Glioblastoma–natural killer cell crosstalk: insights from dynamic spheroid models reveal the importance of secreted cytokines and the CD155 axis

1 Gliobla s to ma–na t ural k iller cell cros st alk: insights f rom dyna mic spheroid models reveal the 1 i mpor tanc e of secreted c ytokines and th e CD155 axis 2 Ana marija H abič (1,2) , Tin a Kol enc Mil av ec (1,2 ), Pia Ži ž ek (1,2 ), Špe la Klad nik (1), Be r n a r da Ma jc (1) , 3 E manu el a Se nj or (3 , 4), Milic a Peri šić Na n ut (3, 4 ), A n dr e j P orčni k (5) , Borut Pre s t o r (5) , Urb an Šv ajger 4 (6), Me tka N ovak (1 ,7), Barb ara B rez nik ( 1,8) 5 Co rresponding author s : Ba rbara Br e z nik (b a r ba r a .b r e z nik @ n ib. s i ) , Anama r i ja H a bi č 6 (a nama rija. habic @ ni b. s i ) 7 8 (1) Na t i on al I n st it ut e o f Bi ology , Dep a r t men t o f Gen e tic Tox icolo gy and C ance r Biolo gy, Ljub ljana , 9 Sl o v en i a 10 ( 2 ) J o ž ef St efa n In t e rn at i on a l Po st gr a duat e S c hool, L jublj an a, Sl ove ni a 11 (3) Jože f S t e f an I n sti t u te , De par tme nt o f Biotec h nolo gy, L jubljan a, Slo v eni a 12 (4) Facu l t y of P ha r m ac y, U niv ers i t y of L jubljana , Ljub ljana , S love ni a 13 (5) De partme nt o f Ne uro s ur ge r y , U n iver s ity Medi cal C entr e Lju blja na, Ljubl ja na, S l oven i a 14 (6) Slov e nia n I n s ti tu te f or Tran sf u s i on Me dic ine, Ljubl jan a, S love ni a 15 (7) Bio techn ical F acu lty, Univ ersi ty o f Ljubl jana , L jublja na , Sl oveni a 16 (8) Facu l t y of Che mi stry a nd Che m ical Tec hnology , Uni ve rsity of L jublj a na, L jublja na , Slo ve nia 17 18 19 20 21 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 2 Abstr ac t : 1 Glio bla s t oma (G B) i s a n a ggre s s iv e pr i m ary brain can ce r with poo r p atie nt pr og nos i s . N atu r a l kill e r 2 (NK) c ell s can r ec ogni se a nd elimi na te a rang e o f mali gnant c ells , incl uding GB s te m cel ls, whic h drive 3 GB r ecurr enc e. N K c ell -ba s ed immu n ot h era py ha s emerged a s a promi sin g app r oa ch fo r GB 4 trea tmen t, bu t a b e t te r unde r stand ing o f th e c omplex cros stalk be tween GB and NK c ells i s n e eded , 5 pa r tic ula rly w it h in the immun o s upp re s s i ve G B tumour mi croe nvi r onm en t. In thi s s tudy, we 6 e stabli s h ed a repr oduci ble pr ot oc ol for the prod uction and dy namic c ultur e o f uniformly s ized GB 7 sph eroid s u s i ng the Celv ivo Clino s tar s y stem. O ur spher oid s reca pi tula ted the het erog ene ou s 8 st ructur e o f G B a nd expr es se d ligan d s fo r N K c ell rec e ptor s at lev el s di stinc t f ro m tho se ob served i n 9 c orr e s pond i ng GB ce ll li ne s i n s t anda r d c ulture , i mplica t i ng al t e red s en si tivi t y of GB cel l s to N K c ell s i n 10 dy namic 3D cul tur es . GB -NK cell cr o sstal k w as GB ce ll type dep e nde nt and the a bili t y o f NK ce ll s t o 11 i nfilt r a te G B did no t n ec e ssarily c or r e la t e with t heir cyt otox icity a gain s t G B ce ll s. S phe roid s de rived 12 from di ff eren tia ted GB cel l s se cre ted h i gher l evel s o f immunom odul ato ry cytoki nes co mpar ed t o 13 sph eroid s fr om GB stem -lik e cel l s , and a promin en t i ncr ea se i n th e s ec re t i on o f imm une -at tra cting 14 fa cto r s was ob serv ed i n t heir co -c ultu re s with N K c ells . F inall y, the C D155 -D N AM 1 / TI GIT ax i s wa s 15 i ndica t e d a s a n impor tan t r egula tor o f N K c ell c ytot oxic ity aga in st GB s tem -like c ells . Col lec tiv ely, ou r 16 re sult s highl ight impo rta nt fac to rs in G B- NK c ell communic atio n a nd provi d e a groundwork f o r 17 fur ther ta rget ed re s ea r c h a s w ell a s the r a p eutic evalu ati on o f NK c ell-b a s ed approac h e s in th e 18 e stabli s h ed dynami c 3D c ultur e s . 19 Key w o r d s : gliobla stoma, immuno the rap y , dyna mic i n vitro mode l s , na tural ki ll er ce lls , cyto tox icity, 20 i nfilt r a t i on, cy tokine s, CD155 21 22 23 24 25 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 3 1. Backg r oun d 1 G l i o b l a s t o m a ( G B ) i s t h e m o s t c o m m o n a n d a g g r e s s i v e p r i m a r y b r a i n c a n c e r i n a d u l t s . D e s p i t e t h e 2 st andar d o f c ar e tre a tment tha t i nclu des surgery, rad io t h e r a py an d chemothe rap y wit h 3 temo z ol omi de, th e progno si s of GB pati ent s is po o r an d le ss than 10% of pa tien ts a r e still al ive at 5 4 y ear s a f t er th e diag no s is (1 ,2) . Dur i ng th e past dec ade s , t h ere h ave bee n n o ma jor a dvanc es in GB 5 trea tmen t whic h would s ignific an tly improve pati e nt su r v ival , i ndica t in g a n ur ge nt ne ed f or n ove l 6 ther ap eu tic a ppr o a che s. 7 I n th e la s t dec ade, immunoth erapy h a s r e voluti onized th e t rea tmen t o f ma ny c anc er s , e s p ecia lly 8 ha emat olog ical malig nanc i e s (3 ) . Nev er the le s s , a pplying th e s e appr oac he s t o soli d tumou rs ha s of ten 9 show n l imit ed s u cce ss. To a gr eat exten t , thi s can be a ttribu ted to the immuno supp re ssive t umo u r 10 mi croenvi r on men t (TME ), w hich not onl y repre sen t s a physic al ba rrie r for imm une c ell infil tra tion, 11 bu t al so dir ectly reg ul ate s the ac t iv ity of i mmune cel l s by secr eting immun om odulating a nd of te n 12 i mmunosupp r e ssive fa cto r s (4 ). De spi t e s everal chall eng es such a s in trac ran i al loc ati on o f th e 13 tumou r , immuno s uppr e s s i ve TME, and hete roge n eity of G B , immun ot he r a py holds pr omi se for GB 14 trea tmen t (5 ). 15 S everal immuno the rapeu tic app roac h es ha ve al rea dy be e n te st ed in the s etti ng of GB (5). H ow ever , 16 a lthough a numb er o f pr e-c linic a l studi es s how ed enc ou r a ging re sul ts, to da te , no durabl e c linic al 17 be ne fit s ha ve b e en ob s e r v ed in rando mi z ed clini cal tr ial s in pa t i en t s . T hi s i ndica tes a l ack of r el evan t 18 prec linic al model s th at would reli ab ly reflec t the compl exity and bi ologic al pro pert i e s of th e G B i n 19 hu man and c ould ac curat ely pre dict t he ef fic acy of immuno the rapie s (6) . Unlik e 2 D c ell c ult ure s , 3 D 20 c ulture mod el s, s u ch a s s ph e r oi d s, orga n ot y pic c ultur e s , a nd or g anoi ds , more clo sely mimic tumo u r 21 a r c hitec tur e, nut r i en t and oxy gen gradi en ts a nd me t a boli sm a nd thu s al low f or a mor e ac cur at e 22 a s s e s s m e n t o f i m m u n o t h e r a p e u t i c a p p r o a c h e s ( 7 , 8) . N e v e r t h e l e s s , t h e a b s e n c e o f a c o m p l e t e T M E , 23 sys temic immuni ty and func t i o nal v a sc ulatur e still l imi t th e pr edic tive pow er o f t he se mo del s. 24 (Human iz e d) i mmunoc ompe te nt mo u s e mod els o f GB, altho ug h more rel eva nt i n t he se r ega r d s, ar e 25 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 4 e xpensive and t ime-co n suming and o ft en do no t f ull y repr oduc e hu m an pa t h o logy and immune 1 re spon se s (9 ). To a n ext ent , blo od flow and immun e c ell in filt rati on may b e m imic ked in vi t ro b y 2 c ulturing hu man c ance r c ells in s tatic o r perfus ed m icro fluidic pla tfo rms (10). 3 Natu ral kille r (N K )-c ell b a s e d immuno t he r a p eutic app roach e s m ay be pa rtic ul arly fe asi ble in the 4 se tting o f GB . NK ce ll s ar e in nat e ly mphoid c ell s tha t c an elimina t e tumo ur ce lls withou t pr i o r 5 sen si tiz a t i on and r ec ogniti on o f t um o ur -a s soci a t e d an tigen s (11) . T hei r ac ti v ity is regula t e d by 6 ac t i v at in g a n d i n h i b i t or y r e ce p to r s , wh i c h in t e r ac t w it h l i g a n ds o n t ar get c e l ls . If th e in t e r ac t i o ns w i t h 7 a ctiva t i ng rec ept or s pr eva il over the in hi bitor y sign al s, the i ntr ace llula r cel l c y totoxic m achin e r y i s 8 a ctiva t e d , lea di ng to el imin ation of t he t arget ce ll s ( 1 1,12) . Mo s t common ly, NK c ells oper ate thr ough 9 the per for in -g ra nz y me pa thw ay, rele a si ng c ytolytic granul e s a n d induc ing ta r get c ell apop to s i s . 10 Alte rnativ ely, t a r g et cel ls c an a l s o b e elimi nat ed th rough t he i nduc tion o f the d eath rece pt o r 11 pa thway (13,1 4) . A noth er import ant m e chani s m employ ed by N K c ell s w hich l ink s the inna te and 12 a daptive immun e re spon s es and can al so b e exploi te d the rap eutic ally i s t h e a ntib ody dep enden t 13 c ellul ar c yt otoxic ity (ADC C) (15 ). The t e r m rel ate s t o th e abili t y o f NK c ell s to elimi nat e an tibody -14 c oated ta rget c ell s, whic h a r e r eco gni s ed b y th e Fc r eg ion -binding r ec ept or C D16 (Fcγ RII I) o n the 15 s u rf a c e of N K ce l ls . La s t l y, N K c e l l s als o se cr e te a v a r i et y of c y t o ki ne s s u ch as i n ter f er on- ga m m a (I FN -16 γ ) a nd t umou r nec ro sis fa c tor-a lph a (T NF- α) , w hich direc tly a ff ect ca nce r cell s and modula te both 17 i nnate a nd adap tiv e immune r e s po ns e s ( 11,12). 18 As GB i s ch arac t er i sed by a l ow tumour mutation al bur den (16), th e u se o f NK ce lls may be ben efic ia l 19 c ompared to T c ell -ba se d t he r a pie s an d va cc ine-ba sed a pproac h e s to by pa s s t h e la ck of tumo ur-20 a ssoci a ted an t ig en s and al so the la ck of func tional a ntig en -pre s en ting c ells in the hi ghl y 21 i mmunosupp r e ssive G B T M E . Ad dition all y, NK ce lls may repre s ent a po we rful t oo l ag ains t t he pool o f 22 h e t e r o g e n e o u s G B c e l l s , a s ( u n l i k e T c e l l s ) t h e y c a n r e c o g n i z e a b r o a d s p e c t r u m o f t a r g e t c e l l s 23 ( 1 7, 1 8). I m p o r ta n t l y , N K ce l ls h av e b een s h ow n t o k il l G B s t e m ce l ls – a s u bs et of ca n c er c el l s t ha t a re 24 i nt rin sic ally r e si st a n t to a number o f s t a nda r d ther ape u t i c app roac h es a nd r e p re sent key dr i ver s of 25 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 5 GB rec urrenc e (19,20 ). A s no g r a f t -ve rsu s-ho s t di se as e is induc ed by NK c ells, all og eneic of f -t h e -sh el f 1 NK c ell produc t s ma y be u sed fo r pa t i ent trea tme nt, p rovidin g a c heap e r , mo r e s c alabl e and 2 reprod uc ible al t e r n a t i ve to p a tien t- s p ec ific T c ell produc ts . Du e to thei r immedi ate avai l abili t y , o ff-3 the - shel f produc t s a r e p articul a r ly con ve nient for th e tr ea tment o f aggre s s i ve typ e s o f c anc e r (17) . 4 I n th e GB TME, th e a bunda nce o f NK ce lls i s g ene rally low, ac countin g fo r ap pr ox imately 2 –10% o f 5 l eukoc yte s w ithin t he tumo ur (21 ) , b u t the pr e sence o f a c tivated NK ce ll s po siti ve ly correla te s wit h 6 pa tien t s urv ival (2 2) . H o weve r , i n the im munosupp re s s iv e TME, fu e lled by G B c el ls, the inf il t r ate d NK 7 c ells o ft en exhib it a lt ere d pheno type s a nd re st r ain ed cyto tox ic ac t i vity. S everal immu nos up p r e s sive 8 fa cto r s in the GB TME w er e r epor t e d to downregul at e the exp r e s sion of ac t i va ti ng NK rec e pt o rs and 9 de crea s e NK c ell secr eti on o f IFN -γ (23– 2 5). A d diti onally , N K c ell numb er and fu nc t i on may al so b e 10 i mpaired by hypox ia (26 ) . 11 A deep e r unde r st a nd ing of th e GB - NK cel l c r os s t a lk is n eed e d to d eve lop e f f e c tive NK c ell-ba s ed 12 ther ap eu tic a pproa che s for GB and pr ev ent o r reve rt i t s po t e n t i al r e sistanc e mec hanism s . Rel ev ant in 13 vi t r o mod el s shoul d be u s e d to inv e s tig ate the G B- N K c ell in te rplay a nd NK c e l l-ba se d t h e r a p eutic 14 a pproach e s in th e human se tti ng. To ad dr e s s t h e se ne e d s , w e e s tabl i s h e d a r ep roduc ible a ppr o ac h 15 for g en era tion an d dynam ic c ulture of G B s p h er o id s from GB c ell l ine s in the Cel v ivo Clino sta r s y s t em. 16 GB s ph e r o id s wer e co-c ult ured with N K ce lls in direc t co-c ul tur e a nd o n a dy namic organ - o n - c hip 17 pl at for m to m odel and explor e th e GB- N K ce ll cros stal k. Mor eover, the imp orta n ce of the CD 1 55 axi s 18 f o r th e cy t o to x i ci t y o f N K ce l ls w as st u d i ed i n m o r e d e ta il . 19 20 2. M e t hods 21 Ce ll lines and c ulture 22 NIB140 i s a di ffe ren tia te d GB cel l lin e derive d fr o m p ati ent tumour ti s s u e . Tumour ti s sue wa s 23 ob t a in ed f rom Univer si ty Medic al Cent re L jubljana . The s tudy w as a pproved by th e N a t i onal M edic a l 24 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 6 E t hi cs Co mmi tt e e of th e R epubli c of S l oveni a ( a ppr oval N o. 0120-190 /2018-27 11-41) . N IB140 c el l s 1 w ere cultur ed in su rfac e -tr eat ed ti s s ue culture fla sks ( Je t Bio fil ) i n high-gl uco s e Dulbec co’ s modi fie d 2 E agle’ s medium ( DMEM; Gibc o, Th er m o Fishe r S cienti fic ) su ppleme n t e d wi t h 10% foetal bo vine 3 se rum (FBS ; Gibc o, Th ermo Fi s her Sc i ent ific ) and 1 × peni cill in /stre ptom ycin (S igma-Ald ric h). Prio r t o 4 rea ching con flue ncy, c ell s we re r egu larly pas sag ed by 0.125% tr y p s in – EDTA s ol ut ion ( G ibc o, The rmo 5 Fi s h er S c ie nt i f i c) . 