Characterization of Platelet-Derived Growth Factor-A Expression in Mouse Tissues Using a lacZ Knock-In Approach

Expression of the platelet-derived growth factor A-chain gene (Pdgfa) occurs widely in the developing mouse, where it is mainly localized to various epithelial and neuronal structures. Until now, in situ mRNA hybridization (ISH) has been the only reliable method to identify Pdgfa expression in tissue sections or whole mount preparations. Validated protocols for in situ detection of PDGF-A protein by immunohistochemistry is lacking. In particular, this has hampered understanding of Pdgfa expression pattern in adult tissues, where ISH is technically challenging. Here, we report a gene targeted mouse Pdgfa allele, Pdgfaex4 COIN, which is a combined conditional knockout and reporter allele. Cre-mediated inversion of the COIN cassette inactivates Pdgfa coding while simultaneously activating a beta-galactosidase (lacZ) reporter under endogenous Pdgfa transcription control. The generated Pdgfaex4 COIN-INV-lacZ allele can next be used to identify cells carrying a Pdgfa null allele, as well as to map endogenous Pdgfa expression. We evaluated the Pdgfaex4 COIN-INV-lacZ allele as a reporter for endogenous Pdgfa expression patterns in mouse embryos and adults. We conclude that the expression pattern of Pdgfaex4 COIN-INV-lacZ recapitulates known expression patterns of Pdgfa. We also report on novel embryonic and adult Pdgfa expression patterns in the mouse and discuss their implications for Pdgfa physiology.


