SEARCH

SEARCH BY CITATION

Keywords:

  • Fluorescence-activated cell sorting;
  • Neurosphere;
  • Prominin;
  • Laminin;
  • Extracellular matrix

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results and Discussion
  6. Disclosures
  7. Acknowledgements
  8. References
  9. Supporting Information

The identification of markers for the isolation of human neural stem cells (hNSCs) is essential for studies of their biology and therapeutic applications. This study investigated expression of the integrin receptor family by hNSCs as potential markers. Selection of α6hi or β1hi cells by fluorescence-activated cell sorting led to an enrichment of human neural precursors, as shown by both neurosphere forming assays and increased expression of prominin-1, sox2, sox3, nestin, bmi1, and musashi1 in the β1hi population. Cells expressing high levels of β1 integrin also expressed prominin-1 (CD133), a marker previously used to isolate hNSCs, and selection using integrin β1hi cells or prominin-1hi cells was found to be equally effective at enriching for hNSCs from neurospheres. Therefore, integrin subunits α6 and β1 are highly expressed by human neural precursors and represent convenient markers for their prospective isolation.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results and Discussion
  6. Disclosures
  7. Acknowledgements
  8. References
  9. Supporting Information

Human neural stem cells (hNSCs) have the ability to form all the major cell types of the central nervous system, making them candidates for cell-based therapies in neurodegenerative disorders, such as Parkinson's disease or multiple sclerosis. However, two factors have hampered the development of clinical applications, such as cell transplants. First, the very limited availability of fetal tissue containing large numbers of hNSCs. Second, the inability to purify these cells from more differentiated types. The first can be overcome by the expansion of primary cells, as human neural stem cells can be maintained in vitro within free-floating aggregates, termed neurospheres. The second factor, however, remains a key problem, especially as neurospheres represent a complex mixture of neural stem cells (NSCs) and more differentiated cell types. Cell markers appropriate for selection strategies are therefore required. The intracellular location of factors enriched in NSCs, including Musashi1, Nestin, and Sox1, has limited their usefulness as prospective markers [1, [2], [3], [4]5]. Attempts have been made to circumvent these problems by using adenoviral constructs with green fluorescent protein under the control of various promoters or enhancers, but the incorporation of adenoviral sequences limits the potential applications of such cells [6]. A more optimal strategy, and one that has allowed the near-homogeneous purification of the hematopoietic stem cell (HSC) [7, 8], would be the identification of cell surface antigens suitable for fluorescence-activated cell sorting (FACS). Although markers have been defined for the mouse neural stem cell (mNSC), enabling the mNSC to be enriched to 80% [9, [10]11], few markers have been developed for hNSCs.

One method used to identify potential markers is to examine those cell surface receptors likely to be required for stem cell maintenance in vivo. The NSCs are localized to the ventricular and subventricular zone of the developing brain, where specialized areas called niches are responsible for maintaining the multipotency and self-renewal capacity of NSCs [11, 12]. The study of other stem cell niches has allowed the identification of common components, including an extracellular matrix (ECM)-rich basal lamina, that are likely to form part of the hNSC niche [13]. Interactions with the ECM are mediated mainly by cell surface receptors called integrins. Integrins are transmembrane αβ heterodimers involved in the regulation of proliferation, survival, migration, and differentiation [14]. The expression of high levels of integrin β1 has been used successfully to enrich human epidermal and rodent neural stem cells from more restricted progenitor populations [15, [16]17]. Recent studies comparing the gene expression profiles of mouse embryonic stem cells, NSCs, and HSCs have found the expression of laminin-binding integrins α6 and β1 to be elevated in all three populations [18, 19]. Similarly, integrin α6 has been found to be a marker for hepatic and spermatogonial stem cells [20, [21], [22]23]. In this study, we therefore asked whether integrins are highly expressed by hNSCs and provide a marker for their isolation.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results and Discussion
  6. Disclosures
  7. Acknowledgements
  8. References
  9. Supporting Information

