Molecular Human Reproduction, Vol.22, No.4 pp. 229–239, 2016
Advanced Access publication on January 20, 2016 doi:10.1093/molehr/gaw006
ORIGINAL RESEARCH
Isolation and gene expression analysis
of single potential human
spermatogonial stem cells
K. von Kopylow 1,*, W. Schulze 1,2, A. Salzbrunn 1, and A.-N. Spiess 1
1
Department of Andrology, University Hospital Hamburg-Eppendorf, 20246 Hamburg, Germany 2MVZ Fertility Center Hamburg GmbH,
amedes-group, 20095 Hamburg, Germany
*Correspondence address. Department of Andrology, University Hospital Hamburg-Eppendorf, Martinistr. 52, 20246 Hamburg, Germany.
Tel: +49-40-7410-51583; Fax: +49-40-7410-51554; E-mail: [email protected]
Submitted on September 25, 2015; resubmitted on January 8, 2016; accepted on January 15, 2016
study hypothesis: It is possible to isolate pure populations of single potential human spermatogonial stem cells without somatic
contamination for down-stream applications, for example cell culture and gene expression analysis.
study finding: We isolated pure populations of single potential human spermatogonial stem cells (hSSC) without contaminating somatic
cells and analyzed gene expression of these cells via single-cell real-time RT– PCR.
what is known already: The isolation of a pure hSSC fraction could enable clinical applications such as fertility preservation for prepubertal boys and in vitro-spermatogenesis. By utilizing largely nonspecific markers for the isolation of spermatogonia (SPG) and hSSC, previously
published cell selection methods are not able to deliver pure target cell populations without contamination by testicular somatic cells. However,
uniform cell populations free of somatic cells are necessary to guarantee defined growth conditions in cell culture experiments and to prevent
unintended stem cell differentiation. Fibroblast growth factor receptor 3 (FGFR3) is a cell surface protein of human undifferentiated A-type
SPG and a promising candidate marker for hSSC. It is exclusively expressed in small, non-proliferating subgroups of this spermatogonial cell
type together with the pluripotency-associated protein and spermatogonial nuclear marker undifferentiated embryonic cell transcription
factor 1 (UTF1).
study design, samples/materials, methods: We specifically selected the FGFR3-positive spermatogonial subpopulation
from two 30 mg biopsies per patient from a total of 37 patients with full spermatogenesis and three patients with meiotic arrest. We then
employed cell selection with magnetic beads in combination with a fluorescence-activated cell sorter antibody directed against human FGFR3
to tag and visually identify human FGFR3-positive spermatogonia. Positively selected and bead-labeled cells were subsequently picked with a
micromanipulator. Analysis of the isolated cells was carried out by single-cell real-time RT–PCR, real-time RT –PCR, immunocytochemistry
and live/dead staining.
main results and the role of chance: Single-cell real-time RT–PCR and real-time RT– PCR of pooled cells indicate that beadlabeled single cells express FGFR3 with high heterogeneity at the mRNA level, while bead-unlabeled cells lack FGFR3 mRNA. Furthermore, isolated
cells exhibit strong immunocytochemical staining for the stem cell factor UTF1 and are viable.
limitations, reasons for caution: The cell population isolated in this study has to be tested for their potential stem cell
characteristics via xenotransplantation. Due to the small amount of the isolated cells, propagation by cell culture will be essential. Other potential
hSSC without FGFR3 surface expression will not be captured with the provided experimental design.
wider implications of the findings: The technical approach as developed in this work could encourage the scientific community
to test other established or novel hSSC markers on single SPG that present with potential stem cell-like features.
study funding and competing interest(s): The project was funded by the DFG Research Unit FOR1041 Germ cell potential
(SCH 587/3-2) and DFG grants to K.v.K. (KO 4769/2-1) and A.-N.S. (SP 721/4-1). The authors declare no competing interests.
Key words: fibroblast growth factor receptor 3 / spermatogonia / spermatogonial stem cells / human spermatogonial stem cells / cell isolation /
Dynabeads / UTF1 / single-cell real-time RT –PCR / immunocytochemistry
& The Author 2016. Published by Oxford University Press on behalf of the European Society of Human Reproduction and Embryology. All rights reserved.
For Permissions, please email: [email protected]
230
von Kopylow et al.
Introduction
Materials and Methods
Spermatogonial stem cells (SSCs) represent the basis of nearly life-long
spermatogenesis and encompass a small germ cell population with selfrenewal potential. Their surgical removal from the testis prior to nontesticular anti-cancer treatment of prepubertal boys may prevent toxic
injury of healthy testis cells. Re-transplantation after convalescence
could preserve children’s fertility, enabling them to father their own
genetic children (Faes et al., 2013; Goossens et al., 2013). Furthermore,
SSCs could facilitate in vitro-spermatogenesis in non-obstructive azoospermia (Yang et al., 2014), be used for fertility treatment via autotransplantation of cell culture-propagated SSCs into men with azoospermia
factor c gene (AZFc) deletions (Nickkholgh et al., 2015), serve as a cell
source in regenerative medicine, provide useful targets for male contraception (Guo et al., 2014) and may foster investigations on the mechanisms of spermatogenesis.
Current strategies for the isolation of human spermatogonia (SPG)
and human SSCs (hSSC) are differential plating, magnetic-activated cell
sorting (MACS), using a fluorescence-activated cell sorter and
STA-PUT gravity sedimentation (Guo et al., 2014; Liu et al., 2015).
None of these methods alone or in combination is suitable for delivering
pure SPG/SSC populations free from contaminating testicular somatic or
differentiating germ cells, while being feasible for small biopsies (He et al.,
2010; Guo et al., 2014; Valli et al., 2014). However, contamination-free
hSSC cell isolates are absolutely essential for defined growth conditions
in cell culture experiments (Kanatsu-Shinohara and Shinohara, 2013) and
for preventing contamination of autograft transplants with cancer cells,
e.g. after hematological cancer treatment (Jahnukainen et al., 2015).
To address this problem, we developed a method for obtaining a pure
and somatic contamination-free SPG cell population with potential
stem cell-like features from two 30 mg testis biopsies per patient. This
is accomplished by using fibroblast growth factor receptor 3 (FGFR3,
JTK4, CD333; Eswarakumar et al., 2005; Lanner and Rossant, 2010) as
a SPG-specific cell surface marker (Juul et al., 2007; von Kopylow et al.,
2010). Within the human testis, FGFR3 expression is restricted to a
small subpopulation of non-proliferating, non-differentiating A-type
SPG, organized only in small clusters of mostly two or four cells (von
Kopylow et al., 2012a, b). These SPG co-express the undifferentiated
embryonic cell transcription factor 1 (UTF1; von Kopylow et al.,
2012a, b), a pluripotency-associated protein (Okuda et al., 1998; Kooistra et al., 2009) and established human SPG marker (Kristensen et al.,
2008; von Kopylow et al., 2010; Valli et al., 2014). UTF1 is expressed
in many stem cell types (Guan et al., 2006). FGFR and fibroblast
growth factor (FGF) expression is crucial for maintenance of the male
reproductive system by supporting germ cell proliferation, survival,
growth arrest, differentiation, migration and apoptosis (L’Hôte and
Knowles, 2005; Cotton et al., 2008; Lanner and Rossant, 2010). Investigations on seminiferous tubules of elderly men with somatic FGFR3
gain-of-function mutations in SPG (paternal age-effect mutations) uncovered an increased co-expression of FGFR3 and the SSC self-renewal
marker, phosphorylated AKT (pAKT; Lim et al., 2012; Maher et al.,
2014). These findings point to FGFR3 as a candidate marker for the enrichment and isolation of undifferentiated human SPG, including SSCs.
