Small Ubiquitin-Like Modifier Conjugating Enzyme with Active Site Mutation Acts as Dominant Negative Inhibitor of SUMO Conjugation in Arabidopsis


  • Konstantin Tomanov,

    1. Department of Biochemistry and Cell Biology, Max F. Perutz Laboratories, Center for Molecular Biology, University of Vienna, A-1030 Vienna, Austria
    Search for more papers by this author
  • Christian Hardtke,

    1. Department of Plant Molecular Biology, University of Lausanne, CH-1015 Lausanne, Switzerland
    Search for more papers by this author
  • Ruchika Budhiraja,

    1. Department of Plant Developmental Biology, Max Planck Institute for Plant Breeding Research, D-50829 Cologne, Germany
    Search for more papers by this author
  • Rebecca Hermkes,

    1. Department of Plant Developmental Biology, Max Planck Institute for Plant Breeding Research, D-50829 Cologne, Germany
    Search for more papers by this author
  • George Coupland,

    1. Department of Plant Developmental Biology, Max Planck Institute for Plant Breeding Research, D-50829 Cologne, Germany
    Search for more papers by this author
  • Andreas Bachmair

    Corresponding author
    1. Department of Biochemistry and Cell Biology, Max F. Perutz Laboratories, Center for Molecular Biology, University of Vienna, A-1030 Vienna, Austria
    2. Department of Plant Developmental Biology, Max Planck Institute for Plant Breeding Research, D-50829 Cologne, Germany
      Tel: +43 1 4277 74811; Fax: +43 1 4277 9748; E-mail:
    Search for more papers by this author

Tel: +43 1 4277 74811; Fax: +43 1 4277 9748; E-mail:


Small ubiquitin-like modifier (SUMO) conjugation affects a broad range of processes in plants, including growth, flower initiation, pathogen defense, and responses to abiotic stress. Here, we investigate in vivo and in vitro a SUMO conjugating enzyme with a Cys to Ser change in the active site, and show that it has a dominant negative effect. In planta expression significantly perturbs normal development, leading to growth retardation, early flowering and gene expression changes. We suggest that the mutant protein can serve as a probe to investigate sumoylation, also in plants for which poor genetic infrastructure precludes analysis via loss-of-function mutants.


Protein modification by the small ubiquitin-like modifier (SUMO) is an essential process in which a hetero-dimeric SUMO-activating enzyme (SAE) forms a thioester with the carboxyl terminus of SUMO, and transfers the activated protein to SUMO-conjugating enzyme (SCE), again forming a thioester bond to an active site Cys residue. SCE can either alone, or with the help of SUMO ligases, catalyze formation of an isopeptide bond between the SUMO carboxyl terminus and protein lysine side chains (for recent reviews, see Lois 2010; Miura and Hasegawa 2010; Park et al. 2011a; Castro et al. 2012; Novatchkova et al. 2012). The essentiality of SUMO conjugation precludes the use of loss-of-function mutants in SCE as models for analysis (Saracco et al. 2007). In contrast, loss-of-function mutants in SUMO ligases SIZ1 (Miura et al. 2005; Catala et al. 2007) and HPY2/MMS21 (Huang et al. 2009; Ishida et al. 2009) do not compromise plant viability, and generate a number of interesting phenotypes that allow for dissection of the complex pattern of SUMO activity. Other viable, often characteristic mutants in the pathway are compromised in the function of SUMO-specific proteases, ESD1 (Murtas et al. 2003), ELS1 (Hermkes et al. 2011), and OTS1/2 (Conti et al. 2008). In addition to mutant analysis, substrates of sumoylation have also been identified (Budhiraja et al. 2009; Elrouby and Coupland 2010; Miller et al. 2010; López-Torrejón et al. 2012), and sumoylation of specific substrates was linked to cold stress response, abscisic acid (ABA) signaling and nitrogen metabolism (Miura et al. 2007; Miura et al. 2009; Park et al. 2011b).

In this study, we investigate plants that overexpress a mutant version of SCE with the active site Cys replaced by Ser. This protein cannot accept SUMO from SAE in an in vitro reaction, and acts in vivo in a dominant-negative fashion. We found that the effects of overexpression resemble the consequences of defects in SUMO ligase SIZ1.


