Expedited CO2 respiration in people with Miltenberger erythrocyte phenotype GP.Mur

In Southeast Asia, Miltenberger antigen subtype III (Mi.III; GP.Mur) is considered one of the most important red blood cell antigens in the field of transfusion medicine. Mi.III functions to promote erythrocyte band 3 expression and band 3-related HCO3− transport, with implications in blood CO2 metabolism. Could Mi.III affect physiologic CO2 respiration in its carriers? Here, we conducted a human trial to study the impacts of Mi.III expression in respiration. We recruited 188 healthy, adult subjects for blood typing, band 3 measurements, and respiratory tests before and after exercise. The 3-minute step exercise test forced the demand for CO2 dissipation to rise. We found that immediately following exercise, Mi.III + subjects exhaled CO2 at greater rates than Miltenberger-negative subjects. Respiration rates were also higher for Mi.III + subjects immediately after exercise. Blood gas tests further revealed distinct blood CO2 responses post-exercise between Mi.III and non-Mi.III. In contrast, from measurements of heart rates, blood O2 saturation and lactate, Mi.III phenotype was found to be independent of one’s aerobic and anaerobic capacities. Thus, Mi.III expression supported physiologic CO2 respiration. Conceivably, Mi.III + people may have advantages in performing physically enduring activities.

GPA serves as a chaperone that is responsible for the expression of band 3 [13][14][15][16] . GPB, a structural homologue of GPA, expresses one-fifth the quantity of GPA on the red cell surface, and is functionally unclear 14,17 . In people bearing the Mi.III blood type, half or all of their GPB protein molecules on the RBC membrane are replaced by GP.Mur 18 . GP.Mur, encoded for Mi.III, is a naturally-occurred hybrid protein comprised of GPB and GPA. The sequence of the GYP.Mur gene contains a fragment of GYPA inserted in the sequence of GYPB 19 , and this unique glycophorin B-A-B hybrid structure is capable of promoting band 3 protein production and surface expression, just like GPA 18 . Our previous ex vivo study identified superior band 3-related functions (e.g. larger HCO 3 − /Cl − transport capacities) in Mi.III + RBCs 18,20 . Since people with the Mi.III blood type express more band 3 on their red cells, their capacity of CO 2 respiration were expected to be larger 18,20 .
The highest occurrence frequency of Mi.III worldwide is among the Taiwan Ami tribespeople (88%) 8 . Anecdotally, in Taiwan, Ami tribespeople are generally recognized for their athleticism and superior physical endurance, compared to people of other ethnic origins 21 . A disproportionally high percentage of the Amis are professional or elite athletes 22,23 . Ami people represent only 0.6-0.7% of the Taiwanese population, but over 50% of Taiwanese track athletes who had won international competitions were Amis 22 . Conceivably, these general impressions regarding the Ami people (athleticism and superior physical endurance) might be associated with the Miltenberger blood type that most of them bear.
The main goal of this study was to determine the impacts of Mi.III blood type in respiration through a large-scale human study. Faster CO 2 expiration after a brief, physical challenge was observed in people with the Mi.III blood type. Ami tribespeople lacking the Mi.III phenotype did not have this respiratory advantage, indicating that the involvement of other tribal genes or factors is unlikely.

