PAX7 expression defines germline stem cells in the adult testis

Abstract

Spermatogenesis is a complex, multistep process that maintains male fertility and is sustained by rare germline stem cells. Spermatogenic progression begins with spermatogonia, populations of which express distinct markers. The identity of the spermatogonial stem cell population in the undisturbed testis is controversial due to a lack of reliable and specific markers. Here we identified the transcription factor PAX7 as a specific marker of a rare subpopulation of Asingle spermatogonia in mice. PAX7+ cells were present in the testis at birth. Compared with the adult testis, PAX7+ cells constituted a much higher percentage of neonatal germ cells. Lineage tracing in healthy adult mice revealed that PAX7+ spermatogonia self-maintained and produced expanding clones that gave rise to mature spermatozoa. Interestingly, in mice subjected to chemotherapy and radiotherapy, both of which damage the vast majority of germ cells and can result in sterility, PAX7+ spermatogonia selectively survived, and their subsequent expansion contributed to the recovery of spermatogenesis. Finally, PAX7+ spermatogonia were present in the testes of a diverse set of mammals. Our data indicate that the PAX7+ subset of Asingle spermatogonia functions as robust testis stem cells that maintain fertility in normal spermatogenesis in healthy mice and mediate recovery after severe germline injury, such as occurs after cancer therapy.

Full text

The Journal of Clinical Investigation ReseaRch aRticle
3929
jci.org Volume 124 Number 9 September 2014
Introduction
The functional unit of the mammalian testis, the seminiferous
tubule, is a multilayered epithelium that matures from spermato-
gonial precursors located at the basal layer to more advanced
cell types that migrate toward the tubular lumen, where sper-
matozoa are released . Classically, type Asingle spermatogonia,
which reside on the basement membrane i.e., the basal layer,
were thought to represent the stem cell population of the testis,
as these cells were the earliest identiable morphological progeni-
tors , . Meticulous histological studies have shown that Asingle
spermatogonia progress through multiple rounds of mitoses with
incomplete cytokinesis to produce “chains” of Apair and “aligned”
Aal, Aal, and Aal spermatogonia, which consist of , , , and 
interconnected cells, respectively . Asingle–Aal spermatogonia
are sometimes called “undierentiated” spermatogonia, a term
that is useful but also somewhat misleading, in that this popula-
tion encompasses the true stem cells as well as a progressive series
of dierentiating, transit-amplifying intermediates. Interestingly,
time-lapse imaging studies of mouse testes have clearly docu-
mented that Asingle–Aal spermatogonia are highly migratory, capa-
ble of moving across large distances on the basement membrane
. Aal spermatogonia dierentiate to give rise to type A–A
and then to type B spermatogonia, which become spermatocytes
that initiate meiosis. Round haploid spermatids, the products of
meiosis, initiate a dramatic cytoskeletal rearrangement to pro-
duce elongate spermatids, which at the end of this maturational
sequence are released within the tubular lumina as spermatozoa
Supplemental Figure ; supplemental material available online
with this article; doi:./JCIDS; and ref. .
The continuous production of spermatozoa throughout adult
life, as well as the multitude of cell divisions from Asingle spermato-
gonia to mature spermatozoa, clearly implies the existence of
a dynamic germline stem cell capable of self-maintenance, but
also dierentiation into the transit-amplifying intermediates that
constitute the spermatogenic series . The identity of this adult
testis stem cell remains unknown . As stated above, some mod-
els have posited that all Asingle spermatogonia represent functional
stem cells, consistent with their status as the earliest known mor-
phological precursor. Asingle spermatogonia can be reliably identi-
ed by morphologic criteria i.e., their singularity by confocal
microscopy of intact tubules but have remained largely undened
at the molecular level, although recently ID was described as a
marker of Asingle spermatogonia . On the other hand, some stud-
ies have suggested that only a subset of Asingle spermatogonia are
functional stem cells . If so, then this would suggest that Asingle
spermatogonia encompass the true stem cells a distinct fraction
of Asingle spermatogonia, along with other Asingle subsets that serve
as transit-amplifying descendants prior to their eventual dieren-
tiation to Apair spermatogonia.
Transplantation of spermatogonia from a donor mouse to a
germ cell–decient recipient testis  has been extensively used
to explore the properties and biology of spermatogonial stem cells
Spermatogenesis is a complex, multistep process that maintains male fertility and is sustained by rare germline stem cells.
Spermatogenic progression begins with spermatogonia, populations of which express distinct markers. The identity of the
spermatogonial stem cell population in the undisturbed testis is controversial due to a lack of reliable and specific markers.
Here we identified the transcription factor PAX7 as a specific marker of a rare subpopulation of Asingle spermatogonia in
mice. PAX7+ cells were present in the testis at birth. Compared with the adult testis, PAX7+ cells constituted a much higher
percentage of neonatal germ cells. Lineage tracing in healthy adult mice revealed that PAX7+ spermatogonia self-maintained
and produced expanding clones that gave rise to mature spermatozoa. Interestingly, in mice subjected to chemotherapy and
radiotherapy, both of which damage the vast majority of germ cells and can result in sterility, PAX7+ spermatogonia selectively
survived, and their subsequent expansion contributed to the recovery of spermatogenesis. Finally, PAX7+ spermatogonia were
present in the testes of a diverse set of mammals. Our data indicate that the PAX7+ subset of Asingle spermatogonia functions
as robust testis stem cells that maintain fertility in normal spermatogenesis in healthy mice and mediate recovery after severe
germline injury, such as occurs after cancer therapy.
PAX7 expression defines germline stem cells
in the adult testis
Gina M. Aloisio,1 Yuji Nakada,1 Hatice D. Saatcioglu,1 Christopher G. Peña,1 Michael D. Baker,1 Edward D. Tarnawa,2
Jishnu Mukherjee,1 Hema Manjunath,1 Abhijit Bugde,3 Anita L. Sengupta,1 James F. Amatruda,4
Ileana Cuevas,1 F. Kent Hamra,5 and Diego H. Castrillon1
1Department of Pathology, 2Department of Obstetrics and Gynecology, Division of Reproductive Endocrinology and Infertility, 3Department of Cell Biology, 4Departments of Internal Medicine,
Molecular Biology, and Pediatrics, and 5Department of Pharmacology, UT Southwestern Medical Center, Dallas, Texas, USA.
Conflict of interest: The authors have declared that no conflict of interest exists.
Submitted: March 3, 2014; Accepted: July 1, 2014.
Reference information: J Clin Invest. 2014;124(9):3929–3944. doi:10.1172/JCI75943.

