Loss-of-function lesions impact B-cell development and fitness but are insufficient to drive CLL in mouse models

TO THE EDITOR: Loss-o - unc ion lesions impac B-cell de elopmen and fi ness bu a e insu ficien o d i e CLL in mouse models Elisa en Hacken, 1,2, *Shanye Yin, 1,2, *Robe Redd, 3 Ma ía He n´ andez S´ anchez, 4,5 Kendell Clemen , 6 Gab iela B uns ing Ho mann, 1 Fa a F. Regis, 1 Elizabe h Wi en, 1 Shuqiang Li, 7,8 Donna Neube g, 3 Luca Pinello, 6 Kenne h J. Li ak, 7,8 and Ca he ine J. Wu 1,2,8,9, * 1 Depa men o Medical Oncology, Dana-Fa be Cance Ins i u e, Bos on, MA; 2 Ha a d Medical School, Bos on, MA; 3 Depa men o Da a Science, Dana-Fa be Cance Ins i u e, Bos on, MA; 4 The Ins i u e o Cance Molecula and Cellula Biology–Cance Resea ch Cen e , Ins i u o de In es igaci ´ on Biom ´ edica de Salamanca, Uni e si y o Salamanca, Salamanca, Spain; 5 Depa amen o Biochemis y and Molecula Biology, Pha macy School, Uni e sidad Complu ense de Mad id, Mad id, Spain; 6 Cen e o Cance Resea ch, and Cen e o Compu a ional and In eg a i e Biology, Molecula Pa hology Uni , Massachusse s Gene al Hospi al, Bos on, MA; 7 T ansla ional Immunogenomics Lab, Dana-Fa be Cance Ins i u e, Bos on, MA; 8 B oad Ins i u e o Massachuse s Ins i u e o Technology and Ha a d, Camb idge, MA; and 9 Depa men o Medicine, B igham and Women’s Hospi al, Bos on, MA Mouse models a e indispensable ools o s udy he impac o gene ic al e a ions in cance . Only a ew s udies ha e hus a unde aken unc ional cha ac e iza ions o ecu en genomic-disco e ed gene mu a ions o ch onic lymphocy ic leukemia (CLL) 1 despi e ecen ad ances in gene ic enginee ing ha ha e inc easingly acili a ed an imp o ed abili y o assess he in i o e ec s o candida e disease d i e s. In his con ex , condi ional knockin and knockou s a egies ha e been applied o he s udy o indi idual lesions, including del(13q), modeled in he MDR mice, 2 he B-cell de elopmen al ac o Ikz 3, 3 o in he splicing ac o S 3b1 in combina ion wi h A m dele ion, 4 all o which ecapi ula e indolen CLL de el- opmen , consis en wi h he pheno ype o human disease. We hypo hesize ha he mos common loss-o - unc ion (LOF) pu a i e d i e s ypical o CLL (A m,T p53, Chd2, Bi c3, Mga, o Samhd1) 1,5 (supplemen al Figu e 1A-B) may a ec p eleukemic signaling, hus dys egula ing B-cell–in insic p osu i al pa hways. To add ess his ques ion, we de eloped a e sa ile pla o m o he in oduc ion o indi idual LOF lesions in i o in mu ine B cells. We gene a ed a ansplan model ha e ficien ly deli e s single guide RNAs (sgRNAs) o hema opoie ic s em and p ogeni o cells o s ains exp essing B-cell– es ic ed Cas9 (ie, Cd19-Cas9; Figu e 1A [le ]), hus gene a ing gene edi s only in de eloping Cas9-exp essing B cells and allowing he in oduc ion o inse ions-dele ions (indels) a he a ge locus, leading o LOF o he a ge gene. Fi s , we in e c ossed mice ca ying B-cell– es ic ed C e exp ession (Cd19-c e) wi h hose ca ying condi ional Cas9–g een fluo escen p o ein (Cas9-GFP) 6 o gene a e a s ain exp essing B-cell– es ic ed Cas9 (Cd19-Cas9) (Figu e 1A [le ]). Nex , we op imized he in i o enginee ing o ea ly s em and p ogeni o cells (ie, Lin - Sca-1 + c-ki + [LSK]) om Cd19-Cas9 mice using len i i us exp essing sgRNAs (mChe y + ) a