Deletion of the Homocysteine Thiolactone Detoxifying Enzyme Bleomycin Hydrolase, in Mice, Causes Memory and Neurological Deficits and Worsens Alzheimer’s Disease-Related Behavioral and Biochemical Traits in the 5xFAD Model of Alzheimer’s Disease

Background: Bleomycin hydrolase (BLMH), a homocysteine (Hcy)-thiolactone detoxifying enzyme, is attenuated in Alzheimer’s disease (AD) brains. Blmh loss causes astrogliosis in mice while the loss of histone demethylase Phf8, which controls mTOR signaling, causes neuropathy in mice and humans. Objective: To examine how Blmh gene deletion affects the Phf8/H4K20me1/mTOR/autophagy pathway, amyloid-β (Aβ) accumulation, and cognitive/neuromotor performance in mice. Methods: We generated a new mouse model of AD, the Blmh-/-5xFAD mouse. Behavioral assessments were conducted by cognitive/neuromotor testing. Blmh and Phf8 genes were silenced in mouse neuroblastoma N2a-APPswe cells by RNA interference. mTOR- and autophagy-related proteins, and AβPP were quantified by western blotting and the corresponding mRNAs by RT-qPCR. Aβ was quantified by western blotting (brains) and by confocal microscopy (cells). Results: Behavioral testing showed cognitive/neuromotor deficits in Blmh-/- and Blmh-/-5xFAD mice. Phf8 was transcriptionally downregulated in Blmh-/- and Blmh-/-5xFAD brains. H4K20me1, mTOR, phospho-mTOR, and AβPP were upregulated while autophagy markers Becn1, Atg5, and Atg7 were downregulated in Blmh-/- and Blmh-/-5xFAD brains. Aβ was elevated in Blmh-/-5xFAD brains. These biochemical changes were recapitulated in Blmh-silenced N2a-APPswe cells, which also showed increased H4K20me1-mTOR promoter binding and impaired autophagy flux (Lc3-I, Lc3-II, p62). Phf8-silencing or treatments with Hcy-thiolactone or N-Hcy-protein, metabolites elevated in Blmh-/- mice, induced biochemical changes in N2a-APPswe cells like those induced by the Blmh-silencing. However, Phf8-silencing elevated Aβ without affecting AβPP. Conclusions: Our findings show that Blmh interacts with AβPP and the Phf8/H4K20me1/mTOR/autophagy pathway, and that disruption of those interactions causes Aβ accumulation and cognitive/neuromotor deficits.


INTRODUCTION
Bleomycin hydrolase (Blmh), named for its ability to deaminate and inactivate the anticancer glycopeptide drug bleomycin, is a thiol-dependent cytoplasmic aminopeptidase expressed in human and rodent organs, including the brain [1,2].In addition to the aminopeptidase activity, Blmh has a thiolactonase activity and takes part in homocysteine (Hcy) metabolism by detoxifying Hcy-thiolactone [3,4].
BLMH has been linked to AD and Huntington's disease.Specifically, BLMH has the ability to process amyloid-␤ protein precursor (A␤PP) to amyloid-␤ (A␤) [21] and to further process A␤ [22].BLMH has also the ability to generate N-terminal fragments of huntingtin, thought to be important mediators of the pathogenesis of Huntington's disease [23].In the human brain, BLMH is localized in neocortical neurons and in dystrophic neurites of senile plaques [24].A single nucleotide polymorphism in human BLMH gene, resulting in I443 V substitution in the BLMH protein, is associated with an increased risk of AD [25,26]; however, no association was reported in other studies [27][28][29].
The Hcy-thiolactonase and aminopeptidase activities of BLMH are decreased in brains of AD patients, suggesting that the attenuated BLMH activity could contribute to the pathology of AD [30].In mice, deletion of the Blmh gene causes astrogliosis and behavioral changes [31].Furthermore, Blmh −/− mice exhibit diminished ability to detoxify Hcy-thiolactone, which elevates brain Hcythiolactone levels, and increases neurotoxicity of Hcy-thiolactone injections [4].Studies of Blmh −/− mouse brain proteome demonstrated that Blmh interacts with diverse cellular processes, such as synaptic plasticity, cytoskeleton dynamics, cell cycle, energy metabolism, and antioxidant defenses that are essential for brain homeostasis [9].Collectively, these findings suggest that Blmh plays a key role in the central nervous system (CNS).
Plant homeodomain finger protein 8 (PHF8) has been identified as one of the X chromosome genes linked to intellectual disability syndrome, autism spectrum disorder, attention deficit hyperactivity disorder [32], and severe mental retardation [33].PHF8 is a histone demethylase that can demethylate H4K20me1, H3K9me2/me1, and H3K27me2.Demethylation of H4K20me1 by PHF8 is important for supporting homeostasis of mTOR signaling.The phenotype of human PHF8 deficiency has been replicated in Phf8 −/− mice, which show impaired hippocampal long-term potentiation and behavioral deficits in learning and memory [34].
In the present work we examined the role of Blmh in the CNS by studying behavioral and biochemical consequences of Blmh gene deletion in mice.Since dysregulated mTOR signaling and autophagy have been implicated in A␤ accumulation in AD brains [35][36][37][38], and H4K20me1 demethylation by PHF8 is important for maintaining homeostasis of mTOR signaling, we examined how these processes are affected in brains of Blmh −/− versus Blmh +/+ mice as well as transgenic Blmh −/− 5xFAD versus Blmh +/+ 5xFAD mice.We also examined how biochemical changes in these processes and in A␤PP/A␤ expression relate to behavioral performance in Blmh −/− 5xFAD mice.We studied underlying molecular mechanisms by manipulating Blmh or Phf8 expression or Hcy-thiolactone and N-Hcy-protein levels in mouse neuroblastoma N2a-APPswe cells.

