Monday, 30 March 2015

Fatty Acid Trafficking in Starved Cells: Regulation by Lipid Droplet Lipolysis, Autophagy, and Mitochondrial Fusion Dynamics

 Angelika S. Rambold, Sarah Cohen and Jennifer Lippincott-Schwartz

Under nutrient starvation, fatty acids (FAs), which are often stored in lipid droplets, move into mitochondria to drive beta oxidation-based metabolism to sustain energy levels. Exactly how FA become mobilized and delivered into mitochondria is unclear. In this paper they investigate which mechanisms are used to release FAs into the cytoplasm and how FAs move into mitochondria.

They find that FAs are mainly released from lipid droplets by lipolysis (as opposed to lipophagy).
In starved MEFs, almost all lipid droplets were closely associated with mitochondria and this allows FAs to move directly from the lipid droplets into mitochondria. Mitochondria were also highly fused in the starved cells, enabling equilibration of FAs throughout the mitochondrial population. In cells with fusion deficiencies (through knockout of Mfn1 or Opa1), many mitochondrial elements were not close to lipid droplets and FAs did not become homogeneously distributed across the mitochondrial population.

In starved wild-type cells, a rapid increase in FA oxidation was seen, and cells could almost maintain total mitochondrial respiration levels over the entire starvation period (24 hours). Mfn1 knockout cells initially showed an increase in FA oxidation but within 24 hours FA oxidation reduced significantly, causing total mitochondrial respiration levels to drop over time.

It is suggested that delivery of FAs to mitochondria occurs at limited sites, only there where lipid droplets are in close proximity to mitochondria. This may explain why lipid droplets and mitochondria are so closely associated. Having too many free FAs in the cytoplasm can cause damage, so efficient movement of FAs from the droplets into mitochondria is beneficial. Having mitochondria all fused up then enables the FAs to distribute themselves homogeneously throughout the mitochondrial population.

Thursday, 26 March 2015

Lineage correlations of single cell division time as a probe of cell-cycle dynamics

Oded Sandler, Sivan Pearl Mizrahi, Noga Weiss, Oded Agam, Itamar Simon & Nathalie Q. Balaban

Heterogeneity amongst populations of cells is a fundamental observation in cell biology. One example of this is cell cycle duration, which can be thought of as being determined stochastically, perhaps by inheritance of mitochondrial content at mitosis. According to this intuition, mathematical models can be constructed to describe the expected correlation between mothers, daughters and cousin cell cycle periods, such as the bifurcating autoregression model. This model predicts that the correlation between cousins is less than that between a mother and daughter cell

However, time series which appear stochastic in nature can sometimes be derived from underlying deterministic, chaotic, behaviour. The authors set out to determine whether cell cycle distributions are indeed stochastic, or deterministically chaotic. Using Fucci markers, the authors generated lymphoblasts which fluoresced red at G1 phase, yellow at S phase, green at G2 and no fluorescence during mitosis. Using single-cell microscopy, they found that the cell cycle period between cousins had a greater Spearman's correlation (0.63) than between mothers and daughters (0.04). In other words, this is the reverse of what is to be expected from the bifurcating autoregression model, which they label as the 'cousin-mother inequality'.

The authors go on to suggest that the cousin-mother inequality is evidence for deterministic inheritance. They use the Grassberger-Procaccia algorithm on the cell cycle periods, to examine whether the data is stochastic in nature, or in fact deterministic chaos. This returns a quantity called the 'correlation dimension', which is thought to be low for deterministic systems and high for stochastic ones (although see here for subtleties associated with this). They find a small correlation dimension (~3) for the cell cycle period data (whereas random noise is >10), and use this to conclude that cell cycle times are deterministic. They suggest a dynamical model (containing 6 parameters) to explain their data: the 'kicked cell cycle' model. This states that cell-cycle duration is drawn from a deterministic circadian clock oscillator (see Fig. 3D).

It remains to be seen whether there exist alternative stochastic models to the bifurcating autoregression model, which can explain the heterogeneity in cell cycle times. For instance, can the inheritance of mitochondrial content at cell division also recover the correlation between cousins?

