Sequence analysis of Rgs1
The Sz. pombe sequencing project identified Rgs1 as a hypothetical protein with similarity to members of the RGS family of proteins (Fig. 1). Family members share a ≈ 130 residue domain that is similar at the primary level and is predicted to fold into highly conserved secondary and tertiary structures (the RGS fold) (Tesmer et al., 1997). Structural analysis of the RGS fold of rat RGS4 revealed nine α-helices that fold into a small terminal subdomain and a bundle subdomain that includes most of the residues that interact with the Gα subunit (Tesmer et al., 1997). Rgs1 can be fitted to this structure by introducing insertions in regions not thought to be involved in the folding of the RGS domain (Fig. 1A; Tesmer et al., 1997). Such insertions are found in several RGS proteins from lower eukaryotes and in at least one mammalian RGS protein (Fig. 1B), but their functional significance remains unknown.
Figure 1. . Sequence analysis of Rgs1. A. Predicted amino acid sequence of Rgs1 from Sz. pombe. The rgs1 gene was identified as part of the Sz. pombe sequencing project (accession number Q09777). Residues conserved among the RGS family members are indicated by * and ^; residues marked by * are believed to make direct contact with the Gα subunit and those marked by ^ form the hydrophobic core of the RGS domain (Tesmer et al., 1997). Underlining and dashed lines indicate insertions in the Sz. pombe sequence that maximize similarity to the RGS domain of mammalian RGS4 (Tesmer et al., 1997). B. Comparison of the RGS domain of Rgs1 with other members of the family. Pairwise comparisons performed with the bestfit programme from the GCG package (Genetics Computer Group) using a Gap creation penalty of 3.0, a Gap extension penalty of 0.1, and the default amino acid comparison table. RGS domains were identified by comparison with that determined for rat RGS4 (Tesmer et al., 1997). Thus, the RGS domain of rat RGS4 includes residues 51–178 while that for Sz. pombe Rgs1 is composed of four sections covering residues 311–322, 345–383, 394–437 and 450–480. The following sequences were used for the comparison: S. cerevisiae SST2 (accession number P11972); A. nidulans FlbA (P38093); C. elegans EGL-10 (P49809); human GAIP (P49795); human RGS1(Q08116); human RGS2 (P41220); human RGS3 (P49796); human RGS4 (P49798); human RGS5 (O015539); human RGS6 (O75576); human RGS7 (P49802); rat RGS8 (D1024561); human RGS9 (O75916); human RGS10 (O43665); human RGS11 (O75883); human RGS12 (O14924); human RGS13 (O14921); rat RGS14 (O08773); human RGS16 (O15492); bovine RET-RGS1 (G1895060); human RGSZ1 (O76081); human axin (O15169); mouse D-AKAP2 (O88845); C. elegans C05B5.7 (P34295); C. elegans C29H12.3 (Q18312); C. elegans F16H9.1 (P49808).
Download figure to PowerPoint
Rgs1 contains a long N-terminal extension that we believe is either involved in regulating the RGS domain or required for additional activities that remain to be characterized (J. Davey, P. Watson and C. Pateman, in preparation). Although isolated RGS domains can function as GAPs both in vitro and in vivo (Faurobert and Hurley, 1997; Popov et al., 1997), there is considerable evidence to suggest that sequences outside these conserved domains are required for correct biological activity. Truncation of the N-terminal domains of several RGS proteins, including SST2 (Dohlman et al., 1996), eliminates their activity (Chen et al., 1997; Srinivasa et al., 1998). The mapping of a gain-of-function allele within this region of SST2 further illustrates its functional importance (Dohlman et al., 1995). No catalytic activity has been demonstrated for these additional sequences, and most appear to be involved in facilitating interactions with other proteins or with cell membranes. For example, both axin (Itoh et al., 1998), D-AKAP2 (Huang et al., 1997) and GAIP (De Vries et al., 1998) bind to target proteins through regions outside their RGS domains, and other family members contain motifs known to promote protein–protein interactions (Snow et al., 1998a). Targeting to the membrane could provide additional regulation. Several RGS proteins become palmitoylated (De Vries et al., 1996; Srinivasa et al., 1998), whereas others can interact directly with the Gβ subunits (Snow et al., 1998b) — at least one is thought to contain a transmembrane domain (Faurobert and Hurley, 1997). Establishing the function(s) of the N-terminal domain of the Sz. pombe Rgs1 will be important for a complete understanding of the protein.
