Conflicting findings on the effectiveness of hydrogen therapy for ameliorating vascular leakage in a 5-day post hypoxic-ischemic survival piglet model

Neonatal hypoxic-ischemic encephalopathy (HIE) is a major cause of morbidity and mortality in newborns in both high- and low-income countries. The important determinants of its pathophysiology are neural cells and vascular components. In neonatal HIE, increased vascular permeability due to damage to the blood–brain barrier is associated with seizures and poor outcomes in both translational and clinical studies. In our previous studies, hydrogen gas (H2) improved the neurological outcome of HIE and ameliorated the cell death. In this study, we used albumin immunohistochemistry to assess if H2 inhalation effectively reduced the cerebral vascular leakage. Of 33 piglets subjected to a hypoxic-ischemic insult, 26 piglets were ultimately analyzed. After the insult, the piglets were grouped into normothermia (NT), H2 ventilation (H2), therapeutic hypothermia (TH), and H2 combined with TH (H2-TH) groups. The ratio of albumin stained to unstained areas was analyzed and found to be lower in the H2 group than in the other groups, although the difference was not statistically significant. In this study, H2 therapy did not significantly improve albumin leakage despite the histological images suggesting signs of improvement. Further investigations are warranted to study the efficacy of H2 gas for vascular leakage in neonatal HIE.

. Timeline of the experiment. On day 0, an HI insult was performed after surgical preparation and stabilization. The HI insult was followed by rescue with 100% O 2 . After the rescue, the piglets were randomized into NT, H 2 , TH, and H 2 -TH groups and the treatment was given accordingly for 24 h. In the TH and H 2 -TH groups, rewarming was performed. From day 1 to day 5, the piglets were allowed to recover, nursed, and fed. Neurological function was observed every 6 h from day 1 to day 5 after the insult. On day 5, the piglets were euthanized and their brains were harvested for histological analysis.
Physiological and biochemical data. There were no significant differences among the four groups in HR, MABP, or rectal temperature (RT) at baseline ( Table 1). The NT, H 2 , and H 2 -TH groups showed a significant reduction in HR at 0 h (end of insult), whereas the TH group showed a nonsignificant reduction in HR compared with baseline. HR had returned to baseline values by 1 h after the insult in all four groups. At 6 and 12 h after the insult, HR was significantly higher in the NT group than in the H 2 group. At 6, 12, and 24 h after the insult, the H 2 -TH group had a significantly reduced HR compared with baseline.
There was a significant decline in MABP at the end of the insult compared with baseline in all groups. In the NT group, MABP at 1 and 24 h after the insult was significantly reduced compared with baseline. The piglets in the H 2 group had a significantly lower MABP at 24 h after the insult compared with baseline. MABP values were not significantly different among any of the groups at any time point.
The baseline values for RT were similar in all groups, although the RT was slightly lower in the TH group. In the NT group, the RT was higher than at baseline at 6 h after the insult. The RT of the NT group was significantly higher than that of the TH and H 2 -TH groups at 1, 6, 12, and 24 h after the insult. In the TH group, the RT was significantly lower than at baseline and at the end of insult at 1, 6, 12, and 24 h after the insult. At 24 h after the insult, the TH group had a lower temperature than the NT and H 2 groups. In the H 2 group, the RT was largely constant throughout the experiment, except at 24 h. At 0, 1, 6, 12, and 24 h, the RT of the H 2 group was higher than that of the TH and H 2 -TH groups. In the H 2 -TH group, the RT was lower at 6, 12, and 24 h than at baseline. From 1 h after the insult, the RT of the H 2 -TH group was lower than that of the NT and H 2 groups.
Biochemical parameters such as pO 2 , pCO 2 , pH, base excess, lactate, and glucose at baseline showed no significant differences among the four groups (Table 2). pH, pO 2 , and base excess were significantly reduced at 0 h (end of insult) and blood lactate was significantly higher at 0 h in all groups compared with their respective showing no statistically significant differences among the groups (Fig. 3).
Albumin immunoreactivity under a low-power magnification is presented in Fig. 4. The color-binarized images show reactive areas in red. The red solid dots outside of the brain in all slides and the red areas in the ventricles (A7, B4, B5, C6, D5, and D6) were excluded from the analysis. In the NT group, the entire section was immunoreactive in almost all of the slides and the leakage was more apparent in the central regions than in the periphery. A similar pattern of immunoreactivity was observed in the remaining groups.
