Unexpected Intracellular Localization of the AMD-Associated Cystatin C Variant


Luminita Paraoan, lparaoan@liv.ac.uk


Cystatin C is abundantly expressed by the retinal pigment epithelium (RPE) of the eye. Targeting of cystatin C to the Golgi apparatus and processing through the secretory pathway of RPE cells are dependent upon a 26-amino acid signal sequence of precursor cystatin C. A variant with an alanine (A) to threonine (T) mutation in the penultimate amino acid of the signal sequence (A25T) was recently correlated with increased risk of developing exudative age-related macular degeneration. The biochemical consequence of the A25T mutation upon targeting of the protein is reported here. Targeting and trafficking of full-length mutant (A25T) precursor cystatin C–enhanced green fluorescent protein fusion protein were studied in living, cultured retinal pigment epithelial and HeLa cells. Confocal microscopy studies were substantiated by immunodetection. In striking contrast to wild-type precursor cystatin C fusion protein conspicuously targeted to the Golgi apparatus, the threonine variant was associated principally with mitochondria. Some diffuse fluorescence was also observed throughout the cytoplasm and nucleus (but not nucleoli). Secretion of fusion protein derived from the threonine variant was reduced by approximately 50% compared with that of the wild-type cystatin C fusion protein. Expression of the variant fusion protein did not appear to impair expression or secretion of endogenous cystatin C.

Cystatin C is a cysteine protease inhibitor that is widely distributed in body fluids (1–8). The mature protein consists of a single polypeptide of 120 amino acids with two disulfide bonds (9). Nucleotide sequence data indicate that the protein is synthesized as a precursor, precystatin C, bearing a 26-amino acid N-terminal leader sequence (10,11).

Cystatin C mRNA is abundant in retinal pigment epithelial (RPE) cells (12). In vivo, RPE cells form the outermost layer of the retina, occupying a strategic position between the neurosensitive layers of the retina and the highly vascularized choroid (13). The apical side of the polarized RPE cells faces the outer segments of the photoreceptor cells and the basal membrane is separated from the choroid by the collagenous Bruch's membrane, to which RPE cells adhere firmly. Visual processing and the functional integrity of the neuroretina depend critically on functions carried out by the RPE (reviewed in (14,15)). Studies in which the cDNA for cystatin C, with or without the nucleotides encoding the leader sequence, was fused to cDNA for Enhanced Green Fluorescent Protein (EGFP) demonstrated that the leader sequence targets the precystatin C fusion protein to the Golgi apparatus and secretory pathway of cultured RPE cells, with secretion of the cystatin C-EGFP fusion protein into the medium (16,17). The findings thereby established that the leader sequence functions as a signal sequence and the designation signal sequence is used in the present paper.

Recently, it has been reported that a mutation resulting in an alanine to threonine substitution at position 25 (A25T) in the signal sequence of precystatin C is correlated with increased risk for relatively early onset of Exudative Age-related Macular Degeneration (AMD) (18). It was therefore of interest to test the effect of the mutation upon the targeting of precystatin C to the secretory pathway of RPE cells and also, for comparison, HeLa cells. Here we address this question by site-directed mutagenesis coupled with confocal microscopy to localize the protein product in living cells. Our findings show that intracellular targeting is substantially altered by the A25T mutation.


Generation of mutant Ala25Thr precursor cystatin C–EGFP construct

The construct encoding the full-length precursor cystatin C fused to EGFP (pCysC-EGFP) was subjected to site-directed mutagenesis in order to generate the fluorescently labeled precursor cystatin C with the penultimate amino acid residue of the signal sequence substituted from alanine to threonine. Mutagenesis involved a G to A substitution of the first base of the codon for the respective amino acid residue. The mutant construct was named Ala25Thr_ pCysC-EGFP. (Note that three-letter abbreviations are used for amino acids in the name of the clone to avoid confusion with nucleotide bases, but one-letter amino acid abbreviations are used elsewhere in the text where the context is clear. Note also that the gene in human populations that encodes threonine at position 25 of the signal sequence of precystatin C was designated ‘variant B’ (18)).

Intracellular localization of A25T precursor cystatin C–EGFP fusion protein

The localization and trafficking of mutant A25T precursor cystatin C was studied in RPE cells and HeLa cells transiently transfected with the plasmid Ala25Thr_pCysC-EGFP. Transfections in identical conditions with pCysC-EGFP encoding the wild-type precursor cystatin C and the vector plasmid pEGFP-N3 encoding EGFP on its own were carried out as reference and controls. The efficiency of transfection, evaluated by fluorescence microscopy 24 h post-transfection, was similar for all constructs and was in the range of 20–25% of all cells in culture (as exemplified in Figure 1i). Transfected cells were maintained in culture for up to a week. No qualitative changes in the specific localization patterns described below were noticed during this time.

Figure 1.

Figure 1.

