The α-lipoic acid improves hepatic metabolic dysfunctions in Acute Intermittent Porphyria: a proof-of-concept study

Background: Acute intermittent porphyria (AIP) is caused by haploinsufficiency of porphobilinogen deaminase (PBGD) enzymatic activity. Acute attacks occur in response to fasting and alterations in glucose metabolism, insulin resistance and mitochondrial turnover may be involved in AIP pathophysiology. Therefore, we investigated the metabolic pathways in PBGDsilenced hepatocytes and assessed the efficacy of an insulin-mimic, the α-lipoic acid (α-LA) as a potential therapeutic strategy. Methods: HepG2 cells were transfected with a siRNA targeting PBGD (siPBGD). Cells were cultured with low glucose concentration to mimic fasting and exposed to α-LA alone or with glucose. Results: At baseline, siPBGD cells showed lower expression of genes involved in glycolysis and mitochondrial dynamics along with reduced total ATP levels. Fasting further unbalanced glycolysis by inducing ATP shortage in siPBGD cells and activated DRP1, which mediates mitochondrial separation. Consistently, siPBGD cells in fasted state showed the lowest protein levels of Complex IV which belong to the oxidative phosphorylation (OXPHOS) machinery. α-LA upregulated glycolysis and prompted ATP synthesis and triglyceride secretion, thus possibly providing energy fuels to siPBGD cells by improving glucose utilization. Finally, siPBGD exposed to α-LA plus glucose raised mitochondrial dynamics, OXPHOS activity and energy production. Conclusions: α-LA-based therapy may ameliorate glucose metabolism and mitochondrial dysfunctions in siPBGD hepatocytes.


Introduction
Acute Intermittent Porphyria (AIP) is the most common and severe form of acute hepatic porphyria (AHP), a heterogeneous group of hereditary metabolic disorders characterized by defects of heme biosynthesis. Clinical manifestations include acute neurovisceral attacks (abdominal pain, nausea, constipation, vomiting) occurring in the insulin-mimic alpha-lipoic acid (α-LA), which has been already tested for the treatment of metabolic disorders as type 2 diabetes, turning out to be effective and safe at improving glucose handling, insulin sensitivity and hepatic inflammation [19][20][21].

RNA interference
HepG2 human hepatoma cells, which represent the most used in vitro model to study liver metabolism and related disorders, were transiently transfected for 48 hours by pooling 3 different target-specific siRNA oligo duplexes (MyBioSource, Inc., USA) of human HMBS gene (siPBGD) at final concentration of 10 µM. Cyclophilin B (10 µM) was used as scramble negative control (Horizon Discovery, Italy). To mimic fasting, both scramble and siPBGD cells were cultured in Dulbecco's Modified Eagle's Medium (DMEM) with low glucose concentration (1.125 mg/L) for 6 hours. Cultured medium of siPBGD cells was supplemented with 0.5 mM α-LA alone (siPBGD+ α-LA) or in combination with 0.33 mM glucose (α-LA+Gluc) for 24 hours before inducing fasting. Treatments were freshly prepared and administered when appropriate. The potential efficacy of α-LA alone and α-LA+Gluc were compared to 0.33 mM glucose (siPBGD+Gluc) [17].
Evaluation of PBGD enzymatic activity siPBGD and scramble cells were lysed in 0.1 M Tris-HCl buffer (pH 8.0) containing 0.2% Triton-X 100. 50 µl cell lysates were incubated with 200 µl 0.1 M Tris-HCl and 25µl of 1 mM PBG for 1 hour at 37°C, in order to evaluate whether PBGD enzyme could convert PBG substrate into uroporphyrin according to Hsiao et al. procedure [22]. After incubation, the enzymatic reaction was stopped through 10% trichloroacetic acid and centrifuged at 13.000 rpm for 10 minutes. The uroporphyrin fluorescence emission (Ex/Em = 405/655nm) was measured by a spectrofluorometer (Shimadzu Corporation, Japan) and uroporphyrin concentration was determined using uroporphyrin I standard calibrator (Sigma). The PBGD activity (pmol Uro/h) was normalized to the total amount of proteins (mg) and expressed as percentage of residual activity for each condition.