Mutation in SUMO-conjugating enzyme SCE

Small ubiquitin-like modifier-conjugation in Arabidopsis requires SCE as an essential component. Mutations in the active site of enzymes are a classical way to investigate enzyme functions. Thus, the active site Cys residue of SCE was replaced by Ser (henceforth called SCE(C94S)). The ensuing open reading frame (ORF) was cloned into a vector for overexpression in Arabidopsis. The design of this experiment was inspired by similar experiments with ubiquitin-conjugating enzymes, because Cys to Ser active site mutants of ubiquitin-conjugating enzymes can act in a dominant-negative fashion (Sung et al. 1991; Banerjee et al. 1995; Townsley et al. 1997).

Growth habit of SCE(C94S)-expressing plants

Expression of SCE(C94S) in Arabidopsis could be maintained over several generations before the silencing of transgenes, and resulted in a significant change in growth habit: plants were stunted and had smaller leaves (Figure 1). In contrast, plants that overexpress the wild-type (WT) version of SCE from the same vector are indistinguishable from untransformed Col-0 plants (Figure 1G, H). SCE(C94S) expression also resulted in an early flowering phenotype, which was particularly pronounced under short-day conditions: the plants started to flower after producing approximately half the number of leaves (44 leaves; SD 6.5, n = 16) compared to plants transgenic for WT SCE (90 leaves; SD 4, n = 16) (see also Figure 1E, F). Because the growth habit of SCE(C94S)-expressing plants resembled that of siz1 mutants, we wondered whether a potential decrease in SUMO conjugation in these plants exerts its effects on growth through the same substrates. Park et al. (2011b) recently found that nitrate reductase is a critical substrate of SUMO conjugation, and that supplying plants with additional ammonium ions mitigates the effects of a mutation in the SUMO ligase SIZ1 on growth. We thus compared the growth of SCE(C94S)-overexpressors in the presence and absence of 5 mM (NH4)2SO4. As shown in Figure 2, plants with an additional ammonia supply had more shoots and siliques than those without, suggesting that nitrogen metabolism is perturbed in these plants, and that this perturbation most likely occurs because of suboptimal SUMO conjugation to nitrate reductase. We also quantified the effect (Figure S1) by determining the fresh weight of plants germinated and grown on soil for three weeks either with or without (NH4)2SO4 in the water. The soil contained nitrate fertilizer as a nitrogen source, which was apparently an optimal nitrogen supply for normal plants, because a plant line expressing transgenic WT SCE showed reduced growth in the presence of (NH4)2SO4 (average fresh weight was 88% of the value without ammonium sulfate). In contrast, a plant line that moderately overexpressed SCE(C94S) displayed increased growth (average fresh weight was 176% of the control without ammonium sulfate). As a control, we determined that siz1 mutant plants had an average growth increase to reach 198% fresh weight under the same conditions (for further details, see Figure S1).

Figure 1.

A mutant form of small ubiquitin-like modifier (SUMO)-conjugating enzyme changes growth characteristics and time of flowering.
Plants overexpressing dominant-negative SUMO-conjugating enzyme SCE(C94S) (B) are smaller than plants containing a transgene with the wild-type (WT) version of this gene (A). Under short days, a similar size difference is apparent (C vs. D). SCE(C94S) plants also flower earlier in short days (F vs. E). In contrast, high expression of WT SCE does not cause any difference in growth habit compared to WT Col-0 plants (G, left and middle pots, two different lines of WT SCE overexpressing plants; right pots, WT Col-0 plants, 3 weeks after sowing; H same as G, but plants 5 weeks old).

Figure 2.

Additional ammonium supply mitigates reduced growth of SCE(C94S)-expressing plants.
Seedlings expressing SCE(C94S) were grown on soil with standard nitrate-containing, but ammonium-free fertilizer. (A, C) Size of plants reflects expression level of the transgene. The pot of panel A contains an additional wild-type (WT) seedling for comparison, emphasizing differences in flowering time and growth. (B) The pot of panel A was further grown for one m under the same conditions. The WT plant was removed to prevent overgrowth. (D) The pot of panel C was grown for one m with additional 5 mM (NH4)2SO4 in the supplied water, leading to improved growth and fertility.

