Tyrosine Kinase Signal Modulation: A Matter of H₂O₂ Membrane Permeability?

Abstract

H₂O₂ produced by extracellular NADPH oxidases regulates tyrosine kinase signaling inhibiting phosphatases. How does it cross the membrane to reach its cytosolic targets? Silencing aquaporin-8 (AQP8), but not AQP3 or AQP4, inhibited H₂O₂ entry into HeLa cells. Re-expression of AQP8 with silencing-resistant vectors rescued H₂O₂ transport, whereas a C173A-AQP8 mutant failed to do so. Lowering AQP8 levels affected H₂O₂ entry into the endoplasmic reticulum, but not into mitochondria. AQP8 silencing also inhibited the H₂O₂ spikes and phosphorylation of downstream proteins induced by epidermal growth factor. These observations lead to the hypothesis that H₂O₂ does not freely diffuse across the plasma membrane and AQP8 and other H₂O₂ transporters are potential targets for manipulating key signaling pathways in cancer and degenerative diseases. Antioxid. Redox Signal. 19, 1447–1451.

Introduction

Reactive oxygen species can be highly cytotoxic. At low concentrations, however, they are essential for cell function and survival. H₂O₂ inhibits phosphatases and can activate certain kinases (4). How it can diffuse in and between cells is, thus, a fundamental question in pathophysiology. Many pathways can generate H₂O₂: NADPH oxidases in the plasma membrane, oxidative phosphorylation in mitochondria, and oxidative protein folding in the endoplasmic reticulum (ER) (5). In all cases, H₂O₂ must cross a membrane to reach its cytosolic targets. H₂O₂ has long been thought to cross lipid bilayers freely. However, the slightly larger dipole moment of H₂O₂ makes its simple diffusion through membranes less likely than for water. Its size and electrochemical properties qualify H₂O₂ as a possible substrate of aquaporins (AQP). In fact, some members of the AQP family (8) have been shown to transport H₂O₂ (2, 6). In this study, we investigate the intracellular transport of H₂O₂ exploiting a panel of organelle-targeted ratiometric HyPer sensors (1, 3).

Innovation.

Our observations that efficient transport across the plasma and endoplasmic reticulum membranes requires AQP8, or other membrane proteins, introduce a novel level of regulation in redox signaling. Indeed, in our cell model, AQP8 silencing inhibited not only epidermal growth factor -induced entry of H₂O₂, but also downstream tyrosine phosphorylation of target proteins. Thus, H₂O₂ transporters may be potential targets in cancer and degenerative diseases.

Results

To investigate H₂O₂ transport across the plasma membrane, we expressed a cytosolic HyPer sensor (1) into HeLa cells. Addition of 50 μM H₂O₂ clearly activated the sensor, as determined by the profound 488/405 nm shifts observed in live cell imaging (Fig. 1A). A C199S HyPer mutant that has lost redox sensitivity did not shift its fluorescence emission, excluding pH-dependent events. Pretreatment with HgCl₂, a compound that inhibits many membrane transporters, prevented HyPer oxidation (data not shown). Given the poor specificity of HgCl₂, we performed silencing experiments targeting members of the AQP family, membrane channels involved in the transport of water and other solutes (8), including H₂O₂ (2, 6). Silencing AQP8 efficiently prevented H₂O₂ entry (Fig. 1B, blue trace), suggesting that this protein plays a major role in our cell model. Transport of H₂O₂ was rescued by expression of a Halo-AQP8 chimeric protein, in which the third bases of triplets 12–16 were mutated (red trace). In this way, transgene transcripts were not targeted by silencing oligonucleotides and yet encoded the same amino acids. A similarly engineered C173A-AQP8 mutant was inactive, excluding off-target effects (green trace). Having confirmed that the 488/405 nm ratio remained constant independently from the sensor expression levels (Fig. 1C), we analyzed cells at the population level using single wavelength-automated Typhoon imaging (see Notes). Measuring the increase in 488 nm fluorescence in wells containing >10⁵ cells confirmed that AQP8 promotes H₂O₂ membrane transport. In contrast, silencing AQP3 or AQP4 had only marginal effects (Fig. 1D). The above experiments provided additional evidence that H₂O₂ cannot freely permeate through the plasma membrane and identified AQP8 as an efficient transporter.

