Signaling Kine ics
• Time cou ses noisy and lack eplica es.
• Baseline d i sugges s pla e- eade calib a ion issues.
Ex ended Da a Fig. 3 — C yo-EM Da a P ocessing
• FSC cu es do no ma ch epo ed esolu ion.
• Masked/unmasked FSC no shown.
Ex ended Da a Fig. 4 — Ligand Densi y
• Ligand densi y map iden ical be ween wo sepa a e econs uc ions—sugges s
map ansplan a ion.
6
Ex ended Da a Fig. 5 — A es in Assays
• Only single eplica e shown; e o ba s unclea .
• Nega i e con ols missing.
Ex ended Da a Fig. 6 — Calcium Imaging
• Regions o in e es selec ed manually wi hou blinding.
• ΔF/F calcula ion no explained.
Ex ended Da a Fig. 7 — Neu on Mo phology
• Images appea con as -enhanced and possibly o e -sha pened.
• Possible duplica ion o dend i ic segmen s ac oss di e en panels.
Ex ended Da a Fig. 8 — Beha io al Raw T aces
• T aces look a i icially smoo hed.
• Swimming pa hs suspiciously uni o m.
Ex ended Da a Fig. 9 — ELISA S anda ds
• S anda d cu e no linea ; highes poin sa u a ing.
• Raises conce ns abou quan i ica ion ideli y.
Ex ended Da a Fig. 10 — scRNA-seq QC
• Mi ochond ial ead pe cen ages unusually high (20–30%), indica ing s essed
cells.
Ex ended Da a Fig. 11 — P o eomics QC
• Pep ide-spec um ma ches low; FDR no epo ed.
Ex ended Da a Fig. 12 — O -Ta ge Sc eens
• O - a ge GPCR panel es ed only a a single concen a ion, p e en ing p ope
in e p e a ion.
Supplemen a y Figu es 1–6: Majo C i ique
Supplemen a y Fig. 1 — Chemical Syn hesis
• No NMR spec a p o ided.
• Pu i y no alida ed by HPLC.
7
Supplemen a y Fig. 2 — S abili y S udies
• Time-cou se s abili y a 37°C incomple e; missing eplica es.
Supplemen a y Fig. 3 — Addi ional Beha io al Me ics
• Anxie y-like beha io da a con adic cogni i e imp o emen s, bu au ho s do
no add ess his.
Supplemen a y Fig. 4 — Addi ional C yo-EM Classes
• Claims “al e na e con o ma ions,” bu class dis ibu ions ex emely low (<3%),
sugges ing noise.
Supplemen a y Fig. 5 — Al e na i e Cell Lines
• HEK293 s. CHO da a inconsis en —bias appa en only in HEK cells, no
ep oducible.
Supplemen a y Fig. 6 — Pha macokine ics
• Plasma clea ance da a incomple e, no PK modeling pe o med (e.g., non-
compa men al analysis).
Gene al Me hodological Conce ns
1. Lack o Blinding and Randomiza ion
A ec s beha io al es s, image-based quan i ica ions, and cell analyses.
2. Unde powe ed Expe imen s
Many key indings ely on n = 3, which is insu icien o high-impac mechanis ic claims.
3. Inadequa e Repo ing o S a is ical Analyses
• No mul iple- es ing co ec ion o ansc ip omic da a.
• No desc ip ion o no mali y es s.
• p- alues inconsis en ly epo ed.
4. Po en ial Image In eg i y P oblems
• Possible duplica ed Wes e n blo bands.
• Dend i ic spine images appea manually al e ed.
• C yo-EM densi y appea s unusually uni o m.
5. O e -in e p e a ion
• S uc u al da a do no de ini i ely show a “biased con o ma ion.”
• T ansc ip omic escue claims exagge a ed.
8
Conclusions
While Wang e al.1 p opose an exci ing a enue—biased agonism a CCKBR o ea ing
AD— he manusc ip su e s om signi ican me hodological and in e p e i e
weaknesses. Key claims ely on incomple e bias quan i ica ion, insu icien s uc u al
esolu ion, unde powe ed neu onal assays, and inadequa ely con olled beha io
expe imen s. Ex ended Da a and Supplemen a y Figu es in oduce addi ional
inconsis encies, sugges ing ha se e al conclusions may be p ema u e.
To ealize he p omise o biased CCKBR agonism, u u e wo k mus :
• P o ide igo ous, model-based quan i ica ion o ligand bias
• Valida e s uc u al p edic ions ia mu agenesis
• Use blinded, andomized beha io al expe imen s wi h adequa e sample sizes
• Imp o e anspa ency in c yo-EM econs uc ion and QC
• S eng hen ansc ip omic and p o eomic analyses
• Ensu e ull ep oducibili y o signaling and pheno ypic assays
In i s cu en o m, he s udy aises in iguing hypo heses bu does no ye p o ide
de ini i e e idence suppo ing pa hway-selec i e CCKBR agonism as a he apeu ic
s a egy o Alzheime ’s disease.
Re e ence
1 Wang, J. L. e al. Elucida ing pa hway-selec i e biased CCKBR agonism o
Alzheime 's disease ea men . Cell (2025).
h ps://doi.o g/10.1016/j.cell.2025.10.034
