Monitoring phagocytic uptake of amyloid β into glial cell lysosomes in real time

Phagocytosis by glial cells is essential to regulate brain function during health and disease. Therapies for Alzheimer's disease (AD) have primarily focused on targeting antibodies to amyloid β (Aβ) or inhibitng enzymes that make it, and while removal of Aβ by phagocytosis is protective early in AD it remains poorly understood. Impaired phagocytic function of glial cells during later stages of AD likely contributes to worsened disease outcome, but the underlying mechanisms of how this occurs remain unknown. We have developed a human Aβ1–42 analogue (AβpH) that exhibits green fluorescence upon internalization into the acidic organelles of cells but is non-fluorescent at physiological pH. This allowed us to image, for the first time, glial uptake of AβpH in real time in live animals. We find that microglia phagocytose more AβpH than astrocytes in culture, in brain slices and in vivo. AβpH can be used to investigate the phagocytic mechanisms responsible for removing Aβ from the extracellular space, and thus could become a useful tool to study Aβ clearance at different stages of AD.


Introduction
Glial cells make up more than half of the cells of the central nervous system (CNS) and are vital to the regulation of brain function. 1 Microglia are specialized CNS-resident macrophages that respond to pathogens and injury by clearing cell debris, misfolded protein aggregates and damaged neurons by the process of phagocytosis. 2 Mature microglia in the adult brain exhibit a ramied morphology and constantly survey their surroundings for "eat me" signals 3 present on or released from apoptotic cells, microbes, protein deposits, dysfunctional synapses and other target substrates. Aer CNS injury or during neurodegenerative diseases like Alzheimer's disease (AD) microglia become "reactive", and change morphology, becoming rod-like or amoeboid, 4 and actively engage with their environment by secreting inammatory cytokines like TNFa and IL-1a/b. These cytokines cause functional changes in astrocytes, microglia themselves, and other cells. 5,6 During phagocytosis, proteins on the microglial cell surface, such as the Toll-Like Receptors (TLRs), Fc receptors, and scavenger receptors including CD36 and the receptor for advanced glycation end products (RAGE) among others, recognize the "eat-me" signals and engulf the target substrates into intracellular compartments called phagosomes. [7][8][9][10] The phagosomes mature by fusing with lysosomes to form highly acidic phagolysosomes and mobilize the phagocytosed material for enzymatic degradation. The pH of phagosomal organelles during this maturation process is progressively reduced 11 from 6.0 to around 5.0-4.5. Although microglia are the "professional phagocytes" of the CNS, astrocytes are also competent phagocytic cells with important roles both during health, and in response to injury or in disease. [12][13][14][15] Recent evidence has demonstrated the phagocytic abilities of reactive astrocytes towards cellular debris in CNS injury. 16 Together, reactive microglia and astrocytes play a crucial role in clearing extracellular debris and cellular components and aid in remodeling the tissue environment during disease.
AD is characterized by the generation of soluble oligomers of amyloid b (Ab) that have numerous downstream actions, including reducing cerebral blood ow, 17 inhibiting glutamate uptake which may cause hyperexcitability of neurons, 18 and inducing hyperphosphorylation of the cytoskeletal protein tau 19,20 which leads to synaptic dysfunction and cognitive decline. Ultimately Ab oligomers are deposited as extracellular plaques in the brain, a hallmark of AD, which contribute to neuroinammation and neuronal death. 21 The main Ab species generated excessively in AD is Ab  , which is a small $4.5 kDa peptide produced by the cleavage of amyloid precursor protein on neuronal membranes by band g-secretases. 22,23 Removal of Ab from the extracellular space by phagocytosis into microglia and astrocytes, as well as by clearance across endothelial cells into the blood or lymph vessels, is thought to limit the build-up of the extracellular Ab concentration. However, AD pathology occurs when Ab generation outweighs its removal. 20 Thus, to understand the onset of plaque deposition during AD (and perhaps how to prevent it) it is essential to understand molecular mechanisms underlying glial phagocytosis and degradation of Ab. A method that can monitor this process, especially in real time in vivo, will facilitate identication of the receptors that bind to Ab and initiate its phagocytic clearance. This will allow investigation of why glia that surround Ab plaques in AD show impaired phagocytic function, 5,24,25 and why in inammatory conditions microglia may increase their phagocytic capacity depending on their state of activation. 26,27 Current methods to study glial phagocytosis involve the use of uorescent latex beads, 28 particles of zymosan, 29 or E. coli 30 conjugated to uorophores like uorescein and rhodamine. 31 A non-pH dependent particle makes it difficult to clearly determine whether the particle is inside the phagosomes or outside the cell during live-cell monitoring (Fig. S1A †). Some bioparticles can be labeled with pH-sensitive dyes such as pHrodo, however, these currently available pH-sensitive dyes are not suitable for labeling disease-specic pathogenic molecules like Ab for in vivo use. While non-pH sensitive uorophore conjugates of Ab have been used to evaluate Ab phagocytosis, 28,32 they have several disadvantages for live-cell imaging and cannot be used for selective identication and isolation of phagocytic cells in vivo. First, acidic pH-insensitive uorophore-conjugated Ab peptides exhibit sustained uorescence in the extracellular space (at physiological pH) thus contributing a noisy background that hinders the clear visualization of live phagocytic cells (Fig. S1B †). Second, in live-cell imaging and in uorescence-activated cell sorting (FACS) of live cells, it is difficult to differentiate between Ab molecules that are internalized by the cells versus Ab molecules that are stuck to the cell surface.
To address these issues, we have developed a pH-dependent uorescent conjugate of human Ab  , which we call Ab pH , and characterized it using mass spectrometry, atomic force microscopy and imaging of its uptake into cells in vitro and in vivo. We show the functionality of the Ab pH probe for identifying phagocytic microglia and astrocytes in several different biological model systems such as cell lines, primary cell cultures, brain tissue slices, and in vivo in brain and retina. Ab pH retains an aggregation phenotype similar to that of synthetic Ab in vitro and exhibits increased green uorescence within the acidic pH range of 5.0 to 4.5 but not at the extracellular and cytoplasmic physiological pH values of 7.4 and 7.1, respectively. Ab pH can be used to visualize phagocytosis in live cells in real time without the use of any Ab-specic antibody. It is internalized by glial cells (both astrocytes and microglia) in live rat hippocampal tissue sections in situ. Stereotaxic injection of Ab pH into the mouse somatosensory cortex in vivo leads to its uptake by astrocytes and microglia, following which microglia retain the Ab pH within the cells up to 3 days in vivo unlike astrocytes. Similarly, microglia in retinal tissues retain the Ab pH within the cells for up to 3 days but no signal was detected in astrocytes. Finally, we show, for the rst time, real-time phagocytosis of Ab into microglia and astrocytes in mouse cortex in vivo by twophoton excitation microscopy.