In this work, we show that SPG obtained by our developed isolation
protocol co-express the stem cell factor UTF1 and are equipped with
a moderate but highly variable number of FGFR3 transcripts, as demonstrated by single-cell real-time RT–PCR (qPCR).
Patients, testicular biopsies and
ethical approval
Testis biopsies were surgically extracted in parallel with therapeutic testicular
sperm extraction (TESE) for diagnostic purposes, as described previously
(Jezek et al., 1998; Schulze et al., 1999; Feig et al., 2007), between July
2013 and September 2015. Informed and written consent and Ethic Committee Approval was obtained (WF-007/11 and WF-005/13) prior to the
study, which was conducted in accordance with the ethical principles laid
down in the 1964 Declaration of Helsinki and its later amendments. In
total, testis biopsies from 37 normogonadotropic patients with full spermatogenesis presenting obstructive azoospermia and three patients with meiotic
arrest were used for magnetic cell isolation and subsequent micromanipulator picking of bead-labeled cells after optimization of the isolation protocol.
A total of 26 patients were used for experiments with at least one of the
described methods (cell counting of bead+ SPG: 7; qPCR: 14; immunocytochemistry: 7; live/dead staining: 4), while another 14 different patients were
used for the evaluation of the isolation/bead-labeling protocol. Two of the
three patients with meiotic arrest were used for cell counting of bead+
SPG and single-cell qPCR (see legends to Supplementary Figs S2 and S5).
The sample from one patient with spermatogenic arrest at the level of
spermatocytes was employed only for cell isolation.
Preparation of cell suspensions from
testis biopsies
Two small fragments of testis tissue (30 mg each per left and right testis)
were collected separately in 1 ml Dulbecco’s modified Eagle’s medium
(DMEM, Life Technologies Gibco, Paisley, UK) and stored at 378C. Enzymatic
digestion with Collagenase D (1 mg/ml; Roche, Mannheim, Germany) was
performed at 378C for 30 min with shaking every 10 min. The reaction
was stopped by addition of 600 ml 378C pre-warmed Hank’s balanced salt
solution (HBSS, Life Technologies Gibco, Paisley, UK) to each collagenasedigested biopsy and 3× 6 min centrifugation at 300g (room temperature).
The digestion state was microscopically visualized. The biopsy was transferred to a preheated glass Petri dish containing 1 ml DMEM at 378C and
minced using microscopy scissors. Subsequent up and down pipetting
(10×) and filtration through a Falconw 70 mm-cell strainer (BD Biosciences,
Franklin Lakes, NJ, USA) generated aggregate-free cell solutions.
Magnetic cell isolation (positive selection)
and selective cell separation by
micromanipulation
Volume restriction to 500 ml was achieved by centrifugation at 300g for 10 min
prior to FGFR3 antibody incubation (dilution 1:200; anti-hFGF R3 mouse
monoclonal immunoglobulin IgG #MAB766, R&D Systems, Minneapolis,
MN, USA) at 378C and 5% CO2 for 1 h. The cell culture dish was manually
shaken after 30 min during incubation with the first antibody. Cell isolation
with Dynabeadsw (Dynabeadsw goat anti-mouse IgG #110.33, Thermo
Fisher Scientific, Oslo, Norway) was carried out according to the manufacturer’s instructions (see manual for Dynabeadsw Goat anti-mouse IgG:
https://www.thermofisher.com/order/catalog/product/11033; Manuals &
Protocols; Chapter ‘Isolate Cells – Indirect Technique’; Steps 3–10), but
with modified volumes, incubation times and using DMEM as resuspension
medium: Cells were washed in 750 ml Isolation Buffer (for recipe see Dynabeadw manual) via centrifugation for 16 min (Step 3), re-suspended in
500 ml DMEM (Step 4) and incubated for 10 min at 378C for cell recovery. Prewashed Dynabeadsw (5 ml) were transferred into a 2 ml-round bottom tube
containing the cell suspension (Step 5). Tilting and rotation was carried out
Single spermatogonial stem cell isolation and analysis
231
Figure 1 Isolation of bead-labeled cells. (A – C) Overview images of the positive cell fraction after magnetic cell isolation, with bead-labeled (bead+) cells
(red arrows) and bead-unlabeled (bead2) cells (black arrows) from three different patients. (D– F) Micromanipulator picking of a bead+ single cell, a
doublet and a quadruplet, respectively, from three other patients. Examples for bead+ single cells (H, I), doublets (J, K) and quadruplets (L, M). Scale
bars: A – F: 100 mm, H – M: 10 mm.
for 80 min at 48C (Step 6). After incubation with the magnetic beads, the cell
suspension was diluted with 1100 ml Isolation Buffer (Step 7) to minimize
nonspecific bead binding. To avoid significant cell loss from minute amounts
of material due to washing procedures, the unwashed positive cell fraction
(plus 1200 ml Isolation Buffer) was utilized for the micromanipulation procedure after placing the tube in the magnet and discarding the supernatant (Steps
8–10). Under a light-microscope (Axiovert 100; Achrostigmat 10×/0.25 objective; Carl Zeiss AG, Oberkochen, Germany), visually selected bead+ germ
cells (Fig. 1) were consecutively picked from 300 ml fractions with a micromanipulator capillary (VACU Tip FCH #5175240.006, Eppendorf AG, Hamburg,
Germany; Fig. 1D–F) using a PatchMan NP2 micromanipulator combined with
a Cell Tram Vario microinjector (Eppendorf AG, Hamburg, Germany). Each
cell considered for selection was captured with an Axiocam ICc3 digital
camera (Carl Zeiss AG, Oberkochen, Germany) prior to picking. Cells with
only loosely attached beads were not included in the bead+ cell fraction.
Very rarely (not more than two/patient), morphological structures similar
to Sertoli cell nuclei were surrounded by Dynabeads. In this case, cells
were also not selected for micromanipulation. Counting of bead+ cells was
carried out for seven patients.
