The formation of new blood vessels from preexisting ones—angiogenesis—occurs during embryonic development, wound healing, and many pathological conditions, including tumor growth (Adams and Alitalo,2007). Anti-angiogenic drugs are currently being used to treat patients, but blood vessel growth still cannot be modulated therapeutically with a curative result. One possible reason for this is simply that we know neither all key proteins involved, nor their respective role in the angiogenic process. To this end, we have previously identified 58 genes enriched in the vasculature using transcription profiling of vascular fragments (Wallgard et al.,2008). Of these 58 genes, 26 were known regulators of blood vessel growth, including Tie1, Robo4, Eng, Epas1, Notch4, Esam1, EphB4, Fgfr, Pecam, and Vegfr2. However, 32 genes had not previously been implicated in angiogenesis and among them was the gene encoding the Paladin protein (X99384 in mice, KIAA1274 or PALD in humans). Previously, we have also identified Paladin in a reverse genetic screen for genes necessary for developmental angiogenesis, where morpholino knockdown of Paladin in zebrafish embryos led to reduced perfusion of intersegmental arteries (Kalen et al.,2009). Furthermore, Paladin has been identified independently as a gene preferentially expressed in endothelial cells using mRNA expression profiling (Bhasin et al.,2010). Interestingly, Paladin is the only common denominator in the three above-mentioned screens. Apart from being implicated in vascular biology, Paladin has also been identified in a screen for modulation of insulin signaling (Huang et al.,2009). Huang and co-workers showed that Paladin negatively regulates insulin receptor expression and phosphorylation of the insulin receptor, as well as the downstream kinase AKT.
Paladin is a putative cytoplasmic protein with four predicted protein phosphatase (PTP) domains. To unravel the role of Paladin, we mapped the gene- and protein expression during mouse development and human tumor angiogenesis. Paladin is expressed in the vasculature, but is also found in hematopoietic cells and other nonvascular cells. Interestingly, the expression is dynamic and initially detected in developing endothelial cells, whereas in adult tissues it is found predominantly in the mural cells of the vasculature.
Gene Targeting Strategy and Generation of Paladin LacZ Reporter Mice
In mouse, only one Paladin isoform exists, containing 859 amino acids (aa). In humans, there are three isoforms; the full-length protein (856 aa), a long splice variant lacking exon 17 (832 aa), and a short splice variant lacking exons 1–14 (237 aa; Ensembl database, release 65, December 2011; Fig. 1a,b). To map the expression of Paladin we generated a mouse null allele with a functional β-galactosidase (LacZ) reporter. The mutant allele was generated using the VelociGene technology (Valenzuela et al.,2003), where bacterial artificial chromosome-based targeting vectors were used to replace exons 1–18 of the Paladin gene in frame with LacZ followed by a loxP-flanked neomycin selection cassette (Fig. 1a–c). Chimeric mice were bred to C57BL/6 males to produce roman offspring. PaladinLacZ/+ mice were subsequently intercrossed to produce homozygous mutants. Polymerase chain reaction (PCR) genotyping of wild-type, heterozygous, and knockout mice is shown in (Fig. 1d). The presence of Paladin mRNA and protein was investigated in wild-type, heterozygous, and knockout mice. The Paladin mRNA and protein were found to be absent in the knockout mice, while the PaladinLacZ/+ animals had approximately half the amount of Paladin transcripts and protein compared with wild-type littermate controls (Fig. 1e,f).
Paladin LacZ Reporter and Paladin Protein Expression During Mouse Development
First, the LacZ reporter expressed from the mouse Paladin locus was analyzed during embryonic development. Embryos harvested at embryonic day (E) 9.5 and 10.5 were whole-mount stained with X-gal. At E9.5, LacZ expression was detected at the midline, along the entire longitudinal axis of the embryo, and weakly in the intersomitic mesenchyme (Fig. 2a). LacZ expression was also visible as two broad stripes at the hindbrain level corresponding to rhombomere 2 and 4, extending ventrally into the first and second pharyngeal arches (Fig. 2a). In addition, LacZ staining was observed in pharyngeal arch 3 and the heart. There was no apparent LacZ expression in the neural tube at this stage (Fig. 2a). At E10.5, the intersomitic reporter expression had evolved into a distinct striped pattern (Fig. 2b). The blood vessels of the head and the intersomitic vessels displayed clear LacZ expression (Fig. 2c,d). X-gal staining of sectioned E10.5 embryos revealed strong LacZ expression in the roof- and floor plate of the neural tube, cervical nerve trunk, perineural vascular plexus, as well as in blood vessels of the developing brain (Fig. 2e–g). Moreover, we observed weak expression in the notochord and the intersomitic mesenchyme, as well as strong expression in intersomitic vessels. No expression was observed in the somites (Fig. 2g). At the E10.5 stage, we also detected strong expression in the cardinal veins and weak expression in the dorsal aorta (Fig. 2h). The analysis of LacZ reporter expression at E9.5 and 10.5 suggests that Paladin has a weak but general expression in the intersomitic mesenchyme, and that there is a higher level of expression in the intersomitic vessels and other developing small and large blood vessels. In addition, the LacZ reporter indicates that Paladin is expressed in the notochord and the roof- and floor plates of the neural tube. The reporter expression in the roof plate, the staining along rhombomeres 2 and 4, as well as pharyngeal arches 1 and 2, is compatible with expression in neural crest cells.
