Hematopoietic stem cells from induced pluripotent stem cells – considering the role of microRNA as a cell differentiation regulator Free

Stem cells are undifferentiated cells that present an indeterminate expansion potential to produce progeny through self-renewal or differentiation (Sakaki-Yumoto et al., 2013). Furthermore, stem cells have a low-turnover profile, in contrast to their differentiated progeny (Eckfeldt et al., 2005; Cheng et al., 2000). We focus here on the role of miRNAs in cell reprogramming and induced pluripotent stem cell (iPSC) differentiation. The first study to demonstrate the formation of iPSCs upon viral-mediated transduction of murine embryonic and adult fibroblasts with octamer-binding transcription factor 4 (OCT-4; also known as POU5F1), sex-determining region Y-box 2 (SOX-2), v-myc avian myelocytomatosis viral oncogene homolog (MYC) and Kruppel-like factor 4 (KLF-4) was published in 2006 (Takahashi and Yamanaka, 2006). Although iPSCs are similar in morphology and pluripotent potential to embryonic stem cells (ESCs), these types of stem cells are not identical and their gene expression profile, microRNA (miRNA) expression and epigenetic markers are distinct, indicating that the stem cell differentiation process needs to be further investigated.

Stem cells have proven to be a powerful tool in studies aimed at understanding in vitro cell differentiation, and advances in developing cell reprogramming protocols have meant that these cells can now be induced to differentiate into a number of different tissues in vitro (Takayama et al., 2010; Teng et al., 2014; Menon et al., 2016). iPSC generation has now been achieved by several methods, including through integration of viruses and episomal plasmids (Meng et al., 2012; Slamecka et al., 2016). Key aspects of iPSC production, such as the target tissue, reprogramming factors, method of cell delivery, culture conditions and the biological assays to confirm the resulting cell pluripotency potential, are all time-consuming, arduous and expensive (Maherali and Hochedlinger, 2008). In fact, the technology used to integrate viruses for the generation of iPSCs represents the main bottleneck for the therapeutic application of iPSCs owing to the possibility of viral vectors being integrated into the genome, which can result in tumorigenesis (Maherali and Hochedlinger, 2008). By contrast, episomal plasmids do not integrate in the genome and typically disappear from iPSCs after 10 to 14 passages (Chou et al., 2011; Meng et al., 2012).

Somatic cell reprogramming and iPSC differentiation have the potential to be used in a wide range of therapeutic applications in vitro, including disease modelling, drug screening and cellular replacement therapies (Maherali and Hochedlinger, 2008; Giani et al., 2016; Tiyaboonchai et al., 2014); however, the culture conditions of stem cells and the particular protocols applied have a great influence on whether the desirable final results are obtained (Fig. 1).

During differentiation, it is expected that stem cells lose the expression of the pluripotency-related genes OCT4 and NANOG and begin to express markers associated with differentiation, such as GATA4 and GATA6 (Miyamoto et al., 2015). An important issue regarding epigenetic markers in stem cells is that the current methods of iPSC cultivation allow the maintenance of the epigenetic profile over a long period (Philonenko et al., 2017). Such epigenetic control may be also mediated by miRNAs (Fig. 1).

miRNAs are endogenous small non-coding RNA (ncRNAs) consisting of 20 to 22 nucleotides that impair protein expression by binding to mRNAs and interfering with their translation (Ambros, 2004). In this way, miRNAs are involved in fundamental biological processes, including tissue development, cell differentiation, proliferation and apoptosis. Besides all this knowledge with regard to miRNA function, their role in stem cells differentiation is not that well understood (Kim et al., 2017). Here, we aim to highlight miRNAs as a relevant regulator of iPSC generation and reprogramming.

Protocols for cell differentiation in vitro aim to generate specific cell types from undifferentiated stem cells by using embryoid bodies as an initial step (Brickman and Serup, 2017). Embryonic bodies are a three-dimensional, multicellular aggregates consisting of the three germ-layers – endoderm, mesoderm and ectoderm – and are obtained from spontaneous differentiation of iPSCs when they are cultured under low-oxygen conditions (Hawkins et al., 2013).

