Cell type–specific labeling of newly synthesized proteins by puromycin inactivation

Abstract

Puromycin and its derivative O-propargyl puromycin (OPP) have recently found widespread use in detecting nascent proteins. Use of these metabolic labels in complex mixtures of cells leads to indiscriminate tagging of nascent proteomes independent of cell type. Here, we show how a widely used mammalian selection marker, puromycin N-acetyltransferase, can be repurposed for cell-specific metabolic labeling. This approach, which we named puromycin inactivation for cell-selective proteome labeling (PICSL), is based on efficient inactivation of puromycin or OPP in cells expressing puromycin N-acetyltransferase and detection of translation in other cell types. Using cocultures of neurons and glial cells from the rat brain cortex, we show the application of PICSL for puromycin immunostaining, Western blot, and mass spectrometric identification of nascent proteins. By combining PICSL and OPP-mediated proteomics, cell type–enriched proteins can be identified based on reduced OPP labeling in the cell type of interest.

Tissues and other experimental preparations consisting of different cell types present a challenge for molecular analysis when assigning detected biomolecules to correct cellular entities. The overall translational output of a tissue or complex cell mix to a particular stimulus is shaped by the distinct response of the cells it comprises, which may vary considerably among cell types. Moreover, while alterations in the global proteome provide clues into different biological processes, profiling changes in the nascent proteome is critical for understanding the rapid changes following a particular stimulus. Tracking these changes in a cell type–specific manner has proven particularly difficult for minor or low-abundance cell types. Cell type–specific transcriptome analysis is now commonplace and researchers can choose between two strategies: either affinity purification of ribosome-associated RNA using cell type–specific expression of tagged ribosomes (1, 2) or single-cell RNA-seq of enzymatically dissociated tissues (3). In proteomics, mass spectrometric (MS) analysis of single cells has recently been demonstrated (4), but the technology is still in early development and not widely accessible. Similarly to RNA analysis, affinity purification of proteins labeled in a cell-specific manner offers another possibility. Today, the most widely used approach for this is based on cell-specific expression of engineered aminoacyl-tRNA synthetases, which label newly synthesized proteins with noncanonical amino acids (5). The latter can be conjugated with biotin by click chemistry for affinity purification. While this has been demonstrated in several model organisms (6, 7, 8), its use is limited by the requirement of extensive labeling times and depletion of endogenous amino acids that compete with the noncanonical ones. As an alternative to labeling with noncanonical amino acids, puromycin (PMY) has emerged as an efficient label for newly synthesized proteins (9). In addition, a click-reactive derivative, O-propargyl puromycin (OPP), enables similar measurements of global protein synthesis (10) as well as MS analysis of newly synthesized proteins by biotin conjugation and affinity purification (11). Cell-specific puromycinylation has been achieved using “blocked” (or “caged”) PMY derivatives that become active when uncaged by an enzyme orthogonally expressed in targeted cells (12) in an antibody-assisted manner (13) or by activating light (14). Here, we present an alternative strategy for cell type–specific labeling of de novo–synthesized proteins based on enzymatic, cell-selective inactivation of PMY or OPP labels mediated by puromycin N-acetyltransferase (PAT).

Results

PAT efficiently inactivates PMY and OPP

PAT (also known as PAC) (15) is frequently used as a selection marker in cell culture experiments, allowing cells expressing this enzyme to survive for extended growth periods in media containing PMY (16). We argued that effective inactivation of PMY by PAT, working to clear up free PMY or OPP in the cell (Fig. 1A), can be repurposed for cell-specific labeling of newly synthesized proteins, facilitating the quantification of overall translation as well as the isolation and identification of nascent proteins in a cell-specific manner (Fig. 1B). To test this, HEK 293 cells transfected with mCherry-tagged PAT were mixed in a 1:1 ratio with nontransfected cells, labeled with 5 μM PMY for 10 min, and subsequently stained with anti-PMY antibody. An absence of PMY incorporation in PAT-expressing cells can be observed, whereas nontransfected cells are strongly labeled (Fig. 2, A and B). A similar reduction of PMY signal can be observed in either unlabeled cells or cells labeled in the presence of cycloheximide or anisomycin (Fig. S1). Next, we tested if OPP—a non-natural PAT substrate—can be similarly inactivated by PAT. For this, HEK 293 cells stably expressing PAT (“HEK-PAT”) and regular HEK 293 cells (“HEK 293”) were pulse-labeled with increasing amounts of PMY or OPP for 15, 60, or 120 min, and labeled, newly synthesized proteins were detected by Western blot. We determined that PAT can very efficiently block the labeling of de novo–synthesized proteins with 2 h pulses of up to 20 μM OPP (Fig. 2, C and D and S2, A and C) or 5 μM PMY (Fig. S2, B and D). Higher concentrations of OPP or PMY are incompletely inactivated at longer incubation times (Fig. S2). Overall protein synthesis was not inhibited by labeling with up to 5 μM PMY or 20 μM OPP for 2 h, as measured by ³⁵S Met/Cys incorporation (Fig. S3). Note that higher concentrations of PMY also cause a shift toward lower sizes in labeled proteins (Fig. S2B). To confirm that inactivation of OPP by PAT would effectively prevent the isolation of newly synthesized proteins, HEK 293 and HEK-PAT cells were pulse-labeled and newly synthesized proteins tagged with biotin by click reaction and isolated through streptavidin pull-down. We confirmed that PAT expression prevents pull-down of biotinylated de novo–synthesized proteins, as revealed by Western blot with streptavidin-horseradish peroxidase (HRP), as well as two tested proteins (GAPDH and β-tubulin), efficiently pulled down in OPP-labeled regular HEK 293 cells (Fig. 2E). In conclusion, by introducing PAT in a mix of cells, de novo protein synthesis can be highlighted in PAT-negative (PAT-) cells by subtracting (erasing) PMY signal from PAT-expressing ones (PAT+). We named this approach puromycin inactivation for cell-selective proteome labelling—PICSL (/ˈpixel/).

Figure 1.

