This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Novel Regulation of Aquaporins during Osmotic Stress1
Rosario Vera-Estrella*, Bronwyn J. Barkla, Hans J. Bohnert, and Omar Pantoja
Departamento de Biologıa Molecular de Plantas, Instituto de Biotecnologıa, Universidad NacionalAutonoma de Mexico, Cuernavaca, Morelos, 62250, Mexico (R.V.-E., B.J.B., O.P.); and Departmentsof Plant Biology and Crop Sciences, University of Illinois, Urbana, Illinois 61801–3838 (H.J.B.)
Aquaporin protein regulation and redistribution in response to osmotic stress was investigated. Ice plant (Mesembryanthemumcrystallinum) McTIP1;2 (McMIPF) mediated water flux when expressed in Xenopus leavis oocytes. Mannitol-induced waterimbalance resulted in increased protein amounts in tonoplast fractions and a shift in protein distribution to other membranefractions, suggesting aquaporin relocalization. Indirect immunofluorescence labeling also supports a change in membranedistribution for McTIP1;2 and the appearance of a unique compartment where McTIP1;2 is expressed. Mannitol-inducedredistribution of McTIP1;2 was arrested by pretreatment with brefeldin A, wortmannin, and cytochalasin D, inhibitors ofvesicle trafficking-related processes. Evidence suggests a role for glycosylation and involvement of a cAMP-dependent signal-ing pathway in McTIP1;2 redistribution. McTIP1;2 redistribution to endosomal compartments may be part of a homeostaticprocess to restore and maintain cellular osmolarity under osmotic-stress conditions.
It seems intuitively obvious that plant waterchannels, aquaporins (AQP), ought to play a dynamicrole in maintaining cellular water homeostasis underconditions that necessitate modifications in water flux.Changes in water uptake and allocation would berequired to balance alterations in the cellular osmoticpotential and, therefore, AQP activity and/or expres-sion should be tightly regulated.
Environmental stimuli, including drought, dehy-dration, desiccation, and salinity, as well as a rise inabscisic acid (ABA), which accompanies the percep-tion of osmotic stress, have been shown to regulate theexpression of both tonoplast (vacuolar membrane; TP)and plasma membrane (PM) AQP at the transcriptlevel (for review, see Maurel et al., 2002).
Recent evidence supports a direct role for aquapor-ins in plant water relations and provides informationon their involvement in drought stress tolerance.Manipulation of PM intrinsic protein (PIP) transcriptlevels through overexpression (Aharon et al., 2003),or through gene silencing by antisense suppression(Kaldenhoff et al., 1998; Martre et al., 2002; Siefritzet al., 2002) or T-DNA insertion (Javot et al., 2003),resulted in changes in root hydraulic conductivity,transpiration rates, and cellular osmotic water poten-tial and, in some cases, directly affected the plant’sability to recover from water deficit conditions (Martreet al., 2002; Siefritz et al., 2002; Aharon et al., 2003).
Few studies have addressed the direct regulation ofAQP expression by osmotic stress at the protein level,and fewer still have focused on AQP dynamic behav-ior. However, when factors such as mRNA stabilityand, conceivably, altered turnover under stress con-ditions are considered, evidence for transcriptionalregulation may be less compelling than observationsindicating that protein amounts are directly affected.
In the ice plant (Mesembryanthemum crystallinum),regulation of AQP protein amount by salinity stresswas observed using peptide-specific antibodies.McTIP1;2 amounts in leaf TP decreased during saltstress and levels of McPIP2;1 (McMIPC) in root PMfractions were induced, while McPIP1;4 (McMIPA)showed no change (Kirch et al., 2000). For McTIP1;2,the decrease in protein amount in the presence of NaClwas contrasted by an increase in the presence ofmannitol (Vera-Estrella et al., 2000), suggesting precisedeciphering and discrimination of osmotic and ionicsignaling pathways.
Little is known about the synthesis of AQP and thepathway delivering them to their destination mem-branes. Endosomal trafficking of plant AQP may playa role in the regulation and turnover of these proteinsat their target membrane under conditions of osmoticstress. In animals, regulatory cycling between mem-branes has been demonstrated for several AQP.Vasopressin-induced increases in water permeabilityof kidney-collecting duct cells have been shown tooccur via shuttling of AQP2 from intracellular vesiclesto the apical PM through exocytosis (Nielsen et al.,1995; for review, see Knepper and Inoue, 1997). Also,acetylcholine-induced increases in cytoplasmic Ca21
appeared to trigger the translocation of AQP5 fromintracellular membranes to the apical PM of ratparotid tissue (Ishikawa et al., 1999). In the ice plant,changes in membrane distribution of the TP AQPMcTIP1;2, correlatedwith changes in osmotic potential,
1 This work was supported by the Consejo Nacional de Ciencia yTecnologıa (grant nos. 31794N to R.V.-E. and 33054N to O.P.) andby the National Science Foundation International Program (USA-Mexico; to O.P. and H.J.B.).
Article, publication date, and citation information can be found atwww.plantphysiol.org/cgi/doi/10.1104/pp.104.044891.
