Glycosyl-phosphatidylinositol-anchor addition signals are processed in Nicotiana tabacum


  • Adam M. Takos,

    1. Centre for Plant Membrane Biology, School of Biological Sciences, Flinders University, GPO Box 2100, Adelaide, SA 5001, Australia, and
    Search for more papers by this author
  • Ian B. Dry,

    1. CSIRO Division of Horticulture, Hartley Grove, Urrbrae, SA 5064, Australia
    Search for more papers by this author
  • Kathleen L. Soole

    Corresponding author
    1. Centre for Plant Membrane Biology, School of Biological Sciences, Flinders University, GPO Box 2100, Adelaide, SA 5001, Australia, and
      *For correspondence (fax + 61 88 201 3015; e-mail
    Search for more papers by this author

*For correspondence (fax + 61 88 201 3015; e-mail


Recent studies have demonstrated the existence of glycosyl-phosphatidylinositol (GPI)-anchored proteins in higher plants. In this study we tested whether GPI-addition signals from diverse evolutionary sources would function to link a GPI-anchor to a reporter protein in plant cells. Tobacco protoplasts were transiently transfected with a truncated form of the Clostridium thermocellum endoglucanase E reporter gene (celE′) fused with a tobacco secretion signal (PR-1a) at the N-terminus and either a yeast (GAS1), mammalian (Thy-1) or putative plant (LeAGP-1) GPI-anchor addition signal at the C-terminus. The yeast and plant C-terminal signals were found to be capable of directing the addition of a GPI-anchor to the endoglucanase protein (EGE′) as shown by the sensitivity of the lipid component of GPI to phosphatidylinositol-specific phospholipase C (PI-PLC) digestion. In contrast, the mammalian signal was poorly processed for anchor addition. When EGE′ was fused to a truncated form of the LeAGP-1 signal (missing three amino acids predicted to be critical to signal cleavage and anchor addition), a GPI-anchor was not linked to the EGE′ protein indicating the necessity for the missing amino acids. Our results show the conservation of the properties of GPI-signals in plant cells and that there may be some similar preferences in GPI-addition signal sequences for yeast and plant cells.


The plant plasma membrane is not static, but represents a dynamic site of regulation of exchange of substances and information between the intracellular and extracellular environments. This interaction often involves cell surface proteins, many of which are unknown or poorly characterised ( Sussman 1994). The mode of attachment of proteins to the plasma membrane may impact significantly on the functions they perform. Whilst attachment of proteins by direct insertion into the plasma membrane of one or more hydrophobic domains has long been accepted dogma, it has been recognised more recently that many eukaryote proteins may be anchored to the plasma membrane by a direct covalent linkage to lipid molecules ( Ferguson & Williams 1988; Low 1989). A common type of lipid anchor is glycosyl-phosphatidylinositol (GPI) and it consists of a conserved glycan linkage containing ethanolamine linked to the C-terminal amino acid of the protein, three mannose residues and glucosamine linked to the inositol group of phosphatidylinositol which is embedded in the external leaflet of the plasma membrane ( Englund 1993). Often the lipid component of this structure is susceptible to cleavage by phosphatidylinositol-specific phospholipase C (PI-PLC) ( Brodbeck & Bütikofer 1994).

Proteins destined to receive a GPI-anchor are synthesized containing cleavable signal peptides at each terminus. The N-terminal signal localises the preproprotein to the lumen of the ER and is cleaved to form the proform of the protein. The C-terminal GPI-addition signal of the proprotein is then a substrate for a transamidase complex which catalyses the cleavage of the signal and replacement en bloc with a preformed GPI-anchor ( Kinoshita & Takeda 1994). The GPI-linked protein then transverses the default pathway of secretion and is retained on the external leaflet of the plasma membrane solely by the lipid component of GPI ( Takeda & Kinoshita 1995).

Whilst there is no sequence homology between GPI-anchor addition signals from different sources, some characteristic properties of the signal are conserved. At the GPI-attachment site (designated the ω amino acid) only small amino acids such as Ser, Asn, Ala, Gly, Asp or Cys may be present while at the ω+2 site only Ala, Gly, Thr or Ser are commonly found ( Udenfriend & Kodukula 1995). The ω+1 site has far less specific amino acid requirements ( Gerber et al. 1992 ). The region between ω+3 and ω+10 is frequently rich in charged amino acids and forms a putative hinge domain ( Furukawa et al. 1997 ). This hinge region is then followed by 8–20 amino acids predominantly hydrophobic in nature, which must traverse the ER membrane before processing of the ω site by the transamidase ( Wang et al. 1999 ).

