On the role of Mitofusin 2 in endoplasmic reticulum–mitochondria tethering

The recent paper by Naon et al. (1) claims that their new data “definitively” prove the role of Mitofusin 2 (Mfn2) as an endoplasmic reticulum (ER)–mitochondria tether, supporting their original proposal (2) and arguing against evidence presented by ourselves and others (3–6) suggesting that Mfn2 is a negative regulator of tethering. A careful reading of the paper highlights that Naon et al.’s (1) claims are not supported by the data presented. Below are some pivotal examples.

First, the key parameter (number of ER–mitochondria contacts) that both we and others (3, 4) found doubled in Mfn2^−/− cells is simply not addressed.

Second, the ER–mitochondria distance was determined by electron microscopy to be 2 nm larger in Mfn2^−/− cells, compared with wild-type; using the same cells, no difference was found by two independent groups (3, 4). This difference does not explain the apparent reduction in organelle juxtaposition found by confocal microscopy (1–4), the resolution of which is at best ∼200 nm. The reduced organelle interface length (not observed in refs. 3–5) cannot be an alternative explanation, because (i) Mfn1^−/− cells display a similar reduction and (ii) Mfn2 re-expression in Mfn2^−/− cells does not recover it (tables S1 and S2 in ref. 1).

Third, using an ER–mitochondria artificial tether, Naon et al. (1) show the FRET_ratio [calculated as (FRET_max − FRET_basal)/FRET_basal] being lower in Mfn2^−/− cells (figures 1H and 2A and figure S2D in ref. 1). Based on their own definition, the inevitable mathematical inference is that FRET_ratio tends to approach 0 as FRET_basal approaches the value of FRET_max. Naon et al.’s own data thus demonstrate that in Mfn2^−/− cells organelle apposition is higher than in controls; this is the opposite of the authors’ conclusion.

The Naon et al. (1) paper also contains data inconsistencies; we list only the more substantial cases.

In figure 3 of ref. 1, the peak of ATP-induced mitochondrial Ca²⁺ rise (in Ca²⁺-free medium) in controls (mtYFP) is: ∼60 nM in panel B, ∼400 nM in panel C, and ∼4,000 nM in panel F. Thus, the same parameter, in the same cells, under the same experimental conditions spans two orders-of-magnitude.

In figure 3 I and J of ref. 1, the speed of mitochondrial Ca²⁺ accumulation was calculated with 50 μM CaCl₂ additions, a condition where the rate of Ca²⁺ uptake [in the absence of Mg²⁺, as quoted (7)] is close to the apparent V_max (8). However, in respiring mitochondria, V_max is limited by the activity of the respiratory chain (9, 10), thus being not informative on the intrinsic mitochondrial Ca²⁺ uniporter (MCU) kinetics. In the “typical” experiment presented (figure 3J of ref. 1), the rate of signal decay (representing mitochondrial Ca²⁺ uptake) is clearly slower in Mfn2^−/− liver mitochondria, as we (3) reported in Mfn2^−/− cells (in agreement with lower MCU expression). The data presented by Naon et al. (1), accordingly, do not provide evidence supporting that MCU expression is unchanged in Mfn2^−/− cells.

Naon et al. (1) argue that our finding of lower MCU expression in Mfn2^−/− cells (3) is a result of different cell densities. However, their own data show that, at high density, MCU levels indeed decrease in controls but not in Mfn2^−/− cells (figure S5C in ref. 1).

In short, the paper of Naon et al. (1) fails to address the multiple accrued data (by independent groups) against their proposed role of Mfn2 as a tether and does not add any new evidence reinforcing their hypothesis, but instead, the opposite.