Integration of single‐cell transcriptome and proteome technologies: Toward spatial resolution levels

Cells are basic building blocks of life with vast heterogeneity. Nowadays, the rapid development of single‐cell multiomics (scMulti‐Omics) has facilitated comprehensive understanding of gene regulatory networks, cellular characteristics, and temporal dynamics. However, simultaneous analysis of transcriptome and proteome at single‐cell level still faces huge challenges due to their differences in molecular modalities. Recent technological advances in single‐cell manipulations, barcoding, and ultrasensitive instrument recently offer unprecedented opportunities for the co‐profiling of genes and proteins. In this review, multiple types of single‐cell isolation, lysis, and molecular separation technologies are first introduced. Second, various approaches for co‐measurement of transcriptome and proteome in single‐cells are summarized, with their advantages, limitations, and capacity for targeted or unbiased deep analysis. Then we highlight the cutting‐edge spatial multiomics methodologies that operate at the single‐cell or subcellular resolution level, providing a comprehensive understanding of cell function and heterogeneity within the tissue spatial environment. The emerging biomedical applications of multiomics are also discussed. Finally, the challenges and prospects of this field are proposed.


INTRODUCTION
Cellular heterogeneity is pervasive in various cellular systems and which observed not only in different tissues but also within the same cell type. 1,2Since single-cell sequencing was selected as the method of the year in 2013 3 , the rapid development of single-cell analysis technique has contributed to biology research revealing cellular heterogeneity and its impact on cell functions, now permitting single-cell genome 4 , epigenome 5 , transcriptome 6 , proteome 7 , or metabolome 8 analysis of heterogeneous cell types in tissue samples and different cell states.However, the extracting and analyzing of one molecule type in single cells can only provide incomplete information because the performance of individual cells is determined by the interaction of multiple types of molecules (DNA, RNA, or protein).][14] In recent years, with single-cell RNA sequencing (scRNA-seq) as the intermediate mediator included in most scMulti-Omics studies, genome-transcriptome 15,16 and epigenome-transcriptome 17,18 measured simultaneously have been relatively mature, mainly relying on the amplification of DNA and RNA.In contrast, the "central dogma" reveals that proteins are translated by mRNA and are the direct executor of cellular functions 19 , which indicate that protein expression level plays an essential role in cellular phenotype analysis.Nevertheless, mRNAs and their cognate proteins are non-linear, with low-or even negative correlation due to the protein translation rate, half-life of proteins or mRNAs, or degradation of product, resulting that difficulty in directly predicting the expression of proteins from transcriptome data. 20,21inking transcriptome and proteome scMulti-Omics measurement can enhance the analysis of exploring the relationship between mRNA and protein abundance and leading to more robust definitions in different cell types/states and functions, which is of great value and significance in biological fields such as screening drug targets, exploring pathogenesis and prognosis assessment.However, mRNA and proteins are very different molecular modalities due to the proteome methods are not available for amplification like oligonucleotides, moreover the low protein content (only about 150 pg of protein per cell) as well as the difficulty to implement multiplexing and high-throughput barcoding techniques in scRNA-seq. 22,23 a result, an increasing number of researchers dedicate to developing effective methods for simultaneous analysis of the single-cell transcriptome and proteome recently to bring this longstanding discussion to the fore. 24Moreover, recent developments in spatial resolution level omics techniques based on two-dimensional cellular mapping can comprehensively describe the heterogeneity of different single cells within three-dimensional tissues of origin, whereas maintaining their spatial and morphological contexts intact. 25any excellent reviews have comprehensively summarized the single-cell and spatial omics strategies, respectively [26][27][28] , however, most of them are either too broad in scope and cover various types of omics integration, or only focus on a specific single-cell or spatially resolved level mono-omics, there are few about simultaneous integration transcriptome and proteome from single-cell level to spatial resolution.In this review, we first summarize various single-cell isolation approaches, lysis strategies, as well as molecule separation approaches between mRNA and protein.We then present the integration of targeted-based single-cell transcriptome and proteome joint analysis technologies including their advantages and limitations, along with breakthrough technologies that the measurement in a nontargeted deep manner.We also focus on multiomics analysis strategies at the spatial resolution including methodologies and applications, and the potential value of combining with scMulti-Omics level.Finally, the challenges and prospects of the field are discussed (Figure 1).

Single-cell isolation technologies
The initial step for scMulti-Omics analysis is to isolate the single-cell randomly from a population with heterogeneity which can be used in downstream processing (e.g., lysis and encode) and measuring.Therefore, it is highly essential to develop strategies for the efficient and mild isolation of single-cell from tissue sample matrixes or cell populations. 29,30In recent years, several popular methods used for single-cell isolation have been developed based on different technologies and can be predominantly divided into two categories   F I G U R E 2 Joint profiling of single-cell isolation methods.A The limiting dilution method isolates single cells.B Micromanipulation involves collecting single cells by microscope-guided capillary pipettes.C Single-cell manipulation system using plasmonic tweezer.Reproduced with permission from Kang et al. 31 Copyright 2023 John Wiley & Sons, Inc. D LCM utilizes microscopic visualization laser operation system to isolate cells from solid tissue samples.E FACS isolates single-cell by labeling cells with fluorescent marker proteins.F Microfluidic system for single-cell high throughput isolation.An example of microfluidics (e.g., droplet-seq).

