Functional subdivisions of the hypothalamus using areal parcellation and their signal changes related to glucose metabolism

NeuroImage 162 (2017) 1–12

Contents lists available at ScienceDirect

NeuroImage journal homepage: www.elsevier.com/locate/neuroimage

Functional subdivisions of the hypothalamus using areal parcellation and their signal changes related to glucose metabolism Takahiro Osada a, Ruriko Suzuki b, Akitoshi Ogawa a, Masaki Tanaka a, Masaaki Hori c, Shigeki Aoki c, d, e, Yoshifumi Tamura b, d, Hirotaka Watada b, d, Ryuzo Kawamori b, d, Seiki Konishi a, d, e, * a

Department of Neurophysiology, Juntendo University School of Medicine, 2-1-1 Hongo, Bunkyo-ku, Tokyo 113-8421, Japan Department of Metabolism and Endocrinology, Juntendo University School of Medicine, 2-1-1 Hongo, Bunkyo-ku, Tokyo 113-8421, Japan c Department of Radiology, Juntendo University School of Medicine, 2-1-1 Hongo, Bunkyo-ku, Tokyo 113-8421, Japan d Sportology Center, Juntendo University School of Medicine, 2-1-1 Hongo, Bunkyo-ku, Tokyo 113-8421, Japan e Research Institute for Diseases of Old Age, Juntendo University School of Medicine, 2-1-1 Hongo, Bunkyo-ku, Tokyo 113-8421, Japan b

A R T I C L E I N F O

A B S T R A C T

Keywords: Boundary mapping Autonomic nervous system OGTT Diabetes mellitus

The hypothalamus consists of numerous nuclei, and is regarded as the highest center for various autonomic functions. Although each hypothalamic nucleus implements a distinct function, it remains difficult to investigate the human hypothalamus at the nucleus level. In the present high-resolution functional MRI study, we utilized areal parcellation to discriminate individual nuclei in the human hypothalamus based on areal profiles of restingstate functional connectivity. The areal parcellation detected ten foci that were expected to represent hypothalamic nuclei, and the locations of the foci were consistent with those of the hypothalamic nuclei identified in previous histological studies. Regions of interest (ROI) analyses revealed contrasting brain activity changes following glucose ingestion: decrease in the ventromedial hypothalamic nucleus and increase in the lateral hypothalamic area in parallel with blood glucose increase. Moreover, decreased brain activity in the arcuate nucleus predicted future elevation of blood insulin during the first 10 min after glucose ingestion. These results suggest that the hypothalamic nuclei can putatively be determined using areal parcellation, and that the ROI analysis of the human hypothalamic nuclei is useful for future scientific and clinical investigations into the autonomic functions.

1. Introduction The hypothalamus has been recognized as the highest-level center for autonomic functions (Kandel et al., 2013). Historically, the ventromedial nucleus of the hypothalamus (VMH) and the lateral hypothalamic area (LHA) act as a “satiety center” and a “feeding center”, respectively (Hetherington and Ranson, 1940; Anand and Brobeck, 1951; Delgado and Anand, 1953; Miller, 1960). Recent investigations have revealed the arcuate nucleus (ARC) as a key structure regulating feeding behavior and energy homeostasis (Kubota et al., 2007; Williams and Elmquist, 2012; Sternson, 2013; Abraham et al., 2014; Krashes et al., 2016). The hypothalamus is also thought to modulate the release of insulin, a pancreatic hormone secreted in response to elevated blood glucose level (Ahr
en, 2000; Teff, 2011; Chan and Sherwin, 2012). Even though the blood glucose level has not yet elevated, insulin is known to be secreted soon

after food ingestion during the first 10 min (Berthoud et al., 1981; Bellisle et al., 1983; Teff et al., 1991; Ahr
en, 2000). It has been reported that lesions to the hypothalamus enhance the insulin release (Louis-Sylvestre, 1976, 1978). Most of the studies have been performed in rodents, and nucleus-level analysis of human hypothalamus remains relatively underinvestigated. Functional MRI (fMRI) has recently been utilized to investigate involvement of the human hypothalamus in energy homeostasis (Matsuda et al., 1999; Liu et al., 2000; Smeets et al., 2005; Batterham et al., 2007; Vidarsdottir et al., 2007; Teeuwisse et al., 2012; Page et al., 2013; Heni et al., 2014, 2015; Little et al., 2014; Jastreboff et al., 2016). It has been clarified that specific regions of the hypothalamus are associated with glucose metabolism. To further clarify the functional roles of the hypothalamus, it is important to segregate brain activity from the individual hypothalamic nuclei involved in distinct feeding and metabolic

* Corresponding author. Department of Neurophysiology, Juntendo University School of Medicine, 2-1-1 Hongo, Bunkyo-ku, Tokyo 113-8421, Japan. E-mail address: [email protected] (S. Konishi). http://dx.doi.org/10.1016/j.neuroimage.2017.08.056 Received 16 May 2017; Received in revised form 20 July 2017; Accepted 21 August 2017 Available online 24 August 2017 1053-8119/© 2017 Elsevier Inc. All rights reserved.

T. Osada et al.

NeuroImage 162 (2017) 1–12

2. Materials and methods

functions. Higher spatial resolution of the fMRI images and areal parcellation methods that determine hypothalamic nuclei will be useful in segregating brain activity from the hypothalamic nuclei. Technological advancements in MRI have allowed us to apply highresolution fMRI (Feinberg et al., 2010). Moreover, developments in resting-state functional connectivity (Fox and Raichle, 2007) and diffusion tractography (Le Bihan and Johansen-Berg, 2012) analyses have provided useful tools for the parcellation of the brain into various functional areas (Eickhoff et al., 2015). Areal parcellation using boundary mapping methods identify sharp changes in resting-state functional connectivity profiles as areal boundaries (Cohen et al., 2008; Biswal et al., 2010; Hirose et al., 2012, 2013, 2016; Wig et al., 2014a, 2014b; Laumann et al., 2015; Poldrack et al., 2015; Glasser et al., 2016). It has been demonstrated that the resting-state fMRI is more accurate in parcellation than other structural and functional features such as task-based fMRI, myelin maps or cortical thickness (Glasser et al., 2016). Moreover, resting-state functional connectivity has revealed region-specific patterns of connectivity between the hypothalamus and the cerebral cortex (Kullmann et al., 2014; Hirose et al., 2016). In the present study, high-resolution fMRI was applied to the hypothalamus to investigate glucose metabolism based on individual hypothalamic nuclei using the areal parcellation (Fig. 1A, Supplementary Fig. 1). During fMRI scans, human subjects drank glucose solution or water, and the plasma glucose, insulin and free fatty acid were sampled (Fig. 1B).