Cell Culture

Human fetal tissue (8–10 weeks postconception) was collected in accordance with the arrangements for informed consent recommended by the Polkinghorne Committee [24] and the U.K. Department of Health guidelines [25] and with the approval of the Addenbrooke's Hospital Medical Research Ethics Committee. The hNSCs from the cortex were propagated as free-floating aggregates (neurospheres), and cellular expansion prior to use in the studies described here was achieved using a passaging method described in detail elsewhere [26]. In brief, fresh tissue was initially dissociated in trypsin and seeded at a density of 200,000 cells per milliliter in a T75 flask containing Dulbecco's modified Eagle's medium/Ham's F-12 medium (3:1; Gibco, Paisley, U.K., http://www.invitrogen.com), supplemented with N2 (1:100; Gibco), epidermal growth factor (EGF), fibroblast growth factor (FGF)-2 (both at 20 ng/ml; R&D Systems Inc., Abingdon, U.K., http://www.rndsystems.com), and heparin (5 μg/ml; Sigma-Aldrich, Poole, U.K., http://www.sigmaaldrich.com). Spheres had grown to a radius of >0.35 mm after 14 days of expansion, and passaging of cells was routinely undertaken at this time point by sectioning spheres into 200-μm sections using a McIlwain tissue chopper, after which they were reseeded into fresh growth medium. Cultures were fed every 4–5 days by replacing half the medium. To induce differentiation, growth factors were withdrawn, and neurospheres were plated on poly(d-lysine) and laminin-1 (10 μg/ml; Sigma-Aldrich)-coated eight-well chamber slides (Nunc, Roskilde, Denmark, http://www.nuncbrand.com) for 7 days prior to immunostaining.

Semiquantitative Reverse Transcription-Polymerase Chain Reaction

Total RNA was extracted from neurospheres using RNeasy minikit (Qiagen, Crawley, U.K., http://www1.qiagen.com) before being treated with RNase-free DNase I (Roche Diagnostics, Lewes, U.K., http://www.roche-applied-science.com) to remove genomic DNA contamination. cDNA was generated from 1 μg of total RNA using a First Strand cDNA Synthesis kit (GE Healthcare Life Sciences, Little Chalfont, U.K., http://www.amersham.com). Each polymerase chain reaction (PCR) consisted of 1/30th reverse transcription reaction, 1× PCR buffer (Qiagen), 0.25 mM each dNTP (GE Healthcare Life Sciences), 2 μM each primer, and 1.25 units of Taq polymerase (Qiagen). PCR mixtures then were denatured at 95°C for 2 minutes before being cycled at 94°C for 1 minute, the annealing temperature for 1 minute, and 72°C for 1 minute. A final extension of 72°C for 10 minutes occurred after cycling. Details of the primer sequences, annealing temperatures, and cycle numbers are provided in supplementary online Table 1. The PCR products were separated by electrophoresis on 2.5% agarose gels.

Immunohistochemistry

Neurospheres were fixed in 4% paraformaldehyde for 1 hour at room temperature before being washed with phosphate-buffered saline (PBS) and placed in 25% sucrose overnight at 4°C. Spheres were then cut into 14-μm sections using a Leica CM3000 cryostat (Leica, Milton Keynes, U.K., http://www.leica.co.uk). Sections were treated with 10% normal goat serum (DakoCytomation, Ely, U.K., http://www.dakocytomation.co.uk) and 0.2% Triton X-100 in PBS for 1 hour at room temperature, before being incubated with anti-nestin (1:500; a kind gift from Dr. H. Okano, Tokyo, Japan) or anti-β1 integrin (1:200; clone P5D2; Chemicon, Chandlers Ford, U.K., http://www.chemicon.com) overnight at 4°C. Subsequently, the sections were incubated with the appropriate Alexa Fluor-conjugated secondary antibody (Molecular Probes Inc., Leiden, The Netherlands, http://www.probes.invitrogen.com), diluted 1:200 in Hoechst 33258, for 2 hours at room temperature. Sections were viewed under a Zeiss Axioplan fluorescence microscope (Carl Zeiss, Welwyn Garden City, U.K., http://www.zeiss.co.uk), and images were captured using Openlab software (Improvision, Coventry, U.K., http://www.improvision.co.uk). Differentiated neurospheres were fixed as described above and then labeled with anti-βIII tubulin (1:500; Tuj1; Sigma-Aldrich) and anti-glial fibrillary acidic protein (1:500; DakoCytomation) followed by appropriate secondary antibodies as described above.

