INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Insulin-dependent diabetes mellitus (IDDM) can lead to an overall depression in ventilatory control mechanisms. In humans with IDDM, ventilatory depression is characterized by impaired respiratory muscle function (1) and by decreased responses to hypoxia and hypercapnia (2, 3). Rats with streptozotocin (STZ)-induced IDDM also exhibit decreased baseline ventilation, as well as attenuated hypercapnic and hypoxic ventilatory responses (4). Apart from these few findings, the impact of IDDM on respiratory control is poorly understood. In particular, it is unclear whether metabolic factors other than insulin deficiency play a role in respiratory abnormalities associated with diabetes.

The purpose of the present study was to examine the link between metabolic factors associated with IDDM, including insulin deficiency and hyperglycemia, and alterations in respiratory control mechanisms. Specifically, we were interested in a putative role for leptin, a metabolic hormone that controls satiety and food intake (5), but which is also modulated by insulin levels (6, 7). We have previously shown that leptin is a powerful respiratory stimulant, and that the absence of leptin is associated with severe respiratory depression (8, 9). On this basis we hypothesized that leptin may provide a common link between IDDM and alterations in respiratory control. Specifically, we proposed that ventilatory depression in IDDM is dependent on a reduction in circulating leptin levels.

To examine this hypothesis, we studied the influence of diabetes and leptin on respiratory control in two mouse models of IDDM. In the first part (Part A) of our study, IDDM was induced by administration of STZ to C57BL/6J mice, a procedure that destroyed the pancreatic islet cells over a period of hours to days (10). Using this model, we characterized the changes in respiratory control that occurred in response to IDDM in C57BL/6J mice fed a regular diet (low circulating leptin levels) and in C57BL/6J mice fed a high-fat diet to induce obesity (high circulating leptin levels) (5, 11). The role of leptin in ventilatory responses to IDDM was further studied by administering STZ to weight-matched, obese, leptin-deficient C57BL/6J-Lep^ob mice. In these studies of STZ-induced IDDM, respiratory control parameters were measured during wakefulness and sleep to control for sleep/wake state and examine whether state changes influence ventilatory responses to IDDM. In the second part (Part B) of the study, we examined the natural history of changes in respiratory control that occur over weeks to months in spontaneous IDDM in nonobese diabetic (NOD) Ltj mice (12). The combination of these two mouse models of IDDM has enabled us to gain insight into the mechanisms and time course of ventilatory dysfunction in diabetes.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animals

Forty-eight wild-type male C57BL/6J mice, 11 male C57BL/6J-Lep^ob mice, and 10 female NOD-Ltj mice from Jackson Laboratory (Bar Harbor, ME) were used in the study. The study was approved by the Johns Hopkins University Animal Use and Care Committee and complied with the American Physiological Society Guidelines for Animal Studies. For all surgical procedures, anesthesia was induced and maintained with halothane administered through a face mask.

Surgical Procedures

C57BL/6J and C57BL/6J-Lep^ob mice had chronically implanted polysomnographic electrodes for determination of their sleep/wake state as previously described (9). A midline incision was made to expose the skull, and two bilateral pairs of holes were drilled through the skull in the frontal and parietal regions. Four electroencephalographic electrodes were fashioned from Teflon-coated stainless steel wire. The end of each electrode was stripped of 0.15 cm of Teflon, bent at 90 degrees, and inserted into the skull through one of the four predrilled holes. The four electrodes were bonded to the dorsal surface of the skull with dental acrylic (Land Dental, Wheeling, IL). Two nuchal electromyographic electrodes were stitched (size 6-0 silk) in a flat postition to the surface of the muscle immediately posterior to the dorsal area of the mouse skull. The skin overlying the skull and posterior muscles was reapposed, and the six electrodes were led out of the skin dorsally between the shoulderblades.

Retroorbital sinus puncture, for glucose measurement in peripheral blood (50 µl), was performed three times under halothane anesthesia in C57BL/6J and C57BL/6J-Lep^ob mice (before administration of STZ, and on Days 2 and 7 after STZ) and weekly in NOD-Ltj mice. Arterial blood (1 to 1.2 ml) was obtained by direct cardiac puncture under halothane anesthesia. After blood withdrawal, the animals were euthanized with pentobarbital (60 mg, intraperitoneally).