FIG. 1.

Entry of exogenous H₂O₂ into HeLa cells requires AQP8. (A) Time course response of wt cytosolic HyPer (red line) or a redox-insensitive mutant (C199S, blue line) to 50 μM H₂O₂. (B) Kinetics of H₂O₂-dependent HyPerCyto activation in HeLa cells silenced for AQP8 expression (8i), and then transfected with silencing-resistant constructs driving the expression of wt (8i+wt) or mutant AQP8 (8i+C173A, see Methods for details). Graphs in A and B show mean fold change±SEM of the 488/405 nm ratio measured by confocal laser scanning. (C) The HyPer 488/405 nm ratio is not influenced by the expression levels of the probe. Transiently transfected HeLa cells were analyzed by confocal laser scanning. Despite cells being varied in the extent of probe expression, the ratio remained rather constant, as demonstrated by the linear correlation of 405 and 488 nm light excitation (r=0.92). (A.U. stands for Arbitrary Units). (D) H₂O₂-dependent activation of HyPerCyto was analyzed in HeLa cells treated with specific silencing oligos for AQP8, AQP4, or AQP3 by Typhoon scanning (see Methods). Data are expressed as fold increase in the 488 nm fluorescence±SEM, relative to cells that were not treated with H₂O₂. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

Next, we investigated whether AQP8 is needed for H₂O₂ transport across the mitochondria and ER membranes. Since the silencing of this protein prevented H₂O₂ passage across the plasma membrane (Fig. 1B), we treated cells expressing specific sensors in mitochondria or ER (HyPerMito or HyPerERlum) with digitonin, a detergent that permeabilizes the cholesterol-enriched plasma membrane without significantly affecting the integrity of intracellular membranes (7), and followed the import of added H₂O₂. AQP8 silencing perturbed H₂O₂ transport into the ER lumen, but had no significant effects on mitochondria (Fig. 2). The causes and potential physiological significance of the delayed H₂O₂ entry into the ER of AQP8-silenced cells are presently unclear. Taken together, the above observations suggest that AQP8 is important for H₂O₂ to cross the plasma and ER membranes, but dispensable for mitochondrial import.

FIG. 2.

H₂O₂ transport into the endoplasmic reticulum (ER) is facilitated by AQP8. HeLa cells expressing HyPerERLum (A) or HyperMito (B) were silenced with AQP8-specific (blue line) or irrelevant (red line) oligos before semipermeabilization with 40 μg/ml digitonin (7) to bypass the plasma membrane and allow access to ER and mitochondria. The kinetics of H₂O₂ entry into the two organelles were measured by confocal laser scanning. Graphs show mean fluorescence±SEM. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

Epidermal growth factor (EGF)-induced H₂O₂ inhibits phosphatases through sulfenylation of their active-site cysteines, potentiating kinase signaling (4). Accordingly, EGF induced a transient spike of cytosolic H₂O₂ in our cells (Fig. 3A, red trace). This increase was strongly diminished in AQP8-silenced cells (blue trace), suggesting that part of the H₂O₂ measured derives from extracellular sources. Hence, when catalase was added to the cell medium before EGF stimulation, the intensity of bands corresponding to tyrosine-phosphorylated proteins was decreased (Fig. 3B, C). Silencing AQP8 expression diminished tyrosine phosphorylation to a level similar to that obtained with catalase. In this way, our results corroborate that, during EGF signaling, AQP8 allows entry of exogenously generated H₂O₂, amplifying signal transduction.

FIG. 3.