Since the core GPI structure was first solved for the variant surface glycoprotein of Trypanosoma brucei over 10 years ago ( Ferguson et al. 1988 ), numerous other GPI-anchored proteins of heterogeneous function have been identified and/or cloned from mammals, insects, protozoan and fungi ( Low 1989). The core structures of GPIs have now been solved for most eukaryote kingdoms ( Ferguson 1994) and genes for a biosynthetic pathway comprising at least 10 separate steps have been cloned ( Takeda & Kinoshita 1995). GPI-anchored proteins and the GPI structure has been implicated in diverse processes, such as signal transduction ( Peles et al. 1997 ; Trupp et al. 1998 ), protein targeting to plasma membrane microdomains ( Hoessli & Robinson 1998), parasitic evasion of host immune responses ( Ferguson 1994), cell-cell recognition and nutrient uptake ( Rothberg et al. 1990 ). However, an explanation as to why such a diverse array of proteins should benefit from the addition of a GPI-anchor, rather than being associated with the plasma membrane by polypeptide domains or a less complex anchor structure, remains elusive.

Whilst more than 150 GPI-anchored proteins have been identified in most other eukaryotes ( Low 1989), until very recently there were no reports of these proteins in plant cells. The first evidence that such proteins exist in terrestrial plants was the detection of several unidentified GPI-anchored proteins in tobacco by biotinylation of cell surface proteins and PI-PLC cleavage of the GPI-anchor ( Takos et al. 1997 ). Other evidence of plant GPI-anchored proteins has now emerged, including a nitrate reductase in sugar beet and barley leaves ( Kunze et al. 1997 ), a purple acid phosphatase in the aquatic monocot Spirodela oligorrhiza ( Nakazato et al. 1998 ) and classical arabinogalactan-proteins (AGPs) in Nicotiana alata styles and Pyrus communis suspension culture ( Youl et al. 1998 ) and ‘Paul’s Scarlet’ rose suspension culture ( Svetek et al. 1999 ). Analysis of full-length cDNA clones of classical AGPs indicates that they encode C-terminal sequences consistent with the properties of other eukaryote GPI-addition signals ( Schultz et al. 1998 ).

Despite the common evolutionary features of GPI-addition signals, there appears to be sufficient heterogeneity to affect the efficiency of processing of signals in different kingdoms. For example, protozoan GPI-addition sequences are apparently poorly processed in mammalian cells ( Moran & Caras 1994) and slime moulds ( Reymond et al. 1995 ). Whilst yeast and mammalian signals have been recognized interchangeably in both types of organism ( Guadiz et al. 1998 ; Morel & Massoulié 1997), there is evidence to suggest a more restricted specificity for yeast cells (H. Riezman, personal communication).

In this study, we have used a reporter gene to conclusively demonstrate operation of the GPI-anchor addition pathway in plant cells and have characterised this pathway in terms of processing of GPI-addition signals from yeast, mammalian and plant sources. Our data confirm that the properties of C-terminal signals directing GPI-addition are conserved between plants and other eukaryotes.


EGE′ is stably produced and exported in plant cells

A truncated form of the Clostridium thermocellum celE gene encoding the catalytic domain of endoglucanase E (EGE′) has previously been used as a GPI-anchored reporter molecule to monitor apical sorting of GPI-anchored proteins in polarized epithelial cells ( Soole et al. 1995 ). This reporter protein tolerates N and C-terminal fusions and retains activity upon the addition of a GPI-anchor. In addition, the EGE′ protein is highly thermostable and resistant to proteases and so retains activity under conditions that may be detrimental to other reporter proteins ( Hazelwood et al. 1990 ).

As the celE′ gene had not been previously used as a reporter gene in plants, the stability of EGE′ activity produced in plant cells and the effect of translocation through the ER, upon EGE′ activity, was assessed. Tobacco protoplasts were transfected by electroporation with the plasmid pcelE, which directs the production of EGE′ to the cytosol. A linear increase in extractable EGE′ activity was observed from 24 to 72 h post-transfection which plateaued after 72 h ( Fig. 1a). This indicates that the EGE′ activity produced in the tobacco system is stable over several days and, in subsequent experiments, cells were harvested for enzyme activity measurements within 72 h of transfection. No endogenous EGE′ activity was detectable in extracts of untransfected cells electroporated in the absence of plasmid DNA (data not shown).

Figure 1.

Stable production and export of EGE′.

(a) Time course of intracellular EGE′ production in pcelE transfected cells. EGE′ activity is expressed as pmol 4-methyumbelliferone (4-mUMB) produced per hour.

(b) Time course of EGE′ export to the bathing media of pcelE and pPR-celE transfected cells.

A time course experiment was also conducted to monitor EGE′ activity exported into the media surrounding transfected cells. For this purpose, a construct designed to direct EGE′ to the default pathway of secretion, designated pPR-celE, was produced by the introduction of an N-terminal signal peptide from the tobacco PR-1a gene ( Hammond-Kosack et al. 1994 ). There was a linear increase in EGE′ accumulated in the bathing media of protoplasts transfected with pPR-celE to at least 166 h post-transfection ( Fig. 1b). In contrast, there was no detectable EGE′ activity in the bathing media of protoplasts transfected with pcelE for at least 72 h. A slight increase in EGE′ activity was then observed after 72 h and suggests that some cell lysis may have occurred after this time.