Ventilatory Measurements

Plethysmography. During data-collection periods, the animals were placed in a whole-body barometric plethysmography custom-made chamber to measure ventilation as previously described (9). The chamber (823 ml³) was customized to allow the animal free movement while electroencephalogram (EEG) and electromyogram (EMG) electrodes were exited through a sealed port. The plethysmograph was referenced to a second chamber and flushed with compressed, humidified air (80% relative humidity) at a flow rate of approximately 600 ml/min. To measure ventilation, the ports through which the gases entered and exited the chamber were closed to produce a constant chamber volume. Once the chamber was at constant volume, tidal volume (VT) and respiratory frequency (f ) were measured from changes in pressure (PM15E differential pressure transducer; Statham Gould, Hato Ray, Puerto Rico) caused by inspiratory and expiratory temperature fluctuations. The body temperature was assumed to be constant at 37° C. Calibration injections of 10, 20, and 30 µl of room air were made with the animal inside the constant-volume chamber. Minute ventilation (E) is reported as the product of f and VT.

Respiratory control protocol. In C57BL/6J and C57BL/6J-Lep^ob mice, sleep/wake state was assessed from EEG and EMG recordings as previously described (9). Wakefulness was characterized by low-amplitude, high-frequency (approximately 10 to 20 Hz) EEG waves and high levels of EMG activity as compared with the sleep state. NREM sleep was characterized by high-amplitude, low-frequency (approximately 2 to 5 Hz) EEG waves and by considerably less EMG activity than during wakefulness.

Ventilation was measured during wakefulness and non-rapid-eye-movement (NREM) sleep in response to a range of hypercapnic gases (0, 3, 5, and 8% CO₂ in 40% O₂, to ensure the absence of any hypoxic stimulus). The animals were allowed to freely cycle through sleep/ wake states. In each sleep/wake state, a single gas challenge was introduced to the mouse for 3 to 4 min before the ports through which the gases entered and exited the chamber were closed to produce a constant chamber volume and to record ventilation. Over the course of the experiment, we challenged the mice twice during wakefulness and NREM sleep, over a range of from 0 to 8% CO₂.

VT and f were assessed over 5- to 8-s periods during wakefulness and NREM sleep. The hypercapnic ventilatory response (HCVR) was determined in each animal from the slope of the relationship between E and inspired CO₂ (0 to 8%) during wakefulness and NREM sleep, using linear least-squares regression analysis.

NOD-Ltj mice did not have implanted EEG or EMG electrodes, since serial measurements were required over several weeks. Before the ventilatory measurements, the mice were permitted to acclimate within the plethysmograph for at least 30 min. Ventilation was measured during quiet wakefulness in response to a range of hypercapnic gases (0, 3, 5, and 8% CO₂ in 40% O₂, to ensure the absence of any hypoxic stimulus). Each mouse was challenged twice with each gas during the experiment, as described earlier.

Blood Glucose, Urine Acetoacetate, Plasma Leptin, Insulin, and -Hydroxybutyrate Determination

Blood glucose was measured in blood from the retroorbital plexus, using the Accu-Chek Easy kit (Boehringer Mannheim, Inc., Indianapolis, IN). Urine acetoacetate was determined with Multistix 10 SG reagent strips (Bayer, Inc., Elkhart, IN). Heparinized arterial blood (1.0 to 1.2 ml) was centrifuged and the plasma was decanted. Plasma leptin levels were measured with a mouse leptin radioimmunoassay (RIA) kit (Linco Research, Inc., St. Charles, MO). Plasma insulin levels were measured with an ultrasensitive rat insulin RIA kit (100% cross-reactivity with mouse insulin) (Linco Rersearch). Plasma -hydroxybutyrate levels were determined with -hydroxybutyrate kits from Sigma (St. Louis, MO).

Diaphragm Glycosylation

NOD-Ltj mice were killed with pentobarbital (60 mg, intraperitoneally) under halothane anesthesia. Their diaphragms were excised, washed with 0.01 M Tris-HCl, pH 7.4, until the wash solution was clear, and then quick frozen in a dry ice/100% methanol slurry and stored at 70° C. Before biochemical analysis, the diaphragms were washed in 0.01 M Tris-HCl (pH 7.4), dehydrated in a stepwise gradient of methanol solutions and delipidated overnight in a chloroform-methanol 2:1 mixture. On the following morning the diaphragms were homogenized in 1 ml of 0.01 M Tris-HCl, pH 7.4. Hydroxyproline was determined according to Huszar and colleagues (13) and then converted to collagen, assuming a hydroxyproline content of 14%. Total protein was measured by the method of Lowry and associates (14) using albumin as a standard. The glucose content of diaphragm tissue was measured with the anthrone reaction (15), using glucose as a standard. Glycosylation of diaphragm was calculated as both the glucose/total protein and glucose/collagen (wt/wt) ratios.