Neurological scores. The neurological score (NS) was assessed every 6 h from day 1 to day 5 after the HI insult. The NSs of all four groups showed a progressively increasing trend from day 1 to day 5. The NT group had Table 2. Arterial blood gas data at baseline, 0 h (end of insult), 1 h, 6 h, 12 h, and 24 h after the insult. Values are expressed as the mean ± SD. NT normothermia, TH therapeutic hypothermia, H 2 hydrogen ventilation, TH-H 2 therapeutic hypothermia with hydrogen ventilation. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 versus baseline; # p < 0.05 versus TH; § p < 0.05 versus H 2 -TH; † p < 0.05 versus H 2 . Cerebral hemodynamics and oxygenation. There were no statistically significant differences in cerebral hemodynamics compared with baseline in any of the four groups. At the end of the insult, CBV in all four groups increased and then gradually returned to around the respective baseline values 1 h later. Thereafter, the values continued to decrease until 6 h after insult with the TH group showing the lowest CBV. At 12 h after insult, the CBV value was significantly higher in the H 2 group than in the TH group. At 24 h after insult, CBV in the TH-H 2 group was the highest among the four groups, with significantly higher values compared with the NT and TH groups. CBV was also significantly higher in the H 2 group than in the NT and TH groups at 24 h after insult (Fig. 6).

Occurrence of seizures within 24 h after insult.
Analysis of aEEG data showed that seizure occurred within 24 h after insult in all 7 piglets (100%) in the NT group, in 4 of 6 piglets (66.7%) in the TH group, in 2 of 6 piglets (33.3%) in the TH-H 2 group, and in 6 of 7 piglets (85.7%) in the H 2 group.

Discussion
This is the first study in a neonatal HIE piglet model to examine the ability of H 2 therapy to ameliorate vascular leakage via an analysis of albumin immunohistochemistry. Our findings provide conflicting evidence as to whether H 2 therapy has potential in alleviating albumin leakage.
In HIE piglet models, HI insult severity is crucial for assessing the efficacy of new therapeutic approaches. Even though few studies have directly used the total duration of LAEEG as an index of HI insult severity, it is associated with biomarkers that determine brain injury severity. In previous work, the total duration of LAEEG during the HI insult was correlated with the cortical/WM injury 31 . In addition, one piglet with an LAEEG duration less than 20 min did not have any brain damage while piglets with a long duration of LAEEG developed seizures. In another study, the HI protocol involved a reduction in the FiO 2 by between 3 and 4% and its adjustment to maintain the cerebral function monitor at ≤ 5 µV. The duration of the HI insult, including the 10 min of hypotension, determines the severity of the brain damage. A mild insult was defined as a duration of 20 min while a severe insult was defined as a duration of 30 min. The severity was consistent with the histological brain damage 36 . In our model, the longer the duration of LAEEG after the insult, the larger the increase in CBV, which indicated severe brain injury 37 . Based on the literature, we considered that a total duration of LAEEG between 20 and 60 min would indicate moderate-to-severe severity in our model. In our piglet model, H 2 did not significantly reduce the albumin extravasation in response to the moderateto-severe HI insults, as determined by the ratio of albumin stained to unstained areas, even though histological images showed signs of improvement.
Regarding vascular leakage, several mechanisms can be considered. The BBB is a unique physiological barrier formed by the endothelium of cerebral microvessels in which tight junctions regulate the transport of substances between blood and the central nervous system. The BBB is responsible for protecting the brain from neurotoxic substances and delivering essential nutrients [38][39][40] . The BBB also participates in neural signaling, innate immune responses, and cellular repair and seeks to preserve optimal brain function. Normal BBB function is maintained by neurovascular units comprising vascular cells (endothelial cells, pericytes, and smooth muscle cells), glial cells (astrocytes and microglia), and neurons 41 . When the brain is exposed to certain insults, including HI insults, their impact on the BBB leads to disruption of the above-mentioned critical functions. Because of the complex interactions involved, better understanding of the mechanisms and cells helping to maintain BBB structure and function can also be key to a new therapeutic approach in neonatal HIE research.