At E14.5, the overall staining pattern of the Paladin LacZ reporter recapitulated the endogenous mRNA expression pattern previously disclosed by Genepaint (www.genepaint.org), set ID EB1838 (Fig. 3). Furthermore, the weak staining in the mesenchyme was no longer detectable, and notably we observed strong vascular LacZ expression in essentially all vessels, and of all vessel types, i.e., capillaries (Fig. 3b,c,g,h,j) and arteries and veins (Fig. 3e). Strong vascular staining was observed in neural tissue, skeletal muscle, skin, kidney, liver, and along the gastrointestinal tract. In the liver, there was also LacZ expression in a spotty pattern, indicative of expression in solitary cells (Fig. 3i). Nonvascular staining was also detected in the heart myocardium (Fig. 3d), lung and kidney mesenchyme (Fig. 3f,k). Apart from the LacZ expression in vessels, also nonvascular staining was detected in the heart myocardium (Fig. 3d), and lung and kidney mesenchyme (Fig. 3f,k). Expression of the Paladin protein was confirmed using polyclonal serum raised against human Paladin. In general, the polyclonal serum revealed the same pattern of expression as the LacZ Paladin reporter, confirming the validity of the reporter construct. Paladin staining of sectioned wild type E14.5 embryos replicated capillary LacZ expression in the brain (Fig. 3l), myocardium (Fig. 3m,n), lung mesenchyme (Fig. 3o), as well as expression in the kidney glomeruli, mesenchyme, and vasculature (Fig. 3p). With the Paladin antibody, we detected only the full-length protein in mouse tissues (Fig. 1f) and HUVECs (human umbilical vein endothelial cells) in culture. More detailed information about the antibody is found at www.proteinatlas.org and www.atlasantibodies.com.
Paladin LacZ Reporter Expression in the Adult Mouse
Next, we performed X-gal staining of several adult mouse tissues. Distinct vascular expression of the Paladin LacZ reporter was detected in the central nervous system (CNS), kidney, lung, heart, skeletal muscle, white adipose tissue (WAT), brown adipose tissue (BAT), liver, pancreas, and spleen (Fig. 4). In the spleen, a large number of small caliber vessels were LacZ positive (LacZ+; Fig. 4a). Several large vessels (arteries/veins) and arterioles/venules expressed LacZ in the brain (Fig. 4b). LacZ expression was also found in the walls of the lateral ventricles in the brain (Fig. 4c). We did not detect any LacZ expression in brain capillaries (Fig. 4b,c). In the heart, WAT, liver, pancreas, and kidney, only a small fraction of all blood vessels expressed LacZ, and the expression was mainly localized to large vessels (Fig. 4d,f–k), and kidney glomeruli (Fig. 4j). In contrast, skeletal muscle displayed only a minor fraction of small caliber LacZ+ vessels (Fig. 4l). Nonvascular reporter expression was observed in the myocardium (Fig. 4d) and lung mesenchyme (Fig. 4e).
Paladin Expression Shifts From Capillary and Venous Vessels to Arteries During Postnatal Vascular Development
To analyze the vascular expression of Paladin over time, we studied tissues from embryonic to adult stages. To molecularly distinguish arteries from veins, we used mice heterozygous for the nuclear EphrinB2-GFP (green fluorescent protein) reporter, which labels arterial endothelial cells and certain mural cells (Davy and Soriano,2007). At E14.5, Ephrin-B2 was detected in the arterial endothelial cells of the aorta and intercostal blood vessel, but weak expression could also be seen in the aortic vascular smooth muscle cells (vSMC; Fig. 5a). Paladin was highly expressed in endothelial cells in vena cava, and to a lower degree in aortic endothelial cells and vSMC (Fig. 5a). Paladin was also expressed in both intercostal veins and arteries (Fig. 5b).
The highly patterned vasculature of the mouse retina makes it an excellent model system for studies of sprouting angiogenesis and arterial–venous differentiation, as the development of the vasculature is stereotyped and clearly visualized in whole-mount. The arteries and veins are easily distinguished morphologically by vessel diameter, morphology of the branches, vasculature-free area around the arteries, and amount of α-SMA-positive vSMC coating. In the postnatal (P) retina day 5, strong LacZ expression was detected primarily in retinal veins and the expanding capillary plexus, and only weak expression was observed in the arteries (Fig. 5c–e). At P21 and in the adult, the expression pattern of the Paladin reporter had shifted to the large arteries and their principal branches (Fig. 5f–h).
