High-level expression of mammalian G-protein-coupled receptors (GPCRs) is a necessary step toward biophysical characterization and high-resolution structure determination. Even though many heterologous expression systems have been used to express mammalian GPCRs at high levels, many receptors are improperly trafficked or are inactive in these systems. En route to engineering a robust microbial host for GPCR expression, we have investigated the expression of 12 GPCRs in the yeast Saccharomyces cerevisiae, where all receptors are expressed at the mg/L scale. However, only the human adenosine A2a (hA2aR) receptor is active for ligand-binding and located primarily at the plasma membrane, whereas other tested GPCRs are mainly retained within the cell. Selective receptors associate with BiP, an ER-resident chaperone, and activated the unfolded protein response (UPR) pathway, which suggests that a pool of receptors may be folded incorrectly. Leader sequence cleavage of the expressed receptors was complete for the hA2aR, as expected, and partially cleaved for hA2bR, hCCR5R, and hD2LR. Ligand-binding assays conducted on the adenosine family (hA1R, hA2aR, hA2bR, and hA3R) of receptors show that hA2aR and hA2bR, the only adenosine receptors that demonstrate leader sequence processing, display activity. Taken together, these studies point to translocation as a critical limiting step in the production of active mammalian GPCRs in S. cerevisiae.
The G-protein-coupled receptors (GPCRs) are seven transmembrane-spanning proteins whose roles in physiological functions are diverse and widespread.1 GPCRs participate in cell communication and signaling in a variety of organisms, and thus are implicated in many diseases. Rational drug design for GPCRs has proven challenging because of the absence of detailed structural information available for the majority of these proteins. Notable advances in crystallization strategies for these membrane proteins have led to the determination of three high-resolution GPCR structures, the β12 and β2 adrenergic3 and adenosine A2a receptors,4 which have been solved recently. Together with the structure of rhodopsin, which was crystallized 9 years ago,5 these four structures now serve as homology models for the nearly 1000-member GPCR superfamily. However, all of these structures display distinct differences from each other,2–4 such as differences in helix packing and substrate-binding pockets, and from predicted structures based on rhodopsin,5 further reinforcing the need to obtain high-resolution structural information for other GPCRs on an individual basis.6, 7
As almost all human GPCRs do not exist naturally in high abundance, heterologous expression systems are required to achieve sufficient protein yields for structural characterization, generally at the mg/L scale or greater. Bacterial, yeast, and insect hosts have successfully been implemented for high-level expression of soluble proteins, and similar approaches have been applied toward the expression of membrane proteins.7–10 Although all of these systems have proven useful for expression of some heterologous membrane proteins, each host has advantages and disadvantages associated with its use. For instance, microbial hosts such as E. coli and yeast offer well-understood genetics, low cost of culture, and relatively easy scale up. Insect and mammalian cells can perform many more complicated protein processing and post-translational modifications that may be necessary for proper function, but prove costly and time-intensive. However, even with several host systems available, heterologous expression has not yet systematically allowed for high-level expression of any given GPCR of interest, and typically relies on trial-and-error methods.7–10
Even though all GPCRs share a commonality in their seven transmembrane domain segments and in their ability to couple to trimeric G-proteins, they also display great diversity in their overall function, ligand preference, tissue location, and physiological prevalence.11 Furthermore, significant differences exist in how proteins are expressed and processed in various heterologous systems, which may have a direct impact upon folding and activity of heterologously expressed GPCRs.9
Given its ease of genetic manipulation, rapid growth, and eukaryotic secretory pathway, yeast are an attractive host system for the development of a robust GPCR expression system. Yeast have been successfully used for the heterologous expression of membrane proteins, specifically GPCRs.12, 13 Recent expression of the human adenosine A2a receptor (hA2aR) in S. cerevisiae has yielded active protein at greater than 10 mg/L of culture, which has facilitated its purification14 and biophysical characterization (O'Malley, Naranjo, Lazarova, and Robinson, manuscript in preparation). However, under identical culture and expression conditions, the human neurokinin NK1 (hNK1) receptor fails to traffic to the plasma membrane in this system and does not display ligand-binding activity.15 Instead, these receptors are trapped in punctuate structures that are present just below the plasma membrane.15 Other studies have also cited improper trafficking of recombinant membrane proteins in yeast,16, 17 which suggests that differences between the native mammalian secretory pathway and the yeast secretory pathway can adversely affect the production of proteins in this non-native system. At this point, little information has been published regarding the factors that govern folding and cellular trafficking of heterologously expressed GPCRs in yeast. By understanding the bottlenecks in plasma membrane targeting, we can perhaps employ cellular engineering approaches to enable high-level expression of active GPCRs, to create a more robust and desirable host.
In this work, we expressed 12 GPCRs from the rhodopsin family of receptors in the yeast Saccharomyces cerevisiae, to determine limitations associated with GPCR expression in this system. The GPCRs chosen for this screen possess different predicted post-translational modifications, couple to several different trimeric G-proteins, and encompass a range of primary sequence homology to the actively expressed human adenosine A2a receptor. Our results indicate that problems with GPCR folding and trafficking begin at the point of translocation within the ER membrane, which leads to the activation of downstream cellular stress responses.
GPCR expression levels and membrane integration
The presence of a C-terminal GFP tag on expressed GPCRs made possible the direct comparison of expression levels among different strains. This technique has been successfully applied toward the estimation of A2aR expression levels in S. cerevisiae18 and was further corroborated by other methods.18 Whole-cell fluorescence measurements indicate most GPCRs were readily expressed at high levels within yeast, based on total GFP fluorescence (Table I). We did not observe any degradation fragments via SDS-PAGE analysis followed by in-gel fluorescence for early time-points (1–5 hrs). For 24-hour time points, the degradation was between 0–15%, depending on the receptor identity. However, within the error of both measurements, we believe total fluorescence gives a reasonable value on total protein yields.
Table I. Summary of Data Obtained for Heterologous Expression of GFP-Tagged GPCRs in S. cerevisiae Including Relative Expression Yields, Cellular Localization, Leader Sequence Processing, and Glycosylation
Expression Level (thousands of molecules per cell)
Estimated expression yielda (milligrams per liter of culture)
Extent of pre-pro cleavage
ND, not determined.
