Dwarfism and Increased Adiposity in the gh1 Mutant Zebrafish vizzini

Abstract

Somatic growth and adipogenesis are closely associated with the development of obesity in humans. In this study, we identify a zebrafish mutant, vizzini, that exhibits both a severe defect in somatic growth and increased accumulation of adipose tissue. Positional cloning of vizzini revealed a premature stop codon in gh1. Although the effects of GH are largely through igfs in mammals, we found no decrease in the expression of igf transcripts in gh1 mutants during larval development. As development progressed, however, we found overall growth to be progressively retarded and the attainment of specific developmental stages to occur at abnormally small body sizes relative to wild type. Moreover, both subcutaneous (sc) and visceral adipose tissues underwent precocious development in vizzini mutants, and at maturity, the sizes of different fat deposits were greatly expanded relative to wild type. In vivo confocal imaging of sc adipose tissue (SAT) expansion revealed that vizzini mutants exhibit extreme enlargement of adipocyte lipid droplets without a corresponding increase in lipid droplet number. These findings suggest that GH1 signaling restricts SAT hypertrophy in zebrafish. Finally, nutrient deprivation of vizzini mutants revealed that SAT mobilization was greatly diminished during caloric restriction, further implicating GH1 signaling in adipose tissue homeostasis. Overall, the zebrafish gh1 mutant, vizzini, exhibits decreased somatic growth, increased adipose tissue accumulation, and disrupted adipose plasticity after nutrient deprivation and represents a novel model to investigate the in vivo dynamics of vertebrate obesity.

Energy metabolism and somatic growth are closely coordinated in humans and other vertebrates. For example, childhood obesity is associated with increased linear growth, whereas undernutrition is associated with growth retardation (1). GH is an endocrine factor secreted from the anterior pituitary gland that serves to synchronize growth and metabolism. GH-deficient children exhibit short stature and are mildly obese (2), and supplementation with GH reduces adipose mass in these children (2, 3). In rodents, hypophysectomy (4) or inactivation of a GH transgene (5, 6) leads to obesity independent of food intake. Thus, GH deficiency in mammals is associated with stunted somatic growth and increased adipose tissue (3).

All known effects of GH are mediated through the transmembrane GH receptor (GHR) (7). In mammals, signaling downstream of GHR is mediated by the GHR-associated tyrosine kinase Jak2 (8, 9) and transcriptional activation of IGF by Stat5B (10, 11). Indeed, mutations in GHR, JAK2, STAT5B, and IGF lead to severe disorders in humans, characterized by GH insensitivity, profound growth failure, short stature, obesity, and increases in specific adipose tissues (review in Ref. 11). GH likely arose early during vertebrate evolution (12, 13), but most of our understanding about GH function derives from studies of humans and other mammals. The activities of GH and IGF in teleost fish appear to be largely conserved with mammals (14, 15). However, the genetic requirements for GH on development, growth, and fat storage and mobilization in this group have not been defined. A genome duplication event early in teleost evolution resulted in copies of the igf genes, and the zebrafish genome contains igf1 (also known as igf1a), igf3 (also known as igf1b), and igf2a and igf2b (16). Despite the genome duplication, zebrafish possess only a single gene for GH: gh1 (17).

In fish as in mammals, white adipose tissue expands via 2 mechanisms: adipocyte hyperplasia (addition of new small fat cells) and adipocyte hypertrophy (growth of existing fat cells) (18). The number and size of adipocytes have important implications for health, because excessive hypertrophy is associated with increased susceptibility to metabolic disorders in humans (19). GH-deficient humans show a reduced number of adipocytes, and this symptom is corrected by GH treatment (20–22). This relationship suggests that GH stimulates adipocyte hyperplasia. However, in vitro, the role of GH during adipocyte proliferation is unclear: although GH stimulates adipocyte differentiation of 3T3-F442A cells (23), the hormone inhibits adipocyte differentiation in primary preadipocyte cultures (24, 25). The differential effects of GH on adipose cellularity are linked to the nutritional state of the organism. GH is the major anabolic hormone during nutrient deprivation, increasing lipolysis and stimulating mobilization of stored lipids from adipose tissue (26). Consistent with this role, mammalian tissues deficient for GH show both hypertrophic morphology and reduced rates of lipolysis (20). However, in the fed state, GH exerts acute insulin-like effects by inhibiting noradrenaline-stimulated lipolysis (27). Thus, GH exerts complex and antagonistic effects on adipose tissue growth and morphology relative to nutritional state.

The zebrafish is emerging as an important model for studying developmental endocrinology and adipose tissue metabolism (28–33). As in mammals, zebrafish fat is stored as lipid droplets (LDs) in adipocytes located within multiple, regionally distinct, adipose tissues situated in both visceral and subcutaneous (sc) locations (28, 33). The appearance of these different adipose tissues, or “depots,” is tightly correlated with somatic growth and the attainment of well-characterized postembryonic developmental stages (33, 34). Zebrafish adipocytes share extensive molecular and morphological conservation with mammalian white adipocytes (28, 33), and lipids stored in zebrafish adipocytes are mobilized in response to a negative energy balance (28). Adipocyte LDs can be visualized in live zebrafish using a range of fluorescent lipid dyes (28, 35). Because the defining characteristics of adipocyte hyperplasia and hypertrophy are LD accumulation and growth, longitudinal investigation of LD dynamics provides an account of hyperplastic and hypertrophic adipose growth. Thus, zebrafish provide a tractable system to investigate GH regulation of both overall somatic development and adipose tissue growth.

