Role of the EF-hand and coiled-coil domains of human Rab44 in localisation and organelle formation

Rab44 is a large Rab GTPase that contains an amino-terminal EF-hand domain, a coiled-coil domain, and a carboxyl-terminal Rab GTPase domain. However, the roles of the EF-hand and coiled-coil domains remain unclear. Here, we constructed various deletion and point mutants of human Rab44. When overexpressed in HeLa cells, the wild-type Rab44 (hWT) formed ring-like structures, and partially localised to lysosomes. The dominant negative mutant, hT847N, localised to lysosomes and the cytosol, while the constitutively active mutant, hQ892L, formed ring-like structures, and partially localised to the plasma membrane and nuclei. The hΔEF, hΔcoil, and h826-1021 mutants also formed ring-like structures; however, their localisation patterns differed from hWT. Analysis of live imaging with LysoTracker revealed that the size of LysoTracker-positive vesicles was altered by all other mutations than the hC1019A and hΔEF. Treatment with ionomycin, a Ca2+ ionophore, induced the translocation of hWT and hΔcoil into the plasma membrane and cytosol, but had no effect on the localisation of the hΔEF and h826-1021 mutants. Thus, the EF- hand domain is likely required for the partial translocation of Rab44 to the plasma membrane and cytosol following transient Ca2+ influx, and the coiled-coil domain appears to be important for localisation and organelle formation.

www.nature.com/scientificreports/ raised fundamental questions about the differences in localisation and function of Rab44 between humans and mice. Indeed, our recent study using the murine macrophage cell line, RAW-D cells, showed that mouse Rab44 localised mainly to lysosomes and the Golgi complex 17 . There are currently no studies on the localisation of human Rab44. Moreover, the roles of the amino-terminal EF-hand and coiled-coil domains in the localisation of human Rab44 remain unclear.
In this study, we investigated the mechanisms involved in the localisation of human Rab44 by ectopically expressing various Rab44 mutants in HeLa cells, which are human epithelial cells most often used in intracellular localisation analyses.

Results
Ectopic expression and localisation of wild type and mutants of human Rab44 in HeLa cells. To investigate the subcellular localisation of human Rab44, we constructed various mutants of human Rab44 by site-directed mutagenesis (Fig. 1a), including wild-type human Rab44 (hWT), an EF-hand motif deletion (hΔEF), a coiled-coil domain deletion (hΔcoil), a Rab domain only (h826-1021), a constitutively active (CA) mutant (hQ892L), a dominantly negative (DN) mutant (hT847N), and C-terminal lipidation site mutants with point mutations at residues 1019 and 1020 (C1019A, C1020A, and C1019A/C1020A). The hWT and mutant constructs were exogenously expressed as N-terminal green fluorescent protein (GFP)-fusion proteins in HeLa cells (Fig. 1a). The protein levels of the mutants in the HeLa cells were confirmed by western blot analysis (Fig. 1b). The expressed proteins were detected at their predicted molecular weights, although the expression levels of h826-1021, hT847N, and C1019A were relatively low.
The ectopically expressed hWT formed ring-like structures, and partially surrounded the LAMP1-positive lysosomes in HeLa cells (Fig. 2a). However, the hWT hardly merged with EEA1 (a marker for early endosomes)and GM130 (a marker for the Golgi complex)-positive compartments. (Fig. 2a). Interestingly, the hWT partially merged with KDEL [a marker for endoplasmic reticulum (ER)]-positive compartments. Quantitative analysis of colocalisation between the hWT and these organelle markers is shown in Fig. 2b. We determined that an average score of 0.4 or more is colocalised, 0.2 to 0.4 is partially colocalised, and score of 0.2 or less is not colocalised. Similar determinations were performed in subsequent experiments.
The DN mutant, hT847N, was mostly diffuse, indicating cytosolic localisation, and colocalised partially with LAMP1-positive lysosomes and slightly with the KDEL-positive ER (Fig. 3a). However, hT847N failed to merge with EEA1-positive early endosomes and the GM130-positive Golgi complex (Fig. 3a). Intriguingly, HeLa cells expressing the DN mutant hT847N were rounder in shape than those expressing hWT (see Fig. 2). Quantification of colocalisation between the hT847N and these organelle markers is also shown in Fig. 3b.
The CA mutant, hQ892L, formed ring-like structures and was partially detectable in LAMP1-and EEA1positive compartments, the plasma membrane, and the nuclei, but hardly detectable in the GM130-and KDELpositive compartments (Fig. 3c). Of note, in hQ892L-expressing HeLa cells, the shape of the cells and the nuclei were distorted, the cell-cell contact sites were unclear, and hQ892L appeared to promote cell-cell adhesion (Fig. 3c). These results were confirmed by quantitative analysis of colocalisation between the hQ892L and these organelle markers (Fig. 3d).
One of the lipidation-site mutants, hC1019A, was diffusely distributed throughout the cytoplasm, although it partially localised to the KDEL-positive compartments (Fig. 3e). However, the hC1019A mutant was hardly detectable in the LAMP1-, GM130-, and EEA1-positive compartments (Fig. 3e). Consistent with these results, hC1019A colocalisation analysis revealed a colocalisation score of 0.4 with KDEL, and only 0.2 or less with other organelle markers (Fig. 3f).
The double mutant hC1019/1020A was also mostly diffusely distributed in the cytoplasm with partial colocalisation with KDEL-positive compartments, and was undetectable in other organelles (Fig. 3i). Quantification of colocalisation between the hC1019/1020A and these organelle markers validated these observations (Fig. 3j).
The hΔcoil mutant formed ring-like structures and mainly merged with LAMP1-positive lysosomes, but not with EEA1-, GM130-, or KDEL-positive compartments (Fig. 4c). Quantification of the colocalisation of hΔcoil mutant with these organelle markers showed that its localisation pattern differed from hWT (Fig. 4d). The colocalisation score of the hΔcoil mutant with LAMP1-positive lysosomes was higher than that of hWT, and that with KDEL-positive compartment was lower than that of hWT, suggesting that fusion of the hΔcoil mutant with lysosomes is promoted but fusion with the ER is reduced (Fig. 4d).