Principle of PICSL—puromycin inactivation for cell-selective proteome labelling.A, puromycin N-acetyltransferase (PAT, also called PAC) catalyzes acetylation of the reactive amino group (red) in puromycin (PMY) or its click-handled O-propargyl derivative (OPP). By blocking PMY/OPP incorporation in PAT-expressing cells, this property can be used for cell-specific analysis of de novo protein synthesis. B, uses of PICSL. Pulse-labeling of cells with PMY allows for the quantification of protein synthesis by immunocytochemistry or Western blot analysis. OPP labeling in combination with biotin azide conjugation enables affinity purification of newly synthesized proteins and analysis by mass spectrometry or Western blot. Cells not expressing PAT (white) are metabolically labeled (orange), whereas labeling is prevented (gray) in cells expressing PAT (blue).

Figure 2.

PAT effectively quenches incorporation of PMY and OPP into newly synthesized proteins. A, immunocytochemistry of HEK 293 cells transfected with mCherry-PAT (magenta) mixed with nontransfected cells (no magenta signal), after 10 min labeling with 5 μM PMY. Anti-PMY is shown in white and cell territories delineated by phalloidin staining are shown in teal. The scale bar represents 20 μm. B, quantification of (A). Signal intensities from the anti-mCherry and anti-PMY channels were quantified under a phalloidin mask for transfected (magenta dots) and untransfected cells (gray dots) and plotted against each other. Ten different images were used for quantification. C, Western blot analysis of puromycinylated proteins (anti-PMY) in regular HEK 293 or HEK 293 cells stably expressing PAT (HEK-PAT) labeled for the indicated time periods with 20 μM OPP. The corresponding Coomassie-stained membrane is presented on the right panel. D, quantification of (C). Graph depicts mean ± SD; n = 3 independent labeling experiments. E, streptavidin pull-down of newly synthesized proteins from HEK 293 and HEK-PAT cells. (Western blot analysis with indicated antibodies and HRP-conjugated streptavidin, Strep-HRP). HEK 293 and HEK-PAT cells were labeled for 2 h with 20 μM OPP (+) or DMSO as a control and biotinylated by click-reaction. Biotinylated proteins were isolated by streptavidin pull-down and analyzed by Western blot alongside input samples. DMSO, dimethyl sulfoxide; OPP, O-propargyl puromycin; PAT, puromycin N-acetyltransferase; PMY, puromycin.

PICSL proteomics

Next, we aimed to test the applicability of PICSL for cell-specific profiling of the newly synthesized proteome. For this, we used a test system whereby two cell lines from different species—mouse Neuro2a (N2a) and human HEK 293 cells—were mixed and subsequent unambiguous mapping of origin for a large number of peptides was enabled by MS analysis. Mouse–human mixed samples have previously been successfully resolved in MS analysis, including a similar experimental design mixing 3T3 (mouse) and HeLa (human) cells (17). This setup enabled us to probe the proteome-wide effect of PAT-mediated suppression of labeling when coculturing PAT+ and PAT- cells. Here, N2a cells were mixed with either regular HEK 293 cells or HEK-PAT cells in a 1:1 ratio. Cell mixes were then pulse-labeled with 20 μM OPP or dimethyl sulfoxide (DMSO) for 2 h, followed by lysis and click-conjugation with biotin azide. Newly synthesized proteins were then purified by streptavidin pull-down and analyzed by MS as described before (11, 18) (Fig. 3A). In these pull-down samples, 4031 proteins were identified. Taxonomy mapping of individual peptides facilitated the analysis of peptide species uniquely mapping to mouse or human proteins (Fig. 3B and Table S1). Spectral counts (Fig. 3B) and MS intensity distributions showed an enrichment in OPP-labeled samples over DMSO-treated background (Fig. 3, C–E). Intensity distributions for peptides common to mouse and human proteins showed a similar intensity profile, with a shift toward lower intensities in the HEK-PAT+N2a sample, likely reflecting the effect of PAT inactivating OPP in HEK-PAT cells (Fig. 3C). Analysis of species-specific peptides displayed a major reduction in human-specific peptide intensities in HEK-PAT+N2a compared to HEK293+N2a (Fig. 3E) showing effective suppression of de novo–synthesized proteins in human cells. In contrast, mouse proteins from N2a cells were efficiently labeled in the HEK-PAT+N2a sample (Fig. 3D). These results show that PICSL is suitable for proteome-wide identification of cell type–specific proteins.

Figure 3.

Mixed-species mass spectrometry experiment shows proteome-wide efficiency of PICSL. A, schematic of the experiment. Neuro2a cells (N2a, mouse) were mixed at a 1:1 ratio with either regular HEK 293 cells (human) or HEK 293 cells stably expressing PAT (HEK-PAT). After cell attachment (4 h), cells were labeled for 2 h with 20 μM OPP or DMSO, lysed, and clicked with biotin azide. Newly synthesized proteins were pulled down using streptavidin beads and analyzed by LC-MS/MS. PAT-expressing cells are shown in blue and OPP-incorporating cells in orange. B, peptides from de novo–synthesized proteome pull-downs (four samples as indicated) were mapped to UniProt taxonomy identifiers. Shown are spectral counts for all peptides, peptides mapping to both mouse and human proteins, and peptides unique to mouse or human proteins. Bovine serum proteins and streptavidin account for the difference in counts in “All” and combined counts of mouse and human peptides. C–E, distribution of intensity values for individual peptides common to mouse and human (C), unique to mouse (D), or human (E). Number of peptides in log(2) intensity value bins (dots, bin width 0.2) are shown with fitted curves (lines). The difference in mouse-specific peptides identified in the HEK-PAT+N2a/OPP compared to HEK293 + N2a/OPP may be caused by more efficient pull-down of mouse proteins from the click-conjugated HEK-PAT+N2a/OPP lysate due to better availability of streptavidin-binding sites. DMSO, dimethyl sulfoxide; OPP, O-propargyl puromycin; PAT, puromycin N-acetyltransferase; PICSL, puromycin inactivation for cell-selective proteome labeling.

