Developmental Analysis and Influence of Genetic Background on the Lhx3 W227ter Mouse Model of Combined Pituitary Hormone Deficiency Disease

Abstract

Combined pituitary hormone deficiency (CPHD) diseases result in severe outcomes for patients including short stature, developmental delays, and reproductive deficiencies. Little is known about their etiology, especially the developmental profiles and the influences of genetic background on disease progression. Animal models for CPHD provide valuable tools to investigate disease mechanisms and inform diagnostic and treatment protocols. Here we examined hormone production during pituitary development and the influence of genetic background on phenotypic severity in the Lhx3^{W227ter/W227ter} mouse model. Lhx3^{W227ter/W227ter} embryos have deficiencies of ACTH, α-glycoprotein subunit, GH, PRL, TSHβ, and LHβ during prenatal development. Furthermore, mutant mice have significant reduction in the critical pituitary transcriptional activator-1 (PIT1). Through breeding, the Lhx3^{W227ter/W227ter} genotype was placed onto the 129/Sv and C57BL/6 backgrounds. Intriguingly, the genetic background significantly affected viability: whereas Lhx3^{W227ter/W227ter} animals were found in the expected frequencies in C57BL/6, homozygous animals were not viable in the 129/Sv genetic environment. The hormone marker and PIT1 reductions observed in Lhx3^{W227ter/W227ter} mice on a mixed background were also seen in the separate strains but in some cases were more severe in 129/Sv. To further characterize the molecular changes in diseased mice, we conducted a quantitative proteomic analysis of pituitary proteins. This showed significantly lower levels of PRL, pro-opiomelanocortin (ACTH), and α-glycoprotein subunit proteins in Lhx3^{W227ter/W227ter} mice. Together, these data show that hormone deficiency disease is apparent in early prenatal stages in this CPHD model system. Furthermore, as is noted in human disease, genetic background significantly impacts the phenotypic outcome of these monogenic endocrine diseases.

The pituitary gland is critically important for growth, reproduction, and the maintenance of physiological homeostasis in mammals. The anterior pituitary (AP) secretes polypeptide hormones from five cell types: corticotropes (releasing ACTH), gonadotropes (FSH and LH), thyrotropes (TSH), somatotropes (GH), and lactotropes (prolactin [PRL]). Hormones produced by the AP regulate growth and metabolism (GH and TSH), the stress response (ACTH), reproduction (FSH, LH, and PRL), the thyroid (TSH), and lactation (PRL). During development, after inductive events between the pituitary primordium (Rathke's pouch) and the diencephalon, the actions of transcription factors such as HESX1, PITX1/2, LIM (lin-1, Isl-1, and mec-3)-homeodomain 3 (LHX3), LHX4, PIT1, and PROP1 are key to establishment of the hormone-secreting cell types of the AP (1–4).

The LHX3 transcription factor is expressed during the early stages of pituitary and nervous system development and plays essential roles in establishment of the differentiated cells of the AP and specialized neurons in the brain and spinal cord (5–10). Mice lacking function of the entire Lhx3 gene exhibit structural arrest of pituitary gland development, failed differentiation of the lactotrope, gonadotrope, somatotrope, and thyrotrope AP lineages, and die shortly after birth presumably due to nervous system deficits (5–7). Pituitary target genes of LHX3 include those encoding the β-subunits of FSH and TSH (FSHβ and TSHβ), α-glycoprotein subunit (αGSU, the common subunit of the FSH, LH, and TSH glycoprotein hormones), PRL, GnRH receptor, and the later-stage PIT1 transcription factor (eg, 9, 11–14).

Mutations in the human LHX3 gene are associated with severe combined pituitary hormone deficiency (CPHD) diseases. Affected patients have insufficiencies in GH, TSH, LH, FSH, and PRL hormones and also frequently have associated syndromic symptoms such as ACTH deficiency, mental retardation, and sensorineural hearing loss (15–23). In addition to CPHD, most characterized patients with LHX3 mutations also display a rigid cervical spine resulting in limited head rotation, a symptom that is considered to result from loss of LHX3 function during early spinal cord neuron development. Described LHX3 gene mutations in humans, mice, and dogs act in recessive fashion (eg, 6, 18, 19, 24, 25); however, a recent report indicates that symptoms in the heterozygous condition are possible (26).