ge ing A m,T p53, Chd2, Bi c3, Mga, o Samhd1. We chose LSKs because o hei high ansducibili y and long- e m epopula ing po en ial and designed sgRNAs a ge ing ea ly exonic egions in each o he 6 genes o allow ea ly dis up ion o he p o ein sequence (supplemen al Table 1). Las ly, we ans- plan ed he single sgRNA-exp essing LSKs in o suble hally i adia ed CD45.1 ecipien mice. We confi med he consis en p esence o ~45% o 85% gene-edi ed sequences (>70% ca ying ame- shi mu a ions, he majo i y esul ing om 1 o 2 base pai indels; supplemen al Figu e 2) in edi ed B cells (GFP + mChe y + , isola ed om he pe iphe al blood by cell so ing), bu no in T cells o monocy es, a 4 mon hs a e ansplan by polyme ase chain eac ion-based a ge ed deep sequencing (analyzed by CRISPResso2 so wa e 7 ;Figu e 1A [ igh ]; supplemen al Figu e 1C-D). A mino i y o lesions (<30%) led o in- ame al e a ions (ie, ±3 nucleo ides in leng h). Rep oducible on- a ge edi ing Submi ed 11 Oc obe 2022; accep ed 23 No embe 2022; p epublished online on Blood Ad ances Fi s Edi ion 7 Decembe 2022; inal e sion published online 11 Augus 2023. h ps://doi.o g/10.1182/bloodad ances.2022009135. *E.H., S.Y., and C.J.W. con ibu ed equally o his s udy. Mouse RNA-seq da a epo ed in his a icle ha e been deposi ed in he Gene Exp ession Omnibus da abase (accession numbe GSE197061). Da a a e a ailable on eques om he co esponding au ho , Ca he ine J. Wu (cwu@ pa ne s.o g). The ull- ex e sion o his a icle con ains a da a supplemen . © 2023 by The Ame ican Socie y o Hema ology. Licensed unde C ea i e Commons A ibu ion-NonComme cial-NoDe i a i es 4.0 In e na ional (CC BY-NC-ND 4.0), pe mi ing only noncomme cial, nonde i a i e use wi h a ibu ion. All o he igh s ese ed. RESEARCH LETTER 4514 22 AUGUST 2023 •VOLUME 7, NUMBER 16 Downloaded om h p://ashpublica ions.o g/bloodad ances/a icle-pd /7/16/4514/2072565/blooda_ad -2022-009135-main.pd by gues on 20 Feb ua y 2024 was de ec ed ac oss he 6 coho s (wi h no de ec able gene edi s in 24 pu a i e o - a ge s), pe single-cell DNA-sequencing analysis (DNA-seq) o 1787 single B cells (supplemen al Figu e 1E). All LOF mu a ions we e ound o con e a su i al ad an age o B cells compa ed wi h he non- a ge ing sc amble con ol when exposed o he mi ogenic s imuli lipopolysaccha ide (LPS) and in e leukin-4 (IL-4) o 72 hou s in i o (n = 5 pe g oup; P< .05; Figu e 1B). Likewise, ci cula ing edi ed B cells om animals ollowing ans- plan we e ele a ed ac oss he mu a ed models compa ed wi h he non- a ge ing con ol line (N -1;P< .05; RM-ANOVA; Figu e 1C). Despi e hese ea ly changes, mos ly ma ked a 4 mon hs a e ansplan , a ime poin a which B cells a e conside ed o achie e op imal hos econs i u ion, none o he single-edi ed animals (n = 10 pe g oup) de eloped clonal B220 + CD5 + Igκ + ci cula ing B-cell expansions o e 24-mon hs o obse a ion. Mga- and Chd2- deple ed animals showed ajec o ies ha we e pa icula ly sho - li ed (supplemen al Figu e 1F; P<.0002), in line wi h limi ed in i o pe sis ence obse ed in p e ious cell line s udies. 