Safety of inhaled CO, defined by the incidence of pre-specified administration-related AEs (as defined below) and spontaneously reported AEs through study day 7.

1. Acute myocardial infarction within 48 hours of study drug administration
2. Acute cerebrovascular accident (CVA) within 48 hours of study drug administration
3. New onset atrial or ventricular arrhythmia requiring DC cardioversion within 48 hours of study drug administration
4. Increased oxygenation requirements defined as: an increase in FiO2 of ≥ 0.2 AND increase in PEEP ≥ 5 cm H2O within 6 hours of study drug administration
5. Increase in COHb ≥ 10%
6. Increase in lactate by ≥ 2 mmol/L within 6 hours of study drug administration

All subjects were analyzed for safety measures.

The Lung Injury Score (LIS) is a composite 4-point scoring system including the PaO2/FiO2, PEEP, quasi-static respiratory compliance, and the extent of infiltrates on the chest X-ray. Each of the four components is categorized from 0 to 4, where a higher number is worse. The total Lung Injury Score is obtained by dividing the aggregate sum by the number of components used. Previous randomized clinical trials in ARDS have shown that a decreased LIS correlates with improvement in lung physiology as well as important clinical outcomes including mortality and ventilator-free days (VFDs). The number presented is the average difference from beginning to end of treatment. We present the average of data from the 2 available subjects in each group and report as "Mean" with standard deviation.

PaO2/FiO2 was to be measured on days 1-5 in ventilated subjects. We are providing percent change in PaO2/FiO2 ratio from baseline to day 5. We present the average of data from the 2 available subjects in each group and report as "Mean" with standard deviation.

We are providing percent change in PaO2/FiO2 ratio from baseline or day 1 to day 5.

The oxygenation index will be measured on days 1-5 in ventilated subjects. Oxygenation index is calculated as (FiO2 X mean airway pressure)/PaO2. We provide change in Oi from baseline. Oi is only measured when subjects are ventilated, therefore not all timepoints are available. Limited number of measurements prevents variance and measures of dispersion analyses; therefore we present the data from the subjects available in each group and report as "Mean" without standard deviation, where measures of dispersion cannot be calculated.

We provide change in Oi from baseline. Oi is only measured when subjects are ventilated, therefore not all timepoints are available. Limited number of subjects precludes further analyses.

The dead space fraction will be measured days 1-3 in ventilated subjects. We present change in dead space fraction between initial and final measurements that were available (measurements only taken while the subjects were intubated). We present the data from the subjects available in each group and report as "Mean" with standard deviation.

We present change in dead space fraction between initial and final measurements that were available (measurements only taken while the subjects were intubated).

Organ failure will be assessed using the SOFA score. SOFA scores will be assessed daily on days 1-5, as the SOFA score has been shown to be a reliable prognostic indicator of outcomes in critically ill patients. To calculate the Sequential Organ Failure Assessment (SOFA) score, each of the six components (Respiratory, Coagulation, Liver, Cardiovascular, Central Nervous System, Renal) is categorized from 0-4, where a higher number is worse. The SOFA score (0-24) will be calculated by summing all six components. We present changes in SOFA score over the time of hospitalization, over the time of ICU admission (up to days 3 and 5 for the enrolled subjects). We present the data from the subjects available in each group and report as "Mean" with standard deviation.

We present changes in SOFA score over the time of hospitalization, over the time of ICU admission (up to days 3 and 5 for the enrolled subjects).

Ventilator-free days to day 28 are defined as the number of days from the time of initiating unassisted breathing to day 28 after randomization, assuming survival for at least two consecutive calendar days after initiating unassisted breathing and continued unassisted breathing to day 28. If a subject returns to assisted breathing and subsequently achieves unassisted breathing to day 28, VFDs will be counted from the end of the last period of assisted breathing to day 28. Participants who do not survive to day 28 are assigned zero ventilator-free days. We present data of ventilator free days for enrolled subjects. Note that one subject in the medical air group died at day 9. We present the data from the subjects available in each group and report as "Mean" with standard deviation.

We present data of ventilator free days for enrolled subjects. Note that one subject in the medical air group died at day 9.

ICU-free days will be assessed on day 28. ICU-free days is defined as the number of days between randomization and day 28 in which the patient is in the ICU (for any part of a day). We present average number of ICU free days. Please note that one subject in the medical air group died at day 9. We present the data from the subjects available in each group and report as "Mean" with standard deviation.

We present average number of ICU free days. Please note that one subject in the medical air group died at day 9.

Hospital-free days will be assessed on day 60. Hospital-free days are days alive post hospital discharge through day 60. Patients who die on or prior to day 60 are assigned zero hospital-free days. We present hospital free days at day 60. Please note that one subject in the air group died at day 9. We present the data from the subjects available in each group and report as "Mean" with standard deviation.

We present hospital free days at day 60. Please note that one subject in the air group died at day 9.

Mortality will be assessed on day 28 and day 60.

Montreal Cognitive Assessment - Blind Version (MoCA-Blind) The MoCA-Blind is a remote adaptation of the Montreal Cognitive Assessment (MoCA) used as a screening assessment for detecting cognitive impairment. It is administered via telephone interview.

