bioRxiv preprint doi: https://doi.org/10.1101/2025.01.29.635500; this version posted February 1, 2025.

The copyright holder for this preprint

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

Simultaneous profiling of native-state proteomes and

transcriptomes of neural cell types using proximity
labeling
Christina C. Ramelow1-3, Eric B. Dammer2,6, Hailian Xiao1,2, Lihong Cheng2,4, Prateek Kumar5,
Claudia Espinosa-Garcia5, Maureen M. Sampson3, Ruth S. Nelson5, Sneha Malepati5, Dilpreet
Kour5, Rashmi Kumari5, Qi Guo2,6, Pritha Bagchi7, Duc M. Duong6,7, Nicholas T. Seyfried1,2,6,7,
Steven A. Sloan3,#, Srikant Rangaraju1,5,#
1
Department of Neurology, Emory University, Atlanta, GA
2
Center for Neurodegenerative Disease, Emory University
3
Department of Human Genetics, Emory University
4
Department of Pharmacology and Chemical Biology, Emory University
5
Department of Neurology, Yale University, New Haven, CT
6
Department of Biochemistry, Emory University
7
Emory Integrated Proteomics Core, Emory University
#
Co-senior authors
Proximity labeling | transcriptomics | proteomics | microglia | Introduction
astrocytes | neurons | mRNA-protein concordance
The transcriptome and proteome of a cell provide com-
Correspondence: S.R. ([email protected])
plementary molecular information about its phenotype,
S.A.S ([email protected])
state, and cellular functions. Quantifying RNA and pro-
tein levels in cell types using high-throughput -omics ap-
Abstract proaches uncover mechanisms and molecular path-
Deep molecular phenotyping of cells at transcriptomic ways that contribute to development, aging, injury, and
and proteomic levels is an essential first step to under- disease. RNA-sequencing can be leveraged to record
standing cellular contributions to development, aging, different transcript species, expression patterns, gene
injury, and disease. Since proteome and transcriptome structure, splicing patterns, and post-transcriptional
level abundances only modestly correlate with each modifications. On the other hand, mass spectrometry
other, complementary profiling of both is needed. We (MS)-based quantitative proteomics provides infor-
report a novel method called simultaneous protein and mation on protein abundances, localization, post-trans-
RNA -omics (SPARO) to capture the cell type-specific lational modifications, structure, cell-to-cell interactions,
transcriptome and proteome simultaneously from both and effector functions. Importantly, the number of RNA
in vitro and in vivo experimental model systems. This transcripts for a given gene does not always correlate
method leverages the ability of biotin ligase, TurboID, to with its protein abundance1,2. This is straightforward to
biotinylate cytosolic proteins including ribosomal and explain from a biological standpoint by considering dy-
RNA-binding proteins, which allows enrichment of bioti- namics of RNA stability, translational regulation, protein
nylated proteins for proteomics as well as protein-asso- half-life, chaperone proteins, cellular sequestration, as
ciated RNA for transcriptomics. We validated this ap- well as other post-transcriptional and post-translational
proach first using well-controlled in vitro systems to ver- events3–7. This contributes to observed discordances
ify that the proteomes and transcriptomes obtained re- between mRNA and protein abundances and molecular
flect the ground truth, bulk proteomes and transcrip- signatures in cells and tissues.
tomes. We also show that the effect of a biological stim- Measuring RNA and protein levels simultane-
ulus (e.g., neuroinflammatory activation by lipopolysac- ously can reveal changes at the RNA-level that are not
charide) can be faithfully captured. We also applied this reflected at the protein-level, and vice-versa, and ena-
approach to obtain native-state proteomes and tran- bles exploration of mechanisms regulating mRNA-pro-
scriptomes from two key neural cell types, astrocytes tein correlation. This can have biological implications as
and neurons, thereby validating the in vivo application well as highly practical consequences that impact
of SPARO. Next, we used these data to interrogate pro- choice of RNA or protein markers in experiments (e.g.,
tein-mRNA concordance and discordance across these choice of RNA probe for in situ hybridization or choice
cell types, providing insights into groups of molecular of protein marker for spatial studies). Also, the level of
processes that exhibit uniform or cell type-specific pat- discordance between mRNA and protein, and the
terns of mRNA-protein discordance. mechanisms that control mRNA and protein levels may
vary across different cell types, tissues, and biological
contexts, and cannot be ascertained from bulk tissue -
omics approaches. Therefore, we need new
 bioRxiv preprint doi: https://doi.org/10.1101/2025.01.29.635500; this version posted February 1, 2025. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

approaches to jointly measure mRNA and protein levels the RiboTag approach34,35. Therefore, we hypothesized
in cell type-specific contexts. that transcripts associated with these proteins could
Several attempts have been made to quantify also be purified simultaneously with biotinylated pro-
this general concordance (or discordance) between teins. We developed and validated the SPARO ap-
transcriptomic and proteomic levels2,8–16. Simultaneous proach to capture cell type-specific transcriptomes and
sampling of the transcriptome and proteome can be proteomes using a microglial cell line in vitro, and neu-
readily accomplished under in vitro monoculture condi- rons and astrocytes in vivo. This allowed us to investi-
tions. However, under in vivo settings, these methods gate the concordance and discordance between mRNA
involve collecting RNA and protein from separate phys- and protein levels, and to examine how patterns of
ical samples and rely on cell type purification using me- mRNA-protein discordance are conserved or vary
chanical approaches, which, by themselves, impact across cell types. SPARO represents a novel approach
transcriptomic and molecular characteristics of the to simultaneously quantify RNA and protein profiles with
cells17. The ability to purify intact cells from complex tis- potentials for broad applications in vitro and in vivo.
sues like the brain also varies across cell type. For ex-
ample, adult neurons are difficult to isolate without loss Results
of their cellular and synaptic architecture18,19.
A general approach to obtain cell type-specific Validation of SPARO in BV2-TurboID cells in vitro
transcriptomes or proteomes from tissues that are inde- To test whether we could leverage TurboID-based cy-
pendent of cell type isolation involves protein tagging. tosolic proteomic labeling to simultaneously capture
cellular proteomes and transcriptomes, we first used
For cell type-specific in vivo transcriptomics, ribosomal
our previously validated, stably transduced mouse mi-
subunits (e.g., Rpl22) can be tagged (e.g., HA) in a cell
type-specific manner using Cre/lox genetics, allowing croglia BV2 cell line that expresses V5-TurboID-NES
capture of ribosome-associated mRNA species (trans- (BV2-TurboID)35,49. In this construct, the TurboID biotin
latome). For cell type-specific proteomics, metabolic la- ligase is restricted to the extranuclear compartment via
a nuclear export sequence (NES) to bias towards bioti-
beling with non-canonical amino acids20–25 and proxim-
ity-labeling by biotin ligases26–37 have been employed. nylation of the cytosolic proteome (Fig. 1a). By starting
There have also been several efforts to achieve joint with a homogeneous cell line, we could quantitatively
mRNA and protein measurements, using RiboTag38–41 assess how the transcriptome and proteome of BV2
and Translating Ribosome Affinity Purification cells using the SPARO approach compare with whole-
(TRAP)42,43 to capture mRNAs and nascent peptides cell transcriptomes and proteomes from bulk cell ly-
from a single tagged ribosomal protein under a cell sates. Furthermore, to test whether SPARO could also
capture cellular changes in microglia induced by an in-
type-specific promoter. However, these approaches
flammatory stimulus, we treated BV2 control and BV2-
limit profiling to only ribosome-bound transcripts and
nascent peptides rather than the broader proteome and TurboID cells with lipopolysaccharide (LPS) (Fig. 1b).
transcriptome of the cell. Further, mammalian cells con- We lysed cells in a buffer that maintains RNA-protein
tain at least 79 different ribosomal proteins for cytosolic interactions and then processed for simultaneous tran-
scriptomics (mRNA-seq) and proteomics (label-free
translation44, and ribosomes with different protein com-
positions translate different groups of mRNAs45. There- quantitative mass spectrometry or LFQ-MS). Immunob-
fore, ribosome affinity purification-based methods that lot analyses confirmed the presence of biotinylated pro-
target a single ribosomal protein may not capture the teins (via streptavidin) and the TurboID protein (via V5)
full complexity of the RNA species within a cell. in BV2-TurboID cells, which were absent in non-Tur-
Another method to achieve joint mRNA and boID control cells both before (Supplementary Fig. 1a)
and after streptavidin bead affinity purification (hence-
protein measurements is Single-Cell Protein And RNA
Co-profiling (SPARC)46, which uses poly(A) oligo-dTs forth “pulldown”) (Fig. 1c). We detected only endoge-
conjugated to magnetic beads that hybridize to mRNAs nously biotinylated proteins in the BV2 control cells
in conjunction with a homogeneous protein extension (Fig. 1c). After streptavidin pulldown, we confirmed pro-
tein enrichment using a silver stain (Supplementary
assay (PEA)47,48. SPARC, however, is limited by the 96-
Fig. 1b). RNA gel electrophoresis showed comparably
plex capability of the targeted proteins of the PEA as-
say. Due to the limitations of current methods, new tools high levels of RNA in bulk BV2 lysates (Supplementary
for complementary molecular profiling of cell type-spe- Fig. 1c). Importantly, we found high levels of RNA in
cific transcriptomes and proteomes are needed to gain BV2-TurboID pulldowns but negligible RNA in BV2 con-
a more global characterization of cell types in their na- trol cell pulldowns without TurboID (Fig. 1d), confirming
tive-states in a complex tissue environment. that RNA is enriched only when proteomic biotinylation
To develop an approach for simultaneous pro- occurs in BV2 microglia.
We next performed LFQ-MS and RNA-seq on
tein and RNA omics (SPARO), we leveraged the biotin
ligase, TurboID. Using MS-based proteomics, we previ- bulk BV2-TurboID samples and pulldowns. Raw inten-
ously found that cytosolic TurboID biotinylates many sity and processed intensity values are found in Supple-
RNA-binding proteins (RBPs) and ribosomal proteins mentary Data 2 and 3, respectively. Raw RNA-seq and
processed counts values are found in Supplementary
both in vitro and in vivo, including Rpl22 that is used in
Data 4 and 5, respectively. We first assessed the
 bioRxiv preprint doi: https://doi.org/10.1101/2025.01.29.635500; this version posted February 1, 2025. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

Figure 1: Simultaneous transcriptomics and proteomics profiling of BV2 microglia under homeostatic and LPS-stimulated conditions
a. Schematic for leveraging cytosolic TurboID to create a simultaneous protein and RNA -omics (SPARO) approach. b. Experimental design: BV2 control and BV2-
TurboID cells (n = 3 biological replicates/genotype/condition) were treated with 100 ng/mL of LPS for 48 hours and 200 μM of biotin for 24 hours during the second
day of LPS treatment. c. Immunoblot visualization of biotinylated proteins from BV2-TurboID cells at the pulldown level not seen in non-TurboID control cells (probed
with streptavidin-680). Presence of TurboID recombinant protein (via V5 tag) was also detected at the pulldown level. d. RNA gel electrophoresis of RNA eluted off
biotinylated proteins after streptavidin pulldown. High levels of rRNA and mRNA were detected in BV2-TurboID cells, but not seen in non-TurboID control cells. e.
Venn diagram representation of protein numbers identified by LFQ-MS present in the BV2-TurboID bulk and pulldown proteomes without LPS treatment (top) and
 bioRxiv preprint doi: https://doi.org/10.1101/2025.01.29.635500; this version posted February 1, 2025. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

