Open Access

Differential effects of antidepressants escitalopram versus lithium on Gs alpha membrane relocalization

  • Robert J Donati1, 3Email author,
  • Jeffrey Schappi1,
  • Andrew H Czysz1,
  • Alexander Jackson^1 and
  • Mark M Rasenick1, 2
^Deceased
BMC Neuroscience201516:40

https://doi.org/10.1186/s12868-015-0178-y

Received: 16 February 2015

Accepted: 6 July 2015

Published: 11 July 2015

Abstract

Background

Plasma membrane localization can play a significant role in the ultimate function of certain proteins. Specific membrane domains like lipid rafts have been shown to be inhibitory domains to a number of signaling proteins, including Gsα, and chronic antidepressant treatment facilitates Gs signaling by removing Gsα form lipid rafts. The intent of this study is to compare the effects of the selective serotnin reuptake inhibitor, escitalopram, with that of the mood stabilizing drug, lithium.

Results

There are a number of mechanisms of action proposed for lithium as a mood stabilizing agent, but the interactions between G proteins (particularly Gs) and mood stabilizing drugs are not well explored. Of particular interest was the possibility that there was some effect of mood stabilizers on the association between Gsα and cholesterol-rich membrane microdomains (lipid rafts), similar to that seen with long-term antidepressant treatment. This was examined by biochemical and imaging (fluorescence recovery after photobleaching: FRAP) approaches. Results indicate that escitalopram was effective at liberating Gsα from lipid rafts while lithium was not.

Conclusions

There are a number of drug treatments for mood disorders and yet there is no unifying hypothesis for a cellular or molecular basis of action. It is evident that there may in fact not be a single mechanism, but rather a number of different mechanisms that converge at a common point. The results of this study indicate that the mood stabilizing agent, lithium, and the selective serotonin reuptake inhibitor, escitalopram, act on their cellular targets through mutually exclusive pathways. These results also validate the hypothesis that translocation of Gsα from lipid rafts could serve as a biosignature for antidepressant action.

Background

Despite several decades of research, no unifying hypothesis for a molecular or cellular basis of action for antidepressant drugs or mood-stabilizing drugs has emerged. Several studies in both human and animal tissue suggest that there is lower cAMP production in depressed individuals (see reviews [1, 2]) while a recent report using PET imaging suggests a global decrease of cAMP in brains from depressed subjects [3]. The reviews above also suggest that effective antidepressant treatment increases adenylyl cyclase (AC) activity. It has also been observed that effective antidepressants induce movement of Gsα out of lipid rafts, thus increasing the association between Gsα and AC, elevating AC activity, and increasing cAMP [46]. Gsα from C6 rat glioma cells migrates from a Triton X-100 (TX-100) resistant lipid raft containing membrane domain to a TX-100 soluble non-lipid raft membrane domains in response to chronic antidepressant treatment [4, 7] revealing Gsα as a preferential target for antidepressant action [4]. Taken together, these studies suggest that the lipid environment of the G protein may play an important role in its localization and function, and that chronic antidepressant treatment alters the membrane localization of Gsα, augmenting coupling to AC [7].

Human post-mortem data reveals that depressed subjects have more Gsα in the TX-100 resistant (raft enriched domains) compared to control subjects [8]. These results suggest that Gsα is less available for AC signaling in the depressed brain and is consistent with the observation that a therapeutic effect of antidepressants may be to move Gsα out of TX-100 resistant regions of the membrane and into a membrane domain where it is more available to interact with AC. Additionally, Gsα-mediated AC activity is significantly lower in platelets from depressed subjects [9, 10]. Similar to results seen in cell culture these studies suggest that during depression, Gsα is sequestered in lipid raft-like domains and antidepressant treatment liberates the Gsα from these inhibitory domains allowing it to more freely couple to AC. Furthermore, antidepressant drugs concentrate in raft-like plasma membrane domains [11]. Thus, antidepressants may exert their observed effects on cAMP signaling by liberating Gsα from lipid rafts, where it accumulates during the course of depression [8, 12].

Lithium continues to be the primary pharmacological treatment for bipolar disorder. It has been shown that lithium and other mood stabilizers inhibit AC activity [13, 14] with lithium having a preference for Types V and VII AC [15]. There are numerous hypotheses on the mechanism of action of lithium (review by Marmol [16]) with one of the more current mechanisms involving inhibition of GSK-3 (review by Machado-Vieira et al. [17]). The involvement of serotonin and dopamine regulation of GSK-3 activity in the action of psychotropic drugs has also been recently reviewed [1821]. Previous studies have shown that the effects of lithium in the treatment of affective disorders may be due to inhibition/decrease of the G-protein Giα activity and an increase in AC types I and II activity [22].

