Elevated Carbon Monoxide in the Exhaled Breath of Mice During a Systemic Bacterial Infection or Endotoxicosis Alan G. Barbour1, Arash Ghalyanchi Langeroudi1, Charlotte M. Hirsch2, Azadeh Shojaee Estabragh1, Simone Meinardi2, and Donald R. Blake2

1Departments of Medicine and Microbiology & Molecular Genetics and 2Department of Chemistry University of California Irvine, Irvine, CA

#1729

Background: Blood is the specimen of choice for most laboratory tests for diagnosis and disease monitoring. Sampling exhaled breath is a non-invasive alternative to phlebotomy and has potential for real-time monitoring at the bedside. To evaluate breath analysis as a biomarker for infection and sepsis under experimental conditions, we crafted a means for noninvasively collecting breath samples from individual awake animals. Methods: Containers flushed with ultra-pure air were used to collect breath samples from nose-only manifolds, and the contents were subjected to gas chromatography and mass spectrometry procedures developed for analysis of trace volatile organic compounds (VOCs) in the atmosphere. We first evaluated the system with experimental infections of BALB/c scid mice with the bacterium Borrelia hermsii. Infected mice had high density bacteremia by day 5 and in comparison to uninfected controls had hepatosplenomegaly and elevations acute phase reactants and of both inflammatory and anti-inflammatory cytokines. Results: Whereas breath samples from individual infected and uninfected animals did not significantly differ for 72 different VOCs, carbon monoxide (CO) was elevated in samples from infected mice, with a mean (95% confidence limits) effect size of 4.2 (2.8-5.6), when differences in CO2 in the breath were taken into account (CO/CO2). CO/CO2 values declined to the uninfected range after 1 day of ceftriaxone. CO was not elevated in B. hermsii culture headspaces. CO/CO2 in the breath also increased dose dependently within 4 hr of a single injection of purified Escherichia coli endotoxin in wildtype BALB/c mice, and these elevations persisted for several hours. Correlated with elevated CO in breath samples were blood levels of heme oxygenase-1 protein as measured by ELISA and HMOX1 cDNA as measured by PCR. Conclusion: Our findings with single, unanesthetized animals with masses of 20-30 grams provide a rationale for further studies of CO in the human breath as a biomarker that is informative for diagnosis, staging, and monitoring effects of resuscitation and therapy of systemic infection and sepsis.

ABSTRACT

Experiments with Borrelia hermsii infection of mice are described in the following publication: Barbour AG, Hirsch CM, Ghalyanchi Langeroudi A, Meinardi S, Lewis ER, Estabragh AS, Blake DR. Elevated carbon monoxide in the exhaled breath of mice during a systemic bacterial infection. PLoS One. 2013 Jul 31;8(7):e69802. PMC3729689. @alanbarbour Copies of this publication, with full methods and references, are also in the handouts envelope. In this poster we provide details and further results of immunocompetent mice injected with single doses of E. coli lipopolysaccharide or B. burgdorferi OspA lipoprotein.

Animal models: Male 8-10 week BALB/c mice were injected intraperitoneally with (a) E. coli O111:B4 lipopolysaccharide (LPS) at doses of 50 or 250 µg, (b) endotoxin-free Borrelia burgdorferi OspA at 50 µg, or (c) endotoxin-free water at hour 0. The mice were euthanized and bled 4 h (LPS) or 24 h (LPS, OspA, or water) after the injection, and serum samples were analyzed by Myriad RBM with RodentMAP v. 3.0 bead-based immunoassays for biomarkers. Selected assays are shown for mice below, which also gives the Lower Limit of Quantitation (LLOQ) for each assay. MIP, Macrophage Inflammatory Protein; MCP, Monocyte Chemotactic Protein; TNF, Tumor Necrosis Factor; VEGF, Vascular Endothelial Growth Factor.

There was a greater pro-inflammatory response to the LPS, e.g. IL-1, IL-6 and TNF-α and the chemokines, than to the lipoprotein OspA, but C-Reactive Protein levels were higher with the lipoprotein, and there were elevations of CD40-Ligand and some chemokines with OspA as well as with LPS. There were no changes in the hematocrits of the mice, but elevations of haptoglobin were seen after LPS and OspA injections as they were during the bacterial infection.

Email: abarbour@uci.edu @alanbarbour

The fate of heme and origin of endogenous CO. One biomarker of sepsis that has not received much attention is heme oxygenase, the enzyme that recycles iron from heme and is the principal source of endogenous CO in the body. The induction of heme oxygenase expression and heightened production in CO in response or counter-response to sepsis, oxidative stress, and inflammation has garnered increased attention for several disease states, including oxidative stress in general and malaria in animal models and humans. While understanding of the role of enodgenous CO as a modulator of inflammation remains incomplete, there is evidence that by affecting heme oxygenase-1 levels genetically or pharmacologically the outcome of a disease state can be altered.

Collection and analysis. Awake mice are gently placed at nose-only ports of an inhalation chamber for ~4 min. Connected to the inlet is a pressurized cylinder of ultra-pure air. Samples of exhaled breath are collected using stainless steel canisters under controlled dynamic flow conditions via a HEPA-filter. CO was analyzed using a gas chromatograph equipped with a flame ionization detector (FID) and a molecular sieve column. A switching valve directed the column outflow to a nickel catalyst where the CO reacted with H2 to form methane (CH4) detected by the FID set at 250°C. CO2 was measured with the gas chromatograph equipped with a thermal conductivity detector set at 230°C. For CO, accuracy and precision were 1% and 2 parts per billion by volume (ppbv); values for CO2 were 1% and 3 parts per million by volume.

