Mice Lacking Platelet-Derived Growth Factor D Display a Mild Vascular Phenotype

Platelet-derived growth factor D (PDGF-D) is the most recently discovered member of the PDGF family. PDGF-D signals through PDGF receptor β, but its biological role remains largely unknown. In contrast to other members of the PDGF family of growth factors, which have been extensively investigated using different knockout approaches in mice, PDGF-D has until now not been characterized by gene inactivation in mice. Here, we present the phenotype of a constitutive Pdgfd knockout mouse model (Pdgfd-/-), carrying a LacZ reporter used to visualize Pdgfd promoter activity. Inactivation of the Pdgfd gene resulted in a mild phenotype in C57BL/6 mice, and the offspring was viable, fertile and generally in good health. We show that Pdgfd reporter gene activity was consistently localized to vascular structures in both postnatal and adult tissues. The expression was predominantly arterial, often localizing to vascular bifurcations. Endothelial cells appeared to be the dominating source for Pdgfd, but reporter gene activity was occasionally also found in subpopulations of mural cells. Tissue-specific analyses of vascular structures revealed that NG2-expressing pericytes of the cardiac vasculature were disorganized in Pdgfd-/- mice. Furthermore, Pdgfd-/- mice also had a slightly elevated blood pressure. In summary, the vascular expression pattern together with morphological changes in NG2-expressing cells, and the increase in blood pressure, support a function for PDGF-D in regulating systemic arterial blood pressure, and suggests a role in maintaining vascular homeostasis.


Introduction
The platelet-derived growth factor (PDGF) family consists of four ligands, PDGF-A to -D.They are all secreted as disulfide-bonded homo-or heterodimers, PDGF-AA, -AB, -BB, -CC and-DD, and exert their biological functions by binding to and signaling via two tyrosine kinase receptors, PDGF receptor (PDGFR)α and PDGFRβ [1][2][3][4].The biological function of the PDGF family and their receptors has been extensively investigated, with the exception of PDGF-D [5].
PDGF-D shares the conserved growth factor domain with the other members of the family [3,4].PDGF-C and PDGF-D also contain an additional N-terminal CUB domain that requires proteolytic removal to enable receptor binding, and are thus secreted as latent dimers [6].Serine proteases such as urokinase-type plasminogen activator (uPA) and matriptase can activate PDGF-D [7][8][9].PDGF-D is the only ligand that binds specifically to, and induces downstream signaling via PDGFRβ [3].Deletion of the genes encoding the different PDGF ligands and receptors have shown that they are essential during embryonic development.PDGF-A and -C signaling via PDGFRα are broadly required during organogenesis [5].The PDGF-B and PDGFRβ deficient mice show similar phenotypes, as they die prenatally due to hypoplasia of mural cells (pericytes and vascular smooth muscle cells, vSMCs) leading to vascular defects and other complex cardiovascular malformations [5,10,11].
The expression pattern of PDGF-D in normal tissue is not well studied, although in adult human tissue, PDGFD mRNA expression has been detected in relatively high levels in adrenal gland, pancreas and heart [3,4].In mice, heart, kidney and lung have been reported as high Pdgfd expressing organs [12].Similar to PDGF-B and PDGFRβ, PDGF-D is described to have a vascular expression pattern, with expression in normal human arterial endothelium and arterial vSMCs [13,14].PDGF-D expression has also been reported in certain epithelial cells [15,16].
Evidence is pointing to an important function of PDGF-D signaling through PDGFRβ in the vasculature, where PDGF-D emerges as a vascular modulator during pathological conditions.In a rat model of chronic rejection, adeno-associated virus mediated transfer of PDGF-D, but not PDGF-B, resulted in accelerated cardiac allograft vasculopathy and fibrosis [17].Heart-specific overexpression of PDGF-D caused vascular remodeling with dilated microvessels and proliferation of vSMCs, resulting in cardiac fibrosis, dilated cardiomyopathy and cardiac failure, whereas overexpression of PDGF-B yielded a much milder phenotype [12].PDGF-D is also capable of stabilizing newly formed vessels through effects on vSMCs and it contributes to neointimal hyperplasia [13,18].Furthermore, in several genome-wide association studies, PDGF-D has been associated with coronary artery disease [19][20][21].
Here we investigate the Pdgfd expression pattern and effects of the PDGF-D deficiency in a new constitutive Pdgfd knockout (Pdgfd -/-) mouse line, containing a LacZ reporter gene allowing visualization of Pdgfd expression sites.We show that PDGF-D deficient mice are viable, and thus offer the opportunity to study the physiological and pathological roles of PDGF-D/ PDGFRβ signaling in adult mice.Moreover, we demonstrate a vascular, predominantly arterial, expression pattern of Pdgfd in all investigated organs.Furthermore, we show morphological changes in NG2-expressing cells within cardiac tissue and demonstrate an increase in blood pressure in Pdgfd deficient mice.Taken together, these findings suggest a regulatory function for PDGF-D in the systemic arterial blood pressure, and furthermore suggest a role in maintenance of vascular homeostasis.