Gene expression analysis
RNA was extracted from cell cultures using Trizol reagent (Life Technologies-Ther-moFisher Scientific, Carlsbad, U.S.A). 1µg of total RNA was retro-transcribed with VILO random hexamers synthesis system (Life Technologies-ThermoFisher Scientific, Carlsbad, U.S.A). Quantitative real time PCR (qRT-PCR) was performed by an ABI 7500 fast thermocycler (Life Technologies), using the TaqMan Universal PCR Master Mix (Life Technologies, Carlsbad, CA) and TaqMan probes ( Table S1). The SYBR Green chemistry (Fast SYBR Green Master Mix; Life Technologies) was used for cDNA amplification through human primers (Table S2). All reactions were delivered in triplicate. Data were normalized to β-actin gene expression and results were expressed as arbitrary units (AU) or fold increase as indicated in bar graphs.

Western Blot Analysis
Total protein lysates were extracted from cell cultures, using RIPA buffer containing 1 mmol/L Na-orthovanadate, 200 mmol/L phenylmethyl sulfonyl fluoride and 0.02 µg/µL aprotinin. Samples were pooled prior electrophoretic separation and all reactions were performed in duplicate. Then equal amounts of proteins (50 µg) were separated by SDS-PAGE, transferred electrophoretically to nitrocellulose membrane (BioRad, Hercules, CA) and incubated with specific antibodies overnight. At least, three independent lots of freshly extracted proteins were used for experiments. Antibodies and concentration used are listed in Table S3.

Statistical analysis
Differences between two groups were calculated by two -way ANOVA, followed by post hoc t-test (two-tailed). Differences among multiple groups were analyzed by two-way ANOVA followed by two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli, which corrects for the number of comparisons and controls the False Discovery Rate (FDR). Adjusted p values <0.05 were considered statistically significant. Statistical analyses were performed using JMP 16.0 (SAS, Cary, NC) and Prism software (version 9.1, GraphPad Software).

α-Lipoic acid improved heme production in siPBGD cells
At baseline, PBGD silencing reduced both PBGD mRNA levels by 60% (p<0.0001 at ANOVA; adjusted p<0.0001 vs Scramble; Figure 1A) and its enzymatic activity by 42% (p=0.0001 at ANOVA; adjusted p=0.001 vs Scramble; Figure 1B). siPBGD cells showed lower ALAS1 expression (p=0.003 at ANOVA; p=0.01 vs Scramble; Figure 1C) and intracellular heme content compared to the scramble probably due to the delay in the heme biosynthetic pathway induced by PBGD downregulation (p<0.0001 at ANOVA; adjusted p<0.0001 vs Scramble; Figure 1D).
Fasting further reduced PBGD activity (p=0.0001 at ANOVA; adjusted p=0.04 vs siPBGD untreated (NT) and vs fasted Scramble; Figure 1B To assess the potential efficacy of α-LA as prophylactic treatment, we exposed siPBGD cells to α-LA 0.5 mM for 24 hours before fasting and its efficacy was compared to glucose supplementation (0.33 mM) at the same timing. Additionally, we explored whether α-LA could enhance glucose efficacy by evaluating the potential synergism between the two treatments.

α-Lipoic acid stimulates glucose utilization and provides energy supplies during fasting
Alterations in glucose metabolism might precipitate the AIP acute attack and compromise the efficacy of glucose therapy. However, metabolic aberrancies occurring in hepatocytes at baseline have not been described yet and whether they could be refrained through an insulin-mimic agent have not been reported in previous studies.
As expected, glucose administration regulated the expression of GCK (p=0.004 at ANOVA; adjusted p=0.02 vs siPBGD fasted cells, Figure 2E), PFK-L (p=0.0002 at ANOVA, adjusted p=0.0001 vs siPBGD fasted cells; Figure 2F) and PK (p<0.0001 at ANOVA, adjusted p<0.0001 vs siPBGD fasted cells; Figure 2G) although it resulted ineffective at promoting ATP production, whose levels matched with those produced in siPBGD cells at baseline (p<0.0001 at ANOVA; adjusted p=0.0009 vs siPBGD fasted cells; Figure 2H) thereby supporting that alterations of glucose metabolism occur in presence of PBGD downregulation.