Constructing a mat1-Pm>lacZ reporter
The mat1-Pm gene encodes a protein required for entry into meiosis (Kelly et al., 1988). It is only expressed in P-cells and the low level of expression observed during mitotic growth is increased dramatically by pheromone stimulation (Nielsen et al., 1992; Aono et al., 1994; Davey and Nielsen, 1994). To identify and characterize factors involved in regulating the pheromone response, we constructed a P-cell carrying a mat1-Pm>lacZ reporter gene (JY464) (Fig. 2). JY464 has several differences to a mat1-Pm>lacZ reporter strain used previously to assay the response to M-factor (Aono et al., 1994). First, the mat1-Pm>lacZ reporter construct is integrated into the chromosomal mat1-Pm locus so that each individual gives a more uniform response when stimulated, and we avoid the variability that can complicate the use of plasmid-borne reporters (Aono et al., 1994). Second, our strain lacks the mat2 and mat3 loci and is therefore unable to switch from the M-type information encoded at the mat1 locus (Klar and Miglio, 1986). Finally, the strain lacks the cyr1 gene that encodes adenylate cyclase (Yamawaki-Kataoka et al., 1989; Young et al., 1989). Nutritional regulation of the pheromone response pathway complicates mating-related studies in Sz. pombe as cells are derepressed for mating during mitotic growth. The mechanism by which cells detect nutritional status is unclear, but mating functions are inhibited by high intracellular levels of cAMP. Mutants defective in cyr1 respond to pheromones during mitotic growth (Davey and Nielsen, 1994; Imai and Yamamoto, 1994).
Figure 2. . Constructing the mat1-Pm>lacZ reporter strains. The mat1-Pm ORF was first replaced with a 1.8 kb Sz. pombe ura4+ cassette (Grimm et al., 1988). The complete mat1-Pm locus was amplified by PCR using the sense primer JO531 (gggcaTATGCGCTCTAACTTGG; lower-case letters are not complementary to mat1-Pm, but the oligonucleotide includes an emboldened NdeI site so that digestion leaves ends that are fully homologous to the chromosomal sequence) and the antisense primer JO532 (gggcaTATGAAAAACAAAAACCGAATG; cleavage at the NdeI site again leaves ends that are fully homologous to the chromosome). The resulting PCR product was cloned into the NdeI site of pGEM5Zf (Promega) and the clone used as template for PCR with JO748 (AATGAATTGCTTAAAATAAAAC; an antisense primer complementary to a region immediately upstream of the ATG initiator codon for mat1-Pm) and JO747 (TGAATTATGTTAGCTTAG; a sense primer that includes the TGA stop codon and immediate downstream region of mat1-Pm). This product (corresponding to the upstream and downstream regions of mat1-Pm cloned into the NdeI site of pGEM5Zf) was ligated to either the ura4+ cassette (to generate JD699, pGEM5Zf containing a construct suitable for disruption of mat1-Pm) or a PCR product corresponding to the lacZ ORF (to generate JD806, pGEM5Zf containing the mat1-Pm>lacZ reporter construct). The lacZ ORF was prepared by amplification using the sense primer JO660 (ATGCAGCTGGCACGACAGGTTTCCCGAC; includes the ATG initiator codon and the next 25 bases of the lacZ ORF) and the antisense primer JO706 [TTTTTGACACCAGACCAACTGGTAATGGTAGCGACCGGCGCTCAGCTGGAATTCCGCCGATACTGACGGGCTCCAGGAG TCGTCGCCACCAATCCCCATgTGGAAACCGTCG; complementary to the 3′ end of the lacZ ORF except for a single base change (a T-to-c change when considered in the sense direction) that destroys the NdeI site without changing the coding potential of lacZ [both the original CAT and the newly created CAc encode histidine]. JY357 (a mating stable P-cell lacking cyr1) was transformed with an NdeI fragment corresponding to the mat1-Pm::ura4+ construct (isolated from JD699), and stable Ura+ transformants were initially screened by PCR and replacement of the mat1-Pm locus was confirmed by Southern blot. A correct mat1-Pm disruptant (JY412) was then transformed with an NdeI fragment corresponding to the mat1-Pm>lacZ reporter construct (isolated from JD806). Stable Ura-transformants (selected by their ability to grow in the presence of 5′-fluoroorotic acid; Boeke et al., 1987) were screened by PCR and homologous integration of the reporter construct at the mat1-Pm locus confirmed by Southern blot (JY464).