Besides the ability to obtain embryonic bodies from iPSCs, another advantage in using iPSCs for in vitro cell differentiation is that technical manipulations, such as selection of clones or cell sorting, do not interfere with their ability to differentiate into hematopoietic tissue (Philonenko et al., 2017). Even if iPSCs require genetic manipulations (e.g. the introduction of reprogramming factors to induce cell pluripotency and self-renewal, such as OCT-4, SOX-2 and KLF-4) their properties as iPSCs are maintained due to their high genetic and epigenetic stability (Philonenko et al., 2017).

 Although numerous studies have linked iPSCs or iPSC-differentiated cells with cell-based therapy, there are, however, still challenges and methodology limitations to be overcome. We consider the main challenges in generating iPSCs to be: choosing the ideal somatic cell source for iPSC generation, the reprogramming method used, and the control and the optimization of iPSC differentiation (Box 1).

Here, we discuss the processes and protocols currently employed to obtain hematopoietic stem cells (HSCs) in vitro from iPSCs, and also highlight the potential roles of miRNAs in cell reprogramming and differentiation.

The current therapies for hematological neoplasms are chemotherapy, immunotherapy, tyrosine kinase inhibitors and transplantation of hematopoietic stem cells (HSCs) obtained from bone marrow, peripheral blood or umbilical cord (Jaramillo et al., 2017; Ye et al., 2017; Baron and Nagler, 2017).

Patients who receive allogeneic transplants of bone marrow or umbilical cord may exhibit a rejection of the bone marrow transplanted cells due to graft-versus-host disease (GVHD), toxicity owing to the conditioning regimen used, disease relapse or infections (van Bekkum and Mikkers, 2012). In addition, patients receiving HSCs by allogeneic bone marrow transplantation should be under 55 years old and should have a compatible human leukocyte antigen (HLA) donor, which restricts the applicability of the procedure.

In cases of autologous HSC transplantation, patients are not affected by severe GVHD nor by risk of rejection. However, the disease relapse index is higher in comparison with allogeneic HSC transplantation, as no GVHD and graft-versus-leukemia effect (GVL) will occur (i.e. immune response against neoplastic cells) and any remaining leukemic cells could then induce the disease relapse. Taken together, all these disadvantages of bone marrow HSC transplantation highlight the requirement of an alternative source of HSCs, which aims to reduce the rate of HSC rejection, disease relapse and bone marrow failure syndromes (graft failure or poor graft function), as well as increasing the possibility of obtaining HSCs more easily (van Bekkum and Mikkers, 2012; Masouridi-Levrat et al., 2016). In this regard, iPSCs cells represent a suitable source that may be used to generate sufficient amounts of HSCs in vitro with limited immunogenicity (Araki et al., 2013).

Despite the success in obtaining in vitro hematopoietic cells from iPSCs, the strategy used is laborious and expensive (Yang et al., 2017). Furthermore, there is a the lack of a specific surface marker for the human hemangioblast, a single mesodermal cell that gives rise to blood cells and endothelium, which hinders the derivation of HSCs from iPSCs in practice (Lacaud and Kouskoff, 2017).

A particular advantage for the use of iPSC-derived HSCs is that they do not induce GVHD because they are autologous. However, these cells nevertheless exhibit some bias, such as their inability to self-renew in culture and a failure to engraft and survive long-term after the transplant (Shepard and Talib, 2014). Another advantage of iPSCs as a source of HSCs is their ability to differentiate into primitive cells, such as erythrocytes, which express fetal-type hemoglobin, and definitive cells including lymphocytes (Seiler et al., 2011).

iPSCs might be also useful as disease models, as the maintenance of human primary cells in culture over long periods is difficult, and the use of animal models often involves interspecies variabilities (De Lázaro et al., 2014). Moreover, numerous reports have demonstrated the therapeutic potential of iPSCs in diverse hematological conditions, such as myelodysplastic syndrome (MDS), sickle cell anemia and hemophilia (Kotini et al., 2015; Hanna et al., 2007; Park et al., 2014). iPSCs derived from a variety of monogenic diseases can accurately recapitulate disease phenotypes in vitro when differentiated into disease-relevant cell types (Hamazaki et al., 2017).

Based on these properties, iPSCs should be considered as a potential and relevant source for HSCs in vitro. Below, we will discuss the current protocols used for hematopoietic differentiation of pluripotent stem cells, both ESCs and iPSCs. We will also emphasize the role of miRNAs as reprogramming and differentiation regulators.

