Vascular pathology in the aged human brain
- First Online:
- Cite this article as:
- Grinberg, L.T. & Thal, D.R. Acta Neuropathol (2010) 119: 277. doi:10.1007/s00401-010-0652-7
Cerebral atherosclerosis (AS), small vessel disease (SVD), and cerebral amyloid angiopathy (CAA) are the most prevalent arterial disorders in the aged brain. Pathogenetically, AS and SVD share similar mechanisms: plasma protein leakage into the vessel wall, accumulation of lipid-containing macrophages, and fibrosis of the vessel wall. CAA, on the other hand, is characterized by the deposition of the amyloid β-protein in the vessel wall. Despite these differences between CAA, AS and SVD, apolipoprotein E (apoE) is involved in all three disorders. Such a pathogenetic link may explain the correlations between AS, SVD, CAA, and Alzheimer’s disease in the brains of elderly individuals reported in the literature. In addition, AS, SVD, and CAA can lead to tissue lesions such as hemorrhage and infarction. Moreover, intracerebral SVD leads to plasma protein leakage into the damaged vessel wall and into the perivascular space resulting in a blood–brain barrier (BBB) dysfunction. This SVD-related BBB dysfunction is considered to cause white matter lesions (WMLs) and lacunar infarcts. In this review, we demonstrate the relationship between AS, SVD, and CAA as well as their contribution to the development of vascular tissue lesions and we emphasize an important role for apoE in the pathogenesis of vessel disorders and vascular tissue lesions as well as for BBB dysfunction on WML and lacunar infarct development.
KeywordsAtherosclerosisSmall vessel diseaseCerebral amyloid angiopathyBrain infarctionHemorrhagesWhite matter lesions
The three most important vessel disorders in the elderly brain are cerebral atherosclerosis (AS), small vessel disease (SVD), and cerebral amyloid angiopathy (CAA). All three disorders can lead to infarction and hemorrhage [56, 87, 102]. SVD is also associated with white matter lesions (WMLs) . Infarction, hemorrhage, and WMLs cause the destruction of brain tissue and, thereby, frequently cause neurological symptoms.
This review is aimed at identifying pathogenetic mechanisms involved in AS, SVD, and CAA that may explain their correlation with one another and with vascular lesions, such as hemorrhage, infarction, and WMLs. To address this aim, we will first summarize the current knowledge about these vessel disorders. Second, we will discuss the correlation between these vessel diseases and the proteins involved in the pathogenesis of all three disorders. Third, we will summarize the current concepts about the involvement of apolipoprotein E (apoE) in these vessel disorders, and discuss potential pathogenetic mechanisms that may represent a link among them. Then, the relationship between the vessel disorders and vascular brain lesions will be reported and pathogenetic mechanisms will be discussed.
Pathology of major vessel disorders in the aging brain
Distribution of frequencies of autopsy cases with a given extent of atherosclerosis (AS) within the arteries of the circle of Willis (CaW) among different age groups
Extent of AS
Frequency of cases/% in the age group
81 years and older
1–25% of CaW vessels with AS
26–50% of CaW vessels with AS
51–75% of CaW vessels with AS
76–100% of CaW vessels with AS
ApoE and its receptors are critically involved in the pathogenesis of AS. ApoE-knockout mice and low-density lipoprotein (LDL) receptor knockout mice develop AS [8, 156]. Dysfunctional uptake of LDLs may, thereby, lead to the accumulation of oxidized LDLs in the atherosclerotic vessel wall . Thus, oxidized LDLs may be candidates to trigger the development of atherosclerotic plaques. ApoE and LDL receptors [e.g. LDLR and A2M receptor/LDL receptor-related protein (LRP = CD91)] were found in AS plaques (Fig. 1c) [8, 43]. The apoE ε4 allele is controversially discussed as a possible genetic risk factor for AS [64, 75, 116]. The second ligand of LRP, the plasma protein A2M, also accumulates within AS plaques (Fig. 1d)  and an association of its deletion/insertion polymorphism with the expansion of AS plaques within the circle of Willis has been reported . The accumulation of A2M within the AS plaque did not correlate with the type of the AS lesions  suggesting that plasma protein influx including A2M may play a role in the initiation of AS plaques but not for plaque rupture. In summary, apoE, LDL receptors and A2M are involved in the pathogenesis of AS. Leakage of plasma apoE and A2M possibly contributes to the development of AS plaques.