One form of LHX3 gene mutation (W224ter) introduces a stop codon resulting in a predicted truncated form of the LHX3 protein lacking the transcriptional activation domains required for pituitary gene activation (19, 27). Patients with the W224ter mutation have symptoms associated with CPHD, but they do not have the cervical spine rigidity, suggesting that the nervous system functions of LHX3 are preserved (19). To test the hypothesis that the pituitary and nervous system functions of LHX3 are separable, a knock-in mouse strain was created carrying an equivalent mutation (Lhx3^W227ter) (24). The resulting Lhx3^{W227ter/W227ter} homozygous mice are viable and provide a useful animal model for CPHD, having deficiencies of ACTH, GH, PRL, TSH, LH, and FSH (24). The mice display dwarfism, lethargy (due to thyroid insufficiency), and reproductive problems but no nervous system deficits (24).

Pediatric CPHD diseases associated with genomic mutations in transcription factor genes such as LHX3 are developmental in nature. However, because cases are typically diagnosed postnatally and due to the inaccessible location of the pituitary gland, little is known about the nature and time course of disease onset in human patients, especially at the molecular and cellular levels. Furthermore, a range of disease symptoms and severity of hormone deficiencies has been observed in patients with LHX3 mutations (15–23). This phenotypic disease variation may be due in part to differences in genetic background and the actions of modifier genes. The viable Lhx3^{W227ter/W227ter} mouse model has hormone deficiencies that model the human disease, providing a valuable reagent to analyze the molecular nature and progression of the human disease condition. We therefore examined the expression of pituitary hormones during development and the influence of genetic background in Lhx3^{W227ter/W227ter} mice. Additionally, we conducted a proteomic analysis to identify and quantify proteins that might be affected by the CPHD-causing mutation.

Materials and Methods

Animals and tissue harvesting

Lhx3^{W227ter/W227ter} knock-in mice were housed in standard conditions with access to food and water provided ad libitum and genotyped using PCR as reported (24). Animals were fed 2018SX Teklad Global 18% protein rodent diet. All procedures were approved by the Indiana University Committee on Use and Care of Animals and followed National Institutes of Health Guidelines for the Care and Use of Experimental Animals. Embryos were dissected from females of heterozygous pairs at embryonic day 13.5 (E13.5), E15.5, and E17.5. The day after conception was set as E0.5. Embryos were genotyped by PCR. Tissues were fixed in 4% paraformaldehyde in PBS (137mM NaCl, 5.4mM KCl, 16.2mM Na₂HPO₄, 2.9mM KH₂PO₄). After fixation, all tissues were rinsed in PBS and incubated in 20% sucrose in PBS at 4°C overnight. Tissues were rapidly frozen in Peel-A-Way disposable embedding molds (Polysciences, Warrington, PA) using Tissue-Tek O.C.T. Compound (Sakura, Torrance, California), oriented for sagittal sectioning and stored at −80°C until cryosection.

Cryosections and immunohistochemistry

Frozen tissue blocks were sectioned in 7-μm sections using a CM 1950 cryostat (Leica Microsystems, Buffalo Grove, Illinois). Procedures for fluorescent immunohistochemistry (IHC) using pituitary hormone antibodies from the National Hormone and Pituitary Program (NHPP, Torrance, California) were as reported (24) except the recently generated rabbit antirecombinant PRL antibody (NHPP AFP10712402Rb) (1:100) and rat anti-FSHβ (NHPP 85GP9691bFSHb) (1:500) were used for hormone staining. Similar procedures were used with the anti-PIT1 antibody (28) at 1:1000. At least 2 sections per animal type were examined.

Genetic background analyses

Lhx3^W227ter/+ (129/Sv + C57BL/6 mixed background) (24) were separately backcrossed into 129/Sv or C57BL/6 strains for 6 generations to obtain a 98.4% pure background in each strain. Lhx3^W227ter/+ from generation 6 within each strain were separately intercrossed to obtain wild-type (WT), Lhx3^W227ter/+, and Lhx3^{W227ter/W227ter} progeny. Animals from 129/Sv litters (n = 18) and C57BL/6 litters (n = 11) were genotyped and analyzed in comparison with observations on a mixed background (24). E17.5 embryos were dissected from females of heterozygous pairs for IHC as described above.

Quantitative PCR analyses

Tissue harvest and cDNA preparation using triplicate samples were as described (24). Parallel negative control reactions contained no reverse transcriptase enzyme. The relative abundance of the genes of interest was determined using real-time quantitative PCR in triplicate reactions as described (24). Mouse Pit1 (Pou1f1) gene (Mm00476852_m1) primers and probes were from Applied Biosystems (Carlsbad, California). Data were normalized by determining the relative abundance of 36B4 mRNA. The mouse 36B4 (also known as Rplp0) gene encodes an acidic ribosomal phosphoprotein P0. WT data were normalized to 1, and heterozygote and homozygote data were subsequently adjusted.