8 These obse a ions collec i ely sugges he dual possibili y o limi ed cell- in insic dys egula ion o p osu i al pa hways in oduced by hese lesions and/o insu ficien cell-ex insic con ibu ion o leukemia pheno ypes in hese o he wise heal hy animals. We asked whe he any o he modeled lesions (also equen d i e s o o he agg essi e B-cell malignancies, including di use la ge B-cell lymphoma) 9 would a ec B-cell de elopmen al ajec o ies (supplemen al Figu e 3A; supplemen al Table 2). Al hough ansi ional B-cell popula ion abundance was unchanged ac oss geno ype- defined g oups (supplemen al Figu e 3B-C), Mga-deple ed mice showed inc eased ma ginal zone (P= .003) (wi h conco dan ly dec eased ollicula B cells) and ge minal cen e (P= .009) splenic subpopula ions (Figu e 2A-B; supplemen al Figu e 3D). We also iden ified an abundance o pe i oneal B1a and a conco dan dec ease in B1b cells (P<.05;Figu e 2C; supplemen al Figu e 3E), sugges ing a ole o Mga in egula ing ma u e B-cell de elopmen and lineage-specifica ion. In pa icula , he inc eased spon aneous ge minal cen e o ma ion may ela e o he lymphomogenic ac i i y o his lesion, whe eas inc eased CD5 + B1a pe i oneal B cells may ins ead ela e o CLL p edisposi ion, as B1a cells a e conside ed o be he cell o o igin o mu ine CLL. 10 Gi en he known ole o Mga as nega i e egula o o MYC a ge s in o he cellula con ex s, 11 we u he asked whe he he obse ed B-cell de elopmen al pheno ypes could esul om abe an MYC a ge ac i a ion in ma u e B cells. To his end, we analyzed he BN -1 Ga ed on li e 1 5 4 3 0 –3 2345 GFP (log) mChe y (log) 10.9% T p53 indel 1 5 4 3 0 –3 2345 GFP (log) mChe y (log) 88.7% Bi c3indel A mindel Samhd1indel T p53indel Chd2indel Mgaindel N -1 0123 P (all s. N -1) < .05 n = 5/g oup Fold inc ease a day 3 o e day 0 (edi ed s. unedi ed) C %GFP+mChe y+B cells 30 20 10 0 248121824 Mga P = .01 N -1 248121824 Chd2 P = .05 24 8 12 18 24 T p53 P = .04 30 20 10 0 24 8 12 18 24 Samhd1 P = .03 24 8 12 18 24 A m P = .002 24 8 12 18 24 Bi c3 P = .05 Mon hs pos - ansplan Cd19c e/c e Cas9/Cas9 Cd19-Cas9 (8-12 week) X CD45.2 LSKs CD45.2 LSKs CD45.1 N -1 Mga Chd2 A m Bi c3 Samhd1 T p53 LSK isola ion sgRNA (single) T ansplan T ansduc ion LSK enginee ing Es ablishmen o B-cell es ic ed single-edi ed coho s A Bi c3indel A mindel Samhd1indel T p53indel Chd2indel Mgaindel N -1 unedi ed indels 0 % 89.1% 80.6% 90.8% 84.7% 45.7% 63.9% F ameshi In- ame 87.779.189.676.2 72 86.6 Figu e 1. Indi idual CLL LOF lesions al e ma u e B-cell su i al bu a e insu ficien o d i e leukemia in mouse models. (A) T ansplan schema and gene a ion o singly-edi ed mouse lines. LSKs we e len i i ally ans ec ed in i o wi h sgRNAs a ge ing he single LOFs (o a non- a ge ing sc amble con ol, ie, N -1) in addi ion o he mChe y ma ke be o e he eng a men in o suble hally i adia ed CD45.1 ecipien mice. PCR-based a ge ed deep sequencing o DNA om pe iphe al blood edi ed B cells (ie, GFP + mChe y + ) was used o confi m p esence o ~45% o 85% gene edi s (>70% ameshi mu a ions) ac oss he 6 genes. (B) In i o su i al o no mal B cells isola ed om he spleen o 5 animals pe g oup (6 LOF-exp essing s ains and 1 N -1) and cul u ed in i o in p esence o LPS+IL4 o 3 days. Fold inc ease o e day 0 is displayed alongside ep esen a i e flow cy ome ic plo s om 1 N -1 and 1 T p53-deple ed sample. P- alue lowe o equal o 0.05, ANOVA wi h Dunne ’s co ec ion o mul iple compa isons. (C) Pe cen (%) GFP + mChe y + B cells as assessed on longi udinal bleeds o e he cou se o 24 mon hs, in 10 animals pe g oup. *P- alue lowe o equal o 0.05; lowe o equal o 0.05; RM-ANOVA, epea ed measu es analysis o a iance. 