The regulation of ion and pH homeostasis of endomembrane organelles is critical for functional protein trafficking, sorting and modification in eukaryotic cells. pH homeostasis is maintained through the activity of vacuolar H+-ATPases (V-ATPases) pumping protons (H+) into the endomembrane lumen, and counter-action by cation/proton exchangers, such as the NHX family of Na+(K+)/H+ exchangers. In plants, V-ATPase activity at the trans-Golgi network/early endosome (TGN/EE) is important for secretory and endocytic trafficking; however, the role of the endosomal antiporters NHX5 and NHX6 in endomembrane trafficking is unclear. Here we show through genetic, pharmacological and live-cell imaging approaches that double knockout of NHX5 and NHX6 results in the impairment of endosome motility and protein recycling at the TGN/EE, but not in the secretion of integral membrane proteins. Furthermore, we report that nhx5 nhx6 mutants are partially insensitive to osmotic swelling of TGN/EE induced by the monovalent cation ionophore monensin, and to late endosomal swelling by the phosphatidylinositol 3/4-kinase inhibitor wortmannin, demonstrating that NHX5 and NHX6 function to regulate the luminal cation composition of endosomes.
The endomembrane system of eukaryotic cells is composed of a complex series of interconnected compartments that function in the synthesis, sorting, transport and degradation of proteins. For these cellular processes to operate efficiently, endomembrane organelles must control their luminal pH by balancing the activity of proton pumps and cation channels (Casey et al., 2010). In plants, the trans-Golgi network also acts as an early endosome, and functions to sort and transport both newly endocytosed and secretory proteins (Viotti et al., 2010). The trans-Golgi network/early endosome (TGN/EE) maintains a large proton gradient for the sorting and secretion of functional proteins, which is achieved through V-ATPase-mediated proton pumping into the TGN/EE lumen (Schumacher, 2014). This acidification results in the TGN/EE being the most acidic endomembrane compartment in plants (Martinière et al., 2013; Shen et al., 2013). By contrast, cation/proton exchangers, including the NHX family of Na+, K+/H+ exchangers and cation/(H+) exchanger (CHX) family of K+/H+ exchangers act as a proton leak to alkalinise the lumen of endosomes and assist in fine-tuning of organelle pH (Bassil et al., 2012; Brett et al., 2005b; Chanroj et al., 2012). Moreover, cation/proton exchangers also function in cation detoxification and are important for salt tolerance in plants (Rodríguez-Rosales et al., 2009).
Intracellular NHX-type exchangers have evolutionarily conserved roles in ion and pH homeostasis and, in plants, also function in development and in protein trafficking to the vacuole (Bassil et al., 2012; Chanroj et al., 2012; Dragwidge et al., 2018). Arabidopsis thaliana has eight NHX genes; four encode tonoplast-localised proteins (NHX1, NHX2, NHX3, NHX4), two are endosomal-localised (NHX5, NHX6), and two are present on the plasma membrane (NHX7 and NHX8) (Brett et al., 2005b). In A. thaliana, double knockouts of the endosomal NHX isoforms nhx5 and nhx6 show defects in vacuolar transport, with delayed trafficking of the endocytic tracer dye FM4-64 and mis-secretion of a vacuole-targeted yeast carboxypeptidase-Y (CPY) fragment (Bassil et al., 2011a). Similar vacuolar trafficking defects have been described in yeast, with the knockout of the single endosomal nhx1 gene causing altered CPY secretion, delayed vacuolar trafficking, and defects in late endosome/ multi-vesicular body (LE/MVB) formation and sorting (Bowers et al., 2000; Brett et al., 2005a; Kallay et al., 2011; Mitsui et al., 2011). Furthermore, silencing by using RNA interference (RNAi) of mammalian endosomal orthologues NHE6 and NHE8 leads to disruptions in endosome trafficking and recycling (Lawrence et al., 2010; Ohgaki et al., 2010), demonstrating that eukaryotic endosomal NHX proteins have a conserved role in subcellular protein trafficking and recycling.
In plants, endosomal NHX antiporters have been implicated in the trafficking of soluble cargo proteins to the vacuole. Soluble proteins, such as seed storage proteins, are synthesised in the ER, bind vacuolar sorting receptors (VSRs) and transit towards the TGN/EE (Künzl et al., 2016), before budding and maturation of the TGN/EE into the LE/MVB, which then ultimately fuse with the vacuole (Scheuring et al., 2011). In A. thaliana nhx5 nhx6 double knockout mutants have inhibited VSR-cargo interactions and mis-processing of seed storage proteins (Ashnest et al., 2015; Reguera et al., 2015). These defects are believed to be caused by hyper-acidification of endomembrane luminal compartments in nhx5 nhx6 knockouts (Reguera et al., 2015).
