Pere Puigserver, Zhidan Wu, Cheol Won Park, Reed Graves, Margaret Wright, and Bruce M. Spiegelman Cell. 1998, 92: 829-839.
Dana- Farber Cancer Institute and Department of Cell Biology Harvard Medical School, Boston, Mass. 02115

Introduction:
Before examining the paper, the basics of thermogenesis and coactivators must be understood. Adaptive thermogenesis is the process in which your body responds to changes in temperature or excessive caloric intake. The energy that is used up in this process comes from brown adipose tissue (BAT), commonly called fat. However, BAT is actually a good type of fat since it burns up calories trying to maintain homeostasis (stability) in the body, while white adipose tissue (WAT), the energy storing fat, is what people normally battle to lose. Unfortunately, humans and other larger mammals do not have large BAT deposits as do mice, but BAT might be mixed in with the WAT (Garruti and Ricquier, 1992). Adaptive thermogenesis and BAT are affected by beta- adrenergic agents and thyroid hormone (Himms- Hagen, 1989) which regulate fuel oxidation and ATP synthesis in the mitochondria (Nicholls et al., 1986). Also located in the mitochondria and linked to adaptive thermogenesis are three types uncoupling proteins (UCPs). This paper deals only with UCP-1 which is specific for BAT (Jacobsson et al. 1985).

The definition of a coactivator is somewhat intuitive in that it is a molecule that works with other molecules to activate a certain function. Coactivators can do this even when relatively distant from the promoters for the nuclear receptors. However, there are some typical characteristics that a coactivator of a nuclear receptor has. Some examples are homologous sequences, low specificity for regulation of their expression, low specificoty for the receptors with which they interact, and an LXXLL sequence that binds at the ligand- regulated helix 12 in the carboxy terminal AF- 2 domain (Heery et al., 1997). Coactivators normally have either histone acetyltransferase or histone deacetylase activities (Pazin and Kadonaga, 1997). These traits are important in order to decide whether or not Puigserver et al. have actually found a coactivator of nuclear receptors.

As a last introductory note, here are the names of some nuclear receptors that are important in adaptive thermogenesis and will be used through out the review of this paper: thyroid hormone receptor (Silva, 1995), PPARd (Sears et al., 1996), PPARd/ RXR (retinoid X receptor) heterodimer (Tontonoz et al., 1994).

Research by Puigserver et al.
Identification of Novel Protein and Anaylsis of PGC-1 Sequence (Figure 1)

The first step to finding an activator or coactivator was to find a protein that interacts with a known factor in adaptive thermogenesis. PPARd was chosen as the known protein because it activates the Ucp-1 gene in brown fat cells. The yeast two- hybrid system was used to screen for the potential activator. PPARd was the known protein bound to GAL4 (1-147). A cDNA library (130 clones) provided the unknown proteins bound to GAL4 (768-881). By screening for lac Z activity, a new protein was identified that interacts with PPARd and was named PGC-1. Figure 1 is the sequence of the open reading frame of PGC-1 cDNA as well as the schematic representation of the gene. Some sequences were easily identified. PGC-1 includes a SR and RNA binding domain, three consensus sites for phosphorylating protein kinase A, and an LXXLL motif. Pertinent information obtained from this analysis is the molecular weight of PGC-1, 92 kDa, which will allow the protein to be identified on a gel. Also, it seems that the LXXLL motif, a common trait of coactivators of nuclear receptors, led Puigserver et al. to define PGC-1 as a potential coactivator. This assignment could have been a little premature, as you will see in the later results.

Although Puigserver et al. had identified an interaction between PPARd and PGC-1, little was known about it. Therefore, the researchers designed a series of experiments to see exactly what and when PGC-1 would bind to. Figure 2a tests which nuclear receptors PGC-1 will bind to and whether the reaction is ligand dependent. PGC-1 is fused to the protein GST which will be held in place by glutathione agarose beads. PPARd, thyroid hormone receptor (TRb), retinoid X receptor (RXRa), retinoic acid receptor (RARa), and estrogen receptor (ERa) were the nuclear receptors tested for binding affinity to PGC-1. Each receptor corresponded to four lanes: GST only, GST and ligand, GST fused with PGC-1 only, and GST fused with PGC-1 and ligand. The two GST lanes were negative controls, and no bands were seen as was expected. From the bands seen, we can conclude that PPARd interacts with PGC-1 and is not ligand dependent since the band intensities have remained constant. TRb also binds to PGC-1, but there seems to be a slightly higher affinity for PGC-1 in the presence of the T3 ligand. They cite this increase as being a 2 to 3 fold higher affinity. RXRa seems to have little interaction with PGC-1 with or without the ligand. RARa does not bind much with PGC-1 alone, but with the RA ligand a significant increase in intensity is seen. Finally, PGC-1 and ERa do interact, but seem to have a higher binding affinity with a ligand. Therefore, figure 2a reveals that PGC-1 does interact with a range of nuclear receptors, but the interaction with one nuclear receptor is very specific to that nuclear receptor. The controls used for this experiment were adequate. The basic difficulty with this figure is the report of the quantitative increase in binding for TRb and not for any other ligand dependency. Besides being inconsistent with reporting results, it does not explain how the quantitative number was obtained.

