Amide ¹H and ¹⁵N NMR signal assignments of all naturally-occurring di-ubiquitins

Abstract

Ubiquitin acts as a building block for a wide variety of poly-ubiquitin chains. Decoding the role of poly-ubiquitin chains in different cellular processes remains an active area of research. Here, we report amide ¹H and ¹⁵N signal assignments of each ubiquitin unit in di-ubiquitins of all seven lysine linkages and in M1-linked di-ubiquitin determined by our lab over the last decade. These assignments can aid in NMR studies of the structure, dynamics, and function of various di-ubiquitins. Comparison of the NMR resonance assignments among all the di-ubiquitins revealed linkage-specific chemical shifts and isopeptide signals that can be used as “fingerprints” to directly identify using NMR spectroscopy the linkage type in a di-ubiquitin and potentially longer poly-ubiquitin chains. Our data highlight both the similarities and dissimilarities of NMR signals of ubiquitin units in di-ubiquitins of different linkages, as well as the importance of selective isotopic labeling of specific ubiquitin units in a poly-ubiquitin chain for NMR studies.

Biological context

Ubiquitin (Ub) is a highly conserved 76 amino-acid signaling protein found in all eukaryotes. Post-translational covalent attachment (ubiquitination) of a single Ub or of a polymeric chain of Ubs (polyUb) to a lysine of a target protein can determine the fate of that protein. Ubiquitination is involved in the regulation of a wide range of vital cellular processes, including protein degradation, progression through the cell cycle, transcriptional activation, DNA damage repair, antigen processing, vesicular trafficking of proteins etc. (Hershko and Ciechanover 1998; Muratani and Tansey 2003; Wickliffe et al. 2009; Rape 2018; Zhong et al. 2023). PolyUb chains are formed when the C-terminal glycine (G76) of one Ub is attached to another Ub through either an isopeptide bond with an ε-amine of one of the seven lysine residues in Ub (K6, K11, K27, K29, K33, K48, or K63) or a peptide bond with the N-terminal methionine (M1) (Pickart and Fushman 2004). The functional diversity of polyUb chains reflects differential recognition by cellular proteins/receptors of the chains of different linkage compositions and topologies (Fushman and Wilkinson 2011; Komander and Rape 2012; Dikic and Schulman 2023). From the structural/dynamics perspective, polyUb chains are multidomain protein systems where each Ub unit (domain) is relatively rigid, while the linker connecting adjacent Ub units (C-terminal tail of Ub and the isopeptide-linked lysine) is flexible (Fushman et al. 2004; Ryabov and Fushman 2006, 2007; Castaneda et al. 2016a; Fushman 2017). Due to the arrangement of M1 and the lysines across the Ub’s surface, polyUb chains with different linkages can adopt a range of conformations (Berlin et al. 2013; Castaneda et al. 2016a, b; Boughton et al. 2020), and this conformational diversity of polyUb chains is responsible for the variety of ubiquitin-mediated cellular processes.

Since polyUb chains are dynamic and most proteins that interact with them have binding affinities in the K_d range from single-digit to hundreds of µM (Raasi et al. 2005), NMR spectroscopy is particularly suited for studying these chains. However, although monomeric Ub is one of the most well-studied proteins by NMR(e.g., (Tjandra et al. 1995; Tjandra and Bax 1997; Fushman et al. 1998; Massi et al. 2005; Lange et al. 2008), characterizing the structure, dynamics, and ligand interactions of polyUb chains using NMR presents a significant challenge because all Ub units in the chain are chemically identical (or nearly identical), making it almost impossible to distinguish between NMR signals from different Ub units in the chain. To address this challenge, we developed several strategies of polyUb chain assembly that enable unit-specific isotopic labeling (Varadan et al. 2002, 2005; Castaneda et al. 2011a, b; Dixon et al. 2013). This allowed us to study by NMR each individual Ub unit in di-ubiquitins (Castaneda et al. 2016a) and larger chains (e.g., (Nakasone et al. 2013; Boughton et al. 2020; Lemma et al. 2023). To facilitate future NMR studies of polyUb chains, here we report amide ¹H-¹⁵N NMR signal assignments for each Ub unit in all seven lysine-linked di-ubiquitins (Ub₂), as well as in M1-linked Ub₂. Our results reveal linkage-specific similarities and differences in chemical shifts between the two Ub units in Ub₂. We show that the chemical shifts of the C-terminal residues involved in the Ub-Ub linkage, as well as of the isopeptide-bond signals, can be used as characteristic NMR “fingerprints” of different linkage types. We also illustrate the challenge in distinguishing signals coming from the two Ub units when studying M1-linked Ub₂ constructs, where the Ub units have not been individually isotopically labeled.

