Significance of oxygen transport through aquaporins

Aquaporins are membrane integral proteins responsible for the transmembrane transport of water and other small neutral molecules. Despite their well-acknowledged importance in water transport, their significance in gas transport processes remains unclear. Growing evidence points to the involvement of plant aquaporins in CO2 delivery for photosynthesis. The role of these channel proteins in the transport of O2 and other gases may also be more important than previously envisioned. In this study, we examined O2 permeability of various human, plant, and fungal aquaporins by co-expressing heterologous aquaporin and myoglobin in yeast. Two of the most promising O2-transporters (Homo sapiens AQP1 and Nicotiana tabacum PIP1;3) were confirmed to facilitate O2 transport in the spectrophotometric assay using yeast protoplasts. The over-expression of NtPIP1;3 in yeasts significantly increased their O2 uptake rates in suspension culture. In N. tabacum roots subjected to hypoxic hydroponic conditions, the transcript levels of the O2-transporting aquaporin NtPIP1;3 significantly increased after the seven-day hypoxia treatment, which was accompanied by the increase of ATP levels in the apical root segments. Our results suggest that the functional significance of aquaporin-mediated O2 transport and the possibility of controlling the rate of transmembrane O2 transport should be further explored.

Following formaldehyde fixation 13 , paraffin embedding and preparation of sectioned yeast cells for immunodetection with the anti-human aquaporin 1 antibody, strong immunofluorescence was detected in the periphery of the yeast cell section of the HsAQP1 strain in comparison with the relatively weak intracellular fluorescence signal (Fig. 2), pointing to the plasma membrane as the likely localization site. This is consistent with the subcellular localization prediction by TargetP 14 , suggesting the absence of mitochondrial targeting peptide and pointing to the secretory pathway as the most likely location of HsAQP1 in eukaryotic cells (Supplementary Information  Table S1). O 2 transport. Of the yeast strains that were examined, those expressing HsAQP1, NtPIP1;3 NtPIP1;4, NtPIP2;1, and NtXIP1;1 showed statistically significant increases in O 2 permeability with preliminary spectrophotometric measurements as evidenced by higher rates of change in myoglobin A 541 absorbance (Figs S2 and S3A) compared with mock control (Fig. S3B). Over-expression of the A. thaliana (AtPIP1;1, AtPIP1;2, AtPIP 1;3, AtPIP1;4, and AtPIP2;1) and Laccaria bicolor (LbAQP1, LbAQP3, LbAQP5, LbAQP6, and LbAQP7) aquaporins did not alter A 541 absorbance (Fig. S3B), indicating no significant effect on O 2 permeability.
Two of the most promising O 2 -transporters (HsAPQ1 and NtPIP1;3) and one that did not show O 2 -transporting properties in preliminary experiments (AtPIP1;2), as well as the mock strain were further analyzed in yeast protoplast assay. Based on the spectrum scanning on purified myoglobin (Fig. S2) and yeast protoplasts ( Fig. S4), ∆ A 541 /∆ A 600 and ∆ A 319 /∆ A 341 (Fig. S5) were chosen to indicate myoglobin oxygenation. ∆ A 541 /∆ A 600 at 90 s showed the same trend across the strains with the preliminary assay: the strain expressing NtPIP1;3 had the highest value, followed by HsAQP1, mock and AtPIP1;2 in order (Fig. 3). ∆ A 319 /∆ A 341 after 5 min with 5 times of 30 s aeration demonstrated more distinct statistical difference between HsAQP1 and mock, and between all of the myoglobin-expressing strains and untransformed strain INVSc1 (Fig. 4).
Since the conversion of deoxymyoglobin to oxymyoglobin is iron-dependent, and may be affected by the cell redox status, the redox state of selected strains after being pretreated for O 2 transport assay was measured using CM-H 2 DCFDA. Fluorescence intensity generated by CM-H 2 DCFDA showed no significant difference between yeast strains after the pretreatment of O 2 transport assay (Fig. S6). This suggested that the cell redox state in the mock, HsAQP1, NtPIP1;3 and AtPIP1;2 strains was similar prior to the O 2 transport assay. Similarly to the earlier report 6 , increased H 2 O 2 permeability was detected in NtPIP1;2 strain (Fig. S7). However, no increase in H 2 O 2 permeability was measured in NtPIP1;3, HsAQP1 or mock strains, whereas a slightly higher fluorescence intensity suggesting increased H 2 O 2 permeability in AtPIP1;2 strain was not statistically significant (Fig. S7).
