Dosing profile profoundly influences nicotinic acid's ability to improve metabolic control in rats.

Acute nicotinic acid (NiAc) administration results in rapid reduction of plasma FFA concentrations. However, sustained NiAc exposure is associated with tolerance development resulting in return of FFA to pretreatment levels. The aim of this study was to determine whether a 12 h rectangular exposure profile (intermittent dose group) could avoid tolerance development and thereby reverse insulin resistance induced by lipid overload. FFA lowering was assessed in male Sprague Dawley (lean) and obese Zucker rats (obese) in response to a 5 h NiAc infusion, in either NiAc-naïve animals or after 5 days of continuous (24 h/day) or intermittent (12 h/day) NiAc dosing (via implantable, programmable minipump). We found that intermittent dosing over 5 days preserved NiAc-induced FFA lowering, comparable to dosing in NiAc-naïve animals. By contrast, following 5 days continuous administration, NiAc-induced FFA lowering was lost. The effect of intermittent NiAc infusion on insulin sensitivity was assessed in obese Zucker rats using hyperinsulinemic-isoglycemic clamps. The acute effect of NiAc to elevate glucose infusion rate (vs. saline control) was indeed preserved with intermittent dosing, while being lost upon continuous infusion. In conclusion, an intermittent but not continuous NiAc dosing strategy succeeded in retaining NiAc's ability to lower FFA and improve insulin sensitivity in obese Zucker rats.-Kroon, T., A. Kjellstedt, P. Thalén, J. Gabrielsson, and N. D. Oakes.

The GPR109A agonist nicotinic acid (NiAc) has been used clinically ever since its antidyslipidemic effects (HDL elevation and reductions of total cholesterol, LDL-cholesterol, and TG) were discovered more than 50 years ago (11)(12)(13)(14)(15). Although NiAc potently lowers FFA acutely, large-scale clinical studies, with repeated oral NiAc administration, often report increased levels of fasting glycemia (16)(17)(18)(19). NiAc has not been optimized to achieve durable and therapeutically meaningful FFA lowering. By this, we specifi cally mean reducing around-the-clock FFA area under the curve (AUC). In theory, this might be achieved by sustained NiAc exposure; however, the FFA-lowering effect seen initially appears to be lost over time despite maintained NiAc exposure (tolerance development) ( 20 ). Time-dependent loss of both FFA lowering and glucose control improvement also occur in patients with type 2 diabetes, treated with the NiAc analog acipimox ( 21,22 ). To avoid tolerance development, drug holidays are needed. However, at the end of each NiAc exposure period, there is the risk of FFA rebound (here referring to the situation where FFA overshoots pretreatment levels in connection with NiAc decline) due to the short NiAc plasma half-life ( 23 ). FFA rebound is associated with impaired glucose control ( 24,25 ). The question of whether there might be an optimal balance between periods of continuous exposure (which would minimize rebound) and drug holidays (which would minimize tolerance) in order to achieve maximal FFA lowering has not been addressed.
Abstract Acute nicotinic acid (NiAc) administration results in rapid reduction of plasma FFA concentrations. However, sustained NiAc exposure is associated with tolerance development resulting in return of FFA to pretreatment levels. The aim of this study was to determine whether a 12 h rectangular exposure profi le (intermittent dose group) could avoid tolerance development and thereby reverse insulin resistance induced by lipid overload. FFA lowering was assessed in male Sprague Dawley (lean) and obese Zucker rats (obese) in response to a 5 h NiAc infusion, in either NiAcnaïve animals or after 5 days of continuous (24 h/day) or intermittent (12 h/day) NiAc dosing (via implantable, programmable minipump). We found that intermittent dosing over 5 days preserved NiAc-induced FFA lowering, comparable to dosing in NiAc-naïve animals. By contrast, following 5 days continuous administration, NiAc-induced FFA lowering was lost. The effect of intermittent NiAc infusion on insulin sensitivity was assessed in obese Zucker rats using hyperinsulinemic-isoglycemic clamps. The acute effect of NiAc to elevate glucose infusion rate (vs. saline control) was indeed preserved with intermittent dosing, while being lost upon continuous infusion. In conclusion, an intermittent but not continuous NiAc dosing strategy succeeded in retaining NiAc's ability to lower FFA and improve insulin sensitivity in obese Zucker rats.-Kroon, T., A. Kjellstedt, P. Thalén, J. Gabrielsson, and N. D. Oakes. Dosing profi le profoundly infl uences nicotinic acid's ability to improve metabolic control in rats. J. Lipid Res. 2015. 56: 1679-1690.
Supplementary key words lipolysis and fatty acid metabolism • diabetes • insulin • adipose tissue • drug therapy/hypolipidemic drugs • niacin • GPR109A • tachyphylaxis Lipid overload in nonadipose tissues has been linked to the pathogenesis of insulin resistance and atherogenesis (1)(2)(3)(4). A potential means for reversing peripheral lipid overload is to restrict the release of FFAs from adipose tissues. A number of independent mechanisms have been dosing solutions were ‫ف‬ 1 M. Vehicle, for control animals, consisted of sodium chloride solutions at equimolar concentrations. Freshly prepared formulations were loaded into the infusion pump (see below), via a 0.2 µm sterile fi lter (Acrodisc ® ; Pall Corporation, Ann Arbor, MI) just before pump implantation.