Introduction
The platelet-derived growth factor (PDGF) family plays fundamental roles during several stages of vertebrate development (reviewed in [1]. The mammalian PDGFs encompass 5 protein isoforms, which are dimers of 4 distinct, but related, polypeptide chains (PDGF-A-D) encoded by separate genes. PDGF-A-D chains assemble into 4 homodimers (PDGF-AA, BB, CC, DD) and one heterodimer (PDGF-AB). PDGFs exert their biological activities through two receptor tyrosine kinases, PDGF receptoralpha (PDGF-Ra) and beta (PDGF-Rb) (reviewed in [2]. Whereas ligand-receptor interactions mapped in vitro suggest a significant degree of redundancy in PDGF ligand-receptor interaction, in vivo gene knockout analyses show that PDGF-AA and PDGF-CC are the principal ligands for PDGF-Ra, at least during development, whereas PDGF-BB is the key ligand for PDGF-Rb [3][4][5][6][7]. The developmental roles of PDGFs mapped to-date suggest paracrine modes of signaling, i.e. PDGFs released from one type of cells act on neighbors of a different type (reviewed in [8]. Thus, various developing epithelia express PDGF-A and PDGF-C, whereas the neighboring mesenchyme expresses PDGF-Ra [9][10][11]. Similarly, PDGF-B is expressed in angiogenic vascular endothelial cells, and triggers responses in adjacent murals cells (vascular smooth muscle cells (VSMC) and pericytes) [12,13]. The paracrine mode of action of PDGF-AA and PDGF-BB depends in part on their extracellular distribution. This, in turn, is regulated by C-terminal heparan sulfate proteoglycan-binding motifs that may be present or absent in the PDGF protein depending on alterative splicing (in PDGF-A) [14] or alternative proteolytic processing (in PDGF-B) [15]. The activity of PDGF-C and PDGF-D in tissues further depends on extracellular proteolytic processing; both factors carry N-terminal CUB domains that require removal for receptor-binding to occur (reviewed in [16]).
PDGFs have also been implicated in the pathogenesis of a number of different diseases. With few exceptions, mainly involving various cancers, the evidence for involvement is based on correlations between expression and disease. Functional evidence through specific gene inactivation, or the use of highly specific inhibitors, is generally lacking. Nevertheless, a wealth of data suggests the involvement of different PDGFs in different types of fibrotic conditions affecting the lung, liver, skin, kidney and heart (reviewed in [32]). PDGF signaling has also been implicated as a pathogenic driver in vascular disorders, including atherosclerosis, pulmonary hypertension and retinopathy (reviewed in [1]). In all of these conditions, the assumed mode of signaling is paracrine. A similar mechanism has also been proposed for the involvement of PDGFs in the formation of tumor stroma (reviewed in [33]). However, in addition to the paracrine functions, autocrine PDGF signaling is also known to play a role in some cancers. This evidence is particularly strong in the case of dermatofibrosarcoma protuberans, a human skin tumor caused by chromosomal translocations that fuse PDGFB coding sequences with transcriptional control elements from the COL1A1 gene [34]. This leads to production of PDGF-B in collagen-I producing cells (fibroblasts/fibrocytes). These cells carry endogenous PDGF receptors, hence forming the basis for an autocrine growth stimulatory loop.
In determining the mode of action and function of PDGFs in adult tissues in physiological and pathophysiological settings, two hurdles appear: (i) the lack of well-validated tools and techniques for the determination of gene and protein expression (especially the expression of PDGF-A and PDGF-B) and (ii) the lack of specific and validated inhibitors for studies in vivo. For PDGF-A, the most broadly expressed of the PDGF ligands, this void is noteworthy: no validated specific immunohistochemistry protocols for in vivo PDGF-AA detection have yet been reported, to our knowledge. Moreover, the use of a floxed Pdgfa allele has not been reported previously. The embryonic-to-early-postnatal lethality of full Pdgfa knockout [5] prevents analysis of adult roles of PDGF-A using this model.