The large amount of activity in the bathing media of cells transfected with pPR-celE demonstrates that passage through the ER is not detrimental to EGE′ activity and that there are no apparent cryptic signals within the EGE′ protein to prevent protein secretion. The EGE′ reporter protein was subsequently used to test the ability of GPI-addition signals to catalyse the addition of a GPI-anchor in tobacco suspension cells.

Yeast and mammalian GPI-addition signal sequences have different effects on EGE′ localisation in plant cells

GPI-targeting in plant cells was initially tested with previously characterised GPI-addition signals from yeast and mammalian sources. The GAS1 protein is a predominant plasma membrane protein of Saccharomyces cerevisiae and the sequence requirements of the C-terminal GPI-addition signal in yeast cells have been determined by saturation mutagenesis of the GAS1 signal ( Nuoffer et al. 1991 ; Nuoffer et al. 1993 ). The Thy-1 protein from Mus musculus was one of the initial mammalian GPI-anchored proteins to be characterised ( Homans et al. 1988 ) and its C-terminus has previously been utilised in GPI-targeting reporter protein studies in animal cells ( Soole et al. 1995 ).

The GAS1 and the Thy-1 C-terminal GPI-addition sequences were ligated in frame with the 3′ end of the celE′ gene in the secretion vector pPR-celE to produce pPR-celE-GAS and pPR-celE-Thy, respectively. Cells were transfected with these constructs and EGE′ production and localisation assessed in comparison to the cytosolic (pcelE) and secretion (pPR-celE) constructs.

Almost the entire EGE′ activity from pcelE was retained within the cells, whilst almost no activity was exported to the bathing media ( Fig. 2). In contrast, for cells transfected with pPR-celE, there was a dramatic increase in the export of EGE to the bathing media, demonstrating that secretion is the default pathway for proteins which are targeted to the ER. Even so, significant levels of activity are retained within cells transfected with pPR-celE. This may represent activity contained in various compartments of the default secretory pathway. This is consistent with other studies of protein secretion in tobacco cells ( Denecke et al. 1990 ).

Figure 2.

Comparison of the distribution of EGE′ activity associated with cells transfected with cytosol, secretion and GPI-targeting constructs (n = 5).

Values in each replicate experiment are normalised as a percentage of activity of pellets of pcelE transfected cells.

The introduction of the yeast GAS1 GPI-addition signal to the celE′ gene (pPR-celE-GAS) resulted in prevention of the export of EGE′ to the bathing media and almost complete association of EGE′ activity with the tobacco cells. This result is consistent with the addition of a GPI-anchor leading to retention of EGE′ on the outer surface of the plasma membrane. The level of EGE′ production with pPR-celE-GAS was found to be consistently higher than that observed with the cytosolic construct, pcelE. This may be due to a more favourable protein folding environment in the ER or a constraint on the amount of enzyme which can be ‘packaged’ in the cytosol, as such a constraint would not apply upon targeting to the ER and cell surface.

In contrast to the results obtained with the GAS1 signal, the addition of the Thy-1 GPI-signal to the celE′ gene (pPR-celE-Thy) did not greatly diminish EGE′ export to the bathing media. This indicates that the Thy-1 signal may not catalyse GPI-anchor addition in plant cells.

The EGE′-GAS1 reporter protein fusion shows hydrophobic properties

It would be expected that if a proportion of EGE′ activity, associated with the pellet fraction of pPR-celE-GAS or pPR-celE-Thy transfected cells, was linked to a GPI-anchor, it would display hydrophobic properties due to the presence of the lipid moiety. To test this, transfected cells were removed from the bathing medium and extracted in the detergent TritonX-114. This detergent has the property of undergoing phase partitioning (condensing into separate detergent and aqueous phases) at temperatures above 20°C ( Bordier 1981). Using this method, GPI-linked proteins and other hydrophobic membrane proteins are concentrated in the detergent-rich phase whilst hydrophilic proteins (including peripheral membrane proteins) are concentrated into the aqueous phase ( Lisanti & Rodriguez-Boulan 1990).

For the construct containing the GAS1 GPI-addition signal, a significant amount of cellular activity was found to partition into the detergent-rich phase ( Fig. 3). In contrast, there was negligible EGE′ activity recovered in the detergent phase of cells transfected with the cytoplasmic construct (pcelE), and only slightly more for the secretion construct (pPR-celE). The differences in detergent-partitioned activity cannot be accounted for by carry over contamination from the aqueous phase as the detergent phases were re-partitioned with fresh aqueous buffer multiple times until no more activity was re-partitioned into the aqueous phase. These data are consistent with the GAS1 signal catalysing the addition of a GPI-anchor to EGE′.

