|Home | About | Journals | Submit | Contact Us | Français|
A predictive allergenicity test system for assessing the contact allergenicity of chemicals is needed by the cosmetic and pharmaceutical industry to monitor product safety in the marketplace. Development of such non-animal alternative assay systems for skin sensitization and hazard identification has been pursued by policy makers and regulatory agencies. We investigated whether phenotypic and functional changes to a subset of dendritic cells (DC), plasmacytoid DC (pDC), could be used to identify contact allergens. To achieve this goal, normal human DC were generated from CD34+ progenitor cells and cryopreserved. Frozen DC were thawed and the pDC fraction (CD123+/CD11c-) was harvested using FACS sorting. The pDC were cultured, expanded, and exposed to chemical allergens (N=26) or non-allergens (N=22). Concentrations of each chemical that resulted in >50% viability was determined using FACS analysis of propidium iodide stained cells using pDC from 2-5 donors. Expression of the surface marker, CD86, which has been implicated in dendritic cell maturation, was used as a marker of allergenicity. CD86 expression increased (≥ 1.5 fold) for 25 of 26 allergens (sensitivity = 96%) but did not increase for 19 of 22 non-allergens (specificity = 86%). In a direct comparison to historical data for the regulatory approved, mouse local lymph node assay (LLNA) for 23 allergens and 22 non-allergens, the pDC method had sensitivity and specificity of 96% and 86%, respectively, while the sensitivity and specificity of the LLNA assay was 83% and 82%, respectively. In conclusion, CD86 expression in pDC appears to be a sensitive and specific indicator to identify contact allergenicity. Such an assay method utilizing normal human cells will be useful for high throughput screening of chemicals for allergenicity.
Langerhans cells (LC)/dendritic cells (DC) represent a heterogeneous group of motile cells including skin and mucosal LC, blood DC, interdigitating cells in the T-cell areas of the lymph node, and interstitial (tissue) DC (Ayehunie et al, 1997). DC monitor “danger” signals in the epidermal, mucosal, and other tissue microenvironments by sampling antigens and transporting them to regional lymph nodes to induce T-cell responses. The activation of DC is considered one of necessary events for the initiation of primary and secondary immune responses (Shi et al, 2003). It has also been postulated that chemicals or microbial components provide signals that alert the immune system to initiate appropriate immune responses. These danger signals initiate the immune response and lead to DC activation, maturation, and presentation of antigens to T-lymphocytes.
The use of LC to study contact allergens has been hampered because: 1) LC constitute only 2-4% of the epidermis, 2) LC are difficult to isolate and propagate in vitro, and 3) LC have a short life span in culture. These difficulties have led researchers to use peripheral blood, bone marrow, and umbilical cord blood derived progenitor cells and monocytes derived DC as alternative sources for LC/DC isolation and generation. However, variation in the methodology for generating of LC/DC has led to variability of results.
The utility of utilizing changes in surface marker expression levels (e.g. HLA-DR, CD54, CD80, CD83, and CD86) on LC/DC in the identification of chemical allergens has been investigated by others. For instance, Hulette et al (2005) showed that exposure of monocyte-derived DC to chemical allergens (with slight to moderate effects on cytotoxicity) increased CD86 expression levels but treatment of DC with non-cytotoxic concentrations of chemicals resulted only in modest upregulation of CD86. Ashikaga et al, (2002) also showed increased CD86 expression and enhanced internalization of MHC class II molecules by DCs using THP-1, a human monocyte cell line.
DC can be divided into two distinct subset of cells: 1) myeloid DC (MDC, CD123-/CD11c+) and 2) plasmacytoid DC (PDC, CD123+/CD11c-) based on phenotypic marker expression, ultrastructural features, cytokine responses, and anatomical distribution. However, it is unclear whether the two DC subsets represent a distinct cell linage or a different state of cell maturation (Summers et al, 2001). While both DC populations can induce strong proliferation of allogeneic naïve CD4+ T-cells (Kadowaski et al, 2000), MDC induce the production of TH1 type cytokines (IFN-γ and IL-12) and the pDC induce TH-2 type cytokine (IL-4, IL-10) release. However, others have questioned this view and suggested that pDC are flexible in directing TH1 or TH2 responses depending on the nature of stimuli including inflammatory mediators like cytokines, pathogens, or dose of antigens (Celia et al, 2000).