Low throughput-based single-cell isolation technologies
Manual cell picking is a conventional method for isolating single cells.Common devices for manual cell isolation typically rely on visualized devices such as microscopes and other image-based systems.Serial limiting dilution is the most used method in which pipettes are used to isolate single cells from the cell suspension 32 and can help limit the degradation of more volatile molecules such as RNA or protein (Figure 2A), but this method is limited by efficiency and randomness.The capillary-based system has been a reachable modular assembly component for single-cell isolation and is compatible with automated micromanipulation (Figure 2B).For example, an integrated optically guided in situ subcellular capillary microsampling method was developed for direct sampling of single cells of living vertebrate embryos. 33However, the potential limitation of this method is difficult to achieve a targeted selection of interest single cells from suspension.
Lately, our group designed a plasmonic tweezer optical method (Figure 2C) consisting of an optical fiber probe, gold thin film, and thermosensitive hydrogel layer to realize selective capture of target cells, and it displays excellent 3D biocompatibility as well as low energy consumption. 31esides, the strategy of manual isolation namely laser capture microdissection (LCM) 34 (Figure 2D), can be available for collecting individual cells 35 or tissue regions of interest (ROI) from solid tissue samples 36,37 under microscopic visualization laser operation system.Laser-induced forward transfer (LIFT) 38 is a similar method to LCM, which has been successfully used to isolate single-cell 39,40 combined with other optical techniques.The most important advantage of the LCM/LIFT technique is speed, precision, and retaining spatial information.Despite this, the challenge of low throughput and reproducibility cannot be ignored.Thus, the strategies mentioned above are more suitable for collecting small cell samples or when labeling is not required.

High throughput-based single-cell isolation technologies
Recently, increasing research based on flow-cytometry have made significant contributions to high-throughput single-cell analysis. 41,42Fluorescence-activated cell sorting (FACS), a specialized form of flow-cytometry 43 , is the most affinity and non-destructive technique for defining different cell types based on specific light scattering and/or fluorescent properties of each cell (Figure 2E), and has become the most common strategy for isolating single cells currently. 44Notably, FACS also allows linking the phenotype (e.g., intercellular or intracellular marker expression) of single cells with multimodal omics analysis. 45For example, Katzenelenbogen et al. 46 developed an integrated technology for simultaneously massively parallel scRNAseq and intracellular protein measurements.Briefly, the cells were tagged intracellularly with fluorescence-labeled antibodies and sorted by FACS according to their intracellular fluorescent signal intensity.Despite the great advances in FACS single-cell isolation, it still remains challenges, including the requirement for a large amount of starting volumes (>10,000 cells) in suspension and the limited number of available fluorescence-labeled antibodies to target dozens of interest proteins.
Currently, microfluidic is recognized as a popularity technology for single-cell isolation due to device miniaturization, low cost, higher throughput, and precise fluid control 47 (Figure 2F).Various microfluidic-based platforms have been developed for single-cell isolation based on different techniques, including hydrodynamic flow 48 , droplet-based 49 , valve-based 50 , microwell techniques 51 , and so on.There have been several reviews in the field of microfluidic platforms that have detailed reports on various technologies. 29,52,53Herein, what we mainly want to emphasize is that microfluidics has been utilized in scMulti-Omics analysis (transcriptomics-genomics 16 , transcriptomics-epigenomics 54 , transcriptomics-proteom-ics 55 ) due to the development of individual control of valves and switches.Despite the microfluidics platform may suffer from potential challenges (for example, droplet microfluidics-based platforms suffer from low cell utilization (25%) due to the Poisson distribution to ensure that co-encapsulation of single cells and single barcoded beads), it has great promising prospects in single-cell isolation and multiomics research in the future.

Single-cell lysis methods
Single-cell lysis is a crucial step following cell isolation to efficiently extraction and the downstream omics analysis of intracellular components.Careful selection of the proper lysis method is essential to collect accurate data from single cells, as each method has its respective strengths and weaknesses.

Chemical lysis
Chemical lysis is a common method that mainly uses various hydrolases, denaturants, or detergents to disrupt the membrane structure of cells, resulting in cell lysis.Detergent-based lysis results from the incorporation of detergent into the cell membrane, dissolving lipids and proteins in the membrane, creating pores within the membrane, and ultimately completing cell lysis.Detergent lysis has been well applied to single-cell omics analysis.For example, sodium dodecyl sulfate is the most used powerful lysis strategy which provides cell lysis of the order of seconds.However, it is limited in the level of singlecell protein analysis due to the ability to inhibit pancreatic enzyme activity.Triton X-100 is a milder non-ionic detergent that is compatible with applications involving protein activity. 56n-Dodecyl-b-D-maltoside (DDM) is a family of mild nonionic surfactants which is similar to Triton X-100.It can adsorb strongly on hydrophobic surfaces and form a monolayer, making the surface hydrophilic and nonionic, thereby reducing sample losses due to adsorption between the protein and the surface. 57,58Zhu's group has confirmed that the DDM lysis buffer and dropletchip combination could be efficiently and sensitively applied to single-cell proteomic and even spatial proteomic levels. 36Alternatively, RapiGest has been receiving considerable attention in the field of shotgun-mass spectrometry (MS) because it allows for efficient solubilization/digestion of protein and is easy to remove from peptide samples before MS analysis. 59Fang's group lysed single cells with RapiGest in nanoliter-scale oil-air-droplet chip and later with shotgun-MS mode for single-cell proteomic analysis. 60More recently, based on this lysis strategy, they have achieved simultaneous analysis of transcriptome and proteome in the same cell. 61Although chemical lysis is widely used in single-cell processing due to its simplicity and economical, it is noteworthy that the chemical treatment of the membrane takes a long time and its residue needs to be removed to avoid effects on downstream analysis.