Ethics statement
This study was carried out in accordance with the Swedish legislation of animal welfare.The protocols for this study were approved by the Stockholm North Ethical Committee on Animal Research (ethical permit numbers N187/13, N140/13, N451/12, N127/11, N128/11, N325/11, N140/10, N188/10 and N231/09).All animals used in this study were monitored daily by the animal facility staff.Whenever illness was seen, or suspected, animals were assessed according to the Karolinska Institutet extended assessment checklist for small rodents and rabbits, and euthanized if a total score of more than 0.3 was reached [22,23].Animals were kept in a pathogen free animal facility with controlled temperature and 12h/12h day-night cycle lightning, communally housed in standard cages with enrichment in form of nesting material and had free access to food/water.All invasive procedures were terminal, and performed under deep anesthesia, or after euthanization.For blood collection using heart puncture, animals were sedated by isoflurane gas.For all other experiments involving organ collection, animals were euthanized by intraperitoneal injection of Hypnorm (2,5 mg/ml Fluanisone, 0,07875 mg/ml Fentanyl citrate) Midazolam (1,25 mg/ml), in dH 2 O, 10 ml/kg.Animals that were not assigned to experiments involving necropsy were euthanized by CO 2 followed by cervical dislocation.

Generation, validation and maintenance of Pdgfd null mice
The Pdgfd +/-mouse strain was generated by Regeneron Pharmaceuticals Inc, using the Veloci-Gene technology [24].In the Pdgfd gene 0.1 kb from the ATG start codon to the 3'-end of exon1 was replaced by a LacZ reporter gene cassette generated using a BAC-based targeting vector containing a Neomycin resistance cassette (Ub1 Em7 Neo cassette) flanked by two LoxP sites.Successfully targeted 129S6SvEv/C57BL6F1 ES clones were selected based on neomycin resistance, and screened by a "loss-of-native-allele" (LOA) assay described in [24], using realtime quantitative PCR (qPCR).A Pdgfd-targeted positive clone (5120C-F10) with a LOA score of 0.92 was injected into mouse blastocysts to obtain chimeric mice.F1 heterozygotes males and females were shipped to Karolinska Institutet and backcrossed for >10 generations with C57BL/6J (Jackson Laboratories) to establish a knockout mouse line, in which the Neomycin resistance cassette was not removed.Genotyping was done by conventional PCR of genomic DNA from ear, or tail biopsies to detect the Pdgfd wildtype allele (generating a 289-bp fragment), or the Pdgfd knockout allele (generating a 387-bp fragment), respectively.A forward primer, located upstream (-252 bp) of the ATG start codon within exon 1 of Pdgfd, was used to detect both alleles.The genomic reverse primer to detect the wildtype allele is located 15 bp downstream of the ATG-codon and the cassette primer to detect the inserted allele is located 111 bp downstream of the ATG within the cassette.Absence of Pdgfd exon1 mRNA in Pdgfd -/- mouse hearts was confirmed by quantitative real time-PCR (qPCR) (n = 3 per genotype).All primer sequences are found in S1 Table .The mouse strain was maintained by heterozygous breeding and continuous backcrossing with C57BL/6J mice.A cohort of animals from heterozygous breeding was kept up to the maximum age of 18 months according to our ethical permit.For rapid expansion, offspring from different heterozygous breeding were set up as !5 homozygous breeding pairs per genotype (Pdgfd +/+ x Pdgfd +/+ ; Pdgfd -/-x Pdgfd -/-).First generation animals were used for experiments, but not for further breeding.Pups were counted at birth and surviving animals were then weighed weekly, or once every second weeks for a period of up to 16 weeks.After the 16 weeks monitoring period, animals were assigned to other experimental procedures in order to reduce the number of animals used.

Organ and serum collection
For RNA extraction, organs were snap frozen and further processed as described in the next section (see below).For histology, animals were perfused with dPBS or HBSS (Gibco) followed by 0,2% paraformaldehyde (PFA) in dPBS (for cryosectioning), or in 4% PFA in dPBS (for paraffin embedding) prior to organ collection.Before paraffin embedding, organs were postfixed in 4% PFA for 24 h.

RNA isolation and qPCR
Snap frozen organs were homogenized in TRIzol1 Reagent (Ambion).After addition of 0.2 volumes of chloroform, RNA was extracted using the RNeasy Mini Kit (Qiagen), according to manufacturer's instructions.Genomic DNA was removed using the RNAse-free DNase kit (Qiagen), and cDNA was generated using the iScript™ cDNA Synthesis Kit (Bio-Rad) using a VWR UnoCycler Thermal Cycler.Final RNA concentrations were measured using a Nanodrop ND-1000.qPCR was performed using KAPA SYBR1 FAST Universal 2X qPCR Master Mix (Kapa Biosystems), on a Rotor-Gene Q real-time cycler (Qiagen).L19 was used as reference gene, and the ΔΔCT method was used to calculate relative expression.Pdgfd expression analysis was performed in heterozygotes (n!3 except for bone, cerebrum, eye, kidney, liver, soleus, quadriceps, skin, spleen, thymus and trachea, n = 2, and uterus, n = 1).The expression analysis of Pecam1, Cspg4, Desmin, Notch1 and Gata4 in heart was based on three animals per genotype.The primer sequences are presented in S1 Table .X-gal staining X-gal staining was performed on whole mounts (postnatal day 4 or 3 months old heterozygous animals), or 12-16 μm thick sections (3-6 months old heterozygous animals).Tissues were postfixed in 1-2% PFA in PBS, or in 0.2-0.5% glutaraldehyde PBS (for sections, 10 min; for whole mounts, 20 min) at +4°C and permeabilized with 2 mM MgCl 2 , 0.02% Nonidet P40 substitute, 0.01% deoxycholic acid sodium salt in PBS.Incubations with the X-gal reagent (5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside, Promega) was performed o/n at +37°C.Sections were then either subjected to further staining according to standard protocols below, or mounted.

