In their target article, Mather and colleagues propose a molecular model, glutamate amplifies noradrenergic effects (GANE), through which arousal enhances or inhibits perception, attention, and memory consolidation. In this model, arousal precipitates phasic release of norepinephrine (NE) throughout the brain, but “hotspots” of NE release are generated near activated glutamate circuits, sufficient to engage low-affinity β-adrenoceptors, which further increase glutamate release and enhance postsynaptic plasticity by increasing cAMP signaling. The model provides an impressive integration across several fields but will require empirical investigation to test its validity. Furthermore, although the model is presented as universally applicable throughout the brain, NE actually has very heterogeneous actions in different brain circuits. In particular, although the GANE model addresses the effects of normal arousal mechanisms in sensory cortex and hippocampus, it is important to discuss how this model may relate to NE actions in higher cognitive circuits and to conditions of psychopathology.

The noradrenergic system plays an essential role in the pathophysiology and treatment of psychiatric disorders. For example, noradrenergic dysregulation is associated with post-traumatic stress disorder (PTSD), and α₁-antagonists can reduce these symptoms (Arnsten et al. Reference Arnsten, Raskind, Taylor and Connor2015b; Southwick et al. Reference Southwick, Bremner, Rasmusson, Morgan, Arnsten and Charney1999). Many antidepressants target the noradrenergic system (Klimek et al. Reference Klimek, Stockmeier, Overholser, Meltzer, Kalka, Dilley and Ordway1997), and α_2A-agonists enhance cognition in patients with attention deficit hyperactivity disorder (ADHD) (Arnsten & Wang Reference Arnsten and Wang2016). Similarly, accumulating evidence implicates glutamate in the etiology and treatment of mental disorders (Chambers et al. Reference Chambers, Bremner, Moghaddam, Southwick, Charney and Krystal1999; Krystal et al. Reference Krystal, Sanacora and Duman2013). Whether the GANE model applies to traumatic stress conditions is not clear; the research Mather et al. cite utilized subtle arousing conditions – for example, an emotional word. It is, however, likely to explain several aspects of PTSD: for example, enhancement of the consolidation of traumatic events that may contribute to flashbacks and intrusive memories. However, additional, higher brain changes during trauma may not be captured by this model, as NE actions in the brain are more heterogeneous than described.

Most important for human cognition, the newly evolved circuits in layer III of the dorsolateral prefrontal cortex (dlPFC) that underlie higher cognitive operations are modulated in a unique manner that is often opposite that of classic synapses in sensory cortex, amygdala, and hippocampus (Arnsten et al. Reference Arnsten, Wang and Paspalas2012). Indeed, these newly evolved “delay cell” circuits in the dlPFC are even regulated differently than sensory/response-related neurons within the dlPFC. For example, delay cell persistent firing is mediated by NMDAR with NR2B subunits that are exclusively in the postsynaptic density, not extrasynaptic as they are in classic synapses (Wang et al. Reference Wang, Yang, Wang, Gamo, Jin, Mazer, Morrison, Wang and Arnsten2013). Furthermore, delay cells are only subtly influenced by AMPA receptors and show reduced, rather than increased, neuronal firing following systemic ketamine (Wang et al. Reference Wang, Yang, Wang, Gamo, Jin, Mazer, Morrison, Wang and Arnsten2013). In contrast, response feedback cells in the dlPFC (likely layer V) have a more classic profile, with large AMPA receptor influences and increased firing with systemic ketamine (Wang et al. Reference Wang, Yang, Wang, Gamo, Jin, Mazer, Morrison, Wang and Arnsten2013). These marked differences extend to intracellular cAMP signaling events as well. In classic synapses, activation of cAMP signaling, for example, arising from β-adrenoceptor stimulation, increases glutamate release from axon terminals and strengthens long-term potentiation (LTP) postsynaptically. However, in layer III dlPFC circuits, increased cAMP signaling weakens connections by opening cAMP-PKA-regulated potassium channels in dendritic spines (Arnsten Reference Arnsten2015; Arnsten et al. Reference Arnsten, Wang and Paspalas2012). Instead, it is inhibition of cAMP signaling via postsynaptic α_2A-adrenoceptors that strengthens network connectivity by closing potassium channels near the synapse (Wang et al. Reference Wang, Ramos, Paspalas, Shu, Simen, Duque, Vijayraghavan, Brennan, Dudley, Nou, Mazer, McCormick and Arnsten2007). There is currently no evidence of NE “hotspots” in these circuits; for example, blockade of β-receptors within the primate dlPFC has no effect on working memory performance (Li & Mei Reference Li and Mei1994), even though there are likely high levels of glutamate release in dlPFC arising from the persistent firing of these neuronal networks. Thus, the model in Figure 6 of the target article is misleading because it does not differentiate NE actions in classic synapses from those in more newly evolved dlPFC circuits.

Mather et al. also provide an oversimplified discussion of NE actions at α₁-adrenoceptors. Although they focus on α₁ mechanisms that weaken plasticity, α₁ promotes synaptic actions in many synapses – for example, in somatosensory cortex (Mouradian et al. Reference Mouradian, Seller and Waterhouse1991; Waterhouse et al. Reference Waterhouse, Moises and Woodward1981; Reference Waterhouse, Mouradian, Sessler and Lin2000). There are also key circuits where α₁-receptor activation potentiates β-receptor actions: For example, in amygdala, α₁-receptors facilitate β-adrenergic enhancement of memory consolidation (Ferry et al. Reference Ferry, Roozendaal and McGaugh1999a; Reference Ferry, Roozendaal and McGaugh1999b). These effects are opposite those described by Mather and colleagues. Their model also does not capture the important finding that high levels of NE release in PFC during stress decrease persistent firing and working memory abilities through stimulation of α₁-receptors (Birnbaum et al. Reference Birnbaum, Yuan, Wang, Vijayraghavan, Bloom, Davis, Gobeske, Sweatt, Manji and Arnsten2004). All of these actions likely have a key effect in switching control of behavior from thoughtful, flexible, top-down control by PFC under conditions of safety (moderate levels of arousal) to reflexive, unconscious habits mediated by sensorimotor cortex and subcortical structures during uncontrollable stress (very high levels of arousal).

These mechanisms have particular relevance to the symptoms of PTSD, for which there is extensive evidence of elevated noradrenergic activity (Southwick et al. Reference Southwick, Bremner, Rasmusson, Morgan, Arnsten and Charney1999). For example, the α₂-antagonist yohimbine worsens symptoms and induces hypofrontality in subjects with PTSD at doses that have little effect in control subjects (Bremner et al. Reference Bremner, Innis, Ng, Staib, Salomon, Bronen, Duncan, Southwick, Krystal, Rich, Zubal, Dey, Soufer and Charney1997; Southwick et al. Reference Southwick, Krystal, Morgan, Johnson, Nagy, Nicolaou, Heninger and Charney1993). These drug actions may arise from a combination of neural events, for example, loss of dlPFC top-down control from blockade of post-synaptic α_2A-receptors and increased NE stimulation of α₁-receptors in dlPFC, as well as increased NE release in “hotspots” in the amygdala, hippocampus, and sensory cortex that may exacerbate anxiety and flashbacks (Arnsten et al. Reference Arnsten, Raskind, Taylor and Connor2015b). Thus, the GANE model may apply to NE actions in classic brain circuits, but not to those in higher cortical circuits, which are strengthened by α_2A- rather than β-adrenoceptor mechanisms.