Download figure to PowerPoint
Exposing the JY464 reporter strain to M-factor results in a time- and pheromone-dependent production of β-galactosidase that can be detected by a colony colour assay or quantified using a liquid assay (Fig. 3). Low-level production of β-galactosidase in the absence of M-factor increased after about 8 h and reached a plateau after 16 h. Because of differences outlined in the previous section, our results cannot be compared with an earlier study that also used a mat1-Pm>lacZ reporter construct (Aono et al., 1994). In some respects, our strain shows greater similarity to one in which the pheromone-dependent induction of mat1-Pm was monitored using Northern blot analysis (Davey and Nielsen, 1994); both strains are cyr1− and both studies effectively monitor the activity of the mat1-Pm promoter at its correct chromosomal locus. However, a direct comparison is still difficult as the first study measured transcript levels, whereas the current work monitors the activity of a reporter enzyme — a process that requires transcription, translation and assembly into an active conformation. These differences probably explain why the pheromone-dependent increase in mat1-Pm transcripts occurred about 6 h before any increase in β-galactosidase activity (compare Fig. 3 with results in Davey and Nielsen, 1994). Furthermore, differences in the stabilities of the β-galactosidase protein and the mat1-Pm transcript presumably explain why β-galactosidase activity remained high (Fig. 3), even though the level of mat1-Pm transcripts decreased rapidly following a peak at 5 h (Davey and Nielsen, 1994). Monitoring the level of mat1-Pm>lacZ transcripts rather than the activity of β-galactosidase could avoid some of these problems, but differences in the stability of the two transcripts would still be likely to complicate any comparison between the two studies.
Figure 3. . Pheromone-dependent production of β-galactosidase. The mat1-Pm>lacZ reporter strain (JY464) was exposed to M-factor and β-galactosidase production assayed using o-nitrophenyl-β-d-galactoside as substrate. A. Cells were exposed to M-factor at either 0 or 10 units ml−1 and assayed at hourly intervals. B. Cells were exposed to various concentrations of M-factor for 16 h before being assayed. β-Galactosidase activity was determined colorimetrically and expressed as the ratio of o-nitrophenol product to cell density (see Experimental procedures). The values shown are the means of triplicate determinations.
Download figure to PowerPoint
Pheromone–dependent transcription of the mat1-Pm>lacZ reporter
Strains with and without rgs1 were assayed for β-galactosidase activity following exposure to M-factor (Fig. 5). Loss of rgs1 did not affect the maximum level of β-galactosidase activity, but did increase production in the absence of pheromone, and increased the sensitivity of the cells to pheromone stimulation. These changes are due to the loss of rgs1 as they were overcome by expression of a plasmid-borne rgs1 gene. Furthermore, expression of SST2 (encodes an RGS protein from the budding yeast S. cerevisiae) also overcame the loss of rgs1. Our results suggest that Rgs1 is a negative regulator of the pheromone response pathway and, consistent with predictions from sequence analysis, that it probably functions as an RGS protein.
