AD is one of the most common neurodegenerative diseases worldwide and is considered the leading cause of dementia in the world's adult population (Hirtz et al., 2007). From a genetic point of view, AD can be classified into two types, familial cases with autosomal dominant inheritance (FAD), and sporadic AD which primary cause are still unknown (SAD) (Barage and Sonawane, 2015). The accumulation of misfolded proteins such as the amyloid beta peptide (Aß) and hyperphosphorylated tau are considered essential components in the pathogenesis of AD (Querfurth and LaFerla, 2010; Huang and Mucke, 2012). Current evidence suggests that mitochondrial impairment is also a hallmark of AD, as mitochondrial dysfunction appears before the establishment of tau and Aβ pathology contributing to synaptic impairment observed in AD (Knott et al., 2008; Cabezas-Opazo et al., 2015, Yao et al., 2011).

AD mainly affects memory and cognitive functions, and currently, there is an urgent need for an early biomarker with the reliability and accuracy to diagnose this disease. Nowadays, AD diagnosis is based on neuropsychological surveys and in the exclusion of other age-related dementias in consultant patients (Lanfranco et al., 2012). Analysis of this data has determined three stages of AD progression: preclinical AD, mild cognitive impairment type Alzheimer's, and Alzheimer's type dementia (Lanfranco et al., 2012). Unfortunately, the confirmation of AD is only possible during the patient's autopsy observing the characteristic pathological brain lesions (Khan and Alkon, 2015).

Several groups had suggested the use of peripheral tissues as a reliable source of biomarkers that could reflex the metabolic changes in the brain (Yao et al., 2011; Perez et al., 2016). In fact, there are several examples of using skin fibroblasts to study metabolic abnormalities related to neurological (Zoumakis et al., 2007), psychiatric (Kalman et al., 2016), neurometabolic (Khan and Alkon, 2015) and neurodegenerative diseases (Lopez-Erauskin et al., 2012; Ambrosi et al., 2014), in particular Huntington's disease (Marchina et al., 2014), Parkinson's disease (Rakovic et al., 2010; Cooper et al., 2012; Papkovskaia et al., 2012; Mcneill et al., 2013), and genetic forms of Amyotrophic Lateral Sclerosis (Bartolome et al., 2013; Prause et al., 2013; Allen et al., 2014). For that reason, it has been suggested that biochemical changes in brain cells could be reflected in peripheral tissues derived from ectoderm, such as skin (Zoumakis et al., 2007), and that this tissue represents a suitable model to study potential new biomarkers in AD and other neurodegenerative diseases (Khan and Alkon, 2015).

In AD brains, mitochondrial dysfunction is commonly presented as a respiratory failure, with a reduced ATP production, increased reactive oxygen species (ROS) production, abnormal calcium regulation, and mitochondrial fragmentation (Hu et al., 2016). Interestingly, evidence in cortical samples from AD patients suggests an unbalance of mitochondrial fission and fusion, as mitochondrial dynamics is inclined toward fission (Manczak et al., 2011). Therefore, we will evaluate mitochondrial deficits associated with neurodegeneration in fibroblasts of young, age-matched healthy and sporadic Alzheimer's disease patients.

In this work, we evaluated changes in mitochondrial bioenergetics and morphology to identify possible differences in mitochondrial function of fibroblasts from sporadic AD and match-aged healthy patients. Evaluation of mitochondrial bioenergetics showed that AD patients fibroblasts presented higher levels of ROS (Figure 3D) and with no differences in the basal levels of mitochondrial membrane potential (Figure 3A). These data are in agreement with other groups that had shown that AD fibroblasts presented higher levels of basal ROS compared to non-AD patients (Gibson et al., 1996; Naderi et al., 2006; Murphy and Steenbergen, 2011). Also, further analysis of total ATP levels for groups (Control vs. AD group) did not show differences between control and AD patients (Figure 3E). However, the analysis of each patient (Figure S2) indicated that mild-to-moderate and severe AD patients had decreased levels of ATP compared to age-matched healthy patients and preclinical AD patient, suggesting that the compromised bioenergetics state in AD fibroblasts could be related to the progression of the disease.

