Heterogeneous Disease Progression in a Mouse Model of Vascular Cognitive Impairment.

Recently, an asymmetric vascular compromise approach that replicates many aspects of human vascular cognitive impairment (VCI) has been reported. The present study aimed to first investigate on the reproducibility in the disease progression of this newly reported VCI model using wild-type C57BL6/J mice. The second aim was to assess how this approach will affect the disease progression of transgenic Alzheimer’s disease (AD) 5XFAD mice subjected to VCI. C57BL6/J and 5XFAD mice were subjected to VCI by placing an ameroid constrictor on the right CCA and a microcoil on the left CCA. Infarcts and hippocampal neuronal loss did not appear predominantly in the right (ameroid side) as expected but randomly in both hemispheres. The mortality rate of C57BL6/J mice was unexpectedly high. Inducing VCI reduced amyloid burden in the hippocampi of 5XFAD mice. Since VCI is known to be complex and complicated, the heterogeneous disease progression observed from this current study shares close resemblance to the clinical manifestation of VCI. This heterogeneity, however, makes it challenging to test novel treatment options using this model. Further study is warranted to tackle the heterogeneous nature of VCI.


Introduction
Cognitive impairment or dementia caused by cerebrovascular injury is referred to as vascular cognitive impairment (VCI). Second to Alzheimer's disease (AD), VCI is the most common cause of cognitive impairment in the elderly population. VCI is heterogeneous depending on the pathophysiology and distribution of cerebrovascular injury [1,2]. The pathogenesis of VCI is multifactorial. Common pathology underlying VCI include thromboembolism and hypoperfusion.

C57VCI Mice Displayed a Wide Variation in Development of Infarcts
Magnetic resonance imaging (MRI) was utilized to assess infarct characteristics such as progression over time and volume. T2 weighted MR images were acquired weekly (post 8, 15, 22, and 29 days) following VCI surgery ( Figure S1). Microbleeds and hemorrhages were not observed from the C57VCI mice. Out of the 17 C57VCI mice, infarcts were detected from a total of six mice ( C57VCI 4,5,9,15,16,and 17) (Table S1). A variation in size and appearance of cerebral infarcts was, however, noted from the MR images ( Figure 1). The appearance and size of the infarcts differed among the C57VCI mice. For example, infarcts were first observed at Day 15 for mouse C57VCI 4, Day 29 for C57VCI 5, and Day 8 for C57VCI 9 ( Figure 1A). Interestingly, there was one mouse that displayed infarcts in both hemispheres (C57VCI 17) and the septal nuclei ( Figure 1A). For this mouse, infarcts were first observed and detected only in the right hemisphere at day 22 but when images were taken subsequently at day 29, infarcts were also observed from the opposite hemisphere (left) and the septal nuclei ( Figure 1A). from the C57VCI mice. Out of the 17 C57VCI mice, infarcts were detected from a total of six mice ( C57VCI 4,5,9,15,16,and 17) (Table S1). A variation in size and appearance of cerebral infarcts was, however, noted from the MR images ( Figure 1). The appearance and size of the infarcts differed among the C57VCI mice. For example, infarcts were first observed at Day 15 for mouse C57VCI 4, Day 29 for C57VCI 5, and Day 8 for C57VCI 9 ( Figure 1A). Interestingly, there was one mouse that displayed infarcts in both hemispheres (C57VCI 17) and the septal nuclei ( Figure 1A). For this mouse, infarcts were first observed and detected only in the right hemisphere at day 22 but when images were taken subsequently at day 29, infarcts were also observed from the opposite hemisphere (left) and the septal nuclei ( Figure 1A). A total of 18 infarcts were detected from the six mice and many of the infarcts were observed from the right hemisphere (on the ameroid side) ( Figure 1B). Many of the cerebral infarcts were detected in the following areas of the mouse brain: caudate putamen (CPu), corpus callosum (CC), anterior commissure (AC), cortex, and hippocampus (hippo). The highest number of infarcts were observed from the right CC followed by the right and left CPu, respectively ( Figure 1C). The volume of the cerebral infarcts ranged from 0.051 to 4.774 mm 3 ( Figure 1D). The infarct with the greatest volume was detected in the left CC ( Figure 1D). Cerebral infarcts visualized from the MR images  C57VCI 4,5,9,17). Cerebral infarcts appeared at 4 different time points ( Day 15,22,8,and 29,respectively) for each of the mice. Hyperintense signals (solid yellow arrows) indicate the location of infarcts. L indicates left and R indicates the right hemisphere of the mouse brain. The numbers on the sagittal section of the mouse brain (top illustration) indicate the localization of cerebral infarcts in representative areas of the C57VCI parenchyma: (1) forceps minor (2) external capsule of the corpus callosum (3) caudate putamen, and (4) hippocampal fimbria. (B) Total number of infarcts observed from 6 of the 17 C57VCI mice (C57VCI 4,5,9,15,16,17). (C) Distribution of infarcts. CC = corpus callosum, CPu = caudate putamen, AC = anterior commissure, Hippo = hippocampus, and HF = hippocampal fimbria. (D) Volume of the cerebral infarcts (mm 3 ).
