Mitochondria-endoplasmic reticulum contacts (MERCs) play an essential role in multiple cell physiological processes. Although Mfn2 was the first protein implicated in the formation of MERCs, there is debate as to whether it acts as a tether or antagonizer, largely based on in vitro studies. To understand the role of Mfn2 in MERCs in vivo, we characterized ultrastructural and biochemical changes of MERCs in pyramidal neurons of hippocampus in Mfn2 conditional knockout mice and in Mfn2 overexpressing mice, and found that Mfn2 ablation caused reduced close contacts, whereas Mfn2 overexpression caused increased close contacts between the endoplasmic reticulum (ER) and mitochondria in vivo. Functional studies on SH-SY5Y cells with Mfn2 knockout or overexpression demonstrating similar biochemical changes found that mitochondrial calcium uptake along with IP3R3-Grp75 interaction was decreased in Mfn2 knockout cells but increased in Mfn2 overexpressing cells. Lastly, we found Mfn2 knockout decreased and Mfn2 overexpression increased the interaction between the ER-mitochondria tethering pair of VAPB-PTPIP51. In conclusion, our study supports the notion that Mfn2 plays a critical role in ER-mitochondrial tethering and the formation of close contacts in neuronal cells in vivo.
A significant proportion of the mitochondrial surface is closely apposed to the endoplasmic reticulum (ER) membrane and forms a specialized structure called mitochondria-ER contact (MERC), which provides a stable platform to carry out many important physiological functions, including the regulation of intracellular calcium homeostasis, phospholipid synthesis, mitochondrial dynamics, autophagy, inflammasome activation and apoptosis (Hailey et al., 2010; Hamasaki et al., 2013; Muñoz et al., 2013; Rowland and Voeltz, 2012; Zhou et al., 2011). Different types of MERCs with varying thickness (i.e. the distance between the mitochondrial outer membrane and the ER surface) ranging from narrower smooth ER-mitochondria contacts (∼10-50 nm) to wider rough ER-mitochondria contacts (∼25-80 nm) (Wang et al., 2015) have been described previously (Giacomello and Pellegrini, 2016). The thickness of MERCs is an important factor that defines its biological function as the 50 nm MERC were found to contribute to autophagy (Hamasaki et al., 2013), and average thickness coordinating calcium transfer was ∼15 nm and optimal thickness for phospholipid synthesis was ∼10 nm (Csordás et al., 2010; Giacomello and Pellegrini, 2016). MERCs appear to be dynamic structures, as evidenced by the induced changes in MERC thickness in cells exposed to apoptotic or metabolic stimuli (Bravo et al., 2011; Csordás et al., 2006; Sood et al., 2014), suggestive of adaptation to metabolic transitions.
Many proteins present in the mitochondria-associated membrane (MAM) of the MERCs are essential for the formation and function of MERCs. Pairs of integral membrane proteins localized on the mitochondrial outer membrane and ER tether the two organelles together to form MERCs through physical interactions, and several pairs of these tethering proteins were identified, including the VAPB-PTPIP5 pair and the BAP31-Fis1 pair (De Vos et al., 2012; Iwasawa et al., 2011; Szabadkai et al., 2006). Mfn2, a large GTPase involved in mitochondrial fusion, was also enriched in MAM (Merkwirth and Langer, 2008). Initial studies found MAM-located Mfn2 tethers the ER to mitochondria through in trans interaction with mitochondrial mitofusins (Mfn1 or Mfn2) in mouse embryonic fibroblasts (MEFs), which mediated efficient ER-mitochondrial calcium transfer (de Brito and Scorrano, 2008). This was further strengthened by later electron microscopy studies from the same group (Naon et al., 2016). Phospholipid synthesis and transport was also impaired in Mfn2-deficient MEFs (Area-Gomez et al., 2012). Studies from another group demonstrated that mitochondrial ubiquitin ligase MITOL ubiquitylates Mfn2 and further revealed that MITOL regulates ER tethering to mitochondria by activating Mfn2 via K192 ubiquitylation (Sugiura et al., 2013). A critical role for Mfn2 in ER-mitochondrial tethering was confirmed by multiple groups, and ER-Mfn2 and mitochondria-Mfn1/2 are thus widely accepted as a major tethering pair of MERC in different tissues (Alford et al., 2012; Chen et al., 2012; Göbel et al., 2020; Li et al., 2015; Schneeberger et al., 2013). However, this view was initially challenged by an electron microscopy study in MEFs (Cosson et al., 2012) followed by additional biochemical and functional studies from several other groups (Cosson et al., 2012; Filadi et al., 2015, 2016; Puri et al., 2019; Cieri et al., 2018). For example, Filadi et al. (2015) reported that Mfn2 ablation increased close contacts along with elevated calcium flow from the ER to the mitochondria but did not alter loose contacts in MEF cells, which thus supports an antagonizing role for Mfn2 in the formation of MERCs. Obviously, further studies are needed to fully elucidate the role of Mfn2 in MERCs.