Immunocytochemical staining
Bead+, micromanipulator-picked cells from the positive cell fraction were
transferred in a drop of 20 ml isolation buffer [1× Dulbecco’s phosphate-
buffered saline (DPBS, Life Technologies Gibco, Paisley, UK) containing
0.1% bovine serum albumin (PAA Laboratories GmbH, Pasching, Austria)
and 2 mM EDTA, pH ¼ 7.4 (Sigma-Aldrich, St. Louis, MO, USA)] to polylysine coated microscope slides (Menzel-Gläser Thermo Scientific, Braunschweig, Germany). For fixation, cells on the slides were covered with
Bouin’s Solution (Sigma-Aldrich, St. Louis, MO, USA) and left at room temperature overnight until dry. Slides were rinsed with distilled water for 30 min
and with 1× Tris-buffered saline (TBS) for 5 min, permeabilized with 1 mM
Digitonin (Sigma-Aldrich, St. Louis, MO, USA) in 1× TBS for 10 min and
rinsed with 1× TBS for 5 min again. Endogenous peroxidases were
blocked with DAKO Dual Endogenous Block (DAKO, Carpinteria, CA,
USA) for 10 min. After washing with 1× TBS and permeabilization with
1× TBS plus 1 mM Digitonin as described above, nonspecific antibody
binding with 1× TBS plus 2% normal goat serum (Sigma-Aldrich, St. Louis,
MO, USA) for 1 h at room temperature was followed by an incubation
with UTF1 antibody (dilution 1:200; #MAB4337, Millipore, Temecula, CA,
USA) at 48C overnight. Incubations with omission of the primary antibody
as well as replacement of the primary mouse antibody against UTF1 with
normal mouse IgG (#SC-2025, Santa Cruz Biotechnology, Inc., Dallas, TX,
USA) at the same concentration served as negative controls. Slides were
rinsed and permeabilized once more and incubated with DAKO Envision+
System- HRP Labelled Polymer Anti-mouse (DAKO, Carpinteria, CA, USA)
for 30 min at room temperature continued by further washing and permeabilization. All incubation steps were conducted in a humified chamber.
232
The peroxidase activity was developed using the Metal Enhanced DAB Substrate Kit (Thermo Scientific, Rockford, IL, USA). After mounting, cells were
visualized on a Zeiss Axiovert 100 (Carl Zeiss AG, Oberkochen, Germany).
Live/Dead Cell Viability Assay
The Live/Deadw Cell Viability Assay (Invitrogen, Eugene, OR, USA) was
carried out according to the manufacturer’s instruction following an overnight incubation at 358C with a CO2 level of 5% for cell recovery. Sample
evaluation and image capturing was performed using a Biozero fluorescence
microscope (Keyence, Osaka, Japan).
Single-cell qPCR/qPCR
Micromanipulator-picked bead+ or bead2 single cells, or alternatively pools
of 7 – 11 bead+ or bead2 cells, were transferred into a 0.5 ml Eppendorf
DNA LoBind Tubew (Eppendorf AG, Hamburg, Germany) and immediately
frozen at minus 208C. After thawing, 10 ml DNA-free water (Qiagen, Venlo,
Netherlands) were added before the cells were heat-lysed (2 min at 808C or
4 min at 908C as protocol optimization). Alternatively, ultrasonic cell lysis
was performed (2 min). An initial RNA unfolding step (10 min 708C plus
5 min 258C) was conducted prior to cDNA synthesis due to difficulties in
amplification of the low copy number of FGFR3 transcripts in SPG as a consequence of a very high GC-content in the 3′ untranslated region (up to 84%; see
Supplementary Fig. S8, and Perez-Castro et al.,1997). Second strand synthesis was carried out using SuperScriptTM III First-Strand Synthesis SuperMix for
qRT – PCR (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s
instruction, but without working on ice and skipping the ‘chilling on ice’ reaction termination step. Instead, for termination of cDNA synthesis, samples
were cooled down to room temperature. All of these steps should be conducted as fast as possible to minimize RNA degradation. Subsequently,
2 ml undiluted cDNA were used for each qRT – PCR reaction which was performed on a LightCyclerTM 1.0 (Roche, Basel, Switzerland) employing SYBRw
Premix ExTaqTM or SYBRw Premix DimerEraserTM (Takara Bio Europe,
Saint-Germain-en-Laye, France). In a second experimental approach for
evaluation of the FGFR3 qPCR results from the single cells, the complete
cDNA preparation of a cell (22 ml) was divided into 11 aliquots of 2 ml
and employed for 11 technical replicate single-cell qPCRs. Normal testis
tissue served as a positive control and DNA-free water served as a negative
control in all PCR experiments. Cell-free isolation buffer from the micromanipulation procedure was also tested for FGFR3 expression. Working procedures were conducted according to manufacturer’s instructions, but with
primer concentrations of 0.5 mM for each primer in a total volume of
20 ml. Primers (Eurofins Genomics, Ebersberg, Germany) were designed
using Primer3 (http://www.ncbi.nlm.nih.gov/tools/primer-blast/). All
primers were intron-overlapping with exception of the FGFR3 primers
(FGFR3 primer pair 2; see Supplementary Fig. S8), and were designed to
bind to a homologous sequence of all transcript variants with NM_### reference sequence numbers in the NCBI database (http://www.ncbi.nlm.nih.
gov/nucleotide/). Primer sequences (5′ to 3′ ) are as follows:
ACTA2_2_F: TGTGAAGAAGAGGACAGCACT,
ACTA2_2_R: TTGTCCCATTCCCACCATCA,
FGFR3_2_F: GCCGTGAATTCAGTTGGTTC,
FGFR3_2_R: CGTCGCTGGGTTAACAAAAT,
INSL3_F: GTCCACCGAAGCCAGGAG,
INSL3_R: GAGGGTCAGCAGGTCTTGTT,
INSL3_R_TV1: TTTGGAAGGTCAAGGTGGGA,
RPS27 forw: AACATGCCTCTCGCAAAGGA,
RPS27 rev: TGTGCATGGCTAAAGACCGT,
WT1_F: ATAACCACACAACGCCCATC,
WT1_R: GCTGTGCATCTGTAAGTGGG.
von Kopylow et al.
Amplicon sizes for the transcripts are 150 bp for actin alpha 2 smooth muscle
aorta (ACTA2), 172 bp for FGFR3, 232 bp (transcript variant 2) or 116 bp
(transcript variant 1) for insulin-like 3 (INSL3), 155 bp for ribosomal
protein S27 (RPS27) and 212 bp for Wilm’s tumor antigen 1 (WT1). The
first primer pair for INSL3 (INSL3_F, INSL3_R) resulted only in the amplification of transcript variant 2 (NM_005543.3), although originally designed
for both transcript variants. Therefore, a second reverse primer (INSL3_
R_TV1) to be used with the first INSL3_F primer was designed for specific
amplification of transcript variant 1 (NM_001265587.1). For the FGFR3 transcript, it was impossible to design a robust intron-overlapping primer pair for
reliable detection of FGFR3, especially in single-cell qPCR (see Supplementary
Fig. S8). As five attempts to design a primer pair without primer dimerization
or for efficient amplification failed, our established non-intron-overlapping
primer pair for FGFR3 (von Kopylow et al., 2010) was used. In our experience,
the formation of primer dimers in single-cell RT – PCR must be strictly
avoided since their presence most likely entails a shift of PCR kinetics from
the desired and specific low copy template to the already formed high
copy number primer dimers.