Comparison of colocalisation of the human Rab44 mutants with LAMP1 and LysoTracker.
Effects of Ca 2+ modulators on localisation of human Rab44. Next, we assessed whether Ca 2+ influx affects the subcellular localisation of human Rab44, since Rab44 encodes an EF-hand domain. As shown in Fig. 7, under basal conditions, hWT formed ring-like structures and partially surrounded the LAMP1-positive lysosomes in HeLa cells. However, when we treated hWT-expressing HeLa cells with ionomycin, a selective Ca 2+ ionophore, the hWT partially translocated to small vesicles in the marginal region and to the plasma membrane and cytosol (Fig. 7a). We further examined the effects of other Ca 2+ -related reagents on the localisation of hWT. Upon treatment with thapsigargin, a Ca 2+ -ATPase inhibitor in the ER, hWT was detected in many vesicles distinct from LAMP1-positive lysosomes (Fig. 7a). ML-SA1 is a specific agonist for the lysosomal calcium channels transient receptor potential channel mucolipins (TRPML1-3). When hWT-expressing HeLa cells were treated with ML-SA1, the hWT localised to the plasma membrane and cytosol, and was also partially detectable in the lysosomes and non-lysosomal compartments (Fig. 7a). Following a quantitative analysis, we found that all the Ca 2+ modulators significantly decreased colocalisation of hWT and LAMP1-positive lysosomes (Fig. 7b). These results indicate that transient Ca 2+ influx induced by ionomycin or ML-SA1 causes partial translocation of hWT from the lysosomes to the plasma membrane and cytosol. In contrast, inhibition of intracellular Ca 2+ with thapsigargin induces localisation of hWT to non-lysosomal vesicles.

Effects of ionomycin-mediated Ca 2+ influx on localisation of various human Rab44 mutants.
We further assessed whether ionomycin-mediated Ca 2+ influx affects the subcellular localisation of human Rab44 and its mutants. Localisation of the DN mutant hT847N was unaffected by ionomycin treatment, and was partially detectable both in the cytosol and with LAMP1-positive lysosomes (Fig. 8a). Without stimulation, the CA mutant hQ892L was detected in LAMP1-positive lysosomes surrounded by ring-like structures and partially in the plasma membrane (Fig. 8b). However, following ionomycin treatment, the formation of ring-like structures surrounding the LAMP1-positive lysosomes was slightly enhanced -though the difference was not statistically significant -and its localisation to the plasma membrane decreased (Fig. 8b). Importantly, the hΔEF mutant formed ring-like structures even after ionomycin treatment, and its localisation was nearly unchanged by stimulation (Fig. 8c). Therefore, the EF-hand domain is likely to be important for Ca 2+ -mediated localisation of human Rab44. The hΔcoil mutant localised partially to lysosomal and non-lysosomal vesicles without ionomycin, and was partially localised to the cytosol with ionomycin treatment (Fig. 8d). Ionomycin treatment significantly reduced the colocalisation ratio of the hΔcoil mutant with LAMP1-positive lysosomes (Fig. 8f). Ionomycin had no effect on the localisation of the h826-1021 mutant to non-lysosomal vesicles or its partial colocalisation with LAMP1-positive lysosomes (Fig. 8e). Quantification of the effects of ionomycin on colocalisation of Rab44 mutants with LAMP1-positive lysosomes is shown in Fig. 8f. Ionomycin treatment significantly reduced the colocalisation of hWT or hΔcoil mutant with LAMP1-positive lysosomes, but had little effect on the mutants lacking the EF-hand domain, such as the hΔEF and h826-1021 (Fig. 8f). Thus, the localisation of Rab44 is altered by ionomycin, and the EF-hand domain of human Rab44 is important for Ca 2+ -mediated regulation of Rab44 localisation.

