Changes in gene expression are intrinsic to how cells respond to various enviornmental stimuli. An increased awareness of temporal and spatial contributions to these mechanisms has generated an interest in knowing how gene expression changes in individual living cells as they respond to stimuli in real time. Green fluorescent protein (GFP) from the jellyfish Aequorea victoria would, in principle, be ideal as a reporter gene to facilitate such studies. GFP is readily expressed in a wide range of organisms, is non-disruptive to their growth, and its detection is independent of any added external substrate. However, a major drawback to the use of GFP is that it is very stable (>420 min in yeast) and, therefore, is not optimal for applications involving the kinetics and patterning of gene expression (Natarajan et al., 1998).
To make GFP more suitable as a reporter of transient gene expression, destabilized versions with half-lives as short as 30 min have been generated, using strategies based on fusions to destabilization domains or to peptides that direct protein destruction by C-tail-specific proteases (Andersen et al., 1998; Li et al., 1998; Mateus and Avery, 2000). GFP derivatives with a broader range of stabilities would be useful for studies of transient gene expression. For example, different applications are likely to have different temporal constraints and demands on the amount of reporter required for detection. Flexibility in fluorescent protein reporter stabilites could be realized by exploiting the N-end rule pathway for programmable protein degradation. This pathway targets proteins to the proteasome for their destruction with efficiencies that are determined by the identity of a bipartite ‘N-degron’ signal (Bachmair et al., 1986; Bachmair and Varshavsky, 1989; Gonda et al., 1989; Park et al., 1992; Suzuki and Varshavsky, 1999). Because the N-end rule pathway is evolutionarily conserved, the utility of this strategy for generating families of short-lived fluorescent reporter proteins should be generally applicable (Gonda et al., 1989).
Studies by Varshavsky and colleagues using test proteins expressed in S. cervisiae were the first to reveal that the N-terminal residue of a protein is either stabilizing or destabilizing (Bachmair et al., 1986; Bachmair and Varshavsky, 1989). These studies utilized ubiquitin-β-galactosidase fusion proteins with different amino acids engineered at the fusion junction (Ubi-X-βgal) (Figure 1A). Different residues are exposed at the amino terminus of otherwise identical β-gal proteins because the ubiquitin is cleaved off the nascent fusion protein by de-ubiquitinating enzymes (DUB) during translation (Figure 1A). This cleavage takes place regardless of the amino acid residue at the junction. Characterization of the different X-βgal test proteins identified stabilizing N-terminal residues (Met, Gly, Val, Pro, Cys, Ala, Ser, Thr) that confer a long half-life (>20 h) on the test protein and destabilizing residues (Glu, Gln, Asp, Asn, Ile, Leu, Phe, Trp, Tyr, His, Lys, Arg) that confer half-lives of 2–30 min (Figure 1C; for review, see Varshavsky, 1996).
Subsequent studies with N-end rule test substrates revealed that the signal for degradation is bipartite. One determinant is the destabilizing N-terminal amino acid that mediates binding to the N-end rule pathway ubiquitin ligase, Ubr1. The second determinant is the presence of one or more lysine residues that are sterically positioned for conjugation to ubiquitin. The number, positioning and context of these lysine residues directly influences the efficiency of ubiquitination and, consequently, degradation of the protein by the 26S proteasome (Bachmair and Varshavsky, 1989; Suzuki and Varshavsky, 1999).
A 40 amino acid sequence with a central KRK motif from the Escherichia coli lac repressor serves as an efficient N-degron signal (Figure 1B). This particular sequence was identified because it is fortuitously at the N-terminus of the original β-galactosidase test protein. The investigators who discovered its function called this extension (e) bearing lysines (K) an ‘ek’ element (Bachmair and Varshavsky, 1989; Suzuki and Varshavsky, 1999). The ek element has been used as a portable N-degron signal to destabilize unrelated proteins, such as DHFR and Ard1, by the ubiquitin fusion strategy (Bachmair and Varshavsky, 1989; Park et al., 1992). Mutational variants of the ek sequence have been characterized and shown to either increase or decrease efficiency of degradation (Bachmair and Varshavsky, 1989; Suzuki and Varshavsky, 1999). For example, more efficient N-degrons result from the addition of extra KRK motifs to the ek sequence. Less efficient N-degron signals result from mutation of the ek lysine 17 to arginine (ek-R17) or deletion of ek residues C-terminal to threonine 24 (Δk) (Figure 1B). Such manipulation of the N-degron signal sequence allows for fine tuning of decay rates and expands the range of destabilization that is theoretically feasible for any given protein.
We applied the ubiquitin fusion strategy for the N-end rule degradation to generate a family of destabilized cyan fluorescent proteins (CFP) to be used as transcriptional reporters in living cells. We chose the CFP colour variant to be compatible with a constitutive internal yellow fluorescence protein (YFP) reference reporter, because this colour pair has no emission overlap and is therefore suitable for dual-colour imaging. CFP reporter expression from the carbon source-regulated GAL1 promoter was used to determine the relative half-lives of different N-degron variants and the speed with which they can report changes in transcription. The proteins were also expressed under control of the pheromone-induced FUS1 promoter to assess their suitability as reporters of transient transcription.
