news & views PROTEIN AGGREGATION
Close encounters of the greasy kind Aggregation of a-synuclein (aSN) is critical to the development of Parkinson’s disease (PD), but the role of membranes in this process has been unclear and controversial. Galvagnion et al. demonstrate and model how lipids can stimulate aSN aggregation over a narrow range of lipid:protein stoichiometries.
Daniel Otzen
a 7
[α-synfree] fixed
6 [α-syn]fibrils (µm)
npg
© 2015 Nature America, Inc. All rights reserved.
T
here is widespread agreement that aggregation of the natively unfolded protein α-synuclein (αSN) to form fibrils critical to the development of PD. However, scientists have long struggled with an apparent paradox: αSN is often found aggregated together with phospholipid molecules. Yet on most membranes, monomeric αSN assumes an α-helical conformation rather than the β-sheets found in the final fibrils; indeed, lipids can inhibit αSN fibrillation in vitro. In this issue, Galvagnion et al.1 come close to resolving this quandary by showing that there is a ‘window of opportunity’ for lipids to bind αSN in the helical state and also stimulate the formation of fibrils. Although the normal function of αSN in vivo still remains to be firmly defined2, it clearly has a membrane connection3: αSN is most likely linked to neurotransmitter release via small synaptic vesicles and also plays a role in the biogenesis of synaptic membranes in general. It’s also fairly clear how membrane attraction is mediated: αSN is natively unfolded and contains an amphipathic N terminus that drives it to bind to negatively charged phospholipid membranes. Membrane binding is also Increasing [α-synbound]
connected with αSN pathology. αSN has an unfortunate tendency to aggregate, forming oligomeric species as well as large mature fibrils via small fibrillation nuclei4. Oligomers are believed to be cytotoxic by virtue of their very ability to bind to and disrupt membranes, something which they do much better than monomeric αSN5. There is a very large body of literature on how αSN binds to membranes as well as (to a smaller extent) how membranes affect αSN aggregation, but few studies (for example, ref. 6) rigorously address how lipids can affect the mechanism of aggregation. Over the past few years, Knowles’ group has developed robust theoretical approaches to fit different kinds of aggregation processes in a global fashion, i.e., simultaneous analysis of many different sets of aggregation time profiles7, and in the present study these analytical tools have been combined with the world-class amyloid expertise in the Dobson and Vendruscolo groups to great effect. As a model membrane, the authors used the negatively charged phospholipid DMPS. In the absence of lipid, αSN did not aggregate over the experimental time scale (2–3 days). The eye-opener (previously
b
seen only when using anionic detergents like SDS) is that over a fairly narrow range of DMPS:protein molar ratios, the lipid strongly stimulated αSN fibrillation. The effect peaked around 8 DMPS per αSN. DMPS has a large binding capacity for αSN, since binding of monomeric αSN continues to increase up to DMPS:αSN ratios of around 28, but clearly excess DMPS inhibits aggregation. The authors provide a straightforward explanation for this: although DMPS concentrates αSN locally on the vesicle surface (by a factor of almost 103 compared to the concentration of αSN in solution), higher DMPS concentrations actually dilute out the bound αSN as well as depleting it from solution, effectively immobilizing αSN. If, on the other hand, the concentration of DMPS is kept constant, the amount of fibrils formed scales in a simple proportional manner with the amount of bound αSN. So how do the growing fibrils look? They have two remarkable features: the fibrils are very thin and short, and each vesicle gives rise to only one fibril. Lipidfree fibrils are much thicker and longer, possibly due to lateral binding or secondary nucleation (growth on the sides of the
Homogeneous primary nucleation (slow)
Free monomers
5 4
Heterogeneous primary nucleation
3 2 1 0
0
5
10 15 20 25 30 Time (h)
Lipid vesicle surface
Nucleus conversion and elongation
Fragmentation
Figure 1 | Modelling lipid-assisted αSN fibrillation. (a) Kinetic data for the early stages of formation of aSN amyloid at increasing concentrations of lipid-bound of aSN in the presence of the anionic lipid DMPS. The lines show best fits to a model involving nucleation of aSN on the lipid surface. The lipid concentration is not included directly in the model, but only indirectly through its effect on the concentration of active aSN. (b) Model of lipid-assisted aSN aggregation used to fit the data in a. The lipid surface attracts aSN, increasing the local concentration. Only one nucleus is formed on each vesicle, leading aSN to growth of fibrils from the vesicle surface. Growth does not involve fragmentation of growing fibril ends (to yield new growing ends) or secondary nucleation, that is, growth of fibrils from other parts of the surface of the initial fibril. 176
nature chemical biology | VOL 11 | MARCH 2015 | www.nature.com/naturechemicalbiology
npg
© 2015 Nature America, Inc. All rights reserved.