Figure 3.

The effect of GPI-addition signals on the partitioning of EGE′ activity into a detergent-rich phase (n = 4).

Values in each replicate experiment are normalised as a percentage of activity in the cell lysate of cells transfected with pcelE.

The detergent partitioning of activity for the Thy-1 signal fusion with EGE′ was similar to that observed for the secretion construct and these data suggest that the Thy-1 signal fusion has resulted in less GPI-linked EGE′ protein than the GAS1 signal fusion.

The GAS1 signal directs GPI-anchor addition in plant cells

The presence of a GPI-anchor on a detergent-partitioned protein can be conclusively demonstrated using PI-PLC, an enzyme that specifically cleaves the lipid component of GPI-anchored proteins ( Brodbeck & Bütikofer 1994). On the addition of aqueous buffer to the PI-PLC treated detergent phase, GPI-anchored proteins re-partition into the aqueous phase because of the removal of the lipid moiety of the anchor structure ( Lisanti et al. 1990 ). The PI-PLC sensitivity of detergent-partitioned activity obtained from pPR-celE-GAS transfected cells is shown in Fig. 4.

Figure 4.

Phase partitioning and PI-PLC treatment of EGE′ from pPR-celE-GAS transfected cells (n = 5).

Values for each phase are expressed as a percentage of the sum of activity in the detergent-rich and aqueous phases of each treatment.

On average, over eight separate experiments, 70.5 ± 2.7% of EGE′ activity, which initially partitioned into the detergent phase, re-partitioned into the aqueous phase upon incubation with PI-PLC. Under identical incubation conditions, but in the absence of PI-PLC, only a small amount of EGE′ activity (11.4 ± 4.7%) re-partitioned into the aqueous phase, with the remaining activity retained within the detergent phase. This background activity may be due to a low level of carry over contamination from the detergent-rich phase, as aqueous phases were not re-partitioned a second time with fresh buffer. Therefore, on average at least ∼60% of the detergent partitioned EGE′ protein is sensitive to PI-PLC digestion. These data conclusively demonstrate that the GAS1 GPI-addition signal from yeast can catalyse the addition of a GPI-anchor to a reporter protein in the plant cell culture system.

The C-terminus of the LeAGP-1 protein from tomato is a plant GPI-addition signal

Having demonstrated that a GPI-addition signal from yeast was processed in tobacco cells, the EGE′ reporter protein was used to evaluate a putative plant GPI-addition signal. The classical AGPs, AGPNa1 and AGPPc1, isolated from N. alata styles and P. communis suspension culture, respectively, have been postulated to be GPI-anchored proteins based on proteolytic cleavage of the C-terminus and the presence of ethanolamine at their C-terminal amino acid ( Youl et al. 1998 ). The C-terminal amino acid sequences derived from cDNA sequences of other classical AGPs also show characteristics similar to GPI-addition sequences from other eukaryote sources ( Schultz et al. 1998 ). Interestingly, however, in the case of AGPNa1 and AGPPc1, neither protein was found to have the lipid component of GPI attached following protein purification. A soluble form and a plasma membrane form of AGPs have been observed in ‘Paul’s Scarlet’ rose suspension cells with the membrane form susceptible to PI-PLC digestion, indicating the presence of the lipid component of GPI ( Svetek et al. 1999 ). We decided to determine if a C-terminal fusion of the putative GPI-addition signal from a classical AGP would lead to GPI-anchor attachment to the EGE′ reporter protein.

The LeAGP-1 protein from Lycopersicon esculentum (tomato) has been predicted by Schultz et al. (1998) to contain a C-terminal GPI-anchor addition signal with the site of GPI-anchor attachment (the ω site), to be located at position Ser192. Two EGE′ expression constructs containing different C-terminal coding regions of LeAGP-1 were produced: (a) pPR-celE-AGPFLT containing the full-length putative GPI-addition signal from Ser192-Phe215; and (b) pPR-celE-AGPTRU containing a truncated form of the signal from Glu195-Phe215 which excludes the predicted ω to ω+2 sites which have been shown to be a requirement for GPI anchor attachment in all other systems ( Udenfriend & Kodukula 1995).

As Table 1 shows, the level of EGE′ activity associated with cells transfected with both the full-length and truncated plant signals was comparable with the GAS1 signal. In this series of experiments, however, slightly higher levels of activity were observed in the bathing media for the GAS1 signal in comparison to earlier experiments, although this level of activity was negligible in comparison to the secretion construct, pPR-celE ( Fig. 2). A similar level of EGE′ activity was also observed in the bathing media with the full-length plant signal. It is unlikely that this activity could represent cell lysis as almost no activity was found in the bathing media with the truncated plant signal.

Table 1. . Comparison of the properties of EGE′ activity for cells transfected with constructs containing the GAS1 and LeAGP-1 (full-length and truncated) GPI-addition signals (n = 3)
  • *

    Activity normalised as percentage of pcelE cell lysate activity.