Maintenance of functional pH homeostasis at the TGN/EE has also been shown to be important for the trafficking of transmembrane proteins (Luo et al., 2015). Reduction of V-ATPase activity at the TGN/EE through Concanamycin-A treatment or in the det3 mutant results in alkalinisation of the TGN/EE and, subsequently, leads to delayed vacuolar transport, reduced Golgi and TGN/EE motility, and defects in the secretion and recycling of the receptor BRI1 (Dettmer et al., 2006; Luo et al., 2015; Viotti et al., 2010). These data suggest that maintaining correct pH of the TGN/EE by V-ATPases is important for TGN/EE function and in the trafficking of integral membrane proteins. Despite NHX5 and NHX6 regulating the pH of multiple endosomal compartments and being broadly implicated in vesicle trafficking (Bassil et al., 2011a; Reguera et al., 2015), evidence for their involvement in the trafficking of integral membrane proteins has not been shown so far.
Here, we investigated the effects of disrupted endomembrane pH and ion balance through dissection of the secretory and endocytic transport pathways in the double knockout nhx5 nhx6, in A. thaliana. Through live cell-imaging, we reveal that nhx5 nhx6 mutants have reduced Golgi and TGN/EE motility, and defects in the recycling of transmembrane receptors from the TGN/EE. Furthermore, our results reveal that nhx5 nhx6 endosomes are insensitive to osmotic swelling induced by the ionophore monensin or the late-endosome inhibitor wortmannin.
NHX5 and NHX6 are involved in endomembrane compartment motility
In plant cells, the movement of endomembrane organelles through the secretory and endocytic pathways is essential for functional protein delivery, and is highly dependent on the cytoskeleton network of actin filaments and microtubules (Brandizzi and Wasteneys, 2013). As alkalinisation of the TGN/EE has been shown to reduce Golgi and TGN/EE motility (Luo et al., 2015), we questioned whether the hyper-acidification of endomembrane organelles in nhx5 nhx6 could also negatively affect their motility. We generated stable A. thaliana nhx5 nhx6 lines expressing the endosomal markers YFP-Got1p (Golgi), and YFP-VTI12 (TGN/EE) and examined endosomal motility using spinning disk confocal microscopy of live cells. Quantitative analysis revealed that the movement of Golgi and TGN/EE vesicles was significantly reduced in nhx5 nhx6 cells compared to that in wild-type cells (Fig. 1A-E; Fig. S1A,B). Additionally, the proportion of slower moving bodies (<10 μm min−1) was more than double in nhx5 nhx6 compared to wild type, demonstrating that a high proportion of vesicles exhibited minimal movement in nhx5 nhx6 (Fig. S1C). We also assessed the straightness of particle tracks, in order to indirectly assess whether organelle behaviour or their potential association with the cytoskeleton may be altered. Quantification revealed a significant reduction in the straightness of TGN/EE trajectories in nhx5 nhx6 cells, indicating that these vesicles displayed more disordered, non-continuous movement (Fig. 1F), typical of cytoskeletal independent endosome movement (Akkerman et al., 2011).
BRI1 recycling but not secretion is reduced in nhx5 nhx6 double mutants
Next, we investigated whether the transport or recycling of transmembrane receptors at the TGN/EE is inhibited in nhx5 nhx6 cells. We employed the well-characterised receptor kinase BRASSINOSTEROID-INSENSITVE 1 (BRI1) as it is constitutively endocytosed from the plasma membrane to the TGN/EE, where it is sorted for recycling back to the plasma membrane or for degradation towards the vacuole (Dettmer et al., 2006; Geldner et al., 2007; Irani et al., 2012). We treated root cells expressing BRI1-GFP with the fungal toxin brefeldin-A (BFA) to reversibly inhibit TGN/EE and Golgi trafficking (Geldner et al., 2001; Richter et al., 2007), and assessed the recycling of BRI1-GFP out of BFA bodies after washout (Fig. 2A). Quantification revealed that nhx5 nhx6 cells had larger BFA bodies and a higher proportion of cells containing BFA bodies after washout (Fig. 2B,C). These results indicate that BRI1 recycling from the TGN/EE is impaired in nhx5 nhx6 and, together, with similar BRI1-recycling defects in det3 mutants (Luo et al., 2015), suggests that pH-sensitive trafficking machineries are required for efficient BRI1 recycling. We also assessed growth response of nhx5 nhx6 treated with the V-ATPase inhibitor Concanamycin A (ConcA), which disrupts TGN/EE structure, and endocytic and secretory trafficking (Dettmer et al., 2006; Viotti et al., 2010). Moreover, nhx5 nhx6 seedlings displayed reduced hypocotyl elongation compared to wild type, and showed similar hypersensitivity to ConcA as det3 mutants (Fig. S2), implying that nhx5 nhx6 knockouts have impaired TGN/EE function.
