Induced Autoimmunity against Gonadal Proteins Affects Gonadal Development in Juvenile Zebrafish

A method to mitigate or possibly eliminate reproduction in farmed fish is highly demanded. The existing approaches have certain applicative limitations. So far, no immunization strategies affecting gonadal development in juvenile animals have been developed. We hypothesized that autoimmune mechanisms, occurring spontaneously in a number of diseases, could be induced by targeted immunization. We have asked whether the immunization against specific targets in a juvenile zebrafish gonad will produce an autoimmune response, and, consequently, disturbance in gonadal development. Gonadal soma-derived factor (Gsdf), growth differentiation factor (Gdf9), and lymphocyte antigen 75 (Cd205/Ly75), all essential for early gonad development, were targeted with 5 immunization tests. Zebrafish (n = 329) were injected at 6 weeks post fertilization, a booster injection was applied 15 days later, and fish were sampled at 30 days. We localized transcripts encoding targeted proteins by in situ hybridization, quantified expression of immune-, apoptosis-, and gonad-related genes with quantitative real-time PCR, and performed gonadal histology and whole-mount immunohistochemistry for Bcl2-interacting-killer (Bik) pro-apoptotic protein. The treatments resulted in an autoimmune reaction, gonad developmental retardation, intensive apoptosis, cell atresia, and disturbed transcript production. Testes were remarkably underdeveloped after anti-Gsdf treatments. Anti-Gdf9 treatments promoted apoptosis in testes and abnormal development of ovaries. Anti-Cd205 treatment stimulated a strong immune response in both sexes, resulting in oocyte atresia and strong apoptosis in supporting somatic cells. The effect of immunization was FSH-independent. Furthermore, immunization against germ cell proteins disturbed somatic supporting cell development. This is the first report to demonstrate that targeted autoimmunity can disturb gonadal development in a juvenile fish. It shows a straightforward potential to develop auto-immunization-based technologies to mitigate fish reproduction before they reach maturation. However, the highly variable results between treatments and individuals suggest significant optimization should be performed to achieve the full potential of this technology.


Introduction
Control of reproduction in farmed, pest, and pet animals is an important issue with multiple implications in economy, environment, and welfare [1,2,3]. Unwanted sexual maturation of fish before they reach market size remains a significant constraint in aquaculture. For example, in farmed Atlantic cod (Gadus morhua), maturation often occurs in the second year, while in Atlantic salmon (Salmo salar) it can happen after just one winter in seawater. Both cases result in loss of growth, condition, and flesh quality [4,5,6]. In addition to economic consequences, precocious sexual maturation of farmed fish creates a threat of genetic pollution to natural populations because of escapes and uncontrolled gamete production [7]. The creation of mono-sex stocks can mitigate some concerns regarding gamete production and growth in some species such as catfish (Pelteobagrus spp.) or tilapia (Oreochromis spp.) [8,9]. However, it does not resolve concerns about reproductive potential of escapees and flesh quality of farmed fish.
Therefore, a method to mitigate or possibly eliminate gonadal growth in farmed fish is highly demanded. Transgenic [10], morpholino-based knockdown [11], or induced triploidization [12] approaches, resulting in partial or complete gamete ablation and consequently in reproduction disability, have their apparent limitations related to welfare, food safety, costs, or application potential to massscale production. A method based on intrinsic properties of the immune system, which does not require genetic modifications, is cost-effective and applicable at the industrial production level, would overcome the aforementioned limitations.
The majority of efforts in immunocontraception and immunosterilization have been focused on the mammalian reproductive system with a purpose to control wildlife populations [3,13,14,15]. Usually the treatments stimulate the production of autoimmune antibodies to either bind to the gametes directly and prevent their function, or interfere with the GnRH pathway to prevent gamete production; some mammalian vaccines have resulted in significantly impaired fertility [14]. While immunocontraception serves the purpose of preventing outbreak of mammalian pest populations, it is not suitable for aquaculture as it does not affect gonadal development and the associated deteriorated growth and flesh quality. A study on rainbow trout (Oncorhynchus mykiss) demonstrated that injection of a gonadal extract resulted in a limited autoimmune response in gonads [16].
However, that study was performed on few individuals only and no recent published studies have explored this concept further.
In the current study, we have asked whether the vaccination against specific targets in a juvenile gonad will produce an autoimmune response, and consequently, if it will disturb gonadal development. We chose zebrafish (Danio rerio) as a model. Zebrafish undergo a juvenile hermaphroditism sex differentiation pattern [17,18,19]. The gonad first develops perinucleolar oocytes around 4 weeks post-fertilization (pf). While the timing of sexual differentiation can vary, a minimum length of 8 mm is a pre-requisite for development beyond an undifferentiated gonad [17]. Then either ovary is developed, or the oocytes degenerate and the gonad transforms into testis, with the first spermatocytes observed around 7 weeks pf. Completion of sexual differentiation occurs by weeks 11-12 pf [17]. However, the timing of this process is widely variable between individuals, as well as depending on fish strain and conditions.
After preliminary studies, we selected three proteins essential for gonad development as potential targets for induced autoimmunity: gonadal somaderived factor (Gsdf), growth differentiation factor 9 (Gdf9), and lymphocyte antigen 75 (Cd205/Ly75). Gsdf is a member of TGFb superfamily expressed in supporting somatic cells of ovary and testis, granulosa and Sertoli cells, respectively [20]. Gsdf expression in both sexes is highest when germ cells are in the spermatogonial or oogonial stages of gametogenesis, while expression decreases when germ cells enter the meiotic and postmeiotic stages [20]. This mechanism appears to be conserved in teleosts, as it is similar in rainbow trout, medaka (Oryzias latipes), and zebrafish [20,21,22,23]. In rainbow trout, it was demonstrated that gsdf knock-down during embryogenesis suppressed germ cell populations, while recombinant Gsdf was used for dose-dependent proliferation of type A spermatogonia in testis cell culture [21]. It has been suggested that since oocytes do not proliferate under this treatment, gsdf may have a different function in ovaries than testes.
Gdf9 is another member of the TGFb superfamily secreted from germ cells in both sexes and is involved in follicle development [24,25,26,27,28]. In gdf9 knock-out mice, folliculogenesis stops at the primary follicle stage [29]. In male cats, gdf9 is expressed specifically within the germ cell lineage and is involved in signaling to Sertoli and Leydig cells to regulate tight junctions [28]. In zebrafish, gdf9 is expressed predominantly in testes [30], but also plays a role in ovarian follicle development, where it regulates tight junction integrity between oocytes and follicle cells [26].
Lastly, Cd205 is an endocytic type I C-type lectin-like receptor belonging to the macrophage mannose receptor (MMR) family and consisting of a single polypeptide chain. In mammals, Cd205 is strongly expressed in dendritic cells, where the protein acts as an endocytic receptor, binding with antigens and carrying them through the MHC class I and II pathways to present them to CD4+ and CD8+ T cells [31,32,33]. In fish, cd205 is expressed in multiple tissues. However, within the gonad it is expressed exclusively in the germline [34,35]. In rainbow trout, the protein was localized to the surface of early germ cells, mainly primordial germ cells, type A spermatogonia, oogonia, and cortical alveoli oocytes. It was not found in spermatocytes, spermatids, spermatozoa, and vitellogenic oocytes [34]. While its function in teleost germ cells is unknown, the protein likely maintains a role in antigen presentation and related immune function.
In the current study, we immunized 6 week-old juvenile zebrafish against Gsdf, Gdf9, or Cd205 during early sexual development in order to induce autoimmune response in gonads. We gave a booster injection 15 days later, and sampled the fish at both 15 and 30 days after the initial treatment. We used in situ hybridization to localize transcripts encoding the target proteins. We performed gonadal histology, quantitative real-time PCR (qPCR) of immune-, apoptosis-, and gonad development-related genes, as well as whole-mount immunohistochemistry for the pro-apoptotic protein Bcl2-interacting-killer (Bik). We have demonstrated that gonadal development in juvenile zebrafish was affected via autoimmune mechanism.