Localization of cystatin C, wild-type and variant B, in RPE and HeLa cells. Living, cultured human RPE cells (a–i) and HeLa cells (j–l) expres sing pCysC-EGFP (a d g and j), Ala25Thr_pCysC-EGFP (b e h and k), and EGFP on its own (c f i and l) were imaged 24 h post-transfection. The emission specific for EGFP constructs is shown as green. The cells were incubated with Mitotracker Red CM-H2Xros probe, which upon oxidation in actively respiring cells appeared concentrated in the mitochondria and fluoresced red. For each field of cells imaged, green and red fluorescence were recorded as separate images and then merged for assessing the overlap of respective signals (g h j k and l). Colocalization of the green fluorescence with the mito-chondria-specific red fluorescence is observed in cells expressing the mutant precursor cystatin C fusion, demon-strated by yellow color in (h) and (k). The morphology of the RPE cells and the transfection efficiency are illustrated in the bright field (f) and the corresponding overlaid image of RPE cells transfected by EGFP on its own (i). Scale bars represent 50 μm.

The mutant precursor cystatin C construct gave rise to an entirely unpredicted intracellular localization of fluorescence, clearly and consistently defined in all cells expressing the mutant fusion protein and remarkably different from that generated by the wild-type precursor cystatin C fusion. The specific fluorescence appeared partly concentrated in a prominent network of interconnected, branched structures spread throughout the cytoplasm, both in RPE cells and HeLa cells (Figure 1b,k, respectively). This network was identified as mitochondria by incubating the living cells with the active mitochondria-specific probe, Mitotracker Red CM-H2Xros, whose red fluorescence overlapped the green fluorescence signal of the fusion protein as demonstrated by the yellow color in Figure 1 (h,k). Besides the localized, concentrated distribution described above, a lighter, diffuse green fluorescence indicated the presence of the mutant precursor cystatin C in the cytoplasm and nucleus (but not nucleoli) of RPE cells. A similar diffuse fluorescence was evident in HeLa cells as well, albeit seemingly less intense.

The localization of fluorescence from the mutant construct was in striking contrast with that from the wild-type construct. The latter appeared, in both RPE and HeLa cells, clearly targeted to a perinuclear region of the cytoplasm, previously identified and reported as being the Golgi apparatus in RPE cells (16,17), decreasing towards the periphery of the cells and being absent from the nucleus (Figure 1a,j). In further agreement with findings reported previously (16), punctate green fluorescent foci were also apparent in RPE cells, first appearing about 24 h post-transfection (Figure 1a). The wild-type localization was completely independent of the mitochondrial staining (Figure 1g,j).

In control transfections with the plasmid encoding EGFP on its own, cells showed, as previously characterized in RPE cells (17), a completely diffuse fluorescence throughout the cytoplasm and nucleus, but not nucleoli, with no accumulation in the mitochondria (Figure 1c,i,l). The lack of intracellular targeting of the EGFP protein provides further evidence that the specific localization seen with the cystatin C fusion protein is indeed dictated by the cystatin C component.

Characterization of the dynamic state of localization of variant precursor cystatin C

The above findings indicate a degree of association of A25T mutant precursor cystatin C with mitochondria in living cells. To provide an indication of the strength of this association, the movement of the respective fusion protein was studied after bleaching parts of transfected cells. A representative bleaching experiment of a live, Ala25Thr_pCysC-EGFP-transfected RPE cell is presented in Figure 2. Before bleaching, the mutant cystatin C-EGFP fusion protein was, as described above, partly associated with mitochondria with a fraction of it present diffusely in the cytoplasm and nucleus of the cell (Figure 2a). Following bleaching of approximately half of the cell for 50 s, the diffuse cytosolic fluorescence was simultaneously obliterated in both the irradiated and nonirradiated parts of the cell (Figure 2b). This indicates rapid movement of mutant fusion protein through the cytosol, with practically all molecules entering the irradiated region and being bleached during the 50s irradiation period. The fluorescence associated with the mitochondria appeared less affected, with about 40% of it remaining present in the nonbleached mitochondria. The remaining mitochondrial fluorescence did not readily redistribute from the nonbleached into the bleached area (Figure 2c), suggesting some degree of association of the fusion protein with the cellular organelle. Quantification of the fluorescence intensity in various nonbleached regions of the cell (Figure 2d) confirmed the differential mobility of fluorescently tagged cystatin C as a function of its localization. Given the concomitant decrease of mitochondrial fluorescence with the disappearance of the cytoplasmic fluorescence upon bleaching, a degree of exchange between the two pools of fluorescently tagged cystatin C cannot be ruled out.

Figure 2.

Figure 2.

