TREM2 regulates purinergic receptor-mediated calcium signaling and motility in human iPSC-derived microglia

  1. Amit Jairaman
  2. Amanda McQuade
  3. Alberto Granzotto
  4. You Jung Kang
  5. Jean Paul Chadarevian
  6. Sunil Gandhi
  7. Ian Parker
  8. Ian Smith
  9. Hansang Cho
  10. Stefano L Sensi
  11. Shivashankar Othy
  12. Mathew Blurton-Jones  Is a corresponding author
  13. Michael D Cahalan  Is a corresponding author
  1. Department of Physiology and Biophysics, University of California, Irvine, United States
  2. Department of Neurobiology and Behavior, University of California, Irvine, United States
  3. Sue and Bill Gross Stem Cell Research Center, University of California, Irvine, United States
  4. UCI Institute for Memory Impairments and Neurological Disorders, University of California, United States
  5. Institute for Neurodegenerative Diseases, University of California, San Francisco, United States
  6. Center for Advanced Sciences and Technology (CAST), University "G. d'Annunzio" of Chieti-Pescara, Italy
  7. Department of Neuroscience, Imaging and Clinical Sciences, University G d'Annunzio of Chieti-Pescara, Italy
  8. Department of Mechanical Engineering and Engineering Science, University of North Carolina, United States
  9. Institute of Quantum Biophysics, Department of Biophysics, Dept of Intelligent Precision Healthcare Convergence, Sungkyunkwan University, Republic of Korea
  10. Institute for Immunology, University of California, Irvine, United States

Abstract

The membrane protein TREM2 (Triggering Receptor Expressed on Myeloid cells 2) regulates key microglial functions including phagocytosis and chemotaxis. Loss-of-function variants of TREM2 are associated with increased risk of Alzheimer’s disease (AD). Because abnormalities in Ca2+ signaling have been observed in several AD models, we investigated TREM2 regulation of Ca2+ signaling in human induced pluripotent stem cell-derived microglia (iPSC-microglia) with genetic deletion of TREM2. We found that iPSC-microglia lacking TREM2 (TREM2 KO) show exaggerated Ca2+ signals in response to purinergic agonists, such as ADP, that shape microglial injury responses. This ADP hypersensitivity, driven by increased expression of P2Y12 and P2Y13 receptors, results in greater release of Ca2+ from the endoplasmic reticulum stores, which triggers sustained Ca2+ influx through Orai channels and alters cell motility in TREM2 KO microglia. Using iPSC-microglia expressing the genetically encoded Ca2+ probe, Salsa6f, we found that cytosolic Ca2+ tunes motility to a greater extent in TREM2 KO microglia. Despite showing greater overall displacement, TREM2 KO microglia exhibit reduced directional chemotaxis along ADP gradients. Accordingly, the chemotactic defect in TREM2 KO microglia was rescued by reducing cytosolic Ca2+ using a P2Y12 receptor antagonist. Our results show that loss of TREM2 confers a defect in microglial Ca2+ response to purinergic signals, suggesting a window of Ca2+ signaling for optimal microglial motility.

Editor's evaluation

Overall, this is a significant advance in the field of microglial regulation by calcium signaling.

https://doi.org/10.7554/eLife.73021.sa0

Introduction

As the primary immune cells of the central nervous system, microglia survey their local environment to maintain homeostasis and respond to local brain injury or abnormal neuronal activity. Microglia are strongly implicated in several neurodevelopmental and neurodegenerative diseases (Andersen et al., 2021; Crotti et al., 2014; Fahira et al., 2019; Jansen et al., 2019; McQuade and Blurton-Jones, 2019; Pimenova et al., 2021; Tan et al., 2013), warranting further study of human microglial dynamics. Purinergic metabolites (ATP, ADP, UTP, UDP) in the brain constitute key signals driving microglial activation and chemotaxis, and are detected by microglial cells over concentrations ranging from hundreds of nM to μM (Davalos et al., 2005; De Simone et al., 2010; Honda et al., 2001; Koizumi et al., 2007; Haynes et al., 2006; Yegutkin, 2008). ATP released from both homeostatic and damaged cells is hydrolyzed locally by nucleosidases such as the ectonucleotidase NTPDase1 (CD39) or pyrophosphatase NPP1 to produce ADP (Dissing-Olesen et al., 2014; Madry and Attwell, 2015; Zhang et al., 2014). ADP is then detected by P2Y purinergic receptors on microglia, causing IP3-dependent Ca2+ release from the endoplasmic reticulum (ER) lumen. Ca2+ depletion from the ER in turn activates ER STIM1 proteins to translocate proximally to puncta where closely apposed plasma membrane (PM) Orai1 channels are activated. This mechanism underlies store-operated Ca2+ entry (SOCE) in many cell types (Prakriya and Lewis, 2015), including microglia (McLarnon, 2020; Mizuma et al., 2019; Gilbert et al., 2016).

Purinergic signaling is central to microglial communication with other brain cell types and has been negatively correlated with the onset of disease-associated microglia (DAM) transcriptional states (Hasselmann et al., 2019; Keren-Shaul et al., 2017; Krasemann et al., 2017; Olah et al., 2020; Sala Frigerio et al., 2019). P2Y12 and P2Y13 receptors are highly expressed by microglia and are activated predominantly by ADP (Zhang et al., 2014; Weisman et al., 2012). P2Y12 receptors are essential for microglial chemotaxis and have been implicated in the microglial response to cortical injury (Haynes et al., 2006; Cserép et al., 2020), NLRP3 inflammasome activation (Suzuki et al., 2020; Wu et al., 2019), neuronal hyperactivity and protection (Cserép et al., 2020; Eyo et al., 2014), and blood-brain barrier maintenance (Lou et al., 2016; Bisht et al., 2021). While purinergic receptors have been broadly identified as markers of microglial homeostasis (Krasemann et al., 2017; Weisman et al., 2012), mechanisms by which receptor expression may drive or maintain homeostatic microglial states remain incompletely understood.

Neuroinflammatory pathologies are often associated with altered Ca2+ signaling (Leissring et al., 2000). Microglia, in particular, show altered Ca2+ responses in mouse models of Alzheimer’s disease (AD) by mechanisms that are not fully understood (Brawek et al., 2014; Demuro et al., 2010; Mustaly-Kalimi et al., 2018). Ca2+ responses to purinergic metabolites have been extensively studied in cultured murine microglia, acute brain slices, and, more recently, in anesthetized mice (Davalos et al., 2005; Honda et al., 2001; Brawek et al., 2014; Eichhoff et al., 2011; Irino et al., 2008; Milior et al., 2020). However, our understanding of how specific patterns of Ca2+ signals in microglia correlate with and tune downstream microglial responses such as cell motility or process extension remains incomplete. There is also a paucity of knowledge on how regulators of purinergic Ca2+ signals in microglia might play a role in the dysregulation of Ca2+ signaling associated with aging and neuroinflammation.

TREM2 encodes a cell surface receptor that binds a variety of ligands, including various lipids, apolipoprotein E (ApoE), and amyloid-β peptides. Upon ligand binding, TREM2 signals through its adaptor protein DAP12 to activate a host of downstream pathways (Krasemann et al., 2017; Cheng-Hathaway et al., 2018; McQuade et al., 2020; Ulrich et al., 2014). Loss of TREM2 function is thought to promote a more homeostatic-like state (Krasemann et al., 2017; Andrews et al., 2020; Karch et al., 2012). Indeed, microglia lacking TREM2 expression exhibit greatly diminished activation against disease pathology, correlating with increased risk of Alzheimer’s disease (AD) (Krasemann et al., 2017; McQuade et al., 2020; Cheng et al., 2018). Purinergic receptor hyperexpression has been reported at the transcriptome level across multiple TREM2 loss of function models, including human patient mutations (Hasselmann et al., 2019; Keren-Shaul et al., 2017; Krasemann et al., 2017; Sala Frigerio et al., 2019; McQuade et al., 2020; Gratuze et al., 2020). For example, P2Y12 receptor protein expression was found to be elevated in the cortical microglia of Trem2-/- mice and in a preclinical mouse model of AD (Götzl et al., 2019; Griciuc et al., 2019), although the mechanistic link between purinergic receptor expression and TREM2 function remains poorly understood.

We previously developed methods to generate human induced pluripotent stem cell-derived microglia (iPSC-microglia) (Abud et al., 2017; McQuade et al., 2018; McQuade and Blurton-Jones, 2021), which can be used to model human microglial behavior. While iPSC-microglia are proving increasingly useful to investigate neurodegenerative disorders (McQuade et al., 2020; Andreone et al., 2020; Cosker et al., 2021; Konttinen et al., 2019; Piers et al., 2020; You et al., 2022), Ca2+ signaling has not yet been extensively profiled in these models. In this study, we compared purinergic Ca2+ signaling and motility characteristics in wild type (WT) and TREM2 knockout (KO) human iPSC-microglia, and examined the mechanisms that underlie enhanced purinergic Ca2+ signaling in microglia lacking TREM2. We find that motility is differentially tuned by Ca2+ in TREM2 KO cells with consequences for chemotaxis.

Results

Purinergic receptor Ca2+ signaling is enhanced in TREM2 KO human iPSC-microglia

To determine if TREM2 plays a role in microglial Ca2+ signaling, we compared cytosolic Ca2+ responses to the purinergic agonist ADP in isogenic, CRISPR-modified wild type (WT) and TREM2 KO human iPSC-microglia. ADP stimulation induced a biphasic Ca2+ response – a rapid initial peak followed by a secondary phase of sustained Ca2+ elevation lasting several minutes, in line with previous observations in mouse microglia (Michaelis et al., 2015; Visentin et al., 2006). Both phases of the Ca2+ response were significantly elevated in TREM2 KO microglia, raising the possibility that augmentation of the initial Ca2+ response to ADP in TREM2 KO microglia may be coupled to a larger sustained component of Ca2+ entry (Figure 1A and B). These results were corroborated in iPSC-derived microglia cell line expressing the genetically encoded Ca2+ indicator Salsa6f (Dong et al., 2017; Jairaman and Cahalan, 2021; Figure 1C and D). The Salsa6f probe showed the expected increase in the GCaMP6f fluorescence in response to Ca2+ elevation without any change in the tdTomato signal, and it did not perturb microglial activation and function (Figure 1—figure supplement 1A–G). TREM2 KO microglia also showed exaggerated Ca2+ responses to the purinergic agonists ATP and UTP at similar low μM concentrations, although the secondary Ca2+ elevations were not as long-lasting as with ADP (Figure 1E and F, Figure 1—figure supplement 2).

Figure 1 with 2 supplements see all
Microglia lacking TREM2 show exaggerated Ca2+ responses to purinergic stimulation.

(A) Representative red-green channel overlay images of wild type (WT) (top) and TREM2 knockout (KO) (bottom) induced pluripotent stem cell (iPSC)-microglia loaded with Fluo-4 (green) and Fura-red (red) showing resting cytosolic Ca2+ before ADP, and Ca2+ levels 15 s and 5 min after ADP addition. Scale bar = 20 μm. (B) Average traces (left panels) showing changes in cytosolic Ca2+ in response to 2.5 μM ADP in 1 mM Ca2+ buffer (n = 39–44 cells). Baseline-subtracted peak Ca2+ response and cytosolic Ca2+ levels 5 min after ADP shown on the right (n = 250–274 cells, five experiments, Mann–Whitney test). (C, D) Cytosolic Ca2+ response to ADP as in (A) and (B) but in iPSC-microglia expressing the GCaMP6f-tdTomato fusion Ca2+ probe Salsa6f (n = 41–53 cells, two independent experiments, Mann–Whitney test). Images in (C) are overlay of GCaMP6f (green) and tdTomato (red) channel images. Scale bar = 20 μm. (E) Ca2+ responses to 2.5 μM ATP in WT and TREM2 KO iPSC-microglia. Average traces (left panel, n = 63–71 cells) and bar graph summary of peak cytosolic Ca2+ and Ca2+ after 5 min (right panel, 165–179 cells, three experiments, Mann–Whitney test). (F) Ca2+ responses to 10 μM UTP. Average traces (45–55 cells) and summary of peak cytosolic Ca2+ and Ca2+ after 5 min (175–269 cells, three experiments, Mann–Whitney test). Data shown as mean ± SEM for traces and bar graphs. p-Values indicated by *** for p<0.001, ****p<0.0001.

Figure 1—source data 1

Microglia lacking TREM2 show exaggerated Ca2+ responses to purinergic stimulation.

In this dataset, the results of microglial stimulation with purinergic agonists and validation of Salsa6f isogenic microglia are included.

https://cdn.elifesciences.org/articles/73021/elife-73021-fig1-data1-v3.xlsx

Increased P2Y12 and P2Y13 receptor expression drives increased peak Ca2+ in TREM2 KO microglia

Given the critical importance of ADP signaling in several aspects of microglial function, we investigated the mechanisms driving higher ADP-evoked Ca2+ signals in TREM2 KO microglia by focusing on specific steps in the purinergic Ca2+ signaling pathway (Figure 2A). The initial Ca2+ response to P2Y receptor engagement results from G protein-coupled phospholipase C activation and IP3-mediated ER Ca2+ store release. To test this, we treated cells with ADP in Ca2+-free solution buffered with the Ca2+ chelator EGTA to isolate Ca2+ signals from store release and eliminate Ca2+ influx across the PM. Both WT and TREM2 KO cells exhibited a single Ca2+ peak, with TREM2 KO cells showing significantly higher peak Ca2+ response to ADP (Figure 2B, Figure 2—figure supplement 1A and B). Moreover, the amplitude of the Ca2+ peak was not significantly different in the presence or absence of external Ca2+, strongly suggesting that it is driven primarily by release of Ca2+ from intracellular stores even when external Ca2+ is present (Figure 2—figure supplement 1C). Dose–response curves for the peak Ca2+ response showed a steep leftward shift in TREM2 KO cells (Figure 2C). The EC50 value for WT microglia was 650 nM, whereas TREM2 KO microglia reached their EC50 by 15 nM. This stark difference was driven at least in part by a diminished percentage of WT cells responding to ADP at low μM doses (Figure 2D). However, limiting the analysis to cells that showed a Ca2+ rise revealed that ‘responding’ TREM2 KO cells still exhibited higher Ca2+ responses to ADP than ‘responding’ WT cells (Figure 2E). TREM2 KO microglia are thus significantly more sensitive to ADP than WT cells, which may be critical in sensing ADP and detecting ADP gradients.

Figure 2 with 1 supplement see all
Higher sensitivity of TREM2 knockout (KO) microglia to ADP is driven by increased purinergic receptor expression.

(A) Schematic highlighting key downstream Ca2+ signaling events triggered by ADP. Cytosolic Ca2+ response to ADP is determined by functional expression and activity of P2Y12 and P2Y13 receptors, IP3 receptors, endoplasmic reticulum (ER) store Ca2+ content, and store-operated Ca2+ entry (SOCE) regulated by STIM and Orai proteins. (B) Representative images (left panel) showing overlay of Fluo-4 (green) and Fura-red (red) channels in wild type (WT) (top) and TREM2 KO (bottom) induced pluripotent stem cell (iPSC)-microglia before and peak Ca2+ response after ADP addition in Ca2+-free buffer. Scale bar = 20 μm. Average trace showing Ca2+ response to ADP in Ca2+-free buffer (middle panel, 64–83 cells). Quantification of peak signal (right panel, n = 264–289 cells, four experiments, Mann–Whitney test). (C–E) Dose–response curves showing baseline-subtracted peak Ca2+ responses to ADP in Ca2+-free buffer (C), percent of ‘responding’ cells (D), and peak Ca2+ responses only in ‘responding’ cells (E). N = 84–474 WT cells and 70–468 TREM2 KO cells, 2–5 experiments. (F) RNA normalized read counts of P2Y12 and P2Y13 receptor expression from bulk RNA-sequencing of WT and TREM2 KO iPSC-microglia (n = 4, adjusted p-values from DESeq2). (G) Representative histogram (left panel) showing plasma membrane (PM) expression of P2Y12 receptor in WT and TREM2 KO microglia. Cells were stained with BV421-labeled anti-human P2Y12 receptor antibody. Isotype control is shown as dashed line. Right panel shows summary of median fluorescence intensity (MFI) of P2Y12 receptor-labeled cells (n = 10 samples each, Student’s t-test). (H) Ca2+ traces (left panel) showing response to 1 μM ADP in Ca2+-free buffer after 30 min pretreatment with a combination of P2Y12 receptor antagonist PSB 0739 (10 μM) and P2Y13 receptor antagonist MRS 2211 (10 μM). Summary of the peak Ca2+ response (right panel, n = 40–79 cells, two experiments, Mann–Whitney test). Data are mean ± SEM. p-Values indicated by ****p<0.0001.

Figure 2—source data 1

Higher sensitivity of TREM2 knockout (KO) microglia to ADP is driven by increased purinergic receptor expression.

