Rapid on-site dual optical system to measure specific reactive oxygen species (O2-• and OCl-) in a tiny droplet of whole blood

Oxidative stress has been implicated in various disorders and controlling it would be important for healthy life. We have developed a new optical system for easily and accurately measuring oxidative stress in whole blood. It is optimized for simultaneously detecting reactive oxygen species (ROS) and highly reactive ROS (hROS), elicited mostly by white blood cells in a few microliters of blood. Results obtained by using this system show at least four important findings. 1) chemiluminescence of MCLA was confirmed to be attributable to O2-•. 2) PMA-stimulated cells released O2-• longer and more slowly than fMLP-stimulated ones. 3) fluorescence produced by APF oxidation was confirmed to be attributable to hROS, mostly OCl-, produced by myeloperoxidase. 4) the generation of OCl- was found to be a slower process than the O2-• generation. We also conducted pilot studies of oxidative stress in healthy volunteers.

Typically, a 525 μl of anticoagulated blood was loaded on a 450 μl of Mono-Poly resolving medium in a microcentrifuge tube (Thermo Fisher Scientific, Waltham, MA, USA). The blood sample was then centrifuged at 400×g for 20 min at room temperature using a swing rotor. Since the middle white layer in the centrifuged tube contains neutrophils, it was recovered carefully, washed once with 750 μl of RH buffer, centrifuged again, and finally dissolved in cold RH buffer at 4˚C.

Neutrophil counting
Neutrophil counts in whole blood and that of the isolated cells were determined with Pentra MS CRP (Horiba, Kyoto, Japan) according to the manufacturer's instruction. The isolated fraction from the whole blood mainly contained neutrophils (average 73.23% out of 16 experiments). We double-checked the total number of isolated neutrophils by direct microscope observation.

Cell culture
HL-60 cells, a human acute promyelocytic leukemia cell line, obtained from American Type Culture Collection (Manassas, VA, USA) were maintained and differentiated to neutrophillike cells with DMSO as previously described [10,11,15]. The cells thus differentiated were suspended in RH buffer and kept at 4˚C until measurement. The cells were counted by microscope observation with Trypan Blue staining.

Monitoring procedure of CL and FL
Basically, CL MCLA and FL APF were simultaneously monitored in the following procedure (see also S1 Table) with the newly developed system (CFL-P2200, Hamamatsu Photonics K.K., Hamamatsu, Japan): 1. RH buffer containing 0.5 μM MCLA, 10 μM APF and 1 mM CaCl 2 was pre-incubated on a dedicated glass slide.
3. For negative control samples, adding SOD and ABAH (please see next section) to the solution on the slide. 4. Adding 1.5~3 μl of whole blood or suspension of 1× 10 4 cells of isolated neutrophils, or 2× 10 4 cells of cultured neutrophil-like cells to the solution. 5. Stirring the solution with gentle pipetting. 6. Setting the glass slide in the system. 7. Incubate for 30 seconds in the system. 8. Start the measurement. 9. Stimulation of the sample cells by adding fMLP or PMA automatically. 10. Simultaneous recording of CL MCLA and FL APF . 11. Data analysis by the dedicated software and Excel. Stimulant enhancements of the signals of CL MCLA and FL APF were determined by calculating the peak areas under the curves (AUC) [10,11].
Optical configuration of the system CL MCLA and FL APF were both detected by using a single photomultiplier tube (PMT; H10682-210, Hamamatsu Photonics K.K.; Fig 1A), with 480 nm excitation light (band-pass filter 480 nm, FWHM 10 nm), which is optimized for the FL reagent APF (Ex-Max 490nm). Further, we placed light emitting diodes (LED) as the excitation lights at the photodetector side ( Fig 1A  and Panel b in S1 Fig). Same as the previous system [10,11], the lens set was arranged in front of the PMT (Fig 1A). This lens set transmits FL APF (Em-Max 515 nm) and CL MCLA (Em-Max 465 nm). The band rejection filters ( Fig 1A) were optimized for removing excitation light efficiently. The signal separation between FL APF and CL MCLA by using a single PMT was attained by repeating on/off of the excitation lights at a high speed, which was already used in the previous system [10,11].

Blood sample holder
In order to ensure the stability of sample droplet shapes, the glass slide used in the newly developed CFL-P2200 has two circular areas (19 mm diameter) with printed rims made of a highly water-repellent material (a custom product by Matsunami Glass Ind., Ltd.). Fig 1A and Panel b in S1 Fig show that sample mixtures containing blood were spread over the predetermined areas on it. Since the signal crosstalks were suppressed effectively, two adjacent samples can be measured simultaneously (e.g., measuring a test and a control samples).

