OUP user menu

Glia-Specific Activation of All Pathways of the Unfolded Protein Response in Vanishing White Matter Disease

Barbara van Kollenburg MSc, Jantine van Dijk MSc, James Garbern MD, PhD, Adri A. M. Thomas PhD, Gert C. Scheper PhD, James M. Powers MD, Marjo S. van der Knaap MD, PhD
DOI: http://dx.doi.org/10.1097/01.jnen.0000228201.27539.50 707-715 First published online: 1 July 2006

Abstract

Leukoencephalopathy with vanishing white matter (VWM) is a childhood white matter disorder with an autosomal-recessive mode of inheritance. The clinical course is chronic progressive with episodes of rapid neurologic deterioration after febrile infections. The disease is caused by mutations in the genes encoding the subunits of eukaryotic initiation factor 2B (eIF2B), a protein complex that is essential for protein synthesis. In VWM, mutations in the eIF2B genes are thought to impair the ability of cells to regulate protein synthesis under normal and stress conditions. It has been suggested that the pathophysiology of VWM involves inappropriate activation of the unfolded protein response (UPR). The UPR is a protective mechanism activated by an overload of unfolded or malfolded proteins in the endoplasmic reticulum. Activation of one pathway of the UPR, in which eIF2B is involved, has already been described in brain tissue of patients with VWM. In the present study, we demonstrate activation of all 3 UPR pathways in VWM brain tissue using real-time quantitative polymerase chain reaction and immunohistochemistry. We show that activation occurs exclusively in the white matter, predominantly in oligodendrocytes and astrocytes. The selective involvement of these cells suggests that inappropriate UPR activation may play a key role in the pathophysiology of VWM.

Key Words
  • Childhood white matter disorders
  • Eukaryotic initiation factor 2B
  • Immunohistochemistry
  • qPCR
  • Unfolded protein response

Introduction

An abnormal activation of the unfolded protein response (UPR) in the brain has been described in several different neurodegenerative conditions, including Pelizaeus-Merzbacher disease (PMD) (1), Alzheimer disease (2), and the neurologic disorders caused by unstable trinucleotide (CAG) repeats (3-6). Recently, we added vanishing white matter disease (VWM) to this list (7).

VWM, also called childhood ataxia with central hypomyelination (CACH) (8), is one of the most prevalent inherited childhood white matter disorders (9). The disease has a chronic progressive course with additional episodes of major and rapid neurologic deterioration after minor head trauma and fever. Its main clinical features are cerebellar ataxia and spasticity, whereas cognitive abilities are relatively preserved. Magnetic resonance imaging of the brain shows a diffuse abnormality of the cerebral white matter with evidence of progressive rarefaction and cystic degeneration (9, 10). The disease has an autosomal-recessive mode of inheritance (11).

VWM is caused by mutations in any of the 5 genes encoding the subunits of the eukaryotic initiation factor 2B (eIF2B) (12, 13). The eIF2B protein complex has an essential function in the regulation of protein synthesis. It catalyses the GDP-GTP exchange on another initiation factor, eIF2, that enables the binding of initiator methionyl-transfer-RNA (Met-tRNAi) to the small ribosomal subunit to start protein synthesis (14). The ability to regulate protein synthesis through eIF2 is also part of the cellular stress response or heat shock response, a mechanism induced by a variety of stress conditions, including viral infection, malnutrition, and fever. In the downregulation of protein synthesis under hyperthermia, eIF2B is the main factor involved (15).

The endoplasmic reticulum (ER) is mainly responsible for the folding of polypeptides and posttranslational modification of proteins in the secretory pathway (16). Eukaryotic cells have evolved multiple regulatory mechanisms to maintain ER homeostasis. If the amount of unfolded or malfolded polypeptides in the ER exceeds its folding or processing capacity, the physiological state of the ER is perturbed (16). Signaling pathways, termed the UPR, are activated to protect cells against the accumulation of denatured proteins and subsequently to return the ER to its normal physiological state (16-19). Three ER transmembrane proteins, IRE1, PERK, and ATF6, transduce the signals triggering the UPR (for a schematic overview, see Fig. 1).

FIGURE 1.

An overview of the main signaling pathways of the unfolded protein response. Unfolded proteins present in the endoplasmic reticulum (ER) will bind to BiP, leading to activation of the ER transmembrane proteins IRE1, PERK, and ATF6. Activation of these 3 different pathways to restore the cell homeostasis is called the unfolded protein response.