In this dataset, the results of ADP stimulation in 0 Ca2+, dose curve of ADP in wild type (WT) and TREM2 KO, P2Y receptor expression, expression of key calcium signaling proteins, and inhibition of P2Y receptors are included.

https://cdn.elifesciences.org/articles/73021/elife-73021-fig2-data1-v3.xlsx

RNA-sequencing revealed significantly increased transcripts for P2Y12 and P2Y13 receptors, the main P2Y receptor subtypes in microglia that bind ADP, in TREM2 KO microglia (Abud et al., 2017; McQuade et al., 2018; Figure 2F). In comparison, relative mRNA levels of common mediators of Ca2+ signaling – including predominant isoforms of IP3 receptors, SOCE mediators Orai and STIM proteins, and SERCA and PMCA Ca2+ pumps – were either similar or modestly reduced in TREM2 KO in comparison with WT iPSC-microglia (Figure 2—figure supplement 1D and E). We therefore considered the possibility that signal amplification in microglia lacking TREM2 results primarily from increased expression of P2Y12 and P2Y13 receptors. Consistent with this, expression of P2Y12 receptors in the PM was significantly increased in TREM2 KO cells (Figure 2G). Furthermore, Ca2+ responses to ADP in Ca2+-free medium were completely abolished following treatment with a combination of P2Y12 and P2Y13 receptor antagonists (PSB 0739 and MRS 2211, respectively) in both WT and TREM2 KO microglia (Figure 2H). Treatment of cells with P2Y12 and P2Y13 receptor antagonists separately produced partial inhibition of peak ADP-mediated Ca2+ signals, implicating involvement of both receptor subtypes (Figure 2—figure supplement 1F and G). In summary, deletion of TREM2 results in a larger cytosolic Ca2+ peak in response to ADP due to increased expression of P2Y12 and P2Y13 receptors.

SOCE through Orai channels mediates the sustained phase of ADP-evoked Ca2+ elevation

To probe the basis for the increased sustained component of ADP-evoked Ca2+ signal in TREM2 KO microglia, we examined SOCE using pharmacological and genetic approaches. Synta66, a reasonably specific inhibitor of Orai channels, significantly reduced the rate of SOCE following Ca2+ readdition after ER store depletion by the sarco-endoplasmic reticulum Ca2+ ATPase (SERCA pump) inhibitor, thapsigargin (TG), in both WT and TREM2 KO microglia (Figure 3A, Figure 3—figure supplement 1A). Using a similar Ca2+ readdition protocol with ADP, we found significant inhibition of ADP-induced SOCE by Synta66 in both WT and TREM2 KO cells (Figure 3B, Figure 3—figure supplement 1B). The ADP-evoked sustained Ca2+ phase in TREM2 KO iPSC-microglia was also blocked by less specific Orai channel inhibitors, Gd3+ and 2-APB (Figure 3—figure supplement 1C and D). To further confirm the specific role of Orai1 channels in mediating SOCE, we generated an Orai1 CRISPR-knockout iPSC line. Deletion of Orai1 abrogated SOCE and significantly reduced the sustained Ca2+ response to ADP (Figure 3—figure supplement 1E and F). These results confirm that Orai1 plays an important role in mediating SOCE and ADP-evoked Ca2+ signals in iPSC-microglia.

Figure 3 with 2 supplements see all
Regulation of ADP-evoked store-operated Ca2+ entry (SOCE) in wild type (WT) and TREM2 knockout (KO) microglia.

(A) SOCE in WT microglia triggered with thapsigargin (TG, 2 μM) in Ca2+-free buffer followed by readdition of 1 mM Ca2+ in the absence (control, gray trace) or presence (red trace) of the Orai channel inhibitor Synta66 (n = 34–48 cells). Cells were pretreated with Synta66 (10 μM) for 30 min before imaging. Bar graph summary of the rate of Ca2+ influx (n = 80–137 cells, two experiments, Mann–Whitney test). (B) SOCE evoked by ADP (2.5 μM) in WT microglia (gray trace) using a similar Ca2+ addback protocol as in (A). Red trace shows the effect of Synta66 on ADP-evoked SOCE. Right panel shows bar graph summary of the rate of ADP-triggered Ca2+ influx after readdition of 1 mM Ca2+ (n = 148–155 cells, two experiments, Mann–Whitney test). (C) Comparison of SOCE evoked with TG (2 μM) in WT and TREM2 KO cells (n = 90–129 cells). Bar graph summaries of endoplasmic reticulum (ER) store release quantified as area under the curve, rate of SOCE, and peak SOCE (n = 187–266 cells, two experiments, Mann–Whitney test). (D) Traces showing ADP-evoked SOCE in WT and TREM2 KO microglia after depleting stores with 100 nM ADP in Ca2+-free buffer and readdition of 1 mM Ca2+ (left panel, n = 97–114 cells). Comparison of ADP-evoked cytosolic Ca2+ peak, peak SOCE and SOCE rate (right panel, n = 234–313 cells, three experiments, Mann–Whitney test). (E) Ionomycin pulse experiment to measure residual ER Ca2+ pool in cells after initial treatment with ADP. WT and TREM2 KO cells were pulsed sequentially with ADP first (200 nM) and subsequently treated with ionomycin (1 μM) to empty and measure the residual pool of ER Ca2+. Imaging was done entirely in Ca2+-free buffer to prevent Ca2+ influx across the plasma membrane (PM). Average trace (left panel), peak ADP Ca2+ response (middle panel), and peak ionomycin-induced Ca2+ response (right panel) (n = 38–60 cells, 3–4 experiments, Mann–Whitney test). (F) Average trace (left, 71–117 cells) and summary of ER store release after 2 μM ionomycin treatment in Ca2+-free buffer (right, 146–234 cells, two experiments; nd, nonsignificant p>0.05, Mann–Whitney test). (G) Same as (H) but in response to UV IP3 uncaging (167–200 cells, ns, nonsignificant p>0.05, nonparametric t-test). Data shown as mean ± SEM for traces and bar graphs. Data are mean ± SEM. p-Values indicated by ns, nonsignificant, *p<0.05, and ****p<0.0001.

Figure 3—source data 1

Regulation of ADP-evoked store-operated Ca2+ entry (SOCE) in wild type (WT) and TREM2 knockout (KO) microglia.

In this dataset, the results of blocking SOCE on ADP stimulation and investigation of store content as well as the correlation between original calcium store release and SOCE are included.

https://cdn.elifesciences.org/articles/73021/elife-73021-fig3-data1-v3.xlsx

To determine if SOCE is increased in TREM2 KO microglia and contributing to the higher sustained Ca2+ response to ADP, we compared the rate of store-operated Ca2+ influx after store depletion with TG and found that both the rate and amplitude of SOCE were modestly reduced in TREM2 KO cells (Figure 3C). In keeping with this, RNA-sequencing revealed a modest reduction in STIM1 mRNA expression in TREM2 KO cells, although Orai1 mRNA was similar in WT and TREM2 KO microglia (Figure 2—figure supplement 1C and D). We further conclude that the elevated secondary phase of ADP-driven Ca2+ signals in TREM2 KO microglia is not primarily due to the differences in the expression of STIM and Orai.

ADP depletes ER Ca2+ stores to a greater extent in TREM2 KO microglia, leading to greater SOCE activation

We hypothesized that the exaggerated secondary Ca2+ phase in response to ADP in TREM2 KO microglia may be driven by increased ER Ca2+ store release, leading to greater SOCE activation. Consistent with this possibility, peak cytosolic Ca2+ in response to partial store depletion with ADP and after Ca2+ readdition was elevated in TREM2 KO microglia (Figure 3D). To examine if the higher magnitude of SOCE in TREM2 KO cells is due to depletion of ER Ca2+ stores by ADP, we sequentially treated cells with ADP followed by ionomycin to completely release stores in Ca2+-free buffer. While TREM2 KO cells showed greater peak Ca2+ with ADP as expected, the ionomycin Ca2+ peak – which reflects the residual ER Ca2+ pool – was significantly reduced, indicating that ADP depletes ER Ca2+ stores to a greater extent in TREM2 KO cells (Figure 3E). Similar results were obtained when residual ER store content was depleted using TG instead of ionomycin (Figure 3—figure supplement 2A and B). We plotted cytosolic Ca2+ levels 5 min after addition of varying doses of ADP to indicate the degree of SOCE as a function of the initial peak Ca2+, a readout of ER store release (Figure 3—figure supplement 2C and D). Both WT and TREM2 KO microglia showed similar linear relationships between SOCE and store release, further suggesting that SOCE is activated by similar mechanisms in the two cell lines, but is recruited to a greater extent in TREM2 KO cells due to increased ER store release. We also note that increased sustained Ca2+ in TREM2 KO cells is unlikely to be due to differences in Ca2+ pump activity based on similar Ca2+ clearance rates (Figure 3—figure supplement 2E and F), consistent with comparable transcriptomic expression of major SERCA and PM Ca2+ ATPase (PMCA) isoforms in WT and TREM2 KO cells (Figure 2—figure supplement 1C and D).

Finally, quantification of cumulative cytosolic Ca2+ increases after maximally depleting ER stores with ionomycin alone suggested that overall ER store content is not altered in microglia lacking TREM2 (Figure 3F). Comparison of Ca2+ responses to IP3 uncaging also ruled out major differences in the pool of functional IP3 receptors between WT and TREM2 KO cells (Figure 3G), as further substantiated by similar transcriptomic expression of IP3 receptor type 2 (the major IP3R subtype expressed in iPSC-microglia) in WT and TREM2 KO cells (Figure 2—figure supplement 1C and D; McQuade et al., 2020; Abud et al., 2017). In summary, deletion of TREM2 in iPSC-derived microglia leads to upregulation of P2Y12 and P2Y13 receptors and renders the cells hypersensitive to ADP signaling, consequently leading to greater IP3-mediated ER store depletion and increased coupling to SOCE in response to purinergic metabolites.

ADP potentiates cell motility and process extension in human WT iPSC-microglia

ADP is a potent chemoattractant for microglia (Honda et al., 2001). Analogous to a previous study in fibroblasts (Borges et al., 2021), we found that ADP treatment alters cell motility and leads to increased rates of scratch wound closure in WT iPSC-microglia (Figure 4A). To investigate the cellular mechanism of accelerated wound closure, we used time-lapse imaging to track open-field microglial cell motility (Figure 4B). Mean cell track speed and track displacement (defined as the overall change in position from the origin at a given time) were both increased after application of ADP. On the other hand, average track straightness, an indicator of how frequently cells change direction, was unaltered by ADP (Figure 4C). These data suggest that ADP-driven changes in motility in WT iPSC-microglia primarily arise from increases in microglial speed, and not altered turning behavior. ADP-dependent increases in speed were reversed in the presence of P2Y12 (PSB 0739) and P2Y13 (MRS 2211) receptor antagonists, confirming the role of these two purinergic receptors in ADP enhancement of microglial motility (Figure 4D). To determine if Ca2+ influx regulates ADP-mediated increases in motility, we measured cell migration with ADP in Ca2+-free medium and found that removing extracellular Ca2+ significantly decreased cell speed, displacement, and track straightness, suggesting that sustained Ca2+ signals are required for maximal increase in motility in response to ADP (Figure 4E).

Figure 4 with 1 supplement see all
Nondirectional ADP exposure increases wild type (WT) microglial speed and process extension.

(A) Average trace showing closure of scratch wound produced with IncuCyte S3 WoundMaker. Induced pluripotent stem cell (iPSC)-microglia imaged every 30 min after scratch wound with or without ADP stimulation (n = 4 wells; two images per well). (B) Representative image of WT iPSC-microglia motility 30 min after ADP exposure with cell tracks overlain (left). Pseudocolored images (center) across time: 0 min (red), 4 min (orange), 8 min (yellow), 12 min (green), 16 min (cyan), 20 min (blue), 24 min (purple), and 28 min (magenta). Scale bar = 100 μm. White boxes zoomed in at right to demonstrate motile (top) and nonmotile (bottom) cells. (C) Representative color images (top left) and displacement vectors (bottom left) of WT iPSC-microglia at baseline (no ADP, gray) and 30 min after 2.5 μM ADP treatment (red). Summary of mean speed (µm/min), Displacement over 10 min (μm/10 min) and track straightness (track length/track displacement) (414–602 cells, two experiments). (D) Representative images, displacement vectors, and quantification of WT iPSC-microglia motility for 20 min following ADP addition. Cells were pretreated with vehicle (gray), MRS 2211 (10 μM, gold), or PBS 0739 (10 μM, blue) (180–187 cells, two experiments). (E) Representative images, displacement vectors, and quantification of WT iPSC-microglia motility after ADP in 1 mM Ca2+ (light gray) or Ca2+-free buffer (dark gray) (401–602 cells, three experiments). (F) Representative images (left) and process extension (right) of iPSC-microglia (cytoplasmic GFP, gray) before or 30 min after ADP addition. Cells were pretreated with vehicle (gray), MRS 2211 (10 μM, gold), or PBS 0739 (10 μM, blue) (52–163 cells, 3–4 experiments). (C–F) One-way ANOVA with Tukey post hoc test. Data shown as mean ± SEM (A, F) and as violin plots with mean, 25th and 75th percentile (C–E). p-Values indicated by ns, nonsignificant, *p<0.05, **p<0.01, ***p<0.001, and ****p<0.0001.

Figure 4—source data 1

Nondirectional ADP exposure increases wild type (WT) microglial speed and process extension.

In this dataset, the results of motility experiments and process extension in WT cells are included.

https://cdn.elifesciences.org/articles/73021/elife-73021-fig4-data1-v3.xlsx

In addition, some microglia responded to ADP by extending processes and altering their morphology rather than increasing motility (Figure 4—figure supplement 1). Microglia have been observed to extend processes in response to injury and purinergic stimulation in brain slices (Davalos et al., 2005; Haynes et al., 2006). Therefore, we compared process complexity before and 30 min after ADP exposure in WT microglia and observed significant increases in both the number of branches per process and total length of these processes (Figure 4F). Similar to effects on cell motility, ADP-mediated process extension was inhibited by P2Y12 and P2Y13 receptor antagonists (PSB 0739 and MRS 2211, respectively). Furthermore, even before process extension was activated with ADP, cells treated with P2Y antagonists showed significantly fewer and shorter processes, suggesting that baseline purinergic signaling may regulate resting microglial process dynamics. Altogether, these results demonstrate that activation of purinergic signaling through P2Y12 and P2Y13 receptors is required for ADP-driven microglial process extension and motility.

ADP-evoked changes in cell motility and process extension are enhanced in TREM2 KO microglia

To characterize differences in motility characteristics between WT and TREM2 KO microglia responding to ADP, we plotted mean squared displacement (MSD) vs. time and compared cell track overlays (flower plots), which showed that ADP enhances motility in TREM2 KO cells to a greater extent than in WT microglia (Figure 5A and B). Baseline motility characteristics in unstimulated cells, however, were similar in WT and TREM2 KO cells (Figure 5—figure supplement 1A and B). To further understand the basis of differences in ADP-induced motility between WT and TREM2 KO cells, we analyzed mean track speed, track displacement, and track straightness. Although mean track speeds were similar, TREM2 KO microglia showed greater displacement than WT cells (Figure 5C and D), raising the possibility that TREM2 KO cells may turn with lower frequency. Consistent with this, analysis of track straightness revealed that TREM2 KO microglia move farther from their origin for the same total distance traveled (Figure 5E). Vector autocorrelation, an analysis of directional persistence (Gorelik and Gautreau, 2014), further confirmed that WT cells turn more frequently than TREM2 KO microglia in response to ADP (Figure 5—figure supplement 1C and D). To assess if these differences in TREM2 KO cells require sustained Ca2+ influx, we analyzed microglial motility in response to ADP stimulation in the absence of extracellular Ca2+ (Figure 5F–J). MSD and cell track overlay plots showed that motility is constrained when Ca2+ is removed from the external bath in both WT and TREM2 KO cells (Figure 5A and B vs. F and G). In the absence of extracellular Ca2+, TREM2 KO microglia showed similar mean speed, displacement, and track straightness as WT cells (Figure 5C–E vs. H–J). We conclude that increases in microglial motility (mean speed, displacement, and straightness) require sustained Ca2+ influx and that deletion of TREM2 reduces microglial turning in response to ADP.

Figure 5 with 2 supplements see all
ADP-driven process extension and cell displacement are increased in TREM2 knockout (KO) induced pluripotent stem cell (iPSC)-microglia.

(A–E) Motility of wild type (WT) (gray) and TREM2 KO (green) iPSC-microglia over 20 min following ADP addition in 1 mM Ca2+-containing buffer. (A) Plots of track displacement in μm centered from point of origin at (0,0). (B) Mean squared displacement (MSD) vs. time. Mean cell track speeds (C), total track displacement in 10 min interval (D), and track straightness (E) for 130–327 cells, seven experiments, Student’s t-test. (F–J) Same as (A–F) but in Ca2+-free medium (125–279 cells, two experiments, Student’s t-test). (K) Representative images of GFP-expressing WT (top) and TREM2 KO (bottom) iPSC-microglia, before and 30 min after 2.5 μM ADP addition. (L) Quantification of total number of branches per cell before and after ADP treatment (left) and paired dot plots showing fold change in branch number from pre-ADP levels (right). Each data point represents an imaging field in the paired plots. (M) Total process length before and after ADP treatment displayed as raw values per cell (left) and as fold change from baseline conditions per imaging field (right). For (L) and (M). n = 151–158 cells, WT; 133–167 cells, KO; 9–10 imaging fields, 3–4 experiments. One-way ANOVA with multiple comparisons for single-cell data, two-tailed paired t-test for the paired plots. Data shown as mean ± SEM (B, G, L, M) and as violin plots with mean, 25th and 75th percentile (C– E, H–J). p-Values indicated by ns, nonsignificant, *p<0.05, **p<0.01, and ****p<0.0001.

