The pHi (intracellular pH) is an important physiological parameter which is altered during hypoxia and ischaemia, pathological conditions accompanied by a dramatic decrease in pHi. Sensors of pHi include ion transport systems which control intracellular Ca2+ gradients and link changes in pHi to functions as diverse as proliferation and apoptosis. The annexins are a protein family characterized by Ca2+-dependent interactions with cellular membranes. Additionally, in vitro evidence points to the existence of pH-dependent, Ca2+-independent membrane association of several annexins. We show that hypoxia promotes the interaction of the recombinant annexin A2–S100A10 (p11) and annexin A6 with the plasma membrane. We have investigated in vivo the influence of the pHi on the membrane association of human annexins A1, A2, A4, A5 and A6 tagged with fluorescent proteins, and characterized this interaction for endogenous annexins present in smooth muscle and HEK (human embryonic kidney)-293 cells biochemically and by immunofluorescence microscopy. Our results show that annexin A6 and the heterotetramer A2–S100A10 (but not annexins A1, A4 and A5) interact independently of Ca2+ with the plasma membrane at pH 6.2 and 6.6. The dimerization of annexin A2 within the annexin A2–S100A10 complex is essential for the pH-dependent membrane interaction at this pH range. The pH-induced membrane binding of annexins A6 and A2–S100A10 might have consequences for their functions as membrane organizers and channel modulators.
Biological systems are finely tuned and optimized to function within tightly constrained pH ranges. Changes in cytoplasmic H+ concentration control the activities of many enzymes, transporters and ion channels, and influence such processes as muscle contraction, post-translational processing of secretory proteins, cleavage of pro-hormones  and ultimately cell survival and apoptosis . Cellular pH homoeostasis and pHi (intracellular pH)-dependent control of cellular functions involve pHi sensors, and cells have developed multiple parallel mechanisms for pH regulation. The main acid/base transporters of mammalian cells are vacuolar H+-ATPases, Na+/H+ exchangers, lactate−/H+ symporters and HCO3− transporters . This redundancy underlines the importance of pH homoeostasis for cellular function.
pHi values, which in the resting cells are in the neutral range (7.0–7.2), can change rapidly during muscle contraction: intramuscular pH declines from 7.1 to 6.5 due to skeletal muscle fatigue . Hypoxia and ischaemia are accompanied by a dramatic drop of pHi. This decrease probably results from an immediate blockade of oxidative phosphorylation and activation of glycolysis . In cancer, the expansion of tumours progressively distances the cells from the vasculature, and thus from oxygen and nutrients. The pHi of tumours becomes acidic as a result of lactate and carbonic acid accumulation, and the cells counteract this by activation of acid extrusion mechanisms, such as the growth factor and amiloride-sensitive Na+/H+ exchanger and carbonic anhydrase, which converts CO2 into carbonic acid. Uptake of carbonic acid contributes to the intracellular alkalinization, critical for cancer cell survival, and ensures an acidic extracellular milieu typical of the malignant phenotype . Importantly, the expression of many proteins responsible for pH homoeostasis in tumour cells is controlled by HIFs (hypoxia-inducible factors), therefore the new approaches in cancer therapy target tumour metabolism and pHi-control systems .
Ca2+ is an important co-mediator of pH-induced effects in the cells. Cellular pHi sensors include a variety of ion transport systems, which control intracellular Ca2+ gradients and might link changes in pHi to cellular functions, including long-term effects such as cell proliferation and apoptosis. The annexins are a multigene family of Ca2+- and charged phospholipid-binding proteins, which have been implicated in numerous Ca2+-regulated processes inside the cell. Their carboxy-terminal cores are evolutionarily conserved and contain Ca2+-binding sites, whereas their amino-terminal tails are unique and enable the proteins to interact with distinct cytoplasmic partners. Eukaryotic cells express several annexins which are involved in the regulation of multiple functions within a single cell type . Previously, using biochemical methods and imaging in live cells, we have shown that four different annexins target membrane sites of distinct lipid composition, each at a different [Ca2+]i (intracellular calcium concentration) [8,9]. Given the fact that several annexins are present within any one cell, it is likely that they form a sophisticated [Ca2+]i sensing system, with a regulatory influence on other signalling pathways.
Parallel to these observations, there is evidence of Ca2+ -independent membrane association of several annexins [10–12] that points to the existence of pH-dependent binding mechanisms, most likely mediated by hydrophobic interactions. Isas et al.  report that human annexin A5 and hydra annexin XII are able to bind phospholipid vesicles in vitro at pH 5.3 and 5.8 respectively. Similar interactions were described for hydra annexin B12 [11,14] and human annexin A6 [10,15]. It should be noted that these studies were performed in vitro at pH values ranging from 4 to 6, i.e. under conditions which rarely occur in living cells. However, the observations that annexin A6 interacts with membranes of late endosomes  and annexin A5 induces pH-dependent vesicle aggregation and fusion  point to the fact that the local acidic microenvironment of certain intracellular compartments allows these Ca+-independent membrane interactions to occur in vivo.