Experimental Design: Part A

Regular diet. Eighteen C57BL/6J wild-type mice were maintained on a regular ad libitum chow diet. IDDM was induced in 12 mice (90 ± 4 d; study group) by administration of STZ (Sigma) injected intraperitoneally at 200 mg/kg in 0.5 ml of citrate buffer, pH 4.8, as previously described (10). Six mice received citrate buffer alone (86 ± 9 d [mean ± SEM]; control group). Ventilatory control was assessed before STZ or sham injection (Day 0) and 7 d after the injection (Day 7).

High-fat diet. Twenty wild-type C57BL/6J mice were fed a high-fat diet (49% fat; 5.8 kcal/g) for approximately 16 wk until they reached 40 to 45 g in body weight, and then electrodes were implanted for polysomnography. Fourteen mice (129 ± 9 d of age) were injected with STZ at 200 mg/kg, the remaining six mice (136 ± 16 d of age) were food-restricted to match the weight loss in the STZ group. Ventilatory control measurements were performed before (Day 0) and after STZ administration (Day 7), or at the beginning (Day 0) and end (Day 7) of the food-restriction period.

Seven C57BL/6J-Lep^ob mice (96 ± 9 d of age) were fed a high-fat diet for 1 wk, after which their baseline ventilatory control was assessed (Day 0). Subsequently, all C57BL/6J-Lep^ob mice were injected with STZ as described earlier, and ventilatory control was reassessed 7 d later (Day 7). Four remaining animals from this group were used as controls.

E was determined as the change in E per unit body weight from Day 0 to Day 7, and was expressed as a percent of E per unit body weight on Day 0.

Experimental Design: Part B

We maintained 10 female NOD-Ltj mice on a regular ad libitum chow diet, and measured their ventilatory control and fasting blood glucose on a weekly basis starting from the age of 10 wk. Diabetes was diagnosed when the fasting blood glucose exceeded 200 mg/dl. The last experiment was performed on Week 4 of diabetes or at the age of 30 wk. HCVR was determined as the change in HCVR per unit body weight from the onset of diabetes (Day 0), and was expressed as a percent of HCVR per unit body weight on Day 0.

Measurement of Metabolism

Metabolic parameters were assessed in two groups of animals. Six wild-type male C57BL/6J mice (age 84 ± 7 d; body weight 27.5 ± 0.5 g) were fed a regular chow diet and four wild-type male C57BL/6J mice (age 140 ± 21 d; body weight 43.9 ± 1.9 g) were fed a high-fat diet according to the experimental protocol for Part A. Metabolic parameters were measured over a 2-h period, using indirect, open-circuit calorimetry with a two-chamber Oxymax system (Columbus Instruments, Columbus, OH). Data from the first half-hour of equilibration were discarded. Upon completion of the experiment, mice were given 200 mg/kg STZ. The second measurement of metabolism was performed on Day 7 after STZ administration.

Statistical Analyses

Data were analyzed with Microsoft Excel (Microsoft Corporation, Seattle, WA) and Statview (SAS Institute Inc., Cary, NC) software, and results are presented as the mean ± SEM. The statistical significance of between-group comparisons was determined with t tests. The statistical significance of pre- versus postdiabetes data was assessed within each sleep/wake state, using intrasubject analysis of variance. Correlation analyses were done with linear regression based on the least-squares method.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Part A: STZ-Induced Diabetes in C57BL/6J Mice Fed with a Regular Chow Diet

Metabolic parameters. Diabetes with a fasting blood glucose level of more than 200 mg/dl was observed in 11 of 12 mice on Day 2 after injection with STZ. By Day 7, all mice in this group had diabetes, with fasting blood glucose levels averaging 485 ± 14 mg/dl (mean ± SEM) (Table 1). Mice with STZ-induced diabetes had reduced plasma insulin levels (0.23 ± 0.09 versus 0.96 ± 0.13 ng/ ml in the control group, p < 0.01). As compared with the control group, the animals with diabetes increased their food intake and lost weight (Table 1). Their plasma leptin level was the same as in control mice, despite a lower final body weight. Notably, none of the mice with STZ-induced diabetes had urinary ketones or increased plasma -hydroxybutyrate levels (Table 1).

Metabolic measurements were made before and after STZ administration. There were no significant differences between VO₂, VCO₂, and the respiratory exchange ratio on Day 0 (2,605 ± 152 ml/min, 2,312 ± 133 ml/min, and 0.89 ± 0.02, respectively) and on Day 7 (2,953 ± 210 ml/min, 2,505 ± 153 ml/min, and 0.85 ± 0.02, respectively).