A total of 18 infarcts were detected from the six mice and many of the infarcts were observed from the right hemisphere (on the ameroid side) ( Figure 1B). Many of the cerebral infarcts were detected in the following areas of the mouse brain: caudate putamen (CPu), corpus callosum (CC), anterior commissure (AC), cortex, and hippocampus (hippo). The highest number of infarcts were observed from the right CC followed by the right and left CPu, respectively ( Figure 1C). The volume of the cerebral infarcts ranged from 0.051 to 4.774 mm 3 ( Figure 1D). The infarct with the greatest volume was detected in the left CC ( Figure 1D). Cerebral infarcts visualized from the MR images were corroborated by H&E staining (Figure 2A). The MR images and the corresponding H&E stains were visualized from four representative areas: forceps minor of CC (indicated as 1), external capsule of CC (indicated as 2), internal capsule and caudate putamen (indicated as 3), hippocampal fimbria (indicated as 4), and or the cerebral cortex (not noted in Figure 2A). When immunostaining was performed, the infarct site was easily demarcated (black broken line; Figure 2A) as an empty void and signs of inflammatory cell infiltration (Iba-1 and GFAP) were not observed. were corroborated by H&E staining (Figure 2A). The MR images and the corresponding H&E stains were visualized from four representative areas: forceps minor of CC (indicated as 1), external capsule of CC (indicated as 2), internal capsule and caudate putamen (indicated as 3), hippocampal fimbria (indicated as 4), and or the cerebral cortex (not noted in Figure 2A). When immunostaining was performed, the infarct site was easily demarcated (black broken line; Figure 2A) as an empty void and signs of inflammatory cell infiltration (Iba-1 and GFAP) were not observed.  Along with cerebral infarcts, C57VCI mice also showed selective neuronal death in the CA1 and CA2 areas of the hippocampus ( Figure 2B). The lesion was usually unilateral (out of the six C57VCI mice that survived up to Day 32, three showed loss on the ameroid side and one on the microcoil side) and compared to the contralesional hippocampus, pyknosis and TUNEL positive cells were detected from the ipsilesional pyramidal neurons ( Figure 2B). Along with hippocampal neuronal death, alterations in the hippocampal structure such as hippocampal folding (indicated by a yellow arrow) were also evident on the ipsilesional side in a few mice ( Figure 2C).

Differences in DTI Parameters Were Not Significant without the Presence of Infarcts
Interestingly, when comparing the C57VCI (six that survived up to 32 days; C57VCI 1,2,5,6,11, and 12) to the sham group, statistically significant differences in DTI indices (normalized = entire corpus callosum (CC)/whole brain) were not observed ( Figure 4A). Although tract density was slightly reduced (normalized: CC/whole brain), the difference was not statistically significant ( Figure 4B). We further investigated whether the presence of infarcts in the CC affect DTI parameters (normalized: entire CC including infarct/whole brain).
Three C57VCI mice that died before the 32 days with cerebral infarcts (C57VCI <32 days: C57VCI 15, 16, and 17) were included in the analyses. Compared to the sham group, a statistically significant reduction in FA (* p < 0.05) and AD (* p < 0.05) were observed from the C57VCI <32 days group ( Figure 4A). Interestingly, significant changes in demyelination (RD) and tissue integrity (MD) were not observed. When observing the tractography of the corpus callosum, reduced fiber density (yellow solid arrow; Figure 4B) was noted on the ipsilesional side of the C57VCI <32 days group. However, when compared to the tract density of the sham group, differences were not significant. Instead of measuring the DTI indices of the entire corpus callosum, DTI indices of the contralesional side were also compared to that of the ipsilesional side ( Figure 4C). Pronounced effects on the DTI parameters were noted when excluding the sham group and comparing the contralesional to the ipsilesional side ( Figure 4C). Compared to the contralesional side, a significant reduction in TD (** p < 0.01), AD (*** p < 0.001), RD (*** p < 0.001), and ADC (*** p < 0.001) were noted in the ipsilesional side.
Interestingly, when comparing the C57VCI (six that survived up to 32 days; C57VCI 1,2,5,6,11, and 12) to the sham group, statistically significant differences in DTI indices (normalized = entire corpus callosum (CC)/whole brain) were not observed ( Figure 4A). Although tract density was slightly reduced (normalized: CC/whole brain), the difference was not statistically significant ( Figure  4B). We further investigated whether the presence of infarcts in the CC affect DTI parameters (normalized: entire CC including infarct/whole brain).  The normalized values of the CC (the CC value were divided by the value of the whole brain) were used for group comparisons: Sham, C57VCI (sacrificed at 32 days), and C57VCI <32 days are shown in bar graphs on the right. * p < 0.05 vs. sham; mean ± S.E.M. L indicates left and R indicates the right hemisphere of the mouse brain. (B) Tractography of a representative animal from each of the groups (from left to right: axial, dorsal, left hemisphere, right hemisphere) and the normalized (CC/whole brain) tract density shown as bar graphs. The solid yellow arrows indicate reduced fiber density. (C) Normalized DTI parameters (the value on the side of the infarct was divided by the value of the contralesional side of the C57VCI <32 days group) quantitated by comparing the ipsilesional to the contralesional side of the C57VCI <32 days group. ** p < 0.01, *** p < 0.001 vs. contralesional; mean ± S.E.M.
