The Human Growth Hormone Gene Cluster Locus Control Region Supports Position-independent Pituitary- and Placenta-specific Expression in the Transgenic Mouse*

The human growth hormone (hGH) cluster contains five genes. The hGH-N gene is predominantly expressed in pituitary somatotropes, whereas the remaining four genes, the chorionic somatomammotropin genes (hCS-L, hCS-A, and hCS-B) and hGH-V, are expressed selectively in the placenta. In contrast, the mouse genome contains a single pituitary-specific GH gene and lacks any GH-related CS genes. Activation of the hGH transgene in the mouse is dependent on its linkage to a previously described locus control region (LCR) located 2 15 to 2 32 kilobases upstream of the hGH cluster. The sporadic, nonreproducible expression of hCS transgenes lacking the LCR suggests that they may be dependent on hGH LCR activity as well. To determine whether the hCS genes could be expressed with appropriate placental specificity, a series of five transgenic mouse lines carrying an 87-kilobase human genomic insert encompassing the majority of the hGH gene cluster and the entire contiguous LCR was established. All of the hGH cluster genes were appropriately expressed in each of these lines. High level expression of hGH was restricted to the pituitary and hCS to the labyrinthine layer of the placenta. The expression of the GH cluster genes in their respective tissues paralleled transgene copy numbers irrespective of the transgene insertion site in the host mouse genome. These studies have extended the utility of the transgenic mouse model for the analysis of the full spectrum of hGH gene cluster activation. Further, they support a role for the hGH LCR in placental hCS, as well as pituitary hGH gene activation, and expression. The between hGH-N and hGH-V but do not anneal to hCS-A or hCS-L. After RT with hGH-N 3 9 and 3 cycles of cold PCR with hGH-N/V5 9 A and hGH-N 3 9 , [ g - 32 P]ATP-labeled hGH-N/V5 9 B was added before the final 27 cycles of PCR. PCR was carried out with 2.5 units of Taq DNA polymerase at 1 m M Mg 2 1 . The temperatures for annealing, elongation, and denaturation were 55, 72, and 94 °C, respec- tively. The 5 9 -end-labeled PCR products were digested with Taq I; a specific 38-bp hGH-N end-labeled cDNA fragment and a specific 120-bp hGH-V fragment were expected. The samples were analyzed on 8% denaturing polyacrylamide gels. The RT-PCR analyses for comparing hCS-A and hGH-N expression in the placenta were performed by using the hGHNCSA II primer set followed by a nested amplification between hGHNCSA II 5 9 and 32 P-labeled hGHNCSA II 3 9 B. The 3 9 -end-labeled products were cut with Hha I and fractionated on 6% polyacrylamide gels to generate 141-bp hGH-N- and 130-bp hCS-A-specific fragments. Intensities of these two bands were compared by phosphorimager analysis. of in transgenic may functionally parallel the site of hCS-A expression in placenta and is clearly distinct from the sites of expression of the prolactin gene family in mouse placenta.

The human growth hormone (hGH) cluster contains five genes. The hGH-N gene is predominantly expressed in pituitary somatotropes, whereas the remaining four genes, the chorionic somatomammotropin genes (hCS-L, hCS-A, and hCS-B) and hGH-V, are expressed selectively in the placenta. In contrast, the mouse genome contains a single pituitary-specific GH gene and lacks any GHrelated CS genes. Activation of the hGH transgene in the mouse is dependent on its linkage to a previously described locus control region (LCR) located ؊15 to ؊32 kilobases upstream of the hGH cluster. The sporadic, nonreproducible expression of hCS transgenes lacking the LCR suggests that they may be dependent on hGH LCR activity as well. To determine whether the hCS genes could be expressed with appropriate placental specificity, a series of five transgenic mouse lines carrying an 87-kilobase human genomic insert encompassing the majority of the hGH gene cluster and the entire contiguous LCR was established. All of the hGH cluster genes were appropriately expressed in each of these lines. High level expression of hGH was restricted to the pituitary and hCS to the labyrinthine layer of the placenta. The expression of the GH cluster genes in their respective tissues paralleled transgene copy numbers irrespective of the transgene insertion site in the host mouse genome. These studies have extended the utility of the transgenic mouse model for the analysis of the full spectrum of hGH gene cluster activation. Further, they support a role for the hGH LCR in placental hCS, as well as pituitary hGH gene activation, and expression.
The human growth hormone (hGH) 1 gene family is encoded within a 48-kilobase (kb) cluster on chromosome 17q22-24. It contains five genes that share greater than 94% sequence identity (1,2). hGH-N is selectively expressed in the somatotrope and lactosomatotrope cells of the anterior pituitary, whereas the other four genes, hCS-L, hCS-A, hGH-V, and hCS-B, are expressed in the syncytiotrophoblastic layer of the mid-to late gestational placenta (3)(4)(5). Among the placental genes, hCS-A and hCS-B are highly expressed, whereas hCS-L and hGH-V are expressed at trace levels (6). The mutually exclusive tissue specificities and distinctive developmental controls of the closely spaced and structurally related hGH and hCS genes make the hGH gene cluster a highly informative model system for the study of gene expression in mammalian development.