Small vessel disease
Small vessel disease encompasses degenerative alterations in the vessel wall of the small arteries and arterioles that are assumed to be pathogenetically linked [51, 67, 142]. SVD includes changes also described as small vessel arteriosclerosis/atherosclerosis, arteriolosclerosis, and lipohyalinosis [51, 67, 142]. These three subforms of SVD and how they are pathogenetically linked to one another will be discussed hereunder. CAA is not included in this group basically because of differences in the pathogenesis and the location of affected vessels.
SVD is first seen in the arteries of the basal ganglia, mainly in the putamen and the globus pallidus, which exhibit small vessel AS and lipohyalinosis. Secondly, small white matter arteries become affected by arteriolosclerosis or lipohyalinosis. Leptomeningeal arteries of the hemispheres and the cerebellum develop small vessel AS in parallel. Brain stem arteries, including those located between the pontine nuclei, usually develop lipohyalinosis or arteriolosclerosis only in the end stage of SVD . Cortical vessels are usually free of SVD . Enlarged perivascular spaces are often associated with SVD-affected vessels. As such, all subforms of SVD contribute to this hierarchical expansion of SVD changes throughout the brain and seem to be pathogenetically linked.
Arterial hypertension and diabetes mellitus are risk factors for SVD . However, this association is lacking in a significant number of cases . Plasma protein leakage into the vessel wall including apoE and A2M is a common feature in all three subforms of SVD (Fig. 2c–e, h, j). Genetically, an association between SVD and the apoE ε4 allele has been reported .
Cerebral amyloid angiopathy
CAA begins in cortical and leptomeningeal vessels of neocortical areas and then, expands into vessels of allocortical areas and the cerebellum. In single cases, even white matter arteries and vessels of the basal ganglia, the diencephalon, and/or the brain stem exhibit CAA [129, 132]. This hierarchical pattern resembles the one described for Aβ plaques [2, 131, 134].
Physiologically, brain-derived Aβ is drained along the perivascular space [15, 147] and the vascular basement membranes [16, 146]. Alterations of this clearance pathway may result in the deposition of Aβ near the basement membrane . In addition, smooth muscle cell-derived Aβ is potentially capable of contributing to the deposition of Aβ in the vessel wall [85, 86]. ApoE plays an important role in the pathogenesis of CAA. It is found in vascular Aβ deposits  and it is physiologically drained along the perivascular space co-occurring with Aβ [133, 139]. ApoE binds Aβ . Alterations of the perivascular drainage of Aβ and apoE likely lead to CAA [133, 146, 147]. CAA, especially capillary CAA, is associated with the apoE ε4 allele [37, 100, 103, 126, 130, 132]. This finding points to an important role of apoE for the development of CAA because apoE4 is less effective in the receptor-mediated clearance of Aβ when compared to apoE3 . This property of apoE4 presumably results in capillary Aβ deposition in apoE ε4 carriers as soon as alterations in the perivascular drainage occur. In addition, the apoE ε4-genotype promotes Aβ aggregation in vascular smooth muscle cell cultures . Finally, CAA-related hemorrhage is reported to be associated with the apoE ε2 and ε4 allele [37, 88, 95, 97].
In familial cases, other amyloidogenic proteins can also aggregate in the cerebral blood vessels and cause other forms of CAA, i.e., ABri, ADan, transthyrretin, gelsolin, cystatin D, and prion protein . Unlike Aβ-related CAA, these forms of CAA are usually not restricted to the cerebral and leptomeningeal vessels .
Relationship between SVD subforms, AS, and CAA
Correlations between extent of cerebral atherosclerosis (AS), small vessel disease (SVD), cerebral amyloid angiopathy (CAA), and Alzheimer’s disease (AD)
SVD and AS exhibit similar morphological alterations. For example, atherosclerotic lesions in small arteries are defined either as AS or as SVD (subform: small vessel arteriosclerosis/atherosclerosis). Partly, this overlap is explained by the similarities of the pathogenetic mechanism of AS in large- and medium-sized vessels as well as in small vessels , such as plasma protein leakage into the vessel wall. On the other hand, there are morphological differences that justify the distinction between AS and the SVD subform small vessel arteriosclerosis/atherosclerosis: AS plaques in large- and medium-sized vessels often contain cholesterol clefts and calcifications, which are usually not seen in small vessel arteriosclerosis/atherosclerosis.