Figure 5. . The influence of Rgs1 on pheromone-dependent transcription. Pheromone-dependent transcription of a mat1-Pm>lacZ reporter construct was assayed by monitoring β-galactosidase activity in cells with different combinations of RGS proteins. Control strains possessing the wild-type rgs1 gene (JY464, rgs1+), or strains lacking rgs1 (JY508, Δrgs1), were transformed with the expression vector pREP3X (JY572 and JY584) or with constructs containing rgs1 from Sz. pombe (JY580 and JY592) or SST2 from S. cerevisiae (JY576 and JY588). Cells were grown to mid-exponential phase and treated with M-factor for 16 h before being harvested. β-Galactosidase activity was determined colorimetrically and expressed as the ratio of o-nitrophenol product to cell density (see Experimental procedures). The values shown are the means of triplicate determinations.
Download figure to PowerPoint
The pheromone-independent expression of mat1-Pm involves the same basic elements as those responsible for pheromone-induced transcription (Aono et al., 1994), and it is perhaps not surprising that Rgs1 plays a role in suppressing signalling in the absence of pheromone. Spontaneous dissociation of the receptor-coupled G protein, for example, would activate expression of mat1-Pm>lacZ, and this would be increased in the absence of Rgs1. In contrast, overexpression of rgs1 (or SST2) did not reduce pheromone-independent signalling in rgs1+ cells [Fig. 5, compare JY572 (rgs1+) with JY580 (rgs1+ and Rgs1) or JY576 (rgs1+ and SST2)], indicating that at least some of this signalling is independent of Rgs1. Very similar results were reported for SST2 in S. cerevisiae (Dohlman et al., 1996).
RGS proteins do not necessarily decrease signalling intensity, and loss of rgs1 did not affect the maximum level of pheromone-dependent β-galactosidase production (Figs 3 and 5). It did, however, make the cells more sensitive to low levels of M-factor. Pheromone-dependent activity in rgs1+ cells required at least 0.1 U ml−1 M-factor whereas the Δrgs1 strain was some 100-fold more sensitive to stimulation and responded to pheromone at 0.001 U ml−1 [Fig. 5, compare JY572 (rgs1+) with JY584 (Δrgs1)]. An alternative measure of a cell's sensitivity is given by the EC50, the concentration of ligand required to induce a half-maximal response. This is not easily calculated for some of the strains but, allowing for differences in the pheromone-independent production of β-galactosidase, the EC50 for the Δrgs1 strain (JY584) was ≈ 0.04 U ml−1 and that for the rgs1+ strain (JY572) ≈ 1.5 U ml−1. This suggests that the Δrgs1 strain is about 40-fold more sensitive than the rgs1+ strain.
Expression of rgs1 from the nmt1 promoter (Maundrell, 1990; 1993) overcame the increased pheromone sensitivity of rgs1-strains. Indeed, the cells were less sensitive than wild-type (rgs1+) strains, and the EC50 for JY592 (Δrgs1 and Rgs1) was ≈ 6 U ml−1 [compared with ≈ 1.5 U ml−1 for JY572 (rgs1+)]. The level of Rgs1 protein appears to be limiting even in wild-type (rgs1+) cells, as expressing rgs1 from the powerful nmt1 promoter further reduced the sensitivity of these cells to stimulation [EC50 for JY580 (rgs1+ and Rgs1) is ≈ 6 U ml−1]. Increasing the level of Rgs1 did not affect the maximum level of β-galactosidase activity. Further work is required to determine whether the removal of Gα-GTP is indeed the rate-limiting step in the recovery process.
