Intracellular oxygen : Similar results from two methods of measurement using phosphorescent nanoparticles

David Lloyd*, Catrin F. Williams*, K. Vijayalakshmi, M. Kombrabail, Nick White*, Anthony J. Hayes*, Miguel A. Aon and G. Krishnamoorthy *Biosciences and School of Optometry and Vision Sciences Cardi® University, Cathays Park and Maindy Road, Cardi®, Wales, UK Department of Chemical Sciences, Tata Institute of Fundamental Research Homi Bhabha Road, Mumbai 400 005, India The Johns Hopkins University, Institute of Molecular Cardiobiology,720 Rutland Av., 1059 Ross Bldg., Baltimore MD, USA lloydd@cf.ac.uk


Introduction
Despite 50 years of research, uncertainties and controversies revolve around published values for tissue and intracellular pO 2 .Many diverse methods have been employed; all are subject to problematic aspects and minimally non-invasive approaches are continually being devised.The complexity of the processes of human energy generation and homeodynamics is nowhere more evident than in the O 2 -delivery system to tissues, cells, and reaction sites. 1 This process displays a series of step-wise decreasing O 2 levels from alveolar air in equilibrium with arterial blood (containing the equivalent of 100 M O 2 ) to the discharge of blood oxyhaemoglobin (depending on CO 2 tension) occurring with the range 80-20 mm Hg ¼ 60-30 M O 2 .The oxymyoglobin store of red muscle begins to unload at below 10 M O 2 and the working pO 2 in active skeletal muscle is only about 3 mm Hg. 2 The mean O 2 levels of some other human tissues are also unexpectedly low, e.g., 6-16 mm Hg in white brain cortex, 3 and 8 mm Hg in kidney medulla. 4In the perifovea, the P(O 2 ) was 48 AE 13 mm Hg (mean and SD) at the choroid and fell to a minimum of 3:8 AE 1:9 mm Hg around the photoreceptor inner segments in dark adaptation, rising again toward the inner retina, and 5 mm Hg at the retina. 5Even in the rat gut, values of 58, 32, 11, and 3 mm Hg were determined in stomach, duodenum, small intestine-colon junction, and hind-colon. 6The measurement of intracellular O 2 for cells in vitro should not only be appropriate to these values for their in vivo niches but also take into account the further complexities of heterogeneous dissolved gas distributions inside cells.
Oxygen and its reduced products are essential not only for energy production, 7 but also for the maintenance of appropriate redox balance 8 and cell signalling. 9Reasons for wishing to resolve and map O 2 inside cells fall into several categories.The high respiratory activity of mitochondria is responsible for > 90% of O 2 consumption of the human organism, and oxidative phosphorylation is the major mitochondrial role.The high a±nity of cytochrome c oxidase, the binuclear Cu-Fe haemoprotein, reacts by way of a multi-stage cyclic mechanism whereby a concerted and highly e±cient four electron reduction of O 2 on the inner surface of the inner mitochondrial membrane (Km O 2 < 0:1 M), 10 leads to a residual steady-state level of O 2 that has not yet been accurately assessed.This can, however, be estimated indirectly by in situ noninvasive observations of the redox states of respiratory chain components using °uorimetric monitoring for NAD(P)H, 11 and spectrophotometric measurements for °avoproteins and cytochromes. 12,13hus respiratory chain mediated electron transport leads to depletion in the matrix space and the zones outside their outer membranes in their cytosolic vicinities.It has often been assumed that the sensing of O 2 is one of the many functions of mitochondria.However the discovery of a less sensitive but more suitably poised system, the HIF signalling pathway 14 that reacts during impending hypoxic conditions, and cooperates with mitochondrial O 2 sensing 15 emphasises the roles of several distinct proline hydroxylases with poorer oxygen a±nities.Thus complexity of multicomponent fuctional reactivities can ensure a graded scale of responses.
