Topographical plots of esophageal distension and contraction: effects of posture on esophageal peristalsis and bolus transport

Abstract

Each swallow induces a wave of inhibition followed by contraction in the esophagus. Unlike contraction, which can easily be measured in humans using high-resolution manometry (HRM), inhibition is difficult to measure. Luminal distension is a surrogate of the esophageal inhibition. The aim of this study was to determine the effect of posture on the temporal and quantitative relationship between distension and contraction along the entire length of the esophagus in normal healthy subjects by using concurrent HRM, HRM impedance (HRMZ), and intraluminal ultrasound (US). Studies were conducted in 15 normal healthy subjects in the supine and Trendelenburg positions. Both manual and automated methods were used to extract quantitative pressure and impedance-derived features from the HRMZ recordings. Topographical plots of distension and contraction were visualized along the entire length of the esophagus. Distension was also measured from the US images during 10-ml swallows at 5 cm above the lower esophageal sphincter. Each swallow was associated with luminal distension followed by contraction, both of which traversed the esophagus in a sequential/peristaltic fashion. Luminal distension (US) and esophageal contraction amplitude were greater in the Trendelenburg compared with the supine position. Length of esophageal breaks (in the transition zone) were reduced in the Trendelenburg position. Change in posture altered the temporal relationship between distension and contraction, and bolus traveled closer to the esophageal contraction in the Trendelenburg position. Topographical contraction-distension plots derived from HRMZ recordings is a novel way to visualize esophageal peristalsis. Future studies should investigate if abnormalities of esophageal distension are the cause of functional dysphagia.

NEW & NOTEWORTHY Ascending contraction and descending inhibition are two important components of peristalsis. High-resolution manometry only measures the contraction phase of peristalsis. We measured esophageal distension from intraluminal impedance recordings and developed novel contraction-distension topographical plots to prove that similar to contraction, distension also travels in a peristaltic fashion. Change in posture from the supine to the Trendelenburg position also increased the amplitude of esophageal distension and contraction and altered the temporal relationship between distension and contraction.

INTRODUCTION

Each swallow induces a wave of inhibition followed by a wave of contraction, both of which traverse from the top to the bottom of the esophagus in a sequential or peristaltic fashion. During the inhibition phase, there is relaxation of the esophageal musculature that allows entry of the bolus into the lumen of the esophagus, following which esophageal contraction propels the bolus toward the stomach. The duration of the inhibition increases from the cranial to the caudal direction in the esophagus, resulting in a delay of contraction at each location, which is referred to as the latency of contraction (11). Initial inhibition followed by contraction is the basis of peristalsis, or the “law of intestine,” described by Bayless and Starling (2) 120 years ago. Each swallow induces primary peristalsis in the esophagus in a monotonous fashion, even though the characteristics of inhibition and contraction can be modulated by central and peripheral factors, e.g., a second swallow soon after the first swallow inhibits esophageal contractions, known as deglutitive inhibition. Volume and viscosity of swallowed bolus affects the velocity and amplitude of esophageal contractions.

Studies show that the degree of distension induced by the bolus during the inhibition phase of peristalsis may have major effects on the amplitude of esophageal contraction (7). In human subjects, one can record esophageal contractions at every 1 cm along the entire length of the esophagus using high-resolution manometry (HRM), which can be displayed and visualized as topographical plots. On the other hand, even in the year 2018, it is difficult to record the inhibition phase of peristalsis in the clinical esophageal manometry studies. Distension of the esophagus that occurs before the contraction during peristalsis is a surrogate marker of the esophageal inhibition; greater inhibition should result in a greater degree of esophageal distension. Our study shows that, similar to contraction, distension of the esophagus also occurs in a sequential or peristaltic fashion (1). One can visualize esophageal distension, but not the luminal cross-sectional area (CSA), using fluoroscopic imaging, as the latter is generally performed in one plane. Luminal CSA can be measured using catheter-based intraluminal ultrasound (US) imaging at one (15, 19) or two locations in the esophagus simultaneously but not along the entire length of the esophagus (1).