Results
Localization of gsdf, gdf9, and cd205 transcripts within the gonad In situ hybridization (ISH) targeting transcripts of gsdf, gdf9, and cd205 was performed on sections of 7 month-old zebrafish gonads. As a reference for normal gonadal development, hematoxylin-eosin (HE) staining was also performed. The zebrafish ovary was composed of germ cells at various stages of developmental advancement (Fig. 1A). ISH of gsdf showed the transcripts were localized in supporting somatic cells surrounding stage 1 and stage 2 primary oocytes, as well as larger previtellogenic oocytes (Fig. 1B). No signal was detected around vitellogenic oocytes. ISH for gdf9 and cd205 showed the same expression pattern (  Fig. 1C, D).
The mature zebrafish testis was composed of clusters of germ cells at various stages of spermatogenesis ( Fig. 2A). gsdf transcripts showed strong signal in cells surrounding germ cell clusters (Fig. 2B). Although the cell type from which the signal originated was not determined, no signal in spermatocytes, spermatids, or spermatozoa was observed. Very weak signal was detected for gdf9 ISH. It was not certain what cell types expressed gdf9 transcripts (Fig. 2C). ISH for cd205 resulted in a comparatively weak signal, which appeared to originate from spermatogonia, but not spermatocytes, spermatids or sperm (Fig. 2D).

Preliminary trial: effect of autoimmunization on adult female growth
To determine whether immunization against gonadal targets influences gonad weight and overall growth of zebrafish, a preliminary trial was performed on adult zebrafish females. Fish were immunized against four different proteins: Zona pellucida C (Zpc), Cd205, Insulin-like growth factor 3 (Igf3), or Riboflavin carrier protein (Rcp). At 10 days post-treatment (10 dpt), significant effect of treatment on fish weight (p50.004) and gonadosomatic index, GSI% (p,0.001) was found. Fish from anti-Zpc and anti-Cd205 treatments had significantly lower weight compared to controls; the length of treated fish was not different from controls (Table S1). The reduction in total weight in anti-Zpc and anti-Cd205 treatments likely resulted from a loss in gonadal weight, because the GSI% of fish in both treatments was significantly lower than in controls.
At 20 dpt, no significant differences in weight and length of fish were found. The effect of treatment on GSI% was significant (p,0.001). Fish from anti-Zpc variant had significantly lower GSI% than controls. The GSI% of fish from anti-Zpc and anti-Cd205 treatments was significantly lower than in PBS-injected controls. Other treatments did not differ from either non-injected or PBS-injected controls (Table S1).
At 30 dpt, significant effects of treatment on fish weight (p50.003) and GSI% (p,0.001) were observed. Fish from the anti-Zpc and anti-Igf3 treatments were significantly heavier than control fishes; however, their GSI% did not differ Figure 1. Sections of ovaries from 7 month-old zebrafish. A) Representative haematoxylin-eosin (HE) staining of the mature zebrafish ovary. B) in situ hybridization (ISH) of gsdf transcripts in the ovary; staining shows gsdf transcripts are localized in somatic cells surrounding primary oocytes. The staining is strongest in stage 2 primary oocytes (Poc2) and previtellogenic oocytes (Poc3). C) ISH of gdf9 transcripts in the ovary; strong signal is detected in stage 1 and 2 primary oocytes (Poc1 and Poc2). Weak signal is detected in previtellogenic oocytes (Poc3), while no signal was detected in vitellogenic oocytes (Poc4). D) ISH of cd205 transcripts in the ovary; strong signal is detected in stage 1 and 2 primary oocytes, with weak signal in previtellogenic oocytes, and no transcript detected in vitellogenic oocytes. All scalebars represent 200 mm. significantly from control fish (Table S1). The GSI% of fish treated with anti-Cd205 was significantly lower than in any other treatment or in control fish.