SUMO conjugation patterns in SCE(C94S)-expressing plants

As the experiment with ammonium sulfate was suggestive of reduced SUMO conjugation to some in vivo substrates, we investigated levels of free SUMO and of SUMO conjugates. Western blot analysis showed that free SUMO levels were reduced (Figure 3A). Interestingly, there was also a reduction in higher molecular weight (MW) SUMO conjugates that is particularly obvious in one of the transgenic lines (Figure 3A, lane 6; black bar to the right of the gel). SCE(C94S) plant extracts also contained two bands not prominently present in WT, demonstrating occasional exceptions to the overall reduction of SUMO conjugate abundance (Figure 3A, arrows). We also probed plant extracts with anti SCE antibody to detect SCE protein. We could detect elevated levels of free SCE in some SCE(C94S)-expressing plants (Figure 3B). In contrast, we did not detect an SCE(C94S)-SUMO ester conjugate (data not shown), suggesting that the SCE(C94S) protein pool is not significantly conjugated to SUMO in vivo, which is consistent with the in vitro results presented below. We also wanted to know whether transient expression of SCE(C94S) causes changes in the SUMO conjugate pattern. Tsuda et al. (2012) have recently shown that efficient Agroinfection experiments can be done using an Arabidopsis line expressing AvrPto. Using this plant line, we obtained leaf extracts after 2 d of Agrobacterium application. Figure 4 shows that transient expression of SCE(C94S) also causes a decrease in high MW SUMO conjugates.

Figure 3.

Small ubiquitin-like modifier (SUMO)1/2, SUMO1/2 conjugates and SCE levels in transgenic plants.
(A) Leaf extracts from plants containing a wild-type (WT) SCE transgene (SCE(WT), lane 2), and four different lines with an SCE(C94S) transgene (lanes 3–6), were subjected to Western blotting with anti-SUMO1/2 antibodies. Lines with the SCE(C94S) transgene contain less free SUMO1/2 (band at the position of the dot to the right of the picture), and sometimes less SUMO conjugates (size range indicated by the black line to the right). Note two prominent bands present in SCE(C94S) lanes 3, 4 and 6, but not in the SCE(WT) lane 2 (arrows to the right). Lane 1, molecular weight marker bands with sizes indicated to the left. (B) Western blot using anti-SCE antibody indicates that transgene expression levels vary considerably. Some SCE(C94S) transgenic plants accumulate the transgene-encoded protein. Lanes 1, 6, molecular weight markers (only 17 kDa band visible); lane 2, plant transformed with WT SCE transgene (no overexpression detectable); lanes 3–5, different plant lines containing the SCE(C94S) transgene; lane 7, SCE protein purified from Escherichia coli as a standard.

Figure 4.

Transient expression of SCE(C94S) influences protein sumoylation.
Leaves of a plant line expressing AvrPto were infiltrated with an Agrobacterium strain containing either a vector for wild-type (WT) SCE expression (lane 1), or for SCE(C94S) expression (lane 3). Extracts were taken for Western blotting with anti small ubiquitin-like modifier (SUMO)1 antibody. Lane 2 shows extract from uninfected leaf. Top panel, Western blot; bottom panel, loading control (RUBISCO large subunit band of a Coomassie-stained gel).

Mechanistic implications of SCE(C94S) expression

We next wanted to better understand the mechanism or interference of SCE(C94S) with SUMO conjugation. Previously, in an in vitro SUMO conjugation system consisting entirely of Arabidopsis proteins overexpressed in Escherichia coli, we established that SCE(C94S) cannot transfer SUMO to substrates (Budhiraja et al. 2009). Here, we used specific reaction conditions that allow accumulation of the SUMO-SCE thioester conjugate to find out whether SCE(C94S) can accept SUMO from SAE in vitro. Figure 5 indicates that this is not the case. Both antibodies detecting the SUMO tag and anti SCE antiserum show the presence of SCE-SUMO, but a relative absence of SCE(C94S)-SUMO under identical reaction conditions. As a control, we show that the SCE-SUMO band disappears upon treatment of the sample with reducing agent, supporting the identity of the SCE-SUMO band as a thio ester. We therefore conclude that the dominant-negative effect of SCE(C94S) on in vivo sumoylation is exerted by the free protein, and not by the SCE(C94S)-SUMO oxyester.

Figure 5.

