TBC1D10C is a cytoskeletal functional linker that modulates cell spreading and phagocytosis in macrophages

Cell spreading and phagocytosis are notably regulated by small GTPases and GAP proteins. TBC1D10C is a dual inhibitory protein with GAP activity. In immune cells, TBC1D10C is one of the elements regulating lymphocyte activation. However, its specific role in macrophages remains unknown. Here, we show that TBC1D10C engages in functions dependent on the cytoskeleton and plasma membrane reorganization. Using ex vivo and in vitro assays, we found that elimination and overexpression of TBC1D10C modified the cytoskeletal architecture of macrophages by decreasing and increasing the spreading ability of these cells, respectively. In addition, TBC1D10C overexpression contributed to higher phagocytic activity against Burkholderia cenocepacia and to increased cell membrane tension. Furthermore, by performing in vitro and in silico analyses, we identified 27 TBC1D10C-interacting proteins, some of which were functionally classified as protein complexes involved in cytoskeletal dynamics. Interestingly, we identified one unreported TBC1D10C-intrinsically disordered region (IDR) with biological potential at the cytoskeleton level. Our results demonstrate that TBC1D10C shapes macrophage activity by inducing reorganization of the cytoskeleton-plasma membrane in cell spreading and phagocytosis. We anticipate our results will be the basis for further studies focused on TBC1D10C. For example, the specific molecular mechanism in Burkholderia cenocepacia phagocytosis and functional analysis of TBC1D10C-IDR are needed to further understand its role in health and disease.


Results
Immune cells show morphological defects in the cytoskeletal architecture upon changes in TBC1D10C expression. To investigate TBC1D10C, we generated TBC1D10C knockout (KO) mice by the Cre-Lox recombination system (Supplementary Fig. 1a-c). TBC1D10C knockout (KO) mice developed normally in the expected Mendelian ratio, and no abnormalities were detected, but they were more docile than WT (wild type) animals (data not shown). Initially, we studied immune cells from the spleen (splenocytes) 24,25 . Splenocytes were stained with phalloidin to label F-actin, and a morphometric examination was carried out. Our analysis showed that the spread area of KO splenocytes was reduced compared to that of their WT counterparts (Fig. 1a). Subsequently, we conducted the same morphometric analysis for immune cells from bone marrow, and we studied BMDMs (bone marrow-derived macrophages). We also observed that KO BMDMs showed a reduced spread area (Fig. 1b), thus corroborating that alterations related to the actin cytoskeleton of KO splenocytes (Fig. 1a) were not specific to immune cells obtained from the spleen. We focused our research on BMDMs because the role of TBC1D10C in macrophages has not been investigated. Consequently, we overexpressed TBC1D10C in RAW 264.7 cells to compare the effect of TBC1D10C elimination vs. overexpression ( Supplementary Fig. 1d). In contrast to KO BMDMs, overexpression of TBC1D10C was associated with an increased spread area (Fig. 1c). Our initial data suggested that TBC1D10C influences morphological changes in macrophages through actin-cytoskeleton changes.
TBC1D10C modifies cytoskeleton-plasma membrane dynamics. Morphology and cytoskeleton dynamics are intimately related 26,27 . To gain more insights into the participation of TBC1D10C in the actin cytoskeleton, we measured cell spreading dynamics (aspect ratio dynamics) in macrophages by live cell imaging. This analysis indicated that KO BMDMs showed fewer cell spreading dynamics. Biologically, these KO cells were significantly less efficient at modifying their cellular size compared to WT BMDMs, which in turn tended to retract their cellular bodies during the analyzed time ( Fig. 2a,b). Unexpectedly, despite the inefficiency of KO macrophages, the velocity at which these cells changed their cellular size was statistically greater than that of WT BMDMs (Fig. 2c). In the first instance, we expected that this velocity would also be lower. Why, then, was a greater velocity of aspect ratio dynamics recorded in KO BMDMs? To shed light on this effect, we analyzed multiple kymographs (through the cellular body of every cell; horizontal and vertical ROI every 1 µm). This analysis indicated that KO BMDMs frequently showed a phenotype consisting of detached filopodia that suddenly moved toward the cell body ( Fig. 2d and Supplementary Movies 1-2). This phenotype was present in a significantly higher number of KO BMDMs than WT BMDMs (Fig. 2e) and seemed to affect cell spreading dynamics. In addition, quantification of filopodia retraction velocity showed more than double the velocity in KO cells (118.1% more; Fig. 2f), thus possibly increasing the quantification of the velocity of cell spreading dynamics (Fig. 2c). Afterward, we analyzed RAW 264.7 macrophages similarly and found that TBC1D10C overexpression was associated with an increase in both cell spreading dynamics and the corresponding velocity. In addition, macrophages overexpressing TBC1D10C showed a tendency to spread their cellular bodies (Fig. 2g,h,i). Together, our data strongly propose that TBC1D10C participates in macrophage spreading dynamics and that such participation correlates with defects of the cytoskeleton architecture.