Flow Cytometry and Neurosphere Formation Assay

Neurospheres were dissociated with Accutase (Chemicon) before being resuspended in PBS and stained with antibodies against β1 integrin (clone TDM29-fluorescein isothiocyanate; Chemicon) or prominin-1 (clone 293C3-phycoerythrin; Miltenyi Biotec, Bisley, U.K., http://www.miltenyibiotec.com) at 4°C for 1 hour. For α6 integrin staining, cells were incubated with the primary antibody (clone NKI-GoH3; Chemicon) for 1 hour at 4°C before addition of the R-phycoerythrin-conjugated secondary antibody (1:300; Molecular Probes) for 45 minutes at 4°C. All primary antibodies were used at 2 μg/106 cells. Dead cells were excluded by propidium iodide staining (2.5 μg/ml; Sigma-Aldrich), and doublets were identified by the pulse-width parameter. The specificity of each antibody was confirmed by the lack of significant labeling if the fluorochrome-conjugated anti-integrin or anti-CD133 antibody was omitted, as shown in Figures 2E and 4A. The cells were sorted using a MoFlo flow cytometer (DakoCytomation) into 96-well plates (Nunc) at a density of either 250 or 500 live cells per well (750 or 1,500 cells per cm2) or, in the experiments using a limiting dilution assay to determine the frequency of neurosphere-forming cells, progressively decreasing numbers of cells per well. Each well contained 200 μl of neurosphere medium, supplemented with 2% B27 (Gibco) in place of the N2. Cells were fed every 4–5 days by replacing 50% of the medium, and the number of neurospheres formed was counted after 21 days in culture.

Statistical Analysis

For the analysis of expression (PCR or FACS), three fetal brain samples were analyzed, and a representative experiment is shown. For the cell sorting studies, eight experimental replicates were performed for each of at least two fetal brain samples tissues, and data are presented as mean ± SE. All statistics were calculated using Student's t test, with p < .05 considered significant.

Results and Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results and Discussion
  6. Disclosures
  7. Acknowledgements
  8. References
  9. Supporting Information

Human neural stem cells, like rodent NSCs, can be maintained in vitro as free-floating aggregates, termed neurospheres. These spheres grow from single cells following dissociation at each passage in culture, and when withdrawn from mitogenic growth factors, each sphere will generate neurons, astrocytes, and oligodendrocytes. As such, these single cells have stem cell properties, and currently, the identification of NSCs is predominantly based on assays defining NSCs as neurosphere-initiating cells [27]. To investigate a role of integrins as markers for hNSCs, it was first necessary to define which subunits are expressed. Using a reverse transcription-polymerase chain reaction (RT-PCR)-based approach, Figure 1 shows the integrin subunits found to be consistently expressed by human neurospheres, regardless of the growth factors used (EGF, FGF-2, or a combination) or the length of time in culture. The mRNA for α subunits 1–3, 5–10, and V were expressed in all samples (n = 3), along with β subunits 1, 5, and 8, whereas integrins β4 and β6 were never seen.

thumbnail image

Figure Figure 1.. Expression of integrin subunits by human neurospheres. Reverse transcription-polymerase chain reaction (PCR) analysis of mRNA prepared from human neurospheres. The numbers over each lane indicate the specific α and β subunits expressed (i.e., 1 is α1, 2 is α2, etc.). Note that α6, αV, β1, β5, and β8 were detected after only 30 PCR cycles, whereas other subunits required 40 cycles. Also, note that splice variants of α3 and α6 are expressed. Size markers are shown in base pairs on the left.