A pathogenetic link between AS and SVD on the one hand and CAA on the other is less obvious. However, apoE occurs in the SVD- and AS-related vessel wall lesions and in association with CAA-related Aβ deposits. Moreover, the apoE ε4 allele is a risk factor for CAA [100, 115, 126], SVD , and is discussed as a risk factor for AS [64, 75, 116]. The absence of apoE in apoE-knockout mice results in AS and hypercholesterolemia . These points indicate that apoE is a common player in the pathogenesis of all three vessel disorder and may represent a pathogenetic link.
The role of apoE in vessel pathology of the aging brain
Apolipoprotein E appears to be a common player in the pathogenesis of AS, SVD, and CAA. It is a plasma protein produced by hepatocytes and macrophages  and it functions as a transporter for cholesterol and other lipids . In the brain, apoE is produced by astrocytes [10, 77]. Under physiological conditions, an intact BBB prevents the influx of plasma apoE into the brain [22, 27]. Brain-apoE is drained along perivascular channels and vascular basement membranes . It can bind cholesterol, lipids, and proteins such as Aβ [77, 123], which are cleared together with apoE [6, 106], suggesting that apoE may have a transporter function not only for cholesterol and lipids but also for proteins such as Aβ.
In AS- and SVD-affected vessels, plasma proteins including apoE leak into the vessel wall and accumulate in the respective lesion [1, 18, 91, 92, 139, 150] (Figs. 1, 2, 4). Plasma proteins also leak into the perivascular space of vessels affected by SVD indicating an alteration of the pre-capillary segment of the BBB [127, 139]. Thus, SVD presumably impairs perivascular clearance, including that of Aβ and apoE, by several mechanisms: (1) SVD-induced fibrosis of the vessel wall may decrease the capacity for drainage along the basement membranes , (2) the increasing stiffness of the SVD-affected arteries leads to a less effective outward transport of the perivascular fluid due to a cessation of the pulsations [146, 147], and (3) the SVD-induced leakage of plasma proteins including apoE into the vessel wall and the perivascular space competes for perivascular drainage with the extracellular fluid of the brain [127, 136, 137, 139] (Fig. 4). Therefore, SVD presumably contributes to the accumulation of Aβ in the brain leading to CAA and Alzheimer’s disease-related Aβ plaque deposition. This hypothesis is supported by the association between SVD, CAA, and Aβ plaque deposition [13, 129] (Table 2) and by the increase of plasma Aβ1–40 in patients with SVD . Since SVD and CAA are both localized in intracerebral small arteries and arterioles, which have a perivascular space, and since both disorders develop apoE-containing lesions [93, 139] and are associated with the apoE ε4 allele as a genetic risk factor, it is tempting to assume that apoE represents a pathogenetic link between SVD and CAA. The hypothesis that apoE contributes to vessel wall destruction is further supported by the association of the apoE ε2 and ε4 alleles with CAA-related hemorrhages [37, 88, 95, 97].
One argument against the hypothesis that SVD-related BBB alterations contribute to the development of CAA and parenchymal Aβ deposition is the lacking increase of apoE in the brain . However, in the event that Aβ accumulates in the brain, e.g. in AD cases, apoE clearance switches from perivascular drainage to enzymatic cleavage as indicated by the predominance of C-terminal truncated apoE over full-length apoE . Hence, the absence of an apoE increase in the brain of AD patients or of patients with perivascular drainage deficits does not argue against the hypothesis that altered perivascular drainage of extracellular fluid leads to a congestion of extracellular fluid and proteins in the brain. Moreover, the increase of 27-hydroxycholesterol levels in AD brains indicates an influx of plasma steroids through the BBB . Another argument against a pathogenetic link between SVD and Aβ accumulation in the brain is that AD-related Aβ deposition in aged individuals first occurs in the cortex [11, 104, 129, 134] whereas SVD affects mainly white matter, leptomeningeal and basal ganglia vessels . However, the cortex is a distal area in relation to the areas with SVD-affected vessels, and Aβ is mainly produced by neurons located in this distal area [40, 94]. Thus, a proximal clearance block for peripheral extracellular fluid drainage containing neuron-derived Aβ will lead to a protein accumulation, e.g. of Aβ, in distal areas such as the cortex. This is exactly where parenchymal and vascular Aβ deposits occur first [11, 104, 129, 134]. Therefore, it is tempting to speculate that the distribution of Aβ deposits in aging and AD brains may represent the result of insufficient clearance of Aβ, rather than an anterograde neuronal expansion as previously discussed . A strong argument in favor of this hypothesis is the fact that normal brain transplants into APP-transgenic mice, which produce a high number of Aβ plaques, also develop Aβ plaques in the absence of neuronal connections .
