Response to "Is the His6-HA-SUMO1 knock-in mouse a valid model system to study protein SUMOylation?"
James A Daniel (1), Benjamin H Cooper (1), Jorma J Palvimo (2), Fu-Ping Zhang (3), Nils Brose (1), Marilyn Tirard (1)
(1) Max Planck Institute of Experimental Medicine, Molecular Neurobiology, Göttingen, Germany; 2) Institute of Biomedicine, University of Eastern Finland, Kuopio, Finland; 3) Institute of Biomedicine, and Turku Center for Disease Modeling, University of Turku, Turku, Finland
Wilkinson et al. provide a critical commentary regarding our study on SUMO1-conjugation in neuronal synapses (Daniel et al., 2017), in which we report that the evidence for SUMO1-conjugation at neuronal synapses and of several synaptic proteins is equivocal. The corresponding study was based on the analysis of three mouse models: (1) wild-type (WT) mice, (2) a knock-in mouse line that expresses His6-HA-tagged SUMO1 from the endogenous Sumo1 locus (His6-HA-SUMO1-KI) (Tirard et al., 2012), and (3) a knock-out mouse that lacks SUMO1 expression (SUMO1-KO) (Zhang et al., 2008).
Summary of biochemical data on proposed SUMO1 substrates
Reflecting our general focus on the development and function of neuronal synapses, we generated the His6-HA-SUMO1-KI with the aim of studying synaptic SUMO1-conjugation. We were surprised by the fact that the His6-HA-SUMO1-KI model provided no evidence for synaptic SUMO1-conjugation (Tirard et al., 2012). Given the discrepancy between our study (Tirard et al., 2012) and several previous reports proposing synaptic SUMO1 substrates, we sought to reconcile this controversy through the present study by focusing specifically on some of the synaptic proteins that had been proposed as SUMO1 substrates. We used immunoaffinity purification of proteins from WT and His6-HA-SUMO1-KI brain tissue to test eight candidate SUMO1-conjugated proteins for His6-HA-SUMO1-conjugation - the transcription factor Zbtb20 and seven candidate synaptic SUMO1 substrates (synapsin 1A, gephyrin, GluK2, RIM1, syntaxin 1A, synaptotagmin 1, and mGluR7). Corresponding ‘reverse experiments’, in which the target proteins were immunopurified from WT and His6-HA-SUMO1-KI brains and detected by Western blotting with anti-HA and candidate-protein-specific antibodies, were performed for synapsin-1, gephyrin, GluK2, RIM1, and syntaxin-1A. Of the proteins we examined, only Zbtb20, which we had identified in our initial unbiased proteomic screen for SUMO1-conjugated proteins in brain (Tirard et al., 2012), yielded evidence of SUMO1-conjugation as assessed by Western blot analysis. For the synaptic proteins we tested, no bands with the appropriate SUMO1-conjugation-induced size shift were detected by Western blotting (Figures 1-5; Daniel et al., 2017).
Based on the data and arguments outlined above, our publication challenges the notion that synapsin-1, gephyrin, GluK2, RIM1, syntaxin-1A, synaptotagmin-1, and mGluR7 are bona-fide SUMO1-conjugation substrates in mouse neurons in vivo (Daniel et al., 2017). Our present candidate screen does not provide information on any of the other types and classes of extranuclear proteins that the introduction of the commentary by Wilkinson et al. refers to, and we do not propose that our study unequivocally resolves the issue. However, we conclude that based on our data the role of SUMO1-modifications in the function of synaptic proteins and synapses remains - at least - unclear. Accordingly, by using landmark studies in other areas of SUMO biology for guidance, we propose in the discussion part of our publication a set of consensus criteria that should be met so that a candidate protein can be considered a bona fide SUMO substrate in neurons in vivo (Daniel et al., 2017). We expect that adherence to these criteria, and the development of genetically engineered mice that allow the unequivocal mass spectrometric identification of SUMO-conjugated peptides in proteolytic digests of proteins from mouse brain, subcellular brain fractions, or purified protein fractions (e.g. by introducing specific proteolytic cut sites in SUMOs proximal to the isopeptide link), will ultimately resolve the present controversy.
