Enhanced Ca2+ influx in mechanically distorted erythrocytes measured with 19F nuclear magnetic resonance spectroscopy

We present the first direct nuclear magnetic resonance (NMR) evidence of enhanced entry of Ca2+ ions into human erythrocytes (red blood cells; RBCs), when these cells are mechanically distorted. For this we loaded the RBCs with the fluorinated Ca2+ chelator, 1,2-bis(2-amino-5-fluorophenoxy)ethane-N,N,N′,N′-tetraacetic acid (5FBAPTA), and recorded 19F NMR spectra. The RBCs were suspended in gelatin gel in a special stretching/compression apparatus. The 5FBAPTA was loaded into the cells as the tetraacetoxymethyl ester; and 13C NMR spectroscopy with [1,6-13C]d-glucose as substrate showed active glycolysis albeit at a reduced rate in cell suspensions and gels. The enhancement of Ca2+ influx is concluded to be via the mechanosensitive cation channel Piezo1. The increased rate of influx brought about by the activator of Piezo1, 2-[5-[[(2,6-dichlorophenyl)methyl]thio]-1,3,4-thiadiazol-2-yl]-pyrazine (Yoda1) supported this conclusion; while the specificity of the cation-sensing by 5FBAPTA was confirmed by using the Ca2+ ionophore, A23187.


General
We provide here some context and motivation for studying Ca 2+ and its signalling in metabolism and cell shape by using NMR spectroscopy in the way that we developed in the current project.

Calcium
Mixed phosphates and hydroxides make up ~99% of the total calcium content of the human body. It is remarkable that in the face of this huge pool, it is free Ca 2+ inside cells (ionically dissociated and in the micromolar-nanomolar concentration range) that regulates metabolism, and the organization of cytoskeletal networks.
In human blood plasma, the ~2.2 mM Ca 2+ is distributed as 45% free, 45% bound to protein, and 10% bound to various ligands. Again, it is the free cation in plasma that has various signalling roles in cells that are perfused by it 1 ; e.g., activation of phagocytosis in neutrophils is driven by Ca 2+ uptake into their cytoplasm 2 . Based on extensive work, Lew and Tiffert have identified amongst other effects on RBC ionic homeostasis 3 , the key role of Ca 2+ influx in human RBCs during their senescence (and by analogy, in other mammals) 4 . Consequently, the methods developed in the present work will aid the understanding of factors that affect the role of Ca 2+ in RBC shape, flexibility, and survival in the circulation.

Supplementary Results Part 1
Direct experiment Figure S1 shows 19 F NMR spectra obtained from a sample of RBCs that had been loaded with 5FBAPTA and then compressed in gelatin gel, as used previously when enhanced glycolysis was discovered 5 . Figure S1c was from a control sample that had been prepared identically but not compressed, while Fig. S1b shows a superposition of the two spectra. The spectra clearly show a level of S/N like that reported by others; they imply the need to refine the method of measuring Ca 2+ influx to improve the precision of the rate estimates.
The spectrum in Fig. S1a was recorded in 1.5 h at 20 o C after the sample (RBCs and gelatin) had been compressed and stored for 17 h (overnight) at 4 o C and then maintained at 20 o C for 32.75 h. There are three main spectral features on which to focus (apart from the noise): (1) the broad peak with S/N = 3.6 and width-at-half-height (Dn½) = ~500 Hz was assigned to free 5FBAPTA inside the RBCs and designated a chemical shift (d) = 0.0 ppm; (2) the peak 3 located at d ~ 1.76 ppm to lower frequency with S/N = 4.74 and Dn½ = ~80 Hz was assigned to Fdespite the fact that none of this anion was separately added to the sample; and (3) the broad peak at ~6 ppm with S/N = 2.3 and Dn½ = ~500 Hz was assigned to the intracellular Ca 2+ complex of 5FBAPTA (5FBAPTA-Ca).
The ratio of the peak integrals of 5FBAPTA-Ca to free 5FBAPTA was ~0.19:1.0. Given that the loaded amount of 5FBAPTA-AM was 4.0 mmol (L RBC) -1 , the concentration of the complex was deduced from the integral ratio to be 0.64 mmol (L RBC) -1 . Since this complex primarily accumulated over 32.75 h (ignoring the overnight storage at 4 o C), the net influx rate of the Ca 2+ was calculated to be ~20 µmol (L RBC) -1 h -1 . (In the Discussion, we present caveats associated with this estimate, and its relationship to the operation of the RBCs at 37 o C.). 4 Figure S1. 19  arbitrarily set to a chemical shift of 0.0 ppm is from free 5FBAPTA; and the sharper peak at ~1.9 ppm was assigned to free Ffrom the gelatin. b shows the superposition of a (green) and c (blue). Sample details: 3.