PCR cycling conditions were: 60 cycles of 10 s at 958C, 20 s at 608C and
30 s at 728C. As a control transcript, we used RPS27, a 40S ribosomal
subunit component protein, which is ubiquitously expressed (Thorrez
et al., 2008). Commonly, ribosomal proteins or other housekeeping genes
are applied for normalizing gene expression between samples (Thellin
et al., 1999). Here, RPS27 was employed for confirming (i) whether the cell
was correctly deposited in the tube and (ii) if the reverse transcription reaction was successful. This is sufficient because inter-sample normalization is
not required when comparing expression levels between single cells, since
biological variance is considerably higher than technical variance.
Identification of amplified transcripts was carried out by melting point
determination, sizing of amplicons on a Bioanalyzer (Agilent Technologies,
Santa Clara, CA, USA) and sequencing (Eurofins Genomics, Ebersberg,
Germany).
Estimation of FGFR3 copy numbers
in single cells and aggregates
The FGFR3 amplicon was purified from two pooled qPCR reactions using the
QIAquick PCR Purification Kit (Qiagen, Hilden, Germany). The sequencespecific molecular weight of the FGFR3 amplicon was estimated to be
106170.5 g/mol by the Oligocalc online tool (http://www.basic.north
western.edu/biotools/oligocalc.html) for double-stranded DNA. The conversion of the optical density (OD) obtained from an Ultrospec 3000 UV
Spectrophotometer (Pharmacia Biotech, Piscataway, NJ, USA) to copy
numbers was calculated by
50
ng
· OD · V · 10−9
ml
· NA
M
where V ¼ Volume of amplicon in ml, M ¼ molar mass in g/mol and NA ¼
Avogadro’s constant (6.022E23 mol21).
The amplicon was serially diluted from 108 to 1.25 copies. For each dilution, 2-9 replicate qPCRs were conducted. A calibration curve was calculated
by linear regression of all replicate Cq values per dilution against log10 copy
numbers. To estimate copy numbers from unknowns, the inverse function
using slope and intercept estimates of the least-squares fit were used.
Statistics
Calibration curves and variance analysis were performed with the R statistical
programming environment (www.r-project.org). Cq values were estimated
using the qpcR package (Ritz and Spiess, 2008) for R.
Single spermatogonial stem cell isolation and analysis
233
Figure 2 Immunocytochemistry on isolated bead-labeled cells. (A – C) Overview images of micromanipulatory picked, bead+ cells with positive undifferentiated embryonic cell transcription factor 1 (UTF1) nuclear staining from two different patients. (D) Larger depiction of cells from (A– C). Scale bars
(A– D): 10 mm.
Results
Cell isolation and counting of bead1 SPG
+
After magnetic cell isolation, bead-bound (bead ) cells from the positive
fraction (Fig. 1A –C; red arrows) could be clearly distinguished from
bead-unbound (bead2) cells (Fig. 1A– C; black arrows). While in the
negative fraction less than 1% of all cells were bead-labeled (Supplementary Fig. S1), 13% of cells with SPG-like morphology were labeled by
beads (data not shown). Figure 1H –M depict typical examples for
bead+ single (H, I), double (J, K) and quadruple (L, M) cells/aggregates
presenting a largely homogeneous phenotype compared with the heterogeneous appearance of cells within the negative cell fraction (Supplementary Fig. S1). Bead+ cells were selected under a light-microscope,
separately photographed for documentation, counted and consecutively
picked from the positive fraction with the capillary of a standard micromanipulator (Fig. 1D –F). The number of bead+ SPG per patient that
was microscopically visible ranged from 54 to 138 in patients with full
spermatogenesis (Supplementary Fig. S2), whereas the patients with
meiotic arrest displayed a higher number of bead+ cells in the range
of 220– 280 cells. This higher number may be explained by the lower
tubular diameter and thus a higher percentage of tubules per biopsy
compared with patients with full spermatogenesis (Behre et al., 2000;
Cerilli et al., 2010).
Analysis of the isolated cells by
immunocytochemistry and live/dead staining
To inspect stem cell characteristics of bead+ cells and to confirm the
isolation of the correct target cell population, immunostaining for the
pluripotency-associated transcription factor UTF1 (Okuda et al., 1998;
Kooistra et al., 2009), which is co-expressed with FGFR3 in rarely dividing
undifferentiated human A-type SPG (von Kopylow et al., 2012b), was conducted. UTF1 was chosen instead of FGFR3 as we failed to get a clear and
interpretable immunoreaction for FGFR3 on isolated single cells with an
independent antibody (sc-13121; Santa Cruz Biotechnology; data not
shown). UTF1 immunocytochemistry was carried out for the raw cell
suspension as well as for the picked bead+ cells, resulting in 1–2% (not
shown) and 100% positive staining (Fig. 2A–D), respectively. In contrast,
the vast majority of bead2 cells within the negative cell fraction exhibited
no UTF1 staining (Supplementary Fig. S3A and B). Also no staining was
observed when replacing the primary mouse antibody against UTF1
with normal mouse IgG (Supplementary Fig. S4B and C).
The Live/Deadw Cell Viability Assay revealed that almost all (95%)
bead+ and micromanipulator-isolated germ cells were viable (green;
Fig. 3A– G). Morphology of red-stained cell-like structures indicates
that these were cell debris (not shown). Live/dead staining of the negative cell fraction with viable (green) and dead (red) cells as depicted in
Supplementary Fig. S5 indicates over 90% viable cells.
234
von Kopylow et al.
Figure 3 Live/dead staining on isolated bead-labeled cells. (A– F) Representative images of live/dead staining from three patients with bead+, micromanipulatory picked cells that are viable (green). (G) Enlarged depiction of cells from (A – F). Red/green/bright field overlay. Scale bars: 100 mm (A– F) or
10 mm (G).
Single-cell qPCR/qPCR
A prerequisite for reliable single cell gene expression analysis is an
efficient extraction of all transcripts contained in the cell. Therefore,
we compared heat versus ultrasonic cell lysis with respect to their transcript liberation efficiency. We found no difference in the two methods
(P ¼ 0.96, t-test) based on Cq values obtained from RPS27-qPCRs of
bead2 cells (Supplementary Fig. S6A and B).
To investigate if bead+ cells express FGFR3 at the mRNA level, qPCR
was performed on single or pooled SPG. In an initial step, qPCR for the
control transcript RPS27 was conducted to demonstrate whether the
cells were correctly deposited in the tube and/or the reverse transcription reaction was successful (Fig. 4A and B). A subsequent FGFR3 qPCR
on single cells revealed that bead+ cells express FGFR3-mRNA (Fig. 4C,
‘B’, Fig. 4D) in contrast to bead2 cells (Fig. 4C, ‘NB’), which are only
positive for the RPS27 control gene (Fig. 4A, ‘NB’, Fig. 4B). Here, we
also observed a Pearson Correlation of 0.7 between these two,
suggesting that the variation of technical replicates may be based on
variations in cDNA synthesis.