2.1. Subjects Twelve right-handed subjects [six males and six females, age: 26.6 ± 8.3 years (mean ± s.d.) ranging from 20 to 39] participated in the experiments. They were confirmed to be healthy by annual medical checkups. Written informed consent was obtained from all the subjects according to the Declaration of Helsinki. The body mass index (BMI) of the subjects was 21.6 ± 1.3 kg/m2 (mean ± s.d.), ranging from 19.9 to 23.7. None of the subjects had been diagnosed with diabetes mellitus. The experimental procedures were approved by the Institutional Review Board of Juntendo University School of Medicine. 2.2. Experimental schedule Subjects underwent two conditions: Glucose and Water conditions (Fig. 1B). The order of the two conditions was counterbalanced across subjects. Each condition consisted of two consecutive days of fMRI scanning (Day 1 and Day 2). Scanning on Day 1 was conducted in the evening, and scanning on Day 2 was conducted in the next morning following an overnight fast. During the Day 1 scanning, the subjects underwent task-free fMRI scans while resting quietly, and during the Day 2 scanning, the subjects drank glucose solution or water (Glucose or Water condition), while plasma glucose, insulin and free fatty acid were

Fig. 1. Overview of the experiments. (A) The human hypothalamus was parcellated using high-resolution fMRI (1.25 mm voxel), and the associations between brain activity in each nucleus and blood levels of glucose, insulin, and free fatty acid were analyzed. (B) Subjects underwent two conditions: Glucose and Water conditions. Each condition consisted of two consecutive days of fMRI sessions: Day 1 and Day 2. During Day 1 scanning, the subjects underwent task-free fMRI scans, and during Day 2 scanning, the subjects drank glucose solution or water (Glucose or Water condition). Two runs of functional scans were acquired on both Day 1 and Day 2. On Day 2, blood samples were obtained inside the scanner at T ¼ 10, 5, 1, 10, 20, and 30 min (the onset of drinking was defined as T ¼ 0 min). 2

T. Osada et al.

NeuroImage 162 (2017) 1–12

hypothalamus or the data analyses employed in the present study. Importantly, the phase encoding direction in the present study was posterior to anterior, and the third ventricle was intact (Supplementary Fig. 2A). In the Human Connectome Project (HCP), where a large amount of imaging datasets is provided in 2-mm voxel resolution (https://www. humanconnectome.org/), the phase encoding direction was right to left or left to right. The midline in ventral slices was distorted toward one either direction in the HCP datasets, and thus the third ventricle was distorted (Supplementary Fig. 2B) and was still distorted and additionally blurred after topup correction (Supplementary Fig. 2C). Therefore, it is critical to employ the posterior-to-anterior direction for the analyses of the nuclei in the medial hypothalamus applied in the present study. Moreover, as shown in Supplementary Fig. 3, the most ventral part of the hypothalamus was well preserved under the scanning parameters of the present study.

sampled. The image data from Day 1 were used for parcellation of the hypothalamus, and the image data from Day 2 were used in various analyses of MRI signals related to drinking of the glucose solution or water. This study was conducted in a within-subject design, and all the subjects participated in both glucose and water experiment. Thus, the subjects underwent the experiment four times. Two runs of functional scans were acquired on both Day 1 and Day 2. Each run lasted for 28 min, with an inter-scan interval of 2 min. On Day 2, 14 min after the start of the first run, the subjects were instructed to drink either glucose solution or water. The onset of drinking was defined as T ¼ 0 min. Therefore, the fMRI scanning took place from T ¼ 14 min to T ¼ 14 min in the first run, and from T ¼ 16 min to T ¼ 44 min in the second run. Blood samples were obtained inside the scanner at T ¼ 10, 5, 1, 10, 20, and 30 min. The overall experimental schedule, including the scanning, ingestion and blood sampling, was controlled using a stop watch.

2.6. Data analysis: Day 1 overview 2.3. Ingestion procedures Resting-state fMRI data were used to parcellate the hypothalamus. The data consisted of two daily sessions of two runs of 28 min (Day 1). In total, we analyzed 112-min fMRI data per subject. An overview of the parcellation analysis is shown in Fig. 2. The analyses consisted of two steps. First, the parcellation method yielded probabilistic center maps in individual subjects. Second, following spatial transformation to a common group structural image, the individual center maps were group averaged, and the foci that are expected to represent hypothalamic nuclei were determined. More details are described below.

Inside the scanner, the subjects ingested 225 ml of 75 g glucose solution (TRELAN-G75, AY Pharmaceuticals, Tokyo, Japan) or 225 ml of water. A peroral rubber tube of 1 m in length was used for drinking, connecting the subject's mouth with the solution bottle placed outside the scanner. At T ¼ 0.5 min, the subjects were gently tapped once on their feet, as a sign of preparation for drinking. At T ¼ 0 min, the subjects were tapped three times on their feet, as an indication to start drinking. Following ingestion, which took approximately 2 min, the subjects were again tapped three times to indicate cessation of drinking.

2.7. Preprocessing for resting-state functional connectivity analyses 2.4. Blood sampling Functional images were preprocessed for the resting-state functional connectivity, as previously described (Fox et al., 2005; Fair et al., 2007). Images were corrected for slice timing and realigned using SPM8 (http:// www.fil.ion.ucl.ac.uk/spm/). Temporal filters (0.009 Hz < f < 0.08 Hz) were applied to the functional images using FSL (Smith et al., 2004). A general linear model (Worsley and Friston, 1995) was used to regress out nuisance signals that correlated with head motion, whole-brain global signal, averaged ventricular signal and averaged white matter signal.

An intravenous cannula was placed in the forearm for blood sampling prior to MR scanning. Blood samples were obtained through the catheter at T ¼ 10, 5, 1, 10, 20, and 30 min inside the scanner in order to measure levels of glucose, insulin, and free fatty acid at each time point. Some of the blood samples for the insulin measurement were damaged by hemolysis. However, data points in Water condition were recovered by interpolation of the adjacent time points, based on stable plasma insulin levels (see Results). Particularly, at T ¼ 10 min, three samples in Glucose condition were damaged by hemolysis, but one sample damaged by hemolysis in Water condition was recovered due to the flat time course. Therefore, there were nine samples for insulin at T ¼ 10 min.

2.8. Areal parcellation analyses The probabilistic center maps were generated by the areal parcellation method based on boundary mapping (Cohen et al., 2008; Biswal et al., 2010; Hirose et al., 2012, 2013, 2016; Wig et al., 2014a, 2014b; Laumann et al., 2015; Poldrack et al., 2015; Glasser et al., 2016). The method was applied to the 3D structure in the present study. For each subject, the hypothalamus was manually delineated based on anatomical landmarks as described in previous MRI studies of the hypothalamus (Makris et al., 2013; Schindler et al., 2013; Sch€ onknecht et al., 2013; Hirose et al., 2016). Each voxel in the hypothalamus of each subject was used as a seed to calculate its correlation with target voxels in the gray matter of the cerebral cortex. A part of the cerebral cortex that suffered from overlap due to aliasing artifact, as described earlier, was removed from the target voxels. A voxel-wise correlation map of the target region was generated for each seed voxel, and the correlation coefficient was then converted to Fisher's z (Fox et al., 2005; Fair et al., 2007). After calculation of the correlation maps, the similarity of the spatial correlation pattern was then quantified using an index of eta2, and similarity maps were generated. After spatial smoothing (FWHM ¼ 2.0 mm), spatial gradients of the similarity maps were computed for each seed voxel. Note that spatial smoothing was applied separately to the left and right sides of the hypothalamus in this study, to ensure that MRI signals from one hemisphere did not contaminate those from the other. The local minima in the gradient map were detected, and binary local minima