Effect of treatment on juvenile zebrafish growth
In the main experiment, five treatments against three gonadal proteins were tested: anti-Gsdf (two treatments, -a and -b), anti-Gdf9 (two treatments, -a and -b), and anti-Cd205. Fish were measured for total length prior to the first injection (day 0) and at 15 and 30 dpt, and weighted at 15 and 30 dpt. Length of fish at the day 0 was homogenous among treatment groups (p.0.05). At 15 dpt, fish from the anti-Gsdf-b treatment had significantly lower weight than controls, but no significant differences in length were recorded. At 30 dpt, more distinct retardation in growth was observed: fish from 3 out of 5 treatment variants had significantly lower weight than control fish; of them, fish from two variants (anti-Gsdf-b and anti-Cd205) were significantly shorter than control fish (Fig. 3). staining shows strong signal throughout the testis, although it is unclear from which cell type it originates; no staining is visible in spermatocytes or spermatids. Spermatogonia and spermatozoa could not be identified. C) ISH of gdf9 transcripts in testis; faint signal was found throughout the testis. It is unclear which cell types the signal is localized to. D) ISH of cd205 transcripts in the testis; the signal is strongest in spermatogonia, while it is not visible in spermatocytes or spermatids. Spermatozoa could not be identified. All scalebars represent 100 mm.

Sexually dimorphic gene expression
Total RNA was extracted from individuals sampled at both 15 and 30 dpt. Quantitative real-time PCR was then performed to compare relative expression of eight immune-, apoptosis, and gonadal-related genes. All investigated genes showed highly significant sexually dimorphic expression in juvenile zebrafish gonads (p,0.001). Expression of T-cell receptor alpha constant (tcrac), vasa, and cytochrome P450 gonadal isoform (cyp19a1a) was female-enhanced, whereas expression of immunoglobulin M heavy chain secretory form (igm), bcl-2interacting-killer (bik), gsdf, inhibin alpha subunit (inha), and anti-Müllerian hormone (amh) was male-enhanced (Table 1 and Table S2). Correlations between transcript relative abundances followed this pattern: testis-enhanced bik, gsdf, inha, and amh showed high and positive correlation with each other in both control and treated groups, while igm transcript level showed high and positive correlation with bik, gsdf, and amh ( Table 2). Ovary-enhanced tcrac, vasa, and cyp19a1a transcript levels correlated highly and positively with each other (  Table 2). Therefore, analysis of the effect of autoimmunization on gene expression was conducted on males and females separately.

Effect of treatment on gene expression
Generally, transcript levels showed high variation within all treatments; therefore, the differences in average expression between groups, although frequently considerable, were not always significant. Nevertheless, the effect of the autoimmunization on gene expression was found in all treatments. Certain sexand time-specific patterns in gene expression were observed. In males, downregulation in expression of vasa, inha, and amh at 30 dpt was observed in all treatments; whereas igm expression at 30 dpt was up-regulated in both sexes (  Table 3).
At 15 dpt, in ovaries, up-regulation of igm and bik transcripts correlated with down-regulation of vasa transcripts in anti-Gsdf-b, anti-Gdf9-a, anti-Gdf9-b, and anti-Cd205 treatments. In the anti-Gdf9-b and anti-Cd205 treatments this relationship was significant for all three genes. In testes, significant up-regulation of igm was observed in the anti-Gsdf-b treatment. No other significant changes in expression were observed. However, down-regulation of cyp19a1a transcripts was observed in all treatments (Table 3).
At 30 dpt, many significant changes in gene expression were observed. In ovaries, there was significant up-regulation of tcrac and vasa expression in the anti-Gsdf-a treatment. In the anti-Gdf9-a treatment, there was significant upregulation of igm transcripts. The anti-Gdf9-b treatment resulted in significant upregulation of gsdf and down-regulation of amh transcripts. In the anti-Cd205 treatment, there was significant up-regulation of igm and down-regulation of amh and cyp19a1a transcripts. In testes, all treatments resulted in significant downregulation of inha and amh transcripts. The anti-Gdf9-b and anti-Cd205 treatments also resulted in significant down-regulation of vasa and gsdf transcripts. Significant up-regulation of igm expression was observed in the anti-Gsdf-a treatment ( Table 3).

Effect of treatment on gonadal histology
At each sampling point, sections of gonads were taken and stained with haematoxylin-eosin. In total, 64 ovary sections were examined: 21 control (7 at 15 dpt, and 14 and 30 dpt), 7 anti-Gsdf-a (2 at 15 dpt and 5 at 30 dpt), 7 anti-Gsdf-b (4 at 15 dpt and 3 at 30 dpt), 11 anti-Gdf9-a (7 at 15 dpt and 4 at 30 dpt), 6 anti-Gdf9-b (30 dpt only), and 12 anti-Cd205 (2 at 15 dpt and 10 at 30 dpt). In controls at 15 dpt, theovaries consisted of a uniform mass of previtellogenic oocytes ( Fig. 4A), whereas by 30 dpt, the ovaries often contained previtellogenic and sometimes vitellogenic oocytes with a thick zona radiata surrounded by even follicular layers (Fig. 4B). No deviation from this pattern was observed in any control fish. In the treated fish, however, developmental pathology related to zona radiata development and oocyte atresia was observed in a number of samples from the anti-Cd205 and anti-Gsdf-a treatments (Fig. 4C, D).
Testis sections from 41 males were examined: 9 control (6 at 15 dpt and 3 and 30 dpt), 7 anti-Gsdf-a (2 at 15 dpt and 5 at 30 dpt), 4 anti-Gsdf-b (1 at 15 dpt and 3 at 30 dpt), 11 anti-Gdf9-a (4 at 15 dpt and 7 at 30 dpt), 2 anti-Gdf9-b (30 dpt only), and 8 anti-Cd205 (4 at 15 dpt and 4 at 30 dpt). The treatment effect was primarily through retarded growth and developmental advancement. Control testis at 15 dpt typically consisted of several stages of well-developed clusters of spermatogonial, spermatocyte, and spermatid cells (Fig. 5A). At 30 dpt, testes were further developed and clusters of spermatozoa were visible (Fig. S1A). In contrast, fish from the anti-Gsdf-a and -b treatments at 30 dpt had distinctively smaller and less developed testes (Fig. 5B, C). In anti-Gdf9-a treated fish, some individuals had underdeveloped testes (Fig. 5D); in one extreme case, the testis consisted of undifferentiated gonocytes only (Fig. S1B). There was only a single male examined from the anti-Gdf9-b treatment; its testis was comparatively well developed (Fig. S1C). No obvious developmental retardation was found in the anti-Cd205 treatment (Fig. S1D).