Baseline minute ventilation. During both wakefulness and NREM sleep, baseline E significantly declined from Day 0 to Day 7 after STZ injection (Figure 1A). The decline in baseline E was due to a decrease (p < 0.01) in breathing frequency from 3.24 ± 0.21 Hz to 2.65 ± 0.27 Hz during wakefulness, and from 3.34 ± 0.11 Hz to 2.63 ± 0.20 Hz during NREM sleep. There was no significant change in baseline VT when it was corrected per body weight. Because rapid-eye-movement (REM) sleep was not consistently observed in diabetic animals, only NREM sleep data are presented. In control, sham-injected mice, baseline E did not change in either the sleep or waking state between Day 0 and Day 7 (Figure 1A). Since animals in the STZ group lost weight and control animals did not, ventilatory measurements were normalized per body weight to account for weight-related changes in E.

View larger version (16K): [in this window] [in a new window]   Figure 1.   Effect of STZ-induced diabetes on baseline E (A) and HCVR (B) in lean C57BL/6J mice during wakefulness and NREM sleep. Results presented show the mean ± SEM before (Day 0) and after (Day 7) STZ administration. Statistical significance of the difference between pre- and post-STZ parameters during each sleep/wake state was derived with two-way intrasubject ANOVA. * p < 0.05 for difference between Day 0 and Day 7.

HCVR. During both wakefulness and NREM sleep, HCVR significantly declined from Day 0 to Day 7 after STZ injection, but did not change in either the sleep or waking state in control sham-injected mice (Figure 1B).

These data show that STZ-induced diabetes leads to ventilatory depression during both wakefulness and NREM sleep. However, the decreases in baseline E and HCVR were not associated with a decrease in plasma leptin levels (Table 1) or a change in metabolic rate.

Part A: STZ-Induced Diabetes in C57BL/6J Mice Fed a High-Fat Diet

Metabolic parameters. Diabetes with a fasting blood glucose level of more than 200 mg/dl was observed in 17 of 18 wild-type mice on a high-fat diet by Day 2 after STZ injection. By Day 7, all mice in this group had diabetes, with a fasting blood glucose level of 442 ± 22 mg/dl (Table 1). After STZ administration, the mice lost approximately 20% of their body weight and reduced their food intake by more than 50%. Plasma insulin levels in these mice were almost threefold lower and plasma leptin levels were more than twice as high as in food-restricted mice with matched body weights (Table 1). All wild-type C57BL/6J diabetic mice maintained on a high-fat diet had ketones in their urine and increased -hydroxybutyrate levels in their plasma. Plasma -hydroxybutyrate levels in the C57BL/ 6J diabetic mice on the high-fat diet were significantly higher than in diabetic C57BL/6J-Lep^ob mice or in food-restricted wild-type mice (Table 1). STZ-treated C57BL/6J-Lep^ob mice also developed diabetes, with a significant increase in their fasting blood glucose level (Table 1). Unlike wild-type mice on a high-fat diet, these leptin-deficient animals did not exhibit significant decreases in body weight or food intake. None of the diabetic C57BL/6J-Lep^ob mice had urinary ketones or a significant increase in plasma -hydroxybutyrate as compared with food-restricted wild-type mice (Table 1). Plasma insulin levels in C57BL/6J-Lep^ob mice after STZ administration (22.1 ± 8.2 ng/ml) were higher than in either wild-type group (Table 1), but lower than previously reported normal values in ob/ob mice (16).

CO₂ production and O₂ consumption were measured in four wild-type C57BL/6J mice maintained on a high-fat diet before and 7 d after STZ injection. There were no significant differences between O₂, CO₂, and the respiratory exchange ratio on Day 0 (2,124 ± 219 ml/min, 1,731 ± 152 ml/min, and 0.82 ± 0.03, respectively) and on Day 7 (1,998 ± 111 ml/min, 1,531 ± 106 ml/min, and 0.76 ± 0.02, respectively).

Baseline E. In contrast to the wild-type STZ mice on the regular diet (Figure 1A), wild-type mice on the high-fat diet showed a trend toward increasing baseline E during both wakefulness and NREM sleep, although the increase did not reach statistical significance (Figure 2A). This trend in baseline E was due to an increase in VT, both during wakefulness (from 7.5 ± 0.3 µl/g body weight on Day 0 to 8.6 ± 0.5 µl/g body weight on Day 7; p < 0.05) and NREM sleep (from 6.8 ± 0.2 µl/g body weight on Day 0 to 7.8 ± 0.6 µl/g body weight on Day 7, p < 0.05). The respiratory rate did not change significantly from Day 0 to Day 7 after STZ administration. When individual mice were compared, it appeared that six of the 14 C57BL/6J obese wild-type mice had an obvious increase in E during both NREM sleep (Figure 3A, thick lines) and wakefulness (not shown), whereas the other eight mice had either a decrease or did not have a change in E during either stage (Figure 3A, thin lines). This heterogeneity was not observed among C57BL/6J-Lep^ob mice injected with STZ (Figure 3B) or among food-restricted wild-type mice (Figure 3C), as E either did not change consistently or declined from Day 0 to Day 7 in both groups of mice. These data are further analyzed in the subsequent section on diabetic ketoacidosis, leptin, and hyperventilation.