In an inactive state, the luminal domains of IRE1, PERK, and ATF6 are associated with the ER-resident chaperone BiP (also called GRP78). Unfolded proteins in the ER lumen bind BiP in a competitive manner, leading to dissociation of BiP from IRE1, PERK, and ATF6 (20). Subsequently, oligomerization of IRE1 and PERK result in activation of these proteins (21-23). BiP regulates the activity of 2 independent and redundant Golgi localization sequences (GLS1 and GLS2) on ATF6 that retain the protein in the ER (24). In the absence of BiP, ATF6 is constitutively translocated to the Golgi.

A downstream effect of PERK is the phosphorylation of its substrate, the α-subunit of initiation factor 2 (eIF2α) (25, 26). Phosphorylation of eIF2α inhibits protein translation as it sequesters eIF2B in an inactive state, blocking GDP-GTP exchange on eIF2 (26). The eIF2α phosphorylation and eIF2B inactivation also induce preferential translation of mRNA containing several short upstream open reading frames (uORF) (27) such as the mRNA for the transcription factor ATF4 in mammalian cells (28). Important downstream targets of ATF4 are CHOP (also called GADD153), mainly known as a proapoptotic protein, and GADD34, which enables dephosphorylation of eIF2α (29).

On induction of the UPR, IRE1 recognizes and cleaves 2 specific stem loop sequences in the XBP-1 mRNA, removing an internal 26-nucleotide sequence from the XBP-1 mRNA, which results in a translational frame shift (30, 31). The resulting larger XBP1 protein is a potent transcription factor that activates genes involved in restoring ER homeostasis (32).

On ATF6 translocation to the Golgi, it is cleaved by 2 proteases to release the cytosolic domain of the protein, which then can translocate to the nucleus and activate transcription, leading to enhanced expression of regulatory proteins, including BiP, CHOP, and ER chaperones (16).

Our previous study on activation of the UPR in VWM concerned only one of the 3 known UPR pathways and showed that the PERK-eIF2α-ATF4-CHOP pathway is activated in VWM (7). Increased activation of this pathway was also shown in tunicamycin-treated fibroblasts derived from patients with VWM (33). The increased expression of ATF4 and its downstream target CHOP in VWM cells can be explained by the decreased eIF2B activity and uORF-mediated translational control, as mentioned previously. The activation of PERK is more surprising because this occurs upstream of the regulation of eIF2B activity. Activation of PERK suggests ER perturbation or faulty regulation of UPR activation. In both cases, activation of the ATF6 and IRE1 pathways would be expected as well.

The present study is focused on the question of whether the ATF6 and IRE1 pathways of the UPR are also activated in VWM white matter cells.

Materials and Methods

Total RNA was extracted from frozen brain tissue from 5 patients diagnosed with VWM and from 7 age-matched control subjects (Table 1). Patient and control tissue was frozen in liquid nitrogen immediately at autopsy. As a positive control for activation of the UPR, brain tissue from a patient diagnosed with PMD was included (1).

View this table:
TABLE 1.

The immunohistochemical study was performed using brain samples from the frontal lobe of 4 patients with VWM and 4 age-matched control subjects; the postmortem intervals were chosen to be as low and as similar as possible (Table 1). The brain samples were formalin-fixed and paraffin-embedded. Standard hematoxylin and eosin staining confirmed the presence of both gray and white matter in the sections.

RNA Isolation

Total RNA was isolated from 0.1 g frozen brain tissue containing both gray and white matter. The frozen tissue sample was homogenized using a high-speed agitation blender (Ultra Turrax, IKA, T18 basic; Wilmington, NC) at 18,000 rpm in the presence of TRIzol reagent (RNA-Bee; Campro Scientific, Veenendaal, The Netherlands) and chloroform. Residual DNA was removed using a cleanup protocol (RNeasy Mini protocol for RNA from QIAgen, Venlo, The Netherlands).