Surgical preparation
In order to acclimate to individual housing, animals were moved to separate cages 3 days prior to surgery. To prevent potential infections in conjunction with surgery, oral antibiotic treatment was initiated 1 day prior to surgery and then once daily for 3 days (sulfamethoxazole and trimethoprim 40  ; RB Pharmaceuticals Ltd., Berkshire, United Kingdom). Animals were then housed individually and allowed 3 days of recovery before start of the preprogrammed pump infusion. Throughout the study, body weight and general health status was monitored and recorded daily.

Treatment
Both lean and obese animals were divided into three dose groups, and NiAc was given acutely (NiAc naïve) or following 5 days with either continuous (Cont. NiAc) or intermittent (Inter. NiAc) administration. Each dose group was matched with corresponding saline-infused controls. Infusions were given subcutaneously at 0. 17  . The intermittent infusion protocol was programmed as a 12 h on-off cycle (infusion on at 13:00 h). During the last day of the treatment period, an overnight fast was initiated (food removed at 24:00 h with water freely available), and animals entered into one out of three terminal acute experiments: study I, study II, or study III. An overview of study protocols is summarized schematically in Fig. 1 .

Study I (NiAc-induced FFA lowering)
In the morning of the acute experimental day, the jugular catheter was connected to a swivel system to enable blood sampling in unrestrained animals. Jugular catheter patency was maintained by continuous infusion (5 µl min Ϫ 1 ) of Na-citrate solution (20.6 mM). After a 3-4 h adaptation period, at ‫ف‬ 12:00 h, the basal phase of the acute experiment (basal period) commenced with two to three blood samples drawn between Ϫ 60 and Ϫ 5 min relative to commencement of NiAc/saline infusion (note that, in the Cont. NiAc groups, infusion pumps were on throughout this sampling period; Fig. 1 ). Samples were then drawn during the infusion phase (infusion period) at 30, 60, and 90 min, as well as at 2.0, 3.0, 4.0, and 4.5 h with infusion stop at 5.0 h. During the postinfusion phase (postinfusion period), blood samples were drawn at 5.2, 5.3, 5.5, 6.0, 7.0, and 8.0 h. Blood samples were In this study, effects of daily NiAc dosing profi le on NiAc's acute ability to lower FFA levels and improve insulin sensitivity were examined. More specifi cally, NiAc-induced FFA lowering following pretreatment with two well-defi ned NiAc dosing regimens was compared with the response in NiAc-naïve animals. The pretreatment regimens were a 12 h rectangular exposure profi le (intermittent dose group) and a sustained exposure profi le (continuous dose group). These profi les were produced using implantable, programmable minipumps in lean healthy and obese insulin-resistant Zucker rats. Insulin sensitivity in obese Zucker rats was assessed using hyperinsulinemic-isoglycemic clamps. Additionally, we examined the impact of the alternative dosing regimens on expression of selected adipose tissue genes involved in FFA mobilization.

Animals
Experimental procedures were approved by the local Ethics Committee for Animal Experimentation (Gothenburg region, Sweden). Male Sprague Dawley (lean) and obese Zucker rats (fa/fa, obese) were purchased from Harlan Laboratories B.V. (The Netherlands). Animals were housed in an Association for Assessment and Accreditation of Laboratory Animal Care accredited animal facility with climate-control: room temperature 20°C-22°C, relative humidity 40-60%, and with a 12 h light-dark cycle (lights on at 06:00 h). The animals were housed in groups of fi ve and given free access to standard rodent chow (R70; Laktamin AB, Stockholm, Sweden) and regular tap water. Perioperative body weight was well matched within each group (lean, 471 ± 4 g; obese, 649 ± 7 g) and stable prior to the acute study, with insignifi cant changes following surgery and 5 days of saline/NiAc pretreatment (lean, increase of 3.7 ± 0.5%; obese, increase of 0.1 ± 0.2%).

Studies overview
We wanted to compare the ability of continuous versus intermittent NiAc administration to suppress FFA levels in metabolically healthy and insulin-resistant rats. To this end, study I was conducted in conscious lean and obese animals to compare acute FFA lowering with the situation following 5 days of either continuous or intermittent NiAc administration. In obese only, an additional group was studied following 11 days of intermittent NiAc dosing. In study II, the effects of acute, continuous, and intermittent NiAc infusion on whole body insulin sensitivity were assessed in obese animals under hyperinsulinemic-isoglycemic clamp conditions. Study III was performed to examine the impact of the alternative dosing regimens on expression of selected adipose tissue genes involved in FFA mobilization in lean and obese groups using quantitative RT-PCR (Taqman).