TBC1D10C contributes to Burkholderia cenocepacia phagocytosis by RAW 264.7 macrophages. The cytoskeleton participates in multiple cellular activities, including phagocytosis. In this process, macrophages must change shape through substantial membrane and cytoskeleton remodeling 28 . To determine whether TBC1D10C plays a role in this cellular function, we infected KO and WT BMDMs with Burkholderia cenocepacia 29 , an opportunistic pathogen that infects and survives in macrophages by altering actin polymerization 30 . To quantify only ingested bacteria, we used the gentamicin protection assay. TBC1D10C www.nature.com/scientificreports/ elimination in BMDMs was not associated with a significant increase or decrease in the amount of phagocytosed B. cenocepacia (Fig. 3a). However, its overexpression in RAW 264.7 macrophages correlated with a higher number of phagocytosed bacteria (Fig. 3b). Likewise, it has been reported that macropinocytosis contributes to the internalization of this bacterium by macrophages 31 ; therefore, we evaluated this endocytic pathway using 10 kDa dextran (a fluid-phase marker ingested by macropinocytosis) 32 , but no significant differences were observed ( Supplementary Fig. 2). Furthermore, by evaluating the capacity of macrophages to kill ingested bacteria, we determined that both WT and KO BMDMs eliminated intracellular bacteria similarly (Fig. 3c), whereas RAW 264.7 cells transfected with control vector and TBC1D10C were unable to efficiently kill phagocytosed microorganisms (Fig. 3d). This indicates a differential capacity to control B. cenocepacia by BMDMs and RAW 264.7 macrophages; consequently, the higher number of internalized bacteria upon TBC1D10C overexpression was not specifically related to differential elimination of bacteria or increased macropinocytosis. On the other hand, to validate the specific effect of TBC1D10C on B. cenocepacia phagocytosis, we transfected RAW 264.7 cells with Myosin 1G, an unrelated protein to TBC1D10c, which is not known to participate in the phagocytosis of B. cenocepacia. Again, we evaluated phagocytosis and found that Myo1g overexpression in RAW 264.7 macrophages was not associated with increased phagocytosis of this bacteria (Fig. 3e), thus indicating a specific contribution of TBC1D10C to B. cenocepacia phagocytosis. As further evidence, we evaluated phagocytosis by flow cytom-  TBC1D10C is a protein intimately associated with cytoskeleton-plasma membrane reorganization and tension. Cell spreading and phagocytosis are related cellular processes that notably require cytoskeleton-plasma membrane reorganization as well as the participation of multiple proteins 28,33,34 ; our data indicate that TBC1D10C participates in such cellular functions. To better understand the mechanistic contribution of TBC1D10C to cell spreading and phagocytosis, we carried out proteomic analysis using stably transfected RAW 264.7 cells (GFP tag-TBC1D10C). TBC1D10C was immunoprecipitated to identify proteins in the complex by mass spectrometry followed by bioinformatics analysis. This led us to identify 27 TBC1D10C-interacting proteins (Table 1), which were functionally grouped in cytoskeleton organization, protein localization to orga- www.nature.com/scientificreports/ nelles, protein translation, and ATP hydrolysis coupled to proton transport ( Fig. 4a), in addition to others (Supplementary Table 1). The proteins identified and functionally grouped in cytoskeletal organization are tubulin, plectin, Gapdh, and Rab35. The identification of Rab35 as a TBC1D10C-interacting protein in macrophages, in principle, suggests a specific role for TBC1D10C GAP activity in cell spreading and phagocytosis. Therefore, we performed experiments to evaluate the contribution of this GAP activity as well as its biological effect on Rab35. This activity was represented by the use of the TBC1D10C-GAP mutant (R141K) and Rab35 DN (Rab35 dominant negative, S22N, which theoretically could represent the biological effect of TBC1D10C's GAP activity) to modulate such cellular functions. These experiments