Currently, it is not possible to directly and noninvasively monitor the activity of AC or the partitioning of Gsα. However, if a biomarker can be identified for the process of making Gsα available to complex with AC, it would then be possible to quantify whether a given drug treatment is altering Gsα-AC coupling in a shorter time period, weeks vs months, predicting a likely therapeutic success. It is our current hypothesis that chronic antidepressant treatment, but not chronic treatment with other classes of drugs used to treat mood disorders, may alter lipid raft composition (lipids, proteins or both) so that the anchor for Gsα is lost and Gsα moves into non-raft fractions, increasing its availability to activate AC [12]. Escitalopram, the therapeutically active S-enantiomer of citalopram, liberated Gsα from detergent resistant membranes of chronically treated C6 glioma cells, while R-citalopram or acute drug treatment had no effect [23].

The results of the current study demonstrate that escitalopram facilitates the release of Gsα, but not Giα, from detergent resistant membrane domains while lithium and valproic acid do not have this effect. In fact, lithium and valproic acid may actually increase the movement of Gsα into these detergent resistant membrane domains.

Results

Effect of antidepressant vs lithium on Gsα TX-100 resistant membrane localization

Chronic antidepressant treatment has been shown to decrease the localization of Gsα in detergent resistant membrane domains (lipid rafts), both in cerebral cortex membranes from rats and in C6 glioma cells [4, 7, 23]. Despite its structural similarity to tricyclic antidepressants, chlorpromazine did not have this effect [7, 25]. To test whether two of the more common mood stabilizing drugs, lithium and valproic acid, had similar effects as antidepressants, C6 cells were exposed to escitalopram, lithium, or valproic acid. Extraction of the membranes with TX-100 followed by sucrose density gradient centrifugation allows for the purification of signaling molecules within detergent resistant membranes [24]. The results show that while escitalopram reduced the amount of Gsα in the lipid raft compared with drug-free control cells, lithium and valproic acid had the opposite effect (Figure 1). There is translocation of Gsα out of lipid raft domains caused by the antidepressant escitalopram and a shift of Gsα into the lipid raft domains by the mood stabilizing agents, lithium and valproic acid compared to control. Additionally, there is a significant difference compared to escitalopram treatment (p < 0.005 for valproic acid and lithium).This experiment demonstrated that an antidepressant stimulated the movement of Gsα out of lipid rafts, while the two mood stabilizers had the opposite effect.
Figure 1

Gsα is removed from lipid rafts subsequent to antidepressant treatment while lithium has the opposite effect. Cells were treated as indicated in “Methods” and lipid rafts were isolated using sucrose gradient flotation. The amount of Gsα was determined by Western blot. a The figure shows the percentage of change in Gsα protein compared to control in the lipid raft membrane fractions (n = 7 for control, escitalopram and lithium; n = 5 for valproic acid). Data were analyzed by one-way ANOVA (p < 0.0006) followed by Tukey’s multiple comparison test for post hoc comparison of means. Data are represented as mean ± SEM (**, p < 0.005 compared to escitalopram). b A representative immunoblot of lipid raft Gsα from the various treatment paradigms as explained in “Methods”. C control, E escitalopram, L lithium, V valproic acid.

Effect of antidepressant vs lithium on Giα TX-100 resistant membrane localization

Previous studies have shown that, while a number of different antidepressants alter the association of Gsα with lipid rafts, the localization of Giα remains unaffected [4, 7, 25]. This suggested that the antidepressants were not altering lipid rafts, but were altering the anchoring of Gsα to those rafts. Figure 2 demonstrates that neither escitalopram nor lithium alters the membrane localization of Giα in C6 glioma cells (Figure 2). While valproic acid appears to increase the amount of Giα in the lipid rafts, there is a much greater variability between trials compared to escitalopram and lithium treatment. Congruent with this result and the results of the study mentioned above, it appears that enhancing Gsα mobility out of lipid raft domains may in fact, be a hallmark of antidepressant action.
Figure 2

Giα membrane compartmentalization is not altered by chronic antidepressant or lithium treatment. Purified lipid rafts isolated using sucrose gradient flotation revealed that neither chronic antidepressant treatment nor chronic lithium treatment move Giα out of lipid rafts. a The figure shows the percentage of change in Giα protein in the lipid raft membrane fractions compared to control (n = 4 for control, escitalopram, and lithium; n = 3 for valproate). Data were analyzed by one-way ANOVA followed by Tukey’s multiple comparison test for post hoc comparison of means. Data are represented as mean ± SEM b A representative immunoblot of lipid raft Gsα from the various treatment paradigms as explained in “Methods”. C control, E escitalopram, L lithium, V valproic acid.