Three C57VCI mice that died before the 32 days with cerebral infarcts (C57VCI <32 days: C57VCI 15, 16, and 17) were included in the analyses. Compared to the sham group, a statistically significant reduction in FA (* p < 0.05) and AD (* p < 0.05) were observed from the C57VCI <32 days group ( Figure  4A). Interestingly, significant changes in demyelination (RD) and tissue integrity (MD) were not observed. When observing the tractography of the corpus callosum, reduced fiber density (yellow solid arrow; Figure 4B) was noted on the ipsilesional side of the C57VCI <32 days group. However, when compared to the tract density of the sham group, differences were not significant. Instead of measuring the DTI indices of the entire corpus callosum, DTI indices of the contralesional side were also compared to that of the ipsilesional side ( Figure 4C). Pronounced effects on the DTI parameters The normalized values of the CC (the CC value were divided by the value of the whole brain) were used for group comparisons: Sham, C57VCI (sacrificed at 32 days), and C57VCI <32 days are shown in bar graphs on the right. * p < 0.05 vs. sham; mean ± S.E.M. L indicates left and R indicates the right hemisphere of the mouse brain. (B) Tractography of a representative animal from each of the groups (from left to right: axial, dorsal, left hemisphere, right hemisphere) and the normalized (CC/whole brain) tract density shown as bar graphs. The solid yellow arrows indicate reduced fiber density. (C) Normalized DTI parameters (the value on the side of the infarct was divided by the value of the contralesional side of the C57VCI <32 days group) quantitated by comparing the ipsilesional to the contralesional side of the C57VCI <32 days group. ** p < 0.01, *** p < 0.001 vs. contralesional; mean ± S.E.M.

5XVCI Mice Displayed Pathological Heterogeneity and Reduced Aβ Levels in the Hippocampus
Following the experiment using C57VCI mice, a second experiment was performed to assess whether it would be feasible to detect changes in AD pathogenesis by 5XFAD AD mice to VCI (5XVCI) ( Figure S2). Interestingly, the survival rate of the 5XFAD mice was strikingly higher (83.3%) when compared to that of the C57VCI mice ( Figure S2). Out of a total of six 5XVCI mice, only one mouse died at Day 22 (5XVCI 5) while the remaining five mice survived up to Day 32 (5XVCI 1,2,3,4, and 6) ( Figure S2). Only two out of the six 5XVCI mice (5XVCI 5 and 6) exhibited infarcts in the parenchyma (Table S2). Like the C57VCI mice, 5XVCI mice also exhibited signs of cerebral infarcts in regions such as the corpus callosum (CC), caudate putamen (CPu), and hippocampal fimbria (HF) ( Figure 5A). Most of the infarcts were observed from the right hemisphere of the 5XVCI mice ( Figure 5B,C). The volume of the cerebral infarcts ranged from 0.102 to 4.339 mm 3 ( Figure 5D). Like the C57VCI group, the infarct with the greatest volume was detected in the left CC of the 5XVCI group ( Figure 5D).
Hippocampal neuronal death was also observed unilaterally in the CA1 and CA2 regions of the right hippocampus (yellow arrow; Figure 5E). Hippocampal folding was also discernible ( Figure 5E). Like the C57VCI mice, a heterogeneity was noted in the manifestation of pathological features (presence of infarcts and neuronal loss) in 5XVCI mice (Table S2). Only two of the five 5XVCI mice that survived up to 32 days (5XVCI 1 and 4) showed signs of hippocampal neuronal loss (Table S2). A slight reduction in NeuN positive neuronal cell density was noted but impairment in spatial working memory (SAP) was not observed from the 5XVCI mice ( Figure 6A,B). In comparison to the 5X-Sham group (63.0% ± 5.8%) the SAP % was higher for the 5XVCI group (73.9% ± 10.1%). Moreover, statistically significant differences in AAR% and number of entries were also not noted ( Figure 6A). NeuN neuronal density levels were as follows: 5X-Sham: 77.7% ± 3.0%, 5XVCI: 63.9% ± 4.3%, * p < 0.05) ( Figure 6B). Although the fold change was not as high as the C57VCI mice, a slight increase in Iba-1 microglia expression was observed from the hippocampal region of the 5XVCI group (13.4% ± 3.2%) in comparison to the 5X-Sham group (6.2% ± 0.7%). Dramatic differences in GFAP positive astrocyte expressions were not noted between the two groups: 5X-Sham: 8.2% ± 0.7%, 5XVCI: 6.8% ± 0.9% ( Figure 6C). Following the experiment using C57VCI mice, a second experiment was performed to assess whether it would be feasible to detect changes in AD pathogenesis by 5XFAD AD mice to VCI (5XVCI) ( Figure S2). Interestingly, the survival rate of the 5XFAD mice was strikingly higher (83.3%) when compared to that of the C57VCI