The derivation of blood cells from iPSCs has been reported by several research groups, including ours (reviewed by Zhang, 2013; Smith et al., 2013; Sweeney et al., 2016).

The embryoid bodies formed from iPSCs, as well as the culture of iPSCs in a differentiation medium supplemented with human bone morphogenetic protein 4 (hBMP-4), human vascular endothelial growth factor (hVEGF) and human WNT3A Wnt family member 3A (hWnt3a), have been used to differentiate iPSCs into the hematopoietic lineage in vitro (Sweeney et al., 2016; Smith et al., 2013). Dissociation of embryonic bodies and culture of their cells with OP9 cells (a marrow stromal cell line) have been successfully used to obtain myeloid precursors, macrophages, eosinophils and neutrophils (Sweeney et al., 2016). These authors also reported that the potential for colony formation in iPSCs is associated with the co-expression of the hematopoietic markers CD34 and CD45 (also known as PTPRC) during iPSC differentiation; however, these iPSCs were not able to promote long-term hematopoietic engrafting in mice (Sweeney et al., 2016). In order to generate hematopoietic cells from iPSCs, growth factors, such as prostaglandin-E2 (PGE2) and StemRegenin 1 (SR1), have been used to increase the iPSC differentiation efficiency and to give rise to a long-term HSC phenotype (Zhang, 2013).

Another approach that has been reported for the generation of hematopoietic cells in vitro is the co-culture of ESCs with pre-adipocytic stromal cells, favoring conditions conducive for hematopoietic differentiation without the need for to generate embryonic bodies (reviewed by Seiler et al., 2011).

However, the differentiation of iPSCs in vitro might be impaired by the transgenes that are used to reprogram somatic cells, such as SOX-2, OCT-4 and KLF-4, as these have been shown to reduce the ability of mesodermal-like cells to differentiate in hematopoietic progenitors (Ebina and Rossi, 2015). In fact, SOX-2 downregulation is important in hematopoietic development, because its expression appears to be inversely related to the hematogenic potency of a cell (Seiler et al., 2011).

In addition to the need of establishing a standardized protocol for the generation of HSCs in vitro, there is also a great demand for mature blood cells, such as erythrocytes and platelets, that have been generated from hematopoietic progenitor cells, obtained either from the bone marrow, peripheral blood or cord blood. Pluripotent HSCs and early multipotent progenitors (MPPs) all originate from erythroblasts, which generate erythrocytes (Palis, 2008).

Mouse ESCs have been successfully used for in vitro differentiation into erythroid cells by using two strategies (Nakano, 1996; Carotta et al., 2004; Kitajima et al., 2003). The first one utilizes disaggregated embryonic bodies that have been cultured with erythropoietin (EPO) and Kit ligand (KL-1) to stimulate the growth and differentiation of erythroid progenitors (Carotta et al., 2004). EPO stimulates cell growth by binding to EPO-R in burst-forming unit-erythrocytes (BFU-Es) and colony-forming unit-erythrocyte (CFU-Es) (Elliott et al., 2008; Metcalf, 2008). KL-1, in synergy with other cytokines, stimulates growth of hematopoietic progenitors in vitro and increases blood cell production in vivo in animals (Broxmeyer et al., 1991). The protocol involving EPO and KL-1 requires 10 days of culture to obtain erythroid colonies (Carotta et al., 2004). Using only EPO without KL-1 results in primitive erythroid colonies, which are characterized by their nucleated morphology and expression of embryonic globin (Carotta et al., 2004). By contrast, definitive erythroid colonies are composed of cells that express adult globin, which is an important feature when considering the use of these cells for hemotherapy (Carotta et al., 2004).

The second protocol uses OP9 cells, which are able to create an environment that can induce hematopoietic differentiation when co-cultured with ESCs. After 5 days of co-culturing, the presence of colonies formed of hematopoietic tissue can be noticed (Kitajima et al., 2003). In this setup, the addition of cytokines, such as EPO and KL-1, also increases the potential of the culture to produce erythroid colonies (Kitajima et al., 2003). Indeed, after 14 days, definitive erythroid colonies are obtained, which can be separated from the primitive cells by simply removing the nonadherent cell population representing the primitive erythroid progenitors (Carotta et al., 2004).