We investigated the influence of a hypoxia-induced decrease of pHi on the membrane binding of fluorescent protein-labelled human annexins. We studied the pH-dependent localization of annexins A1, A2, A4, A5 and A6 in living cells and characterized their membrane interaction biochemically in a microsome binding assay. Here we demonstrate for the first time that annexin A6 and tetrameric A2–S100A10 (p11) interact with plasma membranes of living cells upon intracellular acidification independently of the [Ca2+]i and that dimerization of annexin A2 within the annexin A2–S100A10 heterotetramer is essential for pH-dependent membrane interaction. Submitting the cultures to hypoxic conditions caused intracellular acidification and consequently promoted the interaction of fluorescently-labelled annexin A2–S100A10 and A6 with the plasma membrane.
Reagents and antibodies
For Western blotting and immunofluorescence, monoclonal antibodies against annexins and S100A10 were purchased from Transduction Laboratories and Santa Cruz Biotechnology. Anti-GFP (green fluorescent protein) monoclonal antibody was from Clontech. Chicken anti HIF-1α antibody was a gift from Prof. M. Gassmann (University of Zurich, Zurich, Switzerland) and rabbit polyclonal anti-Sp1 antibodies were from Santa Cruz. Unless otherwise indicated, routine chemicals were from Sigma. Restriction endonucleases, Taq polymerase and T4 DNA ligase were purchased from New England Biolabs. Fluo-3/AM (Fluo-3 acetoxymethyl ester) carboxy SNARF-4F/AM (seminaphthar-hodafluor-4F acetoxymethyl ester), nigericin and BAPTA/AM [1,2-bis-(o-aminophenoxy)ethane-N,N,N′,N′-tetra-acetic acid tetrakis(acetoxymethyl ester)] were from Molecular Probes.
Tissue preparation and immunofluorescence
Human urinary bladder smooth muscle tissue was prepared as described previously . For treatment at low pH, the solution was adjusted to pH 5.0 and 100 μM BAPTA. The muscle was fixed in 4% paraformaldehyde (at pH 7.4 or 5.0) and processed as described previously . For the visualization of endogenous annexins, HEK (human embryonic kidney) 293 cells were seeded onto glass coverslips, washed with Ca2+-free Na+-Tyrode's buffer, incubated at pH 7.4 or pH 5 for 15 min, fixed and stained with antibodies as described previously .
Expression of annexins as fusions with fluorescent proteins
The coding sequences of annexins A1, A2, A4, A5 and A6 as well as S100A10 were cloned into the Living Colors Fluorescent protein vectors pEGFP-N1, pECFP-N1 and pEYFP-N1 (where CFP, cyan fluorescent protein and YFP, yellow fluorescent protein) (Clontech) following the PCR as described previously . A2core-YFP fusion protein was amplified from the full-length annexin A2 sequence using the forward primer 5′-ATAAAGCTAGCATGTTTGATGCTGAGCGGGATGC-3′ corresponding to nt 145 of the annexin A2 cDNA and containing NheI cloning site (underlined); and the reverse primer 5′-ATAAACTCGAGCCAGATCTGTCATCTCCACCACACAGGTA-3′ containing BglII and XhoI cloning sites (underlined) for insertion into pEYFP-N1, digested with NheI and XhoI. For tandem attachment of the additional copies of the A2 core sequence, the core was amplified using the same reverse primer, and a forward primer containing a BamHI site, and the PCR product inserted into pEYFP-A2core vector, digested with BglII and XhoI. Dimerization of the A4 sequence was performed similarly using a pECFP-N1 vector. Following PCR amplification of the A4 coding sequence using the forward primer 5′-AAATAAGATCTCATGGCCATGGCAACCAAAG-3′ and the reverse primer 5′-TAAATAAGCTTATCATCTCCTCCACAGAGAACA-3′, it was inserted in-frame into the pEN1A4-CFP vector digested with BglII and HindIII (restriction sites underlined).
Cell culture and transfections
HEK293 cells were maintained in DMEM (Dulbecco's modified Eagle's medium), and HepG2 (human hepatocellular carcinoma) cells in RPMI medium, containing 2 mM glutamine, 100 units penicillin/ml, 100 μg streptomycin/ml and 10% foetal calf serum. All cells were grown in 5% CO2 at 37 °C in a humidified incubator. HEK 293 and HepG2 cells were transiently transfected with plasmids using electroporation (BioRad) and analysed after incubating at 37 °C for 48 h.
Live cell imaging and intracellular calcium measurements
Live cell imaging and calcium imaging recordings were performed as described in . Cells were observed under an Axiovert 200 M microscope with laser scanning module LSM 510 META (Zeiss) using a ×63 oil immersion lens. Images were acquired at a scanning speed of 6 s, with 4 s intervals between frames. Images of the Fluo-3 loaded cells were recorded and analysed using the Physiology evaluation software package (Zeiss). An average of three independent experiments is presented.
