Soluble Brain Homogenates from Diverse Human and Mouse Sources Preferentially Seed Diffuse Aβ Plaque Pathology When Injected into Newborn Mouse Hosts

Seeding of pathology related to Alzheimer’s disease (AD) and Lewy body disease (LBD) by the injection of tissue homogenates, puried proteins, or recombinant proteins into model systems has revealed prion-like seeding of the protein aggregates that dene these disorders. Most commonly these homogenates are injected into adult mice stereotaxically. Injection of brain lysates into newborn mice represents an alternative approach of delivering seeds that could be used to direct the evolution of amyloid-β (Aβ) pathology co-mixed with either tau or α-synuclein (αSyn) pathology in vulnerable mouse models.


Results
We observed that the lysates from the brains of individuals with AD (Aβ+, tau+), AD/LBD (Aβ+, tau+, αSyn+), or Pathological Aging (Aβ+, tau-, αSyn-) e ciently seeded diffuse Aβ deposits, composed primarily of Aβ42 peptides, in our transgenic host animals. Moderate seeding of cerebral amyloid angiopathy (CAA) was also observed. No animal of any genotype developed discernable tau or αSyn pathology. Fear conditioning, cognitive, outcome was not signi cantly altered in APP transgenic animals injected with AD brain lysates compared to nontransgenic controls.

Conclusions
These ndings demonstrate that diffuse Aβ pathology, which is a common feature of AD, AD/LBD, and PA brains, can be easily induced by injecting newborn APP mice with crude brain homogenates. Seeding of tau or αSyn comorbidities was disappointingly ine cient in the models we used, indicating additional methodological re nement will be needed to e ciently seed AD or AD/LBD mixed pathologies by injecting newborn mice.

Background
The amyloid cascade hypothesis posits that deposition of the Aβ peptide is an initiating factor in Alzheimer's disease (AD) with downstream effects of increased in ammatory response, altered neuronal homeostasis, altered tau phosphorylation leading to tangle formation, neuronal dysfunction, and cognitive decline (1,2). Aβ containing seeds, derived from human AD and Pathological Aging (PA) brain lysates, have been shown to accelerate the time-course and severity of Aβ deposition in amyloid precursor protein (APP) transgenic models as well as impact the morphology of the seeded deposits (for review see (3)). The literature contains numerous studies in which different types of seeding preparations have been injected in different APP mouse models [reviewed in (3)]. Seeding preparations can consist of human brain homogenates, APP transgenic mouse brain homogenates, enriched preparations derived from such homogenates and synthetic Aβ brils (3). In most of these published studies, seeds are injected into adult APP mice that are a few months away from developing pathology on their own and in such cases the inherent pathological tendencies of the APP host model can contribute to the pathological outcome.
Recent studies have begun to use seeding approaches to create models of mixed AD or Lewy body disease (LBD) pathology. Partially puri ed tau aggregates from AD brains were found to seed enhanced neuritic tau pathology in the 5xFAD mouse model and the NL-G-F/human MAPT double knock-in model (4). In Parkinson's disease and LBD, the presence of Aβ plaques is associated with cortical αSyn aggregation and spreading (5,6). A number of recent studies suggest a possible synergistic relationship between αSyn and Aβ (7). Aβ deposits accelerated αSyn pathogenesis and spread throughout the brain of 5xFAD mice (8), whereas reducing αSyn expression in transgenic APP mice rescued and reversed the amyloid pathology and neurodegenerative phenotype (9). One of the goals of our study was to explore the potential to use seeding as a means to generate novel mouse models of mixed AD and LBD pathology.
A subset of cognitively normal aged individuals exhibit substantial Aβ pathology at autopsy (10). These individuals, which we use the designation of Pathologic Aging (PA), may represent prodromal, presymptomatic AD (11)(12)(13)(14). The amyloid pathology of PA brains has typically been described as diffuse in nature whereas AD brains contain both diffuse and compact/neuritic deposits of Aβ (14)(15)(16). Cognitively normal individuals with PA generally have very low levels of tau pathology (14). Rigorous biochemical examination of Aβ from the pre-frontal cortex of PA and AD brain revealed extensive overlap in Aβ levels, peptide pro les, solubility, and SDS-stable oligomeric assemblies (13). Interestingly, brain homogenates from a human PA case were shown to seed diffuse Aβ deposition in APP transgenic mice, albeit at somewhat less severe levels (17). Thus, PA brains provide a source of Aβ seeds that are less contaminated with tau seeds and therefore provide a means to examine the potential for seeded Aβ pathology to trigger tau pathology in APP/tau mice.
Our approach for this study was to inject homogenates prepared from AD, PA, and AD/LBD donors into newborn APP, APP/tau, or APP/αSyn host mice. In prior studies, we had established that injection of recombinant adeno-associated virus into newborn mice was an effective approach to obtaining widespread dispersion of injected virus (18,19). We therefore hypothesized that misfolded seeds of Aβ, tau, or αSyn might similarly be more widely dispersed to more effectively seed pathology. In fact, we observed e cient seeding of αSyn pathology in newborn mice, expressing wild-type human αSyn (M20 line), by injecting brain lysates from patients with multisystem atrophy (20). Here we performed brain lysate injection studies at P0 in the lateral ventricles of host transgenic mice that co-express humanized mutant APP with either mutant human tau-P301L or wild-type human αSyn (M20 line). Importantly, the recipient transgenic mice used in the current studies develop pathology at late ages or not at all; enabling sensitive evaluation of potential seeding. Seeding of APP, APP/tau, or APP/αSyn mice with these brain homogenates failed to produce tau or αSyn pathology but e ciently induced diffuse Aβ pathology. Notably, homogenates from human brain with high levels of cerebral amyloid angiopathy (CAA) also seeded modest CAA. We compared the data from human lysates to lysates prepared from transgenic mouse donors nding again that diffuse Aβ deposits were preferentially seeded. These ndings demonstrate that diffuse Aβ pathology can readily be induced by injecting newborn APP, APP/tau, or APP/αSyn mice with homogenates from diverse pathological specimens. Though the human homogenates used contained coexisting αSyn and tau pathologies, we were not able to directly induce LBD or tau pathology in recipient models tested here. Moreover, the induced diffuse Aβ pathology failed to secondarily induce LBD or tau pathology in mice that were bigenic for APP/tau or APP/αSyn. APP mice seeded with AD brain lysates, exhibiting diffuse Aβ pathology, also showed no signi cant impairments in a contextual fear paradigm of cognitive performance. These ndings are consistent with the idea that diffuse Aβ pathology may be less pathogenic than cored-neuritic deposits (21).