Activation and expression of genes in their native chromatin context reflect the combined actions of multiple determinants. The most well characterized cis-acting determinants of gene expression are those located in close proximity to the respective promoter. There are a number of well defined and functionally characterized promoter proximal determinants critical to the expression of hGH-N. Among these are a pair of binding sites for the tissue-restricted POU-homeodomain protein, Pit-1/ GHF-1 (7), and a binding site for the more widely expressed zinc-finger protein factor (Zn-15) (8). In the case of the hGH-N gene these proximal promoter elements, necessary for gene expression in cell transfection assays, are insufficient to mediate in vivo, tissue-specific expression in transgenic mouse lines (9 -11). This observation points to the existence of determinants located at a greater distance from the hGH gene that are essential to the establishment of a transcriptionally active chromatin domain. A set of such determinants, located from 14.5 to 32 kb 5Ј to the hGH-N gene promoter, were identified by DNaseI hypersensitivity (HS) mapping of chromatin from expressing tissues. The closely spaced HSI and HSII are specific to pituitary chromatin, HSIII and HSV are shared by pituitary and placental chromatin, and HSIV is specific to placental chromatin. When linked to the hGH-N gene, the full set of determinants directs high level, position-independent, somatotrope-specific expression of the hGH-N gene in transgenic mice in a consistent and predictable manner (10,11). The ability of these determinants to overcome site of integration effects on the hGH-N gene fulfills the functional criteria for a locus control region (LCR). These criteria were initially defined for the human ␤-globin LCR (12)(13)(14)(15) and subsequently for a small group of additional gene systems including the chicken lysozyme LCR (16,17) and the LCR for the human red and green visual pigment genes (18,19). A similar dependence of the placental genes in the hGH gene cluster on LCR activation has not been demonstrated.
The hCS-A transgene, with extensive 5Ј-and 3Ј-flanking sequences, is either not expressed at all or is expressed at low and unpredictable levels in the mouse placenta (10). These observations might reflect two alternative, or coexisting, mechanisms: 1) the mouse, which lacks CS-like genes, is unable to support hCS transgene expression in the placenta because of a species-specific array of placental transcription factors poorly suited for hCS gene activation or 2) the consistent activation of the hCS transgene is dependent on linkage to the hGH LCR. This second possibility is supported by our previous observation that a specific subset of hGH LCR HS form in the chro-matin of primary human placental syncytiotrophoblast nuclei (10). In the present report we have isolated a human genomic clone that contains the entire LCR as well as much of the contiguous gene cluster encompassing both hGH-N and hCS-A. By introducing this 87-kb transgene into the mouse genome we are able to document the competence of the mouse transgenic model to reliably support appropriate expression of the hGH gene cluster in both the pituitary and the placenta. These data support the role of the hGH LCR in activation of both hGH and hCS gene expression.
The hGH-N genomic probe was a 1.37-kb SmaI fragment (coordinates Ϫ494 to 876 bp, relative to the hGH-N transcription initiation site) (10); the HSI and -II genomic probe was a 1.6-kb BglII fragment (coordinates Ϫ14.56 to Ϫ16.16 kb) (11); and the HSIV probe was a 2.38-kb SmaI fragment (coordinates Ϫ31.23 to Ϫ28.85; GenBank TM Accession number AC005803). The MX probe, which detects the unique sequence 3Ј-flanking region of the m-globin gene, was released as a 1.3-kb BamHI fragment from pMX plasmid (20). The mouse rPL32 cDNA probe was released as a 0.32-kb EcoRI and HindIII fragment from the rpL32 plasmid (21). The P1 vector 3Ј-end probe was a 0.23-kb BglI segment corresponding to position 15.25-15.65 kb in the P1 vector. All oligos used as hybridization probes or as primers were synthesized by the Nucleic Acid and Protein Research Core Facility of Children's Hospital of Philadelphia or by Life Technologies, Inc. All primer sequences are listed in Table I.

Screening the Human Genomic P1 Library and Mapping the P1 Clone
A human genomic P1 library (Genome Systems, PAD10SacBII vector) was screened with a primer pair (CSA(E) primers, Table I) corresponding to the hCS-A gene 3Ј-enhancer and a second primer pair (HSV primers) corresponding HSV (Table I). A doubly positive clone, P1 6057, was obtained. P1 6057 was mapped by Southern blot analysis of complete restriction enzyme digestions (see "Results") and by indirect end labeling of partial digestions. In the latter case, the DNA was linearized with NotI and subsequently partially digested with ClaI or EcoRI. Digests were electrophoresed on pulsed field gels (FIGE Mapper Electro-phoresis System), DNA was transferred to Zetabind, and the membranes were incubated with a 32 P-labeled oligomer corresponding to the T7 site in the P1 vector or to a probe corresponding to the 3Ј-end of linearized P1 6057. This mapping served to screen for any deletions or rearrangements in the insert compared with the sequence (GenBank TM AC005803) or the known restriction map of the LCR and gene cluster (1,10).