In this study, we identify the dwarf zebrafish mutant vizzini and map it to a premature stop codon in gh1. We show that in the absence of GH1, postembryonic somatic growth, development, and adipogenesis become decoupled, resulting in severely retarded somatic growth, increased adiposity, and an inability to appropriately mobilize stored fats during starvation. Our results provide novel insights into the requirements for GH1 in development, adipocyte morphology, and lipid mobilization in a teleost.

Materials and Methods

Fish rearing, genetic mapping, and genotyping

All experiments using zebrafish were conducted in conformity with the Public Health Service Policy on Humane Care and Use of Laboratory Animals using protocols approved by the Institutional Animal Care and Use Committee of the University of Washington and the University of North Carolina at Chapel Hill. All fish were reared at 28.5°C with a 12-hour light, 12-hour dark cycle. Mutations were induced in the inbred AB^wp genetic background by ethyl-N-nitrosourea mutagenesis, and mutant screening was accomplished by early pressure gynogenesis (36). Male vizzini produce viable sperm, and females produce small numbers of viable eggs. Mapping families were generated by crossing vizzini mutants to the commonly used inbred genetic background wik, followed by intercrossing of these F1s to generate a mapping panel of F2 individuals. vizzini was assigned to a chromosome by bulked segregant analysis and subsequently mapped at higher resolution using existing and newly developed microsatellite markers. Exons of candidate genes were amplified from genomic DNA, sequenced directly on a 3730 Genetic Analyzer, and compared with sequences derived from the unmutagenized AB^wp and other genetic backgrounds. A subset of map cross individuals was subsequently genotyped for a premature stop codon identified within the gh1 open reading frame by amplifying a 236-bp fragment, then digesting this product to detect an MseI restriction site introduced by the presumptive lesion.

Recombinant GH injections

Recombinant chicken GH (rGH) (Sigma, St Louis, Missouri) was reconstituted in sterile ddH₂O at 5 ng/nL. At 38 days postfertilization (dpf), the smallest fish from a vizzini/+ intercross were selected to enrich for vizzini mutants (9/30 fish). These fish were individually housed in glass beakers, and their growth assessed over 9 days. Fish were categorized as mutant or wild-type sibling according to growth during this period; 4 fish were determined to be vizzini mutants, and 5 fish were wild-type siblings. In a separate experiment, individual zebrafish of different standard lengths (SLs) were weighed, and quadratic polynomial regression was used to predict body weight from SL (weight in milligrams = −48.10 + 7.36*SL + 0.68*(SL − 8.94)², R² = 0.98). Exogenous injection of 10-100-μg/g body weight rGH induces somatic growth in multiple fish species (37–39). Therefore, 50-μg/g rGH was injected into the abdominal cavity of each fish every 2 days; SL was recorded after 9 days. Somatic growth over the 9-day period without rGH (−rGH) and with 50-μg/g exogenous rGH (+rGH) was compared using a paired t test.

Analyses of growth and developmental timing

For analyses of growth before the attainment of independent feeding, a homozygous male vizzini mutant was crossed with a vizzini/+ female by in vitro fertilization, and the resulting backcrossed progeny were used for analyses of embryonic growth. Twenty-four individuals were placed in 3% methyl cellulose and imaged at each of 3 stages: 50, 74, and 100 hours after fertilization. Each embryo was genotyped after imaging, and images were measured in ImageJ (40) to determine embryonic length and head-trunk angle (41). Values for head-trunk angle were ln-transformed to correct for heteroscedasticity of residuals; embryonic length and transformed head-trunk angle were analyzed by ANOVA in R (42) to test for effects of age and genotype.

To analyze postembryonic growth, fish from a backcrossed clutch were reared individually in plastic cups and imaged daily. Images were measured in ImageJ for SL and staged according to external anatomical features (34). Differences in the age and SL of wild-type and vizzini mutants upon attaining defined stages of postembryonic development were analyzed by ANOVAs to test effects of genotypes, developmental stages, and interaction between the two; tests controlled for repeated measures of individuals (nested within genotypes). Both dpf and SL were ln-transformed to normalize residual variances across factor levels. Independent analyses of the likelihoods of fish attaining specific stages as functions of their genotype, age, SL, and interactions among these variables by multiple logistic regression gave concordant results (details available from D. Parichy on request). Statistical analyses were performed in JMP 8.0 (43).

All imaging was performed with an Olympus SZX-12 stereomicroscope (Olympus, Tokyo, Japan), and digital images were collected with a Zeiss Axiocam HR using Zeiss Axiovision software (Zeiss, Oberkochen, Germany).

