Using the cre-lox recombination system to assess functional impairment caused by amino acid substitutions in yeast proteins

Abstract

A method was developed to assess the functional significance of a sequence motif in yeast Upf3p, a protein required for nonsense-mediated mRNA decay (NMD). The motif lies at the edge of the Upf3p-Upf2p interaction domain, but at the same time resembles the canonical leucine-rich nuclear export sequence (NES) found in proteins that bind Crm1p exportin. To test the function of the putative NES, site-directed mutations that cause substitutions of conserved NES-A residues were first selected to identify hypermorphic alleles. Next, a portable Crm1p-binding NES from HIV-1 Rev protein that functions in yeast was fused en masse to the C-terminus of variant Upf3 proteins using loxP sites recognized by bacterial cre-recombinase. Finally, variant Upf3-Rev proteins that were functional in NMD were selected and examined for the types of amino acid substitutions present in NES-A. The mutational analysis revealed that amino acid substitutions in the Upf3 NES impair both nuclear export and the Upf2p-Upf3p interaction, both of which are required for Upf3p to function in NMD. The method described in this report could be modified for the genetic analysis of a variety of portable protein domains.

Introduction

UPF3 is one of three genes in the yeast Saccharomyces cerevisiae that is required for the nonsense-mediated mRNA decay pathway (NMD) (1-5) Aberrant mRNAs where translation is prematurely terminated become substrates in the NMD pathway and are rapidly degraded. Mutations in UPF3 stabilize nonsense mRNAs resulting in rates of decay similar to wild-type transcripts (5). We identified a sequence motif in Upf3p called NES-A that resembles the canonical leucine-rich nuclear export sequence (NES) found in many other proteins that binds to Crm1p, the exportin that carries cargo through nuclear pores into the cytoplasm (6). However, NES-A is located in a complex region at one edge of a domain that mediates a protein-protein interaction between Upf3p and Upf2p. This region also shows weak sequence similarity to canonical RNA recognition domains (RRM) (6, 7).

The segment containing NES-A was shown to direct a nuclear-localized reporter to the cytoplasm (6). Alanine substitutions of evolutionarily conserved leucine and isoleucine residues in NES-A abolished export of the same reporter. Three simultaneous alanine substitutions in NES-A (upf3-triple) inactivated NMD and caused the accumulation of Upf3p in the nucleolus, a region within the nucleus. These findings support a role for NES-A in nuclear export. However, there were three unexplained features of NES-A that distinguished it from the canonical NES known to drive export of other proteins (6, 7). Although the NES-A sequence matches the consensus leucine-rich NES sequence, it is present in reverse orientation relative to the N- and C- terminus compared to all other NES domains that direct export. Also, the canonical NES binds to the Crm1p exportin (8-10), but Upf3p is not dependent on Crm1p for export and presumably does not bind to Crm1p. Finally, when export was fully restored by adding a heterologous NES at the C-terminus of Upf3-Triple, NMD was only partially restored, suggesting that NES-A may have one or more functions in addition to promoting export (6).

In order to resolve these conflicting observations and to separate the possible functions of NES-A, we developed a unique way to analyze mutations in NES-A. The method, which we refer to as the dual selection for loss/regain of function, requires two steps. The first is a selection for mutations in NES-A that impair the function of UPF3. The second step is based on the observation that the canonical leucine-rich NES is a self-contained and portable binding site for Crm1p (10). Mutations in NES-A that specifically impair NMD by blocking Upf3p export were identified by fusing a bona fide Crm1p-binding NES to the mutant Upf3p proteins and then selecting for those that regain function in NMD.

To accomplish this, we utilized the cre-lox recombination system commonly used for genomic manipulation. Cre-mediated recombination between loxP sites has been used successfully to eliminate or activate gene expression in a tissue-specific and time-specific manner, reduce the number of copies of integrated transgene arrays, and to direct the site-specific insertion of transgenes (11, 12). In our system, we used cre-lox recombination to cause a fusion of the leucine-rich NES from the mammalian HIV-1 Rev protein (13-16), which directs export in yeast, to the C-terminus of Upf3 mutant proteins. Two loxP sites were positioned such that induction of cre-recombinase eliminated the intervening sequence resulting in an in-frame translational fusion between sequences coding for the Rev NES and Upf3p. Following recombination, the alleles were analyzed to see if defects in nuclear export and NMD were corrected by the C-terminal addition.

In this report we focus on the details of the two-step selection and describe the characterization of three mutations causing amino acid substitutions in NES-A that impair the function of Upf3p. We suggest that a similar approach could yield valuable information when applied to the analysis of other sequence motifs and domains in other proteins.

Materials and Methods

Strains, genetic methods, and supplemental information

Strain RSy5 (MATα ade2-1 leu2-1 tyr7-1 can1-100 upf3-Δ1 trp1 ura3 his3 GAL2), which carries a null allele of UPF3 (7), was used as a transformation recipient in the dual selection for loss/regain of function. Further information on plasmids, strains, and mutant alleles of UPF3 beyond what is described in this report can be found in Shirley et al. (7). Yeast was transformed by electroporation (17), which is described in the Protocol section. Supplemental information including the complete sequences for the plasmids pRR2 and pRLS207 described below can be found at http://www.molbio.wisc.edu/culbertson/creloxplasmids/.

