A spike in mechanotransductive adenosine triphosphate release from red blood cells in microfluidic constrictions only occurs with rare donors

Abstract Objective Wan et al (Proc Natl Acad Sci USA, 105, 2008, 16432) demonstrated that RBCs rapidly and transiently release a spike of 300% more ATP shortly downstream from a short microfluidic constriction where the cells experience a sudden increase in shear stress. More recent work by Cinar et al (Proc Natl Acad Sci USA, 112, 2015, 11783), however, yielded no evidence for a similar spike in ATP release downstream of the constriction. Our aim was to determine whether a transient spike in mechanotransduction is the typical response of RBCs to the sudden onset of increased shear. Methods We investigate ATP release downstream of a microfluidic constriction for 15 participants using a luciferase‐based photoluminescent assay. Results While we observe mechanotransductive ATP release from blood drawn from all donors, we find evidence of a spike in ATP concentration after the microfluidic constriction for only 2 of 15 participants. No clear trends in ATP release are found with respect to the magnitude of the applied shear stress, or to the gender, age, or physical activity (Baecke) index of the donor. Conclusions In aggregate, all data acquired to date suggest that a spike in mechanotransductive ATP due to a suddenly applied increase in shear stress occurs in blood drawn from only 14% of the population.

and many diseases have been linked to decreased ATP release such as cystic fibrosis, 20 pulmonary hypertension, 21 and diabetes. 22 Early studies by Sprague et al 2 investigated deformation-induced release of ATP using filter paper with varied pore diameter. They found that as the pore diameter decreased, the amount of ATP released by RBCs increased.
Subsequent studies by Sprung et al 8

and Edwards et al 7 corroborated
this result using microbore tubing with inner diameters of 25-75 μm. 7,8 This work improved on the filter paper experiments by allowing for collection of ATP release in a continuous flow system, but did not explore geometries consistent with those in the microvasculature.
The advent of microfluidics allowed for an investigation of ATP release in more physiologically relevant geometries. The earliest work investigating ATP release by RBCs in straight microfluidic channels was performed by Price et al, 9 who found that decreasing cross-sectional area and increasing Hct both led to increased ATP release. Later work by the same group 10 utilized a channel that narrowed uniformly along the flow direction. They found that detected ATP in these channels was not governed by the smallest dimension of the channel, but was instead dependent on overall cross-sectional area. Moehlenbrock et al 11 improved on both designs using hydrodynamic focusing to investigate different cross-sectional areas without physically changing the placement or geometry of the microchannel. This approach increased reproducibility, and again indicated that a decrease in crosssectional area increased detected ATP. 11 Subsequent work by Wan et al 12 focused on probing the dynamics of ATP release, using a microfluidic channel with a constriction that served as a mechanical stimulus at a prescribed location. Given the time-invariant flow rate, measurements of ATP concentration as a function of position in the channel (using a standard luciferase-based photoluminescent assay) allowed quantitative estimates of the time required for mechanotransduction to occur. They found that the ATP concentration in the channel remained low upstream of the constriction, reached a sharp peak following a delay after the onset of increased shear stress, and then decreased back to levels consistent with ATP concentration prior to the constriction. The magnitude of this spike varied with the length and width of the constriction and had a maximum value of 1.57 ± 0.17 μmol/L for a channel that was 800 μm long and 20 μm wide. The peak value for this channel, found at x ≈ 1800 μm (measured from the constriction entrance), was approximately 260% greater than the ATP concentration prior to the constriction. Although the results in this study were highly reproducible, with 65 distinct experimental trials, all of the RBCs were drawn from a single participant.
Later work by Cinar et al 13 was performed to identify specific ion channels responsible for the shear-induced mechanotransduction. They utilized the same microfluidic channel dimensions that had yielded the maximum ATP release observed by Wan et al, 12 a constriction 800 μm long and 20 μm wide. Although Cinar et al 13 were able to demonstrate that the Piezo-1 channel was associated with overall increased ATP release, the RBCs in the experiments did not exhibit any pronounced ATP release dynamics. Specifically, their observed peak was 0.81 ± 0.17 μmol/L, only approximately 20% greater than the initial magnitude of ATP. The "spike" was also located further downstream from the constriction at x ≈ 3300 μm; most importantly the magnitude of the peak was smaller than the surrounding error bars, raising the question of whether a spike in mechanotransductive ATP release had even occurred. Notably, Cinar et al 13 report a sample size of n = 11 measurements from 6 participants. The change in magnitude of the peak, in conjunction with the increase in sample size, suggests that the dynamics of ATP release may be more complicated than originally thought.  12 observed in blood drawn from only a single subject, are reproducible in the larger population. Furthermore, doubt has been cast on the mechanism for mechanotransduction, with some evidence pointing toward hemolysis as the primary mechanism for ATP release. 24 However, there is some question as to whether the use of 1-to 14-day-old cells in work investigating hemolysis has obfuscated other mechanisms of ATP release. 25 The main goal of this work was to assess whether a large spike in ATP mechanotransduction occurs downstream of a short constriction, using blood drawn from a larger cohort of participants. We show that RBCs drawn from n = 15 different participants all exhibited shearinduced ATP release, consistent with previous results by numerous researchers who employed a variety of different mechanical stimuli. We go on to address the more specific question of whether or not there is an observable spike in ATP concentration immediately after a very short mechanical stimulus of shear stress using a microfluidic constriction, similar to that reported by Wan et al. 12 We find that for 13 of the 15 participants, the ATP concentration is spatially invariant along the entire length of the channel, that is, no mechanotransductive spike was observed. We find no evidence that the magnitude of the flow rate or the characteristics of the donors (age, gender, and physical activity index) has a statistically significant impact on the spike (or lack thereof) in ATP release. Taking this work in conjunction with the previous efforts of Wan et al 12 and Cinar et al, 13 it appears that a pronounced spike in ATP mechanotransduction in response to a suddenly applied shear increase only occurs in RBCs drawn from about 14% of the population.