Yeast O 2 consumption capacity. Yeast cells heterologously expressing NtPIP1;3 and HsAQP1 showed 2.3-fold and 1.8-fold higher O 2 uptake rates, respectively, compared with mock control (Fig. 5A) and depleted oxygen from the solution significantly faster (P ≤ 0.05) (Figs 5B and S8). The O 2 uptake rates of yeast cells expressing AtPIP1;2 and the time for O 2 depletion from the solution were not significantly (P ≥ 0.05) different from the mock controls (Fig. 5). Yeast cell diameter was not significantly affected by the heterologous expression of aquaporins and measured 2.87 ± 0.05, 2.74 ± 0.06, 2.98 ± 0.06, and 2.87 ± 0.08 μ m (mean, n = 50 ± SE) in mock, HsAQP1, NtPIP1;3, and AtPIP1;2 strains.  the high rate of O 2 transport) increased by about four-fold compared with well-aerated (≈ 500 μ mol L −1 O 2 ) plants ( Fig. 6A,B). In well-aerated plants, aquaporin transcript levels remained similar on days two and seven in leaves and roots ( Fig. 6A-D). Relatively minor increases were also measured for transcript levels of NtPIP1;4 in leaves and NtPIP1;1 and NtPIP1;2 in roots ( Fig. 6A,B). After seven days, a sharp increase of NtPIP1;3 was measured in hypoxic leaves (about 12-fold higher than aerated control) and roots (about 22-fold higher than aerated control) (Fig. 6C,D). There was also about three-fold increase in NtPIP2;1 in the leaves (Fig. 6C). Under the hypoxia treatment, from day two to day seven, the NtPIP1;3 transcript levels sharply increased in both leaves (P = 0.0001) and roots (P = 0.0037) (Fig. 6), which was accompanied by a significant increase in ATP levels in the apical root segments ( Fig. 7; P = 0.0267). After seven days of treatment, hypoxic and well-aerated roots had similar ATP levels in each root segment (Fig. 7B). Hypoxic plants showed healthy and green appearance, without chlorosis or other visible signs of O 2 deficiency.

Discussion
In this study, we investigated the potential contribution of aquaporins to transmembrane O 2 transport in yeast whole cells and yeast protoplasts by measuring absorbance near the peak wavelengths of myoglobin over time.
In the whole-cell assay, A 541 increased over the first 60 s, which enabled us to screen strains that expressed putative O 2 -transporting aquaporins (Fig. S3B). Changes in A 541 likely represent a combination of several processes including O 2 diffusion, oxygenation of deoxymyoglobin, conversion between oxymyoglobin and metmyoglobin, and O 2 consumption. The presence of cell walls might hinder the changes in absorbance in myoglobin and lead to an underestimation of O 2 diffusion in the preliminary screening. In addition, possible artifacts on absorbance reading might be caused during the mixing of the yeast suspension and aerated buffer. The yeast protoplast assay aimed to eliminate these potential pitfalls with numerous precautions and more replications and to maximize the signal of myoglobin oxygenation. The results suggested that ∆ A 319 /∆ A 341 in yeast protoplast assay may also be a highly sensitive parameter in measuring O 2 transport through aquaporins.
Human aquaporin HsAQP1, which we found to enhance myoglobin oxygenation by facilitating O 2 passage, was also reported to facilitate CO 2 transport when heterologously expressed in Xenopus laevis oocytes 15 . However, other major CO 2 -transporting aquaporins including AtPIP1;2 5 , NtPIP1;2 3 and LbAQP1 6 did not facilitate O 2 transport when expressed in yeast (Fig. S3B). This suggests that aquaporin orthologues have developed certain degree of specificity for O 2 transport. The alignment of all the 20 analyzed aquaporins does not show consensus residues that are exclusive to O 2 -transporting aquaporins (Note S3). It appears that the conserved residues are species-dependent rather than being relevant to transport capacity. However, it is noteworthy that all six O 2 -transporting aquaporins have well-conserved 29 amino acid residues across species, including most of the Asn-Pro-Ala (NPA) signature motifs and the selective filters of Ar/R residues (Note S4). In NtPIP1;3, the asparagine residue commonly in the second NPA motif is substituted by threonine (Thr-235), which may be potentially relevant to its highly enhanced O 2 -transporting capacity.
Calculations of permeation of hydrophobic gases (O 2 , CO 2 , and NO) have consistently shown similar values of an energy barrier of 5-6 kcal mol −1 through water pores 16,17 . Membrane protein simulation systems of the human HsAQP1 tetramer have demonstrated the presence of a pore located in the center between the four monomers that is lined by largely hydrophobic residues and may be involved in the transport of gases rather than water 16,18 . It could be speculated that the presence and the exact structure of this pore imparts gas transport specificity to different aquaporins. In proven correct, the gating properties of this pore could be targeted to alter rates of the transmembrane passage of gases.
