Mechanical hyperalgesia correlates with insulin deficiency in normoglycemic streptozotocin-treated rats

www.elsevier.com/locate/ynbdi Neurobiology of Disease 24 (2006) 384 – 394

Mechanical hyperalgesia correlates with insulin deficiency in normoglycemic streptozotocin-treated rats Dmitry Romanovsky, a Nancy F. Cruz, c Gerald A. Dienel, c and Maxim Dobretsov a,b,⁎ a

Department of Anesthesiology, University of Arkansas for Medical Sciences, 4301 West Markham St., Little Rock, AR 72205, USA Department of Neurobiology and Developmental Sciences, University of Arkansas for Medical Sciences, 4301 West Markham St., Little Rock, AR 72205, USA c Department of Neurology, University of Arkansas for Medical Sciences, 4301 West Markham St., Little Rock, AR 72205, USA b

Received 1 June 2006; revised 20 July 2006; accepted 21 July 2006 Available online 28 August 2006 The triggers and pathogenesis of peripheral diabetic neuropathy are poorly understood, and this study evaluated the role of insulinopenia in nociceptive abnormalities in the streptozotocin (STZ) rat model of diabetes to test the hypothesis that, in addition to hyperglycemia, impairment of insulin signaling may be involved in progression of neuropathy. We measured blood glucose, plasma insulin, and sciatic nerve glucose and sorbitol levels, and withdrawal thresholds for hind limb pressure pain and heat pain in STZ-injected rats that developed hyperglycemia or remained normoglycemic. The pressure pain threshold did not change in vehicle-injected controls, but during the 2 weeks after STZ, it decreased by 25–40% in STZhyperglycemic and STZ-normoglycemic animals (P < 0.05). Mean heat pain threshold did not change in STZ-normoglycemic rats, but increased by about 1.5°C in STZ-hyperglycemic rats (P < 0.05). These pain thresholds did not correlate with blood or nerve glucose or sorbitol levels, but both correlated with plasma insulin level in STZ-normoglycemic rats, and low-dose insulin replacement normalized the pressure threshold without affecting blood glucose level. Thus, at least one of early signs of diabetic neuropathy in STZtreated rats, mechanical hyperalgesia, can be triggered by moderate insulinopenia, irrespective of glycemic status of the animals. © 2006 Elsevier Inc. All rights reserved. Keywords: Mechanical hyperalgesia; Pain; Diabetic neuropathy; Insulin; Hyperglycemia; Streptozotocin; Type I diabetes

Introduction Chronic, systemic hyperglycemia has been historically viewed as a major trigger for the pathogenic mechanisms of neurologic complications of type I and type II diabetes (Ishii, 1995; Sugimoto et al., 2000a). For example, distal peripheral neuropathy is a ⁎ Corresponding author. Department of Anesthesiology, Slot 515, University of Arkansas for Medical Sciences, 4301 West Markham Street, Little Rock, AR 72205, USA. Fax: +1 501 603 1951. E-mail address: [email protected] (M. Dobretsov). Available online on ScienceDirect (www.sciencedirect.com). 0969-9961/$ - see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.nbd.2006.07.009

frequent symptom of diabetes, and most symptoms of this neuropathy may be explained as either immediate or secondary consequences of hyperglycemia-associated activation of aldose reductase (the first enzyme in polyol pathway), which causes nerve sorbitol accumulation followed by osmotic, metabolic, oxidative and circulatory imbalance in peripheral nervous tissue (Sugimoto et al., 2000a). Further support for the hyperglycemic hypothesis comes from clinical trials showing that stringent glycemic control decreases the incidence of peripheral neuropathy by as much as 60% to 70% over several years (Boulton et al., 2004; DCCT, 1993), and from animal research demonstrating parallel development of hyperglycemia, nerve conduction slowing, and nociceptive impairments and an apparent association between hyperglycemia, sensitivity of cutaneous nociceptors, and mechanical hyperalgesia (Biessels et al., 1999; Dobretsov et al., 2001, 2003; Khan et al., 2002; Sugimoto et al., 2000a; Suzuki et al., 2002). In spite of persuasive supporting evidence, the glucose hypothesis alone does not fully explain the variety and complexity of clinical presentations of distal peripheral neuropathy. While glucose control is an effective preventive measure, during 5 years of follow up study at least 30% of diabetic patients with apparently satisfactory control of blood glucose level still developed neuropathy (DCCT, 1993). Furthermore, there is a high incidence of painful neuropathies in pre-diabetic patients who have impaired glucose tolerance but are normoglycemic or moderately hyperglycemic (Singleton et al., 2001). Moreover, neither clinical (Chan et al., 1990; Thye-Ronn et al., 1994) nor animal (Courteix et al., 1996; Dobretsov et al., 2003; Maneuf et al., 2004; Romanovsky et al., 2004) studies have shown a correlation between hyperglycemic status and pain/ hyperalgesia. Taken together, these observations suggest that factors in addition to hyperglycemia must play a role in the pathogenesis of distal peripheral neuropathy, and identification and evaluation of the causative mechanisms of peripheral neuropathy are important, not only for understanding the clinical course of the disorder, but also for the development of new strategies for its treatment and prevention. Recent data from several independent laboratories implicate insulinopenia as a contributory factor in development of peripheral neuropathy (Brussee et al., 2004; Huang et al., 2003; Schmidt et

D. Romanovsky et al. / Neurobiology of Disease 24 (2006) 384–394

al., 2004), and our initial studies suggest that mechanical hyperalgesia, a symptom of distal peripheral neuropathy, is caused by insulinopenia without accompanying hyperglycemia in a rat model using the pancreatic toxin streptozotocin (STZ) (Romanovsky et al., 2004). Establishing a neuropathic role for insufficiency of insulin signaling may help explain the apparent inconsistencies related to the glucose dependence of the origin of pain, especially in pre-diabetic patients, and the present study was, therefore, designed to examine relationships among insulinopenia, hyperglycemia, and early signs of neuropathy reflected by abnormalities of nociception in STZ-treated rats. Materials and Methods All experimental procedures followed “Principles of laboratory animal care” (NIH publication no. 85-23, revised 1985) and were reviewed and approved by the Institutional Animal Care and Use Committee.

