Backbone and side chain resonance assignment of the intrinsically disordered human DBNDD1 protein

The dysbindin domain-containing protein 1 (DBNDD1) is a conserved protein among higher eukaryotes whose structure and function are poorly investigated so far. Here, we present the backbone and side chain nuclear magnetic resonance assignments for the human DBNDD1 protein. Our chemical-shift based secondary structure analysis reveals the human DBNDD1 as an intrinsically disordered protein.

The canonical human protein DBNDD1 (UniProtKB: Q9H9R9), the focus of our study, is 158 amino acids long with a high content of the acidic residues glutamate and aspartate (13% and 7%, respectively) as well as serine (6%) and threonine (8%). Unlike other dysbindin family proteins, DBNDD1 isoforms are probably non-classical secretory proteins (Talbot et al. 2009).

3
Additionally, it is a proline-rich (10% prolines) cytoplasmatic protein with expression in nearly all organs and e.g., neuronal cells. No expression could be detected in the ovary, the adipose tissue, and the bone marrow [ (Uhlen et al. 2015), https:// www. prote inatl as. org].
The Pfam database [(Mistry et al. 2021), https:// pfam. xfam. org/] predicts human DBNDD1 mainly as an intrinsically disordered protein (IDP) and also the recently released AlphaFold database (Jumper et al. 2021;Varadi et al. 2022) predicts human DBNDD1 -with a short stretch of helical propensity between residues L77 and S95 -entirely as an IDP. Interestingly, S95 (beside S119) is one of the two reported phosphorylation sites. Along with S65, S95 is proposed to constitute a casein kinase 1 interaction site while S119 might be modified by cyclin-dependent kinase 5 (Talbot et al. 2009).
We performed a Basic Local Alignment Search Tool (BLAST) analysis to identify regions of local similarity between the human DBNDD1 and protein sequences from other species (Fig. 1). As an outcome human DBNDD1 revealed a high sequence identity to dysbindin domaincontaining proteins from other Hominidae (e.g., G. gorilla gorilla and P. paniscus 99% and 97% identity, respectively). Likewise, the proteome of Old and New World monkeys contains DBNDD1-like proteins with sequence identities to human DBNDD1 of approximately 95%. Proteins with high sequence identity to human DBNDD1 can also be found in evolutionarily more distant species (e.g., M. musculus and X. laevis 80% and 61% identity, respectively). The sequence conservation of the putative dysbindin domain across all selected species is notable ( Fig. 1 shaded region).
Although, a high sequence conservation also suggests a conservation of structure and function, current experimental insights into the structure or function are missing on human DBNDD1 with the exception of some experimental data indicating that the DBNDD1 gene is associated with melanoma risk and that the DBNDD1 level is decreased in Parkinson's disease mouse models (Auburger et al. 2016;Fang et al. 2020). Also, a negative regulation of protein kinase activity is predicted for DBNDD1.
In contrast, its paralog dysbindin-1, the first family member discovered (Benson et al. 2001), is more intensively described. Dysbindin-1 contains a coiled-coil domain, a structural component known to e.g., facilitate biological maintenance, repair, replication, trafficking processes and enzymatic activities (Truebestein and Leonard 2016). Expression of dysbindin-1 is ubiquitous in the body and in virtually all neuronal cells (Talbot et al. 2009). For instance, dysbindin-1 was shown to be involved in neurite extension and synaptic vesicle trafficking (Auburger et al. 2016). Mutations in dysbindin-1 are responsible for the Hermansky-Pudlak syndrome (Li et al. 2003) and genetic variations of dysbindin-1 are associated with psychiatric conditions like psychosis, bipolar disorder, major depression, and schizophrenia (Straub et al. 2002;Talbot et al. 2009).
From the data available for DBNDD1 and its paralogs, it becomes clear that DBNDD1 may be involved in essential cellular processes. Thus, investigation of human DBNDD1 can broaden our understanding of the exact function of this protein and help to explain the previously observed associations with pathological manifestations.