Immunohistochemistry
For immunohistochemical (IHC) staining of cryosections, antigen retrieval was performed using 0.05% trypsin-EDTA (Gibco, Life Technologies), for 20 min at +37°C.Endogenous peroxidases were quenched by incubation in 0.3% H 2 O 2 in TNT buffer (0.05% Tween 20, 0.1 M Tris pH 7.5, 0.15 M NaCl), for 30 min.Sections were incubated with primary antibody (+4°C, o/n), and secondary antibody (RT, for 30 min), using the TSA indirect kit (NEN Life Science), according to the manufacturers instructions.The signal was developed using 3,3'-diaminobenzidine (DAB Vector Laboratories, Inc.).Sections were either directly dehydrated and mounted with X-tra-kit mounting medium (Medite), or further counterstained with hematoxylin and/or eosin (H&E) according to standard protocol, followed by bleaching by brief dipping into 36 mM HCl in 70% EtOH.
For detection of PDGF-D expression, an affinity-purified polyclonal antibody (anti-PDGF-D rabbit 626) made in-house was used.The antibody was generated by injecting human full-length PDGF-D into a rabbit and affinity-purified as described in [2,3].Paraffin sections were rehydrated and subjected to heat induced antigen retrieval in citrate buffer pH 6.1, (Target retrieval solution S1699, DAKO) for 60 min in steamer prior to the trypsin step, quenching of peroxidases and antibody incubations as described for cryosections.The antibody signal was amplified using the Tyramide Signal Amplification (TSA) kit (PerkinElmer) according to manufacturers instructions.The signal was developed and sections were mounted as described above.

Intravenous injection and visualization of fluorescent tracer
Alexa Fluor1 555-conjugated Cadaverine (Invitrogen) was injected intravenously into the tail vein of adult mice using 0.2 mg/animal in 100 μl dPBS (Gibco), (3 months, n = 4 per genotype), and allowed to circulate for 2-4 h.For in situ detection of tracer, heart and kidney were postfixed in 4% PFA at +4 °C for 4 h.50 μm thick vibratome sections were immunostained with an anti-podocalyxin antibody as described above.To verify successful systemic distribution of injected tracer, kidney was used as positive control.

Imaging of stained tissues
Regular brightfield and immunofluorescence imaging was performed using an inverted AxioObserver Z.1 microscope and equipped with an Apotome unit (Zeiss).The latter was used for immunofluorescence-stained paraffin sections.The X-gal/podocalyxin stainings were imaged with fluorescent and transmitted light simultaneously.Imaging of X-gal/ASMA, PECAM/PDGFRβ, PECAM/NG2 and cadaverine/podocalyxin was performed using an LSM700 (Zeiss) confocal microscope.The blue X-gal precipitate in co-staining with ASMA was captured (as black) using transmitted light.Whole mounts were imaged using a Lumar.V12 with an AxioCam HRc (Zeiss).The software used for acquisition was AxioVision, Zen Black and Zen Blue (Zeiss).Representative images are shown in the figures.
Presented data is obtained from tissues stained for PDGFRβ/PECAM (heart), NG2/PECAM (heart) and NG2/podocalyxin (exocrine and endocrine pancreas) and based on at least three animals per genotype and a total frame number of 20-146.Representative images of podocalyxin-stained endocrine pancreas are found in S3 Fig. Pancreatic islet morphology was analyzed on two sections, at least 200 μm apart, from each animal (n = 4 per genotype).All islets were imaged with a 40x objective and encircled to obtain the total islet area.Whole section area was obtained by mosaic imaging at 5x followed by area measurement.Total insulin per section area, insulin area per islet area and insulin/glucagon pixel ratios were calculated from the parameters above.

Blood pressure measurements
Blood pressure measurements were performed non-invasively using a two-channel CODA Tail cuff High Throughput system (Kent Scientific), as described by the manufacturer.Mice were allowed one week of acclimatization to the equipment to reduce stress.Mice were placed in a restrainer on a heating pad, on day 1-2 for 8-10 min, day 3 for 8-10 min with cuffs on tail, and day 4-5 for 5 min with cuffs on tail followed by 10 measurement cycles (30 s per cycle with 15 s rest in between).On the day of the experiment, the mice were acclimatized 5-10 min, with the equipment in place, prior the measurements.Twenty blood pressure measurement cycles were performed on each animal at the same time each day, for five consecutive days.The first 5 recordings, and any recording showing a tail blood flow below 20 or where the animal had moved during the measurement, were excluded.The experiments were repeated three times with the same magnitude of increase between the groups (n = 4-6 male animals per group, total n = 15 per genotype).

Dry and wet heart weight
Hearts were weighed upon harvest, then dried for 10 days at 56°C and weighed again, using an analytical scale (Sartorius).Pdgfd +/+ n = 15 and Pdgfd -/-n = 16.