385

Paw pressure withdrawal threshold Pressure pain threshold was measured with an Ugo Basil Analgesy-meter (Stoelting, Wood Dale, IL, USA) as described previously (Romanovsky et al., 2004). Briefly, each test session consisted of five trials (separated by at least 15 min) on both the left and right paws of each rat. In each trial, pressure was applied to the center of the hind paw at a linearly increasing rate of 16 g s− 1 until the animal withdrew the paw, struggled, or in rare cases vocalized; to avoid tissue damage, the pressure cutoff of the device was set at 250 g. Any trial in which pressure cutoff was reached or animal’s behavioral reaction could not be unequivocally attributed to withdrawal reflex was discounted and repeated. In successful trials, nociceptive pain threshold (expressed as mass units, g) was recorded from the Analgesy-meter and the mean bilateral pressure threshold for each test session was calculated for each animal as average of ten (five per paw) threshold readings. Hot plate withdrawal threshold

Animal procedures All behavioral tests were carried out between 2 and 6 PM on male Sprague–Dawley rats (200–300 g, Harlan Inc., Indianapolis, IN); only one test per day was given to any animal. During the week before injection of STZ, rats were first acclimatized to the testing apparatus, then baseline nociceptive threshold measurements were conducted; baseline results of 3–4 pressure pain threshold or 1–2 heat pain threshold test sessions (see below) were averaged and designated as day 0 values. After acclimation and baseline tests, rats were randomly assigned to experimental or control groups. Experimental rats were injected with STZ (65 mg kg− 1 body weight in 33 mM citrate-buffered saline given i.p.; pH 4.5) while control animals received an equivalent injection of vehicle, citrate-buffered saline. Tail blood samples were taken for glucose determination from overnight-fasted animals on the day before STZ injection and on days 3, 8 and 13 thereafter. Based on the day 3 fasting blood glucose level, the rats were categorized as normoglycemic or hyperglycemic, using a cutoff value of > 6.9 mM to define hyperglycemia (American Diabetes Association, 2006). Pressure pain threshold was measured on days 3, 6, 9, 11 and 14 after STZ treatment, whereas heat pain threshold was measured on days 4 and 12. When behavioral testing followed the fasting glucose measurements, at least 6 h of free access to the food was allowed to animals before the behavioral data collection. Samples for determination of plasma insulin and sciatic nerve metabolite levels were collected within 6 h after the last pain threshold test on day 14 after STZ from animals that were deeply anesthetized with halothane. Insulin replacement therapy was carried out in separate groups of rats that were first carried through the procedures described above. Then, starting on day 8 after STZ injection, each control and STZ-injected rat was given two daily i.p. injections of 0.2–0.4 U of insulin (human recombinant insulin, Novolin R, Novo Nordisk Pharmaceuticals Inc., Princeton, NJ) in physiological saline for 7 days. This insulin dose is about 1/10 that required to correct hyperglycemia in diabetic animals, and a lower dose was used to avoid causing hypoglycemia in the normoglycemic rats (Brussee et al., 2004). In the insulin replacement rats, PPT was measured on days 4, 6, 8, 11 and 15, and blood for insulin level determination was drawn on day 15.

Heat pain threshold was measured using an incremental Hot/Cold Plate Analgesia Meter (IITC Life Science, Woodland Hills, CA). The animal was placed on a clean and dry aluminum plate maintained at 28°C in the clear plastic chamber of the device. The test started as soon as the rat relaxed, either standing on all four paws, sitting, or grooming its head or forepaws. During the test, the plate temperature increased at a linear rate of 10°C min− 1 with a cutoff temperature of 55°C. Hind limb heat pain threshold was defined as a temperature at which the animal abruptly withdrew either of its hind feet from the plate surface in a sharp move, typically followed by licking of the lifted paw. Trials were discounted if the paw withdrawal appeared to associate with normal grooming behavior or repositioning rather than with reflex behavior. Testing sessions continued at 15–20 min intervals until seven hind limb heat threshold readings were obtained for each animal. The results of these determinations were averaged, and the mean value was used for further analysis. Blood glucose and insulin measurements Blood glucose in tail-prick samples was measured using the colorimetric Accu-Chek blood glucose monitoring system (Roche Diagnostics Corp., Indianapolis, IN, with a manufacturer’s stated 97–99% accuracy and precision over the device operating range of 0.6–33.3 mM glucose). For insulin determination, blood samples were obtained by cardiac-puncture from halothaneanesthetized animals. Insulin was measured in duplicate in each sample with the Ultra Sensitive Rat Insulin ELISA Kit according to the manufacturer’s protocol (Crystal Chem Inc., Downers Grove, IL; range 0.1–64 ng, sensitivity 5 pg insulin/ml). Sciatic nerve glucose, lactate, and sorbitol measurements About 1 cm of each sciatic nerve was exposed by blunt dissection at mid-thigh level, frozen in situ by Fisher’s Freeze’It, excised, and stored at about −80°C until weighed at −25°C in a cryobox (Cahn microbalance). The frozen nerves were minced, thawed at −12°C in 100% ethanol, and homogenized, then serially extracted (4×) with 65% ethanol, centrifuged, lyophilized, re-dissolved in deionized water (Dienel et al., 1990), flash frozen, and stored at −80°C. Glucose and lactate levels were assayed with a YSI Biochemistry Analyzer (Yellow Springs, OH). Sorbitol levels were assayed by

386

D. Romanovsky et al. / Neurobiology of Disease 24 (2006) 384–394

minor modification of the procedure of Sherman and Stewart (1966). The assay mixture contained sample or authentic sorbitol standards in a final volume of 1 ml containing (in mM) 50 Tris–HCl, 2.0 MgCl2, 0.4 NAD+, and 0.8 unit sorbitol dehydrogenase (pH 8.6; Roche Diagnostics Corporation, Indianapolis, IN); NADH fluorescence was assayed with a Farrand (New York) fluorometer. Statistical analysis Data were analyzed using Student’s one- or two-populations t tests, and two-way ANOVA for comparisons between multiple groups. Regression analysis and non-linear curve fitting procedures were performed using a Levenberg–Marquart algorithm of χ2 minimization (Marquardt, 1963) and Origin 7.0 software package (MicroCal, Northampton, MA). Values are mean ± SEM; significant differences were defined by P < 0.05.

cemic) and exhibited progressive weight loss thereafter (Figs. 1A, B); these animals became less sensitive to higher temperatures within 4 days (increased heat pain threshold [HPT], Fig. 1C) and less tolerant to pressure applied to the paw (decreased pressure pain threshold [PPT], Fig. 1D). All these changes were statistically significant in comparison to their within-group control animals and to their respective baseline parameters (P < 0.05, two-population t test). In contrast, 39% of the STZ-injected rats remained normoglycemic (Fig. 1B) for at least 2 weeks; these rats gained weight normally (Fig. 1A) and their heat thresholds were equivalent to those at day 0 and the control group (Figs. 1A, C). However, the STZ-normoglycemic animals did exhibit mechanical hyperalgesia, reflected by decreased pressure pain threshold (P < 0.05; ANOVA and t test; Fig. 1D). Sciatic nerve metabolite levels and plasma insulin