As dysregulation of TGN/EE V-ATPase activity has been implicated in defects of protein secretion and delivery of proteins to the plasma membrane (Dettmer et al., 2006; Luo et al., 2015), we questioned whether these pathways are sensitive to general pH disruptions and can be similarly perturbed in nhx5 nhx6 mutants. We first quantified steady-state levels of BRI1-GFP at the plasma membrane in wild-type and nhx5 nhx6 roots, but found no significant differences in fluorescence levels (Fig. 3A,B). Similarly, fluorescence recovery after photobleaching (FRAP) experiments revealed no clear difference in recovery of BRI1-GFP to the plasma membrane in nhx5 nhx6 (Fig. 3C,D), suggesting that the delivery of newly synthesised BRI1-GFP to the plasma membrane is not impaired in nhx5 nhx6. Taken together, these findings indicate that NHX5 and NHX6 activity at the TGN/EE is important for the recycling of BRI1-GFP but not for its synthesis and delivery to the plasma membrane.
NHX5 and NHX6 antiporter activity is required for wortmannin induced swelling of MVBs
As NHX5 and NHX6 have been reported to localise to the late endosome/multi-vesicular body (LE/MVB) (Reguera et al., 2015), we questioned whether trafficking or function at the LE/MVB are affected in nhx5 nhx6 mutants. Wortmannin at high concentrations (>30 µM) inhibits phosphatidylinositol 3-kinases (PI3K) and PI4K, and has been used extensively to investigate trafficking of late endosomes in plants (Jaillais et al., 2006; Vermeer et al., 2006). We examined root cells expressing the Rab5 GTPase YFP-ARA7 as a LE/MVB marker. In wild-type cells wortmannin induced enlarged ring-like structures (Fig. 4A) that were associated, but not fused with TGN/EE labelled with the SNARE VTI12 (Fig. S3). Surprisingly, although a similar number of wortmannin bodies were present in nhx5 nhx6 cells compared to those in wild type, they were significantly smaller and denser, and did not exhibit a characteristic ring-like shape (Fig. 4A,B). To test if these wortmannin-induced swelling defects are dependent on the antiporter activity of NHX6, we generated an antiporter-inactive NHX6 by mutating the highly conserved acidic Asp194 residue, which is crucial for binding of Na+ and H+, to Asn (D194N) (Lee et al., 2013; Wang et al., 2015) and fused it to EGFP (Fig. S4A). This mutated NHX6D194N-EGFP reporter did not rescue growth impairment in nhx5 nhx6 (Fig. S4B) but localised correctly to core and peripheral BFA compartments, as previously reported by using functional NHX5 and NHX6 reporters (Fig. S4C) (Bassil et al., 2011a). Wortmannin-treated cells expressing NHX6D194N-EGFP had smaller wortmannin bodies in the nhx5 nhx6 background compared to wild type and NHX5-CFP rescue line (Fig. 4C,D), suggesting that NHX antiporter activity was required for swelling of LE/MVBs in response to wortmannin.
Since wortmannin at high concentrations inhibits both PI3K and PI4K, we assessed whether nhx5 nhx6 swelling insensitivity was present during PI3K-specific inhibition. Specifically targeting PI3K pathways by using the inhibitor LY294002 or low concentrations of wortmannin (Fujimoto et al., 2015; Simon et al., 2016; Takáč et al., 2013) also caused similar fusion of MVBs into smaller, more densely labelled bodies in nhx5 nhx6 cells (Fig. 4E). This finding indicates that wortmannin and LY294002 induced swelling is caused by inhibition of PI(3)P on the LE/MVB (Fig. 4F).
Next, we examined the ultrastructure of nhx5 nhx6 MVBs through high-pressure freezing and freeze substitution of A. thaliana roots. We used the constitutively active GTP-bound mutant of Rab5 (ARA7QL) to label MVBs present in ultra-thin sections due to the large size of MVBs expressing this reporter (Ebine et al., 2011; Jia et al., 2013). Consistent with confocal microscopy analysis (Fig. 5A), we also regularly observed fused MVBs in close contact with the vacuole in wortmannin-treated samples (Fig. 5B) under transmission electron microscopy (TEM). These enlarged MVBs were also consistently smaller in nhx5 nhx6 root cells compared to those in the wild-type control. Furthermore, no obvious ultrastructural abnormalities were detected in MVBs from nhx5 nhx6 root cells (Fig. 5B).