mice ( Figure S2). Out of a total of six 5XVCI mice, only one mouse died at Day 22 (5XVCI 5) while the remaining five mice survived up to Day 32 (5XVCI 1,2,3,4, and 6) ( Figure S2). Only two out of the six 5XVCI mice (5XVCI 5 and 6) exhibited infarcts in the parenchyma (Table S2). Like the C57VCI mice, 5XVCI mice also exhibited signs of cerebral infarcts in regions such as the corpus callosum (CC), caudate putamen (CPu), and hippocampal fimbria (HF) ( Figure 5A). Most of the infarcts were observed from the right hemisphere of the 5XVCI mice ( Figure  5B,C). The volume of the cerebral infarcts ranged from 0.102 to 4.339 mm 3 ( Figure 5D). Like the C57VCI group, the infarct with the greatest volume was detected in the left CC of the 5XVCI group ( Figure 5D).  Hippocampal neuronal death was also observed unilaterally in the CA1 and CA2 regions of the right hippocampus (yellow arrow; Figure 5E). Hippocampal folding was also discernible ( Figure 5E). Like the C57VCI mice, a heterogeneity was noted in the manifestation of pathological features (presence of infarcts and neuronal loss) in 5XVCI mice (Table S2). Only two of the five 5XVCI mice that survived up to 32 days (5XVCI 1 and 4) showed signs of hippocampal neuronal loss (Table S2). A slight reduction in NeuN positive neuronal cell density was noted but impairment in spatial working memory (SAP) was not observed from the 5XVCI mice ( Figure 6A,B). In comparison to the 5X-Sham group (63.0% ± 5.8%) the SAP % was higher for the 5XVCI group (73.9% ± 10.1%). Moreover, statistically significant differences in AAR% and number of entries were also not noted ( Figure 6A). NeuN neuronal density levels were as follows: 5X-Sham: 77.7% ± 3.0%, 5XVCI: 63.9% ± 4.3%, * p < 0.05) ( Figure 6B). Although the fold change was not as high as the C57VCI mice, a slight increase in Iba-1 microglia expression was observed from the hippocampal region of the 5XVCI group (13.4% ± 3.2%) in comparison to the 5X-Sham group (6.2% ± 0.7%). Dramatic differences in GFAP positive astrocyte expressions were not noted between the two groups: 5X-Sham: 8.2% ± 0.7%, 5XVCI: 6.8% ± 0.9% ( Figure 6C). Aβ IHC staining was carried out to evaluate changes in amyloid burden of 5XFAD after inducing VCI. The 6E10 Aβ monoclonal antibody was used to stain all Aβ expressing cells and plaques in the hippocampi and thalamus of both 5X-Sham and 5XVCI mice. The area of amyloid burden was quantitated afterwards. While no significant changes were observed from the thalamus (5X-Sham: 7.2% ± 1.7%, 5XVCI: 5.0% ± 0.3%, * p < 0.05), a significant reduction in area of amyloid burden in the hippocampus was noted from the 5XVCI group ( Figure 6D; 5X-Sham: 7.2% ± 0.9%, 5XVCI: 5.6% ± 0.3%). Hippocampal neuronal death was also observed unilaterally in the CA1 and CA2 regions of the right hippocampus (yellow arrow; Figure 5E). Hippocampal folding was also discernible ( Figure 5E). Like the C57VCI mice, a heterogeneity was noted in the manifestation of pathological features (presence of infarcts and neuronal loss) in 5XVCI mice (Table S2). Only two of the five 5XVCI mice that survived up to 32 days (5XVCI 1 and 4) showed signs of hippocampal neuronal loss (Table S2). A slight reduction in NeuN positive neuronal cell density was noted but impairment in spatial working memory (SAP) was not observed from the 5XVCI mice ( Figure 6A,B). In comparison to the 5X-Sham group (63.0% ± 5.8%) the SAP % was higher for the 5XVCI group (73.9% ± 10.1%). Moreover, statistically significant differences in AAR% and number of entries were also not noted ( Figure 6A). NeuN neuronal density levels were as follows: 5X-Sham: 77.7% ± 3.0%, 5XVCI: 63.9% ± 4.3%, * p < 0.05) ( Figure 6B). Although the fold change was not as high as the C57VCI mice, a slight increase in Iba-1 microglia expression was observed from the hippocampal region of the 5XVCI group (13.4% ± 3.2%) in comparison to the 5X-Sham group (6.2% ± 0.7%). Dramatic differences in GFAP positive astrocyte expressions were not noted between the two groups: 5X-Sham: 8.2% ± 0.7%, 5XVCI: 6.8% ± 0.9% ( Figure 6C).  Aβ IHC staining was carried out to evaluate changes in amyloid burden of 5XFAD after inducing VCI. The 6E10 Aβ monoclonal antibody was used to stain all Aβ expressing cells and plaques in the hippocampi and thalamus of both 5X-Sham and 5XVCI mice. The area of amyloid burden was quantitated afterwards. While no significant changes were observed from the thalamus (5X-Sham: 7.2% ± 1.7%, 5XVCI: 5.0% ± 0.3%, * p < 0.05), a significant reduction in area of amyloid burden in the hippocampus was noted from the 5XVCI group ( Figure 6D; 5X-Sham: 7.2% ± 0.9%, 5XVCI: 5.6% ± 0.3%).