Effect of treatment on gonadal cell apoptosis
Whole gonads were excised at both 15 and 30 dpt from fish of the tg(vas::egfp) line. Whole mount immunohistochemistry (IHC) was subsequently performed using a primary antibody specific for pro-apoptotic Bcl2-interacting killer (Bik) protein. In addition, several samples were stained with 49,6-diamidino-2phenylindole (DAPI) to visualize associated somatic cells. IHC was performed on 40 ovarian samples: 11 control (5 at 15 dpt, and 6 and 30 dpt), 8 anti-Gsdf-a (3 at 15 dpt and 5 at 30 dpt), 5 anti-Gsdf-b (2 at 15 dpt and 3 at 30 dpt), 9 anti-Gdf9-a (5 at 15 dpt and 4 at 30 dpt), and 7 anti-Cd205 (4 at 15 dpt and 3 at 30 dpt). The general organization of the germ cell and somatic lineages in the ovary is shown in Fig. 6 (A, B). In control ovaries, there was generally a weak signal originating from primary oocytes, but not oogonia (Fig. 6C, D). Also, small, unidentified somatic cells containing Bik at a low level were detected around the primary oocytes at both 15 and 30 dpt. In the treatment groups, Bik signal was clearly stronger and with a more widespread distribution. In both anti-Gsdf-a andb treatments at both 15 and 30 dpt, Bik was localized in stage 2 primary oocytes, but not in stage 1 oocytes (Fig. 6E R H). At 30 dpt, Bik signal was also detected in stage 3 previtellogenic oocytes. In the anti-Gdf9-a group, Bik protein had a relatively The pattern of correlations is similar in controls (n532) and in the treated groups (n577). Significant r-values (p,0.05) are marked with bold font. Significant correlations among male-enriched and female-enriched transcripts are denoted with M or F letters, respectively.
weak signal restricted to stage 1 and stage 2 primary oocytes. No variation in signal was observed between 15 and 30 dpt. However, at both time points the ovaries appeared to develop abnormally, with many smaller primary oocytes in a scattered distribution when compared to control fish (Fig. 6I, J). In the anti-Cd205 treatment, there was a strong Bik signal primarily localized within somatic cells surrounding stage 2 primary oocytes (Fig. 6K, L). This pattern was observed in six of the seven ovaries examined. While the identity of somatic cells could not be confirmed, their distribution suggested supporting follicular cells of the theca and/or granulosa The relative gene expression is shown as fold change compared to control fish. The values in bold represent a significant difference in expression (ANOVA and LSD test at p,0.05). Different antigens for the same target protein are indicated using -a or -b following the target protein name. All data were normalized using the reference genes rpl13a and ef1a. doi:10.1371/journal.pone.0114209.t003 lineage. In primary and previtellogenic oocytes, the Bik signal was similar to that in other treatment groups. There were 8 testis samples examined using IHC: 1 control at 30 dpt, 1 anti-Gsdf-a at 15 dpt, 1 anti-Gsdf-b at 30 dpt, 4 anti-Gdf9-a (3 at 15 dpt and 1 at 30 dpt), and 1 anti-Cd205 at 30 dpt. No Bik protein signal was found in the control fish (Fig. 7A). In the anti-Gsdf males, weak or no Bik signal was detected in the testis, which appeared underdeveloped as compared to the control fish (  Fig. 7B, C). In the anti-Gdf9-a treatment, weak Bik signal was detected at both 15 and 30 dpt, primarily in spermatogonia (Fig. 7D, E). In the anti-Cd205 treatment, the Bik signal in testis was strong compared to other treatment groups (Fig. 7F). However, it was not possible to determine the cell type the signal originated from.

Discussion
Immunization of juvenile zebrafish against its own gonadal proteins generally induced an autoimmune response, which consequently affected gonadal development, although the success within and between treatments was variable. The treatment success could be evidenced from differential expression of genes being immune-related or gonadal cell markers, gonad pathology, cell apoptosis, and retarded development. To our knowledge, this is the first demonstration that a peptide-based vaccine against structural or signaling components of a juvenile teleost gonad can affect gonadal development.

Transcripts encoding target proteins are localized in gonads
We confirmed that transcripts encoding the target proteins were expressed in gonads of zebrafish. gsdf expression was localized in the gonadal somatic cells (  Figs. 1B and 2B), which is in agreement with the previous report showing gsdf expression localized in granulosa and Sertoli cells of zebrafish [23]. Previous studies showed that strong expression of gdf9 in zebrafish primary growth oocytes (stages 1 and 2) was declining progressively with the follicle development [36]. Our results also demonstrated an inverse relationship between oocyte growth and signal intensity (Fig. 1C). Furthermore, although gdf9 transcripts were not detected in vitellogenic oocytes, this could be likely due to a lack of sensitivity in the ISH procedure, as previous work showed through RT-PCR that gdf9 transcripts were present in mature oocytes [36]. Although we found a weak signal of gdf9 expression in testis, its specific cell localization could not be established (  Fig. 2C). ISH for cd205 transcripts was generally consistent with findings in rainbow trout and bluefin tuna (Thunnas orientalis) [34,35]. In rainbow trout, it was reported that cd205 transcripts were detected in oogonia, nucleolus-stage oocytes, and type A spermatogonia, whereas in bluefin tuna testis, the transcripts were detected in type A spermatogonia. We detected cd205 transcripts in stage 1 and 2 primary oocytes, some stage 3 previtellogenic oocytes, as well as spermatogonia (Figs. 1D and 2D).