Total RNA isolation, cDNA synthesis, primer design, and quantitative RT-PCR (qPCR)

Homozygous and heterozygous vizzini siblings were collected at 10 dpf after 4 days of feeding (∼10-mm SL). Larvae were anesthetized in MS222, 1-1.5 mm of tail tissue were collected for DNA extraction and genotyping, and the rest of each larva was stored individually in RNAlater (Ambion, Inc, Austin, Texas) while each individual was genotyped. RNA was isolated from 3 vizzini homozygotes and 3 heterozygotes and deoxyribonuclease treated using the RNAqueous Micro kit (Ambion, Inc) according to the manufacturer's protocol. Superscript III (Invitrogen, Carlsbad, California) was used in cDNA synthesis reactions primed with oligo dT according to the manufacturer's protocol.

Primer sets (forward, reverse, 5′ to 3′) were designed using Primer3 (44) to span intron-exon boundaries, and each qPCR primer was designed with a 5′ tail (AATAAATCATAA) to improve specificity (45). To detect gh1 transcripts, RT-PCR was performed using primers with the sequence GAAAGCCTCCGAAAACCAG and GCAGAACGACAGAGGGAAGA. For qPCR, the following primers were used: igf1a GTGGAGACAGGGGCTTTTATTT, CTTTGAAAGCAGCATTCGTC; igf2a AGTGTCACAGGCTCTTCACAAG, GATGGGACTCCTCTCCTTAACC; igf2b AAACGTTATGTGGCGGAGAG, ATCCCACGATTTTGAGAACG; igf3 GCGGACGAGAACTAGTGGAC, CTACGAGCTGCTCCAGGTTT; and β-actin AATCCCAAAGCCAACAGAGA, CGACCAGAAGCGTACAGAGA.

RT-PCR to assay the presence of gh1 transcripts was performed in 25-μL reactions using 0.25-μL Taq polymerase with 2.5-μL 10× buffer and 0.5-μL 10mM free deoxyribonucleotides (all from NEB, Ipswich, Massachusetts) with 2-μL 1:10 diluted cDNA and 0.5 μL of 10μM. Reactions for β-actin and gh1 were amplified simultaneously at 94°C for 3 minutes followed by 30 cycles of 94°C for 30 seconds, 58°C for 30 seconds, and 72°C for 20 seconds, followed by a final step of 72°C for 5 minutes.

qPCR was performed on 3 technical replicates for each of 3 biological replicates. Reactions were performed 25-μL reaction volumes using 12.5-μL Power SYBR Green PCR Master Mix (Applied Biosystems, Foster City, California), 1.5-μL specific primer mix (5μM) and 3-μL diluted cDNA. Cycling was performed on a StepOnePlus Real-Time PCR System (Applied Biosystems) with an initial denaturing step at 95°C for 20 minutes followed by 40 cycles of 95°C for 20 seconds, 58°C for 30 seconds, and 72°C for 30 seconds. This was followed by a melt curve stage beginning annealing at 58°C and increasing 3°C with each step. No-template control reactions were performed for each primer set, and all values were normalized to β-actin. qPCR data were collected and analyzed using StepOne Software (version 2.1; Applied Biosystems).

Adipose tissue visualization, area morphometrics, and Folch lipid assay

To characterize accumulation of neutral lipids within adipose depots during developmental progression, larvae were reared under standard conditions until they had attained approximately 4.5-mm SL, at which point they were stained with Nile Red as described (28, 35). Images of stained larvae, anesthetized using 0.4× MS222 supplemented with 0.003% isoflurane (Baxter, Deerfield, Illinois), were taken using a Leica MZ 16F fluorescence stereomicroscope (Leica, Heerbrugg, Switzerland) equipped with a green fluorescent protein long-pass emission filter set (HQ480/40× and HQ510LP), a red-green-blue liquid crystal color filter module, and a Retiga 2000R Fast 1394 cooled digital monochrome camera (QImaging, Tucson, Arizona). Two-dimensional area measurements of pancreatic, abdominal, perirenal (2D), and sc adipose tissues (SATs) were taken in ImageJ and correlated with SL, also measured in ImageJ (32). Adipose areas were assessed by analyses of covariance in JMP testing effects of SL, genotype, an SL x genotype interaction, as well as repeated measures of individuals (nested within genotypes). Areas were square-root transformed before analysis to correct for differing variances across treatments and sizes.

To determine whether wild-type and vizzini mutants reached specific stages of adipose development at significantly different sizes, SLs were analyzed by mixed model ANOVA after ln-transformation to normalize variances in residuals and after controlling for differences attributable to main fixed effects of stage and interactions between stage and genotype, as well as effects attributable to individuals (nested within genotypes).

To generate cross-sections for visualizing gross internal morphology, vizzini mutants and size-matched wild-type heterozygotes were fixed with 4% paraformaldehyde and embedded in 4% agarose, then sectioned by vibratome to 400 μm.

Total lipid extraction from approximately 10-mm SL wild-type siblings (at ∼1 mo of age) and vizzini mutants (at ∼2.5 mo of age) was performed using the Folch procedure as described (28).