Figure 5—source data 1

ADP-driven process extension and cell displacement are increased in TREM2 knockout (KO) induced pluripotent stem cell (iPSC)-microglia.

In this dataset, the results of motility experiments and process extension in wild type (WT) and TREM2 KO cells, as well as baseline motility and directional persistence, are included.

https://cdn.elifesciences.org/articles/73021/elife-73021-fig5-data1-v3.xlsx

We next analyzed the effects of TREM2 deletion on process extension in microglia. Treatment with ADP induced a dramatic increase in the number of branches and length of processes extended in both WT and TREM2 KO microglia (Figure 5K and L). Comparison of the absolute number of branches and process length after ADP treatment, as well as the relative fold increase in these parameters from baseline, indicated that process extension is not affected in TREM2 KO microglia (Figure 5K–M, Figure 5—figure supplement 2A and B). We note that the greater fold change in process extension in TREM2 KO cells can be attributed to the reduced morphological complexity of these cells prior to stimulation. Finally, ADP stimulation in Ca2+-free medium did not induce process extension in WT cells, and only a modest increase in TREM2 KO cells (Figure 5—figure supplement 2A and B vs. C and D). Together, these results indicate that sustained Ca2+ entry across the PM is required for optimal microglial process extension in both WT and TREM2 KO microglia.

Cytosolic Ca2+ levels tune motility in TREM2 KO iPSC-microglia

To further characterize the effects of sustained Ca2+ signals on microglial motility, we used Salsa6f-expressing iPSC WT and TREM2 KO reporter lines to monitor cytosolic Ca2+ and motility simultaneously in individual cells (Figure 6—figure supplement 1). To isolate the effects of sustained Ca2+ elevations on microglia motility and eliminate any contribution from Ca2+ independent signaling pathways, we used a protocol that relies on triggering SOCE and varying external Ca2+ to maintain cytosolic Ca2+ at ‘low’ or ‘high’ levels in the Salsa6f reporter line (Figure 6A–C), similar to our previous study in T lymphocytes (Negulescu et al., 1996). In WT cells, lowering extracellular Ca2+ from 2 to 0.2 mM predictably decreased the G/R ratio but did not influence mean track speed, 10 min track displacement, or track straightness (Figure 6C and D, top). However, in TREM2 KO microglia, reducing Ca2+ to a lower level significantly increased speed, displacement, and track straightness (Figure 6C and D, bottom). These data suggest that motility characteristics of TREM2 KO microglia are more sensitive to changes in cytoplasmic Ca2+ levels than in WT cells. Similar results were obtained upon addition of ADP in this paradigm, suggesting that long-lasting Ca2+ elevations may override effects of Ca2+-independent ADP signaling on cell motility (Figure 6—figure supplement 2A).

Figure 6 with 2 supplements see all
Cytosolic Ca2+ levels tune microglial motility in TREM2 knockout (KO) cells.

(A) Schematic of traditional store-operated Ca2+ entry (SOCE) pathway with store refilling (left) and protocol for sustaining cytoplasmic Ca2+ to ‘low’ and ‘high’ levels with 0.2 and 2 mM extracellular Ca2+ and using thapsigargin (TG) to inhibit store refilling (right). (B) Average SOCE traces in wild type (WT) Salsa6f induced pluripotent stem cell (iPSC)-microglia showing changes in cytoplasmic Ca2+ after addition of either 0.2 or 2 mM extracellular Ca2+ (n = 78–110 cells). (C) Average change in cytoplasmic Ca2+ levels in WT and TREM2 KO microglia over 25 min after SOCE activation. (D) Comparison of Ca2+ levels and microglia motility in WT (top) and TREM2 KO (bottom) microglia. Cytosolic Ca2+ levels indicated by instantaneous single-cell G/R ratio (n = 74–158 cells). Mean of instantaneous speeds, track displacement, and track straightness calculated as before in Figures 3 and 4. Yellow (0.2 mM Ca, TG), green (2 mM Ca, TG). Student’s t-test ****p<0.0001; **p=0.0062; *p=0.432; ns > 0.9999. (E) Correlation of instantaneous Ca2+ and instantaneous speed in WT and KO cells. Red line denotes 10 μm/s (cells above this threshold considered ‘fast moving’). For WT: p<0.0001; r = –0.1316; number pairs = 5850. For KO: p<0.0001; r = –0.1433; number pairs = 6,063 (Spearman’s correlation). (F) Mean speed of cells binned by instantaneous G/R Ca2+ ratio (one-way ANOVA ****p<0.0001). Each data point is calculated for a bin increment of 0.5 G/R ratio. (G) Percentage of fast-moving cells quantified as a function of G/R Ca2+ ratio. X-axis G/R ratios binned in increments of 0.5 as in (F). In (E–G), n = 78–100 cells. Data shown as mean ± SEM (B, F) and as violin plots with mean, 25th and 75th percentile (D). p-Values indicated by ns, nonsignificant, *p<0.05, **p<0.01, and ****p<0.0001.

Figure 6—source data 1

Cytosolic Ca2+ levels tune microglial motility in TREM2 knockout (KO) cells.

In this dataset, the results showing the effect of calcium levels on motility in TREM2 wild type (WT) and KO cells are included.

https://cdn.elifesciences.org/articles/73021/elife-73021-fig6-data1-v3.xlsx

To further analyze the Ca2+ dependence of microglial motility, we plotted Salsa6f G/R Ca2+ ratios for each individual cell at every time point against the instantaneous speeds of that cell (Figure 6E). These data revealed a stronger dependence of instantaneous speed on Ca2+ levels in TREM2 KO microglia (Figure 6F). Furthermore, when stratifying cell speed arbitrarily as ‘fast’ (>10 μm/min) or ‘slow’ (<10 μm/min), we observe a marked reduction in the percentage of ‘fast’ cells when Ca2+ levels are high in TREM2 KO microglia (Figure 6G). Interestingly, frame-to-frame cell displacement correlated with cytosolic Ca2+ to the same degree in both WT and KO cells (Figure 6—figure supplement 2B and C). Together, TREM2 KO human microglia are more sensitive to tuning of motility by cytosolic Ca2+ than WT cells.

Chemotactic defects in TREM2 KO microglia are rescued by dampening purinergic receptor activity

To assess the physiological significance of TREM2 deletion on microglial motility over longer time scales, we performed a scratch wound assay. At baseline, both WT and TREM2 KO microglia migrated into the cell-free area at similar rates, consistent with our previous findings (McQuade et al., 2020; Figure 7—figure supplement 1). Addition of ADP to this system accelerated the scratch wound closure rates to the same extent in WT and TREM2 KO. In vivo, directed migration of microglia is often driven by gradients of ADP from dying or injured cells (Haynes et al., 2006; Eyo et al., 2014). Because no chemical gradient is formed in the scratch wound assay (Liang et al., 2007), we studied microglial chemotaxis toward ADP over a stable gradient using two-chamber microfluidic devices. Consistent with previous findings, WT iPSC-microglia directionally migrated up the concentration gradient of ADP, resulting in higher numbers of cells within the central chamber (McQuade et al., 2020; Park et al., 2018). In the absence of a chemotactic cue, this directional migration was lost (Figure 7A). This assay revealed a deficit of chemotaxis in TREM2 KO microglia (Figure 7A), mirroring reports that TREM2 KO microglia are unable to migrate toward amyloid plaques in AD (Cheng-Hathaway et al., 2018; McQuade et al., 2020; Meilandt et al., 2020). Given that ADP hypersensitivity in TREM2 KO cells is driven by increased expression of P2Y receptors, we examined the effects of dampening P2Y signaling to WT levels. Treatment with the P2Y12 receptor antagonist, PSB 0739, reduced Ca2+ responses in TREM2 KO cells and rescued the migration deficit in the chemotaxis assay (Figure 7B and C). These results link the increased Ca2+ signals and altered motility characteristics evoked by ADP in TREM2 KO cells to microglial chemotaxis toward areas of tissue damage, a vital functional response in microglia.

Figure 7 with 1 supplement see all
Migration deficits in TREM2 knockout (KO) microglia are rescued by inhibition of purinergic signaling.

(A) Migration toward ADP in a two-chamber microfluidic device. Representative images of RFP-expressing microglia that migrated into the central chamber 3 days after 100 ng/mL ADP addition. Dotted circle delineates separation of inner and outer chambers. Scale bar = 500 μm. Quantification of microglial migration (right panel). Migrated cell counts are normalized to wild type (WT) cells treated with ADP (n = 3–4 experiments; one-way ANOVA with multiple comparisons). (B) Baseline-subtracted peak ratiometric Ca2+ signal in response to 2.5 μM ADP in 1 mM extracellular Ca2+, and in the presence or absence of 10 μM PSB 0739 (44 cells, WT; 39–43 cells, KO; representative of three independent experiments; one-way ANOVA with multiple comparisons). (C) Two-chamber migration to 100 ng/mL ADP with or without 10 μM PSB 0739. Values are normalized to WT cells with ADP (n = 3–4 experiments; one-way ANOVA with multiple comparisons). Representative images shown on the left. Scale bar = 500 μm. Data shown as mean ± SEM. p-Values indicated by ns, nonsignificant, *p<0.05, **p<0.01, and ****p<0.0001.

Figure 7—source data 1

Migration deficits in TREM2 knockout (KO) microglia are rescued by inhibition of purinergic signaling.

In this dataset, the results of directional migration and inhibition of purinergic receptor activity are included.

https://cdn.elifesciences.org/articles/73021/elife-73021-fig7-data1-v3.xlsx

Discussion

This study focuses on two aims: understanding the roles of purinergic signaling in regulating human microglial motility behavior and elucidating the impact of TREM2 loss of function on this Ca2+ signaling pathway. We find that sustained Ca2+ influx in response to ADP regulates microglial process extension, motility speed, and turning behavior. A key observation in our study is that microglia lacking TREM2 are highly sensitive to ADP-mediated signaling and show exaggerated cytoplasmic Ca2+ responses. Using novel iPSC-microglia lines that express a ratiometric, genetically encoded Ca2+ probe, Salsa6f, we found that the motility characteristics of human WT and TREM2 KO microglia are differentially tuned by Ca2+ signaling. Informed by these discoveries, we were able to rescue chemotactic deficiencies in TREM2 KO microglia by dampening purinergic receptor signaling.

We provide several lines of evidence to show that hyper-responsiveness to purinergic ADP signaling in TREM2 KO microglia is driven primarily by increased purinergic P2Y12 and P2Y13 receptor expression: (1) Ca2+ response is completely abrogated in the presence of P2Y12 and P2Y13 receptor inhibitors; (2) RNA-sequencing data shows significant increase in expression of P2Y12 and P2Y13 receptor transcripts but minimal fold change in other regulators of Ca2+ signaling (IP3R, STIM, Orai, SERCA, and PMCA); and (3) labeling of surface P2Y12 receptors shows greater PM expression in the TREM2 KOs. Furthermore, functional assays rule out any role for Ca2+ clearance mechanisms or any difference in maximal IP3 and SOCE activity as a cause of increased sustained Ca2+ signal in TREM2 KO cells. Mechanistically, this increase in Ca2+ signals is driven by enhanced IP3-mediated ER store release coupled to SOCE. Indeed, based on the dose–response curves for peak ADP-Ca2+ responses in Ca2+-free buffer, TREM2 KO cells have an EC50 at least 10-fold lower than WT cells. As a functional consequence, TREM2 KO microglia exhibit a defect in turning behavior and show greater displacement over time despite moving with similar speeds as the WT cells. The increased frequency in turning in WT microglia (relative to TREM KO cells) reflects greater canceling of the velocity vectors, which take the direction of motility into account. This restricts cell motility to more confined regions, potentially allowing for more frequent path correction. It is important to note that these motility differences with ADP are observed after acute treatment and in the absence of any gradient.

Interestingly, deletion of TREM2 had no significant impact on scratch wound closure rates, over a time scale of 24 hr in the presence of a constant concentration of ADP (Ilina and Friedl, 2009). However, we find in a directional chemotaxis assay toward a gradient of ADP concentration that TREM2 KO cells are unable to migrate as efficiently as WT cells, concordant with previous studies showing reduced migration of TREM2 KO cells toward Aβ plaques (McQuade et al., 2020). Enhanced ADP signaling likely abolishes the ability of TREM2 KO cells to distinguish gradations of the agonist, and this loss of gradient sensing results in an inability to perform directed migration. We speculate that increased ADP Ca2+ signaling in TREM2 KO cells may result in Ca2+ signaling domains that are no longer restricted to the cell region near to the highest ADP concentrations and disrupt the polarity of key signaling molecules that drive directed cell motility.

The amplitude and duration of Ca2+ signals shape specificity of downstream cellular responses. Our experiments with ADP in Ca2+-free medium revealed that a transient Ca2+ signal is insufficient to induce microglial motility in either WT or TREM2 KO cells. Previous studies have shown that mouse microglia with genetic deletion of STIM1 or Orai1 also show defects in cell migration to ATP (Michaelis et al., 2015; Lim et al., 2017), likely because diminished SOCE renders them unable to sustain Ca2+ signals in response to ATP. The dependence of motility on prolonged purinergic Ca2+ signals may thus be a general feature of microglia. In contrast, a Ca2+ transient can initiate some process extension in TREM2 KO but not in WT microglia, suggesting a threshold for ADP signaling that is reached in KO but not WT cells, and highlighting subtle differences in the Ca2+ requirement for motility and process extension in TREM2 KO microglia.

To directly monitor Ca2+ signaling and motility simultaneously in individual cells, we developed a novel iPSC-microglia cell line expressing a genetically encoded, ratiometric Ca2+ indicator Salsa6f, a GCaMP6f-tdTomato fusion protein. Because Salsa6f allows simultaneous measurement of Ca2+ signal and tracking of processes, this Salsa6f iPSC line is likely to be a useful tool to dissect the relationship between Ca2+ signaling and the function of various iPSC-derived human cell types, including neurons, astrocytes, and microglia. In addition, this line may be readily xenotransplanted for use with human/microglia chimeric models to examine functional Ca2+ responses to injury and pathology in vivo. Using Salsa6f-expressing microglia, we uncovered critical differences in how Ca2+ levels tune motility in WT and TREM2 KO microglia. By tracking instantaneous velocity at the same time as Salsa6f Ca2+ ratios in individual cells, we found that TREM2 KO cell motility showed a greater sensitivity to changes in cytosolic Ca2+ levels with significantly higher speeds than WT cells at lower Ca2+ and a more dramatic reduction in cell speed at high Ca2+ levels. It is possible that high cytosolic Ca2+ serves as a temporary STOP signal in microglia similar to its effects on T cells (Negulescu et al., 1996); we further speculate that TREM2 KO cells may be more subject to this effect with ADP, given the higher expression of P2RY12 and P2Y13 receptors. Accordingly, reducing cytosolic Ca2+, resulted in increased mean speed, displacement, and straighter paths for TREM2 KO iPSC-microglia, but had no effect on these motility metrics in WT cells, suggesting that TREM2 KO cells may display a greater dynamic range in regulating their motility in response to sustained Ca2+ elevations. Consistent with this observation, chemotaxis in TREM2 KO cells was restored by partially inhibiting P2Y12 receptors. In response to neurodegenerative disease, microglia downregulate P2Y12 receptors (Krasemann et al., 2017; Sala Frigerio et al., 2019; Lou et al., 2016). Active regulation of purinergic receptor expression is critical for sensing ADP gradients and decreasing motility near the chemotactic source. In vivo studies (Hasselmann et al., 2019; Krasemann et al., 2017; McQuade et al., 2020) suggest that TREM2 KO microglia are unable to downregulate P2Y receptor expression upon activation, which may lead to the known chemotactic deficits in these cells.

The studies presented here provide evidence that reducing purinergic receptor activity may be clinically applicable in Alzheimer’s patients with TREM2 loss-of-function mutations (Cheng-Hathaway et al., 2018; Piers et al., 2020; Parhizkar et al., 2019). Pharmacologically targeting P2Y12 receptors to dampen both the Ca2+-dependent (PLC) and -independent (DAG) arms of the GPCR signaling pathway may be useful to control microglial activation and motility. However, our results suggest that altering downstream Ca2+ flux may be sufficient, and thus, CRAC (Orai1) channel blockers that would specifically inhibit the sustained Ca2+ signals without affecting the initial Ca2+ transient or the activation of DAG may provide a more targeted approach.

Currently, TREM2 activating antibodies are being examined in early stage clinical trials for AD (Alector Inc, 2021; Wang et al., 2020), making it critically important to understand the broad consequences of TREM2 signaling. Therefore, an understanding of how TREM2 influences responses to purinergic signals and regulates cytosolic Ca2+ in human iPSC-microglia is critical. Beyond TREM2, we have found that protective variants in MS4A6A and PLCG2 gene expression also decrease P2Y12 and P2Y13 receptor expression (unpublished data), suggesting that this mechanism of microglial activation could be common across several microglial AD risk loci.