α-Lipoic acid recovered mitobiogenesis: the dual role of PGC1α
Mitochondrial dysfunctions, in terms of bioenergetic failure, have been described in AIP mice with overt symptomatology [11,12], but alterations of mitochondrial dynamics need to be further elucidated.
The mRNA and protein levels of PCG1α, master regulator of mitobiogenesis, were induced by fasting in both scramble and siPBGD cells (p<0.0001 at ANOVA, p=0.008 vs siPBGD-NT, Figure 3A-B) accompanied by downregulation of Optical Atrophy 1 (OPA1), which mediates fusion of mitochondrial inner membranes, and Mitofusin 2 (MFN2), which joins mitochondrial outer membranes (p=0.01 at ANOVA, p<0.05 vs Scramble-NT and vs siPBGD-NT, Figure S1A-B). Conversely, we found that dynamin 1-like protein (DRP1), which regulates mitochondrial separation, markedly localized in the cytoplasm of siPBGD cells in fasting condition ( Figure 3C). Since DRP1 activation is usually associated to low OXPHOS capacity and energy shortfall, we assessed the expression of Complex IV of the respiratory chain, which is encoded by mtDNA and represents the core of OXPHOS functionality. In keeping with previous findings, siPBGD cells showed the lowest levels of COXI and COXII, both subunits composing the mtDNA-encoded Complex IV (p<0.0001 at ANOVA, p<0.0.0001 vs siPBGD-NT and p=0.0007 vs scramble fasted cells, Figure 3D; p<0.05 vs siPBGD-NT and p<0.01 vs scramble fasted cells, Figure 3E), thereby supporting that fasting may activate mitochondrial biogenesis and exacerbate mitochondrial injury in siPBGD cells by shifting towards fission rather than fusion.
In siPBGD cells, pre-treatment with α-LA before fasting enhanced mitochondrial biogenesis by upregulating PCG1α (p<0.0001 at ANOVA, p=0.01 vs siPBGD fasted cells, Figure 3F) more than glucose administration alone (p<0.0001 at ANOVA, p=0.004, Figure 3F). The OPA1 and MFN2 mRNA expression were increased with the α-LA and glucose supplementation compared to siPBGD in fasting condition (p<0.0001 at ANOVA, p<0.0001; Figure 3G-H), showing similar levels among the two treatments ( Figure 3G-H). Unexpectedly, DRP1 expression was strongly pulled down by α-LA treatment (p<0.0001 at ANOVA, p=0.0002 vs siPBGD-NT, p=0.0001 vs siPBGD fasted cells and p<0.0001 vs siPBGD+Gluc, Figure 3I) supporting that it may counteract the effect of fasting by switching mitochondrial biogenesis towards fusion rather than fission.