Longitudinal confocal analysis of SAT growth dynamics

To visualize adipose growth at cellular resolution, we used longitudinal in vivo imaging in live animals. Every 6 days, individual wild-type siblings and vizzini mutants were stained with the neutral lipid dye, LipidTOX (1:5000; Invitrogen) (35), and their SAT was imaged by confocal microscopy. Superficial SAT was chosen for this analysis, because unlike visceral or perirenal adipose, it can be readily imaged at juvenile stages. At each time point, numbers and sizes of LDs in sc tissues were measured, producing a longitudinal description of LD hyperplasia (ΔSAT-LD#) and hypertrophy (ΔSAT volume) for each animal. Larvae were anesthetized using 0.4× MS222 supplemented with 0.003% isoflurane to facilitate visualization, 1-mg/mL epinephrine (Sigma) was used to condense melanosomes of overlying melanophores. Between imaging time points, larvae were housed within individual wells of a 6-well plastic culture plate; 90% water changes were conducted daily together with scraping, and removal, of debris collecting on the bottom surface of the well. Each fish was fed twice daily with approximately 30 live brine shrimp supplemented with a 3:2 mixture of Golden Pearl powdered food (Brine Shrimp Direct, Ogden, Utah) and spirulina (Aquatic Eco-Systems, Apopka, Florida). LD size and counts were measured in ImageJ from Z-stacks (635-μm² field) obtained at the anteriormost extent of the SAT. The 635-μm² region of interest was kept constant across all time points and fish. LD diameters were ln-transformed before ANOVAs testing effects of genotype, time points and their interaction, and individuals (nested within genotypes).

Nutrient deprivation

Juvenile wild-type siblings and vizzini mutants were identified at approximately 2 months of age based on size. Fish were stained with Nile Red, imaged, and then placed in 200 mL of system water in glass beakers. Fish were imaged before starvation (fed state) and days 8 and 21 of starvation. The average size of wild-type siblings was 17-mm SL and vizzini mutants 11-mm SL. Body area was measured by tracing the 2D outline of larvae in ImageJ. LD number was quantified within a 1300- × 800-μm area immediately dorsal to the anus. LD diameters were ln-transformed before ANOVAs as described above.

Results

A premature stop codon within gh1 causes the vizzini phenotype

The recessive, homozygous viable vizzini mutant exhibits a severe growth defect. By 6 months after fertilization, wild-type zebrafish have grown to approximately 30-mm SL, whereas homozygous vizzini mutants reach only approximately 10-mm SL and never exceed approximately 12-mm SL (Figure 1A). We mapped the vizzini mutation to chromosome 3 within 0.9 cM of a microsatellite located in BAC clone AL929222 at 23.5 Mb (18 recombinants of 2112 individuals genotyped). Inspection of this genomic region revealed gh1 at 22.5 Mb, and sequencing this locus revealed an A→T transversion in the third exon (Figure 1B), resulting in the substitution of a premature stop codon for lysine-64 in the GH1 protein. Sequencing exons from additional candidate genes near AL929222 (col1a1, 23.5 Mb; igf2bp1, 23.8 Mb; nr1d1, 23.8 Mb; and zgc:73293, 23.9 Mb) failed to reveal either missense or nonsense mutations. To further test cosegregation of the vizzini mutant phenotype and gh1 lesion, we genotyped 370 individuals (including 18 individuals recombinant between vizzini and the microsatellite within AL929222) for the presence of the gh1 A→T transversion, using an MseI restriction site introduced by the lesion. These analyses revealed no recombinants, consistent with allelism of vizzini and gh1 (see Figure 1C). The premature stop codon within gh1 was predicted to result in nonsense-mediated mRNA decay and reduced levels of gh1 transcript in vizzini mutants as compared with the wild type; this prediction was confirmed by RT-PCR (Figure 1D). Additionally, we tested whether injection of exogenous rGH could rescue growth in vizzini mutants. Repeated injection with rGH induced a significant increase in growth of mutants (P < .01) (Supplemental Figure 1, published on The Endocrine Society's Journals Online web site at http://endo.endojournals.org), demonstrating that exogenous rGH is sufficient to rescue somatic growth in vizzini mutants. Together, these data support the conclusion that the vizzini phenotype results from a premature stop codon in the zebrafish orthologue of the GH gene, gh1.

Figure 1.

Dwarfism in vizzini mutants linked to lesion in gh1. (A) Heterozygous and homozygous vizzini at their respective full sizes. Scale bar, 5 mm. (B) A→T transversion in the vizzini mutant (arrow) introduces a premature stop codon at K64 in the third exon of gh1. S, serine; K, lysine; I, isoleucine; * indicates stop codon. (C) Genotyping of individuals using the MseI restriction site introduced by the mutation. (D) Transcripts of gh1 are undetectable in vizzini mutants by RT-PCR. WT, wild-type; viz, vizzini homozygote; viz/+, vizzini heterozygote.