with LPS treatment (bottom). f. Venn diagram representation of RNA quantities identified by RNA-seq present in the BV2-TurboID bulk and pulldown transcriptomes
without LPS treatment (top) and with LPS treatment (bottom). g. MDS plot of normalized LFQ-MS intensity data (left) in each BV2-TurboID proteome group. h.
Correlation analysis of protein abundances from normalized LFQ-MS mean intensity values between the BV2-TurboID bulk and pulldown proteomes (n = 1455
proteins, Pearson correlation coefficient r = 0.30). i. Correlation analysis of RNA abundances from normalized RNA-seq mean count values between the BV2-TurboID
bulk and pulldown transcriptomes (n = 12742 genes, Pearson correlation coefficient r = 0.95). j. Volcano plot representation of DEPs between non-LPS and LPS-
treated BV2-TurboID cells at the pulldown level (n=3/group). Black symbols (two-sided t-test, p ≤ 0.05 and ≥ 1-fold change, n = 99) represent DEPs in non-LPS-treated
BV2-TurboID cells. Green symbols (two-sided t-test, p ≤ 0.05 and ≥ 1-fold change, n = 292) represent DEPs in LPS-treated BV2-TurboID cells. k. Proteome GO
analysis of DEPs visualized in panel j, showing overrepresented ontologies in non-LPS treated (black) and LPS-treated (green) BV2-TurboID cells at the pulldown
level. l. Correlation analysis of all shared DEPs (two-sided t-test, p ≤ 0.05) between the LPS-treated BV2-TurboID cells at the bulk and pulldown level (n = 644 proteins,
Pearson correlation coefficient r = 0.34). Green symbols represent highly enriched shared DEPs (two-sided t-test, p ≤ 0.05 and ≥ 1-fold change, n = 83, Pearson
correlation coefficient r = 0.68). Examples of pro-inflammatory and anti-inflammatory canonical microglia markers labeled and in bold. m. Volcano plot representation
of differentially expressed genes (DEGs) between non-LPS and LPS-treated BV2-TurboID cells at the pulldown level (n=3/group). Black symbols (two-sided t-test, p
≤ 0.05 and ≥ 1-fold change, n = 418) represent DEGs enriched in non-LPS-treated BV2-TurboID cells. Green symbols (two-sided t-test, p ≤ 0.05 and ≥ 1-fold change,
n = 820) represent DEGs enriched in LPS-treated BV2-TurboID cells. n. Transcriptome GO analysis of DEGs visualized in panel m. of overrepresented ontologies of
non-LPS treated black) and LPS-treated BV2-TurboID cells (green scale) at the pulldown level. o. Correlation analysis of all shared DEGs (two-sided t-test, p ≤ 0.05)
between the LPS-treated BV2-TurboID cells at the bulk and pulldown level (n = 6798 genes, Pearson correlation coefficient r = 0.82). Green symbols represent highly
enriched shared DEGs (two-sided t-test, p ≤ 0.05 and ≥ 1-fold change, n = 999, Pearson correlation coefficient r = 0.89). Examples of pro-inflammatory and anti-
inflammatory canonical microglia markers are labeled and in bold.

overlap of the number of proteins between the TurboID inability of cytosolic TurboID to label intra-mitochondrial
pulldowns and the bulk proteome. We observed 44% proteins and their associated transcripts. At the proteo-
(untreated) and 51% (LPS-treated) overlap of TurboID mic level, the bulk proteome was enriched in nuclear
pulldown proteomes with the ground truth bulk BV2 pro- and mRNA-splicing related terms, consistent with un-
teomes, fairly consistent with prior observations in BV2- der-sampling of the nuclear proteome by TurboID label-
TurboID cells35. This is likely a result of the NES tag on ing (p ≤ 0.05, ≥ 1-fold change, Supplementary Fig. 3a).
TurboID that biases the proteome towards the cytosolic SPARO also detected proteomic and tran-
compartment and against the nuclear and mitochondrial scriptomic changes driven by LPS treatment (Fig. 1j
compartments (Fig. 1e). Next, to benchmark the fidelity and m, Supplementary Fig. 2 and 3). As expected,
of our TurboID pulldown transcriptome, we compared GO analysis of differentially enriched proteins (DEPs)
our RNA-seq datasets between the bulk BV2 samples between TurboID pulldowns revealed overrepresented
and our pulldown populations. We observed between ontologies of LPS-treated cells including response to
95% (untreated) and 98% (LPS-treated) overlap of Tur- chemokine, defense response to symbiont and cytoki-
boID pulldown transcriptomes with the ground truth bulk nesis (Fig. 1k, Supplementary Data 8). The changes
BV2 transcriptome (Fig. 1f). We next asked whether the that we observed upon LPS treatment measured by
pulldown proteomes and transcriptomes could ade- TurboID pulldown were still modestly correlated with
quately capture changes in BV2 cells upon exposure to LPS changes in the bulk proteome (Pearson correlation
LPS (Fig. 1g, left). Multidimensional scaling (MDS) coefficient r = 0.34, n = 644 shared proteins, p ≤ 0.05,
analysis of the transcriptomes identified distinct un- Fig. 1l). However, this correlation was substantially im-
treated and LPS-treated transcriptomes regardless of proved when specifically considering proteins that ex-
sample type (bulk vs pulldown) (Fig. 1g, right) suggest- hibit high differential abundance between LPS-treated
ing that LPS-effect was equally captured by the TurboID and control BV2-TurboID cells (Pearson correlation co-
pulldown and bulk transcriptomes. efficient r = 0.83, n = 83 shared proteins with p ≤ 0.05
To test the fidelity of the pulldowns to their re- and ≤ -1 or ≥ 1-fold change, Fig. 1l). We also found ca-
spective bulk samples, we next performed correlation nonical pro-inflammatory microglial activation markers
analysis of our LFQ-MS and RNA-seq data, respec- (e.g., Ikbke, Nfkb2, and Ehd1) enriched in the LPS-
tively. At the proteomic level, we observed modest cor- treated BV2-TurboID pulldowns and anti-inflammatory
relation between LPS-treated pulldown and the bulk microglial activation markers (e.g., Mrc1/Cd206, Mgl2
LPS-treated BV2-TurboID proteomes (Pearson correla- and Lgals3) enriched in the control pulldowns (p ≤ 0.05,
tion coefficient r = 0.30, n = 1455 shared proteins, Fig. ≥ 1-fold change, Fig.1l).
1h). Notably, we observed a much higher correlation Similarly to the proteome, we performed differ-
between the LPS-treated pulldown transcriptomes and ential enrichment and GO analyses of differentially ex-
bulk BV2-TurboID transcriptomes (Pearson correlation pressed genes (DEGs) in the LPS-treated and non-LPS
coefficient r = 0.95, n = 12742 shared transcripts, Fig. treated pulldown transcriptomes (Supplementary Data
1i). We also compared the correlation between the un- 9). We observed the same canonical reactive microglial
treated bulk and pulldown proteomes and transcrip- response enriched in LPS treated and untreated sam-
tomes, respectively (Supplementary Fig. 1e and h). ples (Fig. 1m-o). Unlike at the protein level, the LPS-
To confirm the cytosolic bias of the V5-TurboID-NES re- driven transcriptomic changes correlated highly be-
combinant protein, we differentially compared the pull- tween the TurboID RNA pulldown and the LPS-en-
down and bulk BV2-TurboID transcriptomes and prote- riched bulk transcriptome (Pearson correlation coeffi-
omes. Differential enrichment analysis of BV2 tran- cient r = 0.82, n = 6798 shared transcripts with p ≤ 0.05,
scriptomic and proteomic data are found in Supplemen- Fig. 1o). This correlation was even more improved
tary Data 6 and 7, respectively. At the transcriptomic when specifically considering genes that exhibit high
level, we found that mitochondrial-encoded genes (e.g., differential enrichment between LPS-treated and con-
COX1, CYTB and ND2) were highly enriched in the bulk trol BV2-TurboID cells (Pearson correlation coefficient r
samples (p ≤ 0.05, ≥ 1-fold change, Supplementary = 0.89, n = 999 shared genes, p ≤ 0.05 and ≤ -1 or ≥ 1-
Fig. 2a) as compared to the pulldowns, consistent with fold change, Fig. 1o). Additional transcriptomic and
 bioRxiv preprint doi: https://doi.org/10.1101/2025.01.29.635500; this version posted February 1, 2025. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

proteomic GO analysis data comparing either LPS- negligible RNA in non-TurboID controls (Fig. 2d), con-
treated versus untreated bulk or pulldown samples and firming transcriptome enrichment only in the presence
pulldown versus bulk samples can be found in Supple- of biotinylation. We next prepared samples for LFQ-MS.
mental Data 10-15. Raw intensity and processed intensity values are found
Overall, these results suggest that SPARO can in Supplementary Data 16-18, respectively. Notably,
simultaneously capture the cellular transcriptome and MDS analysis of the proteomes showed separation be-
proteome and effectively captures inflammatory effects tween neuronal and astrocytic proteomes (Fig. 2e). We
of LPS in an in vitro mammalian system. also confirmed the cell type specificity of the proteomes
including enrichment of canonical astrocytic proteins
Validation of SPARO in vivo using cortical astrocytes e.g., Hepacam, Glu, Aqp4, Plpp3, p ≤ 0.05, ≥ 1-fold
and neurons change) and GO terms (e.g., astrocyte end-foot, glial
We next wanted to test whether we could capture cell cell projection and astrocyte differentiation) in astrocytic
type-specific transcriptomes and proteomes simultane- pulldowns and neuronal proteins (e.g., Map2, Ncam1,
ously in vivo while retaining the native state of these Mapt, p ≤ 0.05, ≥ 1-fold change) and GO terms (e.g.,
cells. We crossed Rosa26TurboID/wt (TurboID) mice with synaptic signaling dendrite and neuron projection,) in
appropriate inducible Cre-ERT2 mice, to drive Cre ex- neuronal pulldowns (Fig. 2f and g, Supplementary
pression specifically in Camk2a-expressing excitatory Data 19 and 20). These data are consistent with our
neurons or in Aldh1l1-expressing astrocytes. We chose prior work using astrocyte-TurboID and neuron-TurboID
the Aldh1l1-Cre/ERT2 mice and Camk2a-Cre/ERT2 systems34. It is important to highlight that the lysis buffer
mouse lines for their extensive validation and non-leak- used in the current studies was a homogenization buffer
iness of Cre40,50. Our breeding scheme resulted in the intended to maintain RNA-protein interactions, as op-
generation of Aldh1l1CreERT2/wt/Rosa26TurboID/wt (astro- posed to 8 M urea lysis buffer in prior work34.
cyte-TurboID) and Camk2aCreERT2/wt/Rosa26TurboID/wt Like the proteome, we conducted MDS, differ-
(neuron-TurboID) mice. We used Aldh1l1CreERT2/wt or ential enrichment and GO analysis of the neuronal and
Camk2aCreERT2/wt as Cre-only controls. We induced Cre- astrocytic transcriptomes. Raw RNA-seq and pro-
mediated recombination by intraperitoneal tamoxifen at cessed counts values are found in Supplementary Data
7 weeks of age, followed by a 3-week gap, and then 21 and 22, respectively. Like the proteome, MDS anal-
biotin water supplementation for 2 weeks (Fig. 2a). ysis identified distinct neuronal and astrocytic transcrip-
Based on our prior study, V5-TurboID-NES presence tomes (Fig. 2h), each of which were enriched for ca-
and biotin treatment does not have adverse effects on nonical cell type specific markers (Fig. 2i and j, Supple-
the mice and does not impact homeostatic cellular func- mentary Data 23 and 24), suggesting that the pulldowns
tions29,31,34. Following biotin treatment (age 2.5-3 are cell type enriched. To further validate the cell type
months), we harvested and lysed cortical tissue in the specificity of the pulldown transcriptomes, we com-
same buffer as the in vitro studies to maintain RNA-pro- pared the relative gene expression of top astrocytic and
tein interactions and then processed for simultaneous neuronal genes from Zhang et al., 201451 between the
transcriptomics and proteomics using SPARO (Fig. 2a). bulk cortex and the cell type enriched pulldowns (Sup-
Immunofluorescent imaging of the cortex showed ro- plementary Fig. 5). The cell type enriched gene lists
bust biotinylation in TurboID brains compared to con- can be found in Supplementary Data 25. The astrocyte-
trols (Fig. 2b and Supplementary Fig. 4f). Moreover, TurboID pulldown transcriptomes exhibited a high ex-
the biotinylation signal co-localized with the astrocytic pression of astrocytic genes (e.g., Tnc, Sox9 and Aqp4)
marker, S100b (in astrocyte-TurboID mice) and neu- and a low expression of neuronal genes (e.g.,
ronal marker, b-III-tubulin (in neuron-TurboID mice) Tmem130, Slc10a4 and Npy) (Supplementary Fig.
confirming cell targeting (Fig. 2b). Importantly, our prior 5a), and the converse was true for the neuron-TurboID
work validated that the biotinylation signal present in as- pulldown transcriptomes (Supplementary Fig. 5b).
trocyte- and neuron-TurboID animals is cell type-spe- We included a 5-day repeated LPS paradigm in
cific via the absence of signal in other brain cell types, our original astrocyte SPARO study to examine effects
and without reactive gliosis34. Immunoblot analysis of of LPS-induced neuroinflammation of astrocyte profiles
cortical lysates at the bulk level (Supplementary Fig. (Supplementary Fig. 6). We observed modest effects
4a and b) and after streptavidin affinity purification dis- of LPS at both the proteome (p ≤ 0.05, ≥ 1-fold change,
played biotinylated proteins from astrocyte-TurboID and n = 11 up with LPS, n = 35 up in non-LPS controls, Sup-
neuron-TurboID pulldowns not seen in non-TurboID plemental Data 18) and transcriptome levels (p ≤ 0.05,
controls (Fig. 2c). The presence of the TurboID recom- ≥ 1-fold change, n = 141 up with LPS, n = 41 up in non-
binant protein via V5 presence was also only present in LPS controls, Supplemental Data 21), suggesting that
astrocyte-TurboID and neuron-TurboID pulldowns and the LPS effect may have been partially washed out dur-
not non-TurboID controls (Fig. 2c) providing biochemi- ing the one-week interval between the last LPS dose
cal verification of cell type-specificity. and euthanasia (Supplementary Fig. 6b and c). De-
Following streptavidin pulldown, we determined spite this, we observed modest yet meaningful effects
protein enrichment using a silver stain (Supplementary of LPS on the astrocyte pulldown transcriptome using
Fig. 4c and d) and eluted high-quality RNA in both as- the SPARO approach, including upregulation of
trocyte-TurboID and neuron-TurboID cortical tissue, but canonical reactive genes (e.g., Lcn2 and Serpina3g,
Supplementary Fig. 6b). We also performed GO
 bioRxiv preprint doi: https://doi.org/10.1101/2025.01.29.635500; this version posted February 1, 2025. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