To determine the localization of Paladin expression in adult brain arteries and veins, we used whole-mount brain preparations. To visualize all large vessels of the brain, we perfused adult mice with India ink (Fig. 5j,l). Mice heterozygous for the EphrinB2-LacZ reporter (Wang et al.,1998) labeled the arterial vasculature of the brain (Fig. 5j,m), but not the veins. Similarly, Paladin LacZ reporter expression was detected in the middle cerebral artery and other arteries, but not in the veins (Fig. 5k,n). In addition, Paladin LacZ expression was observed in the arcuate arteries of the adult kidney, but not the arcuate veins (Fig. 5o,p).
Cellular Expression and Subcellular Localization
To reveal the cell types expressing Paladin, we stained mouse tissues with X-gal, the Paladin antibody, or β-galactosidase antibody in combination with markers of endothelial cells, vSMCs and pericytes. Confocal microscopy was performed on brain (E14.5, P5), retina (P5, adult), and kidney (adult) (Figs. 6, 7). During active angiogenesis in E14.5 embryos, Paladin reporter expression was detected in essentially all vessels (Fig. 6a). Staining with X-gal (Fig. 6a,b) or Paladin antibodies (Fig. 6c) overlapped with endothelial markers PECAM1/CD31 or isolectin B4 (Fig. 6a–c). We did not find any overlap with the mural cell marker NG2, even though the NG2-expressing cells were in very close proximity to LacZ-expressing cells (Fig. 6d–f). In the P5 brain and retina there was a dominant endothelial expression of Paladin, and there were only a few vessels that were negative for LacZ (Fig. 6g–j). The endothelial specificity of the expression was confirmed using β-galactosidase antibodies (Fig. 6k). However, at this point in time a few large vessels clearly expressed the Paladin reporter in the vSMC compartment, visualized by staining with either β-galactosidase antibody or X-gal (Fig. 6l,m). This suggests that angiogenic endothelial cells of E14.5 embryos express Paladin, and at later time points the expression is shifted from endothelial cells to vSMC, in certain large vessels.
In the adult brain, the Paladin LacZ reporter was primarily detected in arterial vSMC (Fig. 7a–f). Outside the CNS, e.g., the kidney, Paladin expression was also located to vSMC of large vessels in the adult (Fig. 7g). However, at a few locations in the brain vasculature, weak Paladin expression in the endothelium could still be detected (Fig. 7h, arrow).
To address the subcellular localization of the protein, the human full-length Paladin protein tagged with the V5 epitope, was overexpressed in human dermal fibroblasts. As expected from the Paladin sequence, i.e., the lack of signal peptide and transmembrane domains, the staining indicated a cytosolic localization (data not shown).
Paladin Expression in Cells of Hematopoietic Lineage
In the retina, brain, lung, spleen, and embryonic liver, we found solitary LacZ-positive cells that were not in direct association with blood vessels. mRNA expression profiling has previously been used to show that Paladin is expressed by cultured T- and B-cells (Bhasin et al.,2010). We detected Paladin protein in cell lysates from normal human peripheral blood mononuclear cells (data not shown). Paladin RNA expression was detected using quantitative RT-PCR on flow cytometry-sorted cells of hematopoietic origin obtained from tumor bearing mice, i.e., CD11b+ gr1+, immature myeloid cells (IMC), CD8+ T-cells, CD4+ T-cells, and B220+ B-cells (Fig. 8a). Staining for the hematopoietic cell marker CD45 in combination with X-gal showed that a quarter of the CD45+ cells in the adult lung were LacZ-positive as well (Fig. 8b). In the adult normal lung, solitary LacZ+ cells were also found to be positive for CD68, CD11b or CD4 (data not shown). In addition to endothelial cells, isolectin B4 stains retinal microglia. In the P4 retina, isolectin B4 stained solitary cells with distinct microglial morphology that were LacZ-positive (Fig. 8c). While the embryonic liver contained numerous solitary LacZ-positive cells, not associated with vessel structures, these were essentially lacking in the adult liver (Figs. 8d, 4f). This fits with a hematopoietic expression of Paladin, as the embryonic liver is a site of hematopoiesis, while the adult liver is not.
Paladin Is Expressed in Human Tumor Vasculature
To analyze Paladin expression in malignant human tissue, we stained customized tissue microarrays (TMA) containing normal adult human brain tissue as well as human brain tumors of astrocytic origin with polyclonal anti-Paladin or anti-CD34 antibodies. As expected, the endothelial marker CD34 (Baumheter et al.,1993) was detected in blood vessels in the normal human brain samples (Fig. 9b), as well as in tumor tissue, including astrocytoma and glioblastoma (Fig. 9f,h,j). The Paladin antibody generated a staining pattern similar to that of CD34 in normal human brain, as well as brain tumor endothelium (Fig. 9a,e,g,i). Certain larger vessels expressed Paladin in mural cells (Fig. 9c arrows), in addition to the endothelial staining (Fig. 9c). As expected, CD34 was only expressed in endothelial cells (Fig. 9d), and not in vSMC (Fig. 9d). In addition, in normal human brain tissue we detected Paladin expression in solitary cells with multiple slender processes, which could suggest that these cells are microglia (Fig. 9).