Results Temporal progression of weight loss, blood glucose, and altered pain thresholds Out of 18 rats injected with STZ in the first group of rats, 61% developed overt hyperglycemia by day 3 (STZ-hypergly-

Sciatic nerve glucose and sorbitol levels STZ-hyperglycemic rats were markedly elevated, rising 3.8- and 21.7-fold above their respective control values, whereas they were normal in STZnormoglycemic rats (Figs. 2A, B). Nerve lactate concentrations were similar in all groups, indicating a lack of pronounced endoneurial hypoxia in STZ-hyperglycemic rats (Fig. 2C). STZ-hyperglycemic

Fig. 1. Physiological and behavioral changes at 1–2 weeks after STZ treatment. Baseline values and changes in body weight (A), blood glucose (B), and heat pain (HPT) (C) and pressure pain (PPT) (D) withdrawal thresholds at intervals after STZ injection (see Materials and Methods); values are means ± SEM. Open, hatched, and filled columns represent control, STZ-normoglycemic, and STZ-hyperglycemic groups of animals (n = 6, 7, or 11, respectively). Labels “b” and “c” indicate statistically significant differences from baseline (day 0) or from within-group control means, respectively (P < 0.05, t test).

D. Romanovsky et al. / Neurobiology of Disease 24 (2006) 384–394

rats exhibited marked insulinopenia, whereas STZ-normoglycemic rats were moderately insulinemic (Fig. 2D). Correlative relationships among glucose and insulin levels and nociceptive impairment Animal-to-animal comparisons and regression analysis were applied to gain further insight into relationships among insulinopenia, glucose metabolism and nociceptive pain thresholds. The results of this type of analysis are shown in Fig. 3, which illustrates two significant correlations that predict a greater sensitivity of nociception to plasma insulin level compared to blood glucose concentration. Blood glucose levels of individual STZ-hyperglycemic animals were significantly and inversely proportional to plasma insulin concentration (Fig. 3A; linear regression coefficient, R = − 0.78; P < 0.05). Extrapolation of the solid regression line for the “insulin–blood glucose” relationship in STZ-hyperglycemic rats to the baseline glucose level in control and STZ-normoglycemic rats (dashed line) identifies a threshold value (0.2 ng/ml; arrow in Fig. 3A) below which plasma insulin should fall before fasting hyperglycemia is predicted to become apparent. The pressure pain threshold in STZ-normoglycemic rats is strongly and highly significantly correlated with plasma insulin level (Fig. 3B, triangles and solid line; R = 0.94; P < 0.01).

387

Furthermore, the thermal threshold is tentatively considered to be higher in the group of STZ-normoglycemic rats with lower plasma insulin because a significant negative correlation (R = − 0.94; P < 0.01; Fig. 3C) was obtained if one outlier that exceeded 2 SD of the population mean (circled triangle, Fig. 3C) was excluded. Extrapolation of the two regression lines describing insulin dependencies of both pressure and heat thresholds to their respective baseline values (arrows, Figs. 3B, C) predicts that a plasma insulin concentration of about 1.8 ng/ml is the minimum required to maintain normal thermal and pressure nociception in the rat, and as insulin level falls, there is a progressively larger disruption of both sensory modalities that become maximal in the hyperglycemic groups that have 10-fold lower plasma insulin levels. No significant correlations were found between pressure or heat thresholds and either blood glucose, or nerve glucose, lactate, or sorbitol in any of these groups of animals (data not shown). Insulin replacement studies To determine if low-dose insulin replacement therapy that would have no effect on the animal’s glycemic status could reverse impairment of pressure threshold in STZ-normoglycemic animals, a second group of 16 rats was injected with vehicle or STZ; they were characterized during the first week after STZ as described

Fig. 2. Sciatic nerve metabolite levels at 2 weeks after STZ treatment. Glucose (A), sorbitol (B), lactate (C), and plasma insulin (D) concentrations were measured at 14 days after the STZ injection. Open, hatched, and filled columns represent control, STZ-normoglycemic, and STZ-hyperglycemic groups of animals (the respective group numbers were n = 6, 7, or 11 for nerve metabolites and n = 5, 7, or 8 for insulin levels). Labels “c”, “HG”, and “NG” above the columns indicate statistically significant differences from the mean control (c), STZ-hyperglycemic (HG), and STZ-normoglycemic (NG) group values (P < 0.05, t test).

388

D. Romanovsky et al. / Neurobiology of Disease 24 (2006) 384–394

Fig. 3. Plasma insulin deficit required to evoke abnormal pain withdrawal responses. Relationships between plasma insulin and fasting blood glucose concentrations (A), and pressure pain (PPT) (B) and heat pain (HPT) (C) withdrawal thresholds at 2 weeks after STZ treatment. Open circles, filled triangles, and filled circles represent data points from individual control, STZ-normoglycemic, and STZ-hyperglycemic rats (n = 5, 7, and 8 respectively). Note that some symbols overlap in the hyperglycemic group and that samples for plasma insulin level were drawn within 6 h after the last pressure threshold test on day 14 (see Materials and Methods). The dashed line in A represents the average overnight-fasting blood glucose level in control animals. Dashed and dot-dash lines in B and C represent the average behavioral threshold levels in control and STZ-hyperglycemic rats, respectively. The solid regression lines describe relationships between insulin and blood glucose in STZ-hyperglycemic rats (A; R = −0.78, P = 0.02) and between insulin and pressure threshold or heat threshold in STZnormoglycemic rats (B; R = 0.94, P = 0.002, and C; R = −0.94, P = 0.005, respectively). One data point (circled in C) was excluded from regression analysis of the heat pain threshold data because it deviated by more than 2 SD from the group mean. Arrows and labels indicate the intercepts of the regression lines with the normal mean values in control rats and identify predicted minimal concentrations of plasma insulin required for maintenance of normal fasting glucose and pain thresholds.

above to establish pre-treatment values against which therapeutic efficacy that was initiated on day 8 could be compared. As in the first group (Fig. 1), 60% of the rats injected with STZ developed hyperglycemia and lost weight whereas the normoglycemic rats did not; both groups showed mechanical hyperalgesia on days 4–8 (Figs. 4A–C). One week of insulin replacement restored the pressure threshold of STZ-normoglycemic animals to a level statistically indistinguishable from control values, whereas insulin therapy did not significantly change the pressure threshold in either

the control or STZ-hyperglycemic rats compared to their corresponding pre-treatment values (Fig. 4C). Furthermore, lowdose insulin replacement did not restore weight gain or normalize blood glucose level of STZ-hyperglycemic rats, nor did it affect glucose levels in control or STZ-normoglycemic animals (Fig. 4B). Plasma insulin concentration measured immediately after the last pressure threshold assay was similar in control and STZnormoglycemic animals, but remained very low in the STZhyperglycemic rats (Fig. 4D).