2318 Plant Physiology, August 2004, Vol. 135, pp. 2318–2329, www.plantphysiol.org � 2004 American Society of Plant Biologists www.plantphysiol.orgon April 3, 2019 - Published by Downloaded from
have been suggested (Barkla et al., 1999) and observed(Vera-Estrella et al., 2000). Here, we further character-ize regulation of the ice plant AQP McTIP1;2 underosmotic stress, and begin to dissect the mechanismsresponsible for the stress-induced changes inMcTIP1;2membrane distribution.
RESULTS
Ice Plant McTIP1;2 Is a Functional Water Channel
in Xenopus Oocytes
To study and begin to physiologically interpret thechanges in abundance and distribution of McTIP1;2from the ice plant exposed to osmotic stress, it wasimportant to functionally characterize the protein inmore detail. We used expression studies in Xenopusoocytes to determine if McTIP1;2 could indeed medi-ate water transport. Oocytes expressing McTIP1;2showed a 37-fold increase in swelling rate comparedto oocytes injected with ribonuclease (RNase)-freewater (Fig. 1A). Included as a positive control wereoocytes expressing AtTIP1;1, which characteristicallyexhibits high water permeability in oocytes (Maurelet al., 1993). McTIP1;2 showed comparable rates ofswelling as AtTIP1;1-injected oocytes (Fig. 1A). Basedon the rate of swelling and the imposed osmoticgradient, we calculated the osmotic water permeabil-ity (Pf) for oocytes expressing McTIP1;2 and AtTIP1;1and compared these values to the Pf calculated fromwater-injected oocytes (Fig. 1B). McTIP1;2-injectedoocytes had a calculated Pf of 13.9 3 1023 cm s21,11-fold higher than that of the water-injected oocytesand a value close to the Pf calculated for AtTIP1;1expressed in oocytes (15.0 3 1023 cm s21; Fig. 1B).These values indicated that McTIP1;2 facilitated thetransport of water. Whether or not it is strictly selectivefor water requires further study.
Ice Plant McTIP1;2 Is Differentially Regulated
by Salt and Osmotic Stress
Previous results with peptide-specific antibodieshad suggested a complex regulation of McTIP1;2(Kirch et al., 2000; Vera-Estrella et al., 2000). Extendingthese studies, we have observed that McTIP1;2 isslightly down-regulated by salt (200 mM) and up-regulated by mannitol (200 mM), sorbitol (200 mM),andABA (25mM; Fig. 2; Vera-Estrella et al., 2000).Underosmotic-stress conditions, in addition to the 34-kDband, a second polypeptide of 41 kD appeared thatcross-reacted with the McTIP1;2 antibody, possiblya glycosylated form of McTIP1;2 (see below).
Membrane Distribution of McTIP1;2 Is Alteredby Osmotic Stress
Suc density gradient separation of microsomalmembranes from untreated cell suspensions showedthat McTIP1;2 cofractionated with markers for the TP
(highest expression in fractions 16–25, correspondingto 13%–20% Suc; Fig. 3, A, B, and G). However, underosmotic stress (mannitol, Fig. 3, A and B; sorbitol orABA; data not shown), McTIP1;2 amounts in TPfractions increased, and it was also detected in frac-tions that overlapped with Golgi (UDPase activity;Fig. 3F), endoplasmic reticulum (ER; calreticulin andcytochrome-c reductase activity; Fig. 3, A, C, and F),prevacuolar compartments (PVC; Pep12; Fig. 3, A andD; Sanderfoot et al., 1998), and the vacuolar-sortingreceptor, BP80, that traffics between clathrin-coatedvesicles, Golgi, and PVC (Fig. 3, A and E; Paris et al.,1997). Interestingly, in the presence of mannitol, BP80expression also increased (Fig. 3, A and E). The peak ofMcTIP1;2 expression was shifted to fractions 25 to 31
Figure 1. Water permeability of McTIP1;2 and AtTIP1;1 expressed inXenopus oocytes. A, Time courses of changes in the relative volumeincreases of Xenopus oocytes injected with 50 ng of the indicatedcRNA or RNase-free water. The relative volume increases weremeasured with reference to initial volume (set as 1) and plotted asa function of time after the introduction of oocytes into 40 mosmol kg21
hypotonic saline solution at time zero. B, Pf values calculated from theinitial rate of oocyte swelling as described in ‘‘Materials and Methods.’’Values are the mean 6 SE (n 5 10).
Aquaporin Regulation under Osmotic Stress
Plant Physiol. Vol. 135, 2004 2319 www.plantphysiol.orgon April 3, 2019 - Published by Downloaded from
(20%–23% Suc), which overlapped closely to the peakfor Golgi markers (Fig. 3, A, B, and F).
Differences in the response of McTIP1;2 to salt andosmotic stress (Figs. 2 and 3A, McTIP1;2-M and -S)suggest that the ionic component of the salt stressinitiates a separate and independent signaling path-way resulting in a decrease in expression of theMcTIP1;2 protein, while osmotic stress, in the absenceof an ionic signal, results in the up-regulation andapparent redistribution of McTIP1;2.