In vitro transfer of small ubiquitin-like modifier (SUMO)1 from SUMO activating enzyme (SAE) to SUMO conjugating enzyme SCE(WT) vs. SCE(C94S).
SUMO1 equipped with Strep tag, SAE and SCE, or Strep-SUM1, SAE and SCE with a Cys-to-Ser change in the active site, were incubated in the presence of ATP in vitro. Whereas wild-type (WT) SCE can form a thioester with SUMO1 (dot, lanes 2 and 6), SCE(C94) can not (lanes 3, 7). Heating of reactants in the presence of a reducing agent hydrolyzes thioester bonds, showing the nature of the linkage between SCE and SUMO (lanes 4, 8). Both antibody directed against the Strep tag of SUMO1 (lanes 1–5), and anti SCE antibody (lanes 6–9) detect the SCE-SUMO thioester intermediate, but no comparable SCE(C94S)-SUMO oxy-ester intermediate. One asterisk indicates a contaminating band in the SUMO1 preparation (lane 1 shows Strep-SUMO1 preparation only). Two asterisks indicate SCE-thioester conjugate to a SUMO protein with (proteolytically) shortened amino-terminal extension (lane 6).

Impact of SCE(C94S) on transcription

To further characterize SCE(C94S)-expressing plants, we analyzed transcript levels (Figure 6). An SCE probe indicated that growth inhibition correlated with high SCE(C94S) mRNA expression, because the plant used for lane 2 was smaller than the age-matched plant of lane 4. SCE(C94S) plants have changes in their gene expression pattern. We tested genes previously studied for the influence of sumoylation on their expression level, and/or known to be affected by the siz1 mutation in connection with stress response (Lois et al. 2003). Figure 6 shows that RD29A (At5g52310) and COR47 (At1g20440), two cold-induced genes, show decreased levels at room temperature in overexpressing plants.

Figure 6.

Increased levels of SCE(94S) mRNA inversely correlate with expression of cold response genes.
RNA from wild-type (WT) plants (lane 1, WT), and from three plant lines with SCE(C94S) transgene (lanes 2–4) was used for gel blotting. The expression level of SCE(C94S) influences the abundance of two cold-regulated transcripts (RD29A, COR27). At4g18480 mRNA was used as a loading control.


In this study, we investigated Arabidopsis plant lines with changes in the sumoylation system. Transgenic lines overexpressing variant SUMO conjugation enzyme SCE, SCE(C94S), show reduced growth, early flowering and changes in the pattern of SUMO conjugates. It is of note that a previous publication reported a similar approach, but did not report any physiological changes (Lois et al. 2003). We ascribe this difference to the fact that Lois et al. overexpressed SCE(C94S) with an attached His tag. Experiments with an in vitro sumoylation system in our lab suggest, however, that extensions at either the amino-, or the carboxyl-terminus of SCE compromise functionality (R Budhiraja and A Bachmair unpubl. data, 2004; Budhiraja et al. 2009), and may therefore also compromise the ability of SCE(C94S) to act in a dominant-negative fashion.

The variant SCE enzyme in this study has the active site Cys residue replaced by Ser. A similar strategy has previously been used for a number of ubiquitin-conjugating enzymes (Sung et al. 1991; Banerjee et al. 1995; Townsley et al. 1997). A dominant-negative effect of such a protein can arise if the mutant conjugating enzyme uses partner proteins that would, in the unperturbed situation, interact with the free enzyme. Alternatively, a mutant conjugating enzyme linked via (oxy-)ester bond to the modifier could compete with the respective WT complex, but inability to transfer the modifier to substrates blocks further complex dynamics. The latter case was demonstrated for a number of ubiquitin-conjugating enzymes (Banerjee et al. 1995; Eddins et al. 2006; Bremm et al. 2010). Although the Cys-to-Ser mutant of SCE was used in mammalian tissue cultures (Mo et al. 2004), the issue of SUMO loading was not specifically addressed. The available data, however, suggested accumulation of the native protein, but not of the SUMO-linked version. We have investigated this question for the plant SUMO conjugation system through in vivo and in vitro experiments. We could not detect in vivo accumulation of SCE(C94S) covalently linked to SUMO. In support of this finding, in vitro incubation of WT SCE with SUMO activating enzymes gives rise to the SUMO-SCE covalent complex, but a comparable complex is not formed between SUMO and SCE(C94S) (Figure 5). Thus, SCE(C94S) presumably perturbs SUMO conjugation by competing with un-loaded WT SCE, for instance for binding to SUMO activating enzyme, resulting in the observed net decrease in SUMO conjugation.

