Common Impact of Chronic Kidney Disease and Brain Microhemorrhages on Cerebral Aβ Pathology in SHRSP

Abstract

While chronic kidney disease seems to be an independent risk factor for cognitive decline, its impact on cerebral amyloid‐β (Aβ) depositions, one hallmark of Alzheimer's Disease (AD) pathology, has not been investigated.

Utilizing 80 male nontransgenic spontaneously hypertensive stroke prone rats (SHRSP) at various ages (12 to 44 weeks), tubulointerstitial renal damage, prevalence of cerebral microhemorrhages and Aβ accumulations were quantified. Using age‐adjusted general linear models we investigated the main and interaction effects of renal damage and cerebral microhemorrhages on cerebral Aβ load. In addition, using post mortem human brain tissue of 16 stroke patients we examined the co‐localization of perivascular Aβ deposits and small vessel wall damage.

Statistical models revealed an age‐independent main effect of tubulointerstitial kidney damage on brain Aβ accumulations, which was reinforced by the consecutive presence of cerebral microhemorrhages. Moreover, cerebral microhemorrhages independently predicted brain Aβ burden in SHRSP. In up to 69% of all human cases perivascular Aβ deposits were detected in the direct vicinity of small vessel wall damage.

Our results support the associations between vascular pathology and Aβ deposition, and demonstrate a relationship between chronic kidney disease and cerebral Aβ pathology. Hence, our data suggest that prevention of chronic renal damage may reduce cerebral Aβ pathology.

INTRODUCTION

During the last two decades, the incidence of both chronic kidney disease (CKD) and dementia increased remarkably in the aging population, and both diseases have been recognized as a significant global health burden 1, 30, 54. Related to this, CKD has been discussed as an independent risk factor for cognitive decline and is associated with an increased risk of various dementia subtypes, especially in patients with vascular risk factors 16, 43, 62.

Age‐dependent spontaneous systemic arteriolar and endothelial dysfunction 68 as well as common susceptibilities of the renal and cerebral microvasculature to cardiovascular risk factors 69 have been proposed to be the main link between CKD and cognitive decline, basically explaining the association between renal disease and the vascular dementia subtype. Recently reported relationships between impaired renal function, cerebral microbleeds and white matter hyperintensities account for a more direct link between cerebral small vessel pathology and CKD 58, 63, 64, 69, 80. In addition, direct associations between CKD and neurodegeneration, which are independent from microvascular pathology have been discussed 8, as have associations between CKD and increased serum amyloid‐β (Aβ) levels 22, 39, further suggesting an impact of kidney damage on Alzheimer's disease (AD) pathology. Both old age and arterial hypertension seem to link CKD, cerebral small vessel disease and (vascular) Aβ depositions 2, 52, 69, 78. However, so far no studies investigated the interaction between CKD, small vessel wall damage and cerebral Aβ pathology.

Spontaneously hypertensive stroke‐prone rats (SHRSP) develop arterial hypertension, ischemic and hemorrhagic strokes 76, and are a valid model for the interactions between CKD, cerebral microvascular damage and Aβ accumulations 59, 60, 61, 66. While our previous work showed a relationship between CKD and cerebral small vessel disease in SHRSP 59, we were now interested to further elucidate associations between CKD, microhemorrhages and cerebral Aβ pathology in that rat model. In addition, we aimed to test the relevance of parts of the data derived from our animal model, namely the occurrence of cerebral Aβ depositions in the presence of small vessel wall damage, on human brains by using post mortem brain tissue of stroke patients suffering from arterial hypertension.

MATERIALS AND METHODS

Samples and tissue preparation

SHRSP. Animal procedures were conducted after obtaining the approval of the Animal Care Committee of Sachsen‐Anhalt, Germany (reference number of license for animal testing 42502‐2‐1148DZNE, 42502‐2‐943FAN).

Groups of 80 male SHRSP (Charles River Laboratories International, Inc., Wilmington, MA, USA) aged from 12 to 44 weeks (12 weeks (w) n = 6, 14w n = 5, 16‐18w n = 7, 20‐22w n = 4, 24‐26w n = 9, 28w n = 9, 30‐31w n = 5, 32‐34w n = 9, 36w n = 8, 40w n = 5, 44w n = 13) were investigated. Animals were bred in standard conditions including free access to food and water in the natural day light cycle, and principles of laboratory animal care were kept. Neurological status (such as decreased spontaneous activity, coordination failure, falling to one side and hunched posture) was assessed daily and body weight was monitored twice a week.

Rats were anesthetized with Pentobarbital (40 mg/kg body weight), transcardially perfused with 120 mL of phosphate buffered saline (PBS) followed by 120 mL of paraformaldehyde (PFA, 4%); brains and kidneys were removed, post‐fixed for 48 h in the same fixative, cryoprotected in 30% sucrose for 6 days and frozen in methylbutane at −80°C. Tissue sections were cut using a cryostat; coronal brain sections were taken from 11 planes (from the frontal to the occipital pole) and renal sections from respective three planes (from the cranial pole to the mid of both kidneys).