cre/lox-mediated recombination of plasmid pRR2

The plasmid pRR2 (Figs. 1 and 3) was designed to analyze the functional effects of mutations in NES-A. pRR2 contains two loxP recognition sites oriented in the same direction. In cells expressing the site-specific cre-recombinase, double-strand breaks occur at both loxP sites. The free DNA ends recombine, resulting in a re-circularized plasmid in which the DNA between the two loxP sites is deleted. Recombination also results in the deletion of one of the loxP sites. loxP sites must be in the same orientation in order to delete the intervening segment. If they are in opposite orientation, recombination causes an inversion of the intervening segment. In pRR2, one loxP site is located within the UPF3 open reading frame directly 5' of the stop codon (designated UPF3-loxP). The second loxP site is located 5' of sequences coding for a functional leucine-rich NES from the viral HIV-1 Rev protein and the 3'-UTR of UPF3 (loxP-Rev-UTR). In S. cerevisiae, the HIV-1 Rev-NES domain binds to yeast Crm1p and promotes the nuclear export of any protein to which the NES is fused (8, 9). Expression of cre-recombinase results in the excision of the DNA sequence between the loxP sites and the fusion of the Rev-NES and UPF3 3'-UTR to UPF3-loxP. The recombined plasmid expresses a version of Upf3p that has a functional NES at the C-terminus. The pre-recombined UPF3-loxP and post-recombined UPF3-loxP-Rev alleles both express functional versions of an otherwise wild-type Upf3p as evidenced by the ability of UPF3-loxP and UPF3-loxP-Rev to complement a upf3-Δ1 null allele (data not shown). Plasmid-mediated complementation was assayed by suppression of nonsense mutations and by accumulation of CYH2 pre-mRNA as described below. The S. cerevisiae ADE2 gene is located in the segment between the two loxP sites. In the ade^- strain RSy5, the ADE2 gene on the plasmid allows selection of the pre-recombined plasmid and the ability to monitor recombination following cre expression (Fig. 3). Colonies expressing the pre-recombined UPF3-loxP plasmid are white. After recombination and excision of the ADE2 gene, colonies turn red because ade2 mutants accumulate a red pigment.

Fig. 1. Plasmids used in the dual screen for loss/regain of function.

Both plasmids are circular with critical elements shown in linear format. Both plasmids are shuttle vectors that replicate in both yeast and E. coli (A) The diagram shows the organization of elements important for expression of cre recombinase in pRLS207. (B) The diagram shows the organization of elements in pRR2 used in mutagenisis and cre-mediated recombination. Wild-type amino acids targeted by mutagenesis are shown in red.

Fig. 3. Design of the dual screen for loss/regain of function.

Recombination mediated by bacterial cre recombinase expressed in yeast places a function nuclear export sequence (NES) derived from the mammalian HIV-1 Rev protein at the C-terminus of mutant Upf3p. Regain of function after recombination implies that a mutation in the Upf3p NES motif impairs nuclear export and that nuclear export is required for the function of Upf3p. The petri plates shown at the bottom represent idealized results. Loss of function is indicated by co-suppression of leu2-1 tyr7-1 (white colonies before recombination that grow in the absence of leucine and tyrosine) and suppression of can1-100 (grey imprints of colonies that fail to grow). Regain of function after recombination is indicated by loss of suppression allowing growth in the presence of canavanine (red colonies). White colony color results from the presence of the ADE2 gene on plasmid pRR2 before recombination and red colony color results from deletion of the ADE2 gene after recombination. The cre-recombinase expressed from plasmid pRLS207 is controlled by the GAL1 promoter and is induced with galactose as carbon source.

The pRR2 plasmid, which was constructed starting with the vector pRS316 (7, 18), is a single copy plasmid containing an S. cerevisiae origin of replication (ARS4), the centromere of S. cerevisiae chromosome IV (CEN4), and the S. cerevisiae URA3 gene which confers selective growth of Ura⁺ transformants in medium lacking uracil. pRR2 carries a bacterial ampicillin resistance gene and the pBR322 origin of replication allowing selection of E. coli transformants.

Expression of cre recombinase

Plasmid pRLS207 expresses cre, the site-specific recombinase from bacteriophage P1 (19) (Fig. 1). In pRLS207, the cre open reading frame is flanked by the S. cerevisiae GAL1 promoter and CYC1 transcription termination signal sequence. When cells harboring pRLS207 are grown in glucose, GAL1^P-cre-CYC1^T is not expressed, but in the presence of galactose transcription is induced. pRLS207, which was constructed starting with the vector pRS423 (18) is a multicopy plasmid carrying the origin of replication from the S. cerevisiae 2αm endogenous plasmid and the yeast HIS3 gene, which confers selective growth of His⁺ transformants in medium lacking histidine. pRLS207 carries the ampicillin resistance gene and the pBR322 origin of replication allowing selection of E. coli transformants.

Phenotypic assays

The phenotypes of mutations in NES-A were assessed by three different assays: growth tests (suppression of nonsense mutations), accumulation of a pre-mRNA whose abundance and stability depends on UPF3 function, and cytological localization of the protein products of mutant upf3 alleles. All three assays summarized here are described in greater detail in previous reports (6, 7).

Nonsense suppression was monitored in strains carrying the leu2-1 (UAA) and tyr7-1 (UAG) nonsense mutations, which prevent growth on medium lacking leucine and tyrosine. Mutations in UPF3 that cause loss of function suppress leu2-1 and tyr7-1 and allow growth on medium lacking leucine and tyrosine. The can1-100 (UAA) nonsense mutation confers resistance to canavanine (Can^r) in a wild-type UPF3 strain, whereas loss-of-function mutations in UPF3 suppress can1-100 and confer sensitivity to canavanine (Can^S). In each of these three cases, suppression results from increased stability of the nonsense mRNA and from increased read-through of translation termination at the premature stop codon.