Results
In this study, we recruited 266 non-athlete adults for questionnaire survey, physical assessment, blood tests (including Miltenberger phenotyping, band 3 measurements, and complete blood count), and a 3-minute step test (a standardized cardiorespiratory fitness test). The total sample number was down to 188, after excluding subjects who failed to complete the 3-minute exercise test, or who have one or more of the following conditions: hypertension, BMI > 30, cardiovascular disease(s), asthma, HBV/HCV hepatitis, joint/bone problems, cancer(s), or other major diseases. Among the selected, physically-healthy subjects, 39% are positive with the Mi.III blood type. Their initial physical assessment and RBC-relevant complete blood count data were summarized in Tables 1 & 2. Mi.III and the non-Mi.III participants were not distinguishable in most of the categories in these two Tables. On the other hand, Mi.III+ subjects had slightly higher blood pressure than non-Miltenberger subjects in general (Table 1). Additionally, compared to non-Mi.III female, Mi.III+ female subjects had in average smaller sizes of red cells (Table 2, MCV) and less hemoglobin per RBC ( Table 2, MCH). The smaller MCH values in Mi.III+ female were likely due to their smaller RBC sizes. The differences in MCV and MCH between Mi.III and non-Mi.III female however were not observed among the male subjects (Table 2). Because the differences in MCV and MCH were restricted to female, they were unlikely due to the expression of Mi.III. The gender differences in red cell sizes were probably related to other Ami-unique genetic factors, as many of our Mi.III+ subjects are Ami tribespeople. These variations in blood pressure, MCV and MCH were still within the normal ranges  18 . Here, we used a DIDS labeling method to verify band 3 expression levels in the red cells of these subjects. DIDS is a fluorescent compound that binds covalently and specifically to lysine-539 of band 3 24 . By DIDS labeling, we found that Mi.III+ RBC samples expressed in average 25.2% more band 3 than the blood type-negative samples ( Fig. 1), which is in agreement with our previous measurements by quantitative iTRAQ mass spectrometry, Western blot, and flow cytometry 18,25 . Because many of our Mi.III+ subjects are Ami, Ami tribespeople who do not express Mi.III ("non-Mi.III Amis" in Fig. 1) were compared. As shown in Fig. 1, non-Mi.III Amis expressed significantly lower levels of band 3 than Mi.III+ subjects. These results confirmed the direct association between Mi.III expression and higher band 3 levels, and excluded the possibility that other Ami-unique genes could also promote the expression of band 3.

Expedited CO 2 expiration following exercise in Mi.III+subjects. To determine the impacts of
Mi.III in respiration, the subjects were asked to participate in a standardized 3-minute step test, and their cardiorespiratory parameters (end-tidal CO 2 , breathing rates, heart rates, and blood O 2 saturation) were monitored before and immediately after the test. This 3-minute exercise test evaluates one's cardiorespiratory fitness levels, and is backed up by a large population database on fitness collected by Taiwan National Council on Physical Fitness & Sports 28 . Among the physiological parameters measured, expired or end-tidal CO 2 (Figs. 2,3) and respiration rates (Fig. 4) immediately following the step test were distinguishable between Mi.III and non-Mi.III. Mi.III+ test participants generally exhaled CO 2 faster right after the step test than Miltenberger-negative participants (Fig. 2). Because the amount of exhaled CO 2 (as expressed in end-tidal CO 2 or EtCO 2 in the figures) varied greatly from individual to individual (data not shown), the minute-to minute changes of CO 2 expiration (as expressed in Δ EtCO 2 /min) were calculated per individual, and then grouped for comparison (Fig. 2B) Table 2. A summary of the results of complete blood count (CBC) test for the subjects 3 . 1 MCV (mean corpuscular volume), the average volume of a RBC in fL, was significantly different among Mi.III+ and Mi.III-negative female subjects ( # p < 0.01), but not among the male subjects. 2 MCH (mean corpuscular hemoglobin), the average quantity of hemoglobin in a RBC, had significant variations between Mi.III and non-Mi.III in female (*p < 0.05), but not in male. 3 Data were presented in mean ± S.D.   . That is, the rate changes of CO 2 expiration for Mi.III+ male had decreased by the 2 nd minute post-exercise, but that for Miltenberger-negative male participants remained substantial (Fig. 2B). Similar kinetic differences in CO 2 expiration between Mi.III and non-Mi.III female were observed. From the 1 st to the 2 nd minute post-exercise, Δ EtCO 2 /min for Mi.III+ female subjects had already dropped to 2.4 ± 0.7 mmHg, compared to 4.1 ± 0.5 mmHg at the same time interval for non-Mi.III female (Fig. 2B, right). For both male and female, the kinetic differences in CO 2 expiration (Δ EtCO 2 /min) between Mi.III and non-Mi.III diminished by the 3 rd minute following exercise. By comparing the differences in CO 2 expiration kinetics between male and female, we also found slightly faster recovery in male than in female after the exercise challenge. At the 4 th minute post-exercise, Δ EtCO 2 /min have dropped to 0.6 ± 0.8 mmHg for Mi.III male and 1.1 ± 0.7 mmHg for non-Mi.III male, indicating plateauing CO 2 expiration kinetics (Fig. 2B, left). When Δ EtCO 2 /min approaches zero, there would be no more change in the rate of CO 2 expiration. On the other hand, for female subjects, Δ EtCO 2 / min were 1.9 ± 0.5 mmHg (Mi.III) and 1.5 ± 0.6 mmHg (non-Mi.III) at the 4 th minute post-exercise, and continued to drop (Fig. 2B, right).
