Oxygen-18 isotope of breath CO2 linking to erythrocytes carbonic anhydrase activity: a biomarker for pre-diabetes and type 2 diabetes

Carbonic anhydrase (CA), a well-characterized metalloenzyme, is associated with oxygen-18 ( 18O)-isotopic fractionations of CO2. To investigate how CA activity links the 18O of breath CO2 to pre-diabetes (PD) and type 2 diabetes (T2D) during metabolism, we studied pre- and post-dose CA activities in erythrocytes with simultaneous monitoring of 18O/ 16O-isotope ratios of breath CO2 and thereafter elucidated potential metabolic pathways underlying CA alteration in the pathogenesis of T2D. Here we show that the post-dose CA activity in both T2D and PD was markedly enhanced, whereas the non-diabetic controls (NDC) exhibited a considerable reduction in post-dose CA activity when compared with their basal CA activities. However, T2D and PD exhibited isotopic enrichments of 18O in breath CO2, while a marked depletion of 18O in CO2 was manifested in NDC. Thus, the isotopic enrichments and depletions of 18O in breath CO2 were well correlated with the changes in CA activities for controls, PD and T2D. Our findings suggest the changes in CA activities in erythrocytes may contribute to the pathogenesis of T2D and the breath C 18O 16O regulated by the CA activity as a potential biomarker for non-invasive assessment of T2D, and thus may open a new method for treating T2D.

Carbonic anhydrase (CA), a well-characterized metalloenzyme, is associated with oxygen-18 ( 18 O)-isotopic fractionations of CO 2 . To investigate how CA activity links the 18 O of breath CO 2 to pre-diabetes (PD) and type 2 diabetes (T2D) during metabolism, we studied pre-and post-dose CA activities in erythrocytes with simultaneous monitoring of 18 O/ 16 O-isotope ratios of breath CO 2 and thereafter elucidated potential metabolic pathways underlying CA alteration in the pathogenesis of T2D. Here we show that the post-dose CA activity in both T2D and PD was markedly enhanced, whereas the non-diabetic controls (NDC) exhibited a considerable reduction in post-dose CA activity when compared with their basal CA activities. However, T2D and PD exhibited isotopic enrichments of 18 O in breath CO 2 , while a marked depletion of 18 O in CO 2 was manifested in NDC. Thus, the isotopic enrichments and depletions of 18 O in breath CO 2 were well correlated with the changes in CA activities for controls, PD and T2D. Our findings suggest the changes in CA activities in erythrocytes may contribute to the pathogenesis of T2D and the breath C 18 O 16 O regulated by the CA activity as a potential biomarker for non-invasive assessment of T2D, and thus may open a new method for treating T2D. C arbonic anhydrase (CA), a well-characterized pH-regulatory metalloenzyme found in most tissues including human erythrocytes (red blood cells), rapidly catalyzes the hydration of carbon dioxide (CO 2 ) to form bicarbonate (HCO 3 2 ) and the reversible dehydration 1,2 . It also plays an important role in the transport of CO 2 and ions (such as H 1 , Na 1 and Cl 2 ) along with pH-regulation in a variety of physiological processes ranging from respiration to intermediary metabolism at the cellular level [3][4][5] . Some early evidences suggest that the changes in CA activities in erythrocytes may be an initial step of altered metabolism in diabetes mellitus 2,6 . However, the precise role of CA activity, especially in the pathogenesis of type 2 diabetes mellitus (T2D), the most common deleterious metabolic disease at present worldwide 7 , is not currently known. Furthermore, the potential links between CA activity and T2D have not yet been fully elucidated.
Some early studies [8][9][10] demonstrated that the oxygen-16 ( 16 O) isotope in 12 C 16 O 2 and the oxygen-18 ( 18 O) isotope of body water (H 2 18 O) are rapidly exchanged during the respiration process in humans, catalyzed by carbonic anhydrase. This efficient exchange suggests the possibility of exploiting the oxygen-isotope fractionations of CO 2 in exhaled breath for non-invasive assessment of early-stage pre-diabetes prior to the onset of T2D. It also suggests a tantalizing hypothesis that monitoring stable 18 O 16 O in breath influenced by the enzymatic activity of CA in erythrocytes remains a challenge, whenever an individual is at high-risk for altered insulin action or for the acute onset of T2D.