Previous studies of Mfn2 in MAM mainly used cultured Mfn2-deficient MEF cells, and not much is known in vivo. Alterations in Mfn2 have been associated with various neurological diseases (Lee et al., 2012; Stuppia et al., 2015; Wang et al., 2009), many of which also involve disturbed MERCs (Liu and Zhu, 2017; Wang et al., 2020). Mfn2 ablation leads to neuronal loss in mouse brain (Han et al., 2020; Jiang et al., 2018), whereas Mfn2 overexpression protects neurons against neurotoxin-induced neuronal death in vivo (Zhao et al., 2017), although the effects of manipulated Mfn2 expression on MERCs were not elucidated. To better understand the role of Mfn2 in MERCs in neurons in vivo, we characterized MERCs in the pyramidal neurons of hippocampi in Mfn2 conditional knockout (cKO) mice and in Mfn2 overexpressing mice, and further explored the functional consequence of MERC changes in SH-SY5Y cells with manipulated Mfn2 expression.
Mfn2 ablation differentially impacted close contacts and loose contacts between the ER and mitochondria in vivo
We previously characterized Mfn2 cKO mice (CaMKII-Cre+/−/Mfn2loxP/loxP) where Mfn2 was knocked out in neurons in the hippocampus and cortex (Jiang et al., 2018). To understand the role of Mfn2 in mediating ER-mitochondria interaction, we performed a quantitative electron microscopy analysis of ER-mitochondria association in pyramidal neurons of the CA1 region of hippocampi from Mfn2 cKO mice by determining the number, length, thickness and proportion of the mitochondrial surface that was apposed to the ER with a distance less than 80 nm (Fig. 1A,B). Mfn2 cKO mice began to demonstrate subtle mitochondrial alterations without any neurodegeneration at 8 weeks of age and displayed significant mitochondrial deficits and extensive neuronal death at 28 weeks (Jiang et al., 2018). These parameters were thus measured in Mfn2 cKO mice at both of these ages. As previously reported, ER-mitochondria association was categorized as close contacts (≤15 nm) and loose contacts (15-80 nm). Approximately 2.63±0.43% (mean±s.e.m.) of the mitochondrial surface formed close contacts with the ER in 4- to 8-week-old wild-type control mice (Fig. 1C). There was a significant decrease of close contact coverage to 1.51±0.19% in neurons of 4- to 8-week-old Mfn2 cKO mice (Fig. 1C). The close contact coverage was similarly decreased in 28-week-old Mfn2 cKO mice compared to age-matched wild-type control mice, although this difference was not significant (Fig. 1C). The number of close contacts per mitochondria was significantly reduced in 4- to 8-week-old Mfn2 cKO mice and further reduced in 28-week-old Mfn2 cKO mice (Fig. 1D). The average length of close contacts showed a decreasing trend in 4- to 8-week-old Mfn2 cKO mice but increased in 28-week-old Mfn2 cKO mice compared with age-matched wild-type control mice (Fig. 1E). The average distance of close contacts between ER and mitochondria also increased in 28-week-old Mfn2 cKO mice (Fig. 1F). Interestingly, differential effects on loose contacts were noted in Mfn2 cKO mice: the percentage of mitochondrial surface forming loose contacts increased from 2.25±0.34% in 4- to 8-week-old wild-type mice to 3.68±0.40% in age-matched Mfn2 cKO mice but significantly decreased to 0.67±0.24% in the 28-week-old cKO mice compared to 2.29±0.34% in 28-week-old wild-type mice (Fig. 1C). Similarly, the number of loose contacts per mitochondria increased in 4-week-old Mfn2 cKO mice but significantly decreased in 28-week-old Mfn2 cKO mice compared with their age-matched wild-type control mice (Fig. 1D). The average length of loose contacts was not changed in 4- to 8-week-old Mfn2 cKO mice but significantly increased in 28-week-old Mfn2 cKO mice compared with their age-matched wild-type control mice (Fig. 1E). There was no change in the average distance of loose contacts between the ER and mitochondria in Mfn2 cKO mice (Fig. 1F).
As some of these ER-mitochondrial contact parameters may be affected by mitochondrial morphology, mitochondrial parameters were also measured. As previously reported (Jiang et al., 2018), mitochondria become round and slightly larger but with a reduced perimeter in 4- to 8-week-old Mfn2 cKO mice (Fig. 1B, left, Fig. 1G,H). However, mitochondria became large and round with a significantly increased perimeter and area in 28-week-old Mfn2 cKO mice (Fig. 1B, right, Fig. 1G,H), likely representing a swollen phenotype. Considering the significant changes in mitochondria size and perimeter in Mfn2 cKO mice, to control for the effects of changes in mitochondrial morphology on the ER-mitochondrial interaction, an ER-mitochondria contact coefficient (ERMICC) that computes the interaction length and the perimeter of the mitochondria involved in the interaction was analyzed as described previously (Naon et al., 2016). ERMICC of close contacts was significantly reduced in 4- to 8-week-old Mfn2 cKO mice compared to age-matched wild-type mice (Fig. 1I). ERMICC was similarly decreased in 28-week-old Mfn2 cKO mice compared to age-matched wild-type control mice, although this difference was significant (Fig. 1I). ERMICC of loose contacts was slightly increased in 4- to 8-week-old Mfn2 cKO mice but significantly reduced in 28-week-old Mfn2 cKO mice compared to their age-matched wild-type control mice (Fig. 1I). Overall, ERMICC results were consistent with the percentage of mitochondrial surface coverage in these mice (Fig. 1C).