NHX5 and NHX6 do not regulate root vacuole morphology
In plants, the primary vacuolar transport pathway is marked by the coordinated sequential activity of the Rab GTPases Rab5 (RabF) and Rab7 (RabG) (Cui et al., 2014, 2016). This pathway is also implicated in the transport of storage proteins to the protein storage vacuole (PSV) in seeds (Ebine et al., 2014; Singh et al., 2014), which has shown to be disrupted in nhx5 nhx6 (Ashnest et al., 2015; Reguera et al., 2015). The constitutively active GTP-bound mutant of Rab5 (ARA7QL) transits through the LE/MVB to the tonoplast (Ebine et al., 2011), allowing us to assess whether Rab5 recruitment or maturation could be affected in nhx5 nhx6 root cells. We observed similar localisation of ARA7QL to the LE/MVB and tonoplast in wild-type and nhx5 nhx6 cells (Fig. 5A), suggesting that Rab5 recruitment and delivery to the tonoplast occurs normally. As endosomal NHX-type antiporter activity has been previously reported to be implicated in vacuolar trafficking and fusion in the yeast nhx1 mutant (Qiu and Fratti, 2010), we assessed vacuole morphology in nhx5 nhx6 root cells by using the tonoplast-localised marker VAMP711 (Fig. S5A). Vacuolar morphology and vacuole expansion upon wortmannin treatment appeared unaffected in nhx5 nhx6 root cells (Fig. S5A,B). Together, these results suggest that NHX5 and NHX6 are not important for vacuole function in root tissue.
nhx5 nhx6 endosomes have reduced sensitivity to monensin
Since NHX proteins regulate Na+ and K+ accumulation in the endosomal lumen, we questioned whether the reduced swelling of LE/MVB induced by wortmannin in nhx5 nhx6 could be due to an altered ionic composition in the LE/MVB lumen. The monovalent cation ionophore monensin induces rapid osmotic swelling of trans-Golgi stacks and TGN/EE through exchange of Na+/K+ for H+ across the endomembrane (Zhang et al., 1993). Consistent with these data, we observed rapid swelling TGN/EE but not of LE/MVB in root cells (Fig. 6A). Next, we assessed the pH of TGN/EE after treatment with monensin using the TGN/EE localised ratiometric pH sensor SYP61-pHusion (Luo et al., 2015). TGN/EE of monensin-treated root cells were less acidic than untreated cells (Fig. 6B), consistent with monensin-induced luminal import of Na+/K+ in exchange for H+. Furthermore, we assessed the pH of TGN/EE of nhx5 nhx6 root cells and found a ∼0.5 pH reduction (acidification) of TGN/EE (Fig. 6C), similar to those previously reported in Arabidopsis protoplasts (Reguera et al., 2015).
Given the increased acidity of nhx5 nhx6 TGN/EE, we questioned whether nhx5 nhx6 root cells could be hypersensitive to swelling induced by monensin, given they produce a stronger proton gradient for monensin to act upon. We, therefore, assessed the response to monensin by using our established Golgi and TGN/EE markers in nhx5 nhx6. We could not detect any significant swelling of Golgi in either wild-type or nhx5 nhx6 monensin-treated root cells (Fig. 6D), probably because the trans-most Golgi stack only swells slightly upon monensin treatment (Zhang et al., 1993). TGN/EE in monensin-treated roots showed clear clustering and swelling in wild-type; however, in nhx5 nhx6 only minor swelling was present despite clustering of TGN/EE, suggesting these TGN/EE had a reduced capacity to swell (Fig. 6D). Furthermore, similar results were obtained using NHX6D194N-EGFP as a marker for NHX6 activity in wild-type and nhx5 nhx6 backgrounds (Fig. 6D). Together, these results suggest that TGN/EE in nhx5 nhx6 cells slowed osmotic swelling induced by monensin, probably originating from a disruption to the intraluminal ion (Na+/K+) balance in these endosomes.
In this study, we investigated the function of NHX5 and NHX6 in endomembrane trafficking and ion homeostasis in A. thaliana. Previous studies identified that NHX5 and NHX6 localise to the Golgi, TGN/EE and LE/MVB, and play roles in soluble cargo trafficking in seeds (Bassil et al., 2011a; Reguera et al., 2015). Here, we showed that NHX5 and NHX6 play additional roles in important endomembrane processes including endosomal motility and trafficking at the TGN/EE. Furthermore, we reported that monensin and wortmannin induces osmotic swelling of endosomes, which is affected by endosomal ion homeostasis maintained by NHX5 and NHX6. Our results demonstrate that NHX antiporters are important regulators of endomembrane pH and ion homeostasis required for efficient endomembrane trafficking.