Somatic growth decoupled from development and age in vizzini mutants

To assess roles for gh1 in zebrafish development, we compared growth and developmental progression of wild-type and vizzini mutant individuals from embryonic through larval and adult stages. Before the onset of independent feeding, vizzini mutant embryos and early larvae (50, 74, and 100 hours postfertilization) developed at a rate statistically indistinguishable from the wild type, reminiscent of the GH1-independent growth during embryonic, fetal, and early postnatal periods in humans (46). Nevertheless, size differences became apparent after the onset of feeding (5 dpf) and were increasingly pronounced with age (Figure 2A). To determine whether gh1 promotes developmental progression in addition to growth, we assessed the appearance of morphological features diagnostic for different postembryonic stages (34). We predicted that if developmental events are promoted by GH1 signals, or if different developmental events exhibit differential GH1 dependencies, then vizzini mutants might exhibit heterochronic changes in stage progression, with disordered or delayed appearance of developmental milestones. We found that stage-specific milestones appeared in their normal order in vizzini mutants but that stages were attained at progressively older ages than in wild-type fish (Figure 2B). Nonetheless, stages of development were reached by mutants at sizes somewhat smaller than those of wild-type controls (Figure 2C and Supplemental Figure 2). Thus, in the absence of gh1, fish growth was retarded relative to age, but developmental stage was advanced relative to size.

Figure 2.

vizzini mutants show decoupled growth and development but similar igf expression at an early stage. (A) In comparison with their wild-type vizzini/+ siblings (gray), vizzini mutants (black) exhibited retarded growth beginning shortly after first feeding. (B and C) Relationships between developmental stage and age (B) and developmental stage and size (C) differed between vizzini mutants and wild-type siblings. X-axes show developmental milestones describing early larval stages (left) through late larval stages (right) as defined in Ref. 34. (B) vizzini mutants (black bars) took longer to reach defined developmental stages than their wild-type siblings (open bars; F_1,24 = 43.1, P < .0001), and the magnitude of this difference was increasingly severe at later stages (stage x genotype interaction, F_12,598 = 32.1, P < .0001). (C) vizzini mutants also were smaller than their wild-type siblings upon reaching defined developmental stages (F_1,28 = 48.0, P < .0001), with differences greater at later stages (stage x genotype interaction, F_12,606 = 5.2, P < .0001). N = 642 observations, 24 individuals. Stages as described in Ref. 34; FLe, notochord flexion; CR, caudal fin ray appearance; AC, anal fin condensation; DC, dorsal fin condensation; MMA, metamorphic melanophore appearance; aSB, inflation of anterior swim bladder lobe; AR, anal fin ray appearance; DR, dorsal fin ray appearance; PB, pelvic fin bud appearance; PR, pelvic fin ray appearance; SP, squamation through posterior; SA, squamation through anterior. Bars indicate means ± 1 SD. (D) Fold differences in expression of igf loci in vizzini mutants relative to heterozygous siblings, all standardized internally to β-actin. Bars indicate means ± 1 SEM. WT, wild-type; viz, vizzini mutants.

No detectable change in igf expression in vizzini mutants at 10 dpf

In mammals and in fish, the growth-promoting effects of GH are mediated by GH-dependent transcriptional regulation of IGF family genes (47–50). To investigate GH1 regulation of IGF expression in zebrafish, we used qPCR to compare expression of igf genes between wild-type and vizzini mutants at 10 dpf (∼4-mm SL), when overall morphological defects in mutants were still relatively subtle (see Figure 2A). These analyses did not reveal significant differences in igf gene expression between vizzini mutants and their heterozygous siblings at this early stage, although overall expression levels were low (delta threshold cycle = 4.8-11.1), limiting our power to detect subtle differences that might have been present (Figure 2D).

vizzini mutants exhibit obesity

GH regulates adiposity in mammals (51, 52), and our identification of vizzini as a gh1 mutant provided the opportunity to test whether GH1 regulates adipose development or accumulation in zebrafish. To test these possibilities, we visualized neutral lipids in live vizzini mutants and siblings (28, 35). The first zebrafish adipocyte LDs appear in a visceral location associated with the pancreas (Figure 3, A and B) (28, 33). LDs next develop intra-abdominally, associated with the swim bladder (Figure 3, D and E), followed by perirenal (Figure 3, G and H) and sc locations (Figure 3, G and H). Consistent with analyses of postembryonic stages (Figure 2C), vizzini mutants attained specific stages of adipose development in their normal order but attained these stages while smaller and older than wild-type siblings (Supplemental Figure 3). We conclude that GH1 is not required for establishing adipose tissue depots nor for initiating morphogenetic events that characterize these depots.