Results
Properties of a novel pH-dependent uorescent conjugate of human Ab  We synthesized a new pH-sensitive uorescent dye-labelled phagocytic Ab probe for imaging both in vitro and in vivo, and for cell sorting which allows for downstream analysis of functional subtypes of cells. We used a facile bioconjugation strategy to make our new probe safe for use with live cells in vitro and with live animals in vivo. The Ab pH conjugate was synthesized at the microgram scale by linking the synthetic human Ab  peptide to the amine-reactive Protonex Green 500, SE (PTXG) uorophore (Fig. 1A). We selected PTXG based on its ability to gain uorescence in acidic environments and thus emit uorescence specically at the low pH of 5.0-4.5 found within lysosomesa property of only a few commercial dyes. Conjugation of the uorophore was performed at the side chain amine groups of the lysine residues within, and at the Nterminal of, human Ab 1-42 peptide. The conjugation was conrmed with matrix-assisted laser desorption/ionizationmass spectrometry (MALDI-MS) ( Fig. S2A and B †) that indicated the conjugation of PTXG with the Ab 1-42 peptide (molecular weight >4.5 kDa) by removal of succinimidyl ester (SE) as a leaving group. Additionally, proton nuclear magnetic resonance ( 1 H-NMR) analysis of Ab pH also demonstrated the presence of PTXG as well as Ab 1-42 peptide ( Fig. S3A-C †). The spectrum of Ab 1-42 peptide obtained from attenuated total reection Fourier transform infrared spectroscopy (ATR-FTIR) also shows a strong absorption peak at 1625 cm À1 conrming carbonyl functional group of amide bonds (Fig. S4A †) and the The Ab pH is synthesized by conjugating the amine-reactive pH-sensitive Protonex Green dye to the side chain amine groups of the lysine residues and the N-terminal of human Ab 1-42 peptide. (B) The pH-sensitivity of the Ab pH probe characterized at different concentrations from 0.1 mM to 5.0 mM. Increased fluorescence is observed at acidic pH values of $5.0 to $2.0, covering the pH range of the intracellular acidic organelles. (C) Atomic force microscopy topographic images of Ab pH oligomers compared to synthetic Ab oligomers. Left-2D topographic image of Ab pH and synthetic Ab oligomers. Right-3D image (2 Â 2 mm x-y). (D) Live cell imaging of the phagocytic uptake of 1 mM Ab pH by BV2 and N9 mouse microglia and by HMC3 human microglia over 24 hours. (E) Quantification of Ab pH phagocytic score by BV2, N9, and HMC3 microglial cells from the live cell images. (F) The phagocytic uptake of Ab pH by BV2 cells is measured and quantified via flow cytometry analysis. Dot plot shows live (PI À ) and Ab pH+ cells. No green fluorescence is measured in unstained cells (UC) and in dead cells stained with the PI only whereas green fluorescence is measured in cells treated with 0.5 and 5.0 mM Ab pH for 1 hour (higher fluorescence is seen in cells exposed to the higher concentration of Ab pH ). Data shown in terms of % max, by scaling each curve to mode ¼ 100% (yaxis). spectrum of PTXG shows the presence of amide and ester group with absorption peaks at 1668 and 1727 cm À1 , respectively (Fig. S4B †). The conjugated product Ab pH shows a distinct peak at 1674 cm À1 conrming amide bond formation between the Ab 1-42 peptide and PTXG dye, as expected (Fig. S4C †). Collectively, these experiments conrm the formation of the peptidedye conjugate. We also synthesized another conjugate of Ab  with the pHrodo-Red, NHS uorophore (RODO) and conrmed conjugate formation from the MALDI-MS spectrum (Fig. S5 †).
The pH-sensitivity of the PTXG and RODO-conjugated Ab was assessed by measuring their uorescence intensities at various pH values. Notably, the pK a of a dye may shi when conjugated with a protein or peptide and this would change the pH-sensitivity of the conjugated product. 33 At the concentrations of 0.5, 1.0, 2.0, and 5.0 mM, the PTXG-conjugated Ab showed increased uorescence at pH less than $5.5 whereas PTXG alone showed increased uorescence at pH $4.5 with excitation/emission wavelengths of 443/505 nm. The pK a of PTXG-conjugated Ab was 5.2-5.9 and the pK a of PTXG alone was 4.4-4.7 at concentrations of 0.5, 1.0, 2.0 and 5.0 mM (Fig. S6A †). The PTXG dye clearly shows increased uorescence around pH $4.5 whereas PTXG-conjugated Ab shows a pK a of $5.5 that lies within the pH range of lysosomes thus making it suitable for our study. On the other hand, the RODO-conjugated Ab had a pK a value of 6.2-6.4 at four concentrations (5, 2, 1, 0.5 mM) (Fig. S6B †). It is known that the amine-reactive forms of the RODO dye have a pK a of $7.3 in solution and shi to about $6.5 upon conjugation 34 which is similar to the pK a value we observed for RODO-conjugated Ab. Furthermore, the PTXG-Ab conjugate showed a maximum uorescence intensity between 500 and 510 nm in the acidic pH range that covers the pH values of the lysosomal organelles (Fig. S7A †). The PTXG-Ab conjugate exhibited low uorescence intensity at pH values more alkaline than 6.0 including at the physiological extracellular pH of 7.4. In contrast, the RODO-conjugated Ab does not show a consistent pattern of higher orescence at similar acidic pH range and the pattern of pH sensitivity is dependent on concentration ( Fig. S7B †). Overall, these experiments show that the pK a shis towards phagosomal pH range when PTXG is conjugated to Ab (average pK a $5.5) compared to the value for PTXG alone (average pK a $4.5) whereas RODO-conjugated Ab has a pK a of $6.3 that is more alkaline and close to the endosomal pH range of around 6.4-6.5. 35 We also tested to see if the unconjugated PTXG and RODO dyes were endogenously taken up by the cells. BV2 microglia treated with PTXG alone showed very low uorescence indicating minimal uptake of the dye compared to cells treated with the PTXG-Ab conjugate. The cells treated with RODO alone showed high uorescence indicating higher dye uptake (at 1 mM, the cells showed almost 50% uorescence with RODO compared to the RODO-Ab conjugate) (Fig. S8 †) limiting its in vivo use. Neither of the two dyes were toxic to the cells in culture (Fig. S9 †).
In order to validate that the PTXG-Ab uorescence increase is due to the acidic environment of the phagosome, we measured the uorescence of the PTXG-Ab conjugate in cells treated with balomycin A (BF), a compound that inhibits lysosomal acidication by blocking phagosome-lysosome fusion during late stages of phagocytosis. 36 As expected, we measured a decrease in cellular PTXG-Ab uorescence with BF treatment compared to the control cells (Fig. S10 †). The reduction in PTXG-Ab uorescence in the presence of BF indicates that the uorescence of PTXG-Ab is dependent on the acidic pH of the lysosomal organelles. Thus, summarizing the above-experiments, we believe that the PTXG-Ab conjugate outperforms the RODO-Ab conjugate due to the following reasons: (i) a narrower range of uorescence, (ii) minimal background uptake, (iii) the long-term sustained uorescence intensity of PTXG-Ab (Fig. S11 †), and (iv) a more suitable pK a value. Thus, the PTXG-Ab conjugate performed better and was chosen for all further experiments (termed Ab pH henceforth in the paper). Lastly, we wanted to determine whether Ab pH exhibits aggregation properties similar to the aggregation of synthetic, non-conjugated Ab. The Ab 1-42 and Ab pH oligomers were prepared 37 from hexauoroisopropanol (HFIP) treated peptide lms in PBS pH 7.4 buffer at 4 C. Atomic force microscopy (AFM) revealed that the ability of Ab pH to aggregate is similar, in size and height, to that of the non-conjugated Ab ( Fig. 1C) suggesting that Ab pH is suitable for biological use. 38,39 Ab pH uptake into human and mouse microglial cell lines To visualize phagocytosis of Ab pH in real time in live microglial cells, immortalized human microglial clone 3 (HMC3) cells and mouse BV2 and N9 microglial cells were treated with 1, 2 and 5.0 mM concentrations of Ab pH and live-cell images were acquired every 30 minutes for 24 hours (Fig. 1D). We observed internalization and increased uorescence of Ab pH (implying phagocytosis) by HMC3 cells. The uorescence was quantied as a phagocytic score, i.e. relative uorescence compared to initial time (t ¼ 0) normalized over the 24 hour period (see Methods). For HMC3 cells there was an initial rapid phase of uorescence (score) increase followed either by a slower increase in uorescence at 5 mM Ab pH concentration or a slow decrease of uorescence from its peak value at 1 mM and 2 mM Ab pH concentration ( Fig. 1E and S12 †). This suggests rapid initial uptake of Ab pH , followed by intracellular degradation of Ab pH which occurs either more rapidly than the inux (giving a slow decline) or less rapidly than the inux (giving a slowed increase) (Fig. S12 †). Cells that did not phagocytose Ab pH did not display any green uorescence thereby differentiating Ab pH -specic phagocytic and non-phagocytic microglial cells in real time. Rodent microglial cell lines (BV2 and N9) showed a peak of phagocytic score at 12-16 hours for N9 and 16-20 hours for BV2 at 5 mM Ab pH treatment, compared to the HMC3 human microglial cell line that showed a gradual increase in phagocytosis over the 24 hour treatment period for the same concentration. Interestingly, for the lower Ab pH doses of 1 mM and 2 mM, the peak value of phagocytic score for HMC3 cells was within the initial 4 hours compared to the gradual increase for the rodent cell lines over the 24 hour period (Fig. S12 †). Using live-cell imaging, we also observed interesting morphological differences over time between phagocytic and non-phagocytic microglial cells. During the initial 2 hours, many cells displayed an elongated, branched morphology followed by acquisition of an amoeboid morphology during subsequent time points when phagocytosis was occurring, as indicated by increased uorescence (Movies S1-S3 †). Thus, the Ab pH reporter can be used to visualize Ab-specic phagocytosis in real-time and can be used in experiments to evaluate enhancement or inhibition of microglial phagocytosis for in vitro screening of drug candidates for AD.