Regulation of endosomal ion composition by NHX5 and NHX6
Here, we identified a novel mechanism of swelling of late endosomes induced by wortmannin, which shares a striking resemblance to osmotic swelling of TGN/EE induced by the cation ionophore monensin. Whereas wortmannin has been well-described to cause enlargement of LE/MVBs from homotypic membrane fusion of MVBs (Wang et al., 2009; Zheng et al., 2014), the typical size of these wortmannin bodies is much larger than what can be achieved through membrane fusion alone. Previous reports show that monensin-induced TGN/EE swelling occurs through Na+/K+ transport into the endosomal lumen across a proton gradient (Zhang et al., 1993), resulting in luminal alkalinisation of endosomes, consistent with our reported pH measurements. Similarly, wortmannin treatment induces alkalinisation of the LE/MVB (Martinière et al., 2013), and suggests swelling induced by wortmannin may be a consequence of rapid luminal cation (e.g. K+) import.
It is currently not known how the presumed cation import is induced by wortmannin in fused late endosomes. Since LY294002 and wortmannin at low concentrations specifically inhibit PI3K, this would result in reduced PI3P levels on the late endosome and, consequently, a loss of membrane identity due an altered lipid composition. This loss of membrane identity could result in the de-repression of inactive cation transporters present on the MVB/LE endomembrane, such as CHX17 (Chanroj et al., 2013), NHX5 and NHX6. This would then facilitate rapid ion influx and induce osmotic swelling at the late endosome, similar to the cation influx induced by monensin at the TGN/EE. Investigation of changes in endosomal K+ concentrations in vivo, by using recently established genetically encoded fluorescent K+ probes may facilitate greater understanding of cation compositions and dynamics of endosomal compartments (Bischof et al., 2017).
While it has been predicted that the rate of monensin-induced swelling corresponds to the acidity of secretory pathway compartments (Zhang et al., 1993), we propose here that the ionic composition of the vesicle also contributes to the rate of swelling. As plant NHX proteins function in intracellular internalisation of Na+ and K+ (Bassil et al., 2011a,b), lack of NHX-mediated Na+ /K+ import into endosomes within nhx5 nhx6 would result in reduced luminal Na+/K+ concentration. Thus, whereas monensin retains functional activity as a cation importer and causes aggregation of TGN/EE in nhx5 nhx6, osmotic swelling occurs slower due to the reduced luminal cation concentration at the TGN/EE (Fig. 7). Similarly, the hyper-acidified MVB/LE and reduced Na+/K+ composition in nhx5 nhx6 mutants contributes to insensitivity of the late-endosome swelling to wortmannin. Therefore, we have demonstrated that NHX5 and NHX6 antiporter activity regulates both pH and cation composition of the TGN/EE and MVB/LE.
NHX5 and NHX6 are important for endosome motility
Here, we identified that Golgi and TGN/EE motility is significantly reduced in nhx5 nhx6 hypocotyl cells, and that the behaviour of TGN/EE is altered – which is suggestive of a reduction in the ability of TGN/EE to associate with the cytoskeleton. These findings are similar to reported defects of Golgi and TGN/EE motility in det3 mutants, which also have an altered TGN/EE pH (Luo et al., 2015); however, the functional consequences of a reduction in endosome motility are unclear. Previous reports indicate that the growth of root hair cells is significantly reduced in nhx5 nhx6 knockouts (Bassil et al., 2011a). As tip-directed growth of root hair cells relies on rapid endosomal transport along actin filaments (Szymanski and Staiger, 2018), we speculate that the reduced endosomal motility in nhx5 nhx6 may lead to slower endosomal transport to the growing cell tip and, consequently, inhibit the rate of cell expansion. Similarly, reduced endosome motility in the hypocotyl could explain the reduction in hypocotyl cell elongation in both nhx5 nhx6 and det3 mutants, and suggests a pH-sensitive mechanism that governs endosome-cytoskeleton association.
NHX5 and NHX6 play roles in recycling at the TGN/EE
Pharmacological BFA and FRAP experiments indicate that nhx5 nhx6 have defects in BRI1-GFP recycling from the TGN/EE, but not in general BRI1 secretion to the plasma membrane. These results contrast with findings in det3 V-ATPase mutants, which have both impaired secretion and recycling of BRI1 (Luo et al., 2015). Since the de novo synthesis and secretion of BRI1 is likely to be the main contributor to the pool of BRI1 present at plasma membrane, this explains why we found no significant differences in BRI1-GFP levels at the plasma membrane in nhx5 nhx6 mutants, despite clear defects in BRI1 recycling to the cell surface from the TGN/EE. Given that det3 mutants have alkalinised TGN/EE owing to reduced V-ATPase activity and that nhx5 nhx6 have hyper-acidified TGN/EE, these findings show that maintaining correct homeostasis of TGN/EE pH is essential for functional BRI1 recycling from the TGN/EE but not necessarily for BRI1 secretion.