CYH2 coding for ribosomal protein L29 contains an intron that is inefficiently spliced (20). An in-frame stop codon at position 19 in the intron triggers rapid decay of unspliced pre-mRNA (21). The pre-mRNA accumulates to a 5-6 fold higher level in upf3 ^- strains due to an increase in the pre-mRNA half-life but without any change in the mature mRNA half-life. The CYH2 pre-mRNA/mRNA ratio therefore provides a quantitative assessment of function of the NMD pathway. Methods for quantitative measurement of RNA accumulation using Northern blotting are described by Leeds et al. (1) and Atkin et al. (22). To determine relative accumulation, the CYH2 pre-mRNA/mRNA ratio in upf3 mutants was normalized to the ratio in a wild-type UPF3 strain to calculate the fold change. Fold changes were averaged across three or more trials and evaluated statistically as described previously (7).

Epitope-tagged Upf3p was visualized by immunofluorescence. Early log phase cells were processed as described previously (6, 23). The ability of mutant Upf3 proteins to interact with Upf2p was assayed using the yeast two-hybrid system as described by James et al. (24). The details of the two hybrid experiments are described by Shirley et al. (7). The results are not presented in detail here but are summarized in Table 2.

Table 2.

Phenotypic properties of NES-A mutants.

UPF3 allele	NES-A amino acid substitutions	Nonsense suppression	Percent function in NMD**	Nuclear export defect***	Upf2/Upf3 interaction****
UPF3 (wild-type)	88-LVI RLLPPNL-97	-	100	no	++
Upf3-Δ1 (null)	no protein made	+++	0	N/A	N/A
upf3-Triple-HA	L88A I90A L93A	++	25	yes	-
upf3-Triple-HA-Rev		+	46	no	ND
upf3-Triple-HA-M10		++	25	yes	ND
upf3-nes2	L88D I90A L92G L93S	+	46	yes	-
upf3-nes2-Rev		+	46	no	ND
* “+++” represents full suppression of the leu2-1 and tyr7-1 nonsense alleles characteristic of the upf3-Δ1 null allele. “–“ indicates the complete lack of suppression characteristic of wild-type UPF3. “+” and “++” indicate weaker and stronger levels of suppression intermediate between upf3-Δ1 and UPF3. **The percent function was determined using the CYH2 assay for pre-mRNA accumulation (Materials and Methods and (7)). ***Epitope-tagged Upf3 proteins were localized by immunofluorescence microscopy (Materials and Methods and (7)). Wild-type Upf3p was visualized throughout the cytoplasm with no staining detected in the nucleus (6, 7). A defect in nuclear export was indicated by the increase in nucleolar accumulation (7). ****The ability of the Upf3 proteins to interact with Upf2p was assayed using the yeast two-hybrid system (7). “++” and “-“ indicate robust growth and lack of growth, respectively, on selective media. ND = Not Determined.

Mutagenesis

Multi-round PCR was used to generate mutations in the sequences coding for residues L88, I90, L92 and L93 in NES-A (Fig. 2). The mutagenic oligonucleotide RR03 was degenerate for nucleotides in each of four codons such that the mutations in each codon could potentially result in 15 different amino acid substitutions (not including Met, Phe, Tyr, Cys, and Trp). Variants where U was the first nucleotide of a codon were avoided so that UAA, UAG, and UGA termination codons would not be generated. Alternative schemes designed to produce all 20 amino acid substitutions were not attempted because they necessitate the inclusion of stop codons.

Fig. 2. Strategy for mutagenesis of NES-A in pRR2 using multi-round PCR.

The structure of UPF3 is shown at the top including known or suspected domains: NLS, nuclear localization signal sequence; Upf2p binding, the region required for binding to Upf2p; RRM, a region showing similarity to an RNA recognition motif identified in studies of the human ortholog to yeast Upf3p (27); NES-A, the motif resembling the leucine-rich nuclear export domain corresponding to the binding domain for the exportin Crm1p.

In the first round of PCR, oligonucleotides primers RRO3 and LSO183 (Fig. 2) and template plasmid pLS17 (CEN URA3 UPF3) (7) were used to amplify DNA fragments containing random mutations at each of the four codons. The first-round PCR products were used as a megaprimer in the second round along with oligonucleotide T7 (Fig. 2) and template pLS17. The PCR fragments amplified in the second round contained recognition sites for SnaBI and SacII. After digestion with these enzymes, the resulting DNA fragments were used to replace the SacII–SnaBI fragment containing wild-type NES-A in plasmid pRR2.

The ligation reaction between SacII-SnaBI PCR fragments and SacII-SnaBI-digested pRR2 was transformed into XL10-Gold ultra-competent E. coli cells (Stratagene, La Jolla, CA) and plated on standard Luria-Bertani (LB) medium plus 100 μg/ml ampicillin. Plasmid DNA isolated from pools of E. coli transformants were used to transform yeast strain RSy5 as described below. The number of E. coli transformants to be pooled in order to obtain with 95% confidence every possible amino acid combination caused by the mutagenic oligonucleotide was estimated using the formula:

where P = the probability that each combination of amino acid substitutions will be represented once, N = the number of bacterial colonies pooled, n = the number of the rarest combination, and T = the total number of combinations. Since some of the amino acids are represented by only one of the 24 possible codon combinations, 70 colonies must be pooled in order to have a 95% probability of obtaining at least one of each of the 24 combinations at a given codon. This assumes that each combination is represented equally after PCR amplification, which might not occur if any fragments are preferentially amplified. When four codons are mutated, a pool of ~10⁶ E. coli transformants would be required to ensure detection of the rarest combinations. In the experiments described below, we pooled far fewer than this number and did not, therefore, exhaust the potential to uncover mutants of interest. In designing a screen of this type, there are trade offs between the advantages of mutating multiple codons and the disadvantages of needing to screen large numbers of transformants to recover all possible mutant combinations.