  • **

    Activity expressed as percentage of detergent partitioned activity sensitive to PI-PLC.

  • n.d. No activity detected.

Cell lysate *182 ± 21.8219.8 ± 28.2230.3 ± 23.6
Bathing media *29 ± 14.832.7 ± 13.91.6 ± 1.6
Detergent partitioned activity *41.9 ± 4.543.9 ± 2.674.0 ± 1.4
PI-PLC sensitive activity **59 ± 6.241.4 ± 2.8n.d.

Phase partitioning of cells extracted in TX-114 resulted in the recovery of almost identical levels of EGE′ activity into a detergent-rich phase for both the GAS1 and full-length plant signal ( Table 1). In both cases a large proportion of this activity was sensitive to PI-PLC digestion and subsequently repartitioned into an aqueous phase. This indicates that the full-length LeAGP-1 signal can catalyse the addition of a GPI-anchor and that the lipid component of GPI is present. In contrast, for the truncated plant signal almost double the level of EGE′ activity was recovered in the detergent-rich phase than was observed with the GAS1 and full-length plant signal. However, none of this activity was sensitive to PI-PLC digestion and hence subsequently remained in the detergent-rich phase after incubation with PI-PLC, indicating that a GPI-anchor was not present.


We have previously demonstrated the presence of GPI-anchored proteins in terrestrial plant cells using a biotin-labelling approach with N. tabacum suspension cells ( Takos et al. 1997 ). Other groups have subsequently identified GPI-anchored proteins in a number of plant species based upon protein sensitivity to PI-PLC digestion and a subsequent shift in hydrophobic profile, or labelling of proteins with radioactive components of GPI, or structural analysis of GPI-anchor components associated with purified proteins ( Kunze et al. 1997 ; Nakazato et al. 1998 ; Svetek et al. 1999 ; Youl et al. 1998 ). In this study we have taken an alternative approach of fusing GPI-addition signals to a reporter gene in an attempt to further characterise the GPI-anchor processing pathway in plant cells.

To our knowledge, the celE′ gene from C. thermocellum has not previously been used as a reporter gene in plant cells. The EGE′ protein may prove to be an exceptional choice for delineating molecular processes in plant cells, particularly those involving transport of proteins through the secretory pathway. Other commonly used plant reporters such as firefly luciferase and β-glucuronidase (GUS) have limitations in this field of study because of poor protein stability, inability to withstand N-and C-terminal fusions or to traverse secretory organelles without diminished activity ( Iturriaga et al. 1989 ). Our results with a tobacco transient expression system showed that EGE′ activity was stable over several days ( Fig. 1a), the activity was maintained in the presence of both N- and C-terminal fusions within the same construct and high levels of activity could be exported to the bathing media by the default pathway of protein secretion ( Fig. 1b).

Our results show that C-terminal signals from yeast (the GAS1 protein) and plant (the LeAGP-1 protein) sources could each direct the addition of a GPI-anchor to the EGE′ reporter protein in tobacco protoplasts. These signals appear to be recognised with approximately equal efficiency as they both prevent the export of significant levels of EGE′ to the media bathing cells and resulted in the production of similar levels of hydrophobic EGE′ activity (partitioned into a detergent-rich phase) which was susceptible to PI-PLC digestion ( Table 1). The results obtained for the mammalian Thy-1 signal, however, seemed to indicate that this signal was poorly processed in tobacco cells. EGE′ fused to the Thy-1 signal was exported in high levels to the media bathing cells ( Fig. 2) and resulted in the production of only a relatively low amount of hydrophobic EGE′ activity ( Fig. 3).

A truncated LeAGP-1 signal, containing a deletion of three amino acid residues from Ser192 to Ala194, was not processed for GPI-anchor addition to the EGE′ protein ( Table 1). Schultz et al. (1998) predicted that these three residues comprised of a motif that has been found to be common to all other eukaryote GPI-addition signals and to be essential for cleavage of the C-terminal signal and replacement with a GPI-anchor. The data support that this cleavage motif is also conserved in plant GPI-addition signals, and that Ser192 is the site of GPI-anchor linkage to the LeAGP-1 protein. Unlike the Thy-1 signal, however, the truncated LeAGP-1 signal completely prevented the export of EGE′ activity to the bathing media surrounding cells ( Fig. 2 and Table 1). These data are most likely explained by the retention of EGE′ at the plasma membrane or within an ER compartment by the uncleaved hydrophobic polypeptide domain of the LeAGP-1 signal, as has been observed to occur for uncleaved GPI-addition signals in other systems ( Caras et al. 1989 ; Moran & Caras 1992).

