AltitudeOmics: Resetting of Cerebrovascular CO2 Reactivity Following Acclimatization to High Altitude

Previous studies reported enhanced cerebrovascular CO2 reactivity upon ascent to high altitude using linear models. However, there is evidence that this response may be sigmoidal in nature. Moreover, it was speculated that these changes at high altitude are mediated by alterations in acid-base buffering. Accordingly, we reanalyzed previously published data to assess middle cerebral blood flow velocity (MCAv) responses to modified rebreathing at sea level (SL), upon ascent (ALT1) and following 16 days of acclimatization (ALT16) to 5260 m in 21 lowlanders. Using sigmoid curve fitting of the MCAv responses to CO2, we found the amplitude (95 vs. 129%, SL vs. ALT1, 95% confidence intervals (CI) [77, 112], [111, 145], respectively, P = 0.024) and the slope of the sigmoid response (4.5 vs. 7.5%/mmHg, SL vs. ALT1, 95% CIs [3.1, 5.9], [6.0, 9.0], respectively, P = 0.026) to be enhanced at ALT1, which persisted with acclimatization at ALT16 (amplitude: 177, 95% CI [139, 215], P < 0.001; slope: 10.3%/mmHg, 95% CI [8.2, 12.5], P = 0.003) compared to SL. Meanwhile, the sigmoidal response midpoint was unchanged at ALT1 (SL: 36.5 mmHg; ALT1: 35.4 mmHg, 95% CIs [34.0, 39.0], [33.1, 37.7], respectively, P = 0.982), while it was reduced by ~7 mmHg at ALT16 (28.6 mmHg, 95% CI [26.4, 30.8], P = 0.001 vs. SL), indicating leftward shift of the cerebrovascular CO2 response to a lower arterial partial pressure of CO2 (PaCO2) following acclimatization to altitude. Sigmoid fitting revealed a leftward shift in the midpoint of the cerebrovascular response curve which could not be observed with linear fitting. These findings demonstrate that there is resetting of the cerebrovascular CO2 reactivity operating point to a lower PaCO2 following acclimatization to high altitude. This cerebrovascular resetting is likely the result of an altered acid-base buffer status resulting from prolonged exposure to the severe hypocapnia associated with ventilatory acclimatization to high altitude.