The well established signalling functions of the single electron reduction products of O 2 utilization (especially O 2 À., and H 2 O 2 ) provides another reason for determining local O 2 concentrations, as redox states and reaction rates both hinge on these values.During pathophysiological deprivation or transient interruption of oxygenation, the O 2 demand of tissue function cannot be satis¯ed so that adaptation to hypoxia necessitates forewarning.Thus a range of mechanisms each responsible for sensing di®erent levels of low O 2 covers a wide ranging necessity for survival of viable cells and tissues: clearly the regulation is tuned for the particular needs of individual sites within the organism.In many microorganisms, and especially in bacteria inhabiting low-oxygen environments, a±nities for O 2 are very high. 16Thus, in Escherichia coli, the cytochrome bd has a Km O 2 of 5 nM O 2 17 during infections that involve invasion by intracellular bacteria, or in heavilly infected tissues intracellular O 2 may reach extremely low levels.
The \Holy Grail" of cellular physiology is to resolve the spatio-temporal complexity of intracellular O 2 distribution by constructing 4D maps.Although two-photon excitation enables their dynamic imaging 18 quanti¯cation of the various species of reactive O 2 (OH., O 2 À., H 2 O 2 and singlet O 2 ) in the past has largely relied on indirect data and has been often overestimated. 19On the macroscopic scale, the dynamic distribution of O 2 in vivo in body °uids, tissues and organs continues to be an important research aim, and a number of °uorophores and luminophores have been used for this purpose 1,20,21 : many of the most important features of delivery of O 2 to tissues have emerged.Fluorescence lifetimes of organic °uorophores are mostly of the order of 10-20 ns and the di®usion path of a small molecule is very short. 22Extra time during which di®usion may occur greatly increases the volume of solution that in°uences the excited state of each °uorophore molecule. 23Thus for some exceptionally long-lived cases, e.g., pyrenebutyric acid (100 ns in air-saturated water), the O 2 -sensing sphere of in°uence of the probe is increased from 100 to 1000 times over that volume a®ecting the shorter lived excited molecules.Exceptionally longlived °uorescent states such as this have been employed for O 2 -mapping in living cells, e.g., in cultured cells of mouse liver by the use of quantitative °uorescence microscopy, 24 and in Amoeba proteus. 25The even longer lifetimes provided by O 2quenchable phosphorescent luminophores, [26][27][28][29] together with their other desirable photophysical properties, [30][31][32][33][34][35] make them substances of choice for a burgeoning body of applications.
For O 2 sensing, Ru(II) complexes show many advantages. 27,36,37][45][46][47] In this paper, we show that two independent approaches, namely pinhole shifting in a two-photon excitation beam and epi-phosphorescence timecorrelated single photon counting (TCSPC), give similar estimates of local O 2 concentrations within several di®erent single-celled organisms and in suspensions of cultured mammalian cells.Some of these data have been published in preliminary form in a \Technical Design Note" note. 48