In summary, deletion of TREM2 renders iPSC-microglia highly sensitive to ADP, leading to prolonged Ca2+ influx, which increases cell displacement by decreasing cell turning. Despite this, TREM2 KO microglia show a defect in chemotaxis that is likely due to their inability to sense ADP gradients and make appropriate course corrections. Decreasing purinergic signaling in TREM2 KO microglia rescues directional chemotactic migration. We suggest that purinergic modulation or direct modulation of Ca2+ signaling could provide novel therapeutic strategies in many AD patient populations, not solely those with reduced TREM2 function.

Materials and methods

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
Cell line (human)WT iPSC-microgliaUCI ADRC iPSC CoreADRC5;orgin: Blurton-Jones labiPSC-derived microglial line
Cell line (human)TREM2 KO iPSC microgliaBlurton-Jones labADRC5 Clone 28-18;orgin: Blurton-Jones labCRISPR-mediated knockout of TREM2 on the WT iPSC line
Cell line (human)WT GFP-expressing iPSC-microgliaCoriellAICS-0036;RRID:CVCL_JM19iPSC-line with GFP tagged to αtubulinOriginally developed by Dr. Bruce Conklin
Cell line (human)TREM2 KO GFP-expressing iPSC-microgliaBlurton-Jones labGFP Clone 1from above RRIDCRISPR-mediated knockout of TREM2 on the WT GFP+ iPSC line
Cell line (human)WT RFP-expressing iPSC-microgliaCoriellAICS-0031-035;RRID:CVCL_LK44iPSC-line with RFP tagged to αtubulinOriginally developed by Dr. Bruce Conklin
Cell line (human)TREM2 KO RFP-expressing iPSC-microgliaBlurton-Jones labRFP Clone 6from above RRIDCRISPR-mediated knockout of TREM2 on the WT RFP+ iPSC line
Cell line (human)WT Salsa6f-expressing iPSC-microgliaUCI ADRC iPSC CoreADRC76 Clone 8;orgin: Blurton-Jones labiPSC-line expressing a GCaMP6f-tdTomato fusion construct (Salsa6f)
Cell line (human)TREM2 KO Salsa6f-expressing iPSC microgliaBlurton-Jones labADRC76 Clone 8, Clone 98;orgin: Blurton-Jones labCRISPR-mediated knockout of TREM2 on the WT Salsa6f+ iPSC line
Cell line (human)Orai1 KO iPSC microgliaBlurton-Jones labADRC76;orgin: Blurton-Jones labCRISPR-mediated knockout of Orai1 on the WT ADRC76 iPSC line
Transfected construct (transgene)Salsa6fAddgenePlasmid# 140188;RRID:Addgene_140188A genetically encoded calcium indicator with tdTomato linked to GCaMP6f by a V5 epitope tag.
OtherDMEM/F12, HEPES, no Phenol redThermo Fisher Scientific11038021Microglia differentiation cell culture medium
OtherTeSR-E8STEMCELL Technologies05990Stem cell culture medium
OtherStemDiff Hematopoietic kitSTEMCELL Technologies05310
Peptide, recombinant proteinNonessential amino acidsGibco11140035
Peptide, recombinant proteinGlutaMAXGibco35050061
Peptide, recombinant protein (human)InsulinSigmaI2643
Peptide, recombinant proteinB27Gibco17504044
Peptide, recombinant proteinN2GibcoA1370701
Peptide, recombinant proteinInsulin-transferrin-seleniteGibco41400045
Peptide, recombinant proteinIL-34PeproTech200-34
Peptide, recombinant proteinTGFβ1PeproTech100-21
Peptide, recombinant proteinM-CSFPeproTech300-25
Peptide, recombinant proteinCX3CL1PeproTech300-31
Peptide, recombinant proteinCD200NovoproteinC311
Peptide, recombinant proteinFibronectinSTEMCELL Technologies07159
OtherMatrigelCorning356231
OtherReLeSRSTEMCELL Technologies5872Human pluripotent stem cell selection and passing reagent
OtherGoat serumThermo Fisher Scientific10,000C
OtherFluorescent beta-amyloid 1–42 (647)AnaSpecAS64161
OtherpHrodo tagged zymosan A beadsThermo Fisher ScientificP35364
OtherpHrodo tagged S. aureusThermo Fisher ScientificA10010
OtherHuman Stem Cell Nucleofector kit 2LonzaVPH-5022
OtherAlt-R CRISPR-Cas9 tracrRNAIDTDNA107253
OtherAlt-R HiFi Cas9 NucleaseIDTDNA1081061
AntibodyAnti-human IBA1(rabbit monoclonal)Wako019-19741;RRID:AB_839504(1:200)
AntibodyGoat anti-rabbit 555(secondary antibody)Thermo Fisher ScientificA21429;RRID:AB_2535850(1:400)
OtherHuman TruStain FcXBioLegendCat# 422301Fc blocking solution5 μL per test
AntibodyBrilliant Violet 421 anti-human P2RY12Primary antibody(mouse monoclonal)BioLegend392105;clone 16001E;RRID:AB_2783290(5 μL) per test
AntibodyBrilliant Violet 421 mouse IgG2aκ Isotype control mouseBioLegend407117;clone MOPC-173;RRID:AB_2687343(5 μL) per test
Chemical compound, drugFluo-4 AMThermo Fisher ScientificF14201
Chemical compound, drugFura-red AMThermo Fisher ScientificF3021
Chemical compound, drugPluronic F-127Thermo Fisher ScientificP3000MP
Chemical compound, drugCal-520 AMAAT Bioquest21130
Chemical compound, drugCal-590 AMAAT Bioquest20510
Chemical compound, drugci-IP3/PMSiChem6210Caged-inositol triphosphate analog
Chemical compound, drugHoeschstThermo Fisher ScientificR37165
Chemical compound, drugADPSigma-AldrichA2754
Chemical compound, drugATPSigma-AldrichA9187
Chemical compound, drugUTPSigma-AldrichU1006
Chemical compound, drugPSB 0739Tocris3983
Chemical compound, drugMRS 2211Tocris2402
Chemical compound, drugSynta66Sigma-AldrichSML1949Orai channel inhibitor
Chemical compound, drug2-APBSigma-AldrichD9754
Chemical compound, drugGadoliniumSigma-AldrichG7532
Chemical compound, drugEGTASigma-AldrichE8145
Chemical compound, drug1-ThioglycerolSigma-AldrichM6145
Chemical compound, drugCloneRSTEMCELL Technologies05888Defined supplement for single-cell cloning of human iPS cells
Chemical compound, drugThiazovivinSTEMCELL Technologies72252ROCK inhibitor
Other35 mm glass-bottom dishMatTekP35G-1.5-14C1.5 coverslip, 14 mm glass diameter
OtherIncubation perfusion Lid for 35 mm dishesTokai HitLV200-D35FMEPerfusion lid with inlet and outlet
OtherLaser Scanning Confocal MicroscopeOlympusFV3000Equipped with Resonant Scanner, IX3-ZDC2 Z-drift compensator, 40× silicone oil objective, 20× air objective
OtherStage Top Incubation SystemTokai HitSTXGTemperature and humidity control for FV3000 microscope stage
OtherNikon Eclipse Ti microscope systemNikonEquipped with a 40× oil immersion objective (NA 1.3; Nikon) and an Orca Flash 4.0LT CMOS camera (Hamamatsu)
OtherChemotaxis Assay ChamberHansang Cho Lab
OtherIncuCyte S3 Live-Cell Analysis SystemSartorius
OtherEssen Incucyte WoundMakerSartorius4493
Software, algorithmGraphPad Prism 9.1.0Data analysis, statistical analysis
Software, algorithmFiji (ImageJ)Image analysis
Software, algorithmIncucyte 2020CImage acquisition and analysis
Software, algorithmImaris 9.7.0Cell tracking and image analysis
Software, algorithmFlikaImage analysis
Software, algorithmDiPer Excel MacrosPMID:25033209Data analysis, directional persistence

Generation of iPSCs from human fibroblasts

Request a detailed protocol

Human iPSC lines were generated by the University of California, Irvine Alzheimer’s Disease Research Center (UCI ADRC) Induced Pluripotent Stem Cell Core from subject fibroblasts under approved Institutional Review Boards (IRB) and human Stem Cell Research Oversight (hSCRO) committee protocols. Informed consent was received from all participants who donated fibroblasts. Reprogramming was performed with nonintegrating Sendai virus in order to avoid integration effects. To validate the karyotype and identity of iPSC lines, cells were examined via Microarray-based Comparative Genomic Hybridization (aCGH, Cell Line Genetics). Sterility and confirmation of mycoplasma negativity was examined every 10 passages and proceeding experimentation via MycoAlert (Lonza). Pluripotency was verified by Pluritest Array Analysis and trilineage in vitro differentiation. Additional GFP- and RFP-αtubulin-expressing iPSC lines (AICS-0036 and AICS-0031-035) were purchased from Coriell and originally generated by Dr. Bruce Conklin. Each Coriell line is provided with a corresponding certificate of analysis that verifies the correct reporter sequence insertion site, lack of plasmid integration, growth rate, expression of pluripotency markers, normal karyotype, sterility including mycoplasma negative, and identity of line via short tandem repeat (STR). See here and here.

CRISPR-mediated knockout of TREM2 and ORAI1

Request a detailed protocol

Genome editing to delete TREM2 was performed as in McQuade and Blurton-Jones, 2021. Briefly, iPSCs were nucleofected with ribonucleoprotein complex targeting the second exon of TREM2 and allowed to recover overnight. Transfected cells were dissociated with pre-warmed Accutase then mechanically plated to 96-well plates for clonal expansion. Genomic DNA from each colony was amplified and sequenced at the cut site. The amplification from promising clones was transformed via TOPO cloning for allelic sequencing. Knockout of TREM2 was validated by Western blotting (AF1828, R&D) and HTRF (Cisbio) (McQuade et al., 2020). A similar strategy was used to delete ORAI1 using an RNP complex of Cas9 protein coupled with a guide RNA (5′ cgctgaccacgactacccac) targeting the second exon of ORAI1. The resulting ORAI1 clones were then validated to exhibit a normal for karyotype, identity, pluripotency, and sterility via Microarray-based Comparative Genomic Hybridization (aCGH, Cell Line Genetics), tri-lineage differentiation, and MycoAlert mycoplasma testing.

iPSC-microglia differentiation

Request a detailed protocol

iPSC-microglia were generated as described in McQuade et al., 2018 and McQuade and Blurton-Jones, 2021. Briefly, iPSCs were directed down a hematopoietic lineage using the STEMdiff Hematopoietic kit (STEMCELL Technologies). After 10–12 days in culture, CD43+ hematopoietic progenitor cells were transferred into a microglia differentiation medium containing DMEM/F12, 2× insulin-transferrin-selenite, 2× B27, 0.5× N2, 1× GlutaMAX, 1× nonessential amino acids, 400  μM monothioglycerol, and 5  μg/mL human insulin. Media was added to cultures every other day and supplemented with 100  ng/mL IL-34, 50  ng/mL TGF-β1, and 25 ng/mL M-CSF (PeproTech) for 28 days. In the final 3 days of differentiation, 100  ng/mL CD200 (Novoprotein) and 100  ng/mL CX3CL1 (PeproTech) were added to culture.

Confocal laser scanning microscopy

Request a detailed protocol

Unless otherwise stated, cells were imaged on an Olympus FV3000 confocal laser scanning inverted microscope equipped with high-speed resonance scanner, IX3-ZDC2 Z-drift compensator, 40× silicone oil objective (NA 1.25), and a Tokai-HIT stage top incubation chamber (STXG) to maintain cells at 37°C. To visualize Salsa6f, 488 nm and 561 nm diode lasers were used for sequential excitation of GCaMP6f (0.3% laser power, 450 V channel voltage, 494–544 nm detector width) and TdTomato (0.05% laser power, 450 V channel voltage, 580–680 nm detector width), respectively. Fluo-4 and Fura-red were both excited using a 488 nm diode laser (0.07% laser power, 500 V channel voltage, 494–544 nm detector width for Fluo-4; 0.07% laser power, 550 V channel voltage, 580–680 nm detector for Fura-red). Two high-sensitivity cooled GaAsP PMTs were used for detection in the green and red channels, respectively. GFP was excited using the same settings as GCaMP6f. Other image acquisition parameters unique to Ca2+ imaging, microglia process, and cell motility analysis are indicated in the respective sections.

Measurement of intracellular Ca2+

Cell preparation

Request a detailed protocol

iPSC-microglia were plated on fibronectin-coated (5 μg/mL) glass-bottom 35 mm dishes (MatTek, P35G-1.5-14C) overnight at 60% confluence. Ratiometric Ca2+ imaging was done using Fluo-4 AM and Fura-red AM dyes as described previously (McQuade et al., 2020). Briefly, cells were loaded in microglia differentiation medium with 3 μM Fluo-4 AM and 3 μM Fura-red AM (Molecular Probes) in the presence of Pluronic Acid F-127 (Molecular Probes) for 30 min at room temperature (RT). Cells were washed with medium to remove excess dye, and 1 mM Ca2+ Ringer’s solution was added to the 35 mm dish before being mounted on the microscope for live-cell imaging. We note that iPSC-microglia are sensitive to shear forces and produce brief Ca2+ signals in response to solution exchange that are dependent on extracellular Ca2+, and that these are more prominent at 37°C. To minimize these confounding effects, cells were imaged at RT and perfusion was performed gently. Salsa6f-expressing iPSC-microglia were prepared for Ca2+ imaging in the same way as conventional microglia, but without the dye loading steps. The following buffers were used for Ca2+ imaging: (1) 1 or 2 mM Ca2+ Ringer’s solution comprising 155 mM NaCl, 4.5 mM KCl, 1 mM CaCl2, 0.5 mM MgCl2, 10 mM glucose, and 10 mM HEPES (pH adjusted to 7.4 with NaOH); (2) Ca2+-free Ringer’s solution containing 155 mM NaCl, 4.5 mM KCl, 1.5 mM MgCl2, 10 mM glucose, 1 mM EGTA, 10 mM HEPES, pH 7.4. Live-cell imaging was performed as described earlier. Cells were treated with ADP as indicated in the ‘Results’ section.

Data acquisition

Request a detailed protocol

Time-lapse images were acquired in a single Z-plane at 512 × 512 pixels (X = 318.2 μm and Y = 318.2 μm) and at 2–3 s time intervals using Olympus FV3000 software. Images were time averaged over three frames to generate a rolling average and saved as .OIR files.

Data analysis

Request a detailed protocol

Time-lapse videos were exported to Fiji-(ImageJ; https://imagej.net/Fiji), converted to TIFF files (16-bit), and background-subtracted. Single-cell analysis was performed by drawing ROIs around individual cells in the field, and average pixel intensities in the green and red channels were calculated for each ROI at each time point. GCaMP6f/ TdTomato (G/R Ratio) and Fluo-4/Fura-red ratio was then obtained to further generate traces showing single-cell and average changes in cytosolic Ca2+ over time. Single-cell ratio values were used to calculate peak Ca2+ signal and responses at specific time points after agonist application as previously reported (Jairaman and Cahalan, 2021). Peak Ca2+ signal for each cell was baseline-subtracted, which was calculated as an average of 10 minimum ratio values before application of agonist. SOCE rate was calculated as Δ(ratio)/Δt(s–1) over a 10 s time frame of maximum initial rise after Ca2+ addback. Area under the curve (AUC) was calculated using the AUC function in GraphPad Prism.

Microglia process extension analysis

Data acquisition

Request a detailed protocol

GFP-expressing iPSC-microglia were plated overnight on 35 mm glass-bottom dishes at 40–50% confluence. Cells were imaged by excitation of GFP on the confocal microscope at 37°C as described earlier. To study process extension in response to ADP, two sets of GFP images were obtained for each field of view across multiple dishes: before addition of ADP (baseline) and 30 min after application of ADP. Images were acquired as a Z-stack using the Galvo scanner at Nyquist sampling. Adjacent fields of view were combined using the Stitching function of the Olympus FV3000 Software and saved as .OIR files.

Process analysis

Request a detailed protocol

The basic workflow for microglia process analysis was adapted from Morrison et al., 2017. Image stacks (.OIR files) were exported to Fiji (ImageJ) and converted into 16-bit TIFF files using the Olympus Viewer Plugin (https://imagej.net/OlympusImageJPlugin). Maximum intensity projection (MIP) image from each Z-stack was used for further processing and analysis. MIP images were converted to 8-bit grayscale images, to which a threshold was applied to obtain 8-bit binary images. The same threshold was used for all sets of images, both before and after ADP application. Noise reduction was performed on the binary images using the Process -> Noise -> Unspeckle function. Outlier pixels were eliminated using Process -> Noise -> Outliers function. The binary images were then skeletonized using the Skeletonize2D/3D Plugin for ImageJ (https://imagej.net/plugins/skeletonize3d). Sparingly, manual segmentation was used to separate a single skeleton that was part of two cells touching each other. The Analyze Skeleton Plugin (https://imagej.net/plugins/analyze-skeleton/) was then applied to the skeletonized images to obtain parameters related to process length and number of branches for each cell in the imaging field. Processes were considered to be skeletons > 8 μm. The data was summarized as average process length and number of branches, before and after ADP application for a specific imaging field, normalized to the number of cells in the field that allowed for pairwise comparison. Additionally, single-cell data across all experiments were also compared in some instances.