2.5. MRI procedures All MRI data were acquired using a 3-T MRI scanner and a 32-channel RF head coil (Siemens Skyra, Erlangen, Germany). T1-weighted structural images were obtained for anatomical reference (resolution ¼ 0.8  0.8  0.8 mm3). Functional images were obtained using multi-band gradient-echo echo-planar sequences (Feinberg et al., 2010) (TR ¼ 5 s, TE ¼ 41.6 ms, flip angle ¼ 73 deg, FOV ¼ 160  160 mm2, matrix size ¼ 128  128, 120 contiguous slices, voxel size ¼ 1.25  1.25  1.25 mm3, multi-band factor ¼ 4). For parcellation of the hypothalamus, we acquired high-resolution functional images with a resolution of 1.25 mm, rather than standard resolution of 2 mm. To attain a maximal signal to noise ratio, a small FOV (160  160 mm2) was set. This small FOV did not always cover the whole brain along the anterior-posterior axis. However, aliasing artifact in the frontal lobe (from the occipital lobe) was minimal, and the alias overlap artifact observed in the frontal lobe was 1.3–4.9% of the whole cerebral cortex. It has also been demonstrated that the parcellation procedures do not require the entire cerebral cortex for appropriate detection of the connectivity transitions (Hirose et al., 2012, 2013). Therefore, it is unlikely that the small FOV influenced signal acquisition of the

3

T. Osada et al.

NeuroImage 162 (2017) 1–12

Fig. 2. An overview of the areal parcellation applied to the hypothalamus. Functional images were processed to determine the centers of nuclei in the hypothalamus of each subject. After obtaining single-level center maps of the hypothalamus, a group-level center map of the hypothalamus was generated.

2.9. Comparison of the detected foci with those from a previous study of histology

maps were generated. The binary maps were then averaged across seeds to generate a probabilistic center map, and the local maxima in the probabilistic center map were detected. Next, a group structural image was generated by averaging the structural images of all the subjects after spatial normalization to the MNI template. We obtained a group-level center map of the hypothalamus that matched to the group structural image with regard to two visually identifiable landmarks, the anterior commissure (AC) and the mammillary body (MB). More specifically, the center map for each subject, which was calculated from functional images, was linearly transformed to the group structural image, interpolated to a resolution of 1  1  1 mm3, based on visual identification of the AC and the MB on the original functional images. This transformation allowed us to compare the detected foci in the present study with those reported in Baroncini et al. (2012), in which hypothalamic nuclei in the structural MRI images were validated using histological images. The transformed center maps were then averaged across subjects. After spatial smoothing (FWHM ¼ 1.5 mm), the local maxima in the averaged center map were defined as foci that are expected to correspond to hypothalamic nuclei. To select symmetrical pairs of bilateral centers, the X coordinates of the centers in one hemisphere were flipped, and the distances between the left and right centers were calculated. Bilateral centers apart by less than 2.5 mm were regarded as symmetrical, which rejected bilateral centers apart by 2 original voxels.

The locations of the nuclei from the present study and Baroncini et al. (2012) were compared. We investigated whether the distance between the nuclei from the two studies was shorter when the nuclei were closer to the line connecting the AC and MB than when they were farther from the AC-MB line. A significant correlation (r ¼ 0.70, p < 0.05) (Supplementary Fig. 4) suggests that, although the locations in the two studies were closely matched, the misregistration between the functional and structural MRI images requires the parcellation analysis to identify the hypothalamic nuclei in functional images.

2.10. Data analysis: Day 2 Data from Day 2 were analyzed to evaluate fMRI responses following glucose/water ingestion. The time courses of the functional images were preprocessed using SPM8. Slice timing and head movement were corrected, and the images were spatially transformed to the group structural image, interpolated to a resolution of 1  1  1 mm3. Spatial smoothing was applied to the left and right sides of the hypothalamus separately (FWHM ¼ 1.5 mm). To investigate the signal time courses in each focus, the regions of interest (ROIs) were defined as those voxels within a 4

T. Osada et al.

NeuroImage 162 (2017) 1–12

following glucose ingestion, despite no significant increases in plasma glucose levels. No significant difference in the baseline levels of free fatty acid was

1.5 mm radius of the foci determined using the data from Day 1. When the ROI invaded the other hemisphere (the ROIs in the medial part of the hypothalamus), the voxels in the other hemisphere were excluded from the ROI to ensure statistical independence of the bilateral ROI pairs. The signals from the voxels in the hypothalamus were compared to the mean 10 min baseline before ingestion, yielding a percentage signal change from the mean baseline. A general linear model was used to extract specific effects, such as the subsequent effect, from the signal time courses.

observed between Glucose and Water conditions [t (11) ¼ 0.3, p > 0.05]. As shown in Fig. 3C, the plasma free fatty acid levels showed significant decrease at 30 min after glucose ingestion, relative to Water condition [t (11) ¼ 2.5, p < 0.05]. 3.2. Parcellation of the hypothalamus

2.11. Evaluation of head movement

Image data acquired on Day 1 were used to determine the hypothalamic nuclei putatively by generating a group-level center map (Fig. 2). Fig. 4A shows the probabilistic map used to detect local maxima as foci that are expected to correspond to hypothalamic nuclei. Ten bilateral pairs of foci in the hypothalamus were detected that were located symmetrically across hemispheres. To test the reliability of the parcellation results, we divided the Day 1 data into two halves for each individual, and the mean correlation coefficient was 0.43 ± 0.13 (mean ± s.d.), and

We evaluated the amount of head motion by employing frame-wise displacement (FD) (Power et al., 2012). FD is instantaneous head motion that can be calculated as a locational difference between two successive images, and is an important measure for quality control of resting-state data. There were no Day 1 sessions where large FD (>0.5 mm) occurred in more than 10% of the acquired functional images, with the mean FD across subjects being 0.14 ± 0.04 mm (mean ± s.d.). Although FD is an important measure for quality control of the resting-state analyses of Day 1, we also calculated FD for the data of glucose/water ingestion in Day 2. There were no Day 2 sessions where large FD (>0.5 mm) occurred in more than 10% of acquired images. The mean FDs in Glucose and Water conditions were 0.16 ± 0.04 and 0.17 ± 0.04 mm (mean ± s.d.), respectively. The mean FDs from T ¼ 0–2 min (i.e., ingestion period) were 0.41 ± 0.11 and 0.45 ± 0.11 mm (mean ± s.d.) in Glucose and Water conditions, respectively. The FDs seem to be within the acceptable range of head movement, and we did not exclude the functional scans right after swallowing.

significantly greater than zero [t (11) ¼ 9.6, p < 0.001].