Yeast and Spironucleus cultures
The yeasts Schizosaccharomyces pombe 972 h-and Saccharomyces cerevisiae were maintained and grown on Petri dishes containing Sabouraud Maltose Agar (Difco).Organisms were transferred to unbaf-°ed Ehrlenmayer °asks (50 mL) containing 10 mL YEPD liquid medium (0.3% yeast extract, 1% peptone, and 1% glucose) and grown at 30 C for two days on a rotary shaker at 150 rpm.Before electroporation, repeated centrifugation for 3 min at 2000 rpm (3000 g min) and re-suspension in 1.0 M sorbitol solution removed all nutrients.Spironucleus vortens was cultured as described by Millet et al. 49

Human cells
Cultured cell lines, human mammary adenocarcinoma, MCF-7, were from the European Collection of Cell Cultures, Porton Down, Wiltshire, UK.Human ¯broblasts (human dermal ¯broblast (HDF) cells) were from TIFR Biology stocks.Cancer cells and HDF were maintained in Eagles Minimal Essential Medium supplemented with 10% foetal bovine serum, penicillin and streptomycin.Detachment from the plastic culture °asks was carried out using trypsin/EDTA and was followed by re-suspension in growth medium.

Nanoparticle preparation
The Ru coordinate complex employed as a phosphorescent probe here has been extensively used and has proven a highly e±cient O 2 sensor.Reagents (from Sigma-Aldrich) were RuCl 3 (57 mg, 0.0276 mmol) and 4,7-diphenyl-1, 10-phenanthroline disulfonic acid disodium salt (518 mg, 0.0965 mmol) in 20 mL distilled water re°uxed with stirring for two days.After ¯ltration and rotary evaporation, the product, Ru([dpp(SO 3 Na) 2 ] 3 Þ Cl 2 Á 6H 2 O, was passed down a Sephadex G25 size exclusion column and eluted with water.The ¯rst brown and purple fractions were discarded and the red fraction dried.The product is Ru([dpp (SO 3 Na) 2 ] 3 )Cl 2 Á 6H 2 O and it contains several isomers with the -SO 3 Na substituents randomly positioned on the phenyl rings.However, previous reports indicate that the mixture of isomers gives no evidence of heterogeneous °uorescent lifetimes.Characterization of the compound by standard methods was performed as described. 37The nanosensors were prepared by incorporation of Ru([dpp (SO 3 Na) 2 ] 3 Cl 2 Á 6H 2 O by encapsulation using radical polymerization added drop-wise into a solution of 43 mL hexane, 3.08 g AOT and 1.59 g Brij30 in a round bottom °ask.The monomer solution contained 2.7 g acrylamide, 0.8 g N;N-methylenebis(acrylamide), and 9 mL of 10 mM sodium phosphate bu®er, pH 7.2, were added to the microemulsion.The solution was stirred under argon throughout the preparation and deoxygenated by three freeze-vacuum-thaw cycles using liquid nitrogen as freezing medium.To initiate the polymerization, 50 L of a 10% (w/w) sodium bisul¯te solution was added.The solution was kept under argon and stirred at room temperature for 2 h to ensure complete polymerization.
Hexane was removed by rotary evaporation and the remaining solution was re-suspended in 96% ethanol and transferred to an Amicon ultra-¯ltration cell model 8200 (Millipore Corp., Bedford, MA, USA).The solution was washed with 600 mL 96% ethanol in order to separate surfactants, unreacted monomers and excess proteins from the sensors using a 100 kDa ¯lter under 2 bar pressure.The polymer particles containing Ru complex were then re-suspended in ethanol and passed through a suction ¯ltration system (Millipore Corp., Bedford, MA) with a 0:025 m nitrocellulose ¯lter membrane and rinsed with 100 cm 3 ethanol.Hydrodynamic particle diameter (45 nm) was determined by rightangle light scatter of 632.8 nm laser light (HeNe) using a BI-200SM Goniometer (Brookhaven Inst., New York).Nanoparticles were stored at -18 C. Re-suspension (18 mg mL À1 ) was in 1.0 M sorbitol and dispersion was by treatment with 20 kHz ultrasound for 30 s at 4 C at an amplitude providing the maximum cavitation intensity.

Electroporation of nanoparticles into yeast, Sp. vortens, rat neonatal cardiomyocytes, or human cells
The Biorad Labs Gene Pulser Transfection Apparatus (Bio-Rad Laboratories, CA) was ¯tted with a Capacitor Extender set at 450 V for yeasts and 340 V and 320 V for human cells or S. vortens (capacitance 900 F, resistance 200 ).Electroporation vessels were of 400 L working volume; three high voltage pulses were employed.

Pinhole shifting with two-photon excitation confocal microscopy
Ramshesh 50 and Ramshesh and Lemasters 51 have described time-resolved imaging of long-lifetime luminescence of Europium (Eu 3þ ) microspheres, and explored some aspects of the potential usefulness of the method for biological applications (e.g., O 2 sensing in feline myocytes).As in the original method used by these authors, we employed a Zeiss LSM510META NLO confocal microscope with a Coherent Chameleon 140 fs pulsed Ti-sapphire laser system (with integral pump lasers) for twophoton excitation at 900 nm.This excitation avoids much of the auto°uorescence expected from shorter wavelengths and also the longer wavelengths reduce the level of detection of any un-blocked (re°ected) IR since the PMTs are less sensitive.The single photon excitation peak is at 450 nm.While this is not de¯nitely predictive of a two-photon peak at 900 nm (i.e., 2Â) this was a natural wavelength to choose in the absence of other information.An analysis of excitation e±ciency at all available wavelengths is quite di±cult, given the many variables involved, and because of this we have yet to seek a better wavelength to use.We used about 20% of the maximum laser intensity at the specimen plane.The ruthenium nanoparticles were imaged through a 685 nm shortpass emission ¯lter after beam separation by a 545 nm longpass dichroic mirror using a 40Â1:3NA Plan-Neo°uor (Zeiss) oil immersion.The 685 nm shortpass ¯lter is an IR blocking ¯lter speci¯cally designed for multiphoton imaging, i.e., to block the re°ection of the titanium-Saphire laser.By collecting successive images from the shifted confocal of the raster scan, a series of images was collected.Image intensities were determined using the standard Zeiss LSM software and plotted as a function of decay times (Fig. 2) calculated from scan speed, pinhole o®set and consequent image displacement from the optical axis).