The ER morphology was also analyzed in pyramidal neurons from the brains of Mfn2 cKO mice and their age-matched controls to determine whether changes in ER morphology may impact the ER-mitochondrial contacts. No obvious changes in ER morphology were noted. Indeed, quantification analysis revealed no significant changes in the number, average width of individual cisternae or cell volume coverage by ER in 4- to 8-week-old Mfn2 cKO mice compared to age-matched wild-type control mice (Fig. S1). However, there was a significant reduction in the cell volume coverage by ER and a decreasing trend in the number of ER (P=0.08) in 28-week-old Mfn2 cKO mice, although the average width remained unchanged. These data indicate that ER integrity is not impacted in 4- to 8-week-old Mfn2 cKO mice but becomes compromised in 28-week-old Mfn2 cKO mice.
Mfn2 overexpression increased close contacts but did not change loose contacts between the ER and mitochondria
A similar quantitative electron microscopy analysis of ER-mitochondria association in pyramidal neurons of the CA1 region of hippocampi from 6- to 7-month-old mice overexpressing Mfn2 (Zhao et al., 2017) was also performed (Fig. 2A). A significant increase in the percentage of mitochondrial surface coverage (Fig. 2B) and ERMICC (Fig. 2H) of close contacts was noted in Mfn2 overexpressing mice. The number of close contacts also increased in Mfn2 overexpressing mice (Fig. 2C). There was an increasing trend in the average length of close contacts but this difference was not significant (Fig. 2D). No changes in the thickness of close contacts were found (Fig. 2E). As for loose contacts, only the average length was significantly increased but no changes in other parameters, including ERMICC, were found in Mfn2 overexpressing mice (Fig. 2B-E,H). Mfn2 overexpression caused enhanced mitochondrial fusion, and mitochondrial perimeter was significantly increased in Mfn2 overexpressing mice (Fig. 2G). Mitochondrial size was also increased in Mfn2 overexpressing mice, although this difference was not significant (Fig. 2F).
Mfn2 expression modulated the expression of MAM-related proteins involved in calcium signaling
Close MERCs regulate essential cellular functions, such as calcium transfer between the ER and mitochondria (Csordás et al., 2010; Vance, 2014). To corroborate the role of Mfn2 in modulating close contact, the expression of MAM-related proteins involved in calcium signaling were determined by western blotting in hippocampal tissues from either Mfn2 cKO mice or Mfn2 overexpressing mice (Fig. 3A-D). In 8-week-old Mfn2 cKO mice, Mfn2 levels were significantly reduced. The expression of many MAM proteins were either significantly reduced [i.e. Sig1R (also known as Sigmar1), IP3R3 (also known as Itpr3), and VAPB] or displayed a decreasing trend (i.e. VDAC1, DJ-1 (also known as Park7), ATG5 and FACL4) in Mfn2 cKO mice (Fig. 3A,B). In the Mfn2 overexpressing mice, the expression of Mfn2 was increased about 4-fold compared to wild-type control mice (Fig. 3C,D). Interestingly, the expression of many of these same MAM proteins was significantly increased (i.e. IP3R3, GRP75, DJ-1 and VAPB) or displayed an increasing trend (i.e. Sig1R, FACL4) (Fig. 3C,D).
The expression of these MAM-related proteins in the MAM fraction was also determined (Fig. 3E-H). The quality of MAM preparation from mouse hippocampi was confirmed by the enrichment of known MAM proteins, such as VAPB and CNX, and the lack of contamination of mitochondria (i.e. COX IV) or cytosolic proteins (i.e. GAPDH) (Fig. S2). In Mfn2 cKO mice, the protein levels of IP3R3, DJ-1 and VAPB were significantly decreased, and the levels of Sig1R showed a decreasing trend (P=0.06) in MAM fractions (Fig. 3E,F). In contrast, the expression levels of Sig1R, VDAC1, Grp75 and DJ-1 were significantly increased in the MAM fractions of Mfn2 overexpressing mice (Fig. 3G,H).
We also established clonal lines of human SH-SY5Y neuroblastoma cell lines with stable knockout or overexpression of Mfn2. Although the expression of VDAC1, IP3R3, DJ-1 or VAPB did not change in the total lysates of Mfn2 knockout cells (Fig. 4A,B), they were significantly decreased in the MAM fractions (Fig. 4E,F). On the other hand, the expression levels of all of these proteins were significantly increased in the MAM fraction from Mfn2 overexpressing cells (Fig. 4G,H) and some were also increased in the total lysates (Fig. 4C,D).