The Journal of Clinical Investigation
ReseaRch aRticle
3930 jci.org Volume 124 Number 9 September 2014
ducted a number of investigations to explore the contribution of
these rare PAX spermatogonia to spermatogenic recovery after
germ cell ablation, such as occurs after chemotherapy or radio-
therapy. Remarkably, PAX spermatogonia selectively survived
cytotoxic drugs and radiation despite widespread germ cell death.
Not only did their numbers not decrease in the immediate after-
math of such treatments, but instead, PAX spermatogonia rap-
idly expanded to replenish normal spermatogenesis. Thus, our
results demonstrated that PAX spermatogonia are stem cells that
maintain fertility in the healthy adult, and also serve particularly
important roles in replenishing spermatogenesis following treat-
ments that damage the germline.
Results
PAX specically marks a small subset of Asingle spermatogonia in vivo.
We reasoned that a bona de testis stem cell marker should be
highly expressed in SSC cultures, but at low perhaps undetect-
able levels in the adult testis, where true stem cells are a rare
subpopulation. An RNA-based approach previously used for
marker discovery in ovarian cell subpopulations  led us to the
identication of Pax Figure A. At the mRNA level, Pax was
highly expressed in SSC cultures, but undetectable in adult testis
-fold dierence; Figure B. In vivo, Pax transcripts were
detectable in type A early spermatogonia, but not in dierenti-
ated type B spermatogonia, spermatocytes, or round spermatids.
Pax transcripts were absent in testes at embryonic days ..
e.–e. and rst detected at postnatal day  PD. Among
adult tissues, Pax was expressed only in skeletal muscle, consis-
tent with PAX’s eminence as a marker of satellite cells, the dor-
mant tissue stem cell population that regenerates skeletal muscle
after injury , . In comparison, the pan-germ cell marker
Ddx also known as VAS A was expressed in adult testis  and
in all germ cell subpopulations, but not in any somatic tissues,
and transcripts were markedly decreased in germ cell–decient
KitlSl/KitlSl–d testes, as expected Figure C. Thus, in contrast to
Pax, Ddx did not exhibit a stem cell signature.
We sought to visualize PAX cells in sections of adult testis
with an anti-PAX monoclonal antibody. PAX cells were rare:
in  complete testis cross-section, a single PAX cell might be
detected within seminiferous tubules Figure A. Despite this
rarity, several observations conrmed that the detection of PAX
spermatogonia was specic, dening a novel population of sper-
matogonia. First, the PAX cells always rested on the basement
membrane and were isolated, single cells i.e., consistent with
Asingle spermatogonia; see also below. Second, PAX protein in
spermatogonia was always nuclear, as expected based on its func-
tion and nuclear localization within satellite cells , .
To further dene these PAX spermatogonia, we compared
their abundance with that of other subsets of spermatogonia
dened by well-characterized markers Supplemental Figure .
KIT dierentiating spermatogonia were the most abundant .
cells/tubule, with FOXO and PLZF spermatogonia being more
restricted, as expected, given that FOXO and PLZF are both
markers of undierentiated Asingle→Aal spermatogonia, a less
abundant population , . RET spermatogonia were rarer
still, consistent with RET’s more restricted expression in Asingle and
Apair spermatogonia . However, PAX spermatogonia were
SSCs . In these assays, the regeneration of complete sper-
matogenesis occurs via the formation of spermatogenic colonies
thought to arise from a single transplanted cell. Clonogenicity is
a notable strength of the assay, permitting assessment of stem
cell numbers in the donor population. However, transplantation
has not proven decisive in identifying the true presumably rare
stem progenitors in the adult testis. Most strategies to enrich SSCs
in transplantation assays to date have used cell surface selec-
tion markers such as THY  or α/β integrins  that are
expressed across broad subsets of spermatogonia, limiting their
precision in pinpointing rarer subsets of stem progenitors , ,
. Furthermore, transplantation assays do not mirror stem cell
functionality in the undisturbed testis. Donor germ cells are disso-
ciated into single-cell suspensions, resulting in chain fragmenta-
tion, a phenomenon that occurs in vivo and has been proposed as a
distinct mechanism promoting stem cell renewal, though this has
not yet been conclusively demonstrated . Additionally, germ
cells in the recipient are ablated by treatments that also damage
the somatic environment, which may induce stemness by increas-
ing the number of available niches or by eliminating negative
feedback signals that emanate from other germ or somatic cells to
regulate stem cell numbers within the testis , , .
A notable strength of the mouse testis as a model system for
stem cell biology is the ability to establish cultured SSCs ex vivo.
Cultured SSCs are capable of self-renewal and are truly immor-
tal: they can be passaged and expanded indenitely, maintaining
genetic stability. The dynamics of stem cell maintenance in the
murine SSC culture system are incompletely understood, but the
cultures contain a much higher fraction of stem cells relative to
the adult testis , . Cultured SSCs can be transplanted into a
host testis, where they function as tissue stem cells, reestablishing
functional spermatogenesis . The limitless expansion of SSCs
in culture is dependent on the growth factor GDNF, which acts
through the RET/GFRα receptor complex, although additional
growth factors are also necessary .
Here we report on our identication of a rare subset of Asingle
spermatogonia characterized by expression of the paired box tran-
scription factor PAX. PAX has been previously identied and
extensively used as a marker of satellite cells, which function as a
normally quiescent stem cell population within adult skeletal mus-
cle . In contrast, PAX spermatogonia were highly prolifera-
tive during steady-state spermatogenesis. To explore the contri-
bution of PAX spermatogonia to normal spermatogenesis in the
undisturbed testis, we performed a variety of cell lineage–tracing
studies with an inducible PaxCreE RT allele and  dierent report-
ers. These studies revealed that PAX spermatogonia normally
serve as robust stem cells that contribute to full-lineage matura-
tion, as evidenced by the formation of clones including all stages
of spermatogenesis from Asingle spermatogonia to spermatozoa.
Furthermore, labeled i.e., lineage-traced descendants including
Asingle spermatogonia were observed even after prolonged intervals
 weeks, which demonstrated that PAX spermatogonia func-
tion as bona de stem cells that self-maintain and dierentiate to
sustain spermatogenesis, and are not merely transit-amplifying
intermediates. Lineage-tracing studies with neonatal mice sug-
gest that adult PAX spermatogonia are derived from an initial
cohort of PAX spermatogonia present at birth. Finally, we con-

The Journal of Clinical Investigation ReseaRch aRticle
3931
jci.org Volume 124 Number 9 September 2014
approximately  orders of magnitude rarer than RET
spermatogonia Figure B. PAX spermatogonia were
FOXO and GFRα, while most FOXO or GFRα
cells were PAX– Figure C, which demonstrated that
PAX spermatogonia represented a subset of undier-
entiated, GFRα spermatogonia.
Confocal microscopy of intact seminiferous tubules
further showed that PAX spermatogonia were a subset
of Asingle spermatogonia. PAX spermatogonia were sin-
gular, and larger chains of undierentiated spermatogo-
nia i.e., Aal–Aal never contained PAX spermatogonia
Figure D and Supplemental Video . Additional confo-
cal microscopy studies conrmed that PAX spermato-
gonia were always KIT–; no KITPAX spermatogonia
were ever observed Figure E. Thus, PAX dened a
rare but specic subset of Asingle spermato gonia, revealing
striking heterogeneity within Asingle spermatogonia in vivo.
PAX spermatogonia are rare in the adult testis, but
constitute a much higher fraction of germ cells in the neo-
natal testis. Interestingly, a much higher percentage of
germ cells dened by the pan–germ cell marker germ
cell nuclear antigen GCNA were PAX at birth 
in neonates; however, this fraction steadily decreased
postnatally, stabilizing at  weeks of age Figure A. The
much higher fraction of PAX germ cells at birth further
underscores their rarity in adults, and also demonstrat-
ed that PAX spermatogonia can be reliably identied
in tissue sections; analyses of conditional knockout tes-
tes also conrmed antibody specicity see below.
This age-dependent decrease in the proportion of
PAX cells per total GCNA cells could reect decreased
absolute numbers of PAX spermatogonia, versus their
dilution due to the massive expansion of spermatogenic
cells that normally occurs during postnatal life e.g., tes-
tes weights increase from ~ mg at birth to ~ mg in adult
males . To distinguish between these possibilities,
we serially sectioned and immunostained entire PD and
adult testes and documented similar numbers of PAX
spermatogonia per testis    and   , respec-
tively; mean  SEM; Figure B. Thus, the dramatic age-
dependent decrease in the fraction of PAX germ cells
reects mainly the rapid expansion of spermatogenesis,
and not a large decrease in absolute numbers of PAX
cells. These results also strongly suggest that the post-
natal PAX spermatogonia represent the initial founder
population for PAX spermatogonia in adult testes.
Figure 1. Digital Northern analysis identifying PAX7 as poten-
tial adult testis germline stem cell marker. (A) General RNA-
based approach to identify markers that were highly expressed
in cultured SSCs relative to adult testis. (B and C) Relative
expression levels of (B) Pax7 and (C) Ddx4 across multiple
samples. Error bars denote SEM. Pax7 mRNA levels were
>180-fold higher in established spermatogonial cultures rela-
tive to adult testis. SSC, cultured SSCs; Sl/Sl(d), KitlSl/KitlSl–d
germ cell–deficient adult testes; ES, embryonic stem; HS,
hematopoietic stem.

The Journal of Clinical Investigation
ReseaRch aRticle
3932 jci.org Volume 124 Number 9 September 2014
inducible recombinase CreERT  was knocked into the Pax locus,
driving CreE RT expression in cells that express Pax, such as sat-
ellite cells Supplemental Figure A and refs. , . We gener-
ated mice harboring PaxCreERT  and the Rosa β-galactosidase
lox-stop-lox reporter, RR . -week-old adult males were
treated with tamoxifen to activate Cre in PAX cells. Untreated
PaxCreE RT;RR males exhibited no Cre-mediated recombina-
tion in testis or skeletal muscle, demonstrating tight control of
Cre. Expression of PAX in labeled clones conrmed faithful Pax-
CreERT  expression in PAX spermatogonia Supplemental Fig-
ure , BD. To characterize PAX descendants within the tes-
PAX spermatogonia are rapidly cycling during normal sper-
matogenesis and function as robust stem cells that give rise to all stages
of spermatogenesis. We considered the possibility that by analogy
with satellite cells adult PAX spermatogonia might represent
a quiescent subset of Asingle spermatogonia. To our surprise, how-
ever, EdU labeling showed that PAX cells, like other subsets of
spermatogonia, were rapidly cycling Figure C.
We then sought to explore the contribution of PAX sper-
matogonia and their descendants to normal, steady-state sper-
matogenesis in the undisturbed adult testis through lineage
tracing. We used a PaxCreERT allele, in which the tamoxifen-
Figure 2. PAX7+ spermatogonia in normal testes. (A) Rarity of PAX7+ cells in adult (6-week-old) testis cross-section. Arrow denotes a PAX7+ cell within a
single seminiferous tubule. (B) Relative number of spermatogonia positive for known markers compared with PAX7 in tissue sections, with >90 tubules
counted per testis. Error bars denote SEM of averages derived from 3 6-week-old animals. (C) Double-labeling (confocal microscopy) showing that PAX7+
cells were FOXO1+ and GFRα1+ (representative examples among ≥10 PAX7+ cells). Basement membrane staining is nonspecific. (D) FOXO1 and PAX7 double-
labeling of Asingle, Apair, and Aal4–Aal16 chains, visualized by confocal microscopy. PAX7+ spermatogonia were Asingle spermatogonia; larger chains did not contain
PAX7+ spermatogonia. Arrow indicates an Aal4 chain. Bottom panels show dierent channels for the same field of a single PAX7+FOXO1+ spermatogonium
(asterisk in DAPI channel). (E) KIT and PAX7 double-labeling (confocal microscopy) showed that PAX7+ spermatogonia (arrow) were isolated (i.e., Asingle) and
KIT–. No KIT+PAX7 + spermatogonia were observed. PAX7 was nuclear, whereas KIT was membrane-associated, as expected. Image shows 3 tubules optically
sectioned close to the level of the basement membrane to visualize large numbers of spermatogonia. Scale bars: 10 μm (A); 25 μm (C and D); 50 μm (E).