Determined by extrapolating fluorescence levels and OD600s from 5 mL cultures, verified from direct scale up to 200 mL cultures.
Confirmed by N-terminal sequencing. Note that for hNK1R, data are for N-terminal sequencing alone, as problems that arose in sample preparation precluded analysis via gel mobility assay.
Overall expression yields were somewhat variable among the strains—this results from a combination of potential differences in integration number for the pITy vector, as well as differences in final cell culture densities. However, all GPCRs tested had predicted yields greater than 1 mg/L of culture with the exception of hFSHR-GFP-His10, at ∼0.7 mg/L. To verify that these membrane proteins were embedded into the lipid bilayer, an integral membrane protein assay19 was applied for GPCRs expressed in this system. Membrane fractions were treated with 0.1M NaCO3 (pH 11.0) and 2M urea, to dissociate peripherally associated proteins from membranes, or 1% SDS, which is capable of solubilizing integral membrane proteins from a membrane environment. If GPCRs expressed in this system exist as integral membrane proteins, only treatment of membrane preparations with SDS would liberate the protein from the membrane. In applying this assay toward the 12 GPCRs expressed in this system, we find that only treatment with SDS releases these proteins from the membrane environment, as shown for hA2aR-GFP-His10 and hA1R-GFP-His10 (Fig. 1). Thus, each mammalian GPCR expressed in this system was found within a membrane environment.
GPCR trafficking in S. cerevisiae
With the exception of the human A2a receptor [Fig. 2(A–C)], the GPCRs tested in this study appear predominantly within the cell. These receptors are mainly retained internally as determined through confocal microscopy of GFP-tagged receptors [Fig. 2(D,E,G,H,J,K)]. It has been previously observed that heterologous expression of NK1, D2s, and neuromedin U GPCRs in yeast systems led to protein retention within subcellular compartments.15–17
Early time points after induction of expression (∼5 h postinduction) were examined to investigate whether the localization in cell populations was different early in expression. As expected, at 5-h postinduction hA2aR-GFP appears to uniformly label the plasma membrane [Fig. 2(A)]. Heterologous expression of other GPCRs in S. cerevisiae led to two typical appearances, which are apparent in the images of hA1R, hA2bR, and hA3R [Fig. 2(D–L)]. In many individual cells, receptors accumulate at the cell periphery. Additionally, larger concentrations of the GPCR-GFP fusions accumulated more centrally within the cell. Each of the 11 GPCR-GFP fusions showed a combination of these patterns (summarized in Table I; confocal data not shown for hD2LR, hCB1R, hCB2R, hFSHR, hNK1R, hNK2R, hCCR5R, and hCXCR4R). However, from confocal imaging alone, we cannot rule out the possibility that a fraction of expressed receptors reached the plasma membrane.
Each GPCR construct contains an N-terminal pre-pro leader sequence to facilitate protein targeting to the ER. The presence of particular hydrophobic leader sequences strongly favors cotranslational insertion of membrane helices.20 However, to ensure that this leader sequence did not negatively impact cellular trafficking of mammalian GPCRs in S. cerevisiae, cells expressing GPCR-GFP constructs lacking this pre-pro leader were examined. In many cases, the absence of the pre-pro sequence resulted in decreased expression yields (determined by whole-cell GFP). The hA2aR and hA1R [Fig. 2(C,I)]. The hA2a receptor appeared plasma membrane localized, and most other receptors displayed a similar intracellular localization to that observed when the leader sequence was present [Fig. 2(C,I,L)]. However, in the expression of the hA2bR-GFP-His10 lacking the pre-pro leader sequence [Fig. 2(F)], more fluorescence is observed at the cell periphery when compared with those expressed with the leader sequence [Fig. 2(D,E)].
Cellular stress responses are activated during heterologous receptor expression
We investigated the UPR activation within the endoplasmic reticulum (ER) of S. cerevisiae during heterologous GPCR expression, as this organelle is implicated in protein retention within the secretory pathway. The UPR pathway is activated when the folding capacity of the ER is exceeded by the presence of unfolded or misfolded protein within the organelle.21–23 UPR activation promotes expression of numerous ER-resident chaperones via transcriptional upregulation, to reduce aggregation and promote secretion.24 Recently, it has been reported that overexpression of the P2 transporter in S. cerevisiae, a membrane protein, can trigger the UPR.25
In these studies, we used a UPRE-GFP reporter plasmid to monitor the activation of the UPR.26, 27 Activation of the UPR in cells expressing recombinant non-GFP-tagged GPCRs was monitored over time by measuring whole-cell fluorescence. We find that expression of most GPCRs activated the UPR pathway after 24 h of expression in S. cerevisiae [Fig. 3(A, solid bars)], relative to a BJ5464 parental control. UPR activation monitored via GFP fluorescence of cells producing hA1R, hA2bR, and hA3R was observed to increase over time [Fig. 3(B)]. In contrast, GFP fluorescence did not increase over the same time course during hA2aR expression, similar to background GFP levels observed for the parental BJ5464 strain [Fig. 3(B)]. The fluorescence signal due to UPR activation in strains expressing hA1R, hA2bR, and hA3R leveled off ∼15 h after expression was induced and did not decrease up to 24 h past expression induction, likely because of low GFP turnover in the cell. Similar increases over time in UPR activation (monitored via GFP fluorescence) were observed in the other eight GPCR-expressing strains (data not shown).
To ensure that the absence of UPR activation in cells expressing hA2aR was not due to plasmid loss, these cells were incubated with 4 mM DTT, which produced a significant fluorescence response (220 au). Therefore, the absence of this cellular stress response during heterologous expression of hA2aR was not due to failure of the reporter system in the hA2aR strain.