indicated that functional TBC1D10C's GAP domain is dispensable for macrophage spreading dynamics (Supplementary Fig. 3) but appears necessary for internalization of B. cenocepacia in the phagocytic process in RAW 264.7 cells (Supplementary Fig. 4). To shed light on these differences, we conducted another set of proteomic analyses for cells expressing the TBC1D10C-GAP mutant and Rab35 DN. We only identified Rab35, Histone H2A.V, and Histone H4 as TBC1D10C-R141K-interacting proteins in macrophages (TBC1D10C itself was also identified; 3 different experiments), thus indicating that the functional GAP domain in TBC1D10C is not essential for the TBC1D10C interaction with Rab35 in macrophages but is somehow important for the interaction with other TBC1D10C-interacting proteins. In the case of Rab35 DN, we identified 41 Rab35-S22N-interacting proteins in RAW 264.7 macrophages (Supplementary Table 2) and grouped them functionally (Supplementary Table 3). Two of the Rab35-S22N interactions we identified were particularly interesting: CD14 and Myo1G ( Supplementary Fig. 5). Previously, we reported that Myo1G elimination in lymphocytes affected cell spreading and phagocytosis via a mechanism involving membrane tension, a physical property generated by the interaction between the actin cytoskeleton and the plasma membrane 35 . Furthermore, some of the TBC1D10C-interacting proteins (plectin, actin, tubulin) are known linkers of the plasma membrane to the cytoskeleton 26,36-39 , constituting a mechanism that influences plasma membrane tension. Accordingly, to demonstrate whether membrane tension was influenced by TBC1D10C, we measured this physical characteristic in transfected cells by atomic force microscopy. TBC1D10C overexpression in RAW 264.7 Table 1. List of TBC1D10C-interacting proteins in samples from RAW 264.7 macrophages. Conserved proteins identified between the GFP-tagged control vector and GFP-tagged TBC1D10C were discarded (results from 4 independent experiments). SC: Sequence coverage. TBC1D10C was identified but not included in the www.nature.com/scientificreports/ macrophages was associated with increased membrane tension (Fig. 4b) and was significantly independent of functional GAP activity. Together, our results show that TBC1D10C is a functional link of the cytoskeletonplasma membrane that impacts its tension.
TBC1D10C has a predicted intrinsically disordered region with protein-binding properties at the cytoskeleton level. TBC1D10C has an interacting calcineurin domain and a GAP domain. This GAP domain is responsible for the inactivation of small GTPases such as Ras and Rab35 2,5 , therefore playing a significant role in the GAP and small GTPase interaction 40,41 . To obtain more insights into the interaction between TBC1D10C and TBC1D10C-interacting proteins, we searched for other domains or motives in TBC1D10C by bioinformatics analysis. We identified two intrinsically disordered regions (IDRs) in TBC1D10C (Fig. 5a): one on the C-terminus that coincided with the interacting calcineurin domain (residues 406-446) and another in the N-terminal region (residues 1-34). Interestingly, the (N-terminal)-IDR in TBC1D10C is similar to the recently reported (N-terminal)-IDR in Src kinase, which is essential to form Src dimers as a mechanism for autophosphorylation and phosphorylation of Src substrates, as well as for protein and/or cytoskeletal macromolecule interactions 42 . In addition, similar to the (N-terminal)-IDR in Src kinase, the (N-terminal)-IDR in TBC1D10C has multiple putative phosphorylation sites (Fig. 5b), thus reinforcing the notion of a biological role of the (N-terminal)-IDR in TBC1D10C. Importantly, we found that the disorder profile for TBC1D10C was similar to that for the TBC1D10C GAP mutant (Fig. 5c), but one of the sites in the (N-terminal)-IDR was predicted to be significantly different as a consequence of the GAP mutation (R141K) (Fig. 5d). These results suggest that the functional GAP domain in TBC1D10C is important for the interaction with TBC1D10C-interacting proteins in RAW 264.7 macrophages.

