Mechanism of Human Apohemoglobin Unfolding - Biochemistry (ACS

Feb 20, 2017 - *Department of BioSciences, Rice University, P.O. Box 1892, MS-140, Houston, TX 77251. E-mail: [email protected]. Telephone: (713) 348-476...
0 downloads 0 Views 5MB Size
Subscriber access provided by University of Newcastle, Australia

Article

The Mechanism of Human Apohemoglobin Unfolding Premila P. Samuel, William Christopher Ou, George N. Phillips, and John Steven Olson Biochemistry, Just Accepted Manuscript • DOI: 10.1021/acs.biochem.6b01235 • Publication Date (Web): 20 Feb 2017 Downloaded from http://pubs.acs.org on February 23, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Biochemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

The Mechanism of Human Apohemoglobin Unfolding

Premila P. Samuel 1, William C. Ou1, George N. Phillips, Jr.1,2, John S. Olson1*

1

From the Department of BioSciences and 2

Department of Chemistry

Rice University, Houston, TX 77251

*To whom correspondence should be addressed: John S. Olson, Department of BioSciences, Rice University, Post Office Box 1892, MS-140, Houston, TX, USA, Tel.: (713) 348-4762; Fax: (713) 348-5154; E-mail: [email protected]

Funding Source Statement: NIH Grants HL110900 (JSO) and GM109456 (GNP) and Grant C0612 from Robert A. Welch Foundation (JSO).

1

ACS Paragon Plus Environment

Biochemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ABBREVIATIONS: GdnHCl: guanidine hydrochloride; hemoglobin: Hb; adult human hemoglobin A: HbA; fetal human hemoglobin F: HbF; recombinant hemoglobin: rHb; Mb: myoglobin; K2,1: heterodimer equilibrium disassociation constant; K4,2: tetramer-dimer equilibrium disassociation constant; CD: circular dichroism; NMR: nuclear magnetic resonance; DTT: dithiothreitol; rHb0.0: recombinant hemoglobin with V1M mutations in both α and β globin chains; rHb0.1: recombinant hemoglobin with V1M mutations in both di-α and β globin chains and a single glycine linker joining the C-terminus of the first α chain and the N-terminus of the second α chain; MALDI-TOF: matrix assisted laser desorption/ ionization time of flight; HPLC: high performance liquid chromatography; SDS-PAGE: sodium dodecyl sulfate polyacrylamide gel electrophoresis; metHb: oxidized hemoglobin with heme iron in the Fe3+ form.

2

ACS Paragon Plus Environment

Page 2 of 52

Page 3 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

ABSTRACT Removal of heme from human hemoglobin (Hb) results in formation of an apoglobin heterodimer. Titration of this apo-dimer with GdnHCl leads to biphasic unfolding curves indicating two distinct steps. Initially the heme pocket unfolds and generates a dimeric intermediate in which ~50% of the original helicity is lost, but the α1β1 interface is still intact. At higher [GdnHCl], this intermediate dissociates into unfolded monomers. This structural interpretation was verified by comparing GdnHCl titrations for adult human hemoglobin A (HbA), recombinant fetal human hemoglobin (HbF), recombinant Hb crosslinked with a single glycine linker between the α chains (rHb0.1), and recombinant Hbs with apolar heme pocket mutations that markedly stabilize native conformations in both subunits. The first phase of apoHb unfolding is independent of protein concentration, little affected by genetic crosslinking, but significantly shifted toward higher [GdnHCl] by the stabilizing distal pocket mutations. The second phase depends on protein concentration and is shifted to higher [GdnHCl] by genetic crosslinking. This model for apoHb unfolding allowed us to quantitate subtle differences in stability between apoHbA and apoHbF, which suggest that the β and γ heme pockets have similar stabilities, whereas the α1γ1 interface is more resistant to dissociation than the α1β1 interface.

3

ACS Paragon Plus Environment

Biochemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 52

Protein folding and assembly occur through various non-covalent interactions, including secondary and tertiary structure formation, cofactor binding, and oligomerization. These interactions are critical for biological function and inhibit protein disorder in localized regions1-3. Human Hb provides an ideal model framework for studying the effects of subunit interactions and heme binding on protein folding and assembly. Approximately 90-95% of the total protein in human erythrocytes is HbA, which is composed of α and β globin chains. Each HbA subunit has 7-8 alpha helical segments with a myoglobin (Mb) like fold and heme bound in a hydrophobic pocket4, 5. The subunits assemble first into a heterodimer forming an α1β1 interface with extensive inter-subunit contacts involving 34 different amino acid side chains6. Based on past measurements of the rates of subunit association and dissociation, the estimated value for the heterodimer to monomer equilibrium dissociation constant (K2,1) is on the order of 10-11 M

7-10

,

but this value has not been measured directly. These α1β1 dimers then assemble further into a compact globular tetrameric structure with two additional α1β2 interfaces, each of which contains only 19 amino acid side chain contacts for ligand bound HbA6. The tetramer-dimer dissociation constants (K4,2) of met- and oxyhemoglobin are in the range of 10-6 M11, 12.. Despite the high structural homology with monomeric mammalian myoglobins, hemoglobin subunits are much less stable in both their apo- and hologlobin forms as monomers. There have been reports that suggest heme binds to the nascent α and β chains as they are being synthesized on ribosomes in order to promote and stabilize globin folding13, 14. However, there is no direct experimental evidence for initial heme binding to unfolded monomers before subunit dimerization. Previous studies have shown that partially folded apoHbA monomers are stabilized markedly by oligomerization to form αβ heterodimers14, 15. Similar stabilization occurs when partially folded apo-α chains bind to the alpha hemoglobin stabilizing protein (AHSP), and when

4

ACS Paragon Plus Environment

Page 5 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

apo-β chains self-assemble into homo-dimers and tetramers

16-18

. The extraction of heme from

HbA appears to result in complete dissociation into apo-α1β1 dimers with little or no trace of either monomeric or tetrameric species at neutral pH14,

19, 20

. Because free apoglobin

heterodimers were found in erythrocytes, Winterhalter et al.21 suggested that apo-heterodimers are likely to be the initial post-ribosomal species during the Hb folding and assembly pathway20. Mammalian apoMbs show an ~20-30% loss of helical content compared to their respective holoproteins 22, and, in our CD spectroscopy measurements, human apoHb shows a similar ~20% loss of helix content compared to holoHb. Eliezer and Wright23 characterized sperm whale apoMb through high resolution NMR methods and observed increased disorder that is localized around the proximal side of the heme pocket at the EF loop, F helix, FG loop, and the beginning of G helix. By analogy, we assume that the enhanced disorder in human apoHb also occurs mainly around the F helix region and the FG loop, both regions that are critical for α1β2 interface interactions. Previous studies on ligand binding and cooperativity established the critical role of the proximal histidine and F helix in the quaternary changes that occur at the α1β2 interface24, 25. Thus, in both Mb and Hb, the binding of heme and its coordination with the proximal histidine (F8) "rescues" these disordered regions, and, in the case of hemoglobin, leads to formation of the α1β2 interface by interactions near the FG loops and F and C helices. Past studies of human holoHb unfolding and denaturation were not able to discriminate in a quantitative manner between heme affinity, apoglobin stability, and equilibrium subunit association constants, nor did they identify any molten globule intermediates26-35. For example, thermal denaturation curves for holoHb were usually fitted to 2-state models (folded and unfolded, and sometimes an irreversible aggregated state), and the partially unfolded transition states in between were not clearly defined26,