View larger version (22K): [in this window] [in a new window]   Figure 2.   Effect of STZ-induced diabetes on baseline E (A) and HCVR (B) in C57BL/6J wild-type and C57BL/6J-Lepob mice on a high-fat diet. Results presented show the mean ± SEM before (Day 0) and after (Day 7) STZ administration. Control data for mean ± SEM E and HVCR are shown for C57BL/6J wild-type mice before (Day 0) and after (Day 7) food restriction. *p < 0.05 versus Day 0; two-way ANOVA.

View larger version (14K): [in this window] [in a new window]   Figure 3.   Effect of STZ-induced diabetes on baseline E during NREM sleep in wild-type mice (A) and in C57BL/6J-Lepob mice (B) maintained on a high-fat diet. (C) Effect of food restriction on E during NREM sleep in wild-type C57BL/6J on a high-fat diet. Each line represents one mouse, and bold lines represent animals that increased their E (hyperventilators).

HCVR. HCVR during both wakefulness and NREM sleep decreased in wild-type diabetic mice on a high-fat diet after STZ injection (Figure 2B). This finding was consistent with the data for the STZ-injected mice fed a regular diet (Figure 1B). C57BL/6J-Lep^ob mice had a markedly depressed HCVR before STZ injection, as expected (8, 9). Neither C57BL/6J- Lep^ob mice with STZ-induced diabetes nor food-restricted wild-type mice had a significant change in HCVR from Day 0 to Day 7 (Figure 2B).

Diabetic ketoacidosis, leptin, and hyperventilation. We compared the six diabetic wild-type mice on a high-fat diet and with an increased baseline in the E from Day 0 to Day 7 (hyperventilators) with the eight mice in which in the in the E was unchanged or decreased (nonhyperventilators; Figure 3A). Mice in both subgroups lost a similar percentage of their body weight (22.0 ± 1.5% and 19.0 ± 2.0% in the hyper- and nonhyperventilators groups, respectively), and had similar levels of plasma insulin (0.57 ± 0.15 ng/ml and 0.63 ± 0.13 ng/ ml, respectively) and -hydroxybutyrate (5.7 ± 1.1 mmol/L and 7.4 ± 0.5 mmol/L, respectively). However, significant differences between the two subgroups were found in plasma leptin levels and food consumption by Day 7. Hyperventilating mice became anorectic, with a food intake that was approximately 40% and a plasma leptin level that was approximately 250% of that in nonhyperventilating mice (Figure 4). Linear regression analysis of the data for all 14 wild-type mice in the STZ group showed a positive correlation between plasma leptin levels and the change in E from Day 0 to Day 7 both during NREM sleep (Figure 5) and wakefulness (r = 0.74, p < 0.01). There was a negative correlation between last-day food intake and change in E during NREM sleep and wakefulness (r = 0.66 and 0.63, respectively; p < 0.02). Blood glucose levels, plasma insulin, and urinary and plasma ketones did not correlate significantly with the degree of hyperventilation.

View larger version (12K): [in this window] [in a new window]   Figure 4.   Food intake (A) and plasma leptin level (B) on Day 7 in C57BL/6J wild-type mice with STZ-induced diabetes and on a high-fat diet. Mice with E on Day 7   E on Day 0 (nonhyperventilators), and mice with E on Day 7 >  E on Day 0 (hyperventilators) were compared. Results presented show mean ± SEM values. Statistically significant differences between hyperventilators and nonhyperventilators were derived with unpaired t tests.

View larger version (13K): [in this window] [in a new window]   Figure 5.   Relationship between plasma leptin level and E during NREM sleep in C57BL/6J wild-type mice with STZ-induced diabetes and on a high-fat diet. E was determined as the change in E per body weight from Day 0 to Day 7, and was expressed in percent of E per body weight on Day 0. Each data point represents one animal.