Effect of antidepressant vs lithium on Gsα and Giα membrane mobility

Lateral mobility within the plasma membrane of Gsα as determined by fluorescence recovery after photobleaching (FRAP), changed following 1–3 days (depending on concentration) of antidepressant treatment [12]. Specifically, the half-time to recovery of GFP-Gsα is increased, likely due the increased association of Gsα, with the slow-moving adenylyl cyclase. In contrast, FRAP of GFP-Giα1 was unchanged after antidepressant treatment. A typical membrane bleaching and recovery of fluorescence is shown in Figure 3a. Escitalopram treated cells demonstrate a shorter total recovery (increased “immobile fraction”), perhaps because Gsα is bound to adenylyl cyclase (Figure 3b). Additionally, escitalopram treated cells have a less steep recovery curve than either control or lithium treated cells, which have similarly shaped curves. That is, they recover their fluorescence more slowly (increased half time of recovery). Figure 4 shows GFP-Gsα FRAP after 3 days of treatment with lithium chloride (3 mM), valproic acid (300 μM), escitalopram (10 μM), or escitalopram plus lithium combined (same doses as individual treatments with these compounds). While lithium and valproic acid treatment show no effect on Gsα membrane mobility compared to vehicle, treatment with escitalopram or escitalopram plus lithium shows the characteristic slowing of Gsα mobility as demonstrated by an increase in recovery half-time compared to vehicle (p < 0.0005 for escitalopram and p < 0.0005 for escitalopram plus lithium). In contrast, FRAP of GFP-Giα1 is unaltered by the aforementioned treatments (Figure 5), suggesting again that the effects of these drugs are not mediated through actions on Giα1.
Figure 3

FRAP suggests increased release of GFP-Gsα from lipid rafts after chronic antidepressant but not lithium treatment. a Typical course of photobleaching with representative images of cell before photobleaching (t = −3 s), immediately after photobleaching (t = 0 s), and after maximal recovery of fluorescence (t = 45 s). b Demonstration of typical fluorescence recovery after photobleaching in cells treated with escitalopram 10 µM or LiCl 3 mM for 72 h as described in “Methods”. Yellow arrows indicate area of bleach and recovery.

Figure 4

FRAP suggests that lithium combined with antidepressant does not reverse the effect of antidepressant GFP-Gsα was stably expressed in a C6 glioma cell line. Half-time to recovery of GFP-Gsα is increased after chronic escitalopram treatment, suggesting increased coupling with its effector, adenylyl cyclase. Lithium combined with escitalopram does not affect this value, suggesting an alternate locus of action for lithium. GFP-Gsα recovery is not altered by lithium or valproic acid treatment, suggesting unaltered coupling of GFP-Gsα with adenylyl cyclase. Sample size represents the number of cells assayed, with a minimum of 21 and a maximum of 101 cells assayed per experiment. Data were analyzed by one-way ANOVA (p < 0.0001) followed by Tukey multiple comparison test for post hoc comparisons of means. Data are represented as mean ± SEM (***, p < 0.0005 compared to vehicle).

Figure 5

GFP-Giα FRAP is unaffected by chronic antidepressant or mood stabilizer treatment. Giα-GFP was stably expressed in a C6 glioma cell line. Half-time to recovery of Giα-GFP is not altered by chronic treatment with escitalopram, lithium, or valproic acid. This suggests that the change in membrane compartmentalization of Gsα is not due to direct effects of these agents on lipid rafts, but that the effects of antidepressants are specific to Gsα. Sample size represents the number of cells assayed, with a minimum of 26 and a maximum of 46 cells assayed per experiment. Data were analyzed by one-way ANOVA followed by Tukey multiple comparison test for post hoc comparisons of means. Data are represented as mean ± SEM.

