This study used Phos-Tag staining and 32
P labeling of porcine heart and liver mitochondria to further characterize the extensive network of mitochondrial phosphoproteins established in previous gel-based screening studies5,6,10,11
. showed that a large number of mitochondrial phosphoproteins were detected with both Phos-Tag and 32
P labeling. Since Phos-Tag stains the steady-state phosphate incorporation into proteins, whereas highly sensitive 32
P labeling is dependent on phosphate turnover rate and pool building, different labeling patterns were observed for these techniques in both heart and liver mitochondria. The inconsistencies revealed from comparing these two techniques have provided insight into the feasibility of using each methodology as well as the inherent characteristics of the heart and liver mitochondrial phosphoproteomes.
The number of reported mitochondrial phosphoproteins has grown extensively. Mass spectrometry studies13,14
have revealed that most phosphoproteins contain multiple phosphorylations, resulting in hundreds of new phosphorylations sites. Importantly, there is good correlation between these mass spectrometry results and phosphoproteins identified from 2D gels labeled with 32
P or phospho-specific dyes5,6,10,11
. With these hundreds of phosphorylation sites, we suggest that determining the turnover of these sites using 32
P will identify the dynamic phosphorylation sites that may represent acute regulatory sites that warrant further attention. Thus, a major focus of this study was to optimize the 32
P labeling methodology in intact mitochondria to better characterize the biochemistry associated with 32
In order to achieve reproducible 32P labeling patterns, the importance of maintaining matrix [ATP] was determined. Independent of whether the mitochondria held a membrane potential, exhibited a strong RCR or showed an appropriate net state 3 rate, the matrix [ATP] upon warming and re-energization was found to be critical for reproducible 32P labeling patterns. The major variable in this process was matrix [Pi], which was depleted in our isolation procedures, but was replenished adequately by a brief Pi incubation during the isolation process.
The dynamic nature of 32
P incorporation is an advantage, but also a complication, as outlined in the time-course and chase experiments. If 32
P association resulted solely from rapid turnover, then the extent of labeling would be relatively independent of the order of additional 32
P and cold phosphate. However, the majority of 32
P labeling was not exchanged out by a cold Pi chase. A good example of this was PDHE1α, which remained heavily phosphorylated after the cold Pi chase in both heart and liver mitochondria. This data are consistent with the de-phosphorylation of PDH during the mitochondrial isolation process, followed by a re-phosphorylation with warming and de-energization, as demonstrated in earlier functional studies from Randle’s lab18
. Several other phosphoproteins revealed this pattern of non-exchanging pools, including VDAC1 and elongation factor, tau in the heart and electron transfer flavoprotein, β-subunit and serine hydroxymethyl-transferase in the liver. In the heart experiments, Succinyl-CoA Synthetase, α-subunit and the 23kDa subunit of Complex I were exceptions; these proteins did significantly exchange out 32
P implying a rapid exchange of phosphate. These chase experiments demonstrated that 32
P incorporation into proteins was not merely the result of rapid phosphate exchange, but also due to the recovery of phosphorylated protein pools upon re-energization.
Using both Phos-Tag staining and 32P labeling provided specific information on the nature of the phosphate-protein association. Good correlation of 32P and Phos-Tag provided evidence that the degree of phosphorylation was significant with a significant turnover, or pool expansion. Examples of strong labeling with both 32P and Phos-Tag in heart and liver mitochondria include PDHE1α, heat shock proteins 60 and 70, aconitase, and Complex V, β-subunit PDH E2, VDAC1, Complex I 23kDa subunit, aconitase and Complex IV Va subunit (, and ).
Several proteins exhibited high 32P labeling with low Phos-Tag staining, suggesting a rapid turnover into a fraction of the phosphorylation sites or simply a low total abundance of the protein (, and ). These two conditions can be differentiated using the total protein stains. For example, succinyl-CoA synthetase, α-subunit, γ subunit of Complex V, and aconitase in the heart and carbamoyl-phosphate synthase and 3-hydroxy-isobutyrate dehydrogenase in the liver all exhibited strong 32P labeling and relatively weak Phos-Tag staining. However, the total protein levels suggested that the lack of Phos-Tag labeling in succinyl-CoA synthetase and 3-hydroxy-isobutryate dehydrogenase was consistent with low protein content. In contrast, the significant amount of aconitase and carbamoyl-phosphate synthase protein suggested a low overall percentage of steady-state phosphorylation, with a relatively high turnover. Another explanation for high 32P incorporation and low Phos-Tag labeling may also be ascribed to proteins undergoing 32P metabolite associations, and not protein phosphorylation (discussed below). Such 32P associations include enzyme catalytic sites that retain high affinity, even in the presence of denaturing SDS.