Description of AD, PA and NDC cases
For this study we utilized a subset of the human brain samples that our group previously analyzed (13). Tissue sections and frozen pre-frontal cortex (AD = 2, PA = 4, and NDC = 2) were obtained from the Mayo Clinic Brain Bank with informed consent, in accordance with the Mayo Clinic institutional review board, using previously described acquisition and diagnostic analyses (14,22,23). Samples were initially cryopulverized to allow for multiple studies. We previously analyzed the Aβ peptide pro le, quantity, and solubility in these brain samples (13). In this study, we analyzed two brains from AD patients (both aged 84), four PA brains from subjects (age range = 78 to 83) without clinical evidence of dementia, and two brains with rare or no AD lesions from elderly individuals without clinical evidence of a neurological illness (ages 75 and 78). In addition, human brain samples were also obtained from the University of Florida Neuromedicine Human Brain Tissue Bank following institutional regulations and previously described classi cation (24)(25)(26). We analyzed tissue sections and frozen pre-frontal cortex from two AD patients (ages 82 and 86), two cases with AD/LBD (ages 81 and 83), and two control individuals without pathology (ages 52 and 82). Table 1 summarizes the cases studied in this report.

Transgenic mice
The transgenic mice used in this study have been described previously and are summarized in Table 2. The PrP.APPsi mice express human/mouse APPswe/ind using the MoPrP.Xho vector, which was coinjected with a vector to express eGFP in skin ( rst described in (27)). The iTau-P301L mice express human 4R0N tau-P301L using a tet-regulated vector that includes a minimal CMV promoter (28). Tau levels were assessed by Western blotting. PBS lysates of 2.5-6 month old P301L mice were heated at 95ºC for ve minutes in the presence of denaturing SDS sample buffer, were separated on a 4-12% Bis-Tris gel (Bio-Rad, Hercules, CA) in 3-(N-morpholino)propanesulfonic acid (MOPS) running buffer (Bio-Rad) and transferred onto PVDF membrane. The membrane was blocked in casein blocking buffer and incubated overnight at 4ºC with primary antibodies to tau, Tau-13 (anti-human tau2-18; Covance) and 3026 (rabbit polyclonal antibody raised against full length 0N3R human tau and also reacts with 0N4R human tau (29,30)), and anti-β-actin (Sigma-Aldrich, St. Louis, MO) before incubation with secondary antibody, uorophore conjugated AlexaFluor 680 anti-mouse IgG (Thermo Fisher Scienti c, Waltham, MA). Protein bands were detected and quanti ed using the multiplex Odyssey Infrared Imaging system (Bio-Rad, Hercules, CA).
These mice were bred in-house. PrP.APPsi and M20 were maintained on hybrid C57BL6/C3H backgrounds, following a breeding scheme in which transgene positive males were breed to F1 B6/C3 mice. The iTau-P301L were maintained on a background of FVB mice. APPsi/Tau-P301L and APPsi/αSyn mice were generated by intercrossing mice that where hemizygous for the respective transgenes.
PrP.APPsi mice were genotyped by visualizing GFP expression, which is possible because these mice were co-injected with a transgene that expresses GFP in the skin integrated next to APP transgene.
PrP.APPsi/iTau-P301L mice were genotpyed for APP by illumination and visulation with special lter goggles (BLS Ltd, Budapest, Hungary) and for tau by PCR of tail DNA. iTau-P301L and M20 mice were genotyped by PCR of tail DNA. PrP.APPsi destined for behavioral studies were backcrossed one generation onto C57BL6. All animals were housed 1-5 per cage with unlimited access to food and water with a 14-hour light and 10-hour dark cycle. All experiments involving mice were approved by the University of Florida Institutional Animal Care and Use Committee (IACUC) and conducted in accordance with NIH guidelines.