SST2 partly overcame the loss of rgs1, increasing the EC50 from ≈ 0.04 U ml−1 for JY584 (Δrgs1) to ≈ 0.7 U ml−1 for JY588 (Δrgs1 and SST2). Such a result is not unexpected given the sequence similarity between the Sz. pombe Gα subunit, encoded by the gpa1 gene, (Obara et al., 1991) and its S. cerevisiae counterpart (GPA1) (Miyajima et al., 1987; Nakafuku et al., 1987). Indeed, many RGS proteins have relatively broad specificities and can act as GAPs for several Gα subunits. Mammalian RGS4, for example, stimulates GTP hydrolysis for all Gα subtypes except Gs (for which no RGS protein has yet been identified) (Berman et al., 1996; Hepler et al., 1997; Huang et al., 1997) and can partly suppress the loss of SST2 in S. cerevisiae (Druey et al., 1996). Several other mammalian RGS proteins can also complement the loss of SST2 (Druey et al., 1996; Chen et al., 1997).
There are several reasons that could explain the inability of SST2 to rescue fully the loss of rgs1, and also its lack of effect when expressed in an rgs1+ strain [there was little difference between JY572 (rgs1+) and JY576 (rgs1+ and SST2)]. Differences in the level of expression of rgs1 and SST2 from the nmt1 promoter, for example, or differences in the stabilities of the two proteins or in their respective GAP activities towards Gpa1 could all contribute to the effect observed. Alternatively, the differences could be due to Sz. pombe being able to regulate the activity of Rgs1 but not SST2. It is also possible that Rgs1 is not only a GAP for Gpa1, but that it plays additional roles in regulating the pheromone response, roles that cannot be performed by SST2. These possibilities are discussed further in a later section.
We next investigated the involvement of Rgs1 in events that occur after the induction of mat1-Pm. Shmoo formation, the pheromone-dependent elongation of a responding cell, can be quantified using an assay that monitors the increase in cell volume following exposure to pheromone (Davey, 1992). Although a non-synchronous culture contains cells of various sizes, the increase in cell volume extends to the median cell volume of the culture. This is the volume that divides the distribution into two equal groups such that 50% of the cells are smaller than the median volume and 50% of the cells are larger than the median volume. Cells are exposed to pheromone and the size distribution of the responding culture is monitored using a Coulter Channelyser. The percentage of cells that are larger than the median volume of a culture not exposed to pheromone is then calculated, and subtracting 50 from this value provides the ‘shmoo response’. An additional feature of the assay is that it is performed under conditions of nitrogen starvation, and the increase in volume is irreversible because the cells are unable to divide and recover their original size. The auxotrophic requirements and the lack of functional cyr1 and mat1-Pm genes precludes the use of the original mat1-Pm>lacZ reporter strain in the shmoo assay, and we therefore disrupted the rgs1 gene in a more appropriate strain (JY383, to generate JY471) using the strategy described in Fig. 4. Homologous integration of the disrupted construct was confirmed using Southern blot analysis and the results of the cell volume assay are shown in Fig. 6.
Figure 6. . The influence of Rgs1 on shmoo formation. The pheromone-dependent increase in cell volume was assayed in cells with different combinations of RGS proteins. Control strains possessing the wild-type rgs1 gene (JY330, rgs1+) or strains lacking rgs1 (JY471, Δrgs1) were transformed with the expression vector pREP3X (JY510 and JY478) or with constructs containing rgs1 from Sz. pombe (JY527 and JY535) or SST2 from S. cerevisiae (JY514 and JY518). Growing cells were used to inoculate SSL at 1 × 106 cells ml−1 and cultured for 24 h at 29°C before being used to inoculate SSL containing appropriate levels of M-factor. The increase in cell volume was monitored using a Coulter Channelyser after incubation at 29°C for 48 h (Davey, 1992). Calculation of the ‘Shmoo response’ value is described in the text.
Download figure to PowerPoint
Loss of rgs1 did not affect the maximum level of shmoo response but increased the sensitivity of cells to pheromone. The EC50 for JY478 (Δrgs1) was ≈ 0.04 U ml−1 and that for JY510 (rgs1+) was ≈ 0.8 U ml−1, an increase in sensitivity of about 20-fold. Overexpression of rgs1 in either rgs1+ or Δrgs1 strains had little effect on the maximum level of response but increased the EC50 to ≈ 4 U ml−1. This is consistent with the results of the β-galactosidase assay, and further supports the suggestion that the amount of Rgs1 in wild-type strains is limiting with respect to its ability to regulate the pheromone response.