The interactions in figure 2a took place in vitro and not in vivo. In order to verify that the results were not just artifact, the interaction of PGC-1 with PPARd and RXRa in vivo are shown in figure 2b. Cell extracts were immunoprecipitated with an anti- PPARd antibody. The bound proteins were then analyzed by a Western blot in which the probe was anti- PGC-1. The results confirmed that there is an interaction between the two proteins in a living organism. In figure 2c, PCG-1 was fused with green fluorescent protein in order to see the localizaton in the cell. PGC-1 is found in the nucleus where it should be to bind the nuclear receptors.

Since the LXXLL motif is the suspected place on coactivators for binding to nuclear receptors, specific deletions of PGC-1 were constructed and then added to a GST- PPARd fusion protein and autoradiographed to check for interactions. Figure 3a is an autoradiograph of the different lengths of PGC-1. From the figure, the reader can see that PGC-1 is decreasing in size. However, there are no molecular weight markers and the reader must trust Puigserver et al. that the lengths (i.e. 797, 675, 503, 403, 338, 292 amino acids) they report are true. Figure 3b shows the interaction between the deleted sections of PGC-1 and GST- PPARd. Once again, the negative controls are GST only. However, there is a problem with the negative control corresponding to the full-length PGC-1 protein. A small band shows up that is the same length as the interaction between PGC-1 and PPARd. No mention of this is made, even though it is very important whether PGC-1 binds to GST or if some other problem occured. Interactions are seen with all PGC-1 portions except for the smallest one, which is 292 amino acids. This is significant because includes only the LXXLL motif which is the normal region for binding between a coactivator and a nuclear receptor. Therefore, it seems that PGC-1 binds with the residues from 292- 338. This region is mostly proline and is very significant because no other coactivator is known to bind nuclear receptors in this manner. Figure 3c is a schematic diagram showing where the deletions occur and a quantitative percent of binding.

Likewise, PPARd also has a normal place of binding to coactivators. Figure 4a and b, show three PPARd deletions that were made to see if the full-length PGC-1 protein would continue to associate with it. The only section of PPARd that interacted with PGC-1 was actually the original section used to screen for novel proteins in the yeast two hybrid system. Therefore, the C- terminal AF-2 domain that docks other coactivator is not sufficient to bind PGC-1. Instead, the DNA binding and hinge domains are necessary. According to Puigserver et al., no mistake was made in this experiment because the deletions of PPARd continued to bind other coactivators. This data was not shown. Puigserver et al. could have strengthened their interpretation of the data if a construct containing only the DNA binding and hinge domains interacted with PGC-1 because they have based their conclusions on a negative result instead of a positive result.

As written in the introduction, adaptive thermogenesis deals with changes in environmental conditions. Therefore, placing the system in a colder temperature should increase the activity of molecules associated with adaptive thermogenesis. The next step taken was to confirm that PGC-1 was upregulated in colder temperatures. At a normal temperature, mRNA of PGC-1 is found only in the heart, kidney, brain and brown fat tissue in mice (figure 5a). When brown fat tissue is isolated and exposed to cold, an increase in PGC-1 mRNA is observed from no detectable band at zero hours to a very large band at twelve hours. UCP-1 which has already been linked to adaptive thermogenesis also shows an increase of mRNA at the same times. Likewise figure 5c has other key mitochondrial proteins (ATP synthase, COX-II and COX IV) in different tissues which have been exposed to cold. No upregulation of these proteins was seen. This result was explained that longer exposure to cold was needed. The researchers could have reinforced there interpretation of results by repeating the experiment, but increasing the amount of time the tissues were exposed to cold. Since they did not prove that the adaptive thermogenesis proteins would increase in expression with cold, it allows the reader to doubt some of their claims.1 Still, the upregulation of PGC-1 and UCP-1 can be accepted because a control RNA is shown that remains constant.

Beta- Adrenergic agonists simulate cold exposure. Figure 6 shows that PGC-1 and UCP-1 are both present with beta adrenergic added and that these proteins increase proportionately as beta- adrenergic increases with respect to the control. This connects beta- adrenergic to the regulation of the proteins.