Glucose and insulin tolerance tests
Glucose and insulin tolerance tests were performed on age-matched males (20±3 weeks of age).Glucose (2 mg/g BW in 200 μl) was administered to animals that were fasted o/n (12 h), and insulin (0.75 mU/g BW in 100 μl), was administered to non-fasted animals, by intraperitoneal injection, as previously described [25] (n !12 animals per genotype).

Clinical chemistry
Clinical chemistry was performed by the University Animal Hospital, Swedish University of Agricultural Sciences, Uppsala, Sweden.In mouse serum (n = 5 per sex and genotype), ASAT, ALAT, ALP, albumin and calcium were analyzed with spectrophotometry and the electrolytes (chloride, sodium and potassium) with an ion specific electrode on an automatic multi-analyzer Architect 4000c (Abbott Laboratories), all with commercial reagents (Abbott Laboratories).

Statistics
The Mendelian ratio was calculated from 350 pups born from heterozygous breeding pairs and statistics were calculated by chi-square test.All other statistics were calculated by Students ttest (for p 0.05, data was considered significant, Ã p 0.05, ÃÃ p 0.01, ÃÃÃ p 0.001), using either Microsoft Excel or Prism GraphPad.

Pdgfd -/-mice are viable and fertile
To investigate the biological function of PDGF-D, a constitutive Pdgfd knockout mouse strain was generated by replacing exon1 in the Pdgfd gene with a LacZ expression cassette (Fig 1A).Genotypes were confirmed using PCR of genomic DNA (Fig 1B).The lack of Pdgfd exon1 mRNA expression in Pdgfd -/-tissue from heart was verified by qPCR (Fig 1C).In heterozygous breedings of mice on C57BL/6 background, the offspring (Pdgfd +/+ , Pdgfd +/-, and Pdgfd -/-) were born in the expected Mendelian ratio (S3 Table ), and the mice displayed an overall good general health (followed up to 18 months of age).The finding that lack of Pdgfd expression does not generate a severe or lethal phenotype showed that PDGF-D signaling is not essential during embryonic development.To study possible phenotype(s) in the PDGF-D deficient mice, first generation homozygous mice derived from the heterozygous main breedings were set up as new breeding pairs and proved to be fertile.Growth and development of the offspring were studied for 16 weeks, and no gross abnormalities were observed during the study period.However, Pdgfd -/-mice were born at a lower average number per litter, compared with wildtype control litters (

Pdgfd promoter-driven LacZ expression displays a consistent vascular localization
To identify sites of interest for studying PDGF-D function, a general expression analysis was performed.Pdgfd mRNA from a variety of adult tissues was analyzed by qPCR (Fig 2A).The result showed that Pdgfd mRNA was most abundantly expressed in adrenal gland, spinal cord, aorta, heart, uterus, cerebellum and lung.Moderate expression was found in white adipose tissue (WAT), kidney, trachea, testis, eye, red and white muscle and liver, while lower expression was observed in cerebrum, bone, liver, pancreas, skin, intestine, spleen and thymus.
To further determine the expression pattern of Pdgfd, a comprehensive histological analysis was performed.The Pdgfd knockout construct contains the LacZ reporter gene encoding βgalactosidase that enables detection of Pdgfd expression.Through its enzymatic activity the substrate X-gal yields a blue insoluble product in the tissue sections.From here on, when discussing X-gal staining, Pdgfd promoter driven expression of the LacZ reporter gene in heterozygous animals will be denoted as LacZ expression.
Enzymatic staining for LacZ expression on frozen sections from Pdgfd +/-tissues (Fig 2B -2U) showed expression in adrenal gland, aorta, heart, uterus, testis, kidney, liver and pancreas.Corresponding sections from wildtype mice were negative for X-gal staining.The adrenal gland (  1. Furthermore immunohistochemical stainings of wildtype heart, kidney and pancreas, using a polyclonal anti-PDGF-D antibody verified protein expression in most, but not all, structures identified by X-gal staining, while Pdgfd -/-tissues showed no, or only limited, staining at the corresponding sites (S2  In summary, several interesting Pdgfd expression patterns were found, such as strong staining in the vascular system of all tissues investigated, and in various epithelial tissues.The staining intensity varied in the different vascular beds, but the common expression in the blood vessels suggested that PDGF-D might have an important function in the vasculature.A more detailed expression analysis and physiological experiments were therefore performed with special emphasis on the vasculature.

