Pregnancy-associated plasma protein A (PAPPA) is a metzincin superfamily metalloproteinase in the insulin-like growth factor (IGF) system. PAPPA increases IGF bioavailability and mitogenic effectiveness in vitro through regulated cleavage of IGF-binding protein 4 (IGFBP4). To determine its function in vivo, we generated PAPPA-null mice by gene targeting. Mice homozygous for targeted disruption of the PAPPA gene were viable but 60% the size of wild-type littermates at birth. The impact of the mutation was exerted during the early embryonic period prior to organogenesis, resulting in proportional dwarfism. PAPPA, IGF2 and IGFBP4 transcripts co-localized in wild-type embryos, and expression of IGF2 and IGFBP4 mRNA was not altered in PAPPA-deficient embryos. However,IGFBP4 proteolytic activity was completely lacking in fibroblasts derived from PAPPA-deficient embryos, and IGFBP4 effectively inhibited IGF-stimulated mitogenesis in these cells. These results provide the first direct evidence that PAPPA is an essential growth regulatory factor in vivo, and suggest a novel mechanism for regulated IGF bioavailability during early fetal development.
Introduction
The insulin-like growth factors (IGF1 and IGF2) are important determinants of fetal growth and postnatal development(Baker et al., 1993; Stewart and Rotwein, 1996). IGF bioactivity is modulated by a family of six IGF-binding proteins (IGFBPs)(Firth et al., 2002), the structure and function of which can be regulated by specific IGFBP proteases (Wetterau et al.,1999; Bunn and Fowlkes,2003). Recently, pregnancy-associated plasma protein A (PAPPA) was identified as a novel zinc-binding metalloproteinase secreted by normal human fibroblasts with IGFBP4 as its substrate(Lawrence et al., 1999). IGFBP4 is an inhibitor of IGF action and, in this capacity, may serve as a pericellular reservoir for IGFs (Mohan et al., 1989; Pintar et al.,1998). Cleavage of IGFBP4 by PAPPA results in increased bioavailability and mitogenic effectiveness of IGFs in vitro(Conover et al., 1995; Byun et al., 2001; Ortiz et al., 2003). Along with fibroblasts, PAPPA proteolytic activity has been identified in cultured osteoblasts (Conover et al.,1995; Qin et al.,2000; Ortiz et al.,2003), vascular smooth muscle cells(Bayes-Genis et al., 2001a) and ovarian granulosa cells (Conover et al.,2001). Furthermore, increased PAPPA expression in vivo has been shown to be associated with conditions of heightened IGF activity, such as neointimal hyperplasia following balloon angioplasty of pig coronary arteries(Bayes-Genis et al., 2001a),active atherosclerotic plaques in human coronary arteries(Bayes-Genis et al., 2001b),and healing human skin (Chen et al.,2003). Although PAPPA was originally described as a protein of placental origin circulating in human pregnancy(Lin et al., 1974), these data indicate additional roles for PAPPA, outside of pregnancy, in localized and finely controlled growth states. However, direct experimental evidence to date has been lacking.
Human PAPPA has an elongated zinc-binding motif, with residues coordinating the catalytic zinc ion of the active site, and a structurally important methionine residue located downstream in the so-called Met-turn, both of which are strictly conserved within the metzincin superfamily of metalloproteinases(Stocker et al., 1995; Boldt et al., 2001; Overgaard et al., 2003). Metzincins are remarkably similar in their tertiary structure, although they have only limited sequence identity. PAPPA is distinct from the other four metzincin groups (astacins, serralysins, adamalycins or reprolysins, and matrix metalloproteinases) because of a characteristic residue directly following the zinc-binding motif, and the unusual distance between the zinc-binding motif and the Met-turn (Boldt et al., 2001). The overall sequence identity between murine and human PAPPA is 91% (Soe et al.,2002), with the coding of all residues of the zinc binding and Met-turn consensus conserved in exon 4(Overgaard et al., 2003).
In this study we generated PAPPA-null mice by gene targeting and demonstrate a crucial role for PAPPA during fetal development.