CAT assays test for acetyltransferase activity within a molecule (Figure 7). The UCP-1 promoter, which has binding sites for PPARd and TR, was used for the assays. Mixtures of different nuclear receptors and PGC-1 were analyzed to see which combination produced the largest migration. This corresponds to the highest acetyltransferase activity. This is the figure that determined PGC-1 was a COactivator and not an activator because no difference was seen in the CAT assay when just PGC-1 was used. However, when used in conjunction with other ligands, a much higher activation was seen. It can be concluded that PGC-1 is involved in some synergistic effort to promote the transcription of PPARd and TR. However, it is still unknown how it induces the transcription. Therefore, PGC-1 was fused with GAL4 and examined for activation of transcription. PGC-1 could still activate the transcription while SRC-1 (another coactivator) could not (Figure 7d). Puigserver et al. believe this indicates that PGC-1 does not need to bind to nuclear receptors for initiate transcription and the nuclear receptors may just bring PGC-1 to the correct place on the DNA.

Figure 8 is a functional test of PGC-1. In order to show that PGC-1 is linked to adaptive thermogenesis, Puigserver et al. have inserted a vector coding for PGC-1 into white adipose tissue and examined it for adaptive thermogenesis behavior. The control was a cell transfected with a plasmid that did not contain PGC-1. This test resulted in an obvious increase in band intensity of RNA for the proteins PGC-1, UCP-1, and less obviously, an increase in the band intensities of the proteins COX-II, COX-IV, and ATP synthase. Analysis of mitochondrial DNA of cells with or without the PGC-1 plasmid was also supposed to show an increased amount of the DNA with control levels remaining constant. This last test of PGC-1 is a very important test to see whether adaptive thermogenesis can be induced in a foreign system since this shows gain of function. However, this is also the weakest figure. The doubling in band intensities for mtDNA, COX II, and ATP synthase is really not seen well. In order to convince readers of some increase, it might have been useful to quantitate the amount hybridized with the probe as was done with the TRb in figure 2a. Another problem with this figure is that the control, aP2, does not seem constant in all lanes. Still, Puigserver et al. provide another control that does remain constant so the slight chane in aP2 can be ignored.

Conclusions and Remarks:
Puigserver et al. have convincingly found a novel protein that does interact with PPARd, TRb, RARa, and ERa. The protein is localized in the correct area of the cell, the nucleus, for these interactions to occur. PGC-1 behaves in a manner that is like adaptive thermogenesis since it is upregulated upon cold or beta- adrenergic agonists and does activate transcription of PPARd and TR which are linked to adaptive thermogenesis. PGC-1 also induces some characteristics of adaptive thermogenesis in ectopic expression experiments. Before PGC-1 can be definitively connected with adaptive thermogenesis, a better gain of function test is needed than the one provided in figure 8.

A major concern that I have is the classification of PGC-1 being a coactivator of nuclear receptors. As explained in the introduction, there are certain traits common to coactivators of nuclear receptors. PGC-1 does not fit them. PGC-1 does have an LXXLL motif, but that is not involved in binding. Instead the region, that seems to be involved in binding is rich in proline residues. PGC-1 has specific tissues it is located in and different binding affinities, perhaps even different binding regions, for nuclear receptors (figure 2). This is totally unlike the broad range of tissues and general binding of nuclear receptors in identified coactivators. PGC-1 demonstrates a temperature dependency that is not found in other coactivators. Numerous other papers have cited the presence of acetyltransferase or histone deacetylase activity in coactivators, but PGC-1 has no sequence that would indicate the presence of either activity. In CAT assays, PGC-1 did not show any acetyltransferase activity without the presence of other ligands. Also, and perhaps most importantly, PGC-1 has no significant sequence homologies with other nuclear coactivators or corepressors even within the same species. Puigserver et al. believe that all of these differences from existing nuclear coactivators can be ignored by saying that PGC-1 is a coactivator that "falls outside of the current paradigm of ligand- dependent AF-2 docking proteins". If it falls outside of the current requirements for coactivators of nuclear recpetors, on what basis was the classification made? PGC-1 has specific characteristics and might have other unknown functions. From the evidence given by Puigserver et al., I am convinced that PGC-1 is associated with nuclear receptors and mimics some processes of thermogenesis, but I am not convinced that it is a coactivator.

Ideas for Future Research
· Do knock out mice for adaptive thermogenesis regain this function when PGC-1 is added to the system?
Puigserver et al. have "linked" PGC-1 to adaptive thermogenesis, but have not completely shown that PGC-1 is a necessary portion of this process. First, an organism should lose the ability to deal with changes in temperature if the PGC-1 protein was removed from the chromosomes. This could be achieved through a knock out mouse. Then, to verify this result, the addition of PGC-1 back into the mouse should result in adaptive thermogenesis.