Plants with sufficiently high overexpression of SCE(C94S) show reduced growth, reminiscent of mutants devoid of SUMO ligase SIZ1, or of plants with mutation in SUMO protease ESD4 (Murtas et al. 2003; Miura et al. 2005; Hermkes et al. 2011). It was recently shown that siz1 mutants are perturbed in nitrate reductase function, and an additional supply of reduced nitrogen can at least partially relieve the growth defect (Park et al. 2011b). In addition, introduction of a salicylic acid (SA)-degrading nahG transgene, or a mutation in SA biosynthesis, also counterbalances the growth defects of siz1 plants (Lee et al. 2007). We investigated whether an additional supply with reduced nitrogen has an influence on the growth of SCE(C94S) plants (Figure 2). We found that, under our soil and watering regimen, there is a mitigating effect comparable to that observed in siz1 mutants. We therefore conclude that, similar to siz1 mutants, SCE(C94S) expression impacts the performance of nitrate reductase, presumably by decreased SUMO conjugation to this enzyme.

We also investigated transcripts of the cold response for altered abundance in the SCE(C94S) transgenic lines. In particular, RD29A and COR 47 mRNA abundance was decreased by high levels of SCE(C94S) (Figure 6). Interestingly, the related RD29B protein was previously identified as an in vivo sumoylation substrate (Budhiraja et al. 2009), suggesting that certain stress proteins are regulated both at the transcriptional level and by direct modification by the SUMO conjugation system. Another known cold stress-induced gene, transcription factor DREB1A/CBF3 (At4g25480), which has also previously been implicated in SUMO-dependent regulation (Miura et al. 2007), shows no change in SCE(C94S) transgenics (data not shown), suggesting qualitative or quantitative differences between these transgenics and siz1 or esd4 mutants. In line with the less severe phenotype of the SCE(C94S) transgenics compared with siz1 or esd4 mutants, we did not find changes in cold resistance compared to WT plants (data not shown).

Whereas SUMO conjugation defects of Arabidopsis mutants such as siz1 offer an opportunity to analyze sumoylation in the model plant, other plants with higher genetic redundancy do not allow similar approaches. Dominant interference, on the other hand, might offer an opportunity to study sumoylation deficiency in such plant species as well. Also, many plants with complex genomes and poor genetic accessibility are suited for transient overexpression studies. We therefore asked whether in vivo expression of SCE(C94S) is also effective in a transient setting. Transient expression by Agroinfection is possible in Arabidopsis leaves containing AvrPto (Tsuda et al. 2012). Figure 4 shows a significant decrease in higher molecular weight conjugates after transient expression of SCE(C94S) in Arabidopsis. We thus suggest that transient expression of SCE(C94S) can be used to quickly find out whether the dominant negative approach introduced in this study is effective in a particular plant species, before transgenic plants are used for more detailed studies. In summary, we show that expression of an SCE mutant with the active site Cys replaced by Ser can act as a dominant-negative inhibitor of sumoylation, and we point to ways of extending this method to other plant species.

Materials and Methods

Plant genotypes and growth conditions

Arabidopsis plants of genotype Col-0 were grown in the greenhouse under long-day conditions (16 h light, extended by illumination during darker seasons). For determination of flowering time, plants were germinated and grown on soil in an 8 h light, 16 h dark controlled environment at 25 °C.

Vector constructs

SCE (At3g57870) cDNA was cloned from RNA by reverse transcription and polymerase chain reaction (PCR), and inserted into plant binary vector pTCSH1 (Becker et al. 1992; Hardtke et al. 2000; a derivative of pGPTV-BAR) as a Sac I Xba I fragment. The vector allows in planta expression of ORFs using cauliflower mosaic virus (CaMV) 35S promoter sequences and octopine synthase termination signals, and confers Basta resistance to plants. A similar construct containing the Cys 94 to Ser (TGT to AGT) active site mutant was also generated using standard methods. Vectors for expression of SAE, SCE, SCE Cys to Ser (TGT to AGT) active site mutant, and SUMO1 in Escherichia coli were as described previously by Budhiraja et al. (2009).