Flow cytometry of Ab pH phagocytic cells and staining of Ab pH aer phagocytosis in xed primary cultured microglia and astrocytes We next determined whether Ab pH can be used to analyze phagocytosing cells using ow cytometry. By adding Ab pH to BV2 cells in vitro, we show that live microglial cells that phagocytose Ab pH can be easily analyzed with ow cytometry without the need for traditional dyes or antibodies to detect Ab ( Fig. 1F and S13A †). Phagocytic uptake of Ab pH by BV2 microglia was evident with a green uorescence peak within live cells when the cells were treated with 0.5 mM and 5.0 mM Ab pH for 1 hour in culture, with unstained and live/dead stained cells as controls. There was a slight increase in Ab pH uorescence at 1 hour when the Ab pH concentration was increased 10-fold. The green uorescence signal indicates internalization of Ab pH into the cellular acidic organelles, thereby avoiding detection of peptide sticking to the cell surface. Ab pH uorescence aer phagocytosis is sufficiently bright to enable FACS experiments, allowing single cell analysis of Ab pH+ phagocytic and Ab pHÀ nonphagocytic microglia and other glial cells.
Fixing cells that have internalized Ab pH , and using specic antibodies and dyes to study specic molecular processes, can help identify molecular mechanisms involved during Ab phagocytosis. To determine whether Ab pH can maintain its uorescence in xed cells, we used primary mouse CD11b + microglial cells isolated from 3-5 month old mice as well as BV2, N9, and HMC3 microglia. Aer Ab pH (5.0 mM) treatment for 2 hours, the cells were xed in 4% paraformaldehyde followed by addition of red phalloidin dye (a podosome core marker for F-actin) to visualize actin laments and the cell body with confocal microscopy ( Fig. S14A and B †). Confocal imaging of the xed Ab pH -treated microglia showed green uorescence within the red actin laments thereby conrming the uptake of Ab pH peptides by the cells. Further, LysoTracker DND-99 was used to conrm localization of Ab pH within acidic organelles such as phagosomes or phagolysosomes. The co-localization of green uorescent Ab pH along with the red signal from the LysoTracker dye conrmed the presence of Ab pH within acidic phagolysosomes aer 2 hours in HMC3, N9, and BV2 microglial cell lines ( Fig. 2A and S14C †). Similarly, uptake of Ab pH into intracellular acidic organelles was also conrmed in CD11b + primary microglia that were cultured in reduced-serum TIC medium 40 (Fig. 2B). Finally, we also detected the green uorescence signal within CD11b + primary microglia (Ab pH+ cells) at 0.5, 1.0, and 2.0 mM Ab pH concentrations via ow cytometry ( Fig. 2C and S13B †) aer 1 hour of treatment, and found an increase in uorescence with Ab pH concentration. The unstained and live/dead stained cells were used as controls.
Microglia recognize and phagocytose Ab peptides through scavenger receptors such as Toll-Like Receptor 2 (TLR2), Cluster of Differentiation 14 (CD14), and Triggering Receptors Expressed on Myeloid Cells 2 (TREM2). [41][42][43] Studies have shown that deletion of, or mutations in Ab-specic receptors, such as TREM2, leads to increased Ab seeding. 44,45 Thus, the Ab pH reporter that we have developed could serve as a valuable chemical tool to delineate the role of receptor proteins involved in Ab uptake by glial cells, using specic antibodies or CRISPR-Cas9 genetic screens.
In addition to microglia, astrocytes have been shown to exhibit phagocytic characteristics aer ischemic injury 16 and can phagocytose extracellular Ab. 28 Therefore, we tested whether primary immunopanned cultured rat astrocytes phagocytosed Ab pH . Indeed, this was the case, with retention of the phagocytic Ab pH signal aer methanol-xation and staining of cells with GFAP antibody (Fig. 2D). Ab pH phagocytosis could also be detected with live cell imaging (Fig. 2E). Quantication of Ab pH uptake at different concentrations showed internalization increasing for approximately 1 hour, followed by a sustained uorescence within the cells which may reect a balance between phagocytosis and degradation of the probe (Fig. 2F). Thus, Ab pH may be a viable candidate for concentration-and time-dependent studies of glial cell phagocytosis in vitro. The sustained uorescence of Ab pH seen for up to 6 hours inside cultured microglia and astrocytes suggests that it is chemically stable under physiological conditionsallowing for long-term use in vivo.
Ab pH uptake by microglia and astrocytes in hippocampus, cortex, and retina To assess phagocytic uptake of Ab pH in the hippocampus, a brain area that is crucial for learning and memory, we applied 5 mM Ab pH (for 1.5 hours at 37 C) to live hippocampal slices from postnatal day 12 (P12) rats. The tissue slices were then xed and stained with glial cell specic antibodies (Fig. 3A). Microglia phagocytosed Ab pH as seen by the localization of Ab pH within the IBA1 + myeloid cells in these tissues (Fig. 3B). Our experiments revealed green uorescent signal both within IBA1 + microglial cells as well as outside the IBA1 + cells ( Fig. S15 †), presumably reecting phagocytic uptake of Ab pH by cells other than microglia, such as astrocytes. Indeed, staining with GFAP antibody demonstrated internalization of Ab pH within GFAP + astrocytes (Fig. 3C). Summing over cells, approximately 60% of the internalized Ab pH was phagocytosed by microglia and 40% by astrocytes (Fig. 3D). Interestingly, in astrocytes the phagocytosed Ab pH was distributed more homogeneously throughout the GFAP area than the phagocytosed Ab pH in IBA1 stained microglia. This is expected since in microglia the acidic compartments (where Ab pH is present) are mostly juxtanuclear whereas IBA1 is known to be enriched in podosomes and podonuts. 46 In contrast, in astrocytes the acidic compartments occur all over the cell body. 47 We next tested the Ab pH probe in different in vivo settings. Aer intracranial injection of Ab pH into the somatosensory cortex of wild-type P7 C57BL/6J mice (Fig. 4A), phagocytosis of the Ab pH by IBA1 + microglia and GFAP + astrocytes was assessed by xing the corresponding tissue sections at 24 and 72 hours aer injection. At 24 hours, Ab pH was observed in the injected area and appeared to be enclosed within cell bodies. There was less astrocyte uptake of Ab pH compared to uptake by microglia, which may be a result of a reactive response by astrocytes that has been shown to downregulate phagocytic pathways in some states 5 (Fig. 4B and D). At 72 hours aer the Ab pH injection there was also Ab pH visible at the pial surface and in the periventricular white matter. With increased magnication, we observed cell bodies containing Ab pH that were positive for the IBA1 microglial marker (Fig. 4C) but the GFAP + astrocytes localized at the injection site showed very little Ab pH signal (Fig. 4D). Thus, microglia engulf the majority of the Ab pH under these conditions in vivo with astrocytes also contributing to removal at early times.
Next, we injected Ab pH into the vitreous of the eye in postnatal rats to eliminate possible complications due to glial  scarring resulting from cortical injection, and low penetration through the blood-brain barrier as a result of peripheral delivery (Fig. 4E). We injected 1 mL of Ab pH intravitreally and le the animals for 3, 24, or 72 hours. At the end of the experiment, retinae were removed, xed in 4% paraformaldehyde, and whole mount retinal preparations were made. Astrocytes and microglia in the retinal ganglion cell layer were labelled with GFAP and IBA1 antibodies respectively, and co-localization with the uorescent signal from the injected Ab pH probe was determined. We observed that IBA1 + microglia contained Ab pH at all time points, but we did not detect any Ab pH signal in GFAP + astrocytes ( Fig. 4F and G). The Ab pH positive microglia were less visible in the retina compared to other brain regions, suggesting that clearance from the eye was more rapid than clearance from the brain parenchyma as observed in other experiments (above).
The in vivo investigation of Ab phagocytosis described above consists of imaging xed tissue sections with confocal microscopy. To date, phagocytosis of Ab has not been observed in live animals in real time. Using Ab pH , we observed phagocytic uptake in live mice in real time using two-photon microscopy (Movies S4 and S5 †). Here, 5.0 mM Ab pH was applied onto the cortical surface of live mice through a cranial window for 10 minutes during two-photon imaging of the barrel cortex (Fig. 5A). This showed an increase in green uorescence in the cell somata aer Ab pH application, indicating phagocytic uptake of Ab pH (Fig. 5B). Quantication of the uorescence within the cell somata showed rapid Ab pH uptake aer the peptide was added to the cranial window followed by stabilization of the signal at around 20 minutes (Fig. 5C). The brains were later xed by cardiac paraformaldehyde perfusion and stained with antibodies to identify the cell types mediating the Ab pH uptake. Phagocytosed green Ab pH was present in IBA1 labeled microglia and colocalized with the microglial/macrophage lysosomal protein CD68, and Ab pH was also seen in GFAP + astrocytes (Fig. 5D). Integrating the uorescence over the different classes of labeled cells showed that about 70% of the phagocytosed Ab pH was taken up by microglia and about 25% by astrocytes (Fig. 5E). These percentages are similar to those seen in the brain slice experiments described above. Thus, microglial uptake of Ab pH dominates over astrocyte uptake.