6 G B s t e m - l i k e c e l l l i n e N C H 4 2 1 k w a s p u r c h a s e d f r o m C e l l L i n e s S e r v i c e G m b H ( C L S ) . C e l l s w e r e 7 c ultured as flo ating sphe re s i n s u spen si on ce ll cultur e fl a sks (Sar s t e d t ) in Neur o bas al Mediu m (NBE ; 8 Gibc o, The r mo Fi sh er Sc ie nti fic) sup ple m ented with 1 × pen ic illi n/st r ep tomy cin ( S i gma-Ald r ic h ), 2 mM 9 L -glutami ne ( S igma-Al dric h) , 1 × B-27 ( In vi t rog en) , 1 U/ mL heparin (Sigma -A l dr i c h), 20 ng/mL basic 10 fib r obl a s t grow th f a cto r ( Inv itr og en) a n d 20 ng/mL epid e r ma l g r o wt h f ac tor ( I nvitroge n ) . Sp h ere s 11 w ere regul a r l y dis soci ated wit h TrypL E Express ( Gibc o, The rmo F i s he r Sci e nt i fic) . 12 Ce ll lin e NK - 9 2 wa s purc ha s e d f r o m th e America n Type Cultu re Colle c tion (ATC C) . C ell s w ere g row n in 13 N K - 9 2 c o m p l e t e m e d i u m , i . e . , R P M I 1 6 4 0 m e d i u m w i t h G l u t a M A X ( G i b c o , T h e r m o F i s h e r S c i e n t i f i c ) 14 suppl e ment ed with 12 .5% F BS ( G i bco, T hermo Fi sh er Sc ie nti fic), 12 .5% he a t in activ ated hor s e s eru m 15 (Gibc o , Th er m o Fi she r S cie nti f i c ), 1 × pen icill in/s trep tomyci n (Sig ma- Aldrich ) and 200 I U/mL 16 rec ombina n t hum an int erleuki n 2 ( IL-2 ; Milteny i B iot ec). F or a ctiva t i on, c el l s w e r e grown fo r 24 h in 17 NK-92 c omple te medium supple m ent ed with 1,000 IU/mL rec ombin an t h uman IL -2 (Mil tenyi Bio tec) . 18 I sola te d hea l thy dono r pe r ip h eral bl oo d mononucl ear c e lls (P B M C s ) w er e obta ined a s bu ffy c oa t s 19 from the Blo od Tran s fu si on Ce n t re o f Sl ovenia . Expe r i ment s w ere per for me d in ac cordanc e with th e 20 a pproval o f N atio nal M edic al E t hi c s C ommitte e o f th e Re p ublic o f Sl oveni a (appr ova l no. 0120 -21 2 79/ 2 017-3 ). Cell s w ere w a s h ed twi ce in PBS with 1 mM E DTA a nd 2% F BS to remove pla t e l et s . 22 Hea lthy don or N K c ells were i so lat ed f r om PBM C s u sing t h e E a s y Sep Human N K Cell Enrichmen t Kit 23 (S t e m C ell Te chnolo gie s ) and the Ea syS e p cel l s e par ation mag ne t (St em Cell T ec hnol ogie s) acc ording 24 to th e ma nu factu rer ’ s in s tr uction s. For i sola tion of N K c ell s f r o m whole bl ood o f GB pa tien t s, PBM C s 25 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 7 w ere is olat ed u sing t he Ly mpholyte® C ell Se para tion Medi a ( Ceda rlane ) and NK c e lls wer e i sola ted a s 1 de s c r i bed abov e. Expe r i men ts w ere per formed in acc ordanc e with the approv al of N a ti onal M edic al 2 E t hi cs Commit te e o f the R epubl ic of S loveni a (appr oval no . 0120-19 0 / 20 18-2711 - 41). Af ter i sola tion , 3 NK c ells w er e grow n in N K comple te medium, i.e ., RP MI 164 0 medi um wi th G l ut a MA X ( G i bco, 4 T hermo F i s h e r Sc i enti fic) suppl eme nte d with 1 2.5% FBS (Gibc o, The rmo Fi sh e r Sc ien tific ) an d 1× 5 pe nici llin/ strep tomyc in (S igma -Ald ric h). For a ctiva t i on, c ell s w e r e grown for 18h in NK c omplet e 6 me dium suppl eme nte d with 1 ,000 IU /mL rec ombinant hu m an IL - 2 (Milt enyi B iote c). 7 Normal hum an a st r o cyte s (NHA) w er e p urcha sed from L on z a and w ere cul t ur ed i n A s tr ocy t e M edium 8 (S cienc ell ) supple m ent ed wi t h 10% F B S (Gibco , Thermo F i s h er S cien ti fic ), 1 % astroc yte grow th 9 suppl e ment (Scie n Cell ), an d 1× p enic illin /s t r eptomy cin (S igma -Ald ric h) . Prio r to r eac hing c onflu enc y, 10 c ells we re pa ssag ed by 0 .125% t ryp s in – E D T A solu t io n (G ib co , Th er m o Fi sher Sc ie ntific ). Low passag e s 11 w ere u sed in expe rime nts . 12 K5 62 cells were pu rcha sed from th e A merica n T ype Cult ure Co llec tion ( ATC C) and were grown in 13 RP M I 164 0 me dium with G l u t a MA X (Gi bco , T hermo F i s h e r Sc i enti fic ) su pplem en t e d w ith 10 % FB S 14 (Gibc o , Thermo F i s h e r S ci enti fic ) and 1× penic illin/ strep tomyc in (Si gma -Ald rich) . Ce ll s wer e pa s s a g ed 15 ev e r y 2 - 3 d a ys . 16 Ce ll line s were r egula r ly c hec ked for M y copla s ma c ont aminatio n using the My c oAler t Myc oplasm a 17 Det ec tio n Kit (L onz a ). 18 Im m unophenotyping of PBMCs and NK c ells is ola t e d fr om P BM C s 19 8- C ol or Immunoph e no typing Kit, a nt i - h uman ( Milt eny i Bio t e c ), w a s us ed ac cording to th e 20 ma nufac ture r’ s in struc t io n s to de te rmin e t h e fre quenci e s o f i mmune c ell po pul a tion s (T cel l s , B cell s , 21 NK c ells, monoc yt e s , neu tr o phil s , eo s i no phils, CD4 +, C D8+, a nd C D56+C D3+ T cell s ub se ts) in PB MC s. 22 T he kit was al so u sed to a sse s s th e pur ity of t he NK cel l s i sola ted f rom PBMC s. Stain e d c ells wer e 23 a nalys ed usi ng the MA CS Quan t Analy z e r 10 F low Cy t ome t e r ( Milt eny i Biotech ) a nd Flow J o so ftwa re 24 (BD L i fe Sc ienc e s). The pu rity o f N K cells i sola ted fr om PBM C s w as > 90% . 25 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 8 E s tablishment and c ultur e of GB s pheroi ds in the Cel vivo Cl inos tar sys tem 1 GB sph eroid s w ere e s t abl i s he d b y t h e forc ed floa ting me thod i n 9 6 -well ro und bottom micropla t e s 2 (F alco n, C orni ng). Th e s ta rting num b ers of cell s p er s ph e r oi d w e re op timiz e d for eac h ce l l l ine a nd 3 w ere s e t a t 5,000 an d 3 0 ,000 ce ll s fo r N CH421k an d NIB1 40, r e s p ec tivel y. Th e s elec ted num b er of 4 c ells wa s s eed ed i n 100 ul of c or r es pond ing comple t e medium w ith 4% methy lce ll ulose . Pl at es we re 5 c entri fuged for 90 minu te s a t 900 g a nd incuba t e d f o r 3 da ys u n der s tatic c on di tion s at 37 ° C in 5% 6 CO 2 a t mo spher e. S phe roi d s wer e th en re su s p end ed in fr es h c omple te medi a , tran s ferred to p re -7 e quilibra ted ClinoR ea cto r s (C e lviv o) an d cultured i n a dynami c sy st em in th e Clino S tar cl ino s t at 8 i ncuba t or (Ce l vivo) at 37 °C a nd 5% CO 2 a t mo sph ere . The sp eed of ClinoR ea ctor r ota tion wa s 9 a djusted daily t o en sure op timal sphe roi d dis p e rsion . Med i a in th e Clin oRea c tor s w ere cha nged e very 10 2 –3 day s. Spher oid s tha t had be en grow n in Cli noRe ac t o rs for 8 day s w ere u s ed f or ex perimen t s . 11 Monitoring spheroid grow th 12 S pheroid growth w a s moni t o red by mea suring t h e s pher oid a rea . Imag e s of sphe roid s i n 13 Cl inoR eac to rs wer e t ak en on N i kon Ec lipse Ts 2 R i nver ted mic ro scope . Sphe roid are a wa s quan ti fie d 14 in F i j i ( 2 7 ) . 15 Diss oc iation of G B spher oi ds into single -c ell solution 16 F o r f l o w c y t o m e t r y a n d c a l c e i n r e l e a s e a s s a y s p h e r o i d s w e r e d i s s o c i a t e d i n t o s i n g l e c e l l s . N C H 4 2 1 k 17 s p h e r o i d s w e r e d i s s o c i a t e d i n T r y p L E E x p r e s s ( G i b c o , T h e r m o F i s h e r S c i e n t i f i c ) . A f t e r 5 m i n u t e s o f 18 i ncuba t io n a t ro om tempe rat ure, s ph er oids we re di s s oc ia ted by pip e t tin g. Cell s w er e wash ed wit h 19 P BS and r e su s pe nd ed in de sir ed stai ning solu tion /mediu m. Fo r di ssoc iatio n of NI B140 sphe roid s, w e 20 use d T r y pLE E xpress (Gibc o, Thermo Fishe r Sc ien tific ) a nd C olag ena s e, Ty pe I (The rmo F ishe r 21 S cie nt i f i c) dil ute d in DMEM (Gibco , Thermo F ishe r Sci en ti fic). S ph eroid s wer e i ncuba t e d in th e 22 e n z y m e m i x t u r e a t 3 7 ° C a n d w e r e o c c a s i o n a l l y m i x e d b y p i p e t t i n g u n t i l t h e y d i s s o c i a t e d . S i n g l e c e l l 23 su spe nsi on wa s w a sh ed wi t h PBS an d re susp ended in de si red st aining sol ut i o n/ m ed ium. 24 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 9 F low-cy t om etric determination of ce ll via bility and de ath in GB s phe r oids 1 Aft er s phe roi d dissoci ation , ce lls w er e incuba t e d with ann ex inV -FIT C ( Milt en yi Biotec ) for 15 min at 2 4 ° C . T he n, ce ll s w ere dilu ted and stain e d fo r 5 min wi t h 7- AA D (M i lt enyi Bio tec ). S tain ed c ells wer e 3 di luted a nd a na lys ed u sing t h e M ACS Qua nt An al yz e r 10 F low Cytome ter (M iltenyi Biot ec h) an d 4 F lowJo s of tware (B D L ife Sc ienc e s ). 5 Prepa r ation and immunofluo rescent st a ining of formalin -fixe d, paraf fin-e mbedded (FF PE) sections 6 F or prep a r a t io n o f th e FFP E sec t i on s , s pheroid s (w ithou t or with in filt rat ed NK c ells ) were wa sh ed 7 w ith P BS and fixed in buf fe red 4% fo r maldehy de s ol ut i on (M erck ) for 72 h a t 4 °C. Fixe d sampl e s 8 w ere embe d ded in 0 .5 % aga r o se in PBS to ea s e fu rth er proc e s s ing and pr eve nt sampl e lo s s . Ag a r o se 9 bl ocks we re d ehy dr a t e d a nd em bedd ed in pa raf fin . FF PE sec t io n s (4 μm thic k ) were pr e p are d f o r 10 sub se quen t imm unofluo re s c en t st a ini ng. 11 F FPE sec tion s we re d e par af fini sed i n x ylene (Ch em-L ab ) and reh ydra ted in a s er i e s o f e than ol 12 solu tion s o f d ecre a s i ng conc ent r a t io n s ( 100% , 96%, a nd 70%). S ample s we re h e a t e d at 95 °C for 20 13 mi n in 10 mM sodium c itr a te b uffe r (pH 6.0) to re tr i eve a nt i gen s . Block ing w as pe r f ormed for 1 h at 14 room temp e r a ture in 10 % (v/v) n ormal goat s erum ( S igma -Ald ric h) , 0.1% T r iton X-100 (v / v ) (S igma -15 Ald r ic h) a nd 1% bovi ne serum alb umin ( BS A ; w / v ) (S igma -Ald ric h) i n PBS . Samp le s we re inc uba te d 16 w ith pr i mary an tibodi e s dilut ed in 1% B SA ( w / v ) i n P BS (Tabl e S 1) ov e rnight at 4 ° C . A ft er w a shing 17 w ith 0.5% BSA (w /v) in PBS , flu ore sce n tl y label led s e con dary a n t i bodi es d ilut ed in 1% BSA (w/v) in 18 P BS (T able S1 ) we r e adde d to the slid e s and in cubat ed for 1 h a t room t emp era ture. Th e slide s wer e 19 then w a s h e d in PBS an d nuc lei wer e sta ined with Hoe ch st 332 58 s olution (Sig ma-Ald rich) dilu te d 20 1 :1,000 in PBS . A fte r w as hing with PBS , sample s we re moun ted in ProLo ng Gol d AntiF ad e re agent 21 (In vitro gen, L if e Tec hnologi e s ) , c over slipp ed and s ealed w ith nail poli sh . I nverted fluor e scen t 22 mi croscop e (Niko n Ec lips e Ti, Tok yo, Jap an ) a nd NIS - E lemen t s, N ik on s o ftwa r e , w ere u s e d to im age 23 flu ore sc ence . 24 F low c ytometr y to assess the e xpres sion of rec ept ors on NK ce lls a nd their ligands on c ancer cells 25 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 10 T he exp r e s sion o f rec ep t o r s on live N K c ells (C D16, CD96, DN A M -1, K IR2 DL1, KIR2 D L 2 / L 3/S2, 1 KI R2DL4, KIR3 DL 1, NK G 2D , NKp30, P D-1 , TI G IT ) w a s mea su r e d on N K -92 cell s , N K cel ls i s ola ted from 2 PB M Cs o f 6 - 9 hea lt h y d on o r s a n d N K ce lls is o l at e d f ro m P BM Cs of 6 G B p at i e nts . T h e sa m e r e ce p to rs 3 w ere a l s o analy s e d on N K -92 cel l s tha t hav e bee n c ultu r e d for 48 h in a 1 :1 mix tur e of NK-9 2 4 c omplete medi u m a nd N CH421k/N IB14 0 condi t i oned m edium. The ex pre ssi on of NK cell r ece pto r 5 l igands (B7 -H6, CD112 , C D15 5, P D-L1, CD54, H L A- A , B, C, HLA-E, MI CA / MI CB, U L BP-1 , UL B P-2 / 5/6 , 6 UL BP-3 ) wa s mea s ur ed on live fla sk - and sph eroid -c ultu red ca nce r c ell l ine s NCH4 21k and NIB14 0. 7 Ce ll s we re r esus pend e d in P BS with 1 % BS A and s t ai ned w i th an tibo die s li s t ed in Tabl e S 2 . Ig G 8 is ot y p ic c o n tr o ls we r e use d as c o n t r ols . C e l ls w e r e s t a ine d f or 3 0 m i n a t 4° C . A f te r wa s h i n g , c e lls 9 l abelle d with flu o r e sce n t l y c onjugated pri mary antibo di e s were a n alys ed by the MACS Qu ant A nal yz e r 10 1 0 F low Cytometer ( M i lte nyi Biotec h ). I n the case o f u nconjug ated pr i ma ry antibodie s, cell s wer e 11 w ashed a nd a ddi tion ally stain ed with Alexa Flu or 488 r a bbit an t i -mou s e Ig G seco n dary a ntibo dy 12 (Th ermo F i s he r Sc ienti fic ), inc ubat ed fo r 30 min at 4°C, w a sh ed a nd analy s ed on the MA CS Q u an t 13 A n a l y z e r 1 0 F l o w C y t o m e t e r ( M i l t e n y i B i o t e c h ) . F l o w J o s o f t w a r e ( B D L i f e S c i e n c e s ) w a s u s e d t o 14 de t e r mi ne the % o f c el ls po sitive fo r s p eci fic r ece pto r o r ligan d . G ate s w er e set ba s ed on t he 15 c orr e s pond i ng IgG i sotypic c ontrol s amp l es. 16 E s tablishment of direct NK c ell c o- c ultures w ith G B spheroids 17 S pheroid s tha t h a d be e n gr ow n in Clin oRea ct or s for 8 da y s we re u se d for e s tabli shmen t o f co -18 c ulture s w ith N K c ell s ( NK -92 or h ealt hy d onor NK ce ll s). A t lea s t 5 s phe roid s of eac h cell l ine we r e 19 di ssoci a ted to d et ermine av erage num b er o f c ell s p er sph eroid and to c alc ula te the numb e r o f (non -20 )a ctiva te d NK ce ll s n e ed ed for th e sel ec ted ef fec tor :ta rget r a t i o s (2.5:1 or 5 :1) . For th e co -c ultu r e s, 21 sph eroid s wi t h 1 00 µl of c o rr e sp ondin g c ompl ete media we re tran s fe r red to w ell s o f a round b o ttom 22 9 6-well plat e (Falc on, Cornin g ). The co ncentra tion s o f (no n - ) a ctiva te d NK c ells we r e adju st ed t o 23 a chie ve the de sir ed fin al e ffec tor : ta r g e t ratio s a ft er t he addi tion o f 100 µl o f NK ce ll su s pen sio n pe r 24 sph eroid in 96 -wel l pl a te . F or ac t i vat ed N K ce ll s , IL - 2 w a s inc lu ded in the m e dia t o achi e ve final 25 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 11 c once ntratio n o f 1,0 00 IU/mL . Sphe roid s and NK cell s wer e c o-cul tur ed fo r 24 h at 37 ° C i n 5 % C O 2 1 a t m o s phe re. 2 F or t h e a naly s i s of NK ce ll in filtr ation an d c ytotoxic ity by f l ow c ytome tr y , NK cell s w e r e fluo re s c en tly 3 l abelle d pri or t o the on s e t o f c o-cult ure s t o e nable the ir d i r e ct d etec t io n and di sc rimin ation . NK cel l s 4 w ere wa shed with P BS and sta ined w ith 5 µM Cell T r a ck er Blu e CM AC (Thermo F i s h e r S cie nt i fic) o r 10 5 µM Ce llTr ac ker Gr e en CMF DA (The rmo F is he r S cie n t i fi c) in R P M I m edium w ith out s u p plemen t s. Af t e r 6 3 0 mi nutes o f inc ubation , cel ls w ere w ash ed, and th eir conc ent r a t i o n w as a dj us t ed to ac hieve th e 7 de s i red fina l e ff ector :ta rget r a t i o s a ft er the ad dition of 10 0 µl o f N K c ell su s pen s ion p e r sph eroid i n 8 9 6-well plat e. 9 Co -c ultur e s o f G B s p her oid s a nd nona c t i vated o r I L-2 -ac tiv ated P BMC s f rom GB pati ent s wer e 10 e stabli s h ed in the sam e m anner . 