Article

Clonal inactivation of TERT impairs stem cell

competition

https://doi.org/10.1038/s41586-024-07700-w Kazuteru Hasegawa1,2,3, Yang Zhao4, Alina Garbuzov1,2,3, M. Ryan Corces4, Patrick Neuhöfer1,2,3,
Victoria M. Gillespie5, Peggie Cheung1,2,3, Julia A. Belk6, Yung-Hsin Huang4, Yuning Wei4,
Received: 13 August 2021
Lu Chen1,2,3,5, Howard Y. Chang4,7 & Steven E. Artandi1,2,3 ✉
Accepted: 11 June 2024

Published online: xx xx xxxx

Telomerase is intimately associated with stem cells and cancer, because it catalytically
Open access elongates telomeres—nucleoprotein caps that protect chromosome ends1.
Check for updates Overexpression of telomerase reverse transcriptase (TERT) enhances the proliferation
of cells in a telomere-independent manner2–8, but so far, loss-of-function studies have
provided no evidence that TERT has a direct role in stem cell function. In many tissues,
homeostasis is shaped by stem cell competition, a process in which stem cells compete
on the basis of inherent fitness. Here we show that conditional deletion of Tert in the
spermatogonial stem cell (SSC)-containing population in mice markedly impairs
competitive clone formation. Using lineage tracing from the Tert locus, we find that
TERT-expressing SSCs yield long-lived clones, but that clonal inactivation of TERT
promotes stem cell differentiation and a genome-wide reduction in open chromatin.
This role for TERT in competitive clone formation occurs independently of both its
reverse transcriptase activity and the canonical telomerase complex. Inactivation of
TERT causes reduced activity of the MYC oncogene, and transgenic expression of MYC
in the TERT-deleted pool of SSCs efficiently rescues clone formation. Together, these
data reveal a catalytic-activity-independent requirement for TERT in enhancing stem
cell competition, uncover a genetic connection between TERT and MYC and suggest
that a selective advantage for stem cells with high levels of TERT contributes to
telomere elongation in the male germline during homeostasis and ageing.

Telomerase is enriched in tissue stem cells and activated in many tissue function by promoting the replacement of less-fit stem cells by
cancers by somatic TERT promoter mutations9–11. The core of the tel- more-robust neighbouring stem cells17–21. This competitive behaviour
omerase enzyme comprises the catalytic subunit TERT and the tel- is also characteristic of carcinogenesis, during which oncogenic muta-
omerase RNA component Terc—a small non-coding RNA scaffold that tions drive cells to clonally expand their territory through a process of
encodes the template for telomere addition1. The crucial requirement super-competition22. Among tissues in which competitive repopulation
for telomerase in long-term cell viability is conserved from single-cell has been observed, the testis shows high telomerase activity and exhib-
eukaryotes to humans12,13. Cell proliferation in the absence of telomer- its unusual telomere dynamics in that telomere lengths are preserved
ase results in a lag phase that is at first well tolerated, while telomere with ageing, in contrast with the progressive shortening that is seen in
reserves are ample, but which culminates in senescence or cell death other human tissues23,24. We previously found that SSCs express high
as telomeres become short and dysfunctional. In laboratory mice with levels of TERT and that TERT is downregulated with lineage commit-
very long telomeres (40–80 kb, as compared with 5–15 kb in humans), ment16. Here, to investigate the telomere-independent role of TERT
germline inactivation of Tert or Terc results in viable mice, but sub- in stem cells, we developed a system to mark single SSCs expressing
sequent intergenerational breeding is accompanied by progressive TERT, coupled with the ability to conditionally inactivate TERT using
telomere shortening, which results in severe tissue defects owing to a lineage-tracing approach, and examined how clonal TERT deletion
critically short telomeres after six generations14–16. In contrast with affects stem cell competition.
these loss-of-function studies, overexpression studies have revealed
non-canonical functions of TERT, which were separable from telomere
synthesis because they also occurred either with a catalytically inac- TERT expression in spermatogenesis
tive Tert allele or in mice lacking Terc (refs. 3,4,8). In this context, TERT Spermatogenesis is a dynamic process through which SSCs give rise to
promoted proliferation or impaired apoptosis, and was shown to acti- sperm through hierarchical mitotic divisions, meiosis and post-meiotic
vate the MYC, WNT and NF-κB pathways2–8. Many renewing tissues are maturation (Extended Data Fig. 1a,b). In the testis, SSCs reside in a
characterized by stem cell competition, a mechanism that optimizes functionally and morphologically heterogeneous population termed
1
Stanford Cancer Institute, Stanford University School of Medicine, Stanford, CA, USA. 2Department of Medicine, Stanford University School of Medicine, Stanford, CA, USA. 3Department of
Biochemistry, Stanford University School of Medicine, Stanford, CA, USA. 4Center for Personal Dynamic Regulomes, Stanford, CA, USA. 5Nuclear Dynamics and Cancer Program, Cancer
Epigenetics Institute, Fox Chase Cancer Center, Philadelphia, PA, USA. 6Department of Computer Science, Stanford University, Stanford, CA, USA. 7Howard Hughes Medical Institute, Stanford
University, Stanford, CA, USA. ✉e-mail: sartandi@stanford.edu

Nature | www.nature.com | 1
Article
undifferentiated spermatogonia (US). Singly isolated Asingle (As) US six months, deletion of TERT caused a marked reduction in the number
undergo incomplete cytokinesis, subsequently producing progres- of tdTomato+ clones and diminished the mean patch length (Fig. 1l–n).
sively elongating chains of Apaired (Apr: 2 connected US) and Aaligned (Aal: Correspondingly, the total aggregated tdTomato+ patch length was
4 (A4), 8 (A8) or 16 (A16) connected US)25 (Extended Data Fig. 1b). Matu- sharply reduced (Fig. 1o). Genotyping of tdTomato+ patches at three
ration of US yields differentiating spermatogonia (DS)—committed months revealed that 36.2% (34/94) of clones did not fully recombine
progenitors that lack stem cell potential. To measure Tert expression in the Tert allele (Extended Data Fig. 3d), indicating an enrichment of
distinct subpopulations of spermatogonia, we purified MCAMhighKIT− patches that retain TERT. At three months after injection with higher
US (US-h), MCAMmedKIT− US (US-m) and MCAMmedKIT+ DS (Extended doses of tamoxifen (5 mg), tdTomato+ areas in seminiferous tubules
Data Fig. 1a–c), as we reported previously26. Strong expression of MCAM remained unchanged, suggesting that the elimination of TERT-deleted
closely overlapped with the expression of GFRA1, a marker enriched SSCs is alleviated in a less competitive environment (Extended Data
in As and Apr cells (Extended Data Fig. 1d,e). Among these subpopula- Fig. 3e,f). To further differentiate between a role for TERT in stem cell
tions, Tert mRNA expression was high in both US-h and US-m cells, and competition versus a requirement for TERT in SSC function, mice were
decreased in DS cells (Extended Data Fig. 1f). Similarly, in Tert-Tdtomato treated with repeated high doses of tamoxifen (5 mg for three consecu-
knock-in reporter mice, both US-h and US-m showed high tdTomato tive days) and analysed after a three-month trace. tdTomato expres-
expression (Extended Data Fig. 1g). These data indicate that TERT sion was detected in all seminiferous tubules, and telomerase activity
expression is enriched in the entire population of US cells. was sharply reduced, whereas spermatogenesis proceeded normally
(Extended Data Fig. 3g–i). This indicates that in this context of quan-
titative deletion throughout all US cells, TERT is not essential for SSC
Clone formation from TERT+ SSCs maintenance, which is consistent with results from first-generation
To functionally study TERT-expressing spermatogonia, we developed germline Tert-knockout mice16. Tert mRNA analysed by quantitative
a lineage-tracing assay using TertCreER/+;Rosa26lsl-Tdtomato/+ (TertCreER/+) reverse transcription PCR (qRT–PCR) at both exon 2 and exon 6 was
mice, in which TERT-expressing cells are permanently labelled after markedly reduced, which indicates that transcripts from the Tert-flox
tamoxifen-dependent activation of CreER to express tdTomato27,28 allele are lost by nonsense-mediated decay after Cre-dependent recom-
(Fig. 1a,b). Marking SSCs results in long-lived tdTomato+ clones, also bination, precluding notable expression of a truncated protein from
known as patches, which are composed of many daughter cells pro- the recombined allele (Extended Data Fig. 3j). Together, these findings
duced by the labelled SSC29. Conversely, if committed progenitor cells indicate that focal deletion of TERT in a subset of SSCs compromises
are labelled, only a small transient clone is generated and these cells are clone formation through a process of competition.
lost through differentiation (Fig. 1b). Sparse labelling allows rare SSCs
to be marked, and in this context each patch derives from a single SSC.
Thus, measuring the number of tdTomato+ patches allows a quantita- A role independent of catalytic activity
tive assessment of stem cell self-renewal activity. To determine whether the effect of TERT in enhancing SSC competition
At two days after tamoxifen injection, tdTomato was detected depends on its catalytic role in elongating telomeres, we performed
similarly throughout the population of US cells, but not in KIT+ cells several studies. Terc encodes the RNA template for telomere addition
(Extended Data Fig. 2a–c). At three months, marking TERT+ cells and serves as the central scaffold for assembly of the telomerase com-
yielded labelled patches, whereas no-tamoxifen controls did not pro- plex. Therefore, in the absence of Terc, TERT and the other components
duce patches (Fig. 1c and Extended Data Fig. 2d,e). As we varied the of telomerase do not associate1,30. To determine whether formation of
dose of administered tamoxifen from 0.25 mg to 4 mg, patch number the telomerase complex is required for TERT-dependent SSC clone for-
increased in a dose-dependent manner (Fig. 1d and Extended Data mation, we produced first-generation Terc−/−:TertCreER/flox:Rosalsl-Tdtomato/+
Fig. 2f). Patch length remained constant at doses lower than 1 mg but (G1 Terc−/−:TertCreER/flox) mice (Fig. 2a). In this strain, telomerase was inac-
increased for those higher than 2 mg, reflecting the fusion of two inde- tive in the testes, consistent with disruption of the complex through
pendently labelled clones with high dose (Fig. 1e and Extended Data Terc deletion (Extended Data Fig. 4a). Despite the absence of telomerase
Fig. 2f). In mice that were treated with 1 mg tamoxifen and traced for activity in these mice, conditional deletion of Tert in Terc-deficient mice
one year, patch length increased, clone number decreased and the total impaired patch number, patch length and total patch length when
aggregated tdTomato+ patch length (mean patch number times mean analysed three months after tamoxifen treatment (Fig. 2b–e). These
patch length) remained constant, compared with those traced for three findings indicate that SSCs depend on TERT for effective stem cell
months, consistent with previously described stochastic competition competition even in mice lacking telomerease, thereby uncoupling
with unlabelled clones29 (Fig. 1d–f and Extended Data Fig. 2f). At both the requirement for TERT in stem cell competition from engagement
three months and one year, patches were composed of cells throughout in the classical telomerase complex.
the spermatogenic lineage (Extended Data Fig. 2g). Together, these To understand whether the reverse transcriptase activity of TERT
data show that TERT-expressing SSCs generate long-lived clones and is required for stem cell compeitition, we developed a system to
exhibit competition within the stem cell pool. rescue the defect in TERT-deleted cells using tetracycline-regulated
TERT transgenes: either wild-type (TetO-TERT) or catalytically inac-
tive (TetO-TERTci) owing to a single amino acid substitution in the
Impaired clone formation with TERT loss catalytic site8. We produced compound mouse strains in which
To investigate a direct role for TERT in SSCs, we developed a competi- activation of CreER simultaneously deletes the floxed Tert gene
tive clone formation assay using TertCreER/flox:Rosalsl-Tdtomato/+ (TertCreER/flox) while inducing the expression of transgenic Tert by triggering the
mice (Fig. 1g and Extended Data Fig. 3a–c). In this strain, activation of expression of the tetracycline transactivator (tTA) that binds to and
CreER induces tdTomato labelling and concomitantly inactivates Tert activates the TetO promoters (Fig. 2f). We examined how TERT loss
in the same TERT+ cell. Sparse labelling using this strain enables one to affected clone formation in TertCreER/+:Rosalsl-Tdtomato/lsl-tTA (TertCreER/+)
trace the fate of cell clones deriving from SSCs in which Tert has been mice versus TertCreER/flox:Rosalsl-Tdtomato/lsl-tTA (TertCreER/flox) controls, and
somatically deleted in an environment in which most neighbouring we compared these results with the restoration of TERT expression
cells retain TERT expression. At two days after treatment with 1 mg in TertCreER/flox:Rosalsl-Tdtomato/lsl-tTA:TetO-Tert (TertCreER/flox + Tert) mice or
tamoxifen, GFRA1+ As and Apr clones were marked indistinguishably in TertCreER/flox:Rosalsl-Tdtomato/lsl-tTA:TetO-Tertci (TertCreER/flox + Tertci) mice
both TertCreER/+ and TertCreER/flox mice (Fig. 1h,i). Pulse labelling efficiently (Fig. 2f). Transgenic Tert and Tertci expression were induced at a physi-
eliminated Tert mRNA in tdTomato+ US (Fig. 1j,k). At three months and ological range by treating mice with a low dose of doxycycline, which