IP3 uncaging

Request a detailed protocol

Whole-field uncaging of i-IP3, a poorly metabolized IP3 analog, was performed as previously described (Lock et al., 2016) with minor modifications. Briefly, iPSC-microglia were loaded for 20 min at 37°C with either Cal520 AM or Cal590 AM (5 μM, AAT Bioquest), and the cell-permeable, caged i-IP3 analog ci-IP3/PM (1 μM, SiChem) plus 0.1% Pluronic F-127 in Microglia Basal Medium. Cells were washed and incubated in the dark for further 30 min in a HEPES-buffered salt solution (HBSS) whose composition was (in mM) 135 NaCl, 5.4 KCl, 1.0 MgCl2, 10 HEPES, 10 glucose, 2.0 CaCl2, and pH 7.4. Intracellular Ca2+ ([Ca2+]i) changes were imaged by employing a Nikon Eclipse Ti microscope system (Nikon) equipped with a 40× oil immersion objective (NA 1.3; Nikon) and an Orca Flash 4.0LT CMOS camera (Hamamatsu). Cal520 or Cal590 were excited by a 488 or a 560 nm laser light source (Vortran Laser Technologies), respectively. i-IP3 uncaging was achieved by uniformly exposing the imaged cells to a single flash of ultraviolet (UV) light (350–400 nm) from a xenon arc lamp. UV flash duration, and thus the amount of released i-IP3 was set by an electronically controlled shutter.

Image acquisition was performed by using Nikon NIS (Nikon) software. After conversion to stack TIFF files, image sequences were analyzed with Flika, a custom-written Python-based imaging analysis software (https://flika-org.github.io/; Ellefsen et al., 2014). After background subtraction, either Cal520 or Cal590 fluorescence changes of each cell were expressed as ∆F/F0, where F0 is the basal fluorescence intensity and ∆F the relative fluorescence change (Fx – F0). Data are reported as superplots (Lord et al., 2020) of at least three independent replicates. Experiments were reproduced with two independent lines. Comparisons were performed by unpaired nonparametric t-test.

Immunocytochemistry

Cells were fixed with 4% paraformaldehyde for 7 min and washed 3× with 1× PBS. Blocking was performed at RT for 1 hr in 5% goat serum, 0.1% Triton5 X-100. Primary antibodies were added at 1:200 overnight 4°C (IBA1, 019-19741, FUJIFILM Wako). Plates were washed 3× before addition of secondary antibodies (goat anti-rabbit 555, Thermo Fisher Scientific) and Hoechst (Thermo Fisher Scientific). Images were captured on an Olympus FV3000RS confocal microscope with identical laser and detection settings. Images were analyzed with Imaris 9.7.0 software. We note that our attempt to verify Orai1 expression at the protein level was unsuccessful as the antibody used (Alomone, Cat# ALM-025, clone# 3F11/D10/B9) did not stain WT microglia in either immunostaining or western blot experiments.

Flow cytometry

Request a detailed protocol

iPSC-derived microglia were seeded on fibronectin-coated 12-well plates at 200,000 cells/well. Cells were harvested and centrifuged in FACS tubes at 300 × g for 5 min at 4°C. The cell pellet was subsequently resuspended in FACS buffer (1× PBS + 0.5% FBS). Fc receptors were blocked with a blocking buffer (BioLegend TruStain FcX in 1× PBS + 10% FCS). Cells were then incubated with Brilliant Violet 421-labeled anti-human P2Y12 receptor antibody (clone S16001E, BioLegend, Cat# 392106) or with IgG2a isotype control antibody (clone MOPC-173, BioLegend, Cat# 400260) for 30 min at 4°C. Cells were washed, pelleted, and then resuspended in FACS buffer. Clone S16001E binds to the extracellular domain of the P2Y12 and permits labeling of PM P2Y12 receptors. Data were acquired using Novocyte Quanteon flow cytometer (Agilent) and analyzed using FlowJo analysis software (FlowJo v10.8.1 LLC Ashland, OR).

Scratch wound assay

Request a detailed protocol

Nondirectional motility was analyzed using Essen Incucyte WoundMaker. iPSC-microglia were plated on fibronectin (STEMCELL Technologies) at 90% confluence. Scratches were repeated 4× to remove all cells from the wound area. Scratch wound confluency was imaged every hour until scratch wound was closed (15 hr). Confluence of cells within the original wound ROI was calculated using IncuCyte 2020C software.

Imaris cell tracking

Request a detailed protocol

For motility assays, iPSC-microglia were tracked using a combination of manual and automatic tracking in Imaris 9.7.0 software. For videos of GFP lines, cells were tracked using spot identification. For videos of Salsa6f lines, surface tracking was used to determine ratiometric Ca2+ fluorescence and motility per cell. In both conditions, tracks were defined by Brownian motion with the maximum distance jump of 4 µm and 10 frame disturbance with no gap filling. Tracks shorter than 3 min in length were eliminated from analysis. After automated track formation, tracks underwent manual quality control to eliminate extraneous tracks, merge falsely distinct tracks, and add missed tracks. After export, data was plotted in Prism 9.1.0 or analyzed in Excel using DiPer Macros for Plot_At_Origin (translation of each trajectory to the origin) and MSD(t) = 4D(t-P(1-e^(-t/P))), where D is the diffusion coefficient, t is time, and P represents directional persistence time (time to cross from persistent directionality to random walk) (Gorelik and Gautreau, 2014). From Imaris, speed was calculated as instantaneous speed of the object (μm/s) as the scalar equivalent to object velocity. These values were transformed to μm /min as this time scale is more relevant for the changes we observed. Mean track speed represents the mean of all instantaneous speeds over the total time of tracking. 10 min displacement is calculated by (600) * (TDL/TD), where TDL is the track displacement length (distance between the first and last cell position) represented as TDL = p(n) - p(1) for all axes, where the vector p is the distance between the first and last object position along the selected axis, and TD is the track duration represented as TD = T(n) - T(1), where T is the time point of the first and final time point within the track. Frame-to-frame displacement is calculated as p(n) – p(n-1) for all the different frames in a cell track. Track straightness is defined as TDL/TL, where TDL is the track displacement as described above and TL is the track length representing the total length of displacements within the track TL = sum from t = 2 to n of |p(t)-p(t-1)|.

Generation of Salsa6f-expressing iPSC lines

Request a detailed protocol

iPSCs were collected following Accutase enzymatic digestion for 3 min at 37°C. 20,000 cells were resuspended in 100 μL nucleofection buffer from Human Stem Cell Nucleofector Kit 2 (Lonza). Salsa6f-AAVS1 SHL plasmid template (2 μg; Vector Builder) and RNP complex formed by incubating Alt-R S.p. HiFi Cas9 Nuclease V3 (50 μg; IDTDNA) was fused with crRNA:tracrRNA (IDTDNA) duplex for 15 min at 23°C. This complex was combined with the cellular suspension and nucleofected using the Amaxa Nucleofector program B-016. To recover, cells were plated in TeSR-E8 (STEMCELL Technologies) media with 0.25 μM thiazovivin (STEMCELL Technologies) and CloneR (STEMCELL Technologies) overnight. The following day, cells were mechanically replated to 96-well plates in TeSR-E8 media with 0.25 μM thiazovivin and CloneR supplement for clonal isolation and expansion. Plates were screened visually with a fluorescence microscope to identify TdTomato+ clones. Genomic DNA was extracted from positive clones using Extracta DNA prep for PCR (Quantabio) and amplified using Taq PCR Master Mix (Thermo Fisher Scientific) to confirm diallelic integration of the Salsa6f cassette. A clone confirmed with diallelic Salsa6f integration in the AAVS1 SHL was then retargeted as previously described (McQuade et al., 2020) to knock out Trem2.

Phagocytosis assay

Request a detailed protocol

Phagocytosis of transgenic iPSC-microglia was validated using IncuCyte S3 Live-Cell Analysis System (Sartorius) as in McQuade et al., 2020. Microglia were plated at 50% confluency 24 hr before substrates were added. Cells were treated with 50  μg/mL pHrodo tagged human AD synaptosomes (isolated as described in McQuade et al., 2020), 100  ng/mL pHrodo tagged zymosan A beads (Thermo Fisher Scientific), 100 ng/mL pHrodo tagged Staphylococcus aureus (Thermo Fisher Scientific), or 2  μg/mL fluorescent beta-amyloid (AnaSpec). Image masks for fluorescence area and phase were generated using IncuCyte 2020C software.

Chemotaxis assay

Request a detailed protocol

iPSC-microglia were loaded into the angular chamber (2–5K cells/device) to test activation and chemotaxis toward the central chamber containing either ADP (100 ng/mL or 234 nM) or vehicle. When noted, PSB 0739 (10 μM) was added to both the central and angular chamber to inhibit P2Y12 receptors. To characterize motility, we monitored the number of recruited microglia in the central chamber for 4 days under the fully automated Nikon TiE microscope (10× magnification; Micro Device Instruments, Avon, MA).

Statistical analysis

Request a detailed protocol

GraphPad Prism (versions 6.01 and 8.2.0) was used to perform statistical tests and generate p-values. We used standard designation of p-values throughout the figures (ns, not significant or p≥0.05; *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001). Traces depicting average changes in cytosolic Ca2+ over time are shown as mean ± standard error of the mean (SEM). Accompanying bar graphs with bars depicting mean ± SEM provide a summary of relevant parameters (amplitude of Ca2+ response, degree of store release, rate of Ca2+ influx, etc.) as indicated. Details of the number of replicates and the specific statistical test used are provided in the individual figure legends.

Data availability

RNA sequencing data referenced in Figure 1- figure supplement 2 is available through Gene Expression Omnibus: GSE157652.

The following data sets were generated
    1. McQuade A
    (2020) NCBI Gene Expression Omnibus
    ID GSE157652. Transcriptomic and functional deficits in human TREM2-/- microglia impair response to Alzheimer's pathology in vivo [RNA-seq].

References

    1. Honda S
    2. Sasaki Y
    3. Ohsawa K
    4. Imai Y
    5. Nakamura Y
    6. Inoue K
    7. Kohsaka S
    (2001)
    Extracellular ATP or ADP induce chemotaxis of cultured microglia through Gi/o-coupled P2Y receptors
    The Journal of Neuroscience 21:1975–1982.

Decision letter

  1. Murali Prakriya
    Reviewing Editor; Northwestern University, United States
  2. Richard W Aldrich
    Senior Editor; The University of Texas at Austin, United States
  3. Mohamed Trebak
    Reviewer; University of Pittsburgh, United States

Our editorial process produces two outputs: i) public reviews designed to be posted alongside the preprint for the benefit of readers; ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "TREM2 regulates purinergic receptor-mediated calcium signaling and motility in human iPSC-derived microglia" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Richard Aldrich as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Mohamed Trebak (Reviewer #2).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission. Reviewers had concerns on two broad aspects of the paper: the calcium signaling studies, and the results relating to cell motility. In addition, other concerns were raised which are described in the individual reviewer comments.

Essential revisions:

1) A more rigorous examination of protein expression should be performed with antibody validation to determine if there are compensatory changes at the protein level of other relevant Ca2+ signaling proteins (Figure 1 Suppl2) that could affect the results. Westerns should be performed for STIM1 and STIM2, Orai1 and IP3R 1/2/3, for which there are reliable Abs. Although the IF as performed for Orai1 is acceptable, the authors do not show positive and negative controls and a thorough validation of this Orai1Ab in IF. This is strongly recommended given the poor quality of Orai1 Abs in general. Likewise, staining for P2YR12 appears internal and the Ab should be validated. A WB on P2YR12 and P2RY13 in addition would help.

2) Important: It is clear that the TREM2 KO cells have significantly lower STIM1 expression, which would be expected to impact SOCE under conditions of physiological agonists stimulation (ADP, ATP). The conclusion based on maximal activation with thapsigargin does not address this issue. What is the impact of reduced STIM1 expression on the Ca signals and motility?

3) A Ca2+ off/on protocol with ADP (Figure 1) as done for thapsigargin should be performed. In Figure 2B, ADP was applied in the absence of external Ca2+, but sadly those recordings did not subsequently replenish Ca2+ to gauge Ca2+ entry. Low concentrations of Gd3+ (which should not alter Ca2+ release) reduce the initial Ca2+ release peak, suggesting that there is a contribution of Ca2+ entry to this initial response. Also, it would have been helpful to include in Figure 1 Suppl1 the control (non KO cells) in these experiments for reference.

4) The conclusion that SOCE is mediated by Orai1 should be validated with a genetic approach as 2-APB and La3+ are non-specific.

5) Conversely, it is not clear if store content is depleted to a greater extent following ADP receptor stimulation in the KO cells. Assess ER calcium content in response to ADP in WT and KO cells or with an ionomycin pulse.

6) Figure 3: The WT data in this figure don't include KO cells. How do the TREM2 KOs compare to WT in the scratch wound closure assay? The KO cells seem to be tested, but only in the absence of ADP (Figure 4 Sup 1A). Is there a difference in wound closure after ADP treatment? This point should be evaluated to assess the effects of altered ADP Ca signaling on non-directional cell movement in a physiologically relevant paradigm.

7) Terminology: The term "Ca clamp" in the way it has been used here could be misconstrued by readers. Cellular Ca is not clamped. It varies within each cell depending on the activity of the pumps and other clearance mechanisms in the cell. The motility measurements are simply done after extracellular Ca is restored to the cells after thapsigargin treatment, at time points where the cytosolic Ca is lower in 0.2 mM Ca compared to 2 mM Ca. This phrasing (Ca clamp) needs to be changed to eliminate confusion that cellular calcium is "clamped" analogous to voltage-clamp.

8) Figure 6D TREM2 KOs (but not WT) cells show differences in motility between 0.2 mM and 2 mM extracellular Ca even after thapsigargin treatment with no ADP exposure. This is strong indication that motility differences in TREM2 KO cells are due to more than just alterations in purinergic signaling. (Consistent with lack of additional ADP effects in Figure 6 Supplement 1). What is the role of these intrinsic changes in cell motility in response to cellular Ca elevations in the presence and absence of nucleotides?

9) Figure 6D: The data for WT cells should be directly compared to the KOs at each extracellular Ca concentration to dissect out differences in KO cells at each Ca condition.

10) The data in Figure 4K should be reanalyzed to compare differences in the absolute levels of branch complexity (# of branches and process length) in the two groups.

11) The apparent disconnect between the larger mean square distance travelled in the absence of nucleotides in the KO with no change in the mean speed needs to be clarified. Is this due to cancelling of speed vectors with opposite directions?

Reviewer #1 (Recommendations for the authors):

1) Figure 3: The WT data in this figure don't include KO cells. How do the KOs compare to the WT in the scratch wound closure assay? The KO cells seem to be tested, but only in the absence of ADP (Figure 4 Sup 1A). Is there a difference in wound closure after ADP treatment? This point should be evaluated to assess the effects of altered ADP Ca signaling on non-directional cell movement in a physiologically relevant paradigm.

2) Figure 4: This reviewer is not convinced that reduced turning accounts for the similar speed in the KO despite the larger mean square distance travelled. If turning is reduced and the mean square distance is doubled in the KO, then effectively, the average distance travelled over time (speed) should be greater in the KO than in the WT. The logic of this inference and how it explains the data (similar average speed but much greater root mean square displacement in the KO) needs to be better clarified. Since the mean square distance is so much larger, it suggests that the speed vectors with opposite directions are cancelling out in the KO cells to yield the same average velocity (which could theoretically even occur from increased, not decreased turning).

3) Figure 4K: It seems that the baseline branch complexity is low to start with in the KOs, hence the fold change from baseline seems larger. From the images shown, there appears to be no difference following ADP treatment between the two groups. This data should be reanalyzed to compare differences in the absolute levels of branch complexity (# of branches and process length) in the two groups.

4) The use of the term Ca clamp is misleading in the way it has been used here. Cellular Ca is, in fact, not clamped, but changes within each cell depending on the activity of pumps and other calcium clearance mechanisms. The motility measurements are simply done after extracellular Ca is restored to cells after thapsigargin treatment, at time points where the cytosolic Ca is lower in 0.2 mM Cacompared to 2 mM Ca. This phrasing (Ca clamp) needs to be changed to eliminate confusion that cellular Ca levels are "clamped" to defined values analogous to voltage clamp.

5) Figure 6D: It is noteworthy that TREM2 KOs, but not WT cells, show differences in motility between 0.2 mM and 2 mM extracellular Ca even after thapsigargin treatment with no ADP exposure. This is strong suggestion that motility differences in TREM2 KO cells are due to more than just alterations in purinergic signaling. In fact, this inference is consistent by the lack of additional effects of adding ADP in Figure 6 Supplement 1. These intrinsic changes in cell motility in response to cellular Ca elevations need further clarification.

6) Figure 6D: Related to this figure, the data for WT cells should be directly compared to the KOs at each extracellular Ca concentration to dissect out differences in KO cells at each Ca condition.