Proteomic analyses

WT and Lhx3^{W227ter/W227ter} female adults (n = 5 in each group) were rapidly killed and perfused with ice-cold saline (0.9%). The age of mice ranged from 3 to 6 months (2 WT and 3 Lhx3^{W227ter/W227ter} were 3 months, 2 WT and 1 Lhx3^{W227ter/W227ter} were 4 months, and 1 WT and 1 Lhx3^{W227ter/W227ter} were 6 months). Pituitaries were frozen in liquid nitrogen and then solubilized in lysis buffer (8M urea, 10mM dithiothreitol). Protein concentration was determined by the Bradford method (Bio-Rad, Hercules, California). Samples were reduced and alkylated by triethylphosphine and iodoethanol and subjected to trypsin digestion as described (29). Digested samples were analyzed using a Thermo-Finnigan linear ion-trap (linear trap quadrupole) mass spectrometer coupled with a Surveyor autosampler and mass spectrometry HPLC system (Thermo-Finnigan, San Jose, California). Tryptic peptides were analyzed using a C18 RP column as described (29). Data were searched against the International Protein Index mouse database (IPI Mouse version 3.77 FASTA format) using SEQUEST (version 28 revision 12) algorithms in Bioworks (version 3.3). General parameters were set to: peptide tolerance 2.0 atomic mass units, fragment ion tolerance 1.0 amu, enzyme limits set as “fully enzymatic–cleaves at both ends,” and missed cleavage sites set at 2. Searched peptides and proteins were validated by PeptideProphet (30) and ProteinProphet (31) in the Trans-Proteomic Pipeline (TPP version 3.3.0) (http://tools.proteomecenter.org/software.php). Only proteins with P ≥ .90 and peptides with P ≥ .80 were reported. Label-free protein quantification was performed using IdentiQuantXL software as described (32). The retention time of peptide for its intensity extraction was carried out with an experiment-based algorithm, RetentionTimeXL. A Student's t test was performed to determine the significance of differences between the two groups. False discovery rate (33) was estimated using Q-value software. Data were further analyzed using Interactive Pathway Analysis software (Ingenuity Systems, Redwood City, California).

Results

W227ter knock-in mice have decreased expression of AP hormones during development

LHX3 is critical for the development of the hormone-producing cells of the AP. Therefore, we hypothesized that Lhx3^{W227ter/W227ter} animals would have decreased expression of AP hormones during developmental stages. Using IHC, AP hormone expression was assessed using sections of embryonic pituitary glands in Lhx3^{W227ter/W227ter} and littermate WT embryos from timed pregnancies. ACTH was the only hormone readily detectable at E13.5, and it was notably decreased in the AP of Lhx3^{W227ter/W227ter} animals (Figure 1A) and also later at E15.5 (Figure 1B). Whereas the αGSU, PRL, and TSHβ were detectable at low levels in WT animals at E15.5, these hormones were not observed in the AP glands of Lhx3^{W227ter/W227ter} animals (Figure 1B and data not shown). At both E15.5 and E17.5, the pituitary glands of Lhx3^{W227ter/W227ter} animals often had morphological abnormalities involving hypoplastic AP lobes and sometimes forked/bifurcated structures (eg, Figures 1B and 2–5). At E17.5, ACTH, αGSU, and GH had reduced expression in Lhx3^{W227ter/W227ter} animals, and PRL, TSHβ, and LHβ were not detected in the knock-in mice (Figure 2).

Figure 1.

Delayed onset of AP hormone expression in Lhx3^{W227ter/W227ter} mice during development. ACTH, PRL, and TSHβ hormones were labeled using fluorescent IHC in pituitary sections of homozygote Lhx3^{W227ter/W227ter} and WT mice at E15.5 (n = 5 per group). ACTH was also labeled in E13.5 sections (n = 5 per group). A and B, No specific ACTH expression was observed in Lhx3^{W227ter/W227ter} animals at E13.5, and ACTH expression was decreased at E15.5. B, Low levels of PRL and TSHβ (asterisks) were detectable at E15.5, but no specific staining was seen in Lhx3^{W227ter/W227ter} mutants. Lhx3^{W227ter/W227ter} anterior pituitary lobes often were hypoplastic compared with WT controls (eg, Δ).

Figure 2.

The expression of AP hormones is reduced or absent in embryonic Lhx3^{W227ter/W227ter} mice. ACTH, αGSU, GH, TSHβ, LHβ, and PRL pituitary hormone markers were labeled using IHC in E17.5 pituitary sections of Lhx3^{W227ter/W227ter} homozygote and WT mice. Decreased expression of ACTH, αGSU, and GH was observed for Lhx3^{W227ter/W227ter} animals, and no labeling was detected for TSHβ, LHβ, and PRL (n = 5 per group). Hematoxylin and eosin (H&E) staining (top panels) revealed that Lhx3^{W227ter/W227ter} APs often were hypoplastic with structural abnormalities such as bifurcations (asterisk). PP indicates the posterior lobe; rt, rostral tip.

