Ac a Bioma e ialia 170 (2023) 202–214
Con en s lis s a ailable a ScienceDi ec
Ac a Bioma e ialia
jou nal homepage: www.else ie .com/loca e/ac bio
Full leng h a icle
T apping me as a ic cance cells wi h mechanical a che a ays
Da id Caballe o
a , b , ∗, Rui L. Reis
a , b
, Subhas C. Kundu
a , b
a
3B’s Resea ch G oup, I3Bs – Resea ch Ins i u e on Bioma e ials, Biodeg adables and Biomime ics, Uni e si y o Minho, Headqua e s o he Eu opean
Ins i u e o Excellence on Tissue Enginee ing and Regene a i e Medicine, A ePa k - Pa que da Ciência e Tecnologia, 4805-017 Ba co, Guima ães, Po ugal
b
ICVS/3B’s-PT Go e nmen Associa e Labo a o y, B aga, Guima ães, Po ugal
a i c l e i n o
A icle his o y:
Recei ed 16 Ma ch 2023
Re ised 26 July 2023
Accep ed 17 Augus 2023
A ailable online 23 Augus 2023
Keywo ds:
Cance
T aps
Di ec ed mig a ion
Ra che
Mic o ab ica ion
a b s a c
Cu en ea men s o cance , such as chemo he apy, adio he apy, immuno he apy, and su ge y, ha e
posi i e esul s bu a e gene ally ine ec i e agains me as a ic umo s. T ea men e ec i eness can be
imp o ed by employing bioenginee ed cance aps, ypically u ilizing chemoa ac an -loaded ma e ials,
o a ac infil a ing cance cells p e en ing hei uncon olled sp ead and po en ially enabling e adi-
ca ion. Howe e , he encapsula ed chemical compounds can ha e ad e se e ec s on o he cells causing
unwan ed esponses, and he gene a ed g adien s can e ol e unp edic ably. He e, we epo he de elop-
men o a cance ap based on mechanical a che s uc u es o cap u e me as a ic cells. The aps use
an a ay o asymme ic local ea u es o mechanically a ac cance cells and di ec hei mig a ion o
p olonged pe iods. The apping efficiency was ound o be g ea e han iso opic o in e se aniso opic
a che s uc u es on ei he dissemina ing cance cells and umo sphe oids. Impo an ly, he aps ex-
hibi ed a educed e ec i eness when a ge ing non-me as a ic and non- umo igenic cells, unde sco ing
hei pa icula sui abili y o cap u ing highly in asi e cance cells. O e all, his o iginal app oach may
ha e he apeu ic implica ions o figh ing cance , and may also be used o con ol cell mo ili y o o he
biological p ocesses.
S a emen o significance
Cu en cance ea men s ha e limi a ions in ea ing me as a ic umo s, whe e cance cells can in-
ade dis an o gans. Bioma e ials loaded wi h chemoa ac an s can be implan ed o a ac and cap-
u e me as a ic cells p e en ing uncon olled sp ead. Howe e , encapsula ed chemical compounds can
ha e ad e se e ec s on o he cells, and g adien s can e ol e unp edic ably. This pape p esen s an o ig-
inal concep o “cance aps” based on using mechanical a che -based s uc u es o cap u e me as a ic
cance cells, wi h g ea e apping efficiency and s abili y han p e iously s udied me hods. This inno-
a i e app oach has significan po en ial clinical implica ions o figh ing cance , pa icula ly in ea ing
me as a ic umo s. Addi ionally, i could be applied o con ol cell mo ili y o o he biological p ocesses,
opening new possibili ies o biomedicine and issue enginee ing.
©2023 The Au ho (s). Published by Else ie L d on behal o Ac a Ma e ialia Inc.
This is an open access a icle unde he CC BY license ( h p://c ea i ecommons.o g/licenses/by/4.0/ )
1. In oduc ion
The ad en o mode n nano echnologies and issue enginee -
ing s a egies has enabled he de elopmen o inno a i e ools
o figh cance [1–8] . Despi e he demons a ed benefi s o cu -
en he apeu ic app oaches, such as su ge y, chemo he apy, adio-
he apy, and immuno he apy, hey o en p esen limi a ions (e.g.,
accessibili y o umo s, side e ec s, o lack o e ec i i y –can-
∗Co esponding au ho .
E-mail add ess: [email p o ec ed] (D. Caballe o) .
ce cells migh emain a e ea men ), pa icula ly in ea ing
me as a ic umo s. These issues ha e spu ed in e es in pu su-
ing inno a i e issue-enginee ed app oaches o a ge highly infil-
a ing cance cells. Consequen ly, cance aps ha e eme ged as
a issue-enginee ed app oach o selec i ely cap u ing me as a ic
cells when implan ed wi hin he body and p e en ing hei uncon-
olled sp eading, as well as allowing hei analysis and/o po en-
ial e adica ion [9] . These aps a e ypically made o biocompa -
ible bioma e ials loaded wi h chemoa ac an s o a ac agg es-
si e cance cells [ 10 , 11 ]. A my iad o chemokines and g ow h ac-
o s ha e been employed o igge he mig a ion o cance cells
owa ds he ap [12–14] . Howe e , he li e ime o his ype o ap
h ps://doi.o g/10.1016/j.ac bio.2023.08.034
1742-7061/© 2023 The Au ho (s). Published by Else ie L d on behal o Ac a Ma e ialia Inc. This is an open access a icle unde he CC BY license
(
h p://c ea i ecommons.o g/licenses/by/4.0/ )
D. Caballe o, R.L. Reis and S.C. Kundu Ac a Bioma e ialia 170 (2023) 202–214
is limi ed due o he una oidable deg ada ion o he chemokine
g adien s, and chemokines hemsel es may p oduce undesi ed e-
sponses in o he cell ypes. To o e come hese issues, chemical-
ee aps ha e been p oposed ha ely on opog aphical and mo -
phological ea u es (e.g., fib illa y mic o/nano-s uc u es, acks, o
g oo es) o guide and cap u e cance cells [ 11 , 15 ]. These aps a e
mo e s able bu can s ill ace challenges in e ms o cap u ing cells
efficien ly. Fo ins ance, mic o-fil e memb anes wi h size-sensi i e
po es ha e been employed o selec i ely ap cance cells wi hin
mic ofluidic sys ems [16–19] . Despi e hei po en ial o be im-
plan ed in blood essels o nex o he umo a e su ge y o se-
lec i ely cap u e ci cula ing o highly infil a ing umo cells, hei
ansla ion o clinical se ings has been impeded by he possibili y
o clogging, which could lead o se e e consequences. Physical en-
apmen ep esen s ano he mechanical-based s a egy aimed a
cap u ing me as a ic cells. This me hod in ol es igge ing a con-
o ma ional change o he apping ma e ial, ypically a hyd ogel,
in esponse o en i onmen al cues, such as he acidic condi ions
p esen in he umo mic oen i onmen [20] . Despi e i s p omis-
ing capabili ies, a significan limi a ion o physical en apmen is
i s inabili y o di e en ia e be ween cance ous and heal hy cells.