NiAc dose selection and formulation
A key aspect of the study design was to achieve steady-state plasma NiAc concentrations corresponding to therapeutically relevant levels in the rat ( ‫ف‬ 1 µM). The dose selection was based on the previously obtained relationship between FFA lowering and NiAc infusion rate in NiAc-naïve lean and obese rats ( 26 ). NiAc (pyridine-3-carboxylic acid; Sigma-Aldrich, St. Louis, MO) was dissolved in sterile water and adjusted to physiological pH using sodium hydroxide. The fi nal concentrations of the NiAc was determined for each animal to be equal to its own basal level (isoglycemia). This was obtained from the average of at least three stable consecutive samples during the preclamp period. Plasma glucose was clamped with a variable rate infusion of 20% (w/v) glucose using a syringe pump (Model 22 I/W; Harvard Apparatus Inc., South Natick, MA) via a dedicated jugular vein catheter. During the fi rst hour of the clamp, arterial plasma glucose was measured every 5 min, and during the second h every 10 min, using a glucose analyzer (ACCU-CHEK ® Compact Plus; Roche Diagnostics, Indianapolis, IN; <10 µl blood per sample). Steady state, in both plasma glucose level (within ± 10% of the target level) and glucose infusion rate (GIR), was generally achieved within 90 min of clamp start. Additional blood samples (100 µl) were collected into potassium-EDTA-coated tubes during the basal period (at Ϫ 90 and Ϫ 75 min relative to insulin infusion start), the preclamp period (at Ϫ 30, Ϫ 15 min), and the clamp (60 and 120 min). Blood samples were centrifuged immediately and stored at Ϫ 20°C pending analysis of plasma FFA and insulin concentrations.

Study III (NiAc-induced changes in adipose tissue gene expression)
Following 5 days of continuous or intermittent saline/NiAc treatment ( Fig. 1A ) and 5 h of continuous infusion of NiAc/ saline ( Fig. 1B ), animals were anesthetized with isofl urane, and tissues (liver and epididymal adipose tissue) were dissected, snap frozen in liquid nitrogen, and stored at -80°C pending analysis.

Analytical methods
Plasma FFA was analyzed using an enzymatic colorimetric method (Wako Chemicals GmbH, Neuss, Germany). Plasma glucose was measured using a portable blood glucose monitoring device (ACCU-CHEK ® Compact Plus; Roche Diagnostics). Obese plasma insulin was analyzed with an RIA kit (rat insulin RIA kit; Millipore Corporation, St. Charles, MO), while lean plasma insulin concentrations were determined using a colorimetric ELISA kit (Ultra Sensitive Rat Insulin ELISA Kit; Crystal Chem Inc., Downers Grove, IL). The ELISA was used for lean rats to minimize blood sample volume (only 5 µl plasma required vs. ‫ف‬ 50 µl plasma for RIA). The RIA was used for the obese rats because high lipid levels in the plasma of these animals interfere with the ELISA but not the RIA measurement. Because of the hyperinsulinemia in the obese animals only 5 µl of plasma was required. For lean rat plasma (with low lipid levels) the absolute insulin measurement are equivalent for the RIA and ELISA assays based on an in-house comparison. For plasma samples collected during the glucose clamp study (study II), total (rat + human) insulin was determined using the rat RIA, and human insulin was determined by a species-specifi c RIA (human insulin-specifi c RIA kit; Millipore). Plasma NiAc concentrations were analyzed using LC/ MS/MS with a hydrophilic interaction liquid chromatography approach, separated on a 50 × 2.1 mm Biobasic AX column, with 5 m particles (Thermo Hypersil-Keystone, Runcorn, Cheshire, United Kingdom). TG content of liver was measured using an enzymatic colorimetric method (Horiba ABX, France). Area under the concentration-time curves for FFA, insulin, and glucose were calculated by trapezoidal approximation, using GraphPad Prism 6.01 (GraphPad Software Inc., La Jolla, CA). Homeostasis model assessment for insulin resistance (HOMA-IR) was calculated for study II based on the product of basal period plasma insulin and glucose.
RNA was extracted and isolated according to the manufacturer's instructions using RNeasy ® Mini Kit (Qiagen AB, Solna, Sweden). RNA concentration was measured and purity assessed using a Nanodrop (ThermoFisher, Wilmington, DE). cDNA was reverse transcribed from up to 2.5 g RNA using a High Capacity used to determine plasma concentrations of NiAc, FFA, glucose, and insulin. Blood sample volume ranged between 30 and 150 µl with a combined loss of <5% of the total blood volume. Blood was collected in potassium-EDTA coated tubes and briefl y kept on ice until centrifugation and storage at Ϫ 20°C pending analysis.