Thus, SVD-related alterations of the BBB may contribute to the development of CAA and possibly of parenchymal Aβ deposition. In general, the results reviewed here for the fate of Aβ and apoE after SVD-related BBB alteration point to two different effects of protein retention in the brain (1) accumulation, as documented for Aβ, and (2) degradation as presumed for apoE. Therefore, it is likely to posit that accumulation or degradation of other proteins also takes place and may be related with other SVD-associated brain lesions.
Vascular brain lesions and their relationship to AS, SVD, and CAA
Brain infarction, hemorrhage, and WMLs are the major brain lesions related to AS, SVD, and CAA. The location and size of the lesions vary according to the underlying vessel disorders and the potential risk factors. Here, we will briefly describe the vascular brain lesions and their association with AS, SVD, and CAA.
Brain infarction is a circumscribed brain tissue necrosis resulting from insufficient blood supply [73, 102]. Infarcts are subdivided by nature into ischemic (anemic) or hemorrhagic infarcts. Alternatively, the descriptive classification distinguishes large infarcts (those involving the supply territories of major cerebral arteries and veins), lacunar infarcts (i.e. small vessel infarcts), and microinfarcts.
First, we will describe the relationship between ischemic and hemorrhagic infarcts to AS, SVD, and CAA regardless of the size of the infarct. Ischemic infarcts occur when focal brain perfusion falls below critical levels and no relevant collateral or remaining blood flow is available. This type of infarct is seen after thrombosis of large- to medium-sized arteries with atherosclerotic plaque rupture with or without subsequent embolism or after occlusion of SVD- or CAA-affected vessels [14, 73, 141]. Cardiac embolism also causes ischemic brain infarction [29, 33, 73]. Hemorrhagic infarcts are characterized by blood influx into the infarct territory. Different mechanisms can be responsible for hemorrhagic infarcts: (1) embolism due to AS or cardiac thrombosis [29, 81] with insufficient blood influx into the infarcted tissue (the mechanism is not clear, partial lysis of the embolus with resting blood flow, which is insufficient to save the tissue, is considered), (2) reperfusion of an anemic infarct (e.g. following lysis therapy or resuscitation), (3) collateral blood influx (insufficient to save the tissue but enough to bleed), (4) infarction in areas in which the remaining vessels are fragile due to a vessel disease such as SVD or CAA , and (5) venous obstruction with subsequent congestion and extravasation of blood into the infarct area. Thus, AS, SVD, and CAA are related to infarcts, regardless of their nature. The underlying vessel disorder is, therefore, not predictive of the ischemic (anemic) or hemorrhagic nature of an infarct.
On the other hand, the underlying vessel disorders have influence on the infarct size. Large infarcts are bigger than 15–20 mm3. They are frequently ischemic (70–80%) and due to thrombotic or embolic artery occlusion . AS plaque rupture with subsequent thrombosis and/or embolism usually occurs in the extracranial parts of the vertebral and the internal and common carotid arteries [73, 81]. Large infarcts due cardiac embolism are less frequent [73, 81]. About 10% of all large infarcts are watershed infarcts. It is a distinct type of infarct located between two vessel territories that results from insufficient blood supply during hypotensive episodes [52, 57, 157]. SVD and CAA usually do not lead to large infarcts . Therefore, large infarcts are mainly linked to AS and its complications. Lacunar infarcts are cavitating infarcts, measuring up to 5–15 mm3 in volume or 5–10 mm in diameter [30, 44, 73, 102]. They are largely confined to the cerebral white matter and subcortical structures, most commonly found in the putamen, caudate nucleus, thalamus, pons, internal capsule and the cerebral white matter. Lacunar infarcts are associated with SVD [17, 74] but not with CAA. Pathogenetically, hypertension, diabetes, previous brain infarcts, an increase of high-density lipoproteins and triglycerides in the blood  were considered risk factors, but a recent review reported that this risk is not greater than that for large infarcts . However, SVD-affected vessels showing lipohyalinosis or small vessel atherosclerosis are often found in the center of lacunar infarcts [30, 31, 111] presumably indicating that SVD is a major cause of lacunar infarcts. The risk factors described for lacunar infarcts are similar with those of AS and SVD. AS-related or cardiac embolism can also cause lacunar infarcts . As opposed to large and lacunar infarcts, microinfarcts are usually not visible at gross examination. They are smaller than 5 mm in diameter. Multiple microinfarcts in the cortex and the white matter rather than a single one likely impact cognition . Microinfarcts result either from AS-related embolism, SVD, or CAA [14, 33, 53, 61, 99]. Due to the location of the underlying vessel disorder, cortical microinfarcts are often associated with CAA whereas subcortical infarcts are mainly linked to SVD [7, 14, 53, 99, 118, 129, 132, 143].