pDC constitute only 0.4% of the peripheral blood mononuclear cell population (Ronnblom and Alm, 2001) and do not express the myeloid lineage marker CD11c. Phenotypically, pDC resemble IgG secreting plasma cells and can be identified by their high expression of the Interleukin 3 receptor alpha chain (CD123) combined with other cell surface markers such as CD45RA, HLA-DR, and CD68 (Farkas et al, 2001). pDC have also been associated with allergic reactions (Jahnsen, et al, 2000). For instance, an increased number of pDC in skin lesions are present in patients with contact allergic reactions such as psoriasis vulgaris (Wollenberg et al, 2002), and Lupus erythematosus (LE) (Farkas L, 2001) suggesting their role in proinflammatory responses. In a short-term homing assays, mDC precursors migrated to peripheral tissues and subsequently to draining lymph nodes (LN), whereas pDC precursors directly enter the LN in an E-selectin dependent manner (Yoneyama, etal, 2004). This finding suggests that the distinct trafficking pathway of pDC to lymph nodes may be important in the initiation of primary immune responses to antigens/chemicals with allergenicity potential.
Since pDC are low expressors of CD86 but are involved in the allergic, inflammatory response, we hypothesized that CD86 expression could be used to develop an allergenicity assay of high sensitivity and specificity. Here, we report the development of a human cell based in vitro pDC assay method for detecting chemical allergens. First, we developed a technology from which a large number of pDC can be generated. Then, we developed a pDC assay protocol which was highly sensitive and specific for identifying contact allergens (n=48 chemicals). We anticipate that the pDC-based assay will prove to be useful for large scale screening of chemicals for their allergenicity potential. Development of such a non-animal alternative assay system for hazard assessment directly addresses European legislation such as REACH (Registration, Evaluation, and Authorization of Chemicals) and the 7th Amendment to the Cosmetics Directive which significantly limit and or ban animal testing for safety assessment.
DC were generated from umbilical cord blood derived CD34+ progenitor cells. Umbilical cord blood was obtained from a local hospital or from the National Disease Research Interchange (NDRI, Philadelphia, PA) following Institutional Board Approval (IRB). Isolation, propagation, and maturation of DC progenitor cells from CD34+ progenitor cells were performed as previously described (Romani et al, 1994;, Fraissnette et al 1988, Caux et al, 1992) with modification. Briefly, umbilical cord derived blood mononuclear cells (BMC) were isolated by Ficoll-hypaque density gradient method. Cells were then washed and CD34+cells were isolated using Dynal magnetic beads (Dynal, Lake Success, NY) following the manufacturer's recommendation. Briefly, BMCs were incubated with blocking reagent and CD34 antibody coated microbeads. Cells were incubated at 4°C for 30 min, washed 2× with buffer, and the cell bead complexes were loaded onto a positive selection column (AS depletion column, Miltenyi Biotech) and placed in a strong magnetic field. The column was washed with buffer to remove cells that were not attached to CD34+ magnetic beads, and then the column was removed from the magnetic source to elute the CD34+ cells. Higher purities of progenitor cells were obtained by loading the CD34+ fraction onto an additional column (MS/RS separation column, Miltenyi Biotech). In this manner, progenitor cells purity in excess of 95% were routinely obtained. To generate dendritic cells, the harvested CD34+ cells were plated into 12-well plates at a density of 1 × 105 cells/well and fed with DC-MM medium (MatTek Corporation). The cells were cultured and expanded for 17 days to generate Langerhans cells/dendritic cells (LC/DC) phenotype. The generated LC/DC were then cryopreserved until further fractionation into pDC.
To characterize the DC population, cells were washed 2× with wash buffer consisting of phosphate buffered saline (PBS) containing 2% fetal bovine serum (FBS). Next, cells were re-suspended in cold PBS containing 2% FBS and 2 × 105 cells were seeded into 96-well plates. The appropriate amount of fluorescent dye-conjugated monoclonal antibody (mAb) was added to the respective wells and cells were incubated for 30 min on ice. Finally, cells were washed 2× with wash buffer, fixed with 10% formalin, and analyzed by FACS at the Dana Farber Cancer Institute (Boston, MA) core facility. For each test article, 5000 -10,000 events were collected and analyzed.
