MainCellular protein composition and how it is modified in response to physiological and pathological signals and events is key to understanding how cells function. It is clear that the same extra- or intracellular signals can lead to distinct proteomic responses in different cell types1,2. Indeed, bulk analyses of proteomes can ‘average out’ intrinsic cellular differences as well as cell type-specific differences in responses. One way to obtain cell type-specific proteomes exploits cell type-specific markers and fluorescent labeling, followed by enrichment using fluorescence-activated cell sorting for subsequent mass spectrometry (MS) studies. A drawback of this approach, however, is the loss of processes and compartments, such as dendrites or axons, during sample processing3,4. By contrast, BioID- and TurboID-based methods are explicitly designed to identify proteins from targeted subcellular regions. Because protein populations are labeled in a proximity-dependent manner at a specific time point, this method includes spatial and temporal information but does not distinguish between newly synthesized and preexisting proteins5.Bio-orthogonal methods based on the use of artificial amino acid analogs as baits for protein visualization and purification are particularly amenable to the study of cell type-specific proteomes6,7. The incorporation of artificial amino acids into proteins can be genetically controlled to enable cell type-specific labeling8,9. For instance, the expression of a mutant methionyl-tRNA synthetase with an enlarged methionine recognition pocket (L274G point mutation; hereafter referred to as MetRS*) allows the charging of the methionine analog azidonorleucine (ANL) to the corresponding methionine tRNA. ANL has a slightly larger structure compared with methionine. This size difference is sufficient to exclude ANL recognition by the endogenous MetRS; thus, ANL is not incorporated into proteins in cells where MetRS* is not present10. When MetRS* expression is driven by cell type-specific promoters, it enables specific labeling of proteomes with ANL and, after tissue lysis or fixation, derivatization of the azide group present in ANL with an alkyne by click chemistry. The alkyne can be used to visualize cell type-specific proteomes by immunofluorescence (FUNCAT) or western blot (BONCAT) and to purify these proteomes. The MetRS* system overcomes the above-mentioned challenges: it is possible to obtain cell type-specific proteomes without losing proteins located in dendrites or axons11. In addition, the time window of protein synthesis is determined by the exposure time of ANL, and protein degradation or subcellular location of ANL-labeled proteins is not altered; using this method, it is even possible to measure protein half-lives8,12.Based on this system, we originally created a mouse model carrying a knock-in of the MetRS* gene fused to green fluorescent protein (GFP) by a P2A peptide under an enhanced actin promoter. Expression of the cassette is switched on by Cre-dependent excision of a transcriptional stop. Crossing this mouse line with cell type-specific Cre drivers enables cell type-specific protein labeling and its subsequent visualization, purification and identification. Given the considerable number of Cre drivers, the described system can be used to study cell type-specific proteins from almost any tissue in any field. Proteins can be labeled with ANL in live animals or in vitro. Using this first generation of mice, we demonstrated that it is possible to purify and identify cell type-specific proteomes allowing the identification of the respective cell type of origin. Furthermore, we identified proteins whose expression in the excitatory hippocampal neurons is modified in response to an enriched environment, a well-established paradigm for synaptic plasticity enhancement8. It is also possible to identify changes in protein expression in models of prion diseases13. However, we observed limitations when we aimed to obtain proteomes from relatively low-abundance cell types. Similarly, the identification of proteins labeled for short periods was not possible in brain tissue using this mouse model. Furthermore, given that the line expresses a single copy of the MetRS*, we probably observed substantial competition with methionine incorporation by the endogenous MetRS. Indeed, to obtain good in vivo labeling, a low-methionine diet and homozygous MetRS* alleles were required. The need for homozygosity complicates the crossing of this mouse line into a disease or any other background of interest. To overcome these limitations, we describe here a second generation of the MetRS* mouse tool designed to increase the expression of the mutant enzyme. The new design of the line boosts ANL incorporation into proteins, allowing the isolation and identification of low-number neuronal populations, such as the dopaminergic (DA) neurons, and proteins synthesized in the excitatory neurons of the cortex after just 3 h of a single intraperitoneal (IP) injection