Mitochondrial calcium transport is a highly conserved phenomenon that is linked to a diverse set of cellular processes, including cellular metabolism 
and programmed cell death 
. Given the lack of cell-permeant uniporter inhibitors and the previously unknown molecular identity of the uniporter, the functional studies to date have been largely correlative. The discovery of MICU1 and MCU 
opens the door to an endless set of targeted genetic and biochemical studies that will enable a detailed molecular understanding of mitochondrial calcium uptake and the mechanisms that govern its regulation.
MICU1 is a part of a duplicated gene family found in all major branches of life, including metazoans, plants, protozoa and fungi, with lineage specific losses 
. In the current paper, we show that MICU1, MICU2 and MICU3 are conserved in vertebrates and that they exhibit distinct patterns of expression (), suggesting complementary roles in controlling uniporter physiology. Based on MitoCarta, MICU2 is a high confidence mitochondrial-localized protein whereas MICU3 is likely mitochondrial but with lower confidence 
. We prioritized MICU2 for functional studies, but MICU3 likely has a role in mitochondrial calcium handling in a subset of tissues, notably in the CNS and skeletal muscle, though this remains to be formally proven. Although we showed in previous work that loss of MICU1 in HeLa cells results in impaired mitochondrial calcium handling 
, it was unclear if this property extended to other cell types. Our current study extends this finding to mouse liver, establishing the central role of MICU1 in calcium handling in mammalian mitochondria.
Our biochemical and genetic studies strongly support the notion that MCU, MICU1 and MICU2 reside within a complex. This idea is supported by co-IP experiments as well as by the novel observation of cross-stabilization of these proteins. This was most evident in HEK293T cells in which we performed the majority of our biochemical experiments. Silencing MICU1 resulted in loss of MICU2 protein (), and forced expression of either MICU1 or MCU led to apparent stabilization of MICU1 and MICU2 protein (). Similar effects are observed in HeLa cells, where MICU1 protein levels were also sensitive to MICU2 knockdown (Fig. S2a
). In mouse liver, we observed that silencing MICU1 or MICU2 led to a shift in the large molecular weight MCU complex () previously described 
. Together, these studies reveal that MICU2 is a genuine member of the uniporter complex.
At present, the precise molecular function of MICU1, MICU2 and MICU3 remain unclear. In this study, we performed in vivo
silencing of MICU1 and MICU2, which resulted in reduced mitochondrial calcium clearance in response to large 50 µM calcium pulses. Given the reconstitution data suggesting that MCU is the pore-forming subunit of the uniporter, possible roles for MICU2 could include: (i) Ca2+ sensing and regulation of MCU, (ii) calcium buffering with a secondary impact on transport or (iii) assembly and stabilization of MCU. A recent study provides compelling evidence that MICU1 sets the threshold for mitochondrial calcium uptake without affecting the kinetics properties of the pore 
. How MICU2 contributes to this mechanism remains an important outstanding question.
Although our results show that MICU1 and MICU2 play complementary roles at a physiological level, it remains unclear whether they have distinct or redundant molecular functions. The evolutionary conservation of MICU1, MICU2 and MICU3 in vertebrates, their distinct patterns of expression across organs and the presence of MICU1 and MICU2 in two different cell types indicate complementary roles in cellular physiology. However, it is unclear if they are redundant on a molecular level. We attempted to complement a strong MICU1 phenotype in HeLa cells that we previously reported by expressing MICU2 on a MICU1 knockdown background. Although MICU2 was able to rescue this phenotype, we found that MICU2 also stabilized the protein expression of the small amount of MICU1 (Fig. S2d
), confounding the interpretation. Resolving this question will require the use of a null genetic background in which the activity of one paralog can be rigorously assessed in the absence of the other paralog.
Our biochemical findings have important implications for the interpretation of functional studies of the uniporter that employ genetic silencing or overexpression. The current study demonstrates that when uniporter protein components are genetically silenced, this perturbation may impact the protein stability of companion proteins, and that cross-stabilization may represent a cell-type specific phenomenon. For example, forced expression of MCU has been reported to give a gain of function phenotype, yet we observe that forced expression of MCU also leads to elevated levels of MICU1 and MICU2 in HEK293T cells (). Silencing of MICU1 and MICU2, either alone or in combination, in mouse liver appears to have an impact on the abundance of MCU protein levels (), which likely contributed to decreased mitochondrial calcium clearance in these assays (). Cross-stabilization is not unusual in large, protein complexes, including those of the respiratory chain, where the expression of subunits is nucleated and stabilized by other partners. Future studies of the uniporter need to be cognizant of the ability of MCU, MICU1, and MICU2 to impact each other's protein expression, as genetic silencing studies may misattribute a molecular function to the target protein when indeed the impact may be indirect. In addition, it will be important to consider other mechanisms, including those involving MCUR1 
, LETM1 
and NCLX 
that may additionally influence mitochondrial calcium physiology.
It is tempting to speculate that the relative expression of MICU1, MICU2 and MICU3 differ in a cell and state-specific manner to regulate mitochondrial calcium handling. Under this model, multiple paralogs could be constitutively expressed in a single cell type, but variation in their relative abundance could give rise to functional differences in mitochondrial calcium handling, as has been previously documented across tissues 
. If this model proves to be correct, it may open up the possibility of therapeutically targeting the uniporter in a tissue specific manner.