General validity of the His6-HA-SUMO1-KI model
The core of the criticism by Wilkinson et al. focuses on the use of the His6-HA-SUMO1-KI mice. Specifically, Wilkinson et al. point to our reports that overall SUMO1-conjugation is reduced by ~20-30% in these mice, as assessed by Western blot and immunocytochemical analyses (Daniel et al., 2017; Tirard et al., 2012). In view of these changes, the validity of the His6-HA-SUMO1-KI as a reporter for SUMO1-conjugation was dismissed in previous reviews on neuronal and synaptic SUMO biology (e.g. Henley et al., 2014; Luo et al., 2013) - and is dismissed again in the present commentary. In all these cases, the lack of evidence for synaptic SUMO1 conjugation in our studies has been attributed to the reduction in SUMO1-conjugation levels and the inclusion of the His6-HA-tag in the His6-HA-SUMO1-KI.
We have repeatedly acknowledged that overall SUMO1-conjugation is reduced by ~20-30% in His6-HA-SUMO1-KI brain and that certain SUMO1-conjugated protein variants might be too transient, unstable, or rare to be detected reliably with our methodology (Tirard et al., 2012; Daniel et al., 2017). However, many previously identified SUMO1-conjugation substrates were unequivocally detected in our unbiased proteomic screens and immunoaffinity purification experiments using the His6-HA-SUMO1-KI (Tirard et al., 2012). Further, multiple novel SUMO1-substrates were identified by us, including Zbtb20 (Tirard et al., 2012), which was also found in other proteomic screens for SUMOylation substrates (Becker et al., 2013; Hendriks et al. 2017) and which we stringently verified in the present study (Daniel et al., 2017). We therefore regard it as unlikely that all seven synaptic candidate SUMO1-conjugates we tested escaped our detection method, e.g. due to a complete occlusion effect of the ~20-30% reduction in overall SUMO1-conjugation levels or the influence of the His6-HA-tag in His6-HA-SUMO1-KI mice. Furthermore, when we tested SUMO1-conjugation of recombinant Zbtb20, synapsin-1, gephyrin, and GluK2 in fibroblasts that co-expressed HA-tagged SUMO1, Western blot analyses showed that only Zbtb20 was SUMO1-conjugated (Figures 1-4; Daniel et al., 2017). In these experiments, the candidate proteins and HA-SUMO1 were strongly overexpressed under the control of a cytomegalovirus promoter. Given this overexpression, we consider it unlikely that a lack of SUMO1-conjugation of the synaptic candidate proteins would be due to an intrinsic ~20-30% decrease of HA-SUMO1-conjugation (as proposed by Wilkinson et al. for the His6-HA-SUMO1-KI). While one cannot solely rely on such data from cultured non-neuronal cells in vitro, they support our in vivo data. Furthermore, replacement of SUMOs by tagged variants - in some cases involving tags that are much larger than the His6-HA- and HA-tags tag we used for the His6-HA-SUMO1-KI - is well tolerated by Saccharomyces cerevisiae (Panse et al., 2004), Arabidopsis (Miller et al., 2010), and Caenorhabditis elegans (Kaminsky et al., 2009).
Use of wild-type material in analyses of candidate protein SUMOylation
Along with their general criticism of our His6-HA-SUMO1-KI model, which we discussed above, Wilkinson et al. allege that we did not use WT material in our analysis of candidate SUMO1-conjugated proteins.
In this context, we would like to point out that we immunopurified Zbtb20 (Figure 1B), synapsin-1 (Figure 2B; Daniel et al., 2017), gephyrin (Figure 3B; Daniel et al., 2017), GluK2 (Figure 4B; Daniel et al., 2017), RIM1 (Figure 5B; Daniel et al., 2017), and syntaxin-1A (Figure 5D; Daniel et al., 2017) from WT mouse brain and assessed the input and immunoisolated proteins by Western blotting using antibodies against the different proteins with the aim of detecting higher molecular weight bands that could be attributed to SUMO-conjugation of the respective proteins. It is of note that these experiments do not allow us to determine whether SUMO-conjugation is due to SUMO1, SUMO2, or SUMO3. Only in the case of Zbtb20 did we obtain unequivocal evidence of protein species with molecular weight shifts that likely represent SUMO-conjugation. In all other cases, no protein species with an apparent molecular weight shift indicative of SUMO1/SUMO2/SUMO3-conjugation were detected in WT (or His6-HA-SUMO1-KI) samples. These data also address another criticism by Wilkinson et al. - that we did not examine SUMO2/SUMO3-conjugation of the synaptic proteins. Since no molecular weight shift was detected for any of the synaptic proteins, we can rule out conjugation to any of the three SUMO paralogues. In general, we did not specifically focus on SUMO2/SUMO3 in our study because - with the exception of mGluR7 (Choi et al., 2016) and gephyrin (Ghosh et al., 2016) - most of the previous studies on SUMOylation of synaptic proteins had focused exclusively on SUMO1.