In this study, we first investigated whether the total enzymatic activity of CA in erythrocytes is altered when individuals are in pre-diabetic and T2D states, and subsequently we assessed the precise role of CA activity in erythrocytes in response to glucose-stimulated insulin secretion that might influence the change in oxygen-isotope fractionations of CO 2 in exhaled breath. We further explored the potential metabolic pathways underlying CA alteration in the pathogenesis of T2D and the mechanisms linking breath oxygen-isotopes to pre-diabetes (PD) and T2D.

Results and Discussion
To investigate the potential role of CA in the pathogenesis of T2D, we first tested whether the total basal CA activity altered in the prediabetic state and T2D after overnight fasting. We measured the total erythrocytes CA activity, which includes primarily three isozymes of CA (i.e. CAI, CAII and CAIII), spectro-photometrically in patients with non-diabetic controls (NDC) (n 5 32), pre-diabetes (PD) (n 5 39) and T2D (n 5 48). In this investigation (Fig. 1a), individuals with T2D exhibited significantly lower basal CA activity as compared with PD and NDC, whereas no significant difference in basal CA activities was evident between subjects with PD and NDC. Several lines of evidence suggest that glycosylation of CA decreases its enzymatic and immunological activities and the glycosylation of CA is also enhanced during the peripheral circulation of erythrocytes 2,6 . Therefore, the decrease in basal CA activity for T2D individuals is possibly attributed to the increased level of glycosylation of the enzyme, caused by the exposure of erythrocytes to a higher concentration of plasma glucose compared to the PD and NDC (see Supplementary Fig. 1). Taken together, these findings indicate that the glycosylation plays a vital role in reduction of CA activity in individuals with T2D, whereas the glycosylation of CA does not play a significant role in either causing or facilitating the considerable differences in basal CA activities when an individual is in a normal or pre-diabetic state owing to their slight difference in blood glucose levels. Moreover, our results are coincidence with the previous study 11 , where the both CA I isozyme levels and total erythrocytes CA activities in fasting state have been reported to be downregulated in T2D as compared to control individuals.
We next explored the effect of CA activity in erythrocytes after administration of glucose, as the potential role of CA in response to glucose ingestion for NDC, PD and T2D remains unknown. To investigate this, we performed the 2h-oral glucose tolerance test (2h-OGTT). The post-dose CA activity (Fig. 1b), in both T2D and PD individuals, was markedly enhanced as compared with that of basal CA activity, although no significant difference of CA activities between these two groups was evident. In contrast, the NDC individuals exhibited a considerable reduction in post-dose CA activity when compared with basal CA activity. Erythrocyte receptor bound insulin has been proposed to alter the carbonic anhydrase activity 12 , suggesting that post-dose glucose stimulated plasma insulin levels may play an important role in changing the post-dose CA activities in erythrocytes in NDC, PD and T2D.
It is noteworthy that only pre-dose or post-dose erythrocyte CA activities do not allow one to distinguish the exact pathological state of the diabetes mellitus. We therefore, examined the absolute changes in CA activities (i.e. DCA) between the pre-dose (basal) and post-dose of glucose ingestion among all individuals (Fig. 2c). We found marked differences in DCA activities between subjects with NDC and PD, and subjects with PD and T2D, suggesting a potential link between changes in CA activities and altered metabolism responsible in individuals with PD and T2D. We have also established the previous hypothesis that the changes in CA activities in erythrocytes might signify the onset of altered metabolism in diabetes mellitus 2 . In view of this result, our findings suggest DCA activity in erythrocytes may contribute to the pathogenesis of T2D and might be considered as a potential biomarker for the detection of T2D.
We next investigated the oxygen-isotope fractionations of CO 2 by monitoring the 18  O% values in breath samples from NDC, PD and T2D (Fig. 2a), using an optical cavity-enhanced integrated cavity output spectrometer. In this investigation, individuals with T2D exhibited significantly higher isotopic enrichments of 18 O in CO 2 compared with PD during the 2h-OGTT, while a marked depletion of 18 O in CO 2 was manifested in individuals with NDC (Fig. 2b). These outcomes suggest oxygen-isotope fractionations of CO 2 in breath play an important role for non-invasive assessment of diabetes mellitus and hence these findings may open a new route to treat diabetes mellitus. However, the observations of isotopic depletion of 18 O in CO 2 for NDC and enrichments of 18 O in CO 2 for both individuals with T2D and PD are likely to be the effects of inhibition and promotion of catalytic activities of CA respectively, thus unveiling a potential link between CA activity in erythrocytes and the 18 O-isotopic www.nature.com/scientificreports SCIENTIFIC REPORTS | 5 : 8137 | DOI: 10.1038/srep08137 exchange in exhaled breath. As a result, we therefore posit that the monitoring of 18 O-exchange between carbon dioxide (C 16 O 2 ) and body water (H 2 18 O) influenced by the enzymatic activity of CA may distinctively track the evolution of pre-diabetes prior to the onset of T2D in a non-invasive way (Fig. 3a).