To add ess his challenge, i has been p oposed modi ying he hy-
d ogel wi h a ecogni ion molecule ha specifically a ge s ecep-
o s o e exp essed in cance cells [ 21 , 22 ]. Al hough his app oach
has demons a ed success in apping specific cance cell ypes, he
a iabili y in ecep o exp ession ac oss di e en ypes o cance
may es ic i s applicabili y o all cance cells.
Recen ly, mechanical aps based on he mig a ion concep
o a che axis ha e been p oposed [ 23 , 24 ], which use a ays o
aniso opic ea u es o mechanically pola ize and ein o ce he o -
ganiza ion o he inne cell cy oskele on, gene a ing a di ec ional
memo y [25–29] . Owing o his simplis ic mechanism o ac ion,
mechanical a che -based aps o e unique ad an ages o selec-
i ely cap u ing cance cells [30] , bu hei abili y o do so has no
ye been ully demons a ed in p ac ice. In his wo k, we p esen
he ab ica ion o mechanical a che -based annula aps ha use
asymme ic ea u es o p omo e he cap u e and di ec ed mig a-
ion o me as a ic cance cells. Ou analysis e eals he wo king
mechanism o he aps o biasing cell mo ion and how cell dy-
namics is impac ed by he ap a chi ec u e. Nex , we show ha
he efficacy o he aps can be modula ed by p ope ly scaling he
ap. We also showcase he diminished efficacy o he aps when
aimed a non-me as a ic and non- umo igenic cells, highligh ing
hei pa icula sui abili y o cap u ing in asi e cance cells. Fi-
nally, we demons a e ha he aps can e ec i ely im pede he
in asion o dissemina ing mul i-cellula umo sphe oids. O e all,
hese esul s sugges ha mechanical a che -based cance aps
ha e he po en ial o become a aluable adjunc o cu en cance
ea men s. Mo eo e , hese aps could find applica ions in o he
biological con ex s whe e he selec i e ec ui men o specific cell
ypes, such as immune cells, s em cells, and o he s, is desi able. Fo
example, he aps could be used o selec i ely isola e and collec
hese cell ypes om a mul i-cellula issue, o o enhance o gan
egene a ion p ocesses.
2. Ma e ials & me hods
P epa a ion o he a che -based aps : The aniso opic a che -
like mic os uc u es we e designed using CAD so wa e (Au oCAD,
Au odesk). An O- ing (i.e. annula ) mo phology was designed o
concen a e he cap u ed cells in he cen e o he ap, wi h
he iangula ea u es poin ing inwa ds. Fo compa ison (con-
ol), iangles poin ing ou wa ds and iso opic g oo es we e also
designed. Two di e en iangle sizes we e also conside ed o
scaling a gumen s (Table S1). UV-pho oli hog aphy (Mask aligne
MDA-400M, SPS-Eu ope) was used o ab ica e a 10 μm hick
SU8 mold (SU8 2010, Mic ochemicals GmbH) on o a Si wa e and
di ec ly eplica ed wi h PDMS (Sylga d 184, Dow Co ning, 10:1
w/w p e-polyme /c oss-linke ; 70 °C o 1 h) ( Fig. 1A ). The PDMS
eplica was ende ed hyd ophilic by O
2
plasma (Diene ) o 30 s a
200 W, silanized wi h ichlo o (1H,1H,2H,2H pe fluo ooc yl) silane
(Sigma-Ald ich) by apo phase o 1 h a oom empe a u e, and
cu ed o 1 h a 70 °C. Nex , a small olume o PDMS (50 μL)
was spin-coa ed a 20 0 0 pm o 30 s on op o a glass co e slip
( hickness #1, Menzel-Gläse ) o c ea e a hin polyme ic laye . The
silanized PDMS eplica was ca e ully loca ed on op and hen ac-
uumed o 1 h o emo e all bubbles, ollowed by cu ing o 1 h
a 70
ºC. Then, he PDMS eplica was ca e ully eleased o ob ain
a hin PDMS laye wi h he mic o- a che ap s uc u es ( Fig. 1A ).
Las ly, he mic o ab ica ed a che s we e ende ed hyd ophilic by
O
2 plasma ea men (30 s, 200 W) and unc ionalized wi h 25
μg mL
−1 fib onec in om bo ine plasma (Sigma-Ald ich) in PBS
o p omo e cell adhesion. Bo h nega i e and posi i e PDMS eplicas
we e cha ac e ized by whi e ligh in e e ome y o confi m he ac-
cu a e eplica ion o he ea u es (see supplemen a y me hods and
Fig. S1).
Cell cul u e: Human MDA-MB-231 (me as a ic) and MCF-7 (non-
me as a ic) b eas adenoca cinoma cells (ATCC), along wi h L929
mu ine fib oblas s, we e cul u ed using high glucose DMEM media
(Sigma-Ald ich). Human non- umo igenic MCF10A b eas epi he-
lial cells (ATCC) we e cul u ed in MEGM Bulle ki media (Lonza).
All media was supplemen ed wi h 10% e al bo ine se um (FBS;
Gibco), NaHCO
3 (3.7 g L
−1
), and 1% penicillin/s ep omycin (Al a-
gene). All cell ypes we e cul u ed a low passages unde physio-
logical condi ions (37
ºC and 5% CO
2
). Cells we e ypsinized using
T ypLE Exp ess Enzyme (1X) (Gibco).
P epa a ion o umo sphe oids : Ul a-low adhesion U-bo omed
96 well-pla es (Co ning) we e used o o m 3D mic o-sized
sphe oids by spon aneous agg ega ion. Fo his, 2500 cells we e
deposi ed in each well in 100 μl o cul u e media, and he
sphe oids we e g own o 4 days esul ing in sphe oids wi h an
a e age diame e o 545 ±51 μm ( n = 36). Fo he expe imen s,
he sphe oids we e ca e ully emo ed wi h a mic opipe e.