Results and Discussion

Rationale

The combined mutagenesis and dual selection requires five steps (Figs. 2 and 3): (1) mutagenic PCR to create mutations in pre-chosen residues in NES-A, (2) selection for yeast co-transformants that carry both a mutagenized pRR2 plasmid and the pRSL207 plasmid that codes for cre-recombinase, (3) plate tests for loss of function before recombination, (4) induction of cre-mediated plasmid recombination that places the HIV-1 Rev-NES at the C-terminus of mutant Upf3 proteins, and (5) plate tests to assess whether the C-terminal Rev NES improves function.

Since mutant alleles were identified based on loss and regain of function of mutant Upf3 proteins in the absence or presence of C-terminal additions, respectively, it was important to determine whether the additions of loxP and Rev NES affect function by themselves. The addition of an HA epitope at the N-terminus was known to cause a slighty impairment of Upf3 function (6).

Appropriate alleles were tested for their ability to complement a upf3-Δ1 null allele as measured by nonsense suppression and CYH2 pre-mRNA accumulation (see below). The C-terminal addition of loxP and the HIV-1 Rev-NES had no significant effect on the function of wild-type Upf3p (data not shown), indicating that the NMD phenotype of the mutant Upf3-loxP and Upf3-loxP-Rev proteins would be due solely to their amino acid substitutions in NES-A.

Dual selection for loss/regain of function

Plasmid DNA was isolated from a pool of ~8,600 E. coli transformants carrying PCR-mutagenized pRR2 plasmids. The pRR2 DNA was mixed with pRLS207 plasmid DNA and used to co-transform strain RSy5 (MATα ade2-1 leu2-1 tyr7-1 can1-100 upf3-Δ1 trp1 ura3 his3 GAL2) to a Ura⁺ His⁺ phenotype (pRR2 carries URA3 and pLRS207 carries HIS3, see Fig. 1). Omission of adenine in the growth medium and white colony color ensured that transformants contained pre-recombined plasmids. Transformants were plated directly onto medium lacking uracil, histidine, adenine, leucine and tyrosine in order to select for transformants harboring pRR2 plasmids that suppress the tyr7-1 and leu2-1 nonsense mutations. Out of ~84,000 yeast transformants plated, 7,200 had a Tyr⁺ Leu⁺ phenotype, indicating that at least one mutation causing impairment of UPF3 function was carried on the pRR2 plasmids.

Two alternative approaches can be used to analyze the Tyr⁺ Leu⁺ transformants. The transformants can be replica-plated to medium containing canavanine. Complete loss of UPF3 function leads to a Can^S phenotype due to suppression of the canavanine resistant nonsense allele can1-100. Mutations that partially impair function cause an intermediate phenotype of weak resistance (increased sensitivity) compared to wild-type. The strength of the phenotype can be adjusted by varying the concentration of canavanine. Growth on canavanine prior to recombination can be compared to growth of the transformants on canavanine following cre-mediated recombination between the loxP sites. Recombination can be induced by replica-plating the Tyr⁺ Leu⁺ transformants to medium containing galactose to activate the expression of cre. In a second approach, the phenotypic analysis of primary transformants on canavanine before recombination can be skipped. Primary transformants can be replica-plated directly to medium containing galactose to induce recombination and then to plates containing canavanine. Using the latter approach, we identified 389 can^r colonies from the original 7,200 transformants.

The mutations in the pRR2 plasmids from seven of these transformants were analyzed by sequence analysis of PCR-amplified DNA fragments (Table 1, (7)). All seven transformants were chosen based on their robust growth on medium lacking leucine and tyrosine and on medium containing canavanine prior to and after cre-induced recombination, respectively. The seven alleles designated upf3-nes1 through upf3-nes7 contained three to four amino acid substitutions of the conserved leucine and isoleucine residues in NES-A (Table 1). The amino acid substitutions were analyzed by replacing the wild-type NES in Upf3p with each of the seven NES variants to quantitate their extent of NMD impairment. Quantifying the CYH2 pre-mRNA/mRNA ratio in transformants carrying the upf3 alleles revealed that the ability of the mutant proteins to function in NMD ranged from 27% to 87% compared to wild-type Upf3p (Table 1.

Table 1.

Amino acid substitutions in NES-A.

Allele	NES-A amino acid sequence	Percent function in NMD*
UPF3	88-LVI RLLPPNL-97	100
upf3-Triple-HA	L88A I90A L93A	25
upf3-nes1	L88V I90V L92G L93V	87
upf3-nes2	L88D I90A L92G L93S	46
upf3-nes3	L88N I90D L93R	nd
upf3-nes4	L88V I90V L93G	nd
upf3-nes5	L88R I90L L92G L93S	37
upf3-nes6	L88G I90V L92P L93K	36
upf3-nes7	L88Q I90R L92V L93G	27
*The percent function was determined using the CYH2 assay for pre-mRNA accumulation (Materials and Methods and (7)).

Phenotypic analysis of selected mutants

The genetic, molecular, and cytological phenotypes of two alleles, upf3-nes2 (L88D I90A L92G L93S) and upf3-nes7 (L88Q I90R L92V L93G) were characterized in the presence or absence of a functional NES at the C-terminus (Table 2). In this analysis we included a previously described allele called upf3-Triple (L88A I90A L93A) (7). We reported previously that Upf3p-Triple tagged with the HA epitope was reduced in abundance compared to Upf3p-HA by 50% (6). To increase the abundance of the mutant protein to a level similar to Upf3p-HA, the experiments were done using a plasmid that carries two copies of the upf-Triple-HA gene, which results in two-fold over-expression of the mutant protein (6). The upf3-nes2-HA and upf3-nes-HA alleles produced proteins that were similar in abundance to the wild-type protein when present in a single gene copy.

