Perfusion Single Photon Emission Computed Tomography in a Mouse Model of Neurofibromatosis Type 1: Towards a Biomarker of Neurologic Deficits

Abstract

Neurofibromatosis type 1 (NF1) is a single-gene disorder affecting neurologic function in humans. The NF1+/– mouse model with germline mutation of the NF1 gene presents with deficits in learning, attention, and motor coordination, very similar to NF1 patients. The present study performed brain perfusion single-photon emission computed tomography (SPECT) in NF1+/– mice to identify possible perfusion differences as surrogate marker for altered cerebral activity in NF1. Cerebral perfusion was measured with hexamethyl-propyleneamine oxime (HMPAO) SPECT in NF1+/– mice and their wild-type littermates longitudinally at juvenile age and at young adulthood. Histology and immunohistochemistry were performed to test for structural changes. There was increased HMPAO uptake in NF1 mice in the amygdala at juvenile age, which reduced to normal levels at young adulthood. There was no genotype effect on thalamic HMPAO uptake, which was confirmed by ex vivo measurements of F-18-fluorodeoxyglucose uptake in the thalamus. Morphologic analyses showed no major structural abnormalities. However, there was some evidence of increased density of microglial somata in the amygdala of NF1-deficient mice. In conclusion, there is evidence of increased perfusion and increased density of microglia in juvenile NF1 mice specifically in the amygdala, both of which might be associated with altered synaptic plasticity and, therefore, with cognitive deficits in NF1.

Keywords

amygdala cerebral perfusion histology mouse model neurofibromatosis type 1 small animal SPECT

INTRODUCTION

Genetic disorders that result in impaired neuronal plasticity by disrupting signaling pathways are a common cause of cognitive impairment in children.¹ The most frequent single-gene disorder affecting neurologic function in children is neurofibromatosis type 1 (NF1).^2–4 The mutated gene that causes NF1 is referred to as the NF1 oncogene. The product of the NF1 oncogene is neurofibromin, a multidomain molecule that is involved in the regulation of several intracellular processes including RAS (‘Rat sarcoma’ oncogene family) activity and adenylyl cyclase-mediated signal transductions. By this, neurofibromin has a key role in signal transduction and regulation involved in synaptic plasticity, memory, and learning.⁵

Besides impaired motor coordination,⁶ learning disability is the most frequent neurologic complication in children with NF1, occurring in 30% to 60% of cases.⁵ More precisely, children with NF1 often show a considerable lag of achievement/performance despite a normal or only slightly reduced IQ (intelligence quotient).^5,7 Learning capabilities tend to improve towards normal levels during adolescence and young adulthood, and late maturation of the IQ can be observed.⁸ Nevertheless, the quality of life is often reduced in the long term because of lack of education at school and vocational training during childhood, because deficits acquired in this early formative years are not re-compensated later. Thus, there is a need for treatment of impaired learning/learning disability in children with NF1. Normal IQ and the fact that learning disabilities tend to decrease with age suggest that effective treatment might be possible.

Further evidence for good treatability of learning disabilities in NF1 has been provided by research in mouse models that are heterozygous for a null mutation of the NF1 gene.⁹ The NF1+/– mice with a germline mutation resulting in hemizygote knockout of the NF1 gene exhibit cognitive abnormalities such as deficits in learning, attention, and motor coordination that resemble deficits seen in children with NF1.^9–12 Costa and co-workers found the learning deficits in NF1+/– mice to be associated with impaired long-term potentiation due to (i) excessive RAS activity that leads to increased γ-amino butyric acid (GABA) inhibition and (ii) impaired N-methyl-D-aspartate receptors.¹³ The same authors also showed that learning deficits in NF1+/– mice can be reduced not only by genetic manipulation but also by pharmacological interventions that inhibit RAS function.^9,13,14 The H-RAS (Harvey rat sarcoma viral oncogene homolog) and other guanosine triphosphatase-activating proteins also have been implicated in long-term potentiation via their modulation of excitatory neurotransmission. These deficits in synaptic plasticity have also been linked to NF1-associated impairment of cognition in humans.¹⁵ Furthermore, neurofibromin has been demonstrated to modulate excitatory synaptic function, namely by interacting with the N-methyl-D-aspartate receptor in a large postsynaptic complex.¹³ In very recent work, NF1-associated learning deficits were linked to disruption in the regulation of the classic MAPK (mitogen-activated protein kinase) pathway in the amygdala leading to impaired synaptic plasticity in terms of GABA-mediated inhibition and glutamate excitation, as well as altered expression of key synaptic proteins in the amygdala.¹⁶ These alterations could be rescued by deletion of the Pak1 (p21 protein-activated kinase 1) gene or pharmacological blockade of Pak1 function.¹⁶ Thus, there are various promising targets for the treatment of learning deficits in NF1.

