Abstract
Vascular abnormalities are a key feature of Alzheimer's disease (AD). Imaging of cerebral vascular reactivity (CVR) is a powerful tool to investigate vascular health in clinical populations although the cause of reduced CVR in AD patients is not fully understood. We investigated the specific role of tau pathology in CVR derangement in AD using the rTg4510 mouse model. We observed an increase in CVR in cortical regions with tau pathology. These data suggest that tau pathology alone does not produce the clinically observed decreases in CVR and implicates amyloid pathology as the dominant etiology of impaired CVR in AD patients.
Introduction
Vascular abnormalities are fundamentally entwined in the pathogenesis of Alzheimer's disease (AD). 1 Estimates of vascular health (such as regional measurements of perfusion) can be sensitive correlates of disease progression, with potential to discriminate control, mild cognitive impairment and AD cohorts. 2 Indeed, there is a persuasive argument as to the causal nature of these correlations,3, 4, 5 with vascular pathology initiating downstream symptoms. However, understanding the precise link between vascular health and other features of AD pathophysiology remains a key challenge, as efforts continue to unearth novel therapeutic pathways to treat this devastating illness.
Non-invasive imaging of regional cerebral vascular reactivity (CVR) using magnetic resonance imaging (MRI) is a powerful tool to investigate vascular health in clinical populations. 6 Cerebral vascular reactivity represents the potential of blood vessels to dilate or constrict in response to a challenge; typically a change in arterial [CO2] driven by a controlled adjustment to the CO2 balance of inhaled gas. Several studies have observed a reduction in CVR in AD patients in comparison with healthy controls.7, 8, 9, 10 Moreover, there is tentative evidence that measures of CVR could reveal vascular dysfunction before abnormalities in baseline cerebral blood flow (CBF),7, 8 thus meeting the urgent need for an early biomarker of disease, preceding established measures of neurodegeneration, such as structural MRI.
The cause of impaired CVR in AD patients is not fully understood. A number of mechanisms have been proposed, some directly linked to irregular amyloid-β (Aβ) deposition 11 and others more broadly associated with the tau-driven neurodegenerative cascade.6, 11 The purpose of the present study was to estimate CVR in the presence of tau-induced pathology but without the amyloid abnormalities that are found in AD patients, to investigate the specific effect of tau pathology on CVR measures of vascular health. To disentangle the contributions of amyloid pathology and tau-driven neurodegenerative processes to compromised vascular health in AD, we used arterial spin labelling (ASL) MRI to image CVR in the rTg4510 model of AD.12, 13 The rTg4510 mouse presents with tau pathology (and marked neurodegeneration) but no irregular Aβ deposition. Our observations were counterintuitive; cortical CVR was not only preserved in regions of tau pathology, but was significantly increased in comparison with litter matched wild-type (WT) controls. To our knowledge, this is the first study to measure CVR in a tau model of AD and provides evidence that neither tau pathology, nor neurodegeneration per se, are major drivers of impaired CVR in AD. These data suggest that other pathologic mechanisms, notably amyloid pathology, are more important factors underlying the compromised CVR observed in AD patients.
Materials and methods
Transgenic Mice
Generation of homozygous rTg4510 transgenic mice has been reported previously. 12 rTg4510 mice were licensed from the Mayo Clinic (Jacksonville, FL, USA) and bred for Eli Lilly by Taconic (Germantown, MD, USA). Mice were imported to the United Kingdom for imaging studies at the UCL Centre for Advanced Biomedical Imaging (CABI), London. All studies were performed in accordance with the United Kingdom Animals (Scientific Procedures) act of 1986. All studies were conducted under the approval of University College London.