Study II (hyperinsulinemic-isoglycemic clamps)
Whole body insulin sensitivity was assessed in anesthetized obese animals using hyperinsulinemic-isoglycemic clamps. Animals were anesthetized at ‫ف‬ 08:00 h (Na-thiobutabarbitol, Inactin ® , 180 mg kg Ϫ 1 , ip; RBI, Natick, MA), tracheotomized with PE 240 tubing, and breathed spontaneously. One catheter (PE 50 tubing) was placed in the left carotid artery for blood sampling, as well as recording of arterial blood pressure and heart rate. Four catheters (PE 10 tubing) were placed together in the right external jugular vein for infusions of NiAc, insulin, and glucose and for administering top-up doses of anesthetic, if needed. The catheters were fi lled with Na-citrate solution (20.6 mM) in normal saline to prevent clotting. The arterial catheter patency was maintained by continuous infusion of Na-citrate (20.6 mM in saline, 5 µl min Ϫ 1 ) from shortly after carotid catheterization until the conclusion of the experiment. Body temperature was monitored using a rectal thermocouple and maintained at 37.5°C by means of servo controlled external heating.
Animals were allowed a stabilization period of ‫ف‬ 150 min from surgical completion. Following this, a 30 min basal period (with no NiAc/saline infusions in any group except Cont. NiAc) preceded commencement of intravenous-infused NiAc/saline, performed by external syringe pumps (CMA 1100; Carnegie Medicin, Solna, Sweden). An initial 60 min preclamp period preceded the start of insulin infusion for the clamps. Human insulin (Actrapid ® ; Novo Nordisk, Bagsvaerd, Denmark) was infused at a constant rate based on estimated lean body mass [lbm; ( 27 ) at 60 pmol kg lbm Ϫ 1 min Ϫ 1 ] via one dedicated jugular vein catheter using a syringe pump. The target plasma glucose level for the clamp In the lean NiAc-naïve group, during the postinfusion period, there was clear evidence of a marked rebound ( Fig. 3 ) with FFA AUC greater than saline control for the same period ( P < 0.05; Fig. 4B ). Interestingly, the Cont. NiAc group also exhibited a rebound ( Fig. 3 ) with AUC greater than saline control ( P < 0.001; Fig. 4B ), consistent with a previous study ( 20 ).
Remarkably, in the lean animals over the whole 8 h observation period, none of the NiAc protocols reduced total FFA AUC ( Fig. 4C ). Thus, in the NiAc-naïve and Inter. NiAc groups, the FFA rebound during the postinfusion period tended to cancel the FFA-lowering achieved during the infusion period. In the Cont. NiAc group, the total FFA AUC was actually elevated compared with both NiAcnaïve and Inter. NiAc groups ( P < 0.05; Fig. 4C ).
OBESE RATS. As expected, obese exhibited higher average FFA levels than lean during the basal period (increase of 86% vs. lean saline, P < 0.001; Fig. 3 ). Previous exposure to NiAc had no effect on basal period FFA ( P > 0.05, vs. saline control). As in lean, NiAc-induced FFA lowering was completely lost in the Cont. NiAc group after 5 days of continuous, uninterrupted NiAc infusion. Thus, during the infusion period the FFA AUC was similar in Cont. NiAc versus saline control ( P > 0.05; Fig. 4A ). Intermittent NiAc dosing succeeded in retaining significant FFA lowering during the infusion period, with FFA AUC in the Inter. NiAc group less than saline control ( P < 0.001). Unlike lean though, there was some loss of the extent of FFA AUC lowering (Inter. NiAc vs. NiAc naïve, P < 0.001; Fig. 4A ). Importantly, an additional group of animals studied following 11 days of intermittent NiAc (Inter. NiAc Day 11) showed that there was no further development of tolerance (Inter. NiAc Day 11 vs. Inter. NiAc, P > 0.05; Fig. 4A ). Total 8 h FFA AUC lowering was achieved only in the NiAc-naïve group (decrease of 36%, P < 0.001 vs. saline control; Fig. 4C ) while being unaffected in both intermittent NiAc groups and actually increased in the Cont. NiAc group compared with saline control ( P < 0.05; Fig. 4C ).
Insulin and glucose. Plasma insulin and glucose concentration-time profi les are shown in Fig. 5 . Comparisons between groups are based on analysis of AUC for insulin ( Fig. 6 ) and glucose ( Fig. 7 ). To satisfy homogeneity of variance assumption, insulin AUC data were expressed as percent of respective saline control values. An important objective was to assess the potential of NiAcinduced FFA lowering to improve glucose control. Indeed, cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA). Differences in gene expression (relative to 36B4) were determined by quantitative RT-PCR (Taqman) using a 7900HT system with SDS2.3 software ( Table 1 ).

Statistics
Statistical signifi cance of post hoc comparisons were evaluated based on one-way ANOVA with Sidak's multiple comparisons test, performed using GraphPad Prism 6.01 (GraphPad Software Inc.). For study I, statistical comparisons were made on area under concentration-time curves estimates, for FFA, insulin, and glucose. Throughout, results are reported as mean ± SE. P < 0.05 was considered statistically signifi cant.

Study I: metabolic effects of continuous versus intermittent NiAc dosing in lean and obese rats
NiAc exposure. The target steady-state plasma NiAc concentration of ‫ف‬ 1 µM was successfully achieved by subcutaneously infusing 0.17 µmol·min Ϫ 1 ·kg Ϫ 1 in all lean and obese groups ( Fig. 2 ). This exposure level was selected based on previous data showing near maximal FFA suppression in NiAc-naïve Sprague Dawley rats ( 26 ). Plasma NiAc concentrations declined rapidly across all dosed groups during the postinfusion period. In control (salineinfused) animals, endogenous NiAc levels were below the detection limit (6 nM).
FFAs. Plasma FFA concentration-time profi les are shown in Fig. 3 . Comparisons between groups are based on analysis of area under the concentration-time curves for the 5 h infusion period, 3 h postinfusion period, and the combined 8 h observation period ( Fig. 4 ).
LEAN RATS. Previous exposure to NiAc, either intermittent or continuous, had no effect on average basal period FFA ( P > 0.05, vs. saline control). During the infusion period, FFA AUC was lower in the NiAc-naïve group compared with saline control ( P < 0.05; Fig. 4A ). In the Cont. NiAc group, FFA AUC was not reduced compared with saline controls ( P > 0.05; Fig. 4A ) despite ongoing NiAc infusion/exposure ( Fig. 2 ) indicating complete tolerance after 5 days of NiAc exposure. In contrast, the FFA-lowering response was fully preserved following intermittent infusion for 5 days compared with NiAc-naïve animals, with similar AUC values in the two groups ( P > 0.05; Fig. 4A ). Thus, intermittent dosing was successful in avoiding FFAlowering tolerance.