For efficient evaluation of these treatment options in NF1+/– mice, the identification of a biomarker might be useful that strongly correlates with cognitive performance but provides higher test–retest stability than cognitive testing, particularly in mice. Such biomarkers might be derived from functional radionuclide imaging using positron emission tomography (PET) or single-photon emission computed tomography (SPECT) with tracers that are sensitive to synaptic activity. Positron emission tomography with the glucose analog F-18-fluorodeoxyglucose (FDG) measuring the regional cerebral metabolic rate of glucose (rCMRGlc) has been shown to be a useful marker of synaptic activity.¹⁷ Recent proposals to revise guidelines for the diagnosis of neurodegenerative diseases, for example Alzheimer's disease, recommend FDG PET as a biomarker of synaptic dysfunction to complement symptom-based criteria.^18,19 In patients with NF1, FDG PET has revealed reduction of rCMRGlc specifically in the thalamus, as large as approximately 30% in children and considerably smaller, approximately 10%, in adults.^20–22 However, FDG PET of the mouse brain is limited by the spatial resolution of small animal PET systems, which is approximately 1.5 mm. Given the recent advances in instrumentation and reconstruction software, small animal SPECT systems allow spatial resolution of 0.7 mm at adequate statistical image quality, which enables brain imaging in mouse models.²³ The SPECT might be used with tracers for regional cerebral blood flow (rCBF), which also is coupled with synaptic activity. The rCBF and rCMRGlc are highly correlated not only in the healthy brain but also in many diseased states.²⁴ In fact, some guidelines for the diagnosis of neurodegenerative diseases in humans suggest perfusion SPECT and FDG PET as biomarkers for synaptic activity on an equal footing.²⁵ Our group recently evaluated small animal SPECT with Tc-99m-labeled hexamethyl-propyleneamine oxime (HMPAO) for measuring rCBF in mice.²⁶

The primary aim of the present study was to compare rCBF as measured by HMPAO perfusion SPECT between NF1+/– mice and wild-type littermates to identify surrogate markers of synaptic dysfunction in NF1 that might be evaluated as biomarkers of cognitive function in further studies. Histologic analyses were performed to identify structural correlates of altered rCBF.

MATERIALS AND METHODS

Mice

Heterozygous NF1n31 mice, C57BL/6J mice with a targeted NF1 gene mutation,¹⁰ were kindly provided by Professor Dr D Kaufmann (Institute of Human Genetics, University Hospital Ulm, Albert-Einstein-Allee 11, 89070 Ulm, Germany). The NF1n31 mice show deficits in both synaptic plasticity and learning.^9,27 Genotype of all newborn mice was determined using PCR (AccuPrime SuperMix II, Invitrogen, Darmstadt, Germany) with the appropriate primer set (NFX4, NFX414, and MC1-Out III, Invitrogen).

The study included 81 mice in total: 40 heterozygous NF1n31 mice (NF1+/–) and 41 wild-type littermates (WT). Thirty-nine of these mice were included in the in vivo perfusion SPECT experiments with SPECT imaging at two different time points (see subsection ‘SPECT imaging’); 20 NF1+/– and 19 WT mice. Baseline SPECT was performed in all mice, follow-up SPECT was performed in 16 mice from the NF1+/– group and in 18 mice of the WT group. At baseline, the NF1+/– group comprised eight female and 12 male mice, the WT group 11 female and eight male mice. At follow-up, the NF1+/– group comprised six female and 10 male mice, the WT group 10 female and eight male mice. The gender distribution was not significantly different between the NF1+/– and the WT group, neither at baseline (Chi-square test, P = 0.343) nor at follow-up (P = 0.327). Ex vivo well-counter measurements of thalamic FDG uptake were performed in 31 mice (14 NF1+/–, 17 WT; 5.9 to 6.4 weeks of age). The remaining 11 mice were included in histology and immunohistochemistry analyses (six NF1+/– aged 7.2 ± 1.8 weeks, five WT aged 7.7 ± 2.7 weeks). All the animals were kept in the local animal facility on a 12:12 hours light/dark cycle for at least 3 days before the experiment. The temperature was kept at 22 °C, humidity at 50%. The mice had free access to mouse chow pellets and water. All procedures had been approved by the local authorities (Landesamt für Gesundheit und Soziales Berlin, reference number G 0167/10, and Behörde und Gesundheit und Soziales Hamburg, reference number G059-08) and conducted in accordance with the German Animal Protection Act of 18 May 2006 (BGBl. I S. 1206, 1313), last changed 9 December 2010. The manuscript was written up in accordance with the ARRIVE (Animal Research: Reporting In Vivo Experiments) guidelines.

Single-Photon Emission Computed Tomography Imaging

Perfusion-SPECT was performed at two different time points in the same mice. The baseline scan was performed at juvenile age, younger than 10 weeks (6.9 ± 2.0 weeks), the follow-up scan at young adult age, older than 12 weeks (16.8 ± 4.0 weeks). Mice were anesthetized by intraperitoneal injection of ketamine/xylazine (100 mg/kg ketamine and 10 mg/kg xylazine) 10 minutes before tracer injection.

The HMPAO was prepared in high-activity concentrations using fresh (< 2 hours old) eluent of Tc-99m-pertechnetate taken from a standard generator. Aliquots of Ceretec (GE Healthcare, Arlington Heights, IL, USA) were used for the preparation of HMPAO, as previously described.²⁶ The injected dose was 110 ± 64 MBq at baseline, 119 ± 30 MBq at follow-up.