Treatment against gonadal proteins induces an immune response
Changes in expression of tcrac, which is a mature T-cell receptor marker gene [37], and igm, which is expressed in both secreted and membrane forms of zebrafish Immunoglobulin M (IgM) antibodies [38], have shown the effect of treatments on the immune system. The zebrafish immune system is both morphologically and functionally mature by the age of 4-6 weeks; by this time, mature T and B lymphocytes can be detected, as well as secreted immunoglobulins [37]. The three organs responsible for the majority of immune cell development are the spleen, thymus, and head kidney [37,39]. Because these organs were not included in the trunk section from which RNA was extracted, we only measured immune cell concentration within the kidney, trunk muscle tissues, digestive system, and gonads.
The T lymphocytes are fundamental to the vertebrate specific immune response. These cells identify antigens through a CD3-associated antigen (Ag)specific and heterodimeric T cell receptor (TCR) expressed on the cell surface [40]. The majority of mature T-cells displays a ab TCR which recognizes peptides presented on the surface of antigen-presenting cells (APC) and reside in secondary lymphoid organs. The qPCR data showed that the autoimmunization strategy used in the present study was generally not successful in stimulating increase in Tcell concentration within the gonad (Table 3). Only the anti-Gsdf-a treatment resulted in a significant up-regulation of tcrac transcripts in females at 30 dpt. In contrast, while not statistically significant, the other treatments generally showed down-regulation of this marker. In cancer vaccine studies on mice, the use of vaccines with oil-based adjuvant showed that T-cells were far more likely localized in the site of vaccination rather than the sites of tumors, as the injected antigen persisted in oil droplets for extended periods of time [41]. In the current study, peptides were also immersed in an oil-based adjuvant and consequently oil droplets from the vaccines were still observable around the injection site at the time of sampling (not shown). This may explain the general down-regulation of tcrac expression, since T-cells could be localized in the site of injection rather than the gonad, and were not included in the sample examined using qPCR. If true, this would be a major limitation of an oil-based adjuvant.
B-cells are lymphocytes in the adaptive immune system, with primary roles in the production of specific antibodies, as well as memory retention of past infections. B-cell activity is performed through B-cell antigen receptor, a multicomponent complex on T-cell surfaces comprised of cell surface immunoglobulins, which mediates antigen binding and is responsible for B-lymphocyte maturation, maintenance, activation, and silencing [42,43]. In the present study, expression of igm, a B-cell marker, was generally up-regulated in the treated males and females (Table 3). B-cells can be present in the kidney as well as the blood; consequently, their specific recruitment to the gonad cannot be confirmed. However, secreted IgM antibodies in the blood would be able to interact with gonadal targets. Much of the elevation in igm concentration was likely due to immunization against keyhole limpet hemocyanin (KLH), which was used as an immunogenic carrier protein in the study. However, the differential expression of igm between different treatment and control groups can be explained by immune response to the specific antigen peptide. Furthermore, transcripts of the apoptosis-promoting gene bik were significantly up-regulated in females in anti-Cd205 and anti-Gdf9-b treatments at 15 dpt; it correlated with significant upregulation of igm expression and could suggest a direct link between increased antibody production and programmed cell death.

Effect of autoimmunization in gonads is FSH-independent
Follicle stimulating hormone (FSH) is a major regulator of both ovarian and testicular development. Major regulatory pathways are regulated by FSH, such as the Tgfb pathway, which includes gdf9, gsdf, amh and inha, and is involved in germ cell proliferation and differentiation [44]. Also, FSH is suggested to stimulate Sertoli cell proliferation [45]. In rainbow trout testes, FSH up-regulated inha expression while down-regulated amh expression [44]. In the present study, all treatments synergistically resulted in a significant down-regulation of both inha and amh transcripts in males at 30 dpt when compared to control fish (Table 3). These findings suggest a mechanism independent of FSH stimulation. There is evidence that the Gsdf role in germ cell proliferation does not involve the FSH regulatory pathway [44]. In the present study, the anti-Gsdf treatments resulted in significant down-regulation in amh and inha expression in testes at 30 dpt, indicating that the effect of the treatment on gene expression was not gonadotropin-mediated. Consequently, disturbances in gonadal development were likely the result of direct autoimmunization against gonadal proteins.

Immunization against germ cell proteins affects somatic supporting cells
Sertoli and granulosa cells derive from a common progenitor cell type and form the first layer of supporting somatic cells surrounding the germ cells in a testis and ovary, respectively [46]. In the gonad, inha and amh expression is specific to somatic supporting cells and not germ cells. Inhibitin functions as an endocrine hormone from the gonads to regulate FSH secretion in the pituitary. In zebrafish, inha is predominantly expressed in somatic follicle cells, increasing in expression as folliculogenesis progresses [47]. Amh expression is not detected in somatic cells of germ cell-ablated gonads of medaka, which indicates that signaling from germ cells is necessary to sustain amh expression in soma [48]. In the present study, both treatments against germ cell-specific proteins (anti-Cd205 and anti-Gdf9) resulted in down-regulation of inha and amh in testis and ovary at 30 dpt, while gsdf was down-regulated in the testis only (Table 3). This decrease in expression of somatic supporting cell-specific genes could result from a disturbance of somagermline cross-talk in gonads, caused by an immune response towards the germ cells. Disruption of signaling from germ cells to Sertoli and granulosa cells, affecting further development and functions of Sertoli and granulosa cells, is further supported by follicular histopathology (Fig. 4C, D), intensive apoptosis (  Fig. 6K, L), as well as developmental retardation (Fig. 5).

Effect of anti-Gsdf treatments
Anti-Gsdf treatments resulted in retarded body growth and retarded growth and development of the testes, particularly at 30 dpt (Fig. 5B, C and Fig. 7B, C). Consequently, there was significant down-regulation of inha and amh transcripts (Table 3). In contrast, the treatments had little to no effect on female fish at the macroscopic level. Interestingly, histology of some anti-Gsdf-a treated females showed that despite having well-developed ovaries, there were signs of atresia and inflammation (Fig. 4C). Because of the role of gsdf in spermatogonial proliferation [21], our results suggest that the treatment interrupted Gsdf signaling, and this suppressed spermatogonia proliferation and subsequent development. It is likely that the treatment was applied too late (6 week-old fish) to hamper ovarian germ cell proliferation, as there were already many primary oocytes present in the ovaries at that time. Since gsdf is still strongly expressed in somatic cells surrounding primary oocytes in 7 month-old zebrafish (Fig. 1B), the gene likely retains some function during the ovarian development. However, it is unclear whether the treatment failed to interrupt the signaling, or if the gsdf role is not essential for ovarian maturation.