Mfn2 expression modulated mitochondrial Ca2+ uptake and impacted the IP3R3-VDAC1 complex in neuronal cells
We then evaluated the functional impacts of Mfn2 expression on ER-mitochondrial contacts by determining the ER-mitochondria calcium transfer upon histamine treatment, which led to an increase in cytosolic followed by mitochondrial Ca2+ levels through mitochondrial calcium uptake. To this end, SH-SY5Y cells were transfected with a mitochondrial matrix-targeted Ca2+ fluorescent sensor CEPIA4, and mitochondrial Ca2+ uptake was measured following histamine treatment in Ca2+-free Hank's balanced salt solution (HBSS) buffer with 20 mM HEPES (Fig. 5A). As a control, these cells were also transfected with a cytosol-targeted version of the Ca2+ indicator Cyto-GCaMP3 to measure cytosolic calcium uptake following histamine treatment in parallel experiments (Fig. 5B). Under the condition that a similar cytosolic calcium peak was induced (i.e. 1 µM histamine treatment) between Mfn2 knockout and negative control cells (Fig. 5B), Mfn2 knockout cells showed significantly slower and lower mitochondrial Ca2+ uptake following histamine treatment compared to the negative controls (Fig. 5A). To exclude the possibility that there might be changes in intrinsic mitochondrial calcium uptake ability in Mfn2 knockout cells, SH-SY5Y cells were treated with 2 μM ionomycin, a widely used calcium ionophore, and 15 mM Ca2+, which caused a pronounced increase in both cytosolic and mitochondrial Ca2+ levels, as expected (Fig. S3). Importantly, no difference in mitochondrial calcium uptake between Mfn2 knockout cells and the negative controls cells were found under this condition (Fig. 5C), which thus supports the notion that decreased mitochondrial calcium uptake of Mfn2 knockout cells in response to histamine-induced ER calcium release is specifically caused by decreased ER-mitochondria tethering but not by changes in intrinsic mitochondrial calcium uptake ability. Similarly, Mfn2 overexpression did not impact intrinsic mitochondrial calcium uptake activity, as evidenced by similar mitochondrial calcium uptake upon treatment of ionomycin and Ca2+ (Fig. 5F). However, under the condition that similar cytosolic calcium uptake and cytosolic calcium peak were induced (i.e. 50 µM histamine treatment) between Mfn2 overexpression and the controls cells expressing empty vector (Fig. 5E), Mfn2 overexpressing cells showed significantly increased mitochondrial Ca2+ levels by histamine treatment (Fig. 5D), indicating the mitochondria Ca2+ uptake is increased by Mfn2 overexpression. Collectively, these data demonstrate that Mfn2 expression modulates ER-mitochondrial calcium transfer.
The IP3R-Grp75-VDAC1 complex in MAM plays a crucial role in ER-mitochondria Ca2+ transfer (Szabadkai et al., 2006). We therefore investigated whether changes in Mfn2 expression impacted the expression and assembly of the IP3R-Grp75-VDAC1 complex. We first examined the interaction between IP3R and VDAC1 by proximity ligation assay (PLA) in SH-SY5Y cells (Fig. 5G). The IP3R-VDAC1 PLA signals were significantly decreased in Mfn2 knockout cells compared to negative control (NC) cells (Fig. 5G,H). In contrast, the IP3R-VDAC1 PLA signals were significantly increased in Mfn2 OE cells compared to control cells expressing empty vector (Fig. 5G,H). To corroborate this finding, a co-immunoprecipitation assay was performed. Consistently, the amounts of Grp75 co-immunoprecipitated by IP3R3 antibody were significantly reduced in Mfn2 knockout cells compared to NC cells (Fig. 5I), and Grp75 levels co-immunoprecipitated by IP3R3 were significantly increased in Mfn2 overexpressing cells (Fig. 5J).
Mfn2 expression impacted VAPB-PTPIP51 interaction
VAPB and PTPIP51 (also known as Rmdn3) were identified as a pair of tethering proteins between the ER and mitochondria (Stoica et al., 2014). Because manipulated Mfn2 expression changed the expression of VAPB in MAM, VAPB-PTPIP51 interactions were investigated in Mfn2 knockout and overexpressing cells by co-immunoprecipitation (Fig. 6). Western blot revealed significantly reduced levels of PTPIP51 in the VAPB immunoprecipitates of the Mfn2 knockout cells (Fig. 6A). Reversed co-immunoprecipitation found significantly reduced levels of VAPB in the PTPIP51 immunoprecipitates of the Mfn2 knockout cells (Fig. 6B). Consistently, VAPB-PTPIP51 interaction was increased in Mfn2 overexpressing cells (Fig. 6C,D). However, no Mfn2 immunoreactivity could be detected in the VAPB immunoprecipitates, suggesting that there was no interaction between VAPB and Mfn2 (Fig. S4).
In this study, we investigated the role of Mfn2 in the regulation of mitochondria-ER association in neuronal cells both in vitro and in vivo. Based on quantitative electron microscopy analysis, we found that Mfn2 ablation caused reduced close contacts but increased loose contacts between mitochondria and the ER in pyramidal neurons of the CA1 region of hippocampi from Mfn2 cKO mice at 4-8 weeks of age. On the other hand, Mfn2 overexpression caused increased close contacts but not significant changes in the loose contacts in vivo. Corroborating these electron microscopy findings, biochemical analysis revealed that the expression of MAM-related proteins involved in calcium transfer was reduced in brain homogenates and MAM fractions from the hippocampi of Mfn2 cKO mice but was increased in Mfn2 overexpressing mice. Similar changes in the expression of these MAM proteins were also observed in the stable SH-SY5Y cells with Mfn2 knockout or overexpression. Indeed, mitochondrial calcium uptake after histamine stimulation was decreased in Mfn2 knockout cells but increased in Mfn2 overexpressing cells. Consistently, we found that the interaction between IP3R3 and Grp75 was reduced in Mfn2 knockout cells but increased in Mfn2 overexpressing cells. Lastly, we found that Mfn2 knockout decreased and Mfn2 overexpression increased the interaction between the ER-mitochondria tethering pair of VAPB-PTPIP51.