Fluorescein isothiocyanate (FITC) conjugated monoclonal antibodies to Langerhans cell surface markers CD1a (clone HI149, isotype IgG1), CD40 (clone 5C3, isotype IgG1), CD80 clone L307.4, isotype IgG1) CD83 (HB15e, isotype IgG1) CD86 (2331(FUN-1), isotype IgG1), HLA-DR (clone G46-6, isotype IgG2a) were purchased from BD Pharmingen (San Diego, CA). CD54 (ICAM 1; Clone BBIGI1, isotype IgG1) were purchased from R&D Systems (Minneapolis, MN). For FACS sorting of myeloid and plasmacytoid dendritic cells, FITC conjugated CD11c (clone KB90, Dako, Carpinteria, CA) and phycoerythrin (PE) conjugated CD123 (clone 7G3; Isotype IgG2a; BD Pharmingen) were used.
DC were washed 2× with wash buffer and 2 × 107 cells were suspended in fresh, cold buffer. To isolate two subsets of DC, pDC (CD123+ / CD11c-) and mDC (CD11c+ / CD123-), the appropriate amount of fluorescent dye-conjugated CD11c and/or PE-conjugated CD123 mAb (10 μl/1 million cells) was added and cells were incubated for 30 min on ice. Labeled cells were then washed 2× with wash buffer, re-suspended in serum free RPMI1640 medium, filtered (mesh size = 30 μm), and volume adjusted to 10-20 × 106 cells/ml. Cell sorting was performed using MoFlow (Cytomation, Fort Collins, CO) at the Dana Farber Cancer Institute core facility. The CD123+ CD11c-cells were further expanded in culture using pDC expansion medium (DCP-MM, MatTek Corporation, Ashland, MA) for 10-14 days. The cells were then exposed to test articles
For each new test article (TA), the TA was dissolved in three different vehicles, ultrapure water, DMSO, and ethanol. If the TA was soluble in water, water was used as the solvent. If DMSO or ethanol was necessary to solubilize the TA, the final concentration of DMSO or ethanol in the culture medium was ≤ 0.01%. At these concentrations, DMSO and ethanol both tested resulted in > 90% pDC cell viability.
Cryopreserved pDC-100 cells were thawed, centrifuged, and transferred to 6-well plates containing pre-warmed assay medium and incubated overnight at 37°C, 5% CO2. After a 24 hour equilibration period, the cells were re-suspended in fresh medium, counted and cultured in 24-wells at 150K cells/well; different concentrations of test materials and the control were added to wells for an exposure time of 18 hr. Initial dose range finding was performed using 3 concentrations (at 4 fold dilutions starting from 1% of the test article). Tissue viability was measured by the reduction of the vital dye 3-[4,5-dimethylthiazol-2-yl]2,5-diphenyltetrazolium bromide (MTT) or by FACS analysis following propidium iodide staining. When MTT was used, cell viability was determined from the ratio of MTT optical density (OD) for test chemical treated cells versus the OD for concurrent negative control cells as per the following equation:
For viable/dead cell discrimination by FACS analysis, 2 μl of propidium iodide (PI) solution (10 ug/ml) were added to each well containing 2 × 105 cells. Living cells with intact cell membranes exclude PI, while dead cells or cells with compromised membranes allow dye entry. Cells that actively take up the dye fluoresce brightly in the red range of the visible color spectrum. The % of viable cells was quantified using FACS.
If all concentrations used resulted in <50% viability, additional dose rang findings were performed using 3 concentrations of 10 fold dilution below the lowest concentration utilized in the initial dose range finding test. Based on the viability/ dose range data, concentrations that resulted in >50% viability were selected for allergenicity analysis.
To study allergenicity of chemicals and chemical formulations, common contact allergens known to cause allergic contact dermatitis (ACD) were selected from the North American Contact Dermatitis Tray or from published datasets of skin allergens (Gerberick et al, 2004). Both contact allergens (N=26) and non-allergens (N=22) were investigated. Concentrations that resulted in >50%cell viability following 18 hour incubation at 37°C were used in the various experiments. All test articles were purchased from commercial sources in analytical grade purity either from OmniDERM (Quebec, Canada), ICN Biochemicals (Plainview, N.Y.), and Sigma (St. Louis, MO) For FACS analysis, pDC (1 × 105 cells) were incubated with test material for 18 hr at 37°C. After exposure, the pDC were collected and phenotypic changes were examined by FACS analysis using specific antibodies to CD86
Changes in CD86 expression were quantified by calculating the fold increase (FI), defined as a percentage of CD86 expression of treated pDC divided by the percentage CD86 expression of untreated (control) pDC.