Materials and methods
Construction of replacement vector
To construct a vector for targeting of the PAPPA gene, we first screened a 129 mouse embryonic stem (ES) cell genomic library (using Lambda FIX II, Stratagene, LaJolla, CA), with human PAPPA cDNA(Overgaard et al., 2000), and isolated a phage λ clone carrying DNA that included exon 4 of the mouse PAPPA gene, which encodes the protease domain(Overgaard et al., 2003). Fragments of the cloned PAPPA gene and a pKO Scrambler Series vector(Stratagene) were used for construction of the replacement vector. The neomycin-resistant gene (neo) cassette, replacing 1.6 kb of PAPPA gene sequence, including most of exon 4, was flanked by a 6 kb PstI (P) fragment and a 2 kb NsiI (N) fragment of mouse PAPPA locus DNA (5′ and 3′, respectively). Addition of neo also introduced a novel BamHI (B) restriction site. A cassette for thymidine kinase gene selection was located upstream of the first set of polylinker restriction sites. A schematic of the vector and targeting strategy is shown in Fig. 1A.
Generation of PAPPA-null mice by homologous recombination
Linearized replacement vector DNA was introduced into 129-derived ES cells by electroporation (BioRad Gene Pulsar at 230 V, 500 uF capacitance), and the cells seeded and selected on feeder layers of irradiated fibroblasts. The neo and thymidine kinase gene markers in the vector allowed the application of a positive-negative selection protocol in the presence of the drugs, G418 and FIAU. Drug-resistant colonies were picked and expanded(without selection) for further analysis. Seven prospective ES clones with targeted PAPPA alleles were identified by Southern blot analysis of ES cell DNA. Four of these seven independent targeted clones were microinjected into C576Bl/6 blastocysts and transferred into the uterine horn of foster mothers to generate chimeric mice, scored by presence of agouti coat hair. The frequency of formation of overt chimeria was high (>50%) for three of the four strains of blastocysts.
Male chimeras from these three clones (designated E3, E7, D10) were then cross-bred with C57Bl/6 females and germ-line transmission was obtained for all three. Heterozygous mutants were identified by Southern analysis of tail tip DNA. After transmission of the mutations, intercrosses between heterozygous progeny yielded homozygous mutants for E3, E7 and D10. Littermates obtained by breeding heterozygous males and females were used for all phenotypic analyses.
Genotyping
Southern analysis
ES cells and mouse tail tip DNA were digested with BamHI, run on a 0.8% agarose gel and transferred to Hybond (Amersham Pharmacia, Arlington Heights, IL). Membranes were prehybridized for 1 hour at 65°C in RapidHyb and then hybridized overnight at 65°C in the same solution containing 32P-labeled 3′ probe (see Fig. 1A). Membranes were washed at 65°C in 1×SSC/0.1% SDS, 0.3×SSC/0.1% SDS and 0.1×SSC/0.1% SDS, and then exposed to film. With this probe, homologous recombination in ES cells would be expected to show both the wild-type 15 kb fragment and a mutant 2.6 kb fragment of BamHI-digested DNA. For mouse tail DNA, we would expect wild-type mice to have a single 15 kb fragment, heterozygous mice to have both 15 kb and 2.6 kb fragments, and homozygous mutants to have single 2.6 kb fragments (see insert Fig. 1A).
PCR
PCR on mouse tail DNA was performed using primers: 5′-ATG ATT CAT GAG ATT GGG CAT AG-3′ and 5′-TGT TGT AAG GAG TGT TGA AGA AGC-3′,to detect exon 4 in the mouse PAPPA gene; and 5′-AGG ATC TCC TGT CAT CTC ACC TTG CTC CTG-3′ and 5′-AAG AAC TCG TCA AGA AGG CGA TAG AAG GCG-3′, to detect neo. PCR reactions containing these primers generated fragments in ethidium bromide-stained agarose gels of 223 bp for the endogenous exon 4-containing PAPPA gene and 492 kb for the recombinant neo-containing gene.
PCR-based sexing of mouse embryos was performed according to the method of McClive and Sinclair (McClive and Sinclair, 2001), using yolk sac DNA and primer pairs for Sry, the master sex determining gene on the Y chromosome, and myogenin, a control gene. It is known that there is a great deal of variability in embryo sizes even among littermates, and that male embryos may develop faster than females. Therefore, yolk sacs from embryos were sexed by PCR to rule out possible gender bias.