Imaging was performed with a four-head nanoSPECT/CTplus system (Bioscan Europe, Paris, France).²⁸ The anesthetized mouse was positioned on the mouse bed of the Minerve Small-Animal Environment System (Bioscan Europe) with integrated heating to maintain normal body temperature during the scan. General-purpose mouse apertures were used for SPECT imaging, each with nine holes of 1 mm diameter. This configuration provides a sensitivity of 1.2 c.p.s./kBq for the detection of Tc-99m decays. The SPECT imaging was started 10 minutes after HMPAO injection. Projection data were acquired in step-and-shoot mode for 120 seconds at each of the 20 angular steps resulting in a total scan duration of 40 minutes. The energy window was centered at 140 keV with a window width of ± 10%. Transversal images were reconstructed with the iterative algorithm of the HiSPECT multi-pinhole software (SCIVIS) with a pixel size of 0.3 × 0.3 mm² and slice thickness of 0.3 mm. Spatial resolution in the reconstructed images was approximately 0.7 mm full-width-at-half-maximum.²³

Single-Photon Emission Computed Tomography Image Processing

The HMPAO SPECT images were processed as described previously.²⁶ In short, SPECT images were stereotactically normalized using the ‘Normalize’ tool of the Statistical Parametric Mapping software package (SPM8, Wellcome Trust Centre for Neuroimaging, London, UK) with a custom-made HMPAO SPECT template to define the target space. For the scaling of voxel intensities, mean global tracer uptake (mean tracer uptake over the whole brain) was computed as the weighted average of all voxel intensities. Each voxel was weighted by the probability that it belongs to the brain according to the whole-brain probability map of the atlas developed by Ma et al (see next paragraph).^29,30 Voxel intensities of the stereotactically normalized SPECT were scaled to the individual mean global tracer uptake, i.e., each voxel value was divided by the mean global tracer uptake.

The ‘3D digital atlas database of the adult C57BL/6J mouse brain’ provided by Ma et al^29,30 was used for region of interest (ROI) analyses. This database includes a volumetric probabilistic atlas in the anatomic space of the reference brain. The probabilistic atlas is created from 10 C57BL/6J brains and represents 20 segmented bilateral ROIs. The probabilistic atlas consists of 21 images in the reference brain space, one for each ROI plus one for the whole brain. The voxel intensities in these images specify the probability that a given voxel belongs to the corresponding brain structure. The sum over all voxel intensities was scaled to one in each ROI image. Mean uptake of HMPAO in a given ROI was obtained by multiplying the SPECT image voxel-wise by the corresponding ROI image and summing up all voxel intensities of the product.

Statistical Analysis

Analysis of variance was performed to compare the relative HMPAO uptake between NF1 and WT mice within the following ROIs (notation as defined by Ma et al^31,32): neocortex, hippocampus, amygdala, olfactory bulb, basal forebrain and septum, caudate and putamen, thalamus, hypothalamus, central gray matter, superior colliculi, inferior colliculi, rest of midbrain, cerebellum, and brain stem. SPSS Statistics (version 19, IBM, Armonk, NY, USA) was used for this purpose. Age was taken into account as covariate. Bonferroni adjustment for the number of ROIs was not applied.

In addition to the analysis with these a priori defined ROIs, explorative testing for differences in HMPAO uptake between NF1+/– and WT mice was performed on a voxel-by-voxel base using the t-test implemented in SPM 8. The SPECT images were smoothed with an isotropic Gaussian kernel with 1.0 mm full-width-at-half-maximum before testing. The differences were considered significant for P ≤ 0.01 (uncorrected for multiple testing) and a minimum cluster size of 1,000 voxels corresponding to a volume of 1.0 mm³. Voxel-based testing was restricted to the cross-sectional group comparison of NF1+/– versus WT mice separately at juvenile age and at young adulthood. This avoids confounding effects of age-related changes in brain size and structure, which affect testing of longitudinal changes of cerebral perfusion and are difficult to control reliably.

Ex vivo Measurement of Thalamic F-18-Fluorodeoxyglucose Uptake

In vivo SPECT imaging did not reveal reduced blood flow in the thalamus of NF1+/– mice (see Results), which was somewhat unexpected, as PET in humans with NF1 has shown a rather pronounced reduction of FDG uptake specifically in the thalamus (and normal FDG uptake in all other brain structures).^20–22 To test whether the lack of an effect in the thalamus of NF1+/– mice in the in vivo SPECT experiments was because of targeting perfusion rather than glucose metabolism (or a true interspecies difference), we performed postmortem well-counter measurements of FDG uptake in the thalamus of NF1+/– and WT mice. Postmortem well-counter measurements avoid partial volume effects inevitable in in vivo PET imaging. The latter reduces the power to detect effects in substructures of the mouse brain.

