analytical. . 'chemistry pubs.acs.org/ac Single Cerebral Organoid Mass Spectrometry of Cell-Specific Protein and Glycosphingolipid Traits Markéta Nezvedova,* Durga Jha,# Tereza Váňová, Darshak Gadara, Hana Klímová, Jan Raška, Lukáš Opálka, Dáša Bohačiaková, and Zdeněk Spáčil* Cite This: Anal. Chem. 2023, 95, 3160-3167 Read Online ACCESS lihl Metrics & More HÜ Article Recommendations © Supporting Information LC-MS/MS analysis Cerebral organoid development y FABP7 Proteins NE AO s> TB NEFM SCX2 MAP2. TTR GFAP TUBB3 : CD44| -1|J GAPDH S100B i ; 3 j : s f ! 4 10 l'l ounu vi Acaunita , rt 15 16 17 13 ,S 20 CORTICAL NEUROGENESIS A B S T R A C T : Cerebral organoids are a prolific research topic and an emerging model system for neurological diseases i n human neurobiology. However, the batch-to-batch reproducibility of current cultivation protocols is challenging and thus requires a highthroughput methodology to comprehensively characterize cerebral organoid cytoarchitecture and neural development. W e report a mass spectrometry-based protocol to quantify neural tissue cell markers, cell surface lipids, and housekeeping proteins i n a single organoid. Profiled traits probe the development of neural stem cells, radial glial cells, neurons, and astrocytes. W e assessed the cell population heterogeneity i n individually profiled organoids i n the early and late neurogenesis stages. Here, we present a unifying view of cell-type specificity of profiled protein and lipid traits i n neural tissue. O u r workflow characterizes the cytoarchitecture, differentiation stage, and batch cultivation variation o n an individual cerebral organoid level. Cerebral organoids ( C O s ) generated f r o m i n d u c e d pluripotent stem cells (iPSCs) are an emerging i n vitro model system i n neurobiology.1 C O s recapitulate human brain cytoarchitecture a n d cell diversity during neurogenesis, mimicking brain development i n three dimensions.2 C O s are increasingly used to model diseases-in-the-dish with recent viral applications toward Zika virus or S A R S - C O V - 2 . 4 However, current cultivation protocols are notorious for substantial intraand interbatch variation i n differentiation, morphology, and cell composition.5 CO-based disease models expanded our ability to study neurodevelopment and degeneration via cell lineage-specific protein and lipid markers (Table SI and Figure S i ) . A t the early cortical neurogenesis stage, neural stem cells ( N S C s ) differentiate into radial glial cells ( R G C s ) , giving rise to neurons, astrocytes, and oligodendrocytes.5 R G C s divide asymmetrically to generate neurons directly or indirectly through intermediate progenitor cells (IPCs), later differentiating symmetrically into immature neurons.5 N S C s express the early neurogenesis marker, a transcription factor S O X 2 . S O X 2 is downregulated in post-mitotic neurons. Glial hallmarks (fatty acid-binding protein, F A B P 7 ) begin to emerge during later differentiation simultaneously with primary astrocyte markers—calcium-binding proteinB (S100B), glial fibrillary acidic protein ( G F A P ) , and C D 4 4 antigen.6 Astrocytes express S100B during the proliferative and migration phase.' A microtubule-associated protein 2 ( M A P 2 ) i n neurons' dendrites and reactive astrocytes stabilizes the microtubules against depolymerization.8 Tubulin beta-3 chain ( T U B B 3 ) , the principal constituent of microtubules i n neuronal axons, and microtubule-associated protein doublecortin ( D C X ) are characteristic of the immature neuronal population. D C X ceases with neuronal maturation.9 Mature neurons express neurofilaments containing intermediate filament proteins ( l i g h t — N E F L , m e d i u m — N E F M ) and synapsin- 1 ( S Y N l ) . T h e choroid plexus's epithelial cells representing the non-neuronal cells express the transthyretin ( T T R ) . 9 - 1 1 Received: March 1, 2022 Accepted: January 16, 2023 Published: February 1, 2023 A / - C n . . U K . . 4 - 1 ^ ^ ^ e 2023 American Chemical Society https://doi.org/10.1021/acs.analchem.2c00981 A L o rUDIIC3TIOnS 3i60 Ami»em.2023,95,3100-3157 Analytical Chemistry pubs.acs.org/ac Article (T) Harvesting organoids (2) Mass spectrometry analysis (3) Data processing J J .3 j j . u j j j •••••••••lilii •••1 Single step extraction (80% I PA) Lipid extract Cell culture 96-well plate Single organoid for LC-MS/MS analysis w Protein pellet IProteolysis 1Multiplex targeted proteomics and lipidomics 0 Lipid 0 protein 0 Protein ° Protein 0 Protein Cell-specific protein and lipid markers Harvested individual organoids embedded in Geltrex C RS treatment Geltrex-free organoid 0.005c 0.004- o 2 0.003o 0.002- o O 0.001- 0.000Outlier removal Figure 1. Single cerebral organoid mass spectrometry-based protein and lipid profiling. The workflow overview. Cerebral organoids were harvested after 48, 76, 95, 110, 135, and 160 days of differentiation, treated with the cell recovery solution to remove the cell culture matrix, and lipids were extracted using 80% IPA, and the protein pellet was subjected to the bottom-up S R M protein assays. Single-organoid protein and lipid profiling was the basis for cell-specific population characterization and the outlier removal to mitigate the intra- and interbatch variability. Like