Discussion
The phagocytic capacity of glial cells has been measured using traditional dyes and uorophore-labeled particles such as uorescein or rhodamine-coated E. coli and beads, however, these experiments entail some major drawbacks: (1) difficulty in differentiating between adherent versus internalized particles, (2) use of additional reagents (like trypan blue or ethidium bromide) and additional experimental steps required to quench the extracellular uorescence prior to analysis by ow cytometry and imaging, 48 (3) quenching of the uorescence within the acidic environment of the phagosomes, and (4) lack of speci-city of the neuronal target substrate (beads, E. coli, zymosan, etc.). 49 Here, we have synthesized a pH-sensitive uorogenic Ab reporter, Ab pH , at a microgram scale and characterized the reporter in detail (Fig. 1A-C and S2-S11 †). We show that the Ab pH is well suited for detection in the acidic environment of the lysosomal organelles with pH of 5.0-4.5. The PTXG dye has an acidic pK a of 4.4-4.7 in DMEM/F12 cell culture medium and exhibits a gain of uorescence in the acidic environments. Upon conjugation with Ab, the pK a of Ab pH probe shis to 5.2-5.9, which is in the range of acidic lysosomal organelles and is thus ideal for future mechanistic studies. Moreover, the unconjugated PTXG dye exhibits low to no background uorescence signal, compared to other commercial dyes such as pHrodo-Red SE, justifying its in vitro and in vivo use. We demonstrated the functionality of Ab pH in mouse and human microglial cell lines, in primary microglial and astrocyte cultures in vitro, in acute hippocampal slices from mouse brains, in mice in vivo using stereotaxic injections followed by xation, and in live mice in vivo via two-photon imaging. The Ab pH reporter is a powerful tool for answering questions related to mechanisms of Ab phagocytosis and cellular inammatory responses, and for developing therapeutic strategies to promote Ab clearance.
We observed phagocytic uptake of Ab pH in three different microglial cell lines and in primary glial cell cultures in realtime (Fig. 1D, E, 2E and F) and by using confocal imaging ( Fig. 2A, B and D). We showed the utility of Ab pH to be used without the need for any antibodies to identify or detect Ab using ow cytometry (Fig. 1F and 2C) which will greatly benet experimental design and outcome by reducing the number of steps required in assays. While microglia cultured in vitro and ex vivo do not completely recapitulate the transcriptomes of microglia in vivo, 50 microglial cell culture models provide a convenient system for screening chemical and biological molecules in a rapid and high-throughput manner. We also demonstrated the utility of Ab pH in functional assays to evaluate cellular phagocytosis. Further, we demonstrated the phagocytic uptake of Ab pH in live tissue hippocampal slices that preserve the three-dimensional nature of the cell microenvironment and can serve as an additional useful model to study Ab-related biology in a tissue environment. Microglia were more efficient in internalizing Ab pH than astrocytes in situ (Fig. 3A-D). During the period of this assay it is unlikely that astrocytes would have become reactive so it appears that Ab clearance is not just a property of reactive astrocytes. [51][52][53] The astrocytes most closely associated with neurodegeneration appear to be of a reactive phenotype that downregulates phagocytic pathways, 5 suggesting that the engulfment of Ab pH by astrocytes is a normal physiological function of these cells (although this may change with disease progression).
Most importantly, we demonstrated the utility of using of Ab pH in various in vivo models. Stereotaxic injection of Ab pH into the mouse somatosensory cortex was followed by uptake by microglia and astrocytes (Fig. 4A-D), injecting Ab pH into the vitreous of the eye was followed by uptake by retinal microglia (Fig. 4E-G), and the real-time uptake of Ab pH by microglia and astrocytes was demonstrated in vivo in live mice using twophoton microscopy ( Fig. 5A-C). We conrmed the identity of the cells observed with two-photon excitation by using IBA1 to label microglia, CD68 (a phagocytosis-specic marker) to label microglia and macrophages, and GFAP to label astrocytes ( Fig. 5D and E). The time course of uptake in vivo was rapid, with phagocytosed Ab pH uorescence reaching a peak level within 20 minutes (this may reect a balance between continued phagocytosis and intracellular degradation). Approximately two thirds of the Ab pH was phagocytosed by microglia and one third by astrocytes, again on a time scale too rapid for astrocytes to have become reactive.
The development of Ab pH , in particular the ease with which it can be produced in the lab, will facilitate the characterization of different populations of cells that remove (or do not remove) Ab by phagocytosis in various conditions, including AD, traumainduced amyloidopathy and Down's syndrome. Combining this tracer with transgenic labeling of microglia (e.g. Tmem119-tdTomato mice 54 which exhibit red uorescence in microglia) or other cell types will allow the transcriptome and proteome of different subpopulations of heterogeneous phagocytic cells to be dened (including cells of different age, 25,55,56 and sex 57 ), facilitating the discovery of new mechanisms, targets and functional biomarkers. Understanding the clearance of Ab is fundamental to understanding the onset of AD, 58,59 and having a quantitative technique to assess Ab phagocytosis should contribute signicantly to this. We expect that Ab pH will become a useful tool to facilitate the study of impaired phagocytic function mechanisms in vivo during chronic inammation in neurodegenerative diseases.