The mice were injected with approximately 1 MBq FDG in the tail vein, and were killed after an uptake period of 20 minutes. The brains were removed immediately. The bilateral thalamus was dissected from a 2 mm coronal slice (Bregma – 0.5 to – 2.5 mm) using a standard stereotactical matrix. This approach was verified histologically in two additional mice. Weight was obtained for thalamus and for the rest of the cerebrum (without cerebellum). The FDG uptake of each tissue sample was measured using a γ-well-counter (Wallac Wizard 1470, 5 minutes) and then scaled to its weight (specific FDG uptake). Dead-time and statistical count error were low in all the cases. The thalamus-to-cerebrum ratio was computed, in analogy to the proportional scaling approach used in FDG PET in humans.²¹

Histology and Immunohistochemistry

For histology and immunohistochemisty, mice were anesthetized with sodium pentobarbital (0.8 mg/g body weight, intraperitoneal; Narcoren; Merial, Hallbermoos, Germany) and transcardially perfused with physiologic saline for 1 minute, followed by 4% formaldehyde in 0.1 M sodium phosphate buffer, pH 7.3, for 15 minutes. Brain hemispheres were than post-fixed in 4% paraformaldehyde for at least 48 hours. Paraffin-embedded, 4-μm thick sections were deparaffinized and conventionally stained with hematoxylin–eosin stain. Immunohistochemical analysis was performed as described previously^31,33 using a Bond-Max (Leica Microsystems GmbH/Menarini, Wetzlar, Germany) with antibodies against ionized calcium-binding adapter molecule 1 (IBA1, 1:2,000, Wako 019-19741, Neuss, Germany) to label microglia, glial-fibrillary acid protein (1:1,000, DAKO Z033401, Hamburg, Germany) to label astrocytes, NeuN (1:1,000, Millipore MAB377, Darmstadt, Germany) to label neurons, synaptophysin (1:200, clone Z66, Invitrogen) to label synapses. Slides were developed using the Bond Polymer Refine Detection kit (Menarini/Leica Microsystems, Germany). For the evaluation, whole-tissue sections and TMAs were digitized at 230 nm resolution using a MiraxMidi Slide Scanner (Zeiss MicroImaging GmbH, Oberkochen, Germany).³¹

RESULTS

In vivo Perfusion Single-Photon Emission Computed Tomography At baseline, a significant intersubject genotype effect was detected in the amygdala (P = 0.028), and a tendency towards significance in the olfactory bulb (P = 0.082). In both ROIs, relative HMPAO uptake was increased in NF1+/– mice compared with WT mice (Table 1, Figures 1A and 1B,). The increase of HMPAO uptake in the amygdala of juvenile NF1+/– mice was even more significant (P = 0.004) when the highest value and the lowest value (‘outliers’) were removed in both the NF1+/– and the WT group, as might have been expected from Figure 1A.

Table 1

Relative HMPAO uptake (in %) in the amygdala, olfactory bulb, and thalamus

	Amygdala			Olfactory bulb			Thalamus
	NF1+/−	WT	P	NF1+/−	WT	P	NF1+/−	WT	P
Baseline	87.8±4.1	83.4±5.2	0.028	110.7±5.1	106.0±4.3	0.082	105.8±2.8	106.8±4.9	0.979
Follow-up	85.0±3.6	83.8±5.7	0.282	109.7±4.7	108.3±4.1	0.531	104.3±3.2	105.9±3.7	0.571

HMPAO, hexamethyl-propyleneamine oxime; NF1, neurofibromatosis type 1; WT, wild type.

Figure 1.

Cross-sectional ROI-based comparison of NF1+/– versus WT mice; significantly increased HMPAO uptake in the amygdala (A) and tendency towards increased HMPAO uptake in the olfactory bulb (B) in NF1+/– mice at juvenile age (baseline), but not at young adulthood (follow-up, D and E). No difference of HMPAO uptake in the thalamus between NF1+/– and WT mice, neither at juvenile (C) nor at adult age (F). No difference in thalamic FDG uptake at juvenile age (G). All ROIs were bilateral, i.e., including the corresponding brain structure in both hemispheres. FDG, F-18-fluorodeoxyglucose; HMPAO, hexamethyl-propyleneamine oxime; NF1, neurofibromatosis type 1; ROI, region of interest; WT, wild type.

Relative HMPAO uptake in the thalamus was not different between NF1+/– and WT mice (P = 0.979, Table 1, Figure 1C). The results of the ROI-based analyses were confirmed by the voxel-based tests; there was a significant cluster of increased HMPAO uptake in the NF1 mice in both the amygdala (Figure 2A) and the olfactory bulb (Figure 2B).

Figure 2.

Cross-sectional voxel-by-voxel group comparison of NF1+/– versus WT mice; increased HMPAO uptake in amygdala (A) and olfactory bulb (B) in juvenile NF1+/– mice. Reduced HMPAO uptake in the frontal cortex of young adult NF1+/– mice (C). HMPAO, hexamethyl-propyleneamine oxime; NF1, neurofibromatosis type 1; WT, wild type.

At follow-up, there was no genotype effect on HMPAO uptake in the amygdala nor in the olfactory bulb; HMPAO uptake in NF1+/– mice was reduced to the normal level in WT mice (Table 1, Figures 1D and 1E,). This was confirmed by voxel-based testing, which did not detect a significant effect in these brain regions at follow-up. The only significant genotype effect in the ROI analysis at follow-up was a reduction of HMPAO uptake in the basal forebrain and septum in the NF1+/– mice (P = 0.004). This finding was confirmed by a cluster of significantly reduced HMPAO uptake in the voxel-based testing (Figure 2C).