On the other hand, studies in AD cell models have been shown that mitochondrial potential is only affected under stressor conditions such as calcium or amyloid beta peptide (Quintanilla et al., 2012; Vargas et al., 2014). In that context, we found that in all analyzed patients showed equal levels of basal calcium (Figures 4A,B), and after treatment with thapsigargin produced a significant increase in relative calcium levels in AD patients respect to control cells (Figure 4). Also, thapsigargin treatment induced a further decrease in mitochondrial length and a significant reduction in mitochondrial potential in AD patients (Figures 5A,B), indicating a reduced capacity of the mitochondria to respond to calcium stress. In fact, treatment with the mitochondrial uncoupler, FCCP completely reduced the enhanced calcium increase in AD fibroblasts (Figures 5C,D), suggesting that the incapacity of calcium buffering is related to mitochondrial impairment in AD patients. Interestingly, studies in young healthy patients revealed that young fibroblasts presented higher ATP levels, less ROS production, and intact calcium buffering capacity compared to healthy elderly patients (Figure 6). These findings suggest that in these fibroblasts the aging process could be associated with a reduction of mitochondrial bioenergetics similar to what is observed in mice models (Brunk and Terman, 2002; Lores-Arnaiz et al., 2016; Wyss-Coray, 2016).

Studies in AD brain samples have shown fragmented mitochondria (Wang et al., 2009b), a decrease in mitochondrial density in synaptic structures (Du et al., 2010), and an altered expression of fission and fusion proteins which could explain the mitochondrial fragmentation (Wang et al., 2009a). Here we showed that fibroblasts of sporadic AD patients presented a significant decrease in mitochondrial length and an increase in the percentage of shorter mitochondria respect to the control patients (Figure 1). However, we did not detect differences in the levels of the fission protein DRP1 (Figure 2). These results are opposite to those described previously that shows that levels of DRP1 were decreased in sporadic AD fibroblasts and that was possibly related to an increase in elongated mitochondria accumulated in perinuclear areas of the cells (Wang et al., 2008). However, our results are in agreement with findings presented in AD mice models and cortical AD samples that showed mitochondrial fragmentation (Wang et al., 2008; Cai and Tammineni, 2016), accompanied with an increase in DRP1 (Manczak et al., 2011) and a decrease in total OPA1 levels (Manczak et al., 2011).

Moreover, our results strongly suggest that AD fibroblasts present a specific deregulation of mitochondrial fusion proteins through the reduction of OPA1 and MFN1 expression (Figure 2). In AD fibroblasts the expression of the 84 kDa isoform of MFN1 is decreased and the 71 kDa isoform is significantly increased (Figure 2). Interestingly, it was described that cortical samples of AD patients showed a reduction in the 84 kDa isoform which is related to the severity of the disease (Manczak et al., 2011), but the 71kDa isoform is not present in all the patient's brains (Manczak et al., 2011). Complementary studies showed that the expression of a MFN1-splice variant was specifically upregulated in tumorigenic cells (Chung et al., 2001), suggesting that this different pattern could be related to a disease status (Chung et al., 2001). On the other hand, AD fibroblasts presented a decrease in the expression of total OPA1 and a specific pattern of proteolytic processing, consisting of a decrease in L1, L2, S3, and S4 and an increase in S5 forms (Figure 2). Interestingly, it was described that mitochondrial stress affects the distribution of OPA1 and trigger the complete conversion of L-OPA1 into S-OPA1, leading to a stimulation of mitochondrial fission (Anand et al., 2014). Indeed, this specific increase of the S5 form was described as a result of OPA1 processing induced by H₂O₂ (Anand et al., 2014). These are a novel an interesting findings that suggest that mitochondrial fragmentation in AD fibroblasts could be mediated by an increase in MFN1 71kDa isoform and the irregular processing of OPA1.