Human brain lysates
Human and mouse brain lysates were prepared as previously described (32). Brie y, frozen tissue was homogenized at 10% (w/v) in sterile PBS, subjected to vortex and sonication (3 x 5 sec) and then centrifuged 3000 x g for 5 min at 4ºC. Lysates were immediately aliquoted and stored at -80ºC until needed. As previously described (13), the Aβ levels and solubility in the lysates was assessed by Western blotting. Brie y, lysates were heated at 50ºC for three minutes in the presence of denaturing sample buffer, were separated on a 4-12% Bis-Tris gel (Bio-Rad, Hercules, CA) in 1X2-(Nmorpholino)ethanesulfonic acid (MES) running buffer (Bio-Rad) and transferred onto nitrocellulose membrane (Bio-Rad, Hercules, CA). The membrane was boiled in PBS for 5min, blocked in Starting Block (Thermo Fisher Scienti c (formally Thermo Scienti c), Waltham, MA) and incubated overnight at 4ºC with primary antibody Ab5 (human Aβ1-16 speci c; T.E. Golde) before incubation with the secondary antibody, uorophore conjugated Alexa Fluor 680 anti-mouse IgG (Thermo Fisher Scienti c, Waltham, MA). Protein bands were detected using the multiplex Odyssey Infrared Imaging system (Bio-Rad, Hercules, CA).

Brain seeding with neonatal cerebral ventricle injections
Injections of human brain lysates and lysates from a 27mo PrP.APPsi mouse, a 25mo PrP.HuAβ/PS1 mouse, a 24mo Tet.MoAβ mouse, and a 24mo PrP.MoAβ/PS1 mouse were performed as described previously with recombinant adeno-associated virus (33). Brie y, P0 pups were cryoanesthisized and 2 uL of human or mouse brain lysates were bilaterally injected into the cerebral ventricle using a 10 uL Hamilton syringe with a 30-inch needle (Hamilton Company, Reno, NV). Pups were placed on a heating pad for recovery and returned to their home cage. Cohorts of mice were euthanized at 6, 9, 12, and 18 months for analysis.

Mouse brain tissue collection
Mice were anesthetized with iso urane and perfused transcardially with 20 mL of cold PBS. The brains were cut sagittally through the midline, and one hemibrain was drop-xed in 4% paraformaldehyde in PBS (pH 7.5) for ~48 hrs at 4°C followed by processing and para n embedding. The other hemibrain was snap frozen with isopentane on dry-ice and then stored at -80 C until it was thawed and homogenized in preparation for ELISA measurements of Aβ peptide levels.

Contextual fear conditioning
Transgenic PrP.APPsi and nontransgenic littermates, seeded with AD cases 1 and 2, and NDC case 7, were aged to 12 months and subjected to contextual fear conditioning as previously described (35,36). Brie y, in the contextual fear conditioning test, a mouse learns the association between the training chamber, which represents an initially neutral conditional stimulus (CS) and an aversive, brief electric foot-shock, unconditional stimulus (US), which takes place in the training context. We have previously established that 0.45 mA electric current elicits robust avoidance response and results in strong freezing response after 2 CS-US pairings (35). During the training session, a mouse was allowed to explore the training chamber for 120 sec, with a 2 sec foot shock immediately following a 30 sec tone (80 dB) (day 1; see Fig. 4A). The mouse recovers for 60 sec and then receives another 2 sec foot shock following a 30 sec tone. The nal post-US period is 60 sec. For the context test session, on day 3, the mice were exposed to the same training context and the freezing presentation during the 300 sec was recorded by an image analysis system (FreezeFrame, Actimetrics). On day 4 the mice experienced the tone session, they were placed in a modi ed chamber (altered inserts and smell), during the rst 180 sec the mouse was allowed to explore the new environment. During the second 180 sec period the tone was delivered and the percent freezing was recorded. No shock was applied during both the test days.