As predicted from the mat1-Pm>lacZ results, expressing SST2 decreased the sensitivity of both rgs1+ and Δrgs1 strains, although not quite as effectively as rgs1. Unexpectedly, however, SST2 limited the maximum shmoo response attained in the assay: increasing the M-factor beyond 10 U ml−1 did not increase the shmoo response above 16%. A 16% shmoo response can be monitored easily using the Coulter Channelyser; however, this is too small an increase for differentiation between responding cells and control cells under microscopic examination. Rgs1 is not required for the formation of highly elongated shmoos. [JY478 (Δrgs1) forms such elongated shmoos, however it does so at pheromone concentrations lower than those required by rgs1+ strains.] Therefore, the reduced elongation of strains overexpressing SST2, regardless of whether they are rgs1+ or Δrgs1, suggests that SST2 partially inhibits shmoo formation. We are currently investigating whether this inhibition is due to the inability of Sz. pombe to correctly regulate SST2 activity.
The effect of Rgs1 on conjugation and sporulation was assessed using a quantitative mating assay (Fig. 7). Mating was completely inhibited by loss of rgs1, but could be recovered (to 80% of the control level) by expression of rgs1 from the nmt1 promoter. Expression of SST2 failed to rescue the mating defect of a Δrgs1 strain but did not inhibit mating of an rgs1+ strain.
Figure 7. . The influence of Rgs1 on mating. Strains containing different combinations of RGS proteins were tested for their ability to mate with an M-strain (JY291). P-strains possessing the wild-type rgs1 gene (JY330, rgs1+) or strains lacking rgs1 (JY471, Δrgs1) were transformed with the expression vector pREP3X (JY510 and JY478) or with constructs containing rgs1 from Sz. pombe (JY527 and JY535) or SST2 from S. cerevisiae (JY514 and JY518). All strains were grown to mid-exponential phase in minimal medium, washed and resuspended in distilled water at 1 × 106 cells ml−1. Equal volumes of the two strains to be tested were mixed and 10 μl spotted onto SSA plates. After 4 days at 29°C, the plates were exposed to iodine vapour to stain spores dark brown and provide a qualitative measure of mating (Egel et al., 1994). To provide a more quantitative comparison, mating mixtures were resuspended in water and the numbers of spores and vegetative cells determined by light microscopy. Under the conditions used, rgs1+ cells (JY510) have a mating efficiency of ≈ 35%.
Download figure to PowerPoint
The loss of mating ability of Δrgs1 strains contrasts with the residual mating ability of S. cerevisiae strains lacking SST2 (3–10% of wild-type efficiency; Chan and Otte, 1982). This may reflect a difference in the roles of the two proteins or could simply be due to differences in the efficiencies with which the two yeast initiate mating. In natural cultures, Sz. pombe exists primarily as a haploid organism whereas S. cerevisiae generally exists as a diploid, and their promiscuity may reflect these preferences. The essential requirement for Rgs1 in the mating process is yet to be defined, but our earlier results suggest that it is likely to be a relatively late event in the process as loss of rgs1 prevents neither pheromone-dependent transcription nor shmoo formation. Examination of the mating mixtures involving Δrgs1 strains reveals that the cells fail to agglutinate with an appropriate partner (not shown); it will be interesting to discover whether the mating-specific agglutinins are expressed and whether they are targeted to the leading tip of the elongated shmoo.
Expression of SST2 in an rgs1+ strain did not inhibit mating, despite the fact that it limited shmoo formation in the same cells (see Fig. 6). The shmoos of reduced elongation, produced in the presence of SST2, are able to mate. Indeed, examination of a standard mating mixture of M-cells and P-cells indicates that the majority of cells in the mixture do not become highly elongated. Only a few cells, possibly those that fail to identify a mating partner, develop into the highly elongated shmoos observed in the cell volume assay.