|Home | About | Journals | Submit | Contact Us | Français|
The primitive foregut is patterned in a manner that spatially promotes proper organ specification along the anterior-posterior foregut axis. However, the molecular pathways that specify foregut endoderm progenitors are poorly understood. We show that Wnt2/2b signaling is required to specify lung endoderm progenitors within the anterior foregut. Embryos lacking Wnt2/2b expression exhibit complete lung agenesis and do not express Nkx2.1, the earliest marker of the lung endoderm. In contrast, other foregut endoderm derived organs including the thyroid, liver, and pancreas are correctly specified in Wnt2/2b null animals. We show that this phenotype is recapitulated by an endoderm restricted deletion of β-catenin, demonstrating that Wnt2/2b signaling through the canonical Wnt pathway is required to specify lung endoderm progenitors within the foregut. Moreover, activation of canonical Wnt/β-catenin signaling results in reprogramming of esophagus and stomach endoderm to a lung endoderm progenitor fate. Together, these data reveal that canonical Wnt2/2b signaling is uniquely required for specification of lung endoderm progenitors in the developing foregut.
The vertebrate gut tube is patterned such that organs are specified in a precise spatial location along the anterior-posterior axis of the developing embryo. Signaling molecules expressed in the surrounding lateral plate mesoderm (LPM) are thought to promote development and patterning of these organs including the thyroid, lung, liver, and pancreas in part through proper specification of endoderm progenitors. Although several important signaling pathways have been implicated in the regulation of foregut endoderm development the pathways that uniquely specify the lung within the anterior foregut are unknown.
Wnt signaling is one such pathway known to be important for early tissue morphogenesis. Multiple roles for β-catenin in cell proliferation and differentiation have been reported in the endodermal components of multiple tissues including the liver, pancreas, and lung (Apte et al., 2007; Dessimoz et al., 2005; Mucenski et al., 2003; Murtaugh et al., 2005; Shu et al., 2005; Tan et al., 2006). However, whether Wnt signaling plays a role in specification of foregut derived tissues remains unclear. In this report, we show that Wnt2 and Wnt2b play an essential and cooperative role in specifying lung endoderm progenitors within the anterior foregut without affecting the specification of other foregut derived tissues. Moreover, we show that activation of Wnt/β-catenin signaling can reprogram posterior endoderm to a lung progenitor fate indicating the potent role of Wnt signaling in specifying early lung endoderm progenitors. Thus, our studies reveal a unique role for Wnt/β-catenin signaling in promoting lung endoderm specification in the foregut.
Previous studiers have reported the expression of several Wnt ligands in the lung including Wnt2, Wnt5a, Wnt7b, and Wnt11 (Li et al., 2002; Weidenfeld et al., 2002). Given the importance of the Wnt pathway in endoderm development, especially in regulation of tissue specific progenitors, we explored whether these ligands are expressed in the appropriate spatial and temporal pattern to regulate specification and development of lung endoderm progenitors in the foregut. These studies revealed that Wnt2 and Wnt2b are expressed in the mesoderm surrounding the ventral aspect of the anterior foregut from E9.0–E10.5 during the period when the lung is specified (Supplemental Figure 1). Wnt2 and Wnt2b are expressed later in the developing lung mesenchyme with Wnt2 expression persisting into adulthood (Supplemental Figure 1 and (Monkley et al., 1996; Zakin et al., 1998)). These data suggest that Wnt2/2b are expressed in the appropriate spatial and temporal manner in the LPM to regulate lung specification and development.
