GENERAL anesthesia usually requires mechanical ventilation to maintain adequate gas exchange. However, mechanical ventilation may impose stress on the lung parenchyma, triggering a proinflammatory response that damages the lung, a phenomenon known as ventilator-induced lung injury (VILI).¹  In patients with the acute respiratory distress syndrome, the use of protective mechanical ventilation with tidal volumes (V_T) of 4 to 8 ml/kg and positive end-expiratory pressure (PEEP) has been shown to reduce lung inflammation and mortality.² 

Most surgical patients undergoing general anesthesia differ from intensive care patients in terms of the absence of lung injury. However, because inflammatory mediators can be released into the circulation during surgery, making lungs vulnerable to VILI,³  the use of protective mechanical ventilation has been recommended during general anesthesia.⁴ 

Mechanical stress of the lung parenchyma is a major factor determining VILI.¹  In lungs, stress is proportional to strain, that is, proportional to the change in lung volume at end-inspiration in relation to the resting volume at barometric pressure and to the specific lung elastance.⁵  Some investigators claim that as long as stress is kept below a certain threshold, lung injury might be avoided.⁵  It has been shown that not only the magnitude but also the duration of stress triggers an inflammatory genomic response in cultured primary alveolar epithelial cells.⁶  This suggests that even below what is considered a safety threshold for magnitude of stress, proinflammatory and profibrotic responses may still be triggered in lungs if stress is applied for longer periods of time—as a result of a phenomenon that can be called stress versus time product (STP). The use of the term STP may be advantageous over “pressure versus time product,” because it applies to both tensile and compressive forces that result in a biological response. Furthermore, the term STP can be used beyond the setting of mechanical ventilation—for example, in reference to cell preparations subjected to stretching. To the best of our knowledge, however, the biological impact of STP on lung tissue in vivo has not yet been characterized.

In the current study, we investigated the effects of STP on gas exchange, respiratory system mechanics, biological markers of inflammation, fibrosis, apoptosis, as well as alveolar epithelial and endothelial damage in mechanically ventilated rats with intratracheal lipopolysaccharide (acute lung inflammation) or saline (control) administration. Modulation of STP was accomplished by using different inspiratory-to-expiratory (I:E) ratios, with I:E = 1:2, 1:1, and 2:1 corresponding to low, middle, and high STP levels (STP_low, STP_mid, and STP_high, respectively). In the mild acute lung inflammation model used, but not in controls, we expected that STP might modulate the alveolar epithelial and endothelial biological responses without major impact on lung function. Accordingly, our primary hypothesis, which was used for sample size calculation, was that STP_high would lead to increased gene expression of interleukin-6 (IL-6) in lung tissue compared with STP_mid and STP_low. Our secondary hypothesis was that STP_high, compared with STP_mid and STP_low, would increase the gene expression of markers of fibrosis, apoptosis, and mechanical stress, as well as lung tissue damage.

This study was approved by the Health Sciences Center Ethics Committee (CEUA-CCS, 019) at the Federal University of Rio de Janeiro, Rio de Janeiro, Brazil. Animals received humane care in compliance with the “Principles of Laboratory Animal Care” formulated by the National Society for Medical Research and the “Guide for the Care and Use of Laboratory Animals” prepared by the U.S. National Academy of Sciences.

Forty-six male Wistar rats (300 to 360 g) were used for this study. Animals were anesthetized with sevoflurane (2.5 Vol%) and subjected to intratracheal instillation of Escherichia coli lipopolysaccharide (Serotype 055:B5; Sigma Aldrich, St. Louis, MO; 0.55 to 0.66 mg/kg suspended in 0.9% saline with total volume equal to 20 μl, n = 18) or saline (20 μl, n = 18). After recovering from anesthesia, animals remained under close observation.