Previous studies reported enhanced cerebrovascular CO 2 reactivity upon ascent to high altitude using linear models. However, there is evidence that this response may be sigmoidal in nature. Moreover, it was speculated that these changes at high altitude are mediated by alterations in acid-base buffering. Accordingly, we reanalyzed previously published data to assess middle cerebral blood flow velocity (MCAv) responses to modified rebreathing at sea level (SL), upon ascent (ALT1) and following 16 days of acclimatization (ALT16) to 5260 m in 21 lowlanders. Using sigmoid curve fitting of the MCAv responses to CO 2 , we found the amplitude (95 vs. 129%, SL vs. ALT1, 95% confidence intervals (CI) [77,112], [111,145], respectively, P = 0.024) and the slope of the sigmoid response (4.5 vs. 7.5%/mmHg, SL vs. ALT1, 95% CIs [3.1, 5.9], [6.0, 9.0], respectively, P = 0.026) to be enhanced at ALT1, which persisted with acclimatization at ALT16 (amplitude: 177, 95% CI [139,215], P < 0.001; slope: 10.3%/mmHg, 95% CI [8.2, 12.5], P = 0.003) compared to SL. Meanwhile, the sigmoidal response midpoint was unchanged at ALT1 (SL: 36.5 mmHg; ALT1: 35.4 mmHg, 95% CIs [34.0, 39.0], [33.1, 37.7], respectively, P = 0.982), while it was reduced by ∼7 mmHg at ALT16 (28.6 mmHg, 95% CI [26.4,30.8], P = 0.001 vs. SL), indicating leftward shift of the cerebrovascular CO 2 response to a lower arterial partial pressure of CO 2 (PaCO 2 ) following acclimatization to altitude. Sigmoid fitting revealed a leftward shift in the midpoint of the cerebrovascular response curve which could not be observed with linear fitting. These findings demonstrate that there is resetting of the cerebrovascular CO 2 reactivity operating point to a lower PaCO 2 following acclimatization to high altitude. This cerebrovascular resetting is likely the result of an altered acid-base buffer status resulting from prolonged exposure to the severe hypocapnia associated with ventilatory acclimatization to high altitude.
During prolonged exposure to hypoxia, the hyperventilationinduced hypocapnia and associated respiratory alkalosis lead to a compensatory renal HCO − 3 excretion, thereby lowering [HCO − 3 ] and returning pH toward homeostatic values in the arterial and CSF compartment (Dempsey et al., 1974(Dempsey et al., , 1978Forster et al., 1975). At sea-level, reducing CSF [HCO − 3 ] with oral ammonium chloride administration increases the cerebrovascular and ventilatory responsiveness to CO 2 , and elicits a resetting of these responses to lower PaCO 2 values (Fencl et al., 1969). This reduction in CSF [HCO − 3 ] and associated base buffering change leads to a greater increase in [H + ] for a given rise in PaCO 2 (Siesjö, 1972). Such changes in acid-base buffering in the CSF compartment accounts for the enhanced CBF and ventilatory responsiveness to CO 2 observed at high altitude (Mathew et al., 1983;Schoene et al., 1990;Fan et al., 2014), along with the leftward shift in the ventilatory CO 2 sensitivity (Crawford and Severinghaus, 1978;Mathew et al., 1983;Fan et al., 2012Fan et al., , 2014. Pioneering work by Severinghaus (1965) estimated that CBF would be 78% above pre-altitude values if arterial partial pressures of O 2 (PaO 2 ) and CO 2 were returned to sea-level values following 3-5 days at 3800 m. Similarly, data from Fan et al. (2010Fan et al. ( , 2012 indicated a leftward shift in the CBF response to CO 2 following prolonged high altitude exposure. Since there is a delayed onset of compensatory metabolic acidification we would expect to observe no such resetting of the cerebrovascular CO 2 reactivity during acute high altitude exposure. However, no studies have identified the magnitude of this cerebrovascular resetting during acute and chronic exposure to high altitude. Under constant arterial blood pressure, the overall relationship of PaCO 2 and CBF is sigmoidal with a linear portion between PaCO 2 of 25-65 mmHg (Madden, 1993). While the majority of the literature is focused on the linear portion of this response, sigmoid fitting might be more appropriate for the assessment of cerebrovascular CO 2 reactivity at high altitude due to the severe hypocapnia resulting from ventilatory acclimatization to high altitude. Further, the use of a linear model may miss a number of physiological parameters (e.g., the optimal operating point) and physical constraints of the cerebral vasculature (e.g. vessels' finite dilation/constriction and associated vascular reserves; Battisti-Charbonney et al., 2011). Accordingly, sigmoid curve fitting could provide additional insight to the changes in cerebrovascular CO 2 reactivity at high altitude (such as changes in vascular dilation or constriction reserves and a shifting of the operating point).
Using the sigmoid curve fitting method described by Battisti-Charbonney et al. (2011), we reanalyzed our cerebrovascular CO 2 reactivity data from rebreathing tests as previously reported in Fan et al. (2014). We tested the hypotheses that: (1) ascent to high altitude is accompanied by a leftward shift as well as an increase in the cerebrovascular response to CO 2 upon ascent to 5260 m and following acclimatization; (2) these changes in cerebrovascular CO 2 reactivity would be abolished once we account for the acidbase change by plotting CBF against [H + ] instead of PCO 2 ; and (3) the use of sigmoid fitting will reveal changes in vascular reserve at high altitude.

Subject Recruitment and Screening
This study was conducted as part of the AltitudeOmics project. Following institutional ethics approval, young (19-23 years old), healthy, sea level residents were recruited from the greater Eugene, Oregon area (elevation 130 m). Potential subjects were screened to exclude anyone who was born or had lived at altitudes >1500 m for more than 1 year or had traveled to altitudes >1000 m in the past 3 months. After obtaining written consent, physical exams and the Army Physical Fitness Test (push-ups, sit-ups, and a 3.2-km run) were performed to verify health and fitness status (Subudhi et al., 2014b).

Ethical Approval
The study was performed according to the Declaration of Helsinki and was approved by the Institutional Review Boards of the Universities of Colorado and Oregon and by the Human Research Protection Office of the U.S. Department of Defense. All participants were informed regarding the procedures of this study, and written informed consent was given prior to participation.

Experimental Design
After familiarization with the experimental procedures outlined below (visit one), twenty-one subjects underwent three experimental trials: near sea level (SL: 130 m, barometric pressure: 749 mmHg) and on day 1 and 16 following ascent to 5260 m (ALT1 and ALT16). Approximately 4 weeks following SL measurements in Eugene, Oregon, the subjects were flown to La Paz, Bolivia. They spent two nights at low altitude (1525 m in Coroico, Bolivia) before being driven to the Chacaltaya Research Station at 5260 m, while breathing supplemental oxygen (ALT1). Subjects acclimatized to altitudes ranging from 3800 to 5260 m over the next 15 days, with most of the time (75%) spent at 5260 m. On the 16th day (ALT16), measurements were repeated (Subudhi et al., 2014b).