Human brains

The study included 16 subjects (mean ± standard deviation, range for age in years, 69.33 ± 8.98, 53‐85; male, 44%) presented to the departments of Neurology (n = 14), Cardiology (n = 1) and Oncology (n = 1) of the University of Medicine and Pharmacy of Craiova, Romania.

Ten patients (63%) have suffered from macroangiopathic (n = 8) or microangiopathic (n = 2) ischemic stroke localized in cortical and subcortical regions, four (25%) have presented with intracerebral hemorrhages localized in the temporal cortex (n = 1), the striatum (n = 1) and the medulla oblongata (n = 2), while two (12%) had no neurological disease and presented with chronic bronchitis and colonic cancer. Twelve patients (75%) had a history of co‐morbid arterial hypertension. Five patients (31%) died from ischemic brainstem stroke or from large hemorrhagic stroke, nine (56%) from myocardial infarction, one (6%) from cardiorespiratory failure following bilateral chronic bronchitis and one (6%) from non‐CNS metastasizing colonic cancer.

With the exception of one patient with AD (primary diagnosis chronic bronchitis, 85‐year old, Mini Mental State Examination score 21 [MMSE, 15], moderate neuritic plaque load according to Consortium to Establish a Registry for Alzheimer's Disease [CERAD] criteria 42), none of the patients had a history of dementia; moreover, none of the patients' relatives reported on impairment of daily living activities suggestive of dementia and neuropathological examinations performed by an experienced neuropathologist (DP) excluded advanced AD pathology.

Our human study was approved by the local ethics committee of the University of Craiova, Romania (reference number 02/17.02.2009) and written informed consent was obtained from each patient or her/his relatives, accepting tissue preservation for research purposes.

Histology and immunohistochemistry (IHC)

SHRSP. Thirty three brain sections (30 µm, three per plane) per animal were stained with hematoxylin and eosin (HE) and Congo red (CR). Furthermore five brain sections adjacent to those with suspected CR‐positive Aβ depositions were double stained with CR/Prussian blue staining (Perls‐iron‐staining), Thioflavine T/Prussian blue staining and Thioflavine S/Prussian blue staining to simultaneously detect cerebral Aβ accumulations and iron deposits indicative of small hemorrhages 55. CR, Thioflavine T and Thioflavine S are markers for the anti‐parallel β‐pleated conformation of Aβ fibrils composing dense Aβ 14. For Prussian blue staining sections were incubated with a 10% potassium ferrocyanide solution for 20 minutes at room temperature.

For the kidney, 12 sections (30 µm, two per plane) per animal were stained with HE.

Five brain sections per animal adjacent to those with suspected CR‐positive Aβ deposits were stained with STL‐FITC (solanum tuberosum lectin‐fluorescein isothiocyanate, Axxora, Enzo Life Sciences GmBH, Loerrach, Germany, 1:500, endothelial marker), anti‐rodent Aβ (Covance, Dedham, MA, USA, 1:500, specific for rodent Aβ), and DAPI (4′.6‐diamidino‐2‐phenylindole, MoBiTec GmbH, Germany, 1:10 000, nuclear staining). In brief, tissue was pretreated with citrate buffer (65°C, 30 minutes), blocked with 10% donkey serum and incubated with anti‐rodent Aβ overnight at 4°C. Cy3‐tagged donkey anti‐rabbit IgG (Dianova GmbH, Hamburg, Germany, 1:500) was used as secondary antibody. After dehydration with increasing concentrations of alcohol sections were mounted on slides with Histomount (Fisher Scientific GmbH, Schwerte, Germany).

Human brains

We investigated formalin‐fixed paraffin‐embedded sections from frontal and temporal cortices, hippocampus, striatum and pons, all of which showed no signs of cerebrovascular pathology at macroscopic examination.

Two or three 4‐µm‐thick histological sections per brain were initially stained with HE, and one additional section per brain was immunohistochemically stained with anti‐immunoglobulin G (IgG) to detect small vessel wall damage; briefly, sections were subjected to antigen retrieval in EDTA, incubated for 32 minutes at room temperature with a rabbit polyclonal anti‐human IgG antibody (Thermo Fisher Scientific Germany BV & Co KG, Braunschweig, Germany, 1:500) and IgG positivity was detected using an ultraView 3.3′‐diaminobenzidine (DAB) Universal Detection Kit (Ventana Medical Systems, Inc., Tucson, AZ, USA). In addition, sections adjacent to HE sections showing small vessel wall damage were stained for combined HE staining and Aβ IHC. In three cases (two with macroangiopathic cortical ischemic lesions and one with chronic bronchitis/AD) with Aβ accumulations (= Aβ positive in the combined HE staining and Aβ IHC) two additional sections were stained for combined Prussian blue staining and Aβ IHC. Briefly, after antigen retrieval in 70% formic acid for 5 minutes, sections were washed in distillate water and incubated for 30 minutes in a 0.1% hydrogen peroxide solution. Next, the sections were blocked for 1 h in 3% skimmed milk (Merck, Darmstadt, Germany), then incubated overnight with a mouse monoclonal anti‐human Aβ antibody [amino acid residues 17–24, clone 4G8, Merck‐Millipore, Bucharest, Romania, 1:30 000, detects diffuse and dense Aβ accumulations 23]. On the next day the signal was amplified for 30 minutes utilizing a species specific human‐adsorbed peroxidase polymer‐based system (Nikirei Bioscience, Tokyo, Japan), and then detected with DAB (Dako, Glostrup, Denmark). Sections were coverslipped in a styrene based mounting medium (DPX, Merck, Darmstadt, Germany) after additional HE staining. Further on, after being photographed, Aβ immunopositive sections were de‐coverslipped and counterstained with a prolonged Prussian blue protocol 12.