The heat shock response (HSR) within S. cerevisiae serves to activate cellular pathways in response to cellular cytosolic stress. In addition to elevated cell culture temperature (∼37–42°C), the HSR pathway may also be triggered by other stress-inducing conditions, such as accumulation of misfolded protein within the cytosolic compartment.29 We investigated whether this cytosolic stress pathway was activated during heterologous expression of mammalian GPCRs in S. cerevisiae, as this would indicate that systemic cellular stress occurs. The heat shock reporter element exists as a translational fusion to lacZ, which links activation of the HSR pathway to expression of β-galactosidase.28
In contrast to our observations with UPR activation, we find that only 9 of the 12 receptors (hA1, hA3, hCCR5, hCB2, hD2L, hFSH, hCXCR4, hNK1, and hNK2) expressed in this system trigger a HSR [>0.8 U β-galactosidase; Fig. 3(A, open bars)]. As expected from the lack of UPR induction, expression of hA2aR does not result in a HSR (<0.20 U β-galactosidase) compared to parental cells. Interestingly, expression of either hA2b or hCB1 receptors also does not produce a HSR from this reporter system (<0.20 U β-galactosidase). To verify that reporter plasmid loss in yeast strains expressing the hA2a, hA2b, and hCB1 receptors was not responsible for this negative result, these strains were heat shocked at 42°C for 1 h, which produced a significant rise in β-galactosidase activity (1.11 U β-galactosidase for hA2aR, 0.65 U β-galactosidase for hA2bR, and 2.25 U β-galactosidase for hCB1R), indicating that the heat shock reporter was present and functional within these cell strains.
Most receptors maintain an association with BiP
As the ER-resident protein BiP is implicated in the UPR, affinity pull-down experiments were conducted as a complementary method to the UPR studies. As all cells exhibited a significant UPR during the expression of GPCRs except for hA2aR, we sought to identify whether the ER-resident chaperone BiP was associated specifically with the receptors.
When affinity pull-down experiments were performed, generally high levels of BiP were found to be associated with all receptors, except for the hA2a receptor [Fig. 4(A)]. As BiP is a molecular chaperone that is upregulated during ER stress, it may have been possible that the BiP binding observed here was partially influenced by the rise in cellular BiP levels within strains in which the UPR pathway was activated. To address this issue, yeast cells expressing each of the adenosine receptors were lysed, and their corresponding BiP levels were measured and compared to untransformed, parental yeast cells. BiP protein measured in the BJ5464 parental strain was similar to that in cells expressing hA2aR [Fig. 4(B)], although BiP levels were generally more elevated in cells expressing other GPCRs [Fig. 4(B)]. Although activation of the UPR in these strains does increase the amount of available BiP relative to wild-type or A2aR-bearing strains [up to twice the amount contained in the wild-type strain, Fig. 4(B)], when these differences are taken into account, they do not account for the differences in BiP association between A2aR and most other GPCRs that were observed in pull-down samples. In the association experiment (Fig. 4), BiP levels for these receptors were 3–16 times greater than the parental strain [Fig. 4(A)], which is in semiquantitative agreement with independent biological replicates. As a consequence, it can be inferred that BiP maintains a specific association with most receptors that is not a byproduct of UPR upregulation.
Leader sequence processing suggests a problem with translocation
All constructs used for GPCR-GFP expression in these studies include an N-terminal pre-pro leader sequence upstream of the GPCR gene to facilitate targeting to the ER. This leader sequence is cleaved from mature protein in two steps, which corresponds to processing and trafficking of the expressed membrane protein along the secretory pathway. As illustrated in Figure 5(A), the prepeptide leader sequence is cleaved in the ER by signal peptidase, following signal sequence exit from the translocon, whereas the remaining propeptide is cleaved by the Kex2 protease after the membrane protein enters the trans-Golgi.30
To determine the extent of leader sequence processing throughout the course of expression, GFP-tagged receptors were expressed and analyzed at 2, 5, and 24 h postinduction via in-gel fluorescence of samples separated via SDS-PAGE, and the migration of these proteins was compared to those of the corresponding GFP-tagged “mature” receptor (subcloned and expressed without the pre-pro sequence) [Fig. 5(B)]. Earlier time points were included to complement data from UPR experiments, which generally showed an upregulation of the UPR 5–6 h postinduction for all cells expressing GPCRs, except hA2aR.
As shown in Figure 5(C), samples from cells expressing the hA2a-GFP-His10 receptors with the N-terminal pre-pro leader sequence migrated to the same distance as mature hA2aR at all time points tested, indicating that the pre-pro sequence was uniformly cleaved from this receptor throughout the course of its expression. However, samples from cells expressing hA1-GFP-His10 and hA3-GFP-His10 receptors with the pre-pro sequence migrated slower relative to their mature forms at each time point, which is consistent with the mobility shift expected from incomplete processing of the leader sequence. In the case of the hA2b-GFP-His10 receptor, although a band with decreased mobility relative to the mature protein was observed, a fraction of receptors migrates to the same distance as mature hA2b, suggesting that both mature and pre-Kex2 forms of the hA2bR were present.
As a control in these experiments, samples of cells expressing receptors with and without the pre-pro leader were digested with the deglycosylating agent PNGaseF to determine if any expressed receptors were modified by N-linked glycosylation. Digestion of the adenosine receptors with PNGaseF did not alter the mobility of these proteins, which confirms that glycosylation of these receptors was not responsible for the observed shift in mobility from the mature form. Table I summarizes the results for each GPCR from the deglycosylation studies and the mobility shift assay. Notably, we find that the N-terminal pre-pro leader sequence was cleaved from a population of hCCR5-GFP-His10 and hD2L-GFP-His10 receptors, similar to behavior observed for the hA2b-GFP-His10 receptor, although in all of these cases there is a significant fraction of receptors that migrated slower than the mature form.
To confirm the results of the mobility shift assay, some fluorescently tagged GPCRs were expressed for 24 h, purified via nickel affinity chromatography, and N-terminally sequenced. The entire pre-pro leader sequence was found to be properly cleaved from the hA2a receptor, revealing the amino acid sequence of mature hA2aR (E-A-R-P-D-V-M-P-I).14 N-terminal sequencing also identified a population of processed hA2bR (E-A-R-P-D-V-M-L). In contrast, hCB2, hNK1, and hA3 receptors were shown to have intact N-terminal pre-pro sequences (M-K-V-L-I-V-L-L-A-I), indicating that signal peptidase did not cleave the pre sequence following its exit from the translocon, consistent with the results from the mobility shift experiments.