Wednesday, 25 March 2015

Proportionality: A Valid Alternative to Correlation for Relative Data

David Lovell, Vera Pawlowsky-Glahn, Juan José Egozcue, Samuel Marguerat, Jürg Bähler

When trying to understand when a quantity covaries with another, a standard set of tools which come to mind are correlation coefficients. But consider three statistically independent variables: X, Y and Z which have no correlation. Plotting a large number of samples from X vs Y, Y vs Z or X vs Z will indeed give small correlation coefficients. However, the quantities X/Z and Y/Z must be correlated due to their common divisor, which can mislead us in believing that X is correlated with Y, which we know is untrue (this is clearly shown in Fig. 1A). Thus, if we are interested in relationships between X and Y, searching for correlation between X/Z and Y/Z can be misleading. It should be noted that this is only a concern when Z is a random variable, with a large enough variance. If Z is a constant number across experiments, then our intuition for correlation coefficients is restored. The lesson is 'correlation between relative abundances is meaningless, if we are using different normalisations for each condition'.

This statistical trap is easy to overlook, as it is commonplace to search for correlations in quantities which are normalised (say mRNA of gene 1/total mRNA, i.e. X/Z).  The authors highlight that if X/Z and Y/Z are proportional across each sample, then X must be proportional to Y. They therefore suggest a 'goodness-of-fit to proportionality' as a more appropriate statistic when searching for covariation in relative abundances. This is defined as ϕ = var(log(A/B))/var(log A), where A = X/Z and B = Y/Z. ϕ is zero when A and B are perfectly proportional.


Update: For enthusiasts!

Let's use some Monte Carlo to test this out! Using the notation Unif(p,q) as a uniform distribution with p as the minimum and q as the maximum. I have generated draws from three uniform random variables: X ~ Unif(1,2), Y ~ Unif(4,8), Z ~ Unif(5, 300). We see that none of the variables correlate with each other. However, when we create new variables A = X/Z and B = Y/Z, we see a striking correlation (0.95). So one cannot claim that X is correlated with Y, just because A is correlated with B.

This is a particularly pathogenic example, since Z has a huge variance. This simulation yielded ϕ=0.11. Check out the comments on this post to see some back-and-forth between David and I on this.

Tuesday, 17 March 2015

Leptin modulates mitochondrial function, dynamics and biogenesis in MCF - 7 cells

Blanquer-Rosselló MD, Santandreu FM, Oliver J, Roca P, Valle A.

Leptin is a hormone that regulates energy expenditure and suppresses food intake. The concentration of leptin in the blood rises as body weight and fat mass increase. Leptin is also involved in many other processes including sex maturation, lactation, immune response and the development of mammary gland. Leptin can increase cell proliferation and inhibits apoptosis in breast cells.

In this paper they investigate the link between leptin and metabolism in breast cancer cells. Cancer cells must rewire cell metabolism to satisfy demands of growth and proliferation. They analyze the effects of a physiological dose of leptin in several features of cellular and mitochondrial metabolism in MCF-7 breast cancer cells.

They find that cellular ATP levels become more reliant on mitochondria in leptin-treated cells and rates of glycolysis decreased. Mitochondrial oxygen consumption increases, but no changes are seen in mitochondrial volume density, respiratory chain proteins or proton leak. ROS levels were decreased and autophagy was increased.
They conclude that leptin ameliorates oxidative stress and increases mitochondrial ATP production in breast cancer cells, which may benefit growth and survival.

Thursday, 12 March 2015

Stable heteroplasmy at the single-cell level is facilitated by intercellular exchange of mtDNA

Anitha D Jayaprakash, Erica Benson, Swapna Gone et al.

Deep sequencing allows the measurement of mtDNA heteroplasmy at a single-cell level. However, regions of mtDNA exist in the nuclear genome (nDNA), called Numts. This is due to ancestral transfer of genetic material from the mitochondria to the nucleus. Numts may have variable sequence and copy number, so failure to separate nDNA from mtDNA can cause inaccuracy in heteroplasmy measurement. Although methods currently exist to deal with this, it is unclear whether they are able to resolve heteroplasmies below 5%.

The authors present a method called Mseek, to enzymatically digest linear nDNA and leave behind circular mtDNA, to a purity >98% (whereas endogenously, mtDNA can form <1% of genetic material). Thus ability to resolve different haplotypes is only limited by sequencing depth.  By inspecting multiple cell lines, the authors found that heteroplasmy is a ubiquitous phenomenon, with most mutations being transitions, indicating replication errors of polymerase-γ.

