Microglia are parenchymal innate immune cells and vital remodelers of the central nervous system (CNS) (Prinz et al., 2021). They have a myriad of critical functions during development, including elimination of viable or dying cells (Anderson and Vetter, 2019c; Anderson et al., 2019b). Similar to other macrophages, microglia are extremely attuned to changes in their neural niche (Lavin et al., 2014; Gosselin et al., 2017; Bennett et al., 2018) and rapidly respond by phagocytosis or secreting soluble factors. Recent single-cell RNA sequencing (scRNAseq) has demonstrated that microglia are particularly diverse during development (Hammond et al., 2019; Li et al., 2019a). However, we still lack an understanding of the drivers of this heterogeneity, including the impact of environmental cues or the act of phagocytosis itself. Therefore, connecting microglial states to specific developmental events and determining the pathways involved remains a central challenge.

One developmental process fundamental for nearly every organ in the body is apoptotic death of excess or dysfunctional cells (Jacobson et al., 1997). In the CNS, death of neurons and glia is an essential and well-documented feature of development (Oppenheim, 1991). Efferocytosis, or the phagocytosis of apoptotic cells, is a rapid and carefully orchestrated process designed to minimize damage to surrounding cells (Morioka et al., 2019; Boada-Romero et al., 2020). Largely mediated by microglia, efferocytosis is important for maintaining CNS homeostasis not only in development but in aging and disease as well (Galloway et al., 2019; VanRyzin, 2021). The developing mouse retina is an excellent CNS model system to link remodeling events such as efferocytosis and microglial state and function (Li et al., 2019b). During the postnatal period in the retina, as circuits are being established, waves of neuronal death occur (Braunger et al., 2014). For example, retinal ganglion cells (RGCs), the projection neurons connecting retina and brain, undergo a well-characterized culling during the first postnatal week which is dependent on pro-apoptotic factor, Bcl-2-associated X protein (Bax) (Péquignot et al., 2003). An estimated ~50% of RGCs will undergo apoptosis and need to be cleared (Farah and Easter, 2005; Farah, 2006). We previously found that neuronal death significantly influences microglia properties in postnatal retina, resulting in a majority of microglia expressing a distinct gene signature and showing altered dependence on CSF1R signaling for survival (Anderson et al., 2019a). Relative to adult microglia, homeostatic genes were reduced, while genes associated with phagocytosis and lipid metabolism were increased, similar to aging and disease-associated microglia (DAM) (Holtman et al., 2015; Keren-Shaul et al., 2017; Krasemann et al., 2017), and to developmental microglia residing in postnatal white matter tracts (PAM) (Li et al., 2019a) or (ATM) (Hammond et al., 2019) or CD11c microglia (Benmamar-Badel et al., 2020).

Consistent with either apoptotic cell recognition or phagocytosis playing a central role in driving this gene signature, microglia in disease states increase select DAM genes in response to apoptotic cells (Krasemann et al., 2017), and ATM/PAM populations in the white matter tract engulf oligodendrocyte precursor cells (Li et al., 2019a; Nemes-Baran et al., 2020). In disease, the TREM2 receptor is partially required for acquisition of the DAM signature (Keren-Shaul et al., 2017; Krasemann et al., 2017), but not for the developmental signature in retina (Anderson et al., 2019a) or brain (Li et al., 2019a). Thus, it remains unclear to what extent the process of phagocytosis is altering properties of developmental microglia or whether distinct recognition pathways are involved. Microglia, like other macrophages, can recognize and engulf apoptotic (or viable) cells via a variety of ligand-receptor systems including C1q/C3 to complement receptor 3 (CR3) and exposed phosphatidylserine (PtdSer) to TAM family of tyrosine kinase receptors Mer (gene Mertk) and Axl (Lemke, 2019), but how these pathways mediate changes in microglia properties and responses is an active area of research.

Here, we ask how neuronal apoptosis drives key properties of developmental microglia such as CSF1R dependence and expression of DAM/ATM/PAM-related genes, and test the role of phagocytosis and specific recognition pathways. Through scRNAseq on postnatal retinal microglia, we find considerable transcriptional heterogeneity, with multiple distinct microglial populations expressing DAM/ATM/PAM-related, lysosomal, and lipid metabolism genes. We further show, by analyzing microglia from Bax knockout (KO) retinas, that exposure to dying neurons drives several of these states and that most are more resistant to CSF1R inhibition. We find that CR3 and Mer are required for clearance of dying RGCs, but not for altering dependence on CSF1R signaling for survival. Conversely, Axl is not required for clearance of dying RGCs, but for augmenting microglial survival in the absence of CSF1R signaling. Loss of Mer did not have a widespread impact on expression of microglial remodeling genes, and neither did loss of Axl or Mertk/Axl double KO (dKO), suggesting that TAM receptor-mediated signaling or reduced clearance are not sufficient. Thus, we find that interactions with apoptotic neurons drives developmental microglial diversity, and that distinct recognition receptors mediate phagocytosis of dying cells versus specific microglial properties, including microglial survival.

scRNAseq has revealed that microglia in development and disease are transcriptionally diverse across the CNS (Masuda et al., 2020). Due to their remarkable plasticity, a major challenge is to determine the impact of local cues on microglial state and properties. Here, using the postnatal retina as a model system, we establish neuronal cell death as a key factor driving developmental microglia diversity. We show by scRNAseq that multiple microglial states coexist within this discrete CNS region, encompassing a spectrum from homeostatic to remodeling states with high lysosomal and lipid metabolism gene expression. We establish that multiple distinct remodeling states are driven by neuronal apoptosis, and further show that several of these remodeling states survive CSF1R inhibition, while more homeostatic microglia are susceptible. CR3 and Mer are important for clearance of apoptotic retinal neurons, while Axl facilitates microglial survival following inhibition CSF1R, but neither significantly regulates expression of microglial remodeling genes. Thus, we conclude that cell death is a critical cue for driving diverse microglial properties and that multiple pathways contribute.

We identify several states of remodeling microglia characterized by reduced homeostatic gene expression and elevated lipid metabolism and lysosomal genes, with more discrete expression of various genes including chemokines/cytokines. Many of the genes in these remodeling states are shared with microglia in developing white matter tracts of the brain (CD11c/ATM/PAM) (Wlodarczyk et al., 2017; Hammond et al., 2019; Li et al., 2019a) and disease (DAM) (Holtman et al., 2015; Keren-Shaul et al., 2017; Krasemann et al., 2017). Unlike in brain, we find they represent a majority of microglia in the early postnatal retina, with most of them dependent upon neuronal apoptosis. Due to the large proportion of microglia expressing these remodeling genes, we broaden our understanding of microglia subsets present during normal development. For example, we find genes such as Apoe and Ctsd are widely expressed at varying levels, while Fabp5 and Spp1 are confined to more discrete subsets, suggesting that regulation of these genes and microglial states is multifaceted.

While our previous work found that microglia expressing several of these remodeling genes were dynamically regulated over development, future work analyzing the localization of these populations would further inform understanding of function and population dynamics. At present, it remains unclear whether they transition from state to state. Since we find an increase in the proportion of homeostatic microglia in Bax KO retinas, this argues that microglia in the context of neuronal death shift from a homeostatic state to more remodeling ones. We found that loss of Bax had an effect on microglial density, consistent with zebrafish studies (Casano et al., 2016), suggesting that neuronal apoptosis is important for regulating other microglial properties. Whether this is due to a reduction in proliferation is unknown, however we did not see a change in the proportion of cycling cells by scRNAseq.

In other contexts, efferocytosis can have a major impact on the transcriptional profile and state of the phagocyte. For example, phagocytosis promotes distinct transcriptional states in macrophages in a tissue-specific manner throughout the body (A-Gonzalez et al., 2017), and tissue macrophages modify their repertoire of phagocytic machinery in response to phagocytosis (Zent and Elliott, 2017). In the CNS, there is much to learn about how phagocytosis impacts microglial state and function (Márquez-Ropero et al., 2020; VanRyzin, 2021). One study addressing this found engulfment of newborn neurons by microglia in adult neurogenic zones stimulates a microglial secretome that subsequently regulates neurogenesis (Diaz-Aparicio et al., 2020). Here, in the postnatal retina, phagocytosis and digestion appear to be a major aspect of microglial heterogeneity. We postulate that some remodeling states may even represent distinct stages of the clearance and digestion process. Our analysis also identified microglial states that did not change with loss of Bax. While perhaps they represent microglia that have not had the opportunity to interact with dying neurons, they may be involved in other important microglial functions during postnatal retinal development. For example, astrocytes are also pruned by microglia during this time period, which has been shown to be Bax-independent and also not require CR3, Mer, or CX3CR1 pathways (Puñal et al., 2019). Future studies could target these specific microglia subsets and try to link them to other important remodeling functions in the developing retina. Some clusters did have higher expression of ex vivo activation genes (Marsh et al., 2020) but the impact on developmental datasets has not been directly tested. exAM genes did not drive the formation of a single cluster and some genes were differentially expressed between samples, arguing against solely a technical effect. Further, we found exAM genes Ccl3 and Ccl4 were expressed in vivo, suggesting biological relevance (Figure 1—figure supplement 5 and data not shown).

Despite the fact that postnatal RGC death has been a well-characterized model of developmental cell death of the CNS (Péquignot et al., 2003), the recognition pathways required for the clearance of apoptotic RGCs was previously undefined. CR3 and Mer are well-known recognition receptors in the clearance of stressed or dying neurons in various contexts (Lemke, 2019). Here, we show that both receptors are important for clearance of apoptotic RGCs. Previous work has suggested that complement component C1q may get deposited on exposed PtdSer (Païdassi et al., 2008), suggesting the pathways might work together to signal for phagocytosis, but whether complement and TAM pathways are working redundantly or cooperatively is unknown in any context. It is also unclear whether other mechanisms can compensate for the loss of either receptor. We predict this is likely, since microglia express a broad array of phagocytic receptors, and other cell types such as astrocytes can phagocytose, particularly in contexts of deficient microglia (Puñal et al., 2019). Furthermore, since Mer and Axl are expressed in other cell types such as astrocytes (Fourgeaud et al., 2016), we cannot exclude the possibility of indirect effects. While receptors Mer and Axl are both members of the TAM receptor family of tyrosine kinases (Lemke, 2013), it has been appreciated that Mertk is more stably expressed in microglia, while Axl is more dynamic and upregulated with inflammation or disease (Fourgeaud et al., 2016). Both have previously been implicated in phagocytosis and in regulating inflammatory responses (Elliott et al., 2017), however this is cell type-dependent (Seitz et al., 2007). Here, we find Mertk/Axl dKO did not have an enhanced clearance deficit for apoptotic RGCs compared to Mertk KO, although this could be due to a ceiling effect since RBPMS likely gets downregulated as apoptotic programs continue (Elliott and Ravichandran, 2016).