Testing for age effects on HMPAO uptake by paired comparison of baseline versus follow-up using the general linear model for repeated measures with genotype as intersubject factor revealed a significant decrease in basal forebrain and septum (P = 0.001), caudate and putamen (P = 0.003), central gray matter (P = 0.005), inferior colliculi (P = 0.035), rest of midbrain (P < 0.0005), cerebellum (P = 0.004), and brain stem (P < 0.0005; Figure 3). A significant increase of HMPAO uptake with age was observed in the neocortex (P < 0.0005) and the hippocampus (P < 0.0005). The interaction between genotype and scanning session (baseline, first follow-up) did not reach the level of statistical significance in any of the ROIs. A tendency for a significant interaction was observed in the olfactory bulb (P = 0.068).

Figure 3.

Longitudinal ROI data of the 16 NF1+/– mice and the 18 WT mice with follow-up perfusion SPECT. Shown are mean values of relative HMPAO uptake ± s.e.m. HMPAO, hexamethyl-propyleneamine oxime; NF1, neurofibromatosis type 1; ROI, region of interest; SPECT, single-photon emission computed tomography; WT, wild type.

F-18-Fluorodeoxyglucose Uptake

NF1+/– and WT mice did not differ with respect to any of the following variables: sex, age, body weight, injected volume or FDG dose, weight of thalamus or cerebrum, total counts in thalamus or cerebrum. The thalamus-to-cerebrum ratio was 1.019 ± 0.124 in the NF1+/– mice, 0.976 ± 0.098 in the WT mice. The difference was not statistically significant (t = – 1.083, df = 29, P = 0.288; Figure 1G). Post hoc analysis showed that the power to detect a reduction in NF1+/– mice of 10%, 15%, 20% at α = 0.05 was 78%, 98%, 99.9%, respectively.

Morphologic Analysis

Animals were first analyzed using conventional histology and hematoxylin–eosin stain. Mouse brains of both groups did not exhibit gross pathologic changes in the structural and cellular composition (Figures 4A and 4B). Distinct vascular changes seen as slight enlargement of capillaries were recognized in the amygdala of NF1+/– mice (Figure 4B, lower panel). To detect neuronal loss or changes in neuron distribution, NeuN-labeling of neurons was used (Figures 4C and 4D). Cell counting showed no differences in neuronal density between the two groups (P = 0.804, Figure 4F). Synaptophysin labeling also revealed no gross differences between the groups (Figures 4G and 4H). Both groups showed normal amounts of glial-fibrillary acid protein-positive astrocytes without any localized accumulation in the cortex or the basal ganglia (Figures 5A and 5B). The amygdala did not show any astrogliotic reaction as well. The Iba1 stain revealed microglia appearance in the amygdala with a thickening of the cellular body and processes of some cells, whereas the rest of the brain was inconspicuous (Figures 5C and 5D). Cell counting revealed a higher density of microglial somata in the amygdala of NF1+/– compared with WT mice (Figures 5E to 5G). In two-tailed testing with the t-test for unpaired groups and unequal variances, because Levene's test for homogeneity of variance indicated different variance in the two groups (P = 0.014), the difference did not reach the level of statistical significance (P = 0.174). One-tailed testing for increased microglial cell density in NF1+/– mice, which might be justified by previous reports on microglial activation in NF1,³² revealed a tendency towards significance (P = 0.087).

Figure 4.

Morphological appearance of WT (left) and NF1+/– (right) mice. Lower panels show magnified regions of the amygdala. Scale bar, 100 μm (highest magnification), 200 μm (higher magnification), and 1,000 μm (lower magnification). (A and B) H&E stains reveal only minor changes in the amygdala of NF1+/– mice (B–lowest panel). Some capillaries show enlarged vessel diameter (arrows). (C and D) NeuN-labeling of neurons does not show differences between the groups. (E) Higher magnification of NeuN positive neurons in amygdala (F) Scatter plot of NeuN cell density (cells/mm²) in the amygdala of NF1+/– (filled symbols) and wild-type (open symbols) mice showing no significant genotype effect. (G and H) Also synaptophysin-labeling of synapses did not reveal a difference between NF1+/– and WT mice. H&E, hematoxylin–eosin; NF1, neurofibromatosis type 1; WT, wild type.

Figure 5.

Morphologic appearance of WT (A and C) and NF1+/– (B and D) mice. (A and B) GFAP-labeling of astrocytes does not show differences between the groups. (C and D) Iba1-labeling of microglia shows slight enhancement of microglial cells with enlargement of cell body and processes in NF1+/–. Lower panels show magnified regions of amygdalae. Scale bar, 30 μm (highest magnification, only lowest panel), 200 μm (middle magnification), and 1,000 μm (lower magnification, upper panels). Further magnification of Iba1-labeling in (E, NF1+/–) and (F, WT). (G) Microglial cell density (cells/mm²) in the amygdala of NF1+/– (filled symbols) and WT (open symbols) mice. GFAP, glial-fibrillary acid protein; NF1, neurofibromatosis type 1; WT, wild type.