Several PDGF-A antibodies are available commercially, and their use in immunohistochemistry (IHC) has been reported in tissues from human [35][36][37], rat [38], chicken [39], and mouse [40]. To our knowledge, none of the reported PDGF-A immunohistochemistry protocols have been validated using Pdgfa knockout tissue as negative control. In theory, even with the access to specific antibodies and staining protocols, the characterization of PDGF-A expression in tissues by IHC is likely going to be problematic since PDGF-A is rapidly secreted from the producer cell. Moreover, most PDGF-A is expressed as a short, diffusible, splice isoform, whereas the long heparan sulfate proteoglycanbinding isoform is rare in most instances [41,42]. Therefore, developmental expression studies have primarily utilized RNA in situ hybridization (ISH). In this way, PDGF-A expression has been mapped to e.g. CNS neurons [43], developing embryonic organs [9], embryonic lung [5,44] and intestinal epithelium [11], tubular epithelium of testis and epididymis [25], embryonic epidermis and hair follicle epithelium [10]. Whereas in some instances the spatial resolution of non-radioactive ISH has permitted mapping of the expression with single cell resolution, this is usually not the case. Also, ISH techniques are prone to nonspecific background signals; in our own hands this was especially problematic in tissues rich in extracellular matrix, as occurs commonly in both normal and pathological adult tissues. Although we successfully applied non-radioactive ISH to uncover embryonic Pdgfa mRNA expression patterns in several instances [10,11,24], we experienced notorious difficulties in maintaining comparable signal intensities and signal-to-noise ratios from one experiment to the other.
To overcome the mentioned problems in elucidating specific PDGF-A expression patterns and functions, we have now generated, and performed an initial characterization of, a mouse Pdgfa allele (Pdgfa ex4COIN ), which combines the features of a conditional null and expression reporter allele. After Cre-mediated recombination and functional inactivation, the allele (Pdgfa ex4 -COIN-INV-lacZ ) expresses lacZ from endogenous regulatory elements, thus providing a reliable proxy for Pdgfa expression. Pdgfa ex4COIN-INV-lacZ also provides a marker for cells in which Pdgfa gene inactivation has occurred. This is of great importance since Cre-mediated recombination in somatic cells is generally chimeric. Pdgfa ex4COIN therefore provides a new useful tool for studies of PDGF-A functions in mice, particularly in adults. Here, we show that Pdgfa ex4COIN functions as a conditional null allele. We also use the Cre-recombined allele (Pdgfa ex4COIN-INV-lacZ ) to confirm previously reported embryonic Pdgfa expression patterns, as well as to provide new information about Pdgfa expression patterns in healthy adults.

Ethics statement
The Pdgfa ex4COIN mice were generated at Regeneron Pharmaceuticals Incß, USA, and shipped to Karolinska Institute and Uppsala University, Sweden, were all analyzes were done. The protocol for this study was approved by the Stockholm's North Committee on the Ethics of Animal Experiments (permit numbers N33/10 and N15/12) and by the Uppsala Committee (permit number C224/12). All efforts were made to minimize animal suffering, and all surgery was performed under Hypnorm/ Midazolam anesthesia.