The test articles were rank ordered according to their level of CD86 expression and these data were compared to historical allergenicity data obtained from LLNA and human testing. A 1.5 fold increase in CD86 expression minimized the differences between sensitivity and specificity of the in vitro and in vivo assay methods. Hence, the cutoff value for allergenicity was set to 1.5 fold increase in order to optimize discrimination between allergens and non-allergens.
As per the recommendations of the ECVAM Biostatistics Task Force (Holzhutter, 1996), the degree of concordance between the in vitro predicted and the in vivo classification were evaluated by calculating the sensitivity, specificity, and accuracy of test results. Data for all 48 test materials were summarized in a 2 × 2 contingency table (Table 7a). Separate contingency tables for 45 test materials for which LLNA data were available were also constructed (Table 7b and 7c). Contingency tables for the pDC assay were based on predictions from 50% or more of the donors or the total number of assays performed. If a test article is positive for one donor and negative in the second donor, a third donor was used to verify the allergenicity potential of the test article.
A large number of normal human DC were generated from umbilical cord blood CD34+ progenitor cells and cryopreserved. From samples of umbilical cord blood (55-100 ml), 691± 582 million DC (n=16 donors) were obtained following a 15-18 day culture process. Although the average number of DC generated from an umbilical cord sample was >190 fold more than the starting number of CD34+ progenitor cells, donor to donor variation in cell yield was noted. Overall, the DC number obtained per UBC sample ranged from 280 million to 2.5 billion (Table 1). This result is a significant achievement because previous attempts to develop in vitro assays utilizing normal human DC were hampered by the inability to harvest large numbers of cells (Regnier et al, 1997, Xu, et al 1995). The ability to harvest a large number of dendritic cells from a single donor will aid in assay reproducibility and large scale screening of chemicals for potential allergenicity.
To characterize the DC population, surface marker expression levels were determined by FACS analysis. Results indicate that the DC express: 1) CD1a (49-70%), HLA-DR (80-92%), as well as co-stimulatory molecules such as CD80 (39-70%), CD40 (37-74%), and CD86 (36-67%), CD206 (20-42%), and low levels of intracellular adhesion molecule-1 (ICAM-1) (22%). Ultrastructural features of the DC were also examined using transmission electron microscopy and the results showed that the DC express both cytoplasmic and plasma membrane bound Birbeck granules (Figure 1), a unique characteristic of Langerhans cells (Fahrbach et al 2007).
As shown in Figure 2a, the DC consisted of two distinct subsets, plasmacytoid dendritc cells (CD123+ / CD11c-) and myeloid dendritic cells (CD11c+ / CD123-). The ratio of pDC to mDC averaged 1.2:1. Morphologically, pDC are spherical and of low dendriticity compared to the mDCs which are more dendritic with irregular shape (Figure 2b). pDC had low CD1a expression (12%) while mDC had significant CD1a expression (67%); both cell fractions expressed high levels of HLA-DR (> 70%). Results of surface marker characterization by FACS for DC from a representative donor are shown in Figure 2c.
Some of the previous limitations associated with the use of dendritic cells for allergenicity studies were: 1) short lifespan of cells in culture and 2) insufficient number of DC obtained from a single donor. These limitations were overcome by developing a specific pDC culture medium (DCP-MM, MatTek Corporation, Ashland) that allows expansion and maintenance of pDC in vitro. In the DCP-MM, the cells were maintained in culture for 10-14 days and increased 12-34 fold in cell number. The growth curves of 2 representative pDC preparations are shown in Figure 3.
Since the maturation state of DC is critical for the induction of T cell function (Fahrbach, KM et al., 2007,, Lapointe et al 2000, Labeur et al, 1999) and lipopolysaccharide (LPS) and TNF-α are known to induce DC maturation, (Labeur SM et al, 1999), the effect of LPS and TNF-α on pDC surface marker expression was examined. pDC were stimulated with LPS (5 ug/ml) and TNF-α (30 ng/ml) overnight and analyzed using FACS. As shown in Table 2, the expression of the LC-associated co-stimulatory molecules, CD80, CD83, and CD86, increased following exposure to LPS and TNF-α. These results showed that pDCs were functional and responsive to exogenous stimuli.