Materials and methods
Recombinant DNA procedures and plasmid constructions
Bacterial transformations, bacterial DNA preparation, plasmid constructions, and DNA restriction enzyme digestions were performed by standard methods (Sambrook et al., 1989). The tail-to-tail PCR method was used to introduce various base-pair substitutions, insertions or deletions, using template DNA and primers with PFU Turbo under conditions specified by the vendor (Stratagene) (Ling and Robinson, 1997).
Plasmids with the different CFP reporter alleles were assembled by subcloning fragments from different UAS cassette plasmids into CFP cassette plasmids (Figure 2A–D). All cassettes and assembled reporter genes are in the multiple cloning site of pUC118 (Yanisch-Perron et al., 1985). The salient features of the fluorescent reporter alleles and the plasmids from which they were assembled are outlined here. The details for their construction are provided in Supplementary methods. Plasmid maps and assembled DNA sequences for these plasmids can also be accessed from our laboratory website.
UAS cassette plasmids
The FUS1 UAS plasmids with (pNC820, pNC824, pNC929) and without (pNC987) the downstream UBI4 coding sequence were derived from pGA1815 and pGA1706, respectively (Figure 2A; Lydall et al., 1991; Rhodes et al., 1990). The GAL1 UAS plasmids with (pNC825, pNC840 and pNC866) and without (pNC989) the downstream UBI4 coding sequence have a 667 bp fragment with the GAL1,10 promoter derived from pBM150 that replaces the XhoI–BglII fragment from the corresponding FUS1UAS plasmid (Figure 2A) (Johnston and Davis, 1984). The cassettes with the UBI4 coding sequence were engineered to have an ATG, GAA or TAT codon immediately following the C-terminal codon for ubiquitin. UAS cassettes without the UBI4 coding sequence have an ATG codon at the equivalent position with respect to the BamHI cleavage site. The respective ATG, GAA or TAT codons specify the N-terminal methionine, glutamate or tyrosine that is exposed in the resulting fluorescent reporter protein.
CFP cassette plasmids
Plasmids with the cyan-fluorescent variant of GFP (CFP) were derived from pDH3 (provided to us by Trisha Davis, University of Washington, Seattle, WA, and The Yeast Resource Center) (Yoder et al., 2003). This version of CFP has seven amino acid substitutions (F64L, S65T, Y66W, N146I, M153T, V163A, N164H) compared with the GFP sequence from A. victoria. The CFP coding sequence of pNC829 was manipulated so that the BamHI site maintains the same reading frame as that for the ubiquitin (or ATG) of the UAS cassettes (Figure 2A, B). The cassettes with an in frame ek (pNC860) or Δk (pNC1053) extension to CFP were constructed by inserting a BamHI–BglII fragment encoding the specified linker at the BamHI site of pNC829 (Figure 2B). The CFP cassettes also carry the HIS3MX6 selectable marker from S. kluyveri that complements his3 mutations of S. cerevisiae (Wach et al., 1997). To enable targeted integration specifically to the SmaI site that is 3′ to the URA3 coding region, we flanked these reporter alleles with sequences derived from the URA3–TIM9 locus (Figure 2B). The 5′ targeting sequence is a 410 bp SacI–KpnI fragment generated by PCR amplification of the locus, beginning at URA3 codon 161 and terminating at the SmaI site, that is 76 nucleotides 3′ to the URA3 stop codon. The 3′ targeting sequence is a 366 bp SmaI–SalI fragment from PCR amplification of the region beginning at the SmaI site 3′ to URA3 and terminating 40 nucleotides 3′ to the TIM9 stop codon (the SacI, KpnI and SalI recognition sites in the two fragments are derived from the primers used for their amplification).
Fluorescent reporter genes
PGAL1 and PFUS1 reporter plasmids with ubiquitin fusions to the fluorescent protein (pNC843, pNC844, pNC952, pNC953, pNC951, pNC862, pNC863, pNC861, pNC948 and pNC955) carry the KpnI–BamHI fragment from one of the UAS cassettes (pNC840, pNC866, pNC825, pNC820 or pNC824) inserted at the KpnI–BamHI site of the CFP, ekCFP or ΔkCFP cassette (pNC829, pNC860 and pNC1053, respectively). PGAL1 reporter plasmids without fusion to UBI4 (pNC887, pNC960 and pNC864) were also made to express reference proteins that do not require processing of the nascent polypeptides. These have the KpnI–BamHI fragment from pNC989 inserted at the BamHI–KpnI site of the CFP, ekCFP or ΔkCFP cassette (pNC829, pNC860 and pNC1053, respectively).
A constitutive reference reporter gene to be used as an internal reference for normalization of CFP expression levels has the pH-tolerant yellow-fluorescent variant of GFP (YFP) from pDH5 (provided to us by Trisha Davis, University of Washington, Seattle, WA, and The Yeast Resource Center) (Yoder et al., 2003). This YFP allele has five amino acid substitutions (S65G, V68L, Q69K, S72A, T203Y) compared with the GFP sequence from A. victoria. A 444 bp SalI–SmaI ADH1 promoter fragment is located 5′ to the YFP coding region and a 450 bp SacI–EcoRV ADH1 terminator fragment is present 3′ to the kanMX6 selectable marker. These 5′ and 3′ flanking sequences target the reference reporter cassette for replacement of the ADH1 coding sequence at chromosome XV (the gene replacement places YFP expression under control of the endogenous ADH1 promoter).