Intraluminal esophageal impedance has been in use to record gastroesophageal reflux and bolus movements in the esophagus for more than 20 years (12, 16). Studies show that one may also record luminal distension using intraluminal impedance measurements (8, 20). HRM impedance (HRMZ) catheters allow impedance measurements at every 2 cm along the entire length of the esophagus. Therefore, it is possible that, similar to contraction, one can record esophageal distension along its entire length using intraluminal impedance measurements. The goal of our study was to determine the temporal and quantitative relationship between distension and contraction along the entire length of the esophagus in normal healthy subjects using concurrent impedance and HRM recordings. We also compared the effects of swallowing between the supine and Trendelenburg positions on the movement of bolus, distension, and contractions of the esophagus in normal healthy subjects.

METHODS

Study Population

Healthy subjects (n = 15; mean 35.6 ± 12 yr; 6 men) with no history of gastrointestinal disease or surgery were enrolled in the study. The human investigation committee of the University of California, San Diego approved the study protocol, and each subject signed an informed consent before enrollment in the study.

Simultaneous HRMZ and High-Frequency Intraluminal Ultrasonography Image Recordings

All subjects were studied using a catheter assembly that consisted of an HRMZ catheter (4.2-mm diameter; Medtronic, equipped with 36 pressure transducers (1 cm apart) and 18 impedance electrodes (2 cm apart), taped to a 6F, 15-MHZ, high-frequency intraluminal ultrasonography catheter (Boston Scientific Instruments, Boston, MA). The US catheter was interfaced to the HP Sonos100 Ultrasound machine (Hewlett Packard Sonos Intravascular, Andover, MA). The US images were acquired on the Manoscan computer along with the HRMZ recording using Manoscan V (Model A-400), as well as on a DVD recorder (for back up). The US transducer was positioned to acquire images at 5 cm above the lower esophageal sphincter (LES). Liquid lidocaine spray (2% lidocaine topical solution, USP) and viscous lidocaine (1% lidocaine hydrochloride topical solution, USP) were administered orally and nasally for local anesthesia, followed by placement of the HRMZ-US catheter assembly through the nose. Swallows (n = 8–10) with 10 ml 0.5 N saline were performed with the US transducer located 5 cm above the LES, with the subject in the supine and then in the Trendelenburg position. For the supine position, the stretcher was in the perfect horizontal condition (parallel to the ground). The stretcher was then tilted to a −20°-Trendelenburg position (head end of the subject lower than the feet). An additional 8–10 swallows were recorded in the Trendelenburg position.

Data Analysis

Manual and automated methods were employed to analyze these data. For manual, the analysis was carried out at 5 cm above the proximal end of the LES. For the automated method, features from all (36 pressure or 18 impedance) channels were extracted. In brief, the 20-mmHg pressure isocontour lines were first extracted using the MATLAB “contour” function. The period between the onset of pharyngeal contraction and the onset of an isocontour of 20 mmHg was defined as P1, and between the onset and the end of isocontour of 20 mmHg was period 2 (P2). For each channel during P1 and P2, descriptive statistics such as mean, minimum, and maximum were extracted. Moreover, velocity of the propagation of peak pressure and nadir impedance were also extracted. The esophageal length between the distal end of the upper esophageal sphincter and the proximal length of the LES was next divided into three equal segments and denoted by proximal, middle, and distal esophageal segments. Contractile integrals, defined as the sum of the pressure signal values, of each esophageal segment that fell between the boundaries of period P2 were also extracted and denoted by proximal, middle, and distal contractile integrals (PCI, MCI, and DCI, respectively). For each subject, per posture, all swallows were exported, and mean swallows were calculated for both pressure and impedance channels in the proximal, middle, and distal segments.

Bolus Arrival, Bolus Clearance, and Peak Distension Times in the Distal Esophagus

Impedance recordings can be used to determine the bolus arrival time (BAT; fall of impedance to 50% of the baseline) and bolus clearance time (BCT; return of impedance to the same value as bolus arrival impedance) at a given site in the esophagus (16). The BAT at 5 cm above the LES is the time difference between the onset of swallow (onset of upper esophageal sphincter relaxation) and the drop in esophageal impedance to 50% of the baseline value (denoted by BAT). The BCT is the time difference between BAT and return of impedance to 50% of baseline. Nadir impedance time corresponds to the peak bolus distension (8); it is the time between the onset of swallow and the nadir impedance. The time interval between nadir impedance and contraction peak was also determined.