Immunodetection of SUMO1 in neurons, synapses, and subcellular fractions
In the introduction of their commentary, Wilkinson et al. mention several publications that reported the apparent presence of SUMO1 and/or SUMO1-conjugated proteins (and of components of the SUMOylation machinery) in synapses of neurons in culture and brain sections, and in biochemically isolated subcellular brain fractions. The corresponding studies were based on immunostaining and Western blotting approaches with a range of anti-SUMO1 antibodies that Wilkinson et al. refer to as 'validated'. However, we feel that the data that Wilkinson et al. cite cannot be taken at face value because the anti-SUMO1 antibodies used were not validated by comparing WT samples to SUMO1-KO samples as negative control. We regard this as a major omission, particularly in view of the many different fixation, permeabilization, and staining protocols that have been used, which affect antigen detection and antibody cross-reactivity.
Lack of antibody specificity is a notorious problem. Large-scale studies indicate that only ~50% of commercially available antibodies can be used to reliably assess protein distribution in tissue (see Baker, 2015, for a critical commentary). In line with this, we show in our study that 'synaptic' signals generated with a 'validated' anti-SUMO1 antibody in cultured neurons (Figures 12-15; Daniel et al., 2017) show no significant difference to the 'synaptic' signal in SUMO1-KO samples. Similarly, most of the anti-SUMO1-positive bands in synaptic fractions of mouse brain were evident in both His6-HA-SUMO1-KI and SUMO1-KO samples (Figure 6; Daniel et al., 2017). These data indicate that the anti-SUMO1 antibody generates non-specific signals that can be erroneously interpreted as synaptic SUMO1-conjugation. As discussed by us (Daniel et al., 2017) but essentially ignored before, the conclusion that widely used anti-SUMO1 antibodies generate non-specific signals is supported by the fact that anti-SUMO1 immunolabelling in neuronal dendrites has been reported to be punctate in some studies (Craig et al., 2015; Konopacki et al., 2011, Loriol et al., 2013; Martin et al., 2007) and relatively homogeneous in others (Craig et al., 2012; Ghosh et al., 2016; Kantamneni et al., 2011). In addition, one of the publications cited in the commentary of Wilkinson et al. in the context of evidence for synaptic localization of SUMO1-conjugates (Hasegawa et al., 2014) provides beautiful images to demonstrate that SUMO1, SUMO2, and SUMO3 are present in many cell types throughout the brain. However, the anti-SUMO immunoreactivity in these images appears to be exclusively localized to cell nuclei. The authors did not employ antibodies against synaptic markers in their study and did not make claims about the presence of SUMO-immunoreactivity at synapses.
As regards our neuron immunolabelling experiments to detect SUMO1, Wilkinson et al. mention the low SUMO1 signal intensity in nuclei of WT and His6-HA-SUMO1-KI neurons and argue that “these low detection levels would almost certainly rule out visualization of the far less abundant, but nonetheless functionally important, extranuclear SUMO1 immunoreactivity”. In our experiments, we made a considerable effort to recapitulate the methods of immunolabelling used in previous studies regarding permeabilization conditions and anti-SUMO1 antibody concentration to rule out a methodological basis for the differences between our observations and previous studies. We acknowledge that the nuclear anti-SUMO1 immunolabelling is in fact somewhat higher in WT neurons than in His6-HA-SUMO1-KI neurons (Figure 1 - Figure Supplement 1; Daniel et al., 2017). This relates to the ~20-30% reduction in endogenous SUMO1-conjugation levels the His6-HA-SUMO1-KI brain. We note that in some previous studies (e.g. Jaafari et al, 2013; Gwizdek et al., 2013) images of anti-SUMO1 immunolabelling are saturated, which we wanted to avoid. Due to the very large number of images that were acquired per neuron we also used minimal laser power to prevent bleaching of the anti-SUMO1 signal.