Methods and experiments

Protein expression and assembly of Ub₂s with different linkages

Human Ub and Ub variants were expressed in the BL21(DE3) strain of E. coli carrying a helper plasmid, pJY2, and purified as previously described (You et al. 1999; Pickart and Raasi 2005). Di-ubiquitins of all lysine-linkages were assembled in vitro, with a single selected Ub unit isotope-labeled per sample (Varadan et al. 2005). To distinguish between the two Ub units in a di-ubiquitin, the Ub with the free C terminus is termed ‘proximal’ while the other Ub whose C terminus is linked to a lysine or M1 of the proximal Ub is termed ‘distal’. Di-ubiquitins linked via K11, K48, and K63 were made enzymatically using lysine-specific E2/E3 enzymes and Ub units containing chain-terminating mutations (Varadan et al. 2002, 2004; Pickart and Raasi 2005; Castaneda et al. 2011b, 2013). K27-, K29-, and K33-linked Ub₂s were made by a non-enzymatic assembly method that uses silver-mediated condensation reaction to form the isopeptide bond (Castaneda et al. 2011a). This method utilizes, as the proximal Ub unit, recombinant Ub in which unlabeled Boc-lysine protected on the side chain, Lys(Boc), is incorporated as a genetically-encoded unnatural amino acid. K6-linked Ub₂s were assembled using either method. M1-linked Ub₂ was expressed as a fusion of His₆-tagged Ub and wildtype Ub.

NMR experiments

NMR samples (50–200 µM) of purified Ub₂s were prepared in 20 mM sodium phosphate buffer at pH 6.8, containing 7–10% D₂O and 0.02% NaN₃. NMR data were acquired at a temperature of 296 or 298 K on a Bruker Avance III 600 spectrometer equipped with a cryoprobe. NMR spectra were processed with TopSpin 4.1.4 and analyzed with Sparky 3.190. Backbone signal assignments for each Ub unit in Ub₂ were obtained based on ¹H-¹⁵N HSQC, SOFAST-HMQC, and TROSY spectra by using the signal assignments of wildtype monomeric Ub (WT Ub) and the respective Ub mutants as a starting point. Signals of the mutated Ub variants were only slightly shifted compared to WT Ub. The assignments were adjusted based on TOCSY/NOESY spectra. An illustrative example of 2D ¹H-¹⁵N NMR spectra and amide signal assignments of the distal and proximal Ubs in K48-linked Ub₂ is shown in Fig. 1.

Fig. 1.

¹H-¹⁵N HMQC spectra of K48-linked di-ubiquitin with only one Ub unit ¹⁵N-labeled: (a) the distal Ub and (b) the proximal Ub, colored teal and black, respectively, as indicated by the cartoons. To distinguish between the two Ubs in Ub₂, the Ub unit with the free C terminus is termed ‘proximal’ while the other Ub attached to it through its C-terminal G76 is termed ‘distal’. Amide signals are labeled by the residue type and number in the respective Ub unit. The peak marked as “K48-iso” in panel b corresponds to the side-chain H_Z-N_Z group of residue K48 in the proximal Ub involved in the isopeptide bond with G76 of the distal Ub

Amide chemical shift differences between the proximal and distal Ubs in Ub₂ for each observed residue were quantified as , where and are the differences in ¹H and ¹⁵N chemical shifts, respectively, for the same residue in the proximal and distal Ub.

Extent of assignment and data deposition

Despite di-ubiquitins being composed of chemically identical (or nearly identical) Ub monomers, the ¹H-¹⁵N spectra of the two Ub units in Ub₂ show significant differences, caused by both linkage-associated chemical modifications and non-covalent effects characteristic for each type of Ub₂ (Fig. 2). The non-covalent interactions involve residue-specific contacts/interfaces between the two Ub units, primarily mediated by the hydrophobic residues on the surface of each Ub. The Ub-Ub interface can be symmetric, as in K48-linked Ub₂, where both Ub units utilize the “canonical” hydrophobic surface patch (Beal et al. 1996) formed by residues L8, I44, V70 (Fig. 3a) (Cook et al. 1992; Varadan et al. 2002), or asymmetric, as in K6-linked Ub₂, where a “non-canonical” hydrophobic patch (residues L8, I36, L71) of the distal Ub contacts the canonical hydrophobic patch on the proximal Ub (Fig. 3b) (Hospenthal et al. 2013), or there might be no defined interface, as in K63-Ub₂ (Fig. 3c) (Varadan et al. 2004). Conformations of Ub₂ (observed in crystals, e.g., Fig. 3a, b) where a noncovalent interface exists between the two Ub units are often referred to as “closed” states (Varadan et al. 2002). It is important to note that the non-covalent/hydrophobic interactions between Ubs are relatively weak, and the C-terminal tail of the distal Ub is flexible (Castaneda et al. 2016a), which means Ub₂ adopts multiple conformational states in solution and is not restricted to the closed state (Berlin et al. 2013). The dynamic interconversion among various states of Ub₂ is fast on the NMR time scale (Ryabov and Fushman 2006), and as a result, each residue is represented by a single amide signal in the ¹H-¹⁵N spectra (Fig. 1).