Reverse Transcriptase-Polymerase Chain Reaction and Polymerase Chain Reaction

cDNA was synthesized from 5 μg total RNA using random hexanucleotides and Superscript III (Invitrogen, Breda, The Netherlands) according to the vendor's protocol. RNAse H (Invitrogen) was added after the reverse transcription to hydrolyze the RNA template. Specific cDNAs were amplified by polymerase chain reaction (PCR) with primer combinations corresponding to different UPR markers. Primers used (5′-3′ forward and reverse primer, A of ATG [start codon] at +1) were as follows:

XBP1: (374/383) CCTTG TAGTT GAGAA CCAGG, (770/789) GGGCT TGGTA TATATG TGG (as described before 34).

Quantitative Polymerase Chain Reaction

The quantitative (qPCR) experiments were performed using an ABI PRISM 7700 sequence detector (Applied Biosystems, Nieuwerkerk aan de IJssel, The Netherlands). Transcript-specific primers were generated with Primer Express software (Applied Biosystems) and designed to overlap exon-exon boundaries to prevent genomic DNA amplification. The PCR reaction was carried out in a volume of 10 μL using SYBR green PCR mix, 3.0 μM primers, and 0.1 μg cDNA. The PCR program followed the guidelines of the vendor (Applied Biosystems).

The cycle of threshold value (Ct), defined as the cycle number at which the fluorescent emission reaches a fixed threshold during the exponential phase of amplification, was used to calculate the relative expression level of the genes of interest and normalized to the transcript for the housekeeping gene β-actin. Calculations were done as instructed by the vendor and as described previously (35).

qPCR primers (5′-3′ forward and reverse, A of ATG = +1) were as follows:

  • BiP: (584/607) TTATG AGGAT CATCA ACGAG CCTA, (614/634) CCAGG GCCAT AAGCA ATAGC AG

  • CHOP: (−58/−36) CATAC ATCAC CACAC CTGAA AGC, (−29/−7) AGTTG GATCA GTCTG GAAAA GCA

  • ATF4: (−97/−81) TGGCG CTTCT CACG GC, (−68/−47) GTCTT TGTCG GTTAC AGCAA CG

  • ATF6: (221/243) GGGAC ATCAA CAACC AAATC TGT, (264/285) TGGAG AAAGT GGCTG AGGTT CT

  • GADD34: (391/411) GAAGT CAATT TGCAG ATGGC C, (422/441) TCAGA AGGCT GGGAG ACAGG

  • β-actin: (990/1007) GCTCC TCCTG AGCGC AAG, (1045/1064) CATCT GCTGG AAGGT GGACA

Statistical Analysis

qPCR samples were analyzed using a 2-sample Student t-test. Ct values were considered to be significantly different between VWM and control sample if p = 0.05 (with p adjusted according to Bonferroni correction).

Immunohistochemical Staining

Sections of patients and controls were deparaffinized and stained according to a standard immunohistochemical streptavidin protocol (36). Primary antibodies against the following antigens were used: carbonic anhydrase II (CAII) from Rockland (Gilbertsville, PA; 1/15,000 dilution, with retrieval), glial fibrillary acidic protein (GFAP) from DakoCytomation (Carpinteria, CA; 1/15,000), myelin basic protein (MBP) from Chemicon International (Temecula, CA; 1/5,000), CD68 from DakoCytomation (1/500 with retrieval), activating transcription factor 6 (ATF6) from Imgenex (San Diego, CA; 1/200), binding protein GRP78 (BiP) made by the late Dr. T. Schulz (Utrecht University, The Netherlands; dilution 1/15,000), and x-box binding protein 1 (XBP1) from Santa Cruz Biotechnologies (Santa Cruz, CA; 1/150). Biotinylated secondary antibodies were obtained from Vector Laboratories (Burlingame, CA; 1/200 diluted) and streptavidin HRP from Jackson ImmunoResearch (West Grove, PA; 1/1,000). Detection was carried out using the AEC chromogen (ScyTek laboratories, Logan, UT). For double immunostaining, a second enhancer, MACH3 HRP polymer kit from Biocare Medical (Walnut Creek, CA), was used with Vector SG substrate kit for peroxidase (Vector Laboratories) as the second chromogen.

Criteria for Oligodendrocytic Positivity

CAII was used as a marker for oligodendrocytes. It is an established marker for oligodendrocytes, including some that are mature, usually the smaller type I/II oligodendrocytes that myelinate small-diameter fibers in rodents (37). CAII labels approximately 20% to 30% of the oligodendrocytes in human cerebral white matter according to Morris et al (38). In our well-fixed, paraffin-embedded neonatal immunostaining control, approximately 80% to 90% of white matter cells morphologically consistent with oligodendrocytes were labeled with the CAII antibody, but the intensity of the immunoreactivity varied considerably. Some reported that all oligodendrocytes in VWM and controls could be labeled with a CAII antibody (36, 39).