We now provide unprocessed sample images from the dataset used in our study showing the intensity of immunolabelling as a heat map (Figures A and B). The corresponding images show that immunolabelling of the nucleus of Triton-X-100-permeabilized neurons from WT mice is much stronger than in neurites and the surrounding cytoplasm. In SUMO1-KO neurons the specific nuclear labeling is not visible and anti-SUMO1 immunolabelling appears uniformly weak throughout the cell (Figure A). This is also clearly represented quantitatively in Figure 16 (Daniel et al., 2017). This validates the conclusion that the SUMO1-KO results in a dramatic loss of specific nuclear anti-SUMO1 immunolabelling. As regards immunolabelling in neurites, we found the anti-SUMO1 immunolabelling in Triton-X-100-permeabilized cells to be very weak. This was also noted by the Henley-group and led to their preferential use of digitonin as a permeabilisation agent in SUMO1 immunolabelling. We now present unaltered heat map images of neurites of digitonin-permeabilized neurons (Figure B). These images were automatically generated by the Fiji macro that we used for the anti-SUMO1 intensity quantification at synapses and show the regions defined as 'synaptic' based on anti-synapsin immunolabelling. While anti-SUMO1 immunolabelling is sparse in WT neurons, we note that it is punctate. However, most anti-SUMO1 puncta do not appear to correspond to synapsin-positive synaptic sites, and puncta are equally evident in WT and SUMO1-KO neurons (Figure B).
These images, and the quantitative analyses presented in our study, demonstrate again that antibodies can generate non-specific signals that may be erroneously interpreted as specific. Indeed, regarding the notion that the "low detection levels would almost certainly rule out visualization" of synaptic SUMO1, we have not only visually (qualitatively) examined these images but quantified the synaptic anti-SUMO1 signal using two independent, unbiased methods. One would assume that if anti-SUMO1 immunolabelling at synapses were specific but relatively weak, the anti-SUMO1 immunolabelling should be higher in WT than in SUMO1-KO neurons. Thus, we conclude that specific anti-SUMO1 immunolabelling is either absent or is of such low abundance as to be undetectable using our methods. Overall, the relatively weak anti-SUMO1 immunolabelling in neurons also highlights one of the advantages of the His6-HA-SUMO1-KI model, given the strong and specific labelling achieved with an antibody against HA in this system as compared to anti-SUMO1 immunolabelling. In agreement with our findings, a recent publication (Matsuzaki et al., 2015) also shows virtually no specific overlap between anti-SUMO1 and anti-synaptophysin immunolabelling in brains of mice that overexpress SUMO1 (Figure 2; Matsuzaki et al., 2015), even with highly saturated images.
Regarding our Western blot analyses of brain subcellular fractions from His6-HA-SUMO1-KI and SUMO1-KO mice, Wilkinson et al. argue that we should have used WT mice and not His6-HA-SUMO1-KI mice to compare to the SUMO1-KO samples, presumably because WT mice have a ~30% higher overall SUMO1-conjugation level as compared to His6-HA-SUMO1-KIs. What Wilkinson et al. do not acknowledge is the fact that our analysis shows that apparent (or alleged) SUMO1 signals in synapses of cultured neurons and in synaptic subcellular fractions are equally well detected in SUMO1-KO samples, indicating that they are of non-specific and/or artefactual nature (Figures 6 and 12-15; Daniel et al., 2017). Nevertheless, we generated new subcellular fractions from WT and SUMO1-KO brains, and assessed them by Western blotting with six different anti-SUMO1 antibodies (Figure C). Apart from the fact that such subcellular fractions always contain non-cognate contaminants (see for instance the co-purificatiopn of RanGAP1 in synaptic plasma membrane fractions), our data show that (i) all antibodies show non-specific cross-reactivity with proteins in SUMO1-KO samples, that (ii) there is very little correspondence between datasets obtained with the different antibodies, and that (iii) the vast majority of protein bands that are detectable in WT synaptic fractions by these anti-SUMO1 antibodies are equally well detectable in synaptic SUMO1-KO fractions. These observations stress again the requirement of KO controls for any immunodetection assay and demonstrate that the evidence for the presence of SUMO1-conjugated proteins in synaptic fractions is truly equivocal.