Although Mer is important for clearance, we did not see a dramatic reduction in lysosomal or lipid metabolism genes in either Mertk or Axl KO microglia. These results are consistent with disease contexts where loss of Mertk and Axl did not have a profound effect on the DAM signature in mouse model of Alzheimer’s disease (Huang et al., 2021). This suggests that transcriptional changes in postnatal retinal microglia may depend upon other pathways, although Mer and Axl could still have essential roles in post-transcriptional regulation of microglial states. Additionally, Mertk KO microglia are likely still clearing, but at a reduced rate, since live-imaging has shown that loss of Mer results in delayed engagement with apoptotic cells but does not abolish phagocytosis completely (Damisah et al., 2020).

While in adulthood, retinal microglia can be ablated by blocking CSF1R signaling (MacDonald et al., 2010; Anderson et al., 2019a), we previously found that during the early postnatal period a large proportion of microglia expressing select remodeling genes are more resistant to inhibition or loss of this important survival pathway (Anderson et al., 2019a). One hypothesis is that specific microglial states, which are transient, have altered CSF1R dependence. Here, we identify specific remodeling states of microglia that persist following CSF1R inhibition showing reduced homeostatic gene expression. Other groups recently found small populations of CSF1R-independent or repopulating microglia in the brain, and we note similar patterns of gene expression, with reduced homeostatic genes such as Tmem119 and P2ry12, and increased expression of genes such as Lyz2 (Zhan et al., 2020; Hohsfield et al., 2021). In these microglia repopulation studies, it was proposed that more immature microglia survive CSF1R inhibition, but microglia maturation may not be a key factor in the context of development since we previously found greater dependence of retinal microglia on CSF1R at embryonic stages (Anderson et al., 2019b) than in postnatal retina (Anderson et al., 2019a). We find that most of these resistant microglia subsets are induced by neuronal apoptosis, consistent with our prior analysis showing increased microglial dependence on CSF1R in Bax KO retina (Anderson et al., 2019a). Interestingly, in culture, peritoneal and bone marrow macrophages that phagocytose apoptotic cells also survive in the absence of serum or survival factors (Reddy et al., 2002). In periods of elevated cell death and possible lack of extrinsic survival factors, such as neuronally expressed CSF1R ligand, IL-34 (Nandi et al., 2012), phagocytes such as microglia might have evolved strategies to survive in order to maintain homeostasis or reduce inflammation (Reddy et al., 2002). Importantly, cholesterol, along with CSF-1/IL-34 and TGF-ß2, has been shown to be an important survival factor for microglia in culture (Bohlen et al., 2017). However, other possibilities exist for the overlap between clusters dependent on neuronal apoptosis (0,3,4,5) and clusters more resistant to CSF1R inhibition (3,4,5,6,10). For example, PLX-treatment itself could be inducing transcriptional changes reminiscent of remodeling states. We did see an expansion of PLX-enriched (3) microglia with the lowest levels of homeostatic gene expression and high levels of lysosomal and lipid metabolism genes. Thus, CSF1R signaling may also be important for promoting microglial homeostasis (Hu et al., 2021). Consistent with this, mutations in CSF1R can lead to leukoencephalopathy, a neurodegenerative disorder that is associated with loss of homeostatic microglia phenotype (Kempthorne et al., 2020). PLX-enriched (3) microglia may also represent cells that are undergoing other changes, even progressing toward death. It will be valuable for future studies to explore the contributions of CSF1R signaling to microglial survival and homeostasis during development.

Here, we find that Axl expression, which is regulated by neuronal apoptosis, enhances microglial survival in the absence of CSF1R signaling. Since loss of Axl did not substantially alter expression of other candidate apoptotic pathways in microglia, one possibility is that signaling directly downstream of Axl is involved in promoting survival. Axl has been shown to inhibit apoptosis in a variety of cell types (Axelrod and Pienta, 2014) including fibroblasts (Goruppi et al., 1997), oligodendrocytes in contexts of growth factor withdrawal or TNF toxicity (Shankar et al., 2006), and in Gonadotropin-releasing hormone neurons in the brain (Pierce et al., 2008). Its role in various types of cancer has been well established for promoting tumor growth and metastasis in part due to its ability to inhibit apoptosis of cancer cells (Zhu et al., 2019). Axl has diverse downstream signaling cascades depending on context (Axelrod and Pienta, 2014), but one candidate signaling cascade is the PI3K-Akt-NfkB-Bcl2 pathway. Activation of PI3K and Akt has shown to be important for survival of cultured macrophages (Reddy et al., 2002), oligodendrocytes (Shankar et al., 2003; Shankar et al., 2006), neurons (Pierce et al., 2008) as well as cancer cells (Axelrod and Pienta, 2014). Another candidate pathway for regulating microglial survival is Spp1/osteopontin since we found it was significantly downregulated in Axl KO microglia. Spp1/osteopontin has been shown to promote survival of other cell types including peripheral immune cells (Denhardt et al., 2001) and cancer cells (Saleh et al., 2016). Furthermore, Spp1/osteopontin can influence aspects of microglial properties including enhancing survival under conditions of stress in culture (Yu et al., 2017).

Altogether, we show neuronal cell death is a key factor driving multiple states of microglia in developing retina, with distinct transcriptional profiles and altered dependence from CSF1R signaling for survival. Apoptotic cell recognition by microglia is thus a critical developmental event driving diverse effects on gene expression, clearance, and survival, with distinct pathways involved in mediating each of these responses.

Sequencing data have been deposited in GEO under the reference series GSE192602. https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE192602 Data generated or analyzed during this study are included in the manuscript.

1. 1. Vetter ML

   (2022) NCBI Gene Expression Omnibus

   ID GSE192602. Neuronal apoptosis drives remodeling states of microglia and shifts in survival pathway dependence.

   https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE192602

1. 1. A-Gonzalez N
   2. Quintana JA
   3. García-Silva S
   4. Mazariegos M
   5. González de la Aleja A
   6. Nicolás-Ávila JA
   7. Walter W
   8. Adrover JM
   9. Crainiciuc G
   10. Kuchroo VK
   11. Rothlin CV
   12. Peinado H
   13. Castrillo A
   14. Ricote M
   15. Hidalgo A

   (2017) Phagocytosis imprints heterogeneity in tissue-resident macrophages

   The Journal of Experimental Medicine 214:1281–1296.

   https://doi.org/10.1084/jem.20161375

   • PubMed
   • Google Scholar
2. 1. Anderson SR
   2. Roberts JM
   3. Zhang J
   4. Steele MR
   5. Romero CO
   6. Bosco A
   7. Vetter ML

   (2019a) Developmental Apoptosis Promotes a Disease-Related Gene Signature and Independence from CSF1R Signaling in Retinal Microglia

   Cell Reports 27:2002–2013.

   https://doi.org/10.1016/j.celrep.2019.04.062

   • PubMed
   • Google Scholar
3. 1. Anderson SR
   2. Zhang J
   3. Steele MR
   4. Romero CO
   5. Kautzman AG
   6. Schafer DP
   7. Vetter ML

   (2019b) Complement Targets Newborn Retinal Ganglion Cells for Phagocytic Elimination by Microglia

   The Journal of Neuroscience 39:2025–2040.

   https://doi.org/10.1523/JNEUROSCI.1854-18.2018

   • PubMed
   • Google Scholar
4. 1. Anderson SR
   2. Vetter ML

   (2019c) Developmental roles of microglia: A window into mechanisms of disease

   Developmental Dynamics 248:98–117.

   https://doi.org/10.1002/dvdy.1

   • PubMed
   • Google Scholar
5. 1. Axelrod H
   2. Pienta KJ

   (2014) Axl as a mediator of cellular growth and survival

   Oncotarget 5:8818–8852.

   https://doi.org/10.18632/oncotarget.2422

   • PubMed
   • Google Scholar
6. 1. Benmamar-Badel A
   2. Owens T
   3. Wlodarczyk A

   (2020) Protective Microglial Subset in Development, Aging, and Disease: Lessons From Transcriptomic Studies

   Frontiers in Immunology 11:430.

   https://doi.org/10.3389/fimmu.2020.00430

   • PubMed
   • Google Scholar
7. 1. Bennett FC
   2. Bennett ML
   3. Yaqoob F
   4. Mulinyawe SB
   5. Grant GA
   6. Hayden Gephart M
   7. Plowey ED
   8. Barres BA

   (2018) A Combination of Ontogeny and CNS Environment Establishes Microglial Identity

   Neuron 98:1170–1183.

   https://doi.org/10.1016/j.neuron.2018.05.014

   • PubMed
   • Google Scholar
8. 1. Boada-Romero E
   2. Martinez J
   3. Heckmann BL
   4. Green DR

   (2020) The clearance of dead cells by efferocytosis

   Nature Reviews. Molecular Cell Biology 21:398–414.

   https://doi.org/10.1038/s41580-020-0232-1

   • PubMed
   • Google Scholar
9. 1. Bohlen CJ
   2. Bennett FC
   3. Tucker AF
   4. Collins HY
   5. Mulinyawe SB
   6. Barres BA

   (2017) Diverse Requirements for Microglial Survival, Specification, and Function Revealed by Defined-Medium Cultures

   Neuron 94:759–773.

   https://doi.org/10.1016/j.neuron.2017.04.043

   • PubMed
   • Google Scholar
10. 1. Braunger BM
    2. Demmer C
    3. Tamm ER

    (2014) Programmed cell death during retinal development of the mouse eye

    Advances in Experimental Medicine and Biology 801:9–13.

    https://doi.org/10.1007/978-1-4614-3209-8_2

    • PubMed
    • Google Scholar
11. 1. Bult CJ
    2. Blake JA
    3. Smith CL
    4. Kadin JA
    5. Richardson JE
    6. Mouse Genome Database Group

    (2019) Mouse Genome Database (MGD) 2019

    Nucleic Acids Research 47:D801–D806.

    https://doi.org/10.1093/nar/gky1056

    • PubMed
    • Google Scholar
12. 1. Casano AM
    2. Albert M
    3. Peri F

    (2016) Developmental Apoptosis Mediates Entry and Positioning of Microglia in the Zebrafish Brain