Discussion
TBC1D10C is functionally related to the oncogenic proteins Ras and Rab35 2,5,15 ; however, little is known about the function of this GAP in cellular physiology; in particular, its role in macrophages is not known. Here, we analyzed the participation of TBC1D10C in the cellular physiology of macrophages and identified TBC1D10C as a functional linker for cytoskeleton-plasma membrane reorganization in cellular pathways for cell spreading and phagocytosis, thus modulating the activity of macrophages. www.nature.com/scientificreports/ The first insights into the role of TBC1D10C in macrophage physiology came from important differences observed in cytoskeletal architecture. Specifically, the spreading area (size) of macrophages visualized by F-actin is one of the main components of the cytoskeleton 36 . These differences under TBC1D10C elimination and overexpression correlated with less and more cell spreading in these immune cells, respectively (Fig. 1). We inferred that TBC1D10C positively influenced the spreading ability of macrophages in a manner related to cytoskeletal dynamics. However, how can TBC1D10C positively regulate cell spreading? In this process, cytoskeletal and plasma membrane modifications are essential and impact morphology [43][44][45] . By studying cytoskeleton-plasma membrane dynamics by live cell imaging in macrophages, we found that TBC1D10C KO cells were deficient in cell spreading dynamics. Unlike WT BMDMs, KO BMDMs had defects when dynamically reshaping their morphology, either toward cell spreading or cell retraction (Fig. 1a,b). These functions are in balance and are opposite cellular processes. Specifically, cell spreading is characterized by the formation of filopodia and actin polymerization 45 . KO BMDMs were associated with a significant phenotype consisting of filopodia that abruptly moved toward the cell body (Fig. 2d,e and Supplementary Movies 1-2). Such defects observed in KO cells seemed to affect the balance between cell spreading and cell retraction (Fig. 2b). In addition, it is possible that the phenotype described above may have contributed to an increase in the velocity of cell spreading dynamics in BMDM KO (Fig. 2c), perhaps by adding acceleration (Fig. 2f) derived from abrupt filopodia retraction to the quantified velocity 46 . Furthermore, TBC1D10C overexpression in RAW 264.7 cells was associated with the opposite outcome in the same cellular process; transfected macrophages were more dynamic in modifying their morphology, especially toward cell spreading (Fig. 2g,h). Furthermore, our proteomic analysis allowed us to identify multiple TBC1D10C-interacting proteins ( Table 1) that were functionally grouped by cytoskeleton organization (Fig. 4), including Rab35 and actin, thus highlighting the role of TBC1D10C in cytoskeleton-plasma membrane dynamics. www.nature.com/scientificreports/ Additionally, identification of Rab35 suggested that, as in lymphocytes 5 , this small GTPase and cellular processes in which it participates could also be modified by TBC1D10C's GAP activity in macrophages. A role for Rab35 in cell spreading has been documented; for example, when this function (also referred to as frustrated phagocytosis; in IgG-coated coverslips) was evaluated in BMDMs transfected with Rab35 DN and Rab35 CA (Rab35 constitutively active), transfected BMDMs-Rab35 DN showed decreased spreading ability 20 . Our results showed that Rab35 DN was not associated with increased cell spreading in RAW 264.7 macrophages and that the effect of TBC1D10C in promoting cell spreading was independent of TBC1D10C's GAP activity ( Supplementary  Fig. 3). Additionally, there is an antagonistic operating relationship between Rab35 and Arf6 in many cell types that regulates diverse cellular processes [18][19][20][21]23,[47][48][49][50][51][52] . TBC1D10C is functionally related to Arf6, seemingly as an effector protein (meaning a protein that can interact with active small GTPases, bound to GTP, and promote and/ or participate in cellular pathways) 13,14 , and Arf6 also participates in cell spreading. It was reported that ectopic expression of Arf6 DN (dominant negative) on fibroblasts was able to block spreading, but activation of Arf6 by QS11 (ARF GAP inhibitor used to stimulate Arf6 activation) restored cell spreading and behaved similarly to control cells (on integrin ligands) 53 . Moreover, regardless of functional (TBC1D10C) or inactive (TBC1D10C-R141K) GAP activity, Arf6 was not identified as one of the TBC1D10C-interacting proteins in macrophages, thus suggesting other possible protein targets. In conclusion, TBC1D10C promotes cell spreading in macrophages in a manner closely associated with actin-cytoskeleton dynamics but significantly independent of GAP activity.