Use of PICSL in mixed cultures of brain cells

To demonstrate the use of PICSL in a coculture system of primary cells, we chose a mixed rat brain cell preparation obtained from prenatal cerebral cortices. This culture consists of neurons, astroglia, oligodendrocyte precursor cells, and a small number of microglia (Fig. S4A). We transduced the mixed cortical culture with adeno-associated virus (AAV) carrying the mCherry-tagged PAT enzyme under the control of the Synapsin 1 promoter (“[Syn]-PAT”) to drive the expression of PAT exclusively in neurons (19). At 7 days in vitro, these cultures were pulse-labeled with PMY, and cell-specific PMY incorporation was analyzed by immunocytochemistry. The specificity of PAT expression was confirmed by coimmunodetection of mCherry and NeuN, a nuclear neuronal marker (20), and the absence of mCherry signal in astrocytes, detected by Aldh1L1 staining (21) (Figs. 4A and S5). PMY labeling was restricted entirely to nonneuronal cells as detected by a completely nonoverlapping immunofluorescence signal of PMY and mCherry (Figs. 4A and S5). Therefore, in the shown example, protein synthesis can be detected only in glial cells. Both neurons and glial cells develop a dense meshwork of processes in these cultures. PICSL allows for the detection of newly synthesized proteins in these processes in a cell type–specific manner.

Figure 4.

Neuronal-targeted expression of PAT in mixed rat cortical cell cultures. A, mixed cortical neuron-glia cultures from embryonic day 21 (E21) rat pups were transduced at seeding with AAV-[Syn]-PAT-mCherry to direct expression of PAT to neurons. At 7 days in vitro, cells were pulse-labeled for 10 min with 5 μM PMY and costained with anti-mCherry (PAT, in magenta), anti-NeuN (neuronal marker, in green), anti-Aldh1L1 (astrocyte marker, in green), or anti-PMY (in white) in the indicated combinations. Bottom right image shows an enlargement of the area indicated above. Full images with separate channels are shown in Fig. S5. The scale bar represents 50 μm. B, untransduced or AAV-[Syn]-PAT–transduced cocultures were labeled for 2 h with 20 μM OPP (“OPP”, “OPP + [Syn]-PAT”) or DMSO as a control (“DMSO”), and newly synthesized proteins were biotinylated by click-reaction and isolated by streptavidin pull-down. Western blot analysis of three neuronal markers (NeuN, calbindin 2, and β-III-tubulin) and overall biotinylation (detected by streptavidin-HRP) showed signal reduction in cultures transduced with AAV-[Syn]-PAT-mCherry. C and D, quantification of (B). C. signal intensities of the three tested neuronal markers were determined in the pull-down fractions of OPP-labeled untransduced (blue) or AAV-[Syn]-PAT–transduced (red) samples. Bars indicate mean ± SD of 4 to 5 biological replicates. Statistical differences were determined by multiple paired t-tests and the Holm-Šídák correction for multiple comparisons. D, streptavidin-HRP signal intensity (representing biotinylated proteins) was quantified for pull-down and input fractions in OPP-labeled untransduced (blue) or OPP-labeled AAV-[Syn]-PAT–transduced (red) samples. Strep-HRP signal intensity was not quantified for DMSO-treated samples since this background signal arises from endogenous biotinylated proteins. Bars show mean ± SD of 3 to 4 biological replicates. One-way ANOVA and the Holm-Šídák’s multiple comparisons test was used to determine statistically significant differences. ∗adjusted p-value ≤ 0.05, ∗∗adjusted p-value ≤ 0.01. AAV, adeno-associated virus; DMSO, dimethyl sulfoxide; HRP, horseradish peroxidase; OPP, O-propargyl puromycin; PAT, puromycin N-acetyltransferase; PMY, puromycin; Syn, synapsin.

Next, we examined the efficiency of PAT in quenching OPP incorporation in mixed cortical cell cultures. We labeled AAV-[Syn]-PAT–transduced cocultures with OPP, clicked with biotin azide, and isolated de novo–synthesized proteins by streptavidin pull-down. Western blot analysis showed a significant PAT-dependent reduction in the amount of biotinylated newly synthesized proteins in total protein lysates and pull-down samples (Fig. 4B, bottom panel, Fig. 4D). In addition, three tested neuronal marker proteins (NeuN, calbindin 2, and β-III-tubulin) were detected at reduced levels in pull-down fractions of cultures transduced with AAV-[Syn]-PAT-mCherry, indicating that the neuron-specific expression of PAT effectively reduced puromycinylation and subsequent pull-down of OPP-labeled nascent proteins (Fig. 4, B and C).

Next, we tested the application of PICSL for cell-specific proteomic analysis of newly synthesized proteins in neuron-glia cocultures by OPP-ID (11, 18). To improve the detection of nonneuronal proteins, we prepared cocultures with higher glial content. Cultures prepared as above (Fig. S4A) were seeded on a preplated monolayer of astroglia, which supports the expansion of different nonneuronal cells such as microglia and oligodendrocytes (Fig. S4B). Next, cultures were transduced with AAV-[Syn]-PAT-mCherry or left untransduced, pulse-labeled for 2 h with 20 μM OPP (or DMSO in untransduced cells), newly synthesized proteins tagged with biotin through click reaction, and isolated by streptavidin pull-down. MS analysis was performed on samples from two independent biological replicates, and 3990 protein entries were identified (see Fig. S6 for the correlation of signal intensities between the two repeats). The intensity distribution of the identified proteins showed an enrichment in the OPP-labeled samples compared to the DMSO-treated control (Fig. 5A), similar to that observed in the N2a/HEK experiment (see Fig. 3 for reference). Additionally, PAT expression in neurons caused a reduction in enrichment over background (DMSO-labeled sample) (Fig. 5A). A clear overlap was observed between the proteins detected in OPP-labeled, PAT-expressing samples and DMSO-treated samples, in agreement with nonspecifically binding proteins (DMSO) and PAT-resistant specifically labeled proteins (OPP + PAT) being a subset of all proteins isolated by pull-down (OPP) (Fig. 5B). For subsequent analyses, detected proteins were filtered for entities highly enriched in pull-down (no signal intensity in the DMSO sample; 2035 proteins). We further focused on proteins that showed no signal intensity in PAT-expressing samples, that is, proteins whose labeling was prevented by neuronal PAT (see the “Proteins in OPP only” list in Table S2). Gene ontology enrichment analysis performed using g:Profiler (22) showed a significant enrichment in terms associated with several neuronal structures and processes (Fig. 5C and Table S2), suggesting an overrepresentation of neuronal proteins in this subset. As an example, proteins associated with the GO term “synapse” are shown in the fly out of Figure 5C. This indicates that by implementing PICSL and comparing PAT+ and PAT- cells, it is possible to identify cell-specific (in this case neuron-specific) proteins.