Protein expression and purification
We ordered a synthetic gene coding for full-length human DBNDD1 from Thermo Fischer Scientific (Germany). The coding sequence was optimized for expression in E. coli.
The gene was subcloned into a pET28a expression vector using Ndel and Xhol restriction enzymes, thereby introducing a N-terminal hexahistidine fusion. The resulting construct was verified by DNA sequencing (LGC Genomics GmbH, 120 Germany). For expression, transformed Escherichia coli BL21 (DE3) cells were plated onto kanamycin plates. A single colony was picked to inoculate a first LBmedium preculture. At an OD 600 of 0.6 cells were diluted 1:50 in M9 mineral salts medium grown again. This step was repeated with fresh M9 medium. Subsequently, cells were diluted 1:70 in 250 mL M9 medium main culture supplemented with 1 g/l 15 NH 4 Cl and 4 g/l 13 C 6 -labeled glucose. Gene expression was induced at an OD 600 of 0.6-0.8 by adding 1 mM IPTG (isopropyl-1-β-d-galactopyranoside). Cells were harvested after 4 h by centrifugation (5250xg, 30 min, 4 °C). All cultures were grown at 37 °C and supplemented with 50 μg/ml kanamycin.
For purification, cells were resuspended in 40 mL lysis buffer (11.5 mM Na 2 HPO 4 , 8.5 mM NaH 2 PO 4 , 500 mM NaCl, 10 mM imidazole, pH 7.0) containing a protease inhibitor cocktail (cOmplete Mini from Roche Diagnostics GmbH, Mannheim, Germany). Cells were disrupted by sonification while placed on ice and then centrifuged (40,000 rpm, 40 min, 4 °C, Beckman Coulter Optima L-90 K Ultracentrifuge). The supernatant was loaded onto a pre-equilibrated Ni-NTA affinity chromatography column (ÄKTA prime plus, QIAGEN Ni-NTA Superflow Cartridge 1 × 5 ml) at 4 °C. After washing with 10 column volumes of the lysis buffer human DBNDD1 was eluted with 11.5 mM Na 2 HPO 4 , 8.5 mM NaH 2 PO 4 , 500 mM NaCl, 500 mM imidazole, pH 7.0. Further purification was done by size exclusion chromatography (HiLoad 16/60 SD75, GE Healthcare) using 10 mM sodium phosphate buffer at pH 6.5, 150 mM NaCl. Fractions containing human DBNDD1 were pooled and concentrated. Sample purity was verified by SDS-PAGE and mass spectrometry. The final concentration of the human DBNDD1 NMR sample was about 300 µM.
Of note, the used construct has a thrombin cleavage site between the N-terminal His 6 tag and the native human DBNDD1 sequence. Although no canonical thrombin cleavage site is predicted within human DBNDD1 sequence, the addition of thrombin led to the rapid protein degradation. Therefore, the removal of the purification tag was omitted, and the amino acid numbering is as follows: −19 to 0 indicates the purification tag, the native human DBNDD1 sequence starts with methionine number 1.

NMR spectroscopy
1 H-detected NMR spectra on human DBNDD1 were recorded at 283.2 K on a 700.5 MHz Bruker AvanceIII NMR spectrometer system equipped with a 5 mm TXI triple resonance probe (Bruker Biospin GmbH, Rheinstetten, Germany). Spectra with direct 13 C detection were recorded at 293.2 K on a Bruker AvanceIII 700 MHz spectrometer equipped with cryogenic TXO probe at CERM/CIRMMP (Florence, Italy). The spectrometers were locked on D 2 O.
For direct 1 H chemical shift referencing as 0.00 ppm we added 3-(trimethylsilyl)propane-1-sulfonate (DSS) at a final concentration of 0.1 mM to the NMR samples. 13 C and 15 N chemical shifts were referenced indirectly to the 1 H DSS standard by the magnetogyric ratio (Wishart et al. 1995).
The three-dimensional 1 H-detected experiments were recorded with 25% non-uniform sampling. Compressed sensing with an iteratively reweighted least squares algorithm was used for data reconstruction (Kazimierczuk and Orekhov 2011;Holland et al. 2011). All spectra were processed using Bruker Topspin 3.6.2 or 4.1.1 and analyzed using CcpNmr Analysis 2.5 (Vranken et al. 2005) within the NMRbox virtual environment (Maciejewski et al. 2017).