2.12. Statistics Statistical test on the maps generated by SPM8 was performed without correction for multiple comparisons. Time courses from regions of interest were extracted and were subject to statistics using Excel (Microsoft) and Matlab (Mathworks), with correction for multiple comparisons by the number of regions detected as bilateral lateral and medial foci (n ¼ 20). To test the lateral/medial dissociation of signal changes after drinking glucose or water, we employed a two-way ANOVA with foci (lateral and medial) and conditions (Glucose and Water) as main effects. We also tested whether the MRI signals (Glucose minus Water) correlated with the subsequent plasma insulin levels (Glucose minus Water) using both statistical maps from SPM8 and extracted time courses from regions of interest. 3. Results 3.1. Metabolic and hormonal responses after ingestion The plasma glucose, insulin and free fatty acid levels were averaged across the three baseline time points (T ¼ 10, 5 and 1 min). No significant difference in the baseline plasma glucose levels was observed between Glucose and Water conditions [t (11) ¼ 0.2, p > 0.05]. As shown in Fig. 3A, the plasma glucose levels showed significant increase at 20 and 30 min after glucose ingestion, relative to Water condition [t (11) ¼ 3.6, p < 0.005 at T ¼ 20 min]. The plasma glucose levels at 10 min after ingestion did not show significant increase [t (11) ¼ 1.7, p > 0.05]. No significant difference in the baseline levels of insulin was observed between Glucose and Water conditions [t (11) ¼ 1.2, p > 0.05]. As shown in Fig. 3B, the plasma insulin levels showed significant increase at 10, 20 and 30 min after glucose ingestion, relative to Water condition [t

Fig. 3. Plasma glucose, insulin and free fatty acid levels following ingestion of glucose solution or water. (A) Time courses of plasma glucose levels after ingestion of glucose solution (blue) and water (red). Error bars indicate s.e.m. **p < 0.01, ***p < 0.001. (B) Time courses of plasma insulin levels. The format is the same as (A). (C) Time courses of plasma free fatty acid levels. The format is the same as (A). *p < 0.05.

(8) ¼ 4.7, p < 0.01 at T ¼ 10 min; see Blood sampling in Materials and Methods]. These results are in accordance with those of previous studies, which demonstrated significant increases in plasma insulin 10 min 5

T. Osada et al.

NeuroImage 162 (2017) 1–12

detecting centers of the two subdivisions, that were separated by only 2–3 mm (Table 1).

To test the validity of the detected foci, the coordinates of the foci were compared with those described by Baroncini et al. (2012), in which hypothalamic nuclei in structural MRI images were validated using histological images. Fig. 4B shows the locations of the nuclei in the medial region of the hypothalamus (averaged across the left and right hemispheres) calculated from the functional images of the present study and the structural images of Baroncini et al. (2012). These nuclei were closely matched with one another, with an average distance of 1.1 mm. However, the slight mismatch that may be derived mainly from the misregistration between the functional and structural MRI images used in the two studies suggests the need of the parcellation analysis to identify the hypothalamic nuclei in functional images (see also Supplementary Fig. 4). The full coordinates, putative nucleus names of the bilateral foci, and the distance between the foci identified by the two studies are listed in Table 1. We also delineated the spatial extent of the ten subdivisions using the watershed algorithm (Supplementary Fig. 5). Although most of the reported subdivisions were identified bilaterally, the two foci (M-F1 and F4) were merged into one cluster, indicating the relatively greater difficulty in drawing boundaries between two subdivisions, rather than

3.3. MRI signals following ingestion The symmetrical foci in the hypothalamus were used as seeds for regions of interest (ROIs), and the MRI signal time courses were extracted from the image data acquired on Day 2. Fig. 5A and B shows the time courses from the ROIs at ten detected foci in the medial (M-F1 to F8) and the lateral hypothalamus (L-F1 and F2). In some foci, a sharp signal decrease was observed immediately after the onset of drinking. To test the possibility that head motion during drinking (T ¼ 0–2 min) caused the signal decrease, MRI signals from the whole hypothalamus at T ¼ 0–2 min were inspected (Supplementary Fig. 6). Head motion artifacts are typically seen along the boundaries between the hypothalamus and the regions outside the hypothalamus; however, the spatial signal patterns were not characteristic of such head motion. The VMH is historically regarded as the satiety center, whereas the LHA is regarded as the feeding center. We examined whether the two foci

Fig. 4. Parcellation of the hypothalamus. (A) A group-averaged probabilistic center map in the hypothalamus. Local maxima were defined as foci that are expected to correspond to hypothalamic nuclei. The color scale reflects the probability of existence of hypothalamic centers. L, left; R, right. (B) Locations of the nuclei, averaged across the left and right hemispheres, in the medial region of the hypothalamus were calculated from the functional images of the present study and the structural images of Baroncini et al. (2012). A, anterior; P, posterior; D, dorsal; V, ventral. L-F#, lateral foci #; M-F#, medial foci #. 6

T. Osada et al.

NeuroImage 162 (2017) 1–12

Table 1 Hypothalamic foci detected by parcellation analyses. Left

M-F1 M-F2 M-F3 M-F4 M-F5 M-F6 M-F7 M-F8 L-F1 L-F2

Right

X

Y

Z

X

Y

Z

1 1 1 1 2 1 1 1 4 4

4 1 8 4 2 5 1 8 5 0

18 18 17 15 14 12 11 10 13 10

1 1 2 1 1 1 1 1 4 4

5 1 8 5 2 3 0 7 4 0

18 16 15 16 14 12 13 10 13 12

Putative nucleus names

Distance (Y-Z plane)

ARC AH MB VMH MPO DMH PVH PH LHAt LHAa

0.5 1.6 0.6 0.9 1.6 1.0 0.9 1.6 1.6 1.4

The coordinates of the detected hypothalamic nuclei (left and right), putative nucleus names of the bilateral foci, and distance between the foci in the present study and Baroncini et al. (2012) in Y-Z plane are listed. ARC, arcuate nucleus of the hypothalamus; AH, anterior nucleus of the hypothalamus; MB, mammillary body; VMH, ventromedial nucleus of the hypothalamus; MPO, medial preoptic nucleus; DMH, dorsomedial nucleus of the hypothalamus; PVH, paraventricular nucleus of the hypothalamus; PH, posterior hypothalamic nucleus; LHAt, lateral hypothalamic area (tuberal part); LHAa, lateral hypothalamic area (anterior part). L-F#, lateral foci #; M-F#, medial foci #.

M-F1, indicating that the greater the decrease of the MRI signals, the greater the subsequent plasma insulin increase.

exhibited dissociable time courses reflective of the two contrasting functions. Fig. 6A shows the statistical maps depicting the difference between Glucose and Water conditions. The time courses at the M-F4 (VMH, left and right averaged) revealed a signal decrease at T ¼ 10–40 min in Glucose condition, relative to Water condition, consistent with the time courses for the plasma glucose data (Fig. 6B, left). Conversely, the foci at the L-F1 and F2 (LHA, left and right averaged) exhibited signal increases at T ¼ 10–40 min (Fig. 6B, right). A twoway ANOVA with foci (the M-F4 and the L-F1 and F2) and conditions (Glucose and Water) as main effects revealed a significant interaction