Figure 5.

Application of PICSL for cortical coculture proteomics. A, LC-MS/MS analysis of de novo–synthesized proteins isolated from cortical cocultures with high glial content (see main text for details) with or without AAV-[Syn]-PAT transduction upon seeding. After 7 days in vitro, cells were pulse-labeled for 2 h with 20 μM OPP (blue and red for untransduced and AAV-[Syn]-PAT transduced, respectively) or DMSO (black) and processed for streptavidin-mediated pull-down. Shown are log2 intensity distributions of identified proteins. Bars indicate mean ± SD of two biological replicates. The distribution for DMSO-treated samples indicates the background binding of unlabeled proteins to beads in affinity purification. B. Venn diagram showing the overlap of proteins identified in the DMSO-treated samples (gray) and OPP-treated samples (blue and red for untransduced and AAV-[Syn]-PAT transduced, respectively). Only proteins with an intensity > 0 for each condition in at least one replicate were included. The diagram was generated using the BioVenn web application tool (46). C, top 10 cellular compartment gene ontology (GO) enriched terms for proteins detected only in the untransduced, OPP-labeled samples. Genes in the GO term “synapse” are shown in the fly out. The list of proteins identified in the MS analysis was refined by including only those that showed no intensity in the DMSO-treated or the OPP-labeled PAT-expressing samples but did so in the untransduced, OPP-labeled samples (see list of “Proteins in OPP only” in Table S2). Enriched GO terms for this subset of proteins were determined using g:Profiler (22). Shown here are GO cellular compartment terms and their corresponding adjusted p-values (in –log₁₀ scale). D, list of the top 10 proteins (ranked by signal intensity) detected only in PAT negative, OPP-labeled samples (i.e., neuron-enriched proteins). Highlighted in blue are those that were selected for further validation. E, schematic of the RiboTag assay for validation of neuron-enriched candidate proteins. Neuron-glia cocultures transduced with AAV-[Syn]-Rpl22-HA express HA-tagged ribosomes only in neurons, allowing neuron-specific immunopurification of ribosome-associated mRNA. F, RiboTag analysis of the top five neuron-specific proteins from the OPP mass spectrometry experiment (shown in D). Cortical cocultures were transduced with AAV-[Syn]-Rpl22-HA on the day of seeding. At 7 days in vitro, neuronal ribosome-associated mRNAs were immunoprecipitated with anti-HA antibody (RiboTag +), using untransduced cultures (RiboTag -) as negative controls. For reference, glia-enriched marker genes Aldh1L1 (astrocyte), Iba1 (microglia), Cspg4 (oligodendrocyte precursor cells), Pdgfra (oligodendrocyte precursor cells), and Cnp (oligodendrocytes) were analyzed. Bars depict mean ± SD of n = 3 independent biological replicates. Statistical significance between RiboTag + and RiboTag - groups was determined by multiple unpaired t tests and the Holm-Šídák correction for multiple comparisons. ∗∗∗∗adjusted p-value ≤ 0.001, ns: not significant. AAV, adeno-associated virus; DMSO, dimethyl sulfoxide; MS, mass spectrometry; OPP, O-propargyl puromycin; PAT, puromycin N-acetyltransferase; PICSL, puromycin inactivation for cell-selective proteome labeling; Syn, synapsin.

Next, we independently validated the top five neuron-enriched proteins from the MS analysis (“Proteins in OPP only” in Table S2 and Fig. 5D). For this, we implemented the RiboTag tool (1, 23), which enables cell-specific isolation of ribosome-associated mRNAs, based on immunoprecipitation of tagged ribosomes (Fig. 5E). Cortical cocultures were transduced upon seeding with AAV-[Syn]-Rpl22-HA, directing the expression of tagged ribosomal proteins to neurons. At 7 days in vitro cells were lysed, ribosomes immunoprecipitated using an anti-HA antibody, and ribosome-associated mRNA was extracted. A total RNA sample was collected in parallel, and both samples were then analyzed by quantitative real-time PCR (RT-qPCR). RiboTag analysis showed significant enrichment of all five candidates within neuronal mRNAs, supporting the specific expression of these proteins in neurons (Fig. 5F). Of note, analyzed glial markers (Aldh1L1, Iba1, Cspg4, Pdgfra, and Cnp) were not enriched in immunoprecipitated fractions, despite being expressed in the cultures (Fig. S4), corroborating the specificity of the validation approach.

Taken together, the examples presented above show two ways of implementing PICSL in mixed-cell scenarios. First, PICSL enables cell-specific visualization of newly synthesized proteins, which is useful for the analysis of global protein synthesis activity. Second, in using PICSL for pull-down of OPP-labeled proteins, one can identify proteins enriched in the particular cell type expressing PAT—these will be detected by a reduction in OPP-mediated pull-down compared to nontargeted cell types.