of ANL.ResultsThe 3xMetRS* mouse lineTo achieve more efficient ANL labeling to purify proteomes from sparse cell populations or from proteomes after short ANL labeling times, we focused on improving the expression of MetRS*. We therefore modified and optimized the allele design of the previous MetRS* mouse line (termed 1xMetRS* here). We maintained the overall strategy of a Cre-dependent excision of a floxed transcriptional stop to switch on transcription of a cassette with MetRS* and GFP in the ROSA26 locus (Fig. 1a). Whereas the 1xMetRS* line carried a GFP sequence fused to the MetRS* coding region, separated by the self-cleaving 2A peptide sequence (P2A) (Fig. 1b), the new MetRS* mouse line described here (hereafter referred to as 3xMetRS*) expresses a cassette carrying three copies of MetRS* coding sequences and one copy of GFP per allele, with all four parts separated by sequences coding for different self-cleaving 2A peptides (Fig. 1c). As this expression cassette codes for a very large mRNA (and protein before separation of the single units by self-cleavage), we added an mRNA stabilization element (WPRE) to boost expression of the exogenous mRNA. In addition, we shifted the GFP to the C-terminus of the encoded protein (Fig. 1c). As such, we ensured that, if GFP is detected, MetRS* is also translated. Mice carrying the 3xMetRS* allele and the Cre::3xMetRS* lines used in this work were viable and fertile, and both heterozygous and homozygous mice showed no obvious behavioral differences compared with wild-type mice. We use the MetRS* copy numbers to refer to the genotype: 1xMetRS* and 2xMetRS* when studying heterozygous or homozygous animals, respectively, in the previously described mouse line; and 3xMetRS* or 6xMetRS* when using heterozygous or homozygous animals of the new line (Fig. 1).Fig. 1: Comparison of the two mouse lines for ANL cell type-specific protein labeling.a, Insertion of the MetRS* cassette in the ROSA26 locus of wild-type mice (WT). b, Scheme of the cassette introduced in the first generation of the floxed-STOP-MetRS* mouse line. The right part shows mice expressing one (1xMetRS*, heterozygous animals) or two (2xMetRS*, homozygous animals) copies of the transgene after recombination by Cre. c, Scheme of the cassette introduced in the second-generation floxed MetRS*, showing all the introduced genes separated by 2A peptides, and the mRNA stabilization element (WPRE) at the end of the cassette. The right part shows mice expressing three (3xMetRS*, heterozygous animals) or six (6xMetRS*, homozygous animals) copies of the gene after recombination by Cre. SA, splice acceptor.Full size imageNeuronal ANL labeling in vitro with different MetRS* copy numbersTo analyze the utility of increasing the MetRS* copy number, we compared ANL incorporation between the original and second-generation lines. First, we evaluated whether there was increased ANL labeling in the 3xMetRS* line compared with the 1xMetRS* line in vitro. We crossed animals carrying the respective alleles to the Nex-Cre driver line, driving Cre expression to excitatory neurons14, and performed experiments on cultured cortical neurons derived from the offspring. Experiments in culture have the advantage that ANL labeling is direct and circumvents any variability arising from ANL administration or uptake into the brain. Comparison of 2-h ANL labeling in either 1xMetRS* or 3xMetRS* excitatory neurons (Nex-Cre::1xMetRS* and Nex-Cre::3xMetRS*) evaluated by FUNCAT showed that ANL labeling was approximately three times higher in neurons expressing 3xMetRS* (Fig. 2a,b)12. FUNCAT experiments in the presence of the protein synthesis inhibitor anisomycin (Fig. 2c) confirmed that the observed fluorescent signal was due to ANL incorporation into newly synthesized proteins and not due to detection of free ANL or ANL-charged tRNAs. Interestingly, we observed that the increased ANL labeling in the Nex-Cre::3xMetRS* line compared with Nex-Cre::1xMetRS* was accompanied by higher GFP expression—even though both lines carry one copy of GFP (Fig. 2d). Similarly, western blot comparisons of ANL labeling in 1xMetRS* or 3xMetRS* cultured neurons by BONCAT also showed an increase in ANL labeling (Fig. 2e,f). To examine whether the observed increase in labeling with the 3xMetRS* line was also evident in tissue, we prepared acute slices from brain tissue of adult Nex-Cre::1xMetRS* or 3xMetRS* mice and performed in vitro slice metabolic ANL (2 h) labeling experiments. As was observed in the cultured neurons, FUNCAT labeling was substantially elevated in slices expressing the 3xMetRS* transgene (Fig. 2g).Fig. 2: MetRS* copy number influences ANL labeling.a, Representative confocal images showing ANL labeling by FUNCAT after 2 h of incubation, in primary cortical neurons from Nex-Cre::1xMetRS* or Nex-Cre::3xMetRS* mice with (+) or without (−) ANL, showing an increase in labeling in the 3xMetRS* mouse line. Scale bar, 100 µm. b, Quantification of the experiments shown in a. Multiple-comparison ANOVA, ****P