The upf3-nes2, upf3-nes7, and upf3-Triple-HA alleles all cause suppression of the leu2-1, tyr7-1, and can1-100 nonsense mutations, but the efficiency of suppression was less than that of upf3-Δ1 (Table 2, (6)).The growth tests suggest that all three alleles cause partial impairment of function. When DNA coding for the HIV-1 Rev NES was present following the last codon of each mutant gene, the level of suppression was unchanged for upf3-nes2 and upf3-nes7, but was reduced for upf3-Triple-HA. This suggests that the functional impairment caused by the nes2 and nes7 amino acid substitutions was unaltered when a functional NES is added to the C-terminus, whereas the functional impairment caused by the triple alanine substitutions is partially alleviated by the C-terminal addition.

To quantitate the effects of the mutations in the presence and absence of a C-terminal NES, the CYH2 pre-mRNA/mRNA ratio was determined (Table 2, (6)). The upf3-nes2 and upf3-nes7 alleles produce proteins that were 46% and 27% as functional as wild-type Upf3p, respectively (Table 2). The addition of a functional NES at the C-terminus caused no change in the CYH2 pre-mRNA/mRNA ratio. However, the addition of a C-terminal NES to the Upf3-Triple-HA protein improved function from 25% to 46% (Table 2). To verify that the improved function was due to the addition of a functional C-terminal NES, a control experiment was done using the HIV-1 Rev-M10, which carries a mutation that abolishes the ability of the NES to promote nuclear export. There was no change in the CYH2 pre-mRNA/mRNA ratio when HIV-1 Rev-M10 was placed at the C-terminus of Upf3-Triple-HA (Table 2).

To examine the effects of the mutations on nuclear export, HA-tagged versions of the mutant proteins were localized by immunofluorescence using HA monoclonal antibodies (Table 2, (6)). In wild-type strains, Upf3p-HA localizes to the cytoplasm without any detectable signal in the nucleus. However, nuclear

Upf3p-HA is detected either when the wild-type protein is over-expressed or when mutant Upf3p-HA is impaired in its ability to export from the nucleus. Under these circumstances, the immunofluorescent signal is concentrated in the nucleolus. Since the nucleolus is a sub-region within the nucleus, nucleolar accumulation serves as a marker for impaired nuclear export. HA-tagged versions of upf3-nes2, upf3-nes7, and upf3-Triple-HA produced proteins that accumulate in the nucleolus with some Upf3p still detectable in the cytoplasm. These results are consistent with the hypothesis that the functional impairment of the alleles in NMD is caused at least in part by a defect in nuclear export.

To ask whether the HIV-1 Rev-NES restores nuclear export when added to the C-terminus, we analyzed Upf3p-Triple-HA, Upf3p-Triple-HA-M10 (non-functional NES), and Upf3p-Triple-HA-Rev (functional NES) by immunofluorescence (Table 2, (6)). The nucleolar signal indicative of an export defect was detected in cells carrying the upf3-Triple-HA and upf3-Triple-HA-M10 alleles. However, the nucleolar signal was abolished in cells carrying the upf3-Triple-HA-Rev allele. This evidence supports the conclusion that the HIV-1 Rev-NES restores the ability of Upf3p-Triple to export from the nucleus to the cytoplasm. However, the restoration of export results in only a partial restoration of function in NMD as determined by the suppression and mRNA accumulation tests described above.

The NES-A region was shown previously to be important but not sufficient for the protein-protein interaction known to occur between Upf2p and Upf3p (7, 25). The ability of Upf3-Triple, Upf3-nes2, and Upf3-nes7 proteins to interact with Upf2p was determined using the yeast two-hybrid system. All three mutant proteins expressed with Upf2p failed to promote growth on selective media indicating a significant decrease in the strength of the interaction (Table 2, (7)).

To summarize, the triple alanine substitutions cause partial loss of function without the C-terminal HIV-1 Rev-NES and partial regain of function when the C-terminal NES is present. The upf3-nes2 and upf3-nes7 alleles also cause partial loss of function where upf3-nes2 is less impaired than upf3-Triple-HA and upf3-nes7. However, both alleles differ from upf3-Triple-HA in that a partial re-gain of protein function does not occur in the presence of a C-terminal NES. All three alleles produce a protein that is impaired for export and for the physical interaction with Upf2p. Overall, the results suggest either that the export of Upf3p occurs by binding to Upf2p or a specific export pathway for Upf3p is necessary to promote an interaction with Upf2p.

Recently, the crystal structures for human Upf3p and Upf2p were reported (26). The amino acid sequence within the NES-A region (residues 87-95 in S. cerevisiae) are 78% identical and 89% similar to human UPF3, suggested that the folded structure is likely to be similar in both organisms. The NES-A region lies within a three-dimensional structure that resembles an RNA recognition domain, but despite this, human Upf3p fails to bind to RNA. Instead, the NES-A region forms a β-sheet that is critical for the binding of Upf3p and Upf2p. Binding studies revealed that human Upf2p is capable of simultaneous binding to Upf3p and RNA, suggesting that a Upf2p/Upf3p complex may bind to RNA during export.

Our genetic analyses combined with the structural analyses reported in Kadlec et al. (26) support a model in which Upf3p exports from the nucleus bound to Upf2p. Amino acid substitutions that impair the export of Upf3p result from an impaired interaction between Upf3p and Upf2p. Thus, NES-A directs nuclear export not by serving as a binding site for an exportin but rather as a binding site for Upf2p. The addition of the HIV-1 Rev-NES promotes export of mutant Upf3 proteins but it fails to restore the activity of the NMD pathway because the essential Upf3p/Upf2p interaction remains impaired.

