Abrogation of type-I interferon signalling alters the microglial response to Aβ1–42

Neuroinflammation and accompanying microglial dysfunction are now appreciated to be involved in Alzheimer’s disease (AD) pathogenesis. Critical to the process of neuroinflammation are the type-I interferon (IFN) family of cytokines. Efforts to phenotypically characterize microglia within AD identify distinct populations associated with type-I IFN signalling, yet how this affects underlying microglial function is yet to be fully elucidated. Here we demonstrate that Aβ1–42 exposure increases bioactive levels of type-I IFN produced by primary microglia alongside increased expression of type-I IFN related genes. Primary microglia isolated from brains of APPswePS1ΔE9 mice with ablated type-I IFN signalling show an increased phagocytic ability to uptake FITC-Aβ1–42. Correlative assessment of plaque sizes in aged APPswePS1ΔE9 mice with abrogated type-I IFN signalling show unchanged deposition levels. Microglia from these mice did however show alterations in morphology. This data further highlights the role of type-I IFN signalling within microglia and identifies a role in phagocytosis. As such, targeting both microglial and global type-I IFN signalling presents as a novel therapeutic strategy for AD management.


Treatment of microglial cells with type-I IFN inhibits phagocytosis. Recently, type-I IFNs have
been reported to be involved in the modulation of microglial phenotype. To investigate how type-I IFN stimulation directly affects microglial function, a phagocytosis assay utilizing pHrodo E. coli bioparticles was performed with immortalised BV-2 murine microglial-like cells. Dose response curves revealed that IFNα significantly decreased phagocytosis at concentrations of 100 (untreated: 100 ± 0% vs. 100U: 73.57 ± 5.656%, p = 0.0297, n = 5) and 1000 units (untreated: 100 ± 0% vs. 1000U: 78.07 ± 5.215%, p = 0.0423, n = 5) (Fig. 1a). Regression analysis also showed a negative relationship between phagocytosis and concentration. IFNβ treatments showed a similar trend for decreased phagocytosis with increasing concentration, however this was not statistically significant (Fig. 1b). To compare this to a classical pro-inflammatory response, this assay was repeated with lipopolysaccharide (LPS) (10-10,000 ng/mL) as a stimulus (Fig. 1c). Contrasting to the type-1 IFNs, significant increases in phagocytosis were seen with LPS concentrations greater than 100 ng/mL (untreated: 100 ± 0% vs. 100 ng/ mL: 121.6 ± 2.89%, p = 0.0157, 500 ng/mL: 126.4 ± 3.249%, p = 0.0124, 1000 ng/mL: 133 ± 5.99%, p = 0.0305, 5000 ng/mL: 128.5 ± 2.978%, p = 0.0077, 10,000 ng/mL: 133 ± 5.2961%, p = 0.0261). This suggests that type-I IFNs are indeed eliciting a unique functional effect on microglia that differs from a classical pro-inflammatory response. For increased sensitivity and to confirm these findings, we also performed experiments using a high content imaging platform. The workflow is described in Fig. 1d. Regression analysis showed a negative relationship for both IFNα and IFNβ treatments in number of spots per cell, confirming our initial findings. We then repeated these experiments using primary CX3C chemokine receptor 1 (CX3CR1 eGFP/+ ) microglia. Similarly, we are able to show a negative dose-response relationship. To then investigate if this inhibition can be reversed, cells were pre-treated with either IgG controls or an anti-IFNAR1 (MAR1) antibody before treatment with either 10 4 units IFNα/β. MAR1 was able to increases the number of spots per cell in IFNα treated cells (9.632 ± 4.023, p = 0.0436).
Monomeric Aβ 1-42 triggers a type-I IFN response in wild type microglia that is ameliorated in IFNAR1 −/− microglia. We have previously demonstrated that ablation of type-I IFN signalling in APP swe PS1 ΔE9 mice slowed cognitive decline and altered global glial phenotype 33 . Here, we focused specifically on examining the microglial response. To confirm that microglia do indeed mount a type-I IFN response, primary microglia of wild type and IFNAR1 −/− genotypes were subjected to monomeric Aβ 1-42 treatment in vitro. Media was collected and analysed by a bioactive type-I IFN measurement assay (Fig. 2a). IFNAR1 −/− microglia showed decreased levels of bioactive type-I IFN compared to wild type at both 24 (wild type: 3.719 ± 0.299 U/mg total protein vs. IFNAR1 −/− : 1.716 ± 0.585 U/mg total protein, p = 0.0427, n = 5-6) and 48 (wild type: 3.889 ± 0.391 U/mg total protein vs. IFNAR1 −/− : 1.416 ± 0.585 U/ mg total protein, p = 0.0094, n = 5-6) hours after treatment.