To exclude other Ami-related factors than the Mi.III phenotype, we analyzed the data obtained from Ami subjects who lack Miltenberger expression. It is estimated that ~12% of Ami people do not express Mi.III blood type 8 . We found that the non-Mi.III Ami subjects showed significantly slower CO 2 clearance than Mi.III+ subjects, regardless of their gender (Fig. 3). Thus, expedited CO 2 expiration following exercise was unique to people with Mi.III expression, and appeared unrelated to other Ami factors.
Besides CO 2 expiration, breathing rates immediately following exercise were also significantly higher for Mi.III+ subjects, regardless of their gender (Fig. 4). The differences in breathing rates between Mi.III and non-Mi.III diminished by the 2 nd minute following the test (Fig. 4A). We also calculated the minute-to-minute changes of breathing rates, and found that the changes between Mi.III and non-Mi. III male were significantly different during the interval from the 1 st to the 2 nd minute post-exercise (Fig. 4B, left). The kinetics of breathing rates for Mi.III+ male had already dropped significantly by the 2 nd minute post-exercise (Fig. 4B, left). On the other hand, the female subjects, unlike the male, did not show significant differences in the kinetics of breathing rates between Mi.III and non-Mi.III (Fig. 4B, right); this is probably because the differences between Mi.III and non-Mi.III female were relatively small (Fig. 4A). Breathing frequencies are primarily controlled in response to changes of blood CO 2 concentrations by the respiratory center in the central nerve system 33  Aerobic capacity not associated with Mi.III expression. The two other cardiorespiratory parameters measured before and immediately after exercise-heart rate and blood O 2 saturation (SpO 2 ), did not differ between Mi.III and non-Mi.III (Tables 3 & 4). Heart rate is an important indicator for one's aerobic fitness 27 . One's cardiorespiratory endurance index (CREI) score for each test participant could be calculated using three heart rate measurements after the step test, and be compared with the CREI norm charts published online by the Section of Physical Fitness & Sports of the Taiwan Sports Administration 28 . Mi.III and non-Mi.III participants exhibited similar levels of cardiorespiratory fitness (Table 3). Besides, the percent maximal heart rates (% max HR) at any time points measured post-exercise (from the 30 th second to the 5 th minute following the step test) also did not differ significantly between Mi.III and non-Mi.III (Table 3) 29 . So there appeared no associations between Mi.III expression and one's cardiorespiratory fitness.
As for blood O 2 saturation, there was a slight drop to 97.7-98.3% within 30 seconds post-exercise for all the subjects (Table 4). Their SpO 2 levels recovered within 1 minute to the pre-test levels, and remained relatively the same from the 2 nd to the 5 th minute post-exercise in all 4 groups (Mi.III male/ non-Mi.III male/Mi.III female/non-Mi.III female). There were also no significant variations in SpO 2 observed at any time points before and after exercise between Mi.III and non-Mi.III (Table 4). In sum, these results suggested that Mi.III expression only expedited CO 2 respiration (Fig. 2) and did not affect aerobic capacity (Tables 3 & 4).    Distinct blood CO 2 responses following exercise in Mi.III versus non-Mi.III. We further examined how Mi.III erythrocyte expression could affect one's CO 2 respiration by performing additional blood tests immediately before and after 3-minute step exercise on 30 male subjects. The tests included venous blood gas tests (CO 2 , pH, HCO 3 and base excess) and lactate measurements, and the results were summarized in Table 5. For all the subjects tested, 3-minute exercise resulted in small changes in venous CO 2 , HCO 3 − , and pH. This was expected, as the standardized 3-minute stepping test is considered moderate for healthy adults. For most of our test subjects, their heart rates remained below 60% HR max immediately after exercise (Table 3). Anaerobic respiration in untrained people generally begin at 50-60% HR max 30 , so it was not surprising that there were very small increments of plasma lactate found immediately after the step test (Table 5).
