nt12dreuar3esd

Nature | Vol 579 | 12 March 2020 | 275

tumour and NAT (Fig. 1a). To quantify our observations in these four
patients, we classified clones not only by their expansion in tissue
but also by their peripheral expansion, as measured by clone sizes in
blood (Fig. 1b, Extended Data Fig. 2e). Clones with single or multiple
constituent cells in blood were termed blood non-expanded and blood-
expanded, respectively, whereas clones without any associated cells
in blood were termed blood-independent. With this schema, we found
that an aggregate measure of peripheral expansion over all clones—as
determined by the fraction of blood T cells from blood-expanded
versus non-expanded clones (Fig. 1c)—associated with higher levels
of lymphocytic infiltration by blood-expanded clones into tumour
and NAT samples, and served as a proxy for the extent of dual expan-
sion (Fig. 1d). Notably, in patients with strong dual expansion, dual-
expanded clones in tissue could frequently be detected as expanded
in blood (Fig. 1e).
Our scTCR-seq data allowed us to make observations at the level
of clones rather than individual T cells. By studying clones, we found
that peripheral and intratumoural clone sizes were significantly cor-
related (P = 6.3 × 10−53, P = 8.2 × 10−10 and P = 0.0038) in our dataset
as well as in two external datasets^3 ,^4 (Fig. 1f–h), which indicates that
the relationship between peripheral expansion and tumour infiltra-
tion held not only for aggregate cell fractions but also for individual
clones. More notably, a general and significant (P = 5.5 × 10−4) associa-
tion was observed across patients in the combined datasets between
the total amount of peripheral expansion and parallel dual expansion
(Fig. 1i, Extended Data Fig. 3). Collectively, our results show a strong

relationship between peripheral clonal expansion and parallel infiltra-

tion into tumours and NAT.

We explored the phenomenon of clonal expansion further by simul-

taneously measuring gene expression using single-cell RNA sequenc-

ing (scRNA-seq). Transcriptional profiles of individual T cells allowed

grouping of similar cells into clusters (Fig. 2a), which we characterized

by cross-labelling our cells against reference gene signatures from

published datasets^3 –^5 (Extended Data Fig. 4a), and by profiling genes

we found to be cluster-specific (Extended Data Fig. 4b, Supplementary

Table 4). Cluster analysis identified T effector cells (8.1-Teff), T effector

memory cells (8.2-Tem), and regulatory T cells (4.6a-Treg, 4.6b-Treg), as

well as subtypes or physiological states of resident memory T (Trm)

cells^6 , which shared abundant expression of ITGAE (also known as

CD103) and ZNF683 (Hobit), a regulator of tissue residency^7 , as well as

positive scores on a published gene signature of Trm cells^8 (Extended

Data Fig. 4c). These subclusters, designated 8.3a-Trm, 8.3b-Trm and 8.3c-

Trm, differed by low, intermediate and high expression of activation-

exhaustion markers, respectively, including PD1 (also known as PDCD1)

(Extended Data Fig. 4d), and also by a published gene signature for ter-

minally exhausted versus stem-like exhausted CD8+ T cells^9 (Extended

Data Fig. 4e), consistent with recently identified phenotypes of pro-

genitor exhausted and terminally exhausted Trm cells^10. Subcluster

8.3c had lower expression of activation marker XCL1 (which encodes

lymphotactin), while 8.3b had exclusively high expression of KLRC2

(also known as NKG2-C) and KLRC3 (NKG2-E) (Extended Data Fig. 4b),

observed in CD8+ T cells with both TCR-dependent and innate-like,

0210 50

0

5

50

0210 50

0

2

10

50

0550

0

10

200

0.20.4 0.60.8

−0.4

0.4

Peripheral expansion

Parallel dual exp

Renal2

nD = 139 rw = 0.93

00 .001 0.01

Renal1

nD = 108 rw = 0.81

00 .01

Renal3

nD = 119rw = 0.58

0 0.01

Lung 6

nD = 213rw = 0.31

0 0.001

Blood singleton (b)

Blood multiplet (B)

