White matter alterations in Alzheimer’s disease without concomitant pathologies

Aims Most individuals with AD neuropathological changes have co‐morbidities which have an impact on the integrity of the WM. This study analyses oligodendrocyte and myelin markers in the frontal WM in a series of AD cases without clinical or pathological co‐morbidities. Methods From a consecutive autopsy series, 206 cases had neuropathological changes of AD; among them, only 33 were AD without co‐morbidities. WM alterations were first evaluated in coronal sections of the frontal lobe in every case. Then, RT‐qPCR and immunohistochemistry were carried out in the frontal WM of AD cases without co‐morbidities to analyse the expression of selected oligodendrocyte and myelin markers. Results WM demyelination was more marked in AD with co‐morbidities when compared with AD cases without co‐morbidities. Regarding the later, mRNA expression levels of MBP, PLP1, CNP, MAG, MAL, MOG and MOBP were preserved at stages I–II/0–A when compared with middle‐aged (MA) individuals, but significantly decreased at stages III–IV/0–C. This was accompanied by reduced expression of NG2 and PDGFRA mRNA, reduced numbers of NG2‐, Olig2‐ and HDAC2‐immunoreactive cells and reduced glucose transporter immunoreactivity. Partial recovery of some of these markers occurred at stages V–VI/B–C. Conclusions The present observations demonstrate that co‐morbidities have an impact on WM integrity in the elderly and in AD, and that early alterations in oligodendrocytes and transcription of genes linked to myelin proteins in WM occur in AD cases without co‐morbidities. These are followed by partial recovery attempts at advanced stages. These observations suggest that oligodendrocytopathy is part of AD.

However, most studies of WM in the human ageing brain have not considered that in the elderly, most individuals have AD pathology that cannot be visualized with current neuroimaging methods. Therefore, we cannot be certain that a reasonable percentage of aged individuals subjected to neuroimaging studies suffer concomitant AD pathology. Moreover, the presence of a certain number of b-amyloid plaques and neurofibrillary tangles (NFTs) in the elderly have been considered until recently to be normal brain ageing; these cases might have been classified at post mortem as normal aged individuals.
This study was designed to analyse molecular alterations in the WM linked to oligodendrocytes and myelin in cases with AD pathological changes without clinical or pathological co-morbidities. These cases were first selected from a consecutive series of autopsies in a general hospital in which AD was one of the post mortem neuropathological diagnoses. Cases with clinical and pathological co-morbidities were not included in the second part of the study.
In the second part of the study, gene transcription was assessed with targeted RT-qPCR, and protein expression of altered genes by immunohistochemistry in the WM of the centrum semi-ovale of the frontal lobe in cases with AD pathology without clinical and pathological co-morbidities at different Braak stages of disease progression. Selected genes for analysis included markers of the oligodendrocyte lineage, genes encoding structural proteins of myelin and genes involved in energy metabolism and axon maintenance. Adult NG2-glia have the capacity to produce myelinating oligodendrocytes [57,58] thus contributing to oligodendrocyte and myelin turnover and regeneration in the adult CNS [59]. Stages of oligodendrocyte lineage are identified by the expression of platelet-derived growth factor receptor a polypeptide (PDGF-Ra), SRY-Box-10 (Sox10), NK2 homeobox 2 (Nkx2.2), oligodendrocyte transcription factor 1 (Olig1) and oligodendrocyte lineage transcription factor 2 (Olig2), among others [60]. After oligodendrocyte differentiation, myelination is triggered by myelin regulatory factor (MYRF) which is expressed in postmitotic oligodendrocytes [61]. Myelination is linked to increased expression of myelin basic protein (MBP), myelin-associated glycoprotein (MAG), proteolipid protein 1 (PLP1), myelin oligodendrocyte glycoprotein (MOG) and 2 0 , 3 0 -cyclic nucleotide 3 0 phosphodiesterase (CNP), among others [62][63][64][65][66][67][68]. Finally, blood-derived glucose is taken up by oligodendrocytes through glucose transporter 1 (GLUT1) encoded by SLC2A1. Glucose is metabolized via glycolysis to produce pyruvate and lactate which are delivered to the axons through specific solute carriers, the monocarboxylate transporters (MCT) located in cell membranes [69,70]. MCT1 is mainly expressed in oligodendrocytes [71,72]. Thus, both GLUT1 and MCT1 produced in oligodendrocytes are involved in axon maintenance independently of the complementary myelin/axon trophic alliance. These markers were analysed in this study.