5

29-31

. Almost all of these studies measured

ACS Paragon Plus Environment

Biochemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

irreversible unfolding in the absence of reducing agents that inhibit disulfide formation. In the case of holoHb, irreversible hemichromes crosslinked by disulfide bonds can also occur in the absence of reducing agents32, 33, 36, 37. Thus, oxidative processes can trap Hbs in non-native folding states that vary with sample preparation and eventually lead to globin denaturation. Our long-term goal is to define the parameters involved in human hemoglobin expression, folding, and assembly. Unfolding mechanisms for both apo- and holoHbA have to be established and analyzed quantitatively. A 6-state unfolding mechanism for holoMb has been derived, is based on the 3-state apoMb unfolding scheme, and has been used to estimate heme affinity constants for the native, partially unfolded, and unfolded states of apoMb22, 38-40. This approach has also been used successfully to examine the structural features that regulate holoMb expression in vivo and in vitro22, 41. We are in the process of doing a similar analysis for HbA folding and assembly, which have the added complexity of hetero-dimer and tetramer formation. In our initial work, we have focused on the mechanism of human apoHb unfolding. In this study, we have (1) measured GdnHCl-induced unfolding curves for apohemoglobins using CD signals to measure loss of secondary structure; (2) established a general mechanism for unfolding of the apo-heterodimer; (3) verified our structural interpretations using a series of recombinant apoglobin variants; and (4) used this approach to examine differences in stability between human adult (HbA) and fetal (HbF) hemoglobin. HbF is composed of α and γ chains. Previous studies with both holoHbA and holoHbF suggests that the α1γ2 tetramer interface is stronger than the corresponding α1β2 interface based on the observation that the equilibrium constant for tetramer to dimer dissociation (K4,2) of HbF is significantly smaller than that for HbA in the liganded forms42. However, the relative strengths of the α1γ1 versus α1β1 dimer interfaces are more controversial due to the difficulty of measurements and interpretations8, 42-44.

6

ACS Paragon Plus Environment

Page 6 of 52

Page 7 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

In this work, we have attempted to make a direct comparison of strengths of these dimer interfaces during unfolding of apoHbA and apoHbF.

EXPERIMENTAL PROCEDURES Expression and purification of holo- HbA and rHb samples - Native HbA was isolated by lysis of human erythrocytes obtained from expired blood units from the Gulf Coast Regional Blood Bank, Houston, TX. Recombinant HbF was expressed in E. coli JM109 cells (Promega) from the pHE9 plasmid45 according to protocols developed by Shen et al.45,

46

.

αH58L/V62F and

βH63L/V67F apolar heme pocket mutations were introduced by site-directed mutagenesis of human α and β genes and the mutants were then expressed in E. coli JM109 cells (Promega) using the pHE2 vector

45, 46

. The pHE2 and pHE9 plasmids were gifts from Dr. Chien Ho’s

laboratory (Carnegie Mellon University)45, 46. The globins expressed from the pHE2 and pHE9 plasmids have an "extra" Met residue initially at their N-terminus. This initiator Met is later cleaved from the globin chains by E. coli Met aminopeptidase, which is co-expressed with the globins from the same plasmid45, 46. rHb0.1 was expressed from the pSGE1.1-E4 expression plasmid, whereas rHb0.0 was expressed from the pDL111-13e expression plasmid. Both of these plasmids were gifts from Somatogen Inc. (later Baxter Hemoglobin Therapeutics) which also provided the SGE1661 E. coli cells for rHb 0.0 and rHb 0.1 expression47-49. rHb0.0 and rHb0.1 were engineered with V1M mutations as the initiator Met for expression in E. coli. rHb0.1 was genetically crosslinked with a glycine linker between Arg-141 on the C-terminus of the first α chain (with the V1M mutation) and the Val-1 N-terminus of the second alpha chain

47-49

. Cells transformed for expression of

rHb0.0, rHb0.1, and rHbs with heme pocket mutants were grown in a Biostat C20 bioreactor (B

7

ACS Paragon Plus Environment

Biochemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Braun Biotech International, Melsungen, Germany), using cell growth and protein expression protocols described initially by Looker et al.49 and later modified by Maillett 50. Purification of the various rHbs and native HbA was done following the protocols of Maillett50, Looker et al.49, and Shen et al.45, 46. MALDI-TOF mass spectroscopy and reverse phase HPLC analyses confirmed both the identity and purity of the Hb variants45, 46. An example of the quality of the heme pocket rHb mutants and rHb0.1 samples is shown in the reverse-phase chromatographs in Figure 1. The increase in hydrophobicity of the heme pocket α and β mutants and the increase in the molecular weight of the di-α subunit of rHb0.1 are apparent from the shifts to longer elution times for the individual subunit peaks in the α(H58L/V62F), β(H63L/V67F), and di-α samples in the reverse phase HPLC chromatograms shown in Figure 1.

Preparation of apoglobin samples - Apoglobin was prepared from ferric samples by heme extraction into 2-butanone at low pH as described by Ascoli et al.51. The apoglobin remained in the aqueous phase, which was then dialyzed overnight in cold 10 mM potassium phosphate pH 7 buffer. The apoprotein concentration was determined using ε280 = 12.7 cm-1 mM-1 per subunit for apoHbA and the other apo-rHb variants51. ApoHbF concentration was calculated using a higher extinction coefficient, ε280 = 15.2 cm-1 mM-1 per subunit, due to an additional Trp residue in the γ subunits compared to β chains. This value was approximated using the formula described by Gill and von Hippel52.

Optimization of temperature and anion concentrations for GdnHCl unfolding titrations GdnHCl was chosen as the denaturant because it prevents precipitation of dissociated hemin22, allowing titrations of both holo- and apoHbA for direct comparisons in future studies. Removal

8

ACS Paragon Plus Environment

Page 8 of 52

Page 9 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

of heme from human HbA results in an ~20% loss of secondary structure, and the resultant apoHb is unstable at room temperature and readily precipitates at temperatures above 10° C14, 53. The instability of apoHb is the major cause of scatter for the initial unfolding data points because irreversible denaturation can occur before the [GdnHCl] is high enough to solubilize unfolded states54. High anion concentrations have an overall stabilizing effect on apoglobins, especially for the molten globule intermediates55. Therefore, the most reproducible GdnHCl induced unfolding curves for apoHb were obtained at 10°C in 200 mM potassium phosphate, pH 7 buffer. Stock solutions of ~6 M GdnHCl were prepared in 200 mM potassium phosphate, pH 7, taking care to re-adjust the pH after addition of the denaturant. [GdnHCl] was varied from 0 to 4.8 M. The individual apoHb/GdnHCl mixtures were incubated in a water bath at 10°C for 1 hour to ensure equilibrium22, 38. Spectral changes upon addition of apoHb to the denaturant solution were complete in a few minutes and reversed rapidly when the apoHb samples in concentrated GdnHCl were diluted into buffer. CD spectra were recorded for each sample using a Jasco J-810 CD spectropolarimeter, and unfolding was monitored as a decrease in negative ellipticity at 222 nm (peak for α helical secondary structure). Unlike apoMb, apoHbA does not show a bell-shaped increase and decrease in Trp fluorescence during GdnHCl-induced unfolding22, 39, 41, 55. Instead, only very small changes in fluorescence are observed and difficult to analyze quantitatively.

