Developmental tracing of luteinizing hormone β-subunit gene expression using green fluorescent protein transgenic medaka (Oryzias latipes) reveals a putative novel developmental function


  • Jon Hildahl,

    1. Norwegian School of Veterinary Science, Department of Basic Science and Aquatic Medicine, Oslo, Norway
    Search for more papers by this author
  • Guro K. Sandvik,

    1. University of Oslo, Department of Molecular Biosciences, Oslo, Norway
    Search for more papers by this author
  • Rikke Lifjeld,

    1. Norwegian School of Veterinary Science, Department of Basic Science and Aquatic Medicine, Oslo, Norway
    2. University of Oslo, Department of Molecular Biosciences, Oslo, Norway
    Search for more papers by this author
  • Kjetil Hodne,

    1. Norwegian School of Veterinary Science, Department of Basic Science and Aquatic Medicine, Oslo, Norway
    2. University of Oslo, Department of Molecular Biosciences, Oslo, Norway
    Search for more papers by this author
  • Yoshitaka Nagahama,

    1. National Institute for Basic Biology, Division of Reproductive Biology, Okazaki, Aichi, Japan
    2. Institution for Collaborative Relations, Ehime University, Matsuyama, Ehime, Japan
    Search for more papers by this author
  • Trude M. Haug,

    1. University of Oslo, Department of Molecular Biosciences, Oslo, Norway
    Search for more papers by this author
  • Kataaki Okubo,

    1. University of Tokyo, Graduate School of Agriculture and Life Science, Department of Aquatic Bioscience, Bunkyo, Tokyo, Japan
    Search for more papers by this author
  • Finn Arne Weltzien

    Corresponding author
    1. Norwegian School of Veterinary Science, Department of Basic Science and Aquatic Medicine, Oslo, Norway
    2. University of Oslo, Department of Molecular Biosciences, Oslo, Norway
    • Norwegian School of Veterinary Science, Department of Basic Science and Aquatic Medicine, Oslo, Norway
    Search for more papers by this author


Background: Luteinizing hormone (LH) and follicle stimulating hormone (FSH), produced in gonadotrope cells in the adenohypophysis are key regulators of vertebrate reproduction. The differential regulation of these hormones, however, is poorly understood and little is known about gonadotrope embryonic development. We developed a stable transgenic line of medaka with the LH beta subunit gene (lhb) promotor driving green fluorescent protein (gfp) expression to characterize development of LH-producing gonadotropes in whole larvae and histological sections. Additionally, developmental and tissue-specific gene expression was examined. Results: The lhb gene is maternally expressed during early embryogenesis. Transcript levels increase by stage 21 (36 hours post fertilization [hpf]) and then decrease during continued larval development. Examination of the expression of pituitary marker genes show that LH-producing cells are initially localized outside the primordial pituitary, and they were localized to the developing gut tube by 32 hpf. At hatching, lhb-GFP is clearly detected in the gut epithelium and in the anterior digestive tract. lhb-GFP expression later consolidate in the developing pituitary by 2 weeks postfertilization. Conclusions: During embryonic development, lhb is primarily expressed outside the central nervous system and pituitary. The novel expression of lhb in the embryonic gut suggests that LH has a hitherto unidentified developmental function. Developmental Dynamics 241:1665–1677, 2012. © 2012 Wiley Periodicals, Inc.


Pituitary gonadotropins are key players in the endocrine control of vertebrate puberty and reproduction as part of the brain—pituitary—gonad axis (b-p-g axis). In fish, as opposed to mammals, luteinizing hormone (LH) and follicle-stimulating hormone (FSH), are synthesized and secreted from distinct cells in the pituitary. The synthesis and secretion of LH and FSH is largely controlled by gonadotropin releasing hormone (GnRH), gonadal feedback loops and a variety of stimulating and inhibiting factors, such as kisspeptin, neuropeptide y, dopamine, and gonadotropin inhibiting hormone (Levavi-Sivan et al., 2010; Clarke, 2011). Upon release, LH and FSH bind to their specific receptors, LHR and FSHR, in the gonads, stimulating steroidogenesis and gametogenesis. Gonadotropin receptors, however, are expressed in multiple tissues also outside the b-p-g axis (Kumar et al., 2001; Vischer and Bogerd, 2003; Wong and Van Eenennaam, 2004; So et al., 2005; Rocha et al., 2007; Mittelholzer et al., 2009), suggesting LH and FSH could be involved in many physiological processes, although nonreproductive functions are largely unknown.

LH and FSH are part of a larger family of cysteine knot-forming polypeptide glycoproteins which form noncovalently linked heterodimers between a common α, glycoprotein alpha (GPA), and a hormone-specific β subunit. Other members of the family include thyroid stimulating hormone (TSH) and the placenta-specific chorionic gonadotropin (CG). In addition, two recently discovered glycoproteins have been identified in invertebrates and vertebrates, ranging from nematodes to humans, termed glycoprotein beta 5 (GPB5) and glycoprotein alpha 2 (GPA2) (Hsu et al., 2002). As these proteins are found in invertebrates, it has been suggested that they represent an ancestral glycoprotein (Roch et al., 2011). In addition to a putative thyrotropic effect in mammals, GPB5 and GPA2 have been suggested to function as an insect anti-diuretic hormone (Sellami et al., 2011) and to act as a neural signaling molecule controlling intestinal function in nematodes (Oishi et al., 2009).