For pHi measurements, cells were loaded with 6 μM carboxy SNARF-4F/AM (Molecular Probes) for 45 min at room temperature. Cells were excited at 543 nm and dual wavelength measurements (587 and 650 nm) were performed on a laser scanning confocal microscope (LSM 510 META, Zeiss, Germany) with intervals of 10 seconds for 5 frames. Calibration of the carboxy SNARF-4F signal was performed by the method of Thomas et al.  using nigericin (20 μM) in high K+ salines. The ratios at saturating acidic and basic conditions (pH 2.6 and 9.68 respectively) were determined, and calibration carried out using the equation
where R is the ratio Fλ1/Fλ2 of fluorescence intensities (F) measured at the two wavelengths λl and λ2 and the subscripts A and B represent the limiting values at the acidic and basic endpoints of the titration respectively, following the manufacturer's protocol (Molecular Probes). pKa for carboxy SNARF-4F was 6.4, and the calibration procedure yielded a RB at pH 9.68 of 0.2134, a RA at pH 2.6 of 1.4791 and a FB(λ2)/FA(λ2) of 171.215/77.97.
Isolation of smooth muscle microsomes and HEK 293 membranes; SDS/PAGE and Western blotting analysis; protein aggregation test
Unless otherwise stated, all procedures were performed at 4 °C or on ice. Smooth muscle microsomal membranes were isolated from porcine stomach smooth muscle according to the protocol described in [9,18]. Membrane preparation from HEK 293 cells was performed using a detergent-free protocol described in . For membrane binding assays, the membranes were pelleted by centrifugation (10 min at 12000 g), resuspended in a buffer containing 100 mM Mes, 100 mM EGTA, 100 mM KCl at different pH (from 5.0 to 7.0, in 0.2 pH unit increments) and incubated at 4 °C for 30 min. Membrane-bound annexins were then separated from the unbound proteins by centrifugation (20 min, 12000 g at 4 °C). The membrane-bound proteins in the pellet and the soluble proteins in the supernatant were analysed by SDS/PAGE followed by Western blotting with anti-annexin antibodies. Image analysis to estimate the protein content of the individual bands following SDS/PAGE and Western blotting was performed using PowerLook 1120 scanner and ImageQuant TL (v2003) software from Amersham Biosciences. In order to test for pH-dependent protein aggregation, microsomes were stripped of annexins by incubating them in buffer containing 100 mM EGTA at pH 7.0, and the soluble annexins were separated from the membranes by centrifugation. The pH of the resulting supernatants was adjusted to the same values as in the binding experiments (ranging from 2.0 to 7.0, in 0.2 pH unit increments), and following incubation at 4 °C, insoluble protein aggregates were pelleted by centrifugation.
Induction of hypoxia in the cultured cells and HIF-1α detection
Cells were incubated under either normoxic conditions (20% O2), using a standard CO2 incubator, or hypoxic conditions (1.5% O2), using a hypoxia workstation (Ruskinn Technology) for 1, 6, and 24 h. Induction of HIF-1α was detected in nuclear-enriched protein extracts by Western blotting as described . Nuclear-enriched proteins were separated by SDS/PAGE (7.5% gel) and analysed by Western blotting with anti-HIF-1α antibody. The membranes were re-incubated with an anti-Sp1 antibody to control for equal loading. To localize the fluorescent protein-tagged annexins following hypoxia, the coverslips were fixed with cold ethanol inside the hypoxia workstation. To induce hypoxia during live cell imaging, cells were grown on coverslips, placed in Na+-Tyrode's buffer (pHo=7.4, where pHo is extracellular pH) in the recording chamber (LaCon) on a heated stage and sealed from the outside air using a lid. Oxygen in the chamber was then substituted by a mixture of air and N2 (1.5% O2, 98.5% N2) created using a Digamix 2M 302/a-F pump (H. Woesthoff GmbH, Germany) at a constant flow rate of 3.7 litre/min.
Acidification of pHo promotes a decrease of pHi and induces translocation of annexin A6 to the plasma membrane
In order to visualize the interactions of annexins with the plasma membrane of live cells, we expressed them in HEK 293 cells as fusions with fluorescent proteins and tracked their translocation in response to various stimuli. HEK 293 cells were co-transfected with annexin A2-CFP, and annexin A6-YFP, and the cells were exposed to acidic environment (pHo=3.0) for a period of several minutes. A series of confocal images (frames) was taken every 4 s at a scanning speed of 6 s (Figure 1A and Video 1 at http://www.BiochemJ.org/bj/409/bj4090065add.htm). In non-stimulated cells (200 s), both annexins show a diffuse, predominantly cytoplasmic localization. Approx. 3 min after reducing the pHo from 7.4 to 3, a translocation of annexin A6 to the plasma membrane was observed (350 s), and this association remained until the end of the experiment (500 s). Surprisingly, the localization of annexin A2 did not change under these conditions, which allowed the assumption that the process was Ca2+-independent, since the Ca2+-sensitivity of annexin A2 exceeds that of annexin A6 . Therefore, in a parallel experiment, we investigated the effects of extracellular acidification on the pHi using the cell-permeable fluorescent pH indicator carboxy SNARF-4F (Figure 1B). The cells were subjected to a pHo of 3.0, the fluorescence emission ratios measured and the results converted into pH values following calibration. In cells exposed to an acidic milieu, the pHi gradually dropped from approx. 7.1 to 6.1 within 5 min, probably due to the decreased rate of acid extrusion by Na+/H+ exchangers , and translocation of annexin A6 to the plasma membrane occurred at values below 6.2 (Figure 1B, arrow). These results indicate that a decrease in pHo caused the reduction of pHi, which is in agreement with previously published results [5,25], and that the translocation of annexin A6-YFP to the plasma membrane is likely to be pH-dependent.