Although Mfn2 is widely accepted as an essential molecule involved in ER-mitochondria juxtaposition, it is hotly debated whether Mfn2 acts as a physical tether or antagonizer between the two organelles (Cosson et al., 2012; de Brito and Scorrano, 2008, 2009; Filadi et al., 2015; McLelland et al., 2018; Naon et al., 2016). It should be noted that previous studies were largely based on analysis in MEF cells in vitro (Cosson et al., 2012; de Brito and Scorrano, 2008; Filadi et al., 2015; Naon et al., 2016) in which mitochondrial morphology and MERC structures may be sensitive to the changes in culture conditions due to their dynamic nature. In this study, we employed a thorough ultrastructural and biochemical analysis of ER-mitochondrial contacts in hippocampi of Mfn2 cKO mouse and Mfn2 overexpressing mouse to characterize the role of Mfn2 in ER-mitochondria interaction in vivo. We have measured MERC parameters in Mfn2 cKO mice at both 4-8 and 28 weeks of age and found the impact of Mfn2 knockout on some of these factors were completely opposite between 4-8 weeks and 28 weeks. For example, the number, ERMICC and percentage of mitochondria surface coverage of loose contacts were increased in 4- to 8-week-old Mfn2 cKO mice but significantly decreased in 28-week-old Mfn2 cKO mice. Additionally, the average MERC length of close contacts was decreased in 4- to 8-week-old Mfn2 cKO mice but increased in 28-week-old Mfn2 cKO mice. Given that mitochondria became significantly swollen and damaged at 28 weeks, paralleling extensive neuronal loss (Jiang et al., 2018), changes in the MERCs in 28-week-old Mfn2 cKO mice likely reflected degenerative changes rather than the direct impact of Mfn2 knockout. Therefore, we focused on the analysis of Mfn2 cKO mice at 4-8 weeks. The quantitative ultrastructural analysis of hippocampal neurons revealed that the number, ERMICC and percentage of mitochondria surface coverage of close contacts were decreased but these parameters of loose contact were increased in Mfn2 cKO mice at 4-8 weeks of age, demonstrating that Mfn2 deficiency led to reduced close contact (≤15 nm) but increased loose contacts (15∼80 nm) in hippocampal neurons in vivo. On the other hand, we also found that Mfn2 overexpression caused increased number, ERMICC and percentage of mitochondrial surface coverage of close contacts but not much change in loose contacts in hippocampal neurons in the brain, suggesting that Mfn2 overexpression increased close contacts in vivo. Close contacts support essential functions, such as calcium transfer and phospholipid transfer between the ER and mitochondria (Csordás et al., 2010; Vance, 2014). Consistent with changes in the close contacts, the expression of MAM proteins involved in calcium signaling, including Sig1R, IP3R3 and DJ-1, was decreased in the total lysates and in the MAM preparation from hippocampal tissues of Mfn2 cKO mice but increased in the total lysates and/or MAM preparation from Mfn2 overexpressing mice. Together, these data provided evidence in vivo to support the notion that Mfn2 plays an essential role in ER-mitochondrial tethering and in the formation of close MERC contacts. Indeed, one recent study found that Mfn2 deficiency caused a reduced transfer of PS from ER to mitochondria, a function supported by close contact, and liver-specific ablation of Mfn2 in mice impaired phospholipid metabolism and provoked triglyceride accumulation, fibrosis, and liver cancer (Hernández-Alvarez et al., 2019), which appeared to also support an essential role for Mfn2 in the regulation of close contacts in liver in vivo.
The impact of Mfn2 expression on MERC function was directly demonstrated by changes in mitochondrial uptake of calcium in SH-SY5Y neuronal cells with manipulated Mfn2 expression: Mfn2 knockout reduced mitochondrial calcium uptake upon histamine-induced ER calcium release, whereas Mfn2 overexpression increased it. Consistently, these SH-SY5Y cells recapitulated many of the biochemical changes observed in vivo in which Mfn2 knockout reduced levels of Sig 1R, VDAC1, IP3R3, DJ-1 and VAPB in the MAM fraction, whereas Mfn2 overexpression increased them. Furthermore, we found that the interaction between IP3R3-VDAC1 was decreased by Mfn2 knockout but increased by Mfn2 overexpression, as evidenced by a PLA, and co-immunoprecipitation analysis also confirmed reduced IP3R3-Grp75 interaction, which likely specifically underlies Mfn2-induced changes in mitochondrial calcium uptake as IP3R3-Grp75-VDAC1 complex plays a critical role in ER-mitochondrial calcium transfer (Szabadkai et al., 2006). Importantly, the IP3R3-VDAC1 complex was reduced in Mfn2 knockout SH-SY5Y cells, as well as in the hippocampi of Mfn2 cKO mice, demonstrating similar biochemical changes caused by Mfn2 knockout both in vitro and in vivo. Overall, these in vitro studies corroborated the in vivo findings and suggest that the biochemical and structural changes we observed in vivo in Mfn2 cKO mice or Mfn2 overexpressing mice likely lead to functional changes, such as calcium homeostasis between the ER and mitochondria in neurons in vivo.
Our biochemical analysis demonstrated that Mfn2 knockout caused reduced VAPB levels in the total lysates and in the MAM both in vitro and in vivo, whereas Mfn2 overexpression increased them. More importantly, co-immunoprecipitation analysis between the tethering pair of VAPB and PTPIP51 revealed decreased interaction in Mfn2 knockout cells but increased interaction in Mfn2 overexpressing cells, paralleling changes in the close contacts between the ER and mitochondria. These data suggest that changes in VAPB and VAPB-PTPIP51 interaction may contribute to Mfn2-induced changes in close contacts. How Mfn2 affects VAPB expression and VAPB-PTPIP51 interaction warrants further investigation.
