Wild-Type and Mutant Forms of Rab7 Are Inducibly Expressed in Stable BHK Transfectants
Long-term overexpression of dominant negative forms of the rab proteins can be deleterious to cell viability (van der Sluijs et al., 1992
). Therefore, we chose to use a tetracycline-controlled expression system that has been shown to be very tightly regulated (Gossen et al., 1994
; Gossen and Bujard, 1992
). This system uses a transactivator-tetracycline repressor chimera to regulate expression of the gene of interest from a tetracycline operator and minimal CMV promoter. The hybrid transactivator does not induce transcription in the presence of tetracycline, but upon removal of the drug, the chimeric transactivator binds to the operator and induces expression of the recombinant protein. This system has been used previously to express mutant forms of the dynamin GTPase and the activating rab7Q67L mutant in HeLa HtTA1 cells (Damke et al., 1994
; Méresse et al., 1995
For our purposes, a new BHK parental line expressing the transactivator-tetracycline repressor chimera was generated. This cell line was subsequently used to derive stable lines that could be induced to overexpress wild-type rab7 or a dominant negative mutant rab7N125I (Fig. A
). The mutant protein was readily distinguishable from the wild-type protein because of its increased mobility on SDS-PAGE. The cause for this change in migration has not been investigated, but point mutations can cause anomalous migration on SDS-PAGE. Recombinant protein expression was tightly regulated in the newly isolated cell lines. In the presence of tetracycline, and for the first several hours after transfer of the cells to tetracycline-free media, the recombinant protein was not detectable (Fig. , A
). Beginning at 6 h after transfer, however, there was a steady accumulation of the recombinant protein, reaching maximal levels after 48 h (Fig. B
; data not shown). The recombinant rab proteins were overexpressed modestly (approximately threefold) after 18 h, obviating concerns that grossly overexpressed protein might have pleiotropic effects (Fig. B
). This level of mutant protein expression was sufficient to cause a 2.5-fold decrease in CI-MPR–mediated internalization when compared to duplicate samples cultured in the presence of tetracycline (Press, B., and A. Wandinger-Ness, manuscript in preparation). This is consistent with what was observed after transient overexpression (Feng et al., 1995
). The 18-h induction period was also short enough to avoid compensatory changes caused by long-term overexpression of deleterious proteins. For these reasons, recombinant protein expression was induced consistently for 18 h in all subsequent experiments.
Figure 1 Wild-type and mutant forms of rab7 are inducibly expressed in stable transfectants. Stable BHK fibroblasts were cultured in the absence of tetracycline (−Tet) for various times to induce the expression of recombinant rab7 proteins. Duplicate (more ...)
Overexpression of Mutant Rab7N125I Alters the Steady-State Distribution of CI-MPR, but Not That of Lgp120
The first hint that the expression of a dominant negative rab7 mutant might perturb lysosomal transport came from an examination of the subcellular localization of CI-MPR in the newly isolated cell lines. The distribution of CI-MPR was assessed by immunofluorescence microscopy. Normal, perinuclear staining of the CI-MPR was evident in cells expressing wild-type rab7 (Fig. , top left). Strikingly, when mutant rab7N125I protein was overexpressed, a significant fraction of the CI-MPR was localized in large peripheral vesicles (Fig. , bottom left). Identical results were also obtained using a stable line expressing a second dominant negative mutant, rab7T22N (data not shown).
Figure 2 CI-MPR distribution is altered in cells expressing a mutant form of rab7. Stable BHK fibroblasts were cultured in the absence of tetracycline for 18 h to allow for overexpression of wild-type (rab7wt) or mutant rab7N125I proteins. Cells were then fixed (more ...)
In contrast, the distribution of lgp120 was not noticeably altered in cells expressing mutant rab7N125I as compared to control cells or those expressing wild-type rab7 (Fig. , right). To further evaluate the alteration in CI-MPR distribution and to determine its impact on the trafficking of molecules to lysosomes, cell fractionation studies were conducted.