Figure 3.

Expression of the PIT1 pituitary transcription factor is reduced in Lhx3^{W227ter/W227ter} animals. A, PIT1 protein was labeled using IHC in pituitary sections of Lhx3^{W227ter/W227ter} homozygote and WT mice at E15.5 and E17.5. Lhx3^{W227ter/W227ter} animals had decreased amounts of nuclear staining in PIT1-positive cells (n = 5 per group). Examples of a hypoplastic, bifurcated AP lobe also are noted (asterisks). B, Measurement of Pit1/Pou1f1 transcript levels in pituitary glands of 12-week-old adult mice using real-time quantitative PCR revealed a significant decrease in levels in the Lhx3^{W227ter/W227ter} homozygote knock-in (KI/KI) mice. The asterisk indicates significance when compared with WT controls with P < .05. Data were normalized by determining the relative abundance of 36B4 (acidic ribosomal phosphoprotein P0 [Rplp0 gene]) mRNA. Examples of hypoplastic, bifurcated anterior pituitary lobes are noted (*, asterisks).

Figure 4.

Genetic background impacts the expression of AP hormones in the Lhx3^{W227ter/W227ter} mouse. Pituitary hormone markers were analyzed by IHC in animals on C57BL/6 and 129/Sv backgrounds. AP hormone expression in Lhx3^{W227ter/W227ter} homozygote knock-in and WT mice at E17.5 on a 98.4% C57BL/6 or 129/Sv genetic background. Lhx3^{W227ter/W227ter} animals had decreased expression of ACTH and αGSU and no expression of PRL, FSHβ, TSHβ, or LHβ in both strains. Neither Lhx3^{W227ter/W227ter} nor WT animals on a 129/Sv background had any PRL labeling. Lhx3^{W227ter/W227ter} animals on a 129/Sv background had no GH expression, whereas those on a C57BL/6 background had decreased expression of GH compared with WT. Examples of hypoplastic, bifurcated AP lobes are noted (asterisks).

Figure 5.

Reduced expression of PIT1 protein in Lhx3^{W227ter/W227ter} animals on C57BL/6 and 129/Sv backgrounds. PIT1 protein was detected using IHC in pituitary sections of Lhx3^{W227ter/W227ter} homozygote and WT mice at E17.5 on a 98.4% C57BL/6 or 129/Sv background. Decreased expression was seen in Lhx3^{W227ter/W227ter} animals on a C57BL/6 background, and no expression was seen in Lhx3^{W227ter/W227ter} mice on a 129/Sv background. An example of a hypoplastic, bifurcated AP lobe is labeled (asterisk).

Lhx3^{W227ter/W227ter} mice have decreased expression of PIT1

The PIT1 transcription factor is essential for the development of differentiated somatotropes, lactotropes, and thyrotropes and acts at later stages of pituitary development than LHX3 (4). PIT1 also is known to synergize with LHX3 to activate transcription of other pituitary genes (9). Using IHC, we saw decreased expression of PIT1 protein in Lhx3^{W227ter/W227ter} knock-in embryos at E15.5 and E17.5 (Figure 3A). Additionally, 12-week-old adults had significantly lower expression of Pit1/Pou1f1 mRNA measured by quantitative RT-PCR (Figure 3B).

Genetic background significantly influences the effect of the Lhx3^{W227ter/W227ter} mutation

The W227ter mutation was originally assessed in mice of a mixed background (129/Sv + C57BL/6) (24). To assess the influence of genetic background on the W227ter mutation, we bred Lhx3^{W227ter/W227ter} mice into the individual 129/Sv and C57BL/6 strains for six generations (98.4% pure background). We counted the number of WT, Lhx3^W227ter/+, and Lhx3^{W227ter/W227ter} animals from heterozygous matings that survived to weaning at 21 days. From 18 litters on the 129/Sv background, there were 18 WT and 50 heterozygous animals born and weaned; however, no Lhx3^{W227ter/W227ter} animals survived to weaning. This indicates that W227ter homozygotes are not viable on this mouse background (Table 1). By contrast, animals born on the C57BL/6 background from 11 litters followed the expected 1:2:1 Mendelian ratio, with 22 WT, 43 Lhx3^W227ter/+, and 21 Lhx3^{W227ter/W227ter} knock-ins generated and surviving to weaning (Table 1).

Table 1.