To confirm if these changes in messenger RNA (mRNA) expression translated to alterations in protein levels, enzyme-linked immunosorbent assays (ELISA) were performed on media from these same samples. Due to differences in purity between commercially available Aβ 1-42 , all results are expressed as a ratio compared to total media protein levels as measured by Bradford analysis. Between genotypes, significant differences were observed in levels of IL6 ( Fig. 3d) at 24 hours (wild type: 2.83 ± 1.026 pg/mg total protein vs. IFNAR1 −/− : 0.44 ± 0.144 pg/ mg total protein, p = 0.0164, n = 5-6) and IL1β (Fig. 3e) at 48 hours (wild type: 4.83 ± 2.682 pg/mg total protein vs. IFNAR1 −/− : 0.228 ± 0.077 pg/mg total protein). TNFα measurements showed a trend for decreased levels in IFNAR1 −/− microglia across all observed timepoints (Fig. 3f). This further confirms the pro-inflammatory response elicited by Aβ 1-42 in vitro and identifies type-I IFNs as potential regulators of this response.
To confirm if this finding was specific to Aβ or an increase in overall phagocytic ability, we then performed the same experiment using the reverse peptide FITC Aβ 42-1 . Interestingly, APP swe PS1 ΔE9 × IFNAR1 −/− microglia showed no similar increases in uptake, but rather IFNAR1 −/− microglia showed significant increases in both percentage of parent (  Plaque burden in 9 and 13-month APP swe PS1 ΔE9 × IFNAR1 −/− mice is unchanged. To investigate whether the observed increased Aβ phagocytosis translated in vivo, we performed immunofluorescence analysis of Aβ plaques in both 9 and 13-month APP swe PS1 ΔE9 and APP swe PS1 ΔE9 × IFNAR1 −/− mice. Sagittal brain sections were stained with an anti-Aβ antibody and imaged before being subjected to an automated analysis utilizing a watershed segmentation approach (Fig. 5b). Both APP swe PS1 ΔE9 and APP swe PS1 ΔE9 × IFNAR1 −/− mice show significant deposition throughout the brain. However, no significant difference was observed in either 9 or 13-month old mice when quantitation of plaque size, plaques per mm 2 or % of plaque burden was performed in either the cortical or hippocampal areas ( Fig. 5c-h). Both APP swe PS1 ΔE9 and APP swe PS1 ΔE9 × IFNAR1 −/− showed similar age-related increases in plaques. These findings suggest that removal of IFNAR1 does not alter plaque burden in either 9 and13-month old APP swe PS1 ΔE9 mice.

Discussion
Neuroinflammation is seen as a chronic and detrimental process within AD, and it is suggested that microglia are the critical cell type responsible for this response 34,35 . This is also seen in congruence with microglial dysfunction. As such the identification of key regulators of this neuroinflammatory process present as novel therapeutic targets 36 . Type-I IFNs are known to act as "master regulators" of the innate immune response and are able to regulate levels of IL1β, IL6 and TNFα which are consistently upregulated within AD brains 3,[27][28][29] . Recent work investigating microglial phenotype has identified type-I IFNs as regulators of unique and conserved microglial populations in both ageing and disease 8,31,32,37 . Here, we further examine the modulation of these phenotypic microglia by the type-I IFNs, critically focussing on investigating their functional roles.
We firstly identified that type-I IFN treatment decreases the ability of microglial phagocytosis in a dose-dependent manner. Through genetic abrogation of type-I IFN signalling by targeting its receptor IFNAR1, www.nature.com/scientificreports www.nature.com/scientificreports/ we identify that IFNAR1 −/− primary microglia exhibit a decreased IL1β, IL6 and TNFα response to monomeric Aβ 1-42 , and that targeting IFNAR1 within primary microglia isolated from the APP swe PS1 ΔE9 AD model leads to an enhanced ability to phagocytose FITC Aβ 1-42 . These aged APP swe PS1 ΔE9 × IFNAR1 −/− mice also show altered microglial morphology in vivo.