Results
Characterization of AD/LBD, AD, PA, and NDC cases In this study, we selected a subset of AD, PA, and NDC donors that were previously analyzed (13) for examination of seeding activity in APP, APP/tau, and tau transgenic mice ( Table 1, AD 1 and 2, PA 1,2,3,4, and NDC 1 and 2). All 8 of these donors exhibited some degree of Lewy body pathology (LBD). Two of the donors selected for the study were sub-categorized with cerebral amyloid angiopathy (CAA); PA 3 had a high abundance of vascular amyloid with a CAA score of 2+-3+ while AD 1 had moderate levels of vascular Aβ deposition with a score of 1+ (Table 1). As expected, the two AD brains had signi cant accumulation of tau, indicated by Braak staging (5.5 and 6) ( Table 1). The PA and NDC brains had lower abundance of tau than the AD brains, but were similar to each other (2, 2.5, 3, 2, and 2, 3, respectively) ( Table 1). In this set of cases, all also exhibited incidental, diffuse, αSyn pathology. To these 8, we added 6 additional donors that included two AD (no LBD), two with Lewy-body variant AD (AD/LBD), and 2 additional controls that were free of all types of pathology (Table 1, AD 3 and 4, AD/LBD 1 and 2, and NDC 3 and 4). These brains were used in seeding APP, APP/α-synuclein (αSyn), and αSyn mice so that we would be able to compare AD and NDC that lacked αSyn pathology to AD/LBD brains with high levels of αSyn pathology.
To characterize the pathology in the AD and PA brains used to seed the APP/tau, APP, and tau mice, sections were stained with a pan-Aβ antibody (Fig. S1a), Thio avin S (Thio-S) (Fig. S1b), and with a phospho-speci c tau antibody (Fig. S1c). We observed widespread Aβ deposition of both diffuse and compact amyloid in both AD and PA (Fig. S1a, b), with a subset showing striking CAA. PA 3 had numerous, Thio-S positive, vascular amyloid deposits, consistent with the assessment of CAA as 2+-3+ (Fig. S1a, b). Similarly, we observed Thio-S positive amyloid staining surrounding several vessels within AD 1, consistent with a CAA score of 1+ (Fig. S1a, b). AD 2 contained several cored Thio-S positive deposits while the compact deposits in PA 1,2, and 4 had little to no Thio-S staining. As expected, AD 1 and 2 contained both substantial phosphorylated tau in form of dystrophic neurites, neuropil threads and neuronal inclusions while all four of the PA were negative for tau deposits (Fig. S1c). NDC 1 and 2 were negative for Aβ and tau deposits by both immunostaining and Thio-S staining.
Cerebral injection of brain lysates into newborn APP/tau and APP mice results in widespread, robust amyloid deposition To gain a better understanding of the type of Aβ pathology that unseeded PrP.APPsi mice produce, we harvested breeder mice at various ages to assess phenotypic variation (Online Resource, Fig. S2a and b) (27). At 12 months of age, when deposition begins to occur in the brains of PrP.APPsi mice, any given section through the cortex and hippocampus may exhibit 1 or 2 diffuse tufted deposits and/or cored deposits (score +/-). At this age, Aβ deposition also begins to appear in the meninges surrounding the cerebellum (Online Resource 2, Fig. S2a). Approximately 50% of PrP.APPsi mice at 11-13 months of age exhibit no Aβ pathology or show only meningeal deposition in the cerebellum (Online Resource 2, Fig.  S2a). To concurrently examine whether tau pathology could be augmented by seeding from these lysates, the PrP.APPsi mice were crossed to a line of mice that express human tau P301L (iTau-P301L) ( Table 2) (28, 37). The iTau-P301L mice express mutant human tau at levels 2-3 fold higher than nontransgenic mice (Fig. S3).
To compare the relative ability of homogenates prepared from AD and PA cases to seed Aβ deposition in these mice, lysates from the pre-frontal cortex of the AD, PA, and NDC were injected into the cerebral ventricles of newborn mice at P0 (20,32,33). We con rmed that the lysates contained Aβ by immunoblotting and ELISA (Fig. S1d and Table 1). Aβ was detectable by immunoblot in each of the lysates used for injection (Fig. S1d), and enzyme-linked immunosorbent assay (ELISA) measurements con rmed high levels of SDS-soluble and formic acid (FA) soluble Aβ42 in each brain lysate (Table 1). We hypothesized that injection of these lysates in P0 mice would result in widespread dispersion of the Aβ or tau seeds, and potentially offer the best chance to extensively alter the type of pathology that would be induced by seeding.