Figure 2: TurboID-based dual transcriptomic and proteomic analysis of cortical Aldh1l1-expressing astrocytes and Camk2a-expressing neurons in vivo.
a. Experimental overview of dual affinity purification of RNA and protein using TurboID proximity labeling in Aldh1l1CreERT2/wt or Camk2aCreERT2/wt (control, n = 2 mice),
Aldh1l1CreERT2/wt/Rosa26TurboID/wt (astrocyte-TurboID, n = 3 mice) and Camk2aCreERT2/wt/Rosa26TurboID/wt (neuron-TurboID, n = 3 mice) mouse models. Astrocyte-TurboID,
neuron-TurboID and control mice received tamoxifen (75 mg/kg/day via intraperitoneal (i.p.) injections x 5 days). After 3 weeks of Cre recombination, mice received
biotin-containing water (37.5 mg/L) for two weeks. Cortical tissue was isolated for streptavidin affinity purification, followed by LFQ-MS and RNA-sequencing. b.
Representative immunofluorescence images from sagittal brain sections of control, astrocyte-TurboID and neuron-TurboID mice (n = 3 mice/experimental group)
confirms astrocyte and neuron specific biotinylation via streptavidin-488 (green) overlapping with astrocytic marker S100b (red) and neuronal marker b-III-tubulin (red),
respectively. Scale bar = 50 μm. c. Immunoblot visualization of cortical lysates after streptavidin affinity purification displaying biotinylated proteins from astrocyte-
TurboID and neuronal-TurboID pulldowns not seen in non-TurboID controls (probed with streptavidin-680). The presence of the TurboID (via V5) was only present in
astrocyte-TurboID and neuron-TurboID cortical pulldown tissue but not non-TurboID controls. d. RNA gel electrophoresis of RNA eluted off biotinylated proteins after
streptavidin pulldown. High levels of rRNA and mRNA were detected in astrocyte-TurboID and neuronal-TurboID cortical tissue, not seen in non-TurboID controls. e.
 bioRxiv preprint doi: https://doi.org/10.1101/2025.01.29.635500; this version posted February 1, 2025. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

MDS plot of normalized LFQ-MS intensity data in astrocyte-TurboID and neuronal-TurboID pulldowns. f. Volcano plot representation of DEPs between astrocyte-
TurboID and neuron-TurboID pulldowns (n=3/group). Blue symbols (two-sided t-test, p ≤ 0.05 and ≥ 1-fold change, n = 95) represent pulldown proteins enriched in
neuron-TurboID cortical samples. Purple symbols (two-sided t-test, p ≤ 0.05 and ≥ 1-fold change, n = 139) represent pulldown proteins enriched in astrocyte-TurboID
cortical samples. g. Proteome GO analysis of DEPs visualized in panel f. of overrepresented ontologies of neuron-TurboID (blue) and astrocyte-TurboID (purple)
pulldowns, highlighting ontologies related to neuronal (e.g., synaptic signaling and neuron projection) and astrocyte function (fatty acid synthesis and glial cell projec-
tion), respectively. h. MDS analysis of normalized RNA-seq count data in astrocyte-TurboID and neuron-TurboID pulldowns. i. Volcano plot representation of DEGs
between astrocyte-TurboID and neuron-TurboID pulldowns (n=3/group). Blue symbols (two-sided t-test, p ≤ 0.05 and ≥ 1-fold change, n = 1887) represent pulldown
proteins enriched in neuron-TurboID cortex. Purple symbols (two-sided t-test, p ≤ 0.05 and ≥ 1-fold change, n = 679) represent pulldown proteins enriched in astrocyte-
TurboID cortex. j. Transcriptome GO analysis of DEGs visualized in panel i. of overrepresented ontologies of neuron-TurboID (blue scale) and astrocyte-TurboID
(purple scale) pulldowns, highlighting ontologies related to neuronal (e.g., neurotransmitter receptor activity and neuron to neuron synapse) and astrocyte function
(e.g., lipid binding and astrocyte end-foot), respectively.

analysis of DEGs and found synapse- and translation- between the astrocyte-TurboID and astrocyte-RiboTag
related terms enriched with LPS treatment (Supple- pulldowns was 95% (Fig. 3c). The cell type specificity
mentary Fig. 6d, Supplementary Data 26). In contrast, of RiboTag and SPARO transcriptomes (Supplemen-
LPS-induced changes were minimal at the level of the tary Fig. 7c), as well as the correlation across datasets
astrocyte proteome (Supplementary Fig. 6c), suggest- (Pearson correlation coefficient r = 0.93, n = 17050
ing a discordance between LPS-induced effects on as- shared transcripts, Fig. 3d) were both robust. Next, we
trocytes at the transcriptomic and proteomic levels. We examined whether highly cell type-specific astrocytic
did not find any GO terms enriched from DEPs, due to genes from Zhang et al., 2014 were correlated across
the small number of DEPs found within each sample the TurboID- and RiboTag-enriched transcriptomes.
group. We found that the mRNA abundances were highly cor-
These results demonstrate the application of related (Pearson correlation coefficient r = 0.69, n = 50
the SPARO approach to obtain biologically meaningful genes, Fig. 3e).
proteomes and transcriptomes of astrocytes and neu- We also performed differential enrichment
rons while retaining their native states in adult mouse analysis and GO analysis between the astrocyte-Tur-
cortex. boID and astrocyte-RiboTag pulldown transcriptomes
(Supplementary Fig. 7d-f, Supplementary Data 29
The SPARO- and RiboTag-enriched astrocytic tran- and 30) and identified several differences in signatures.
scriptomes are similar We hypothesize that these differences arise from the
We next benchmarked SPARO against RiboTag, a fact that RiboTag enriches all RNAs bound to ribosomal
gold-standard in vivo cell type-specific transcriptomic complexes that contain the ribosomal protein, Rpl22,
profiling method38,39. RiboTag mice express a hemag- whereas the SPARO approach captures mRNAs inter-
glutinin tag (HA) modified allele of the Rpl22 gene acting with many biotinylated proteins that interact with
(Rpl22-HA), a major component of the polyribosome RNA in astrocytes, which include 66 ribosomal proteins
complex38,39. In the presence of Cre recombinase, HA- (including Rpl22) (Fig. 3f) and 67 RBPs (Supplemen-
tagged polyribosomes can be isolated from target cell tary Data 31).
types using an anti-HA antibody and immunoprecipita-
tion methods. The mRNAs interacting with the ribo- Correlation of the SPARO transcriptomes and proteo-
somes during translation can then be extracted and mes of cortical astrocytes and neurons in vivo
quantified to obtain a snapshot of actively translated We next evaluated the concordance between the in vivo
transcripts (translatome) in specific cell types. To as- proteomes and transcriptomes of astrocytes and neu-
sess the similarity between the TurboID-enriched and rons obtained using SPARO (Fig. 4a). The raw correla-
RiboTag-enriched transcriptomes, we compared both tion between 1,934 mRNA and protein pairs was mod-
methods using cortical Aldh1l1-expressing astrocytes in est in both astrocytes (Pearson correlation coefficient r
vivo (Fig. 3a). We bred Aldh1l1CreERT2/wt mice with = 0.27, Supplementary Fig. 8a, Supplementary Data
Rpl22tm1.1Psam or TurboID mice to generate astrocyte-Ri- 32) and neurons (Pearson correlation coefficient r =
boTag and astrocyte-TurboID mice, respectively. We 0.26, Supplementary Fig. 8b, Supplementary Data
administered tamoxifen to astrocyte-TurboID, astro- 33). To determine if these generally low correlations
cyte-RiboTag and Aldh1l1CreERT2/wt or Rpl22tm1.1Psam were simply a product of the complexity of an in vivo
(control) mice at 7 weeks of age, followed by a 3-week brain environment, we also performed similar transcrip-
gap, then performed cortical tissue lysis, pulldown of bi- tome and proteome correlation analysis of the in vitro
otinylated or HA-tagged proteins, and RNA-seq (Fig. BV2-TurboID cells and found similar results (Pearson
3a). correlation coefficient r = 0.24, Supplementary Fig. 8c,
Immunofluorescent imaging of the cortex Supplementary Data 34). We did observe the highest
showed the presence of HA recombinant protein in as- concordance between bulk cortical transcriptomes and
trocyte-RiboTag brains, which co-localized with the as- proteomes (Pearson correlation coefficient r = 0.47,
trocytic marker, S100b (Fig. 3b). RNA extractions were Supplementary Fig. 8d, Supplementary Data 35),
of similarly high quality across all SPARO and RiboTag which lack any cell type specificity.
pulldowns, with low levels seen in the RiboTag only Quantifying concordance is further complicated
controls (Supplementary Fig. 7a and b). We next pre- by innate technical differences between LFQ-MS and
pared samples for RNA-seq. Raw RNA-seq and pro- RNA-sequencing. For example, the dynamic range and
cessed counts values are found in Supplementary Data sensitivity of both approaches differ vastly. Therefore,
27 and 28, respectively. Importantly, transcript overlap to better normalize against these biases, we calculated
 bioRxiv preprint doi: https://doi.org/10.1101/2025.01.29.635500; this version posted February 1, 2025. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.
 bioRxiv preprint doi: https://doi.org/10.1101/2025.01.29.635500; this version posted February 1, 2025. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

Figure 3: The TurboID-enriched transcriptome is comparable to the RiboTag-enriched transcriptome of cortical astrocytes.
a. Experimental overview to compare the TurboID-enriched astrocyte transcriptome to the RiboTag-enriched astrocyte transcriptome. Aldh1l1CreERT2/wt or Rpl22tm1.1Psam
(control, n = 3 mice), Aldh1l1CreERT2/wt/Rosa26TurboID/wt (astrocyte-TurboID, n = 3 mice) and Aldh1l1CreERT2/wt/ Rpl22tm1.1Psam (astrocyte-RiboTag, n = 3 mice) mouse models
(aged 2.5-3.5 months). Astrocyte-TurboID, astrocyte-RiboTag and control mice received tamoxifen (75 mg/kg/day via intraperitoneal (i.p.) injections x 5 days). After 3
weeks astrocyte-TurboID mice received biotin-containing water (37.5 mg/L) for two weeks. Cortical tissue was isolated and lysed for affinity purification of either
biotinylated ribosomal and RNA-binding proteins from astrocyte-TurboID mice, or HA-tagged ribosomal protein Rpl22, followed by RNA extraction and RNA-sequenc-
ing. b. Representative immunofluorescence images of Aldh1l1CreERT2/wt control and astrocyte-RiboTag mice sagittal brain sections (n = 2-3 mice/experimental group)
confirms the presence of the HA recombinant protein (green) in the astrocyte-RiboTag mouse cortex and not the controls. The HA tag colocalizes with the astrocytic
marker, S100b (red), confirming astrocytic specificity. Scalebar = 50 μm. c. Venn diagram representation of mRNA numbers identified by RNA-seq present in the
TurboID-enriched and RiboTag-enriched astrocyte pulldown transcriptomes. d. Scatter plot visualization comparing the expression of 17,050 transcripts present in
the TurboID (x-axis) or RiboTag (y-axis) pulldown transcriptomes (Pearson correlation coefficient r = 0.93). Genes that are not present in both lists were removed from
the analysis. e. Scatter plot visualization comparing the expression of the top 50 postnatal day 7 (P7) astrocyte-specific transcripts enriched in the TurboID (x-axis) or
RiboTag (y-axis) pulldown transcriptomes after bulk RNA subtraction (Pearson correlation coefficient r = 0.69). The astrocyte-specific gene lists were acquired from
Zhang et al., 2014. f. Heatmap visualization displaying the relative abundance (via Row Z score, purple = relative high abundance or enrichment and blue = relative
low abundance or depletion) of ribosomal proteins (n = 65) present in the Aldh1l1CreERT2/wt (Cre only) and astrocyte-TurboID pulldowns.