To further investigate the role of Wnt2 and Wnt2b in development of the anterior foregut, we generated null alleles in mice of both genes using homologous recombination in embryonic stem (ES) cells (Supplemental Figure 2). The majority of Wnt2−/− null mutants are cyanotic at birth and die within a few minutes whereas Wnt2b−/− null mutants are viable with no discernable phenotype (Supplemental Table 1 and data not shown). To explore the reason for perinatal lethality in Wnt2−/− null mutants, histological analysis was performed on embryos from E10.5–E18.5. These studies revealed significant lung hypoplasia in Wnt2−/− null mutants (Fig. 1A–H). Despite the hypoplasia, branching of the terminal airways was relatively normal in Wnt2−/− null mutants (Fig. 1I–L). Poor development of the lung mesenchyme was observed leading to a dilated and dysfunctional vascular endothelial plexus by birth (Fig. 1M–P). Cell proliferation is significantly reduced in both epithelial and mesenchymal cell lineages in Wnt2−/− lungs (Fig. 1Q–S). Several signaling pathways and transcription factors known to be important for lung growth and differentiation including Fgf10, Nkx2.1, Bmp4, N-myc, and cyclin D1 were significantly reduced in Wnt2−/− null mutants (Fig. 1T and (Eblaghie et al., 2006; Kimura et al., 1996; Okubo et al., 2005; Sekine et al., 1999; Zhang et al., 2008)). In contrast, proximal-distal patterning was unperturbed in Wnt2−/− null mutants as noted by normal expression of SP-C, a marker of distal alveolar epithelial cells, and CC10, a marker of proximal bronchiolar epithelial cells (Fig. 1U–X).
To address the combined role of Wnt2 and Wnt2b in lung development, we generated Wnt2/2b double knock-out (DKO) mutants. Examination of Wnt2/2b DKO mutants revealed complete lung agenesis (Fig. 2A–L). While the esophagus is readily apparent in wild-type embryos at E11.5 and E14.5, neither lung nor tracheal development could be found in Wnt2/2b DKO mutants (Fig. 2A–L). To determine whether the lung endoderm lineage was specified within the anterior foregut, we assessed the expression of Nkx2.1, a homeobox transcription factor that is the earliest marker of the developing lung endoderm (Kimura et al., 1999; Minoo et al., 1995). Nkx2.1 expression is first observed by immunohistochemistry and in situ hybridization at E9.5 and by Q-PCR at E8.5 in the ventral aspect of the foregut demarcating where the trachea will bud off of the anterior foregut (Fig. 2M and N and (Kimura et al., 1996; Serls et al., 2005; Yuan et al., 2000)). Nkx2.1 expression was absent in the anterior foregut region of Wnt2/2b DKO mutants confirming the loss of tracheal and lung development (Fig. 2N and R and Supplemental Figure 3). Q-PCR confirms loss of Nkx2.1 expression in anterior foregut (Supplemental Figure 4). In contrast, Nkx2.1 expression was observed in the thyroid primordium of both wild-type and Wnt2/2b DKO mutants (Fig. 2O and S). Specification of foregut endoderm was not lost as demonstrated by expression of Foxa2 in Wnt2/2b DKO mutants nor was cell proliferation or apoptosis affected in the anterior foregut endoderm of Wnt2/2b DKO mutants (Supplemental Figure 4). The esophagus was specified normally in Wnt2/2b DKO mutants as determined by expression of p63, a marker of esophageal endoderm (Fig. 2P and T). Expression of Wnt7b, an additional marker of early lung endoderm progenitors in the anterior foregut (Shu et al., 2002), was also lost in Wnt2/2b DKO mutants (Fig. 2U and V and Supplemental Figure 4). E-cadherin immunostaining for foregut endoderm reveals lack of tracheal budding in Wnt2/2b DKO mutants (Fig. 2W and X). Development of the liver, stomach, intestine, pancreas and kidneys was grossly normal in Wnt2/2b DKO mutants (Supplemental Figure 5). Together, these data reveal that Wnt2/2b are necessary for lung specification but specification of other foregut derived tissues including the thyroid, esophagus, liver, pancreas, kidney or stomach is not affected.
Wnt ligands can signal through several distinct pathways to regulate cell specification and tissue development. The best understood of these is the β-catenin dependent canonical pathway which has been demonstrated to regulate development and differentiation of several tissues including hair follicles, intestinal epithelium, and the heart (Andl et al., 2002; Cohen et al., 2007; Pinto et al., 2003). To assess whether the canonical Wnt pathway was affected by loss of Wnt2/2b we crossed the BAT-GAL (Maretto et al., 2003) canonical Wnt reporter line to Wnt2/2b mutants and performed lacZ staining in wild-type BAT-GAL embryos, Wnt2−/−:BAT-GAL, Wnt2b−/−:BAT-GAL, and Wnt2/2b:BAT-GAL DKO null mutants. LacZ expression from the BAT-GAL Wnt reporter line was reduced in Wnt2−/− and Wnt2b−/− null mutants and completely lost in the anterior foregut endoderm in Wnt2/2b DKO mutants (Fig. 3A–F). To further address whether canonical Wnt signaling was necessary in the developing foregut endoderm for lung specification, we genetically deleted the Ctnnb1 (β-catenin) gene using the Shh-cre mouse line which expresses the cre recombinase as early as E8.75 in the anterior foregut endoderm and effectively deletes β-catenin expression by E9.5 (Supplemental Figure 6 and (Harfe et al., 2004; Harris et al., 2006). Ctnnb1:Shh-cre mutants exhibited a phenotype identical to Wnt2/2b DKO mutants and completely lacked lung specification and tracheal budding (Fig. 3G–P). However, specification of other gut-derived tissues including the esophagus, liver and thyroid was unaffected (Supplemental Figure 6 and data not shown). These data demonstrate that Wnt2/2b act in the β-catenin dependent canonical Wnt pathway, which is required to specify lung endoderm progenitors.