To assess amyloid pathology, hemibrains of seeded mice were stained with a pan-Aβ antibody and Thio-S. In mice injected with either AD or PA lysates, we observed signi cant induction of Aβ deposition, comprised primarily of diffuse deposits throughout the cortex and hippocampus, with some vascular deposition in the pia surrounding the cortex and within the hippocampal ssure (Fig. 1b, c). Mice injected with homogenates from NDC 1 and 2, and mice that were not injected, had little Aβ pathology by 12 months of age (Fig. 1b, c). Examples of cases with the most severe pathology are shown in Figure 1, while images of animals with the least pathology are shown in Figure S4. In all cases, however, the induced Aβ pathology exhibited a diffuse, Thio-S negative morphology (Fig. S5). The levels of induced Aβ pathology in mice that co-expressed APPsi and Tau-P301L were similar to that of mice that expressed only APPsi (Fig. 1b versus c, S4a versus b). Mice that were transgenic for only Tau-P301L, or were nontransgenic, showed no evidence of Aβ deposition (data not shown). To compare the data across all of the mice examined, Aβ pathology was qualitatively scored by three independent observers (Fig. 1d, e). At 12 months, only PA 4, was scored as having a level of Aβ pathology that approached mice injected with the AD cases; both of which exhibited widespread, diffuse pathology (Fig. 1b, c). Aβ pathology in mice seeded by PA 1, 2, and 3 was scored as less abundant than in mice seeded by the AD cases (Fig. 1d, e). We con rmed the high seeding activity of AD 1 and 2, and PA 4 by assessing Aβ pathology at 9 months post-injection. Although the severity of pathology varied, multiple animals injected with each of these lysates exhibited Aβ deposition (Fig. S6a). Notably, dilution of the brain lysates by 10-fold prior to injection greatly diminished seeding activity (Fig. S6b). To further examine the relative seeding activity of PA lysates, we injected newborn PrP.APPsi mice with each of the 4 PA lysates and NDC 1, and analyzed pathology at 18 months post-injection. Although the mice injected with the PA lysates appeared to have more severe Aβ deposition, we also observed some signi cant Aβ pathology in animals injected with the NDC lysate (Fig S7). Whether this outcome was due to some small amount of Aβ seed in the NDC 1 will require further study. Collectively, these studies demonstrated that AD and PA brains contain misfolded forms of Aβ that preferentially seed diffuse Aβ pathology.
Two of the donor cases were identi ed as having CAA pathology, AD 1 and PA 3, and were scored 1+ and 2+-3+, respectively (Table 1). To determine if CAA pathology was seeded in the recipient mice, we stained sections with Thio-S and searched for evidence of vascular amyloid. In mice seeded with PA 3, we observed Thio-S positive, vascular pathology in 6 of the 7 recipient mice however, the incidence of this pathology was limited (Fig. S8). Although AD 1 also contained CAA pathology, we did not observe Thio-S positive, vascular pathology in mice seeded by this homogenate. These ndings suggest that it may be possible to selectively increase CAA using seeding, but enhancing such pathology may require puri cation of cerebral vessels before preparation of the seeds. We biochemically con rmed our histological data by analyzing Aβ levels from sequentially extracted brain lysates using C-terminal speci c antibodies in sandwich ELISAs to measure Aβ40 and Aβ42 speci cally. In our Aβ ELISAs the detection limit was approximately 0.04 pmol/g. In the brains of all seeded mice, the levels of Aβ40, both SDS-soluble and FA were 10-100 fold lower than that of Aβ42 ( Fig. 2a, b). Consistent with the presence of abundant diffuse Aβ pathology, we detected elevated levels of Aβ42 in SDS-soluble fractions from mice seeded with AD 1 and 2, and PA 4 (Fig. 2c). Somewhat surprisingly, the levels of Aβ42 in the FA-soluble fractions were similar to that of the SDS fraction despite the absence of cored Aβ deposits ( Fig. 2c and d). Notably, there was considerable variation in the levels of Aβ in these seeded animals, which rendered relatively few indications of statistical differences (Fig. 2eh). For Aβ40 measures, the only instance in which the seeded mice had levels that were higher than uninjected controls, or mice injected with NDC lysate, were mice seeded by AD 1 (Fig. 2e, f). For measurements of SDS-soluble Aβ42, the brains of mice seeded with AD 1, 2, and PA 4 were the only examples in which the levels were statistically higher than the levels in mice seeded with NDC 1 and 2, or uninjected mice (Fig. 2g). In the FA-soluble fractions, only the AD lysates possessed higher levels of Aβ42 than the brains of mice seeded with the two NDC lysates or uninjected mice (Fig. 2h). Collectively, these ndings demonstrated that both AD and PA brains have the potential to seed diffuse Aβ pathology, with the AD brains appearing to be slightly more potent.
To determine whether the injection of these lysates into the brains of newborn PrP.APPsi/iTau-P301L mice also induced tau pathology, we stained the brains of the seeded animals with antibodies speci c for phosphorylated tau (AT8) and misfolded tau (MC1). Immunostaining with the antibody CP27, which is speci c for human tau, con rmed the presence of human P301L tau in the bigenic mice, but we observed no obvious reactivity with AT8 or MC1 antibodies (Fig. S9).