Functional studies
In the introduction of their commentary, Wilkinson et al. state that the SUMOylation of many classes of extranuclear neuronal proteins has been functionally 'validated'. The authors criticize that our own experiments were confined to immunolabeling and Western blotting analyses (Daniel et al., 2017), and state that an examination of the functional effects of target protein SUMOylation by electrophysiology or transmitter release assays would have been at least as important. They argue "that simply because SUMO1-ylation of a protein is beneath the detection sensitivity in a model system that exhibits sub-endogenous levels of SUMO1-ylation, does not mean that protein is not a functionally important and physiologically relevant SUMO1 substrate".
We did not perform experiments to assess the functional consequences of SUMO1-conjugation of the seven candidate synaptic proteins we studied because we did not obtain any evidence that these proteins are SUMO1-conjugated in the first place. Accordingly, we did not discuss previously published functional analyses of SUMO1-conjugation of these candidate proteins; in the absence of any functional analyses of our own, it did not seem appropriate to us to discuss the validity of corresponding functional studies conducted by others.
In more general terms, the typical strategy employed for functional analyses of SUMO1-conjugation of a given protein has a major intrinsic limitation. In the currently most stringent approach, the functional characteristics of the WT form of a given protein is compared to the characteristics of a variant in which an identified or alleged SUMO-conjugated lysine residue is mutated so that SUMO-conjugation at this position is abolished. Under optimal conditions, this functional comparison is conducted as a 'rescue experiment', i.e. on a background where the expression of the endogenous protein at hand is blocked by KO or knock-down. The general problem with this approach is that it is in principle impossible to conclude with certainty that the functional consequences of the corresponding lysine mutation are specifically and only due to the blockade of SUMOylation at this site - because the lysine mutation itself might have SUMOylation-independent consequences, e.g. by affecting the secondary/tertiary/quaternary structure of the protein, its interaction with other proteins, or its modification at this site by other lysine modifications. In view of this limitation, corresponding functional studies are only valuable if the lysine residues in focus represent truly validated SUMOylation sites. Particularly in cases where the in vivo SUMOylation of a given protein and the identity of a given SUMOylation site are equivocal, the functional consequences of a corresponding lysine mutation must be interpreted with care as they do not necessarily relate to changes in protein SUMOylation. This problem is even more profound in cases where WT and lysine-mutant variants of proteins are strongly overexpressed and/or compared in a WT background. There is currently no easy way to circumvent these intrinsic limitations of typical functional analyses of protein SUMOylation.
Anti-GluK2 antibodies
A final critical issue raised in the commentary by Wilkinson et al. concerns the detection of SUMOylated GluK2 by Western blotting. After pointing out that GluK2 is a prototypic synaptic SUMO1 substrate that has been validated in exogenous expression systems, neuronal cultures, and rat brain (Chamberlain et al., 2012; Konopacki et al., 2011; Martin et al., 2007; Zhu et al., 2012), Wilkinson et al. state that a key flaw in our attempts to assess the SUMO1-conjugation of GluK2 is "that the C-terminal anti-GluK2 monoclonal rabbit antibody used does not recognise SUMOylated GluK2 because its epitope is masked by SUMO conjugation". They conclude that "due to [these] technical reasons, [our experiments] could not possibly detect SUMOylated GluK2 whether or not it occurs in the KI mice".