In angiogenesis, the importance of kinase signaling is well established, both at the receptor level as well as within the cell, reviewed in (Olsson et al.,2006). The termination of kinase signaling is clearly essential and therefore phosphorylation of target proteins is balanced by protein de-phosphorylation and degradation. There are several ubiquitous protein phosphatases but currently only one vascular specific protein phosphatase is known; vascular endothelial-protein phosphatase (VE-PTP). This protein is a transmembrane tyrosine phosphatase essential for normal angiogenesis and embryogenesis (Dominguez et al.,2007). Hence, it is of great interest to identify and characterize the roles of additional phosphatases selectively expressed in the vasculature. To this end, we have shown that Paladin, a putative protein phosphatase, is a cytoplasmic protein expressed in the vasculature of developing and adult mouse tissues, as well as in normal human and brain tumor vessels. We identified Paladin in two independent screens (Wallgard et al.,2008; Kalen et al.,2009). In one approach, Paladin was among the 58 genes identified as a core set of genes with broad and general vascular expression in the mouse (Wallgard et al.,2008). In this set, 26 genes were known regulators of the vascular biology. Despite the great interest in the genetic regulation of angiogenesis, we found—much to our surprise—that only a few genes out of the 26 had been thoroughly characterized with regards to their expression pattern, i.e., Ptprb (VE-PTP), EphB4, CD31, VE-cadherin, Vegfr2, Alk1, endoglin, Notch4, Robo4, Tie1, and Nrp1. To examine the expression of Paladin, we used antibodies against Paladin as well as a novel targeted mouse allele, harboring a LacZ reporter gene.
Analysis of gene expression based on a LacZ reporter has its pros and cons. In general, it is a sensitive and very specific method for detecting gene activity, but there is always the risk that the gene deletion will affect promoters and therefore not recapitulate the true mRNA expression pattern of the endogenous gene. In addition, the half-life of the LacZ protein is long and therefore quick on/off changes in gene expression cannot be tracked. In light of this, we have compared the LacZ reporter expression with the mRNA in situ hybridization of Paladin as disclosed by Genepaint. At E14.5, the mRNA in situ hybridization and LacZ expression patterns were essentially the same. In addition, we have used polyclonal antisera against human Paladin, which also cross-reacts with the mouse protein, to stain mouse tissues. Comparing the LacZ reporter expression and Paladin protein staining in embryos at E14.5 and in the brain at P5 resulted in almost identical staining patterns. Therefore, we conclude that the Paladin LacZ reporter faithfully recapitulates Paladin mRNA and protein expression during embryogenesis and early postnatal development.
The antisera against Paladin was generated by the Human Protein Atlas project, which has been set up for the systematic generation of specific antibodies on a global scale combining high-throughput generation of affinity-purified polyclonal antibodies with immunohistochemistry-based protein profiling on tissue and cell microarrays (Ponten et al.,2011). All protein expression data, are published on the free and publically available Human Protein Atlas portal (www.proteinatlas.org; Uhlen et al.,2005,2010). Within this effort, two different antibodies toward two nonoverlapping epitopes of the human Paladin protein have been generated (HPA015696 and HPA017343). Stainings using the anti-Paladin antibody HPA015696 clearly display a better signal- to background ratio compared with those of HPA017343. Therefore, we analyzed only images resulting from the use of HPA015696. There was a distinct vascular staining in brain, lung, kidney, pancreas, placenta, and skin. However, the staining was mainly localized to endothelial cells, even though a subset of vSMC was positive as well. Notably, the endothelial staining was detected in a larger number of both capillaries and larger vessels as compared to the mouse Paladin LacZ reporter. Nonvascular staining was observed in cardiomyocytes, lung macrophages, microglia in the CNS, and in cells in the walls of the lateral ventricles of the brain. In comparison to the LacZ reporter expression in mouse, analysis of the human tissues stained in the Human Protein Atlas showed additional staining in intestinal and kidney tubular epithelium, neuronal cells in the cerebral cortex, as well as occasional staining in the pancreatic islets of Langerhans, and myocytes in striated skeletal muscle. We conclude that the mouse and human gene and protein expression seem to overlap to a large degree. However, the antibody staining on human tissue suggests a broader expression of the human protein, compared with the mouse gene and protein, both within and outside the vasculature. In addition, the human Paladin staining in adult tissue is predominantly localized to endothelium, in contrast to the staining in mouse. Given that we have verified the Paladin LacZ reporter expression in mouse by both the use of Paladin antibody staining and comparison to mRNA in situ hybridization, we would argue that the difference observed between mouse and man reflect to true differences in gene and protein regulation, rather than technical aspects of the methods used.