Animal ethics
Animal maintenance and isolation of primary microglia were performed according to Purdue Animal Care and Use Committee guidelines and approval (protocol number 1812001834). Injection of Ab pH into the retina of rats were completed in accordance with the National Institute of Health Stanford University's Administrative Panel on Laboratory Animal Care. Purication of rat primary astrocyte was completed in accordance with NYU Langone School of Medicine's Institutional Animal Care and Use Committee (IACUC) guidelines and approval. All rats were housed with ad libitum food and water in a 12 hour light/dark cycle. Standard Sprague Dawley rats (Charles River, #400) were used in all retinal experiments. Animal maintenance and experimental procedure for the intracranial stereotaxic injections of Ab pH in mice were performed according to the guidelines established by the IACUC at Harvard Medical School (protocol number IS00001269). Experiments on brain slices and in vivo 2-photon imaged mice were carried out under a UK government license (PPL 70/8976, awarded aer local ethical review and UK Home Office assessment) to David Attwell, in accordance with all relevant animal legislation in the UK.

Synthesis of pH-sensitive uorescent human Ab conjugate with Protonex Green™ (AbpH)
Human amyloid-beta (Ab 1-42 ) was purchased from AnaSpec., Inc (Cat. #AS-20276, extinction coefficient at 280 nm based on single tyrosine residue is 1490 M À1 cm À1 ), Protonex™ Green 500 SE was from AAT Bioquest, Inc. (Cat. #21215, absorbance maximum 457 nm, extinction coefficient 4000 cm À1 M À1 , correction factor at 280 nm 1.069). To 200 mL aliquot of monomeric Ab 1-42 (1 mg mL À1 in 1 M NaHCO 3 , pH $8.3) was added 10 equivalents of Protonex-Green 500 (PTXG), SE dye (88 mL from 5 mM stock in anhydrous DMSO) and incubated at room temperature for 3 hours in the dark (vial wrapped in aluminum foil: note: add 100 mL ultrapure water if the solution becomes viscous). The additional 5 equivalents of PTXG, SE dye (44 mL from 5 mM DMSO stock) was added and incubated under the same conditions for 3 hours. The crude reaction mixture was diluted with 1 mL ultrapure water and the conjugated product was dialyzed by Pierce Protein Concentrators at 4500 g for 30-45 minutes in a swinging bucket centrifuge to remove the small molecular weight fragments [Pierce Protein Concentrators PES, 3K Molecular Weight Cut-Off (MWCO); Thermo Fisher Scientic, Cat #PI88514, note: prior use, wash and centrifuge protein concentrator with 1 mL ultrapure water to remove any preservatives]. The resulting concentrated solution was diluted with 0.5 mL ultrapure water and dialyzed again for 15-30 minutes as done previously. Then the concentrated solution was diluted with 0.2 mL ultrapure water and lyophilized overnight to get the Ab pH powder. Protein or dye concentration can be calculated using parameters mentioned above. Finally, MALDI-MS spectrum was recorded to conrm the chemical conjugation of PTXG with Ab 1-42 .

Synthesis of pH-sensitive uorescent human Ab 1-42 conjugate with pHrodo (RODO-AbpH)
The pHrodo-Red SE was purchased from Thermo Fisher Scientic (Cat. #P36600, absorbance maximum 560 nm, extinction coefficient 65 000 M À1 cm À1 , correction factor at 280 nm 0.12). A solution containing monomeric Ab 1-42 (570 mg, 12.63 nmol) was prepared in 1 M NaHCO 3 (pH 8.3, 570 mL) and the pHrodo Red-NHS (1 mg) stock solution was prepared in anhydrous DMSO (150 mL) ($10.2 mM). Next, the Ab solution (570 mL) and pHrodo stock solution (0.6314 mmol, 61.5 mL of stock solution) were mixed and incubated at room temperature for 6 hours while wrapped with aluminum foil. This crude reaction mixture was then diluted with ultrapure water (1 mL) and the conjugated product was dialyzed by Pierce Protein Concentrators (PES, 3K MWCO, note: prior use, wash and centrifuge protein concentrator with 1 mL ultrapure water to remove any preservatives) at 4500 g for 30-45 minutes in a swinging bucket centrifuge to remove the small molecular weight fragments. The resulting concentrated solution was diluted with ultrapure water (0.5 mL) and dialyzed again for 15-30 minutes as before. Then the concentrated solution was diluted with 0.2 mL ultrapure water and lyophilized overnight to get the RODO-Ab pH powder. Protein or dye concentration can be calculated using parameters mentioned above. Finally, the MALDI-MS spectrum was recorded to conrm the chemical conjugation of pHrodo-Red uorophore to the Ab peptide.

Preparation of HFIP-treated Ab and Ab pH stocks
Ab or Ab pH was dissolved in HFIP and prepared as previously described. 37 Briey, 1 mM Ab solution was prepared by adding HFIP directly to the vial (0.5 mg Ab or Ab pH in 93.35 mL HFIP). The peptide should be completely dissolved. The solution was incubated at room temperature for at least 30 min. HFIP was allowed to evaporate in the open tubes overnight in the fume hood and then dried down under high vacuum for 1 hour without heating to remove any remaining traces of HFIP and moisture, leaving a thin clear lm of peptide at the bottom of the tubes. The tubes containing dried peptides were stored at À20 C until further use. To make oligomers, 5 mM Ab DMSO stock was prepared by adding 22 mL fresh dry DMSO to 0.5 mg of dried peptide lm. To ensure complete resuspension of peptide lm, the mixture was pipetted thoroughly, scraping down the sides of the tube near the bottom. The suspension was vortexed well ($30 seconds) and pulsed in a microcentrifuge to collect the solution at the bottom of the tube and the 5 mM Ab DMSO solution was sonicated for 10 minutes. This preparation was used as the starting material for preparing the aggregated Ab incubated at 4 C for 24 hours for analysis by atomic force microscopy.

Atomic force microscopy for analysis of Ab and Ab pH aggregates
We followed the previously published detailed protocols for analyzing Ab and Ab pH aggregates by atomic force microscopy (AFM). 37,38 Briey, sample preparation was performed with sterile techniques using sterile media and MilliQ-water. A 10 mL syringe with ultrapure water equipped with a 0.22 mm lter was lled and the initial 1-2 mL was discarded though syringe lter output. 1 M HCl and 1Â PBS buffer were also ltered through 0.22 mm lter. The samples were prepared for spotting on mica by diluting to nal concentrations of 10-30 mM in water. Immediately before sample delivery, top few layers of mica were cleaved away using an adhesive tape to reveal a clean, at, featureless surface. The fresh surface was pretreated with $5-8 mL of ltered 1 M HCl for 30 seconds and rinsed with two drops of water (note: the mica was held at a 45 angle and washed with water to allow the water coming out of the syringe lter to roll over the mica). If necessary, the remaining water was absorbed with ber-free tissue paper/wipes by keeping paper on the edge of the mica. Immediately, the sample was spotted onto mica and incubated for 3 minutes followed by rinsing with three drops of water and blow drying with several gentle pulses of compressed air. Samples were then kept in a dust-free box and incubated on benchtop for a few minutes to hours at room temperature until analysis. AFM imaging was performed with Veeco Multimode with NanoScope V controller with NanoScope Soware using the Silicon AFM probes, TAP300 Aluminum reex coating (Ted Pella, Inc. Cat# TAP300AL-G-10) at $300 kHz resonant frequency and $40 N m À1 force constant in the tapping mode.

pH-dependent emission spectra of Ab conjugated with Protonex Green® and Ab conjugated with pHrodo at various concentrations
The cell culture medium was supplemented with dilute HCl and NaOH solutions to obtain different solutions of pH ranging from 1.0 to 9.0 for the assay. Lyophilized powder of Ab conjugated with pHrodo (RODO-Ab pH ) and Ab conjugated with Protonex Green® (PTXG-Ab pH ) was dissolved in cell culture medium to make stock solutions and kept at 37 C for 24 hours to pre-aggregate the peptide conjugates. From the stock solutions, different dilutions for each pH condition were prepared at concentrations of 0.5, 1.0, 2.0, and 5.0 mM in a 96-well plate (100 mL per well). Fluorescence intensity of each well containing Ab pH was obtained on a Cytation™ 5 imaging multi-mode reader (BioTek Instruments) at 443/505 nm excitation/ emission wavelengths. The uorescence intensity of each pHsolution and Ab pH -concentration in relative uorescence units (RFU) was plotted using GraphPad Prism soware. pK a of Protonex Green 500, SE® (PTXG), Ab pH , and RODO-Ab pH The cell culture medium DMEM/F12 (Corning Cat.# MT15090CV) was supplemented with dilute HCl and NaOH solutions to obtain different pH solutions (pH range 3.85, 4.03, 4.26, 4.46, 4.88, 5.02, 5.30, 5.67, 6.22, 6.44, 6.94, 7.40, 8.31) for the assay. From the stock solutions (5 mM of PTXG and 100 mM of Ab pH ) different dilutions for each pH condition were prepared at concentrations of 5.0, 2.0, 1.0, 0.5 mM in a 384-well plate with 40 mL per well total volume (Corning NBS plate, cat #3575). Fluorescence intensity of each well containing PTXG and Ab pH was obtained on a Varioskan LUX imaging multimode reader (Thermo Scientic) at excitation/emission wavelengths 443/505 nm for PTXG or Ab pH and excitation/emission wavelengths 466/590 nm for RODO-Ab pH . The uorescence intensity was normalized by diving each value by the highest value obtained for 5 mM concentration. The uorescence intensity of each pH-solution vs. Normalized uorescence intensity for each concentration was plotted using three replicates and processed using GraphPad Prism soware.