proteins, lipids constitute the primary structural aspect of neuronal membranes. M a j o r cerebral lipids consist o f phospholipids, glycolipids, cholesterol, and triglycerides. H o w ever, our work focused o n membrane glycosphingolipids, particularly gangliosides, i n examining primary neural development and maturation traits as they parallel protein cell-specific markers. Gangliosides are ubiquitous i n vertebrate tissues and highly abundant i n neural cells, essential for cellular signal transduction, adhesion, proliferation and differentiation, i m mune response, and apoptosis.1 2 Neuronal membranes and myelin sheaths contain 10—12% of gangliosides arranged in microdomains, referred to as lipid rafts.1 3 T h e perturbed composition of neuronal gangliosides i n the membrane triggers neurodegeneration.1 4 T h e gangliosides' distribution is associated with specific cell types and characterizes the cortical neurogenesis stage and cytoarchitecture i n C O s . Cell-specific protein markers are frequently profiled i n C O s using antibody-based immunoaffinity assays, i.e., E L I S A , Western Blot, or immunofluorescence staining.1 5 However, the quantitative performance, robustness, multiplexing capacity, and throughput of immunoaffinity assays are l i m i t e d . 1 6 Similarly, the thin-layer chromatography ( T L C ) immunostaining, liquid or gas chromatography ( L C / G C ) methodology to probe lipid composition often lacks sensitivity and selectivity.1 7 Few studies utilized organoid sections for immunostaining but struggled with the lack of diversity in information regarding the lipid subclasses.1 8 '1 9 Organoids were often pooled before T L C analysis, hiding the level of heterogeneity.2 0 O n the contrary, mass spectrometry ( M S ) proteomics2 1 and lipid profiling2 2 via selected reaction m o n i t o r i n g ( S R M ) assays are h i g h l y reproducible and quantitative. W e present a workflow to simultaneously profile cell-specific protein markers and glycosphingolipids i n a single cerebral organoid to characterize cytoarchitecture and to identify outliers and the batch-to-batch variation5 (Figure l ) . W e used the bottom-up S R M protein assays, selecting surrogate proteotypic peptides to generate an S R M library. A s the consensus on proteotypic peptide selection is missing, we report on the design of S R M protein assays (Figure S2). • EXPERIMENTAL SECTION Lipid Extraction for Mass Spectrometry Assays. C O s harvested for S R M analysis were immediately washed with PBS, treated (4 ° C ; 1 h) with cell recovery solution ( C R S , Corning, N e w Y o r k ) , and washed again. C O s were freeze-dried (y 1—16 LSCplus, M a r t i n Christ G m B H , Germany) and stored at —80 °C until further processing. F o r lipid and protein analysis, a single C O was used; biological replicates (n = 4) per time point were analyzed i n duplicates. Freeze-dried C O was homogenized by adding 100 ßL of water i n a Protein L o B i n d (Eppendorf, Germany) microtube with a glass bead (Benchmark Scientific, Edison, N e w Jersey), sonicated, and vortexed. T h e homogenate was centrifuged briefly, and 10 ßL of the supernatant was used to determine the total protein content b y the B C A assay. T h e remaining homogenate was dried (Savant S D P 121 P, SpeedVac, Thermo Fisher Scientific). For lipid extraction, we added 100 ßL of 80% I P A to the dry homogenate, vortexed ( l min), sonicated (37 H z , 5 min), and mixed (10 min, 2000 rpm). T h e sample was centrifuged (12.3 R C F for 5 min), and 85 ßL of the lipid extract was removed from the residual protein pellet. Lipid extracts were stored at —20 ° C until analysis. After lipid extraction, protein pellets were dried (SpeedVac, 37 ° C ) and processed for S R M protein assays (Figure l ) . Mass Spectrometry Ganglioside Assays and Data Processing. L i p i d extracts were twofold diluted by adding 0.3 of isotopically labeled G M 1 and G M 3 i n 10% isopropanol 3161 https://doi.org/10.1021/acs.analchem.2c00981 Anal. Churn. 2023, 95, 3160-3167 Analytical Chemistry pubs.acs.org/ac Article (IPA). Sample volume (2 (iL) was injected i n a U H P L C system (1290 Infinity II; Agilent Technologies, California) equipped with C 1 8 precolumn and analytical column ( C S H T M , 5 X 2.1 m m 2 X 1.7 fim and 50 X 2.1 m m 2 X 1.7 fim from Waters C o r p ) thermostated at 40 ° C . U H P L C system was coupled to a triple quadrupole mass spectrometer (Agilent 6495B, A g i l e n t Technologies). The mobile phase for the positive ion mode analysis consisted of buffer A (0.5 m M ammonium fluoride i n the water) and B (methanol: I P A (50:50 v / v ) ) . T h e gradient elution (17.1 m i n ) at a flow rate of 0.3 m L / m i n was 30% B for 2 min, 70% B from 2 to 9 min, 95% B maintained from 9 to 13.3 min, 5% B at 13.3 min, and 5% B at 14.3 m i n with re-equilibration from 14.5 to 17.1 m i n at 30% B. T h e electrospray source capillary voltage was 3500 V , and the i o n source parameters for positive i o n mode were: gas flow rate 16 L / m i n at 190 °C, sheath gas pressure 20 PSI at 350 °C, and nozzle voltage 1300 V . The mobile phase for the negative ion mode analysis consisted of buffer A (0.5 m M a m m o n i u m fluoride and 10 m M ammonium acetate i n water) and B (acetonitrile: I P A (50:50 v / v ) ) . T h e gradient elution (19.1 min) at a flow rate of 0.3 m L / m i n was 10% B for 4 min, 85% B from 4 to 6.2 min, 95% B maintained from 6.2 till 10.2 min, and changed to 10% B at 10.4-14.4 min, 95% B from 14.4 to 16.2 min, maintained till 16.4 m i n with re-equilibration from 16.4 to 19.1 m i n at 10% B. The E S I source capillary voltage was 3000 V , and the ion source parameters for negative ion mode were: gas flow rate 14 L / m i n at 190 °C, sheath gas pressure 25 PSI at 400 °C, and nozzle voltage 1500 V . Commercial 1 3 C isotopically labeled standards for gangliosides are not commercially available. Protocols for1 3 C 1 8 labeled G M 1 and G M 3 gangliosides in-house synthesis are i n the supporting information and respective mass spectra in Figure S3. W e used the labeled G M 3 internal standard to determine the concentration of all gangliosides, except for G M 1 , determined using the corresponding labeled G M 1 internal standard. Respective response factors ( R F ) to the labeled G M 3 were calculated for all gangliosides. W e processed raw data i n Skyline (Version 20.1.0.76, M a c C o s s Lab., U W ) . A l l concentrations are the average of technical duplicates relative to the A C T B level. The S R M library is shown i n Table S2. Chromatograms for all ganglioside species and internal standards are shown i n Figure S4a. Protein Extraction and Enzymatic Proteolysis. After lipid extraction, the dried protein pellet with a glass bead was powdered (4 m / s , 10 s, two cycles with 10 s inter-time, BeadBlasterTM 24, Benchmark), solubilized i n the ammonium bicarbonate ( A m B i c ) buffer (50 m M ) with sodium deoxycholate (5 m g / m L ) , 2 3 vortexed (10 s, 2000 r p m , V E L P Scientifica), mixed (10 min, 2035 rpm, H e i d o l p h T M M u l t i Reax), and sonicated ( l m i n , 80 k H z , Elmasonic P, E l m a Schmidbauer G m b H ) . T h e total protein concentration was adjusted toO.Sfig/fiLby adding the A m B i c buffer. Samples were centrifuged ( l min, 12300 R C F , Micro-Star 12, V W R , Radnor, Pennsylvania), and the volume of 60 flL (equivalent to 30 fig of total protein) was used to reduce (20 m M D T T i n 2.5 m M A m B i c ; 10 min; 95 ° C ) and alkylate (40 m M I A A i n 2.5 m M A m B i c ; 30 m i n ; ambient, i n the dark) proteins. T h e remaining volume of individual C O homogenates was pooled into a quality control ( Q C ) sample. Identical to the analysis of individual C O s , we used 60 flL aliquots of the Q C sample (30 fig of protein). Trypsin was added i n the ratio of 1:60 (enzyme: total protein content, w / w ) , and the Parafilm sealed samples were incubated (37 ° C ; 16 h ; gentle shaking). The trypsin digestion efficacy was tested i n Q C samples after 2, 4, and 16 h (Figure S5). T h e isotopically labeled ( S I L ) synthetic peptides were added (sample cone. « 260 n m o l / L ) before quenching the digestion with 200 flL of 2% formic acid ( F A ) . Samples were centrifuged (5 min, 12300 R C F ) , and the supernatant was loaded o n the mixed-mode cartridge (Oasis P R i M E H L B - 30 mg, Waters Corp. Milford, Massachusetts) for solid-phase extraction (SPE). Peptides were washed with 2% F A and eluted with 500 fcL of 50% acetonitrile ( A C N ) with 2% F A , and the samples were dried in SpeedVac. S I L standard peptides ( S T ) response i n the Q C sample before and after the S P E was compared to determine the S P E recovery for tryptic peptides: peak area of ST (before S P E ) / peak area of ST(after S P E ) X 100. T h e average S P E recovery was 87% for all 14 quantifier proteotypic peptides (Figure S6a and Table S3). Mass Spectrometry Protein Assays and Data Processing. Dried SPE-purified peptides were reconstituted in 15 flL of 5% A C N with 0.1% F A . T h e Q C sample homogenates with 40 fig total protein were reconstituted in 60,40, and 20 flL to load 2, 3, a n d 6 fig total protein equivalent to U H P L C - S R M , respectively (Figure S6c,d). Peptides were analyzed i n positive ion detection mode using the same U H P L C - M S system as for ganglioside assays. A sample volume (3 flL, equivalent to 6 fig of total protein) was injected into the C 1 8 analytical column (Peptide C S H 1.7 fim, 2.1 X 100 m m 2 , Waters Corp., Milford, Massachusetts). T h e mobile phase flow rate was 0.3 m L / m i n ; b u f f e r A ( 0 . 1 % F A ) and buffer B (0.1% F A in 95% A C N ) . Linear gradient elution: initial 5% B ; 25 m i n 30% B ; 25.5 m i n 95% B ; 30 m i n 95% B ; and from 31 to 35 m i n with 5% B. T h e E S I source temperature was 200 °C, and the capillary voltage was 3500 V . S R M protein assays were designed utilizing the neXtProt database (online, www.nextprot.org) to select proteotypic peptides (2—4 per protein), preferably w i t h experimental evidence i n the PeptideAtlas. S R M library (3—4 transitions per proteotypic peptide) was selected i n the S R M A t l a s (www. srmatlas.org), (Figure S2). T h e dwell time (10 ms) and a cycle time (<1 s) allowed for up to 100 transitions i n every acquisition method. W e tentatively identified peptides in Q C samples using a retention time prediction model and verified the identifications using isotopically labeled synthetic analogues. W e used a dynamic S R M ( d S R M ) mode with a 2 min-wide window centered at a peptide