Activity of adenosine receptors expressed in S. cerevisiae
Given that some GPCRs appeared to be retained intracellularly, we wanted to test for ligand-binding activity. We focused our analysis primarily on the family of adenosine receptors, as members of this family include the full range of trafficking patterns, stress response upregulation, and leader sequence processing observed in these studies. To determine whether adenosine receptors were folded properly, radioligand-binding experiments were performed on membrane preparations. Previously, we have shown that hA2aR is folded in an active, ligand-binding conformation in whole-cell assays, in membrane preparations, and once purified from the cells.14, 15, 18, 31 Using 3H-NECA, an adenosine analog and receptor agonist for hA2aR, hA2bR, and hA1R,32 enabled ready comparison in these ligand-binding studies.
Figure 6(A) shows that addition of 3H-NECA to yeast membranes expressing GFP-tagged hA2aR resulted in saturation binding with a Kd of 36 ± 2 nM. Based on 0.2 mg of membrane protein used in the assay and the total bound complexes obtained (Bmax), the strain produces 855 pmol active A2aR/mg membrane protein. The Kd compares well to expected values for this radioligand, because the estimated Kd for NECA with hA2aR is within the range of 1–20 nM.32 Specific binding of 3H-NECA was also observed for yeast membranes containing recombinant, GFP-tagged hA2bR [Fig. 6(B)]. However, saturation was not achieved in this case because of technical limitations of this experiment; the expected Kd for NECA binding to hA2bR is ∼5 μM,32 which is near the highest working concentration of radioligand available. For A2bR, the observed bound complexes correspond to ∼11 pmol/mg membrane protein, although we have not reached saturation, so this is likely an underestimate of the true yield. In contrast to the results observed for hA2aR and hA2bR, no appreciable binding was seen for yeast membranes expressing recombinant hA1 or hA3 GFP-tagged receptors in the presence of 400 nM3H-NECA (data not shown) (Kd for hA1R is 3–30 nM32), and CFP-tagged hA3 receptors also did not demonstrate appreciable activity when incubated in excess of 4 nM125I-AB-MECA (Kd for hA3R with AB-MECA is 1.7 nM) (data not shown). To validate that the conditions of the radioligand-binding assay for both hA1 and hA3 receptors were adequate to detect binding in yeast membranes, recombinant mammalian membranes containing these receptors were analyzed (refer to Materials and Methods) and specific activity was observed for these preparations (data not shown). This indicates that the lack of activity observed for both hA1 and hA3 receptors expressed in yeast was not due to inadequacies of the ligand-binding assay. If there were active A1R and A3R present in yeast cells, we would have expected to detect a minimum of 10% of the active complex (Bmax) values measured from mammalian controls. Using this as the limit of detection suggests that active A1R levels were less than 5 pmol/mg (based on NECA), and for A3R, less than 0.2 pmol/mg (IB-MECA).
In these studies, we examined cellular expression of 12 GPCRs in S. cerevisiae to gain a better understanding of limitations to plasma membrane localization, with the goal of achieving high functional yields for these proteins. We focused on determining why receptors trafficked differently, and if intracellular retention could be linked to protein processing and cellular stress responses. Of the 12 mammalian GPCRs expressed in this host, only the human adenosine A2a receptor escaped quality control responses and was localized primarily at the plasma membrane. Eleven other receptors, three of which are closely related to hA2aR, did not appreciably reach the membrane and triggered the UPR. These receptors also specifically associated with the ER-resident molecular chaperone BiP.
Under these conditions, S. cerevisiae was able to produce the desired milligrams of GPCRs/L of culture. To our knowledge, only the human dopamine D2,16, 34 human adenosine A2a,14, 15, 18, 35 hA2bR,33 and human neurokinin NK115 receptors of these 12 have been previously expressed in S. cerevisiae. Although expression yields for the hNK1 receptor were not reported, expression levels for hA2aR reached 10–15 mg/L of culture,14, 18 whereas hD2R expression was 2.8 pmol/mg,34 which corresponds to ∼0.1 mg/L of culture, using the conversion given by Sarramegna et al.9
Many of the other receptors investigated in this study have been expressed in different host systems previously. For example, a maltose-binding protein fusion to hCB1R produced in E. coli yielded purified protein at ∼2.0 mg/L,36 while similar approaches have been used to produce hCB2R at yields up to 39.5 pmol/mg,37 which has also led to its purification in milligram quantities.38Pichia pastoris produced 3.6 pmol/mg (∼0.15 mg/L) of hCB2R,39 14.0 pmol/mg (∼0.56 mg/L) of hD2R in Schizosaccharomyces pombe,34 and 80 pmol/mg (∼0.62 mg/L) of the hNK1R in Sf9 insect cells.40
To our knowledge, we are the first to report high levels of heterologous expression of the hA1, hA2b, hA3, hCCR5, hCXCR4, and hFSH receptors. Collectively, the expression yields achieved for each GPCR investigated in this study are among the highest levels yet reported among available host systems, and most represent major improvements in yields compared to those achieved in other studies. Although other studies have emphasized the utility of screening fluorescence GFP fusions as a means to achieve membrane protein overproduction in yeast,41 in our system we do not observe a correlation between fluorescence of GFP-fusion protein and ligand-binding activity. For instance, although GFP is folded and optically active when fused to hA1R and hA3R (Fig. 2), we were unable to detect activity for these receptors. However, we do not observe aggregated material, for example, through pelleting of material at 16,000g. Therefore, we stress that yields estimated from GFP fluorescence represent total protein and are not necessarily active (ligand-binding competent) yields.
Although the protein yields attained here were generally excellent, we found that many GPCRs did not reach the cell periphery. Of the GPCRs that were examined here in S. cerevisiae, only the hA2a receptor localized primarily to the plasma membrane. In previous studies, for yeast cells expressing both hA2bR and a chimeric Gαs, cell cycle arrest upon ligand addition was observed, which suggested the presence of active receptors at the plasma membrane.33 This is consistent with our observed ligand-binding activity for membrane preparations [Fig. 6(B)], although we do not yet know whether any of the receptors couple to the endogenous yeast Gα subunit, which could alter measurable Kd values that describe the association of ligands with these receptors.