Study II: effects of continuous versus intermittent NiAc dosing on whole body insulin sensitivity in obese Zucker rats
Whole body insulin sensitivity was assessed in anesthetized obese animals using hyperinsulinemic-isoglycemic clamps. During the basal period, FFA, glucose and insulin levels were similar across all groups ( Table 2 ), confi rming complete tolerance development in the Cont. NiAc group. NiAc infusion lowered FFA and insulin in both the NiAcnaïve and Inter. NiAc groups ( P < 0.05; Table 2 ) while these variables remained stable in the saline group. In saline-infused animals, insulin infusion increased total plasma insulin by ‫ف‬ 80%. Despite this, plasma FFA levels remained stationary, that is, a complete loss of antilipolytic action of insulin and supporting the phenotype of this insulin-resistant animal model ( 28 ). Importantly, acute NiAc exposure restored insulin's ability to suppress plasma FFA levels in both NiAc-naïve and Inter. NiAc groups ( P < 0.001 vs. saline control; Table 2 ). The extent of FFA suppression was similar in NiAc-naïve and Inter. NiAc groups ( P > 0.05).
Clamps were performed at similar levels of glycemia ( Table 2 ). Steady state GIRs needed to maintain basal glucose levels during the hyperinsulinemic-isoglycemic clamp are summarized in Fig. 8 . In the NiAc-naïve group, GIR was markedly increased compared with saline-infused controls (increase of 92%, P < 0.01). Importantly, the Inter. NiAc group also had an elevated GIR (increase of 71%, compared with their respective saline-infused controls, we observed a reduction in insulin AUC during the infusion period ( Fig. 6A ) in lean NiAc naïve ( P < 0.01), obese NiAc naïve ( P < 0.05), and obese Inter. NiAc ( P < 0.05). This occurred in the absence of change in infusion period glucose AUC ( P > 0.05 for all NiAc-infused groups vs. respective saline controls; Fig. 7A ), suggesting an improvement in insulin sensitivity, which was particularly remarkable in the obese groups. Although insulin profi les in the different groups were broadly similar in pattern to the FFA profi les described previously, one difference was in the obese animals in which the lowering of insulin AUC was fully preserved at 5 days in Inter. NiAc versus NiAc-naïve groups ( P > 0.05; Fig. 6A ), compared with FFA lowering, which was partially lost at this time point ( Fig. 4A ). The rebound phenomenon was not just restricted to FFA. Insulin rebounds were observed in all NiAc-dosed groups ( Fig. 5A, B ) with the exception of obese Cont. NiAc ( Fig. 5B ). As for FFA (described previously), reduction in insulin AUC achieved during the infusion period tended to be cancelled during the postinfusion period, with the result that total 8 h insulin AUC is similar in all NiAc groups compared with respective saline control groups ( Fig. 6C ). NiAc succeeded in moderately lowering blood glucose AUC in the obese NiAc-naïve group (decrease of 11%, P < 0.001; Fig. 7C ), although this effect was not maintained with either intermittent or continuous NiAc dosing. given acutely (NiAc naïve, n = 7/group) or following 5 days continuous (Cont. NiAc, lean n = 4, obese n = 8) or intermittent (Inter. NiAc, lean n = 4, obese n = 9) or 11 days intermittent (Inter. NiAc Day 11, obese n = 4) dosing. The black horizontal bar represents the period of acute NiAc/saline infusion. Data presented as mean ± SE. ) given acutely (NiAc naïve, n = 7/group) or following 5 days continuous (Cont. NiAc, lean n = 4, obese n = 8), intermittent (Inter. NiAc, lean n = 4, obese n = 9), or 11 days intermittent (Inter. NiAc Day 11, obese n = 4) dosing. The black horizontal bar represents the period of acute NiAc/saline infusion. Data presented as mean ± SE. levels to be lower in obese saline versus lean saline, with signifi cantly lower levels for PDE-3B ( P < 0.05) and GPR109A ( P < 0.05). In lean rats, continuous NiAc infusion induced an increase in expression of genes promoting both FFA mobilization (ATGL, P < 0.05) and FFA storage (GPR81, P < 0.05) compared with saline controls. By contrast, intermittent NiAc administration had limited impact on gene expression compared with saline control. Unlike the lean situation, continuous NiAc infusion in obese rats had no signifi cant impact on gene expression. Neither was there any apparent effect of intermittent dosing on the genes measured. Overall there was no evidence that a coordinated alteration in expression of genes was responsible for the tolerance development in either lean or obese ( Table 3 ).

Liver TGs
Liver TG content, at the end of the infusion period, is presented in Table 4 . As expected, saline-infused obese displayed elevated liver TG content ( P < 0.001; Table 4 ) P < 0.05), similar in magnitude to the NiAc-naïve group ( P > 0.05), compatible with a sustained insulin sensitization of the intermittent dosing approach. In stark contrast, upon continuous dosing, this effect was completely lost ( P > 0.05 vs. saline control; Fig. 8 ). Across all groups, GIR was negatively correlated with prevailing clamp FFA levels (individual data linear regression, r 2 = 0.35, P < 0.01; data not shown).