HI damage is associated with inflammation and oxidative stress, which have impacts on BBB permeability and angiogenesis. Angiogenesis is essential for the replacement of old vessels and the formation of new immature vessels, which eventually develop into mature and stable vasculature to maintain permeability 12 . Increased BBB permeability allows the infiltration of peripheral lymphocytes and macrophages, which augment the inflammatory response and cause further damage, leading to vasogenic edema 42,43 . Various biomarkers have been combined to detect increased BBB permeability, such as dyes and radiolabeled substances and the immunohistochemical detection of plasma proteins, including albumin 44 . In general, a BBB breach can be assessed by using albumin immunohistochemistry. Being an endogenous plasma protein, albumin avoids the nonphysiological www.nature.com/scientificreports/ conditions of dyes. Although there is no single and ideal biomarker for the assessment of BBB integrity, more than 500 publications have used albumin as a marker of BBB permeability 44 .
In translational neonatal HIE models, the BBB is damaged after an HI injury, which results in the extravasation of large endogenous plasma proteins such as albumin and smaller injected molecules 45 . In neonatal mice exposed to an HI insult, increased BBB permeability is associated with elevated albumin extravasation, which increases with time after the HI insult. In the same study, infarct area and albumin extravasation were correlated, showing that BBB damage is associated with neuropathological damage 39 . In fetal HIE models, chronic hypoxia reduces pericytes and astrocytic end-feet associated with the increased extravasation of albumin 46 . In clinical HIE, an elevated cerebrospinal fluid/blood albumin ratio (BBB permeability) is observed and a significant correlation is also seen between BBB permeability and free radical injury markers 14 .   www.nature.com/scientificreports/ Numerous studies have examined the neuroprotective effects of H 2 . Its anti-oxidative, anti-inflammatory, and anti-apoptotic properties are interconnected in a complex manner. Therefore, we focus here on some important studies that emphasize how H 2 protects the BBB as a speculation. One possible mechanism involves modulation of nuclear factor erythroid 2-related factor (NRF2), which is the key transcription factor protecting cells against reactive oxygen species, pro-inflammatory stimuli, and subsequent apoptosis. In the BBB, NRF2 promotes the expression of tight junction proteins, maintains mitochondrial functions, and enhances ATP production. Activation of NRF2 is associated with protection of the BBB integrity and improvements in cognition 47,48 . In septic mice, inhalation of 2% H 2 alleviates BBB damage, decreases pro-inflammatory cytokines, increases antiinflammatory factors, and improves survival by enhancing NRF2-dependent downstream signaling pathways 49 . In another study, BBB disruption and brain edema were improved by 2% H 2 via increased NRF2 expression 19 .
Here, we examined the patterns of albumin-reactive areas (Fig. 4). In general, at a low-power magnification, the signs of albumin leakage were more common in the central areas, likely around the choroid plexus, compared with the periphery. In perinatal HI insult, the choroid plexus shows extensive necrosis after the insult 50 . Neutrophils contribute to the brain swelling in perinatal HI brain injury and most neutrophils are located in the choroid plexus 51 . The choroid plexus produces cerebrospinal fluid, plays a role in secretory and immune function, and transports nutrients and metabolites across the barrier 52 . Even though we focused on the consequences of vascular leakage in the BBB in this study, understanding of the alterations in permeability across the choroid plexus of the blood-cerebrospinal fluid barrier should also be useful for studying pathogenesis and disease progress.
In terms of cerebral hemodynamics, cerebral blood volume increased from 12 to 24 h after insult in the H 2 and H 2-TH groups, suggesting that H 2 may exert an effect, whether alone or combined with TH, that could aid in the neuronal survival by promoting greater blood flow to the neurons, and increased consumption of oxygen might be reflected in the form of lower ScO 2 53 . Regarding the occurrence of seizure, most of the piglets in the H 2 -TH group did not develop seizures within 24 h after insult. By contrast, most of the piglets that received H 2 ventilation alone had seizures within 24 h. However, in this study, albumin leakage was assessed at day 5 after insult, meaning that the relationship between seizure occurrence within 24 h and vascular leakage at day 5 is difficult to interpret.