Effect of the anti-Gdf9 treatments
Treatments against Gdf9 protein resulted in retarded testes development and abnormal ovary development. Furthermore, significant loss in weight, but not in total length was recorded (Fig. 3), which could result from smaller gonads. In testes, the significant down-regulation of inha and amh transcripts in both treatments, as well as vasa and gsdf transcripts in the anti-Gdf9-b treatment (  Table 3) was accompanied by retarded development (Fig. 5D). While it has not been confirmed in zebrafish yet, the role of Gdf9 in testes is likely similar to that in ovaries, which is to regulate tight junction integrity between cells of the germ cell lineage and gonadal somatic supporting cells [26,28]. Interestingly, the strong Bik signal observed in spermatogonia, but not in clusters of spermatocytes and spermatids which had already recruited Sertoli and Leydig cells (Fig. 7E), suggests that an earlier application of the treatment would be more efficient. In femals, significant up-regulation of igm and down-regulation of vasa transcripts showed that the effect was strongest at 15 dpt, suggesting that the anti-Gdf9 treatment should be done earlier in females than in males (Table 3). Furthermore, the significant up-regulation of bik transcripts at 15 dpt (Table 3), along with localization of Bik protein in primary oocytes (Fig. 6I, J) suggests that preventing Gdf9 signaling between primary oocytes and somatic cells promotes oocyte apoptosis. It is unclear if the treatment had any effect on preventing oocyte maturation, since the fish sampled for histology and IHC were smaller than the control fish, and therefore could not be used for direct comparisons. However, while control ovaries were compact, with a uniform gradient of growth, many anti-Gdf9-treated ovaries appeared abnormal, with a scattered distribution of primary oocytes of various sizes. This could result from impaired Gdf9 signaling, preventing the early stage oocytes from recruiting gonadal somatic cells.

Effect of the anti-Cd205 treatment
In females, the treatment resulted in reduced growth, elevated expression of immune-related genes, reduced expression of gonadal marker genes, localized signal of pro-apoptotic proteins within the gonad, and irregular gonadal development. The retarded growth at 30 dpt (Fig. 3) could likely result from a reduced ovary mass; as demonstrated in the preliminary experiment, anti-Cd205 treatment resulted in a significant reduction of GSI% in adult females (Table S1). The significant up-regulation of igm and bik, combined with significant downregulation of vasa transcripts in 15 dpt ovary (Table 3) suggests a targeted immune response resulting in germ cell ablation. This is supported by histopathological features observed in 30 dpt ovaries (Fig. 4D). The treatment had little to no effect to testes at 15 dpt, whereas significant down-regulation of germ cell (vasa) and supporting Sertoli cell (gsdf, inha, and amh) transcript markers (  Table 3) along with a strong Bik signal (Fig. 7F) was found at 30 dpt.
While the function of the Cd205 protein in teleosts has not been studied yet, it may retain some role in antigen presentation to immune cells. In Pacific bluefin tuna, cd205 transcripts were found not only in the gonad, but also the gills, liver, spleen, and pyloric caeca [35]. In teleosts, the spleen is one of the major lymphoid tissues [49], whereas the gills, spleen and pyloric caeca are important regions for exogenous antigen uptake [50]. In mice, aside from its role in antigen presentation through the MHC pathways, Cd205 is also involved in uptake of ''self'' antigens from cells undergoing apoptosis and necrosis to induce tolerance to these cells and prevent autoimmunity [33]. Cellular apoptosis is a normal part of gonadal differentiation and development in teleost fishes [17,18,19,51]. Therefore, teleost Cd205 may have an important role in promoting tolerance to self-antigens during early sexual development.

Conclusions
The present study demonstrates that immunization against specific targets in juvenile zebrafish gonads can induce an autoimmune response and, in consequence, cause disturbances in gonadal development manifested by growth and developmental retardation, intensive apoptosis, atresia, and disturbed transcript production. The treatment effect was likely FSH-independent. Targeting germ cell-specific proteins resulted in disturbed functions of somatic supporting cells, likely through disturbed cross-talk signaling. The anti-Gsdf treatments had little to no effect on females. However, testes were remarkably underdeveloped. The anti-Gdf9 treatments promoted apoptotic signals in testes, and caused abnormal development of ovaries. Finally, anti-Cd205 treatment stimulated a strong immune response in both sexes, resulting in strong proapoptotic signals in somatic cells and oocyte atresia. Further optimization of the treatment, including timing, dose, and type of antigen peptide, as well as investigating long-term effects of the treatment, should deliver a highly promising tool to control reproduction in fish.

Fish and husbandry
All husbandry and experimental procedures were performed in accordance with the Norwegian Regulation on Animal Experimentation (The Norwegian Animal Protection Act, No. 73 of 20 December 1974) and were approved by the National Animal Research Authority (Utvalg for forsøk med dyr, forsøksdyrutvalget, Norway) General License for Fish Maintenance and Breeding (Godkjenning av avdeling for forsøksdyr) no. 17. For the preliminary adult zebrafish trial, adults of mixed line were obtained from Febo Norge AS (Oslo, Norway). For the juvenile trial, adult zebrafish from three inbred lines were obtained from the Norwegian School of Veterinary Science (Oslo, Norway). These lines were: AB Tübingen (TAB), serving as a reference line; tg(vasa::vasa-EGFP)zf45 [52], showing GFP expression in the germ cell lineage; and the nacre (-/-) mutant line, lacking melanophores, thus being transparent. Fish were housed in a zebrafish recirculating system (Aquatic Habitats, Apopka FL, USA) and maintained using standard zebrafish procedures [53]. Water temperature was maintained at 28.0¡1.0˚C. Broodfish were housed in 10 L tanks at a density of 20 fish/tank with a 1:1 sex ratio. One month prior to spawning the fish were conditioned with SDS 400 zebrafish specific diet (Special Diet Services, Essex, United Kingdom) and freshly hatched Artemia sp. nauplii. Juvenile zebrafish were produced using standard breeding techniques [53]. Post hatch, zebrafish larvae were housed in 1L tanks with fine mesh baffles and restricted water flow. Starting at 5 days post hatch, fish were weaned using SDS 100 zebrafish specific diet (Special Diet Services). From days 14 to 21 they were given a mix of SDS 100 and Artemia nauplii. Post day 21 they were only fed Artemia nauplii.