2 | Nature | www.nature.com
 a TertCreER/+
b Winner clone Loser clone d f
(Self-renewing) (Differentiating) Seminiferous tubule
Rosalsl-Tdtomato/+

Patch number per testis

150 250

Total patch length (mm)

Tert TERT+
CreER +Tam CreER cells 200
100
Rosa 150
Stop Tdtom Tdtom

Time
100
50
TERT+ 50
cells
0 0
Tam 0.25 0.5 1 2 4 1 Tam 0.25 0.5 1 2 4 1
Lineage trace (mg) Three months One (mg) Three months One
year year

c Bright field Merge

e P = 1.9 × 10–15
tdTomato
10 P = 3.0 × 10–9

Patch length (mm)

8 P = 0.002
6 NS
NS
4
2
0
Tam 0.25 0.5 1 2 4 1
(mg) Three months One year

g h TertCreER/+ TertCreER/flox
TertCreER/flox
Rosalsl-Tdtomato/+ GFRA1 tdTomato i TertCreER/+
Tert CreER CreER 15 TertCreER/flox
+Tam
Tert

GFRA1+ (%)
Rosa

tdTomato+
Stop Tdtom Tdtom 10
m TertCreER/+
5 TertCreER/flox
Tert –/f –/f –/Δ –/f –/f P = 5.9 × 10–5
0 100 P = 7.2 × 10–8

Patch number
Lineage trace As Apr 80

per testis
60
40
20
j l TertCreER/+ TertCreER/flox 0
TertCreER/CreER tdTomato Three Six
TertCreER/+ TertCreER/flox
months months
n
Three months

P = 2.2 × 10–3
2.4
P = 0.016

Patch length
1.8

(mm)
1.2
0.6
0
Three Six
months months
k 100
Tert mRNA foci (%)

80
Foci o
>3 P = 2.7 × 10–7
Total patch length

60 150
P = 2.2 × 10–5
Six months

2
40 100
1
(mm)

20 0 50
0
CreER/+ CreER/ CreER/ 0
Tert flox CreER Three Six
months months

Fig. 1 | Deletion of TERT impairs SSC-mediated clone formation. Arrows indicate tdTomato+GFRA1+ cells. Scale bars, 100 μm. i, Quantification
a, Tert CreER/+Rosa lsl-Tdtomato/+ mice. Tam, tamoxifen. b, Lineage tracing using of tdTomato+ cells among GFRA1+ As and Apr cells (n = 4 mice per group). j, In situ
Tert-CreER. c, Epifluorescence of tdTomato and bright-field image of untangled hybridization against Tert mRNA in purified tdTomato+ US at five days after
seminiferous tubules in a single whole testis at three months after the injection tamoxifen treatment. US from TertCreER/CreER mice were used as a negative control.
of 1 mg tamoxifen. Scale bars, 5 mm. d–f, Mean patch number (d) (n = 6, 6, 6, 4, Scale bars, 40 μm. k, Quantification of foci of Tert mRNA (n = 232, 242, 341 cells,
6, 5 mice from left to right), mean patch length (e) (n = 6, 6, 6, 4, 6, 5 mice from 3 mice per group, from left to right). l, Epifluorescence of tdTomato in untangled
left to right) and total patch length (f) (n = 30, 140, 351, 337, 687, 98 patches seminiferous tubules at three months and six months after tamoxifen injection.
from left to right) after pulse labelling with the indicated dose of tamoxifen for Scale bars, 5 mm. m–o, Quantification of mean patch number (m), mean patch
the indicated time. P values determined by one-way ANOVA with Tukey’s test. length (n) and total patch length (o) (n = 7, 7, 9, 8 mice from left to right). P values
NS, not significant. g, Tert CreER/flox :Rosa lsl-Tdtomato/+ mice. h, Immunofluorescence determined by two-sided unpaired t-test. Data are mean ± s.e.m.
of GFRA1 and tdTomato (day 2) using whole-mount seminiferous tubules.

suppresses the binding of tTA to the TetO promoter (Extended Data Somatic deletion of Tert in mice with very long telomeres is unlikely
Fig. 4b). Transgenic expression of wild-type Tert, but not Tertci, restored to cause telomere dysfunction because achieving sufficient telomere
telomerase activity (Extended Data Fig. 4c). Notably, expression of shortening to cause telomere dysfunction requires generations of
either Tert or Tertci transgenes fully rescued patch number, average interbreeding for more than one year16,31,32. Critical telomere short-
patch length and total patch length (Fig. 2g–j). Furthermore, expression ening in mouse and human cells triggers a DNA damage response
of TERTci significantly increased total patch length, compared with con- and the activation of the p53 tumour suppressor protein, induc-
trol TertCreER/+ mice (Fig. 2j). These results indicate that TERT promotes ing either cell death or senescence33. To understand whether the
enhanced stem cell competition independent of its catalytic activity. impaired clone formation in TERT-deleted SSCs is mediated by p53,

Nature | www.nature.com | 3
Article
a b Terc+/–:TertCreER/+ Terc+/–:TertCreER/flox
c P = 0.0011
TertCreER/flox tdTomato P = 0.004

Patch number per testis

Rosalsl-Tdtomato/+ 120 P = 0.0075 P = 5.1 × 10–4
Terc–/–
80
Tert
CreER
+Tam CreER
Tert 40
Rosa
Stop Tdtom Tdtom e
0 P = 1.2 × 10–4
Terc KO KO
Tert CreER CreER CreER CreER P = 9.8 × 10–4

Total patch length (mm)

KO KO 150
/+ /flox /+ /flox
P = 7.0 × 10–4 P = 1.6 × 10–4
Terc+/– G1 Terc–/–
100
Tert –/f –/f –/Δ –/f –/f G1 Terc–/–:TertCreER/flox
Terc –/– –/– –/– –/– –/–
G1 Terc–/–:TertCreER/+ d P = 0.0035
50
P = 0.018
1.8
P = 0.0049 P = 0.013

Patch length (mm)

0
Lineage trace Tert CreER CreER CreER CreER
1.2
/+ /flox /+ /flox
Terc+/– G1 Terc–/–
0.6

0
Tert CreER CreER CreER CreER
/+ /flox /+ /flox
Terc+/– G1 Terc–/–

f g TertCreER/+ TertCreER/flox h P = 9.7 × 10–6

P = 7.7 × 10–5
TertCreER/flox tdTomato
120 P = 0.0022
Rosalsl-Tdtomato/lsl-tTA

Patch number
TetO-Tert or -Tertci
80

per testis
Tert
CreER +Tam CreER j P = 0.048
Tert 40
Rosa P = 5.8 × 10–6
Stop Tdtom Tdtom
0
Stop tTA tTA P = 4.6 × 10–5
Tert CreER CreER CreER CreER

Total patch length (mm)

Tg tTA /+ /flox /flox /flox 150 P = 0.0029
Tert Tert
TetO or or + Tert + Tertci
prom. tTA
Tertci Tertci 100
TertCreER/flox + Tert TertCreER/flox + Tertci i P = 0.002
50
P = 0.0022
Tert –/f –/f –/Δ –/f –/f 1.8

Patch length (mm)

P = 0.037 0
Tg – – Tert or – – Tert CreER CreER CreER CreER
Tertci
1.2 /+ /flox /flox /flox
+ Tert + Tertci
Lineage trace 0.6

0
Tert CreER CreER CreER CreER
/+ /flox /flox /flox
+ Tert + Tertci

P = 0.002
P = 0.021
k P = 0.0034 P = 0.0034 l m 40 P = 0.026
20 20
P = 0.0028 P = 8.6 × 10–4 P = 2.5 × 10–4 BrdU+tdTomato+
PLZF+ US (%) 30
GFRA1+ Apr (%)

30
GFRA1+ As (%)

TertCreER/+
tdTomato+ clones

15 P = 0.0067 15 P = 0.0031
PLZF+ cells (%)
tdTomato+

tdTomato+

TertCreER/flox 20
containing

20
10 10 TertCreER/flox + Tert
TertCreER/flox + Tertci 10 10
5 5
0
0
0 0
Tert CreER CreER Tert CreER CreER CreER CreER
Day 7 Day 14 Day 7 Day 14 /+ /flox /flox /flox
/+ /flox
+ Tert + Tertci

