A soluble secretory reporter system in Trypanosoma brucei. Studies on endoplasmic reticulum targeting.

A homolog of the endoplasmic reticulum (ER) hsp70 protein, binding protein (BiP), from the parasitic protozoan Trypanosoma brucei (Bangs, J. D., Uyetake, L., Brickman, M. J., Balber, A. E., and Boothroyd, J. C. (1993) J. Cell Sci. 105, 1101-1113) is further characterized. In co-precipitation experiments, BiP transiently associates with newly synthesized secretory proteins, including variant surface glycoprotein (VSG), confirming its role as a molecular chaperone. To study the molecular signals targeting BiP to the ER, we have developed soluble secretory reporters for expression in transformed procyclic trypanosomes. Deletion of the BiP C-terminal tetrapeptide (MDDL) and the glycosylphosphatidylinositol-anchor addition sequence of VSG converts these proteins to secreted forms. Attachment of MDDL to VSG results in intracellular retention confirming that MDDL is a trypanosomal ER localization signal. Secretion of both reporters is inefficient, but further truncation of the BiP C-terminal peptide-binding domain allows quantitative export (t1/2 ~1 h) of the N-terminal ATPase domain (BiPN), consistent with the conserved domain structure of hsp70 proteins. This is the first demonstration of soluble protein secretion in African trypanosomes. Using the BiPN reporter, the sequence specificity of C-terminal tetrapeptide retention signals in trypanosomes is analyzed and found to be similar to higher eukaryotes. These results indicate that the basic signals mediating protein targeting to the ER lumen are conserved throughout the wide range of eukaryotic evolution.

African trypanosomes, parasitic protozoa of the order Kinetoplastida, represent one of the most ancient of all eukaryotic lineages (1). They have a life cycle that alternates between an insect vector, tse-tse flies, and mammalian hosts, in which they cause serious disease. Considerable effort has been made in the study of trypanosome molecular biology resulting in the elucidation of many unusual molecular and biochemical pathways, such as trans-splicing and RNA editing. More recently, interest in the basic cell biology of these organisms has grown, and significant progress has been made in areas such as organellar biogenesis (2) and endocytosis (3). One area, however, that has lagged behind is that of secretion and secretory protein trafficking.
Most of what is known about secretory trafficking in trypanosomes comes from biosynthetic studies of VSG, 1 a homodimeric GPI-anchored protein that is the major surface antigen of the bloodstream stage (4 -6). Although other exported plasma membrane proteins are known (7,8), no soluble secretory polypeptides have yet been identified in Trypanosoma brucei. Export of various "factors" (9,10), as well as soluble protease activities (11), has all been reported. Unfortunately, none of these have been characterized at the molecular level. All exocytic and endocytic trafficking in trypanosomes occurs at an invaginated plasma membrane microdomain called the flagellar pocket (3,12). Since the opening of the lumen of the pocket to the external environment is restricted, it may be that exit of soluble molecules is not possible. However, fluid phase markers can enter the pocket (13), and secretion of soluble polypeptides does occur in the related trypanosomatids, Trypanosoma cruzi (14) and Leishmania donovani (15,16). Thus, the absence of soluble secretion in T. brucei most likely reflects the failure to detect such molecules rather than the true biological situation.
The VSG biosynthetic studies suggest that a relatively typical eukaryotic secretory pathway is operative in trypanosomes, and membranous structures morphologically resembling the ER and Golgi can be discerned by electron microscopy (13,17). However, molecular markers for the secretory subcompartments are few. A trypanosomal homolog of BiP has been characterized, and this serves as an ER marker (18), but no other immunological marker proteins are available for cell biological investigations. Not surprisingly then, little is known about the molecular signals that direct the intracellular trafficking and targeting of the protein components of the trypanosomal secretory pathway.
One of the best documented examples of protein targeting in eukaryotic cells is the localization of soluble proteins, such as BiP, to the lumen of the ER (reviewed in Ref. 19). These proteins have C-terminal tetrapeptide sequences that serve as specific ER localization signals. The consensus tetrapeptide motif is XDEL where the amino acid X varies in a speciesspecific, and sometimes protein-specific, manner. Typical examples are KDEL in mammals, HDEL in Saccharomyces cerevisiae, DDEL in Kluyveromyces lactis, and ADEL in Schizosaccharomyces pombe (19,20). All of the aforementioned sequences have been tested functionally in transfection experiments. Sequences in other species such as SDEL in Plasmo-dium falciparum (21) and KDEL in Giardia lamblia (22) are presumed by analogy to be retention signals. Two putative signals have been found in trypanosomes, MDDL and KQDL at the C termini of BiP (18) and a protein disulfide isomerase homolog (23), respectively. Given the relative divergence of the trypanosomal sequences and the great phylogenetic distance between trypanosomes and species where ER targeting has been assayed directly, it is essential that these sequences be demonstrated to actually mediate ER localization.