Supplementary Methods
Because of the very low S/N (~4:1) in the 19 F NMR spectra of RBCs loaded with 5FBAPTA, when they were suspended in gelatin, a protocol was devised that raised this to ~15:1. This meant that we had to forego the aim to record spectral time courses of the influx of Ca 2+ from a single sample held in the NMR spectrometer in the 10 mm probe, over 10's of hours. Signal enhancement was ultimately achieved by removing the RBCs from the gelatin, washing them centrifugally, and recording spectra in a 5-mm NMR probe that had substantially higher sensitivity. Specifically, for a 1 M solution of 19 Fin the 10 mm probe (glass 10-mm NMR sample tube) and the optimized (dedicated for 19 F) 5 mm probe the S/N was approximately the same. The 10 mm sample volume was typically 6-10 times greater than for the 5 mm probe for which the volume was much smaller, so if the same amount of solute were present in each sample then the optimized 19 F NMR probe would give ~6-10 times the S/N. In addition, the RBCs in the gels had maximum packing densities of ~20% (Ht = 0.2), so extraction of them from the gel enabled the Ht to be raised four-fold to ~0.8. Moreover, the washing procedure removed any extracellular 5FBAPTA-Ca complex that could have formed if there had been haemolysis, thus avoiding mis-assignment of peak area to intracellular 5FBAPTA-Ca. Recording the spectra at 37 o C, which was not feasible if retention of gel was required, meant that the extraneous 19 Fresonance from the gelatin (Fig. S3) was also removed. Overall, the protocol used for extracting the RBCs from compressed or stretched gels after prolonged time courses used to assess Ca 2+ influx is shown in Fig. S2, with the various steps described in the caption.
6 Figure S2. Experimental protocol for studying stretched (or compressed) RBCs that were extracted from the gel prior to 19 F NMR spectroscopy. The RBCs were loaded with 5FBAPTA, suspended in gelatin that was drawn into a silicone rubber tube that was held inside a glass NMR tube (Step 1). The silicone tube and its contents  RBCs were separated from the liquid gelatin by centrifugation (Steps 7 -9) including resuspension and recentrifugation in warm saline. The RBC pellet was then transferred to a 5-mm NMR tube (Step 10) for acquisition of a 19 F NMR spectrum (Step 11).

