In this study, we have performed the first proteomic profiling of rapamycin treatment in S. cerevisiae,
and used this information for comparative expression analysis with existing expression data measured under different conditions. Our aim was to use this information for identifying novel relationships between regulators of known biological pathways and TOR function. Additionally, we also sought to identify protein abundance changes that could not be predicted from previous microarray analyses of rapamycin treatment 
to gain new insights into TOR function. Although the total number of proteins identified with high confidence (578) was relatively small compared to other proteomic studies in yeast, (most likely due to the charge-neutralizing effect on peptide n-termini of the PIC label incorporated for quantitative analysis 
), we were able to identify abundance changes for 127 proteins upon rapamycin treatment. Among these, 17 proteins were found increased in abundance upon rapamycin treatment that do not show similar changes in their corresponding mRNA transcripts. Among these, increased abundance of Ppx1 and Inh1 upon rapamycin treatment is of particular interest, since Ppx1 overexpression inhibited mTOR activity in mammalian cells 
, whereas inh1Δ
cells were reportedly rapamycin resistant 
. Our proteomic findings thus suggest that the induction of these proteins might potentiate TOR inhibition and promote rapamycin sensitivity in yeast, although further study is necessary to confirm this possibility.
Using comparative expression analysis of our proteomic dataset and existing microarray gene expression data, we observed extensive overlap in gene products affected by rapamycin treatment and conditions of heat/oxidative stress. Although the activation of stress genes by rapamycin treatment has been noted by other groups previously, it has been attributed mostly to the activation of Msn2/4 under these conditions 
. However, a majority of the affected proteins we identified are not known to be regulated by Msn2/4. Additionally, little information currently exists about the other known downstream responses of TOR inhibition to explain the extent of overlap observed between rapamycin treatment and heat/oxidative stress. Preiss et al 
have demonstrated that rapamycin and heat shock induced changes in the transcriptome are amplified at the translational level. However, to the best of our knowledge a direct comparison of the specific genes affected under each of these conditions, as done here has not been reported previously.
Based upon the results of our comparative expression analysis, we hypothesized that the activation of a regulator(s) of heat shock/oxidative stress response inhibits TOR function and/or signaling. Because these stress responses in yeast are controlled by three main transcription factors, Msn2/4 
, Hyr1 
, and Hsf1 
, we explicitly tested for a putative role of their activation in the inhibition of TOR signaling and rapamycin resistance. Unlike other transcription factors tested, Hsf1 is unique since cells constitutively activated for Hsf1 (hsf1-R206S, F256S
cells) specifically display multiple phenotypes consistent with reduced TOR function. Several lines of evidence support this conclusion. First, genes representing five different biological functions (Stress genes, RTG signaling, NCR genes, Glycogen synthesis, and Autophagy) which are inhibited by Tor1/2 in yeast, are all elevated for expression in hsf1-R206S, F256S
cells. Second, multiple ribosomal protein genes (which are known to be down-regulated upon TOR inhibition) are also reduced for expression in hsf1-R206S, F256S
cells. Third, western blotting indicates a faster migrating form of Gln3p in these cells, consistent with reduced phosphorylation of this physiological substrate of TORC1. Fourth, genetic data support that the TORC1 inhibited transcription factors, Msn2/4 and Gln3/Gat1 are activated in hsf1-R206S, F256S
cells. Finally, hsf1-R206S, F256S
cells are hypersensitive to rapamycin treatment in an FPR1
-dependent manner, indicating sensitivity to TOR inhibition.
Elevated expression of specific Hsf1 target genes in hsf1-R206S, F256S
cells contributes to the TOR-regulated phenotypes seen in these cells. This conclusion is based on our finding that deletion of PIR3
suppresses rapamycin sensitivity and PIR3
deletion also augments TOR signaling in hsf1-R206S, F256S
cells. In contrast, their deletion has no effect in wild-type cells (where their expression is baseline compared to hsf1-R206S, F256S
cells). This also explains why PIR3
have not previously been identified in global screens of rapamcyin fitness in yeast 
. Also, neither of these genes have been identified in studies using galactose-inducible overexpression of yeast genes to identify regulators of rapamycin resistance 
. Potential reasons for this include the possibility that galactose-inducible library used by this group did not express PIR3
, or that their overexpression does not inhibit rapamycin resistance on alternative carbon sources such as galactose, or that they act in concert with other Hsf1 target genes to affect TOR signaling and rapamycin resistance. Finally, hypomorphic or dysregulated alleles of hsf1
were unaffected for rapamycin resistance, further supporting a role for Hsf1 activation induced targets specifically in inhibiting yeast TOR.