Figure 3.

vizzini mutants exhibit greater size-specific adiposity than wild-type larvae. (A–L) Images of Nile Red-stained neutral lipid in wild-type and vizzini mutant larvae. Wild-type siblings (A, B, D, E, G, H, J, and K) undergo expansion of pancreatic (arrows), abdominal (arrowheads), sc (open arrowheads), and perirenal (open arrows) adipose tissues. These depots are larger in equivalently sized vizzini mutants (C, F, I, and L). sb, swim bladder; white dotted outline, gall bladder. (M–P) Quantification of adipose depot areas reveals that at a given SL, vizzini mutants exhibit significantly greater adiposity than the wild type, as evidenced by significant SL x genotype interactions (M, F_1,93 = 171.2; N, F_1,106 = 173.1; O, F_1,93 = 161.4; P, F_1,33 = 23.8; all, P < .0001; N = 446 observations; 20 wild-type, 20 vizzini mutants). WT, wild-type; viz, vizzini mutants.

To test whether GH1 regulates the accumulation of lipids within adipose depots, we measured 2D areas of depots during larval growth. Comparing vizzini mutants and size-matched wild-type larvae revealed substantially more adipose tissue in the mutants (Figure 3, A–L). Moreover, the 2D areas of the four examined adipose tissues became increasingly large relative to size-matched wild-type individuals as body length increased (Figure 3, M–P). In larger larvae (eg, 10.1-mm SL), both ventral abdominal and SATs were markedly expanded in vizzini mutants compared with the wild type; indeed, we found lipids to be ectopically deposited along the entire flanks of mutants (Figure 4, A–F). The expansion of adipose tissue was clearly apparent even in unstained transverse sections (Figure 4, G–J). Measurement of absolute lipid levels after Folch extraction further demonstrated significantly increased lipid contents of vizzini mutants compared with size-matched wild-type controls (Figure 4K). These findings demonstrate that for a given size, vizzini mutants exhibit increased amounts of adipose tissue compared with wild type, and confirm roles for GH1 in the regulation of both sc and visceral adiposity in zebrafish.

Figure 4.

Increased visceral and SAT in late larval vizzini mutants. Images of Nile Red-stained neutral lipid in wild-type siblings and vizzini mutant zebrafish (A–F). Cross-sections through trunk of wild-type and vizzini mutants (G–J). (A–F) Nile Red fluorescence (white signal) is increased in both visceral (asterisks, A and D) and SAT (arrows, C and F) in vizzini mutants. White asterisks denote melanophores of the adult pigment stripes. (G–J) Visceral and SAT, visible as globular LDs, is increased in vizzini mutants (arrows in I and J). Region shown is just anterior to that shown in C and F. Scale bar, 1 mm. (K) Folch extraction significantly increased total lipid in vizzini mutants (wild-type sibling, n = 4; vizzini mutants, n = 5). Student's t test, P < .0001. Bars indicate means ± 1 SEM. WT, wild-type; viz, vizzini mutants.

Increased LD hypertrophy in vizzini mutants

GH is implicated in regulating mammalian adipose morphology through both hyperplastic and hypertrophic mechanisms (53, 54). To determine whether zebrafish GH1 plays a role in adipocyte hyperplasia and hypertrophy in vivo, we used fluorescent lipid dyes to label adipocyte LDs (28, 35) and performed longitudinal confocal live imaging to visualize changes in LD number and size in SATs in wild-type and vizzini mutant larvae (Figure 5, A–E). Mature adipocytes contain a single large LD, and the number and size of LDs serve as a proxy for hyperplasia and hypertrophy, respectively (35, 55). These cellular-level confocal analyses of longitudinal in vivo adipose morphology and dynamics in zebrafish provide a complement to existing mammalian in vivo imaging techniques, including nuclear magnetic resonance, bioluminescence, and computed tomography (54, 56, 57).

Figure 5.

Longitudinal confocal imaging of SAT growth in wild-type and vizzini mutant siblings. Confocal imaging of LipidTOX-stained SAT in wild-type sibling (B and C) and vizzini mutants (D and E). (A) Schematic of imaging procedure. Animals were imaged 3 times at 6-day intervals (d1, day 1; d6, day 6; d12, day 12). (B–E) Wild-type sibling (B and C) showing progressive SAT growth over 3 time points. vizzini mutant showing extreme SAT-LD hypertrophy over 3 time points (D and E). Regions within white boxes are shown at higher magnification. (F) Average LD diameter varied among time points (P < .0001), and vizzini mutants exhibited larger LD diameters than wild-type, particularly evident at the final time point (as evidenced by a significant genotype x time-point interaction, F_2,1538 = 37.4, P < .0001). Groups with the same letters are not significantly different (P > .05) by Tukey honestly significant difference test. (G) Mean number of LDs exhibited an overall increase at later time points (F_2,24 = 4.9, P < .05) but did not differ significantly between genotypes (F_1,25 = 0.9, P = .4), nor did differences between time points depend on genotype (genotype x time point interaction, F_2,21 = 0.6, P = .6); accordingly, post hoc pairwise comparisons of genotype, and time point means revealed no significant differences, although when genotypes were combined, d12 means were significantly greater than d1 means. Scale for all white outlined boxes is 100 μm. Bars indicate means ± 1 SEM. N = 5 wild-type, 4 vizzini mutants. WT, wild-type; viz, vizzini mutants.

