In the past years, reports of clinical signs suggestive of mitochondrial dysfunction in non-infected children exposed to NRTIs have prompted studies addressing the issue of developmental toxicity of such drugs and, in particular, of AZT.
Occurrence of subclinical mitochondrial dysfunction has been consistently shown by either clinical and experimental studies in different tissues and organs, including the brain 
. In particular, a recent study provided evidence that in utero
exposure to the NRTI, AZT and Lamivudine, is associated with brain mitochondrial impairment, which progresses over time and might possibly end up in delayed neurobehavioral effects 
. In laboratory rodents, developmental exposure to AZT produces both early and delayed behavioral changes in offspring including alteration in sensorimotor maturation, social/aggressive behavior, responses to environmental stressors and learning abilities 
. Altogether, these data pointed to the potential risk of sub-clinical side effects of ARV therapy on the developing brain that may go undetected in clinical and epidemiological studies. Thus, current recommendations call for long-term clinical follow-up for any child with in utero
exposure to ARV drugs 
to monitor significant side effect of ARV drugs that may augment risk in children and adolescents born to HIV-1-positive women, who represent a vulnerable group as for psychiatric health 
The main results of the present preclinical study contribute to clarify this controversial issue, as they show that 1) transplacental exposure to AZT enhances oxidative stress and causes a consistent and permanent alteration of hippocampal glutammatergic neurotransmission functionally associated to a deficit in spatial learning and memory, 2) LAC administration from the beginning of pregnancy prevents/ameliorates the adverse effects of AZT on both neurochemical and behavioral parameters.
In line with previous findings in rodent models, we show that transplacental AZT exposure has significant long-term effects on behavior of exposed offspring. In particular, AZT treatment has detrimental effects on performances in the water maze task, one of the most widely used tests to measure hippocampal-dependent spatial-based learning and memory in rodent models with high translational value with respect to humans 
. We observed a marked deficit in acquisition and memory recall, as both acquisition of the platform location and retrieval of the acquired information in the probe trial were impaired in AZT-treated mice. Since swimming speed was not affected by AZT treatment, we can exclude motor impairment possibly associated with the well-known effects of this drug on muscle tissue 
Different neurotransmitter systems have been implicated in the effects exerted by AZT and other ARV drugs on behavior but a correlation between specific neural mechanisms and the behavioral changes observed has not been attempted yet. Extensive evidence indicates that learning and memory components of spatial navigation in rodents rely on hippocampal synaptic plasticity 
that is largely modulated by hippocampal mGlu 
and by AMPA receptors 
. Our findings indicate that hippocampal glutamate neurotransmission is permanently affected by early AZT exposure and support a role of this important neurochemical pathway in the behavioral effects here reported.
The reduced expression of metabotropic group I and ionotropic Glu1 subunit of AMPA receptors in the hippocampus is well in agreement with the specific impairment of spatial learning and memory performances observed in AZT treated mice. Recent data strongly support a role for the different glutamate receptor subtypes in spatial cognition: administration to rats of mGluR1 antagonists affects synaptic plasticity and impairs acquisition in the Morris water maze 
, while blockade of mGlu5 receptors interferes with long-term potentiation and has detrimental effects on spatial learning 
. As well as for metabotropic group I receptors, there is ample evidence that the dynamic regulation of AMPA receptors - which mediate most of the fast excitatory synaptic transmission - can change synaptic function and regulate storage of information 
. Based on our behavioral and neurochemical findings, we hypothesize that the lower expression of different subgroups of glutamate receptors in AZT mice results in reduced plasticity of hippocampal glutamatergic synapses and, as an ultimate consequence, in diminished capacity to encode relevant information during the spatial learning task.
Such effect could be further amplified by the enhanced corticosterone responsiveness to stress in AZT exposed mice. Our findings show that administration of AZT during pregnancy caused increased release of plasma corticosterone following acute restraint stress at adulthood. As water maze training induces release of high corticosterone levels shortly after completion of the task in mice 
, an abnormal physiological reaction to stress could interfere with both acquisition and consolidation of information in AZT mice. An inverse relationship between spatial learning and memory and elevated corticosterone levels has been widely demonstrated in rodents: exposure to mild stressor or moderate glucocorticoid levels facilitates spatial memory 
while chronic stress or higher glucocorticoid levels exert detrimental effects on neuronal plasticity and spatial memory 
. The increased corticosterone release after acute stress observed in AZT treated mice deserves further consideration, as it is suggestive of altered HPA axis response caused by early AZT exposure. A recent study indicates that chronic AZT is able to increase corticosterone release in adult rats possibly by a central action 
. In a developmental perspective, AZT treatment could interfere with the physiological setting of HPA axis activity, thus reducing the capability to respond adaptively to environmental challenges. Previous studies from our laboratories indicate that mice exposed to AZT during development present abnormal reaction to novelty, and that pre- and/or postnatally AZT exposed adult male mice show maladaptive intraspecific social/aggressive behavior depending on the length of exposure 
Overall, our findings confirm the behavioral effects of transplacental AZT in rodent models. More importantly, they point for the first time to a specific and permanent action of this drug on glutamate neurotransmission. We propose that such effect is linked to the well-known mitochondrial toxicity of AZT 
. NRTI-induced alterations in mitochondrial structure and function result from interference with various mechanisms involved in the normal maintenance of mitochondrial function. In particular, it has been suggested that AZT directly impairs the electron transport chain (ETC) thus increasing production of reactive oxygen species (ROS) and oxidative stress, which will eventually lead to a loss of the mtDNA integrity 
. Our data confirm that transplacental exposure to AZT induces oxidative stress in the fetal brain and determines long-term effects on the oxidative status of brain tissue as two different oxidative stress markers, 15-F2t
-isoP and protein carbonyl levels at birth and at adulthood, respectively, were enhanced in AZT-treated offspring.