percentile ranked abundances for proteins and RNA, high protein) were present in both cell types (Sup-
transcripts in each dataset (1,934 gene and protein plementary Fig. 8i). Many discordant mRNA-protein
pairs that overlap) and classified these mRNA-protein pairs were also shared across cell types (Fig. 4f and
pairs into distinct groups based on level of mRNA- g). Interestingly, we observed a pattern in which the
protein rank discordance (Fig. 4b and Supplementary types/categories of discordant genes were generally
Fig. 8e, Supplementary Data 32 and 33). In both astro- conserved across cell types, with specific subtypes of
cytes and neurons, and in the bulk cortex, we observed genes that are cell type specific. For example, within the
that most mRNA-protein pairs fell into either concord- discordant group where protein exceeded mRNA (pro-
antly low or high groups (Log2 % rank values between tein >> mRNA) abundance, we found overrepresented
75 and 100 for both the x and y-axis). Together, these ontologies of cytoskeletal pathways in both astrocytes
findings verify modest levels of concordance between and neurons. However, within astrocytes, only terms re-
protein and mRNA abundances52–54, although mRNAs lated to microtubules were enriched, as opposed to ac-
that tend to be highly expressed in neurons and astro- tin machinery in neurons (Fig. 4f). For the discordant
cytes, and in the cortex in general, also tend to be highly group of genes in which mRNA exceeded protein
abundant at the proteomic level. (mRNA >> protein) abundance, we found a consistent
We next classified mRNA and protein pairs signature in both neurons and astrocytes of ontologies
from the astrocyte and neuronal pulldowns into either related to mitochondria. However, within astrocytes this
concordant (mRNA ~ protein, > 0.75 in black) or dis- signature comprised of genes involved in aerobic respi-
cordant (protein >> mRNA, > - 0.50 in blue, or mRNA ration, whereas neurons were biased towards genes in-
<< protein levels, > 0.50 in red) groups based on a rank volved with mitochondrial fusion and GTPase activity
differential of abundance values (Supplementary Fig. (Fig. 4g). This pattern most likely reflects the metabolic
8f, Supplementary Data 32 and 33). We wondered differences that exist between these two cell types. Af-
whether there were examples of canonical cell type- ter performing GO analyses on the mRNA and protein
specific markers that fall into either highly concordant or pairs in each concordant and discordant quadrant, we
discordant quadrants. To designate astrocytic and neu- also looked at the overlap of the GO terms across the
ronal enriched canonical markers, we used a unionized two cell types. The cell type-specific GO enrichment
marker list55–57 generated from the Zhang et al., 201451 patterns of mRNA and protein pairs in the concordant
transcriptomic data and the Sharma et al., 201554 pro- and discordant groups generally matched our results
teomic data from acutely isolated astrocytes and neu- from GO analyses at the level of mRNA-protein pairs
rons from the mouse brain (Fig. 4c and Supplemen- (Fig 4f and g, Supplementary Fig. 9, Supplementary
tary Fig. 8g and h, Supplementary Data 36). A subset Data 38). After evaluating the correlation between the
of example markers belonging to each category are paired proteomes and transcriptomes of each TurboID
shown in Fig. 4d. We repeated the same analysis for model, we aimed to uncover new mechanisms mediat-
the neuronal pulldown proteomes and transcriptomes ing concordance and discordance between mRNA and
(Fig. 4e). These findings nominate cell type-specific as- protein abundances in the bulk cortex, astrocytes and
trocytic and neuronal markers that may be preferentially neurons. We found no relationship between the level of
suitable for either transcriptomic or protein-based stud- mRNA-protein discordance with the average RNA
ies, as well as those that are highly abundant at both traits: % GC content, number of exons and transcript
mRNA and protein levels. length or protein half-life58 (Supplementary Fig. 10,
mRNA-protein discordance can arise from Supplementary Data 39 and 40), suggesting that these
many biological phenomena, which may potentially also mRNA characteristics and protein synthesis dynamics
vary across cell types. We next were curious wondered are unlikely on their own to explain the observed
whether discordant proteins and mRNAs were con- discordance.
sistent across astrocytes and neurons. We assessed Together, these findings emphasize an im-
the level of overlap in mRNA-protein pairs across astro- portant observation about discordant mRNAs and pro-
cytes and neurons in 3 quadrants (high mRNA and high teins across these two cell types. In general, the broad
protein, high mRNA and low protein, low mRNA and classes of mRNAs and proteins that are highly discord-
high protein, Supplementary Data 37). Across all 3 ant in abundance are well conserved between neurons
groups, most concordant mRNA-protein pairs (high
 bioRxiv preprint doi: https://doi.org/10.1101/2025.01.29.635500; this version posted February 1, 2025. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

Figure 4: Correlation analysis of TurboID-based transcriptomes and proteomes of Aldh1l1-expressing astrocytes and Camk2a-expressing neurons in vivo.
a. Schematic of analysis approach to compare SPARO of cortical astrocytes and neurons (n = 3 animals/group). b. Percent rank transformation and density plot
visualization of 1934 gene and protein pairs between the astrocyte-TurboID (left) and neuron-TurboID (right) pulldown transcriptome (x-axis, via normalized, log2-
transformed and percent ranked RNA-seq mean count values) and proteome (y-axis, via normalized, log2-transformed and percent ranked LFQ-MS mean intensity
values). c. Schematic of analysis approach to nominate astrocytic and neuronal markers into highly concordant (mRNA ~ protein in black) or highly discordant (protein
>> mRNA in blue or mRNA >> protein in red) groups. The top astrocyte-specific and neuronal-specific markers were acquired from the unionization of the Zhang et
al., 2014 and Sharma et al., 2015 datasets. d. Scatter plot visualization of examples of astrocytic markers that are concordant where mRNA and protein abundances
are high and similar (top right, black, e.g., Gfap, Aqp4, and Slc1a2), discordant protein >> mRNA (top left, blue, e.g., Tnc, Acot2 and Hadha) and discordant mRNA
>> protein (bottom right, red, e.g., Apoe, Tthy1 and Aldoc). e. Scatter plot visualization of examples of neuronal markers that are concordant where mRNA and protein
abundances are similar (top right, black, e.g., Camk2a, Syn1, and Snca), discordant protein >> mRNA (top left, blue, e.g., Eef1a2, Syngap1, and Sncb) and discordant
mRNA >> protein (bottom right, red, e.g., Gng3, Eno2, and Gap43). f. Venn diagram representation of mRNA and protein pair quantities present in the discordant (top
left, blue, protein >> mRNA) group between astrocyte-TurboID and neuron-TurboID paired pulldown transcriptomes and proteomes. Gene-set enrichment analysis
visualization of GO (via -log10(p value) of unique mRNA and protein pairs enriched in neurons and astrocytes. g. Venn diagram representation of mRNA and protein
pair quantities present in the discordant (bottom right, red, mRNA >> protein) group between Aldh1l1-expressing astrocytic and Camk2a-expressing neuronal paired
pulldown transcriptomes and proteomes. Gene-set enrichment analysis visualization of GO (via -log10(p value) of unique mRNA and protein pairs enriched in neurons
and astrocytes.
 bioRxiv preprint doi: https://doi.org/10.1101/2025.01.29.635500; this version posted February 1, 2025. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

and astrocytes. It is only in the nuanced subtypes of transcriptome of enriched mRNAs after streptavidin
these gene classes where each cell type exhibits a pulldown showed very high overlap and correlation with
uniquely discordant repertoire of proteins and tran- the bulk transcriptome, highlighting that the SPARO
scripts. transcriptome is highly representative of the bulk tran-
scriptome. As predicted, we found that mitochondrial
Discussion DNA-encoded genes (e.g., COX1, CYTB and ND2)
were under sampled in the pulldown transcriptomes,
Measuring transcriptomic and proteomic levels is im- consistent with the inability of cytosolic TurboID-NES to
portant to identify cellular and molecular mechanisms of label the intramitochondrial matrix compartment.
biology. Less than 5% of gene products exhibit cell Using LPS as a neuroinflammatory stimulus to
type-specific expression patterns (via > 4-fold enrich- activate BV2 microglia, we used SPARO and verified
ment), and as a result, bulk tissue -omics studies cannot LPS effect at the pulldown transcriptome level strongly
completely resolve changes occurring at the level of in- reflected LPS effect at the bulk transcriptome level. At
dividual cell types51,54. Since proteome and transcrip- the proteomic level, LPS effect at the bulk level was
tome level abundances only modestly correlate with modestly recapitulated by the pulldown level, potentially
each other52–54, complementary profiling of both levels explained by the cytosolic bias of TurboID-NES. These
of abundances are needed. We describe a new ap- observations highlight major differences in LPS effects
proach to capture the cell type-specific transcriptome at the proteomic and transcriptomic levels in BV2-mi-
and proteome simultaneously that is applicable to in croglia, although canonical markers of microglial activa-
vitro and in vivo models. This method takes advantage tion induced by LPS tend to be conserved at both the
of the biotin ligase, TurboID, to biotinylate cytosolic pro- protein and mRNA levels. Based on successful demon-
teins that interact with RNA including ribosomal and stration of the feasibility and validity of the SPARO ap-
RNA-binding proteins, which allows for enrichment of proach in vitro in a mammalian cell system, we next ap-
biotinylated proteins for proteomics as well as protein- plied the SPARO approach in vivo to contrast the na-
associated mRNA for transcriptomics. This approach, tive-state proteome and transcriptome of astrocytes
called SPARO, addresses a major methodological gap and neurons from the adult mouse cortex. As seen in
in the field by providing a single pipeline for dual tran- our in vitro studies, RNA bound to biotinylated proteins
scriptomics and proteomics from a desired cell type, were enriched from astrocyte-TurboID and neuron-Tur-
while retaining the cell’s native state in the tissue. In this boID cortical tissues. We confirmed that the astrocyte-
study, we establish the SPARO approach to obtain valid and neuron-TurboID pulldown proteomes and transcrip-
proteomes and transcriptomes from BV2 microglia in tomes were cell type enriched via high enrichment of
vitro and further capture the effect of a neuroinflamma- astrocytic or neuronal markers (e.g., Hepacam and
tory stimulus on microglia, using whole-cell bulk profiles Aqp4 in astrocytes and Map2 and Tmem130 in neu-
of BV2 cells as gold standard references. Based on rons) and GO terms related to their unique cellular func-
these in vitro results, we extended SPARO to the in vivo tions. One important observation that we noted was that
setting, using Camk2a-expressing neurons and the effect of LPS on astrocytes in vivo (as well as BV2
Aldh1l1-expressing astrocytes as brain cell types of in- cells in vitro) was far more dramatic at the tran-
terest, further validating the method for in vivo applica- scriptomic level. This is well reflected by the enrichment
tions. By comparing SPARO-derived with RiboTag-de- of the canonical inflammatory reactive marker, Lcn2, at
rived astrocyte transcriptomes, we further confirmed the transcriptomic level only, providing one potential ex-
that both approaches yield comparable results. Lever- planation for why protein-based readouts of reactivity
aging simultaneously enriched transcriptomes and pro- are oftentimes more difficult to demonstrate. Overall,
teomes of neurons and astrocytes, we also investigated these observations provide further validation that
patterns of mRNA-protein concordance across cell SPARO in vivo can be used to quantify cell-type differ-
types. ences at the RNA and protein levels.
Feasibility and validity of the SPARO approach
The SPARO with RiboTag transcriptomes are similar
To first validate SPARO in vitro, we used a validated and complementary
BV2-TurboID microglial cell line35,49 where we could Our benchmarking of SPARO with RiboTag noted some
benchmark our findings against the “ground truth” of a interesting differences. A notable difference between
homogenous cell population. Consistent with previous the SPARO and RiboTag transcriptomes is that in
work35, the TurboID-enriched pulldown proteome cap- SPARO, TurboID biotinylates 133 proteins that interact
tured approximately 50% of proteins identified by bulk with RNA (e.g., ribosomal and RBPs), while the Ri-
proteomics. The level of correlation between the bulk boTag approach labels one ribosomal protein, suggest-
proteome and the cytosolic biotinylated proteome from ing that the TurboID-mediated SPARO approach cap-
BV2 microglia was also modest. We attribute these to tures more than ribosome-bound transcripts. Addition-
the cytosolic bias of TurboID-NES used in our experi- ally, TurboID biotinylates the Rpl22 protein, suggesting
ments, which under-samples nuclear and mitochondrial that the transcripts associated with Rpl22 can be cap-
proteins26. More sensitive MS approaches are also tured similarly to the RiboTag approach. When we
likely to increase the depth of the proteome. The
 bioRxiv preprint doi: https://doi.org/10.1101/2025.01.29.635500; this version posted February 1, 2025. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