The main obstacle to the study of protein trafficking in trypanosomes has been the inability to express recombinant genes in these organisms. In recent years this technological hurdle has been overcome (24 -26), and stable transformation vectors have been exploited to study both mitochondrial and glycosomal protein targeting (reviewed in Ref. 2). In this report, we use similar techniques to investigate secretory protein trafficking in T. brucei. First, we have confirmed that trypanosomal BiP functions as a molecular chaperone by demonstrating its transient physical interaction with an endogenous secretory protein, VSG. Second, we have developed soluble secretory reporters that are efficiently exported to the extracellular environment when expressed in procyclic cells, explicitly demonstrating for the first time that soluble secretion in African trypanosomes is possible. Finally, we have exploited these recombinant reporters to study the sequence specificity of ER targeting signals. Our results demonstrate that C-terminal tetrapeptides mediate ER localization in trypanosomes and define the sequence specificity of this phenomenon. This work suggests that ER localization occurs by a mechanism that is common to most, if not all, eukaryotic organisms.
Immunoprecipitation and Electrophoresis-Affinity purified rabbit anti-VSG117 antibody and whole rabbit anti-BiP antiserum have been described (18). Rabbit anti-VSG 221 serum and affinity purified rabbit anti-HA-9 peptide antibody were generous gifts of Dr. John Boothroyd (Stanford University) and Dr. David Russell (Washington University), respectively. Rabbit anti-␤-glucuronidase (GUS) was purchased from CLONTECH Laboratories, Inc. (Palo Alto, CA).
Construction of Trypanosome Expression Vectors-Procedures were performed by protocols according to either Ref. 29 or manufacturer's instructions. PCR products were cloned into pBSIISK (Stratagene Inc., San Diego, CA), and the ends were sequenced (Sequenase kit, United States Biochemical Corp.) to confirm identity.
The stable transformation vector, pXS2, was constructed from p205CATϩA (a generous gift of Dr. John Boothroyd, Stanford University), which contains in order (5Ј to 3Ј in pBSIIKS) the PARP B locus promoter region (Ϫ1044 to ϩ22 bp relative to the B2␣ splice acceptor site (30)) in the pBSIIKS HincII site; the CAT gene in the pBSIIKS HindIII site; and an EcoRI/BglII fragment containing the ␤␣-tubulin intragenic region (TUBint, Ϫ108 to ϩ532 bp relative to the ␤-tubulin 3Ј end) (31) in the pBSIIKS EcoRI/BamHI sites. p205CATϩA was first modified by deletion of the CAT-bearing HindIII fragment and by elimination of three restriction sites (AscI at 896 bp upstream of the procyclin B2␣ splice acceptor site, EcoRI at the 5Ј end of TUBint, and BstXI in the 3Ј pBSIIKS polylinker) with separate rounds of cutting, filling in, and ligation. Second, the 795-bp Neo gene was PCR-amplified using kinased primers JB55/JB56 and the plasmid pRC/RSV (Invitrogen, San Diego) as template. Third, the PARP B2␣␤ intragenic region (PARPint, 937 bp from the B2␣ 3Ј end to 3 bp upstream of the B2␤ start codon) was PCR-amplified using kinased primers JB57/JB58 and the cloned procyclin B2 locus (a generous gift of Dr. Keith Wilson, Stanford University) as template. Both PCR products were EcoRV/SmaI digested and cloned into the EcoRV site of pBSIISK generating pNeo and pPARPint. The PARPint EcoRV/SmaI fragment was then inserted into the EcoRV site of pNEO to generate a PARPint:NEO fusion (pPIN) with unique flanking 5Ј EcoRV and 3Ј SmaI sites. This fragment was then inserted into the EcoRV site of modified p205CATϩA to create pXS2 with a multicloning region between the B2␣ trans-splice acceptor site and PARPint (Fig. 1A).