Another labeling pattern involved strong Phos-Tag staining, but weak 32
P incorporation (, and ). This result suggested that a protein was abundantly phosphorylated in steady-state, but that its phosphate turnover rate was relatively slow in the 20min time-course of these experiments. This behavior was demonstrated in both heart and liver mitochondria by Mn-SOD, Complex I 30kDa subunit, succinate dehydrogenase flavoprotein subunit, and thioredoxin-dependent peroxide reductase. Interestingly, the branched-chain α-ketoacid dehydrogenase complex E1α (BCKDH) labeled strongly with Phos-Tag in heart, but was not detected with 32
P staining, as has been previously demonstrated42,43
. Given the proximity of BCKDH to the strongly 32
P labeled elongation factor, Tu and PDH E1α, it is likely that BCKDH’s 32
P labeling may be masked in our 2D gel studies.
Additionally, many proteins showed strong 32P labeling in regions where no Phos-Tag or total protein stain was observed. This pattern suggested a very high specific activity and turnover rate for these low abundant proteins. In general, these proteins were not identified in this study due to their low concentration and the limits of our mass spectroscopy identification. The unidentified 32P labeled proteins deserve further investigation, since they may represent signaling molecules that are in low concentration in order to permit rapid changes in phosphorylated protein levels during signaling processes.
There were also some interesting discrepancies between 32P incorporation, Phos-Tag and IEV elements within given proteins. For many proteins, 32P preferentially incorporated into the far acid shifted IEVs, while Phos-Tag stained all IEV elements, including the most alkaline, parent IEV ( and ). 32P labeling of the most acidic IEV species implied that the turnover rate of the acid shifted IEVs were much higher than the alkaline IEVs species. Examples of this behavior are observed in heart and liver mitochondria for heat shock protein 70 and aconitase, and in the purified protein studies for the α- and d-chain subunits of Complex V and the 51kDa subunit of Complex I (). The reason for high phosphate turnover in the acidic IEV species is unknown and not predicted by conventional interpretation of the IEV phenomenon being simple summing of individual post-translational modifications.
Another discrepancy between 32
P labeling and Phos-Tag within a protein involved an increased molecular weight shift of a protein’s 32
P labeled component, relative to its Phos-Tag or total protein stain. An example of this was Complex V’s β-subunit, which showed 32
P labeling well above its Coomassie stained IEVs (), but Phos-Tag staining on all components, including the parent protein IEV (). Comparing the Phos-Tag and 32
P labeling patterns of Complex V suggested that a large fraction of the β-subunit had steady-state phosphorylation, while a small component of the protein experienced very rapid phosphate exchange with slightly higher molecular weight incorporation, consistent with metabolite-binding such as ATP or ADP. Since sites of phosphorylation have been identified for Complex V’s β-subunit using mass spectrometry44
, these sites most likely represent the abundant, steady-state phosphorylations depicted using Phos-Tag. The increasing molecular weight shift of the 32
P labeled components of Complex V’s β-subunit was evaluated further in an isolated protein study, which is discussed below.
Comparing the phosphoproteomes of heart and liver mitochondria () revealed that 32
P labeling is generally weaker in liver mitochondria relative to heart, with the exception of PDHE1α. The DIGE image presented in demonstrates that there are differences in protein content between tissues, consistent with the notion that mitochondria are fine tuned to the functionality of a given tissue24
. However, these differences in protein content cannot alone explain the radical differences in 32
P incorporation between heart and liver mitochondria. Interestingly, proteins associated with energy-metabolism (i.e., aconitase, Complex IV, Va subunit, and Complex V, α- and β-subunits) are labeled much more intensely with 32
P in the heart, whereas proteins involved in biosynthetic processes (i.e., carbamoyl-phosphate synthase, aldehyde dehydrogenase and 3-hydroxy-isobutyrate dehydrogenase) incorporate 32
P intensely in the liver. Whether phosphorylation is regulated in a tissue-specific manner to meet the different metabolic demands of heart and liver mitochondria is relatively unknown and requires further investigation.
