Loss of aquaporin-4 results in glymphatic system dysfunction via brain-wide interstitial fluid stagnation

  1. Ryszard Stefan Gomolka
  2. Lauren M Hablitz
  3. Humberto Mestre
  4. Michael Giannetto
  5. Ting Du
  6. Natalie Linea Hauglund
  7. Lulu Xie
  8. Weiguo Peng
  9. Paula Melero Martinez
  10. Maiken Nedergaard  Is a corresponding author
  11. Yuki Mori  Is a corresponding author
  1. Center for Translational Neuromedicine, University of Copenhagen, Denmark
  2. Center for Translational Neuromedicine, University of Rochester Medical Center, United States
  3. Department of Neurology, University of Pennsylvania, United States
  4. School of Pharmacy, China Medical University, China
8 figures, 4 tables and 2 additional files

Figures

3D-CISS MRI, CSF and interstitial space volumetry in vivo.

Overlaid 3D surface images of the co-registered and averaged 3D-CISS brain volumes, and whiskers-box plots comparing (A) the brain volumes and (B) segmented CSF space volumes from 6 WT and 5 AQP4 KO …

Diffusion-weighted MRI in vivo and ex vivo, vascularity, and AQP4 cellular surface expression ex vivo.

Radar plots showing statistical significances for the differences between the slow diffusion measures among 15 parenchymal ROI assessed, for average ADC (A) and D, and in (B) 3 diffusion-encoding …

Figure 2—source data 1

DWI and IVIM-DWI estimates in vivo - source data.

https://cdn.elifesciences.org/articles/82232/elife-82232-fig2-data1-v3.xlsx
Figure 2—source data 2

DWI, IVIM-DWI, and T1 estimates in the phantom - source data.

https://cdn.elifesciences.org/articles/82232/elife-82232-fig2-data2-v3.xlsx
Figure 2—source data 3

Immunohistochemistry: vascular labeling - source data.

https://cdn.elifesciences.org/articles/82232/elife-82232-fig2-data3-v3.xlsx
Figure 2—source data 4

Immunohistochemistry: AQP4 channel labeling - source data.

https://cdn.elifesciences.org/articles/82232/elife-82232-fig2-data4-v3.xlsx
Dynamic contrast-enhanced MRI in vivo (Figure 3—source data 1).

(A) 3D multiplanar reconstructions of dynamic-contract enhanced (DCE) MRI – sagittal slices from mean images of 6 WT (top) and 5 AQP4 KO (bottom) using 3D-FISP, after applying gadobutrol injection …

Appendix 1—figure 1
Supplementary results for MR CSF space volumetry and MR diffusion evaluation.

(A) Whiskers-box plot comparison of segmented CSF regions, where no statistical difference between KO and WT was found, along with overlaid 3D surface images of separate compartments (colored) from …

Appendix 1—figure 2
Supplementary results for dynamic contrast-enhanced MRI in vivo and correlational analysis.

(A) 3D multiplanar reconstructions of dynamic-contract enhanced (DCE) MRI – midsagittal, parasagittal, and lateral along with orthogonal axial and coronal slices from mean 3D FISP images from 5 AQP4 …

Appendix 1—figure 3
Schematic flow chart of the algorithm for the 3D-CISS-based CSF space volumetry and cisternography in Aqp4(-/-) (KO) and Aqp4(+/+) (WT) mice, including initial preprocessing (bias field correction and brain extraction) of the computed 3D-CISS volume (A) and subsequent automatic CSF space segmentation (B), based on representative 3D-CISS volume from a single animal.
Author response image 1
Inverse Laplace transform (ILT) performed using INSPECT algorithm assuming presence of 2 spectral components.

Presences of fast diffusivities (>10 um2/ms, perfusion-like) components was confirmed in aggregated ADC spectra from 3 regions analyzed (A-C, see ‘Distribution diffusion-encoding direction-wise’). …

Author response image 2
3D-CISS-based regional brain morphometry in Aqp4 (+/+) and Aqp4 (-/-) mice brains.

Legend: OLF-olfactory, CA-cingulate, VIS (V1)-visual, SS (S1)-somatosensory, M-motor; AUD-auditory, HIP-hippocampus, PERI-perirhinal, TH-thalamus, HAB-habenula, HY-hypothalamus, MB-midbrain, …

Tables

Table 1
Summary of demographic characteristics of the animals used in 8 experiments performed, and (B) details of MRI protocols and acquisition parameters for evaluation of the brain CSF and ISF spaces, employed in the current study.