Twenty-four hours after induction of acute lung inflammation, animals were anesthetized by intraperitoneal injection of ketamine (100 mg/kg) and midazolam (5 mg/kg). An intravenous line (24 gauge) was placed in the tail vein and anesthesia was maintained intravenously with ketamine (50 mg kg⁻¹ h⁻¹) and midazolam (2.5 mg kg⁻¹ h⁻¹). Animals were placed and kept in supine position throughout the whole experiment. After median neck incision, a polyethylene catheter (PE 50) was introduced into the right internal carotid artery for blood sampling and mean arterial blood pressure measurement. Heart rate, mean arterial blood pressure, and rectal temperature were continuously recorded (Networked Multiparameter Veterinary Monitor LifeWindow 6000V; Digicare Animal Health, Boynton Beach, FL). Body temperature was maintained at 37.5° ± 1°C using a heating blanket. Administration of lactated ringer solution (8 ml kg⁻¹ h⁻¹) was performed via the tail vein to keep fluid homeostasis and mean arterial blood pressure higher than 70 mmHg. A 14-gauge cannula was used for tracheostomy and a fluid-filled tube was inserted in the esophagus for esophageal pressure measurements. Muscle paralysis was achieved with continuous infusion of pancuronium, 1 mg kg⁻¹ h⁻¹. Lungs were mechanically ventilated (Servo-I; MAQUET, Solna, Sweden) in pressure-controlled ventilation mode with inspiratory driving pressure to reach tidal volumes (V_T) of 6 ml/kg, fraction of inspired oxygen (Fio₂) of 0.4, and PEEP of 5 cm H₂O. Respiratory rate was set at 70 breaths/min and adjusted to arterial pH of 7.35 to 7.45. The I:E time ratio was set to 1:1. After the onset of mechanical ventilation, a recruitment maneuver (30 cm H₂O constant inflation for 15 s) was performed in all animals to achieve similar baseline conditions. Thirty-six animals were then randomly assigned to low, middle, and high STP levels (STP_low, STP_mid, and STP_high) for 2 h (n = 12 each). Modulation of STP was accomplished by changing I:E ratios to 1:2, 1:1, and 2:1, corresponding to STP_low, STP_mid, and STP_high, respectively. At the end of this experimental protocol, animals were sacrificed by exsanguination and the lungs were removed for further analysis. The remaining 10 animals (n = 5 each in the acute lung inflammation and control groups) did not receive mechanical ventilation and were used for measurement of wet-to-dry ratio and molecular biology analysis only. Blood (300 μl) was drawn into a heparinized syringe to determine arterial partial pressure of oxygen, arterial partial pressure of carbon dioxide, and arterial pH (i-STAT; Abbott Laboratories, Chicago, IL).

Airflow, tracheal, and esophageal pressures were measured. Airway pressure (P_aw) was measured with a SCIREQ differential pressure transducer (UT-PDP-300; SCIREQ, Montreal, Canada). Changes in esophageal pressure (ΔP_es), which reflect changes in chest wall pressure, were measured with a 30-cm long water-filled catheter (PE205) with side holes at the tip connected to a SCIREQ differential pressure transducer (UT-PL-400; SCIREQ). The catheter was passed into the stomach and then slowly returned into the esophagus. Its proper positioning was assessed using the occlusion test. In brief, this method consists of comparing the variation between P_es and P_aw during spontaneous inspiratory efforts made against a closed airway. The presence of comparable changes in P_es and P_aw (difference lower than 5%, phase angle close to nil) indicates that the changes in P_es accurately reflect changes in pleural pressure. Airflow and tracheal and esophageal pressures were continuously recorded using LabView-based software (National Instruments, Austin, TX).⁷  Transpulmonary pressures were calculated during inspiration and expiration as the difference between airway and esophageal pressures. All signals were filtered (100 Hz), amplified in a four-channel conditioner (SC-24; SCIREQ), and sampled at 200 Hz with a 12-bit analogue to digital converter (NI-DAQmx 8.7.1; National Instruments). Peak and mean airway pressures, as well as peak and mean transpulmonary pressures (P_aw,p, P_aw,m, P_L,p, and P_L,m, respectively), were computed. The work of breathing performed by the ventilator was calculated using the area enclosed by the inspiratory and expiratory part of the airway pressure volume loop. Also, we calculated the pressure–time product using airway pressure (PTP_aw) and transpulmonary pressure (PTP_L). The elastance (E_rs) and resistance (R_rs) of the respiratory system were calculated using the equation of motion.⁷  All respiratory variables were computed from continuous recordings (5 min) of P_aw, P_es, and airflow by routines written in MATLAB (Version 7.14; The Mathworks, Natick, MA). All functional measurements were obtained at baseline and at 1 and 2 h.

A laparotomy was performed immediately after blood sampling at the end of experiments and heparin (1,000 IU) was injected intravenously in the caval vein. The trachea was clamped at end-expiration at a continuous airway pressure of 5 cm H₂O, and the abdominal aorta and vena cava were sectioned to quickly sacrifice the animal by exsanguination.