Generation of mice
The Pdgfa ex4COIN allele was generated by inserting a TMLacZ-COIN-f1neo cassette as an artificial intron into Pdgfa exon 4 in a BAC clone. Exon 4 was thereby split into exon4a (78 bp) and exon4b (110 bp). The lacZ gene was inserted antisense and flanked by lox71 and lox66 sites. These modified loxP sites enable irreversible inversion of the intermediate sequence in the presence of Cre. The engineered BAC was recombined into ES cells with 129S6SvEv/C57BL6F1 background using VelociGene technology (Valenzuela et al., 2002). Two ES cell clones with a correctly integrated Pdgfa ex4COIN allele (clone B3 and D5) were obtained and used to generate mouse lines that were subsequently confirmed to be indistinguishable. One of these lines (D5) was kept for further analysis.

X-gal staining
Visualization of beta-galactosidase expression was done in whole mount embryos or dissected organs. X-gal staining of muscle tissue to be further used for IHC was performed on freefloating sections. Embryos were immersion fixed in 4% paraformaldehyde for 1 h. For staining of inner organs, embryos were decapitated and the skin was partially removed before fixation. Postnatal mice (older than P12) were perfusion fixed through the heart for 3 min; inner organs were then cut out and postfixed for 1 h. Prior to X-gal staining, tissues were washed in PBS and permeabilized in PBS/2 mM MgCl 2 /0.02% Igepal/0.01% Nadeoxycholate for at least 1 h at room temperature, with change of the solution 3 times. Staining was performed for 2-16 h at 37uC in 50 mg/ml X-gal (Promega) in PBS/2 mM MgCl 2 /0.02% Igepal/ 0.01% Na-deoxycholate 5 mM K 4 Fe(CN) 6 /5 mM K 3 Fe(CN) 6 . Washings in PBS ended the reaction.

Fluorescent detection of neuromuscular junctions
Whole mount X-gal stained muscle was soaked in 30% sucrose, frozen with dry ice and sectioned in a freezing sleigh microtome. Free-floating 100 mm thick sections were re-stained with X-gal for 30 min to enable the staining to uniformly penetrate the section. Sections were blocked in PBS/1% bovine serum albumin/0.5% triton X-100 for 1 h at room temperature. Sections were stained at room temperature with alpha-bungarotoxin conjugated to Alexa Fluor-555 (Molecular Probes) diluted to 3 ng/ml in PBS/0.5% BSA/0.25% triton X-100 followed by washings in PBS and mounting in ProLong Gold anti-fade reagent with DAPI (Invitrogen). Imaging was done using a LSM700 confocal microscope (Zeiss). Muscle morphology and positive X-gal staining were visualized using transmitted light in the empty A647 channel.

Pdgfa gene expression in public databases
Expressed Sequence Tag (EST) data were extracted from the EST profile of the NCBI UniGene databases (http://www.ncbi. nlm.nih.gov/UniGene/). Human PDGFA expression data was extracted from Hs.535898, and mouse Pdgfa data from Mm.2675.

Statistics
The genotype distribution of mice that were born alive, from heterozygous crossings of Pdgfa ex4COIN-INV-lacZ/+ was compared to the expected Mendelian distribution. Pups from .10 litters were compared using Chi-square test (www.graphpad.com). P,0.05 was considered to be statistically significant.

Generation of a conditional Pdgfa ex4COIN allele
A conditional Pdgfa knockout allele was generated using the ''conditional-by-inversion'' (COIN) strategy [46,47]. This strategy is based on the insertion of an inverted (and inactive) COIN module into the gene of interest, by a single targeting event in embryonic stem (ES) cells. In Pdgfa ex4COIN , the COIN module, along with a neomycin cassette, was placed as an artificial intron into Pdgfa exon 4. As a result, exon 4 was split in two new exons, 4a and 4b (Fig. 1). The Pdgfa ex4COIN allele was expected to be functional, since splicing between exons 4a and 4b resulted in an RNA sequence identical to that encoded by the original exon 4. Exon 4-encoded sequences are absolutely required for the production of a functional PDGF-A protein, and thus, this splicing event is necessary for the functionality of the Pdgfa ex4COIN allele. Two targeted Pdgfa ex4COIN carrying lines were generated using VelociGene technology [48], and both lines were bred to homozygosity (Pdgfa ex4COIN/ex4COIN ). Initial experiments confirmed that the two lines behaved identically, and one was therefore selected for further breeding. Pdgfa ex4COIN/ex4COIN mice were born in Mendelian numbers (data not shown) and found to be viable and normal (followed up to 10 months) as expected for a functional Pdgfa allele [5].