Finally, histone deacetylases (HDACs) remove specific acetyl groups on a histone enabling it to interact with DNA thereby modulating gene transcription. HDAC1 and HDAC2 are expressed in oligodendrocytes [73,74]. HDAC1 immunoreactivity was also assessed in the nuclei of glial cells in the WM.
Our aim was to discern which alterations in the WM in the general population affected by AD pathological changes were linked to AD and not to concomitant clinical and pathological co-morbidities.

Selection of samples
First series Cases for study were obtained at the Bellvitge University Hospital following the guidelines of the Spanish legislation (Real Decreto 1716/2011) and the approval of the local ethics committee. Clinical parameters were retrieved after the revision of the complete clinical history in every case. Only cases with comprehensive clinical information were separated for further analysis. The first series of cases was chosen from consecutive autopsies carried out from 2009 to 2015. The current protocol for the autopsies in adult donors was as follows: one hemisphere was immediately cut in coronal sections, 1 cm thick and selected areas of the encephalon were rapidly dissected, frozen on metal plates over dry-ice, placed in individual air-tight plastic bags and stored at À80°C until use for biochemical studies. The other hemisphere was fixed by immersion in 4% buffered formalin for 3 weeks for morphological studies. For the current neuropathological study, 4-µm-thick sections from 20 representative brain regions were stained with haematoxylin and eosin, periodic acid-Schiff (PAS) and Kl€ uver-Barrera, or processed for immunohistochemistry for microglia Iba1, glial fibrillary acidic protein (GFAP), b-amyloid, phospho-tau (clone AT8), a-synuclein, TDP-43 and ubiquitin, using EnVision + System peroxidase (Dako), and diaminobenzidine and H 2 O 2 . In addition, 1-cm-thick coronal sections of the frontal lobe at the level of the head of the caudate and putamen were obtained in every case. Blocks were embedded in paraffin, cut at a thickness of 7 µm, de-waxed and stained with haematoxylin and eosin, and with Kl€ uver-Barrera. Details of the 20 selected regions and the methodological protocols for current neuropathological studies are described elsewhere [75]. From among autopsies following this procedure, including the availability of appropriate clinical information and suitability of white matter sections stained with Kl€ uver-Barrera for densitometric studies, 470 of 713 autopsies were initially selected. Three groups were defined. First, all cases with NFT pathology and/or with cerebral b-amyloid deposition, covering NFT pathology stages I-VI of Braak and b-amyloid deposition in the form of diffuse and/or neuritic plaques Braak stages 0 (no deposits) to C [76], were chosen for further classification. These were 128 men and 78 women (total n = 206); mean age (AESEM): 73.3 AE 9.6 years. AD cases were categorized as ADI-II/0-A (n = 113, men: 83, women: 30; age: 69.1 AE 8.9 years); ADIII-IV/0-C (n = 70, men: 36, women: 34; age: 78.4 AE 7.6 years) and ADV-VI/B-C (n = 23, men: 9, women: 14; age: 78.5 AE 8.5 years). Most of these cases encompassed a diversity of co-morbidities including metabolic disorders such as HTA, type II diabetes, hyperlipaemia, renal or liver failure and chronic respiratory failure; peripheral and central vascular pathology; age-related neurodegenerative diseases, such as other tauopathies, Lewy body pathology and TDP-43 proteinopathy, among others; hippocampal sclerosis and long agonic state, or vegetative state. Regarding cerebrovascular pathology, cases with cerebral infarcts, including micro-infarcts, lacunes, infarcts in the watershed areas, vascular leucoencephalopathy, hippocampal sclerosis and status cribosus; and cases with severe atherosclerosis, moderate or severe artheriolosclerosis, hypertensive angiopathy, inflammatory vascular diseases and vascular malformations were considered as AD with vascular co-morbidity. This group was classified as AD with concomitant pathology, AD-Co, and was made up of 173 cases, categorized as ADI-II/0-A (n = 104), ADIII-IV/0-C (n = 63) and ADV-VI/B-C (n = 7).