In vertebrates, the pituitary is a key endocrine organ generally producing and secreting six to eight hormones from specific cell types, driving a wide range of essential physiological processes, including growth, metabolism, and reproduction. The pituitary is composed of anterior and posterior lobes called the adenohypophysis and neurohypophysis, respectively, which arise from different embryological origins. In tetrapods the anterior pituitary, where LH and FSH are produced, is derived from an invagination of the oral ectoderm called Rathke's pouch. In fish, however, the adenohypophysis arises by anterior–posterior patterning of preplacodal ectoderm, which maintains a linear subepithelial organization (Pogoda and Hammerschmidt, 2009). The neurohypophysis, on the other hand, is derived from a ventral outgrowth of the brain originating from neuroectoderm.

Specific pituitary cell types are determined by differential expression of a suite of transcription factors during early embryonic development (Pogoda and Hammerschmidt, 2009). Cell specification of many fish adenohypophysis cell types has been clearly defined in zebrafish by forward mutagenesis screens and gene knockdown experiments (Herzog et al., 2004; Nica et al., 2004, 2006; Pogoda et al., 2006). Information concerning the early development of pituitary gonadotropes producing LH and FSH, however, is limited. Gonadotropes and thyrotropes are present in the developing adenohypophysis anlage of zebrafish by 32 hours post fertilization (hpf) as determined by gpa gene expression (Nica et al., 2006); however, fshb is first detected later at 4 days postfertilization (dpf) and lhb is not detected until around 25 dpf by in situ hybridization (ISH; Chen and Ge, 2012). The ontogeny of pituitary cell types has also been followed in rainbow trout (Saga et al., 1993), American shad (Laiz-Carrion et al., 2003), and Ayu (Saga et al., 1999) by immunohistochemistry. These studies found that gonadotropic hormones arise after the other pituitary cell types, first occurring at hatching or posthatch. lhb transcripts, however, have been detected much early by 72 hpf in zebrafish larvae using the more sensitive polymerase chain reaction (PCR) method (Nica et al., 2006), so the ontogeny of fshb and lhb during fish larval development remains to be fully elucidated.

The Japanese medaka (Oryzias latipes) is a powerful model for developmental and functional genomic research due to its rapid development in transparent eggs, short generation time, sequenced genome, and availability of advanced forward and reverse genetic techniques (Kinoshita et al., 2009). As a teleost fish, the medaka has the additional advantage of possessing distinct gonadotropes producing LH and FSH. Thus, this provides a good model system for characterizing gonadotropin regulation. Gonadotrope regulation is still poorly understood and a good characterization of these cells during embryonic development is lacking. We thus chose to use the medaka to address our aim to characterize the expression and function of lhb during embryonic development. To accomplish this, we generated a transgenic line of medaka, tg(lhb:GFP), stably expressing green fluorescent protein (GFP) under the control of the lhb promotor, providing a sensitive method to trace developmental regulation of lhb.


Co-expression of lhb and GFP

Two transgenic founder fish were identified by crossing the 170 surviving F0 generation fish injected with the transgenic construct (Fig. 1), which had identical temporal and spatial expression patterns in their F1 progeny. The homozygote offspring of one founder fish were identified in the F3 generation and used to make the stable transgenic line. Expression of GFP in cells where lhb was transcribed was confirmed by determining the co-localization of GFP protein and lhb transcripts in specific adult pituitary cells by ISH (Fig. 2). A layer of GFP-expressing cells were detected on the lateral surface of the proximal pars distalis of the pituitary. These cells were localized more anteriorly and dorsally on the distal surface (Fig. 2A) than toward the midline of the pituitary where they extended posteriorly along the ventral surface of the pars intermidia (Fig. 2B,D,F). Cells expressing fshb were localized dorsal to the lhb-producing cells and did not colocalize with GFP-positive cells (Fig. 2E,F).

Figure 1.

Schematic of lhb-hrGFPII transgenic construct. A bacterial artificial chromosome (BAC) was identified (golwb108_H20 BAC) containing the medaka lhb gene and 25-kb 5′ sequence and 78-kb 3′prime sequence. The transgenic construct was generated by inserting a hrGFP-II gene, with a strong Kozak sequence and bovine growth hormone poly-A signal (BGHpA), 3-bp upstream of the endogenous lhb translation start site at the start of exon 2. GFP, green fluorescent protein.

Figure 2.

A–F: In situ hybridization detection of lhb mRNA in adult medaka pituitaries. A,B,D,F: Green fluorescent protein (GFP) was detected in cells along the lateral surface of the pituitary seen in sections along the mediolateral axis, lateral surface of the pituitary (A) extending medially (B) to midline sections (D,F). C,D: Cells expressing the endogenous lhb gene were detected by in situ hybridization (C) and shown to colocalize with GFP-expressing cells along the distal margin of the proximal pars distalis (D) in adult medaka pituitaries. E,F: Note that fshb-producing gonadotropes are not colocalized with GFP. Pituitaries are orientated with the anterior surface to the left. Scale bars represent 50 μm.