Lhx3^{W227ter/W227ter} Mice Are Not Viable on the 129/Sv Genetic Background

	C57BL/6	129/Sv
WT	22	18
Lhx3W227ter/+	43	50
Lhx3W227ter/W227ter	21	0
Total	86	68

Pups born to Lhx3^W227ter/+ (heterozygous) parents from generation 6 backcrosses (98.4% pure) of C57BL/6 or 129/Sv strains were weaned and tallied for 11 and 18 litters, respectively. The table lists numbers of weaned (viable) pups of each genotype (WT, heterozygous [Lhx3^W227ter/+], and homozygous [Lhx3^{W227ter/W227ter}]) observed on either a separate C56BL/6 or 129/Sv background. C57BL/6 pups followed an approximately 1:2:1 expected Mendelian ratio, whereas no Lhx3^{W227ter/W227ter} pups survived to weaning from 129/Sv animals.

To further examine the effects of genetic background on the W227ter mutation, we used IHC to detect hormone proteins and PIT1 in 129/Sv and C57BL/6 embryos at E17.5. Overall, hormone expression profiles in the individual strains were similar to those observed during the development of animals on a mixed-strain background (Figures 2 and 4). Addition of IHC experiments detecting FSHβ (as a further marker of gonadotrope cells) revealed comparable expression profiles to LHβ staining (Figure 4). Although GH was detected at a low level in Lhx3^{W227ter/W227ter} animals on either the mixed background or C57BL/6, it was not detected in Lhx3^{W227ter/W227ter} mice on the 129/Sv background (Figures 2 and 4). PRL was not detected in WT 129/Sv animals but was detected in WT controls on the mixed and C57BL/6 backgrounds (Figures 2 and 4). PIT1 transcription factor protein was not detected in Lhx3^{W227ter/W227ter} knock-ins on the 129/Sv background, but it was detected in Lhx3^{W227ter/W227ter} mice on the mixed background or the C57BL/6 background (Figures 3 and 5).

Proteomic analysis of pituitary hormone markers in adult WT and Lhx3^{W227ter/W227ter} animals

To investigate molecular differences between WT and Lhx3^{W227ter/W227ter} knock-in animals, we conducted a quantitative proteomic analysis of the proteins expressed by adult pituitary glands. A total of 2416 unique proteins were identified and quantified in the screen. Differential expression in 425 of these was observed between knock-in and WT animals at P < .00876 (corresponding to q < 0.05). Within this group, 58 proteins were down-regulated in the knock-in, and 367 proteins were up-regulated (Supplemental Table 1, published on The Endocrine Society's Journals Online web site at http://endo.endojournals.org).

Of the AP hormones, PRL was 17.1-fold lower in the Lhx3^{W227ter/W227ter} animals when compared with WT (Figure 6). Pro-opiomelanocortin (POMC, the precursor peptide containing ACTH) and αGSU also were significantly lower in the Lhx3^{W227ter/W227ter} mice, at 3.5-fold and 1.6-fold, respectively (Figure 6). These results correlate with quantitative PCR measurements of POMC (mouse Pomc gene) and αGSU (Cga gene) mRNA levels that showed modest reductions in Pomc and Cga mRNA levels in homozygous pituitary mRNA levels (ref. 24; and (S.C.C. and S.J.R., data not shown). LHβ and TSHβ were significantly higher in the Lhx3^{W227ter/W227ter} mice, at 1.8- and 1.7-fold. GH, the most abundant protein detected, was not significantly different between Lhx3^{W227ter/W227ter} and WT animals at this age (Figure 6). A negative control, ACTA1 (alpha actin) showed no significant change between the two groups. Unfortunately, pituitary transcription factors such as PIT1, ISL1, LHX3, and LHX4 were not detectable by this approach. The whole proteomic data set was further analyzed using Interactive Pathway Analysis software to investigate whether proteins that significantly changed between Lhx3^{W227ter/W227ter} and WT pituitaries were associated with specific cellular functions or pathways. The highest scoring affected network functions were cell cycle, cellular assembly and organization, cellular function and maintenance, molecular transport, and RNA modification and trafficking.

Figure 6.

Proteomic analysis of pituitary hormone markers in Lhx3^{W227ter/W227ter} animals. Pituitaries were harvested from adult female Lhx3^{W227ter/W227ter} or WT animals, and protein extracts were analyzed by label-free quantitative liquid chromatography tandem mass spectroscopy (see Materials and Methods). Fold changes in relative abundances of PRL, POMC (representing ACTH), αGSU (from the mouse Cga gene), LHβ, TSHβ, GH, and ACTA1 (α-actin, as a comparator negative control) proteins/peptides were calculated (Lhx3^{W227ter/W227ter} homozygote/WT). PRL, POMC, and αGSU were significantly lower in Lhx3^{W227ter/W227ter} animals at −17.1-fold, −3.5-fold, and −1.6-fold respectively. (Note that in proteomic method development, the PRL result depicted here was shown as an isolated example for the technique.) (32). GH showed no significant change. LHβ and TSHβ were increased in Lhx3^{W227ter/W227ter} animals at 1.8-fold and 1.7-fold, respectively. The control ACTA1 showed no significant change. Asterisks indicate significance at P < .00876 (q < 0.05) (n = 5 per group).