news & views fibrils, not just the ends). Somehow the lipids control fibril growth, blocking sideways extension and preventing limitless extension of the growing ends—this remains a rather mysterious phenomenon. Thus although the vesicles strongly attract αSN monomers, at most one fibril is allowed to form per vesicle, for reasons as yet not understood and consequently not modeled by Galvagnion et al. Modeled this way, the rate of fibrillation does not depend on free αSN concentration, unlike for fibrillation in solution; the trick is the high degree of binding to vesicles—in fact, vesicle binding increases nucleation around 5 million times! The authors have also been involved in another study indicating that different αSN oligomers interconvert enroute to the fibrils8; this conversion can also be included in the model, but the fit is very decent even without this step (Fig. 1). Where does this leave us? It is a significant advance that we can incorporate the role of lipids into fibrillation mechanisms at the mechanistic level. As is often the case, this is just the first step and much remains to be refined. Ultimately models like these are interesting only if they are physiologically relevant. That may be
questioned at this stage. Synaptic vesicles are not 100% anionic like the DMPS vesicles, but rather are complex lipid mixtures containing around 20% anionic lipids while the rest are zwitterionic or neutral9. Will complex mixtures have a similar effect? Moreover, the estimated synaptic lipid:αSN ratio in vivo is estimated to 400, well above the optimal ratio seen here. But perhaps higher lipid ratios could compensate for lower charge? Moreover, these experiments are conducted under conditions under which DMPS is in a state of flux between the ordered and disordered states: DMPS normally melts around 41 °C, but binding of αSN reduces this to ~31 °C, so that the lipid is actually in a rather ill-defined physical state; this is not necessarily unphysiological but adds to the complexity of the interpretation. A further mismatch is that all these exciting experiments were conducted at 0 mM NaCl; even 25–50 mM NaCl (well below physiological concentrations) strongly inhibit the effect of lipids. Reducing the salt concentration can also lead to αSN fibrils with very different structural and functional properties10, but although this illustrates the diversity of αSN conformations, the link to in vivo conditions seems rather doubtful.
Nevertheless, the article is a trailblazer and should inspire many interesting follow-up studies that in the end may lead to realistic models of what actually happens when αSN fibrillates in the cell. ■ Daniel Otzen is in the Interdisciplinary Nanoscience Center (iNANO), Department of Molecular Biology and Genetics, Center for Insoluble Protein Structures (inSPIN), Aarhus University, Aarhus, Denmark. e-mail: [email protected] References
1. Galvagnion, C. et al. Nat. Chem. Biol. 11, 229–234 (2015). 2. Spillantini, M.G. & Goedert, M. Ann. NY Acad. Sci. 920, 16–27 (2000). 3. Auluck, P.K., Caraveo, G. & Lindquist, S.L. Annu. Rev. Cell Dev. Biol. 26, 211–233 (2010). 4. Lorenzen, N. et al. J. Am. Chem. Soc. 136, 3859–3868 (2014). 5. Volles, M.J. & Lansbury, P.T. Biochemistry 42, 7871–7878 (2003). 6. Last, N.B., Rhoades, E. & Miranker, A.D. Proc. Natl. Acad. Sci. USA 108, 9460–9465 (2011). 7. Cohen, S.I., Vendruscolo, M., Dobson, C.M. & Knowles, T.J. in Amyloid Fibrils and Prefibrillar Aggregates: Molecular and Biological Properties (ed. Otzen, D.E.) 7–9 (Wiley-VCH, 2013). 8. Cremades, N. et al. Cell 149, 1048–1059 (2012). 9. Benfenati, F., Greengard, P., Brunner, J. & Bähler, M. J. Cell Biol. 108, 1851–1862 (1989). 10. Bousset, L. et al. Nat. Commun. 4, 2575 (2013).
Competing financial interests The author declares no competing financial interests.
NATURAL PRODUCTS
Untwisting the antibiotic’ome
Microbial natural products and the specific subset with antibiotic activity, ‘the antibiotic’ome’, consist of a dizzying array of structures and exert their effects by many known modes of action. In this issue, Cociancich et al. describe a unique natural product that—along with a compound identified in a recent publication by Baumann et al.—defines a new antibacterial chemical scaffold that acts on a rarely hit target, DNA gyrase subunit A.