Experimental Protocol
A 22-gauge catheter was inserted into a radial artery at least 1 h prior to instrumentation. Subjects were seated in an upright position for 15 min, while sensors were placed to measure physiological variables of interest (Subudhi et al., 2014b). Following 10-15 min of quiet rest in a seated position, each experimental testing session comprised of: (a) instrumentation; (b) 10 min room air baseline; (c) 10 min steady-state with partial pressure of end-tidal CO 2 (PETCO 2 ) and O 2 (PETO 2 ) clamped at 40 mmHg and 300 mmHg, respectively, and (d) assessment of cerebrovascular CO 2 reactivity with a modified rebreathing test. For each subject, all ALT measurements were carried out around the same time of day to minimize any confounding effect by circadian rhythm. Measurements were taken upon arrival at ALT1 to minimize the influence of acute mountain sickness (AMS). Likewise, no symptoms of AMS were observed at ALT16.
Throughout the protocol, the subjects sat upright and breathed through a mouthpiece attached to a two-way nonrebreathing valve (Hans-Rudolph 2700, Hans-Rudolph Inc., Shawnee, KS, USA). The breathing circuit allowed switching from room air to either an end-tidal clamping system or a rebreathing system.

Modified Rebreathing Method
The rebreathing bag was filled with gas to achieve inspired PCO 2 and PO 2 of 0 and 300 mmHg, respectively, upon rebreathing at each altitude. Hyperoxia was used to abolish the potentially confounding influence of changes in PaO 2 at SL, ALT1, and ALT16 on our cerebrovascular CO 2 reactivity estimates. Subjects were instructed to hyperventilate for 3 min to lower and then maintain PETCO 2 at ∼18 mmHg at both sea level and 5260 m (in background PETO 2 >250 mmHg). Subjects were then switched to the rebreathing bag, and following two initial deep breaths to mix the gas from the bag with that in the respiratory system, they were instructed to breathe ad libitum. The rebreathing tests were terminated when PETCO 2 reached 50 mmHg, PETO 2 dropped below 200 mmHg or the subject reached the end of his/her tolerance.

MEASUREMENTS Cerebral Blood Flow Velocity
Middle cerebral artery velocity (MCAv, an index of cerebral blood flow) was measured in the left middle cerebral artery using a 2-MHz pulsed Doppler ultrasound system (ST3, Spencer technology, Seattle, WA, USA). The Doppler ultrasound probe was positioned over the left temporal window and held in place with an adjustable plastic headband (Marc 600 Headframe, Spencer technology, Seattle, WA, USA). The signal was acquired at depths ranging from 43 to 54 mm. Signal quality was optimized and an M-mode screen shot was recorded to facilitate subsequent probe placements.

Arterial Blood Pressure
Beat-to-beat arterial blood pressure (ABP) was measured from an arterial catheter inserted in a radial artery, and connected to a calibrated, fluid-filled, disposable pressure transducer positioned at the level of the heart (DELTRAN II, Utah Medical, Salt Lake City, UT, USA).

End-Tidal Gases
PETCO 2 and PETO 2 were measured using fast responding gas analyzers (O 2 Cap Oxygen analyzer, Oxigraf, Mountain View, CA, USA). The gas analyzers were calibrated using gas mixtures of known concentrations of O 2 and CO 2 prior to each testing session.

Arterial Blood Gases
For each subject, arterial blood samples were taken four times at SL, ALT1 and ALT16 (room air breathing, PETCO 2 clamped at 40 mmHg, hyperventilation and end of rebreathing). Hydrogen ion concentration ([H + ]) was calculated from the arterial pH values using the Siggaard-Andersen equation (OSA.exe, freely available at http://www.siggaard-andersen.dk). The [H + ] values were plotted against PaCO 2 (Figure 1) and the relationship between these two variables was obtained using linear regression (Prism 6.0b, GraphPad Software Inc., San Diego, CA, USA). The equation of this [H + ]-PaCO 22 slope was subsequently used to calculate the corresponding [H + ] values for the range and the midpoint of the sigmoid curves for each rebreathing maneuver (Figure 1).

Data Acquisition
All analog data were sampled and recorded at 200 Hz on a PC for off-line analysis (ADInstruments Powerlab 16/30, Dunedin, New Zealand).