Generation and Analysis of P1 Transgenic Mice
P1 6057 plasmid DNA was linearized at the NotI site ( Fig. 1), purified by Elutip (Schleicher & Schuell), adjusted to 2 ng/l in 10 mM Tris-HCl, pH 7.6, 0.1 mM EDTA, and microinjected into the male pronucleus of fertilized mouse eggs (University of Pennsylvania Transgenic and Chimeric Mouse Core). Positive founders were identified by dot blot analysis of tail DNA using the hGH-N probe. The transgene was mapped in F1 progeny, and the transgene copy number for each line was determined by Southern blot analysis (described in "Results").

Reverse Transcription-Polymerase Chain Reaction (RT-PCR)
hGH-N expression in the pituitary was documented and quantified relative to endogenous mGH mRNA by the RT-PCR approach generating an hGH-N to mGH mRNA ratio as described previously (10). Briefly, reverse transcription and subsequent PCR were carried out beginning with 0.2-0.5 g of total RNA using primers corresponding to regions perfectly conserved between mGH and hGH (mhGH primer set, Table I). After 30 cycles of PCR the 3Ј-end-labeled cDNA products were digested with BstNI to generate fragments specific to hGH-N and mGH mRNAs. Intensities of the two bands were compared by phosphorimager analyses.
The RT-PCR analysis of hCS-A expression was performed with the hGHNCSA I primer set (Table I). These primers annealed to both hCS-A and hGH-N mRNAs. The 5Ј-end-labeled cDNA products were cut with BamHI to generate a 104-bp end-labeled hCS-A cDNA band.
The nested RT-PCR analysis for distinguishing hGH-N and hGH-V mRNAs was performed by using primers hGH-N/V5Ј A and hGH-N 3Ј and a nested [␥-32 P]ATP-labeled hGH-N/V5Ј B. These primers correspond to the regions conserved between hGH-N and hGH-V but do not anneal to hCS-A or hCS-L. After RT with hGH-N 3Ј and 3 cycles of cold PCR with hGH-N/V5Ј A and hGH-N 3Ј, [␥-32 P]ATP-labeled hGH-N/V5Ј B was added before the final 27 cycles of PCR. PCR was carried out with 2.5 units of Taq DNA polymerase at 1 mM Mg 2ϩ . The temperatures for annealing, elongation, and denaturation were 55, 72, and 94°C, respectively. The 5Ј-end-labeled PCR products were digested with TaqI; a specific 38-bp hGH-N end-labeled cDNA fragment and a specific 120-bp hGH-V fragment were expected. The samples were analyzed on 8% denaturing polyacrylamide gels.
The RT-PCR analyses for comparing hCS-A and hGH-N expression in the placenta were performed by using the hGHNCSA II primer set followed by a nested amplification between hGHNCSA II 5Ј and 32 Plabeled hGHNCSA II 3Ј B. The 3Ј-end-labeled products were cut with HhaI and fractionated on 6% polyacrylamide gels to generate 141-bp hGH-N-and 130-bp hCS-A-specific fragments. Intensities of these two bands were compared by phosphorimager analysis.

Transgenic Expression of the hGH Gene Cluster
In Situ Hybridization Generation of hCS-A cDNA-A pair of primers (hcs cDNA 5Ј and hcs cDNA 3Ј; Table I) were used to amplify a 614-bp hCS-A cDNA fragment extending from coordinates 114 to 738 of the hCS-A cDNA clone (pG12F extends from 102 to 654 nucleotides of hCS-A cDNA plus the 3Ј-UTR and 77 base poly(A) tail (6)). PCR products were purified by recovery from the gel and ligation into pGEM-T easy vector (Promega).
Generation of Riboprobes-The subcloned hCS-A cDNA plasmid (641 pGEM-T-A) was linearized with SpeI (for T7 transcription) or NcoI (for Sp6 transcription). The linearized plasmids were transcribed in vitro using MEGAshortscript T7 Kit (for the antisense probe) or Sp6 Kit (for sense probe), in the presence of [␣-33 P]UTP.