Analytical gel filtration analysis - Our initial apoHbA unfolding experiments showed significant variability, particularly for the second phase of unfolding at higher protein concentrations. We were concerned that irreversible, non-native oligomers were present due to disulfide formation, which would be promoted at higher protein concentrations. HbA and HbF have sulfhydryl

9

ACS Paragon Plus Environment

Biochemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 52

groups at β Cys93 and γ Cys93, respectively, which are located at the α1β2 and α1γ2 interfaces, and at α Cys104 and β Cys112, which are located at the α1β1 and α1γ1 dimer interfaces. Because apoHbA exists predominantly as an α1β1 dimer

11, 14, 19, 20

, the β and γ Cys93 thiol groups are

exposed and are able to form disulfides even in the native, folded apodimer state. The other thiol side chains are exposed during dissociation of the dimer interface into monomers, which can lead to even more di-sulfide crossed linked states at high GdnHCl. To examine this problem, apoHb samples were loaded onto a 24 ml Superose-12 HR 10/30 GL column that was equilibrated with 200 mM potassium phosphate, pH 7 at 4°C. The final concentration of eluted protein was calculated using a dilution factor computed as the ratio of the average elution peak width at the half-height to the volume of the sample that was loaded on the column56. The value of K4,2 has been reported to be ~10 µM for metHbA11. Concentrated metHbA (final concentration = 82 µM) eluted from the gel filtration column at 13.67 ml, representing primarily a tetrameric species (MW = 64.6 kD; Figure 2A). For dilute metHbA at a final elution concentration of 0.52 µM, the elution peak shifted to 14.17 ml due to the presence of > 85% dimers. Monomeric apomyoglobin (MW=17.3 kD) eluted at ~14.93 ml, irrespective of protein concentration. These positions serve as approximate standard elution volumes for tetramers, dimers, and monomers, respectively. When our initial apoHbA samples were loaded on the column at different concentrations, there were always two prominent elution peaks, one at ~13.23 ml and another at ~14.44 ml (Figure 2B). The 14.44 ml elution peak was always the major peak and represents a dimer species (apo-α1β1 MW = 31 kD). The smaller and more variable 13.23 ml elution peak appears to represent a tetrameric species. Interestingly, these 2 peaks for apoHbA were completely distinct. If the dimers and tetramers had been in equilibrium, a single broad elution peak would have

10

ACS Paragon Plus Environment

Page 11 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

appeared somewhere between the 13.23 ml and 14.44 ml and been dependent on the initial apoHb concentration42, 57. When apo-Hb0.1 was loaded onto the gel filtration column, the major elution peak appeared at ~13.50 ml (Figure 2C), demonstrating that this apoprotein was a tetramer. However, a significant amount of this recombinant protein eluted with a peak at ~12.43 ml (Figure 2C), which probably represents disulfide crosslinked tetramers. The apoHbA stock sample for Figure 2B was incubated with 50 mM TCEP (tris(2carboxyethyl)phosphine), a disulfide reducing agent, at 10 °C for 30 minutes and then loaded onto the gel filtration column. In this case, the higher molecular weight peaks at ~13.23 ml for apoHbA became significantly diminished and the smaller peaks near the void volume disappeared (Figure 2D). Similarly, after apoHbF was treated with reducing agent, the major elution peak represented a dimer. These results indicated that we needed to include a reducing agent during apoHb sample preparations and the unfolding titration experiments. Dithiothreitol (DTT) was chosen because TCEP is unstable in phosphate buffers58, and at 1 mM concentration, DTT does not interfere with the far UV CD signals59. Immediately after heme extraction into 2-butanone, all freshly isolated apoHb samples were dialyzed in 10 mM potassium phosphate buffer, 1mM DTT, pH 7 buffer overnight at 4°C. When these DTT-treated samples were analyzed on the gel filtration column, no elution peaks characteristic of non-native, higher-order oligomeric species were observed (see Figures 2E, 2F for apoHbA and apo-rHb0.1). Then all apoHb samples were buffer exchanged into pH 7 200 mM potassium phosphate buffer containing 1mM DTT. The final apoHb unfolding conditions for all of the variants examined (Figures 4, 6, 7, 8, 10, S1, and S2) were 200 mM potassium phosphate, 1mM DTT at pH 7, 10° C. All the DTT containing solutions were freshly prepared before use to prevent oxidation and generation of radical oxygen species prior to use.

11

ACS Paragon Plus Environment

Biochemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 52

RESULTS Quantitative analyses of apoHbA unfolding - The key mechanistic issue for human apoHb unfolding is whether or not a molten globule forms, and, if it does form, whether it occurs before or after dissociation of human apoHb dimers into monomers. There are three basic models for unfolding of the apoHbA heterodimer: (1) partial unfolding of the dimer (D) to generate a hetero-dimeric intermediate folding state (ID) followed by a second phase involving ID dissociation into 2 unfolded monomers, with each unfolded monomer (UM) representing either the α or β subunit (Figure 3A); (2) dissociation of D into 2 partially unfolded, intermediate monomers (2IM) followed by a second phase representing complete unfolding of each IM to UM (Figure 3B); and (3) dissociation of the D into 2UM in a single bimolecular dissociation process (Figure 3C). The simulations in Figure 3 were carried out as described by Culbertson and Olson22, 38 to show how protein concentration should affect GdnHCl-induced unfolding curves for each model. The simulations used expressions similar to those described below in eqs. 1-4.22, 38, 60

In order to distinguish between these mechanisms, we examined the unfolding of apoHbA at

protein concentrations ranging from 1 µM to 140 µM. All our unfolding experiments with apoHbA show two major phases, with only the second process showing a dependence on protein concentration (Figure 4). These results are consistent with the two-step Model 1 shown in Figure 3A. The apoHbA dimer (D) partially unfolds into a dimeric molten globule (ID), which is followed by dissociation into 2 unfolded monomers (2UM). The CD signals were normalized to the total α helical content of the folded apoHbA dimer and then analyzed by computing the fractions (Y) of D, ID, and UM populations in

12

ACS Paragon Plus Environment

Page 13 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

terms of the equilibrium folding constants for the first two steps in Figure 5, using eqs 1-4 described below. The equilibrium constants at a given [GdnHCl] are defined in terms of folding starting from UM, and the m values define the dependence of the folding free energy on [GdnHCl] and absolute temperature (T)61. KI

D ,D

K 2U

=

M ,I D

⎛ −m I ,D [GdnHCl] ⎞ [D] D = K 0I ,D exp ⎜ ⎟ D RT [I D ] ⎝ ⎠ =

⎛ −m 2U ,I [GdnHCl] ⎞ [I D ] 0 M D = K exp ⎜ ⎟ 2U M ,I D RT [U M ]2 ⎝ ⎠

(1)

The total monomeric protein concentration, P0 is defined by a quadratic expression in [UM] (eq 2). At each data point, [UM] can be computed as the positive root of eq 2, defined by P0 and the equilibrium constants at a given [GdnHCl] (eq 3). P0 = 2[D] + 2[I D ] + [U M ] [I D ] = K 2U

M ,I D

[U M ]2 ; [D] = K I

P0 = 2K I

D ,D

K 2U

M ,I D

0 = 2 KI

D ,D

K 2U

M ,I D

(

[U M ] =

[U M ]2 + 2K 2U + K 2U

M ,I D

(

D ,D

−1+ 1+ 8P0 K I

(

D ,D

4 KI

D ,D

K 2U

)[U

K 2U

M ,I D

K 2U

M ,I D

M

M ,I D

[U M ]2

[U M ]2 + [U M ]

]2 + [U M ] − P0

M ,I D

+ K 2U

+ K 2U

M ,I D

)

M ,I D

(2)

) (3)

The total CD signal S[GdnHCl] is then computed from the fractions of each folding species (Y) multiplied by the corresponding intrinsic CD signal of each species (i.e. SD ≈1.0; SID ≈ 0.5; and SUM ≈ 0) by inserting eq 3 into eq 4.