Epidermal growth factor (EGF) induces transport of H₂O₂ through AQP8. (A) Time course analysis of H₂O₂ uptake by HeLa cells expressing HyPerCyto by confocal laser scanning after treatment with 10 μg/ml EGF. Color code as in Figure 2; (B). Western blot analysis with the indicated antibodies showing changes in phosphorylation of total cell tyrosines on whole HeLa cell extracts that were either silenced for AQP8 (8i) or treated with catalase before treatment with EGF. (C) The intensity of protein phosphotyrosine signals in the blots shown in B were quantified by densitometry and normalized to Tubulin. Bars show average band intensities relative to untreated control cells±SEM. (D) A schematic view of H₂O₂ generation and transport during EGF signaling. NOX and DuOX enzymes on the cell surface are activated to produce H₂O₂. As demonstrated by the inhibitory effects of catalase, extracellularly generated H₂O₂ is important for efficient amplification of signaling and enters the cell through AQP8, as indicated by silencing (B, C). The residual activity in silenced cells could reflect incomplete AQP8 downregulation, the presence of additional H₂O₂ transporters in the plasma membrane and/or its generation in mitochondria. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

Discussion

Our results indicate that H₂O₂ cannot freely diffuse through the plasma and ER membranes and identify AQP8 as an efficient transporter. In our HeLa cell model, entry of exogenous H₂O₂ is largely sustained by AQP8. The efficient rescue of H₂O₂ entry by AQP8 re-expression excludes off-target effects. Presumably, AQP8 directly transports H₂O₂; a less likely possibility is that it controls the expression and/or activity of other transporter(s). Silencing AQP3, another family member thought to mediate H₂O₂ transport (6), had minor effects, possibly reflecting lower expression levels in our cells. If both AQP3 and AQP8 can allow H₂O₂ entry into cells, a differential tissue expression of the two molecules may reflect other metabolic requirements, like glycerol or ammonia transport (8).

Since AQP8 folds in the ER, it is not surprising that downregulation of this channel impacts H₂O₂ transport across the membrane of this organelle. The observation that mitochondrial entry was not affected reinforces the view that no functional AQP8 resides in this organelle (9). It will be of interest to determine whether other H₂O₂ transporters be present in mitochondrial membranes. The above observations have important functional implications, since both the ER and mitochondria can be sources of abundant H₂O₂. During intense metabolic activity or hypoxia, generation of H₂O₂ from electron leakage in mitochondria (4) might act as a signaling rheostat. Likewise, it is conceivable that exuberant disulfide bond formation in the ER could modify cytosolic H₂O₂ levels with anti- or proapoptotic effects (5). The observation that devoted transporters are needed also for inter-organellar transport delineates yet another possible level at which eukaryotes can exert homeostatic control.

Elegant work from many laboratories demonstrated that H₂O₂ increases the strength of tyrosine kinase signaling [(4) and references therein]. AQP8 silencing strongly reduced the H₂O₂ fluxes observed soon upon EGF stimulation. Importantly, also, downstream tyrosine phosphorylation was inhibited. AQP8 silencing was almost as efficient as extracellular catalase, confirming that most physiologically EGF-related H₂O₂ entered through the AQP8 channels. Thus, the regulation of H₂O₂ entry could be an additional means to tune signal transduction (Fig. 3D).

Methods

Cell culture, plasmids, siRNAs, and transfection

HyPer sensors targeted to the cytosol (Cyto) or mitochondria (Mito) and the HyPerC199S mutant (5) were generous gifts of Dr. V. Belousov (IBCh, Moscow, Russian Federation), while HyPer ER luminal (ERLum) plasmid (3) was kindly provided by Drs. E. Margittai and M. Geistz (Semmelweis University, Budapest, Hungary). Human AQP8 cDNA was amplified from pCMV6AC-GFP-AQP8 (NM_001169; Origene) and cloned into HaloTag^® vector pHTN (Promega). Silencing-resistant Halo-AQP8 was generated by introducing five silent point mutations in the 21 bp AQP8-siRNA target sequence, to obtain 5′ TTCGGGAACGATAAA 3′, and used to create the C173A mutant. All constructs and mutations were validated by sequencing. Reagents to silence AQP3 (5′ GGG UCGUCACUCCUUUAAUU 3′), AQP4 (5′ GAUCAGCAU CGCCAAGUCU 3′), and AQP8 (5′ UUUGGCAAUGACA AGGCCA 3′), and an unrelated control (Block-it™) were purchased from Ambion (Life Technologies).