The MoCA-Blind assesses the following cognitive domains (points):

• Attention (0-6)
• Language: (0-3 (Repetition (0-2) and fluency (0-1))
• Abstraction (0-2)
• Memory: Delayed recall (0-5)
• Orientation (0-6)

The minimum score is 0 and maximum score is 22 points. Calculated as the sum of all domains. Interpretation: Higher scores indicate better cognitive functioning. Lower scores indicate worse cognitive functioning (greater cognitive impairment). A score of 18 or above is within the normal range. A limited number of measurements prevents variance and dispersion analyses; therefore, we report available group data as "means" without standard deviation where dispersion cannot be calculated.

Montreal Cognitive Assessment - Blind Version (MoCA-Blind) The MoCA-Blind is a remote adaptation of the Montreal Cognitive Assessment (MoCA) used as a screening assessment for detecting cognitive impairment. It is administered via telephone interview.

The MoCA-Blind assesses the following cognitive domains (points):

• Attention (0-6)
• Language: (0-3 (Repetition (0-2) and fluency (0-1))
• Abstraction (0-2)
• Memory: Delayed recall (0-5)
• Orientation (0-6)

The minimum score is 0 and maximum score is 22 points. Calculated as the sum of all domains. Interpretation: Higher scores indicate better cognitive functioning. Lower scores indicate worse cognitive functioning (greater cognitive impairment). A score of 18 or above is within the normal range. A limited number of measurements prevents variance and dispersion analyses; therefore, we report available group data as "means" without standard deviation, since dispersion cannot be calculated.

Please note that one of the CO subjects was alive and well at 6 months but declined participation in the MOCA test.

Mitochondrial DNA (mtDNA) plasma levels will be measured on days 1-3 by quantitative PCR of human NADH dehydrogenase 1. Limited number of measurements prevents variance analyses; therefore we present the data from the subjects available in each group and report as "Mean" without standard deviation, since measures of dispersion cannot be calculated.

Percentage reduction pre-day 1 Vs. post-day 3 exposure for those subjects with 3 exposures.

Plasma IL-18 levels will be measured on days 1-3 and day 5 by ELISA. We present the data from the subjects available in each group and report as "Mean" with standard deviation.

We present percent change over time from baseline to final measurement for a representative marker of inflammasome activation.

Plasma RIPK3 levels will be measured on days 1-3 and day 5 by ELISA. We present the data from the subjects available in each group and report as "Mean" with standard deviation.

Due to missing values we calculated percent change over time.

Lipid mediators (LM) and specialized pro-resolving mediators (SPMs) will be measured in plasma on days 1-3 and day 5 using liquid chromatography-tandem mass spectrometry (LC-MS-MS) based methods. We present the data from the subjects available in each group and report as "Mean" with standard deviation.

We present percent change over time from baseline to final measurement for a representative marker of pro-resolving lipid mediators.

The Hayling Sentence Completion Test assesses executive functioning (response initiation and inhibition), administered via telephone. With 30 sentence-completion items split into 2 sections (15 each).

Sections:

• 1: Response initiation (time to provide a contextually appropriate word).
• 2: Response inhibition (time and error score for providing an unrelated word).

Scores (response time and errors) from both sections are combined and converted to an age-adjusted standardized total score.

Total combined standardized score ranges from 1-10:

1. Impaired
2. Abnormal
3. Poor
4. Low Average
5. Moderate Average
6. Average
7. High Average
8. Good
9. Superior
10. Very Superior

Higher scores= Better executive functioning Lower scores= Greater impairment. A limited number of measurements prevents variance and dispersion analyses; therefore, we report available group data as "means" without standard deviation where dispersion cannot be calculated.

The Hayling Sentence Completion Test assesses executive functioning (response initiation and inhibition), administered via telephone. With 30 sentence-completion items split into 2 sections (15 each).

Sections:

• 1: Response initiation (time to provide a contextually appropriate word).
• 2: Response inhibition (time and error score for providing an unrelated word).

Scores (response time and errors) from both sections are combined and converted to an age-adjusted standardized total score.

Total combined standardized score ranges from 1-10:

1. Impaired
2. Abnormal
3. Poor
4. Low Average
5. Moderate Average
6. Average
7. High Average
8. Good
9. Superior
10. Very Superior

Higher scores= Better executive functioning Lower scores= Greater impairment. A limited number of measurements prevents variance and dispersion analyses; therefore, we report available group data as "means" without standard deviation, since dispersion cannot be calculated.

Please note that one of the CO subjects was alive and well at 6 months but declined participation in the Hayling test.

This was assessed by comparing the measured 90-minute COHb level and the target COHb level of 6-8% post exposure. We present average data from the two available subjects in the CO group and report as "Mean" with standard deviation.

Percentage change in measured COHb from targeted range.

This was assessed by comparing the measured 90-minute COHb level and the target COHb level of 6-8% post exposure. The limited number of measurements prevent the variance and measures of dispersion calculations; therefore we present the data from one available subject in the CO group and report as "Mean" without standard deviation, since measures of dispersion cannot be calculated.

The percent change in measured COHb from targeted range.

This was assessed by comparing the measured 90-minute COHb level and the target COHb level of 6-8% post exposure. The limited number of measurements prevent the variance and measures of dispersion calculations; therefore we present the data from one available subject in the CO group and report as "Mean" without standard deviation, since measures of dispersion cannot be calculated.

The percent change in measured COHb from targeted range.