Behavioral testing
Novel Object Recognition test.NOR is a test of recognition memory [44].The test was conducted in two sessions, divided by a 6-h intersession interval.During the first session (familiarization session), the animal was free to explore two similar objects, and during the second session (test session), one of the objects was replaced by a novel, unfamiliar object.No habituation phase was performed.A minimal exploration time for both objects during both the familiarization and test phase (∼20 s) was used, with a maximal time of 10 min to reach the criterion.Mice were evaluated in a white plastic box (33×33×20 cm).W used objects that differ in shape and texture: towers of Lego bricks (8-cm high and 3.2-cm wide, built-in blue, yellow, red, and green bricks) and Falcon tissue culture flasks filled with sand (9.5 cm high, 2.5 cm deep and 5.5 cm wide, transparent plastic with a yellow bottle cap).We scored object exploration whenever the mouse sniffed the object or touched the object while looking at it (i.e., when the distance between the nose and the object was less than 2 cm).Climbing onto the object (unless the mouse sniffs the object it has climbed on) or chewing the object did not qualify as exploration.
Hindlimb test.The hindlimb clasping test is used to assess neurodegeneration in mouse models [45].For this test, mice were suspended by the base of the tail and videotaped for 10 s.Three separate trials were taken over three consecutive days.Hindlimb clasping was scored from 0 to 3 : 0 = hindlimbs splayed outward and away from the abdomen, 1 = one hindlimb retracted inwards towards the abdomen for at least 50% of the observation period, 2 = both hindlimbs partially retracted inwards towards the abdomen for at least 50% of the observation period, 3 = both hindlimbs completely retracted inwards towards the abdomen for at least 50% of the observation period.Hindlimb clasping scores were added together for the three separate trials.
Ledge test.The ledge test is used to assess motor deficits in rodent models of CNS disorders [46].Typically, mice walk along the ledge of a cage and try to descend back into the cage.Three separate trials were taken for each mouse.Ledge test was scored from 0 to 3 points: 0 = a mouse walked along the ledge without slipping and lowered itself back into the cage using paws; 1 = the mouse lost its footing during walking along the ledge but otherwise appeared coordinated; 2 = the mouse did not effectively use its hind legs and landed on its head rather than paws when descending into the cage; 3 = the mouse fell of the ledge or was shaking, barely moving.
Cylinder test.The cylinder test is used to assess sensorimotor function in rodent models of CNS disorders.A mouse was placed in the transparent 500 ml plastic cylinder.The number of times the mouse reared up and touched the cylinder wall during a period of 3 min was counted.A rear is defined as a vertical movement with both forelimbs off the floor so that the mouse is standing only on its hindlimbs.At the end of 3 min, the mouse was removed and placed back into its home cage.Because spontaneous activity in the cylinder is affected by repeated testing resulting in reduced activity over time, mice were only once in their lifetime.

Brain protein extraction
Mice were euthanized by CO 2 inhalation, the brains collected and frozen on dry ice.Frozen brains were pulverized with dry ice using a mortar and pestle and stored at -80 • C. Proteins were extracted from the pulverized brains (50 ± 5 mg; 30 ± 3 mg brain was used for A␤ analyses) using RIPA buffer (4 v/w, containing protease and phosphatase inhibitors) with sonication (Bandelin SONOPLUS HD 2070) on wet ice (three sets of five 1-s strokes with 1 min cooling interval between strokes).Brain extracts were clarified by centrifugation (15,000 g, 30 min, 4 • C) and clear supernatants having 8-12 mg protein/mL were collected (RIPA-soluble fraction).Protein concentrations were measured with BCA kit (Thermo Scientific).
After cells reached 70-80% confluency, the monolayers were washed twice with PBS and overlaid with DMEM medium without methionine (Thermo Scientific), supplemented with 5% dialyzed fetal bovine serum (MilliporeSigma) and non-essential amino acids.L-Hcy-thiolactone (MilliporeSigma) or N-Hcy-protein, prepared as described in [49], were added (at concentrations indicated in figure legends) and the cultures were incubated at 37 • C in 5% CO 2 atmosphere for 24 h.
For western blots analyses of A␤, brain protein extracts (2 L) were separated on 10% Tricine gels, and then transferred (0.5 A, 25 V 10 min) onto 22 m PVDF membranes (Bio-Rad).The membranes were washed 3 times with 1x TBST and then blocked with 5% bovine serum albumin for 1 h at RT.After blocking, membranes were washed 3 times with 1x TBST and then incubated with primary anti-A␤ antibody (D54D2, CS #8243).Membranes were washed 3 times with 1x TBS-T and incubated with anti-rabbit IgG HRP-linked antibodies (CS#7074) for 1 h at RT. Signals were collected using clarity Max Western ECL Substrate (Bio-Rad) and GeneGnome XRQ -Chemiluminescence imaging (Syngene).

RNA isolation, cDNA synthesis, RT-qPCR analysis
Total RNA was isolated using Trizol reagent (MilliporeSigma).cDNA synthesis was conducted using Revert Aid First cDNA Synthesis Kit (Thermo Fisher Scientific) according to manufacturer's protocol.Nucleic acid concentration was measured using NanoDrop (Thermo Fisher Scientific).RT-qPCR was performed with SYBR Green Mix and CFX96 thermocycler (Bio-Rad).The 2 (− Ct) method was used to calculate the relative expression levels [50].Data analysis was performed with the CFX Manager™ Software, Microsoft Excel, and GraphPad Prism7.RT-qPCR primer sequences are listed in Supplementary Table 1.