Two or 10×10⁴ HeLa cells were grown overnight in 24- or 6-well plates, transfected with 18 or 90 pmol of siRNA, respectively, using RNAiMAX lipofectamin (Life Technologies) and analyzed after 72 h. Transient transfections were performed by Polyethylenimine as described (7).

Imaging HyPer-oxidation

Cells were equilibrated in Ringer buffer (140 mM NaCl, 2 mM CaCl₂, 1 mM MgSO₄, 1.5 mM K₂HPO₄, 10 mM glucose, pH 7.3) for 15 min at RT and images taken using a Typhoon FLA 9000 fluorescence scanner (GE Healthcare) equipped with a 488 nm excitation filter 0, 3, 6, or 9 min after treatment with or without 50 μM H₂O₂ (Sigma-Aldrich). Untransfected HeLa cells were used to calibrate laser sensitivity. Images were analyzed using ImageJ and results presented as fold change increase in 488 nm intensity relative to control cells.

HeLa transfectants were seeded on glass coverslips. Forty-eight hours after transfection, coverslips were equilibrated in Ringer buffer for 10 min at RT before H₂O₂ exposure or stimulation with 10 μg/ml of EGF (Sigma-Aldrich) at 37°C. Confocal images were collected every 2 s for 3 min by excitation with 488 nm argon and 405 nm violet diode lasers on an Ultraview confocal laser scanning microscope using a 40× oil-immersion lens (Perkin Elmer). The 488/405 nm ratios were calculated by ImageJ software in triplicate wells, always averaging in each ≥10 cells, and plotted against time. HaloAQP8-expressing cells were identified by labeling with 5 μM HaloTag TMR Ligand (Promega) and analyzed at 532 nm.

To measure intracellular H₂O₂ fluxes, transfectants expressing HyPerMito or HyPerERLum were treated and silenced as above. Coverslips were placed in a live cell chamber at RT, equilibrated in Ringer Buffer for 5 min, and then treated for 5 min with 40 μg/ml of digitonin (Sigma-Aldrich). The integrity of the ER and mitochondrial membranes was assessed as described previously (7).

Antibodies and western blotting

Mouse anti-PhosphoTyrosine (4G10), mouse anti-tubulin, goat anti-mouse IgG-HRP, or AlexaFluor-647 were from Millipore, Sigma-Aldrich, Zymed, and Invitrogen, respectively. Images were acquired by Chemidoc-it Imaging System (UVP) or Typhoon FLA 9000, processed with ImageJ and densitometrically quantified by ImageQuant 5.2 (Molecular Dynamic).

To follow EGF responses biochemically, subconfluent HeLa cells were cultured in the FCS-free DMEM for 3 h before incubation with or without 5000 U/ml of catalase and EGF addition. Cells were then washed with ice-cold PBS containing 0.4 mM Na₃VO₄ to block phosphatases, lysed in sample buffer and analyzed by electrophoresis and western blot.

Further details will be made available upon request (milena.bertolotti@hsr.it)

Abbreviations Used

AQP

aquaporins

EGF

epidermal growth factor

ER

endoplasmic reticulum

SEM

standard error of the mean

Acknowledgments

We apologize to several colleagues whose seminal work could not be cited due space limitations. We thank Drs. V. Belousov, E. Margittai, M. Geistz, T. Dick, and M. Toledano for providing reagents and/or advice, Claudio Fagioli for technical help, and Cesare Covino and colleagues at the ALEMBIC facility of University Vita-Salute San Raffaele Scientific Institute (Milan) for assistance in imaging. This work was supported by grants from Associazione Italiana Ricerca Cancro (AIRC; IG and 5×1000 programs), Ministero della Salute, Regione Lombardia (ASTIL), and Telethon (GGP11077). The authors declare no conflict of interests.