Epi-phosphorescence TCSPC
The time-resolved °uorescence microscope was a combination of a picosecond time-resolved °uorescence spectrometer and an inverted epi°uorescence microscope. 52The time-resolved epi-phosphorescence measurements were carried out with a TCSPC setup coupled with a picosecond laser (Fig. 3).
A titanium-sapphire picosecond laser beam (Tsunami, Spectra Physics) pumped by a diode pumped CW Nd-vanadate laser (532 nm) (Millenia X, Spectra Physics), and frequency doubled by a doubler, was used to excite a single cell at 495 nm.The pulse width of the excitation laser beam was typically $ 1 ps.A pulse repetition rate of 80 MHz was reduced to a repetition rate of 0.5 MHz by a pulse picker.The ps pulses obtained after frequency doubling were guided to the objective lens by a dichroic mirror and focused onto a single cell.The diameter of the focused beam was $ 0:5 m and hence several locations within a single cell could be probed.Time-resolved epi-phosphorescence measurements were carried out on a Nikon Diaphot 300 microscope ¯tted with a 20Â objective Fig. 2. Two examples of calculations of phosphorescence lifetime measurements from experiments with Ru nanoparticles in suspension or after electroporation into ethanol-¯xed Schizosaccharomyces pombe that used the pinhole shifting method of Ramshesh and Lemasters. 51ith 0.75 NA maintained at room temperature (23 C).Phosphorescence emission collected by the same objective lens was passed through a combination of a cut-o® ¯lter (475 nm longpass and a shortpass ¯lter (600 nm), and a polarizer.Timeresolution of the phosphorescence signal was obtained by coupling the microscope to a TCSPC card (Becker and Hickl, SPC-630).The temporal resolution of the setup is $ 50 ps and the spatial resolution is $ 0:5 m in the xy plane.The measurements typically require $ 100 luminophore molecules in the observation volume.The instrument response function was estimated by the use of oxonol VI whose °uorescence lifetime is < 50 ps.The full width at half maximum height of the instrument response function estimated in this way was $ 160 ps.Phosphorescence decays were analyzed for a single exponential function by using the nonlinear ¯tting.Since the repetition rate (0.5 MHz) of the excitation pulse was shorter than the measured lifetimes we ensured that our analysis took care of the incomplete decay coming from the previous excitation pulse.However, for single exponential decays this overlap does not cause any problem in the analysis even if the incomplete decays from previous pulses are not taken into account.

Electroporation gives minimallyinvasive permeabilization
Figure 4 shows that electroporation is an e®ective method for loading cells and organisms with polyacrylamide nanoparticles containing the Ru coordinate complex.For all four populations of organisms, nanoparticles have entered all cells.Controls neither incubated nor electroporated with nanoparticles showed no auto°uorescence at wavelengths in the red range of visible spectra (by confocal microscopy), or in the red channel in the two-photon scans.Some endosomal uptake of nanoparticles into MCF-7 cells occurred in the absence of electroporation.
The images indicate that, after electroporation in the presence of Ru nanoparticles using the mild conditions indicated, intracellular distributions are not con¯ned to any speci¯c subcellular organelles, although the distribution within a cell is heterogeneous and some organelles (e.g., nuclei) show lower intensity of emission.Absence of blebbing indicates vitality of the carcinoma cells.Similarly the continued motility of Spironucleus suggests the relatively innocuous nature of the permeablization procedure.Yeasts continued to proliferate so as to produce colonies on solid medium, and cardiomyocytes continued to spread and exhibit regular contractility.
For cardiomyocytes (Fig. 5), the persistence of NAD(P)H auto°uorescence also con¯rmed continued cellular integrity.

Response of the Ru nanoparticle probe to O 2
Figure 6 shows images of the intensity of emission of nanoparticles in three MCF-7 cells during transitions from a N 2 gas phase to air and back.Both response times are limited by gaseous di®usion into the hanging drop cell suspension.

Two-photon laser excitation scanning microscopy with pinhole shifting
The optical setup whereby laser scanning microscopy may be to acquire sequential images with delayed lifetimes 50,51 is shown in Fig. 1.From these images (Fig. 7), intensity plots of delayed emission (Fig. 2) permit determination of a delay time for each image and further calculation of phosphorescence lifetime.These images provide adequate resolution of the distribution of phosphorescence quenching within cells.