Part B: Spontaneous IDDM in NOD-Ltj Mice

Metabolic parameters. To control for nonspecific detrimental effects of STZ on ventilation, we studied NOD-Ltj mice with spontaneous IDDM. Seven of 10 NOD-Ltj mice spontaneously developed IDDM with a fasting blood glucose level of more than 200 mg/dl by 14 ± 1 wk of age. By Week 2 of hyperglycemia, diabetic mice significantly increased their food intake, from 3.6 ± 0.1 g/d to 9.3 ± 1.0 g/d (Table 2). Despite a high level of food consumption, diabetic mice lost significant amounts of weight, and by Week 4 they weighed almost 30% less than their nondiabetic counterparts. Interestingly, the plasma insulin level in diabetic NOD-Ltj mice was not significantly different than in the nondiabetic control mice (p = 0.15; Table 3). On the other hand, nondiabetic NOD-Ltj mice had lower plasma insulin levels than did control wild-type C57BL/6J mice of comparable body weight (Table 1; p < 0.05), which was consistent with an insulin deficiency even in those NOD-Ltj mice that did not have overt diabetes. The plasma leptin level in diabetic mice was three times as high as in nondiabetic mice despite the approximately 30% lower body weight of the diabetic animals (Table 3). Urine ketones were detected in four of seven diabetic mice, which was reflected in the increased levels of plasma -hydroxybutyrate in this group (Table 3).

Baseline E. E was measured weekly in NOD-Ltj mice, and did not change over the 4-wk course of diabetes (Figure 6A). These data differed from those for C57BL/6J mice with STZ diabetes and fed a regular diet, which exhibited a decline in E (Figure 1A).

View larger version (17K): [in this window] [in a new window]   Figure 6.   Effect of spontaneously developed IDDM on respiratory control in diabetic NOD-Ltj mice (n = 7). IDDM was diagnosed when the fasting blood glucose level exceeded 200 mg/dl. E and HCVR (A and B, respectively) were measured weekly during quiet wakefulness. After retrospective analysis, the data points of 2 wk (2 wk before the onset of diabetes), 0 wk (the week of onset of diabetes), and 1, 2, 3, and 4 wk of diabetes were selected and analyzed. Results presented show mean ± SEM values. Statistically significant differences between ventilatory parameters depending on the time course of diabetes were derived with paired t tests. * p < 0.001 versus Week 2;  p < 0.01 versus Week 2.

HCVR. All diabetic NOD-Ltj mice exhibited a steady decrease in HCVR over the 4-wk course of diabetes (Figure 6B). This finding is consistent with the decrease in HCVR with STZ-induced diabetes in C57BL/6J mice on either a regular diet (Figure 1B) or a high-fat diet (Figure 2B), confirming that the depression in HCVR is a complication of diabetes rather than a result of nonspecific STZ toxicity. In nondiabetic NOD-Ltj mice, HCVR did not change significantly: at the age of 13 wk, HCVR was 0.59 ± 0.06 ml/min/%CO₂/g, and at 30 wk HCVR was 0.49 ± 0.05 ml/min/%CO₂/g.

Hyperglycemia and glycosylation of the diaphragm. Linear regression analysis showed a lack of significant correlation between the decline in HCVR and plasma levels of insulin (r = 0.21, p > 0.1), leptin (r = 0.04, p > 0.1) and -hydroxybutyrate (r = 0.44, p > 0.01) in NOD-Ltj mice, suggesting that hypoinsulinemia, hyperleptinemia, and ketoacidosis did not account for the ventilatory depression in these animals. On the other hand, a significant correlation existed between the decline in HCVR and the time course of hyperglycemia (r = 0.79, p < 0.001; Figure 7).

View larger version (13K): [in this window] [in a new window]   Figure 7.   Relationship between the duration of hyperglycemia, defined as fasting blood glucose > 200 mg/ dl, and HCVR. The HCVR was determined as the change in HCVR per body weight from Day 0, and was expressed in percent of HCVR per body weight on Day 0. Each animal is represented by four data points, corresponding to the daily time course of diabetes.

Since hyperglycemia may lead to glycosylation of proteins, and particularly collagen in the diaphragm (17), we determined levels of protein glycosylation in the diaphragms of NOD-Ltj mice. Both glucose/total protein and glucose/collagen ratios in diabetic NOD-Ltj mice were significantly higher than in nondiabetic mice, in accord with significant glycosylation of the diaphragm (Figure 8).