Two 20‐µm‐thick sections adjacent to the Aβ immunopositive sections (see above) were selected from three cases (two diagnosed with macroangiopathic cortical ischemic lesions and the third with chronic bronchitis/AD, the same cases that have been chosen for combined Prussian blue staining/Aβ IHC) with Aβ deposits, were incubated with the 4G8 primary antibody (as already described), but the signal was detected with a goat anti‐mouse Alexa Fluor 594 secondary antibody (Invitrogen, Darmstadt, Germany, 1:300, 1 h at room temperature), counterstained with DAPI and coverslipped in anti‐fading mounting medium (Invitrogen, Darmstadt, Germany).

Data analysis

SHRSP

Numbers of microhemorrhages/small perivascular bleeds and amount of parenchymal/perivascular Aβ accumulations were quantified in different brain regions (including cortical areas, basal ganglia, hippocampus, corpus callosum, thalamus), examining various stainings (HE, CR and Aβ‐IHC).

Renal damage was analyzed as follows: 12 sections per rat and 10 fields of view (FOV) per section were investigated for the occurrence of vascular pathologies, for the occurrence of peritubular capillary erythrocyte accumulations and tubular protein cylinders. Peritubular capillary erythrocyte accumulations and tubular protein cylinders were separately ranged from 0 to 3 with 0 = no peritubular capillary erythrocyte accumulations or no tubular protein cylinders per FOV and 1 ≤ 5%, 2 = 5–30%, 3 ≥ 30% per FOV were affected by peritubular capillary erythrocyte accumulations or tubular protein cylinders 4, 59. For peritubular capillary erythrocyte accumulations and tubular protein cylinders mean scores of all FOVs derived from all sections (= 10 FOVs x 12 sections = 120 FOVs) were calculated. For statistical analysis, for each rat the mean score of peritubular capillary erythrocyte accumulations and the mean score of tubular protein cylinders were summed up, resulting in one score representing combined tubulointerstitial damage. Combined tubulointerstitial damage score ranged from 0 to 6.

Human tissue

Analysis was performed in brain areas not affected by any disease‐specific/‐related neuropathology (e.g., investigation of brain regions contralateral to ischemic lesions). HE sections were assessed to detect small vessel wall damage, including small vessel wall discontinuities and erythrocyte leakages throughout the small vessel walls, and for the presence of microhemorrhages. Perivascular space enlargement was also assessed as it indicates increased fluid content that may result from subtle small vessel wall damage 73. However, since postmortal perivascular edema and other artefacts can lead to perivascular space enlargement, we exclusively focused on perivascular spaces around small vessel pathologies, for example, around small vessel thromboses. In anti‐IgG stained sections small vessel wall damage was indicated by (i) IgG positive plasma cells, and/or (ii) IgG positive neurons and/or (iii) IgG positive microglia. In all HE/Aβ stained sections the occurrence of perivascular Aβ depositions in the presence of small vessel wall damage was evaluated. In all Prussian blue/Aβ stained sections, we traced the co‐occurrence of iron depositions and cerebral Aβ accumulations. In order to identify the occurrence of extravasated erythrocytes in Aβ deposits, Aβ accumulations were followed on the red‐channel immunofluorescence in the 20‐µm‐thick sections (prepared as described above), together with the green‐spectra signal given by the erythrocyte autofluorescence. Z‐stacks were grabbed utilizing a Nikon Eclipse 90i motorized microscope equipped with a Rolera‐XR cooled CCD camera (Q‐Imaging, Surrey, BC, Canada), together with the Image ProPlus AMS 7 image analysis software (Media Cybernetics, Bethesda, MD, USA). Randomly chosen Aβ accumulations were visualized in the entire depth of the sections and images have been captured on DAPI, Alexa 488 and Alexa 594 spectra. All z‐stacks were subjected to a blind deconvolution algorithm based on a multi‐pass adaptive point spread function (PSF) subtraction of diffracted light (AutoDeblur, Image ProPlus). All final image collages were prepared for publication in Corel Draw 12 (Corel Corp., Ottawa, Canada).