Discussion
In the present study, we observed the heterogeneous disease progression and inconsistent reproducibility of both WT C57BL6/J and transgenic AD 5XFAD mice subjected to asymmetric vascular compromise. One finding that was noticeable in our study was that the mortality rate of the C57VCI mice was strikingly high. In comparison to a previously reported study [7] where the survival rate of the VCI-induced mice was around 80% at post 28 days, the survival rate for the current study was around 27%. It has been reported in the past that the C57BL/6 strain is highly susceptible to cerebral ischemia when compared to other strains such as the ICR, BALB/c, and C3H strains [20]. Previous studies also showed that severe ischemia occurs in the C57BL/6 strain when bilateral CCA occlusion was performed by temporarily occluding the carotid artery using microaneurysm clips for 20 minutes [20]. The poorly developed or even absent posterior communicating artery in close to 90% of the C57BL6/J mice [8,21] could have accounted for this high

Discussion
In the present study, we observed the heterogeneous disease progression and inconsistent reproducibility of both WT C57BL6/J and transgenic AD 5XFAD mice subjected to asymmetric vascular compromise. One finding that was noticeable in our study was that the mortality rate of the C57VCI mice was strikingly high. In comparison to a previously reported study [7] where the survival rate of the VCI-induced mice was around 80% at post 28 days, the survival rate for the current study was around 27%. It has been reported in the past that the C57BL/6 strain is highly susceptible to cerebral ischemia when compared to other strains such as the ICR, BALB/c, and C3H strains [20]. Previous studies also showed that severe ischemia occurs in the C57BL/6 strain when bilateral CCA occlusion was performed by temporarily occluding the carotid artery using microaneurysm clips for 20 minutes [20]. The poorly developed or even absent posterior communicating artery in close to 90% of the C57BL6/J mice [8,21] could have accounted for this high susceptibility. Since the C57BL/6 strain is more prone to developing cerebral infarcts, it makes it difficult to control the onset, size, and number of infarcts that will develop in the mouse brain. This could partly explain the pathological heterogeneity observed specifically in relation to the presence and severity of cerebral infarcts observed from the VCI-induced mice of the current study. It is interesting to note that a unilateral application of a single ameroid constrictor (inner diameter of 0.5 mm with a microcoil applied around the opposite CCA) was able to exert effects of similar magnitude (high mortality rate and multiple cerebral infarcts) comparative to models where ameroid constrictors were applied bilaterally to the CCAs [8].
Previously reported [7] pathological features such as cerebral infarcts and hippocampal neuronal loss were reproduced from both C57VCI and 5XVCI mice. However, there was a wide variation in terms of which hemisphere the infarcts appeared at. Out of the six C57VCI mice that survived up to 32 days, only one of the mice (C57VCI 5) displayed signs of infarct that was on the left or microcoil side. Out of the entire 17 C57VCI mice (including mice that died before 32 days), one displayed infarcts on the microcoil side (L), three on the ameroid side (R), and one on both hemispheres. Such results contrasted strongly to past findings where 81% of VCI-induced mice (day 32) displayed multiple infarcts in the right hemisphere [7]. Similarly, out of the six 5XVCI (including one mouse that died before 32 days) mice, one mouse exhibited infarcts on the microcoil side (L) side while infarcts were detected on the ameroid side (R) of another mouse. With respect to anatomical localization, the highest number of infarcts were detected from the right CC of C57VCI mice while an even distribution of infarcts was observed from the 5XVCI mice. Interestingly, the greatest infarct volumes were observed from the left CC for both groups. Such results are not surprising in that the CC is part of the watershed territory which makes it a highly susceptible region to ischemia.
Like subcortical infarcts, it was expected that hippocampal neuronal loss would be predominantly observed from the ameroid side. However, out of the six C57VCI mice that survived up to 32 days, only three mice showed signs of hippocampal neuronal death in the ameroid side (R) and one mouse on the microcoil side (L). Out of the entire 17 C57VCI mice, when including mice that died before 32 days, there were two other mice that showed signs of neuronal loss in the microcoil side. Hippocampal neuronal death for the 5XVCI mice was discernible from three mice (including one mouse that died before 32 days) in the ameroid side (R), while unlike the C57VCI mice, no signs of neuronal loss were visible from the microcoil side (L) from the other remaining mice. A noticeable feature is that previously 69% of the VCI-induced mice (day 32) showed signs of hippocampal neuronal death on the ameroid side [7] while for the current study it was 50% (three of the three C57VCI mice that survived up to 32 days) and 40% (two of the five 5XVCI mice that survived up to 32 days) for the C57VCI and 5XVCI groups, respectively.
