Ultra-high resolution X-ray structures of two forms of human recombinant insulin at 100 K
© The Author(s) 2017
Received: 5 March 2017
Accepted: 12 July 2017
Published: 1 August 2017
Success in the use of pig insulin to control diabetes ultimately lies in its ability to mimic the activity of the human form, which is a consequence of near perfect structural isomorphism. However, the use of non-human forms of insulin to control diabetes is known to lead to both allergic reactions and other complications resulting from antibody production in some patients . For this reason the use of recombinant forms of human insulin which have now been developed is becoming more commonplace, on the assumption that their structure–function properties are even more closely related to the natural hormone. There are 2 independent molecules in the asymmetric unit of the crystal structure of hexagonal porcine insulin : molecule 1 comprising peptide chains A1 and B1, and molecule 2, comprising peptide chains A2 and B2 (the 4 chains are now usually designated A, B, C and D). Peptide chains A and C are identical in sequence, as are chains B and D. Chains A and B, and chains C and D are linked by disulphide bridges Cys7A–Cys7B, Cys7C–Cys7D, Cys20A–Cys19B and Cys20C–Cys19D, respectively. Chain A also has an internal stabilizing disulphide bridge Cys6A–CysA11 and there is a corresponding S–S bridge in chain C, Cys6C–CysC11. As shown in Additional file 1: Figure S1 there are 3 AB and 3 CD dimers in the unit cell grouped around a crystallographic threefold axis. In the 2Zn crystals, three non-crystallographic insulin dimers are assembled around two Zn ions on the threefold axis. Each Zn ion is coordinated to three symmetry-related Nε atoms of HisB10 and to three water molecules. Water oxygen atoms (282) were also assigned and included in the refinement which converged to a value of R = 0.153 for 10,119 significant Iobs(hkl). Seven of the amino acid side-chains were assigned less ordered conformations, refined with separate atomic coordinate sets and occupancy factors. Commercial human recombinant insulin is now available from several sources. The present study describes the ultra-high resolution (0.92 Å) low temperature structure of Insugen (I) human recombinant insulin, Fig. 1 and Additional file 1: Figure S2a. The unpublished structure of a second recombinant form of human recombinant insulin, from Intergen, at the same resolution, deposited as structure 3W7Y in the Protein Data Base (in June 2013) shows a number of surprising differences when compared with the Insugen (I) structure reported here. These two structures will be referred to as Insugen (I) and Intergen (II). The Insugen (I) and Intergen (II) 2Zn hexagonal HR insulin structures are predominantly isomorphous with that of porcine 2Zn insulin . In both of these new structures the A and B-chains of molecule 1 are in the T-state .
Implications for biological activity
See also “General comments”, “Peptide side chain electron density and conformations in Intergen (II) [PDB 3W7Y]”, “Comments on the solvated propanol in Insugen (I)”, “PheB24 and PheB25 in Insugen (I) and Intergen (II)” for further discussions of the implications of structure for biological activity.
Materials and methods
Materials Insugen (I)
Human recombinant insulin (Insugen-30/70) was supplied by Biocon (India) Ltd. See Additional file 1: Table S1a. Human recombinant insulin, Intergen (II) was produced by the INTERGEN Company and purchased by Sakabe  from the SEIKAGAKU Company. Details are to be found in Additional file 1: Figure S2b. Other chemicals including HCl, zinc acetate, acetone, trisodium citrate and NaOH were purchased from Fisher Scientific (UK) and Sigma-Aldrich (UK).
Crystallization of Insugen (I)
The crystals were prepared at room temperature by a batch method similar to that of Baker et al. , modified as follows: 0.01 g of insulin as a fine powder was placed in a clean test tube; 1 mL of 0.02 M HCl was added to dissolve the protein; on addition of 0.15 mL of 0.15 M zinc acetate the solution became cloudy due to precipitation of the protein; 0.3 mL of acetone and then 0.5 mL of 0.2M trisodium citrate together with 0.8 mL of water were added and the solution became clear; the pH was checked and increased with NaOH to a pH between 8 and 9 for different batches, thus ensuring complete dissolution. It was then adjusted to the required value of pH 6.3. If any slight turbidity occurred, it was removed by warming the solution. The solution was then filtered using a Millipore membrane/acetate cellulose acetate filter. This removes any nuclei which will encourage precipitation or formation of masses of small crystals.
The solution was then warmed to 50 °C by surrounding the test tube with preheated water in a Dewar. This allowed the solution to cool slowly to room temperature. The test tube was lightly sealed with cling film; crystals formed within a few days and were of a suitable size for X-ray diffraction within 2 weeks; the test tube containing crystals was kept at 4 °C prior to data collection. The crystal used for data collection was about 0.2 mm3.
Crystallization of Intergen (II)
The following details were supplied by Sakabe . In contrast to the Insugen (I) crystals, Intergen (II) crystals were grown using the vapour diffusion hanging drop method at 293 K. The reservoir solution contained 0.1 M sodium citrate, and 22% (w/v) DMF, and 0.08% (w/v) zinc chloride, pH 8.67 while the protein solution was insulin, Intergen (II) dissolved in 0.02 N HCl to a final concentration of 10 mg/mL. The starting volume of the reservoir solution was 1 mL, and the volume of the drop was 20 μL of protein and reservoir solution in a 1:1 ratio. In 4 or 5 days, crystals were observed to have formed, and after 10 days to 2 weeks, insulin crystals of a size suitable for X-ray diffraction studies were present, typically about 0.5 mm × 0.5 mm × 0.3 mm. The crystal used for 3W7Y data collection was about 1.2 mm × 0.7 mm × 0.5 mm .
X-ray data collection
Insugen (I) crystal at Diamond Light Source, MX beamline I02
Crystals grown at room temperature were passed through a 30% glycerol solution, prepared in mother liquor, prior to cryo cooling in liquid nitrogen. Crystals were screened with three test shots, separated by 45° using 0.5 s exposure and 0.5° oscillation. Data were collected at 16,000 keV (λ = 0.77 Å) and 100 K with the Pilatus 6 M detector as close to the sample as possible (179.5 mm). The EDNA strategy  was used to obtain a start angle and 180° of data were collected with 0.1° oscillation and 0.1 s exposure. The resolution of useful diffraction data achieved and used for structure analysis was 0.92 Å. The spacegroup is H3 (146) and the unit cell is a = b = 81.827 Å, c = 33.849 Å, α = β = 90° γ = 120°. Further details can be found in Additional file 1: Table S1.