Chad W Johnston & Nathan A Magarvey
M
icrobial secondary metabolism has been a wellspring of chemical diversity and antibiotic lead compounds since the 1940s, providing thousands of bioactive molecules and the majority of clinically relevant antibacterial chemical scaffolds1. Two of the major biosynthetic classes present in this ‘antibiotic’ome’ are polyketides and nonribosomal peptides—complex small molecules produced by modular enzymatic assembly lines. Although these privileged compounds have traditionally been isolated from prolific producer families such as Bacilli, Actinobacteria and Fungi, widespread microbial genome sequencing is revealing similar biosynthetic potential from previously untapped sources. Given the current need for new antibacterial leads,
gaining access to new sources of microbial chemistry should be considered a high priority. In this issue of Nature Chemical Biology, Cociancich et al.2 describe the structure and biosynthesis of albicidin, a powerful antibiotic from the sugarcane microbe Xanthomonas albilineans. By heterologously expressing the biosynthetic genes required for the compound’s production, these researchers were able to unlock exotic chemistry from this similarly exotic organism, providing a potential lead for antibacterial drug development. Although Cociancich et al.2 are the first to elucidate the structure of albicidin, it has been known for decades as the phytotoxin responsible for leaf scald disease in sugarcane3. Genetic data obtained in 2001 first demonstrated that albicidin was
nature chemical biology | VOL 11 | MARCH 2015 | www.nature.com/naturechemicalbiology
a polyketide synthase–nonribosomal peptide synthetase (PKS-NRPS) product4, with the biosynthetic gene cluster including genes encoding both a trans-acting acyltransferase and a much rarer transacting NRPS5. Much like Pseudomonas syringae—another biosynthetically talented plant pathogen—X. albilineans appears to utilize carefully developed chemistry to gain a foothold in its choice environmental niche. Because it inhibits the function of DNA gyrase through interactions with the A subunit, albicidin’s actions are distinct from other natural-product DNA gyrase inhibitors that hit the B subunit. Before now, researchers had been unable to isolate sufficient quantities of albicidin for structure elucidation, likely because albicidin production is carefully controlled in the host 177
Close encounters of the greasy kind Aggregation of a-synuclein (aSN) is critical to the development of Parkinson’s disease (PD), but the role of membranes in this process has been unclear and controversial. Galvagnion et al. demonstrate and model how lipids can stimulate aSN aggregation over a narrow range of lipid:protein stoichiometries.
Daniel Otzen
a 7
[α-synfree] fixed
6 [α-syn]fibrils (µm)
npg
© 2015 Nature America, Inc. All rights reserved.
T
here is widespread agreement that aggregation of the natively unfolded protein α-synuclein (αSN) to form fibrils critical to the development of PD. However, scientists have long struggled with an apparent paradox: αSN is often found aggregated together with phospholipid molecules. Yet on most membranes, monomeric αSN assumes an α-helical conformation rather than the β-sheets found in the final fibrils; indeed, lipids can inhibit αSN fibrillation in vitro. In this issue, Galvagnion et al.1 come close to resolving this quandary by showing that there is a ‘window of opportunity’ for lipids to bind αSN in the helical state and also stimulate the formation of fibrils. Although the normal function of αSN in vivo still remains to be firmly defined2, it clearly has a membrane connection3: αSN is most likely linked to neurotransmitter release via small synaptic vesicles and also plays a role in the biogenesis of synaptic membranes in general. It’s also fairly clear how membrane attraction is mediated: αSN is natively unfolded and contains an amphipathic N terminus that drives it to bind to negatively charged phospholipid membranes. Membrane binding is also Increasing [α-synbound]
connected with αSN pathology. αSN has an unfortunate tendency to aggregate, forming oligomeric species as well as large mature fibrils via small fibrillation nuclei4. Oligomers are believed to be cytotoxic by virtue of their very ability to bind to and disrupt membranes, something which they do much better than monomeric αSN5. There is a very large body of literature on how αSN binds to membranes as well as (to a smaller extent) how membranes affect αSN aggregation, but few studies (for example, ref. 6) rigorously address how lipids can affect the mechanism of aggregation. Over the past few years, Knowles’ group has developed robust theoretical approaches to fit different kinds of aggregation processes in a global fashion, i.e., simultaneous analysis of many different sets of aggregation time profiles7, and in the present study these analytical tools have been combined with the world-class amyloid expertise in the Dobson and Vendruscolo groups to great effect. As a model membrane, the authors used the negatively charged phospholipid DMPS. In the absence of lipid, αSN did not aggregate over the experimental time scale (2–3 days). The eye-opener (previously
