Biological context

Calbindin D9k is a small monomeric protein (Mr 8.5 kDa, 76 amino acids) which belongs to the EF-hand family, and consists of two helix-loop-helix motifs that bind one calcium ion each (Kretsinger and Nockolds 1973). Calbindin D9k undergoes small structural changes upon calcium-binding, involving the rearrangement of non-polar side chains (Ikura 1996). The protein is predominantly found in the mammalian epithelial cells of the small intestine and placenta, and it has been implicated to facilitate the transport of calcium across the intestinal epithelial cells (Christakos et al. 1989).

High resolution three dimensional structures of Calbindin D9k in various calcium-loaded states have been characterized extensively by X-ray crystallography and solution state NMR spectroscopy (Kordel et al. 1997; Kordel et al. 1993; Szebenyi and Moffat 1986). Although the complete resonance assignment of 13C, 15N and 1H nuclei for this protein was not used for structure determination, it will facilitate a comprehensive study of its dynamics, structure, dihedral angle distributions and electrostatic interactions, as well as supplying data for comparisons with solid-state NMR and chemical shift calculations by quantum chemical methods.

Methods and experiments

MM294 E. coli cells transformed by the PCBWR plasmid containing the calbindin Pro43Gly gene (Bos taurus) were used for protein expression. A single colony was picked from agar plate and grown overnight in 100 mL LB medium with ampicillin at 30°C. 20 mL of the overnight culture was added to 500 mL minimal medium containing U-13C-glucose and U-15N ammonium chloride at 30°C. Protein production was started by ten-fold dilution of the cells into medium containing U-13C glucose, U-15N ammonium chloride and 0.1 mg/mL IPTG at 37°C. Purification of Calbindin D9k P43G was performed as in a previous study (Thulin 2002).

All experiments (see Table 1) were carried out on Varian Unity INOVA 500 and 600 MHz spectrometers equipped with pulsed field gradient probes. The spectra were recorded at 301 K. The NMR sample contained ~2.5 mM [13C,15N]-enriched Calbindin D9k P43G, pH 6.0, 7% D2O. The spectra were processed using NMRPipe (Delaglio et al. 1995) and analyzed using Sparky (Goddard and Kneller 2003).

Table 1 List of experiments
Full size table

Assignments and data deposition

Relation to previous assignment (BMRB entry 327): Only 1H signals of Calbindin D9k P43G had been assigned. Here the 100% complete assignment of non-labile 1H, 13C and 15N signals for calcium-loaded D9k P43G was achieved using an extensive suit of standard and non-standard 2D and 3D NMR experiments.

Backbone and aliphatic side chains

Backbone assignments were obtained using 3D experiments such as HN(C′)N and HNN to obtain the amide proton and nitrogen chemical shifts (Fig. 1 displays the dispersion of NH–HN chemical shift of Calbindin D9k from 1H–15N HSQC). Several 2D projections of triple resonance experiments; H(N)CO, HA(CA)CO, and HACA(N) were recorded to assign the carbonyl, alpha proton and alpha carbon resonances. Most aliphatic side chain signals were assigned using 3D experiments; (H)C(CO)NH-TOCSY and H(CCO)NH-TOCSY, which correlate the backbone nitrogen and amide proton shifts to aliphatic carbon and proton frequencies of residue i − 1. For residues which are preceded by proline, the side chain nuclei were assigned using a 3D 15N–1H TOCSY-HSQC experiment which correlates NH, HN, and proton aliphatic side chain in the same residue. The HCCH-COSY experiment, which gives information about chemical shifts of protons bound to carbon, enabled us to verify the assignment for long side chains like lysine, leucine, glutamate, isoleucine, valine and proline. The specific assignment of lysine Nζ were obtained using the 2D H2(C)N experiment. The signals in this spectrum are well dispersed between 31 and 34 ppm (see Fig. 2a). The 15N shift of the N terminal methionine residue was obtained using 2D H2(C)N pulse sequence where the final shaped carbon inversion pulse was replaced by a full power rectangular 180º pulse (Andre et al. 2007). The carbonyl side chains of glutamate, glutamine, aspragine and aspartate were detected using (HBGCBG)CO(CBGCABCACON)H experiments which correlates the side chain carbonyl of residue i to the amide proton of residue i+1 (see Fig. 3). This experiment is very powerful to get the unambiguous chemical shift assignment of carbonyl/carboxyl side chains due to the excellent dispersion of amide proton chemical shifts in folded proteins. However, the sensitivity of the experiments was not sufficient to detect all signals (Q22 was absent) and a H2(C)CO experiment was used to detect the Q22 Cδ–Hγ correlation.

