A major task of functional genomics is to study cell function at the level ofmRNA and protein expression.Routine approaches of identification and quantification include DNA microarrays, expressed tag sequencing (EST), serial analysis of gene expression (SAGE), subtractive cloning and differential display (DD) for mRNA, and two-dimensional gel electrophoresis, mass spectrometry and protein microarray based antibody-binding for protein. Traditionally, each approach requires relatively large numbers of cells. These traditional methods have been extensively utilized to study parallel gene expression in different cell lines. It is known that cell lines used as models in some fields, such as tumorigenesis, have limitations. For example, after certain tumour cell lines go through several hundred passages, many properties of tumour cells have changed; the genome expression from the cell line does not accurately reflect properties of in vivo tumour cells. However, some primary tumour cells from tumour tissue display the intrinsic function and properties of tumour cells. In addition, tumour cell formation and development involves the accumulation of multiple-gene mutations as a tumour grows from a single cell or a very small number of cells (clonality). If the single cell or a small number of cells from primary tumour cells can be employed for genome analysis, it can address questions such as how tumour cells form, how tumour tissue develops, and how some agents can block tumour formation and development. Based on the requirement for analyzing tumorigenesis, some strategies have been developed to study functional genomics at the single-cell level. In order to clearly introduce functional genomic analysis in the single cell, we will briefly review traditional functional genomic methods and then discuss genomic analysis at the unicell level.

Traditional techniques used to measure gene expression are directly related to the quantitative detection of mRNA and protein among parallel samples. Relating these techniques to the central dogma can help us to categorize these methods. Briefly, they can be divided into four fields: (1) DNA level such as genomic sequencing and single nucleotide polymorphism (SNP); (2) mRNA level including microarray, expressed tag sequencing (EST), serial analysis of gene expression (SAGE), subtractive cloning and differential display (DD); (3) protein level, for example, two-dimensional gel electrophoresis, mass spectrometry, and protein arrays based on antibody binding; and (4) post-translational level such as protein-protein interaction via the yeast two-hybrid or repressor system.

Specimens of animal and human tissue often contain multiple cell types with different gene expression profiles. Theoretically and practically, potentially important findings in the gene expression profiles in the multiple cell types will be obscured or unclear. Therefore, studies of representative single cells will provide the most precise analysis possible of these subtle gene expression patterns. Here, in order to discuss functional genomics in single cells clearly, the following three fields: (1) single-cell sampling; (2) mRNA and protein isolation or amplification from a single cell; and (3) application of genomic expression of single cells, will be systematically reviewed.

Single-cell sampling can be divided into flow-cytometric cell sorting and laser-based microdissection of tissues. In flow cytometry, cells in solution are labelled with fluorescent signals.  These signals can be derived from a specific biomarker such as a tumour antigen attached to an antibody that is labeled with a fluorescent signal or a recombinant DNA construct encoding modified proteins with a fluorescent signal. At present, multi-coloured fluorescence-activated cell sorters can selectively separate and collect homogeneous cells with identical phenotypic features in a collection tube in order to increase sensitivity for gene expression profile in a given cell type. Although flow-cytometric cell sorting and multi-coloured fluorescence-activated cell sorters can isolate and sort homogeneous cells, there are three limitations to using these techniques: (1) some cell types such as neurons are not amenable to separation and sorting by flow-cytometry; (2) internal cell localization of sub-cellular components cannot be well-defined using flow cytometry; and (3) the microenvironment of a cell (such as a good blood supply or a bad blood supply) cannot be evaluated and studied using flow cytometry.  The microdissection technique, in part, avoids these problems. At present, three microdissection methods have been developed: (1) laser-assisted mechanical tissue microdissection, (2) laser pressure catapult microdissection and (3) laser capture microdissection. Laser-assisted mechanical tissue microdissection can focus on small target cell areas, reducing the chance of contamination with neighbouring cells.

The quantity of mRNA in a single cell is approximately 1.0 pg (about 5x10⁵ molecules). Although some scientists try to isolate RNA from single cells, we prefer to use a crude cell lysate without purifying procedures. This protocol has two important advantages. First, it ruptures the cells and releases the RNA directly into a cell lysis buffer without loss of RNA. Moreover, the heating step to rupture cells inactivates endogenous RNase, further protecting RNA from degradation. Theoretically, in order to observe subtle differences in parallel gene expression, genome information amplification should be applied in single-cell studies. At present, there are two strategies to employ genome information amplification: mRNA amplification (aRNA) and PCR-based cDNA amplification. We have also developed a more facile strategy to screen the genome at the single cell level. To illustrate, three techniques (RNA directly from cell lysis, randomized primer design as differential display and single cell genome cloned into plasmids) are simultaneously combined. Following cell lysis and reverse transcription PCR, a 3’ end oligo (dT)_n primer and a set of 5’ end arbitrary primers (both containing restriction enzyme terminals) are used in an amplification by PCR. After double digestion, the genome from a small number of cells is introduced into plasmids and transformed into cells. As Figure 2 and Figure 3 indicate, subtractive hybridization from a reference cell genome is employed in the modified method so that artifacts of cDNA amplification from test cells are minimized. The technique has been used in genome expression analysis from 10 to 100 cells. The advantage is that the expression results are very sensitive and accurate because they exclude problems of artifacts. The disadvantage is similar to TPEA, that is, expression has a 3’ bias. Over the past two years, as aRNA techniques have been developed, cDNA detection sensitivity has significantly increased so that we can use either cells from microdissection or single cells obtained from culture to analyze their genome expression.

