Science - USA (2022-02-11)

(fig. S4, E to J). Induction of leaders by p21
overexpression was also sufficient to up-regulate
PI3K and ITGb1(fig.S4,OtoR),whichwasalso
up-regulated by addition of the CDK1 inhibitor
RO-3306 (fig. S4, S and T). Notably, chemical
inhibition of PI3K and Rac1 prevented the
active migration of MMC-generated leaders
(fig. S5, A and B, and movie S7), indicating
that these pathways are also necessary for
p53- and p21-mediated leader cell migration.
Thus, p53 and p21 induce leader specifica-
tion by means of elevation of the leader fate
effectors PI3K and ITGb1, likely through CDK
inhibition.

Injury-induced p53 drives collective cell
migration during epithelial repair

Because p53- and p21-induced leaders share the
same molecular mechanisms as spontaneous

leaders to drive migration, we wondered

whether p53, p21, and the resulting CDK in-

hibition also play a role in the migration of

leaders driving epithelial repair. First, we

sought to determine the cell cycle status of

injury-induced leaders. We generated a cell

line expressing the cell cycle reporter FUCCI

(fluorescent ubiquitination-based cell cycle in-

dicator) ( 29 ) and monitored cell cycle progres-

sion in leaders emerging upon epithelial scratch.

We found that cells could become leaders in any

phase of the cell cycle (fig. S6A). However, they

spent more time in the cell cycle phase they

were in than did nonleader cells (Fig. 5A). This

indicates that leaders tend to experience cell

cycle delay and, therefore, likely also CDK in-

hibition. This delay was not a general feature

of migrating cells, because migrating followers

led by spontaneous leaders did not cycle more

slowly than nonfollower cells in the same cul-

ture (fig. S6B). Thus, a slower cell cycle is not an

obligate feature of cells engaging in migration.

In addition, scratch-induced leaders showed

high levels of both p53 (Fig. 5, B and C) and p21

(fig. S6, C and D). Furthermore, leaders induced

by barrier release (fig. S6E) also showed elevated

p53 (fig. S6, F and G) and p21 (fig. S6, H and I),

suggesting that the p53-p21 pathway is a general

driver of leader cell migration in MDCK cells.

We next sought to determine how p53 be-

comes activated at the edge of the injured

epithelial monolayer and hypothesized that

mechanical damage, inflicted when epithelia

are scratched, could be triggering p53 elevation.

To measure p53 activation in live and migrating

epithelial sheets, we generated a cell line carry-

ing a live fluorescent p53 reporter consisting of

short-lived nuclear localization signal (NLS)ÐGFP

Kozyrskaet al.,Science 375 , eabl8876 (2022) 11 February 2022 3 of 10

A

0hrs 12hrs

WT :

p53KO

MMC

+DNA damaging B

agent MMC

WT

WT (untreated)

WTMMC

(high p53)

WT

(low p53)

WT :

WT

MMC

0hrs 4hrs 8hrs 12hrs

C E

F G H I

-200 200

-300

-200

-100

100

WT/ WT: WTGFPMMC

WT/ WT: p53KOGFPMMC

(0,0) : Point of contact

x(μm)

y(μm)

0

0.5

1.0

Persistence

p = 0.0147

After Contact

WTMMCp53KOMMC

n= 7

Time (h)

Speed (WT)

Speed (WTGFPMMC)

Distance between WTGFPMMC and WT

T0 : Point of contact

-12 -8 -4 0 4 8 12 16

0

5

10

15

0

10

20

30

40

Speed(μm/h)

Dista

n

ce

(^) (μm)
Time (h)
Speed (WT)
Speed (p53KOGFPMMC)
Distance between p53KOGFPMMC and WT
T0 : Point of contact
n= 7
-12 -8 -4 0 4 8 12 16
0
5
10
15
0
10
20
30
40
)
m
μ(
e
c
na
ts
i
D
)h
/
m
μ(
de
ep
S
0hrs 12hrs
WT :
WT
D
0
20
40
60
80
100
% acting as leadersn=328 n=283
WTMMCp53KOMMC
p<0.0001
Fig. 2. p53 instructs leader cell behavior.(A) Experimental design for
coculture of untreated and mitomycin C (MMC)Ðtreated cells. (B) Movie stills
of cocultures of MMC-treated GFP-positive wild-type cells and untreated
unlabeled wild-type cells. (C) Movie stills from cocultures of untreated unlabeled
wild-type cells and untreated wild-type cells expressing nuclear GFP. (D) Percentage
of MMC-treated wild-type orp53KOcells acting as leaders. Mean values shown.
Here and in subsequent figures, empty circles indicate values obtained in each
repetition. (E) Movie stills of cocultures of untreated unlabeled wild-type and
MMC-treated GFP-positivep53KOcells. (FandG) Tracking of mean migration
speed of and mean respective distance between pairs of untreated wild-type
and MMC-treated wild-type cells (F) orp53KOcells (G). (H) Tracks of wild-type
cells upon contact with MMC-treated wild-type (blue) orp53KO(gray) cells.
(I) Persistence of migration of wild-type followers upon contact with MMC-
treated wild-type orp53KOcells. Black bars indicate median. Thenvalues along
thexaxis in (D) and (F) to (I) indicate the number of contacts. Data are
pooled from three biological replicates (D) or from selected movies of three
biological replicates [(F) to (I)]. Error bars indicate ±SEM in (D), (F), and (G).
Pvalues from logistic regression (D) or Mann-WhitneyUtest (I).
RESEARCH | RESEARCH ARTICLE