experimental retention time i n the Q C sample. W e relatively quantified target proteins preferably using >5 j-ions with peak area >10 000 and reproducible response across technical duplicates (% coefficient of variation ( C V ) < 15), as shown i n Figure S2. T h e d S R M assay included 251 transitions to monitor 41 unique peptides of 18 proteins (Table 54) . T h e lowest total protein content i n analyzed C O s (n = 24) was 30 fig. Data were processed i n Skyline and manually inspected, Figure S4b. A single quantifier transition (Table S4) was used to determine relative concentrations (light peptide peak area/ST peptide peak area X S T peptide concentration). T h e protein levels i n individual samples are reported as an average of technical duplicates normalized to A C T B levels. Ganglioside and Protein Assays Validation. Detailed information o n assay validation is described i n the Supporting Information: ganglioside assay validation and protein assay validation. F o r gangliosides, 10-point matrix-matched calibration curves were prepared and analyzed (Figure S7 and Table 55) . Precision was <12.1% of % C V (Table S5b), ganglioside recovery was high (>82.3%), and matrix effects were negligible 3162 https://doi.org/10.1021/acs.analchem.2c00981 Anal. Chem. 2023, 95, 3160-3167 Analytical Chemistry pubs.acs.org/ac Article o u Q. 1 6 0 1 140 120 100 80 60 40 20 10 Extracted pellet I PA extract ill idelf <£> <$ & $ & ^ j r Jß ^ J ? d> J >^ v ^ ^ ^ 4 Non- treated CRS- treated i i r Figure 2. Analytical figures of merit of the mass spectrometry-based workflow, (a) Loss of targeted proteins to isopropanol (IPA) after lipid extraction. The protein content in the residual pellet and the IPA extract was compared to the total protein amount in the homogenate not subjected to IPA extraction. The result is expressed as protein yield in %. Four target proteins were detected in the IPA extract, and the protein loss was <5%. (b) Geltrex removal using cell recovery solution (CRS). Housekeeping protein levels were analyzed in 30 /ig of processed cerebral organoid total protein (n = 2). O n average, 2-fold higher levels were found in CRS-treated organoids relative to untreated. DO I t Cell plating *9 D15 —I—• D45 —I— Cultivation on shaker in maturation medium t t Sample collection at selected time points D160 5 SOX2 DCX SOX2 J 20 5 15 «? 10 TUBB3 O < MAP2 SYN1 NEFL S100B G FAP ACTB 60 40 20 0 0 4 B 0 3 f 0.2 N 0.1 6000 O 4000 < S, 2000 U l i TUBB3 .11 GFAP L a TTR 1.1 ö «S «A A v 0.99 linear response range 1.02-81.25 n M for S O X 2 , 1.02-1300 for A C T B , G A P D H , and 1.02 or 5 . 0 8 - 3 2 5 n M for other proteins. The matrix effects were moderate, o n average 32% (Figure S6b and Table S3), and signal reproducibility i n the sample matrix was <11 % C V (Table S3). Data Analysis and Visualization. Cluster analysis for protein and lipid markers was prepared i n MataboAnalyst 5.0 (online, https://www.metaboanalyst.ca/ (2021)). T h e graphs were prepared using G r a p h P a d P r i s m version 8.0.2 for W i n d o w s , G r a p h P a d Software, California (www.graphpad. com). Figures 1—4a, SI, S2, S4, S6, S10, and S l l were created with BioRender.com. • RESULTS AND DISCUSSION Characterization of Cell-Specific Markers via qPCR, Immunoblotting Assay, and Indirect Immunofluorescence. W e used a protocol modified b y Lancaster et al.2 to differentiate C O s 2 (Figure 3a). A n average D 8 5 C O can range from 3 to 5 m m i n diameter and consists of 2.5 million cells (Figure 3b). T h e C O s ' morphology o n D 8 5 was characterized 3163 https://doi.org/10.1021/acs.analchem.2c00981 Anal. Churn. 2023, 95, 3160-3167 Analytical Chemistry pubs.acs.org/ac Article Intra-batch variability g? 200-1 £ 100 J- Mean: 46% - 1 — I — I — I — I — [ — Mean: 46% ••• •.. :.; •'• •.. Outlier removal ZUU-i 150-^ 100 JNEFM GM1 GD2 100 50 Inter-batch variability I | Batch 01 • Batch 02 ID ra 15u! 10- uj z D) 5- C 0 100T T 50^ Q 1 0. s 20 j f 15- 1 10z 5- 0individual samples Outlier removal I | Batch 01 • Batch 02 individual samples TTR GM3 Days N e u r o n s 76 95 110 135 160 Days 2500- 2000- 4S 76 95 110 135 160 Days Q 3 e Q n — i — i — i Astrocytes 95 110 135 160 Days a id9 76 95 110 135 160 Days Non-neuronal cells 76 95 110 135 160 Days GM1 1000 1500 GD2 r =0.61 p<0.0001 600 800 Figure 4. Single cerebral organoid characterization by mass spectrometry assays for cell-specific protein and lipid markers, (a) Intra-batch variability of target proteins and lipids in organoids from timeline experiment before and after the application of outlier removal, (b) Interbatch variability of neuronal population in organoids from two cultivation batches, (c) Single-organoid time trends in levels of specific traits (after outlier removal) for neurons ( N E F M , G M l ) , astrocytes (CD44, GD2), and non-neuronal cells (TTR, G M 3 ) , n = 3 per time point. A significant increase in neuronal and astrocyte populations was visible (*p-value < 0.05, **p-value < 0.01). (d) Correlation plots for protein and lipid markers for neurons ( N E F M , G M l ) , astrocytes (CD44, GD2), and non-neuronal cells (TTR, G M 3 ) . by indirect immunofluorescence. T h e cell-specific marker expression was assessed via W B (Figure 3c) and q P C R (Figure 3d), pooling 5—7 C O s per assay. Consistently