RT-PCR
Total RNA was extracted from whole embryos and tissues using RNeasy Mini kit (Qiagen, Valencia, CA) and treated with DNase (DNA-free, Ambion, Austin,TX). RNA (400 ng) was reverse-transcribed using TaqMan Reverse Transcription reagents (PE Biosystems, Foster City, CA), according to manufacturer's instructions. Primer sequences for assessment of PAPPA mRNA expression were as above for a predicted PCR product of 223 bp. Having established the linear range, amplifications were performed for 32 cycles. The initial denaturation was performed at 94°C for 5 minutes, cycles were at 94°C for 30 seconds, 62°C for 30 seconds and 72°C for 1 minute,and full-length products were obtained by a final elongation period of 10 minutes at 72°C. PCR reaction products were analyzed by agarose gel electrophoresis and visualized by ethidium bromide staining.
Primary cell cultures
Primary cultures of mouse embryo fibroblasts (MEF) were derived from E13.5 day embryos from heterozygous matings. Tissue from each embryo was used for genotyping. Embryos were washed, minced, trypsinized and single cell suspensions plated in high glucose DMEM, containing glutamine, penicillin,streptomycin, β-mercaptoethanol and 10% ES cell-tested FCS. Cells at passage 2-4 were used for experiments.
IGFBP4 protease activity assay
Primary cultures of MEF were washed and changed to serum-free medium. After 24 hours, conditioned medium was collected for cell-free assay. IGFBP4 proteolysis was assayed as described previously(Conover et al., 1995; Conover et al., 2001; Lawrence et al., 1999; Overgaard et al., 2000), by incubating MEF-conditioned medium samples at 37°C for 6 hours with 125I-IGFBP4 in the absence and presence of 5 nM IGF2. Proteins were separated by SDS-PAGE and visualized by autoradiography.
Cell proliferation
[3H]Thymidine incorporation was performed as described previously (Conover et al.,1995; Ortiz et al.,2003). MEF cultures were grown to 80% confluence, washed twice and changed to 0.1% FCS for 48 hours prior to experimental additions.[methyl-3H]Thymidine (0.5 μCi/ml; DuPont-NEN, Boston, MA) was added for 22-26 hours after the experimental additions. For the experiments in Table 2, cultures were washed three times immediately before addition of IGFs. For the experiments in Fig. 5B, 25 nM IGFBP4 or IGFBP3±5 nM IGF were directly added to the 48-hour-conditioned medium. Results are calculated as the percentage of total counts in the incubation medium that are incorporated into acid-precipitable material.
. | . | % of control . | . | |
---|---|---|---|---|
. | . | Wild type . | PAPPA−/− . | |
IGF1 | 2 nM | 162±28 | 139±17 | |
5 nM | 191±18 | 196±21 | ||
10 nM | 224±17 | 264±13 | ||
IGF2 | 2 nM | 154±22 | 144±36 | |
5 nM | 206±37 | 200±27 | ||
10 nM | 227±41 | 256±20 |
. | . | % of control . | . | |
---|---|---|---|---|
. | . | Wild type . | PAPPA−/− . | |
IGF1 | 2 nM | 162±28 | 139±17 | |
5 nM | 191±18 | 196±21 | ||
10 nM | 224±17 | 264±13 | ||
IGF2 | 2 nM | 154±22 | 144±36 | |
5 nM | 206±37 | 200±27 | ||
10 nM | 227±41 | 256±20 |
Ten separate MEF cultures (five wild type, five PAPPA−/−, assayed in triplicate), were treated with the indicated concentrations of IGF1 and IGF2, and [3H]thymidine incorporation determined as described in the Materials and methods. Data(mean±s.e.m., n=5) are expressed as percent of control, i.e. no IGF treatment.
