IRS-1 and -2 as intracellular signaling nodes contain numerous serine and threonine residues that can be modified by protein kinases and protein phosphatases. For IRS-1 this has been shown extensively 
. Additionally, the interaction of IRS-1 and 14-3-3 has been investigated in some studies. 14-3-3 binding to IRS-1 was found to be influenced by insulin in HepG2 cells 
but not in 3T3-L1 adipocytes 
. Xiang et al. proposed a role for 14-3-3 in IRS-1 trafficking in COS-7 cells 
, whereas Oriente et al. and Kosaki et al. observed an influence of 14-3-3 on IRS-1 associated kinases and their activity in NIH-3T3 fibroblasts and 3T3-L1 adipocytes 
Less is known about individual serine/threonine phosphorylation sites on IRS-2, the interaction of 14-3-3 with IRS-2 and the influence on cellular metabolism. The IRS-2 interaction with 14-3-3 proteins has been documented in large-scale proteomic studies, but only as one of many interaction partners without any further characterization 
. Ogihara et al. showed interaction of overexpressed IRS-2 and 14-3-3 in Sf9 cells 
and Dubois et al. showed IRS-2 as 14-3-3 interaction partner upon IGF-1 stimulation involving the PI 3-kinase pathway in HEK293 cells 
This interaction was verified here in different cell culture models and in the liver of mice in vivo. In addition, we present here a pattern of 24 phospho-sites of IRS-2 after IGF-1 stimulation and the IGF-1/insulin-regulated 14-3-3 binding region on IRS-2 involving the novel IRS-2 serine phosphorylation site 573. We showed that upon insulin or IGF-1 stimulation 14-3-3 associated with IRS-2 in transiently transfected HEK293 cells, in Flp-In HEK293 cells that stably expressed IRS-2 and in Fao cells, which expressed IRS-2 endogenously. In addition, we could provide evidence that the IGF-1/insulin-dependent activation of the PI 3-kinase-Akt/PKB pathway was necessary for the IGF-1/insulin-stimulated phosphorylation of serine 573 and the 14-3-3 binding of IRS-2. Our data suggest that serine 573 potentially is involved in the negative regulation of IGF-1/insulin signaling, since the 573 alanine mutant led to increased Akt/PKB signaling. In mice liver 14-3-3 associated with IRS-2 upon insulin stimulation and refeeding, thus indicating the physiological occurrence of this interaction.
We used the IRS-2 mouse sequence to identify novel phosphorylation sites on IRS-2 and the mass spectrometry approach revealed 24 phosphorylated serine/threonine residues. These sites seem to be conserved in human IRS-2 since all sites could also be found in the human IRS-2 protein sequence. In addition, certain sites could be found in the corresponding human and mouse protein sequences of IRS-1. Phosphorylation of serine 675 and serine 907 after insulin stimulation has already been investigated and described by our group 
. Serine 303 has a homologue in the human IRS-2 protein sequence - serine 270. This site has been described as substrate for S6K1 and as part of TNF-α-induced insulin resistance 
. Little is known about the other sites that were identified by our screen. Truncated versions of the IRS-2 protein, spanning 5 different segments, revealed the area residing between amino acid position 301 and 600 to be important for 14-3-3 binding. Within this region our mass spectrometric approach identified several phosphorylated residues, but only mutation of serine 573 to alanine abrogated the 14-3-3 interaction with IRS-2 in the overlay assay. Ser-573 and the surrounding amino acids (RKRTYS573
SLTT) do not completely conform to a 14-3-3 binding motif with the lack of proline at position +2 relative to the phosphorylated serine being the most prominent feature to be different. However, a proline in positin +2 is found in only 50% of the target sites as has been reported by Johnson et al 
. In addition, if the target sequence would match the consensus motif completely, the strength of binding would not give appropriate regulation of this protein 
. The phosphorylation of serine 573 was increased upon IGF-1/insulin stimulation in a PI 3-kinase-Akt/PKB-dependent manner, thus under the same conditions when the 14-3-3/IRS-2 interaction took place. Notably, pharmacological inhibition of PI 3-kinase using wortmannin or PI-103 or inhibition of Akt/PKB using Akti-1/2 prevented the enhanced phosphorylation of serine 573 and the increased 14-3-3 binding of IRS-2.
We focused on the characterization of serine 573 since a physiological importance of serine 573 is suggested by the fact that sequence alignments of different mammalian species showed complete sequence accordance, and even an amphibian sequence showed partial accordance. We could observe phosphorylation of serine 573 of mouse IRS-2 with insulin and IGF-1 and also phosphorylation of the corresponding residue serine 577 in human IRS-2. A time course of serine 573 phosphorylation in Fao cells showed peak phosphorylation levels after 60, 90 and 120 min of insulin treatment. Moreover, mutation of this site to alanine displayed clear effects on IGF-1/insulin signal transduction.