Adapting the dual selection to study other protein domains

The results of our genetic analyses provide a proof of principle for using a dual selection to analyze the function of protein domains. Modifications could be made to study functional domains in other proteins. The best systems would be those where the crystal structure of a protein is known so that the effects of different amino acid substitutions could be more readily interpreted. The following considerations should guide the design of a dual selection to analyze domains in other proteins:

(1) The domain to be studied must be self-contained, portable, and able to function out of context. The best way to test portability is to attach the domain of interest to a reporter. The C-terminus of a protein is likely the best position for attachment of a portable domain, but this would have to be tested empirically for any given system. Binding domains for DNA, RNA, protein, and small molecules are often self-contained and portable. However, active centers for enzymatic activities or other types of complex domains are often not portable since the active center is the result of protein folding rather than a contiguous amino acid sequence.

(2) Since the selection is conducted in vivo, it is most readily adapted to study the products of genes that are dispensable for growth. In our studies, UPF3 is not essential for growth hence mutations do not reduce viability (5). For essential genes where the mutations of interest are potentially lethal, a condition would be needed where mutants are viable. Such a condition would vary depending on the gene being studied.

(3) The phenotypic assays are critical in terms of the functions they monitor and the ease of scoring. At a minimum, the assays should include (a) a plate test for overall function of the protein that is amenable to mass screening (such as our assays for nonsense suppression), and (b) a specific assay for the function of the domain in which the mutations reside (such as the immunofluorescent localization assay we used for nuclear export and the two hybrid assay for protein-protein interactions).

Mutations that cause partial impairment of function can be difficult to distinguish from wild-type, as was the case for upf3-nes1, which is impaired only slightly (Table 1). Even for mutations causing greater impairment of function, suppression levels before and after recombination were not readily distinguished by examining the growth of colonies after replica-plating. We resolved this by resorting to more sensitive serial drop tests to assay growth rates (7), but drop tests precluded the ability to screen large numbers of candidate mutants. Despite the fact that our system fell short of the ideal for phenotypic screening as described in Fig. 3, the basic idea behind the dual screen for loss/regain of function could prove useful in analyzing mutations in domains of other proteins using different phenotypic assays.

Appendix

Protocols

Protocols

Protocols are written as performed for the sets of experiments described in the paper. General suggestions are given in brackets. To adapt the methods to study other genes, modifications of restriction sites, PCR conditions, choices of vectors, or other experimental parameters may be required.

Plasmid Construction

Insertion of loxP site after last codon and prior to termination codon of gene

1. Clone UPF3 into centromeric plasmid pRS314 between restriction sites SpeI and BstEII.

2. Removal of NotI site in the multiple cloning site of pRS314. {These steps can be eliminated if the vector containing the gene of interest does not contain a NotI site}.

1. Digest 5 μg of pLS17 with NotI (NEB) in a total volume of 50 μl at 37°C for hours.

2. Add 1μl of T4 DNA polymerase (NEB) and 100 μM of dNTPs at 16°C for 45 minutes to fill in 5' overhangs. {According to NEB, decreasing temperataure and incubation time to 12°C for 15 minutes reduces the occurrence of recessed ends caused by the 3' → 5' exonuclease activity of T4 DNA polymerase}.

3. Add 3μl of DNA, 4μl of T4 Ligase Buffer (Invitrogen), 13μl of H₂O and 1μl of T4 Ligase, mix, incubate for 5 hours at 16°C.

4. Transform 2μl into subcloning competent DH5α cells (Invitrogen) according to manufacturer’s instructions. Plate 50, 100 and 200μl of cell suspension following incubation at 37°C for 1 hour on LB plates containing 100 μg/ml ampicillin. Incubate at 37°C for 18 hours.

5. Pick individual colonies and grow in 1 - 2 ml of LB liquid media containing 100 μg/ml ampicillin.

6. Isolate DNA by miniprep using Qiagen kit according to the manufacturer’s instructions.

7. Screen colonies for removal of NotI site and addition of FseI site. {If plasmids without NotI are a small percentage of plasmids recovered, pool colonies isolated in Step d. Isolate the DNA from the pooled colonies, cut with NotI and transform into DH5α.}.

1. Insertion of NotI site using Chameleon double-stranded site-direected mutagenesis kit (Strategene).

1. Synthesize an HPLC-purified olgionucleotide containing a 5'-end phosphate to insert a NotI site (GCGGCCGC) following the last codon and prior to the termination codon of UPF3. Synthesize a second oligonucleotide that changes the unique restriction site within pRS314.

2. Insert a NotI site into UPF3 using the oligonucleoties described above following manufacturer’s instructions.

1. To insert loxP, synthesize complementary oligonucleotides that contain a loxP site (underlined in the sequence below), generate NotI overhangs, and maintain the reading frame after ligation into plasmid generated in Step 3. To aid further cloning the oligonucleotides destroy the NotI site upon ligation and generate new restriction sites ( AvrII , Cla I).

1. 5' - GGC CTA GGA ATA ACT TCG TAT AAT GTA TGC TAT ACG AAG TTA TAT CGA TAT

2. 5' - GGC CAT ATC GAT ATA ACT TCG TAT AGC ATA CAT TAT ACG AAG TTA TTC CTA

3. Combine 250 pmoles of each primer in a total volume of 100μl. Boil for 5 minutes and place immediately on ice.

4. Digest 1μg of vector generated in Step 3 with NotI in a total of 20μl. Clean-up digested vector using QIAquick PCR Purification Kit following manufacturer’s instructions.

5. Combine 2μl of digested vector, 1μl of annealed oligonucleotides diluted 1:10, 4 μl of T4 Ligase Buffer and 1μl T4 Ligase in 20μl total. Mix gently. Maintain at room temperature for 1 hour.

6. Digest 3μl of the ligation reaction with NotI for 2 hours at 37°C.

7. Transform 2.5μl of the Not1-digested ligation into DH5α according to manufacturer’s instructions.

8. Isolate DNA by miniprep using Qiagen kit following manufacturer’s instructions.

9. Screen colonies for removal of NotI site, the addition of AvrII and ClaI restriction sites and by sequence analysis.

10. Sequence plasmid DNA to determine direction of the loxP site. The directional loxP sites must be in the same orientation for excision of the intervening sequence to occur after recombination.