We have previously reported that ablation of type-I IFN signalling within the APP swe PS1 ΔE AD model is neuroprotective 33 . In this model, reduced type-I IFN signalling was associated with an attenuated whole-brain inflammatory profile and a rescue in cognition as assessed via the Morris water maze. The cell-specific contributions that resulted in this altered response were unknown. This suggested that microglia were indeed the cell type responsible.
Microglial function follows phenotype, and phagocytosis is recognised as a critical function of microglia 38 . We first employed a phagocytosis assay on microglial cells to investigate how type-I IFN stimulation affects this process. A negative phagocytic dose-response relationship was observed in cells treated with IFNα, with a trend for a decrease in IFNβ also seen. Previous work investigating IFN and phagocytosis has shown differing results.
In an autoimmune encephalomyelitis model of multiple sclerosis, IFNβ was shown to increase microglial phagocytosis of myelin 39 . Furthermore, increased myelin was observed within the brains of IFNAR1 −/− knockout mice. www.nature.com/scientificreports www.nature.com/scientificreports/ Phagocytosis is not limited to a single pathway, and both myelin and Aβ may indeed invoke separate cellular mechanisms that result in engulfment 40 . There may also be other kinetic measures that explain these differences which are masked in end-point assays. Investigation in macrophages and other mononuclear phagocytes also demonstrate a role for type-I IFNs in promoting phagocytosis [41][42][43] . Similarly, these studies also use different particles to measure phagocytosis. Regardless, further work is required to investigate these findings. Our results confirm work that type-I IFNs are involved in microglial function, with data demonstrating that IFNα in particular affects phagocytosis. Interestingly, these decreases contrasted to the increases seen with the pro-inflammatory LPS. To further investigate this finding, we also performed similar experiments using a high content imaging platform for increased sensitivity. We are able to confirm our initial findings with BV-2 cells. Levels of type-I IFNs are known to increase with normal ageing, with our study supporting previous data identifying an impaired phagocytic function of aged microglia 31,44 . We are able to demonstrate that type-I IFNs elicit differential functional effects on microglia when compared to classical inflammatory stimuli, suggesting that type-I IFNs are able to shift microglia to a unique phenotype. This is in line with recent findings identifying unique microglia populations within the CNS that are enriched with a number of IFN regulated genes 45 . www.nature.com/scientificreports www.nature.com/scientificreports/ Consistent with our previous findings examining Aβ treatments on primary mixed glial cultures, our study confirmed that Aβ 1-42 elicits a type-I IFN response in wild type microglia, with this response attenuated in IFNAR1 −/− microglia. Increases were seen in levels of bioactive type-I IFN at 24 and 48 hours between genotypes, which is mirrored in the mRNA transcripts of both IFNα and IFNβ. Within IFNAR1 −/− microglia transcript levels of IFNα were observed to remain constant, but IFNβ levels actively downregulated. This suggests differing roles for IFNα and IFNβ in AD, and that IFNα is the critical subtype that affects microglia within AD. Type-I IFNs are known to amplify their responses in an autocrine manner and elicit differing responses depending on both type and subtype 46 . In congruence, levels of IL1β and IL6 are actively decreased within these same IFNAR1 −/− microglia. This suggests that expression of these hallmark proinflammatory cytokines may be downstream of type-I IFN signalling. Furthermore, alteration of microglial phenotype through reduction of these Figure 6. 13-month APP swe PS1 ΔE9 × IFNAR1 −/− microglia adopt a stellate-like morphology. 30 μm thick brain sections of 9 and 13-month old APP swe PS1 ΔE9 and APP swe PS1 ΔE9 × IFNAR1 −/− mice were stained with both IBA-1 and WO2 antibodies before 15 μm z-stacks were taken from of the hippocampal region. Representative images are shown for 13-month APP swe PS1 ΔE9 and APP swe PS1 ΔE9 × IFNAR1 −/− in (a) and (b) respectively. A minimum spanning tree algorithm was then employed to measure morphological characteristics with an overview of the skeletonising process in (c) and (d) for both genotypes respectively. Measurements are then shown for (e) total branch length, (f) cell area and (g) cell radius. Data is displayed as mean ± SEM (*p < 0.05, Students t-test, APP swe PS1 ΔE9 vs. APP swe PS1 ΔE9 × IFNAR1 −/− , n = 3-8). Scale bar = 500 μm. (2020) 10:3153 | https://doi.org/10.1038/s41598-020-59917-0 www.nature.com/scientificreports www.nature.com/scientificreports/ hallmark pro-inflammatory cytokines is well established in shifting microglial function 47 . Levels of IRF7 were upregulated between wild type and IFNAR1 −/− genotypes at both 24 and 48 hours respectively. In primary macrophages, it has been reported that IRF7 is able to elicit differential type-I IFN responses, with IRF7 associated with an overall heightened response level 48 . Furthermore, IRF7 is known to be a major IRF induced downstream of type-I IFN responses and is critical in further induction of the IFN response 19 .