We used analyses of covariate (ANCOVA) to test if any of the 5 parameters listed in Table 5 (CO 2 , pH, HCO 3 − , base excess, and lactate) were significantly affected by Mi.III expression following the exercise challenge. We found exercise-induced changes of venous CO 2 to be significantly different between Mi.III and non-Mi.III subjects (*P < 0.05 by ANCOVA in Table 5). The other blood gas parameters and lactate were not affected by Mi.III expression, according to ANCOVA. On the other hand, by comparing the pre-exercise measurements between Mi.III and non-Mi.III subjects, or comparing their post-exercise blood measurements alone using t-test, we found no significant differences between Mi.III and non-Mi. III (Table 5).
We also found that a roughly direct correlation between venous CO 2 and plasma HCO 3 − levels (Fig. 5,  top). This plot shows that despite the considerable variability in blood CO 2 /HCO 3 − levels among individuals, blood CO 2 levels were indiscriminately directly related to blood HCO 3 − levels. To decipher the data, we plotted pre-exercise and post-exercise data points of all subjects again using vector representation (Fig. 5). Each vector composes of the pre-exercise and post-exercise measurements for a single participant. The beginning of a vector is one's pre-exercise value, and the arrowhead is his post-exercise value. The length of a vector indicates the magnitude of change, and its directionality indicates how relevant or irrelevant the 2 parameters (represented by x-and y-axes) are.
From vector analyses (Fig. 5, bottom), we found that majority of the non-Mi.III subjects showed reduced levels of blood CO 2 and HCO 3 − following exercise. In comparison, half of the Mi.III subjects showed small increases of blood CO 2 , and the rest of the Mi.III subjects showed small decreases of blood CO 2 after exercise. The magnitudes of blood CO 2 changes among Mi.III subjects were generally smaller, compared to that in non-Mi.III (Fig. 5, bottom). The findings by vector representation supports the  Table 5. Blood gas and lactate immediately before and after the stepping test. 1 The number inside parentheses indicated the number of blood test subjects. 2 "Difference" referred to the change of the measured values due to stepping for individual subjects; the difference was one's post-exercise value subtracted by his pre-exercise value. Shown here were mean± S.D. for each parameter. 3 To test if the preexercise measurements alone (or the post-exercise measurements alone) were significantly different between non-Mi.III and Mi.III groups, t-test was used. Another statistical method-ANCOVA was used to assess differences between the pre-exercise and the post-excise measurements for each subject, and to determine whether Mi.III significantly affected the parameter following exercise. "n.s.", not significant.  Table 5). Though CO 2 production increased during exercise, experimentally we and others observed reduced blood CO 2 immediately after exercise in most non-Mi.III subjects 31 . It has been suggested that exercise-induced hyperventilation decreases body CO 2 stores, in which bicarbonate is a main constituent 31,32 . This explains the reduction of blood CO 2 and blood bicarbonate that we observed after the moderate exercise test (Fig. 5). Bicarbonate is also utilized to buffer excessive acidic metabolites generated during exercise, which would decrease the content of plasma HCO 3 − following exercise, too 33 . Conceptually, Mi.III RBCs express more AE1 on the cell surface than non-Mi.III RBCs (Fig. 1) 18 , and therefore Mi.III cells are expected to facilitate blood CO 2 /HCO 3 − equilibrium more efficiently. Implied from our results (Fig. 5B), Mi.III expression might somehow affect blood CO 2 sensing. Blood CO 2 levels are primarily set by the respiratory center in the CNS 33 . Because Mi.III expression allows for faster reach of blood HCO 3 − homeostasis and HCO 3 − /CO 2 conversion, the respiratory center in Mi.III+ people would need to cope with faster changes of blood CO 2 levels at times. From what we found experimentally-the different venous P CO 2 responses post-exercise between Mi.III and non-Mi.III (Fig. 5B &  Table 5), conceivably the respiratory center in Mi.III+ subjects could have higher tolerance for blood P CO 2 . This also explains, at least in part, why our healthy Mi.III subjects generally had slightly higher blood pressure than the non-Mi.III subjects (Table 1).