Generation of the Pdgfa ex4COIN-INV-lacZ allele
The COIN-module was flanked by lox66 and lox71 sites oriented head-to-head (Fig. 1). Therefore, Cre recombinase was expected to mediate irreversible inversion of the COIN-module, resulting in fusion of the lacZ sequences with Pdgfa coding sequences (Fig. 1). We crossed Pdgfa ex4COIN mice with the Cre deleter strain EIIa-Cre (Xu et al., 2001) and identified offspring with an inverted COIN-module using PCR. The resulting Pdgfa ex4COIN-INV-lacZ allele was expected to be a null allele. Exon 4a was predicted to splice into the activated COIN module and, as a result, produce a fusion protein consisting of an N-terminal exon 4a-derived portion of PDGF-A, a transmembrane (TM) domain, a lacZ cassette, and a polyadenylation site. A schematic illustration of the expected expression and processing of the PDGF-Aex4a-TM-lacZ fusion protein is shown in Figure 2. As expected from a Pdgfa null allele, no homozygous Pdgfa ex4COIN-INV-lacZ/ex4COIN-INV-lacZ mice were found alive after birth (Table 1).
Since no deletion of endogenous Pdgfa genomic sequences occurred in the Pdgfa ex4COIN-INV-lacZ allele, we expected the encoded PDGF-Aex4a-TM-lacZ fusion protein to reproduce the endogenous Pdgfa expression pattern. In order to confirm that Pdgfa and Pdgfa ex4COIN-INV-lacZ were similarly expressed, we compared quantitative real-time PCR (qPCR) data for Pdgfa and LacZ across a panel of tissues. Using Taqman probes against Pdgfa and LacZ the relative levels of expression of the wildtype Pdgfa and the Pdgfa ex4COIN-INV-lacZ alleles in different organs were compared in Pdgfa ex4COIN-INV-lacZ/+ mice at two different ages, P5 and adult. This showed that Pdgfa and lacZ expression had highly similar organ distribution at both ages (Fig. 3a, b). Lung tissue from a wildtype control showed no lacZ expression, as expected (Fig. 3a, b). To further validate the comparison, an Elastin Taqman probe was used to asses an irrelevant gene in the the same RNA samples. As expected, Elastin mRNA showed a completely different relative organ distribution (Fig. 3c). Furthermore, high levels of Pdgfa and lacZ mRNA expression correlated well with stronger X-gal staining, e.g. in kidney, lung and brain at P5 (as described later in this paper).