Mechanical lysis
Mechanical lysis is an effective method to destroy cell membranes with contrivable force, with high lysis efficiency and low selectivity, which can lyse cells more vigorously and comprehensively.Ultrasonicated singlecell lysis is a classic method, using ultrasonic waves to generate localized areas of high pressure which results in cavitation and can shear cells. 62Additionally, after several studies of optimization, it has been possible to set reasonable ultrasonic time and gap time of lysis to allow efficient separation of cell contents and prevent effects on downstream analysis. 63Multiple washing steps are omitted in cell lysis using sonication which can eliminate clean-up-related losses in proteome involved in multiomics.For instance, Slavov's group quantified a thousand proteins in differentiating mouse embryonic stem cells with the combination of ultrasonic lysis method and Single Cell ProtEomics by Mass Spectrometry (SCoPE-MS) platform. 64Freeze-thaw cycle-based strategy is also commonly used in single-cell lysis which mainly through swelling and causes the disruption of the cell structure.For example, Specht et al. 65 reported an automated and miniaturized cell lysis method by Minimal ProteOmic sample Preparation (mPOP) for single-cell proteomics analysis.It uses a freeze-thaw cycle ranging from −80 • C to 90 • C in water droplet that can efficiently extract proteins, and the step of cleanup before MS analysis can be omitted.Similarly, Zhu's group 66 proposed a nanodroplet microchip combined with freeze-thaw cycle cell lysis to realize the parallel measurement of single-cell transcripts and proteins multimodal omics.It has been demonstrated that the freeze-thaw cycle-based cell lysis strategy is mild and effective, but can damage large-sized biomolecules and result in possible RNA fragmentation, so this strategy is more compatible with single-cell 3′ mRNA-seq instead of full-length mRNA sequencing. 67xcept for the widely used strategies described above, there are other methods such as optical and electrical methods that can achieve efficient lysis of single-cell. 68,69verall, many elements should be considered when choosing the lysis strategy, including lysis efficiency, cost, degree of impact on downstream analysis, etc., as each method has significant advantages and drawbacks.Briefly, no best solution for all situations can be utilized for single-cell lysis, the proper lysis techniques should be decided under specific details such as the type of tissue, the cost, and even a combination of several different lysis methods could also be considered.

Molecular separation methods of RNAs and proteins in single-cell
After isolating of single-cell from the population, multiple types of molecules from the same cell need to be separated for subsequent measurements.For mRNA and protein molecules, some reports used the splitting strategy to split cell lysates into two parts after lysis and then quantify the transcriptomes and proteomes of the same cell, respectively. 70,71Apart from this, there are also strategies reported that allow the capture of protein and RNA directly after cell lysis for a "one-pot" assay step without splitting into two parts. 72Another strategy used antibodyfluorescence/antibody-oligos complexes to combine with the target protein in single-cell, followed by imaging of protein fluorescence intensity or indirect conversion to oligos.Then the cells were lysed to release the mRNA for transcripts measurement (parallel measurements if there are oligo-tags by protein conversion). 73Besides, there is an interesting strategy that added oligo-dT magnetic beads into the cell lysis system to capture the mRNA in the lysate, since the mRNA has a poly(A) sequence at the 3′ end tail.Then the cell lysate containing the protein supernatant was transferred through magnetic enrichment, allowing mRNA and protein to be efficiently collected in single cells and for further downstream analysis. 74We are confident that an efficient separation strategy for RNAs and proteins can provide analyte feedstock for subsequent scRNA-seq and targeted proteins measurements or label-free nontargeted measurements, enabling truly scMulti-Omics analysis of transcriptome and proteome within the same single cell.

INTEGRATIVE scMULTI-OMICS ANALYSIS OF TRANSCRIPTOME AND PROTEOME
The selection of method for proteome quantification is a key element in achieving scMulti-Omics analysis.Currently, the two main approaches of single-cell proteome analysis are targeted-based and nontargeted assays. 75The most widely used target label-based approaches for specific target proteins detection are as follows: fluorescent probes combined with high-resolution imaging 76 ; antibodies-based strategies coupled with flow-cytometry 77 , mass cytometry (CyTOF) 78 , or microfluidics. 79][82][83] Joint profiling of transcriptome and proteome multiomics in the same cell offers the possibility to reveal the relationship between cellular state and actual phenotypes of single cells. 84Several methods that can be simultaneous for scMulti-Omics analysis of transcriptome and proteome have been developed, including target labels-based and nontargeted-based strategies (Figure 1B).Each of these approaches has advantages and disadvantages related to levels of sensitivity, multiplexing or analytical throughput (Table 1).

3.1
Targeted-based single-cell transcriptome and proteome analysis

3.1.1
Fluorescence-labeled antibodies Immunofluorescence (IF) imaging is widely used to profile protein levels directly with ultra-high sensitivity and resolution.Combining fluorescence-based imaging proteins with RNAs analysis methods provides the possibility for simultaneous detection of transcriptome and proteome in the same cells.
The microwell-based device allows the isolation of single cells into physically quarantined confinements, representing a dynamic platform to generate single-cell multidata analysis.Park et al. 73 designed a microwell device for single-cell loading and then simultaneously measuring transcript and proteins level from the same cells (Figure 3A).The target proteins were stained with immunofluorescence antibodies in individual cells, followed by imaging of the protein fluorescence intensities.The cells were then lysed and the extracted transcripts expression was quantified via reverse transcription (RT)-PCR.The targeted protein and mRNA were measured and analyzed for correlation from the same cell with extracted fluorescence signals by using MATLAB software.To address the limitation of the number of detected mRNAs, George et al. 85 developed a splittable singlecell microchip for both genome-wide transcriptome and secreted proteins detection from the same single immune cells (Figure 3B).Single cells were seeded into nanosize microchambers and then sealed with a high-density antibody array for capturing cytokines proteins.Then the single cells with a protein profile were picked up by a syringe needle and lysed immediately for sequencing of mRNAs.By measuring both transcriptome and cytokine proteins from the same single macrophage cells, a deeper understanding of the complex regulation of heterogeneous cell populations can be achieved.Recently, a FACS-based method called INs-seq 46 , is used for intracellular protein immunodetection followed by scRNA-seq (Figure 3C).The fixed and permeabilized cells were labeled intracellularly with fluorophore-conjugated antibodies and sorted by the FACS technique of single-cell isolation according to the intracellular fluorescent signal intensity, then the labeled cells were collected for scRNA-seq using plate-based or droplet-based approaches.The INs-seq technology allows discovering new immune subsets and metabolic pathways by profiling transcriptome and intracellular posttranslational modification of proteins in relevant cell pathways analysis.However, the fluorescence-labeled imaging methods above are fundamentally limited in the detected number and throughput due to spectral overlap and sensitivity.To address the shortcoming of these methods, cognate antibody pairs-based proximity ligation assay (PLA) and proximity extension assay (PEA) methods have been developed to improve the throughput and abundance of both mRNAs and proteins analysis levels.