D. Romanovsky et al. / Neurobiology of Disease 24 (2006) 384–394

389

Fig. 4. Insulin replacement normalizes pressure pain threshold. Changes in body weight (A), blood glucose (B), and pressure pain thresholds (C) before (day 0–8) and after (days 12–15) insulin replacement therapy. Samples for plasma insulin (D) were collected immediately after the last behavioral test session on day 15 after STZ treatment (after 7 days of insulin therapy). Values are means ± SD. In panels A, B, and D, the open, hatched, and filled columns represent control, STZnormoglycemic, and STZ-hyperglycemic groups of animals (n = 6, 4, or 6 animals, respectively, except for n = 4 for STZ-hyperglycemic group in panel D). Labels “b”, “c”, and “NG” indicate statistically significant differences from baseline (day 0) measurements, from within-group control means, or STZnormoglycemic rats, respectively (P < 0.05, t test). In panel C, open circles, filled circles, and filled triangles represent control, STZ-normoglycemic, and STZhyperglycemic rats (n = 6, 4, or 6), respectively. Mean pressure pain thresholds (PPT) of STZ-hyperglycemic and-normoglycemic rats are statistically significantly different from controls at each test day after STZ injection (P < 0.05, t test), except for post-STZ days 11 and 15 in STZ-normoglycemic group of rats. The horizontal bar above the data indicates the 7-day interval of insulin replacement treatment (0.2–0.4 U/rat/day). Biological variability was evident in some of the baseline values and weight gain. For example, fasting glucose level in this set of 16 rats, averaging 4.3 mM, was normal but somewhat higher (P < 0.05) than the mean of 3.1 mM in the first set of 24 rats (see Fig. 1). Furthermore, the control and normoglycemic rats in the second set of rats gained weight more slowly than animals studied in the first group (compare Fig. 1A and 4A), perhaps due to their slightly higher (20–25 g) entry weights and the slowing of weight gain with age as Sprague–Dawley rats approach and exceed about 300 g (Harlan Laboratories growth data). This difference however did not appear to influence either plasma glucose or pain thresholds values or their changes in different groups of STZ-treated rats. Finally, the mean baseline pressure threshold for the STZ-normoglycemic group was 4% lower (P < 0.05) than the HG group mean in this set of rats, whereas there were no inter-group differences for any variable in the set shown in Fig. 1.

Differential insulin dependence of pressure threshold and blood glucose levels Design of insulin replacement therapy for normoglycemic subjects that have abnormal sensory pain responses requires identification of ranges of insulin levels that are sufficient to maintain normal function with minimal risk of inadvertent hypoglycemia during insulin supplementation therapy. As an initial step toward this goal, the Hill equation was used to obtain best-fit curves describing insulin “dose–response” relationships for blood glucose and pressure threshold (solid curves, Fig. 5) based on all of our data from the untreated and insulin-treated sets of rats (see Fig. 5 and legend for explanation of symbols). This analysis

yields estimates for “half-effective” concentrations plasma insulin necessary to maintain normal glucose levels and pressure threshold of about 0.3 and 1.4 ng/ml, respectively (Fig. 5). Our results from fasted rats are in good agreement with a literature data set comprised of randomly sampled blood glucose and insulin concentrations (denoted by stars); the concentrations in random samples are, as expected, somewhat higher than the fasting levels (Fig. 5A). Within this context, the STZ-normoglycemic rats clearly represent an intermediate “insulinemic–normoglycemic state” (denoted as “NG”, Fig. 5), the borders of which (vertical dashed lines in Fig. 5) contain the abnormal pressure threshold responses. The lower limit of this range corresponds to the plasma insulin level (1 ng/ml) at which the maximal impairment of PPT is

390

D. Romanovsky et al. / Neurobiology of Disease 24 (2006) 384–394

Fig. 5. Model fitting to define the normoglycemic–insulinopenic state and establish goals for insulin replacement therapy. Pooled data from Figs. 3 and 4 are plotted along with mean values from the literature to characterize insulin dependence of blood glucose concentration (A) and of pressure pain threshold (B). Our data are from overnight-fasted rats and are denoted by filled (non-treated rats) and open (insulin-treated rats) symbols corresponding to control, STZnormoglycemic, and STZ-hyperglycemic rats (circles, squares, and triangles, respectively). Randomly sampled glucose and insulin values from the literature (open stars) are from the following references: Burcelin et al., 1992, Dachicourt et al., 1997, Gelling et al., 2004, Gerbi et al., 1998, Grey et al., 1970, Kobayashi and Kamata, 1999, Qiang et al., 1998, Rossetti and Giaccari, 1990, Rossetti et al., 1987 and Sechi et al., 1994. The solid smooth curves are results of the best-fit of Hill's equation to the data obtained in the present study; all parameters of the equation were allowed to vary during the fitting procedure. For “insulin–glucose” relationships in the upper panel, the half-effective insulin concentration (ED50) to maintain normal blood glucose levels calculated by the best-fit procedure is 0.27 ± 0.06 ng/ml, whereas the ED50 for “insulin–pressure pain threshold” relationships in the lower panel is 1.35 ± 0.06 ng/ml to maintain normal pain thresholds. The horizontal dotted line in the panel A represents the 11 mM glucose cutoff value which, when exceeded, is diagnostic of diabetes for randomly sampled glucose concentrations. Vertical dashed lines mark the approximate borders of the STZ-normoglycemic state defined here as impaired pressure pain threshold and normal blood glucose concentration. Horizontal arrows represent direction and magnitude of changes in mean plasma insulin levels in control and STZ-hyperglycemic (dashed arrows), and STZ-normoglycemic (solid arrows) groups of rats in response to 1 week of low-dose insulin replacement treatment in the present study.

observed in the normoglycemic subset of STZ rats, and the upper limit is the extrapolated value of about 1.75 ng/ml above which heat and pain thresholds are normal. Note that plasma insulin concentration within this range (1–1.75 ng/ml) is sufficient to maintain all of the reported glucose levels below the random glucose level of 11 mM used to define diabetic hyperglycemia (Fig. 5A, horizontal dashed line; American Diabetes Association,