Discussion

This study reveals that the Lhx3^{W227ter/W227ter} mouse model of CPHD disease has pituitary hormone deficiencies that start during embryonic development (Figures 1, 2, and 4), consistent with an important role for LHX3 in organogenesis of the pituitary gland (6, 7, 34, 35) and observations that the carboxy terminus of the protein (removed due to the mutation) is important for pituitary gene regulation (9, 27). These findings add to the profile of hormone deficiencies also observed in juvenile and adult Lhx3^{W227ter/W227ter} animals (24). Together, these observations establish the Lhx3^{W227ter/W227ter} mouse as a useful model for pituitary hormone deficiency diseases that will facilitate future molecular and cellular studies of disease onset and mechanisms to obtain data that cannot be practically obtained from human patients.

Structural abnormalities were noted in many of the analyzed sections of Lhx3^{W227ter/W227ter} embryonic pituitary glands. A hypoplastic AP region was evident in many cases as early as E15.5 (Figure 1) and in all cases by E17.5 (Figure 2). As seen in juvenile and adult Lhx3^{W227ter/W227ter} mice (24), the posterior pituitary was a similar size in WT and mutant animals, indicating that it is unaffected by loss of LHX3 carboxyl-terminal function. Bifurcated or forked pituitary structures were observed in some Lhx3^{W227ter/W227ter} embryonic pituitaries, but not in every case. Figures 1B and 2–5 show labeled examples of these abnormal structures within the developing AP. Structural deformities including bifurcations have been seen during pituitary development in other mouse models with mutations in pituitary transcription factor and regulatory genes, such as the Aes (36), Prop1 (37, 38), Wnt5a (39), Tcf4 (40), Sox2 (41), and Hesx1 genes (42). It is postulated that such structural defects reflect loss of signaling pathway specificity or abnormalities in signaling protein gradients. These dysmorphologies appear at different time points depending upon the regulatory gene that is impacted, suggesting that multiple pathways may be disrupted to form these unusual structures.

Studies of mice with complete Lhx3 gene ablation or with Lhx3 gene action disrupted by Cre insertion, suggest roles for LHX3 in preventing cell death and cellular proliferation. The developing pituitaries of these animals display reduced proliferation and differentiation and excessive cell death appears to be a major factor in the morphological defects (eg, 6, 7, 35, 43). Lhx3^−/− mice were first analyzed by in situ hybridization of hormone marker mRNAs. These experiments demonstrated almost complete absence of AP hormone transcripts except for a few Pomc-staining corticotrope cells (6). Immunostaining of developing Lhx3^−/− mouse pituitaries also indicates very few corticotrope cells (35). Lhx3^{W227ter/W227ter} mice seem to have less severe structural and differentiation defects during pituitary development than Lhx3^−/− mice. Our observations here of hypoplastic AP lobes and reduced differentiation in developing Lhx3^{W227ter/W227ter} mouse pituitaries are consistent with roles for LHX3 in AP cellular survival, proliferation, and specification. They indicate that the knock-in mutation causing loss of the LHX3 carboxyl terminus compromises these roles but does not overall result in as severe a pituitary phenotype. The carboxyl terminus contains the most potent pituitary activation domain in the LHX3 protein (27), but other regions of the protein have transcription modulatory activities (27, 44, 45) and may be able to maintain some functions. It is also possible that in some genetic environments, compensatory changes in other genes may assist LHX3W227ter function. Lhx4 might be a candidate for such roles, but preliminary experiments measuring Lhx4 mRNA levels in Lhx3^{W227ter/W227ter} pituitaries have not shown any significant changes in Lhx4 gene expression (S.C.C. and S.J.R., data not shown).