PEA/PLA-labeled cognate antibody pairs
PLA is performed by the ligation of two antibodyconjugated oligonucleotides in proximity to the same target protein. 86Based on this, Stahlberg et al. 87 utilized a strategy by using reverse transcription and PLA coupled with qPCR for simultaneous quantification of DNA, mRNAs, miRNAs, noncoding RNAs, and proteins in the same cell.Similarly, Frei et al. 88 reported another approach called proximity ligation assay for RNA (PLAYR) for the simultaneous quantification of more than 40 different mRNAs and proteins in thousands of single cells by using CyTOF.Pairs of PLAYR oligonucleotide probes were designed to hybridize with two adjacent regions of target mRNAs after cell fixation and permeabilization.The backbone and inserted oligonucleotides could bind and circularize with two adjacent probe pairs.After ligation, the rolling circle amplification (RCA) reaction could be triggered by polymerase to generate a large amount of replicate complementary strands of the template.Different RCA products generated by each mRNA could be hybridized with specific isotope-labeled oligonucleotides complementary strands and measured together with isotope-labeled antibodies on proteins using CyTOF.However, these methods are limited by relative quantification in the workflow.To overcome this, Albayrak et al. 70 designed a workflow that combines PLA and digital PCR for sensitive and absolute quantification of both  mRNA and proteins in single cells (Figure 4A).In digital PLA, the oligonucleotide-bound antibodies bound to the target protein from lysed cells, then the connector oligonucleotide hybridized and ligated with proximity probes to form a double-stranded DNA (dsDNA).After proteolysis step, the remaining dsDNA (and the cDNA in RT-droplet digital PCR workflow) was emulsified to create 20,000 droplets under limiting dilution and these molecules were amplified by PCR for simultaneous detection.PEA utilizes a similar method but relies on the hybridization rather than ligation, of two antibodyconjugated oligonucleotides with complementary 3′ends on the same protein target, which convert proteins into DNA oligos, and random RT is carried out for mRNAs using RT primers to generate cDNAs. 71,89Both DNA oligos and cDNAs are extended by polymerization to generate an amplifiable molecule for detection by qPCR or sequence.For example, Genshaft et al.90 presented an approach for a couple of PEA-based detection of protein and RT-based RNA analyses from single cells using the Fluidigm C1 platform, which had enabled parallel measurement of 38 proteins and 96 transcripts, and further explored the correlation and synergies between RNAs and proteins of the human breast adenocarcinoma cell line (Figure 4B).As an improvement, an approach called single-cell protein and RNA Co-profiling (SPARC) that combined scRNAseq with PEA was presented recently to enable parallel measurement of whole mRNA and high multiplex, targeted intracellular proteins in single cells. 74After cell lysis, mRNA was captured by the added oligo-dT beads and reacted using the Smart-seq2 method for full-length transcripts measurement.The cell lysate containing the protein supernatant was transferred and then use the PEA strategy-based cognate antibody pairs to bind the target protein, which allowed for the simultaneous measurement of 89 proteins in single cells.