Insertion of 2^nd loxP site downstream of gene insertion

1. Synthesize complementary oligonucleotides that contain a loxP site (underlined in the sequence below) and generate KpnI overhangs (Note: KpnI was chosen because it is a unique restriction site within the vector only). To aid in the cloning of ADE2 between the loxP sites and the addition of the NES element, destroy the 3' KpnI restriction site and replace with an XhoI restriction site (Note: depending on your sequence, it might be necessary to use a different restriction site).

1. 5' - CAT AAC TTC GTA TAA TGT ATG CTA TAC GAA GTT ATC TCG AGC GTA C

2. 5' - GCT CGA GAT AAC TTC GTA TAG CAT ACA TTA TAC GAA GTT ATG GTA C

1. Follow steps 4c to 4h (omit Step 4f) as described above.

2. Screen colonies for addition of loxP site by PCR, restriction digest and sequence analysis.

Insertion of ADE2 between the loxP sites

1. Synthesize oligonucleotides to amplify ADE2 and include restriction sites for BstEII and KpnI. {We include six random nucleotides (N) at the end of the primer to increase efficiency of digestion of the PCR product by BstEII and KpnI.}

1. 5' - NNN NNN GGT TAC CCA CCA ACA TAA CAC TGA CAT C

2. 5' - NNN NNN GGT ACC CCT GAC GTA GCG CTA TCC

1. Set up a 50μl PCR reaction containing 1 – 2μl (0.2-0.4 μg) of S. cerevisiae genomic DNA, 1x Pfu Buffer, 50 pmoles primers, 200μM dNTPs. Begin cycling at 94°C for 3 minutes. After denaturation, pause cycle and add 1 μl of Pfu. Continue with 30 cycles of 94°C for 45 seconds, 50°C for 45 seconds, 72°C for 3 minutes followed by 72°C for 7 minutes. Check 10% of the PCR reaction on a 1% agarose gel to confirm size and purity of the amplified PCR product.

2. Add 50μl of TE(pH 8.0)-buffered-phenol:CHCl₃, vortex, centrifuge on max speed for 1 minute, remove aqueous layer, repeat. Add 50μl CHCl₃, vortex, centrifuge on max speed for 1 minute, remove aqueous layer. Add 5μl NaOAc, 200μl EtOH to aqueous layer, place on ice for 10 minutes. Centrifuge on max speed for 5 minutes, wash pellet with 70% EtOH. Resuspend pellet in 50μl H₂O. {Alternatively, clean up the remaining PCR reaction using QIAquick PCR Purification Kit following manufacturer’s instructions.}

3. Digest 1μg plasmid isolated in Step 7 and 4 μl (~0.2 μg) ADE2 PCR product islolated above with BstEII and KpnI. Clean-up digested vector and PCR product using QIAquick PCR Purification Kit following manufacturer’s instructions.

4. Combine 5μl of digested vector, 6μl ADE2 PCR product, 4μl of T4 Ligase Buffer and 1μl T4 Ligase in 20μl total. Maintain at room temperature for 1 hour.

5. Transform 2μl of the ligation reaction into DH5α according to the manufacturer’s instructions.

6. Screen colonies for addition of ADE2 by restriction digest.

Insertion of HIV-1 Rev NES 3' of 2^nd loxP site

1. Synthesize complementary oligonucleotides that contain the HIV-1 Rev NES sequence (underlined in sequence below) with overhangs compatible with XhoI, generate a NotI restriction site and maintain the reading frame after loxP recombination.

1. 5' – TCG AGT GCC ATT GCA ATT ACC ACC ATT GGA AAG ATT GAC TTT AGA TAG C

2. 5' – GGC CGC TAT CTA AAG TCA ATC TTT CCA ATG GTG GTA ATT GCA ATG GCA C

1. Follow steps 4c-4h (omit step 4f) as describe above.

2. Screen colonies for addition of domain by PCR, restriction digest and sequence analysis.

Insertion of the 3'-UTR of choice downstream of the domain of choice

1. Synthesize oligonucleotides to amplify the UPF3 stop codon and 3'-UTR and include restriction sites for NotI.

1. 5' - NNN NGC GGC CGC GTC GAC TAG TAA AAG CTC ATG GCT TCT TAT AT

2. 5' - NNN NGC GGC CGC GTA CCG TTT TTT AGT TCT ATG GTA ACC GTC TC

1. Set up a 50μl PCR reaction containing 1-2μl (0.2-0.4 μg) of S. cerevisiae genomic DNA, 1x Pfu Buffer, 50 pmoles primers, 200μM dNTPs. Begin cycling at 94°C for 3 minutes. After denaturation pause cycle and add 1μl of Pfu. Continue with 30 cycles of 94°C 45 seconds, 50°C 45 seconds, 72°C 2 minutes followed by 72°C for 3 minutes.

2. Run 10% of the reaction on a 1% gel to confirm the size. Clean-up the remaining PCR reaction using QIAquick PCR Purification Kit following manufacturer’s instructions. Elute PCR product in 30μl H₂O. {Alternatively, use the clean-up procedure described in step 10 to purify PCR fragment}.

3. Digest 15μl of the PCR product with NotI for 2 hours at 37°C. Clean-up digested PCR product using QIAquick PCR Purification Kit following manufacturer’s instructions. Elute the PCR product in 30μl H₂O.

4. Digest 1μg of the vector with 1 unit of NotI for 2 hours at 37°C. After digestion add 1 unit of calf intestinal alkaline phosphatase (CIP). Incubate at 37°C for 15 minutes. Clean-up digested plasmid using QIAquick PCR Purification Kit following manufacturer’s instructions. Elute the vector in 30μl H₂O.