Pdgfd is highly expressed in cardiac vasculature, particularly in arterial bifurcations
The general expression analysis showed that the heart is an organ with high expression of Pdgfd, which is supported by previously published data [3,4,12].
To complement these observations, Pdgfd driven LacZ expression was investigated in whole mount hearts from postnatal (P4) and adult animals (Fig 3A -3C).Both postnatal and adult hearts showed distinct and strong vascular LacZ expression.In postnatal heart, LacZ expression was undetectable in the atrium, while distinct staining was observed in epicardial coronary vessels (Fig 3A).As the heart developed from postnatal to adult, LacZ expression became more widespread throughout the heart, including the atria.LacZ expression in the epicardial coronary vessels became less prominent while smaller vessels displayed a more pronounced LacZ expression (Fig 3B and 3C).Wildtype littermate controls were negative for LacZ expression (Fig 3B -3D).
A more detailed analysis of heart sections stained for LacZ expression showed a widespread expression in the myocardial vasculature (Fig 3E).At higher magnification, the LacZ expression appeared to be strong in medium-sized blood vessels, likely arterioles.Interestingly, the LacZ expression was particularly prominent at vascular bifurcations, especially in the smaller vessel just after the branching point (Fig 3F , arrow).To provide more details of the tissue localization of the staining, X-gal stained sections were counterstained with hematoxylin and eosin.By morphology, the vascular LacZ expression appeared strong in arterioles, particularly at branching points, and weaker as the vessels became smaller (Fig 3G).
In order to classify the blood vessels that express Pdgfd and describe its cellular expression more closely, X-gal stained heart sections were subsequently stained for podocalyxin expression to identify endothelial cells.Intense LacZ expression was observed in arteries (Fig 3H  Confocal images from sections stained for LacZ expression and immunofluorescence-labeled αSMA to visualize vSMCs, verified that predominantly arteries were stained (Fig 3L -3O).Moreover, in smaller αSMA-expressing arterioles, the LacZ expression was often found within the αSMA staining (3M-O, arrows), indicating that endothelial cells express Pdgfd at these sites.
In summary, Pdgfd was abundantly expressed in the heart vasculature, and the expression localized mainly to arteriolar vessels.Moreover, both endothelial cells and perivascular cells, such as vSMCs were found to express Pdgfd.

Pdgfd is predominantly expressed in arterial endothelial cells
To further explore vascular beds expressing Pdgfd, we analyzed postnatal and adult mesenteric tissue for LacZ expression.The mesentery has a well-structured vasculature where arteries, veins and lymphatic vessels are clearly distinguishable.Staining of whole mounted postnatal mesenteric tissue revealed a strong vascular LacZ expression pattern (Fig 4A).In order to differentiate between arteries and veins, co-staining for LacZ expression and αSMA was performed.In postnatal mesentery LacZ expression (black arrows) was predominantly found in arteries, characterized by strong, concentric αSMA staining, while veins appeared negative for LacZ expression (Fig 4B -4D).Interestingly, as previously observed in the heart (Fig 3A -3C), the LacZ expression pattern changed during development into adult tissue.In adult mesentery, LacZ was still strongly expressed by arteries.However, distinct LacZ expression could be found also in veins (Fig 4E -4G).Furthermore, the LacZ expression in adult mesenteric veins was wrapped around the vessels in a typical vSMC-like fashion.The arterial expression was enclosed by the αSMA expressing cells, and thus, likely corresponding to endothelial cells, in both postnatal and adult tissues.
To verify whether the arterial and endothelial expression patterns for Pdgfd could be found in other vascular beds, postnatal skin tissue was analyzed, and indeed, a similar endothelial-like arterial LacZ expression pattern was found (Fig 4H -4J).
Altogether, the histological analysis of LacZ expression in different vascular beds showed that Pdgfd is predominantly expressed in arteries, and that both endothelial cells and vSMCs expressed Pdgfd, implicating a spatially controlled regulation of Pdgfd expression as well as possible region-specific functions of PDGF-D.
NG2-expressing pericytes are disorganized in the cardiac vasculature of Pdgfd -/-mice PDGFRβ signaling is important for proper cardiac development and recruitment of pericytes/ vSMCs to blood vessels [10,11,26], and previous studies have shown that cardiac overexpression of PDGF-D potently modulates both vascular and connective tissue growth [12].To investigate potential alterations in the hearts of Pdgfd-deficient animals, histological analyses were performed on hearts from age-matched wildtype and Pdgfd -/-animals.Hematoxylin and eosin staining showed normal morphology in Pdgfd -/-hearts compared to control hearts (Fig 5A -5D).To further evaluate the vascularization in Pdgfd-deficient hearts, cryosections were stained with antibodies against PECAM and either PDGFRβ, or NG2, and visualized by immunofluorescence.The Pdgfd -/-hearts showed a normal vascular bed with no obvious abnormalities (Fig 5E and 5I).This was confirmed by quantification of vascular density as PECAM pixels per frame (Fig 5H).
To evaluate whether populations of vSMCs/pericytes are missing or defective in Pdgfd -/-animals, expression of PDGFRβ, and NG2 were quantified.In heart sections stained for PDGFRβ, no obvious differences in the staining pattern were found between wildtype and Pdgfd Pdgfd -/-pericytes displayed a rounder morphology, which was perceived as more detached As PDGF-B deficiency is known to gives rise to detached pericytes and leaky vessels [10,11], vessel permeability was investigated by intravenous injections of an inert fluorescent tracer (Alexa Fluor1 555 Cadaverine).No increase in leakage was found in Pdgfd -/-cardiac vasculature compared to control mice (S5 Fig) , suggesting that the decrease in NG2 coverage did not influence the permeability of cardiac vessels.
In summary, the Pdgfd -/-hearts showed a normal vascular network, however pericyte/ vSMC morphology appeared disorganized and vascular coverage by NG2 expressing cells was significantly lower suggesting that PDGF-D might have an important regulatory function for a selected population of mural cells.