We observed that most receptors that were retained intracellularly activate stress responses (UPR and HSR), which links trafficking to cellular events typically associated with quality control. Although high levels of hA2aR are produced in this system, heterologous expression of this particular GPCR escapes both quality control checkpoints under these culture conditions. In light of this observation, we hypothesize that the trafficking pattern observed for receptors other than A2aR is the result of ER retention and quality control pathways, which are closely linked to activation of the UPR. For most expression constructs studied, a significant UPR response is detectable 4–6 h post induction. It is somewhat difficult to ascertain if a significant lag phase exists between expression induction and UPR activation, because detection of the UPR in our system is also coupled to hac1 splicing and to the expression and folding of GFP, which have inherent time delays.
Although UPR activation in this system corresponds with a high level of receptors retained intracellularly, it is less clear why differential activation of the HSR pathway occurs among the GPCRs studied. Upregulation of the HSR for most receptors could be caused by misfolding of the intracellular loop regions of these membrane proteins (exposed to the cytoplasm). For example, expression of the hA2bR and hCB1R may not trigger the HSR pathway if intracellular loop regions of these proteins are folded in such a way that they do not result in Hsp binding within the cytosol. Alternatively, prolonged UPR could result in systemic cellular stress, activating the HSR as a secondary response.
Previously, it has been shown that activation of the UPR pathway in the expression of both soluble42 and membrane-bound proteins25 in S. cerevisiae results when unfolded or partially misfolded proteins are present within the ER. We explored this possibility indirectly for the GPCRs expressed in our system by comparing their level of association with the ER-resident chaperone, BiP, which is known to bind to exposed hydrophobic regions of misfolded proteins. As expected, hA2aR maintained a negligible association with BiP, which is consistent with its proper folding and trafficking out of the ER. As we have observed that BiP preferentially associates with most other receptors studied (Fig. 4), we can infer that at least some population of these highly hydrophobic receptors are folded in a non-native conformation within the ER.
As several of our initial observations pointed to problems within the ER, we chose to investigate the processing of the N-terminal pre-pro leader sequence fused to GPCRs. We found that the leader sequence was not uniformly cleaved, which is unexpected considering that this leader sequence should aid in membrane protein translocation. A recent study regarding hCB2R has also suggested that signal sequence processing may prove critical to GPCR production in the methylotrophic yeast P. pastoris.43 Our analysis of leader sequence processing among receptors expressed in S. cerevisiae (Table I) shows that all receptors other than the hA2aR display either a homogeneous population of unprocessed (immature) receptors or a heterogeneous population of both processed (mature) and unprocessed (immature) proteins. It is unlikely that slow translocon gating results in cytosolic synthesis, as we find that these proteins are inserted into a membrane environment. In addition, activation of the UPR and BiP binding for these receptors verifies that expressed receptors at least partially enter the ER membrane. As the leader sequence remains fused to expressed GPCRs in this system, signal peptidase must not cleave efficiently, which suggests that it does not have access to the pre sequence on the lumenal side of the ER membrane. The lack of leader sequence processing could be attributed to receptor misfolding and/or the adoption of an unexpected topology within the ER membrane shortly after receptor synthesis, which precludes leader sequence cleavage.
Radioligand-binding assays conducted on membrane preparations for the adenosine receptor family link leader sequence processing with ligand-binding activity of expressed GPCRs in this system. We have found that only the hA2aR and hA2bR bind to the adenosine analog NECA (Fig. 6), whereas closely related receptors hA1R and hA3R that do not demonstrate leader sequence processing also do not display appreciable activity. In light of this observation, we believe that the hD2LR and hCCR5R also are likely to show some amount of activity, because a fraction of these receptors are properly processed. Although hCCR5R has not yet been heterologously expressed in yeast, Michel and coworkers reported activity of hD2R in S. cerevisiae,16 which supports this hypothesis.
Although GPCRs are characterized by the presence of a conserved seven alpha-helical transmembrane domain, various motifs, coupling partners, and post-translational modifications differentiate these receptors (Table II). We sought to correlate differences between these features with subcellar expression patterns and cellular responses that occur in response to heterologous GPCR expression within S. cerevisiae. However, there is no obvious correlation between these properties (N-terminal length, putative N-linked glycosylation sites, G-protein coupling, homology to hA2aR, or presence of rare codons) and leader sequence cleavage, although each may serve to contribute to receptor-specific effects.
Table II. Summary of Characteristics Among GPCRs Used in This Study
Calculated through BLASTp using a BLOSUM62 matrix (open gap: 11; extension gap: 1 penalty; gap x_dropoff 15; expect 10.00; word size 3).
2 (153, 163)
3(3, 4, 12)
2 (11, 176)
3 (77, 83, 112)
3 (5, 17, 23)
4 (191, 199, 293, 318)
2 (14, 18)
12, 13, 11, q, s
2 (11, 19)
11, q, s
Taken together, these data support the conclusion that although high yields are readily achieved, problems with mammalian GPCR trafficking in this system arise very early in the secretory pathway. Incomplete processing of the pre-pro leader sequence in the expression of all GPCRs except the hA2a receptor suggests that improper ER translocation causes transmembrane helices to form improper contacts with each other, which may drive misfolding and ER retention. We are currently investigating protein topology within the ER membrane to better understand the mechanism that gives rise to differential processing, folding, and localization for these receptors. We are also developing methods to facilitate proper translocation and trafficking for the hA2bR, hD2LR, and hCCR5R, which demonstrate a population of properly processed receptors.
Materials and Methods
Subcloning and construction of GPCR strains
The human adenosine hA1 (hA1R), human adenosine A2b (hA2bR), human cannabinoid CB1 (hCB1R), human cannabinoid CB2 (hCB2R), human dopamine D2L (long isoform) (hD2sR), human chemokine CXCR4 (hCXCR4R), human chemokine CCR5 (hCCR5R), human neurokinin NK1 (hNK1R), and human neurokinin NK2 (hNK2R) receptors were amplified using polymerase chain reaction (PCR) from mammalian shuttle vectors (UMR cDNA resource center). Genes for the human adenosine A2a (hA2aR), human adenosine A3 (hA3R), and human follicle stimulating hormone receptor (hFSHR) were provided by Merck Pharmaceuticals.