4. Discussion In the present study, areal parcellation was applied to the hypothalamus, and ten bilateral pairs of foci were dissociated that are expected to represent hypothalamic nuclei, including the VMH, LHA and ARC. The locations of the detected foci were reasonably well matched with the histology-based hypothalamic nuclei reported in a previous study (Baroncini et al., 2012). The ROI analysis revealed significant functional dissociation between the VMH and the LHA, with decrease and increase of brain activity 10–40 min after glucose ingestion observed in the VMH and the LHA, respectively. Moreover, brain activity in the ARC during the first 10 min after glucose ingestion significantly predicted the increase in plasma insulin levels observed 10 min after glucose ingestion. These results suggest that ROI analyses based on areal parcellation results can be used to reveal the diverse metabolic and endocrinological functions of the hypothalamic nuclei in humans. Resting-state functional connectivity and diffusion tractography analyses have parcellated the brain into various functional areas (Behrens et al., 2003; Cohen et al., 2008; Draganski et al., 2008; Biswal et al., 2010; Mars et al., 2011; Yeo et al., 2011; Hirose et al., 2012, 2013, 2016; Zhang and Li, 2012; Zhang et al., 2012; Bzdok et al., 2013, 2015; Onoda and Yamaguchi, 2013; Shen et al., 2013; Long et al., 2014; Wig et al., 2014a, 2014b; Eickhoff et al., 2015, 2016; Finn et al., 2015; Hutchison et al., 2015; Laumann et al., 2015; Poldrack et al., 2015; Wang et al., 2015; Glasser et al., 2016; Jackson et al., 2016; Reid et al., 2016). It has been demonstrated that the parcellation analysis detects areal boundaries in the cerebral cortex that are closely matched to those identified using histological methods (Glasser et al., 2016). Indeed, the foci detected in the present study were largely matched to the results of a previous study (Baroncini et al., 2012), in which structural images were combined with histological analyses (Fig. 4B). It is generally difficult to compare two images from different studies, mainly due to differences in imaging modalities (structural vs. functional), scanners used and subject groups. The misregistration between structural and functional images in the present study was minimized by aligning the AC and the MB of the two images, and the foci near the line connecting the AC and MB were particularly well matched (Supplementary Fig. 4). The relatively larger misregistration in the foci far from the AC-MB line that was left uncorrected, however, indicates the need to parcellate the hypothalamus using the same imaging modality (i.e., functional), the same scanner and the same group of subjects. The functional dissociation in the VMH and the LHA also demonstrates the potential usefulness of the ROI analysis of the hypothalamus based on parcellation results. The signal changes reflective of blood glucose levels were not necessarily spatially selective in the present

effect [F (1, 11) ¼ 20.1, p < 0.001] (Fig. 6C), demonstrating the functional dissociation between the VMH and the LHA. 3.4. MRI signals in the hypothalamus that predict plasma insulin levels Blood data showed that, at T ¼ 10 min, the increase in plasma insulin levels preceded the increase in plasma glucose levels (Fig. 3A and B). We examined the effect of the hypothalamic brain activity on the subsequent plasma insulin levels, similar to the “subsequent memory effect” (Wagner et al., 1998). More specifically, we investigated whether the MRI signals (Glucose minus Water) at T ¼ 0–10 min predicted the subsequent plasma insulin levels (Glucose minus Water) at T ¼ 10 min. To account for daily variations, the baseline signals were subtracted from signals after ingestion. As shown in Fig. 7A, a voxel-wise analysis revealed significant subsequent effects (i.e., the greater the decrease of the MRI signals, the greater the subsequent plasma insulin increase) near the right ARC at coordinates (1, 5, 18) [t (7) ¼ 5.5, p < 0.001; see also Blood sampling in Materials and Methods] and near the left ARC at (2, 5, 17) [t (7) ¼ 6.4, p < 0.001]. A ROI-based analysis also revealed significant subsequent effects in the right M-F1 (ARC) [t (7) ¼ 5.0, p < 0.005] and the left M-F1 (ARC) [t (7) ¼ 2.7, p < 0.05] (Fig. 7B). After correction for multiple comparisons for the number of ROIs (n ¼ 20), the subsequent effect was still significant in the right M-F1 (p ¼ 0.03). We further investigated whether the MRI signals at T ¼ 10–20 min accounted for the subsequent plasma insulin levels at T ¼ 20 min, and whether those at T ¼ 20–30 min accounted for the subsequent plasma insulin levels at T ¼ 30 min. No such effects were observed for the later MRI signals at M-F1 (p > 0.05) (Fig. 7C), suggesting that the subsequent insulin effect was restricted to the early phase. To more precisely investigate the timing of the subsequent insulin effect, the MRI signals at T ¼ 0–10 min were divided into two halves (T ¼ 0–5 and 5–10 min). The subsequent effect at the right M-F1 was significant at both T ¼ 0–5 min [t (7) ¼ 4.3, p < 0.005] and T ¼ 5–10 min [t (7) ¼ 4.7, p < 0.005], indicating a sustained effect during the cephalic phase, over the first 10 min (Fig. 7D). Fig. 7E and F depict scatter plots of MRI signals at T ¼ 0–10 min and plasma insulin levels at T ¼ 10 min in the left and right 7

T. Osada et al.

NeuroImage 162 (2017) 1–12

Fig. 5. Signal time courses in the hypothalamus following ingestion of glucose solution. (A) MR signal change maps in Glucose and Water conditions at T ¼ 0–10 min, 10–20 min, 20–30 min and 30–40 min. The color scales indicate the level of signal change. (B) Time courses of MRI signal changes in Glucose (blue) and Water (red) conditions at each focus in the hypothalamus (mean ± s.e.m.). L-F#, lateral foci #; M-F#, medial foci #. 8

T. Osada et al.

NeuroImage 162 (2017) 1–12

Fig. 6. Signal time courses in M-F4/L-F1 and F2 following ingestion of glucose solution. (A) Statistical maps of signal changes (Glucose minus Water) at T ¼ 0–10 min, 10–20 min, 20–30 min and 30–40 min. The color scales indicate the significance level. (B) Time courses of MRI signal changes in Glucose and Water conditions in M-F4/L-F1 and F2, left and right averaged (mean ± s.e.m.). (C) Functional double dissociation in M-F4/L-F1 and F2. Error bars indicate s.e.m. ***p < 0.001, two-way ANOVA. L-F#, lateral foci #; M-F#, medial foci #.

9

T. Osada et al.

NeuroImage 162 (2017) 1–12

Fig. 7. Signals in the ARC predict subsequent plasma insulin increase. (A) A statistical map of the subsequent insulin effect (MRI signals at T ¼ 0–10 min vs. subsequent insulin levels at T ¼ 10 min). The format is similar to Fig. 6A. Significant effects were observed near the bilateral ARC. (B) Z-values of the subsequent insulin effect are shown for each of the hypothalamic foci in Table 1. ***p < 0.005, *p < 0.05. L-F#, lateral foci #; M-F#, medial foci #. (C) Z-values of the subsequent insulin effect (MRI signals at T ¼ 0–10, 10–20, and 20–30 min vs. subsequent plasma insulin levels at T ¼ 10, 20 and 30 min) are shown for the left M-F1 (green) and right M-F1 (yellow). (D) Z-values of the subsequent insulin effect (MRI signals in MF1 at T ¼ 0–5 and 5–10 min vs. subsequent plasma insulin levels at T ¼ 10 min) are shown. (E) Scatter plots of MRI signals at T ¼ 0–10 min and plasma insulin levels at T ¼ 10 min in the left M-F1. One dot represents data from one subject. (F) Scatter plots of MRI signals at T ¼ 0–10 min and the plasma insulin levels at T ¼ 10 min in the right M-F1.