11 F low c ytometr i c ana lys i s of N K ce ll spher oi d infiltr a t ion and cy t otox icity 12 S pheroid s w ith infi l t ra ted fluor esc ently pre -la bell ed NK c ells w er e w ashe d i n PBS an d ind ividua lly 13 di ssoci a ted a s d es c ribed abo ve . C ell s w e r e wa shed a nd r e s u sp ended i n 100 µl PB S with 1% BS A. 1 µ l 14 7 - AA D (Milt eny i Bio tec ) wa s add ed and i nc ubated fo r 5 min to la be l d ead cell s. C e lls were dil ute d and 15 a nalys ed by the M ACS Q uant Analy z e r 10 Flow Cytomet er (Milt enyi Biot ech) a n d Flow Jo sof twa r e (B D 16 L ife S cienc e s ) to de te rmine NK c ell in f i ltratio n and thei r c yto toxic ity a gain st GB ce ll s . 17 Anal ys is of cytoki nes , c hemokines, growth factor s a nd cytotoxic it y -r el ated f a ctors in the media of 18 c o-cultures 19 Aft er the 24 h incu b atio n, me dia o f t he c o-cul tur es w ere coll ect ed fo r t h e ana lysi s o f sec ret ed 20 c ytoki nes, c hemokin es, grow th fac t o r s a nd c ytot oxic ity-r elat ed fac tor s. Me dia were cen tri fuge d (5 21 mi n, 5,000 rpm) and sup er n ata nt s w e r e store d a t -80 °C . Cytok ine /chem ok ine an al ysi s wa s 22 pe r f ormed in tw o multipl e x a ssay s (H uman Cytoki ne /Che mokin e 96- Plex Di s cov er y As s ay ® A rray 23 (HD96 ) and TGFB 3 - Plex Di scove r y A s s a y® Multi Spec ie s Ar ray (TGF β1 -3) ) by t h e E ve Tec hnologie s 24 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 12 Co r po rati on (C alga r y , AB Can a da) on t he Lumin ex® 200™ pl at form . The fo llo wing targe t s wer e 1 a nalys ed: BAF F , CCL 1, CCL 13, C CL 17, CCL 19, CCL 2, C CL 20, CCL 21, C CL 22, C CL 23, C CL 24, CCL 26, C CL 27, 2 CC L28, C CL 3, C CL 4, C CL5 , C CL7, CC L8, CXCL 10, CXCL 11, CX CL 13, CXCL 16, CX CL 5, C XCL 9, EGF, Eo t a xin , 3 F G F - 2 , F L T - 3 L , F r a c t a l k i n e , G C P - 2 , G - C S F , G M - C S F , G r a n z y m e A , G r a n z y m e B , G R O α , H M G B 1 , I F N - α 2 , 4 I FN β , I FNγ , IF N d=FO, IL-10 , IL -11, IL -12p40, I L-12p70, IL -13, IL-15, IL -16, IL -17A , IL-1 7 E/ I L-25, IL -17F, I L-18, 5 I L-1RA , IL - 1 α, IL -1β, IL -2, IL -20 , IL-21 , IL-2 2, IL-23, IL -24 , IL -27, IL -28A, IL-29, IL -3, I L -31, IL -33, IL - 3 4, IL -6 3 5, IL-4, IL-5 , IL -6, IL -7, IL-8, IL -9, L IF , Ly mphotactin , M -CS F, M IP -1d=FO, PD GF -AA , P DGF -A B/ B B, Pe rfori n , 7 sC D137, sC D40L, SCF , S DF -1, s F a s, s F a s L , TGF- α, TGF -β1 , TGF- β2, TGF -β3 , T N F SF1 3, TN F α, TNF β, TP O , 8 T RAIL, T SLP, and VEGF -A . 9 Anal ys ing N K-9 2 cel l infiltr a t ion in a dy n amic or g an-on -chip pla t for m 10 W e s e t u p a d y n a m i c M I V O ® S i n g l e - O r g a n P l a t f o r m ( M I V O ® p l a t f o r m , R e a c t 4 l i f e ) – a c o m m e r c i a l l y 11 a vail able set up, c on s i sting o f tran swell c hamb ers c arr y ing s pher oid s in Matr igel ( Corning ) under flow 12 c ondition s, mimi cki ng the in flu x of immun e c ell s i nto th e tumou r s . NC H 42 1k o r NIB140 s ph er o id s 13 w ere prep ared by th e fo r c ed fl oa t i ng method ( a s de scribed a bove) and embe dded i n Ma trige l 14 (Co rning ) in the tumou r ch a mber a bo ve a mi croci r c ula tion o f NK -92 ce ll s . N K - 9 2 c el ls a t 15 e ffe ct or: t a r g e t r atio 2. 5:1 we re app lied to th e circu la tion at flow r ate 1 mL / min for 24 h a t 37 °C i n 16 5% C O 2 a tmosphe re. IL -2 w as i nclud e d in the medi a to ac hi eve fi n al co ncent rati on of 200 IU /mL . N K -17 9 2 c e l l s w e r e p r e - l a b e l l e d w i t h 1 0 µ M C e l l T r a c k e r G r e e n C M F D A ( T h e r m o F i s h e r S c i e n t i f i c ) o r 5 µ M 18 Ce llTrac ke r Blu e CM AC (The rmo Fi sh er S c ientif ic) in R PMI me dium wi thout suppl ement s . A fter 24 h 19 of NK -92 cell c ircula t io n, c onte nt s o f t h e tumour ch a mber w e re w a shed wi th c old PBS to re move 20 M a t r i g e l . S p h e r o i d s w e r e d i s s o c i a t e d a n d c e l l s w e r e r e s u s p e n d e d i n 2 % F B S i n P B S . D e a d c e l l s w e r e 21 st ained by 7- AAD (Milt eny i Bio tec ). MA CS Quan t A nalyzer 10 Fl ow Cytom ete r ( Miltenyi Bi ote ch) and 22 F l o w J o s o f t w a r e ( B D L i f e S c i e n c e s ) w e r e u s e d t o d e t e r m i n e t h e i n f i l t r a t i o n o f l a b e l l e d N K c e l l s i n t o 23 the t u mou r chambe rs a nd s p he r oi ds a nd the vi ability/d ea t h of G B and NK-92 c ell s . 24 Ca lcein r e lease ass ay 25 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 13 Ca lce in r el e ase a ssay wa s u s e d t o te st cy t otox ici t y of (IL-2 -pr ea ctiva ted ) NK -92 and heal thy donor 1 (HD ) N K c ell s ag ain st NCH421k and NI B1 40 c ells fr om s tanda rd or sp heroi d cu ltu r e s. I t wa s a l s o u s ed 2 to t e st w he ther N K-92 a nd HD NK cel l c ytotox icity a gain st K562 c ell s i s af fec ted by pre-cul t uring N K 3 c ells in NCH421k or NI B140 c onditi oned me dium. 4 S erial dilution s o f e f fec t o r cel l s ( N K-9 2 or H D N K c e lls ) in at lea s t thre e t e chni c al r e plica t e s w er e 5 prep are d in a r ound b ot t o m 96-w ell p la te (Falc on, C orning) in fina l volum e o f 1 00 µl cor re s p on ding 6 co m p l e t e me d i u m pe r we l l . T ar get c e lls we r e l ab e ll ed w it h Ca l c ei n- A M ( Si g m a) a t f in a l con c e n tr a t io n 7 1 5 µM i n s e rum- fr ee RPM I 164 0 mediu m (Gibco , The rmo Fi sh er Sc i enti fic) f or 3 0 mi n at 37 °C . A f t e r 8 the inc uba tion, c ell s w er e w ash ed and r e s u spend ed in c or re sp onding c omplet e mediu m to a fin al 9 c once ntratio n 0.1 ×10 6 ce l l s/ mL . 5 , 00 0 l a b el l e d t ar g et ce l ls ( 5 0 µ l ) p e r w e l l w e r e add e d t o t h e eff e ct or 10 c ells and c on tr ol w ell s. Th e pla t e w as ce ntri fuged a t 200 g for 2 min and i ncub at ed a t 37 °C a nd 5% 11 CO 2 for 3 h. Aft er inc ubati on, t he pla te w as cen tri fuged at 700 g for 7 min a nd 50 µL of the 12 s u pe r n at a n t f r o m e a ch wel l was t r ans f e rr e d t o a b l ac k 9 6- w e ll p la t e ( T he r mo Fis h er S c ie nt if i c ). 13 F luore scenc e ( ex cita t io n : 496 nm, emi s s i on : 51 6 nm) wa s m ea sure d on a Syn e r g y Mx mi croplat e 14 rea de r (Bi o -Tek I ns tr u m ent s Inc . ). Ba s e d on t he ca lc ein fl uore s c ence me a s u rem e nts, t he p erc en tag e 15 of c yto toxic ity wa s c alc ul at ed a s 100 × ( te st rel e a s e – s pon tan e ou s rel e a s e ) / (t otal rel ea se – 16 spon tan eo u s rel e a s e ). Spon tane ou s rele a se was mea sur ed in w ells c on t a ining 10 0 µl NK-92 c omplet e 17 or NK c omplet e med ium and 5 0 µl c alce i n-AM-la belled t a r g et cell s. F or total r e le as e, 2% Triton X-10 0 18 w as add ed to t he NK -92 com plet e or N K c omplete me dium t o ach ieve c omple te ly s i s of c alce i n- A M -19 l abelle d ta r g e t cel ls. L ytic units (L U ) w ere cal cula t e d u s in g th e inve rse of the nu mb er of e ffe cto r cel l s 20 ne ede d to l yse 30% of th e ta rget c ell s m ultiplie d by 10 0. 21 Ca lce in r ele a se as sa y with block ing of the CD155 axi s was e xec ut ed a s de scrib ed abov e w ith th e 22 a ddition o f C D155 b lock ing antib o die s ( c lone L95, Merc k) , D N AM -1 bl ocki n g antibodi e s (c l one 23 1 02511, R&D S ystem s) , TIGIT bloc king a nt i bodie s ( c l one MBS A43 , T hermo Fi she r S cie nt i fi c) or th ei r 24 c ombina t i on s . Targe t cell s (NCH421k o r NI B140 c ell s) we r e pr einc ubat ed for 45 mi n w it h C D155 25 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 14 bl ocki ng antibodi e s and NK ce ll s w ere p reincub a ted f or 45 mi n with DN AM -1 or / a nd TIG IT block ing 1 a ntibodi e s. All block ing antib odi e s we re used a t conc ent r a t i on o f 10 ug / mL . I s o type c ontro l a t th e 2 same conc entra tio ns wa s u s e d a s con trol . 3 S t a t istical analy s is 4 Unle s s s tat ed oth erw i s e , grap h s wer e plott ed a nd sta ti s tic s ca lcul at ed i n Gra phPad P ri sm v er si o n 5 1 0.4.1. At le a s t thre e biol ogic al replic a te s (N≥3 ) w er e per fo rmed pe r ex perim ent . Dat a ar e pre s ent ed 6 a s m e a n ± s t a n d a r d e r r o r o f t h e m e a n ( S E M ) ; s a m p l e s i z e s a n d s t a t i s t i c s u s e d a r e i n c l u d e d i n f i g u r e 7 l egend s. p v alue sd=FO<d=FO0 . 05 we re con sid ered to in di cat e signi fica nt di ffe ren ce s . The p - v alue s ar e 8 i ndica t e d in the fig ure s a s f ol lows : ∗ p < 0.05; ∗∗ p < 0. 01; ∗∗∗ p < 0. 001 and ∗∗∗∗ p < 0. 0001. 9 Hea t ma p vis ual i s a tion of secr et ed fac t or s a nd hi erarch ica l clust ering ba s ed on the “ one mi nu s 10 P ea rson co rrel ation ” m etr i c w ere pe rform ed i n M o r ph e u s, 11 http s: //so ftw are .br oadin sti t u te. o r g / mor pheu s . 12 3. Res ults 13 3. 1 G B spheroids cultured in the dynamic Celviv o Clinost a r s yste m recap it ula t e t he 14 he t eroge neous ar c hitecture of GB 15 We e st abli shed a n ov el sp her oid model deriv ed f r om hum an GB c ell line s wh ich cl os e ly mimic s t he 16 a r c hitec tur e of GB tis s u e (Fig. 1 a ). To ou r k nowle dge, w e are the fir st to maintain GB sp heroi ds in th e 17 dy namic C elviv o C linos t a r sy stem , w hich provid es a con t roll e d a nd p hysi ologic ally rele vant cultu re 18 e nvironmen t. By ensu ring low shea r str es s and ev en di stributi on of nut r i en ts, it enabl e s long -t e rm 19 sph eroid cul t ur e. In ou r study , sph eroi ds w e r e s ucc ess fully p repa red and mai ntaine d f rom thre e 20 di s t inc t GB c e ll lin e s: N C H 42 1k, a su s p e ns ion G B ce ll line w i th stem-lik e cha r a ct eri s tic s , NI B140, ou r 21 i n-hou se pa tien t- derived adh eren t GB ce ll line, a nd U87 , a c ommonly u s e d adhe r ent GB c el l lin e ( F ig . 22 1 b, Fig . S1), i ndica t i ng a g ener al a ppli ca bility of t h e appro ac h. The p r od ucti on an d cultu re o f 23 sph eroid s in C lin oReac tors wa s r ep rodu c ible and ga ve round and unif ormly s i zed sphe roid s w hich 24 c ould be cul t ur ed in the Cel vivo Clino s t a r s ys tem f or at l e a s t seve r a l we ek s (Fig . S1). 25 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 15 S pheroid s de r i ved from cel l line s NCH42 1k and NIB140 were c haract eriz ed in more det ail a nd were 1 use d fo r furt her experime n ts. Cell li ne s N CH421 k and NI B140 exhibi t ma jo r dif fe rence s in ex pre ssio n 2 of s temne s s-r ela te d gen e s, inc lu ding OL I G 2 , SO X2 , PR OM 1 , CD9, a n d NO TC H ( F i g . S 2 ) a n d w e r e u s e d 3 a s p r o x i e s o f G B s t e m - l i k e a n d d i f f e r e n t i a t e d G B c e l l s , r e s p e c t i v e l y . M a j o r d i f f e r e n c e s w e r e o b s e r v e d 4 i n growth ki netic s of NC H 42 1k and NIB1 4 0 s ph e r o i ds. NCH421k sphe roid s grew much f a ster th an the 5 NIB140 s phe r oid s (Fi g . 1c, Fig. S 1) an d p os se ss ed a l oo se struc ture , whi le th e NI B140 sph eroid s wer e 6 mo r e c ompac t. On d ay 8 i n th e Cl ino Re ac tors, i .e ., th e tim epoi nt a t w hich sphe roid s were u se d f o r 7 fur ther ex pe r i men ts, a pproxima tel y 10% of cel l s w ere de tec ted in ea rly o r la te st ag es o f a pop to s i s in 8 sph eroid s o f both c e ll l ine s (Fig. 1d ). 