Real-time analysis of differential intracellular movement of fusion protein expressed from the Ala25Thr_pCysC-EGFP construct as a function of its localization. The experiment was conducted by complete bleaching of fluorescence in half of an RPE cell and monitoring the changes in the distribution of fluorescence thereafter. Images were taken at 15 s intervals, both before and after bleaching. Representative images are presented, as follows: a) before bleaching b) immediately after bleaching c) 60 s postbleaching. The bleached area is indicated by the white rectangle in (b). Quantification of the average intensity of fluorescence per pixel, normalized to the intensity of the respective area at the start of experiment, is presented in (d) for the bleached area, cytoplasm in the nonbleached area, and mitochondria in the nonbleached area, throughout the experiment. In the nonbleached area, cytoplasmic fluorescence was completely obliterated at the same time with bleaching the other half of the cell; fluorescence associated with mitochondria was reduced to approximately 40% of the level before bleaching and the remaining fluorescence level appeared relatively steady following bleaching. Scale bar, 20 μm.

The effect of disruption of mitochondrial membrane potential upon mitochondrial localization of variant B precursor cystatin C-EGFP fusion protein was also investigated. Treatment with the uncoupler of mitochondrial function FCCP caused a rapid loss of mitochondrial membrane potential, which was monitored in real time by the loss of retention of the red, oxidized form of Mitotracker Red CM-H2Xros. Approximately 20 s after addition of FCCP, the red mitochondrial probe, whose retention in the mitochondria is membrane potential-dependent, started to leak out of the mitochondria and appeared diffused in the cytoplasm of cells (Figure 3, middle panel). Distribution of mutant precursor cystatin C-EGFP fusion protein, however, remained unchanged and maintained its structured appearance at all observation times (Figure 3, top panel). As the Mitotracker probe ceases to indicate the morphology of mitochondria, the fluorescence of the mutant cystatin C-EGFP fusion protein in merged images appears green rather than yellow (Figure 3, lower panel). These experiments indicate that the association of A25T precursor cystatin C with mitochondria, at least once established, is not dependent on the inner mitochondrial membrane potential.

Figure 3.

Figure 3.

Disruption of mitochondrial membrane potential does not abolish association of variant B cystatin C-EGFP fusion protein with mitochondria in RPE cells. Mitochondrial membrane potential was disrupted with FCCP in live RPE cells expressing A25T mutant precuror cystatin C-EGFP fusion protein. Once the mitochondrial staining with Mitotracker Red CM-H2Xros probe was established (approximately 5 min after addition of probe), FCCP was added to the culture medium at a final concentration of 500 nm and RPE cells were imaged continuously for up to 10 min. Times shown indicate time lapsed after addition of FCCP. Green fluorescence (indicative of variant B cystatin C-EGFP fusion protein) and red fluorescence (indicative of the localization of the oxidized form of Mitotracker probe) were recorded as separate images and then merged. Yellow indicates colocalization of the green and red signals. Disruption of mitochondrial membrane potential causes the oxidized form of Mitotracker probe to leak out of the mitochondria, as indicated by the increasingly diffuse appearance of the red fluorescence with time (middle panel). However, the green fluorescence maintains its structured appearance (top panel) and the corresponding overlap with red fluorescence is lost (lower panel). The remaining yellow color in the merged image at t2 is mainly due to the high level of red fluorescence throughout the cytoplasm of cells. Field of cells shown is representative of four experiments carried out at either 24 h or 48 h post-transfection. Scale bars represent 50 μm.

Immunodetection of mutant cystatin C-EGFP fusion protein in mitochondrial fraction and sensitivity to trypsin digestion

To confirm the association of mutant cystatin C-EGFP fusion protein with mitochondria and to test its accessibility on the surface of isolated mitochondria, crude mitochondrial fractions were prepared from RPE cells transiently transfected with Ala25Thr_pCysC-EGFP construct. Mitochondrial proteins were subjected to SDS/PAGE/blotting, followed by immunodetection with anticystatin C, anti-EGFP and anti-cytochrome c antibodies. Mutant cystatin C-EGFP fusion protein was detected in crude mitochondrial fraction as a protein of about 40–42 kDa, immunoreactive with both anti-cystatin C antibody (Figure 4, lane 1) and anti-EGFP antibody (data not shown). Upon treatment with trypsin, the band corresponding to the fusion protein was only slightly reduced in intensity (Figure 4, lanes 2–4). The lack of detection of cytochrome c, an inner mitochondrial membrane protein, in lanes 1–4 of the same blot (Figure 4) indirectly confirmed the membrane integrity of mitochondrial preparations. Detergent solubilization of mitochondria with CHAPS rendered cystatin C-EGFP fusion protein susceptible to trypsin action, the corresponding band disappearing completely (Figure 4, lane 6). The dissapearance of the fusion protein was clearly attributable to proteolysis, since in the presence of protease inhibitors cocktail (added after CHAPS solubilization of mitochondria) the fusion protein was not degraded (Figure 4, lane 7). Detection of cytochrome c in solubilized mitochondria, as a band of approximately 14 kDa (Figure 4, lane 5), confirmed the nature of the subcellular preparations; trypsin degradation was effective on solubilized cytochrome c (Figure 4, lane 6) and was impeded in the presence of protease inhibitors (Figure 4, lane 7).