In Situ Hybridization-Placentas were collected from mouse embryos generated by crossing an hGH P1 line 813I male with a CD-1 female. The mother was euthanized at embryonic day 16 and the placentas were fixed in 4% paraformaldehyde in PBS (pH 7.2) at 4°C overnight. Genotypes were determined by dot blot analyses of embryonic tissues. Placentas of the transgenic and wild-type mice were embedded in paraffin, and 5-m sections were prepared and treated for in situ hybridization as described previously with minor modifications (22). Briefly, wax was removed, and the sections were rehydrated, fixed with 4% paraformaldehyde in PBS (pH 7.2) for 20 min, and washed twice with PBS for 5 min each. The sections were treated with 20 g/ml proteinase K in 50 mM Tris-HCl, pH 8.0, 5 mM EDTA for 7.5 min, washed with PBS, and fixed again in 4% paraformaldehyde for 5 min. After washing with PBS, the slides were treated in 0.1 M triethanolamine for 5 min followed by incubation in 0.25% acetic anhydride in 0.1 M triethanolamine for 5 min. Following washing with 2ϫ SSC, the slides were incubated overnight with 4 ϫ 10 6 cpm of probe in 100 l of hybridization buffer (40% formamide, 0.48 M NaCl, 8 mM Tris-HCl, pH 7.5, 1.6 mM EDTA, 0.8ϫ Denhardt's solution, 0.4 mg/ml yeast tRNA, 80 g/ml poly(A) ϩ RNA, and 10% dextran sulfate) overnight at 65°C. The slides were washed with 5ϫ SSC for 10 min at 50°C, then in 50% formamide, 2ϫ SSC for 20 min at 65°C, next in RNase buffer (0.3 M NaCl, 10 mM Tris-HCl pH 8.0, 5 mM EDTA) for 30 min at 37°C, and then digested with RNase A (50 g/ml) in the same buffer for 30 min at 37°C. Finally the slides were washed with RNase buffer, 50% formamide in 2ϫ SSC, 2ϫ SSC alone, and 0.1ϫ SSC. The slides were dehydrated and exposed to emulsion for about 2 weeks and then developed and stained with hematoxylin and eosin. Bright field and dark field micrographs were generated using a Nikon Microphot-FX microscope.

Isolation and Mapping of a P1 Clone Encompassing the hGH
Locus and Its Contiguous LCR-A genomic fragment containing the entire hGH LCR and the majority of the contiguous hGH gene cluster was isolated from a human genomic P1 library. The library was screened with amplimers specific for the region immediately 5Ј to HSV of the LCR and a second primer set specific for a site 3 kb 3Ј to the hCS-A gene. A clone positive with both probes, P1 6057, was isolated and named "hGH/P1." The restriction map of this clone was compared with the known genomic structure (1, 10). The hGH-N probe, which cross-hybridizes to each of the genes in the cluster, identified unique 10.5 and 7.8 kb BglII bands corresponding to hCS-A and hCS-L and a more intense 2.6-kb band containing the two identically sized fragments with the hGH-N and hGH-V genes, respectively (Fig. 1B, right). Missing was the most 3Ј-gene in the cluster, hCS-B, which would have been present on a 3-kb BglII fragment. The HSI and -II probe detected the expected 1.6-kb BglII fragment (Fig. 1B, middle), and the HSIV probe hybridized to the expected 22 kb EcoRI band encompassing HSIII, HSIV, and HSV (Fig. 1B, left). Additional mapping of the hGH/P1 insert DNA by complete and partial digestions failed to reveal any rearrangements or deletions when compared with the genomic structure (data not shown). The orientation of the insert relative to the vector (Fig. 1C) was established by demonstrating that the vector-derived T7 polymerase segment and the HSIV site were located on the same ClaI restriction fragment (data not shown). Thus the 87-kb P1 6057 (hGH/P1) insert spanned coordinates Ϫ45.8 to 40.6 kb of the hGH locus (hGH-N transcription start site as coordinate 1) and contained all five HS sites of the LCR and the contiguous hGH-N, hCS-L, hCS-A, and hGH-V genes (Fig. 1, A and C).
Generation of hGH/P1 Transgenic Mouse Lines-The hGH/P1 plasmid was mapped for rare cutting endonucleases sites to identify a unique site for linearization prior to zygotic microinjection. NotI, SalI, and SfiI were tested because each cleaves a single site adjacent to the P1 vector cloning site (23). The banding pattern of the restriction digestions on 1% agarose pulsed field gels demonstrated that the insert contained two SfiI sites and one SalI site but lacked a NotI site (data not  Table I)  shown). NotI was therefore used for linearization (Fig. 1C). The size of the NotI-linearized plasmid was consistent with the 87-kb genomic insert size (see above). The linearized hGH/P1 plasmid was microninjected into fertilized mouse oocytes, and five transgenic founders were obtained: 809C, 809F, 811B, 811D, and 813I. Each founder was crossed with a CD1 mate to generate an F1 transgenic. The integrity and copy number of the transgene in each line was determined by Southern blot of F1 tail DNA. Membranes were sequentially hybridized with probes for hGH-N, HSI, -II, and HSIV. Fig. 2 shows a representative Southern blot using the 1.37-kb hGH-N probe. In all cases, the bands predicted by the map (Fig. 1) were obtained. The transgene copy number was estimated for each line by normalizing the transgene signals to a loading control, the mouse -globin probe (m-MX), and to human genomic DNA. The transgene copy numbers determined in this way ranged from 2 to 19 (Fig. 2).
