RESEARCH NEWS & VIEWS
50 Years Ago Despite some impressive demonstrations of root-pressure, ‘pumping’ or positive pressure is generally thought to be of little importance for the ascent of sap in trees … In ten species of palms at Calcutta I was able to detect and measure positive root-pressures throughout the year, sometimes enough to lift sap to heights exceeding that of the trees … The positive pressure exceeds the negative pressure considerably. It often exceeds the plant’s needs … The distal part of a cut root of a fruited banana plant was seen to develop pressure. The plant was decapitated at 2.25 m. above ground-level, and the cut root reconnected with a rubber tube. Other agencies such as transpiration are probably less important than root pressure. From Nature 21 October 1961
100 Years Ago My attention has been directed to three elm trees at Ettington … which it is said have been “killed by wasps.” It appears that the wasps were attracted by the sweetness of the sap, and attacked the trees in such swarms, and so drained them of sap, that the death of the trees seems imminent, all the leaves having gone yellow long before the usual time. I should be glad to know if others have noticed attacks on elm trees, and whether the averred sweetness of the sap is due to some previous degenerative change in the tissues of the tree, or whether wasps would attack a normal tree if they could get access to the sap. The elms are all three comparatively young trees, and belong to the common variety. My informant tells me that he has previously noticed the same thing happen with an elm tree in one of his fields, which died the next winter. From Nature 19 October 1911
species does not necessarily apply to all mammals, and especially not to humans. Despite great similarities in brain organization, there are also significant differences that could have profound functional consequences10. Obviously, one such difference is a decrease in adult neurogenesis and neuronal turnover that seems to have accelerated with primate evolution and that may be beneficial for human mental capabilities11. Examination of apes such as chimpanzees should answer the question of whether reduced neurogenesis in the adult SVZ is a uniquely human trait or a more general trend in hominoids. The present work also underscores the importance of considering species-specific differences in cellular mechanisms, and in their timing and function, in seemingly similar structures. The general decrease in adult neurogenesis with vertebrate evolution is associated with a diminished capacity for neuronal replacement and regeneration11. Reprogramming of neural progenitors in the SVZ and redirecting of neuronal migration to achieve regeneration in injured brain areas are already proving to be formidable challenges in rodents. The
present results3, demonstrating the lack of significant neurogenesis and migration in the human SVZ beyond infancy, suggest that such strategies might be even more difficult to apply in our species. ■ Jon I. Arellano and Pasko Rakic are in the Department of Neurobiology and the Kavli Institute, Yale University School of Medicine, New Haven, Connecticut 06510, USA. e-mail:
[email protected] 1. Ihrie, R. & Álvarez-Buylla, A. Neuron 70, 674–686 (2011). 2. Kornack, R. D. & Rakic, P. Proc. Natl Acad. Sci. USA 98, 4752–4757 (2001). 3. Sanai, N. et al. Nature 478, 382–386 (2011). 4. Weickert, C. S. et al. J. Comp. Neurol. 423, 359–372 (2000). 5. Sanai, N. et al. Nature 427, 740–744 (2004). 6. Curtis, M. A. et al. Science 315, 1243–1249 (2007). 7. Sanai, N., Berger, M. S., Garcia-Verdugo, J. M. & Alvarez-Buylla, A. Science 318, 393 (2007). 8. Lois, C., García-Verdugo, J.-M. & Alvarez-Buylla, A. Science 271, 978–981 (1996). 9. Sidman, R. L. & Rakic, P. Brain Res. 62, 1–35 (1973). 10. Rakic, P. Nature Rev. Neurosci. 10, 724–735 (2009). 11. Rakic, P. Science 227, 1054–1056 (1985).