b
seen only when using anionic detergents like SDS) is that over a fairly narrow range of DMPS:protein molar ratios, the lipid strongly stimulated αSN fibrillation. The effect peaked around 8 DMPS per αSN. DMPS has a large binding capacity for αSN, since binding of monomeric αSN continues to increase up to DMPS:αSN ratios of around 28, but clearly excess DMPS inhibits aggregation. The authors provide a straightforward explanation for this: although DMPS concentrates αSN locally on the vesicle surface (by a factor of almost 103 compared to the concentration of αSN in solution), higher DMPS concentrations actually dilute out the bound αSN as well as depleting it from solution, effectively immobilizing αSN. If, on the other hand, the concentration of DMPS is kept constant, the amount of fibrils formed scales in a simple proportional manner with the amount of bound αSN. So how do the growing fibrils look? They have two remarkable features: the fibrils are very thin and short, and each vesicle gives rise to only one fibril. Lipidfree fibrils are much thicker and longer, possibly due to lateral binding or secondary nucleation (growth on the sides of the
Homogeneous primary nucleation (slow)
Free monomers
5 4
Heterogeneous primary nucleation
3 2 1 0
0
5
10 15 20 25 30 Time (h)
Lipid vesicle surface
Nucleus conversion and elongation
Fragmentation
Figure 1 | Modelling lipid-assisted αSN fibrillation. (a) Kinetic data for the early stages of formation of aSN amyloid at increasing concentrations of lipid-bound of aSN in the presence of the anionic lipid DMPS. The lines show best fits to a model involving nucleation of aSN on the lipid surface. The lipid concentration is not included directly in the model, but only indirectly through its effect on the concentration of active aSN. (b) Model of lipid-assisted aSN aggregation used to fit the data in a. The lipid surface attracts aSN, increasing the local concentration. Only one nucleus is formed on each vesicle, leading aSN to growth of fibrils from the vesicle surface. Growth does not involve fragmentation of growing fibril ends (to yield new growing ends) or secondary nucleation, that is, growth of fibrils from other parts of the surface of the initial fibril. 176
nature chemical biology | VOL 11 | MARCH 2015 | www.nature.com/naturechemicalbiology
npg
© 2015 Nature America, Inc. All rights reserved.
news & views fibrils, not just the ends). Somehow the lipids control fibril growth, blocking sideways extension and preventing limitless extension of the growing ends—this remains a rather mysterious phenomenon. Thus although the vesicles strongly attract αSN monomers, at most one fibril is allowed to form per vesicle, for reasons as yet not understood and consequently not modeled by Galvagnion et al. Modeled this way, the rate of fibrillation does not depend on free αSN concentration, unlike for fibrillation in solution; the trick is the high degree of binding to vesicles—in fact, vesicle binding increases nucleation around 5 million times! The authors have also been involved in another study indicating that different αSN oligomers interconvert enroute to the fibrils8; this conversion can also be included in the model, but the fit is very decent even without this step (Fig. 1). Where does this leave us? It is a significant advance that we can incorporate the role of lipids into fibrillation mechanisms at the mechanistic level. As is often the case, this is just the first step and much remains to be refined. Ultimately models like these are interesting only if they are physiologically relevant. That may be
questioned at this stage. Synaptic vesicles are not 100% anionic like the DMPS vesicles, but rather are complex lipid mixtures containing around 20% anionic lipids while the rest are zwitterionic or neutral9. Will complex mixtures have a similar effect? Moreover, the estimated synaptic lipid:αSN ratio in vivo is estimated to 400, well above the optimal ratio seen here. But perhaps higher lipid ratios could compensate for lower charge? Moreover, these experiments are conducted under conditions under which DMPS is in a state of flux between the ordered and disordered states: DMPS normally melts around 41 °C, but binding of αSN reduces this to ~31 °C, so that the lipid is actually in a rather ill-defined physical state; this is not necessarily unphysiological but adds to the complexity of the interpretation. A further mismatch is that all these exciting experiments were conducted at 0 mM NaCl; even 25–50 mM NaCl (well below physiological concentrations) strongly inhibit the effect of lipids. Reducing the salt concentration can also lead to αSN fibrils with very different structural and functional properties10, but although this illustrates the diversity of αSN conformations, the link to in vivo conditions seems rather doubtful.