Chromatin immunoprecipitation assay
For the CHIP assays, CUT&RUN Assay Kit #86652 (Cell Signaling Technology, Danvers, MA, USA) was used following the manufacturer's protocol.Each assay was done in triplicates.cells.Cells (100 000/ assay) were trypsinized, harvested, washed 3x in ice-cold PBS, and bound to concanavalin Acoated magnetic beads for 5 min, RT.Cells were then incubated (4 h, 4 • C) with 2.5 g of anti-PHF8 anti-body (Abcam, ab36068) or anti-H4K20me1 antibody (Abcam, ab177188) in the antibody-binding buffer plus digitonin that permeabilizes cells.Next, cells are treated with pAG-MNase (1 h, 4 • C), washed, and treated with CaCl2 to activate DNA digestion (0.5 h, 4 • C).Cells were then treated with the stop buffer and spike-in DNA was added for each reaction for signal normalization, and incubated (30 min, 37 • C).Released DNA fragments were purified using DNA Purification Buffers and Spin Columns (CS #14209) and quantified by RT-qPCR using primers targeting the promoter, upstream, and downstream regions of the mTOR gene (Supplementary Table 1).

Confocal microscopy, Aβ staining in N2a-APPswe cells
Mouse neuroblastoma N2a-APPswe cells were cultured in Millicell EZ SLIDE 8-well glass slides (Merck).After 24 h treatments, cells were washed with PBS (3 times, 10 min each) and fixed with 4% PFA (Sigma-Aldrich) (37 • C, 15 min).After fixation, cells were again washed 3 times with PBS buffer and permeabilized in 0.1% Triton X-100 solution (RT, 20 min), blocked with 0.1% bovine serum albumin (RT, 1 h), and incubated with anti-A␤ antibody (CS #8243; 4 • C, 16 h).Cells were then washed 3 times with PBS and stained with secondary antibody Goat Anti-Rabbit IgG H&L (Alexa Fluor® 488) (Abcam, ab150077; RT, 1 h) to detect A␤.DAPI (Vector Laboratories) was used to visualize nuclei.Fluorescence signals were detected by using a Zeiss LSM 880 confocal microscope with a 488 nm filter for the Alexa Fluor® 488 (A␤) and 420-480 nm filter for DAPI, taking a z stack of 20-30 sections with an interval of 0.66 m and a range of 15 m.Zeiss Plan-Apochromat X40/1.2Oil differential interference contrast objective were used for imaging.Images were quantified with the ImageJ Fiji software (NIH).Negative controls without primary or secondary antibody, which yielded no fluorescent signals, and positive controls with an authentic A␤ 42 standard (1 g/mL medium) (Abcam, ab120301), which showed fluorescent signals, were done to verify the specificity of the assay.

Statistical analysis
Data from in vivo and in vitro experiments are reported as mean ± standard deviation (SD).Values for each experimental/treatment group were normalized to controls.Data were analyzed using one-way analysis of variance (ANOVA) with Tukey's multiple comparisons post-test using GraphPad Prism7 software (GraphPad Holdings LLC, San Diego CA, USA, https://www.graphpad.com).A two-sided unpaired Student t test was used for analysis of NOR data.

RESULTS
Blmh gene deletion impairs recognition memory, induces neurodegeneration, and sensorimotor deficits in Blmh −/− and Blmh −/− 5xFAD mice Blmh −/− mice.To examine effects of Blmh depletion on cognition and sensorimotor activity, Blmh −/− and Blmh +/+ mice fed with a control or high Met diet were assessed in the NOR, hindlimb clasping, and ledge tests.We found that 4-month-old Blmh −/− mice did not differentiate between novel and familiar objects in the NOR test, regardless of the diet (Fig. 1A, B), showing impaired recognition memory.As expected, 4-month-old control Blmh +/+ mice showed normal preference for novelty (Fig. 1A); however, the preference for novelty disappeared when the mice were fed with a high Met diet (Fig. 1B).In contrast, 2-month-old Blmh −/− mice showed significant preference for a novel object, regardless of the diet, as did 2-month-old Blmh +/+ mice (Fig. 1A, B).These findings show that a longer exposure (>2-month) is needed for the manifestation of detrimental effects of Blmh depletion or high Met diet on cognition.
Blmh −/− 5xFAD mice.We also examined effects of Blmh depletion on cognition and neuromotor activity in 1-year-old Blmh −/− 5xFAD versus Blmh +/+ 5xFAD mice.We found that Blmh −/− 5xFAD mice fed with a standard chow or high Met diet did not differentiate between novel and familiar objects in the NOR test (Fig. 1E), showing impaired recognition memory in Blmh −/− 5xFAD animals.In contrast, Blmh +/+ 5xFAD mice fed with a standard chow diet or high Met diet showed normal preference for novelty (Fig. 1E).