compared the two methods using cortical Aldh1l1-ex- bulk. Therefore, the modest in vivo concordance be-
pressing astrocytes, we found that the transcriptomes tween the pulldown proteomes and transcriptomes may
are comparable when looking at the presence and be because of the cytosolic localization of TurboID and
abundance of transcripts. Furthermore, the abundance the biotinylation of a subset of the proteome. We identi-
of canonical astrocytic enriched genes (e.g., Gfap, fied astrocytic and neuronal markers that are highly
Aqp4, Sox9 and Aldh1l1) are highly correlated. When concordant or discordant. Some of the discordant mark-
we differentially compared the RiboTag and SPARO- ers are derived from secreted proteins (e.g., Apoe and
enriched transcriptomes, we found overrepresented on- Gap43). Although cytosolic TurboID has been found to
tologies involved in translation-, splicing- and mitochon- biotinylate secreted proteins29,31,34,35,49, it is not well un-
drial-related terms in the RiboTag pulldowns and syn- derstood if TurboID is able to localize within extracellu-
apse- and ion transport-related terms in the SPARO lar vesicles (EV) found in the endolysosomal secretory
pulldowns. These differences may be attributed to pathway. However, there is evidence that proteins can
SPARO capturing mRNA species beyond only those be secreted without being vesicle-bound60, which would
that are actively being translated. Notably, the highest enable cytosolic TurboID to biotinylate such proteins.
DEP in our SPARO astrocytic pulldown proteome is Future studies using different TurboID constructs local-
Aqp4, which is a highly abundant water channel found ized to various subcellular compartments or throughout
in the astrocytic end-feet59. Perhaps, the differences the whole cell, will allow for a more wholistic quantifica-
highlight methodological differences between the two tion of the concordance between paired proteomes and
approaches. For example, we observed that biotinyla- transcriptomes using SPARO. Future studies that will
tion by TurboID in astrocytes preferentially labels astro- investigate post-transcriptional regulatory processes
cytic end-feet, which is where Aqp4 is mostly localized. within astrocytes and neurons in their native environ-
Despite these differences, we show that the SPARO ment will also likely shed light on the determinants of
and RiboTag astrocytic transcriptomes are indeed com- discordance between mRNA and protein levels.
parable and serve as appropriate tools to measure as- We also found that many of the cellular pro-
trocytic genes in adult cortical astrocytes in vivo. cesses involved in the concordant and discordant
mRNA and protein pair groups are shared between as-
SPARO to assess the concordance between paired trocytes and neurons. For example, in the discordant
proteomes and transcriptomes group (protein >> mRNA) astrocytes are enriched in mi-
One of the major advantages of SPARO is the ability to crotubule proteins and neurons are enriched in actin
assess concordance between proteomes and transcrip- proteins. It is not fully clear whether the abundance of
tomes of the same cells. In our in vivo astrocyte and these cytoskeletal proteins differ across the two cell
neuron datasets, we observed only modest concord- types in general, across subcellular compartments or
ance between the pulldown proteomes and transcrip- across different brain regions. It has been well reported
tomes, which is in line with prior studies52–54. There are that actin filaments are an important part of the neuronal
many possible reasons as to why the concordance is cytoskeleton and contribute to the structure of axons
modest. From a technical perspective, the sensitivity of and dendrites61. In contrast to our study in the cortex, a
MS detection limits likely contributes to challenges of cytosolic BioID2-based LFQ-MS proteomics study
accurately measuring protein abundances and there- found actin-filament-based processes enriched in
fore, the correlation with transcript abundances. Addi- mouse striatal astrocytes33, suggesting that there may
tionally, transcriptomic and proteomic datasets are usu- be brain region differences. Another study found that
ally analyzed in isolation with different data processing microtubules are present in perivascular astrocytic end-
and statistical analysis approaches that may contribute feet62. Perhaps the high enrichment of microtubule pro-
to the poor correlation between mRNA and protein lev- teins in the SPARO astrocytic pulldowns is due to cap-
els. There are also limited resources to determine which turing more end-feet-associated proteins, as suggested
allele of a gene yields a particular protein isoform. More- by the high abundance of Aqp4 protein as well. In the
over, the dynamic range of abundance is much higher discordant group (mRNA >> protein), different mito-
for proteins than for transcripts1,2,13. chondrial-related terms were enriched in astrocytes and
From a biological standpoint, other possibilities neurons. The transcripts associated with mitochondrial
for the mRNA-protein discordance are post-transcrip- function are nuclear-encoded, as opposed to mitochon-
tional regulatory mechanisms such as mRNA and pro- drial DNA-encoded. Therefore, we predict that the mi-
tein stability, mRNA half-life transcription and transla- tochondrial transcripts are higher across both cell types
tion rates and efficiency, post-translational modifica- because SPARO utilizes a cytosolic TurboID and under
tions, and vesicle-bound secretory pathways. Because samples mitochondrial proteins. Interestingly, we found
we did not find a pattern associated with protein half-life ganglioside-induced differentiation associated protein 1
and mRNA-protein discordance, we predict that other (Gdap1), a neuronal mitochondrial outer membrane
mechanisms or a collection of many cellular and molec- protein implicated in Charcot-Marie-Tooth disease63–65,
ular processes contribute to the discordance between to be highly enriched in the SPARO neuronal transcrip-
mRNA and protein levels. We also observed that the tome, suggesting that SPARO may be a useful tool to
BV2-TurboID bulk and pulldown transcriptomes in vitro study disease-associated genes and proteins. Although
are highly correlated, suggesting that the TurboID-en- the cellular processes are similar across the two cell
riched transcriptome is generally representative of the types, our findings suggest that astrocytes and neurons
 bioRxiv preprint doi: https://doi.org/10.1101/2025.01.29.635500; this version posted February 1, 2025. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

may use different molecular machinery. The differences used for cause-and-effect mechanistic studies using re-
are likely due to the unique functions astrocytes and ductionist models in multi-cellular model systems. Addi-
neurons play in the brain66. tionally, SPARO can be used to study how transcrip-
tional and translational processes are fully regulated,
Broader applications of SPARO changes in the efficiency of protein biosynthesis, pro-
While our studies provide the foundation for the cesses that contribute to the discordance between
SPARO approach and demonstrate its validity in vivo, mRNAs and proteins, mRNA and protein degradation,
the approach offers several future potential applications and much more. A major advantage of SPARO is that it
and extensions with broad implications within and be- can be adapted to study different cell types, brain re-
yond the neuroscience field. One interesting finding is gions, and other tissues outside the brain, making the
the quantity of DEGs found in the astrocytic and neu- approach a highly versatile and broadly applicable tool
ronal pulldown transcriptomes. Nearly 3 times more for the field of molecular biology.
DEGs were found in astrocytes when compared to neu-
rons. The difference between cell types may be at- Methods
tributed to differences in the heterogeneity of neuronal In vitro studies
and astrocytic populations, or differences in the abun- Mouse microglia BV2 cells were cultured in 0.2 µm vac-
dance of enriched biotinylated proteins that interact with uum filter sterilized Dulbecco’s Modified Eagle Medium
RNA (e.g., ribosomal proteins). Additionally, because supplemented with high glucose and L-glutamine con-
astrocytes are highly metabolically active81 and since taining 1% penicillin-streptomycin, and 10% fetal bovine
many of the top astrocytic DEGs are involved in metab- serum. Cells were incubated in 100 mm cell culture dish
olism-related processes (e.g., lipid binding), the at 37°C and 5% CO2 until reaching 80-90% confluency.
SPARO method may be more efficient in capturing tran- Cells were treated with 100 ng/mL of LPS for 48 h and
scripts related to metabolism that are more highly abun- 200 µM of biotin for 24 h during the second day of LPS
dant in astrocytes. Further research will be needed to treatment.
parse out cell-type differences in transcript abundances
and the biases of the SPARO method. Additionally, with In vivo studies
the rapid evolution of MS instrumentation and MS pipe- Mice were housed in the Division of Animal Resources
lines, it is anticipated that the SPARO approach can be vivarium at Emory University under a standard environ-
extended to small brain regions and subcellular com- ment (12 h light/ 12 h dark cycle, temperature 72°F, hu-
partments with significantly deeper proteomic coverage midity range 40-50%) with access to food and water ad
in future studies67,68. libitum. All animal-related studies (PROTO-201700821)
TurboID biotinylates RBPs, therefore, SPARO were conducted with the approval of Emory’s Institu-
also has the potential to quantify RNA species other tional Animal Care and Use Committee following the
than mRNAs including miRNAs, snRNAs, and other National Institute of Health’s “Guide for the Care and
lncRNAs. Recent work has highlighted that lncRNAs Use of Laboratory Animals” and reported in accordance
play a significant role in various cellular processes in- with the ARRIVE guidelines.
cluding transcription, translation, metabolism and sig- Transgenic approach for Cre recombinase ex-
naling70. Interestingly, Xist, a lncRNA that is important pression in neurons and astrocytes using Rosa26Tur-
for mammalian females to transcriptionally silence one boID/wt
mice:
of the pair of X chromosomes71, was highly enriched in Neuronal TurboID cohort: Rosa26TurboID/wt (Jackson
the neuronal pulldown transcriptome in which 2/3 of the Labs, Strain No. 037890) were crossed with Camk2aCre-
mice were female. The astrocytic samples were derived ERT2
(Jackson Labs, Strain No. 012362) mice to obtain
from male mice. This observation highlights that heterozygous Rosa26TurboID/wt/Camk2aCreERT2/wt (“neu-
SPARO may be a useful tool to study sex differences at ron-TurboID”, n = 3, 1 M and 2 F littermate mice. Heter-
the transcriptomic and proteomic levels. Lastly, future ozygous littermate mice Camk2aCreERT2/wt (“Camk2a”, n
modifications of TurboID in vivo models using both = 2, 1 M and 1 F) were used as controls.
transgenic72 and AAV-based approaches33,73, may pro-
vide opportunities to restrict TurboID to other cellular Astrocytic TurboID cohort: Rosa26TurboID/wt (Jackson
compartments or remove the cytosolic restriction by ex- Labs, Strain No. 037890) were crossed with
cluding the NES tag. These future endeavors are likely Aldh1l1CreERT2 (Jackson Labs, Strain No. 031008) mice
to extend the SPARO approach to interrogate compart- to obtain heterozygous Rosa26TurboID/wt/Aldh1l1CreERT2/wt
ment-specific protein and mRNA processes such as lo- (“astrocyte-TurboID”, n = 8, 6 M and 2 F). Heterozygous
cal translation that occurs in synapses or glial pro- littermate mice Aldh1l1CreERT2/wt (“Aldh1l1”, n= 3, 2 M
cesses74–80. TurboID-EV was recently developed69 and and 1 F) were used as controls. For astrocyte-TurboID
a TurboID-EV-SPARO approach will be an exciting fu- LPS studies, systemic LPS at 0.75 /kg/dose x 5 days
ture tool to better understand the RNA and protein com- was given via i.p. injection during the first week of biotin
positions of EVs. water supplementation. Based on previous work and
In conclusion, SPARO is an exciting new tool our own, we validated neuron- and astrocyte-specificity
that can be used to study many facets of cellular and and nonleaky Cre activity of the Camk2aCreERT2 and
molecular biology. For example, SPARO in vitro can be Aldh1l1CreERT2 models, respectively40,50.
 bioRxiv preprint doi: https://doi.org/10.1101/2025.01.29.635500; this version posted February 1, 2025. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

