Transcription Factor USF1 Is Required for Maintenance of Germline Stem Cells in Male Mice

Abstract

A prerequisite for lifelong sperm production is that spermatogonial stem cells (SSCs) balance self-renewal and differentiation, yet factors required for this balance remain largely undefined. Using mouse genetics, we now demonstrate that the ubiquitously expressed transcription factor upstream stimulatory factor (USF)1 is critical for the maintenance of SSCs. We show that USF1 is not only detected in Sertoli cells as previously reported, but also in SSCs. Usf1-deficient mice display progressive spermatogenic decline as a result of age-dependent loss of SSCs. According to our data, the germ cell defect in Usf1^−/− mice cannot be attributed to impairment of Sertoli cell development, maturation, or function, but instead is likely due to an inability of SSCs to maintain a quiescent state. SSCs of Usf1^−/− mice undergo continuous proliferation, which provides an explanation for their age-dependent depletion. The proliferation-coupled exhaustion of SSCs in turn results in progressive degeneration of the seminiferous epithelium, gradual decrease in sperm production, and testicular atrophy. We conclude that the general transcription factor USF1 is indispensable for the proper maintenance of mammalian spermatogenesis.

During spermatogenesis, haploid spermatozoa are continually produced from diploid spermatogonia through several rounds of mitotic divisions and two meiotic divisions. This complex process initiates from a population of undifferentiated germ cells, referred to as spermatogonial stem cells (SSCs). SSCs either self-renew or give rise to committed progenitors that are primed to differentiate under steady-state. A balance between SSC self-renewal and differentiation is critical for proper maintenance of spermatogenesis and for fertility (1). Heretofore, mechanisms underlying SSC quiescence, as well as seminiferous cycle-dependent cell cycle entry and exit, remain essentially undefined.

Spermatogenic cells are organized in the seminiferous epithelium in highly defined cell associations, or stages. In mice, there are 12 stages (I to XII) that together constitute the cycle of the seminiferous epithelium. Proliferation of spermatogonia initiates from a population of isolated type A-single (A_s) spermatogonia. Cell division of A_s spermatogonia first gives rise to a two-cell cyst, that is, A-paired (A_pr) spermatogonia, and then to A-aligned (A_al) spermatogonia, typically consisting of 4, 8, or 16 interconnected cells. Collectively, these cells are referred to as A-undifferentiated (A_undiff) spermatogonia. The A_undiff spermatogonia comprise spermatogonial stem cells (SSCs) and transit-amplifying progenitor spermatogonia that are primed to differentiate but possess a latent self-renewal capacity (2–6). A_undiff spermatogonia mitoses are not strictly bound to the progress of the seminiferous epithelial cycle, but they are, however, restricted to stages X to II (7, 8). In contrast, their irreversible commitment toward meiosis is spatiotemporally strictly regulated and confined to stages VII and VIII of the seminiferous epithelial cycle (9). At this point type A1 differentiating spermatogonia are formed that then undergo five additional mitotic divisions (A1–A2–A3–A4–In-B) before giving rise to preleptotene spermatocytes that enter meiosis.

Spermatogenesis is for a large part orchestrated by Sertoli cells that transduce endocrine signals (e.g., FSH and testosterone) and other cellular cues into paracrine regulation of male germ cell differentiation (10). Sertoli cells display unparalleled plasticity in terms of cellular function during the course of development and under steady-state spermatogenesis. Sertoli cell cyclical activity is a key to successful spermatogenesis; in addition to nursing up to five generations of differentiating germ cells, Sertoli cells also provide a niche for the A_undiff spermatogonia, including SSCs. The SSC niche is defined by molecular criteria. Glial cell line–derived neurotrophic factor (GDNF) is the most important single paracrine regulator of SSC fate. While Gdnf haploinsufficiency results in loss of SSCs, A_undiff spermatogonia accumulate when Gdnf is overexpressed (11, 12). In the testis, GDNF is derived from Sertoli, peritubular myoid, and vascular endothelial cells, and its secretion is partially under endocrine regulation (11, 13–18). Besides GDNF, ∼12 other paracrine factors have been implicated in the regulation of SSC fate decisions (1, 19).

Transcription factors, expressed by germ cells intrinsically and by somatic supporting cells, have also been implicated in the regulation and maintenance of spermatogenesis. Promyelocytic leukemia zinc finger (PLZF) (20, 21), TATA-box binding protein associated factor 4b (TAF4B) (22), spalt-like transcription factor 4 (SALL4) (23, 24), and forkhead box O1 (FOXO1) (25) are among germ cell–intrinsic transcription factors whose function is essential for lifelong spermatogenesis. Here, we dissect the requirement of upstream stimulatory factor (USF)1, a general transcription factor of the basic helix–loop–helix leucine zipper family, for mouse spermatogenesis. USF proteins are encoded by two ubiquitously expressed genes, Usf1 and Usf2, in mammals (26, 27). USF1/USF2 heterodimers bind the enhancer box in the promoter region of target genes (28, 29). In Sertoli cells of 5- to 11-day postpartum rats, USF proteins bind with increased affinity to FSH receptor (Fshr), Gata4, Nr5a1 [more commonly known as steroidogenic factor 1 (Sf1)], and sex hormone–binding globulin (Shbg) promoters, implying a role for USF in spermatogenesis (30).

As expected for a ubiquitously expressed transcription factor, USF1 has multifaceted roles in biological systems. In humans, USF1 polymorphisms are associated with regulating arterial blood pressure, synaptic plasticity in the central nervous system, and lipid metabolism (31–34). Recently, Laurila et al. (35) demonstrated that Usf1^−/− mice have beneficial lipid profiles, featuring reduced plasma triglycerides and elevated high-density lipoprotein cholesterol, and are protected against diet-induced weight gain. These findings indicate USF1 as a therapeutic target in cardiometabolic diseases in humans. However, whether USF1 is dispensable for regulatory pathways involved in reproductive processes has remained elusive.

Very limited data exist on USF1 target genes in the testis. In silico predictions [University of California Santa Cruz genome browser and SABioscience’s DECODE database (http://www.sabiosciences.com/)] indicate that USF proteins in mammals regulate expression of thousands of genes, of which all USF2 target genes are also USF1 targets but not vice versa. In other words, there are many genes that are predicted to be regulated by USF1 only, which further highlights the importance of USF1 over USF2 in mammals. Using Usf1 knockout (KO) mice (35), we now show that transcription factor USF1 is indispensable for proper maintenance of spermatogenesis and, more specifically, that USF1 is essential for maintaining a balance between self-renewal and differentiation of spermatogonial stem cells.

