of the isoprenylated isoform, by contrast, had no effect. These observations make a per- suasive case for palmitoylated Cdc42 as the dominant isoform regulating dendritic spine morphology, with palmitoylation being crucial for this function. Kang et al. propose that the functional differences between palmitoylated and iso- prenylated Cdc42 could be ascribed to their different membrane localization, governed by their specific lipid modifications. For example, dendritic spines are enriched in cholesterol– sphingolipid membrane microdomains called lipid rafts 6 , and palmitoylated proteins are known to preferentially associate with rafts 7 . Perturbing rafts in neurons leads to a loss of dendritic spines 6 , an effect similar to the con- sequences of depleting palmitoylated Cdc42 in neurons 1 . It would be interesting to determine whether palmitoylated Cdc42 localizes within membrane microdomains. If palmitoylation is a bona fide regulatory modification, then a protein’s palmitoyla- tion, and thus its function, should change in response to molecular signals. Kang et al. show that palmitoylation is a dynamic process — they document rapid changes in the palmi- toylation state of selected synaptic proteins in neurons grown in culture in response to treat- ment with the neurotransmitter glutamate. Not all of the ten proteins examined were affected, however, with those that were affected showing either increased or decreased palmi- toylation. So glutamate does not seem to have a general effect on protein palmitoylation in neurons. As for Cdc42, on treatment with glutamate its palmitoylation decreased, and this GTPase was lost from dendritic spines. A previous study 8 showed that dendritic spines contract and collapse in cultured neurons under the same experimental conditions as those used by Kang and colleagues. This correlative evidence therefore supports the proposal that regulated palmitoylation of Cdc42 contributes to changes in spine morphology in response to synaptic activity. Establishing the broader significance of palmitoylation in regulating communi- cation at synaptic junctions will require an understanding of the functional consequences of dynamic palmitoylation of other synaptic targets identified by the authors. Characterization of the neural palmitoyl- proteome is a technical tour de force and provides a wealth of data to mine. Many investigators will find their favourite pro- teins on Kang and colleagues’ list and will be challenged to consider how palmitoylation might alter those proteins’ function. More- over, changes in the palmitoylation state of proteins in response to synaptic activity imply that enzymes that add palmitate to, or remove it from, proteins may also be regu- lated by synaptic activity. The first order of business, however, will be to identify enzymes responsible for reversible palmitoylation of Cdc42 and other synaptic proteins. Palmitoylation is catalysed by a large fam- ily of palmitoyltransferases — enzymes that share an evolutionarily conserved domain with a signature motif of aspartate–histidine– histidine–cysteine amino-acid residues 7,9 . Once animals lacking a specific palmitoyltrans- ferase are engineered, Kang and colleagues’ approach can be used both to identify the full spectrum of palmitoylated proteins that are substrates for that enzyme and to examine the functional consequences of ablating specific palmitoylation events in vivo. Less is known about enzymes that remove the palmitoyl group from proteins 7 , and their discovery will also be essential for elucidating the pathways that make palmitoylation dynamic. ■ Maurine E. Linder is in the Department of Cell Biology and Physiology, Washington University School of Medicine, St Louis, Missouri 63110, USA. e-mail: mlinder@wustl.edu 1. Kang, R. et al. Nature 456, 904–909 (2008). 2. Roth, A. F. et al. Cell 125, 1003–1013 (2006). 3. Drisdel, R. C. & Green, W. N. Biotechniques 36, 276–285 (2004). 4. Ethell, I. M. & Pasquale, E. B. Prog. Neurobiol. 75, 161–205 (2005). 5. Marks, P. W. & Kwiatkowski, D. J. Genomics 38, 13–18 (1996). 6. Hering, H., Lin, C. C. & Sheng, M. J. Neurosci. 23, 3262–3271 (2003). 7. Smotrys, J. E. & Linder, M. E. Annu. Rev. Biochem. 73, 559–587 (2004). 8. Halpain, S., Hipolito, A. & Saffer, L. J. Neurosci. 18, 9835–9844 (1998). 9. Tsutsumi, R., Fukata, Y. & Fukata, M. Pflügers Arch. 456, 1199–1206 (2008). BIOGEOCHEMISTRY Nitrous oxide in flux Sharon A. Billings In drought conditions, forest soils can serve as a small but surprisingly persistent sink for the greenhouse gas nitrous oxide. The effect highlights a research avenue necessary for predicting Earth’s climate. Increasing amounts of reactive nitrogen 1 are entering the environment through human agency. One consequence is increased produc- tion of the powerful greenhouse gas nitrous oxide — N 2 O — by microorganisms in soils. We do not understand the intricate dynamics of N 2 O production and consumption in soils, prompting research such as that reported in Global Change Biology 2 by Goldberg and Gebauer. They tracked N 2 O fluxes in Euro- pean spruce-forest soils under experimental conditions of a predicted climate pattern — increasing episodes of drought followed by heavy rainfall. The burning of fossil fuel, planting of crops associated with bacteria that can cap- ture atmospheric dinitrogen (N 2 ), and use of increasing amounts of fertilizer, all result in more nitrogen in Earth’s biological cycles 1 . Atmospheric concentrations of N 2 O have risen by 18% since the middle of the eighteenth century, in part because of these activities 3 . In the atmosphere, N 2 O lasts for an average of 114 years before undergoing reactions result- ing in its destruction. However, this atmos- pheric N 2 O sink accounts for only about 71% of known sources, leaving 5.2 million tonnes ‘missing’ from the atmosphere annually 3 . This discrepancy means that we are either over- estimating N 2 O sources or underestimating N 2 O sinks. Goldberg and Gebauer offer evi- dence that addresses this point. Reports of soil N 2 O fluxes typically describe net emissions to the atmosphere 4 , but Goldberg and Gebauer conclude that forest soils may serve as net N 2 O sinks to a greater extent than previously thought. Nitrous oxide is both produced and consumed in soils by a complex suite of microbially mediated processes. For example, denitrifiers perform the step-by-step chemical reduction of nitrate (NO 3 – ) to N 2 , producing N 2 O as an intermediate by-product. Some N 2 O escapes from the soil, but under anaerobic con- ditions some is completely transformed into N 2 . The denitrification pathway is thus associated with both N 2 O production and consumption. Moreover, other groups of microorganisms 4–6 can also transform nitrogen in ways that produce and consume N 2 O. Moisture availability can promote the pro- duction of N 2 O (ref. 7), but investigators often presume that N 2 O consumption — transforma- tion into N 2 — is also highest when soil moist- ure levels are high because of the anaerobic nature of this process 4 . There are relatively few field studies that report net N 2 O consumption at the soil surface, but those that do describe comparatively small fluxes that show no pre- dictable relationship with soil moisture. These small fluxes are often dismissed as an indica- tion of the challenges of quantifying N 2 O levels near detection levels, and not robust evidence of soil N 2 O sink strength 4 . Thus, the influence of soil moisture on N 2 O production and its ultimate fate — release or consumption — remains unclear. Goldberg and Gebauer 2 have tackled this issue by inducing soil drought in a spruce for- est in Germany, and measuring net N 2 O fluxes at the soil surface and concentrations within the soil profile. They found that net N 2 O con- sumption at the soil surface was enhanced by drought, and, surprisingly, that it occurred 888 NATURE|Vol 456|18/25 December 2008 NEWS & VIEWS