Astrocytic RiboTag cohort: RiboTag mice, Cell and tissue homogenization

Rpl22tm1.1Psam (Jackson Labs, Strain No. 029977) were BV2 control and BV2-TurboID cells (1.0 x 106 cells)
crossed with Aldh1l1CreERT2 (Jackson Labs, Strain No. were washed three times with cold 1X PBS containing
031008) mice to obtain heterozygous 100 µg/mL of cycloheximide. Next, cells were incubated
Rpl22tm1.1Psam/Aldh1l1CreERT2/wt (“astrocyte--RiboTag”, n with 1 mL of cold homogenization buffer (10 mM
= 3, 3 F). Heterozygous littermate mice Rpl22tm1.1Psam HEPES pH 7.4, 150 mM KCl, 10 mM MgCl2 supple-
(“RiboTag”, n= 2, 1 M and 1 F) were used as controls. mented with 0.5 mM DTT, 0.1% v/v RNasin, 0.1% v/v
All mice were given tamoxifen (75 mg/kg) intra- SUPERasin, 1X HaltTM protease and phosphatase in-
peritoneally (i.p.) for 5 days at 6 weeks of age and al- hibitors, 100 µg/mL cycloheximide) for 10 minutes,
lowed 3 weeks of Cre recombination. After Cre recom- scaped and collected into a 1.5 mL Eppendorf LoBind
bination, TurboID mice were given water supplemented tube. For brain samples, frozen Camk2a and Aldh1l1
with biotin (37.5 mg/L) for 2 weeks until euthanasia at controls and neuron-TurboID and astrocyte-TurboID
3-4 months of age. RiboTag mice were euthanized at mouse cortices (up to 50 mg) were lysed in same cold
2.5 months of age. Mice from each cohort were anes- homogenization buffer as the BV2 cells with RNase-
thetized using ketamine/xylazine (ketamine 87.5 mg/kg, free glass beads and a Bullet Blender Homogenizer
xylazine 12.5 mg/kg) followed by transcardial perfusion (Next Advance). RiboTag control and astrocyte-Ri-
with ice-cold 1X PBS. The brain was immediately re- boTag mouse cortices were lysed in cold RiboTag ho-
moved and hemisected along the mid-sagittal line. The mogenization buffer (50 mH Tris-HCl pH 7.5, 100 mM
left hemisphere was immediately dissected to obtain KCl, 12 mM MgCl2, 1% NP-40, 1 mM DTT, 1% v/v so-
the cortex and then snap frozen using dry ice. The right dium deoxycholate 0.1% v/v RNasin, 0.1% v/v SUPER-
hemisphere was used for immunohistological studies. asin, 1X Halt protease and phosphatase inhibitors, 100
µg/mL cycloheximide) similarly to the TurboID samples.
Immunohistochemistry All homogenates underwent 3 rounds of sonication (5 s
For immunohistochemistry studies, mice were eu- of active sonication at 25% amplitude followed by a 10
thanized and perfused with ice-cold 1X PBS, and the s incubation on ice) followed by centrifugation for 10 min
brain was dissected. One hemisphere was immediately at 2000 x g at 4°C. The supernatants were transferred
transferred into 4% PFA (Cat No. J19943.K2) and fixed to a new tube and protein concentrations were calcu-
overnight for immunohistological analysis. After PFA lated using a Pierce BCA assay for proteomic studies.
fixation, the brain was washed with 1X PBS 3 times to
remove residual PFA and transferred into 30% sucrose Immunoblotting
solution until sectioning. After removing extra sucrose, To confirm the presence of biotinylated proteins, 17 µg
the fixed and sucrose-saturated hemisphere was em- of cell or tissue lysates were resolved on a 4-12% Bis-
bedded in the Tissue-Tek optimal cutting temperature Tris gel, transferred onto a nitrocellulose membrane,
compound (Cat No. 4583, Sakura) and frozen on dry blocked with StartingBlock for 30 min and then probed
ice. Serial 40 µm sagittal sections of the brain were gen- with a streptavidin-AlexaFluor 680 antibody (Ther-
erated using a cryostat (Leica Biosystems, moFisher, S32358, dilution: 1:10,000 in StartingBlock)
CM1850) and stored free-floating in cryoprotective me- for 1 h at room temperature. To confirm TurboID con-
dia (Glycerin, Ethylene Glycol and 0.1 M Phosphate struct presence, the membrane was probed with a rab-
Buffer in ratio of 2.5: 3.0: 5.0) at -20°C. For immunoflu- bit anti-V5 primary antibody (dilution: 1:500 in Start-
orescence (IF) staining, 3-4 brain sections from each ingBlock) overnight and then incubated with a goat anti-
mouse were washed 3 times with 1X TBS and then rabbit HRP-conjugated secondary antibody (Jackson
blocked and permeabilized by incubating in 1X TBS ImmunoResearch, 111-035-003, dilution: 1:10,000 in
containing 0.25% Triton X-100 (TBS-T) and 5% goat se- StartingBlock). Immunoblots were imaged with an Od-
rum for 1 h at room temperature. Following, sections yssey Infrared Imaging System (LI-COR Biosciences),
were incubated with primary antibodies diluted in 1X or BioRad chemiluminescence system.
TBS-T containing 1% goat serum overnight at 4°C.
Next, the sections were washed 3X with 1X TBS and Biotinylated protein and ribosomal protein enrich-
then incubated with fluorophore-conjugated secondary ment for transcriptomics
antibodies for 2 h at room temperature in the dark. All Following confirmation of biotinylated proteins and V5
primary and secondary antibodies were used at dilution presence via immunoblotting, biotin-tagged proteins
1:500 including Streptavidin Protein, DyLight™ 488. were enriched for using streptavidin-coated magnetic
Brain sections were then incubated for 5 minutes in nu- beads as described previously34. For each sample, 25
clear staining reagent DAPI (1 µg/mL) dissolved in 1X µL of streptavidin beads (ThermoFisher, 88817) in a 1.5
TBS. The samples were further washed 3X with 1X mL Eppendorf LoBind tube were washed 3 times with 1
TBS, mounted onto glass slides and coverslipped using mL of homogenization buffer on rotation for 2 min. The
ProLong Diamond Antifade Mountant (P36965, Ther- streptavidin beads were incubated with 300 µg of pro-
moFisher). Images of the same brain regions for each tein lysate from each sample with additional wash buffer
mouse were captured using a Keyence fluorescence (10 mM HEPES pH 7.4, 150 mM KCl, 10 mM MgCl2
microscope (Keyence BZ-X810). Images were pro- supplemented with 0.5 mM DTT, 1% v/v NP-40 substi-
cessed and analyzed using Image J software (FIJI Ver- tute, 0.1 % v/v RNasin, 0.1% v/v SUPERasin, 1X HaltTM
sion 2.14). protease and phosphatase inhibitors, 100 µg/mL
 bioRxiv preprint doi: https://doi.org/10.1101/2025.01.29.635500; this version posted February 1, 2025. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

cycloheximide) to create a final volume of 250 µL on 100) on rotation for 2 min. The beads were incubated
rotation for 1 h at 4°C. Next, the beads were washed with 500 µg of protein lysate from each sample with ad-
four times with cold high salt buffer (10 mM HEPES pH ditional RIPA lysis buffer to create a final volume of 500
7.4, 350 mM KCl, 10 mM MgCl2 supplemented with 0.5 µL on rotation for 1 h at 4°C. Next, the beads were
mM DTT, 1% v/v NP-40 substitute, 0.1 % v/v RNasin, quickly centrifuged, placed on a magnetic rack and the
0.1% v/v SUPERasin, 1X HaltTM protease and phospha- supernatant was transferred to a new 1.5 mL Eppendorf
tase inhibitors). After RiboTag control and Astrocyte-Ri- LoBind tube and stored at -80°C. The beads were
boTag cortical lysis, 800 µL of lysate was incubated with washed with the following buffers on rotation at room
4 µL of anti-hemagglutinin (HA) antibody (Biolegend, temperature: twice with 1 mL of RIPA lysis buffer for 8
901513) for 4 hours on end-over-end rotation at 4°C. min, once with 1 M KCl for 8 min, once with 1 mL 0.1 M
For each sample, 25 µL of A/G beads (ThermoFisher, sodium carbonate for ~10 s, once with 1 mL 2 M urea
88803) in a 1.5 mL Eppendorf LoBind tube were in 10 mM Tris-HCl (pH 7.6) for ~10 s, and twice with 1
washed 3 times with 400 µL of RiboTag homogeniza- mL RIPA lysis buffer for 8 min. After the final RIPA
tion buffer. The A/G beads were incubated with 100 µL wash, the beads were resuspended in 1 mL of 1X PBS,
of HA-tagged lysate overnight on end-over-end rotation transferred to a new tube and washed two more times
at 4°C. Next, the A/G beads were washed four times with 1X PBS on rotation for 2 min. To confirm biotinyl-
with cold RiboTag high salt buffer (50 mM Tris-HCl pH: ated protein enrichment, 10% of the streptavidin bead
7.5, 300 mM KCl, 12 mM MgCl2, 1% v/v NP-40, 0.50 volume was transferred to a new 1.5 mL Eppendorf
mM DTT, 100 µg/mL cycloheximide).10 mM HEPES pH LoBind tube and boiled in 30 μL of 2X Laemmli protein
7.4, 350 mM KCl, 10 mM MgCl2 supplemented with 0.5 loading buffer (Bio-Rad, 1610737) supplemented with 2
mM DTT, 1% v/v NP-40 substitute, 0.1 % v/v RNasin, mM biotin and 20 mM dithiothreitol (DTT) at 95°C for 10
0.1% v/v SUPERasin, 1X HaltTM protease and phospha- min to elute the biotinylated proteins. Following, 1/3 or
tase inhibitors). After final washes, the streptavidin and the eluate was resolved on a 4-12% Bis-Tris gel, trans-
A/G beads were resuspended in 100 µL of wash buffer ferred onto a nitrocellulose membrane, blocked with
and then added to 700 µL of Trizol and stored at -80°C StartingBlock for 30 min and incubated overnight with a
until RNA extraction. For bulk/input lysate/animal, 50 µL rabbit anti-V5 antibody (Abcam, ab206566) on a shaker
of lysate remained at 4°C during the protein enrichment at 4°C. Next, the membrane was incubated with a goat
protocols. anti-rabbit HRP-conjugated secondary antibody (Jack-
son ImmunoResearch, 111-035-003, dilution: 1:10,000
RNA extraction in StartingBlock) and streptavidin-AlexaFluor 680 anti-
After protein enrichment using streptavidin and A/B body (ThermoFisher, S32358, dilution: 1:10,000 in
beads, RNA was extracted from the beads and or bulk StartingBlock) for 1 h at room temperature. The remain-
lysates using a miRNeasy Mini Kit (Qiagen Inc., ing 2/3 of protein eluate was resolved on a 4-12% Bis-
217004) following the manufacturer’s instructions. RNA Tris gel for a silver stain (ThermoFisher, 24612). Im-
was eluted in RNase-free water and analyzed for purity munoblots were imaged with an Odyssey Infrared Im-
and concentration using a Bioanalyzer (Agilent) prior to aging System (LI-COR Biosciences), or BioRad chemi-
RNA-sequencing library preparation. luminescence system. Silver-stained gels were imaged
using a Canon scanner (CanoScan LIDE 300).
RNA-sequencing and processing
After assessing RNA quality, cDNA library preparation Protein digestion
was performed. The SMART-Seq® v4 PLUS Kit (Takara To prepare enriched biotinylated proteins for MS-based
Bio) was used following the manufacturer’s protocol. proteomics, the remaining 90% volume of the streptav-
Bulk sequencing of 40 million paired-end reads per idin beads were resuspended in 50 mM ammonium bi-
sample was completed using the Illumina-based plat- carbonate (Na3CO2). Next, the bound proteins were re-
form at Admera Health. FASTQ files were evaluated for duced with 1 mM DTT for 30 min at room temperature
quality using FastQC82 and then mapped to the mouse and then alkylated with 5 mM iodoacetamide (IAA) in
genome, GRCm38 (mm10) using the Spliced Tran- the dark for 30 min. Following, proteins were subse-
scripts Alignment to a Reference (STAR)83 aligner with quently digested overnight with 0.5 µg of lysyl (Lys-C)
the paired-end option. The featureCounts84 package endopeptidase (Wako, 127-06621) and then 1 µg of
was used to generate a raw counts data matrix from trypsin (ThermoFisher, 90058) on the shaker at room
mapped reads. Lowly abundant transcripts across all temperature. After digestion, the resulting peptide mix-
samples (average count value ≤ 10) were filtered out. tures were acidified to a final concentration of 1% formic
After generating the filtered, raw counts data matrices, acid and 0.1% trifluoroacetic acid, desalted with an HLB
we used DESeq285 to normalize the matrices and per- column (Waters, 186003908), and dehydrated using a
form differential gene expression-based statistics. vacuum centrifuge (SpeedVac Vacuum Concentrator).
To prepare bulk lysates for MS-proteomics, 50 µg of
Biotinylated protein enrichment for proteomics protein from each sample was reduced in 5 mM DTT for
For each sample, 42 µL of streptavidin beads in a 1.5 30 min at room temperature and then alkylated with 10
mL Eppendorf LoBind tube were washed 2 times with 1 mM IAA for 30 min in the dark. After, each sample was
mL of RIPA lysis buffer (50 mM Tris, 150 mM NaCl, diluted (4-fold) with 50 mM ammonium bicarbonate
0.1% SDS, 0.5% sodium deoxycholate, 1% Triton X- (ABC) buffer and then digested with 1 µg of Lys-C
 bioRxiv preprint doi: https://doi.org/10.1101/2025.01.29.635500; this version posted February 1, 2025. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