Concentration-dependent response of Protonex Green 500, SE® and Ab pH at acidic pH over time
Solutions of Protonex Green 500, SE® (PTXG) and Ab conjugated with PTXG (Ab pH ) were prepared in the cell culture medium at concentrations of 0, 0.1, 0.5, 1.0, 2.0, 5.0 mM and a 50 mL aliquot of each solution was transferred in duplicates into the wells of a 96-well plate. To measure the orescence of PTXG and Ab pH solutions under acidic conditions at different concentrations, 7.5 mL of pH 1.0 solution (hydrochloric acid in media) was added to each well to obtain a nal pH of 3.0. Fluorescence intensities of the acidic solutions were measured at an excitation/emission wavelength of 443/505 nm on A Cytation 5 multimode plate reader (BioTek, Inc). Next, to initiate aggregation of Ab pH , the plate was incubated at 37 C with 5% CO 2 and uorescence was measured at 2, 6, 12, and 24 hour time points. The change in uorescence intensities of the PTXG and Ab pH aggregates was analyzed over time using GraphPad Prism soware.
Cell linesculture and maintenance BV2 and N9 mouse microglial cell lines were generously gied by Dr Linda J. Van Eldik (University of Kentucky, USA). The BV-2 cell line was developed in the lab of Dr Elisabetta Blasi at the University of Perugia, Italy. Cells were maintained at 37 C and 5% CO 2 in DMEM (Dulbecco's Modied Eagle's Medium)/Hams F-12 50/50 Mix (Corning #10-090-CV) supplemented with 10% FBS (Atlanta BiologiBiologics), 1% L-glutamine (Corning #25-005-CI), and 1% penicillin/streptomycin (Invitrogen). HMC3 human microglial cell line was a gi from Dr Jianming Li (Purdue University, USA) who originally obtained the cells from ATCC. These cells were maintained at 37 C and 5% CO 2 in DMEM supplemented with 10% FBS and 1% penicillin/ streptomycin.
Background uorescence of PTXG and RODO in cells BV2 cells were seeded at a concentration of 10 000 cells in 200 mL per well in a 96-well at-bottom plate (Falcon) for 16 hours (overnight). The next day, the media was aspirated, and the cells were treated with 1.0 mM or 0.5 mM of PTXG, PTXG-Ab pH , RODO, and RODO-Ab pH and placed in a 37 C incubator for 2 hours respectively. Next, the uorescence of PTXG and PTXG-Ab pH was measured at 443/505 nm and the uorescence of RODO, and RODO-Ab pH was measured at 560/585 nm using a uorescence plate reader (Attune NxT, Thermo Fisher Scien-tic, USA) to evaluate the cellular uorescence indicating uptake of the unconjugated and Ab-conjugated dyes.

Lactate dehydrogenase (LDH) activity cytotoxicity assay
The cytotoxicity of the PTXG and RODO dyes were evaluated with LDH assay per manufacturers protocol (CyQUANT LDH Cytotoxicity Assay; Thermo #C20300). Briey, 5000 BV2 cells/100 mL were seeded onto the wells of a 96-well at bottom plate (Falcon) for 16 hours (overnight). The next day, the cells were treated with the PTXG and RODO dyes (25, 12.5, 6.25, 3.125, 1.56, 0.78, and 0.39 mM) for 24 hours. Cells without any dye treatment were used as a negative control (to measure the spontaneous LDH activity). Cells treated with the provided lysis buffer were used as the positive controls (to measure the maximum LDH activity that is later set to 100%). The % cytotoxicity of the PTXG and RODO dyes were calculated per the equation below and the data was plotted for n ¼ 3 biological replicates, mean + sd.