No lhb transcripts were detected by ISH in transgenic larvae. Transcript levels of lhb could be detected, however, in pooled whole larvae by the more sensitive quantitative PCR (qPCR) method (Fig. 3A). We, therefore, dissected GFP-positive cells from larvae to see if endogenous lhb and gfp gene expression could be detected in the same cells by PCR. Dissected GFP-positive cells from 2–4 dpf larvae showed co-expression of endogenous lhb and the gfp transgene by qPCR (Table 1). The expression of related genes was analyzed in the dissected cDNA to determine their co-expression with lhb (Table 1). LH receptor (lhr) was expressed in all of the dissected samples. Similarly, a marker of gut endoderm, forkhead box protein A2 (foxa2), was expressed in all nine samples analyzed. The genes for the common alpha subunit, gpa, was expressed in three of nine dissected samples. Additionally, genes encoding enzymes involved in LH dependent steroidogenic pathways—steroidogenic acute regulatory protein 1 (star1), cytochrome P450 cholesterol side-chain cleavage enzyme (cyp11a1), cytochrome P450 17α-hydroxylase (cyp17a1), and 3β-hydroxysteroid dehydrogenase (hsd)—were all expressed in some but not all of the samples. The star2 gene was not expressed in any of the samples.

Figure 3.

lhb quantitative polymerase chain reaction (qPCR) analysis. A,B: Relative lhb (A) and lhr (B) gene expression in pooled medaka through embryonic development to posthatch larvae. Ligand and receptor gene expression was normalized to 16s gene expression using an efficiency adjusted relative quantification method. Data are presented as mean relative expression + SEM, n = 4. Data were analyzed by one-way analysis of variance and differences were considered significant at P < 0.05. The number of individuals per pool is given in Table 2.

Table 1. Gene Expression in Dissected GFP-Positive Cellsa
  1. + indicates that the gene was detected and − indicates that the gene was not detected. GFP, green fluorescent protein.


Developmental Expression

The lhb gene was expressed early during embryogenesis, initially as maternal expression and transcripts continued to be detected throughout larval development (Fig. 3A). lhb was clearly detected at all stages investigated. Transcript levels increased up to the six-somite stage (stage 21, 36 hpf) just before the onset of gut tube formation and then decreased again, reaching significantly lower levels by 120 hpf and remained low in hatched larvae. The lhr transcripts were also detected throughout larval development (Fig. 3B), showing a significant increase at 14 dpf. gfp transcripts were detected together with lhb in a subset (n = 2) of pooled lhb-GFP larvae between 1 hpf and 36 hpf (data not shown), showing a similar increase in expression at 36 hpf.

The first GFP protein was detected at the two-somite stage (32 hpf, approximately developmental stage 19). GFP expression started posterior of the eyes as paired lateral clusters at the anterior margin of the otic vesicles. These cells then increased in number, extending to the midline and posteriorly as a continuous group of cells (Fig. 4). By the 16-somite stage (48 hpf), GFP-producing cells extended posteriorly from the metencephalon–myelencephalon boundary of the developing hindbrain to behind the otic vesicles (Fig. 5A,B) as a band of cells along the ventral surface of the larvae (Fig. 5D,E). The majority of the cell formed a band structure along the midline, with a few dispersed cells more anteriorly and some more laterally in the same plane, ventral to the brain (Fig. 5E). As the larva grew, the main band of GFP cells elongated only slightly relative to the growing larva, maintaining its alignment with the otic vesicles. By 96 hpf, there were fewer dispersed GFP-producing cells which extended to the mesencephalon–metencephalon boundary, and a few distinct cells were located at the posterior margin of the eye (Fig. 5G,H).

Figure 4.

A,B: First expression of lhb mRNA measured by in vivo green fluorescent protein (GFP) detection. The initial development of lhb-GFP expression was followed in vivo from 30 to 39 hours post fertilization (hpf) in synchronized groups of lhb-GFP transgenic larvae. Larvae are situated with the head to the right in the frontal plane. B–H: GFP could first be detected by 32–33 hpf (B), and GFP-expressing cells then expanded to the midline and posteriorly. Scale bars = 200 μm.

Figure 5.

A–H: Developmental expression of lhb-green fluorescent protein (GFP) in transgenic medaka. lhb-GFP detection from 48 hours post fertilization (hpf; see Fig. 5) to 96 hpf using confocal microscopy (A,D) on fixed material and using in vivo fluorescent imaging (B,E,G–H). A,D: Confocal analysis allowed for high-resolution imaging, clearly showing GFP signal at the midline between the two otic vesicles (ot) at 48 hpf. The basic GFP expression pattern is largely established by 48 hpf, with GFP distribution expanding only slightly as development progresses. Light microscopic images (C,F) show the orientation and anatomy of 48 hpf larvae. Larvae are orientated with the head to the right. Arrows identify dispersed cells anterior of the main central strip of GFP-positive cells. Scale bars = 70 μm for confocal images and 200 μm for in vivo images.

Extrapituitary Localization of lhb-gfp

ISH of pituitary marker genes showed that GFP-producing cells were initially localized outside the primordial pituitary. Both pit1 and lhx3 were expressed in the midline ventral to the eye at 48 hpf and 72 hpf (Fig. 6A,B). However, GFP was detected posterior to the signal for the pituitary marker genes (Fig. 6D,E). The otic vesicles are located at the ventral margin of the brain (Fig. 6C), and GFP was detected ventral to the otic vesicles on the basal surface of the larvae (Fig. 6F). During embryonic development, thus, GFP was detected outside the central nervous system and pituitary.