Interestingly, the recycling of polar localised auxin carrier proteins PIN1-GFP and PIN2-GFP from the TGN/EE is not significantly impaired in nhx5 nhx6 root cells (Dragwidge et al., 2018). These findings are consistent with emerging evidence that the TGN/EE is composed of functionally segregated domains that sort distinct membrane cargoes for delivery to the plasma membrane (Li et al., 2016; Singh and Jürgens, 2017), and suggest that only some recycling pathways are sensitive to endomembrane pH disruptions. Further work to characterise and clearly identify the specific TGN/EE-to-cell surface delivery pathways and the distinct sets of receptors and transporters that are transported will be vital for a better understanding of receptor-mediated endosome sorting in plants.
Our results showing that nhx5 nhx6 are hypersensitive to treatment with the V-ATPase inhibitor ConcA further support a role for NHX5 and NHX6 at the TGN/EE. As ConcA leads to alkalinisation of TGN/EE (Luo et al., 2015), ConcA treatment should, instead, counteract the hyper-acidification of TGN/EE in nhx5 nhx6 and, potentially, restore TGN/EE function. Instead, the hypersensitivity of nhx5 nhx6 and det3 mutants to ConcA probably originates from impaired TGN/EE function in these mutants and, therefore, an increased sensitivity to the secondary effects of ConcA, including disruptions to TGN/EE structure and identity (Viotti et al., 2010). These findings are supported by reports showing det3 mutants have impaired TGN/EE recycling and motility (Luo et al., 2015), as well as similar defects in nhx5 nhx6 mutants presented here.
How NHX5 and NHX6 affect trafficking at the TGN/EE is unknown. In mammals and Drosophila, NHX and V-ATPase activity has been shown to effect electrostatics at the cytosolic surface of endomembranes, which can affect membrane signalling or protein recruitment. Specifically, endosomal V-ATPase activity is required for recruitment of the small GTPase Arf6 In mammalian cells (Hurtado-Lorenzo et al., 2006), while pH- and charge-dependent Wnt signalling at the plasma membrane is regulated by Nhe2 activity in Drosophila (Simons et al., 2009). Thus, potential disruptions to endomembrane electrostatics in nhx5 nhx6 may influence the charge-dependent recruitment of small GTPases or Arfs involved in protein trafficking or recycling. Further investigation of endosomal trafficking pathways through selective trafficking inhibitors such as endosidin compounds may uncover a clearer understanding of the specific trafficking pathways which require NHX5 and NHX6 activity (Hicks and Raikhel, 2010; Li et al., 2016).
In conclusion, we have shown that endosomal pH and ion regulation by plant NHX proteins is important for functional endosomal behaviour. We demonstrate that NHX5 and NHX6 activity is necessary for functional Golgi and TGN/EE motility, protein recycling at the TGN/EE, and for regulation of endosomal ion balance. This work sheds light on the complex nature of the plant endomembrane system and demonstrates the importance of regulation of endomembrane ion and pH homeostasis.
MATERIALS AND METHODS
Plant material and growth conditions
A. thaliana lines were all in the Columbia-0 (Col-0) accession background. Plant lines used have been previously described. These are nhx5-1 nhx6-1 (Bassil et al., 2011a), nhx5-2 nhx6-3 (Ashnest et al., 2015), det3 (Schumacher et al., 1999), p35S:NHX5-CFP (Ashnest et al., 2015), pBRI1::BRI1-GFP (Geldner et al., 2007), GFP-ARA7Q69L (Ebine et al., 2011), pUBQ10::xFP-ARA7 (RabF2b) (Wave2Y/R), pUBQ10::YFP-Got1p (Wave18Y), pUBQ10::xFP-VAMP711 (Wave 9Y/R), pUBQ10::YFP-VTI12 (Wave13Y) (Geldner et al., 2009) and VHA-a1-GFP x mRFP-ARA7 (Singh et al., 2014). The nhx5-2 nhx6-3 allele combination was used for all experiments except for the SYP61-pHusion pH measurements and ConcA hypocotyl assay where nhx5-1 nhx6-1 was used. BRI1-GFP in nhx5 nhx6 was generated by crossing BRI1-GFP into NHX5/nhx5 nhx6 plants, and nhx5 nhx6 plants homozygous for BRI-GFP were identified in the following generations. The nhx5 nhx6 lines expressing other fluorescent markers were obtained by floral dripping (Martinez-Trujillo et al., 2004) using Agrobacterium tumefaciens GV3101 strain cultures containing the given constructs.