X-ray data collection for Intergen (II) crystal at the Photon Factory beamline BL-6C (Ibaraki, Japan)
The following details were supplied by Sakabe . A synchrotron data set to 0.7 Å was collected at the Photon Factory beamline BL-6C using wavelength λ = 0.97974 Å. Data were measured on a specially designed Weissenberg type instrument known as “Galaxy”, employing a fully automated high speed imaging plate detector. The detector comprised a vertically focussing 1 m long bent mirror of Pt-coated fused silica at a distance of 21 m from the SR source point and 7 m from the focal point. The low resolution limit was 50.0 and high resolution limit 0.7 Å; the number of reflections observed was 91.73%; Rmerge for Iobs = 0.05579 for 57006 hkl’s. The resolution of useful diffraction data achieved and used for structure analysis was 0.92 Å [10–14]. The space group is: H3 (146); the unit cell is: a = b = 81.120 Å, c = 33.930 Å, α = β = 90° γ = 120°.
X-ray data processing for Insugen (I) crystal
Manual processing of the data was carried out using XDS  to integrate and Aimless  to scale and merge intensities. The purpose of manual scaling was to optimise the included data to maximise the final resolution to 0.92 Å.
Structure solution and initial refinement
Further details can be found in Additional file 1: Table S1.
Presence of Zn in the Insugen (I) Crystal
Model building and further least squares refinement
Data-collection and final refinement statistics
Insugen (I) (Biocon)
Intergen (II) (Intergen)
a = b = 81.8270 Å
a = b = 81.120 Å
c = 33.8490 Å
c = 33.930 Å
α = β = 90°
α = β = 90°
γ = 120°
γ = 120°
VM (Å3 Da−1)
Solvent content (%)
22 (223 waters)
25 (275 waters)
Resolution range (Å)
No. of measurements
No. of unique reflections
R merge a
Residues in allowed regions of Ramachandran plot (%)
R.m.s.d. bonds (Å)
R.m.s.d. angles (°)
All-atom Baverage (Å2)
Model building and further least squares refinement for Intergen (II)
The structure for 3W7Y was determined by molecular replacement and refined using the program REFMAC . Non-hydrogen atoms were refined anisotropically. Several residues were modelled as two clear conformers with complementary occupational parameters having a sum of 1.0. At the end of the refinement the R factor and Rfree were 0.162 and 0.180, respectively. Inspection of the Ramachandran plot revealed that 96.81% of the residues are in the allowed regions. All coordinates and data are deposited in the Protein Data Bank, with identification code 3W7Y.
Superficially the ultra-high resolution structure of HR insulin (Insugen I), as expected, strongly resembles that of 2Zn porcine insulin (see “Introduction”) having an asymmetric unit with 2 independent molecules: molecule 1, comprising peptide chains A and B; and molecule 2, comprising peptide chains C and D. Peptide chains A and C are identical in sequence, as are chains B and D.
As described below there are significant and interesting differences between the detailed ultra-high resolution structures of Insugen (I) and Intergen (II) and also between the two human recombinant insulin structures and the less detailed porcine insulin . For example in the porcine insulin structure  289 waters were assigned and in Intergen (II) 275. However after intense scrutiny and assessment 220 water molecules have been included and refined in the Insugen (I) structure. Further features of interest in the Insugen (I) structure are: (i) an acetate molecule ACT2101 (or simply ACT) has been assigned in the neighbourhood of Zn2100 in molecule 1 and is in fact coordinated with this Zn. This unexpected feature is described below and is presumably a consequence of the zinc acetate used in the crystallization procedure. The acetate molecule has excellent refinement parameters and geometrical features. To the best of our knowledge acetate has not been assigned to any other published insulin structure; further evidence for this assignment can be found in Additional file 1: Text S1 and Figure S3: (ii) a solvated propanol molecule has been assigned as described below in detail. The propanol molecule POL5001 (or simply POL) forms H-bonds with the prominent Oγ1A of ThrD27 located on the A conformation of ThrD27 which has two clearly defined conformations A and B, of which A has 0.645 occupancy and B 0.355 occupancy. POL is also H-bonded to water 6007. Further evidence for the assignment of propanol can be found in Additional file 1: Text S2 and Figure S4. There is no evidence of propanol solvate close to ThrB27 in chain B which has a single fully occupied conformation (see below). Intergen (II) shows no evidence of either acetate or propanol in the electron density for the deposited 3W7Y structure. To the best of our knowledge solvated propanol has not been reported as present in any other determined insulin structure. Possible origins of the solvated propanol are examined. As discussed below other differences occur between the two human recombinant insulin crystal structures. Such differences may ultimately be of importance with respect to the hormonal and biological activities of these synthetic therapeutics .
Description of the secondary structure regions in Insugen (I)
The ultra-high resolution refinement of HR insulin, Insugen (1) undertaken in the analysis described above has enabled a study of the secondary structure motifs in the insulin molecule to be carried out in detail which exceeds all previous studies.
Chain A (Fig. 4a)
Helix A1: This involves the first 9 residues GlyA1–SerA9 and comprises about 2 turns of a distorted α-helix. Although GlyA1 involves a bifurcated H-bond and its (φ, ψ) values are indeterminate because it is N-terminal, this residue does seem to be part of the helix. SerA9 is at the C-terminal end of the helix, its side chain H-bonding to the peptide N of IleA10. Details are in Fig. 4b.
Strand A2 (Fig. 4a)
Strand A2 runs from IleA10–SerA12 forms an anti-parallel sheet with strand B1 in the B-chain (see below). Note there is only one β-bridge, at CysA11.
Helix A3 (Fig. 4a, c): this secondary structure involves LeuA13–TyrA19 and is a 7 residue 310 helix. The SerA12 side-chain caps the N-terminal end of the helix by H-bonding to the peptide N of GlnA15, whose side-chain in turn forms an H-bond to the N of SerA12. The carbonyl of SerA12 forms the first H-bond of the helix, but the (φ, ψ) values of SerA12 suggest it is part of the preceding strand and not this helix.
Strand A4 (Fig. 4a): CysA20 and AsnA21 appear to form a mini strand and participate in an anti-parallel sheet with strand B4 (Fig. 6a) in the B-chain strand. The carbonyl oxygen of TyrA19 forms the first H-bond of the strand although it is part of the preceding helix.
Chain C (Fig. 5a)
Strand C2 (Fig. 5a): IleC10–SerC12 form a strand, based on the (φ, ψ) values. SerC9 is probably not part of this strand as its φ, ψ value is at the edge of the β-strand region (φ = −90o,ψ = −164o). The strand extends to SerC12.
Helix C3 (Fig. 5a, c): LeuC13–TyrC19 is a 7 residue 310 helix comprising about 2 turns. SerC12 caps the N-terminus end with its side-chain forming an H-bond with the peptide N of GlnC15, while the side-chain of GlnC15 forms an H-bond with the peptide N of SerC12.
Strand C4 (Fig. 5a): CysC20 and AsnC21 comprise a mini strand and this forms an anti-parallel sheet with strand D4 in the D-chain (see below).