Other:

– Stim1 is expressed at different levels between the WT and TREM2 KO. It has a significant p-value (Figure 1 Sup2D). This decrease could affect multiple aspects of microglia including cell motility in the presence of physiologically relevant agonists. This needs to be addressed.

– A technical point related to Figure 4 is that figures 4C and 4G should be labelled "velocity" rather than speed, as it seems the only way in which the mean squared distance can be twice as large in the KO but not the velocity if the speed vectors with opposite directions are cancelling out in the KO.

– The dose of ADP used in Figure 7 is extremely high. 100ng/mL (which is 234uM by my calculation). By contrast, Figure 4 used 2.5uM. The 100 fold higher concentration in the chemotaxis assay needs to be justified. Further, it should be determined whether the ADP used diffuses into both chambers might be expected. This super high dose of ADP may have non-physiological effects that could explain the seemingly differing phenotypes in Figures 4 and 7.

– Figure 5 is a great set of experiments to show that the Salsa6f does not affect end points of the study, but may be better as a supplementary figure rather than a main figure. If the paper is about the TREM2 KO versus WT then the main figures should stick with looking at differences between the two with maybe a sub figure to show validation.

Reviewer #2 (Recommendations for the authors):

– A more rigorous approach could be used to document the presence or absence of compensatory changes at the protein protein level of other Ca2+ signaling proteins (Figure 1 Suppl2). Preferably Westerns should be performed for STIM1 and STIM2, Orai1 and IP3R 1/2/3, for which there are reliable Abs. Although IF as performed for Orai1 is acceptable, the authors do not show positive and negative controls and a thorough validation of this Orai1Ab in IF.

– Along the same lines as comment above, Staining for P2YR12 appears internal. Has the Ab been validated? Can authors perform a WB on P2YR12 and P2RY13 in addition?

– Did the authors check on whether the expression of SECRA is altered at the protein level? Did the authors perform CRAC channel recordings in Control and TREM2 KO cells?

– It is quite surprising that the authors did not opt to use the Ca2+ off/on protocol with ADP (Figure 1) as they have done with thapsigargin. In Figure 2B, ADP was applied in the absence of external Ca2+, but sadly those recordings did not subsequently replenish Ca2+ to gauge Ca2+ entry. One can rationalize the result of 2-APB (Figure 1 Suppl1), a non-specific channel blocker that also inhibits IP3Rs. However, the fact that low concentrations of Gd3+ (which should not alter Ca2+ release) reduce the initial Ca2+ release peak, suggests that there is a contribution of Ca2+ entry to this initial response. Also, it would have been helpful to include in Figure 1 Suppl1 the control (non KO cells) in these experiments for reference.

Additional thoughts to the attention of the authors:

– The authors have speculated on the fact that increased cytosolic Ca2+ might lead to a "spill" outside the restricted Ca2+ nanodomain, thus disrupting polarity. Do the authors have any evidence of altered location of leading edge vs trailing edge proteins or focal adhesion proteins in TREM2 KO cells?

– Have the authors performed RNAseq on WT control and TREM2 KO cells? If so, does pathway analysis shows changes in the Cell motility genes between these two groups? This might offer some interesting clues to pursue in future studies.

Reviewer #3 (Recommendations for the authors):

This study by Jairaman et al. describes how iPSC-derived microglia exhibit exaggerated cytosolic Ca2+ responses to ADP stimulation in TREM2KO cells, and that this leads to a defect in turning behaviour and hence no directed migration to a chemotactic signal.

Overall, the experiments are well conducted, carefully controlled and the findings are new and exciting. The authors nicely dissect out the underlying molecular basis for the exaggerated Ca2+ responses to ADP and then extend their findings to cell movement and directed migration. Given the substantial body of evidence linking microglia to the pathogenesis of Alzheimer's disease, and the role for TREM2, the work by Jairaman et al. is of translational significance. As an aside, the introduction of the calcium-sensitive reporter Salsa6F is a welcome new tool in the arsenal for recording cytosolic calcium. Overall, this is an elegant, novel and important study. Nevertheless, I have a few comments/suggestions.

1. The authors argue that the increased Ca2+ plateau to ADP in TREM2KO cells is due to enhanced Ca2+ release from the stores ie greater store depletion. Evidence is presented that maximal SOCE is not compromised in TREM2KO cells, that store content under resting conditions is unaffected, InsP3R activities are similar. But the authors do not demonstrate that store content falls more following ADP receptor stimulation in the KO cells. This could be shown by directly measuring ER calcium to ADP in WT and KO cells or by applying ADP in Ca-free solution and then assessing store content with an ionomycin pulse. The ionomycin response should be smaller in the TREM2KO cells, after ADP exposure.

2. The authors suggest that an increase in P2Y12 and P2Y13 receptor expression in the KO cells accounts for the increased Ca2+ release from the ER. However, the increase in P2Y12 protein levels is modest at best, and the increase for P2y13 is less than 2-fold. The authors' point would be strengthened by showing InsP3 levels are increased in the KO cells compared with WT ones, for the same dose of ADP. Single cell InsP3 probes are available. Alternatively, a population measurement could be carried out.

3. The evidence for an involvement of SOCE is based on two rather non-specific inhibitors (2-APB and Gd3+). P2YRs, like other GPCRs, activate TRPC proteins and these Ca2+-permeable channels could contribute to the Ca2+ plateau. The authors should therefore knock down Orai1 or at least use more selective inhibitors such as Synta66 or the GSK compound.

4. The authors should include some controls to show the P2Y antibodies are indeed specific. For example, they could use a cell line that lacks P2Y12/13. In Figure 2I, J, what happens to the ADP response when both P2Y inhibitors are present at the same time? It would be important to show that the combination of P2Y12 and P2Y13 suppress ADP responses fully.

5. The authors should add a justification for the concentrations of agonists they have used (ADP, ATP, UTP). A dose-response curve is included for ADP but a comment on the doses selected would be helpful.

6. It is nicely shown that Ca2+ influx in TREM2 KO cells leads to microglia motility and process extension to a greater extent than WT cells. This is attributed to the increased SOCE. If so, then one might expect raising external Ca2+ in WT cells to have the same effect. Is this the case?

7. Is anything known mechanistically how the exaggerated Ca2+ signal leads to a chemotaxis defect? The authors' data would suggest that chemotaxis might have a bell-shaped dependence on cytosolic Ca2+, with too little or too much impeding migration towards a cue.

8. Do the authors know whether the effects of SOCE are mediated through a local Ca2+ rise or via a global cytosolic Ca elevation?

9. Perhaps I missed something but Figure 1 S2(D) seems to show a significant decrease in STIM1 levels in the KO cells.

10. The authors present SOCE as the peak signal remaining after 5 minutes. It is not clear whether the value has been subtracted from the pre-stimulation levels. In some graphs, the signal after 5 minutes looks the same as the resting level, but the bar charts show higher values.

https://doi.org/10.7554/eLife.73021.sa1

Author response

Reviewers had concerns on two broad aspects of the paper: the calcium signaling studies, and the results relating to cell motility. In addition, other concerns were raised which are described in the individual reviewer comments.

Essential revisions:

1) A more rigorous examination of protein expression should be performed with antibody validation to determine if there are compensatory changes at the protein level of other relevant Ca2+ signaling proteins (Figure 1 Suppl2) that could affect the results. Westerns should be performed for STIM1 and STIM2, Orai1 and IP3R 1/2/3, for which there are reliable Abs. Although the IF as performed for Orai1 is acceptable, the authors do not show positive and negative controls and a thorough validation of this Orai1Ab in IF. This is strongly recommended given the poor quality of Orai1 Abs in general. Likewise, staining for P2YR12 appears internal and the Ab should be validated. A WB on P2YR12 and P2RY13 in addition would help.

Our approach in this study relies on functional Ca2+ imaging readouts in combination with pharmacological and genetic tools to assess P2Y receptor sensitivity to agonists and antagonists and downstream signaling events that include IP3R activation, ER store-release, and SOCE in WT and TREM2 KO microglia. Given our focus on functional data that are concordant with RNA expression results and point toward the differential expression of P2Y receptors, we did not see a strong rationale for further dissecting the expression of various STIM, Orai, IP3R or pump isoforms at the protein level. New data in the paper strengthen this conclusion. These include the ionomycin and TG-pulse experiments (Figure 3E, and Figure 3—figure supplement 2A, B) together with the ADP-Ca2+ addback experiments (Figure 3D), showing that ADP depletes ER Ca2+ stores to a greater extent in TREM2 KO cells. The new data reinforce the conclusion that increased expression of P2Y receptors in TREM2 KO microglia drives downstream calcium signaling events.

We concur with the reviewer that appropriate antibody controls are needed to support data related to protein expression, and we attempted to address this issue for the Orai1 antibody and P2Y12 and P2Y13 receptor antibodies as described below.

In response to reviewer comments, we generated Orai1 CRISPR-knockout iPSC-microglia and further tested the Orai1 Ab (Alomone, Cat# ALM-025, Clone 3F11/D10/B9) for immunostaining in WT and Orai1 KO iPSC-microglial cell line; antibody staining was found to be nonspecific. We therefore agree with the reviewer that, because of poor quality, Orai1 antibody staining does not provide solid and quantifiable evidence for Orai1 expression or function. Accordingly, we have removed Orai1 immunofluorescence staining data (old Figure 1—figure supplement 2E) from the revised manuscript. Instead, we performed Ca2+ measurements in the newly generated Orai1 KO iPSC-microglia cell-line, and with a more specific pharmacological inhibitor of Orai channels (Synta66) to unambiguously establish the role of Orai1 in mediating SOCE and in maintaining sustained Ca2+ signals by ADP in microglia (new Figure 3—figure supplement 1E and F).

Concerning P2Y expression, we performed control experiments using iPSC microglia treated with siRNA against P2Y12 and P2Y13 receptors to validate the P2Y12 and P2Y13 receptor antibodies (P2Y12 receptor Ab: Sigma, cat# HPA014518, polyclonal; P2Y13 receptor Ab: Alomone, cat# APR-017, polyclonal) for immunostaining. We agree with the reviewer that there is significant non-membrane staining of the cells with the P2Y12 receptor antibody making visualization of P2Y12 receptors on the membrane problematic. Upon further investigation, the P2Y13 receptor antibody was also found to be non-specific, in line with recent reports on the widespread lack of efficacy and specificity of available antibodies to label P2Y13 receptors in ex vivo and in vitro settings including the one used in our study (Alomone, Cat# APR-017; see PMID: 31520551; Suppl. Figure 1). We have therefore removed P2Y12 and P2Y13 immunofluorescence data from the paper. Instead, we now include flow cytometry data using a different antibody that targets the extracellular domain of the P2Y12 receptor (clone 16001E) to show that these receptors are expressed to a greater level on the plasma membrane (PM) of TREM2 KO cells (new Figure 2G). The appropriate isotype control is included in this assay.

2) Important: It is clear that the TREM2 KO cells have significantly lower STIM1 expression, which would be expected to impact SOCE under conditions of physiological agonists stimulation (ADP, ATP). The conclusion based on maximal activation with thapsigargin does not address this issue. What is the impact of reduced STIM1 expression on the Ca signals and motility?

We thank the reviewer for bringing attention to this issue. The volcano plot comparing the transcriptomic expression of STIM1 in WT and TREM2 KO cells shows that STIM1 mRNA expression is below the threshold for what would be considered significant in an RNA seq experiment. In fact, this is the case for all the STIM and Orai isoforms. For RNAsequencing experiments, tens of thousands of genes are tested against the null hypothesis which has led to the use of false discovery rates (FDR) rather than traditional p-values (PMID: 12883005). However, statistical significance based on FDR alone is not enough to determine a meaningful result in RNA-sequencing experiments. Due to highly accurate sequencing, samples may show low variability leading to highly “significant” FDRs with a fold change less than 1. In many cases, this small change in expression would not yield biological differences. For these reasons, we consider only genes that reach both the FDR threshold and the fold change threshold. The data shown here were originally published in (PMID: 33097708) with cutoffs of FDR < 0.05 and -1 < FC < 1.

Unfortunately, the bar-graph in old Figure 1—figure supplement 2D may have given the false impression that differences in STIM1 expression are quite large. We have now remade this bar-graph (Y-axis range from 0 – 1.2) and have also included relative expression of P2Y12 and P2Y13 receptor transcripts for comparison (new Figure 2—figure supplement 1E). STIM1 mRNA expression is modestly reduced in TREM2 KO cells which is consistent with the modestly reduced maximum functional SOCE response (measured after store-depletion with TG). To address the issue about SOCE under conditions of physiological agonist stimulation, we also measured SOCE in response to store-depletion with ADP; new Figure 3D shows that ADP produces greater store-release and therefore engages SOCE to a greater extent in TREM2 KO cells. Based on this, we conclude that the modestly reduced STIM1 expression in TREM2 KO microglia does not play a significant role in determining the differences in ADP-mediated Ca2+ signals between WT and TREM2 KO cells.

3) A Ca2+ off/on protocol with ADP (Figure 1) as done for thapsigargin should be performed. In Figure 2B, ADP was applied in the absence of external Ca2+, but sadly those recordings did not subsequently replenish Ca2+ to gauge Ca2+ entry. Low concentrations of Gd3+ (which should not alter Ca2+ release) reduce the initial Ca2+ release peak, suggesting that there is a contribution of Ca2+ entry to this initial response. Also, it would have been helpful to include in Figure 1 Suppl1 the control (non KO cells) in these experiments for reference.

We thank the reviewer for three excellent suggestions (see our replies i, ii, and iii below).

3a (i) To demonstrate SOCE in response to ADP, we now include a Ca2+ off/on protocol in the presence or absence of Synta66, a more specific inhibitor of Orai channels than Gd3+ or 2-APB. Synta66 significantly inhibited SOCE triggered by TG and ADP in WT and TREM2 KO microglia (Figure 3A, B and Figure 3—figure supplement 1A, B). Additionally, we compared ADP-induced SOCE in WT and TREM2 KO microglia (Figure 3D). The implications of this experiment have been discussed above in Essential revision comment 1 and 2.

3a (ii) We re-examined the issue of whether Ca2+ influx contributes to the initial Ca2+ peak in two ways. First, we compared the height of the initial Ca2+ peak after application of ADP in 1mM Ca2+ and Ca2+ free buffer (Figure 2—figure supplement 1C); there was no significant difference between the two conditions in either WT or TREM2 KO microglia suggesting that the initial Ca2+ peak is driven primarily by store-release. We further tested this by acute addition of Gd3+ or 2-APB with ADP without pre-incubation (Figure 3—figure supplement 1C, D). In this instance, we did not find an inhibition of initial Ca2+ peak. We speculate that the pre-incubation of cells with Gd3+ before addition of ADP in old Figure 1—figure supplement 1 might have non-specifically caused reduction of the initial Ca2+ peak.

3a (iii) Given the complex effects of Gd3+ and 2-APB on Ca2+ signaling in microglia, we did not repeat the Gd3+ and 2-APB experiments in WT microglia. Instead, we include new data as outlined above in Figure 3 and Figure 3—figure supplement 1 showing involvement of Orai channels using Synta66 and using the Orai1 KO line (Figure 3—figure supplement 1E, F).

4) The conclusion that SOCE is mediated by Orai1 should be validated with a genetic approach as 2-APB and La3+ are non-specific.

Thank you for the suggestion. As described above, we now include data using a more specific inhibitor (Synta66) and a newly generated Orai1 KO microglial cell line (Figure 3A, B and Figure 3—figure supplement 1A, B, E, F).

5) Conversely, it is not clear if store content is depleted to a greater extent following ADP receptor stimulation in the KO cells. Assess ER calcium content in response to ADP in WT and KO cells or with an ionomycin pulse.

We thank the reviewer for this suggestion. We examined the extent of ER store-depletion in response to ADP by sequentially pulsing the cells first with ADP and then with ionomycin in Ca2+ free buffer. The Ca2+ release peak in response to ADP was higher in the TREM2 KO cells as expected, and the subsequent ionomycin peak was significantly reduced. These results (Figure 3E and Figure 3—figure supplement 2A, B) indicate greater ER Ca2+ store-release by ADP in TREM2 KO cells. As for the suggestion to directly monitor ER calcium content, transfection of iPSC-microglia shifts them from a resting to a highly activated state in which they downregulate their P2Y receptors, as reported (PMID: 17115040, 28602351, 28930663). Thus, transfection of genetically-encoded probes to measure ER Ca2+ or to measure IP3 levels is problematic.

6) Figure 3: The WT data in this figure don't include KO cells. How do the TREM2 KOs compare to WT in the scratch wound closure assay? The KO cells seem to be tested, but only in the absence of ADP (Figure 4 Sup 1A). Is there a difference in wound closure after ADP treatment? This point should be evaluated to assess the effects of altered ADP Ca signaling on non-directional cell movement in a physiologically relevant paradigm.

As suggested by the reviewer, we performed the scratch wound assay in the presence of ADP (data now included as Figure 7—figure supplement 1, and described in lines 309-313). Interestingly, while ADP speeds up closure of scratch wounds, we found no differences in the wound closure rates between WT and TREM2 KO microglia. The differences in purinergic signaling are more decisive in shaping chemotaxis to ADP. The effects of TREM2 deletion, and the subsequent effects of increased purinergic signaling on microglial motility appear to depend on the specific physiological context, and this will require further investigation in a follow up study.