View larger version (13K): [in this window] [in a new window]   Figure 8.   Glycosylation of the diaphragm in NOD-Ltj mice. Glucose/ collagen and glucose/ total protein ratios were calculated on a weight (mg)/weight (mg) basis (A and B, respectively). Horizontal lines represent the mean. Statistically significant differences between diabetic and nondiabetic mice were derived with unpaired t tests.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In this study, we used mouse models of IDDM to assess the effect of diabetes on respiratory control. We made a number of novel observations about the impact of IDDM on ventilation. First, we showed that depression of HCVR occurs in both STZ-induced and spontaneous IDDM in the mouse, and that ventilatory depression in spontaneous IDDM follows the time course of the disease. Second, decreases in HCVR in both obese and nonobese C57BL/6J mice and NOD-Ltj mice were unrelated to decreases in leptin levels, in contrast to our original hypothesis. Rather, the time course of the depression of HCVR in the NOD-Ltj model was related to the duration of hyperglycemia, and was associated with glycosylation of the diaphragm. In contrast to the case with HCVR, considerable variability was observed in E in mice on a high-fat diet that developed diabetic ketoacidosis (DKA). A major finding of our study was that the ability to hyperventilate in response to DKA depended on the presence of increased leptin levels. Thus, leptin did not affect HCVR in IDDM, but played an essential role in stimulating baseline E in response to DKA. In the discussion that follows, we outline our experimental approach, the mechanisms for the depression of HCVR in IDDM, the role of leptin in the hyperventilatory response to DKA, and the potential implications of this work.

Experimental Approach to Ventilatory Control in Mouse Models of IDDM

STZ is commonly used to induce IDDM by targeting  cells of the pancreatic islets (10). In order to examine the effect of IDDM on respiratory control, we measured ventilation in mice before and 1 wk after STZ administration. Besides its obvious advantages, including technical simplicity and a very high percentage of animals acquiring diabetes (10), an STZ model of IDDM imposes certain limitations. These limitations include potential nonspecific effects related to toxicity from STZ, significant morbidity that precludes long-term experiments, and variability between animals in physiologic responses to a specific dose of STZ (10, 18). In an effort to overcome these limitations, we concurrently studied IDDM in NOD-Ltj mice. NOD-Ltj mice acquire IDDM spontaneously from autoimmune injury of the pancreatic -cells, and about 85% of female NOD-Ltj mice develop IDDM by 30 wk of age (12). With these mice, we confirmed that ventilatory depression was associated with the development of IDDM.

Another potential limitation of ventilatory measurements concerns changes in E and HCVR during different sleep/ wake states, since both E and HCVR are depressed during sleep (9). In order to overcome this limitation, we determined sleep/wake state simultaneously with ventilatory measurements in C57BL/6J mice. Our data demonstrate that respiratory depression was present during both wakefulness and NREM sleep after STZ administration (Figures 1 and 2B), indicating that the depression in ventilation was not a nonspecific effect of increased somnolence.

Potential Mechanisms for HCVR Suppression in IDDM

Our findings in both STZ-treated mice and NOD-Ltj mice confirmed previous observations in rats (4) and humans (2, 3) that IDDM leads to a decrease in HCVR. Our data indicate that insulin deficiency, per se, is unlikely to play a direct role in suppression of HCVR, since no significant difference in insulin levels was observed between diabetic (depressed HCVR) and nondiabetic (normal HCVR) NOD-Ltj mice (Table 2; Figure 6). Furthermore, no correlation existed between HCVR and insulin levels in C57BL/6J mice with STZ-induced diabetes. Thus, some factor(s) other than decreased insulin levels probably accounts for ventilatory depression in IDDM.

One possible mechanism is that leptin provides a link between the metabolic disturbances of IDDM and changes in ventilatory control. We have previously shown that leptin deficiency leads to a severe reduction of HCVR (8, 9). Furthermore, it has been proposed that insulin deficiency may lead to a decrease in leptin levels in IDDM (6, 7). On the basis of these facts, we initially hypothesized that leptin was the link between diabetes and depressed ventilatory control. However, our study data showed that plasma leptin levels were not decreased in either obese or nonobese C57BL/6J mice with STZ-induced diabetes, or in NOD-Ltj mice with spontaneous diabetes, indicating that insulin deficiency does not necessarily result in leptin deficiency. Indeed, our data suggest that leptin levels were actually increased in IDDM, particularly in spontaneous NOD-Ltj diabetes and in STZ-induced diabetes with ketoacidosis. On the basis of the data presented in the present study, a decrease in leptin levels cannot account for the decreased HCVR in IDDM.

A second potential mechanism of ventilatory depression involves changes in metabolic rate that accompany IDDM (4). However, there were no significant differences in O₂ or CO₂ in mice before and after administration of STZ, indicating that ventilatory suppression was not due to a decrease in metabolic rate. A third possibility is that metabolic acidosis impairs the diaphragm and therefore results in ventilatory depression, as previously shown by Howell and colleagues in dogs (19). In our study, however, it is unlikely that metabolic acidosis contributed to the development of respiratory failure, since STZ-treated mice on a regular diet had significant ventilatory impairment without metabolic acidosis. Thus, metabolic changes associated with IDDM do not appear to have been responsible for the ventilatory depression in our study.