The third group (n = 20) was made up of young cases (middle-aged: MA) chosen at random among the remaining 264 cases (470 minus 206). This group consisted of 10 men and 10 women; age: 48.1 AE 7.7 years. MA did not have the clinical risk factors and co-morbidities mentioned in previous paragraphs; they did not have neurological or mental diseases, and the neuropathological study did not show abnormalities. The control group must be not interpreted as an age-matched control group, but as a control group of normal WM in MA individuals.
Second series Only cases with AD pathology without co-morbidities were the focus of this part of the study. These cases were the same as those of the second group of the first series. In addition, 10 MA cases (6 men, 4 women; age: 51.7 AE 4.7 years) from the third group of the first series were included and processed in parallel. The total number of MA and AD cases in this series is detailed in Table 1. All these cases were used for specific immunohistochemical studies. However, frozen samples were available for 24 AD cases and 10 controls. In these cases, the brain pH at the time of the autopsy was between 6.2 and 6.4, and the RNA integrity number (RIN) higher than 6 (excepting one case). AD cases at stages I-II/0-A had no neurological symptoms; AD cases at stages III-IV/0-C had no neurological symptoms nor were they affected by mild cognitive impairment; AD cases at stages V-VI/B-C had severe cognitive impairment or dementia.
Quantitative densitometric studies of the WM stained with Kl€ uver-Barrera Photomicrographs of the centrum semi-ovale of Kl€ uver-Barrera-stained sections, at the level of the head of the caudate and putamen as described above, were obtained at a mid-distance between the inferior frontal sulcus and the cingulate sulcus, approximately at 0.7-1 cm from the external vertex of the lateral ventricle. Figures were obtained at a magnification of 9 200, covering an area of 0.126 mm 2 , using a DP25 camera adapted to an Olympus BX50 light microscope . The pictures, two areas per case in every case, were analysed using Photoshop software. The density of myelin was calculated as the intensity of blue normalized for the total area and expressed as arbitrary units per area. The normality of distribution was analysed with the Kolmogorov-Smirnov test. Results were analysed with one-way ANOVA and post hoc Tukey. Differences between MA and the two groups of AD, and between the different stages in pure AD cases, were considered statistically significant at *P < 0.05, **P < 0.01, ***P < 0.001 when comparing MA with AD and AD-Co, and ## P < 0.01 when comparing AD-Co with 'pure' AD.

RNA purification and RT-qPCR
The WM was dissected from the grey matter in 34 cases, corresponding to 9 ADI-II/0-A, 8 ADIII-IV/0-C, 7 ADV-VI/B-C and 10 MA. WM samples at different AD stages and MA cases were processed in parallel. RNA from frozen WM was extracted following the instructions of the supplier (RNeasy Mini Kit, Qiagen â GmbH, Hilden, Germany). RNA integrity and 28S/18S TaqMan RT-qPCR assays were performed in duplicate for each gene on cDNA samples in 384-well optical plates using an ABI Prism 7900 Sequence Detection system (Applied Biosystems, Life Technologies, Waltham, MA, USA). For each 10 ll TaqMan reaction, 4.5 ll cDNA was mixed with 0.5 ll 20 9 TaqMan Gene Expression Assays and 5 ll of 2 9 Taq-Man Universal PCR Master Mix (Applied Biosystems). TaqMan probes used in the study are detailed in Table 2. Values of GUS-b were used as internal controls for normalization [77]. The parameters of the reactions were 50°C for 2 min, 95°C for 10 min and 40 cycles of 95°C for 15 s and 60°C for 1 min. Finally, capture of all TaqMan PCR data was made using the Sequence Detection Software (SDS version 2.2.2, Applied Biosystems). For the data analysis, threshold cycle (CT) values for each sample were processed to obtain the double delta CT (DDCT) values. First, delta CT (DCT) values were calculated as the normalized CT values for each target gene in relation to the CT of endogenous controls GUS-b. Then, DDCT values were obtained from the DCT of each sample minus the mean DCT of the population of control samples.