DISCUSSION

As first major finding of the present study, rCBF in the thalamus did not differ between NF1+/– and WT mice. Given that rCBF is a marker of synaptic activity similar to cerebral glucose metabolism, this finding is in contrast to FDG PET studies in humans with NF1, which consistently showed reduced FDG uptake specifically in the thalamus of NF1 subjects.^20–22 However, ex vivo measurements of thalamic FDG uptake also did not show an effect in NF1+/– mice, suggesting that the lack of an effect in the SPECT experiments was not caused by limitations of HMPAO as perfusion tracer nor by limitations of in vivo small animal SPECT imaging, but most likely indicates a real interspecies difference. Thus, the NF1n31 mouse is not useful as model for thalamic hypometabolism in NF1. In fact, the lack of a thalamic effect in NF1n31 mice suggests that the neurologic deficits in humans with NF1 are not directly associated with thalamic hypometabolism, as NF1n31 mice show NF1-typical neurologic and learning deficits without thalamic hypoactivity. Furthermore, NF1n31 mice do not develop neurofibromas. This might suggest that thalamic hypometabolism in humans with NF1 is caused by reduced peripheral input to the thalamus secondary to disturbance of fiber tracts by neurofibromas. This hypothesis might be tested in further studies.

As a second major finding, relative HMPAO uptake was increased in the amygdala of juvenile NF1+/– mice. The ROI used in the analysis was bilateral, i.e., included the amygdala in both hemispheres, and, therefore, does not allow detection of possible laterality. We tested the amygdala effect for hemispheric laterality as described in our previous work²⁶ and found no indication of left-right asymmetry, neither in ROI- nor in the voxel-based analyses (results not shown). This suggests that the observed relative hyperperfusion in the amygdala of juvenile NF1+/– mice is a symmetric effect affecting both hemispheres more or less equally. The fact that the voxel-based group comparison revealed a significant cluster in one hemisphere only (Figure 2A) is an effect of thresholding the statistical parametric map at the significance level P ≤ 0.01 (at the slightly more liberal threshold P ≤ 0.05, there was a significant cluster of hyperperfusion in the amygdala of juvenile NF1+/– mice in both hemispheres, result not shown). Ex vivo measurement of FDG uptake was restricted to the thalamus, i.e., there were no data of FDG uptake in the amygdala available to corroborate the in vivo findings of perfusion SPECT.

Robinson et al³⁴ performing T2-weighted magnetic resonance imaging at 7 T in 46 NF1+/– mice and 39 NF1+/+ controls, found significantly increased T2 values in NF1+/– mice in the striatum, the brainstem, and the thalamus, brain regions often affected by so-called ‘unidentified bright objects’ in patients with NF1.^35,36 Significant reduction of T2 values in NF1+/– mice was observed in motor- and vision-related brain regions as well as in the hippocampus and the amygdala. The 10.5% reduction in the amygdala was the largest among all significant genotype effects on the T2 signal. Voxel-by-voxel testing for correlation between T2 intensity and motor performance (wire-hanging test) revealed a significant negative correlation between T2 intensity in the amygdala and wire-hanging performance in the group of NF1+/– mice. There was no significant correlation in any other brain region in NF1+/– mice, neither negative nor positive. Although elevation in T2 signal intensity is generally assumed to reflect increase of the water fraction in tissue, a decrease in T2 signal might indicate increased tissue density and low water fraction. It might be worth noting that the sign of the effect in the amygdala of NF1+/– mice is different between the present study and the study of Robinson et al, i.e., increase of HMPAO uptake versus decrease of T2 intensity. This suggests that these effects are not secondary to partial volume effects because of different size of the amygdala in NF1+/– and WT mice, but are rather primary effects in perfusion and tissue density, respectively.

Recently, Molosh et al¹⁶ identified the amygdala as a key structure involved in social learning deficits in NF1+/– mice. They found aberrant glutamate and GABA neurotransmission in BLA (basolateral amygdala nuclei) neurons in NF1+/– mice resulting in functional changes in amygdala networks. Our data do not allow to establish a direct association between these effects and the present finding of hyperperfusion in the amygdala of NF1+/– mice. However, synaptic activity is known to trigger metabolic and vascular changes that can be detected by in vivo brain imaging.³⁷ Thus, we hypothesize that the hyperperfusion in SPECT reflects altered synaptic activity in the amygdala of NF1+/– mice.

Morphologic analyses showed no major structural abnormalities in NF1+/– mice. However, there was some evidence for increased density of microglial somata in the amygdala of NF1+/– mice (Figures 5E to 5G) accompanied by enlarged capillaries (Figure 4B). Statistical testing of the difference in microglial density revealed only a tendency towards statistical significance. The fact that the difference did not reach the level of statistical significance might be explained not only by small sample size, but also by clustering in the NF1+/– group (Figure 5G); the NF1+/– group might be divided into two subgroups, one subgroup with clearly increased microglial density (four of six mice) and one subgroup with normal microglial density (two of six mice). The clustering of microglial cell density into ‘increased’ and ‘normal’ might indicate intersubject variability of the expression of disease-specific pathology.