- Isoprostanes are lipid peroxidation products; they are among the most sensitive in vivo biomarkers of oxidative stress and are the marker of choice for the evaluation of oxidative damage in small size samples such as newborn mouse brain 
. Similarly, immunodetection of carbonyl groups provides a measure of oxidation status of proteins, and in turn, of oxidant injury. The increased levels of 15-F2t
-isoP and protein oxidation are indicative of increased levels of ROS, likely deriving from AZT-dependent mitochondrial impairment, as described in several previous studies 
The evidence of early oxidative stress in AZT exposed offspring allows us to advance a mechanistic explanation for the endocrine and neurobehavioral effects here reported. Impairment of mitochondrial function and production of increased levels of ROS may reduce the capacity of the cell to generate ATP, and induce excess glutamate release at the synapse and, eventually, excitotoxicity 
. In turn, an excess of glutamate release in the early stage of life results in overstimulation of glutamatergic receptors that can be possibly permanently modified in their expression and functionality. Besides, it should be considered that a significant portion of the control of HPA axis activity is mediated by glutamate 
thus enhanced glutamate release during early development could result in lifelong alteration of HPA axis responsiveness. The link between AZT-induced oxidative stress in the fetal stage and the long-term consequences on neurobehavioral functions at the adult stage is strongly supported by the protective effect exerted by LAC. Indeed, we show that most of the alterations induced by AZT are prevented or reduced by administration of LAC from the beginning of pregnancy.
Several lines of evidence indicate that administration of antioxidant agents, including vitamin C, E and coenzyme Q10 prevents or alleviates oxidative stress and symptoms of AZT-induced myopathy in both rats and humans (see 61 for a review).
More relevant to our study, LAC is known to decrease oxidative stress by 1) exerting a protective effect on mitochondrial structure and function, making the ETC less prone for electron leak and superoxide production, and promoting mitochondrial biogenesis, as indicated by increased mitochondrial size and number 
; 2) stimulating the endogenous cellular antioxidant defense mechanisms. These effects possibly rely on the ability of LAC to act as a donor of acetyl groups. In a recent work, a proteomic survey of protein acetylation identified that 20% of mitochondrial proteins presented with acetylation sites and these proteins included those involved in the tricarboxylic acid cycle, fatty acid β-oxidation, amino acid and carbohydrate metabolism, membrane transport, and ETC 
. These data clearly show that acetylation can control the activity of mitochondrial enzymes and, since acetyl-CoA is the acetylation donor for all known acetyltransferases, the concentration of mitochondrial acetyl-CoA could be a limiting factor in the acetylation reaction. By increasing the acetyl-CoA content, supplemented LAC should increase the acetylation status of mitochondrial proteins thus improving mitochondrial function and integrity. Interestingly, low levels of circulating LAC have been found in HIV-1 patients experiencing a clinically manifest peripheral neuropathy while staying on AZT and other nucleoside analogue treatment 
, and in patients with AZT-induced mtDNA depletion 
. Hence, LAC deficiency could play a role in neurotoxic effects of these drugs. AZT treatment reduces carnitine levels in blood and tissues 
and LAC treatment could mediate an improvement of mitochondrial dysfunction in a straightforward manner, namely, by replenishing carnitine levels.
Other clinical studies have shown that treatment with LAC improved NRTI-induced polyneuropathy symptoms in HIV-1-positive patients, including electrophysiological variables relating to motor conduction velocity 
. In a morphological study in HIV-1 patients with distal symmetrical polyneuropathy, regeneration of all fiber types and in particular of small sensory fibers in the dermis, epidermis and sweat glands was observed after oral LAC treatment 
. All this considered, LAC can be viewed as a pleiotropic molecule metabolically active, with neuromodulatory, neurotrophic, cytoprotective, and antioxidant activity in brain.
The use of ARV drugs to prevent mother-to-child transmission of HIV-1 infection is one of the most successful achievements in HIV-1 prevention and, based on several clinical and epidemiological studies, benefits of this therapy appear to outweigh its adverse side effects in exposed infant/children 
. However, there is still considerable uncertainty on the prevalence and severity of mitochondrial toxicity associated to ARV treatment in pregnancy 
. In this framework, research on therapeutic intervention effective in reducing side effect risks of this life-saving regimen should be implemented. The present preclinical findings, besides providing a mechanistic hypothesis for the neurobehavioral effects of AZT, strongly suggest that preventive administration of LAC already shown to improve NRTI-induced polyneuropathy in HIV-1-positive patients 
might also be effective in reducing the neurological side-effects of ARV therapy in fetus/newborn.
It is known that an excessive production of pro-inflammatory cytokines and free radicals might occur during HIV infection. Because pro-inflammatory cytokines and free radicals can travel through placenta to the fetus, they could contribute to increase the fetal oxidative stress induced by NRTI drugs. Nonetheless, our results highlight the importance of counteracting the neurotoxic effects of NRTIs in terms of rescue of neuronal function based on the maintenance of mitochondrial activity. Since LAC treatment has generally shown a good profile of safety and tolerability, our study strongly supports the rationale for the development of clinical trials aimed at investigating the efficacy of the administration of LAC to HIV-1 infected pregnant women treated with NRTIs to counteract potential adverse consequences on their children.