13

ACS Paragon Plus Environment

Biochemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

YD =

2[D] 2[D] + 2[I D ] + [U M ]

YI =

2[I D ] 2[D] + 2[I D ] + [U M ]

YU =

[U M ] 2[D] + 2[I D ] + [U M ]

D

2

Page 14 of 52

S[GdnHCl] = SD YD + SI YI + SU YU S[GdnHCl] =

(

D

2 SD K I

(

D

D

,D K 2U M ,I D

2 KI

D

M

M

+ SI K 2U

,D K 2U M ,I D

D

+ K 2U

M

M

,I D

,I D

)[U

)[U

M ] + SU M

M]+1

(4)

Algorithms for fitting observed data to eqs 1-4 were written in Gnuplot 5.0462. As shown in Figure 4, fitting to the simple 2-step model provides a preliminary description of the observed unfolding curves and an interpretation of the two major processes (a complete description of the fitting statistics is given in Supporting Information and Table S1). However, a third unfolding phase is observed experimentally and represents loss of the last 10-15% of the CD signal change (Figure 4). This third phase does not appear to depend on protein concentration and requires additional steps in the apoHbA unfolding mechanism. The observed data also show significantly less dependence on protein concentration for the second unfolding phase than predicted by the 2step model (see inset to Figure 4). These observations led us to propose additional steps that specify a third distinct phase and lead to less protein concentration dependence for the second phase. The final model is presented in Figure 5 and includes 5 states and 4-steps. The first two steps still involve partial unfolding of the dimer (D→ID) followed by dissociation into unfolded monomers (ID→2UM). The third phase in the observed data implies that the initial unfolded monomer, UM, still retains a small fraction (~10%) of α-helicity relative to the native state D. Further addition of GdnHCl leads to complete unfolding of UM into a polypeptide chain, UC, with no secondary structure.

14

ACS Paragon Plus Environment

Page 15 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

However, the addition of a third monomeric unfolding step (UM→UC) alone cannot explain the smaller than expected dependence on protein concentration for the second phase, which appears to reach a limit at ~60 µM P0 in the observed data (Figures 4 and S1). In order to reduce the dependence on P0, we added a fourth step, 2UM→U2, representing transient non-specific association of unfolded monomers with residual helicity to form a mostly unfolded dimeric species (U2 in Figure 5). Transient formation of the U2 species will facilitate unfolding at higher protein concentrations, partly compensating for the inhibition of the ID→2UM transition as P0 is

increased. However, as [GdnHCl] increases, the weak non-specific interactions in the U2 state will be disrupted and lead to dissociation back to monomers, which then completely unfold in the final UM→UC step. The equilibrium constants for these two new steps are defined in eq 5. KU KU

C ,U M

2 ,2U M

=

⎛ −m U ,U [GdnHCl] ⎞ [U M ] C M = K 0U ,U exp ⎜ ⎟ C M [U C ] RT ⎝ ⎠

=

⎛ −m U ,2U [GdnHCl] ⎞ [U M ]2 2 M = K 0U ,2U exp ⎜ ⎟ 2 M RT [U 2 ] ⎝ ⎠

(5)

The equilibrium constant KU2,2UM is defined for the dissociation of U2 back into 2 UM species unlike K2UM,ID, which is defined for association of monomers to form the dimer intermediate. Eq 2 was modified in order to include the additional U2 and UC folding species in the quadratic expression defining P0 (eq 6). P0 = 2[D] + 2[I D ] + [U M ] + 2[U 2 ] + [U C ] ⎛ ⎞ ⎛ 1 1 ⎞ 2 0 = 2 ⎜ K I ,D K 2U ,I + K 2U ,I + ⎟ [U ] − P0 ⎟ [U M ] + ⎜ 1+ M D M D ⎜⎝ D ⎜⎝ K U ,U ⎟⎠ M K U ,2U ⎟⎠ 2 M C M

[UM] was then derived (eq 7) as the root of this quadratic expression (eq 6).

15

ACS Paragon Plus Environment

(6)

Biochemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

[U M ] =

⎛ 1 ⎞ − ⎜ 1+ ⎟+ ⎜⎝ K U ,U ⎟⎠ C M

Page 16 of 52

2

⎛ ⎛ ⎞ 1 1 ⎞ ⎜ 1+ ⎟ + 8P0 ⎜ K I ,D K 2U ,I + K 2U ,I + ⎟ M D M D ⎜⎝ D ⎜⎝ K U ,U ⎟⎠ K U ,2U ⎟⎠ C M 2 M ⎛ ⎞ 1 4 ⎜ K I ,D K 2U ,I + K 2U ,I + ⎟ M D M D ⎜⎝ D K U ,2U ⎟⎠ 2 M

(7)

The population fractions of U2 and UC (eq 8) were then inserted into eq 4 to obtain a final expression for the CD signal (eq 9).

YU = C

YU

2

[U C ] 2[D] + 2[I D ] + [U M ] + 2[U 2 ] + [U C ]

2[U 2 ] = 2[D] + 2[I D ] + [U M ] + 2[U 2 ] + [U C ]

S[GdnHCl] = SD YD + SI YI + SU YU + SU YU + SU YU D

D

M

M

2

2

C

(8)

C

⎛ ⎞ 1 1 2 ⎜ SD K I ,D K 2U ,I + SI K 2U ,I + SU ⎟ [U M ] + SU + SU D M D D M D 2 M C ⎜⎝ K U ,2U ⎟⎠ K U ,U 2 M C M S[GdnHCl] = ⎛ ⎞ 1 1 2 ⎜ K I ,D K 2U ,I + K 2U ,I + ⎟ [U ] + 1+ M D M D ⎜⎝ D K U ,2U ⎟⎠ M K U ,U 2 M C M

(9)

As shown in Figure 6, the observed equilibrium GdnHCl titration data for apoHbA fit well to eq 9, which was derived from the 5 state, 4-step model shown in Figure 5. Both reduced χ2 values63 and small sample Akaike's information criteria64 (AICc and ΔAICc values) indicate that the 4step model is required for an accurate representation of the observed data (see Supporting Information and Table S1). The final fitted parameters for HbA are given in Table 1. All of the equilibrium constants in this table represent the K0 values defined in eqs 1 and 5 (K extrapolated to [GdnHCl]=0). The superscript 0 in eqs 1 and 5 is implied when these K values are discussed in the text. The fitted curves in Figure 6 support the idea that the dampened protein concentration dependence for the second unfolding phase is due to transient formation of the U2 species and

16

ACS Paragon Plus Environment

Page 17 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

that the final third phase represents loss of residual secondary structure in monomers to form polypeptide chains. We did explore 3-step models in which either the UM→UC or UM→U2 step was included but not both. Descriptions of the fitting statistics for these models are presented in Supporting Information (Table S1) and help justify why both processes are needed to explain the last two phases of apoHb unfolding. Including just the UM→UC step for the third unfolding phase still misrepresents the protein concentration dependence of the second phase (see Figure S1). Including just the UM→U2 step, where U2 has lost all secondary structure, results in too small of a dependence on protein concentration for the second phase and also misrepresents the third unfolding phase at low protein concentrations (Figure S2). However, regardless of which model is used, the main conclusions that the first unfolding phase involves formation of a molten globule dimer intermediate and that the second phase involves dissociation into unfolded monomers are valid. The data points in Figure 6 for the apoHbA concentrations 1.9, 12, and 61 µM represent the average of triplicate titrations and the data points for P0 = 108 µM and 140 µM represent single titrations because of the large amount hemoglobin required for these concentrations. The same array of [GdnHCl] values was used for each run. The error bars represent the standard deviations for the titrations at 1.9, 12, and 61 µM. The K parameters and all the m values were determined from global fitting of the overall data points shown in Figure 6 across varying P0 and [GdnHCl]. The values for m2UM,ID, mID,D, mU2,2UM, and mUC,2UM for the 5-state model were determined to be 11.55 kJ mol-1M-1, 16.22 kJ mol-1M-1 , -3.91 kJ mol-1M-1, and 5.39 kJ mol-1M-1, respectively, while KUC,UM is 1190 and KU2,2UM is 5.55 × 10-7 M (Table 1). The negative m value for the dissociation of the U2 dimers describes the inhibitory effect of the denaturant on the

17

ACS Paragon Plus Environment

Biochemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 52

formation of unfolded dimer aggregates. S, K2UM,ID, and KID,D values obtained from global fitting to the data point averages were used as initial values for fitting globally individual sets of single titration curves across the 3 different apoprotein concentrations used for the triplicate measurements. The values for KUC,UM, KU2,2UM, m2UM,ID, mID,D, mU2,2UM, and mUC,2UM were fixed from the previous analysis of the data point averages. By fitting the individual sets of titration curves, we estimated the experimental errors for KID,D and K2UM,ID, which were 164 ± 3 and 6.6 ± 0.1 x 108 M-1, respectively (Table 1). For the S parameters, the following values and errors were estimated: SD =1.003 ± 0.002; SID =0.430 ± 0.002; SUM = 0.089±0.0001; SUC = 0.0034 ± 0. 0002; and SU2 =0.021 ±0.001 (Table 1).