The Journal of Clinical Investigation ReseaRch aRticle
3933
jci.org Volume 124 Number 9 September 2014
Lineage-tracing experiments with PaxCreERT  and a double-
uorescent tdTomato/eGFP reporter mT/mG  gave nearly
identical results. Clones began as single cells. At  week after Cre
induction, labeled Asingle, Apair, and Aal–Aal chains were identied.
At  weeks, larger clones were visualized, and elongate spermatid
tails were rst identied in tubular lumina. By  weeks, clones
were even larger Figure A, and motile labeled sperm were
present in epididymides Supplemental Video . Labeled Asingle
spermatogonia were observed at all time points Figure A and
Supplemental Videos  and . To more clearly delineate clonal
architecture, we also analyzed frozen tissue sections of intact
testes, which permitted better visualization of spermatogenic
layers and cell types. In some clones, all of the germ cells in the
entire tubular cross-section were clearly labeled green; Figure
B, which demonstrated that all the germ cells spermatogonia,
spermatids, and spermatocytes were derived from a PAX pro-
genitor. These results indicated that PAX spermatogonia give
rise to full-lineage maturation Figure C, thereby fullling a key
criterion of an adult testis stem cell. As with RR-based lineage
tracing, clone numbers did not decrease, even when the analyses
were extended to  weeks after tamoxifen treatment Figure , D
and E. We concluded from these analyses with  distinct report-
ers that PAX spermatogonia are rare but robust tissue stem cells.
They give rise to other Asingle spermatogonia that persist even after
very long intervals of  weeks, and also give rise to all stages of
spermatogenesis, including motile sperm.
Lineage-tracing studies of neonatal animals show that neonatal
PAX spermatogonia have long-term stem cell potential in vivo and
also have stem cell activity in transplantation assays. Lineage-trac-
ing studies initiated with neonatal animals PDPD conrmed
that PAX spermatogonia were rapidly expanding by PD and
tis, labeled clones were analyzed after dened time intervals. By
“clone,” we refer not to individual spermatogonial chains, but
rather to completely isolated groups of labeled cells that could
constitute multiple chains, but were clearly descendants of a com-
mon progenitor — despite their sometimes complex arrangements
— because of their close proximity. Notably, clones were very dis-
tant to their nearest neighbors and were much more isolated than
the gures convey, with no labeled clones evident in either direc-
tion along the tubule. Thus, there is no question that these clones
were separate tracing events at every time point analyzed.
At  days after tamoxifen treatment, clones were very small,
most consisting of single, isolated cells, similar to the PAX immu-
nostaining pattern Figure A. The presence of slightly larger clones
after  weeks and their subsequent rapid expansion was consistent
with rapid cycling. There was striking clone expansion at succes-
sive time points, such that by  weeks of age, very large clones were
readily visualized. Interestingly, larger clones were sometimes asso-
ciated with distinctive “trails” of cells, some of which were clearly
Asingle spermatogonia based on their distance from other labeled
descendants i.e., multiple cell diameters; Figure A and Supple-
mental Video . This is indicative of complex patterns of migration
of PAX cells and their descendants, but could also reect chain
fragmentation. Labeled elongate spermatid tails were rst iden-
tied  weeks after induction Figure , B and C. Average clone
size i.e., number of cells per clone increased over time, but clone
numbers remained stable over this prolonged interval. This argues
that PAX spermatogonia are bona de stem cells. If PAX sper-
matogonia were instead transit-amplifying intermediates, labeled
cells would become diluted out with time and eventually disappear,
because the total duration of spermatogenesis i.e., from Asingle sper-
matogonium to sperm release is only  days in the mouse .
Figure 3. PAX7+ spermatogonia
make up a higher percentage of
germ cells in the neonatal testis.
(A) PAX7+ cell frequency, expressed
as a percentage of total GCNA+
cells. Inset: PAX7 IHC at PD7, show-
ing several PAX7+ spermatogonia
(asterisks). Magnification, ×250. (B)
Counts of total PAX7+ cells by IHC
of serially sectioned PD1 and adult
testes (n = 3). Each bar represents
a single testis. Total numbers
were similar in neonatal and adult
testes. (C) 6-week-old animals were
injected with EdU, and testes were
harvested after the indicated times.
The experiment was repeated twice
with similar results.

The Journal of Clinical Investigation
ReseaRch aRticle
3934 jci.org Volume 124 Number 9 September 2014
were progenitors of subsequent Asingle spermatogonia and sper-
matogenesis; clones grew in size over time and persisted into
adulthood Figure , A and B. Concordantly, PAX spermato-
gonia were rapidly proliferating by PD, as shown by EdU incor-
poration rate, without signicant cell death Figure C. Finally,
although ow sorting of live PAX spermatogonia was not possi-
ble with available reagents, transplantations were conducted with
unsorted cells from tamoxifen-treated PaxCreERT ;tdTomato
reporter mice . Labeled, lineage-traced germ cells were
observed in every host n  ; Figure D, which indicated that
PAX spermatogonia and their descendants have stem cell activ-
ity in transplantation assays.
PAX spermatogonia are selectively resistant to anticancer ther-
apies that kill other germ cells in the adult testis radiotherapy and
chemotherapy and also contribute to spermatogenic recovery after
ablation of most germ cells. Spermatogenesis is highly sensitive
to systemic genotoxic stresses, such as cytotoxic chemotherapy.
Chemotherapy-induced ablation of germ cells has been studied in
rodent models. After treatment with the alkylating agent busulfan
also known as Myleran; used to treat hematopoietic malignan-
cies, germ cells undergo massive cell death in a dose-dependent
manner, with higher doses leading to near-total germ cell deple-
tion. This results in an interval of azoospermia and infertility,
followed by a gradual recovery of spermatogenesis and, in most
Figure 4. Lineage tracing of PAX7+ descendants in Pax7-CreERT2;R26R testes. Mice were treated with tamoxifen at 6 weeks of age then aged for the indi-
cated intervals. (A) Representative clones at 4 days, 3 weeks, and 6 weeks. Dashed lines demarcate tubule borders. Arrows denote a 1-cell clone at
4 days or isolated Asingle spermatogonia at the periphery of larger clones at 6 weeks. All marker-expressing cells in these fields are considered clonal
because there were no additional marked cells in the tubule. The H&E-stained section of Xgal-treated testis at 6 weeks showed a single-labeled (blue)
elongate spermatid tail among many unlabeled (pink) spermatid tails in the tubular lumen. Other labeled spermatid tails were outside the field of view.
Scale bars: 100 μm; 10 μm (H&E). (B) Average clone numbers in n = 4 testes. The 1-week time point may represent undercount due to diculty in visual-
izing single-cell clones, or marker expression lag. Clone numbers were consistent with the presence of approximately 400 PAX7+ cells per testis and 10%
recombination eciency for Pax7-CreERT2 after tamoxifen administration (31). (C) Clone size. Red bars denote means. Larger clones were composed of large
labeled zones and peripheral smaller chains including Asingle spermatogonia.

The Journal of Clinical Investigation ReseaRch aRticle
3935
jci.org Volume 124 Number 9 September 2014
animals, restoration of fertility even after high doses . Such
spermatogenic recovery poses a paradox: the germline is almost
entirely ablated, yet the restoration of spermatogenesis implies
the existence of rare stem cells that not only survive, but replenish
spermatogenesis during the recovery period .
To study the contribution of PAX spermatogonia to sper-
matogenic recovery, adult mice  weeks of age were treated
with busulfan. As expected, testes underwent massive germ cell
death with dose- and time-dependent germ cell loss, as visual-
ized by immunohistochemistry IHC with the pan–germ cell
Figure 5. Lineage tracing of PAX7+ descendants in Pax7-CreERT2;mT/mG testes. Adult males were treated with tamoxifen at 6 weeks of age, then aged
for the indicated intervals. (A) Clone morphology by confocal microscopy of isolated tubules. Representative Asingle, Apair, Aal4, and Aal8 clones 1 week after
tamoxifen administration are shown. Other panels show larger clones at 6 weeks; arrows indicate detached Asingle spermatogonia that were part of larger
clones. Inset: elongated spermatid tails in tubular lumen. Clones >500 cells could not be reliably counted. Larger clones were associated with smaller
separate chains at their periphery, including Asingle spermatogonia. The few 1-cell clones at 6–16 weeks represent Asingle spermatogonia too distant from the
nearest clone to be confidently identified as part of it, but may reflect long-distance migration. Green motile sperm were observed in the epididymis.
(B) Clone morphology in tissue sections (16 weeks after tamoxifen). PAX7+ descendants gave rise to all spermatogenic stages, as evidenced by circumfer-
ential full-thickness labeling of all spermatogenic stages throughout the tubule. Tissue sections were counterstained with DAPI. (C) Labeled sperm from
epididymis showing bright green fluorescence and characteristic hook morphology. The majority of sperm did not exhibit fluorescence, and control epididy-
mides did not contain spermatozoa with comparable fluorescence (i.e., the signal shown is not background autofluorescence of sperm). (D) Average clone
number in n = 4 testes. Clone numbers did not decrease over time. (E) Clone size. Red bars denote means. Larger clones included detached smaller chains
and Asingle spermatogonia. Scale bars: 25 μm.

