Benzodiazepines (drug class)

First synthesized accidentally by Leo Sternbach at Hoffmann-La Roche in 1955 (chlordiazepoxide / Librium, found in a 1957 lab cleanup), benzodiazepines are GABA-A positive allosteric modulators that produce rapid anxiolysis, sedation, anticonvulsant, and muscle-relaxant effects within 30–60 minutes of oral dosing. The acute efficacy is real and A-tier — they are first-line for status epilepticus and unmatched for acute panic — but every chronic-use signal is bad: A-tier evidence for tolerance (weeks), physical dependence (days to weeks even at therapeutic doses), severe withdrawal (seizures, delirium, akathisia, depersonalization), cognitive impairment that does not fully resolve in some users (BIND / protracted withdrawal in ~10–15% of long-term users, lasting 6–18+ months), elevated dementia risk in older adults, hip-fracture and motor-vehicle-crash signal, and a 2020 FDA Boxed Warning for the entire class covering abuse/misuse/addiction/dependence/withdrawal — added on top of the pre-existing Boxed Warning for opioid co-administration (where ~70% of US benzodiazepine overdose deaths in 2023 also involved illicitly manufactured fentanyl). For Dylan (20yo, no seizure history, no diagnosed anxiety disorder, brain-priority + MMA reaction-time goals): SKIP-PERMANENT for chronic use, full stop. PRN single-dose under a prescriber for a real medical event is a different conversation, and not what this file is gating against.

The binding site (the part everyone gets wrong)

Benzodiazepines do not activate GABA-A receptors directly. They bind to a distinct allosteric site at the interface of an α-subunit and the γ-subunit (most commonly γ2) of the pentameric GABA-A receptor. When GABA itself is bound at the orthosteric (β/α interface) site, benzodiazepine occupancy of the α/γ allosteric site increases the frequency of chloride channel opening — without changing the maximum chloride conductance per opening. Barbiturates, by contrast, increase the channel-open duration and at high enough doses become direct GABA-A agonists, which is why they have a much narrower therapeutic index and more easily cause respiratory depression in monotherapy overdose. This frequency-vs-duration distinction is why benzodiazepine-only oral overdoses are usually survivable in opioid-naive adults and barbiturate overdoses historically were not.

Subunit selectivity and the BZ1/BZ2 nomenclature

GABA-A receptors are heteropentamers built from a menu of 19 subunits (six α, three β, three γ, plus δ, ε, π, θ, ρ). The α-subunit in the receptor determines the bulk of the benzodiazepine pharmacology, and benzodiazepines bind to receptors containing α1, α2, α3, or α5 (NOT α4 or α6 — those are diazepam-insensitive, which is why pharmacology textbooks distinguish "BZ-sensitive" α-subunits from "BZ-insensitive"):

• α1 (BZ1, ω1) — sedation, hypnosis, anterograde amnesia, ataxia, much of the abuse liability. Densest in cortex, thalamus, cerebellum. Zolpidem (Ambien, a Z-drug) is α1-selective, which is why it sedates without doing much for anxiety.
• α2 — anxiolysis (the part anxious patients actually want). Densest in limbic system and motor neurons.
• α3 — anxiolysis + muscle relaxation. Co-mediator of anxiolysis with α2.
• α5 — learning, memory, hippocampal LTP. Implicated in the cognitive-impairment side of benzodiazepine pharmacology. α5 inverse agonists are an active research target as cognitive enhancers (basmisanil, MRK-016, α5IA — all failed for AD/Down syndrome cognition but the rationale was that removing α5 GABA-A tone would lift cognition; relevance here: if α5 modulation tunes cognition, chronically amplifying α5 GABA-A inhibition with a benzodiazepine is plausibly bad for cognition, and that aligns with what we see clinically).

No marketed clinical benzodiazepine is meaningfully subunit-selective. All of them — alprazolam, clonazepam, diazepam, lorazepam — hit α1, α2, α3, and α5 with broadly similar affinity, which is why they all sedate AND anxiolyze AND impair memory. The pharmacological holy grail of an α2/α3-selective anxiolytic without α1 sedation or α5 cognitive cost has been pursued for 30+ years (TPA023, MRK-409, others) and as of 2026 has produced no marketed drug.

Plain-English summary

GABA is the brain's main "off" signal. Benzodiazepines don't add a new "off" signal — they make the existing GABA "off" signal hit harder, by holding the receptor in a more responsive shape. Because GABA-A receptors with different α-subunits do different jobs (calm fear, induce sleep, suppress seizures, impair memory, relax muscle), and because every clinical benzodiazepine hits all of them at once, you get all of those effects bundled — desirable and undesirable — every time. With chronic exposure, the receptors downregulate, change subunit composition, and uncouple GABA from benzodiazepine binding, which is the molecular substrate of tolerance and the reason abrupt cessation produces an acute glutamate/GABA imbalance — and seizures.

Pharmacokinetics — the four agents Dylan would encounter

The half-life table is the single most clinically relevant fact about this class. Short-half-life agents (alprazolam, triazolam, midazolam) produce stronger inter-dose withdrawal, more rebound anxiety, more conditioned PRN reinforcement (you feel it work and feel it wear off, several times a day) — and that is precisely the abuse-pattern profile. Long-half-life agents (diazepam, clonazepam) build up over days, produce less inter-dose oscillation, and are easier to taper from — which is why Heather Ashton's protocol substitutes diazepam for short-acting benzodiazepines before starting the dose-reduction taper. Lorazepam is unusual in that it is metabolized by glucuronidation (UGT) rather than the CYP system, so it is the standard ICU/elderly choice when CYP3A4 interactions or hepatic impairment make alprazolam/diazepam risky.