Indirect immunofluorescence microscopy of livingcells gave a clearer picture of McTIP1;2 redistributionunder osmotic stress. In control cells the fluorescenceassociated with McTIP1;2 recognition was presentonly in the TP (Fig. 4A,McTIP1;2-C), showing a similardistribution to that observed for the TP V-PPase (Fig.4A, V-PPase). Upon mannitol treatment, McTIP1;2-associated fluorescence was altered and could be seenin the cytosol, possibly associated with vesicles (Fig.4A, McTIP1;2-M), but not associated with the PM (Fig.4A, McHKT1-M). This distribution was similar tothat observed for BP80, in control (not shown) andmannitol-treated cells (Fig. 4A, BP80-M), and forPep12 (data not shown), confirming results obtainedfrom the Sucdensity gradient separation ofmembranes(Fig. 3). This mannitol-induced membrane distribu-tion was unique for McTIP1;2 as no alterations wereobserved for P-ATPase, V-PPase, McHKT1; PEP12,AtTIP2;1, and calreticulin (Fig. 3, A and B; Fig. 4, Aand B; data not shown).
Double-labeling of cells with McTIP1;2 and BP80antibodies allowed us to define the relationship be-tween the structures carrying both antigens. As ob-served in Figure 4B,McTIP1;2 and BP80 do not appearto colocalize in control cells as McTIP1;2 primarilylabeled the TP, and BP80, cytosolic structures (Fig. 4B,control). In cells treated with mannitol, partial colo-calization of McTIP1;2 and BP80 is indicated as ob-served by the merging of colors in the cytosol (Fig. 4B,mannitol). However, McTIP1;2 was the only antigenthat labeled the spherical vesicular compartmentspresent in mannitol-treated cells (Fig. 4B, mannitol[arrows]). We have so far been unable to determine theorigin of this compartment as it does not label withany of the available markers, including TP (V-PPase,V-ATPase), PM (P-ATPase, McHKT1), protein storagevacuoles (PSV; AtTIP2;1), and PVC (BP80 and Pep12;Fig. 4B; data not shown).
McTIP1;2 Membrane Redistribution Is Organ Specific
and Temporally Regulated
Changes in McTIP1;2 similar to those observed insuspension cells (Fig. 5, cells) were also observed inmembranes from leaves of osmotically stressed plants(Fig. 5, leaves). Whereas membranes from rootsshowed a mannitol-induced increase in McTIP1;2amounts in TP fractions, no change in membranedistribution or appearance of the 41-kD polypeptidewas observed (Fig. 5, roots). This contrasting responsesuggests that the mannitol-induced redistribution ofMcTIP1;2 in the leaves is organ specific and controlled,and not primarily due to an accumulation of newlysynthesized proteins in membranes.
To further investigate mechanisms involved in reg-ulating the distribution of McTIP1;2 during osmoticstress, we determined the time course of the observedchanges. Modifications in protein amount and mem-brane distribution were observed within 2 h of expo-sure to 200 mM mannitol (Fig. 6A), while increases inmRNA levels were observed only in the first h ofmannitol treatment (Fig. 6B) and then remained ata steady level. Over time, McTIP1;2 protein amountswere seen to increase in heavier fractions (peak ob-served in fraction 22), while levels in fraction 13,corresponding to TP, initially increased but subse-quently decreased (Fig. 6A). These data suggest thatthe redistribution of McTIP1;2 is due to posttransla-tional regulation and not the further synthesis ofmRNA or protein.
The 41-kD Polypeptide Is a Glycosylated Formof McTIP1;2
In addition to the observed changes in proteinamount and the changes in membrane distribution ofMcTIP1;2 upon mannitol treatment, the appearance ofa 41-kD protein that also cross-reacted with theMcTIP1;2 antibody was observed (Figs. 2, 3, 5, and 6A). The possibility that this second immunoreactivepolypeptide was a glycosylated form of McTIP1;2 wasinvestigated. Membrane fractions collected from con-tinuous Suc gradients were immunoprecipitated withthe anti-McTIP1;2 antibody, separated by SDS-PAGE,transferred onto nitrocellulose membranes, and sub-sequently stained for the detection of both N-linkedand O-linked oligosaccharides (see ‘‘Materials andMethods’’). Under control conditions, no glycosylationof the 34-kD McTIP1;2 was detected (Fig. 7A, ProQFuchsia); however, immunoprecipitated fractionsfrom the mannitol-treated cells showed glycosylationof the 41-kD polypeptide in all fractions in which theprotein was detected (Fig. 7B, ProQ Fuchsia). Treat-ment with tunicamycin, which blocks the formation ofN-glycosidic protein carbohydrate linkages, was ableto block both the membrane redistribution ofMcTIP1;2and the appearance of the 41-kD glycoprotein (Fig. 7B,McTIP1;2 and ProQ Fuchsia 1 tunicamycin). Incontrast, tunicamycin had no effect on the TP locationof McTIP1;2 in the control-treated cells (Fig. 7A,
Figure 2. Differential expression of McTIP1;2 under salt, osmotic, orABA treatment in TP fractions of ice plant cell suspensions. Cells wereincubated for 5 d with 200 mM of either NaCl (Na), mannitol (M),sorbitol (S), or 25 mM ABA (A) before isolation of TP on a discontinuousSuc density gradient at the 0/16% (w/v) Suc interface. Blots arerepresentative of five independent experiments.