Discussion

Accessible methods for cell-specific analysis of nascent proteomes are increasingly needed. The most commonly used strategies rely on the use of noncanonical amino acids that are introduced into newly synthesized proteins by engineered aminoacyl-tRNA synthetases specifically targeted to the cell type of interest (5). Different variants of this strategy (such as BONCAT or SORT) have been successfully applied to model organisms like Caenorhabditis elegans, Drosophila, and mice ((24) and references therein). A Cre-inducible mouse line expressing a mutant methionyl-tRNA synthetase is currently available, opening up the possibility to specifically label newly synthesized proteins in a variety of cell types, limited by the Cre driver lines available (25). However, labeling times in the range of days (or weeks) have limited its application for tracking short-term changes in the proteome (24, 26). More recently, owing to excellent temporal resolution and absence of bias toward amino acid composition of target proteins, PMY and its derivatives have found extended use for the efficient labeling of nascent peptides. PMY has been previously employed to tag newly synthesized proteins in a cell-specific manner using “blocked” or “caged” PMY derivatives that become active when uncaged in an antibody-assisted manner (13), through photoactivation (14) or by an enzyme orthogonally expressed in targeted cells (12). Of these solutions, only the latter is likely suitable for in vivo application, and it remains to be demonstrated how well the caged substrate used by Barrett et al. is biodistributed upon systemic delivery. Here, we leverage the good bioavailability of PMY (see (27) and references in (28)) to develop PICSL, a labeling method based on cell-specific substrate inactivation rather than cell-specific release. This method allows highly efficient tagging of newly synthesized proteins by unmodified PMY or its click-handled derivative OPP. Importantly, intraperitoneal administration of OPP for 1 h prior to tissue collection has been shown to be effective for labeling of de novo–synthesized proteins in bone marrow cells (29, 30), small intestine (10), skeletal muscle (31), skin (32), and fetal liver in pregnant mice (33), indicating good bioavailability. Here, we show that OPP is also efficiently used as a PAT substrate and suitable for PICSL proteomics. In comparison with the noncanonical amino acid approach, PICSL does not need extensive labeling times or prior methionine depletion. Another benefit of PICSL is that the key component of the system, that is, the PAT enzyme, is readily available within the PMY resistance cassette found in a wide variety of vectors. Given its extensive use as a selection marker, it has been well-established that PAT expression has no deleterious effects on expressing cells (28), eliminating the need to establish its safety for in vitro and in vivo use. Finally, the crystal structure of PAT has been recently obtained, and it served as the basis for the generation and characterization of several mutants with either increased or decreased activity (34), opening the possibility of using different PAT variants that better suit the translational activity of the system of interest.

By implementing PICSL, instead of attempting to restrict labeling to a cell type, labeling is specifically removed from the targeted PAT-expressing cells. This allows for the following: (1) to highlight the proteome of other, potentially lower abundance, cells present in the mix and (2) by comparing the PAT-quenched versus nonquenched proteome, to assign the origin of a protein or subset of proteins of interest to the PAT-expressing cell. For a successful PICSL experiment, it is important to ensure even expression levels in the target cell type using viral vectors or genetic drivers in mice. Another limitation of the PICSL strategy is that, in principle, it permits only the quenching of the proteome of one cell population at a time. This could be addressed by combined use of PAT under different cell type–specific promoters. On the other hand, inactivation of puromycinylation in a single cell type can be used as an advantage when addressed as a “mirror image”, a negative of the commonly used enrichment approaches. If a protein shows large differences in signal intensity between PAT+ and PAT- conditions, that protein can be identified as enriched in the PAT-expressing cell type. We could verify this in our study system, focusing on neurons, by detecting several well-known neuronal proteins that are depleted from the pool of affinity-purified newly synthesized proteins in PAT-expressing cells (Fig. 5, C and D for examples). By applying the same principle, previously unrecognized cell type–enriched proteins could be identified from high PAT-/PAT+ pull-down ratios. Such is the case of Tmem130, a poorly studied protein with a potential link to neurological disorders (35), which is highly enriched in neurons in our mixed-culture system.

Our results indicate that labeling by PMY/OPP is suppressed over a wide range of incubation times and concentrations in PAT-expressing cells, without interfering with labeling of neighboring cells. Furthermore, using a neuron-glia coculture as a test system, we confirmed that targeting PAT to neurons effectively prevents puromycinylation of nascent proteins in these cells, while allowing labeling in glial cells present in the culture. Similarly, PICSL could be applied in a variety of conditions, such as profiling normal and diseased tissue in mouse models and in organoid research (36, 37). Another interesting possibility is the introduction of PAT into established human cancer cell lines used for xenograft models. Tumor-specific expression of PAT would prevent labeling of tumor-derived newly synthesized proteins and highlight the changes in neighboring host tissue.

In conclusion, PICSL offers a technically simple alternative to other methods for cell type–specific labeling of newly synthesized proteins. The key advantage of PICSL over existing methods is using unmodified PMY or OPP as a substrate, which does not require precursor uncaging or competition with endogenous amino acids (as in noncanonical amino acid-based methods).

Experimental procedures

Chemicals and plasmids

Puromycin hydrochloride (Cat. #13884), cycloheximide (Cat #14126), and anisomycin (Cat #11308) were from Cayman Chemical, Tris(3-hydroxypropyltriazolylmethyl)amine (THPTA, Cat. #762342), Tris(2-carboxyethyl)phosphine hydrochloride (TCEP, Cat. #C4706), and CuSO₄ from Sigma-Aldrich, and Pierce streptavidin magnetic beads (Cat. #88817) from Thermo Fisher Scientific. Mammalian expression vector expressing PAT-mCherry fusion protein under the cytomegalovirus promoter (pcDNA-[CMV]-PAT-3x-mCherry) was prepared by PCR-amplifying (Phusion Hot Start II DNA Polymerase, Thermo Fisher Scientific) the PAT ORF from Addgene plasmid #48141 (sequence labeled as the “Puro” selection marker) and subcloning it into N-terminal fusion with a triple-mCherry tag in Addgene plasmid #64108, then subcloned the PAT-3x-mCherry fragment into a pcDNA mammalian expression vector under the cytomegalovirus promoter. Neuron-specific AAV expression construct AAV-[Syn]-PAT-mCherry was prepared by NEBuilder HiFi DNA Assembly (New England Biolabs) by recombining an AAV genomic vector backbone with the human synapsin promoter (Syn) and PAT-mCherry fragment amplified by PCR using Phusion Hot Start II DNA Polymerase (Thermo Fisher Scientific) from the pcDNA-PAT-3xCherry construct. For generating a neuron-specific RiboTag AAV vector (AAV-[Syn]-RiboTag), the [Syn] promoter was subcloned from AAV-[Syn]-PAT-mCherry into Addgene plasmid #111811 to replace the [GfaABC1D] promoter. Plasmid constructs were verified by Sanger sequencing.