10

T. Osada et al.

NeuroImage 162 (2017) 1–12

study; this is consistent with previous literature reporting a widespread distribution of hypothalamic neurons expressing glucose receptors (Burdakov et al., 2005). Nonetheless, the ROI-based analyses in the VMH and LHA demonstrated contrasting time courses following glucose ingestion. The subsequent insulin effect, on the other hand, was more selective for the ARC. The results are inconsistent with those of previous rodent studies reporting that lesions to the VMH enhance the insulin release (Louis-Sylvestre, 1976, 1978). The discrepancy may be explained by difference in species or by insufficient spatial selectivity of lesions to one of two adjacent nuclei, particularly in animals with small brain. Obviously, the present study did not dissociate all of the hypothalamic nuclei, particularly those located closely to one another. Due to limitations in the voxel size of functional images and the statistical power of the data set, the present study only dissociated nuclei that were apart by approximately 2 mm. However, the present study was able to cover major nuclei in humans, including the VMH, LHA and ARC. Moreover, the present high-resolution scanning was conducted in a clinical 3-T MRI system widely available in many hospitals, which indicates that the ROI analysis based on parcellation results can be utilized for future investigation of human patients. In particular, the dissociation of hypothalamic nuclei may be useful for the localization of drug targets and for uncovering the mechanisms underlying the pathophysiology of various metabolic and endocrine disorders.

Bzdok, D., Langner, R., Schilbach, L., Jakobs, O., Roski, C., Caspers, S., Laird, A.R., Fox, P.T., Zilles, K., Eickhoff, S.B., 2013. Characterization of the temporo-parietal junction by combining data-driven parcellation, complementary connectivity analyses, and functional decoding. Neuroimage 81, 381–392. Chan, O., Sherwin, R.S., 2012. Hypothalamic regulation of glucose-stimulated insulin secretion. Diabetes 61, 564–565. Cohen, A.L., Fair, D.A., Dosenbach, N.U., Miezin, F.M., Dierker, D., Van Essen, D.C., Schlaggar, B.L., Petersen, S.E., 2008. Defining functional areas in individual human brains using resting functional connectivity MRI. Neuroimage 41, 45–57. Delgado, J.M.R., Anand, B.K., 1953. Increase of food intake induced by electrical stimulation of the lateral hypothalamus. Am. J. Physiol. 172, 162–168. Draganski, B., Kherif, F., Kl€ oppel, S., Cook, P.A., Alexander, D.C., Parker, G.J., Deichmann, R., Ashburner, J., Frackowiak, R.S., 2008. Evidence for segregated and integrative connectivity patterns in the human basal ganglia. J. Neurosci. 28, 7143–7152. Eickhoff, S.B., Laird, A.R., Fox, P.T., Bzdok, D., Hensel, L., 2016. Functional segregation of the human dorsomedial prefrontal cortex. Cereb. Cortex 26, 304–321. Eickhoff, S.B., Thirion, B., Varoquaux, G., Bzdok, D., 2015. Connectivity-based parcellation: critique and implications. Hum. Brain Mapp. 36, 4771–4792. Fair, D.A., Dosenbach, N.U.F., Church, J.A., Cohen, A.L., Brahmbhatt, S., Miezin, F.M., Barch, D.M., Raichle, M.E., Petersen, S.E., Schlaggar, B.L., 2007. Development of distinct control networks through segregation and integration. Proc. Natl. Acad. Sci. U.S.A. 104, 13507–13512. Feinberg, D.A., Moeller, S., Smith, S.M., Auerbach, E., Ramanna, S., Gunther, M., Glasser, M.F., Miller, K.L., Ugurbil, K., Yacoub, E., 2010. Multiplexed echo planar imaging for sub-second whole brain FMRI and fast diffusion imaging. PLoS One 5, e15710. Finn, E.S., Shen, X., Scheinost, D., Rosenberg, M.D., Huang, J., Chun, M.M., Papademetris, X., Constable, R.T., 2015. Functional connectome fingerprinting: identifying individuals using patterns of brain connectivity. Nat. Neurosci. 18, 1664–1674. Fox, M.D., Raichle, M.E., 2007. Spontaneous fluctuations in brain activity observed with functional magnetic resonance imaging. Nat. Rev. Neurosci. 8, 700–711. Fox, M.D., Snyder, A.Z., Vincent, J.L., Corbetta, M., Van Essen, D.C., Raichle, M.E., 2005. The human brain is intrinsically organized into dynamic, anticorrelated functional networks. Proc. Natl. Acad. Sci. U.S.A. 102, 9673–9678. Glasser, M.F., Coalson, T.S., Robinson, E.C., Hacker, C.D., Harwell, J., Yacoub, E., Ugurbil, K., Andersson, J., Beckmann, C.F., Jenkinson, M., Smith, S.M., Van Essen, D.C., 2016. A multi-modal parcellation of human cerebral cortex. Nature 536, 171–178. Heni, M., Kullmann, S., Ketterer, C., Guthoff, M., Bayer, M., Staiger, H., Machicao, F., H€aring, H.U., Preissl, H., Veit, R., Fritsche, A., 2014. Differential effect of glucose ingestion on the neural processing of food stimuli in lean and overweight adults. Hum. Brain Mapp. 35, 918–928. Heni, M., Kullmann, S., Preissl, H., Fritsche, A., H€aring, H.U., 2015. Impaired insulin action in the human brain: causes and metabolic consequences. Nat. Rev. Endocrinol. 11, 701–711. Hetherington, A.W., Ranson, S.W., 1940. Hypothalamic lesions and adiposity in the rat. Anat. Rec. 78, 149–172. Hirose, S., Osada, T., Ogawa, A., Tanaka, M., Wada, H., Yoshizawa, Y., Imai, Y., Machida, T., Akahane, M., Shirouzu, I., Konishi, S., 2016. Lateral-medial dissociation in orbitofrontal cortex-hypothalamus connectivity. Front. Hum. Neurosci. 10, 244. Hirose, S., Watanabe, T., Jimura, K., Katsura, M., Kunimatsu, A., Abe, O., Ohtomo, K., Miyashita, Y., Konishi, S., 2012. Local signal time-series during rest used for areal boundary mapping in individual human brains. PLoS One 7, e36496. Hirose, S., Watanabe, T., Wada, H., Imai, Y., Machida, T., Shirouzu, I., Miyashita, Y., Konishi, S., 2013. Functional relevance of micromodules in the human association cortex delineated with high-resolution fMRI. Cereb. Cortex 23, 2863–2871. Hutchison, R.M., Culhamd, J.C., Flanagan, J.R., Everling, S., Gallivan, J.P., 2015. Functional subdivisions of medial parieto-occipital cortex in humans and nonhuman primates using resting-state fMRI. Neuroimage 116, 10–29. Jackson, R.L., Hoffman, P., Pobric, G., Lambon Ralph, M.A., 2016. The semantic network at work and rest: differential connectivity of anterior temporal lobe subregions. J. Neurosci. 36, 1490–1501. Jastreboff, A.M., Sinha, R., Arora, J., Giannini, C., Kubat, J., Malik, S., Van Name, M.A., Santoro, N., Savoye, M., Duran, E.J., Pierpont, B., Cline, G., Constable, R.T., Sherwin, R.S., Caprio, S., 2016. Altered brain response to drinking glucose and fructose in obese adolescents. Diabetes 65, 1929–1939. Kandel, E.R., Schwartz, J.H., Jessell, T.M., Siegelbaum, S.A., Hudspeth, A.J., 2013. Principles of Neural Science, fifth ed. McGraw Hill, New York. Krashes, M.J., Lowell, B.B., Garfield, A.S., 2016. Melanocortin-4 receptor-regulated energy homeostasis. Nat. Neurosci. 19, 206–219. Kubota, N., Yano, W., Kubota, T., Yamauchi, T., Itoh, S., Kumagai, H., Kozono, H., Takamoto, I., Okamoto, S., Shiuchi, T., Suzuki, R., Satoh, H., Tsuchida, A., Moroi, M., Sugi, K., Noda, T., Ebinuma, H., Ueta, Y., Kondo, T., Araki, E., Ezaki, O., Nagai, R., Tobe, K., Terauchi, Y., Ueki, K., Minokoshi, Y., Kadowaki, T., 2007. Adiponectin stimulates AMP-activated protein kinase in the hypothalamus and increases food intake. Cell Metab. 6, 55–68. Kullmann, S., Heni, M., Linder, K., Zipfel, S., H€aring, H.U., Veit, R., Fritsche, A., Preissl, H., 2014. Resting-state functional connectivity of the human hypothalamus. Hum. Brain Mapp. 35, 6088–6096. Laumann, T.O., Gordon, E.M., Adeyemo, B., Snyder, A.Z., Joo, S.J., Chen, M.Y., Gilmore, A.W., McDermott, K.B., Nelson, S.M., Dosenbach, N.U., Schlaggar, B.L., Mumford, J.A., Poldrack, R.A., Petersen, S.E., 2015. Functional system and areal organization of a highly sampled individual human brain. Neuron 87, 657–670.