Discussion
In this study, we constructed various mutants of human Rab44 using site-directed mutagenesis. Wild-type human Rab44 (hWT) and the Rab44 mutant constructs were expressed exogenously in HeLa cells.
The expression level of the h826-1021 mutant was extremely low, for which there may be a few possible causes. First, whereas the wild-type protein contains 1021 amino acid residues, the h826-1021 is reduced to 19% of its original size with only 196 amino acid residues. Second, this fragment alone may be unstable as a protein. In other words, the N-terminal domains may be important for stable formation of the protein, even though the structure of this mutant is similar to that of other small Rab proteins.
Upon analysing the DN and CA mutants, Rab44 was found to share some common features with other small Rab GTPases. In many small Rab GTPases, differences in localisation have been observed between the DN and CA mutants. For example, when human Rab21 was overexpressed in HeLa cells, it predominantly localised to EEA1-containing early endosomes. The DN mutant Rab21(T33N) was concentrated in the perinuclear region, whereas the CA mutant Rab21(Q78L) localised to the reticular tubular network 20 . Therefore, it is possible that differences in localisation between the DN mutant hT847N and CA mutant hQ892L affect cell shape: hT847N caused a round cellular shape with clear contacts between cells, whereas hQ892L resulted in unclear cell-cell contact sites.
The lipidation-site mutants of Rab44 also shared some common features with those of the small Rab GTPases. The prenylatable cysteines are key factors for membrane targeting of Rab GTPases. For example, lipidation-site mutations in both Rab4 and Rab5 cause complete cytosolic localisation 21 . Similarly, the lipidation-site mutant, hC1019A, was mostly localised to the cytosol and partially to lysosomes (Fig. 3c), whereas the lipidation-site mutants hC1020A and hC1019/1021A completely localised to the cytosol (Fig. 3d,e). Interestingly, hC1019A partially localised to LAMP1-positive lysosomes, whereas hC1020A was mostly diffusely distributed throughout the cytoplasm. Therefore, hC1020A appears to have a greater effect on subcellular localisation than hC1019A. www.nature.com/scientificreports/ The EF-hand domain, which is known to be the Ca 2+ binding site 22 , is likely to be required for the partial translocation of Rab44 into the plasma membrane and cytosol under conditions of transient Ca 2+ influx. Consistent with this notion, transient Ca 2+ influx caused partial translocation of the constructs containing the EF-hand domain, including hWT and hΔcoil, into the plasma membrane and cytosol. Considering that the CA mutant, hQ892L, partially localised to the plasma membrane and cytosol without stimulation, the EF-hand domain of Rab44 may be an important activation factor for translocation induced by Ca 2+ mobilisation. In contrast, translocation of constructs lacking the EF-hand domain, including hΔEF and h826-1021 mutants, were virtually unaffected by ionomycin-mediated transient Ca 2+ influx conditions. Taken together, the EF-hand domain of Rab44 is probably important for Ca 2+ -mediated translocation.
The coiled-coil domain appears to be important for localisation and formation of LysoTracker-positive vesicles as described below. Given that mutants lacking the coiled-coil domain, such as hΔcoil and h826-1021 mutants, had a higher colocalisation coefficient with LAMP1-positive lysosomes and a lower one with KDEL-positive www.nature.com/scientificreports/ compartment compared with hWT, it is likely that fusion of the mutants lacking the coiled-coil domain with lysosomes is increased but fusion with the ER is decreased. Generally, the coiled-coil domain-containing proteins have been proposed to function as tethering factors to target organelles prior to fusion or as scaffolds for the assembly of other factors important for fusion 23,24 . Therefore, it would be interesting to investigate the molecular mechanisms through which the Rab44 coiled-coil domain assists in localisation and organelle formation. The size of LysoTracker-positive vesicles was affected by Rab44 expression and mutation. hWT induced larger vesicles compared to the control, the DN mutant hT847N induced smaller vesicles than the WT, and the CA mutant hQ892L induced larger vesicles than hWT; therefore, it is likely that Rab44 has effects on a large formation of LysoTracker-positive vesicles. Moreover, it is interesting that the combination of hC1020A and hC1019/1020A mutants and the combination of hΔcoil and h826-1021 mutants had the exact opposite effects on formation of LysoTracker-positive vesicles, although the mechanisms behind these effects remain unknown.
Interestingly, differences in colocalisation between LAMP1 and LysoTracker were observed in the mutants that are diffusely localised, such as hQ892L, hC1019A, hC1020A, and hC1019/1020A (Fig. 6). Specifically, these mutants had decreased colocalisation with LAMP1, and increased colocalisation with LysoTracker compared to hWT. There are known to be some differences between LAMP1-positive compartments and LysoTrackerstained organelles 25 . Therefore, we speculate that LysoTracker may detect LAMP1-negative acidic vesicles. Indeed, LysoTracker has reported to be an acidotropic dye that stains intermediate compartments, including autolysosomes, and phagolysosomes as well as late endosomes/lysosomes 26,27 .
Ionomycin-and ML-SA1-induced transient Ca 2+ influx caused partial translocation of hWT from the lysosomes into the plasma membrane and cytosol. In contrast, inhibition of intracellular Ca 2+ using thapsigargin induced translocation of hWT to non-lysosomal vesicles. Thus, the translocation of Rab44 is regulated by the intracellular Ca 2+ level, and the EF-hand domain of Rab44 is important for the Ca 2+ -mediated translocation.
In conclusion, the roles of the EF-hand, coiled-coil domains, and lipidation sites in human Rab44 were analysed using deletion and point mutants. The EF-hand domain is required for partial translocation of Rab44 into the plasma membrane and cytosol under conditions of transient Ca 2+ influx, and the coiled-coil domain is important for localisation, formation of LysoTracker-positive vesicles. The lipidation-site is essential for localisation to the membrane. The PCR products were cloned into the retroviral vector, pMSCVpuro (Clontech, Mountain View, CA, USA), using an In-Fusion cloning kit (Clontech). The pMSCVpuro vector with an eGFP fragment insert yielding N-terminal eGFP-fusion proteins was kindly provided by Prof. Kosei Ito (Nagasaki University, Japan). The vectors were transfected into HEK293T cells using Lipofectamine 3000 (Life Technologies, Gaitherburg, MD, USA), according to the manufacturer's instructions. After incubation at 37℃ in a 5% CO 2 atmosphere for 48 h, the supernatants containing the viral particles were collected and used to infect HeLa cells. Cells stably expressing Rab44 were selected using puromycin (5 μg/mL) in the culture medium, and media was changed every third day after 3 weeks.