Oligonucleotide barcode-labeled antibodies/aptamers
Although the PEA and PLA-based method circumvents some of the limitations in immunofluorescence imaging, the comprehensiveness of mRNA needs to be improved, the number of proteins is still limited because cognate antibody pairs are required to label the target protein and the single-cell analysis throughput is also to be considered.A high throughput droplet microfluidic-based single-cell analysis strategy has been proposed recently to achieve co-measurement of mRNA and protein through F I G U R E 4 Workflow of PEA/PLA-labeled cognate antibody pairs strategy for single-cell transcriptome and proteome analysis.A Schematic diagrams of digital PLA protocol for absolute protein and mRNA quantification from single cells.Reproduced with permission from Albayrak et al. 70 Copyright 2016.Elsevier.B Workflow for PEA/STA detection in single cells.Reproduced with permission from Genshaft et al. 90 Copyright 2016.Part of Springer Nature.oligonucleotide barcodes-labeled antibodies. 91Cellular Indexing of Transcriptomes and Epitopes by sequencing (CITE-seq) was developed by using oligonucleotide sequence barcodes-labeled antibodies to profile the cellular proteins and unbiased transcriptome for thousands of single cells in parallel 92 (Figure 5A).The antibody-derived DNA tags (ADTs) containing PCR handles, antibody barcodes, and poly(A) tails, which were first bound to target proteins on the cell surface.After cells were sorted and lysis within the droplets, both mRNAs and antibody anneal-derived oligos were captured by drop-seq beads containing oligo-dT primers.The library is generated by using cDNA molecule and PCR amplification, allowing for simultaneous quantifying of both mRNAs and proteins via sequencing.RNA expression and protein sequencing assay (REAP-seq) 93 is similar to CITE-seq but differs only in the barcode conjugated to the bead, which reduces steric hindrance and potential crosstalk.Based on this, the method could quantify proteins with 82 barcoded antibodies and over 20,000 genes (Figure 5B).To overcome the limitation of the input cell number and high-reagent consumption, Yang's group developed a multipaired-seq microfluidic platform for about 1000 cellular indexing of mRNAs and membrane proteins (Figure 5C). 94The basic principle was similar to CITE-seq and REAP-seq but the pump/valve structure was designed to remove free DNA-barcoded antibodies and mRNAs for more accurate measurements.Based on hydrodynamic differential flow resistance, the cell utilization was improved.Nevertheless, the above methods are unfortunately still limited to cell surface proteins.To address this challenge, single-cell RNA and immunodetection 95 (Figure 5D) allows for codetection of transcriptome and intracellular (phospho-) proteins level, primarily after crosslinking and permeabilization, immunostaining, and conversion of proteins into RNA using antibody RNA-Barcode Conjugates, followed by reverse crosslinking and cDNA synthesis for library sequencing of both transcriptomes and proteomes detection.
Despite the highly specific and affinity of antibody labeling proteins methods, they are limited by cost and stability.As an alternative, Delley et al. 96 presented a more simple, high-purity generation and lower-cost approach called Apt-seq (Figure 5E) to simultaneously characterize the transcriptomes and proteomes of single cells using oligonucleotide barcodes-labeled aptamers, which allow F I G U R E 5 Scheme of oligonucleotide barcode-labeled antibodies/aptamers strategy for single-cell transcriptome and proteome analysis.A CITE-seq used DNA-barcoded antibodies for the simultaneous detection of single-cell transcriptome and proteins.Reproduced with permission from Stoeckius et al. 92 Copyright 2017, Springer Nature.B Schematic workflow of REAP-seq.Reproduced with permission from Peterson et al. 93 Copyright 2017, Springer Nature.C Schematic overview of the multipaired-seq workflow based on microfluidic platform.Reproduced with permission from Xu et al. 94 Copyright 2022, American Chemical Society.D Schematic overview of the RAID workflow.with permission from Gerlach et al. 95 Copyright 2019, Springer Nature.E Schematic workflow of Apt-seq using aptamer probes to label protein markers.Reproduced with permission from Delley et al. 96 Copyright 2017, Springer Nature.single cell surface binding and profiling via droplet-based sequencing.

Nontargeted single-cell transcriptome and proteome analysis
Nevertheless, multiomics approaches that rely on affinity reagents such as antibodies or aptamers for protein analysis still prevent unbiased analysis, which can only enable tens to hundreds of targeted protein measurements in single cells.Incomplete data may limit the research of many potentially interesting proteins and subsequent multiomics in-depth analysis.MS-based single-cell proteomics technique that enables minimal sample requirements, high sensitivity, unbiased, and in-depth measurement of protein analysis.Combining scRNA-seq with MS-based proteomics techniques can overcome the limitations of targeted specificity, enabling the simultaneous quantification of thousands of proteins and mRNAs from the same cell.
In a recent study, Fang's group developed a singlecell simultaneous transcriptome and proteome (scSTAP) analysis platform based on full-length RNA-seq and MSbased technology to achieve a deep and joint quantitative analysis of transcriptome and proteome at the singlecell level, with an average quantitative depth of 19,948 genes and 2663 protein groups in single mouse oocytes 61 (Figure 6A).Briefly, precise sample splitting of single-cell lysates in the nanoliter range and multistep sample pretreatment was performed using the sequential operation droplet array system after single-cell capture, isolation, and enzyme-assisted cell lysis.Subsequently, two evenly distributed aliquots of single-cell components such as RNAs and proteins, one for transcriptome measurement by the MATQ-seq workflow, and the other aliquot for proteome analysis using MS-based shotgun proteomics method.Zhu's group developed a similar multiomics platform named nanoSPLITS (nanodroplet Splitting for Linkedmultimodal Investigations of Trace Samples; 66 Figure 6B).Briefly, single-cell sorting and freeze-thaw cycle-based cell lysis were performed first.Then the microchip containing single-cell lysate was manually merged and mixed with a separate chip containing only cell lysis buffer.Nanoliterscale cell lysates were divided equally via two droplet microarrays and used for downstream scRNA-seq and shotgun-MS proteomics measurements, respectively.This method demonstrated that the nanoSPLITS can reliably profile more than 5000 genes and 2000 proteins per single cell, and identified cell-type-specific markers from both modalities.
Analyzing in unprecedented depth of transcriptome and proteome at the single-cell level can provide comprehensive insight into cellular features and cell-cell interaction.Fang's group explored the regulatory features of transcription and translation during oocyte meiotic maturation, and 30 transcript-protein pairs were identified as specific oocyte maturational signatures. 61Similarly, Zhu's group performed an in-depth multiomics analysis of mouse alveolar epithelial cells, observing the dynamic nature of mRNA relative to its protein counterparts as well as differentially expressed markers at the protein and gene levels. 66e believe that multiomics measurement of mRNAprotein that incorporates both scRNA-seq and MS data represents an exciting area of development for the future of single-cell research.

Integrative multiomics analysis of single-cell transcriptome and proteome data
The maturation of scMulti-Omics technologies also requires the development of new computational methods to integrate information from different data types.Fortunately, various matched (that multimodal data is measured from the same cell) data integration methods of single-cell transcriptome and proteome multiomics had been developed for normalization and modeling, including BREM-SC 97 , TotalVI 98 , CiteFuse 99 and Seurat 4.0. 100Up to now, there have been excellent reviews analyzing multiomics data integration models elsewhere 26,84,101 , which are not repeated here.