Heme pocket mutations - Previous work on reversible unfolding of apomyoglobin has shown that introducing large apolar amino acids in the heme pocket increases apomyoglobin stability, partially compensating for the loss of heme22, 23, 39, 41. NMR characterization of the intermediate folding state of H64F apoMb showed that replacement of the highly polar E7 imidazole side chain with an apolar benzyl group leads to additional stabilization of the E helix65. In contrast to apoMb, there have been no structural characterizations published for native human apoHb dimers (D) or intermediate folding structures (ID). Building on the work with apoMb41, 66, we introduced large apolar and aromatic amino acids at the E7 and E11 helical positions in both the α and β subunits by constructing recombinant hemoglobins with His(E7)→Leu and Val(E11)→Phe

replacements:

α(H58L/V62F)β(wt),

α(wt)β(H63L/V67F),

and

α(H58L/V62F)β(H63L/V67F). These mutations did not alter apoHb quaternary structure based on analytical gel filtration (i.e. the α(H58L/V62F)β(H63L/V67F) aporHb, eluted at 14.43 ml at a final concentration of 60 µM, corresponding to a dimer elution peak in Figure 2).

18

ACS Paragon Plus Environment

Page 19 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

Triplicate GdnHCl unfolding titrations for α(H58L/V62F)β(wt), α(wt)β(H63L/V67F), and α(H58L/V62F)β(H63L/V67F) apo-rHbs and apoHbA at P0 = 12µM are shown in Figure 7. There is a dramatic right shift of the first phase of the unfolding curve towards much higher [GdnHCl] as the heme pocket is made more apolar. These results suggest that the first phase involves unfolding of the heme pockets to generate dimeric molten globule forms. These titration curves were analyzed in terms of the expanded apoHbA unfolding mechanism described in Figure 5 and eq 9. All the m, KUC,UM, and KU2,2UM parameters were set to the values determined from the global fittings of apoHbA unfolding data (Table 1). The K2UM,ID, KID,D, and the various folding state signal (S) values were allowed to vary during the curve fittings. The S values remained very similar to those for apoHbA (Table 1). The S, K2UM,ID, and KID,D values obtained from fitting to the single curve for the data point averages (Figure 7) were used as initial values for fitting each of the individual titration curves in order to estimate experimentally the errors in the fitted values of these key parameters (standard deviations in Table 1). KID,D for α(H58L/V62F)β(H63L/V67F) apo-rHb is 14,900 ± 2,300, which is ~90 fold larger than the value for refolding of the heme pockets in apoHbA (Table 1). Clearly, the apolar mutations in the E helix markedly increase resistance to unfolding during the first phase to form the ID state. The KID,D values for α(H58L/V62F)β(wt) and α(wt)β(H63L/V67F) aporHbs are 5,500 ± 150 and 590 ± 110 respectively (Table 1), showing that larger stabilizing effects occur in the α subunits (Table 1). The results suggest the increase in free energy of heme pocket folding for the individual mutant subunits is roughly additive for the quadruple mutant dimer, i.e. KID,D(quadruple)/KID,D(native) ≈

(KID,D(α-mutant)/KID,D(native)U(KID,D(β-mutant)/KID,D(native))).

Putting these heme pocket mutations in both subunits together results in a dramatic shift of the 19

ACS Paragon Plus Environment

Biochemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 52

unfolding curve during the first phase to very high values of [GdnHCl], which begins to obscure the second major phase involving dimer dissociation (Figure 7). Because these mutations along the E helix are not at the α1β1 interface, we expected that the fitted value of heterodimer association constant, K2UM,ID, would not be affected. The K2UM,ID

values for both apoHb α(H58L/V62F)β(wt) and α(wt)β(H63L/V7F) aporHbs are 6.5 ± 0.3 × 108 M-1 and 8.6 ± 0.2 × 108 M-1 respectively and are almost identical to value for native apoHbA, 6.6 ± 0.3 × 108M-1. In comparison, the K2UM,ID for α(H58L/V62F)β(H63L/V67F) aporHb was ~3 fold higher, 19.3±4.6 × 108 M-1, suggesting that stabilizing the heme pockets of both subunits together has a small synergistic effect on promoting the association of unfolded α and β monomers to form the dimeric intermediate, ID (Table 1).

Unfolding of crosslinked rHb0.1 - Crystal structures of holo-rHb0.1 have shown that the addition of a glycine linker between the two α subunits does not have a significant effect on the tertiary structures of the hemoglobin subunits nor on the interactions at both the α1β1 and α1β2 interfaces67, 68. However, the glycine crosslink causes apo-rHb0.1 to remain a tetramer even after heme removal (Figures 2C, 2F). Apo-rHb0.1 is really a trimer composed of 3 subunits (2 β chains and 1 di-α chain) and two α1β1 interfaces, whereas apoHbA contains only 2 subunits and one α1β1 interface. Thus, more free energy (i.e. more GdnHCl) should be required to dissociate rHb0.1 into its three monomeric units than for native apoHbA to dissociate into two units. As shown in Figure 8A, the effect of crosslinking is quite dramatic. Unfolding of apo-Hb0.1 still occurs in two major phases, with the first phase being very similar to that observed for apoHbA and apo-rHb0.0, the wild-type rHb control in which the subunits have V1M mutations. The most notable effect of crosslinking occurs for the second phase, which is significantly shifted to the

20

ACS Paragon Plus Environment

Page 21 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

right towards higher [GdnHCl]. Thus, the molten intermediate for the crosslinked apoglobin is much more resistant to dissociation (Figure 8A). Regardless of the detailed analyses described below, the right-shift of the second unfolding phase due to crosslinking verifies our assumption that for human apoHb unfolding, the intermediate state represents an oligomeric molten globule, which then dissociates into monomeric units. The unfolding curve for α(H58L/V62F)β(wt) apo-rHb was included in Figure 8A to emphasize the effect of stabilizing the heme pocket and, together with the apo-rHb0.1 curve, provides the experimental verification of our model for the first two steps in human apoHb unfolding (Figure 5). The unfolding data, obtained in triplicate measurements at P0=12 µM, for dimeric aporHb0.0 were fitted (Figure 8A) to the same model used for native apoHbA and the distal pocket mutants (i.e. Figure 5, and eq 9). Again, the m, KUC,UM, and KU2,2UM parameters were fixed to values obtained for apoHbA unfolding data, while all other parameters were allowed to vary. The V1M mutations cause little effect on the overall stability of apoHb (i.e. at most a 2-fold increase in KID,D). The KID,D and K2UM,ID values for apo-rHb0.0 were 319 ± 68 and 7.99 ± 0.84 × 108 M-1 respectively, and the S values were also very similar to same signal parameters for HbA (Table 1). Analysis of the unfolding of trimeric apo-rHb0.1 required modification of the scheme shown in Figure 5 for apoHb dimers. In the modified model shown in Figure 9, the apo-rHb0.1 trimer (T) undergoes unfolding via its heme pockets to form a trimeric molten globule (IT). As described in the INTRODUCTION, the α1β2 interfaces in apoHb are disrupted due to unfolding of the F helix following heme removal. Thus, we predict that apo-Hb0.1 exists as two dimers held together by the glycine linker, i.e. α1β1-α2β2. During the 2nd phase of unfolding, IT