Prevention of evaporation from thin sample suspensions set in the system chamber
Stirring and mixing are necessary for adding stimulants or for preventing blood cells from aggregation, but placing a magnetic bar into the thin sample suspensions in the system may disrupt the thin optical paths and interfere with the optical detection from below. So, as a means of mixing, we adopted air flows with the air-injecting nozzles placed above the samples ( Fig 1A). For keeping the sample temperature at 37˚C and preventing evaporation in thin samples under constant air flows, we used warmed air flows and installed a humidifier inside the sample holding chamber (100% humidity). The chamber was covered with rubber heaters which can automatically control the chamber temperature.

Comparison of the cuvette-type cell and the thin suspensions on a glass slide
The previous system (CFL-C2000) [10,11] used a cuvette (10 mm square) as the sample container (Panel a in S1 Fig). However, such cuvettes loaded with colored samples containing light absorbing substances such as blood are susceptible to strong light absorptions. To circumvent this problem, we selected a glass slide as the sample holder in the new system (Panel b in S1 Fig). The optical path length of a sample placed on it becomes about one fourth shorter than that of the cuvettes. To find out the improvements in optical signal strengths we compared the CL MCLA data obtained with the previous (CFL-C2000, cuvette-type) and the new (CFL-P2200, glass slidetype) systems. The highest peak area of the CL MCLA signals was obtained with 750-fold dilution of whole blood in cuvettes and the corresponding data of the samples on a glass slide was obtained with 500-fold dilution.

Effects of air-flow stirring on solution mixing and CL/FL signals
For the air-flow stirring to work effectively and for obtaining a stable light signals, we carefully adjusted the amounts, the timings and the directions of the air flows. When the air flows were too strong, the FL APF signal intensity fluctuated noticeably due to liquid surface disturbances. Conversely, absence of air flows or unsuitable directions resulted in erroneous measurements, due to insufficient diffusion of the added stimulant and blood cell aggregations (Panel c in S2

Selectivity of MCLA, APF and HPF for monitoring O 2 -• and OCl -
We used MCLA as a probe to measure extra-cellular O 2 -• generated by NADPH oxidase in the cell membrane as previously reported [10]. MCLA is known to have a weak nonspecific luminescence derived from auto-oxidation with phosphate buffer [21]. To minimize the auto-oxidation, we used RH buffer in this system. For OCldetection, our first choice was APF, which is a fluorescence probe monitoring some of the highly reactive ROS (hROS) in a cumulative way. APF selectively and dose-dependently releases a strongly fluorescent compound, fluorescein, upon oxidation by hROS such as • OH, ONOOand OCl -, but not by other ROS [14]. Another similar fluorescence probe HPF, which also releases fluorescein upon oxidation by • OH, ONOO -, on the other hand, is not reactive to OCl - [14]. Therefore, OClcan be specifically detected by using HPF and APF together [14]. We compared FL APF and FL HPF from isolated neutrophils. S3  The FL increase of ΔHPF was less than that of ΔAPF and AUC of the ΔHPF was 90.7% less than that of the ΔAPF. Next we measured CL MCLA and FL APF from neutrophil-like cells derived from cultured HL-60. Fig 2A shows that, soon after the stimulation (dotted line), CL MCLA intensity was instantly increased and then immediately returned to the baseline level (Fig 2A, black line). This pattern was almost the same as that previously described [10,11]. To find out whether or not the signal is attributable to the generated O 2 -• , we scavenged O 2 -• by adding SOD. Addition of three different concentrations of SOD significantly decreased the CL MCLA intensity in a dose-dependent manner (Fig 2A, blue and similar colored lines). In contrast to the transient nature of the CL MCLA , the FL APF intensity accumulated gradually by oxidation of APF ( Fig 2B, black line). Likewise, to find out whether or not this signal is attributable to OClproduced by MPO, we added ABAH as an MPO specific inhibitor [19,20]. Addition of three different concentrations of ABAH markedly decreased the FL APF intensity in a dose-dependent manner as expected ( Fig 2B, red and similar colored lines).