Figure 6.

Extrapituitary green fluorescent protein (GFP) detection. A,B: In situ hybridization with two pituitary marker genes, pit1 (A) and lhx3 (B), show the developing pituitary at the midline of the head and ventral to the eye (main expression identified with arrows). D,E: GFP expression starts around the anterior margin of the otic vesicle along the surface of the larva. D,E: The base of the larval brain is at the midpoint of the otic vesicle (C), and GFP is clearly detected ventral to this boundary (E). GFP is not detected in the pituitary before 14 days post fertilization (dpf). Larvae are situated with the head to the left. Identified tissues are b, brain (the base of the brain is delineated by a hatched line); ot, otic vesicle (identified by arrow and hatched oval) and y, yolk. Scale bars = 100 μm for fixed embryos (A,B,D,E) and 200 μm for in vivo images (C,F).

Optical sections and three-dimensional rendering generated by confocal microscopic analysis in addition to histological fluorescent imaging showed that GFP cells form a tubular structure (Fig. 7A). Confocal images are taken as optical sections, a z-stack, through a section of tissue, such that individual planes of the image can be analyzed separately. GFP-positive cells changed shape moving through the tissue, with a round form more dorsal and ventral (Fig. 7B) and an elongated columnar form medially (Fig. 7C). Three dimensional rendering of the entire confocal z-stack (Fig. 7A) indicated that these cells formed a tubular structure. Taken together these data indicate that the GFP-positive cells are columnar shaped and organized in a tubular arrangement. The anatomical organization of GFP-positive cells was supported by fluorescent and histological imaging of sections. GFP-positive cells were detected as a tubular structure in the abdomen of 5 dpf larvae (Fig. 7D–F), which was identified as the developing gut by histological staining (Fig. 7F), in addition to the expression of the foxa2 gene in this tissue. GFP was expressed intermittently throughout the radial axis of the developing gut tube, which formed a singular radial layer of cells. By 5 dpf the lumen was clearly seen (Fig. 7F). After hatching, the intestinal epithelium thickened and became more convoluted (Fig. 8D–F). Some residual yolk was present in the yolk-sac for 3–4 days (8–11 dpf). GFP expression was more limited posthatch (Fig. 8A,B,E) and was detected as isolated cells in the intestine wall up to 11 dpf (Fig. 8E), when the yolk was almost completely absorbed. Individual cells were also identified just posterior to the eye (Fig. 8B), which was localized to the dorsal and posterior surface of the first gill arch in histological sections (Fig. 8H,I) up to 11 dpf. Because there were only a few isolated GFP-positive cells posthatch, we were not able to measure the expression of endogenous lhb expression in the intestine at this time; however, the persistence of isolated GFP-positive cells provides an indirect measure of lhb expression specifically in the gastrointestinal tract.

Figure 7.

Embryonic tissue-specific distribution. B–F: Tissue-specific green fluorescent protein (GFP) expression was analyzed by confocal microscopy (B,C) and fluorescent microscopy (D,E) and hematoxylin/eosin staining (F) of histological sections. A: Three-dimensional rendering of confocal image analysis shows these cells are arranged in a tubular-like structure. B,C: Optical sections obtained by confocal microscopy display changing cell morphology in distal (B) and medial (C) planes. D–F: In transverse sections of 5 days post fertilization (dpf) larvae, GFP is expressed in the gut tube in the ventral margin of the larva. Arrows identify GFP-expressing cells. Tissues are labeled as brain (b), gut tube (gt), notochord (nt) and otic vesicle (ot). Scale bars = 50 μm in D, 25 μm in B,C,E,F.

Figure 8.

Lhb-GFP expression posthatch. A,B: At hatching, green fluorescent protein (GFP) expression is greatly reduced relative to the whole larva with isolated clusters at the dorsal–anterior margin of the yolk-sac (A) and posterior margin of the eyes (B). C,E: Sagittal histological sections show that GFP is first detected in the pituitary by 14 days post fertilization (dpf) (C) and is expressed in the intestine at hatch (E). D,F,G:: Hematoxylin/eosin staining shows anatomical structures of histological sections. Arrows identify individual cells or isolated cell clusters near the eyes (B,H,I), in the gut (A,E), and in the pituitary (C). Tissues are labeled as brain (b), gill (g), intestine (i), and yolk (y). Scales bars = 200 μm in A,B, 100 μm in G, 50 μm in D, 25 μm in C,E,F,H,I.

Transition to Pituitary Expression

Pituitary GFP expression was first detected in fry at 14 dpf in a few isolated cells. The yolk was fully absorbed at this time, with no GFP detected in extrapituitary tissue.


We have established a transgenic line of medaka expressing GFP under the control of the lhb promotor. Colocalization of lhb transcripts and GFP in pituitaries from sexually mature fish and in dissected cells from larvae confirms that GFP expression specifically reflects the regulation of endogenous lhb gene and is not ectopic. We have thus established a sensitive method for tracing the development of lhb-producing cells using in vivo and in situ fluorescent imaging of GFP. By this means, we reveal novel regulation of the lhb gene during early larval development, suggesting a possible developmental function for LH.