Magnetic Resonance Imaging
Magnetic resonance imaging was performed with a 9.4 T VNMRS horizontal bore scanner (Agilent Inc., Santa Clara, CA, USA). A 72-mm inner diameter volume coil (Rapid Biomedical, Columbus, OH, USA) was used for radio frequency transmission and signal was received using a 2 channel array head coil (Rapid Biomedical). Mice were anesthetized under 2% isoflurane in medical air. Anesthesia was then maintained at 1.5% throughout imaging and animals were free breathing. Core temperature and respiration were monitored using a rectal probe and pressure pad (SA Instruments, Stony Brook, NY, USA). Mice were maintained at ∼36°C to 37°C using heated water tubing and a warm air blower with a feedback system (SA Instruments). In this study, two cohorts of 7-month-old female rTg4510 and WT litter matched control mice were imaged (n=6 per group, scanned in an interleaved manner). Histologic analysis of cortical neurofiblirary tangle density from a cohort of ten 7.5-month rTg4510 and WT female mice from the same litter returned a mean (s.d.) cortical neurofiblirary tangle density of 35.5 cell/mm2 (16.5) and 0.03 (0.02), respectively. Female rTg4510 mice develop tau pathology at a faster rate than males and therefore present with more pronounced pathology at 7 months of age. 14 An anatomic reference scan was acquired using a fast spin echo sequence with the following parameters: field of view=25 mm × 25 mm, 20 slices (1 mm slice thickness), repetition time=3.1 seconds, echo time=48 ms, echo train length=8, and matrix size=256 × 256. Cerebral blood flow images were acquired using a flow-sensitive alternating inversion recovery sequence with the following parameters: 5 slices, 1 mm slice thickness, repetition time=3.5 seconds, inversion time=1.5 seconds, slice selective inversion=12 mm (pulse bandwidth 20,000 Hz 15 ), 3 shots, interleaved tag-control images, kzero=4, matrix size=64 × 64, field of view=25 mm × 25 mm; giving a multi-slice ASL (perfusion weighted) image every 21 seconds. Baseline ASL images were acquired for 5 minutes before the inhaled gas mixture was switched from 100% medical air to 95% medical air and 5% CO2 for 5 minutes, imaging was then continued for 3 minutes after switching back to medical air. This paradigm was repeated three times for each mouse with the average temperature and respiration before and after each acquisition reported. All 12 mice underwent a single imaging session and no data were excluded from the analysis.
Statistical parametric Mapping (SPM, http://www.fil.ion.ucl.ac.uk/spm/) was applied to the perfusion-weighted ASL images to investigate possible differences in the CBF response to hypercapnia in the rTg4510 cohort in comparison with the WT control animals. Processing steps included spatial smoothing (0.8 mm full width at half maximum Gaussian kernel), first-level analysis of each subject's time series using an on/off regressor derived from the applied CO2 stimulus paradigm (and convolved with a standard haemodynamic response function), and spatial registration to a template image (http://www.spmmouse.org/). Identical steps were applied to each data set from each animal and no subjective input was required apart from a manually drawn mask to exclude signal from outside the brain for each mouse. Statistical inference used a voxelwise second-level, mixed-effects, two sample t-test to assess differences in CVR between the rTg4510 and WT groups. In addition, a region of interest (ROI) was manually drawn in the frontal cortex on the perfusion-weighted images at baseline for each of the animals to extract ROI-level time series data with minimal postprocessing.
Results
Figure 1A shows the mean CBF across the cohort of rTg4510 and WT mice within the frontal cortical ROI at baseline and during the hypercapia challenge (5% CO2 in air). The data are averaged over all three repeated CVR challenges and the error bars represent the between-subject variance. At baseline, the CBF was highly comparable between the groups (t-test, P=0.95). However, during hypercapnia the CBF increase (mean % increase (s.e.m.)=61% (10)) from baseline was greater in the rTG4510 cohort in comparison with the WT control group (mean % increase (s.e.m.)=26% (7)) (t test, P=0.02). Figure 1B shows representative perfusion-weighted images for a single animal in each cohort (every fifth image in the time series is displayed (1 image acquired every 21 seconds)). The mean (s.d.) temperature and respiration rate within the WT and rTg4510 groups were 36.6 (0.5) and 36.3 (0.4) and 98 (28) and 87(14), respectively, immediately before the ASL acquisitions and 36.7(0.5) and 36.7 (0.6) and 105(13) and 92 (11) after the acquisitions. There were no significant differences between the groups.
(
Figure 2 shows statistical maps from the SPM mixed-effects, two-sample, voxel-wise comparison between the CVR to hypercapnia in the rTg4510 and WT cohorts, overlaid on a registered anatomic reference scan from one of the rTg4510 subjects. The rTg4510 mice displayed significantly increased CVR across the cortex in comparison with the WT controls, confirming the regional nature of the phenomenon, which is concordant with the dominant cortical expression of tau pathology in this model.12, 13Figure 3 shows T2-weighted structural images. Marked cortical atrophy and ventricular enlargement is evident due to tau-induced neurodegeneration as previously described.11, 12
Statistical maps after second-level group analysis to investigate differences in the cerebral blood flow (CBF) response to hypercapnia between the rTg4510 and wild-type (WT) groups (P<0.01 uncorrected). Regions in red/yellow represent a greater CBF response to hypercapnia in the rTg4510 mice in comparison with the WT control group. Regions in blue/green represent a greater CBF response to hypercapnia in the WT control mice in comparison with the rTg4510 group. Statistical maps are overlaid on a representative anatomic reference scan from a single rTG4510 mouse.