Rebreathing Analysis (Figure 2)
As first detailed by Battisti-Charbonney et al. (2011), the following assumptions were made regarding the analysis of the rebreathing data: (i) hyperventilation lowers PETCO 2 sufficiently to produce full hypocapnia-induced vasoconstriction; (ii) as PETCO 2 increases from a hypocapnic value during hyperventilation to increasingly hypercapnic values during rebreathing, full vasodilation is achieved; the shape of this relationship is assumed to be sigmoidal with a linear portion in the middle of the curve (Claassen et al., 2007); and (iii) as hypercapnia increases beyond the point of maximal CO 2induced vasodilation, CBF becomes directly dependent on perfusion pressure. Based on these assumptions, cerebrovascular CO 2 reactivity can be summarized as the vasodilatory and vasoconstrictive response of the cerebral blood vessels. The "midpoint" of the sigmoid curve then represents the optimal point of the vessels' capacity to dilate and constrict (i.e., the optimal operating point), while the upper and lower plateaus of the sigmoid response represent the physical constraints of the cerebral vascular reserve. Meanwhile, the range of the sigmoid curve reflects the PCO 2 range of the linear portion of the sigmoid curve, and the amplitude represents the overall extent of this response.
The rebreathing data were first reduced to 1-s averages across the entire rebreathing period. MCAv values were then converted to a percentage change from MCAv values observed at resting PETCO 2 . Since the ABP-CO 2 response is enhanced at high altitude Fan et al., 2014;Willie et al., 2015) and known to potentially confound estimates of cerebrovascular CO 2 reactivity (Betz, 1968;Regan et al., 2014), a PETCO 2 threshold (T ABP ) was first identified as the point from which mean ABP rose more than 0.5 mmHg per mmHg change in PETCO 2 . This threshold was identified by fitting the mean ABP response to PETCO 2 , with two straight lines above and below the T ABP using linear regression, and restricting the slope of the lower portion to < 0.5 mmHg change in ABP per mmHg change in PETCO 2 . The MCAv response to PETCO 2 was then divided into two portions below and above the T ABP and the portion of MCAv-PETCO 2 response below T ABP was fitted with a sigmoid curve as described by Battisti-Charbonney et al. (2011), thereby minimizing the confounding influence of ABP on the MCAv-CO 2 response. To minimize the sum of squares for nonlinear regression (Levenberg-Marquardt algorithm) we used the equation: Where MCAv is the dependent variable in %, PETCO 2 is the independent variable in mmHg, minimum is the minimum MCAv determined from the mean MCAv of the hypocapnic (hyperventilation) region, amplitude is the amplitude of the response (i.e., the maximum MCAv value), midpoint is the PCO 2 value at the center of the sigmoid curve, and range is the PCO 2 range of the linear portion of the sigmoid (an inverse reflection of the slope of the linear portion). The slope of the sigmoid curve (i.e., cerebrovascular CO 2 reactivity) is subsequently calculated from the linear portion of the response. The vasoconstriction reserve was calculated as the difference between minimum MCAv and MCAv at resting PaCO 2 , while the vasodilation reserve was calculated as the difference between the MCAv at T ABP (or amplitude if no T ABP was observed during rebreathing) and MCAv at resting PaCO 2 , expressed at percentage MCAv (Figure 2).
The alignment and fitting processes were performed using custom written graphic analysis routines (LabVIEW, Nation Instruments, Austin, TX, USA). The fitting program provided an r-square of the sigmoid and linear fits.

Statistical Analysis
Due to logistical impacts on planning and transportation, not all subjects were able to participate in all high-altitude studies; please see the Figures and Table for complete sample size reporting for each procedure. After calculating descriptive statistics [mean, 95% confidence interval (CI)], mixed model linear regression analysis (diagonal repeated covariance assumed) (IBM R SPSS R Statistics version 21, IBM R Corporation, Armonk, NY, USA) was used to evaluate the effects of altitude exposure and acclimatization (between SL, ALT1 and ALT16) on the midpoint, amplitude and range of the fitted sigmoid curves. Post hoc tests were performed on the midpoint, amplitude and range of the sigmoid curves between SL, ALT1 and ALT16 using the Holm-Sidak adjustment for multiple comparisons (α = 0.05) (IBM R SPSS R Statistics version 21, IBM R Corporation, Armonk, NY, USA). Cohen's d, as a measure of effect size, was calculated as d = (M 1 -M 2 )/SD, where M 1 and M 2 are means of group 1 and 2; SD is standard deviations of pooled data. The effect size was considered negligible when d < 0.2, small when d ≥ 0.2, moderate when d ≥ 0.5, large when d ≥ 0.8, and very large when d ≥ 1.3 (Sullivan and Feinn, 2012). Results were considered significant at the alpha level <0.05. A priori power calculations (α = 0.05, β = 0.20) were used to determine sample size and limit Type II error.