9 We f u rth er c harac ter ized GB s pher oid s a t the mol ec ular a nd s p a tial le vel by immunofl uo r e s c en t 10 st ainin g of s t e mne ss ma r k e rs SRY -box tr a nscrip tion fac to r 2 (SOX2) and oligode n d r oc yte tr an scripti on 11 fa cto r 2 (OLI G2) , dif fe ren tiati on marker g li al fibrill a r y ac idic pr o tei n ( GFA P) , mes en chy mal subtype 12 ma r k er C D44, prol if era tion marke r pro tei n Ki-6 7, and hypoxi a marke rs hyp oxia ind uci ble fac tor 1 13 subuni t a lpha (HIF1 α) a nd ca r b onic a nhy dra se 9 ( CA9 ). The sta ining was pe rf orme d at two dif fe ren t 14 time point s, i.e ., a fte r 8 days a nd 18 da ys of cultur e in the ClinoR e act or s ( F ig. 2 , Fi g . S3). W e 15 c onfirmed the e xpec ted ex pre ss ion o f G B ce ll mark ers an d di ff er ence s be tw een t he NCH421k and 16 NIB140 cel l line s (Fi g. 2a ). Low level s of d iffe r e nt i a tion marke r G F A P and me s enc h ymal s ubty pe 17 ma r k er C D44 we r e d et ected in NCH 4 21k sph eroid s, whi le bo th w er e pro foundly e xpres s e d in NIB140 18 sph eroid s. In N CH421k s ph er o id s , p r a ctic a lly a ll c ells wer e po s itiv e for s t em cell marker s SOX2 and 19 OLI G 2, while no OLIG2 a nd lowe r l evel s o f S O X2 were ob s erved in N IB140 spher oi ds. Int ere s t i ngly, 20 the expre s s i o n of S OX2 w as no t unifo rm among the N IB140 c ell s, indi ca ting an int rinsic h e t e r og en eity 21 of cell s ta te s . In li ne wi t h t he s l ower gr o wth rate of NIB1 40 sphe roid s compa red t o NCH421k 22 sph eroid s, the a bun danc e o f Ki-67 - po s it i ve cel ls wa s lowe r in th e NI B140 sp heroi ds. In sphe roid s tha t 23 ha d been cul tured i n the ClinoR e acto rs fo r 1 8 day s , h e t e r og en eou s zon es we re o bse rved, inclu ding 24 reg ion s of highl y proli fe rativ e c el ls a s wel l a s some hypox ic a nd nec r o tic region s d ee pe r in the 25 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 16 sph eroid st ructur e (Fig . 2b ). Th u s , o u r GB sphe roid s reca pitula ted the spa tia lly he terog en eou s 1 a r c hitec tur e of GB in v i v o . 2 3. 2 Spheroid c ult ure a f fec ts the e xpres sion of li gands for N K c ell rec ept ors on GB ce lls a nd their 3 sensitivity to NK cel ls 4 A s w e a i me d t o use o ur s p he r o i d m ode l t o s t u d y t he c r oss ta lk b e t w e e n G B a n d NK c el l s , w e ana l y se d 5 t h e e x p r e s s i o n o f s e v e r a l l i g a n d s f o r N K c e l l r e c e p t o r s i n G B c e l l s p h e r o i d s a n d , i n p a r a l l e l , i n G B c e l l 6 l ines f r om s t anda r d c ultu re (Fig . 3 a) . T he ex pr e s si on o f p ol ioviru s re cep tor (C D155 ), n e ctin -2 ( C D112) , 7 B7 homolog 6 (B7 - H6) , inte rcellu la r ad he s i on molec ule 1 ( CD54 / I CA M -1) , U L 16 bi nding pr otei n s 8 (ULBP s ), MH C cla s s I polype pti de -r e l at e d se quenc e s A and B (M IC A -B) , HLA c la ss I his tocompa t i bili ty 9 a ntigen, a lph a c hain A/B /C and E (HL A -A /B/C and HLA-E ), a nd progr ammed c ell death r ece ptor li gan d 10 1 (P D-L 1) wa s ana ly sed . A numbe r o f NK ce ll -r elat ed li gand s were promin e ntly ex pre s s ed on 11 NCH421k and NI B140 c ells a nd seve r a l diffe renc e s we re ob served b etwee n t he two ce ll l ines in 12 st andar d cul t ur e: C D155, B7 -H6, CD54 , ULBP1, an d ULBP -2, 5,6 w ere expre s s ed at highe r lev el s i n 13 NIB140 cell s c ompar ed to N CH421k c ells , wh ich e xpress ed mo r e C D 1 12. I n sphe roid c ulture s , 14 NCH421k c ells expr e sse d more HLA- A/ B / C and C D155 and le ss UL B P-2 ,5,6 and B7-H6 compa r e d t o 15 the NIB1 40 ce ll s . 16 T he expre s sion leve l s of s everal li gand s were signi fic antly a ffe ct ed by the c ulture meth od (stand a rd 17 v s . Cel vivo sph eroid cultu re) (Fig . 3a ). F or ex ample , hig h p erce ntag e s o f CD 1 55 po s itiv e cell s wer e 18 ob serve d in sta ndardly cultu red NIB140 and NCH421k cultu re s (85% a nd 63% , r e spe ctivel y), wh ile the 19 propor tion of C D155 po si tive c ell s w a s m uch low er i n sp heroi d-c ul ture d NIB14 0 and NC H 42 1k cel l s 20 ( 2 0% a n d 3 5 % , r e s p e c t iv e l y) . A s i mi l a r d ecr eas e i n t h e p er ce n ta g e o f l i g a n d- p os it iv e c e l ls in s p h e r o ids 21 w as ob s e rve d fo r ULBP -2 ,5,6 (in both c e l l lines ), B7 -H6 ( i n N IB1 40 s phe r oi d s), HLA-A / B /C (in N IB140 22 sph eroid s ) and U L BP3 (in N CH421k s ph e roid s). 23 T o furt her t e s t wh eth er GB c ell s e n s i tiv ity agai nst N K c ell s i s a ff ec te d by t he c ulture ty pe, cal c ein-24 rel ea se a ssay was p er formed o n singl e cell s di s socia ted f rom GB C elv ivo sph eroi ds a s wel l a s on cel l s 25 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 17 from s t anda r d cu l t u re of b oth ce ll li n e s (Fig . 3b, Fig . S 4a -c) . N K-9 2 c ells ex hibited s l ightly highe r 1 c ytotoxic ity a gain s t NCH421k and NIB 140 c ell s from stan d ard cul tur e c omp ared to ce ll s f ro m 2 sph eroid s; N I B140 c el l s in sphe roid s s ho w ed s i gni fic antly low e r sen si tivity to NK -92 cel l cyto toxi city 3 a t t h e h i g h e s t e f f e c t o r : t a r g e t r a t i o ( 2 0 : 1 ) ( F i g . 3 b ) . C o n v e r s e l y , t h e c a l c e i n - r e l e a s e a s s a y r e a d - o u t 4 show ed no signi f i ca nt dif fe rence s be tw ee n c ytotoxic ity of HD N Ks agai n st cell s from s tanda rd and 5 sph eroid c ul ture (Fig. S 4b-c ). Activ atio n of HD NK s w ith I L-2 inc re a s e d t he ir c yto toxic ity, par tic ularly 6 i n cas e o f the N CH 42 1k ce ll s. In compa r i son, norma l human a stroc yte s we re re si s t ant ag ain st NK -92 7 c ells (Fig . S4d ). 8 3. 3 N K-92 cells show diffe rences in i nf i lt rati on rates between G B c ell li nes in direct c o- c ultures 9 a nd in a dy namic or g an-on-chip platf orm 10 T o s tudy how N K c ell s in ter ac t with G B ce lls in a 3 D microe nv ironmen t, we s e t u p GB sphe roid -NK c el l 11 c o-cul t u re s. NK -92 and HD N K c e lls we r e ap pl ied t o the GB s phe r oi d s from t he C elvi vo C lin o st a r 12 sys tem fo r 24 h. Th en, c o -cul ture s wer e d issoci a ted to s i ngle c ell s or fixe d for t he prep ara tion of F FP E 13 slide s and NK cell infil tra tion wa s me a s ur ed by fl ow cytometry o r immunof lu o r e sce nce stai ning, 14 re spec t i vely ( F ig. 4a ). N K-92 cell s p oorly infilt rat ed th e N CH 4 21k spher oid s, while t h e in filt rati on wa s 15 profo und in t he NI B140 s ph e r oi d s (Fig . 4b). Nev erthe l e ss, infil tr a t i on le vel s w ere simila r at bo th 16 e ffe ct or: t a r g e t ra tio s. In con tra st, a we ak infiltr ation o f H D NK s wa s ob s erved in NIB140 s ph er o i d s 17 a nd s l ightly hig her in th e NCH421 k s p he r o id s (F ig. 4c ) . Al thou gh no t r e achin g s t ati stical sig nific an ce , 18 I L-2 pr e- activ ati on s l ightly imp r ov ed the infilt rati on o f H D N K s . The e xten t o f N K-92 an d H D N K c el l 19 i nfilt r a t i on w a s al so confi rmed by imm uno f l uo r e s c en t d e t e c tion o f C D45-po si t ive NK c ell s i n F FPE 20 sec t i on s o f s p her oid- NK c ell c o-cul tur es ( Fig. 4d-e ). Overall , mo st NK c ell s we re conc entr a t e d n ear th e 21 sph eroid sur f a ce, bu t some NK c ell s we re a l s o ob se rved d e ep within th e sph eroi ds, whic h indic a te s 22 thei r a bili ty t o p e net r a te the ce ll -den s e GB s ph er o id s t ructur e. We a l s o s how ed t hat in fil tra ted NK-9 2 23 c ells proli fe rat ed in the G B sph eroid s (Fi g . S 5). To furt h e r c ompar e the immune - a t t rac t i ng pot enti al 24 of N CH421k a nd N IB14 0 s phe roid s, we a lso se t thei r direct co -c ultu r e s wi t h GB patient PBM C s ( F ig . 25 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 18 S 6). Si milar to NK-92 ce ll s, IL-2 -ac tiva t ed P B M C s in fi ltrat ed at much high er le vels in to NIB140 1 sph eroid s, c ompa red t o NC H421k sph ero ids. 2 T he abili ty o f NK-92 c el l s to in fil tra te GB sphe roid s w a s ad di tionally te s ted in a dyna mic organ -on -3 c hip platfo rm mimic king the blood flow a nd influx of immune c ells f rom blood c ircula t io n i nt o th e 4 tumour (F ig. 4f -h ). N CH421k or NIB14 0 sphe r oi d s w ere emb ed ded in Ma trige l in t he porou s tra ns w e ll 5 i ns e rt and NK -92 cel ls w ere appli ed to t he c ir c ula tion sys tem (Fig. 4f ). A ft er 24 h of c on sta nt NK -92 6 c ell ci r c ulation, fl ow c ytometry was use d to quanti fy NK-92 cel ls th at infil tra te d i nt o the tra ns w ell 7 i ns e rt a nd G B sph eroid s . In line wi t h t he re sult s in direc t c o-cultu re s, a l s o in the se tt in g o f t h i s 8 ph ysiologic ally r eleva n t i n v it r o mod e l, NI B140 sphe roid s exhibi t e d a hig h er attra c tion for N K-9 2 cel l s 9 c ompared to N CH421k sp heroi ds (Fig. 4 g). Import antly, th e i nfil tr a t io n of N K-92 c ells wa s mea s u re d 10 a t t he s am e conc en tra tion of FBS in th e cha mber s for bo th sp heroi d type s , a s FBS per se signi fic antl y 11 i ncrea sed in filt rati on (Fig . S 7a ). The pe rc entage o f dea d GB ce ll s in the in ser t w as slig h t l y higher in 12 NCH421k s ph eroid s co mpa red to NIB14 0 s phe roid s (Fig . S7b) . Int ere s t i ngly, i nfil tr a t e d N K c ell s were 13 l ess viabl e in c hambe rs con taini ng NCH421k s ph eroid s compa r e d to cha mber s wi t h NI B140 14 sph eroid s, bu t v iabili t y was s till app r ox i mately 80% or higher (F ig . 4h). 15 3. 4 N K c ells s how differences in c ytotoxici t y agai ns t GB s phe r oids derive d fr o m d if fe rent c ell 16 li n e s 17 I n a dditi on t o m ea s uring t he abil i t y o f NK c ells t o in fil t r at e GB sph eroid s , we also m ea sure d th ei r 18 c ytotoxic ity af ter 24 h o f di r ec t c o -cultu re (Fig . 5) . I n c on tr a st to the ob serve d sph eroid in filt ra tion 19 l evels , NK -92 c ell s ex hibi ted muc h hig her cy t o tox ici t y aga in s t NCH421k s ph e r oi d s c ompar ed t o 20 NIB140 sphe roid s (Fig. 5 a ). N K-92 c ytot oxic ity aga inst N CH421k s pher oid s incre as ed wi t h i ncrea sin g 21 E :T ratio. IL - 2 -a ctiva te d N K-9 2 cel l s a t th e 5 :1 E:T ratio k illed o ver 30% of N CH42 1k cel ls i n s phe r oi d s. 22 I n t h e ca se o f N IB140 sph eroid s, n o signi fic ant i nc r e a se in th e pr opor t i on o f de ad c ells wa s o bs erved 23 i n the NK -92 co -c ultur e s c ompar ed to c o ntr o l, in dic ating a p r o f o und r e s i stanc e of NIB140 c ell s to N K -24 9 2 cell -medi at ed ly s i s . In te rms o f HD NK c ell c ytotoxic ity, ther e we re sub st antia l dif fer enc e s amo ng 25 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 19 the d onors (Fig . 5b) . Ove rall, simil ar t o t he NK -92 c ell s, al so the HD N K s we re m ore c yto toxic aga in st 1 NCH421k s phe roid s co mpar ed to s ph e r o i ds from N IB140 c ell s, whic h wer e hig hly re sista n t to H D NK 2 c ell-me di ate d killi ng. The cytotox ic ac ti v ity of bo th N K-92 c ell s a nd H D NK s a g ains t N CH421k wa s 3 slig ht l y increa s ed by IL-2 pre -activ ati on. Consi sten t with h igher cytotoxi city of NK-92 and HD NK s 4 a gains t NCH421 k sph eroid s c ompa red to NIB140 spher oid s, we a l s o de t e c ted an inc rea se o f IF N- γ in 5 the medi a o f N CH421k c o-cul tur es , whi l e it s l evel s we r e unch ang ed o r eve n d ec rea sed i n N I B1 40 co-6 cu l t u r es ( Fi g . S8). 7 3. 5 G B spheroids de rived from dif ferent cel l lines s e crete variabl e level s of soluble f a ctors that 8 a f fec t N K cell infilt ration a nd cytotoxic ity 9 S ince t he infil tr a t i on and c ytoto xic a ctivi t y of N K c ells may be i nfluenc ed by solu ble fac t ors rele a sed 10 from ca nce r ce ll s , w e te sted w he the r c ondition e d me dia from GB c ell li nes af fec t s N K c ell 11 c ytotoxic ity. Ca lce in rel ea se a ss ay w as p erfo r me d wi th NK -92 c ell s and HD NK s p re - c ul tured in GB c ell 12 c onditione d medi a fo r 48h prior t o the ons e t of th e as say . K5 62 c ells w ere u sed a s ta rget cel l s. 13 NIB140 - a nd N CH421k -condi tion ed medi a signi f i ca ntly d ecre a sed cy totox ici t y of NK-92 c ell s by ove r 14 o r n e a r l y 5 0 % , r e s p e c t i v e l y ( F i g . 6 a ) . I n p a r a l l e l , l o w e r l e v e l s o f a c t i v a t i n g r e c e p t o r N K p 3 0 w e r e 15 de t e c ted on NK-92 cell s cu l t ured in GB cel l-condi tio ne d medi a (Fi g. 6b ). N I B 140 (bu t no t NCH421k) 16 c onditione d medium a lso i ncrea s ed NKG 2D a nd dec r e a s e d P D-1 on N K-9 2 cel l s (F i g. S9). Co ndition e d 17 me dium from eith er o f th e c ell l ine s did not a ff ec t the cy t ot ox icity of HD NK s (Fig . S 10), pr obabl y du e 18 to h igh ac tivatio n of HD NK c ell s. 19 T o f u rthe r in sp ect the po ssibly immuno modul ating fac tor s rel e a s e d f r om G B ce ll s , w e an aly sed th e 20 c once ntratio n s of c yt okin e s , c hemokin e s and cy t o t ox ici t y -r e l at ed fa ctor s in me dia s upern ata nt s o f 21 G B s p he r o i ds a n d t h e ir 2 4 h c o- cu l t u r es w it h N K- 9 2 a n d H D N K c e ll s ( F i g . 6 c -f , Fi g. S1 1- 1 3 ) . S p h e r o ids 22 de r i ved fr o m NC H421k (i. e., G B st e m-like ) an d NIB140 (i. e. , di ff eren tia t ed) GB c ells v aried 23 sign ific antly in thei r c ytokin e se cre tion p rofil e s (Fi g . 