Cell s aining: Cells we e fixed wi h 10x o malin (Sigma-Ald ich)
o 20 min a oom empe a u e and washed wi h PBS o 5
min (x3). Cells we e hen pe meabilized wi h 0.5% T i on (Sigma-
Ald ich) o 3 min. Nex , he cells we e washed wice o 5 min
wi h PBS. Fo s aining, phalloidin– e ame hyl hodamine B iso h-
iocyana e (1:250; 1 h) (Sigma-Ald ich) and DAPI (4.6-diamidino-2-
phenylidole) (1:10 0 0; 20 min) (Sigma-Ald ich) we e used o imag-
ing ac in filamen s and nuclei, espec i ely. The s ained samples
we e washed in PBS o 5 min (x3). To e ec i ely inhibi cell p oli -
e a ion in li e cell mig a ion s udies ( umo sphe oid expe imen s),
cells we e ea ed wi h high DAPI concen a ions (10 μg/mL; 30
min) dilu ed in cell cul u e media, ollowed by insing wi h esh
media p io o s a ing he expe imen . This app oach ha nesses
DAPI’s abili y o specifically bind o DNA, p o iding s able fluo es-
cence and low oxici y [ 31 , 32 ]. Finally, cell-pe mean Calcein AM
(10 μM; The mo Fishe Scien ific) was employed o li e mul i-
cellula (cance s. non-cance cells) expe imen s.
Op ical mic oscopy: Time-lapse expe imen s we e pe o med in
a Zeiss Axio Ve in e ed op ical mic oscope using 10 ×0.45 N.A.
ai objec i e a a 1 image/15 min o acquisi ion a e o a minimum
o 60 h. The mic oscope was equipped wi h a cha ge-coupled de-
ice came a (Zeiss), an HXP 120 C fluo escence ligh uni (Zeiss)
o epifluo escence expe imen s, and an in eg a ed shu e o p e-
en pho o oxici y. Fo expe imen s, aps we e loca ed in a Pe i
P35 dish and mechanically fixed wi h a home-made PDMS O- ing
o he same ex e nal diame e as he dish. 2 mL o cell cul u e me-
dia was added and a hin laye o mine al oil (Sigma-Ald ich) was
ca e ully deposi ed on op co e ing he medium o minimize me-
dia e apo a ion.
203
D. Caballe o, R.L. Reis and S.C. Kundu Ac a Bioma e ialia 170 (2023) 202–214
Fig. 1. Mic opa e ned cance a che aps. (A) Scheme desc ibing he s ep-by-s ep ab ica ion o he cance ap. (B) A schema ic ep esen a ion o he design and wo king
mechanism o he cance ap based on in e connec ed aniso opic a che -like opog aphic s uc u es. The pa ame e s b and h co espond o he heigh and wid h o he
iangula a che uni , espec i ely. MIN and MAX deno e he lowes and highes z posi ions, espec i ely. (C-D) B igh field images and magnified iews showing he di e en
designs o he O- ing cance aps (IN, OUT, and NULL – con ol). (E) S aining illus a ing he ap wo king mechanism desc ibed in (A) o cells mig a ing di ec ionally om
one
apping mo i o he neighbo ing one (Phalloidin –F-ac in: ed; DAPI – nucleus: blue). Scale ba s: 100 μm. The bo om schemes illus a e he p ofile o he po en ial
well in which he cells a e apped and unde going di ec ed mig a ion. Scale ba s: (B) 1.2 mm; (C) D) 500
μm; (D) 100 μm.
T ap assay: Two di e en assays we e pe o med: seeding o in-
di idual cells iso opically dis ibu ed ac oss he aps, and seeding
o mul i-cellula umo sphe oids in he inne pa o he ap. A -
e seeding, cells and sphe oids we e allowed o adhe e o 1–2 h
be o e he s a o he expe imen .
Biophysical analysis : To analyze he efficacy o he ap in p o-
mo ing he di ec ed mig a ion and apping o cance cells, he ol-
lowing pa ame e s we e defined:
1. La ice uni (l.u.): Leng h o he pe iodic iangula uni wi hin
he a che pa e n. I se es as a undamen al measu emen
o disc e izing he sys em and analyzing cell mig a ion wi hin
he a che s uc u e. The la ice uni defines he spa ial scale
o e which cellula mo emen and di ec ional bias a e quan i-
fied and e alua ed.
2. Mean bias pe la ice uni ( p ): Ne di ec ional bias o cells mi-
g a ing wi hin he a che pa e n, indica ing hei consis en
mo emen in a specific di ec ion. I is quan ified by coun ing
he numbe o s eps (in l.u.) ha he cells mig a e owa d he
poin ed (n + ) o wide (n-) edge di ec ion [33] . This alue is a -
e aged ou o all cells and condi ions.
3. Pe sis ence leng h/ ime (L
pe
/T
pe
): Leng h and ime du ing
which a cell mig a es s aigh wi hou e e sing i s mo ion, e-
spec i ely. To measu e hese biophysical pa ame e s, cell a-
jec o ies we e acked using he Poin ing Cell T acking Plug-in
(Fiji, NIH) a 15 min in e als. The gene a ed x-y poin s we e
connec ed o gene a e a se o mig a ion pa hs used o calcu-
la e L
pe and T
pe
.
4. Mean eloci y: A e age speed o cells mig a ing along he ap-
ping ea u es in bo h di ec ions.
5. Confinemen ac o ( ): Ra io o he numbe o cells loca ed
in he inne egion o he o al numbe o cells loca ed in all
egions.
6. % inc ease: Cell en ichmen wi hin he aps compa ing he be-
ginning and end o he expe imen .
7. Size a io (S
): Ra io o he cell size ela i e o he ea u e.
Image p ocessing – nuclei densi y hea map: An a e age nuclei
hea map was gene a ed using a Py hon code de eloped in-house
(Py hon 3.11.4) wi hin Jupy e Lab ( 3.5.3). The code u ilized se -
e al lib a ies, including NumPy, Ma plo lib, and sciki -image. The
loaded images unde wen p ep ocessing s eps o imp o e hei
quali y, including alignmen . The a e age in ensi y ac oss all im-
ages was calcula ed using he np.mean() unc ion om he NumPy
lib a y. The esul ing a e age in ensi y da a was isualized as a
hea map using he Ma plo lib lib a y. The pl .imshow() unc ion
was employed o plo he hea map, using he ’ho ’ colo map o
in ensi y mapping. The in ensi y ange o he hea map was de e -
mined using he np.pe cen ile() unc ion, which p o ided he lowe
and uppe pe cen ile alues. The comple e code implemen a ion
can be ound in he Supplemen a y Me hods.
S a is ical analysis : Da a a e p o ided as he Mean ±S.E.M. S a-
is ical analysis was pe o med using a one-way ANOVA wi h a
Dunn
´
s pos -hoc es , and significance was accep ed a P < 0.05
(O iginP o). The da a se indica ing he numbe o expe imen al
epe i ions and cells analyzed is desc ibed in he Fig. cap ions.