Measurements of CL MCLA and FL APF in a tiny droplet of whole blood
Next we measured CL MCLA and FL APF elicited by a droplet of whole blood. PMA was used as a stimulant instead of fMLP because PMA was found to be more effective for stimulating neutrophils than fMLP. Blood samples were collected from a volunteer's fingertip and kept at room temperature. They were then diluted into the reaction mixture before measurement as described in Materials and Methods and S1  Rapid on-site dual optical system to access oxidative stress responses in a tiny droplet of whole blood CL MCLA (blue and pale blue lines) and FL APF (red and pink lines), in which the left vertical axis indicates CL MCLA intensity while the right vertical axis indicates FL APF intensity. CL MCLA slowly started to increase at around 90 seconds after PMA stimulation and then decreased gradually ( Fig  3A,   Rapid on-site dual optical system to access oxidative stress responses in a tiny droplet of whole blood line). ABAH inhibition experiments confirmed that FL APF was attributable to OCl - [19,20]. The ABAH suppression became noticeable at about 600 seconds (Fig 3A, pale pink line). Next, we calculated AUCs of CL and FL of these experiments, which shows that AUC of CL MCLA in the presence of SOD was 97.9% less than that without it (Fig 3B, left columns) and AUC of FL APF in the presence of ABAH was 98.3% less than that without it (Fig 3B. right columns). We then tested whether or not CL MCLA signal can be increased when O 2 -• are independently generated in blood samples by using the hypoxanthine xanthine oxidase mechanism [10,11] (Panel a in S4 Fig). When xanthine oxidase together with different concentrations of hypoxanthine was injected to the blood samples, the CL MCLA signals were increased in a dose dependent manner (Panel a in S4 Fig, columns 4 and 5). Surprisingly, the notable CL MCLA signal was detected when only xanthine oxidase was injected to the samples without hypoxanthine (Panel a in S4 Fig, column 3), which suggests that some amount of hypoxanthine was already in the blood samples. We also added OCl-to blood samples and confirmed that the

Reproducibility of CL MCLA and FL APF measurements in blood samples
Next we tested reproducibility of data with four different blood samples from a single volunteer preserved at room temperature in BD Microtainer Tube (see Materials and Methods; all blood samples in this experiment were measured within two hours after collection). Fig 4A shows time courses of CL MCLA (blue and pale blue lines) and FL APF (red and pink lines). The line color brightness indicates the order of measurement (the brightest is the latest). The reproducibility was so good that all four time courses almost overlapped with each other (Fig 4A). We calculated all four AUCs of CL MCLA and FL APF . The relative errors of the four measurements were 7.94% for FL APF and 7.06% for CL MCLA (Fig 4B).

Correlation analysis of the neutrophil data sets
We then performed correlation analysis of CL MCLA and FL APF of whole blood samples from three healthy volunteers. The analysis was focused on the following three perspectives, (1) intensity (the area of increased CL MCLA or FL APF ) obtained from the whole blood samples, (2) intensity obtained from isolated neutrophils in the blood samples and (3) the concentration of the neutrophils in them. Blood samples from the three healthy volunteers were collected after fasting for 12 hours and then neutrophils were isolated as described in Materials and Methods. Neutrophil concentrations in the blood samples were determined by the conventional method (see Materials and Methods). First, we tested correlations between the whole blood samples and the isolated neutrophils. We measured CL MCLA and FL APF using the same amount of the whole blood (3 μl) or the same number of isolated neutrophils (1× 10 4 cells) obtained from the volunteers. Fig 5A and 5B show their scatter plots, which shows correlations between the intensities obtained from the whole blood (horizontal axis) and those from the isolated neutrophils (vertical axis). Clear linear correlations were found in both CL MCLA (Fig 5A, R = 0.981) and FL APF (Fig 5B, R = 0.990). We also examined whether or not the intensities obtained from the whole blood samples were correlated with the neutrophil concentrations in each of them. Fig 5C and 5D show scatter plots which indicate correlations between the intensities obtained from the whole blood (horizontal axis) and the neutrophil concentrations in each blood (vertical axis). Again clear linear correlations were found in both CL MCLA (Fig 5C, R = 0.991) and FL APF (Fig 5D, R = 0.993). Then, by measuring neutrophil concentrations in the blood samples, we adjusted the intensities obtained from 1× 10 4 cells of the isolated neutrophils (vertical axes in Fig 5A and 5B) to those from the corresponding cell counts in each blood sample (vertical axes in Fig 5E and 5F). Evidently, the most predominant correlations were found in both CL MCLA (Fig 5E,