A Dominant Negative Rab7 Mutant Causes CI-MPR and Its Ligands to Accumulate in Light Membrane Fractions
Individual subcellular compartments are readily separated on Percoll gradients, providing a convenient means of monitoring the proteins or enzyme activities associated with various membrane fractions. To analyze the subcellular distribution of CI-MPR and its ligands, it was essential to use conditions that could resolve early and late endosomes. This was made possible using iso-osmotic homogenization and Percoll solutions prepared without sucrose (Czekay et al., 1997
). Under these conditions, rab5-positive early endosomes were routinely recovered near the top (fractions 1–3) of a 20% Percoll gradient and were well resolved from rab7-positive late endosomes present in the denser fractions (5–12) (Fig. A
). The profile of a peripheral membrane protein EEA1, recently shown to be associated with early endosomes and recognized by human autoimmune serum (Mu et al., 1995
), overlapped with that of rab5 (Fig. A
). Thus, two independent markers confirmed that early endosomes were confined to the top of the gradient. The profile of integral lgp120 established that lysosomes were confined to the densest fractions (Fig. A
). Additional markers were used to demonstrate that the Golgi was confined to fractions 1–4 and the plasma membrane was confined to fractions 1–3 (see Table ). The results presented in Fig. A
, using cells overexpressing wild-type rab7, were identical to those obtained with control BHK cells (not shown).
Figure 3 CI-MPR and its ligands accumulate in light membranes in cells expressing the mutant protein rab7N125I. Stable BHK fibroblasts were cultured in the absence of tetracycline for 18 h to allow for overexpression of wild-type and mutant rab7 proteins. PNS (more ...)
When the fractionation was performed with cells induced to overexpress the dominant negative mutant rab7N125I protein, the distribution of the markers for various endocytic compartments was largely unchanged. Lysosomes marked by lgp120 were still found primarily in the densest fractions (Fig. B). Late endosomes marked by the endogenous rab7 protein remained in the lower third of the gradient (Fig. B, wt). The majority of early endosomes were detected in the top three to four gradient fractions, based on the distributions of both rab5 and EEA1 markers (Fig. B). A second peak of rab5-containing endosomes was also observed at slightly higher densities. This modest alteration in the density of a subset of early endosomes most likely results from the transport block induced by expression of the mutant rab7N125I. Increased protein accumulation in early endosomes would be expected under these circumstances and could affect early endosome density. In spite of these differences, there was never any overlap between early and late endosomes in the top four gradient fractions.
It is noteworthy that the overexpressed mutant rab7N125I protein was recovered in the uppermost gradient fractions (Fig. B, N125I). This could be caused by a propensity of the mutant protein to be cytosolic and associated with structures other than late endosomes, including vesicles aligned on the actin cytoskeleton (Wandinger-Ness, A., unpublished observation). We do not believe that this represents a subpopulation of late endosomes because there was no trace of any endogenous, wild-type rab7 protein (distinguished by its slower mobility) associated with these fractions.
The gradient fractions shown in Fig. A
were also assayed for CI-MPR and its ligands. As expected, CI-MPR was found to be associated with fractions containing rab7-positive late endosomes (fractions 6–12). Lysosomal hydrolases known to be targeted by CI-MPR (including β-hexosaminidase and cathepsin D) cofractionated with rab7 and lgp120 markers, indicative of their late endosomal/lysosomal localization under steady-state conditions (Fig. , A
, closed squares
). The cathepsin D detected by immunoblotting is the intermediate form of the protein, and it represents a major species of the protein that is present in hamster cells under steady-state conditions (Isidoro et al., 1991
). The antibody used for these studies was unable to detect the mature hamster cathepsin D protein on immunoblots.
The situation was remarkably different when the steady-state distributions of both the CI-MPR and its ligands were examined after overexpression of the mutant rab7N125I protein. Approximately 40–50% of the receptor and ligands (β-hexosaminidase and cathepsin D) were now found to be associated with the top four gradient fractions containing Golgi and early endosomes, but devoid of late endosomes, as monitored by the distribution of the wild-type rab7 protein (Fig. , B and C, open squares). Clearly, this shift in the hydrolase distributions cannot be accounted for by a change in lysosome density because the lysosomal membrane protein lgp120 was still present exclusively in the densest fractions (Fig. B). These results showed that the expression of the mutant rab7N125I protein clearly perturbed the lysosomal accumulation of some markers while leaving others unaffected. One unresolved question concerned the issue as to whether the light membranes containing the lysosomal hydrolases were derived from the Golgi or endosomes.