However, in contrast to the fractions of pooled bead+ cells, not
all FGFR3-qPCR reactions for bead+ single cells were FGFR3-positive
Single spermatogonial stem cell isolation and analysis
235
Figure 4 Real-time RT – PCR on isolated single cells. (A) Bar plots represent quantification cycles (Cq) of bead+ cells (‘B’; n ¼ 1 to n ¼ 3, color-coded)
and bead2 cells (‘NB’; n ¼ 1 to n ¼ 3, color-coded) from five different patients for ribosomal protein S27 (RPS27; control transcript) versus testis positive
control (PC). (B) Amplification curves for RPS27 for the same samples as shown in (A; same color code) with testis positive (PC) and no template negative
control (NC). (C) Same as in (A), but for fibroblast growth factor receptor 3 (FGFR3). (D) Same as in (B), but for FGFR3.
within a set of three technical replicates (data not shown). Technical
replicates are defined here as several qPCRs from the same cDNA preparation of a single cell. To inspect this finding in more detail, the complete
cDNA preparation of a cell (22 ml) was divided into 11 aliquots of 2 ml
each and used for parallel PCR reactions. This approach was undertaken
to ensure amplification of all transcripts contained in a single cell and not
only a potentially unrepresentative aliquot that is affected by low-copy
Poisson heterogeneity (Tellinghuisen and Spiess, 2015). Here we
found distinct results for different cells: For some bead+ cells, all qPCR
reactions from 11/11 cDNA aliquots (Fig. 5, columns Aliquot 1–11)
showed successful PCR amplification (Fig. 5; 11 black numbered Cq
values in white table fields: patient 1, cells 8– 10; patient 2, cells 9–10),
whereas in other cases fewer samples were amplified (Fig. 5; 1–10
black numbered Cq values: patient 1, cells 1–7; patient 2, cells 1–8).
When calculating the averages of the quantification cycles (Cq) from all
successful amplifications of a single cell, it was observed that lower
averages correlated with a higher number of successful amplifications
(data not shown).
We speculated that low copy numbers of FGFR3 in SPG single cells are
the cause for inconsistent amplification, which is only successful when the
amount of transcript exceeds a certain copy number threshold per PCR
capillary/tube. For further examination of this assumption and to correlate Cq values with transcript copy numbers, we created qPCR calibration
data by serial dilution of a defined copy number-containing amplicon solution. First, we calculated FGFR3 copy numbers based on the measured
DNA concentration of a purified FGFR3 PCR product, its molar mass and
Avogadro’s constant (see Materials and Methods). We then prepared a
serial dilution from 108 FGFR3 copies downward to 1.25 copies for qPCR
calibration. The serial dilution qPCR, starting from 104 copies, delivered
Cq values of approx. 24, 28, 31 and 35 for 104, 103, 102 and 101 copies,
respectively (Fig. 6A, black squares). We noticed a threshold of five
copies for robust and reproducible PCR product formation with a
Cq 35. For 2.5 copies, PCR product formation occurred only in 22%
of all PCR reactions with Cq 36. Contrasting this, the dilution step containing 1.25 copies always failed to amplify. By applying these calibration
results to our FGFR3 qPCRs, 20 different single bead+ cells (from a total
of 75 cells, from 8 patients) (Fig. 5) were estimated to contain 2–141
FGFR3 copies (as calculated by summing copy numbers of all 11 cDNA
aliquots, Fig. 5, last column). This confirmed that higher copy numbers
correlate with a higher number of successful amplifications as well as
lower Cq values. Two of the bead+ cells (Patient 1, Cell #1 and Patient
2, Cell #1) with 2 copies suggest amplification of just the two genomic
copies and hence the presence of false positives (Ståhlberg and Bengtsson,
2010). However, since all unlabeled cells are qPCR negative (Fig. 4C, ‘NB’),
together with the pools of 10 unlabeled cells in Fig. 6B (Patient 2), there
seems to be a clear tendency for amplification only if a significant
236
von Kopylow et al.
Figure 5 Distribution of Cq values as obtained from FGFR3 qPCRs of 10 single SPG for two different patients. The cDNA of each cell (22 ml) was divided
into 11 aliquots A1 – A11 of 2 ml (left to right) and subjected to FGFR3 qPCR. In case of successful amplification, the Cq value is noted or left blank otherwise.
Estimated copy numbers (from the calibration curve; summed from the single values) are given in the last column (‘Copies’). The table is sorted by ascending
copy number. Patient 1 showed meiotic arrest, patient 2 had full spermatogenesis. Note the increasing number of positive amplifications in the aliquots in the
presence of lower Cq values and thus higher copy numbers.
number of transcripts add to the genomic copies. This indicates a likely
underestimation of copy numbers, which we discuss later on.
Next, a total of 51 bead2 single cells from 8 patients, morphologically
similar to bead+ cells, were tested with respect to FGFR3 expression.
First, we interrogated FGFR3 copy numbers of 39 bead2 cells from 6
patients by using one or more cDNA aliquots. Here, only 3/118
(2.5%) PCR reactions were positive (data not shown). For another 12
bead2 single cells from 2 patients, an evaluation of copy numbers for
all 11 cDNA aliquots similar to the bead+ cells indicated a low 0–8
copies (data not shown). In this context, it must be mentioned that we
had to reduce the Dynabead volume recommended by the manufacturer
to circumvent ‘over-labeling’ of the FGFR3-positive cells (accompanied
by a blackberry-like appearance; not shown), which would hamper
their applicability for downstream methods such as cell culture. Therefore, it can be expected that some bead2 cells are false negative
because they present FGFR3 on their cell surface, are not bead-labeled
but appear FGFR3-positive in qPCR.
A final copy number analysis of bead+ pools containing 7 –11 cells and
bead2 pools containing 10 cells gave a calculated 46– 134 and 0–8 FGFR3
copies, respectively (Fig. 6B). Interestingly, the estimated bead+ pool
copy numbers per cell (46–134 copies) tally nicely with those obtained
from the bead+ single-cell analysis (2– 141 copies, Fig. 5, last column).
In order to obtain an overview on how much of the observed variation
is attributed to biological or technical variance, we calculated technical
and biological errors of estimated copy numbers as % coefficient of variation (s.d./mean * 100) where possible (Supplementary Fig. S7). Here,
technical variation was calculated for the calibration curve replicates,
estimated from the regression curve, and from the 11 aliquots of each
cell in Fig. 5. Biological variation between the different cells was calculated
from the summed copy numbers of each cell in Fig. 5 and the estimated
copy numbers for all 10– 11 bead+ cell pools of Fig. 6B. As expected, in
the case of single-cell analysis, the biological variation is indeed higher
than the technical. The largest variation is evident for those single-cell
samples in which only a few qPCR reactions are positive, probably
from an increased Poisson effect. The calibration curve technical error
is increasing with higher dilutions (lower copy numbers), while biological
variation for the pooled cells is low because the inter-sample variance is
removed by the pooling procedure. Hence, minimization of technical
variation in low-copy qPCR should be a primary goal for the future.
Cell-free isolation buffer in the same volume as aspirated with a single
picked cell showed a background of approximately two FGFR3 copies,
probably caused by lysed cells. No significant increase of FGFR3 copy
numbers was observed when unbound beads were specifically tested.