SPATIALLY-RESOLVED MULTIOMICS OF TRANSCRIPTOME AND PROTEOME
Despite the explosion of scMulti-Omics techniques which offer unprecedented opportunities to explore cellular diversity and heterogeneity more comprehensively, it loses spatial location information for cell-cell interactions and overall cell phenotype/state.Actually, in tissue systems, the functions of cells are closely related to their spatial location and surrounding microenvironments.Exploring the way cells function in the spatial environment is crucial for uncovering biological information processes.In order to preserve the position information of cells in spatial and reconstruct the spatial expression map, various spatial omics methods have been developed to date.After singlecell multimodal omics was selected as the 2019 method of the year 102 , spatial multiomics (SM-Omics) technology was also voted as one of the seven technologies to watch by Nature in 2022 103 , which could map multiomics data into situ expression profiles to deeply analyze the spatial distribution characteristics of cellular and molecular expression profiles that provide new insights for revealing physiological and pathological processes.Spatial proteomics 27,104 and transcriptomics 105,106 techniques have been reviewed independently in previous reports.In this section, we mainly discuss SM-Omics methods that integrate transcriptome and proteome in recent years, which will improve the overall understanding of complex tissue with high dimensions or cellular structures with precise localization requirements.Typical spatially-resolved multiomics of transcriptome and proteome in the past five years are listed here (Table 2).

Spatial transcriptome and proteome analysis methods
The aim of developing spatially resolved techniques is to maximize the number of markers that can be observed simultaneously.These methods can be divided into (1) next-generation sequencing-based whole-transcriptome level using massively parallel sequencing barcode and (2) multiplexed probe-or antibody-based in situ labeling.
For the first type of method, in 2020, Fan's group reported deterministic barcoding in tissue for spatial omics sequencing (DBiT-seq) 107 for co-mapping of whole transcriptome and a panel of 22 proteins in tissue (Figure 7A).Two perpendicular microfluidic chips with 50 parallel channels were designed.The first chip was placed against a fixed tissue slide to introduce oligo-dT-tagged DNA barcodes A 1 -A 50 to enable in situ RT of mRNA.After removal of the first chip, another chip was placed on the same tissue slide to introduce DNA barcodes B 1 -B 50 and then ligate with barcodes A 1 -A 50 in situ, which can form 2D tissue pixels containing unique coordinates barcodes AiBj.Target proteins were co-measured by adding ADTs to the fixed tissue slide before flow barcoding.Afterward, the tissue was digested to recover spatially barcoded cDNAs and for preparation and sequencing.DBiT-seq with 10-μm pixel size for rapid spatial distributions, and cell types identification through integration with scRNA-seq.The study of whole mouse embryos revealed all major tissue types during early organogenesis, and this method can achieve close to single-cell spatial mapping.To further improve the analysis level of protein expression, on this basis, Fan's group further developed spatial co-indexing of transcriptomes and epitopes (spatial-CITE-seq) which used 200-300 panels of ADTs to stain a tissue slide, achieving spatially resolved high-plex spatial protein (189 proteins in mouse tissue types and 273 proteins in human tissues) and whole transcriptome co-profiling. 108A similar method called SM-Omics, is a fully automated and high-throughput platform to achieve spatial transcriptomics and spatial antibodybased multiplex protein detection (Figure 7B). 109After tissue staining and imaging by fluorescently (H&E, IF) or DNA-barcoded antibodies, the slide was loaded into the SM-Omics platform.By combining liquid handler robot devices, capture of mRNAs on the spatial array and prebuild RNA-Seq library preparation.SM-Omics platform demonstrated a high-throughput capability with the ability to perform 64 in situ spatial reactions or up to 96 sequencing-ready libraries in two days, although it is currently limited to frozen tissues.
In addition to multiplexed capabilities, resolving the spatial context of RNAs and proteins in tissues at TA B L E 2 Multimodal spatial transcriptome and proteome methods.subcellular resolution is also a challenge in the field of basic research and clinical applications.Microscopy is a classical strategy that was based on probe hybridization for spatial transcriptome and compatible with fluorescent/DNA conjugated antibodies for proteome readouts, such as seqFISH 110 , MERFISH 111 , DNA-MERFISH 112 , and DNAseqFISH+. 113Recently, a spatial molecular imager (SMI) automated and integrated platform was developed that relies on in situ hybridization (ISH) probes and fluorescent readout reporter probes, allowing highly sensitive spatial in situ profiling of 980 RNAs and 108 proteins in formalin-fixed, paraffin-embedded (FFPE) tissues at single-cell and subcellular resolution (∼50 nm; Figure 8A). 114ISH probe consists of a fragment that captures RNA (target-binding domain) in the tissue and a fragment that identifies the RNA (called readout domain).Meanwhile, The SMI encoded probe utilized for RNA allowed to high-plex protein detection by conjugating the oligonucleotide readout sequences to antibodies via a site-specific linker.For the preparation of tissue samples, exposed RNA by the FISH method was hybridized with ISH probes.After washing the slides, they were placed in a fluidic cell of SMI for cyclic readout with 16 sets of fluorescent reporters (each of them contains four single-color reporter pools which the RNA and protein SMI imaging barcodes over-background pixels were in one of them).

Method
Following each set of reporter incubation, high-resolution Z-stacked images were acquired for downstream analysis.
Similarly, a fluorescence imaging-based spatial omics technology called MOSAICA (Multi Omic Single-scan Assay with Integrated Combinatorial Analysis) was developed for simultaneous co-detection of protein and mRNA in FFPE tissues by integration in situ labeling of mRNA and protein markers with combinatorial fluorescence spectral and lifetime labeling encoded probes. 115xcept for that subcellular resolution at the tissue level, techniques have been developed to interrogate the spatial transcriptome and proteome within subcellular compartments of cells, which not only provides insight into the organization of cellular compartments but allows exploration of cellular function through subcellular localization. 116,117For example, Benhalevy et al. 118 developed a proximity-CLIP method that combines mature compartment-specific protein biotinylation with ultraviolet-medicated crosslinking, followed by in vivo crosslinking to 4-thiouridine labeling RNA-proteins in living cells (Figure 8B).This strategy achieved simultaneous profiling of both RNA and interest proteins in subcellular compartments of HEK293 cells which the localized proteome was determined by MS and transcriptome defined by RNA-seq.