Expression of Mutant Rab7N125I Causes Lysosomal Hydrolases to Accumulate in Endosomes
DAB can be used to cross-link endocytic compartments that contain internalized HRP (Ajioka and Kaplan, 1987
; Courtoy et al., 1984
). After the formation of a dense, cross-linked reaction product within their lumen, endosomes are readily sedimented at low speed. The time required for HRP to fill early endosomes is on the order of min, and HRP does not reach Golgi compartments during incubation periods of <1 h (data not shown). Therefore, this method provided a convenient means to distinguish between marker residence in Golgi membranes or endosomes.
Cells were induced to express rab7N125I and were maintained continuously in medium containing mannose 6–phosphate to prevent recapture of any secreted enzymes. HRP was internalized for 30 min to ensure that the endosomes would be well labeled. After Percoll gradient fractionation, the top four fractions were collected and divided into two equal aliquots. One aliquot was subjected to DAB cross-linking, and the second was left untreated as a control. After a short incubation period, cross-linked membranes were removed by low speed centrifugation (see schematic, Fig. A
). The resulting soluble fraction was assayed for various markers because the insoluble nature of the cross-linked material makes analysis of the pellet fractions intractable. Cathepsin D, present in the top four gradient fractions without DAB treatment, was significantly depleted after DAB cross-linking, as was the early endosome marker rab5 (Fig. B
). Inclusion of mannose 6–phosphate in the culture medium was noted to cause an increase in procathepsin D levels, perhaps reflecting its premature dissociation from CI-MPR in early endosomes caused by the presence of mannose 6–phosphate in the endocytic system. The effect of DAB cross-linking on cathepsin D was dramatic, but it could not be quantified because of the use of chemiluminescence detection. Therefore, a quantitative measure of the fraction of lysosomal enzymes depleted by DAB cross-linking was obtained by assaying the activity of β-hexosaminidase in the top three fractions. A reduction in activity of >40% was measured after DAB cross-linking (Fig. C
). Because the maximal depletion of endocytic markers that can be achieved after DAB cross-linking is in the vicinity of 70– 80%, a significant fraction of the β-hexosaminidase can be considered sensitive to DAB cross-linking (Futter et al., 1996
). In marked contrast, two Golgi markers, α-mannosidase II and NBD-sphingolipids, were largely unaffected by the cross-linking and remained soluble; their activity decreased <6% in the presence of DAB (Fig. C
). These results strongly indicate that the hydrolases recovered from the top of the gradient were primarily associated with endosomes and not the Golgi.
Overexpression of Mutant Rab7N125I Impairs the Endosomal Processing of the Lysosomal Hydrolase Cathepsin D
Cathepsin D processing was analyzed in an effort to pinpoint the site of lysosomal enzyme delivery from the TGN. Cathepsin D is the major aspartyl protease of lysosomes and a preferred ligand for CI-MPR (Ludwig et al., 1994
; Pohlmann et al., 1995
). Cathepsin D is initially synthesized in the ER as an inactive proenzyme (~53 kD) that is subsequently converted into an active, single-chain intermediate (~46 kD) (Delbrück et al., 1994
; Richo and Conner, 1994
; Rijnboutt et al., 1992
). Processing to the single chain form most likely occurs in late endosomes. In specialized cell types, activation may even occur in early endosomes (Diment et al., 1988
). A final cleavage in the lysosome generates the two-chain mature form, which consists of one 14-kD (ML
) and one 31-kD (MH
) subunit. Thus, the processing of cathepsin D can be used to gauge the progress of its intracellular transport.
Immunoprecipitation of endogenous hamster cathepsin D from pulse-labeled cells after various chase periods (from 0 to 4 h) revealed that processing to the intermediate form began ~1 h after synthesis, and formation of the lysosomal mature forms was detected within a 2-h chase period (Fig. A
lanes). After a 4-h chase period, the procathepsin D species was mostly processed, and the intermediate and mature forms prevailed. This is in agreement with other studies performed on the processing of hamster cathepsin D (Isidoro et al., 1991
). Expression of the mutant rab7N125I protein resulted in a kinetic delay in the processing of procathepsin D to the intermediate species, and formation of the mature species was undetectable even after a 4-h chase (Fig. A
lanes). Alterations in cathepsin D processing persisted and were detectable in cells that were metabolically labeled for as long as 6 h (Press, B., and A. Wandinger-Ness, manuscript in preparation). Procathepsin D levels were elevated by three- to fourfold, while the levels of the mature species were similarly decreased in cells expressing rab7N125I as compared to cells expressing wild-type rab7. This is consistent with the fact that procathepsin D was clearly discernible in cells expressing rab7N125I, even at steady state (Figs. B
Figure 5 Rab7N125I expression leads to diminished cathepsin D processing in endosomes, but Golgi processing remains normal. Stable BHK fibroblasts were cultured in the absence of tetracycline for 18 h to allow for overexpression of wild-type and mutant rab7 (more ...)