In the characterization of the mouse expression pattern of Paladin, we were struck by its dynamic expression during vascular development; a predominant expression starting in the endothelial cells of primitive capillary plexa and veins, but over time shifting to pericytes and vSMCs of arteries in the mature vasculature, summarized in Table 1. To our knowledge, this type of expression pattern has not been described in detail previously. Among the 26 known vascular regulators pertaining to the core cluster, there are at least three other genes that are expressed in both endothelial and mural cells; Ephrin-B2 (Gale et al.,2001), Notch4 (Lindner et al.,2001), Alk1 (Yao et al.,2007), and the adrenomedullin co-receptor Calcrl (calcitonin like receptor-like; Kamitani et al.,1999). However, the dynamics of their expression pattern is not known. Furthermore, when taking genes outside of the core cluster into account, there are additional vascular enriched genes expressed in both endothelial and mural cells, e.g., CD13 (aminopeptidase N; Rangel et al.,2007), and Gpr124 (Kuhnert et al.,2010). Thus, vascular genes are expressed; (i) in both endothelium and mural cells (Paladin, Gpr124), (ii) predominantly in endothelium (CD31, VE-cadherin) or (iii) preferentially in smooth muscle/pericyte (α-SMA, RGS5). Interestingly, Paladin—and likely other genes—shifts expression from one cell type to another over time. The transcriptional regulation of these three basic expression patterns will be different. For the endothelial specific expression, Black and co-workers showed that the combination of two transcription factors (not expressed selectively in endothelial cells themselves) led to a synergistic activation of endothelial specific transcription (De Val et al.,2008). The transcriptional control of genes shifting their expression from endothelial to mural cells has yet to be elucidated. To unravel this mechanism, it will be important to further characterize the gene expression of other vascular enriched genes.
Table 1. Summary of Paladin Expression in the Vasculaturea
Regarding the function of Paladin, Huang and co-workers have previously shown that Paladin negatively regulates insulin receptor abundance, signaling and AKT phosphorylation (Huang et al.,2009). Even though we could not detect a Paladin LacZ reporter expression in the islets of Langerhans of the mouse, it is possible that the knockout construct has deleted essential promoter regions for this expression, or that there is expression differences between mouse and man. Now, knowing that the expression pattern of Paladin is abundant in the vasculature in late embryogenesis and adulthood, one could speculate that one additional function of Paladin is to regulate insulin signaling in the vasculature. Furthermore, we have previously shown that, in the zebrafish, Paladin is essential for normal vascular development and vascular perfusion (Kalen et al.,2009). Therefore, it is possible that Paladin has a broader role in regulating receptor tyrosine kinase and AKT phosphorylation in the vasculature. Even though Paladin has four predicted S/T/Y phosphatase domains and negatively regulates insulin receptor tyrosine kinase and AKT phosphorylation Huang and co-workers could not confirm phosphatase activity of the protein (Huang et al.,2009). The members of the PTP superfamily can be divided into four main subfamilies, type 1 through 4, based on their amino acid sequence and presence of functional domains, reviewed in (Alonso et al.,2004) and (Mustelin,2007). The human full-length Paladin gene displays a weak homology to DUSP23 (dual specificity phosphatase 23), which belongs to PTP subfamily 1. DUSP23 has been reported to enhance MAPK signaling independent of the catalytic site, through acting as a scaffold and enhancing MAPK binding to JNK and p38, reviewed in (Takagaki et al.,2004) and (Wu et al.,2004). Another catalytically inactive phosphatase-domain containing protein, with a role in endothelial biology is the His-domain-containing protein tyrosine phosphatase (HD-PTP). HD-PTP is a classic nontransmembrane PTP, expressed broadly, and shown to regulate endothelial migration through interaction with Src kinase and modulation of FAK phosphorylation (Castiglioni et al.,2007, 2009). HD-PTP is catalytically inactive due to incorrect aa-sequence in the PTP domain (Gingras et al.,2009). This suggests that there might be several catalytically inactive PTPs that have crucial functions in vascular biology. It will be an important task to determine if Paladin is one of these catalytically inactive phosphatase-like proteins, and how Paladin as well as other phosphatase-like proteins, affect signaling and angiogenesis.