Pressure Time-Series Analysis

The HRM recordings were analyzed to determine the following: 1) contractile integral in the PCI, MCI, and DCI esophagus (integrated contraction amplitude values of the three regions were calculated using a custom-developed computer software program); 2) mean and peak pressures during P1 (time between swallow onset and isobaric contour of 20 mm Hg) and P2 (time period between onset and end of isobaric pressure of 20 mm Hg) (Fig. 1); and 3) the length of the breaks in esophageal peristalsis at the transition zone, based on the contraction amplitude <20 mm Hg.

Fig. 1.

High-resolution manometry and impedance topographs of a sample 10-ml swallow of 0.5 N saline during the supine (A) and Trendelenburg (B) position. Contours of iso-pressure (i.e., 20 mmHg) are also displayed. White dashed lines on the pressure topographs represent the level at which the ultrasound (US) probe was positioned. The corresponding US M-mode is also shown for both postures. Here, period 1 corresponds to the time between the onset of swallow and onset of contraction (20 mm Hg isocontour). Period 2 is the time between the isocontours of 20 mm, at the beginning and end of the contraction wave. Note that the lowering of impedance values and the increase of esophageal pressure in the Trendelenburg position. Furthermore, the length of the transition zone breaks is reduced in the Trendelenburg position.

Distension and HRM Plots

Based on the principles that impedance recordings at each location in the esophagus can be used to measure luminal CSA of the esophagus with the equation described previously (8), we developed a custom computer software to visualize the distension and contraction plots along the length of the esophagus. At each location in the esophagus, we determined the temporal relationship between esophageal distension and peak contraction and the amplitude of distension and amplitude of peak contraction, and bolus velocity was defined as the velocity of the 95% of the nadir impedance point, moving along the length of the esophagus, between adjacent channels. The 95% value was used to avoid local minima in the impedance recordings, which usually happen in the distal section of the esophagus.

US Image Analysis

For the manual calculations (10 subjects), for each swallow, the M-mode US image was generated from the B-mode US images and converted into 16 equally spaced M-mode US images (every 22.5° apart) using a custom designed software (21). The M-mode US image, orthogonal to the esophageal wall, in which both the circular and longitudinal muscle layers were clearly visualized, was selected for the data analysis. To isolate impedance waveforms at the level of US transducer, the impedance data were exported in excel sheets and converted into line drawings. The M-mode US image and impedance recordings at the location of US were temporally aligned with the HRMZ recording for each swallow. Three swallows were analyzed in each subject at 5 cm above the LES in such a fashion. Esophageal muscle thickness (sum of both circular and longitudinal muscles) was measured dynamically from baseline (before each swallow) to the peak of esophageal contraction (5 cm above the LES). Furthermore, luminal CSA of the esophagus was also measured from the US images at 5 cm above the LES, from the B-mode US images using the methodology described previously (21).

Statistical Analysis

Quantitative data are reported as mean (±SD) or median [interquartile range (IQR) 25–75] when appropriate. The normality of the distributions was checked by the Shapiro-Wilk test. Wilcoxon signed rank test and paired t-tests were used for statistical analysis. P < 0.05 was deemed significant.

RESULTS

Both the manual and automated analysis methods produced the same statistical significance results for all features used in the analysis of 10 subjects in whom the data were analyzed by both methods.

Effect of Posture on BAT, BCT, and Nadir Impedance Time

Single location manual analysis (5 cm above LES).

This analysis was carried out in 10 subjects. The bolus arrival time was not statistically different between the supine and Trendelenburg positions [0.85 (0.5) and 1.3 (0.9) s, respectively; median (IQR); P = 0.19]. Bolus clearance time was also not significantly different in the two positions (P = 0.33). The time from the onset of swallow to nadir impedance was longer in the Trendelenburg as compared with the supine position [5.25 (1.5) vs. 4.45 (1.5) s, respectively; P < 0.05].

Automated analysis.

This analysis was carried out in 15 subjects. The BAT was significantly longer in the Trendelenburg position in the proximal, middle, and distal esophagus (P < 0.01). BCT was also different in the middle (P < 0.01) and distal (P < 005) regions between the supine and Trendelenburg position, shorter in the later position (Fig. 2). The time from the onset of the swallow to nadir impedance was longer in the Trendelenburg position. Swallow-induced bolus generally consists of air and liquid, with air arriving first in the distal esophagus in the Trendelenburg position (seen as a rise in the impedance value and loss of image in the US recording).

