Mitochondrial Respiration in Response to Iron Deficiency Anemia: Comparison of Peripheral Blood Mononuclear Cells and Liver

Iron is an essential component for metabolic processes, including oxygen transport within hemoglobin, tricarboxylic acid (TCA) cycle activity, and mitochondrial energy transformation. Iron deficiency can thus lead to metabolic dysfunction and eventually result in iron deficiency anemia (IDA), which affects approximately 1.5 billion people worldwide. Using a rat model of IDA induced by phlebotomy, we studied the effects of IDA on mitochondrial respiration in peripheral blood mononuclear cells (PBMCs) and the liver. Furthermore, we evaluated whether the mitochondrial function evaluated by high-resolution respirometry in PBMCs reflects corresponding alterations in the liver. Surprisingly, mitochondrial respiratory capacity was increased in PBMCs from rats with IDA compared to the controls. In contrast, mitochondrial respiration remained unaffected in livers from IDA rats. Of note, citrate synthase activity indicated an increased mitochondrial density in PBMCs, whereas it remained unchanged in the liver, partly explaining the different responses of mitochondrial respiration in PBMCs and the liver. Taken together, these results indicate that mitochondrial function determined in PBMCs cannot serve as a valid surrogate for respiration in the liver. Metabolic adaptions to iron deficiency resulted in different metabolic reprogramming in the blood cells and liver tissue.


Introduction
Anemia is a global burden affecting more than 1.5 billion people worldwide [1][2][3]. The most prevalent form of anemia is iron deficiency anemia (IDA), mostly occurring in preschool children and women in reproductive age [1,2]. The main symptoms include fatigue, weakness, and pale skin along with reduced cardiovascular performance. Moreover, IDA may negatively affect the growth and cognitive development of children [1, 4,5]. IDA is characterized by absolute iron deficiency caused by chronic blood loss, insufficient dietary intake, or a combination of both [1,5].
Iron is an essential trace element for life, as it is necessary for hemoglobin biosynthesis and for key metabolic enzymes involved in DNA replication, hormone synthesis, and mitochondrial bioenergetics [6,7]. Especially in mitochondria, iron is needed for heme synthesis, iron sulfur (Fe-S) cluster formation, and oxidative phosphorylation [8][9][10][11][12]. In the electron transfer system (ETS) complexes CI, CII, and CIII possess Fe-S clusters which are crucial for the synthesis of adenosine triphosphate (ATP).
Iron metabolism is tightly regulated by the liver-derived peptide hormone hepcidin, which mediates the degradation of ferroportin, the only cellular iron exporter known so far, thereby controlling iron absorption and iron recycling from macrophages [13]. Expression of hepcidin is controlled by multiple factors. Both iron deficiency and anemia reduce hepcidin expression, therefore increasing circulating iron levels and delivery to erythroid progenitor cells [3,13].
It has been shown previously that systemic iron deficiency and IDA reduce mitochondrial respiratory capacity in cardiomyocytes and skeletal muscle via the reduction of iron-rich mitochondrial electron transfer components and morphologic changes of mitochondria, such as reduced cristae structure [14][15][16][17][18]. In addition, mitochondria in hepatocytes of iron-deficient rats exhibit ultrastructural abnormalities, including an enlarged and rounded shape, therefore occupying an increased proportion of the cytoplasm due to their larger size, but not due to an increased number of mitochondria [19,20].
Despite the possible implications of these morphologic alterations on mitochondrial function, the impact of IDA on mitochondrial respiration in circulating peripheral blood mononuclear cells (PBMCs) and the liver has not been analyzed thus far. We studied these two systems, since the determination of mitochondrial function in PBMCs might represent an easily accessible surrogate reflecting mitochondrial functionality in the organs, such as the liver. Changes in the mitochondrial respiration of PBMCs have been observed in various diseases, including fatty liver disease, depression, or sepsis, and even in restless legs syndrome, where impaired mitochondrial function was linked to the indication of mitochondrial iron deficiency [21][22][23][24]. Therefore, the analysis of PBMCs could be an efficient and minimally invasive procedure when relating information from blood samples to organ function or disease state instead of collecting biopsies from the particular organ itself.
The aim of this study was a comparative analysis of mitochondrial respiration in PBMCs and the liver under steady-state control conditions and in the course of IDA.