C H EM I CA L B I OLOGY
Many faces of a cancersupporting protein The protein Hsp90 is a target of promising anticancer drugs. An analysis of the components of Hsp90 complexes in tumours reveals a path that may lead to predictive assays of drug sensitivity in cancer patients. J O H N F. D A R B Y & PA U L W O R K M A N
T
he protein Hsp90 is a molecular chaperone — it assists in the correct folding of other cellular proteins. Many of these Hsp90 ‘client’ proteins are over expressed and/or mutated in cancer and are involved in maintaining the cancerous state1. Hsp90 inhibition is therefore an attractive strategy for simultaneously blocking multiple abnormal pathways that are crucial for many tumour types2. Indeed, early clinical trials have confirmed the therapeutic potential of this approach. But understanding the mechanism that ensures selective targeting of tumour cells by Hsp90 inhibitors, and finding biomarkers to predict which cancers will be most sensitive to such treatment, has proved challenging. Writing in Nature Chemical Biology, Moulick et al.3 use chemical proteomics — a combination of chemical precipitation and multi-protein profiling — to shed light on these questions. There are some 20 Hsp90 inhibitors now
3 3 4 | NAT U R E | VO L 4 7 8 | 2 0 O C T O B E R 2 0 1 1
© 2011 Macmillan Publishers Limited. All rights reserved
in clinical trials. Their effects have been most impressive in breast cancers that overexpress a highly sensitive Hsp90 client, the HER2 oncoprotein, but that are resistant to the HER2antibody drug trastuzumab4. They are also promising in non-small-cell lung tumours that express the mutated oncogenic protein EML4-ALK, a similarly sensitive Hsp90 client 5. However, in other cancer types Hsp90 inhibition is less effective, despite the fact that the oncogenic constituents of such cancers are among Hsp90’s clientele. The best path to obtaining approval for widespread clinical use of these inhibitors may therefore be to apply them in particular tumour subtypes that are driven by highly Hsp90-dependent oncoproteins6. To predict an individual patient’s responsiveness, it will be crucial to define all of Hsp90’s oncoprotein clientele and to understand the make-up and function of chaperone– client complexes, together with the molecular networks in which they are involved. Large-scale genetic and messenger RNA/ protein-profiling efforts have revealed that
NEWS & VIEWS RESEARCH many cellular processes require Hsp90 activity under both normal and stress conditions7. These screening studies also identified8 many of the molecules that interact with Hsp90 but have done little to explain the relevance of particular client proteins to traits seen in cancer or to help us understand drug sensitivity. Moreover, although new Hsp90 clients have been found9, such approaches could not easily distinguish inhibitor-sensitive chaperone– client complexes from the ‘noisy’ background of other interactions involving Hsp90. A significant study10 published in 2003 provided insight into why tumour cells are so much more sensitive to Hsp90 inhibition than healthy cells. It showed that, in cancer cells, Hsp90 exists in a highly active, multi-chaperone complex that has much greater affinity for inhibitors than does Hsp90 in normal cells. However, the role of drug-sensitive Hsp90 complexes and the identity of their constituent clients and co-chaperones are still unclear some eight years later, highlighting the need for improved identification techniques. To discover specific, cancer-associated Hsp90 clients and pathways involved in chronic myeloid leukaemia (CML), Moulick et al.3 have profiled complexes that selectively bind to the Hsp90 inhibitor PU-H71, a drug that is undergoing clinical trial. Specifically, they used chemical precipitation, which selectively retrieves only those Hsp90 complexes that are ‘available’ for inhibition. This is in contrast to immunoprecipitation of Hsp90, which recovers both drug-binding and non-binding complexes (Fig. 1). The authors find that PU-H71 binds to a limited fraction (10–20%) of cellular Hsp90, suggesting that two distinct Hsp90 subpopulations are present in CML, with only one being available for inhibition by the drug. Across the various leukaemic cell lines analysed, the abundance of PU-H71-bound Hsp90 complexes correlated with sensitivity to the inhibitor. Indeed, the results of another study11 have also shown a correlation between inhibitor binding and drug sensitivity. Moulick and colleagues next performed protein profiling using mass spectrometry — a powerful proteomic screening approach — to identify components of the inhibitor-bound Hsp90 complexes. They find that PU-H71 binds to Hsp90 that is in complex with wellknown co-chaperones (Hsp40, Hsp70, HOP and HIP), but not to Hsp90 complexes lacking these co-chaperones. With regard to Hsp90 client proteins, Bcr-Abl — a product of gene translocation and an Hsp90 client that is the driver oncoprotein in CML — is also preferentially present in PU-H71-bound Hsp90 complexes. By contrast, its non-oncogenic precursor, Abl, resides in complexes that did not bind this inhibitor. It seems, therefore, that Bcr-Abl is highly reliant on an active subpopulation of Hsp90 that is susceptible to inhibition in CML, consistent with the emerging evidence for the
Hsp90
Client protein
Co-chaperone 10–20% PU-H71
80–90%
Hsp90 complexes
Chemical precipitation
Anti-Hsp90 Immunoprecipitation
Proteomics
Co-chaperones and known and newly identified clients
Bioinformatics
Pathways and cancer traits
Figure 1 | Analysis of Hsp90-containing complexes using chemical proteomics. Moulick et al.3 find that the Hsp90 inhibitor PU-H71 selectively precipitates Hsp90 that is in complex with well-known co-chaperones and client proteins — a subpopulation that makes up roughly 10–20% of total cellular Hsp90. The abundance of this subpopulation correlates with inhibitor sensitivity. By contrast, immunoprecipitation using antiHsp90 antibodies recovers all Hsp90 complexes. Further proteomic and bioinformatic analyses reveal that the client proteins in these complexes include members of many signalling pathways that are deregulated in chronic myeloid leukaemia, as well as proteins, such as CARM1, that had not previously been associated with Hsp90.