The Paladin gene was targeted using the VelociGene technology as described before (Valenzuela et al.,2003). Shortly, coding exons 1–18 were used for bacterial homologous recombination approach to incorporate the reporter cassette containing a β-galactosidase gene (LacZ) and a Neomycin selection cassette, flanked by loxP sites. Targeted ES cells were used for production of chimeric mice, which were subsequently bred to gain the heterozygous mice used in this study. Genotyping was performed using the HotStarTaq Plus Master Mix Kit (Qiagen), and Paladin universal reverse primer (AACCAGAAACCCTAGGAAGG), Paladin wild-type forward primer (GGTGGTGATGAAGGTGGTACA), and Paladin knockout forward primer (TCATTCTCAGTATTGTTTTGCC).
Mice heterozygous for ephrin-B2 and either the GFP or LacZ reporter gene were used as previously described in (Davy and Soriano,2007; Wang et al.,1998).
All animal experiments were performed in compliance with the relevant laws and institutional guidelines and were approved by local animal ethical committees in Sweden and the United Kingdom.
Quantitative Real-Time PCR
RNA was isolated from retinas (n = 5 per genotype) and cells of hematopoietic origin (n = 3 per cell type) using the RNeasy Micro Kit (Qiagen). Homogenization was performed using rotor-stator homogenization, followed by disruption using a shredder column (Qiagen) and on-column DNase digestion. Samples were quality controlled on an Agilent 2100 Bioanalyzer using the Agilent 2100 Expert software (version B.02.02.SI238) and the Agilent RNA 6000 Nano Kit (Agilent). In vitro transcription reactions were performed using the SuperScript III First-Strand Synthesis Kit for real-time (RT)-PCR (Invitrogen) according to the manufacturer's instructions. Three RT quantitative PCR reactions/TaqMan assays were carried out from every in vitro transcription reaction using TaqMan Universal PCR Master Mix (Applied Biosystems) and the StepOnePlus Real-Time PCR System (Applied Biosystems), quantitative application, according to manufacturer's manual. The TaqMan Assays (Applied Biosystems) used were Gapdh (4352932E) and X99384 (Mm00600187_m1 and Mm00600181_m1) targeting Paladin exon boundaries 7–8 or 19–20, respectively.
Lungs were collected from wild-type, heterozygous, and knockout mice at postnatal day 5. The lungs were homogenized using a Tissue Tearor in lysis buffer (20 mM Tris, 0.1% Triton, 0.1% sodium dodecyl sulfate, 50 mM NaCl and 2.5 mM ehtylenediaminetetraacetic acid (EDTA), 1 nM Na3VO4 and ×1 protease inhibitor cocktail [Roche]). Total protein lysates were denatured in sample buffer and reducing agent (Invitrogen). The samples were separated on a bistris 8–12% gel (Invitrogen). Proteins were transferred to a hybond-C extra membrane (Amersham Biosciences). Unspecific binding to the membrane was blocked using 5% skimmed milk in TBS 0.1% Tween. Primary rabbit anti-KIAA1274, HPA017343 (Atlas Antibodies) or goat anti-actin (Santa Cruz Biotechnology) antibody was added in blocking buffer in a 1:1,000 dilution overnight. Membranes were washed in TBS 0.1% Tween and horseradish peroxidase (HRP) conjugated anti-rabbit (GE healthcare) or anti-goat secondary (DAKO) antibodies were added at 1:4,000 dilution in blocking solutions. Membranes were washed in TBS 0.1% Tween and developed using ECL plus (GE healthcare).
Whole-mount X-gal Staining and Immunohistochemistry
X-gal staining of embryo, retina, and brain whole-mounts was performed according to previously described protocols (Hogan et al.,1994). Shortly, embryos were dissected and fixated in paraformaldehyde (PFA) as follows: E9.5 in 0.2% PFA overnight at 4°C; E10.5 in 0.2% PFA 6 hr at room temperature; E11.5 in 0.2% PFA overnight at 4°C, and E14.5 in 4% PFA, 5 hr at 4°C. P5-adult pups were killed and eyes were harvested, fixated in 4% PFA in phosphate buffered saline (PBS) at 4°C for 45 min and washed in PBS, then the retinas were dissected. Embryos and retinas were permeabilized and incubated in X-gal (1 mg/ml; Promega) diluted in staining solution at 37°C, protected from light. After that, the staining pattern of the embryos was documented. Retinas were subsequently stained with isolectin B4 as described previously (Hellstrom et al.,2007), and with various antibodies. Briefly, retinas were blocked in 1% bovine serum albumin (BSA), 0.5% Triton X-100 in PBS, washed twice in PBlec (1% Triton X-100, 0.1 mM CaCl2, 0.1 mM MgCl2, 0.1 mM MnCl2 in PBS at pH 6.8), and incubated in biotinylated isolectin B4 followed by staining with streptavidin conjugated to Alexa Fluor488. The retinas were then stained with either anti-α-SMA antibodies conjugated to Cy3 diluted 1:500 in 0.5× blocking buffer, or rabbit anti-NG2 antibodies diluted 1:200 in blocking buffer at 4°C overnight, followed by staining with goat anti-rabbit antibodies conjugated to Alexa Fluor 568 diluted 1:100 in 0.5× blocking buffer. Retinas were finally flat-mounted in ProLong Gold Antifade (Invitrogen). Adult brains were dissected and fixated in PFA in 4% PFA for 1 hr at 4°C, then permeabilized and stained with X-gal. India ink perfusion was performed as described previously (Leypoldt et al.,2009). Shortly, animals were anesthetized and perfused with PBS followed by 4% PFA and India ink (50% India ink, 5% gelatin in PBS). After incubation at 4°C overnight, brains were dissected and photographed using a stereo microscope.