This is clearly an interesting, important, and possibly problematic piece of information. However, it has so far only been conveyed anecdotally, and we did not see this issue systematically assessed, explained, or proven in any of the four papers cited above, nor in any other publication. Further, it is difficult to reconstruct from available published records why the antibody we used (i.e. MerckMillipore rabbit monoclonal anti-GluK2 antibody NL9; rmAb-MerckMillipore-NL9) might not recognize SUMOylated GluK2. We assume, based on the time of publication and corresponding information in the methods text, that the anti-GluK2 antibody used in the first study to detect SUMO1-conjugated GluK2 (Martin et al., 2007) was Upstate rabbit polyclonal anti-GluK2 antibody 06/309 (rpAb-Upstate-06/309). This antibody, which has unfortunately been discontinued and is no longer available, was raised against a lysine-linked peptide representing the C-terminal 15 amino acid residues of rat GluK2 (Lys-HTFNDRRLPGKETMA), which are positioned seven residues downstream of the proposed SUMO1-conjugation site in GluK2, K886. As stated above, we used for our study rmAb-MerckMillipore-NL9, which was raised against the exact same sequence of the C-terminus of rat GluK2 (linked to keyhole limpet hemocyanin, KLH) as rpAb-Upstate-06/309 (i.e. KLH-HTFNDRRLPGKETMA). Given that the two antibodies relevant in this controversy were raised against exactly the same C-terminal GluK2-sequence, which is proximal to but does not include the proposed SUMO1-conjugation site K886, it is not immediately obvious why SDS-denatured, SUMO1-conjugated GluK2 should be readily detectable on Western blots by one antibody (i.e. rpAb-Upstate-06/309) but not by the other (i.e. rmAb-MerckMillipore-NL9).
Further complicating the issue, we interpret the available published information to indicate that two of the studies cited by Wilkinson et al. in the context of problems with the ability of rmAb-MerckMillipore-NL9 to detect SUMO1-conjugated GluK2 on Western blots did actually employ rmAb-MerckMillipore-NL9 to detect SUMO-conjugated GluK2 (Konopacki et al, 2011; Zhu et al, 2012): (i) Konopacki et al. (2011) state under 'Materials and Methods' in the 'Supporting Information' part of their publication that rmAb-MerckMillipore-NL9 was used to detect purified GluK2 C-termini in in vitro assays, and show in Figure 3A and Figure S2 of the same publication Western blots of apparently in-vitro-SUMO1-conjugated C-terminal fragments of GluK2. (ii) According to the 'Materials and Methods' part of the corresponding publication, Zhu et al. (2012) used rmAb-MerckMillipore-NL9 for all analyses of SUMO-conjugation of GluK2. Ignoring, for the sake of the argument, other issues with this study (Zhu et al., 2012) and taking the data provided at face value, one has to assume again that rmAb-MerckMillipore-NL9 can detect SUMO1-conjugated GluK2.
Despite the arguments above, we further examined whether we can detect SUMO-conjugated GluK2 with antibodies other than rmAb-MerckMillipore-NL9. We performed additional Western blot analyses with two alternative anti-GluK2 antibodies, (i) Abcam rabbit polyclonal anti-GluK2 antibody 66440 (rpAb-Abcam-66440), which was raised against an N-terminal epitope, and (ii) Alomone rabbit polyclonal anti-GluK2 antibody AGC-009 (rpAb-Alomone-AGC-009), which was raised against a C-terminal epitope (amino acid residues 858-870) that excludes the proposed SUMOylation site in GluK2 and is distinct from the epitope used to generate rmAb-MerckMillipore-NL9. The corresponding Western blots (Figure D) show no evidence of a GluK2-positive band with shifted molecular weight, supporting our initial conclusion (Daniel et al., 2017) that evidence for GluK2 SUMOylation remains equivocal.
Conclusion
We published our study on synaptic SUMO1-conjugation (Daniel et al., 2017) to highlight discrepancies in the published record that should not be simply dismissed, to stimulate a corresponding discussion, and to encourage activities towards a consensus set of criteria based on which a candidate protein can be considered a bona fide SUMO substrate in neurons in vivo. Accordingly, we are grateful for the commentary by Wilkinson et al., but feel, as outlined above, that important aspects of our paper were not considered in sufficient detail and depth.
Based on the arguments made in the preceding paragraphs, we maintain that the His6-HA-SUMO1-KI mouse line is a reliable and useful tool for the localization and identification of SUMO1-conjugation substrates, particularly when used alongside WT and SUMO1-KO mice. Further, we maintain that our data and the arguments made above indicate that evidence for SUMO1-conjugation at neuronal synapses and of several synaptic proteins remains equivocal.
Figures
https://cdn.elifesciences.org/annotations-media/3691261705-001-b15d2cee192f951a0c91b2817386ecf306f49d78772cc0ba931dd522478270d4.jpg
Figure A: Anti-SUMO1 immunolabelling in nuclei of WT and SUMO1-KO neurons.