Discussion

While primary targets of many antidepressant drugs may be proteins mediating monoamine uptake or catabolism, there may also be postsynaptic actions, including increased cAMP production and a cascade of events resulting from sustained increase in cAMP or the activation of Gsα [1]. Recent results suggest that Gsα/adenylyl cyclase coupling may be one of the targets of antidepressant action, and that chronic treatment is required to observe this effect via altered membrane localization [10]. Previous studies demonstrated that chronic antidepressant treatment-induced increases in Gsα movement from a TX-100 insoluble lipid raft rich domain to a TX-100 soluble membrane domain with a concurrent increase in coupling to adenylyl cyclase [7]. These results are consistent with a study revealing that a number of antidepressant drugs concentrate in detergent insoluble membrane domains, like lipid rafts, subsequent to chronic treatment [11]. Preclinical, platelet, and postmortem brain data indicate that, in depression, the G-protein Gsα is more likely to reside in detergent insoluble lipid rafts [5, 8, 9]. There is evidence that other proteins potentially involved in a mechanism of depression also show altered lipid raft localization. SERT clustering in lymphocyte lipid rafts is altered in a rat model of depression, relative to controls. [26].

The model system used in these studies, C6 glioma cells, has been used for antidepressant studies by this and other labs [4, 7, 11, 12, 23, 25, 37]. Direct comparisons have demonstrated similar effects of prolonged drug treatment as seen in rats, except that 3 weeks in rats were identical to 3 days in cell culture [7]. While the presumption is that neurons are the dominant factor in both depression and antidepressant response, a role for glia (alone or in combination with neurons) is both likely and unresolved.

Chronic antidepressant treatment (in rats and cultured cells) moves Gsα out of TX-100 insoluble lipid raft rich domains and into a domain more amenable to an association with adenylyl cyclase [1, 4], this action may explain the sustained increases in cAMP signaling that accompany antidepressant treatment [2]. A previous study also provided evidence that Gsα signaling was inhibited in TX-100 insoluble lipid raft rich domains [24]. These data are consistent with the hypothesis that liberation of Gsα from these detergent insoluble inhibitory membrane domains by chronic treatment with antidepressants leads to increased coupling between signaling molecules in the detergent soluble non-raft membrane regions.

Lithium has been used as a treatment for bipolar disorder for over 50 years and has been effective in treating both acute manic and depressive symptoms, in addition to reducing their occurrence [27, 28]. A number of studies have demonstrated that lithium inhibits adenylyl cyclase activity [2932] and more specifically type V adenylyl cyclase [13]. Additionally, one potential mechanism of the antidepressant action of lithium is via inhibition of GSK-3 [14, 3335]. Recent evidence suggests that lithium has a membrane delimited dual function in inhibiting the activation of G-protein activated K+ channels [36]. This was suggested to be due to decreased affinity of the Gα subunit for the Gβγ subunit and diminished GDP-GTP exchange on the Gα subunit. What is unknown is whether chronic lithium treatment has any effect on the TX-100 insoluble lipid raft membrane localization of Gsα. In this vein, we set out to test whether lithium has the ability to alter Gsα localization in TX-100 insoluble lipid raft rich domains in an antidepressant manner. In addition, we used a structurally disparate mood stabilizer, valproic acid, to see if the effects were similar to lithium. In light of mood stabilizers’ effect on Gsα membrane disposition as determined by membrane fractionation and immunoblotting (Figure 1), these data are not sufficient to suggest that Gsα movement into lipid raft fractions and the decreased cellular production of cAMP associated with lithium or valproic acid treatment [13] are due to a decreased association of Gsα with adenylyl cyclase. Rather, it likely reflects a G protein-independent pathway of adenylyl cyclase regulation.

Curiously, while the above membrane fractionation data demonstrates movement of Gsα into lipid rafts upon treatment with lithium (Figure 1), we did not observe changes in membrane localization of Giα (Figure 2), nor FRAP recovery half-time of either GFP-Gsα or GFP-Giα (Figures 3, 4, 5). One interpretation of this is that the physical coupling between GFP-Gsα or GFP-Giα and adenylyl cyclase is not altered by lithium treatment, and that regulation of cellular cAMP levels by lithium is not effected by direct G protein regulation of adenylyl cyclase. The lack of effect by lithium on escitalopram-mediated Gsα FRAP (Figure 4) further suggests a different locus of action. These effects on Gsα raft domain localization may partially explain why lithium is not effective as a stand alone treatment for chronic depression.