Expression of the hGH/P1 Transgene in the Mouse Pituitary-In previous studies we documented that the hGH-N gene is appropriately expressed in the pituitaries of transgenic mice when linked to its contiguous LCR (10). To confirm and extend this observation, expression of the hGH cluster containing the hGH-N gene as well as the linked and placentally expressed hCS-L, hCS-A, and hGH-V genes was analyzed. RNA was isolated from the pituitaries of adult F1 transgenic mice from each of the five hGH/P1 lines. These samples were analyzed for hGH-N, hCS, and hGH-V mRNAs. hGH-N mRNA was identified and quantified relative to endogenous mGH mRNA using an RT-PCR co-amplification assay (10) (Fig. 3A). This approach co-amplifies hGH and mGH as identically sized fragments that can be subsequently differentiated and quantified by restriction analysis (the primers do not amplify hGH-V or hCS). The analyses of all five transgenic mouse lines revealed the ex-pected three cDNA fragments; two (170 and 125 bp) corresponding to the two major splice products of the hGH transcript and the third (110 bp) corresponding to mGH cDNA. The ratios of the hGH to mGH signals were quantified and normalized to transgene copy number. The pituitary expression from the hGH-N gene was consistent from line to line, varying less than 3-fold (0.34 -0.96) (Fig. 3A). This range of hGH-N transgene expression was remarkably similar to that observed previously for the LCR linked to the isolated hGH-N gene (10). This result demonstrated reproducible, copy number-dependent expression of the hGH-N gene from the 87-kb hGH P1 transgene.
The genes in the hGH gene cluster are highly similar in structure and yet are expressed in mutually exclusive tissue distributions, predominantly in the pituitary or the placenta. To determine whether this tissue specificity of expression from the hGH gene cluster is maintained in the mouse model, the expression of hCS and hGH-V was tested in the pituitary. RT-PCR was performed on transgenic pituitary RNA using a pair of primers that were complementary to both hGH-N and hCS-A mRNAs; the co-amplified cDNAs were then distinguished by restriction digestions (Fig. 3B). In contrast to the strong hGH-N signals in each sample, there was a complete absence of hCS-A expression in all lines. The total lack of hCS-A expression was confirmed using a technically independent hCS-specific nested RT-PCR assay (data not shown). A similar RT-PCR approach designed to detect ectopic expression of the placental hGH-V revealed total absence of expression in transgenic mouse pituitaries from each of the five transgenic lines (Fig. 3C). Expression of the hCS-L gene was not tested, because the function of this gene has been lost during the evolution of the cluster via multiple alternative splicing errors (24). The high levels of hGH-N and absence of detectable expression of hGH-V or hCS-A indicated that the genes in the hGH gene cluster were expressed with appropriate tissue specificity in the pituitaries of all five of the hGH/P1 lines.
Phenotype of the hGH/P1 Mouse Lines-All five of the hGH/P1 transgenic mouse lines displayed normal fertility, fecundity, and lifespan. Males and females from each line were weighed at intervals starting at 3 weeks of age. Although there was minor variations in body weight from line to line, only line 813I displayed a growth curve that differed significantly from sex-matched wild-type littermates. The 813I transgenic mice (both males and females) were approximately 10% larger than the matched controls after 7 weeks of age. This difference did not increase with time. The normal body growth rate of all the remaining groups indicated that hGH expression from the P1 transgene was under normal physiologic control and that no significant ectopic and/or unregulated transgene expression was occurring in other tissues.
Transgene Expression in the Placenta-In a previous study, we had generated six mouse lines transgenic for a 15-kb Hin-dIII fragment encompassing the hCS-A gene along with 5.4 kb of contiguous 5Ј-flanking region and 7.2 kb of contiguous 3Јflanking region (10). When assessed for expression in the placenta, only two of these lines expressed hCS mRNA at clearly detectable levels, two at trace levels, and two not at all (10). This lack of consistent transgene expression suggested that the transgene was subject to site of integration effects. To test the hypothesis that hGH LCR element(s) might be necessary for consistent locus activation in the placenta, gene expression was assessed in the placentas of fetuses from each of the five lines carrying the hGH/P1 transgene. F1 transgenic mice from each of the lines were crossed with wild-type mates to generate embryonic day 18 and 19 fetuses. The genotype of each embryo was determined by DNA dot blotting, and the corresponding placentas were assessed for hCS-A transgene expression by an RT-PCR restriction endonuclease assay as well as by Northern blot analyses (Fig. 4, A and B, respectively). By RT-PCR, all five lines were positive for placental expression of hCS-A mRNA. Of note, low levels of hGH-N mRNA were also detected in all samples, but in each case these levels were approximately 1% that of the hCS-A mRNA. The levels of hCS-A expression in the placenta were quantified for each of the five lines by Northern blot analysis. An hCS-A mRNA band was detected in placental mRNA from each of the five lines (Fig. 4B). The signal intensity quantified by phosphorimager was normalized to a loading control signal and then divided by the transgene copy number. These normalized hCS-A expression values were remarkably consistent among four of the mouse lines (2-fold range). A fifth line, 809C line, expressed hCS at a 10-fold higher level. Thus the hCS-A gene in the hGH/P1 transgene was expressed in all lines irrespective of its site of integration, and with the exception of line 809C, the levels of expression were closely related to transgene copy number.