In summary, AS is capable of causing infarcts of all size, whereas SVD is only related to lacunar infarcts and microinfarcts and CAA is mainly restricted to microinfarcts. These relationships can be easily explained by the vessel types affected by the respective vessel disorders and do not seem to reflect a more specific pathomechanism.
Cerebral hemorrhages are blood extravasations into the brain parenchyma larger than 10 mm in diameter and are differentiated from microbleeds. Together with other disorders, such as saccular aneurysms of the circle of Willis, vascular malformations, or coagulation disorders, AS, SVD, and CAA can lead to hemorrhage. In particular, AS results in hemorrhage after rupture of an atherosclerotic aneurysm , although this is a rare cause of intracerebral hemorrhage. More often, SVD leads to intracerebral hemorrhage, especially when it is associated with arterial hypertension. These hemorrhages preferentially occur after rupture of the lenticulostriate artery [125, 153]. Lobar hemorrhages with destruction of cortical tissue are frequently associated with severe CAA [50, 79, 141, 143]. Microaneurysms can be formed in SVD- and CAA-affected vessels (Fig. 3c) and are prone to rupture [87, 141, 142]. Microbleeds are blood extravasations into the perivascular and/or Virchow-Robin space without further tissue displacement and/or small intracerebral hemorrhages. They usually measure less than 10 mm in diameter  and are found in ~6% of the population with an increasing tendency during aging, especially in the presence of CAA [38, 63, 138]. Hemosiderin-laden macrophages within the perivascular space indicate prior microbleeds . Hypertension-related microbleeds are preferentially located in the basal ganglia, thalamus, and the pons in association with SVD-affected vessels . Thus, SVD and CAA are the major causes of cerebral hemorrhage and microbleeds in the aged human brain, whereas AS complications rarely lead to cerebral bleedings.
White matter lesions
White matter lesions, synonymous with leukoaraioses when found in imaging examinations, are present in up to 65% of the subjects over 65 years of age. They are more prevalent in patients with cerebrovascular disease or with cardiovascular risk [58, 96]. Clinically, WMLs are often asymptomatic [12, 25], but they can also impair cognition [26, 105]. Leukoencephalopathy, Binswanger’s disease (synonymous with subcortical arteriosclerotic encephalopathy), periventricular arteriosclerotic leukoencephalopathy, and leukomalacia are clinical correlatives of WMLs [26, 48, 138]. Risk factors for the development of WMLs are advanced age, female sex , increased plasma homocysteine , increased ICAM levels , and arterial hypertension . Some of these risk factors such as hypertension and increased plasma homocysteine levels also act as risk factors for AS and SVD. WMLs usually display, in variable degrees, white matter rarefaction (i.e., demyelination and axon loss), mild reactive astrocytosis, edema, and macrophage reaction [28, 39, 48, 124]. These changes are frequently observed in the frontal, parietal, temporal, and occipital deep white matter, especially in the centrum semiovale, and as a rule, spare the subcortical U-fibers, which are located close to the border between the cortex and the white matter [59, 98, 101, 113]. It is not clear whether periventricular WMLs have the same nature as those in the deep white matter . White matter infarcts, hemorrhages and enlarged perivascular spaces do not fall into the category of WMLs . SVD-affected vessels are often seen together with perivascular WMLs . AS and CAA, on the other hand, are not directly associated with WMLs because these vessel diseases are usually not seen in white matter vessels. Pathogenetically, SVD-related chronic hypoperfusion of the white matter and BBB alterations are presumed to lead to the degeneration of axons and myelin sheets, i.e. to WMLs [59, 98, 101, 136, 137, 140].