    Cell Reports 16:897–906.

    https://doi.org/10.1016/j.celrep.2016.06.033

    • PubMed
    • Google Scholar
13. 1. Choi HMT
    2. Schwarzkopf M
    3. Fornace ME
    4. Acharya A
    5. Artavanis G
    6. Stegmaier J
    7. Cunha A
    8. Pierce NA

    (2018) Third-generation in situ hybridization chain reaction: multiplexed, quantitative, sensitive, versatile, robust

    Development (Cambridge, England) 145:12.

    https://doi.org/10.1242/dev.165753

    • PubMed
    • Google Scholar
14. 1. Coxon A
    2. Rieu P
    3. Barkalow FJ
    4. Askari S
    5. Sharpe AH
    6. von Andrian UH
    7. Arnaout MA
    8. Mayadas TN

    (1996) A novel role for the beta 2 integrin CD11b/CD18 in neutrophil apoptosis: A homeostatic mechanism in inflammation

    Immunity 5:653–666.

    https://doi.org/10.1016/s1074-7613(00)80278-2

    • PubMed
    • Google Scholar
15. 1. Damisah EC
    2. Hill RA
    3. Rai A
    4. Chen F
    5. Rothlin CV
    6. Ghosh S
    7. Grutzendler J

    (2020) Astrocytes and microglia play orchestrated roles and respect phagocytic territories during neuronal corpse removal in vivo

    Science Advances 6:eaba3239.

    https://doi.org/10.1126/sciadv.aba3239

    • PubMed
    • Google Scholar
16. 1. Danecek P
    2. Bonfield JK
    3. Liddle J
    4. Marshall J
    5. Ohan V
    6. Pollard MO
    7. Whitwham A
    8. Keane T
    9. McCarthy SA
    10. Davies RM
    11. Li H

    (2021) Twelve years of SAMtools and BCFtools

    GigaScience 10:giab008.

    https://doi.org/10.1093/gigascience/giab008

    • PubMed
    • Google Scholar
17. 1. Denhardt DT
    2. Noda M
    3. O’Regan AW
    4. Pavlin D
    5. Berman JS

    (2001) Osteopontin as a means to cope with environmental insults: regulation of inflammation, tissue remodeling, and cell survival

    The Journal of Clinical Investigation 107:1055–1061.

    https://doi.org/10.1172/JCI12980

    • PubMed
    • Google Scholar
18. 1. Diaz-Aparicio I
    2. Paris I
    3. Sierra-Torre V
    4. Plaza-Zabala A
    5. Rodríguez-Iglesias N
    6. Márquez-Ropero M
    7. Beccari S
    8. Huguet P
    9. Abiega O
    10. Alberdi E
    11. Matute C
    12. Bernales I
    13. Schulz A
    14. Otrokocsi L
    15. Sperlagh B
    16. Happonen KE
    17. Lemke G
    18. Maletic-Savatic M
    19. Valero J
    20. Sierra A

    (2020) Microglia Actively Remodel Adult Hippocampal Neurogenesis through the Phagocytosis Secretome

    The Journal of Neuroscience 40:1453–1482.

    https://doi.org/10.1523/JNEUROSCI.0993-19.2019

    • PubMed
    • Google Scholar
19. 1. Dobin A
    2. Davis CA
    3. Schlesinger F
    4. Drenkow J
    5. Zaleski C
    6. Jha S
    7. Batut P
    8. Chaisson M
    9. Gingeras TR

    (2013) STAR: ultrafast universal RNA-seq aligner

    Bioinformatics (Oxford, England) 29:15–21.

    https://doi.org/10.1093/bioinformatics/bts635

    • PubMed
    • Google Scholar
20. 1. Doran AC
    2. Yurdagul A
    3. Tabas I

    (2020) Efferocytosis in health and disease

    Nature Reviews. Immunology 20:254–267.

    https://doi.org/10.1038/s41577-019-0240-6

    • PubMed
    • Google Scholar
21. Preprint

    1. Dorman LC
    2. Nguyen PT
    3. Escoubas CC
    4. Vainchtein ID
    5. Xiao Y
    6. Lidsky PV
    7. Nakajo H
    8. Silva NJ
    9. Lagares-Linares C
    10. Wang EY
    11. Taloma SE
    12. Cuevas B
    13. Nakao-Inoue H
    14. Rivera BM
    15. Schwer B
    16. Condello C
    17. Andino R
    18. Nowakowski TJ
    19. Molofsky AV

    (2022) A Type I Interferon Response Defines A Conserved Microglial State Required for Effective

    bioRxiv.

    https://doi.org/10.1101/2021.04.29.441889

    • Google Scholar
22. 1. Duncan JL
    2. LaVail MM
    3. Yasumura D
    4. Matthes MT
    5. Yang H
    6. Trautmann N
    7. Chappelow AV
    8. Feng W
    9. Earp HS
    10. Matsushima GK
    11. Vollrath D

    (2003) An RCS-like retinal dystrophy phenotype in mer knockout mice

    Investigative Ophthalmology & Visual Science 44:826–838.

    https://doi.org/10.1167/iovs.02-0438

    • PubMed
    • Google Scholar
23. 1. Elliott MR
    2. Ravichandran KS

    (2016) The Dynamics of Apoptotic Cell Clearance

    Developmental Cell 38:147–160.

    https://doi.org/10.1016/j.devcel.2016.06.029

    • PubMed
    • Google Scholar
24. 1. Elliott MR
    2. Koster KM
    3. Murphy PS

    (2017) Efferocytosis Signaling in the Regulation of Macrophage Inflammatory Responses

    Journal of Immunology (Baltimore, Md 198:1387–1394.

    https://doi.org/10.4049/jimmunol.1601520

    • PubMed
    • Google Scholar
25. 1. Farah MH
    2. Easter SS

    (2005) Cell birth and death in the mouse retinal ganglion cell layer

    The Journal of Comparative Neurology 489:120–134.

    https://doi.org/10.1002/cne.20615

    • PubMed
    • Google Scholar
26. 1. Farah MH

    (2006) Neurogenesis and cell death in the ganglion cell layer of vertebrate retina

    Brain Research Reviews 52:264–274.

    https://doi.org/10.1016/j.brainresrev.2006.04.002

    • PubMed
    • Google Scholar
27. 1. Fourgeaud L
    2. Través PG
    3. Tufail Y
    4. Leal-Bailey H
    5. Lew ED
    6. Burrola PG
    7. Callaway P
    8. Zagórska A
    9. Rothlin CV
    10. Nimmerjahn A
    11. Lemke G

    (2016) TAM receptors regulate multiple features of microglial physiology

    Nature 532:240–244.

    https://doi.org/10.1038/nature17630

    • PubMed
    • Google Scholar
28. 1. Fricker M
    2. Tolkovsky AM
    3. Borutaite V
    4. Coleman M
    5. Brown GC

    (2018) Neuronal Cell Death

    Physiological Reviews 98:813–880.

    https://doi.org/10.1152/physrev.00011.2017

    • PubMed
    • Google Scholar
29. 1. Galloway DA
    2. Phillips AEM
    3. Owen DRJ
    4. Moore CS

    (2019) Phagocytosis in the Brain: Homeostasis and Disease

    Frontiers in Immunology 10:790.

    https://doi.org/10.3389/fimmu.2019.00790

    • PubMed
    • Google Scholar
30. 1. García-Moreno A
    2. López-Domínguez R
    3. Ramirez-Mena A
    4. Pascual-Montano A
    5. Aparicio-Puerta E
    6. Hackenberg M
    7. Carmona-Saez P

    (2021) In-depth profiling of COVID-19 risk factors and preventive measures in healthcare workers

    Bioinformatics (Oxford, England) 1:962.

    https://doi.org/10.1101/2021.04.15.439962

    • Google Scholar
31. Preprint

    1. Garrison E
    2. Marth G

    (2012) Haplotype-Based Variant Detection from Short-Read Sequencing

    arXiv.

    https://doi.org/10.48550/arXiv.1207.3907

    • Google Scholar
32. Preprint

    1. Garrison E
    2. Kronenberg ZN
    3. Dawson ET
    4. Pedersen BS
    5. Prins P

    (2021) Vcflib and Tools for Processing the VCF Variant Call Format

    bioRxiv.

    https://doi.org/10.1101/2021.05.21.445151

    • Google Scholar
33. 1. Goruppi S
    2. Ruaro E
    3. Varnum B
    4. Schneider C

    (1997) Requirement of phosphatidylinositol 3-kinase-dependent pathway and Src for Gas6-Axl mitogenic and survival activities in NIH 3T3 fibroblasts

    Molecular and Cellular Biology 17:4442–4453.

    https://doi.org/10.1128/MCB.17.8.4442

    • PubMed
    • Google Scholar
34. 1. Gosselin D
    2. Skola D
    3. Coufal NG
    4. Holtman IR
    5. Schlachetzki JCM
    6. Sajti E
    7. Jaeger BN
    8. O’Connor C
    9. Fitzpatrick C
    10. Pasillas MP
    11. Pena M
    12. Adair A
    13. Gonda DD
    14. Levy ML
    15. Ransohoff RM
    16. Gage FH
    17. Glass CK

    (2017) An environment-dependent transcriptional network specifies human microglia identity

    Science (New York, N.Y.) 356:eaal3222.

    https://doi.org/10.1126/science.aal3222

    • PubMed
    • Google Scholar
35. 1. Hafemeister C
    2. Satija R

    (2019) Normalization and variance stabilization of single-cell RNA-seq data using regularized negative binomial regression

    Genome Biology 20:296.

    https://doi.org/10.1186/s13059-019-1874-1

    • PubMed
    • Google Scholar
36. 1. Hammond TR
    2. Dufort C
    3. Dissing-Olesen L
    4. Giera S
    5. Young A
    6. Wysoker A
    7. Walker AJ
    8. Gergits F
    9. Segel M
    10. Nemesh J
    11. Marsh SE
    12. Saunders A
    13. Macosko E
    14. Ginhoux F
    15. Chen J
    16. Franklin RJM
    17. Piao X
    18. McCarroll SA
    19. Stevens B

    (2019) Single-Cell RNA Sequencing of Microglia throughout the Mouse Lifespan and in the Injured Brain Reveals Complex Cell-State Changes

    Immunity 50:253–271.