PIT1 is an essential pituitary transcription factor located downstream of LHX3 in the transcription factor cascade that is associated with anterior lobe development (reviewed in Refs. 1, 2, and 4). We observe that PIT1 expression is decreased in the Lhx3^{W227ter/W227ter} animals (Figures 3 and 5). Proper expression of PIT1 is crucial to ensure differentiation of somatotropes, lactotropes, and thyrotropes. Mutations in the human PIT1/POU1F1 gene are associated with CPHD diseases featuring loss of GH, PRL, and TSH hormones, a more restricted outcome compared with LHX3-based diseases (1). The observed decreases in PIT1 (Pou1f1 gene) expression in the Lhx3^{W227ter/W227ter} mouse model are consistent with prior suggestions that PIT1 lineage-specific actions are downstream of LHX3 transcription factor action (9). The Snell and Jackson mouse models, which carry mutations in the mouse Pit1 (Pou1f1) gene, are dwarfed and have hypoplastic pituitary glands in which little or no GH, PRL, and TSH are detected (46). Studies of the Snell mouse have shown that in the developing AP, two separate populations of thyrotropes appear at different times. First, a small cluster of cells that are not dependent upon PIT1 appears rostrally, and later, cells that are PIT1 dependent appear more centrally (47). We did not detect cells producing TSHβ in the Lhx3^{W227ter/W227ter} mice at E15.5 or E17.5 (Figures 1, 2, and 4), suggesting that the PIT1-independent population is perhaps affected by the loss of LHX3 function in the homozygous mutants.

Using immunostaining to detect hormone and regulatory factor proteins, we noted that Lhx3^{W227ter/W227ter} 129/Sv embryos had less staining for ACTH than those on the C57BL/6 background and had no detectable GH or PIT1 protein, but we did observe low levels of GH and PIT1 in the C57BL/6 animals (Figures 4 and 5). Consistent with our observations of viability, the 129/Sv background appears to perhaps be more sensitive to loss of LHX3 functions, reflected in the poor development of AP hormone-producing cells. Wild-type 129/Sv animals did not have any PRL or PIT1 staining at E17.5, whereas both were present in the C57BL/6 WT sections. This may indicate that those genes activated slightly later in the 129/Sv strain, or the levels were so low that they were undetectable by IHC.

Our genetic backcross study suggests that Lhx3^{W227ter/W227ter} mice are not viable when they are on the 129/Sv background, but on the C57BL/6 or original mixed backgrounds, they are born and weaned in the expected Mendelian ratios. We chose weaning as the time point of measurement for this study, but only one live pup genotyped as an Lhx3^{W227ter/W227ter} mouse was born on the 129/Sv background. This animal died shortly before the weaning date of unknown causes. Additionally, three stillborn 129/Sv knock-in pups were identified. These four homozygous knock-in pups were born to three different 129/Sv breeder pairs. None of these animals had obvious problems such as neural tube closure defects. It is also possible that there may have been additional stillborn pups that were cannibalized by parents. Consistent with this idea, analysis of embryo frequencies at E17.5 suggests that genotype ratios are roughly as predicted at that time. Loss of viability has been seen before in mouse models of pituitary disease. In the case of Prop1 gene knockout mice, those on a 129S1 background were less viable when compared with animals on a mixed background or other strains such as C56BL/6 (48). Differences in phenotypic outcomes in different strains also have been reported involving mouse models of other diseases, such as kidney diseases and stress-related disorders, including noted differences between 129/Sv-based and C57BL/6 animals (49–51). Some strain differences are specific to treatments, but in some cases, anatomical or physiological variations in pituitary biology are observed between untreated WT animals. For example, WT 129/Sv animals have a larger intermediate lobe than C57BL/6 animals (52). Together, our observations in this study are consistent with described variations in human patients with LHX3 gene alterations (15–23), and in other diseases such as cystic fibrosis (53) and Huntington's disease (54), suggesting that the actions of modifier genes in different genetic backgrounds influence disease onset and character and in human patients may sometimes result in nonviable pregnancies.

It is postulated that the Lhx3 whole gene knockout mice die due to nervous system deficits (5–7). By contrast, the viability of Lhx3^{W227ter/W227ter} animals in mixed (24) and C57BL/6 genetic backgrounds (this study) suggests that LHX3 functions in the pituitary and nervous systems can be separated, thereby impacting viability. In the nervous system, LHX3 synergizes with partner basic helix-loop-helix class transcription factors and the LIM-homeodomain factor ISL1 to specify motor neuron subtypes during development (55). It is possible, for example, that there is a threshold environment for effective function of the truncated W227ter LHX3 protein in the nervous system and that the levels of partner proteins (ie, the actions of modifier genes) in 129/Sv animals are below the levels required to permit appropriate nervous system development and viability. Other models might involve higher levels of proteins with possible compensatory functions (such as LHX4) in the mixed and C57BL/6 backgrounds.