Figure 4.

Figure 4.

Immunodetection of mitochondrial variant B cystatin C-EGFP fusion protein and trypsin treatment of mitochondrial preparations. Crude mitochondrial fractions were prepared from RPE cells transiently transfected with Ala25Thr_pCysC-EGFP construct. Approximately 20 μg of intact and CHAPS-solubilized mitochondria were treated with trypsin (+, 20 μg/mL; + +, 40 μg/mL) for 20 min on ice, with or without preliminary addition of an inhibitors cocktail, as indicated in the Figure. Mitochondrial proteins were resolved on 10% SDS-PAGE gel. Top panel shows the mutant cystatin C-EGFP fusion protein (40–42 kDa) identified by anti-cystatin C antibody and lower panel shows the band corresponding to cytochrome c (14 kDa), identified by anti-cytochrome c antibody on the same blot following membrane stripping off. Each experiment (blot) used samples from the same mitochondrial preparation. Four mitochondrial preparations were obtained from two independent transfections.

Immunodetection of wild-type and mutant cystatin C-EGFP fusions in cell lysates and conditioned culture media of transfected RPE cells

Transiently transfected RPE cells were lysed 24 h post-transfection and their total protein content was quantified as indicated in Materials and Methods. Western blot analysis (Figure 5a) indicated comparable amounts of wild-type and mutant cystatin C fusion proteins (identified by both anti-cystatin C and anti-EGFP antibodies as bands of about 40 kDa) present in the respective cell lysates. Endogenous cystatin C, identified as a band of approximately 14 kDa reacting only with the anti-cystatin C antibody, appeared also in comparable amounts in all RPE cells, including those transfected with EGFP vector only. In each cell lysate, the band representing the fusion protein appeared much stronger than that corresponding to endogenous cystatin C, presumably due to increased expression driven by the cytomegalovirus promoter of the EGFP vector. EGFP on its own was identified as a single band of about 27 kDa in cells transfected by EGFP only, indicating that no free EGFP is present in cells transfected with either wild-type or mutant precursor cystatin C constructs. Probing with anti-GAPDH antibody confirmed the equal loading of samples.

Figure 5.

Figure 5.

Immunodetection of cystatin C (wild-type and mutant)-EGFP fusion proteins and of endogenous cystatin C in cell lysates and culture medium. Samples subjected to SDS-PAGE electrophoresis were cell lysates (a) and conditioned media (b), of RPE cells transiently transfected with the plasmid encoding EGFP only and with the constructs pCysC-EGFP or Ala25Thr_pCysC-EGFP. Complete media conditioned by untransfected RPE cells were used as controls for endogenous cystatin C secretion. All samples were quantified as described in Materials and Methods and aliquots used for gel electrophoresis were normalized to total protein concentration. All gels were 10% SDS-PAGE, but cell lysates were also run on a 16% high resolution gel (presented in (a)) to resolve a slight difference in the size of wild-type cystatin C-EGFP and A25T cystatin C-EGFP, first identified on 10% gels. Western blots were probed with anti-cystatin C antibody, which detected the cystatin C-EGFP fusions and the endogenous cystatin C, and with anti-EGFP antibody, which identified the fusion proteins and, in cell lysates, EGFP on its own. Anti-GAPDH and anti-PEDF antibodies were used to quantitatively probe the cell lysates, respectively, media blots, for equal loading. Arrows point to bands of slightly different size identified in cell lysates containing the wild-type vs. mutant cystatin C fusion proteins (see text for further description).

The 10% PAGE analyses indicated that there may be a slight difference in size between wild-type and mutant cystatin C fusions. High-resolution Western blot analysis resolved a slight delay in the migration of the corresponding band in cell lysates of RPE cells transfected with the mutant cystatin C fusion. We interpreted this as being due to the lack of processing by cleavage of the leader sequence in the case of A25T mutant cystatin C. Thus, we deduced that the band of about 40 kDa identified by both anti-cystatin C and anti-EGFP antibodies in pCysC-EGFP-transfected cells corresponds to the mature cystatin C-EGFP fusion protein (calculated molecular weight of 40.3 kDa), following cleavage of the 26-amino acid signal sequence from the wild-type precursor cystatin C in the secretory pathway; the slightly slower band, identified by the same antibodies in Ala25Thr_pCysC-EGFP-transfected cells, represents the unprocessed mutant precursor cystatin C fusion of about 42.8 kDa.