RESULTS
Detailed baseline characteristics of the 21 (nine women; age 21 ± 1 years) subjects participating in AltitudeOmics were presented elsewhere (Subudhi et al., 2014a). All 21 subjects performed the modified rebreathing test at SL and ALT16. However, 4 of the 21 subjects did not perform the modified rebreathing at ALT1. After excluding data due to unstable ABP during rebreathing, 13, 9, and 7 trials met the inclusion criteria at SL, ALT1, and ALT16, respectively, and were included in the analysis. Sigmoid curve FIGURE 2 | A representative example of an individual rebreathing test and sigmoid curve fitting upon ascent to 5260 m (ALT1).
Frontiers in Physiology | www.frontiersin.org fitting was applied to the remaining trials with R 2 of 0.89 ± 0.05 (mean ± SD).

Cerebrovascular CO 2 Reactivity
No difference was observed in the PETCO 2 range of the linear portion of the sigmoid between SL and ALT1 (d = 1.0, P = 0.149). Likewise, the PETCO 2 range was unaltered following acclimatization at ALT16 when compared to SL (d = 1.2, P = 0.060) or to ALT1 (d = 0.3, P = 0.149 vs. ALT1, Table 1). Upon ascent to 5260 m (ALT1), the amplitude of the MCAv response was increased by ∼34% (d = 5.8, P = 0.024 vs. SL), and further increased by ∼82% at ALT16 (d = 11.1, P < 0.001 vs. SL, Table 1). However, the amplitude of the sigmoid MCAv response was not different between ALT1 and ALT16 (d = 7.3, P = 0.101, Table 1). As a result of these increases in amplitude of the MCAv response, the linear portion of the sigmoid curve was enhanced by ∼67% at ALT1 (d = 1.8, P = 0.026) and by ∼129% at ALT16 (d = 3.0, P = 0.003) compared to SL, while no difference was observed from ALT1 to ALT16 (d = 1.7, P = 0.175, Table 1). The PETCO 2 midpoint of the sigmoid curve was unchanged upon ascent to 5260 m (d = 0.5, P = 0.982 vs. SL), while it was reduced by ∼8 mmHg (i.e., shifted leftward) following acclimatization at ALT16 (d = 3.3, P = 0.001 vs. SL, Table 1). Accordingly, the PETCO 2 midpoint was lower at ALT16 by ∼7 mmHg compared to initial exposure to high altitude (d = 3.1, P = 0.005 vs. ALT1, Table 1). These findings indicate that cerebrovascular CO 2 reactivity was enhanced upon ascent to 5260 m, resulting in both upward shifting and steeping of the sigmoid curve compared to sea level (Figure 3). Following acclimatization, at ALT16, there was a leftward shift of the sigmoid curve ( Figure 4A), while, cerebrovascular CO 2 reactivity did not differ greatly between ALT1 and ALT16 (Figure 3).

DISCUSSION
In the present study, we examined the changes in cerebrovascular parameters during modified rebreathing using sigmoid curve fitting analysis. Using sigmoid fitting, we found: (1) the increase in cerebrovascular CO 2 reactivity at ALT1 is mediated by an enhanced sensitivity of the cerebral blood vessels to H + ( Figure 4B); (2) there is a leftward shift of MCAv-CO 2 response with acclimatization at ALT16 (Figure 4A), indicating a resetting of the cerebrovascular CO 2 reactivity sigmoid midpoint to a lower PaCO 2 with chronic hypoxia; (3) this leftward shift in cerebrovascular CO 2 reactivity is abolished once we account for the changed acid-base buffering status; and 4) despite this cerebrovascular resetting, the vascular constriction reserve is reduced at high altitude. From our findings, it appears that this cerebrovascular resetting to a lower range of PaCO 2 , mediated by changes in acid-base buffering, is likely a serendipitous consequence of acclimatization to high altitude. Importantly, this resetting restores cerebrovascular control during the persistent  severe hypocapnia associated with prolonged exposure to high altitude. CBF regulation during both acute and chronic hypoxia can be summarized as the net effect of hypoxia-induced vasodilation and hypocapnia-induced vasoconstriction of the cerebral blood vessels (Xu and Lamanna, 2006;Brugniaux et al., 2007). During initial exposure to severe hypoxia, the vasodilatory drive in the cerebral blood vessels typically exceeds that of the constrictor drive, thereby elevating CBF (Severinghaus et al., 1966;Subudhi et al., 2014c). During acclimatization, the progressive increase in ventilatory drive lowers PaCO 2 and elevates PaO 2 (Brugniaux et al., 2007), resulting in greater vasoconstrictor drive and restoring of CBF to pre-exposure values (Severinghaus et al., 1966;Subudhi et al., 2014c). In the absence of a cerebrovascular resetting, the progressive increase in ventilatory drive and associated hyperventilationinduced hypocapnia with acclimatization would eventually lead to cerebral hypoperfusion. Therefore, the observed resetting of cerebrovascular CO 2 reactivity to a lower PCO 2 operating point mitigates the adverse effects of hypocapnia-induced vasoconstriction, thus enabling the cerebral vessels to better regulate CBF during prolonged high altitude exposure.