The Journal of Clinical Investigation
ReseaRch aRticle
3936 jci.org Volume 124 Number 9 September 2014
onstrating that busulfan treatment stimulated PAX cell division
acutely and suggesting that cell division is one mechanism under-
lying the formation of PAX cell clusters. In contrast to PAX
spermatogonia, FOXO undierentiated spermatogonia counts
fell more than -fold  days after  mg/kg busulfan administra-
tion, but then recovered coincident with the peak of PAX expan-
sion Figure A. These data indicate that spermatogonia are sen-
sitive to genotoxic stress as previously reported , emphasizing
the unique properties and survival of PAX spermatogonia after
treatments that ablate the vast majority of germline cells. That the
increase in FOXO spermatogonia coincided with the decrease of
PAX spermatogonia Figure , B and C is further evidence that
PAX spermatogonia eventually dierentiate. In control experi-
ments, neither tamoxifen nor the DMSO solvent had a signicant
eect on testis weight or morphology or PAX spermatogonia
Supplemental Figure , AD.
We then analyzed the response of PAX spermatogonia to
ionizing radiation and a second chemotherapeutic agent com-
monly used in the clinic, cyclophosphamide. Cyclophosphamide
marker GCNA . In striking contrast, both relative and absolute
numbers of PAX spermatogonia increased several-fold, also in
a time- and dose-dependent manner Figure A. Absolute num-
bers of PAX cells peaked -fold higher than untreated mice 
days after treatment with the highest dose of busulfan  mg/kg.
PAX cell counts then decreased between  and  days, most
likely a consequence of dierentiation see below. Thus, while
germ cells as a whole were largely ablated by busulfan, PAX cells
not only survived, but expanded in number.
Histology and immunostaining conrmed massive loss of
germ cells. Whereas in untreated animals, virtually all PAX cells
were single, isolated cells with only extremely rare cells being
present as pairs, and never in clusters of , larger PAX clusters
of  to  cells were observed after busulfan treatment Figure ,
B and C. This dierence in cluster sizes  versus  was highly
statistically signicant in untreated animals versus those  days
after treatment with  mg/kg busulfan P    –. The PAX
fraction undergoing DNA replication, based on EdU incorpora-
tion, increased  days after busulfan treatment Figure D, dem-
Figure 6. PAX7+ spermatogonia have long-term stem potential in vivo, and their descendants function as stem cells in transplantation assays. For
PD3 time points, tamoxifen injections were performed at PD1 and PD2; for later time points, tamoxifen administration was performed for 3 consecutive
days starting at PD3. (A) Clone size. Each column represents 1 testis from separate animals (total n = 15); red bars denote means. Note that many labeled
Asingle spermatogonia were present at PD21, demonstrating that PD21 PAX7+ spermatogonia are derived from neonatal PAX7+ spermatogonia. Clones grew
over time and persisted in aged animals (12 weeks). (B) Representative clone morphologies by confocal microscopy (n denotes number of cells in labeled
chain shown); z stacks confirmed cell counts and Asingle status. (C) Mitotic and apoptotic indices of PAX7+ cells at PD3 (n = 3 animals) demonstrated that
early PAX7+ cells were highly proliferative and not characterized by significant apoptosis. Error bars denote SEM. (D) Transplantation assay. A Pax7-
CreERT2;tdTomato donor was treated with tamoxifen at PD3. Testes were disaggregated at PD14 and transplanted into germ cell–deficient KitW/KitW–v hosts,
which were sacrificed after 4 weeks (n = 3). All hosts (but no controls) showed multiple labeled clones (i.e., 15–20); representative examples are shown.
Scale bars: 25 μm (B); 100 μm (D).

The Journal of Clinical Investigation ReseaRch aRticle
3937
jci.org Volume 124 Number 9 September 2014
after the last treatment. With the tam→bu protocol, the number
of clones was fewer than in untreated control mice P  .,
whereas the number of clones was greater than controls in the
bu→tam protocol P  .. These results were consistent with
the expansion of PAX cells observed at  days after busulfan
Figure A and Figure , B and C. The somewhat smaller mean
clone sizes in the bu→tam versus tam→bu experiments  vs.
 clones could be explained by the administration of tamoxi-
fen  days after busulfan, whereas in the tam→bu protocol,
busulfan was administered only  days after tamoxifen. This - to
-day dierence might permit  or more additional cell doublings
to occur in the tam→bu protocol, thus accounting for the modestly
increased clone size ~-fold dierence. Clone morphology was
similar to that observed in the prior lineage-tracing experiments;
an example of a large clone is shown in Figure D. In tissue sec-
tions of the larger clones, the labeled cells made up all of the germ
cells in a tubular cross section Figure E, demonstrating that, as
in normal spermatogenesis in untreated mice, PAX spermato-
gonia contributed to full-lineage maturation after busulfan treat-
ment. Together, these data showed that PAX spermatogonia not
only selectively survive, but also contribute to the reestablishment
of spermatogenesis after diverse genotoxic insults to the germline,
including radiotherapy and chemotherapy .
Preliminary observations demonstrate that PAX is dispensable
for spermatogenesis. To study the genetic requirements for Pax
in spermatogenesis, we performed conditional genetic knock-
is less toxic to the germline, necessitating a longer treatment
protocol  mg/kg i.p. every  days for  days than busulfan,
which was administered as a single dose. Radiation was adminis-
tered in a single nonfractionated dose of  Gy. Selective survival
and clustering of PAX spermatogonia P  – similar to that
observed after busulfan were also observed after either external
irradiation or cyclophosphamide Supplemental Figures  and .
After cessation of each treatment, the number of PAX spermato-
gonia increased, then subsequently declined, as was observed
with busulfan. These results are signicant in that they demon-
strate that in the mouse, PAX spermatogonia selectively survive
rst-line cancer therapies that often result in reversible or perma-
nent sterility in men and boys . These ndings make PAX
spermatogonia strong candidates as the “spermatogenic recov-
ery” cells postulated to be responsible for the restoration of fer-
tility after cytotoxic/genotoxic treatments in rodent models .
To further explore this possibility, lineage-tracing studies
were performed with adult mice treated with  mg/kg busulfan.
In the rst experiment, lineage tracing was initiated by tamoxifen
usual -day regimen followed by busulfan referred to herein
as the tam→bu protocol. In a second experiment, the order of
treatments was reversed, and tamoxifen was administered 
days after busulfan administration, the time point coinciding with
the peak of PAX cells bu→tam protocol; Figure A. In each
experiment, PAX spermatogonia contributed to spermatogenic
recovery, as evidenced by the presence of labeled clones  weeks
Figure 7. Germline ablation with busulfan has dose-dependent eects on PAX7+ spermatogonia. (A) Number of cells expressing GCNA (pan-germ cell
marker) and PAX7. Error bars denote SEM for n = 3 animals at 6 weeks of age. (B) H&E and immunostained sections 16 days after a single dose of 40 mg/kg
busulfan. GCNA stains all germ cells to the round spermatid stage. Busulfan resulted in expansion of PAX7+ clusters never observed in untreated testes; an
example of a 4-cell group is shown (insets; enlarged ×4). (C) Fractions of PAX7+ clusters of dierent sizes. For each time point, fractions add up to 1. The
dierence in cluster sizes (1 versus ≥2) was highly statistically significant in untreated animals versus those treated with 40 mg/kg busulfan after 32 days
(P = 2 × 10–9). (D) Percent EdU incorporation in PAX7+ spermatogonia 4 days after busulfan administration. Scale bar: 10 μm.