Utilizing minimally disruptive native gel electrophoresis, we observed that a large fraction of the 32
P association was removed upon denaturing with heat or acid (). For example, Complex V had the most intense 32
P labeling in the native gels. However, upon denaturing, the 32
P labeling of this band was dramatically reduced to a level consistent with that observed in our 2D gel studies, which denature proteins with SDS. This pattern held true for all five complexes of oxidative phosphorylation, and suggested that many mitochondrial protein complexes have a high degree of weak, phosphate-metabolite interactions in their native form. These weak associations may be related to active enzyme sites or allosteric binding sites on individual proteins or within protein complexes. Furthermore, these associations potentially provide a mechanism for phosphate/phosphate-metabolites to modulate oxidative phosphorylation at several levels, as has been suggested in previous studies for Complex IV45,46
and, naturally, Complex V which uses these metabolites in its catalytic activity (discussed in more detail below). Thus, the native gel system may provide an extremely useful tool in screening for weak, phosphate-metabolite interactions that may regulate enzyme activities. It is important to note that Phos-Tag labeling in ghost native gels was considerably less sensitive to washing with acid (), implying that this approach primarily detected abundant, steady-state phosphoproteins and not the weak interactions observed with 32
Although this study aimed to screen for mitochondrial protein phosphorylation, the native gel studies (discussed above) and the purified protein studies (discussed below) clearly demonstrate that phosphate-metabolites are formed and likely contribute to the overall 32
P labeling pattern observed in our 2D gel studies. The binding of phosphate-metabolites may be responsible for the 32
P labeled components that increase in molecular weight above a given protein’s Coomassie stained IEVs. In addition to the formation of radioactive ATP and ADP upon the incubation of intact mitochondria with 32
P, several other metabolites may form in the 20min time-course of our experiments and result in additional covalent protein modifications, such as ADP-ribosylation. Interestingly, the biosynthesis of NAD in mitochondria is on the order of minutes47
. Furthermore, nicotinamide monoculceotide adenylyltransferase-3 (NMNAT3), a central enzyme of NAD biosynthesis, is localized to the mitochondria48
. Thus, while the majority of 32
P labeling observed in our 2D gel studies is likely due to phosphorylation, the contribution of phosphate-metabolites is possible and will require further study.
We focused on Complexes V and I to confirm identifications of their 32
P-labeled subunits in the more complex gels and to better characterize the nature of these Complex’s phosphate-associations. Phos-Tag staining of purified Complex V revealed labeling of the α, β, δ, d, OSCP, and e-chain subunits (). Good correlation between Phos-Tag labeling and total protein IEVs with modest, implied that these proteins are abundantly phosphorylated. 32
P association was found in α, β, γ, and d-chain subunits () suggesting these elements have rapidly exchanging protein phosphorylation sites. Since the purified 32
P labeled components of the β-subunit are shifted in molecular weight above its IEVs (as observed in the total protein study; ), this supports the notion that the 32
P labeled protein originates from Complex V, and is most likely an ATP or ADP association. Our observation that the β-subunit labels in intact mitochondria in the presence of oligomycin (results not shown) suggests that this labeling may primarily result from ADP incorporation. The minute fraction of the β-subunit involved in this binding may represent the active catalytic sites or other metabolite association sites. Purified Complex V was also used to confirm identification of the far acid-shifted 32
P components of the α- and d-chain subunits in the more complex 2D gel studies. As observed for several matrix proteins, this acid-shifted 32
P incorporation pattern indicates that a tiny fraction of the α- and d-chain subunits exchange phosphate rapidly, whereas Phos-Tag labeling on the most alkaline components of these proteins suggests that a large steady state fraction is phosphorylation. Thus far the role that phosphorylation may play in the regulation of Complex V’s activity has yet to be resolved. However, these studies imply that phosphorylation of the γ-subunit may be the most interesting with regards to the extent of its 32
P label and its reported sensitivity to dephosphorylation with calcium5
Purifying Complex I from porcine heart mitochondria resulted in the identification of 13 subunits (). Although Complex I 45–46 subunits29,49–52
, previous studies in bovine heart have demonstrated that 26 of these subunits migrate on SDS gels in the molecular weight range of 10–20kDa29,53,54
, which is below the level of detection in our system. Phos-Tag staining of purified Complex I () revealed labeling for the 75kDa, 51kDa, 49kDa, 42kDa, 30kDa, 24kDa, 23kDa, 19kDa, 18kDa, 15kDa, 13a kDa, and 8B subunits. As observed in the total protein study, this purified Complex I study revealed good correlation between the Phos-Tag staining and the total protein IEVs, implying abundant steady-state phosphorylation. To the best of our knowledge, 7 of these Phos-Tag labeled subunits are novel phosphorylations, including the 51kDa, 24kDa, 19kDa, 15kDa, 13a kDa, 13b kDa, and B8 subunits. Although the nature of phosphorylation for Complex I’s 18kDa subunit is still a matter of debate11,55–63
, our purified study demonstrates good resolution and strong Phos-Tag labeling for the 18kDa subunit, consistent with it being phosphorylated in steady-state. 32
P incorporation was observed in the 75kDa, 51kDa, 42kDa, 23kDa, and 13a kDa subunits. 32
P incorporation into 51kDa and 42kDa subunits was partially masked by the proximity of PDHE1α. However, 32
P incorporation was observed for the far acid-shifted IEVs of the 51kDa protein, suggesting that a relatively small fraction of the protein is turning over rapidly. Although the exact role that phosphorylation may play in regulating Complex I remains largely elusive, studies have demonstrated that mutations disrupting phosphorylation sites in specific Complex I subunits (i.e., 18kDa64,65
) can result in lethal phenotypes. To this effect, phosphorylation of the 51kDa may be the most interesting because it labels intensely with 32
P, carries the NADH-binding site66
, has a high degree of conservation—underlining its functional importance67
, and its mutation has specifically been implicated in several clinical ailments68