Lungs were removed en bloc and the end-expiratory lung volume was measured as described elsewhere.⁸  In brief, a jar containing sufficient 0.9% saline with a submerged surplus weight was placed on a common laboratory scale, which was subsequently set to zero. The lungs were fixed to a laboratory stand by means of a thread with the surplus weight and completely submerged in saline 0.9%. The liquid displaced by the submerged lungs corresponds to the weight on the scale. Because the specific gravity of saline 0.9% differs no more than 2 to 3% from 1 g/cm³, the volume of the organ may be expressed directly by the weight gain registered on the scale.

The right lower lung lobe was fixed in 4% buffered formaldehyde solution, paraffin embedded, cut in slices of 4-μm thickness, and stained with hematoxylin–eosin. Photomicrographs at magnifications of ×25, ×100, and ×400 were obtained from four nonoverlapping fields of view per section using a light microscope. Diffuse alveolar damage (DAD) was quantified using a weighted scoring system by an expert in lung pathology (M.K.) blinded to the experimental protocol, as described elsewhere.⁹  In brief, values from 0 to 4 were used to represent the severity of alveolar edema, hemorrhage, inflammatory infiltration, and alveolar overdistension, with 0 standing for no effect and 4 for maximum severity. In addition, the extent of each score characteristic per field of view was determined with values from 0 to 4, with 0 standing for no appearance and 4 for complete involvement. Scores were calculated as the product of severity and extent of each feature, being situated in the range 0 to 16. Cumulated DAD score was calculated as sum of single score characteristics yielding score values from 0 to 64.

The wet-to-dry ratio was determined in the right middle lobe as described elsewhere.¹⁰  In brief, the right middle lobe was separated, weighed (wet weight), and then dried in a microwave at low power (200 W) for 5 min. The drying process was repeated until the difference between two consecutive lung weight measurements was less than 0.002 g. The last weight measured represented the dry weight.

Quantitative real-time reverse-transcription polymerase chain reaction was performed to measure the messenger RNA (mRNA) expression of IL-6, procaspase-3, receptor for advanced glycation end-products (RAGE), surfactant protein-B, type III procollagen, intercellular adhesion molecule (ICAM)-1, and vascular cell adhesion molecule (VCAM)-1. Glycerylaldehyde-3-phosphate dehydrogenase was used as a housekeeping gene. Central slices of the left lung were cut, collected in cryotubes, quick-frozen by immersion in liquid nitrogen, and stored at −80°C. Total RNA was extracted using the spin or vacuum total RNA isolation system (Promega, Fitchburg, WI). RNA concentration was measured by spectrophotometry in Nanodrop^® ND-1000 (Wilmington, DE). First-strand complementary DNA was synthesized from total RNA using M-MLV Reverse Transcriptase Kit (Invitrogen, Carlsbad, CA). The primers used are shown in the Supplemental Digital Content 1, http://links.lww.com/ALN/B83. Relative mRNA levels were measured with a SYBR green detection system using ABI 7500 real-time polymerase chain reaction (Applied Biosystems, Foster City, CA). Samples were measured in triplicate. Relative gene expression was calculated as a ratio of the average gene expression levels compared with the reference gene (glycerylaldehyde-3-phosphate dehydrogenase) and expressed as fold change relative to nonmechanically ventilated animals.

The sample size calculation for testing the primary hypothesis (the gene expression of IL-6 in lung tissue is increased with STP_high compared with STP_low and STP_mid in acute lung inflammation) was based on effects estimates obtained from pilot studies as well as on previous measurements made by us (mean value and dispersion, respectively). Accordingly, we expected that a sample size of six animals per level of STP would provide the appropriate power (1-β = 0.8) to identify significant (α = 0.05) differences in IL-6 gene expression, considering an effect size d = 2.2, two-sided test, and multiple comparisons (n = 3) (α* = 0.0167, α* Bonferroni adjusted).

All data are expressed as mean and SD. The primary and secondary hypotheses were tested with one-way and two-way ANOVAs, as appropriate (Prism for Mac, Version 5.0a; GraphPad Software, La Jolla, CA). DAD scores are shown as median and interquartile range. Statistics were calculated by Kruskall–Wallis tests. Adjustments for multiple comparisons were performed according to Bonferroni. Stepwise curve-fit regression analyses using linear, logarithmic, quadratic, and exponential functions with PTP_L as the nondependent and morphological and molecular biology data as the dependent variables were performed using IBM SPSS Statistics 20.0 (IBM Corp., Armonk, NY). The global significance level for all tests was P value less than 0.05.