The effects of diet and exercise on CL MCLA and FL APF of a tiny amount of whole blood: A pilot study of oxidative stress in volunteers
The ROS/hROS dual monitoring system we have developed may be applicable to evaluations of various factors affecting the O 2 -• generation and the MPO activity in blood. To address this Rapid on-site dual optical system to access oxidative stress responses in a tiny droplet of whole blood possibility, we studied the effect of diet and exercise on ROS/hROS in volunteers' blood. Foods and supplements containing anti-oxidative compounds are expected to countervail ROS induced by over-activation in WBC [10,11,22,23]. In addition, the activity of MPO is known to be affected by various factors including exercise [24][25][26][27], aging [28,29], alcohol [30] and smoking [31]. We chose diet (i.e., food intake) and exercise among them since these are regarded as two of the most influential factors in our daily life. First, we compared ROS (CL MCLA ) and hROS (FL APF ) before and after diet for 5 consecutive days. The blood sampling times were (1) at the end of fasting for more than 12 hours (around 7:30) and (2) within an hour after lunch (around 13:00). Lunch menus were not fixed during the experiment. Panels a (the end of fasting) and b (after lunch) in S6 Fig show day-today variations of CL MCLA (blue), the FL APF (red) and the neutrophil concentration (green). To confirm the reproducibility of the decrease in the correlation coefficients, we conducted a similar analysis with another healthy volunteer for arbitrary 8 days (within 3 weeks). The blood sampling times were (1) at the end of fasting for more than 12 hours (around 10:00) and (2) within an hour after lunch (around 13:00). Similarly, after food intake, significant linear correlations decreased in both CL MCLA (R = 0.856 to 0.477) and FL APF (R = 0.844 to 0.512) (Panels e and f in S6 Fig).
Next we tested the blood ROS/hROS just before and just after exercise (within 10 minutes) for 5 consecutive days. This experiment was carried out in a hunger condition (more than 12 hours of fasting). At about the same time during these 5 days, the volunteer rode a bicycle, as exercise, in the same route for about 20 minutes. The cardiac rate and the blood pressure always showed higher values just after exercise.