2006). The striking finding is that low-dose insulin replacement therapy that raised insulin in STZ-normoglycemic rats above 1.75 ng/ml threshold level (from 1.3 ± 0.1 to 2.4 ± 0.5 ng/ml; the shifts in insulin level due to low-dose insulin injections are denoted by the horizontal arrows in Fig. 5) also normalized the pressure pain threshold in these animals. The same treatment had little effect on insulin level in STZ-hyperglycemic rats (increase

D. Romanovsky et al. / Neurobiology of Disease 24 (2006) 384–394

from 0.10 ± 0.01 to 0.15 ± 0.02 ng/ml) and decreased insulin in control rats from 3.6 ± 0.1 to 2.1 ± 0.2 ng/ml (dashed arrows in Fig. 5). In both cases however, no borders of “insulinemic” state was crossed, and no effect of treatment on either blood glucose level or pressure threshold was observed. These data provide further support to the idea that insulinopenia has independent “metabolic” and “neuropathic” effects, and the thresholds of these effects are different. Discussion The two major findings of the present study are that nociceptive pressure thresholds in STZ-normoglycemic rats correlate with the level of circulating insulin and low-dose insulin replacement therapy corrects both insulinopenia and the deficiency of pressure pain threshold without modulation of blood glucose level. Distal peripheral neuropathy is one of most frequent and troublesome complications of diabetes mellitus (Boulton et al., 2004), and our observations are relevant to three questions important for understanding the etiology of this disease. First, our data support the emerging concept that insulin is involved in normal functions of the peripheral nervous system; second, hyperglycemia is not the only trigger of distal peripheral neuropathy, and third, the onset of some symptoms of peripheral neuropathy may take place before diagnostic criteria of either diabetes or pre-diabetes are met. Diabetes is defined by fasting blood glucose levels exceeding 6.9 mM or random levels above 11 mM, whereas pre-diabetes is characterized by slightly elevated fasting glucose (5.6–6.9 mM) and/or impaired glucose tolerance (American Diabetes Association, 2006). Human distal peripheral neuropathy is a slowly progressing disease that frequently develops on a background of chronic hyperglycemia, and there is a wealth of evidence that hyperglycemia itself may be responsible for progression of many of symptoms of neuropathy (Sugimoto et al., 2000a). The hyperglycemic hypothesis of diabetic neuropathy is, however, not without challenges (reviewed in Ishii, 1995), one of which is frequent occurrence of tingling, lancinating or burning paresthesias, and mechanical allodynia (pain due to a stimulus that does not normally provoke pain) in patients with impaired glucose tolerance but not fasting hyperglycemia (Singleton et al., 2001). These observations suggest two important possibilities, first, abnormal pain thresholds appear to be “a uniquely sensitive indicator of early axonal injury” in distal peripheral neuropathy (Singleton et al., 2001), and second, pathogenic mechanisms of diabetic neuropathy other than hyperglycemia may exist. Clinical studies designed to evaluate early or alternative mechanisms of neuropathy are quite limited due to the uncertainties in establishing the duration of symptomatic and, especially, asymptomatic neuropathy in human subjects, so animal models studies are valuable to evaluate factors that contribute to early changes in pain thresholds. STZ rat models for stages of diabetes Overt insulinopenia and hyperglycemia in human Type I diabetes usually arise from an autoimmune attack that destroys about 90% of the pancreatic β-cells (Eisenbarth, 1986), and use of the selective toxin streptozotocin to kill β-cells has led to development of animal models for type I diabetes, including the well-characterized STZ-hyperglycemic rat (Animal Models of Diabetes, 2001). A key characteristic of diabetes is insulinopenia, but normal rat plasma insulin levels can vary considerably,

391

depending on the time since last food intake, species, and various experimental factors. Thus, there is a rather wide range of literature values for random samples from normal adult Sprague–Dawley and Wistar rats that fall between 1 and 6 ng/ml; the mean literature value, 2.6 ± 0.4 ng/ml, is similar to our control levels, 3.6 ± 0.2 ng/ml (see Romanovsky et al., 2004). Importantly, neither fasting (our work) nor random (literature) glucose levels rose above the cutoff diagnostic levels for diabetes over the entire range of all of these measurements unless insulin fell below 0.3 ng/ml (Fig. 5; STZhyperglycemic rats). This large (10-fold) excess of insulin in plasma above the minimal level for physiological control of blood glucose level explains the normoglycemic subset of STZ rats; STZnormoglycemic rats have four times the predicted threshold level of insulin for hyperglycemia to become manifest (1.28 ± 0.09 ng/ml vs. 0.3 ng/ml, Figs. 2 and 5). The STZ-normoglycemic rat model is not commonly studied and, therefore, is not as extensively characterized as the STZ-hyperglycemic model. Based on available data, the STZ-normoglycemic rat might correspond to early stage III of human type I diabetes, when insulin production is decreased but still adequate to counteract a glucose load challenge (Eisenbarth, 1986); these rats are currently not considered to be “pre-diabetic” on the basis of blood glucose levels and glucose tolerance measurements (Figs. 1, 4 and Romanovsky et al., 2004). Mechanical hyperalgesia In STZ-hyperglycemic rats, the rapid onset of hyperglycemia is quickly followed by mechanical hyperalgesia and tactile allodynia (Romanovsky et al., 2004). Abnormal nociception manifesting itself as decreases in both pressure (superficial and deep tissue pain), and the von Frey filament (filaments with graded bending force used to test for skin sensitivity to light touch and tactile allodynia) thresholds in STZ-normoglycemic rats has been consistently reported by a number of laboratories (Figs. 1D, 4C, and Calcutt et al., 1996, 2004; Dobretsov et al., 2003; Maneuf et al., 2004; Wuarin-Bierman et al., 1987). A fall in pressure pain threshold linked to aldose reductase activity can also be induced by local in vivo hyperglycemia or galactose feeding (Dobretsov et al., 2003; Wuarin-Bierman et al., 1987). However, in spite of these associations, changes in mechanical nociceptive thresholds cannot be simply ascribed to hyperglycemia in animals, as observed in humans. No correlation between tactile allodynia or mechanical hyperalgesia and the degree of systemic hyperglycemia was detected (Dobretsov et al., 2003; Maneuf et al., 2004) and treatment of STZ-hyperglycemic rats with aldose reductase inhibitors at a dosage that blocked accumulation of nerve sorbitol failed to prevent tactile allodynia (Calcutt et al., 1996, 2004). Taken together, these findings suggest the existence of additional pathogenic mechanisms that are independent of blood sugar level, even in the hyperglycemic rats. One possibility is an unrecognized neurotoxic effect of STZ, but the efficacy of insulin replacement to restore the normal pressure threshold in STZ-normoglycemic rats during the initial 2 weeks of treatment (Fig. 4C) and as late as after 20 months of diabetes in STZ-hyperglycemic rats (Calcutt et al., 1996; Courteix et al., 1996) renders this possibility unlikely. Therefore, to explain the apparently conflicting findings outlined above and our current and previous data obtained in STZnormoglycemic animals (Romanovsky et al., 2004 and Fig. 4), we propose an alternative hypothesis, that moderate insulinopenia and abnormal insulin signaling underlie mechanical allodynia and hyperalgesia.