6c ) . W h il e sphe roid s fr om bo th cel l lin e s secr et e d 24 h ig h l e ve l s o f v as c ul a r e n do t he l ia l g r o wt h f ac t o r A (V EG F-A ) , h i g h m o bi l i t y gr o u p p r o t ein B 1 ( H M G B 1 ) 25 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 20 a nd tr an sfo r mi ng growth fac to r be ta (TGF -β ), N IB140 sphe roid s secr et ed much more p r o -1 i nflammat ory c ytokine s such a s C -C moti f c hemok ine liga nd 2 (C CL2), IL -6 , IL-8, st r oma l cel l -derive d 2 fa cto r 1 ( SDF -1) , and mac ropha ge c olon y- s timul ating facto r ( M- CSF ) c ompared t o N CH421 k 3 sph eroid s. 4 T he s ec r e tio n of s ev er a l f ac tors was modu lat ed i n the media of GB sp her oid- NK c ell co -c ultu re s 5 c ompared t o mono -c ultur e c ont r ol s (Fi g. 6d-e, Fig . S11 , F ig. S12) . In bo th N K -9 2 an d HD NK c ell 6 c ocul t ure s wi t h GB sphe roid s, cy t ot ox icit y - rel a ted facto r s includ ing granzyme s A and B, per for i n, and 7 solubl e F a s l igand ( sFa sL) clu ste red with IL-16 a nd C CL5 (Fig . 6 d -e ). Addi t i onal ly , CC L3, C CL4 , I L-10 , 8 a nd l ymphotac t i n w er e w ithin th e cl ust er in NK-92 co -c ultu r e s . IL -8 and M -CS F were signi fic an tly 9 up r e gula ted i n all co -cul t u re s, ex cept fo r NIB140 sph eroid s wi t h HD NK s (Fig. S12 ). Downr eg ulation o f 10 tumou r ne c ro s i s fa c tor b eta (TNF -β ), IL -1 0, IL -13, p er forin, S DF - 1 , s F a sL and VEGF - A, and up reg ulati on 11 of C - X -C mot i f chemok in e li gand 11 (CX C L11 ) were speci f i call y obs erved i n N K -92 co-c ultu r e s, bu t not 12 HD NK co -c ul ture s. P lat ele t-d erive d grow th fac to r ( P DGF ) w a s s ig ni fica nt l y d ow nre gulate d o nly in N K-13 9 2 co-c ulture with NC H 42 1k. Ove rall, th e m ost s t rik ing upr egul ati on s w ere o bs e r v ed for IL-8, CX CL9, 14 CX CL11, C CL8, and CX CL 10 in N K-92 cel l c o-c ultu r e s w ith N IB140 sphe roid s (Fig . 6 f, Fig . S13 ). 15 3. 6 G B pa t ie nt N K c ells expres s l es s ac t iv ati ng NK cel l receptors and TIGIT c ompa re d to N K cells 16 fro m hea lt hy donors 17 As t he func tion o f NK c ell s al so signi f i c antly de p end s on the pre sen ce o f a c tiva t i ng and i nhibito ry 18 rec ep tor s on t he i r s ur fac e, we analy sed the expre ssi on of selec ted N K rec e pto rs o n IL-2 ac tivated N K -19 9 2 cell s , HD N K c ells and , i n addition , GB patie n t NK c ells ( F ig. 7). T he ex pre s sion of rec epto rs vari ed 20 be t w ee n th e thre e NK c ell s o ur c e s (Fig . 7 a, F ig. S14 a ). NK -92 cel l s exp r e s sed a ctiva t i ng rec ep tor s 21 NKp30 an d NKG2 D, bu t als o inhibit ory (co)r ec ep tors C D94, CD 9 6 a n d T IG I T . A majority of HD NK s 22 e xpre ssed F c rec epto r C D16; varyin g lev el s o f ac tiv ating rec e pt o r D N AM -1 and i nhi bitor y rec ept or s 23 CD94 , TI GIT, and PD -1 w ere a l s o de tec t e d. C ompa re d to HD NKs , GB p ati ent NKs ex pre ssed le s s 24 CD16 , NKp30 , T IGI T, N K G2D, and D NA M -1, a nd mo r e i mmune ch e ck-po in t P D -1 (Fig . 7a, Fig . S14b) . 25 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 21 E xpec tedly, IL - 2 a c tivatio n of H D NK s si gnific antly inc r e a s e d the ex pre ssi on o f a ctiva t in g r ec epto r s 1 NKG2D an d NKp30 and de cr ea sed the e x pre ssion o f P D -1 (Fig. 7b, Fig . S14 c) . 2 3. 7 The CD155-T I GIT/ D NA M- 1 axi s is an imp ortant media t o r of H D N Ks’ cy t otox icity a gainst 3 NCH421k cells 4 As hi gh C D 15 5 ex pr essi on was ob s erv e d in st a nda rd a nd sph eroid cultu re s of G B ce lls and sinc e 5 de crea s ed expr e ss i o n o f C D 15 5 rec ep t or s T IGIT an d DNAM -1 wa s d e t e c ted o n NK c ell s from G B 6 pa tien t s , w e further ex plored the role o f t h e C D 1 55 axi s i n regul a tion of N K c ell c ytotoxic ity aga in s t 7 GB c ell s. Acc ording to t he TC G A a nd C G GA da ta , PV R (th e g en e th at e nc ode s CD1 55) is upregul ate d in 8 GB compa re d t o low-g ra de glioma s (Fig . 8a, F ig. S15 a ) a nd hig h level s o f PVR co r rela t e w it h poo r 9 prog no s i s of g lioma p a tien t s (F ig. 8b , Fig . S15b). 10 We u sed ca lce in r e le a se a s say t o te st th e e f f e ct s o f antibo dy -ba s ed bloc k ade o f C D155 o n 11 NIB140 / N CH421k c ell s and i ts r ece pto rs TIGIT o r D NA M -1 on H D NK ce ll s (Fig . 8c, Fi g. S1 6). The a s say 12 w as p er formed with N K s f r om t hre e h ealthy d onor s, w hich ex pr e ss ed varying le vel s o f TI GIT and 13 DNA M- 1 but did not expre s s CD96 (Fig . S 17). Bl ock ing comp onen ts of t h e C D1 55 a xis h ad a much 14 g r e at er e f fec t on cy t ot ox icity ag ains t N CH421k ce ll s (Fi g. 8c ) compar ed to NIB 140 c ells (F ig. S16) . 15 Bl ocki ng CD155 d ec r e a sed c ytot oxic ity ag ains t N CH421k c ells i n N K s f r o m tw o o f th ree d onor s ( F ig . 16 8c ) . B l o ck a de o f D N A M- 1 c ons is te n t l y d e c r e as e d HD N K ce l l c y t ot o x ic i t y ag a i ns t N C H 4 2 1k c el l s in N Ks 17 from al l t h ree dono rs, whi le bloc king T IG I T upregula ted NK c ell ac tivity onl y in the ca s e o f NK s from 18 he althy dono r t hre e, w hich ex pr e ss ed the hig hes t lev el of TI GIT amo ng the do nor s. S imultan e ou s 19 bl ocka de of C D155 and TIG IT or, to a g r e at er ex t e n t , C D155 and D N AM -1, s e v erely impai r e d th e 20 a bility of HD NK s to elimi n ate NCH 42 1k c ells . 21 4. D iscussion 22 I n thi s s tudy, we e s tabli sh ed a p roto col f or rep roduc ibl e prod uc tion and d ynamic cu lt u r e of u nifo rmly 23 s i z ed G B s ph e ro i ds in t h e C e l vi v o C li n os ta r s ys t e m. U n i f or m s p her o id s iz e i s p ar t icu l a r l y im p o r t ant f or 24 rel iabili ty and rep roducibi lity o f fu rt h e r s tudi e s , s uc h a s a sse s s ing the r e s pon s es o f sphe roid s t o 25 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 22 ther ap ie s , in whic h their va r i abl e size may sign ifican t ly di st o rt ex perim enta l re sult s (28). A n o t he r k e y 1 a dvantag e of our a ppro ach i s t hat t h e spheroid s i n th e Celv ivo Cli n ostar s y stem c an be c ul ture d for 2 sub sta nti ally lon ger time p eriod s comp a r ed to o the r pro tocol s in whic h sphe r oids a r e ma inta ined 3 un der st a t ic c onditio ns . Thu s , our me t hod i s well s u ite d fo r sp eci fic l ong -t er m ex periment s and 4 stu die s of c hr onic tre atmen t s. I n the pr es ent s tu dy, sph eroid s wer e produce d f rom seve ral di stinc t 5 c ell li ne s , in cludin g s u spen sion -cul t ure d as wel l a s a dher ent GB ce ll lin es , imply ing a broad 6 a pplic ability o f t he app roa ch . In time, t he archi t e ctu re o f ou r sphe roid mod els beca me incr ea s ing ly 7 he t e r og en eou s. L ike GB tum our s i n v iv o , ou r mo del s d evel oped nec ro t i c an d hypo x ic reg ions an d a l so 8 (a t l ea s t par tly) re capi tula te d the h et er oge neity o f canc er c ell s t a t e s . Spe ci fic al ly, di stinc t z one s o f 9 hi ghly and poo rly proli fe r a t i ve cell s wer e ob se rved and ce ll s ex pr e ssed va ryin g l evel s o f st emne s s 10 ma r k er SOX2. 11 Ca ncer c ell sph eroi ds r epr oduc e th e 3D conforma tion o f th e tumour , incl uding c ell-c ell in te rac t io n s, 12 e xtracel lul ar ma t rix , and gr adie n ts o f nutr i en t s and ox yge n. A s immune c ell func tio n i s sev erely 13 a f f e ct ed by nut r i en t an d ox ygen dep riv ation a nd s uc h c onditi on s are c ommon in t umo urs du e t o 14 thei r e xt en sive grow t h a nd abn ormal vascula t ure (29 –32) , usi ng sphe roid model s in i mmunothe rap y 15 r es e ar c h m a y r e ve a l i ns i g ht s i n t o i m m u n e c e ll r es po ns es un d e r c ondi tion s clo s e t o th ose ob s e r v ed in 16 vi v o . Im portant ly , al so the e xpr e ss i on o f immune -rel at ed mol ec ule s on canc er c ells c a n b e af fec ted 17 by the c ultur e type (2D vs . 3D ) (33) . I t has p r ev iou sly bee n repor t e d that GB c ells c ultu red in 3 D 18 u pr eg u l at e t h e e x p r ess io n of im m u n os upp r ess i v e m o l ec u les , s uc h as H LA - E and M ICA /B ( 3 4 , 3 5) . I n 19 l ine, w e o bs erved c on sisten t downr eg ula tion of li g and s f or seve ral ac tiv atin g NK c ell r ec ept or s 20 (C D155 , B7-H6 and ULB Ps ) in GB sph eroid c ulture s . In ter e stingly , major dif fer enc e s were a l so 21 ob serve d in the NCH421k c ell lin e, whi ch is cl a ssic ally cul t ured in th e form o f small mul tic ellula r 22 a ggrega t e s, imply ing th at n ot j u st int er -c ellu lar c on nec ti on s , bu t al so s phe r oi d size i s an im portan t 23 fa cto r in f l uenci ng t h e c anc er c ell immu ne phe no type. Compa red t o the standa r d adhe re nt NIB14 0 24 c ulture, ce ll s i n NIB14 0 s phe r oi d s ex pre ssed signific an tly lo wer leve l s of HLA- A/ B / C . D ownreg ula tion 25 or lo s s o f HLA cla ss I m ol ecul e s i s o ne of the ma in m echa ni s m s of t u mou r immu ne eva sion, w hic h 26 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 23 hi de s canc er cel l s from rec ogni tion by T ce lls, bu t may simultan e ou s l y inc r ea s e the ir s en sitivity t o NK 1 c ells (36). 2 Ba sed on the dif fe renc es in ex pre ssi on of s t emne ss r el at ed gen e s, N CH421 k an d N I B140 s ph e r oi d s 3 w ere u sed a s p roxie s of GB ste m-like and dif fer enti at ed GB c ell s, r e s p ec tivel y . We s how ed t ha t 4 NIB140 s ph e r oi d s (i .e . , di ff ere ntia ted GB ce lls) p r odu ced muc h hig he r level s of pr o-in fla mmat or y an d 5 i mmune-a ttra cting facto r s (CC L2, I L -8, IL -6, S DF-1 , M - C SF ) c ompar ed to the s t em-lik e NC H421k 6 sph eroid s. A s al so in dic at ed by o the r s ( 37 ) , o u r r e sult s st r ongl y s ug ge s t that di f fer enti ate d GB cel l s 7 should b e c on s i de r e d in d evel op ment o f nov el immunoth erap eutic ap proach e s . Although G B stem 8 c ells a re gen erall y vi ewed a s t he mo st c ritical ther apeu tic t a r g et s , d if f e r e n t i a ted GB ce ll s may 9 sign ific antly a ffe ct the imm une l and sca pe w ithin t h e TME , shap e ada p tive im mune re spon se s and 10 thu s als o influ e nce the suc ce s s of i mmu nothe rapie s . For ex ample, C CL2 ma y attr act seve ral e ffe cto r s 11 of i mmuno sup pr e ssion , inc luding mac ropha ge s, regula tory T cel l s (Treg s ) , a nd myeloi d -derive d 12 s u p p r e s s o r c e l l s ( M D S C s ) , ( 3 8– 4 0 ) . S i m i l a r l y , M D S C s c a n a l s o b e r e c r u i t e d b y I L - 8 ( 4 1 , 4 2 ) a n d t h e i r 13 a cc umulation a nd immune inhi bi t o ry f u n ction s are enh anced by IL -6 (43) . 14 I n t h e 24 h di r e c t co -c ulture o f GB s pher oids and N K cell s, w e ob se r v ed sub s ta nti a l dif fer enc e s i n N K-15 9 2 a nd HD N K c ell in filtr ation and cy t o tox ic c apac itie s. Whil e the in filtr ation o f HD N K s cor rela te d 16 w ith their c yto toxic ity, no s uc h tr end w as ob s e r v ed w ith NK - 9 2 ce lls . Eve n though N K-92 cel l s 17 profo undl y kil led NCH421k c ells in th e sphe roid s, f ew NK-92 c ell s were in f i ltra ted within th e 18 sph eroid s. Conve rsely , N K -92 c ell s s u c ces s f u lly infil t ra ted in to t he de ns e s t ruc t u re o f N IB140 19 s p h e r o i d s , w h i c h w e r e h i g h l y r es i s t a n t a g a i n s t N K - 9 2 m e d i a t e d k i l l i n g . T h e s e r e s u l t s s u g g e s t t h a t t h e 20 c ytotoxic ac t i vity doe s no t nec e s s ar ily correla te w ith the abil i t y of N K cell s to pen e t ra te the sphe roi d 21 a nd t ha t me a suring NK ce ll infil t r ation and th eir cyto tox icity s hould b e con sid ered a s tw o di s tinc t 22 rea d -ou t s . 23 T he r em a r k able in fil t r atio n o f N K-92 c e lls i n t o N IB140 s p h er o id s co mpar ed to NCH 4 21k s p h er o id s 24 ma y be attribut ed t o a nu mber o f pote ntia lly NK cell a t trac t i ng cytok ines sec re ted a t hig h level s in 25 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 24 NIB140 s ph e r oi d cul t u re s. The s e i nclu de , for exa mple, C CL 2 (44 ), IL-8 (45, 4 6), an d S DF-1 (47 ). 1 Addi t i on ally , in NK -92 c o-cul tur es wi th N IB140 s pher oid s th ere wa s signi fic ant inc rea se in l ev el s o f IL-2 8, C C L8 and IF N- γ-i nd ucible c ytoki ne s C XCL 9, CX CL1 1 a nd CXCL10 . C X CL9, CXCL 10 and CXCL 11 may 3 fur ther pot enti at e NK c ell i n filtra t i o n a s th ey bin d to CXCR3, a re cep tor wi th a fun dame nt a l rol e in N K 4 c ell tr af fick ing (48). S p ecula t i ng th at t he CXCR3 liga nd s a re a l so upr eg ulat ed up on int erac ti on o f GB 5 an d N K c e lls in G B TM E i n viv o , they coul d als o a ttra ct ac t i vated T cell s an d oth er leukoc ytic s u btyp e s 6 w hich are kno wn to ex pre ss CXCR3 . Of note, CXCR3 i s al so ex pre ssed on G B c ell s and it s i sof orm s may 7 ha ve e ith er pro - or a nti -tumo r i genic func tion s (49 ). 