Blmh depletion downregulates histone demethylase Phf8 and upregulates H4K20me1 epigenetic mark in brains of Blmh
The present findings that Blmh-deleted mice showed behavioral deficits in memory that are like cognitive deficits in memory seen in Phf8-deleted mice [34] suggests that Blmh may interact with Phf8.To examine this possibility, we quantified Phf8 protein in brains of young Blmh −/− mice and their Blmh +/+ sibling controls by western blotting.We also examined how HHcy, induced by supplying 1% methionine (Met) in drinking water, affects the Blmh-Phf8 interaction.We found that Phf8 protein was significantly downregulated in brains of Blmh −/− mice versus Blmh +/+ sibling controls in mice fed with a standard chow diet (Fig. 2A).The reduced Phf8 expression in Blmh −/− versus Blmh +/+ brains was attenuated in mice fed with a high Met diet (Fig. 2A).However, Phf8 expression in Blmh −/− mice was not affected by high Met diet (Fig. 2A).In contrast, in Blmh +/+ mice, high Met diet significantly downregulated Phf8 levels (Fig. 2A), showing that the Blmh-Phf8 interaction is affected by high Met diet.
We saw a similarly upregulated A␤PP in 5and 12-month-old Blmh −/− 5xFAD mice carrying a mutated human A␤PP transgene (Supplementary Figure 1I).High Met diet had no effect on A␤PP in brains of 5-month-old Blmh −/− 5xFAD and Blmh +/+ 5xFAD mice.However, A␤PP was upregulated in brains of 12-month-old Blmh +/+ 5xFAD mice by high Met diet, which abrogated the difference in A␤PP between Blmh −/− 5xFAD versus Blmh +/+ 5xFAD animals seen in animals fed with a control diet (Supplementary Figure 1I).

Blmh controls the expression of mTOR-, autophagy-related proteins, and AβPP in N2a-APPswe cells
To elucidate the mechanism by which Blmh depletion impacts Phf8 and its downstream effects on mTOR, autophagy, and A␤PP, we first examined whether our findings in Blmh −/− mice can be recapitulated in cultured mouse neuroblastoma N2a-APPswe cells carrying a mutated human A␤PP transgene.We silenced the Blmh gene in these cells by RNA interference using two different siRNA (siRNA Blmh#1 and siRNA Blmh#2) for transfections; controls were mock-transfected or transfected with scrambled siRNA (siRNAscr).The changes in gene expression at the protein and mRNA levels in Blmh-silenced versus control cells were analyzed by western blotting and RT-qPCR using Gapdh protein and mRNA, respectively, as a reference.
We found that the Blmh protein level was significantly reduced in Blmh-silenced cells (by 80%; Supplementary Figure 2A).We also found that the histone demethylase Phf8 protein level was also significantly reduced (Supplementary Figure 2B), while the histone H4K20me1 level was significantly elevated (Supplementary Figure 2C) in Blmh-silenced cells.
To elucidate whether Blmh exerts a transcriptional control over the expression of A␤PP, mTOR-and autophagy-related proteins, we quantified by RT-qPCR mRNAs for these proteins.We found that the changes in mRNA levels in Blmh-silenced N2a-APPswe cells (Supplementary Figure 3) were like the changes in the corresponding protein levels (Supplementary Figure 2).Specifically, Blmh mRNA was significantly reduced (by 95%; Supplementary Figure 3A), as was Phf8 mRNA (Supplementary Figure 3B).mTOR mRNA was significantly upregulated (Supplementary Figure 3C) as was A␤PP mRNA (Supplementary Figure 3D) in Blmh-silenced cells, reflecting changes in the corresponding protein levels in these cells (Supplementary Figure 2A-F).mRNAs for autophagy-related proteins Atg5, Atg7, and Becn1 (Supplementary Figure 3E, F, and G, respectively).

Hcy-thiolactone and N-Hcy-protein modulate the expression of AβPP, mTOR-and autophagy-related proteins in N2a-APPswe cells
The effects of Blmh deletion on Phf8 and its downstream targets can be caused by the absence the Blmh protein or by Hcy-thiolactone and N-Hcy-protein, which are known to be elevated in Blmh −/− mice [4].To distinguish between these possibilities, we treated N2a-APPswe cells with different concentrations of Hcy-thiolactone or N-Hcy-protein.
A␤PP levels were significantly upregulated in N2a-APPswe cells treated with Hcy-thiolactone or N-Hcy-protein compared to untreated controls (Fig. 3H).
To find whether Hcy-thiolactone or N-Hcy-protein can affect the autophagy flux, we quantified Lc3 and p62 in N2a-APPswe cells treated with these metabolites.We found that treatments with Hcy-thiolactone or N-Hcy-protein reduced levels of soluble Lc3-I (Fig. 3I) and an autophagosome membrane-bound lipidated Lc3-II (Fig. 3J).The Lc3-II/Lc3-I ratio, a measure of autophagy flux, was also reduced (Fig. 3K), while p62 was upregulated (Fig. 3L).Representative images of western blots are shown in Fig. 3M.Taken together, these findings suggest that Hcythiolactone and N-Hcy-protein, metabolites that are elevated in Blmh −/− mice [4], contribute to the detrimental effects of Blmh depletion on Phf8, mTOR signaling, autophagy flux, and A␤PP in N2a-APPswe cells.