Combined analysis of single-cell and spatial multiomics
Multimodal intersection analysis is a useful method approach that can effectively overcome the limitations of both single-cell omics, which lose spatial information, and spatial omics, which are not at individual cell resolution by linking single-cell data with spatial omics data. 119For example, Liu et al. analyzed over 60,000 cells from four patients with breast cancer and paired metastatic axillary lymph nodes using scRNA-seq and spatial transcriptomics to describe the dynamic metabolic evolvement of early disseminated breast cancer. 120 I G U R E 8 Schematic of spatial multi-omics analysis of proteome and transcriptome at the subcellular level.A Workflow of SMI chemistry platform.Reproduced with permission from He et al. 114 Copyright 2022.Springer Nature.B Schematic of proximity-CLIP for the simultaneous profiling of localized RNA and proteins in subcellular comportments.Reproduced with permission from Benhalevy et al. 118 Copyright 2018, Nature Research Reporting.
F I G U R E 9 Schematic of study design and fidelity of cryopreserved/fresh biopsy for single-cell and spatial multiomics of proteomes-transcriptomes.Reproduced with permission from Mennillo et al. 121 Copyright 2023.bioRxiv.
Inspired by this, Mennillo et al. 121 performed comprehensive single-cell and spatial transcriptomic and proteomic multimodal analysis of peripheral blood and colonic biopsies in healthy controls (HC) and patients with ulcerative colitis (UC; Figure 9).The combination of CITE-seq and CyTOF establishes a surface protein cell atlas for the colon, and spatial atlas of intestinal tissue subsets was described by using multiplex ion beam imaging, co-detection by indexing , and highly multiplexed RNA-ISH on FFPE samples.The single-cell level revealed that mononuclear phagocytes as a primary target of vedolizumab anti-integrin antibody in ulcerative colitis, with associated changes in stromal and epithelial populations.And spatial information results showed increased density and proximity of mononuclear phagocytes and fibroblast subsets in UC biopsies compared with HC, which was inhibition by vedolizumab.Obviously, the combination of single-cell and SM-Omics of transcriptome and proteome approaches could provide deep immunophenotyping and lead to more accurate algorithms.

BIOMEDICAL APPLICATIONS
The fast-growing single-cell and SM-Omics technologies of transcriptome-proteome greatly advance our under-standing and applications of immunology, oncology, and neuroscience.For immunology study, Wu et al. developed a timeresolved assessment of protein secretion from single cells by sequencing methods that enable concurrent measurement of cellular protein secretion, phenotypes and the transcriptome at the single-cell level. 122The method revealed that phenotypic and transcriptional determinants of the secretion of pleiotropic T H 1 cytokines (IFNγ, IL-2, and TNF) in activated T cells, demonstrating that early central memory T cells with CD45RA expression were essential to the production and maintenance of polyfunctional cytokines.For spatial resolution, using the spatial-CITE-seq strategy for immunology study, Fan's group revealed spatially distinct germinal center reaction in tonsil and identified major skin and immune cell types as well as a subset of peripheral helper T cells highly enriched from the COVID-19 mRNA injection site, which may contribute to local immune activation that initiates systemic vaccine response. 108Chetrit et al. 123 developed a Spatial PrOtein and Transcriptome Sequencing platform, by examining two regions enriched in macrophage/myeloid markers (Mac1-and Mac2-enriched), it was shown that M2-macrophages form an immunosuppressive barrier at the tumor border, leading to immune failure and evasion.
For oncology, Darmanis et al. 71 studied the response of early-passage glioblastoma cells (U3035MG cell line) to bone morphogenetic proteins 4 (BMP4).They found significant heterogeneity in mRNA and protein abundance after BMP4 treatment.The results demonstrated that the overall poor correlation between RNA and protein across single cells and protein level was more accurate in analyzing the response to tumor treatment.Moreover, the development of SM-Omics techniques has made it possible to understand the heterogeneity within tumor and stromal regions, and the acquisition of SM-Omics profiles enables the reconstruction of key processes of tumor formation in a holistic manner.For example, digital spatial profiling technology has been developed for the simultaneous profiling of proteins and RNA in FFPE tissues within an ROI, allowing automated profiling of tumor and stroma compartments in the tumor microenvironment and rare localized cell types within different locations in a tissue. 124he analysis of colorectal cancer tissues reveals immune cell-related markers with the highest levels of expression at the invasive margin, including those related to adaptive and innate immune response, whereas the tumor center was associated with tumor markers such as PanCK, β2microglobulin, and Ki-67.Besides, nonsmall cell lung and breast cancer samples were analyzed at the subcellular resolution level and identified 10 unique TME and 100 pairwise ligand-receptor (LR) interactions were identified, and 16 unique LR pairs were significantly enriched at the interface between tumor and T cells in at least one of five tumors. 114or neuroscience research, Reimegård et al. 74 measured mRNA by scRNA-seq data and targeted protein expression in human embryonic stem cells unperturbed or at fixed time points after induction of directed neuronal differentiation.The results indicated that an overall poor correlation of mRNA and protein which mRNA expression failed to accurately reflect protein abundance, and protein levels of transcription factors could be better predicting their downstream effects.Kaufmann et al. 125 utilized human progressive multiple sclerosis brain tissue for spatial transcriptomic and proteomic analysis.The results revealed the multicellular mechanisms underlying the pathogenesis.They found that local failure of trophic and anti-inflammatory cellular communication in the early stages of neurodegeneration as well as unclear determination of the role of proinflammatory factors in disease pathogenesis.