It is known that hypoperfusion can lead to border zone infarcts and hippocampal sclerosis [22]. In accordance with past findings, increased levels of Iba-1positive microglia and GFAP positive astrocytes were observed from the hippocampi of the C57VCI group. An increase in microglia number is known to be a noticeable feature of hypoperfusion models [6,23,24]. Microglia is reported to be indicative of the severity of ischemic injury [25]. Along with microglia, levels of GFAP which is a marker used to label reactive astrocytes [26], is also increased in chronic hypoperfusion models [27]. A reduction in NeuN neuronal density and presence of apoptotic neurons in the CA1 and two hippocampal regions of our C57VCI mice was observed via IHC and TUNEL staining, respectively. This increase in cell apoptosis could have accounted for the activation of the astrocytes and microglial cells at the site of hippocampal lesion [28] considering that both astrocytes and microglia cells have been reported to be involved in the clearance of apoptotic neurons [29].
A past study reported that bilateral hippocampal lesions impair the spatial working memory of a rat model [30] while another group presented findings that large dorsal hippocampal lesions which encompass 40%-60% of the total hippocampal volume impair the spatial working memory of rats [31]. Smaller lesions, however, preserved the memory. In the current study, neuronal loss was evident from the unilateral hippocampus for both the C57VCI and 5XVCI mice, but the loss involved only the CA1 and 2 regions, not the entire hippocampus. Since the extent of hippocampal damage was not widespread, partial lesions might not have been enough to further impair the spatial working memory of the VCI groups. Furthermore, Y-maze was conducted at Day 32, while only three of the six C57VCI and two of the five 5XVCI mice that survived up to Day 32 showed signs of neuronal loss. Thus, there were more mice with an intact hippocampus and preserved spatial working memory remaining in each of the groups. This would have factored in increasing instead of decreasing the overall SAP of both the C57VCI and 5XVCI groups.
Diffusion tensor imaging (DTI) is known to be sensitive towards the motion of water molecules. One of the parameters of DTI is fractional anisotropy (FA). A decrease in FA indicates decrease in anisotropic water diffusion (direction-dependent) which is indicative of damage to the white matter region [32,33]. Axonal degradation can be measured by axial diffusivity (AD), demyelination can be assessed by measuring radial diffusivity (RD), and tissue integrity can also be assessed by measuring apparent diffusion coefficient (ADC) [33]. Tract density (TD) was quantitated to determine whether the presence of lesions also affects the number and the density of the tracts of the corpus callosum.
In this present study, signs of white matter damage, specifically in the corpus callosum, were assessed non-invasively by utilizing DTI. We expected that white matter rarefaction would occur predominantly on the left hemisphere (microcoil side), which, however, could not be determined due to the heterogeneous pathology of the C57VCI mice. Infarcts and hippocampal neuronal loss were observed randomly in both hemispheres and thus it was difficult to isolate the effects of ameroid constrictors from that of microcoils. Out of the six C57VCI mice that survived up to 32 days, only one mouse displayed infarcts in the microcoil side. For that mouse, infarcts were not detected in the corpus callosum. The overall lack in presence of infarcts could have accounted for the insignificant difference observed between the sham and C57VCI group (all four DTI indices; CC/whole brain). When the three C57VCI mice that died before the 32 days with cerebral infarcts (C57VCI<32 days: C57VCI 15, 16, and 17) were compared to the sham group, a reduction in FA and AD values was evident. The results suggest that the presence of infarcts in the corpus callosum increased both white matter damage (FA) and axonal degeneration (AD). Moreover, past groups have reported that a decrease in FA and AD values is indicative of early stage white matter degeneration [34,35].
Various studies have reported on how it is common to observe the co-occurrence of Alzheimer's disease and vascular pathology [36][37][38] in human patients. However, few studies have investigated on creating an AD-VCI mouse model. To the best of our knowledge, this is the first study to induce VCI in the 5XFAD transgenic AD mouse model and to study the resulting effects. The 5XFAD mouse model is a well-characterized, suitable model to study AD amyloid pathology [39]. An interesting observation made from our 5XVCI experiment was the reduction in Aβ area in the hippocampus. These results were similar to a previous report, where subjecting mice overexpressing the APP Swedish and Indiana mutation to BCAS reduced Aβ deposition and cored plaque (insoluble Aβ) formation but expedited neuronal loss and memory impairment [40]. The authors proposed that chronic hypoperfusion could have led to a reduction in insoluble Aβ via increased inflammation or shift in Aβ solubility [40]. However, the effects of chronic hypoperfusion in AD pathology is debatable. One group has noted an increase in Aβ deposition, a month after BCAS was induced, in a transgenic AD mouse model with Swedish, Dutch, and Iowa mutations [41]. Furthermore, an increase in total number of Aβ plaques was also demonstrated in a mouse of model of severe chronic cerebral hypoperfusion (SCCH), which was induced by ligating the right CCA with a silk thread and applying a vessel clamp to the left CCA of APPswe/PS1 mice [13].