Study III: NiAc-induced changes in adipose tissue gene (mRNA) expression
Quantitative RT-PCR (Taqman) was conducted to assess whether changes in mRNA expression of adipocyte proteins could explain the differences observed in the NiAcinduced FFA lowering in study I ( Table 3 ). Degradation of TG to FFA predominantly involves ATGL and HSL. The activity of these lipases is governed by the intracellular concentration of cAMP. Production and degradation of cAMP is controlled via adenylate cyclase and PDE-3B, respectively. There was a general tendency for expression Fig. 4. Plasma FFA AUCs during the infusion period (A), postinfusion period (B), and the sum of these two periods (C) in lean (open bars) and obese (black bars). * P < 0.05, *** P < 0.001 versus lean saline; † P < 0.05, † † P < 0.01 versus lean NiAc naïve; ‡ P < 0.05 versus lean Cont. NiAc; + P < 0.05, ++ P < 0.01, +++ P < 0.001 versus obese saline; ¤ P < 0.05, ¤¤ P < 0.01, ¤¤¤ P < 0.001 versus obese NiAc naïve; ¥¥ P < 0.01, ¥¥¥ P < 0.001 versus obese Cont. NiAc. Data presented as mean ± SE . ) given acutely (NiAc naïve, n = 7/group) or following 5 days continuous (Cont. NiAc, lean n = 4, obese n = 8) or intermittent (Inter. NiAc, lean n = 4, obese n = 9) or 11 days intermittent (Inter. NiAc Day 11, obese n = 4) dosing. The black horizontal bar represents the period of acute NiAc/saline infusion. Data presented as mean ± SE. importance of therapeutically relevant exposures ( 29 ), a key aspect of the current studies was that they were performed at plateau plasma NiAc concentrations of ‫ف‬ 1 µM, which in NiAc-naïve animals are just suffi cient to maximally suppress FFA levels ( 26 ). This was done because loss of FFA lowering might theoretically be exacerbated by sustained supramaximally effective levels of target engagement [e.g., by ligand-induced GPR109 desensitization and internalization ( 30 )]. Oh et al. ( 20 ) previously demonstrated, in Wistar rats, complete return of FFA to pretreatment levels during a ‫ف‬ 10-fold higher NiAc continuous infusion rate than we have used in the current study. Based on a pharmacokinetic analysis (data not shown), and the fact that NiAc clearance exhibits saturation kinetics ( 26 ), we estimate that this infusion rate would result in plateau concentrations >20-fold above those achieved in the current study.
Markedly improved insulin sensitivity was seen in association with NiAc-induced FFA lowering, either in NiAcnaïve or previously intermittently dosed obese Zucker rats. Adult male obese Zucker rats exhibit extreme whole body insulin resistance associated with tissue lipid overload. Elevated FFA mobilization from adipose tissue, seen under both basal fasting as well as hyperinsulinemic clamp conditions, is an important mechanism driving this condition compared with lean saline. In both lean and obese, NiAc (either intermittent or continuous exposure) had no signifi cant impact on liver TG content.