Chain B (Fig. 6a)
Helix B2 (Fig. 6a, b): This extends from GlyB8 to CysB19 (12 residues), about 3-turns of α-helix. Note GlyB8 does not have helical (φ, ψ) values but does have a 310 turn H-bond.
Central Loop B3 (Fig. 6a): There is a type I turn from GlyB20–GlyB23 and an open α-turn from CysB19 to GlyB23.
Strand B4 (Fig. 6a): In terms of (φ, ψ) values, this strand could be considered to extend from PheB24 to ThrB30, but in terms of H-bonds in the sheet, it ends at LysB26. It forms an anti-parallel sheet with D4. Note that strands A4 and C4 are part of this four-strand sheet.
Chain D (Fig. 7a)
Helix D2 (Fig. 7a, b): This is a 12 residue α-helix from GlyD8 to CysC19. Note CysD7 is part of strand D1, GlyD8 does not have helical (φ, ψ) values but does have a bifurcated H-bond and CysD19 is helical.
Central Loop D3 (Fig. 7a): The region CysD19–GlyD23 forms an open-α turn and GlyD20-GlyD23 form a type I turn.
Strand D4 (Fig. 7a): This extends from PheD24 to TyrD26. It forms a sheet with strand B4 and this sheet also comprises strands A4 and C4.
Type I Turn D5 (Fig. 7a, c): This is a type I turn and comprises ThrD27, ProD28, LysD29 and ThrD30.
Solvated water molecules in Insugen (I)
Types of H-bond involving water and their numbers: W–SC water–side chain, W–W water–water
Type of H-bond W–SC, W–W
Type of H-bond W–SC, W–W
Salt bridges in Insugen (I)
Insugen (I) residues involved in salt bridge formation
Distance in Å
Water–side chain interactions in Insugen (I)
The table above indicates the 18 residues in Insugen (I) which do not form any hydrogen bond interactions with solvated water molecules. There are 2 in both chains A and C; 8 in chain B; and 6 in chain D. Entries in bold are common to two chains, either A and C, or B and D.
It is of interest to note that 14 of the 18 residues that do not associate with solvated water molecules are located in α-helical structures. These are: CysA6 (helix A1); LeuA16 (helix A3); LeuB11, ValB12, AlaB14, LeuB15 and CysB19 (helix B2); IleC2 (helix C1); LeuC16 (helix C3); LeuD11, ValD12, AlaD14, LeuD15 and CysD19 (helix D2).
Survey of the peptide side chain electron density and conformations in Insugen (I) and Intergen (II)
Peptide side chain electron density and conformations in Insugen (I)
It is well known that ultra-high resolution protein structures derived from X-ray diffraction data using cryo cooled crystals often reveal amino acid residues which display more than a single ordered conformation. See for example Smith et al.  and Addlagatta et al. . When such effects are observed it is possible that the use of these harsh high speed experimental conditions have both caused and allowed these alternative structures to be captured for detailed examination. It is also possible that such alternative conformations may have a bearing on the biological activity of the protein. As described below, the present ultra-high resolution structures of human recombinant insulin Insugen (I) and Intergen (II) both display several amino acid residues having two distinct ordered conformations. As described in detail below the same residues are not necessarily affected in corresponding protein chains in either the Insugen (I) or Intergen (II) structure. Thus, somewhat surprisingly, the disordered regions do not match 1:1 between the two recombinant structures or between corresponding protein chains in the same structure. A detailed analysis and comparison is given below. It is possible that these structural features may affect the biological functions of these recombinant insulins .
Insugen (I) chain A
The electron density of Insugen (I) chain A is of very high quality (mainly blue) with few problems associated with fitting the amino acid residue structures; only the C-terminal residue N21 exhibits a double conformation with two weak regions of density at the end of the chain.
Insugen (I) chain C
In contrast chain C exhibits the following characteristics: residues Q5, Y14 and Q15 have mainly good density but with some poorly defined regions; residues C6 and C11 participating in an S–S bridge, and L16 demonstrate clear electron density but corresponding to double residue conformations with good geometry (orange). The remaining residues are clearly defined in strong electron density (blue).
Insugen (I) chain B
The electron density of Insugen (I) chain B exhibits the following characteristics: residues L11, V12, and E13 and T27 have clear electron density with two distinct conformations (orange); residues Q4 and L17 show mainly clear double conformations but with some poor density at the extreme end; residue F25 clearly adopts two conformations but both phenyl rings A and B occupy very weak regions of density; residues K29 and T30 are mainly clear single conformations but with some terminal disorder. The remaining residues are clearly defined in strong electron density (blue).
Insugen (I) chain D
In contrast the electron density of Insugen (I) chain D can be described as follows: residues F1, V2, Q4, E21 and K29 have overall poorly defined electron density; residues V12 and V18 have clear double conformations (orange); residue T27 is mainly a clear double conformation but with some missing terminal density. The remaining residues are clearly defined in strong electron density (blue).
Overall comments on Insugen (I)
Why is chain A so well ordered while the related chain C shows a number of double conformations and poorly defined residues?
Chains B and D both show a number of double conformations. Double conformations L11, E13 and R22 occur only in chain B; the double conformation V18 occurs only in chain D; double conformations V12 and T27 occur in both chains B and D. L17, R22, F25 and T30 are disordered in chain B alone; F1, V2 and Q4 are disordered in chain D alone; E21, T27 and K29 are disordered in both chain B and D.
It may be possible to rationalise these differences for example via molecular dynamics simulations.
Implications for the biological activity
The residues most likely to affect biological activity in an adverse way are those which display conformational differences between the corresponding chains A and C, or between chains B and D, particularly with respect to the way the residues have been rated in Fig. 2.
being disordered: PheB25, and to a lesser extent GlnC5, AsnA21, LysB29, LysD29 and ThrB30;
exhibiting two clear conformations: CysC6–CysC11, LeuB11 and ValB12, and to a lesser extent LeuC16 and GluB13.
Peptide side chain electron density and conformations in Intergen (II) [PDB 3W7Y]
Properties of the electron density for Intergen (II) are summarised in Fig. 11b.
Intergen (II) chain A
The electron density of Intergen (II) chain A is of very high quality with no major problems associated with fitting the amino acid residue structures and no multiple conformations or other disorder.
Intergen (II) chain C
In contrast chain C exhibits the following characteristics: residues 1–4, 7, 8, 12, 13, 16,17,19–21, have clear well defined density; residue Q5 has mainly clear density but with missing terminal density; residues S9, I10, N18 and C6–C11 are modelled as single conformations but are probably well ordered double conformations (all shown in orange in Fig. 11b; Y14 has very poor electron density and is fitted as Ala; Q15 also has very poor density and is disordered).