According to recent publications, the use of skin-derived cells to investigate neurodegenerative diseases with the aim of identifying early biomarkers had to attract increased interest (Clos et al., 2012; Ambrosi et al., 2014; Khan and Alkon, 2015). Indeed, several groups have used use fibroblasts as a model to investigate the mechanisms underlying the pathology of these diseases, with the idea of describing how this cell type present the molecular and metabolic defects that are typically reported at central level (Squitieri et al., 2010; Zanellati et al., 2015; Lopez-Fabuel et al., 2017). For example, Squitieri and collaborators had shown that fibroblasts of Huntington disease's (HD) patients provide a potential peripheral marker as skin biopsies presented pathological mitochondrial features that may predict neuronal pathology (Squitieri et al., 2010). In Parkinson disease (PD), several studies had shown that patient's fibroblasts may be a reliable model system to study mitochondrial dysfunction (Mortiboys et al., 2008; Grunewald et al., 2010; Pacelli et al., 2011; Van Der Merwe et al., 2014), as this cell type present alterations in mitochondrial bioenergetics (Zanellati et al., 2015; Lopez-Fabuel et al., 2017), dynamics (Zanellati et al., 2015), and in the expression and assembly of mitochondrial complexes (Lopez-Fabuel et al., 2017). All these alterations were largely confirmed in mitochondria of cultured neurons and forebrain samples of several PD mouse models (Mortiboys et al., 2008; Grunewald et al., 2010; Pacelli et al., 2011; Van Der Merwe et al., 2014; Lopez-Fabuel et al., 2017). In the case of AD, several studies had shown that fibroblasts of patients with FAD presented a reduction in the mitochondrial number and respiratory function (Gray and Quinn, 2015), high levels of ROS (Richardson, 1993; Moreira et al., 2007), reduced levels of antioxidant defenses (Cecchi et al., 2002), and mitochondrial calcium handling defects (Richardson, 1993; Ito et al., 1994); Therefore, our results presented here indicate that fibroblasts obtained from SAD patients could also present the mitochondrial dysfunction features observed in the AD brain.

Despite the fact that our observations correspond to a limited number of samples, we consider these findings novel and they suggest that mitochondrial function is significantly compromised in AD fibroblasts, presenting identical alterations showed in neuronal cells during AD progression. From our part, further studies will be made increasing the number of samples from control, AD, and MCI patients to corroborate our observations regarding defects in mitochondrial dynamics and bioenergetics.

AD is a disease commonly associated with cerebral pathology. However, it is accepted that several neurodegenerative diseases where mitochondrial dysfunction is compromised as HD, PD, and ALS, also present also a systemic component that affects peripheral tissues outside the central nervous system. The possibility that skin fibroblasts could reflect these metabolic changes produced in the brain makes this cell type a suitable model to study new biomarkers of AD.

Despite the small group of samples, we were able to analyze several changes that are characteristic of mitochondrial damage. Interestingly, fibroblasts obtained from patients with sporadic AD show specific defects in mitochondrial dynamics and bioenergetics suggesting a new possible target in search of peripheral biomarkers for AD.

To our knowledge, the present study represents an interesting characterization of the mitochondrial status of patient's fibroblasts and suggests that the evaluation of mitochondrial function could differentiate normal aging of pathological AD aging. Our work opens the possibility of a new target for the development of AD biomarkers and presents a novel strategy for epidemiological studies in this disease.

A total of 7 patients participated in this study after signed an informed consent approved by the Bioethics Committee of the Hospital Clínico de la Universidad de Chile. In severe cases of dementia, their caregivers provided the consent. This study as been endorsed by the Bioethics Committee of the Hospital Clínico de la Universidad de Chile and validated by the Universidad Autónoma de Chile.

MP: Performed research, designed experiments, collected data, analyzed data, wrote paper. DP: Collected data, analyzed data. CO: Collected data, revised paper. MB: Contributed with patients samples, analyzed data, revised paper. RQ: Directed the project, designed experiments, contributed important reagents, analyzed data, and revised paper.