If the ω, ω+1 and ω+2 sites of the yeast GAS1 signal are considered in isolation, it may not be surprising that this signal is processed efficiently in the plant system. The residues at these sites are asparganine, alanine and alanine, respectively, and this same cleavage motif is predicted for the putative plant GPI-signals of AGPNa1 and AtAGP5 ( Table 2). In contrast, the Thy-1 signal has cysteine at the ω site followed by two glycine residues ( Table 2). Cysteine is not found at this position in any of the yeast or putative plant GPI-addition signals identified to date ( Furukawa et al. 1997 ; Hamada et al. 1998 ; Schultz et al. 1998 ). Furthermore, the analysis of GPI-signal sequence requirements at the ω site in S. cerevisiae, conducted by Nuoffer et al. (1993) , is strikingly consistent with the processing of the signals observed in this study. An asparagine or serine residue at the ω site were the most efficiently processed amino acids for GPI-anchor addition in the yeast system, and these amino acids are found at the ω site of the GAS1 and LeAGP-1 signals, respectively ( Table 2). In contrast, a cysteine residue at the ω site (as is the case with Thy-1) was found to be the least efficiently processed amino acid of the six possible residues that could be processed at the ω position in yeast cells. It is also interesting to note that in the yeast study, poorly processed signals resulted in increased export of protein to the media bathing cells, as was the case for Thy-1 in this study ( Fig. 2). Saturation mutagenesis of a single defined GPI-addition signal is necessary before any definite signal requirements for GPI-anchor addition can be determined in plants. However, the preliminary data suggest a common preference against cysteine at the ω site in yeast and plant cells.

Table 2. . C-terminal amino-acid sequences of GPI-anchor addition signals
SourceGPI-anchor addition amino acid sequence
  1. The ω, ω+1 and ω+2 sites are shown in bold.


For both the GAS1 and full-length LeAGP-1 signal, it was perhaps unexpected to observe that only about 25% of the EGE′ protein synthesized could be partitioned into a detergent rich phase ( Table 1). These data can be explained by saturation of the GPI-addition pathway by EGE′ protein produced under the direction of the strong viral transcription elements in the expression vector. We suspect that the GPI-addition pathway may not be highly active in this plant cell line. Presumably the remainder of activity, which partitioned into the aqueous phase, was retained within cells (or at the cell surface) by the unprocessed GPI-addition signal, as was the case for the truncated plant signal. For the truncated plant signal, however, there was significantly more EGE′ activity in the cell lysate and almost double the amount of EGE′ activity was recovered in the detergent phase than for the full-length GPI-signals, although none of this activity is sensitive to PI-PLC and is probably due to the hydrophobic domain of the unprocessed signal ( Table 1). A possible explanation for this observation would be cleavage of the full-length GPI-addition signals without anchor addition. Although GPI-signal cleavage has generally been thought to only occur concomitant with anchor addition, it has been shown for alkaline phosphatase (ALP), in mammalian cells, that upon inhibition of GPI-anchor synthesis, cleavage without anchor addition may occur thus resulting in secretion of ALP into the media bathing cells ( Takami et al. 1992 ). Similarly, in a mutant lymphoma cell line deficient in Dol-P-Man synthetase (an enzyme required for GPI-anchor synthesis), Thy-1 protein lacking GPI-anchor modification is also secreted into the bathing media ( Fatemi & Tartakoff 1986). These studies seem analogous to the saturation of the pathway with EGE′ protein in the plant system. Thus, some of the EGE′ protein fused to the GAS1 and full-length LeAGP-1 signal would have the hydrophobic C-terminal domain of the signal cleaved and instead of EGE′ retained at the surface by a GPI-anchor, it would proceed by the default pathway of secretion, and hence less protein is partitioned into a detergent-rich phase. This also explains the low level of EGE′ activity observed in the bathing media for the GAS1 and the full-length plant signal but not for the truncated plant signal.

The export of high levels of EGE′ to the bathing media for the Thy-1 signal suggests that the hydrophobic domain of the Thy-1 signal must be removed. If the cysteine residue at the ω site is indeed not processed efficiently then the removal of the hydrophobic domain may be due to processing without concomitant anchor addition, but in this case to a greater extent than that which may occur for the GAS1 and LeAGP-1 signals. Replacement of the ω site amino acid of a plant signal (or even the GAS1 signal) with cysteine would determine if indeed the cysteine residue is the cause of the apparent poor processing of Thy-1 in plant cells. This would eliminate other possibilities that may account for EGE′ export to the bathing media such as proteolytic cleavage at an alternate site in the Thy-1 signal fusion, removal of the lipid anchor by an endogenous phospholipase, or other characteristics of the Thy-1 signal, such as properties of the spacer region or hydrophobic domain.