Receptor phosphorylation
MEF cultures were washed and changed to 0.1% FCS for 48 hours. Twenty-five nM recombinant wild-type or protease-resistant IGFBP4(Overgaard et al., 2000; Ortiz et al., 2003) was added and incubation continued for an additional hour. Cultures were then washed with ice-cold PBS containing 2 mM vanadate and solubilized in lysis buffer [20 mM Tris (pH 7.6), 137 mM NaCl, 2 mM sodium orthovanadate, 1% NP40, 10%glycerol, 10 mM sodium pyrophosphate, 10 mM sodium fluoride, 2 mM EDTA, 40 mMβ-glycerolphosphate, 10 μg/ml leupeptin, 10 μg/ml pepstatin, 10μg/ml aprotinin]. The lysed material was precleared with protein A-Sepharose (Oncogene Research Products, Boston, MA) and non-specific rabbit IgG before overnight incubation at 4°C with 4 μg/ml antibody to type I IGF receptor β-subunit (C-20, Santa Cruz, Biotechnology, Santa Cruz, CA)or non-specific rabbit IgG. The immune complexes were precipitated by incubation with protein A-Sepharose for 6 hours at 4°C, followed by centrifugation and four washes with PBS containing Triton X-100, sodium deoxycholate and SDS. The proteins were separated by SDS-PAGE under reducing conditions, transferred to PVDF membrane, probed with a 1:1000 dilution of anti-phosphotyrosine antibody (PY20, Transduction Laboratories, Lexington, KY)and a 1:2000 dilution of secondary antibody (goat anti-mouse HRP, Transduction Laboratories), and visualized using enhanced chemiluminescence (Amersham,Biosciences, Piscataway, NJ). The membranes were reprobed with anti-type I IGF receptor antibody (1 μg/ml), with goat anti-rabbit HRP secondary (Johnson ImmunoResearch Laboratories, West Grove, PA). Signal intensities were quantified by scanning densitometry (UltraScan XL laser densitometer,Pharmacia LKB Biotechnology, Piscataway, NJ).
Skeletal staining
Embryos were eviscerated, skinned, fixed in ethanol and then stained with Alcian Blue 8GS (cartilaginous elements) and Alizarin Red S (mineralized elements) at 37°C for 3-5 days, as adapted from McLeod(McLeod, 1980). The tissues were cleared with 1% KOH and the skeletons stored in glycerol.
In situ hybridization
Whole-mount embryos were hybridized with digoxigenin-labeled RNA probes as described previously (Fuchtbauer et al., 1995). cDNA clones, all contained in vectors with the T7 promoter, were used as templates for in vitro transcription. Murine PAPPA clone E11 (Soe et al., 2002) was used for PAPPA. Rat IGF2 and mouse IGFBP4 cDNAs were kindly provided by Dr C. Bondy (NIH-NICHD,Bethesda, MD) and Dr S. Drop (Rotterdam, The Netherlands), respectively.
Results
Generation of PAPPA-null mice
The general scheme for the generation of PAPPA-null mice through homologous recombination in embryonic stem cells is presented in Fig. 1A. Interbreeding of heterozygous mice gave rise to the expected mendelian distribution of the PAPPA gene. The male:female ratio for homozygous mutants was approximately 50:50. Homozygous PAPPA-deficient mice were viable, but only 60%of the size of their wild-type littermates at birth(Fig. 1B). They remained 40%smaller during postnatal development (Fig. 1C). This growth-deficient phenotype was the same in three independently targeted mouse lines, indicating a consistent effect of PAPPA deletion. There was no apparent size or growth difference between wild-type and heterozygous mice. Allometric measures of various major internal organs of 8-week-old mice indicated a general proportionate decrease in the wet weight of all organs, as well as in body and bone length in PAPPA-/- versus wild-type mice(Table 1). Apart from their small size, the PAPPA-deficient mice appeared normal and were fertile.
. | Wild type . | PAPPA−/− . | % . | P value . |
---|---|---|---|---|
Body weight (g) | 33.7±1.5 | 18.0±0.9 | 53 | <0.0001 |
Heart (g) | 0.1760±0.0075 | 0.1070±0.0058 | 61 | <0.0001 |
Spleen (g) | 0.0930±0.0067 | 0.0400±0.0033 | 43 | <0.0001 |
Liver (g) | 1.674±0.093 | 0.922±0.084 | 55 | <0.0001 |
Kidney (g) | 0.208±0.008 | 0.121±0.009 | 58 | <0.0001 |
Body length(mm)* | 95.0±1.3 | 84.0±2.1 | 88 | 0.0012 |
Femur length (mm) | 15.9±0.3 | 14.0±0.2 | 88 | 0.0004 |
. | Wild type . | PAPPA−/− . | % . | P value . |
---|---|---|---|---|
Body weight (g) | 33.7±1.5 | 18.0±0.9 | 53 | <0.0001 |
Heart (g) | 0.1760±0.0075 | 0.1070±0.0058 | 61 | <0.0001 |
Spleen (g) | 0.0930±0.0067 | 0.0400±0.0033 | 43 | <0.0001 |
Liver (g) | 1.674±0.093 | 0.922±0.084 | 55 | <0.0001 |
Kidney (g) | 0.208±0.008 | 0.121±0.009 | 58 | <0.0001 |
Body length(mm)* | 95.0±1.3 | 84.0±2.1 | 88 | 0.0012 |
Femur length (mm) | 15.9±0.3 | 14.0±0.2 | 88 | 0.0004 |
Body and tissue weights and lengths (mean±s.e.m.) of six wild-type and six PAPPA−/− male mice at 8 weeks.