The phenotype of Wnt2/2b DKO and Ctnnb1:Shh-cre mutants is distinct from other lung hypoplasia phenotypes, including the loss of Fgf10 and loss of Gli2/Gli3 expression, in that the lung is uniquely affected and specification is completely lost. Fgf10 null mutants form a trachea which does not branch, indicating that the lung endoderm lineage is specified but fails to grow and branch (Min et al., 1998; Sekine et al., 1999). Gli2/Gli3 double null mutants fail to form a lung but other aspects of foregut development are severely affected, including the loss of the esophagus (Motoyama et al., 1998). To develop a hierarchical model of lung specification, we assessed expression of Fgf10, Gli2, and Gli3 to determine whether their expression was affected by loss of Wnt2/2b expression. Expression of Fgf10 was reduced in Wnt2/2b DKO mutants, suggesting that it acts down-stream of canonical Wnt signaling in the anterior foregut (Fig. 3Qand R). Although we have demonstrated that Fgf10 is a direct target of Wnt/β-catenin signaling (Cohen et al., 2007), it remains possible that loss of Fgf10 expression is secondary to a loss of lung specification. In contrast to Fgf10 expression, Gli2 and Gli3 expression were unchanged in the anterior foregut region of Wnt2/2b DKO mutants (Fig. 3S and T and Supplemental Figure 4). These data suggest that Wnt2/2b act upstream of Fgf10 but not Gli2/Gli3 in the regulation of lung specification.
The potent role for Wnt/β-catenin signaling in specifying lung endoderm progenitors suggested that ectopic activation of this pathway might dominantly expand lung endoderm progenitor identity outside the normal region in the foregut. To test this hypothesis, we generated Ctnnb1:Shh-cre mutants. These mutants express the stabilized form of β-catenin lacking the phophorylation sites required for its degradation which leads to strong activation of Wnt/β-catenin signaling (Harada et al., 1999). Ctnnb1:Shh-cre mutants displayed defects in tracheal-esophageal septation compared to wild-type controls (Fig. 4A–C, E–G). Immunostaining revealed expansion of Nkx2.1 positive lung progenitors into the hindgut region corresponding to stomach endoderm in Ctnnb1:Shh-cre E10.5 mutants (Fig. 4D and H). This expansion is also evident at E11.5 in Ctnnb1:Shh-cre mutants where the esophagus as well as proximal stomach are populated with Nkx2.1 positive lung progenitors (Fig. 4I–P). To determine whether this expansion of Nkx2.1 lung progenitors represented a reprogramming of foregut endoderm or an increase in Nkx2.1 expression in esophagus and stomach endoderm, E11.5 wild-type and Ctnnb1:Shh-cre mutants where immunostained to detect expression of the esophagus and stomach marker p63. These data reveal that p63 expression is lost in the esophagus and proximal stomach of Ctnnb1:Shh-cre mutants suggesting that activation of Wnt/β-catenin signaling reprograms esophagus and stomach endoderm to a lung endoderm progenitor fate (Fig. 4Q–V). Together, these data indicate that activation of Wnt/β-catenin signaling reprograms posterior regions of the foregut endoderm to a lung endoderm progenitor fate, suggesting that activation of this pathway drives lung endoderm specification in a dominant manner. Thus, our work identifies Wnt2/β-catenin dependent activity as the required signal for specification of lung endoderm progenitors within the foregut (Fig. 4W).