Construction and Epitope Tagging of Reporter Genes-Epitope-tagging vectors were constructed for the introduction of the nine amino acid (YPYDVPDYA) HA9 epitope (32) into reporter constructs. Deoxyoligonucleotides JB52/JB53 were annealed to create a double-stranded insert encoding the HA9 epitope and having EcoRI-compatible singlestranded overhangs. This was cloned into the EcoRI site of pBSIISK in both orientations such that the 5Ј end of the insert coding strand is proximal to the KpnI (pHA9F) or SacI (pHA9R) sites of the pBSIISK polylinker. The EcoRI site is preserved at the insert 3Ј end only. These vectors allow the in-frame fusion of any coding region to the HA9 epitope using unique NsiI and NheI sites at the 5Ј end of the insert in conjunction with any site in the pBSIISK polylinker.
Reporter gene cassettes (Fig. 1B) were generated by PCR using the following cloned genes as template: pA4 containing the T. brucei BiP gene (18); pGB117 containing the T. brucei VSG 117 basic copy gene (a generous gift of Dr. John Boothroyd, Stanford University); and pGUSN358-S (CLONTECH Laboratories) containing the bacterial ␤-glucuronidase gene. Products were cloned into either pBSIISK, pHA9F, or pHA9R, and the ends were sequenced to confirm identity. Using unique flanking restriction sites, reporter cassettes were then inserted into the pXS2 multicloning region to generate pXS2:reporter constructs. In some cases, products were cloned directly into pXS2, and the 3Ј end of the inserts were sequenced using primer JB85.
Transformation of Trypanosomes-Mid-log phase (ϳ5 ϫ 10 6 cells/ml) procyclic trypanosomes were washed once in PBS and resuspended at 4 ϫ 10 7 /ml in Optimem medium (Life Technologies, Inc.). Qiagenpurified (QIAGEN Inc, Chatsworth, CA) pXS2:reporter plasmids were linearized using a unique BstXI site in the ␤␣-tubulin intragenic region cassette, precipitated, washed in 70% ethanol, air-dried, and dissolved at 1.0 mg/ml in Optimem. 0.5 ml of cells and 0.1 ml of linear DNA were mixed in 0.4-mm cuvettes and electroporated (24 ⍀, 1.5 kV) in a BTX600 cell manipulator (BTX Inc., San Diego, CA). Cells were transferred to 10-ml TM-P cultures, and, at 48 h, G418 was added to 50 g/ml. Stable cell lines typically grew out in 2-3 weeks and were maintained in 25 g/ml G418.

RESULTS
Functional Analysis of Endogenous BiP-Previous analyses, based on sequence homology and subcellular localization, indicate that we have identified the trypanosomal homolog of the ER hsp70 protein, BiP (18). As such, it would be predicted to act as a molecular chaperone by facilitating the folding and assembly of secretory proteins (33). Therefore, prior to investigation of the molecular signals responsible for intracellular targeting of the trypanosomal BiP homolog, we wished to confirm that it has chaperone function as judged by transient physical interaction with newly synthesized secretory proteins. VSG, the major secretory protein of bloodstream trypanosomes, comprising 10 -20% of total cell protein (34), is the ideal choice for such an assay.
Bloodstream trypanosomes were pulse/chase-radiolabeled, and polypeptides were immunoprecipitated from cell extracts with anti-BiP or anti-VSG ( Fig. 2A). The total amount of each labeled protein (lanes 1-5, VSG; lanes 6 -10, BiP) remained constant throughout the chase period. More importantly, at the earliest chase time, an abundant newly synthesized protein of the same mobility as VSG co-precipitated with anti-BiP (lane 6). This association was transient, disappearing rapidly during the chase period (lanes 7-10). Additionally, a set of high M r proteins were also BiP-associated in a transient, although more prolonged manner ( Fig. 2A, lanes 6 -10). The identity of the major BiP-associated protein was investigated by sequential immunoprecipitation (Fig. 2B). Following denaturation of a primary anti-BiP immunoprecipitate (lane 2), VSG could be specifically reprecipitated (lane 3) confirming that it is the major BiP-associated protein. Radiolabeled VSG is readily detected in anti-BiP precipitates since a significant portion of newly synthesized VSG is associated with the total BiP pool (compare lanes 1 and 2). However, the converse is not true (lane 1), since BiP is synthesized at a much lower rate and little of the radiolabeled pool is likely to be associated with VSG. These results indicate that BiP does interact physically with VSG, consistent with its role as a secretory molecular chaperone, and validate its use as a reporter for ER localization studies.