Intergen (II) chain B
Chain B exhibits the following characteristics: residue F1 has mainly clear density but with missing terminal density; residues 2–10, 13–26 and 28–30, have clear well defined density; residues L11 and V12 are modelled as single conformations but are probably well ordered double conformations, whereas residue T27 is in clear well defined density and is modelled as a double conformation but has missing terminal density (all three are shown in orange in Fig. 11b).
Intergen (II) chain D
Chain D exhibits the following characteristics: residues 2,3, 5–11, 13–20, 22–28, and 30 are all well defined in clear electron density; F1 is in clear but weak density; Q4 is largely well defined but has missing terminal density; V12 is modelled in a clear single conformation but is in density that strongly suggests it is disordered in two clear conformations (orange in Fig. 11b); E21 and K29 are poorly defined with weak density that does not include all atoms in the residue chains.
Overall comments on Intergen (II)
As for the Insugen (I) structure the following points can be made for Intergen (II). Why is chain A so well ordered while chain C shows a number of double conformations and poorly defined residues? Chain B shows one double conformation. There are no double conformations in chain D.
Comparison of the Insugen (I) and Intergen (II) structures
Both A-chains have mostly well-defined electron density with very few problems in their interpretation.
For the C-chains the only notable difference here lies in the assignment of a double conformation for the C6–C11 disulphide bridge in Insugen (I). As mentioned above the electron density for Intergen (II) in this region, Fig. 16, strongly suggests that it might be possible to model a double conformation here as well.
Comparison of the B-chains of Insugen (I) and Intergen (II): differences here occur for residues L11, V12 and R22 which have double conformations in Insugen (I) and T27 which has a double conformation in Intergen (II). Insugen (I) chain B also has problem residues E13, L17, E21, F25, T27, K29 and T30, which are well behaved in Intergen (II). Intergen (II) chain B has one residue T27 modelled as a double conformation but which is single in Insugen (I).
There are a number of differences between Insugen (I) chain D and Intergen (II) chain D. In Insugen (I) F1, V2 and T27 all have problem electron density but are well behaved in Intergen (II); Q4, E21 and K29 have weak or poorly defined electron density in both structures; residues S9, V12 and V18 have double conformations in Insugen (I) but not in Intergen (II).
General comments on Insugen (I) and Intergen (II) structures
The above analysis has indicated that in both the Insugen (I) and Intergen (II) structures the sequence equivalent protein chains A and C, and B and D, respectively exhibit significant differences with respect to their corresponding amino acids such as double conformations and quality of the electron density. It is of interest to note that Baker et al.  in discussing the 1.5 Å X-ray structure of porcine insulin, report the presence of seven disordered amino acid residues: two in chain B (ArgB22 and LysB29) and five in chain D (GlnD4, ValD12, GluD21, ArgD22 and ThrD27). Of these only two amino acids in Insugen (I) ArgB22 and ValD12, have double conformations. The question of double conformations and poorly defined or absent electron density in the recombinant human insulin structures and the widespread lack of correspondence between the two raises two questions: (1) what is the origin of these differences? And (2) do they affect the therapeutic properties of these preparations? With respect to question (1) the possibilities include (a) method of preparation including folding of the recombinant amino acid-chains and (b) the forces in play when the crystal is cryo cooled prior to X-ray data collection. With respect to question (2) it is well known that differences in the form of a therapeutic insulin preparation with respect to the naturally occurring insulin can induce the production of antibodies in patients. No such indication has been noted with respect to the widespread use of either Insugen (I) or Intergen (II) but is nevertheless a possibility which should be borne in mind.
Some further selected details
Insugen (I) structure: chain C(3) S–S bridge between Sγ6–Sγ11
Figure 14b shows the geometry of part B of the disordered S–S bridge in chain C(3) for Sγ6–Sγ11B: Cα6–Cβ6 = 1.532 Å, Cβ6–Sγ6 = 1.815 Å, Sγ6–Sγ11B = 1.966 Å, Sγ11B–Cβ11B = 1.771 Å, Cβ11B–Cα11B = 1.534 Å; Cα6–Cβ6–Sγ6 = 114.16°, Cβ6–Sγ6–Sγ11B = 112.79°, Sγ6–Sγ11B–Cβ11B = 103.44°, Sγ11A–Cβ11A–Cα11A = 114.26°; torsion angle χ3 = Cβ6–Sγ6–Sγ11B–Cβ11B = −79.10° corresponds to left handed chirality . Additional file 1: Figure S7 shows views of the major and minor conformations of this S–S bridge with respect to the secondary structure of the protein.
Figure 15b shows the geometry of the ordered internal S–S bridge in Insugen (I) chain A(1) Sγ6–Sγ11: Cα6–Cβ6 = 1.532 Å, Cβ6–Sγ6 = 1.797 Å, Sγ6–Sγ11 = 2.051 Å, Sγ11–Cβ11 = 1.806 Å, Cβ11–Cα11 = 1.533 Å; Cα6–Cβ6–Sγ6 = 116.37°, Cβ6–Sγ6–Sγ11 = 103.66°, Sγ6–Sγ11–Cβ11 = 101.86°, Sγ11A–Cβ11A–Cα11A = 113.83°; torsion angle χ = Cβ6–Sγ6–Sγ11B–Cβ11B = 106.49° is close to the value of χ = 108.23° in the major conformation in chain C(3) and again corresponds to right handed chirality  the minor conformation being left handed, χ3 = −79.10°.
Conclusions on the comparison between Insugen (I) and Intergen (II) structures
Possible explanations for the observed bifurcation of chain C(3) Sγ6–Sγ11 disulphide are as follows: CysC6 is hydrogen bonded to a water molecule and there are several other waters modelled in this region which may be associated with greater conformational flexibility compared to CysA6. In addition CysA6 is in a hydrophobic pocket devoid of solvate molecules and consequently the disulphide may be more restricted by this environment. This is supported by the fact that the section of chain D close to chain A(1) Sγ6–Sγ11 disulphide is disordered (residues D1, 2 and 4) whereas the section of chain B close to chain C(3) Sγ6–Sγ11 disulphide is not, Fig. 11a and Additional file 1: Tables S3c, d. In “Molecular dynamics” the results of a molecular dynamics study of this observed order/disorder in the Sγ6–Sγ11 disulphides are presented.
Intergen (II) structure: chain C(3) S–S bridge between Sγ6–Sγ11
Inspection of the deposited X-ray structure of Intergen (II) (3W7Y), indicates that no attempt was made to model CysC6–CysC11 in chain C in a double conformation. However the superposition of the refined Insugen (I) chain C with the 3W7Y chain C indicates that the alternate conformation of the disulphide from residues CysC6–CysC11 is likely also to be present, but not modelled, in the 3W7Y structure. This is indicated by the presence of negative electron density (green) in the same position as the CysC11 γ sulphur atom in the second (minor) conformation and positive (red) electron density in the over modelled main conformation, Fig. 16.