Seeds were surface sterilised with 70% ethanol for 5 min, 10% bleach for 5 min, washed three times in ddH2O and grown on ½ strength Murashige and Skoog (½ MS) medium containing 1.0% (w/v) agar pH 5.8, without sucrose unless indicated. Seedlings were stratified for 48 h at 4°C in the dark and grown in a 16 h light/8 h dark photoperiod at 22°C at 100 µmol m−2 s−1 under cool-white fluorescent lights.
For live cell microscopy and chemical treatments, seedlings were grown vertically on ½ MS plates with 1.0% (w/v) agar without sucrose. For time-lapse motility assays, etiolated seedlings were grown at 22°C in the dark for 4 days.
For all cloning procedures, the Gateway recombination system (Thermo Fisher Scientific) was used. NHX6D194N-EGFP was generated through site directed mutagenesis using partially overlapping primers (Table S1). D194N-F1 and D194N-R1 primers were used to amplify from a NHX6-EGFP pENTR-D-TOPO template using PfuUltra high-fidelity DNA polymerase, then LR recombined into pEarlyGate100 to generate p35S:NHX6D194N-EGFP.
Confocal microscopy and drug treatments
Confocal images were acquired with a Leica SP5 or Zeiss LSM 780 laser scanning confocal microscope using a C-Apochromat 40×/1.3 water objective at 2× digital zoom and 1024×1024 pixels per image. Excitation and emission detection settings were as follows: ECFP 458 nm/460-520 nm; GFP/YFP 488 nm/490-560 nm; mRFP/mCherry 561 nm/565-630 nm; FM5-95 561 nm/565-650 nm. For all quantification experiments, identical settings were used to acquire each image. Chemical stock solutions were made in DMSO at the following concentrations: brefeldin-A (BFA) 50 mM (Sigma-Aldrich), cycloheximide (CHX) 50 mM (Sigma-Aldrich), FM5-95 4 mM (FM4-64 analogue, Thermo Fisher Scientific), wortmannin 33 mM (LC Laboratories), LY294002 50 mM (Sigma-Aldrich), 1 mM Concanamycin A (ConcA) (Santa Cruz Biotechnology). Monensin stock was made in EtOH at 10 mM (Santa Cruz Biotechnology). An equal concentration of DMSO or EtOH was used in control treatments. For drug treatments, 6-7-day-old seedlings were incubated in 6-well plates for 60 min for BFA treatments, 90 min for wortmannin and LY294002 treatments, or 30 min for monensin treatments.
For quantification of BRI1-GFP BFA bodies, maximum intensity projections were generated from 3-4 slices spaced at 2 μm. BFA body size was quantified with Fiji (Schindelin et al., 2012) based on ImageJ v1.48 g, using automated thresholding and the Analyse Particles tool with a minimum size of 0.7 μm2 and circularity of ≥0.3. At least 15 epidermal cells were counted from each root. For quantification of wortmannin bodies, circular regions of interest (ROIs) were manually drawn and the body area was measured using ImageJ. For analysis of bodies in YFP-VTI-12 lines, only clearly fused bodies (≥1 μm2) were used for analysis. For quantification of BRI1-GFP plasma membrane fluorescence, plasma membrane regions from ≥6 cells per root were quantified using fixed ROIs, and mean fluorescence was calculated after subtracting background fluorescence. For ConcA hypocotyl length measurements, etiolated seedlings were scanned and measured using the ‘segmented line’ tool in ImageJ. Confocal images were post-processed using the Gaussian blur filter at 0.6-1.0 sigma.
Fluorescence recovery after photobleaching (FRAP)
FRAP analysis was performed on a Zeiss LSM 780 laser confocal microscope using a 40×/1.2 water objective. Seedlings that were 7 days old were transferred from plates onto single well Lab-Tek™ Chambered Coverglass slides, and a thin agar slice was placed on top. A. thaliana root epidermal cells were bleached using a 488 nm argon laser at 100% power for 60 s with a circular 30 μm diameter ROI. Images were acquired using 512×512 pixel resolution pre-bleach at 0 min, 10 min, 20 min, 30 min, 40 min and 50 min after bleaching.
Image series were aligned using Stackreg and Linear Stack Alignment with SIFT in ImageJ. For analysis, a fixed ROI was used to select plasma membrane from completely bleached cells. Fluorescence values from bleached cells were normalised to fluorescence from ROIs from two unbleached cells in the same image to account for minor photobleaching during image acquisition. Plasma membrane fluorescence before bleaching was set to 100%, and directly after bleaching as 0%. Percent of fluorescence recovery after photobleaching was calculated by dividing the normalised bleached fluorescence value minus background (t=0 min) from the pre-bleached value minus background (t=0 min). The experiment was repeated, and similar results were obtained.