Legend: CryoProbe – cryogenically-cooled MR coil; TA - time of acquisition; Tx/Rx - transmit/receive; TR - time to repetition; TE - time to echo; FA - flip angle; FOV - field of view; Av. / Rep. – …

ANo.of animalsAge [weeks]Body weight [g]Respiration [bpm]
KOWTMale [%]Overall (mean ±SD)pOverall (mean ±SD)pOverall (mean ±SD)p
3D-CISS volumetry5672.713.8±1.8NS27.3±2.1NS176±30NS
MR-DWI6641.710.4±0.7NS22.4±3.0NS175±11NS
DCE-MRI5645.513.4±1.6NS27.3±2.2NS182±14NS
AQP4 expression-45010.0±0.0NS25–30N/A--
Vascular density66014.6±1.0NS24.3±4.6NS--
CSF production5645.515.4±0.5NS28.7±4.2NS--
ISF space volume estimation using TMA8205510.0±0.0NS25–30N/A--
Brain water content3562.512.0±1.0NS25–30N/A--
BMR sequences and acquisition parameters
Sequence (Tx/Rx coil; slice orientation)TR [ms]TE [ms]FA [deg]Av./Rep.Voxel size [mm3](interpolation)Bandwidth [Hz/pix]Matrix sizeTA
MR CSF space volumetry
3D-TrueFISP (CryoProbe; sagittal)5.22.65020.033×0.033×0.033 (2.0×1.6×1.0)26019.2×12.8×12.827 min
MR-DWI (in vivo and ex vivo) (δ=3 and Δ=10ms for gradient duration and separation times)
*(b-values (Av.>1)): 40, 50, 59, 70, 92, 113, 165, 197, 238, 342, 445, 649, 854, 1057 (2), 1564 (2), 2071 (2), 3081 (2) s2/mm
2D-EPI (volumetric; axial)3600309030.15×0.15×0.5 (0.2 mm gap, 16 slices)330716.2×14.4×11.220–30 min (respiratory-gated)
DCE-MRI via CM-injection
3D-FISP (CryoProbe; sagittal)3.261.631510.1×0.1×0.178119.2×12.8×12.890 min
Microbeads phantom ex vivo - T1 mapping
(VTR: 12000, 9000, 6500, 4000, 2000, 1000, 800, 500, 300, 100, 80, 50, 15ms) (VFA: 45° and 90°)
2D-RARE (volumetric; axial)VTR3.19020.1×0.1×3.067116.2×16.2×13 h 20 min
2D-RARE (volumetric; axial)120003.1VFA30.1×0.1×3.067116.2×16.2×11 hr 40 min
  1. *

    Presented averages of measured values from 3 diffusion encoding directions are slightly higher than the set-up, due to gradient preparation time.

Table 2
Summary of findings for (A) average ADC and D-IVIM, (B) direction-wise MR diffusion and pseudodiffusion among 21 ROI assessed, along with statistical scoring.

Asterisks reflect ‘p’ significance values from the nonparametric Mann-Whitney test comparing diffusion measures between KO vs. WT animals ROI-wise (total n=21, balanced groups), along with the …

AROIAverage ADCAverage D-IVIMIVIM
FindingSignificanceFindingSignificanceD*, Fp,Fp x D*
Cerebral cortexOLFKO>WT*-NS-
CA / RSP-NS-NS-
VIS-NS-NS-
SSKO>WT*KO>WT*-
AUDKO>WT*KO>WT*-
HIPKO>WT**KO>WT*-
PERI-NS-NS-
Brain stemTHKO>WT**KO>WT**-
HAB-NS-NS-
HY-NS-NS-
MBKO>WT*KO>WT*-
PAGKO>WT*-NS-
HBKO>WT**KO>WT**-
Cerebral nuclei and tractsCPKO>WT**KO>WT*-
WMKO>WT*KO>WT*-
CSF space3V-NS-NS-
LV-NS-NS-
4V-NS-NS-
PCS-NS-NS-
CoW-NS-NS-
CerebellumCB-NS-NSKO>WT
Fp = 0.0649
BROIIVIMADC / D
Direction Z(cranio-caudal)Direction X(bilateral)Direction Y(ventral-dorsal)Direction Z (cranio‐caudal)Direction X (bilateral)Direction Y (ventral‐dorsal)
FindingSignif.FindingSignif.FindingSignif.FindingSignif.FindingSignif.FindingSignif.
Cerebral cortexOLF-NSFp x D*
KO>WT
*-NSKO>WT**/*-NS-NS
CA / RSPD*
KO>WT
*-NS-NS-NS-NS-NS
VIS-NS-NS-NS-NSKO>WT*/-KO>WT*/-
SS-NS-NS-NSKO>WT-/*KO>WT**/
P=0.056
KO>WT*/-
AUD-NS-NS-NSKO>WT*/*KO>WT**/*KO>WT*/*
HIP-NS-NS-NSKO>WT*/*KO>WT*/-KO>WT**/*
PERID*
KO>WT
**-NS-NS-NSKO>WT*/--NS
Brain stemTH-NS-NS-NSKO>WT**/**KO>WT*/-KO>WT*/-
HAB-NS-NS-NS-NS-NS-NS
HY-NS-NS-NS-NS-NS-NS
MB-NS-NS-NSKO>WT-/*KO>WT*/-KO>WT**/-
PAG-NS-NS-NSKO>WT*/**KO>WT**/-KO>WT*/-
HB-NS-NS-NSKO>WT*/*KO>WT**/-KO >WT*/*
Cerebral nuclei and tractsCP-NS-NS-NSKO>WT**/**KO>WT*/--NS
WMD*
KO>WT
**-NS-NS-NS-NSKO>WT*/-
CSF space3V-NSFp /
Fp x D*
KO>WT
** / *-NS-NS-NS-NS
LV-NS-NS-NS-NS-NS-NS
4V-NS-NS-NS-NS-NS-NS
PCS-NS-NS-NS-NS-NS-NS
CoW-NS-NS-NS-NS-NS-NS
CerebellumCB-NS-NS-NS-NS-NS-NS
Table 3
DCE-derived parameters from 21 ROI (matching those analyzed by means of MR-DWI) in 5 KO and 6 WT littermate mice, along with associated nonparametric pair-wise statistics using a two-tailed Wilcoxon signed-rank test and median ± standard deviation (SD) values strain-wise.