Phagocytosis is a critical functional process of microglia impaired in AD 3 . Here we observed that APP swe PS1 ΔE9 × IFNAR1 −/− microglia, but not IFNAR1 −/− microglia alone, exhibited greater phagocytic ability towards FITC Aβ 1-42 . This was compounded by the increases in uptake on a per-cell basis of IFNAR1 −/− microglia of the reverse peptide, FITC Aβ 42-1 . Whether this result is a unique interaction between APP swe PS1 ΔE9 and IFNAR1 or indeed due to other processes remains unknown, and as such warrants further investigation. One such explanation may be microglial "priming", a concept explored within neuroinflammation and AD 49 . Levels of Aβ have been observed in culture from transfected cells that contain human APP, which may be responsible for alterations in microglial phenotype 50 . Bias for particular Aβ species may also be involved. The nature of FITC conjugation denotes that aggregation dynamics differ between it and unlabelled peptides 51 . This oligomerisation bias may also explain the observations in vivo, with APP swe PS1 ΔE9 × IFNAR1 −/− mice showing decreases in monomeric levels of Aβ 1-42 only 33 . The observed increases in the IFNAR1 −/− microglia may indeed be due to differential phagocytic processes. Critically, the reverse peptide has differential aggregation dynamics when compared to the forward peptide sequence 52 . Multiple receptors are involved in microglial phagocytosis that differ in alterations in cytoskeletal elements, phagosome maturation and inflammatory responses in response to binding 38,40 . It is important to note that previous in vitro work investigating immune-related gene knockouts and fluorescent Aβ phagocytosis have not considered the role of their respective AD genotype 13,53 .
Due to these observed differences in uptake of Aβ 1-42 , we then examined Aβ plaque burden as a surrogate measure of in vivo microglial phagocytosis. Critically, we expand our analysis from our previous work. We observed no differences between genotypes, which follow our previous findings 33 . It is well established that plaque burden is not a measure of disease state, as levels do not correlate with cognition. Furthermore, plaque reduction does not alter pathology and cognitively normal individuals containing plaques within their brains [54][55][56] .
We further investigate these mice though use of a sophisticated, and critically, verified, algorithm to examine a number of morphological features 57 . Increased cellular radius and diameter without increases in branch length indicate that APP swe PS1 ΔE9 × IFNAR1 −/− microglia adopt a stellate-like morphology. The exact nature of how this relates to underlying phenotype however remains unknown. Classically, ramified-like morphologies have classically been used to identify "quiescent" or "resting" microglial states and amoeboid-like morphologies for "reactive" states 58 . Similar morphologies can however exhibit altered phenotypes 59 . As such, the extent to which this classic paradigm holds true remains unknown. However, quantified morphological analyses add an additional trait to further our understanding of microglia. Further work is required to link various morphological features and microglial functions, both in normal and disease states, as well as across ageing.
Our observations combined with our in vitro data suggest that these microglia possess an altered anti-inflammatory phenotype. The observation of altered morphology at 13 but not 9 months of age in APP swe PS1 ΔE9 × IFNAR1 −/− microglia is notable and calls for further examination. It is established that APP swe PS1 ΔE9 mice begin to show a decreased cognitive phenotype at 6 months, and as such 13-months is an advanced stage for disease 60 . This heightened disease state may overcome protective effects due to loss of IFNAR1 −/− as we have previously observed, in turn altering microglial phenotype and subsequent morphology. Future work will focus on the direct analysis of microglia isolated from these mice with a transcriptomic based approach utilised, similar to that reported in other related studies focused on microglial phenotypes 8,31 . This combined with morphological data will allow for deeper insight into how the type-I IFN signalling pathway is involved in regulating microglial function.