To investigate the precise transition from normal to pre-diabetes and then on to type 2 diabetes, we determined optimal diagnostic cut-off points of d DOB 18 O% values in breath and DCA activity in erythrocytes using receiver operating characteristics curve (ROC) analysis ( Fig. 3b and 3c). Individuals with d DOB 18 O% . 2.77% and d DOB 18 O% , 21.14% were considered T2D and NDC respectively, whereas subjects with 2.77 $ d DOB 18 O% $ 21.14 were suggested to be PD and these corresponded to the diagnostic sensitivity and specificity of ,95% and ,91%, respectively. Conversely, a diagnostic cut-off point of DCA 5 3.15 U/min/mL between individuals with PD and T2D, exhibited the sensitivity and specificity of 91.7% and 94.9%, respectively, whereas DCA 5 21.31 U/min/mL correctly diagnosed individuals with NDC and PD corresponding to similar levels of diagnostic sensitivity and specificity. Taken together, these findings may have a broad clinical efficacy for accurate evaluation of diabetes mellitus either invasively or non-invasively and therefore provide a unique approach to treat the world's most common metabolic disease.
Finally, we elucidated the potential metabolic pathways (Fig. 4) both in basal and postprandial (2h-OGTT) states underlying the mechanisms responsible for the alteration of CA activity and 18 Oisotopic exchange in breath CO 2 . Under basal conditions, the majority of endogenous glucose disposal takes place in insulin-independent tissues like the brain, liver and gastrointestinal tract 13,14 . In T2D state, an elevated rate of hepatic glucose production is the major abnormality for the increased basal plasma glucose concentration [14][15][16] and erythrocytes are then exposed to excessive glucose concentration which in turn makes a higher glycosylated form of CA than that in healthy individuals, resulting in reduction of total enzymatic activity of CA. However, in the basal state, more than 60% of cellular fuel for insulin-dependent muscle tissues is derived from free fatty acids for whole-body fuel oxidation 17 . The major metabolite CO 2 produced by cellular oxidation reaches erythrocytes through the bloodstream and reacts with body water to form carbonic acid (H 2 CO 3 ). The isotopes 16  However, under postprandial condition, the exogenous glucose stimulates the release of insulin which facilities the majority of ingested glucose to uptake predominantly in peripheral tissues 18 through the GLUT 4 transporter and a small fraction to adipose tissue. For T2D, the inhibition of insulin-receptor signalling due to insulin resistance and accordingly the lack of GLUT 4 recruitment plays a central role for the blunted rate of glucose transportation especially in insulin-dependent muscle tissues 15 . After glucose enters into muscle cell, it is phosphorylated to produce glucose-6-phosphate which later equilibrates between glycogenesis and glycolytic metabolic pathways. About one-third of the total glucose-6-phosphate, which enters into the glycolytic pathway, is predominantly oxidised into CO 2 which is then transported to the erythrocyte and eventually excreted as 12 C 16 O 18 O and 12 C 16 O 16 O in breath. In case of prediabetes and newly diagnosed type 2 diabetes, the pancreas produces more insulin to compensate the insulin resistance. Therefore, erythrocyte is exposed to the environment of elevated receptor bound insulin in pre-diabetes and newly diagnosed type 2 diabetes, creating an ionic imbalance within the erythrocytes owing to increased proton production, Na 1 ion uptake and possibly exchange of other ions such as K 1 and Cl 2 between plasma and erythrocytes 2,12 , leading to the enhancement of CA activity and accordingly alter the 18 O-isotopic fractionations of CO 2 in breath.