All animals survived the intratracheal administration of lipopolysaccharide and saline, as well as the 24-h period that followed.

In animals subjected to mild acute lung inflammation, heart rate, mean arterial blood pressure, and the amount of fluid given did not differ significantly among groups (table 1). The same was true for gas exchange, end-expiratory lung volume, and lung wet-to-dry ratio (fig. 1).

Respiratory variables are shown in figure 2. P_aw,m was higher in STP_high compared with other STPs, whereas P_aw,p did not differ among STP levels. P_L,p was also similar among the groups, but P_L,m was higher with STP_high than STP_low. Intrinsic PEEP, peak flow values, volume-independent elastance (%E1), or lobe E_rs did not differ among STP levels.

As depicted in table 2, the work of breathing performed by the ventilator was comparable among STP levels. The PTP_aw was higher in STP_high compared with STP_low and STP_mid. The PTP_L differed significantly among STP levels, achieving the highest value in STP_high and the lowest value in STP_low.

Histological evaluation of DAD score characteristics revealed increased alveolar edema in STP_high compared with STP_mid (fig. 3). Other features as well as cumulated DAD score did not differ statistically among STP levels.

Gene expression of IL-6 was higher with STP_low and STP_high than STP_mid, whereas mRNA expression of procaspase-3 was comparable among STP levels (fig. 4). The mRNA expression of RAGE was higher in STP_high than STP_mid, whereas mRNA levels of surfactant protein-B did not differ significantly among STP levels. The analysis of gene expression of type III procollagen showed no statistically significant differences among STP settings. STP_low yielded increased mRNA expression of VCAM-1 compared with STP_high, and of ICAM-1 compared with STP_mid.

PTP_L was significantly associated with cumulated DAD score, inflammatory infiltration, and mRNA expression of VCAM-1 (table 3).

Data obtained in control animals are shown in Supplemental Digital Content 2, http://links.lww.com/ALN/B84, and Supplemental Digital Content 3, http://links.lww.com/ALN/B85. PTP_aw was higher in STP_high compared with other STPs (see table 1, Supplemental Digital Content 2, http://links.lww.com/ALN/B84). No significant differences were observed among levels of STP in all other functional (see figs. 1–3, Supplemental Digital Content 3, http://links.lww.com/ALN/B85), morphological (see table 2, Supplemental Digital Content 2, http://links.lww.com/ALN/B84), and biological variables (see fig. 4, Supplemental Digital Content 3, http://links.lww.com/ALN/B85). Additional data showing the significant regressions between PTP_L and morphological and molecular biology in controls and in controls pooled with lipopolysaccharide-treated animals are reported in figure 5, Supplemental Digital Content 3, http://links.lww.com/ALN/B85.

The main finding of this study was that in animals with mild acute lung inflammation, but not in controls, (1) gene expression of IL-6 was higher in STP_high and STP_lowversus STP_mid; (2) expression of RAGEs was increased in STP_highversus STP_low; (3) alveolar edema was decreased in STP_low compared with STP_high; and (4) expressions of VCAM-1 and ICAM-1 were higher in STP_low, compared with STP_high and STP_mid, respectively. Therefore, our data suggest that mechanical ventilation with I:E = 1:1, compared with I:E = 1:2 and 2:1, minimized VILI in the model of mild acute lung inflammation used in this investigation.

To assess the effects of STP on acute lung inflammation, we chose a model with intratracheal instillation of E. coli lipopolysaccharide with lung mechanical and histological impairment.^11,12  Animals were ventilated with a protective ventilatory strategy to rule out the possible effects of higher V_T, whereas the I:E ratio was modulated to accomplish different levels of STP. During mechanical ventilation, stress/strain of lung tissue occurs at all phases of the respiratory cycle. However, inspiration represents the phase of the respiratory cycle that exposes the lung tissue to higher stress. We chose to modulate STP by using I:E = 1:2, 1:1, and 2:1 in a pressure-controlled ventilation mode for the following reasons. First, the airflow magnitude and profile during pressure-controlled ventilation is similar across different I:E settings and affects VILI.¹³  Second, we aimed at maintaining the expiratory strain in the lungs—that is, end-expiratory lung volume—comparable across the groups, whereas modulating only the total amount of inspiratory stress.