Figure 5. Mi.III and non-Mi.III people exhibited distinguishable blood CO 2 responses following the step test. Top: A roughly direct correlation between venous CO 2 and HCO 3
− levels in Mi.III and non-Mi. III subjects alike. The concentrations of venous CO 2 and HCO 3 − immediately before 3-minute step test (shown in solid symbols) and after the test (empty symbols) were plotted for individual Mi.III (red symbols) and non-Mi.III subjects (black). Bottom: To further dissect the changes of venous CO 2 and HCO 3 − , as well as their correlations, for each subject, his pre-exercise and post-exercise data points were connected and represented by a vector. Almost all the subjects, Mi.III and non-Mi.III alike, showed reduced HCO 3 − levels after exercise. But noticeably, up to 70% of the non-Mi.III subjects had decreased blood CO 2 levels after exercise, and only 40% of the Mi.III subjects showed slight decreases of blood CO 2 following exercise. 50% of the Mi.III subjects showed slightly higher blood CO 2 after exercise. The numbers of the test subjects per group are specified in parentheses.  Table 5). Blood lactate begins to accumulate when tissue oxygen is low and insufficient to provide energy for muscle contraction. Lactate acid is generated from pyruvate during anaerobic glycolysis to release anaerobic energy, and is a major metabolite of anaerobic respiration. Overproduction of lactic acid in muscle tissues (e.g. from strenuous exercise) may lower blood pH, which would then require buffering by metabolites like phosphocreatine, bicarbonate, and proteins 32 . From the vector analyses (Fig. 6), all the subjects, Mi.III and non-Mi.III alike, exhibited reduced plasma bicarbonate (or base excess) with increased lactate post-exercise. Statistical analyses using either t-test or ANCOVA found no significant differences in the lactate levels between Mi.III and non-Mi.III (Table 5). Thus, Mi.III expression does not seem to affect anaerobic respiration (Fig. 6).

Discussion
Mi.III is the second most important erythrocyte blood type following ABO in the fields of transfusion medicine in Taiwan 5,7 . Mi.III-encoded protein, GP.Mur, structurally is a hybrid of glycophorin B and glycophorin A; it functionally resembles that of glycophorin A in facilitating band 3 expression on the erythrocyte membrane 18,25 . Higher expression of band 3 on Mi.III erythrocytes enlarges the capacity of HCO 3 − permeation across the red cell membrane 18 . Since band 3 is the rate-limiting factor for blood CO 2 metabolism 12 , Mi.III+ people are expected to have more efficient CO 2 respiration. Here, we showed experimentally that Mi.III+ subjects exhibited faster CO 2 respiration and clearance immediately following a physical challenge (Figs. 2-4). Exercise-induced venous CO 2 responses were also distinguishable between Mi.III and non-Mi.III (Table 5 & Fig. 5). In contrast, Mi.III phenotype does not affect either aerobic parameters (i.e. heart rates and SpO 2 ) (Tables 3,4) or anaerobic respiration (Table 5 & Fig. 6). To put in context with our current understanding of respiratory physiology, we propose a model to illustrate how CO 2 expiration could be expedited especially in Mi.III+ people following a physical challenge (Fig. 7). In lung alveoli (P CO 2 < 40 mmHg), the driving force to convert blood bicarbonate to CO 2 is substantial. Since this conversion mainly takes place inside red cells, with larger bicarbonate influx capacities, Mi.III+ RBCs in the lungs are expected to produce CO 2 faster than non-Mi.III erythrocytes. We indeed observed faster CO 2 expiration in people bearing the Mi.III blood type (Figs. 2,3). As Figure 7. A proposed model on how CO 2 expiration could be faster in people bearing the Mi.III blood type than those lacking this phenotype. (A) During exercise, there is increasing CO 2 output, along with an increasing demand for ATP expenditure from contracting muscles. CO 2 produced by muscle tissues enters erythrocytes primarily through passive diffusion (shown in red dotted arrows). Erythrocytes contain abundant carbonic anhydrase, which functions to greatly accelerate the chemical conversion between CO 2 and HCO 3 − . Inside circulating red cells in this vicinity, CO 2 is rapidly converted to HCO 3 − by carbonic anhydrase. Much HCO 3 − produced is passively transported out of RBCs via AE1 (shown as the blue gates embedded on the cell membrane). For every bicarbonate ion moving out of RBCs through AE1, a Cl − ion is transported into RBCs to maintain electroneutrality across the cell membrane. The direction and magnitude of anion transport via AE1 follow the Donnan equilibrium, and thus this bi-directional anion transport does not consume energy. Mi.III RBCs express significantly more AE1 molecules on the membrane, allowing more efficient export of bicarbonate and faster reach of HCO 3 − homeostasis across the cell membrane. (B) P CO 2 drops to below 40 mmHg in the lungs. The P CO 2 