Discussion
The binomial association between porphyrins and acute neuropsychiatric attacks has been at the basis of AIP pathogenesis for decades. Still, there are several reports which doubt about the exclusive pathogenic role of heme precursors in disease, as AIP patients who accumulate ALA and PBG may not manifest any acute attack throughout their life [17,23]. Thus, other factors may be involved in the pathophysiology of the acute event. A link may exist between metabolic homeostasis and biosynthesis of heme, as the latter is an essential precursor for the correct nutrients' handling to meet energy requirements during basal and stressful situations.
Attempting to shed light on this issue, we silenced PBGD gene in HepG2 cells (siPBGD) and characterized metabolic alterations occurring at baseline and after glucose deprivation, a condition which could mimic fasting in vitro. Transient PBGD silencing in HepG2 cells reduced PBGD and ALAS1 levels paralleled by low hepatocellular heme content thus supporting it may reproduce the slowdown of the hepatic heme biosynthesis observed in carriers of the HMBS genetic mutations [24]. We revealed that siPBGD cells at baseline showed low expression of glycolytic enzymes and mitochondrial ATP production, possibly indicating that PBGD downregulation may cause early aberrancies of glucose metabolism and OXPHOS functionality. The basal characterization of hepatic metabolism in AIP context has been poorly explored and just one study reported that the hepatic transcriptomic profile was similar between AIP and the congenic wild-type (WT) mice in absence of the stress-induced attack [13]. Nonetheless, characterization of nutritional status of AIP subjects highlighted that the inadequate glucides and carbohydrates consumption less than 45-60% of the total energy intake, was associated with disease severity [25,26]. In addition, sub-clinical OXPHOS defects, resulting in high circulating lactate levels, were found outside of the crisis in AIP patients in clinical remission [27]. Therefore, there may be some discrepancies in gene expression at baseline between transcriptomic findings by Chen et al [13] and ours, mainly due to the limitations of transient PBGD silencing in vitro which could not fully resemble features of a stable knockout model. However, the downregulation of glycolysis and the biochemical evaluation of energetic status of siPBGD cells could, in broad terms, mirror the energy failure featuring AIP patients at higher risk to develop symptoms [13,[25][26][27].
Previous evidence has outlined that most dysregulated pathways in the liver which may precipitate the AIP acute attacks are under the transcriptional regulation of PCG1α, a powerful nutrient sensor activated in response to stressful factors [28]. In AIP context, PGC1α attempted to re-establish glucose homeostasis by activating hepatic gluconeogenesis and increasing mitochondrial mass, but it also induced ALAS1 expression [12,13,17], thereby exacerbating the porphyrins 'overproduction. In keeping with these findings, we found that the induction of fasting through glucose deprivation halved PBGD enzymatic activity at 50% and sensitized siPBGD cells to upregulate ALAS1, likely via PGC1α, without increasing heme production similar to what has been reported in AIP rodents and patients exposed to precipitating factors [11,17].
Fasting even downed glycolysis and total ATP levels, possibly aggravating energetic imbalance in siPBGD cells as occurs in AIP mice during an attack which lowers glucose availability and runs into cataplerosis of the Krebs cycle, resulting unable to provide reducing equivalents (NADH, FADH2) to the OXPHOS [29]. Beyond bioenergetic failure, alterations of mitochondrial dynamics, including a cycle of fission and fusion events, could exacerbate metabolic dysfunctions as it regulates the intracellular mitochondrial mass, shape, and metabolic status of these organelles [30]. Here, we firstly revealed that fasting-induced PCG1α activation in siPBGD cells was accompanied by low expression of fusion genes and overexpression of DRP1, the main inducer of mitochondrial separation. Fission of mitochondria is commonly associated with low OXPHOS performance and ATP synthesis consistent with the reduced expression of complex IV which has been observed in siPBGD fasted cells. Homedan et al showed that hepatic complexes I, II and III, but not complex IV, decreased their activities in PBGD-deficient mice treated with phenobarbital, a drug which induces the acute episodes [12,29]. Conversely, morphological alterations of mitochondria presenting para-crystallin inclusions have been described in liver biopsies of AIP patients [30] and lack of complex IV activity alongside collapse of ATP were found in the hippocampus of AIP mice knocked-in for the Hmbs c.500G > A (p.R167Q) mutation, resulting in a severe phenotype with neuropsychiatric behavior [31]. Most recently, it has been demonstrated that the deficiency of ferrochelatase (FETCH), a gene involved in the last step of heme production and responsible for the development of Erythropoietic protoporphyria, damaged both glycolysis and OXPHOS along with a decrease in mitochondrial fusion [32]. Still, glucose metabolism and mitochondrial dysfunction might precipitate the acute attacks but not all studies share the same alterations possibly reflecting the huge phenotypic variability of AIP symptomatology [12,[29][30][31][32]. Overall, our model may support the hypothesis that fasting on one hand worsens glycolytic alterations and on the other unmasks mitochondrial damage by shifting mitochondrial dynamics towards fission rather than fusion and, consequently, contributing to OXPHOS aberrancies.