Vera-Estrella et al.
2320 Plant Physiol. Vol. 135, 2004 www.plantphysiol.orgon April 3, 2019 - Published by Downloaded from
McTIP1;2 1 tunicamycin). We have also observed thatthedistribution andexpression ofV-ATPase, P-ATPase,McHKT1, BP80, and calreticulin did not change upontunicamycin treatment (data not shown).
McTIP1;2 Membrane Redistribution Is the Resultof Complex Signaling Mechanisms
To decipher mechanisms involved in the changes inMcTIP1;2 membrane distribution induced by osmoticstress, we used brefeldin A (BFA), a specific inhibitorthat causes disassembly of Golgi complexes and dis-ruption of vesicular trafficking in eukaryotes (Chardinand McCormick, 1999; Ritzenthaler et al., 2002). Pre-incubation of cells with BFA (10 mg/mL) for 30 minwas sufficient to inhibit both the increase in McTIP1;2protein in the TP fractions and the mannitol-induced
membrane redistribution of the AQP (Fig. 8B). Theinhibition of osmotic stress-induced protein increasescould indicate that BFA inhibited the trafficking ofMcTIP1;2 between membranes and affected proteinsynthesis and the normal targeting and delivery ofproteins to their resident membranes. To investigatethis possibility, we followed the expression of markerenzymes under the same treatment. The presence ofBFA did not affect protein levels or distribution of theV-ATPase (VHA-E; E-subunit) or P-ATPase (PMA1),markers for the TP and PM, respectively, pointing toa specific effect of BFA onMcTIP1;2 trafficking (Fig. 8).Further evidence that this was not a general effect onprotein synthesis was obtained by comparing theeffects of BFA with those of the protein synthesisinhibitor cycloheximide. Pretreatment with cyclohex-imide for 30 min, followed by 2 h mannitol treatment,
Figure 3. Expression of McTIP1;2 and marker enzyme colocalization under mannitol stress. Western-blot analysis of microsomalfractions from ice plant cell suspensions. Individual blots (eight lanes/blot) were digitally photographed (Kodak DC-120; Kodak,Mexico City), and then images were aligned and joined using the imaging software PhotoImpact SE 3.01 (Ulead Systems,Torrance, CA) in order to enable visualization of all representative fractions. A, Immunological detection in the respectivefractions of (from top to bottom), calreticulin (57 kD), P-ATPase (PMA-1; 100 kD), V-PPase (PAB-HK; 70 kD), BP80 (80 and 85kD), Pep12 (37 kD), McTIP1;2 (34 kD and 41 kD); C, control, M, mannitol, or S, salt treatment. B, Mean intensity of the bandsfrom McTIP1;2 -control (:), and McTIP1;2 -mannitol (d, 41 kD and h, 34 kD)-treated microsomal fractions. C, Mean intensityof the immunolocalized markers (=) calreticulin, (d) P-ATPase, (n) V-PPase in microsomal fractions. D, Mean intensity of theimmunolocalized marker Pep12 in control (=) and mannitol-treated (n) microsomal fractions. E, Mean intensity of theimmunolocalized marker BP80 in control (d) and mannitol-treated (n, 80 kD and ,, 85 kD) microsomal fractions. F,Distribution of marker enzyme-specific activity in microsomal fractions. (h) NADH cytochrome c reductase and (d) UDPase. G,Suc concentrations in collected fractions. The blots are representative of at least four independent experiments.
Aquaporin Regulation under Osmotic Stress
Plant Physiol. Vol. 135, 2004 2321 www.plantphysiol.orgon April 3, 2019 - Published by Downloaded from
did not result in changes to protein amount. Moreimportantly, redistribution of McTIP1;2 and theV-ATPase or P-ATPasemarkers was not affected by thepresence of this protein synthesis inhibitor. (Fig. 9A;data not shown).
Pulse chase labeling experiments followed by im-munoprecipitation of total protein also confirmed thatthe redistribution of McTIP1;2 was not due to normaltrafficking to the TP of newly synthesized McTIP1;2.Met labeling of the 34-kD McTIP1;2 protein wasobserved after 30 min incubation with mannitol, andno further increase in protein synthesis was detectedafter this time. Rather there was a decrease in theamount of 34-kD McTIP1;2 protein with the subse-quent appearance of the 41-kD glycosylated form ofMcTIP1;2 (Fig. 9B). No Met labeling of the 41-kDglycosylated form of McTIP1;2 was observed in cellsincubated in the absence of mannitol (Fig. 9B).