Western blot analysis of culture media conditioned by the same cells as above was carried out to assess the presence of cystatin C-EGFP fusions and of endogenous cystatin C secreted by RPE cells (Figure 5b). The amount of endogenous cystatin C was similar in all cases, indicating the same level of secretion by RPE cells, untransfected and transfected with various constructs. In contrast, the level of cystatin C-EGFP fusion protein was reduced to approximately half in the media conditioned by RPE cells transfected with the mutant cystatin C construct, compared with that in culture media of wild-type pCysC-EGFP-transfected cells (Figure 6). No EGFP in the medium of cells transfected with EGFP on its own was detected by probing with the anti-EGFP antibody (data not shown), thus eliminating the possibility of cellular leakage. Probing with anti-pigment epithelium derived factor (PEDF) antibody confirmed the equal loading with respect to protein content.

Figure 6.

Figure 6.

Secretion of cystatin C-EGFP fusion protein in the culture medium of RPE cells transfected with wild-type (pCysC-EGFP) and mutant (Ala25Thr_pCysC-EGFP) cystatin C constructs. The levels of secreted cystatin C-EGFP fusion protein and endogenous cystatin C were evaluated by densitometry of Western blots probed with anti-cystatin C antibody. Data from six independent transfection experiments (three with wild-type cystatin C fusion and three with mutant cystatin C fusion) were analyzed. In each case, the level of secreted cystatin C-EGFP fusion was normalized to that of secreted endogenous cystatin C. The results are presented as percentages with normalized level of secreted wild-type cystatin C fusion set to 100%. RPE cells transfected with Ala25Thr mutant cystatin C construct secreted approximately 50% less cystatin C fusion protein the culture medium (the levels ranged from 40.8% to 56.6%, and averaged 50.4%, standard deviation indicated in the graph is 8.4%).


Our previous and current findings can be summarized as follows. When cultured RPE cells were transfected with a DNA construct encoding precystatin C fused to EGFP, the fusion protein was transported through the secretory pathway into the culture medium (16). In RPE cells transfected with a cystatin C-EGFP construct lacking the precystatin signal sequence, the fusion protein did not enter the secretory pathway but instead became diffusely distributed in the cytoplasm and nucleus (17). In the present study, cells transfected with the mutant construct Ala25Thr_pCysC-EGFP yielded strikingly different findings from either of the above. Fluorescence was conspicuously associated with mitochondria and was present with a lower intensity diffusely in the cytoplasm and nucleus. The bleaching experiment indicated that the fluorescent protein was mobile in the cytosol, but less so when associated with mitochondria. The association of fusion protein with mitochondria, which was not dependent on inner mitochondrial membrane potential, was substantiated by immunodetection of the protein in crude mitochondrial preparations. Mutant precursor cystatin C-EGFP fusion protein was slightly susceptible to trypsin digestion in intact mitochondria and fully degraded by trypsin in solubilized mitochondria. Immunochemical data on cell lysates indicated that the EGFP remained covalently attached to the cystatin C moiety, and hence that fluorescence reports the distribution of cystatin C. Immunochemical data on the culture medium of transfected cells revealed a reduction of approximately 50% of the secreted fusion protein expressed from the mutant construct, compared with that from the wild-type construct. Immunochemical analysis of the medium also indicated that transfection of cells with the mutant construct did not appear to interfere with secretion of endogenously produced cystatin C into the medium. In control experiments in which RPE cells were transfected by EGFP on its own, EGFP did not accumulate in the medium ((16) and present study). This implies that the protein expressed from the mutant construct that entered the medium did so through secretion and not through leakage from the cells. Lastly, in transfection experiments with HeLa cells the mutant construct again gave rise to aberrant distribution of fluorescence both in association with mitochondria and diffusely in the cytoplasm and nuclei, the diffusely distributed fluorescence being relatively weaker than in RPE cells.

The findings are consistent with the following interpretation. A proportion of the fusion protein from the mutant construct undergoes incorrect trafficking to intracellular destinations. A fraction of this protein becomes associated with mitochondria. This fraction is fluorescently conspicuous and readily immunodetected in mitochondrial preparations, indicating a relatively high local concentration. It is less mobile than the diffusely distributed fluorescent protein and its association with mitochondria is not abolished by the collapse of mitochondrial membrane potential. Furthermore, both cystatin C and EGFP epitopes are detected in intact mitochondria preparations in denaturing gels by the respective antibodies. These findings exclude the possibility of the presence of fusion protein in the mitochondrial matrix but suggest some degree of interaction, most likely indirect, of the fusion protein with mitochondria (discussed below). Nevertheless, some fusion protein from the mutant construct is secreted into the medium, as is the case for most or all of the fusion protein from the wild-type construct and for cystatin C itself. Thus the A25T mutation appears to be ‘leaky’: it does not completely exclude precystatin C from the secretory pathway, but it substantially reduces the efficiency of targeting for secretion, and the rest of the protein is redirected intracellularly, especially in association with mitochondria.