Results

PCR and qPCR Shows Upregulation of Several Unfolded Protein Response Components in VWM Brain

Using a qualitative PCR technique, we found upregulation of the spliced form of XBP1 mRNA in brain tissue from 4 of 5 patients with VWM and a complete absence of this spliced form in 6 control subjects (Fig. 2). Positive control brain tissue was derived from a PMD patient, in which activation of the UPR was shown using immunofluorescence and Northern blot analysis (1).

FIGURE 2.

Splicing of XBP1 mRNA in vanishing white matter (VWM) brain. Frozen brain material (including both gray and white matter) of 6 control subjects (c1-c6) and 5 patients with VWM (p1-p5) was analyzed for splicing of XBP1 mRNA using reverse transcriptase-polymerase chain reaction. Brain material from a patient diagnosed with Pelizaeus-Merzbacher disease (PMD) was used as a positive control. HPRT mRNA levels were included as a loading control. The upper arrowhead indicates the unspliced form of XBP1 mRNA, and the arrowhead indicates the smaller, spliced XBP1 mRNA. In the PMD sample, a doublet for HPRT was observed. The nature of this band is unknown; it does not affect the interpretation of the XBP1 splicing data in VWM samples.

To study all 3 UPR pillars, the expression levels of mRNA for BiP, ATF4, CHOP, GADD34, and ATF6 were determined by qPCR (Fig. 3). The lower ΔCt values for patients with VWM indicate that fewer cycles were needed to reach the threshold and are an indication of upregulation of mRNA in VWM brain. The increase was found to be significant for BiP, CHOP, and GADD34. Figure 4 shows the increased expression separately for all patients per mRNA.

FIGURE 3.

mRNA expression of unfolded protein response markers in brain. Expression levels from the indicated mRNA were determined by quantitative polymerase chain reaction as described in "Materials and Methods." The Y-axis shows the mean ΔCt values; i.e. the mean Ct value (the cycle number at which the threshold is reached) corrected for the Ct values of the housekeeping gene β-actin. The gray bars represent the controls (n = 7); the black bars the patients with vanishing white matter (n = 5). The asterisk indicates a statistically significant (p ≤ 0.05) difference in Ct value.

FIGURE 4.

Gene expression in individual patients with vanishing white matter. To assess the relative gene expression (2-ΔΔCt) of the indicated genes, the ΔCt value from each patient was compared with the average ΔCt value from the 7 controls. The average gene expression of the controls was set at 1, indicated by the horizontal line. Values above 1 indicate an increase in expression and values below 1 indicate a decrease. Total mRNA from brain tissue from a patient with Pelizaeus-Merzbacher disease (PMD) was included as a positive control.

Immunohistochemical Staining Showed Upregulation of ATF6, BiP, and XBP1

To verify the increased expression of UPR markers at the protein level and determine in which cells UPR activation occurred, the protein expression of the UPR markers ATF6, BiP, and XBP1 was assayed in formalin-fixed, paraffin-embedded tissue from patients with VWM and from age-matched control subjects (Fig. 5). Cerebral white matter from the frontal lobe, a region that is normally prominently affected in VWM, was studied.

FIGURE 5.

Immunostaining for unfolded protein response (UPR) markers. Cells positive for ATF6 (A), BiP (B), and XBP1 (C) in white matter from patient 6 (original magnification: 400x) are red. White matter from control no. 8 shows no staining for XBP1 (D); BiP (E), and ATF6 (F) (original magnification: 200×). Cells positive for ATF6 (G), BiP (H), and XBP1 (I) in the white matter from patient 7 (original magnification: 400x) are red. Gray matter from patient 7 shows no staining for XBP1 (J); BiP (K) and ATF6 (L) (original magnification: 200×). The open arrowheads indicate cytoplasmic staining and the full arrows point to nuclear staining.