21

ACS Paragon Plus Environment

Biochemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 52

dissociates into three mostly unfolded monomers (3UM), i.e. 1 di-α monomer (UM(di-α)) and 2 β monomers (2UM(β)). During the 3rd phase, the monomers either non-specifically associate transiently as a trimer (U3) or each monomer unfolds completely into a peptide chain (UC) that retains no secondary structure. As in the apoHb dimer model (Figure 5), the formation of the U3 state occurs by non-specific association and is disrupted at high [GdnHCl]. In order derive a relatively simple expression for the unfolding curve of apo-rHb0.1, the di-α and β monomers are considered to be equivalent unfolded species. The equilibrium constants for folding into apo-rHb0.1 trimers are then defined as: KI

T ,T

K 3U KU KU

=

M ,I T

C ,U M

3 ,3U M

⎛ −m I ,T [GdnHCl] ⎞ [T] T = K 0I ,T exp ⎜ ⎟ T RT [I T ] ⎝ ⎠ =

⎛ −m 3U ,I [GdnHCl] ⎞ [I T ] 0 M T = K exp ⎜ ⎟ 3U M ,I T RT [U M ]3 ⎝ ⎠

=

⎛ −m U ,U [GdnHCl] ⎞ [U M ] C M = K 0U ,U exp ⎜ ⎟ C M [U C ] RT ⎝ ⎠

⎛ −m U ,3U [GdnHCl] ⎞ [U M ]3 3 M = = K 0U ,3U exp ⎜ ⎟ 3 M RT [U 3 ] ⎝ ⎠

(10)

P0 at each [GdnHCl] is defined in terms of the equilibrium constants and [UM] as shown in eq 11.

P0 = 3[T] + 3[I T ] + 3[U 3 ] + [U C ] + [U M ] 0 = 3(K I ,T K 3U T

M ,I T

+ K 3U

M ,I T

+

1 K 3U

M ,I T

)[U M ]3 + (1+

1 KU

C ,U M

)[U M ] − P0

(11)

The absolute value of [UM] is computed from the cubic root of this expression (eq 11) using a Newton-Rhapson algorithm written in MATLAB R2016a (The MathWorks®, Inc., Natick, MA). The value of [UM] at a given [GdnHCl] is then used to compute population fractions of the different folding states (Y values), which in turn is used to define the overall CD signal (S) in eq 12, as was done for apoHbA (eqs 4 and 9).

22

ACS Paragon Plus Environment

Page 23 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

S[GdnHCl] = ST YT + SI YI + SU YU + SU YU + SU YU T

3(ST K I ,T K 3U =

T

M ,I T

3(K I ,T K 3U T

T

3

+ SI K 3U

M ,I T

T

+ K 3U

3

M ,I T

M ,I T

C

+

+

SU KU

C

3

3 ,3U M

1 KU

3 ,3U M

M

)[U M ]2 +

)[U M ]2 +

M

SU KU

C

C ,U M

1 KU

+ SU

M

(12)

+1

C ,U M

We also examined the dependence of apo-rHb0.1 unfolding on total protein concentration (ranging from 0.85µM to 50µM, Figure 8B). Remarkably, apo-rHb0.1 unfolding showed virtually no dependence on P0. Global fitting of the 5 titration curves in Figure 8B to eq 12 was also done in MATLAB R2016a by allowing all the K, m, and S values to vary, and the results are summarized in Table 1. A large K3UM,IT value of 6.8×1016 M-2 was obtained, consistent with the observed high resistance of the molten globule IT state to unfolding (Figure 8A, Table 1). Additionally, the trimeric molten globule (IT) and the partially unfolded monomers both show a higher fraction of secondary structure compared to apoHbA and apo-rHb0.0 (Table 1, Figure 8A), with SIT and SUM equal to ~0.7 and ~0.3 respectively, whereas the other S values were comparable to those for apoHbA. Lack of dependence of the second phase of unfolding on P0 was due to the small value of KU3,3UM, 3.6×10-16 M2, which describes the dissociation of the U3 state (Table 1). Thus, although increasing protein concentration inhibits IT dissociation, this inhibition is compensated by promotion of transient aggregation of the partially unfolded monomers to the U3 state. These results suggest that the di-α linkage not only stabilizes the oligomeric molten globule, but also promotes re-association of the mostly unfolded di-α and β subunits (UM states), inhibiting their complete unfolding into chains with no secondary structure. Genetic crosslinking has been used in the past to optimize oligomer stability for the gene v protein and the arc repressor. The resulting enhanced stability in the cross-linked proteins was attributed to the linker’s role in driving the association of the subunits by keeping them at high 23

ACS Paragon Plus Environment

Biochemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

local concentration and reducing the entropy of the unfolded subunits

Page 24 of 52

3, 69-71

. This type of

mechanism seems to be consistent with the small fitted KU3,3UM parameter obtained for the association of apo-rHb0.1 monomers. The value of K3UM,IT for the formation of the trimeric intermediate state from unfolded subunits is determined by the strength of the two α1β1 interactions that are formed. Presumably, the α1β2 and α2β1 interfaces do not exist in this intermediate because the heme pockets are unfolded.

This idea is strongly supported by noting that the square root of K3UM,IT is

2.6×108 M-1, which is very similar to the values of K2UM,ID for formation of a single dimer interface in apoHbA, aporHb0.0, and the distal pocket apo-mutants (all ~ 6 x 108 M-1, Table 1). Similar considerations apply when comparing the formation of the U3 and U2 states. The square root of the fitted value of KU3,3UM is 1.9 ×10-8 M-1, which is similar in magnitude to KU2,2UM (0.5 x 10-8 M-1, Table 1) for apoHbA and supports the idea of two interface interactions occurring in U3. Perhaps the most remarkable result is that the KIT,T value of 158 for apo-rHb0.1 is virtually identical to that for the heme pocket stabilities of apoHbA and apo-rHb0.0 dimers. This result quantitatively supports our mechanistic interpretation that the initial phase of apoHb unfolding involves unimolecular "melting" of the heme pockets and not dimer or tetramer dissociation. This result also shows that crosslinking has little effect on the average stabilities of the α and β heme pockets (Figure 8A, Table 1).

Comparison of ApoHbA and ApoHbF - In order to test the utility of the mechanism in Figure 5 for understanding Hb biochemistry and physiology, we chose to compare the apoglobin stabilities of adult and recombinant fetal human Hbs. Triplicate measurements of GdnHCl 24

ACS Paragon Plus Environment

Page 25 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

induced apoHbF unfolding curves were made at three different P0 values (Figure 10). Global fitting of all three sets of curves to the 3-step apoHbA unfolding model (Figure 5) was achieved by optimizing the S, K, and m values as previously described for apoHbA unfolding data. Again two major phases are observed, but, in the case of apoHbF, the dependence of the second phase of unfolding on P0 is smaller than for apoHbA. A direct comparison of apoHbA and apoHbF unfolding curves at P0 = 12 µM is shown in Figure 10B. The differences are subtle, with the second phase for apoHbF shifting to higher [GdnHCl]. The latter result implies that the α1γ1 interface is more resistant to dissociation than the α1β1 interface in the apo-dimer intermediate. The S coefficients and m values for the various transitions of apoHbF were very similar to those for apoHbA (Table 1). The KID,D of 128 ±11 for apoHbF suggests similar stabilities of the heme pockets between γ and β subunits (Figure 10B, Table 1). In contrast, the K2UM,ID value for apoHbF is ~10-fold larger, 84 ± 22 × 108 M-1, compared to the value for apoHbA (6.6 ± 0.1 × 108 M-1, Table 1). Previous workers have suggested that the γ1γ2 and α1γ1 interfaces in γ homo-oligomers and HbF are stronger than the β1β2 and α1β1 interfaces in β homo-oligomers and HbA due to the increased hydrophobicity of γIle116 compared to βHis116 at these interfaces43, 44, 72. Interestingly, the value of KU2,2UM for dissociation of the apoHbF U2 species decreased to 0.16 x 10-7 M compared to 5.5 x 10-7 M for apoHbA (Table 1). The new results in Figure 10 suggest that the increased apolar character of γ chains does stabilize the α1γ1 interface in the ID state of apoHbF (K2UM,ID ~10 fold higher for HbF). This increase in surface hydrophobicity of γ chains also stabilizes the unfolded U2 state, which partially compensates for its favorable effect on α1γ1 dimer formation and decreases the dependence of the second phase on protein concentration, consistent with our experimental observations (Figure 10).