204
D. Caballe o, R.L. Reis and S.C. Kundu Ac a Bioma e ialia 170 (2023) 202–214
3. Resul s
3.1. De elopmen and wo king mechanism o he cance aps
We designed an assay o ap me as a ic cance cells by me-
chanically di ec ing hei mo ion owa ds a specific cap u ing loca-
ion. The aps we e based on an a ay o mic o ab ica ed a che -
like aniso opic s uc u es wi h a heigh (h) la ge han i s base,
and in e connec ed h ough a na ow b idge ( Fig. 1B ). Wi h his
configu a ion, he s uc u es physically pola ized he cells and ig-
ge ed hei di ec ed mig a ion owa ds he inne pa o he ap
(cap u e a ea). The apping e ec was enhanced by he h ee-
dimensional (3D) a chi ec u e o he ea u es, which ac ed as po-
en ial wells o dep h z
max - z
min
confining he cells wi hin he
a che -like a ay channels. These we e pa e ned ollowing an O-
ing, o annula , dis ibu ion o concen a e he cells in he inne
pa o he ap (cap u e a ea). To maximize he numbe o cells
apped wi hin he s uc u es and he a che ing e ec , and min-
imize hei mig a ion ou side he ea u es, we ab ica ed a che
uni s uc u es ha we e la ge han he a e age cell size. Fo his,
we measu ed he a e age a ea o he cells seeded on he ap’s fla
su ace ob aining a alue o < A
cell
> = 1072 ±47 μm
2 (Fig. S2).
To demons a e he a che ing e ec , we designed h ee di e en
configu a ions, namely ap IN, OUT, and NULL ( Fig. 1C ). In b ie ,
ap IN and OUT included he a che a ays poin ing inside and
ou side he O- ing, espec i ely, whe eas in he NULL – con ol –
condi ion he iso opic eng a ed linea g oo es guided he mo ion
o cells unidi ec ionally combining con ac guidance and confine-
men ( Fig. 1D ). In his case, no p e e en ial di ec ion o mo ion was
obse ed and cells could mig a e inwa ds and ou wa ds he O- ing
ap wi h equal p obabili y (see below). We deposi ed highly in-
asi e MDA-MB-231 b eas cance cells on op o he ap and a -
e 1-2 h, mos o hem adhe ed uni o mly h oughou he su ace.
Non-adhe ed cells we e emo ed by eplacing he media by esh
one. We obse ed ha a e sho ime pe iods, cells we e mainly
loca ed wi hin he wells, and in some cases, he ou e egions we e
almos deple ed o cells (Fig. S3A). We hen ini ia ed he cell ap-
ping expe imen whe e cells we e allowed o mig a e o se e al
hou s wi hou in e up ion. Cells ha adhe ed di ec ly wi hin he
aps we e confined and immedia ely unde wen mig a ion, ol-
lowing an adhesion- unneling- apping mechanism by c ossing he
na ow in e connec ion b idge be ween neighbo ing mo i s ( Fig. 1B
and Fig. 1E ). In con as , cells ha adhe ed o he uppe fla egions
beha ed as fluc ua ing ou -o -equilib ium pa icles and mig a ed
andomly. E en ually, hey we e cap u ed by he opog aphic ea-
u es ha ac ed as apping po en ial wells (Fig. S3B and Mo ie
S1). Using fixed samples, we measu ed he ac ion o cells loca ed
in he uppe (MAX) and lowe (MIN) z-posi ions (i.e., ou side and
inside he channels), finding a pe cen age o 65%, 62%, and 59% in
he IN, OUT, and NULL condi ions, espec i ely (Fig. S4A-B). Sim-
ila ly, cells loca ed a he ou e egion and en ance o he ap
we e engul ed by he ea u es (Fig. S3C and Mo ie S2). No e ha
cance cells could e en ually escape he ea u es due o he limi ed
heigh o he channels and mig a e ac oss he op o he ap un il
hey we e apped again (Fig. S3D and Mo ie S3).
Nex , we in es iga ed he dynamics o cells wi hin indi idual
mo i s and ound ha cells exhibi ed s ochas ic mig a ion owa ds
ei he he poin ed o wide edge ( ip o base o he iangula mo-
i ). This mig a ion was acili a ed by he con ac guidance mecha-
nism, whe e cells adhe ed o he walls o he opog aphical s uc-
u e. Al hough cells had he abili y o mig a e in ei he di ec ion,
he asymme y o he iangula uni a o ed cell mig a ion o-
wa ds he poin ed edge. This p e e ence can be a ibu ed o he
mechanical in e ac ion be ween he cell and he wall a an app ox-
ima ely 80
ºangle, which inc eased he likelihood o dis up ing cell
pola i y and e e sing i s di ec ion o mo ion when mig a ing o-
wa ds he wide edge ( Fig. 2A and Mo ie S4). In con as , when
cells mig a ed owa ds he poin ed edge o he iangula s uc-
u e, he shallow angle be ween he walls o he ea u e and he
b idge induced apid hopping ( unneling) owa ds he neighbo ing
mo i , making i he p e e ed condi ion o guiding and apping
he cells. Th oughou his p ocess, he nucleus played a pi o al ole
in de e mining he di ec ion o cell mo ion by allowing o p e en -
ing cell passage h ough he b idge. In doing so, he nucleus unde -
wen mechanical de o ma ion ( Fig. 2B ), which did no occu when
cells mo ed along he g oo es (NULL). This mechanical de o ma-
ion is an ene gy-in ensi e p ocess and is mo e a o able when
cells a e mo ing owa ds he poin ed edge, as indica ed by he en-
e gy p ofiles depic ed in Fig. 1E . Fu he mo e, his in e ac ion also
led o pauses in mo ion as cells ansi ioned om one mo i o he
neighbo ing one.
To alida e his obse a ion, we conduc ed an analysis o he lo-
ca ion and in e ac ion o cell nuclei wi h he opog aphical walls.
Ou findings confi med ha he nucleus was he p ima y o ganelle
in e ac ing wi h he walls in he wide edge when cells mig a ed
owa ds his di ec ion ( Fig. 2C ). As men ioned ea lie , his in e ac-
ion esul ed in he loss o pola i y and p omo ed he e e sal o
cell mo ili y owa ds he poin ed edge, which is he p e e ed di-
ec ion o mo ion. The p e e ence o he poin ed edge was also
e iden in he measu emen o cell nuclei loca ion. The densi y
map o he mean nuclei posi ion on a iangula s uc u e e ealed
a non-uni o m dis ibu ion, p ima ily concen a ed in he poin ed
edge egion ( Fig. 2D ). This finding sugges s ha he opog aphical
ap cap u es he nucleus-cell complex in a manne analogous o
an ene gy po en ial well, as explained abo e.
To explo e whe he he in e ac ion wi h he opog aphical wall
s uc u e influenced he o ien a ion o he nuclei, we quan ified
he o ien a ion angle ( α) wi h espec o he a che labo a o y
amewo k axis on fixed samples. Ou analysis e ealed a peak dis-
ibu ion a ound 0
º, which con as ed wi h he andom o ien a ion
obse ed o cell nuclei loca ed in he uppe fla egions ( Fig. 2E ).