Pdgfd -/-mice display mild circulatory defects
To investigate whether the disorganized PECAM/NG2-positive vascular network in the cardiac vasculature of Pdgfd -/-mice has implications on heart function, a gene-expression analysis on agematched wildtype and Pdgfd -/-hearts was performed.A broad spectrum of different immunological, developmental and vascular genes was analyzed.Most of the genes tested were not differentially regulated in Pdgfd -/-mice.However, Pdgfd -/-mice showed reduced mRNA expression of the mural cell markers Cspg4 (NG2) and Des (Desmin) (Fig 6A ), confirming the significantly lower levels of NG2 found in the immunofluorescence stainings and supporting the idea that PDGF-D is needed for correct function/differentiation of a certain population of pericytes/vSMCs.The normal appearance of PECAM staining in the histological analysis was also verified by unaltered mRNA expression of the corresponding gene Pecam1 (Fig 6A).Notably, two additional genes were found to have reduced mRNA expression, namely Gata4 and Notch1 (Fig 6B).GATA4 and Notch1 are known to be important for cardiac development and growth, control of cell fate and maintenance of cardiac function in the adult [27][28][29][30][31]. Since the Pdgfd -/-animals survive, the role of the reduced expression of Gata4 and Notch1 in these mice needs to be further investigated.
To evaluate the cardiac function in Pdgfd -/-animals we analyzed heart growth, by measuring heart wet and dry weights (Fig 6C and 6D).The hearts from Pdgfd -/-animals appeared to be smaller compared with control animals, although the differences were not statistically significant (wet, p = 0.06 and dry, p = 0.09).
Pdgfd -/-mice show normal glucose homeostasis The observed Pdgfd-driven LacZ expression around vascular bifurcations suggested that PDGF-D might regulate blood vessel function, and more specifically, regional blood flow, possibly affecting the access of blood borne nutrients.
To assess possible metabolic changes in Pdgfd -/-mice, the pancreas was analyzed.Similar to heart, both whole mounts and frozen sections from adult Pdgfd The general morphology of Pdgfd -/-pancreas was investigated by hematoxylin and eosin staining and found to be normal compared to wildtype controls (Fig 7E and 7F).To verify whether the alterations in Pdgfd -/-cardiac blood vessel architecture could also be found in other vascular beds, vascularization and pericyte coverage were investigated in Pdgfd -/-pancreas.The visualization of blood vessels with podocalyxin and NG2 revealed no significant differences in Pdgfd To further assess the islet morphology, immunohistochemical stainings for insulin and glucagon, the two major regulators of glucose homeostasis, were performed (Fig 7G and 7H).Quantitative analysis of the stainings showed that the number of islets was unaltered in Pdgfd -/- mice as compared to wildtype controls (Fig 7I).Notably, Pdgfd -/-mice showed a trend towards smaller islets, although the differences did not reach statistical significance (p = 0.21) due to large standard deviations (Fig 7J).Pdgfd -/-mice also showed a trend towards smaller total insulin producing area compared with control mice (p = 0.19), measured as total insulin staining per section area (Fig 7K).However, when measuring individual islets, the area proportion of insulin per islet appeared normal (Fig 7L).Pdgfd -/-mice also had a normal ratio of insulin to glucagon staining within the islet compared to control mice (Fig 7M).
To investigate the function of endocrine pancreas in Pdgfd -/-mice, a classical glucose tolerance test was performed.Pdgfd -/-mice demonstrated a capacity for glucose clearance comparable to wildtype animals (Fig 7N ), suggesting that the pancreatic islets have a normal function.
To exclude that Pdgfd -/-mice have higher insulin sensitivity, they were also subjected to an insulin tolerance test, but showed no alterations in insulin sensitivity compared to controls (Fig 7O).In summary, the Pdgfd -/-pancreas showed a normal morphology and vascular network.Although no alterations were seen in the capacity for glucose clearance, or insulin sensitivity, a trend towards smaller islets was observed in Pdgfd -/-animals.