7) Terminology: The term "Ca clamp" in the way it has been used here could be misconstrued by readers. Cellular Ca is not clamped. It varies within each cell depending on the activity of the pumps and other clearance mechanisms in the cell. The motility measurements are simply done after extracellular Ca is restored to the cells after thapsigargin treatment, at time points where the cytosolic Ca is lower in 0.2 mM Ca compared to 2 mM Ca. This phrasing (Ca clamp) needs to be changed to eliminate confusion that cellular calcium is "clamped" analogous to voltage-clamp.

We agree with this comment, and have accordingly removed the term “Ca2+ clamp” from the manuscript. Instead, we refer to this protocol as a method to investigate the effects of Ca2+ elevation that bypasses purinergic receptor activation on microglial motility.

8) Figure 6D TREM2 KOs (but not WT) cells show differences in motility between 0.2 mM and 2 mM extracellular Ca even after thapsigargin treatment with no ADP exposure. This is strong indication that motility differences in TREM2 KO cells are due to more than just alterations in purinergic signaling. (Consistent with lack of additional ADP effects in Figure 6 Supplement 1). What is the role of these intrinsic changes in cell motility in response to cellular Ca elevations in the presence and absence of nucleotides?

We agree with the comment. The question raised has great potential for further work that is beyond the scope of this study. Purinergic stimulation produces a complex downstream response that includes activation of both Ca2+-dependent and -independent (Gβ/γ -> PI3K -> PIP3, Ras, cAMP etc) pathways. Experiments in Figure 6 (using a protocol to maintain different levels of cytosolic Ca2+ over time using thapsigargin) were done with the goal of isolating the effects of sustained cytosolic Ca2+ levels on motility. The key observation is that motility in TREM2 KO microglia responds to changes in cytoplasmic Ca2+ levels to a greater extent than in WT cells, suggesting that Ca2+ tunes motility differently in WT and TREM2 KO cells. Addition of ADP had no further effect, as the reviewer correctly observes (Figure 6—figure supplement 1). We speculate in the Results section (lines 291-292) that the rise in Ca2+ in this assay may override the complex effects of ADP on motility; this may reflect intrinsic differences in Ca2+-dependent regulation of motility between WT and TREM2 KO cells. We note that the baseline motility characteristics are similar between WT and TREM2 KO microglia (Figure 5—figure supplement 1A).

9) Figure 6D: The data for WT cells should be directly compared to the KOs at each extracellular Ca concentration to dissect out differences in KO cells at each Ca condition.

Yes, this is shown in Figure 6F: instantaneous speeds of WT and TREM2 KO microglia over a range of different cytosolic Ca2+ levels. Additionally, the histogram in Figure 6G compares the percent of cells with instantaneous speeds > 10 µm/min as a function of cytosolic Ca2+ in WT and TREM2 KO cells. This comparison reveals a unimodal relationship in TREM2 KO cells with low cell speeds at high and low cytosolic Ca2+ levels and higher speeds at intermediate Ca2+ levels.

10) The data in Figure 4K should be reanalyzed to compare differences in the absolute levels of branch complexity (# of branches and process length) in the two groups.

We fully agree with the reviewer that the lower baseline branch complexity in the KOs explains the greater fold change in the KOs after ADP application. We also agree that the absolute number and length of branches (normalized to cell number) after ADP treatment is similar between WT and KOs. Based on these comments, we made the following change in the Results section (lines 267-270), “Comparison of the absolute number of branches and process length after ADP treatment, as well as the relative fold-increase in these parameters from baseline indicated that process extension is not affected in TREM2 KO microglia”. We note that the absolute number of branches and the process length (normalized to cell number in each imaging field) was calculated and is shown in the left panels of Figure 5L and M. These results are also shown as paired plots comparing the absolute increase in the average number of branches and length of the processes after ADP treatment in the WT and KO groups per imaging field (Figure 5—figure supplement 2A and B, top row).

11) The apparent disconnect between the larger mean square distance travelled in the absence of nucleotides in the KO with no change in the mean speed needs to be clarified. Is this due to cancelling of speed vectors with opposite directions?

We thank the reviewer for bringing attention to this. The MSD vs time plots (Figure 5B) show that TREM2 KO cells travel farther away from the origin than WT cells after ADP treatment. This could either be because (1) KO cells move faster than WT cells or (2) because they change direction less frequently (or travel in straighter paths) than WT cells while moving at similar speed. Comparing mean track speeds and track straightness (Figure 5C) affirms the latter possibility. The mean track speed (Figure 4C-E, Figure 5C, H) is the mean of all instantaneous speeds for a given track. It is calculated without any regard to the direction of cell motility, and therefore does not have negative values. We agree with the reviewer that the velocity vectors (which take into account the direction of cell movement) cancel each other out to a greater degree in the WT cells, which is reflected in lower values of track straightness. Our data suggest that WT cells remain confined to a smaller region of random walk because they turn more frequently. We have observed analogous distinctions in motility patterns between regulatory T (Treg) cells and inflammatory Th17 cells in the spinal cord of mice, with Treg cells executing a back-n-forth motion with similar mean speeds as Th17 cells, and being confined to smaller regions of the cord while Th17 cells traverse larger regions because they change directions less frequently (PMID:32732436). We now include a sentence in the Discussion section (lines 349-350) to indicate that velocity vectors cancel each other out to a greater extent in WT cells. For clarity, we added a new paragraph in the methods section (lines 922-937) detailing how the motility parameters were calculated.

Reviewer #1 (Recommendations for the authors):

1) Figure 3: The WT data in this figure don't include KO cells. How do the KOs compare to the WT in the scratch wound closure assay? The KO cells seem to be tested, but only in the absence of ADP (Figure 4 Sup 1A). Is there a difference in wound closure after ADP treatment? This point should be evaluated to assess the effects of altered ADP Ca signaling on non-directional cell movement in a physiologically relevant paradigm.

We thank the reviewer for this suggestion. We have addressed this under Essential revision comment no. 5.

2) Figure 4: This reviewer is not convinced that reduced turning accounts for the similar speed in the KO despite the larger mean square distance travelled. If turning is reduced and the mean square distance is doubled in the KO, then effectively, the average distance travelled over time (speed) should be greater in the KO than in the WT. The logic of this inference and how it explains the data (similar average speed but much greater root mean square displacement in the KO) needs to be better clarified. Since the mean square distance is so much larger, it suggests that the speed vectors with opposite directions are cancelling out in the KO cells to yield the same average velocity (which could theoretically even occur from increased, not decreased turning).

We thank the reviewer for bringing attention to this issue. We have addressed this under Essential revision comment no. 10.

3) Figure 4K: It seems that the baseline branch complexity is low to start with in the KOs, hence the fold change from baseline seems larger. From the images shown, there appears to be no difference following ADP treatment between the two groups. This data should be reanalyzed to compare differences in the absolute levels of branch complexity (# of branches and process length) in the two groups.

We thank the reviewer for bringing attention to this issue. We have addressed this under Essential revision comment no. 9.

4) The use of the term Ca clamp is misleading in the way it has been used here. Cellular Ca is, in fact, not clamped, but changes within each cell depending on the activity of pumps and other calcium clearance mechanisms. The motility measurements are simply done after extracellular Ca is restored to cells after thapsigargin treatment, at time points where the cytosolic Ca is lower in 0.2 mM Cacompared to 2 mM Ca. This phrasing (Ca clamp) needs to be changed to eliminate confusion that cellular Ca levels are "clamped" to defined values analogous to voltage clamp.

We agree with this comment, and have accordingly removed the term “Ca2+ clamp” from the manuscript (Essential revision comment no. 6).

5) Figure 6D: It is noteworthy that TREM2 KOs, but not WT cells, show differences in motility between 0.2 mM and 2 mM extracellular Ca even after thapsigargin treatment with no ADP exposure. This is strong suggestion that motility differences in TREM2 KO cells are due to more than just alterations in purinergic signaling. In fact, this inference is consistent by the lack of additional effects of adding ADP in Figure 6 Supplement 1. These intrinsic changes in cell motility in response to cellular Ca elevations need further clarification.

We have addressed this under Essential revision comment no. 7.

6) Figure 6D: Related to this figure, the data for WT cells should be directly compared to the KOs at each extracellular Ca concentration to dissect out differences in KO cells at each Ca condition.

We have addressed this under Essential revision comment no. 8.

Other:

– Stim1 is expressed at different levels between the WT and TREM2 KO. It has a significant p-value (Figure 1 Sup2D). This decrease could affect multiple aspects of microglia including cell motility in the presence of physiologically relevant agonists. This needs to be addressed.

We thank the reviewer for this comment. This point was made by other reviewers as well and needs better clarification. We have addressed this under Essential revision comment no. 2.

– A technical point related to Figure 4 is that figures 4C and 4G should be labelled "velocity" rather than speed, as it seems the only way in which the mean squared distance can be twice as large in the KO but not the velocity if the speed vectors with opposite directions are cancelling out in the KO.

These graphs (current Figure 4C-E, current Figure 5C, H) plot the mean track speed, which represents the mean of all instantaneous speeds over the total time of tracking. The instantaneous speed is calculated at each time point for a given track as the scalar equivalent to object velocity, without taking into account directionality (hence no negative values). We agree that MSD is lower in WTs; this is because WT cells turn more frequently (reflected as lower track straightness compared with TREM2 KOs), leading to greater cancellation of velocity vectors. For clarification, we now include definition of the different motility parameters in lines 922-937 of the methods section.

– The dose of ADP used in Figure 7 is extremely high. 100ng/mL (which is 234uM by my calculation). By contrast, Figure 4 used 2.5uM. The 100 fold higher concentration in the chemotaxis assay needs to be justified. Further, it should be determined whether the ADP used diffuses into both chambers might be expected. This super high dose of ADP may have non-physiological effects that could explain the seemingly differing phenotypes in Figures 4 and 7.

The reviewer’s calculation of concentration is incorrect. 100 ng/ml of ADP is equivalent to a concentration of 234 nM, not 234 µM. In the methods section on the chemotaxis assay (line 962), we now also report the concentration of ADP in nM. This dose of ADP is not super high.

– Figure 5 is a great set of experiments to show that the Salsa6f does not affect end points of the study, but may be better as a supplementary figure rather than a main figure. If the paper is about the TREM2 KO versus WT then the main figures should stick with looking at differences between the two with maybe a sub figure to show validation.

We thank the reviewer for this suggestion and are happy to comply. Because we ended up using the Salsa6f expressing lines in several additional experiments to confirm key findings in the study, the Salsa6f validation data are now presented in Figure 1 figure supplement 1.

Reviewer #2 (Recommendations for the authors):

– A more rigorous approach could be used to document the presence or absence of compensatory changes at the protein protein level of other Ca2+ signaling proteins (Figure 1 Suppl2). Preferably Westerns should be performed for STIM1 and STIM2, Orai1 and IP3R 1/2/3, for which there are reliable Abs. Although IF as performed for Orai1 is acceptable, the authors do not show positive and negative controls and a thorough validation of this Orai1Ab in IF.

We thank the reviewer for these comments and suggestions. We have addressed this under Essential revision cmment no. 1.

– Along the same lines as comment above, Staining for P2YR12 appears internal. Has the Ab been validated? Can authors perform a WB on P2YR12 and P2RY13 in addition?

We have addressed this under Essential revision comment no. 1.

– Did the authors check on whether the expression of SECRA is altered at the protein level? Did the authors perform CRAC channel recordings in Control and TREM2 KO cells?

We have not looked at the expression of SERCA at the protein level. However, we further examined our published RNAseq data on WT and TREM2 KO iPSC-microglia (PMID: 33097708) and compared the relative read counts of PMCA and SERCA isoforms that are known to be expressed in WT and TREM2 KO microglia and found no significant differences. Data for the relative expression of relevant Ca2+ signaling molecules (STIM1, Orai1, IP3R2, SERCA2 and 3, PMCA1) are now shown in Figure 2—figure supplement 1D and E. We note that among these, only P2Y12 and P2Y13 receptor transcripts showed significant fold change in expression between WT and TREM2 KO cells, which prompted us to focus our efforts in that direction. Additionally, we found that Ca2+ clearance rate after SOCE is similar in WT and TREM2 KO cells (Figure 3—figure supplement 2E, F). Based on this, we find it unlikely that the higher sustained Ca2+ level in TREM2 KO cells is due to differences in Ca2+ clearance mechanisms, including Ca2+ pump activity. We did not attempt to record Icrac in iPSC microglia.

– It is quite surprising that the authors did not opt to use the Ca2+ off/on protocol with ADP (Figure 1) as they have done with thapsigargin. In Figure 2B, ADP was applied in the absence of external Ca2+, but sadly those recordings did not subsequently replenish Ca2+ to gauge Ca2+ entry. One can rationalize the result of 2-APB (Figure 1 Suppl1), a non specific channel blocker that also inhibits IP3Rs. However, the fact that low concentrations of Gd3+ (which should not alter Ca2+ release) reduce the initial Ca2+ release peak, suggests that there is a contribution of Ca2+ entry to this initial response. Also, it would have been helpful to include in Figure 1 Suppl1 the control (non KO cells) in these experiments for reference.

We thank the reviewer for these comments and suggestions. We have accordingly performed new experiments and address this under Essential revision comment no. 3a.

Additional thoughts to the attention of the authors:

– The authors have speculated on the fact that increased cytosolic Ca2+ might lead to a "spill" outside the restricted Ca2+ nanodomain, thus disrupting polarity. Do the authors have any evidence of altered location of leading-edge vs trailing edge proteins or focal adhesion proteins in TREM2 KO cells?

No, we did not investigate the interesting question regarding the polarity of membrane proteins or focal adhesion proteins in TREM2 KO cells. Ca2+ signals play a role in generating cell polarity and regulating membrane protrusion and retraction in some migratory cell-types (PMID: 25977921). Local Ca2+ pulses have been detected at the leading edge of migrating fibroblasts in response to PDGF gradients (PMID: 19118385), and in migrating sheets of endothelial cells in response to a scratch wound (PMID: 24463606). More recently, localized Ca2+ signaling has been proposed to correlate with process extension in murine microglia based on intravital imaging, though it’s role in chemotaxis was not explored (PMID: 32716294). We have now rephrased the relevant sentence in the Discussion section (lines 360-363) to make it clear that we are speculating on the possible mechanism by which excessive Ca2+ signaling can impede gradient sensing. Reflecting the uncertainties, we have removed Figure 8 considering it may be overly speculative.

– Have the authors performed RNAseq on WT control and TREM2 KO cells? If so, does pathway analysis shows changes in the Cell motility genes between these two groups? This might offer some interesting clues to pursue in future studies.

The authors have performed and previously published RNA-seq on WT and TREM2 KO cells (Figure 1 in PMID: 33097708). Using gene ontology analysis, the authors did find differences in motility including “regulations of natural killer cell chemotaxis”, “Positive regulation of cell migration”, and “regulation of smooth muscle cell migration”. Even in the more selective gene ontology of genes which change in opposite directions after TREM2 KO and TREM2 antibody stimulation, we find “positive regulation of leukocyte chemotaxis” as one of the most significant gene family suggesting differences in motility. These gene lists included P2RY12 and P2RY13 receptors and were part of what stimulated our work presented here. However, some other chemotactic molecules including CCL2 and CCL3 are also included in this reciprocal gene list and warrant further study.

Reviewer #3 (Recommendations for the authors):

This study by Jairaman et al. describes how iPSC-derived microglia exhibit exaggerated cytosolic Ca2+ responses to ADP stimulation in TREM2KO cells, and that this leads to a defect in turning behaviour and hence no directed migration to a chemotactic signal.

Overall, the experiments are well conducted, carefully controlled and the findings are new and exciting. The authors nicely dissect out the underlying molecular basis for the exaggerated Ca2+ responses to ADP and then extend their findings to cell movement and directed migration. Given the substantial body of evidence linking microglia to the pathogenesis of Alzheimer's disease, and the role for TREM2, the work by Jairaman et al. is of translational significance. As an aside, the introduction of the calcium-sensitive reporter Salsa6F is a welcome new tool in the arsenal for recording cytosolic calcium. Overall, this is an elegant, novel and important study. Nevertheless, I have a few comments/suggestions.

We thank the reviewer for the positive and insightful comments. We have done new experiments to address specific suggestions as outlined below.

1. The authors argue that the increased Ca2+ plateau to ADP in TREM2KO cells is due to enhanced Ca2+ release from the stores ie greater store depletion. Evidence is presented that maximal SOCE is not compromised in TREM2KO cells, that store content under resting conditions is unaffected, InsP3R activities are similar. But the authors do not demonstrate that store content falls more following ADP receptor stimulation in the KO cells. This could be shown by directly measuring ER calcium to ADP in WT and KO cells or by applying ADP in Ca-free solution and then assessing store content with an ionomycin pulse. The ionomycin response should be smaller in the TREM2KO cells, after ADP exposure.

We thank the reviewer for suggesting the ionomycin pulse experiment. We have accordingly performed new experiments, and address this under Essential revision comment# 3b.

