INTRODUCTION

Protein-coding eukaryotic genes possess exon–intron structure, and their primary transcripts undergo splicing—excision of introns from pre-mRNA to form mature mRNA consisting of exons only. Importantly, most genes of higher eukaryotes produce multiple isoforms of mature mRNA because of so-called alternative splicing, the process by which exons can be spliced in different arrangements (1). Alternative splicing provides a way to increase proteome complexity and to regulate gene expression in a tissue-, stage- or stimuli-dependent manner (2).

Regulation of alternative splicing is a complex process, which is still not completely understood. Short cis-acting sequence elements referred to as exonic splicing enhancers (ESEs) and exonic splicing silencers (ESSs) or intronic splicing enhancers and silencers located within alternative exons and nearby introns, as well as adjacent constitutive exons, were found to play important roles in the alternative splicing regulation (1,3). Also, other structural features, such as RNA secondary structure and length of the exons, can be involved in determination of the alternative splicing patterns.

Classical methods to study alternative splicing include northern blot analysis and reverse transcriptase–polymerase chain reaction (RT–PCR). Modern high-throughput approaches, such as hybridization to oligonucleotide microarrays and large-scale sequencing of transcriptomes in combination with extensive bioinformatic examination, enable global analyses of alternative splicing (1,3).

Last decade, several approaches to analyze alternative splicing using green fluorescent protein (GFP) and other fluorescent proteins, e.g., red fluorescent protein (RFP), were developed (4–9). These methods enable visualization of results of alternative splicing in live cells and organisms at single-cell resolution. However, none of them gives the possibility of calculating relative amounts of mRNA species by measuring single cell fluorescence intensities (see ‘Results and Discussion’ section below).

Here, we suggest a new approach to analyze alternative splicing of a target cassette exon using two fluorescent proteins of different colors. A percentage of the full length and alternative (exon-skipping) forms can be estimated from the ratio of the fluorescent signals at single cell level.

MATERIALS AND METHODS

RESULTS AND DISCUSSION

The proposed method is aimed to analyze alternative splicing of cassette exons, skipping or including of which results in an open reading frame shift (i.e. exon length is not a multiple of 3). Two fluorescent proteins of clearly different colors are used for quantitative analysis at single live cell level. A fragment of target gene (minigene) consisting of an alternative exon (AE_n) of interest flanked by adjacent upstream and downstream introns (I) and constitutive exons (E) is cloned between RFP- and GFP-encoding sequences in the following order: rfp-E_(n−1)-I_(n−1)-AE_n-I_n-E_(n+1)-gfp (Figure 1A) so that the ORF is conserved from RFP to GFP. Thus, translation of the normally spliced full-length transcript results in a RFP-tORF-GFP fusion protein. In contrast, alternative exon skipping results in a frame shift within exon E_(n+1) and thus leads to a truncated protein RFP-tORF without GFP.

Cells expressing this construct develop different levels of red and green fluorescence (which can be quantified using fluorescence microscopy or flow cytometry), and the ratio of red to green can be used to estimate a proportion of normal and alternative transcripts. To this end, a control construct encoding RFP-tORF-GFP without introns should be used. Expression of this control provides a reference ratio between the red and the green signals observed in a particular model and detection settings when only normal transcript is present. In the experimental construct undergoing alternative splicing, additional red signal, which is above the estimated control red/green ratio, corresponds to the alternative transcript.

Experimental results can be represented as a plot where each cell gives a dot of particular intensities in green (abscise axe) and red (ordinate axe) channels (Figure 1B). On this plot, the control cells would form a diagonal cluster, whereas experimental cells would lie either on or above this diagonal. Portions of normal and alternative transcripts in each cell can be estimated using a simple calculation using its position relative to the control diagonal (Figure 1B).

To test this new method, we analyzed alternative splicing of human PIG3 (p53-inducible gene 3) gene. This gene represents an interesting example of a regulatable alternative splicing. Expression of PIG3 produces two splice variants: major full-length transcript consisting of all 5 exons and minor short transcript lacking exon 4 of 197-bp length (10). Absence of exon 4 induces a frame shift in exon 5.