Traditional methodologies for protein detection and quantificationinclude two-dimensional gel electrophoresis, mass spectrometry,and antibody (Ab) binding. As we discussed previously, the application of these traditional proteomics-oriented technologies at the single-celllevel has been limited because each methodology needs relatively large amounts of tissue.

Single-cell gene expression analysis can be carried out both at the specific profile and global genome profile. In situ hybridization and rtPCR belong to the specific profile. In situ PCR combined with immunohistochemical detection is frequently used as a measurement of single-cell gene activity. Multiplex rtPCR is also effective for observing gene expression at the single-cell level. At present, rtPCR using real-time detection of PCR products can quantify gene expression at the single-cell level with reduced risk for artefacts resulting from contamination or illegitimate transcript amplification. However, because the primers are pre-selected, expression profiles will not contain previously unreported transcripts or novel sequences. 

Global genome profile expression analysis at the single-cell level holds new promise to analyze disease pathogenesis and tumorigenesis. At present, four techniques are utilized to advance the global genome profile of a single cell (in addition to the previously described specific profiles such as in situ hybridization and multiplex rtPCR). These global genome profiles include differential display, subtractive cloning, microarray, and protein array. Differential display and subtractive cloning can be employed with a small number of cells in which the resolution is from one cell to 10⁴ cells. Because both of these methods may have an artefact contamination after amplification, it may result in variable genome expression at the single-cell level. As discussed above, we introduced a strategy combining amplifying RNA, randomized primers (with restriction terminals for cloning into plasmid) and subtractive hybridization (for eliminating some artefacts), which has successfully been used in genome expression at single-cell level. DNA microarrays used for single-cell DNA have emerged. One method involves modifying some procedures to increase resolution such as aRNA and cDNA amplification, and another involves remodelling microarray platform materials. In remodelling microarray platform materials, recently, a high-density fibre optic DNA microarray has been developed, in which there are 6,000 to 50,000 fused optical fibres, and each fibre terminates with an etched well.

Protein arrays based on antibody-binding technology shall emerge by development of immuno-PCR and T7 RNA amplification to screen protein expression at the single cell level. If intact Abs, ScFv fragments or exocyclic peptide-based complementarity’s determiningregion (CDR) subunits can serve as antigen detectors in the protein array, it can likely facilitate the development of a robotic platform for proteomics. If a large scale of antibodies can be produced, the usage of this approach will have a good potential.

Gene expression profiles of single cells are providing tremendous insights into disease pathogenesis, especially in tumorigenesis. The progress is very slow due to current technological limitations. Most scientists screen global profiles, as discussed above, by using microarrays from a large number of cells and then confirming them by analyzing the specific profile at the single-cell level by using rtPCR or differential display.  In 2000, we reported a strategy to observe the global profile of a small number of T-cells (10 to 100 cells). Recently, some investigators have suggested using tiny tissue samples from laser-captured microdissection to analyze genomic profiles, especially in complex multifactorial diseases such as neuropsychiatric disorders. According to the idea of micro-dissection to obtain the tiny tissue or single cell, a genomic change of aberrant crypt foci have been successfully observed in our laboratory.

Although technical developments and clinical application in single-cell gene expression have been established and developed, the techniques and applications still need to be optimized. A mature genome expression analysis at the single-cell level needs the following: (1) a good method for isolating pure homogenous cells; (2) intact bio-molecule harvest for mRNA and proteins along with a high-fidelity amplification system; and (3) a sensitive method to detect the biomarker at the single-cell level. As we indicated above, no single method can currently satisfy all of these requirements. For example, in situ hybridization and multiplex rtPCR are limited to the analysis of a known profile. Also, DNA microarrays require the development of more sensitive platform materials. In other words, after developing immuno-PCR and T7 RNA amplification, it is possible to screen protein expression at single cell levels, but production of many thousands of intact protein Abs and peptide Abs still face great challenges. We have designed randomized primers and combined strategies including amplifying RNA, cloning into plasmids and subtractive hybridization to minimize some artefacts, and this strategy has been successfully applied from several hundred cells to a single-cell level. However, the problem with 3’ bias still needs to be resolved.

Fortunately, many scientists and companies are focused on overcoming these challenges. It is believed that single-cell global genome profiling will become an important tool for scientists and physicians to study pathogenesis, early clinical diagnosis and treatment in the near future.