To determine whether heteroplasmy in cull culture originates from a population of cells homoplasmic for different mutations, or a population of heteroplasmic cells, the authors performed the following experiment. They extracted two single cells from a colony, and passaged them for ~25 generations to derive two new colonies. They then measured single-cell heteroplasmy levels for the two new colonies. They found that the derived colonies were heteroplasmic and had similar haplotype distributions. They show that a simple computational model of random genetic drift would shift the haplotype distribution by a large extent over this many generations. Their observations imply that individual cells are heteroplasmic, and this heteroplasmy is relatively stable over a time scale of ~25 generations. There is therefore a mechanism to counteract random drift.

Exchange of mtDNA between cells could bring the haplotype distribution closer to the average across the population. To test this hypothesis, the authors co-cultured cell lines with distinct haplotype distributions, one of which was GFP-labelled. After 4 weeks of co-culture, one of the cell lines was selected for (either with FACS or antibiotic-resistance), and then cultured for a further 4 weeks. In two of four pairs of cell lines tested, they found transfer of mtDNA from one cell line to the other. In the other two pairs, there was no genetic transfer. Thus mtDNA transfer does appear to occur, but is not universal. Furthermore, it is not a necessary mechanism to counteract genetic drift, but it is perhaps sufficient.

Tuesday, 10 March 2015

The Mitochondrial-Derived Peptide MOTS-c Promotes Metabolic Homeostasis and Reduces Obesity and Insulin Resistance

Changhan Lee, Jennifer Zeng, Brian G. Drew, Tamer Sallam, Alejandro Martin-Montalvo, Junxiang Wan, Su-Jeong Kim, Hemal Mehta, Andrea L. Hevener, Rafael de Cabo, Pinchas Cohen

Summary: A short open reading frame (sORF) encoding the signaling peptide humanin was previously identified in mitochondria, and suggested the possible existence of others. The authors identified an sORF in 12S rRNA which encodes 16-aa peptide they name MOTS-c, which regulates insulin sensitivity & metabolic homeostasis. This peptide targets skeletal muscle, inhibiting folate and purine biosynthesis, which activates AMPK.

Retrograde signalling via ROS, calcium and cytochrome C is a well-known phenomenon, however this signalling mechanism, dubbed 'mitokines' by the authors, represents a novel means of mitonuclear signalling. sORFs have recently emerged as an area of interest as techniques to study them have improved. sORFs have been found throughout mRNAs, and may be translated through leaky scanning or transcriptional re-initiation.

The authors used an in silico search to identify potential sORFs in human 12S mitochondrial rRNA, which they named MOTS-c (mitochondrial open reading frame of 12S rRNA type-c). It appears that although MOTS-c is transcribed in the mitochondrion, it must be translated in the cytoplasm. The possibility that MOTS-c has arisen from nuclear mtDNA transfer (NUMT) was ruled out as no NUMT peptides are homologous to MOTS-c. Mitochondrial depletion showed elimination of 12S rRNA and MOTS-c transcripts in HeLa cells. Fasting reduced MOTS-c in skeletal muscle, testes and plasma, but heart and brain tissues showed no change.

MOTS-c expression alters metabolite levels, as well as changing gene expression profiles. The authors assess that its target is the folate-methionine cycle, which also affects de novo purine biosynthesis. MOTS-c action leads to a significant decrease in purines. The changes to the folate cycle also have the effect of activating AMPK and causing fatty acid oxidation, as well as increasing glucose utilisation.

MOTS-c treatment in mice on a normal diet led to body weight, food intake, and blood glucose reductions, as well as a decrease in circulating signalling proteins implicated in pathogenesis of both obesity and insulin resistance. When mice were fed a high-fat diet, however, MOTS-c treatment prevented obesity without reducing caloric intake. Heat generation significantly increased, as did respiratory exchange ratio (RER). An increase in RER is indicative of a shift towards higher glucose utilisation. This suggests that MOTS-c treatment leads to avoidance of obesity through increase of energy expenditure due to heat generation. MOTS-c appears to act in skeletal muscle. Promisingly, treating mice for 7 days restored the insulin sensitivity of 12 month old mice to nearly the level of 3 month old mice.

Changes in mitochondrial morphology and bioenergetics in human lymphoblastoid cells with four novel OPA1 mutations.