·Do the SR and RNA binding domains allow PGC-1 to bring RNA polymerase II to the gene that needs transcription?
Figure 7 tested the ability of PGC-1 to activate translation. Although PGC-1 was enhanced by the presence of other molecules, it was also shown that PGC-1 does not require a nuclear receptor to activate transcription (7d). One possible way to test this is to delete those two domains and see whether PGC-1 still initiates transcription. However, this deletion is problematic in that it would rearrange the structure of the protein. It might be better to do mutagensis near those sights and see what is needed for this activation of transcription.

· Does PGC-1 function by interfering with a repressor?
The portion that PGC-1 binds to in HR and PPARd is the DNA binding and hinge domain. Hinge domains in TR are known to bind corepressors (Horlein, 1995). Therefore, it could be that PGC-1 competes with the corepressors for binding sites to the nuclear receptors. In order to determine if this occurs, one might determine whether PGC-1 is upregulated when there is an abundance of corepressors such as N-CoR. Also, to determine that PGC-1 and the corepressor bind to the same section of TR, you could make an antibody to the hinge region of the TR and then construct an anti-idiotype (which would look like the hinge region of TR). If the anti- idiotype is used as a probe, you will be able to see if PGC-1 and the corepressor bind to the same thing.

· Do the three phosphorylation consensus sites identified in figure 1 result in a posttranslational activation when a beta- adrenergic agonist is added to the system?
Puigserver et al. have identified that beta- adrenergic agonists induce PGC-1 expression as well as helping the system ectopically. However, this question must not always apply because PGC-1 is active without the agonists as well. Still, this could be tested by removing the beta- adrenergic agonists from the WAT and see whether PGC-1 could still activate adaptive thermogenesis. Another approach would be to see what removing the consensus sites would do to the protein.

Does PGC-1 act directly or indirectly on the mitochondrial proteins involved in adaptive thermogenesis?
Puigserver et al. determined that PGC-1 was localized in the nucleus of a cell (Figure 2c). However, in their discussion they reveal that some PGC-1 is transported to the mitochondria. Although they do not reveal how this was discovered, it raises interesting possibilities such as what the method of transport is, what percentage is transported, and whether the amount of PGC-1 in the mitochondria effects the upregulation of the proteins. This last possibility could be tested by artifically increasing the amount of PGC-1 in the mitochondria and observing the results by looking at the relative amount of mRNA to determine expression.

Garruti, G. and Ricquier D. (1992). Analysis of uncoupling protein and its mRNA in adipose tissue deposits of adult humans. International Journal of Obesity Related Metabolic Disorders 16, 383-390.

Heery, D. M., Kalkhoven, E., Hoare, S., and Parker, M.G. (1997). A signature motif in transcriptional co- activators mediates binding to nuclear receptors. Nature 387, 733- 736.

Himms- Hagen, J. (1989). Role of thermogenesis in the regulation of energy balance in relation to obesity. Canadian Journal of Physiological Pharmacology 67, 394- 401.

Horlein, A. J., Naar, A. M., Heinzel, T., Torchia, J., Gloss, B., Kurokawa, R., Ryan, A., Kamei, Y., Soderstrom, M., Glass, C. K., and Rosenfeld, M. G. (1995). Ligand -independent repression by the thyroid hormone receptor mediated by a nuclear receptor corepressor. Nature 377, 397- 404.

Jacobsson, A., Stadler, U., Glotzer, M. A., and Kozak, L. P. (1985). Mitochondrial uncoupling protein from mouse brown fat. Molecular cloning, genetic mapping, and mRNA expression. Journal of Biological Chemistry 260, 16520- 16524.

Nicholls, D., Cunningham, S., and Weisinger, H. (1986). Mechanisms of thermogenesis in brown adipose tissue. Biochem. Soc. Trans. 14, 223-225.

Pazin, M. J., and Kadagona, J. T. (1997) What is up and down with histone acetylation and transcription. Cell 89, 325- 328.

Sears, I. B., MacGinnitie, M. A., Kovacs, L. G., and Graves, R. A. (1996). Differentiation- dependent expression of the brown adipocyte uncoupling protein gene: regulation by peroxisome proliferator- activated receptor gamma. Molecular Cell Biology 16, 3410- 3419.

Silva, J. E. (1995). Thyroid hormone control of thermogenesis and energy balance. Thyroid 5, 481- 492. Tontonoz, P., Hu, E., and Speigelman, B. M. (1994). Stimulation of adipogenesis in fibroblasts by PPARd2, a lipid- activated transcription factor. Cell 43, 1271- 1278.

Yuryev, A., Patturajan, M., Litingtung, Y., Joshi, R. V., Gentile, C., Gebara, M., and Corden, J. L. (1996). The C- terminal domain of the large subunit of RNA polymerase II interacts with a novel set of serine/ arginine- rich proteins. Proc. Natl. Acad. Sci. USA 93, 6975-6980.