Spatial and temporal regulation of membranebudding, movement, and fusion can be achievedby localized, rapid changes in phosphoinositidelevels through the action of protein kinases andphosphatases (Simonsen et al., 2001). One of these,
Figure 5. McTIP1;2 membrane redistribution is tissue specific. Rootsand leaves were obtained from plants grown in hydroponic culture5 d after addition of 200 mM mannitol. Cell suspension cultures weretreated for 5 h with 200 mM mannitol. Microsomal membranes fromcontrol, salt-, or mannitol-treated cell suspensions (cells), leaves, androots were separated on continuous Suc density gradients. The blots arerepresentative of at least four independent experiments.
Figure 4. Indirect immunofluorescent labeling of mannitol-treated iceplant cells. A, Cell suspensions were grown for 5 d and then incubatedfor 5 h in the absence (McTIP1;2-C) or presence of 200 mM mannitol(McTIP1;2-M; V-PPase [PAB-HK]; BP80; McHKT1). Cells were pre-pared as described in ‘‘Materials and Methods’’ for immunolocaliza-tion. Cells were immunolabeled with anti-McTIP1;2 antibodies(McTIP1;2-C and McTIP1;2-M) or antibodies directed against specificmembrane markers: V-PPase as a marker for TP; BP80 as a marker ofPVC; and McHKT1 as a marker for PM, followed by labeling with theAlexa fluor 568 secondary antibody. On the merged images Alexa fluor568 immunofluorescence is shown in green and the phase contrastimage (PC) appears in red. B, Double labeling of ice plant cells withMcTIP1;2 and BP80. Cells were prepared as described in ‘‘Materialsand Methods’’ for immunolocalization. Control and mannitol-treatedcells were labeled with McTIP1;2 and BP80. On the merged imagesMcTIP1;2 is shown in red, BP80 in green, the colocalization of bothproteins in yellow, and the vesicular compartments are indicated byarrows. The images are representative of at least four independentexperiments. Bar, 50 mM.
Vera-Estrella et al.
2322 Plant Physiol. Vol. 135, 2004 www.plantphysiol.orgon April 3, 2019 - Published by Downloaded from
phosphatidylinositol 3-phosphate kinase, plays a rolein regulating the availability of phosphatidylinositol3-phosphate. Changes in the level of phosphatidylino-sitol 3-phosphate candisrupt endocyticmembrane traf-ficking (Simonsen et al., 2001). Wortmannin, a specificinhibitor of phosphatidylinositol 3-phosphate kinase(Woscholski et al., 1994), disrupted the mannitol-induced changes in McTIP1;2 distribution; McTIP1;2was exclusively detected in fractions associated withTP (Fig. 10, wortmannin). Disruption of cytoskeletalcomponents, including microtubules or actin micro-filaments, can also affect membrane trafficking.
Cytochalasin D-induced depolymerization of actinfilaments significantly inhibited McTIP1;2 membraneredistribution under osmotic stress (Fig. 10, cytocha-lasin D); however, the McTIP1;2 34-kD polypeptideand the 41-kD glycopeptide were still detected infractions coinciding with TP markers.
In animals, the translocation of AQP2-laden intra-cellular vesicles to the apical PM is triggered by vaso-pressin binding to an adenylyl cyclase (AC)-V2 coupledreceptor, resulting in an increased level of cAMP,which subsequently activates cAMP-dependentprotein kinase A (PKA). PKA phosphorylation ofAQP2 regulates delivery of the protein to the apicalPM (Brown et al., 1998). Sequence analysis of ice plantMcTIP1;2 identified a possible PKA phosphorylationsite motif at the N terminal of the protein (positions4–7). The possibility that similar signaling moleculesto those regulating AQP2 in animals may be involvedin controlling McTIP1;2 membrane redistribution wasinvestigated. Treatment with forskolin, an activator ofAC, known to elicit cAMP-dependent physiologicalresponses by increasing intracellular cAMP (Hanouneand Defer, 2001), and 8-Br cAMP, a cell permeablecAMP analog that activates PKA (Baumgarten et al.,1998), in the absence of mannitol treatment mimickedand enhanced the effect of mannitol-induced osmoticstress on McTIP1;2 membrane redistribution (Fig. 10,forskolin and 8-Br cAMP). In contrast, pretreatmentwith 2#,5#-dideoxyadenosine (2,5-D), an inhibitor ofAC (Hanoune and Defer, 2001), prior to mannitolexposure appeared to restrict the McTIP1;2 protein toTP fractions on the Suc gradient with no protein de-tected in fractions of higher density (Fig. 10, 2,5-D).Direct inhibition of PKA by the specific inhibitorH89 prevented the redistribution of McTIP1;2 protein(Fig. 10, H89).
DISCUSSION
Heterologous expression of AQP in Xenopusoocytes has provided substantial information towardfunctional characterization of these proteins. A num-ber of plant AQP are permeable to water to varyingdegrees. Arabidopsis AtTIP1;1, belonging to the tono-plast intrinsic protein (TIP) subfamily, exhibits one ofthe highest Pf values of those characterized (Maurelet al., 1993). McTIP1;2, which also belongs to thisfamily, has similarly high water permeability (Fig. 1B).However, preliminary data indicate that McTIP1;2does not transport Na1, nitrate, or Zn1 but appearsto be permeable to K1 (data not shown). This maysupport a role for this AQP in osmotic adjustment inplant cells.