The Journal of Clinical Investigation
ReseaRch aRticle
3938 jci.org Volume 124 Number 9 September 2014
PAX signals but not nonspecic background bands; Figure B,
which conrmed that this was the epitope detected by the PAX
monoclonal antibody. Alignment of corresponding aa sequenc-
es from diverse species revealed that the -aa PAX epitope is
conserved across all  mammalian PAX homologs evaluated,
but not in the zebrash Danio rerio or the fruit y Drosophila
melanogaster Table . We then performed immunolocalization
of PAX in tissue sections of paran-embedded, formalin-xed
testes from diverse mammalian species, including companion
and domestic animals, nonhuman primates, and humans. Rare
basal PAX spermatogonia were present in these species Figure
C. Interestingly, PAX cells were more abundant in juvenile
testes which were available for cat and baboon, with multiple
cells in some tubules, similar to our observations in mice. These
results suggest that PAX spermatogonia serve important roles
as adult testis stem cells and contribute to spermatogenesis in a
wide range of species.
Discussion
Our data revealed surprising heterogeneity in cells previously
identied through morphologic criteria as Asingle spermatogonia.
This nding also suggests the existence of further Asingle subtypes,
which may be characterized by the expression of other distinct
markers, such as ID or ERBB , , . PAX dened an unex-
pectedly small subset of Asingle spermatogonia. PAX spermatogo-
nia were highly proliferative in steady-state spermatogenesis and
fullled criteria of self-renewal and complete lineage dieren-
tiation in the adult testis. That PAX is a marker of germline stem
cells in the testis is notable in light of extensive studies of PAX as
a marker of satellite cells , . Our work shows some common-
alities between PAX stem cells in the testis and skeletal muscle,
but also some important dierences. PAX cells were rarer in the
testis, making them dicult to detect. In skeletal muscle, a tissue
characterized by little cellular proliferation, PAX cells are nor-
mally quiescent, only to become reactivated after injury. In con-
trast, in the testis, PAX spermatogonia were highly proliferative
and continually replenish spermatogenesis.
out cKO with the germline-specic VASACre VC, which we
had previously generated and characterized , and a condi-
tional oxed Pax  allele . The resulting Pax germline cKO
mice VC;Pax–/ had testes that were morphologically normal
and exhibited normal spermatogenesis, as evidenced by normal
weights and histological analyses. Furthermore, all males n  
were fertile, with normal litter sizes Figure , AD, which indi-
cated that Pax is dispensable for male fertility in mice. Pax cKO
males treated with busulfan n   showed a signicant lag in
spermatogenic recovery  weeks after treatment Supplemental
Figure , although this lag was somewhat variable. In Pax cKO
males, testes showed a trend toward smaller size P  ., but
many tubules lacked complete spermatogenesis, while in control
animals, practically all tubules had recovered P  ..
The availability of Pax cKO testes permitted us to conrm the
specicity of PAX immunodetection. Whereas intratubular germ
cells were readily detectable by GCNA immunostaining of control
and Pax cKO testes, PAX expression was abolished in Pax cKO
testes compared with wild-type controls Figure E, conrming
the specicity of PAX immunodetection. Thus, Pax appears to be
dispensable for spermatogenesis, at least in the laboratory setting,
but may make a functional contribution to the recovery of sper-
matogenesis under conditions of germline stress see Discussion.
PAX spermatogonia are present across mammalian species.
We sought to determine whether PAX spermatogonia are phy-
logenetically conserved in spermatogenesis, as many aspects
of spermatogenesis are shared by diverse species . The
monoclonal antibody we used to detect PAX spermatogonia
was generated against chicken PAX Gallus gallus; amino acid
aa , which suggests that the epitope might be broadly
conserved . However, there was no a priori guarantee that this
would be the case. We epitope-mapped the anti-PAX monoclo-
nal antibody with a tiled peptide array of both the chicken and
corresponding mouse aa sequences at -aa resolution. This identi-
ed a distinct -aa peak at identical positions in the chicken and
mouse polypeptides Figure A. Western blotting with a -aa
blocking peptide spanning this epitope eectively eliminated the
Figure 8. Counts of FOXO1+ spermatogonia
after busulfan administration at 6 weeks of
age. For each time point, n = 3 animals were
analyzed; error bars denote SEM. (A) FOXO1+
spermatogonia per tubule. (B) FOXO1+ counts
per tubule compared with PAX7+ counts. (C)
IHC showing FOXO1+ spermatogonia (arrows).
Unlike PAX7+ spermatogonia, FOXO1+ (undif-
ferentiated) spermatogonia were not resis-
tant to busulfan. Scale bar: 25 μm.

The Journal of Clinical Investigation ReseaRch aRticle
3939
jci.org Volume 124 Number 9 September 2014
Some have argued that SSC transplantation represents a gold
standard and is the only reliable assay for studying testis stem
cell activity. Reconstitution of a self-maintaining cellular clone in
a host organ is indisputable evidence that the cell of origin func-
tioned as a stem cell in the assay. However, other investigators in
the stem cell eld have challenged the assumption that transplan-
tation assays recapitulate stem cell function in native, undisturbed
organs, and have pointed out limitations inherent in transplanta-
tion assays . These concerns are valid for SSC transplantations
, , , , particularly since transplantation requires treatments
cell dissociation in the donor, near-complete germ cell ablation in
the host that may strongly stimulate regenerative potential in ways
that are not fully understood. There is an important distinction to
be made between actual stemness and the potential for stemness.
Transplantation assays are clearly useful for studying the latter, but
do not necessarily accurately reect the former . Future investi-
gations are needed to dene plasticity with respect to actual stem-
ness versus stemness potential  in the adult testis.
PAX as a testis stem cell marker. Our data are consistent with
a model whereby PAX Asingle spermatogonia function as stem
cells in the adult testis. The fact that only a minority of Asingle
spermatogonia were PAX indicated that the Asingle popula-
tion is more heterogenous that some models propose, although
elegant studies previously suggested that only a subset of Asingle
spermatogonia function as true stem cells , . Our present
findings indicate that the fraction of Asingle spermatogonia that
are PAX is in the range of  to  . We speculate that
PAX spermatogonia sit at the top of the differentiation hier-
archy, further suggesting that there are other subsets of Asingle
spermatogonia — perhaps defined by currently unknown mark-
ers — that function as transit-amplifying intermediates prior
to differentiating to Apair spermatogonia Figure . However,
other models are possible , , necessitating future investiga-
tions to gain a complete understanding of the cellular hierar-
chies underlying stem cell maintenance and differentiation in
the mammalian testis.
Figure 9. Lineage tracing of PAX7+ spermatogonia following busulfan (20 mg/kg) treatment of Pax7-CreERT2;mT/mG males at 6 weeks of age. (A) Sche-
matic showing both busulfan lineage-tracing experiments. Testes were harvested 8 weeks after the last drug dose for each experiment. (B) Number of
clones 8 weeks after busulfan administration. Each point represents 1 testis from 1 animal; red bars denote means; P values were determined by unpaired
t test. (C) Clone size 8 weeks after tamoxifen administration. Red bars denote means; P values were determined by unpaired t test. (D) Composite image
of representative large clone from tam→bu experiment. Tubule borders are highlighted with dashed lines. Sp, elongate spermatids (arrows denote indi-
vidual cells or small groupings forming a “trail” of cells). (E) Cryosection of testis from tam→bu experiment showing germ cell clone spanning the entire
tubule. ST, seminiferous tubule; LC, Leydig cells. Scale bar: 200 μm (D); 25 μm (E).

The Journal of Clinical Investigation
ReseaRch aRticle
3940 jci.org Volume 124 Number 9 September 2014
Lack of genetic requirement for Pax in normal spermatogenesis in
the mouse. In preliminary genetic studies, we did not nd evidence
for a functional requirement for Pax in spermatogenesis. Germ
cell–specic Pax inactivation conrmed by the apparent lack
of PAX protein in germ cells did not result in male infertility or
have a discernible eect on spermatogenesis. It will be interesting
to study the eect of Pax inactivation in SSC cultures, which may
exhibit phenotypes not apparent in vivo. Challenging these Pax
cKO mice with busulfan, however, demonstrated that lack of PAX
delayed spermatogenic recovery, a nding that should be further
explored, particularly as the number of animals analyzed was rela-
tively small and busulfan was tested at only  concentration.
The lack of a functional requirement for fertility in the undis-
turbed testis — at least under laboratory conditions — may be
surprising given the phylogenetic conservation of PAX sper-
matogonia. On the other hand, several other useful conserved
and canonical stem cell markers, like Lgr gut and CD
hematopoiesis are, also dispensable for stem cell function in the
respective organs in which they serve . Although Lgr con-
ditional inactivation in intestinal epithelium yielded no apparent
phenotype, simultaneous inactivation of Lgr and Lgr which is
expressed more broadly than Lgr enhanced an intestinal crypt
loss phenotype observed with Lgr . It is similarly possible
Here, we took advantage of lineage tracing as a method to
explore the stem behavior of a novel population of spermatogo-
nia dened by PAX expression. We propose some criteria by
which stem cell lineage–tracing studies should be evaluated in
the context of the adult testis, in the addition to the requirement
for full-lineage maturation. First, we believe one important crite-
rion is that the labeled germ cell clones begin as single cells. For
example, lineage tracing initiated with a Cre driver expressed
in broad subsets of spermatogonia e.g., FOXO or PLZF sper-
matogonia would initiate labeling in Asingle–Aal spermatogonia,
including larger chains of spermatogonia, only a few of which
represent actual stem cells. Extending this logic to our present
study, it is possible that only a subset of PAX spermatogonia
function as stem cells, although the remarkable rarity of PAX
spermatogonia is one argument against this possibility. Another
criterion we believe should be considered is the long-term perdu-
rance of labeled clones. Such perdurance excludes the possibility
that the labeled cells represent transit-amplifying intermediates,
which would be diluted out and thus disappear with time. In the
lineage-tracing studies with the mT/mG reporter, we studied
clones for up to  weeks  days and observed no decrease in
clone numbers, while the duration of spermatogenesis in mice is
approximately  days .
Figure 10. Pax7 cKO in the male germline. For each analysis, n = 3 animals were evaluated per genotype. (A) Testes from floxed Pax7fl/fl control and Pax7
cKO (VC;Pax7–/fl) males at 6 months of age. No abnormalities or size dierences were noted. (B) Testis weight expressed as percent of total body weight
(6 months of age). (C) Fertility assays of 6 month-old males. Bars denote means. All floxed control and Pax7 cKO males were fertile and sired litters of
normal size. (D) Histological analyses at 6 months of age. No abnormalities in spermatogenesis or testis morphology were noted. (E) IHC analyses of
Pax7 cKO males at PD7. PAX7+ spermatogonia were abundant and present in most tubules in wild-type controls (arrow), but absent in Pax7 cKO tubules
(multiple sections were stained and examined for each), confirming the specificity of the PAX7 antibody in testis sections. GCNA shows the presence of
germ cells. Scale bars: 50 μm.