Fig. 2 | TERT promotes SSC competition independent of the telomerase determined by one-way ANOVA with Tukey’s test. k, Quantification of
complex and TERT catalytic activity. a, Tert CreER/flox :Rosalsl-Tdtomato/+:Terc−/− mice. tdTomato+GFRA1+ As and Apr cells at 7 or 14 days after labelling (n = 5, 5, 5, 5, 6, 6,
b, Epifluorescence for tdTomato in untangled seminiferous tubules at three 5, 5 mice from left to right). P values determined by one-way ANOVA with Tukey’s
months after injection. Scale bars, 5 mm. c–e, Quantification of mean patch test. l, Quantification of the percentage of tdTomato+ clones containing PLZF+
number (c), mean patch length (d) and total patch length (e) (n = 7, 6, 8, 7 mice cells at six weeks after tamoxifen injection (n = 5, 4 mice from left to right).
from left to right). P values determined by one-way ANOVA with Tukey’s test. P values determined by two-sided unpaired t-test. m, Quantification of BrdU+
f, TertCreER/flox:Rosalsl-Tdtomato/lsl-tTA:TetO-Tert or -Tertci mice. g, Epifluorescence cells in tdTomato+PLZF+ US at seven days after tamoxifen injection (n = 6, 5, 5, 6
for tdTomato in untangled seminiferous tubules at three months after tamoxifen mice from left to right). P values determined by one-way ANOVA with Tukey’s
injection. Scale bars, 5 mm. h–j, Quantification of mean patch number (h), test. Data are mean ± s.e.m.
mean patch length (i) and total patch length ( j) (n = 6 mice per group). P values

we generated Trp53flox/flox:TertCreER/flox:Rosalsl-Tdtomato/+ mice and exam- the canonical telomerase complex, catalytic activity and the DNA
ined competitive clone formation (Extended Data Fig. 4d). At day damage response.
five, the homozygous deletion efficiency of the Trp53-flox alleles in
tdTomato+ US was 44.4% (Extended Data Fig. 4e,f). We found that
deleting p53 did not rescue the impaired clone formation associated Differentiation of US cells lacking TERT
with Tert inactivation, as measured by patch number, patch length The preferential elimination of conditionally TERT-deleted SSCs could
and total patch length (Extended Data Fig. 4g–j). Consistent with be caused by a promotion of differentiation, impaired proliferation
these findings, we found no accumulation of γH2AX—a marker of or increased apoptosis. To distinguish among these possibilities, we
DNA damage—in TertCreER/flox mice, whereas γH2AX was increased in investigated how clonal deletion of TERT influences SSC fates across
G6 TertTdtomato/Tdtomato mice with dysfunctional telomeres (Extended time points. At 14 days after tamoxifen administration, whole-mount
Data Fig. 4k,l). Together, these results show that the effect of TERT immunofluorescence showed that tdTomato+ As and Apr US were signifi-
in promoting stem-cell-derived clone formation is independent of cantly decreased in TertCreER/flox mice and were rescued by TERT or TERTci

4 | Nature | www.nature.com
(Fig. 2k and Extended Data Fig. 5a). Flow cytometry analysis revealed DS (Fig. 3a,b,d,e and Extended Data Fig. 7e). By contrast, peak number
that tdTomato+ US-h and US-m cells were significantly decreased, was unaffected by TERT deletion in DS (Fig. 3a,b,d,e and Extended Data
whereas the KIT+ DS population was reciprocally increased, and the Fig. 7e). Pathway analysis revealed that genes in the MAPK signalling
alterations in both the US and the DS populations were rescued by pathway, which promotes self-renewal of SSCs35, were particularly
transgenic Tert or Tertci (Extended Data Fig. 5b,c). At six weeks, when enriched among those showing a loss of open chromatin peaks in the
a comparable frequency of labelled patches was found in TertCreER/+ TERT-deleted US (Extended Data Fig. 7f). The reduction in open chro-
and TertCreER/flox mice, labelled clones contained significantly fewer matin peaks was also evident in genes associated with stemness in SSCs,
PLZF+ US cells (Fig. 2l and Extended Data Fig. 5d–f). These results including Ret, Gfra1, Cdh1 and Zbtb16, whereas those associated with
indicate that clonal TERT deletion promotes the differentiation of US differentiation, including Kit, Prm2 and Prm3, remained unchanged
cells to DS committed progenitor cells. To understand how the loss (Extended Data Fig. 7h–j). Immunostaining revealed similar levels of
of TERT affects cell proliferation, we measured BrdU incorporation GFRA1 and PLZF expression between sorted US-h, US-m and DS from
seven days after tamoxifen administration. The percentage of US cells TertCreER/+ and TertCreER/flox mice, suggesting that the alteration of global
incorporating BrdU was significantly reduced in TertCreER/flox mice, and chromatin accessibility is not due to changes in the composition of
proliferation was rescued with transgenic expression of Tert or Tertci sorted cells (Extended Data Fig. 7k,l). Together, these data show that
(Fig. 2m and Extended Data Fig. 6a). The reduction of BrdU incorpo- clonal inactivation of TERT causes a loss of open chromatin selectively
ration was also observed in GFRA1+ short-chain US (Extended Data in the stem-cell-containing population, but not in committed pro-
Fig. 6b,c). There was no increase in apoptosis by cleaved-PARP staining genitors, consistent with a model of enhanced stem cell differentiation
in TertCreER/flox mice, but apoptosis was increased in testes from control caused by TERT deletion.
G6 TertTdtomato/Tdtomato mice with critically short telomeres (Extended Data
Fig. 6d,e). Together, these data show that the impaired clone forma-
tion that is seen in TERT-deleted SSCs occurs because the loss of TERT Rescue of clone formation with MYC
promotes differentiation and decreased proliferation. To understand how TERT promotes competitive clone formation in
SSCs, we examined gene expression in TERT-deleted US-h cells by RNA
sequencing (RNA-seq) seven days after tamoxifen administration.
Reduced chromatin accessibility in US cells PCA showed that TERT-deleted US-h clustered separately from TERT+
To understand how TERT deletion affects global chromatin struc- controls (Fig. 3f). Using strict cut-offs for significance, there were 23
ture, we performed the assay for transposase-accessible chromatin genes downregulated and 116 genes upregulated in TERT-deleted US-h
with sequencing (ATAC-seq), which allows chromatin accessibility (Extended Data Fig. 8a and Supplementary Table 1). Gene set enrich-
genome-wide to be assessed34. To first define the patterns of chromatin ment analysis (GSEA) revealed that spermatogenesis-related genes were
changes during normal spermatogenesis, we purified US-h, US-m and upregulated in TERT-deleted US-h, consistent with enhanced differen-
DS from TertCreER/+ mice that had been injected with tamoxifen seven tiation (Extended Data Fig. 8b). Several gene sets were downregulated
days before isolation, and spermatocytes (SP) and round spermatids in TERT-deleted US-h cells, including E2F targets and G2M checkpoints,
(RS) were purified on the basis of differential Tert promoter activity reflecting the quantitative reduction in proliferation in these cells
in TertTdtomato/+ mice16. Principal component analysis (PCA) revealed (Extended Data Fig. 8c). The most significantly downregulated gene set
that US-h and US-m clustered together in the bottom left quadrant, was ‘MYC targets v.1’, and a second gene set, ‘MYC targets v.2’, was also
consistent with their similar patterns of gene expression26 (Extended represented (Fig. 3g and Extended Data Fig. 8c). These changes were
Data Fig. 7a). DS cells localized in the top left quadrant, whereas SP not caused by alterations in the composition of sorted cells, because
and RS populations were clustered together in the bottom right quad- the expression levels of stem cell and differentiation marker genes
rant (Extended Data Fig. 7a), indicating that the PC1 axis captures the remained unchanged (Extended Data Fig. 8d). Consistent with the
changes in global chromatin state associated with differentiation. GSEA, MYC protein levels were significantly decreased in TERT-deleted
Similarly, Pearson correlation hierarchical clustering showed a high US by immunofluorescence analysis, and MYC levels were restored by
correlation in open chromatin patterns among spermatogonia sub- transgenic expression of TERT or TERTci (Fig. 3h and Extended Data
populations, but abrupt changes of chromatin accessibility globally Fig. 9a,b). Myc mRNA levels remained unchanged, which suggests
after entry to meiosis (Extended Data Fig. 7b). The number of unique that TERT promotes MYC expression at the post-transcriptional level
ATAC-seq peaks and promoter chromatin accessibility surrounding (Extended Data Fig. 9c). Notably, MYC protein was also reduced in US
transcription start sites (TSSs) were highest in US-h and US-m and cells from germline Tert-knockout mice (TertCreER/CreER G1 mice) (Extended
decreased significantly during differentiation into DS and SP (Fig. 3a,b Data Fig. 9d,e), indicating that a reduction in MYC levels is linked to
and Extended Data Fig. 7c,d). The promoter region of Tert was acces- TERT deletion, and is not a feature of impaired cell competition.
sible in US-h and US-m cells, less accessible in DS cells and inaccessible MYC is a transcription factor that promotes cell competitiveness
in SP and RS, consistent with the expression pattern of TERT during by regulating cell proliferation, growth and metabolism20,36–38. MYC
spermatogenesis16 (Extended Data Fig. 7g). Together, these data reveal has been shown to promote SSC self-renewal and is also an impor-
that the population of US cells exhibits a markedly increased pattern tant oncogene39–42. Given the reduction in MYC seen with clonal TERT
of chromatin accessibility, and that a global reduction in chromatin inactivation, we hypothesized that overexpressing MYC might rescue
accessibility occurs during lineage differentiation. the failure of clone formation in TERT-deleted SSCs. To test this idea,
To understand how clonal TERT deletion influences chromatin acces- we intercrossed TertCreER/flox:Rosalsl-Tdtomato/lsl-tTA mice with TetO-human
sibility, we performed ATAC-seq on US-h, US-m and DS isolated from MYC transgenic mice (TertCreER/flox + MYC) (Fig. 3i). This system allows
TertCreER/flox mice and controls. The overall pattern of chromatin accessi- simultaneous deletion of the residual Tert allele and activation of trans-
bility in TERT-deleted cells remained similar to that of TertCreER/+ controls, genic MYC selectively in a lineage of TERT-expressing stem cells. To
on the basis of Pearson correlation hierarchical clustering (Extended limit the expression levels of transgenic MYC, mice were treated with
Data Fig. 7b). However, PCA revealed that TERT deletion caused a shift doxycycline, which reduced transgenic MYC mRNA levels by 5.6-fold
in the US-h and US-m along the differentiation axis, whereas the loss (Extended Data Fig. 9f). Three months after tamoxifen treatment, the
of TERT had no discernible effect on committed DS (Fig. 3c). Deletion defects in clone formation associated with TERT loss were consid-
of TERT in the purified populations of US caused a marked reduction erably rescued by MYC expression, as measured by patch number,
in the number of open chromatin peaks, and diminished chromatin patch length and total patch length (Fig. 3j–m). MYC overexpres-
accessibility surrounding TSSs to a level resembling that of the control sion did not impair spermatogenesis (Extended Data Fig. 9g). MYC

Nature | www.nature.com | 5
Article
a b c

Read count per million mapped reads

Tert Tert
2.0 US-h: CreER/+ 150 CreER CreER
Distal intergenic US-h: CreER/flox /+ /flox
7 US-m: CreER/+ US-h
Number of peaks (×104)

PC2: 14.2% variance

Gene body 100
6 US-m: CreER/flox US-m
Promoter (<3 kb) 1.5 DS: CreER/+ DS
5 DS: CreER/flox 50
4 SP: Tdtom/+
1.0 RS: Tdtom/+
3
0
2
1 0.5 –50
0
Tert CreER CreER CreER CreER CreER CreER Tdtom Tdtom
/+ /flox /+ /flox /+ /flox /+ /+ –100
US-h US-m DS SP RS –1,000 –500 TSS 500 1,000 –100 –80 –60 –40
Genomic region (5′–3′) PC1: 40.7% variance

d US-h US-m DS SP RS e f TertCreER/+ TertCreER/flox

Tert CreER CreER CreER CreER CreER CreER Tdtom TertCreER/+

PC2: 27% variance

/+ /flox /+ /flox /+ /flox /+ TertCreER/flox 5
US-h 1,675
0 h P = 8.6 × 10–10
30,989 P = 1.1 × 10–7

MYC signal intensity (a.u.)