The possibility of this effect being accounted for by radiation damage in the Insugen (I) structure was investigated by closely inspecting the intensity data collected. This led to the conclusion that there is no global suggestion of radiation damage in the data. Next a number of subsets of data were integrated and scaled and the minimum set of data with acceptable completeness was assembled by using images 1–600 (the first third of the data). When solved and initially refined there was still evidence for the second conformation at this disulphide bond. As two clear conformations, rather than complete disorder have been assigned successfully it may be concluded that this is a reflection of the true state in the crystal, rather than radiation damage. Further examination of the difference in the disulphides A6–A11 (single ordered conformation) and C6–C11 (clear ordered double conformation) may be explained by the difference in solvent exposure. C6 is less than 4 Å from the nearest solvent molecule and there are several waters modelled in that area which may give greater conformational flexibility to the region. A6 is in a hydrophobic pocket and consequently the disulphide may be more restricted by that environment. This is supported both by the fact that the section of chain D in this vicinity of the part of the molecule is also disordered (see above).
Solvated propanol in Insugen (I)
Figure 17c shows the propanol molecule in Insugen (I) lying in a binding pocket formed by rigid body movement of the first helix of chain C with respect to the structure of Baker et al.  (PDB 4INS). Interestingly, there is a similar movement of chain A in spite of there being no propanol solvate in this region. The C-terminal of chain D is also displaced towards the propanol binding pocket, while on chain B the movement is in the opposite direction.
Comments on the solvated propanol in Insugen (I)
It is interesting to note that Step 12 of US Patent Application Number US 13/032,797  describes the use of n-propanol in a process for producing improved preparations and methods for manufacturing substantially liquid preparations of RH insulin API. It is possible that the manufacture of Insugen (I) has included a similar step and this is the origin of the bound propanol revealed in the ultra-high resolution X-ray structure described here. It is possible that the presence of propanol in this insulin preparation may have consequences with respect to its biological/therapeutic characteristics . Further evidence for the assignment of propanol in this pocket of electron density was obtained by modelling in a number of different likely possibilities. Of these propanol emerged as the most likely candidate (see Additional file 1: Text S2, Figure S4).
The use of the molecular modelling procedures described in “Molecular dynamics” to investigate reasons for the presence of propanol in the binding site located on chain D of Insugen (I) described here is currently in progress.
The Zn sites in molecules 1 and 2
Insugen (I) and Intergen (II) have been synthesised to include the Zn ions present in naturally occurring insulins. The Zn ions are an essential feature in the formation of the crystal structure and are located on a crystallographic three-fold axis. In porcine insulin 2 Zn crystals , three insulin dimers are assembled around two zinc ions, 15.82 Å apart on the threefold axis. Each zinc is coordinated to three symmetry related Nε atoms of residue His10B, both at 2.05 Å, and to three water molecules at 2.36 and 2.21 Å, in molecules 1 (chains A and B) and 2 (chains C and D), respectively. During the course of the X-ray analysis of Insugen (I) the Zn sites in molecules 1 and 2 were carefully examined.
The Zn site in Insugen (I) molecule 1
The Zn site in Insugen (I) molecule 2
Figure 19c shows the electron density in the vicinity of Zn1 in Insugen (I) molecule 2. This shows Zn4100 coordinated to His 4010 Nε and two water molecules. This is the normal mode of Zn binding in insulins . Note: as for molecule 1 the complete coordination sphere around the zinc ion is generated by application of the crystallographic three-fold symmetry.
Additional file 1: Figure S8 shows the arrangement of the Zn sites in Insugen (I) with respect to peptide chains B and D and the propanol associated with chain D ThrD27.
The Zn sites in Intergen (II) molecules 1 and 2
Figure 19d shows the electron density in the vicinity of Zn501 in the Intergen (II) structure molecule 1. Zn501 is coordinated by His10B Nε as usual and a single water molecule water 617. There is no other solvate in this site. The Zn site in Intergen (II) molecule 2 is structured in the same way. Note: as previously stated the complete coordination sphere around the zinc ion is generated by application of the crystallographic three-fold symmetry.
As discussed previously in the ultra-high resolution X-ray structure of Insugen (I) in the internal disulphide bridge of chain C (CysC6–CysC11) CysC11 is disordered into two sites: A (80%) and B (20%), Figs. 13, 14a, b. However the corresponding disulphide bridge in chain A is not disordered, Fig. 15a, b. In this section molecular dynamics calculations have been employed in order to investigate and find a rational explanation for this difference.
Materials and methods
In order to prepare for the molecular dynamics (MD) simulations, two pdb files, Cys11_80percent.pdb and Cys11_20percent.pdb were generated from the original high resolution crystal structure by editing the atom records for Cys11, generating two separate pdb files, one with the chain C CysC6–CysC11 disulphide in the major (80%) conformation, the other with the minor (20%) conformation. Following this, both structures, including water molecules in the crystal structure were subjected to energy minimisation using HyperChem 8 Professional(™) . Energy minimisation was performed using the AMBER3 force field , using the Polak-Ribere conjugation gradient , with only the original contents of the crystal structure contained in a periodic box, since the object of the MD simulation was to explain the disorder in the original unit cell of the high resolution crystal structure, rather than a protein under normal solvated biological conditions.
RSMD values (Å) for HR insulin structures after the MD simulation
Molecule 1 (chain A & B)
Molecule 2 (chain C & D)
(a) Insugen (I)
(b) Intergen (II)
Intergen (II) 3W7Y
Materials and methods
The structure of Intergen (II) was also minimised using the method described above for Insugen (I), but was not split into two pdb files beforehand as no disordering was modelled for this structure, (Fig. 16). Following this, MD simulations were run at 310 K and 320 K, again using the methods described above for Insugen (I).
From the results of both the torsion angle plots and the RMSD kinetics for the Cys6–Cys11 SS bonds for Insugen (I), it is clear that both the Cys6–Cys11 internal disulphide bridges in chains A and C possess flexibility. However the flexibility of Cys6–Cys11 of chain C appears to be much greater, as both the MD simulations for the minor conformation at 310 K and the major conformation show times when the torsion angle of Cys6–Cys11 of chain C shows rapid decreases followed by rapid increases. In contrast, Cys6–Cys11 of chain A only showed one major change, in the major conformation at 310 K, at all other times staying relatively constant. The rapid changes in torsion angles shown by Cys6–Cys11 of chain C of Insugen (I) would appear to explain why it shows disorder in the original crystal structure.