Fu he mo e, e en when only a po ion o he cell was loca ed
wi hin he a che s uc u es, he nucleus emained apped inside,
he eby influencing he o e all cell locomo ion p ope ies ( Fig. 2F ).
Nex , o u he demons a e he c ucial ole o he nucleus, we
acked i s posi ion and o ien a ion by fi ing i o an ellipse a pe-
iodic ime poin s. This analysis highligh ed he mechanical in e -
ac ion be ween he nucleus and he walls, u he suppo ing i s
impo ance in he a che ing mechanism ( Fig. 2G and Mo ie S5).
Las ly, we quan ified he pe cen age o nuclei loca ed inside and
ou side he aps, finding a clea bias o IN/OUT condi ions, sug-
ges ing he key ole o he nuclei in he mechanism o apping
and cell guidance ( Fig. 2H ).
3.2. Guiding he long- e m mo ion o cance cells in he a che aps
The efficiency o he cance ap is di ec ly p opo ional o he
abili y o cells o unde go long- e m di ec ed mig a ion, so we nex
moni o ed how cance cells mig a ed along he aps. Gene ally,
cells we e smalle han he a ea o an indi idual ea u e, as ea lie
indica ed, and he e o e, se e al could be alloca ed wi hin a sin-
gle ea u e a e seeding (Fig. S4C). Ne e heless, when he sys em
eached he s eady s a e a e ew hou s, cells dis ibu ed along he
a ay and mainly one o wo cells we e ound wi hin an indi id-
ual mo i ( Fig. 3A ). Unde hese condi ions, cells mig a ed mainly
a ached o he opog aphical walls ia a con ac guidance mecha-
nism. Fig. 3B shows ime-sequence images o a cell mig a ing along
he opog aphical s uc u es (Mo ie S6). Fo IN and OUT a che
aps, indi idual cells mig a ed pe sis en ly o se e al hou s wi h-
ou s opping o e e sing mo ion. No e ha cells we e also capa-
ble o s ochas ically e e sing hei mo ion and mig a ing owa ds
he di ec ion se by he base o he iangula ea u es (Fig. S5
205
D. Caballe o, R.L. Reis and S.C. Kundu Ac a Bioma e ialia 170 (2023) 202–214
Fig. 2. The ole o cell nuclei in he apping and a che ing o cance cells. (A) Re e sing he mo ion o cance cells wi hin an indi idual a che s uc u e. (Le ) Scheme
illus a ing he p ocedu e o e e sing he di ec ion o mo ion:
(1) Di ec ed mig a ion – owa d he base o he iangle –along he walls h ough con ac guidance; (2) Me-
chanical in e ac ion wi h he wall and pola i y dis up ion;
(3) Re e sal o mo ion; (4) Di ec ed mig a ion and hopping owa ds he neighbo ing s uc u e. (Righ ) Phase
con as image showing a ime sequence o his e e se e en . Whi e a owheads indica e he posi ion o he cell and la ge a ows he di ec ion o cell mo ion (see also
Mo ie S5. Scale ba : 100
μm. (B) Mechanical de o ma ion o he cell nucleus squeezing h ough he b idge walls. (C) Mechanical in e ac ion o he nucleus wi h he walls
om he wide and poin ed edge. (D) Densi y map o cell nuclei posi ion wi hin an indi idual ea u e ( n = 231 nuclei). (E) Nuclei o ien a ion inside and ou side he a che
s uc u e (n
a che
= 141; n
fla
= 227 cells; N = 3). (F) Immunos aining image depic ing he apping o he nuclei and he pa ial sp eading o he cell ou side he ap, as indi-
ca ed by he whi e a owhead (in ed, F-ac in; in blue, DAPI). (G) Time-sequence images highligh ing he posi ion and in e ac ion o he nucleus wi h
he a che walls. (H)
Pe cen age o nuclei loca ed inside o ou side he ap o all condi ions ( N = 3; n = 227). Scale ba s: 100
μm.
and Mo ie S7). In he NULL condi ion, cells spon aneously pola -
ized and mig a ed di ec ionally in bo h di ec ions wi h equal p ob-
abili y, displaying a fluc ua ing beha io as expec ed ( Fig. 3C and
Mo ie S8). We u he in es iga ed whe he he efficacy o guid-
ing and apping cance cells was a ec ed by cell densi y, ha is,
wi h mul iple cells wi hin he ap. In e es ingly, we ound ha
he capaci y o he ap o cap u e (and bias) he mo ion o cells
was main ained, e en hough a e y high densi ies, cell jamming
had a p onounced e ec on cell locomo ion capabili ies by in en-
si ying cell confinemen and mechanical in e ac ions ( Fig. 3B - 3C ,
Mo ie S7, and Fig. S6). I is impo an o no e ha de e mining he
p ecise cell densi y h eshold o jamming is challenging in such a
dynamic and e ol ing sys em. Fo mode a e densi ies, cells could
mig a e di ec ionally as indi idual en i ies, e en hough he physi-
cal in e ac ion be ween he cells in e e ed wi h he efficacy o he
ap, ansien ly delaying he a che ing e ec . This was mo e e i-
den o he NULL condi ion, whe e only one cell could fi wi hin
he g oo e and he in e ac ion be ween wo cells caused hem o
e e se hei mo ion (Mo ie S8). Howe e , as men ioned ea lie ,
his ansien jamming e ec disappea ed when he s eady s a e
was eco e ed in he a che ap. Al oge he , on a e age, he cells
mig a ed owa ds he poin ed edge o he opog aphical ea u es in
IN/OUT condi ions.
We nex quan ified he p obabili y o a cell o mig a e owa ds
one o he opposi e di ec ion. Fo his, we measu ed he posi ion
o a cell a he beginning and he end o he expe imen and mea-
su ed he o al ne mo ion owa d he poin ed o wide edge. To
assess he deg ee o di ec ionali y and pe sis ence, we defined se -
e al biophysical pa ame e s ( Fig. 4A and Me hods). We disc e ized
ou sys em in o la ice uni s (l.u.) finding ha he IN/OUT a che
ap had a biasing e ec , wi h an a e age cell mig a ion owa d he
poin ed edge (60 ±5 % s 40 ±5 %) ( Fig. 4B ). In con as , he con-
ol NULL condi ion did no p o ide any significan bias (52 ±10
% s 48 ±7 %), confi ming he a che ing e ec o he asymme -
ic s uc u es. Nex , we measu ed he mean bias pe l.u. defined as
ollows [33] (see Me hods):
p =
n
+
−n
−
n
+
+ n
−
(1)
206

D. Caballe o, R.L. Reis and S.C. Kundu Ac a Bioma e ialia 170 (2023) 202–214
Fig. 3. Long- ange di ec ed mig a ion o cance cells along a che cance aps. (A) Cell cy oskele on s aining showing ha he majo i y o cells we e loca ed ( apped) wi hin
he opog aphical ea u es. (B-C) Time-sequence illus a ing he long- e m mig a ion o indi idual and mul iple cells o IN/OUT and NULL condi ions. Whi e a owheads
indica e he posi ion o he cells. Las panel shows he comple e mig a ion pa h acking o e ime. Time in hh:mm. Scale ba : 100 μm. (S a is ics: N = 4; n
cells
= 100).