X-gal Staining and Immunohistochemistry of Frozen Sections
Embryos were dissected and fixated in PFA as follows: E9.5 in 0.2% PFA overnight at 4°C; E10.5 in 0.2% PFA 6 hr at room temperature; and E14.5 in 4% PFA, 5 hr at 4°C. A small cut was made in the neck of the E14.5 embryos to facilitate penetration of liquid. Tissues from adult mice were collected and fixated either in 4% PFA for 4 hr at room temperature (kidney, heart, white adipose tissue, spleen, lung, liver), overnight (brain, interscapular brown adipose tissue), or in 0.4% PFA for 4 hr at room temperature (skeletal muscle, pancreas). Embryos were collected at day E10.5 and E14.5 and fixated as stated above. A small cut was made in the neck of the E14.5 embryos to facilitate penetration of liquid. Tissues and embryos were incubated in 30% sucrose and 2 mM MgCl2 in PBS overnight and 10 or 14 μm (E10.5 only) and frozen sections were obtained as described previously (Hogan et al.,1994). LacZ staining of sections was performed according to previously described protocols (Hogan et al.,1994). Shortly, sections were post-fixated in 0.2% PFA, permeabilized, and then stained with 1 mg/ml X-gal (Promega) diluted in staining buffer for 17–22 hr at 37°C protected from light. Frozen sections were then stained for CD31 (1:500), NG2 (1:200), CD45 (1:500), and/or α-SMA (1:100). For fluorescent secondary detection, Alexa Fluor 488, -555, or -568 conjugated secondary antibodies were used, diluted 1:200 in 1.5% BSA, 0.05% Tween-20. For chromogenic detection, sections were incubated in biotinylated secondary antibodies (1:200) diluted in 0.5× blocking buffer with 5% NDS, 5% normal mouse or goat serum (Invitrogen) for 1–2 hr at room temperature. Next, sections were incubated in Streptavidin-HRP (SA-HRP; 1:200; Vector labs) diluted in 0.5× blocking buffer for 1 hr at room temperature, and then incubated 5–10 min in sterile filtered DAB solution (3′3′-diaminobenzidine; 1 DAB tablet (Sigma) dissolved in 15 ml 1× TBS and 12 μl H2O2 (30%) before mounting.
Immunohistochemistry of Paraffin Sections
Adult and postnatal day (P) 5 brains and E14.5 embryos were harvested and fixated in 4% PFA overnight at 4°C. Tissues were then dehydrated and infiltrated and embedded with paraffin as previously described (Hogan et al.,1994). Paraffin sections (5 μm) were then deparaffinized and heat induced epitope retrieval (Pressure cooker, Decloaking Chamber, Biocare Medical, Histolab) in Citrate buffer pH6 (S2369, Dako) was performed before blocking with 3% BSA, 0.1 Tween, 5% normal donkey serum (NDS) in PBS or PBlec (only before isolectin B4 staining) for 1 hr at room temperature. Before staining with Paladin or β-galactosidase, sections were additionally blocked using a streptavidin/biotin blocking kit from Vector Laboratories. Sections were then stained with Paladin (HPA015696, Atlas Antibodies; 1:50, β-galactosidase (1:400), α-SMA, CD31, anti-GFP (E14.5 Ephrin-B2-GFP only), and/or with isolectin B4 diluted in 0.5× blocking buffer containing 5% NDS or PBlec (only isolectin B4) overnight at 4°C. After staining with Paladin or β-galactosidase a signal enhancement step was introduced (sections stained with any other antibodies continued to directly to secondary labeling): biotinylated anti-rabbit antibody diluted in 0.5× blocking buffer containing 5% NDS and 5% normal mouse serum (NMS) was incubated for 2 hr at room temperature. Signal detection was done with streptavidin (SA)-biotin system using SA-conjugated horse radish peroxidase (1:200; Vector Labs) followed by a tyramide-biotin enhancement step according to manufacturer's instructions (Molecular Probes,) and binding with Alexa Fluor 555 conjugated SA (1:200; Molecular Probes). Sections were then washed twice in PBlec and co-stained with fluorescein isothiocyanate (FITC) -conjugated α-SMA (1:250; Sigma), or isolectin B4 (1:50; Sigma). If applicable, streptavidin/biotin block from Vector Laboratories were used before staining.