Statistics

SHRSP

We aimed to assess the independent and common effects of CKD and microhemorrhages on cerebral Aβ pathology and a general linear model (ANOVA, univariate analysis of variance) was used for statistical analysis. Our statistical model included the number of cerebral Aβ accumulations (dependent variable), the number of cerebral microhemorrhages (independent variable, main effect), the degree of combined tubulointerstitial renal damage (independent variable, main effect), age as a covariate and interaction effects between (i) the number of cerebral microhemorrhages × age, (ii) the degree of combined tubulointerstitial damage × age and (iii) the number of cerebral microhemorrhages × the degree of combined tubulointerstitial damage, as further independent variables. To interpret the directionality of the interaction effects, frequency of Aβ deposits was analyzed as a function of all those interaction effects which became significant (severity of tubulointerstitial damage × age, number of microhemorrhages × degree of tubulointerstitial damage). F‐ratios (ratio of the explained to the unexplained variance of the model) are given and P‐values ≤ 0.05 were considered to be statistically significant. Statistical analysis was conducted using SPSS, version 23.0.

RESULTS

Qualitative and quantitative tissue analysis

SHRSP

In SHRSP dense Aβ deposits were seen in the brain parenchyma, surrounding capillaries and arterioles (pericapillary and perivascular Aβ accumulations) and were co‐localized with iron accumulations indicative of sites of small (pericapillary/perivascular) hemorrhages (Figure 1). Quantitative tissue analysis including CR‐ and Thioflavine stainings and Aβ‐IHC revealed the occurrence of cerebral Aβ depositions in 18 (23%) of all investigated SHRSP (Figure 2 ). The Aβ pathology predominantly affected cortical regions and the basal ganglia and was detectable first at an age of 20 weeks (Supporting Information Table S1). Cerebral small perivascular bleeds/microhemorrhages were verifiable in 26 (33%) of the animals, occurred first at 24 weeks of age and predominantly affected cortical regions, the basal ganglia and the hippocampus formation (Figures 1 and 2 ; Supporting Information Table S1). Seven (9%) SHRSP showed cerebral small vessel thromboses, adjacent confluent tissue infarcts and an abnormal neurological status (Figures 1 and 2 ; Supporting Information Table S1).

Figure 1.

Aβ CNS deposition, microhemorrhages and tubulointerstitial renal damage in SHRSP. Dense parenchymal (A–A³ & C¹), pericapillary (B, C (arrows) & C²–C⁴) Aβ deposits, small perivascular bleeds (D, E & F (arrows) & D¹), infarcts (arrowheads in G) and IgG leakage throughout the small vessel walls (K–M) are part of the age‐dependent mixed cerebral pathologies in SHRSP. Iron and Aβ depositions are colocalized in the rat brain (A¹–A³ & C¹–C⁴). In the brains of these animals, recent microhemorrhages co‐occur with small ischemic lesions (arrowheads in E & F), thromboses (* in F) and with liquefaction ischemic lesions containing haem‐laden macrophages (arrowheads in G). Peritubular capillary erythrocyte accumulations (H, arrow), tubular protein cylinders (I, arrow) and hyperplastic arteriolosclerosis (J, arrows) indicate combined tubulointerstitial and vascular/hypertensive kidney damage in SHRSP. B, K, L, M – IHC, STL ‐ solanum tuberosum lectin‐fluorescein isothiocyanate (endothelial marker), zymo – in situ zymography showing matrix metalloproteinases 2,9 activity visualized with the use of gelatin conjugated with FITC (DQ gelatin, Invitrogen), IgG – immunoglobulin G, representing the IgG leakage, visualized with the use of an anti‐rat IgG secondary antibody Cy3 conjugated (Dianova, Hamburg, Germany, 1:500), DAPI ‐ 4′.6‐diamidino‐2‐phenylindole (nuclear staining), A–A³ – Prussian blue/CR staining, C, C²–C⁴ – Prussian blue/Thioflavine S staining, C¹ ‐ Prussian blue/Thioflavine T staining, D–J – HE staining. A & A¹ – corpus callosum, A², A³, C¹, C² & D–G – cortex, B, C, C ³ & C⁴ – hippocampus, D¹ ‐ basal ganglia, H & I – kidney medulla, J – kidney cortex. A¹, B¹, C³, C⁴ – magnifications of the respective subfigures.

Figure 2.

Frequencies of perivascular Aβ accumulation, cerebral small vessel wall damage and renal pathology in the animal model lot. The overall frequencies of perivascular Aβ formations, cerebral small perivascular bleeds, small vessel thrombosis and infarcts, as well as of combined tubulointerstitial renal damage and of renal vascular pathologies are depicted for the total number of investigated animals (n = 80) (A). The relative frequencies of mild tubulointerstitial damage, severe tubulointerstitial damage and cerebral microhemorrhages are illustrated for the total number of Aβ depositing (n = 18) and respectively negative (n = 62) animals (B).