The data presented here indicate that lhb is not expressed in the pituitary of medaka until approximately 14 dpf, after first feeding and the complete absorption of the yolk. gnrh1 and gnrh3 are expressed by 2 dpf in medaka, and developmental tracing in transgenic medaka reveals that Gnrh neurons extend ventrally into the pituitary by between 10 and 20 dpf (Okubo et al., 2006), around the time that lhb-GFP is detected in the pituitary. This suggests that lhb is expressed relatively late in the pituitary development, when the gross morphology of the brain and pituitary have been established and hypothalamic neurons innervate the pituitary. Thus, LH gonadotropes could require hypothalamic inputs for final gonadotrope activation similar to in sheep (Brooks et al., 1992; Szarek et al., 2008) and rat (Aubert et al., 1985) gonadotrope maturation. This is further supported by the observation that gonadotropin expression starts later than other anterior pituitary hormones in multiple fish and mammal species (Asa et al., 1988; Saga et al., 1993, 1999; Japon et al., 1994; Laiz-Carrion et al., 2003). However, zebrafish, one of the species with the best characterized adenohypophysis development, provides partially conflicting results. Gnrh3 neurons reach the pituitary earlier in zebrafish, by 5 dpf, than Gnrh1 neurons in medaka. Of interest, this corresponds to the onset of fshb pituitary expression but not lhb pituitary expression in zebrafish (Chen and Ge, 2012). This suggests that the b-p-g axis is established earlier in zebrafish, although details of early gonadotrope development remain to be clarified in this species. Zebrafish pituitary gonadotropes are determined to differentiate early in larval development by 32 hpf (Pogoda and Hammerschmidt, 2009), but it is unknown at what point these primordial gonadotropes express lhb or fshb. In zebrafish transiently expressing GFP under the control of the 5′-flanking region of lhb, GFP signal was detected in the area of the pituitary by 48 hpf (Chen and Chiou, 2010), although this signal was not confirmed by colocalization with the endogenous lhb gene or generation of a stable transgenic line. Similarly, down-regulation of lhb in eya1 mutant zebrafish suggests that lhb is being produced in gonadotropes by 72 hpf, although cell-specific expression could not be confirmed by in situ hybridization (Nica et al., 2006). This is likely due to the lower sensitivity of in situ hybridization relative to PCR. Likewise, in the current study, lhb was detected by qPCR in whole larvae but could not be detected by in situ hybridization. In zebrafish, the presence of gpa in the absence of tshb is an indication of early gonadotropes, but final maturation when synthesis and production of functional LH occurs may come later. For example, in mammals gpa expression precedes lhb and fshb expression in mouse (Japon et al., 1994) and human gonadotropes (Pope et al., 2006), which may explain the inability to detect lhb transcripts in the pituitary of zebrafish when gpa could be detected. In addition, the presence of two lhb genes in zebrafish (So et al., 2005) could lead to partially divergent function or tissue distribution, possibly explaining the proposed earlier onset of lhb gene expression relative to medaka.

It has been hypothesized that the lack of conserved synteny for the lhb gene between teleost fish and tetrapods could explain the divergent expression of LH and FSH in distinct gonadotropes in contrast to the situation in tetrapods (Kanda et al., 2011). The difference in the genomic environment of lhb between teleosts and tetrapods could possibly further explain the divergent extrapituitary developmental expression of lhb in the gut of medaka. It is interesting to note that the recently discovered and closely related glycoproteins, Gpa2 and Gpb5, are expressed in both vertebrates and invertebrates (Dos Santos et al., 2009). The presence of these proteins in invertebrates suggests that they could be an ancestral form of the glycoprotein hormones. In mammals, these have been found to dimerize and activate the thyroid stimulating hormone receptor, leading to the heterodimer being called thyrostimulin (TS; Nakabayashi et al., 2002). Their endogenous physiological function, however, remains to be elucidated. These hormones are attributed various functions in the gut in D. melanogaster (Sellami et al., 2011) and C. elegans (Oishi et al., 2009), including antibacterial defense, intestinal color change, defecation cycling, and possible water reabsorption. A putative function for LH in the gut during development could thus represent an evolutionarily redundant glycoprotein function, whereas in higher vertebrates Gpa2 and Gpb5 could possibly have assumed a thyrotropic function.

Few studies have looked at the expression of lhb outside the pituitary and none that we are aware of during early development. In medaka, the pituitary is located at the ventral margin of the eyes at 48 hpf as indicated by lhx3 and pit1 expression (Fig. 6A,C), similar to zebrafish (Nica et al., 2006; Pogoda et al., 2006); however, GFP is clearly not colocalized with the pituitary marker genes at these embryonic stages. Although lhb-GFP could not be detected in the pituitary of medaka until after hatching, its expression in the developing gut tube, another tissue of endoderm origin, suggests a novel function for LH in the gut during development. Interestingly, lhb transcript content significantly increases and GFP is first detected just before initiation of gut tube formation at stage 22. The developing gut tube then extends caudally, reaching the cloaca by stage 26 (54–58 hpf) (Kobayashi et al., 2006). This process is initiated by migration of endodermal cells to the midline. GFP-producing cells are also initially expressed distally and migrate to the midline and posteriorly at which time they also express foxa2. When the gut tube is fully developed, GFP is clearly expressed in the rostral portion as a cylindrical tube-like structure. The significant decrease in lhb expression at 120 hpf when gut tube formation is complete further supports the notion that LH could be important for early development of the gastrointestinal tract.