Discussion
By retaining the optical configuration of the previous ROS/intracellular Ca 2+ dual monitoring system (CFL-C2000) [10,11] and by improving all other parts of the system, we have developed a new system aiming at measuring the oxidative stress in whole blood easily and accurately. For fulfilling that purpose, the new system is optimized for simultaneously detecting ROS and hROS, elicited mostly by WBC in a tiny amount of blood. Moreover, the system is now able to measure those signals stably for a very long time by an efficient air-flow stirring without evaporating blood samples.
The CL probe, MCLA, is commonly used for detecting one of ROS, O 2 -• [12,13]. It is known to be auto-oxidized and be releasing a nonspecific weak CL, which is not inhibited by SOD [16,18]. It is also known to be reacting with singlet oxygen ( 1 O 2 ), which is not scavenged by SOD [17]. The FL probes, APF and HPF, are also widely used for detecting hROS [14]. APF is known to be oxidized by OCl -, hydroxyl radical ( • OH) and peroxynitrite (ONOO -), releasing a fluorescent molecule fluorescein, while HPF is only oxidized by • OH and ONOO -, releasing the same molecule, fluorescein [14]. Therefore, OClcan be detected by APF and HPF in combination [14]. . Second, the two types of stimulants, fMLP and PMA, activated those cells in quite different time courses (Figs 2A and 3A). The start time of O 2 -• releasing from PMA-stimulated cells was more delayed than that from fMLP. Also, the O 2 -• generating period of PMA-stimulated cells was much longer than that of fMLP. A probable explanation is that these stimulants differently affected protein kinase C, which in turn activated NADPH oxidase for O 2 -• generation [32,33].
Third, we found that ABAH decreased FL APF in isolated neutrophils ( Fig 2B) and sodium hypochlorite solution increased FL APF (Panel b in S4 Fig) in dose dependent manners. In addition, the increase of FL HPF was considerably smaller than that of FL APF (S3 Fig). These results show that the majority of hROS generated by isolated neutrophils were OCl -. Even when whole blood samples were used, the time derivative of FL APF revealed that the increase of FL APF began later than that of CL MCLA (Fig 3A) and ABAH appreciably inhibited FL APF (Fig  3) , whereas FL HPF was not inhibited by ABAH (data not shown). This shows that even in whole blood, FL APF was mostly derived from OClproduced by MPO. However, in the case of blood, there may be more factors contributing to FL APF , such as yet unidentified hROS sources or some unknown artifacts.
Fourth, the generation of hROS, mostly OCl -, was found to be a slower process than the O 2 -• generation. Since instantaneous hROS generation is proportional to the time derivative of FL (i.e., dFL/dt), the dFL APF /dt curve of neutrophil-like cells in Fig 2B clarified that the hROS generation started at about 400 seconds after stimulation (data not shown), and it is later than the peak of the O 2 -• generation which was clearly before 400 seconds (CL MCLA curve in Fig 2A).
This may be explained by the fact that OClis a product of the reaction of O 2 -• catalyzed by MPO.
Since the results obtained with whole blood seem to be more complicated than those obtained with isolated cells, we tried to identify the responsible blood cells generating O 2 -• and OCl -. Results of Fig 5 suggest that the luminescence signals were derived largely from the neutrophils in whole blood we used so far. In fact, both CL MCLA and FL APF were proportional to the counts of neutrophils added to the blood samples (S5 Fig), which enabled us to estimate the CL/FL intensity attributable to the neutrophil count. This was found to be about 90% of the total CL/FL observed. The residual 10% of CL/FL not attributable to neutrophils may include not yet identified hROS sources and/or some unknown artifacts. Although measuring WBC counts has become quite easy by using commercially available devices, such counts cannot directly reveal ROS generation. Many reports suggested that excess amount of ROS by over-activated WBC are linked to various disorders [4,5]. It is also reported that ROS generation can be modified by some factors such as diet and exercise [10,11,[24][25][26][27], which was also shown by our results (S6 and S7 Fig). Therefore, it would be very important to develop a method for detecting ROS directly. For this purpose, isolated neutrophils have so far been used in many studies [24]. However, neutrophil isolation requires several steps and at least a few milliliters of blood. In addition, the procedures might cause stress on neutrophils prior to measurement. Moreover, behavior of neutrophils in blood might be different from that of isolated neutrophils. Since the present system can measure ROS/ hROS in whole blood, these drawbacks would be eliminated. Further, it may be able to monitor other types of cells (such as monocytes) in situations caused by some diseases [34]. In order to prevent such diseases caused by oxidative stress and improve individual physical conditions, it would also be important to detect an imbalance of oxidative states in vivo at early stages.
As /OClgeneration of circulating neutrophils [26] and/or induced enhancement of plasma antioxidant activity [27], depending on a day-to-day variance in the physical conditions. Further studies by increased number of subjects, fixed diet/quantitated exercise are important in the future.
Recently, various biomarkers for excess activities of MPO were developed [35][36][37]. They are good indicators to disease symptoms, which can be regarded as "products" of oxidative stress. On the other hand, what our system can evaluate is ROS and hROS generation, which can be regarded as "substrates" of oxidative stress. Therefore, combining such biomarkers and the present system may reveal the mechanisms and processes of oxidative stress in more detail. We are also planning to clarify detailed relationships between some disorders and ROS/hROS generation.
Another advantage of this system is that it is readily designed to monitor not only blood samples but also various types of cells including adhesive cells. Although we were able to measure O 2 -• generation from floated cultured neurons by using the previous cuvette-type system [38], the present system can do it more easily and effectively. Moreover, since the new system is very sensitive to the optical signals, it may also be applicable to cells not actively producing ROS, such as muscle cells and neurons. Further, this system may be applicable to measurements of more than two luminescence signals simultaneously from various suspension samples including non-biological materials. In spite of these advantages, more improvements are still possible and practical, such as auto-adjustment of air flows and automatic handling of liquids and blood samples. Therefore, we are also hoping to make it even easier to operate so that it can publicly be accessible at various facilities such as citizen centers, drug stores and gyms in order to help people select suitable anti-oxidative foods or evaluate daily physical conditions for preventing diseases. (e) Scatter plots derived from the data of another healthy volunteer. The data were obtained as described in (c). Blood were collected and measured for arbitrary 8 days (within 3 weeks) before diet. (f) Scatter plots after diet. The data were obtained as described in (e). CFL-P2200 was used for the monitoring of luminescence signals and Pentra MS CRP was used for measuring neutrophil concentrations in blood. Scatter plots after exercise. The data were obtained as described in (a). CFL-P2200 was used for monitoring luminescence signals and Pentra MS CRP was used for measuring neutrophil concentrations in blood.

Supporting information
(TIF) S1 Table. Scheme showing the procedure to measure blood samples by using CFL-P2200. The first column in the table shows the reagents in the reaction mixture and the second column shows the stock concentrations of each reagent. The reaction mixture were prepared and measured on the dedicated glass slide and obtained data were analyzed according to the steps 1-11 on the lower side in the table.
(TIF) S1 Movie. Visualization of the air-flow stirring process of the newly developed system. The movie shows how dropped blue ink into a measuring sample on the glass slide is diffused. The ink was fully diffused in about 2 seconds (see also Fig 1B).