8 T he c ytotoxic poten tial o f both NK-92 a nd HD N K cel l s was hig he r agai n s t NCH421k sphe roi d s 9 c ompared to NI B140 sph eroid s. T hi s i s i n line with a numb er o f studi e s w hich d emo nstra te th at N K 10 c ells c an recog nize and k ill GB stem -li ke c ells in vit r o a nd tha t G B s t em-lik e c e ll s a re mor e su s c eptibl e 11 to NK ce lls c ompar ed to thei r dif fer enti a ted c ount er p a r t s (19,50,5 1 ). We ha ve sh own that GB s tem -12 l ike cell s upon d i f f e re nt i a tion w ith se r um and IFN- γ beco me mor e r e s i st ant to NK-c ell m ediat ed 13 k illing (19). Int ere sting ly, de s pi te a pr o fo un d r e s i s tance o f NIB14 0 s phe r oi d s ag ai nst H D NKs a nd N K -14 9 2 cel ls , bo th N K cell s ou rc e s show e d dos e-d epen de n t cytotoxi city a gain st single NIB14 0 cel l s 15 d i s s o c i a t e d f r o m s p h e r o i d s i n c a l c e i n r e l e a s e a s s a y . T h i s m a y i m p l y t h a t t h e a c t i v i t y o f N K - 9 2 c e l l s i s 16 spec i fica lly ob str uc ted within t he 3D en vironmen t o f NIB140 ce ll s an d h ighlig hts th e importanc e o f 17 the 3 D re s e arch mo del s. In v i v o , th e im mune phe not y pe o f G B -in fi ltrating N K c ells i s di s tinct to N K 18 c ells f rom matche d per ipher al bloo d. W i t hi n t h e tumou r, N K c ells downr eg ulat e ac tivating r ec ep tor s 19 a nd re duc e I FN -γ s ecre tion (52–54 ) , w h ich may al so occ ur upon in fil trat ion in to GB s p he r o id s in 20 vi t r o . In the sph eroid s , N K -92 cel l ac tiv it y may be sup pre sse d by eit her 1 ) phy s ic al inte rac ti on s within 21 the cell -de n s e s t ructur e of th e sph eroid or/an d 2) by NK- sup pre s s iv e fac t o rs rel e as ed by GB cell s tha t 22 de mon s tr ate a g r e a ter e f f ec t in t he 24 h co - c ul ture exp er i m ent compa red t o the 3 h c alc ein r elea se 23 a ssay. Ind e ed, NI B140 s p h er o i d s (but n ot NC H421k sph eroid s ) sec re t e d high le v els o f IL -6 a nd IL-8, 24 w hich are know n t o i mpair N K c ell a ctivi ty (55) . Comp ared to NCH 4 21k sph eroid s, N IB1 40 s ph er o id s 25 a ls o sec r e t e d mo re TGF - β1, one of the major s up p r e s sors of NK cel l fu nction i n G B (24 ,25). Cul turin g 26 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 25 NK-92 c ell s in N IB1 40 (a s w ell a s N C H421k ) c ondition e d medium inh ib ite d t h e ir c ytotoxic ity aga in st 1 K5 62 c ells , indic a ting a pro fou nd ef fec t of G B solubl e fac t ors in r egu latio n o f NK-92 c ell ac t i vity. 2 I nt e r e sti ngly , neithe r NIB140 nor N CH 4 2 1k c onditioned me dium af fec te d t h e cyt otox icity of HD N K s , 3 i mplying that (compa r e d to N K-92 cell s) the se ce ll s a re in hig h ac t i va tion st ate an d ma y be l ess pr one 4 to i nactiv a tion by G B s olub l e fac t or s . 5 T he rep ert oire of ac tivating an d in hibi tory rece pt o r s on the NK c ell sur face i s a n i mportan t f ac t o r tha t 6 l argely a f fect s N K c ell func t i on . N K cell r ec epto r s may be dysregula t e d on N K cell s in c anc er p a tien t s 7 (56 ,57) . In spec t i n g NK cel ls from GB pa t i en ts rev eal ed tha t , compar ed t o H D NKs, the se cell s 8 e xpre ssed l e ss C D16, NKp30, T IGIT , NKG2D, and DNAM -1 and mor e PD -1. Down r eg ula tion o f 9 a ctiva t i ng r e cep tor s on GB patie n t NK s was desc ribed pr eviously (23 ) and we previ ou s ly repor ted 10 redu ced c yto toxic ity and dec rea s ed IFN - γ secr etio n in GB pati ent NK s w hen co-c ultured with c ance r 11 c ells (19) . H ow eve r, our a tt enti on wa s c aug ht by r edu ced l ev el s o f TI GIT and D NAM-1, w hich b oth 12 bi nd CD155 – a lig and tha t wa s hi ghly expre s s ed in s t a nd ard a s wel l as s ph e r oi d G B cultur e s. W e t hu s 13 de cide d to furthe r i nspec t the rol e of the CD 1 55 ax is in NK ce ll cy tot o xici t y aga in st GB c ell s. 14 CD15 5 (al so know n a s t h e p oliov iru s r ec eptor o r “ PV R”) i s a ty pe I tran s me mb ra ne gl ycoprot ein from 15 the n e ctin and n e ctin -like pr o tei n f a mily . I ts expre s s i o n i s up regu la ted i n many c anc er s w hich o ft en 16 c orr e la te s w i th po or p atie nt p r o gno si s ( 58,59 ). I n G B, C D155 ha s bee n a ssoci a te d with a numbe r o f 17 di verse func tion s s uc h a s ca nce r ce ll adhe sion, migra tion, inv a sio n, and p roli fe r a t i on (60 –62) . 18 I mpor t an t l y, CD155 al so regul a te s im mu ne re s po n se s of N K c ell s and o ther ly mp hocy t e s . I t c an bind 19 to seve ra l rec ep t o r s, inc luding t h e inhibi t ory T IG I T, the ac tivating DN A M-1 , a nd C D96, a rec ep tor with 20 du al role s (58,59). In specting t he rol e of the C D155 ax is in reg ula tion o f NK c ell c ytotoxic ity aga ins t 21 GB c ell s s how ed th at although the e xpr es s io n o f th e li gand w a s h igh er on NI B1 40 ce lls comp ared to 22 NCH421k cel l s , bl ocki ng CD155 o r i ts r ec ep t or s only s how ed co nvinc ing ef fec t s o n NK c ell c ytotoxic ity 23 a gains t N CH 42 1k c ells . 24 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 26 T he inhi bito r y recep tor TI GIT gene rall y h as a stronge r af fini ty for CD 1 55 co mpa re d wi t h t he ac tiv ating 1 rec ep tor D N AM -1 (63 ). Alt hough TI GIT a nd, to a gr eat er exten t , DN AM - 1 wer e b oth e xpr e ssed on NK 2 c ells f r o m a ll thre e heal thy don ors, a d e c rea se in cy t o tox icity u pon block ade o f CD155 on NCH421k 3 c ells sugg e sts tha t i n n et e f f e c t, the li ga nd predomi n antly induc e d ac tiva ting signal ling . Block ing 4 DNA M- 1 signi fi cantly de cr ea se d H D N K c ytotoxic ity agai nst NCH421k c ell s, with an ef fect stronge r 5 tha n t h a t ob s erved w ith bloc king C D155 i t s elf . Thi s impl i e s tha t D NA M -1 prob ab l y also ac t i va te s N K 6 c ells by bi nding ot her lig a nd s , e .g., C D11 2, w hich we al so d et ec t e d on N C H 42 1k ce lls. Bl ocki ng TI GIT 7 on ly mode r a t e ly inc r ea sed cy t ot ox ici t y o f NK s from h e alt hy don or thr ee, whic h e xpres se d th e hig he st 8 l evel of TIG IT ( a nd D NAM -1) am ong t h e thre e donor s , wh ile it showe d no e ff ec t in NKs f r o m th e 9 othe r tw o donor s. O ur o bse r v ati on s a re i n li ne wit h a r e cen t s t u dy wh ich c onclu d ed th at T IG IT pe r s e 10 do e s no t di r ec tly inh ibit N K c ell s upon i n t e r a c tion wi t h C D 15 5 on GB cell s , bu t r a ther ac ts a s a d ec oy 11 rec ep tor, p r e ven ting the i nte rac ti on wit h DN A M-1 (64) . Thu s, bloc king TI GIT a lo ne may n ot r elia bly 12 re sto r e NK ac tivity ag ain s t GB . All in all, our re s u lts may im ply that the C D15 5 a xis is an i mpo r ta nt 13 reg ulat or o f NK c ell c ytot oxic ity agai n s t G B st em-lik e c ell s, a ltho ugh addi tio nal cell l ine s should b e 14 te st ed fo r c onfirm ation o f ou r finding s . Never thel es s , w e can c onclud e t ha t the importanc e o f th e 15 a xis dep end s on the ove rall i mmune ph enotyp e of c anc er c ell s con si s t i ng of a d ivers e r e p er t oi re o f 16 a ddition al ligan d s fo r NK c ell rec e pto r s . 17 5. Conclusions 18 I n the pr e s e n t s t u dy, we e sta bli shed a g eneral pr ot o col for repr oduci ble pro d uction and dyna mic 19 c ulture of unif ormly s i z ed GB sph eroid s in the Ce lvivo Clino st ar sy stem . T he mod el w a s u sed t o 20 e xplore t he cro s stalk be tween GB a n d NK cell s. S ev eral lig and s fo r NK c ell rec e pto r s w ere 21 di ffe re nt i ally ex pr e s se d i n s pher oid s c ompared to cel l s in s ta nda r d cul t u re, af fec t in g t hei r 22 i nt e ra ction s w ith N K ce ll s. Al thoug h NK ce lls in fil trat ed G B spher oid s, th eir in fil tr a t i on c apa ci ty did 23 no t ne ce s s a r il y correla t e w ith their a bili ty t o e limina t e GB c ell s. G B c ell li ne s sec re t e d vari abl e leve l s 24 of s oluble fac t ors tha t i mpaired NK c ell c ytotoxic ity. Sphe roi d s derive d from di ff eren tiat ed GB cel l s 25 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 27 s e c r e t e d h i g h e r l e v e l s o f i m m u n e - a t t r a c t i n g , p r o - i n f l a m m a t o r y a n d N K c e l l - i n h i b i t i n g c y t o k i n e s 1 c ompared to G B stem - l ike sph eroi ds , ind ic ating a profo un d abili ty of dif fe r e n tiat e d GB cel l s to shap e 2 the GB i mmune mili eu. L a stly , the C D155-TIG IT/ DNAM -1 ax is wa s f ound to b e an i mportan t medi a to r 3 of NK ce ll cy t ot ox icity a gains t G B st em-li ke c ells . Colle c tive ly, our re s ul ts highli ght import ant f a c tor s 4 i n the G B- N K c ell communi cation and p re sen t a ground wo rk for fur t h er t a r g e ted r ese arch . 5 6 Abbrevia t i ons 7 AD CC – a ntibody de pend en t cellu l ar cy t o tox icity 8 B7 - H6 – B7 homol og 6 9 BS A – bov ine se rum albu min 10 CA9 – ca r b o nic a nhydra s e 9 11 CC L – C -C moti f chem okine ligand 12 CD11 2 – n ec tin -2 13 CD15 5 – pol i oviru s r e ce ptor 14 CD16 ( Fcγ RI II ) – c lus t e r o f di f fere ntia tion 16 ( F c g amma rec epto r II I) 15 CD44 – CD44 pr ot ein 16 CD54 ( IC AM-1) – cl u ster o f di ff eren tia tio n 54 (interc e llular ad he s i on mol ec ule 1) 17 CD96 (T ACTILE ) – cl uster of di ff ere ntia t i o n 96 (T-c ell sur face pro tei n tac tile ) 18 CX CL – C -X-C moti f che mokine ligan d 19 DNA M- 1 – DNAX Acce ssory Molec ul e-1 20 ED TA – e th y l en e d ia m in e te tr aa c et i c a ci d 21 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 28 F BS – fo etal b ov ine s erum 1 F FPE – fo rmalin fi xed p a r a ffi n embedd e d 2 GB – gliobl a stoma 3 GFA P – glia l fib rilla r y ac idic pr o tei n 4 HD – he a lthy do no r 5 HIF 1α – hy poxia induc ibl e fac tor 1 s ub un it alpha 6 HL A – human leuko cyte a n tigen 7 HMGB1 – hi gh mobi li ty group p r ot ein B1 8 I FN-γ – in ter fe ron -gamm a 9 I L-2 – int er l eukin 2 10 Ki -67 – proli f e ra t i o n marker pr ot ein K i-6 7 11 KI R – ki ller ce ll immu noglobuli n -like rec e ptor 12 M- CS F – ma croph ag e col ony- s t im ulati ng fa cto r 13 MIC A – MHC clas s I p oly peptid e -rel ate d seque nce A 14 MICB – MHC cla ss I po ly peptid e- rela ted sequ ence B 15 NHA – no rmal human a s t rocy t e s 16 NK– n atu r a l kil ler 17 NKG2D – NKG2 -D t y pe II in t e gr al mem bra ne pro tein 18 NKp30 (NCR3) – natu ral c ytot oxic ity trigge r in g rece p t or 3 19 OLI G 2 – olig od endr ocy t e tran sc r ip tion f a ctor 2 20 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 29 P BMC s – peri pher al bl ood mononuc l ear c ells 1 PB S – p hos p hat e - buf f er e d s a li ne 2 P DGF – pla tel e t-der iv ed grow t h fa cto r 3 P D-L 1 – progr amm ed ce ll dea t h 1 liga nd 1 4 S DF-1 – s tr o mal c ell-d erive d facto r 1 5 S EM – s tanda rd er ror o f th e mea n 6 sFa s L – solubl e Fa s liga nd 7 S O X2 – SRY -box tr an scripti on fac tor 2 8 T G F -β – tr an s f orming growth f ac tor b eta 9 T IG I T – T cell immunorec epto r with Ig an d ITI M domain s 10 T M E – tumour mic roenvir onme nt 11 T N F -α – tumou r nec r o si s facto r- α 12 TN F- β – t u m o ur n e cr os is fa ct or b et a 13 UL BP – UL16 binding pro t e in 14 V EG F - A – v as c ul ar e n d ot h e l ia l g ro wt h f ac t o r A 15 16 Dec larations 17 Ethic s approv al and c ons e nt t o par tic ipate 18 T he st udy w as app rov ed by the N atio na l Medi cal Ethic s C om mitte e o f the Republi c of Sl ov eni a 19 (a ppro val N o . 0120-19 0 / 20 18-2711 -41 ) . Exp er i ment s with human bu ffy c oat s were pe rfo r me d i n 20 a cc or da nce with the a pp r o val no . 0120 -2 79 / 201 7-3 . 21 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 30 Consent for publication 1 Not app lic ab le . 2 Availa bility of data and materials 3 No da ta set s wer e gene rat ed or a n alys ed during the c urren t s tudy. 4 Competing int e r e s ts 5 T he autho rs d ecl ar e no c ompeti ng intere s t s. 6 Funding 7 T his s t udy was s up ported by th e S lov eni an R e sea rch and Innov a tion Agen c y (programm e and 8 re sea rch grant s P1 - 0 245, J3 -4504 , NC-25 002, N3-0394, and youn g re s e a r c he r gra nt to AH ) , Europe an 9 Union’ s projec ts CutC anc er 1010 7911 3, SP ACE TIME 10113 6552, UNCAN - CONNECT 10121 5206, 10 Genom e MET 22HL T06. 11 Author s' c ont ributions 12 A.H. , M.N, and B .B. conc ep tuali s ed the s t ud y. A.H ., T.K.M. , P.Ž ., B .M. , Š.K. , E.S. , a nd M.P. N. de sign ed 13 a nd p erform ed ex perime nt s an d a n a lysed t he o b t a ine d dat a. U. Š., A .P ., and B. P. prov ided 14 P BMC s/blo od s a mple s fr om hea l t h y do nor s and GB p atie nt s a nd perfo rmed ex perimen t s. M.N and 15 B. B. sup ervi sed the study . A .H., M. N., a nd B.B . w rote th e fi r s t d raf t o f th e ma nuscrip t. All au thor s 16 rev iewe d, edi ted a nd appr ov ed the sub mitted man u s c r i p t . 17 Ack nowledge ment s 18 We ack nowled ge t he suppor t o f CO ST ac tion s IMM U N O -mod el C A21135 and Ne t 4Brain C A2 2103. 19 20 6. Refer e nces 21 1 . van den Bent M J, Geur t s M, Fr ench P J , S mits M, Capp er D , Bromb erg J E C , et al . P rimary brai n 22 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 31 tumours in adul t s. La nce t. 20 23;402(104 12):1564– 79. 1 2 . Miller KD, O s tr om QT, Kruc hko C, P atil N , Tiha n T, Ci offi G , et al . Brai n and oth er c entra l 2 nervou s sy s t em t um or st ati s t ic s, 2021. C A Canc e r J Clin . 