Osmotic stress-induced increases in McTIP1;2 pro-tein amount were accompanied by rapid and pro-longed changes in AQP membrane distribution (Figs.3A, 4, A and B, and 5). These observations suggestedinvolvement of membrane trafficking in the posttrans-lational control of McTIP1;2 and presented evidence
Figure 6. Time dependence of mannitol-induced changes in McTIP1;2synthesis and protein redistribution. Ice plant cells were grown for5 d and then incubated for the indicated times with 200 mM mannitol.A, Microsomal fractions were isolated and separated on continuousSuc density gradients. Protein samples were separated by 12.5% SDS-PAGE, transferred to nitrocellulose membranes, and probed with anti-McTIP1;2 antibodies. B, RNA gel-blot analysis of McTIP1;2 andMcHKT1 transcripts (1,250 and 1,907 nucleotides, respectively) hy-bridized to total RNA from cell suspensions treated with 200 mM
mannitol for the times indicated. Ethidium bromide stained gel(bottom) shows equal RNA loading prior to blotting. Blots are repre-sentative of three independent experiments.
Aquaporin Regulation under Osmotic Stress
Plant Physiol. Vol. 135, 2004 2323 www.plantphysiol.orgon April 3, 2019 - Published by Downloaded from
for a novel mechanism in the regulation of AQPactivity under stress conditions in plants.
Membrane trafficking between different compart-ments, including ER, Golgi, TP, and PM can followa number of routes. Aside from the TP of large centralvegetative vacuoles, AQP belonging to the TIP sub-family, which includes McTIP1;2, have been shown toreside on membranes of other vesicles, including PSV,lytic vesicles, autophagic vacuoles, and electron-densevesicles (Jauh et al., 1999). It appears that these TIPdenote their respective compartments in a develop-mentally preprogramed sense with a specific TIP de-
fining a particular vacuole type, although it is alsopossible to detect more than one TIP colocalized to thesame membrane (Jauh et al., 1999). The complexityincreases with the evidence that different stages ofvacuole biogenesis and different types of vacuolarcompartments can be present in a single cell type andchanges can be triggered by environmental conditions(Paris et al., 1996; Marty, 1999). Vesiculation has alsobeen reported to occur in animal cells as a means toprevent cytosolic dilution (Hazelton et al., 2001). Evi-dence also supports several independent pathways forthe biogenesis and maintenance of vacuoles (Jiang andRogers, 1998; Chrispeels and Herman, 2000).McTIP1;2could reside on a novel vacuole-type compartment thatis induced by mannitol stress. Support for this viewcomes from electron microscopy analyses that showedthe appearance of multivesicular bodies in the cyto-plasm of mannitol-treated cells (data not shown). Thelocation ofMcTIP1;2 to a different membranemay helpto maintain osmotic balance within the cytoplasmduring the stress-induced uptake of specific solutes orions into these vesicles. The Arabidopsis AQPAtPIP1;1has been shown to reside on plasmalemmasomes,endosomal vesicles thought to arise as a result of PMinvaginations, whose physiological significance is un-known (Robinson et al., 1996). While the corn (Zeamays) ZmPIP1b-sGFP transgene expressed in tobacco(Nicotiana tabacum) is localized to the PM and theperinuclear region, suggesting that this PIP may betargeted to organelles involved in a secretory pathway(Chaumont et al., 2000). TheMcTIP1;2-associated com-partments in this studymay also be part of an extensiveand complex endomembrane system capable of seg-menting and insertingmembraneswhere needed at theonset of the osmotic-stress response (Hazelton et al.,2001).
Vesicular trafficking of proteins involves a complexsystem that ensures the correct sequence of eventsleading to the final step of docking of the protein withits target membrane. Any signal transduction pathwaythat would relay the information and trigger relocationof the proteins will add another level of complexity.Disrupting components involved in protein sortingand targeting provided a way for deciphering thepathway involved in the redistribution of McTIP1;2.While transport ofMcTIP1;2 (and other markers) to theTP was unaffected by BFA, a complete inhibition of themannitol-induced redistribution of McTIP1;2 was ob-served (Fig. 8). Previous results have already shownthat the sorting of TIPs to the TP is insensitive to BFA(Gomez and Chrispeels, 1993). This indicates that theosmotic stress-induced redistribution of McTIP1;2 fol-lows a route that is independent of the normal deliveryof this protein to the TP and, moreover, that it is BFAsensitive. This further suggests the requirement forindependent functional vesicular sorting machineryinvolved in protein targeting to other endomembranesunder osmotic-stress conditions. McTIP1;2 redistri-bution under osmotic stress was also specificallyperturbed by the presence of wortmannin (Fig. 10),
Figure 7. Glycosylation of McTIP1;2 in mannitol-treated cells isnecessary for membrane redistribution. Microsomal proteins fromcontrol (A) or mannitol-treated (B) cells were layered onto continuousSuc density gradients, and fractions were either prepared directly forSDS-PAGE or immunoprecipitated with the anti-McTIP1;2 antibodiesas described in ‘‘Materials and Methods.’’ Samples were then resus-pended in Laemmli (1970) sample buffer, separated by SDS-PAGE, andtransferred to nitrocellulose membranes. Blots of SDS-PAGE-separatedmicrosomal proteins were used for immunological detection ofMcTIP1;2 (McTIP1;2 2/1 tunicamycin) as described in ‘‘Materialsand Methods.’’ Blots of immunoprecipitated protein were used forglycoprotein staining with the ProQ Fuchsia kit (ProQ Fuchsia 2/1tunicamycin) as described in ‘‘Materials and Methods’’. Tunicamycin(25 mM) was added to cells 30 min prior to the 12-h incubation periodin the absence or presence of mannitol. Blots are representative of threeindependent experiments.