Structure prediction
For the sequence-based prediction of structural disorder we used the ODiNPred web server (https:// st-prote in. chem. au. dk/odinpred) (Nielsen and Mulder 2019; Dass et al. 2020). Figure 5A(I-II) shows the ODiNPred disorder prediction of human DBNDD1. ODiNPred predicts fully disorder approximately for the first 50 amino acids (residues M1-A49) in the N-terminal part, followed by a stretch of roughly 50 amino acids where the fractional formation of local order is predicted. After a short stretch (residues E98-R113) of fully disorder the partial formation of local order is also predicted for the C-terminal part (residues E140-D158) of DBNDD1.
According to the predicted structural disorder, we used the POTENCI tool (https:// st-prote in02. chem. au. dk/ poten ci) to calculate the random coil chemical shifts for human DBNDD1 based on the amino acid sequence considering temperature, pH value and ionic strength (Nielsen and Mulder 2018).
Additionally, we used the programs SSP (Marsh et al. 2006) and TALOS-N (Shen and Bax 2013), respectively, to examine potential secondary structure elements of DBNDD1 based on the assigned backbone chemical shifts.

Extent of assignments and data deposition
By using a set of two-and three-dimensional NMR experiments (s. Methods and experiments) we achieved the sequence specific resonance assignments for nearly all backbone 1 H, 13 C and 15 N spins of human DBNDD1. We could assign 99% of the backbone resonances (C α , C′, N′, H N ). For the side chain protons and carbons (ß, γ, δ, and ε positions) the assignment could be completed to 73% and 76%, respectively. Table 1 summarizes the extent of assignment.
In agreement with a predicted low overall secondary structure content, the [ 1 H, 15 N]-HSQC spectrum of human DBNDD1 shows limited signal dispersion in the 1 H N dimension (Fig. 2).
The backbone 13 CO, 15 N-correlations of neighboring residues in the 2D CON experiment are given in Fig. 3.
We assigned the 13 C β and 13 C γ resonances for 15 out of the 16 proline residues in DBNDD1. The 13 C γ resonance assignment of proline residue P120 is missing due to signal ambiguity. All assigned proline residues show 13 C β and 13 C γ values in the range of 32.09 ± 0.08 ppm and 27.42 ± 0.09 ppm, respectively, with a mean difference of the proline 13 C β and 13 C γ chemical shifts of 4.68 ± 0.08 ppm. The obtained proline 13 C β chemical shift values are plotted versus the 13 C γ chemical shift values in Fig. 4. Based on the 13 C β and 13 C γ chemical shift values, we assume that in its major conformation all completely assigned proline residues of human DBNDD1 are in a trans configuration (Schubert et al. 2002;Shen and Bax 2010). Moreover, the absence of an additional subset of peaks with lower intensity in the proline specific region of the CON spectrum (Fig. 3) supports the statement that all prolines are exclusively in trans configuration. We used the obtained chemical shifts of human DBNDD1 for an initial structural analysis based secondary chemical shifts. The differences between the secondary 13 C α and 13 C β chemical shifts and the secondary structure propensity (SSP), respectively, (Fig. 5A), III-IV) were calculated using the SSP script (Marsh et al. 2006). An overall intrinsic disorder of DBNDD1 is supported by the application of secondary chemical shifts and the sequence specific SSP method. Although consecutive positive and negative differences of secondary 13 C α and 13 C β chemical shifts are observable, their magnitude are comparatively low to predict reliably secondary structure elements. The SSP method combines C α , C β and H α chemical shift values into single residue specific scores. The calculated SSP scores predict the entire human DBNDD1 protein as highly disordered (Fig. 5A, IV). In contrast to the sequence-based disorder prediction, an analysis based on the measured chemical shifts also reveals the proposed dysbindin domain (residues L53-D97) as highly disordered. The mean SSP score is  − 0.016 ± 0.113 and by averaging the calculated SSP scores, an overall total of only 3.2% α-helical and 5.5% β-sheet structure is estimated for human DBNDD1. Additionally, we compared the experimentally determined chemical shifts with random coil chemical shifts, predicted at our experimental conditions using the POTENCI web server (Nielsen and Mulder 2018). The measured and predicted C α , C β , C′, N′, H N , H α and, H β chemical shift values agree remarkably (Fig. 5B, I-VII). The mean differences between the experimental and POTENCI-predicted random coil chemical shift values for human DBNDD1 are ΔC α = 0.04 ± 0.19 ppm, ΔC β = 0.03 ± 0.27 ppm, ΔC′ = 0.04 ± 0.18 ppm, ΔN′ = 0.14 ± 0.53 ppm, ΔH N = -0.02 ± 0.08 ppm, ΔH α = 0.04 ± 0.04 ppm, and ΔH β = 0.05 ± 0.05 ppm.