The selective susceptibility of the hippocampus towards ischemic brain injury [42] could partly explain why changes in amyloid deposition of the 5XVCI mice were only distinct in the hippocampus and not the thalamus. The common carotid artery branches out into many arteries including the middle cerebral arteries (MCA) and the posterior cerebral arteries (PCA) [43]. The PCA plays a role in hippocampal formation by providing blood supply to the hippocampal region [44]. Thus, although only two of the five 5XVCI mice displayed cerebral infarcts, all of the 5XVCI mice would have undergone stenosis of the CCA over the course of 32 days which would have affected the blood flow and the dynamics of amyloid deposition in the hippocampus. Furthermore, the 6E10 antibody used to evaluate amyloid deposition is reactive towards Aβ proteins which are expressed not only in plaques but also in the neuronal cells of the hippocampus. Since neuronal loss was evident from two of the five 5XVCI mice, this could have factored in reducing the overall level of amyloid burden (in the hippocampus) of the 5XVCI group. It has been reported in the past that at injured or affected areas of the brain, there is an intense expression of activated microglia while amyloid deposition is absent [45]. This may be consistent with our results that the 5XVCI group which exhibited decreased levels of amyloid burden in the hippocampus showed, although not statistically significant, a 2.2-fold increase in Iba-1 positive microglia expression in comparison to the 5X-Sham group.
Limitations of this study included small sample size and absence of a baseline. The heterogeneous disease progression observed from this current study underscore the need of setting a baseline. Temporal profiles of the cerebral blood flow (CBF) were not recorded. Measuring the hemodynamics at baseline and at different time points could have provided the means of making a criterion to exclude certain mice subjected to VCI below a set cut-off. Such exclusions could have reduced heterogeneity and improved the overall reproducibility of the model in this study.
Taken together, the pathological heterogeneity observed from the current study recapitulates the heterogeneous nature of VCI that is observed clinically. The complicated progressions of two different pathologies (AD and vascular) could attribute to differences in disease progression that were observed between the C57VCI and 5XVCI mice. Further research is warranted to elucidate this difference and the complex mechanisms underlying asymmetric vascular compromise. Such investigations may provide solutions to increase the reproducibility of mouse VCI models which will be essential for the model to be used as a platform to test novel treatments for human VCI.

Mice
This study consisted of a total of 20 C57BL6/J (female, 3-4-month-old, Sham: n = 3, C57VCI: n = 17) and 8 transgenic AD 5XFAD (female, 3-4 month old; 5X-Sham: n = 2, 5XVCI: n = 6) mice. Both C57BL6/J and 5XFAD were originally purchased from the Jackson Laboratory (Bar Harbor, ME, USA). Both strains were maintained by mating a 5XFAD male with a C57BL6/J female mouse. Genotyping was performed from mouse tail biopsies of the offspring and the transgenic and non-transgenic littermates were separated accordingly. Mice were maintained in a 12-hour light/12-hour dark cycle and were fed ad libitum.

VCI Modeling
Initially, anesthesia was induced by placing the mice in a closed induction chamber. The induction setting was 5% isoflurane (Hana Pharmaceutical Co., Ltd., Seoul, Republic of Korea). Mice were then placed in a supine position and anesthesia was maintained at 1.5~2% during the surgical procedure. VCI was induced (Figure 1) by referring to previously reported studies [7,46]. Once the fur was removed and the incision area was sterilized with povidone iodine, a paramedian skin incision was performed right above the thyroid bone. Then, the omohyoid and sternomastoid muscles were retracted. Both the left and right carotid arteries were exposed by detaching the carotid arteries from the overlying fascia and surrounding sheath including the vagus nerve. An ameroid constrictor with a titanium jacket, an inner diameter of 0.5 mm, outer diameter of 3.25 mm, and length of 1.28 mm (Research Instruments SW, Escondido, CA, USA) was placed around the right common carotid artery (CCA) by lifting the distal and proximal ends of the CCA by using a 6-0 silk thread (Ethicon, Cincinnati, OH, USA). Surrounded by a circular, ring-shaped titanium jacket, the ameroid constrictors were applied to the CCAs through an opening slit. With the same procedure carried out on the left CCA, once the artery was exposed, a microcoil (Wuxi Samini Spring Co., Ltd., Wuxi, China) with an inner diameter of 0.18 mm was implanted by rotating the coil around the CCA right below the bifurcation point. The sham operated animals underwent the same procedure as the C57VCI group but no microcoil or ameroid constrictors were applied after exposing the bilateral CCAs. After the surgical procedure, the midline incision was closed by suturing the area with a 5-0 suture silk thread (Ethicon, Cincinnati, OH, USA). Mice that have undergone VCI surgery were sacrificed at Day 32. A Kaplan-Meier survival curve was drawn using the GraphPad Prism software to assess survival rates.