Materials and Methods

Mice

The KO construct and generation of Usf1^−/− mice were as described previously (35). Briefly, embryonic stem cells deficient for Usf1 were obtained from the German Genetrap Consortium (clone M121B03), in which vector ROSAbetageo+2 was retrovirally delivered into the fourth exon of the Usf1 gene. The resulting M121B03 cells were injected into C57BL/6J blastocysts to obtain Usf1 heterozygous mice. These mice were further crossed to obtain Usf1 KO mice. All experiments in this study were performed following all applicable national and institutional guidelines (Animal Experiment Board in Finland and Laboratory Animal Center of the University of Helsinki, respectively). The numbers of mice used in different experiments are detailed in an online repository (36).

Genotyping

PCR primers and cycling conditions for genotyping Usf1 were published previously (35). See the online repository (36) for details.

Histological analysis

For basic histology, testes were fixed with 4% paraformaldehyde (PFA) in 1× PBS for 4 hours at room temperature, followed by Bouin’s solution (Sigma-Aldrich, catalog no. HT10132) overnight. Testes were then dehydrated in 50% ethanol for 4 hours, 70% ethanol for 4 hours, and 70% ethanol overnight, embedded in paraffin, and cut into 5-µm-thick sections. Tissue sections were deparaffinized using standard xylene and alcohol series (absolute, 95%, 90%, and 70% ethanol), and finally into sterile water. After staining with Mayer’s hematoxylin solution (Sigma-Aldrich), sections were washed, counterstained with eosin (Sigma-Aldrich), and dehydrated using a standard procedure (once with 70%, 90%, 95%, and absolute ethanol, and twice with xylene) and then mounted using a xylene-based mounting medium.

Assessment of the spermatogenic defect

Testes were collected and fixed overnight in 4% PFA followed by embedding into paraffin. Five-micrometer-thick sections were prepared for histological analysis and stained with 4′,6-diamidino-2-phenylindole (DAPI) and then analyzed for integrity of the seminiferous epithelium. At least 64 cross-sections of seminiferous tubules per mouse [at the ages of 8, 12, 20, and 30 weeks; n = 2 to 3 for wild-type (WT), n = 3 for KO] from two nonconsecutive histological sections were analyzed for the extent of spermatogenic defect and classified into three categories (normal, one to three layers missing, only basal layer) based on the presence or absence of hierarchical layers of differentiating germ cells.

Immunofluorescence labeling on cryosections

Testes were dissected and fixed overnight in 4% PFA, followed by dehydration in 20% sucrose solution in 1× PBS and embedding into OCT compound (Tissue-Tek). Ten-micrometer-thick sections were prepared for immunofluorescence labeling. Slides containing testis cross-sections were washed briefly in PBS and boiled in 10 mM sodium citrate buffer (pH 6.0) for 15 to 20 minutes in a microwave oven. Sections were then washed twice in PBS and blocked for 1 hour at room temperature in a blocking buffer containing 5% BSA and 5% normal serum (from the same species in which the secondary antibody was raised) in 0.2% PBST (0.2% Tween 20 in 1× PBS). Primary antibody was diluted in antibody dilution buffer (1% BSA in 0.2% PBST), incubated overnight in a cold room, and washed four times with 0.1% PBST the next morning. Secondary antibodies were diluted in the same antibody dilution buffer as the primary antibody and applied on the sections. Sections were incubated with secondary antibody solution for 1 hour at 37°C, washed four times with 0.1% PBST, and mounted using Vectashield mounting medium containing DAPI (Vector Laboratories). Sections were imaged on a Zeiss Axioimager microscope, captured with ZEN2 software, and further processed with CorelDraw (version X7) image editing software. The following primary antibodies were used in this study: AR [RRID: AB_11156085 (37)], Claudin-11 [RRID: AB_639330 (38)], GATA1 [RRID: AB_627663 (39)], GATA4 [RRID: AB_2108747 (40)], Ki-67 [RRID: AB_10854564 (41)], SOX9 [RRID: AB_2239761 (42) and RRID: AB_2574463 (43)], Wilms tumor 1 [WT1; RRID: AB_2216233 (44)], PLZF [RRID: AB_2304760 (45)], USF1 [RRID: AB_2213986 (46)], GDNF family receptor α1 [GFRα1; RRID: AB_2110307 (47)], Espin [RRID: AB_399174 (48)], and DNMT3A [RRID: AB_1149786 (49)]. Antibody dilutions are provided in an online repository (36).

Immunofluorescence labeling on paraffin-embedded sections

Immunofluorescence labeling on 4% PFA-fixed, paraffin-embedded testis sections (see above) were used to analyze the total and proliferating number of Sertoli and Leydig cells at different time points (36). Double labeling of cells with proliferation marker Ki-67 antibody and cell type–specific antibodies (SOX9 for Sertoli cells, GATA4 for Leydig cells) was done. Briefly, slides were dewaxed using serial incubations in xylene and ethanol. Permeabilization was carried out in a pressure cooker in 0.1 M citrate buffer (pH 6.0), and autofluorescence was blocked with 100 mM NH₄Cl. Unspecific binding of the primary antibody was blocked by incubation in a buffer containing 5% normal serum (from same species in which the secondary antibody was raised) in 0.05% TBST (0.05% Tween 20 in 1× TBS) for 1 hour. Primary antibodies were diluted in a blocking solution (5% normal serum in 0.05% TBST) and incubated overnight in a cold room. Secondary antibodies were diluted in the same blocking solution and applied on the sections for 1 hour at 37°C. DAPI was used as a nucleic counterstain. Finally, the sections were mounted in ProLong^® Diamond antifade mountant (Thermo Fisher Scientific). Sections were imaged using a Pannoramic MIDI FL slide scanner with a 40×/Korr 0.95 Plan Apochromat objective (Zeiss).

Isolation of stage-specific segments of seminiferous tubules

The testes of 8-week-old Usf1^−/− (n = 3) and WT control (n = 3) mice were dissected and decapsulated. Using a transillumination-assisted microdissection method, seminiferous tubule segments representing stages II to V, VII and VIII, and IX to XI were dissected and snap-frozen in liquid nitrogen (50, 51).

RNA extraction and RT-qPCR

RNA was extracted from snap-frozen pieces of testicular tissue or staged segments of seminiferous tubule using a mini NucleoSpin RNA extraction kit (Macherey-Nagel, catalog no. 740955.50) or TRIzol (Thermo Fisher Scientific), respectively, following the manufacturers’ protocols. One microgram of extracted RNA was reversed transcribed using either a SuperScript VILO cDNA synthesis kit (Thermo Fisher Scientific, catalog no. 11754050) or a SuperScript IV VILO master mix cDNA synthesis kit (Thermo Fisher Scientific, catalog no. 11756050). RT-qPCR was performed using SsoAdvanced Universal SYBR Green Supermix (Bio-Rad Laboratories, catalog no. 1725270), and expression data were normalized to housekeeping genes. The list of primers used in this study is provided in an online repository (36). Data were analyzed using Bio-Rad CFX manager software (version 3.1).