Preliminary experiments were performed to examine CD86 expression and the chemokine receptor-7 (CCR7; a chemokine receptor expressed by activated DC/LC that is essential for DC-T cell interaction), levels in pDC and mDC following exposure to chemicals with or without allergenicity potential. As shown in Table 3, pDC exposed to allergens showed increased expression of CD86 (Fold increase, FI >1.5); CD86 expression in mDC did not increase following exposure to allergens. Also, non-allergens showed no increase in CD86 expression (FI < 1.5) following exposure to pDC. In contrast, CCR7 expression increased for BB exposure but not for the other allergens or non-allergens (data not shown). Based on these observations, CD86 was deemed a potential marker for allergenicity detection and CCR7 was not further tested.
To further examine the utility of CD86 as a marker for allergenicity, experiments were performed using 4 additional allergens, hydroquinone (HQ), nickel sulfate (NS), cinnamaldehyde (CA), and Balsam of Peru (BP), and 3 non-allergens, propylene glycol (PG), Tween 20 (TW), and ethanol (ET). The results confirmed our previous observations: an FI of >1.5 was obtained for all the allergens but not for the non-allergens (Figure 4) with the exception of Tween-20 (FI = 1.8).
Based on the results of Table 3 and Figure 4, testing of the pDC method was expanded to include a total of 48 test articles (26 allergens and 22 non-allergens). All materials were tested using pDC from a minimum of 2 donors. Data from a representative donor for the 48 test materials are presented in Table 4. Materials were categorized as “allergen” for FI > 1.5 and as “non-allergen” for FI < 1.5. As shown in Table 4, all 25 of the 26 allergens tested gave positive results in the pDC assay (assay sensitivity = 96%) whereas 19 of 22 non-allergens were correctly identified as non-allergens (assay specificity = 86%). Assay results for all 48 test materials are summarized in Table 7a.
The reproducibility of the pDC assay method was evaluated using the 20 test chemicals shown in Table 5 with pDC obtained from 2-5 different donors. As shown in Table 5, identical assay results were obtained for 16 of 20 test articles regardless of the pDC donor; test results from 1 donor differed from the other donors for the remaining 4 test materials. In total, 81 of 85 (95.3.3%) assay results were identical. Further, the additional 28 test materials shown in Table 4 were tested using pDC from 2 donors, and in all cases, identical results were obtained. Assay-to-assay variability using pDC cells from a single donor for 10 of the test articles was also found to be minimal; no discordant test results were observed (Table 6).
Data for the pDC assay method were compared to historical results from the local lymph node assay, a murine based assay which has been validated as an acceptable alternative to guinea pig tests previously used to predict human allergenicity (Basketter et al, 2003). Historical LLNA data for 45 of the 48 test articles used in pDC assays here reported (Table 4) were available (Basketter et al, 1999, Gerberick, 2004). These data are included in Table 4 and were used to construct contingency tables comparing pDC and LLNA assay results to human allergenicity classifications. As shown in Table 7b & 7c, the sensitivity of the pDC assay was higher than that of the LLNA (96 % vs. 83%) and specificity was similar (86% vs. 82%); overall accuracy of the pDC and LLNA assays was 91% and 82%, respectively. In short, the pDC based assay method gave comparable results to that of the validated animal-based LLNA allergenicity test method and exhibited high levels of sensitivity and accuracy versus historical human test results.
Previous reports (Fahrbach et al 2007; Maciejewski-Lenoir, 2006) have shown that CD34+ progenitor cells derived from human umbilical cord blood can be used as a source for large scale production of dendritic cells resembling epidermal LC that contain Birbeck granules (unique marker of LC), CD1a, and Langerin. Production of these cells in larger quantity enabled us to: 1) perform large-scale screening of chemicals for their allergenicity potential, 2) determine intra-assay or inter-assay reproducibility using cells from the same donor, and 3) monitor donor-to-donor variability.