The results of yeast O 2 uptake rate corroborate those of the O 2 transport assays, pointing to the significance of pore-mediated transport for respiration. Increased transcript levels of HsAQP1, also sometimes accompanied by other aquaporins, have been commonly reported for cancerous cells 19,20 , with the level of HsAQP1 expression often correlated with cell growth, grade of tumor 20,21 , and metastasis 22,23 . It has been also reported that the deletion of HsAQP1 was effective in reducing breast tumor size and lung metastasis 23 and HsAQP1 silencing inhibited the proliferation and invasiveness of osteosarcoma cells 24 . Although the proposed explanations for the links between HsAQP1 expression and cancerous growth have largely focused on water transport, the association between high O 2 demand of rapidly growing cancerous cells and facilitation of O 2 transport by HsAQP1 should also be considered.
Since the reports of hypoxia-induced expression of HsAQP1 25,26 suggest that aquaporin-mediated transport processes may be especially important under low-O 2 conditions, we examined transcript levels of N. tabacum plants subjected to flooding-induced hypoxia. Although the ATP levels showed some decline in well-aerated plants after 7 days compared with 2 days (Fig. 7A,B), the reverse trend was observed in plants subjected to root hypoxia resulting in similar ATP levels in leaves and all root segments of hypoxic and well-aerated plants after 7 days of hypoxia (Fig. 7A,B). The results suggest that after the initial hypoxic stress, plants likely received sufficient oxygen to support aerobic respiration, as hypoxic plants had healthy and green appearance and did not show chlorosis or other visible signs of O 2 deficiency. While the resistance to root hypoxia can be explained in some plants by an increased supply of O 2 to the root cells through the development of specialized aerating structures such as aerenchyma, the processes of plant resistance to hypoxia in the absence of obvious structural changes remain obscure. In our study, there were no structural features present in the roots and stems of plants exposed to root hypoxia that could be indicative of improved O 2 delivery. Therefore, the increase in NtPIP1;3 transcript levels ( Fig. 6) could be among important factors contributing to improved root aeration, similarly to the increased transcript levels of HsAQP1 in hypoxic human tissues 25,26 . Clearly, the link between pore-mediated O 2 transport and hypoxia deserves further attention.
In conclusion, our results indicate that some of the studied plant and human aquaporins are likely to be involved in O 2 transport. Yeast cells heterologously expressing these aquaporins maintained higher O 2 uptake rates in liquid culture and tobacco plants exhibited sharp increases in the putative O 2 -transporting aquaporin after their roots were subjected to hypoxic conditions. These increases in O 2 transporting aquaporins after the seven-day hypoxia treatment were accompanied by increases in ATP levels in hypoxic apical root segments. The results of the study support the notion that functional significance of pore-mediated O 2 transport should receive more attention.

Methods
Expression of myoglobin and aquaporins in yeast. The complete ORF of sperm whale (Physeter macrocephalus) myoglobin (NCBI accession number J03566.1) was sub-cloned from pMB413 12 into the yeast expression vector pAG425GAL-ccdB (http://www.addgene.org/yeast-gateway/), by the Gateway technology (Invitrogen, Carlsbad, CA, USA). The complete ORFs of the 20 aquaporin genes of interest were sub-cloned from pGEM-T Easy into the yeast expression vector pAG426GAL-ccdB (http://www.addgene.org/yeast-gateway), by the same method, respectively. These genes include three animal aquaporins from Homo sapiens -HsAQP1 (DQ895575), HsAQP2 (CR542024) and HsAQP3 (CR541991), 12 plant aquaporins -NtPIP1;1 (AF440271), NtPIP1;2 (AF024511), NtPIP1;3 (U62280), NtPIP1;4 (DQ914525), NtPIP2;1 (AF440272) and NtXIP1;1 (HM475294) from Nicotiana tabacum, and AtPIP1;1 (AT3G61430), AtPIP1;2 (AT2G45960), AtPIP1;3 (AF348574), AtPIP1;4 (AT4G00430), AtPIP1;5 (AT4G23400) and AtNIP2;1 (AT2G34390) from Arabidopsis  Selection was based on ura3 and leu2 complementation. Transformed yeasts were cultured in glucose containing synthetic complete medium without Ura/Leu (United States Biological) (2 g of yeast nitrogen base, 2 g of dropout amino acids, 5 g of (NH 4 ) 2 SO 4 , 30 g of glucose in 1 L of SD-L-U + glucose medium, pH = 6) for 24 h at 1.2× g and 30 °C. Cultures were diluted to an optical density of OD 600 = 0.6. Heterologous protein expression was induced by changing the carbon source of the medium from glucose to galactose (30 g in 1 L O 2 Transport Assay. The yeasts were washed in KH 2 PO 4 buffer (0.1 M, pH 6) twice, and then suspended in N 2 -bubbled KH 2 PO 4 buffer. The yeast suspension was bubbled with N 2 for 30 s and vacuumed for 30 min, to minimize the soluble O 2 in yeast suspension, which is crucial to maintain the state of deoxymyoglobin 30,31 . The spectrum between 500 nm and 600 nm of yeast suspension was scanned after O 2 depletion and re-aeration. Compared to the spectrum of purified myoglobin 31 , the spectrum of yeast suspension suggests that the state of metmyoglobin was likely dominant over deoxymyoglobin and oxymyoglobin. In addition, the absorbance spectrum of yeast cell suspension is expectably more complex than purified myoglobin proteins. Despite these limitations, the change in A 541 after re-aeration was noticeable (Fig. S3A), which was in the range of 541-543 nm, i.e., the absorption peak of purified oxymyoglobin in the study of Zhao et al. 30 as well as in our observation (Fig. S2). The increase in A 541 reflects the conversion of deoxymyoglobin into the oxygenated state upon Fe 2+ -O 2 binding, which can be attributed to O 2 influx. The rate of increase in A 541 (∆ A 541 s −1 ) can reflect the capacity of O 2 uptake by the yeast strains expressing different aquaporins. Therefore, absorbance of 1 mL yeast suspension of each strain was recorded at 541 nm for 2 min at 1 s interval immediately after the addition of 1 mL of air-saturated KH 2 PO 4 buffer or N 2 -saturated KH 2 PO 4 as negative control, respectively, using a spectrophotometer (Thermo Genesys 10S V4.002, ThermoFisher Scientific). All measurements were carried out at 22 °C. The mean and standard error were calculated based on six biological replications. CM-H 2 DCFDA 32 was used to indicate oxidative state of selected yeast strains due to such pretreatment and to determine H 2 O 2 transport capacity of selected aquaporins 33 (Methods S4).
After screening the strains that expressed putative O 2 -transporting aquaporins by measuring the increase in A 541 over 60 s with the spectrophotometer (Fig. S3B), yeast protoplasts were prepared for the selected ones for the refined O 2 transport assay. After induction, 4 mL of yeast culture at OD 600 = 2 of each strain was harvested, pre-incubated, washed and treated with zymolyase (Yeast lyticase 100 T, United States Biological) at 37 °C, 50 rpm for 2 hr. Yeast protoplasts were re-suspended in 10 mL of enzyme buffer (1.2 M sorbitol, 50 mM magnesium acetate, 10 mM CaCl 2 in autoclaved deionized distilled water). For the initial O 2 -depleted state, the absorbance spectrum was scanned from 300 nm to 650 nm immediately after mixing 500 μ L of protoplast suspension with 500 μ L of isosmotic sodium ascorbate buffer (0.6 M sodium ascorbate, 50 mM magnesium acetate, 10 mM CaCl 2 in autoclaved deionized distilled water). Sequential scanning was conducted after each 30 s of direct aeration at time points of 30 s, 90 s, 150 s and every minute up to the 10 th min. At 541-543 nm, myoglobin of the oxygenated state has a pronounced absorbance peak 31 (Fig. S2). At 319-330 nm, both myoglobin (Fig. S2) and myoglobin-expressing yeast protoplasts showed a second, much more pronounced absorbance peak in the oxygenated state (Fig. S4B-E), which was absent in untransformed yeast strain (Fig. S4A). The value of A 319 /A 341 increased dramatically along with multiple aeration (Fig. S5C), suggesting ∆ A 319 /∆ A 341 be a good indicator for myoglobin oxygenation. Therefore, both ∆ A 541 /∆ A600 and ∆ A 319 /∆ A 341 were calculated to present the change in absorbance due to myoglobin oxygenation. Statistically significant difference across all the strains was analyzed in ∆ A 541 /∆ A 600 of yeast protoplasts at 90 s (ANOVA, Tukey test, P ≤ 0.05, n = 19; P values shown in the table of Fig. 3) and in ∆ A 319 /∆ A 341 after 5 min with 5 times of 30 s aeration (ANOVA, Tukey test, P ≤ 0.05, n = 6; P values are shown in the table of Fig. 4).