Another possibility is that the hyperglycemia that occurs during IDDM causes a depression of ventilatory responses to CO₂. Indeed, our present data for NOD-Ltj mice showed that the depression of HCVR strongly correlated with the duration of hyperglycemia (Figure 7). Human data also show a correlation between blood glucose levels and decrements in respiratory function such as, FEV₁, FVC, and lung volume (20, 21). One effect of prolonged hyperglycemia is glycosylation of proteins, as demonstrated in diabetic patients and animals (17). Nonenzymatic glycosylation of collagen in the diaphragm, as well as in other organs and tissues, leads to formation of fructoselysine, (3,4)N -(carboxymethyl) lysine, and pendosidine (17, 22). These compounds in turn cause cross-linking of collagen and stiffening of the diaphragm (22). As we show in Figure 8, there was a significant increase in the glucose content of the diaphragms from diabetic NOD-Ltj mice, suggesting that diaphragm glycosylation may play a role in ventilatory suppression in diabetes. Although we demonstrate an association between hyperglycemia and glycosylation of the diaphragm, it is possible that other complications of prolonged hyperglycemia, in particular autonomic neuropathy (23), also contribute to ventilatory depression.

DKA and Baseline Ventilation

In addition to demonstrating a depressed HCVR, our study demonstrated substantial variability in baseline E in response to IDDM. In particular, E varied markedly in STZ-treated mice on a high-fat diet (Figure 3), despite a uniformly two- to threefold higher -hydroxybutyrate level than in other groups of mice (Table 1). Only a proportion of these animals with increased ketones, and presumably reduced pH, developed hyperventilation. Although we were unable to directly measure arterial pH because of the morbidity associated with catheterizing animals already compromised by administration of STZ, the markedly increased -hydroxybutyrate levels would be consistent with the development of a significant metabolic acidosis (Table 1). Serum -hydroxybutyrate levels have been previously shown to correlate directly with the anion gap and inversely with serum bicarbonate (24, 25). Since both hyperventilating and nonhyperventilating wild-type mice had similar levels of -hydroxybutyrate, we propose that mice in both groups had a similar degree of metabolic acidosis (26). Therefore, it is unlikely that the degree of acidemia alone could account for the observed differences in ventilation in wild-type mice during DKA. Rather, the major difference between mice that hyperventilated and those that did not hyperventilate in the presence of DKA was the level of circulating leptin. Hyperventilation occurred during ketoacidosis only when accompanied by elevated leptin levels.

The mechanism(s) responsible for an increase in leptin in animals exhibiting DKA is not clear. There is no evidence in the literature that acidic pH or an excess of -hydroxybutyrate can affect leptin levels (27, 28). However, one putative factor that could enhance leptin levels in diabetes is cholecystokinin (CCK) (29). Since both a high-fat diet and IDDM are known to increase plasma levels of CCK (30, 31), it is likely that STZ-induced diabetes in mice on a high-fat diet resulted in elevated CCK levels. As noted earlier (29), such an increased CCK level would be expected to cause a corresponding increase in plasma leptin levels. It is not clear, however, why leptin was increased only in some mice on a high-fat diet after STZ treatment. Nevertheless, it was evident that hyperventilation occurred only when DKA and hyperleptinemia were present simultaneously.

In conclusion, the association between leptin, ketones, and hyperventilation suggests that hyperleptinemia may be essential for hyperventilation to occur in ketoacidosis. We have previously reported that leptin-deficient ob/ob mice are prone to hypoventilation and respiratory acidosis, both of which could be reversed by leptin infusion (9). Our present results, in combination with our previous data, strongly suggest that leptin is important for ventilatory responses to acidosis in general. Indeed, recent studies using in situ hybridization and immunohistochemistry have demonstrated leptin receptors in abundance in the nucleus tractus solitarii and other respiratory control centers in the medulla (32, 33) involved in ventilatory responses to CO₂ and pH (34). Thus, it is conceivable that leptin enhances ventilatory responses to acidosis by interacting with leptin receptors on respiratory neurons in the medulla.

The results of our study provide at least two implications of clinical relevance, as follows:

1. Ventilatory depression in IDDM follows the time course of hyperglycemia and is associated with glycosylation of the diaphragm. This ventilatory depression is further enhanced during sleep, which suggests that hyperglycemia may have a negative impact on diabetic patients who have breathing abnormalities during sleep. Rigorous glucose control may improve the management of respiratory failure and sleep-associated breathing abnormalities in patients with diabetes mellitus.

2. Hyperventilation in response to ketoacidosis is dependent on elevated leptin levels. Taken together with our previous finding that leptin reverses respiratory acidosis (9), this raises the question of whether leptin can be used as a respiratory stimulant in patients with acidosis.