Yeast genetic procedures, strains and culture conditions
Unless otherwise specified, yeast growth media and genetic manipulations were employed as described by Sherman et al. (1986). Yeast transformations, gene replacements and targeted integrations were done using standard procedures (Gietz et al., 1995; Rothstein, 1983). All gene replacements or integrations were confirmed by PCR analysis of genomic DNA.
Details of strain constructions and a list of all strains used in these studies are provided in the section on Supplementary methods. Strains were derived from one of the three following parental strains: C699-94 (MATaade2-1 bar1Δ::HisG can1-100 his3-11,15 LEU2 trp1-1 ura3-1), which is a Leu+ revertant of strain C699-5 (Esch and Errede, 2002); ∑182 (MATahis3 LEU2 trp1 ura3-52), which is a Leu+ revertant of strain 10560-4A (provided by G. Fink, Whitehead Institute, MIT, Cambridge, MA); or ∑244 (MATα leu2Δ::HisG ura3-52), which is a segregant from a cross between strains SC112 and SKY2610 (Chandarlapaty and Errede, 1998; Palecek et al., 2000). MATa strains with different PGAL1-X-CFP alleles were derived from C699-94 by targeted integration of the SacI–SalI fragment encompassing the alleles of the plasmids shown in Figure 2C (strains C699-95–C699-139; Supplementary Table 3).
From here forward, we use X to denote sequence variations in reporters that include the different codons at the fusion junction as well as the presence or absence of UBI4 and/or the ek or Δk extension to CFP. MATa strains with PFUS1-UBI-M-ΔkCFP and PFUS1-UBI-Y-ΔkCFP reporter alleles were derived from ∑182 by targeted integration of the SacI–SalI fragment encompassing the alleles from plasmids shown in Figure 2D (strains ∑275 and ∑284, respectively; Supplementary Table 3). A MATα strain with the PADH1-YFP reference allele was derived from ∑244 by gene replacement of the ADH1 locus with the SalI–EcoRV fragment from pNC891 (Figure 2E) (strain ∑265; Supplementary Table 3). ∑277 and ∑292 are Ura+ revertants of ∑265 and ∑244, respectively.
MATa haploid strains without the reference PADH1-YFP allele were used for experiments designed to measure CFP expression by Western blot (CFP protein) or Northern blot (CFP mRNA) methods. Prototrophic strains carrying both a CFP reporter allele and the reference PADH1-YFP allele were used for all experiments that measure CFP expression by fluorescence imaging methods. Prototrophic diploid strains for dual-colour imaging (LS24–LS37; Supplementary Table 3) were made by crossing each of the MATa haploid PGAL1-X-CFP strains to the MATαPADH1-YFP reference reporter strain. MATa haploid strains with PADH1-YFP reference and either the PFUS1-UBI-M-ΔkCFP or PFUS1-UBI-Y-ΔkCFP alleles (∑303 and ∑333, Supplementary Table 3) were obtained as segregants from crosses between MATαPADH1-YFP reference reporter strain and ∑275 or ∑284, respectively. Strains ∑303 and ∑333 were made prototrophic by transformation with plasmids that complemented nutritional markers.
Prototrophic strains were grown in synthetic medium without additional supplements. Strains with auxotrophic markers were grown in synthetic medium supplemented with the appropriate nutritional requirements.
The steady state amount and the half-life of different N-degron derivatives of CFP were determined from analysis of the PGAL1-X-CFP reporter genes. Strains were grown in medium with 2% (w/v) galactose as the sole carbon source to a density of 1–2 × 107 cells/ml. Samples were removed for determination of steady-state expression levels of the different variants. Further transcription from the GAL1 UAS was blocked by the addition of dextrose (2% w/v, final concentration) to the remaining culture. Samples were taken for analysis at indicated times after the addition of dextrose to measure the half-life characteristic of each N-degron derivative.
To monitor pheromone induction profiles, cultures of FUS1-UBI-M-ΔkCFP or FUS1-UBI-Y-ΔkCFP strains were grown in medium with 2% dextrose as the carbon source to a density of 5–7 × 106 cells/ml. Samples were removed for determination of basal reporter gene expression. Mating pheromone α-factor (1 µM final concentration) was added to the remaining culture. Samples were removed for analysis at the indicated times after the addition of α-factor to measure reporter gene expression during the induction and attenuation phases of the pheromone response.
