Mitochondrial dysfunction in an animal model of diabetic neuropathy is associated with a reduction of neurosteroid synthesis.

Background : Recent work in a model of diabetic neuropathy revealed that layer 2/3 cortical pyramidal neurones of the pain pathway exhibited reduced endogenous neurosteroid modulation of the GABA AR and exogenously applied neurosteroids had an exaggerated impact. It is postulated that this is related to reduced precursor synthesis, due to mitochondrial dysfunction in diabetic neuropathy. Benzodiazepines are also known to activate neurosteroidogenesis by binding to mitochondrial translocator protein (TSPO). This study explored the differential effect of diazepam on GABA AR modulation via neurosteroidogenesis in diabetic and wild type (WT) mice. Methods : Whole-cell patch-clamp technique was used on slices of neural tissue. Electrophysiological recordings were obtained from layer 2/3 cortical pyramidal neurons of the pain pathway from mice with type-II diabetic neuropathy ( ob/ob) and WT controls aged 60-80 days. Results : There was a key difference in the response of the WT and ob/ob cortical neurons to simultaneous incubation with diazepam and flumazenil. In contrast, diazepam and the 5a-reductase inhibitor finasteride, individually or in combination, produced the same response in both strains. Conclusions : The exaggerated effect of diazepam on GABAergic inhibitory tone in the ob/ob, despite the presence of the GABA AR benzodiazepine antagonist flumazenil is likely observed due to physiological upregulation of key neurosteroidogenic enzymes in response to the reduced pregnenolone synthesis by the mitochondria. By increasing pregnenolone via TSPO activation, it is possible to promote enhanced neurosteroidogenesis and increase GABAergic inhibitory tone via an alternate route. In diabetic neuropathy, mitochondrial dysfunction may play an important role. Enhancing the GABAergic neurosteroid tone could be of potential therapeutic benefit.


Amendments from Version 1
Changes have been made in response to comments made by the reviewers. The title has been made more specific to the data presented. In the Abstract, "neuropathic pain" has been changed to "neuropathy" as the concept of pain is specific to humans, in contrast to nociception, which applies to mice and other species including humans. The Discussion has been edited to facilitate the reader's understanding by putting the data in the context of recent work (Humble, 2016a). Similarly, Figure 1 and its legend have been edited in order to highlight action sites of diazepam, flumazenil and finasteride and therefore enhance readability. We have removed the mention of Dr Ladas from the Introduction Diabetic neuropathy is a common cause of painful neuropathy, and treatment is often suboptimal because the underlying aetiology is poorly understood. Peripheral and central sensitisation are implicated in the development of neuropathic pain with neuroplasticity occurring at multiple levels of the pain pathway (Harvey & Dickenson, 2008). GABAergic neurones at all levels of the pain pathway have a vital role in the transmission of painful stimuli in the perception of pain itself (D'Mello & Dickenson, 2008). Endogenous and exogenous neurosteroids may act as potent positive allosteric modulators of GABA A receptors (GABA A Rs) and consequently exhibit analgesic, anxiolytic, anticonvulsant, and sedative properties (D'Hulst et al., 2009).
Within inhibitory synapses, the presynaptic fusion of a single vesicle releases the inhibitory neurotransmitter GABA to activate synaptic GABA A Rs. Under voltage-clamp conditions this causes a miniature inhibitory postsynaptic current (mIPSC). Drugs that enhance GABA A R function cause a prolongation of the mIPSC decay phase. Recent work in a model of type-II diabetic neuropathy (ob/ob) revealed that layer 2/3 cortical pyramidal neurones of the pain pathway exhibited reduced endogenous neurosteroid modulation of the GABA A R, and exogenously applied neurosteroids had an exaggerated impact (Humble, 2016a). The mechanism responsible appeared unrelated to GABA A R sensitivity, but instead was associated with a reduction of neurosteroid precursors, such as pregnenolone, which is metabolised sequentially to the active compound allopregnanolone ( Figure 1) (Humble, 2013;Humble, 2016a). Pregnenolone is synthesised in the mitochondrion from its precursor cholesterol by the side chain cleavage enzyme P450 located in the inner mitochondrial membrane (Do Rego et al., 2009;Schumacher et al., 2012), and it is postulated that diabetic neuropathy may be associated with a reduction in mitochondrial activity. Cholesterol is translocated across the mitochondrial membrane by the 18 kDa translocator protein (TSPO) in a coordinated fashion with the steroidogenic acute regulatory (StAR) protein (Do Rego et al., 2009;Gatliff & Campanella, 2016;Rupprecht et al., 2010;Stocco et al., 2017; Figure 1).
The present study explored the impact of the benzodiazepine diazepam, a positive allosteric modulator of the GABA A R (D'Hulst et al., 2009), on GABA A R modulation via neurosteroidogenesis in diabetic and wild type (WT) mice. Benzodiazepines are also known to activate neurosteroidogenesis by binding to TSPO (Rupprecht et al., 2010;Tokuda et al., 2010).