GPCRs were subcloned into the multi-integrating pITy4-wt plasmid44 for expression in S. cerevisiae, as performed previously for high-level expression of the human A2aR.14, 15, 18 Each GPCR was subcloned directly into the pITy-MC-His10 plasmid14 and the pITy-MC-eGFP-His10 plasmid. Both vectors contain a multiple cloning (MC) site consisting of EagI, AatII, and SacII immediately following the Gal1-10 promotor and synthetic pre-pro N-terminal leader sequence.45 The pITy-MC-eGFP-His10 plasmid also contains the gene for enhanced green fluorescent protein (eGFP) after the MC site and before an encoded decahistidine purification tag. hA2aR, hA2bR, and hA3R were PCR-amplified with a 5′ AatII site upstream of the gene and a 3′ SacII site downstream of the gene. The PCR product was digested with AatII and SacII, purified using a PCR clean-up kit (Promega, Madison, WI), and then ligated into pITy-MC-His10 and pITy-MC-eGFP-His10 plasmids, which were previously digested with AatII and SacII. All other GPCRs were PCR-amplified with a 5′ EagI site and a 3′ SacII site, digested with EagI and SacII, and similarly ligated into pITy-MC-His10 and pITy-MC-eGFP-His10 digested with EagI and SacII. After insertion into the above vectors, the following amino acids are introduced between the pre-pro leader sequence and the GPCR N-termini: E-A-R-P-D-V for hA2aR, hA2bR, and hA3R, and E-A-R-P for all other GPCRs. For GFP-tagged receptors, an additional P-R amino acid linker is introduced between the C-terminus of each receptor and GFP.
GPCR genes were also subcloned into the pITy-MC-eGFP-His10 plasmid without the synthetic N-terminal pre-pro leader sequence. This construct was created by digesting pITy-A2aR-eGFP-His10 with KpnI and AatII to remove the Gal1-10 promotor and pre-pro leader sequence. Following this digest, the Gal1-10 promotor with an added 5′ KpnI site and a 3′ AatII site was PCR-amplified, digested with KpnI and AatII, and then ligated into the previously digested vector. Other GPCR genes were inserted into this plasmid in place of A2aR, as detailed for the constructs mentioned earlier.
Additional constructs were created for some receptors (hA3R and hNK1R) where cyan fluorescent protein (CFP) was present at the C-terminus instead of GFP. These constructs were created in the same manner as the GFP constructs listed earlier, but contain an L-E amino acid linker between the receptor C-terminus and CFP. All GPCR constructs were verified through DNA sequencing (University of Delaware DNA Sequencing Facility or University of Pennsylvania DNA Sequencing Facility).
GPCR constructs (both with and without GFP tags) were transformed into S. cerevisiae BJ5464 (MATα ura3-52 trp1 leu2Δ1 his3Δ200 pep4::HIS3 prb1-Δ1.6R canI GAL), which is a vacuolar protease-deficient yeast strain. hA1R, hA2aR, hA2bR, hFSHR, hCB1R, hCB2R, and hNK2R constructs were linearized with BsaBI, whereas hA3R, hD2sR, hNK1R, hCXCR4R, and hCCR5R constructs were linearized with AhdI before transformation through electroporation. Transformation and selection of transformed yeast cells were carried out as performed previously for the human A2a receptor,14 via selection with G418.
Additionally, nonfluorescently tagged GPCR strains were transformed with pRS316-UPRE-GFP27 or pwB215,28 which are centromeric plasmids that serve as reporters of the UPR and HSR, respectively. pRS316-UPRE-GFP contains four UPR elements (UPRE) immediately upstream of the GFP gene, enabling detection of UPR induction by monitoring GFP fluorescence.26 Similarly, pwB215 contains the promoter region of SSA4, a heat shock inducible element, as a translational fusion to lacZ that produces β-galactosidase when the HSR is induced.28 Plasmids were transformed using a standard lithium acetate chemical transformation method.46 Cells transformed with pRS316-UPRE-GFP were selected on SD-2XSCAA47 plates lacking uracil, whereas cells transformed with pwB215 were selected on SD-2XSCAA plates lacking tryptophan.
Yeast cultures were typically grown overnight in YPD media (2% Bacto peptone, 2% glucose, 1% yeast extract) in 5-mL culture tubes in a water bath shaker, set to 30°C and 275 rpm. In BJ5464 cells that also contained heat shock or UPR reporter plasmids, cultures were instead grown in selective media to maintain plasmid selection. SD-2XSCAA media lacking uracil were used for cells with pRS316-UPRE-GFP, and SD-2XSCAA media lacking tryptophan were used for cells with pwB215. GPCR expression was induced by resuspension of cells to an OD600 of 0.5 in YPG media (2% Bacto peptone, 2% galactose, 1% yeast extract). Cultures expressing GPCRs were grown at 30°C at 275 rpm, and samples were collected at various times. Cell pellets were collected by centrifugation at 2000g for 2 min, and the following assays were performed as needed.
Expression levels of GFP-tagged GPCRs were determined by measuring whole-cell fluorescence of GFP-tagged GPCRs. Yeast cells expressing C-terminally, GFP-tagged GPCRs were grown for ∼24 h, the OD600 of the cultures was measured and recorded, and 1 OD-mL of cells was centrifuged for 2 min at 2000g. The supernatant was decanted and cells were resuspended to an OD600 of 0.5 in 2 mL PBS buffer (1.44 g/L Na2HPO4, 0.24 g/L KH2PO4, 0.2 g/L KCl, 8.0 g/L NaCl, pH 7.4). Samples were placed in a quartz cuvette, excited at 488 nm, and fluorescence was detected at 511 nm on a Hitachi F-4500 fluorescence spectrophotometer. Relative per cell expression levels and culture yields were estimated by comparison to an enhanced GFP standard curve, previously used to estimate concentrations of A2aR-GFP in yeast cells.18 A total of 1 OD was estimated to contain ∼2.5 × 107 cells per mL of culture.31 Three or more replicates were used to calculate average expression levels, and the error is the standard deviation from the mean. At this early time points (1–5 hours) and for samples grown at scales of 5–50 mLs no degradation of the GPCR-GFP constructs was observed via SDS-PAGE, suggesting that GFP is a good indicator of total heterologous protein expression. For 24-hour time points in larger size cultures, the degradation was between 0–15%, depending on the receptor identity. However, within the error of both measurements, we believe total fluorescence gives a reasonable value on total protein yields.