In conclusion, our findings point to a fundamental mechanism underlying the cause of altered CA activity in erythrocytes in individuals with dysglycemia. We have also taken a step towards unravelling the potential link between erythrocyte CA activity and 18 O-isotopic fractionations of breath CO 2 , thus suggesting that 18 O in CO 2 , the major metabolite of human breath, could be used as a potential molecular biomarker for the identification of accurate metabolic transition from normal to pre-diabetes and then on to type 2 diabetes in a non-invasive approach. Although many important gaps remain in our understanding of the exact metabolic pathways involved in causing the 18 O-isotopic exchange and in the pathophysiology of T2D, our findings may open new perspectives in the molecular diagnosis of diabetes mellitus with broad clinical applications. Moreover, new insights into the mechanisms linking changes in CA activities to 18 O-isotopic exchange are fostering exploration of the molecular basis of the pathogenesis of type 2 diabetes and new methods along with new pharmacological targets to prevent or treat the deleterious effects of this metabolic disease that threatens modern society.

Methods
Subjects. 119 participants (n 5 32 non-diabetic control, n 5 39 pre-diabetes, n 5 48 type 2 diabetes) were selected for the study. Individuals with hypertension, chronic respiratory disorder, interstitial lung disease, asthma or taking any type of medication that may alter the glucose metabolism are strictly excluded from the present study. Members from families with previous history of diabetes were also excluded from the study. Subjects were classified into three groups of type 2 diabetes (T2D), pre-diabetes (PD) and non-diabetic controls (NDC) according to the current criteria outlined by the American Diabetes Association 19 . Individuals with T2D were determined with glycosylated hemoglobin (HbA1c %) $ 6.5 and 2 hr glucose $ 200 mg/dL during an oral glucose tolerance test (OGTT), whereas PD was determined with 5.7 # HbA1c (%) , 6.5 and 140 mg/dL # 2 hr OGTT glucose , 200 mg/dL. Non-diabetic controls were defined with normal glucose tolerant, with 5.7 . HbA1c (%) and 2-hr OGTT glucose ,140 mg/dL. The detailed subject characteristics are presented in Table 1 Study protocol. After overnight fasting (,12 hours), subjects were placed on chairs, and a baseline breath sample from each subject was taken in a breath sample collection bag (QUINTRON, USA, SL No.QT00892). A baseline blood sample (10 mL) was also collected simultaneously. A test meal containing 75-gm normal glucose dissolved in 150-mL water was orally administered and consumed within 1 minute. From this starting point, breath samples were collected in every 30-min interval, whereas blood samples were drawn after 2-hr of glucose load. During exhalation, subjects were instructed in such a way that they breathed normally, held for a few seconds and exhaled completely into a breath sample collection bag so that the oral breath first passed to dead space and the endogenously produced deep-breath (end expiratory breath) entered into the 750 mL reservoir bag through an one-way valve. The blood samples were utilized for measurement of glycated hemoglobin i.e. HbA1c (%), carbonic anhydrase activities and insulin levels. Breath samples were immediately analyzed to measure d 18 O of breath CO 2 .   16 O 1 CO 2 ) compounds, catalyzed by CA activity. Erythrocytes of T2D are exposed to higher concentrations of glucose (G), which converts CA to GCA (glycosylated-CA), resulting in a substantial decrease of CA activity. Following the glucose load, the exogenous glucose stimulates insulin (I) production from the pancreatic islet. For PD and T2D, the insulin secretion increases in proportion to insulin resistance; elevated glycogenolysis, gluconeogensis and lipogenesis and diminished rate of insulin-dependent muscle tissue glucose uptake. When insulin action is suppressed for PD and T2D, there is a substantial increase in erythrocyte receptor-bound insulin, which enhances the CA activity. This alteration of CA activities, associated with insulin resistance during transition from NDC to T2D through PD, is reflected by the isotopic fractionation d 18  Breath sample analysis. Breath samples were drawn from the breath sample bags by an air-tight syringe (QUINTRON) through a T-connector fitted onto the reservoir bag. All breath samples were analyzed by a high-resolution isotopic CO 2 integrated cavity output spectrometer (ICOS).