with the previous reports,2 '2 4 we demonstrate the expression of markers for neuroectodermal cells ( S O X 2 , P A X 6 ) , neurons ( M A P 2 , T U B B 3 , D C X , and N E U N ) , deep-layer neurons ( C T I P 2 ) , synaptic junctions ( S Y N l ) , neurofilaments (light chain, N E F L ) , and astrocytes (S100B, G F A P ) . T h e protein expression of S O X 2 reached a maximum o n D 5 0 and later declined. Neuronal (i.e., M A P 2 , D C X , and T U B B 3 ) markers and astrocytic G F A P were 3164 https://doi.org/10.1021/acs.analchem.2c00981 Anal. Chem. 2023, 95, 3160-3167 Analytical Chemistry pubs.acs.org/ac Article at the maximum level o n D U O . Detailed information o n cultivation a n d analysis is described i n the Supporting Information. Extraction of Gangliosides and Proteins from Cerebral Organoids for Mass Spectrometry Assays. C O s were C R S treated to remove the Geltrex matrix before M S analysis (Figure l ) . Isopropanol ( I P A ) was added to homogenized C O s to extract gangliosides and precipitate proteins. T h e protein loss due to I P A extraction was <5% (Figure 2a). W e compared housekeeping protein ( H K P ) levels i n CRS-treated a n d nontreated C O s to assess the efficiency of matrix removal. H K P levels (Figure 2b) and gangliosides' internal standards' signals (Figure S9) i n CRS-treated samples were up to 2-fold higher than i n nontreated samples. T h e enriched cellular proteins and gangliosides i n CRS-treated C O s improved assay sensitivity. Heterogeneity in Cerebral Organoids. W e profiled cell populations using protein and lipid markers i n individual C O s after 48, 76, 95, 110, 135, and 160 days of differentiation (Figures 4, S10, and S l l ) . C O s were analyzed individually at each time point (n = 4) to remove one outlier per time point (n = 3). Detailed information o n heterogeneity is described i n the Supporting Information: Heterogeneity i n individual cerebral organoids. T h e outlier removal reduced C V within the batch from 46 to 34 and 46 to 30% for protein and lipid markers, respectively (Figure 4a). After outlier removal, we observed stronger correlations between markers, such as G D 2 vs C D 4 4 (before r = 0.45, p = 0.0003 and after outlier removal r = 0.71, p < 0.0001) and G M 3 vs T T R (before r = 0.32, p = 0.0042 and after outlier removal r = 0.61, p < 0.0001); data not shown. In addition, we analyzed two batches of C O s (n = 5) derived from the same cell line and harvested at identical time points to perform an interbatch variability analysis. T h e interbatch variability was reduced substantially after outlier removal (Figure 4b), which is not feasible i n pooled samples. The protein and lipid markers panel were characterized i n individual C O s , and results are shown after outlier removal (n = 3 per time point) (Figures 4c and S l l ) . Cell-Specific Protein Expression in Cerebral Organoids. Cell markers f o r N S C s , radial glial cells, neurons, astrocytes, and the ubiquitously present housekeeping proteins were relatively quantified i n C O s . T h e total cell mass estimated using H K P (i.e., G A P D H and A C T B ) levels reached a maximum between D 7 6 and D 9 5 (Figure S12). O n D 4 8 , S O X 2 was the most abundant marker in C O s and later downregulated (Figure S l l ) . In parallel with the S O X 2 decline, the expression of R G C s marker F A B P 7 increased until D 7 6 and later remained steady (Figure S l l ) . T T R expression attributed to the choroid plexus epithelial cells reached a maximum i n D 9 5 (Figure 4c). Neuronal markers' expression increased from D 7 6 until D U O , followed b y a steady state or decline, while astrocyte markers' expression increased until D160 (Figures 4c and S l l ) . N e u r o n specific proteins D C X , T U B B 3 , M A P 2 , N E F L , and N E F M , emerged early (D48), culminated o n D U O , and later declined, except for M A P 2 (Figure S l l ) . T h e mature neurons' marker S Y N 1 emerged from D 9 5 (Figure S l l ) . T h e astrocytic markers (S100B, G F A P , and C D 4 4 ) , negligibly expressed o n D 4 8 , gradually increased until D 1 6 0 (Figures 4c and S l l ) . W e characterized some protein markers using W B and q P C R assays to align with the reported L C - M S - b a s e d workflow (Figure 3c,d) and previous studies demonstrating the development of neuron and astrocyte populations to mimic the neurogenesis in v i v o . 1 0 ' 2 4 W B and M S assays identically show the highest N S C s ' population ( S O X 2 ) at an early stage (48D) of C O s ' proliferation (Figures 3c and S l l ) . Temporal trends of neuronal markers (i.e., D C X , T U B B 3 , M A P 2 ) a n d astrocytic markers (i.e., S100B, G F A P ) determined by W B and q P C R mainly agreed with M S based assays, except for q P C R assessed S100B and G F A P showing an earlier onset (Figure S I 3 ) . However, only a limited number of cell-specific markers can be determined i n pooled C O s b y immune-based a s s a y s 2 5 - 2 7 without assessing the variability i n individual C O s . O n the other hand, the S R M assay allows the characterization of multiple analytes i n a single organoid w i t h high specificity a n d multiplexing capability for protein quantification. Membrane Glycosphingolipids in Cerebral Organoids. Apart from