The Ectopic Expression of the Transgenes in the Different Tissues-The above data demonstrated that hGH-N was selectively expressed in the pituitary and hCS-A in the placentas of hGH/P1 mice. The possibility of ectopic expression of the GH cluster outside of the pituitary or placenta was next investigated. By Northern blotting a strong hGH mRNA signal was detected in the pituitary and a weaker hCS mRNA signal in the placenta. In contrast, no signals were obtained in any of multiple other tissues surveyed (a single exception was a very weak signal in the testis of line 813I, the highest transgene copy line) (data not shown). To assess ectopic expression at a higher level of sensitivity, we carried out RT-PCR analyses with three sets of primers specific for hGH-N, hCS-A, or ␤-actin. A representative study using line 809C is shown in Fig. 5. Strong signals were seen for hGH-N in the pituitary. Low level expression of hGH-N was detected in the placenta as observed previously. Similarly low levels were detected in the brain in all lines and in the testis, ovary, and spleen in a subset of lines. A strong hCS-A signal was generated from placental RNA and very low levels of expression were observed in the brain, testis, and ovary of several lines. These data support specific and mutually exclusive tissue specificities for the hGH and hCS loci in the context of the hGH/P1 transgene.
Sublocalization of hCS Transgene Expression within the hGH/P1 Mouse Placenta-In prior studies we have demonstrated by immunohistochemical staining the co-localization of hGH and mGH expression in pituitary somatotrope cells of transgenic lines carrying the hGH-N gene linked to the LCR (11). A similar study was carried out to identify the site of hCS expression in the placenta of hGH/P1 transgenic mice. The mouse placenta is composed of a fetal portion and a maternal portion. The fetal portion can be divided three regions: labyrinth, spongiotrophoblast, and trophoblast giant cells (Figs. 6,  A and B). In situ histohybridization with 33 P-labeled antisense RNA probes demonstrated that hCS-A mRNA was confined to the labyrinth of P1 transgenic mouse placentas. Control studies with the 33 P-labeled sense probes and stains of wild type placentas were all negative (Figs. 6, C and D). By high power FIG. 3. Selective, copy number-dependent expression of the hGH-N locus from the hGH/P1 transgene in mouse pituitaries. The framed diagram to the right of each autoradiograph indicates the RT-PCR approach used to detect, specify, and quantify the respective mRNAs. In each case the end-labeled (*) PCR product was digested with the indicated restriction enzyme to differentiate between two co-amplified mRNA species. A, relative expression of hGH-N and mGH. hGH and mGH mRNAs were co-amplified from hGH/P1 pituitary RNA samples using the mhGH primer set (Table I). Digestion of the end-labeled cDNAs with BstNI generated a strong 170-bp hGH-N cDNA band, a weak 125-bp alternatively spliced hGH-N band, and a 110-bp mGH cDNA band in the pituitary sample from each line. Wild-type mouse pituitary showed only the 110-bp mGH cDNA band, and mRNA from the pituitary of the 149C mouse line (expressing an isolated hGH-N transgene (10)) was used as positive control for hGH mRNA. The levels of hGH-N mRNA normalized to mGH mRNA and to transgene copy numbers (Fig. 2) are indicated below the respective lanes. This value represents the level of expression from a single transgene copy as a percentage of a single endogenous mGH gene. B, selective expression of hGH-N but not hCS-A in the hGH/P1 transgenic pituitary. The pituitary RNA samples were as in A with the exception that RNA isolated from the placenta of the 809C line (Fig. 4) was used as a positive control for hCS-A RNA. hGH-N and hCS-A RNAs were co-amplified with the hGHNCSA I primer set (Table I) and were distinguished by digestion with BanI (as shown to the right). C, selective expression of hGH-N but not hGH-V in the hGH/P1 transgenic pituitary. hGH-N and hGH-V RNAs were co-amplified using the hGHN3Ј and hGHN/V5ЈA primers and then with a second, nested and end-labeled 5Ј-hGHN/V5ЈB primer (Table I). The respective cDNAs were distinguished by digestion with TaqI (as shown to the right). The pituitary samples were as in A with the exception that human placentas were used as positive controls for the detection of hGH-V. magnification, the positive cells in the labyrinth were mononuclear, of moderate size, elongated, and usually located adjacent to the maternal blood sinuses (Figs. 6, E and F). There was some heterogeneity of expression levels among positive cells. This was best seen by variation of signal strength on dark field analyses. The morphology, location, and distribution of the positive cells all suggested that they represented mononuclear labyrinthine trophoblast cells. DISCUSSION Previous studies have demonstrated that expression of the hGH-N transgene is dependent on remote regulatory elements. These elements were originally identified as a set of DNaseI HS in pituitary chromatin located from 14.5 to 32 kb 5Ј to the hGH gene cluster. Subsequent functional analyses demonstrated that these elements were able to overcome site of integration position effects and establish autonomously regulated chromatin domains in mouse pituitary chromatin. The activated hGH-N transgene was expressed at consistent and uniform levels that were comparable to that of endogenous mouse GH. Based on their ability to impart copy number dependence on the hGN-N transgene, these elements fulfilled the criteria for LCR action (14,25).