Transformation of Trypanosomes-A vector for the stable transformation of cultured procyclic trypanosomes (pXS2, Fig.  1A) was constructed based on the previous work of others (24 -26). Prior to transfection, the vector is linearized at a unique BstXI site in the tubulin intragenic region which targets homologous recombination to the tubulin locus. pXS2 was tested using GUS as an expression reporter gene. A stably transformed GUS cell line was prepared and analyzed in two ways. First, correct targeted integration into the chromosomal tubulin locus was confirmed by Southern analysis. 2 Second, GUS expression was readily detected by both enzymatic assay of cell extracts 2 and by specific immunoprecipitation of GUS protein from metabolically radiolabeled cells (Fig. 3A, lane 1).
Secretion of Epitope-tagged BiP-In order to analyze the molecular signals mediating ER localization in trypanosomes it was first necessary to develop soluble reporters that are efficiently exported. To accomplish this, endogenous polypeptides of the trypanosomal secretory pathway, albeit cell-associated, were adapted as soluble reporters. First, modified BiP genes (Fig. 1B) were generated in which the putative C-terminal ER localization tetrapeptide, MDDL, was replaced with the viral HA9 immuno-epitope to allow discrimination of recombinant from endogenous BiP protein. This reporter was prepared as a secretory form with the native N-terminal signal sequence (BiPHA9) and as a cytoplasmic form (BiPHA9⌬ss) with the signal sequence deleted. If MDDL is indeed a retention signal, then it is predicted that BiPHA9 will be secreted from transformed cells in a soluble manner. Using pXS2, stable cell lines bearing each reporter gene were prepared and analyzed by specific immunoprecipitation of radiolabeled recombinant proteins (Fig. 3) , lanes 8 and 9). As expected, no BiPHA9 reporter was detected with this reagent in control GUS cells (Panel A, lane 7). A set of high M r cellular proteins, similar to those seen in bloodstream trypanosomes (Fig. 2), co-precipitated with BiP (Panel A, lanes 4 -6); in longer exposures 2 an identical set of proteins is associated with recombinant secretory BiPHA9, but not cytoplasmic BiPHA9⌬ss (Panel A, lane 8 and 9, respectively). The identity of these proteins, which are consistently associated with BiP (see also Fig. 7), is not known, but the transient nature of the association (see Fig. 6) suggests that they may be endogenous secretory proteins or other ER chaperones involved in secretory protein folding. Significant BiP polypeptide was detected in the culture supernatant of BiPHA9 cells (Panel B, lane 5); three results indicate that this represents "classical" secretion of recombinant BiPHA9 polypeptide. First, no endogenous BiP was detected in supernatants of control GUS cells (Panel B, lane 4). Second, anti-HA9 detects a polypeptide of identical size in BiPHA9 culture supernatants (Panel B, lane 8). Third, export of BiPHA9 is signal sequence-dependent (compare Panel B, lanes 5 and 6). These results clearly demonstrate that BiPHA9 is being secreted in a soluble manner from transformed T. brucei and also suggest that the C-terminal tetrapeptide, MDDL, does indeed serve as an ER localization signal in trypanosomes.
Although BiPHA9 is secreted, export of this reporter is not robust since the amount detected in the medium is low relative to the total synthesized during the 4-h labeling period. This was confirmed by a pulse/chase experiment in which export of BiPHA9 was monitored by immunoprecipitation with anti-HA9 antibody over a 24-h period (Fig. 4). Two observations can be made from this experiment. First, all of the labeled reporter disappears from the cell fractions during the chase (lanes 1-6) with an estimated t 1 ⁄2 of ϳ8 h. Thus, although export is efficient, it is very slow. Second, the amount of exported reporter apparent after 24 h (lane 12) is less than the total at time 0 (lane 1) suggesting that some intracellular degradation may be occurring. No attempt was made to accurately determine the efficiency of export since anti-HA9 antibody does not give quantitative recovery of the recombinant protein (see Fig. 3). Identical experiments were performed in which export of endogenous BiP from untransformed Pro1 cells was monitored for 24 h. 2 These experiments revealed that no endogenous BiP is exported, even in prolonged chases, and that endogenous BiP is extremely stable (t 1 ⁄2 for turnover Ͼ24 h).