The normality of distribution of fold-change values was analysed with the Kolmogorov-Smirnov test. Pearson's correlation coefficient was used to assess a possible linear association between two continuous quantitative variables. To determine the relationship between gene expression and RIN values according to pathologic variables, we used the analysis of covariance (ANCOVA) in the 16 probes. Statistical analysis of the expression data between groups was made using oneway analysis of variance (ANOVA) followed by Tukey posttest, or Kruskal-Wallis test followed by Dunn's post hoc test when required using the SPSS software (IBM Corp. Released 2013. IBM SPSS Statistics for Windows, Version 21.0; Armonk, NY: IBM Corp). Outliers were detected using the GraphPad software QuickCalcs (P < 0.05). All data were expressed as mean values AE SEM. Differences between MA and AD cases were considered statistically significant at *P < 0.05, **P < 0.01, ***P < 0.001 vs. MA; # P < 0.05, ## P < 0.01, ### P < 0.001 vs. ADI-II/0-A and $ P < 0.05 vs. ADIII-IV/0-C and ADV-VI/B-C.

Immunohistochemistry
Formalin-fixed, paraffin-embedded, de-waxed sections 4-µm thick of the frontal white matter of cases listed in Table 1 were processed for specific immunohistochemistry. The sections were boiled in citrate buffer (20 min) to retrieve protein antigenicity. Endogenous peroxidases were blocked by incubation in 10% methanol, 1% H 2 O 2 solution (15 min) followed by 3% normal horse serum solution. Then the sections were incubated at 4°C overnight with one of the primary rabbit polyclonal antibodies: NG2 (used at a dilution of 1:200; Sigma-Aldrich, Merck, Darmstadt, GE), Olig2 (used at a dilution of 1:500, Sigma-Aldrich, Merck); HDAC2 (diluted 1:100, Abcam, Cambridge, UK); GLUT 1 (diluted 1:100, Abcam); or with one of the primary mouse monoclonal antibodies: PLP1 (used at a dilution of 1:100, Lifespan Biosci, Seattle, WA, USA), CNPase (used at a dilution of 1:100, Sigma-Aldrich, Merck), MBP (diluted 1:1000, Abcam) and GFAP (diluted 1:1000, Diagnostic BioSyst, CA, USA). Following incubation with the primary antibody, the sections were incubated with EnVision + system peroxidase (Dako, Agilent Technologies, Santa Clara, CA, USA) for 30 min at room temperature. The peroxidase reaction was visualized with diaminobenzidine and H 2 O 2 . Control of the immunostaining included omission of the primary antibody; no signal was obtained following incubation with only the secondary antibody. Sections were slightly counterstained with haematoxylin. Quantification of NG2-, Olig2-and HDAC2-immunoreactive cells in the WM was done by counting the number of positive cells in areas of the WM located about 1 cm below the subcortical U fibres of the frontal lobe at the level of the head of the caudate and putamen chosen at random. Microphotographs were obtained at magnifications of 9 200 or 9 400, covering areas of 0.126 mm 2 and 0.038 mm 2 , respectively, in three nonconsecutive sections per case using a DP25 camera adapted to an Olympus BX50 light microscope. The number of positive cells was counted directly on the figures and expressed as the number of positive cells per area (0.038 mm 2 for NG2-and HDAC2-immunoreactive cells, and 0.126 mm 2 for Olig2-positive cells). The normality of distribution of fold-change values was analysed with the Kolmogorov-Smirnov test. Results were analysed with one-way ANOVA and post hoc Tukey; differences were considered statistically significant at * MA vs. AD; #: ADI-II vs. ADIII-IV or ADV-VI; $: ADIII-IV vs. ADV-VI; significance levels were set at: * , # P < 0.05, **, ## P < 0.01, and ***, ###, $$$ P < 0.001. No attempts were made to quantify the densitometry of GLUT1.