Life-threating attacks featuring AIP patients represent a crippling issue for subject who suffer from this rare disorder and mostly for those who are predisposed to recurrent crisis. The ongoing therapies aim to modulate ALAS1 activity and even showed several constraints [14,15,17,33]. AIP patients with mild attacks may be treated with glucose solution, but there are no clear clinical data showing beneficial effects in AIP prophylaxis [17,34], possibly due to an impaired glucose metabolism [10][11][12]. Conversely, hemin directly represses ALAS1 activity but its long-term exposure has been associated with hepatic iron accumulation, oxidative stress, inflammation and liver fibrosis in AIP patients alongside refractoriness to the attacks [14,33], growing the need of novel medications. In the recent years, it has emerged that AIP patients showed hyperinsulinemia, a condition which appeared protective against a stress-induced attack [17]. It has been also postulated that reduced insulin levels and C-peptide were associated with AIP disease activity, possibly due to the essential role of insulin for the ALAS1 inhibition and blockage of porphyrin overproduction [35].
Therefore, we assessed whether a treatment with an insulin sensitizer, the α-LA, alone or in combination with glucose may improve metabolic abnormalities when supplemented before fasting-induced stress with the goal to offer a potential preventive approach of acute events.
α-LA acts as co-factor of the α-ketoglutarate dehydrogenase in the Krebs cycle, which bridges glycolytic-mitochondrial respiration network and provides succinyl-CoA substrate for heme biosynthesis [36,37]. The main findings of this project were that pre-treatment with α-LA, likely boosting mitochondrial Krebs cycle, ameliorated heme biosynthesis in siPBGD cells and their energetic status by promoting ATP and lipid production. Moreover, we revealed that α-LA may counterbalance the harmful effects of fasting through the stimulation of mitochondrial elongation and inhibition of mitochondrial fission attempting to meet energy demand. Glucose administration alone even recovered the expression of glycolytic enzymes and of genes involved in mitochondrial dynamics but it failed at improving ATP content, possibly explaining the reduced efficacy of glucose therapy reported in humans [17]. When α-LA was combined to glucose, we found that total ATP, triglyceride levels and mitochondrial mass were increased in siPBGD cells supporting that α-LA may additively enhance glucose efficacy and rescue the overall mitochondrial dynamics (Figure 4). By looking at a translational perspectives, α-LA supplementation may potentially sustain the hepatic absorption and maintenance of glucose homeostasis from dietary carbohydrates [17].
To conclude, emerging studies have highlighted that heme depletion may deeply affect glucose metabolism and mitochondrial dynamics [32]. The present study attempted to dissect the role of metabolic aberrancies occurring in hepatocytes in the context of PBGD downregulation, mimicking AIP condition in vitro. We proposed, for the first time, a novel compound which may be an appealing for the prevention of AIP crisis through the improvement of hepatocellular bioenergetics. As proof-of-concept, our findings highlighted that α-LA could re-establish glucose employment, thus restoring the cross-talk among cytosolic glycolysis and mitochondrial respiration and, therefore, the hepatocellular homeostasis.
Author Contributions: M.L. designed, wrote and revised the manuscript, collected, analyzed and interpreted data and prepared figures/tables; E.P. and M.M took part in the experimental design, data interpretation and reviewed the manuscript; L.D contributed to the in vitro study; I.M and A.L.F. contributed to discussion, manuscript revision, data interpretation; E.D.P and P.D. took part to study design, manuscript drafting, data analysis and interpretation, study funding, supervision and has primary responsibility for final content. All the authors read and approved the final draft.  but it improved OPA1 and MFN2 expression, thereby enhancing mitochondrial fusion, a condition which could support OXPHOS activity during energy demand. C) Pre-treatment with α-LA and glucose may recover the overall mitochondrial wellness by acting at multiple levels. On one hand, α-LA may improve heme biosynthesis, a mechanism occurring in mitochondrial matrix. On the other hand, α-LA may enhance glycolysis, ATP production and OXPHOS possibly sustaining Krebs cycle. Restoration of hepatocellular homeostasis could allow to recover mitochondrial dynamics, in terms of fusion, fission and overall mitochondrial mass.