Immune blotting conditions
Culture samples (10 mL) were used for preparation of whole-cell protein extracts (Mattison et al., 1999). The protein concentrations of the resulting extracts were determined by the Lowry method, using the BioRad DC Protein Assay Kit under conditions specified by the vendor. 40 µg samples were fractionated by SDS-PAGE on 12% gels and transferred to nitrocellulose filters; the only exception was that 60 µg samples were needed for analyses to determine the half-life of the Y-Δk-CFP protein. The filters were divided into two portions at the 43 kDa molecular size marker. CFP protein was detected on the lower molecular size portion of the filter by using rabbit anti-A.v. peptide polyclonal primary antibody (1 : 500; BD Biosciences) with goat anti-rabbit-IgG–AP conjugate secondary antibody (1 : 20,000; Promega). As an internal loading reference, yeast α-tubulin was detected on the higher molecular size portion of the filter by using rat anti-tubulin primary antibody (1 : 200; Accurate Chemicals) with rabbit anti-rat IgG-AP conjugate secondary antibody (1 : 10,000; Sigma). Immunoreactive species were detected by colourimetric methods according to procedures detailed for the Promega Protoblot system, with the modification that reagents for colourimetric detection of CFP were used at one-half the recommended concentration. Signal intensities for CFP and tubulin were quantified from scanned blots using NIH Image J software.
Fluorescence microscopy and imaging analysis
Cells for microscopic analysis of fluorescent reporters were specifically designed to have no nutritional requirements because we found that growth in synthetic medium without supplements minimized the autofluorescence background. Prototrophy was achieved either by use of diploid strains with complementing nutritional markers or by transformation of haploid strains with plasmids that complemented nutritional markers. Cells were cultured as described above, harvested by centrifugation, and suspended in medium specified for different experiments to a density of 1–2 × 107 cells/ml. The suspensions were sonicated briefly to disrupt clumps. Standard 3 × 1 inch glass slides were prepared for sample application by pipetting 1 ml agar medium onto the surface of the slide. 5 µl cell suspension was applied to the solidified medium and after the cells settled a cover slip was applied.
Microscopy was performed with a Nikon Eclipse E600 FN, using a 100 × 1.4 NA Plan Apochromatic objective. Images were generated using a Hamamatsu OrcaII Progressive Scan Interline-cooled CCD digital camera (Model c4742-98). CFP and YFP fluorescence was detected by employment of the 86002 V2 filter set from Chroma Technology Corp (Rockingham, VT). Exposure times for differential interference contrast (DIC) and fluorescence images were 20 ms and 1000 ms, respectively. Microscope automation as well as image acquisition and analysis were accomplished with the Metamorph 5.0r7 software package (Universal Imaging Corp., Downingtown, PA).
Amounts of CFP are specified as the ratio of the corrected fluorescence intensity of CFP relative to that of YFP expressed from the reference PADH1-YFP allele. Fluorescence intensities of individual cells were measured from captured images using the Metamorph software. The area used for each measurement is an object defined by tracing the circumference of the cell. The raw fluorescence intensity within the object is measured. The background intensity for an identical area is measured by dragging the same object to the background adjacent to the cell. The total fluorescence intensity specific to the cell is obtained after subtracting the background intensity from the raw fluorescence measurement. Total fluorescence intensity is then corrected for contributions from autofluorescence. To make this correction, total CFP and YFP fluorescence intensities are measured using cells that do not carry either CFP or YFP reporters. The average autofluorescence intensity for each is calculated using measurements from a minimum of 80 cells. Corrected CFP and YFP fluorescence intensities for a given cell are determined by subtracting the product of the respective average autofluorescence values and the area defined by the traced object for that cell. We observed minor variations in autofluorescence during the initial stages of this investigation. Any impact of such variations was ameliorated by conducting analysis of cells for autofluorescence in parallel with the corresponding experimental samples. Steady-state expression values for the different reporters are the average for measurements on a minimum of 140 individual cells. In experiments involving time courses, each data point represents the average value from measurements on a minimum of 22 individual cells.
RNA preparation and Northern blot analysis of UBI-Y-ΔkCFP, FUS1 and ACT1 mRNAs
Cultures of strains ∑182 (no CFP control) and ∑284 (PFUS1-UBI-Y-ΔkCFP) were grown to 1–2 × 107 cells/ml in SD + His + Ura + Trp medium at 30 °C. A 20 ml aliquot from each culture was removed for analysis of steady-state mRNAs in uninduced cultures. Mating pheromone (α-factor, 1 µM) was added to the remaining portion of each culture and samples (20 ml) were removed for analysis at the indicated times. Cells in each aliquot were harvested by centrifugation and total RNA was prepared by the glass bead lysis method (Broach et al., 1979). RNA was glyoxal-denatured, fractionated on 1% agarose gels in 10 mM phosphate buffer (pH 7.0), and transferred to nylon membranes (Immobilon-Ny+, Millipore, Bedford, MA).
Radiolabelled probes for hybridization to the mRNA blots were prepared from purified DNA fragments by random-primed synthesis reactions (Boehringer-Mannheim, Indianapolis, IN). The FUS1 mRNA probe was synthesized using the 1.5 kb EcoRI fragment of pSL589 as template DNA (from G. Sprague Jr, University of Oregon, Eugene, OR). The CFP mRNA probe was derived from the 678 bp BamHI–PvuII fragment of pNC827 as template DNA. The ACT1 mRNA probe was synthesized using the 2.2 kb EcoRI–HindIII fragment of pYACT1 as template DNA (Ng and Abelson, 1980). Each blot was first hybridized to the FUS1 radiolabelled probe. After stripping blots of the first probe, they were then hybridized to the CFP-radiolabelled probe. The process was repeated for hybridization to the ACT1 radiolabelled probe to provide an internal loading reference. Conditions for hybridization to radiolabelled probes, washing and stripping blots were as described by Cameron et al. (1979). The hybridization signals were detected by direct scanning with a Molecular Dynamics STORM 860 PhosphorImager (Amersham, Piscataway, NJ) and quantified using the Molecular Dynamics Image Quant 5.0 software package (Amersham, Piscataway, NJ). Amounts of mRNA in each sample are specified as the ratio of the signal for the query mRNA to that of the ACT1 mRNA reference.