By defini ion, he a e age alue o < p > is -1 < P < 1.
Fig. 4C shows he end obse ed in he o al cell bias, wi h cells
p e e en ially mo ing owa ds he poin ed edge o IN and OUT
condi ions. Despi e obse ing cells mig a ing long dis ances wi h-
ou e e sing in he NULL g oo es, a alue o < p > ∼0 was ob-
ained a e a e aging ou , indica ing a clea s ochas ic beha io as
expec ed. We hen e alua ed he pe sis ence o cell locomo ion in
he a che (IN/OUT condi ion), finding ha cells mig a ed mo e
pe sis en ly owa ds he poin ed edge, co e ing longe dis ances
and pe iods (L
pe
poin ed > L
pe
wide and T
pe
poin ed > T
pe
wide
), confi m-
ing he p e e en ial di ec ion o mo ion ( Fig. 4D , 4E ). This was in
con as o cells mig a ing along g oo es, whe e he e was no sig-
nifican di e ence in L
pe and T
pe in ei he di ec ion. In e es ingly,
we obse ed ha L
pe
NULL < L
pe
IN/OUT
, likely due o he absence
o mechanical cues in he su ounding en i onmen o NULL con-
di ion ha ails o s imula e p e e en ial cell mig a ion. Addi ion-
ally, we obse ed ha T
pe was sligh ly la ge owa ds he poin ed
edge compa ed o he wide edge, wi h T
pe
NULL being significan ly
smalle han T
pe
IN/OUT
. This o e all indica es a mo e s able pola -
i y memo y when cells mig a e owa ds he poin ed edge. In e es -
ingly, cells mig a ed wi h he same eloci y ega dless o he di ec-
ion and mo phology o he aps, sugges ing ha cells mig a ed by
207
D. Caballe o, R.L. Reis and S.C. Kundu Ac a Bioma e ialia 170 (2023) 202–214
Fig. 4. Biophysical analysis o cance cell mig a ion on a che aps. (A) Scheme desc ibing he key biophysical pa ame e s used o quan i y cell bias and pe sis ence. (B)
Quan ifica ion o he pe cen age (%) o cell bias. (C) Plo o < p > o each condi ion as gi en by
Eq. (1) . (D-F) Analysis o he di e en biophysical pa ame e s: cell bias
(B), pe sis ence leng h –L
pe
(C), pe sis ence ime –T
pe
(D), and mean pe sis ence speed –< >
pe
(E). ( N = 3; n
cells
= 229;
∗P < 0.05;
∗∗P < 0.01;
∗∗∗P < 0.001). (G) Pe cen age o
inc ease o cell a ea o all he s udied condi ions as indica o o confinemen efficacy ( N = 3; n
aps
= 5 pe condi ion –15 in o al;
∗P < 0.05;
∗∗P < 0.01).
con en ional locomo ion mechanisms – con ac guidance and con-
finemen – ha a e no a ec ed by he asymme y o he sys em
o de o ma ion ( Fig. 4F ). This implies ha he a che does no im-
pac he capaci y o he cells o apply ac ion o ces bu only hei
abili y o a o mig a ion in one specific di ec ion. Las ly, he e was
no significan change in he biophysical pa ame e s o he NULL
condi ion in ei he di ec ion due o he symme y o he sys em
( Fig. 4D - 4F ). The biophysical pa ame e s we e s ongly dependen
on cell densi y; when he numbe o cells wi hin he a che in-
c eased, he a e o di ec ional swi ching inc eased, consequen ly
leading o a dec ease in pe sis ence (T
pe and L
pe
). Cell jamming,
esul ing om he s ochas ic o ma ion o ho spo s wi h an excess
o cells o as a esul o cell o e -p oli e a ion especially a e long
acquisi ion imes, was ound o u he enhance cell-cell mechan-
ical in e ac ions, causing a highe umbling equency, and hus, a
dec eased efficacy o he a che in sho ime in e als (Fig. S6).
Howe e , when he sys em was allowed o each equilib ium, he
efficacy was pa ially eco e ed.
Nex , o demons a e he efficiency o he ap, we quan ified
cell en ichmen in he h ee dis inc egions o he ap ( Fig. 4G ).
Specifically, we quan ified he cell a ea a he s a and conclusion
o he expe imen , and subsequen ly de e mined he inc ease in
he a ea occupied by he cells as shown in Eq.(2) :
% inc ease =
A
−A
0
A
0
×100 (2)
Whe e A
is he o al a ea occupied by cells a he end o he ex-
pe imen and A
0 is he espec i e a ea a he beginning. In e es -
ingly, he IN condi ion displayed a p onounced accumula ion o
cells in he inne egion (1) , con as ing wi h he OUT and NULL
condi ions. Addi ionally, a mode a e inc ease was also obse ed in
egion (2) , albei wi hou s a is ical significance. The g ow h ob-
se ed in egion (3) could be p edominan ly a ibu ed o cell p o-
li e a ion, which may also impac he esul s in he o he egions.
Howe e , he ob ained di e ences among he dis inc condi ions
sugges he e ec i eness o he ap in confining cance cells.
Nex , we es ed he specifici y o he aps o guide me as a ic
cance cells by es ing hei e ec on non-in asi e cells, such as
MCF-7, which display an epi helial pheno ype, non- umo igenic
MCF10A mamma y epi helial cells, and L929 fib oblas s. Fo non-
in asi e MCF-7, we obse ed ha in all he s udied condi ions (i.e.,
IN, OUT, and NULL) hese cells we e unable o unde go di ec ed
mig a ion (Fig. S7 and Mo ies S9-S11). MCF-7 cells p oli e a ed and
o med densely packed monolaye s, which we e no significan ly
a ec ed by he apping wells. The s onge bias exhibi ed by he
highly me as a ic cell line compa ed o he less agg essi e MCF7
cells sugges s ha mo e poo ly di e en ia ed cells may unde go
a che axis a he han hose wi h a mo e di e en ia ed epi he-
lial pheno ype. This epi helial pheno ype oge he wi h he la ge
size o he colonies inac i a ed he biasing e ec o he a che s.
As a esul , he di e en se o biophysical pa ame e s desc ip-
i e o apping efficiency and cell bias could no be quan ified.