endopeptidase on the shaker overnight at room temper- Data analysis and visualization
ature. Next, samples were diluted (4-fold) with 50 mM We utilized differential enrichment analysis, multidimen-
ABC buffer and digested with 2 µg of trypsin on the sional scaling (MDS), and gene set enrichment analysis
shaker overnight at room temperature. After digestion, (GSEA) to analyze transcriptomic and proteomic data
the peptide mixtures were acidified, desalted, and dried from in vitro and in vivo studies. Data analysis and vis-
down as described above. ualization was completed using R software (version
4.3.1) and Prism (GraphPad, version 10). For tran-
Mass spectrometry scriptomic studies, the DESeq285 Wald test was used to
Dried peptides were prepared for liquid chromatog- identify DEGs between groups. For proteomic studies,
raphy as described previously34. Following, an Orbitrap unpaired 2-tailed (equal variance assumption) T-test
Lumos Tribrid MS with a high-field asymmetric wave- equivalent calculations using F value from ANOVA run
form ion mobility spectrometry (FAIMS Pro)86 interface strictly with 2 comparison groups were performed with
was used to obtain all mass spectra at a compensation the parANOVA suite of functions in R
voltage of -45V with similar parameters as done be- (https://github.com/edammer/parANOVA) to identify
fore34. DEPs between groups. DEGs and DEPs are presented
as volcano plots. GSEA was applied using the GOpar-
Protein identification and quantification allel function (https://github.com/edammer/GOparallel)
Raw MS files for bulk and pulldown protein from BV2- that scrapes monthly updated .GMT formatted GO from
TurboID in vitro and astrocyte- and neuron-TurboID in the Bader Lab website (https://baderlab.org/) as gene
vivo studies along with respective controls were symbol lists submitted to one-tailed Fisher’s exact test
searched separately using the Andromeda87 search en- for GO enrichment of overlapping symbols in gene lists
gine integrated into MaxQuant88 (version v2.4.2.0), of interest, including DEPs and DEGs. GO terms meet-
against the 2020 mouse UniProt database ing Fisher exact significance of p value ≤ 0.05 (i.e., a Z-
(https://www.uniprot.org/help/reference_proteome) in- > 1.96) were considered. Protein and gene input lists
cluding sequences for V5 and TurboID. Search param- were significantly differentially biochemically enriched
eters including: variable and fixed modifications, al- over bulk or background, in addition to gene lists that
lowed miscleavages, minimum/maximum peptide mass were different by treatment condition (p ≤ 0.05, ≥ 1 log2-
and length, mass tolerance for fragment and precursor fold change). To assess average GC content, number
ions and quantification settings were determined as of exons, and average transcript length of tran-
done previously34,35. Peptide spectral match false dis- scriptomic data, we used the biomaRt R package89,90
covery rate (FDR) was set to 1%. The MaxQuant output and the BioMart/Ensembl databases91. To examine pro-
data (raw intensity values) were uploaded to R (version tein half-life information of proteomic data, we used val-
4.3.1) for normalization, log2 transformation, and differ- ues from Fornasiero et al., 201858.
ential abundance analyses. To account for endoge-
nously biotinylated and non-specific proteins captured Data availability
during streptavidin affinity purification, mean intensity The MS proteomics data have been deposited to the
values from TurboID-negative control pulldown sam- ProteomeXchange Consortium92 via the PRIDE93 part-
ples were subtracted from the TurboID-positive pull- ner repository with the dataset identifier PXD059818.
down samples by group. To account for variability in Transcriptomics data is available on the NIH Gene Ex-
TurboID expression, biotinylated protein abundance, pression Omnibus (GEO) repository with the dataset
and streptavidin affinity purification efficiency, intensity identifier GSE287770. Source data are provided with
values were normalized depending on the dataset. For this paper.
BV2 in vitro studies, abundance values from bulk pro-
tein and enriched pulldown samples were normalized References
by taking each sample’s log2-transformed summed pro- 1. Liu, Y., Beyer, A. & Aebersold, R. On the Depend-
tein intensity median and setting it to zero. For astro- ency of Cellular Protein Levels on mRNA Abundance.
cyte-TurboID and neuron-TurboID in vivo studies, en- Cell 165, 535–550 (2016).
riched pulldown samples were normalized by subtract- 2. Franks, A., Airoldi, E. & Slavov, N. Post-transcrip-
ing the difference from the mean of the log2 transformed tional regulation across human tissues. PLoS Comput
summed protein intensity for 71 ribosomal subunit pro- Biol 13, e1005535 (2017).
teins present in all TurboID-positive pulldown samples. 3. Velázquez-Cruz, A., Baños-Jaime, B., Díaz-Quin-
Proteins were filtered out if they had missing values in tana, A., De La Rosa, M. A. & Díaz-Moreno, I. Post-
2/3 samples across groups. Missing values were im- translational Control of RNA-Binding Proteins and Dis-
puted from normal distribution for each sample. Nomi- ease-Related Dysregulation. Front. Mol. Biosci. 8,
nation of proteins as either ribosomal or RNA-binding 658852 (2021).
(via RNA-binding motif or RRM) were determined using 4. Saibil, H. Chaperone machines for protein folding,
the HUGO Gene Nomenclature Committee database unfolding and disaggregation. Nat Rev Mol Cell Biol 14,
(https://www.genenames.org/). 630–642 (2013).
5. Goswami, S. K. Regulation of gene expression in
mammals. in Post-Transcriptional Gene Regulation in
 bioRxiv preprint doi: https://doi.org/10.1101/2025.01.29.635500; this version posted February 1, 2025. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

Human Disease 1–31 (Elsevier, 2022). (BONCAT). Proc. Natl. Acad. Sci. U.S.A. 103, 9482–
doi:10.1016/B978-0-323-91305-8.00019-3. 9487 (2006).
6. Buccitelli, C. & Selbach, M. mRNAs, proteins and 23. Hodas, J. J. L. et al. Dopaminergic modulation of
the emerging principles of gene expression control. Nat the hippocampal neuropil proteome identified by
Rev Genet 21, 630–644 (2020). bioorthogonal noncanonical amino acid tagging (
7. Brito Querido, J., Díaz-López, I. & Ramakrishnan, BONCAT ). Proteomics 12, 2464–2476 (2012).
V. The molecular basis of translation initiation and its 24. Landgraf, P., Antileo, E. R., Schuman, E. M. & Di-
regulation in eukaryotes. Nat Rev Mol Cell Biol 25, 168– eterich, D. C. BONCAT: Metabolic Labeling, Click
186 (2024). Chemistry, and Affinity Purification of Newly Synthe-
8. Bauernfeind, A. L. & Babbitt, C. C. The predictive sized Proteomes. in Site-Specific Protein Labeling (eds.
nature of transcript expression levels on protein expres- Gautier, A. & Hinner, M. J.) vol. 1266 199–215 (Springer
sion in adult human brain. BMC Genomics 18, 322 New York, New York, NY, 2015).
(2017). 25. Link, A. J. et al. Discovery of aminoacyl-tRNA syn-
9. Cenik, C. et al. Integrative analysis of RNA, trans- thetase activity through cell-surface display of non-
lation, and protein levels reveals distinct regulatory var- canonical amino acids. Proc. Natl. Acad. Sci. U.S.A.
iation across humans. Genome Res. 25, 1610–1621 103, 10180–10185 (2006).
(2015). 26. Branon, T. C. et al. Efficient proximity labeling in
10. Chen, G. et al. Discordant Protein and mRNA Ex- living cells and organisms with TurboID. Nat Biotechnol
pression in Lung Adenocarcinomas. Molecular & Cellu- 36, 880–887 (2018).
lar Proteomics 1, 304–313 (2002). 27. Cho, K. F. et al. Split-TurboID enables contact-de-
11. Perl, K. et al. Reduced changes in protein com- pendent proximity labeling in cells. Proc. Natl. Acad.
pared to mRNA levels across non-proliferating tissues. Sci. U.S.A. 117, 12143–12154 (2020).
BMC Genomics 18, 305 (2017). 28. Cho, K. F. et al. Proximity labeling in mammalian
12. Salovska, B. et al. Isoform-resolved correlation cells with TurboID and split-TurboID. Nat Protoc 15,
analysis between mRNA abundance regulation and 3971–3999 (2020).
protein level degradation. Mol Syst Biol 16, (2020). 29. Goettemoeller, A. M. et al. Entorhinal cortex vulner-
13. Schwanhäusser, B. et al. Global quantification of ability to human APP expression promotes hyperexcit-
mammalian gene expression control. Nature 473, 337– ability and tau pathology. Nat Commun 15, 7918 (2024).
342 (2011). 30. Kim, D. I. et al. An improved smaller biotin ligase
14. Specht, H. et al. Single-cell proteomic and tran- for BioID proximity labeling. MBoC 27, 1188–1196
scriptomic analysis of macrophage heterogeneity using (2016).
SCoPE2. Genome Biol 22, 50 (2021). 31. Kumar, P. et al. Native-state proteomics of Parval-
15. Wang, D. et al. A deep proteome and transcriptome bumin interneurons identifies unique molecular signa-
abundance atlas of 29 healthy human tissues. Molecu- tures and vulnerabilities to early Alzheimer’s pathology.
lar Systems Biology 15, e8503 (2019). Nat Commun 15, 2823 (2024).
16. Wei, Y.-N. et al. Transcript and protein expression 32. Mair, A., Xu, S.-L., Branon, T. C., Ting, A. Y. &
decoupling reveals RNA binding proteins and miRNAs Bergmann, D. C. Proximity labeling of protein com-
as potential modulators of human aging. Genome Biol plexes and cell-type-specific organellar proteomes in
16, 41 (2015). Arabidopsis enabled by TurboID. eLife 8, e47864
17. Marsh, S. E. et al. Dissection of artifactual and con- (2019).
founding glial signatures by single-cell sequencing of 33. Soto, J. S. et al. Astrocyte–neuron subproteomes
mouse and human brain. Nat Neurosci 25, 306–316 and obsessive–compulsive disorder mechanisms. Na-
(2022). ture 616, 764–773 (2023).
18. Brewer, G. J. & Torricelli, J. R. Isolation and culture 34. Rayaprolu, S. et al. Cell type-specific biotin labeling
of adult neurons and neurospheres. Nat Protoc 2, in vivo resolves regional neuronal and astrocyte prote-
1490–1498 (2007). omic differences in mouse brain. Nat Commun 13, 2927
19. Brewer, G. J. Isolation and culture of adult rat hip- (2022).
pocampal neurons. Journal of Neuroscience Methods 35. Sunna, S. et al. Cellular Proteomic Profiling Using
71, 143–155 (1997). Proximity Labeling by TurboID-NES in Microglial and
20. Alvarez-Castelao, B. et al. Cell-type-specific meta- Neuronal Cell Lines. Molecular & Cellular Proteomics
bolic labeling of nascent proteomes in vivo. Nat Biotech- 22, 100546 (2023).
nol 35, 1196–1201 (2017). 36. Dumrongprechachan, V. et al. Cell-type and sub-
21. Alvarez-Castelao, B., Schanzenbächer, C. T., cellular compartment-specific APEX2 proximity labeling
Langer, J. D. & Schuman, E. M. Cell-type-specific met- reveals activity-dependent nuclear proteome dynamics
abolic labeling, detection and identification of nascent in the striatum. Nat Commun 12, 4855 (2021).
proteomes in vivo. Nat Protoc 14, 556–575 (2019). 37. Hobson, B. D. et al. Subcellular proteomics of do-
22. Dieterich, D. C., Link, A. J., Graumann, J., Tirrell, pamine neurons in the mouse brain. eLife 11, e70921
D. A. & Schuman, E. M. Selective identification of newly (2022).
synthesized proteins in mammalian cells using 38. Sanz, E. et al. Cell-type-specific isolation of ribo-
bioorthogonal noncanonical amino acid tagging some-associated mRNA from complex tissues. Proc.
Natl. Acad. Sci. U.S.A. 106, 13939–13944 (2009).
 bioRxiv preprint doi: https://doi.org/10.1101/2025.01.29.635500; this version posted February 1, 2025. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