Extracellular acidification leads to a decrease in pHi and the translocation of annexin A6 to the plasma membrane in HEK 293 cells
Intracellular acidification induces annexin translocation independent of [Ca2+]i
Weak acids and bases are routinely used to induce changes in pHi without influencing pHo [26,27]. We used 60 mM sodium butyrate (salt of a weak acid) in Na+-Tyrode's buffer (pH 7.4) containing 80 mM NaCl to adjust the osmolarity. Exposing the carboxy SNARF-4F/AM-loaded cells to butyrate caused a rapid decrease in pHi from approx. 7.2 to 6.7 (Figure 2A). The pHi was further decreased by acidification of pHo from 7.4 to 6.8, which, in the presence of butyrate, resulted in a further decrease of pHi to 6.1. These changes were completely reversible by perfusion with standard Na+-Tyrode's buffer at pH 7.4 (Figure 2A).
Effects of Na+-butyrate and extracellular acidification on pHi, [Ca2+]i and localization of annexins in HEK 293 cells
Changes in pHi are often accompanied by an alteration of [Ca2+]i, where intracellular acidification correlates with a marked increase in [Ca2+]i due to the altered activity of calcium-entry channels, release from intracellular stores [ER (endoplasmic reticulum) and mitochondria] and decreased affinity to Ca2+-binding proteins [26,28]. We investigated the influence of butyrate-induced intracellular acidification on [Ca2+]i by simultaneously loading HEK 293 cells with Fluo-3/AM and carboxy SNARF-4F/AM and perfusing them with 60 mM Na+-butyrate in the absence of extracellular calcium (Figure 2A and 2B). A decrease of pHi was accompanied by an increase in [Ca2+]i, which was reversed when the pHi returned to normal (Figure 2B, solid line). Annexins have the ability to bind cellular membranes in a Ca2+-dependent manner, hence we investigated the role of [Ca2+]i in pH-induced membrane interactions. The cells were loaded with 20 μM BAPTA/AM, a cell-permeable Ca2+ chelator, and again subjected to butyrate perfusion. BAPTA/AM-loaded cells showed intracellular acidosis similar to the control (Figure 2A, dotted line), however, no significant elevation of [Ca2+]i was recorded (Figure 2B, dotted line).
To investigate the pH-induced membrane association of annexins, HEK 293 cells expressing annexin A6-YFP and annexin A2-CFP were loaded with 20 μM BAPTA/AM and exposed to 60 mM butyrate in Ca2+-free buffer at pHo 7.4 (Figure 2C) for 30 s, followed by a pHo of 6.8 for 250 s. A sharp decrease of pHi in butyrate-treated cells exposed to extracellular acidification caused a rapid translocation of annexin A6-YFP to the plasma membrane (Video 2, http://www.BiochemJ.org/bj/409/bj4090065add.htm), whereas annexin A2-CFP retained its mostly cytoplasmic distribution. When the same experiments were performed with cells co-expressing annexin A6 with either annexin A4 (Figure 2D, Video 3, http://www.BiochemJ.org/bj/409/bj4090065add.htm) or annexin A1 (Figure 2E, Video 4, http://www.BiochemJ.org/bj/409/bj4090065add.htm), only annexin A6 showed the ability to interact with the plasma membrane in a pH-dependent, Ca+-independent manner. In contrast, annexin A2-CFP showed an inconsistent localization pattern: it was able to bind to the plasma membrane of cells expressing low amounts of the protein. Annexin A2 forms a heterotetramer with S100A10 protein in vivo, and it is possible that, in transiently-transfected cells, the levels of endogenous S100A10 were insufficient to complex all of the recombinant annexin A2. Therefore we co-expressed annexin A2-CFP with S100A10-CFP and annexin A6-YFP, and monitored their translocation in response to butyrate-induced intracellular acidification (Figure 2F, Video 5, http://www.BiochemJ.org/bj/409/bj4090065add.htm). Under this condition, heterotetrameric annexin A2 was comparable to annexin A6 in localization to the plasma membrane (Figure 2F, 250 s). All these experiments were performed in BAPTA-loaded cells to exclude Ca2+-dependent membrane binding.