Nevertheless, the article is a trailblazer and should inspire many interesting follow-up studies that in the end may lead to realistic models of what actually happens when αSN fibrillates in the cell. ■ Daniel Otzen is in the Interdisciplinary Nanoscience Center (iNANO), Department of Molecular Biology and Genetics, Center for Insoluble Protein Structures (inSPIN), Aarhus University, Aarhus, Denmark. e-mail: [email protected] References
1. Galvagnion, C. et al. Nat. Chem. Biol. 11, 229–234 (2015). 2. Spillantini, M.G. & Goedert, M. Ann. NY Acad. Sci. 920, 16–27 (2000). 3. Auluck, P.K., Caraveo, G. & Lindquist, S.L. Annu. Rev. Cell Dev. Biol. 26, 211–233 (2010). 4. Lorenzen, N. et al. J. Am. Chem. Soc. 136, 3859–3868 (2014). 5. Volles, M.J. & Lansbury, P.T. Biochemistry 42, 7871–7878 (2003). 6. Last, N.B., Rhoades, E. & Miranker, A.D. Proc. Natl. Acad. Sci. USA 108, 9460–9465 (2011). 7. Cohen, S.I., Vendruscolo, M., Dobson, C.M. & Knowles, T.J. in Amyloid Fibrils and Prefibrillar Aggregates: Molecular and Biological Properties (ed. Otzen, D.E.) 7–9 (Wiley-VCH, 2013). 8. Cremades, N. et al. Cell 149, 1048–1059 (2012). 9. Benfenati, F., Greengard, P., Brunner, J. & Bähler, M. J. Cell Biol. 108, 1851–1862 (1989). 10. Bousset, L. et al. Nat. Commun. 4, 2575 (2013).
Competing financial interests The author declares no competing financial interests.
NATURAL PRODUCTS
Untwisting the antibiotic’ome
Microbial natural products and the specific subset with antibiotic activity, ‘the antibiotic’ome’, consist of a dizzying array of structures and exert their effects by many known modes of action. In this issue, Cociancich et al. describe a unique natural product that—along with a compound identified in a recent publication by Baumann et al.—defines a new antibacterial chemical scaffold that acts on a rarely hit target, DNA gyrase subunit A.
Chad W Johnston & Nathan A Magarvey
M
icrobial secondary metabolism has been a wellspring of chemical diversity and antibiotic lead compounds since the 1940s, providing thousands of bioactive molecules and the majority of clinically relevant antibacterial chemical scaffolds1. Two of the major biosynthetic classes present in this ‘antibiotic’ome’ are polyketides and nonribosomal peptides—complex small molecules produced by modular enzymatic assembly lines. Although these privileged compounds have traditionally been isolated from prolific producer families such as Bacilli, Actinobacteria and Fungi, widespread microbial genome sequencing is revealing similar biosynthetic potential from previously untapped sources. Given the current need for new antibacterial leads,
gaining access to new sources of microbial chemistry should be considered a high priority. In this issue of Nature Chemical Biology, Cociancich et al.2 describe the structure and biosynthesis of albicidin, a powerful antibiotic from the sugarcane microbe Xanthomonas albilineans. By heterologously expressing the biosynthetic genes required for the compound’s production, these researchers were able to unlock exotic chemistry from this similarly exotic organism, providing a potential lead for antibacterial drug development. Although Cociancich et al.2 are the first to elucidate the structure of albicidin, it has been known for decades as the phytotoxin responsible for leaf scald disease in sugarcane3. Genetic data obtained in 2001 first demonstrated that albicidin was
nature chemical biology | VOL 11 | MARCH 2015 | www.nature.com/naturechemicalbiology
a polyketide synthase–nonribosomal peptide synthetase (PKS-NRPS) product4, with the biosynthetic gene cluster including genes encoding both a trans-acting acyltransferase and a much rarer transacting NRPS5. Much like Pseudomonas syringae—another biosynthetically talented plant pathogen—X. albilineans appears to utilize carefully developed chemistry to gain a foothold in its choice environmental niche. Because it inhibits the function of DNA gyrase through interactions with the A subunit, albicidin’s actions are distinct from other natural-product DNA gyrase inhibitors that hit the B subunit. Before now, researchers had been unable to isolate sufficient quantities of albicidin for structure elucidation, likely because albicidin production is carefully controlled in the host 177