Densitometric quantification of PLP1 immunoreactivity was acquired in the same areas as those indicated for NG2-, Olig2-and HDAC2-immunoreactive cells following a similar procedure to that described for Kl€ uver-Barrera-stained sections. Photomicrographs were obtained at a magnification of 9 200, covering an area of 0.126 mm 2 , using a DP25 camera adapted to an Olympus BX50 light microscope . The pictures, two areas per case in every case, were analysed using Photoshop software. The density of PLP1 was calculated as the intensity of brown colour normalized by the total area excluding white spaces of the nuclei and expressed as arbitrary units per area. The normality of distribution was analysed with the Kolmogorov-Smirnov test. Results were analysed with one-way ANOVA and post hoc Tukey. Differences between MA and the three stages of pure AD cases were considered statistically significant at: * MA vs. AD; #: ADI-II vs. ADIII-IV or ADV-VI; $: ADIII-IV vs. ADV-VI. Significance levels were set at: * , # P < 0.05, **, ## P < 0.01 and ***, ###, $$$ P < 0.001.

White matter alterations in the general series
Representative examples of the diversity of WM changes in cases with AD pathology without and with co-morbidities (AD-Co) are shown in Figure 1. This illustrates the variability in WM alterations which may concur in cases with AD pathology.
Densitometric studies of the central myelin were carried out in the AD series and in MA (Figure 2A). The density of myelin was significantly reduced in AD without co-morbidities and in AD-Co when compared with MA (P = 0.000 and P = 0.000 respectively). Myelin density was significantly lower in AD-Co when compared with AD cases without co-morbidities (P = 0.003) ( Figure 2B).
Regarding AD cases without co-morbidities, no significant differences, but rather a tendency to reduced myelin density was seen in AD stages V-VI when compared with AD stages I-II and AD stages III-IV (Figure 2C).

RNA expression of genes related to oligodendroglia and myelin in WM of AD without co-morbidities
Results of RT-qPCR are summarized in Figure 3. The expression of several genes linked to the oligodendrocyte lineage was reduced at stages III-IV/0-C when compared with MA and with AD stages I-II/0-A including OLIG1 (P = 0.007, when compared with ADI-II/A-C) and PDGFRA (P = 0.000 when compared with MA, and P = 0.029 when compared with ADI-II/0-A). Levels of PDGFRA were decreased at stages V-VI/B-C when compared with MA (P = 0.000); PDGFRA and OLIG1 levels were also significantly decreased at stages V-VI/B-C when compared with levels at stages ADI-II/0-A (P = 0.015 and P = 0.002 respectively). Curiously, NG2 mRNA expression was reduced only at stages ADV-VI/B-C when compared with levels of MA and AD stages I-II/0-C (P = 0.006 and P = 0.027 respectively).
However, other markers of oligodendrocyte differentiation such as SOX10 and NKX-2, as well as OLIG2, were not significantly altered, although OLIG2 showed a trend to decrease at middle stages of AD.
Regarding genes involved in myelin synthesis, MYRF mRNA expression, the product of which triggers myelination, was significantly increased in ADI-II/0-A and ADV-VI/B-C when compared with MA (P = 0.002 and P = 0.033 respectively). MYRF expression was transiently reduced to normal levels at stages III-IV/0-C when compared with ADI-II/0-A and ADV-VI/B-C (P = 0.005 and P = 0.049 respectively). In contrast, the majority of genes which encode proteins of the myelin sheet showed reduced expression levels at AD stages expression levels of MCT1, the gene coding for solute carrier family 16 (monocarboxylic acid transporter, member 1), were not significantly altered, but did show a trend to reduction at advanced stages of AD (Figure 3).   Table 2. One-way analysis of variance (ANOVA) followed by Tukey posttest or Kruskal-Wallis test followed by Dunn's post hoc test when required using the SPSS software; *P < 0.05, **P < 0.01, ***P < 0.001 vs. MA; # P < 0.05, ## P < 0.01 vs. ADI-II; $ P < 0.05 vs. ADIII-IV and ADV-VI (see Methods for statistical studies).  (Figures 4 and 6). GLUT1 (glucose transporter member 1) immunoreactivity, which decorated the neuropil and the wall of the small blood vessels, decreased in AD with disease progression. However, no attempt was made to quantify the density of the immunostaining due to individual variations ( Figure 4).