Download figure to PowerPoint

In light of the number of integrin subunits that associate with β1 expressed in human neurospheres by RT-PCR, we next asked whether the β1 integrin subunit was highly expressed on hNSCs. Three sets of experiments were conducted to examine this. Nestin is an intermediate filament protein that is highly expressed by neural stem cells [4]. First, therefore, cryostat sections of neurospheres were stained for both nestin and β1 integrin to ask whether the proteins colocalize (Fig. 2A), thus indicating that high levels of β1 integrin expression are seen in cells expressing an established NSC marker. Interestingly, previous studies have shown that rodent neurospheres have a three-dimensional structure, with the stem cells present toward the edge, whereas differentiated cell types are found in the center of the sphere [16]. In agreement, here we found that β1 integrin and nestin are more highly expressed at the edge of human neurospheres, showing that human neurospheres also have a distinct three-dimensional architecture and demonstrating that hNSCs do express high levels of β1 integrin (Fig. 2A).

thumbnail image

Figure Figure 2.. Integrin β1 is highly expressed by human neural stem cells. (A): Cryostat section of a human neurosphere stained for Hoechst 33258 (to label nuclei), β1 integrin, and nestin. A merged image is shown in the lower right panel. Expression of both proteins is colocalized at the edge of the neurosphere. Scale bar = 10 μm. (B): β1(hi), α6(hi), and α6(hi)+β1(hi) were sorted at 250 live cells into a 96-well plate, and the number of neurospheres forming was quantified 21 days later. “Unsorted” refers to sorting live cells with no gating on β1 or α6 expression. Neurosphere formation was significantly increased relative to the unsorted population for all conditions, although no significant additive effect of combining selection for β1(hi) and α6(hi) was found. , p < .05; n = 2. (C): Differentiation of neurons and astrocytes in cultures of the neurospheres grown from β1hi cells and then plated onto poly(d-lysine)/laminin-1 (10 μg/ml) substrates before the withdrawal of growth factors. Note the presence of glial fibrillary acidic protein+ astrocytes (green) and βIII-tubulin+ neurons (red). Nuclei are labeled in blue (Hoechst). (D): Reverse transcription-polymerase chain reaction analysis of markers for neural stem cells and differentiated cells compared for the 20% of cells expressing high and low levels of β1 integrin. Stem cell markers were more highly expressed in the integrin β1hi population, whereas differentiated markers were preferentially expressed by the integrin β1lo cells. (E): Flow cytometry analysis showing unstained cells (left) and cells stained for β1 and α6 integrin (right). Note that all α6+ cells also express β1 integrin, but that not all β1hi cells express high levels of α6, as discussed in the text. Abbreviations: α6(hi), the 10% of cells expressing the highest levels of α6 integrin; α6(hi)+β1(hi), the 10% of cells expressing the highest levels of both β1 integrin and α6 integrin; β1(hi), the 10% of cells expressing the highest levels of β1 integrin; FITC, fluorescein isothiocyanate; hi, high; PE, phycoerythrin.

Download figure to PowerPoint

Second, we asked whether cells expressing high levels of β1 integrin showed the property of NSCs of being able to generate neurospheres. We used the neurosphere-forming assay to examine an integrin β1hi population (the top 10% of cells as determined by β1 integrin expression levels) compared with the total population. Labeled cells dissociated from neurospheres grown from human fetal tissue and passaged by chopping as described under Materials and Methods (primary neurospheres) were sorted by FACS into 96-well plates at a very low density of 1.25 cells per microliter (Fig. 2B) [28]. Twenty-one days later, the number of secondary neurospheres was counted. The results showed that selecting for the integrin β1hi cells within the cells dissociated from the primary neurospheres led to a 2.1-fold increase in the number of secondary neurospheres compared with the number formed when starting with an unselected population of primary neurosphere cells (p < .05; n = 2), indicating that there are more hNSCs in the β1hi population. To confirm that β1hi cells used to generate the secondary neurospheres are enriched for cells capable of repeated self-renewal, we dissociated these secondary neurospheres and replated them at clonal density without any further selection for β1hi cells. As expected, we found that they generated tertiary neurospheres whose frequency (4.88 ± 2.88 per 500 cells plated) was not significantly different from that obtained when the secondary neurospheres were generated from unselected populations (5.69 ± 3.06). To confirm that β1hi cells are multipotent, we demonstrated the presence of both neurons and astrocytes when neurospheres generated from the β1hi cells were plated onto adhesive substrates and withdrawn from growth factor expansion (Fig. 2C). As an alternative means of calculating the frequency of NSC in the β1hi cells, we also performed a limiting dilution analysis as shown in supplementary online Figure 1, in which the frequency of wells without any neurospheres after 21 days was counted at a range of plating densities (10–60 cells per well), and the slope of the best-fit line was used to calculate the frequency of sphere-forming cells, as shown in supplementary online Figure 1. The result obtained in this clonal assay (5.58 ± 2.28%) was similar to that seen in the experiments shown in Figure 2B using 250 cells per well (3.05 ± 0.25%), confirming that the cell density used in this latter assay was sufficiently low to generate clonal neurospheres.