Acknowledgments We thank Prof. Y. Miyashita for helpful comments, and Mr. T. Kamiya and Mr. K. Yamada for technical assistance. This work was supported by JSPS KAKENHI Grant Number 16K18367 to T.O. and a grant from Takeda Science Foundation to S.K. Appendix A. Supplementary data Supplementary data related to this article can be found at http://dx. doi.org/10.1016/j.neuroimage.2017.08.056. References Abraham, M.A., Filippi, B.M., Kang, G.M., Kim, M.S., Lam, T.K., 2014. Insulin action in the hypothalamus and dorsal vagal complex. Exp. Physiol. 99, 1104–1109. Ahr
en, B., 2000. Autonomic regulation of islet hormone secretion–implications for health and disease. Diabetologia 43, 393–410. Anand, B.K., Brobeck, J.R., 1951. Hypothalamic control of food intake in rats and cats. Yale J. Biol. Med. 24, 123–140. Baroncini, M., Jissendi, P., Balland, E., Besson, P., Pruvo, J.P., Francke, J.P., Dewailly, D., Blond, S., Prevot, V., 2012. MRI atlas of the human hypothalamus. Neuroimage 59, 168–180. Batterham, R.L., Ffytche, D.H., Rosenthal, J.M., Zelaya, F.O., Barker, G.J., Withers, D.J., Williams, S.C., 2007. PYY modulation of cortical and hypothalamic brain areas predicts feeding behaviour in humans. Nature 450, 106–109. Behrens, T.E., Johansen-Berg, H., Woolrich, M.W., Smith, S.M., Wheeler-Kingshott, C.A., Boulby, P.A., Barker, G.J., Sillery, E.L., Sheehan, K., Ciccarelli, O., Thompson, A.J., Brady, J.M., Matthews, P.M., 2003. Non-invasive mapping of connections between human thalamus and cortex using diffusion imaging. Nat. Neurosci. 6, 750–757. Bellisle, F., Louis-Sylvestre, J., Demozay, F., Blazy, D., Le Magnen, J., 1983. Reflex insulin response associated to food intake in human subjects. Physiol. Behav. 31, 515–521. Berthoud, H.R., Bereiter, D.A., Trimble, E.R., Siegel, E.G., Jeanrenaud, B., 1981. Cephalic phase, reflex insulin secretion: neuroanatomical and physiological characterization. Diabetologia 20, 393–401. Biswal, B.B., Mennes, M., Zuo, X.N., Gohel, S., Kelly, C., Smith, S.M., Beckmann, C.F., Adelstein, J.S., Buckner, R.L., Colcombe, S., Dogonowski, A.M., Ernst, M., Fair, D., Hampson, M., Hoptman, M.J., Hyde, J.S., Kiviniemi, V.J., K€ otter, R., Li, S.J., Lin, C.P., Lowe, M.J., Mackay, C., Madden, D.J., Madsen, K.H., Margulies, D.S., Mayberg, H.S., McMahon, K., Monk, C.S., Mostofsky, S.H., Nagel, B.J., Pekar, J.J., Peltier, S.J., Petersen, S.E., Riedl, V., Rombouts, S.A., Rypma, B., Schlaggar, B.L., Schmidt, S., Seidler, R.D., Siegle, G.J., Sorg, C., Teng, G.J., Veijola, J., Villringer, A., Walter, M., Wang, L., Weng, X.C., Whitfield-Gabrieli, S., Williamson, P., Windischberger, C., Zang, Y.F., Zhang, H.Y., Castellanos, F.X., Milham, M.P., 2010. Toward discovery science of human brain function. Proc. Natl. Acad. Sci. U.S.A. 107, 4734–4739. Burdakov, D., Luckman, S.M., Verkhratsky, A., 2005. Glucose-sensing neurons of the hypothalamus. Philos. Trans. R. Soc. Lond. B Biol. Sci. 360, 2227–2235. Bzdok, D., Heeger, A., Langner, R., Laird, A.R., Fox, P.T., Palomero-Gallagher, N., Vogt, B.A., Zilles, K., Eickhoff, S.B., 2015. Subspecialization in the human posterior medial cortex. Neuroimage 106, 55–71. 11