On the other hand, in the phagocytosis process, we determined that elimination of TBC1D10C in BMDMs did not affect the phagocytosis of B. cenocepacia (Fig. 3a), perhaps by genetic compensatory mechanisms 54 or by important differences in phagocytosis between BMDMs and RAW 264.7 cells, where BMDMs are described to be more phagocytic in terms of both internalization and phagosome maturation 55 . This agrees with our results showing that BMDMs phagocytosed more bacteria and eliminated the studied microorganisms more efficiently than RAW 264.7 cells (Fig. 3a-d). Conversely, regarding TBC1D10C elimination, the overexpression of TBC1D10C increased the phagocytic activity of RAW 264.7 macrophages for B. cenocepacia (Fig. 3b,e,g), more specifically its internalization (Fig. 3b,g); in this regard, the GAP activity of TBC1D10C seemed to be required ( Supplementary  Fig. 4). Together, these results suggest a specific contribution of TBC1D10C in a cellular pathway involved in B. cenocepacia internalization by RAW 264.7 macrophages. Moreover, cell spreading and phagocytosis are intimately related processes that require cytoskeleton-plasma membrane remodeling as well as a determinant component of membrane tension 28,33,34 . TBC1D10C interacts with proteins associated with cytoskeletal reorganization (Fig. 4a), and specifically, independent of functional GAP activity, TBC1D10C interacts with Rab35 in RAW 264.7 macrophages. We inferred that TBC1D10C could be related to cellular pathways associated with Rab35. One of the Rab35-S22N-interacting proteins we identified is Myo1G (Supplementary Table 2), and we previously reported that membrane tension was affected in Myo1G knockout lymphocytes. Therefore, we analyzed transfected macrophages expressing TBC1D10C to demonstrate whether membrane tension was altered. We determined that TBC1D10C overexpression changes the plasma membrane tension of macrophages (Fig. 4b). This correlation was not impeded by a mutation (R141K) that inactivates the GAP activity of TBC1D10C. Furthermore, it has been reported in RAW 264.7 cells that, derived from actin polymerization during filopodia extension, there is an increase in membrane tension and cell spreading as the phagosome forms 28,33,34 . The above is consistent with our results that relate TBC1D10C with increased membrane tension and increased cell spreading (Figs. 1c, 2h) as well as with increased phagocytosis (Fig. 3b), strongly suggesting the contribution of the TBC1D10C effect on membrane tension to cellular processes where this physical characteristic is a master regulator, including cell spreading and phagocytosis. Additionally, our results suggest that other (than membrane tension) important elements related to TBC1D10C contribute to the phagocytosis of B. cenocepacia, given the significant differences observed in this process when evaluating the functional GAP activity (Supplementary Fig. 4).
For example, the effect of TBC1D10C on phagocytosis of B. cenocepacia seems to occur at the level of initial phases of phagocytosis (macropinocytosis as well as later phases of phagocytosis for bacterial killing were not significantly involved, Supplementary Figs. 2 and 3d). Many works have reported the contribution of Rab35 to phagocytosis; for instance, this small GTPase participates in the early stages of phagosome formation during the uptake of diverse particles in RAW 264.7 cells 20,23,48,55 . Another interesting Rab35-S22N interacting protein we identified is CD14, which was functionally grouped in the innate immune response (Supplementary Fig. 4) and, in turn, is a pattern recognition receptor (PRR) 56 that contributes to phagocytosis of gram-negative bacteria 57,58 . Specifically, it has been reported that part of the immune response against B. cenocepacia is mediated by the association of lipopolysaccharide (LPS) components with CD14 in macrophages 59,60 . These results allow us to infer that the participation of TBC1D10C in phagocytosis of B. cenocepacia could be, at least in part, a cellular pathway associated with bacterial recognition. However, defining the upstream and downstream participation of CD14 in the increased phagocytic activity of B. cenocepacia in macrophages overexpressing TBC1D10C is a future objective.