5. Combine 4μl of digested vector, 12μl 3'-UTR PCR product, 4μl of T4 Ligase Buffer and 1μl T4 Ligase in 20μl total. Maintain at room temperature for 1 hour.

6. Transform 3μl of the ligation reaction into E. coli DH5α following manufacturer’s instructions.

7. Screen colonies for addition of 3’UTR by restriction digest and sequence analysis.

Mutagenesis/library construction

1. Synthesize mutagenic oligonucleotide inserting random nucleotides within the four codons of UPF3 NES-A (see Materials and Methods and Fig. 2 for additional details on NES-A mutagenesis).

1. 5' - CAA CCG AGA ATG AAG GAT TTA AG(A,C,G) (A,C,T,G)(A,C)G TT(A,C,G) (A,C,T,G)(A,C)A GA(A,C,G) (A,C,T,G)(A,C)(A,C,G) (A,C,T,G) (A,C)C CTC CAA ATT TGA CTG CAG ATG

1. Synthesize second oligonucleotide downstream of NES-A and unique restriction site SnaBI.

1. 5' - CCA GTT CCA ACC AGC TCT G

1. Set up a 50μl PCR reaction containing 0.05μg plasmid generated in Step 25, 1x Pfu Buffer, 50 pmoles primers, 200μM dNTP mix. Begin cycling at 94°C for 3 minutes. After denaturation pause cycle and add 1μl of Pfu. Continue with 30 cycles of 94°C 45 seconds, 47°C 45 seconds, 72°C 1 minute followed by 72°C for 3 minutes (PCR Round 1 in Fig. 2).

2. Run 10% of the PCR reaction on a 1% agarose gel to confirm PCR product. Clean-up the remaining PCR reaction using Qiagen QIAquick PCR Purification Kit following manufacturer’s instructions. Include an additional wash with 750μl of 35% guanidine hydrochloride prior to the Buffer PE wash step. Elute DNA in 50μl of H₂O. PCR fragment will be used as a megaprimer in the subsequent PCR reaction.

3. Synthesize T7 oligonucleotide upstream of NES-A and unique restriction site SacII.

1. 5' – TAA TAC GAC TCA CTA TAG GG

1. Set up a 50μl PCR reaction containing 0.05μg plasmid generated in Step 25, 1x Pfu Buffer, 50 pmoles oligonucleotide, 25 - 50μl of megaprimer, 200μM dNTP mix. Begin cycling conditions of 94°for 3 minutes. After denaturation pause cycle and add 1μl of Pfu. Continue with 30 cycles of 94°C 45 seconds, 47°C 45 seconds, 72°C 1 minute followed by 72°C for 3 minutes (PCR Round 2 in Fig. 2).

2. Run 10% of the PCR reaction on a 1% agarose gel to confirm PCR product. Clean-up the remaining PCR reaction using Qiagen QIAquick PCR Purification Kit following manufacturer’s instructions. Elute DNA in 50μl of H₂O.

3. Amplify PCR fragment using outside oligonucleotides from steps 2 and 5. Set up a 50μl PCR reaction containing 2μl PCR product, 1x Pfu Buffer, 50 pmoles oligonucleotides, 200μM dNTP mix. Begin cycling conditions of 94°C for 3 minutes. After denaturation pause cycle and add 1μl of Pfu. Continue with 30 cycles of 94°C 45 seconds, 60°C 45 seconds, 72°C 2 minutes 30 seconds followed by 72°C for 3 minutes.

4. Digest 5μg of PCR fragment containing the mutated sequence in 1x restriction enzyme buffer and 5 units of SacII and SnaBI in 100μl total volume at 37°C for 3 hours. Repeat with 5μg of plasmid generated above.

5. Run the digested products on a 1% agarose gel and gel purify the fragments using Qiagen QIAquick Gel Purification Kit following manufacturer’s instructions. Elute PCR product and digested vector in 50μl of H₂O.

6. Combine 4μl of digested vector, 4μl PCR product, 4μl of T4 Ligase Buffer and 1μl T4 Ligase in 20μl total. Maintain at 15°C overnight. As a control set up the same ligation reaction substituting H₂O for the PCR product.

7. Transform 2μl of ligation reaction in XL-10 Gold Ultracompetent Cells (Stratagene) following manufacturer’s instructions. Plate 200μl of the cell suspension onto each of 5 LB plates containing 100μg/ml ampicillin. Grow for 18 hours at 37°C.

8. Pool the cells on the plates by adding 1-2ml of LB containing 100μg/ml ampicillin to all the plates and gently removing the cells using a rounded-off end of a glass pipet.

9. Combine the cells from all plates and isolate plasmid DNA.

Transformation of yeast

To transform yeast by electroporation, use the Genepulser Electoporation apparatus from Biorad. Prepare cells according to (17) as follows:

1. Grow a fresh patch of the strain to be transformed on solid YEPD (yeast extract/dextrose) medium. Cells can be used for about one month.

2. Scrape cells from the patch with a toothpick and suspend in 20μl of 1M sorbitol plus 20mM HEPES.

3. Wash 2x in 1 mM sorbitol plus 20mM HEPES.

4. Pellet and resuspend in a suitable volume of 1M sorbitol plus 20mM HEPES. Each transformation requires a 40 μl cell suspension.

5. Add up to 1 μg of plasmid DNA in no more than 5μl to each transformation and transfer to a 0.2 cm electroporation cuvette.

6. Electroporate with the Biorad Genepulser set to 1.4 kV, 200Ω, 25μF.

7. Immediately add 0.25 ml of 1M sorbitol. Remove from the electroporation cuvette and plate on selective medium.