Reporter gene construction
We used a modular design to facilitate construction of different N-degron variants of cyan fluorescent protein (CFP) that can be expressed from an upstream activating sequence (UAS) of choice. This system is based on UAS and CFP plasmid cassettes that can be combined to generate families of UAS-UBI-[X-k]-CFP reporter alleles, where [X-k] represents different bipartite N-degron signal sequences (Figure 3). A set of UAS plasmids was created to have unique restriction sites (KpnI or XhoI and BglII) for cloning any UAS (Figure 2A; pNC820, pNC929, pNC824). This set of plasmids also carries the UBI4 coding sequence followed by an engineered codon (X = ATG, Met; GAA, Glu; or TAT, Tyr), which specifies the primary determinant of N-end rule pathway recognition. The second set of plasmids has CFP fused to different linker sequences (k = ek or Δk) for modulating the efficiency with which the N-degron signal is utilized (Figure 2B; pNC829, pNC860 and pNC1053). The CFP cassettes also have a selectable marker (HIS3MX6) and ∼0.4 kb of 5′ and 3′ flanking sequences from the URA3–TIM9 locus. The reporter alleles are assembled by inserting the KpnI–BamHI fragment from one UAS cassette into the corresponding sites of one of the CFP cassettes (Figure 3).
The resulting UAS-UBI-[X-k]-CFP HIS3MX6 alleles can be efficiently targeted to the URA3–TIM9 intergenic region on chromosome V for expression in S. cerevisiae simply by digesting the plasmids with SacI and SalI restriction enzymes prior to transformation of a his3 mutant strain (Figure 3). Cleavage with these enzymes releases the alleles from the pUC118 backbone of the assembled reporter plasmids and generates the free ends that target the chromosomal integration. The ubiquitin moiety of the expressed reporter alleles is cleaved from the nascent proteins during translation. This cleavage exposes the engineered amino acid at the N-terminus of N-degron variants of CFP and confers different relative N-end rule stabilities. Based on the behaviour of other N-degron test proteins, the following hierarchy of stability is expected: X = Met > Glu > Tyr; k = Δk > ek (see Figure 1C, Table 1).
The experiments reported here used either the GAL1 or FUS1 UAS to promote expression of different N-degron CFP reporter proteins. The GAL1 promoter allows for carbon source-regulated transcription, which facilitated the determination of CFP variant half-lives and parameters related to fluorescence maturation. The FUS1 UAS reporters allowed assessment of the CFP variants as temporal reporters of pheromone induced transcription. The different reporter plasmids used in these studies were assembled as described above and are diagrammed in Figure 2C, D. Notice that PGAL1-CFP reporter genes were constructed with and without N-terminal ubiquition fusions. Those without ubiquitin fusions are reference reporters that are not processed by de-ubiquitinating proteases. Each allele was integrated at the URA3–TIM9 intergenic locus for analysis of reporter gene function. This locus was chosen because the sequences abutting the UAS sequences in the integrated reporters are identical to the those abutting the UAS site in the pLGΔ178 promoter trap plasmid (Guarente et al., 1984). pLGΔ178 has been used routinely for reporter studies with many different UAS sequences. Therefore, targeting the fluorescent reporters to this integration site is likely to minimize interference with UAS function.
Steady-state expression levels of N-degron CFP variants
We expressed the different N-degron variants of CFP from the GAL1 promoter under culture conditions where cells were grown continuously in galactose containing medium. This condition fully induces the GAL1 promoter, so that cells achieve a steady state for expression of the different reporter proteins. This steady-state amount reflects the equilibrium between the reporter's rate of synthesis and rate of degradation. The rate of synthesis for variants within the -CFP, -ΔkCFP or -ekCFP groups should be the same because, with the exception of the N-terminal amino acid, the proteins are identical and are expressed from the same promoter. Thus, differences in the abundance of N-degron variants within a group will directly correlate with differences in their relative stabilities. We assessed steady-state reporter abundance, first by Western blot analysis of protein extracts and then by fluorescence detection in individual living cells.