The regulation and physiological significance of loose contacts have not been well defined. It has been shown that close and loose contacts respond differently to the same stimuli, implicating different regulation mechanisms (Cieri et al., 2018). This is further supported by the observation that co-expression of the VAPB-PTPIP51 tethering pair in HeLa cells resulted in increased close contacts but decreased loose contacts (Cieri et al., 2018). Our electron microscopy results demonstrated that Mfn2 knockout led to increased loose contacts yet Mfn2 overexpression had no effect on loose contacts in vivo, depicting a complicated role for Mfn2 in the modulation of loose contacts. Loose contacts were suggested to be involved in the regulation of apoptosis, and Mfn2 ablation resulted in extensive neuronal loss in vivo (Han et al., 2020; Jiang et al., 2018); however, whether increased loose contacts contribute to neuronal death in vivo needs to be further investigated.
It is increasingly recognized that MERCs, as critical signaling platforms that carry out many important physiological functions critical for cellular homeostasis and senescence, play significant roles in human health and disease (Janikiewicz et al., 2018). Genetic defects in genes encoding MAM proteins, such as Mfn2, VAPB and Sig1R (Sigmar1), cause human neurological diseases (Bernard-Marissal et al., 2015; Nishimura et al., 2004), and disturbances in MERCs are involved in many neurodegenerative diseases studied, such as Alzheimer's disease, Parkinson's disease and related diseases (Liu and Zhu, 2017; Wang et al., 2020). Although whether and how changes in MERCs are causally involved in the pathogenesis of these diseases remains to be determined, disturbance in MERCs provides a connection for the various seemingly disparate features of these neurodegenerative diseases. Therefore, better understanding of MERC regulation will be important in the study of these diseases.
In summary, we provide both electron microscopy and biochemical evidence that Mfn2 knockout causes reduced close contacts, whereas Mfn2 overexpression causes increased close contacts between the ER and mitochondria in vivo, which appears to support the notion that Mfn2 acts as an ER-mitochondria tethering protein and regulates close contacts related to Ca2+ transfer in neuronal cells in vivo.
MATERIALS AND METHODS
The wild-type and transgenic Mfn2 cKO mice and Mfn2 overexpressing mice generated by our lab, as described previously, were used in this study (Jiang et al., 2018; Zhao et al., 2017). The CAMKCre gene was introduced into the Mfn2 cKO mice to knockout Mfn2 in the neurons from the forebrain and hippocampus. Human Mfn2 overexpression in neurons was achieved under the control of the Thy-1.2 promoter in Mfn2 overexpressing mice. The wild-type mice with the same background of Mfn2 cKO mice or Mfn2 overexpressing mice served as controls. The animal studies were approved by the Institutional Animal Care and Use Committee of Case Western Reserve University.
Cell culture and transfection
The human neuroblastoma SH-SY5Y cells (American Type Culture Collection, CRL-2266, Manassas, VA, USA) were grown in Opti-MEM medium (Invitrogen; Waltham, MA, USA) at 37°C with 5% CO2, supplemented with 5% (v/v) fetal bovine serum (Gibco, Waltham, MA, USA) and 1% penicillin-streptomycin (Gibco). Cells were transfected using Lipofectamine 2000 Reagent (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's instructions.
Stable cell lines
The stable Mfn2 KO SH-SY5Y cells were constructed by using CRISPR/Cas9 methods. pLentiCRISPR vector (Addgene, Watertown, MA, USA) with gRNA targeting Mfn2 was co-transfected into HEK293 Lenti cells with the packaging plasmids pVSVg (Addgene) and psPAX2 (Addgene). The culture medium was collected 24 h later after transfection, and was added to the normal SH-SY5Y cells for infection. Clonal cell lines were isolated by culturing single cells in 96-well plates and were screened by western blotting with anti-Mfn2 antibody. Human Mfn2 overexpressing plasmid constructed by using the pCMV-Tag3 vector (Stratagene, St Louis, MO, USA) was used to establish a stable Mfn2 overexpressing SH-SY5Y cell line, and the empty vector was also used to generate a stable cell line as control of Mfn2 overexpressing cells.
Electron microscopy was performed as described previously (Jiang et al., 2018; Zhao et al., 2017). For electron microscopy analysis, mice brain samples were collected and fixed. Brain slices from the CA1 region of the hippocampus were sampled and embedded in Poly/Bed 812 embedding resin. The sections were examined using an FEI Tecnai T12 electron microscope equipped with a Gatan single tilt holder and a charged-couple device (CCD) camera. For electron microscopy analysis of Mfn2 cKO mice, 8-week-old to 28-week-old Mfn2 cKO mice and 4-week-old to 28-week-old wild-type mice were sacrificed for hippocampi sample preparation. Mfn2 overexpressing or wild-type mice at 24 to 28 weeks old were used for electron microscopy analysis. SH-SY5Y cells were fixed with 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.2) for 2 h at room temperature and then examined by using an electron microscope. The length and distance of the ER associated with mitochondria, as well as the contact numbers, were measured. The ERMICC contact index was calculated using the following formula: ERMICC=interface length/(mitochondrial perimeter×distance ER-mitochondria) (Naon et al., 2016). The MAM coverage was calculated by MAM length divided by mitochondria perimeter per cell. Mitochondria parameters, including perimeter and size, were also quantified. All the electron microscope images were analyzed using ImageJ.