The proteomic analysis of pituitary hormone proteins is consistent with previous initial characterization of juvenile and adult Lhx3^{W227ter/W227ter} animals using other approaches (24), except in the detected levels of the TSHβ and LHβ glycoprotein hormone subunits. The observed GH levels are as expected because, although the Lhx3^{W227ter/W227ter} dwarf mice have hypoplastic pituitaries, they do express GH (24). Our proteomic results indicate that Lhx3^{W227ter/W227ter} adult females have slightly increased levels of TSHβ and LHβ proteins, whereas IHC using anti-β-subunit antibodies indicates a reduced number of cells producing those markers in adults (24) and in development (this study). Intriguingly, quantitative RT-PCR experiments also show increases in TSHβ-encoding (mouse Tshb gene) and LHβ subunit (Lhb) mRNAs in adult animals (S.C.C. and S.J.R., unpublished). This discrepancy may have several explanations. One possibility may be sex differences; all 10 of the animals used in the proteomic analysis here were females, whereas the IHC data previously reported (24) was for a combination of males and females. Regulation of LH is a complex process, occurring at multiple levels of the hypothalamic-pituitary-gonadal axis. Estrogen affects the synthesis of the LHβ subunit at the hypothalamic level through its effect on GnRH, and there are reports of direct action in gonadotropes (reviewed in Ref. 56). Although the Lhx3^{W227ter/W227ter} females are infertile, evidence suggests that they have mostly normal ovarian function (24). Differences in estrogen levels of the females used may have contributed to the slight increase seen in LHβ during the comparison in our proteomic analysis. TSH regulation also is complex, occurring at the level of the hypothalamus and pituitary (reviewed in Ref. 57). Thyroid hormone has been shown to regulate transcription of the α- and β-subunits at different levels (58). It is therefore possible that complex feedback mechanisms may have contributed to the modestly increased TSHβ (and LHβ) subunits detected in the Lhx3^{W227ter/W227ter} female adults with the complete hormone action still impaired.

The proteomic analysis indicated that affected protein networks included cell cycle, cellular maintenance, and protein translation networks. Intriguingly, some proteins with functions in aspects of the cell cycle (integrin β1, metadherin, small G protein signaling modulator 1, and spectrin α) and elongation of actin filaments (profilin 1 and vasodilator-stimulated phosphoprotein) were up-regulated (Supplemental Table 1). For protein translation components, there were several up-regulated translation initiation-related proteins along with a number of ribosomal proteins, including eukaryotic translation initiation factors 2S, 3H, 3M, and 4G1, and 18 ribosomal proteins including S18, L7, L27, S11, L21, L13A, S24, and S5, among others. For cellular assembly and organization proteins, there was up-regulation of proteins involved in microtubule dynamics and cytoskeletal organization such as β-tubulin, structure-specific recognition protein 1, N-myc downstream regulated 1, major prion protein, dihydropyrimidinase-like 3, hepatoma-derived growth factor, and catalytic cAMP-dependent protein kinase α. For apoptosis-related proteins, overall cell death-related processes were decreased and viability and survival were increased in the dwarf. Seventeen proteins associated with apoptosis were all down-regulated: semaphorin 3B, PRL, polycystic kidney disease 1, peptidyl-tRNA hydrolase 2, and nucleolar and coiled-body phosphoprotein 1 among others. Some apoptosis-related proteins were up-regulated: major prion protein, low-density lipoprotein receptor-related protein-associated protein 1, and glycogen synthase kinase 3β. The observation that many apoptosis-related proteins are down-regulated is interesting considering that apoptosis in increased in Lhx3^−/− mouse developing pituitaries (34, 35) and that developing Lhx3^{W227ter/W227ter} pituitaries are hypoplastic. However, it is important to note that the proteomic studies were done on the pituitaries of adult mice; a similar analysis with developing pituitary tissue (although technically challenging to perform) might reveal a different physiological status for developing normal and Lhx3^{W227ter/W227ter} dwarf pituitary tissue.

Lhx3^{W227ter/W227ter} animals have striking deficiencies in PRL hormone (Figures 1, 2, 4, and 6) (24). In these experiments, we used an anti-PRL antibody that is optimized for developmental studies and is different from the one used by Colvin et al (24) to evaluate the adult and P1 animals but gives consistent data. Our results demonstrate that PRL deficiency begins as early as E15.5, indicating that PRL loss is likely a failure to establish PRL production rather than a later onset of deficiency. POMC is also decreased during development (Figures 1, 2, and 5), consistent with observed low levels in P1 and adult Lhx3^{W227ter/W227ter} animals and reduced Pomc mRNA transcripts in adult mice (24). Supplemental Table 1 provides a complete listing and quantitation of all detected pituitary proteins from the proteomic analysis of the mutant and WT animals. This data set provides a resource for others investigating pituitary physiology and may serve to suggest biomarkers for future studies.