In this work, we used far-red fluorescent proteins Katushka and GFP TagGFP. These fluorescent proteins provide bright and spectrally non-overlapping signals. In the previous work, we showed that the coding region of Katushka contains a strong donor splice site, which can interfere with target gene splicing (11). Therefore, we used a mutated Katushka gene with altered codon usage that does not possess splice sites and, thus, can be used to analyze splicing events (11). We cloned PIG3 minigene (exons 3–5 and corresponding introns) between Katushka and TagGFP-coding sequences (plasmid pSplPIG) in accordance to the scheme depicted in Figure 1A. Also, a control plasmid pCtrlPIG corresponding to the normally spliced transcript (Katushka-exons3–5-TagGFP) was constructed. HEK293T cells were transfected with these plasmids and analyzed using fluorescence microscopy and flow cytometry. Cells expressing pCtrlPIG demonstrated bright green and red fluorescence; the ratio between red and green signals did not vary much from cell to cell. Therefore, overlay of the images in green and red channels gave mainly yellow cells (Figure 2A), and flow cytometry showed that these cells form a quite narrow diagonal on the red–green bivariate plot (Figure 3A).

In contrast, transfection with pSplPIG revealed unexpected cell-to-cell variation in a red/green ratio. Side-by-side analysis of pSplPIG- and pCtrlPIG-expressing cells under the same detection settings showed that some pSplPIG cells were similar to the control cells, some others were almost purely red, and majority of pSplPIG cells gave various relative intensities of red and green signals (with a clear prevalence of red over green fluorescence compared with pCtrlPIG cells) (Figure 2B). We concluded that levels of PIG3 minigene alternative splicing in individual HEK293T cells can vary significantly—from nearly 100% normal transcript (‘yellow’ cells) to nearly 100% alternative transcript (‘red’ cells), but most cells produce both transcripts in various proportions (‘orange’ cells).

Quantitative analysis of flow cytometry data confirmed a broad distribution of the proportion between normal and alternative transcripts in pSplPIG cells (Figure 3B). Two clear peaks at ∼45% and >95% of alternative transcript can be recognized. Thus, at least two distinct populations of cells with different levels of alternative splicing of the target gene exist in HEK293T. A mean percentage of pSplPIG alternative form was estimated to be 50%. To check this value by an independent method, we isolated RNA from HEK293T transfected with pSplPIG, generated cDNA, and performed PCR with primers corresponding to Katushka and TagGFP sequences. As expected, this amplification resulted in two bands corresponding to the normally and alternatively spliced transcripts (Figure 3). Densitometry analysis showed lower band (alternative transcript) to constitute 52% of the total, which is in a good agreement with flow cytometry data.

The production of the PIG3 alternative transcript was found to be enhanced by irradiation of cells with UV light (10). Thus, we tested the effect of UV illumination on pSplPIG splicing. In line with literature data (10), a considerable increase in the percentage of the alternative transcript was observed (Figures 2C and 3C).

Earlier mutagenesis studies on PIG3 minigene revealed regions within exon 4 acting as strong ESE or ESS (10). We generated two mutants, pSplPIG-mutE and pSplPIG-mutH, corresponding (with some modifications, see Materials and Methods for details) to two extreme mutants dubbed e and h in the original publication (10), which produce nearly exclusively alternative and normal transcripts, respectively. pSplPIG-mutE-expressing cells developed only red fluorescence, which corresponds to the presence of alternative and absence of the normal transcripts (Figures 2D and 3D). In contrast, transfection with pSplPIG-mutH resulted in ‘yellow’ cells, which practically overlap with the control pCtrlPIG-expressing cells (purely normal splicing) on the bivariate plot (Figures 2E and 3E). Thus, both mutants showed expected behavior.

We also generated a mutant pSplPIG-mutCU with altered codon usage, which contains no amino acid substitutions but many silent mutations within exon 4 nucleotides 86–133 [designated h–k in (10)]. This region was shown to be ESS (10), and we thus aimed to generate PIG3 variant with no exon 4 skipping. Unexpectedly, pSplPIG-mutCU demonstrated a considerable enhancement of exon 4 skipping (Figures 2F and 3F). Interestingly, the splicing pattern of this mutant was very similar to that observed for UV-treated pSplPIG cells. We believe that pSplPIG-mutCU can potentially be used as a model of UV-induced changes of PIG3 alternative splicing (but without UV irradiation) to study biochemical consequences of this phenomenon.