Altered Acid-Base Buffering
The CSF is a bicarbonate-containing fluid which acts as an important acid-base buffer in the regulation in CSF pH (Siesjö, 1972). Since there are negligible concentrations of protein and other non-bicarbonate buffers in the CSF (Davson, 1967), changes in [HCO − 3 ] reflect changes in buffer base (Siesjö, 1972). As elegantly summarized by Fencl et al. (1969), the regulation of CBF and ventilation are components of the same homeostatic system, where the input consists of [H + ] in the CSF compartment, which is determined by [HCO − 3 ] and PCO 2 . In this system, the respiratory loop regulates PaCO 2 (and therefore [H + ] controlling CBF), while the cerebrovascular loop regulates the PCO 2 gradient between arterial and CSF compartment (and therefore [H + ] at the level of the central chemoreceptors). Importantly, the bicarbonate concentration provides the link between the two regulatory loops of this system. Under conditions of metabolic acidosis or alkalosis, CSF [HCO − 3 ] is regulated to reduce CSF pH variations to one tenth of that occurring in the arterial compartment (Fencl et al., 1969;Dempsey et al., 1978). Since a CSF pH change of 0.2 unit covers the entire ventilatory response from life-threatening hypoventilation to maximal hyperventilation (Pappenheimer, 1967), a close regulation of CSF pH is critical for normal respiratory function. As CSF [HCO − 3 ] is an important link between cerebrovascular and ventilatory control, any bicarbonate-mediated changes in ventilatory CO 2 sensitivity should be reflected in cerebrovascular CO 2 reactivity. Moreover, changes in CSF [HCO − 3 ] would lead to proportional changes in the PCO 2 range within which CBF and ventilation are regulated. In support, we found there was a leftward shift of the PaCO 2 -[H + ] relationship at ALT16 (Table 1).

Enhanced Sensitivity to H + during Acute Exposure
During respiratory alkalosis, as observed during hypoxic exposure, a reduced CSF [HCO − 3 ] leads to a greater increase in [H + ] for a given rise in PaCO 2 (Siesjö, 1972). Previously, we speculated these changes in acid-base buffering following prolonged hypoxic exposure could account for the enhanced cerebrovascular CO 2 reactivity during initial exposure to high altitude (Fan et al., 2010). Accordingly, any increase in cerebrovascular CO 2 reactivity at high altitude should be abolished when plotting MCAv changes against [H + ] instead of PCO 2 . However, in the present analysis we found the increase in sigmoid slope at ALT1 persisted when we plotted MCAv against [H + ] (Figure 4B), which indicates that this increase in cerebrovascular CO 2 reactivity upon ascent to high altitude is mediated by an enhanced H + sensitivity in the cerebral vasculature per se, rather than altered acid-base balance only.
The presence of a cerebrovascular function resetting at high altitude was first alluded to by Severinghaus (1965) and later by Fan et al. (2010). In support, early work by Fencl et al. (1969) reported leftward shifts in both cerebrovascular and ventilation responses to CO 2 following a 5 day administration of sodium bicarbonate (inducing metabolic alkalosis), while chronic metabolic acidosis with ammonium chloride elicited an opposite effect on these CO 2 responses. High altitude studies have reported a leftward shift in the ventilatory response to CO 2 (Crawford and Severinghaus, 1978;Mathew et al., 1983;Somogyi et al., 2005;Fan et al., 2010), which can be predicted by modeling the acidbase changes in the CSF compartment (Duffin, 2005). Using CO 2 supplementation, Cruz et al. (1980) demonstrated that the leftward shift in the ventilatory response to CO 2 observed following 75 h of altitude exposure can be abolished by preventing the respiratory alkalosis (and subsequent negative base excess). These findings suggest that the changes in acidbase balance are responsible for the resetting of resting CBF and the leftward shifts in both cerebrovascular and ventilatory responses to CO 2 during acclimatization to high altitude. In the present study, the leftward shift in the cerebrovascular CO 2 reactivity at ALT16 was abolished when the MCAv changes were plotted against [H + ] instead of PETCO 2 (Figure 4). As a result, the midpoints of the sigmoid curves appeared at similar [H + ] (∼37 nEq/L) across all three time-points ( Figure 4B). From this, we speculate that the observed leftward shift of the sigmoid curve at ALT16 is likely a consequence of reduced  in the CSF, resulting from the prolonged exposure to hypoxia-induced hypocapnia. Moreover, our findings indicate that the cerebral reserves for vasoconstriction and vasodilation are most optimal around 37 nEq/L H + concentration, regardless of altitude or acclimatization status. Our data support the notion that H + -mediated cerebrovascular control is highly conserved.