Blmh depletion by RNA interference increases H4K20me1 biding to the mTOR promoter in N2a-APPswe cells
To find whether increased levels of the histone H4K20me1 mark can promote mTOR gene expression by binding to its promoter in Blmh-depleted cells, we conducted ChIP experiments using anti-H4K20me1 monoclonal antibody in the CUT&RUN assay (Fig. 4).The Blmh gene was silenced by transfecting N2a-APPswe cells with two different Blmh-targeting siRNAs (siRNA Blmh #1 and #2).The cells were permeabilized, treated with the anti-H4K20me1 antibody and a recombinant protein A/G-tagged micrococcal nuclease.DNA fragments released form N2a-APPswe cells were quantified by RT-qPCR using primers targeting the transcription start site (TSS) of the mTOR gene as well as upstream (UP) and downstream (DOWN) regions from the TSS.
We found that the binding of H4K20me1 was significantly increased at the mTOR TSS, mTOR UP, and mTOR DOWN sites in the Blmh-silenced N2a-APPswe cells compared to controls (Fig. 4A).Importantly, we found significantly more DNA fragments from the mTOR TSS compared with the mTOR UP and mTOR DOWN sites (Fig. 4A).Control experiments showed that binding of H3K4me3 to RPL30 intron was not affected by Blmh gene silencing (Fig. 4B).These findings show that Blmh depletion promoted H4K20me1 binding at the mTOR TSS site more efficiently than at the UP and DOWN sites.
The CUT&RUN experiments using anti-Phf8 antibody showed that Blmh depletion did not affect the binding of Phf8 to the mTOR gene (not shown).

Hcy-thiolactone and N-Hcy-protein increase H4K20me1 binding to the mTOR promoter in N2a-APPswe cells
Because treatments with Hcy-thiolactone or N-Hcy-protein, metabolites that are elevated in Blmhdepleted mice [4], upregulated mTOR expression (Fig. 3C), it is likely that each of these metabolites can influence mTOR expression by promoting H4K20me1 binding at its promoter.To evaluate this contention, we conducted the CUT&RUN experiments in N2a-APPswe cells treated with Hcythiolactone or N-Hcy-protein using anti-H4K20me1 antibody and quantified the extent of H4K20me1 binding to the mTOR gene.
We found that Hcy-thiolactone significantly increased binding of H4K20me1 at the mTOR TSS site as well as at the UP, and DOWN sites [51] (Fig. 4C).N-Hcy-protein also significantly increased binding of H4K20me1 at the mTOR TSS and DOWN sites, but not at the UP site [51].Control experiments show that the binding of H3K4me3 to RPL30 intron was not affected by treatments with Hcy-thiolactone or N-Hcy-protein [51] (Fig. 4D).
Binding of Phf8 at the mTOR TSS, UP and down sites was not affected by Hcy-thiolactone or N-Hcyprotein (not shown).

Blmh depletion by RNA interference or treatments with Hcy-thiolactone and N-Hcy-protein promotes Aβ accumulation in N2a-APPswe cells
To find whether Blmh depletion affects A␤ accumulation, we silenced the Blmh gene in N2a-APPswe cells by using RNA interference and quantified A␤ by fluorescence confocal microscopy.Towards this end, we transfected N2a-APPswe cells using two different Blmh-targeting siRNAs.The cells were permeabilized, treated with anti-A␤ antibody, A␤ was visualized with a fluorescent secondary antibody (Fig. 5A) and quantified (Fig. 5B).We found that the silencing of Blmh gene, which reduced Blmh protein levels by 80% (Supplementary Figure 2A), significantly increased the area and average size of fluorescent A␤ puncta in Blmh siRNA-treated cells compared to siRNAscr-treated cells or mock-treated cell without siRNA (Fig. 5B).
Because Blmh depletion elevates Hcy-thiolactone and N-Hcy-protein in mice [4], we next examined whether each of these metabolites can induce A␤ accumulation in N2a-APPswe cells.We found significantly increased area of fluorescent A␤ puncta in cells treated with Hcy-thiolactone or N-Hcy-protein, compared to untreated controls (Fig. 5C, D).However, while treatments with Hcy-thiolactone led to increased average size of the fluorescent A␤ puncta, treatments with N-Hcy-protein did not (Fig. 5C,  D), suggesting different effects of these metabolites on the structure of A␤ aggregates.These findings suggest that Hcy-thiolactone and N-Hcy-protein promote accumulation of A␤ induced by Blmh depletion.

Blmh gene deletion increases Aβ accumulation in brains of 5xFAD mice
To examine if Blmh deletion can promote A␤ accumulation in the mouse brain, we generated Blmh −/− 5xFAD mice and their Blmh +/+ 5xFAD siblings by crossing Blmh −/− mice with A␤overproducing 5xFAD animals.We prepared SDSsoluble and FA-soluble A␤ fractions, which have most of the total A␤ [47], as well as an A␤ fraction extractable with a RIPA buffer, from brains of 5and 12-month-old mice.A␤ was quantified in these brain extracts by western blotting using monoclonal anti-A␤ antibody.Representative western blots are shown in Fig. 6A, B, C, G, and H.

Phf8 depletion upregulates Aβ but not AβPP in N2a-APPswe cells
The findings that Phf8 expression was significantly reduced in the brains of Blmh −/− (Fig. 2A) and Blmh −/− 5xFAD mice (Supplementary Figure 1A) and in the Blmh-silenced (Supplementary Figures 2B  and 3B) or Hcy metabolite-treated (Fig. 3A) mouse neuroblastoma N2a-APPswe cells, suggested that Phf8 depletion by itself can affect biochemical pathways leading to A␤ accumulation.To examine this, we depleted Phf8 in N2a-APPswe cells by RNA interference and quantified by western blotting proteins that we found to be affected in the Blmh −/− (Fig. 2) and Blmh −/− 5xFAD (Supplementary Figure 1A) mouse brains.
We also quantified A␤ by fluorescence confocal microscopy and found that A␤ was upregulated in the Phf8-silenced cells, manifested by significantly increased average size and signal intensity of fluorescent A␤ puncta compared to controls without siRNA or with siRNAscr (Supplementary Figure 4J,  K).These findings show that upregulation of A␤ in Phf8-silenced cells was associated with impaired autophagy and not with the A␤PP levels.