Third, β1hi and β1lo cells were evaluated for their expression of a panel of NSC markers (Fig. 2D). The PCR screen verified that β1hi cells were enriched for well-established markers of rodent NSCs, with the mRNA for prominin-1, nestin, sox2, sox3, musashi1, and bmi1 being upregulated compared with the β1lo cells. No change was seen with sox1 and musashi2, consistent with their expression in both precursors and differentiated neuronal subtypes [29, 30]. Furthermore, the neuronal marker, βIII tubulin, was more highly expressed in the β1lo population (Fig. 2D). Together, these findings indicate that hNSCs express high levels of β1 integrin, whereas differentiated human neural cell types downregulate their expression of β1 integrin.

Next, we asked which αβ1 heterodimer was highly expressed on hNSCs. An attractive candidate is α6β1, a receptor for laminins expressed in the basal lamina found in many stem cell niches [13, 14, 23]. We investigated whether the α6 subunit, like β1 integrin, is highly expressed by hNSCs. Figure 2E demonstrates that all α6+ cells express β1. This was expected, as the other partner for α6, the β4 subunit, was not expressed in human neurospheres (Fig. 1). Selecting for α6hi cells (top 10%) led to a 2.2-fold increase in neurosphere formation compared with the whole population (Fig. 2B; p < .05; n = 2). This was not significantly different from the result obtained when sorting for β1hi alone, with no additive effect observed when selecting for both markers together (Fig. 2B). This indicates that α6 and β1 integrins are both highly expressed on hNSCs, likely as a heterodimer, and that either subunit can be used as a marker for their enrichment. This lack of an additive effect also suggests that only α6β1, and not other β1 heterodimers, is expressed at high levels on hNSCs, although other β1 heterodimers are likely to be expressed at lower levels. Importantly, not all the β1+ cells in the neurospheres express α6 (Fig. 2E); analysis of the FACS data shows that only 20% of the β1+ cells also express α6. This was expected, given the large number of α subunits expressed in the neurosphere cells, as shown in Figure 1. However, taken together with the lack of any additive effect of sorting for β1hi cells and α6hi cells, this suggests that these other β1 heterodimers must be expressed at high levels on more differentiated cell types within the neurospheres. The function of α6β1 on hNSCs remains to be established, but previous studies examining rodent NSCs have revealed roles in maintenance, proliferation, and migration of stem/precursor cells [16, 32, 33].

An important technical issue when sorting for cells expressing high levels of markers such as integrins is that the flow cytometer measures total fluorescence. This means, as shown in supplementary online Figure 2, that large cells are more likely to appear in the markerhi category, as they will have an increased total fluorescence compared with a small cell for an equal marker density on the cell surface. Therefore, it is possible that selecting for β1hi cells merely chooses large cells, and as increased cell size has been shown to select for rodent NSCs [9, 10], the enrichment for hNSCs we observe may simply reflect this selection for size. To exclude this possibility, cells of one particular size were sorted for β1hi and β1lo (Fig. 3). Again, selection of β1hi cells led to a significant enrichment in the neurosphere-forming cell (p < .05; n = 3 for both cell sizes), confirming that hNSCs do express higher levels of β1 integrin regardless of cell size. However, it was notable that the degree of enrichment was increased when selecting for cells of a medium size, rather than of a large size (5.1- and 1.3-fold, respectively). This is likely to be due to the hNSCs being located predominantly among the large cells, indicated by reduced number of neurospheres forming from the total population for the medium-sized cells compared with the large cells (17.79 vs. 22.96 neurospheres per 500 cells, respectively; p < .01; n = 3). Consequently, enrichment based on β1 expression appears to be more efficient from medium-sized cells.

thumbnail image

Figure Figure 3.. Elevated expression of β1 integrin in human neural stem cells is independent of cell size. To demonstrate that the enrichment seen with β1hi cells was not simply a consequence of selecting for large cells, cells of medium (corresponding to the median 10%) or large (top 10%) size were selected and further sorted for high or low β1 expression. Note that increased neurosphere formation was observed when selecting for integrin β1hi cells, regardless of the cell size. , p < .05; n = 2. Abbreviation: Pop, population.