CONCLUSION AND PERSPECTIVES
In this review, we have discussed multimodal single-cell transcriptome and proteome analysis including cell isolation (both low-and high-throughput), various lysis strategies, RNA and protein molecular separation methods, and co-measurement advanced techniques.Compared with mono-omics or genome-focused multiomics, simultaneous analysis of transcriptome and proteome in the same cell can enhance comprehensive characterization of cellular activities and functions.To retain spatial information, we then discussed the spatial resolution multiomics level of transcriptome and proteome breakthrough techniques, even the possibility in combination with scMulti-Omics analysis that can create true cell and tissue atlases.Nevertheless, there are still limitations and challenges that need to be overcome for further research.In terms of scMulti-Omics, first, moving past the target-based strategy will be a challenging but essential frontier for single-cell proteome unbiased analysis.Although MS-based techniques provide the unique advantage of deep protein analysis, it still needs to consider how to improve the measurement throughput and extent of coverage, which is one of the important research directions in single-cell proteomics.The combination of microfluidic chips, automation equipment, and Tandem Mass Tags (label peptides from different samples) is a strategy worth considering and has been initially validated. 126Multiplexed DIA is a recent development approach to achieve higher throughput in single-cell proteome analysis without the need for isobaric chemical tags. 127Second, modified proteoforms have been shown to reveal links to certain diseases and their underlying mechanisms, but the popular bottom-up MSbased method cannot ensure that specific posttranslational modifications of the target protein can be fully identified, which will greatly reduce the chance of posttranslationally modified peptides being identified. 128,129Third, coupling multiple modalities of high-content data into a single cell provides tremendous value while also presenting sophisticated statistical and computational challenges.There are benchmarking frameworks and evaluations of various computational methods for single-cell mono-omics 130,131 ; however, no methods currently are available for scMulti-Omics.In addition to the analytical challenges, the bottleneck of single-cell isolation and lysis techniques needs to be considered.Both single-cell isolation and lysis strategies have their distinct advantages and disadvantages.Especially in lysis selection, one key question is how to utilize MS-compatible reagents to eliminate the sample purification step and balance high-efficient lysis capacity.
In terms of SM-Omics, currently reported spatial transcriptomic and proteomics technologies in tissue samples hardly provide true single-cell/subcellular level resolution while still ensuring high-throughput and multiplexing capacity.Linking single-cell analysis with SM-Omics data may be a way to solve this problem until higher-resolution methods are widely used.We have reason to believe the technology of high-throughput spatial molecular profiling with high resolution will be prevalent in the near future.As for tissue sample preparation, FFPE tissue is the most common technique for clinical tissue preservation currently due to its ability to be stored for decades at room temperature.However, protein profiling technologies rely on FFPE-compatible antibodies when analyzing tissue samples, which can be difficult to generate for some targets.Additionally, protein extraction from FFPE tissues is challenging due to the cross-linked reaction between proteins and formaldehyde.It is crucial to achieve more efficient protein extraction in FFPE as well as compatible MS analysis.SM-Omics processing of samples should support both fresh/frozen tissues and FFPE to extend its applicability.This could be an important topic in fields like clinical applications.
Despite still in its infancy, single-cell and SM-Omics analysis of transcriptome and proteome have blossomed in the past few years and continue to attract attention.We see the future of the combined application of singlecell and spatial resolution levels that can move biomedical analysis forward.Furthermore, given that biological functions arise from the interaction of transcripts, proteins and metabolites, future advanced techniques can be developed to enable triple-omics research or more within singlecell and spatially high-resolution tissue samples, revealing entirely new domains of biology.

C O N F L I C T O F I N T E R E S T S TAT E M E N T
The authors declare no conflict of interest.

F I G U R E 1
Overview of simultaneous transcriptome and proteome analysis in multimodal single-cell and spatial resolution level.A Schematic representation of single-cell multiomics analysis composed of single-cell isolation and lysis, and its spatially resolved multiomics integration.B Classification of various strategies for simultaneous analysis of multimodal single-cell.

F I G U R E 3
Overview of fluorescence-labeled antibody strategies for single-cell transcriptome and proteome analysis.A Schematic diagrams of simultaneous quantification of mRNA and protein for single-cell level analysis by microwell immunostaining.Reproduced with permission from Park et al.73Copyright 2016 The Royal Society of Chemistry.B Schematic diagrams of detecting single-cell secreted proteins and transcriptome by the splittable microchip.Reproduced with permission from George and Wang et al.85Copyright 2016 American Chemical Society.C Schematic diagrams of INs-seq.Reproduced with permission from Katzenelenbogen et al.46Copyright 2020, Elsevier.

F I G U R E 6
Overview of nontargeted strategy for simultaneous analysis of unbiased MS-based proteomes and whole-transcriptomes in the same single cell.A Schematic diagram of the scSTAP platform and workflow for scMulti-Omics analysis.Reproduced with permission from Jiang et al.61Copyright 2022.bioRxiv.B Schematic diagram of the nanoSPLITS-based scMulti-Omics platform.Reproduced with permission from Fulcher et al.66Copyright 2022.bioRxiv.

7
Schematic of spatial proteome and whole-transcriptome multiomics analysis.A Schematic workflow of DBiT-seq for spatial multiomics analysis.Reproduced with permission from Liu et al.107Copyright 2020, Elsevier.B Overview of SM-Omics approach.Reproduced with permission from Vickovic et al.109Copyright 2022.Springer Nature.
This work was supported by the National Natural Science Foundation of China (no.81827901), Central Funds Guiding the Local Science and Technology Development of Shenzhen (2021Szvup024), and Jiangsu Provincial Key Research and Development Program (BE2021664).
Figure 1 was created with BioRender.com. 11 Summary of multimodal single-cell transcriptome and proteome analysis tools.
TA B L E 1