Developmental expression of Pdgfa ex4COIN-INV-lacZ
Since the Pdgfa expression pattern has previously been mapped at relatively high detail during mouse embryogenesis, we first analyzed heterozygous Pdgfa ex4COIN-INV-lacZ/+ mouse embryos for lacZ expression by X-gal staining. Analysis of embryonic day (E)   Table 2). Whole mount X-gal staining of E9.5 Pdgfa ex4COIN-INV-lacZ/+ embryos showed region specific expression in e.g. the 1 st branchial arch, the otic vesicles, in somites and in the tail (Fig. 4). Wholemount X-gal staining of E14.5 Pdgfa ex4COIN-INV-lacZ/+ embryos showed distinct expression in developing hair follicles in the back skin, in whiskers and eyebrows (Fig. 5a, arrowheads). Distinct staining was also observed in developing skeletal muscle, e.g. in limbs and in the thoracic region ( Fig. 5a red arrow). Pdgfa +/+ littermate controls were completely negative for X-gal staining at this age (Fig. 5b). To enable proper penetration of the X-gal staining solution, inner organs were dissected out and individually stained over night. Strong staining was confirmed in developing intestine (Fig. 5c), lung (Fig. 5d), heart (Fig. 5e) kidneys (Fig. 5f) and skeletal muscle (Fig. 5g). At the whole-mount level, strong Xgal staining was also observed in large arteries (e.g. the aorta, Fig. 5f). Apposed tissues with strong and negative/weak staining were observed at the whole-mount level. For example, positive lung epithelium neighbored negative mesenchyme (Fig. 5d), and strongly positive kidney tissue neighbored the weakly positive adrenal gland (Fig. 5f). The liver displayed a weak punctuate staining, which was deemed specific since no staining was observed in Pdgfa +/+ liver (Fig. 5h). Whole organ X-gal stainings of freely dissected organs were repeated at E17.5 and P5 with consistent results (Fig. 6 and 7). Pdgfa ex4COIN-INV-lacZ expression was detected in aorta, vessels, esophagus, thymus, lung, heart, diaphragm, stomach, liver, spleen, pancreas, intestine, adrenal, kidney, skin, bladder and brain. At P5, Pdgfa ex4COIN-INV-lacZ expression was also detected in the retina (Fig. 7q). Background expression due to endogenous beta-galactosidase activity was detected in the intestine (Fig. 6k and 7l), kidney (Fig. 7m) and in the sternum (Fig. 7r). In summary, the X-gal staining pattern in heterozygous Pdgfa ex4COIN-INV-lacZ/+ embryos appeared to reproduce known expression patterns of endogenous Pdgfa previously mapped using non-radioactive ISH ( Table 2). This fact, combined with the strength and localized nature of the staining, suggests that the Pdgfa ex4COIN-INV-lacZ expression reports endogenous Pdgfa expression faithfully. In order to map the Pdgfa ex4COIN-INV-lacZ expression pattern at higher resolution, we analyzed sections from whole-mount stained E14.5 organs (Fig. 8). Previous work has established that Pdgfa mRNA is expressed broadly in embryonic epithelia, in skeletal, cardiac and smooth muscle, and in neuronal cells (Table 2) (reviewed in [1]. These expression patterns were confirmed and extended. We found Pdgfa ex4COIN-INV-lacZ expression in dermal keratinocytes and hair follicle epithelial cells (Fig. 8a, b, c, d), corneal epithelium (Fig. 8b arrowhead), developing inner ear epithelium (Fig. 8e), bronchial epithelium (Fig. 8f), testis and  epididymis epithelium (Fig. 8g), renal epithelium (Fig. 8h), and intestinal and stomach epithelium (Fig. 8i, j). We further documented Pdgfa ex4COIN-INV-lacZ expression in developing visceral smooth muscle (Fig. 8i, j arrowheads), skeletal muscle (Fig. 8d), and cardiac muscle, the latter particularly strong in the cardiac outflow tract (Fig. 8k), and in vascular smooth muscle (Fig. 8h, k arrow heads). In the E14.5 brain, Pdgfa ex4COIN-INV-lacZ expression was restricted to neuroepithelial tissue around the dorsal horn of the lateral ventricle (Fig. 8l).

Adult expression of Pdgfa ex4COIN-INV-lacZ
While the observed patterns of lacZ expression in Pdgfa ex4COIN-INV-lacZ embryos were consistent with, and confirmatory of, already published patterns of Pdgfa mRNA expression in the mouse embryo, only limited information is available about Pdgfa expression patterns in adult mammals. The occurrence of PDGFA/Pdgfa sequences in public human and mouse EST databases suggests widespread but weak expression in many organs (Table 3), but this information does not reveal the cellular source of expression. Moreover, it is not clear if the lack of sequences in certain organs/tissues reflects a true lack of expression, or that expression in underrepresented cells has gone undetected due to dilution.
X-gal staining of adult brain slices (Fig. 9a-c) showed a complex, mainly neuronal, pattern of Pdgfa ex4COIN-INV-lacZ expression, with clear variation between different neuronal subgroups. For example, cerebellar Purkinje cells were one example of neurons with strong Pdgfa ex4COIN-INV-lacZ expression (Fig. 9b, c arrow heads). At a gross level, the adult cerebral and cerebellar Pdgfa ex4COIN-INV-lacZ expression pattern confirms the pattern of Pdgfa mRNA expression previously reported in the adult mouse brain using radioactive ISH [43]. In several other organs, such as the heart (Fig. 9d) and uterus (Fig. 9e), X-gal expression was relatively uniform at the level of a whole-mount perspective, consistent with the low cell type complexity in these organs compared to the brain. In yet other adult organs, Pdgfa ex4COIN-INV-lacZ expression was obviously non-uniform, and displayed distinctive cell type or region specificities, e.g. in the retina (Fig. 9f), adrenal gland (Fig. 9g), liver (Fig. 9h), spinal cord (Fig. 9i) and kidney (Fig. 9j; note the strong staining in the medullary papilla). Whereas the corresponding wildtype tissues were for the most part negative for X-gal staining, some endogenous background was noticed in a few tissues, including cartilage (not shown), kidney cortex (Fig. 9k) and intestinal lumen (not shown). The latter likely represents intestinal bacterial