differences between lung alveoli and capillaries drive CO 2 expiration. As most blood CO 2 is present in the form of HCO 3 − , there is tremendous driving force for HCO 3 − to be converted into CO 2 in the lungs. Notably, the enzymatic activity of carbonic anhydrase to convert HCO 3 − into CO 2 is much faster than the rate of bicarbonate transport through AE1. Larger bicarbonate influxes into Mi.III RBCs that express more AE1, and consequently faster CO 2 production and expiration are expected from Mi.III+ individuals. The directionality of HCO 3 − /Clfluxes through AE1 reverse from tissues to lungs, which is primarily driven by the changes of HCO 3 − concentration gradients across the red cell membrane (as shown left in boxes). The diagram here is intended to illustrate the mechanism of how Mi.III expression could impact physiologic CO 2 respiration, and is not drawn to scale. Mi.III+ RBCs circulate to the tissue vicinities, they are expected to facilitate blood CO 2 /HCO 3 − homeostasis more efficiently than non-Mi.III cells, because of their higher AE1 expression (Fig. 7, top). CO 2 accumulation increases breathing discomfort and reduces one's capability to endure physically 9 . Expedited CO 2 expiration reduces the rate or the degree of CO 2 accumulation in the body. Gorman et al. in their human study using insipiratory resistive loading found that lessening CO 2 accumulation through ventilation prolongs endurance time 9 . Conceivably, when physically challenged, Mi.III+ people with more efficient capabilities of CO 2 respiration are expected to suffer less from CO 2 accumulation or hypercapnia and be more capable of enduring physical stresses. Intriguingly, Ami tribespeople (88% with the Mi.III blood type) historically were recognized by the Japanese colonial government in Taiwan  to outperform other local ethnic groups in forced heavy labor 10,11,26 . One of the main reasons that the Japanese ruling officials in Taiwan at that time favored Ami tribespeople for labor-intensive tasks was Ami's superior physical endurance, according to Yu-Chi Lai, a contemporary historian and anthropologist ethnic of Ami 34 .
To probe into the relations between Mi.III expression and physical endurance, we recently surveyed elite athlete students at the National Taiwan Sport University who had won medals in regional and/or national sports competitions, and found that 22% of them (16/72) have the Mi.III blood type. Moreover, up to a third of the track and field athletes (9/28) are Mi.III+ . Notably, not all these Mi.III+ elite athletes are ethnically Ami. About 11% of the non-Ami elite athletes have the Mi.III phenotype, which is at least double the frequency of Mi.III in the general Taiwanese population (2-6%) 5,7 . The Mi.III phenotype therefore might be associated with athleticism, at least in part through its support of physiologic CO 2 respiration.
The HCO 3 − /Clexchange capacity (permeability) of Mi.III+ RBCs is expandable upon high HCO 3 − / CO 2 stimulation 18 . People generate a lot of HCO 3 − /CO 2 when they are physically stressed (e.g. during rigorous exercise). Our results here indicate that Mi.III expression in erythrocytes facilitate physiologic CO 2 clearance. From this aspect, Mi.III-encoded GP.Mur protein could be perceived as a natural "de-stressor" through its support of band 3 expression. Expression of the Mi.III blood type is thus expected to be beneficial to one's health.

Materials & Methods
This human study aimed to test if GP.Mur phenotype could affect physiologic respiration through its support of band 3 expression.

Ethics Statement.
The study was carried out in accordance with the principles of the Declaration of Helsinki, and was approved by the Institutional Review Board (IRB) of Taiwan Mackay Memorial Hospital (MMH) (MMH-IRB registration number: 11MMHIS038). Written informed consent was obtained from all subjects. Study Design. Local adult subjects between 18-50 years of age were recruited at the three branches of MMHs in Taipei, Tamsui, and Taitung. To minimize interferences from ageing-related pulmonary function decline, people over 50 years old were refrained from participating in this study. The human trial included: (1) a questionnaire survey and initial physical assessments; (2) blood sample studies; (3) a 3-minute, step exercise test that accompanied respiratory physiological measurements, and/or blood tests. A quick prescreening of Mi.III was imposed to ensure that a high percentage of the test subjects have this special blood type. Prior to blood sample collection and the exercise test, each participant was first asked to fill out a questionnaire that surveys one's lifestyle and health conditions. We also conducted initial physical assessments to exclude people who were unfit for the exercise test, i.e. those who were hypertensive (diastolic blood pressure > 90 mmHg or/and systolic blood pressure > 140 mmHg), or/and whose quiet heart rates were above 100.