Seminiferous tubule whole-mount staining

The preparation of seminiferous tubules for whole-mount staining is described in an online repository (36). For immunostaining, seminiferous tubules were blocked for 1 hour using 0.3% PBSX (0.3% Triton X-100 in 1× PBS) supplemented with 2% BSA and 10% fetal bovine serum in a 2-mL round-bottom tube on a rotating table at room temperature. Primary antibodies were diluted in 1% BSA in 0.3% PBSX and incubated overnight in a cold room with rotation. Seminiferous tubules were then washed three times with 0.3% PBSX, incubated with secondary antibody diluted in 1% BSA in 0.3% PBSX for 2 hours on a rotating table at room temperature, and washed again three times. Finally, seminiferous tubules were arranged in linear strips and mounted with Vectashield mounting medium containing DAPI (Vector Laboratories). At least three WT and Usf1^−/− mice were used in the analyses.

Sperm count

The number of cauda epididymal sperm was counted from 8- and 12-week-old WT (n = 3) and Usf1^−/− (n = 3) mice. For each mouse, one cauda epididymis was dissected (few slits were made) and placed in 1 mL of PBS for ∼30 minutes. The solution was pipetted up and down a few times to homogenize and extract any remaining sperm. Sperm in 20 μL of homogenized PBS mixture of sperm were counted on a Bürker chamber, and total sperm from 1 mL of solution were calculated.

Hormone measurements

For intratesticular hormonal level quantitation, testis lysates were prepared according to a protocol described earlier (52). Briefly, testes were mechanically homogenized and lysed in APC buffer [20 mM Tris-HCl (pH 7.7), 100 mM KCl, 50 mM sucrose, 0.1 mM CaCl₂, 1 mM MgCl₂] supplemented with protease inhibitor (cocktail set III, Merck Millipore, catalog no. 535140) at 1:200 dilution. The APC buffer was supplemented with 0.5% Triton X-100 (final concentration) prior to lysate preparation. The lysed samples were centrifuged at 14,000g for 10 to 12 minutes at 4°C to 8°C. A Pierce BCA kit (Thermo Fisher Scientific) was used for measuring the lysate protein concentration. Twenty micrograms of total protein was used for each sample, and the hormone level quantitation was done according to the manufacturer’s instructions. Hormonal concentrations obtained from a standard curve were further normalized to respective testis weights. For serum hormonal level quantitation, 25 µL of serum was used for each reaction, and the concentrations were measured using a standard curve made after the manufacturer’s instructions. Four mice per group were used for intratesticular hormonal measurement, whereas five mice per group were used for serum hormone level quantitation. The following ELISA kits were used: FSH ELISA kit (Novus Biologicals, catalog no. KA2330), LH ELISA kit (Novus Biologicals, catalog no. KA2332), and testosterone ELISA kit (Abcam, catalog no. ab108666).

Statistical analysis

Statistical tests were performed using GraphPad Prism software (version 6). Unpaired t tests were performed to calculate P values; P values <0.05 were considered statistically significant.

Results

USF1 expression within the seminiferous epithelium is detected in Sertoli cells and spermatogonia

As a first step toward unraveling the roles of USF1 in spermatogenesis, we investigated which cells within the testis express USF1. Usf1 mRNA was detected at the whole-testis level at all of the studied time points (Fig. 1A). Moreover, its expression level in adult mice did not depend on the stage of the seminiferous epithelial cycle (Fig. 1B). In adult WT mice, USF1 was detected in Sertoli cell nuclei by indirect immunofluorescence. In agreement with mRNA-level data, USF1 protein expression was not affected by the stage of the seminiferous epithelial cycle (Fig. 1C). To confirm that the USF1-positive cells are Sertoli cells, antibodies against two well-known Sertoli cell markers, GATA1 and WT1, were included in the staining protocol. USF1-positive seminiferous epithelial cells invariably also expressed WT1 but were GATA1-positive only in a subset of cross-sections (Fig. 1C), as expected (53). Interstitial cells also stained weakly for USF1. Reassuringly, Usf1^−/− testes were negative for USF1, but they exhibited normal expression of GATA1 and WT1 (Fig. 1C).

Figure 1.

USF1 expression is limited to testicular somatic cells and a subset of spermatogonia. (A and B) Quantitative RT-PCR analysis of Usf1 expression (A) from whole-testis RNA of WT control mice at the indicated time points and (B) from total RNA of indicated pooled seminiferous tubule epithelial stages. Usf1 expression levels were normalized to respective α-tubulin levels in (A). Data in (B) were normalized against Wt1, which is uniformly expressed by Sertoli cells independent of epithelial stage. Bars represent mean ± SD, and P values are from unpaired t tests. **P < 0.01; ***P < 0.001. (C) Testis cross-sections of the indicated genotypes were stained with DAPI and antibodies against USF1, GATA1, and WT1. GATA1 displayed a stage-dependent pattern of expression in Sertoli cells, and therefore GATA1 expression, unlike WT1, was limited to Sertoli cells of certain seminiferous tubules. Immunofluorescence triple staining confirmed abundant expression of USF1 in Sertoli cells. A low level of USF1 was also detected in the testicular interstitium. USF1 expression was undetectable in the KO testes. Scale bars, 100 μm. (D and E) Representative whole-mount immunofluorescence staining of WT adult seminiferous tubules by antibodies against PLZF and USF1. USF1 was detected in PLZF-negative Sertoli cells as well as in PLZF-positive undifferentiated A_s, A_pr, and A_al (cysts of 4, 8, and 16 cells) spermatogonia. See the online repository (36) for more detailed characterization of the USF1-positive spermatogonial subpopulation. Scale bars, 50 μm. (F) Summary of the whole-mount immunofluorescence staining included in this study. USF1 is ubiquitously expressed by Sertoli cells. USF1 expression in spermatogonia is restricted to A_undiff spermatogonia and to early (up to A4) differentiating spermatogonia. Spermatogonial expression of USF1 closely follows that of PLZF. The solid line indicates ubiquitous readily detectable expression, whereas downregulation of protein expression is marked with a dotted line. For GFRα1, the dotted line is used throughout to illustrate that GFRα1 is limited to the SSC subset of A_undiff spermatogonia. As suggested by immunofluorescence data, (B) Usf1 does not display a seminiferous epithelial stage–regulated pattern of expression.