22 AUGUST 2023 •VOLUME 7, NUMBER 16 RESEARCH LETTER 4515 Downloaded om h p://ashpublica ions.o g/bloodad ances/a icle-pd /7/16/4514/2072565/blooda_ad -2022-009135-main.pd by gues on 20 Feb ua y 2024 ansc ip omes o gene-edi ed ma u e B cells (GFP + mChe y + ) and compa ed hem wi h hose o unedi ed (GFP + mChe y - )B cells isola ed om he same animals (2 animals pe g oup; Figu e 2D). Analyses we e pe o med 4 mon hs a e ansplan because his was obse ed o be he peak o edi ed B-cell abun- dance ac oss coho s (Figu e 1C). Via En ich analysis, using he Molecula Signa u es Da abase (MSigDB), we iden ified en ichmen in PI3K/AKT/mTOR and No ch signaling and dimin- ished in e e on gamma, p53, and apop o ic gene exp ession in Mga-deple ed cells compa ed wi h hose in he con ol unedi ed ac ion ( alse disco e y a e <0.10; Figu e 2E; supplemen al Tables 3-4). To iden i y di ec MYC a ge s a ec ed by loss o Mga, we pe o med an in eg a ed analysis o Mga-deple ed RNA-seq and exis ing ch oma in immunop ecipi a ion–seq (ChIP- seq) da a o MYC om mouse B cells exposed o LPS s imula ion o 8 hou s 12 using CISTROME-GO, 13 which allows so ing o di ec a ge s o specific ansc ip ion ac o s based on he in e- g a i e anking o ansc ip ion ac o –binding (ChIP-seq) com- bined wi h change in gene exp ession (RNA-seq). Ou esul s showed ha Mga deple ion ac i a ed ansc ip ion o se e al MYC a ge genes wi h ele ance o B-cell de elopmen al biology (supplemen al Table 5), including he B-cell ecep o signaling kinase lymphocy e cell-specific p o ein- y osine kinase (Lck), he cell cycle and Wn egula o s S ag3 and P p s and he ibosomal p o ein Rps29 in esponse o LPS (Figu e 2F; supplemen al Figu e 4). No ably, di ec MYC a ge s we e mo e likely o be N -1 –3 5 4 3 0 –3 03 4 5 CD21 (log) CD23 (log) 9.1% Mgaindel –3 5 4 3 0 –3 03 4 5 CD21 (log) CD23 (log) 18.6% Bi c3indel A mindel Samhd1indel T p53indel Chd2indel Mgaindel N -1 0 5 10 15 20 25 % MZ B cells P = .003 A N -1 –3 5 4 3 0 –3 03 4 5 CD38 (log) CD95 (log) <1% Mgaindel –3 5 4 3 0 –3 03 4 5 CD38 (log) CD95 (log) 3.4% Bi c3indel A mindel Samhd1indel T p53indel Chd2indel Mgaindel N -1 012345 % GC B cells P = .009 B C Bi c3 indel A m indel Samhd1 indel T p53 indel Chd2 indel Mga indel N -1 0 20406080 % B1a B cells P = .0005 N -1 –3 5 4 3 0 –3 03 4 5 CD5 (log) B220 (log) 34.3% Mgaindel –3 5 4 3 0 –3 03 4 5 CD5 (log) B220 (log) 62.3% D –3 5B cells 4 3 0 –3 03 4 5 B220 (log) CD45.2 (log) –3 5Unedi ed Edi ed 4 3 0 –3 0 RNA-seq RNA-seq Pai ed DGE/En ich CISTROME-GO 345 GFP (log) mChe y (log) E In e e on Gamma Response Wn -be a Ca enin Signaling In e e on Alpha Response IL-2/STAT5 Signaling Apop osis p53 Pa hway KRAS Signaling Up PI3K/AKT/mTOR Signaling No ch Signaling Hypoxia Mi o ic Spindle 42024 UPDOWN –Log(P) F MYC ChIP Baseline LPS Mgaw _1 Mgaw _2 Mgaindel_1 Mgaindel_2 Lck G MYC a ge s Log2FC 2 1 –1 0 n = 69 P = .001 n = 24 Figu e 2. Mga mu a ion al e s B-cell de elopmen al pa hways. (A) Pe cen (%) abundance o MZ B cells, (B) GC B cells in spleen and (C) B1a cells in pe i oneum p epa a ions om 5 animals