Extracts prepared from cultures of the different PGAL1-X-CFP strains were fractionated by SDS-PAGE and transferred to nitrocellulose filters. From here forward, we use X to denote sequence variations in reporter alleles that include the different codons at the fusion junction, as well as the presence or absence of UBI4 and/or the ek or Δk extension to CFP. The X-CFP proteins were detected by using anti-GFP antibodies. Tubulin was detected on the same filters by using anti-tubulin antibodies to provide an internal reference protein and loading control (Figure 4A). The upper, middle and lower panels show comparisons for variants of CFP, ΔkCFP and ekCFP, respectively. The position of the ΔkCFP and ekCFP signals compared with the reference CFP signal on the corresponding blots reflects the size difference contributed by the respective 24 and 40 amino acid N-terminal extensions to CFP. The ratio of the CFP : tubulin signal of the different variant reporters was calculated and normalized to that for the CFP reference on the corresponding blot (Figure 4B). Reporter genes expressing nascent ubiquitin fusion proteins with a stabilizing methionine residue at the fusion junction (Ubi-M-CFP, Ubi-M-ΔkCFP, Ubi-M-ekCFP) have comparable steady-state abundance to their corresponding CFP, ΔkCFP or ekCFP reference protein without fusion to ubiquitin. The Ubi-Y-CFP variant, which has a strongly destabilizing N-terminal residue but no Δk or ek extension sequence to complete the N-degron signal, shows only a minor decrease in steady-state abundance compared with CFP. By contrast, those reporter genes expressing a nascent fusion protein with a destabilizing N-terminal residue and extension sequence (Ubi-E-ΔkCFP, Ubi-Y-ΔkCFP, Ubi-E-ekCFP and Ubi-Y-ekCFP) show significantly lower steady-state abundance of the reporter compared with the stabilized variants. The relative differences in steady-state amounts of the different N-degron variants are consistent with expectations based on the relative stabilities that were reported for the corresponding β-galactosidase and DHFR N-degron variants (Bachmair et al., 1986; Bachmair and Varshavsky, 1989; see Figure 1C).
To compare reporter expression by fluorescence measurements, we incorporated the PADH1-YFP constitutive reporter gene into strains with the PGAL1-X-CFP reporter genes. The YFP reference reporter allows for normalization in measurements of CFP expression. This normalization corrects for dilution effects related to differences in cell size and any fluctuations due to measurements of cells in different focal planes. Cultures of the dual-reporter strains were grown as above and samples were spread on microscope slides holding a thin layer of solid galactose medium. Images of different fields were captured for quantification of fluorescence, as detailed in Materials and methods. Representative micrographs show differential interference contrast (DIC), CFP and YFP fluorescence images of cells from the reference M-CFP and M-ΔkCFP cultures compared with the Ubi-M-ΔkCFP, Ubi-E-ΔkCFP and Ubi-Y-ΔkCFP cultures (Figure 5A). Images of the Ubi-M-CFP and Ubi-Y-CFP reporters were also analysed but the micrographs are not shown. The steady-state abundance of the destabilized Ubi-E-ekCFP and Ubi-Y-ekCFP reporters proved to be too low to have a useful detection range above the autofluorescence background. For this reason, reporter genes with the ek extension to CFP were not further characterized.
Figure 5B shows the average amount of fluorescence in strains expressing the different reporter genes. The number of cells contributing to the calculated average is given above the bar (each average includes cells from three or more independent cultures). The standard deviation for the cell population is shown by the error bars. The analysis confirms that there is no significant difference in the fluorescence properties of CFP (M-CFP) and the variant with a Δk extension (M-ΔkCFP). However, there does appear to be a slight decrease in the steady-state fluorescence of the corresponding reporters generated from nascent Ubi fusion proteins with a stabilizing N-degron signal (Ubi-M-CFP and Ubi-M-ΔkCFP). Because this effect was not seen for measurement of the corresponding reporter proteins by Western blot analysis, we suspect the de-ubiquitinating processing step may affect CFP maturation and fluorophore development. Nevertheless, the decreased steady-state fluorescence for the destabilizing N-degron CFP variants (Ubi-Y-CFP ∼ Ubi-E-ΔkCFP > Ubi-Y-ΔkCFP) is consistent with these variants having a shortened half-life compared with CFP and ΔkCFP.
Half-life determination of different N-degron CFP variant proteins
To determine the protein half-lives, cultures of strains with the different PGAL1-X-CFP reporter genes were grown in galactose medium to induce their steady-state expression as above. Cultures were then switched to dextrose medium, which immediately inhibits further transcription (Johnston et al., 1994). The persistence of pre-existing reporter protein at different times after the switch to dextrose was then examined by Western blot analysis of protein extracts prepared from culture samples and by fluorescence detection in individual living cells. In contrast to protein stability determinations made in the presence of general translation inhibitors, such as cycloheximide, the strategy used here measures rates of degradation under normal growth conditions. Thus, the apparent relative stabilities reported are those that will apply to the applications for which these reporters are designed.
For the Western blot comparisons, a sample of the starting galactose culture was removed for protein extract preparation and measurement of the steady-state abundance at t = 0. Dextrose was added to the remainder of the cultures and samples were removed at specified intervals for protein extract preparation and comparison to the corresponding t = 0 sample from each culture. The CFP reporter protein and the reference tubulin protein in each sample were detected on Western blots as before (Figure 6A, B). The ratio of CFP to tubulin at each time was calculated from densitometry measurements of the respective signals. The values were then normalized to the amount at t = 0 for a given culture to give the relative amount at each time after dextrose addition. Plots of ln(P) vs. time, where P is the relative amount of protein, were made to compare the decay rates for the different reporters (Figure 6C, D). Half-lives for each protein were calculated from the slopes determined by linear regression fits to the data (Table 1). The relative stabilities of the different N-degron CFP variants (M-CFP ∼ Ubi-M-CFP ∼ M-ΔkCFP ∼ Ubi-M-ΔkCFP > Ubi-Y-CFP ∼ Ubi-E-ΔkCFP > Ubi-Y-ΔkCFP) are as expected from those reported for other N-degron test proteins (Bachmair et al., 1986; Bachmair and Varshavsky, 1989).