Fig. 2.

The effect of posture (A) on bolus arrival time (B), clearance time (C), and nadir impedance time (D) using manual single level estimation [A; 5 cm above upper edge of the lower esophageal sphincter (LES)] and dividing the length in-between the lower edge of the upper esophageal sphincter and upper edge of the LES into three equal segments, denoted by proximal, mid and distal (B–D). *Statistical significance (P < 0.05).

Effect of Posture on the Amplitude of Distension Measured by Nadir Impedance and US Imaging

Based on Ohm’s law, there is an inverse relationship between nadir impedance and luminal CSA. A lower nadir impedance value implies larger luminal CSA. For the single location manual analysis (i.e., 5 cm above the LES), nadir impedance was significantly lower in the Trendelenburg compared with the supine position in the distal esophagus [159 (33) vs. 181 (72) Ω in the Trendelenburg and supine postures, respectively]. Luminal CSA measured from the B-mode US images revealed larger luminal CSA in the Trendelenburg compared with the supine position (P < 0.05; Fig. 3). Based on the automated analysis in 15 subjects, nadir impedance values were also lower in the distal esophagus in the Trendelenburg position as compared with the supine position, suggesting larger distension in the Trendelenburg position in the distal esophagus (Fig. 3). The proximal esophageal nadir impedance values showed a similar significant trend (P = 0.08).

Fig. 3.

Cross-sectional area (CSA) estimation using B-mode ultrasound (US) and impedance (0.5 N). A: sample CSA calculated during the supine (left) and Trendelenburg (right) positions. Note the increase in CSA in the Trendelenburg position. B: comparison of US-based and impedance-based CSA measurements; both modalities are significantly different. C: nadir impedance (in Ω) in the proximal, mid, and distal regions. *P < 0.05. **P = 0.08.

Distension HRM Plots: Relationship Between Esophageal Distension and Contraction

Distension HRM plots reveal the temporal relationship between distension and esophageal contraction along the length of the esophagus (Fig. 4). In the supine as well as Trendelenburg position, peak distension traveled the esophagus in a sequential or peristaltic fashion along the entire length of the esophagus, similar to the onset and peak of esophageal contractions. Bolus velocity was lower in the Trendelenburg position in the proximal esophagus compared with the supine position, ranging from 3.41 to 9.95 [median (IQR); 5.622 (1.58) cm/s] in the Trendelenburg position and 4.72 to 19.7 [6.21 (2.96) cm/s] in the supine position.

Fig. 4.

Sample swallow showing pressure (top) and distension (middle) topographs and esophageal distension (bottom) at regular intervals along the length of esophagus denoted by dashed lines, times 1–10 on the pressure and distension topographs for the supine (A) and Trendelenburg positions (B). Note the “American football-shaped” bolus in the Trendelenburg position.

In the Trendelenburg position, esophageal distension was shaped like an “American football,” wider in the center and tapered toward the two ends. The time difference between peak distension and onset of contraction was significantly shorter in the Trendelenburg position (Fig. 5C). As a result, peak distension moved closer to the onset of contraction in the Trendelenburg position.

Fig. 5.

Time interval between nadir impedance and peak contraction. Supine (A), Trendelenburg (B), and (C) mean time difference between the two in the proximal, mid, and distal esophageal segments of the 15 subjects. *P < 0.05. Secs, seconds.

Effect of Posture on the Pressure During Inhibition and Contraction Phase of Peristalsis and Length of Esophageal Breaks in the Esophageal Transition Zone

The inhibition phase, or period 1 (P1), was defined as the time between the onset of swallow and the contraction phase (onset of isobaric contour of 20 mmHg). Period 2 (P2) is the time between the onset and end of isobaric contour of 20 mmHg (Fig. 1). Mean pressure during P1 was statistically higher in the Trendelenburg compared with the supine position in proximal, middle, and distal regions [proximal supine: 5.50 (11.02) vs. Trendelenburg: 8.48 (11.26) mmHg; middle: 6.67 (4.97) vs. 11.62 (8.54) mmHg; and distal: 4.97 (11.02) vs. 8.55 (11.25) mmHg]. Contractile integrals in the PCI, MCI, and DCI regions were also significantly higher in the Trendelenburg position (Fig. 6). Fifty percent of the subjects demonstrated breaks in the transition zone in the supine position; the length of breaks ranged from 0.4 cm to 1.8 cm [0.5 (0.5)]. These breaks disappeared in the Trendelenburg position.