Microglia surveys the local microenvironment, making direct contact with synaptic spines^38,39 and takes part in remodeling of neuronal circuits.⁴⁰ Synaptic activity is further influenced by microglia through secretion of brain-derived neurotrophic factor, a molecule that is crucial for learning-dependent synapse formation. Activation of microglia might interfere with these interactions and thus result in cognitive dysfunction. At the same time, activated microglia demands more energy, which results in upregulation of local blood flow to supply more oxygen and glucose. Thus, activation of microglia might be associated with both, impaired cognitive function and increased cerebral blood flow. In NF1, microglial activation has been reported previously, mostly associated with brain tumorigenesis.³² Its role in synaptic plasticity and neurologic deficits in NF1 is unknown and needs to be clarified in future studies.

As a third major finding, hyperperfusion in the amygdala of NF1+/– mice was observed at juvenile age only. Perfusion in the amygdala had reduced to normal levels at young adulthood. The normalization of cerebral perfusion in adulthood was observed by follow-up SPECT imaging in the same mice. This longitudinal design is expected to provide more reliable results than cross-sectional experiments in which groups of different animals are examined for each age. The time course of hyperperfusion in the amygdala of NF1+/– mice appears to be in analogy to the time course of neurologic deficits in NF1 patients; most pronounced at juvenile age and dissolving towards adulthood.

CONCLUSION

Juvenile NF1n31 mice show hyperperfusion bilaterally in the amygdala, which normalizes when reaching adulthood. This behavior resembles the typical time course of cognitive deficits in NF1. However, further studies are required to test for causal relationship, which is required to make hyperperfusion in the amygdala a biomarker for cognitive dysfunction in NF1. Further studies are also required to explore hyperperfusion in the amygdala as potential downstream effect of altered synaptic plasticity.

DISCLOSURE/CONFLICT OF INTEREST

The authors declare no conflict of interest.

Footnotes

ACKNOWLEDGMENTS

The authors thank Thomas Brüning for excellent technical assistance with histologic methods.

References

Johnston

. Brain plasticity in paediatric neurology. Eur J Paediatr Neurol 2003; 7: 105–113.

Tucker

Friedman

Friedrich

Wenzel

Funsterer

Mautner

. Longitudinal study of neurofibromatosis 1 associated plexiform neurofibromas. J Med Genet 2009; 46: 81–85.

Friedrich

Kluwe

Funsterer

Mautner

. Malignant peripheral nerve sheath tumors (MPNST) in neurofibromatosis type 1 (NF1): diagnostic findings on magnetic resonance images and mutation analysis of the NF1 gene. Anticancer Res 2005; 25: 1699–1702.

Kolberg

Holand

Agesen

Brekke

Liestol

Hall

Survival meta-analyses for > 1800 malignant peripheral nerve sheath tumor patients with and without neurofibromatosis type 1. Neuro Oncol 2013; 15: 135–147.

Hachon

Iannuzzi

Chaix

. Behavioural and cognitive phenotypes in children with neurofibromatosis type 1 (NF1): the link with the neurobiological level. Brain Dev 2011; 33: 52–61.

Krab

De Goede-Bolder

Aarsen

Moll

De Zeeuw

Elgersma

Motor learning in children with neurofibromatosis type I. Cerebellum 2011; 10: 14–21.

Rosser

Packer

. Neurocognitive dysfunction in children with neurofibromatosis type 1. Curr Neurol Neurosci Rep 2003; 3: 129–136.

Payne

Pickering

Porter

Oates

Walia

Prelog

Longitudinal assessment of cognition and T2-hyperintensities in NF1: an 18-year study. Am J Med Genet A 2014; 164A: 661–665.

Costa

Federov

Kogan

Murphy

Stern

Ohno

Mechanism for the learning deficits in a mouse model of neurofibromatosis type 1. Nature 2002; 415: 526–530.

10.

Jacks

Shih

Schmitt

Bronson

Bernards

Weinberg

. Tumour predisposition in mice heterozygous for a targeted mutation in Nf1. Nat Genet 1994; 7: 353–361.

11.

Cichowski

Shih

Jacks

. Nf1 gene targeting: toward models and mechanisms. Sem Cancer Biol 1996; 7: 291–298.

12.

Gutmann

Giovannini

. Mouse models of neurofibromatosis 1 and 2. Neoplasia 2002; 4: 279–290.

13.

Costa

Silva

. Molecular and cellular mechanisms underlying the cognitive deficits associated with neurofibromatosis 1. J Child Neurol 2002; 17: 622–626; discussion 627–629, 646–651.

14.

Costa

Silva

. Mouse models of neurofibromatosis type I: bridging the GAP. Trends Mol Med 2003; 9: 19–23.

15.

Weeber

Sweatt

. Molecular neurobiology of human cognition. Neuron 2002; 33: 845–848.

16.

Molosh

Johnson

Spence

Arendt

Federici

Bernabe

Social learning and amygdala disruptions in Nf1 mice are rescued by blocking p21-activated kinase. Nat Neurosci 2014; 17: 1583–1590.

17.

Kadekaro

Crane

Sokoloff

. Differential effects of electrical stimulation of sciatic nerve on metabolic activity in spinal cord and dorsal root ganglion in the rat. Proc Natl Acad Sci USA 1985; 82: 6010–6013.