–5 200
P = 5.8 × 10–8
34,228
–5 0 5 10 150
US-m 3,893 PC1: 60% variance
100
33,770 g MYC targets v.1
0.6 50
19,698
0.4
6,487 0.2 0
DS
24,387
0 Tert CreER CreER CreER CreER
6,288 /+ /flox /flox /flox
–3 0 3 + Tert + Tertci
Row z-score nES = 3.22, FWER P = 0

i TertCreER/flox
Rosalsl-Tdtomato/lsl-tTA
TetO-hMYC j TertCreER/+ TertCreER/flox TertCreER/flox + MYC
Tert
CreER +Tam CreER
Tert tdTomato
Rosa
Stop Tdtom Tdtom
Stop tTA tTA
TetO
Tg prom.
MYC tTA MYC

Tert –/f –/f -/Δ –/f –/f

Tg – – MYC – –

Lineage trace

k P = 0.017 l P = 3.3 × 10–5 m P = 0.005 n P = 0.0033

o P = 0.0028
Average patch length (mm)

As and Apr clones (%)

20
Total patch length (mm)

40
tdTomato+GFRA1+
Patch number per testis

P = 1.2 × 10–5 1.6 160 P = 5.7 × 10–6 P = 0.03

120 P = 0.0075 P = 0.039
tdTomato+BrdU+

15
PLZF+ US (%)

30
P = 0.01 1.2 120 P = 0.017
80
0.8 10 20
80
40 0.4 5 10
40
0 0 0 0 0
CreER CreER CreER CreER CreER CreER CreER CreER CreER CreER CreER CreER CreER CreER CreER
Tert Tert Tert Tert Tert
/+ /flox /flox /+ /flox /flox /+ /flox /flox /+ /flox /flox /+ /flox /flox
+ MYC + MYC + MYC + MYC + MYC

Fig. 3 | TERT deletion compromises chromatin accessibility and MYC testicular cross-sections at seven days after labelling (n = 77 cells from 4 mice,
pathway in SSCs. a, Peak calls from ATAC-seq data. Peak calls from each cell type 76 cells from 6 mice, 77 cells from 4 mice, 81 cells from 4 mice, left to right).
are shown individually. Colour indicates the type of genomic region overlapped P values determined by one-way ANOVA with Tukey’s test. a.u., arbitrary units.
by the peak. b, Average tag density of ATAC-seq reads around TSSs. c, PCA of i, Tert CreER/flox :Rosa lsl-Tdtomato/lsl-tTA :TetO-hMYC mice. j, Epifluorescence for
ATAC-seq data from US-h, US-m and DS cells purified from TertCreER/+ versus tdTomato in untangled seminiferous tubules at three months after labelling.
TertCreER/flox mice. d, Heat map representation of 11,656 peaks that are significantly Scale bars, 5 mm. k–m, Quantification of mean patch number (k), mean
different between Tert CreER/+ and Tert CreER/flox in US-h, US-m and DS or significantly patch length (l) and total patch length (m) (n = 6, 8, 6 mice from left to right).
more open either in SP or RS. Each row represents one ATAC-seq peak. Colour P values determined by one-way ANOVA with Tukey’s test. n, Quantification
represents the relative ATAC-seq accessibility. e, Venn diagrams of the peaks in of tdTomato +GFRA1+ A s and Apr clones at 14 days after labelling, detected
US-h, US-m and DS cells from Tert CreER/+ and Tert CreER/flox mice. f, PCA of RNA-seq by whole-mount immunofluorescence (n = 5 mice per group). P values
data from tdTomato+ US-h. g, Significant downregulation of MYC target genes determined by one-way ANOVA with Tukey’s test. o, Quantitative analysis of
in Tert CreER/flox US-h cells. RNA-seq data were analysed using GSEA. FWER, tdTomato +BrdU+PLZF+ US using testicular cross-sections at seven days after
family-wise error rate; nES, normalized enrichment score. h, Quantification labelling (n = 5 mice per group). P values determined by one-way ANOVA with
of mean signal intensity for MYC staining in tdTomato +PLZF+ US using Tukey’s test. Data are mean ± s.e.m.

expression also restored the number of GFRA1+ cells and proliferation transgenes in a TERT-competent TertCreER/+ context. At three months
of TERT-deleted US (Fig. 3n,o). To examine the effects of overexpres- after labelling, patch number, patch length and total patch number were
sion of TERT, TERTci and MYC in SSC competition, we induced these increased (Extended Data Fig. 9h–k), indicating that overexpression of

6 | Nature | www.nature.com
a b P = 3.7 × 10–4 the non-canonical effects of TERT in enhancing stem cell competition
P = 0.0032 can serve to cull cells that have lost TERT and that suffer only modest
100

/centromere sum intensity

10 P = 6.3 × 10–5
telomere shortening (Fig. 4c). These observations provide a new mecha-

Telomere sum intensity

spermatocytes (%)

80 8
nism for favouring cells with long telomeres to ensure that telomeres
tdTomato+

60 6
are maintained in sperm and in the early embryo.
40 4
20 2
0 0 Discussion
Tert CreER CreER

o–

o+

o–

o+
Telomerase serves essential functions in tissue stem cells in maintain-

at

at
at

at
/+ /flox

m
m

m
ing telomeres, although the effects of telomerase loss become evident

To

To
To

To
td

td
td

td
TertCreER/+ TertCreER/flox after an extended lag phase during which telomeres progressively
shorten12,13,15. By tracing the fate of individual SSCs in vivo, we found
c that inactivating TERT in SSCs promotes rapid stem cell differentia-
SSC pool TERT-proficient TERT-deleted
tion, global loss of chromatin accessibility, reduced MYC expression
Cell count

(winner) SSCs SSCs

TERT and impaired clone formation, as these TERT-deleted stem cells are
outcompeted by unlabelled SSCs. This non-canonical effect of TERT
Telomere length MYC MYC MYC
MYC MYC MYC in facilitating stem cell competition is independent of catalytic func-
Competition

Loser Self-renewal ↑ Self-renewal ↓ tion and occurs without a notable lag phase—a fundamental feature of
senescence and crisis enforced by critically short telomeres. Our results
•Differentiation ↓ •Differentiation ↑ lead us to propose a model in which the elimination of clones derived
•Proliferation ↑ •Proliferation ↓ from TERT-deleted SSCs represents a means of culling cells with only
•Chromatin •Chromatin
accessibility ↑ accessibility ↓ modest telomere shortening from the population of sperm, and can
•Telomere ↑ •Telomere ↓
therefore serve to maintain telomere lengths during life in the testes, as
lineages deriving from SSCs with the highest TERT levels outcompete
Fig. 4 | TERT-dependent cell competition eliminates SSCs with shorter
those deriving from SSCs with lower levels of TERT (Fig. 4c). Although
telomeres. a, Quantification of tdTomato+ spermatocytes at one year after
the checkpoint responses to telomere dysfunction effectively elimi-
the administration of 5 mg tamoxifen (n = 4 mice per group). Data are
mean ± s.e.m. b, Quantification of telomere signals in spermatocytes at
nate cells with very short telomeres, the non-canonical mechanisms
one year after labelling (n = 168, 179, 155, 171 cells from 3 mice per group from identified here provide a means of removing cells with only modest
left to right). Total telomere signals per cell (telomere sum signals) were telomere shortening, long before telomere dysfunction ensues. Our
normalized with total centromere signals per cell (centromere sum signals). findings establish TERT as a key determinant regulating competition
P values determined by one-way ANOVA with Tukey’s test. Data are mean between tissue stem cells by favouring self-renewal and disfavouring
values. c, Summary schematic of the functions of TERT in SSC competition. differentiation. This result is noteworthy in that germline inactivation
TERT-deleted SSCs are progressively eliminated from the SSC pool through of Tert in mice is well tolerated while telomeres are sufficiently long.
cell competition, reducing the contribution of TERT-deleted SSCs that have The apparent absence of stem cell defects in telomerase-knockout
shorter telomeres to spermatogenesis over time (left). Green and red indicate mice with long telomeres might occur because of genetic buffering or
TERT+ and TERT-deleted cells, respectively. In wild-type SSCs, TERT promotes compensation, or because of a lack of competition in a tissue in which
competitive clone formation by upregulating MYC protein. Deletion of TERT in
all cells lack TERT. In this context, germline gene inactivation identifies
SSCs downregulates MYC, promotes differentiation, compromises proliferation,
only one layer of activity, whereas clonal, conditional gene deletion
induces a global loss of chromatin accessibility and accelerates telomere
can reveal deeper aspects of gene function.
shortening.
Although TERT levels are important in the clonal expansion of stem
cells, they might have a similar role during pre-neoplastic processes
these transgenes is sufficient to enhance SSC competition. Although and during carcinogenesis. Germline TERT polymorphisms confer an
we did not find defects in spermatogenesis, overexpression of these increased risk of clonal haematopoiesis and an increased susceptibility
transgenes might result in non-physiological effects. Together, these to diverse cancer types43,44. It is notable that somatic TERT promoter
results indicate that TERT promotes stem cell competition through mutations that activate TERT transcription show strong positive selec-
MYC, and establish an epistatic relationship between TERT and MYC. tion during tumour progression, as the prevalence of these mutations
increases substantially from pre-invasive to invasive stages. Our data
showing that TERT-deleted SSCs exhibit reduced levels of MYC provide
Mild telomere shortening with TERT loss support for this model. MYC activity is central in determining outcomes
The non-canonical effects of TERT in promoting stem cell competition in cell competition: cells expressing higher levels of MYC outcompete
by disfavouring clones that have lost TERT could affect tissue telomere those with lower MYC levels, and MYC represents a key node in many
lengths. To investigate this possibility, we measured relative telomere human cancers20,36–38. Thus, upregulation of TERT might promote the
lengths in spermatocytes in TERTCreER/flox mice and TERTCreER/+ controls. clonal expansion of cancer cells by activating the MYC pathway, in addi-
One year after the injection of tamoxifen at a high dose, tdTomato+ tion to maintaining telomere function. Together, these data establish a
male germ cells were less abundant in TERTCreER/flox mice compared with non-canonical role for TERT in promoting clonal competition in stem
TERTCreER/+ controls by immunofluorescence of single cells and by cells in vivo, with implications for understanding tissue homeostasis,
immunohistochemistry on tissue sections (Fig. 4a and Extended Data cancer development and telomere maintenance.
Fig. 10a). Single-cell suspensions from the testes were analysed by
combined fluorescence in situ hybridization (FISH) for telomeres and
centromeres and by immunofluorescence for tdTomato. In tdTomato+ Online content
spermatocytes from TERTCreER/flox mice, telomere length was reduced Any methods, additional references, Nature Portfolio reporting summa-
by 12.7% compared with tdTomato− spermatocytes, owing to the loss ries, source data, extended data, supplementary information, acknowl-
of TERT (Fig. 4b and Extended Data Fig. 10b). For comparison, tel- edgements, peer review information; details of author contributions
omere lengths were indistinguishable in tdTomato+ and tdTomato− and competing interests; and statements of data and code availability
spermatocytes from TERTCreER/+ controls. These findings show that are available at https://doi.org/10.1038/s41586-024-07700-w.