20 21;71(5 ):381 – 406. 3 3 . Tang L, Huan g Z, Me i H, Hu Y. Immu noth er a py in hema tolog ic ma ligna ncie s: a chi e v ement s , 4 chal lenge s a nd futu r e pro s p ec t s . S igna l T ran sduct Ta r g e t Ther . 202 3;8(1 ):1–39 . 5 4 . G u ha P , Hea the rton KR, O ’c onnell K P, Ale x ander IS , Kat z SC. A sse s sing the Fu ture of Sol id 6 Tumor Immunothe r a py. Biom e dicin e s. 2022;10(3) :65 5. 7 5 . Liu Y, Zhou F, Ali H , Lathia JD, Ch en P . I m munothe rapy fo r g liobla stoma : c urre nt s tat e, 8 chal lenge s, a nd futu r e pe r s pe c tives . C e ll Mol Immu nol. 202 4;21 (12):1354 –75. 9 6 . Akter F, S imon B, d e Bo er NL, R e djal N, Wak imoto H, Sh ah K. Pr e-cl inic a l tumor models o f 10 primary brain tumo rs: C ha lleng e s a nd O p p or tuni tie s. Bi ochim B iophy s ac t a R ev canc er. 11 2020;18 75(1):18845 8 . 12 7 . Majc B, N o vak M, Je rala NK, Jew e tt A, Bre z nik B. Immun o t he ra py of Gliobl a stoma : Curren t 13 Stra tegie s and Chall e nge s in Tum or Mod e l Devel o pment . C ell s . 2 021;10(2 ):26 5 . 14 8. W a ng Q , Yua n F, Z uo X, L i M. Break t h r ou g hs and c hall eng e s of o rganoid mod els f or a s s e s s i ng 15 canc er immuno t h e r a py: a cu tting -edg e t ool for adv anci ng per sonali s ed tr ea tmen ts . C ell D e a th 16 D i s c ov. 20 25;11(1 ):1–1 3 . 17 9 . W o ute rs R, Be ve r s S , Riv a M, D e Sm et F, Coo sema n s A. I mmunocomp ete nt Mou se M o del s in 18 the Se a r c h fo r Ef fec t i ve Immun ot h e r a py in G li obla s t o ma. Canc e rs. 2020;13(1 ):1 9 . 19 1 0. Pat er s on K, Za nivan S, Gl as s p o ol R, Cof f el t S B, Zag noni M. Mic r oflui dic technol o gie s f or 20 immunother apy studie s on s ol id tumour s . Lab Ch ip. 2021 ;21 (12) :2306 - 2 329. 21 1 1. W o lf N K, Kis s iov DU, Raule t DH . Role s o f natur al kille r c ell s in immunity to c anc e r, and 22 applic ation s to immunoth e r a py. Nat R ev Immunol. 2022;23(2 ):90 –105. 23 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 32 1 2. Siv or i S, Vac ca P , De l Z ot to G, Muna r i E, Ming ari MC , Mor et t a L. H uma n NK cel l s : sur face 1 r e cep tor s , inhibi tory chec kpoin t s, and tr ansl ation a l appli ca tion s. Cel l Mol Immun ol. 2 2019;16 (5) :430. 3 1 3. Prage r I, W a tzl C. Me ch ani sms o f na tural k iller c ell-m edi at ed ce llula r c ytotoxic ity. J L eukoc 4 Biol. 2 019;105(6 ):1 319 –29. 5 1 4. Ramírez -Labra da A, Pe sini C , Santi ag o L, H i dalgo S, Cal vo -P ére z A, O ñ at e C, e t a l. All About (NK 6 Cell -Medi a te d) D e a th in T wo Act s and an Unexpe cted En cor e: Ini tia t i on, Ex ecutio n and 7 Activa t i on o f Ad aptiv e Immuni ty. Fron t I mmunol. 2022;13 :896228. 8 1 5. O c hoa M C, Mi nu t e L, R odr i gu ez I, Gara s a S, P ere z - Ruiz E, Inogé s S , et al . Anti body - depend en t 9 cel l cy t otox icity : immunoth erapy s tr at egi es e nhanc ing e ff ec to r N K c ell s. Immu nol Cell Biol . 10 2017;95 (4) :347–55. 11 1 6. Alexa ndrov L B, N i k -Zain al S, We dge D C, Apa r ic io S AJR, B ehjati S , Bi ank in A V ., e t al. S ignatu re s 12 of muta tional p r o ce s s e s in human c anc er. N a ture . 2013;500 ( 7 463 ):415– 21 . 13 1 7. Fare s J, Davi s Z B, Rech be r g er J S, To ll SA, S chw art z J D, D aniel s DJ, e t al . Advanc e s i n NK c ell 14 thera py for b rain t um o rs. npj Pre cis O n c ol. 2 023;7(1 ):1–1 7 . 15 1 8. Mor i moto T, Naka z aw a T, Ma eoka R, Na kag awa I, Tsujimura T, Ma t s u da R . Na tur al Ki ller Cell -16 Bas ed Immuno ther apy aga in st G li o bla sto ma . Int J Mol S ci. 2023;24 (3):2 111 . 17 1 9. Br e z nik B, Ko M W , T se C, Chen PC , S enjo r E , Majc B, et al . Infil tr ating na t ural kill e r cells bi nd, 18 lyse an d incr ea s e c hemo the rapy e ffica cy in gli oblas t o ma ste m-like tumoro sph ere s . Commun 19 Biol. 2 022;5(1 ). 20 2 0. Poorva P , Ma st J, C ao B, S ha h M V ., P oll o k K E, She n J. Kill ing the k iller s : Na tur a l kil l er cel l 21 thera py targe tin g gli oma s t e m ce ll s in hig h-g r a d e gli oma. Mol The r. 2025 ;33 (6):24 62– 78. 22 2 1. da S ilva LHR, Cath arino LC C, d a Sil va VJ, E vange lis t a GCM, B arbu t o JAM. The W ar I s on: Th e 23 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 33 Immune S ys t e m aga in s t Gliob la stoma —How Can NK C ell s Drive Thi s Ba ttle ? Bi o medic ine s. 1 2022;10 (2) :400. 2 2 2. Zhong Q Y, Fan EX, Fe ng GY, Che n Q Y , Go u XX, Yue G J, et al. A g en e ex pre ssion -ba sed s tudy o n 3 immune cell s ub typ e s a nd gli oma progn o si s. B MC Canc er . 2019 ;19(1 ). 4 2 3. Cran e CA , Han SJ , Barry J J, Ahn BJ, L ani er L L, P arsa A T . TGF -b eta d ownr egula te s th e activ ating 5 r e cep tor NK G2D on NK cel ls a n d C D8+ T cell s i n gliom a pa tien t s. N euro O n col. 2 0 10;12(1) :7–6 13. 7 2 4. Frie se M A, Wi schhu sen J, Wic k W, W eile r M, Eise le G , Stei n le A, et a l . RNA in ter f e r e nce 8 targe ting tra ns fo r mi ng grow t h fac tor -be t a enha nce s NK G 2 D-me di ate d antig lioma immune 9 r e s pon se , i nhibit s glioma cel l migra tion a nd inv asivene s s, and a broga te s tumor i ge nic ity in 10 viv o. Canc er R es . 2004;6 4(20 ):7596 –6 03. 11 2 5. Shai m H , Shanl ey M, Ba sar R, D a he r M, G umin J, Za mler DB , et al . Targe ting the αv 12 integrin /TGF -β ax i s improve s na tural k ille r c ell func ti on ag ain st gliobl astoma s t em c ells . J Cli n 13 Inves t. 2021 ;131 (14) :e14 2116. 14 2 6. Zhang Y, Guo F , Wang Y . Hypoxi c tumor mic r o environmen t: D e st r o yer o f nat ural k iller c ell 15 functio n. Chi ne se J Canc er Re s . 2024;3 6( 2):138. 16 2 7. Sch indelin J , A r g anda - Carr era s I, Fri se E, Kayni g V, L ongair M, P i etzsch T, e t al. Fiji: an open -17 s ou rc e plat fo rm for bi olo gica l-im age anal y s i s. Na t Me thod s . 2012 ;9(7 ):676–82 . 18 2 8. Ar ora S , Si ngh S, Mit t a l A, De sai N, Khatr i DK, Gugul othu D , et al. Sp her oid s in c ance r r e se arc h: 19 Rece nt a dvan ce s and o ppo r tuni t i e s. J D r ug Del iv Sc i Te chnol. 2024;10 0:10603 3. 20 2 9. G a r c é s - L áza ro I, Ko tzur R, Cerwe n ka A, Ma ndelboim O . NK C el l s U nde r Hyp oxia: The Tw o 21 Fac es of Va sculariz atio n i n Tumor and P re gnanc y. F r on t I mmunol. 20 22;13 :924 775. 22 3 0. Vito A, El-S ayes N, Mo ssman K. Hypox ia - D riv en Immun e E scap e in the T umor 23 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 34 Microenv ironm ent . Cell s . 20 20;9(4 ):992. 1 3 1. Ar ne r E N , Ra thmell J C. Me t a b olic Prog ra mming and Immun e Suppre s s i on in t he T umor 2 Microenv ironm ent . Canc er C ell. 2023;41 ( 3):4 21. 3 3 2. Lobe l GP , Jiang Y, Si mon M C. Tumo r m icr oenvi r on men tal nu trien t s, cel lula r re spo ns e s, a nd 4 canc er. Cel l C hem Bi ol. 2023 ;30 (9) :1015– 32. 5 3 3. Bouche r i t N, Gorvel L, Oliv e D. 3D Tumor Mode l s a nd Thei r Us e fo r the Te s ting o f 6 Immunothe rapi e s. F r o n t Immunol . 20 20; 11:603 640. 7 3 4. H e W, He W , Kuang Y, Xin g X, S impson RJ , Huang H , e t al . Pr ote omic co mpari s on o f 3D and 2 D 8 glioma models r eve al s incr ea s ed H L A - E e xpres sion in 3 D m odel s i s a ssoc iat ed wi th r e s is tanc e 9 to N K cell -m edia ted c yto toxic ity. J P rot e ome Re s . 2014;13 ( 5 ):2272 –81 . 10 3 5. Br a un FK, Ro thhammer -Hampl T , Lor enz J, Pohl S , Men e vs e AN , V o llmann -Zw ere n z A, et al . 11 S c af fo l d- B as e d ( M at r ig e l TM ) 3D Cultu re T echniq ue o f Gliob la s toma R ecov ers a Pat ient -like 12 Immunosu ppre ssiv e Pheno typ e. Cell s . 20 23;12(14) . 13 3 6. Burste r T, Gär tne r F, Bulac h C, Zha napiy a A, Gihring A, Knipp schild U. Regula tio n o f MH C I 14 Molecu le s in Gli obla s t oma C e ll s and th e S ensi tiz in g o f NK Ce ll s. Ph a r m aceu tic al s. 15 2021;14 (3) :236. 16 3 7. Robilli ard LD , Yu J , Ancha n A, Finl ay G, A nge l CE, Gr aham ES. C ompr e h en sive A s se ssme nt o f 17 Sec r e te d Immuno -Mo dula tory Cytokin e s by Serum- Dif fe ren tiat ed a nd St em-li k e Gl iobla s toma 18 Cell s Rev e al s Di s tinc t Di ffe ren ce s be twee n Gliobla st oma Ph enotyp e s. In t J Mol S ci. 19 2022;23 (22) :14164. 20 3 8. Chang AL, Mi ska J , W a inwright DA, Dey M, Rivett a C V., Yu D , e t al . C CL 2 produce d by the 21 glioma mic r oe nvironmen t is es se nti al fo r the rec rui t me n t of r e gula tory t c ell s and myeloi d-22 derived s uppr e sso r c ell s. Can ce r Re s. 201 6 ;76(19) :5671 –82. 23 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 35 3 9. Taka cs G P, Kreige r C J , Luo D, T ian G, G ar cia JS, D ele yroll e L P, e t a l. G lioma -deriv e d CCL 2 and 1 CCL7 media te m igra tion o f imm une su pp re s s ive CC R2+ /CX3 CR1+ M-M DSC s in to t he tumor 2 microenv ironmen t in a redu ndan t ma nn er. Fr ont Immuno l . 2023; 13:99344 4. 3 4 0. Vakil ian A , Kho r ramd e lazad H , Heida ri P , Shei kh Reza ei Z, Has san shah i G. C CL2/ CC R2 sign aling 4 pathwa y in g liobla s toma mul tifo r me . Ne ur o chem I n t . 2017;10 3:1 –7. 5 4 1. Alfa ro C, Te ije ira A, Oña te C , Pe rez G , Sa nma med MF, And u ez a M P, e t al. Tumor - Produc ed 6 Inte rleukin - 8 Attrac ts Human My eloid - D erived Suppr e s s or C el l s and Eli cit s Extru si on of 7 N e utrophi l Extrac ellula r Tr a ps (NET s) . Cli n Canc er R e s . 2016;22 (1 5):3924 –36 . 8 4 2. Jack s o n C , Cher ry C, Bom S, Dyk ema AG , Wang R, Thomps on E, et al . D i s tinct my el oid-d e r iv ed 9 s upp r e s sor cell po pulati on s in human g liobla stoma . S cien ce . 2 025;387(67 31) . 10 4 3. W e be r R, Gro t h C, La s s er S, A rkhy pov I, P etr o va V, Al tev ogt P , e t al . IL -6 as a ma jor r e gulato r 11 of M D S C ac t i vity a nd po s s ibl e t arge t fo r ca ncer imm unoth erapy . Ce ll Immuno l. 20 21; 359. 12 4 4. Mor ri son BE, Park S J, M o on ey JM, M ehra d B. Chem ok ine -media t e d r ecruitm ent o f NK c ells i s 13 a c r itic al h o s t d ef en se me chan i s m i n inva s iv e a spe rgill o s i s. J Cli n Inve s t. 200 3;112( 12):186 2. 14 4 5. H e Q, Shi X, Zhou B, Teng J, Zhang C, Li u S, e t al . Int erle uk in 8 (CX CL8 )- CXC c hemo ki ne 15 r e cep tor 2 ( CXC R2 ) axis c on tr i bu t e s to Mi R-44 37- a ssoc iat ed rec r ui t me n t o f granul oc ytes and 16 natur al k iller c ell s in i sche mic s troke . Mol Immunol. 2 018;101:4 40–9 . 17 4 6. Vujanov ic L , Ball ard W , Thorne S H, Vujan ovi c NL, But te rfi el d LH. Ad enov iru s - eng in eer ed 18 human d endri tic c ell s induc e na tural kille r c ell chemotax i s via CXCL8/ IL -8 and CX C L10/IP -10. 19 O nc oimmunol ogy. 2012;1(4 ):448–57 . 20 4 7. Serran o- Pe rti erra E, Bl anco -Gel a z MA , Ol iva -N a ca rino P , Mar t ín ez- Cam blor P, Villa fani J , 21 Lópe z -La r rea C, e t al . Increa s ed n a tur a l ki l ler cel l che mot axis t o CXCL 12 in pa tien ts with 22 multiple s c lero s i s . J N euroimmunol . 2015;2 82:39–44 . 23 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 36 4 8. Ran G he, Lin Y qing, Tian L, Z hang T , Yan D m ei, Yu J hua, et a l . Na tur a l ki lle r ce ll h oming and 1 tr a ffic king in ti s s ue s a n d tumor s: fr om bio logy t o applic ati on. S ignal Tran sduc t Ta rge t Ther. 2 2022;7(1 ):1 –21. 3 4 9. Chan T YH , Wong JSY, Kiang KMY, Su n C W Y, Le ung GKK. T he dual ity of CXCR3 in gli obla stoma : 4 unvei ling a utocrin e a nd par ac r i ne mec h a nism s f o r no vel th erap e ut i c ap p r oa ch es . Cel l D ea th 5 D i s . 2 023;14(12 ):1 –13. 6 5 0. Avril T, Vaule on E, Hamlat A, Saik ali S , Et che verry A, D elma s C, et al . Human gliobl astoma 7 s t em-li k e c ells are mor e sen sitive t o allog e neic N K and T cell-m e dia ted k illing compa r e d with 8 s e rum-c ul ture d glio bla st o ma c e lls. Br a in Path ol . 2012;2 2 (2):159 –74. 9 5 1. H a s p e l s HN, Ra hma n MA, Jo s e ph JV , Nav arro A G, C heke nya M . G li obl a st o ma ste m - l ike c ell s 10 are mor e su sce ptibl e t h an di f fer enti ate d cells t o na t u ral kill er cell ly si s media ted throug h kill er 11 immunogl obulin-li ke r ec ep tors -human le ukoc yte an t i gen li gand mi s ma tc h and a ct iva t i on 12 r e cep tor -liga nd int erac t io n s . F r o nt Immunol. 2 018;9:134 5. 13 5 2. Clo se HJ, Ste ad L F, Nsengima n a J, R ei lly K A , D r oop A , Wurda k H, e t al . Expre s sion profiling o f 14 s in gle c ell s and p ati ent co ho r t s iden ti fie s mul t ip le immun o s uppr es s i ve p a t hw ay s and an 15 alter ed NK c ell phe notype in gliobl astom a. Clin E xp Immu nol. 202 0;200 (1):33 . 16 5 3. Karimi E , Yu MW, Mari tan S M, Peru s LJ M , Rezan eja d M, Sorin M, et al . S ingle - c ell s pa tia l 17 immune landsc ape s o f p r ima r y and me ta s t atic brai n tumour s . Na t. 2023;6 14 (7948):55 5–63 . 18 5 4. Fu W, Wang W, Li H , J i ao Y, Huo R, Yan Z, et al . Sing le - Cell Atla s Re ve al s Compl exit y of the 19 Immunosu ppre ssiv e Micr oenv ironm ent of Ini tia l and Re curr ent Gliobla s toma. F ron t I mmunol. 20 2020;11 :835. 21 5 5. W u J, Ga o FX, W a ng C , Qin M, H a n F, Xu T, et a l. IL -6 and IL -8 se cre ted by tumou r cel ls i mpai r 22 the f unction o f N K cell s v ia the STAT3 p a th way in oesoph ag eal squamo u s c ell c a rci no ma. J Ex p 23 Clin Ca nc er Re s . 