Protein purification and in vitro activity assays

Tagged SUMO1 (Budhiraja et al. 2009) was purified via His affinity tag as recommended by the resin manufacturer (Qiagen, Valencia, CA, USA), using the following buffer: 10 mM imidazole, 300 mM NaCl, 50 mM Na phosphate pH 8, 10% glycerol, 1 mM phenylmethylsulfonyl fluoride (PMSF) and 1 mg/L each of Aprotinin and Leupeptin. SAE was purified via His affinity tag as SUMO1, except that all buffers contained 10 mM dithiothreitol, 1 mM ATP and 2 mM MgCl2. SCE, which does not contain any tag, was purified by pelleting Escherichia coli cells induced for Arabidopsis SCE. After freezing at −80 °C, 0.03 vol of buffer (50 mM Na phosphate pH 6.5, 50 mM NaCl, 10 mM dithiothreitol) was added to the pellet. After centrifugation (Beckman Optima MAX ultracentrifuge, rotor TLA 100.4, 100 000 g, 1 h), the supernatant was collected for further use. Tagged SUMO1, SAE and SCE preparations were subjected to buffer changes using spin columns (Vivaspin 500, Sartorius Stedim Biotech, Göttingen, Germany) to adjust to reaction buffer conditions (20 mM Tris pH 7.6, 10 mM NaCl, 10 mM MgCl2). For in vitro SUMO thioester formation, 2 μg of SCE and 10 ng of SAE were incubated together with 2 μg of tag3-SUMO1 in 20 μL buffer as indicated above, supplemented with 5 mM ATP, 5 mM Mg(OAc)2 and 1 mM HEPES pH 7.4. After 30 min at 30 °C, samples were withdrawn and incubated in sodium dodecyl sulfate (SDS)-containing sample buffer either with or without dithiothreitol for electrophoresis.

Protein blotting and antisera

Antiserum against SCE was generated in rabbits (Eurogentec, Belgium), and affinity-purified on SCE bound to Affi-Gel 10 (Bio-Rad, Hercules, CA, USA) matrix using standard methods. Electrophoresis was carried out using a Bio-Rad Mini-PROTEAN gel system (Figure 3), or a Life Technologies X-Cell Sure Lock Mini Cell with pre-cast Bis-tris 4–12% gradient gels (Figures 4, 5). Transfer of proteins to a polyvinylidene difluoride (PVDF) membrane (Merck Millipore, Billerica, MA, USA) was carried out in a Bio-Rad Mini-Trans-Blot cell (Bio-Rad). Protein visualization by Western blotting was done essentially as described by Stary et al. (2003) (Figures 3, 5). Anti-rabbit secondary antibody coupled to alkaline phosphatase was purchased from Sigma-Aldrich (St. Louis, MO, USA). Anti-Strep antibody coupled to alkaline phosphatase was purchased from IBA GmbH (Göttingen, Germany). Figure 4 was developed using an Amersham ECL Western Blotting Analysis System (GE Healthcare, Little Chalfont, UK).

Transient assays

Col-0 plants transgenic for the GVG-AvrPto transgene were pre-treated with 7 μM dexamethasone for induction as described by Tsuda et al. (2012). Agrobacteria containing the WT SCE, or the SCE(C94S) transgene were diluted 10-fold from overnight culture and grown for another 4 h. After centrifugation, they were resuspended in 5% sucrose, adjusted to OD600 of 0.1 and used for infiltration of leaf intracellular space. Leaves were harvested after 2 d to obtain protein extracts. For protein extraction, leaves were frozen in liquid nitrogen and ground with quartz sand in the presence of 200 μL extraction buffer (90 mM Hepes, 2% SDS, 30 mM dithiothreitol (DTT), 20 μg/mL pepstatin and one tablet complete mini protease inhibitor cocktail per 7 mL (Roche Applied Science, Penzberg, Germany)). The suspension was heated to 95 °C for 5 min and centrifuged. The supernatant was supplied with 20% glycerol and used for gel loading or for frozen storage.

RNA gel blot analysis

RNA gel blotting was carried out by standard methods as described by Schlögelhofer et al. (2006). The transcript level of At4g18480 (ch42; Koncz et al. 1990) was used as a control.

(Co-Editor: Giovanna Serino)


We thank Michaela Lehnen for technical assistance, Maret Kalda for photography, Andrea Pichler for help with in vitro thioester formation and Karolin Eifler for experimental support. This work was supported by the Max Planck Society, by the German Research Foundation DFG (SFB 635 to G.C., and SPP 1365 and grant BA1158/3–1 to A.B.), by the Austrian Research Foundation FWF (grant P 21215 to A.B.), and by pre-doctoral fellowships from the International Max Planck Research School to R.B. and R.H.