Discussion
The biological function of PDGF-D has so far been mostly characterized from a pathological perspective.Here, we present a Pdgfd knockout mouse line (Pdgfd -/-), carrying a LacZ reporter gene used to visualize Pdgfd promoter activity.This model is the first of its kind, where the role of PDGF-D can be studied under physiological conditions, both during development and adulthood.In addition, the Pdgfd deficient mice provide a useful, unique tool for mapping of Pdgfd expression by visualizing the LacZ reporter gene activity in mouse tissues.Until now, effective tools for such studies have been lacking.
The basic phenotypic studies of Pdgfd -/-mice showed only minor abnormalities, such as a lower birth rate and slightly reduced survival of first generation offspring from homozygous breedings.These differences may originate from a mild developmental defect in the Pdgfd -/- embryos, altered maternal capability of the Pdgfd -/-females to carry out a pregnancy or nurse after delivery.The lack of a well-defined phenotype in C57BL/6 Pdgfd -/-mice shows that PDGF-D is not crucial for embryonic development or postnatal survival.Thus the Pdgfd -/- mouse line offers the opportunity to further study the physiological and pathological roles of PDGF-D in the adult mouse.
We investigated Pdgfd expression in detail in order to locate relevant sites for biological implications of PDGF-D signaling.Our expression mapping revealed that Pdgfd promoter-driven LacZ expression was seen in all analyzed tissues.Although the Pdgfd-expressing cell types varied between tissues, the vasculature was found to be a common expression site of Pdgfd.
Our earlier studies have shown that PDGF-D specifically binds to, and activates, PDGFRβ [3] Interestingly, our present study clearly demonstrates that PDGF-D has a vascular expression pattern, thereby indicating a particular role of PDGF-D in vascular PDGFRβ signaling.
We also found that Pdgfd deficiency in mice is compatible with life, in contrast to Pdgfb or Pdgfrb deficiency.This important, so far missing, piece of information in PDGF biology illustrates that PDGF-B is sufficient for PDGFRβ signaling during embryonic development, and is in line with previous studies showing that mice lacking PDGFRβ, or the ligand PDGF-B, display almost identical, perinatally lethal, phenotypes, characterized by defects in mural cell recruitment [5].
The spatial coexistence of PDGF-D and PDGF-B suggests a specific physiological role of the PDGF-D/PDGFRβ signaling cascade during adult vascular maintenance and homeostasis.We therefore focused our study on adult animals, and a more detailed analysis of Pdgfd expression in different vascular beds revealed a distinct arteriolar expression, with an especially strong patterning around vascular bifurcations.Additionally, we observed that mainly endothelial cells, but also certain populations of vSMCs/pericytes, expressed Pdgfd, suggesting that PDGF-D may provide both paracrine and autocrine signals to PDGFRβ present on vascular mural cells.
The striking Pdgfd expression around vascular bifurcations indicates a region-specific function for PDGF-D at these sites.Since blood flow into capillaries is regulated locally in the arterioles by vSMC contraction and relaxation [32], it is likely that PDGF-D may regulate vSMC/ pericyte tone at arteriolar branching points, and thereby modulates local blood flow into capillaries.To explore this hypothesis, we performed blood pressure measurements, and notably, an elevation in blood pressure in Pdgfd -/-animals was detected.This might be indicative for an increased vascular resistance in peripheral tissues, which could be caused by perturbations in vSMC actions.However, the changes in systemic blood pressure can also be a result of altered cardiac output, and whether the change in systemic blood pressure observed in Pdgfd -/- mice is caused by changes in peripheral vascular resistance or cardiac output, or both, needs to be further evaluated.
Furthermore, as vSMCs regulate the blood flow from arterioles to capillaries upon local demand of oxygen, glucose and nutrients of the tissue [32,33], any severe alterations of capillary blood flow due to lack of PDGF-D should be reflected in the general metabolism.However, as the Pdgfd -/-mice showed no alterations in glucose clearance from the blood, or in insulin sensitivity, the putative role of PDGF-D is likely more in fine-tuning of the regulation of vSMC/pericyte function.
Analysis of the blood chemistry revealed a modest increase in serum Ca 2+ concentrations in Pdgfd -/-mice.Whether this change in Ca 2+ levels comes from hormonal changes or originates in altered kidney function, and the role of PDGF-D remains to be elucidated.However, disturbed, Ca 2+ homeostasis might contribute to altered regulation of vSMC/pericyte function.Ca 2+ is needed for proper regulation of contraction of vSMCs, cardiomyocytes and other muscle cells [32], and dysfunction within Ca 2+ -dependent pathways may lead to abnormal muscle contraction, which is often seen in cardiovascular diseases [34].
PDGFs biology has been associated with both physiological and pathological regulation of SMCs.For instance, in adult rat, PDGF-B is involved in SMC contraction to regulate interstitial fluid pressure [35], a physiological feature that seems to be controlled also by PDGF-D [18].Further, PDGF-D is described in several studies to be a potent modulator of the adult vasculature and a mediator of SMC phenotypic modulation [12,13,18,36].All these data have, however, either been obtained from PDGF-D overexpression studies or from pathological settings.In this study we provide evidence for a vascular function of PDGF-D during physiological conditions where PDGF-D might be involved in normal regulation of vSMC function.
Given that insufficient PDGFRβ signaling is associated with defects in vSMC/pericyte coverage, this was investigated in Pdgfd -/-mice.In heart tissue we found a defective vascular coverage by mural cells, where vSMCs/pericyte morphology appeared disorganized and not properly attached to the blood vessels, but no vascular leakage was observed.The functional significance of the vSMCs/pericyte changes and whether PDGF-D actually is involved in e.g.regulation of vascular tone need to be further investigated.However, we speculate that this could reflect that only a specific population of vSMCs/pericytes is affected in Pdgfd -/-animals, possibly the same population that regulates vascular tone and blood flow to smaller vessels.
Previous studies have shown that lack of PDGFRβ signaling is related to a complex heart phenotype with defects in ventricular septum formation and development of atrioventricular valves, and might be of importance in congenital heart disease [5,26].Even though no gross malformation was observed in adult Pdgfd -/-heart tissue, we found reduced mRNA expression of Gata4 and Notch1, two genes involved in cardiac development, maintenance of adult cardiac function and control of cell fate and differentiation, as well as in congenital heart disease [27][28][29][30][31].The lower levels of Gata4 and Notch1 mRNA were not further investigated, however, this observation suggests a minor disturbance in the maintenance of Pdgfd -/-heart tissue, which might be connected to the alterations in cardiac mural cell maturation and differentiation.
Notably, in tumor settings, down-regulation of PDGF-D has been shown to decrease expression of Notch-1 [37].
In summary, through detailed characterization of Pdgfd expression, followed by a broad vascular analysis and different physiological experiments, we suggest that PDGF-D may have a role in regulation of arterial blood pressure, and we describe a possible role for PDGF-D in mural cell physiology and regulation of vascular homeostasis.The fact that the PDGF family members, including PDGF-D, are implicated in a range of severe diseases augments the importance of fully understanding the diversity and complexity of signaling derived from the PDGF ligand/receptor system.We have created a Pdgfd-deficient mouse that will serve as an important platform for further studying the role of PDGF-D and PDGFRβ signaling, providing new insights into the functional relevance of PDGF-D signaling in both physiological and pathological conditions.
Fig 1D), and the Pdgfd -/-mice showed a trend towards reduced survival during the first 7 days (Fig 1E).Furthermore, surviving Pdgfd -/-animals were significantly heavier than wildtype controls during the first 3 weeks (Fig 1F), and throughout the 16-week period of observation (S1 Fig).The lack of an obvious phenotype in adult mice indicated that PDGF-D is not of vital importance for postnatal life.