pe g oup, including non- a ge ing con ols. **P- alue, lowe o equal o 0.01, ***P- alue, lowe o equal o 0.001, ANOVA wi h Dunne 's co ec ion o mul iple compa isons. (D) Flow-plo o edi ed s unedi ed ac ions analyzed ia RNA-seq and CISTROME-GO. (E) Pa hway en ichmen analysis o Mga-deple ed s unedi ed cells, pe En ich . (F) IGV sc eensho o MYC ChIP-seq da a o mouse B cells wi h o wi hou LPS ea men and RNA-seq da a o Lck gene exp ession in edi ed (Mga indel ) and unedi ed (Mga WT ) ac ions o 2 independen mice.(G) Change in exp ession (Mga indel s Mga WT ) o genes ha we e di ec Myc a ge s upon LPS s imuli, as assessed by CISTROME-GO analysis. ***P- alue, lowe o equal o 0.001; χ 2 es . GC, ge minal cen e ; IGV, in eg a i e genomics iewe ; MZ, ma ginal zone. 4516 RESEARCH LETTER 22 AUGUST 2023 •VOLUME 7, NUMBER 16 Downloaded om h p://ashpublica ions.o g/bloodad ances/a icle-pd /7/16/4514/2072565/blooda_ad -2022-009135-main.pd by gues on 20 Feb ua y 2024 up egula ed han Mga-deple ed cells, consis en wi h MYC ac i- a ion (P= .001; χ 2 es ; Figu e 2G). Toge he , hese esul s suppo he no ion ha Mga plays a unique ole in egula ing MYC a ge s in ol ed in B-cell p oli e a ion and di e en ia ion, esul ing in a di ec impac on cell a e de e mina ion o ma u e naï e B cells. In conclusion, we demons a e ha ou models ep esen a aluable and e sa ile pla o m o he s udy o indi idual gene ic lesions in a p eleukemic B-cell con ex . We show ha in oduc ion o indi idual CLL LOF mu a ions in ma u ing B cells can al e p osu i al and B-cell de elopmen al pa hways and inc ease cellula fi ness in i o and in i o. We u he obse e ha indi idually hey a e insu ficien o d i e leukemia and mo i a e s udies o e alua ing he e ec s o combina- o ial asso men o lesions in B-cell leukemogenesis. Acknowledgmen s: The au ho s hank membe s o he Wu lab o he aluable discussions. The au ho s a e also g a e ul o excellen echnical assis ance om he Dana-Fa be Cance Ins i- u e Animal Resea ch Facili y and he Dana-Fabe Cance Ins i u e Flow Cy ome y Co e. This s udy was suppo ed by g an s om he Na ional Ins i u es o Heal h/Na ional Cance Ins i u e (P01 CA206978 and R01CA216273). E. .H. is a schola o he Ame - ican Socie y o Hema ology. S.Y. is suppo ed by a esea ch ellowship om he Lau i S auss Leukemia Founda ion and Na ional Cance Ins i u e (R21 CA267527-01). M.H.S. has been suppo ed by a Sa a Bo ell pos doc o al con ac (CD19/00222) om he Ins i u o de Salud Ca los III (ISCIII), co ounded by Fondo Social Eu opeo “El Fondo Social Eu opeo in ie e en u u u o.”K.C. is suppo ed by Na ional Human Genome Resea ch Ins i u e Ca ee De elopmen Awa d K99HG011658. S.L. is suppo ed by he Na ional Cance Ins i u e Resea ch Specialis Awa d (R50CA251956). L.P. is pa ially suppo ed by Na ional Human Genome Resea ch Ins i u e Genomic Inno a o Awa d R35HG010717. Con ibu ion: E. .H., S.Y., and C.J.W. designed he s udy and w o e he manusc ip ; E. .H. gene a ed mouse models, pe o med mos expe imen s, and analyzed da a; S.Y. pe o med RNA-seq and ChIP- seq analyses; S.L. pe o med he single-cell DNA-seq expe imen s; R.R., M.H.S, K.C., L.P., K.J.L., and D.N. assis ed wi h s udy design and da a analysis; G.B.H., F.F.R., and E.W. p o ided echnical suppo ; and C.J.W. p