In our study, H 2 therapy (both alone and combined with TH) did not effectively ameliorate vascular leakage. The interpretation of this finding should consider multiple factors, such as target cells or area, severity of insult, and whether or not the mechanisms of action of the therapies overlap with each other at particular phases of HIE. First, this study assessed vascular leakage. The histological outcomes of the neurons and the supporting neural cells were not investigated. Previously, by using the same piglet model, our group investigated the histological outcomes of acute renal injury after TH treatment and found that renal fibrosis was not improved with TH 54 . Because the kidneys are highly vascularized, we speculated that the reduction in renal blood flow with both asphyxia and TH are the main reasons why TH did not ameliorate the renal fibrosis.
Next, our previous study of H 2 concluded that the 5-day neurological outcome was better with combined therapy than with TH alone 20 . In that study, the insult severity was highly variable, from mild to severe, whereas the current study focused on a moderate-to-severe insult.
Regarding combined H 2 and TH therapy, Kovács et al. concluded that neither H 2 nor CO 2 combined with TH showed superior neuroprotective effects in their piglet model. The authors mentioned the possibility of their combination resulting in neutral or even harmful effects 35 . Similar outcomes were observed when erythropoietin was combined with TH, with the High-dose Erythropoietin for Asphyxia and Encephalopathy (HEAL) study concluding that there was no significant reduction in death or neurodevelopmental impairment at 2-3 years of age and that the treatment was associated with serious adverse events in infants with moderate-to-severe HIE 55 . In preclinical studies, erythropoietin had good outcomes when used alone but mixed outcomes when combined with TH due to the activation of similar neuroprotective pathways during the acute phase [56][57][58][59] . Similarly, in the present study, H 2 might reverse the improvement in vascular leakage induced by TH, at least based on the previous findings.
There are several limitations to this study. First, we focused on a moderate-to-severe HI insult. Therefore, the effectiveness of H 2 in mild and very severe insults is unknown. In this study, the histological damage to cerebral vasculatures (e.g., the condition of endothelial cells) was not examined, even though our main focus was on albumin extravasation. The next step is to determine the optimal timing for starting therapy after HI and the duration of H 2 therapy in order to evaluate whether a single or combined approach is effective. Finally, antioxidative, anti-inflammatory, and anti-apoptotic biomarkers were not studied, meaning that the factor underlying the improvement in albumin leakage was not identified. Nonetheless, this translational study revealed the neuroprotective potential of H 2 therapy for neonatal HIE.
To conclude, our study could not prove the efficacy of H 2 ventilation alone or combined with TH for ameliorating vascular leakage. However, H 2 ventilation has potential neuroprotective effects that are worth exploring. As a physiological gas that is potentially harmless in vivo, H 2 may be a suitable candidate for clinical use in the developing and vulnerable brain. H 2 ventilation is also a feasible method compared with H 2 saline in newborns because fluid management can be challenging in HIE. Overall, the benefits of H 2 appear to outweigh the risks. Future studies should focus on a carefully designed translational model with a specific insult severity, target cells (e.g., neurons, astrocytes, and BBB endothelial cells), and an appropriate set of biomarkers to better understand and highlight the effectiveness of H 2 .

Materials and methods
Ethical approval and animal preparation. The study protocol was approved by the Animal Care and Use Committee for Kagawa University (15070-1) and in accordance with Animal Research: Reporting In Vivo Experiments guidelines. All methods were carried out in accordance with relevant guidelines and regulations. www.nature.com/scientificreports/ Thirty-three newborn piglets within 24 h after birth (21 males, 12 females; body weight ranging from 1450 to 2150 g) were anesthetized and surgically prepared. Because the experimental procedures are described in detail in previous articles, the procedures are only briefly noted in this report 20,21 . The piglets were placed under a radiant warmer and their activities and alertness were briefly observed. Anesthesia was induced with 1-2% isoflurane (Forane ® inhalant liquid; Abbott Co., Tokyo, Japan) in air using a facemask. Each piglet was then intubated and mechanically ventilated. An umbilical vein catheter was inserted for blood pressure monitoring, as well as an umbilical artery catheter for blood sampling. After cannulation, the piglets were anesthetized with fentanyl citrate at an initial dose of 10 µg/kg followed by continuous infusion at 5 µg/kg/h and were paralyzed with pancuronium bromide at an initial dose of 100 µg/kg followed by continuous infusion at 100 µg/kg/h. Maintenance solution was infused continuously at a rate of 4 mL/ kg/h via the umbilical vein. Arterial blood samples were taken at critical points and when clinically indicated throughout the experiment. Each piglet was then placed in a copper mesh-shielded cage under a radiant warmer to maintain a RT of 38.0 °C ± 0.5 °C. Inspired gas was prepared by mixing O 2 and N 2 gases to obtain the oxygen concentrations required for the experiment. Ventilation was adjusted to maintain PaO 2 and PaCO 2 within their normal ranges. MABP was measured and recorded via the umbilical arterial catheter.