Simila ly, o non- umo igenic MCF10A, we obse ed no significan
biasing e ec , wi h cells exhibi ing limi ed mig a ion du ing he
long acquisi ion imes. This ou come was again a ibu ed o hei
epi helial pheno ype, which p omo ed s ong cell-cell in e ac ions
and consequen ly inc eased ic ion, inhibi ing he e ec i eness o
he ap (Fig. S8 and Mo ie S12). Su p isingly, L929 fib oblas s did
no exhibi any ec ifica ion o hei mo ion ei he , mos p obably
due o he high dynamics and p oli e a ion a e o his cell line
compa ed o ha o he umo igenic MDA-MB-231 cells. This high
p oli e a ion led o a apid o e c owding wi hin he sys em. As a
esul , he mul iple mechanical in e ac ions among he fib oblas s
caused hem o apidly escape he confinemen o he ap, he eby
208
D. Caballe o, R.L. Reis and S.C. Kundu Ac a Bioma e ialia 170 (2023) 202–214
limi ing i s e ec i eness in e aining and biasing he cell mo ion,
and limi ing he ex apola ion and quan ifica ion o he biophysical
pa ame e s (Fig. S9 and Mo ie S13). This apid o e c owding could
be add essed by p e en ing cell p oli e a ion (e.g., an i-mi o ic
d ugs o nuclea s aining). Howe e , such an app oach migh in e -
e e wi h he na u al dynamics and in e ac ions among cells, which
a e c ucial o unde s anding he ap’s pe o mance unde ealis-
ic condi ions whe e p oli e a ion plays an impo an ole. Hence,
i is an essen ial pa ame e o conside . Finally, we conduc ed an
assessmen o he ap’s efficacy in so ing mixed popula ions o
cells comp ising MDA-MB-231 cells and L929 fib oblas s. Ou find-
ings e ealed dis inc beha io s: ini ially, cance cells we e apped
wi hin he opog aphical ea u es o he ap, o ming egula and
o de ed s uc u es esembling a wo-dimensional colloidal c ys al
(Fig. S10A). In con as , fib oblas s exhibi ed an iso opic dis ibu-
ion h oughou he ap. In e es ingly, as he sys em became mo e
c owded, he p esence o fib oblas s compelled he cance cells o
mig a e ou side he opog aphic wells, impac ing on he o e all e -
ficacy o he ap. This was suppo ed by he quan ifica ion o he
pe cen age o cell bias and < p > a ea ly ime poin s, as well as he
a ia ion in he amoun o MDA-MB-231 cance cells in he ap-
ping egions, which e ealed a dec ease in he ob ained alues o
a mixed popula ion o cells compa ed o hose ob ained o only
cance cells (Fig. S10B-D). This obse a ion highligh s how cell in-
e ac ions and jamming can impac he e ec i eness o he ap,
which demons a es g ea e efficacy in cap u ing isola ed and in-
asi e cance cells.
These findings collec i ely sugges ha he ap exhibi s limi ed
e ec i eness wi h he non- umo igenic and non-me as a ic cells
used and highe efficacy on he in asi e cells es ed, alida ing
he selec i i y o he aps in p ima ily a ge ing me as a ic cance
cells. This indica es ha he mechanisms discussed ea lie , such as
pola iza ion, nuclei posi ioning, and con ac guidance, may no uni-
e sally apply o all cell ypes. To u he in es iga e hese assump-
ions, u u e expe imen s in ol ing a b oade ange o cell ypes
would be necessa y.
3.3. Modula ing he efficacy o cance aps h ough scaling concep s
We obse ed ha ela i ely la ge cells ended o mig a e o-
wa ds he base o he aniso opic ea u es, which is he opposi e
di ec ion om wha we had claimed p e iously, sugges ing ha
cells in eg a e he ela ionship be ween cell and ea u e sizes o
de e mine he di ec ion o mig a ion. Fo his eason, and o u -
he explo e he ole o scaling in cell apping and guidance, we
nex in es iga ed how he size ela ion influenced cell di ec ional-
i y. Fo his, we adjus ed he ap size close o ypical cell dimen-
sions by c ea ing aniso opic ea u es ha we e 1.2- old smalle
han he o iginal ones (deno ed as ap - in Table S1; Fig. S11A-B).
We fi s in es iga ed how a dec ease in size a ec ed he capabili y
o he ap o cap u e cance cells and he abili y o hese cells o
mig a e, since cell densi y inc eased. A e 1-2 h o seeding, mos
cells we e al eady apped wi hin he wells e en hough a la ge
accumula ion o cells was obse ed compa ed o he la ge ap
( ap + ) p e iously u ilized (Fig. S11C). Despi e his highe densi y
o cells ha inc eased he mechanical in e ac ion be ween hem,
di ec ional mig a ion was s ill obse ed once he equilib ium was
eached. Ne e heless, we obse ed a sligh ly la ge pe sis ence o
cells seeded in he ap + a che s uc u es, pa icula ly owa ds
he poin ed edge, as indica ed by he a io o big o small alues
measu ed o all he biophysical pa ame e s (Fig. S11D). No signi -
ican a ia ion was obse ed o he NULL condi ion.
This obse a ion sugges s ha cance aps a e mo e e ec i e a
cap u ing lowe numbe s o cells, while s ill emaining e ec i e o
la ge quan i ies. Indeed, we obse ed ha cells could mig a e di-
ec ionally in he smalle aps ( ap -) bu hei efficacy depended,
no only in he amoun o cells apped, bu mainly on he ela-
i e size be ween he cell and he mo i ( Fig. 5A and Mo ies S14 –
S16). To gain mo e insigh s in o his, we classified he cells based
on hei size ela i e o he ea u e in oducing he size a io S
( Eq.3 ):
S
=
A ea cell
A ea ea u e
(0 < S ≤1) (3)
No e ha only unconfined cells smalle han he a ea o an
indi idual iangula a che uni we e included o he analysis,
which accoun ed o he majo i y o cells. To add ess he dy-
namic na u e o cell a ea, we measu ed he p ojec ed cell a ea us-
ing au oma ic h esholding (ImageJ, NIH) a di e en ime poin s
o each cell wi hin an indi idual a che uni and calcula ed he
mean alue. Fo con enience, we g ouped S
in o h ee ca ego ies:
small (S
= 0 –0.35), medium (0.36 –0.7), and big (0.71 –1)
( Fig. 5B ). Using he s anda d la ge ap, he mos efficien one
(see Fig. S11D), deno ed as ap + in Table S1, we measu ed S
o cells belonging o each o hese ca ego ies and he espec i e
bias. In e es ingly, we ound ha he scaling be ween he ea u e
and he cells influenced hei final di ec ion o mig a ion ( Fig. 5C ).