Based on this localization pattern, testicular USF1 expression appeared to be restricted to testicular somatic cells and missing from germ cells. Owing to the extensive tissue handling procedure, it can be challenging to reveal low-level protein expression in paraffin-embedded tissues sections. Therefore, we performed whole-mount staining of fixed seminiferous tubules, which allows three-dimensional visualization of the tissue and can provide more sensitive detection of low-abundance proteins. Indeed, this approach confirmed the expression of USF1 in Sertoli cells, but it also revealed USF1 expression in PLZF-positive cells (i.e., spermatogonia) on the basement membrane of the seminiferous epithelium (Fig. 1D and 1E). PLZF, originally regarded as a specific marker for undifferentiated stem and progenitor spermatogonia, has since been shown to be expressed also in early differentiating spermatogonia (21, 54, 55).

To further characterize the USF1-positive spermatogonial population we stained for DNMT3A, a protein whose expression is induced upon differentiation commitment in the male germline (56) and maintained in all populations of differentiating spermatogonia (A1 to A4, In, B) and preleptotene spermatocytes. The highest level of USF1 was observed in PLZF-positive/DNMT3A-negative and PLZF-positive/DNMT3A-positive cells, that is, undifferentiated (A_undiff) and early differentiating spermatogonia (36). A more detailed analysis of USF1 expression within the differentiating spermatogonial population revealed that USF1 levels were sharply downregulated in differentiating spermatogonia (36). Because A1 differentiating spermatogonia are derived from A_undiff spermatogonia without mitosis in a retinoic acid–dependent transition (7), USF1 can be justifiably considered a novel marker for A_undiff spermatogonia.

A_undiff spermatogonia are considered to consist of two populations of cells: actual stem cells (SSCs) that undergo self-renewal, and transit-amplifying progenitor cells (2–5). A_undiff spermatogonia expressing GFRα1, most often representing A_s and A_pr cells, are more likely to act as stem cells, whereas differentiation-primed SOX3-positive A_undiff spermatogonia typically represent longer cysts (57). Triple staining of mouse seminiferous tubules with antibodies against GFRα1, USF1, and SOX9 confirmed that a subset of USF1-positive A_undiff spermatogonia also express the stem cell marker GFRα1, which suggests that USF1 is also present in SSCs (36). A summary of the marker expression based on whole-mount immunofluorescence stainings is provided in Fig. 1F.

Reduced testis weight in Usf1^−/− mice

Body and testis weights of Usf1^−/− and control mice were recorded at multiple time points from the first week postpartum to 20 weeks. Decreased body weight and testis size in Usf1^−/− mice was observed at all time points (Fig. 2A) (36). Similar to body weight, testis size and weight were also smaller in KO mice (Fig. 2B and 2C). Furthermore, relative testis weight was lower in Usf1^−/− mice at all of the examined time points as compared with control mice, and it was significantly lower after birth and in adulthood from 12 weeks onward (Fig. 2D). Thus, USF1 deficiency affects body weight and testis growth. Despite USF1-deficient males’ lower testis and body weight, these mice were otherwise healthy and able to sire offspring.

Figure 2.

USF1 is required for normal testis growth. (A) One-week-old male pups of the indicated genotypes. These three males were from the same litter. (B) Representative testes of indicated genotypes from 8-wk-old males. (C) Testis weight of control and Usf1^−/− mice at indicated ages. (D) Relative testis weight of same mice represented in (C). In both (C) and (D), a minimum of three animals per group were included per time point; bars represent mean ± SD, and P values are from unpaired t tests. (E–N) Progressive degeneration of the seminiferous epithelium in the absence of Usf1. (E–J) Representative testis cross-sections of the indicated genotypes. (E–G) Testis sections from (E) 12-wk adult controls and from Usf1^−/− mice at (F) 12 and (G) 30 wk of age stained with hematoxylin and eosin to show morphology of seminiferous tubules. (F) Already at 12 wk a substantial proportion of seminiferous tubules of Usf1^−/− mice had degenerated and hosted only the basal layer, or lacked one or more of the hierarchical layers of differentiating germ cells. (G) Tubule degeneration became more prevalent with age. (H–J) Cross-sections of individual seminiferous tubules representing normal spermatogenesis at stage VII and VIII of the seminiferous epithelial cycle. Compared with (H) seminiferous tubules in controls, (I–J) reduced cellularity is observed in otherwise normal-looking seminiferous tubules of Usf1^−/− mice. Scale bars, (E–G) 500 μm and (H–J) 50 μm. (K–N) Evaluation of spermatogenic defect from testis cross-sections. Error bars represent mean ± SD, and P values are from unpaired t tests. *P < 0.05; **P < 0.01; ***P < 0.001. See also the online repository (36) for scoring criteria.

Usf1^−/− mice display progressive degeneration of the seminiferous epithelium

Testis histology of adult Usf1^−/− and WT control (Usf1^+/+) mice was studied at 8, 12, 20, and 30 weeks of age. The spermatogenic defect of Usf1^−/− mice became obvious in 12-week-old mice. Whereas control mice hosted normal spermatogenesis in nearly all tubules, a substantial proportion of seminiferous tubules of Usf1^−/− mice had degenerated at that time point (Fig. 2E and 2F). The magnitude of this defect increased with age, and in 30-week-old KO mice only a minority of tubules hosted spermatogenesis (Fig. 2G). Moreover, seminiferous tubules with ongoing spermatogenesis typically appeared to contain a lower number of differentiating cells per cross-sectional area and thus displayed lower cellular density (Fig. 2H–2J). Closer examination revealed that many tubule cross-sections were devoid of specific types of germ cells.

A vast majority (92%) of seminiferous tubule cross-sections in 8-week-old Usf1^−/− mice still showed normal layering of the seminiferous epithelium consisting of three to four cohorts of differentiating germ cells. However, in older animals, cross-sections missing one, two, or three layers of differentiating germ cells or containing only the basal layer became significantly more common (Fig. 2K–2N). The cross-sections that lacked one to three layers of differentiating germ cells typically consisted of the basal layer plus elongating spermatids, potentially suggesting a spermiation defect. However, cross-sections lacking just one or two layers of spermatogenic cells were also identified (36). In line with the observations described above, epididymal sperm count in Usf1^−/− mice was only slightly decreased at 8 weeks of age but severely affected at the age of 12 weeks (36).