Altered Cerebrovascular Reserves at High Altitude
The midpoint of the sigmoid cerebrovascular reactivity curve represents the optimization point between maximal vasoconstriction and vasodilation (Battisti-Charbonney et al., 2011). A departure of resting PaCO 2 from the midpoint toward the lower plateau lowers the vascular reserve for further constriction, while increasing PaCO 2 toward the upper plateau lowers the vascular reserve for dilation ( Figure 4A). In addition, outside the linear portion of the sigmoid curve, the CBF responsiveness to changes in CO 2 is blunted. Together these changes in PaCO 2 and associated reductions in vascular reserve for constriction and dilation limit the blood vessels' ability to regulate CBF against perturbations in either PaCO 2 or ABP. As reported in our previous study (Subudhi et al., 2014a), hyperventilation at 5260 m lowered resting PaCO 2 by ∼12 mmHg at ALT1 and by ∼17 mmHg at ALT16. These reductions in resting PaCO 2 lowered the vascular constriction reserve by ∼30% upon ascent to 5260 m and following acclimatization, despite the compensatory leftward shift in the cerebrovascular CO 2 reactivity curve at ALT16 (Figure 4A).
Our present findings indicate that hyperventilation-induced hypocapnia reduced the vascular reserve for vasoconstriction at high altitude, thereby blunting the vessel's ability to respond to changes in CO 2 (and presumably blood pressure) at 5260 m. Importantly, this reduced vascular reserve and associated blunting of the cerebral vessels' responsiveness could potentially account for the impaired cerebral autoregulation observed during acute and chronic exposure to hypoxia, as previously reported by Subudhi et al. (2014b) and others (Jansen et al., 2000;Iwasaki et al., 2011).

Confounding Influence of Background Hypoxia and Blood Pressure
The effect of high altitude exposure on cerebrovascular CO 2 reactivity has been examined by a number of studies (Jensen et al., 1996;Jansen et al., 1999;Fan et al., 2010Fan et al., , 2012Fan et al., , 2014Lucas et al., 2011;Villien et al., 2013;Rupp et al., 2014;Flück et al., 2015;Willie et al., 2015). More recently, studies have assessed cerebrovascular CO 2 reactivity in background hypoxia (Rupp et al., 2014;Flück et al., 2015;Willie et al., 2015), which has been shown to blunt the CBF response to CO 2 (McPherson et al., 1987;Fan et al., 2013;Ogoh et al., 2014)-presumably by exhausting the dilatory response and reducing prostanoid synthesis (Leffler et al., 1986). Those studies have found cerebrovascular CO 2 reactivity to be either reduced (Rupp et al., 2014;Flück et al., 2015), or unchanged (Willie et al., 2015) following ascent to high altitude, while ventilatory responsiveness to CO 2 was enhanced. In contrast, studies assessing cerebrovascular CO 2 reactivity in background hyperoxia have consistently found it to be elevated with high altitude ascent (Fan et al., 2010(Fan et al., , 2012(Fan et al., , 2014 and prolonged exposure to hypoxia (Poulin et al., 2002), coinciding with enhanced ventilatory CO 2 sensitivity. Due to the confounding effects of hypoxia, the cerebrovascular responses to CO 2 assessed in background hypoxia may not necessarily reflect true cerebrovascular CO 2 reactivity per se and thus should be interpreted with caution.
In agreement with our previous report using linear fitting (Fan et al., 2014), we found cerebrovascular CO 2 reactivity to be enhanced by 85% upon ascent at ALT1, while no further increased was observed following the acclimatization period in this study (Table 1). Similar to other high altitude studies (Flück et al., 2015;Willie et al., 2015), we observed large elevations in blood pressure during CO 2 breathing (Fan et al., 2014). Since increases in systemic blood pressure influence the CBF response to CO 2 (Betz, 1968) and thus confounds the measurement of cerebrovascular CO 2 reactivity (Regan et al., 2014), we attribute this smaller additional increase in cerebrovascular CO 2 reactivity at ALT16 to the constant ABP during the rebreathing trials analyzed in this study. Therefore, it should be acknowledged that the previously reported changes of cerebrovascular CO 2 reactivity at high altitude reflect the CBF response to the combined effects of both CO 2 and arterial blood pressure, rather than CO 2 alone. However, we cannot exclude the possibility that the lack of increase at ALT16 compared to ALT1 may be due a lower subject number.