Although the present study did not attempt to identify the actual step at which a proportion of molecules of precystatin C fail to progress through the secretory pathway, some relevant points may be noted here. The A25T mutation affects the 5 amino acid C-terminal (‘c’) region (in the nomenclature of von Heijne (19,20)) of the signal sequence of precursor cystatin C. As this region is not usually involved in signal recognition particle (SRP) (21,22), the A25T mutation is not expected to affect binding to SRP. It is rather more likely that a different step in the translocation of the mutant precursor protein into the lumen of the endoplasmic reticulum may fail or proceed with reduced efficiency. For example, it is conceivable that the A25T mutation might cause a transient unfavorable interaction with Sec61p or an associated component of the translocation channel, resulting in exclusion from transport through the endoplasmic reticulum membrane of a proportion of precystatin C molecules. The fraction of mutant precursor molecules successfully translocated is expected to have the signal sequence correctly cleaved by the signal peptidase. Signal peptidase requires small amino acid side chains at positions − 1 and − 3; the requirement at position − 2 is less critical, though often a bulky and/or polar side chain occurs here (19,20,23), so if anything the A25T mutation would be expected to provide a better substrate than the wild-type sequence for the signal peptidase. The programme of Nielsen et al. (24) gives high scores for predicting correct cleavage by signal peptidase between G26 (the − 1 position) and S27 (the + 1 position) for both the wild-type and the mutant (scores of 0.751 and 0.760, respectively, on a scale of increasing likelihood from 0 to 1). Once the signal sequence is correctly cleaved, the protein should then continue to be processed in the endoplasmic reticulum and secreted as normal. These considerations are in agreement with our findings of reduced secretion of cystatin C fusion protein expressed from the A25T mutant construct.

Whatever the mechanism by which a proportion of molecules of mutant precursor cystatin C are excluded from the secretory pathway, it needs to be considered how these molecules become waywardly associated with mitochondria instead. The program of Emanuelsson et al. (25) gives very low scores for either mutant or wild-type precystatin C signal sequence as a potential mitochondrial targeting sequence (0.083 and 0.082, respectively, on a scale of 0–1, compared with scores of 0.917 and 0.916 for SRP targeting). Clearly the mutation does not confer the required amphiphilic character to the signal sequence (26,27). Therefore an actual direct targeting/import into mitochondria is very unlikely. Instead, mutant precystatin C molecules may interact (possibly via the incorrectly folded hydrophobic region of the signal sequence) with proteins that convey mitochondrially targeted proteins to the outer mitochondrial membrane (28–30). The inaccessibility to trypsin, in our experimental system, may be due to a specific spatial or conformational effect of such an interaction upon the mutant precursor cystatin C-EGFP fusion protein. Eventually, a steady state is established between precystatin C complexes associated with mitochondria and precystatin C molecules in the cytosol. The relative intensities of fluorescence associated with mitochondria and cytosol observed with A25T mutant precursor cystatin C-EGFP fusion protein would reflect this steady state, and this might differ between different cell types, such as RPE and HeLa cells used in the present work. More work is obviously needed to test this hypothesis and the present results will be the basis of further research in our lab. Identification of the precise localization of mutant precursor cystatin C and the mechanistic basis for this localization should prove fertile future grounds of investigation.

Impaired cystatin C secretion has also been observed in cultured human fibroblasts with chromosomally encoded precystatin C homozygous for the A25T mutation (31). These authors attributed the reduced secretion to failure of cleavage of the signal sequence. Our Western blot data from cell lysates also show slight retardation of fusion protein from the mutant construct, indicative of noncleavage of the signal sequence. However, the arguments presented above suggest that the A to T mutation would not be expected to inhibit cleavage of the signal sequence per se, but that failure of cleavage is more likely secondary, for example to failure of translocation of mutant precystatin C molecules across the endoplasmic reticulum membrane.

Concerning the pathological consequences of the mutation, the present findings highlight the question of whether the pathology is linked to insufficient secretion of cystatin C and/or to inappropriate intracellular retention. The following considerations should facilitate the design of further biochemical and microscopy-based experiments addressing this question. The exudative type of AMD is characterized by the migration of choroidal new vessels through Bruch's membrane, initially toward the basal aspect of the RPE monolayer. Previous in vitro experiments suggested that RPE-derived cystatin C is predominantly basally secreted by the cells, raising the possibility that this cysteine protease inhibitor is involved in controlling matrix turnover in Bruch's membrane (16). It is possible that relative insufficiency and/or inactivity of cystatin C in this location could alter the balance of matrix turnover in favor of lysis and hence augment infiltration of the chorioretinal interface by choroidal new vessels. On the other hand, the unexpected association of mutant precursor cystatin C with mitochondria, albeit intriguing, is conceptually coherent with a contributing role in progressive degeneration, especially in cells with high metabolic activity, such as RPE. Lastly, the partial impairment of trafficking is entirely consistent with the hypothesis that cystatin C deficiency is associated with chronic, progressive, age-related diseases such as AMD (18) and Alzheimer's disease (31) and not with acute pathology.