Retention of a Heterologous Reporter-To confirm that MDDL is an ER localization signal it was necessary to demonstrate that this sequence could mediate the intracellular retention of a heterologous secretory reporter. For this purpose we chose VSG, which is expressed only in the bloodstream stage, obviating the need to epitope-tag this reporter for expression in procyclics. To modify VSG for secretion it was necessary to delete the C-terminal hydrophobic peptide sequence that mediates addition of the GPI membrane anchor (35). Therefore, we have engineered this gene in two forms (Fig. 1), one encoding a VSG 117 protein minus the C-terminal tail (117⌬gpi) and one in which the GPI anchor sequence has been replaced with MDDL (117MDDL). Again, transformed cell lines were assayed by specific immunoprecipitation of radiolabeled reporter polypeptides (Fig. 5). Identical amounts of radiolabeled endogenous BiP were detected in all cell lines indicating that labeling conditions were equivalent for each group. 2 Recombinant VSG polypeptides of the expected size (ϳ59 kDa) were detected in cell extracts of both transformed cell lines (lanes 2 and 3) but not in control Pro1 cells (lane 1). An overexposure is deliberately presented for direct comparison of cell and media fractions; lower exposures of these lanes reveal single polypeptide bands of slightly different electrophoretic mobility, 2 117MDDL is larger (ϳ2 kDa) than 117⌬gpi because the HA9 epitope was included in the process of adding MDDL to the C terminus. The 117⌬gpi reporter is exported to the culture supernatant (lane 5) in an N-terminal signal sequence-dependent manner, 2 confirming that soluble export is possible in procyclic trypanosomes. However, as with the BiPHA9 reporter, both the low ratio of VSG secreted to VSG synthesized (compare lanes 2 and 5) and pulse/chase experiments 2 indicate that export is inefficient. Nevertheless, attachment of the tetrapeptide, MDDL, to the C terminus of 117⌬gpi abrogates the observed secretory signal (lane 6) confirming that MDDL can mediate intracellular retention of a heterologous reporter.
Secretion of the BiP ATPase Domain-Having confirmed, as predicted, that MDDL can mediate intracellular retention, we wished to examine the sequence specificity of retention more closely. We initially anticipated that some modifications of this tetrapeptide would result in partial retention phenotypes, that is a retardation of export rather than quantitative retention. Therefore, it was necessary to develop a reporter that is exported with high efficiency so that leaky phenotypes could be detected. In addition, the slow kinetics of secretion of the BiP and VSG reporters leaves the issue of soluble secretion in trypanosomes open to considerable doubt. Failure of the VSG reporter was not so troubling since this protein is not normally expressed in procyclic trypanosomes. However, BiP is endogenous to both stages (18) necessitating some other explanation. By homology to other hsp70 proteins (36), BiP contains a ϳ45-kDa N-terminal ATPase core and a C-terminal peptide binding domain, and we reasoned, in the absence of a bona fide ER localization signal, that export of this reporter may be retarded by virtue of its peptide binding properties.
Therefore, we created a truncated secretory reporter (BiPN) comprised of the N-terminal ATPase domain (Fig. 1). Although still present at the C terminus of this construct, the HA9 epitope is not needed to discriminate between ϳ45-kDa reporter and ϳ72-kDa endogenous BiP since both are detected simultaneously with anti-BiP antibody, a convenient internal control. A pulse/chase experiment with a transformed cell line expressing this reporter is presented in Fig. 6A. Expression of 45-kDa BiPN reporter is very high relative to intracellular BiP protein (lane 1), which remains at constant levels throughout the chase period (lanes 1-5). In contrast, BiPN is rapidly exported from cells (lanes 1-5) into the medium (lanes 6 -10). Again, as mentioned above, a defined set of high M r polypeptides are apparently associated with intracellular BiP in a transient manner (compare lanes 1 with 2-5).
To obtain an accurate estimate of the kinetics of BiPN transport, we analyzed repetitions of this experiment by densitometry (Fig. 6B). Following a short pulse, export of radiolabeled BiPN proceeds in a linear fashion for up to 2 h with a t 1 ⁄2 slightly greater than 60 min and is essentially complete in 4 h. Simultaneously, extracellular reporter increases but never reaches 100% nor does total reporter (intracellular ϩ extracellular). This may simply reflect the difficulties inherent in such assays or may indicate that some portion of BiPN is being degraded intracellularly. Inclusion of protease inhibitors in the medium following fractionation does not increase the recovery of secreted reporter, and this issue has not been further investigated.