Over the course of the observational period, individuals exhibited expected levels of somatic growth, indicative of overall good health (Supplemental Figure 4). In wild-type siblings, SAT LDs increased both in size and number (Figure 5, F and G), demonstrating that SAT growth is derived both from adipocyte hypertrophy and hyperplasia. Strikingly, vizzini mutants exhibited a significantly increased diameter of sc LDs compared with the wild type (Figure 5F), suggesting that GH1 normally constrains hypertrophy. By contrast, vizzini mutants did not exhibit a significant change in LD number in comparison with the wild type (Figure 5G). Taken together, in vivo imaging reveals that GH1 deficiency in zebrafish causes hypertrophic expansion of SAT without increasing LD number. These results are consistent with the enlargement of sc adipocytes observed in GH-deficient humans (20) and the increased sc adipose mass in dwarf mouse lines with disrupted GH signaling (58), suggesting that GH1 plays an analogous role in regulating adipocyte morphology in fish as in mammals.

Adipose stores in vizzini mutants resistant to mobilization during starvation

During starvation in mammals, GH stimulates lipolysis and mobilization of lipids stored within adipose tissues (26). Because of the enlarged sc adipocyte LDs observed in vizzini, we hypothesized that GH1 functions in lipid mobilization in zebrafish. To test lipid mobilization in response to a negative energy balance, juvenile wild-type and vizzini mutants were starved for 21 days (Figure 6). In wild-type siblings, the number of LDs within SATs decreased during nutrient deprivation and were often undetectable after 21 days of starvation (Figure 6, A–F). Concordantly, the size of the remaining LDs decreased over the course of nutrient deprivation (Figure 6, M and N). In contrast, sc lipids in vizzini were not mobilized even after 21 days of starvation (Figure 6, G–L), we observed no measurable reduction in either the size or number of LDs (Figure 6, M and N). Thus, we conclude that GH1 deficiency impairs the mobilization of sc lipid stores during zebrafish nutrient deprivation.

Figure 6.

vizzini mutants exhibit reduced SAT mobilization during nutrient deprivation. Images of Nile Red-stained wild-type and sibling vizzini mutants (A–L). (A–F) Images from a representative wild-type sibling during the course of 21 days of starvation. Note overall reduction in Nile Red stain between A and C. SAT-LDs gradually disappear during nutrient deprivation (arrows, D–F). (G–L) Images from a representative vizzini mutant, illustrating increased SAT during the fed state (G and J) and failure of SAT-LD reduction during starvation (I and L). (M) Mean SAT-LD diameter decreased slightly in wild-type larvae upon starvation but did not change significantly with starvation in vizzini mutants (genotype x time point interaction, F_2,1945 = 175.2, P < .0001). (N) Lipid SAT-LD number decreased after 21 days of starvation in wild-type siblings but not in vizzini mutants (genotype x time point interaction, F_2,27 = 4.5, P < .05). In M and N, bars are means ± 1 SEM, with letters indicating means not significantly different (P > .05) by Tukey honestly significant difference comparisons. N = 6 wild-type, 5 vizzini mutants. WT, wild-type; viz, vizzini mutants.

Discussion

In this study, we demonstrate that dwarf vizzini mutant zebrafish harbor a premature stop codon in gh1, identifying the first loss-of-function GH fish, and we determined that GH1 serves as an essential regulator of size and adiposity in zebrafish. Overexpression of GH1 can substantially increase growth rate and is thus a frequent target for agricultural biotechnology. Indeed, the molecular targets and physiological effects of GH are important in an aquaculture context, and transgenic lines with exogenous GH genes have been established in a number of commercially important fish species, including tilapia and salmon (59–61). Moreover, GH is an important regulator of lipid metabolism in humans, and understanding the cellular roles of this hormone is applicable to the worldwide obesity epidemic. Other teleosts and mammals (including humans) have a second, extrapituitary GH gene (3, 62, 63). In contrast, zebrafish possess only a single GH gene, gh1 (17), allowing us to scrutinize development and lipogenesis in the absence of any potential compensatory effects of a GH homologue (note however that other members of the GH/prolactin family may bind to GHRs) (17). Moreover, the zebrafish model offers unrivaled potential for visualizing vertebrate tissue dynamics, and we have used this capability to evaluate lipid fluctuations at a cellular level in both wild-type fish and in the absence of GH.

Although the pituitary gland begins expressing gh1 at 2 dpf (64), we found no effect of gh1 loss-of-function on embryonic and early larval growth before feeding. This result corroborates previous findings of GH-independent growth during embryonic and early larval development of zebrafish (65) and during the early postnatal period of humans (46). However, the consequences of GH1 deficiency became apparent during later larval development, and significant size differences were manifested between homozygous vizzini mutants and wild-type siblings by 10 dpf, becoming increasingly severe as development progressed. It is well established that the GH-IGF axis regulates postembryonic somatic growth in mammals (66, 67). However, we found that whole-fish expression levels of igf transcripts were comparable between vizzini mutants and heterozygous siblings at 10 dpf, suggesting that IGF expression may be GH independent at these stages in zebrafish. Nonetheless, expression levels at this early stage were low for each of the genes in both wild types and mutants, and this may have precluded the detection of any subtle gh1 dependencies. Analyses of tissue-specific igf gene expression profiles at later stages would be informative for further testing roles for the GH-IGF axis in regulating zebrafish growth.