    https://doi.org/10.1016/j.immuni.2018.11.004

    • PubMed
    • Google Scholar
37. 1. Hohsfield LA
    2. Najafi AR
    3. Ghorbanian Y
    4. Soni N
    5. Crapser J
    6. Figueroa Velez DX
    7. Jiang S
    8. Royer SE
    9. Kim SJ
    10. Henningfield CM
    11. Anderson A
    12. Gandhi SP
    13. Mortazavi A
    14. Inlay MA
    15. Green KN

    (2021) Subventricular zone/white matter microglia reconstitute the empty adult microglial niche in a dynamic wave

    eLife 10:e66738.

    https://doi.org/10.7554/eLife.66738

    • PubMed
    • Google Scholar
38. 1. Holtman IR
    2. Raj DD
    3. Miller JA
    4. Schaafsma W
    5. Yin Z
    6. Brouwer N
    7. Wes PD
    8. Möller T
    9. Orre M
    10. Kamphuis W
    11. Hol EM
    12. Boddeke E
    13. Eggen BJL

    (2015) Induction of a common microglia gene expression signature by aging and neurodegenerative conditions: a co-expression meta-analysis

    Acta Neuropathologica Communications 3:31.

    https://doi.org/10.1186/s40478-015-0203-5

    • PubMed
    • Google Scholar
39. 1. Hu B
    2. Duan S
    3. Wang Z
    4. Li X
    5. Zhou Y
    6. Zhang X
    7. Zhang YW
    8. Xu H
    9. Zheng H

    (2021) Insights Into the Role of CSF1R in the Central Nervous System and Neurological Disorders

    Frontiers in Aging Neuroscience 13:789834.

    https://doi.org/10.3389/fnagi.2021.789834

    • PubMed
    • Google Scholar
40. 1. Huang Y
    2. Happonen KE
    3. Burrola PG
    4. O’Connor C
    5. Hah N
    6. Huang L
    7. Nimmerjahn A
    8. Lemke G

    (2021) Microglia use TAM receptors to detect and engulf amyloid β plaques

    Nature Immunology 22:586–594.

    https://doi.org/10.1038/s41590-021-00913-5

    • PubMed
    • Google Scholar
41. 1. Jacobson MD
    2. Weil M
    3. Raff MC

    (1997) Programmed cell death in animal development

    Cell 88:347–354.

    https://doi.org/10.1016/s0092-8674(00)81873-5

    • PubMed
    • Google Scholar
42. 1. Jung S
    2. Aliberti J
    3. Graemmel P
    4. Sunshine MJ
    5. Kreutzberg GW
    6. Sher A
    7. Littman DR

    (2000) Analysis of fractalkine receptor CX(3)CR1 function by targeted deletion and green fluorescent protein reporter gene insertion

    Molecular and Cellular Biology 20:4106–4114.

    https://doi.org/10.1128/MCB.20.11.4106-4114.2000

    • PubMed
    • Google Scholar
43. 1. Kempthorne L
    2. Yoon H
    3. Madore C
    4. Smith S
    5. Wszolek ZK
    6. Rademakers R
    7. Kim J
    8. Butovsky O
    9. Dickson DW

    (2020) Loss of homeostatic microglial phenotype in CSF1R-related Leukoencephalopathy

    Acta Neuropathologica Communications 8:72.

    https://doi.org/10.1186/s40478-020-00947-0

    • PubMed
    • Google Scholar
44. 1. Keren-Shaul H
    2. Spinrad A
    3. Weiner A
    4. Matcovitch-Natan O
    5. Dvir-Szternfeld R
    6. Ulland TK
    7. David E
    8. Baruch K
    9. Lara-Astaiso D
    10. Toth B
    11. Itzkovitz S
    12. Colonna M
    13. Schwartz M
    14. Amit I

    (2017) A Unique Microglia Type Associated with Restricting Development of Alzheimer’s Disease

    Cell 169:1276–1290.

    https://doi.org/10.1016/j.cell.2017.05.018

    • PubMed
    • Google Scholar
45. 1. Knudson CM
    2. Tung KS
    3. Tourtellotte WG
    4. Brown GA
    5. Korsmeyer SJ

    (1995) Bax-deficient mice with lymphoid hyperplasia and male germ cell death

    Science (New York, N.Y.) 270:96–99.

    https://doi.org/10.1126/science.270.5233.96

    • PubMed
    • Google Scholar
46. 1. Krasemann S
    2. Madore C
    3. Cialic R
    4. Baufeld C
    5. Calcagno N
    6. El Fatimy R
    7. Beckers L
    8. O’Loughlin E
    9. Xu Y
    10. Fanek Z
    11. Greco DJ
    12. Smith ST
    13. Tweet G
    14. Humulock Z
    15. Zrzavy T
    16. Conde-Sanroman P
    17. Gacias M
    18. Weng Z
    19. Chen H
    20. Tjon E
    21. Mazaheri F
    22. Hartmann K
    23. Madi A
    24. Ulrich JD
    25. Glatzel M
    26. Worthmann A
    27. Heeren J
    28. Budnik B
    29. Lemere C
    30. Ikezu T
    31. Heppner FL
    32. Litvak V
    33. Holtzman DM
    34. Lassmann H
    35. Weiner HL
    36. Ochando J
    37. Haass C
    38. Butovsky O

    (2017) The TREM2-APOE Pathway Drives the Transcriptional Phenotype of Dysfunctional Microglia in Neurodegenerative Diseases

    Immunity 47:566–581.

    https://doi.org/10.1016/j.immuni.2017.08.008

    • PubMed
    • Google Scholar
47. 1. Lavin Y
    2. Winter D
    3. Blecher-Gonen R
    4. David E
    5. Keren-Shaul H
    6. Merad M
    7. Jung S
    8. Amit I

    (2014) Tissue-resident macrophage enhancer landscapes are shaped by the local microenvironment

    Cell 159:1312–1326.

    https://doi.org/10.1016/j.cell.2014.11.018

    • PubMed
    • Google Scholar
48. 1. Lemke G

    (2013) Biology of the TAM receptors

    Cold Spring Harbor Perspectives in Biology 5:a009076.

    https://doi.org/10.1101/cshperspect.a009076

    • PubMed
    • Google Scholar
49. 1. Lemke G

    (2019) How macrophages deal with death

    Nature Reviews. Immunology 19:539–549.

    https://doi.org/10.1038/s41577-019-0167-y

    • PubMed
    • Google Scholar
50. 1. Li Q
    2. Cheng Z
    3. Zhou L
    4. Darmanis S
    5. Neff NF
    6. Okamoto J
    7. Gulati G
    8. Bennett ML
    9. Sun LO
    10. Clarke LE
    11. Marschallinger J
    12. Yu G
    13. Quake SR
    14. Wyss-Coray T
    15. Barres BA

    (2019a) Developmental Heterogeneity of Microglia and Brain Myeloid Cells Revealed by Deep Single-Cell RNA Sequencing

    Neuron 101:207–223.

    https://doi.org/10.1016/j.neuron.2018.12.006

    • PubMed
    • Google Scholar
51. 1. Li F
    2. Jiang D
    3. Samuel MA

    (2019b) Microglia in the developing retina

    Neural Development 14:12.

    https://doi.org/10.1186/s13064-019-0137-x

    • PubMed
    • Google Scholar
52. 1. Liao Y
    2. Smyth GK
    3. Shi W

    (2014) featureCounts: an efficient general purpose program for assigning sequence reads to genomic features

    Bioinformatics (Oxford, England) 30:923–930.

    https://doi.org/10.1093/bioinformatics/btt656

    • PubMed
    • Google Scholar
53. 1. Liddelow SA
    2. Marsh SE
    3. Stevens B

    (2020) Microglia and Astrocytes in Disease: Dynamic Duo or Partners in Crime?

    Trends in Immunology 41:820–835.

    https://doi.org/10.1016/j.it.2020.07.006

    • PubMed
    • Google Scholar
54. 1. Love MI
    2. Huber W
    3. Anders S

    (2014) Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2

    Genome Biology 15:550.

    https://doi.org/10.1186/s13059-014-0550-8

    • PubMed
    • Google Scholar
55. 1. Lu Q
    2. Gore M
    3. Zhang Q
    4. Camenisch T
    5. Boast S
    6. Casagranda F
    7. Lai C
    8. Skinner MK
    9. Klein R
    10. Matsushima GK
    11. Earp HS
    12. Goff SP
    13. Lemke G

    (1999) Tyro-3 family receptors are essential regulators of mammalian spermatogenesis

    Nature 398:723–728.

    https://doi.org/10.1038/19554

    • PubMed
    • Google Scholar
56. 1. MacDonald KPA
    2. Palmer JS
    3. Cronau S
    4. Seppanen E
    5. Olver S
    6. Raffelt NC
    7. Kuns R
    8. Pettit AR
    9. Clouston A
    10. Wainwright B
    11. Branstetter D
    12. Smith J
    13. Paxton RJ
    14. Cerretti DP
    15. Bonham L
    16. Hill GR
    17. Hume DA

    (2010) An antibody against the colony-stimulating factor 1 receptor depletes the resident subset of monocytes and tissue- and tumor-associated macrophages but does not inhibit inflammation

    Blood 116:3955–3963.

    https://doi.org/10.1182/blood-2010-02-266296

    • PubMed
    • Google Scholar
57. 1. Márquez-Ropero M
    2. Benito E
    3. Plaza-Zabala A
    4. Sierra A

    (2020) Microglial Corpse Clearance: Lessons From Macrophages

    Frontiers in Immunology 11:506.

    https://doi.org/10.3389/fimmu.2020.00506

    • PubMed
    • Google Scholar
58. 1. Marschallinger J
    2. Iram T
    3. Zardeneta M
    4. Lee SE
    5. Lehallier B
    6. Haney MS
    7. Pluvinage JV
    8. Mathur V
    9. Hahn O
    10. Morgens DW
    11. Kim J
    12. Tevini J
    13. Felder TK
    14. Wolinski H
    15. Bertozzi CR
    16. Bassik MC
    17. Aigner L
    18. Wyss-Coray T

    (2020) Lipid-droplet-accumulating microglia represent a dysfunctional and proinflammatory state in the aging brain

    Nature Neuroscience 23:194–208.

    https://doi.org/10.1038/s41593-019-0566-1

    • PubMed
    • Google Scholar
59. 1. Marsh SE
    2. Kamath T
    3. Walker AJ
    4. Dissing-Olesen L
    5. Hammond TR
    6. Young AMH
    7. Abdulraouf A
    8. Nadaf N
    9. Dufort C
    10. Murphy S
    11. Kozareva V
    12. Vanderburg C
    13. Hong S
    14. Bulstrode H
    15. Hutchinson PJ
    16. Gaffney DJ
    17. Franklin RJM
    18. Macosko EZ
    19. Stevens B