To determine if the induction of amyloid would secondarily induce αSyn pathology, we examined the seeding activity of individuals with AD and LBD pathology. To characterize the pathology of these AD, AD/LBD and NDC brains, we stained sections with a pan-Aβ antibody (Fig. 3a), Thio-S (Fig. 3b), αSyn antibody (Fig. 3c), and a phospho-speci c tau antibody (Fig. 3d). Both AD 3 and 4 contained numerous amyloid deposits that were compact and Thio-S positive (Fig. 3a, b), had widespread tau positive inclusions (Fig. 3d), and lacked αSyn pathology (Fig. 3c). AD 3 showed striking Thio-S positive amyloid staining surrounding vessels, consistent with a CAA score of 3+ (Fig. 3b, Table 1). We also observed CAA staining with AD/LBD 1 (Fig. 3b). The AD/LBD cases showed modest Aβ pathology with some αSyn pathology and sparse tau deposition (Fig. 3a, c, d). NDC 3 contained some amyloid pathology, but both lacked αSyn and tau pathology (Fig. 3).
PrP.APPsi/M20 and PrP.APPsi mice injected with AD brains developed widespread, robust Aβ pathology by 12 months post-injection ( Fig. 4a and b). Aβ deposition was primarily diffuse and Thio-S negative ( Fig  S10). There was no difference in Aβ seeding capacity between PrP.APPsi/M20 and PrP.APPsi mice. Interestingly, AD/LBD 1 and 2 also promoted deposition of Aβ in mice expressing APPsi with and without αSyn (Fig. 4), indicating that the AD/LBD brains contained considerable levels of Aβ seeds despite a much lower burden of Aβ pathology (Fig. 3). None of the injected animals developed appreciable αSyn pathology (data not shown). Although two of the donor cases contained signi cant CAA pathology, CAA was not seeded in the recipient mice (Fig S10).
Seeding diffuse Aβ deposition does not impair performance in a fear-conditioning cognitive task To assess whether the diffuse Aβ pathology seeded by AD brain produces cognitive de cits, an additional cohort of PrP.APPsi mice were seeded with AD 1 and 2, or NDC 1, and then were aged 12 months before behavioral testing in a fear-conditioning paradigm. Control groups were mice that were not injected with brain lysate and nontransgenic littermate mice that were either injected with the same lysates or uninjected. The performance of the animals in the contextual fear memory test was conducted twice. In the rst round of testing, the performance of the seeded mice was not statistically different from that of the control uninjected mice (Fig. 5a). Overall, in this rst round of testing, the percentage of time animals that exhibited freezing behavior was relatively low. In the second round of testing, we expected to reinforce memory of the adverse cue (context or tone) and we observed that re-tested mice showed increased levels of freezing behavior (Fig. 5b); however, there was still no statistically signi cant difference in the performance of the seeded mice relative to any control. These ndings indicate that the diffuse Aβ pathology seeded by these homogenates of human AD brain does not produce a meaningful impairment in memory systems involved fear-conditioning memory tasks. Assessment of Aβ seeding e ciency in APPsi mice.
To examine the relative seeding e ciency of different types of Aβ pathology in PrP.APPsi mice, we used an approach in which homogenates from the brains of 4 different lines of aged APP and APP/PS1 mice were used to seed accelerated Aβ deposition (described in Table 2). At these advanced ages, each of the mice used to produce inoculum had substantial Aβ pathology (Online Resource 2, Fig. S11). Following the strategy used above, we injected the homogenates containing Aβ seeds into newborn mice on neonatal day 0, which were then aged to 9, 12, or 15 months before euthanasia and neuropathological analysis by Campbell Switzer silver stain and Thio-S staining. The type of pathology noted and the relative abundance score are noted at the bottom of each column of images in Figure 6. In all of the PrP.APPsi mice injected with brain homogenate seeds from mice that primarily exhibited cored plaques, we observed an obvious shift in the type of Aβ deposits in recipient mice to match the pathology found in the donor mice used for seeding (Fig. 6). Self-seeding of newborn PrP.APPsi mice with brain homogenates from aged PrP.APPsi mice produced diffuse deposits (Fig. 6), which was the predominant form of Aβ pathology in the PrP.APPsi animal used to generate the seeding homogenate (Online Resource 2, Fig. S11). Similarly, the predominant pathology in the ~25 month old Tet.MoAβ animal used to prepare seeding homogenate was diffuse Aβ deposits (Online Resource 2, Fig. S11)(38), and brain homogenates from this animal e ciently seeded the deposition of human Aβ in the PrP.APPsi mice to produce diffuse deposits (Fig. 6). PrP.APPsi mice seeded with homogenates from donors that had primarily cored deposits had lower Aβ burden scores, but importantly we observed a shift in the neuropathologic features of the recipient mice to a much higher incidence of cored deposits (Fig. 6). Interestingly, the morphology of deposits in the seeded PrP.APPsi mice was not quite identical to the source PrP.HuAβ/PS1mice in that the seeded deposits were smaller and appeared to be more compact than the donor line. These small dense deposits were Thio-S positive (Fig. 6). Similar small dense core deposits that were Thio-S positive were observed in the cortex of PrP.APPsi mice injected with brain homogenates from the PrP.MoAβ/PS1 mice (Fig. 6). Collectively, these ndings demonstrate the relative ease with which diffuse Aβ pathology can be induced in PrP.APPsi mice.