Phagocytosis assay with live microglia
Cells were seeded at 5000 cells per well (200 mL per well) in a 96well at bottom plate (Falcon) for approximately 16 hours (overnight). For all cell assays, the lyophilized Ab pH conjugate was dissolved in the culture medium to prepare a stock solution and was pre-aggregated by placing the stock solution at 37 C for 24 hours. Further dilution for cell treatment was performed in culture media and the diluted solution was ltered using a 0.22 mm syringe lter prior to cell treatment. The adherent cells (BV2, N9, HMC3) were treated with a nal concentration of Ab pH at a nal concentration of 0, 0.1, 0.5, 1.0, 2.0, and 5.0 mM by replacing one-half of the culture medium (100 mL) with a stock Ab pH solution at 2Â concentration. Two technical replicates were used for each treatment concentration. The plates were immediately placed in an IncuCyte S3 Live-Cell Analysis System (Essen BioScience) and four images per well were captured at 30 minute time intervals for 24 hours. The uorescence intensity, cell conuence, and the integrated uorescence intensity data were obtained and analyzed using the GraphPad Prism soware.
The Ab pH uptake was measured as the phagocytic score metric, dened as a normalized value relative to the initial uorescence intensity at t ¼ 0 and calculated as: Phagocytic score ¼ relative total intensityðtÞ maximum relative total integrated intensity where, relative total integrated intensity (t) is dened as total integrated intensity (t) À total integrated intensity (t ¼ 0) for each concentration and cell type.
The total integrated intensity is dened as the total sum of Ab pH uorescence intensity in the entire image and given by the expression CU Â mm 2 Image as dened by Incucyte. We captured 4 images per well with multiple replicates for each Ab pH concentration and for each cell type. These images were used to calculate the value of total integrated intensity, CU Â mm 2 Image : The individual units are dened as: CU ¼ average mean intensity (the average of the Ab pH 's mean uorescence intensity in each cell in an image), mm 2 ¼ average area (the average area of the Ab pH 's in each cell in an image). The maximum relative total integrated intensity is the maximum value of relative total integrated intensity over the 24 hour period. Such a normalization gives phagocytic score values between 0 and 1 to compare different concentrations across different cell types. The maximum peak indicates the time when the degradation is equal to the uptake for each concentration and cell type. All other values show an interplay between uptake or degradation compared to time, t ¼ 0, shown by either increase or decrease in orescence that is also observed visually. The corresponding videos (Movies S1-S3 †) during live cell imaging were taken on IncuCyte S3 Live-Cell Analysis System (Essen BioScience) and stabilized using the Blender version 2.82a soware (https://www.blender.org).
Effect of balomycin-A1 (BF) on uorescence of Ab pH in cells BV2 cells were seeded at a concentration of 50 000 cells/500 mL media per well in a 12-well at bottom plate (Falcon) for 16 hours (overnight). The next day, the media was aspirated, and the cells were treated with 400 nM balomycin (BF) and placed in a 37 C incubator for 1 hour. Aer this time period, the solution was removed and replaced with media containing 100 nM of Ab pH with or without 400 nM BF for 1 hour. Finally, this solution was removed and the cells were detached from the plates using ice-cold 1Â PBS with gentle pipetting. Three minutes before the analysis of each sample, DAPI was added (0.1 mg mL À1 cell suspension) to stain for dead cells. The % cytotoxicity ¼ ½PTXG or RODO-treated LDH activity À spontaneous LDH activity ½maximum LDH activity À spontaneous LDH activity uorescence of Ab pH in the cells were analyzed by ow cytometry (Attune NxT, Thermo Fisher Scientic, USA). Briey, at least 80% of the total cells were rst gated on SSC-A vs. FSC-A plot and the single cells within this gate were selected on the FSC-H vs. FSC-A plot. From the single cells, the live (DAPI Àve) and dead cells (DAPI +ve) were identied. Finally, Ab pH+ or Ab pHÀ cells were identied within the live cell population in the Ab pH+ histogram plot. The data was plotted as median uorescence intensity (MFI) relative to MFI of Ab pH only for n ¼ 2 biological replicates (Fig. S10 †).
Isolation and culture of primary mouse microglia CD11b + primary microglia were isolated from adult mice aged around 7 months (male and female) and cultured as follows.
Mice were euthanized with CO 2 following the Purdue University Animal Care and Use Committee guidelines and brains were transcardially perfused with ice-cold PBS. The perfused brains were dissected and cut into small 1 mm 3 pieces before homogenizing them in Dulbecco's Phosphate Buffered Saline ++ (i.e. with Ca 2+ and Mg 2+ ions) containing 0.4% DNase I on a tissue dissociator (Miltenyi Biotec) at 37 C for 35 min. The cell suspension was ltered through a 70 mm lter and myelin was removed two times, rst using Percoll PLUS reagent followed by myelin removal beads using LS columns (Miltenyi Biotec). Aer myelin removal, CD11b + cells were selected from the single cell suspension using the CD11b beads (Miltenyi Biotec) as per the manufacturer's instructions. The CD11b + cells were nally resuspended in microglia growth media made in DMEM/F12 (Corning Cat. #MT15090CV), further diluted in TIC (TGF-b, IL-34, and cholesterol) media 40 with 2% FBS before seeding 0.1 Â 10 6 cells per 500 mL in a well of a 24-well plate (Corning Cat. #353847). The cells were maintained in TIC media at 37 C and 10% CO 2 with media change every other day until the day (around 10-14 div) of the phagocytosis assay (around 10-14 div).
Flow cytometry analysis of BV2 and primary microglial phagocytosis BV2 microglial cells were seeded at a density of 250k cells per well in a 6-well plate for around 14 hours overnight. The next morning, the cells were treated with 0.5 mM and 5 mM Ab pH and placed in a 37 C incubator for 1 hour, aer which the plate was brought to the hood, placed on ice to stop phagocytosis, and cell culture medium containing Ab pH was aspirated. The cells were washed once with cold PBS. Next, the cells were treated with ice cold PBS containing 2 mM EDTA for 2 minutes on ice to detach the cells from the wells. The cells were then centrifuged at 1400 rpm for 3 minutes. The supernatant was aspirated, and the cell pellets were re-suspended in FACS buffer (PBS, 25 mM HEPES, 2 mM EDTA, and 2% FBS). Five minutes before analysis of each sample, propidium iodide (PI; Thermo Fisher Scientic, Cat. #P1304MP) was added to the sample (40 ng mL À1 cell suspension) for staining of dead cells. Phagocytosis of Ab pH by primary microglial cells was analyzed on div 10-14 in a similar manner. Cells were treated with Ab pH for 1 hour and detached from the plate using cold PBS and 2 mM EDTA. Aer centrifuging the cell suspension, the cell pellet was resuspended in 0.1 mL PBS for live/dead staining. Here, Zombie Violet, ZV, (BioLegend, #423113) was used to evaluate cell viability (1 : 100 per 10 6 cells in 0.1 mL) for 15 min followed by PBS wash. Finally, the cells were resuspended in FACS buffer and taken for analysis. Cells exhibiting green uorescence were captured on the FITC channel upon gating for live cells on Attune NxT ow cytometer (Invitrogen). The les were then analyzed on FlowJo V10 soware. Briey, at least 90% of the total cells were rst gated on SSC-A vs. FSC-A plot and the single cells within this gate were selected on the FSC-H vs. FSC-A plot. From the single cells, the live and dead cells were identied from the viability dye. PI À and ZV À cells were considered as live cells and PI + and ZV + cells were taken as dead cells on the histogram plots. Finally, Ab pH+ or Ab pHÀ cells were identied within the live cell population in the Ab pH histogram plot on the FITC channel.
Confocal imaging of actin laments and nuclei in the paraformaldehyde-xed phagocytic microglial cells For labeling the cells with phalloidin and DAPI, 20 000 cells/250 mL were plated in 14 mm microwells of 35 mm glass bottom dishes (MatTek #P35G-1.5-14-C) and kept overnight. The cells were treated with 5.0 mM Ab pH for 2 hours on the next day. Then the medium was aspirated, and cells were xed with 4% paraformaldehyde for 20 minutes, and then gently washed once with PBS. Phalloidin-iFluor 594 reagent (Abcam, Cat. #ab176757; 1000Â stock) was diluted to 1Â in PBS and added to the xed cells for 10 minutes for staining the actin laments. To label the nuclei, DAPI was diluted to a concentration of 1 mg mL À1 . The cells were washed again with PBS followed by a 10 minute incubation with the diluted DAPI solution. Finally, the DAPI solution (Invitrogen, Cat. #D3571) was aspirated and the xed cells treated with 2-3 drops of ProLong Gold Antifade Mountant (Invitrogen #P36930) before imaging. Fluorescence images of phagocytic microglial cells were captured using 40Â and 60Â objectives on a Nikon AR-1 MP confocal laser microscope. Images were obtained using the NIS Elements microscope imaging soware.

Confocal imaging of intracellular acidic organelles and nuclei in paraformaldehyde-xed phagocytic microglial cells
LysoTracker Red DND-99 (Thermo Fisher Scientic, Cat. #L7528) was used for labeling the intracellular organelles of the cells to observe the subcellular localization of Ab pH sensors inside the cells aer phagocytosis. 20 000 cells/250 mL were plated in 14 mm microwells of 35 mm glass bottom dishes (MatTek #P35G-1.5-14-C) and kept overnight. The cells were treated with 5.0 mM Ab pH for 2 hours on the next day. Then the Ab pH -containing medium was aspirated and replaced with 200 mL of media containing 100 nM concentration of the Lyso-Tracker dye and the cells were incubated in a 37 C, 5% CO 2 incubator for 30 minutes. Finally, the cells were xed and treated with DAPI to stain the nuclei followed by 2-3 drops of ProLong Gold Antifade Mountant using the above-mentioned protocol. Fluorescence images of phagocytic microglial cells were captured using 40Â and 60Â objectives on a Nikon AR-1 MP confocal laser microscope. Images were obtained using the NIS Elements microscope imaging soware.

Intracranial injection of Ab pH , perfusion and immunohistochemistry
Lyophilized Ab pH was dissolved in Hank's Balanced Salt Solution (HBSS no calcium, no magnesium, no phenol red, Ther-moFisher #14175079) to obtain a 100 mM stock solution that was then briey vortexed and sonicated in a bath sonicator for 1 minute and used immediately for intracranial injections or stored at À80 C. For intracranial injections, the stock solution was diluted in HBSS to obtain a 10 mM working solution and kept on ice to prevent aggregation. Postnatal day 7 mice were anesthetized with isourane and were mounted on a stereotaxic frame. Two lots of 250 nL of Ab pH working solution were unilaterally injected in the somatosensory cortex of wild-type C57BL/6J mice using a Nanoliter Injector (anteroposterior À2.3 mm; mediolateral +2.3 mm; dorsoventral À0.5 mm for lower layers and À0.2 mm for upper layers, relative to Lambda) at an injection rate of 100 nL min À1 followed by 2 additional minutes to allow diffusion. 24 or 72 hours aer injections, animals were deeply anesthetized with sodium pentobarbital by intraperitoneal injection and then transcardially perfused with PBS 1Â followed by 4% paraformaldehyde (PFA) in PBS. Brains were dissected out, post-xed for two hours at 4 C, and cryoprotected in a 30% sucrose-PBS solution overnight at 4 C. Then, tissue was sectioned at 40 mm on a sliding microtome (Leica). Free-oating brain sections were permeabilized by incubating with 0.3% Triton X-100 in PBS for 1 hour and then blocked for 3 hours (0.3% Triton X-100 and 10% normal donkey serum), followed by incubation with primary antibodies in 0.3% Triton X-100 and 10% normal donkey serum overnight at 4 C. The next day, brains were rinsed in PBS 1Â for 1 hour, incubated with the appropriate secondary antibodies for 2 hours at room temperature, rinsed again in PBS, incubated with DAPI and mounted using Fluoromount-G (SouthernBiotech, #0100-01). During perfusion and immunohistochemistry all solutions were maintained at neutral pH. The following primary antibodies were used: mouse anti-GFAP (1 : 200, Sigma #G3893-100UL) and rabbit anti-IBA1 (1 : 500, Wako Chemicals, #019-1974). The secondary antibodies used were donkey anti-mouse-IgG1 647 (Invitrogen, #A21241) and donkey anti-rabbit 594 (ThermoFisher, #A-21207). Tissue samples were imaged on a ZEISS Axio Imager and a ZEISS LSM 800 confocal using a 20Â objective. Ab pH uorescence signal was quantied using ImageJ. First, cell contour was manually drawn using IBA1 or GFAP signal and mean uorescence intensity in the Ab pH channel was measured. Next, the selection was moved to a nearby region with no obvious Ab pH uorescence signal and mean uorescence intensity in the Ab pH channel was measured (background). Normalized uorescence intensity for each cell was calculated as Ab pH uorescence signal minus background. Data were analyzed by one-way ANOVA followed by the Sidak's post hoc analysis for comparisons of multiple samples using GraphPad Prism 7 (GraphPad Soware).