Annexins A2 and A6, but not annexins A1, A4 and A5, display pH-dependent membrane interactions in vitro
Next we examined binding of endogenous annexins to smooth muscle microsomes at different pH values in the absence of [Ca2+]i. Since the ability of EGTA to bind Ca2+ is affected by pH, we used the MaxChelator program Winmaxc 2.40 to determine the EGTA concentration required to keep the free Ca2+ concentration below 100 nM at any pH point. In agreement with our results in vivo (above) and previous results [10,15], annexin A6 interacted with microsomal membranes in a Ca2+-independent manner (in the presence of 100 mM EGTA), with a half-maximal binding at pH 6.0 (Figure 3A). Both annexin A2 and S100A10 bound microsomes with half-maximum binding at pH 6.4, and showed an identical distribution between the soluble and membrane-bound states (Figure 3B), indicating that the heterotetrameric rather than monomeric annexin A2 had the ability to bind microsomes in a pH-dependent and Ca2+-independent manner. Annexins A1, A4 and A5 did not show a pH-dependent interaction with microsomes within the tested ranges of pH (results not shown).
pH-dependent binding of annexins A6 and A2–S100A10 to microsomes
Proteins tend to precipitate when the pH of the solution is near their isoelectric point, therefore we performed precipitation controls for annexin A2–S100A10 and annexin A6 (see the Experimental section). We found that annexins A2 and A6 started to aggregate at pH 4.5, which is well below the pH values at which their binding to the membranes was observed (see above).
We have previously shown that annexins A2 and annexin A6 preferentially interact with cholesterol-rich domains at the plasma membrane upon the increase of [Ca2+]i . In contrast with these observations, the Ca2+-independent association of annexin A2–S100A10 and annexin A6 with microsomes was not significantly influenced by an alteration of the lipid composition by extraction of the microsomes with 2% methyl-β-cyclodextrine to sequester cholesterol (not shown).
The A2–S100A10 heterotetramer is essential for pH-dependent membrane association
HEK 293 cells endogenously express both annexin A2 and S100A10, so it is possible that at least some of the recombinant annexin A2 formed a complex with the up-regulated endogenous S100A10. In order to investigate the Ca2+ sensitivities of A2 monomer and heterotetramer separately, we transfected hepatoblastoma HepG2 cells with recombinant fluorescent protein-tagged annexin A2 either alone, or together with S100A10 (Figure 4). These cells express virtually no endogenous annexin A2 or S100A10 . In the absence of S100A10, annexin A2-CFP did not translocate to the plasma membrane of the butyrate-treated HepG2 cells following the decrease of pHo to 6.8 (Figure 4A). When S100A10-YFP was co-expressed with annexin A2-YFP and annexin A6-CFP, a translocation of annexin A2-YFP to the plasma membrane was observed, similar to annexin A6 (Figure 4B). The monomeric annexin A2-YFP was largely pH-insensitive, showing a weak translocation at a pHi of 5.4 (Figure 4C), whereas the annexin A2–S100A10-YFP complex showed the pH sensitivity, which was higher than that of annexin A6-CFP, translocating at a pHi of 6.4 (Figure 4D). The observed binding was mediated by the annexin A2 part of the heterotetramer, rather than by S100A10, because S100A10-YFP expressed alone in HepG2 cells did not show a pH-sensitivity in response to any of the treatments (results not shown).
Heterotetramer formation is necessary for pH-dependent annexin A2 membrane association
Distinct pHi values are required for membrane-binding of individual annexins
To compare our biochemical with the microscopical results, we calibrated the pHi using carboxy SNARF-4F in combination with 20 μM BAPTA. The cells expressing GFP-tagged annexins, and loaded with SNARF and BAPTA, were exposed to high K+ buffers containing 20 μM nigericin at different pH values (ranging from 7.0 to 5.0, at 0.2 pH unit increments), which resulted in a controlled gradual acidification of the cytoplasm. Images of SNARF fluorescence and GFP localization were taken simultaneously and emission ratios of SNARF-4F were converted into pH values. A representative experiment for annexin A6-GFP is shown in the supplemental Figure 1 (http://www.BiochemJ.org/bj/409/bj4090065add.htm).
Similar experiments were performed with GFP-tagged annexins A1, A4, A5, A2–S100A10 and A2 core, a truncated form of annexin A2 lacking the amino-terminus and incapable of binding S100A10, and the results are summarized in Figure 5. In agreement with our biochemical results, recombinant annexin A6 and heterotetrameric annexin A2 were the most pH sensitive, showing a robust plasma membrane interaction at pHi values of 6.2 and 6.6 respectively. The higher pH sensitivity of annexin A2–S100A10-GFP compared to annexin A6-GFP could also be observed in the nigericin-treated cells expressing annexin A2-CFP, S100A10-CFP and annexin A6-YFP (Video 6, http://www.BiochemJ.org/bj/409/bj4090065add.htm).