In contrast, GFAP immunoreactivity showed an increase in the number and intensity in individual astrocytes in the WM in parallel sections (data not shown), in agreement with previous data from several authors, as reviewed elsewhere [78].

Discussion
This study was designed to learn about WM abnormalities in cases with AD pathology without co-morbidities   information and histological quality to carry out densitometric studies of the WM of the frontal lobe were selected. Interestingly, 89% of cases aged 65 and older had AD pathology, a slightly higher figure than the percentage already reported for this age group [79,80]. Review of the clinical history and neuropathological study revealed that 84% had concomitant cerebrovascular pathology; clinical morbidities such as HTA, type II diabetes, hyperlipaemia, renal or liver failure and chronic respiratory failure; cerebrovascular pathology; age-related neurodegenerative diseases, such as other tauopathies, Lewy body pathology and TDP-43 proteinopathy; and long agonic state, or vegetative state, which were putative causes of WM alterations. After further selection, 33 were categorized as AD cases without co-morbidities. It can be argued that not all AD cases had cognitive impairment and dementia, and therefore, they cannot be classified as AD. Moreover, six cases of AD stages I-III without co-morbidities did not have b-amyloid deposits, and they might be classified as Primary age-related tauopathy (PART) [81]. However, PART has also been considered as part of AD [82]. We have followed here this consideration because AD and PART share the same NFT pathology at early stages of the disease.
Densitometric analysis performed on Kl€ uver-Barreraprocessed sections of the central WM of the frontal lobe at the level of the head of caudate and putamen revealed significant differences between MA individuals and cases with AD. It is important to stress that the MA group was composed of individuals with an age of about 25 years younger than that of the AD groups. Therefore, MA must be not considered an age-matched control group, but rather a representation of the myelin in normal MA individuals. A significant reduction in phospholipid myelin in elderly, as revealed with Kl€ uver-Barrera staining, is consistent with the idea that myelin in the cerebral WM decreases with age, as already noted in the Introduction. Since the majority of individuals aged 65 and older had AD pathology, it is difficult to ascertain whether changes in the AD groups were linked to ageing or associated with AD pathology.
More practical is the observation that myelin decay is greater in AD cases with co-morbidities (AD-Co) than in AD cases with no co-morbidities. The age of the individuals has been assessed in every case, and the mean values AE SEM have been obtained for every stage both in the total series and in pure AD. Values are similar in both groups, and then comparisons are not biased by differences in the age, but in the presence or absence of co-morbidities. Since the majority of cases with AD in our series had co-morbidities, it may be inferred that part of the WM lesions commonly reported in the elderly and in AD during life cannot be ascribed solely to age and AD, but rather to common concomitant risk factors and pathologies affecting the ageing brain.

Biochemical alterations in WM in AD without comorbidities
Previous studies have shown alterations in the lipid composition of the WM in AD. Galactosylceramide (Gal-Cer) and sulphatides, synthesized by oligodendroglia in the CNS, are major components of myelin. Reduced Gal-Cer and sulphatide levels, increased cholesterol and increased fatty acid contents occur in cortical grey and WM in AD [83][84][85][86][87][88]. Levels of GalCer and sulphatide slightly decreased in the frontal and temporal cortex, and in WM matter at stages III-IV, and more markedly at stages V-VI in AD [89]. Curiously, the activity of ceramide synthase 2, which catalyses the synthesis of very long chain ceramides, decreased in brain temporal cortex at stages I-II and frontal cortex at stages III-IV preceding neurofibrillary tangle formation, suggesting that alterations of ceramide synthesis occur earlier than previously suspected in the spread of AD [89]. Our observations on Kl€ uver-Barrera-stained sections are in line with biochemical studies showing decreased levels of phospholipid components of myelin in AD when compared with MA individuals. Differences are not as clear among stages in AD cases without co-morbidities, although a trend to reduction is observed at advanced stages.
Regarding myelin proteins, progressive reduction in the levels of myelin basic protein (MBP), myelin proteolipid protein (PLP) and 2 0 ,3 0 -cyclic nucleotide 3 0 -phosphodiesterase (CNP) has been reported in the WM of the parietal and occipital lobes in AD correlating with Braak stages V-VI [90]. Reduced levels of CNP have also been recorded in the WM of the frontal lobes in advanced AD [89,90].