Together, our experimental data and the secondary structure prediction based on them clearly show that human DBNDD1 is an IDP under buffer conditions chosen to somewhat mimic cellular conditions while providing optimal conditions for NMR spectroscopy. However, it is still speculative if the proposed dysbindin domain or parts of the C-terminal region prone for fractional local order are molecular recognition features that might fold upon binding. In addition, the effect of potential post-translational modifications on the structural dynamics of DBNDD1 remains elusive. It is likely that in a cellular context certain serines, threonines or the tyrosine are phosphorylation sites.
The inherent flexibility of IDPs renders NMR spectroscopy a suitable method to study the presence of local conformational preferences at a molecular level. Here, we report the backbone and side chain NMR resonance chemical shift assignments and provide an initial chemical-shiftbased secondary structure analysis of the hitherto structurally "unknown" human protein DBNDD1. Hopefully, we can lay a foundation to adequately describe the fluctuating conformational behavior of DBNDD1 at atomic resolution and, thereby to gain a better understanding of DBNDD1 function and regulation in a cellular context. Fig. 4 Proline 13 C β and 13 C γ chemical shift analysis for human DBNDD1 reveals all Xaa-Pro peptide bonds in trans conformation. Filled circles correspond to the assigned proline 13 C β and 13 C γ chemical shifts. 15 out of 16 prolines were completely assigned (C γ of P120 is unassigned). The open circle and the open triangle indicate the location of the mean (standard deviation shown as error bars) for a proline in trans and cis conformation, respectively (Schubert et al. 2002;Shen and Bax 2010) Acknowledgements Open Access funding provided by Projekt DEAL. Support by the "Institut für Technische Biochemie (ITB) e.V." affiliated at the Martin Luther University Halle-Wittenberg is gratefully acknowledged. The FLI is a member of the Leibniz Association (WGL) and is financially supported by the Federal Government of Germany and the State of Thuringia. We sincerely thank Isabella Felli and Fabio Calogiuri (CERM/CIRMMP, Florence, Italy) for experimental support and helpful discussions.
Funding Open Access funding enabled and organized by Projekt DEAL. This work benefited from access to CERM/CIRMMP, Florence, and was supported by iNEXT-Discovery (No. 871037), a European Additionally, two regions with low predicted disorder propensities are in the C-terminal part of DBNDD1 (shaded in light grey). The N-terminal part is predicted to be fully disordered. The circles show the residue-specific Z-score (A), I) and disorder probability (A), II). A residue specific Z-score larger than 8 (solid line) indicates structural order while a Z-scores below 3 (dashed line) predicts fully disorder. Z-scores between 3 and 8 reflect transient local structure propensity. The Z-score and disorder probability were calculated using the ODiNPred webserver (Dass et al. 2020). The differences between secondary chemical shifts of 13 C α and 13 C β resonances (A), III) and the secondary structure propensity (SSP) prediction (A), IV) based on chemical shifts were calculated using the SSP script (Marsh et al. 2006). A positive and negative SSP score reflect α-helix and β-sheet propensities, respectively. A SSP value of 1 reflects fully formed helical-structure and a value of -1 fully formed β -structure, respectively. Only 13 C α , 13 C β and 1 H α chemical shifts of non-proline preceding residues were applied when running the SSP script. B Secondary chemical shifts analysis reveals that the human DBNDD1 is highly disordered throughout the entire protein sequence. Chemical shift differences calculated from the experimentally determined and predicted C α , C β , C′, N′, H N , H α , H β chemical shifts (B), I-VII). POTENCI (Nielsen and Mulder 2018) was used for sequence-based chemical shift prediction Commission Horizon 2020 project, for the access to the NMR instrumentation (PID: 16180).

Data availability
The assigned 1 H, 13 C and 15 N chemical shift values of the human DBNDD1 are available in the BMRB (https:// bmrb. io) under the Accession No 51301.

Conflict of interest
The authors declare that they have no conflict of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visithttp:// creat iveco mmons. org/ licen ses/ by/4. 0/