Science - USA (2022-02-25)

generate a highly functional and persistent
ILC3 subset, which we refer to hereafter as
“trained ILC3s”(Tr-ILC3s).
Enhanced Tr-ILC3 responses persisted
4 months afterC. rodentiumreinfection of
Abx-cleared mice (Fig. 2A), with reduced bac-
terial loads upon reinfection (Fig. 2B). Abso-
lute numbers of total ILC3s (Fig. 2C) and IL-22+
ILC3s (Fig. 2, D and E) were significantly in-
creased in CRACR mice that were reinfected
4 months later with diverse ILC3 subsets as
dominant IL-22 producers (Fig. 2, F and G).
Because the DNA-binding protein inhibitor
ID2 is highly expressed in differentiated ILCs
( 20 , 21 ), we usedId2-regulated inducible fate-

mapping ( 22 ) to track ILCs under steady state

and after infection (Fig. 2H). All ILCs were red

fluorescent protein (RFP)–labeled, with a some-

what higher percentage of RFP+ILC3s after

C. rodentiuminfection (CR and CRA conditions;

Fig. 2I). By contrast, most RFP+ILC1s and ILC2s

were lost in CRACR mice, whereas labeled RFP+

ILC3 subsets were maintained (Fig. 2I). These

results indicate that diverse Tr-ILC3s can persist

after a limited exposure toC. rodentiumand may

preferentially expand after pathogen reencounter.

We determined if Tr-ILC3s could be gen-

erated in the absence of adaptive immunity

(Fig. 3A). Absolute numbers of IL-22+ILC3s

were significantly increased in CRACRRag2−/−

mice (Fig. 3B), which had reduced bacterial

loads compared with the CR-infectedRag2−/−

mice (Fig. 3C). IL-22 production was required

for Tr-ILC3–mediated protection against

C. rodentium(Fig. 3A) becauseRag2−/−Il22−/−

CRACR mice did not survive reinfection (Fig.

3, D and E). We next compared the protective

capacity of adoptively transferred naïve versus

Tr-ILC3s in vivo (Fig. 3F). CR-infectedIl22−/−

mice that received naïve ILC3s exhibited body

weight loss and succumbed about 2 weeks

after infection (Fig. 3G). By contrast, CR-

infectedIl22−/−mice that received Tr-ILC3s

recovered and survived (Fig. 3G). Thus, the

enhanced functional capacity of Tr-ILC3s

860 25 FEBRUARY 2022•VOL 375 ISSUE 6583 science.orgSCIENCE

F G H

AB C D

GFP TdT Dapi

RorcGFP Il22TdT (SILP; D0)

Rorc

(GFP)

CD3

Intestinal CD45.2+ cells (SI)

11.3 ILC3

NKp46

CCR6 NKp46

+ ILC3 (x10

Time (days)

30 31 32 33 35

0

10

20

30

40

50

28.7 3.63

19.6 48

CR Abx

RorcGFP

Il22TdT

D0 D3 D30

CR

CR CRACR

Abx

Time (days)

CR CR

T cells

100

200

300

400

0 1 2 3 5 30 31 32 33 35

0

20

40

60

80

100

120

140

Cells ( 10ILC3

4 )

CCR6 ILC3 (

10

4 )

Time (days)

30 31 32 33 35

0

10

20

30

40

50

+

0 1 2 3 5 3031323335

0

10

20

30

40

50

0

4

8

12

CR AbxCR

Time (days)

T cells

ILC3

Il22

TdT+

cells (

10

4 )

Small intestine NKp46+ ILC3 CCR6+ ILC3

Day 3 Day 33

CR CRACR

Radiance (x10^6 )

30 31 32 33 35 246

0.0

0.2

0.4

0.6

0.8

1.0

30 31 32 33 35

0

1

2

3

4

5

CR CR

Time (days) Time (days)

Il22

TdT+

cells (% of CD45.2

+)

ILC1 NKp46 CCR6

49.1%

Il22

(TdT)

77.3 %

ILC3

Intestinal CD45.2+ cells (SI)

0123

0

3

6

9

12

15

18

T cells

ILC3

10

20

30

40

50

60

70

T cells

ILC3

Time (days)

E

IL-22

+ cells ( 10

4 )

30 31 32 33

Time (days)

IL-22

+ cells (

10

4 )

0

CD3

Il22

(TdT)

Rorc

(GFP)

CD127

97.2

4 )

CR CRACR CR CRACR

10

10

10

10

10

10

10

10

10

CFU/g of Feces

100

101

102

103

10

105

106

n.d.

Radiance (p/sec/cm

2 /sr)

0

1

2

3

4

5

6

7

8

***

4

3.2 0.9

++

Fig. 1. Tr-ILC3s efficiently control pathogenic bacteria rechallenge.(A) The
experimental design for (B) to (H) is shown at the top.RorcGFPIl22TdTmice
received ciprofloxacin (Abx; 100 mg kg–^1 day–^1 ) afterC. rodentiuminfection (CR).
One month later, the mice were reinfected withC. rodentium(CRACR). A
representative steady-state immunofluorescence analysis of RORgt+(green) and
Il22TdT+(magenta) cells in the small intestine is shown at the bottom. Nuclei were
stained with 4′,6-diamidino-2-phenylindole (DAPI) (blue) (scale bar, 20mm). SILP,
small intestine lamina propria; D0, day 0. (B) Intestinal NKp46+and CCR6+ILC3s
were analyzed by flow cytometry (top). Absolute numbers of ILC3s and T cells
in the small intestine lamina propria (n=4to12foreachtimepoint)areshownat
the bottom. (C) Absolute numbers of intestinal NKp46+(top) and CCR6+(bottom)
ILC3s determined with representative data from three independent analyses
(n= 3 to 7 for each time point). (D) Small intestinalIl22TdT+ILC3 (CD45+CD3−RorcGFP+)
and T cell (CD45+CD3+) frequencies at day 0 (top). Absolute numbers ofIl22TdT+

cells were determined with representative data from six independent analyses

(n= 4 to 12 for each time point) (bottom). (E) Absolute numbers of IL-22+

(protein) cells were analyzed after ex vivo IL-23 (top) and IL-1b(bottom)

stimulation (n= 4 to 9 for each time point). (F)Il22TdTexpression in ILC1s

(CD3−NKp46+NK1.1+), NKp46+ILC3s, and CCR6+ILC3s (left). Representative

data are from three independent analyses. The frequencies ofIl22TdT+NKp46+

and NKp46−ILC3 subsets in the small intestine lamina propria after reinfection

are shown on the right (CRACR;n= 3 to 7 for each time point). (G)C. rodentium

growth monitored by IVIS imaging (CR,n= 12; CRACR,n= 12). Representative

pseudocolor images are shown (color scale in photons s–^1 cm–^2 sr–^1 ). (H) Fecal

C. rodentiumcounts (CR,n= 8; CRACR,n= 13). Data are representative of three

independent experiments. Each graph corresponds to the mean ± SEM of the

values obtained. n.d. indicates not detected and ***P< 0.001 using two-tailed

Mann-Whitney test.

RESEARCH | REPORTS