GPI-anchor biosynthesis occurs by a complex pathway involving at least 10 separate steps on the cytosolic side of the ER membrane before the completed GPI-anchor intermediate is flipped to the lumen side of the ER ( Takeda & Kinoshita 1995). The conservation of GPI-anchor signal processing in plant cells means that an analogous anchor biosynthesis pathway must also exist in plant cells. To date, we have identified several clones homologous to yeast and animal GPI-anchor biosynthesis genes (data not shown). Future work will utilise these clones in gene-disruption studies to attempt to elucidate the role of GPI-anchored proteins in plant cells.

Experimental procedures


All biochemicals including Murashige and Skoog (MS) basal salt mix, 4-methylumbelliferyl-β- d-cellobiopyranoside and 4-methylumbelliferone were purchased from Sigma (St. Louis, MO, USA). Protease inhibitors and Bacillus cereus PI-PLC were from Boehringer Mannheim Australia (Castle Hill, NSW, Australia). Cellulase (Onozuka) RS was obtained from Yakult Pharmaceutical (Tokyo, Japan) and pectolyase Y-23 from Seishin Corporation (Tokyo, Japan). The expression plasmid pRTL2-GUS/NIaδBam and pSLJ6069 containing the PR-1a gene were the kind gifts of Dr James Carrington (Washington State University, USA) and Prof. Jonathan Jones (John Innes Centre, UK), respectively. Plasmids containing the GPI-addition sequences for GAS1, Thy-1 and LeAGP-1 were the kind gifts of Dr L. Popolo (University of Milan, Italy), Prof. H. Gilbert (University of Newcastle, UK) and Dr C. Davies (CSIRO Division of Horticulture, Adelaide, Australia), respectively.

Plasmid construction

Plant expression cassettes were constructed using standard molecular biology techniques and were based on the pRTL2-GUS/NIaδBam vector ( Restrepo et al. 1990 ). Initially the region encoding GUS in this vector was replaced with a NcoI-BglII Photinus pyralis luciferase (luc) gene fragment from the vector pSP-luc + NF (Promega) to produce pRTL2-luc. The BglII site had been introduced by site-directed mutagenesis in place of the XbaI site at the 3′ end of the luc gene. The pRTL2-luc construct was then used to construct celE′ expression vectors by replacement of the luc gene. The celE′ gene from C. thermocellum was amplified from pSRαEGE’THY1 ( Soole et al. 1995 ) and a stop codon following a BglII site introduced in frame with the open reading frame of celE′. The amplified fragment was ligated into the T-tailed EcoRV site of pBluescript II SK ± (Stratagene). From this plasmid designated pBSII-celE, the celE′ gene was excised with BstEII and XbaI and ligated in the place of the luc gene fragment digested with these same enzymes in pRTL2-luc to create a vector for the cytosolic targeting of EGE’, designated pcelE.

To create a vector for the export of EGE′, the sequence of the ER targeting signal of the PR-1a gene from pSLJ6069 ( Hammond-Kosack et al. 1994 ) was amplified. The pcr fragment was cloned into NcoI and BstEII sites at the 5′ end of the luc gene in pRTL2-luc, the luc gene was then excised and replaced with the celE′ gene, as detailed above, and the plasmid designated pPR-celE.

To create vectors for GPI-anchor addition to EGE′, the PR-1a pcr fragment was cloned into pRTL2-luc as above but the luc gene was excised with BstEII and BglII and replaced with the celE′ gene excised from pBSII-celE with these same enzymes. A stop codon immediately after the BglII site in pBSII-celE was therefore not cloned into the pRTL2 vector and in frame fusions could be generated with the 3′ end of the celE′ gene using BglII and XbaI sites. Sequences coding for GPI-anchor addition signals were pcr amplified from the S. cerevisiae GAS1 gene, the M. musculus Thy-1 and the L. esculentum LeAGP-1 gene, using a 5′ primer containing a BglII site and a 3′ primer containing a XbaI site after the stop codon. Primer sequences used in the construction of each plasmid are shown in Table 3.

Table 3. . Primers used to generate DNA fragments for EGE′ expression constructs
Amplified sequence *5′-primer **3′-primer **
  • *

    Refer to text for details of template source.

  • **

    Restriction endonuclease site shown in bold.


Plant cell culture

N. tabacum (NT-1) cells were maintained in suspension culture by subculturing 1 : 25 every 7 days in 25 ml of cell culture medium (CCM) as described in Gao et al. (1991) . For protoplast isolation, 7-day-old cells were subcultured 1 : 10 in 50 ml of CCM and grown for 4–5 days. All cultures were incubated at 25°C on an orbital shaker at 100 rpm with a 12 h day/night cycle.

Protoplast preparation

NT-1 cells (50 ml of suspension culture) were pelleted in a swing out rotor for 5 min at 100× g. The supernatant was discarded and the pellet washed in 50 ml of wash solution (0.4 m mannitol, 20 m m Mes, pH 5.8). Cells were resuspended in a solution containing cellulase RS (1.0% w/v), pectolyase Y-23 (0.1% w/v), 0.4 m mannitol and 20 m m Mes, pH 5.8 at 1–1.5 ml g−1 wet weight of cells. Cell wall digestion was for 1 h at 25°C on an orbital shaker at 40 rpm.