Nature | www.nature.com | 7
Article
1. Roake, C. M. & Artandi, S. E. Regulation of human telomerase in homeostasis and disease. 27. Lin, S. et al. Distributed hepatocytes expressing telomerase repopulate the liver in
Nat. Rev. Mol. Cell Biol. 21, 384–397 (2020). homeostasis and injury. Nature 556, 244–248 (2018).
2. Park, J.-I. et al. Telomerase modulates Wnt signalling by association with target gene 28. Neuhöfer, P. et al. Acinar cell clonal expansion in pancreas homeostasis and carcinogenesis.
chromatin. Nature 460, 66–72 (2009). Nature 597, 715–719 (2021).
3. Shkreli, M. et al. Reversible cell-cycle entry in adult kidney podocytes through regulated 29. Klein, A. M., Nakagawa, T., Ichikawa, R., Yoshida, S. & Simons, B. D. Mouse germ line stem
control of telomerase and Wnt signaling. Nat. Med. 18, 111–119 (2011). cells undergo rapid and stochastic turnover. Cell Stem Cell 7, 214–224 (2010).
4. Sarin, K. Y. et al. Conditional telomerase induction causes proliferation of hair follicle 30. Venteicher, A. S. et al. A human telomerase holoenzyme protein required for Cajal body
stem cells. Nature 436, 1048–1052 (2005). localization and telomere synthesis. Science 323, 644–648 (2009).
5. Koh, C. M. et al. Telomerase regulates MYC-driven oncogenesis independent of its reverse 31. Yuan, X. et al. Presence of telomeric G-strand tails in the telomerase catalytic subunit
transcriptase activity. J. Clin. Invest. 125, 2109–2122 (2015). TERT knockout mice. Genes Cells 4, 563–572 (1999).
6. Khattar, E. et al. Telomerase reverse transcriptase promotes cancer cell proliferation by 32. Liu, Y. et al. The telomerase reverse transcriptase is limiting and necessary for telomerase
augmenting tRNA expression. J. Clin. Invest. 126, 4045–4060 (2016). function in vivo. Curr. Biol. 10, 1459–1462 (2000).
7. Ghosh, A. et al. Telomerase directly regulates NF-κB-dependent transcription. Nat. Cell 33. Chin, L. et al. p53 deficiency rescues the adverse effects of telomere loss and cooperates
Biol. 14, 1270–1281 (2012). with telomere dysfunction to accelerate carcinogenesis. Cell 97, 527–538 (1999).
8. Choi, J. et al. TERT promotes epithelial proliferation through transcriptional control of a 34. Buenrostro, J. D., Giresi, P. G., Zaba, L. C., Chang, H. Y. & Greenleaf, W. J. Transposition of
Myc- and Wnt-related developmental program. PLoS Genet. 4, e10 (2008). native chromatin for fast and sensitive epigenomic profiling of open chromatin, DNA-
9. Horn, S. et al. TERT promoter mutations in familial and sporadic melanoma. Science 339, binding proteins and nucleosome position. Nat. Methods 10, 1213–1218 (2013).
959–961 (2013). 35. Hasegawa, K., Namekawa, S. H. & Saga, Y. MEK/ERK signaling directly and indirectly
10. Huang, F. W. et al. Highly recurrent TERT promoter mutations in human melanoma. contributes to the cyclical self-renewal of spermatogonial stem cells. Stem Cells 31,
Science 339, 957–959 (2013). 2517–2527 (2013).
11. Killela, P. J. et al. TERT promoter mutations occur frequently in gliomas and a subset of 36. Diaz-Diaz, C. et al. Pluripotency Surveillance by Myc-driven competitive elimination of
tumors derived from cells with low rates of self-renewal. Proc. Natl Acad. Sci. USA 110, differentiating cells. Dev. Cell 42, 585–599 (2017).
6021–6026 (2013). 37. Johnston, L. A. Socializing with MYC: cell competition in development and as a model for
12. Lundblad, V. & Szostak, J. W. A mutant with a defect in telomere elongation leads to premalignant cancer. Cold Spring Harb. Perspect. Med. 4, a014274 (2014).
senescence in yeast. Cell 57, 633–643 (1989). 38. Claveria, C., Giovinazzo, G., Sierra, R. & Torres, M. Myc-driven endogenous cell competition
13. Bodnar, A. G. et al. Extension of life-span by introduction of telomerase into normal human in the early mammalian embryo. Nature 500, 39–44 (2013).
cells. Science 279, 349–352 (1998). 39. Kanatsu-Shinohara, M. et al. Myc/Mycn-mediated glycolysis enhances mouse
14. Wong, K. K. et al. Telomere dysfunction impairs DNA repair and enhances sensitivity to spermatogonial stem cell self-renewal. Genes Dev. 30, 2637–2648 (2016).
ionizing radiation. Nat. Genet. 26, 85–88 (2000). 40. Gabay, M., Li, Y. & Felsher, D. W. MYC activation is a hallmark of cancer initiation and
15. Lee, H. W. et al. Essential role of mouse telomerase in highly proliferative organs. Nature maintenance. Cold Spring Harb. Perspect. Med. 4, a014241 (2014).
392, 569–574 (1998). 41. Kanatsu-Shinohara, M., Onoyama, I., Nakayama, K. I. & Shinohara, T. Skp1-Cullin-F-box
16. Pech, M. F. et al. High telomerase is a hallmark of undifferentiated spermatogonia and (SCF)-type ubiquitin ligase FBXW7 negatively regulates spermatogonial stem cell
is required for maintenance of male germline stem cells. Genes Dev. 29, 2420–2434 self-renewal. Proc. Natl Acad. Sci. USA 111, 8826–8831 (2014).
(2015). 42. Wang, J., Hannon, G. J. & Beach, D. H. Risky immortalization by telomerase. Nature 405,
17. Baker, N. E. Emerging mechanisms of cell competition. Nat. Rev. Genet. 21, 683–697 755–756 (2000).
(2020). 43. Kubota, Y. & Viny, A. Germline predisposition for clonal hematopoiesis. Semin. Hematol.
18. Krieger, T. & Simons, B. D. Dynamic stem cell heterogeneity. Development 142, 1396–1406 61, 61–67 (2024).
(2015). 44. Baird, D. M. Variation at the TERT locus and predisposition for cancer. Expert Rev. Mol.
19. Snippert, H. J. et al. Intestinal crypt homeostasis results from neutral competition Med. 12, e16 (2010).
between symmetrically dividing Lgr5 stem cells. Cell 143, 134–144 (2010).
20. Ellis, S. J. et al. Distinct modes of cell competition shape mammalian tissue Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
morphogenesis. Nature 569, 497–502 (2019). published maps and institutional affiliations.
21. Liu, N. et al. Stem cell competition orchestrates skin homeostasis and ageing. Nature
568, 344–350 (2019). Open Access This article is licensed under a Creative Commons Attribution
22. Vishwakarma, M. & Piddini, E. Outcompeting cancer. Nat. Rev. Cancer 20, 187–198 4.0 International License, which permits use, sharing, adaptation, distribution
(2020). and reproduction in any medium or format, as long as you give appropriate
23. Artandi, S. E. & DePinho, R. A. Telomeres and telomerase in cancer. Carcinogenesis 31, credit to the original author(s) and the source, provide a link to the Creative Commons licence,
9–18 (2010). and indicate if changes were made. The images or other third party material in this article are
24. Hjelmborg, J. B. et al. Paternal age and telomere length in twins: the germ stem cell included in the article’s Creative Commons licence, unless indicated otherwise in a credit line
selection paradigm. Aging Cell 14, 701–703 (2015). to the material. If material is not included in the article’s Creative Commons licence and your
25. de Rooij, D. G. The nature and dynamics of spermatogonial stem cells. Development 144, intended use is not permitted by statutory regulation or exceeds the permitted use, you will
3022–3030 (2017). need to obtain permission directly from the copyright holder. To view a copy of this licence,
26. Garbuzov, A. et al. Purification of GFRα1+ and GFRα1− spermatogonial stem cells reveals visit http://creativecommons.org/licenses/by/4.0/.
a niche-dependent mechanism for fate determination. Stem Cell Rep. 10, 553–567
(2018). © The Author(s) 2024