Combined tubulointerstitial renal damage was proven in 68 (85%) of all examined SHRSP and was detectable first in rats aged 14 weeks (Figures 1 and 2 ; Supporting Information Table S1). Thirty six (45%) of all animals additionally exhibited renal vascular pathologies (including thrombotic microangiopathy, small vessel occlusions and hyperplastic arteriolosclerosis [“onion‐skin” concentric vessel wall thickening]) which were detectable from an age of 24 weeks onward (Figures 1 and 2 ; Supporting Information Table S1).

Human tissue

Dense‐diffuse Aβ deposits, including parenchymal and pericapillary/perivascular Aβ accumulations were detected in 69% of the subjects and occurred mostly in cortical regions, but also in the hippocampus and the striatum (Figures 3 and 4). Small vessel wall discontinuities and erythrocyte leakages throughout the small vessel walls occurred in 100% of the cases, microhemorrhages in 81% and enlarged perivascular spaces associated to further small vessel pathologies in up to 94% of the patients, and 81% displayed IgG positivity (Figures 3 and 4). Comparable to the distribution of Aβ, small vessel wall damage was found in cortical and subcortical regions. Of note, at least three of the small vessel lesions described above were invariably seen in the same regions that showed Aβ depositions. Considering the prevalence of Aβ in the direct vicinity of small vessel wall damage, perivascular Aβ was detected in 69% of the cases around small vessels with leakages of single erythrocytes, in 19% around microhemorrhages and in up to 50% of the cases around enlarged perivascular spaces containing small vessel thrombi (Figures 3 and 4). Co‐localization of iron depositions and Aβ immunostaining was demonstrated in all of the three Aβ positive cases under investigation (Figure 3). In randomly sampled Aβ depositions of three Aβ positive subjects more than 80% of the parenchymal Aβ deposits/plaques exhibited extravasated autofluorescent erythrocytes (Figure 3).

Figure 3.

Aβ depositions around small vessel wall damage in a human stroke cohort. Dense‐diffuse perivascular/pericapillary Aβ deposits were localized around small vessels/capillaries with single perivascular/pericapillary erythrocyte infiltrations/leakages into the neuropil (arrow in A), around microhemorrhages (B), around enlarged perivascular spaces without further small vessel/capillary pathologies (C), and around enlarged perivascular spaces bearing small vessels/capillaries with hyaline thrombosis (arrows in D). More diffuse‐like Aβ formations appeared (i) around an apparently intact capillary without perivascular enlargement, that contained red blood cells (arrows in E), and (ii) around what seems to be a remnant neuron that could have produced the Aβ (asterisk in E). Chronic small hemorrhages, indicated by iron depositions, are centered on dense Aβ accumulations as demonstrated in F & G. Birefringent stacked erythrocytes (arrows in G) indicate a capillary near an area of post‐micro hemorrhagic iron deposition that also co‐localised with a small Aβ plaque. Deconvoluted consecutive optical planes captured in 20‐µm‐thick sections revealed that the co‐occurrence of green autofluorescent erythrocytes and Aβ accumulations (asterisks in I) can be identified in consecutive optical planes (arrow in I), where initial imaging showed no apparent association (arrow in H) with the plaque (asterisk in H). Immunoglobulin G (IgG) infiltration in these brains is seen in plasma cells around blood vessels (arrow in J) and in small hemorrhage foci (arrowheads in J), in some neuronal bodies, especially around microhemorrhages (arrow in K), or in perivascular petechial microhemorrhage areas (L and enlarged in its inset). Opening of the blood‐brain barrier also leads to a diffuse neuropil staining around these sites (J and K). A–E – HE staining/DAB detected Aβ IHC, F & G ‐ Prussian blue staining/DAB detected Aβ IHC, H & I, fluorescent Aβ IHC (4G8, red), Ery, erythrocyte autofluorescence (green), DAPI ‐ 4'.6‐diamidino‐2‐phenylindole, nuclear staining (blue), J–L – DAB detected IgG IHC; A–D – temporal cortex, E–L –superior frontal cortex; all insets represent magnifications of the respective figures

Figure 4.

Frequencies of cerebral small vessel wall damage and perivascular Aβ formations in the human stroke sample. The overall frequency of cerebral small vessel wall damage and its combined occurrence with perivascular Aβ formations is depicted for each of the pathological entities of interest. Percentages of affected subjects relating to the whole sample are given. The hatched fill bars represent corresponding small vessel wall damage entities in association with Aβ deposits. EPS = enlarged perivascular spaces, IgG = immunoglobulin G.