Cell culture

All cells were cultured in a humidified incubator at 37 °C with a 5% CO₂ atmosphere. Primary cultures of rat brain cortical cells were prepared as previously described for rat cortical neurons (38), with minor modifications. Relevant procedures were approved by the ethics committee of animal experiments at the Ministry of Agriculture of Estonia (permit number: 45). Briefly, cerebral cortices of embryonic day 21 (E21) Sprague Dawley rat pups were digested with 0.25% Trypsin-EDTA (Invitrogen) for 20 min at 37 °C, with addition of 0.5 mg/ml DNAse I (Roche Diagnostics) and MgSO₄ after the first 10 min. Digestion was terminated by the addition of 0.25% trypsin inhibitor (Invitrogen) and 1% bovine serum albumin (BSA, Pan-Biotech), and tissue was mechanically disrupted with 1 ml pipette tips. Tissue suspension was then diluted in Hanks' balanced salt solution (140 mM NaCl, 5 mM KCl, 1 mM CaCl₂, 0.4 mM MgSO₄·7H₂O, 0.5 mM MgCl₂·6H₂O, 0.3 mM Na₂HPO₄·2H₂O, 0.4 mM KH₂PO₄, 6 mM D-glucose, and 4 mM NaHCO₃), large debris removed by 30 s centrifugation at 200×g, and cells collected by 5 min centrifugation at 200×g. Cells were then seeded on poly-L-lysine–coated dishes and maintained in Neurobasal A medium (Invitrogen), supplemented with B27 (Invitrogen), 1 mM L-glutamine (Invitrogen), and 100 μg/ml Primocin (InvivoGen), with half-medium changes every 48 h. Primary cortical astrocytes were isolated from E21 Sprague Dawley rat pups, using a modified McCarthy and De Vellis (39) method, as described in (40). Cultures were transduced with AAV at seeding (final dilution 2–5 × 10⁸ viral genomes/ml) and treated with 5 μM PMY or 20 μM OPP at 7 days in vitro.

Neuro2A, HEK 293, and HEK-PAT (a kind gift from Dr Illar Pata) cell lines were cultured in Dulbecco's modified Eagle medium (DMEM, Gibco) supplemented with 10% fetal bovine serum (Pan Biotech), 100 U/ml penicillin, and 100 μg/ml streptomycin (Invitrogen). Cells were transfected at ∼80% confluence, using PEI (Sigma-Aldrich) at 2:1 w/v ratio and 1 μg of DNA per 6-well dish. PMY or OPP labeling was performed 24 h after seeding or transfection.

AAV production and purification

Chimeric AAV1/AAV2 virions were produced using equimolar ratios of AAV1 and AAV2 capsid plasmids as described previously (41). HEK 293T cells were grown in DMEM supplemented with 10% fetal bovine serum, 100 U/ml penicillin, and 100 mg/ml streptomycin and transfected at 50 to 60% confluence with PEI (2:1 weight ratio of PEI and DNA). Four 10 cm plates were transfected using PEI with 10 μg of DNA (AAV-[Syn]-PAT-mCherry/AAV-[Syn]-RiboTag plasmid containing the expression cassette, pAdDeltaF6 helper plasmid, AAV1 and AAV2 helper plasmids at 1:1:0.5:0.5 M ratios) in growth medium, supplemented with 1 mM sodium pyruvate. Three days after transfection, virus particles were harvested using AAVpro Purification Midi Kit for All Serotypes from Takara Bio (#6675). Virus titers (as determined by RT-qPCR) were in the range of 10¹¹-10¹² viral genomes/ml.

Immunofluorescence

Following treatment/transfection, cells grown on coverslips were fixed with 4% paraformaldehyde-PBS (137 mM NaCl, 2.7 mM KCl, 8 mM Na₂HPO₄, and 2 mM KH₂PO₄, pH 7.4) for 15 min, treated with 50 mM NH₄Cl for 10 min, and permeabilized with 0.5% Triton X-100 in PBS for 15 min and blocked with 2% BSA–PBS for 1 h. Incubation with primary antibodies was performed overnight at 4 °C, in 0.2% BSA-0.1% Tween 20-PBS. The following antibodies were used: anti-mCherry (Novus Biologicals NBP2-25157, 1:100), anti-NeuN (Millipore MAB377, 1:100), anti-Aldh1L1 (DSHB N103/31, 1:100), anti-puromycin (Millipore MABE343, 1:2000), anti-Iba1 (WAKO 019-19741, 1:300), anti-CNPase (Cell Signaling Technology D83E10, 1:100), anti-MAP2 (Millipore AB5622, 1:2000), and anti-PDGFRa (Novus Biologicals AF1062, 1:500). After washing, incubation with fluorophore-conjugated secondary antibodies (anti-mouse or anti-rabbit Alexa Fluor-594 and Alexa Fluor-488, Thermo Fisher Scientific; anti-goat-FITC, Santa Cruz Biotechnology) was done for 1 to 2 h at room temperature (RT), in 0.2% BSA-0.1% Tween 20-PBS, in the dark. Phalloidin CruzFluor 555 or 647 (SantaCruz Biotechnology, 1:1000 in 1% BSA-PBS) staining was done following secondary antibody incubation, for 1 h at RT. Lastly, cells were counterstained with Hoechst 33342 (Thermo Fisher Scientific, 5 μg/ml) before mounting in ProLong Gold antifade mounting medium (Thermo Fisher Scientific).