8 | Nature | www.nature.com
Methods Syncytium of US (As–A16) were visually judged on the basis of continu-
ous E-cadherin, GFRA1 or tdTomato staining. The following antibod-
Animals ies were used: anti-RFP (Abcam, ab124754, rabbit polyclonal, 1:500
All animal experiments were approved by the Administrative Panel dilution); anti-RFP (MBL, M208-3, mouse monoclonal 1G9 and 3G5,
on Laboratory Animal Care (APLAC) in protocol APLAC-12684, and 1:200 dilution); anti-E-cadherin (R&D systems, AF748, goat polyclonal,
all experiments were in compliance with the ethical regulations of 1:200 dilution); anti-GFRA1 (R&D systems, AF560, goat polyclonal,
Stanford University. Mice were housed at an ambient temperature of 1:200 dilution); and anti-KIT (Cell Signaling Technology, 3074, rabbit
22 °C and at 40% humidity, with a 12-h light–dark cycle (07:00–19:00). monoclonal D13A2, 1:200 dilution).
Tert-CreER, TetO-Tert, TetO-Tertci and Tert-Tdtomato mice were previ-
ously reported3,16,27. Rosa-lslTdtomato45, Rosa-lsl-tTA46, TetO-hMYC47, Section immunostaining
Terc-KO48 and Trp53-flox49 mice were purchased from The Jackson Labo- Testes were detunicated, fixed with 4% PFA at 4 °C overnight, incubated
ratory. Tamoxifen (Cayman) was dissolved in corn oil (Sigma-Aldrich) in a gradient of ethanol and xylen, embedded in paraffin and cut into
at 5–20 mg ml−1 by incubating at 50 °C for 30 min with mixing every 5-μm sections. After rehydration, antigen retrieval was performed using
5 min. Two- to four-month-old mice were administered with 0.25–4 mg either citric acid or tris-based antigen retrieval solution (Vector Labora-
tamoxifen per 25 g body weight by oral gavage or intra-peritoneal injec- tories) for 5 min in a pressure cooker. Sections were blocked with 0.5%
tion. Doxycycline (Sigma-Aldrich) was dissolved in drinking water in BSA in PBST and incubated with primary antibody at 4 °C overnight.
light-protected bottles at 1 or 3 μg ml−1 and changed every three or After washing with PBST, sections were incubated with secondary anti-
four days. BrdU (Sigma-Aldrich) was dissolved in PBS at 10 mg ml−1 and bodies at room temperature for 1 h and mounted in Vectashield with
intraperitoneally injected at 1.25 mg per 25 g body weight two hours DAPI. For BrdU detection, slides were treated with 2 M HCl for 20 min,
before euthanasia. blocked with 0.5% BSA in PBST and incubated with rabbit anti-PLZF and
rat anti-BrdU antibodies at 4 °C overnight, and signals were detected
Generation of Tert-flox mice by Alexa 488-conjugated anti-rat IgG and Cy5-conjugated anti-rabbit
A 9-kb fragment of the Tert locus was subcloned and a Lox-Puro-lox cas- IgG antibodies. For co-staining using rabbit anti-RFP antibodies and
sette from the pBS.DAT-LoxStop plasmid (a gift from D. Tuveson) was rabbit anti-PLZF antibodies, sections were antigen retrieved, blocked
inserted at the BsiWI site in the second intron. Another loxP sequence with 0.5% BSA in PBST, incubated with anti-RFP antibody, then with
and NdeI site were inserted at the KasI site in the sixth intron. The target- HRP-conjugated anti-rabbit secondary antibody as described above,
ing vector was linearized and electroporated into J1 mouse ES cells. After and signals were detected with the TSA Plus Cyanine 3 system (Akoya
positive selection with puromycin, correctly targeted ES clones were Biosciences). After signal detection, the antibodies were stripped off by
selected by long-range PCR and Southern blotting, and then injected antigen retrieval, and sections were further stained with other antibod-
into C57BL6 blastocysts to generate the knock-in line. To remove floxed ies. For triple staining using rabbit anti-RFP, rabbit anti-PLZF and rabbit
puro cassette, the knock-in line was crossed with CMV-cre mice50 and anti-MYC antibodies, sections were stained with anti-RFP antibody
puro-negative Tert-floxed mice were selected by PCR and Southern using the TSA Plus Cyanine 3 system, and antibodies were stripped off
blotting using genomic DNA from tail tips. TERTflox/+ mice were born at with antigen retrieval. Then, those sections were stained with anti-MYC
normal Mendelian frequency. Original uncropped images of Southern antibody with the TSA Plus Fluorescein system (Akoya Biosciences),
blots are provided in Supplementary Fig. 1. followed by antigen retrieval to remove antibodies. Finally, the sec-
tions were further stained with anti-PLZF antibody and Cy5-conjugated
Lineage-tracing assay anti-rabbit IgG. Slides were mounted in Vectashield with DAPI. For
After tamoxifen injection, testes were detunicated and seminiferous chromogenic staining for tdTomato, sections were incubated with
tubules were untangled using fine forceps in PBS containing 1 mg ml−1 anti-RFP antibody followed by HRP-conjugated antiboy. Signals were
collagenase IV (Worthington) for 10 min, and placed in cold PBS. Images detected with a DAB substrate kit (Vector Laboratories). Sections were
were captured with a fluorescent dissection microscope and the patch counterstained with haematoxylin, dehydrated with ethanol and xylene
number and length were measured with ImageJ. Total patch length and then mounted in Clearmount (American MasterTech). To quantify
was calculated by multiplying the patch number by the average patch PLZF+ cells, seminiferous tubules in stage VII–VIII were excluded from
length. the analyses to prevent the inclusion of PLZF+ early DS cells. Images
were captured on a fluorescent microscope and processed in Pho-
Sperm count toshop. The signal intensities of MYC and PLZF were quantified with
Cauda epididymides were dissected into small pieces and incubated ImageJ. Immunofluorescence data were captured using Leica Applica-
in potassium simplex optimized medium (KSOM) at 37 °C for one hour tion Suite AF and immunohistochemistry data were captured using
under 5% CO2 to allow sperm to exude. The collected sperm were then Leica LAS 4.2. The following antibodies were used: anti-RFP (Abcam,
fixed with 4% PFA and counted with a haemacytometer. ab124754, rabbit polyclonal, 1:500 dilution for immunofluorescence
without TSA or 1:3,000 dilution for immunofluorescence with TSA);
Whole-mount immunofluorescence of seminiferous tubules anti-BrdU (Bio-Rad, MCA2483, mouse monoclonal Bu201, 1:500 dilu-
Seminiferous tubules were dissociated using fine forceps in PBS con- tion); anti-GFRA1 (R&D systems, AF560, goat polyclonal, 1:200 dilu-
taining 1 mg ml−1 collagenase IV for 10 min, fixed with 4% PFA at 4 °C tion); anti-PLZF (Santa Cruz, sc-22839, rabbit monoclonal H-300, 1:200
for two hours, cleared with 0.1% Igepal CA-630 (Sigma-Aldrich) in dilution for immunofluorescence without TSA or 1:5,000 dilution for
PBST and dehydrated and rehydrated by immersing in a gradient of immunofluorescence with TSA); anti-cleaved PARP (Cell Signaling
methanol diluted with PBST (25%, 50%, 75%, 100%, 75%, 50%, 25%) at Technology, 9548, mouse monoclonal 7C9, 1:500 dilution); anti-γH2AX
4 °C for 5 min each. After washing in PBST, tubules were incubated in (EMD Millipore, 05-636, mouse monoclonal JBW301, 1:2,000 dilution);
blocking buffer (0.5% BSA in PBST), followed by incubation with anti- and anti-MYC (Cell Signaling Technology, 13987, rabbit monoclonal
bodies in Immuno Shot Immunostaining, Mild (Cosmo Bio) at 4 °C for D3N8F, 1:500 dilution).
two days. After extensive washing with PBST, tubules were incubated
with secondary antibodies in blocking buffer at room temperature TRAP assays
for 90 min, washed with PBST and then mounted in Vectashield with A two-step TRAP (telomeric repeat amplification protocol) proce-
DAPI (Vector Laboratories). Images were captured on a Leica SP5 con- dure was performed as previously reported51. Extracted fractions
focal microscope and processed in Photoshop CC or later versions. from whole testis at three weeks or fluorescence-activated cell sorting
Article
(FACS)-sorted US were incubated with telomeric primers for a 30-min KIT—RED (Advanced Cell Diagnostics) and probes against mouse Tert
initial extension step at 30 °C in a PCR machine, followed by 5 min inac- or Trp53 (Advanced Cell Diagnostics) according to the manufacturer’s
tivation at 72 °C. Without purification, 1 μl of the extended reaction instructions.
was PCR amplified (cycles of 30 s at 94 °C, followed by 30 s at 59 °C)
in the presence of 32P end-labelled telomeric primers that had been qRT–PCR
purified using a micro-spin G-25 column (GE Healthcare). PCR reactions For qRT–PCR, cells were directly sorted into Trizol LS (Thermo Fisher
were resolved by 9% polyacrylamide gel electrophoresis at room tem- Scientific) by FACS. RNA was purified using the Direct-zol RNA Micro-
perature, and the gel was exposed to a phosphor imager and scanned prep kit (Zymo Research) and cDNA was synthesized using oligo-dT
by a Typhoon scanner. The scanned image was quantified using the and the SuperScript IV First-Strand Synthesis system (Thermo Fisher
TotalLab Quant software. Representative gel images were presented Scientific). For qRT–PCR of Tert exon 2, Tert exon 6 and mouse Myc,
among at least two repeats. Original uncropped images are provided TaqMan Fast Advanced Master Mix (Thermo Fisher Scientific) was used,
in Supplementary Figs. 1 and 2. along with Universal Probe Library Probes: no. 66 for Tert exon 2, no. 93
for Tert exon 6 and no. 77 for mouse Myc (Roche). For other qRT–PCR,
FACS analysis the PowerUp SYBR Green Master Mix (Thermo Fisher Scientific) was
Testes were detunicated, lightly dissociated in PBS and incubated in used according to the product manual. PCR analysis was done with a
PBS containing 1 mM EDTA, 1 mg ml−1 collagenase I (Worthington) and 7900HT Fast Real-Time PCR System machine (ABI). Primer information
DNase I (Worthington) at 32 °C for 8 min. Cells were centrifuged at is available in Supplementary Table 2.
250g for 5 min and the supernatant was removed. After repeating the
collagenase I treatment, testicular cells were further digested with Try- RNA-seq
pLE Express (Gibco) at 32 °C for 15 min. During enzymatic digestions, US-h cells were directly sorted into Trizol LS by FACS and RNA was
seminiferous tubules were mechanically fragmented with vigorous purified using the Direct-zol RNA Microprep kit. Genomic DNA was
pipetting every 5 min. Cells were sequentially filtered with 70-μm and digested with on-column DNase treatment. RNA quality was checked
40-μm strainers, resuspended in cold FACS buffer (2% FBS and 1 mM by Bioanalyzer 2100 (Agilent). RNA-seq libraries were constructed
EDTA in PBS) and incubated with antibodies on ice for 30 min. After using the SMARTer Stranded Total RNA-seq Kit v2—Pico Input Mam-
washing with PBS, cells were resuspended in cold FACS buffer contain- malian (Clontech), starting from 5 ng total RNA. cDNA was synthesized
ing DAPI, and analysed and sorted with a BD Aria II (BD Biosciences). and amplified according to the manual. After the rRNA removal step,
Flow cytometry data were acquired using BD FACSDiva software v.8.0. cDNA was amplified with 13 cycles of PCR reactions. The quality of
Data were analysed with FlowJo v.9 software. The following antibod- purified cDNA libraries was confirmed by Bioanalyzer 2100. Libraries
ies were used; anti-α6 integrin with Pe/Cy7 (Biolegend, 313622, rat were sequenced on the Illumina NextSeq platform, generating about
monoclonal GoH3, 1:150 dilution); anti-MCAM with APC (Biolegend, 16 million–24 million 75-bp paired-end reads per library. Four biologi-
134712, rat monoclonal ME-9F1, 1:200 dilution); and anti-KIT with BB515 cal replicates per sample were analysed. Raw reads were trimmed by
(BD Biosciences, 564481, mouse monoclonal 2B8, 1:200 dilution). The TrimGalore v.0.4.0 (Babraham Bioinformatics), mapped to mm10 by
gating strategy is provided in Supplementary Fig. 3. TopHat v.2.0.13 and analysed by DESeq2.