Effects of microhemorrhages and CKD on cerebral aβ depositions in SHRSP

Of those SHRSP accumulating Aβ (n = 18), 56% (n = 10) displayed cerebral microhemorrhages/small perivascular bleeds. Of those animals revealing no cerebral Aβ deposits (n = 62), however, only 26% (n = 16) showed cerebral microhemorrhages/small perivascular bleeds, resulting in significant group differences between Aβ positive and Aβ negative SHRSP (Fisher's exact test, P < 0.02). Additionally, tubulointerstitial renal damage was more pronounced in Aβ positive compared to Aβ negative SHRSP: of the SHRSP accumulating cerebral Aβ, 56% (n = 10) displayed severe combined tubulointerstitial damage (scored as 5 or 6), while only 27% (n = 17) of the Aβ negative animals showed combined tubulointerstitial damage scores of 5 or 6 (Fisher's exact test, P < 0.05). Conversely, none of the Aβ positive SHRSP, but 21% (n = 13) of the Aβ negative rats revealed no or mild combined tubulointerstitial damage (scored as 0 or 1), and group differences became significant (Fisher's exact test, P < 0.04) (Figure 2; for further details please see Supporting Information Table S1).

Age was related to higher numbers of cerebral Aβ depositions in SHRSP (F = 18.7, P < 0.001). Age‐adjusted statistical model revealed significant main effects of the amount of small perivascular bleeds (F = 16.3, P < 0.001) and the degree of combined tubulointerstitial damage (F = 47.3, P < 0.001) on the number of cerebral Aβ accumulations. There were also significant interaction effects between age and combined tubulointerstitial damage (F = 64.7, P < 0.001) and between the amount of small perivascular bleeds and the degree of combined tubulointerstitial damage (F = 29, P < 0.001) on cerebral Aβ pathology. Interaction results have to be interpreted this way, that in SHRSP older age and more severe combined tubulointerstitual damage commonly predict higher amounts of cerebral Aβ accumulations. Higher amounts of cerebral Aβ accumulations are moreover related to combined effects of increased numbers of cerebral microhemorrhages and advanced combined tubulointerstitual damage.

DISCUSSION

We detected age‐independent effects of the amounts of small cerebral perivascular bleeds on the number of Aβ depositions, and qualitative analysis of exemplary human stroke brains supported the common presence of Aβ deposits close to small vessel wall damage. In SHRSP, cerebral Aβ accumulations were additionally related to chronic renal disease and were also predicted by the complementary effects of microhemorrhages and tubulointerstitial damage.

Small vessel wall damage was highly prevalent in our human stroke cohort suggesting a co‐occurrence between small vessel wall damage and large ischemic and hemorrhagic lesions. Our findings thereby support recent imaging data that microbleeds predict intracerebral hemorrhages as well as ischemic stroke 6, 38, 65. This is further supported by our observation that all SHRSP with small vessel thrombosis and confluent large ischemic infarcts also showed microhemorrhages (Supporting Information Table S1). However, in our human cohort the prevalence of small perivascular bleeds and single erythrocyte leakages was considerably higher than reported previously in imaging studies 11. Depending on magnet strength brain imaging data may be biased towards lower microbleed frequencies, while direct pathological investigation can detect even small bleeds and single red blood cell leakages missed by MRI, which may account for the differences between imaging and autopsy data. The prevalence of enlarged perivascular spaces in our cohort prevalence was similar to findings of previous imaging studies 7, 26, 79.

Ischemic and hemorrhagic strokes are associated with increased cortical Aβ retention 77 and poststroke dementia includes up to 60% AD cases 39, 81. Our data show that up to 70% of the stroke patients exhibited diffuse Aβ deposits in brain regions that were not affected by ischemic or hemorrhagic injury. Our results are supported by similar findings indicating diffuse Aβ accumulations around small vessels after cerebral hypoperfusion induced by experimental common carotid artery stenosis 51. As recent data demonstrated that widespread blood‐brain barrier breakdown also affects the contralateral hemisphere in acute, subacute and chronic ischemia stages 17, 18, 53, bi‐hemispherical perivascular Aβ drainage failure (see below) may account for increased global cerebral Aβ which in turn may contribute to AD development in poststroke patients. On the other hand, in our cohort pre‐existing vascular Aβ may have promoted small and large intracerebral bleedings 72.

While our statistics (in SHRSP) and the co‐occurrence of small vessel and Aβ pathology (in the human stroke sample) suggest an association between small vessel wall damage and Aβ, co‐localization of Aβ with iron depositions and extravasated erythrocytes proves the accumulation of Aβ directly at sites of small vessel wall damage. Two interconnected main mechanisms may initiate (peri)vascular Aβ depositions. First, age‐dependent perivascular Aβ drainage failure along the small vessels' basement membranes promotes (peri)vascular Aβ depositions, in particular at sites of small vessel wall discontinuities and small perivascular hemorrhages 19, 34. Second, the components of the blood vessels themselves including vascular smooth muscle cells, adventitial cells, pericytes and perivascular microglia are moreover capable to produce Aβ, that has the potential to accumulate under the glia limitans if perivascular Aβ drainage fails 31, 47. Over time the (peri)vascular Aβ deposits grow and replace the degenerating small vessel wall, which mitigates vessel wall damage and results in an increase of perivascular bleeds (see above). Of note, SHRSP is one of the very few nontransgenic rodent models that deposit endogenous Aβ in the absence of human transgenes, which underlies the important interplay between vascular mechanism and Aβ deposition initiation in our animal model of small vessel disease and arterial hypertension 13, 27, 61, 76.