Overall conclusions from the molecular dynamics study
From the results of both the torsion angle plots and the RMSD kinetics for the Cys6–Cys11 S–S bonds of Insugen (I), it is clear that both the Cys6–Cys11 internal disulphide bridges in chains A and C of possess flexibility. However the flexibility of Cys6–Cys11 of chain C appears to be much greater, as both the MD simulations for the minor conformation at 310 K and the major conformation show times when the torsion angle of Cys6–Cys11 of chain C shows rapid decreases followed by rapid increases. In contrast, Cys6–Cys11 of chain A only showed one major change, in the major conformation at 310 K, at all other times staying relatively constant. The rapid changes in torsion angles shown by Cys6–Cys11 of chain C of Insugen (I) would appear to explain why it shows disorder in the original crystal structure.
From a structural point of view, examination of the secondary structures of chains A and C of Insugen (I) in the original crystal structure may provide an explanation for this increased flexibility. The structure of these chains consists of a single loop between two α-helices. The length of the loop shows differences in chains C and A. In chain C the loop is long enough to contain both Cys residues involved in the internal disulphide bridge, whereas in chain A it is shorter and so one of the Cys residues is located on an α-helix. The longer loop of chain C would be more flexible and therefore may possibly allow for more movement of the Cys residues involved in the disulphide bond. Over the course of the MD simulation changes in secondary structure occur in Insugen (I), most noticeably in chain C, with significant portions becoming converted to coils, which may further affect flexibility.
General comments and further selected examples
The ultra-high resolution X-ray structures of two forms of human recombinant insulin, Insugen (I) and Intergen (II), has revealed several quite unexpected and previously unpredicted features. Both Insugen (I) and Intergen (II) structures exhibit structural features that can be described as: (a) highly ordered; (b) clear and resolved double conformations; (c) badly disordered. The assembled molecule comprises polypeptide chains A, B, C and D where A and C are sequence equivalent, as are B and D. It is somewhat surprising that the occurrence of structural features (a), (b) and (c) between say Insugen (I) chains A and C is by no means one to one but rather almost lacking in correspondence. This observation applies to all pairs of like polypeptide chains in both Insugen (I) and Intergen (II) and to all pairs of like polypeptide chains one from Insugen (I) and one from Intergen (II). It would be of interest (1) to find explanations for these differences and (2) to know whether they affect the therapeutic properties of these preparations? As described in “General comments and further selected examples” below why a given residue should be perfectly ordered in one structure and badly disordered in the other? With respect to question (1) the possibilities include (a) method of preparation including folding of the recombinant amino acid-chains and (b) the forces in play when the crystal is cryo cooled prior to X-ray data collection. With respect to question (2) it is well known that differences in the form of a therapeutic insulin preparation with respect to the naturally occurring insulin can induce the production of antibodies in patients. No such indication has been noted with respect to the widespread use of either Insugen (I) or Intergen (II) but is nevertheless a possibility which should be borne in mind. It may be possible to use molecular dynamics simulations further to resolve some of these considerations.
Previous studies: (i) that of Baker et al.  at room temperature and a resolution of 1.5 Å on porcine insulin and (ii) that of Smith, Pangborn and Blessing on a commercially available biosynthetic form of T6 human insulin (Lilly Research Laboratories) at 120 K and 1.0 Å resolution  have revealed significant differences in a number of the individual amino acid residue conformations between the two structures. The level of refinement achieved in these two analyses, as is also the case with the low temperature structure of Intergen (II) HR insulin, as judged by the final R values (0.153, 0.183 and 0.168, respectively) are all inferior to that achieved here with the Insugen (I) HR insulin (0.1112). Interestingly Smith et al.  list seven side-chains in the porcine room temperature structure  at 1.5 Å resolution and nine side-chains in their 1.0 Å human insulin structure as having two distinct conformations. The residues involved are: porcine GlnB4, ValB12, GluB21, ArgB22, ThrB27 and LysD29; human GlnB4, ValB12, GluB17, GluD21, GluC5, LeuC16, ValD12, ValD18 and GluD21. Only two residues are in common in this list: GlnB4 and ValB12. This result follows the trend reported here for Insugen (I) and Intergen (II) that correspondence between the two structures with respect to multiple or disordered conformations does not follow any fixed pattern. However, interestingly, reference to Fig. 11 reveals that in both Insugen (I) and Intergen (II) GlnB4 and ValB12 presented problems in the interpretation of their electron densities. Of the other residues in the above porcine list some are clear single conformations and others are problematic in either Insugen (I) or Intergen (II). Similar comments apply with respect to the above list for biosynthetic form of T6 human insulin. GlnB4 and ValB12 are the only residues in all four of these insulin structures that presented problems. According to Fig. 2 ValB12 is significantly involved in the interaction of insulin with its receptor.
PheB24 and PheB25 in Insugen (I) and Intergen (II)
The significant observation here is the extremely poor electron density for PheB25 in Insugen (I) which has been modelled as disordered with two distinct conformations, as opposed to PheB24 in Insugen (I) and PheB24 and PheB25 in Intergen (II) which are all excellent examples of strong, clearly resolved single conformation electron density. It is of interest to note that PheB25 in the X-ray structure of porcine insulin  has comparatively weak electron density corresponding to a single well defined conformation but with one missing atom on the phenyl ring (see also Footnote 1; Additional file 1: Figure S6a) and PheB24 is in completely well-defined electron density as are PheD24 and PheD25. What is probably most surprising is that while Insugen (I) PheD25 has strong electron density corresponding to a single ordered conformation, as is also the case for Intergen (II) and porcine insulin , the conformation for Insugen (I) PheD25 uniquely corresponds to that of the ordered porcine B conformation, not the porcine D conformation as displayed by the other two PheD conformations (see also Additional file 1: Text S6, Figure S6a–h). It is planned to investigate the situation with respect to PheB24 and PheB25 in Insugen (I) and Intergen (II) using molecular dynamics as described in “Molecular dynamics” for the Sγ6–Sγ11 disulphides.
Intergen (II) structure: chain C(3) S–S bridge between Sγ6–Sγ11
The intra-chain S–S bridge in chain C(3) Cys6–Cys11 has been observed in Insugen (I) to exhibit two ordered conformations. Cys6 occupies a single site while Cys11 occupies two sites with relative occupancies of 0.8 and 0.2, respectively. The geometry and all other refinement characteristics of this bifurcated cysteine bridge are of excellent quality as discussed previously. The corresponding S–S bridge in Insugen (I) chain A(1) is completely ordered which again poses a question about the origin of the distinction between the two molecular dynamics simulations. Molecular dynamics studies have provided rationale in answer to this question. In fact the difference in the disulphides A6–A11 and C6–C11 may be further explained by the difference in solvent exposure. A6 is less than 4 Å from the nearest water solvent molecule and there are several waters modelled in that area which may give greater conformational flexibility to the region. C6 is in a hydrophobic pocket and consequently the disulphide may be more restricted by that environment. This is supported both by the fact that the section of chain B near this part of the molecule is also disordered. With reference to Intergen (II) the corresponding S–S bridge in chain A(1) is also completely singular and ordered. This S–S bridge in chain C(3) as observed by inspection of PDB 3W7Y has been modelled as a single ordered conformation. However, as discussed previously, there is evidence in the electron density (Fig. 16) that this S–S bridge is actually bifurcated as in the corresponding S–S bridge in Insugen (I). S–S bridges with ordered double conformations have been previously reported. For example Cys14–Cys 38 in the ultra-high resolution (0.86 Å) low temperature, synchrotron structure of bovine pancreatic trypsin inhibitor  is very similar to Cys6–Cys11 in Insugen (I).