Integrated Cavity Output Spectrometer. The cavity-enhanced absorption spectroscopy is an important tool for the measurement of trace gases along with isotopic species with ultra low concentrations. We utilized a high-resolution carbon dioxide isotope analyzer based on off-axis integrated cavity output spectroscopy (ICOS) exploiting a cavity-enhanced laser absorption technique to measure the isotopic compositions of CO 2 of breath samples. There are few naturally occurring isotopes of carbon dioxide among which 12 20,21 . The ICOS instrument is so designed that laser light can be directed into the off-axis of the optical cavity. The present ICOS spectrometer (CCIA 36-EP, Los Gatos research, USA) consists of a continuous wave distributed feedback diode laser operating at ,2.05 mm and a high-finesse optical cavity (,59 cm long) with two high-reflectivity mirrors (R , 99.98%) at the ends of the measurement cell. This arrangement allows the laser light to move back and forth inside the cavity to reach an effective optical path-length of ,3 km and thus enabling high sensitivities. The laser frequency was repeatedly tuned to scan over 20 GHz across the P(36), R(28) and P(16) ro-vibrational lines to record the absorption spectra of 12 21 respectively, in the (2,0u,1) r (0,0u,0) vibrational combination band of the CO 2 molecule 22 . The transmitted laser intensities were recorded by exploiting a photodetector after passing through a breath sample of interest. Absorption was determined from the measurement of voltage from photodetector. Beer-Lambert law was utilized to calculate the concentration after integrating the absorption spectrum. The data were acquired at a rate of 1 Hz. The temperature of the cavity was maintained at 46uC by a resistive heater and feedback control system. The pressure of the cavity also regulated at 30 Torr by a diaphragm pump and solenoid valve in order to analyze the sample. Blood sample preparation. 10 mL of venous blood samples were collected from each participant in EDTA vacutainer tubes. The collected blood samples were centrifuged at 2000 r.p.m for 15 minutes. Plasma was isolated and buffy coat was removed. The red blood cells (RBC) were washed with 0.9% NaCl solution and allowed to spin against 4000 r.p.m for 20 minutes. The erythrocytes were lysed with ice-cold distilled water. The hemolysate was centrifuged at 10,000 r.p.m for 30 minutes to remove the ghost cells and the supernatant liquid was collected. The fresh supernatant liquid was used for analysis of carbonic anhydrase activity.
Total erythrocyte carbonic anhydrase measurement. The esterase activity of carbonic anhydrase was determined spectro-photometrically as mentioned by Armstrong et al. 23 with the modification described by Parui et al. 12 . In this assay, the esterase activity of carbonic anhydrase was determined spectro-photometrically from the hydrolysis rate of p-nitrophenyl acetate to p-nitrophenol. As the other esterases present in the hemolysate also hydrolyze the p-nitrophenyl acetate, a specific inhibitor of CA, acetazolamide was used and the acetazolamide-sensitive activity was taken as the net CA activity. This was performed from the differences between the absorbances before and after adding inhibitor. The protocol consisted of 100 mL hemolysate placed in 1 cm spectrometric cell containing 1.9 mL Tris buffer and 1 ml p-nitrophenyl acetate. The change of absorbance was measured over the period of 3 min at 348 nm before and after adding the hemolysate. The same measurements were performed in the presence of 0.1 mM acetazolamide inhibitor. The absorbance was measured by a UV-Vis spectrophotometer (Shimadzu UV-2600 Spectrophotometer). One unit of enzyme activity was expressed as mmol of pnitrophenol relased/min/mL from hemolysate at room temperature (25uC) 2,12 .The following formula was used to calculate total erythrocyte CA activity: where A 3 is absorbance after 3 min, A 0 is the absorbance at 0 min, 5 3 10 3 M 21 cm 21 is the molar absorptivity of p-nitrophenol. The activity was normalized to 4.5 3 10 9 cells/mL.
Statistical method. Statistical analyses were performed using Origin Pro 8.0 (Origin Lab Corporation, USA) and Analyse-it Method Evaluation software (Analyse-it Software Ltd, UK, version 2.30). To compare normally distributed data, one way analysis of variance (ANOVA) was used whereas non-normal distributed data were analyzed by the Kruskal-Wallis test and Mann-Whitney test. A two sided p-value less than 0.05 was considered as statistically significant. All results were expressed as mean 6 SEM except Fig. 3a, where values are mean 6 SD. Receiver operating characteristic curves (ROC) were generated by plotting sensitivity (true positive rate) vs 1specificity (false positive rate) to obtain the different cut-off points of d DOB 18 O% values and DCA activities (Supplementary Table 4-Supplementary Table 7). The optimal cut-off value was considered to be that point which maximized the sensitivity and specificity (Supplementary Table 8). Data from insulin measurement were analyzed by MasterPlex software (Hitachi Software Engineering America, South San Francisco, CA, USA). Table 1 | Clinical parameters of the study subjects. Data are expressed as mean 6 SD. *Represents statistically significant differences (p , 0.05) among non-diabetic control (NDC), pre-diabetes (PD) and type 2 diabetes (T2D