Fig. 6.

Pressure contractile integral in the proximal, mid, and distal esophagus of the 15 subjects. Note the increase in the contractile integral in the Trendelenburg position. The 3 values are calculated by summing pressures from all of the time/length foci within the field constrained by the box and 20 mmHg isobaric contour line in period 2. *Statistical significance (P < 0.05).

Effect of Posture on Change in Baseline and Peak Muscle Thickness 5 cm Above the LES

The baseline muscle thickness (before the onset of swallow) was not different between the supine [1.2 (0.21) mm] and Trendelenburg positions [1.24 (0.29) mm]. However, the peak muscle thickness during contraction was higher in the Trendelenburg position [2.75 (0.52) compared with supine [2.41 (0.43) mm] at 5 cm above the LES (P < 0.05), suggesting stronger esophageal contraction in the Trendelenburg position.

DISCUSSION

In summary, our data show the following findings: 1) impedance recordings reveal that, similar to contraction, distension peak also travels the esophagus in a peristaltic or sequential fashion along the entire length of the esophagus. 2) Posture has a significant influence on the temporal relationship between distension peak and the onset of esophageal contraction. The distension peak travels closer to the onset of contraction in the Trendelenburg position. 3) Amplitude of distension increases significantly in the Trendelenburg position compared with the supine position. 4) Amplitude of contraction increases and breaks in peristalsis in the “transition zone” are reduced in the Trendelenburg position.

The time period between the onset of swallow and the onset of contraction (P1), so-called the latency period, increases from the proximal to the distal location in the esophagus. Esophageal muscles are relaxed during P1, and this period is also referred to as the period of inhibition. Swallowed bolus traverses the esophagus during P1 and results in distension of the esophagus. The distended segment of the esophagus is shaped like an “American football,” especially in the Trendelenburg position, with maximal distension in the center of the bolus. Greater distension at a point location implies greater relaxation of the esophagus at the site of peak distension. Distension traveling in a sequential fashion along the length of the esophagus implies that the amplitude of inhibition at any given time during P1 is not the same in the entire esophagus. In other words, the wave of inhibition, similar to the wave of contraction, travels the esophagus in a peristaltic fashion. These findings are similar to what we found in an earlier study in which we measured the luminal CSA using two US probes located at two different locations in the esophagus simultaneously (1). The esophageal inhibition can be recorded in humans in vivo using the sophisticated technique of Sifrim et al. (17, 18). However, we suggest that the impedance recordings are a relatively simple way to demonstrate esophageal distension during peristalsis, which can be used as a surrogate of the wave of inhibition along the entire length of the esophagus.

Kronecker and Meltzer (10) were the first to record that the bolus arrives in the distal esophagus much ahead of the peristaltic contraction. They used an “ingenious method,” a litmus paper, to demonstrate the above phenomenon in humans. Fisher et al. used impedance methodology to prove the above and found that the volume of swallowed bolus had significant influence on the mean velocity of the front of the bolus movement in the esophagus. It increased from 10 to 70 cm/s, which was not related to the contraction wave (5). It is clear now that at the onset of swallow, pharyngeal contraction, also called “pharyngeal pump,” can squirt the bolus all the way down, close to the upper edge of the LES. Peristaltic contraction, which arrives in the distal esophagus several seconds after the bolus, propels the bolus into the stomach. In the sitting position, gravity enhances the effect of pharyngeal pump in pushing the bolus further ahead into the esophagus. On the other hand, in the Trendelenburg position, the antigravity effect diminishes the effects of pharyngeal pump in advancing the bolus toward the stomach. Accordingly, we observed that the bolus travels close to the contraction wave in the Trendelenburg position, resulting in a smaller time interval between the nadir impedance and the onset of esophageal contraction. Omari and colleagues (13) have observed the above phenomenon (bolus moving close the onset of contraction) in the supine position in patients with functional dysphagia. We observed the above in patients with achalasia III esophagus (14). The reason for the delayed arrival of bolus in the distal esophagus in the patients with nonobstructive dysphagia and patients with achalasia III esophagus is different from what we describe in this paper in association with the Trendelenburg position. In patients with achalasia III, we found luminal closure of the distal esophagus during P1, which causes delayed arrival of bolus in the distal esophagus. On the other hand, in the Trendelenburg position, it is the antigravity effect on the bolus that diminishes the effect of pharyngeal pump in propelling the bolus in the distal esophagus.