Effects of IDA on Hematological and Iron Parameters
IDA was induced by phlebotomy for five consecutive days, as previously described [25]. As anticipated, the hemoglobin concentration was significantly lower in rats with IDA compared to the controls (Figure 1a). Moreover, the reticulocyte fraction was increased ( Figure 1b). Plasma iron concentration did not differ significantly (Figure 1c). The lymphocyte fraction remained unaffected (Figure 1d), whereas an increased proportion of monocytes altered the PBMC composition in rats with IDA ( Figure 1e). Hepatic mRNA expression (Hamp) of the master regulator of the iron metabolism, hepcidin, was decreased ( Figure 1f). Along with that, the liver iron content was lower in the IDA animals compared to the controls (Figure 1g).

Mitochondrial Respiration Differs in PBMCs and Livers in Response to IDA
We then analyzed the effects of IDA on mitochondrial respiratory capacity in freshly isolated PBMCs and homogenate from fresh liver biopsies (for protocols, see Figure 6a,b). Of note, we observed an increase of the mitochondrial respiration in the PBMCs of the IDA compared to the control rats (  (a) Hemoglobin (Hb) concentration, (b) reticulocyte proportion, (c) plasma iron concentration, (d) lymphocyte fraction, (e) monocyte fraction, (f) liver Hamp mRNA expression relative to glucuronidase beta (Gusβ), and (g) liver iron content. Control: N = 9 rats, IDA N = 10 rats. An unpaired two-tailed Student's t-test was applied for all normal distributed results; a Mann-Whitney U-test was applied for the nonparametric result shown in (g). Values are shown as median ± interquartile range. p-Values are shown in the graphs.

Mitochondrial Respiration Differs in PBMCs and Livers in Response to IDA
We then analyzed the effects of IDA on mitochondrial respiratory capacity in freshly isolated PBMCs and homogenate from fresh liver biopsies (for protocols, see Figure 6a We next investigated whether the tissue-specific alterations of the mitochondrial function could be based on changes in the mitochondrial quality and density in response to IDA. Therefore, flux control ratios (FCR) were calculated in both PBMCs and the liver, whereby the values obtained from the measurements of the mitochondrial respiration were normalized for a common reference state (PBMCs: PGMS E ; liver: SGp E ) showing mitochondrial respiration independent of mitochondrial density (Figure 3a,c) [26,27]. In PBMCs, increased mitochondrial respiration was still observed after normalization independent of mitochondrial density in OXPHOS, indicating a change in mitochondrial quality (Figure 3a, state PM P ) and ET (Figure 3a, states PM E and S E ). Furthermore, citrate synthase (CS) activity, a marker for mitochondrial density, was measured [28][29][30]. In PBMCs, CS activity was significantly increased (Figure 3b). Consequently, the observed changes in the mitochondrial respiration in PBMCs originated from alterations in both mitochondrial quality and mitochondrial density. In the liver, we found a slight increase in the OXPHOS capacity, whereas all other measurements of mitochondrial function remained unaffected when comparing the control to the IDA animals ( Figure 3c, state PM P ). CS activity in the liver did not show any differences (Figure 3d), indicating that the mitochondrial density remained unaltered when comparing the IDA with the control rats, and supporting the conclusion of the unchanged mitochondrial quality drawn from the FCR. We next investigated whether the tissue-specific alterations of the mitochondrial function could be based on changes in the mitochondrial quality and density in response to IDA. Therefore, flux control ratios (FCR) were calculated in both PBMCs and the liver, whereby the values obtained from the measurements of the mitochondrial respiration were normalized for a common reference state (PBMCs: PGMSE; liver: SGpE) showing mitochondrial respiration independent of mitochondrial density (Figure 3a,c) [26,27]. In  in the mitochondrial respiration in PBMCs originated from alterations in both mitochondrial quality and mitochondrial density. In the liver, we found a slight increase in the OXPHOS capacity, whereas all other measurements of mitochondrial function remained unaffected when comparing the control to the IDA animals ( Figure 3c, state PMP). CS activity in the liver did not show any differences (Figure 3d), indicating that the mitochondrial density remained unaltered when comparing the IDA with the control rats, and supporting the conclusion of the unchanged mitochondrial quality drawn from the FCR. Having observed the tissue-specific differences in the mitochondrial function in PBMCs compared to the liver of the control and IDA animals, we next studied for the possibility of mitochondrial damage. As a surrogate, we investigated a potential loss of the mitochondrial outer membrane (mtOM) integrity by quantifying the cytochrome c control efficiency (jcyt c) [27]. The jcyt c did not show significant differences either in PBMCs Having observed the tissue-specific differences in the mitochondrial function in PBMCs compared to the liver of the control and IDA animals, we next studied for the possibility of mitochondrial damage. As a surrogate, we investigated a potential loss of the mitochondrial outer membrane (mtOM) integrity by quantifying the cytochrome c control efficiency (j cyt c ) [27]. The j cyt c did not show significant differences either in PBMCs or in the liver when comparing the control to the IDA rats (Figure 4a,b). Consequently, the damage of the mtOM could be ruled out as a cause for the differences in PBMC mitochondrial respiration between the two groups. Furthermore, in the PBMCs of the IDA rats, an increase in the E-L coupling efficiency was detected (Figure 4c), indicating an improved coupling of the ET to the phosphorylation of ADP. This may likewise be underlying the observed increase in mitochondrial respiration. In the liver, the E-L coupling efficiency remained stable (Figure 4d), whereas the E-P control efficiency was decreased in the IDA liver homogenates as compared to the controls (Figure 4f), indicating a decreased capacity of the oxidative phosphorylation system [27,31]. In PBMCs, the E-P control efficiency remained unchanged (Figure 4e). an increase in the E-L coupling efficiency was detected (Figure 4c), indicating an improved coupling of the ET to the phosphorylation of ADP. This may likewise be underlying the observed increase in mitochondrial respiration. In the liver, the E-L coupling efficiency remained stable (Figure 4d), whereas the E-P control efficiency was decreased in the IDA liver homogenates as compared to the controls (Figure 4f), indicating a decreased capacity of the oxidative phosphorylation system [27,31]. In PBMCs, the E-P control efficiency remained unchanged (Figure 4e).