18.

Jack

Jr Albert

Knopman

McKhann

Sperling

Carrillo

Introduction to the recommendations from the National Institute on Aging-Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease. Alzheimers Dement 2011; 7: 257–262.

19.

McKhann

Knopman

Chertkow

Hyman

Jack

Jr Kawas

The diagnosis of dementia due to Alzheimer's disease: recommendations from the National Institute on Aging-Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease. Alzheimers Dement 2011; 7: 263–269.

20.

Balestri

Lucignani

Fois

Magliani

Calistri

Grana

Cerebral glucose metabolism in neurofibromatosis type 1 assessed with [18 F]-2-fluoro-2-deoxy-D-glucose and PET. J Neurol Neurosurg Psychiatry 1994; 57: 1479–1483.

21.

Buchert

Von Borczyskowski

Wilke

Gronowsky

Friedrich

Brenner

Reduced thalamic 18 F-flurodeoxyglucose retention in adults with neurofibromatosis type 1. Nucl Med Commun 2008; 29: 17–26.

22.

Kaplan

Chen

Lawson

Wodrich

Bonstelle

Reiman

. Positron emission tomography in children with neurofibromatosis-1. J Child Neurol 1997; 12: 499–506.

23.

Lange

Apostolova

Lukas

Huang

Hofheinz

Gregor-Mamoudou

Performance evaluation of stationary and semi-stationary acquisition with a non-stationary small animal multi-pinhole SPECT system. Mol Imaging Biol 2014; 16: 311–316.

24.

Herholz

Schopphoff

Schmidt

Mielke

Eschner

Scheidhauer

Direct comparison of spatially normalized PET and SPECT scans in Alzheimer's disease. J Nucl Med 2002; 43: 21–26.

25.

Rascovsky

Hodges

Knopman

Mendez

Kramer

Neuhaus

Sensitivity of revised diagnostic criteria for the behavioural variant of frontotemporal dementia. Brain 2011; 134: 2456–2477.

26.

Apostolova

Wunder

Dirnagl

Michel

Stemmer

Lukas

Brain perfusion SPECT in the mouse: normal pattern according to gender and age. Neuroimage 2012; 63: 1807–1817.

27.

Silva

Frankland

Marowitz

Friedman

Laszlo

Cioffi

A mouse model for the learning and memory deficits associated with neurofibromatosis type I. Nat Genet 1997; 15: 281–284.

28.

Finucane

Murray

Sosabowski

Foster

Mather

. Quantitative accuracy of low-count SPECT imaging in phantom and in vivo mouse studies. Int J Mol Imaging 2011; 2011: 197381.

29.

Hof

Grant

Blackband

Bennett

Slatest

A three-dimensional digital atlas database of the adult C57BL/6 J mouse brain by magnetic resonance microscopy. Neuroscience 2005; 135: 1203–1215.

30.

Smith

Hof

Foerster

Hamilton

Blackband

In vivo 3D digital atlas database of the adult C57BL/6 J mouse brain by magnetic resonance microscopy. Front Neuroanat 2008; 2: 1.

31.

Scheffler

Stenzel

Krohn

Lange

Hofrichter

Schumacher

Determination of spatial and temporal distribution of microglia by 230nm-high-resolution, high-throughput automated analysis reveals different amyloid plaque populations in an APP/PS1 mouse model of Alzheimer's disease. Curr Alzheimer Res 2011; 8: 781–788.

32.

Daginakatte

Gianino

Zhao

Parsadanian

Gutmann

. Increased c-Jun-NH2-kinase signaling in neurofibromatosis-1 heterozygous microglia drives microglia activation and promotes optic glioma proliferation. Cancer Res 2008; 68: 10358–10366.

33.

Pahnke

Frohlich

Krohn

Schumacher

Paarmann

. Impaired mitochondrial energy production and ABC transporter function-A crucial interconnection in dementing proteopathies of the brain. Mech Ageing Dev 2013; 134: 506–515.

34.

Robinson

Kloog

Stein

Assaf

. Motor deficits and neurofibromatosis type 1 (NF1)-associated MRI impairments in a mouse model of NF1. NMR Biomed 2010; 23: 1173–1180.

35.

Hyman

Gill

Shores

Steinberg

North

. T2 hyperintensities in children with neurofibromatosis type 1 and their relationship to cognitive functioning. J Neurol Neurosurg Psychiatry 2007; 78: 1088–1091.

36.

Moore

Slopis

Schomer

Jackson

Levy

. Neuropsychological significance of areas of high signal intensity on brain MRIs of children with neurofibromatosis. Neurology 1996; 46: 1660–1668.

37.

Sestini

. The neural basis of functional neuroimaging signal with positron and single-photon emission tomography. Cell Mol Life Sci 2007; 64: 1778–1784.

38.

Tremblay

Majewska

. A role for microglia in synaptic plasticity? Commun Integr Biol 2011; 4: 220–222.

39.

Welberg

. Synaptic plasticity: a synaptic role for microglia. Nat Rev Neurosci 2014; 15: 69.

40.

Siskova

Tremblay

. Microglia and synapse: interactions in health and neurodegeneration. Neural Plast 2013; 2013: 425845.