IRS-1 and IRS-2 display a similar architecture and partially overlapping functions, therefore the IRS-1 sequence was screened for a homologue position corresponding to serine 573 of IRS-2. Giraud et al. described the homologue serine 522 in IRS-1 
. They showed insulin-dependent phosphorylation of serine 522 in L6 myoblasts and myotubes and identified Akt/PKB or a kinase downstream of Akt/PKB as kinase for its phosphorylation and provided evidence for a negative role of serine 522 phosphorylation. Serine 573 on IRS-2 also resides in an Akt/PKB consensus motif (RXRXXpS/T). Treating cells with an Akt/PKB inhibitor resulted in significantly reduced phosphorylation of serine 573 and further experiments showed increased Akt/PKB phosphorylation upon IGF-1 stimulation in the S573A mutant compared to IRS-2 wild type. In conclusion, our data suggested comparable regulation and function of the homologue serine 573 and 522 in IRS-2 and IRS-1, but the implication of serine 522 in 14-3-3 binding of IRS-1 has not been reported yet.
The data of the overlay assay indicated that only one phosphorylated residue on IRS-2 was sufficient to trigger IGF-1/insulin-dependent 14-3-3 interaction with IRS-2. This would be in line with several other studies which also identified single phosphorylated residues as 14-3-3 interaction site 
. However, there are also other reports that showed 14-3-3 binding to two simultaneously phosphorylated serine/threonine residues on the same target protein (see 
for a collection of reported binding partners). Serine 556 appeared in the study from Dubois et al. to be phosphorylated after insulin stimulation in HeLa cells and we identified the same residue in our mass spectrometric approach. This opened the possibility that perhaps serine 556 and serine 573 were 14-3-3 binding sites. The S556A mutant showed regulated binding of 14-3-3 in the overlay assay, hence arguing against S556A as a 14-3-3 binding site. To support these findings with another methodical approach GST pulldown experiments were conducted. Unexpectedly, both single mutants S556A and S573A did interact with GST-14-3-3 and also the double mutant S556A/S573A failed to disrupt the interaction. The cause of this discrepancy between the overlay assay and GST pulldown cannot be explained yet, but we speculate that the S573A mutant probably failed to disrupt interaction of 14-3-3 with IRS-2 in the GST pulldown due to the presence of other proteins stabilizing 14-3-3 binding or preventing regulated binding of exogenously added GST-14-3-3 protein. Of course a second IGF-1/insulin-dependent 14-3-3 binding site on IRS-2 cannot be excluded.
How can the persuasive result of the overlay assay be explained? A possibility is partial renaturation of the IRS-2 protein on the nitrocellulose membrane. The renaturation process probably was effective enough to enable binding of 14-3-3 to the phospho-serine 573 motif, whereas the second binding site might not have been easily accessible and full renaturation of the IRS-2 protein would have been required. This may have led to abrogation of 14-3-3 binding in the S573A mutant in the overlay assay, despite the presence of a second binding site.
The 14-3-3 protein family shows high overall sequence conservation, in the central binding groove the residues are strictly conserved 
and the phosphopeptide bonds are build with 3 absolutely conserved residues 
. In addition, the high level of conservation is emphasized by the fact that 14-3-3 isoforms from yeast, plant and mammals are functionally interchangeable 
. The aforementioned properties show that the two 14-3-3 isoforms β and ε used in this study for the GST pulldowns and 14-3-3ζ for the overlay assays were adequate to investigate 14-3-3 binding sites on IRS-2.
How could the interaction of IRS-2 with 14-3-3 proteins influence downstream insulin signaling? The rigid structure of the 14-3-3 proteins led to the hypothesis that 14-3-3 proteins provide themselves as an anvil whereon the target protein can be restructured/reshaped 
. Experimental data from Datta et al. showed that upon 14-3-3 binding to serine 316 of BAD (Bcl-2-associated death promoter) the accessibility of serine 155 of BAD for other kinases increased 
. GlobPlot® analysis revealed unstructured regions in the IRS-2 protein, which could potentially be restructured/reshaped upon 14-3-3 binding. Since IRS-2 functions as a multi-adapter protein, thereby linking the insulin receptor and other receptor tyrosine kinases with intracellular effector molecules, 14-3-3 proteins could modulate the interactions with binding partners by binding to IRS-2 upon phosphorylation of serine 573 and a possible yet to be discovered second binding site.
In summary, we demonstrate the identification of 24 phosphorylation sites including novel sites on IRS-2 and mapped the IGF-1/insulin-dependent 14-3-3 binding region. Serine 573 as novel phosphorylation site was shown to be part of IGF-1/insulin-dependent interaction between 14-3-3 and IRS-2 and its phosphorylation may have an inhibitory role in IGF-1/insulin signaling. The evidence of the interaction between 14-3-3 proteins and the signal transducer IRS-2 in vivo opens several novel perspectives in the (patho)physiological regulation of the biological function of IRS-2 including its role in metabolic disorders such as insulin resistance and type 2 diabetes.