Endosome motility imaging and analysis
Time-lapse motility experiments were performed on a Zeiss Cell Observer spinning disk confocal microscope equipped with a Yokogawa CSU-X1 spinning disk and Photometrics EMCCD camera, using a 63× oil immersion objective. Seedlings grown in the dark for 4 days were placed onto single well Lab-Tek™ Chambered Coverglass (Thermo Fisher Scientific) and a thin agar slice was placed on top. Epidermal cells from the upper hypocotyl were imaged at the cortical focal plane just below the plasma membrane over a 2 min period with 1 s scanning intervals. Only cells of similar size (∼900-1200 μm2) were selected for analysis to minimise variation in particle speeds due to cell size. Cytoplasmic streaming was observed to verify cell viability.
Image drift was corrected using the ImageJ plugin StackReg. Postprocessing and analysis was performed in IMARIS software v7.0 (Bitplane). Particle tracking was performed using the ‘spots’ feature with the autoregressive motion algorithm, with a max distance parameter of 10 pixels, and a gap parameter of 0. Particle tracks <10 s long were filtered and excluded from analysis. Tracks were verified manually, and misaligned tracks were realigned. Mean track speed and track straightness was calculated in IMARIS from pooled particle tracks from ≥5 individual seedlings. Kymographs were created in ImageJ using the MultipleKymograph plugins (http://www.embl.de/eamnet/html/body_kymograph.html).
For vacuolar morphology analysis, 7-day-old seedlings expressing YFP-VAMP711 or RFP-VAMP711 were imaged on a Zeiss spinning disk confocal microscope using a 63× oil immersion objective. Z-stacks from atrichoblast epidermal cells were obtained from 30-45 slices with 200 nm step size and stacked as a maximal intensity projection. Vacuolar morphology index was calculated by measuring the maximal luminal length and width for each cell in ImageJ (Löfke et al., 2015). For quantification, ≥5 cells from ≥5 individual seedlings were analysed.
TGN/EE pH measurements
pH measurements of TGN/EE using the SYP61-pHusion line was performed as previously described (Luo et al., 2015). Briefly, 6-7-day-old seedlings expressing SYP61-pHusion were imaged on a Leica SP5 scanning confocal microscope with a HCX PL APO 63×1.20 water immersion objective with GFP (ex 488, em 490-545), and mRFP (ex 561, em 600-670) settings. Ratios were calculated by dividing the average intensity of GFP/mRFP signals after background subtraction. pH was calibrated in vivo using buffers with pH values between 4.8 and 8.0 by using free cytosolic pHusion, from which a sigmoidal calibration curve was obtained through the Boltzmann fit function in Origin Pro 9.1G and mapped to the corresponding measured values.
High-pressure freezing and electron microscopy
High-pressure freezing and freeze substitution was performed as previously described (Viotti et al., 2010). Briefly, entire root tips of 6-day-old A. thaliana seedlings stably expressing GFP-ARA7Q69L were cut and frozen in a Bal-Tech HPM 010 (Balzers Union AG) high-pressure freezer. Freeze substitution was performed under a Leica EM AFS 2 (Leica) using 0.4% uranyl acetate, and root tips were embedded in Lowicryl resin (HM20). Ultra-thin sections were obtained with a Leica Ultracut S (Leica) ultra-microtome and were observed with a JEOL JEM-1400 (JEOL, Tokyo, Japan) electron microscope operating at 80 kV.
Statistics and software
Statistics were performed using Microsoft Excel or R v3.3.2 and R Studio. Boxplots and stripcharts were generated in R v3.3.2. Images were prepared in Illustrator (Adobe).
We thank the Arabidopsis Biological Resource Center (ABRC), Joanne Chory, and Tomohiro Uemura for sharing published materials, the LIMS Bioimaging Facility (La Trobe University, Australia) and Peter Lock for confocal microscopy assistance, the Electron Microscopy Core Facility (University of Heidelberg, Germany) and Stefan Hillmer for electron microscopy assistance, and Stephanie Gold for technical assistance.
Conceptualization: J.M.D., K.S., A.R.G.; Methodology: J.M.D.; Formal analysis: J.M.D., S.S.; Investigation: J.M.D., S.S.; Writing - original draft: J.M.D.; Writing - review & editing: J.M.D., A.R.G.; Supervision: K.S., A.R.G.; Project administration: A.R.G.; Funding acquisition: A.R.G.
This work was supported by a Department of Education, Employment and Workplace Relations, Australian Government, Australian Postgraduate Award to J.D., an Australian Society of Plant Scientists R.N. Robertson Travelling Fellowship to J.D. and Deutsche Forschungsgemeinschaft (German Research Foundation) grant SFB 1101 to K.S.
The authors declare no competing or financial interests.