%, percent of PAPPA−/− weight or length compared with wild-type weight or length.
Body length was measured as nose-to-anus.
RNA prepared from neonatal tissues verified the complete loss of PAPPA expression by RT-PCR in homozygous mutants(Fig. 1D). Moreover, primary mouse embryo fibroblasts (MEF) derived from E13.5 wild-type and homozygous PAPPA mutant littermates were cultured and the conditioned media assayed for the presence of functional PAPPA, i.e. IGFBP4 protease activity. 125I-IGFBP4 incubated with wild-type MEF conditioned medium was fully proteolyzed into fragments of 18 and 14 kDa(Fig. 1E), similar to the cleavage pattern produced by recombinant PAPPA(Overgaard et al., 2000). IGF2 is a cofactor for the proteolytic activity in vitro as a result of its ability to bind IGFBP4 and increase susceptibility to PAPPA(Byun et al., 2000; Laursen et al., 2001), and fetal mouse fibroblasts secrete high levels of IGF2(Adams et al., 1983),sufficient to promote PAPPA-mediated IGFBP4 proteolysis in these cultures. By contrast, medium conditioned by MEF from PAPPA-/- embryos exhibited no proteolysis of 125I-IGFBP4, either without or with addition of IGF2.
Skeletal development
Investigation of the effect of PAPPA gene disruption on skeletal development during embryogenesis indicated developmental delay in the appearance of ossification centers in the clavicle, facial and cranial bones,vertebrae, and the digits of the forelimbs and hindlimbs(Fig. 2). The initiation of mineralization occurs between E12.5 and E13.5 in the mouse. As expected(Huang et al., 1997), the first bone to mineralize in both wild-type and PAPPA-/- mice was the clavicle. At E13.5, the middle portion of the clavicle was clearly mineralized in the wild type, whereas mineralization was just initiated in PAPPA-/- littermates (Fig. 2A). The delay in mineralization occurred in bones that form via intramembranous ossification (cranial vault) and endochondral ossification(vertebrae, metatarsals) (Fig. 2B,C,D). At E16.5, no caudal vertebrae were mineralized in the PAPPA-/- mouse, whereas three to four caudal vertebral elements were undergoing mineralization in wild-type littermates. Likewise an additional metatarsal had initiated mineralization in the wild type compared with the PAPPA-/- mice. The frontal and interparietal bones of the cranial vault were both undergoing mineralization at E16.5, but these processes were far less complete in PAPPA-/- mice. Examination of several skeletons from E13.5 to E18.5 littermates indicated ossification delays of approximately 1 day.
PAPPA expression during fetal development
Whole embryo in situ hybridization indicated the presence of PAPPAtranscripts in wild-type mice at all embryonic stages investigated(Fig. 3). PAPPA was expressed in the paraxial mesoderm in the presomitic cells shortly before segmentation and in the posterior part of the somites, a pattern seen clearly from E9.5 to E13.5 (Fig. 3B-F,H). In the limbs, PAPPA expression was first detected in the forelimb at E10.5(Fig. 3C) as a diffuse expression in the center of the limb bud. At E11.5(Fig. 3D), there was a clear pattern in both forelimb and hindlimb. As the limb develops, the pattern in the limbs changes in a manner suggesting the expression to be in pre-cartilage cells. At E10.5 (Fig. 3C),defined PAPPA expression can be seen in the brain. The staining seen in the region behind the ear and on the side appeared to be neural crest cells. There was also expression of PAPPA on the distal part of the branchial arches.