Download figure to PowerPoint

Currently, the most effective marker for hNSCs is prominin-1, a pentatransmembrane domain protein of unknown function (CD133) [34, 35]. Analysis by flow cytometry indicated that integrin β1 and prominin-1 are expressed on the same cell populations, with 90% of prominin-1+ cells also expressing β1 integrin, and 10% of cells expressing high levels of both markers (Fig. 4A). As such, our observations in hNSCs are similar to observations that prominin-1+ cells were located within the β1hi population of human prostate and epidermal stem cells [34, 35]. The efficiency of sorting for hNSCs using integrin β1 was compared with the use of prominin-1 by sorting cells into 10% bins based on expression levels of either marker and performing a neurosphere-forming assay (Fig. 4B). The results indicate that β1 integrin is as effective as prominin-1 for purifying hNSCs, leading to 2.8- and 2.4-fold increases in neurosphere numbers, respectively (p = .49; n = 3).

thumbnail image

Figure Figure 4.. Integrin β1 is co-expressed with prominin-1 (CD133). (A): Flow cytometry analysis showing control, unstained cells (left) or the co-expression of prominin-1 and integrin β1 (right). (B): Cells were sorted according to their expression levels of either integrin β1 or prominin-1 and analyzed in a neurosphere forming assay. Bin 91–100 represents the cells expressing the highest levels of a marker; 1–10 represents the lowest. A population of live cells without selection for integrin or prominin-1 expression is also presented for comparison. Cells expressing the highest levels of β1 integrin or prominin-1 were found to form the most neurospheres. Note that integrin β1 is as effective as prominin-1 at enriching for hNSCs. All points, except for bins 51–60 and 61–70, were significantly different from the whole population for both markers (p < .05; n = 3 for β1 integrin; n = 2 for prominin-1), with the five bins with low expression of β1 or prominin-1 forming fewer neurospheres than the unselected population, whereas the three bins with high expression form more neurospheres. Abbreviations: FITC, fluorescein isothiocyanate; PE, phycoerythrin; Pop, population.

Download figure to PowerPoint

From the findings reported here, we conclude that integrin α6β1 is highly expressed by human neural stem cells and can be used as a marker in FACS to enrich for them. These are notable findings for two reasons. First, the extent of this enrichment from neurospheres is equal to that achieved by selecting for prominin-1, currently the most effective marker for selecting hNSCs. Second, they add to the evidence that integrins play an important role in stem cell biology. Previous reports have identified α6 and β1 integrin subunits as being highly expressed by rodent stem cells in testes, liver, skin, and brain [16, [17], [18], [19], [20], [21], [22]23, 38]. Studies have also shown elevated expression of these and other integrins on stem cells within rodent hematopoetic cells (α2, αIIb), cornea (β1 and β4), human colon (β1), and skin (α6,β1) [15, 39, [40], [41], [42]43]. Taken together with the presence of a basement membrane in many stem cell niches [13], these findings indicate that integrins are likely to have a fundamental role in the signaling pathways regulating stemness in human and rodent stem cells.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results and Discussion
  6. Disclosures
  7. Acknowledgements
  8. References
  9. Supporting Information

This study was supported by the Multiple Sclerosis Society (P.E.H.), the NIH-Cambridge Graduate Partnership Program (J.D.L.), the Royal Society (M.A.C.), and the Wellcome Trust (N.G.A.M. and C. ff.-C.).

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results and Discussion
  6. Disclosures
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results and Discussion
  6. Disclosures
  7. Acknowledgements
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
sc05-0595-ffrench_constant_supfig.pdf40KSupplemental Figures

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.