o ided o e all supe ision o he s udy. Conflic -o -in e es disclosu e: C.J.W. is an equi y holde o BioNTech Inc. and ecei es esea ch unding om Pha macyclics. L.P. has financial in e es s in Edily ics Inc. K.C. is an employee, sha eholde , and o fice o Edily ics Inc. The in e es s o L.P. and K.C. we e e iewed and a e managed by Massachuse s Gene al Hospi al and Pa ne s Heal hCa e in acco dance wi h hei conflic - o -in e es policies. The emaining au ho s decla e no compe ing financial in e es s. ORCID p ofiles: E. .H., 0000-0003-1573-8799; S.Y., 0000- 0001-9116-5238; R.R., 0000-0002-1329-5288; M.H.S., 0000- 0001-9968-2782; K.C., 0000-0003-3808-0811; G.B.H., 0000- 0001-7758-898X; K.J.L., 0000-0001-9105-5856; C.J.W., 0000- 0002-3348-5054. Co espondence: Ca he ine J. Wu, Depa men o Medical Oncology, Dana-Fa be Cance Ins i u e, 450 B ookline A e, Dana Building, Room DA-520, Bos on, MA 02115; email: cwu@ pa ne s.o g. Re e ences 1. Knisbache BA, Lin Z, Hahn CK, e al. Molecula map o ch onic lymphocy ic leukemia and i s impac on ou come. Na Gene . 2022; 54(11):1664-1674. 2. Klein U, Lia M, C espo M, e al. The DLEU2/miR-15a/16-1 clus e con ols B cell p oli e a ion and i s dele ion leads o ch onic lymphocy ic leukemia. Cance Cell. 2010;17(1):28-40. 3. Laza ian G, Yin S, Ten Hacken E, e al. A ho spo mu a ion in ansc ip ion ac o IKZF3 d i es B cell neoplasia ia ansc ip ional dys egula ion. Cance Cell. in p ess. 2021;39(3):380-393.e8. 4. Yin S, Gambe RG, Sun J, e al. A mu ine model o ch onic lymphocy ic leukemia based on b cell- es ic ed exp ession o S 3b1 mu a ion and A m dele ion. Cance Cell. 2019;35(2):283-296.e285. 5. Landau DA, Tausch E, Taylo -Weine AN, e al. Mu a ions d i ing CLL and hei e olu ion in p og ession and elapse. Na u e. 2015; 526(7574):525-530. 6. Pla RJ, Chen S, Zhou Y, e al. CRISPR-Cas9 knockin mice o genome edi ing and cance modeling. Cell. 2014;159(2):440-455. 7. Clemen K, Rees H, Can e MC, e al. CRISPResso2 p o ides accu a e and apid genome edi ing sequence analysis. Na Bio echnol. 2019;37(3):224-226. 8. Ten Hacken E, Clemen K, Li S, e al. High h oughpu single-cell de ec ion o mul iplex CRISPR-edi ed gene modifica ions. Genome Biol. 2020;21(1):266. 9. Reddy A, Zhang J, Da is NS, e al. Gene ic and unc ional d i e s o di use la ge B cell lymphoma. Cell. 2017;171(2):481-494.e415. 10. Hayakawa K, Fo mica AM, B ill-Dasho J, e al. Ea ly gene a ed B1 B cells wi h es ic ed BCRs become ch onic lymphocy ic leukemia wi h con inued c-Myc and low Bm exp ession. J Exp Med. 2016;213(13): 3007-3024. 11. Llaba a P, Mi suishi Y, Choi PS, e al. Mul i-omics analysis iden ifies MGA as a nega i e egula o o he MYC pa hway in lung adenoca cinoma. Mol Cance Res. 2020;18(4):574-584. 12. Tesi A, de P e is S, Fu lan M, e al. An ea ly Myc-dependen ansc ip ional p og am o ches a es cell g ow h du ing B-cell ac i a ion. EMBO Rep. 2019;20(9):e47987. 13. Li S, Wan C, Zheng R, e al. Cis ome-GO: a web se e o unc ional en ichmen analysis o ansc ip ion ac o ChIP-seq peaks. Nucleic Acids Res. 2019;47(W1):W206-W211. 22 AUGUST 2023 •VOLUME 7, NUMBER 16 RESEARCH LETTER 4517 Downloaded om h p://ashpublica ions.o g/bloodad ances/a icle-pd /7/16/4514/2072565/blooda_ad -2022-009135-main.pd by gues on 20 Feb ua y 2024