Antibodies and other reagents
Mouse anti-IP3R3 (sc-377518; 1:1000 for western blot, 1:200 for PLA), mouse anti-Mfn2 (sc-100560; 1:1000 for western blot) and mouse anti DJ-1 (sc-55573; 1:4000 for western blot) were purchased from Santa Cruz Biotechnology (Dallas, TX, USA). Rabbit anti-calnexin (CNX) (2679; 1:4000 for western blot), rabbit anti-DJ-1 (5933; 1:4000 for western blot), rabbit anti-Grp75 (3539; 1:3000 for western blot), mouse anti-GAPDH (2118; 1:4000 for western blot) and mouse-anti COX IV (11967; 1:2000 for western bot) were purchased from Cell Signaling Technology (Danvers, MA, USA). Rabbit anti-Sig1R (15168-1-AP; 1:1000 for western blot), rabbit anti-VDAC1 (55259-1-AP; 1:200 for PLA), rabbit anti-VAPB (14477; 1:1000 for WB), rabbit anti-PTPIP51 (20641-1-AP; 1:2000 for western blot) and mouse anti-HSPD1 (66041; 1:5000 for western blot) were obtained from Proteintech (Rosemont, IL, USA). Mouse anti-VDAC1 (ab14734; 1:5000 for western blot), mouse anti-FACL4 (ab155282; 1:1000 for western blot) and mouse anti-PTPIP51 (ab69337; 1:1000 for western blot) were purchased from Abcam (Cambridge, MA, USA). Mouse anti-actin (MA5-11869; 1:5000 for western blot) was purchased from Thermo Fisher Scientific. Histamine (H7250, Sigma-Aldrich) was obtained from Sigma Aldrich (St Louis, MO, USA).
Western blot analysis
Mouse brain hippocampus tissues were manually homogenized using a Teflon pestle in IB-1 buffer [225 mM mannitol, 75 mM sucrose, 0.5% bovine serum albumin (BSA), 0.5 mM EGTA and 30 mM Tris-HCl (pH 7.4)]. Unbroken cells were removed by centrifugation at 750 g for 5 min, and supernatants were further lysed with RIPA buffer (Abcam) supplemented with protease inhibitor cocktail (Cell Signaling Technology) on ice for 30 min. Unsolubilized components were removed by centrifugation at 18,000 g for 20 min. Proteins from total lysates of SH-SH5Y cells were obtained by incubating cells with RIPA buffer together with protease inhibitor cocktail for 30 min on ice. After centrifugation at 18,000 g for 20 min, the supernatants were collected. The concentrations of proteins extracted from mouse tissues and cells were determined using a bicinchoninic acid assay (BCA) protein assay kit (Thermo Scientific). Protein (10 μg) was loaded onto a 4-15% (w/v) SDS-PAGE gel and transferred to PVDF membranes (Millipore). After blocking with 10% nonfat milk in TBS containing 0.1% Tween 20 (TBST) for 1 h at room temperature, primary antibodies were incubated overnight at 4°C. After washing with TBST buffer, the membranes were incubated with secondary antibodies conjugated to horseradish peroxidase (HRP) in wash buffer for 1 h at room temperature on a shaker. The blots were developed with Immobilon Western Chemiluminescent HRP substrate (Millipore) after four final washes in TBST, and bands were detected using a CCD camera (Amersham Imager 600, GE, Chicago, IL, USA). The intensity of the bands was analyzed using ImageJ. Four 4- to 9-week-old Mfn2 cKO mice and four wild-type mice (half male and half female for each group) were sacrificed for western blot analysis. Four 24- to 28-week-old male Mfn2 overexpressing mice and four male wild-type mice were used for western blot analysis.
MAM, mitochondria and ER were prepared from mouse hippocampus tissues and SH-SY5Y cells as described previously (Liu et al., 2019; Wieckowski et al., 2009). Briefly, cells were washed by centrifugation at 500 g for 5 min with PBS, and resuspended in IBcells-1 buffer [(225 mM mannitol, 75 mM sucrose, 0.1 mM EGTA and 30 mM Tris-HCl (pH 7.4)] at 4°C. Mouse hippocampus tissues were washed and resuspended in IB-1 buffer [225 mM mannitol, 75 mM sucrose, 0.5% BSA, 0.5 mM EGTA and 30 mM Tris-HCl (pH 7.4)]. Cells and tissue suspensions were gently homogenized using a glass Dounce homogenizer. The homogenate was then centrifuged twice at 600 g for 5 min to remove nuclei and unbroken cells. The crude mitochondrial fraction was pelleted by centrifugation at 10,000 g for 10 min. The supernatants were centrifuged at 20,000 g for 30 min at 4°C to remove lysosomal and plasma membrane fractions. The supernatants were further centrifuged at 100,000 g for 90 min at 4°C to pellet the ER/microsomes. The fraction of crude mitochondria was washed and resuspended in mitochondrial reconstitution buffer (MRB) [250 mM mannitol, 5 mM HEPES (pH 7.4), and 0.5 mM EGTA], and layered on top of a 30% Percoll gradient buffer [225 mM mannitol, 25 mM HEPES (pH 7.4), 1 mM EGTA, and 30% Percoll]. After centrifugation at 95,000 g for 30 min, a dense band containing the pure mitochondria was recovered at the bottom of the gradient, followed by washing with MRB buffer and centrifugation at 6300 g for 10 min to remove Percoll. The MAM-containing band was retrieved above the mitochondrial band after gradient centrifugation, and resuspended in MRB buffer followed by centrifugation at 100,000 g for 90 min. All final organelle pellets were resuspended and lysed in RIPA buffer supplemented with protease inhibitor cocktail, and the protein concentrations were determined by BCA assay (Thermo Scientific).