Preliminary trial on adult females
Four experimental vaccines consisted of a synthetic peptide conjugated to the carrier protein keyhole limpet hemocyanin (KLH) (Table S3). Each peptide was chosen from the complete respective amino acid sequence based on predicted antigenicity [54]. The peptides were commercially synthesized at .80% purity (Thermo Fisher Scientific, Ulm, Germany) and conjugated to KLH using a maleimide activated conjugation kit following the manufacturer's protocol (Sigma Aldrich, Oslo, Norway). After conjugation, the protein conjugates were isolated through column chromatography (Sephadex G-25M, Sigma Aldrich) and eluted in phosphate buffered saline solution (PBS). Coupling efficiency was determined by comparing the absorbance at 412 nm from a cysteine standard assay with each peptides cysteine absorbance values before and after conjugation. Final protein concentration was estimated by measuring absorbance at 280 nm. Peptide-KLH conjugates were stored at 220˚C until immediately before use. After thawing, the conjugates were diluted in PBS (1.0 mg/mL) and emulsified by vortex at a 1:1 ratio with Freund's complete adjuvant (FCA, Sigma Aldrich).
In total, 485 adult zebrafish were injected with one of the four treatments or a PBS/FCA control. Because of a limited availability of males, only female zebrafish were considered for this experiment. Each treatment group initially consisted of 105 fish whereas the control group had 65 fish. Prior to immunization, zebrafish were anaesthetized in a bath of 150 mg/L MS-222 (Tricaine; Sigma Aldrich) buffered with 150 mg/L sodium bicarbonate (NaHCO 3 ). Fish were considered in surgical anesthesia (stage III) when they became unresponsive to physical stimuli but maintained opercular movement [55]. Fish were removed from the bath and placed on a wet sponge inside a Petri dish. Injections were made intraperitoneally, below the pectoral fin. All injections were 10.0 mL (5.0 mg peptide-KLH) volume. After injection, fish were placed in a recovery tank for 10 min. Injection mortality was 13 out of 485 injected fish. Once recovered, zebrafish were returned to housing tanks.
Zebrafish were sampled immediately prior to immunization. Treated fish were then sampled every 10 days for 30 days. Because of higher mortality, sampling of control fish was only done at 20 dpt. Prior to sampling, fish received an overdose of buffered MS-222 (200 mg/L) and remained in the solution for 10 min following cessation of opercular movement. Upon removal from the bath, the fish were patted dry and weight and fork length measurements were taken. Fish were decapitated before opening the body cavity and the gonads were excised and weighed.

Autoimmunization vaccine preparation
Peptides for three target proteins (Table 4) were designed based on predicted antigenicity [54]. Six custom peptides (two for each target protein) were synthesized and conjugated to KLH by Thermo Fisher Scientific. Upon arrival, lysophilized peptides were dissolved in PBS to make a stock solution of either 10.0 or 5.0 mg/L, depending on protein solubility. Treatments consisted of either a single antigen (anti-Gdf9 and anti-Gsdf) or, where possible, a combination of both antigens (anti-Cd205) emulsified 1:1 in FCA. For anti-Gsdf and anti-Gdf9 treatments, peptides were dissolved individually and had a final concentration of 2.5 mg/mL each. Because of high solubility, both Cd205 peptides were combined and retained a final concentration of 2.5 mg/mL.

Immunization
Zebrafish were first injected at 6 weeks post fertilization, with a booster treatment at 15 dpt and final sampling at 30 dpt. The choice of timing for these treatments was based on the window in which the zebrafish can be safely injected (minimum total length of 15 mm; personal observations), have a mature adaptive immune system [37], and have not yet developed a mature gonad [17,18]. In total, 329 fish (15.5¡2.8 mm average ¡ S.D.) were immunized with one of five vaccines or a PBS/FCA control (Table 5). Fish were anesthetized in a Petri dish filled with tank water containing 50 mg/L MS-222(Sigma Aldrich) buffered with 50 mg/L NaHCO 3 . Each anesthetized fish was first individually photographed and measured for total length using a Zeiss Axio Zoom v.16 microscope. Fish from the tg(vas::egfp) line were also screened for eGFP signal. For injections, fish were transferred to a wet sponge and injected with 2.0 mL of the experimental vaccine using a 50 mL syringe equipped with a 34g needle (Hamilton, Bonaduz, Switzerland). Post injection, fish were transferred to a Petri dish filled with tank water to monitor recovery for 10 min before returning to a housing tank. In total, there were 36 mortalities. For the booster treatment at 15 dpt, the protocol was the same except the vaccination volume was reduced to 1.0 mL and Freund's incomplete adjuvant (FIA) was used as the adjuvant. There was a single mortality from the booster injection.