conclusions
These results demonstrate that type-I IFNs are involved in the modulation of both microglial phenotype and function. This work further expands on emerging data demonstrating a link between microglia, type-I IFNs and AD. Type-I IFNs present as a much-needed novel therapeutic target for the management of AD.

Methods
Animals. C57BL/6J wild type mice were sourced from the Animal Resource Centre. IFNAR1 −/− mice were initially generated by Hwang, et al. 18 . APP swe PS1 ΔE9 transgenic mice were originally sourced from JAX 61 . APP swe PS1 ΔE9 transgenic mice lacking IFNAR1 −/− were generated by Minter et al. 33 . All mice were housed in sterile micro-isolator cages and fed ad-libitum. All animal procedures were performed in accordance with the University of Melbourne animal care committee's regulations and conducted in compliance with the Australian National Health and Medical Research Guidelines. All experiments were approved by the University of Melbourne Animal Ethics Committee (Ethics ID: 1613905).

Mixed cortical and hippocampal glial isolation. Mixed cortical and hippocampus glial cultures were
isolated from P0-P1 pups as described previously 33 . Briefly, cortices and hippocampal structures were isolated, and meninges then surgically removed. This cleaned tissue was then placed into a solution of hanks buffered saline solution (HBSS) containing trypsin ethylenediaminetetraacetic acid (EDTA) (1 × final concentration, 59418 C, Sigma) and deoxyribonuclease (DNAse) (1 mg/mL, D5025, Sigma) for 15 minutes, after which the supernatant was collected. Remaining undigested tissue was then subject to a second round of digestion before the supernatants were combined. Cells were plated at a density of 1.25 × 10 4 cells/mL in culture medium (DMEM containing 20% foetal bovine serum (FBS), 1% penicillin/streptomycin). Media was replaced at days 2, 7 and 14 after which the glial cells formed a monolayer.
Microglial isolation. Microglia were isolated from mixed glial cultures at 18 days in vitro. Isolation was performed using the Saura, et al. 62 mild trypsinization method and fluorescence activated cell sorting (FACS) for CX3CR1 eGFP/+ microglia. Conditioned media was collected from each individual plate before undergoing a wash with DMEM. Media was then replaced with a 1:1 mixture of DMEM and 0.25% trypsin EDTA (25200-072, Gibco) and left for 30 minutes. Once detachment of the upper glial layer was confirmed, cells were again washed with DMEM before the collected conditioned media was returned to the individual plates. Cells were given a further 2 days incubation before use in experiments. Microglial purity was approximately 95% as analysed periodically through flow cytometry using a FITC conjugated CD11b antibody (130-113-796, Miltenyi Biotech). fitc-conjugated amyloid-β treatment. Cells were treated with serum free DMEM containing 2 mg/ mL of FITC conjugated Aβ 1-42 or Aβ 42-1 (1% sodium azide, PBS) for 1 and 3 hours respectively. Cells were rinsed before being scraped in PBS and centrifuged (1500 × G, 5 minutes, room temperature) to form a pellet. Resulting pellets were then resuspended in FACS buffer (1% BSA, 100 ng/mL 4′,6-diamidino-2-phenylindole (DAPI), in PBS) and transferred into 5 mL cytometer tubes. Samples were read on a BD Fortessa flow cytometer.

Amyloid-β preparation and treatment.