Annexins show distinct pH sensitivities of membrane association
Dimerization of the annexin A2 core is essential for its pH-sensitivity
Using an annexin A2core-GFP fusion protein , we found the pHi necessary for its association with the plasma membranes to be significantly lower (pH 5.4) than that of the heterotetrameric annexin A2–S100A10 (pH 6.6) (Figure 5). The heterotetramer has a total of eight annexin repeats, identical to the annexin A6 molecule. In order to assess the possible co-operative effects of core sequence multimerization, we created fusion constructs linking either two (A2c2c-YFP) or three (A2c2c2c-YFP) annexin A2 core sequences in-frame with YFP. Transient transfection of both plasmids in HEK 293 cells yielded proteins of expected molecular weights (96 and 130 kDa respectively, Figure 6A). A2c2c-YFP displayed a similar intracellular distribution as A2core-YFP, and A2c2c2c-YFP was largely excluded from the nucleus, probably due to its molecular mass (Figure 6B). In contrast with the annexin A2 core, both fusion proteins had the ability to bind the plasma membrane upon intracellular acidification, and both were more pH-sensitive than annexin A6 (Figure 6B). The annexin A2 core dimer construct interacted with the plasma membrane at a pH very similar to the annexin A2–S100A10 heterotetramer (Figure 6C), whereas the addition of a third core sequence in the annexin A2 core-trimer rendered the protein slightly more pH sensitive than the core dimer or the annexin A2–S100A10 complex, indicating that the binding was co-operative.
The fusion of two or three annexin A2 core sequences increases the pH sensitivity of the resulting proteins
To investigate whether multimerization of annexin repeats was the reason for the proteins' pH sensitivity, we fused two sequences of annexin A4. Wild-type annexin A4 did not show a pH-induced membrane binding in vitro or as a fluorescent protein fusion in live cells (see above). Two repeats of the annexin A4 sequence were cloned in-frame with CFP, and the resulting protein annexin A4a4-CFP showed the expected size of approx. 90 kDa (supplemental Figure 2A, http://www.BiochemJ.org/bj/409/bj4090065add.htm). However, no acidification-induced membrane translocation was detected or any significant increase in pH sensitivity observed, compared to monomeric annexin A4-CFP (supplemental Figure 2B and 2C).
Recombinant fluorescent protein-tagged annexins A6 and A2–S100A10 translocate to the plasma membrane during hypoxia
HEK 293 cells were co-transfected with annexin A2-CFP and annexin A6-YFP, and placed under hypoxic conditions (1.5% O2) for 24 h, then fixed with ethanol inside the hypoxia workstation and examined on the confocal microscope. Under normoxic (20% O2) conditions, both proteins showed a diffuse cytoplasmic localization, without noticeable association with the plasma membrane (Figure 7A, graph). After 24 h of hypoxia, a significant proportion of annexin A6-YFP became associated with the plasma membrane, in contrast with monomeric annexin A2-CFP, which retained its cytoplasmic distribution (Figure 7B, graph). Adaptation to hypoxia is regulated by hypoxia inducible factor HIF-1, and the expression of its oxygen-regulated subunit HIF-1α was observed after 1 h in the chamber and persisted until 24 h afterwards (Figure 7C). We then performed similar experiments with the cells placed into a sealed chamber assembled at the microscope stage and exposed to the continuous flow of air and N2 (1.5% O2, 98.5% N2). The cells were co-transfected with plasmids expressing annexin A2-YFP and S100A10-YFP, loaded with carboxy SNARF-4F/AM, and simultaneous evaluation of pHi and localization of the annexin A2 heterotetramer were performed by taking 5 frames at 3 seconds intervals after 15, 30, 45 and 60 min of hypoxia. pHi decreased from 7.2 to approx. 6.6 during the first hour of hypoxia (Figure 7D), however, the pHo of the HEPES-buffered Na+-Tyrode's remained at 7.4, indicating that the decline of the pHi was induced by deficiencies of oxidative phosphorylation, or other hypoxia-related mechanisms. Translocation of annexin A2–S100A10 to the plasma membrane was observed after 30 min under hypoxic conditions, and remained stable after 1 h in the chamber (Figure 7E).
Hypoxia induces the membrane association of annexin A6 and the annexin A2–S100A10 heterotetramer
Endogenous annexins in smooth muscle tissue and HEK 293 cells associate with plasma membranes at low pH
Apart from the biochemical results on pH-dependent annexin-binding to smooth muscle microsomes, the results presented above were obtained using recombinant annexins tagged with fluorescent proteins and transiently expressed in cell lines. However, in order to demonstrate that the observed effects were not merely a function of chimeric proteins, we investigated the localization of endogenous annexins in human bladder smooth muscle bundles and in non-transfected HEK 293 cells exposed to a pHo of 5.0 (supplemental Figure 3A, http://www.BiochemJ.org/bj/409/bj4090065add.htm). Exposure of the muscle, ‘relaxed’ in a Ca2+-free solution, to a pHo of 5.0 in the presence of 100 μM BAPTA/AM to ensure the removal of intracellular Ca2+, prompted the relocation of annexin A6 to the plasmalemma. In cultured HEK 293 cells at a low pHo, endogenous annexins redistributed analogous to the recombinant ones: at a pHo of 5.0, annexins A6 and A2 became associated with the plasma membrane and annexin A5 retained its diffuse cytoplasmic localization.