DISCUSSION
Our present findings show that Blmh, a Hcythiolactone-hydrolyzing enzyme [3], has a protective role in the CNS.Specifically, we demonstrated that Blmh depletion, which causes accumulation of Hcythiolactone and N-Hcy-protein in mice [4], downregulated histone demethylase Phf8 and upregulated  the H4K20me1 epigenetic mark.This increased H4K20me1 binding to the mTOR promoter and upregulated mTOR signaling, which in turn inhibited the autophagy flux.We also showed that Blmh depletion upregulated A␤PP and A␤.These biochemical changes, also induced by Hcy-thiolactone an N-Hcy-protein in cultured neural cells, were associated with cognitive and neuromotor deficits in mice.
5xFAD mice accumulate high levels of A␤ beginning around 2 months of age [40] and develop cognitive impairments beginning at 4-5 months of age and sensorimotor impairments at about 9 months of age [52], performing worse than the wild type animals in the memory and sensorimotor tests (https://www.alzforum.org/research-models/5xfad-b6sjl).In the present work, we found that Blmh depletion aggravated those neurological impairments.Specifically, 1-year-old Blmh −/− 5xFAD mice performed worse than Blmh +/+ 5xFAD animals in the NOR test (Fig. 1E), showing impaired memory, and in the hindlimb (Fig. 1F) and cylinder tests (Fig. 1G), showing sensorimotor impairments.We found similar memory and sensorimotor impairments also in 4-month-old Blmh −/− versus Blmh +/+ mice (Fig. 1A-D), which do not accumulate A␤.Our findings suggest that the absence of the Blmh protein is a dominant determinant that causes memory and sensorimotor impairments regardless of the presence or absence of the A␤-generating transgene.
Memory and sensorimotor impairments in Blmh −/− and Blmh −/− 5xFAD mice can be caused, at least in part, by Phf8 depletion, which occurs in Blmh −/− brains (Fig. 2A, Supplementary Figure 1A).Indeed, PHF8 depletion in humans is linked to intellectual disability, autism spectrum disorder, attention deficit hyperactivity disorder [32], and intellectual disability [33], while similar neurological deficits were found in Phf8 −/− mice [34].However, Phf8 was not known to be associated with A␤, a hallmark of AD.Our present findings that Phf8 depletion in mouse neuroblastoma cells, induced by Phf8 gene silencing (Supplementary Figure 4A) or by supplementation with Hcy-thiolactone or N-Hcy-protein (Fig. 3A), significantly increased A␤ accumulation (Supplementary Figure 4J, K; Fig. 5C,  D), suggest that Phf8 depletion can also underly the association of HHcy with AD [53].
Our findings also show that Phf8 is a mediator of Hcy-thiolactone/N-Hcy-protein effects on mTOR signaling, autophagy, and A␤ accumulation (Fig. 7).These findings directly link Hcy-thiolactone and N-Hcy-protein with dysregulated mTOR signaling (Fig. 3C, D) and its downstream outcomes such as impaired autophagy flux (Fig. 3I, J, K, L) and increased A␤ accumulation (Fig. 5) thereby providing a likely mechanism explaining neuropathy resulting from Blmh deficiency (Fig. 7) and accounting for the association of HHcy with AD [53].This role of Phf8 is further supported by findings showing that Phf8 gene silencing by RNA interference had the same effects on mTOR (Supplementary Figure 4C,  D), autophagy (Supplementary Figure 4E-G), and A␤ (Supplementary Figure 4J, K) as did Blmh gene silencing (Supplementary Figure 2D, E, G-M, and Fig. 5A, B) or the treatments with Hcy-thiolactone or N-Hcy-protein (Figs. 3 and 5C and 5D).
In the present study we showed that Blmh gene deletion led to A␤PP upregulation in Blmh −/− (Fig. 2 H) and A␤PP in Blmh −/− 5xFAD mice (Supplementary Figure 1I).These findings were recapitulated in mouse neuroblastoma cells by Blmh silencing, which upregulated A␤PP gene expression at the protein (Supplementary Figure 2F) and mRNA levels (Supplementary Figure 3D).In contrast, Phf8 gene silencing did not affect A␤PP expression (Supplementary Figure 4 H).These findings suggest that Blmh and A␤PP proteins interact with each other in the CNS while Phf8 does not interact with A␤PP.The Blmh-A␤PP interaction is most likely direct, as suggested by findings of other investigators.For example, one study has shown that human BLMH interacts with A␤PP in vitro and that overexpressed BLMH has the ability to process human A␤PP to A␤ in the 293-HEK and CHO cells [21].Another study Fig. 7. Hypothetical pathways leading to A␤ generation in Blmh −/− 5xFAD mice.Panel A illustrates the A␤PP (i) and Phf8 (ii-a) pathways.Panel B highlights the interaction (iii) between autophagy (Becn1) and A␤PP pathways.Up and down arrows show direction of changes in the indicated variables.Blmh, bleomycin hydrolase; Hcy, homocysteine; HTL, Hcy-thiolactone; A␤PP, amyloid-␤ protein precursor; mTOR, mammalian target of rapamycin; pmTOR, phospho-mTOR; Phf8, Plant Homeodomain Finger protein 8. See text for discussion.