Epi-phosphorescence microscopy
The alternative strategy was to determine by single photon counting the phosphorescence emission of Ru nanoparticles from selected 0.5 m diameter spots within cells.Following a single excitation pulse, a monophasic decay curve (Fig. 8) gave similar values for lifetimes to those obtained by the pinhole shifting method (Table 1).Phosphorescence lifetimes give similar values by all three methods.For ethanol-¯xed dead Schizosaccharomyces pombe, pinhole shifting under an anaerobic atmosphere gave a lifetime of 3.19 s.This compared favorably with a corresponding value of 3.28 s, using the epi-phosphorescence-TCSPC method, as well as a previously published value of 3.88 s obtained using spectroscopy by Coogan et al. 32 (Table 1).Slight di®erences in lifetime values obtained from the latter method may be due to more e±cient N 2 saturation of the cell or particle suspensions, whereby three cycles of O 2 evacuation were employed prior to generation of ananaerobic environment.Similar values were also obtained under air.For living organisms or cultured cells, shorter phosphorescence lifetimes indicated  2), possibly due to the higher metabolic rates in the former.These data also suggested slight di®erences between microsites within an individual cell, with higher O 2 concentrations near the periphery of the cytosol (Table 2).As yet, we have no values for intra-mitochondrial pO 2 where at the point of O 2 consumption values would be much lower.
Higher resolution of these subcellular structures would provide detailed maps of O 2 distribution in 3D using xy scans at successive z coordinates.For this, higher power objectives and appropriately adjusted parameters will be employed.Figure 9 shows atypical standard curve obtained from the , where 0 and are the phosphorescence lifetimes in the absence and the presence of quencher respectively and k q is the bimolecular quenching constant.The estimated value of k q ð¼ 1:46 Â 10 9 M À1 s À1 ) is close to di®usion-controlled values as expected.Reproduced from Williams et al., 48 with permission.epi-phosphorescence-TCSPC method.This curve was used in estimations of O 2 concentrations inside a single HDF cell.Time-dependent variations in intracellular O 2 concentration, if con¯rmed in a more systematic study, would be useful in revealing metabolic oscillations.Figure 9 shows a typical standard curve obtained from the epi-phosphorescence-TCSPC method.This curve was used in estimations of O 2 concentrations inside a single HDF cell (Table 2, Fig. 9).Timedependent variations in intracellular O 2 concentration, if con¯rmed in a more systematic study, would be useful in revealing metabolic oscillations.
Major considerations of choice between the two methods employed in this study include the biological information required and cost.While raster scanning two-photon excitation potentially provides 3D rendered images, epi-phosphorescence photon quenching gives single point determinations.However, the cost implications of the former method are considerably greater.
Images obtained from two-photon excitation of a yeast (S. pombe) and a °agellate ¯sh parasite (Sp.vortens) indicated the intracellular location and magnitude of O 2 gradients and con¯rmed the feasibility of optical mapping of cells and organisms under di®erent external conditions.As well as obtaining measurements from single cells, Ru coordinate complex may prove useful to measure O 2 gradients in situ not only from intracellular sites, but also in interstitial extracellular locales (e.g., matrix elements in connective tissues) and in body °uids.However, further work is necessary to make reliable measurements of physiological signi¯cance.David Lloyd wishes to express his indebtedness to Britton Chance for his enormous in°uence during David's pivotal and formative years and his continued interest and inspiration as a mentor and friend for 44 years.Of paramount importance were the daily meetings with visiting leaders in the ¯eld of bioenergetics, from the electric eel to avocado, Vibrio ¯scheri (\photobacterium") to Arum spadix, and from Fasciola hepatica to beef heart and mouse brain.Many major achievements were pioneered by the groups of workers at the Johnson Foundation at the University of Pennsylvania in the years while Britton Chance was its leader.These included the development of mechanical, electronic and optical devices that heralded a new era in the continuous non-invasive monitoring of living systems.As well as the establishment of a range of techniques for rapid reaction kinetics, the elucidation and de¯nition of the respiratory states of mitochondria, 53 the ¯rst quanti¯cation of levels of tissue H 2 O 2 , 54 and the experimental demonstration of biological electron tunneling in membranes, 55 this laboratory was the ¯rst to exploit the diagnostic signi¯cance of intracellular redox states as monitored by NADH and °avin °uorescence, and its personell continue to exploit new clinical applications of this technique. 56atrin F Williams held an EPSRC (EP/ H501118/1) studentship with CASE Partner, Neem Biotech Ltd. G. Krishnamoorthy is a recipient of J. C. Bose National Research Fellowship of the Government of India.The authors thank Dr. Lars Folke Olsen, Dr. Allan Poulsen and Ms. Anita Lunding for expert guidance on nanoparticle preparation.Multiphoton confocal pinhole shifted laser scanning microscopy was carried out at the Vision Science Bioimaging Labs facility, School of Optometry and Vision Sciences, Cardi® University.
Note added after review.The high activity in this ¯eld is indicated by a plethora of new publications dated 2013, e.g., for brain and skin oxygenation in piglets, 57 in cortical arterioles, 58 in the rat retina, 59 and in mouse tumors. 60A recent review of optical oxygen sensing concentrates on the quenching of phosphorescence. 61