    (2020) Dissection of artifactual and confounding glial signatures by single-cell sequencing of mouse and human brain

    Neuroscience 1:542.

    https://doi.org/10.1101/2020.12.03.408542

    • Google Scholar
60. 1. Martin M

    (2011) Cutadapt removes adapter sequences from high-throughput sequencing reads

    EMBnet.Journal 17:10.

    https://doi.org/10.14806/ej.17.1.200

    • Google Scholar
61. 1. Masuda T
    2. Sankowski R
    3. Staszewski O
    4. Böttcher C
    5. Amann L
    6. Scheiwe C
    7. Nessler S
    8. Kunz P
    9. van Loo G
    10. Coenen VA
    11. Reinacher PC
    12. Michel A
    13. Sure U
    14. Gold R
    15. Grün D
    16. Priller J
    17. Stadelmann C
    18. Prinz M

    (2019) Spatial and temporal heterogeneity of mouse and human microglia at single-cell resolution

    Nature 566:388–392.

    https://doi.org/10.1038/s41586-019-0924-x

    • PubMed
    • Google Scholar
62. 1. Masuda T.
    2. Sankowski R
    3. Staszewski O
    4. Prinz M

    (2020) Microglia Heterogeneity in the Single-Cell Era

    Cell Reports 30:1271–1281.

    https://doi.org/10.1016/j.celrep.2020.01.010

    • PubMed
    • Google Scholar
63. 1. Morioka S
    2. Maueröder C
    3. Ravichandran KS

    (2019) Living on the Edge: Efferocytosis at the Interface of Homeostasis and Pathology

    Immunity 50:1149–1162.

    https://doi.org/10.1016/j.immuni.2019.04.018

    • PubMed
    • Google Scholar
64. 1. Nandi S
    2. Gokhan S
    3. Dai XM
    4. Wei S
    5. Enikolopov G
    6. Lin H
    7. Mehler MF
    8. Stanley ER

    (2012) The CSF-1 receptor ligands IL-34 and CSF-1 exhibit distinct developmental brain expression patterns and regulate neural progenitor cell maintenance and maturation

    Developmental Biology 367:100–113.

    https://doi.org/10.1016/j.ydbio.2012.03.026

    • PubMed
    • Google Scholar
65. 1. Nemes-Baran AD
    2. White DR
    3. DeSilva TM

    (2020) Fractalkine-Dependent Microglial Pruning of Viable Oligodendrocyte Progenitor Cells Regulates Myelination

    Cell Reports 32:108047.

    https://doi.org/10.1016/j.celrep.2020.108047

    • PubMed
    • Google Scholar
66. 1. Oppenheim RW

    (1991) Cell death during development of the nervous system

    Annual Review of Neuroscience 14:453–501.

    https://doi.org/10.1146/annurev.ne.14.030191.002321

    • PubMed
    • Google Scholar
67. 1. Païdassi H
    2. Tacnet-Delorme P
    3. Garlatti V
    4. Darnault C
    5. Ghebrehiwet B
    6. Gaboriaud C
    7. Arlaud GJ
    8. Frachet P

    (2008) C1q binds phosphatidylserine and likely acts as a multiligand-bridging molecule in apoptotic cell recognition

    Journal of Immunology (Baltimore, Md 180:2329–2338.

    https://doi.org/10.4049/jimmunol.180.4.2329

    • PubMed
    • Google Scholar
68. 1. Péquignot MO
    2. Provost AC
    3. Sallé S
    4. Taupin P
    5. Sainton KM
    6. Marchant D
    7. Martinou JC
    8. Ameisen JC
    9. Jais J-P
    10. Abitbol M

    (2003) Major role of BAX in apoptosis during retinal development and in establishment of a functional postnatal retina

    Developmental Dynamics 228:231–238.

    https://doi.org/10.1002/dvdy.10376

    • PubMed
    • Google Scholar
69. 1. Pierce A
    2. Bliesner B
    3. Xu M
    4. Nielsen-Preiss S
    5. Lemke G
    6. Tobet S
    7. Wierman ME

    (2008) Axl and Tyro3 modulate female reproduction by influencing gonadotropin-releasing hormone neuron survival and migration

    Molecular Endocrinology (Baltimore, Md.) 22:2481–2495.

    https://doi.org/10.1210/me.2008-0169

    • PubMed
    • Google Scholar
70. 1. Prinz M
    2. Masuda T
    3. Wheeler MA
    4. Quintana FJ

    (2021) Microglia and Central Nervous System-Associated Macrophages-From Origin to Disease Modulation

    Annual Review of Immunology 39:251–277.

    https://doi.org/10.1146/annurev-immunol-093019-110159

    • PubMed
    • Google Scholar
71. 1. Puñal VM
    2. Paisley CE
    3. Brecha FS
    4. Lee MA
    5. Perelli RM
    6. Wang J
    7. O’Koren EG
    8. Ackley CR
    9. Saban DR
    10. Reese BE
    11. Kay JN

    (2019) Large-scale death of retinal astrocytes during normal development is non-apoptotic and implemented by microglia

    PLOS Biology 17:e3000492.

    https://doi.org/10.1371/journal.pbio.3000492

    • PubMed
    • Google Scholar
72. 1. Reddy SM
    2. Hsiao KHK
    3. Abernethy VE
    4. Fan H
    5. Longacre A
    6. Lieberthal W
    7. Rauch J
    8. Koh JS
    9. Levine JS

    (2002) Phagocytosis of apoptotic cells by macrophages induces novel signaling events leading to cytokine-independent survival and inhibition of proliferation: activation of Akt and inhibition of extracellular signal-regulated kinases 1 and 2

    Journal of Immunology (Baltimore, Md 169:702–713.

    https://doi.org/10.4049/jimmunol.169.2.702

    • PubMed
    • Google Scholar
73. 1. Rojo AI
    2. McBean G
    3. Cindric M
    4. Egea J
    5. López MG
    6. Rada P
    7. Zarkovic N
    8. Cuadrado A

    (2014) Redox control of microglial function: molecular mechanisms and functional significance

    Antioxidants & Redox Signaling 21:1766–1801.

    https://doi.org/10.1089/ars.2013.5745

    • PubMed
    • Google Scholar
74. 1. Saleh S
    2. Thompson DE
    3. McConkey J
    4. Murray P
    5. Moorehead RA

    (2016) Osteopontin regulates proliferation, apoptosis, and migration of murine claudin-low mammary tumor cells

    BMC Cancer 16:359.

    https://doi.org/10.1186/s12885-016-2396-9

    • PubMed
    • Google Scholar
75. 1. Seitz HM
    2. Camenisch TD
    3. Lemke G
    4. Earp HS
    5. Matsushima GK

    (2007) Macrophages and dendritic cells use different Axl/Mertk/Tyro3 receptors in clearance of apoptotic cells

    Journal of Immunology (Baltimore, Md 178:5635–5642.

    https://doi.org/10.4049/jimmunol.178.9.5635

    • PubMed
    • Google Scholar
76. 1. Shankar SL
    2. O’Guin K
    3. Cammer M
    4. McMorris FA
    5. Stitt TN
    6. Basch RS
    7. Varnum B
    8. Shafit-Zagardo B

    (2003)

    The growth arrest-specific gene product Gas6 promotes the survival of human oligodendrocytes via a phosphatidylinositol 3-kinase-dependent pathway

    The Journal of Neuroscience 23:4208–4218.

    • PubMed
    • Google Scholar
77. 1. Shankar SL
    2. O’Guin K
    3. Kim M
    4. Varnum B
    5. Lemke G
    6. Brosnan CF
    7. Shafit-Zagardo B

    (2006) Gas6/Axl signaling activates the phosphatidylinositol 3-kinase/Akt1 survival pathway to protect oligodendrocytes from tumor necrosis factor alpha-induced apoptosis

    The Journal of Neuroscience 26:5638–5648.

    https://doi.org/10.1523/JNEUROSCI.5063-05.2006

    • PubMed
    • Google Scholar
78. 1. Stuart T
    2. Butler A
    3. Hoffman P
    4. Hafemeister C
    5. Papalexi E
    6. Mauck WM III
    7. Hao Y
    8. Stoeckius M
    9. Smibert P
    10. Satija R

    (2019) Comprehensive Integration of Single-Cell Data

    Cell 177:1888–1902.

    https://doi.org/10.1016/j.cell.2019.05.031

    • Google Scholar
79. 1. VanRyzin JW

    (2021) Phagocytic microglia in development: Are they what they eat?

    Brain, Behavior, & Immunity - Health 18:100373.

    https://doi.org/10.1016/j.bbih.2021.100373

    • Google Scholar
80. 1. Wlodarczyk A
    2. Holtman IR
    3. Krueger M
    4. Yogev N
    5. Bruttger J
    6. Khorooshi R
    7. Benmamar-Badel A
    8. de Boer-Bergsma JJ
    9. Martin NA
    10. Karram K
    11. Kramer I
    12. Boddeke EW
    13. Waisman A
    14. Eggen BJ
    15. Owens T

    (2017) A novel microglial subset plays A key role in myelinogenesis in developing brain

    The EMBO Journal 36:3292–3308.

    https://doi.org/10.15252/embj.201696056

    • PubMed
    • Google Scholar
81. 1. Wolf Y
    2. Yona S
    3. Kim KW
    4. Jung S

    (2013) Microglia, seen from the CX3CR1 angle

    Frontiers in Cellular Neuroscience 7:26.

    https://doi.org/10.3389/fncel.2013.00026

    • PubMed
    • Google Scholar
82. 1. Xu J
    2. Falconer C
    3. Nguyen Q
    4. Crawford J
    5. McKinnon BD
    6. Mortlock S
    7. Senabouth A
    8. Andersen S
    9. Chiu HS
    10. Jiang L
    11. Palpant NJ
    12. Yang J
    13. Mueller MD
    14. Hewitt AW
    15. Pébay A
    16. Montgomery GW
    17. Powell JE
    18. Coin LJM

    (2019) Genotype-free demultiplexing of pooled single-cell RNA-seq

    Genome Biology 20:290.

    https://doi.org/10.1186/s13059-019-1852-7

    • PubMed
    • Google Scholar
83. 1. Yu H
    2. Liu X
    3. Zhong Y

    (2017) The Effect of Osteopontin on Microglia

    BioMed Research International 2017:1879437.