Mutations in other genes have resulted in either a severe truncation in lung development (i.e. Fgf10 null mutants) or defects in LPM leading to severe foregut agenesis including the lung and esophagus (i.e. Gli2/Gli3 double mutants) (Motoyama et al., 1998; Sekine et al., 1999). Here, we show the Wnt2/2b are distinct in their ability to specify lung progenitors within the developing foregut while sparing other organs including the thyroid, esophagus, liver, and pancreas. Moreover, we show that activation of Wnt signaling can reprogram esophagus and stomach endoderm to a lung progenitor fate. Our data support the importance of mesoderm to endoderm signaling that promotes development of foregut derived tissues and extends these findings to provide a molecular hierarchy of foregut endoderm specification.
Previous reports have elucidated additional roles for Wnt signaling in the developing lung. Loss of β-catenin or expression of the Wnt inhibitor dikkopf1 in lung epithelium after lung specification leads to decreased distal airway epithelial development and an overall proximalization of the lung (Mucenski et al., 2003; Shu et al., 2005). A dermo1-cre mesenchymal specific loss of β-catenin in the lung leads to defective lung mesenchymal proliferation and development (De Langhe et al., 2008; Yin et al., 2008). A previous report on a different Wnt2 allele did not report a lung phenotype although approximately 50% of null animals died by birth (Monkley et al., 1996). This could be explained by the presence of significant levels of a truncated Wnt2 mRNA species observed in this previous allele (Monkley et al., 1996). Expression of several other Wnt ligands besides Wnt2 and Wnt2b has been reported in the lung including Wnt7b and Wnt5a (Li et al., 2002; Weidenfeld et al., 2002). Wnt7b has been shown to regulate mesenchymal proliferation as well as epithelial proliferation and maturation (Rajagopal et al., 2008; Shu et al., 2002). Loss of Wnt7b also disrupts lung smooth muscle development leading to a loss of vascular integrity (Shu et al., 2002). The decreased proliferation observed in Wnt7b mutant lungs is similar to that observed in the Wnt2 null lungs suggesting that one of the major roles for Wnt signaling in the lung post-specification is regulation of organ growth and size. Wnt5a is expressed initially in both the mesenchyme and distal epithelium of the developing lung (Li et al., 2002). After E12.5, however, expression of Wnt5a becomes restricted to the distal epithelium (Li et al., 2002). Loss of Wnt5a leads to increased mesenchymal proliferation and a loss in late airway maturation (Li et al., 2002). Since Wnt5a has been reported to act in the non-canonical Wnt pathway (Topol et al., 2003), which can antagonize β-catenin dependent canonical signaling, the increased proliferation observed in the lung mesenchyme of Wnt5a mutants could be due to increased canonical Wnt signaling in this tissue. The present study shows that in addition to regulation of lung development and growth, Wnt signaling through Wnt2/2b is essential for specification of lung endoderm progenitors in the foregut.
In contrast to previous studies in zebrafish which demonstrated an important role for wnt2b in liver development and specification as well as fin development in zebrafish, our data show that Wnt2/2b are not required for mammalian liver specification (Ng et al., 2002; Ober et al., 2006). The studies described here suggest that the role for Wnt/β-catenin signaling along the anterior-posterior axis of the foregut varies between species which may have occurred as Wnt2/2b and the canonical Wnt pathway were co-opted during evolution to specify the lung during the vertebrate expansion into the terrestrial environment. The specificity for Wnt signaling, in particular Wnt2 and Wnt2b, in regulating specification of the lung is interesting in light of previous reports showing an important role for this pathway in pancreas and liver development (Apte et al., 2007; Dessimoz et al., 2005; Murtaugh et al., 2005; Tan et al., 2006). This may be due to the precise expression pattern of these two Wnt ligands or to an important sensitivity of lung endoderm progenitors to canonical Wnt signaling. Moreover, the phenotype in Ctnnb1:Shh-cre mutants is likely due to the timing and specificity of the Shh-cre line since we do not observe early activity in the liver (Supplemental Figure 6). It is also important to note that since Fgf10 is a direct target of Wnt/β-catenin signaling (Cohen et al., 2007), the ability of Wnt2/2b to regulate its expression in the mesoderm surrounding the anterior foregut in a cell autonomous manner could affect other pathways important for mesoderm-endoderm signaling during lung development. Given the critical importance of Wnt2/2b signaling in lung endoderm specification, it will be interesting in future studies to determine whether simple activation of Wnt signaling can rescue the Wnt2/2b phenotype in foregut endoderm. Previous reports have shown that Wnt/β-catenin signaling is also important in adult lung progenitor expansion after injury (Reynolds et al., 2008; Zhang et al., 2008). Thus, Wnt signaling plays a key role in both embryonic as well as adult lung endoderm progenitor development, which reinforces the importance of critical developmental pathways that are recapitulated upon injury and repair.