The Journal of Clinical Investigation ReseaRch aRticle
3941
jci.org Volume 124 Number 9 September 2014
also very sensitive to radiation-induced damage. Doses of . Gy
and higher are associated with an increased risk of infertility .
Remarkably, murine PAX spermatogonia proved resistant
to both radiotherapy and chemotherapy. They not only survived
the immediate aftermath of these genotoxic insults, but also rap-
idly expanded, forming clusters of PAX spermatogonia never
observed in normal, untreated mice. Lineage-tracing studies con-
rmed that PAX spermatogonia contribute to the restoration
of spermatogenesis. Future studies will be needed to more fully
dene the extent of the role of PAX spermatogonia as spermato-
genic recovery cells and to determine the relative contributions
to spermatogenic recovery of PAX spermatogonia versus other
spermatogonial subtypes. It is also interesting to consider the pos-
sibility that PAX spermatogonia might contribute to the recovery
of fertility in cancer patients, or that their failure to recover might
account for permanent sterilization after chemotherapy or radio-
therapy. If so, improved understanding of the biological pathways
regulating the behavior of PAX spermatogonia might someday
lead to strategies to protect the male germline in cancer patients. It
will also be interesting to explore the biological mechanisms that
render PAX spermatogonia resistant to genotoxic stresses.
The resistance of PAX spermatogonia to both radiation and
chemotherapy argues against models in which resistance is medi-
that other factors — perhaps other Pax genes — are functionally
redundant with Pax and compensate for its loss, although we
did not identify any Pax genes that were specically expressed in
SSCs in gene expression analyses. Furthermore, the oxed allele
used for this study Paxtm.Fan was recently found to give rise to
a hypomorphic mutation that expresses low levels of a truncated
PAX protein from an alternative ATG start site, and thus does not
appear to be a true biological or phenotypic null . In the more
recently described oxed allele Paxtm.Thbr, the transcriptional
start site and the rst  exons are oxed, preventing the generation
of any mRNA from the Pax locus after Cre-mediated recombina-
tion . Thus, even though PAX protein was greatly reduced
in the Pax cKO analysis we performed with the Paxtm.Fan allele
Figure E, it will be of interest to conduct future investigations
with the Paxtm.Thbr allele.
Implications for iatrogenic male infertility. Infertility is a common
and well-known complication of cancer treatment that profoundly
aects men and boys . Virtually all standard therapies e.g.,
cytotoxic chemotherapies and radiotherapy are highly toxic to the
male germline. The likelihood of infertility with chemotherapy is
drug-specic and dose-related. Alkylating agents pose the highest
risk of infertility, with platinum analogs, anthracyclines, and nitro-
soureas posing an intermediate level of risk . The germline is
Figure 11. PAX7+ spermatogonia are conserved in mammals. (A) Epitope mapping of anti-PAX7 mouse monoclonal antibody (generated against chicken
aa 320–523). Chicken and corresponding mouse polypeptide aa sequences were tiled as sequential 12-mers at 1-aa resolution. (B) PAX7 Western blot
(uncropped to show all visible bands in lanes shown). Addition of blocking peptide (22 aa) confirmed that the anti-PAX7 monoclonal antibody bound to the
QPQADFSISP epitope. C2C12, skeletal muscle myoblast cell line. Uterus was included as a negative control. Molecular weight markers denote 75, 50, and
37 kDa. (C) IHC of testes from 7 additional mammalian species, including juveniles for 2 species. PAX7+ spermatogonia (arrows) were rare and localized to
the basement membrane. Scale bar: 25 μm.

The Journal of Clinical Investigation
ReseaRch aRticle
3942 jci.org Volume 124 Number 9 September 2014
embryonic stem cell, embryonic gonad, and spermatogenic cell data
sets were downloaded from GEO accession nos. GSE, GSE,
and GSE . Probe sets were ranked based on signal
strength in SSCs as a proportion of that in the intact adult testis.
Mouse strains and procedures. Mice harboring the PaxCreERT 
B;-Pax tm.cre/ERTFan/J and Pax  B;-Paxtm.Fan/J alleles
as well as the RR FVB.SB-GtROSASortmSor/J, mT/mG
tdTomato/eGFP reporter GtROSASortmACTB–tdTomato,–EGFPLuo/J, and
nuclear tdTomato reporter B.Cg-GtROSASortmCAG–tdTomatoHze/J
alleles were purchased from Jackson Laboratories , . Busulfan
CAS no. , TCI America was dissolved in DMSO and admin-
istered as a single i.p. dose. EdU catalog no. C, Invitrogen was
dissolved in water and injected i.p.  mg/kg. Cyclophosphamide
was dissolved in PBS and administered at  mg/kg i.p. every  days
for  days. Tamoxifen catalog no. T, Sigma-Aldrich was dis-
solved at  mg/ml in  ethanol, then resuspended at  mg/ml
in corn oil.  mg tamoxifen was delivered i.p. to each adult mouse daily
for  days. Neonatal mice PD or earlier were injected i.p. with . mg
tamoxifen daily for  days. Whole-body irradiation was administered
single dose while the mice were restrained in acrylic boxes at a dose
rate of . Gy/min. No specic method for randomization for animal
studies was used; investigators were not blinded.
Transplantation procedure.  testes from  PaxCreE RT;tdTomato
PD donor mixed genetic background were enzymatically digest-
ed with dispase catalog no. , BD to obtain single cells that
were resuspended in DMEM with  FBS plus . Trypan Blue
.  μl of    cells/μl were transplanted by the eerent duct
method into testes of  KitW/KitW–v mice   weeks old, mixed genetic
background; stock no. , Jackson Laboratories. Testis lling per
blue dye was – in each testis. To deplete T cells and promote
engraftment,  μg anti-CD antigen catalog no. MAB, R&D Sys-
tems was injected i.p.  times every other day starting on the day of
transplantation. Testes were analyzed  weeks after transplantation.
Tissue processing, IHC, and immunouorescence IF. For IHC, tis-
sues were xed in  buered formalin overnight, embedded in
paran, and cut into -μm sections except for the serial analysis of
an entire testis and PAX cluster analyses, where -μm sections were
used, with indirect detection performed as described previously .
ated by active drug eux MDR or other transporters, as is the
case with several types of stem cells e.g., the “side population”
eect due to the eux of uorescent dyes , .
In closing, PAX spermatogonia represent a rare but func-
tionally important stem cell population in the healthy adult testis,
and also serve an important role in spermatogenic recovery fol-
lowing injury to the germline, such as occurs after chemotherapy
or radiotherapy. That PAX spermatogonia are rapidly cycling
and yet resistant to such stress is a notable aspect of their biology.
Methods
mRNA analysis and PAX discovery. RNA preparation, microarray
hybridization, normalization, quality control, and digital Northern
analysis was performed as described previously . Additional data
sets included in this analysis were intact PD testes and cultured SSCs
established as previously described , both from FVB/n mice. The
Table 1. The QPQADFSISP PAX7 epitope is perfectly conserved in
mammals, but not zebrafish or Drosophila
Species Sequence
Gallus gallus (chicken) 407 SILSNPSGVPPQPQADFSISPLHGGLDTTNSI
Mus musculus (mouse) 386 SILSNPSAVPPQPQADFSISPLHGGLDSASSI
Rattus norvegicus (rat) 386 SILSNPSAVPPQPQADFSISPLHGGLDSASSI
Canis familiaris (dog) 519 SILSNPSAVPPQPQADFSISPLHGGLDSATSI
Felis catus (cat) 386 SILSNPSAVPPQPQADFSISPLHGGLDSATSI
Homo sapiens (human) 388 SILGNPSAVPPQPQADFSISPLHGGLDSATSI
Pan troglodytes (chimpanzee) 388 SILGNPSAVPPQPQADFSISPLHGGLDSATSI
Macaca mulatta (macaque) 388 SILSNPSAVPPQPQADFSISPLHGGLDSATSI
Papio anubis (baboon) 387 SILSNPSAVPPQPQADFSISPLHGGLDSATSI
Bos taurus (bull) 388 SILSNPSAVPPQPQADFSISPLHGGLDSATSI
Ornithorhynchus anatinus (platypus) 355 SILSNPSGVPPQPQADFSISPLHGGLDTTNSI
Danio rerio (zebrafish), Pax7a 389 SILSNPSAVPSQPQHDFSISPLHGGLEASNPI
Danio rerio (zebrafish), Pax7b 392 SILSNPSAVAPQPQHEFSISPLHSSLEASNPI
Drosophila melanogaster (fruit fly), gsb-n 337 AQHGFPGGFAQPGHFGSQNYYHQDYSKLTID
Figure 12. Models of stemness in mouse spermatogenesis. Spermatogonial subsets proposed
as the bona fide stem cells are shown above each of the 3 models. In the classic Asingle model (As),
Asingle spermatogonia are homogeneous and share stem cell identity (green), having the capacity
for self-maintenance (circular arrows; refs. 3, 4, 66). More recently, models have been proposed
arguing for greater plasticity among undierentiated (Asingle→Aal16) spermatogonia, with chain
fragmentation representing one possible mechanism by which stemness is maintained or regen-
erated (5). Although fragmentation has been shown to occur in vivo, its contributions to stem
cell maintenance under normal conditions or after chemotherapy/radiation have not been for-
mally established. Our findings that only a subset of Asingle spermatogonia expressed PAX7 and
that these spermatogonia functioned as stem cells suggests a new Asingle subset model, whereby
PAX7+ spermatogonia are self-maintaining and may sit atop the hierarchy of spermatogenic dif-
ferentiation. That Asingle spermatogonia were heterogeneous and that only a subset functioned
as stem cells was also suggested by previous studies (10, 67). If so, then this would suggest that
some subset of Asingle spermatogonia represent transit-amplifying (TA) intermediates. The num-
ber of such transit-amplifying steps between PAX7+ Asingle and Apair spermatogonia is unknown. It
will be interesting to determine whether ID4 and ERBB3, expressed in Asingle spermatogonia, are
expressed in overlapping or nonoverlapping subsets of spermatogonia relative to PAX7 (9, 49,
50). Other models are possible, such as ones combining dierent aspects of these models (i.e.,
fragmentation with the presence of PAX7+ spermatogonia, if fragmentation is confirmed as a
functionally significant biological process).