Pathogenetic relations between vessel disorders and brain lesions
According to the issues discussed above, SVD can lead to infarction, hemorrhage, and WMLs. In turn, infarction and hemorrhage can be caused by AS, SVD, and CAA. Thus, the type of brain lesion cannot be related to a specific vessel disorder and vice versa. All vessel disorders result in narrowing of the lumen of the affected blood vessels and can lead to vessel occlusion with subsequent infarction. Likewise, AS, SVD, and CAA damage the vessel wall. Rupture with subsequent hemorrhage can take place.
In addition to these non-specific mechanisms, WMLs and lacunar infarcts are mainly related to SVD. Despite vessel occlusion leading to lacunar infarcts, chronic hypoxia and SVD-related congestion of extracellular fluid may contribute to the development of WMLs and lacunar infarcts. This hypothesis is supported, first, by studies showing that chronic hypoxia presumably causes WMLs and lacunar infarcts given that oligodendrocytes are the most vulnerable cell type under these conditions and that their degeneration contributes to the development of WMLs and lacunar infarcts [48, 82, 83, 101, 122, 152]. Thereby, chronic hypoxia may result from SVD-related concentric fibrosis of the arterial walls and consecutive hypoperfusion [30, 31, 111, 112, 140]. Second, SVD leads to a leakage of plasma proteins, such as IgG, A2M and apoE, into the enlarged perivascular spaces . Recently, BBB alterations in patients with leukoaraiosis (i.e. WMLs) and lacunar stroke have been described [136, 145]. Accordingly, these observations point to a SVD-related alteration of the pre-capillary BBB segment involved in the pathogenesis of WMLs and lacunar infarcts and, thereby, extend the present knowledge of SVD-related BBB alterations and its association with vascular lesions in addition to AD-related changes [127, 139]. Moreover, it is well-known that plasma proteins leak into the brain parenchyma of hypertensive or hypotensive animals [91, 137] and patients with vascular dementia . Endothelial changes have been found in animal models for hypo- as well as hypertension [91, 92, 137] indicating that endothelial cell alterations due to high arterial pressure or due to chronic hypoxia in hypotensive states are capable of promoting alterations of the pre-capillary BBB segment. In the light of all these considerations, it is tempting to speculate that chronic plasma protein leakage into the brain and retention of extracellular fluid due to altered perivascular clearance for the above-mentioned reasons contributes to the development of WMLs and/or lacunar infarcts (Fig. 4b) in addition to chronic hypoxia. Such a mechanism causing alterations of tissue is well-known in liver and lung congestion. Here, chronic hypoxia and a decreased venous drainage of extracellular fluid and blood lead to a perivenous parenchymal necrosis with subsequent fibrosis .
In short, hemorrhage and infarction result from non-specific vessel wall changes seen in AS, SVD, and CAA whereas WMLs and lacunar infarcts appear to be specifically linked to SVD-related BBB leakage and chronic hypoxia.
Despite the etiology of the vessel disorders, infarction and hemorrhage appear to result non-specifically from vessel wall changes of the respective vessel disorders. WMLs and lacunar infarcts can be addressed quite specifically to SVD . The alteration of the pre-capillary segment of the BBB, thereby, allows leakage of plasma proteins into the vessel wall and into the perivascular space. These plasma proteins may compete with the extracellular fluid of the brain for perivascular drainage leading to retention of extracellular fluid, including Aβ in the brain. Such a retention of Aβ may also contribute to the development of CAA and Alzheimer’s disease. ApoE is physiologically involved in the perivascular clearance of the extracellular fluid from brain. It is found in AS, SVD, and CAA lesions, and may represent a link between these disorders. Taken together, SVD-related alterations of the pre-capillary segment of the BBB seem to play an important role in vascular pathology, i.e. WMLs, lacunar infarcts, and CAA.
The results reviewed here were in part results from studies supported by DFG-Grant No. TH624/4-1, FAPESP No. 06/53818-1 and CAPES/DAAD PROBRAL No. 289/08. The authors thank Prof. Helmut Heinsen (University of Würzburg) and Dr. Kelly Del Tredici (University of Ulm) for critical reading of the manuscript. LTG was supported by a scholarship of the Alexander von Humboldt Foundation.
This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.