Cell-type specific expression of Pdgfa ex4COIN-INV-lacZ in adult tissues
In order to provide details about the cellular patterns of Pdgfa expression in adult mice, whole mount X-gal stained tissues were sectioned and counterstained with hematoxylin and eosin. Tissues from more than twenty-five different organs were analyzed in this way. This confirmed the general patterns of cell-type specific expression of Pdgfa ex4COIN-INV-lacZ observed in embryos, namely in various types of epithelia, muscle, and neuronal tissue (Fig. 10).
Epithelial expression of Pdgfa ex4COIN-INV-lacZ was observed throughout the adult body. Expression levels appeared variable and often regionally restricted, implicating localized regulation of expression and possibly also region-specific functions for the produced PDGF-A protein. In the lung, Pdgfa ex4COIN-INV-lacZ expression was observed in the respiratory epithelium in the  (Fig. 10a), in the epithelium of main bronchi and terminal bronchioles (Fig. 10b, red arrowhead), and in alveoli (Fig. 10c). In the latter, expression was non-uniform and localized mainly to cells resembling type II pneumocytes (Fig. 10c, red arrowhead). Epithelial expression of Pdgfa ex4COIN-INV-lacZ was also observed throughout the gastrointestinal tract. Also here, epithelial expression was non-uniform. In the stomach, expression was mainly observed in the corpus, where it sub-localized to cells at the base of the gastric glands (Fig. 10d, e). In the colon, expression was instead localized in the surface epithelial cells, whereas crypts were negative, or showed low expression (Fig. 10f). In the skin, expression was observed in the basal layer of keratinocytes, as well as in hair follicle epithelial cells (Fig. 10g). In the kidney, expression was particularly strong in Henle's loop epithelium in the renal papilla (Fig. 10h, i), but weaker expression was also observed in cells in the distal tubules in the cortex (Fig. 10j,  arrowheads). An unexpected location of Pdgfa ex4COIN-INV-lacZ epithelial expression was asymmetrically located cysts in the pituitary (Fig. 10l, m). These cysts were lined with ciliated epithelium, making them reminiscent of Rathke's cleft cysts (RCC), which are benign remnants of Rathke's cleft, the embryonic origin of the anterior pituitary lobe. RCC have been described in humans where they are often asymptomatic [49]. RCC in mice have been reported previously [50].
Similar to the epithelial expression, the adult neuronal expression of Pdgfa ex4COIN-INV-lacZ was widespread, but nonuniform at the cellular level. In the cerebellum, strong and specific expression was observed in Purkinje neurons, whereas no other neuronal population was positive in this part of the brain (Fig. 10  n, o). The complex and widespread cellular pattern of expression of Pdgfa ex4COIN-INV-lacZ in the cerebrum was primarily neuronal (Fig. 10p), but similar to the situation in the cerebellum, not all neuronal populations were positive.
Muscular expression of Pdgfa ex4COIN-INV-lacZ was observed in skeletal muscle, as exemplified by muscle cells in the diaphragm in (Fig. 10q), in cardiomyocytes ( Fig. 9d and data not shown) and in VSMC, as illustrated in mesenteric arteries (Fig. 10r, s arrowhead), the aorta (Fig. 10t arrowhead) and in bronchial arteries (Fig. 10b). Also in muscle cells, expression was non-uniform, as illustrated e.g. in the kidney, where it was conspicuous in the arteriolar VSMC associated with the juxtaglomerular apparatus (Fig. 10j, k asterisk). Table 3. PDGFA/Pdgfa expression based on publicly available expressed sequence tag (EST) data. PDGF-A is produced by cultured myoblasts [51] and developing skeletal muscle [9,27]. We confirmed the expression of Pdgfa ex4COIN-INV-lacZ in both embryonic (Fig. 5g) and adult (Fig. 10o) skeletal muscle. However, similar to the epithelial cells and neurons, X-gal staining was non-uniform. Analysis of femoral quadriceps muscles from Pdgfa ex4COIN-INV-lacZ and wildtype controls revealed two distinct expression patterns. First, there was a general and uniform X-gal staining in all muscle fibers, which was not seen in the PDGF-A +/+ littermate control (Fig. 11a). Second, we observed a band of intensely stained spots stretching across the approximate middle of the muscle (Fig. 11b,  c, d). Similar bands of stained spots were seen also in other muscles, including the diaphragm (Fig. 7g). This staining was clearly visible already after 30 minutes of X-gal incubation, at which time the more general staining was undetectable or weak (Fig. 11c, d). The localization of the spots suggested a correlation with neuromuscular junctions. Indeed, visualization of the neuromuscular junctions using Alexa Fluor-555-conjugated alpha-bungarotoxin, which binds to acetylcholine receptors, provided a spatial correlation with the X-gal staining (Fig. 11e, f). The X-gal staining was localized to the postsynaptic area of the muscle fiber, suggesting the expression of Pdgfa ex4COIN-INV-lacZ from local synaptic muscle cell nuclei.

