Endogenous and recombinant H-2Kb molecules accumulate in a post-ER compartment
To examine whether the localization of H-2Kb is similar to previously reported data, we stained MEF cells with anti-P8, an antiserum against the cytosolic tail of H-2Kb [22]. Endogenous H-2Kb showed an ER-like stain in addition to a juxtanuclear accumulation (Fig. 1a; arrows). This accumulation resembled the ERGIC and cis-Golgi markers (Fig. 1b, c). However, even blocking the exit of H-2Kb from these compartments by low-temperature incubations (15 °C for the ERGIC, and 20 °C for the Golgi [23]), the pearsons’ coefficients for both the ERGIC and cis-Golgi channels with respect to H-2Kb channel did not show very high correlation with values ranging between 0.68 and 0.77, respectively. The cell surface was also stained, though weakly, in some cells (asterisks). Similar to previously reported data, endogenous H-2Kb exit the ER as partially or fully-loaded and reach post-ER compartments, which cannot be discerned by simple temperature blocks and colocalization comparisons. In addition, a small pool of the molecules can reach the cell surface.
We next compared the endogenous expression pattern with the green fluorescent protein (GFP) fusion of H-2Kb, GFP-H-2Kb, in which the ER-lumenal amino terminus of H-2Kb is fused to the carboxy terminus of GFP [2]. We used an N-terminal fusion, rather than the C-terminal fusion used by Edidin and coworkers [24]. MEF cells expressing recombinant H-2Kb-GFP were stained with different organelle markers. Then the GFP fluorescence was compared with that of the organelle antibody stain. In Fig. 2a, transfected cells were stained with anti-PDI (protein disulfide isomerase), an antibody that labels the ER peripheral tubules. GFP-Kb fluorescence was found in the central rough ER sheets, including the nuclear envelope, rather than the ER peripheral tubules. Upon 15 °C block, again inhibiting exit from the ERGIC, about 30 % of the cells showed a complete overlap of the accumulated molecules with the ERGIC area stained with anti-p58 antibody (Pearson’s coefficient ~ 0.83), and about additional 10 % of the cells showed a partial overlap (Data not shown) (Fig. 2b). A similar distribution pattern was observed when the accumulation of GFP-Kb was compared with the cis-Golgi marker, GM130, at 20 °C (Pearson’s coefficient ~ 0.64; Fig. 2c). The post-ER accumulation was not the result of endocytosed molecules from the cell surface, since there was no overlap between the fluorescence of GFP-Kb and the endosomal marker, EEA1 (Pearson’s coefficient ~ 0.32; Fig. 2d). Both H-2Kb endogenous and recombinant H-2Kb-GFP molecules exhibit similar intracellular distribution though the cell surface signal was not detected with the GFP fusions. The intensity signal at the cell surface could be low due to the high expression level of H-2Kb in the transfected cells. Thus, the ratio of partially to fully loaded molecules is highly masking the presence- if any- of cell surface proteins.
Taken together, our results so far suggest that the trafficking of a large pool of both endogenous and recombinant H-2Kb is jammed outside the ER. The calculated pearsons’ coefficients did not show complete correlation between the colocalized H-2Kb and organelle channels, but was slightly higher for the ERGIC than the cis-Golgi. This underlying fact points up the possibility of the involvement of another quality control compartment that might resemble as well the ERGIC and cis-Golgi compartments. This behavioral phenotype compounded with the cell surface expression of molecules might be restricted to MEF cells and is prone to variation based on the origin and overall characteristics of the tested cell line.