Discussion
In this study, ASL MRI was applied to non-invasively image CVR in the rTg4510 model of AD. It is well acknowledged that the rTg4510 model expresses tau in the forebrain and presents with marked cortical atrophy at 7 months of age.12, 13, 16 Despite pronounced neurodegeneration, we measured an increase in cortical CVR in the 7-month rTg4510 cohort in comparison with litter matched WT controls. This is the first time CVR has been investigated in a selective tau model of AD and suggests that tau pathology (and associated neurodegeneration) in itself does not impair CVR measures of vascular health. The finding in the current study, along with reports of reduced CVR in both AD patients 6 and also in mice specifically expressing Aβ,17, 18 provides evidence that impaired CVR in AD may be pathologically specific to Aβ versus tau.
There is growing evidence supporting the link between Aβ and vascular pathology. 6 The causative role of Aβ is particularly compelling given the location and morphology of amyloid plaques in cases of amyloid angiopathy. However, other mechanisms not specific to aberrant Aβ have also been implicated in vascular pathology in AD such as inflammation, 19 thrombin, 20 and cholinergic dysfunction 21 all of which may coexist with tau pathology and/or neurodegeneration. While the data presented in this study does not directly discount the importance of these contributions, it does suggest that tau itself is not a driver of impaired CVR and supports the hypothesis of a fundamental role of amyloid pathology in the vascular abnormalities observed in AD. This link is further supported by a recent study that reported a strong regional correlation between CVR deficits and mapping of amyloid from earlier Pittsburgh compound B PET (positron emission tomography) studies. 7
The increased cortical CVR in the rTg45010 cohort in comparison with the WT may reflect vascular compensatory processes caused by reduced oxygen extraction fraction due to abnormalities in capillary flow patterns. 22 Regional hyperperfusion has been previously observed in AD patients in the prodromal stage.2, 23 Another study observed an increased blood oxygen level-dependent response in AD patients relative to age-matched controls in specific brain regions during a word retrieval task. 24 However, the precise mechanisms underlying the increase in CVR in the rTg4510 cohort found in this work are currently unknown and warrant further investigation in future studies. Previous telemetry measurements of mean arterial pressure in anesthetized (1.5% isoflurane) rTg4510 and WT mice at 8.5 months of age returned highly comparable mean arterial blood pressure estimates between the two groups (data not shown).
Arterial spin labelling MRI can yield whole brain maps of CVR non-invasively with good spatial and temporal resolution, and as such is a powerful and clinically relevant approach to estimate CVR. Two previous studies have applied ASL to measure CBF changes during hypercapnia in the normal mouse brain, reporting a mean increase of 33% across the whole brain 25 and 30% across the cortex, 26 which are in good agreement with the response to hypercapnia in the WT cohort reported in this work.
Accurate staging of preclinical AD is a major challenge to therapeutic development, where early drug intervention may improve treatment efficacy. Positron emission tomography imaging of brain amyloid using Pittsburgh compound B and invasive cerebrospinal fluid sampling of Aβ or tau proteins can provide early diagnosis but these techniques have limited affinity to clinical status and disease progression. Structural MRI closely correlates to clinical status but captures pronounced neurodegeneration that occurs relatively late in the disease. Abnormalities in brain function detected using PET measures of glucose consumption or MRI estimates of cerebral hemodynamics may be sensitive to processes upstream of gross atrophy, while providing a close correlate of clinical status in earlier stages. 2 This is particularly important in the context of increasing interest in the study of preclinical AD 27 and in conducting efficient prevention trials for novel treatments. Thus, understanding the underlying mechanisms driving impaired CVR and other measures of vascular health in AD patients is becoming increasingly important.
To conclude, we report an increased cortical CVR in the rTg4510 model of AD, despite marked, tau pathology and associated neurodegeneration. This novel finding supports the specificity of irregular Aβ pathology in the etiology of impaired CVR measures of vascular health in AD patients.
Footnotes
The authors declare no conflict of interest.