Methods
The methods are identical to those published by the same author previously (Humble, 2016a), with the exception of the drugs diazepam and flumazenil, which were not used in the previous study. Diazepam and flumazenil were purchased (Tocris, Bristol UK) and prepared as concentrated stock solutions in dimethyl sulfoxide before being added to the artificial extracellular solution as per the previous study (Humble, 2016a).

Figure 1. Modulation of the GABA A R by endogenous neurosteroids.
Cholesterol is taken through the mitochondrial membrane by the translocator protein (TSPO) where it is converted to pregnenolone by the cytochrome P450 side chain cleavage enzyme. Pregnenolone is converted to progesterone by 3β-hydroxysteroid dehydrogenase (3β-HSD), which is in turn reduced to dihydroxyprogesterone by 5α-reductase (5α-R). Dihydroxyprogesterone is converted to allopregnanolone by 3α-hydroxysteroid dehydrogenase (3α-HSD). Postsynaptic GABA A Rs are activated by GABA that has been released from vesicles in the presynaptic nerve terminal. GABA induces a conformational change of the GABA A R, opening its central channel and thereby allowing the passage of chloride ions and the subsequent generation of miniature inhibitory postsynaptic currents (mIPSCs). The negative chloride ions induce hyperpolarisation of the neuronal membrane, which mediates neuronal inhibition. Neurosteroids, such as the active compound allopregnanolone, modulate GABA A R function and facilitate inhibition of the neuronal membrane. (Humble, 2013). The translocation of cholesterol into the mitochondria by TSPO is the first rate-limiting step and is enhanced by the presence of specific ligands such as diazepam. Thus diazepam may enhance GABA A R modulation by binding to the GABA A R directly and separately by increased neurosteroidogenesis. This modulation may be selectively inhibited at the GABA A R itself by the antagonist flumazenil and separately by the 5α-R inhibitor finasteride, which inhibits neurosteroidogenisis.

Discussion
Layer 2/3 cortical neurones from mature type-II diabetic ob/ob are known to have a reduced endogenous pregnane-derived neurosteroid tone in comparison to strain matched WT controls (Humble, 2016a). The present data indicate that by promoting the uptake of pregnenolone's precursor cholesterol by the mitochondria, via TSPO, diazepam may rescue the reduced neurosteroid tone. The restored neurosteroid tone could re-establish GABA A Rmediated neuro-inhibitory tone in cases of neuropathic hypersensitivity. With specific reference to these data, the key result is the difference in response of the WT and ob/ob to simultaneous incubation with diazepam and flumazenil. In contrast, diazepam and the 5α-reductase inhibitor finasteride individually or in combination produced the same response in both WT and ob/ob. This may be interpreted as follows: in the WT, the primary effect of diazepam incubation is direct allosteric modulation of the GABA A R, with negligible contribution from neurosteroidogenesis via mitochondrial TSPO activation. In comparison, diazepam has an exaggerated effect on GABAergic inhibitory tone in the ob/ob, despite the presence of the GABA A R benzodiazepine antagonist flumazenil. This effect is likely observed due to physiological upregulation of the key rate-limiting enzymes involved in neurosteroidogenesis in response to the reduced pregnenolone synthesis by the mitochondria (Figure 1; Humble, 2016a). Thus by increasing the availability of the neurosteroid precursor pregnenolone via TSPO activation, it is possible to promote enhanced neurosteroidogenesis and thereby increase GABAergic inhibitory tone via an alternate route. Benzodiazepines modulate the GABA A R by binding to the α-γ subunit interface (D'Hulst et al., 2009), while neurosteroids bind the GABA A R from a cavity within the α-subunit domain and modulate it directly via the α-β subunit interface (Hosie et al., 2006). There have already been a number of other studies of ligands for the mitochondrial TSPO, as this is a promising target (Gatliff & Campanella, 2016;Giatti et al., 2009;Papadopoulos & Lecanu, 2009;Rupprecht et al., 2010;Zhang et al., 2016). Considering the findings in this paper (B) Superimposed exemplar averaged GABA A R-mediated mIPSCs acquired from a representative ob/ob cortical neurone and from equivalent neurones after ~2 hours pre-incubation of the brain slice with diazepam (1μM), flumazenil (10 μM) and finasteride (50 μM). (C) Histogram illustrating that flumazenil is able to prevent the effect of diazepam to prolong the duration of the GABA A R-mediated mIPSC in WT cortical neurones, but only has a partial efficacy in ob/ob cortical neurones (τ w in ms; one-way ANOVA, P >0.05; Post hoc Newman Keul's test). The persisting effect of diazepam in the presence of flumazenil in the ob/ob mice could be prevented by the presence of the 5α-reductase enzyme inhibitor finasteride (τ w in ms; one-way ANOVA, P >0.05; Post hoc Newman Keul's test). Ctrl = control; Finast = finasteride; Diaz = diazepam; Flumaz = flumazenil.
alongside previous work (Humble, 2016a) it appears that mitochondrial dysfunction may play an important role in the development of type 2 diabetic neuropathy. In this context, it follows that enhancing the GABAergic neurosteroid tone directly or indirectly could be of potential therapeutic benefit for diabetic neuropathic pain and hypersensitivity.