Materials and Methods

Fetal calf serum (FCS), penicillin-streptomycin and glutamine were supplied by Gibco Life Technologies (Paisley, UK). Tissue culture media and all other chemicals were supplied by Sigma-Aldrich Co. Ltd. (Poole, UK), unless otherwise stated.

Site-directed mutagenesis of precursor cystatin C–EGFP construct

The construction of the plasmid encoding the full-length precursor cystatin C fused in frame to the N-terminus of the EGFP, called pCysC-EGFP, made use of the EGFP-expressing vector pEGFP-N3 (Clontech, Palo Alto, CA) and was reported in detail previously (16). In vitro site-directed mutagenesis (G147A, numbers relating to the mRNA sequence with accession number X05607 (10)) was carried out using the QuikChangeTM Site-Directed Mutagenesis kit (Stratagene, La Jolla, CA), following the manufacturer's protocol. The primers used were HPLC purified (Invitrogen Ltd) and had the following sequences (with the nucleotide representing the point mutation in bold):


Both plasmid strands were replicated using the high fidelity PfuTurbo® DNA polymerase and the mutant primers described above. The mutation-containing synthesized DNA (nonmethylated) was selected following digestion of the parental DNA template with the methylated DNA-specific DpnI endonuclease. Following transformation into Escherichia coli XL1-Blue competent cells (Stratagene), kanamycin-resistant colonies were isolated and used to purify the mutated plasmid. The mutant construct, called Ala25Thr_pCysC-EGFP, was sequenced on both strands using vector-specific primers (EGFP-N sequencing primers, BD Biosciences Clontech). Comparison with the sequence of the original plasmid confirmed the desired mutation and the absence of any other mutations in the sequences coding for the precursor cystatin C and EGFP, which were also maintained in-frame.

Cell culture and transfections

Human RPE cells were harvested, characterized and cultured as previously reported (16). The cells were obtained in accordance with the guidelines of the Declaration of Helsinki. The conditions for culturing HeLa cells (ECACC no. 93021013) were as described previously (32). For confocal microscopy, cells were plated in Iwaki 35 mm culture dishes (Bibby Sterilin, Stone, UK) 24 h before transfection; 4 × 104 RPE cells and, respectively, 2.5 × 104 HeLa cells were plated per dish in 2 mL medium. Cells grown to 50–60% confluency were transfected with 1 μg of the appropriate endotoxin-free plasmid DNA (purified using a Qiagen EndoFree Plasmid Maxi Kit, Qiagen Ltd, Crawley, UK) using FUGENE 6TM reagent (Roche Molecular Biochemicals, Lewes, UK). The optimized ratio of DNA:FUGENE 6TM reagent used for transfections was 1 : 3 (v/v) for RPE cells transfections and 1 : 2 (v/v) for HeLa cells transfections. The efficiency of transfection was assessed 24 h post-transfection by visualizing cells under a fluorescence microscope.

Confocal imaging of living transfected cells

Confocal microscopy was carried out on transfected cells in Iwaki dishes placed in a CO2- and humidity-controlled stage-incubator, maintaining the cells at 37 °C and in 5% CO2 for the duration of the observation. Cells were imaged with a Zeiss LSM 510 confocal microscope with either a 40×/1.3 oil Ph3 Fluar objective or a 63×/1.4 oil DIC objective. Excitation of EFGP was achieved with a 488 nm argon ion laser and emitted light was reflected from a 540 nm dichroic mirror and then collected through a 505–530 nm bandwidth filter. The lipophilic probe Mitotracker Red CM-H2XRos (Molecular Probes, Inc., Eugene, OR) was used to stain the mitochondria in living cells. The probe, a nonfluorescent, chemically reduced form of the dihydro-X rosamine dye, diffuses passively across the plasma membrane and, upon oxidation by an actively respiring cell, acquires a deep red fluorescence − easily separable from the green fluorescence of the EGFP; the oxidized, fluorescent form is subsequently concentrated by active mitochondria (33). The mitochondria-selective probe was used at a final concentration of 200 nm in the culture medium. The mitochondria were imaged after 5 min of incubation with the dye, using a Helium Neon laser for excitation (543 nm) and capturing the emission through a 560 nm long-pass filter. Disruption of mitochondrial membrane potential was achieved with FCCP (carbonyl cyanide 4-trifluoro-methoxyphenylhydrazone), a proton ionophore used as an uncoupling agent (34,35). FCCP was added at a final concentration of 500 nm in the culture medium of transiently transfected RPE cells, 24 h or 48 h post-transfection, following incubation with Mitotracker probe. Images were obtained by line averaging four times, with either 512 × 512 or 1024 × 1024 pixel resolution. Data acquisition and analysis were carried out with lsm 510 software, version 3.2 (Zeiss, Welwyn Garden City, UK).