20 19;38(1 ):1–1 5 . 24 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 37 5 6. H a n B, Mao FY, Z hao YL , L v YP, Teng YS, Duan M, et al . A l te r ed N Kp3 0, NKp46, NK G2D, a nd 1 D NA M-1 Ex pr e ssion on Circula ting NK C e lls I s A ssoci a ted wi t h Tumor Progr e ssion i n Human 2 G a s tr ic C ance r. J I mmunol Re s. 20 18;201 8:6 248590. 3 5 7. Mame s s ie r E , Prad el LC, T hibult M -L, D re ve t C, Zouin e A, Jac qu emier J, e t a l. P erip he r a l bloo d 4 N K cell s from br ea st ca nce r pa t i e nt s a re tumo r -ind uc ed c ompo sit e su bset s . J Imm unol . 5 2013;19 0(5) :2424–3 6. 6 5 8. Molf etta R, Zi t ti B , Lec ce M , Mili to N D , St abile H , Fiond a C, e t a l. CD155 : A Multi -F unctiona l 7 Molecu le in T umor Progr essi on. In t J Mo l Sc i. 2020 ;21(3) :922. 8 5 9. Zhan M , Zhang Z, Z hao X, Z hang Y, L iu T, L u L , et al. C D 1 55 in tumor p rogre s sion a nd t a r g e ted 9 thera py. Ca nc er L et t . 2022;54 5 :215830 . 10 6 0. Sloa n KE, E us t ac e BK, S te wart JK, Ze he tmeier C, T or e ll a C, Sime one M, et al . C D 1 5 5/ P V R play s 11 a ke y role i n ce ll motili ty during tumor c e ll inv asion and migra tion . BMC C ance r. 2 004;4:73. 12 6 1. Sloa n KE, S t e war t JK, Trel oa r A F , Mat the ws RT, J a y D G. CD155 /P VR en hanc e s glio ma ce ll 13 disp ersal by regu la ting a dhe sio n signali n g and focal adh esi on dy namic s. Ca nc er R es . 14 2005;65 (23) :10930– 7. 15 6 2. Enloe BM , J ay DG . Inhibi t i on o f Necl -5 (C D15 5/PV R) r educ e s gliob la s t oma di sper sal and 16 decrea se s MMP -2 ex pre ssi on an d ac t iv it y. J Neur oonc ol. 201 1 ;102(2) :225– 35 . 17 6 3. Yeo J, K o M, Le e DH , P ark Y, Jin HS. TI GIT /C D226 Ax is R egula te s Anti -Tumor Im mu nity. 18 Pharmac eutica l s (Ba s el ). 2021;14 (3 ):1 –2 0. 19 6 4. Lupo KB, Torregro sa -Al len S , Elzey B D, Ut tur k ar S, Lanma n N A, Cohen - Gadol AA, e t a l. TI GIT 20 contribu te s to the r egula t i on of 4 -1BB a n d doe s not de fin e N K cel l dys functi on in 21 gliobl astoma . iSc ienc e. 202 3;26 (12) :1083 53. 22 23 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 38 Fi g u r e s 1 2 Fi g u r e 1 : Grow th of G B spher oi ds in the Ce lvivo Clinostar syst e m . a) S chema t i c r e pr e s en ta tio n o f 3 the pr epa ra t i o n and cul t ure of GB s ph e roi ds e stabli sh ed in thi s s tudy . Crea te d w ith bioR end er. com. 4 b) NCH 4 21k and N IB140 spher oid s in Cl i noReac tor s o n da y s 0 an d 8. S cale ba r: 5 00 µm. c) G r o w t h 5 c urves o f N CH421k and NIB140 sph eroi ds. Grow th wa s det ermin ed by mea s uri ng sphe roid a rea i n 6 t i m e . M e a n ± S E M o f a t l e a s t t h r e e b i o l o g i c a l r e p l i c a t e s i s p r e s e n t e d . d) A n aly si s o f c ell via bili ty in 7 NCH421k and N I B140 s p h eroid s o n da y 8. In bo th sphe roid typ es , th e prop or ti o n of c ell s in early o r 8 l ate sta ge s of a popt osi s i s bel ow 10%. 9 10 11 12 13 14 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 39 1 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 40 Fi g u r e 2 : Cha rac t e risation of GB sp heroids cul t ured in the Cel vivo Clinos tar system . a) 1 I mmunofluo re scen t s t a ining o f sele ct e d mark ers in NCH421k and NI B140 s pheroid s on da y 8. 2 NCH421k s phe r oid s ex pre ss high l evel s o f SOX and OL I G 2 ( st em ce ll ma rker s) and low leve l s o f GFA P 3 (di ffe ren tia t i on marker ) and C D 4 4 (me se nchy mal subtype ma r k er) . Sphe roid s ar e hi ghly p r ol ifera tive , 4 a s indi ca ted by a hi gh proport io n o f Ki- 67 - p o s itiv e cell s. Cell s in the NI B140 s p h er o id s ex pre ss high 5 l evels o f GFA P a nd C D44, no O L IG2 an d va r y ing level s o f S O X2. The p ropo rt i on of Ki-67 -po si tive 6 proli fe r a t i ve c ell s i s lowe r c ompar ed t o N CH421k sphe roid s. b) Immu no fluo re s c ent s ta i ning o f 7 sel e cte d marke rs in NC H421k a nd NIB14 0 spher oid s on day 18. Hete rogen eou s z one s were ob se rved 8 i n b ot h type s of s phe r oi d s , such a s reg i ons of low a nd high ce ll proli fe rati on, n ec r otic a nd hypox ic 9 reg ion s . 10 11 12 13 14 15 16 17 18 19 20 21 22 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 41 1 Fi g u r e 3 : Spheroid culture affects the e x pr ession of lig ands for NK cell rec ept or s on GB cel ls and 2 their s ensitivity t o NK ce lls . a) T he ex pre ssion of s elec t e d NK c ell liga nd s wa s a naly sed by flow 3 c y t o m e t r y . F o r e a c h o f t h e t w o c e l l l i n e s , t h e e x p r e s s i o n w a s a l s o a n a l y s e d o n c e l l s i n s t a n d a r d 4 cu l t u r e. D at a ar e p r es en t ed a s m e a n d=FO± d=FO S E M o f t hr ee b i o lo gi c a l re p l i ca tes . Two - w a y AN O V A w i t h 5 u n c o r r e c t e d F i s h e r ’ s l e a s t s i g n i f i c a n t d i f f e r e n c e w a s u s e d t o c a l c u l a t e s t a t i s t i c a l s i g n i f i c a n c e o f t h e 6 di ffe re nce s in ex pre ssio n l evel s o f t he l ig ands be twe en ce ll li ne s a nd cul t ure m eth ods. The symbols i n 7 pa r e n the s e s de signa t e l igand s tha t bind ac tivating ( + ), inhibit ory (- ) o r e ith er a cti va t i ng or inhibi tory 8 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 42 N K r e cep t ors ( +/ -) . b) % c yto toxic ity plo ts o f c alc e in re lea s e a s say o f N K -92 c ells aga in s t di s s oc iat e d 1 NCH421k and N I B140 s p h eroid s a nd NC H421k and NIB140 c ell s from the s t a nda rd cultur e. Dat a ar e 2 pre sen ted a s me an ±d=FOS EM o f 3 –4 biologic al re plic at es. 2-wa y A N O VA wi th Š ídák ’ s 3 mu ltiple c ompari so ns te st wa s u sed for s tati stic al a naly si s. 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 43 1 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 44 Fi g u r e 4 : N K cell infil tration in NC H 42 1 k and NIB140 s phe r oid s . a) Sc hematic re pr e s enta tion of the 1 di r e ct G B spher oid- NK cell c o-cul tur e se t -up . Cre a te d w ith bioR ende r.c om. b) N K - 9 2 ce l l in f i lt r at i on 2 i nt o NCH421k and NI B140 s p he roid s a s d ete rmine d by flow cy t om e t ry a fte r t h e 24 h c o-cultur e. Dat a 3 a r e pre s ent ed as mea ns ±d=FOS E M of 6 biologi cal replic at e s. Ord inary one -wa y AN O VA w as u sed f o r 4 st ati stic al an a lysi s. c) H D N K c ell in filt ra tion in to N CH421 k and N I B140 sphe roi d s as de termin ed by 5 f l o w c y t o m e t r y a f t e r t h e 2 4 h c o - c u l t u r e . D a t a a r e p r e s e n t e d a s m e a n s ± d=FO S E M o f 3 b i o l o g i c a l 6 repl icat e s. E ach dat a poi n t a nd i ts sh a pe corr e s pon d to H D N K ce ll s fr om a n in dividua l don or. 7 Ordina ry one -w ay AN OVA wa s u s ed for st ati s t ic al ana ly si s . d) Immuno flu ore sce nt stai ning of C D45-8 p os it i ve N K- 9 2 c e l l s o n FF P E s l i des of G B s p he r o i d- N K- 9 2 c o- cu l t ur es . e) Im muno fluor e s c ent stai ning 9 of C D45- po sitive HD N K ce ll s on FFPE slide s o f GB sph eroi d-HD N K ce ll c o-cu lture s . f) S c h e m a t i c 10 repr es ent ation o f th e orga n -on -chip p la t for m m imick ing the bl ood fl ow a nd influ x of N K-92 cell s into 11 the tumour . Cr eat ed wit h bio Rend e r .c om. g) In filtr ation o f NK-92 c ell s in to c h amber s co nta inin g 12 NCH421k /NIB140 sph eroid s. A f ter 24 h of N K-92 c ircula t i on , th e perce n ta ge of NK -92 wa s 13 de t e r mi ned by fl ow c ytome try. Da ta ar e s how n as me a n ±d=FOSE M a nd da ta poin ts r epre s ent each o f 14 the 5 –8 bi ol ogic al re plic at es. U npair ed t- te s t wa s u s e d f or s t ati s tical a n alysi s . h) V i a b i l i t y o f N K - 9 2 15 c ells in fil tra ted in to the ch amb ers co nta i ning NCH421k /N I B140 s ph e r oi d s . Da ta are shown as mea n 16 ± d=FOSEM a nd dat a po int s r epr e sen t eac h o f the 6 – 8 biolo gica l re plic at es. 17 18 19 20 21 22 23 24 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 45 1 Fi g u r e 5 : N K ce ll cy t otoxic ity ag ains t NCH421k an d N IB140 s pheroids . a) F l o w c y t o m e t r i c a n a l y s i s o f 2 NK-92 c yto toxic ity ag ain s t N CH421k a nd N I B140 sp hero id s . Da ta a r e p r e s ent ed a s me a n ±d=FO SEM o f 6 3 bi ologic al replic a te s . Ordin ary one -w ay ANOVA wa s u s e d fo r sta tis tica l an aly s i s . b) F low c ytometric 4 a nalysi s o f HD NK ce lls c ytot oxic ity aga ins t N CH 42 1k and NIB140 s p he roid s . Da ta are pr e s e n t e d a s 5 me ans ±d=FOSE M o f 3 biolo gica l r eplic a te s . Eac h data po i nt and i t s s h ape c o r re sp on d to HD NK cel l s from 6 a n indiv idual donor . O rdi na r y one -w ay A NOVA w a s used f or s ta ti st i cal a naly s i s . 7 8 9 10 11 12 13 14 15 16 17 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 46 1 Fi g u r e 6 : Soluble fac t ors i n GB s pheroi d cul t ures a nd their 24 h co-c ult ures w ith NK cell s . a) T h e 2 e ffe ct o f NCH 4 21k - and N I B14 0-condi tio n ed mediu m on cy t o t ox ici t y of NK-92 ce l ls again st K562 cell s. 3 D a t a a r e p r e s e n t e d a s m e a n s ± d=FO S E M o f 3 – 4 b i o l o g i c a l r e p l i c a t e s . P a i r e d t - t e s t w a s u s e d f o r s t a t i s t i c a l 4 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 47 a nalysi s . N – nonc onditio ned medi um, C – c ondit ioned me di um. b) Ex pressi on of N Kp3 0 o n NK -92 1 c ells c ultu r e d in N CH 4 21k- o r NI B140-c o nditione d me dium. D ata a re pre sen ted a s me an ±d=FOS E M of 3–2 4 biolo gica l repli ca te s. P a ir e d t -t e s t w a s u s ed for sta ti st ic al analy s i s o f dif fe re nce s in ex pre ssion 3 be t w ee n noncond i t i on ed a nd c onditi o ned media . N – n on conditi one d m edi u m, C – c onditi oned 4 me dium. c) Ana ly sis o f c ytokin e s ec r e t i o n in N C H 42 1k and NIB140 sphe roid s. log 2(x +1) -tra n s fo rmed 5 c ytoki ne conc entra t i on s ( av eraged from t w o t e chnic al replic a te s) a re pre s ent ed in the heatmap . d) 6 Ana lysi s o f secr et ed cytok ine s in t he 2 4 h sphe roid -HD N K cel l c o-cul tur e s. Only c ytoki ne s with 7 c once ntratio n ≥ 50 pg/mL in a t le a s t o ne o f the condi t i on s wer e inc lud ed. lo g2(x +1)- tr a n sfo rmed 8 c ytoki ne c oncen tra tion s ( av eraged fr o m two technic al r e plic a te s) ar e pre s en ted i n the h eatm ap. 9 Cy tokine s w ere hi e r a r c hic ally clus te red ba s ed on th e “one minus P e ar s on co rrela tion” me t ric in 10 Morpheu s . e) Ana ly si s o f sec re ted c yt okines in t he 24 h s pher oid- NK-92 c e ll c o-cul t ure s . Only 11 c ytoki nes w it h conc ent rati on ≥ 50 pg / m L in at lea st o ne o f th e c onditio ns we re i nc luded. lo g2(x+1 ) -12 tran s forme d cytok ine conc en tra tion s (a v eraged from two techn ic al r epl icat e s) are pr e s ente d in th e 13 he atmap . Cytokine s we re h i era rchic ally cl us t er ed ba se d on the “ on e m inu s P ear son c o r rel ation ” 14 m et r i c i n M o r p he us . f) CX CL9 , CX CL11, CXC L10, IL -8 and CCL8 w e re pr om ine ntl y upregu la ted i n co -15 c ulture s of NK-92 c e ll s w ith N IB1 40 s ph e r oi d s . Da ta are pre s ent ed a s me an ±d=FOS EM of t w o t echnic a l 16 repl icat e s. O r d ina ry one - w ay ANO VA wa s u sed for s t ati st ica l anal y si s . 17 18 19 20 21 22 23 24 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 48 1 Fi g u r e 7: Exp r es s i on of NK ce ll rece pt ors on IL -2 -activ at e d NK cel ls from d i f ferent sou rces . a) 2 Co mpari s o n of rec ep tor ex pre s s io n on I L-2-ac tivated NK-92 c ell s (N=3), H D NK c ell s (N =6–9 ) and N K 3 c ells from G B p ati ent s (N= 6). Da ta are pre sen ted a s mean ±d=FOSE M and dot s r e pres ent biologic a l 4 repl icat e s. Or d ina ry one - w ay A N O VA wa s u sed fo r s ta ti stica l anal y s i s . b) E s tima ti on plot s show ing t he 5 e xpre ss i on of N K cell r e cep tor s on H D N K c ells, a f fec t ed by IL-2 -ac tiva tion (N=6 ). 6 7 8 9 10 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint 49 1 Fi g u r e 8: The C D155 axi s regula t es th e c ytotoxici t y of HD NK c ells ag ainst NC H421k cells . a) 2 Ac cording to th e T CG A da ta, PVR i s up regula te d i n GB compa red to lower -gra de gli omas . T ukey' s 3 Hone st Signi f i c ant Di f f e re nce were c alc ul ated in G li oVi s. b) H ig her e xpres s i on of PV R i s a s so c i a t e d 4 w ith shor ter s u rviv al of glio ma pa tien t s. K aplan- Mei e r e s timat or s u rviv al ana lys i s w a s p er forme d in 5 Glio Vi s. c) Re s u l ts of c alce in r ele as e a ssa y pe r fo rmed w ith NK cel l s from 3 d i ffe re nt he al thy don o rs in 6 w hich CD155 , T IG I T, and D NA M-1 o r th eir combi natio ns w er e block ed by antib odie s and NCH421k 7 c e l l s w e r e u s e d a s t a r g e t c e l l s . D a t a a r e s h o w n a s m e a n ± d=FO S E M o f t h r e e t e c h n i c a l r e p l i c a t e s . 2 - w a y 8 ANO V A with Tuk ey’ s multipl e compa r i s o ns t e st wa s u sed f or s ta ti stica l anal ysi s. 9 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted October 30, 2025. ; https://doi.org/10.1101/2025.10.29.685263doi: bioRxiv preprint