B16 IFN assay. B16-Blue cells (bb-ifnt1, InvivoGen) were used to measure bioactive type-I IFN in collected media as per manufacturers instructions. Briefly, cells were cultured in T75cm 2 flasks in Roswell Park Memorial Institute 1640 (21870-092, Gibco) medium containing 5% FBS, 1% penicillin/streptomycin and zeocin (100 μg/ ml). Cells were then plated in a 96 well plate at 75,000 cells/well. Collected media was added to these cells for 24 hours alongside IFNα (0-1000 U/mL) to generate a standard curve. QUANTI-Blue was then added in 1:1 v/v ratio for a further 24 hours. Following this, 200 μL of media was transferred into a 96 well plate and absorbance was measured on a Multiskan Ascent plate reader. eLiSA. Murine Il1β (559603, BD Biosciences) IL6 (555240, BD Biosciences), and TNFα (DY410, R&D Systems) were used to detect protein levels in collected media as per manufacturer's instructions. Briefly, 96-well plates were coated overnight in capture antibodies diluted in assay diluent (1:1000) at 4°. Plates were washed thrice, then blocked in assay diluent. Wells were then filled with either recombinant standards or 100 μL media from collected samples (2 hours, room temperature. Plates were then washed 5 times after which 100 μL working detector (detection antibody + SAv-HRP reagent) was added (1 hour, room temperature). Plates were then washed 7 times and a 1:1 mixture of hydrogen peroxide and 3,3′, 5,5′ tetramethylbenzidine was added (30 minutes, room temperature). The reaction was then stopped using sulphuric acid (160 mM) and absorbance measured at 450 nm in a Multiskan Ascent spectrophotometer (Thermo Scientific) RnA isolation. All samples underwent a modified RNA isolation protocol using both Trizol ® (Invitrogen) and the illustra RNAspin mini kit (25050071, GE Health Sciences). Cultured cells were rinsed once in DMEM before being scraped in 1 mL Trizol ® and transferred into a fresh autoclaved RNAse/DNAse free tube. Samples were then homogenized using a 23 g needle and 1 ml syringe. 0.2 mL chloroform was then added and samples shaken vigorously for 15 seconds before incubation for 3 minutes at room temperature. Samples were centrifuged (12000 × G, 15 minutes, 4 °C), after which the clear supernatant was collected and transferred into a new ribonuclease (RNAse)/DNAse free tube. 70% ethanol was then added to the supernatant at a ratio of 1:1, after which the mixture was placed onto the illustra RNAspin collection column. Isolation was then performed as per manufacturer's instructions. Sample concentration and purity was assessed using a NanoDrop 1000 spectrophotometer. (ThermoScientific). All samples measured underwent on-column DNAse treatment and had 260/280 ratios above 2.
Reverse transcription. 1-2 μg of total RNA was reverse transcribed using the High Capacity complementary DNA (cDNA) Reverse Transcription Kit (4368814, Applied Biosciences) as per manufacturer's instructions. Individual reactions were prepared, and PCR was performed in a thermal cycler (Eppendorf) under the following conditions: 10 minutes at 25 °C, 120 minutes at 37 °C, 5 minutes at 85 °C and 10 minutes at 4 °C. Each sample group also contained an additional reaction without the addition of reverse transcriptase enzyme (-RT). Final cDNA samples were diluted 1:10 for each 1 μg of initial RNA.
Immunofluorescence. Mice were anesthetized via intra-peritoneal injection of a combinatorial mixture of ketamine (100 mg/kg) and xylazine (10 mg/kg). Deep anaesthesia was confirmed through absence of paw withdrawal reflex. Mice then underwent cardiac perfusion with PBS to remove blood from the cerebral vasculature, after which 4% paraformaldehyde (PFA) was then delivered to fix the brain. Brains were then excised and placed into a solution of 4% PFA for 24 hours at 4 °C after which they were transferred into 30% sucrose solution for 24 hours at 4 °C for cryoprotection. Brains were then submerged in optimal cutting temperature media (OCT) (Sakura) within 2 cm 2 plastic containers and frozen by immersion within an isopropanol dry-ice bath. 30 μm sections were cut sagittally on a cryostat from the initiation of the hippocampal structure to its completion. Sections were then placed into an individual well of a 24 well plate containing PBS for downstream staining. Collected sections were rinsed once in PBS for 5 minutes, after which they were simultaneously permeabilized and blocked by being placed in 200 μL of a solution containing 1.5% Triton-x (Sigma) in 10% goat block (G9023) in PBS (2 hours, room temperature, constant rocking at 20RPM). Sections were then washed thrice in PBS (room temperature, 4 minutes per wash) before incubation with primary antibodies (4 °C, overnight, constant rocking at 20RPM). Details of antibodies and concentrations used are listed in Table 3. Sections were then washed thrice (room temperature, 5 minutes per wash) before incubation with goat fluorescent secondary antibodies occluded from light (1:1000 dilution in PBS, room temperature, A-11012, A-11001, Invitrogen). Sections were washed thrice a final time (room temperature, 5 minutes per wash) before being rinsed in MilliQ H 2 O. Using a wide orifice pipette, sections were transferred onto slides and allowed to dry completely before coverslips were mounted using DAPI containing mounting media (H-1500, Vectashield). Images were obtained using the Zeiss Axio Observer.Z1 or   Table 3. List of antibodies used in this study.