Supplementary Results Part 2 and Discussion
Resonance assignment to 19 F -The relatively sharp peak at 1.53 ppm (relative to 5FBAPTA) in the 19 F NMR spectra shown in Fig. S3 was unexpected and demanded assignment. Assignment was made by recording a 19 F NMR spectrum from a sample of gelatin alone. This was done first at 20 o C, the same temperature as used for the series of stretched and compressed RBC experiments.  It is clear from Fig. S3a that the peak at 1.53 ppm was present in the spectrum from pure gelatin; and that addition of NaF to the sample after it was melted at 37 o C and then reset, enhanced the peak. Further addition of NaF led to a direct quantitative increase in peak intensity, thus confirming the assignment to 19 F -. In other words, the Gelita gelatin that we used in all the experiments herein, and previously 5 , is contaminated with F -.  S3a) and the peak from the K45FBAPTA was broadened.
The latter was concluded to be due to enhanced bonding of the F-atoms on the 5FBAPTA to hydrogen bond donors on the gelatin across a distribution of sites rather than due to increased exchange with free 5FBAPTA because the temperature had been lowered and not raised.
In summary: (1) gelatin contains Fthat gives a peak only when the temperature is lowered to ~20 o C, consistent with hydrogen bonding to gelatin's amino acid side chains; while at 37 o C these bonds are weakened, and exchange broadening occurred leading to loss of the peak into the baseline noise. And (2), 5FBAPTA forms hydrogen bonds with gelatin that means that its chemical shift is the same as for the complex with Ca 2+ . Hence, if 5FBAPTA loaded RBCs in a gelatin gel rupture, a 19 F NMR peak that results from the release on 5FBAPTA to the 9 extracellular medium will be due to a combination of 5FBAPTA-gelatin and 5FBAPTA-Ca and possibly also a ternary complex of both species. Figure S4a shows a 19 F NMR spectrum of RBCs that had been loaded with 5FBAPTA, washed with saline (154 mM NaCl) and 1% w/v BSA, and supplemented with Yoda1. There was no Ca 2+ in this medium. The small peak at 5.9 ppm was identified as intracellular 5FBAPTA-Ca, while the adjacent sharper one at 5.5 ppm according to the argument above ( Fig. S3) was from 5FBAPTA-BSA and any 5FBAPTA-Ca that would have been released by haemolysis that occurred during and after the formation of the RBC pellet. Figure S4b was recorded after Ca 2+ was added to the washed RBCs and it gave rise to the sharp peak at 5.5 ppm that was consistent with formation of 5FBAPTA-Ca outside the RBCs, due to haemolysis. The emergence of the broad peak at 5.9 ppm was consistent with Ca 2+ entry via Piezo1 and the formation of intracellular 5FBAPTA-Ca with a concomitant decline in the intracellular 5FBAPTA peak at 0.0 ppm. In Figs. S4c and S4d not only did the intracellular 5FBAPTA-Ca peak grow over time, but another peak at 7.3 ppm did as well;

Resonance assignment to extracellular 5FBAPTA-Ca
while the sharp peak at 5.5 ppm (assigned to extracellular 5FBAPTA-Ca) did not change in intensity. The latter implied that haemolysis had occurred during sample transfer to the 5-mm NMR tube and on subsequent mixing of the RBCs with the Yoda1 stock solution. The assignment of the peak at 7.3 ppm was concluded to be a ternary complex of Ca 2+ , 5FBAPTA and intracellular protein as its growth occurred in parallel with that of 5FBAPTA-Ca (as is also evident in Fig. 3), and the concomitant decline of free 5FBAPTA inside the cells.
Furthermore, addition of K45FBAPTA to the RBC suspension gave rise to a marked increase in intensity of the peak at 5.5 ppm, underscoring the veracity of the assignment of this peak to extracellular 5FBAPTA-Ca, concluded from Fig. S3. In conclusion: (1) in a simple suspension of RBCs, the sharp peak at 5.5 ppm was from extracellular 5FBAPTA-Ca; (2) in the presence of a high concentration of gelatin, this peak could also be due to 5FBAPTAgelatin complex(es); (3) the peak at 5.9 ppm was from intracellular 5FBAPTA-Ca; and (4) the peak at 7.3 ppm was assigned to a ternary complex of protein-5FBAPTA-Ca, where, based on its high abundance, this was likely to have been haemoglobin. This is now discussed in more detail. giving spectrum a. For b, Ca 2+ was added to sample a to give 4 mM averaged over the whole sample being well in excess of the 5FBAPTA. c was recorded from the sample 25 min later; and then K45FBAPTA was added (0.5 mM averaged over the sample volume) giving spectrum d. NMR settings: 5-mm 19 F probe; total time per spectrum, 10 min; other parameters as for Figs. 1, S1 and 3. Figure S5 shows the spectra acquired in an experiment designed to underscore the assignment of the peak at ~7.3 ppm. First, RBCs were loaded with 4 mmol [L RBC] -1 of 5FBAPTA and the spectrum (Fig. S5a) was recorded with the sample at 20 o C. Only the single broad resonance (width at half the peak height; Dn1/2 = 230 Hz) assigned to free 5FBAPTA was evident, as can also be seen in the spectra in Figure 2 that were recorded at 37 o C with slightly smaller Dn1/2 values of ~200 Hz. This is consistent with the higher temperature causing motional narrowing of the peak. Second, the RBCs were lysed by freezing and thawing in liquid nitrogen and spectrum b was recorded also at 20 o C. No new peaks emerged and the only change was a slight narrowing of the peak at 0 ppm (free 5FBAPTA) to Dn1/2 = 180 Hz.