Discussion
We investigated whether injection of neonatal APP, APP/tau, or APP/αSyn mice with AD or AD/LBD brain homogenates could be used as a paradigm to generate mice that model the mixed pathology associated with each disease. Despite the widespread induction of diffuse Aβ deposition by injection of four different AD brain lysates, and the presence of tau seeds in these lysates, tau pathology was not induced.
For comparison, we also injected PA brain lysates, nding a similar induction of diffuse Aβ deposition without induction of tau pathology. Similarly, injection of AD/LBD brain lysates induced diffuse Aβ deposition, but αSyn pathology was not induced. In all cases, the induced Aβ pathology exhibited a diffuse, Thio-S negative, morphology. PrP.APPsi mice injected with AD brain lysates that develop diffuse Aβ pathology showed no signi cant impairment in a fear-conditioning cognitive task. Our ndings suggest that seeding the brains of neonatal transgenic mice with AD, AD/LBD, and PA brain homogenates can e ciently induce diffuse Aβ deposition but neither AD nor AD/LBD brains seeded concurrent tau or αSyn pathology.
The most common route of administering Aβ seeding preparations is by stereotaxic injection into the hippocampus of adult APP transgenic mice [reviewed in (3)]. In host mice that develop amyloid pathology at relatively young ages, the induced pathology created by seeding may be localized to the site of injection and overlying cortex (3); whereas, in models that do not develop deposits until late in life, the induced pathology may be more wide-spread [reviewed in (3)]. The PrP.APPsi host we used in the present study develops pathology on its own between 12-14 months of age. In testing the approach of injecting Aβ seeds into newborn mice from this model, we aspired to attain a widespread distribution of seeding material and induce Aβ, tau, or αSyn pathology throughout the brain. This approach replicates previous studies where P0 injection of adeno-associated virus resulted in widespread distribution in the brain (39).
Injection of newborn hamsters with scrapie prions has been shown to accelerate onset of prion disease (40). Additionally, we have observed that intracerebral injection of newborn A53T αSyn mice with brain lysates from multiple system atrophy donors induces αSyn pathology and motor impairment (20), and that intraspinal injection of newborn mice expressing mutant superoxide dismutase (SOD1) can accelerate the onset of paralysis and pathology (41,42). Thus, in performing newborn injections our goal was to initiate pathological cascades as early as possible in order to determine whether there may be distinct synergies between Aβ and tau, or Aβ and αSyn, which can be elaborated by seeding.
We recognize that there are a large number of potential mouse models that could have been used for these studies. The APP mouse model we chose to us for these studies was selected because PrP.APPsi mice do not inherently develop Aβ pathology until 12-14 months of age and because this model can exhibit a full spectrum of Aβ pathology including cored-neuritic, diffuse, and vascular deposition (see Supplemental Fig. S2). The tau model we chose to use expresses human P301L tau at relatively low levels and does not develop tau pathology on its own. When paired with mice that express the tetracycline transactivator in the rTg4510 model these develop a robust tau pathology (28). In paradigms in which tau expression in rTg4510 mice is induced early and then suppressed by doxycycline, the low level of "leaky" tau expression in this model is su cient to sustain neuro brillary tangle pathology (28). In our paradigm, we asked whether neonatal seeding of this tau model could induce a sustained pathology in the same way that early expression of the transgene at high levels produced sustained pathology. Alternatively, we were interested to determine whether we could detect any synergy between concurrent Aβ pathology and tau seeding. The αSyn model we chose to use, expresses human WT αSyn at levels that do not cause pathology (31). WT αSyn (M20) mice can be seeded when injected with high levels of puri ed αSyn brils, but are not easily seeded by human brain homogenates (20). Our goal in choosing the M20 WT αSyn mice was to develop a model of human αSyn pathology and assess whether concurrent Aβ pathology could synergize αSyn seeding.
Our method of seed preparation followed commonly used protocols where we injected a soluble fraction; frontal cortex homogenized in PBS (10% w/v) followed by sonication and low speed centrifugation, as previously described (32,43,44). By this method, we expected sonication to fragment all types of seeds with a mixture of seeds remaining in suspension after low-speed centrifugation. As we have observed here, the most common outcome of Aβ seeding with similar preparations is induction of diffuse Aβ pathology [reviewed in (3)].
In order to e ciently seed mixed pathology, it may be necessary to optimize preparations of each type of seed independently. Interestingly, injection of the supernatant and pellet of fractionated APP23 brain homogenate resulted in morphologically different Aβ deposits, with the supernatant fraction seeding diffuse, Congo Red negative Aβ and the pellet seeding deposits similar to the total lysate, a mixture of diffuse and punctate, Congo Red positive deposits (45). Several studies have shown induction of tau pathology after injection of synthetic tau brils, brain extract from mutant P301S tau mouse, or human AD, corticobasal degeneration, and progressive supranuclear palsy brain lysates (46)(47)(48)(49)(50)(51)(52). In these studies, the human brain lysates were enriched for tau seeds by sequential fractionation with the addition of sucrose and/or sarkosyl, creating fractions that contain potent tau seeds. Similarly, injection of the fractionated homogenate from brain lysates of individuals with LBD resulted in the induction of αSyn, indicating that with enrichment it is possible to seed αSyn pathology directly from human brain (8, 53,54). Because one of our goals was to determine whether we could detect synergy between Aβ and tau or αSyn pathology, we chose to avoid enriching for any particular type of seeding activity. The absence of tau or αSyn pathology in our seeded models indicates that the type of Aβ deposition we generated did not synergize to induce, or exacerbate, tau or αSyn pathology.
We demonstrated that the Aβ seeding activity of brain lysates from PA cases was similar to AD and AD/LBD cases. Although mice seeded by PA brains scored as having less amyloid positivity at 12 months post-injection than mice injected with AD brain lysates, the difference between PA and AD brain seeding activity was less evident by 18 months post-injection. Our study was not powered or designed to determine whether homogenates from AD and PA brains have quantitative differences in seeding activity.
All three of the PA cases tested were able to seed amyloid pathology, supporting the hypothesis that the Aβ pathology in PA is similar to AD (13).
Interestingly, several of the brain lysates we used were from tissues that exhibited relatively high levels of CAA pathology (see Table 1). PA 3 and AD 3 were both scored as having the highest levels of CAA, with signi cant CAA in AD/LBD 1 and 2. In mice injected with PA 3, we observed modest seeding of parenchymal CAA pathology in the host mice but in all other mice the only obvious vascular deposition was meningeal. Augmented CAA pathology has been observed in APP23 mice, which develop CAA, by intraperitoneal injection of APP23 lysates (55). CAA pathology was also induced in CRND8 mice by injection of synthetic Aβ42 oligomers, generated in the presence of anionic micelles composed of fatty acids (56). Recent studies have documented early onset cerebrovascular Aβ pathology in individuals receiving dura mater transplants, suggesting that CAA may be transmitted iatrogenically (57). Collectively, these results suggest that CAA could be a distinct conformer of Aβ that may be independently propagated.
Not all Aβ conformers seeded as pro ciently as vascular Aβ. Tissues from AD and AD/LBD brains that were used to prepare these lysates contained both diffuse and compact Aβ deposits, with signi cant levels of formic acid extractable Aβ. Yet, even in mice aged to 18 months of age, diffuse Aβ deposits were the dominant form of pathology. These results mirror similar studies in which injection of human brain homogenates resulted in induction of diffuse Aβ pathology, with relatively few ThioS or congophilic compact deposits (17,39,58,59). The lack of conversion of the diffuse deposits to cored plaques indicates different types of amyloid deposition are not freely interchangeable, supporting the idea that different types of Aβ pathology arise from different amyloid strains with distinct seeding capabilities.
This conclusion is also supported by our studies of Aβ seeding in which transgenic mice were used as the seed source. Newborn PrP.APPsi mice injected with transgenic mouse brain homogenates prepared from mice that exhibit high levels of diffuse Aβ pathology produced a robust induction of diffuse Aβ deposition by 12 months post-injection. By contrast, brain homogenates prepared from mice that predominantly produce cored, Thio-S positive, Aβ deposits induced minimal Aβ deposition by 12 months.
These results are consistent with previous studies with APP/PS1 and APP23 mice. Injection of APPPS1 mice that typically develop compact, punctate plaques, with APP23 seeds results in a mixture of diffuse, lamentous Aβ as well as compact plaques (32,60). Injection of APP23 mice that typically develop mixed pathology, both core and diffuse plaques, with APPPS1 homogenate results in plaques that are more diffuse than the punctate deposits seen in APPPS1 mice injected with APPPS1 homogenate (32,60). Thus, although it is possible to seed cored deposits in APP mice (32,45,(61)(62)(63), it appears that diffuse Aβ pathology is more easily seeded. Together these results suggest that Aβ plaques differ in their morphology, seeding ability, and impact on cognitive function. Current amyloid therapies aim to reduce amyloid deposition without much consideration of the type of plaque that is being targeted. Reduction of diffuse amyloid may at rst appear to be bene cial, since it appears to be the dominate seeding strain, however, it may not be the most pathological strain. Cognitive behavior studies of the seeded APPsi mice reported here, and of a non-seeded Bri-Aβ42 model that also shows primarily diffuse Aβ pathology (64), failed to associate diffuse amyloid with reduced performance in a fear-conditioning memory task. If diffuse Aβ pathology is less damaging, then therapies that only reduce diffuse amyloid would not yield much therapeutic bene t.