Protoplasts were resuspended in 50 ml of wash solution and pelleted for 5 min at 100× g, the supernatant was discarded and the pellet washed 2× in 50 ml wash solution and then 1× in 50 ml of electroporation buffer (150 m m NaCl, 5 m m CaCl2, 0.2 m Mannitol, 10 m m Hepes, pH 7.2) at 4°C. Protoplasts were left resuspended in electroporation buffer at 1.5 ml g−1 wet weight of cells. Into an electroporation chamber (0.4 cm gap, Life Technologies) was placed 20 μg of plasmid DNA, electroporation buffer and 5 × 106 protoplasts in a total volume of 800 μl. The chamber contents were mixed gently, the chamber placed on ice for 5 min, then mixed again immediately before electroporation with the BRL cell porator system 1 (Life Technologies). Electroporation settings were 250 V (625 V cm−1) charge and 800 uF capacitance. Chambers were immediately placed on ice for at least 10 min before transferring protoplasts to tissue culture plates containing 5 ml of protoplast culture medium (which is identical to CCM but with the addition of 0.4 m mannitol and 20 m m Mes). Plates were sealed and incubated in the dark at 25°C for 2–3 days.

Extraction of EGE′

Protoplasts were transferred to 10 ml centrifuge tubes with a wide bore pipette tip and pelleted at 100× g. For assay of activity exported to the bathing media, 2 ml samples of the supernatant were taken and clarified by centrifugation at 100× g and then at 15 000× g. After removal of the remainder of the bathing media, cells were resuspended in 1.5 ml of wash solution, transferred to microfuge tubes and pelleted at 100× g. The supernatant was removed and the pellet snap frozen in liquid nitrogen. Pellets were extracted by grinding in either 600 μl of TX-114 lysis buffer containing 1% (v/v) TritonX-114 in Tris-buffered saline (TBS, pH 7.5) for phase separation experiments or in Cell Culture Lysis Reagent (Promega) containing 25 m m Tris-phosphate pH 7.8, 2 m m DTT, 2 m m 1,2-diaminocyclohexane-N,N,N’,N′-tetraacetic acid, 10% v/v glycerol, 1 v/v TritonX-100 for experiments not requiring phase separation. Extracts were clarified by centrifugation twice at 15 000× g. All manipulations were carried out on ice and all centrifugation steps at 4°C. Extraction buffers and all subsequent solutions used for phase separation and PI-PLC incubation contained protease inhibitors leupeptin (20 μg ml−1), pepstatin A (10 μg ml−1), antipain (10 μg ml−1) and PMSF (0.1 m m).

Phase separation and PI-PLC treatment

The method used for phase separation and PI-PLC treatment has been described previously by Takos et al. (1997) . Briefly, extracts from electroporations were pooled (5 electroporations for each construct to be tested), layered onto a 0.3 m sucrose cushion and phase partitioned at 37°C for 10 min. Detergent pellets containing GPI-anchored and other hydrophobic proteins were then collected by centrifugation at 10 000 ×g at RT. These pellets were then resuspended to a volume of 1 ml in 0.12% v/v TX-114, TBS and the phase partition and collection of pellets repeated as above. For PI-PLC incubation, detergent pellets were resuspended in buffer, split into a final volume of 500 μl each and incubated at 37°C for 6 h with either B. thuringiensis PI-PLC at 6 U ml−1 in a buffer described in Lisanti et al. (1990) or B. cereus PI-PLC at 1.2 U ml−1 in 70 m m triethanolamine, pH 7.5, 1 m m EDTA. Subsequent phase partitioning isolates GPI-anchor cleaved proteins in an aqueous phase.

EGE′ enzymatic assay

EGE′ catalytic activity was determined in cell lysates, media and phase partition samples by fluorometric assay essentially as previously described by Soole et al. (1995) . Samples (100 μl) were mixed with 250 μl of 100 m m phosphate buffer (pH 6.0) containing 1.4 m m 4-methylumbelliferyl-β- d-cellobiopyranoside (1 m m final) and incubated at 60°C for 3 h. The reaction was stopped by the addition of 2 ml of 0.5 m Glycine pH 10.4. Fluorescence was measured at an excitation wavelength of 365 nm and emission wavelength of 460 nm. Fluorescence measurements from samples stopped at zero time were subtracted from the fluorescence of samples incubated for 3 h and enzyme activity calculated using standards of 4-methylumbelliferone. Since TX-114 can quench fluorescence, samples containing TX-114 were clarified after the addition of 0.5 m Glycine by centrifugation at 1000 ×g for 10 min at 37°C and the supernatant was then used for fluorescence measurement.


We thank Prof. Howard Riezman and Dr Christa Niemietz for their helpful advice in the preparation of this manuscript.