reported that rat Blmh can further hydrolyze A␤ in vitro, with fibrillar A␤ 40 and A␤ 42 being more resistant than nonfibrillar peptides [22].Alternatively, BLMH can regulate mTOR expression via binding to the mTOR promoter.This possibility is supported by findings showing that BLMH binds to DNA [54,55].However, the mechanism underlying the regulation of A␤PP by Blmh needs to be elucidated in future studies.
We found that Blmh depletion downregulated Phf8 in mouse brain (Fig. 2A, Supplementary Figure 1A) and mouse neuroblastoma cells (Supplementary Figures 2B and 3B), upregulated A␤PP in mouse brain (Fig. 2H, Supplementary Figure 1I) and mouse neuroblastoma cells (Supplementary Figures 2F and 3D), and A␤ in mouse brain (Fig. 6D-E, I-J) and mouse neuroblastoma cells (Fig. 5A, B).In contrast, Phf8 depletion had no effect on A␤PP (Supplementary Fig- ure 4H, I) but still upregulated A␤ (Supplementary Figure 4J, K).These findings suggest that A␤ accumulation in Blmh-depleted mouse brains can occur via three pathways shown in Fig. 7.
Our findings that A␤PP upregulation (Fig. 2H, Supplementary Figure 1I) was associated with Becn1 downregulation in brains of Blmh −/− (Fig. 2E) and Blmh −/− 5xFAD mice (Supplementary Figure 1E) as well as in N2a-APPswe mouse neuroblastoma cells (Supplementary Figure 2F, G; Supplementary Figure 3D, G) suggest another pathway ((iii) in Fig. 7B) leading to A␤ accumulation.Pathway (iii) involves the interaction between Bcln1 and A␤PP in which Becn1 is a negative regulator of A␤PP expression and processing (Fig. 7B).Becn1, a protein with a key role in the initiation of autophagy, is known to decrease in human AD brains while genetic reduction of Becn1 in transgenic mice that overexpress A␤PP (A␤PP + Becn +/− mice) increased A␤ accumulation in neuronal cells [38].Becn1 was also shown to regulate A␤PP processing and turnover.Specifically, depletion of Bcln1 by RNA interference in rat neuroblastoma cells expressing human A␤PP transgene (B103/hAPPwt cells) increased A␤PP, Lc3, and A␤, while Becn1 overexpression reduced A␤PP levels [37].The involvement of autophagy in Blmh depletion-induced A␤PP and A␤ accumulation needs to be confirmed in future experiments by boosting autophagy (e.g., by treatment with TAT-Beclin1), which should rescue A␤PP and A␤ accumulation in Blmh-depleted cells.
Notably, we found that HHcy induced by a high Met diet and Blmh gene deletion caused similar changes in the Phf8->H4K20me1->mTOR->autophagy pathway (Fig. 2, Supplementary Figure 1) and A␤ accumulation (Fig. 6I, J).These findings can be explained by our earlier findings showing that a common primary biochemical outcome of the Blmh gene deletion and high Met diet was the same: accumulation of Hcy-thiolactone and N-Hcy-protein [4,56].Indeed, as we also found in the present study, treatments with Hcy-thiolactone or N-Hcy-protein induce changes in the Phf8->H4K20me1->mTOR->autophagy pathway and A␤ accumulation in mouse neuroblastoma cells (Figs. 3 and 5C, 5D) like those seen in Blmh −/− mice or mice fed with high Met diet (Fig. 2, Supplementary Figure 1, Fig. 6I, J).These findings also suggest that dysregulation of Hcy metabolism in general would affect the Phf8->H4K20me1->mTOR->autophagy pathway and the central nervous system.Indeed, homocysteine metabolites inhibit autophagy, elevate A␤, and induce neuropathy by impairing Phf8/H4K20me1-dependent epigenetic regulation of mTOR in cystathionine ␤-synthase-deficient mice [57].
Notably, as shown in the present work, Blmh deficiency or HHcy induced by a high Met diet increased H4K20me1 methylation levels due to downregulation of the histone demethylase Phf8 (Fig. 2B, Supplementary Figure 1B).HHcy is also known to affect DNA and protein methylation via S-adenosylhomocysteine (AdoHcy, an inhibitor of cellular AdoMet-dependent methylation reactions), which underlies the pathology of HHcy-associated human disease (reviewed in [58]).However, possible inhibition of H4K20 methylase by AdoHcy would have an opposing effect, i.e., would reduce H4K20me1 levels.The present findings, linking Blmh with the status of the histone H4K20me1 methylation, are reminiscent of our recent findings showing that Pon1 deletion in mice elevated the H4K20me1 methylation levels via downregulation of Phf8 [51].Thus, these two Hcythiolactone-detoxifying enzymes exert similar effects on H4K20me1 levels.Although there is no evidence that Blmh or Pon1 are linked to DNA methylation, our present and earlier findings supply the first evidence that Blmh and Pon1 can affect histone methylation status.
In conclusion, our findings show that Blmh interacts with A␤PP and the Phf8/H4K20me1/ mTOR/autophagy pathway, and that disruption of these interactions leads to A␤ accumulation and cognitive and neuromotor deficits.By revealing the mechanism by which Blmh prevents A␤ generation and cognitive/neuromotor deficits, the hallmarks of AD, our findings significantly expand our understanding of how Blmh maintains brain homeostasis.