Fig. 2.

Comparison of the amide NMR signals of the two Ub units in each di-ubiquitin. (a-g) Top: Overlay of amide ¹H-¹⁵N signal assignments of the proximal Ub (in black) and the distal Ub (in teal) for Ub₂s of various linkages, as indicated in the cartoons. (h) Top: Overlay of amide ¹H-¹⁵N signals of His₆-tagged M1-linked Ub₂ (black filled circles) with the signals of His₆-tagged monomeric Ub (teal unfilled circles). The signals of residue I36 are not shown in the top panels because they are located (δ(¹H) ≈ 6.05 ppm) outside the plotted area. Bottom of panels (a-h): Amide chemical shift differences, Δδ, between the proximal and distal Ubs for each observed residue (stars mark residues with no available Δδ value). Residue G76 in Ub₂ of any linkage has Δδ > 0.35 ppm. Residues Q41 and D52 in K27-linked Ub₂ have Δδ > 0.35 ppm

Fig. 3.

Crystal structures of (a) K48-linked, (b) K6-linked, and (c) K63-linked di-ubiquitins (PDB IDs: 1AAR, 2XK5, and 2JF5, respectively) illustrate different arrangements/contacts of Ub units in Ub₂ chains. The distal and proximal Ub units are colored teal and dark gray, respectively. The linkage lysine side chain is shown in stick representation and marked

There is also a significant variation among different lysine-linked Ub₂s in the ¹H-¹⁵N signals of the distal Ub (Fig. 4a) or the proximal Ub (Fig. 5a). The difference in the types of non-covalent contacts is dictated by the location of the linkage lysine and the imposed limitations (e.g., steric hindrance) on the available inter-domain conformations and contacts (Fushman and Walker 2010; Castaneda et al. 2016a), which in turn can determine how various Ub₂s and longer Ub chains are differentially recognized by cellular receptors (Fushman and Wilkinson 2011).

Fig. 4.

Comparison of the amide signals of the distal Ub in all seven lysine-linked di-ubiquitins. (a) Overlay of scatter plots of backbone amide ¹H-¹⁵N resonances of the distal Ub unit among the Ub₂s with different lysine-linkages, and of WT Ub monomer. (b-c) Zoom in on the boxed areas from panel a to show the effect of isopeptide bond formation or tail-extension on the NMR signals of C-terminal tail residues of Ub. (b) Comparison of the ¹H-¹⁵N signals of C-terminal G76 of the distal Ub in various Ub₂s (squares) and in monoUb variants (triangles) containing C-terminal extension by one residue (A77, D77, or N77). (c) Comparison of the ¹H-¹⁵N signals of residue G75 of the distal Ub in various Ub₂s (diamonds) and in monoUb variants (circles) containing C-terminal extension by one residue (A77, D77, or N77). Note that the G76 signal from the distal Ub of K27 Ub₂ is also seen in the G75 region

Fig. 5.

Comparison of the amide signals of the proximal Ub in all seven lysine-linked di-ubiquitins. (a) Overlay of scatter plots of amide ¹H-¹⁵N resonances of the proximal Ub unit in the Ub₂s of all lysine-linkages, and of WT Ub monomer. (b) Fragment of the structure of K11/K48-linked Ub₃ illustrating an isopeptide linkage through K11 (PDB ID: 6OQ1). (c) Zoom in on the boxed area from panel a showing only the ¹H_Z-¹⁵N_Z resonances of the isopeptide bond in the indicated lysine-linked Ub₂s. The position of the isopeptide signal of K33-Ub₂ is from (Michel et al. 2015)