 Shu-Huei Kao, May-Yung Yen, An-Guor Wang et al

OPA1 is a GTPase protein required for mitochondrial inner membrane fusion. Long and short isoforms of OPA1 exist, a mixture of both of them seems to be necessary for normal mitochondrial fusion. Dissipation of membrane potential induces OPA1 cleavage into its short isoforms and this causes mitochondrial fragmentation.

In this paper they investigate four different human OPA1 mutations in lymphoblastoid cells, to find out whether they affect mitochondrial morphology and bioenergetics in different ways. Two of their mutants only have OPA1 short isoforms, the other two also have some long isoforms.

Normal control cells showed a balanced mitochondrial network between filamentous and fragmented states. In all of the OPA1 mutated cells, mitochondria became more fragmented. The proportions of filamentous, intermediate and fragmented networks were 37%, 44% and 19% in control cells, and 1%, 22%, 77% in the OPA1 mutant cells. Membrane potential and ATP concentrations were reduced in mutant cells (ATP concentration ranged from 56% to 63% of that of control cells).
Additionally, all the OPA1 mutants showed decreases in oxygen consumption rate, maximal respiratory rate and increases (3-5 fold) in proton leak. Higher levels of ROS and oxidative damaged were also found in the mutant cells (a 2-fold increase in hydrogen peroxide and a 3-4 fold increase in superoxide). Also, a 4-8 fold increase in lipid-peroxidation was observed. OPA1 deficient cells preferred glycolysis rather than OXPHOS.

No significant differrences between the different OPA1 mutations were observed.

Monday, 9 March 2015

Mitochondrial function and lifespan of mice with controlled ubiquinone biosynthesis

Ying Wang, Daniella Oxer and Siegfried Hekimi

Ubiquinone (UQ) is a central metabolite of the electron transport chain, accepting electrons from complexes I and II, and passing them to complex III. Using Cre-Lox recombination, the authors were able to knock out the enzyme which generates UQ (MCLK1) in adult mice after allowing them to develop normally, as conventional knock out of MCLK1 is embryonic lethal. Two weeks after the chemically-induced knock out, the mice had almost zero MCLK in all tested tissues, besides the liver which was unaffected.  

One might expect the mice to not survive much beyond the intervention, yet astonishingly the median survival was 276 days. UQ became depleted by as much as 90% over 6 months, in the most heavily affected tissues. Blood lactate levels were elevated, suggesting raised glycolysis. Respiratory rates in isolated tissues were reduced, but by no more that a factor of two. Respiratory enzyme activities were raised, likely in compensation for the low UQ availability.  8 months after treatment, the mice had almost zero UQ in heart tissue, and yet were still able to survive. This is particularly surprising, since the heart is such an energetically demanding tissue. Phenotypically, the animals tended to lose body fat, hair and generate a hunched posture. Further studies will be required to elucidate how such severe depletions in this metabolite can be withstood.

Thursday, 5 March 2015

Mitochondrial control by DRP1 in brain tumor initiating cells

Qi Xie, Qiulian Wu, Craig M Horbinski et al.

Brain tumor initiating cells (BTICs) are capable of self-renewal, they are highly proliferative and show chromosomal abnormalities. BTICs are known to take over the glucose transporter GLUT3 so that they can withstand metabolic stress more easily. All tumor cells have dysregulated metabolic pathways, but the highly proliferative nature of BTICs suggests that these tumor subpopulations have some metabolic features that distinguish them from the tumor bulk. In this paper they look at mitochondrial morphology in the most common primary intrinsic brain tumor, gliblastoma.

They compare mitochondrial morphology from BTICs with non-BTIC tumor cells and found out that mitochondria in BTICs are more fragmented and less tubular : in non-BTIC tumor cells, mitochondria show an elongated tubulated structure whereas in BTICs they are shorter and rounded. This suggest that BTICs have increased mitochondrial fragmentation or decreased fusion. It turns out that BTICs show increased phosphorylation of Drp1 at Ser-616 and decreased phosphorylation at Ser-637, both of these changes enhance fission activity of Drp1.

They then checked whether these changed in Drp1 phosphorylation levels were responsible for increased fission in BTICs. A Drp1 gain-of-function mutant, with increased Ser-616 phosphorylation activity and blocked Ser-637 phosphorylation, was expressed in non-BTIC tumor cells. Mitochondria in non-BTIC tumor cells expressing these mutants indeed became more fragmented and less elongated. It also induced expression of some stem cell regulators and repressed some differentiation markers. Expression of the Drp1 mutant was, however, not sufficient to reprogram non-BTIC tumor cells into BTICs.