PAPPA-deficiency affects IGF bioavailability but not expression
PAPPA expression in the mouse embryo parallels that of IGF2(Stylianopoulou et al., 1988; Lee et al., 1990), the dominant IGF during early fetal development in rodents(Baker et al., 1993). IGF2 mRNA was expressed in the head mesenchyme formed from both the mesoderm and cephalic portion of the neural crest, and in most mesoderm-derived tissues, particularly somites and lateral mesoderm(Fig. 4A,B,D,E). In addition,IGF2 expression in chondrocytes has been shown to be elevated prior to ossification (Stylianopoulou et al.,1988). There appeared to be no difference in the level or pattern of IGF2 mRNA expression between wild-type and PAPPA-/-mice (Fig. 4A,B versus Fig. 4D,E). However, delays in limb and tail development were apparent in E12.5 PAPPA-/- embryos. Most mesodermally-derived tissues of wild-type and PAPPA-/- embryos also expressed IGFBP4 mRNA (Fig. 4C,F). This pattern of IGFBP4 expression was consistent with previous studies (Schuller et al.,1993; Pintar et al.,1998). Similarly, northern analysis for IGF2 and IGFBP4 mRNA showed no difference of expression between wild-type and PAPPA-null mouse embryos (data not shown). Serum IGF1 levels were not significantly different between newborn wild-type mice (32±5, n=11) and PAPPA-deficient mice (35±4, n=13). Fibroblasts from PAPPA-null E13.5 embryos were as responsive as those from wild-type embryos to growth stimulation by IGF1 and IGF2, indicating that the capacity for type I IGF receptor signaling was not altered in cells of PAPPA-null mice (Table 2). Thus, there was no apparent alteration in expression of IGF ligand, IGFBP4 or type I receptor in PAPPA-/- mice.
Nonetheless, loss of PAPPA and the ability to proteolyze IGFBP4 resulted in impaired IGF-mediated growth responses in MEFs from PAPPA-/-embryos. Impairment in type I IGF receptor tyrosine phosphorylation, the initial event in ligand-mediated receptor activation(LeRoith et al., 1995), is shown in Fig. 5A. MEFs from wild-type and PAPPA-/- embryos were incubated for 48 hours, and then recombinant wild-type or protease-resistant IGFBP4 was added to the medium for another hour of incubation. Cells were washed, lysed and immunoprecipitated with antibody against the type I IGF receptor. Immunoprecipitates were subjected to immunoblotting for tyrosine phosphorylation. Addition of wild-type IGFBP4 had no inhibitory effect on endogenous type I IGF receptor phosphorylation in wild-type MEFs but inhibited receptor phosphorylation in PAPPA-/- MEFs by 70%. Protease-resistant IGFBP4 inhibited type I IGF receptor phosphorylation by 70-80% in both wild-type and PAPPA-/- MEFs. Fig. 5B similarly shows that IGFBP4 had no effect on IGF-stimulated [3H]thymidine incorporation in wild-type MEFs, but inhibited stimulation by 80% in PAPPA-/-MEFs. Stimulation of [3H]thymidine incorporation by IGF analogs with markedly reduced affinity for IGFBP4(Francis et al., 1993; Oh et al., 1993) was equally effective in wild-type and PAPPA-/- MEFs. IGFBP3, which is not proteolyzed by PAPPA, inhibited IGF-stimulated growth in both wild-type and PAPPA-/- MEFs. These data are consistent with a model of PAPPA increasing local IGF bioavailability and mitogenic effectiveness through cleavage of IGFBP4.
Discussion
The results of this study demonstrate a causal relationship between deletion of the PAPPA gene in mice and a growth-deficient phenotype that becomes apparent at least as early as embryonic day 12.5 and persists after birth. Thus, PAPPA, a novel metalloproteinase in the superfamily of metzincins (Boldt et al.,2001), is an important fetal growth regulatory factor in vivo.
In many instances, ablation of one member of a gene family has resulted in compensation by one or more other members. We can argue against complete functional redundancy in this case because specific IGFBP4 protease activity is abolished and PAPPA-null mice have a distinct phenotype, i.e. proportional dwarfism. In addition, there was no apparent compensation in other components of the IGF system in the PAPPA-/- mice, when assessed as follows.
(1) In situ hybridization demonstrated the widespread pattern of IGF2 and IGFBP4 mRNA expression during normal embryogenesis,as has been previously reported(Stylianopoulou et al., 1988; Lee et al., 1990; Schuller et al., 1993; Pintar et al., 1998), with no apparent change in spatial pattern or level of mRNA in the PAPPA-deficient embryos. Similarly, there was no significant difference in steady-state levels of IGF2 and IGFBP4 mRNA in wild-type and PAPPA-/-embryos as determined by northern analysis.