The decreased mobility of GFP-Gsα subsequent to antidepressant treatment was initially surprising to us, as we had hypothesized the “liberation” of Gsα from lipid raft domains would produce the opposite effect, increased mobility of GFP-Gsα and faster recovery of fluorescence after photobleaching (i.e., decreased half-time to recovery). Instead, we found, consistently, that the movement of Gsα out of lipid rafts upon antidepressant treatment was associated with a slowing of GFP-Gsα mobility and an increase in half-time to recovery. We have proposed that this phenomenon is due to the increased association of Gsα, a smaller, peripheral membrane protein, with the larger, multi-transmembrane spanning adenylyl cyclase, which displays extremely slow lateral mobility and has been shown to increase physical association with Gsα subsequent to antidepressant treatment [37]. This “molecular signature” is typical of the many antidepressants tested by our lab (all major groups and several atypical compounds) [4, 7, 12, 23]. Other psychotropic molecules (e.g. antipsychotics and anxiolytics) did not alter mobility of Gsα [12].

Conclusions

Taken together, these observations demonstrate that the increased liberation of Gsα from the TX-100 insoluble lipid raft membrane domains is unique to antidepressant drugs. Additionally, this result appears to be Gsα specific as Giα does not have the same response. Previous studies have demonstrated this outcome as well [7, 8, 25]. These findings are supported by evidence suggesting that antidepressant drugs concentrate in raft-like plasma membrane domains [11], which may physically inhibit the localization of Gsα. Thus, it appears that one effect of antidepressants may be to exert their observed effects on cAMP signaling by liberating Gsα from TTX-100 resistant membrane domains, where it accumulates during the course of depression. Mood stabilizing drugs like lithium and valproate do not affect lipid raft membrane domains and do not increase association of Gsα with adenylyl cyclase. The alteration in the membrane localization of Gsα may one day prove to be one of a number of useful biomarkers for human depression and response to antidepressant therapy.

Methods

Cell culture and drug treatment

C6 cells in 150 cm2 flasks were cultured in Dulbecco’s modified Eagle’s medium, 4.5 g of glucose/l, 10% newborn calf serum (Hyclone Laboratories, Logan, UT, USA), 100 mg/ml penicillin and streptomycin at 37°C in humidified 5% CO2 atmosphere. The cells were treated with 10 µM escitalopram (gift from Lundbeck, Copenhagen, Denmark), 3 mM Lithium chloride (Sigma-Aldrich, St. Louis, MO, USA), 300 µM sodium valproic acid (Sigma-Aldrich, St. Louis, MO, USA), dissolved in water. Cells were treated for 3 days as 3 day treatment in culture has given results similar to chronic in vivo treatment for 3 weeks in rats [7]. The culture media and drug were changed daily. There was no apparent change in morphology of cells during the period of exposure to antidepressants.

Cell membrane, lipid raft, and detergent extract preparation

After treatment, cells were washed and harvested in ice-cold Phosphate-Buffered Saline (Mediatech Inc.). TX-100 insoluble membrane fractions were prepared as described by Li et al. [24], with slight modification [4]. In brief, two flasks of C6 cells were scraped into 0.75 ml of HEPES buffer (10 mM HEPES, pH 7.5, 150 mM NaCl, 1 mM DTT, and protease inhibitors) containing 1% TX-100. Cells were homogenized with 10 strokes of a Potter–Elvehjem homogenizer. The homogenate was then mixed with an equal volume of 80% sucrose prepared in HEPES buffer to form 40% sucrose and loaded at the bottom of an ultracentrifuge tube. A step gradient was generated by sequentially layering 30, 15, and 5% sucrose over the homogenate. Gradients were centrifuged at 2,00,000g for 20 h in an SW55 rotor (Beckman, Palo Alto, CA, USA). Two or three opaque bands were confined between the 15 and 30% sucrose layers. These bands were removed from the tube, diluted threefold with HEPES buffer, and pelleted in a microcentrifuge at 16,000g. The pellet was resuspended in HEPES buffer and subsequently analyzed by immunoblotting.