CO/CO2 increases in the exhaled breath of individual BALB/c mice injected with endotoxin. Our study of an experimental bacterial infection showed an increase in CO/CO2 as the infection progressed and the number of bacteria in the blood increased (Barbour et al., PLOS One). There was an event steeper decline in breath CO/CO2 in mice treated with an antibiotic. But to what extent was the phenomenon limited to this particular pathogen? If the effect was general and reflective of the inflammatory and immune state of the host animal, we expected that endotoxin of an enteric bacterium might elicit a similar response in the mice. The study of the cytokines, chemokines, and acute phase reactants in the sera of the mice injected with lipopolysaccharide (LPS) that single injections of this substance could provide this model. The figures above show the results of the experiment. Breath samples were collected on each of 2 days as a baseline before i.p. injection of the E. coli LPS in a single dose of 0 (n=3), 50 µg (n=3), or 250 µg (n=4) at 47 hr. Breath samples were then similarly collected at 51, 57, 72, 77, 96, 120, and 168 h. The left panel is a box-whisker plot where the values plotted are the change from the 0 hr baseline measure of CO/CO2. The middle panel shows values for individual mice injected with 250 µg; one mouse died (†). The right panel is the General Linear Model with ANOVA of CO/CO2 values from control and endotoxin-injected mice at 51, 57, 72, and 96 hr, at time of collection.

Specificity of the CO response to endotoxin. We used C3H/HeJ (Tlr4Lps-d) mice, which are more resistant to endotoxin, and congenic C3H/HeOuJ mice that are wild-type The left panel shows that the HeOuJ mice had similar response in the heme oxygenase-1 expression by blood cells to injections of LPS as was observed in BALB/c mice (Barbour et al., PLoS One), as determined by reverse transcriptase (RT)-quantitative PCR for HMOX1 transcripts and beta actin. We next studied the CO/CO2 in the breath and HMOX1 expression in the blood of individual 8-10 week old male C3H/HeOuJ and C3H/HeJ after i.p. injections of LPS at 10 µg/gm (~250 µg per mouse) (6 HeOuJ mice and 5 HeJ mice) or endotoxin-free buffer; 3 HeOuJ mice) at hour 0. Breath samples were collected before the injections and at 4 h and 24 h.

The middle panel shows the results of the CO/CO2 in the breath of the 3 groups at 24 h after injection. The endotoxin-resistant HeJ mice after LPS injection had similar CO values as the wild-type mice after buffer alone, but much below the normalized CO values for the wild-type mice after LPS. CO/CO2 values were also elevated at early as 4 h after injection for HeOuJ with LPS (324 [293-355]) compared to those with buffer alone (204 [190-217]). The high correlation between HMOX1 transcripts in whole blood and CO/CO2 that we observed in BALB/c mice after LPS was observed as well with wild-type C3H mice with the same LPS dose (right panel). The results showed that the elevated CO was associated with the action of LPS.

CO response to bacterial lipoprotein injection. OspA lipoprotein is a less potent stimulus for inflammation than LPS (as the table shows), but binds to a different Toll-like receptor, TLR-2. Adult male BALB/c in groups of 5 were injected s.c. with 50 µg of either OspA or ovalbumin at hour 0 and then samples were collected at hours 4 and 24. Unlike with LPS, there was no difference between groups in CO/CO2 values at 4 h, but there was a difference at 24 h: 191 (172-211) for ovalbumin mice and 237 (211-263) for lipoprotein group (p = 0.04). The difference between groups was more marked with changes in CO/CO2 values between initial and final measurements, and these correlated with HMOX1 transcript values in the blood (figure to left).

Conclusion: Our findings with single animals with masses of 20-30 grams provide a rationale for further studies of CO in the human breath as a biomarker that is informative for diagnosis, staging, and monitoring effects of resuscitation and therapy of systemic infection and sepsis. While the present study was carried out with complex chromatographic instrumentation, the same performance for CO/CO2 ratios can be obtained real-time with dedicated bench-top instruments (right; Aerodyne Research) and, feasibly, with hand-held instruments.

Table Analyte: C-Reactive Protein

CD40-Ligand Haptoglobin Interleukin

(IL) -1α IL-6 IL-10 MIP-1β MIP-2 MCP-1 MCP-3 TNF-α VEGF-A

Compound-dose-time

Units/ml (LLOQ): µg (2.1) pg (428) µg (121) pg (132) pg (3.8) pg (220) pg (53) pg (11) pg (7.9) pg (4.4) ng (0.11) pg (158)

0 4.4 3460 128 185 5.2 <220 104 13 69 245 <0.11 6440 4.9 1730 132 132 3.8 <220 104 <11 44 210 <0.11 438

LPS-50-4 7.8 8690 136 765 2350 2040 48800 5120 4660 6900 0.33 3720LPS-250-4 5.0 10400 134 . 10400 4440 59000 19400 3350 6100 0.36 3400

LPS-250-24 5.7 21500 170 461 253 1030 999 106 1760 5480 0.16 1850OspA-50-24 20.0 19600 171 175 22 <220 1430 89 620 2310 <0.11 1060OspA-50-24 21.0 25300 163 324 30 <220 1200 101 835 2820 <0.11 1170