(2) Serum levels of IGF1 were not different between wild-type and PAPPA-deficient mice.
(3) Fibroblasts from PAPPA-null E13.5 embryos were as responsive as those from wild-type embryos to growth stimulation by IGF1 and IGF2, indicating that capacity for type I IGF receptor signaling was not altered in the absence of PAPPA gene expression.
Thus, IGF ligands are expressed, type I IGF receptors are present and operational, and the potential reservoir function of IGFBP4 is preserved in both wild-type and PAPPA-/- embryos. The major difference is the presence or absence of PAPPA and its proteolytic activity.
Although PAPPA-deficient mice were significantly smaller than wild-type littermates, the organ-to-body weight ratios were normal. Thus, the impact of the PAPPA mutation could be on control of the timing, rate and/or duration of the growth process early in embryonic development that is important for size control. Skeletal development experiments indicated striking differences in embryonic size and ossification between wild-type and PAPPA-null embryo littermates as early as E13.5. These findings,together with the findings of developmental delay as early as E12.5 as assessed by in situ hybridization and proportional reduction in organ weights,indicate that the impact of PAPPA gene disruption on embryonic growth occurred prior to organogenesis (E10-E14 in the mouse).
The phenotype of the PAPPA-null mouse is strikingly similar to that of the IGF2-null mouse. Thus, mice lacking the ability to express the IGF2gene are small at birth and remain 60% of the size of wild-type littermates during post-natal life (DeChiara et al.,1990). They also show a similar delay in skeletal ossification(Liu et al., 1993). The principal period of IGF2 impact on body size has been suggested to occur during E9-E10 (Burns and Hassan,2001). In the absence of the IGF2 gene there are significant decreases in cell proliferation during this period of time that could account for the smaller size of the embryo, detected at ∼E12.5. Double mutation studies indicated that IGF2 signals through type I insulin and IGF receptors in early embryogenesis (Baker et al., 1993; Louvi et al.,1997), and a similar growth-deficient phenotype occurs with disruption of insulin receptor substrate 1, a key intracellular transducer of type I insulin and IGF receptor signaling(Araki et al., 1994; Tamemoto et al., 1994). The most straightforward explanation for the observed alterations in fetal growth in PAPPA-deficient mice is that PAPPA is necessary for amplification of receptor-mediated IGF2 signaling through specific IGFBP4 proteolysis during a crucial period in embryogenesis. This model is supported by the similar phenotype of PAPPA-, IGF2- and insulin receptor substrate 1-null mice. Interestingly, studies by Pintar et al. indicate that IGFBP4-null mice are 10-15% smaller than wild-type mice (Pintar et al., 1998). This would support the view of IGFBP4 as an important local reservoir of IGFs that can then be released at definitive times and discrete regions through proteolysis. Furthermore, this model is supported by our in vitro studies with fibroblasts from wild-type and PAPPA-/- embryos, demonstrating the importance of PAPPA-induced IGFBP4 proteolysis in regulating endogenous and exogenous IGF bioactivity. Currently, the only identified function of PAPPA is as an IGFBP protease. However, contributions of other IGF system components, as well as possible IGF-independent effects of PAPPA, remain to be determined.
A major conclusion from our genetic data is that PAPPA is clearly involved in optimal fetal growth and development. This growth regulatory mechanism may also underlie the association between PAPPA levels and fetal development recently reported in humans (Smith et al.,2002). Furthermore, PAPPA-deficient mice provide a unique model for testing hypotheses concerning the role of PAPPA in regulating IGF action postnatally, e.g. during bone remodeling and vascular repair (Z. T. Resch, R. D. Simari and C. A. Conover, unpublished).
Acknowledgements
The authors thank Jeffry Harden, Ralitza Mantcheva and Xiaoshong Wu for all their help with cloning and microinjections, and Christopher Ortiz, Shelley Fohrman and Keira Suttcliffe-Stephenson for their excellent technical assistance. All animal studies were performed under protocols reviewed and approved by Mayo's Institutional Animal Care and Use Committee. This work was supported in part by grants from Novo Nordic Foundation (C.O.) and the Mayo Foundation.