The ultra-high resolution Insugen (I) X-ray structure was found to include an ordered solvated propanol molecule which forms H-bonds with the prominent OG1A of ThrD27 in chain D(4) and two water molecules. The electron density for this solvate is clear and the geometry of the refined propanol is excellent. There is no solvated propanol in Insugen (I) chain B(2) or in either chain B or D in Intergen (II). These differences again offer a challenge to a rational explanation. The origin of the solvated propanol in Insugen (I) may be questioned. However it is known that propanol is a minor component used in the manufacturing process and is most likely to have been introduced into the protein at some stage of the synthesis procedure. To the best of our knowledge no other insulin structure has been shown to include structurally ordered propanol.
The Zn sites
Insugen (I) and Intergen (II) have been synthesised to include the essential Zn ions present in naturally occurring insulins. The Zn ions are an essential feature in the formation of the crystal structure and are located on a crystallographic three-fold axis.
The Zn site in Insugen (I) molecule 1
The electron density in the vicinity of Zn2 in molecule 1 revealed an unexpected feature which was shown to be a solvated acetate molecule. Zn2 is coordinated to both His10B Nε in chain B and an oxygen atom of the acetate. It is most likely that the presence of solvated acetate originated during crystallization. There are no other solvated acetate sites in either Insugen (I) or Intergen (II).
In the publication of Baker et al.  the pig insulin asymmetric unit is defined as: molecule 1 (chains A1, B1) and molecule 2 (chains A2, B2). For example residue B25.2 Phe refers to phenylalanine 25 in chain B of molecule 2. However in the PDB deposition of this structure, 4INS, molecule 1 is designated by chains A and B, and molecule 2 as chains C and D. All.pdb files referred to in the present publication follow this later format so B25.2 Phe of Baker et al. becomes PheD25. It should also be noted that to the best of our knowledge in this revised format in the deposited pig insulin 4INS.pdb, unexpectedly Baker et al’s molecule 1 corresponds to chains C + D, and molecule 2 to chains A + B. This means that in Baker et al’s Figure 12.2, for example, the left protruding Phe residue which is supposed to be B25.2 Phe (PheD25) is actually B25.1Phe (PheB25). See Supporting Information Figures S6a and S6b produced from the 4INS file for further details. A second example of this interchange of chain designations occurs in Figure 4.12 of Baker et al.  which describes ValB12.1 (ValB12) as having a single conformation and ValB12.2 (ValD12) as having two conformations. Inspection of 4INS.pdb however confirms the opposite case with ValB12 having double and ValD12 as having a single conformation. From this it seems to be safe to assume that this interchange of chain designations is consistent throughout Baker et al. Comparisons made in the present publication assume this to be so.
DRL was responsible for growing the Insugen (I) crystals used for X-ray data collection at Diamond, monitoring the structure determination and refinement, advising on the graphical analysis of the structure, surveying the progress of the manuscript preparation, checking the Figures and Tables and advising on the Supporting Information. RAP monitored the structure determination and refinement, carried out the graphical analysis of the two structures Insugen (I) and Intergen (II), surveyed the locations and involvement of water molecules, initiated and carried out the manuscript preparation including many of the Figures and References, devised the Graphical abstract and prepared most of the Supporting Information. CMCL selected and mounted the crystal for X-ray diffraction measurements, processed and assessed the data, solved and refined the initial structure and measured fluorescence spectrum to test for the presence of Zn. CEN carried out the further refinement of Insugen (I) including assignment of multiple side chain conformations, conversion to anisotropic thermal parameters for the non-H atoms, modelling H-atoms for inclusion in the refinement, detailed checking and assessment of the structure as it developed, further refinement of the deposited Intergen (II) structure, preparation of some of the Figures and Tables and deposition of the data in the PDB. BZC was responsible (together with RAP and DRL) for the inception of the project, closely followed the progress of manuscript preparation and advised on certain aspects of how to proceed, took responsibilty for the correct useage of scientific units and advised on aspects of submission of the manuscript for publishing including appropriate References, quality of the graphics, Figures and Tables in both the main text and Supporting Information. ZAI-K and AAB were responsible for supplying the original Insugen (I) material for crystallization and advised on its biochemical properties. BJH supervised the MD calculations, interpretation and presentation of the results and implications of the results for the two structures. NCJG undertook the running of the MD calculations, preparation of the graphical outputs and interpretation and implications of the results. JWS undertook the detailed analysis of features of secondary structure in Insugen (I) and advised on the preparation of their graphical images. JNL was largely responsible for characterising the features of the water structure and assisted in devising suitable ways of illustrating and presenting them, he also advised on both special and general aspects of the presentations in the manuscript and in the reading and checking of most of the sections presented. AKB was largely responsible for initiating the further refinement of Insugen (I) and played a significant part in the analysis of the residues with multiple conformations and their refinement, he also carried out a number of checks on the validity of refined structural features and their presentation. All authors read and approved the final manuscript.
The authors wish to sincerely thank the following for help and suggestions at various times: Dr. Hideaki Niwa, Prof. Thomas Sorensen, Prof. Noriyoshi Sakabe, Prof. G David Smith, Prof. Michael Weiss, Prof. Jon Cooper, Prof. Eleanor Dodson, Dr. Mark Ladd, Dr. Kevin Palmer, Andy Fry and Dr. Mark Sanderson.
This article is dedicated to Hilda Thanda Palmer.
The authors declare that they have no competing interests.
Availability of data and materials
All data are readily available (the primary X-ray diffraction data have been deposited at the Protein Data Bank as 5E7W).