In b ie , smalle cells (la ge S ) mig a ed, in a e age, owa ds he
poin ed edge as p e iously obse ed, whe eas bigge ones (small
S ) mig a ed p e e en ially owa ds he wide edge, he eby e e s-
ing hei mo ion. Medium-sized cells (a e age S ) also mig a ed
owa d he poin ed edge, bu expe ienced a highe difficul y in
c ossing o he neighbo ing mo i s, mos likely due o he highe
mechanical in e ac ion be ween he nuclei and he in e connec -
ing b idge ( Fig. 5A - mid and Mo ie S15). This esul ed in la ge
cells acing mo e difficul ies in passing h ough i . As a esul , he e
was an inc eased likelihood o hese cells e e sing hei di ec ion
o mo ion. Howe e , he p e iously discussed a che ing mecha-
nism emained alid in his scena io. Fo e y big cells (small S ),
o he ac o s, such as p o usion adhesions, could syne gize wi h
he nucleus and ac as c ucial de e minan s causing he e e sal
o cell di ec ionali y. To confi m his hypo hesis, addi ional expe i-
men s analyzing p o usion ac i i y and adhesion may be equi ed
[ 24 , 32 ],
Al oge he , hese scaling expe imen s ha e alida ed ou hy-
po hesis and app oach, and may enable he design o mo e effi-
cien aps ha a e mo e cell-specific, o e en ually, could be used
o sepa a e cells based on hei ela i e size [ 30 , 34-37 ].
3.4. Res ic ing he dissemina ion o cance cells om mul i-cellula
sphe oids
While he aps ha e a well-demons a ed capaci y o guide
cance cells owa ds he apping egion, we nex in es iga ed hei
capabili y o e ain he dissemina ion o cance cells. Fo his, we
seeded mul i-cellula cance sphe oids in he cen e o he ap-
ping egion and allowed hem o dissemina e o abou 60 h. To
be e isualize cell sp eading and mig a ion, and o a oid bias
in he quan ifica ion o ap efficacy, cell p oli e a ion was inhib-
i ed by s aining li e cells wi h high concen a ions o DAPI (see
Me hods). Few hou s a e seeding, cance cells s a ed o escape
ou he sphe oid and dissemina e uni o mly as single cells, which
mechanically in e ac ed wi h he inne edge o he ap ( Fig. 6
and Mo ie S17). Cells had a simila p obabili y o in ading bo h
IN/OUT a che a ays and he non-di ec ional NULL g oo es be-
cause he access o he channels (i.e., b idge) was iden ical in all
cases (d
in
= d
ou
= d
NULL
) in all cases ( Fig. 1 ). Once inside he
channel, he cells we e exposed o he same mechanical cues dis-
cussed be o e. Fo he IN/OUT condi ions, cell mo ion was biased
mainly owa d he poin ed edge o he indi idual ea u es, he e-
o e, con ibu ing o confine (IN) o sp ead (OUT) hei mo ili y.
Despi e he enhanced confinemen , cance cells could s ill in ade
209
D. Caballe o, R.L. Reis and S.C. Kundu Ac a Bioma e ialia 170 (2023) 202–214
Fig. 5. Modula ing cell guidance h ough scaling. (A) Time-lapse sequence showing he mig a ion o cells o inc easing sizes in iden ical a che aps as highligh ed by he
di e en S
ac o s. Fo small and med S
alues, cells mig a e on a e age owa ds he poin ed edge, whe eas o la ge S
cells e e se he di ec ion o mo ion and mig a e
mainly owa ds he opposi e di ec ion. The dashed ec angle highligh s a cell expe iencing difficul ies in hopping owa d he neighbo ing mo i due o he mechanical
in e ac ion be ween he nuclei and he walls o he opog aphical ea u e. Time in hh:mm. Scale ba s: 100
μm. (B) Plo o S
o he h ee defined scaling ca ego ies. (C) Bias
plo highligh ing he di e en mig a ion o cells depending on he S
alue. (S a is ics: N = 3; n
cells
= 96;
∗P < 0.01;
∗∗P < 0.05).
he IN- a che channels, bu on a e age, hey e e sed hei mo-
ion and mig a ed back o he inne apping egion. In con as ,
on a e age, he cance cells en e ing he OUT ap displayed longe
mig a ion dis ances owa ds he ou e egion, exhibi ing high L
pe
and T
pe alues (3.6 ±0.5 l.u. and 7.1 ±1.7 h, espec i ely) sim-
ila o hose obse ed o indi idual cells in he same condi ion
(see also Fig. 4D , 4E ). This demons a es he capaci y o he aps o
guide cance cells owa ds he apping egion, and e ain he dis-
semina ion o cance cells ou side he ap. The quali a i e efficacy
o he IN ap o con aining cance cells was fi s es ablished by
imaging he la ge accumula ion o cells in compa ison o he OUT
a che ap (Fig. S12A). The cells ha mig a ed ou wa ds we e p i-
ma ily con ained wi hin he channel. The NULL condi ion also a-
cili a ed he dispe sion o cance cells, p ima ily in he ou wa d di-
ec ion. In his pa icula case, a g ea e numbe o cells we e ob-
se ed ou side he channels, po en ially indica ing he occu ence
o jamming e ec s. To quan i y he amoun o dissemina ed cance
cells o each condi ion, we measu ed he no malized a ea o DAPI-
s ained cells a he beginning and end o he expe imen , showcas-
ing he g ea e capaci y o he IN- a che ap o confine cance cell
mig a ion (Fig. S12B). Finally, o u he suppo hese esul s, we
calcula ed he confinemen ac o ( ) a he inne apping a ea a
he end o he expe imen . The  ac o is he a io o he numbe
o cells loca ed in he inne egion o he o al numbe o cells lo-
ca ed in all egions (inside: 1; channels: 2; ou side: 3; see
Fig. 6D –
le ) as shown in Eq.(4) .
=
A
1

3
i =1
A
i
(4)
Whe e A
i (i = 1..3) is he o al nuclei a ea (p ojec ed) a
= 60
h. By defini ion, is a uni less quan i y wi h a alue in he ange
o 0 ≤≤1; he close o 1, he mo e confined he cells a e in
he inne egion. Fo IN a che aps, we ob ained a IN
= 0.90;
his high  alue was in con as o OUT
= 0.81 and NULL
= 0.83
( Fig. 6D – igh ). O e all, hese esul s demons a e he supe io
e ec i eness o he de eloped a che aps in limi ing he sp ead
o cance cells, o e ing p omising po en ial o con olling cell dis-
semina ion.
4. Discussion
Se e al wo ks ha e epo ed a my iad o (bio-) ma e ials (e.g.,
sca olds, hyd ogels, mic opa icles, e c.) o cap u ing cance cells
[ 11 , 12 , 38 , 39 ]. Howe e , mos o hese cell apping app oaches em-
ploy (bio-) chemical means o a ac he cells, which come wi h
many d awbacks (e.g., limi ed li e- ime) and undesi ed e ec s (e.g.,
in e e ence wi h o he non- umo al cells). To add ess hese issues,
he use o pu ely mechanical means o di ec he mo ion o can-
210