The fact that lhb-GFP expression ceases in the gut soon after hatching and complete absorption of the yolk suggests it could be important for survival in the chorionated microenvironment. For example, LH could drive steroidogenesis in the gut during development. The yolk is the main source of cholesterol for steroidogenesis in yolk-sac larvae, and conversion of cholesterol to 3β-hydroxypregn-5-en-20one (pregnenolone) has been shown to be critical for embryogenesis in zebrafish by stabilizing microtubules for embryonic cell movement (Hsu et al., 2006). A steroidogenic role of lhb-GFP-positive cells is supported by our findings of steroidogenic enzyme gene expression in these cells. The enzyme responsible for the conversion of cholesterol to pregnenolone, Cyp11a1, is expressed in the yolk syncytial layer (ysl) of zebrafish, thus identifying an important steroidogenic tissue during development. LH drives steroidogenesis in the gonad by stimulating the uptake of cholesterol into theca or Leydig cells and into the mitochondrion by stimulating steroidogenic acute regulatory protein (StAR). Subsequently, LH stimulates the production of androgens in these cells by also stimulating cyp17a1 gene expression. Indeed, all three of these genes were expressed in lhb-GFP-positive tissue. Additionally, the expression of gpa and lhr in lhb-GFP tissue suggests that LH acts as a dimer in an autocrine or paracrine manner in this tissue. It is interesting to note that gpa was not detected in all of the dissected GFP-positive cell samples. This suggests that lhb and gpa have different temporal regulation in these cells, but the colocalization of these transcripts supports the notion that LH forms a functional heterodimer in the developing gut. The gut tube during early development is immediately adjacent to the ysl (Kobayashi et al., 2006), which is responsible for initial yolk absorption. The close association of the ysl and the developing gut tube could provide a conduit for cholesterol and steroid exchange. Indeed, the intestine of the green frog (Rana esculenta) incubated with cholesterol stimulated pregnenolone synthesis (Belvedere et al., 2001) and LH receptors are found in the intestine of human fetuses (Abdallah et al., 2004) similar to our present data, suggesting that the intestine is sensitive to LH during development and act as a steroidogenic tissue. LH could, thus, stimulate steroidogenesis in the larval gut of medaka by stimulating cholesterol uptake and cyp17a1 expression, which could be involved in multiple steroidogenic processes.

In summary, we have established a powerful model for characterizing the developmental and reproductive physiological regulation of LH gonadotropes. This has allowed us to localize lhb regulation in developing gut tissue, suggesting a novel function for LH during larval development. This was confirmed by the presence of endogenous lhb transcripts in the developing gut tissue. Furthermore, the transition from extrapituitary to pituitary expression of GFP provides insight into the ontogeny of pituitary gonadotropes. These data lay the framework for future research into the function of LH during early pituitary development. This transgenic line also provides the technological basis for further characterization of pituitary gonadotropes by cell-specific gene expression analysis and electrophysiological experiments.



Japanese medaka of the dr-R strain were used for all experiments. Wild-type and transgenic fish were housed in re-circulating systems with water temperature between 25–28°C and light–dark cycle of L14:D10. When collecting eggs from female fish, they were rolled on paper to remove the connecting fibers and placed in embryo culture (E3) medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl2, 0.33 mM MgSO4) and either kept on ice for staging or placed in an incubator at 26°C until hatching at which point they were transferred to the system tanks. Fish were fed a combination of dry feed and live brine shrimp (Artemia salina). Handling and use of fish was in accordance with the guidelines of the Committee on Life Sciences of the University of Tokyo and approved regulations for the care and welfare of research animals at the University of Oslo.

Generation of Stable Transgenic Line

The lhb-gfp transgenic line, tg(lhb:GFP), was generated using bacterial artificial chromosome (BAC) homologous recombination technology (Copeland et al., 2001; Lee et al., 2001; Nakamura et al., 2008). The BAC clone used in the present study was golwb_108_H20 (insert size 105 kb), containing 103-kb flanking sequence to the lhb gene (Fig. 1). The BAC clone was identified in the medaka genome browser website (http://medaka. and kindly provided by Dr. Kiyoshi Naruse (NIBB, Okazaki, Japan). A hrGFPII-Km cassette containing a Kozak sequence, hrGFPII (Stratagene, Santa Clara, CA), bovine growth hormone poly A signal and kanamycin-resistant gene was constructed. This cassette was used as template in a PCR with primers designed to partially anneal to the hrGFPII-Km cassette, flanked with lhb-specific sequence “arms.” The lhb flanking sequences were designed to anneal to sequences immediately upstream and downstream of the translation initiation site (F: 5′-CCCTCAATGTTTAAAATATCCAGAAATATGTTTTGAGTCAATTTTATTTTTCACAGGATCCACCATGGTGAGCAAGC-3′, R: 5′-CAAACATTTCTTTTACCTGCAGGGGCCAGGGACCAGAGGAAAGTTGAGGTTCCTAGTGGACCAGTTGGTGATTTTGAACTT-3′), thereby replacing the endogenous lhb start site and first 48-bp coding sequence in exon 2 of the lhb gene with the hrGFPII-Km cassette, when homologous recombination with the BAC clone occurred (Fig. 1). The homologous recombination was performed in DY380 E. coli, and the resulting BAC with hrGFPII-Km cassette inserted was extracted with QIAGEN QIAfilter Midi Kit (Qiagen, Hilden, Germany) and carefully eluted in TE buffer (pH = 8). The optimal concentration of injected BAC-DNA was determined experimentally, and 1 ng/μl was injected into one-cell stage medaka embryos with a manual microinjector (Narishige, London, UK). After growing to adults, these fish were crossed with each other and their eggs were checked for fluorescence on 2–6 dpf to identify founders (individual with the BAC-DNA inserted into the germ cells). The F1 progeny of one of the F0 founder fish were crossed with each other. GFP-positive F2 generation fish were then further crossed with wild-type fish, and transgenic fish that produced 90–100% GFP-positive offspring were determined to be homozygous for tg(lhb:GFP). The identified homozygote F2 fish were crossed with each other to generate a stable homozygous line.