H-2Kb exit the ER in COPII vesicles
To characterize the nature of the trafficking pathway that H-2Kb follow to exit the ER, COPII vesicles that carry proteins from the ER to the ERGIC in mammalian cells were targeted [25]. COPII vesicles cannot be isolated from extracts due to their short lifetime in cells. Thus, they were produced from microsomal membranes by an in vitro budding reaction that Springer et al. group has established. [2, 26]. Different COPII formation reactions were performed with microsomes of MEF cells in the presence of cytosol (as a source of COPII components), ATP, and wild type recombinant Sar1 protein (the GTPase that drives COPII budding). Budded vesicles were then isolated by differential centrifugation, lysed, and GFP-H-2Kb or Na+/K+ ATPase, as a control protein for trafficking [3], were immunoprecipitated with anti-GFP serum and anti-Na+/K+ ATPase antibodies, respectively (Fig. 3a). Proteins were then detected by SDS-PAGE and autoradiography. In the absence of cytosol (lane 3), absence of ATP (lane 4), or in the presence of mutant Sar1(T39N) as a specific inhibitor for COPII vesicles (lane 5), the amounts of H-2Kb and ATPase found in the vesicle fraction were reduced (Fig. 3, lanes 3 to 5). The packaging efficiency for all the lanes was quantified relative to lane 7 (Fig. 3b). Comparable budding efficiency for both proteins was measured, with four fold reduction in lanes 3 and 4. This indicates that the isolated vesicles were indeed COPII vesicles. In lane 8, sample was left untreated with EndoF1 revealing a shift of H-2Kb alleles from sensitive form (s)GFP- H-2Kb to glycosylated form (g) GFP- H-2Kb. Thus, H-2Kb in COPII vesicles are EndoF1-sensitive, revealing their trafficking out of the ER en route to the ERGIC and cis-Golgi but they can not travel beyond the median Golgi. Interestingly, addition of peptides to the complete budding reaction (lane 1) or to donor membranes (lane 7) did not increase the intensity of packaged proteins pointing to the existence of mostly unfolded heavy chains H-2Kb in COPII vesicles. This can also point to the ER-egress of not only partially folded proteins, but also heavy chains of H-2Kb molecules.
In vitro FKBP/FRB rapamycin-trapping assay is adopted to further confirm the trafficking of H-2Kb
To follow up on the above observation we used a rapamycin trapping assay, which utilizes the non-covalent crosslink that occurs between two domains, FKBP (FK506 binding protein) and FRB (FKBP-rapamycin binding domain), upon addition of rapamycin [27]. These domains were conjugated to the C-termini of the Venus and Cerulean fluorescent proteins [28]. Cerulean- FKBP and Venus-FRB were then fused to the amino termini of H-2Kb, tapasin, or various organelle markers (Table 1); in this way, triple fusions, such as Venus-FRB-Kb and Venus-FRB-tapasin were generated.
Initially, we examined the proper size of these constructs in 293 T cells (Fig. 4). 293 T cells were either left untranfected (lane 1) or cotransfected with Venus-FRB-Kb and Venus-FRB-tapasin (lane 2) or with Venus-FRB-Kb and wt Tapasin (lane 3). H-2Kb were pulled down with anti-P8 antibody (lanes 1 to 3) followed by blotting with anti-tapasin antibody, #2668. Venus-FRB-Kb had the expected size of approximately 80 kDa and interacted with both Venus-FRB-Tapasin (lane 2) and wild type tapasin (lane 3).
Because of the previous observation by us and others that class I molecules cycle between the ER and the Golgi at steady state [11, 20], we next examined the cycling of accumulated MHC class I molecules in Jurkat cells using the rapamycin trapping assay. MEF cells were substituted by Jurkats as the latter was easier to double and triple transfect. In addition, Jurkats are lymphocytes with a better system that mimics and coordinates the normal behavioral function of MHC class I and the peptide loading complex proteins. Lastly, Jurkats possess a smaller surface area, which facilitates the tracking of recycled molecules. We cotransfected Jurkat cells with Venus-FRB-Kb and Cerulean-FKBP-Ii. Cells showed Cerulean-FKBP-Ii expression in the ER, whereas Venus-FRB-Kb was detected not only in the ER, but also at the cell surface and in intracellular accumulations (Fig. 5, first panel). If Venus-FRB-Kb cycles back to the Golgi from post-ER compartment, then accumulating molecules should be trapped by Cerulean-FKBP-Ii in the ER upon addition of rapamycin. Since, Ii protein is permanently residing in the ER, then we expect any recycling H-2Kb to be trapped by Ii resulting in a decrease of the fluorescence intensity of the post-ER accumulated molecules over time. To rule out the artifact of Ii trapping newly synthesized Kb in the ER, we incubated the cells with cycloheximide for ten minutes prior to rapamycin addition. Cycloheximide inhibits the elongation of protein translation and thus, allows us to evaluate the ER-trapping of accumulated H-2Kb rather than Kb already present in the ER. Consistent with our hypothesis, we observed that the accumulation of Venus-FRB-Kb disappeared after one hour treatment with rapamycin (Fig. 5, second panel). In the controls, without adding rapamycin or in the absence of an FKBP domain, the accumulated pool of H-2Kb persisted even at prolonged incubation period. This trimmed configuration of Cerulean-Ii, “No FKBP”, offers a good control to directly assess the recycling and trapping of Venus-FRB-Kb molecules. It is worth noting that the around 40 % of the cells did not show a disintegration of the accumulated molecules.