gangliosides, the lipid extract was utilized to monitor other major lipid species. W e characterized 351 lipid species from over 24 lipid classes composed of cholesterol, phospholipids, lysophospholipids, ceramides, sphingolipids, triacylglycerols, and carnitines (Figure S14). Gangliosides G M 1 , G M 2 , G M 3 , G D I a, G D l b , G D 2 , G D 3 , and G T l b are abundant i n the nervous tissue.2 8 T h e monosialo G M 3 a n d disialo G D 3 represent N S C s markers.2 9 G D 3 was the most abundant ganglioside i n the C O s (Figure S l l ) , and the levels of G D 3 and G M 3 remained steady at all time points, indicating high N S C reserve even at a late stage of C O proliferation, an analogy with mature brain tissue.3 0 G D 3 interacts with the epidermal growth factor receptor ( E G F R ) and induces neural precursor cell differentiation and neurite formation.3 1 W e observed a progressive increase i n complex neuronal gangliosides from D 4 8 until D U O , followed by a decline i n D135 and D 1 6 0 (Figures 4c and S l l ) . T h e biosynthesis switch possibly indicates the neuronal differentiation stage from G D 3 and G M 3 to complex neuronal gangliosides (i.e., G D l a , G D l b , G T l b , and G M l ) , involved i n signaling neurogenesis and astrocytogenesis.3 2 G M 2 and G D 2 have been associated with astrocytes.3 3 '3 4 G D 2 and G M 2 levels increased gradually i n C O s , with a maximum at D160, paralleled b y astrocyte protein markers, alluding to their possible colocalization i n astrocytes (Figures 4c,d, S l l , and S15 and Table S l l ) . • CONCLUSIONS C O s have been increasingly used as a brain model. However, the 3 D cell cultures suffer from the "batch effect" caused b y variations i n the differentiation, m o r p h o l o g y , a n d cell composition.5 H i g h intra- and interbatch differences limit the reproducibility of experiments and may induce false discoveries. W e developed a mass spectrometry-based profiling of cellspecific proteins (Table S i ) and lipid traits with high selectivity, sensitivity, and reproducibility i n a single C O (Figures 4c and S l l ) . L C - M S can characterize a single cerebral organoid and may be applied repeatedly using different L C separation conditions and S R M assays to profile hundreds of analytes quantitatively. Pre-analytically, we removed the organoid matrix to mitigate a nonspecific binding of small molecules and peptides to cell culture m e d i a , 3 5 reducing interferences with L C M S analysis3 6 (Figure l ) . W e presented a systematic workflow for relative protein quantification (Figure S2). W e demonstrated that the characterization of individual C O s using a panel of cellspecific protein markers and lipid traits could be used to reduce intra-batch a n d interbatch variability post-analytically b y discarding results from abnormally differentiated cerebral organoids. However, our method requires analyzing 3—5 C O s per group/condition to identify outliers, which may lead to extensive cell culture. 3165 https://doi.org/10.1021/acs.analchem.2c00981 Anal. Chem. 2023, 95, 3160-3167 Analytical Chemistry pubs.acs.org/ac Article Our study's protein and lipid traits characterized for various cell populations demonstrate the requisite complexity of C O s to mimic neurodevelopment and aging features. Despite the heterogeneity, our characterization protocol shows the potential of C O s as a model for the neurobiology of human neurological disorders. • ASSOCIATED CONTENT O Supporting Information The Supporting Information is available free of charge at https://pubs.acs.org/ doi/10.1021/acs.analchem.2c00981. A d d i t i o n a l experimental details o n chemicals and reagents; cerebral organoid cell culture; R N A isolation, c D N A synthesis, and real-time quantitative P C R ( q P C R ) assay; immunoblotting assay for cell-specific markers; indirect immunofluorescent staining; ganglioside and protein assay validation; cluster analysis of cell-specific protein markers and glycosphingolipid traits; assessment of heterogeneity i n individual cerebral organoids; major lipids characterization; localization of neuronal ( D C X , T U B B 3 , M A P 2 , N E F L , N E F M , S Y N l ) and astrocytic (S100B, C D 4 4 , G F A P ) markers and housekeeping proteins ( A C T B , G A P D H ) (Figure S i ) ; protein identification and relative quantification workflow (Figure S2); mass spectra for the 1 3 C labeled G M 3 and G M 1 synthesized in-house (Figure S3); representative chromatograms for lipids and proteins (Figure S4); trypsin digestion efficiency after 2, 4, and 16 h of incubation (Figure S5); analytical parameters of proteomic protocol (Figure S6); calibration curves for ganglioside isotopically labeled standards (Figure S7); calibration curves of synthetic isotopically labeled peptide internal standards used for protein assays (Figure S8); cell recovery solution ( C R S ) treatment to remove Geltrex sample matrix (Figure S9); representative chromatograms of a peptide and a ganglioside i n blank, Q C , C O D 4 8 , and C O D 1 6 0 samples (Figure S10); the time trends of neuron-specific markers ( D C X , N E F L , T U B B 3 , M A P 2 , S Y N l , G D I a, G D l b , and G T l b ) ; a marker of choroid plexus epithelial cells ( T T R ) ; N S C s marker ( G D 3 ) ; R G C s