Our observations are restricted to cases with AD pathology without co-morbidities. mRNA expression levels of MBP, PLP1, CNP, MAG, MAL, MOG and MOBP were preserved in the frontal white matter at stages I-II/0-A when compared with MA, but they were transiently decreased at stages III-IV/0-C, and increased thereafter to reach nearly MA levels at stages V-VI/B-C. Regarding protein expression, densitometric studies of PLP1 in the same region revealed a significant decrease in PLP1 immunoreactivity at early stages of AD, which became more marked at middle stages, followed by a slight increase without reaching MA values at advanced stages of AD.
It is worth to stress that differences between AD stages are not related to the age or gender of the individuals, but rather to the stage of the disease. Differences between myelin lipids, as revealed with Kl€ uver-Barrera staining, and myelin proteins as shown by RT-qPCR and immunohistochemistry, in the same cases may be related to differing preciseness of the methods employed. However, the relative recovery of myelin proteins, but not of phospholipids, at advanced stages of the disease deserves further study, including analysis of a possible structural imbalance between lipid and protein components of the myelin sheet in the elderly and in AD. Studies in the aged human post mortem brain have shown decreased internodal distance, reduced axon thickness and greater vulnerability of thin myelinated fibres compared with large myelinated fibres [7]. Little is known about this aspect in AD.

b-amyloid and tau
It has been suggested that cortical atrophy with neuron loss is not the main cause of WM damage in AD, as revealed by pioneering neuropathological studies [29], and later supported by combined MRI and post mortem examination [9,91]. However, this hypothesis has been brought into question by other studies. The impact of b-amyloid on oligodendrocytes is equivocal. On the one hand, b-amyloid is toxic to oligodendrocytes as identified in in vitro models, transgenic mice and familial AD [92][93][94][95][96]. However, Ab oligomers also promote oligodendrocyte differentiation and maturation in isolated oligodendrocytes and in organotypic cerebellar slices [97]. In fact, deterioration of the WM parallels, but does not correlate with either the total amount or the regional localization of b-amyloid plaques [98]. However, soluble b-amyloid is abundant in the WM in the absence of plaques in AD [99]. Therefore, the possibility of oligodendrocyte damage by soluble b-amyloid in AD cannot be ignored.
WM damage has been correlated with tau pathology in the cerebral cortex [10,101]. Therefore, myelin breakdown in AD has been posited as being linked to axonopathy and transport deficits . In favour of WM axonopathy resulting from tau pathology is the presence of WM breakdown in transgenic mice bearing the P301L mutation in the mapt gene [13]. Yet, WM disruption does not correlate exactly with the localization and distribution of NFTs in AD [29]. The present findings further support the concept of early alteration of the WM in AD, as transcription of oligodendrocyte and myelin genes in the frontal WM is altered at stages III-IV/0-C at which point no NFTs or neuron loss are found in this region.

Oligodendrocytes and NG2-glia
Previous studies have shown reduced size of the nuclei of oligodendrocytes [14], and decreased numbers of Olig2-and NG2-glia-immunoreactive cells [15,106]. Moreover, several oligodendroglial nuclei in the WM show oxidative damage (8-OHdG immunoreactivity), whereas other oligodendrocytes exhibit increased expression of p53 as a marker of stress, and a senescent phenotype (SA-b-gal immunoreactive) [17,108].
Reduction in the expression of NG2 and PDGFRA mRNA (stages V-VI/A-C, and stages III-IV/0-C and V-VI/A-C respectively), together with reduced numbers of NG2-and Olig2-immunoreactive cells in the WM, points to progressive decline of the oligodendrocyte lineage in the frontal WM with disease progression, which is in line with the demonstration of early alterations of the oligodendrocyte lineage linked to AD pathology [15,106].
Reduction in NG2-glia is important as it probably compromises the regenerative ability of the WM to restore the number of oligodendrocytes and myelin homeostasis. Reduced numbers of oligodendrocytes and altered control of glucose and lactate metabolites necessary for trophic support for axons may produce, in turn, altered axonal function, and axonal degeneration. This scenario takes place at a limited rate in the ageing human brain [19], but its potentially damaging effect is overwhelmed in AD [110].