To further corroborate our gene expression analysis indicating that
the bead+ cells are non-somatic, the latter were tested by single-cell
qPCR for the absence of markers for testicular somatic cell types, i.e.
Sertoli, peritubular and Leydig cells. For this reason, bead+ cells were
investigated by qPCR for WT1 (Call et al., 1990; Mäkelä et al., 2014),
Single spermatogonial stem cell isolation and analysis
Figure 6 Calibration curve and derived copy numbers for pooled
SPG. (A) Linear regression of FGFR3 dilution qPCR log10 copy
numbers versus Cq values. Error bars depict standard deviations
based on the sample sizes given in red at the upper graph border.
Slope, Intercept, R 2 and significance are given in the inserted box. (B)
Cq values and calibration curve derived copy numbers for two different
patients using five bead+ and seven bead2 pools of 7 – 11 cells. Copies
for pooled cells were calculated from the Cq values multiplied by 10,
because only 1/10 of the complete reverse transcription reaction was
used for qPCR input.
ACTA2 (Kossack et al., 2013) and INSL3 (Burkhardt et al., 1994), respectively. For INSL3, both transcript variants (see Materials and Methods)
were tested. We examined a total of 17 bead+ cells from four patients.
Where possible, all somatic markers were tested on the same cell,
however in some cases (when cDNA was limited), only one marker
was examined. None of the cells exhibited expression of these three
somatic markers (data not shown).
Discussion
FGFR3-positive human SPG represent an undifferentiated SPG subpopulation with potential stem cell character (von Kopylow et al., 2012b). To
isolate this promising cell population for future research, we developed
an approach that is applicable to very small fragments of testis tissue
(60 mg per patient). These fragments constitute significantly smaller
biopsies compared with the 250 –750 mg pieces that are regularly
obtained from micro-TESE extraction (Schlegel, 1999), as used for
GPR125-MACS isolation (Guo et al., 2015). Our method is based on
combining magnetic cell isolation via Dynabeads with subsequent
237
visual selection of bead-labeled cells using a standard micromanipulator.
The Dynabead system has previously been used to isolate murine
GFRA1-positive SSCs in combination with STA-PUT gravity sedimentation and differential plating (Hofmann et al., 2005). However, cell fractions obtained from Dynabead enrichment, among other existing
procedures that either employ (He et al., 2010; Conrad et al., 2014) or
omit (Sadri-Ardekani et al., 2009; Kossack et al., 2009; Mizrak et al.,
2010) specific hSSC markers, are not sufficiently pure to provide
defined cell culture conditions (Kanatsu-Shinohara and Shinohara,
2013). Hence, further procedures are mandatory to deliver meaningful
results that correlate with a pure SPG fraction. Along these lines, the
application of micromanipulator techniques enables the further selection
of enriched cell populations with respect to morphological criteria
(Vasco et al., 2009; Conrad et al., 2014). In our approach, we add
another level of stringency: selection of a tagged SPG subpopulation
expressing a potential human SPG stem cell factor, namely FGFR3. Consequently, this methodology can circumvent the isolation of testicular
cells that do not express FGFR3, mainly somatic cell types (Sertoli, peritubular and Leydig cells). These testicular somatic cells often cause problems within cell culture experiments by overgrowth of germ cells
(Langenstroth et al., 2014), resulting in distinct somatic cell-like structures (Kossack et al., 2013), gene expression patterns typical of
somatic origin (Ko et al., 2010) and misinterpretation of study results
(Zheng et al., 2014).
The isolated SPG subpopulation consists of single cells or low-number
aggregates (bead+) that can be specified as follows: (i) expression of
FGFR3-mRNA in comparison to bead2 cells, as demonstrated by singlecell qPCR, (ii) absence of transcripts typical of testicular somatic cell types
(WT1, ACTA2, INSL3), (iii) strong heterogeneity with respect to FGFR3mRNA copy numbers in single cells, and (iv) co-expression of the
pluripotency-associated protein UTF1.
Although our cell isolation technique is highly specific, a trade-off in
respect of quantity has to be made. Hence, cell cultivation will be a prerequisite for the characterization of hSSC function (Nagano et al., 2002)
by xenotransplantation (Clouthier et al., 1996). Nonetheless, it must be
emphasized that the low amount of material per patient used in this study
is comparable with that available in the fertility treatment of prepubertal
boys (Goossens et al., 2013; Baert et al., 2015).
To the best of our knowledge, this is the first mRNA expression analysis
of a putative stem cell factor in single human SPG. Recently, two other
studies conducted qPCR gene expression analysis on micromanipulatorpicked SPG. The first study investigated an endogenous housekeeping
gene in single mouse SPG and no distinction between SPG cell types/subpopulations was made (Vasco et al., 2009). The second study was
performed on pools of 50 human SPG (Conrad et al., 2014).
It is important to note that the analysis of FGFR3 expression in a single
SPG revealed a large heterogeneity in mRNA level within the FGFR3expressing SPG population (13– 51 copies/cell; 95% confidence interval). The high variability in transcript numbers clearly demonstrates
that morphologically similar and bead-labeled SPG are not a single
entity regarding their transcriptional state. Similar observations have
been made for single human (induced) embryonic stem cells (Narsinh
et al., 2011), but one must be cautious in assuming that these copy
numbers reflect exact values, because of the following issues: (i)
cDNA synthesis is likely to be inefficient in single cells, so that only a
small proportion of FGFR3 mRNA copies are converted to cDNA, and
(ii) an amplicon-derived calibration curve does not deliver exact copy
238
number estimates when the corresponding input is cDNA and not amplicons, consequently resulting in a Cq value shift. This effect is visible in the
intercept of the calibration curve (38.02): this would infer 0 copies, yet
an obtained Cq value indicates that amplification occurs. In addition,
Cell 1/Patient 1 and Cell 2/Patient 2 in Fig. 5 exhibit Cq values considerably lower than the intercept, which is physically impossible (less than 0
copies). Exact quantification in single-cell analysis is an emerging field and
possible solutions to remedy the above issues may be found in more
quantitative methods such as digital PCR (Faragó et al., 2013), which is
independent of calibration curves. Other approaches are based on
using optimized versions of calibration curves, e.g. through weighted
linear regression, quadratic models (Tellinghuisen and Spiess, 2014) or
by increasing the number of replicates in the low copy region. But even
in these cases, the encountered heterogeneity of the SPG stays largely
unaffected if we multiply the stated copy number values by factors that
account for inefficient cDNA synthesis. Therefore, this approach provides a relative quantification of copy numbers that likely represent an
underestimation of the true values.
The transcriptional heterogeneity of the SPG stem cell population might
give rise to a subclassification of stemness based on transcript abundance,
so that ultimately new isolation protocols may be developed for selecting
SPG that exhibit a high expression of specific and membrane-localized
stem cell markers.
Supplementary data
Supplementary data are available at http://molehr.oxfordjournals.org/.