Fig 1 .
Fig 1. Generation, genotyping and general phenotyping of Pdgfd-deficient mice.(A) Schematic outline of the Pdgfd gene with exon numbers indicated.The target construct was inserted into exon1.The locations for genotyping primer binding sites are indicated by left and right arrows below exon1 and construct.(B) Genotyping PCR showing the presence of the three possible genotypes, Pdgfd +/+ , Pdgfd +/-and Pdgfd -/-.The wildtype band is 289 bp, and the knockout band is 387 bp.(C) RT-PCR on heart tissue from Pdgfd +/+ and Pdgfd -/-mice showing the lack of exon1 sequences in Pdgfd mRNA in Pdgfd -/- animals.L19 was used as the reference gene (Pdgfd +/+ n = 3 and Pdgfd -/-n = 3).(D-F) The growth and development of wildtype and Pdgfd -/-animals from homozygote breedings were followed from birth until 16 weeks of age.(D) Observed pups born per litter, from !20 litters per genotype.(E) Survival % of pups born during the first week (Pdgfd +/+ n = 137 and Pdgfd -/-n = 108).(F) Weight curves of surviving offspring.Weight of surviving pups was measured every week (0-3 weeks) (Pdgfd +/+ n!90 and Pdgfd -/-n!80).Error bars indicating standard deviation in (C-D, F). doi:10.1371/journal.pone.0152276.g001 Fig 2B) showed LacZ expression in the fibrous capsule (Fig 2C), as well as in larger blood vessels in the outer adrenal cortex (Fig 2C) and in smaller blood vessels within the medulla (Fig 2D).In the aorta, scattered X-gal staining was observed in the different layers of the aorta (Fig 2E and 2F), with stronger staining found in cells close to the lumen and sparser staining in the outer layers (Fig 2F).In the heart, LacZ expression was generally seen in blood vessels of the myocardium (Fig 2G) and appeared stronger in some, but not all blood vessel bifurcations (Fig 2H).In the uterus, LacZ expression was found mainly in the endometrium (Fig 2I and 2J) although occasional vascular staining was observed.The testis showed expression of LacZ in vascular bifurcations (Fig 2K) and in seminiferous tubules containing spermatids of a more mature state (Fig 2L).In kidney, LacZ expression was observed in the vasculature including the glomerular afferent arteriole, in glomeruli and in certain tubuli (Fig 2M).At higher magnification, LacZ expression revealed a distinct banding pattern in the afferent arterioles, and in the glomerulus, staining appeared to be located in the mesangial cells (Fig 2N).Intense LacZ expression was also observed in the renal papilla (Fig 2O).In the liver (Fig 2P), strong LacZ expression was found in bile ducts (Fig 2Q) and weaker vascular LacZ expression was observed (Fig 2R).In the pancreas (Fig 2S), LacZ expression appeared strong in pancreatic ducts (Fig 2T), and somewhat weaker in blood vessels (Fig 2U).These findings and further description of LacZ expression in additional tissues are summarized in Table Fig).In Pdgfd +/+ heart, PDGF-D expression was detected in coronary arteries and branches (S2A and S2B Fig, arrows) as well as in a pattern corresponding to the capillary X-gal staining seen in the myocardium (S2E and S2F Fig, arrows).In Pdgfd -/-heart tissue weak staining was seen in coronary arteries (S2C and S2D Fig, arrows) while staining in the myocardium was absent (S2G and S2H Fig).In wildtype pancreas, PDGF-D staining was seen in the capillaries of the exocrine pancreas (S2I and S2J Fig, arrows) and in kidney expression was found in certain tubuli (S2M and S2N Fig, arrows).PDGF-D stained tissue from Pdgfd -/-pancreas or kidney showed minor background staining (S2K and S2L and S2O and S2P Fig).The reason that visualization of Pdgfd expression using the LacZ reporter showed a wider expression pattern than the antibody stainings is probably a reflection of the higher sensitivity of the reporter technique, or that the PDGF-D protein has a short half life in the tissues.

Fig 2 .
Fig 2. Pdgfd promoter-driven LacZ expression displays a consistent vascular localization pattern.(A) qPCR analysis on RNA from different tissues from 13 weeks old adult wild-type C57BL/6 mice.Total expression of Pdgfd mRNA measured as % of L19 and normalized to adrenal gland (peak organ).
and 3I), while little expression was found in veins (Fig 3J and 3K, arrows).Furthermore, the expression co-localized with podocalyxin at some sites (Fig 3H, upper arrow), while at branching points strong focal expression was clearly located outside of the podocalyxin staining, likely in the vSMCs surrounding the vessel (Fig 3H, lower arrow).

Table 2 .
Clinical chemistry markers in serum.