Immunopanning and culture of primary astrocytes
Astrocytes were puried by immunopanning from the forebrains of P5 Sprague Dawley rats (Charles River) forebrains and cultured as previously described. 60 In brief, cortices were enzymatically disrupted (using papain) and then mechanically dissociated to produce a single-cell suspension that was incubated on several negative immunopanning plates to remove microglia, endothelial cells and oligodendrocyte lineage cells.
Positive selection for astrocytes was with an ITGB5-coated panning plate. Isolated astrocytes were cultured in a dened, serum-free base medium containing 50% neurobasal, 50% DMEM, 100 U mL À1 penicillin, 100 mg mL À1 streptomycin, 1 mM sodium pyruvate, 292 mg mL À1 L-glutamine, 1Â SATO and 5 mg mL À1 of N-acetyl cysteine. This medium was supplemented with the astrocyte-required survival factor HBEGF (Peprotech, 100-47) at 5 ng mL À1 . 60 Cells were plated at 5000 cells per well in 12-well plates coated with poly-D-lysine and maintained at 10% CO 2 .
Engulfment assay of Ab pH by primary astrocytes Astrocytes were maintained for 1 week in culture and checked for their reactivity state using qPCR 5 before addition of Ab pH . The cells were treated with 0.5, 1.0, or 2.0 mM Ab pH and imaged continuously with still images taken every 5 minutes with an IncuCyte S3 System epiuorescence time lapse microscope to analyze engulfed Ab pH particles. For image processing analysis, we took 9 images per well using a 20Â objective lens from random areas of the 12-well plates and calculated the phagocytic index by measuring the area of engulfed Ab pH particles (uorescence signal) normalized to the area of astrocytes, using ImageJ.
In vivo retinal engulfment of Ab pH P14 Sprague Dawley rats were anaesthetized with 2.5% inhaled isourane in 2.0 L O 2 per min. Once non-responsive, animals received a 1 mL intravitreal injection of Ab pH , or PBS. Retinae were collected for immunouorescence analyses at 3, 24, and 72 hours. At collection, eyeballs were removed, xed in 4% PFA overnight, and washed in PBS, and retinae were dissected and whole-mounts placed on silanized glass slides. Retinae were blocked with 10% heat-inactivated normal goat serum for 2 hours at room temperature. Incubation with primary antibodies to GFAP (Dako, A0063, 1 : 5 000) and IBA1 (WAKO, 019-19741, 1 : 500) diluted in 5% goat serum in PBS was followed by detection with AlexaFluor uorescent secondary antibodies (Thermo, 1 : 1000).

Brain slice experiments
Rats at postnatal day 12 (P12) were sacriced by cervical dislocation followed by decapitation, and 250 mm sagittal hippocampal slices were prepared on a Leica VT 1200S vibratome at 4 C in oxygenated solution containing (mM): 124 NaCl, 26 NaHCO 3 , 2.5 KCl, 1 NaH 2 PO 4 , 10 glucose, 2 CaCl 2 , 1 MgCl 2 , 1 kynurenic acid. Acute slices were allowed to recover for 2.5 hours at room temperature before being transferred to 24-well plates and incubated with 5 mM Ab pH in HEPES-based aCSF (140 mM NaCl, 10 mM HEPES, 2.5 mM KCl, 1 mM NaH 2 PO 4 , 10 mM glucose, 2 mM CaCl 2 , and 1 mM MgCl 2 ) for 1.5 hours at 37 C. Following incubation, slices were quickly rinsed in cold phosphate-buffered saline (PBS) and xed in 4% paraformaldehyde (PFA) for 45 minutes at room temperature. For immunolabeling, slices were permeabilized and blocked in buffer containing 10% horse serum and 0.02% Triton X-100 in PBS for 2 hours at room temperature, followed by incubation with goat anti-IBA1 (Abcam, ab5076) or chicken anti-GFAP (Abcam, ab4674) primary antibodies diluted 1 : 500 in blocking buffer for 12 hours at 4 C. Following four 10 minute washes in PBS, donkey anti-goat IgG 647 (ThermoFisher, A21447) or donkey anti-chicken IgG 649 (Jackson, 703-495-155) secondary antibodies diluted 1 : 1000 in blocking buffer were applied for 4 hours at room temperature. Finally, slices were incubated in DAPI for 30 minutes, rinsed in PBS and mounted. Imaging was done using a Zeiss LSM700 confocal microscope and a 20Â objective, where 10 mm image stacks at 1 mm step interval were acquired. For analysis, the percentage of Ab pH signal within microglial or astroglial cells was calculated by binarizing the microglia or astrocyte channel, creating a mask and multiplying it by the raw Ab pH signal. The uorescence intensity of Ab pH colocalizing with either cell type mask was then expressed as a percentage of the total Ab pH signal across the eld.

In vivo two-photon microscopy
Adult mice bred on a C57BL/6 background aged $4-7 months (P123 to P203) were anesthetized using urethane (1.55 g kg À1 given intraperitoneally). Adequate anesthesia was ensured by conrming the absence of a withdrawal response to a paw pinch. Body temperature was maintained at 36.8 AE 0.3 C and eyes were protected from drying by applying polyacrylic acid eye drops (Dr Winzer Pharma). The trachea was cannulated and mice were mechanically ventilated with medical air supplemented with oxygen using a MiniVent (Model 845). A headplate was attached to the skull using superglue and mice were head xed to a custom-built stage. A craniotomy of approximately 2 mm diameter was performed over the right primary somatosensory cortex, immediately caudal to the coronal suture and approximately 2 to 4 mm laterally from the midline. The dura was removed and 2% agarose in HEPES-buffered aCSF was used to create a well lled with HEPES-buffered aCSF during imaging.
Two-photon excitation was performed using a Newport-Spectraphysics Ti:sapphire MaiTai laser pulsing at 80 MHz, and a Zeiss LSM710 microscope with a 20Â water immersion objective (NA 1.0). Fluorescence was evoked using a wavelength of 920 nm. The mean laser power under the objective did not exceed 25 mW. Image stacks were taken in 2 mm depth increments (50-200 mm from the cortical surface) every 1.5 min for approximately 30 minutes. Image stacks were maximum intensity projected in the z-dimension and registered using the StackReg plugin in FIJI. Fluorescent changes in the Ab pH uorescence were measured from regions of interest (ROIs) drawn around the soma of cells. Five mM Ab pH in HEPES-based aCSF was applied to the cortical surface for 10 minutes and then replaced with HEPES-based aCSF. Animals were transcardially perfused with ice-cold PBS followed by 4% PFA in PBS at 1.5 or 3 hours aer pH Ab pH application. Brains were post-xed in PFA for 12 hours at 4 C and 100 mm sagittal sections were prepared using a vibratome. Slices were permeabilized and blocked for 12 hours and incubated with rabbit anti-IBA1 (1 : 500, Synaptic Systems, 234006), rat anti-mouse CD68 (1 : 500, Bio-Rad, MCA1957) or chicken anti-GFAP (1 : 500, Abcam, ab4674) primary antibodies for 24 hours at 4 C. Following washes in PBS, donkey anti-rabbit IgG 647 (1 : 500, ThermoFisher, A31573), goat anti-rat IgG 647 (1 : 500, ThermoFisher, A21247) or donkey anti-chicken IgG 649 (1 : 300, Jackson, 703-495-155) was applied for 12 hours at 4 C. Image stacks (23 mm deep) were acquired at 1 mm interval in the cerebral cortex and analysis was done as for in situ experiments.