CKD seems to affect cerebral Aβ formation independent of microhemorrhages and therefore small vessel wall damage may not uniquely account for the association between renal pathology and Aβ. There is, however, a relative shortage of data which could help to elucidate the direct link between kidney damage and cerebral Aβ formations, which we will demonstrate shortly in the following sections. First, recent data suggest that peripheral tissues and organs play a pivotal role in the catabolism of peripherally circulating Aβ derived via distinct transport mechanisms from the brain Aβ pool 40, 74. Thereby, kidney and liver have been suggested to be the major organs for exogenous Aβ uptake out of in the blood, and enhanced peripheral Aβ clearance rate was directly related to lower cerebral Aβ levels 74. Conversely, it is very conceivable, that impaired renal clearance of peripherally circulating Aβ (as it should occur in the case of existing CKD) results in elevated cerebral Aβ retention. Second, Cystatin C elevation indicates impaired renal function 75, and may link CKD and cerebral Aβ pathology as it is an endogenous inhibitor of the Aβ_1‐42 degrading enzyme cathepsin B 9, 45, 70. Increased Cystatin C therefore promotes Aβ_1‐42 deposition, which is supported by the co‐localization of Cystatin C with intravascular and parenchymal Aβ 32, and by the fact, that Cystatin C gene ablation lowers the abundance of Aβ_1‐42 formations in a transgenic AD model 67. And last, renal damage related angiotensin converting enzyme (ACE) dysfunction may further link CKD and cerebral Aβ. ACE is an Aβ degrading enzyme by itself 24, 82, and its dysfunction may thus result in lower conversion rates of Aβ_1‐42 to Aβ_1‐40, while its overexpression seems to protect from increased Aβ_1 − 42 burden 3, 36, 44, 48. Aβ_1‐42 is essential for initial fibril genesis and Aβ nidus formation, which is in turn necessary for the later growth of dense Aβ fibrils; Aβ_1‐42 aggregates faster than Aβ_1‐40, and it leads to further cerebral Aβ deposition by co‐optation of other Aβ isoforms 29, 41; it is moreover required for the formation of toxic soluble oligomeric assemblies 10, 33. Aβ_1‐42 and its relative increase resulting from Cystatin C elevation and from ACE dysfunction should thus be capable to account in particular for cerebral Aβ pathology in the presence of CKD. Our approach used for the Aβ detection can however not clearly differentiate between Aβ_1‐40 and Aβ_1‐42, but provides evidence for the existence of Aβ fibrils with β‐sheet conformation in our rat model. But, as SHRSP reveal typical dense and perivascular (pericapillary) Aβ formations, that frequently exhibit Aβ_1‐42 2, the rats' Aβ should mainly contain Aβ_1‐42.

The finding, that SHRSP exhibit renin angiotensin system imbalances, which provoke high blood pressure in that hypertensive model, add further evidence for the association between dysfunctional ACE and Aβ pathology 35, 49. Indeed, it is very conceivable, that arterial hypertension in SHRSP (starting at an age of around 6 weeks, 76) additionally links CKD and cerebral Aβ formation. Blood pressure increase initiates and impairs chronic tubulointerstitial damage, which has been reported to have detrimental effects on cerebral Aβ retention (most likely in combination with age, genetic status and the presence of further cardiovascular risk factors) 46, 56, 57, 71. Advanced CKD itself moreover further aggravates arterial hypertension 71. In our animal model, arterial hypertension may thus additionally either moderate or mediate the impact of CKD on cerebral Aβ accumulation. As we, however, did not perform blood pressure measurements in SHRSP, no final conclusions can be drawn on this hypothesis, which is a shortcoming of the study that should be addressed in future projects.

Our experimental data suggest a temporal pathological cascade beginning with arterial hypertension [at an age of around 6 weeks, 76], proceeding to CKD (and to an early cerebral endothelial dysfunction, see below), microhemorrhages and then to Aβ pathology. Initial (hypertension‐induced) tubulointerstitial damage is supported by the high CKD frequency and the young age of renal disease onset (at 14 weeks) in SHRSP (Supporting Information Table S1), and is in line with published data indicating that renal dysfunction predicts cerebral microhemorrhages 50, 58, 59, 64, 80. Aβ deposits were depicted at 20 weeks of age, and small bleeds occurred first at later ages of 24 weeks, which may account for some independency of Aβ pathology and microhemorrhages. Aβ, indeed, has been related to endothelial dysfunction and altered cerebrovascular autoregulation, promoting cerebral small vessel wall damage itself 21. The 1.4‐fold higher prevalence of microhemorrhages [as well as the detectability of an early endothelial capillary and arteriolar dysfunction in SHRSP aged 12 weeks 60, 61], however, supports a faster progression of small vessel wall damage which conversely further increases Aβ as a result of, for example, impeding the transport of amyloid across the blood‐brain barrier or reducing its drainage along perivascular pathways (see above) 5. Proportions between higher microhemorrhage and somewhat lower Aβ frequencies were similar in our human sample. As renal and small vessel wall damage mutually influence Aβ, one might speculate that once microhemorrhages have occurred, CKD associated Aβ development should progress faster.

