We have explored the structure-activity relationship (SAR) surrounding the clinically efficacious antiepileptic drug topiramate (1), a unique sugar sulfamate anticonvulsant that was discovered in our laboratories. Systematic structural modification of the parent compound was directed to identifying potent anticonvulsants with a long duration of action and a favorable neurotoxicity index. In this context, we have probed the pharmacological importance of several molecular features: (1) the sulfamate group (6-8, 22-25, 27, 84), (2) the linker between the sulfamate group and the pyran ring (9, 10, 21a,b), (3) the substituents on the 2,3- (58-60, 85, 86) and 4, 5-fused (30-38, 43, 45-47, 52, 53) 1,3-dioxolane rings, (4) the constitution of the 4,5-fused 1,3-dioxolane ring (2, 54, 55, 63-68, 76, 77, 80, 83a-r, 84-87, 90a, 91a, 93a), (5) the ring oxygen atoms (95, 96, 100-102, 104, 105), and (6) the absolute stereochemistry (106 and 107). We established the C1 configuration as R for the predominant alcohol diastereomer from the highly selective addition of methylmagnesium bromide to aldehyde 15 (16:1 ratio) by single-crystal X-ray analysis of the major diastereomer of sulfamate 21a. Details for the stereoselective syntheses of the hydrindane carbocyclic analogues 95, 96, 100, and 104 are presented. We also report the synthesis of cyclic imidosulfites 90a and 93a, and imidosulfate 91a, which are rare examples in the class of such five-membered-ring sulfur species. Imidosulfite 93a required the preparation and use of the novel sulfur dichloride reagent, BocN=SCl2. Our SAR investigation led to the impressive 4,5-cyclic sulfate analogue 2 (RWJ-37947), which exhibits potent anticonvulsant activity in the maximal electroshock seizure (MES) test (ca. 8 times greater than 1 in mice at 4 h, ED50 = 6.3 mg/kg; ca. 15 times greater than 1 in rats at 8 h, ED50 = 1.0 mg/kg) with a long duration of action (>24 h in mice and rats, po) and very low neurotoxicity (TD50 value of >1000 mg/kg at 2 h, po in mice). Cyclic sulfate 2, like topiramate and phenytoin, did not interfere with seizures induced by pentylenetetrazole, bicucculine, picrotoxin, and strychnine; also, 2 was not active in diverse in vitro receptor binding and uptake assays. However, 2 turned out to be a potent inhibitor of carbonic anhydrase from different rat tissue sources (e. g., IC50 of 84 nM for the blood enzyme and 21 nM for the brain enzyme). An examination of several analogues of 2 (83a-r, 85-87, 90a, 91a, 93a) indicated that potent anticonvulsant activity is associated with relatively small alkyl substituents on nitrogen (Me/H, 83a; Me/Me, 83m; Et/H, 83b; allyl/H, 83e; c-Pr/H, 83j; c-Bu/H, 83k) and with limited changes in the cyclic sulfate group, such as 4,5-cyclic sulfite 87a/b. The potent anticonvulsants 83a and 83j had greatly diminished carbonic anhydrase inhibitory activity; thus, inhibition of this enzyme may not be a significant factor in the anticonvulsant activity. The alpha-L-sorbopyranoses 67, 68, and 80, which mainly possess a skew conformation (ref 29), were nearly twice as potent as topiramate (1). The L-fructose enantiomers of 1 (106) and 2 (107), synthesized from L-sorbose, were found to have moderate anticonvulsant activity, with eudysmic ratios (MES ED50 in mice at 4 h, po) of 1:106 = 1.5 and 2:107 = 3.5. The log P values for 1 and 2 were determined experimentally to be 0.53 and 0.42, respectively, which are less than the optimal 2.0 for CNS active agents. However, analogues with more favorable calculated log P (clogP) values, in conjunction with just minor steric perturbation according to the developed SAR profile, such as 47 (clogP = 2.09), 83m (1.93), and 86 (1.50), did not display improved potency: 47 is less potent than 1, 83m is equipotent with 2, and 86 is less potent than 2. Although the measured log P value for diethyl analogue 31 is 1.52, this did not translate into enhanced potency relative to 1. (ABSTRACT TRUNCATED)