In addition to the above protocols for generating mice erythroid cells in vitro, human mature erythroid cells had already been obtained through human ESC differentiation in vitro nearly a decade ago (Lu et al., 2008). Although enucleation, the final step in the development of mature erythrocytes remains poorly understood, these authors were able to generate oxygen-carrying erythrocytes on a large scale. Furthermore, the obtained erythroid cells showed the capacity to express the adult β-globin chain upon further maturation in vitro, indicating their potential functionality (Lu et al., 2008). Blood cells that have been generated in vitro could serve as a disease model and, importantly, also pave the way for the large-scale manufacture of red blood cells, which is a global challenge in order to provide a safe supply of transfusable erythrocytes (Timmins et al., 2011).

Although all the methods mentioned above yield functional hematopoietic cells, they also have limitations including the presence of byproducts from murine feeder cells, the need for long culturing periods for embryonic body formation, and the presence of animal-derived culture products, such as bovine serum; all of these will need to be overcome for any future clinical applications (Kim et al., 2017).

Although the literature shows that it is feasible to generate blood cells in vitro, the underlying molecular mechanisms, such as how the Wnt signaling pathway, iron homeostasis and hypoxia affect the expression of transcriptional factors that contribute to cell fate, are still unclear (Tsiftsoglou et al., 2009a,b; Undi et al., 2016).

The current view with regard to hematopoietic differentiation in vitro demonstrates that ESCs and iPSCs are useful sources for blood cells production. However, it would be desirable if blood cell expansion protocols were feeder free, as this would greatly simplify the commercial scalability, as well as reduce the cost of blood cell production.

Besides this knowledge and the considerable advances in protocols that drive derivation of hematopoietic stem cells (HSCs) from iPSCs, the generation of robust transplantable HSCs and mature blood cells production from iPSCs remains elusive. Researchers should discovery strategies to overcome the challenges and obstacles to produce a functional HSC in vitro. It is also important to optimize methodologies to control or stabilize cell epigenetic states and understand the pathways that are linked to functional HSC generation and differentiation. Here, we point out the role of miRNAs in somatic cell reprogramming and HSC differentiation.

As cell reprogramming and stem cell differentiation may be affected by miRNAs, we discuss here how this epigenetic regulator could interfere with these processes (Wang et al., 2015; Ong et al., 2016).

miRNAs have been reported to contribute to somatic cell reprogramming by upregulating the expression of the pluripotent reprogramming, factors OCT-4, SOX-2, KLF-4 and MYC, thus promoting reprogramming (Vitaloni et al., 2014; Wang et al., 2015; Hu et al., 2013). In addition, miRNAs can also induce the pluripotent state of somatic cells in the absence of exogenous factors, as has been reported for miR-302, which inhibits nuclear receptor subfamily 2 group F member 2 (NR2F2) and promotes pluripotency by upregulating OCT-4 (Vitaloni et al., 2014; Wang et al., 2015; Hu et al., 2013).

Furthermore, Miyoshi et al. have described the possibility of reprogramming mature cells by inducing the ectopic expression of miR-200c, miR-302 and miR-369 in human adipose stromal cells and human dermal fibroblasts (Miyoshi et al., 2011). These miRNAs were able to promote iPSCs pluripotency and self-renewal by inducing the overexpression of OCT-3, stage-specific embryonic antigen 3 and 4 (SSEA-3 and -4), SOX-2 and NANOG transcription factors, which resulted in the establishment of a stem-cell-like state (Miyoshi et al., 2011; reviewed by Yao, 2016).

Moreover, in mice, miR-93 and miR-106 have been shown to also give rise to a pluripotent state of somatic cells by increasing the mRNA levels of OCT-4, SOX-2, KLF-4 and MYC (denoted OSKM) through targeting their regulator, the CDKN1A gene (encoding p21), which promotes the formation of iPSC colonies (Li et al., 2011). miR-302 appears to maintain the renewal capacity and pluripotency of human ESCs, as this miRNA was found to be associated with the inhibition of premature zygotic cell differentiation during embryonic development (Miyoshi et al., 2011). Finally, the miR-302/367 cluster is a direct target of OCT-4 and SOX-2 in ESCs and iPSCs (Card et al., 2008) and has been shown to be able to directly reprogram mouse and human somatic cells without the need for any pluripotent stem cell transcription factors (Anokye-Danso et al., 2011).