    https://doi.org/10.1155/2017/1879437

    • PubMed
    • Google Scholar
84. 1. Zent CS
    2. Elliott MR

    (2017) Maxed out macs: physiologic cell clearance as a function of macrophage phagocytic capacity

    The FEBS Journal 284:1021–1039.

    https://doi.org/10.1111/febs.13961

    • PubMed
    • Google Scholar
85. 1. Zhan L
    2. Fan L
    3. Kodama L
    4. Sohn PD
    5. Wong MY
    6. Mousa GA
    7. Zhou Y
    8. Li Y
    9. Gan L

    (2020) A MAC2-positive progenitor-like microglial population is resistant to CSF1R inhibition in adult mouse brain

    eLife 9:e51796.

    https://doi.org/10.7554/eLife.51796

    • PubMed
    • Google Scholar
86. 1. Zhu C
    2. Wei Y
    3. Wei X

    (2019) AXL receptor tyrosine kinase as a promising anti-cancer approach: functions, molecular mechanisms and clinical applications

    Molecular Cancer 18:153.

    https://doi.org/10.1186/s12943-019-1090-3

    • PubMed
    • Google Scholar

Decision letter after peer review:

Thank you for submitting your article "Neuronal apoptosis drives remodeling states of microglia and shifts in survival pathway dependence" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Marianne Bronner as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Your study highlights important and novel functions of cell death in shaping microglial states in the retina, as well as the pathways involved. However, the three reviewers have raised some concerns that you will find detailed below. In particular, they raised some points that should addressed in your revised manuscript:

1) The single cell transcriptomic results appear to have been performed on a single biological replicate, which makes it difficult to convincingly draw strong conclusions especially on the relative shifts in population sizes. To address these concerns, it would be important to either add a biological replicate of the transcriptomic study or to confirm by distinct techniques (such as in vivo staining) that the different microglial states are reduced or increased in different experimental conditions examined by the authors.

2) Several points were raised regarding a need for clarification of the methodological approaches used in the transcriptomic analyses that should be addressed to strengthen the conclusions put forward as well as to provide some of the lists of genes discussed by the reviewers. It would be helpful for the reader to follow the flow of the analyses and manuscript.

3) While the link between CSF1R resistance and the induction of a "remodelling state" could be supported by additional experiments, we believe that this lies outside of the scope of this already important study. Nonetheless, it would be important for the authors to down-tone this specific conclusion and discuss additionnal possibilities.

Reviewer #1 (Recommendations for the authors):

Reviewer concerns and suggestions for improvement:

1. Additional methodological details about scRNAseq processing are needed.

First and most importantly: In line 566 (Methods) the authors state that filtered cells from all 4 datasets were merged and then processed with the [Seurat] SCTransform pipeline. Were the 4 datasets combined using Seurat's integration/batch correction functionality, or were they simply merged? If the former, please provide in the Methods the details of the integration workflow. If the latter, the authors should explain the steps that were taken, if any, to mitigate batch effects. In the absence of such steps, there is a concern that batch effects, rather than true biological differences, could be the major driver of the differences in cluster composition between the WT/KO/vehicle/PLX samples.

2. Second, the authors do not state which count data from the Seurat data object were used to make figures and perform differential expression analysis. The standard in the field is to use log-normalized read counts. But since the SCTransform pipeline was used, the object will also contain "corrected" counts obtained via modeling. The methods should state which were used. If the latter were used the authors should justify this. It is particularly important to consider the appropriate methods for identifying differentially expressed genes (DEGs) because cluster 0, the major remodeling cluster, only had 4 upregulated DEGs that distinguished it from other clusters (Table S1). Two of these, Apoe and Ctsd, are important to the authors argument that cluster 0 cells have a remodeling phenotype. Further, some studies use lack of upregulated DEGs as an argument for "overclustering" i.e. that the clustering algorithm has split the data in a non-meaningful way. So a finding of zero DEGs would be significant for interpretation. For these two reasons it is critical to establish that the cluster 0 markers were identified in a sound manner.

3. Because Cluster 0 lacks definitive markers (Table S1), the reader is left wondering why exactly the clustering algorithm defined it as distinct from other clusters. The clear implication of the text is that cluster 0 is mainly defined by downregulation of homeostatic (cluster 1) markers. Four qualitative examples of this are given in Figure 2b. This point should be supported by a more systematic analysis. There are a number of ways this could be done, e.g. by showing DEGs downregulated in cluster 0 (compared to cluster 1 or to the entire dataset), or by plotting module scores for a larger set of homeostatic genes. Such an analysis is important to establish what genetic differences distinguish the cluster 0 cells, and whether they are biologically plausible or instead might indicate an overclustering artifact.

4. In the section on bulk RNAseq of Mer and Axl mutants, it is stated that there was no effect on lysosomal or lipid metabolism genes or DAM/ATM/PAM genes (lines 310-12 and 318-19). This is not clearly demonstrated by the data in Figure 7. It is great to see the heatmap listing all differentially expressed genes, but the authors should clarify that this list is consistent with the statements in the main text. (The reader cannot be expected to know what all of these genes do!) This could be done either by a gene ontology analysis or by explicitly providing a list of lysosomal and lipid metabolism genes that was used to compare with the differentially expressed gene list from Figure 7D. As for the DAM/ATM/PAM genes -- some are highlighted in the figures and text so this category is better supported. But I still think it would improve clarity and methodological transparency if the authors would provide a list of the genes they defined as DAM/ATM/PAM for this and other analyses used throughout the study.

5. Other scRNAseq studies using enzymatic dissociation have found clusters of ex-vivo activated microglia. See Marsh et al. 2020 biorxiv (cited by the authors). The fact that cells expressing this distinctive profile are missing is noteworthy and merits explanation. Were these cells detected and filtered out? Were transcriptional inhibitors used?

6. The conceptual explanation for the Bax in silico genotyping strategy needs to be expanded, either in the supplementary figure legend or Methods. I appreciated the supplementary figure, but I still don't really understand what the overall strategy was or what the plots in Figure S3C-F are showing. Why was it expected that BaxWT and KO would have distinct SNPs? How were the SNPs identified (at a conceptual level; the explanation of software tools is reasonably clear)? Why are Ftl1 and Ftl-ps1 part of this analysis?

7. The plots in Figure 2B-E (UMAP Feature Plots) give a qualitative sense of how these genes change across clusters. The authors' argument would be strengthened by pairing this with a more quantitative view of how gene expression differs between homeostatic and remodeling groups. E.g. by using dot plots, violin plots, or explicit tests for up/downregulated genes comparing of homeostatic vs. remodeling groups.

8. The analysis of apoptotic clearance in Mer and Axl mutants is compelling. However only one time point, near the end of the death period, was checked (P5). Do apoptotic bodies persist in these mutants or can other mechanisms compensate for the loss of Mer/Axl?

Reviewer #2 (Recommendations for the authors):

In the article titled "Neuronal apoptosis drives remodeling states of microglia and shifts in survival pathway dependence" the authors describe the heterogeneity of the microglial population in the early postnatal retina. In WT mice, the authors find both "homeostatic" microglia (expressing Tmem119 and P2RY12) and several clusters of "remodeling" microglia (expressing lysosomal and phagocytic components). In a Bax KO mouse lacking the significant apoptosis that usually occurs in the early postnatal retina, they find a decrease in the remodeling clusters of microglia, concluding that the lack of apoptosis in this mouse prevents the shift in microglial phenotype from homeostatic to remodeling. Several of these "remodeling" clusters in WT mice are relatively enriched after short term PLX3397 treatment, a CSF1R inhibitor commonly used to deplete microglia in mouse.

This paper provides interesting new phenotyping of retinal microglia and how they change with Bax KO and PLX treatment, but one central conclusion is currently not fully supported by the data provided. The central claim (and a primary new finding compared to a prior study from this group, Anderson et al. 2019, Cell Reports), is that the identified remodeling clusters are not only CSF1R independent, but more specifically that expression of remodeling genes in response to neuronal apoptosis confers this survival advantage.

Two major concerns currently prevent this claim from being supported by the data shown:

1. Lack of biological replicates for single cell data: unless I misinterpreted the methods, there appears to be a single biological replicate for each single cell experimental condition. The authors rely on their sing cell data to measure the abundance of various microglial populations in their models. In the absence of replicates, it is not possible to confidently measure population abundance within and between samples, because there is no way to make an estimate of the error present in cluster sizes. By analogy, a single biological replicate of flow cytometry analysis could not be used to reliably quantify the size of populations. This reviewer does not want to force the authors to double or triple their single cell seq efforts, but this technical limitation really prevents reliable comparison of the abundance of cells in a given cluster within and across conditions, and these comparisons form the central basis for their claims. If not by more biological replicates of single cell data, a potential way to address this is to validate and better quantify the finding of shifted macrophage states using another method to show that remodeling cells go away in BAX ko, and are enriched with PLX treatment, and a possibility would be to use tissue staining or in situ to do this. Or maybe facs with some remodelign cluster markers?

2. Relationship between CSF1R independence and remodeling states: the authors conclude that remodeling states driven by neuronal apoptosis confer CSF1R independence. This conclusion is currently based on inferences from transcriptomic data about the functions of remodeling and PLX resistant microglia (eg that because they have a similar transcriptional profile, they are both engaged in remodeling), which is plausible but there are other equally plausible possibilities, such as that PLX directly induces a state change in microglia that does not involve neuronal apoptosis or remodeling. If the authors could directly show that surviving microglia are indeed engaged in some kind of a remodeling function, this would go a long way to support the claims. If not, careful adjustment of the claims and text would ensure that the claims are truly supported by the data shown, including a discussion of alternative explanations, and the limitations of transcriptomic inference to truly confirm that the "remodeling microglia" and the "plx resistant microglia" are a similar population of cells. This is particularly important because the major cluster in the plx3397 treated sample (cluster 3) is not present in any other condition.

Reviewer #3 (Recommendations for the authors):

The strength of the manuscript lies in its thorough and comprehensive characterization of multiple microglia states in the developing postnatal mouse retina. By the mean of scRNAseq analysis, the authors addressed the question of which environmental cues drive such heterogeneity.