The Journal of Clinical Investigation ReseaRch aRticle
3943
jci.org Volume 124 Number 9 September 2014
sequences were acetyl-capped at the N terminus. PAX antibody
 μg/ml was hybridized to the microarray in  PBS . for  hours
at °C, then washed in  PBS with . Tween- and . Tri-
tonX, pH .. Detection was performed with an anti-mouse IgG/
Alexa Fluor  conjugate  ng/ml in binding buer for  hour at
°C. The array was scanned at  nm on an Axon GenePix B
Microarray Scanner. Sequences were obtained from NCBI Homolo-
Gene  and accession nos. XP_., XP_.,
and XP_. and manually aligned. Gonadal samples were
described previously .
Western blotting. For the blocking experiment, a -residue block-
ing peptide PSAVPPQPQADFSISPLHGGLD, spanning the -aa
epitope, was synthesized. The peptide . μg/μl and antibody
 ng/μl were incubated at °C for  hours in  μl PBS, then
centrifuged at °C for  minutes , g. The supernatants with
and without blocking peptide  μl were added to . ml  milk
 TBST and applied to the Western blots.
Statistics. Statistics were calculated using GraphPad software.
Error bars in all gures indicate SEM for at least  animals. For Fisher
exact tests, -tailed comparisons were performed to calculate P val-
ues. A P value less than . was considered signicant.
Study approval. This study was approved by the UT Southwestern
Institutional Animal Care and Use Committee.
Acknowledgments
This work was supported by the David M. Crowley Foundation,
NCI grants RCA and HD, and the State of Tex-
as Norman Hackerman Advanced Research Program -
. G.M. Aloisio was supported by a UT Southwestern
Cecil H. and Ida Green Center for Reproductive Biological Sci-
ences fellowship. The authors acknowledge the assistance of Kate
Luby-Phelps and the UT Southwestern Live Cell Imaging Facility,
a shared resource of the Harold C. Simmons Cancer Center, sup-
ported in part by NCI Cancer Center Support Grant PCA.
We thank Alexandra Ghaben, Rene Galindo, Lee Kraus, Andrew
Zinn, Thomas Braun, Chen-Ming Fang, Michael Rudnicki, and
Sean Morrison for helpful discussions.
Address correspondence to: Diego H. Castrillon, UT Southwest-
ern Medical Center, Department of Pathology, NB. 
Harry Hines Boulevard, Dallas, Texas , USA. Phone:
..; E-mail: diego.castrillon@utsouthwestern.edu.
For whole-mount IF, seminiferous tubules were mechanically dissoci-
ated in PBS on ice and xed overnight in  paraformaldehyde PFA.
Tubules were dehydrated in a series of methanol washes and stored
at –°C. To rehydrate and permeabilize, tubules were put through
a series of washes with methanol and PBS plus . Tween-, fol-
lowed by incubation with . NP for IF of nuclear proteins.
Tubules were blocked in  BSA and PBS catalog no. , Thermo
Scientic Blocker for  hours, then in MOM block catalog no. MKB-
, Vector Labs, and primary antibody was added in . BSA and
PBS with . sodium azide, followed by incubation at °C over-
night. Tubules were washed  times for  minutes each in PBS at
RT. Secondary antibody Alexa Fluor  anti-rabbit, Alexa Fluor 
anti-mouse, Alexa Fluor  anti-goat; catalog nos. A, A,
and A, respectively, Invitrogen was added at :, in .
BSA and PBS for  hours at RT followed by DAPI staining :, in
PBS; catalog no. , Pierce. Tubules were placed on glass slides
and mounted in Vectashield Vector Laboratories. For visualization of
mT/mG clones in tissue sections, testes were embedded in OCT; sec-
tioned; xed for  minutes in  formalin,  picric acid,  and
sucrose at °C; and then cryosectioned. Microscopy was performed
with a Leica TCS SP confocal microscope.
Antibodies for IF and IHC. Antibodies and titers used were as
follows: PAX : for IHC, : for IF; Developmental Studies
Hybridoma Bank, FOXO : for IHC, : for IF; catalog no.
, Cell Signaling Technology, KIT : for IHC, : for IF;
catalog no. , Cell Signaling Technology, PLZF :, for
IHC; catalog no. AF, R&D Systems, caspase- : for IF;
catalog no. , BD Biosciences — Pharmingen, DSred detects
tdTomato; : for IF; catalog no. , Clontech, GCNA :
for IHC; provided by G.C. Enders, University of Kansas, Kansas City,
Kansas, USA; ref. , RET : for IHC; catalog no. , IBL
America, GFRα : for IF; catalog no. AF, R&D Systems.
X-gal staining. Whole-mount X-gal staining was performed by
manually dissociating tubules, xing in  PFA and PBS for  min-
utes at RT, and staining as previously described  for  hours to
overnight, followed by rexing in  PFA and PBS overnight.
Epitope mapping and phylogenetic analyses. An arrayed microchip
was designed by LC Sciences as described previously . The chip
included -mer tiling peptides sequences with -aa resolution corre-
sponding to the entire chicken polypeptide immunogen aa ;
Genbank NP_.. The corresponding mouse aa sequence was
also arrayed on the same microchip, also at -aa resolution. Peptide
. Oatley JM, Brinster RL. Regulation of spermato-
gonial stem cell self-renewal in mammals. Annu
Rev Cell Dev Biol. ;:.
. de Rooij DG. Proliferation and differentiation
of spermatogonial stem cells. Reproduction.
;:.
. de Rooij DG, Russell LD. All you wanted to know
about spermatogonia but were afraid to ask.
J Androl. ;:.
. Oakberg EF. Spermatogonial stem-cell renewal in
the mouse. Anat Rec. ;:.
. Nakagawa T, Sharma M, Nabeshima Y, Braun RE,
Yoshida S. Functional hierarchy and reversibility
within the murine spermatogenic stem cell com-
partment. Science. ; :.
. Jan SZ, Hamer G, Repping S, de Rooij DG, van
Pelt AM, Vormer TL. Molecular control of
rodent spermatogenesis. Biochim Biophys Acta.
;:.
. Yang QE, Oatley JM. Spermatogonial stem cell
functions in physiological and pathological con-
ditions. Curr Top Dev Biol. ;:.
. Nagano MC, Yeh JR. The identity and fate deci-
sion control of spermatogonial stem cells: where
is the point of no return? Curr Top Dev Biol.
;:.
. Oatley MJ, Kaucher AV, Racicot KE, Oatley
JM. Inhibitor of DNA binding  is expressed
selectively by single spermatogonia in the male
germline and regulates the self-renewal of
spermatogonial stem cells in mice. Biol Reprod.
;:.
. Nagano MC. Homing eficiency and prolifera-
tion kinetics of male germ line stem cells fol-
lowing transplantation in mice. Biol Reprod.
;:.
. Brinster RL, Avarbock MR. Germline transmis-
sion of donor haplotype following spermatogo-
nial transplantation. Proc Natl Acad Sci U S A.
;:.
. Kanatsu-Shinohara M, Shinohara T. Spermato-
gonial stem cell self-renewal and development.
Annu Rev Cell Dev Biol. ;:.
. Kubota H, Avarbock MR, Brinster RL. Culture
conditions and single growth factors affect fate
determination of mouse spermatogonial stem
cells. Biol Reprod. ;: .
. Shinohara T, Avarbock MR, Brinster RL. Beta-