OPP labeling, biotinylation, and streptavidin pull-down of newly synthesized proteins

For isolation of tagged newly synthesized proteins, cells were treated with 20 μM OPP or DMSO for 2 h, washed twice with ice-cold PBS, and lysed in click lysis buffer (100 mM HEPES, 150 mM NaCl, 1% nonylphenyl polyethylene glycol NP-40, pH 7.5, 2 mM PMSF, and EDTA-free protease inhibitors (Thermo Fisher Scientific)). OPP-labeled proteins were then tagged with biotin by click chemistry, following Terenzio et al. 2018 (18), with minor modifications. Reactions were carried out at RT for 2 h, under gentle rotation, in the presence of 1% SDS, 0.1 mM Biotin-PEG3-Azide (for Western blot, Click Chemistry Tools, AZ104-5; Cayman Chemicals, Cat. #23419) or Dde-Biotin-Picolyl-Azide (for MS Click Chemistry Tools, 1186-5), 0.1 mM THPTA, 1 mM TCEP, and 1 mM CuSO₄. Click-conjugated proteins were precipitated with five volumes of acetone, overnight at −20 °C, pellets washed with ice-cold methanol, and resuspended in PBS-1% SDS-1X protease inhibitors. Protein concentration was determined using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific). For streptavidin pull-downs, equal amounts of protein (200–300 μg) were incubated with 20 to 30 μl of Pierce Streptavidin Magnetic Beads (Thermo Fisher Scientific, 88817) in PBS-0.2% SDS-1% NP-40 for 2 h, at RT with gentle rotation. Beads were then washed twice with PBS-0.2% SDS-1% NP-40 for 10 min at 4 °C, three times with PBS-6M urea-0.1% NP-40 for 30 min at 4 °C, once with PBS-0.2% SDS-1% NP-40 for 10 min at 4 °C, and once with PBS for 1 to 2 min at RT. For samples to be used for MS, elution was achieved by treatment with 2% hydrazine (Sigma-Aldrich) 2× for 1 h at RT with gentle shaking, while in samples used for Western blotting, elution was done with 1× Laemmli buffer (62.5 mM Tris–HCl (pH 6.8), 2% SDS, 10% glycerol, and 50 mM DTT) for 5 min at 95 °C.

Western blotting

Equal amounts of protein (or equal volumes from streptavidin pull-down eluted fractions) were separated by Tris–glycine SDS-PAGE following Laemmli’s discontinuous buffer system and transferred to PVDF membranes using Trans-Blot Turbo Transfer system (Bio-Rad). The following antibodies were used for immunoblot: HRP-conjugated streptavidin (Columbia Biosciences HRP-2212, 1:10,000), anti-β-tubulin (Developmental Studies Hybridoma Bank, clone E7, 1:2000), anti-GAPDH (Millipore MAB374, 1:2000), anti-puromycin (Millipore MABE343, 1:2000), anti-β-III-tubulin (Sigma-Aldrich T2200, 1:2000), anti-calbindin 2 (calretinin, Cell Signaling Technology #92635, 1:1000), anti NeuN (Millipore MAB377, 1:750). HRP-conjugated secondary antibodies (anti-mouse IgG 32430 and anti-rabbit IgG 32460, Thermo Fisher Scientific) were used at a 1:5000 dilution. Cell type specificity of antibodies for NeuN, calbindin 2, and β-III-tubulin were validated in comparisons of primary cultures of cortical astrocytes, neurons, and neuron-glia cocultures (Fig. S4C). Immunoblots were developed with SuperSignal West Pico/Femto/Atto Maximum Sensitivity Substrate (Thermo Fisher Scientific) and images captured with ImageQuant LAS 4000 imaging system (GE HealthCare Life Sciences). For equal loading control, membranes were stained with Coomassie solution (0.1% Coomassie brilliant blue R-250 dye, 25% ethanol, and 7% acetic acid). Signal intensity quantification was done using the ImageQuant TL software (https://www.cytivalifesciences.com/en/us/shop/protein-analysis/molecular-imaging-for-proteins/imaging-software/imagequant-tl-10-2-analysis-software-p-28619) (GE HealthCare Life Sciences) on unsaturated images.

Mass spectrometry

Pull-down samples were precipitated with trichloroacetic acid deoxycholate precipitation (42) overnight. Protein pellets were solubilized in 20 μl of 7 M urea, 2 M thiourea, 100 mM ammonium bicarbonate, 20 mM methylamine buffer. Protein reduction was performed with 5 mM DTT by incubating 1 h at RT. Protein alkylation was performed with 10 mM chloroacetamide by incubating 1 h at RT in the dark. Next, protease LysC (Wako) was added to an enzyme:substrate ratio (E:S) of 1:50 (200 ng), and the samples were incubated for 4 h at 25 °C. Samples were then diluted five times with 100 mM ammonium bicarbonate, and dimethylated porcine trypsin (Sigma-Aldrich) was added to 1:50 E:S ratio and incubated overnight at 25 °C. After digestion, samples were acidified with TFA to a concentration of 1% and desalted on in-house–made C18 material StageTips (43). Samples were reconstituted in 0.5% TFA, and peptide concentrations were determined with a Pierce colorimetric peptide assay (Thermo Fisher Scientific). Two micrograms of peptides were then subjected to LC/MS/MS analysis.

Samples were injected to an Ultimate 3500 RSLCnano system (Dionex) using a 0.3 × 5 mm trap-column (5 μm C18 particles, Dionex) and an in-house packed (3 μm C18 particles, Dr Maisch) analytical 50 cm × 75 μm emitter-column (New Objective). Peptides were eluted at 250 nl/min with a 8 to 40% B 60 min gradient (buffer B: 80% acetonitrile + 0.1% formic acid, buffer A: 0.1% formic acid) to a Q Exactive HF (Thermo Fisher Scientific) MS using a nano-electrospray source (spray voltage of 2.5 kV). The MS was operated with a top-20 data-dependent acquisition strategy. Briefly, one 350 to 1400 m/z MS scan at a resolution setting of R = 60,000 at 200 m/z was followed by higher-energy collisional dissociation fragmentation (normalized collision energy of 26) of 20 most intense ions (z: +2 to +5) at R = 15,000 with 1.6 m/z isolation windows. MS and MS/MS ion target values were 3e6 and 1e5 with 50 and 15 ms injection times, respectively. Peptide match was set to preferred and exclusion of isotopes turned on. Dynamic exclusion was limited to 25 s.