The chemical modifications in a lysine-linked Ub₂ are caused by the formation of an isopeptide bond between the carboxyl group of G76 of one Ub (here termed ‘distal’) and the ε-amine of a lysine of the other Ub (termed ‘proximal’). The isopeptide (and peptide) linkage results in large amide ¹H-¹⁵N signal shifts in the C-terminal residues of the distal Ub, most notably of G76 and G75 (Fig. 4a). Importantly, the observed ¹H-¹⁵N chemical shifts of these residues are unique to each Ub₂ (Fig. 4b, c). Interestingly, G76 and G75 in monoUb also exhibit strong amide signal shifts upon extension of the C-terminus by a single residue (Fig. 4b, c). These unique linkage-specific chemical shifts of G76 and G75 can be used as “fingerprints” to directly identify using NMR spectroscopy the linkage type in Ub₂ and longer chains, as well as to monitor the formation of the isopeptide bond upon ubiquitination or its removal as a result of deubiquitination (see, for example, (Wydorski et al. 2023).

In the proximal Ub, the formation of the isopeptide bond converts lysine’s ε-amine into an amide (Fig. 5b), resulting in a new amide ¹H-¹⁵N NMR signal that appears in the typical chemical shift range of backbone amides (Fig. 5a). This is a unique feature of the isopeptide linkage in polyUb that has not been observed in other proteins. Here, we have assigned these (H_Z-N_Z) signals for K6-, K11-, K48-, and K63-linked Ub₂s (Fig. 5c). The isopeptide signals of the remaining three linkages (K27, K29, K33) were not detected in our studies because the respective Ub₂ constructs were assembled non-enzymatically via unlabeled lysine (see Methods). This unique information can be used to identify, through NMR spectra, the type of lysine involved in the linkage and to examine involvement of the isopeptide linkage in ligand binding (see e.g., (Lemma et al. 2023). Similar to the use of G75 and G76 signals of the distal Ub (see above), the ability to directly observe the H_Z-N_Z isopeptide signal can be utilized to monitor by NMR the processes of ubiquitination and deubiquitination (Wydorski et al. 2023).

In-depth NMR-based studies of the structure, dynamics, receptor recognition and, ultimately, function of different Ub₂s require the ability to distinguish NMR signals of the two chemically identical (or nearly identical) Ub units. Achieving this would not be possible without Ub unit-selective isotopic labeling. This can be illustrated using NMR data for M1-linked Ub₂ where Ub unit-specific spectra are currently unavailable. The majority, 72 of the 80 observed amide ¹H-¹⁵N signals of His₆-tagged M1-linked Ub₂ superimpose almost ideally with those of His₆-tagged monomeric Ub, and only eight extra peaks can be seen, reflecting the asymmetry between the proximal and distal Ub units (Fig. 2h). For example, the signal of C-terminal G76 of the distal Ub has shifted as a result of its covalent attachment to M1 of the proximal Ub (see also Fig. 4b). Similarly, the G75 signal that overlaps with that of mono-Ub originates from the proximal Ub (free C-terminus), while the shifted G75 signal belongs to the distal Ub (Fig. 4c). Similar considerations apply to the shifted signals of the N-terminal M1 and Q2 of the proximal Ub. All the observed amide signals of His₆-tagged M1-linked Ub₂ were assigned, either using TOCSY/NOESY spectra or based on their coincidence with those of His₆-tagged monomeric Ub. However, for the majority of Ub residues (63 out of 76) the amide signals of the same residue from the two Ub units are essentially indistinguishable from each other. Thus, the lack of Ub unit-selective isotopically labeled M1-linked Ub₂ and longer chains makes it challenging to study them by NMR in order to gain insight into Ub-unit/residue-specific interactions involving M1-linked polyUb chains.

Author contributions

I.K., C.A.C., D.F. performed NMR experiments and signal assignments. I.K. and D.F. wrote the main manuscript text and prepared the figures. All authors edited and reviewed the manuscript.

Funding

This work was supported by the National Institutes of Health grant GM065334 to D.F. NMR experiments were performed on instruments supported in part by the National Science Foundation grant DBI1040158.

Data availability

The amide ¹H and ¹⁵N chemical shift assignments for all the naturally-occurring di-ubiquitins have been deposited in the Biological Magnetic Resonance Data Bank (BMRB) under the following accession numbers: K6-linked Ub₂ (52078), K6-linked Ub₂ with chain terminating mutations (52105), K11-linked Ub₂ (53493), K29-linked Ub₂ (52108), K33-linked Ub₂ (52115), K48-linked Ub₂ (52139), K48-linked Ub₂ with chain terminating mutations (52127), K63-linked Ub₂ with chain terminating mutations (52140), K63-linked Ub₂ with K63R mutation in the distal ubiquitin (53018), and M1-linked Ub₂ (53492).

Declarations

Competing interests

The authors declare no competing interests.