PBMC Mitochondrial Respiration was Not Correlated with Liver Mitochondrial Respiration
To investigate if the mitochondrial respiration in PBMCs reflects the liver mitochondrial respiration, we analyzed the correlation of the respiratory states shown in Section 2.2. (Figure 5a-j). In none of the states was the PBMC mitochondrial respiration significantly linked to the liver mitochondrial respiration. Consequently, the measurement of the mitochondrial respiration in the PBMCs does not reflect the liver mitochondrial function. Nevertheless, PBMCs may be used as a surrogate for mitochondrial dysfunction in specific diseases affecting other organs or tissues.

PBMC Mitochondrial Respiration was Not Correlated with Liver Mitochondrial Respiration
To investigate if the mitochondrial respiration in PBMCs reflects the liver mitochondrial respiration, we analyzed the correlation of the respiratory states shown in Section 2.2. (Figure 5a-j). In none of the states was the PBMC mitochondrial respiration significantly linked to the liver mitochondrial respiration. Consequently, the measurement of the mitochondrial respiration in the PBMCs does not reflect the liver mitochondrial function. Nevertheless, PBMCs may be used as a surrogate for mitochondrial dysfunction in specific diseases affecting other organs or tissues.

Discussion
We found that IDA led to different effects on the mitochondrial respiratory capacity in permeabilized PBMCs and liver homogenates. Surprisingly, IDA resulted in an increase of mitochondrial activity in PBMCs, for which different factors may be responsible. We observed a relative expansion of monocytes when comparing PBMCs from the IDA versus the control animals. Thus, it will be of interest in future investigations to study if the mitochondrial respiratory capacity differs between the various leucocyte species in general, but also in response to iron deficiency, as suggested in previous studies [32,33]. However, in contrast to these studies, we did not detect any differences in the mitochondrial outer membrane integrity as determined by the j cyt c , but we found an increase in the mitochondrial density in PBMCs, as shown by the higher CS activity, likewise explaining the increase in the mitochondrial respiration. Nonetheless, the increase in the mitochondrial activity in the PBMCs of the IDA rats was surprising because iron deficiency in vitro has been demonstrated to result in impaired Fe-S cluster synthesis for the respiratory complexes and in reduced mitochondrial activity [10,11,34]. However, most of those studies have been performed in vitro and iron deficiency was often induced by the addition of an iron chelator. Nonetheless, the increase in the mitochondrial density and changes in the quality suggests the presence of a compensatory mechanism, by which the mitochondrial number or the longevity of mitochondria and the quality of mitochondria compensate for the impaired mitochondrial activity as a consequence of the iron deficiency. This is in line with a recent observation indicating that the age of mitochondria determines their metabolic activity [35]. Moreover, in future studies, mitochondrial morphology could be investigated in PBMCs and the livers of the control and IDA rats to gain further knowledge about the underlying causes of the differences in the mitochondrial respiration observed in the present study.
Compared to the results of the mitochondrial activity in cardiomyocytes and the skeletal muscle fibers described in the literature, the mitochondria in the liver did not show a decrease in the mitochondrial respiration induced by IDA in our study [14][15][16][17][18]. Plausible explanations might be that those studies on the morphological changes of the mitochondria in different organs and cells used mostly animals whose iron deficiency or IDA was induced over a longer period of time [14][15][16][17][18][19][20]. In contrast, we used a short-term phlebotomy model to induce IDA that might not reduce the amount of iron in the liver to an extent required for restricting the iron supply to the mitochondria, leading to impairment of the mitochondrial respiration. In addition, the liver is a central organ of iron storage and iron regulation. Therefore, short-term induction of iron deficiency by phlebotomy may be compensated by iron mobilization within this organ, thereby delivering enough iron to maintain hepatic mitochondrial function. It will be of interest to study if long-lasting persistence of iron deficiency changes the phenotype of the hepatic mitochondrial function. This is indicated by observations made in mice with genetic and dietary iron overload, where only long-term exposure to those stressors resulted in altered mitochondrial iron status and activity [36]. Finally, it has to be kept in mind that iron availability affects multiple other metabolic pathways, including tricarboxylic acid (TCA) cycle activity, lipid, and protein synthesis, which may affect the mitochondrial structure and function [37].
Our study suggests that the mitochondrial function in PBMCs and the liver is not comparable, at least in the setting of IDA. Therefore, changes in mitochondrial activity in PBMCs are not associated with alterations of the mitochondrial function in the liver. Our results do not exclude, however, that mitochondrial activity in PBMCs may indicate principal defects or alterations of mitochondrial function present in specific diseases affecting other organs or tissues.