Finally, we carried out bioinformatics analysis to gain insights into the interaction of TBC1D10C and TBC1D10C-interacting proteins, which could provide more information about the role of TBC1D10C in cell spreading and phagocytosis. We identified two intrinsically disordered regions (IDRs) in TBC1D10C: one TBC1D10C-IDR was identified on the C-terminus and coincided with the interacting calcineurin domain (carboxy terminus; residues 406-446), and the other was an unreported (N-terminal)-IDR (residues 1-34) (Fig. 5a). It has been reported that IDRs play important roles in modulating membrane properties as well as cytoskeleton-related functions [61][62][63][64] . In addition, a biological role for an IDR in Src kinase was recently revealed. Specifically, this IDR was located in the amino terminus (N-terminal), and it was described to be essential for Src dimerization as a mechanism of regulation. Additionally, it was suggested that this IDR could contribute, at the macromolecular level in the plasma membrane, as a signaling hub for protein-protein interactions 42 . Interestingly, the (N-terminal)-IDR in TBC1D10C is similar to the (N-terminal)-IDR in Src (Fig. 5a). In addition, as the (N-terminal)-IDR in Src kinase, the (N-terminal)-IDR in TBC1D10C has putative phosphorylation sites  . 5b), suggesting a possible biological role for the (N-terminal)-IDR in TBC1D10C 65,66 . Furthermore, the disorder profiles for TBC1D10C and the TBC1D10C GAP mutant were similar (Fig. 5c), but in the (N-terminal)-IDR, a predicted protein-binding site was shown to be significantly different as a consequence of the R141K mutation in TBC1D10C (Fig. 5d); this is possibly related to the difference in the number of TBC1D10C-interacting proteins before the mutation. This result may be due to changes in the complex conformation between the GAP protein and the small GTPase 40,41 . Together, it is possible that the (N-terminal)-IDR in TBC1D10C plays a biological role in the interaction between TBC1D10C and TBC1D10C-interacting proteins and contributes to the functional link between TBC1D10C and the cytoskeletal membrane during spreading and phagocytosis. To summarize, TBC1D10C participates in macrophage physiology through a functional link with the cytoskeletonplasma membrane that impacts its tension and reorganization for cell spreading and phagocytosis (Fig. 6).

Materials and methods
Mice. Mice were maintained in pathogen-free conditions at UPEAL (Unit for Production and Experimentation of Animals of Laboratory) from CINVESTAV. Their manipulation was performed according to the standard and operating protocols approved by the Animal Care Use Committee, Bioethics and Research of CINVESTAV (Protocol 145-15). The general recommendations in the ARRIVE guidelines 2.0 (https:// arriv eguid elines. org/) were followed in our study. Our sample size was small. We did not use any method to generate the randomization sequence. All methods were performed in accordance with the relevant guidelines and regulations. We used an equal number of animals in the WT and KO groups for this study. We used TBC1D10C floxed and knockout (KO) mice generated on the C57BL/6 background. To obtain TBC1D10C KO mice, TBC1D10C floxed mice were crossed with β actin-cre mice (Jackson Laboratories, Bar Harbor, ME, USA) to generate full conventional KO. For genotyping, tail DNA was extracted by the HotShot protocol, and the TBC1D10C gene was amplified by PCR using specific primers: floxed mice-5' loxP side: TBC lxpF, ccttcccctaaagctgaaca and TBC lxpR, tttggctgtcccctaaaaca; 3' side: TBC FrtgtF, tgacagagtgtcaggagagca and TBC FrtgtR, gcacattaccagcccaaact. KO mice-TBC lxpF, ccttcccctaaagctgaaca and TBC FrtgtR, gcacattaccagcccaaact (to detect WT allele band, we used the primer TBC lxpR, tttggctgtcccctaaaaca).   ; S22N), Myo1g WT, and control empty vector (henceforth, control vector) (pDest732, N-terminal GFP tag) 5 using Lipofectamine. Transfected cells with similar GFP expression were sorted (by FL1 mean fluorescence intensity) and cultured in RPMI 1640 supplemented medium with G418 at 600 μg/mL (atmosphere, oxygen, 95%; CO 2 , 5%; 37 °C). Ex vivo: Bone marrow-derived macrophages were obtained from bone marrow from C57BL/6 KO mice treated with medium for differentiation (V/V, supernatant medium of L929 cells, 30%; RPMI 1640 medium supplemented with 10% FBS, 70%), as previously described 67 . For the gentamicin protection assay, we used Burkholderia cenocepacia MH1K 29 , kindly donated by Dr. Miguel A. Valvano (Queens University, Belfast, UK), which is susceptible to gentamicin (this antibiotic does not penetrate eukaryotic cells). To infect RAW 