25

ACS Paragon Plus Environment

Biochemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

DISCUSSION The structural similarity between human Hb subunits and monomeric mammalian myoglobin suggests that there might also be similarities between their folding mechanisms. ApoMb unfolds at neutral pH by a 3-state, 2-step mechanism involving a molten globule intermediate, which retains ~40% of the helical content of the native structure22, 39, 41, 55, 73. The intermediate (I) folding state for sperm whale Mb has been characterized by NMR and consists of an unfolded heme pocket and a folded hydrophobic core of A (N-termini), G, and H (Ctermini) helices with evidence for another minor intermediate with additional partial folding of the B helix74, 75. Results from both our laboratory 22, 39, 41, 66 and those of Wright and Baldwin23, 65, 73-75

established that the initial phase in apoMb unfolding, prior to molten globule formation,

involves “melting” of heme pocket. The introduction of large, apolar amino acids at the 7th and 11th positions of the E helix (i.e. His(E7)Phe, His(E7)Leu, and Val(E11)Phe) on the distal side of the heme pocket markedly inhibits the first phase of unfolding by stabilizing native apoMb. This stabilization also dramatically increases overall holoMb expression in E. coli22, 39, 41, 65, 66. Our results in Figures 4-11 suggest strongly that the pathway for human apoHb unfolding is very similar to that for mammalian apoMb. In the case of human apoHb, the initial step is formation of a dimeric molten globule intermediate via the “melting” of the heme pocket and is followed by concerted dissociation and formation of unfolded monomeric α and β or γ chains. As with apoMb, the introduction of His(E7)Leu and Val(E11)Phe mutations on the distal side of the heme pocket in the α and β subunits of HbA significantly increases the stability of the native apoHb dimer (Table 1, Figure 7). These mutagenesis results for apoHb, combined with previous

26

ACS Paragon Plus Environment

Page 26 of 52

Page 27 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

studies for apoMb, led us to conclude that unfolding of the heme pocket precedes formation of the molten globule heterodimer. For apoHbA and apoHbF, the helical content of the molten globule dimeric state (ID) is approximately ~40% relative to the original apoHb folded dimer (D) (Table 1). The unfolding of the ID intermediate shows a dependence on protein concentration, requiring higher [GdnHCl] at higher P0 (Figures 6 and 10A). This result suggests strongly that the intermediate is still an α1β1 or an α1γ1 heterodimer, and that retention of the secondary structure of the A, G, and H helices is coupled to the strong hydrophobic surface interactions at these interfaces. Our idea that the second phase involves hetero-dimer dissociation into monomers is further strengthened by the dramatic results of genetically crosslinking apoHb. The di-α linker in rHb0.1 markedly shifts this second phase towards much higher [GdnHCl] concentrations due to the presence of two α1β1 interfaces (Figures 8A and 9). During holoMb and holoHb assembly, the folded native and molten globule apoprotein states are greatly stabilized by hemin binding22,

38, 60

. To emphasize this point, a parallel

comparison of apo- and holoprotein unfolding curves for HbA and HbF are shown in Figure 11. Heme binding clearly stabilizes the native folding state and the heme pockets to the same degree in both HbA and HbF, and the initial phases for unfolding of metHbA and metHbF superimpose as shown in Figure 11. However, unexpectedly, hemin binding enhances the stabilizing effect of the stronger, more apolar α1γ1 interface in HbF. When hemin is bound, the second phase of unfolding involving dissociation of the dimer into monomers is shifted towards much higher GdnHCl concentrations for metHbF than for metHbA (Figure 11). This result is consistent with a previous study which showed that metHbF has a higher propensity to form hemichrome intermediates following urea denaturation than metHbA76. These intermediates represent hemin

27

ACS Paragon Plus Environment

Biochemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 28 of 52

binding to a collapsed molten globule state, in which the hemin is coordinated to two amino acid side chains to form a reversible hemichrome state. This hemichrome is an intermediate on the pathway for hologlobin folding22, 38, 60. The data in Figures 10 and 11 represent the first attempt to compare directly the equilibrium association constants for apoHbF and apoHbA heterodimer formation. Past experiments focused on measuring holo-α and holo-β/γ subunit assembly and/or dissociation rate constants, had the added complexity of β4 and γ2/γ4 formation43, 44, 77, and required computing the fractional amount of dimer present in the sample8, 77 The key remaining question is how to incorporate these results for apoHb into a larger scheme for the folding and assembly of human holoHb in vitro and during erythropoiesis in preerythroid cells. In the former case, experiments for both apo- and holoHb need to be analyzed simultaneously to obtain the effects of hemin binding on the various intermediate states in Figure 5. Again, the holoMb studies of Culberston and Olson22 serve as guide for examining the holoHb folding mechanism, and we are in the process of performing the required experiments and deriving the more complex expressions needed to analyze the holoHb results. Even without a complete mechanism for holoHb assembly, our results for human apoHb provide insight into some key physiological observations. Isolated α monomers are unstable in solution, even in the presence of bound heme. The absence of a protein oligomer partner leads quickly to denaturation, a hallmark of the more serious β-thalassemia diseases, in which one or more of the β genes are defective in the patient. In contrast, excess γ and β chains can selfassemble into more stable homo-dimers and tetramers,14,

16-18, 78

accounting for why α-

thalassemia is often less severe in terms of anemia. In effect, the formation of these homodimers and tetramers helps inhibit complete unfolding of β and γ subunits in the same way the U2 and

28

ACS Paragon Plus Environment

Page 29 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

U3 states inhibit complete unfolding of the monomeric subunits in the mechanisms shown in Figures 5 and 9. We believe the U2 or U3 states represent in part non-specific binding that could also occur to form transient homo-γ or β oligomers, which keeps these subunits in a partially folded form until α subunits are present to form stable heterodimers and tetramers. Previous work has shown that γ homo-oligomers are much more stable than β oligomers44, again due to increased hydrophobicity at their dimer interfaces. The ~30 fold decrease in the U2 dissociation equilibrium constant, KU2,2UM, for apoHbF relative to apoHbA is consistent with this idea that γ subunits can form more stable oligomeric interactions with each other (Table 1). Finally, the requirement of globin heterodimerization before formation of the apoHb intermediate state accounts for why the α and β chains have to be translated in equal amounts for efficient expression of holoHb in E. coli and during erythropoiesis. This observation suggests that the partially folded heterodimer has to be formed before hemin can be bound. Isolated apo-α and β subunits do not form stable molten globule states by themselves and, in the unfolded states, are unlikely to bind heme specifically before precipitating. However, quantitative verification of this idea will require a thorough analysis of holoHbA unfolding and expression experiments.

