Nature | Vol 579 | 12 March 2020 | 289into CD45.2+ mice. As expected, 36 h after transfer, the CD45.1+ mono-
cytic MDSCs differentiated towards a more-interstitial macrophage-
like phenotype in the lungs of CD45.2 mice treated with low-dose AET
(Fig. 3e, Extended Data Fig. 8c). In summary, low-dose AET may skew
monocytic MDSCs towards an interstitial macrophage-like pheno-
type in the lung premetastatic microenvironment. This acquisition
of a MDSC-to-macrophage program is associated with CCR2 down
regulation (a known consequence of monocyte-to-macrophage dif-
ferentiation^30 ), thus also establishing the functional implications of
our transcriptional data. Moreover, we found that transferring wild-
type monocytic MDSCs rescued metastases in Ccr2-knockout mice
and resulted in shorter overall survival times, whereas transferring
wild-type polymorphonuclear MDSCs did not. By contrast, transfer-
ring wild-type monocytic MDSCs from mice treated with low-dose
AET brought only limited change to both the lung metastatic rate and
overall survival time of Ccr2-knockout mice (Extended Data Fig. 8d, e).
These findings imply that the decreased accumulation of monocytic
MDSCs in the premetastatic niche owing to low-dose AET is mainly
dependent on CCR2 signalling.
Low-dose AET increases overall survival
In our mouse models, low-dose AET reduces pulmonary metastases,
as shown by H&E staining, gross pathological findings and cone-beam
computed tomography imaging (Fig. 4a, Extended Data Fig. 9a–c). In
all three mouse models, low-dose AET prolongs disease-free survival
and overall survival time (Fig. 4b, Extended Data Fig. 9d). Compared to
adjuvant chemotherapy, low-dose AET confers both a longer disease-
free and overall survival in the LLC model (Fig. 4c). It has previously
been found that epigenetic therapy results in a strong attraction of
CD8+ T cells to NSCLC tumours and a reversal of the immune-exhaustion
profile of these cells^14. However, when we depleted CD4 and CD8 T cells
(from day 1 to day 6) in our LLC model, we observed that there was
no difference during low-dose AET either in disease-free or overall
survival—indicating that the reduction in metastases was independent
of T cells (Fig. 4d, Extended Data Fig. 9e). Finally, the combination of
low-dose AET and CCR2 antagonists in our LLC and HNM007 models
shows synergy in disease-free and overall survival (Fig. 4e, Extended
Data Fig. 9f, g).(^0) D0 D3 D6 D9 D12D 15
12
34
5
2040
6080
100
Days
Area of
me
tastatic nodules(mm
2 ) Mock
60 AE
100
0
036
20
4
2
91215
Mean area of metastaticDaysafterresection
nodules (mm
2 )
(^0) D0 D3 D6 D9 D12D 15
20
40
60
80
100
Numb
er of
me
tastatic nodules
Mock
AE
0 3691215
60
20
0
80
40
Daysafterresection
100
Number of
metastatic nodules
Mock
LD-AET MoLD-AETck
Mo
ck
LD
-AET
Day0 Day3 Day6 Day9 Day12Day 15
a
(^05) time (^1015005101520)
50
100
Percentsurvival
(^00102030)
50
100
Percent survival
0510 15 0102030
0
50
100
0
50
100
Lung-metastasis-free
survival (%) Su
rvival
(%)
0
50
100
Lung-metastasis-free
survival
(%
)
0
50
100
Su
rvival
(%)
0510 15 20 0102030
Daysafterresection Daysafterresection Daysafterresection Daysafterresection
Mock(n= 7)
LD-AET(n= 7)
Mock(n= 7)
LD-AET(n= 7)
Mock(n= 7)
LD-AET(n= 7)
Chemotherapy(n= 7)
Mock(n= 7)
LD-AET(n= 7)
Chemotherapy(n= 7)
bc
(^00102030)
50
100
Days
Percentsurvival
(^0051015)
50
100
Percent survival
(^00510150102030)
50
100
Percent s
urvival
(^00510150102030)
50
100
0
50
100
Lung-metastasis-free
survival
(%
)
Su
rvival
(%)
Daysafterresection Daysafterresection
Mock(n= 7)
LD-AET +
RS5 04393 (n= 8)
RS5 04393 (n= 9)
LD-AET(n= 8)
Mock(n= 7)
LD-AET +
RS 504393 (n= 8)
RS 504393 (n= 9)
LD-AET(n= 8)
de
0510 15 0102030
0
50
100
0
50
100
Lung-metastasis-free
)
%(l
avi
vr
us Su
rvival
(%)
Daysafterresection Daysafterresection
Mock(n= 10)
LD-AET(n= 8)
Anti-CD4/8(n= 8)
IgG(n= 8)
LD-AET +anti-
CD4/8(n= 9)
Mock(n= 10)
LD-AET(n= 8)
Anti-CD4/8(n= 8)
IgG(n= 8)
LD-AET +anti-
CD4/8(n= 9)
**
Fig. 4 | Low-dose AET inhibits pulmonary metastases and prolongs overall
survival in mouse models, mainly by affecting MDSCs. a, Representative
H &E-stained images of lung sections from LLC mice treated with low-dose AET
or vehicle at different time points after surgery. Scale bars, 2 mm. Graph shows
the area and numbers of metastatic nodules. At each time point, three mice
were killed for analysis. For each sample, sections from three levels were
analysed. Tumour area was quantified using Aperio Imagescope software. Two-
sample, two-sided t-test. All bars are mean ± s.e.m. b, Kaplan–Meier curves
showing the disease-free survival and overall survival of LLC mice treated with
low-dose AET or vehicle. c, Kaplan–Meier curves showing the disease-free
survival and overall survival of LLC mice treated with paclitaxel plus cisplatin
chemotherapy, low-dose AET or vehicle. d, Kaplan–Meier curves showing the
disease-free survival and overall survival of mice treated with vehicle, IgG
(isotype control), anti-CD4/CD8 (T-cell-depletion antibody), low-dose AET and
anti-CD4/CD8 in combination with low-dose AET in the LLC model. e, Kaplan–
Meier curves showing the disease-free survival and overall survival of mice
treated with vehicle, CCR2 antagonist (RS504393) (Sigma), low-dose AET and
RS504393 in combination with low-dose AET in the LLC model. In b–e, two-
sided log-rank tests were used. *P < 0.05, *P < 0.01, P < 0.001.