The biochemical characterization of pH-dependent membrane-association of endogenous annexins from HEK293 cells (supplemental Figure 3B) with membranes closely resemble the results obtained for porcine smooth muscle microsomes, demonstrating that both recombinant and endogenous annexins react alike when subjected to low pH values.
A large number of proteins have the ability to interact with cellular membranes in a pH-dependent manner, thus acting as pH sensors: cytochrome c, Bcl-xL and chromogranins A and B have been shown to bind lipid vesicles at acidic pH values [31–33], also, ARF1 is associated with endosomal membranes in a pH-dependent manner .
The annexins comprise over 50 members, are found in all eukaryotic phyla , and exhibit a similar molecular structure. Their characteristic ability to bind negatively charged phospholipids in a Ca2+-dependent manner has been extensively investigated [7,36]. Here, we focus on their ability to interact with membranes in a Ca2+-independent fashion. We constructed fusions of annexins A1, A2, A4, A5 and A6 with fluorescent proteins, and monitored their localization in response to changes in pHi caused by treatment with the pHi-reducing agent sodium butyrate, extracellular acidification and hypoxia. Our study compares for the first time the Ca2+-independent membrane-binding activities of different annexins in live cells, and demonstrates that in the higher, more physiological pH range (6.2 and 6.6) only annexin A6 and annexin A2–S100A10 are capable of Ca2+-independent membrane association. These results are supported by the results obtained for endogenous annexins from human bladder smooth muscle tissue, HEK 293 cells and porcine microsomes, showing a pH-dependent membrane association of heterotetrameric annexin A2 and annexin A6.
The acidification-mediated membrane binding of annexins A6 and A2–S100A10 is Ca2+-independent and not influenced by membrane lipid composition
Extracellular acidification reduces the activity of Na+/H+ exchangers and other regulators of pH homoeostasis, and as a consequence decreases pHi . When the pHi was monitored in HEK 293 cells exposed to a pHo of 3.0, we found that it was gradually declining to approx. 6.1. When the pHi dropped below 6.3, annexin A6-YFP, but not annexin A2-CFP, became associated with the plasma membrane, which allowed the assumption that the pH-induced binding was Ca2+-independent. However, intracellular acidification often correlates with a marked increase in [Ca2+]i due to the altered activity of calcium-entry channels, as well as release from the intracellular stores (ER and mitochondria) and decreased affinity for Ca2+-binding proteins [26,28]. It was therefore necessary to exclude the observed translocations after acidification that were mediated by a rise in [Ca2+]i. We found that a decrease of pHi caused a concomitant increase in [Ca2+]i in the absence of extracellular Ca2+, therefore the cells were loaded with the Ca2+ chelator BAPTA/AM. Loading of cells with 20 μM BAPTA/AM had no effect on the butyrate-induced changes of pHi. Although both the Fluo-3 fluorescence and the BAPTA affinity for Ca2+ are pH-sensitive, in the BAPTA-loaded cells, but not in the BAPTA-free cells, the concomitant increase of Fluo-3 fluorescence was abolished, indicating that Ca2+ was indeed buffered.
When the pHi was reduced in cells treated with sodium butyrate, only annexin A6 and the heterotetrameric form of A2–S100A10 associated with the plasma membrane. In vitro, in the presence of 100 mM EGTA, both annexin A2 and S100A10 bound to microsomes (half-maximum binding at pH 6.4), with an identical distribution between the soluble and membrane-bound states, indicating that the heterotetrameric rather than monomeric form was responsible for pH-dependent, Ca2+-independent association. Annexin A6 interacted with microsomal membranes in a Ca2+-independent manner (half-maximal binding at pH 6.0), while annexin A1, A4, and A5 did not show a pH-dependent interaction with microsomes. Selective extraction of cholesterol had no effect on the binding efficiency of annexin A6 or annexin A2–S100A10. These results differ from previous results on Ca2+-mediated membrane binding at neutral pH, which is strongly affected by the composition of the lipid bilayer [18,19]. They also indicate that the less discriminative hydrophobic interactions might bring about the pH-mediated membrane association, which is in agreement with the biophysical models .
To correlate the biochemical assays with the situation in vivo, we monitored the localization of GFP-tagged annexin A1, A2, A4, A5, A6, A2-core and A2–S100A10 in nigericin-treated cells in high K+ buffers with different pHo. Analogous to the gradient of annexin calcium sensitivities , we have established a pH-gradient for annexin–membrane binding. The pHi required for the translocation of annexin A2–S100A10 was 6.6, followed by annexin A6 (6.2) and annexin A1 (5.9). The pHi of annexin A4 and A5, as well as the annexin A2-core and monomeric annexin A2 (measured in HepG2 cells) was around 5.3, which correlated well with the pH determined by the microsome-binding assays.