Vera-Estrella et al.
2324 Plant Physiol. Vol. 135, 2004 www.plantphysiol.orgon April 3, 2019 - Published by Downloaded from
providing further evidence for two independent path-ways involved in McTIP1;2 delivery to the TP andMcTIP1;2 trafficking under osmotic stress. Matsuokaet al. (1995) demonstrated, using wortmannin, theexistence of distinct routes for sorting of soluble pro-teins to the vacuole, distinguished by their relative sen-sitivities to wortmannin. While we do not suggest thatthese pathways might direct McTIP1;2, our resultsindicate that phosphoinositide regulation of vesicularorganization can show specificity for subsets of trans-port vesicles.Disassembly of the actin cytoskeleton by cytochala-
sin D disrupted the redistribution of McTIP1;2 underosmotic stress (Fig. 10), but not the appearance of the41-kD polypeptide, suggesting that even though theglycosylated form is present, a functional cytoskeletonis essential for redistribution. This in part resemblesvesicular trafficking of mammalian AQP2 and AQP5,which has been shown to be dependent on actinmicrofilaments and microtubules driving the sortingof theAQP-containingvesicles to their targetmembrane(Brown et al., 1998; Ishikawa et al., 1999). Moreover,AQP2 directly binds to actin in vitro, and drugs thatdisrupt cytoskeletal tracks have long been known toinhibit the hormonally induced change in permeabil-ity of the apical PM of renal epithelial cells (Brown
et al., 1998; Ishikawa et al., 1999). Rearrangement of theactin network occurs after hormonal exposure in thesecells (Brown et al., 1998). In Chara, the specific sensi-tivity of endo-osmotic water flow to cytochalasins Band E provided early evidence for actin-mediatedevents that may be involved in regulating the locali-zation and/or activity of plant AQP under certainconditions (Wayne and Tazawa, 1988).
cAMP-regulated signaling has been widely estab-lished in both eukaryotes and prokaryotes; however,the acceptance of its role in plants has been contentious,with only recent compelling evidence provided bymore sensitive and accurate detection techniques (forreview, see Newton et al., 1999) and the cloning ofa putative AC in pollen, which shows homology tofungal ACs (Moutinho et al., 2001). Participation ofa cAMP-signaling pathway in the redistribution ofMcTIP1;2 under osmotic stress is suggested by ourresults. Inhibition of either cAMP synthesis or PKAactivity, by 2,5-D, and H89, respectively, preventedMcTIP1;2 redistribution (Fig. 10). In contrast, stimula-tion of the pathway by either exogenous activation of
Figure 9. The redistribution of McTIP1;2 is not a general effect ofprotein synthesis. A, Microsomal proteins from cells treated for 30 minwith cycloheximide followed by 2 h with 200 mM mannitol treatmentwere isolated and separated on continuous Suc density gradients. Theblot is representative of three independent experiments. B, In vivopulse-chase labeling with 35S Met/Cys for 1 h followed by incubationwith (upper blot) or without (lower blot) 200 mM mannitol and bya chase with unlabeled Met and Cys for an additional 0.5 to 5 h. At theend of the pulse or chase equal amounts of cells were collected. Timezero refers to sampling of cells before the addition of Met. Total cellprotein was extracted, immunoprecipitated with anti-McTIP1;2 anti-bodies, and proteins separated on a 12.5% SDS-PAGE gel. Labeledproteins were detected with a PhosphorImager.
Figure 8. BFA inhibits the changes in McTIP1;2 membrane distributioninduced by mannitol treatment of ice plant cells. Microsomal fractionsof ice plant cell suspensions were separated on continuous Suc densitygradients and probed with antibodies for (from top to bottom)McTIP1;2, V-ATPase E-subunit (VHA-E), and P-ATPase (PMA1). A, Cellsincubated with 200 mM mannitol for 2 h. B, Cells incubated for 30 minwith 10 mg/mL BFA prior to incubation with 200 mM mannitol for 2 h.Blots are representative of five independent experiments.