Fig. 1 .
Fig. 1.The pinhole shifting technique.By shifting the pinhole successively out of its optimally aligned position (a), through a series of steps (e.g., b), collection of sequential time-delayed phosphorescence emission in the lagging direction of the raster scanned two-photon excitation is achieved by employing appropriate laser beam scanning speeds.Reproduced with permission from Ramshesh 2007.50

Fig. 3 .Fig. 4 . 7 J
Fig. 3. Schematic diagram of the °uorescence microscope setup used for acquiring time-resolved epi-phosphorescence data on live cells.The main components are (i) a picosecond laser (ii) an inverted °uorescence microscope (iii) a TCSPC spectrometer.

Fig. 5 . 8 J
Fig. 5. Two-photon excitation of auto°uorescence (excitation 740 nm, total emission < 490 nm), was almost entirely due to NAD (P)H: in Saccharomyces cerevisiae (a) before and (b) after, electroporation.In (c) and (d) yeast and cardiomyocyte °uorescence after electroporation in the presence of Ru nanoparticles con¯rm continued cellular redox balances.Mild ultrasonic treatment gave permeabilized control cells that showed negligible auto°uorescence under these optical conditions.

Fig. 6 .Fig. 7 . 48 Fig. 8 .
Fig. 6.Responses of Ru nanoparticles within MCF-7 cells in an anaerobic suspension (gas phase N 2 ) held in a hanging drop of growth medium to a one minute pulse of air.Time dependencies of intensities of emission from selected areas corresponding to three cells of the confocal image are indicated.

Fig. 9 .
Fig. 9. Standard curve of phosphorescence lifetime vs concentration of oxygen in bu®ers saturated with nitrogen (½O 2 ¼ 0), air (½O 2 ¼ 0:26 mM) and oxygen (½O 2 ¼ 1:31 mM) estimated by the epi-phosphorescence-TCSPC method.Standard deviations of 0.26, 0.02, and 0.02 were obtained respectively for the three values as estimated from eight independent measurements (shown as error bars).The inset shows the Stern-Volmer plot, namely 0= ¼ 1 þ 0 k q [O 2 ], where 0 and are the phosphorescence lifetimes in the absence and the presence of quencher respectively and k q is the bimolecular quenching constant.The estimated value of k q ð¼ 1:46 Â 10 9 M À1 s À1 ) is close to di®usion-controlled values as expected.Reproduced from Williams et al.,48 with permission.

Table 2 .
Estimation of oxygen concentrations in MCF-7 and HDF cells by the epi-phosphorescence-TCSPC method.
Notes: Each measurement corresponds to a di®erent cell.Measurements made at regions either close to the center or close to the plasmamembrane were made in the same cell.Cells were equilibrated with air (½O 2 ¼ 0:26 mM).

Table 1 .
32fetime of Ru nanoparticles in suspension or electroporated into MCF-7, ¯xed yeast or HDF cells under di®erent gas phases (expressed as equivalent dissolved [O 2 ] mM), as assayed by spectrometry (Ru nanoparticles in suspension, or in MCF-7 cells (in parentheses)).aCooganet al.32Pinhole shifting (¯xed yeast or HDF cells under aerobic (air) or anaerobic ( b nitrogen or c argon), and epi-phosphorescence (TCSPC cells in bu®er) methods.