Primary hippocampal neurons were fixed, permeabilized using Triton X-100, immunolabelled with anti-SUMO1, and imaged as detailed in Daniel et al. (2017). The two images on the left show a representative confocal section through the neuronal nucleus/soma of a WT neuron, labelled with DAPI and anti-SUMO1 antibodies. The images on the right show a section through a SUMO1-KO neuron. The fluorescence intensity of the images is represented using the fire LUT from Fiji. The WT neuron shows nuclear anti-SUMO1 immunolabelling, which is absent in the SUMO1-KO neuron. These images are from the same dataset that was used to generate Figure 16 in Daniel et al. (2017). Scale bars, 10 µm.
https://cdn.elifesciences.org/annotations-media/3691261705-004-7badbbfb079d360cabd46204daf4124bee797a1afa8f0e14c98fb0d7aa717b24.png
Figure B: Anti-SUMO1 immunolabelling in neurites of WT and SUMO1-KO neurons.
Primary hippocampal neurons were fixed, permeabilized using digitonin, immunolabelled with anti-SUMO1/anti-synapsin antibodies, and imaged as detailed in Daniel et al. (2017). The images on the left show a representative confocal section through the neurites/synapses of a WT neuron, labelled with anti-synapsin and anti-SUMO1 antibodies. The images on the right show a section through a SUMO1-KO neuron. The images in the lower panels show the detail of an inset region (400 x 400 pixels) from the upper panels. The fluorescence intensity of the images is represented using the fire LUT from Fiji. "Fiji's ‘Fire’ Heat map lookup table was applied to images to visualize fluorescence intensity (0 to 4095, as shown in scale)". White-outlined regions of interest (ROIs) around synapsin-positive puncta were generated by a custom Fiji macro and are shown applied to anti-SUMO1 images as well. Synapsin puncta in which SUMO1 signal is visible are marked with white arrowheads in both WT and SUMO1-KO cultures. Synapsin puncta in which SUMO1 signal was essentially undetectable are marked with open arrowheads in both WT and SUMO1-KO cultures. These images are from the same dataset that was used to generate Figure 12 in Daniel et al. (2017). Scale bars, 10 µm.
https://cdn.elifesciences.org/annotations-media/3691261705-003-77b83859809f5bd4e01352c3d2460b500e39a48a7923460f59180434c53616df.jpg
Figure C: SUMO1-conjugated proteins in subcellular brain fractions.
Brains from adult WT and SUMO1-KO mice were subjected to subcellular fractionation as detailed in Daniel et al. (2017). Western blot analyses of the fractions using anti-GluN1 and anti-synaptophysin antibodies validate the fractionation procedure. Western blot analyses of the fractions using six different anti-SUMO1 antibodies confirm the strong enrichment of SUMO1 candidates in nuclear fractions (P1) but not in synaptic fractions (LP1, SPM). H, homogenate; P1, nuclear pellet; S1, supernatant after P1 sedimentation; P2, crude synaptosomal pellet; S2, supernatant after P2 sedimentation; P3, cellular membrane and organelle fraction; S3, supernatant after P3 sedimentation; LP1, lysed synaptosomal membranes; LS1, supernatant after LP1 sedimentation; LP2, crude synaptic vesicles; SPM, partially purified synaptic plasma membranes. Arrows indicate free SUMO1; stars indicate non-specific bands detected by the anti-SUMO1 antibodies.
https://cdn.elifesciences.org/annotations-media/3691261705-002-321bb61bc21dc123be91cabfff43693483d48c2b0bb0cc5dfbcc24c1526017d8.jpg
Figure D: Analysis of SUMO1-conjugation of GluK2 in HEK cells.
Representative SDS-PAGE (10%) followed by anti-HA and anti-GluK2 Western blotting of input and eluate fractions from anti-HA immunopreciptation in the presence of 20 mM NEM from HEK cells overexpressing HA-SUMO1 and GluK2, alone or in combination. Anti-HA Western blot analysis confirms the enrichment of HA-SUMO1-conjugates (top panel) in the immunopurified fractions (right panels). No SUMO1-GluK2 bands were observed in the eluate fractions, irrespective of the anti-Gluk2 antibody used. Images are representatives of at least three independent experiments.