FSH, LH, and testosterone levels are maintained in Usf1-deficient mice

The testis is an endocrine organ, and pituitary-derived LH and FSH are essential for testicular development and function (58). Whereas the serum testosterone level in Usf1^−/− mice was not different from WT controls, serum levels of LH and FSH were significantly higher (Fig. 3A–3C). These data indicate that the spermatogenic phenotype of Usf1^−/− mice is likely not due to insufficient gonadotropin stimulation. Interestingly, intratesticular testosterone (ITT) levels in 12-week-old Usf1^−/− mice were significantly higher when compared with WT control mice (Fig. 3D). This indicates that degeneration of seminiferous epithelium in Usf1^−/− mice does not result from lack of androgen stimulation. To investigate whether high ITT levels in KO mice are due to Leydig cell hyperplasia, we quantified Leydig cells at different time points. However, no significant differences were observed (Fig. 3E). Moreover, Leydig cell proliferation was not affected, and Leydig cells of both Usf1^−/− and WT mice entered mitotic quiescence by 12 weeks of age (Fig. 3F). Transcript levels of the LH/choriogonadotropin receptor (Lhcgr) were also unaffected (Fig. 3G). High ITT levels can at least in part be explained by the increased proportion of Leydig cells to other cell types in degenerating Usf1^−/− testes.

Figure 3.

Endocrine regulation of the Usf1^−/− testis. (A–C) Serum hormonal levels in mice of the indicated genotypes. Although serum levels of (A) LH and (B) FSH were higher in Usf1^−/− mice, there was no difference in (C) serum testosterone levels. (D) Intratesticular levels of testosterone were significantly elevated in the Usf1^−/− testes when compared with WT control mice. All hormonal levels in (A)–(D) are measured from a minimum of four animals per genotype and at 12 wk of age. (E) Quantitation of Leydig cells at different time points. (F) Quantitation of proliferative Leydig cells at the indicated time points by an antibody against Ki-67. (G) LH/chromogranin receptor (Lhcgr) expression by quantitative RT-PCR analysis; α-tubulin was used as a normalization control. (H and I) Assessment of AR expression by (H) immunohistochemistry and (I) quantitative RT-PCR. AR was found normally expressed within the different testicular somatic cell types. Insets 1 to 4 in (H) show comparable AR expression in Sertoli/myoid cells (insets 1 and 3) and Leydig/myoid cells (insets 2 and 4) between WT control and Usf1^−/− mice. (J) AR expression at the indicated seminiferous tubule epithelial stages as normalized to Wt1, which is uniformly expressed by Sertoli cells independent of the epithelial stage. Error bars represent mean ± SD, and P values are from unpaired t tests. Scale bars, (H) 100 μm (H) and [insets in (H)] 50 μm. *P < 0.05; **P < 0.01.

Testosterone exerts its effect via binding to the androgen receptor (AR) that is expressed by Sertoli, Leydig, and peritubular myoid cells in the testis. Cell type–specific AR action is essential for lifelong fertility, whereas global AR deficiency compromises masculinization (59–63). Immunofluorescence detection indicated that AR expression in the testis between Usf1^−/− and WT control mice does not differ (Fig. 3H). This was further corroborated by quantitative PCR data showing normal AR expression on the whole-testis level (Fig. 3I). Because correct stage-dependent gene expression is arguably essential for efficient progression of the spermatogenic program, we isolated tubules from three pooled epithelial stages (II to V, VII and VIII, and IX to XI) for transcriptomic analyses using the seminiferous tubule transillumination method (50, 51). AR mRNA displayed the highest level of expression in early stages of the seminiferous epithelial cycle both in the KO mice and WT controls (Fig. 3J).

Usf1 deficiency does not substantially affect Sertoli cell maturation and function

Sperm production capacity is determined by the number of Sertoli cells, as a single Sertoli cell is able to host a specific number of germ cells in a species-dependent manner (64, 65). To address whether the low density of germ cells per cross-section of seminiferous epithelium in Usf1^−/− mice (Fig. 2H–2J) can be explained by a reduction in Sertoli cell number, we quantified Sertoli cells per tubular cross-section at different ages. However, no significant differences between KO and WT control mice were found (36).

During the first weeks of postnatal life, Sertoli cells undergo a maturation program that encompasses a shift in cell transcriptome/proteome, loss of mitotic activity, and cell polarization. Because incomplete maturation of Sertoli cells might contribute to the spermatogenic phenotype observed in Usf1^−/− mice, we investigated various aspects of the process. However, no significant differences between Usf1^−/− and Usf1^+/+ Sertoli cells were recorded in mitotic activity (36), in expression of selected Sertoli cell immaturity-related mRNAs [anti-Müllerian hormone (Amh), Podoplanin (Pdpn), and Cytokeratin-18 (Ck18) (66–68)], or in the localization of blood–testis barrier proteins Claudin-11 and Espin (36). Collectively, these data led us to conclude that Sertoli cell maturation in Usf1^−/− mice is not compromised.

Stage-dependent gene expression in Sertoli cells is somewhat deregulated in Usf1^−/− mice

USF1 is a transcriptional activator, and it has been implicated in the regulation of two important testicular genes: Fshr (69–71) and steroidogenic factor 1 (Sf1 or Nr5a1) (30, 72). In silico analyses further predict that there are USF1 binding sites upstream of a number of genes important for Sertoli cell function, including Gata4 (73, 74) and Sox9 (75). RT-qPCR analysis did not reveal any statistically significant differences in the expression of these genes or in another essential Sertoli cell transcription factor, Wt1 (76, 77), in Usf1^−/− testes (36).

Sertoli cells undergo cyclical changes in their transcriptome as a result of the seminiferous epithelial cycle (78), and many genes expressed by Sertoli cells exhibit a variable level of expression, as dictated by the stage of the cycle. Whereas all of the studied genes maintained their typical pattern of expression in different stages, elevated Gata1 and Sox9 mRNA levels, which are potentially biologically important, were observed at stages VII to VIII in KO mice (Fig. 4).

Figure 4.

Seminiferous epithelial stage–specific gene expression patterns are maintained in the absence of Usf1. (A–D) Expression of selected mRNAs: (A) Fshr, (B) Gata1, (C) Sox9, and (D) Stra8 was assessed by RT-qPCR. To control for the observed differences in cellularity between Usf1^−/− and WT mice, data were normalized against Wt1 that is uniformly expressed by Sertoli cells independent of the epithelial stage. Seminiferous tubule segments representing stages II to V, VII and VIII, and IX to XI were isolated by transillumination-assisted microdissection from 8-wk-old mice. Generally, expression of studied genes was maintained in a stagewise manner between Usf1^−/− and WT mice. However, for individual genes, enhanced expression was observed at specific stages in the KO tubules. Three animals per group were used in all experiments. Stra8 was included in the experiment as an internal control because it is known to display a highly stage-dependent pattern of expression at stages VII and VIII of the seminiferous epithelial cycle (17, 79). Error bars represent mean ± SD, and P values are from unpaired t tests. *P < 0.05; **P < 0.01.