As of 1994, more than 100,000 chemicals were in commercial use and 2,000 new chemicals were introduced every year (Koeter, 1994). Furthermore, the number of new product formulations tested or marketed each year exceeds new chemicals by 10-100 fold. The test methods currently available to assess potential contact allergenicity of materials include: 1) human patch tests, 2) guinea pig maximization test (GPMT), and 3) the murine-based local lymph node assay (LLNA). In GPMT tests, hazard identification is done by visual observations of erythema and edema reactions which are subjective, have difficulties in differentiating between contact allergens and strong irritants, and are time consuming (Silva O.D 1994). Human patch tests are expensive and due to ethical considerations, prescreening with animal tests is required prior to human clinical studies. Animal tests frequently require the use of Freund's complete adjuvant, which is painful (Magnusson, et al 1969) or the use of radioisotopes. Furthermore, animal tests will be outlawed in Europe as a result of the 7th Amendment to the Cosmetics Directive (Grindon et al 2005) and due to species-to-species variability, animal tests may not accurately predict allergenicity in humans. Therefore, a cost-effective in vitro assay system that utilizes cells of human origin to predict contact allergens would be of great benefit for screening of new chemicals and formulations.
Most of the work related to allergenicity testing has focused on the use of animal models such as the guinea pig and mouse. Recently, the murine local lymph node assay (LLNA) has been validated as a stand-alone alternative to guinea pig assays for the identification of skin sensitization hazards (Basketter et al, 2003). The LLNA measures lymphocyte proliferation using incorporation of radiolabeled nucleotides (3H-thymidine or uridine) in draining lymph nodes of mice topically exposed to test chemicals as an indicator of sensitization (Kimber and Weisenberger, 1999). Although the LLNA reduces the number of animals used and reduces the time required to obtain results, it remains an animal based model. In this context, the pDC based assay system here reported gave comparable results to those of LLNA assay while avoiding the use of animals for chemical safety assessment.
While previous studies have investigated the use of CD86 expression by dendritic cells as a marker to identify chemicals with allergenicity potential (Gaspari et al, 1998, Hulette et al, 2005), the use of DC for allergenicity studies was limited in part by the low number of DC isolated from a single donor. Here we report that large number of DC can be generated from CD34+ precursor cells from the umbilical cord blood obtained from a single donor. Even though we standardized culture conditions, a large variation in the DC yield from different donors was observed. The reason for this variation is not known but it may be associated with donor-to donor variation similar to what is observed in all biological systems.
Although several researchers have utilized DC for chemical allergenicity studies, the role of CD86 expression level by pDC was not examined. The low level background expression of CD86 by unstimulated pDC makes the pDC-based assay more sensitive and attractive for hazard identification particularly for chemicals with weak allergenicity potential. Here, we have extended earlier studies by specifically focusing on CD123+ CD11c- dendritic cell subsets for the identification/screening of chemical allergens. Furthermore, we have developed methods to expand pDC (CD123+CD11c-) and maintain them in culture for extended time periods. Based on an average yield of 700 × 106 unfractionated DC from each umbilical cord blood (UCB) sample (Table 1), DC to pDC expansion factors of 6-20 fold increase in cell number (Figure 2), and a pDC FACS yield of 20%, approximately 1-3.3 × 109 pDC/UCB sample can be generated. Assuming 1 × 105 pDC are required per assay, 1-3 × 104 tests would be possible from a single UCB sample. Such a large stock of cells from the same donor would enable large scale screening of chemical allergenicity.
In conclusion, upregulation of CD86 by pDC appears to be a valuable biomarker for chemicals with allergenicity potential. This in vitro assay system appears suitable for large-scale screening of chemicals which is difficult to do using existing animal-based methods that requires large number of animals, raise animal welfare concerns, and incur high costs per experiment. The pDC assay method is expected to be cost-effective and, since it utilizes cells of human origin, interspecies extrapolation is avoided. Development and validation of alternative (non-animal) test methods is considered to be a top priority by US and European organizations such as Interagency Coordinating Committee on the Validation of Alternative Methods (ICCVAM) and the European Center for the Validation of Alternative Methods (ECVAM). In short, increased expression of CD86 by pDC appears to be sensitive and specific marker for the identification of chemicals with allergenicity potential.
The research was supported by the National Cancer Institute (NCI) SBIR Grant (#R44 CA106137)., National Institute of Health.
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.