Fig. 2 .
Fig. 2. Blmh depletion affects the expression of histone demethylase Phf8, histone H4K20me1 epigenetic mark, mTOR signaling, autophagy, and A␤PP in the Blmh −/− mouse brain.One-month-old Blmh −/− mice (n = 7 and 14) and Blmh +/+ sibling controls (n = 10 and 10) fed with a control or high Met diet (to induce HHcy) for one or three months were used in experiments.Each group included 7-14 mice of both sexes.Bar graphs illustrating quantification of the following brain proteins by western blotting are shown: Phf8 (A), H4K20me1 (B), mTOR (C), pmTOR (D), Becn1 (E), Atg5 (F), Atg7 (G), and A␤PP (H).Gapdh protein was used as a reference for normalization.Representative pictures of western blots used for protein quantification are shown in panel (I).Each mouse was assayed in three independent experiments and average values were used to calculate the mean ± standard deviation (SD) values for wrote down proteins in each experimental group.*p<0.05,**p<0.01,***p<0.001,or ****p<0.0001were calculated by one-way ANOVA with Tukey's multiple comparisons test.The numbers above bars show p values 0.08 -0.11.NS, not significant.

Fig. 4 .
Fig. 4. Blmh gene silencing or treatment with Hcy-thiolactone or N-Hcy-protein increases H4K20me1 binding at the mTOR promoter in mouse neuroblastoma N2a-APPswe cells.A) CUT&RUN assays with anti-H4K20me1 antibody show specific binding of H4K20me1 at the transcription start site (TSS) of the mTOR gene as well as downstream and upstream sites in Blmh siRNA-silenced N2a-APPswe cells.Bar graphs show the relative H4K20me1 binding at the indicated regions of the mTOR gene in N2a-APPswe cells transfected with two different siRNAs targeting the Blmh gene (siRNA Blmh #1 and #2).Transfections without siRNA (Control -siRNA) or with scrambled siRNA (siRNAscr) were used as controls.B) Control CUT&RUN experiment with anti-H3K4me3 antibody shows that Blmh gene-silencing did not affect the binding of H3K4me3 at the Rpl30 intron.RT-qPCR was conducted on the input and precipitated DNA fragments.Data are averages of three independent experiments.C) N2a-APPswe cells were treated with the indicated concentrations of N-Hcy-protein or Hcy-thiolactone (HTL) for 24 h at 37 • C. Untreated cells were used as controls.The CUT&RUN assays with anti-H4K20me1 antibody show that H4K20me1 binds to the transcription start site (TSS) of the mTOR gene as well as downstream and upstream sites.Bar graphs show relative H4K20me1 binding at the indicated regions of the mTOR gene.D) A control CUT&RUN experiment with anti-H3K4me3 antibody shows that Hcy-thiolactone and N-Hcy-protein did not affect the binding of H3K4me3 at the Rpl30 intron.Panels C and D were reproduced with permission from [51].RT-qPCR was conducted on the input and precipitated DNA fragments.Data are mean ± SD of three biologically independent experiments.p-values were calculated by one-way ANOVA with Tukey's multiple comparisons test.*p<0.05,**p<0.01,***p<0.001.

Fig. 5 .
Fig. 5. Blmh gene silencing or treatment with Hcy-thiolactone or N-Hcy-protein promotes A␤ accumulation in mouse neuroblastoma cells.Analysis of A␤ in mouse neuroblastoma N2a-APPswe cells was performed by confocal immunofluorescence microscopy using an anti-A␤ antibody.A, B) The cells were transfected with siRNAs targeting the Blmh gene (siRNA Blmh #1 and #2).Transfections without siRNA (Control -siRNA) or with scrambled siRNA (siRNAscr) were used as controls.Confocal microscopy images (A) are representative of at least N = 3 biologically independent experiments.Bar graphs (B) show the quantification of A␤ signals from Blmh-silenced and control cells.C, D) N2a-APPswe cells were treated with the indicated concentrations of N-Hcy-protein (N-Hcy) or Hcy-thiolactone (HTL) for 24 h at 37 • C. Untreated cells were used as controls.Confocal microscopy images (C) are representative of at least N = 3 biologically independent experiments.Bar graphs (D) show the quantification of A␤ signals from cells treated with N-Hcy or HTL and untreated cells.Each data point is a mean ± SD of three biologically independent experiments with triplicate measurements in each.Panels (C) and (D) were reproduced with permission from [51].p-values were calculated by one-way ANOVA with Tukey's multiple comparisons test.*p<0.05,**p<0.001,***p<0.0001.

Fig. 6 .
Fig. 6.Blmh gene deletion promotes A␤ accumulation in 5xFAD mice.Analysis of A␤ in brains from 5-month-old (A-F) and 12-month-old (G-J) Blmh −/− 5xFAD and Blmh +/+ 5xFAD mice fed with a standard or high Met diet (to induce HHcy) since weaning at the age of one month is shown.Brain extracts were analyzed on SDS-PAGE gels and A␤ was quantified by western blotting.Representative pictures of western blots of A␤ fractions extracted from brains with RIPA buffer (A, G), 2% SDS (B, H), and 70% formic acid (FA) (C) are shown.Numbers below each lane refer to the amount of brain (mg) from each mouse used in the experiment.A commercial A␤ 42 standard is shown in the first lane from left in each blot.Total A␤ signals in each lane (standing for an individual mouse) were quantified by scanning all chemiluminescent bands.Bar graphs (D, E, F, I, L) show brain A␤ quantification for Blmh −/− 5xFAD and Blmh +/+ 5xFAD mice (n = 8-10 mice/group).Data points for each mouse group represent mean ± SD with four independent measurements for each mouse.p-values were calculated by one-way ANOVA with Tukey's multiple comparisons test.*p<0.05,**p<0.001,***p<0.0001.