Limitations
There are many potential transgenic mouse models available that could be used for the type of seeding studies we conducted. The APP model we used appears to have performed as expected, but the lack of tau or αSyn seeding may be a consequence of the host model chosen to express human tau and αSyn or the dose of the tau and αSyn seeds in the seeding homogenate. The expression of human P301L tau in the tet-tau model used here may have been too low to sustain propagation of tau seeds. In the time between when these studies were initiated and publication, a study of seeding in the JNPL3 model of human P301L tau, which expresses tau at levels similar to the tet-tau model used here, reported poor induction of tau pathology after seeding (36). A higher dose of enriched tau seeds may have been more effective. Likewise, a higher dose of αSyn seeds may be required to induce human WT αSyn to produce pathology. Our study demonstrates that using seeding to produce novel models of mixed pathology will require further re nement in methodology to achieve the desired outcome.

Conclusions
In conclusion, we assessed the relative seeding activity of brain lysates prepared from AD, AD/LBD, and PA brain by injecting neonatal transgenic hosts. We found that P0 injection of transgenic mice expressing humanized APP with Aβ-containing lysates, resulted in robust and widespread induction of Aβ pathology.
The induced Aβ pathology was diffuse, which has been reported in other seeding studies where adult animals were the recipients of injections (3). Our results are consistent with earlier reports in nding that diffuse Aβ is easily seeded by crude homogenates of human brain. Seeds for diffuse Aβ were prevalent in human AD, AD/LBD, and PA brain. The diffuse Aβ pathology induced in these models was not accompanied by secondary tau or αSyn pathology in bigenic mice co-expressing APP/tau or APP/aSyn.
Mice with seeded diffuse Aβ pathology were not impaired in a fear-conditioning memory task. Further re nements in methodology will be required to e ciently seed models of mixed AD and AD/LBD pathology.  Tables   Table 1. List of human brain samples used in this study.

Figures
Mean percentage freezing ± standard error of the mean exhibited by PrP.APPsi and nontransgenic (NTg) littermates injected with AD cases 1 or 2 and NDC case 7. Uninjected mice were the control group. (a) Context and tone paradigm. (b) A subset of mice were re-tested the following week, context and tone test. n = 6-8/Tg, n = 12-16/NTg mice per group.