Resonance assignment to intracellular protein-5FBAPTA-Ca
This change can be ascribed to the sample being magnetically more homogenenous because of the freezing and thawing cycle. In other words, there would no longer have been gradients of magnetic susceptibility across the cell membranes. Third, the haemolysate was supplemented with Ca 2+ (~10 mM) to form 5FBAPTA-Ca. Figure S5c shows the resulting spectrum that had no evidence of free 5FBAPTA but the presence of the relatively sharp peak at ~5.7 ppm (as also evident in Figures 1, 3, 4 and S4) was consistent with assignment to the 5FBAPTA-Ca complex as seen in Figure S3e and f. Furthermore, a smaller peak at ~7.3 ppm, with an integral ratio of 0.2:1.0 relative to the adjacent peak from 5FBAPTA-Ca, appeared as well. Because of its appearance only after adding Ca 2+ it was concluded to be a complex of Ca 2+ with 5FBAPTA that was in a different molecular environment. It is important to note that this spectrum was recorded with the sample at 37 o C. Fourth, when the sample was cooled to 20 o C the spectrum lost the peak at ~7.3 ppm. Fifth, the latter peak, however, reappeared when the same sample was warmed back to 37 o C with almost no change in spectral resolution (line width) and relative intensities. The appearance and disappearance of the peak at ~7.3 ppm with cycling the temperature of the sample was reproducible in many similar experiments. binding reaction to give rise to free 5FBAPTA, as seen by the re-appearance of the peak at 0 ppm while the peaks at ~7.3 and 5.9 ppm showed decreased intensity. This suggested that the EDTA was abstracting Ca 2+ away from 5FBAPTA-Ca and also whatever was giving rise to the peak at ~7.3 ppm. Therefore, we concluded that the peak was due to 5FBAPTA-Ca that was in a molecular environment that was different from free solution (see Figure S3) or in cytosolic water space. We further concluded that the most likely new environment was that of a protein complex with 5FBAPTA-Ca; and in the RBC since haemoglobin constitutes ~97% of the protein in the cytoplasm it would be the most likely candidate. Figure S6 shows a 13 C NMR time course of spectra obtained from RBCs that had been loaded with 5FBAPTA and to which [1,6-13 C]D-glucose was added. The insert shows the corresponding progress curve for L-lactate.  .12] t) from eight points centred on 7 h. Yellow highlights the rise from an initial value and then the fall to 0 of 23BPG; and blue highlights the a-anomer of glucose that declined in parallel with the two other glucose peaks. The integral of the methionine resonance at 13.0 ppm remained constants within experimental error. The concentration of glucose, 23BPG and lactate were measured by importing the spectral data from TopSpin into a

C NMR of glycolysis
Mathematica program that performed automatic baseline correction and scaling to the known amount of 13 CH3-L-methionine (that is not metabolised by human RBCs).
The progress curve shows the initial transient stage that is always seen (e.g., 5 ) in such experiments. It is due to the washout of unlabelled intermediates in the pentose phosphate pathway, glycolysis, and 23BPG into 12 C-L-lactate. The rate of lactate production at ~7 h had stabilized to 2.06 mmol (L RBC) -1 h -1 and was similar 5 h later at 2.40 ± 0.13 mmol (L RBC) -1 h -1 . The time course was carried out at a setting of 30 o C on the NMR-probe thermostat, but some RF heating was known to occur from trial experiments so the estimated sample temperature was ~34 o C. The expected rate of lactate production under the same conditions of sample preparation is 3 mmol (L RBC) -1 h -1 so the RBCs in the present study performed glycolysis normally. This was crucial to establish, because an additional feature of the preparation was inclusion of 10 mM dithioerythritol (DTE) and 1% BSA w/v in the suspension medium. The -SH reagent was included to reduce the possibility of echinocyte formation as we considered that this alone might contribute to a change in mean membrane curvature and hence stimulation of Piezo1, quite apart from mechanically induced distortion.
Thus, DTE was included in the gel-RBC stretching experiments reported in Fig. 1.