Compared with their age-matched controls, all patients showed upregulation of at least one of these UPR markers in the frontal lobe. An overview of these results is given in Table 2. Patients 6 and 7 showed the highest signal for all tested UPR antigens. The deep white matter was most severely affected and showed a higher level of UPR protein expression than the better-preserved subcortical areas. Positive staining for the tested UPR proteins was only observed in white matter of patients with VWM (Fig. 5A-C, G-I) and never in white matter of age-matched control subjects (Fig. 5D-F) or in the cortex of patients with VWM (Fig. 5J-L). In agreement with the expected cellular localization, ATF6 and XBP1 were present in both the nucleus and cytoplasm, whereas BiP was only observed in the cytoplasm (40-43).

View this table:
TABLE 2.

The cell types possibly involved in this UPR activation were assessed according to cell morphology. The highest expression for ATF6, BiP, and XBP1 was apparent in oligodendrocytes, identified by their round dense nuclei and perinuclear clear halo, although the presence of an occasional positive astrocyte or even macrophage could not be excluded (Fig. 5). Variability in the expression levels of the UPR proteins was observed among different patients. Whereas ATF6 expression was most pronounced in patient 6 (Fig. 5A versus Fig. 5G), the staining for BiP and XBP1 was strongest in patient 7 (Fig. 5B and C versus Fig. 5H and I, respectively).

To confirm which cell type showed increased expression of UPR proteins, double immunostaining was performed with antibodies against XBP1 and antibodies against GFAP, CAII, or CD68 in patient 7. Figure 6 shows that XBP1 was indeed mainly expressed in oligodendrocytes (Fig. 6A). However, as expected on the basis of the single staining and morphology of the cells as shown in Figure 5, a few double-positive astrocytes (Fig. 6B) and macrophages (Fig. 6C) were also present. Technical problems made double staining for ATF6 and BiP with GFAP, CAII, and CD68 more difficult to interpret, but they confirmed that the UPR was activated in oligodendrocytes and astrocytes (Fig. 6D).

FIGURE 6.

Unfolded protein response activation in glial cells. Brain material from patient 7 was double stained for XBP1 (red) in combination with blue staining for CAII (A), GFAP (B), or CD68 (C) (original magnification: 400×). The open arrowheads point to cells of which an enlarged view is shown on the right. Double stains for BiP (red, patient 7) and for AFT6 (red, patient 6) with CAII (blue) and GFAP (blue) show that cells positive for both antibodies are present within vanishing white matter (D).

Discussion

Our data show that all 3 known UPR pathways were activated in the brain of patients with VWM. This activation is reflected by both an increase in the mRNA level and in the level of protein expression of the respective UPR components. In most patients with VWM, XBP1 mRNA is spliced, whereas we found no splicing in control subjects. In addition, the protein level of XBP1 was increased in VWM brain (Fig. 5). We demonstrate a significant increase in the mRNA levels of BiP, CHOP, and GADD34 in VWM brain tissue using qPCR. The mRNA levels for ATF4 and ATF6 in VWM brain tissue do not display statistically significant differences from control levels (Fig. 3); but using immunohistochemistry, we demonstrate that the protein levels of ATF4 (7) and ATF6 (Fig. 5) are increased. The apparent discrepancies with the qPCR data are indicative of the posttranscriptional regulation of ATF4 and ATF6 expression with enhanced mRNA translation rather than increased mRNA levels as the basis of the increased protein level.

Despite the fact that the basic molecular defect in VWM resides in eIF2B, a ubiquitously expressed complex, the clinical picture and histopathology of VWM are characterized by an almost exclusive involvement of the cerebral white matter. The neurons within the gray matter are essentially preserved. Within the white matter, oligodendrocytes are predominantly affected followed by astrocytes (7, 36, 44, 45). Using immunohistochemistry, we show that the activation of the UPR pathways is restricted to the white matter and does not involve the cortex. In addition, the activation of the UPR is most obvious in oligodendrocytes but is also observed in astrocytes. The UPR is known to generate both prosurvival and proapoptotic signals. Faulty regulation of this pathway may underlie the abnormal proliferation and apoptotic loss of oligodendrocytes reported in VWM (9, 10, 36, 39, 46-52) and the abnormal morphology and dysfunction of astrocytes (45). It is remarkable that we did not see any positive staining for UPR markers in the cortical oligodendrocytes and astrocytes.