| MATERIALS AND METHODS
Microfluidic channels were fabricated with PDMS using soft photolithographic techniques by Duffy et al. 26 The channels have a constriction from 100 to 20 μm with a taper angle of 60°. The height of the channel is 40 μm, and the length of the constriction is 800 μm.
A schematic of the channel is shown in Figure 1A. We use this geometry to be consistent not only with Wan et al 12  Luciferin/Luciferase, microfluidic channel, and photon counting system as for the RBC experiments (described below). Representative calibration curves are found in Figure 2; note that the spatial variation in PPS is due to the difference in cross-sectional area between the constriction region and the wide region of the channel. When the cross-sectional area exposed to the photomultiplier tube is larger, more light will be captured by the photomultiplier tube. Thus, two calibration curves are calculated from each scan along the channel, the first is for the constriction region, and the second is for the wide region of the channel (cf. Figure 2B,D).

Blood was collected from 15 participants under pre-approved
Institutional Review Board protocols for the study of ATP release by RBCs. A tourniquet was used during blood collection, and whole blood was collected into tubes containing heparin to prevent coagulation.
To isolate RBCs, whole blood was centrifuged at 500 g at 20°C for 1 minute. The supernatant and buffy coat layers were aspirated off, and packed RBCs were resuspended and washed three times in PSS.
These centrifugation parameters are consistent with, or gentler than, those used in previous studies investigating ATP release by RBCs 2,7-12,23 to reduce exposure of RBCs to shear stress prior to microfluidic experiments. These centrifugation parameters produce less than 0.01 μmol/L of ATP as measured in the packed RBCs (ie, over an order of magnitude less than ATP measured in this system, cf.

| RESULTS
Representative plots of the ATP concentration vs position downstream from the constriction for RBCs drawn from three different  RBCs from this minority of participants yielded an observable spike in ATP concentration following the constriction region (as typified in Figure 3A).
As the experimental results suggest that RBCs from some participants yield an appreciable mechanotransductive spike, while most do not, a natural question is as follows: what makes these participants different? Information about each participant is listed in Table S1. Here, we probe whether age, gender, or physical activity index affect ATP release. We compare [ATP] max vs age for Q = 2.5 μL/min in Figure 6A.
There is no significant evidence that increasing age corresponds to decreased ATP release by RBCs (R = −0.25, P = .10) for participants  30 However, the majority of our participants are between the ages of 20-35, and only two participants are above the age of 40.
We analyzed differences in magnitude of ATP released by males and females. A comparison of the two is shown as an inset in Figure 6A. There are two boxplots representing the distribution of [ATP] max for male (blue) and female (red) participants at a flow rate of Q = 2.5 μL/min. We find no statistically significant differences (P = .40) in detected ATP for n m = 9 male and n f = 6 female participants.
Finally, to probe the effect of physical activity level on ATP release by RBCs, we used the BI as a measure of physical activity. BI was determined using the Modified Baecke Questionnaire. 31,32 We compare [ATP] max vs BI for Q = 2.5 μL/min in Figure 6B, which shows a lack of correlation between physical activity and ATP release by RBCs for the range of BIs explored here (R = −0.02, P = .87).  13 with error bars corresponding to the standard deviation of n = 11 measurements from 6 participants. For measurements taken in this work, the initial concentration of ATP is calculated as the average of all measurements taken before the constriction. For previous work, initial concentration of ATP is taken as the first measurement (cf. Figure 3) To estimate the size of this subpopulation, for the sake of specificity, we restrict attention to ATP spikes such that the average of an individual participant's trials yielded a ratio of peak to initial ATP concentrations that exceed a factor of two, that is,

| DISCUSSION
We note this definition is quite conservative compared to the ratio of ~4 originally observed by Wan et al (cf. Figure 3). Another key feature of our results is that, while a slight increase in the median ATP concentration is seen when the flow rate was increased by a factor of 20 from 0.25 to 5 μL/min, it is not statistically significant. This observation broadly agrees with expectations based on previous work; several groups have found that increasing either  comparisons of ATP release between healthy and rigidified cells. 5,[9][10][11] Here, we follow the methodology of Wan et al 12  Another consideration in this work is that the average ATP concentration appears slightly lower than previous results using the same system. As we use the same Hct, it is unlikely this is due to a difference in total number of RBCs. However, it is anticipated that increases to the chosen Hct would increase the detected concentration of ATP in the channel, consistent with earlier investigations. 7,9 It is also unlikely that the differences in detected ATP are due to differences in enzyme efficacy. As stated previously, we prepare a de novo calibration curve for each day of experiments to confirm enzyme activity.
Other possible explanations for the lack of ATP spikes observed here are worth considering. In our work, a tourniquet was used during there is evidence that Ca 2+ influx may be important in the ATP release mechanism. Future work imaging the intracellular Ca 2+ concentration as a function of position may provide further understanding of the underlying differences in participants who display a mechanotransductive spike and those who do not. These future studies would also benefit from additional tests to see whether lack of a mechanotransductive spike is correlated with lack of the ability to transduct ATP in response to other physiological stimuli such as hypoxia or changes in pH.