The disappearance of accumulated Venus-FRB-Kb over time is not due to the disruption of the Golgi apparatus
Treatment of cells for prolonged period with cycloheximide might disrupt the structure of post-ER organelles, since they depend on newly synthesized proteins to maintain their integrity. On a related note, the trafficking of Scy11 between the Golgi and ER is required for the maintenance of the Golgi apparatus [29]. To test whether the relocation of Venus-FRB-Kb molecules to the ER is happening as a result of its trapping and not due to the Golgi disassembly, we triple transfected Jurkat cells with Venus-FRB-Kb, Cerulean-FKBP-Ii, and GalT-mCherry. GalT (1,4-galactosyltransferase) is a marker for the trans-Golgi membranes and the trans-Golgi network. Cells were then either treated with cycloheximide in the presence or absence of rapamycin, or left untreated. The Golgi structure (as detected by the fluorescence of GalT-mCherry) was maintained even after prolonged (6 h) incubation with cycloheximide (Fig. 6, first panel). Thus, the trapping effect is not mediated by the disintegration of the Golgi structure, but rather reflects the recycling of our tracked proteins. Consistent with our hypothesis, MHC class I trapping by Cerulean-FKBP-Ii was again observed (Fig. 6, second panel). This confirms the cycling of Venus-FRB-Kb molecules between the cis-Golgi and the ER and validates our assay.
Live- cell imaging to examine the cycling of recombinant Venus-FRB-Kb
We next followed the fluorescence intensity of accumulated Venus-FRB-Kb molecules in live cell microscopy over a time period of 45 min from the time of rapamycin addition, which is ten to fifteen minutes after addition of cycloheximide. The movie consists of concatenated images taken at time interval of 2 min. Each image is in turn obtained by a compilation of four z-stacks (z = 1 μm). Six different images taken from the movie at time points: 0, 6, 12, 18, 22, and 40 min are depicted in Fig.
7
. The fluorescence intensity (FI) of Venus-FRB-Kb was analyzed by calculating Pearson’s coefficient from a circular region of interest (ROI) drawn around the region in the double transfected cell (red and green = RG), where the green intensity is more concentrated and denser than the red intensity (RG cell in Fig. 7a). As a control, Pearson’s coefficient was calculated from an ROI around a single transfected cell expressing only Cerulean-FKBP-Ii (R cell in Fig. 7). The time-dependent decrease of the accumulated pool of the RG cell is demonstrated by the increase in the Pearson’s coefficient values over time reaching almost 0.9 at around 40 min (Fig. 7b). For the control cell (R cell), Pearson’s coefficient was constant unaffected by rapamycin addition. Interestingly, the percentage of cells showing a morphology of MHC class I only in the ER, without any Golgi or cell surface expression, increased from 30 % to almost 70 % in the presence of rapamycin (Fig. 7c).
Egress of H-2Kb to the cell surface upon addition of peptides
Finally we assessed the localization of Venus-FRB-Kb in MEF cells. Similar to endogenous H-2Kb and of GFP-H-2Kb, they were mostly localized to the ER and showed an additional accumulation close to the nucleus with a weak delineation of the cell surface (data not shown). Cells were pretreated with cycloheximide for 10 min- to avoid cross-linking of the nascent proteins in the ER- prior to their one-hour incubation with or without rapamycin.. As expected, after treatment with rapamycin, 60 % of the cells lacked a juxtanuclear accumulation and mostly showed an ER-like pattern for Venus-FRB-Kb with few vesicular structures that could be late endosomes (data not shown). Trapping failed to occur in cells co-transfected with Cerulean-FKBP-Ii and Venus-Kb lacking the FRB domain (data not shown).
We next tested whether the cell surface presence of Venus-FRB-Kb can be increased by trapping recombinant Kb at the cell surface by Cerulean-FKBP-GPI. The Cerulean-FKBP- protein is modified in the ER by its fusion with a GPI (glycosylphosphatidylinositol) peptide signal moiety that anchors it to the cholesterol-rich lipid microdomains of the plasma membrane and thus maintains its steady state localization at the cell surface [30]. For this, we double transfected cells with Venus-FRB-Kb and Cerulean-FKBP-GPI. To extend the residence time of Venus-FRB-Kb at the cell surface, we added 10 μM of high-affinity peptide (SIINFEKL in the single letter amino acid code) to the medium for one hour. As shown in Fig. 8, the addition of rapamycin to cells incubated with cycloheximide and peptides increased the fluorescence of Venus-FRB-Kb at the cell surface. This advocates the possible peptide-loading of H-2Kb in a post-ER compartment, rescuing their egress to the cell surface. More experiments should be done to reveal a clear statistics on the overall behavior of the cells.