DISCUSSION
The intermittent drug holiday approach succeeded in retaining the ability of a therapeutically relevant NiAc exposure to induce FFA lowering. In lean rats, following intermittent (12 h on/12 h off) infusions for 5 days, the acute NiAc-induced FFA suppression was completely preserved ( Figs. 3, 4 ). While in the obese rats there was a partial loss of FFA-lowering effi cacy, following 5 days intermittent dosing, importantly, this did not appear to be progressive with similar FFA lowering at 11 versus 5 days ( Figs. 3, 4 ). By contrast, continuous NiAc infusion for 5 days resulted in complete return of FFA to pretreatment levels, consistent with the fi ndings of Oh et al. ( 20 ).
NiAc has a very short half-life in the rat ( ‫ف‬ 2 min in plasma), which precludes oral dosing as a means of achieving stable and well-defi ned exposures. Therefore, NiAc was administrated using an implantable, programmable pump infusing via the subcutaneous route. Given the   Measurements were made either in the 30 min period preceding NiAc/saline infusion (basal period) and during the NiAc/saline infusion period, either prior to clamp (preclamp) or during clamp (clamp). a In the basal and preclamp periods, total insulin = endogenous (rat) insulin; in the clamp phase, total insulin = endogenous + human insulin achieved by use of two RIAs (see Materials and Methods). Data presented as mean ± SE (n = 6/group). b P < 0.05 versus saline.  Fig. 8. GIR at clamp steady state in obese. + P < 0.05, ++ P < 0.01 versus saline; ¥¥ P < 0.01 versus Cont. NiAc. Data presented as mean ± SE ( n = 6/group). ( 28,31 ). In the present study, acute suppression of circulating FFA levels by NiAc in the obese animals was associated with reduced fasting hyperinsulinemia ( Fig. 6A ). Reduced insulin secretion might be explained by direct effects on the islets of either NiAc ( 32 ) or the fall in FFA level ( 33 ). However, if the effect was only to decrease insulin secretion, then levels of glycemia should have increased, which was not the case ( Fig. 7A ). Reduced fasting insulinemia in association with normoglycemia suggests instead that NiAc enhanced whole body insulin sensitivity. Indeed, this was confi rmed by the elevated GIRs needed to maintain isoglycemia during the hyperinsulinemic clamps ( Fig. 8 and Table 2 ). Rapid effects of acute modulation of FFA metabolism on insulin sensitivity have been previously reported. The acute blockade of ␤ -oxidation, using etomoxir, increased insulin sensitivity in skeletal muscle and reduced gluconeogenesis ( 34,35 ), which could enhance insulin-mediated suppression of hepatic glucose output. More direct effects of NiAc per se, such as attenuated toll-like receptor 4 proinfl ammatory signaling ( 36 ) and changes in tissue NAD pools to reduce oxidative stress ( 37 ), have been reported. Whether these effects could explain the rapid increase in insulin sensitivity in the present study requires further investigation.
Tissue lipid loading is determined both by the acute circulating lipid levels (plasma FFA and TG), as well as the endogenous intracellular lipid stores. The ability of plasma FFA lowering alone to signifi cantly impact on total tissue fatty acid utilization has been clearly demonstrated by work showing that acute NiAc administration, in the fasting state, induces a major shift from whole body fat to carbohydrate oxidation, in association with the suppression of plasma FFA ( 38 ). We were hoping to reverse lipid overload, not just by reducing acute circulating FFA, but also by lowering endogenous lipid stores via a reduction of net (24 h average) FFA levels with the intermittent dosing protocol. Our quantitative AUC analysis ( Fig. 4 ) suggests, however, that the FFA rebound is of a magnitude sufficient to cancel the acute NiAc-induced FFA lowering in the intermittently dosed groups, despite the fact that this exposure was maintained ( 20 ). Whether this could be explained by the combined effect of acute FFA level during the clamp and a reduction in FFA AUC over the preceding 24 h (both factors determining tissue lipid concentrations involved in interference of insulin signaling) cannot be answered by the available data. At some point though, prolonged exposure might have a negative effect, as shown by a study made in baboons continuously infused for 20 days ( 40 ). An important aspect of the tolerance development, not explored by the current work, is the infl uence of NiAc infusion duration on the minimum drug holiday length (i.e., the time required to restore the acute dosing effects of NiAc on FFA lowering). This could have implications for optimal dosing design. An informative additional experimental group would be to study a 12 h drug holiday at the end of a 5 day continuous NiAc infusion to see whether acute FFA-lowering and insulin-sensitizing effects are restored.
Loss of FFA lowering during prolonged NiAc infusion was not associated with a shift in the balance in expression of adipose tissue genes involved in liberating versus storing fatty acids. The current results differ in some respects from those of Oh et al. ( 20 ), who related apparent changes in lipolysis to changes in adipose tissue gene expression of Wistar rats. In particular, they observed a substantial downregulation of PDE-3B, which the authors argued could be an important mechanism responsible for the return of FFA to preinfusion levels during prolonged NiAc infusion. NiAcinduced suppression of PDE-3B expression and function presumably only occurred once per day. Failure of the intermittent dosing strategy to lower net FFA levels over the 5 day dosing period was also indicated by the lack of lowering of hepatic TG content ( Table 4 ), a relatively slow turnover tissue lipid pool. This may well explain why the NiAc-induced enhancement in GIR was similar in the previously intermittently dosed group compared with the naïve group, because reduced peripheral lipid availability was only effected through reduction of circulating FFA, and the degree to which plasma FFA level was lowered was similar in both groups. The critical role of acute FFA levels is also evidenced by the apparent lack of amelioration of insulin resistance (HOMA-IR) in the intermittent NiAc dose group in the basal state, a period when FFA levels were not reduced ( Table 2 ).
Our results are reminiscent of studies in the NiAc analog, acipimox, in patients with type 2 diabetes. Thus in association with FFA lowering, acipimox acutely enhanced whole body insulin sensitivity in patients naïve to the drug (via a selective increase in oxidative glucose disposal), and moreover, the degree of insulin sensitization was very similar following 3 months of acipimox treatment ( 21 ). The authors of this study also observed FFA rebound, occurring between acipimox doses, and pointed to this as the probable cause of the failure of long-term acipimox treatment to improve glycemia in the patients. A recent study confi rmed insulin sensitization in association with FFA lowering during short, intensive acipimox therapy, despite failure to reduce lipid accumulation in skeletal muscle ( 39 ). We hypothesize that to fully realize the potential of FFA lowering on improving glucose control, net FFA lowering must be achieved, and to do that the FFA rebound issue has to be solved (see below).