Dimerization of the annexin A2 core sequence is obligatory for a co-operative pH-dependent membrane association
We show that the dimerization of annexin A2 within the heterotetramer is essential for acidification-induced membrane binding of the complex, and that the pH sensitivity of the heterotetramer is greater than that of annexin A6. Annexin A2 but not S100A10 was the mediator of the pH-induced membrane binding of the heterotetramer, which gave rise to the idea that it was the number of annexin repeats which rendered the protein pH-sensitive. Both annexin A6 and the annexin A2-heterotetramer have eight repeats, therefore we designed an annexin which exclusively consisted of repeats by fusing two annexin A2 core sequences. We found that the mere dimerization of the annexin A2 core rendered the protein pH-sensitive, and that the pHi necessary for its membrane binding was virtually identical to that of the annexin A2–S100A10 heterotetramer. Interestingly, the annexin A2c2c protein (A2 core dimer) was also significantly more Ca2+ sensitive than the annexin A2 core when binding to the membranes at neutral pH (K. Monastyrskaya, unpublished work). The addition of a third core sequence produced a protein (A2c2c2c) with slightly higher pH sensitivity than that of the dimeric core, indicating that the pH-dependent membrane interaction of these amino acid sequences was co-operative. The amino acid composition of the cores was important for binding: a similar dimerization of annexin A4 did not result in a pH-sensitive protein. Our results are in accordance with the predictions by Golczak et al.  who calculated a pH-sensitive behaviour of annexin A2 based on its amino acid sequence.
Hypoxia causes intracellular acidification and promotes the interaction of annexin A6-YFP and annexin A2–S100A10-YFP with the plasma membrane
Exposure of annexin A6-YFP-expressing cells to hypoxic conditions for 24 h resulted in the association of annexin A6 with the plasma membrane. Hypoxia and ischaemia are associated with a decrease in pHi: a mean pHi of 5.3 was demonstrated in hippocampal neurons [39,40]. Subjecting the cells to hypoxia in our experimental setup induced a robust expression of the oxygen-regulated HIF-1 subunit HIF-1α, observed within 1 h of exposure and with a concomitant drop in pHi. Within the first 30 min, the pHi in the hypoxic cells decreased to approx. 6.6, however no translocation of annexin A6 could be observed during the first hour, in contrast with the more prolonged treatment (24 h). It was not possible to monitor the pHi in the cells placed in the hypoxia workstation, but pHo of the bicarbonate-buffered cultivation medium was decreased during this time (from 7.4 to 6.9). This might have promoted a further decrease of pHi and thus assisted the membrane association of annexin A6. When annexin A2 was co-expressed with S100A10, the resulting heterotetramer gained the ability to translocate to the plasma membrane after 30 min under hypoxic conditions. The pHi values required for the membrane association of the recombinant annexin A2–S100A10 in normal and hypoxic conditions correlate very well, which indicates that it is the hypoxia-induced intracellular acidification and not the reduction of O2per se which drives the interaction.
This is the first report of a pH-induced membrane association of the recombinant annexins in vivo under hypoxic conditions. Although most annexins have the ability to interact with membranes at low pH values, the sensitivity of this interaction varies greatly depending upon the amino acid sequence of the individual proteins, where annexin A2-heterotetramer and annexin A6 are the most pH-sensitive. Annexins A6 and annexin A2–S100A10 have been reported to possess a channel-modulating activity [41–43], the process for which their association with membranes is a necessary first step. Interestingly, short-chain fatty acids including butyrate are known to induce apoptosis in colon cancer cells in vitro and in vivo, however their mode of action is poorly defined . Our observations, carried out in living cells, add an important facet to the mechanisms of the membrane interaction of annexins and open a new investigative avenue for potential functions of these proteins under the low pH conditions occurring in ischaemia and cancer.
We thank Dr A. Keogh, and Mrs A. Hostettler and Mrs C. Allemann for superb technical assistance, Dr F. Burkhard for bladder tissue samples and Prof M. Gassmann for the anti-HIF antibody. We gratefully acknowledge the financial support of the Swiss National Science Foundation (SNF Grant 320000–111778 to K. M. and the National Research Programme NRP 53 “Musculoskeletal Health-Chronic Pain” 405340–104679/1 to A. D.).
[1,2-bis-(o-aminophenoxy)ethane-N,N,N′,N′-tetra-acetic acid tetrakis(acetoxymethyl ester)
intracellular calcium concentration
cyan fluorescent protein
green fluorescent protein
- HEK-293 cells
human embryonic kidney 293 cells
seminaphtharhodafluor-4F acetoxymethyl ester
yellow fluorescent protein