This work allowed the identification of 11 microglia states that coexist in wild type postnatal retina, ranging from a so-called homeostatic state to some remodeling states, with for instance a disease-associated signature. The proportion of the different microglia states is context-dependent and the authors demonstrated that the spectrum of homeostatic to more remodeling clusters is driven predominately by neuronal apoptosis (resulting from waves of neuronal cell death during the postnatal period in the retina). On the other hand, recognition receptors, Mer or Axl, required for clearance of apoptotic cells and survival in the absence of CSF1R signaling, respectively, do not mediate changes in microglia remodeling gene expression.

The manuscript is extremely clear and the data are thoroughly described. They perfectly support the main conclusions. Although the finding that developmental apoptosis impact retinal microglia gene signature is not novel (a manuscript from the authors and based on bulk RNAseq analysis was published in Cell Reports in 2019), the comprehensive scRNA-seq clustering analysis is an important addition, providing a valuable resource for the field.

Some questions listed below should however be addressed for clarifications:

1. The authors nicely identified 11 clusters of microglia in the developing postnatal retina. This raises the question of the spatial distribution of these microglia within the retina and the dynamics during development. Although this goes beyond the scope of the manuscript, the authors show in Supp Figure 5 by in situ hybridization that Ccl3+ microglia are dispersed in the different retinal layers, suggesting a regular mosaic distribution of cluster 3. This question of spatial distribution of microglia clusters could be raised in the Discussion section.

2. On the right of cluster 1 in UMAP plots, there is a non-annotated blue cluster. What does it correspond to?

3. The authors found that density was reduced by nearly half in Bax KO. What is their interpretation of this data? Does this result from less microglia proliferation?

4. Statistical analysis would be useful in Figure 4D, Figure 3B and Figure 4B. Also, indicating ns on the graph for the 3 KO in Figure 6C would be helpful.

5. The authors conclude from their RNAseq analysis that there is no change in genes associated with survival pathways in Axl KO microglia compared to WT. However, they found that Spp1 is downregulated. Given that Osteopontin signaling is associated with survival in different cell types, why the authors do not consider it as a candidate?

6. From their RNAseq analysis, the authors conclude that Mer and Axl are not required to drive microglia remodeling gene expression. However, can the authors really exclude the possibility of post-transcriptional regulation?

7. Line 367, the authors conclude from their data that there are microglial states that are not regulated by neuronal death. Is it really the only interpretation of the data? Could it be a matter of threshold? If there were more apoptotic cells, would these microglial states still be not responsive?

8. The authors mention in the Discussion section that one candidate signaling downstream Axl is the PI3K-Akt-NfkB-Bcl2 pathway. It is not clear whether this is supported by the RNAseq analysis from the present manuscript.

Reviewer #1 (Recommendations for the authors):

Reviewer concerns and suggestions for improvement:

1. Additional methodological details about scRNAseq processing are needed.

First and most importantly: In line 566 (Methods) the authors state that filtered cells from all 4 datasets were merged and then processed with the [Seurat] SCTransform pipeline. Were the 4 datasets combined using Seurat's integration/batch correction functionality, or were they simply merged? If the former, please provide in the Methods the details of the integration workflow. If the latter, the authors should explain the steps that were taken, if any, to mitigate batch effects. In the absence of such steps, there is a concern that batch effects, rather than true biological differences, could be the major driver of the differences in cluster composition between the WT/KO/vehicle/PLX samples.

No batch correction was performed on the single cell datasets. They were simply merged for dimensional reduction, and the methods have been updated to make this clear (lines 881). Batch effects were not obviously improved by batch correction with Seurat’s integration pipeline (see Author response image 1). Batch correction only served to separate Bax KO cells such that clearly similar mitotic cell populations no longer overlapped. While we acknowledge that there are differences between batches, we felt that the uncorrected data are more representative of the experiment.

Author response image 1

Download asset Open asset

2. Second, the authors do not state which count data from the Seurat data object were used to make figures and perform differential expression analysis. The standard in the field is to use log-normalized read counts. But since the SCTransform pipeline was used, the object will also contain "corrected" counts obtained via modeling. The methods should state which were used. If the latter were used the authors should justify this. It is particularly important to consider the appropriate methods for identifying differentially expressed genes (DEGs) because cluster 0, the major remodeling cluster, only had 4 upregulated DEGs that distinguished it from other clusters (Table S1). Two of these, Apoe and Ctsd, are important to the authors argument that cluster 0 cells have a remodeling phenotype. Further, some studies use lack of upregulated DEGs as an argument for "overclustering" i.e. that the clustering algorithm has split the data in a non-meaningful way. So a finding of zero DEGs would be significant for interpretation. For these two reasons it is critical to establish that the cluster 0 markers were identified in a sound manner.

Natural-log normalized counts were used for differential expression analysis and the methods have been updated to make this clear (lines 893-894). We have re-examined the DEGs defining the clusters, and found that the submitted Supplementary Table 1 was incomplete. We very much appreciate the Reviewer helping us to identify this error. Supplementary Table 1 (now called Supplementary File 1) has been updated to include additional upregulated and downregulated genes. It now lists 22 upregulated genes and 85 downregulated genes in Cluster 0.

3. Because Cluster 0 lacks definitive markers (Table S1), the reader is left wondering why exactly the clustering algorithm defined it as distinct from other clusters. The clear implication of the text is that cluster 0 is mainly defined by downregulation of homeostatic (cluster 1) markers. Four qualitative examples of this are given in Figure 2b. This point should be supported by a more systematic analysis. There are a number of ways this could be done, e.g. by showing DEGs downregulated in cluster 0 (compared to cluster 1 or to the entire dataset), or by plotting module scores for a larger set of homeostatic genes. Such an analysis is important to establish what genetic differences distinguish the cluster 0 cells, and whether they are biologically plausible or instead might indicate an overclustering artifact.

We thank the reviewer for drawing attention to Cluster 0. Our updated DEG list (Supplementary File 1) now includes 22 upregulated genes including remodeling markers (Apoe, Lyz2) and 85 downregulated genes including homeostatic marker Tmem119 and chemokines such as Ccl4 and Cxcl2. We are confident that this represents a distinct and relevant cluster, and not an overclustering artifact. We also added differential expression analysis between homeostatic (1 and 2) and remodeling clusters (0,3,4,5,6) to provide a list of differentially expressed genes (Supplementary File 2).

4. In the section on bulk RNAseq of Mer and Axl mutants, it is stated that there was no effect on lysosomal or lipid metabolism genes or DAM/ATM/PAM genes (lines 310-12 and 318-19). This is not clearly demonstrated by the data in Figure 7. It is great to see the heatmap listing all differentially expressed genes, but the authors should clarify that this list is consistent with the statements in the main text. (The reader cannot be expected to know what all of these genes do!) This could be done either by a gene ontology analysis or by explicitly providing a list of lysosomal and lipid metabolism genes that was used to compare with the differentially expressed gene list from Figure 7D. As for the DAM/ATM/PAM genes -- some are highlighted in the figures and text so this category is better supported. But I still think it would improve clarity and methodological transparency if the authors would provide a list of the genes they defined as DAM/ATM/PAM for this and other analyses used throughout the study.

We agree that comparison to a defined list would improve transparency and methodological rigor. We have addressed this by adding citations for GO lists of lysosomal and lipid metabolism genes and adding detail about DAM/ATM/PAM genes to the results (lines 170, 183-186, 188-191, 346-351, 354-356, 359-361). Further, we added a new section in the methods clearly defining these lists (lines 966-976). We used these lists to compare to bulk seq data in Figure 7 and scRNAseq in Figure 2.

5. Other scRNAseq studies using enzymatic dissociation have found clusters of ex-vivo activated microglia. See Marsh et al. 2020 biorxiv (cited by the authors). The fact that cells expressing this distinctive profile are missing is noteworthy and merits explanation. Were these cells detected and filtered out? Were transcriptional inhibitors used?

We agree that this is an important point. We did not filter out activated microglia or use transcriptional inhibitors during the dissociation. We cannot rule out ex vivo transcriptional changes, and do see expression of ex vivo activation (exAM) genes in some of our clusters. However, we found that ex vivo activation did not drive the formation of a distinct cluster as most of the expressed exAM genes were upregulated in multiple clusters. In addition, some of these genes were differentially expressed between samples. This may argue that these genes could have biological relevance. In support of this, histone-related “exAM” genes, for example, were predominately expressed in cycling microglia. When we compared our clusters to 27 exAM markers the authors used for scoring (Table 4 in Marsh et al. 2020), clusters 2 and 4 expressed the most exAM genes with 15/27 and 10/27, respectively. However, some of these genes have been shown to be expressed in vivo. We confirmed that two of these (Ccl3 and Ccl4) were expressed by in situ hybridization (Supplemental Figure 5 and not shown). Because the original study was in adult, the extent to which ex vivo transcription impacts developmental datasets or whether the same signature would emerge, are currently unknown. However, we cannot rule out ex vivo transcriptional changes and have added a sentence to the discussion (lines 427-432).

6. The conceptual explanation for the Bax in silico genotyping strategy needs to be expanded, either in the supplementary figure legend or Methods. I appreciated the supplementary figure, but I still don't really understand what the overall strategy was or what the plots in Figure S3C-F are showing. Why was it expected that BaxWT and KO would have distinct SNPs? How were the SNPs identified (at a conceptual level; the explanation of software tools is reasonably clear)? Why are Ftl1 and Ftl-ps1 part of this analysis?

We agree that the in-silico genotyping strategy was not well explained. We have made changes to Supplementary Figures 2 and 3 in an effort to make the strategy and results clearer. We have expanded our explanation of this process in the methods (lines 897-925), and added a brief note on why Bax-linked genes, Ftl1 and Ftl1-ps1, were considered in these analyses (lines 883-888), but we have removed the plots.

7. The plots in Figure 2B-E (UMAP Feature Plots) give a qualitative sense of how these genes change across clusters. The authors' argument would be strengthened by pairing this with a more quantitative view of how gene expression differs between homeostatic and remodeling groups. E.g. by using dot plots, violin plots, or explicit tests for up/downregulated genes comparing of homeostatic vs. remodeling groups.

We agree with the reviewer that Figure 2 would be strengthened by a more quantitative view. Therefore, we added violin plots of select clusters highlighting genes of interest (Figure 2C’,D’,E’, F’). We also added differential expression analysis between homeostatic (1 and 2) and remodeling clusters (0,3,4,5,6,10) Supplementary File 2.

8. The analysis of apoptotic clearance in Mer and Axl mutants is compelling. However only one time point, near the end of the death period, was checked (P5). Do apoptotic bodies persist in these mutants or can other mechanisms compensate for the loss of Mer/Axl?