SDS-PAGE and western blotting

Samples were assayed for protein via a bicinchoninic acid assay (Pierce Research, Rockford, IL, USA) and equal quantities were loaded onto Stain-Free acrylamide gel for SDS-PAGE (Bio-Rad, Hercules, CA, USA). Gels were transferred to Immobilon-P PVDF membranes (EMD Millipore, Billerica, MA, USA) for western blotting. The membranes were blocked with 5% nonfat dry milk diluted in TBS-T (10 mM Tris–HCl, 159 mM NaCl, and 0.1% Tween 20, pH 7.4) for 1 h. Following the blocking step, membranes were washed with Tris-buffered saline/Tween 20 and then incubated with an anti-Gsα monoclonal antibody (NeuroMab clone N192/12, Davis, CA, USA, catalog # 75-211, RRID #AB_2315846), anti-Gsα polyclonal antibody (EMD Millipore, Billerica, MA, USA, catalog # 06-237, RRID # AB_310078), or anti-Giα polyclonal antibody (Thermo Scientific, Rockford IL, USA, catalog # PA1-1000, RRID # AB_2232440) overnight at 4°C. Membranes were washed with TBS-T and incubated with a secondary antibody [HRP-linked anti-mouse antibody IgG F(ab′)2 or HRP-linked anti-rabbit antibody IgG F(ab′)2] (Jackson ImmunoResearch, West Grove, PA, USA, catalog # 115-036-072 for mouse, RRID # AB_2338525 and catalog # 111-036-047 for rabbit, RRID #AB_2337945) for 1 h at room temperature, washed, and developed using ECL Luminata Forte chemiluminescent reagent (Millipore, Billerica, MA, USA). Blots were imaged using Chemidoc computerized densitometer (Bio-Rad, Hercules, CA, USA) and quantified by ImageLab 3.0 software (Bio-Rad, Hercules, CA, USA). In all experiments, the original gels are visualized using BioRad’s Stainfree technology to verify protein loading.

Fluorescence recovery after photobleaching (FRAP)

C6 glioma cells were transfected with either GFP-Gsα or GFP-Giα1 and cells stably expressing the fluorescent construct were selected with G418 followed by fluorescence activated cell sorting and isolation of clones [12]. Cells were plated on glass microscopy dishes and treated with lithium chloride (3 mM), valproic acid (300 μM), escitalopram (10 μM), or escitalopram plus lithium combined (same doses as individual treatments with these compounds) for 3 days. On the day of imaging, drug was washed out for 1 h prior to imaging and media was replaced with low serum (2.5% NCS) phenol red-free DMEM to reduce background fluorescence. Temperature was maintained at 37°C using a heated stage assembly during imaging, which utilized a Zeiss LSM 710 at 512 × 512 resolution using an open pinhole to maximize signal but minimizing photobleaching. 150 data points approximately 300 ms apart, including 10 pre-bleach values, were measured for each cell. Zeiss Zen software was used to calculate FRAP recovery half-time utilizing a one-phase association fit, correcting for total photobleaching of the analyzed regions.

Statistical analysis

All of the experiments were performed at least three times. Data were analyzed for statistical significance using a one-way ANOVA followed by Tukey test for post hoc multiple comparisons of means. Values of p < 0.05 were taken to indicate significance.

Declarations

Authors’ contributions

RD and MR participated in the conceptualization and organization of the manuscript. JS, AC, and AJ participated in data collection. All of the authors participated in data analysis. RD, MR, JS, and AC participated in manuscript preparation while RD, MR, JS participated in the final review of the manuscript. AJ died prior to the writing of the manuscript. All authors read and approved the final manuscript.

Acknowledgements

The authors would like to thank Lundbeck Research USA, for their financial support for this study and Athanasia Koutsouris for her technical assistance. This study was also supported, in part, by a Veterans Administration Merit Award (BX1149) and by T32 MH067631.

Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Authors’ Affiliations

(1)
Departments of Physiology and Biophysics, College of Medicine, University of Illinois at Chicago
(2)
The Psychiatric Institute, College of Medicine, University of Illinois at Chicago
(3)
Basic and Health Science Department, Illinois College of Optometry