Interleukin-6 and procaspase-3 mRNA expressions were analyzed because of their role as mediators of inflammation and/or apoptosis, respectively, in the pathogenesis of VILI.¹⁴  The lung expression of type III procollagen was evaluated because it is the first collagen to be remodeled in the development of fibrogenesis, and also an early marker of lung parenchyma remodeling.¹⁵  RAGE and surfactant protein-B mRNA were chosen because they closely reflect alveolar types I and II cell injury respectively.^16,17  In the pulmonary endothelium, the gene expression of the adhesion molecules VCAM-1 and ICAM-1 is increased during stress induced by mechanical ventilation.¹⁸ 

In both control animals and in those subjected to mild acute lung inflammation, there was an association between STP and P_aw, as reflected by higher P_aw,m at STP_high and STP_mid compared with STP_low. In mild acute lung inflammation, PTP_L was highest during STP_high and lowest during STP_low, suggesting that STP could be effectively modulated, whereas the work of breathing performed by the ventilator was roughly constant. However, those findings were not accompanied by changes in oxygenation, E_rs and end-expiratory lung volume, suggesting that higher P_aw,m did not yield lung recruitment. Furthermore, P_aw,p was comparable among groups. Taken together, those data seem to indicate that the strain achieved during inspiration did not differ significantly at the different investigated STP levels. Also, the peak airflow and the inspiratory airflow profile itself were comparable among groups, suggesting that alveolar epithelial and endothelial responses were determined mainly by STP.

Previous studies indicate that IL-6, a proinflammatory cytokine, contributes to VILI.^19,20  In the current study, in mild acute lung inflammation, the increase in IL-6 expression with STP_high and STP_low may be related to mechanotransduction in lung tissue by increased distortion of the alveolar-capillary barrier during inspiration and expiration respectively.²¹ 

STP_high increased the gene expression of the marker of alveolar type I cell injury (RAGE) compared with STP_mid. It has been demonstrated that increased mechanical stress, as achieved by proportionally high strain during inspiration, may result in VILI.²²  The current work adds to the previous knowledge that also the duration of the intratidal stress may play a role in the biological impact of mechanical ventilation. In line with our findings, Broccard et al.,²³  using an isolated perfused rabbit lung model, showed that cumulative stress, as measured by airway plateau pressure, contributed more to VILI than did the magnitude of dynamic stress per se, represented by P_L,p.

In mild acute lung inflammation, the markers of mechanical stress in the alveolar endothelium were increased with STP_low as compared with both STP_mid (ICAM-1) and STP_high (VCAM-1). There are different possible explanations for this apparent discrepancy: (1) although the straining of the pulmonary endothelium correlates with the straining of the alveolar epithelium, the magnitude is not identical.²⁴  Because the endothelial layer is surrounded by alveoli, the expansion of the epithelial layer should be more pronounced than that of the endothelium. In fact, excessive alveolar expansion can even decrease the radius of pulmonary capillaries, therefore decreasing the straining of the endothelium during inspiration; and (2) because the duration of expiration was longer in the STP_low group than in the other groups, longer time periods may have changed fluid shear stress, increasing the stimuli to endothelial cells.²⁵ 

Besides fluid shear stress, the intensity of pulmonary perfusion has been identified as an important determinant of VILI in isolated rabbit lungs.²⁶  We previously demonstrated that redistribution of perfusion from dorsal to ventral lung zones is accompanied by decreased damage in a saline lavage model of acute respiratory distress syndrome.^27,28  Also, we showed that hypervolemia worsens lung damage in a model of sepsis-induced acute respiratory distress syndrome.²¹ 

It could be suggested that modulation of I:E may yield changes in pulmonary blood flow, thus altering lung injury heterogeneity. However, if changes in pulmonary blood flow occurred, then we would expect differences in gas exchange and hemodynamics, which were not observed. Furthermore, the DAD was comparable among groups, whereas the gene expression of markers of inflammation, as well as alveolar epithelial and endothelial cell damage did differ according to I:E, suggesting that lung injury heterogeneity is not the main mechanism explaining our findings. In fact, the new experiments in controls, with intratracheal instillation of saline, showed that modulation of I:E did not affect gas exchange and DAD. Taking those facts into account, the biological impact of I:E modulation is more likely determined by STP.