T. Osada et al.

NeuroImage 162 (2017) 1–12 human hypothalamus: a magnetic resonance study with clinical implications. Eur. Arch. Psychiatry Clin. Neurosci. 263, 497–508. Shen, X., Tokoglu, F., Papademetris, X., Constable, R.T., 2013. Groupwise whole-brain parcellation from resting-state fMRI data for network node identification. Neuroimage 82, 403–415. Smeets, P.A., de Graaf, C., Stafleu, A., van Osch, M.J., van der Grond, J., 2005. Functional MRI of human hypothalamic responses following glucose ingestion. Neuroimage 24, 363–368. Smith, S.M., Jenkinson, M., Woolrich, M.W., Beckmann, C.F., Behrens, T.E., JohansenBerg, H., Bannister, P.R., De Luca, M., Drobnjak, I., Flitney, D.E., Niazy, R.K., Saunders, J., Vickers, J., Zhang, Y., De Stefano, N., Brady, J.M., Matthews, P.M., 2004. Advances in functional and structural MR image analysis and implementation as FSL. Neuroimage 23, 208–219. Sternson, S.M., 2013. Hypothalamic survival circuits: blueprints for purposive behaviors. Neuron 77, 810–824. Teeuwisse, W.M., Widya, R.L., Paulides, M., Lamb, H.J., Smit, J.W., de Roos, A., van Buchem, M.A., Pijl, H., van der Grond, J., 2012. Short-term caloric restriction normalizes hypothalamic neuronal responsiveness to glucose ingestion in patients with type 2 diabetes. Diabetes 61, 3255–3259. Teff, K.L., 2011. How neural mediation of anticipatory and compensatory insulin release helps us tolerate food. Physiol. Behav. 103, 44–50. Teff, K.L., Mattes, R.D., Engelman, K., 1991. Cephalic phase insulin release in normal weight males: verification and reliability. Am. J. Physiol. 261, E430–E436. Vidarsdottir, S., Smeets, P.A., Eichelsheim, D.L., van Osch, M.J., Viergever, M.A., Romijn, J.A., van der Grond, J., Pijl, H., 2007. Glucose ingestion fails to inhibit hypothalamic neuronal activity in patients with type 2 diabetes. Diabetes 56, 2547–2550. Wagner, A.D., Schacter, D.L., Rotte, M., Koutstaal, W., Maril, A., Dale, A.M., Rosen, B.R., Buckner, R.L., 1998. Building memories: remembering and forgetting of verbal experiences as predicted by brain activity. Science 281, 1188–1191. Wang, D., Buckner, R.L., Fox, M.D., Holt, D.J., Holmes, A., Stoecklein, S., Langs, G., Pan, R., Qian, T., Li, K., Baker, J.T., Stufflebeam, S.M., Wang, K., Wang, X., Hong, B., Liu, H., 2015. Parcellating cortical functional networks in individuals. Nat. Neurosci. 18, 1853–1860. Wig, G.S., Laumann, T.O., Cohen, A., Power, J.D., Nelson, S.M., Glasser, M.F., Miezin, F.S., Snyder, A.Z., Schlaggar, B.L., Petersen, S.E., 2014a. Parcellating an individual subject's cortical and subcortical structures using snowball sampling of resting-state correlations. Cereb. Cortex 24, 2036–2054. Wig, G.S., Laumann, T.O., Petersen, S.E., 2014b. An approach for parcellating human cortical areas using resting-state correlations. Neuroimage 93, 276–291. Williams, K.W., Elmquist, J.K., 2012. From neuroanatomy to behavior: central integration of peripheral signals regulating feeding behavior. Nat. Neurosci. 15, 1350–1355. Worsley, K.J., Friston, K.J., 1995. Analysis of fMRI time-series revisited again. Neuroimage 2, 173–181. Yeo, B.T., Krienen, F.M., Sepulcre, J., Sabuncu, M.R., Lashkari, D., Hollinshead, M., Roffman, J.L., Smoller, J.W., Z€ ollei, L., Polimeni, J.R., Fischl, B., Liu, H., Buckner, R.L., 2011. The organization of the human cerebral cortex estimated by intrinsic functional connectivity. J. Neurophysiol. 106, 1125–1165. Zhang, S., Ide, J.S., Li, C.S., 2012. Resting-state functional connectivity of the medial superior frontal cortex. Cereb. Cortex 22, 99–111. Zhang, S., Li, C.S., 2012. Functional connectivity mapping of the human precuneus by resting state fMRI. Neuroimage 59, 3548–3562.

Le Bihan, D., Johansen-Berg, H., 2012. Diffusion MRI at 25: exploring brain tissue structure and function. Neuroimage 61, 324–341. Little, T.J., McKie, S., Jones, R.B., D'Amato, M., Smith, C., Kiss, O., Thompson, D.G., McLaughlin, J.T., 2014. Mapping glucose-mediated gut-to-brain signalling pathways in humans. Neuroimage 96, 1–11. Liu, Y., Gao, J.H., Liu, H.L., Fox, P.T., 2000. The temporal response of the brain after eating revealed by functional MRI. Nature 405, 1058–1062. Long, X., Goltz, D., Margulies, D.S., Nierhaus, T., Villringer, A., 2014. Functional connectivity-based parcellation of the human sensorimotor cortex. Eur. J. Neurosci. 39, 1332–1342. Louis-Sylvestre, J., 1976. Preabsorptive insulin release and hypoglycemia in rats. Am. J. Physiol. 30, 56–60. Louis-Sylvestre, J., 1978. Relationship between two stages of prandial insulin release in rats. Am. J. Physiol. 235, E103–E111. Makris, N., Swaab, D.F., van der Kouwe, A., Abbs, B., Boriel, D., Handa, R.J., Tobet, S., Goldstein, J.M., 2013. Volumetric parcellation methodology of the human hypothalamus in neuroimaging: normative data and sex differences. Neuroimage 69, 1–10. Mars, R.B., Jbabdi, S., Sallet, J., O'Reilly, J.X., Croxson, P.L., Olivier, E., Noonan, M.P., Bergmann, C., Mitchell, A.S., Baxter, M.G., Behrens, T.E., Johansen-Berg, H., Tomassini, V., Miller, K.L., Rushworth, M.F., 2011. Diffusion-weighted imaging tractography-based parcellation of the human parietal cortex and comparison with human and macaque resting-state functional connectivity. J. Neurosci. 31, 4087–4100. Matsuda, M., Liu, Y., Mahankali, S., Pu, Y., Mahankali, A., Wang, J., DeFronzo, R.A., Fox, P.T., Gao, J.H., 1999. Altered hypothalamic function in response to glucose ingestion in obese humans. Diabetes 48, 1801–1806. Miller, N.E., 1960. Motivational effects of brain stimulation and drugs. Fed. Proc. 19, 846–854. Onoda, K., Yamaguchi, S., 2013. Small-worldness and modularity of the resting-state functional brain network decrease with aging. Neurosci. Lett. 556, 104–108. Page, K.A., Chan, O., Arora, J., Belfort-Deaguiar, R., Dzuira, J., Roehmholdt, B., Cline, G.W., Naik, S., Sinha, R., Constable, R.T., Sherwin, R.S., 2013. Effects of fructose vs glucose on regional cerebral blood flow in brain regions involved with appetite and reward pathways. JAMA 309, 63–70. Poldrack, R.A., Laumann, T.O., Koyejo, O., Gregory, B., Hover, A., Chen, M.Y., Gorgolewski, K.J., Luci, J., Joo, S.J., Boyd, R.L., Hunicke-Smith, S., Simpson, Z.B., Caven, T., Sochat, V., Shine, J.M., Gordon, E., Snyder, A.Z., Adeyemo, B., Petersen, S.E., Glahn, D.C., Mckay, D.R., Curran, J.E., G€ oring, H.H., Carless, M.A., Blangero, J., Dougherty, R., Leemans, A., Handwerker, D.A., Frick, L., Marcotte, E.M., Mumford, J.A., 2015. Long-term neural and physiological phenotyping of a single human. Nat. Comm. 6, 8885. Power, J.D., Barnes, K.A., Snyder, A.Z., Schlaggar, B.L., Petersen, S.E., 2012. Spurious but systematic correlations in functional connectivity MRI networks arise from subject motion. Neuroimage 59, 2142–2154. Reid, A.T., Lewis, J., Bezgin, G., Khundrakpamb, B., Eickhoff, S.B., McIntosh, A.R., Bellec, P., Evans, A.C., 2016. A cross-modal, cross-species comparison of connectivity measures in the primate brain. Neuroimage 125, 311–331. Schindler, S., Sch€ onknecht, P., Schmidt, L., Anwander, A., Strauß, M., Trampel, R., Bazin, P.-.L., M€ oller, H.E., Hegerl, U., Turner, R., Geyer, S., 2013. Development and evaluation of an algorithm for the computer-assisted segmentation of the human hypothalamus on 7-tesla magnetic resonance images. PLoS One 8, e66394. Sch€ onknecht, P., Anwander, A., Petzold, F., Schindler, S., Kn€ osche, T.R., M€ oller, H.E., Hegerl, U., Turner, R., Geyer, S., 2013. Diffusion imaging-based subdivision of the

12