Fig. 1
figure 1

2D 1H–15N HSQC spectrum of uniformly 15N/13C-labelled Calbindin D9K. All peaks are annotated with the one letter amino acid symbol and their position in the sequence. All amide proton and nitrogen nuclei in the backbone of Calbindin D9k were observed. The 15Nε and 1Hε chemical shits of glutamine and the chemical shifts of 15Nδ and 1Hδ of asparagines are also indicated

Full size image
Fig. 2
figure 2

Some specific assignments of Calbindin D9k side chains. a 2D H2(C)N spectrum showing Lys Nζ–Hε correlations. b 2D CG(CB)HB spectrum to assign aromatic side chain resonances

Full size image
Fig. 3
figure 3

2D 1H–13C (HBGCBG)CO(CBGCABCACON)H spectrum of uniformly labelled 15N/13C Calbindin D9k. Correlation can be observed for carboxyl/carbonyl side chain 13C′ of glutamate, glutamine, asparagine and aspartate of residue i with the amide proton of residue i+1. Peaks labeled AC–H, BC–H and CC–H refer to carbonyl and amide proton peaks from the soluble cyclic enterobacterial common antigen, ECACYC (Erbel et al. 2003)

Full size image

Aromatic side chains

Calbindin D9k P43G contains 5 phenylalanines and 1 tyrosine residue. Some specific strategies were required to assign the aromatic ring 1H and 13C resonances. Aromatic Cγ resonances were assigned using a combination of CG(CB)HB, 3JNCγ and 3JC′Cγ experiments. The sequence-specific side chain 1H assignment of the aromatic side chains was obtained via CB(CGCD)HD and CB(CGCDCE)HE experiments. The information of Hδ and Hε in the aromatic rings from these experiments were used to assign 1H–13C HSQC CP aro, aromatic 1H–13C CT HSQC and 1H–13C HMQC aromatic experiment. F10 Cδ–Hδ was only observed in the non constant time 1H–13C HSQC or 1H–13C HMQC experiment due to the strong coupling within the aromatic ring. A 3D 1H–13C HSQC-NOESY experiment was used to verify the assignment of the aromatic side chains.

To summarize, the 1H, 13C and 15N resonance assignments of Calbindin D9k P43G have been deposited in the BioMagResBank (accession number 16340). Although complete side chain resonance assignments were obtained for Calbindin D9k P43G, it should be mentioned that it does not contain any cysteine, arginine, histidine and tryptophan residues. Those residues, in particular, require specific strategies for their side chain assignment. For arginine guanidine groups, sequence specific assignment of 15N and 1H chemical shifts have been presented by Yamazaki et al. 1995, and for histidine and tryptophan ring, sequence specific assignment of 1H, 13C, 15N have been established by Löhr et al. (2005).

Our study shows that complete assignments of all NMR-active nuclei in small protein can be obtained and describes a suitable strategy for this purpose. In particular, lysine Nζ chemical shifts appear to be difficult to get correct, as witnessed by the BMRB database. Currently (grid update of August 16th, 2010) 110 chemical shift assignments for lysine Nζ are available, but this list contains as many as 25 erroneous assignments. In three cases, a 1H chemical shift was entered for Nz and in 22 instances chemical shifts between 67 and 133 ppm have been listed, either as a result of exchanging the assignment with that of backbone nuclei, or by the incorrect account of spectral aliasing. Even today, the lysine Nζ statistics are heavily polluted and yield 47.8 ± 32.9 ppm for the full set. A restricted set of 14 entries now gives 34.1 ± 3.0 ppm, as opposed to 73.8 ± 50.3 ppm for 7 entries in 2004.