For spectrum scanning of myoglobin, purified horse myoglobin protein at 1 mg/mL was mixed with 1 volume of 10% sodium ascorbate to generate the deoxygenated state, followed by the above-mentioned series of aeration to achieve the state of oxymyoglobin.
Oxygen uptake by yeast. O 2 uptake rates in yeast suspension culture were continuously monitored over 40 min in the over-expression and mock strains (Fig. S8). The time required for the yeast suspension cultures to deplete O 2 from the solution was also measured, with glucose as a carbon source. S. cerevisiae strains INVSc1 for the expression of HsAQP1, or NtPIP1;3, or AtPIP1;2, and the mock control strain, were cultured and induced for heterologous protein expression as described above. The yeasts were washed in KH 2 PO 4 buffer (0.1 M, pH 6) twice, and then suspended in 15 mL of N 2 -bubbled SD-L-U + glucose medium in 50 mL Falcon tubes to reach OD 600 = 5. Air was supplied into the yeast suspension until its concentration of soluble O 2 reached about 235 μ mol L −1 , the stable saturation level of the still medium at 25 °C. Starting from this point, the decrease of O 2 concentration in yeast suspension was monitored and logged per second using an O 2 microsensor with tip diameter of 50 μ m (OX-50) connected to the OXY-Meter, a compact O 2 microsensor amplifier (Unisense, Aarhus, Denmark). Parafilm was used to seal and minimize free air diffusion to the Falcon tubes. The slopes of the decline in O 2 concentration during the initial 0-1000 s were calculated by linear regression, in which absolute values represented the rates of O 2 consumption by different yeast strains during the corresponding intervals. O 2 depletion time of each yeast suspension was recorded. The means and standard errors were calculated based on six biological replications.

Tobacco Root Hypoxia Study: Growth Conditions and Treatment. Tobacco (Nicotiana tabacum L.)
seeds were germinated in soil and plants grown for two weeks in a controlled-environment growth room maintained at 22/18 °C (day/night) temperatures, 60 ± 10% relative humidity, and 18-h photoperiod with photosynthetic photon flux density of approximately 350 μ mol m −2 s −1 . After two weeks of growth, plants were transferred to containers with half-strength modified Hoagland's solution aerated with aquarium pumps (dissolved O 2 of approximately 7.5 mg L −1 ). Thirty-six plants were randomly placed in six containers. After one week, the plants in three containers were subjected to hypoxia by flushing water with nitrogen gas to reach a dissolved O 2 level of 2 mg L −1 , and then left stagnant. The plants in the other three containers continued to be aerated.
Quantitative RT-PCR in tobacco. Roots and leaves were sampled after two and seven days of hypoxia treatment (n = 6). The tissue samples were frozen and homogenized in liquid nitrogen using mortar and pestle. Total RNA was extracted using RNeasy Plant Mini Kit (Qiagen, Valencia, CA, USA). The cDNA synthesis and qPCR were conducted as described in Method S1. The transcript abundance of NtPIP1;1, PIP1;2, PIP1;3, PIP1;4 and PIP2;1 was normalized against geometric mean of that of the two reference genes, EF1-α and ribosomal protein L25. Gene-specific primers were designed using Primer Express 3.0 (Applied Biosystems, Life Technologies) ( Table S2).
Determination of ATP Level. ATP levels were measured in leaves and apical, central and basal root segments after 2 and 7 days of hypoxia and in well-aerated plants. Roots and leaves of tobacco were sampled after two and seven days of treatments. The roots were divided into the basal, apical and central segments. Tissue samples were ground and 50 mg ground samples were placed in 600 μ L of ice-cold 5% Trichloroacetic acid (TCA) 34 in 2 mL centrifuge tubes. The samples were vigorously vortexed for 20 s, left on ice for 10 min and centrifuged at 10,000× g, 4 °C for 10 min. Each 400 μ L of supernatant was collected and added to 400 μ L of ice-cold Tris-acetate buffer (pH = 7.75, 1 M). For the ATP assay, 4 μ L of the mixture was pipetted into 96 μ L of ATP-free water into a well of the 96-well plate (Costar 96 well plate with flat bottom). To quantify ATP, 50 μ L of rLuciferase/Luciferin reagent from ENLITEN ATP Assay Kit (Promega, Madison, WI, USA) was added into each well, and the standard curve was prepared following the manufacturer's protocol. Bioluminescence signal 35 was detected using a microplate reader (Fluostar Optima, BMG Labtech, Ortenberg, Germany).
Statistical Analysis. The means and standard errors were calculated based on the biological replications in each assay by descriptive statistics. Statistical difference was analyzed using one-way ANOVA (Tukey's test, P ≤ 0.05).