Analysis of ER Localization Signals-The development of the BiPN reporter convincingly demonstrates that significant soluble secretion is possible in African trypanosomes and provides the essential tool for secretory protein targeting studies. The data presented earlier strongly suggest that the C-terminal tetrapeptide, MDDL, mediates ER localization in T. brucei. We now wished to exploit the BiPN reporter to investigate the sequence specificity of this targeting signal. The BiPN gene was engineered to place various tetrapeptide sequences at the C terminus, and stably transformed cell lines were assayed for reporter export (Fig. 7). All cell lines were labeled equally as indicated by the constant amount of endogenous BiP and BiPassociated proteins detected in each cell fraction.
Attachment of several tetrapeptide sequences to the C terminus of BiPN resulted in intracellular retention of the re-  1-6) and medium (lanes 7-12) fractions. Labeled polypeptides were immunoprecipitated with anti-HA9 and analyzed as in Fig. 2. A scan of a 4-day exposure is presented. All lanes, 5 ϫ 10 6 cell equivalents. Asterisk denotes the position of BiPHA9. The scale refers to relative molecular mass in kDa.
FIG. 5. Secretion of recombinant VSG. Untransformed procyclic cells (P) and cultured cell lines expressing VSG reporters 117⌬gpi (V) or 117MDDL (M) were metabolically radiolabeled for 4 h and analyzed as described in Fig. 2. Labeled polypeptides were immunoprecipitated from cell (lanes 1-3) or medium (lanes 4 -6) fractions with anti-VSG antibodies. All lanes, 5 ϫ 10 6 cell equivalents. A scan of a 72-h exposure is presented. The scale refers to relative molecular mass in kDa.
porter. These include the native BiP signal, MDDL (lanes 5 and 6), the trypanosome protein disulfide isomerase signal, KQDL (lanes 7 and 8), and the mammalian BiP sequence (37), KDEL (lanes 9 and 10). Sequences that do not mediate retention are the C-terminal tetrapeptide of the HA9 epitope, PDYA (BiPN, lanes 1 and 2), the irrelevant tetrapeptide, AVRG (lanes 3 and  4), the deletion mutant, MDD- (lanes 11 and 12), and the hybrid sequence, AVDL (lanes 13 and 14). Two secreted reporters, BiPNMDD-and BiPNAVDL, were selected for detailed analysis. Since MDD-matches the native BiP sequence in 3 of 4 positions and AVDL is a hybrid between the irrelevant AVRG and the conserved amino acids of the two active trypanosomal tetrapeptides, MDDL and KQDL, we reasoned that they may express a partial retention phenotype. Pulse/chase analyses (Fig. 8) show that the kinetics of export of these two reporters are the same and, more importantly, are essentially identical to the kinetics of the basal BiPN reporter (Fig. 6B). We conclude that these tetrapeptides are completely unable to interact with the trypanosome retention machinery. DISCUSSION We have previously cloned and characterized a homolog of the secretory hsp70 protein, BiP, in the African trypanosome, T. brucei (18). Our identification was based on overall sequence homology with BiP from other organisms and on cytolocalization studies. We demonstrate here that this protein also has a key functional property consistent with its role as a molecular chaperone, the ability to interact transiently with newly synthesized secretory proteins (33). In other systems, BiP has been shown to associate with secretory proteins such as immunoglobulin (38) and VSV-G protein (39). In bloodstream trypanosomes, we show (Fig. 2) that BiP physically associates with newly synthesized VSG and that this association is transient, disappearing rapidly as would be expected for a secretory protein with an export t 1 ⁄2 of 15 min (4,5). In addition, a discrete set of endogenous high M r polypeptides are consistently associated with BiP in both bloodstream and procyclic trypanosomes. The identities of these proteins are not known, but they are presumed to be secretory polypeptides or other chaperones involved in secretory protein folding. Whatever the nature of these proteins, the key feature of their association with BiP, as in the case of VSG, is the transience of binding.