Wnt/β-catenin signaling is one of the critical developmental pathways that are considered important for both self-renewal and differentiation of stem/progenitor cells. With vigorous efforts underway to determine whether agonists or antagonists can be used to manipulate this pathway for therapeutic purposes, our findings that Wnt signaling is central to the specification and ability to reprogram foregut endoderm to a lung endoderm fate provides important information for investigating lung regeneration. In summary, our data provide a molecular hierarchy of foregut endoderm progenitor specification with Wnt2/2b signaling acting dominantly to specify lung endoderm progenitors in the anterior foregut.
Wnt2 mutant mice were generated using recombineering techniques to replace a portion of the coding region of the first exon with the reverse tet-activator cDNA. Three correctly targeted ES clones were used to generate chimeric mice and all three clones transmitted the mutant allele through the germline. The neomycin selection cassette was removed using the flp recombinase expressing mice (Flper) from JAX and confirmed by PCR (Supplemental Figure 2). Analysis was performed with alleles containing and lacking the neomycin cassette. Wnt2b mutant mice were generated using recombineering techniques to insert loxP sites flanking exons 2 and 3. Two correctly targeted ES cells were used to generate chimeric mice which were bred to transmit these mutant alleles through the germline. Wnt2b mutants were crossed to CMV-cre mice to delete exons 2 and 3 and generate a null allele. Both lines were maintained on a C57BL/6:129SVJ mixed background. Genotyping was performed using the oligonucleotides listed in Supplemental Table 2. The generation and genotyping of Shh-cre, CMV-cre, Ctnnb1, BAT-GAL, and Ctnnb1 mice have been previously described (Brault et al., 2001; Harada et al., 1999; Harfe et al., 2004; Maretto et al., 2003).
Embryos were fixed in 4% paraformaldehyde for 24 hours and embedded in paraffin for tissue sectioning. In situ hybridization and immunohistochemistry was performed as previously described (Shu et al., 2001). Antibodies and dilutions used are as follows: Ki67 (Vector Laboratories, 1:50), Nkx2.1 (Santa Cruz, 1:50), p63 (Santa Cruz, 1:50) β-catenin (BD Biosciences, 1:50). Quantitation of positive cell populations was performed using at least three different tissue sections from at least three different embryos of the same genotype. LacZ histochemical staining of embryos was performed as previously described (Shu et al., 2002). TUNEL staining was performed as previously described (Shu et al., 2007).
Total RNA was isolated from lung tissue at the indicated time points using Trizol reagent, reverse transcribed using SuperScript First Strand Synthesis System (Invitrogen), and used in quantitative real time PCR analysis using the oligonucleotides listed in Supplemental Table 2.
Supplemental Figure 1. Expression pattern of Wnt2 and Wnt2b during lung development. Wnt2 and Wnt2b are expressed in the LPM surrounding the ventral aspect of the anterior foregut from E9.0–E9.5 (A, B, G, H). From E12.5–E18.5, Wnt2 is expressed in the developing mesenchyme with higher levels surrounding the distal regions of the branching airways (C–F). Wnt2b expression is observed in the mesothelium encasing the lung and at lower levels in the distal mesenchyme from E12.5–E14.5 after which expression is extinguished (I and J). Scale bars: A, B, G, H=200 μm, C, D, E, I, J=600 μm, F=800 μm.
Supplemental Figure 2. Wnt2 and Wnt2b gene targeting strategy. Schematic of the Wnt2 gene targeting strategy with a representative Southern blot using the indicated probe1/HindIII and probe 2/BamHI digests (A). The reverse tet-activator (rtta) cDNA was used to replace the coding region of exon 1 of Wnt2. The neomycin cassette was removed using Flper mice (flp sites=green circles) and PCR was used to verify its loss (+/neo=Wnt2− with neo cassette, +/Δneo=Wnt2− without neo cassette, neo/neo=Wnt2−/− with neo cassette, +/+=wild type littermate, − = water PCR control)(A). For unknown reasons, the rtta is not active in this animal model. Schematic of the Wnt2b gene targeting strategy with a representative Southern blot using the indicated probe and an Xba1 digest (B).