The Golgi processing of the carbohydrate chains on CI-MPR (Fig. B
) and lgp120 (Fig. E
; data not shown) was identical in both cell lines, serving as a strong indication that expression of the mutant rab7N125I protein had no effect on exocytic transport (Fig. B
). This finding agrees with our previous results showing that rab7N125I had no effect on vesicular stomatitis virus G protein transport along the exocytic pathway (Feng et al., 1995
Figure 7 Newly synthesized CI-MPR and immature cathepsin D are present in early endocytic compartments in cells expressing rab7N125I. Stable BHK fibroblasts were cultured in the absence of tetracycline for 18 h to allow for overexpression of wild-type and mutant (more ...)
Evidence for the Direct Intracellular Delivery of CI-MPR and Newly Synthesized Cathepsin D from the TGN to an Early Endocytic Compartment
To distinguish whether targeting from the TGN entailed direct delivery to an early endocytic compartment, the transport of newly synthesized molecules was analyzed in both wild-type and mutant cell lines. In this regard, it was first important to establish a meaningful time point for the gradient fractionation of metabolically labeled samples (i.e., sufficient time for exit from the TGN to have occurred). Analysis of the kinetics of CI-MPR and cathepsin D processing revealed that a 2-h chase period was sufficient to allow CI-MPR to acquire Golgi-specific carbohydrate modifications and cathepsin D to become processed in endocytic compartments (Figs. , A and B). On the basis of this information, cells were induced to express rab7 wild-type or rab7N125I proteins for 18 h, and they were subjected to a 2-h chase period after brief metabolic labeling. Postnuclear supernatants were prepared and subjected to Percoll gradient fractionation. Individual fractions were collected and analyzed as pools (I–III) to facilitate sample handling and protein detection. Cathepsin D was immunoprecipitated from each set of pooled fractions. The differences between the two cell lines in the endosomal processing of cathepsin D were once again apparent. The mature species were only observed in cells expressing the wild-type protein, and these forms were primarily in pool III, concordant with their formation in lysosomes. The procathepsin D species was prevalent in cells expressing rab7N125I and was enriched in the top gradient fractions (pool I), consistent with its presence in the Golgi and/or early endocytic structures. The distribution of the intermediate form was of primary interest because it is diagnostic of cathepsin D in endosomes.
Quantitative analysis revealed that in cells expressing wild-type rab7, the majority (59.5%) of the intermediate form of cathepsin D was associated with the densest gradient fractions in pool III (Fig. A, rab7wt lanes). The remainder was nearly equally distributed between pools I (18.3%) and II (22.2%). This distribution is consistent with formation of the intermediate species in endosomes. Lysates derived from cells expressing mutant rab7N125I exhibited a remarkably distinct profile for the intermediate cathepsin D species (Fig. A, rab7N125I lanes). In this case, the majority (54%) of intermediate cathepsin D was recovered in pool I together with procathepsin D. The amount of the intermediate form detected in pool III (27.7%) decreased accordingly, while the amount in pool II (18.3%) was similar to that found in pool II using cells expressing wild-type rab7.
Figure 6 Rab7N125I expression causes newly synthesized CI-MPR and immature cathepsin D to accumulate in light membrane fractions. Stable BHK fibroblasts were cultured in the absence of tetracycline for 18 h to allow for overexpression of wild-type and mutant (more ...)
Immunoprecipitation of CI-MPR from the same fractions revealed that the newly synthesized receptor also exhibited an altered distribution in cells expressing mutant rab7N125I protein (Fig. B). Normally, the majority (56.8%) of the receptor was detected in pool III. Pool II contained somewhat less CI-MPR (32.2%), and pool I had low but detectable levels (11.0%). In cells expressing mutant rab7N125I, the amount of CI-MPR present at the top of the gradient increased threefold (41%), and the receptor in the densest fractions decreased correspondingly (23.0%). These data highlight the increased presence of cathepsin D and CI-MPR in light membrane fractions upon expression of mutant rab7N125I protein.