Table 1. Comparison of X-CFP reporter half-lives determined from protein and fluorescence quantification method
72 ± 6
70 ± 2
55 ± 3
72 ± 2
75 ± 2
51 ± 5
12 ± 1
For comparisons of half-lives based on fluorescence measurements, cultures of the different strains with a PGAL1-X-CFP and reference PADH1-YFP reporter were grown to early log phase in galactose medium, as above. A sample of the starting galactose culture was removed for application to a microscope slide with solid galactose medium for measuring the steady-state abundance of each reporter. Dextrose was added to the remainder of the cultures at t = 0 and samples were removed for application to slides prepared with solid dextrose medium. The specified times for fluorescence measurements are the actual times images were captured with respect to t = 0 (the time taken for sample removal, mounting and focusing is ∼2–3 min). Amounts of reporter protein in cells were calculated as described for measurements in Figure 5. The values at different times after dextrose addition were normalized to the amount at t = 0. Plots of ln(F) vs. time, where F is the relative amount of fluorescence, were made to compare decay rates for the different reporters (Figure 6E, F). Half-lives were calculated from the slopes of these plots, as before (Table 1). The relative stabilities of the different N-degron CFP variants determined by fluorescence measurements were in good agreement with those determined by protein measurements.
Induction times for Ubi-M-ΔkCFP detection by protein compared with fluorescence methods
The time for GFP S65T fluorophore maturation and fluorescence detection in E. coli is 27 min (Heim et al., 1995). To estimate this parameter as it applies to the ΔkCFP reporters in yeast, we monitored galactose induction profiles for the PGAL1-UBI-M-ΔkCFP reporter gene by Western blot analysis and by fluorescence in individual living cells (Figure 7). Cultures were grown to early log phase in non-inducing raffinose medium and then switched to galactose medium to induce transcription of the reporter gene. Samples of the liquid culture were removed for protein extract preparation before and at the indicated times after galactose addition. M-ΔkCFP is detectable by 60 min and reaches 50% maximal amounts (1/2Max) at ∼90 min after the switch to galactose medium (Figure 7A, C). Cells from the raffinose culture were applied to a microscope slide with solid raffinose medium and immediately imaged to determine the basal level of reporter gene expression by fluorescence measurements (t = 0). Cells from the culture after galactose addition were applied to slides prepared with solid galactose medium. The specified times for fluorescence measurements are the actual times images were captured with respect to t = 0 (Figure 7B, C). Amounts of reporter protein in cells were determined from the images, as before. Fluorescence above the autofluorescence background is detectable by 150 min and reaches 1/2Max at ∼220 min after galactose addition. The 130 min delay for induction to 1/2Max for fluorescence compared with Western blot measurements suggests that the time it takes for CFP fluorophore maturation and detection in yeast is longer than the time for maturation of the related GFP S65T protein in E. coli. We did not assess whether the apparent difference in maturation time stems from the additional mutations in CFP compared with GFP S65T or from differences between yeast and E. coli that affect chromophore development.
Assessment of fluorescent proteins as a reporter of transient gene expression
Yeast respond to mating pheromone by inducing a transient transcription programme (Roberts et al., 2000). FUS1 encodes a membrane protein that is localized to the tips of mating projections and is required for cell fusion between mating partners. Its expression is strongly induced by pheromones and serves as a standard indicator for mating-specific gene expression. Therefore, the FUS1 UAS was exploited to examine fluorescent proteins as reporters of transient transcription. We first used Northern blot analysis to compare the mRNA profile of the PFUS1-UBI-Y-ΔkCFP gene reporter to that of the endogeneous FUS1 in the same strain. Samples for RNA preparation were taken from a culture before and at the indicated times after addition of pheromone to the culture. RNAs were fractionated, transferred to a membrane and hybridized to radiolabelled probes for detection of FUS1, UBI-Y-ΔkCFP and ACT1 mRNAs. The ACT1 mRNA served as an internal reference and loading control for quantification of the FUS1 and UBI-Y-ΔkCFP mRNAs.
As expected, FUS1 mRNA is rapidly induced ∼40-fold within 15 min after addition of mating pheromone to cultures and then declines to near basal amounts by 90 min (Figure 8). The fluorescent reporter gene mRNA shows induction kinetics identical to that for the FUS1 mRNA but there is a noticeable difference in the attenuation profiles. Although we did not precisely define the induction peak time, to simplify comparison of these induction profiles, we assume that maximal (Max) amounts of mRNA appear at 15 min after pheromone addition. The mRNA levels decline to 1/2Max 15 min sooner for FUS1 mRNA (45 min) than with CFP mRNA (60 min) (Figure 8B). Also, the CFP mRNA persists at higher levels than the FUS1 mRNA (at 180 min after pheromone addition, FUS1 mRNA is only two-fold above the basal amount, whereas CFP mRNA is eight-fold higher than the basal amount). These differences most likely reflect a slight difference in the stability of CFP mRNA compared with that for FUS1 mRNA (the half-life for FUS1 mRNA is estimated to be 14 min; Wang et al., 2002). Such differences are anticipated to affect temporal profiles but this limitation is inherent to most reporter genes commonly in use.