References

  1. Donati RJ, Rasenick MM (2003) G protein signaling and the molecular basis of antidepressant action. Life Sci 73:1–17PubMedView ArticleGoogle Scholar
  2. Malberg JE, Blendy JA (2005) Antidepressant action: to the nucleus and beyond. Trends Pharm Sci 26:631–638PubMedView ArticleGoogle Scholar
  3. Zanotti-Fregonara P, Hines CS, Zoghbi SS, Liow JS, Zhang Y, Pike VW et al (2012) Innis RB population-based input function and image-derived input function for [(1)(1)C](R)-rolipram PET imaging: methodology, validation and application to the study of major depressive disorder. Neuroimage 63:1532–1541PubMed CentralPubMedView ArticleGoogle Scholar
  4. Donati RJ, Rasenick MM (2005) Chronic antidepressant treatment prevents accumulation of gsalpha in cholesterol-rich, cytoskeletal-associated, plasma membrane domains (lipid rafts). Neuropsychopharm 30:1238–1245Google Scholar
  5. Allen JA, Halverson-Tamboli RA, Rasenick MM (2007) Lipid raft microdomains and neurotransmitter signalling. Nat Rev Neurosci 8:128–140PubMedView ArticleGoogle Scholar
  6. Allen JA, Yu JZ, Dave RH, Bhatnagar A, Roth BL, Rasenick MM (2009) Caveolin-1 and lipid microdomains regulate Gs trafficking and attenuate Gs/adenylyl cyclase signaling. Mol Pharm 76:1082–1093View ArticleGoogle Scholar
  7. Toki S, Donati RJ, Rasenick MM (1999) Treatment of C6 glioma cells and rats with antidepressant drugs increases the detergent extraction of G(s alpha) from plasma membrane. J Neurochem 73:1114–1120PubMedView ArticleGoogle Scholar
  8. Donati RJ, Dwivedi Y, Roberts RC, Conley RR, Pandey GN, Rasenick MM (2008) Postmortem brain tissue of depressed suicides reveals increased Gs alpha localization in lipid raft domains where it is less likely to activate adenylyl cyclase. J Neurosci 28:3042–3050PubMedView ArticleGoogle Scholar
  9. Hines LM, Tabakoff B (2005) Platelet adenylyl cyclase activity: a biological marker for major depression and recent drug use. Biol Psych 58:955–962View ArticleGoogle Scholar
  10. Mooney JJ, Samson JA, McHale NL, Pappalarado KM, Alpert JE, Schildkraut JJ (2013) Increased Gsalpha within blood cell membrane lipid microdomains in some depressive disorders: an exploratory study. J Psych Res 47:706–711View ArticleGoogle Scholar
  11. Eisensamer B, Uhr M, Meyr S, Gimpl G, Deiml T, Rammes G et al (2005) Antidepressants and antipsychotic drugs colocalize with 5-HT3 receptors in raft-like domains. J Neurosci 25:10198–10206PubMedView ArticleGoogle Scholar
  12. Czysz AH, Schappi JM, Rasenick MM (2015) Lateral diffusion of Galpha in the plasma membrane is decreased after chronic but not acute antidepressant treatment: role of lipid raft and non-raft membrane microdomains. Neuropsychopharm 40(3):766–773View ArticleGoogle Scholar
  13. Mann L, Heldman E, Bersudsky Y, Vatner SF, Ishikawa Y, Almog O et al (2009) Inhibition of specific adenylyl cyclase isoforms by lithium and carbamazepine, but not valproate, may be related to their antidepressant effect. Bipolar Disord 11:885–896PubMedView ArticleGoogle Scholar
  14. Jope RS (1999) Anti-bipolar therapy: mechanism of action of lithium. Mol Psych 4:117–128View ArticleGoogle Scholar
  15. Mann L, Heldman E, Shaltiel G, Belmaker RH, Agam G (2008) Lithium preferentially inhibits adenylyl cyclase V and VII isoforms. Int J Neuropsychopharm 11:533–539View ArticleGoogle Scholar
  16. Marmol F (2008) Lithium: bipolar disorder and neurodegenerative diseases Possible cellular mechanisms of the therapeutic effects of lithium. Prog Neuropsychopharmacol Biol Psych 32:1761–1771View ArticleGoogle Scholar
  17. Machado-Vieira R, Manji HK, Zarate CA Jr (2009) The role of lithium in the treatment of bipolar disorder: convergent evidence for neurotrophic effects as a unifying hypothesis. Bipolar Disord 11(Suppl 2):92–109PubMed CentralPubMedView ArticleGoogle Scholar
  18. Beaulieu JM, Gainetdinov RR, Caron MG (2009) Akt/GSK3 signaling in the action of psychotropic drugs. Ann Rev Pharm Tox 49:327–347View ArticleGoogle Scholar