Data availability
Open Science Framework: Dataset of 'Neurosteroids are reduced in diabetic neuropathy and may be associated with the development of neuropathic pain', doi: 10.17605/osf.io/bk3tw (Humble, 2016b). Raw data for the present study can be found in Diazepam.zip.
Please refer to (Humble, 2016a) for details of standard software used for data analysis.

Author contributions
Dr Stephen Humble is responsible for this all work, including planning the experiments, performing the experiments and writing the paper. Prof Hales and Lambert, and Dr Belelli assisted Dr Humble with regards to the planning of some experiments.
However, after discussion it was decided that their contributions merited being listed in the Acknowledgements section rather than as co-authors.

Competing interests
No competing interests were disclosed.

Grant information
This research was supported by the Wellcome Trust (090667).
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. ). The current article could be considered as an extended 10.12688/f1000research.9034.1 version providing further clarification on involved mechanisms. Therefore to fully understand the presented results, I recommend reading first the previous article.

Open Peer Review
However, in order to reach a broader leadership illustration and description could be improved. Figure 1 could be improved by highlighting the difference with Humble's 2016 paper; action sites of Diazepam, Flumazenil and Finasterid could be indicated. Likewise, results description focuses on difference between both mice strains without fully presenting individual results (Finast, Diaz, Flumaz…). This required a strong background in the field.
My last comment concerns the primary effect of diazepam on GABA R. The experiment consists in a 2 h incubation and it has been shown ( in other brain structure, Zeitler Eur J Neurosci, 2016) that in vivo et al.
Diazepam acts within 15-30 min then is relayed by neurosteroidogenesis. Therefore, it is difficult to A incubation and it has been shown ( in other brain structure, Zeitler Eur J Neurosci, 2016) that in vivo et al.
Diazepam acts within 15-30 min then is relayed by neurosteroidogenesis. Therefore, it is difficult to conclude on the primary effect alone.

Is the work clearly and accurately presented and does it cite the current literature? Yes
Is the study design appropriate and is the work technically sound? Yes

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility? Yes

Are the conclusions drawn adequately supported by the results? Yes
No competing interests were disclosed. Competing Interests:

I have read this submission. I believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.
Author Response 14 Jun 2017 , Imperial College London, UK

Stephen Humble
Dear Prof Darbon, Thank you for your comments and for taking the time to review the paper. I agree that the illustration and description could be improved and I will do so as per your suggestions. I agree that the brief background description makes the research harder for a broad readership-this is as a result of the article's word limits.
Regarding the 2 hour incubation period for diazepam. This issue was considered at length during the experimental design stage and this is a relevant concern. It is of course true that in vivo diazepam acts within 15-30 minutes and that propofol and etomidate may act within in vivo seconds when given intravenously. However, previous work (Benkwitz 2007;Gredell et al., et al., 2004) has shown that in brain slice preparations the duration of onset may be much longer in lipophilic compounds such as these because the drug has to diffuse through the slice to reach the GABA receptor on the neurone located within the slice. This contrasts to rapid drug delivery via an intact circulation, which may deliver the drug directly to the relevant neurone within in vivo seconds to minutes. Taking all this into account it was decided to use a 2 hour incubation period in order to maximise the impact of diazepam and highlight the difference between the diabetic neurones and the wild type. i.e. that diazepam in the presence of flumazenil (benzodiazepine receptor antagonist) had a greater response in the diabetic neurones, which appears to be mediated via upregulated neurosteroidogenesis and could be blocked by finasteride. Finally, I think A mediated via upregulated neurosteroidogenesis and could be blocked by finasteride. Finally, I think that these results are not inconsistent with the Zeitler , 2016 paper. et al.

Yours sincerely
Stephen My major concern with this study is that title and discussion are not relevant to the data presented. There is no justification to link this study to neuropathic pain or painful diabetic neuropathy. Perhaps more appropriate title would be: "Mitochondrial dysfunction in an animal model of Type 2 diabetes may be involved reduction in neurosteroid synthesis "

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility? Yes benefit.
New: Considering the findings in this paper alongside previous work (Humble, 2016a) it appears that mitochondrial dysfunction may play an important role in the development of type 2 diabetic neuropathy. In this context, it follows that enhancing the GABAergic neurosteroid tone directly or indirectly could be of potential therapeutic benefit for diabetic neuropathic pain and hypersensitivity.

Yours sincerely
Stephen Humble No competing interests were disclosed.

Competing Interests:
The benefits of publishing with F1000Research: Your article is published within days, with no editorial bias You can publish traditional articles, null/negative results, case reports, data notes and more The peer review process is transparent and collaborative Your article is indexed in PubMed after passing peer review Dedicated customer support at every stage For pre-submission enquiries, contact research@f1000.com