Additional work is necessary to determine the mechanism(s) by which Hsf1 activation and the resultant elevated expression of PIR3
putatively impinge on the TOR pathway. The cell wall localization of Pir3 and integral membrane localization of the 7-membrane protein, Yro2, places them in proximity to the TOR kinases which are membrane associated themselves 
. It is noteworthy that both TOR and Hsf1 function have been previously implicated as being involved in aspects of cell wall integrity via effects on the PKC/Mpk1 cascade 
, and deletion of genes affecting cell wall integrity can affect rapamycin resistance, and potentially TOR 
. We found that several putative rapamycin protective genes, were decreased for expression in hsf1-R206S, F256S
cells; however, there was no effect of PIR3
deletions on the reduced expression level of these putative TOR regulators in hsf1-R206S, F256S
cells (data not shown). Thus, alterations in their expression levels are unlikely to represent the basis of PIR3/YRO2
mediated effects in hsf1-R206S, F256S
In yeast, TOR signaling has been shown to bifurcate into at least two distinct effector pathways regulated by Tap42/Sit4 and Ras/cAMP/PKA 
. While the former affects NCR gene expression via Gln3/Gat1 activation, the latter regulates the effect of the TOR pathway on RP gene expression and Msn2/4 activation. We have found that hsf1-R206S, F256S
cells are affected in both of these effector branches of TOR signaling, and that PIR3
deletion suppresses ‘readouts’ of both effector branches. Thus, we propose that Hsf1 activation and its target gene products putatively act upstream of these TOR signaling effectors. However, we cannot formally rule out the possibility that Hsf1 activation might also act parallel to the TOR pathway. Additional targets of Hsf1 might play a role in this regulation as well. Further work is necessary using a combination of genetic and transcriptomic or proteomic analyses to identify the entire spectrum of Hsf1 targets involved, and determine their connections with the known upstream regulators of the TOR pathway in yeast.
We have also tested for the effect of TOR inhibition on Hsf1 transcriptional activity. Cells expressing a plasmid borne synthetic reporter of Hsf1 transcriptional activity (HSE-4Ptt-CYC1-LacZ
) were unaffected for LacZ
expression either upon deletion of TOR1
or treatment with various concentrations of rapamycin (data not shown). Additionally, only about 10% of the 165 known direct targets of Hsf1 
are induced in microarray analyses of rapamycin treatment, arguing against a general activation of Hsf1 
. Thus, unlike the stress regulators Msn2/4 and Hyr1, TOR inhibition does not activate Hsf1 under these conditions. Consistent with these results, dietary restriction (which can cause TOR inhibition) in C. elegans
does not significantly activate expression from a reporter of Hsf1 activity (hsp-16.2:GFP, for example) 
. Rather, our results are consistent with Hsf1 activation inhibiting TOR signaling in yeast.
It would be interesting to test if a similar relationship between Hsf1 and the TOR pathway existed in higher organisms as well. Supporting such a possibility, activation of Hsf1 or TOR inhibition promote lifespan in C. elegans 
. However, the effects of TOR depletion are independent of DAF-16 in C.elegans
(unlike that of HSF-1 activation), raising doubt on the possibility that Hsf1 activation promotes lifespan via a putative inhibitory effect on the TOR pathway. Hsf1 activation or TOR inhibition cause clearance of aggregation-prone proteins in higher organisms 
, but it remains unknown if potential connections between Hsf1 activation and mTOR exist and contribute to these phenotypes. Arguing against such a possibility, we have found that Celasterol treatment of Hela cells, (Celasterol causes pharmacological activation of Hsf1 via an unknown mechanism 
), did not cause reduction in phosphorylation of the mTOR subtrate, S6K protein (Bandhakavi S and Griffin TJ., unpublished results). Future studies will shed further light on the possible conservation of yeast Hsf1/TOR relationship in other organisms.
In conclusion, our findings provide intriguing new insights into the relationship between stress signals and cellular growth inhibition. Additionally, our results highlight the value of performing comparative expression analysis between proteomic and genomic datasets to reveal new regulatory connections. Comparative expression analysis is often used in microarray-based analyses of expression changes due to systematic perturbation to find overlapping effects on biological pathways. However, it is usually not an option in quantitative proteomic profiling based studies because of the paucity of protein expression data obtained under various experimental conditions. Our results show that a qualitative comparison of proteomic and transcriptomic datasets, looking for homodirectional changes between among gene products common to these datasets, has value in identifying novel regulatory connections. Such an approach takes advantage of the wealth of microarray based studies that are currently available and can therefore be a useful tool for enhancing the information gained from proteomic profiling studies.