To study the dynamics of the fluorescently tagged cystatin C, selective bleaching of fluorescence was performed, using the laser at 100% power, directed to specific parts of cells for 30–40 s. Average intensity of fluorescence per pixel was measured in time series pre-and postbleaching, in determined areas of cells, using the LSM 510 software.

Preparation of crude mitochondrial fraction and treatment with trypsin

RPE cells were harvested from Iwaki Plate, 48 h after transfection. Following two washes with phosphate-buffered saline (PBS), pH 7.4, cells were detached in PBS with 0.25% trypsin and 0.02% EDTA, at 37 °C. Trypsinization was stopped after 2 min with FCS. For each preparation, cells from five or six Iwaki plates (approximately 1 × 106 cells) were pooled and pelleted by centrifugation at 120 × g for 10 min, at 4 °C. All subsequent steps were carried out at 4 °C or on ice. Crude mitochondrial fractions were obtained by differential centrifugation, as previously described (36,37). Briefly, cells were resuspended and pelleted five times in ice-cold Ca2+/Mg2+-free PBS, pH 7.4. The resulting pellet, swollen to roughly twice the original size, was resuspended in low ionic isotonic buffer SEM (0.3 m sucrose, 1 mm EDTA, 10 mm MOPS, pH 7.4). KCl was added to a final concentration of 20 mm and cells were mechanically homogenized, monitoring cell rupture by trypan blue staining. Unbroken cells and nuclei were pelleted by centrifugation at 750 × g for 10 min. Mitochondrial pellet was obtained by centrifugation of supernatant at 6800 × g for 20 min and resuspended in SEM buffer. Trypsin digestion of mitochondrial samples was performed with 20 (or 40, as indicated) μg enzyme/mL, on ice for 20 min, with and without preliminary addition of protease inhibitor cocktail, as per manufacturerís protocol (Sigma, P2714). Solubilization of mitochondrial fraction was achieved by incubation with CHAPS detergent at a final concentration of 50 mm for 30 min on ice. Following trypsin treatment, sample buffer was added and samples were incubated at 95 °C for 5 min prior to being subjected to electrophoresis. Immunodetection of cystatin C-EGFP and cytochrome c were carried out by Western blotting.

Western blot analysis

Cell lysates were obtained by washing cells cultured in Iwaki plates twice with PBS and then lysing them in 200 μl of lysis/sample buffer (125 mm Tris-HCl, 2% SDS, 20% v/v glycerol, 2% β-mercaptoethanol, 0.02% bromophenol blue, pH 6.8). Protein quantification in cell lysates and conditioned media were performed using a BIO-RAD Detergent Compatible protein assay kit (BIO-RAD, Hemel Hempstead, UK), following the manufacturer's protocol. To account for the presence of colored components in the respective samples, calibration curves were produced using bovine serum albumin solutions in dilutions of lysis buffer and, respectively, complete culture media. Proteins in cell lysates, culture media or mitochondrial preparations were resolved by SDS-polyacrylamide gel electrophoresis, alongside protein molecular weight markers (MBI Fermentas, Helena Biosciences, Sunderland, UK) and then subjected to Western blot analysis, following the same protocol described previously (16). The primary antibodies used for immunodetection were: anti-human cystatin C antibody (rabbit polyclonal IgG; Upstate Biotechnology, Lake Placid, NY), used at a concentration of 2 mg/mL; anti-GFP antibody (Living ColorsTM Peptide rabbit antibody; Clontech), diluted 1 : 100; anti-cytochrome c monoclonal antibody (Upstate Biotechnology), diluted 1 : 1000; anti-PEDF monoclonal antibody (Chemicon International, Temecula, CA), diluted 1 : 1000; anti-glyceraldehyde phosphate dehydrogenase (GAPDH) monoclonal antibody (Abcam Ltd, Cambridge, UK), diluted 1 : 2000. Detection was achieved by further labeling with horseradish peroxidase-conjugated goat anti-rabbit antibody (Sigma-Aldrich), respectively, anti-mouse antibody (Sigma-Aldrich), and incubation with an enhanced chemiluminescence detection system (ECLTM, Amersham Pharmacia Biotech, Little Chalfont, UK) followed by autoradiography. The densitometric evaluations were made using MCID Basic software (Interfocus Ltd, Cambridge, UK).


The authors are very grateful to Ted Maden for his advice throughout this work, in-depth discussions and contribution to the manuscript. Donna Gray is acknowledged for technical support with immunodetections. This work was supported by The Guide Dogs for the Blind Association and The Foundation for the Prevention for Blindness. Support for confocal microscopy was provided by HEFCE and Carl Zeiss Ltd. While developing the recombinant constructs, L.P. was supported by The Wellcome Trust.