Acknowledgements
We are grateful to Prof. Ingrid Moll for her support. We also thank
Gabriele Hahn for technical support.
Authors’ roles
K.v.K. and A.-N.S. developed the method, analyzed data, interpreted
results and wrote the manuscript. K.v.K. carried out the experiments.
W.S. and A.S. delivered biopsy samples from a therapeutic testicular
sperm extraction and classified the samples via light microscopy.
Funding
The project was funded by the DFG Research Unit FOR1041 Germ cell
potential (SCH 587/3-2) and DFG grants to K.v.K. (KO 4769/2-1) and
A.-N.S. (SP 721/4-1).
Conflict of interest
None declared.
References
Baert Y, Braye A, Struijk RB, van Pelt AM, Goossens E. Cryopreservation of
testicular tissue before long-term testicular cell culture does not alter in
vitro cell dynamics. Fertil Steril 2015;104:1244 – 1252.e1-4.
Behre HM, Yeung CH, Holstein AF, Weinbauer GF, Gassner P, Nieschlag E.
Diagnosis of male infertility and hypogonadism. In: Nieschlag E, Behre HM
von Kopylow et al.
(eds). Andrology. Male Reproductive Health and Dysfunction, 2nd edn. Berlin:
Springer, 2000, 89 –124.
Burkhardt E, Adham IM, Brosig B, Gastmann A, Mattei MG, Engel W.
Structural organization of the porcine and human genes coding for a
Leydig cell-specific insulin-like peptide (LEY I-L) and chromosomal
localization of the human gene (INSL3). Genomics 1994;20:13 – 19.
Call KM, Glaser T, Ito CY, Buckler AJ, Pelletier J, Haber DA, Rose EA, Kral A,
Yeger H, Lewis WH et al. Isolation and characterization of a zinc finger
polypeptide gene at the human chromosome 11 Wilms’ tumour locus.
Cell 1990;60:509– 520.
Cerilli LA, Kuang W, Rogers D. A practical approach to testicular biopsy
interpretation for male fertility. Arch Pathol Lab Med 2010;134:1197–1204.
Clouthier DE, Avarbock MR, Maika SD, Hammer RE, Brinster RL. Rat
spermatogenesis in mouse testis. Nature 1996;381:418– 421.
Conrad S, Azizi H, Hatami M, Kubista M, Bonin M, Hennenlotter J,
Renninger M, Skutella T. Differential gene expression profiling of enriched
human spermatogonia after short- and long-term culture. Biomed Res Int
2014;2014:138350.
Cotton LM, O’Bryan MK, Hinton BT. Cellular signaling by fibroblast growth
factors (FGFs) and their receptors (FGFRs) in male reproduction. Endocr
Rev 2008;29:193 –216.
Eswarakumar VP, Lax I, Schlessinger J. Cellular signaling by fibroblast growth
factor receptors. Cytokine Growth Factor Rev 2005;16:139– 149.
Faes K, Tournaye H, Goethals L, Lahoutte T, Hoorens A, Goossens E.
Testicular cell transplantation into the human testis. Fertil Steril 2013;
100:981– 988.
Faragó N, Kocsis ÁK, Lovas S, Molnár G, Boldog E, Rózsa M, Szemenyei V,
Vámos E, Nagy LI, Tamás G et al. Digital PCR to determine the number
of transcripts from single neurons after patch-clamp recording.
Biotechniques 2013;54:327 – 336.
Feig C, Kirchhoff C, Ivell R, Naether O, Schulze W, Spiess AN. A
new paradigm for profiling testicular gene expression during normal
and disturbed human spermatogenesis. Mol Hum Reprod 2007;
13:33 – 43.
Goossens E, Van Saen D, Tournaye H. Spermatogonial stem cell preservation
and transplantation: from research to clinic. Hum Reprod 2013;28:
897 – 907.
Guan K, Nayernia K, Maier LS, Wagner S, Dressel R, Lee JH, Nolte J, Wolf F,
Li M, Engel W et al. Pluripotency of spermatogonial stem cells from adult
mouse testis. Nature 2006;440:1199 – 1203.
Guo Y, Hai Y, Gong Y, Li Z, He Z. Characterization, isolation, and culture of
mouse and human spermatogonial stem cells. J Cell Physiol 2014;
229:407– 413.
Guo Y, Liu L, Sun M, Hai Y, Li Z, He Z. Expansion and long-term culture of
human spermatogonial stem cells via the activation of SMAD3 and AKT
pathways. Exp Biol Med 2015;240:1112– 1122.
He Z, Kokkinaki M, Jiang J, Dobrinski I, Dym M. Isolation, characterization,
and culture of human spermatogonia. Biol Reprod 2010;82:363 – 372.
Hofmann MC, Braydich-Stolle L, Dym M. Isolation of male germ-line stem
cells; influence of GDNF. Dev Biol 2005;279:114– 124.
Jahnukainen K, Mitchell RT, Stukenborg JB. Testicular function and fertility
preservation after treatment for haematological cancer. Curr Opin
Endocrinol Diabetes Obes 2015;22:217 – 223.
Jezek D, Knuth UA, Schulze W. Successful testicular sperm extraction (TESE)
in spite of high serum follicle stimulating hormone and azoospermia:
correlation between testicular morphology, TESE results, semen analysis
and serum hormone values in 103 infertile men. Hum Reprod 1998;
13:1230 – 1234.
Juul A, Aksglaede L, Lund AM, Duno M, Skakkebaek NE, Rajpert-De Meyts E.
Preserved fertility in a non-mosaic Klinefelter patient with a mutation in the
fibroblast growth factor receptor 3 gene: case report. Hum Reprod 2007;
22:1907 – 1911.
Single spermatogonial stem cell isolation and analysis
Kanatsu-Shinohara M, Shinohara T. Spermatogonial stem cell self-renewal
and development. Annu Rev Cell Dev Biol 2013;29:163– 187.
Ko K, Arauzo-Bravo M, Tapia N, Kim J, Lin Q, Bernemann C, Han DW,
Gentile L, Reinhardt P, Greber B et al. Human adult germline stem cells
in question. Nature 2010;465:E1.
Kooistra SM, Thummer RP, Eggen BJ. Characterization of human UTF1, a
chromatin-associated protein with repressor activity expressed in
pluripotent cells. Stem Cell Res 2009;2:211– 218.
Kossack N, Meneses J, Shefi S, Ngyen HN, Chavez S, Nicholas C, Gromoll J,
Turek PJ, Reijo-Pera RA. Isolation and characterization of pluripotent
human spermatogonial stem cell-derived cells. Stem Cells 2009;
27:138– 149.
Kossack N, Terwort N, Wistuba J, Ehmcke J, Schlatt S, Schöler H, Kliesch S,
Gromoll J. A combined approach facilitates the reliable detection of human
spermatogonia in vitro. Hum Reprod 2013;28:3012 – 3025.
Kristensen DM, Nielsen JE, Skakkebaek NE, Graem N, Jacobsen GK,
Rajpert-De Meyts E, Leffers H. Presumed pluripotency markers UTF-1
and REX-1 are expressed in human adult testes and germ cell
neoplasms. Hum Reprod 2008;23:775– 782.