Unfolded protein response and heat shock response assays
Yeast cells expressing GPCRs with the pRS316-UPRE-GFP reporter plasmid were grown as detailed earlier. ODs were measured and 1 OD-mL was removed from each culture at the time of induction and at ∼4, 6, 8, 12, 16, and 24 h after GPCR expression was induced. Cell aliquots were centrifuged for 5 min at 2000g and the supernatant was removed. Pellets were resuspended in 1 mL of PBS buffer and tested for GFP fluorescence (excitation at λ = 488 nm, emission at λ = 511 nm).
For HSR measurements, ∼1 OD-mL of cells was removed from cultures expressing GPCRs with the pwB215 plasmid at various time points postinduction. β-Galactosidase activity was measured and quantified from permeabilized cell pellets according to a standard protocol developed for yeast.48
Expression of each His-tagged GPCR was induced by growth in galactose (YPG). After 24 h, 5 OD-mL of each strain was collected, centrifuged at 2000g for 2 min, the supernatant was decanted, and cell pellets were frozen at −20°C until use. Pellets were resuspended in 200 μL Buffer A (50 mM Tris-HCl, 1% n-dodecyl-β-D-maltoside (DDM), pH 7.9) including protease inhibitors (1 mM phenylmethanesulphonylfluoride (PMSF) and Complete EDTA-free protease inhibitor tablets (Roche, Indianapolis, IN)) and combined with 600 μL zirconia-silica beads (BioSpec, Barlesville, OK) in a 1.5-mL microcentrifuge tube. We used DDM for this extraction, as it is a commonly used surfactant for membrane protein studies (see Ref.7). Although it is true that membrane proteins can have different solubilization efficiencies in different surfactants, generally DDM has been shown to be a good solubilizing agent, especially at the concentration used in this study.7 Samples were lysed on a vortexer with six cycles of 30 s agitation with 30 s on ice in between cycles. Lysate was separated from the beads and unlysed cells and heavy cell debris were removed through centrifugation of the lysate at 16,000g for 1 min. Lysate was then combined with 800 μL Buffer B (50 mM Tris-HCl, pH 7.9, 180 mM NaCl, 6% nonfat dried milk, 1% DDM with protease inhibitors). Approximately 25 μL of settled Ni-NTA Superflow resin (Quiagen, Valencia, CA) was pre-equilibrated with 1 mL of Buffer AB (1 part Buffer A: 5 parts Buffer B), then was centrifuged at 16,000g for 1 min, and the supernatant was removed. Lysate was then added to the pre-equilibrated nickel resin, and His-tagged GPCRs were allowed to bind to the resin for ∼12 h on an end-over-end mixer at 4°C. Nonspecifically bound material was washed away from nickel-bound protein using five washes of 1 mL Buffer C (50 mM Tris-HCl, 150 mM NaCl, 1% DDM, pH 7.9) with protease inhibitors. Bound protein was eluted through reduction of the nickel ions with SDS loading buffer (50 mM Tris, 0.25 mg/mL bromophenol blue, 10% glycerol, 2% SDS) containing 100 mM EDTA. Eluted protein was heated at 60°C for 5 min in SDS loading buffer with EDTA before electrophoretic separation on 12% SDS-PAGE.
Western immunoassay was performed by transfer to nitrocellulose using a Trans-Blot Cell (Bio-Rad, Hercules, CA). GPCRs were detected with a mouse anti-6xHis primary antibody (Abcam, Cambridge, MA) used at a dilution of 1:1000, whereas BiP was detected with a rabbit anti-BiP primary used at a dilution of 1:5000. The antibody against BiP was generated in rabbits using a MBP-Kar2 fusion protein as the antigen. The C-terminal region of Kar2 (from XbaI site through stop codon) was inserted into pMal-c2g (New England Biolabs). This fragment was purified with amylase resin (New England Biolabs) and eluted with 10 mM maltose. Rabbit polyclonal antibodies were generated against the purified fusion protein (Capralogics) and screened for specificity in yeast lysates. The antibodies were purified from serum using a Protein A Montage kit (Millipore) and stored in aliquots at −80°C. Secondary HRP-conjugated anti-mouse and anti-rabbit antibodies (GE Healthcare, Piscataway, NJ) were used at a dilution of 1:2000, and bands were detected using an ECL Plus Western Blotting Detection Kit (GE Healthcare). Chemiluminescence was visualized on a Typhoon 9400 Variable Mode Imager (GE Healthcare) after excitation at 457 nm with a 520BP40 filter.
Mobility shift assay and N-terminal sequencing
To determine leader sequence cleavage, 5 OD-mL of yeast strains expressing GFP-tagged GPCRs was harvested at 2, 5, and 24 h postinduction. For comparison, 5 OD-mL of yeast cells expressing GPCRs lacking the pre-pro leader sequence was collected 24 h postinduction. Cell pellets were resuspended in 200 μL STE10 buffer (10 mM sucrose, 10 mM Tris, 10 mM EDTA, pH 7.0) supplemented with Complete EDTA-free protease inhibitors (Roche) and 1% DDM and added to 600 μL zirconia/silica beads. Cells were lysed through vigorous vortexing for three cycles of 1-min agitation and 1 min on ice between cycles. Recovered lysate was collected and stored at –20°C before use.
Deglycosylation of expressed proteins was achieved through overnight digestion with PNGaseF (New England Biolabs, Ipswich, MA) at 37°C and compared to mock-prepared samples lacking enzyme. Samples were combined with 3× SDS loading buffer and separated via 10% SDS-PAGE. GFP-tagged receptors were visualized directly in SDS-PAGE gels with a Typhoon 9400 variable mode imager (Amersham Biosciences), equipped with a 488 nm excitation filter.