It has been suggested that miRNA expression may be antagonized by pluripotent factors such as MYC (Chang et al., 2008). Indeed, Yang et al. transduced mouse embryonic fibroblasts (MEFs) with OCT4, SOX2 and KLF4 (denoted OSK) in the presence or absence of MYC and observed that expression of miR-let-7a, miR-16, miR-21, miR-29a and miR-143 decreased during reprogramming, confirming that MYC is involved in the regulation of miRNA expression in MEFs and is able to sustain MEFs reprogramming. Moreover, when the authors tested the efficiency of the process in the presence of a miR-21 inhibitor, they observed a larger number of iPSC colonies, confirming that miRNA inhibition enhances cell reprogramming (Yang et al., 2011).

Similarly, downregulation of miR-29a improves reprogramming in mouse fibroblasts, whereas its overexpression reduced it (Hysolli et al., 2016). Thus, this result suggests that depletion of miR-29a alters the DNA methylation profile in somatic cells and may affect the expression of the genes responsible for cell reprogramming (Hysolli et al., 2016). These findings demonstrate the capacity of miRNAs to regulate DNA methylation and/or demethylation and highlight the need for further studies in order to better characterize the iPSC methylome.

In order to evaluate the importance of Dicer, which is involved in miRNA biogenesis, for cell reprogramming, Kim et al. investigated the reprogramming of Dicer-null MEFs, which do not have a functional miRNA biogenesis pathway (Kim et al., 2012). Use of OCT-4, SOX-2, KLF-4, MYC and LIN-28 to reprogram Dicer-null MEFs was unsuccessful, but reprogramming could be achieved when the human Dicer homolog was introduced in Dicer-null MEFs before their differentiation, raising the hypothesis that miRNAs are essential for iPSC generation (Kim et al., 2012; Judson et al., 2009; Pfaff et al., 2017).

Altogether, the role of miR-125a, miR-125b, miR-155, miR-181, miR-221, miR-222, miR-223 in HSC self-renewal and differentiation has been extensively explored in in vitro experiments (Chen, 2004; Fazi et al., 2005; Felli et al., 2005; Shaham et al., 2012; Kozakowska, et al., 2014, Yao, 2016; Wojtowicz et al., 2016) (Fig. 2). For instance, the ectopic expression of miR-125a in human multipotent progenitors (MPPs) increased their self-renewal, and these cells were also able to repopulate transplanted mice with a robust long-term multi-lineage engraftment (Wojtowicz et al., 2016). Conversely, downregulation of miR-125a decreased HSC self-renewal and so impaired the production of blood cells (Wojtowicz et al., 2014, 2016) (Fig. 2). Furthermore, overexpression of miR-125b has been shown to promote the generation of blood cells such as megakaryocytes in vitro and therefore constitutes a potential therapeutic target in blood disorders (Shaham et al., 2012). However, it is worth noting that the expression of high levels of miR-125b alone in mice causes a very aggressive form of transplantable myeloid leukemia. miR-125b exerts this effect by upregulating the number of common myeloid progenitors, while inhibiting the development of pre-B cells. In this context, miR-125b targets Lin28A, whose downregulation can mimic the preleukemic state in mice (Chaudhuri et al., 2012).

In addition, miR-155, miR-221 and miR-222 also regulate megakaryocytic (Georgantas et al., 2007) and erythroid differentiation (Felli et al., 2005). miR-155 impaired both processes when it was transduced into K562 cells, a model cell line for human hematopoiesis (Georgantas et al., 2007). Here, miR-155 was shown to interfere with the generation of colonies from hematopoietic progenitor cells (CD34⁺) that have been transduced with miR-155, thereby giving rise to only a few myeloid and erythroid colonies, demonstrating its role as a negative regulator of normal myelopoiesis and erythropoiesis (Georgantas et al., 2007). Furthermore, the ectopic expression of miR-221 and miR-222 in CD34⁺ cells from cord blood results in the impaired proliferation and accelerated differentiation of erythroid cells, coupled with down-modulation of Kit protein. miR-221 and miR-222 exert this effect by modulating the expression of the Kit receptor, an important protein involved in HSC maintenance, erythropoiesis upregulation and erythroleukemic cell expansion (Felli et al., 2005; An et al., 2016) (Fig. 2).