Larval Sampling

For staging, eggs were collected just after the lights turned on and spawning commences. The eggs were synchronized at the one-cell stage as follows. Upon harvesting, the eggs were placed on ice to cease development. Synchronized eggs were put in an incubator at 26°C. Subsequently, when the larvae were at the desired stage, they were transferred from E3 medium to RNAlater (Ambion, TX). The samples were stored at −20°C before RNA was extracted. Eggs were collected from 12 different stages with four pooled samples, according to Table 2.

Table 2. Larval Sampling and RNA Yields for qPCR Analysisa
SampleNumber larvae per poolRNA (ng/μl) per poolRNA (ng/μl) per embryo
  • a

    qPCR, quantitative polymerase chain reaction; hpf, hours post fertilization; dpf, days post fertilization.

1 hpf401373.43
18 hpf301354.50
24 hpf20994.95
30 hpf201306.50
36 hpf201356.75
48 hpf2027913.95
72 hpf1534022.67
96 hpf1546731.13
120 hpf1044144.10
8 dpf212964.50
11 dpf212562.50
14 dpf210753.50

lhb-gfp Transgenic Fluorescent Imaging

According to the description above, eggs for confocal microscopic analysis were collected right after spawning and synchronized at the one cell stage. When they reached the desired stage, they were transferred from E3 medium to phosphate buffered saline (PBS) to remove the medium completely. After washing, the eggs were fixed in 4% paraformaldehyde (PFA) at 4°C for 6–12 hr. When fixed, the larvae were de-chorionated and placed in a droplet of 1% agarose to stabilize the larvae for imaging. A FluoView 1000 upright BX61WI confocal laser scanning microscope (Olympus, Center Valley, PA) was used for high resolution confocal analysis. Imaris scientific 3D/4D image processing and analysis software (Bitplane Scientific Software, Zurich, Switzerland) was used for analyzing the confocal images.

In vivo GFP imaging was also carried out using standard fluorescent microscopy. The larvae were synchronized and observed every 30 min from 25 hpf and onward using an Olympus IX81 inverted fluorescence microscope to see when GFP could first be detected. The eggs were placed in an agarose mold submerged in E3 medium to stabilize larvae for imaging. Larvae were subsequently studied from 24 hpf to 14 dpf to investigate the changes in developmental expression.

Due to the difficulty of viewing internal anatomy of juvenile and adult medaka, histological sections were prepared to image the cellular distribution of GFP. Fish were fixed in a mixture of 80% HistoChoice (Sigma-Aldrich, St Louis, MO), 2% PFA, 1% sucrose, 1% CaCl2, and 0.05% glutaraldehyde. Larvae were initially fixed for 5 hr, dissected out of their chorion and then fixed for an additional 4 hours. Hatched juvenile medaka were anesthetized on ice and placed directly in the fixation medium overnight at 4°C. Sexually mature medaka were decapitated and the heads were then fixed in the fixation medium overnight at 4°C. Fixed fish were then dehydrated in serial methanol washes (70%, 80%, 95%, 100%) for 20 min ×2 each and kept in 100% ethanol at −20°C for a minimum of 18 hr. Fixed and dehydrated fish were then cleared in chloroform 30 min ×4 at room temperature and infiltrated with paraffin wax for 30 min ×4. Paraffin blocks were made for the embedded medaka and kept at 4°C until sectioning. 3–6 μm sections were prepared using a horizontal microtome (Thermo Scientific, Waltham, MA), dried at 37°C, and stored at 4°C. Fluorescent imaging was carried out on an Olympus IX81 microscope as described above.

Quantitative Analysis of lhb and lhr mRNA Content

Template preparation.

Pooled embryos of synchronized eggs were collected in RNAlater according to Table 2. RNA was extracted and purified using the QIAGEN RNeasy Lipid Tissue Mini Kit with on-column DNase treatment (Qiagen), according to product specifications. Eggs were homogenized using the FastPrep -24 Tissue and Cell Homogenizer (MP Biomedicals, Solon, OH) in lysing Matrix D and Qiazol (Qiagen). Each sample was eluted in 30 μl RNase-free water.