Legend: Mean AUC – mean from areas under the DCE curves along with associated p-statistical values (p-stat) from nonparametric Mann–Whitney U-test ROI-wise; Duration – duration of significantly …

#WT vs. KO time-series different?Mean AUC [a.u.]Arrival time [min]Time- to-peak [min]Peak intensity [a.u.]Interstitial tracer accumulation time [min]Duration [min]
ROIAqp4(+/+)Aqp4(-/-)p-statAqp4(+/+)Aqp4(-/-)ΔAqp4(+/+)Aqp4(-/-)ΔAqp4(+/+)Aqp4(-/-)rel. Δ [%]Aqp4(+/+)Aqp4(-/-)Aqp4(+/+)Aqp4(-/-)rel. Δ [%]
StartEndStartEnd
CorticalOLFNS72075619NS660615291067925.5438039703832–15.8
CA/RSP**1662616**7708990-1271255.6599066903225–21.9
VIS**1699529**78-18990-133972.7549073903718–51.4
SSNS537438NS67-1201648712.5123211252115–28.6
AUDNS454340NS7708989010550.06790N/AN/A24N/A<-100
HIP**32581622*5.57.5-2638118462252.2459074904617–63.0
PERIP=0.05235132659NS45-18390-7594425.4539057903834–10.5
Brain stemTH*347125570.0522–32–308290-8614526.2539058903833–13.2
HABNS23791992NS911-28990-1453620.0549058903733–10.8
HYNS79086766NS4403934512310514.6235921543734–8.1
MB*61254013*34-1668317875635.6428453904338–11.6
PAGNS2574022474NS810-24338535031111.1357531614131–24.4
HBNS92918124NS2–32–304238413311414.2286627613935–10.3
Nuclei / tractsCPNS17361167NS48-48990-1392535.9549058903733–10.8
WM**69535445**4.56.5-25153-21058023.8377338693732–13.5
CSF spaceCMNS2327619903NS1102424039733914.6155016473632–11.1
PCSNS1516312498NS2–32–3041311022418716.5235920503731–16.2
3VNS1222810441NS56-14546-118015116.1347633614329–32.6
LVNS82886555NS611-59090015712321.7549058903733–10.8
CaudalSSS*58321430**13-2648824772271.4389075905316–69.8
CBNS73696151NS67-1475912978017.5297531674737–21.3
MEDIAN ±SD6125±16534013±13065.0±2.26.5±2.863±2381±27568724±1937±732±7
KO vs. WT difference
(Wilcoxon signed-rank test)
WT >KO, p<0.0001WT <KO, p<0.01WT ~KO, p=0.2971WT >KO, p<0.0001WT >KO, P<0.0001
Table 4
Descriptive summary of findings presented in the current study.

Bold italic font highlights the regions of the largest differences found between KO and WT animals, by means of 3 MRI and 5 physiological and histological assessment methods applied.

Magnetic resonance imaging in vivo
MeasurementGeneral findings in KO compared to WTRegion of largest difference
non-invasive3D cisternography- 5–10% larger brain volumes
- 22–29% smaller CSF space / brain volume ratio
ventricular space, 3rd ventricle
2D diffusion-weighted imaging5–6% higher ADC and D(difference present for average and in all diffusion directions)
- thalamus, hindbrain, periaqueductal gray regions, auditory cortex and hippocampus
Higher Fp and Fp × D* only in the 3rd ventricle
invasiveDynamic CSF tracer imaging via cisterna magna`reduced parenchymal tracer influx and evacuation- influx: cortical ROI, hippocampus
- efflux: superior sagittal sinus
Physiological and histological measurements
MeasurementGeneral findings in KO compared to WTRegion of largest difference
ex vivo and histologyBrain water content~6% larger brain water content
AQP4 expression(only WT) heterogenous AQP4 expression in the brainlargest expression in the thalamus, hippocampus, habenula
Vascular densitysimilar vascular density to WTtrend for larger vascular density in the thalamus and olfactory area
in vivoCSF productionsimilar CSF production to WT
Real-time ionophoresis TMAISF space volume larger

Additional files

Download links