Blood tests and RBC phenotyping.
Blood tests for all subjects participating in the trial included complete blood count (CBC), erythrocyte phenotyping including Mi.III, and a quantitative assessment of erythrocyte band 3 protein. The Mi.III RBC phenotype was serologically determined with anti-Mur, anti-Hil, anti-Anek, and anti-Mi a antisera, each antiserum targeting a distinct epitope on GP.Mur protein.
Mi.III phenotype was further confirmed by direct blood PCR, as described previously 35 . Additionally, venous CO 2 , bicarbonate, pH, base excess, and plasma lactate were tested in some of the male subjects who volunteered to have their blood withdrawn immediately before and after the 3-minute stepping test.
Assessment of erythrocyte band 3 levels by DIDS labeling. RBC ghosts (membranes) were made by hypotonic rupture, followed by repeated washes to remove hemoglobin 18 . For each sample, RBC ghosts were mixed with an equal volume of 50 μ M DIDS (Sigma-Aldrich, St. Louis, MO, USA), and incubated at 37 o C for 30 minutes 36 . The fluorescent compound DIDS binds specifically to lysing-539 of band 3 24 . DIDS unbound to erythrocyte band 3 was then removed by PBS wash. DIDS labeling of band 3 was quantitated per sample by fluorescence emission using Varioskan TM Flash Multimode Reader (Thermo Scientific, Waltham, MA, USA) at the emission wavelength of 450 nm (excitation at 350 nm). After fluorescence background subtraction, each DIDS emission reading was normalized with respect to the averaged value of all the Mi.III-negative samples, and expressed in percentile (the average of non-Mi. III data = 100%).

The 3-minute step test and physiological measurements. The standardized 3-minute step test
designed by the Taiwan National Council on Physical Fitness & Sports is a modification of the original Harvard test 37 . Like the Harvard step test, this standardized step test is to evaluate one's cardiorespiratory fitness levels. Slightly different from the Harvard test or the YMCA step test 27 , the standardized step test used in Taiwan asks test subjects to step onto a 35-cm high sturdy bench at the rate of 24 steps (96 beats) per minute for continuous 3 minutes 37 . Heart rates were measured immediately following the test, and were used to assess one's cardiorespiratory endurance index (CREI), as follows: where step duration (in seconds) is 180 if the 3-minute test is completed. HR1, HR2, and HR3 refer to heart rates (beats/min) at the 1 st , 2 nd , and the 3 rd minute immediately after the step test. The Department of Physical Education under Taiwan Ministry of Education conducts such standardized fitness tests annually on at least tens of thousands of people from 6-65 years of age. These test results were compiled in the norm charts of cardiorespiratory endurance indices for Taiwan populations 28 . In our study, besides heart rates, end-tidal CO 2 (EtCO 2 ), blood O 2 saturation (SpO 2 ), and respiration rates before and right after the step test were also measured using a portable oxi-capnography instrument (MD-667P; COMDEK Inc., New Taipei City, Taiwan).
Δ EtCO 2 /min Refers to the minute-to minute change of EtCO 2 following the step test. Δ EtCO 2 /min at the N th minute post-exercise was calculated by subtracting EtCO 2 at the N th minute from EtCO 2 at the previous or the (N-1) th minute post-exercise. Thus, Δ EtCO 2 /min represents the rate change of CO 2 expiration.
Statistical analyses. For most data, unless specified otherwise, the levels of significance (p) were determined by two-sample t-test. For analyses of blood test results obtained immediately before and after 3-minute step, separate analyses of covariance (ANCOVAs) were performed to examine the effects of Mi.III phenotype on each parameter. In ANCOVA, the fixed factor was the expression or the absence of Mi.III phenotype, pre-exercise measurements were used as covariate, and post-exercise measurements as dependable variables. A p value less than 0.05 was deemed significant.