We observed that the amplitude of distension is greater in the Trendelenburg position in the proximal as well as the distal esophagus. In the supine position, immediately following a swallow, the bolus is spread over the entire length of the esophagus, and thus at any given location, especially in the proximal esophagus, the amplitude of distension is relatively small. On the other hand, bolus remains confined and condensed to a smaller segment of the esophagus in the Trendelenburg as compared with the supine position. As a result, there is an increase in the amplitude of distension at each esophageal location. In addition to a lower nadir impedance, which is a marker of greater esophageal distension, luminal CSA measurement by US imaging confirmed greater amplitude of distension in the distal esophagus in the Trendelenburg position. Other possible mechanisms for greater distension in the Trendelenburg position are greater inhibition/relaxation and greater resistance to bolus flow. The intraluminal pressures are greater during P1 in the Trendelenburg position.

We observed an increase in the amplitude of contraction, more so in the proximal esophagus in the Trendelenburg position. Interestingly, the increase in amplitude of contraction was associated with an increase in the amplitude of distension. Costa and colleagues recently proposed a “neuromechanical loop hypothesis,” according to which neural peristalsis is the consequence of the activation of functional loop involving distension, which activates the polarized enteric circuit. The latter produces contraction, which activates distension further distally, thus closing the neuromechanical loop (4). It is also possible that a greater distension during P1 results in longer lengths of the muscle fibers, which, in accordance with the length-tension principle, generates stronger or greater amplitude of the esophageal contraction. Another possibility is greater resistance to flow in the Trendelenburg position. Routine clinical manometry studies are performed in the supine position and often show the presence of breaks in peristalsis in the “transition zone.” The latter is considered to be the segment where skeletal muscle of the proximal esophagus transition into the smooth muscle of the distal esophagus (3, 6, 9); however, there is no scientific proof of the above association. Our study shows that breaks in peristalsis may be related to the lower levels of esophageal distension in the proximal esophagus in the supine position. With the increase in the distension in the proximal esophagus in the Trendelenburg position, the length of breaks are markedly reduced and often disappear.

In conclusion, simultaneous impedance and HRM recordings are performed routinely in the clinical setting in the year 2018. Generally, the impedance recordings are used to determine the presence or absence of bolus clearance in association with peristalsis in a binary fashion, i.e., yes or no. Our findings suggest that one can measure the degree of luminal distension from the impedance recordings, which has clinical significance because luminal distension can be used as a surrogate marker of esophageal inhibition. Studies by Omari and colleagues suggest alterations in 1) the amplitude of esophageal distension (higher nadir impedance value) and 2) the temporal relationship between distension and contraction in patients with nonobstructive dysphagia. Omari et al. used multiple parameters extracted from the impedance manometry recording to determine a dysphagia index score. We believe that distension HRM plots may be a simpler way to visually inspect the alterations in esophageal distension and bolus flow patterns to diagnose abnormalities of esophageal peristalsis in patients with nonobstructive dysphagia and other motor disorders.

Patients usually complain of dysphagia symptoms while they are eating in the upright position. On the other hand, our technique of measuring esophageal distension is more valid in the Trendelenburg position because impedance methodology to measure luminal CSA is more accurate in this position, the reason for which is the separation of air from liquid in the Trendelenburg position. Future studies in patients with nonobstructive dysphagia need to determine if patients show different distension patterns than normal subjects in the Trendelenburg position.

GRANTS

This work was supported by National Institutes of Health Grant R01-DK-109376.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

R.K.M. conceived and designed research; A.Z. and M.L.-L. performed experiments; A.Z., H.J.S., Y.-H.Y., X.Q., and R.K.M. analyzed data; A.Z., Y.-H.Y., X.Q., and R.K.M. interpreted results of experiments; A.Z. prepared figures; A.Z. and R.K.M. drafted manuscript; A.Z., H.J.S., Y.-H.Y., and R.K.M. edited and revised manuscript; A.Z., H.J.S., Y.-H.Y., X.Q., M.L.-L., and R.K.M. approved final version of manuscript.