264.7 cells (3.5 × 10 5 ; seeded the day prior to infection), we used 12-well plates and 6-well plates for BMDMs (2 × 10 6 ; differentiated as earlier). Macrophages were infected with this bacterium (cultured in LB broth medium with constant shaking; 37 °C, 16 h/180 r.p.m.) at a multiplicity of infection (MOI) of 50 for 2 h (atmosphere, oxygen, 95%; CO 2 , 5%; 37 °C) in RPMI medium without antibiotics (800 µL or 3 mL for 12-and 6-well plates, respectively). Infection was synchronized by centrifugation (1200 r.p.m/2 min). After 2 h of incubation, the cells were washed (twice with sterile PBS; 1 ml and 2 ml according to the well plate), and the medium was changed to RPMI medium plus gentamicin (60 µg/mL) for 1.5 h (to kill extracellular bacteria; same atmosphere). Finally, cells were washed once (same as above) and lysed using sterile water (  Cell spreading and cell spreading dynamics (aspect ratio dynamics). Cell spreading: RAW 264.7-transfected cells (1 × 10 6 ) were seeded onto coverslips in 6-well plates the day before the assay. For BMDMs, we seeded bone marrow cells (1.5 × 10 6 ) for differentiation as previously described. For both types of cells, the medium was discarded, and monolayers were washed (once with PBS). Macrophages were fixed (paraformaldehyde 4%; 10 min), permeabilized (PBS-Tween 0.05%; 5 min) and F-actin stained using phalloidin (ready to use ActinRed™ 555 ReadyProbes™ Reagent; and Alexa phalloidin 488 at dil. 1:625 in PBS) following the manufacturer's instructions. Coverslips were mounted onto slices to acquire confocal images (using a LEICA TCS SP8X confocal microscope; xyz planes), of which we generated projections (sum of slices www.nature.com/scientificreports/ and the linear force vs. displacement was recorded. The vertical probe displacement was 300 nm at a vertical speed of 10 µm/s. The average of the 10 curves was used to obtain the membrane stiffness for each cell. Data adjustment was performed with a homemade MATLAB routine. The elastic model used to estimate the elastic force constant for cell membranes was the mechanical model for two springs in series given by Hooke's law:

Reagents and antibodies. RPMI
where F is the probe applied force, km is the membrane elastic constant, kc is the cantilever constant, z is the vertical probe displacement, and z0 is the point of contact between the cell membrane and probe tip. All measurements were performed at the highest point of the cell, that is, on the nucleus of the cell.
Immunoprecipitation, gel digestion, and mass spectrometry. Immunoprecipitation: Transfected RAW 264.7 cells (10 × 10 7 ; the day prior to the assay; control vector, TBC1D10C, TBC1D10C-R141K, and Rab35 S22N) were cultured in round plates (9 cm) with RPMI medium (9 mL) supplemented with G418 and incubated. On the day of the assay, the medium was discarded, and a monolayer of cells was washed once (PBS; 5 mL). Cells were lysed by adding 1 mL of lysis buffer (150 mM NaCl, 1% Triton X-100, 50 mM Tris HCl, pH 8/protease inhibitors, 1.5x) and detached using a cell scraper. The supernatant was collected and incubated for 30 min on ice (2 mL microtubes), and proteins were obtained from that supernatant by centrifugation (10,000 r.p.m./10 min/4 °C). Immunoprecipitation was carried out using anti-GFP coupled to magnetic beads (MACS, 130-091-125) and microcolumns (MACS, 130-042-701) coupled to a magnetic separator (MACS, 130-042-303). From that procedure, we obtained an eluate or immunoprecipitated proteins (proteins in Laemmli buffer), which were later separated by SDS-PAGE. In gel digestion, bands were excised and cut into small pieces (2 mm) and washed with 1:1 (v/v) 50 mM NH 4 HCO 3 (ammonium bicarbonate) and ACN (acetonitrile) to cover pieces of gel. Then, the proteins were reduced (10 mM DTT, dithiothreitol, in 50 mM NH 4  Bioinformatics. The protein-protein interaction (PPI) network for TBC1D10C-interacting proteins was built using STRING (Search Tool for the Retrieval of Interacting Genes/Proteins; http:// www. string-db. org).
Statistical analysis. Assays were performed with at least 3 independent experiments unless otherwise indicated. We used the D' Agostino-Pearson omnibus normality test to analyze parametric and nonparametric distributions of data. The statistical analyses used were unpaired t-tests or Mann-Whitney tests (two-tailed; statistical significance P < 0.05) for two groups of data with parametric or nonparametric distributions, respectively. In the case of multiple groups, we used Kruskal-Wallis test and ordinary one-way ANOVA followed by Dunn's multiple comparisons test and Tukey's multiple comparisons test, respectively.