Time-resolved near-infrared spectroscopy and analysis.
A portable three-wavelength TRS system (TRS-10; Hamamatsu Photonics K.K., Hamamatsu, Japan) was applied using probes attached to the head of each piglet. The light emitter and detector optodes were positioned on the parietal region of each piglet with a 30-mm interoptode distance. In the TRS system, a time-correlated single-photon-counting technique is used for detection. The concentrations of oxyhemoglobin (oxyHb) and deoxyhemoglobin (deoxyHb) are calculated from the absorption coefficients of oxyHb and deoxyHb, with the assumption that background absorption is due only to 85% (by volume) water. The total cerebral Hb concentration (totalHb), ScO 2 , and CBV were calculated as described previously 60,61 . Amplitude-integrated electroencephalography. Neural activity was measured by aEEG (Nicolet One; Cardinal Health, Inc., Dublin, OH). All electrical devices and the copper mesh shield were grounded. The signal was displayed on a semi-logarithmic scale at a low speed (6 cm/h). Measurements were conducted every second. Gold-plated electrode needles were placed at the P3 and P4 positions, which corresponded to the left and right parietal regions of the head. A maximum amplitude < 5 µV was defined as LAEEG. The aEEG data were examined for the occurrence of seizure within 24 h after insult.

Hypoxic-ischemic insult protocol.
Because the details were reported in our previous work 21 , only an outline of the HI insult protocol is provided here (Fig. 1). Hypoxia was induced by a reduction in the inspired oxygen concentration of the ventilator to 4% after at least 120 min of stabilization from the initial anesthetic induction. To obtain an LAEEG pattern (< 5 µV), the inspired oxygen concentration was further reduced, with adjustments to avoid causing cardiopulmonary arrest. From the beginning of LAEEG, the insult was continued for 30 min. FiO 2 was decreased (1% decrements) or increased (1% increments) during the insult to maintain the LAEEG, HR (> 130 beats/min), and MABP (> 70% of baseline). LAEEG was maintained for 20 min. For the final 10 min of the 30-min insult, if the MABP exceeded 70% of the baseline, hypotension was induced by decreasing the FiO 2 . Resuscitation was performed when the CBV value dropped below 30% and/or the MABP declined below 70% of baseline. Hypoxia was terminated by resuscitation with 100% oxygen. NaHCO 3 was used to correct a base deficit (base excess below − 5.0 mEq/L) to maintain a pH of 7.3-7.5. After 10 min of 100% FiO 2 , the ventilator rate and FiO 2 were gradually reduced to maintain an SpO 2 of 95-98%.
Post-insult treatment. After the HI insult, 33 piglets were randomized into four groups: HI insult with normothermia (NT, n = 9), HI insult with H 2 ventilation (H 2 , n = 8), HI insult with TH (TH, 33.5 °C ± 0.5 °C, n = 8), and HI insult with H 2 ventilation with TH (H 2 -TH, 2.1-2.7% H 2 , n = 8). Whole-body hypothermia was achieved using a cooling blanket (Medicool; MAC8 Inc., Tokyo, Japan) after resuscitation. The piglets were cooled to 33.5 °C ± 0.5 °C for 24 h and then rewarmed at 1 °C/h using a blanket. RT was used as the measure of body temperature. The temperature of the incubator was maintained at 28-32 °C. Once the piglets were weaned off the anesthesia and ventilator and extubated, they were allowed to recover and were maintained for 5 days in the incubator. Piglets were fed 50-100 mL artificial animal milk via a nasogastric tube every 6 h. The presence of seizures was recognized clinically as rhythmic pathologic movements (cycling) and tonic postures sustained between cycling episodes. If seizures occurred, the piglet was treated with phenobarbital (20 mg/kg) via intramuscular injection. If seizures persisted, the piglet was treated with two successive anticonvulsant doses. If seizures persisted after two successive anticonvulsant doses, the piglet was euthanized.