Depletion of undifferentiated spermatogonia contributes to degeneration of the seminiferous epithelium in Usf1^−/− mice

To elucidate the origin of seminiferous epithelial degeneration in Usf1^−/− mice, we quantified the proportion of tubules that host PLZF-positive cells. It steadily decreased in Usf1 KO mice with age (Fig. 5A and 5B), indicating depletion of undifferentiated spermatogonia. Thus, the spermatogenic defect in these mice can at least partially be attributed to an inability of Usf1^−/− testes to maintain undifferentiated spermatogonia, that is, the stem and progenitor cells of the adult male germline.

Figure 5.

PLZF-positive cells are depleted with age in the Usf1^−/− testes. (A) Testis cross-sections, shown here from 12- and 25-wk-old mice, were stained with an antibody against PLZF at different time points. (B) Quantitation of tubules hosting PLZF-positive cells in control and Usf1^−/− testes at the indicated ages. A minimum of two animals per time point were analyzed. Scale bars, 50 μm. (C and D) Gdnf and Scf expression levels at the indicated stages of the seminiferous epithelial cycle. Transcript levels were normalized to Wt1, which is uniformly expressed by Sertoli cells independent of the epithelial stage. A minimum of three animals per group was used in all experiments. Error bars present mean ± SD, and P values are from unpaired t tests. *P < 0.05; **P < 0.01.

Stem cell niche in Usf1^−/− mice

Stem cells are located in a microenvironment that maintains their self-renewal capacity, that is, the stem cell niche. In the mouse testis the niche cannot be defined by anatomical criteria but rather by molecular cues, and the fate of undifferentiated spermatogonia is dictated by the availability of a selection of paracrine factors. A number of factors have been implicated in the regulation of cell fate decisions within the mouse undifferentiated spermatogonia. Although the role of Gdnf among these factors is best characterized, Cxcl12 (80), Csf1 (81), Fgf2 (82), Nrg1 (83), and Wnt4 (84), Wnt5a (80, 85, 86), and Wnt6 (87) are arguably also important regulators of A_undiff spermatogonia, whereas Bmp4 (88) and Scf (89) become critical once the transition into A1 differentiating spermatogonia has taken place. We studied the mRNA expression of these genes at 1, 4, and 8 weeks. Despite considerable variation, no statistically significant changes for any of these genes were recorded, implying that the paracrine milieu that A_undiff spermatogonia are exposed to in the Usf1^−/− testis is not drastically different from that in the control testis (36).

Because of the importance of GDNF and SCF for A_undiff and differentiating spermatogonia, respectively, we studied the expression of these two genes in staged tubules isolated from 8-week-old mice. This time point was selected because in Usf1^−/− testis the first signs of seminiferous epithelial degeneration become apparent by then, but the cellularity still remains largely unaffected (Fig. 2K). Consistent with data above (Fig. 4), the stage-specific expression pattern for both Gdnf and Scf was maintained in Usf1 KO mice (Fig. 5C and 5D). However, mRNA levels were generally higher in Usf1^−/− mice, and the differences reached statistical significance at stages II to V for Gdnf and stages VII and VIII for Scf. These data indicate that spermatogonia in the Usf1^−/− testis may be exposed to physiologically altered levels of paracrine growth factors at specific stages of the seminiferous epithelium, despite that at the whole testis level no changes were recorded.

A-single spermatogonia in Usf1^−/− testes are hyperproliferative

Increased apoptosis and proliferation-coupled stem cell exhaustion are among the obvious mechanisms that may contribute to the observed progressive depletion (Fig. 5B) of germline stem cells within the Usf1^−/− testis. To investigate these options, we employed indirect immunofluorescence on segments of seminiferous tubule from 8-week-old WT and Usf1^−/− mice. As judged by cleaved caspase-3 staining, the incidence of apoptosis within the GFRα1-expressing A_undiff spermatogonia was generally low irrespective of genotype, which is in line with earlier data (19) [and in an online repository (36)]. In contrast, GFRα1-positive A_undiff spermatogonia were proliferatively active both in WT and Usf1^−/− mice, as judged by proliferation marker Ki-67 staining (Fig. 6A and 6B). Interestingly, as illustrated in Fig. 6B and in an online repository (36), areas where GFRα1-positive cells were present at a very high density were occasionally encountered in the Usf1^−/− seminiferous tubules. This prompted us to study proliferation of GFRα1-expressing A_s and A_pr spermatogonial cells that are the main constituents of the stem cell pool under steady-state conditions. While the majority of GFRα1-positive A_s cells were Ki-67–negative in the WT control testis, the situation was the opposite in the Usf1^−/− mice (Fig. 6C and 6D). A similar trend was observed in GFRα1-positive A_pr cells, but this difference was not statistically significant. Based on these data, we conclude that proliferation-coupled exhaustion contributes to the depletion of stem cells in the Usf1^−/− testis.

Figure 6.

Spermatogonial stem cells are hyperproliferative in the absence of Usf1. (A and B) Representative whole-mount immunofluorescence staining of 8-wk (A) control and (B) Usf1^−/− seminiferous tubules showing areas where GFRα1-positive cells were found accumulated. Only a subset of GFRα1-positive cells also stain for Ki-67. GFRα1-negative cells are differentiating spermatogonia that are continuously engaged in the cell cycle and are thus positive for Ki-67. (C) Assessment of proliferation within the GFRα1-positive undifferentiated spermatogonia. Blue arrow points to a Ki-67–positive (proliferatively active)/GFRα1-positive A_s spermatogonium, and yellow arrows indicate Ki-67–negative (nonproliferative)/GFRα1-positive A_s spermatogonia. GFRα1-positive/Ki-67–positive A_pr spermatogonia are indicated by the white arrow. (D) Quantitation of Ki-67/GFRα1 double-positive A_s and A_pr spermatogonia in mice of the indicated genotypes. Scale bars, 50 μm. Error bars represent mean ± SD, and P values are from unpaired t tests. **P < 0.01.