Fish sampling
Fish were sampled at 15 and 30 dpt for quantitative real-time PCR (qPCR), histology, and immunohistochemistry (IHC). Furthermore, 7 month-old control fish from the TAB line were sampled for in situ hybridization (ISH). Prior to sampling, all fish were euthanized in buffered MS-222 (100 mg/L) so they could be photographed, screened for eGFP signal, and measured for weight and total length (Table 5). Comparisons for sex, weight, and total length were made using the reference TAB line only. The proportion of females:males was 55:45, not deviating from the expected 50:50 ratio (Chi-square test). qPCR analysis was performed solely on TAB line fish. Gonads could not be dissected from such a small fish; therefore, sample sections for analysis contained some fragments of trunk muscle tissue, swim bladder, and kidney, in addition to the gonad. Samples were placed in a 1.5 mL eppendorf tube and snap-frozen in liquid nitrogen. Samples were further stored at 280˚C until RNA extraction.
Fish from the nacre -/-and tg(vas::egfp) lines were used for histology and IHC, while the 7 month-old control fish samples for ISH were from the TAB line. The decapitated and truncated body was cut open posterior to the pectoral fins before placing the fish in either Bouin's (for histology and ISH) or 4% paraformaldehyde (PFA, Sigma Aldrich) solution (for IHC). Samples were left in a fixative overnight at 4˚C. Fixed samples were then dehydrated in a gradient series of ethanol (from 50% up to 100%).

Quantitative real time PCR
Total RNA was extracted using TRIzol Reagent (Invitrogen, Paisley, U.K.) following the manufacturer's protocol. RNA integrity was first assessed using electrophoresis on a 1% (w/v) agarose gel. Suitable samples were then quantified using NanoDrop ND-1000 (Thermo Scientific, Saven & Werner AS, Kristiansand, Norway). Approximately 1 mg of total RNA was used for cDNA synthesis using the QuantiTect reverse transcription kit (Qiagen, Nydalen, Sweden). All samples were treated with the gDNA wipeout buffer supplied with the QuantiTect reverse transcription kit for 5 min to remove genomic DNA contamination. 20-fold dilutions of cDNA were used for further analysis.
The eight genes selected for qPCR analysis were: T-Cell receptor alpha constant (tcrac), immunoglobulin M heavy chain secretory form (igm), Bcl2-interactingkiller (bik), vasa (vasa), gonadal somatic cell-derived factor (gsdf), inhibin alpha (inha), anti-Müllerian hormone (amh), and cytochrome P450 aromatase, gonadal isoform (cyp19a1a). The genes used for reference were beta actin (ß-actin), ribosomal protein L13 alpha (rpl13a) and elongation factor 1 alpha (ef1a). Tcrac and igm were both established as markers for T and B-cells, respectively [37,38]. The amplified region of igm corresponds to the CH3 domain, and is shared by the secreted and membrane forms of the zebrafish immunoglobulin heavy chain protein Igm [38]. Bik codes for an early pro-apoptosis inducing protein [58,59]. Vasa is a conserved germ cell marker [60]. Gsdf and inha are specifically expressed in granulosa and Sertoli cells [20,47,61]. Similarly, cyp19a1a is expressed predominantly in granulosa cells, opposite to amh, which is expressed predominantly in Sertoli cells [47]. Specific primers for each target gene for qPCR amplification were either designed manually using Netprimer software (http:// www.premierbiosoft.com/netprimer) or taken from literature (Table 6)  [47,62,63]. When possible, primers were designed to span one intron/exon border to avoid amplification of potential contaminating genomic DNA [64]. All amplified fragments were confirmed as the target gene using Sanger sequencing. The RT-qPCR was performed on LightCycler 480 (Roche, Mannheim, Germany) using 96-well plates (Roche). SYBR green-based detections were done under the thermal cycle conditions of 95˚C for 15 min, followed by 45 cycles of 95˚C for 15 s, 63˚C for 20 s and 72˚C for 10 s. All samples were run in duplicate, together with minus reverse transcriptase, no template and a positive plate control. Five-point standard curves (dilutions 1:1-1:81) were used to calculate the efficiency of the PCR reaction. The reaction specificity was evaluated by melting curve analysis. Cycle threshold (C T ) values were determined using the LightCycler 480 software with a level of fluorescence intensity set to one. The reference genes (ß-actin, ef1a, rpl13a) were examined for data normalization using geNorm [65]. An optimal stability value of 0.486 was obtained using rpl13a and ef1a and target gene expressions were normalized to relative levels.

Gonad histology
Sections and slides were prepared as described above in 5.5. Samples were stained with haematoxylin-eosin (HE) in the Robot Slide Stainer (Microm HMS 760X, Thermo Scientific) and mounted using Pertex mounting media (Leica Biosystems). Evaluation and imaging on the samples done by use of the light microscope Olympus BX 51 (Olympus Optical Co. GmbH, Germany) and Olympus Imaging software Cell B. The identification of structures within the zebrafish gonads was done in accordance with the Histology and Histopathology Atlas of the Zebrafish (http://zfishtoxpat.comoj.com).

Whole mount immunohistochemistry
Whole mount immunohistochemistry was performed to detect Bcl2-interactingkiller (Bik), a pro-apoptotic protein, which is normally suppressed by survivalpromoting factors [58]. The procedure was performed as described by Draper (2012) [66] with the following modifications: the primary antibody used was rabbit polyclonal Immunoglobulin G (IgG) specific for anti-zebrafish Bik protein (Anaspec, Belgium). The primary antibody was diluted 1:150 and administered overnight at 4˚C. The secondary antibody used was Alexa Fluor 594 goat antirabbit IgG (Life Technologies, Eugene, OR, USA) at a 1:500 dilution for 5 hours at room temperature. Staining with 49, 6-diamidino-2-phenylindole (DAPI) was also performed to visualize somatic cells using several washes with a concentration of 2.0 mg/mL. After the completion of the IHC procedure, the gonads were imaged using an AxioZoom V16 microscope equipped with the ApoTome.2 combined with DsRed and eGFP filters. As a positive control to test the procedure, the primary antibody was replaced with an anti-eGFP monoclonal antibody (Invitrogen). Because of a shortage of samples for anti-Gdf9-b treated fish, no whole mount IHC was performed on this treatment group.

Statistical analyses
Statistical analyses were performed according to Zar (1999) [67]. The data are presented as averages ¡ standard deviations (S.D.). ANOVA was used to test the effects of treatments on fish weight, length and relative values of gene expression. Homogeneity of variances was tested using Levene's test. Fisher's least significant difference post-hoc test was used to investigate differences between groups. Pearson's product-moment correlation coefficient r was used to investigate relationships between gene expressions. Significant effects were considered at p,0.05.