clinical effectiveness of Hsp90 inhibitors in this cancer. The authors’ freely available catalogue of the proteins found in PU-H71-bound Hsp90 complexes (see Supplementary Information to the paper3) includes members of molecular signalling pathways that are associated with CML (for example, the PI3K–AKT–mTOR, NF-κB, RAF–MAPK, STAT and focaladhesion signalling pathways). Focusing on a constituent of one of these pathways, the authors show that phosphorylation of the protein STAT5, which is found in the phosphorylated form in CML, requires Hsp90. Furthermore, proteins previously not known to be linked to Hsp90 were also found in PU-H71-bound Hsp90 complexes. For instance, CARM1 (also called PRMT4) — an enzyme that regulates gene transcription and the levels of which are increased in some cancers — is entirely sequestered in these Hsp90 complexes, and, as Moulick et al. show, is required for the viability of CML cells. This finding complements previous proteomic studies9 that identified the CARM1-related enzyme PRMT5 as a novel Hsp90 client.
Chemical proteomics is a powerful way of gaining a functional picture of the bewilderingly complex interactions in which Hsp90 is involved. This approach could also predict which Hsp90 clients are likely to be drugsensitive. Moreover, expansion of the proteomic analysis to compare a wider range of tumour and normal cell types may reveal why some malignant cells are more sensitive to drug inhibition than others. But perhaps the most exciting potential use of this new approach is in personalized medicine, to predict whether a patient would respond to treatment with Hsp90 inhibitors. Thus, it is conceivable that determining the proportion of the Hsp90 population in a patient’s cancer cells that is available for binding to a tagged inhibitor could predict drug sensitivity in the clinic. But this is not the end of the story, and, as always, questions remain. For example, what is the composition and function of the Hsp90 subpopulation in cancer cells that does not bind to PU-H71? What prevents the inhibitor from binding to that subpopulation? How do Hsp90-regulating co-chaperones and posttranslational modifications of Hsp90 control drug binding to the multi-chaperone complex? A puzzling observation made by Moulick et al. is that some potent, but chemically distinct, Hsp90 inhibitors that target the same site on this chaperone have different preferences for Hsp90 subpopulations from PU-H71. How does this relate to the cellular effects and therapeutic efficacy of the various inhibitors? That Hsp90 has a broad range of clients is a potential advantage for therapy, but a challenge for predicting an individual patient’s response. Increased understanding of Hsp90–client complexes in various tumour and healthy cells should help to pave the way for personalized clinical application of Hsp90 inhibitors, and could address a fundamental question — what makes a protein an Hsp90 client? ■ John F. Darby and Paul Workman are in the Cancer Research UK Cancer Therapeutics Unit, Institute of Cancer Research, Haddow Laboratories, Sutton, Surrey SM2 5NG, UK. e-mail:
[email protected] 1. Whitesell, L., Mimnaugh, E. G., De Costa, B., Myers, C. E. & Neckers, L. M. Proc. Natl Acad. Sci. USA 91, 8324–8328 (1994). 2. Workman, P. Cancer Lett. 206, 149–157 (2004). 3. Moulick, K. et al. Nature Chem. Biol. http://dx.doi. org/10.1038/nchembio.670 (2011). 4. Modi, S. et al. Clin. Cancer Res. 17, 5132–5139 (2011). 5. Sequist, L. V. et al. J. Clin. Oncol. 28, 4953–4960 (2010). 6. Neckers, L. & Workman, P. Clin. Cancer Res. (in the press). 7. McClellan, A. J. et al. Cell 131, 121–135 (2007). 8. Zhao, R. et al. Cell 120, 715–727 (2005). 9. Maloney, A. et al. Cancer Res. 67, 3239–3253 (2007). 10. Kamal, A. et al. Nature 425, 407–410 (2003). 11. Tillotson, B. et al. J. Biol. Chem. 285, 39835–39843 (2010).
2 0 O C T O B E R 2 0 1 1 | VO L 4 7 8 | NAT U R E | 3 3 5
© 2011 Macmillan Publishers Limited. All rights reserved