Discussion

To our knowledge, this study constitutes the first in vivo assessment of the role of USF1, a ubiquitously expressed transcription factor, in the maintenance of spermatogenesis. Loss of Usf1 leads to age-related decline in sperm production, most likely due to depletion of spermatogonial stem cells. Even though young Usf1^−/− adult mice still hosted relatively normal spermatogenesis, the spermatogenic defect became obvious by 12 weeks of age and continued to exacerbate thereafter. This is a characteristic of stem cell maintenance failure, as has been previously demonstrated, for example, in Plzf- (20, 21), Taf4b- (22), and Erm- (90) deficient mice. Typically, some areas within the seminiferous tubules are able to maintain stem cells for a longer time, but the number of such areas, or niches that they contain, decreases with age, whereas tubules that contain only the basal layer of the seminiferous epithelium and are devoid of germ cells become more common. As an intermediate step, tubules that lack one to three layers of spermatogenic cells are observed. If stem cells are depleted, 35 days are needed by spermatogenesis to clear the tubule of germ cells. Notably, we also observed tubules that lacked spermatogenic cell layers at the end of differentiation hierarchy (i.e., spermatids) but retained the meiotic and mitotic populations. Similarly, tubule cross-sections missing any single layer were occasionally noted. This implies that in the Usf1^−/− testis not every cycle is able to give rise to differentiating progeny and that the stem cell compartment first functions less efficiently before it collapses.

The significance of mitotic quiescence in the long-term maintenance of stem cells is widely appreciated. Hence, the continued engagement of SSCs in the cell cycle provides an attractive explanation for the progressive spermatogenic failure in Usf1^−/− mice. Normally, A_undiff spermatogonia exit from the cell cycle at epithelial stage II and a subset of them becomes sensitive to retinoic acid as a result of differentiation-priming activity of Wnt signaling, as well as by upregulation of retinoic acid receptor γ (RARg) (7, 87, 91, 92). Expression of RARg and associated genes, including neurogenin-3 (Ngn3) and Sox3, thus delineate A_undiff spermatogonia into differentiation-primed and stem subsets (4–6, 57, 91). Interestingly, the mechanisms responsible for the A_undiff cell cycle exit are essentially undefined. Notably, however, Gdnf is expressed at the lowest level at stages VII and VIII, that is, the same stages where the early phase of differentiation-inducing RA pulse is recorded (17, 92–96). Similarly, A_undiff spermatogonia re-enter the cell cycle at stage X in synchrony with reactivation of Gdnf expression and a sharp decline in RA levels (92, 96). A_undiff spermatogonia mitotic activity thus seems to be intimately coupled with GDNF availability. We speculate that elevated levels of GDNF at stages II to VIII in Usf1^−/− mice may contribute to prolonged proliferation of GFRα1-positive SSCs and to the inability to induce formation of the progenitor subset (97). This scenario would result in smaller cohorts of differentiating progeny and ultimately in fewer sperm, as recently suggested by Sharma and Braun (12). Moreover, prolonged engagement in the cell cycle may eventually lead to proliferation-coupled exhaustion of GFRα1-expressing spermatogonia, thus providing an explanation for the stem cell depletion phenotype (Fig. 7).

Figure 7.

Proposed model for USF1-dependent, proliferation-coupled stem cell exhaustion. In the WT testis, stem cells continually exit (at stage II) and reenter (at stage X) the cell cycle as a result of the progress of the seminiferous epithelial cycle. Spermatogenesis is initiated (i.e., transition from A_undiff to type A1 differentiating spermatogonia) once every epithelial cycle at stages VII and VIII. A delicate balance between self-renewal vs differentiation prevails and the stem cell population is maintained while a sufficient but not excessive number of differentiating progeny are simultaneously produced during every epithelial cycle, enabling lifelong sperm production from the SSC niche. In the Usf1^−/− testis, stem cells become continually engaged in the cell cycle, resulting in their proliferation-coupled exhaustion and inability to maintain the stem cell pool. Once the niche is depleted of stem cells, germ cells are lost from the locale layer by layer as a result of seminiferous epithelial cycle progression and the spermatogenic program. Symbols used to indicate different germ cell types are described in Fig. 1F.

Gdnf is an FSH-regulated gene (14, 16, 17, 98). Interestingly, plasma FSH levels in the Usf1^−/− mouse were elevated, which may contribute to increased Gdnf expression in stages II to V (II to VIII). The significance of this connection, however, is unclear; the role of FSH in Gdnf regulation under physiological conditions has been recently called into question (85). We initially speculated that another major endocrine factor, testosterone, might be more important for the phenotype. Testosterone is crucial for spermatogenesis, and its levels inside the testis are around one order of magnitude higher than in plasma. Once deemed indispensable, recent research has shown that high ITT is not essential for sperm production and that spermatogenesis can be initiated and maintained at a testosterone concentration similar to what is measured in plasma (99). Testosterone has also recently been implicated in regulation of the spermatogonial stem cell niche via GDNF and WNT5A (13, 85). It is therefore possible that the stage II– to stage V–specific elevated Gdnf levels are due to high ITT measured in Usf1^−/− mice. These stages have previously been shown to display a high sensitivity to androgen action (100). WNT5A is a developmental regulator of the spermatogonial stem cell pool, and its expression is downregulated by testosterone (85). We did not, however, detect any changes in Wnt5a mRNA levels in Usf1^−/− testis.

In summary, the paracrine milieu in Usf1^−/− testes was somewhat altered compared with WT testes. However, the changes were modest, and no consistent reduction in the expression of the studied paracrine factor–encoding genes was observed. Moreover, the levels of endocrine factors were at a sufficiently high level to maintain spermatogenesis in the Usf1^−/− testis. Sertoli cells in adult Usf1^−/− mice had matured normally and exhibited all characteristic aspects of adult-type Sertoli cells. Although we cannot rule out a role for a defunct SSC niche in Usf1^−/− testis, it seems likely that the phenotype is mostly of spermatogonial origin and that USF1 is needed for the maintenance of the spermatogonial stem cell pool in a cell-autonomous fashion. We propose that in the Usf1^−/− testis spermatogonial stem cells become continually engaged in the cell cycle, resulting in their depletion with age. This manifests itself as an accumulation of tubules displaying poor spermatogenic differentiation, smaller cohorts of differentiating germ cells, and disrupted layering of the seminiferous epithelium, collectively resulting in age-related reduction in sperm production.

There are numerous different mechanisms by which loss of USF1 may contribute to the loss of spermatogonial stem cells in a cell-autonomous fashion. Namely, among its many functions, USF1 has been implicated in the control of cellular proliferation. Not only have several tumor suppressor genes been recognized as direct USF1 targets [e.g., PTEN, APC, p53 (101–104)], but USF1 also stabilizes p53 (105), opposes the action of Myc at the transcriptional level (106), and may contribute to cellular immortality by maintaining telomerase reverse transcription expression (107). Hence, the effect of USF1 on cellular proliferation is considered growth inhibitory. Although USF1 has been shown to fulfill many aspects of a classical tumor suppressor protein, a connection between USF1 deficiency and increased proliferation or tumor formation has not been demonstrated. To our knowledge, this is the first direct demonstration that loss of USF1 results in higher cellular proliferation in vivo. Paradoxically, however, increased stem cell proliferation does not result in tissue growth but rather in hypoplasia due to a stem cell maintenance defect. It remains to be thoroughly investigated whether (partial) depletion of stem cells contributes to tissue growth defects in tissues other than the testis, in Usf1^−/− mice.