In order to test the role of the BiP C-terminal tetrapeptide, MDDL, in mediating ER localization in trypanosomes, it was necessary to first develop soluble reporters as essential tools for secretory protein targeting studies. This was accomplished by adaptation of two known proteins of the trypanosomal secretory pathway, BiP and VSG. Secretion of VSG was induced by deletion of the C-terminal sequence that directs GPI membrane anchor addition (Fig. 5); secretion of BiP was accomplished by deletion of its C-terminal tetrapeptide, MDDL (Fig. 3). Furthermore, addition of MDDL resulted in the intracellular retention of the heterologous VSG reporter. Although the rates of export for both of these reporters were low, these results indicate in the simplest manner that MDDL is an ER localization signal.
The reason for poor export of the VSG⌬gpi reporter is not clear. Full-length VSG, when expressed in procyclic trypanosomes, is correctly GPI-anchored, dimerized, and transported to the cell surface with reasonable kinetics (t 1 ⁄2 ϳ1 h), 2 ruling out the simple explanation that procyclic cells are just not competent for transport of this bloodstream-specific protein. It may be that in the absence of a GPI anchor VSG cannot be correctly folded and/or dimerized and as a result is retained by ER quality control mechanisms (40). Alternatively, it is possi- Transformed procyclic cells expressing BiPNMDD-or BiPNAVDL were pulse/ chase-labeled, and fluorographs of three representative gels for each reporter were analyzed by densitometry as described in Fig. 6. Data are presented as % of total intracellular report at time 0 (ordinate) versus chase time (abscissa). Closed symbols, cell-associated reporter. Open symbols, secreted reporter. Circles, MDD-. triangles, AVDL.
ble that GPI anchors provide some sort of positive forward transport signal. To answer this question it will be necessary to extend these results to other VSG molecules. However, it is worth noting that similar results were obtained in other systems where GPI attachment was prevented either by mutagenesis of the GPI-addition signal (41,42) or by inositol starvation (43).
Inefficient export of BiP, on the other hand, has been observed in mammalian systems and was attributed to innate peptide binding properties that could retard transport in the absence of an ER localization signal (37). The mammalian cytoplasmic hsp70 protein, hsc70, has been biochemically characterized as having a 45-kDa N-terminal ATPase domain and a ϳ30-kDa C-terminal peptide binding/regulatory domain (36), and this has been directly confirmed for mammalian BiP as well (44). We have shown (Fig. 6) that deletion of the peptidebinding domain results in the rapid and quantitative export of BiPN, providing direct experimental evidence for this conserved hsp70 domain structure in trypanosomal BiP.
The success of BiPN reporter, which is probably due to the compact globular nature of the ATPase domain (45), establishes convincingly that soluble secretion is possible in African trypanosomes. The kinetic half-time of slightly more than 60 min is the first such measurement to be made for any secretory protein, soluble or membrane-bound, in procyclic trypanosomes. Although considerably slower than the transport of membrane-bound VSG in bloodstream trypanosomes, this nevertheless seems reasonable since the doubling time of bloodstream trypanosomes (ϳ6 h) is two to three times faster than procyclics. Bloodstream trypanosomes are also known to be substantially more active in endocytic membrane trafficking (13).
Finally, the development of effective secretory reporters has allowed us to investigate the sequence specificity of ER targeting signals in trypanosomes (Table I). Both known trypanosomal signals, MDDL and KQDL, as well as the related mammalian sequence, KDEL, mediate quantitative intracellular retention. Although the native trypanosomal signals are somewhat divergent from the canonical eukaryotic XDEL motif, the pattern that emerges is consistent with the known sequences found on native ER proteins, as well as variants that have been tested experimentally in other systems (46,47). Trypanosomes are the most ancient eukaryotic organisms in which the sequence specificity of lumenal ER protein targeting has been directly assayed; the results establish the generality of the XDEL targeting signal throughout the broad range of eukaryotic phylogenetics. Presumably, this conservation extends to the machinery that mediates ER localization as well.
Further characterization of the trypanosomal secretory pathway is required for a complete understanding of the cell biology of these important pathogenic organisms. The work we have presented, particularly the development of secretory reporters, provides the foundation for future protein targeting studies. Trypanosomes are ancient organisms and have already provided many eye-popping variations of common eukaryotic processes. It is reasonable to assume that more discoveries await, some of which may be useful in developing new strategies for control of trypanosomatid diseases. Ϫ a Only the C termini are shown and gaps are placed between the HA9 epitope (YPYDVPYDA) and the additional amino acids.