The differentiation of lymphoid cells might also be regulated by miR-181, which is highly expressed in thymus (Li et al., 2007; Henao-Mejia et al., 2013). Li et al. show that increasing miR-181a expression in mature T cells augments their sensitivity to peptide antigens, whereas inhibiting miR-181a expression in the immature T cells reduces sensitivity and impairs both positive and negative selection (Li et al., 2007). Another study has also reported the relevance of miR-181 in natural killer T cell (NKT cell) ontogenesis and lymphocyte T development (Henao-Mejia et al., 2013). The authors described a severe defect in lymphoid development and T cell homeostasis in miR-181-deficient mice, which was linked to deregulation of the phosphoinositide 3-kinase pathway. Similarly, miR-223 has been identified as a hematopoietic-specific miRNA and has crucial functions in myeloid and lymphoid lineage development and their cell fate due to its location in the bone marrow, which contains HSCs and hematopoietic cells at various stages of maturation (Chen, 2004).

Thus, the abovementioned miRNAs might all be involved in regulating differentiation of myeloid and lymphoid hematopoietic cells.

The modulation of miRNAs could also be exploited for the treatment of solid cancer and leukemias (Sun et al., 2017; Fan et al., 2017; Lu et al., 2016). Indeed, miR-223 modulates the differentiation of human myeloid progenitor cells during granulocytic differentiation of acute promyelocytic leukemia (APL) cells in response to their treatment with retinoic acid, as shown both in vitro and in APL patients (Fazi et al., 2005). Myelopoiesis, the generation of myeloid leukocytes, which includes granulopoiesis, monocytopoiesis and megakariocytopoiesis, is controlled by a unique combination of transcription factors, such as nuclear factor 1 A-type (NFI-A), CCAAT-enhancer-binding proteins (C/EBPα), core-binding factor-β (CBF-β) and retinoic acid receptor-α (RAR-α), which cooperatively regulate promoters and enhancers present on myeloid-specific-genes. Fazi et al. showed that granulocytic differentiation is controlled by a regulatory circuitry involving miR-223 and the NFI-A and C/EBPα transcription factors (Fazi et al., 2005). NFI-A maintains miR-223 at low levels, whereas, following retinoic acid (RA)-induced differentiation, it is replaced by C/EBPα, which in turn upregulates miR-223 expression (Fazi et al., 2005). In acute promyelocytic leukemia (APL) cells, overexpression of miR-223 induced by treatment with RA allows leukemic promyelocytes and leukemic progenitors to differentiate. The resulting mature leukemic myeloid cells expressing high levels of miR-223 are more sensitive to cell death, and thus chemotherapy, and, consequently, patients exhibit disease remission in response to retinoic acid (Fazi et al., 2005).

In summary, these observations have clearly demonstrated the role of miRNAs in hematopoiesis, cell reprogramming and differentiation regulation. In addition, these observations have pointed out how miRNAs can be used as a tool or strategy to improve the success of the methods used for iPSCs generation in vitro.

The generation of iPSCs from somatic cells of a patient or from a healthy donor is highly relevant for cell-based therapy approaches and for understanding the molecular and cellular mechanisms involved in disease pathogenesis. It is also worthwhile pointing out that iPSCs may also be used for development and testing of new therapeutic agents in vitro. Here, we have discussed the challenges of some protocols used for iPSC generation in vitro and highlighted the role of miRNAs in iPSC cell reprogramming and hematopoiesis regulation.

The major challenge in this field is to obtain mature functional peripheral blood cells or other tissue-specific mature cells based on iPSC differentiation in vitro.

In the future, the protocols for iPSC generation must be further optimized to allow iPSCs to be used as an unlimited source in disease therapy, for instance, as a source of HSCs. To that end, new strategies must be developed to improve the current protocols for cell reprogramming and differentiation, including the use of ectopic miRNAs as epigenetic regulator.

Clearly, the role of miRNAs in disease pathogenesis and, more generally, in the regulation of cell differentiation and proliferation needs to be explored in greater depth. Considering their relevance, we anticipate that in the near future, miRNAs will be used as biomarkers for several diseases (e.g. autoimmune diseases, leukemia, myeloproliferative neoplasm and solid tumors), and as targets for therapy.

We also thank you, Sandra Navarro for the figure designs, Fernanda Teresinha Udinal and Andy Alastair Cumming for carefully reading the manuscript.