Deficiency of USF1 or USF2 can typically be compensated for by the formation of USF2 or USF1 homodimers, respectively (29, 108). In Usf1^−/− testes, USF2 homodimers are expected to compensate for lack of USF1 at most USF-dependent gene promoters, as demonstrated by Hermann et al. (71) for the Fshr gene in Sertoli cells. In agreement with this study, we also found that Fshr expression was unaffected in the absence of USF1. Furthermore, loss of USF1 activity neither affected expression of genes involved in Sertoli cell maturation or function nor had an overt impact on the stem cell niche. Thus, USF2 is likely sufficient to compensate for USF1 loss in paracrine and autocrine regulation by Sertoli cells. This, however, is likely not the case for the testicular stem cell pool. We speculate that there are USF1-regulated genes in undifferentiated spermatogonia whose transcription cannot be maintained by USF2 homodimers, and that lack of their expression results in the gradual depletion of stem cells in a cell-autonomous fashion.

USF1 deficiency in mice and reduced USF1 expression in humans has been shown to help maintain a beneficial lipid profile (i.e., higher high-density lipoprotein and lower triglycerides), insulin sensitivity, and to protect against hardening of the arteries. Therefore, targeting USF1 has excellent clinical potential in the treatment of obesity, diabetes, and cardiovascular diseases (35). In this study, we have shown that loss of USF1 also has adverse effects on reproductive function, a finding that might intuitively raise doubts about appropriateness of USF1 as a drug target. However, Usf1 heterozygous mice, which also displayed reduced weight gain and more beneficial lipid profiles (35), did not show spermatogenic defects. Thus, whereas complete absence of USF1 leads to impaired spermatogenesis, partial loss (Usf1^+/− mice) does not appear to have these effects. Thus, our present findings are still compatible with our previous proposal (35) of the potential of USF1 modulation as a therapeutic treatment strategy for cardiometabolic disease. We have uncovered a significant novel role for USF1 as a factor required for spermatogenesis, highlighting the varied physiological roles of this transcription factor.

Abbreviations:

Abbreviations:
 
• A_al

  A-aligned

 
• A_pr

  A-paired

 
• A_s

  type A-single

 
• A_undiff

  A-undifferentiated

 
• AR

  androgen receptor

 
• DAPI

  4′,6-diamidino-2-phenylindole

 
• FSHR

  FSH receptor

 
• GDNF

  glial cell line–derived neurotrophic factor

 
• GFRα1

  GDNF family receptor α1

 
• ITT

  intratesticular testosterone

 
• KO

  knockout

 
• PFA

  paraformaldehyde

 
• PLZF

  promyelocytic leukemia zinc finger

 
• SSC

  spermatogonial stem cell

 
• USF

  upstream stimulatory factor

 
• WT

  wild-type

 
• WT1

  Wilms tumor 1

Acknowledgments

We are grateful to Hannu Sariola, Jarkko Soronen, Juha Tapanainen, Taneli Raivio, Pekka Katajisto, and Timo Tuuri (University of Helsinki, Helsinki, Finland) for suggestions and advice on experimental procedures. We thank Elli Aska, Nanna Sarvilinna, Kul Shrestha, and Barun Pradhan (Kauppi laboratory) for assistance with experimental procedures. Assistance was also provided by the following core facilities at the University of Helsinki: the Tissue Preparation and Histology Unit, the Laboratory Animal Center, and the Biomedicum Imaging Unit, as well as the Cell Imaging Core at the Turku Center for Biotechnology.

Financial Support: This work was supported by a Biomedicum Helsinki Foundation grant to I.F.; grants from the Sigrid Jusélius Foundation, Finska Läkaresällskapet, and Paavo Nurmi Foundation to P.-P.L.; Finnish Foundation for Cardiovascular Research, Academy of Finland Grant 257545 and grants from the Magnus Ehrnrooth Foundation and Jane and Aatos Erkko Foundation to M.J.; grants from the Academy of Finland, Sigrid Jusélius Foundation, Novo Nordisk Foundation, and Turku University Hospital to J.T.; an Emil Aaltonen Fellowship to J.-A.M.; and by Academy of Finland Grants 25996, 263870, and 292789 and grants from the Biocentrum Helsinki and the Sigrid Jusélius Foundation to L.K.

Author Contributions: I.F., J.-A.M., and LK: conceptualization. I.F., S.C.-M., G.H., P.-P.L., M.J., J.-A.M., and L.K.: data curation. I.F. and J.-A.M.: formal analysis; I.F., P.-P.L, M.J., J.T., J.-A.M., and L.K.: funding acquisition. I.F., S.C.-M., G.H., M.M.T., P.-P.L., M.T., and J.-A.M.: investigation. I.M., S.C.-M., G.H., M.M.T., P.-P.L., M.T., J.-A.M., and L.K.: methodology. I.F., J.-A.M., and L.K.: project administration. I.F., G.H., M.J., J.T., J.-A.M., and L.K.: resources. I.F., G.H., and J.-A.M.: software. J.-A.M. and L.K.: supervision. I.F., S.C.-M., and J.-A.M.: validation. I.F., G.H., S.C.-M., and J.-A.M.: visualization. I.F. and J.-A.M.: preparation of original draft of manuscript. I.F., S.C.-M., G.H., P.-P.L., M.J., N.K., J.T., J.-A.M., and L.K.: review and editing of final manuscript.

Current Affiliation: I. Faisal’s current affiliation is the Sunnybrook Health Sciences Center, Faculty of Medicine, University of Toronto, Toronto, Ontario M4N 3M5, Canada. P.-P. Laurila’s current affiliation is the École Polytechnique Fédérale de Lausanne, 1015 Lausanne, Switzerland. M. Tuominen, M. Tumiati, and L. Kauppi are currently affiliated with the Research Program in Systems Oncology, Research Programs Unit, Faculty of Medicine, University of Helsinki, 00290 Helsinki, Finland; the Genome-Scale Biology Research Program is closed.

Disclosure Summary: The authors have nothing to disclose.

Author notes

J.-A.M. and L.K. contributed equally to this study.