Lipocalin comes callin’ on the hippocampus

Several decades of clinical and basic research have characterized the complex effects of stress on the brain and body (1–4). Stress, commonly defined as real or implied threat to homeostasis, initially triggers a protective response that helps the organism to survive and adapt to the stressful experience. This short-term response to stress that enables the organism to adapt and regain stability has been termed “allostasis” (4, 5). Prolonged exposure to severe stress, on the other hand, burdens the same protective mechanisms and leads to deleterious effects by increasing the “allostatic load” on both the brain and body (4, 5). Accumulating evidence has identified a range of metabolic, autonomic, and immune mechanisms underlying the body's physiological response to stress (1–4). Early research on the impact of stress on the brain focused largely on a structure that plays a pivotal role in learning and memory—the hippocampus (5, 6). Although impressive progress has been made on both fronts separately, integrating the two lines of investigation has been more challenging. We still have a limited understanding of the interactions between the nervous system and the various physiological mediators that operate throughout the whole body during and after stress. A study in PNAS (7) bridges this gap by identifying a unique role for a protein called lipocalin-2 (Lcn-2) in modulating the effects of stress on the hippocampus. Lipocalins constitute a diverse family of small soluble proteins, many of which are secreted extracellularly (8). Although growing evidence implicates lipocalins in a diverse range of regulatory processes in the body, including apoptosis and inflammation (9, 10), this study identifies a role in stress-induced plasticity in the brain.

Growing evidence gathered from various animal models shows that stress leads to deficits in hippocampal function at multiple levels of neural organization, ranging from molecular and synaptic mechanisms to their functional consequences at the behavioral level (6). Dendritic remodeling, along with changes in spine morphology and number, has emerged as a major cellular metric of stress-induced structural plasticity in the rodent hippocampus (6, 11, 12). Moreover, glucocorticoids, as well as glutamate and its receptors, have been implicated in stress-induced structural plasticity (6). Further evidence for the importance of glutamatergic transmission comes from electrophysiological studies showing how stress and stress hormones modulate hippocampal synaptic plasticity mechanisms, such as long-term potentiation (LTP) (6, 13, 14). As a result of these findings, dendritic spines, the site of excitatory synaptic transmission and plasticity, have been at the center of many earlier studies on stress effects in the hippocampus (11, 12, 15). Indeed, changes in spines are also a key focus of the report by Pawlak and colleagues (7). They find that exposure to 3 days of restraint stress enhances Lcn-2 expression in excitatory principal neurons in the hippocampus, including areas CA1 and CA3 (Fig. 1A). The same stress that leads to this up-regulation in Lcn-2 expression also increases spine density on CA1 pyramidal neurons (Fig. 1A). Strikingly, in vitro treatment with Lcn-2 elicited a robust decrease in spine density in cultured hippocampal neurons (Fig. 1 A and C). Upon closer examination, Lcn-2 seems to trigger two specific effects on the distribution of different classes of spines: a reduction in mushroom spines in parallel with a higher proportion of immature, thin spines (Fig. 1C). Thus, although stress triggers an increase in spine numbers and Lcn-2, the latter by itself has the opposite effect on spine number (Fig. 1A). This suggests a homeostatic “push–pull” mechanism wherein stress pushes spine numbers upward, while Lcn-2 pulls spine density down below control levels (Fig. 1C). Interestingly, stress-induced up-regulation of spine density happens despite the elevated level of Lcn-2 that is expected to suppress spine density by acting in the opposite direction. This raises the intriguing possibility that, in control mice, the two opposing influences are in steady state, with the stress effects having an edge that is manifested as a net gain in spine numbers (Fig. 1C). The authors tested whether this balance is disrupted in mice containing a genetic deletion of the Lcn2 gene (Lcn-2^−/− mice). Consistent with this model, the same restraint stress leads to an even greater increase in spine density in Lcn-2^−/− mice (Fig. 1 B and C). In other words, in the absence of the brakes exerted by Lcn-2, the balance tilts further in favor of stress-induced increase in spine density (Fig. 1 B and C). Furthermore, in Lcn-2^−/− mice, stress triggers the opposite shift in the relative proportion of mushroom and thin spines in CA1 pyramidal cells in vivo than that elicited by Lcn-2 in vitro (Fig. 1C). Although future studies will be required to explore various facets of this model, it will be particularly important to examine the role of glucocorticoids in the effects of Lcn-2 in modulating the shape and number of hippocampal spines.

Fig. 1.

Lcn-2 and behavioral stress exert a “push–pull” effect on dendritic spines in the hippocampus. (A) As described in PNAS (7), stress leads to an increase in both Lcn-2 and spine-density. However, Lcn-2 causes the opposite effect by decreasing spine numbers. (B) The same stress elicits an even greater increase in spine density in Lcn-2^−/− mice. (C) Hippocampal cells in unstressed, WT mice have a certain distribution of mature (mushroom-shaped) and immature (thin) spines adding up to control levels of spine density (gray horizontal line). Lcn-2 treatment lowers spine density and increases the relative proportion of thin spines (green downward arrow). In the absence of the brakes exerted by Lcn-2, stress creates a greater proportion of mushroom spines (Stressed Lcn-2^−/−) and further increases spine density (upward red arrow) relative to WT mice (upward black arrow).