marker ( F A B P 7 ) ; and astrocyte-specific markers ( S 1 0 0 B , G F A P , and G M 2 ) (Figure S11); housekeeping proteins A C T B and G A P D H represent the total cell mass during cerebral organoid proliferation (Figure S12); temporal trends of selected protein markers comparing reference methods (i.e., W B , q P C R ) with M S assays (Figure S13); the time trends of major lipids i n cerebral organoids (Figure S14); the cluster analysis of proteins and gangliosides at different stage of neurogenesis (Figure S15); variability scores the sum of lipid markers levels, the sum of protein markers levels, and neurons to glia ratio of individually profiled cerebral organoids (Figure S16); protein markers profiled i n cerebral organoids (Table S i ) ; selected reaction monitoring ( S R M ) library of precursor/ product i o n transitions for analyzing gangliosides in positive and negative i o n detection modes (Table S2); synthetic isotopically labeled peptides for protein assays: signal reproducibility, recovery, and matrix effects (Table S3); S R M library for synthetic isotopically labeled peptides for protein assays (Table S4); calibration curve for ganglioside assays (Table S5); ganglioside internal standards: signal reproducibility, recovery, and matrix effects (Table S6); calibration curve parameters for protein assays (Table S7); the variability i n quantitation of protein and lipid markers before (a) and after (b) removing outlier cerebral organoids (Table S8); the algorithm to assess heterogeneity i n cerebral organoids (Table S9); a score to assess potential sources of heterogeneity i n cerebral organoids (Table S10); correlation coefficient for the protein and lipid markers (Table S l l ) ; list of q P C R primers (Table S12); and proteomics protocol reproducibility (Table S13) ( P D F ) • AUTHOR INFORMATION Corresponding Author Zdeněk Spáčil — RECETOX, Faculty of Science, Masaryk University, Brno 625 00, Czech Republic; orcid.org/0000- 0002-7505-4332; Phone: (+420) 549 49 7989; Email: spacil@recetox.muni.cz, spacil@iu.washington.edu Authors Markéta Nezvedova — RECETOX, Faculty of Science, Masaryk University, Brno 625 00, Czech Republic Durga Jha — RECETOX, Faculty of Science, Masaryk University, Brno 625 00, Czech Republic Tereza Váňová — Department of Histology and Embryology, Faculty of Medicine, Masaryk University, Brno 625 00, Czech Republic; International Clinical Research Center (ICRC), St. Anne's University Hospital, Brno 656 91, Czech Republic Darshak Gadara — RECETOX, Faculty of Science, Masaryk University, Brno 625 00, Czech Republic; orcid.org/0000- 0002-3141-8990 H a n a Klímová — Department of Histology and Embryology, Faculty of Medicine, Masaryk University, Brno 625 00, Czech Republic Jan Raška — Department of Histology and Embryology, Faculty of Medicine, Masaryk University, Brno 625 00, Czech Republic Lukáš Opálka — Department of Chemistry, Faculty of Pharmacy, Charles University, Hradec Králové S00 OS, Czech Republic Dáša Bohačiaková — Department of Histology and Embryology, Faculty of Medicine, Masaryk University, Brno 625 00, Czech Republic; International Clinical Research Center (ICRC), St. Anne's University Hospital, Brno 656 91, Czech Republic Complete contact information is available at: https://pubs.acs.org/10.1021/acs.analchem.2c00981 Author Contributions # M . N . and D.J. contributed equally. M . N . and D.J.: L C / M S analysis, validation, data processing, and manuscript preparation. T . V . : cerebral organoid differentiation, immunofluorescence; H . K . : cerebral organoid cell culture, western blot analysis; J . R : q P C R and data analysis; D . B . : data consultation, interpretation, and the final reading of the manuscript. Z.S.: conceptualization, methodology, supervision, writing, editing, and final approval of the manuscript. A l l authors have approved the final version of the manuscript. Notes T h e authors declare no competing financial interest. • ACKNOWLEDGMENTS T h e authors thank D r . Gabriela Přibyl Dovrtelova ( R E C E T O X , Faculty of Science, M U ) and D r . H a n a Hribkova (Faculty of Medicine, M U ) for the initial experiments. T h e results of the 3166 https://doi.org/10.1021/acs.analchem.2c00981 Anal. Chem. 2023, 95, 3160-3167 Analytical Chemistry pubs.acs.org/ac Article project were created with the financial support of the provider of the Grant Agency of Masaryk University ( G A M U project N o . M U N I / G / 1 1 3 1 / 2 0 1 7 ) , the Czech Health Research C o u n c i l ( A Z V project N o . NV19-08-00472), the R E C E T O X research infrastructure (the Czech Ministry of Education, Youth, and S p o r t s - M E Y S , L M 2 0 1 8 1 2 1 ) , b y M E Y S , CZ.02.1.01/0.0/0.0/ 17_043/0009632, b y the Czech Science Foundation ( G A C R ; no. 18-25429Y, GA20-15728S, and 21-21510S), and b y the European Union's H o r i z o n 2020 research and innovation program under Grant Agreement N o . 857560 C E T O C O E N E X C E L L E N C E . This publication reflects only the author's view, and the European Commission is not responsible for any use that may be made of the information it contains. D . B . was supported b y funds from N F Neuron, Alzheimer N F , and by Career Restart Grant from Masaryk University ( M U N I / R / 1697/2020), and b y the European Regional Development F u n d — P r o j e c t I N B I O ( N o . C Z . 