Telomere FISH ATAC-seq

For preparing cytospin slides of testicular cells, single-cell suspen- ATAC-seq libraries were made as described previously52 using the
sions were prepared as described above and resuspended in PBS. Omni-ATAC protocol. Adjustments to the protocol were made to
Cells were then fixed with 4% PFA at room temperature for 10 min. reflect two main features of the cell types profiled in this work. First,
After washing, cells were resuspended in PBS and cytospun at 250g the amount of Tn5 transposase added to each reaction was modulated
for 5 min. Slides were stored in 70% ethanol at 4 °C. For telomere to maintain proportionality with the number of cells assayed. For exam-
FISH combined with antibody staining against tdTomato, slides were ple, a normal reaction uses 50,000 cells and 2.5 μl of Tn5 transposase
hydrated in PBS, blocked with PBST containing 0.5% BSA and then in a 50-μl reaction. In the case of rarer SSCs, only 5,000 cells could be
incubated with anti-RFP antibody at 4 °C overnight. After washing obtained so only 0.25 μl of Tn5 transposase was used in a 50-μl reaction.
with PBST, slides were incubated with HRP-conjugated anti-rabbit sec- The difference in volume was adjusted using water. Second, the ploidy
ondary antibody at room temperature for 30 min and washed with of each cell type was taken into account and the amount of Tn5 was
PBST. Signals were detected using the TSA Plus Cyanine 3 system. For adjusted on the basis of ploidy as well. For example, round spermatid
telomere FISH, slides after antibody staining were washed with PBS, cells are haploid, so the transposition of 50,000 cells would require
and dehydrated by immersing in 70%, 90% and 100% ethanol. After 1.25 μl of Tn5 transposase in a 50-μl reaction. Similarly, spermatocytes
drying, slides were incubated in hybridization buffer (70% formamide, are 4N meiotic cells so the amount of Tn5 transposase was increased
10 mM Tris-HCl pH 7.4 and 0.5% blocking reagent (Roche)) containing proportionately and the amount of water in the reaction was reduced.
0.2 μM Alexa 488-conjugated telomere probes (PNA Bio) and 0.2 μM In all cases, regardless of cell number or ploidy, the reaction volume of
Cy5-conjugated centromere probes (PNA Bio) at 80 °C for 10 min and the transposition reaction was kept constant at 50 μl. All ATAC-seq reac-
then at 4 °C overnight. Slides were washed with 70% formamide con- tions were performed using homemade Tn5 transposase and Tagment
taining 10 mM Tris-HCl pH 7.4 for 15 min twice, then with PBS, and were DNA buffer53. Downstream amplification and purification of libraries
then mounted in Vectashield with DAPI. Images were captured on a was performed as described previously34,52,54.
fluorescent microscope and processed in Photoshop. Telomere sum Preprocessing of ATAC-seq data was completed using the PEPATAC
signals and centromere sum signals were determined using Telometer pipeline (https://pepatac.databio.org/). The mm10 genome build
v.3.0.5. Anti-RFP antibody (Abcam, ab124754, rabbit polyclonal, 1:500 (https://github.com/databio/refgenie) was used for alignment. In brief,
dilution) was used. all fastq files were first trimmed to remove the Illumina Nextera adapter
sequence using Skewer with “-f sanger –t 20 –m pe –x” options. FastQC
RNA in situ hybridization (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) was
At five days after labelling, testes were collected and tdTomato+ US used to validate proper trimming and check overall sequence data
were FACS-sorted, and cytospun at 200g for 5 min onto slides. Slides quality. Bowtie2 was then used for pre-alignments to remove reads that
were fixed in 4% PFA for 30 min at room temperature and processed would map to chrM (revised Cambridge Reference Sequence), alpha
for single-molecule RNA FISH using the RNAscope 2.5 HD Reagent satellite repeats, Alu repeats, ribosomal DNA repeats and other repeat
regions with “-k 1 -D 20 -R 3 -N 1 -L 20 -i S,1,0.50 -X 2000 --rg-id” options. study are available in the NCBI Gene Expression Omnibus repository
Bowtie2 was then used to align to the mm10 reference genome using under accession number GSE14659. The resulting fastq files were
“--very-sensitive -X 2000 --rg-id” options. SAMtools was used to sort aligned to the mouse reference genome (mm10).
and isolate uniquely mapped reads using “-f 2 -q 10 -b -@ 20” options.
Picard (http://broadinstitute.github.io/picard/) was used to remove 45. Madisen, L. et al. A robust and high-throughput Cre reporting and characterization
system for the whole mouse brain. Nat. Neurosci. 13, 133–140 (2010).
duplicates. Then the bam files were merged by conditions, and MACS2 46. Li, L. et al. Visualizing the distribution of synapses from individual neurons in the mouse
was used to call peaks with parameter “-q 0.05 --nomodel --shift 0”. The brain. PLoS ONE 5, e11503 (2010).
narrow peaks were then filtered by the ENCODE 7 hg19 blacklist, as well 47. Felsher, D. W. & Bishop, J. M. Reversible tumorigenesis by MYC in hematopoietic lineages.
Mol. Cell 4, 199–207 (1999).
as peaks that extend beyond the ends of chromosomes. Bedtools was 48. Blasco, M. A. et al. Telomere shortening and tumor formation by mouse cells lacking
used to retrieve the reads of the called peaks for each sample with the telomerase RNA. Cell 91, 25–34 (1997).
49. Marino, S., Vooijs, M., van Der Gulden, H., Jonkers, J. & Berns, A. Induction of
multicov module. All of the samples have a similar sequencing depth,
medulloblastomas in p53-null mutant mice by somatic inactivation of Rb in the external
mitochondrial rate and duplication rate. The spermatocyte and the granular layer cells of the cerebellum. Genes Dev. 14, 994–1004 (2000).
round spermatid samples have a similar sequencing depth compared 50. Zinyk, D. L., Mercer, E. H., Harris, E., Anderson, D. J. & Joyner, A. L. Fate mapping of the
mouse midbrain-hindbrain constriction using a site-specific recombination system.
with all other samples, but a slightly higher mitochondrial rate and lower
Curr. Biol. 8, 665–668 (1998).
duplication rate, so have more final reads after initial processing and 51. Chen, L. et al. An activity switch in human telomerase based on RNA conformation and
filtering. To make all the samples comparable for the statistical analysis, shaped by TCAB1. Cell 174, 218–230 (2018).
52. Corces, M. R. et al. An improved ATAC-seq protocol reduces background and enables
we used final reads as the normalization factor. The R package DESeq2
interrogation of frozen tissues. Nat. Methods 14, 959–962 (2017).
was used for statistical analysis to identify significant peaks between 53. Picelli, S. et al. Tn5 transposase and tagmentation procedures for massively scaled
different conditions. The differential peaks were called between US-h sequencing projects. Genome Res. 24, 2033–2040 (2014).
54. Buenrostro, J. D., Wu, B., Chang, H. Y. & Greenleaf, W. J. ATAC-seq: a method for assaying
CreER/+ and US-h CreER/flox samples. Peaks with FDR < 0.01 and a fold
chromatin accessibility genome-wide. Curr. Protoc. Mol. Biol. 109, 21.29.1–21.29.9
change larger than 2 or smaller than −2 were considered significant. The (2015).
R package ChIPseeker was used for peak annotation. The R package
ngsplot was used for visualizing the cumulated peak signal.
Acknowledgements We thank members of the S.E.A. laboratory and H. Sin, J. Sage, R. Nusse,
D. Felsher and M. Fuller for critical comments; and P. Chu for assistance with histology. Confocal
Statistics and reproducibility imaging analysis was performed in the Stanford Cell Sciences Imaging Facility. Cell sorting
No statistical methods were used to predetermine sample sizes. All of and flow cytometry analysis for this project was performed using the Stanford Shared FACS
Facility. Next-generation sequencing for this project was performed using the Stanford
the experiments were replicated more than twice and were statistically
Functional Genomics Facility. This work was supported by NIH grants CA197563 and AG056575
analysed and presented in the paper. When comparing two groups, P (to S.E.A.), GM150538 (to L.C.) and CA209919 (to H.Y.C.). K.H. was supported by a Japan Society
values were determined by two-sided unpaired t-test. When comparing for the Promotion of Science Overseas Research Fellowship.

more than two groups, P values were determined by one-way ANOVA

with Tukey’s test. Values are presented as mean ± s.e.m. The mice were Author contributions Conceptualization: K.H. and S.E.A. Methodology: K.H. and S.E.A. Formal
analysis: K.H., Y.Z., A.G., J.A.B., Y.-H.H. and Y.W. Investigation: K.H., M.R.C., L.C., V.M.G., P.N. and
randomly assigned to each experimental or control group. Statistics P.C. Writing original draft: K.H. and S.E.A. Supervision: S.E.A. and H.Y.C. Funding acquisition:
and plots were generated by ggplot2 in R and GraphPad Prism 8. K.H., S.E.A. and H.Y.C.

Reporting summary Competing interests H.Y.C. is a co-founder of Accent Therapeutics and Boundless Bio, and is
an advisor to 10x Genomics, Arsenal Biosciences and Spring Discovery.
Further information on research design is available in the Nature Port-
folio Reporting Summary linked to this article.
Additional information
Supplementary information The online version contains supplementary material available at
https://doi.org/10.1038/s41586-024-07700-w.
Data availability Correspondence and requests for materials should be addressed to Steven E. Artandi.
Peer review information Nature thanks Robin Hobbs, Miles F. Wilkinson and the other,
All data that support the finding of this study are available in a publicly anonymous, reviewer(s) for their contribution to the peer review of this work.
accessible repository. The source data for the RNA-seq and ATAC-seq Reprints and permissions information is available at http://www.nature.com/reprints.
Article