The strength of our study is its experimental setting allowing for the simultaneous histological investigation of renal and cerebral pathologies at various animal ages, and the concurrent availability of rat and human data. In contrast to previous human studies, we directly considered the relationship between microhemorrhages and perivascular Aβ, and expanded our investigations to even more subtle small vessel wall damage and its direct association to Aβ pathology in our human stroke cohort.

Certainly, as we have chosen a descriptive study design, we currently can just speculate on the pathophysiological causality of the detected relationship between CKD, cerebral microhemorrhages and brain Aβ.

One further limitation is the comparably small animal number exhibiting Aβ, which is explained by the use of a nontransgenic non‐AD animal model. Indeed, the investigation of SHRSP instead of a transgenic experimental model should have provided insights into the more physiological interplay between age, CKD, small vessel wall damage and Aβ pathology. We did not determine the exact proportion between the main Aβ isoforms here; however our model should mimic increased Aβ_1‐42 ratios in the deposited material, as in human sporadic Aβ pathology 20, 28. We, moreover, did not give any details about the inflammatory state in SHRSP, albeit own pilot data reveal increased plasma cytokine levels and microglial activation surround small vessel walls displaying plasma protein leakage (unpublished data). As there are, however, well‐documented associations between cerebral Aβ retention, systemic and local inflammation 25, future studies in SHRSP will have to address this specific connection.

Since we used a human stroke cohort and most of the patients have suffered from arterial hypertension, vascular and Aβ pathologies are most likely biased towards higher frequencies compared to those of aged controls.

Serum creatinine and urea were available in only around 50% of the patients, with only two patients exhibiting a moderate reduction of their glomerular filtration rate (for details please see Supporting Information Table S2). And, although associations between serum Aβ and renal function have been reported recently 22, 39, exploratory statistical analysis failed to reproduce the link between CKD, small vessel wall damage and brain Aβ in our human sample. Missing statistical significances may result from too many subjects revealing kidney function within the normal range, too small sample sizes and dichotomous quantifications of cerebral pathologies. Future studies will thus have to address this issue by providing larger cohorts and by quantifying cerebral pathologies in a more quantitative manner.

Finally, as it was beyond the scope of our study we did not report on the frequencies of cerebral amyloid angiopathy and on the prevalence of pericapillary/perivascular Aβ accumulations around small vessels without discernable wall damage.

In SHRSP, we detected age‐independent effects of chronic kidney damage on brain Aβ, which was moderated by cerebral microhemorrhages. There was moreover a direct association between the occurrence of brain microhemorrhages and cerebral Aβ in SHRSP, and qualitative analysis of exemplary human stroke brains revealed the common presence of cerebral Aβ deposits close to cerebral small vessel wall damage. In conclusion, our results yield important findings concerning the relationship between chronic renal disease, cerebral small vessel wall damage and brain Aβ. They emphasize that chronic renal damage and small vessel wall pathology, each independently and combined with each other may promote cerebral Aβ formation. Prevention, early diagnosis and treatment of kidney damage should be a promising purpose to protect from cerebral Aβ retention, subsequent cognitive decline and dementia development.

CONFLICT OF INTEREST

The authors declare that there are no conflicts of interest.

Supporting information

Additional Supporting Information may be found in the online version of this article at the publisher's web‐site:

Table S1. Frequencies and age of onset are given for tubulointerstitial kidney damage, cerebral microhemorrhages and brain Aβ in all 80 SHRSP under investigation.

For the brain, the absence of microhemorrhages/small perivascular bleeds was coded as 0, and their presence was coded as 1. Cerebral parenchymal/perivascular Aβ accumulations were quantified in the same binary manner (0 = absent, 1 = present). For statistical analysis using a general linear model, number of cerebral microhemorrhages and amount of parenchymal/perivascular Aβ accumulations were additionally counted in cortical and subcortical regions. For the kidney, occurrence of vascular pathologies is given (0 = absent, 1 = existent). Severity of renal peritubular capillary erythrocyte accumulations ranges from 0 to 3 and the severity of renal tubular protein cylinders ranges from 0 to 3, too. Tubulointerstitial renal damage represents the sum of the degree of peritubular capillary erythrocyte accumulations and of the degree of tubular protein cylinders ranging from 0 to 6. See Materials & Methods for further details.

Table S2. Demographics, biomarkers and histopathology in all 16 patients under investigation.

F, female; M, male; NA, not applicable; 0 – absent; 1 – present.