The UPR has been invoked in the pathophysiology of several divergent neurologic disorders, including Alzheimer disease, PMD, and the neurologic disorders caused by unstable CAG repeats (3-6). The UPR involvement, however, appears to be different in each disorder. In Alzheimer disease, the UPR has been described as being downregulated by some authors (2, 53) but activated by others (54); both groups invoked increased ER stress to explain the pathophysiology of the disease. In PMD, there is evidence that the accumulation of unfolded or malfolded proteins exceeds the processing capacity of the ER. The CAG repeat codes for a polyglutamine repeat. An expanded repeat causes misfolding of the mutant protein and formation of insoluble aggregates. Proteins with expanded polyglutamine repeats disrupt the ubiquitin-proteasome pathway and lead to a disturbance of ER-associated degradation. The consequent accumulation of misfolded proteins in the ER has been shown to activate the UPR (55). Another disease known to affect the UPR is ataxia-telangiectasia, in which the UPR is thought to be induced because of oxidative damage to proteins and lipids (56).

The most likely mechanism for the activation of the UPR in VWM is different. Reduced eIF2B activity in VWM (7, 33, 57) induces enhanced expression of ATF4 by a mechanism involving uORFs in the ATF4 mRNA (58). Increased ATF4 expression has been confirmed in experiments with mutated eIF2B genes in transfected cells (59), in stressed VWM patient-derived fibroblasts (33), and in VWM patient-derived white matter (7). Enhanced expression of the ATF4 protein leads to induction of the downstream effector CHOP, which is known to sensitize cells to ER stress (60). Consequently, in VWM cells expressing increased amounts of CHOP, minor stresses could trigger the UPR, whereas these stresses would be insufficient for UPR activation in healthy cells. Inappropriate activation of the UPR may contribute to abnormal activation of cell survival and cell death signaling pathways. The different mechanism of UPR activation in VWM as compared with Alzheimer disease, PMD, and polyglutamine diseases could explain the stress sensitivity of patients with VWM, which is not seen in patients with these other disorders. Why oligodendrocytes and astrocytes are selectively involved in VWM has yet to be explained.

All of the patients display a unique combination of increased UPR markers (Fig. 4). Interestingly, this combination does not seem to be specific for a certain genotype or phenotype. For example, patients 1 and 4 are siblings with a very similar clinical progression. However, patient 4 shows XBP1 processing, whereas patient 1 does not (Fig. 2). Patients 2 and 3 are also siblings with a similar phenotype and their pattern of UPR mRNA expression is different as well. It is difficult to exclude an effect of the disease phase (acute versus chronic) on mRNA expression, because we do not yet know how the disease evolves in the VWM brain. Nor can we completely eliminate an effect by the cause of death or the antecedent events around the time of death such as hypoxia-ischemia or infection. However, it is important to note that the cause of death does not seem to affect the expression of XBP1, ATF6, or BiP mRNA because the controls for the qPCR include material from patients with both sudden and prolonged death and upregulation of the UPR-related mRNA is not observed in either.

In conclusion, our study demonstrates that not only the PERK pathway of the UPR, in which eIF2B plays a role, is activated but that all 3 pathways of the UPR are activated in patients with VWM. We demonstrate that activation of the UPR is restricted to the white matter and that enhanced expression of UPR markers is almost exclusively observed in the cells that are known to be primarily involved in the pathology of VWM: oligodendrocytes and astrocytes. These data suggest that inappropriate and total activation of the UPR may play a key role in the pathophysiology of VWM.

Acknowledgments

The authors thank all of their families for their cooperation. They specifically express their gratitude to patient 5, who personally granted them permission for brain autopsy for scientific purposes. The authors also thank Frances Vito (from the University of Rochester Medical Center, Rochester, NY) for help with immunohistochemical staining. The authors thank Jaap van Veldhuisen and Ron Otsen for graphic support. Human tissue was obtained from the Brain and Tissue Bank for Developmental Disorders at the University of Maryland, Baltimore, and the Human Brain and Spinal Fluid Resource Center, Los Angeles.

Footnotes

  • Financial support for this work was provided by "Princes Beatrix Fonds" (PBF) (grant MAR01-0201), "Stichting Spieren voor Spieren," Dr. W. M. Phelps Stichting (grant 03.030), Stichting Onderzoek Stofwisselingsziekten, and the Optimix Foundation for Scientific Research.

References

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
View Abstract