If the newly synthesized proteins associated with these light membrane fractions are primarily contained in endosomes, they would be expected to exhibit an increased sensitivity to DAB cross-linking, as demonstrated above for the proteins accumulated at steady state. To examine this issue, cells were induced to express the wild-type or mutant rab7 proteins as before, and were then metabolically labeled and incubated for chase periods of 30 and 120 min. HRP was internalized during the last 10 min of each chase period, and samples were subjected to DAB cross-linking, as detailed in Materials and Methods. Replicate samples were left untreated as controls. Quantitative immunoprecipitations of cathepsin D and CI-MPR were conducted in triplicate for each time point, and results from a representative experiment are shown. After a 30-min chase period, the procathepsin D species prevailed in cells expressing either form of rab7, and this species was insensitive to DAB cross-linking (Fig. A). At this time point, the procathepsin D was most likely still in the Golgi and, therefore, inaccessible to HRP. After a 120-min chase period, the intermediate and mature forms of cathepsin D predominated in cells expressing wild-type rab7, and these species were also insensitive to DAB cross-linking (Fig. , A and B). In contrast, the intermediate form of cathepsin D was clearly sensitive to DAB cross-linking after a 120-min chase period in cells expressing rab7N125I (Fig. A). Quantitation showed that the intermediate form was decreased by 40% in the sample treated with DAB (Fig. B). Again, processing to the mature forms did not occur in this cell line within this time frame.
Quantitative immunoprecipitations of CI-MPR yielded analogous results. At the 30-min time point, the immature form of CI-MPR was insensitive to cross-linking in both cell lines because it had not yet traversed through the Golgi (Fig. C). The mature CI-MPR prevalent at the 120-min time point was sensitive to cross-linking, but only in cells expressing mutant rab7N125I protein (Fig. C). In these cells, the mature receptor was decreased by 43% after DAB cross-linking (Fig. D), concurring with the results obtained for the intermediate cathepsin D species.
These DAB–cross-linking experiments using metabolically labeled samples were extremely difficult to perform and, as a result, some of the standard deviations were large (Fig. , B and D). Calculation of the Student's t distribution, with a 90% confidence limit, indicated that the differences in cathepsin D and CI-MPR signals (±DAB cross-linking) upon rab7N125I expression were statistically significant.
The results presented in this section are consistent with the interpretation that the newly synthesized proteins are initially delivered from the TGN to an early endocytic compartment, where they accumulate in cells expressing mutant rab7N125I. Here, slow conversion of procathepsin D to the intermediate form can still take place when transport to late endosomes is diminished. Such early structures would readily be loaded with HRP during a 10-min internalization period and could account for the sensitivity of the newly synthesized proteins to HRP/DAB–mediated cross-linking. On the other hand, when egress from this compartment is normal (as is the case in cells expressing wild-type rab7), these proteins are rapidly transported to later endocytic compartments and are no longer sensitive to DAB cross-linking.
There are numerous indications that lysosomal membrane glycoproteins are transported from the Golgi to lysosomes via a route distinct from that used by CI-MPR and its ligands (Brown et al., 1986
; Griffiths et al., 1988
; Harter and Mellman, 1992
; Mathews et al., 1992
). The finding that expression of rab7N125I had no apparent impact on the steady-state distribution of lgp120 (Figs. and , A
) is in keeping with this. Therefore, we also examined the sensitivity of newly synthesized lgp120 to DAB cross-linking. After brief metabolic labeling and a 30-min chase period, a significant fraction of immature lgp120 was still detectable in both cell lines, but the fully glycosylated mature species was also clearly evident (Fig. E
). Neither the immature nor the mature forms of lgp120 were sensitive to DAB cross-linking at either time point examined in cells expressing wild-type rab7 (Fig. , E
). In contrast to what was observed for CI-MPR and cathepsin D, lgp120 remained insensitive to DAB cross-linking, even in cells expressing the mutant rab7N125I protein. Thus, lgp120 served as a useful control for the DAB cross-linking studies. It is a marker that is transported to lysosomes, yet it was apparently unaffected in its targeting by expression of the mutant rab7N125I protein. These findings also serve as an indication that lgp120 is likely to follow an independent route to lysosomes.