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Baker EN, Blundell TL, Cutfield JF, Cutfield SM, Dodson EJ, Dodson GG et al (1998) The structure of 2Zn Pig insulin crystals at 1.5 Å resolution. Philos Trans R Soc London Ser B Biol Sci 319(1195):369–456View ArticleGoogle Scholar
- Cunha JP (2016) http://www.rxlist.com/humulin-n-side-effects-drug-center.htm
- Ciszak E, Beals JM, Frank BH, Baker JC, Carter ND, Smith GD (1995) Role of C-terminal B-chain residues in insulin assembly: the structure of hexameric LysB28ProB29-human insulin. Structure 15(3):615–622View ArticleGoogle Scholar
- Ishida H, Nagamatsu S (2001) Chemical structure, active sites and receptor binding of the insulin molecule and its signal transduction. Nippon Rinso 59(11):2109–2116 (translated from Japanese by Hideaki Niwa) Google Scholar
- Xu B, Q-x Hua, Nakagawa SH, Jia W, Chu Y-C, Katsoyannis PG, Weiss MA (2002) Chiral mutagenesis of insulin’s hidden receptor-binding surface: structure of an Allo-isoleucine A2 analogue. J Mol Biol 316:435–441View ArticleGoogle Scholar
- Pullen RA, Lindsay DG, Wood SP, Tickle IJ, Blundell TL, Wollmer A et al (1976) Receptor binding region of insulin. Nature 259:369–373View ArticleGoogle Scholar
- Conlon JM (2001) Evolution of the insulin molecule: insights into structure-activity and phylogenetic relationships. Peptides 22(7):1183–1193View ArticleGoogle Scholar
- Sakabe N (2013) Personal communicationGoogle Scholar
- Incardona M-F, Bourenkov GP, Levik K, Pieritz RA, Popov AN, Svensson O (2009) EDNA: a framework for plugin-based applications applied to X-ray experiment online data analysis. J. Synchrotron Radiat 16:872–879. doi:10.1107/S0909049509036681 View ArticleGoogle Scholar
- Sakabe N, Sakabe K, Higashi T, Igarashi N, Suzuki M, Watanabe N, Sasaki K (2001) Automatic Weissenberg data collection system for time-resolved protein crystallography. Nucl Instrum Methods Phys 468:1367View ArticleGoogle Scholar
- Watanabe N, Suzuki M, Higashi Y, Sakabe N (1999) Rotated-inclined focussing monochromator with simultaneous tuning of asymmetry factor and radius of curvature over a wide wavelength range. J Synchrotron Radiat 6:64View ArticleGoogle Scholar
- Sakabe K, Sasaki K, Watanabe N, Suzuki M, Wang G, Miyahra J, Sakabe N (1997) Large-format imaging plate and Weissenberg camera for accurate protein crystallography data collection using synchrotron radiation. J Synchrotron Radiat 4:136View ArticleGoogle Scholar
- Katayama C, Sakabe N, Sakabe K (1972) A statistical evaluation of absorption. Acta Cryst A28:293View ArticleGoogle Scholar
- Higashi T (1989) The processing of diffraction data taken on a screenless Weissenberg camera for macromolecular crystallography. J App Crystallogr 22:9View ArticleGoogle Scholar
- Kabsch W (2010) XDS. Acta Crystallogr D Biol Crystallogr 66(Pt 2):125–132View ArticleGoogle Scholar
- Evans PR (2006) Scaling and assessment of data quality. Acta Crystallogr D Biol Crystallogr 62(Pt 1):72–82View ArticleGoogle Scholar
- Vagin A, Teplyakov A (1997) MOLREP: an automated program for molecular replacement. J Appl Cryst 30:1022–1025View ArticleGoogle Scholar
- Murshudov GN, Skubák P, Lebedev AA, Pannu NS, Steiner RA, Nicholls RA, Winn MD, Long F, Vagin AA (2011) REFMAC5 for the refinement of macromolecular crystal structures. Acta Cryst D67:355–367. doi:10.1107/S0907444911001314 Google Scholar
- Emsley P, Lohkamp B, Scott WG, Cowtan K (2010) Features and development of Coot. Acta Cryst D66:486–501Google Scholar
- Afonine PV, Grosse-Kunstleve RW, Adams PD (2005) CCP4 Newsl. 42, contribution 8. PHENIXGoogle Scholar
- Sheldrick GM (2008) A short history of SHELX. Acta Cryst A64:112View ArticleGoogle Scholar
- Chen VB, Arendall WB, Headd JJ, Keedy DA, Immormino RM, Kapral GJ, Murray LW, Richardson JS, Richardson DC (2010) MolProbity: all-atom structure validation for macromolecular crystallography. Acta Cryst D66:12–21Google Scholar
- Accelerys Corporation (2012) Discovery Studio Software version 3.5. http://accelrys.com/products/datasheets/whats-new-in-discovery-studio.pdf
- Smith GD, Pangborn WA, Blessing RH (2003) The structure of T6 human insulin at 1.0 Å resolution. Acta Crystallogr Sect D 59:474–482. doi:10.1107/S0907444902023685 View ArticleGoogle Scholar
- Addlagatta A, Krzywda S, Czapinska H, Otlewski J, Jaskolski M (2001) Ultrahigh-resolution structure of a BPTI mutant. Acta Crystallogr Sect D 57:649–663View ArticleGoogle Scholar
- Panijpan B (1977) Chirality of the disulfide bond in biomolecules. J Chem Educ 54(11):670View ArticleGoogle Scholar
- Zimmerman RE, Stokell DJ, Akers MP (2012) Elona Biotechnologies Liquid insulin compositions and methods of making the same US Patent Publication Number US20120214964 A1, August 2012Google Scholar
- Owens DR, Landgraf W, Schmidt A, Bretzel RG, Kuhlmann MK (2012) The emergence of biosimilar insulin preparations–a cause for concern? Diabetes Technol Ther 14(11):989–996. doi:10.1089/dia.2012.0105 View ArticleGoogle Scholar
- HyperChem 8 Professional(™) (Hypercube Inc, Gainsville, FL, USA)Google Scholar
- Weiner PK, Kollman PA (1981) AMBER: Assisted Model Building with Energy Refinement. A general program for modeling molecules and their interactions. J Comp Chem 2:287–303View ArticleGoogle Scholar
- Polak E, Ribiere G (1969) Note on the convergence of methods of conjugate directions. Revue Francaise d’Informatique et de Recherche Operationnelle 3(16):35–43Google Scholar
- Berendsen HJC, Postma JPM, van Gunsteren WF, DiNola A, Haak JR (1984) Molecular dynamics with coupling to an external bath. J Chem Phys 81:3684View ArticleGoogle Scholar
- Hockney RW (1970) The potential calculation and some applications. Methods Comput Phys 9:136–211Google Scholar
- Ciszak E, Beals JM, Frank BH, Baker JC, Carter ND, Smith GD (1995) Role of C-terminal B-chain residues in insulin assembly: the structure of hexameric Lys B28 Pro B29-human insulin. Structure 3:615–622View ArticleGoogle Scholar
- Dassault Systèmes BIOVIA (2016) Discovery studio modeling environment, release 2017. Dassault Systèmes, San DiegoGoogle Scholar
- Guex N, Peitsch MC (1997) SWISS-MODEL and the Swiss-PdbViewer. An environment for comparative protein modelling. Electrophoresis 18:2714–2723View ArticleGoogle Scholar