The concept of stress and strain in the lungs has been recently reviewed.²⁹  There are numerous factors contributing to the development of VILI, namely increased plateau pressures, tidal volumes, patient–ventilator asynchrony, high respiratory rate, or airway flow.³⁰  In a porcine model of VILI, Protti et al.⁵  reported that VILI developed only when reaching a strain ≥1.5 to 2. Recently, the same group showed that in addition to global strain, static strain and strain components also contributed to VILI.³¹  Another key factor to understand VILI development is that even if global lung mechanics is not indicative of lung stress and strain, the regional distribution of stress and strain may easily overcome potentially injurious levels in the presence of lung inhomogeneity.³²  This is likely a reason why we did not observe major differences among groups regarding E_rs or plateau pressures, while having clear evidence of differences in mRNA expression of cytokines favoring STP_mid.

The current study was designed to develop a model of mild acute lung inflammation (first hit) with preserved lung function. Therefore, our data are important for better understanding the impact of STP on lung morphological and biological responses. Even though an I:E ratio equal to 1:2 is commonly used during protective mechanical ventilation, our data suggest that, in the presence of mild acute lung inflammation, an I:E of 1:1 may be appropriate. Conversely, in healthy lungs, when a first inflammatory hit is not present, the modulation of I:E does not seem to play a major role in terms of lung protection.

This study has several limitations that should be addressed: (1) we were not able to measure strain, that is, the ratio between tidal volume and aerated lung volume at end-expiration. However, the analysis of aerated area may be regarded as a surrogate of aerated lung volume; (2) although global lung stress was not directly measured, transpulmonary pressure represents the global stress exerted on the lungs,²⁹  and PTP_L was probably a valid surrogate of STP; (3) a model of mild acute lung inflammation induced by intratracheal endotoxin instillation was used. Theoretically, the intravenous administration of lipopolysaccharide could have also been used, but, in our experience, this leads to hemodynamic impairment, which may introduce confounding factors in the molecular biology analysis. Thus, our results may not be extrapolated to other experimental models; (4) a fixed PEEP level was applied and thus we cannot rule out that different results may be obtained at higher PEEP levels; and (5) the observational time was restricted to 2 h, and thus the expression of mediators was quantified using real-time polymerase chain reaction instead of enzyme-linked immunosorbent assay.¹¹  Furthermore, larger observation times may lead to increased mechanical stretch and therefore greater biological impact on the epithelium and endothelium.

In conclusion, in the mild acute lung inflammation model tested in this study, mechanical ventilation with I:E = 1:1 (STP_mid) minimized lung damage, whereas STP_high increased the gene expression of biological markers associated with inflammation and alveolar epithelial cell injury and STP_low increased markers of endothelial cell damage.

The authors thank Andre da Silva, B.S. (Laboratory of Pulmonary Investigation, Carlos Chagas Filho Biophysics Institute, Federal University of Rio de Janeiro, Rio de Janeiro, Brazil), for animal care, Ana Lucia da Silva, B.S. (Laboratory of Pulmonary Investigation, Carlos Chagas Filho Biophysics Institute, Federal University of Rio de Janeiro), for her help with microscopy, and Claudia Buchweitz, Ph.D., Filippe Vasconcellos, B.A., and Moira Schöttler, B.A. (Rio de Janeiro, Brazil), for their assistance in editing the article. The authors are indebted to Hannes Krause, M.S. (Pulmonary Engineering Group, Department of Anesthesiology and Intensive Care Medicine, University Hospital Dresden, Dresden University of Technology, Dresden, Germany), for his assistance with tables and figures.

Supported by the Centers of Excellence Program (PRONEX-FAPERJ; Rio de Janeiro, Brazil), Brazilian Council for Scientific and Technological Development (CNPq; Brasilia, Brazil), Rio de Janeiro State Research Supporting Foundation (FAPERJ; Rio de Janeiro, Brazil), São Paulo State Research Supporting Foundation (FAPESP; São Paulo, Brazil), National Institute of Science and Technology of Drugs and Medicine (INCT-INOFAR; Brasilia, Brazil), Coordination for the Improvement of Higher Level Personnel (CAPES; Brasilia, Brazil), German Academic Exchange Service (DAAD; Bonn, Germany), and departmental funds. MAQUET (Solna, Sweden) provided technical support.

The authors declare no competing interests.