2. The authors suggest that an increase in P2Y12 and P2Y13 receptor expression in the KO cells accounts for the increased Ca2+ release from the ER. However, the increase in P2Y12 protein levels is modest at best, and the increase for P2y13 is less than 2-fold. The authors' point would be strengthened by showing InsP3 levels are increased in the KO cells compared with WT ones, for the same dose of ADP. Single cell InsP3 probes are available. Alternatively, a population measurement could be carried out.

While we agree that the suggested experiment with InsP3 probe would further bolster our conclusion, we have noted in essential revision comment no. 4 that transfection of iPSC-microglia shifts them from a resting to a highly activated state leading to downregulation of P2Y12 receptors, which confounds our ability to accurately compare P2Y receptor activity. Although the antibodies against P2Y12 and P2Y13 receptors were found to be nonspecific (see also essential revision comment No. 1), we now include flow cytometry data showing increased PM expression of P2Y12 receptor in live TREM2 KO cells (Figure 2G); these data complement the functional assays and transcriptomic data (Figure 2—figure supplement 1E) showing >2 fold increase in P2Y12 and P2Y13 receptor RNA expression in TREM2 KO microglia. These differences are likely to be biologically relevant for downstream signaling, given the positive feedback mechanisms and signal amplification associated with GPCR signaling (PMID: 29074251). Increased protein expression of P2Y12 and P2Y13 receptors has also been reported in Trem2-/- mice and in a mouse AD model, and we now include this point in the Introduction section of the manuscript (line 87-89).

3. The evidence for an involvement of SOCE is based on two rather non-specific inhibitors (2-APB and Gd3+). P2YRs, like other GPCRs, activate TRPC proteins and these Ca2+-permeable channels could contribute to the Ca2+ plateau. The authors should therefore knock down Orai1 or at least use more selective inhibitors such as Synta66 or the GSK compound.

We thank the reviewer for this suggestion and have performed additional pharmacological experiments and CRISPRbased Orai1 KO to address this point in Figure 3 and Figure 3—figure supplement 1. The results confirm involvement of Orai channels in SOCE activated by store-depletion with TG and ADP.

4. The authors should include some controls to show the P2Y antibodies are indeed specific. For example, they could use a cell line that lacks P2Y12/13. In Figure 2I, J, what happens to the ADP response when both P2Y inhibitors are present at the same time? It would be important to show that the combination of P2Y12 and P2Y13 suppress ADP responses fully.

Per reviewer suggestion, we examined Ca2+ responses to ADP in the presence of both P2RY12 and P2Y13 inhibitors. This combination inhibited the ADP Ca2+ response completely in both WT and TREM2 KO microglia (Figure 2G).

5. The authors should add a justification for the concentrations of agonists they have used (ADP, ATP, UTP). A dose-response curve is included for ADP but a comment on the doses selected would be helpful.

Extracellular concentration of purinergic signals range from hundreds of nanomolar to μM levels, and are shaped by a variety of factors including baseline secretion, extent of local tissue damage and the prevalence of ectonucleotidases that cleave purinergic ligands (PMID: 18302942). Studies in microglial field have often used tens of μM ADP (PMID: 17115040, 11245682) to study the biology of P2Y12 receptors, but microglial cells experience a range of concentrations depending on the pathophysiological context, and depending on distance from the injury site. We now include a sentence in the Introduction (line 54) about purinergic concentrations in the brain.

6. It is nicely shown that Ca2+ influx in TREM2 KO cells leads to microglia motility and process extension to a greater extent than WT cells. This is attributed to the increased SOCE. If so, then one might expect raising external Ca2+ in WT cells to have the same effect. Is this the case?

We have not examined effects of increasing the extracellular Ca2+ concentration in WTs to test whether that results in a higher magnitude of cell speed and process extension (compared to TREM2 KOs). Our data in Figure 5 and Figure 5 figure supplement 1 shows that Ca2+ influx pathways are required for optimal cell motility and process dynamics in iPSC microglia in general (based on greater increases in cell track speeds, process length and branching in response to ADP in 1 mM Ca2+ compared with 0 mM Ca2+ extracellular solution for both WT and TREM2 KO microglia). We should also note that the effects of TREM2 deletion on process branching/ lengthening is not as drastic as its effects on cell motility. We have addressed this particular point in response to comments from reviewer No. 1 under Essential revision, comment no. 9.

7. Is anything known mechanistically how the exaggerated Ca2+ signal leads to a chemotaxis defect? The authors' data would suggest that chemotaxis might have a bell-shaped dependence on cytosolic Ca2+, with too little or too much impeding migration towards a cue.

The fundamental issue in chemotaxis is how shallow gradients of a chemokine translates to steep polarization of signaling proteins within the cell. This is generally thought to occur downstream of the chemokine receptor and upstream of actin cytoskeleton (PMID: 33990789), but the specific role of Ca2+ signals in this process remains poorly understood. Please also see our earlier response (comment No. 5 by reviewer# 2) on how excessive cytosolic Ca2+ may disrupt cell polarity. The histogram in Figure 6G comparing the percent of fast-moving cells as a function of cytosolic Ca2+ further indicate that there may indeed be a bell-shaped effect with regard to effects of Ca2+ on instantaneous cell speeds in TREM2 KO cells. This is reflected in the chemotaxis assay in Figure 7B and C in which reducing ADP signaling in TREM2 KO cells rescues the defect in chemotaxis. It is possible that high cytosolic Ca2+ serves as a temporary STOP signal in microglia similar to its effects on T cells. We note this point in the Discussion (lines 386-388) and speculate that TREM2 KO cells may be more subject to this effect with ADP, given the higher expression of P2RY12 and P2Y13 receptors.

8. Do the authors know whether the effects of SOCE are mediated through a local Ca2+ rise or via a global cytosolic Ca elevation?

We have not looked at role of local vs. global Ca2+ signals in regulating microglial cell motility in the current study but hope to explore this aspect of Ca2+ signaling as part of a follow up study.

9. Perhaps I missed something but Figure 1 S2(D) seems to show a significant decrease in STIM1 levels in the KO cells.

We thank the reviewer for this comment. This point was made by other reviewers as well and needs clarification. We have addressed this under Essential revision comment no. 2.

10. The authors present SOCE as the peak signal remaining after 5 minutes. It is not clear whether the value has been subtracted from the pre-stimulation levels. In some graphs, the signal after 5 minutes looks the same as the resting level, but the bar charts show higher values.

We assume the reviewer is referring to the traces and bar-graphs in current Figure 1B, D, E and F showing both peak Ca2+ response and Ca2+ at 5 minutes. The data-points (single cell values) in the bar-graphs are all baseline subtracted, as described in the legend. The Y-axis is now labelled “Fluo-4/Fura-Red Ratio (baseline subtracted)” to make this point clear. We also note that the bar-graph summary includes data from multiple experiments, while the average traces are from a single imaging run.

https://doi.org/10.7554/eLife.73021.sa2

Article and author information

Author details

  1. Amit Jairaman

    Department of Physiology and Biophysics, University of California, Irvine, Irvine, United States
    Contribution
    Conceptualization, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing – original draft, Writing – review and editing
    Contributed equally with
    Amanda McQuade
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5206-700X
  2. Amanda McQuade

    1. Department of Neurobiology and Behavior, University of California, Irvine, Irvine, United States
    2. Sue and Bill Gross Stem Cell Research Center, University of California, Irvine, Irvine, United States
    3. UCI Institute for Memory Impairments and Neurological Disorders, University of California, Irvine, United States
    4. Institute for Neurodegenerative Diseases, University of California, San Francisco, San Francisco, United States
    Contribution
    Conceptualization, Formal analysis, Funding acquisition, Investigation, Methodology, Validation, Visualization, Writing – original draft, Writing – review and editing, This author is a co-first author
    Contributed equally with
    Amit Jairaman
    Competing interests
    No competing interests declared
    Additional information
    This author is a co-first author
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5368-6788
  3. Alberto Granzotto

    1. Department of Neurobiology and Behavior, University of California, Irvine, Irvine, United States
    2. Center for Advanced Sciences and Technology (CAST), University "G. d'Annunzio" of Chieti-Pescara, Chieti, Italy
    3. Department of Neuroscience, Imaging and Clinical Sciences, University G d'Annunzio of Chieti-Pescara, Chieti, Italy
    Contribution
    Formal analysis, Investigation, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
  4. You Jung Kang

    Department of Mechanical Engineering and Engineering Science, University of North Carolina, Charlotte, United States
    Contribution
    Formal analysis, Investigation, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
  5. Jean Paul Chadarevian

    Department of Neurobiology and Behavior, University of California, Irvine, Irvine, United States
    Contribution
    Investigation, Methodology
    Competing interests
    No competing interests declared
  6. Sunil Gandhi

    Department of Neurobiology and Behavior, University of California, Irvine, Irvine, United States
    Contribution
    Funding acquisition, Resources, Supervision, Writing – review and editing
    Competing interests
    is a co-founders of NovoGlia Inc
  7. Ian Parker

    1. Department of Physiology and Biophysics, University of California, Irvine, Irvine, United States
    2. Department of Neurobiology and Behavior, University of California, Irvine, Irvine, United States
    Contribution
    Methodology, Resources, Writing – review and editing
    Competing interests
    No competing interests declared
  8. Ian Smith

    Department of Neurobiology and Behavior, University of California, Irvine, Irvine, United States
    Contribution
    Formal analysis, Investigation, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-9910-195X
  9. Hansang Cho

    Institute of Quantum Biophysics, Department of Biophysics, Dept of Intelligent Precision Healthcare Convergence, Sungkyunkwan University, Gyeonggi-do, Republic of Korea
    Contribution
    Formal analysis, Funding acquisition, Investigation, Methodology, Resources, Supervision, Writing – review and editing
    Competing interests
    No competing interests declared
  10. Stefano L Sensi

    1. Center for Advanced Sciences and Technology (CAST), University "G. d'Annunzio" of Chieti-Pescara, Chieti, Italy
    2. Department of Neuroscience, Imaging and Clinical Sciences, University G d'Annunzio of Chieti-Pescara, Chieti, Italy
    Contribution
    Funding acquisition, Methodology, Resources, Supervision
    Competing interests
    No competing interests declared
  11. Shivashankar Othy

    1. Department of Physiology and Biophysics, University of California, Irvine, Irvine, United States
    2. Institute for Immunology, University of California, Irvine, Irvine, United States
    Contribution
    Conceptualization, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Writing – original draft, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-6832-5547
  12. Mathew Blurton-Jones

    1. Department of Neurobiology and Behavior, University of California, Irvine, Irvine, United States
    2. Sue and Bill Gross Stem Cell Research Center, University of California, Irvine, Irvine, United States
    3. UCI Institute for Memory Impairments and Neurological Disorders, University of California, Irvine, United States
    4. Institute for Immunology, University of California, Irvine, Irvine, United States
    Contribution
    Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Writing – original draft, Writing – review and editing
    For correspondence
    mblurton@uci.edu
    Competing interests
    is a co-inventor of patent application WO/2018/160496, related to the differentiation of pluripotent stem cells into microglia. Is a co-founders of NovoGlia Inc
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-7770-7157
  13. Michael D Cahalan

    1. Department of Physiology and Biophysics, University of California, Irvine, Irvine, United States
    2. Institute for Immunology, University of California, Irvine, Irvine, United States
    Contribution
    Conceptualization, Funding acquisition, Methodology, Project administration, Resources, Supervision, Writing – original draft, Writing – review and editing
    For correspondence
    mcahalan@uci.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4987-2526

Funding

National Institutes of Health (R01 NS14609)

  • Michael D Cahalan

National Institutes of Health (R01 AI121945)

  • Michael D Cahalan

National Institutes of Health (R01 AG048099)

  • Mathew Blurton-Jones

National Institutes of Health (R01 AG056303)

  • Mathew Blurton-Jones

National Institutes of Health (R01 AG055524)

  • Mathew Blurton-Jones

National Institutes of Health (core AG066519)

  • Mathew Blurton-Jones

National Institutes of Health (U01 AI160397)

  • Shivashankar Othy

National Institutes of Health (T32 NS082174)

  • Amanda McQuade

National Institutes of Health (RF1DA048813)

  • Sunil Gandhi

The Marie Sklodowska-Curie grant agreement iMIND (no. 84166)

  • Alberto Granzotto

National Research Foundation (2020R1A2C2010285)

  • Hansang Cho

National Research Foundation (2020 M3C7A1023941)

  • Hansang Cho

National Research Foundation (I21SS7606036)

  • Hansang Cho

National Institute of Health (AG059236-01A1)

  • Hansang Cho

UCI Sue & Bill Gross Stem Cell Research Center (Seed Grant)

  • Sunil Gandhi

Susan Scott Foundation (gift)

  • Mathew Blurton-Jones

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Dr. Andy Yeromin for the development of Excel macros to analyze Imaris cell tracking. We would also like to thank Morgan Coburn for sharing Python scripts that aided in the organization of Imaris output files. This work was supported by T32 NS082174 and ARCS foundation (AM); the European Union’s Horizon 2020 research and innovation program under the Marie Sklodowska-Curie grant agreement iMIND – no. 84166 (AG); NIH R01 NS14609 and AI121945 (MDC); NIH U01 AI160397 (SO); NRF 2020R1A2C2010285, 2020 M3C7A1023941, and NIH AG059236-01A1 (HC); NIH AG048099, AG056303, and AG055524 (MBJ); RF1DA048813 (MBJ and SG); UCI Sue & Bill Gross Stem Cell Research Center Seed Grant (SG); and a generous gift from the Susan Scott Foundation (MBJ). iPSC lines were generated by the UCI-ADRC iPS cell core funded by NIH AG066519. Experiments using the GFP-expressing iPSC line AICS-0036 were made possible through the Allen Cell Collection, available from the Coriell Institute for Medical Research.

Ethics

Human subjects: Human iPSC lines were generated by the University of California Alzheimer's Disease Research Center (UCI ADRC) stem cell core. Subject fibroblasts were collected under approved Institutional Review Boards (IRB) and human Stem Cell Research Oversight (hSCRO) committee protocols. Informed consent was received for all participants.

Senior Editor

  1. Richard W Aldrich, The University of Texas at Austin, United States

Reviewing Editor

  1. Murali Prakriya, Northwestern University, United States

Reviewer

  1. Mohamed Trebak, University of Pittsburgh, United States

Publication history

  1. Received: August 13, 2021
  2. Preprint posted: August 24, 2021 (view preprint)
  3. Accepted: February 18, 2022
  4. Accepted Manuscript published: February 22, 2022 (version 1)
  5. Version of Record published: March 9, 2022 (version 2)
  6. Version of Record updated: March 22, 2022 (version 3)

Copyright

© 2022, Jairaman et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,056
    Page views
  • 370
    Downloads
  • 2
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Amit Jairaman
  2. Amanda McQuade
  3. Alberto Granzotto
  4. You Jung Kang
  5. Jean Paul Chadarevian
  6. Sunil Gandhi
  7. Ian Parker
  8. Ian Smith
  9. Hansang Cho
  10. Stefano L Sensi
  11. Shivashankar Othy
  12. Mathew Blurton-Jones
  13. Michael D Cahalan
(2022)
TREM2 regulates purinergic receptor-mediated calcium signaling and motility in human iPSC-derived microglia
eLife 11:e73021.
https://doi.org/10.7554/eLife.73021

Further reading

    1. Immunology and Inflammation
    2. Structural Biology and Molecular Biophysics
    Nathanael A Caveney et al.
    Short Report

    Interleukin 27 (IL-27) is a heterodimeric cytokine that functions to constrain T cell-mediated inflammation and plays an important role in immune homeostasis. Binding of IL-27 to cell surface receptors IL-27Rα and gp130 results in activation of receptor-associated Janus Kinases and nuclear translocation of Signal Transducer and Activator of Transcription 1 (STAT1) and STAT3 transcription factors. Despite the emerging therapeutic importance of this cytokine axis in cancer and autoimmunity, a molecular blueprint of the IL-27 receptor signaling complex, and its relation to other gp130/IL-12 family cytokines, is currently unclear. We used cryogenic-electron microscopy to determine the quaternary structure of IL-27, composed of p28 and Ebi3 subunits, bound to receptors, IL-27Rα and gp130. The resulting 3.47 Å resolution structure revealed a three-site assembly mechanism nucleated by the central p28 subunit of the cytokine. The overall topology and molecular details of this binding are reminiscent of IL-6 but distinct from related heterodimeric cytokines IL-12 and IL-23. These results indicate distinct receptor assembly mechanisms used by heterodimeric cytokines with important consequences for targeted agonism and antagonism of IL-27 signaling.

    1. Developmental Biology
    2. Immunology and Inflammation
    David J Turner et al.
    Short Report Updated

    To identify roles of RNA binding proteins (RBPs) in the differentiation or survival of antibody secreting plasma cells we performed a CRISPR/Cas9 knockout screen of 1213 mouse RBPs for their ability to affect proliferation and/or survival, and the abundance of differentiated CD138 + cells in vitro. We validated the binding partners CSDE1 and STRAP as well as the m6A binding protein YTHDF2 as promoting the accumulation of CD138 + cells in vitro. We validated the EIF3 subunits EIF3K and EIF3L and components of the CCR4-NOT complex as inhibitors of CD138 + cell accumulation in vitro. In chimeric mouse models YTHDF2-deficient plasma cells failed to accumulate.