Aquaporin Regulation under Osmotic Stress
Plant Physiol. Vol. 135, 2004 2325 www.plantphysiol.orgon April 3, 2019 - Published by Downloaded from
an AC (forskolin) or the use of a cAMP analog (8-BrcAMP) in the absence of mannitol treatment mimickedthe osmotically induced changes inMcTIP1;2 distribu-tion (Fig. 10). A role for cAMP signaling has previouslybeen postulated for activation via phosphorylation ofa plantAQP.Animal PKAwas shown tophosphorylatePvTIP3;1 (a-TIP) from kidney bean (Phaseolus vulgaris)in vitro, and increases in cAMP brought about byextrinsic activation of AC specifically increased thewater transport activity of PvTIP3;1 when expressed inXenopus oocytes (Maurel et al., 1995). However, fromthese results it could not be decided whether phos-phorylation by PKA directly affected the gating of thechannel or, alternatively, enhanced targeting of theAQP to the membrane.
Two putative N-linked glycosylation sites are pres-ent in McTIP1;2. The motifs, NISG and NISL, arelocated at positions 78 to 81 and 97 to 100, respectively,which correspond to cytosolic loops. It appears thatglycosylation is essential for McTIP1;2 redistributionunder mannitol treatment as inhibition of glycosyla-tion blocks both the redistribution of the 34-kDMcTIP1;2 polypeptide and the appearance of the41-kD glycopeptide (Fig. 7B,McTIP1;21 tunicamycin).
In eukaryotes, it has been suggested that N-linkedoligosaccharides contain specific targeting informa-tion, and in plants, they have been shown to influenceconformation, stability, and biological activity of pro-teins (Lerouge et al., 1998). Once the glycosylated formof McTIP1;2 is localized to the new compartments, itappears to be deglycosylated, as both the glycosylatedand nonglycosylated forms are observed in fractionsfrom the mannitol-treated cells (Fig. 7B).
It has been reported that tunicamycin may cause themisfolding of glycoproteins. However, many reportshave established that this drug has no effect on thefolding of a large number of glycoproteins and doesnot affect the folding of nonglycosylated proteins(Ceriotti et al., 1998). According to Sparvoli et al.(2000), tunicamycin treatment can affect plant glyco-proteins in different ways depending upon the proteinand the cell type involved. In our case, if McTIP1;2misfolding was occurring as a result of tunicamycintreatment, we would expect an increase of this proteinin fractions corresponding to ER as a consequence oftheir retention and accumulation in this organelle. Thisshould be clearly visualized using western-blot anal-yses of microsomal proteins after tunicamycin treat-ment. However, the fractions which correspond to ER(Fig. 7B; fractions 31–40) show no such accumulationof the 34-kD protein or any other protein that couldrepresent the deglycosylated form of the 41-kD glyco-protein, if it is notMcTIP1;2, but another AQP isoform(Fig. 7B). Rather, we see an increase in the unglycosy-lated McTIP1;2 34-kD polypeptide in the TP-associ-ated fractions.
The existence of water channels has challengedviews about plant water relations (Tyerman et al.,1999), but the large size of the AQP protein family maybe even more puzzling (Santoni et al., 2000; Chaumontet al., 2001; Quigley et al., 2002). Judged by family size,apparently, plants need AQP in many membranes, orin different tissues in a complex pattern, or at differenttimes under different external conditions. While weknow that they transport water only from oocyteexperiments, the list of metabolites that pass throughAQP is growing (Tyerman et al., 2002; Liu et al., 2003;Uehlein et al., 2003). The complexity of AQP expres-sion, from the control of transcript amount to post-translational modification by phosphorylation andglycosylation, is beginning to emerge. We add anothercomponent, the redistribution of a TPAQP dependingon environmental conditions, to the list of mechanismsthat guide the presence and functioning of AQP. Theexact nature of the compartment to which McTIP1;2 isredistributed upon mannitol treatment (Fig. 4, A andB) is still unclear, due to the complexity of vacuolarpathways in plant cells. However, our results indicatethat upon mannitol treatment,McTIP1;2 may reside inseveral membrane compartments (including a BP80-labeled compartment and a unique multi-vesicularcompartment; Fig. 4, A and B) and that McTIP1;2redistribution depends on posttranslational modifica-tions including glycosylation.
Figure 10. Mannitol-induced changes in membrane distribution ofMcTIP1;2 is the result of complex signaling mechanisms. Ice plant cellswere preincubated in the presence of either 10 nM wortmannin, 1.5 mM
cytochalasin D, 6 mM 2,5-D, or 48 nM H89 for 30 min, followed by a 2-hincubation with 200 mM mannitol (blots 3–6). Ice plant cells incubatedfor 2 h with either 8 mM forskolin or 40 mM 8-bromoadenosine-3#,5#-cyclic-monophosphate (8-Br cAMP) in the absence of mannitol treat-ment (blots 7 and 8). Untreated cells (control, blot 1) and cells treatedfor 2 h with 200 mM mannitol (mannitol, blot 2) were included ascontrols. Blots are representative of four independent experiments.
Vera-Estrella et al.
2326 Plant Physiol. Vol. 135, 2004 www.plantphysiol.orgon April 3, 2019 - Published by Downloaded from