Supplemental Figure 3. Serial expression analysis of Nkx2.1 in Wnt2/2b DKO mutants. Serial sections through the wild-type (A–D) and Wnt2/2b DKO mutants (E–H) shown in Figure 2. Panels C and G correspond to panels N and R, respectively, in Figure 2. The other sections are from the same series and confirm lack of Nkx2.1 expression in the anterior foregut of Wnt2/2b DKO mutants. Scale bars=250 μm.
Supplemental Figure 4. Foregut and lung gene expression, cell proliferation, and apoptosis in the anterior foregut of Wnt2/2b DKO mutants. In situ hybridization was performed to assess Foxa2 expression in the foregut of E9.5 wild-type (A) and Wnt2/2b DKO mutants (B). Proliferation as measured by Ki-67 immunostaining is unchanged in the anterior foregut endoderm of E9.5 Wnt2/2b DKO mutants (C–E). TUNEL staining was performed on E9.5 wild-type and Wnt2/2b DKO mutants (F and G). Q-PCR was performed on dissected foreguts of Wnt2/2b double heterozygous (DHET) and Wnt2/2b DKO mutants for Nkx2.1, Wnt7b, Gli2 and Gli3 expression. Wnt2/2b double heterozygous mutants were used as controls since they were the only non-DKO controls obtained in all three litters used to generate this tissue. The residual expression of Nkx2.1 and Wnt7b by Q-PCR is likely due to contamination of Nkx2.1 and Wnt7b expressing cells from other regions of the embryo (i.e. thyroid for Nkx2.1) from the dissection process. Scale bars: 200 μm.
Supplemental Figure 5. Specification of foregut derived organs other than the lung is not affected in Wnt2/2b DKO mutants. As with wild-type litter mates at E14.5 (A), Wnt2/2b DKO mutants have a stomach (St), liver (Li), pancreas (Pa), kidney (Ki), and intestine (Gt) (B and C). Higher magnification pictures showing relatively normal architecture of these organs in both wild-type (D–G) and Wnt2/2b DKO mutants (H–K). Scale bars: A–C=800 μm, D–K=400 μm.
Supplemental Figure 6. Lineage tracing using the Shh-cre line and phenotype of Ctnnb1:Shh-cre mutants. Shh-cre:R26R embryos were stained at E8.75 and E9.25 to show cre activity in the anterior foregut (A and B, arrow). Sectioning of E9.25 Shh-cre:R26R embryos shows lacZ expression in the thyroid primordium (TP, floor plate (FP), notochord (NC), and the ventral aspect of the foregut endoderm (C–E). Sections of a E11.5 Shh-cre:R26R embryo shows lacZ expression in the thyroid primordium (TP), pharyngeal endoderm (PE), esophagus (eso), trachea (tra), endoderm of the lung buds (lb), and the endoderm of the stomach (sto) and hindgut (gut) (F–I). LacZ expression is absent in the early liver (li) (I,). Immunohistochemistry for β-catenin expression in Ctnnb1flox/flox:Shh-cre mutants at E9.5 shows reduced expression of β-catenin in the ventral portion (V) of the anterior foregut (J and K). p63 immunostaining shows that the foregut tube retains esophagus identity in Ctnnb1flox/flox:Shh-cre mutants at E9.5 (L and M). Nkx2.1 immunostaining shows that thyroid specification is retained in Ctnnb1flox/flox:Shh-cre mutants at E9.5 (N and O). Scale bars: C–E, F–H, L–O=100 μm, I=200 μm, J and K=50 μm.
A.M.G. was supported by an American Heart Association Predoctoral fellowship and Y.T. and E.D.C. were supported by American Heart Association Postdoctoral fellowships. This work was support by funding from the NIH to E.E.M. (HL075215 and HL087825). The authors would like to thank the animal care personnel in the Animal Services Unit of the John Morgan Animal facility at the University of Pennsylvania for help in animal husbandry.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.