Genes in a cluster can compete with each other for LCR elements, and such competition can constitute an essential aspect of their normal developmental control. This is well illustrated in the human ␤-globin gene cluster in which the normal developmental silencing of the fetal ␥-globin gene is dependent on the presence of the adult-specific ␤-globin gene in cis (reviewed in Ref. 24) (25). Thus an essential step in the characterization of expression patterns of genes and establishing their mechanisms of control is to determine their expression profiles when they are in the native configuration of their family cluster. In the present studies, we introduced a large human DNA fragment containing four members of the hGH gene family along with an extensive segment of 5Ј-flanking region encompassing all five HS of the LCR into the mouse genome. Analysis of hGH-N expression revealed that five of five hGH/P1 lines expressed hGH-N in the pituitary at levels correlating with transgene copy number. The levels of expression/transgene copy for these lines were tightly grouped and were comparable (34 -96%) to the level of endogenous mGH. This is quite similar to the levels of expression observed when the hGH-N gene with contiguous LCR elements is introduced into the mouse genome in the absence of the rest of the cluster (10). In both cases the transgenic mice grew normally (i.e. no FIG. 4. Selective, copy number-dependent expression of hCS-A mRNA in the placentas of mice carrying the hGH/P1 transgene. A, RT-PCR detection of hCS-A mRNA and trace levels of hGH-N mRNA in hGH/P1 mouse placentas. hCS-A and hGH-N mRNAs were reverse transcribed and co-amplified between the hGHNCSA II 3ЈA and hGHNCSA II 5Ј-primers and then labeled by further amplification with a second, end-labeled and nested 3Ј-primer (hGHNCSA II 3ЈB) ( Table I). The respective cDNA products were distinguished by digestion with HhaI (as diagrammed to the right of the autoradiograph). A strong hCS-A band (130 bp) and a very faint hGH-N band (141 bp) were detected in the placental RNA of each hGH/P1 mouse line. The relative expression of hGH-N to hCS-A is represented as a percentage below the corresponding lanes. Control mRNAs include placental RNA from mouse line 766I carrying an isolated hGH-N transgene with targeted ectopic expression in the placenta (F. Elefant, Y. Su, S. A. Liebhaber, and N. E. Cooke, manuscript in preparation) and line 484E carrying the isolated hCS-A gene previously shown to be expressed in the placenta (10). B, Northern blot analysis demonstrates copy number-dependent expression of the hCS-A gene. Poly(A) ϩ samples, corresponding to sources indicated above each lane, were hybridized with a probe for hCS mRNA and then stripped and rehybridized with a probe for mouse ribosomal protein L32 (mrpL32) mRNA as a loading control. The expression of hCS-A, normalized for RNA loading and for transgene copy number are indicated below the respective lanes. The results showed strong hGH-N and hCS-A signals limited to the pituitary and placenta, respectively, and weak signals in the brain, testes, and ovary. WT, wild type. gigantism), suggesting that there was no significant ectopic, unregulated hGH-N expression and that the pituitary expression was under normal physiologic regulation. Thus comparison of the present report with our prior studies allows us to conclude that the expression of the hGH-N transgene linked to the LCR is insertion site-independent, copy number-dependent, and pituitary-specific whether isolated or in its native gene cluster environment.
The hGH gene cluster contains a set of placentally expressed genes whose expression patterns are distinct and mutually exclusive from the linked and structurally related hGH-N gene. When a 15-kb transgene fragment containing the hCS-A gene with 5.4 kb of 5Ј-flanking sequences and 7.2 kb of 3Ј-flanking sequences were tested for expression in a transgenic assay, the placental expression of hCS-A was sporadic and variable from line to line because of site of integration effects. Of the six lines carrying this transgene, two lines expressed hCS-A at low levels, two at trace levels, and two did not express hCS-A at all (10). This weak and erratic expression profile was reminiscent of hGH-N transgene expression in the absence of the LCR (10). A common dependence on LCR function was further suggested by the observation that HSIII and HSV of the GH LCR HS are present in both placental and pituitary chromatin. Taken together these data suggested that the hCS-A gene, like the linked and closely related hGH-N, was dependent upon LCR function for activation.