Langenstroth D, Kossack N, Westernströer B, Wistuba J, Behr R, Gromoll J,
Schlatt S. Separation of somatic and germ cells is required to
establish primate spermatogonial cultures. Hum Reprod 2014;
29:2018– 2031.
Lanner F, Rossant J. The role of FGF/Erk signaling in pluripotent cells.
Development 2010;137:3351– 3360.
L’Hôte CG, Knowles MA. Cell responses to FGFR3 signalling: growth,
differentiation and apoptosis. Exp Cell Res 2005;304:417 – 431.
Lim J, Maher GJ, Turner GDH, Dudka-Ruszkowska W, Taylor S, Rajpert-De
Meyts E, Goriely A, Wilkie AOM. Selfish spermatogonial selection:
evidence from an immunohistochemical screen in testis of elderly men.
PLoS One 2012;7:e42382.
Liu Y, Niu M, Yao C, Hai Y, Yuan Q, Liu Y, Guo Y, Li Z, He Z. Fractionation of
human spermatogenic cells using STA-PUT gravity sedimentation and their
miRNA profiling. Sci Rep 2015;5:8084.
Maher GJ, Goriely A, Wilkie AOM. Cellular evidence for selfish
spermatogonial selection in aged human testes. Andrology 2014;
2:304 – 314.
Mäkelä JA, Toppari J, Rivero-Müller A, Ventelä S. Reconstruction of mouse
cellular microenvironments in long-term seminiferous tubule culture.
PLoS One 2014;9:e90088.
Mizrak SC, Chikhovskaya JV, Sadri-Ardekani H, van Daalen S, Korver CM,
Hovingh SE, Roepers-Gajadien HL, Raya A, Fluiter K, de Reijke TM et al.
Embryonic stem cell-like cells derived from adult human testis. Hum
Reprod 2010;25:158 – 167.
Nagano M, Patrizio P, Brinster RL. Long-term survival of human
spermatogonial stem cells in mouse testes. Fertil Steril 2002;78:
1225 – 1233.
Narsinh KH, Sun N, Sanchez-Freire V, Lee AS, Almeida P, Hu S, Jan T,
Wilson KD, Leong D, Rosenberg J et al. Single cell transcriptional
profiling reveals heterogeneity of human induced pluripotent stem cells.
J Clin Invest 2011;121:1217 – 1221.
Nickkholgh B, Korver CM, van Daalen SKM, van Pelt AMM, Repping S. AZFc
deletions do not affect the function of human spermatogonia in vitro. Mol
Hum Reprod 2015;21:553 – 562.
Okuda A, Fukushima M, Nishimoto M, Orimo A, Yamagishi T, Nabeshima Y,
Kuro-o M, Nabeshima Yi, Boon K, Keaveney M et al. UTF1, a novel
239
transcriptional coactivator expressed in pluripotent embryonic stem
cells and extra-embryonic cells. EMBO J 1998;17:2019 – 2032.
Perez-Castro AV, Wilson J, Altherr MR. Genomic organization of the human
fibroblast growth factor receptor 3 (FGFR3) gene and comparative
sequence analysis with the mouse Fgfr3 gene. Genomics 1997;41:10– 16.
Ritz C, Spiess AN. qpcR: an R package for sigmoidal model selection in
quantitative real-time polymerase chain reaction analysis. Bioinformatics
2008;24:1549 – 1551.
Sadri-Ardekani H, Mizrak SC, van Daalen SK, Korver CM, RoepersGajadien HL, Koruji M, Hovingh S, de Reijke TM, de la Rosette JJ, van
der Veen F et al. Propagation of human spermatogonial stem cells in
vitro. JAMA 2009;302:2127 – 2134.
Schlegel P. Testicular sperm extraction: microdissection improves sperm
yield with minimal tissue excision. Hum Reprod 1999;14:131 – 135.
Schulze W, Thoms F, Knuth UA. Testicular sperm extraction: comprehensive
analysis with simultaneously performed histology in 1418 biopsies from
766 subfertile men. Hum Reprod 1999;14(Suppl 1):82 – 96.
Ståhlberg A, Bengtsson M. Single-cell gene expression profiling using reverse
transcription quantitative real-time PCR. Methods 2010;50:282 – 288.
Tellinghuisen J, Spiess AN. Comparing real-time quantitative polymerase
chain reaction analysis methods for precision, linearity, and accuracy of
estimating amplification efficiency. Anal Biochem 2014;449:76 – 82.
Tellinghuisen J, Spiess AN. Absolute copy number from the statistics of the
quantification cycle in replicate quantitative polymerase chain reaction
experiments. Anal Chem 2015;87:1889 – 1895.
Thellin O, Zorzi W, Lakaye B, De Borman B, Coumans B, Hennen G,
Grisar T, Igout A, Heinen E. Housekeeping genes as internal standards:
use and limits. J Biotechnol 1999;75:291– 295.
Thorrez L, Van Deun K, Tranchevent LC, Van Lommel L, Engelen K,
Marchal K, Moreau Y, Van Mechelen I, Schuit F. Using ribosomal protein
genes as reference: a tale of caution. PLoS One 2008;3:e1854.
Valli H, Sukhwani M, Dovey SL, Peters KA, Donohue J, Castro CA, Chu T,
Marshall GR, Orwig KE. Fluorescence- and magnetic-activated cell
sorting strategies to isolate and enrich human spermatogonial stem cells.
Fertil Steril 2014;102:566 – 580.e7.
Vasco C, Zuccotti M, Redi CA, Garagna S. Identification, isolation, and
RT-PCR analysis of single stage-specific spermatogenetic cells obtained
from portions of seminiferous tubules classified by transillumination
microscopy. Mol Reprod Dev 2009;76:1173 – 1177.
von Kopylow K, Kirchhoff C, Jezek D, Schulze W, Feig C, Primig M,
Steinkraus V, Spiess AN. Screening for biomarkers of spermatogonia
within the human testis: a whole genome approach. Hum Reprod 2010;
25:1104– 1112.
von Kopylow K, Staege H, Schulze W, Will H, Kirchhoff C. Differential marker
protein expression specifies rarefaction zone-containing human A dark
spermatogonia. Reproduction 2012a;143:45 – 57.
von Kopylow K, Staege H, Spiess AN, Schulze W, Will H, Primig M, Kirchhoff C.
Fibroblast growth factor receptor 3 is highly expressed in rarely dividing
human type A spermatogonia. Histochem Cell Biol 2012b;138:759–772.
Yang S, Ping P, Ma M, Li P, Tian R, Yang H, Liu Y, Gong Y, Zhang Z, Li Z et al.
Generation of haploid spermatids with fertilization and development
capacity from human spermatogonial stem cells of cryptorchid patients.
Stem Cell Reports 2014;3:663 – 675.
Zheng Y, Thomas A, Schmidt CM, Dann CT. Quantitative detection of
human spermatogonia for optimization of spermatogonial stem cell
culture. Hum Reprod 2014;29:2497 – 2511.