  19. Carli M, Afkhami-Dastjerdian S, Reader TA (1997) Effects of a chronic lithium treatment on cortical serotonin uptake sites and 5-HT1A receptors. Neurochem Res 22:427–435PubMedView ArticleGoogle Scholar
  20. Quiroz JA, Machado-Vieira R, Zarate CA Jr, Manji HK (2010) Novel insights into lithium’s mechanism of action: neurotrophic and neuroprotective effects. Neuropsychobiol 62:50–60View ArticleGoogle Scholar
  21. Li X, Jope RS (2010) Is glycogen synthase kinase-3 a central modulator in mood regulation? Neuropsychopharm 35:2143–2154View ArticleGoogle Scholar
  22. Colin SF, Chang HC, Mollner S, Pfeuffer T, Reed RR, Duman RS et al (1991) Chronic lithium regulates the expression of adenylate cyclase and Gi-protein alpha subunit in rat cerebral cortex. Proc Nat Acad Sci 88:10634–10637PubMed CentralPubMedView ArticleGoogle Scholar
  23. Zhang L, Rasenick MM (2010) Chronic treatment with escitalopram but not R-citalopram translocates Galpha(s) from lipid raft domains and potentiates adenylyl cyclase: a 5-hydroxytryptamine transporter-independent action of this antidepressant compound. J Pharm Exp Ther 332:977–984View ArticleGoogle Scholar
  24. Li S, Okamoto T, Chun M, Sargiacomo M, Casanova JE, Hansen SH et al (1995) Evidence for a regulated interaction between heterotrimeric G proteins and caveolin. J Biol Chem 270:15693–15701PubMedView ArticleGoogle Scholar
  25. Donati RJ, Thukral C, Rasenick MM (2001) Chronic treatment of C6 glioma cells with antidepressant drugs results in a redistribution of Gsalpha. Mol Pharm 59:1426–1432Google Scholar
  26. Rivera-Baltanas T, Olivares JM, Calado-Otero M, Kalynchuk LE, Martinez-Villamarin JR, Caruncho HJ (2012) Serotonin transporter clustering in blood lymphocytes as a putative biomarker of therapeutic efficacy in major depressive disorder. J Affect Disord 137:46–55PubMedView ArticleGoogle Scholar
  27. Goodwin FK, Fireman B, Simon GE, Hunkeler EM, Lee J, Revicki D (2003) Suicide risk in bipolar disorder during treatment with lithium and divalproex. JAMA 290:1467–1473PubMedView ArticleGoogle Scholar
  28. Baldessarini RJ, Tondo L, Davis P, Pompili M, Goodwin FK, Hennen J (2006) Decreased risk of suicides and attempts during long-term lithium treatment: a meta-analytic review. Bipolar Disord 8:625–639PubMedView ArticleGoogle Scholar
  29. Dousa T, Hechter O (1970) Lithium and brain adenyl cyclase. Lancet 1:834–835PubMedGoogle Scholar
  30. Ebstein R, Belmaker R, Grunhaus L, Rimon R (1976) Lithium inhibition of adrenaline-stimulated adenylate cyclase in humans. Nature 259:411–413PubMedView ArticleGoogle Scholar
  31. Ebstein RP, Hermoni M, Belmaker RH (1980) The effect of lithium on noradrenaline-induced cyclic AMP accumulation in rat brain: inhibition after chronic treatment and absence of supersensitivity. J Pharm Exp Ther 213:161–167Google Scholar
  32. Mork A, Geisler A (1989) The effects of lithium in vitro and ex vivo on adenylate cyclase in brain are exerted by distinct mechanisms. Neuropharm 28:307–311View ArticleGoogle Scholar
  33. Jope RS (1999) A bimodal model of the mechanism of action of lithium. Mol Psych 4:21–25View ArticleGoogle Scholar
  34. Jope RS, Bijur GN (2002) Mood stabilizers, glycogen synthase kinase-3beta and cell survival. Mol Psych 7(Suppl 1):S35–S45View ArticleGoogle Scholar
  35. Gould TD, Chen G, Manji HK (2004) In vivo evidence in the brain for lithium inhibition of glycogen synthase kinase-3. Neuropsychopharm 29:32–38View ArticleGoogle Scholar
  36. Farhy Tselnicker I, Tsemakhovich V, Rishal I, Kahanovitch U, Dessauer CW, Dascal N (2014) Dual regulation of G proteins and the G-protein-activated K+ channels by lithium. Proc Nat Acad Sci 111:5018–5023PubMed CentralPubMedView ArticleGoogle Scholar
  37. Chen J, Rasenick MM (1995) Chronic treatment of C6 glioma cells with antidepressant drugs increases functional coupling between a G protein (Gs) and adenylyl cyclase. J Neurochem 64:724–732PubMedView ArticleGoogle Scholar

Copyright

© Donati et al. 2015

Advertisement