In contrast to the response to intermittent dosing, continuous NiAc infusion did not enhance whole body insulin action, associated with the loss of FFA lowering. There are few reports of continuous NiAc administration with which to compare. In healthy rats, FFA lowering was achieved for at least the fi rst 5 h but had completely returned to pretreatment levels by 24 h. At the 24 h time point, despite similar FFA levels compared with saline-infused controls, an insulin-sensitizing effect was seen, provided that NiAc insulin administration. This might especially be the case if the proposal that FFA rebound involves a counterregulatory response ( 42 ) to defend substrate supply is correct. Whatever the specifi c strategy, the current experimental paradigm combining a translationally relevant preclinical disease model, chronically catheterized for repeated stressfree blood sampling, the programmable/implantable pump to deliver NiAc with its desirable properties of high potency, high solubility, and rapid clearance, provides a useful experimental paradigm to explore alternative protocols in order to reveal the potential of FFA lowering to improve glucose control. How do we resolve our data with the general view that NiAc dosing in humans worsens glucose control? Current study limitations, either species differences or the relatively short duration compared with the long-term treatments in the clinic, might provide trivial explanations. However, we suggest that the timing of NiAc administration and assessment of FFA level, glucose control, or insulin sensitivity may be critical. In particular, we assessed insulin sensitivity in the presence of circulating levels of NiAc suffi cient to suppress FFA. This differs from almost all clinical reports in which assessments are made when circulating NiAc levels are likely to be very low (overnight fasting, last dose taken at bedtime the night before); for example, when NiAc was orally dosed 12 h before clamp studies, FFA rebound was associated with insulin resistance ( 25 ). We are not the fi rst to suggest the potential impact of timing on niacin therapy. Usman et al. ( 44 ) proposed that giving extended-release (ER) niacin at meal time, instead of bed time, could improve TG lowering. Based on the current results, it seems reasonable that this strategy might also improve glucose handling following mixed meal ingestion.
An intermittent NiAc exposure profi le equivalent to the one used in the present study, with stable and suffi cient NiAc levels to suppress FFA 12 h/day, has to our knowledge not been applied in the clinic. This is important because different profi les may have different effects on average FFA levels. Crystalline NiAc dosing seems to raise net plasma FFA levels, due to marked rebound between doses, and induces insulin resistance ( 25,38 ). ER formulations might do better, but reported effects on glucose control are generally either neutral or negative ( 16,17,42,45 ). This could result from a failure of the ER formulations to lower FFA, either on average or at the time point of glucose control assessment. Several factors probably prevent ER formulations from reducing average FFA levels: 1 ) Dosing has not been designed to lower FFA; rather, the goal has been to ameliorate dyslipidemia via effects now understood to be independent of the FFA-lowering mechanism ( 46 ). 2 ) Although the ER formulation prolongs plasma NiAc exposure, maximum plasma NiAc levels are reduced compared with equivalent crystalline formulation doses ( 47 ), likely resulting in poor FFA suppression over a large fraction of the day [based on the relationship between circulating NiAc levels and FFA suppression ( 23 )]. 3 ) In addition, clinical dosing is also associated with signifi cant FFA rebound [e.g., ( 42 )].
has also been reported to occur in mice ( 41 ). In the lean (Sprague Dawley) rats used in the current study, we observed no signifi cant change in PDE-3B mRNA after 5 days of continuous NiAc infusion. We cannot explain the discrepancy between our study and the previous work, although one possible cause might be that we used substantially lower NiAc doses, ‫ف‬ 1/10 of those used in the earlier studies. Overall our analysis of several genes involved in liberation and storage of FFA in adipose tissue does not point to any involvement of expression as the mechanism for the loss of FFA lowering after 5 days of NiAc administration.
The metabolic responses to NiAc cessation are not restricted to FFA rebound and provide insight into the nature of the loss of FFA lowering during continuous NiAc infusion. To our knowledge, this is the fi rst study to assess the detailed, time-dependent response of plasma insulin and glucose in response to sudden NiAc withdrawal and reveals that in all cases where there was a FFA rebound, this was associated with an insulin rebound, with a particularly pronounced response in the obese NiAc-naïve animals. This phenomenon may be the result of the wellknown potentiation of glucose-stimulated insulin secretion by long-chain FFAs ( 33 ). However, the surprisingly modest reductions seen in plasma glucose in response to the insulin rebound suggest that the complete explanation is likely more complex. Indeed, Vega et al. ( 42 ) suggested that NiAc-induced FFA lowering might trigger a counterregulatory response, perhaps to defend substrate supply, and NiAc has been shown to signifi cantly alter levels of a number of hormones including glucagon and growth hormone ( 43 ). The mechanism of the loss of FFA lowering in response to continuous NiAc exposure appears to be different in the lean and obese animals. Thus, in the lean animals, sudden withdrawal of NiAc after 5 days uninterrupted exposure induced a marked FFA rebound ( Fig. 3 and Fig. 4B ), consistent with the interpretation of Oh et al. ( 20 ) that the return of FFA to preinfusion levels represents the net effect of preserved NiAc action in the presence of an enhanced basal rate of lipolysis. By contrast, in obese Zucker rats there was absolutely no evidence of FFA rebound indicating that lipolysis had become completely tolerant to NiAc (i.e., complete tachyphylaxis).
Further refi nements to NiAc dosing might achieve greater lipid lowering and insulin sensitization. As discussed previously, the simple intermittent protocol for infusion of NiAc has probably not succeeded in lowering average daily FFA levels, with the rebound rise in FFA occurring rapidly in response to NiAc withdrawal quantitatively cancelling the FFA lowering during NiAc infusion, resulting in no reduction of hepatic tissue lipid storage. We anticipate that further refi nements of the infusion protocol or optimal timing of NiAc administration relative to food intake might mitigate this issue. Thus, a programmed, more gradual decline in NiAc concentrations to terminate each infusion period might help to minimize the rebound. Alternatively, FFA rebound might be minimized if NiAc withdrawal is timed to occur in association with feeding/ In conclusion, an intermittent NiAc dosing strategy succeeded in retaining FFA lowering and improving insulin sensitivity in obese Zucker rats. Although these data suggest that FFA lowering is suffi cient to improve insulin sensitivity, further refi nements to the administration regime should be explored to more profoundly reverse lipid-overload-induced insulin resistance.