Magnetic Resonance Imaging and Data Processing
MR images were acquired on a 7T/20 MR System (Bruker-Biospin, Ettlingen, Germany) equipped with a 20 cm gradient set capable of supplying up to 400 mT/m in 100 µs rise-time. A quadrature birdcage coil (inner diameter of 72 mm; Bruker-Biospin) was used for excitation, and an actively decoupled phased array brain coil was used for signal reception. All mice were anesthetized under 2% isoflurane while acquiring the MR images. A T2 weighted spin echo sequence was used to acquire MR images. The parameters were as follows: repetition time (TR)/echo time (TE) = 3000/60 ms, number of averages = 6, echo train length = 4; in-plane resolution = 100 × 100 µm 2 ; slice thickness = 0.5 mm. A diffusion weighted spin echo was used to carry out Diffusion Tractography Imaging ( After the DTI images were acquired, the FSL package was utilized to perform head motion and eddy-current distortions. Once the motion control was performed, the Diffusion Toolkit and TrackVis software (www.trackvis.org) were utilized to draw separate region of interests (ROIs) around the corpus callosum and the whole brain. Both the medial and lateral regions (both hemispheres) of the corpus callosum were included when drawing the ROI. The mean fractional anisotropy (FA), radial diffusivity (RD), apparent diffusion coefficient (ADC), and tract density (TD) were quantitated from the ROIs. Tract density was calculated by dividing the number of tracts by the voxel number of the respective ROI (whole brain or corpus callosum). The TD/FA/AD/RD/ADC of the corpus callosum was divided by the TD/FA/AD/RD/ADC of the whole brain, respectively to obtain normalized values. Three C57VCI mice (C57VCI 15,16, and17) that died before the 32-day endpoint (C57VCI < 32 days) and had cerebral infarcts were also separately analyzed. The TD/FA/AD/RD/ADC of the infarct side of the CC (ipsilesional) was divided by the TD/FA/AD/RD/ADC of the contralesional CC, respectively. The MRICro (www.mricro.com) software was also used to measure the infarct volume of the VCI induced mice.

Histological Staining and Analyses
All sham and VCI-operated mice (both C57VCI and 5XVCI) were sacrificed through cardiac perfusion 32 days after VCI was induced. Post-mortem examination of several mice that died before 32 days was also performed. Brain tissues were harvested and fixated in 4% paraformaldehyde (Biosesang, Republic of Korea) for a day before paraffin blocks were made. Sections of 4 µm were made of the paraffin brain blocks using a micrometer (Leica Biosystems, Wetzlar, Germany). Slides were deparaffinized with xylene and varying percentages of ethanol. 1X Citrate buffer (pH 6; Dako, Carpinteria, CA, USA) was then used to perform heat antigen retrieval. Prior to performing immunohistochemical (IHC) staining, Hematoxylin and Eosin (H&E; Dako, CA, USA) was performed to examine changes in histology such as the presence of cerebral infarcts or neuronal loss following VCI surgery. Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) staining (Millipore, Temecula, CA, USA) was performed according to the manufacturer's instructions.
All H&E and DAB stained slides were scanned using the Scanscope AT scanner (Aperio Technologies, Vista, CA, USA). Images of slides that underwent immunofluorescent staining were acquired using a confocal microscope (Carl Zeiss AG, Jena, Germany). The total number of Iba-1, or GFAP, or NeuN negative and positive cells were manually counted for each image by using the ImageJ image processing program (National Institutes of Health (NIH)). DAB stained slides were also scanned using the Vectra®Automated Imaging System (PerkinElmer Applied Biosystems, Waltham, MA, USA) where a spectral un-mixing algorithm quantitated each of the spectral components (DAB and Hematoxylin). The InForm 2.4.1 image analysis software was used to quantitate the area percentage of Aβ burden in the hippocampi and thalamus of 5X-Sham and 5XVCI mice.

Assessment of Changes in Behavioral Performance following VCI Surgery
The Y-maze test was performed at day 32 after VCI was induced to evaluate changes in spatial working memory. Experimental animals from both the sham and VCI groups were sacrificed right after the behavioral performance was conducted. A previously reported method was referred to carry out the experiment [49]. The Y-maze apparatus consisted of three arms (1, 2, and 3) that were spaced at 120 • angles. Initially, mice were placed in one arm randomly and the sequence and number of arm visits were recorded over an 8-minute period. Spontaneous alternation performance (SAP) and alternating arm return (AAR) % were computed by using the following equation: ((Number of SAP or AAR/(Number of possible alternations) × 100). SAP is an index of spatial working memory. The total number of entries was also recorded.

Statistical Analysis
All results are expressed as mean ± standard error of mean (S.E.M.). P values < 0.05 were considered statistically significant. GraphPad Prism 5.0 (GraphPad, La Jolla, CA, USA) was used for statistical analysis and graphics. A student's t-test (unpaired, two-tailed) or one-way ANOVA (Tukey correction) was used to investigate the differences between (Sham vs. C57VCI; 5X-Sham vs. 5XVCI) or among the groups (sham vs. C57VCI vs. C57VCI <32 days), respectively.