Generally, protein samples were prepared for N-terminal sequencing from 200 mL cultures expressing GFP-tagged GPCRs for 24 h. Cells were lysed as detailed earlier, and protein was purified via immobilized metal affinity chromatography through the attached decahistidine tag, as described previously.14 In some cases, liquid samples of purified protein were sequenced directly (University of Delaware Core Protein Sequencing Facility). Alternatively, protein samples were combined with 3× SDS loading buffer and separated via 12% SDS-PAGE. Samples were then blotted onto PVDF and Coomassie stained to reveal the protein band of interest. Protein bands were then N-terminally sequenced by the Protein Facility at Iowa State University.
Integral membrane protein assay
Membrane association of recombinantly expressed, GFP-tagged GPCRs was determined according to the method of Kuchler et al.19 Briefly, 20 OD-mL of cells was collected after expression for 24 h. All subsequent steps were performed at 4°C. Cell pellets were resuspended in 400 μL lysis buffer (10 mM Tris, 1 mM EDTA, pH 7.8) supplemented with protease inhibitors and combined with 600 μL zirconia/silica beads and lysed as described in the preceding section. Lysate was collected and diluted to ∼1 mL, and unlysed cells and large debris were pelleted for 5 min at 1000g. The resulting supernatant was centrifuged for 1 h at 200,000g using a Beckman Optima L-100 XP ultracentrifuge. Membrane pellets were resuspended to a final volume of 1 mL in lysis buffer and divided into 4 × 250 μL aliquots. These membrane aliquots were resuspended in lysis buffer and concentrated stock solutions of Na2CO3, urea, and SDS to reach a final volume of 1 mL for all treatments, which corresponded to final concentrations of 0.1M Na2CO3 (pH 11.0), 2M urea, 1% SDS. Samples were incubated on an end-over-end mixer at 4°C for 1 h, and then membranes were pelleted for 1 h at 200,000g. Various samples from supernatant and resuspended pellet fractions were combined with 3× SDS loading buffer, separated on 12% SDS-PAGE, and visualized directly in SDS-PAGE gels with a Typhoon variable imager, equipped with a 488 nm excitation filter.
Yeast-expressed mammalian GPCRs with a C-terminal GFP tag were imaged using a Zeiss LSM 5 Duo confocal microscope (Zeiss, Oberkocken, Germany) with a 63×/NA 1.4 objective. Cells were immobilized on eight-chamber LabTek NUNC slides coated with poly-D-lysine (MW > 300,000).
Membrane preparation and radioligand binding
Yeast membranes were prepared from cultures expressing GFP-tagged GPCRs for hA1R, hA2aR, hA2bR, and hA3R, according to a previously published protocol,16 with the following modifications. Isolation of membranes containing GFP-tagged receptors was used, as it facilitated identification of the membranes throughout their purification. Cultures were grown in YPG for 24 h to enable GPCR expression, and 150 mL of cell culture was collected and centrifuged for 2 min at 2000g. The resulting cell pellets were washed in ddH2O and centrifuged as mentioned earlier. Pellets were then resuspended in 5 mL STED10 buffer (10 mM Tris-HCl, 1 mM EDTA, 1 mM DTT, 10% sucrose, pH 7.6) supplemented with Complete EDTA-free protease inhibitors (Roche) and 1 mM PMSF and combined with ∼20 g zirconia/silica beads in a 50-mL conical tube. This resulting suspension was vigorously vortexed at 4°C for 5 min to lyse cells. Following lysis, beads were removed from the homogenate by filtration through a Kontes column, and 15 mL STED10 buffer was added to the mixture. The homogenate was centrifuged at 1600g at 4°C for 30 min, and the supernatant was separated from the mixture and further centrifuged at 27,620g at 4°C for 45 min to pellet membranes. Crude membrane preparations were resuspended in 4 mL STED10 buffer with protease inhibitors, and total membrane protein content was determined using an RC/DC BCA kit (Bio-Rad, Hercules, CA) using BSA as a standard.
Membrane preparations were combined with various amounts of tritiated ethylcarboxyaminoadenosine (3H-NECA) (Perkin Elmer, Boston, MA) to determine ligand-binding activity of hA1R-GFP-His10, hA2aR-GFP-His10, hA2bR-GFP-His10, and hA3R-GFP-His10 in yeast membranes. Additionally, activity of Ni2+ purified, DDM/CHAPS/CHS-solubilized hA3R-CFP-His10 was tested with binding to 125I-IB-MECA (Perkin Elmer). Before activity measurement, membranes were diluted to a concentration of 1 μg membrane protein per μL in ligand-binding buffer (50 mM Tris-HCl, 10 mM MgCl2, pH 7.4) and treated with 4 U/mL adenosine deaminase (Roche) to scavenge free adenosine. Mammalian cell membrane preparations for the rat A3 receptor and the human A1 receptor were obtained from Perkin Elmer and used as positive controls for the assay. HEK293 membranes containing the human adenosine A2b receptor were prepared following transient transfection with the pCEP4-A2bR-CFP plasmid.49
A total of 200 μg membrane extracts were combined with varying amounts of radiolabeled ligand in 96-well plates with glass fiber B filters (Millipore, Bedford, MA) and incubated at room temperature for 1–2 h before filtration. Samples were washed five times in ice-cold ligand-binding buffer, combined with 30 μL scintillantion solution, and radioactive counts were measured on a Perkin Elmer Microbeta Jet scintillation counter. Resulting data of three replicate samples were averaged, with error bars representing the standard deviation. Data were fit to a one-site binding model using KaleidaGraph 3.5 to determine relevant binding parameters.
Confocal microscopy was performed at the Bioimaging Core Facilities at the University of Delaware. The authors thank Dr. Elizabeth Craig (University of Wisconsin) for providing the pwB215 plasmid, Dr. K. Dane Wittrup (MIT) for providing the pITy4 plasmid, and Dr. Marlene Jacobson (Merck) for providing the hA2aR, hA3R, and hFSH genes. They thank Andrea N. Naranjo for providing HEK293 cells expressing A2bR, Yu-Chu (Angel) Huang for the protein sequencing carried out at UD, and Mark J. Richards for help with subcloning.