NAT multiplet (N)

Dual expanded (D)

Tumour singleton (t)

Tumour multiplet (T)

Db NAT singleton (n)

DB

N– D– T–

Nb

NB

Tb

TB

n– t–

tb

tB

nb

nB

b

B

Blood non-expanded (Bl = 1)

Blood-expanded (Bl > 1)

Independent

(Bl = 0)

Associated (Bl ≥ 1)

Tu

01

0> 1

≥1≥1

>1 0

10

001

00 >1

NATBl

Clone size

Tissue expansion pattern

NAT cell fraction

Tumour

cell fraction

Clone size, blood

Clone size,tumour

nNDTt nNDTt nNDTtnNDTt

This study Ref. 3 Ref. 4

Clone size,

blood

Blood singleton

Blood multiplet

Clone size, NAT

Clone size,

tumour

(^0) ∞
1
∞
(^01)
Ind
Non
Exp
Renal1–3, Lung6 Lung (ref. 3) Colon (ref. 4)
Exp = 0.82 0.74 0.50 0.04
0.87 0.83 0.63
0.05
Peripheral clonal expansion. Blood-expanded versus non-expanded
Tissue inltration. Distribution over [n, N, D, T, t, b, B]
Detection in blood-expanded clones. Example: for D, compute DB/(DB + Db + D–)
0
1
2
5
10
20
50
100
D = 0.96 0.94 0.80 0.52
n = 7,141 P = 0 n = 3,451 P = 0 n = 2,488 P = 4.2 × 10–217 n = 9,688 P = 7.6 × 10–115
n = 126 r = 0.51
P = 8.2× 10–10
n = 121 r = 0.26
P = 0.0038
n = 328 r = 0.72
P = 6.3× 10–53
n = 22
P = 5.5× 10–4
c
d
e
b
a
fghi
0
0.01
0.1
0
0.01
0.1
0
0.01
0
0.001
0.01
1
0
1
0
1
0
1
0
Fig. 1 | Parallel dual expansion and peripheral clonal
expansion. a, Clonal expansion scatter plots. Scatter
plots of clonotypes are shown for patients with blood
samples, plotted by normalized clone size, or cell
fraction, in NAT and tumour, with random jitter to
distinguish points. Dots are sized for blood clone size,
and coloured by a palette for tissue expansion pattern.
Diagonal lines indicate equal cell fractions. Other lines
separate the absence and presence of clones within
compartments. rw denotes Pearson’s correlation
coefficient (r), weighted by (1 + blood clone size), on the
dual-expanded clones (nD). b, Schema for clonal
expansion. Tissue and blood expansion patterns are
defined by clone sizes in tumour (Tu), NAT and blood (Bl).
c, Peripheral clonal expansion. Bar plots for each patient
in a show fractions of cells from blood-expanded (values
shown) versus non-expanded clones, coloured as in b. d,
Tissue infiltration. Bar plots show distributions of cells
by tissue expansion patterns from blood-independent
(Ind), non-expanded (Non) and expanded clones (Exp).
Values for dual-expanded clones (D) are shown. P values
were determined by a chi-square test on counts of tissue-
resident cells (those in tumour or NAT but not in blood).
e, Detection of tissue TCRs in blood. Bar plots show
fractions of tissue-resident cells with clonotypes
observed in a blood-expanded clone, for each tissue
expansion pattern (values for D are shown). n, NAT
singleton; N, NAT multiplet; t, tumour singleton;
T, tumour multiplet. f–h, Correlation of blood and
tumour clone sizes. Scatter plots show blood and tumour
clone sizes from patients in a and external datasets^3 ,^4 ,
coloured as in a. Horizontal and vertical lines separate
clone sizes of zero and non-zero. Two-sided P values are
shown from a Pearson correlation coefficient r on log-
transformed clone sizes for clones present in both
tumour and blood. i, Peripheral expansion and parallel
dual expansion. Values are summarized for patients in f–
h. P values were determined by a two-tailed linear
regression t-test, after exponentiating y values to
linearize the data, across patients.