392

D. Romanovsky et al. / Neurobiology of Disease 24 (2006) 384–394

Heat nociception Foot heat pain threshold is increased in about 50% of patients diagnosed with type I diabetes (Navarro et al., 1989), and thermal hypoalgesia affecting the rodent paw is evident in STZ-induced diabetes (Apfel et al., 1994; Calcutt et al., 1994). However, there are also a number of reports of lack of changes, decreases, or transient changes hind limb heat pain withdrawal threshold or latency in STZ-hyperglycemic rats (Calcutt et al., 2004; Fox et al., 1999; Khan et al., 2002; Li et al., 2005). These differences may arise from variations in methodology and test schedules in different laboratories, as well as from the possibility that more than one factor may contribute to pathogenesis. Hyperglycemia-dependent sorbitol formation is likely one of these putative factors (Calcutt et al., 2004) because heat pain latency in STZ-hyperglycemic rats can be normalized by treatment with aldose reductase inhibitors. In our studies, the statistically significant correlation between heat threshold and insulin in STZ-normoglycemic rats (Fig. 3C) suggests that insulinopenia may also contribute to thermal tolerance, but this possibility remains to be evaluated in future studies. Perspective and summary Neuropathic pain is a complex phenomenon that starts with enhanced activity of primary afferent neurons injured by a trauma or disease process and progresses to involve essentially all levels of spinal cord and central nervous system (Bonica, 1990). The mechanisms linking insulin level to pain sensitivity threshold are unknown, but insulin receptors are present at high densities at the paranodal loops of myelinating Schwann cells, endoneurial blood vessels, and small primary afferent neurons (Sugimoto et al., 2000b, 2002). This localization raises the possibility that insulin could act directly via its cognate receptors to regulate axon–glial relationships, endoneurial circulation, or aspects of function of small afferent nerve fibers, most of which are nociceptive. Isoforms of insulin receptors expressed in the peripheral nervous system and tissues (muscle, liver, fat) that are targets for the metabolic effects of insulin are not strongly different in their affinities to the hormone (Frasca et al., 1999), but our data suggest that the concentration of insulin required for normal nerve function is at least 5-fold higher than that needed for control of blood glucose level (Fig. 5). This difference in functional thresholds for the metabolic and neuropathic actions of insulinopenia might be explained by two factors, the notion of existence of spare insulin receptors (that is, not all receptors are needed to obtain the maximal biological response of the ligand) (Olefsky, 1977), and suggestion from our data that 5 times less receptor occupancy of receptors is required for the hormone to produce its maximal metabolic effect compared to its effect on peripheral nervous tissue (e.g., about 10 and 60% occupancy, respectively). Alternatively, insulin could control nerve function via low affinity binding to insulin-like growth factor receptors, or to hybrid insulin-like growth factor receptors. Our low-dose insulin therapeutic protocol was not effective in the severely insulinemic STZ-hyperglycemic rats, and it resulted in a paradoxical decrease in insulin level in normal control animals (Figs. 4 and 5). Further experiments are needed to confirm and explain these observations. However, high-dose insulin treatment restores the normal pressure threshold in STZ-hyperglycemic rats (Courteix et al., 1996). Therefore, it is likely that the failure of

moderate insulin therapy to correct insulin and pressure threshold in STZ-hyperglycemic rats was due to existence of a large pool of unbound insulin receptors in these animals. On the other hand, the decrease of circulating insulin concentration in insulin-treated control animals might result from compensatory feedback inhibition of the hormone release (see Xu and Rothenberg, 1998). The mechanism(s) that link pressure threshold and insulinopenia remain to be established, but the strength of correlation between pressure threshold and insulin in STZ-normoglycemic rats suggests a direct rather multi-component link between the two, and the success of insulin replacement therapy argues against a role for C-peptide in mechanical hyperalgesia. In insulin-resistant Zucker rats, a pressure threshold decrease was observed 4 weeks after spontaneous onset of type II diabetes (Zhuang et al., 1997), but a subsequent study in the same model reported no change in pressure threshold at 5 weeks after hyperglycemic onset (Piercy et al., 1999). If there is a direct role for insulin in the pathogenesis of pain, a pain pressure test may be a useful procedure for early detection of insulin signaling impairment, whether it is associated with insulinopenia or increased insulin resistance. Taken together, results of our and related animal studies suggest that pressure threshold measurements may serve as a relatively simple, low cost, non-invasive clinical tests for longitudinal assessment of early stages of diabetes to help select pre-diabetic individuals for further testing and treatment. The prevalence of increased pressure pain sensitivity of tender points does appear to correlate with the incidence of type I or type II diabetes (Tishler et al., 2003), but pressure threshold tests are not currently a component of the battery of quantitative sensory tests routinely used to diagnose distal peripheral neuropathy (Gracely et al., 2004); our studies support their inclusion. To summarize, evidence is progressively accumulating to reveal the inadequacy of hyperglycemic hypothesis to fully explain the wide variety of symptoms and presentations of diabetic neuropathy. The search for alternative triggers and pathogenic mechanisms of distal peripheral neuropathy (Ishii, 1995) includes functions of insulin-like growth factor and C peptide that are unrelated to glycemic control, as well as direct roles of abnormal insulin signaling in mitochondrial injury, impairment of motor and sensory nerve conduction velocity, and axonal atrophy of sensory and sympathetic neurons (Brussee et al., 2004; Huang et al., 2003; Qiang et al., 1998). Our studies add mechanical hyperalgesia to the list of peripheral neuropathic symptoms that may be directly linked to insulin action in rat and possibly in human, and they emphasize the potential usefulness of normoglycemic-STZ rats as a model to identify and evaluate the roles of factors that contribute to the initial stages of diabetic neuropathology. Acknowledgments This work was supported by NIH National Institute of Diabetes and Digestive and Kidney Diseases (DK067248 to MD) and National Institute of Neurological Disorders and Stroke (NS036728 to GD). References Animal Models of Diabetes, 2001. In: Sima, A.A., Shafrir, E. (Eds.), Harwood Academic Publishers, Amsterdam, pp. 1–364. American Diabetes Association, 2006. Standards of medical care in diabetes. Diabetes Care 29, S4–S42.