Mitochondrial and cytosolic Ca2+ measurements
Mitochondrial calcium uptake was measured as we previously reported (Liu et al., 2019). Control, Mfn2 knockout, EV and Mfn2 overexpressing SH-SY5Y cells were cultured on a 96-well plate with clear bottoms and transfected with pCMV-CEPIA4-mt (Suzuki et al., 2014) or pCMV-mito-R-GECO1 (ex 552 nm/em 589 nm; Addgene, 46021) to monitor the mitochondrial Ca2+ level. Cytosolic Ca2+ level was monitored by transfection of pcDNA3-Cyto-CaMP3 (ex 485 nm/em 528 nm; Addgene, 64853). The mitochondrial and cytosolic Ca2+ levels of the cells were measured 36 h after transfection. The cells were balanced in the HBSS buffer [140 mM NaCl, 5 mM KCl, 0.4 mM MgSO4, 0.5 mM MgCl2, 0.4 mM KH2PO4, 0.6 mM NaHPO4, 3 mM NaHCO3 and 10 mM glucose, with 20 mM HEPES (pH 7.4)] for 30 min, and then loaded on a Gen5 Synergy Microreader (Biotek, Winooski, VT, USA) and monitored with the indicated excitation (ex) wavelength and emission (em) wavelength. The fluorescence value of basal level (F0) was recorded for 2 min. To test mitochondrial Ca2+ uptake from ER, histamine solution was applied to induce Ca2+ release from ER. To test the intrinsic mitochondrial calcium uptake activities, cells were treated with 2 μM ionomycin and 15 mM Ca2+ (added to HBSS with 20 mM HEPES). The mitochondrial and cytosolic Ca2+ uptake (F) were recorded continuously in cells by pCMV-CEPIA4-mt (or pCMV-mito-R-GECO1) and pcDNA3-Cyto-CaMP3 transfection, respectively. The Ca2+ levels were calculated as relative fluorescence change (ΔF/F0).
Immunoprecipitations were performed using a Dynabeads Protein G Immunoprecipitation kit (Invitrogen) following the manufacturer's instructions. Cells were washed with PBS and solubilized in lysis buffer [50 mM Tris (pH 7.4), 250 mM NaCl, 5 mM EDTA, 50 mM NaF, 1 mM Na3VO4, 1% NP40, 0.02% NaN3 and protease inhibitor cocktail, 1 mM phenylmethylsulfonyl fluoride] for 30 min on ice. For immunoprecipitation, 1 μg primary antibodies or normal rabbit or mouse IgG were incubated with 50 μl Dynabeads at room temperature for 10 min. Cell lysates were centrifuged at 18,000 g for 20 min, and the supernatant was subjected to the Dynabeads coated with antibodies and incubated at 4°C overnight. The beads were washed four times in lysis buffer and the precipitated proteins were eluted in 1× SDS sample buffer. The samples were analyzed by SDS-PAGE and immunoblotting with appropriate antibodies.
Proximity ligation assays
In situ PLAs were performed to quantify IP3R3-VDAC1 interactions using a Duolink kit (Sigma-Aldrich) following the manufacturer's protocol. SH-SY5Y cells were fixed in 4% paraformaldehyde probed with mouse IP3R3 and rabbit VDAC1 antibodies after permeabilization and blocking. The secondary antibodies conjugated with oligonucleotide (anti-rabbit PLUS and anti-mouse MINUS) were applied and then ligated. After ligation, the PLA signals were amplified by polymerase and captured using a fluorescence microscope. The numbers of signals were quantified using the particle analysis function in ImageJ.
Blue-native PAGE was performed according to the manufacturer's instructions (Invitrogen). Crude mitochondrial fractions were isolated from mouse hippocampus tissues and SH-SY5Y cells. The fractions were lysed in solubilization buffer (1% digitonin, 4X NativePAGE Sample Buffer, Invitrogen) on ice for 15 min. After centrifugation at 18,000 g for 10 min at 4°C, Coomassie Brilliant Blue G-250 (0.2% final concentration) was added, and the samples were electrophoresed through 3-12% native polyacrylamide gradient gels. The gels were subjected to immunoblotting using appropriate antibodies.
Statistical analysis was performed using GraphPad Prism 6.0. Unless stated otherwise, data were compared using an unpaired two-tailed Student's t-test. P<0.05 was considered statistically significant. For all figures, *P<0.05, **P<0.01, ***P<0.001 and ****P<0.0001.
Conceptualization: X.Z.; Methodology: S.H., F.Z., J.H., H.F., X.Z.; Formal analysis: S.H., F.Z., X.Z.; Investigation: S.H., F.Z., J.H., X.M., Y.L., S.T., H.F.; Writing - original draft: S.H.; Writing - review & editing: X.Z.; Visualization: S.H.; Supervision: X.Z.; Project administration: X.Z.; Funding acquisition: X.Z.
The work was supported in part by the National Institutes of Health (NS083385, AG049479 and AG056363 to X.Z.) and the Alzheimer's Association (AARG-16-443584 to X.Z.). Deposited in PMC for release after 12 months.
The authors declare no competing or financial interests.