39. Sanz, E., Bean, J. C., Carey, D. P., Quintana, A. & fluid reveals early changes in energy metabolism asso-
McKnight, G. S. RiboTag: Ribosomal Tagging Strategy ciated with microglia and astrocyte activation. Nat Med
to Analyze Cell-Type-Specific mRNA Expression In 26, 769–780 (2020).
Vivo. Current Protocols in Neuroscience 88, (2019). 57. Johnson, E. C. B. et al. Large-scale deep multi-
40. Srinivasan, R. et al. New Transgenic Mouse Lines layer analysis of Alzheimer’s disease brain reveals
for Selectively Targeting Astrocytes and Studying Cal- strong proteomic disease-related changes not ob-
cium Signals in Astrocyte Processes In Situ and In Vivo. served at the RNA level. Nat Neurosci 25, 213–225
Neuron 92, 1181–1195 (2016). (2022).
41. Kang, S. S. et al. Microglial translational profiling 58. Fornasiero, E. F. et al. Precisely measured protein
reveals a convergent APOE pathway from aging, amy- lifetimes in the mouse brain reveal differences across
loid, and tau. Journal of Experimental Medicine 215, tissues and subcellular fractions. Nat Commun 9, 4230
2235–2245 (2018). (2018).
42. Heiman, M. et al. A Translational Profiling Ap- 59. Mader, S. & Brimberg, L. Aquaporin-4 Water Chan-
proach for the Molecular Characterization of CNS Cell nel in the Brain and Its Implication for Health and Dis-
Types. Cell 135, 738–748 (2008). ease. Cells 8, 90 (2019).
43. Heiman, M., Kulicke, R., Fenster, R. J., Greengard, 60. Cohen, M. J., Chirico, W. J. & Lipke, P. N. Through
P. & Heintz, N. Cell type–specific mRNA purification by the back door = Unconventional protein secretion. The
translating ribosome affinity purification (TRAP). Nat Cell Surface 6, 100045 (2020).
Protoc 9, 1282–1291 (2014). 61. Dogterom, M. & Koenderink, G. H. Actin–microtu-
44. Zhou, X., Liao, W.-J., Liao, J.-M., Liao, P. & Lu, H. bule crosstalk in cell biology. Nat Rev Mol Cell Biol 20,
Ribosomal proteins: functions beyond the ribosome. 38–54 (2019).
Journal of Molecular Cell Biology 7, 92–104 (2015). 62. Eom, T.-Y. et al. Direct visualization of microtu-
45. Shi, Z. et al. Heterogeneous Ribosomes Preferen- bules using a genetic tool to analyse radial progenitor-
tially Translate Distinct Subpools of mRNAs Genome- astrocyte continuum in brain. Nat Commun 2, 446
wide. Mol Cell 67, 71-83.e7 (2017). (2011).
46. Reimegård, J. et al. A combined approach for sin- 63. Pedrola, L. et al. GDAP1, the protein causing Char-
gle-cell mRNA and intracellular protein expression anal- cot–Marie–Tooth disease type 4A, is expressed in neu-
ysis. Commun Biol 4, 624 (2021). rons and is associated with mitochondria. Human Mo-
47. Darmanis, S. et al. Simultaneous Multiplexed lecular Genetics 14, 1087–1094 (2005).
Measurement of RNA and Proteins in Single Cells. Cell 64. Miressi, F. et al. GDAP1 Involvement in Mitochon-
Reports 14, 380–389 (2016). drial Function and Oxidative Stress, Investigated in a
48. Assarsson, E. et al. Homogenous 96-Plex PEA Im- Charcot-Marie-Tooth Model of hiPSCs-Derived Motor
munoassay Exhibiting High Sensitivity, Specificity, and Neurons. Biomedicines 9, 945 (2021).
Excellent Scalability. PLoS ONE 9, e95192 (2014). 65. León, M. et al. Rapid degeneration of iPSC-derived
49. Bowen, C. A. et al. Proximity Labeling Proteomics motor neurons lacking Gdap1 engages a mitochondrial-
Reveals Kv1.3 Potassium Channel Immune Interactors sustained innate immune response. Cell Death Discov.
in Microglia. Molecular & Cellular Proteomics 23, 9, 217 (2023).
100809 (2024). 66. Mederos, S., González-Arias, C. & Perea, G. Astro-
50. Madisen, L. et al. A robust and high-throughput Cre cyte–Neuron Networks: A Multilane Highway of Signal-
reporting and characterization system for the whole ing for Homeostatic Brain Function. Front. Synaptic
mouse brain. Nat Neurosci 13, 133–140 (2010). Neurosci. 10, 45 (2018).
51. Zhang, Y. et al. An RNA-sequencing transcriptome 67. Breckels, L. M. et al. Advances in spatial prote-
and splicing database of glia, neurons, and vascular omics: Mapping proteome architecture from protein
cells of the cerebral cortex. J Neurosci 34, 11929– complexes to subcellular localizations. Cell Chemical
11947 (2014). Biology 31, 1665–1687 (2024).
52. Maier, T., Güell, M. & Serrano, L. Correlation of 68. Bhushan, V. & Nita-Lazar, A. Recent Advance-
mRNA and protein in complex biological samples. ments in Subcellular Proteomics: Growing Impact of Or-
FEBS Lett 583, 3966–3973 (2009). ganellar Protein Niches on the Understanding of Cell
53. Vogel, C. & Marcotte, E. M. Insights into the regu- Biology. J. Proteome Res. 23, 2700–2722 (2024).
lation of protein abundance from proteomic and tran- 69. Li, Y., Kanao, E., Yamano, T., Ishihama, Y. &
scriptomic analyses. Nat Rev Genet 13, 227–232 Imami, K. TurboID-EV: Proteomic Mapping of Recipient
(2012). Cellular Proteins Proximal to Small Extracellular Vesi-
54. Sharma, K. et al. Cell type– and brain region–re- cles. Anal. Chem. 95, 14159–14164 (2023).
solved mouse brain proteome. Nat Neurosci 18, 1819– 70. Mattick, J. S. et al. Long non-coding RNAs: defini-
1831 (2015). tions, functions, challenges and recommendations. Nat
55. Seyfried, N. T. et al. A Multi-network Approach Rev Mol Cell Biol 24, 430–447 (2023).
Identifies Protein-Specific Co-expression in Asympto- 71. Brockdorff, N., Bowness, J. S. & Wei, G. Progress
matic and Symptomatic Alzheimer’s Disease. Cell Sys- toward understanding chromosome silencing by Xist
tems 4, 60-72.e4 (2017). RNA. Genes Dev. 34, 733–744 (2020).
56. Johnson, E. C. B. et al. Large-scale proteomic anal- 72. Liu, J., Jang, J. Y., Pirooznia, M., Liu, S. & Finkel,
ysis of Alzheimer’s disease brain and cerebrospinal T. The secretome mouse provides a genetic platform to
 bioRxiv preprint doi: https://doi.org/10.1101/2025.01.29.635500; this version posted February 1, 2025. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY-NC-ND 4.0 International license.

delineate tissue-specific in vivo secretion. Proc. Natl. 92. Deutsch, E. W. et al. The ProteomeXchange con-
Acad. Sci. U.S.A. 118, e2005134118 (2021). sortium at 10 years: 2023 update. Nucleic Acids Re-
73. Wei, W. et al. Cell type-selective secretome profil- search 51, D1539–D1548 (2023).
ing in vivo. Nat Chem Biol 17, 326–334 (2021). 93. Perez-Riverol, Y. et al. The PRIDE database re-
74. Vasek, M. J. et al. Local translation in microglial sources in 2022: a hub for mass spectrometry-based
processes is required for efficient phagocytosis. Nat proteomics evidences. Nucleic Acids Research 50,
Neurosci 26, 1185–1195 (2023). D543–D552 (2022).
75. Pilaz, L.-J., Lennox, A. L., Rouanet, J. P. & Silver,
D. L. Dynamic mRNA Transport and Local Translation Acknowledgements
in Radial Glial Progenitors of the Developing Brain. Cur- Research reported in this publication was supported by
rent Biology 26, 3383–3392 (2016). the National Institute on Aging and the National Institute
76. Mazaré, N. et al. Local Translation in Perisynaptic of Mental Health of the National Institutes of Health:
Astrocytic Processes Is Specific and Changes after 1F31AG079597-01A1 (C.C.R.), R01AG075820
Fear Conditioning. Cell Reports 32, 108076 (2020). (N.T.S., S.R.), R01AG071587 (S.R.) and
77. Hobson, B. D. et al. Subcellular and regional local- R01MH125956 (S.A.S.). Research was also supported
ization of mRNA translation in midbrain dopamine neu- in part by the Emory Integrated Proteomics Core (EIPC)
rons. Cell Reports 38, 110208 (2022). which is subsidized by the Emory University School of
78. Herbert, A. L. et al. Dynein/dynactin is necessary Medicine and is one of the Emory Integrated Core Fa-
for anterograde transport of Mbp mRNA in oligodendro- cilities. Additional support was provided by the Georgia
cytes and for myelination in vivo. Proc. Natl. Acad. Sci. Clinical & Translational Science Alliance of the National
U.S.A. 114, (2017). Institutes of Health under Award Number
79. Boulay, A.-C. et al. Translation in astrocyte distal UL1TR002378 (EIPC). Additional support was provided
processes sets molecular heterogeneity at the gliovas- by the National Center for Advancing Translational Sci-
cular interface. Cell Discov 3, 17005 (2017). ences of the National Institutes of Health under Award
80. Sakers, K. et al. Astrocytes locally translate tran- Number UL1TR000454.
scripts in their peripheral processes. Proc. Natl. Acad.
Sci. U.S.A. 114, (2017). Author contributions
81. Zhang, Y. et al. Astrocyte metabolism and signaling Conceptualization: C.C.R., S.A.S., S.R., Methodology:
pathways in the CNS. Front. Neurosci. 17, 1217451 C.C.R., E.B.D., H.X., L.C., P.K., C.E.G., M.M.S.,
(2023). R.S.N., S.M., D.K., R.K., Q.G., P.B., D.M.D., N.T.S.,
82. Andrews, S. FastQC: A Quality Control Tool for S.A.S., S.R., Investigation: C.C.R., E.B.D, N.T.S.,
High Throughput Sequence Data. (2010). S.A.S., S.R., Writing-Original draft: C.C.R., S.A.S., S.R,
83. Dobin, A. et al. STAR: ultrafast universal RNA-seq Writing-Review and Editing: C.C.R., E.B.D., H.X., L.C.,
aligner. Bioinformatics 29, 15–21 (2013). P.K., C.E.G., M.M.S., R.S.N., S.M., D.K., R.K., Q.G.,
84. Liao, Y., Smyth, G. K. & Shi, W. featureCounts: an P.B., D.M.D., N.T.S., S.A.S., S.R., Funding acquisition:
efficient general purpose program for assigning se- C.C.R., N.T.S., S.A.S., S.R., Resources: N.T.S., S.A.S.,
quence reads to genomic features. Bioinformatics 30, S.R., Supervision: S.A.S., S.R.
923–930 (2014).
85. Love, M. I., Huber, W. & Anders, S. Moderated es- Competing interests
timation of fold change and dispersion for RNA-seq data N.T.S. and D.M.D are co-founders of Emtherapro and
with DESeq2. Genome Biol 15, 550 (2014). Arc Proteomics. N.T.S is co-founder of Stitch Rx.
86. Hebert, A. S. et al. Comprehensive Single-Shot
Proteomics with FAIMS on a Hybrid Orbitrap Mass
Spectrometer. Anal. Chem. 90, 9529–9537 (2018).
87. Cox, J. et al. Andromeda: A Peptide Search Engine
Integrated into the MaxQuant Environment. J. Prote-
ome Res. 10, 1794–1805 (2011).
88. Cox, J. & Mann, M. MaxQuant enables high peptide
identification rates, individualized p.p.b.-range mass ac-
curacies and proteome-wide protein quantification. Nat
Biotechnol 26, 1367–1372 (2008).
89. Durinck, S. et al. BioMart and Bioconductor = a
powerful link between biological databases and micro-
array data analysis. Bioinformatics 21, 3439–3440
(2005).
90. Durinck, S., Spellman, P. T., Birney, E. & Huber, W.
Mapping identifiers for the integration of genomic da-
tasets with the R/Bioconductor package biomaRt. Nat
Protoc 4, 1184–1191 (2009).
91. Hubbard, T. J. P. et al. Ensembl 2009. Nucleic Ac-
ids Research 37, D690–D697 (2009).