Limitations
One of the critiques when using TCD-measured MCAv as an index of CBF is the assumption that middle cerebral artery (MCA) diameter does not change. Due to the exponential dependency of the vessel cross-sectional area on the diameter, small changes in MCA diameter would result in considerable errors in CBF estimates with TCD. As mentioned in our previous study (Fan et al., 2014), we assessed MCAv changes in a background of hyperoxia (PETO 2 >250 mmHg), thus excluding any confounding dilatory effects of hypoxia on MCA diameter in our measurements. While early magnetic resonance imaging (MRI) studies found no changes in MCA diameter during changes in PETCO 2 between 24 and 45 mmHg (Valdueza et al., 1997;Serrador et al., 2000), more recent and higher resolution MRI studies reported changes in MCA diameter (1.2-8%) during voluntary hyperventilation (PETCO 2 : 27-24 mmHg) and hypercapnia (PETCO 2 : 45-51 mmHg) (Coverdale et al., 2014;Verbree et al., 2014). The major implications of these MCA diameter changes are overestimation of CBF at low PaCO 2 , and underestimation of CBF at high PaCO 2 with TCD. In agreement, Willie et al. (2015) found TCD to underestimate cerebrovascular CO 2 reactivity when compared to the vascular Doppler ultrasound method. However, we contend that our assessment of CBF with TCD estimates are reliable since: (i) Caputi et al. (2014) have shown similar values of cerebrovascular CO 2 reactivity using MRI and TCD estimates of CBF response to 5% CO 2 breathing; and (ii) all the temporal indices such as the midpoint and range of the sigmoid curve would be unaffected by any underestimation of CBF with TCD. Another important consideration when interpreting our findings is our use of arterial pH and acid-base buffering as a surrogate for CSF pH changes at high altitude. Dempsey et al. (1974) previously reported consistent pH differences (∼0.08), and similar [HCO − 3 ] between arterial and lumbar spinal fluid in seven subjects at sealevel, and following 1, 8 h, and 3-4 weeks at 3100 m. Moreover, they found identical pH compensation (66%) in both arterial and lumbar spinal fluid compartments following acclimatization to high altitude. The data from Dempsey et al. (1974) thus support the use of arterial acid-base changes as an index of CSF changes during acute and chronic exposure to high altitude.

CONCLUSION
The present study is the first to demonstrate a remodeling of the sigmoid cerebrovascular CO 2 reactivity curve following acclimatization to high altitude. Our data indicates that the increase in cerebrovascular CO 2 reactivity upon ascent to high altitude is mediated by enhanced H + sensitivity of the cerebral vasculature. Meanwhile, acclimatization to 5260 m leads to a leftward shift in cerebrovascular CO 2 reactivity which appears to be mediated by an altered acid-base buffering following prolonged exposure to hypoxia. We speculate that such cerebrovascular resetting is critical for maintaining normal cerebrovascular control and preventing cerebral hypoperfusion from the prolonged severe hypocapnia associated with high altitude acclimatization.

ACKNOWLEDGMENTS
This paper is part of a series titled "AltitudeOmics" that together represent a group of studies that explored the basic mechanisms controlling human acclimatization to hypoxia and its subsequent retention. Many people and organizations invested enormous amounts of time and resources to make AltitudeOmics a success. Foremost, the study was made possible by the tireless support, generosity and tenacity of our research subjects. AltitudeOmics principal investigators were Colleen G. Julian, AL, AS, and RR. A complete list of other investigators on this multinational, collaborative effort involved in development, subject management and data collection, supporting industry partners, and people and organizations in Bolivia that made AltitudeOmics possible is available in the first paper in this series (Subudhi et al., 2014a). The authors are extremely grateful to Julia Kern, Oghenero Evero, Jon E. Elliot, Steve S. Laurie, Kara M. Beasley and Nicolas Bourdillon for their invaluable assistance in the data collection for this study. We would like to extend our thanks to Roberto Molinari for his technical assistance in the statistical analysis of the data.