0 2 . 1 . 0 1 / 0 . 0 / 0 . 0 / 1 6 _ 0 2 6 / 0008451). J . R was supported b y funds from Medical Faculty M U to Junior researcher ( R O Z V / 2 3 / L F / 2 0 1 9 , R O Z V / 2 8 / L F / 2020). • REFERENCES (1) Kelava, I.; Lancaster, M . A . Cell Stem Cell 2016, 18, 736-748. (2) Lancaster, M . A ; Renner, M . ; Martin, C. A ; Wenzel, D.; Bicknell, L. S.; Hurles, M . E.; Homfray, T.; Penninger, J. M . ; Jackson, A . P.; Knoblich, J. A Nature 2013, 501, 373-379. (3) Dang, J.; Tiwari, S. K ; Lichinchi, G.; Qin, Y.; Patil, V. S.; Eroshkin, A. M . ; Rana, T. M . Cell Stem Cell 2016, 19, 258-265. (4) Zhang, B.-Z.; Chu, H.; Han, S.; Shuai, H . ; Deng, J.; Hu, Y.; Gong, H.; Lee, A C.-Y; Zou, Z . ; Yau, T.; Wu, W . ; Hung, I. F.-N.; Chan, J. F.W.; Yuen, K . - Y ; Huang, J.-D. Cell Res. 2020, 30, 928-931. (5) Chiaradia, I.; Lancaster, M . A Nat. Neurosci. 2020, 23, 1496- 1508. (6) Dzwonek, J.; Wilczynski, G. M . Front. Cell. Neurosci. 2015, 9, No. 175. (7) Prez, K.; Fan, L. /. Mol. Genet. Med. 2018, 12, No. 366. (8) Geisert, E. E.; Johnson, H . G ; Binder, L . I. Proc. Natl. Acad. Sei. U.SA. 1990, 87, 3967-3971. (9) Zhang, J.; Jiao, J. Molecular Biomarkers for Embryonic and Adult Neural Stem Cell and Neurogenesis. In BioMed Research International; Hindawi Limited, 2015; pp 1 — 14. (10) Bohaciakova, D.; Hruska-Plochan, M . ; Tsunemoto, R ; Gifford, W. D.; Driscoll, S. P.; Glenn, T. D.; Wu, S.; Marsala, S.; Navarro, M . ; Tadokoro, T.; Juhas, S.; Juhasova, J.; Platoshyn, O.; Piper, D.; Sheckler, V.; Ditsworth, D.; Pfaff, S. L.; Marsala, M . Stem Cell Res. Ther. 2019,10, No. 83. (11) Alemi, M . ; Gaiteiro, C ; Rbeiro, C . A ; Santos, L . M . ; Gomes,J. R ; Oliveira, S. M . ; Couraud, P. O.; Weksler, B.; Romero, I.; Saraiva, M . J.; Cardoso, I. Sei. Rep. 2016, 6, No. 20164. (12) Regina Todeschini, A ; Hakomori, S. Biochim. Biophys. Acta, Gen. Subj. 2008, 1780, 421-433. (13) Kolter, T. ISRN Biochem. 2012, 2012, 1-36. (14) Chiricozzi, E.; Lunghi, G ; D i Biase, E.; Fazzari, M . ; Sonnino, S.; Mauri, L. Int. J. Mol. Sei. 2020, 21, No. 868. (15) Kim, H ; Xu, R.; Padmashri, R.; Dunaevsky, A ; Liu, Y.; Dreyfus, C. F.; Jiang, P. Stem Cell Rep. 2019, 12, 890-905. (16) Hale, J. E. Int. J. Proteomics 2013, 2013, 1-6. (17) Li, L.; Han, J.; Wang, Z.; Liu, J.; Wei, J.; Xiong, S.; Zhao, Z. Int. J. Mol. Sei. 2014, 15, 10492-10507. (18) Allende, M . L . ; Cook, E. K ; Larman, B. C ; Nugent, A ; Brady, J. M . ; Golebiowski, D.; Sena-Esteves, M . ; Tifft, C. J.; Proia, R L. /. Lipid Res. 2018, 59, 550-563. (19) Latour, Y. L.; Yoon, R ; Thomas, S. E.; Grant, C; L i , C ; SenaEsteves, M . ; Allende, M . L . ; Proia, R L ; Tifft, C. J. Mol. Genet. Metab. Rep. 2019, 21, No. 100513. (20) Boutry, M . ; Branchu, J.; Lustremant, C ; Pujol, C ; Pernelle, J.; Matusiak, R ; Seyer, A ; Poirel, M . ; Chu-Van, E.; Pierga, A ; Dobrenis, K ; Puech, J. P.; Caillaud, C ; Durr, A ; Brice, A ; Colsch, B.; Mochel, F.; El Hachimi, K H . ; Stevanin, G ; Darios, F. Cell Rep. 2018, 23, 3813- 3826. (21) Shi, T.; Song, E.; Nie, S.; Rodland, K D . ; Liu, T.; Qian, W.-J.; Smith, R D . Proteomics 2016, 16, 2160-2182. (22) Hájek, R ; Jirásko, R ; Lisa, M . ; Cífková, E.; Holčapek, M . Anal. Chem. 2017, 89, 12425-12432. (23) Vidová, V.; Stuchlíkova, E.; Vrbová, M . ; Almasi, M . ; Klanová, J.; Thon, V ; Spacil, Z. /. Proteome Res. 2019, 18, 380-391. (24) Pasca, A M . ; Sloan, S. A ; Clarke, L. E.; Tian, Y ; Makinson, C D . ; Huber, N ; Kim, C. H . ; Park, J. Y ; O'Rourke, N . A ; Nguyen, K. D . ; Smith, S. J.; Huguenard, J. R ; Geschwind, D . H ; Barres, B. A ; Pasca, S. P. Nat. Methods 2015, 12, 671-678. (25) Nascimento, J. M . ; Saia-Cereda, V. M . ; Sartore, R. C ; da Costa, R M . ; Schitine, C. S.; Freitas, H . R ; Murgu, M . ; de Melo Reis, R. A ; Rehen, S. K ; Martins-de-Souza, D . Front. Cell Dev. Biol. 2019, 7, No. 303. (26) Luo, C ; Lancaster, M . A ; Castanon, R ; Nery, J. R ; Knoblich, J. A ; Ecker, J. R Cell Rep. 2016, 17, 3369-3384. (27) Fair, S. R ; Julian, D.; Hartlaub, A . M . ; Pusuluri, S. T.; Malik, G ; Summerfied, T. L.; Zhao, G ; Hester, A B.; Ackerman, W . E.; Hollingsworth, E. W.; Ali, M . ; McElroy, C. A ; Buhimschi, I. A ; Imitola, J.; Maitre, N . L.; Bedrosian, T. A ; Hester, M . E. Stem Cell Rep. 2020,15, 855-868. (28) Sipione, S.; Monyror, J.; Galleguillos, D.; Steinberg, N ; Kadam, V. Front. Neurosci. 2020, 14, No. 1004. (29) Nakatani, Y ; Yanagisawa, M.j Suzuki, Y.j Yu, R. K Glycobiology 2010,20,78-86. (30) Bond, A . M . ; Ming, G ; Song, H . Cell Stem Cell 2015, 17, 3 8 5 - 395. (31) Ryu, J.-S.; Ko, K ; Ko, K ; Kim, J.-S.; Kim, S.-U.; Chang, K - T . ; Choo, Y - K Mol. Med. Rep. 2017, 16, 987-993. (32) Schengrund, C.-L. Trends Biochem. Sei. 2015, 40, 397-406. (33) Marconi, S.; De Toni, L.; Lovato, L.; Tedeschi, E.; Gaetti, L.; Acler, M . ; Bonetti, B. /. Neuroimmunol. 2005, 170, 115-121. (34) Saito, M . ; Wu, G ; Hui, M . ; Masiello, K ; Dobrenis, K ; Ledeen, R. W.; Saito, M . /. Lipid Res. 2015, 56, 1434-1448. (35) Zhang, Y ; Lukacova, V ; Reindl, K ; Balaz, S. /. Biochem. Biophys. Methods 2006, 67, 107-122. (36) Johnson, J.; Sharick, J. T.; Skala, M . C ; L i , L. /. Mass Spectrom. 2020, 55, No. e4452. 3167 https://doi.org/10.1021/acs.analchem.2c00981 Anal. Chem. 2023, 95, 3160-3167