Extended Data Fig. 1 | TERT expression in spermatogonia subpopulations. undifferentiated spermatogonia and an early stage of DS. KIT expression is
a, Spermatogenesis takes place within seminiferous tubules, which are activated when they differentiate into DS. A6-Integrin is expressed from As
composed of four layers of male germ cells and supporting somatic cells. to DS. c, Flow cytometry analysis of wild-type testicular cells stained with
SSC activity resides in undifferentiated spermatogonia, an immature cell α6-Integrin, MCAM, and KIT. α6-Integrin-high cells were further separated
population that undergo self-renewal/mitosis to produce differentiating into US-h, US-m, and DS based on KIT and MCAM expression. d, Whole-mount
daughter cells. Undifferentiated spermatogonia can be divided into US-h and immunofluorescence of GFRA1 and MCAM in seminiferous tubules. Scale bars,
US-m based on MCAM expression. US-h are mainly composed of GFRA1+ As and 100 μm. e, Quantification of GFRA1+ and MCAM-high cells in seminiferous
Apr cells that show higher stem cell potential than longer chained US. US-m are tubules (n = 262, 146, and 26 chains of interconnected US from 4 mice from left
comprised of Aal cells that are more likely to differentiate into DS than shorter to right). f, qRT–PCR of Tert and spermatogonia markers in FACS-sorted US-h,
chained cells. DS are KIT+ differentiation-commited progenitor cells that further US-m, and DS. Expression was normalized with Actb (n = 3 mice per group).
proliferate and differentiate into meiotic spermatocytes and then to spermatids Data are represented as mean ± s.e.m. g, Flow cytometry analysis of testicular
that undergo nuclear elongation. b, A morphological and gene expression cells from TertTdtomato/+ mouse. α6-Integrin-high cells were gated and further
heterogeneity in spermatogonia subpopulations. MCAM and GFRA1 are separated according to KIT and MCAM expression. Tert-Tdtomato expression
strongly expressed in As and Apr cells while NGN3 is mainly expressed in Aal was compared in US-h, US-m and DS.
cells. E-Cadherin and PLZF are expressed in the entire populations of
Extended Data Fig. 2 | Marking and lineage tracing of TERT+ cells. a, Whole- tdTomato and bright-field image of untangled seminiferous tubules in a single
mount immunofluorescence of tdTomato and E-cadherin, a membrane marker whole testis at 3-month after oil injection. Scale bars, 5 mm. e, Singly isolated
of US, in whole-mount seminiferous tubules at day two. Scale bars, 100 μm. tdTomato+ patch. Scale bar, 1 mm. f, Epifluorescence for tdTomato in untangled
b, Quantification of tdTomato+ cells in E-cadherin+ US (n = 4 mice per group). seminiferous tubules at 3-month or 1-year after pulse labelling; injected
Data are represented as mean ± s.e.m. c, Whole-mount immunofluorescence of tamoxifen dose are shown (0.25–4 mg). Scale bars, 5 mm. g, Cross-sections of
KIT and tdTomato in a Tert CreER/+:Rosa lsl-Tdtomato/+ seminiferous tubule at two days testes immunostained with antibodies to tdTomato at 3-month or 1-year after
after high dose tamoxifen injection. Scale bars, 100 μm. d, Epifluorescence of labelling. Scale bars, 50 μm.
Article

Extended Data Fig. 3 | Conditional deletion of TERT in SSCs using Tert-flox area in seminiferous tubules at 3-month after high dose tamoxifen administration
mice. a, Tert-flox targeting strategy and Southern blot strategy. b, Southern (n = 5 mice per group). Scale bars, 100 μm. g, Histology and immunostaining
blot analysis of genomic DNA from mouse tails. Locations of 5’ and 3’ probes are against tdTomato in Tert CreER/flox mice at 3-month after oil or tamoxifen
shown in a. c, TRAP assay using mouse embryonic fibroblasts from Tert-flox administration. h, TRAP assay using whole testes at 3-month after 5 mg x 3
mice. Cells were transfected with adenovirus to express LacZ or Cre. *, first times tamoxifen administration. *, first telomerase addition product; IC,
telomerase addition product; IC, internal PCR control. d, Genotyping of internal PCR control. i, Sperm count in epididymis at 3-month after labelling
tdTomato+ patches at 3-month after labelling. Primers to detect recombined (n = 5 mice per group). j, qRT–PCR analysis on exon 2 and 6 of Tert mRNA in
Tert allele that lacks exon 3-6 were used. Hprt was used as a positive control. whole testes at 3-month after 5 mg × 3 times tamoxifen administration.
e, Immunostaining against tdTomato using testes at 3-month after high dose Expression was normalized with Actb (n = 5 mice per group). P values determined
tamoxifen administration. Scale bars, 100 μm. f, Quantification of tdTomato+ by two-sided unpaired t-test. Data are represented as mean ± s.e.m.
Extended Data Fig. 4 | See next page for caption.
Article
Extended Data Fig. 4 | TERT promotes competitive clone formation of SSCs 5 days post labelling and stained. f, Quantification of foci number of Trp53 mRNA
independent of its enzyme activity and the canonical telomerase complex. (n = 174 and 135 cells from 4 mice per group from left to right). g, Epifluorescence
a, TRAP assay using whole testes from Tert CreER/+:Terc+/+ and G1 Tert CreER/+:Terc−/− for tdTomato in untangled seminiferous tubules at 3-month after labelling.
mice. *, first telomerase addition product; IC, internal PCR control. b, qRT–PCR Scale bars, 5 mm. h–j, Quantitative data of patch number (h), average patch
analysis of Tert and Tertci in purified tdTomato+ US at day 7. Doxycycline was length (i) and total patch length ( j) (n = 6, 6, 7, 7 mice from left to right). P values
added to drinking water to reduce transgenic Tert and Tertci expression. determined by one-way ANOVA with Tukey’s test. k, Immunofluorescence of
Expression was normalized with Actb (n = 3 mice per group). c, TRAP assay using γH2AX, PLZF, and tdTomato in testicular cross-sections. G6 TertTdtomato/Tdtomato
purified tdTomato+ US from indicated genotypes. *, first telomerase addition mice were used as positive control for γH2AX staining. Normal spermatocytes
product; IC, internal PCR control. d, Tert CreER/flox :Rosalsl-Ttomato/+:Trp53 flox/flox mice. are positive for γH2AX. Scale bars, 30 μm. l, Quantification of γH2AX+ PLZF+ US
e, In situ hybridization against Trp53 mRNA. tdTomato+ US were sorted out at (n = 4 mice per group). Data are represented as mean ± s.e.m.
Extended Data Fig. 5 | Differentiation in TERT-deleted US. one-way ANOVA with Tukey’s test. d, Epifluorescence for tdTomato in untangled
a, Immunofluorescence for GFRA1 and tdTomato at 14 days post labelling. seminiferous tubules at 6 weeks after 1 mg tamoxifen injection. Scale bars, 5 mm.
Scale bars, 100 μm. b, Flow cytometry analysis of testicular cells stained with e, Mean patch number at 6 weeks after tamoxifen injection (n = 5 mice per group).
α6-Integrin, MCAM, and KIT. tdTomato+, α6-Integrin-high cells were further f, Immunofluorescence for tdTomato and PLZF using testis cross-sections at
separated by KIT and MCAM expression. c, Quantification of US-h, US-m, and 6 weeks after tamoxifen injection. Scale bars, 50 μm. Data are represented as
DS populations (n = 5, 5, 4, 4 mice per group). P values were determined by mean ± s.e.m.
Article

Extended Data Fig. 6 | Proliferation and apoptosis in TERT-deleted US.

a, Immunofluorescence for tdTomato, PLZF, and BrdU using testis cross-sections
at 7 days after labelling. Scale bars, 30 μm. b, Immunofluorescence for
tdTomato, GFRA1, and BrdU using testis cross-sections at 7 days after labelling.
Scale bars, 30 μm. c, Quantification of GFRA1+ tdTomato+ BrdU+ US (n = 6 mice
per group). P values determined by two-sided unpaired t-test. Data are
represented as mean ± s.e.m. d, Immunofluorescence for PLZF, tdTomato, and
apoptosis marker, cleaved-PARP (cPARP), on testicular cross-sections at 7 days
after labelling. G6 TertTdtomato/Tdtomato mice were used as positive control for
cleaved-PARP staining. Scale bars, 50 μm. e, Quantification of PLZF+ tdTomato+
cleaved-PARP+ US (n = 4 mice per group). Data are represented as mean ± s.e.m.
Extended Data Fig. 7 | See next page for caption.
Article
Extended Data Fig. 7 | Global changes of open chromatin structure in US Tert CreER/flox US-h. g–j, Read distribution of ATAC-seq data across different cell
after acute deletion of TERT. a, PCA of ATAC-seq data. Colours represent types. g, Tert locus. h, Ret and Gfra1 loci. These genes are dominantly expressed
indicated cell types from Tert-heterozygous mice. Arrow indicates direction of in US-h. i, Cdh1 and Zbtb16 loci. These genes are dominantly expressed in both
differentiation. b, Pearson correlation heat maps of ATAC-seq samples. c, Heat US-h and US-m. j, Kit and Prm1/Prm2 loci. Kit is dominantly expressed in DS and
map representation of 7597 peaks showing significant differences across US-h, Prm1/Prm2 are dominantly expressed in RS. k, Immunofluorescence for GFRA1
US-m, DS, SP, and RS. Each row represents one ATAC-seq peak. Colour represents and PLZF in sorted US-h, US-m and DS. l, Quantification of GFRA1+ and PLZF+
relative ATAC-seq accessibility. d,e, Venn diagrams showing the number of expression in sorted cells (n = 132, 154, 201, 125, 160, 180 cells from 3 mice per
peaks found in spermatogonia subpopulations (Tert CreER/+ (d) and Tert CreER/flox group from left to right).
(e)). f, KEGG pathway analysis of differential peaks between Tert CreER/+ and
Extended Data Fig. 8 | RNA-seq analysis of TERT-deleted US-h. a, Volcano plot P < 0.0001) are coloured in red. b, Gene set up-regulated in US-h purified from
of RNA-seq data using purified tdTomato+ US-h from Tert CreER/+ and Tert CreER/flox Tert CreER/flox mice. c, Gene sets down-regulated in US-h purified from TertCreER/flox
mice. Genes showing the significant changes (more than twofold change and mice. d, Gene expression of US and DS marker genes in RNA-seq data.
Article

Extended Data Fig. 9 | See next page for caption.

Extended Data Fig. 9 | Downregulation of MYC in TERT-deleted US. human MYC mRNA in purified tdTomato+ US at day 7 after tamoxifen labelling.
a, Immunofluorescence for MYC, PLZF, and tdTomato at day 7 after tamoxifen The primers and probe used for this assay were human-specific and do
labelling. Scale bars, 50 μm. b, Quantification of mean signal intensities for not detect mouse Myc. Doxycycline was added to drinking water.
PLZF staining in US (n = 77 cells from 4 mice, 76 cells from 6 mice, 77 cells from Expression was normalized with Actb (n = 3 mice per group). n.d.; not detected.
4 mice, 81 cells from 4 mice from left to right). c, qRT–PCR analysis of mouse g, Immunostaining against tdTomato and histology of testes at 3-month after
Myc mRNA in purified tdTomato+ US at day 7. Doxycycline was added to labelling. Scale bars, 100 μm. h, Epifluorescence for tdTomato in untangled
drinking water. Expression was normalized with Actb (n = 3 mice per group). seminiferous tubules at 3-month after 1 mg tamoxifen administration. Scale
d, Immunofluorescence for MYC and PLZF in testicular cross-sections. Scale bars, 5 mm. i–k, Quantitative data of patch number (i), average patch length ( j)
bars, 20 μm. e, Quantification of mean signal intensities for MYC and PLZF and total patch length (k) (n = 7, 6, 6, 6 mice from left to right). P values
staining in US (n = 115, 117 cells from 4 mice per group from left to right). P values determined by one-way ANOVA with Tukey’s test. Data are represented as
determined by two-sided unpaired t-test. f, qRT–PCR analysis of transgenic mean ± s.e.m.
Article

Extended Data Fig. 10 | Telomere length in TERT-deleted male germ cells. immunofluorescens for tdTomato using spermatocytes one year after 5 mg
a, Immunostaining for tdTomato using testes at one year after 5 mg tamoxifen tamoxifen injections. Arrows in the right panels indicate spermatocytes.
injections. Scale bars, 100 μm. b, FISH for telomeres and centromeres and Scale bars, 10 μm.
 α