Materials and methods
Western blot analysis. Western blotting analysis was performed according to the protocol described previously [28][29][30] . Briefly, the cells were lysed in cell lysis buffer supplemented with protease inhibitors. Equal amounts of protein were subjected to sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) www.nature.com/scientificreports/ followed by transfer onto a polyvinylidene difluoride (PVDF) membrane. The blots were blocked with 5% milk in Tris-buffered saline (TBS) for 1 h at 25 °C, incubated with anti-GFP antibody (1:3000) for 2 h at 4 °C and washed four times with TBS containing 0.1% Tween 20. The blots were then incubated with horseradish peroxidase-conjugated secondary antibodies for 1 h at 25 °C, washed, and detected with Immobilon Forte ECL HRP substrate (Merck-Millipore, Burlington, MA, USA). Immunoreactive bands were analysed using a LAS-4000 Mini imaging system (Fujifilm, Tokyo, Japan). The membranes were reprobed with anti-GAPDH antibody, and analysed as above.
Immunofluorescence microscopy. Cells were cultured on cover glasses and fixed with 4.0% paraformaldehyde in phosphate buffered saline (PBS) for 20 min at 25 °C. The fixed cells were then washed with PBS for 5 min twice, and permeabilised with 0.1% Triton X-100 in PBS for 15 min. The cells were blocked with 0.2% gelatin in PBS for 5 min, and subsequently incubated with primary antibodies for 1 h at 4 °C. The cells were washed with PBS-gelatin three times, and then incubated with the secondary antibody, Alexa Fluor 555-conjugated goat anti-mouse IgG. Nuclear staining was then performed using DAPI. The samples were subjected to microscopy using a laser-scanning confocal imaging system (LSM800; Carl Zeiss, AG, Jena, Germany) and analysed by Airyscan processing (ZEN2.3 software, Carl Zeiss). Subcellular distribution and colocalisation were quantified using Pearson's correlation coefficients. Quantitative data are presented as mean ± standard deviation (SD). Statistical analyses were performed using Prism 7 (GraphPad, San Diego, CA, USA). Unpaired t-tests were used to identify differences when a significant difference (*P < 0.05, **P < 0.01, ***P < 0.001, or ****P < 0.0001) was determined by analysis of variance. Ionomycin, thapsigargin, and ML-SA1 treatment. HeLa cells grown on cover glasses were incubated with 2 µM ionomycin, 10 µM thapsigargin, or 10 µM ML-SA1, for 15 min at 37 °C in a 5% CO 2 atmosphere, and then immediately fixed with paraformaldehyde followed by immunofluorescence microscopy.