Localization of Pdgfa ex4COIN-INV-lacZ expression to specific cell types
The expression of Pdgfa ex4COIN-INV-lacZ enables localization to individual cells. We used co-immunofluorescence stainings of paraffin embedded tissue from P5 mice, to confirm expression in type-II pneumocytes and in vascular smooth muscle cells (Fig. 12). Surfactant protein-C (SPC) co-localized with beta-galactosidase in individual cells in the alveolar walls of the lung (Fig. 12a-c). Importantly, beta-galactosidase expression was also detected in the bronchial epithelium, where no SPC was expressed (Fig. 12a, b  arrowheads). In vessels of brown adipose tissue, alpha-smooth muscle actin was co-expressed with beta-galactosidase ( Fig. 12d-f). The fluorescent staining overlapped with the X-gal staining, as shown with transmitted light in the confocal microscope (Fig. 12g,  h). Expression of Pdgfa ex4COIN-INV-lacZ could also be localized to specific cell-types based on morphology. In the liver, strong X-gal staining was detected in megakaryocytes ( Fig. 12i-l) confirming previous data on the expression of PDGF genes during megakaryoblastic differentiation [52].

Discussion
We report on the generation and first analysis of a conditional null and expression reporter Pdgfa allele. The allele was generated using the COIN technique pioneered by scientists at Regeneron Pharmaceuticals [46]. We started by validating that the Pdgfa ex4 -COIN allele was functional by assessing viability and lack of phenotypes associated with PDGF-A deficiency in homozygous Pdgfa ex4COIN/ex4COIN mice. We also confirmed that mice homozygous for the Cre-activated allele Pdgfa ex4COIN-INV-lacZ were not recovered after birth, as expected for Pdgfa null mice on C57Bl6 enriched genetic background. The early postnatal viability originally reported for Pdgfa null mice was observed only in mixed C57Bl6/129Ola hybrid background [5].
We next analyzed heterozygous Pdgfa ex4COIN-INV-lacZ/+ mice as a potential tool for Pdgfa expression analysis, utilizing the lacZ reporter gene inserted into the Pdgfa locus. No endogenous genomic sequences were deleted in the Pdgfa ex4COIN or Pdgfa ex4COIN-INV-lacZ alleles, and hence we were hopeful that the expression of the lacZ-gene from Pdgfa ex4COIN-INV-lacZ would faithfully reproduce the endogenous Pdgfa expression pattern. Indeed, using qPCR analysis, we confirmed that the mRNA levels of Pdgfa and lacZ showed highly similar relative expression levels in different organs, suggesting co-regulation.
PDGF-A is a secreted protein and we therefore aimed for a fusion protein strategy in order to minimize potential deviation from the endogenous pattern of expression. A transmembrane anchoring sequence was inserted, such that the encoded PDGF-A- lacZ fusion protein would become membrane-associated in the expressing cells, with the lacZ domain facing the cytoplasmic compartment. Consequently, X-gal staining would be predicted to mark the cytoplasm of Pdgfa expressing cells. Indeed, our analysis of embryos showed that expression of Pdgfa ex4COIN-INV-lacZ reproduced the patterns of Pdgfa expression that have previously been revealed through ISH analysis. This, together with the strength of the lacZ expression from the Pdgfa ex4COIN-INV-lacZ allele, and the ease with which it could be localized to specific cell types and individual cells, imply that Pdgfa ex4COIN-INV-lacZ is a faithful and powerful Pdgfa expression reporter in the mouse. While our data suggest that X-gal staining of Pdgfa ex4COIN-INV-lacZ/+ mice provides a sensitive and specific proxy for the expression of Pdgfa, the model is less useful for other purposes, such as cell sorting or fate mapping (of Pdgfa-expressing cells). For an overview of the features and advantages with the COIN technique, the reader is referred to the original publication by Economides et al [46].
The possibility to map Pdgfa expression patterns in adult tissues is of particular interest, since, until now, validated tools and protocols for in situ Pdgfa expression analysis in adult mice have not been available. We found abundant Pdgfa ex4COIN-INV-lacZ expression in most analyzed adult organs, which were mapped to distinct cell types and even individual cells. The general tissue/cell type pattern of expression was similar in the adult and embryo, i.e. the predominant sites of expression were various epithelial, muscle, and neuronal cell types. The constitutive expression of Pdgfa ex4COIN-INV-lacZ in quiescent adult tissues challenges the view of PDGF-AA as being mainly a mitogen for mesenchymal cells during development and tissue repair or pathology, such as wound healing, fibrosis and cancer. Indeed, available information on the transcriptional regulation of the PDGF-A gene largely depicts transcriptional elements engaged by mitogenic signaling, tissue injury and tumor promotion (reviewed in [53]). The transcriptional mechanisms behind the normal constitutive cell-type specific expression of Pdgfa observed in the present study remain unknown. Future in vivo analysis of Pdgfa transcriptional regulation will therefore benefit from the access to faithful gene expression reporters, such as Pdgfa ex4COIN-INV-lacZ . Moreover, studies on the role of PDGF-A in cancer, including autocrine growth regulation in the cancer cells themselves, as well as the paracrine recruitment of tumor stroma (reviewed in [33], and the involvement of PDGF-A in tissue fibrosis (reviewed in [32]) will benefit from more precise information about the endogenous PDGF-A expression patterns in both normal and pathological situations.
Two conspicuous physiological expression patterns of Pdgfa ex4 -COIN-INV-lacZ illustrate the power of Pdgfa ex4COIN-INV-lacZ/+ mice for Pdgfa expression analysis. 1) The Pdgfa expression in pituitary RCC's remains functionally unclear but provides a possibility for their easy visualization. This may be of use for the analysis of RCC localization and number in correlation with other developmental abnormalities and pathological processes. The Pdgfa ex4COIN-INV-lacZ expression in these structures is probably a remnant of the developmental situation in which Pdgfa is broadly expressed in the pharyngeal epithelium. 2) The second remarkable Pdgfa ex4COIN-INV-lacZ pattern localized to the neuromuscular junctions. Whereas PDGF-A and PDGF-Ra proteins have been suggested at neuromuscular junctions based on IHC techniques [54], the cellular sources of the proteins were not revealed in this study. The Pdgfa ex4COIN-INV-lacZ pattern is suggestive in this regard, since the X-gal staining was localized to a region of the muscle fiber corresponding to the postsynaptic area. This expression pattern appears consistent with that of other molecules localized to the postsynaptic membrane of the neuromuscular junction, such as acetylcholine receptors (reviewed in [55]). These observations therefore suggest that Pdgfa expression from local (synaptic) nuclei is induced and maintained by synaptic activity and postsynaptic signaling. Further studies using the conditional nature of the Pdgfa ex4COIN allele has the potential to reveal the functional importance of Pdgfa expression at this location.