Developmental progress is highly correlated with size in zebrafish, and size alone predicts the state of most developmental traits with over 90% accuracy (age is far less predictive of developmental stage) (34). To test whether GH1 is required to maintain this close relationship between size and stage, we compared developmental states in mutants with those in age-matched wild-type fish. We found that the order in which anatomical somatic milestones appear is unaffected by the lack of gh1 in zebrafish, suggesting that GH1-independent mechanisms regulate postembryonic developmental progress. Even so, because vizzini mutants grow much more slowly than normal fish, developmental stages are attained at older ages than in wild type. Our results therefore indicate that GH1 is not essential for developmental progression and that GH1-independent mechanisms promote the appearance of stage-specific markers after a requisite developmental state or minimum overall size is reached.

We submit the vizzini mutant as a useful new model for understanding vertebrate obesity. Our data demonstrate that adipose tissue is increased and adipose development is advanced relative to size in vizzini mutants. Thus, vizzini mutants satisfy the common criterion of being overweight or obese: increased adipose tissue relative to a linear measure of size (height in humans, SL in zebrafish). In contrast to mammalian obesity models, the zebrafish system offers excellent potential for high-resolution in vivo imaging, and our study addresses long-established technical limitations concerning in vivo imaging of adipose tissue. Several approaches have been developed to study adipose tissue dynamics in mammals, including ex vivo culture of adipose tissue, intravital confocal microscopy after surgical incision and coverslip implantation, and mathematical modeling using data from reiterated adipose biopsies (68–72). These approaches have provided novel paradigms in adipose growth, particularly in the context of differing endocrinological contexts; however, these methods are inherently invasive and do not permit visualization of cellular dynamics within the intact physiological context of in vivo adipose tissue. To our knowledge, this study provides the first noninvasive and longitudinal visualization of in vivo adipose tissue formation and growth.

In this study, we demonstrated that vizzini mutants have increased sc and visceral adipose tissue relative to body size. The observed increase in sc adiposity is similar to that exhibited by GH-deficient mouse models, including an ethyl-N-nitrosourea-induced missense Gh^sma1 mutation (73), a Ghr mutant (58, 74), and the Snell dwarf line, in which development of the anterior pituitary gland is disrupted (75, 76). Similarly, GH-deficient and GH-insensitive humans develop abnormally large amounts of SAT (77–79). Thus, it appears that GH has a conserved role in regulating lipid storage within SAT from humans to zebrafish. GH-deficient mouse models exhibit visceral depot-specific responses to GH. Retroperitoneal adipose was increased in GHR mutants (58); however, epididymal adipose was either unchanged (58, 80) or reduced (81) when normalized to body weight. Gh^sma1 mutants exhibit increased visceral adipose tissue; however, it is unclear which visceral adipose tissue was analyzed (73). In vizzini, 3 independent visceral adipose tissues (pancreatic, abdominal associated with the swim bladder, and perirenal) were larger relative to body length. Therefore, our data are consistent with a role for GH in regulating the size of visceral adipose tissue.

We found that the increased overall size of the SAT was caused by a dramatic increase in the size of adipocyte LDs. GH impairment in mammals is characterized by large sc adipocytes (2–4), suggesting a role for GH in regulating the size of adipocytes in both mammals and teleosts. The role of GH in regulating the number of adipocytes within adipose tissue is less clear. Although GH-deficient humans show reduced numbers of sc adipocytes and increase the number of adipocytes upon treatment with GH (20), GHR knockout mice exhibit increased numbers of sc adipocytes (82). Our investigations did not reveal significant modifications in the number of SAT LDs in vizzini mutants relative to wild-type controls; further analyses await the development of a transgenic line labeling zebrafish preadipocytes.

Mammalian GH is centrally involved in the use of stored lipid from adipose tissue, and GH stimulates lipolysis and lipid oxidation (26). Our in vivo imaging approach revealed a dramatic increase in size of SAT LDs in vizzini mutants compared with wild type, suggesting that lipolysis is deficient in vizzini SAT. Further, 21 days of starvation did not measurably decrease either the size or number of SAT LDs in vizzini mutants, suggesting that lipolysis and mobilization of adipose tissue are defective in gh1 mutants and that, as in mammals, GH1 plays a critical role in lipid mobilization in teleosts.

The increased LD hypertrophy and greater total quantity of adipose tissue relative to size suggest vizzini may serve as a zebrafish model of human GH-mediated obesity. Several models of zebrafish obesity have previously been described (30, 83, 84). However, obesity in these models was defined only through measures of body mass and size, triglyceride levels, and adipose tissue histology and did not use in vivo imaging nor quantify effects on different adipose tissues. Therefore, to our knowledge, the analysis of adipose tissue growth in vizzini mutants in the current study is the most comprehensive cellular investigation of obesity in zebrafish.