They then tried to find out whether the change in Drp1 activity is crucial for BTIC maintenance, because it is observed that differentiation of BTICs reduces the hyperactivation of Drp1. Drp1 was knocked down in BTICs using small hairpin RNA lentiviral constructs (shDrp1) , which significantly decreased the growth of BTICs, whereas Drp1 knockdown had no effect on non-BTIC tumor cells or normal neuronal progenitor cells (which are, just like BTICs, capable of differentation and self-renewal). Targeting Drp1 lead to a fourfold decrease in tumorsphere size. They evaluated the potential anti-tumor effects of Drp1 knockdown in vivo in mice, and found that mice with BTICs expressing shDrp1 had reduced tumor formation and increased survival compared to mice with BTICs expressing non-targeted shRNA.

Mdivi-1 is an inhibitor of the GTPase activity of Drp1, and using Mdivi-1 to block Drp1 activity also lead to decreased growth of BTICs. BTICs were implanted into brains of mice, and the mice where then injected with Mdivi-1 which increased mice survival compared with control.

AMPK is a cellular stress sensor. They found that activation of AMPK decreased BTIC growth, and both Drp1 knockdown and Mdivi-1 treatment lead to an increase in AMPK activation. This suggests that the hyperactivity of Drp1 in BTICs may lead to a decreased AMPK activity. Downregulation of both Drp1 and AMPK activity did not reduce the growth of BTICs (as opposed to only downregulating Drp1) which indeed suggests that Drp1 is a critical node in the response of BTICs to metabolic stress through AMPK regulation.

They go on to find the molecular mechanisms that activate Drp1 in BTICs, it turns out that CDK5 and CAMK2 are important. CAMK2 inhibits Drp1 in non-BTIC tumor cells, and its expression is lower in BTICs. CDK5 activates Drp1 and is preferentially expressed in BTICs.

Identification and functional prediction of mitochondrial complex III and IV mutations associated with glioblastoma

Rhiannon E. Lloyd, Kathleen Keatley, D. Timothy J. Littlewood et al.

The role of mtDNA mutations in cancer is subtle. Across most cancers, it has been suggested that mutations are negatively selected for. In this study, the authors provide contrary evidence for glioblastoma multiforme (GBM). They searched for heteroplasmic mutations of complexes III and IV from a mixture of cell lines and human biopsies in GBM patients. They clustered mutations according to three parameters: prevalence in the GBM population; prevalence in the general population; and heteroplasmy level, using ANOVA to check statistical significance. This split the >200 identified mutations into a group of 9 rare functional GBM mutations, and nonfunctional mutations. 43% of GBM tumours carried at least one of the functional mutations.

They mapped the mutations onto bovine crystal structures of the corresponding enzymes, to classify each mutation into one of the following groups: i) frameshift, ii) active site, iii) binding pocket, iv) protein interaction region, v) non-functional. 3D modelling provided mechanistic insight into the function of these mutations.

Wednesday, 4 March 2015

Integrity of the yeast mitochondrial genome, but not its distribution and inheritance, relies on mitochondrial fission and fusion

Christof Osman, Thomas R. Noriega, Voytek Okreglak et al.

In this study, the authors develop a new means of imaging mtDNA nucleoids in budding yeast, without using dyes. They do this by introducing genes into the mitochondrial genome, whose protein products fluoresce. They constructed 3D images of mtDNA and mitochondrial content distributions. They found that the total length of the mitochondrial network correlates strongly with the number of nucleoids, with approximately 1 nucleoid per micron of network. They found that i) the inter-nucleoid distance, and ii) the distance between the network tip and the closest nucleoid, are significantly different to randomly distributed nucleoids. Indeed, they found that nucleoids tend to be preferentially distributed at the tips of the network, aiding mtDNA inheritance to daughter cells. 

After investigating the physiological case, the authors investigated the effect of knocking out mitochondrial fission (Dnm1) and fusion (fzo1). Remarkably, none of the above observations changed by a large magnitude, indicating that nucleoid distribution and inheritance in yeast is essentially fusion/fission independent. The authors show that respiratory-deficient cells accumulate in  fusion/fission knockouts. This suggests that the role of fusion and fission in yeast is for complementation of respiratory defects in individual mitochondria, rather than distribution and inheritance of mtDNA.