In conclusion, our method provided an adequate evaluation of PIG3 minigene splicing including expected changes in response to UV and introduced mutations. Moreover, analysis at single-cell level made it possible to reveal a strong heterogeneity of cells with respect to PIG3 alternative splicing. The reasons for this heterogeneity are unclear and call for further investigation. Generally, heterogeneity within seemingly uniform cell populations recently became the focus of extensive research (12–14). It can be revealed by cell staining for several independent markers (15,16). Heterogeneity of alternative splicing can add a new dimension to these studies.

We believe that the new method is potentially useful for the analysis of alternative splicing of target genes in a variety of biological models—cell lines and whole organisms—where fluorescence microscopy and/or flow cytometry can be used to detect fluorescence. The method possesses common drawbacks of GFP-based techniques such as the necessity to (over)express an exogenous gene (that can affect cell physiology including splicing), and potentially altered stability of chimeric mRNA with and without alternative exon. However, worldwide practice of applications of fluorescent proteins shows that it can be used safely in most biological models. Advantages of the method proposed can be summarized as follows:

1. A new scheme of arrangement of two fluorescent proteins is applied. The main novelty of the scheme is that RFP is encoded in both normal and alternative transcripts. This design allows quantitative analysis of alternative splicing in individual cells by decomposing red signal into two portions corresponding to alternative (RFP) and normal (RFP+GFP) transcripts using a simple control sample (RFP+GFP). In contrast, previously published methods do not allow calculating the percentage of transcripts from fluorescence intensities even when two fluorescent proteins are used (5,6,8,9). A quantitative analysis of a cassette exon skipping within single cells using EGFP and DsRed was claimed by Orengo et al. (7). However, this method is based on the comparison of fluorescence intensity of a cell in green channel with that of the same cell in red channel as assessed by fluorescence microscopy or flow cytometry. In fact, such direct comparison of the green fluorescence signal to the red one to estimate molar ratio between EGFP and DsRed (or any other fluorophores) seems to be incorrect for many reasons including the following: (a) different fluorophores possess different intrinsic (per molecule) brightness, (b) excitation filters (lasers) and emission filters fit the excitation–emission spectra of the fluorophores differently, (c) intensity of excitation light (e.g. from a mercury lamp) can differ drastically in different spectral regions and (d) the detector can possess different sensitivity in different spectral regions.

2. A large fragment of the unaltered target gene including an alternative exon with flanked introns and constitutive exons is taken for analysis. This ensures inclusion of nearly all native regulatory elements, which are usually concentrated not only within alternative exons and flanked introns but also can be found in adjacent constitutive exons (3). At the same time, expression cassettes in some existing methods of fluorescent protein-based analysis of alternative splicing do not include adjacent constitutive exon(s) or contain fluorescent protein-coding inserts into alternative exons (5,9). In some cases, this can affect the regulation of alternative splicing.

3. In addition to cassette exons, our method can be potentially adapted for other alternative splicing types such as alternative 5′- or 3′- splice sites, intron retention and mutually exclusive exons. The only requirement is a frame shift in the alternative transcript relative to the normal one. The frame shift was found to occur in more than one-third of all alternative splicing events in human and mouse (1,17). Thus, the method suggested is applicable for a large number of genes. For other genes, artificial alteration of the size of alternative regions by 1–2 nucleotides can be used to induce the required frame shift (6,7).

FUNDING

Molecular and Cell Biology program of the Russian Academy of Sciences; Russian Foundation for Basic Research (10-04-01042-a and 11-04-12187-ofi-m); and Ministry of Education and Science of the Russian Federation (16.740.11.0367 and 16.512.11.2139). Funding for open access charge: Molecular and Cell Biology program of the Russian Academy of Sciences.

Conflict of interest statement. None declared.