The Journal of Clinical Investigation
ReseaRch aRticle
3944 jci.org Volume 124 Number 9 September 2014
and alpha-integrin are surface markers on
mouse spermatogonial stem cells. Proc Natl Acad
Sci U S A. ;:.
. Phillips BT, Gassei K, Orwig KE. Spermato-
gonial stem cell regulation and spermato-
genesis. Philos Trans R Soc Lond B Biol Sci.
;:.
. Zhang Z, Shao S, Meistrich ML. The radiation-
induced block in spermatogonial differentiation
is due to damage to the somatic environment, not
the germ cells. J Cell Physiol.  ;:.
. Yoshida S. Elucidating the identity and behavior
of spermatogenic stem cells in the mouse testis.
Reproduction. ;:.
. McLean DJ. Spermatogonial stem cell transplan-
tation and testicular function. Cell Tissue Res.
;:.
. Brinster RL. Germline stem cell transplantation and
transgenesis. Science. ;:.
. Kanatsu-Shinohara M, et al. Long-term prolif-
eration in culture and germline transmission of
mouse male germline stem cells. Biol Reprod.
;:.
. Kanatsu-Shinohara M, Inoue K, Ogonuki N,
Morimoto H, Ogura A, Shinohara T. Serum- and
feeder-free culture of mouse germline stem cells.
Biol Reprod. ;:.
 . Meng X, et al. Regulation of cell fate decision of
undifferentiated spermatogonia by GDNF.
Science. ; :.
. Relaix F, Zammit PS. Satellite cells are essential
for skeletal muscle regeneration: the cell on
the edge returns centre stage. Development.
;:.
. Gallardo TD, et al. Genomewide discovery and
classiication of candidate ovarian fertility genes
in the mouse. Genetics. ;: .
. Yin H, Price F, Rudnicki MA. Satellite cells
and the muscle stem cell niche. Physiol Rev.
;:.
. Tanaka SS, et al. The mouse homolog of Dro-
sophila Vasa is required for the development of
male germ cells. Genes Dev. ;:.
 . Goertz MJ, Wu Z, Gallardo TD, Hamra FK,
Castrillon DH. Foxo is required in mouse sper-
matogonial stem cells for their maintenance and
the initiation of spermatogenesis. J Clin Invest.
;:.
 . Buaas FW, et al. Plzf is required in adult male
germ cells for stem cell self-renewal. Nat Genet.
;:.
. Naughton CK, Jain S, Strickland AM, Gupta
A, Milbrandt J. Glial cell-line derived neuro-
trophic factor-mediated RET signaling regulates
spermatogonial stem cell fate. Biol Reprod.
;:.
. Vergouwen RP, Huiskamp R, Bas RJ, Roepers-
Gajadien HL, Davids JA, de Rooij DG. Postnatal
development of testicular cell populations in
mice. J Reprod Fertil. ;:.
 . Lepper C, Fan CM. Inducible lineage tracing of
Pax-descendant cells reveals embryonic origin of
adult satellite cells. Genesis. ; :.
. Lepper C, Conway SJ, Fan CM. Adult satel-
lite cells and embryonic muscle progenitors
have distinct genetic requirements. Nature.
;: .
. Soriano P. Generalized lacZ expression with
the ROSA Cre reporter strain. Nat Genet.
;:.
 . Franca LR, Avelar GF, Almeida FF. Spermatogen-
esis and sperm transit through the epididymis in
mammals with emphasis on pigs. Theriogenolog.
;:.
. Muzumdar MD, Tasic B, Miyamichi K, Li L, Luo
L. A global double-fluorescent Cre reporter
mouse. Genesis. ;:.
. Madisen L, et al. A robust and high-throughput Cre
reporting and characterization system for the whole
mouse brain. Nat Neurosci. ;:.
 . Bucci LR, Meistrich ML. Effects of busulfan on
murine spermatogenesis: cytotoxicity, sterility,
sperm abnormalities, and dominant lethal muta-
tions. Mutat Res. ;:.
. van Keulen CJ, de Rooij DG. Spermatogenetic
clones developing from repopulating stem cells
surviving a high dose of an alkylating agent. Cell
Tissue Kinet. ;:.
. Enders GC, May JJ. Developmentally regulated
expression of a mouse germ cell nuclear antigen
examined from embryonic day  to adult in male
and female mice. Dev Biol. ;:.
. Lu CC, Meistrich ML. Cytotoxic effects of che-
motherapeutic drugs on mouse testis cells. Can-
cer Res. ;: .
. Goldfarb S, Mulhall J, Nelson C, Kelvin J, Dickler
M, Carter J. Sexual and reproductive health in can-
cer survivors. Semin Oncol. ;:.
. Withers HR, Hunter N, Barkley HT Jr, Reid BO.
Radiation survival and regeneration characteris-
tics of spermatogenic stem cells of mouse testis.
Radiat Res. ;:.
. Gallardo T, Shirley L, John GB, Castrillon
DH. Generation of a germ cell-speciic mouse
transgenic Cre line, Vasa-Cre. Genesis. ;
:.
. Castrillon DH, et al. Toward a molecular genetic
analysis of spermatogenesis in Drosophila
melanogaster: characterization of male-sterile
mutants generated by single P element mutagen-
esis. Genetics. ;:.
 . Tarnawa ED, Baker MD, Aloisio GM, Carr BR ,
Castrillon DH. Gonadal expression of foxo, but
not foxo, is conserved in diverse mammalian
species. Biol Reprod. ;:.
. Roosen-Runge EC. Comparative aspects of sper-
matogenesis. Biol Reprod. ;:.
. Fuller MT. Genetic control of cell proliferation
and differentiation in Drosophila spermatogen-
esis. Semin Cell Dev Biol. ;:.
. Kawakami A, Kimura-Kawakami M, Nomura T,
Fujisawa H. Distributions of PAX and PAX
proteins suggest their involvement in both early
and late phases of chick brain development. Mech
Dev. ;:.
. Abid SN, et al. A-single spermatogonia heteroge-
neity and cell cycles synchronize with rat semi-
niferous epithelium stages VIIIIX. Biol Reprod.
;:.
. Chan F, et al. Functional and molecular features
of the Id germline stem cell population in
mouse testes. Genes Dev. ;: .
. Snipper t HJ, Clevers H. Tracking adult stem cells.
EMBO Rep. ;: .
. Rojas-Ríos P, González-Reyes A. Concise review:
The plasticity of stem cell niches: a general prop-
erty behind tissue homeostasis and repair. Stem
Cells. ;:.
. Barker N, et al. Identiication of stem cells in
small intestine and colon by marker gene Lgr.
Nature. ;:.
. Kiel MJ, Yilmaz OH, Iwashita T, Yilmaz OH,
Terhorst C, Morrison SJ. SLAM family receptors
distinguish hematopoietic stem and progenitor
cells and reveal endothelial niches for stem cells.
Cell. ; :.
. de Lau W, et al. Lgr homologues associate with
Wnt receptors and mediate R-spondin signalling.
Nature. ;:.
. Wang N, et al. The cell surface receptor SLAM
controls T cell and macrophage functions. J Exp
Med. ;: .
. Gunther S, Kim J, Kostin S, Lepper C, Fan CM,
Braun T. Myf-positive satellite cells contribute
to Pax-dependent long-term maintenance
of adult muscle stem cells. Cell stem cell.
;:.
. Howell SJ, Shalet SM. Spermatogenesis after
cancer treatment: damage and recovery. J Natl
Cancer Inst Monogr. ;:.
. Drach J, Zhao S, Drach D, Korbling M, Engel
H, Andreeff M. Expression of MDR by nor-
mal bone marrow cells and its implication for
leukemic hematopoiesis. Leuk Lymphoma.
;:.
. Petriz J. Flow cytometry of the side population
SP. Curr Protoc Cyt om. ;
Chapter :Unit..
. Falciatori I, Lillard-Wetherell K, Wu Z, Hamra FK,
Garbers DL. Deriving mouse spermatogonial stem
cell lines. Methods Mol Biol. ;:.
. Kurimoto K, et al. An improved single-cell cDNA
ampliication method for eficient high-density
oligonucleotide microarray analysis. Nucleic
Acids Res. ;:e.
. Small CL, Shima JE, Uzumcu M, Skinner MK,
Griswold MD. Proiling gene expression dur-
ing the differentiation and development of
the murine embryonic gonad. Biol Reprod.
;:.
. Namekawa SH, et al. Postmeiotic sex chro-
matin in the male germline of mice. Curr Biol.
;:.
 . Pellois JP, Zhou X, Srivannavit O, Zhou T, Gulari
E, Gao X. Individually addressable parallel
peptide synthesis on microchips. Nat Biotechnol.
;: .
. Huckins C. Cell cycle properties of differentiating
spermatogonia in adult Sprague-Dawley rats. Cell
Tissue Kinet. ;:.
. Suzuki H, Sada A, Yoshida S, Saga Y. The hetero-
geneity of spermatogonia is revealed by their
topology and expression of marker proteins
including the germ cell-speciic proteins Nanos
and Nanos. Dev Biol. ;: .