We thank the reviewer for the interesting question. We think other mechanisms likely compensate in the absence of Mertk or both Mertk/Axl since microglia express a broad array of phagocytic receptors and other cell types such as astrocytes can phagocytose, particularly in the context of deficient microglia (Puñal et al. 2019). However, experiments looking at later timepoints would be difficult to interpret due to cell death of RGCs until P10, albeit at low levels, as well as the continued breakdown of apoptotic cells into smaller and smaller bodies. We added a sentence clarifying this point in the discussion (lines 442-445).

Reviewer #2 (Recommendations for the authors):

In the article titled "Neuronal apoptosis drives remodeling states of microglia and shifts in survival pathway dependence" the authors describe the heterogeneity of the microglial population in the early postnatal retina. In WT mice, the authors find both "homeostatic" microglia (expressing Tmem119 and P2RY12) and several clusters of "remodeling" microglia (expressing lysosomal and phagocytic components). In a Bax KO mouse lacking the significant apoptosis that usually occurs in the early postnatal retina, they find a decrease in the remodeling clusters of microglia, concluding that the lack of apoptosis in this mouse prevents the shift in microglial phenotype from homeostatic to remodeling. Several of these "remodeling" clusters in WT mice are relatively enriched after short term PLX3397 treatment, a CSF1R inhibitor commonly used to deplete microglia in mouse.

This paper provides interesting new phenotyping of retinal microglia and how they change with Bax KO and PLX treatment, but one central conclusion is currently not fully supported by the data provided. The central claim (and a primary new finding compared to a prior study from this group, Anderson et al. 2019, Cell Reports), is that the identified remodeling clusters are not only CSF1R independent, but more specifically that expression of remodeling genes in response to neuronal apoptosis confers this survival advantage.

Two major concerns currently prevent this claim from being supported by the data shown:

1. Lack of biological replicates for single cell data: unless I misinterpreted the methods, there appears to be a single biological replicate for each single cell experimental condition. The authors rely on their sing cell data to measure the abundance of various microglial populations in their models. In the absence of replicates, it is not possible to confidently measure population abundance within and between samples, because there is no way to make an estimate of the error present in cluster sizes. By analogy, a single biological replicate of flow cytometry analysis could not be used to reliably quantify the size of populations. This reviewer does not want to force the authors to double or triple their single cell seq efforts, but this technical limitation really prevents reliable comparison of the abundance of cells in a given cluster within and across conditions, and these comparisons form the central basis for their claims. If not by more biological replicates of single cell data, a potential way to address this is to validate and better quantify the finding of shifted macrophage states using another method to show that remodeling cells go away in BAX ko, and are enriched with PLX treatment, and a possibility would be to use tissue staining or in situ to do this. Or maybe facs with some remodelling cluster markers?

We sequenced a single technical replicate for each condition, but each consisted of several biological specimens (11-13 animals each) from multiple litters, totaling 49 animals. Pooling this large number of retinas was necessary to obtain enough cells for analysis, and consequently, we have many biological samples represented. In addition, the proportions of microglia expressing specific markers in our scRNAseq match our published quantification by in situ hybridization (Cell Reports 2019), giving us confidence that the proportions determined here are representative. For example, we previously found that in P7 WT retinas, 100% of microglia express Apoe and Lyz2 at varying levels, 65% Tmem119, and 25% Spp1 and these numbers are consistent with proportions identified by our scRNAseq. The same is true for both the Bax KO and PLX treated retinas. We thank the reviewer for pointing this out and have emphasized the number of animals/retinas used in the results (line 110-112), figure legend and methods (lines 856-860) and importantly, that these proportions match our previous quantification (results line 158-160).

2. Relationship between CSF1R independence and remodeling states: the authors conclude that remodeling states driven by neuronal apoptosis confer CSF1R independence. This conclusion is currently based on inferences from transcriptomic data about the functions of remodeling and PLX resistant microglia (eg that because they have a similar transcriptional profile, they are both engaged in remodeling), which is plausible but there are other equally plausible possibilities, such as that PLX directly induces a state change in microglia that does not involve neuronal apoptosis or remodeling. If the authors could directly show that surviving microglia are indeed engaged in some kind of a remodeling function, this would go a long way to support the claims. If not, careful adjustment of the claims and text would ensure that the claims are truly supported by the data shown, including a discussion of alternative explanations, and the limitations of transcriptomic inference to truly confirm that the "remodeling microglia" and the "plx resistant microglia" are a similar population of cells. This is particularly important because the major cluster in the plx3397 treated sample (cluster 3) is not present in any other condition.

We agree that function cannot be determined by transcriptomic data alone. The conclusion that remodeling states driven by neuronal apoptosis confer CSF1R independence is bolstered by our previous publication where we found that in the Bax KO animal, a larger proportion of microglia were dependent on CSF1R signaling (Cell Reports 2019) referenced in lines 482-484 of the discussion. We are currently exploring whether “resistant” populations are actively engaged in remodeling activities, which is being prepared for another publication. However, we agree that other possibilities exist for the overlap in PLX-resistant and remodeling clusters and have added alternative explanations in the discussion (line 492-495) and have qualified this conclusion in results (line 270).

Reviewer #3 (Recommendations for the authors):

Some questions listed below should however be addressed for clarifications:

1. The authors nicely identified 11 clusters of microglia in the developing postnatal retina. This raises the question of the spatial distribution of these microglia within the retina and the dynamics during development. Although this goes beyond the scope of the manuscript, the authors show in Supp Figure 5 by in situ hybridization that Ccl3+ microglia are dispersed in the different retinal layers, suggesting a regular mosaic distribution of cluster 3. This question of spatial distribution of microglia clusters could be raised in the Discussion section.

We agree that the spatial distribution of these various populations is of interest since it may help inform function. This is a focus of ongoing investigation. We previously examined the distribution of Apoe, Lyz2, Clec7a, Spp1, Igf1, Tmem119, and Itgax (Cell Reports 2019). While we found that these genes were dynamically regulated over development, a detailed localization analysis at one time point was not done. We agree this would be important and informative. We have added a sentence to the discussion (line 398-400).

2. On the right of cluster 1 in UMAP plots, there is a non-annotated blue cluster. What does it correspond to?

Those cells are members of cluster 6. We added a note in the figure legend of Figure 1.

3. The authors found that density was reduced by nearly half in Bax KO. What is their interpretation of this data? Does this result from less microglia proliferation?

The observation that microglial density is reduced in the Bax KO is an intriguing one. We currently do not know whether it is a result of less proliferation, but we do not see a change in the proportion of cycling microglia by scRNAseq in KO microglia compared to WT at P7 (Figure 3A,B). We added a sentence to the discussion (lines 403-407).

4. Statistical analysis would be useful in Figure 4D, Figure 3B and Figure 4B. Also, indicating ns on the graph for the 3 KO in Figure 6C would be helpful.

We have added statistical information in the figure and legend for Csf1r expression across clusters as well as a Chi-square test looking at the change in distribution between samples for Figures 3B and 4B. In addition, we have added “ns” for comparisons that are not significant to Figure 6C.

5. The authors conclude from their RNAseq analysis that there is no change in genes associated with survival pathways in Axl KO microglia compared to WT. However, they found that Spp1 is downregulated. Given that Osteopontin signaling is associated with survival in different cell types, why the authors do not consider it as a candidate?

We thank the reviewer for this excellent point and for identifying this oversight. We have added several sentences to the discussion highlighting this possibility (lines 519-524).

6. From their RNAseq analysis, the authors conclude that Mer and Axl are not required to drive microglia remodeling gene expression. However, can the authors really exclude the possibility of post-transcriptional regulation?

We focus here on a set of remodeling genes that we know can be transcriptionally regulated in our context as well as others. However, we agree that post-transcriptional regulation is likely involved in driving many microglial responses as well, and cannot rule out the involvement of Mer and Axl. We have added a sentence to the discussion (line 462-463).

7. Line 367, the authors conclude from their data that there are microglial states that are not regulated by neuronal death. Is it really the only interpretation of the data? Could it be a matter of threshold? If there were more apoptotic cells, would these microglial states still be not responsive?

Good point. We changed the text to reflect that an alternative possibility is that these clusters represent cells that have not had the opportunity to interact with dying cells (lines 420-421).

8. The authors mention in the Discussion section that one candidate signaling downstream Axl is the PI3K-Akt-NfkB-Bcl2 pathway. It is not clear whether this is supported by the RNAseq analysis from the present manuscript.

This pathway is primarily regulated post-transcriptionally and thus, we don’t expect to see changes at the gene expression level.

Author details

1. Sarah Rose Anderson

   Department of Neurobiology, University of Utah, Salt Lake City, United States

   Contribution

   Conceptualization, Formal analysis, Investigation, Methodology, Supervision, Validation, Visualization, Writing - original draft

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-7639-4064

2. Jacqueline M Roberts

   Department of Neurobiology, University of Utah, Salt Lake City, United States

   Contribution

   Data curation, Formal analysis, Investigation, Methodology, Software, Visualization, Writing – review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-1024-3976

3. Nathaniel Ghena

   1. Department of Neurobiology, University of Utah, Salt Lake City, United States
   2. Interdepartmental Program in Neuroscience, University of Utah, Salt Lake City, United States

   Contribution

   Formal analysis, Investigation, Visualization, Writing – review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-1890-8036

4. Emmalyn A Irvin

   Department of Neurobiology, University of Utah, Salt Lake City, United States

   Contribution

   Formal analysis, Investigation

   Competing interests

   No competing interests declared

5. Joon Schwakopf

   Department of Neurobiology, University of Utah, Salt Lake City, United States

   Contribution

   Formal analysis, Investigation

   Competing interests

   No competing interests declared

6. Isabelle B Cooperstein

   Department of Neurobiology, University of Utah, Salt Lake City, United States

   Contribution

   Formal analysis, Investigation

   Competing interests

   No competing interests declared

7. Alejandra Bosco

   Department of Neurobiology, University of Utah, Salt Lake City, United States

   Contribution

   Formal analysis, Investigation, Writing – review and editing

   Competing interests

   No competing interests declared

8. Monica L Vetter

   Department of Neurobiology, University of Utah, Salt Lake City, United States

   Contribution

   Conceptualization, Funding acquisition, Project administration, Supervision, Writing – review and editing

   For correspondence

   monica@neuro.utah.edu

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-6262-8504

• 1,686

  Page views

• 348

  Downloads

• 11

  Citations

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