A tissue microarray (TMA), containing duplicate tissue cores (1 mm diameter) from 120 different specimens representing normal human brain tissue and a multitude of various brain tumors was generated and used for immunohistochemistry-based protein profiling as previously described (Paavilainen et al.,2009). The human tissue sections and tumors were obtained from patients diagnosed at the Department of Genetics and Pathology, Uppsala University Hospital, Uppsala, Sweden, in agreement with approval from the Research Ethics Committee at Uppsala University. Histopathologic diagnoses and grade were carefully reevaluated (including use of additional immunohistochemical stainings when needed) by two neuropathologists (Tommie Olofsson, Hannu Kalimo) to get representative samples for the TMA according to the World Health Organisation criteria (Louis et al.,2007). For the purpose of this study, only cores representing grade IV glioma (glioblastoma, N = 11 and gliosarcoma, N = 3), grade III glioma (anaplastic astrocytoma, N = 7 and oligoastrocytoma, N = 5), grade II glioma (diffuse astrocytoma, N = 7), and nonmalignant control brain tissue (gliosis N = 2, control white and gray matter, N = 3) were taken into account. The Paladin antibody HPA015696, and anti-CD34 antibody (1:100, Dako, M7165BD) were used for immunohistochemical staining. In brief, slides were deparaffinized in xylene and rehydrated in a descending ethanol series. A total of 0.3% H2O2 was added to the 95% ethanol step to block endogenous peroxidase activity. Slides were boiled in epitope retrieval buffer (Thermo Scientific, Waltham, MA) in a pressure boiler (Biocare Medical, Walnut Creek, CA) at 125°C for 4 min and subsequently cooled to 90°C for approximately 30 min. IHC was performed using an Autostainer 480 instrument (Lab Vision, Fremont, CA) and 3′3′-diaminobenzidine (DAB) as substrate. Slides were counterstained with hematoxylin, mounted using a Leica Autostainer XL (Leica Microsystems), and scanned using a ScanScope XT system (Aperio Technologies, Vista, CA) at ×20 magnification.
FACS Sorting of Hematopoietic Cells
F9 teratocarcinoma cells (ATCC, Manassas, VA) were maintained in Dulbecco's modified Eagle's medium (DMEM, Invitrogen, Carlsbad, CA) supplemented with 10% FCS (fetal calf serum; Sigma-Aldrich, St. Louis, MO). For tumor inoculation, male 129S6/SvEvTac mice were anesthetized by isoflurane inhalation (Forene, Abbott Laboratories, Abbott, IL) and 106 F9 cells were injected subcutaneously into the left flank. On day 13 after tumor inoculation, spleens were collected and passed through a 70-μm cell strainer to prepare a cell suspension. Erythrocyte lysis was performed using ACK buffer (10 mM KHCO3, 150 mM NH4Cl, 0.1 mM EDTA, pH 7,4). Cells were washed in PBS + 1% FCS and passed again through a 70-μm cell strainer before labeling with 1 μg/ml directly conjugated antibodies (FITC-conjugated Gr-1, PE conjugated CD11b, FITC-conjugated CD8, PE conjugated CD4, PE conjugated B220 [all BD]) for 30 min at 4°C. Cells were subsequently washed and resuspended in PBS + 1% FCS. Directly before sorting of individual populations, cells were passed through a 40-μm strainer and DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride; Sigma) was added for discrimination of dead cells. For sorting, a FACS Vantage Cell Sorter (BD) was used. Tumors were not used in this study.
The following commercial antibodies were used: rabbit anti-β-galactosidase (ICN/Cappel), rabbit anti-NG2 (Chemicon), rat anti-CD31 (BD Biosciences), anti-α-SMA antibodies conjugated to FITC (Cy3 (Sigma), biotinylated or FITC-conjugated isolectin B4 from Bandeiraea simplicifolia (Sigma), anti-rabbit biotin (Invitrogen), anti-KIAA1274, HPA015696 (Atlas Antibodies), anti-KIAA1274, HPA017343 (Atlas Antibodies) FITC-conjugated anti-GFP (Invitrogen), goat anti-mouse CD45 (R&D), FITC conjugated Gr-1 (BD), PE conjugated CD11b (BD), FITC conjugated CD8 (BD), PE conjugated CD4 (BD), PE conjugated B220 (BD),and Alexa Fluor 488, -555, or -568 conjugated secondary antibodies or streptavidin (Molecular Probes). Hoechst (Invitrogen) and DAPI (Invitrogen) were used for nuclear staining.
Dr. Carina Hellberg at the Ludwig institute of cancer research, Uppsala University, who kindly provided immortalized Human foreskin fibroblasts. Neuropathologist Hannu Kalimo for histopathologic diagnoses and grade reevaluation of human TMA:s according to WHO criteria. Ross Smith, Department of Immunology, Genetics, and Pathology, Uppsala University, for proof reading.