An intriguing feature in this scenario is the increased expression of MYRF mRNA at early and late stages of AD. Regarding the number of oligodendrocytes, reduced expression of MYRF mRNA could be expected, whereas with MYRF being a factor regulating the expression of several myelin genes, the expression of such genes would be expected to increase in AD. We have no explanation for the altered MYRF regulation and response in AD.
In all these models, WM alterations characterized by myelin loss and decreased numbers of oligodendrocytes occur at early stages before the appearance of b-amyloid plaques, and they increase for a limited period of disease progression. Oligodendroglial cell death and myelin loss occur at early stages in 3xTg-AD mice [112,113], and region-specific alterations appear before b-amyloid and tau pathology [112]. However, murine models of b-amyloid deposition show reparative responses at later stages of the disease [15,[114][115][116]. No similar increase in Olig2 and NG2-immunoreactive cells occurs in AD, as shown in the present work and in previous studies 109,110]. However, a certain normalization of the mRNA expression of several myelin-related genes here observed in AD without comorbidities has its counterpart in transgenic models. As a working hypothesis, it may be suggested that repair of oligodendrocyte lineage is activated in AD and transgenic models, but regeneration is minimized in AD when compared with transgenic murine models.
In addition to vascular alterations linked to age, which may occur in any patient, hypoperfusion linked to b-amyloid deposition, alteration of the blood vessel walls, increased RAGE, altered microglia and astrocytes with senescent forms are constant in AD. b-amyloid angiopathy is found in the vast majority of cases [124,125] and it shows an early predilection for cortical blood vessels. b-amyloid deposition is accompanied by decreased expression of efflux receptor for b-amyloid and increased influx receptor RAGE in AD [126,127]. Other alterations include atrophy, oedema and increased numbers of pinocytic vesicles in endothelial cells; thickening and focal disruption of the basal membrane; increase in heparan sulphate, proteoglycans, collagen IV and laminin in the basal membrane, with atrophy of smooth muscle fibres and augmented aquaporin expression in perivascular astrocytes [128][129][130][131][132]. These changes, in conjunction with mounting evidence of altered blood-brain barrier in AD [133][134][135][136], lend support to the old proposal of reduced perfusion of the brain as a pathogenic factor in AD [137][138][139].
Our study in AD cases without co-morbidities excludes major vascular pathology and systemic circulatory factors which could have an impact on the integrity of oligodendrocytes and myelin. However, we cannot rule out a role for primary alterations in blood vessels linked to AD since specific vascular pathology is one of the manifestations of AD.

Conclusions
WM alterations in AD have been considered the result of cerebrovascular dysfunction [138,139], axonopathy linked to retrograde neuronal tau deposition , or homeostatic responses to age-related myelin breakdown [140]. These factors are not mutually exclusive, but rather reinforce each other with age and disease progression. This study has tried to minimize co-morbidities in a general population with AD pathology dying in a general hospital. It may be argued that: i. the population dying in a general hospital is not representative of the total population whose health status is better than that represented by in-patients, and that co-morbidities are less frequent and serious in the general population when compared with individuals dying in the hospital; and ii. vascular ageing, and of course vascular changes linked to AD, cannot be ruled out in the pathogenesis of WM disorders in the present series of AD cases without co-morbidities. These are undoubtedly reasonable objections.
Yet, several studies have stressed the role of oligodendrocytes as important players in the pathogenesis of distinct neurodegenerative diseases including AD [141][142][143][144]. In this line, the present observations point to early alterations of oligodendrocytes and transcription of genes linked to myelin proteins in WM in cases with AD pathology without co-morbidities before the appearance of NFTs in these regions, and before the appearance of clinical symptoms of cognitive impairment. This suggests that oligodendrocytes are along with neurons, targets of AD, and that oligodendrocytopathy is therefore part of AD.

Author contributions
IF designed the study, revised the neuropathology of cases, selected the cases without co-morbidities, interpreted the results and wrote the manuscript; PAB carried out the RT-qPCR studies and immunohistochemistry. Both authors revised and approved the final version of the manuscript.