To analyze gene expression specifically in GFP-positive cells, nine larvae (72 hpf, stage 29) were carefully dissected using fine-tip tweezers under a fluorescent dissecting microscope. RNA from the dissected GFP-positive tissue was linearly amplified using MessageBOOSTER cDNA Synthesis Kit (Epicenter Biotechnologies, Madison, WI). Dissected tissue was eluted into 3 μl of MessageBOOSTER Quick Extract solution. Oligo(dT)primer containing a T7 promoter and SuperScript III (Invitrogen) Reverse Transcriptase were used to synthesize first-strand cDNA from poly(A) RNA. Following second-strand cDNA synthesis, a high-yield in vitro transcription reaction amplified the poly(A) RNA (mRNA). Following in vitro transcription, the samples were treated with DNase I to remove the ds cDNA template, as well as a large portion of genomic DNA. The amplified RNA was purified using Qiagen RNeasy Micro Kit (Qiagen).

Developmental series cDNA was prepared using Superscript III reverse transcriptase (Invitrogen) and oligo(dt) primers from 1 μg total RNA according to product specifications. The quantity of DNase treated RNA was measured on a NanoDrop spectrophotometer (NanoDrop, Thermo Fisher scientific, Wilmington, DE). The quality of a subsample of the RNA was controlled using the Bioanalyzer (Agilent Technologies, Santa Clara, CA).

Quantitative gene expression analysis by qPCR.

qPCR was carried out on a LightCycler 480 Real-Time PCR system (Roche, Mannheim, Germany), using the LightCycler 480 Master with SYBR Green (Roche). To avoid detection of genomic DNA (gDNA), the primers or the amplicon was designed to span exon–exon boundaries based on in silico analysis of the medaka genome. A standard dilution curve was run for each primer pair to determine the pair with the best PCR reaction efficiency. Beta actin (bact), 60s ribosomal protein L7 (rpl7), and 16s ribosomal RNA (16s) were run on each sample and analyzed together and separately in the BestKeeper program (Pfaffl et al., 2004) to determine the gene with the most stable expression throughout embryonic development and with an efficiency value similar to the target genes. The best housekeeping gene, 16s, was used to normalize the expression analysis, using an efficiency-corrected relative quantification method (Weltzien et al., 2005). Each sample was analyzed in duplicate reactions consisting of 5 μl of mastermix, 2 μl of forward and reverse primer mix (5 μM each), and 3 μl of each 10 × diluted cDNA sample in a total volume of 10 μl. The cycling parameters were 10 min preincubation at 95°C, followed by 42 cycles of amplification at 95°C for 10 sec, 60°C for 5 sec, and 72°C for 4 and 5 sec (for lhb and 16s, respectively, with amplicon lengths of 100 bp and 120 bp), followed by a melting curve analysis from 65°C to 95°C. A no template control on every plate controlled for nonspecific contamination and melting curve analysis was used to verify that a single specific product was measured in each run. A subsample (n = 5) of no-RT controls were run to confirm the lack of gDNA detection.

Dissected GFP-positive cell cDNA was used to determine whether endogenous lhb and gfp were co-expressed in the same tissue. In addition, expression of related genes presented in Table 1 was measured in the dissected tissue. All genes were analyzed by qPCR, except for foxa2, which was analyzed by 35 cycles of reverse transcriptase-PCR (RT-PCR) using Platinum Taq polymerase (Invitrogen) according to product specifications. Primers are given in Table 3. The lhb assay was designed to only detect endogenous lhb by spanning the gfp insert, such that the transgenic construct would not be detected. PCR products were analyzed by melting curve analysis and run on a 1.0% agarose gel to determine the product size.

Table 3. Primers Used for RT- and qPCRa
NameTypeSequence 5′-3′Amplicon length
  • a

    RT-, reverse transcriptase; qPCR, quantitative polymerase chain reaction; hpf, hours post fertilization.


In Situ Hybridization

Digoxigenin (DIG) -labeled riboprobes were generated by in vitro transcription (Roche) of cloned sense and antisense sequences to detect the tissue distribution of lhb and fshb, as well as the pituitary marker genes pit1 and lhx3 by whole-mount and tissue section ISH.

For whole mount ISH, 48 hpf and 72 hpf larvae were fixed and dissected as described above using 4% PFA. ISH was carried out according to established protocol. Anti-DIG staining was observed using an Olympus IX81 system as described above. Brightfield imaging of the anti-DIG staining was compared with GFP fluorescence detection.

For ISH analysis of mature medaka pituitaries, sections were prepared as described above for imaging. ISH was performed according to standard section protocol. Bright field images of stained sections were compared with fluorescent detection from the same slide to determine colocalization of DIG-labeled probe and GFP when possible.

Statistical Analysis

qPCR data were first tested for homogeneous variance and then were analyzed by one-way analysis of variance, followed by a Tukey-Kramer HSD post-hoc analysis to determine which groups were significantly different (P < 0.05).


We thank Dr. Kiyoshi Naruse of the National Institute for Basic Biology (Okazaki, Japan) for providing us with the lhb containing BAC, and Drs. Lino Tessarollo and Donald L. Court (National Cancer Institute, USA) for the DY380 bacteria. We also thank Dr. Shin-ichi Higashijima for the technical advice on generating the transgenic construct. F.-A.W. and T.M.H. were supported by the Research Council of Norway.