In contrast to the erratic expression profile of the hCS-A transgene lacking LCR elements seen in prior studies (10), five of five mouse lines carrying the hGH/P1 transgene in the present study expressed hCS-A in the placenta. Furthermore, the levels of hCS-A mRNA expression/transgene copy number were highly reproducible (2-fold range) for four of the five lines. The remarkably consistent expression of the hCS-A gene in the context of the hGH/P1 transgene supports a role for the remote 5Ј-flanking region LCR elements in establishing a transcriptionally productive domain for the hCS-A gene in the placenta. The consistent activation of the hCS-A gene in the context of the hGH/P1 transgene is surprising from an evolutionary viewpoint. Both the structure and composition of the GH gene family and the structure and function of the placenta have diverged between mouse and man. The multiple evolutionary rounds of gene duplication that have given rise to a five member GH/CS cluster in humans have not occurred in the rodent lineage. Instead, the GH-related but much more divergent prolactin gene (26) has duplicated in the mouse, and most of these multiple prolactin-related genes are expressed in the mouse placenta (27,28). It is not clear whether the function(s) of hCS in primates and these prolactin-related hormones in rodents are similar. There is also, at this time, no evidence to support an LCR linked to the single mGH gene locus. 2  activate the hCS locus in a consistent manner in the mouse placenta must have been conserved despite this divergence between mouse and primate at the GH gene locus.
The structures of the mouse and human placenta present fundamental similarities and important distinctions (29). The placentas in both species belong to the deciduate group in that they both contain fetal and maternal portions. However, the details of their structures and endocrine functions differ significantly. The mouse placenta is labyrinthine, whereas the human placenta is villous (30). The murine labyrinthine and human villous areas of the respective placentas are the sites at which nutrients and gas exchanges occur between fetal and maternal blood. The maze-like labyrinthine layer of the mouse placenta is composed of anastomosing cords or plates of trophoblasts. These trophoblasts are in direct contact with the maternal blood circulating in the maternal blood sinuses. In this respect the murine labyrinthine trophoblasts are analogous to the human syncytiotrophoblasts that line the outer layer of the fetal placental villi and are also in direct contact with the maternal circulation. In the mouse, the labyrinthine trophoblasts form three layers separating the maternal blood from the fetal blood, a single mononuclear cellular layer and two layers of a trophoblast syncytium. The mononuclear cellular layer comes in direct contact with maternal blood (31). In human placentas, hCS-A is expressed in syncytiotrophoblast cells that line the surface of the villi and contact maternal blood. In the hGH/P1 transgenic mouse, hCS-A-positive cells were mononuclear and lined the maternal blood sinuses within the labyrinth. Thus we conclude that the site of hCS-A transgene expression in the trophoblasts cells lining the maternal blood sinuses in the labyrinth of the mouse placenta is functionally analogous to its natural site of expression in the human placenta.
Overlying the murine labyrinth is the spongiotrophoblast layer (Fig. 6A). The outermost cells in this layer are the trophoblast giant cells that form the interface with maternal decidua. In the present studies, in situ hybridization showed that hCS-A expression was limited to the labyrinth of transgenic mouse placenta. In contrast, the mouse prolactin-related genes, PL-1 and PL-2, are expressed in the trophoblast giant cells (27,28). The prolactin-like proteins, proliferin and proliferin-related protein, are expressed in giant cells and/or spongiotrophoblast cells (32)(33)(34)(35)(36)(37). These data indicate that the cellspecific expression of hCS-A in hGH/P1 transgenic mice may functionally parallel the site of hCS-A expression in human placenta and is clearly distinct from the sites of expression of the prolactin gene family in mouse placenta.
An extensive survey of tissue RNA samples from hGH/P1 mice was carried out to identify tissues in addition to pituitary and placenta that express the transgene. Northern blot analysis demonstrated only pituitary and placental expression of the transgenes. However, a more sensitive RT-PCR approach revealed several additional sites. hGH-N was expressed at trace levels in the brain in all hGH/P1 transgenic lines and in testis, ovary, or spleen in a subset of lines. hCS-A expression was detectable at trace levels in brain, testis, ovary, and kidney. Recent studies have reported low levels of GH gene expression in the rat brain (38) and GH and CS gene expression in human testis (39) and ovary (40). These observations suggest that the low level "ectopic" expression from the hGH/P1 transgene that we have observed may instead reflect normal physiologic patterns of low level gene expression.
In conclusion, the present study established a remarkable parallel between the expression of the hGH gene cluster in situ and its expression from a large continuous transgene encom-passing the cluster and the full LCR. The data further extended the function of the LCR to expression of the hCS genes in the placenta and established an in vivo model system in which the component determinants of LCR function in the pituitary and placenta can be further explored.