Hydrogen therapy. For H 2 ventilation, two types of cylinders were used: one contained a gas mixture comprising 3.8% H 2 and 96.2% N 2 ; the other contained 100% O 2 . The H 2 concentration depended on the oxygen requirement of each piglet. Therefore, the H 2 concentration was usually between 2.1 and 2.7 (FiO 2 range, 0.21-0.4) during the therapy. H 2 was delivered through the ventilator for 24 h. The concentration of H 2 was measured by a portable gas monitor (GX-8000; RIKEN KEIKI Co., Ltd., Japan). After 24 h of treatment, the hydrogen-nitrogen gas mixture was replaced with the air compressor.
For piglets given TH, their temperature was automatically controlled to maintain the target temperature (RT, 33-34 °C) during TH and rewarmed at a rate of 1 °C/h by a cooling blanket. Anesthesia was stopped at the beginning of the rewarming period. For NT piglets, the RT was monitored continuously to maintain a normal Histological assessment. For the euthanization of piglets on day 5 after the insult, their face was inserted into a mask and inhalation anesthesia was administered. The anesthetic agent isoflurane was introduced via a vaporizer. The vapor was inhaled until respiration ceased and death ensued. The brain of each piglet was perfused with 0.9% saline and 4% phosphate-buffered paraformaldehyde via cannulation of the left ventricle and an incision into the right atrial auricle. Brain tissue was histologically evaluated, and irregularities were graded according to a histopathology grading scale for a piglet model of posthypoxic encephalopathy, which has also been validated 31 . Coronal blocks of the GM, WM, hippocampus, and cerebellum were embedded in paraffin and cut with a microtome at 4 μm. Albumin immunohistochemistry was performed using goat polyclonal anti-pig albumin antibody (1:200, Cat A100-110P; Bethyl Laboratories, Inc., Montgomery, TX), as instructed by the manufacturer's protocol. Whole areas in sections were analyzed, and albumin-stained areas were identified using ImageJ software (National Institutes of Health, Bethesda, MD). In our model, albumin leakage was mainly seen in the subcortical structures, especially the basal ganglia and periWM, and superficial cortical GM and subWM. The ratio of albumin stained to unstained areas was analyzed and compared among the four groups.
Neurological assessment. Soon after the piglets were nursed in the incubator, neurological function was observed by examiners who were blinded to the protocols. Neurological examination was conducted every 6 h for 5 days from day 1 to day 5 post-insult. The neurological scoring comprised nine neurological items: a, respiration; b, consciousness; c, orientation; d, ability to walk; e, ability to control the forelimbs; f, ability to control the hind limbs; g, maintenance of tone; and h, pathological movements (scored as: 2, normal; 1, moderately abnormal; or 0, definitely pathological). The minimum score was 0 and the maximum, indicating a normal healthy piglet, was 18 31 . Data analysis. GraphPad Prism 9 (GraphPad Software, La Jolla, CA) was used for all statistical analyses.
To demonstrate the efficacy of H 2 under moderate-to-severe insult conditions, any of the 33 piglets with a total duration of LAEEG below 20 min and above 60 min were excluded from the analysis (2 from the NT group, 1 from the H 2 group, 2 from the TH group, and 2 from the H 2 -TH group). The final total number of piglets was 26 (NT = 7, H 2 = 7, TH = 6, and H 2 -TH = 6). Values are expressed as the mean (standard deviation [SD]) for physiological and blood gas data. For the duration of LAEEG, mean (SEM) was used. Physiological data and blood gas data were compared among the four groups at each time point (baseline and 0, 1, 6, 12, and 24 h after the insult). For the comparison of each time point with the baseline value, Dunnett's multiple comparisons test was used. TRS and ScO 2 values are expressed as the mean (SD). To compare the ratio of albumin stained to unstained areas among the groups, the Steel-Dwass test was performed and the values were expressed as the median (interquartile range). For the NS from day 1 to day 5 among all groups, one-way ANOVA followed by Tukey's multiple comparison test was used. Values are expressed as the median (interquartile range). A p value < 0.05 was considered statistically significant.

Data availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.