D. Romanovsky et al. / Neurobiology of Disease 24 (2006) 384–394 Apfel, S.C., Arezzo, J.C., Brownlee, M., Fedoroff, H., Kessler, J.A., 1994. Nerve growth factor administration protects against experimental sensory neuropathy. Brain Res. 634, 7–12. Biessels, G.J., Cristino, N.A., Rutten, G.J., Hamers, F.P., Erkelens, D.W., Gispen, W.H., 1999. Neurophysiological changes in the central and peripheral nervous system of streptozotocin-diabetic rats. Course of development and effects of insulin treatment. Brain 122 (Pt. 4), 757–768. Bonica, J.J., 1990. In: Loeser, J.D., Chapman, C.R., W.E. (Eds.), The Management of Pain. Lea and Febiger, Philadelphia, London, pp. 1–958. Boulton, A.J., Malik, R.A., Arezzo, J.C., Sosenko, J.M., 2004. Diabetic somatic neuropathies. Diabetes Care 27, 1458–1483. Brussee, V., Cunningham, F.A., Zochodne, D.W., 2004. Direct insulin signaling of neurons reverses diabetic neuropathy. Diabetes 53, 1824–1830. Burcelin, R., Eddouks, M., Kande, J., Assan, R., Girard, J., 1992. Evidence that GLUT-2 mRNA and protein concentrations are decreased by hyperinsulinaemia and increased by hyperglycaemia in liver of diabetic rats. Biochem. J. 288 (Pt. 2), 675–679. Calcutt, N.A., Malmberg, A.B., Yamamoto, T., Yaksh, T.L., 1994. Tolrestat treatment prevents modification of the formalin test model of prolonged pain in hyperglycemic rats. Pain 58, 413–420. Calcutt, N.A., Jorge, M.C., Yaksh, T.L., Chaplan, S.R., 1996. Tactile allodynia and formalin hyperalgesia in streptozotocin-diabetic rats: effects of insulin, aldose reductase inhibition and lidocaine. Pain 68, 293–299. Calcutt, N.A., Freshwater, J.D., Mizisin, A.P., 2004. Prevention of sensory disorders in diabetic Sprague–Dawley rats by aldose reductase inhibition or treatment with ciliary neurotrophic factor. Diabetologia 47, 718–724. Chan, A.W., MacFarlane, I.A., Bowsher, D., 1990. Short term fluctuations in blood glucose concentrations do not alter pain perception in diabeticpatients with and without painful peripheral neuropathy. Diabetes Res. 14, 15–19. Courteix, C., Bardin, M., Massol, J., Fialip, J., lavarenne, J., Eschalier, A., 1996. Daily insulin treatment relieves long-term hyperalgesia in streptozocin diabetic rats. NeuroReport 7, 1922–1924. Dachicourt, N., Serradas, P., Bailbe, D., Kergoat, M., Doare, L., Portha, B., 1997. Glucagon-like peptide-1(7–36)-amide confers glucose sensitivity to previously glucose-incompetent beta-cells in diabetic rats: in vivo and in vitro studies. J. Endocrinol. 155, 369–376. DCCT, 1993. The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulindependent diabetes mellitus. The Diabetes Control and Complications Trial Research Group. N. Engl. J. Med. 329, 977–986. Dienel, G.A., Cruz, N.F., Mori, K., Sokoloff, L., 1990. Acid lability of metabolites of 2-deoxyglucose in rat brain: implications for estimates of kinetic parameters of deoxyglucose phosphorylation and transport between blood and brain. J. Neurochem. 54, 1440–1448. Dobretsov, M., Hastings, S.L., Stimers, J.R., Zhang, J.-M., 2001. Mechanical hyperalgesia in rats with chronic perfusion of lumbar dorsal root ganglion with hyperglycemic solution. J. Neurosci. Methods 110, 9–15. Dobretsov, M., Hastings, S.L., Romanovsky, D., Stimers, J.R., Zhang, J.-M., 2003. Mechanical hyperalgesia in rat models of systemic and local hyperglycemia. Brain Res. 960, 174–183. Eisenbarth, G.S., 1986. Type I diabetes mellitus. A chronic autoimmune disease. N. Engl. J. Med. 314, 1360–1368. Fox, A., Eastwood, C., Gentry, C., Manning, D., Urban, L., 1999. Critical evaluation of the streptozotocin model of painful diabetic neuropathy in the rat. Pain 81, 307–316. Frasca, F., Pandini, G., Scalia, P., Sciacca, L., Mineo, R., Costantino, A., Goldfine, I.D., Belfiore, A., Vigneri, R., 1999. Insulin receptor isoform A, a newly recognized, high-affinity insulin-like growth factor II receptor in fetal and cancer cells. Mol. Cell. Biol. 19, 3278–3288. Gelling, R.W., Overduin, J., Morrison, C.D., Morton, G.J., Frayo, R.S., Cummings, D.E., Schwartz, M.W., 2004. Effect of uncontrolled diabetes