MS raw files were processed with the MaxQuant software package (version 2.0.3.0) (44). Methionine oxidation, asparagine/glutamine deamidation, and protein N-terminal acetylation were set as variable modifications, while cysteine carbamidomethylation was defined as a fixed modification. Search was performed against UniProt (www.uniprot.org) Homo sapiens, Mus musculus, or Rattus norvegicus reference proteome databases using the tryptic digestion rule (cleavages after lysine and arginine without proline restriction). Only identifications with at least 1 peptide ≥ 7 amino acids long (with up to two missed cleavages) were accepted. Normalization with MaxQuant label-free quantification algorithm was also enabled. Label-free quantification ratio count (i.e., number of quantified peptides for reporting a protein intensity) was set to 1. Peptide-spectrum match and protein false discovery rate was kept below 1% using a target-decoy approach. Only unique peptides (with or without variable modifications) were used for protein quantification. All other parameters were default.

RiboTag

Neuron-glia cocultures were prepared as described above and transduced with AAV-[Syn]-Rpl22-HA on the day of seeding in poly-L-lysine–coated 10-cm dishes. Untransduced cultures were used as Ribotag-negative controls for defining the background in these experiments. After 7 days in vitro, cells were washed twice with ice-cold PBS and lysed on ice in 800 μl of supplemented homogenization buffer (50 mM Tris (pH 7.0), 100 mM KCl, 12 mM MgCl₂, 1% NP-40, 1 mM DTT, 1× EDTA-free protease inhibitor cocktail (P8340 Sigma-Aldrich), 300 units/ml RNasin (Promega), 150 μg/ml cycloheximide, and 10 mM ribonucleoside vanadyl complexes, R3380 Sigma-Aldrich). Cells were scraped; lysates were transferred to Eppendorf tubes, rotated for 5 min at 4 °C, and centrifuged at 4 °C for 10 min at 10,000×g. Cleared lysates were used for immunoprecipitation, and 10% of each sample was set aside as input. Eight micrograms of mouse anti-HA antibody (H3663, Sigma-Aldrich) were added to each sample and incubated for 4 h at 4 °C under gentle rotation. Eighty microliters of Protein G Dynabeads Magnetics beads (Thermo Fisher Scientific) pre-equilibrated with homogenization buffer were added to lysates preincubated with the anti-HA antibody and rotated overnight at 4 °C. After this, beads were washed 3 × 5 min at 4 °C with high salt buffer (50 mM Tris (pH 7.0), 300 mM KCl, 12 mM MgCl₂, 1% NP-40, 1 mM DTT, and 150 μg/ml cycloheximide). After washes, RNA was eluted from beads and purified using the RNeasy Mini Kit (Qiagen) and Epoch Life Sciences columns, following the supplier's protocol. Input samples were purified in the same manner. DNase digestion was performed on-column with RNase-Free DNase Set (Qiagen). Complementary DNA synthesis was done with 500 ng of RNA as starting material using an oligo(dT) primer (100 μM, Microsynth) and the Superscript IV system (Invitrogen). RT-qPCR was performed using HOT FIRE Pol EvaGreen qPCR Mix Plus (Solis BioDyne) on a LightCycler 480 II instrument (Roche). For each analyzed gene, the immunoprecipitation (IP) to input ratio (IP/Input) was calculated as 2ˆ^{- Cp-IP}/2ˆ^{- Cp-input}, where Cp is the average of the “crossing point” of the two technical replicates for each sample. Primers used are listed in the Table below.

Primer	Sequence (5′-3′)
r_Tmem130_F	CGTCTGGTCGGACCTCGT
r_Tmem130_R	AGGTTATAGAGACCTGCTGCC
r_Apbb1_F	GGCTAAACCCGTTGGGGTA
r_Apbb1_R	CTGCTGGTGCAAGATGGTGA
r_Stmn4_F	AAACCGGACTGCAGGATTGT
r_Stmn4_R	TCCTTATAGGCTGCGAGGGT
rm_Grin1_F	CTCCAAAGACACGAGCACCG
rm_Grin1_R	CTCCCTCTCAATAGCGCGTC
r_Clstn3_F	CTGGCGTCCTTAACACCATGA
r_Clstn3_R	TGGCTTATGCTTGTTCGCTTT
r_Aldh1l1_F	GCAGGTACTTCTGGATTGCA
r_Aldh1l1_R	GGAAGGTACCCAAGGTCAAA
r_Iba1_F	GCAAAGATTTGCAGGGAGGA
r_Iba1_R	CGTCTTGAAGGCCTCCAGTT
r_Cspg4_F	CCCTGGCTCCACTACAACTG
r_Cspg4_R	CTGCCTCCTGGACTACCTCTA
r_Pdgfra_F	TCCTCCGGGCTATCGGATTT
r_Pdgfra_R	GATGAGGCTCGGCCCTGTG
r_Cnp_F	ACCTGGTCAGCTATTTTGGC
r_Cnp_R	AAGATCTCCTCACCACATCCTG

Metabolic labeling with ³⁵S methionine/cysteine

HEK 293 cells were grown to confluence in 12-well plates in DMEM supplemented with serum and pulse-labeled for 2 h with 100 μCi/ml ³⁵S methionine/cysteine (EasyTag Express, PerkinElmer) in 0.5 ml/well of DMEM lacking methionine and cysteine (Gibco, Cat# 21013-024). PMY (2.5–20 μM), OPP (5–40 μM), or cycloheximide (50 μg/ml) were added as indicated together with ³⁵S Met/Cys label. At the end of incubation, cells were gently washed twice with 1 ml PBS at RT, lysed directly in 100 μl Laemmli buffer, and boiled. Twenty microliters of lysate were separated by 10% SDS-PAGE. Gels were stained with Coomassie solution, dried, and exposed to X-ray film (CP-BU, Agfa) overnight. Autoradiographs were imaged with ImageQuant LAS 4000 (GE HealthCare Life Sciences), and signal intensity over whole lanes was quantified using the ImageQuant TL software (GE HealthCare Life Sciences).

Data availability

The MS proteomics data have been deposited to the ProteomeXchangeConsortium via the PRIDE (45) partner repository with the dataset identifier PXD043494 and PXD043498.

Supporting information

This article contains supporting information.

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.

Reviewed by members of the JBC Editorial Board. Edited by Phillip A. Cole