We then examined pheromone response profiles by fluorescence measurements using cells that express destabilized vs. stable versions of the FUS1 reporter gene. The two strains also express the reference PADH1–YFP reporter gene for quantification of the reporter signals as before. Samples of cells in liquid culture were removed before and at different times after addition of pheromone. DIC, CFP and YFP images were captured and amounts of reporter fluorescence quantified from the images as before (Figure 9A, B). The PFUS1-UBI-Y-ΔkCFP reporter reveals transient expression by fluorescence measurement that is characteristic of pheromone induction (Figure 9B). However, the details of the induction kinetics and time of maximal expression are different from those obtained by measurements of reporter mRNA abundance. Because of the limited sampling at early times discussed above, we use the 15 min time point as the apparent time at which maximal (Max) mRNA expression occurs. The mRNA signal reaches 1/2Max by ∼3 min and declines back to 1/2Max by ∼63 min after pheromone induction. The 1/2Max induction, peak and 1/2Max attenuation times for the Y-ΔkCFP fluorescence profile occur at 60, 150 and 225 min, respectively. Similar to what was seen for the comparison of protein and fluorescence with the GAL1 induction profiles, the pheromone induction kinetics for the fluorescent reporters are delayed compared with the mRNA. Notably, the attenuation kinetics for the mRNA and Y-ΔkCFP fluorescence are more similar, the time from peak amounts back to 1/2Max being 60 and 75 min, respectively. PFUS1-UBI-M-ΔkCFP reporter gene expression measured by fluorescence has the advantage of a stronger signal but the stability of the protein completely masks the attenuation phase of the response (Figure 9B). It is also noteworthy that the M-ΔkCFP and Y-ΔkCFP reporters show similar induction kinetics, suggesting that both require the same time for fluorophore maturation and have similar delays for reporting the onset of transcription. However, the shorter half-life reporter is superior for revealing transient profiles and in this regard is a good option for applications where the objective is to define relative patterns of gene expression.
Reporters of temporal transcription programmes are most useful if they have a short half-life that restricts their detection to the window in which their transcripts are present and translated. The closer this condition is satisfied, the better the reporter is able to reveal transient profiles and the less likely it is to magnify basal expression. To meet this criterion for reporters of transcription in individual living cells, we designed cyan fluorescent reporter proteins (CFP) with different stabilities. The members of this group vary in stability because of N-terminal fusions to different N-degron sequences, which confer different rates of degradation by the N-end rule pathway. The CFP reporters characterized here have half-lives of 75, 50 and 5 min. To illustrate their performance, we compared the 75 and 5 min half-life CFP proteins as temporal reporters for the pheromone-responsive FUS1 promoter. The short half-life reporter revealed a transient profile that was otherwise masked by accumulation of the longer-lived protein. Additionally the attenuation kinetics for the short-lived reporter reasonably followed that of its mRNA.
The speed with which any fluorescent protein-based system can report transcriptional induction is constrained by the time required for fluorophore maturation. Fluorescence maturation entails an oxidation step, which requires ∼2 h for wild-type GFP from A. victoria. A single amino acid subsitution in GFP (S65T) results in a derivative with six-fold greater brightness and four-fold faster oxidation to the mature fluorophore (Cubitt et al., 1995). The version of CFP for the reporters described here is derived from GFP S65T, but has a maturation time that appears to be more similar to that of wild-type GFP. This estimate comes from the comparison of the induction kinetics for GAL1-expressed CFP by Western blot and fluorescence methods. The time at which the 1/2Max amount of protein is detected on Western blots occurs ∼130 min prior to that for fluorescence. This delay confines the utility of these reporters for timing transcriptional programmes to instances where a correction factor can be determined by comparing the induction kinetics for the reporter to that of the mRNA in question. These reporters are nonetheless well suited for applications where the objective is to compare different genetic backgrounds or environmental conditions for their effect on a given transcription profile.
The amount of the reporter protein that is able to mature and produce fluorescence decreases with the decreasing stability of these family members. As a consequence, the shortest-lived reporter of this generation of N-degron CFP proteins is limited to applications with more robust promoters. We anticipate that this limitation will be overcome in the future. Different substitution variants of GFP continue to be generated and selected for brighter fluorescence and improved maturation times (Miyawaki et al., 2003). The same strategy applied to improved GFP variants is expected to generate the similar range of stabilities, but with better sensitivity and speed as reporters of changes in transcription.
We thank Dr T. Davis, University of Washington, Seattle, WA, and The Yeast Resource Center, for the plasmids pDH3 and pDH5, which carry CFP and YFP, respectively. We also thank Dr E. Yeh for discussions and guidance during the initial phases of this work and Drs J. Cook and T. Elston for valuable discussions and comments on the manuscript. This work was supported by NIH Grants GM39852 and GM067809.