393

on plasma ghrelin concentrations and ghrelin-induced feeding. Endocrinology 145, 4575–4582. Gerbi, A., Maixent, J.-M., Jamme, I., Pierlovisi, M., Coste, T., Pieroni, G., Nouvelot, A., Vague, P., Raccah, D., 1998. Alterations of Na,K-ATPase isoenzymes in rat diabetic neuropathy: protective effect of dietary supplementation with n-3 fatty acids. J. Neurochem. 71, 732–740. Gracely, R.H., Eliav, E., Hansson, P., 2004. In: Dostrovsky, J.O., Carr, D.L., Koltzenburg, M. (Eds.), Quantitative Sensory Testing: Clinical Considerations and New Methods. IASP Press, Seattle, pp. 589–600. Grey, N.J., Goldring, S., Kipnis, D.M., 1970. The effect of fasting, diet, and actinomycin D on insulin secretion in the rat. J. Clin. Invest 49, 881–889. Huang, T.J., Price, S.A., Chilton, L., Calcutt, N.A., Tomlinson, D.R., Verkhratsky, A., Fernyhough, P., 2003. Insulin prevents depolarization of the mitochondrial inner membrane in sensory neurons of type 1 diabetic rats in the presence of sustained hyperglycemia. Diabetes 52, 2129–2136. Ishii, D.N., 1995. Implication of insulin-like growth factors in the pathogenesis of diabetic neuropathy. Brain Res. Brain Res. Rev. 20, 47–67. Khan, G.M., Chen, S.R., Pan, H.L., 2002. Role of primary afferent nerves in allodynia caused by diabetic neuropathy in rats. Neuroscience 114, 291–299. Kobayashi, T., Kamata, K., 1999. Effect of insulin treatment on smooth muscle contractility and endothelium-dependent relaxation in rat aortae from established STZ-induced diabetes. Br. J. Pharmacol. 127, 835–842. Li, F., Obrosova, I.G., Abatan, O., Tian, D., Larkin, D., Stuenkel, E.L., Stevens, M.J., 2005. Taurine replacement attenuates hyperalgesia and abnormal calcium signaling in sensory neurons of STZ-D rats. Am. J. Physiol. Endocrinol. Metab. 288, E29–E36. Maneuf, Y.P., Blake, R., Andrews, N.A., McKnight, A.T., 2004. Reduction by gabapentin of K+-evoked release of [3H]-glutamate from the caudal trigeminal nucleus of the streptozotocin-treated rat. Br. J. Pharmacol. 141, 574–579. Marquardt, D.W., 1963. An algorithm for least-squares estimation of nonlinear parameters. J. Soc. Ind. Appl. Math. 11, 431–441. Navarro, X., Kennedy, W.R., Fries, T.J., 1989. Small nerve fiber dysfunction in diabetic neuropathy. Muscle Nerve 12, 498–507. Olefsky, J.M., 1977. Insensitivity of large rat adipocytes to the antilipolytic effects of insulin. J. Lipid Res. 18, 459–464. Piercy, V., Banner, S.E., Bhattacharyya, A., Parsons, A.A., Sanger, G.J., Smith, S.A., Bingham, S., 1999. Thermal, but not mechanical, nociceptive behavior is altered in the Zucker Diabetic Fatty rat and is independent of glycemic status. J. Diabetes Its Complicat. 13, 163–169. Qiang, X., Satoh, J., Sagara, M., Fukuzawa, M., Masuda, T., Sakata, Y., Muto, G., Muto, Y., Takahashi, K., Toyota, T., 1998. Inhibitory effect of troglitazone on diabetic neuropathy in streptozotocin-induced diabetic rats. Diabetologia 41, 1321–1326. Romanovsky, D., Hastings, S.L., Stimers, J.R., Dobretsov, M., 2004. Relevance of hyperglycemia to early mechanical hyperalgesia in streptozotocin-induced diabetes. J. Peripher. Nerv. Syst. 9, 62–69. Rossetti, L., Giaccari, A., 1990. Relative contribution of glycogen synthesis and glycolysis to insulin-mediated glucose uptake. A dose–response euglycemic clamp study in normal and diabetic rats. J. Clin. Invest. 85, 1785–1792. Rossetti, L., Shulman, G.I., Zawalich, W., DeFronzo, R.A., 1987. Effect of chronic hyperglycemia on in vivo insulin secretion in partially pancreatectomized rats. J. Clin. Invest. 80, 1037–1044. Schmidt, R.E., Dorsey, D.A., Beaudet, L.N., Parvin, C.A., Zhang, W., Sima, A.A.F., 2004. Experimental rat models of types 1 and 2 diabetes differ in sympathetic neuroaxonal dystrophy. J. Neuropathol. Exp. Neurol. 63, 450–460. Sechi, L.A., Griffin, C.A., Schambelan, M., 1994. Effect of dietary sodium chloride on insulin receptor number and mRNA levels in rat kidney. Am. J. Physiol. 266, F31–F38. Sherman, W.R., Stewart, M.A., 1966. Identification of sorbitol in mammalian nerve. Biochem. Biophys. Res. Commun. 22, 492–497.

394

D. Romanovsky et al. / Neurobiology of Disease 24 (2006) 384–394

Singleton, J.R., Smith, A.G., Bromberg, M.B., 2001. Painful sensory polyneuropathy associated with impaired glucose tolerance. Muscle Nerve 24, 1225–1228. Sugimoto, K., Murakawa, Y., Sima, A.A.F., 2000a. Diabetic neuropathy—A continuing enigma. Diabetes/Metab. Res. Rev. 16, 408–433. Sugimoto, K., Murakawa, Y., Zhang, W., Xu, G., Sima, A.A.F., 2000b. Insulin receptor in rat peripheral nerve: its localization and alternatively spliced isoforms. Diabetes/Metab. Res. Rev. 16, 354–363. Sugimoto, K., Murakawa, Y., Sima, A.A.F., 2002. Expression and localization of insulin receptor in rat dorsal root ganglion and spinal cord. J. Peripher. Nerv. Syst. 7, 44–53. Suzuki, Y., Sato, J., Kawanishi, M., Mizumura, K., 2002. Tissue glucose level modulates the mechanical responses of cutaneous nociceptors in streptozotocin-diabetic rats but not normal rats in vitro. Pain 99, 475–484. Thye-Ronn, P., Sindrup, S.H., Arendt-Nielsen, L., Brennum, J., HotherNielsen, O., Beck-Nielsen, H., 1994. Effect of short-term hypergly-

cemia per se on nociceptive and non-nociceptive thresholds. Pain 56, 43–49. Tishler, M., Smorodin, T., Vazina-Amit, M., Ramot, Y., Koffler, M., Fishel, B., 2003. Fibromyalgia in diabetes mellitus. Rheumatol. Int. 23, 171–173. Wuarin-Bierman, L., Zahnd, G.R., Kaufmann, F., Burcklen, L., Adler, J., 1987. Hyperalgesia in spontaneous and experimental animal models of diabetic neuropathy. Diabetologia 30, 653–658. Xu, G.G., Rothenberg, P.L., 1998. Insulin receptor signaling in the beta-cell influences insulin gene expression and insulin content: evidence for autocrine beta-cell regulation. Diabetes 47, 1243–1252. Zhuang, H.-X., Wuarin, L., Fei, Z.-J., Ishii, D.N., 1997. Insulin-like growth factor (IGF) gene expression is reduced in neural tissues and liver from rats with non-insulin-dependent diabetes mellitus, and IGF treatment ameliorates diabetic neuropathy. J. Pharmacol. Exp. Ther. 283, 366–374.