Looking beyond structural changes at the level of dendritic spines, these findings also shed light on the electrophysiological and behavioral consequences. CA1 pyramidal neurons from Lcn-2^−/− mice exhibited higher input resistance and fired more action potentials in response to somatic depolarization. Because such an increase in neuronal excitability is expected to serve as an ideal cellular substrate for facilitating the induction of LTP, it would be interesting to examine whether hippocampal LTP is enhanced in Lcn-2^−/− mice. Further analysis will also be needed to explore the interplay between potential changes in LTP in Lcn-2–deficient CA1 neurons and stress-induced modulation of LTP in the same neurons. Previous electrophysiological studies in the hippocampus have suggested an inverted U-shaped dose–response curve for the relationship between corticosterone and LTP, with low levels of corticosterone facilitating synaptic potentiation, whereas higher levels inhibit it in the CA1 region (6, 13–15). Because stress-induced CA1 spinogenesis is enhanced in the Lcn-2–deficient mice, how would stress-induced modulation of LTP pan out in these mice? On the basis of earlier findings, the answer is likely to depend on the levels of glucocorticoids elicited by the stressor, which in turn depends on the duration and intensity of stress. The stress paradigm used in the present study (6 h of restraint per day for 3 consecutive days) is not as brief as previously used models of acute stress involving either a single 2-h episode of immobilization (16) or 6-h session of restraint (6). However, the 3-d paradigm is significantly less severe compared with previously published protocols of chronic immobilization (2 h/d for 10 d) or repeated restraint stress (6 h/d for 21 d) (17). Therefore, the cellular and molecular changes reported in this study are likely to be mediating a hippocampal response to a stressor of moderate intensity rather than a more prolonged and severe stress that causes aberrant plasticity and functional impairment. In this context it would be interesting to investigate where the changes in Lcn-2 and spines fall within the stress-response continuum between allostasis and allostatic load. For instance, in contrast to the increase in spine density reported in the present study (7), a longer duration of repeated restraint stress down-regulates spines in area CA1 (12). In this earlier work, Pawlak et al. (12) identified a role for another extracellular signaling mechanism—the tissue plasminogen activator/plasmin system—in this form of chronic stress-induced reduction in CA1 spines. Because Lcn-2 is also known to be secreted through extracellular release, it too may have a role in the biphasic modulation of spines by acute (spine increase) vs. chronic stress (spine decrease). Thus, future studies need to extend the stress duration to further develop the “push–pull” model emerging from the present study as the basis of a biphasic effect of stress and glucocorticoids (13–15).

Questions related to the temporal features of stress-induced changes are also relevant in light of earlier reports showing that the duration and intensity of the stressor can lead to time-dependent modulation of anxiety-like behavior, a functional outcome that has been characterized in the present study. In Lcn-2^−/− mice, exposure to stress led to anxiety levels that were higher than in wild-type mice. Although earlier studies have demonstrated a role for the hippocampus in anxiety-like behavior, more recent evidence has also linked growth of spines in the basolateral amygdala (BLA)—caused either by stress or genetic manipulations—to enhanced anxiety-like behavior (16, 17). The same time points when stress triggers higher anxiety is also when spine density is elevated on principal neurons in the BLA. For instance, anxiety and spine density are both increased 1 d after chronic immobilization stress (17). Even a shorter duration of acute stress that fails to affect spine density or anxiety 1 d later leads to a significant increase in both ten days later (16). Together, these studies have helped identify unique features of stress-induced structural plasticity in the amygdala that are quite distinct from those observed in the hippocampus (17). In light of the evidence provided by Mucha et al. (7), it will be of considerable interest to see whether Lcn-2 also plays a role in stress-induced spinogenesis in the amygdala. Thus, in addition to the temporal dimensions of stress effects (acute vs. chronic, immediate vs. delayed), the spatial location (i.e., region-specific differences between the hippocampus and amygdala) may also be relevant for future studies (17). Indeed, there are already data suggesting that the role played by Lcn-2 often varies with the context in which it is being activated. For example, after spinal cord injury, Lcn-2 protein level is enhanced rapidly (18). However, in this case Lcn-2 has been shown to have detrimental effects after injury, and the lack of Lcn-2 actually improves locomotor recovery after injury (18). Similarly, Lcn-2 expression and secretion is enhanced after inflammatory stimulation in cultured astrocytes (19). Further, treatment with Lcn-2 protein increases the sensitivity of astrocytes to cytotoxic stimuli (19). In contrast to these examples, the findings reported in PNAS (7) suggest that Lcn-2 may help dampen the impact of stress on hippocampal spinogenesis and anxiety-like behavior. By identifying an intriguing player in stress-induced plasticity, this study opens up the possibility for new experimental strategies for investigating how the brain and body respond to and recover from stressful experiences.