Animal Experiments
All rat experiments were performed as described before [25]. Age-matched female Lewis rats (Charles River Laboratories, Margate, UK) were kept on a standard rodent diet containing 180 mg Fe/kg (Ssniff, Soest, Germany) until they reached an age of 8 to 11 weeks. The animals had free access to food and water and were kept according to institutional and governmental guidelines in the animal housing unit of the Medical University of Innsbruck, with a 12 h light-dark cycle and an average temperature of 20 ± 1 • C. All animal experiments were approved by the Austrian Federal Ministry of Science and Research (BMWFW-66.011/0138-WF/V/3b/2016).
Half of the rats were phlebotomized; 1.8 mL blood was sampled daily for five consecutive days (starting one week before death) to induce IDA. After termination of the experiment, organs and blood were harvested for further analysis. Total blood counts were measured using a VetABC animal blood counter (Scil Animal Care, Viernheim, Germany).

Reticulocyte Quantification
Peripheral reticulocytes were measured using flow cytometry. Briefly, full blood was stained with thiazole orange (Santa Cruz, Dallas, TX, USA), and after identification of the single cells using forward and side scatters, the thiazole-orange-positive cells were identified as reticulocytes.

Isolation of PBMCs
PBMCs were isolated as described before [21]. Briefly, rat full blood was diluted 1:3 with phosphate-buffered saline (PBS, Lonza Bioscience, Basel, Switzerland) and loaded onto Pancoll separating solution (density 1.077 g/mL, PAN-Biotech, Aidenbach, Germany). After centrifugation (1000 g, 10 min, no brake), the buffy coat was collected and the cells were washed twice with PBS.

Plasma and Total Tissue Iron
Plasma iron was measured using QuantiChrom Iron Assay kit (BioAssay Systems, Hayward, CA, USA) according to the manufacturer's instructions. Tissue iron determination was performed as described [27,36,40]. After acidic hydrolysis at 65 • C for 24 h, the iron content was measured using a colorimetric staining solution containing sodium acetate and bathophenanthroline disulfonic acid. Total tissue iron content was normalized by protein content.

CS Activity
CS activity was measured as described [27,36]. A spectrophotometric assay was used to measure the enzyme activity in snap-frozen liver homogenates. The sample enzymatic reaction mix contained 0.25% Triton X-100 in aqua dest, 0.31 mM acetyl-coenzyme A in aqua dest, 0.1 mM 5,50-dithiobis-(2-nitrobenzoic acid) in 1 M Tris-HCl buffer (pH 8.1), and 0.5 mM oxaloacetate in 0.1M triethanolamine-HCl buffer (pH 8.0). The absorbance of the reaction product, thionitrobenzoic acid, was measured at 412 nm over 200 s. The resulting enzyme activities were normalized by the protein content of the samples.

Statistics
Statistical analysis was carried out using the software GraphPad Prism 9. In case of normal distribution, statistical significance was determined by an unpaired two-tailed Student's t-test. Otherwise, a Mann-Whitney U-test was applied. To assess whether PBMC mitochondrial respiration was correlated to liver mitochondrial respiration, a Spearman's rank correlation analysis with a two-tailed Student's test was performed. Data was shown as median ± interquartile range. p-Values below 0.05 were considered significant.