29

ACS Paragon Plus Environment

Biochemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACKNOWLEDGEMENT We thank Eileen Singleton for her assistance with the Biostat C20 bioreactor (B Braun Biotech International, Melsungen, Germany). We also thank Eileen Singleton, Dr. Jayashree Soman, and Dr. Neal Varnado for their help with the purification and quality controls of some of the rHb proteins. SUPPORTING INFORMATION This additional material contains a description of all four models for apohemoglobin unfolding, statistical analyses of the fits to these models to the unfolding data for apoHbA (Table S1), and the final fitted parameters for four different apoHbA unfolding models (Table S2). Comparison of measured and calculated unfolding curves for apoHbA to the 3-step model with just UM→Uc,, and to the 3-step model with just UM→U2 are shown in Figures S1 and S2, respectively. Plots of raw and standardized residual distributions for fits to the four models are given in Figures S3 and S4, respectively.

30

ACS Paragon Plus Environment

Page 30 of 52

Page 31 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

REFERENCES [1] Ofran, Y., and Rost, B. (2003) Analysing six types of protein-protein interfaces, J Mol Biol 325, 377-387. [2] Dyson, H. J., and Wright, P. E. (2005) Intrinsically unstructured proteins and their functions, Nature reviews. Molecular cell biology 6, 197-208. [3] Levy, Y., Wolynes, P. G., and Onuchic, J. N. (2004) Protein topology determines binding mechanism, Proceedings of the National Academy of Sciences of the United States of America 101, 511-516. [4] Schechter, A. N. (2008) Hemoglobin research and the origins of molecular medicine, Blood 112, 3927-3938. [5] Bunn, H. F., and Forget, B. G. (1986) Hemoglobin Structure, In Hemoglobin--molecular, genetic, and clinical aspects, pp 13-35, W.B. Saunders Co., Philadelphia. [6] Perutz, M. F. (1969) The Croonian Lecture, 1968. The haemoglobin molecule, Proc R Soc Lond B Biol Sci 173, 113-140. [7] Antonini, E., Bucci, E., Fronticelli, C., Chiancone, E., Wyman, J., and Rossi-Fanelli, A. (1966) The properties and interactions of the isolated alpha- and beta-chains of human haemoglobin. V. The reaction of alpha- and beta-chains, J Mol Biol 17, 29-46. [8] Mrabet, N. T., Shaeffer, J. R., McDonald, M. J., and Bunn, H. F. (1986) Dissociation of dimers of human hemoglobins A and F into monomers, J Biol Chem 261, 1111-1115. [9] Shaeffer, J. R., McDonald, M. J., Turci, S. M., Dinda, D. M., and Bunn, H. F. (1984) Dimermonomer dissociation of human hemoglobin A, J Biol Chem 259, 14544-14547. [10] McGovern, P., Reisberg, P., and Olson, J. S. (1976) Aggregation of deoxyhemoglobin subunits, J Biol Chem 251, 7871-7879. [11] Hensley, P., Edelstein, S. J., Wharton, D. C., and Gibson, Q. H. (1975) Conformation and spin state in methemoglobin, J Biol Chem 250, 952-960. [12] Edelstein, S. J., Rehmar, M. J., Olson, J. S., and Gibson, Q. H. (1970) Functional aspects of the subunit association-dissociation equilibria of hemoglobin, J Biol Chem 245, 43724381. [13] Weiss, M. J., and dos Santos, C. O. (2009) Chaperoning erythropoiesis, Blood 113, 21362144. [14] Yip, Y. K., Waks, M., and Beychok, S. (1972) Influence of prosthetic groups on protein folding and subunit assembly. I. Conformational differences between separated human alpha- and beta- globins, J Biol Chem 247, 7237-7244. [15] Waks, M., Yip, Y. K., and Beychok, S. (1973) Influence of prosthetic groups on protein folding and subunit assembly. Recombination of separated human alpha-and beta-globin chains with heme and alloplex interactions of globin chains with heme-containing subunits, J Biol Chem 248, 6462-6470. [16] Krishna Kumar, K., Dickson, C. F., Weiss, M. J., Mackay, J. P., and Gell, D. A. (2010) AHSP (alpha-haemoglobin-stabilizing protein) stabilizes apo-alpha-haemoglobin in a partially folded state, Biochem J 432, 275-282. [17] Adachi, K., Zhao, Y., Lakka, V., Weiss, M. J., and Surrey, S. (2007) Assembly of recently translated full-length and C-terminal truncated human gamma-globin chains with a pool of alpha-globin chains to form Hb F in a cell-free system, Archives of biochemistry and biophysics 463, 60-67.

31

ACS Paragon Plus Environment

Biochemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

[18] Adachi, K., Zhao, Y., and Surrey, S. (2002) Assembly of human hemoglobin (Hb) beta- and gamma-globin chains expressed in a cell-free system with alpha-globin chains to form Hb A and Hb F, J Biol Chem 277, 13415-13420. [19] Fanelli, A. R., Antonini, E., and Caputo, A. (1958) Studies on the structure of hemoglobin. I. Physicochemical properties of human globin, Biochim Biophys Acta 30, 608-615. [20] Winterhalter, K. H., and Huehns, E. R. (1964) Preparations, Properties, and Specific Recombination of Alpha-Beta-Globin Subunits, J Biol Chem 239, 3699-3705. [21] Winterhalter, K. H., Heywood, J. D., Huehns, E. R., and Finch, C. A. (1969) The free globin in human erythrocytes. I, Br J Haematol 16, 523-535. [22] Culbertson, D. S., and Olson, J. S. (2010) Role of heme in the unfolding and assembly of myoglobin, Biochemistry 49, 6052-6063. [23] Eliezer, D., and Wright, P. E. (1996) Is apomyoglobin a molten globule? Structural characterization by NMR, J Mol Biol 263, 531-538. [24] Baldwin, J., and Chothia, C. (1979) Haemoglobin: the structural changes related to ligand binding and its allosteric mechanism, J Mol Biol 129, 175-220. [25] Gelin, B. R., Lee, A. W., and Karplus, M. (1983) Hemoglobin tertiary structural change on ligand binding. Its role in the co-operative mechanism, J Mol Biol 171, 489-559. [26] Olsen, K. W. (1994) Thermal denaturation procedures for hemoglobin, Methods in enzymology 231, 514-524. [27] Kawahara, K., Kirshner, A. G., and Tanford, C. (1965) DISSOCIATION OF HUMAN COhemoglobin by urea, guanidine hydrochloride, and other reagents, Biochemistry 4, 12031213. [28] Fotouhi, L., Yousefinejad, S., Salehi, N., Saboury, A. A., Sheibani, N., and MoosaviMovahedi, A. A. (2015) Application of merged spectroscopic data combined with chemometric analysis for resolution of hemoglobin intermediates during chemical unfolding, Spectrochimica acta. Part A, Molecular and biomolecular spectroscopy 136 Pt C, 1974-1981. [29] Harrington, J. P., and Keaton, L. (1993) Unfolding and release of heme from human hemoglobins A, S and F, The International journal of biochemistry 25, 661-664. [30] Yan, Y. B., Wang, Q., He, H. W., and Zhou, H. M. (2004) Protein thermal aggregation involves distinct regions: sequential events in the heat-induced unfolding and aggregation of hemoglobin, Biophysical journal 86, 1682-1690. [31] Michnik, A., Drzazga, Z., Kluczewska, A., and Michalik, K. (2005) Differential scanning microcalorimetry study of the thermal denaturation of haemoglobin, Biophysical chemistry 118, 93-101. [32] Winterbourn, C. C., and Carrell, R. W. (1974) Studies of hemoglobin denaturation and Heinz body formation in the unstable hemoglobins, The Journal of clinical investigation 54, 678-689. [33] Winterbourn, C. C., and Carrell, R. W. (1972) Characterization of Heinz bodies in unstable haemoglobin haemolytic anaemia, Nature 240, 150-152. [34] Jacob, H. S., and Winterhalter, K. H. (1970) The role of hemoglobin heme loss in Heinz body formation: studies with a partially heme-deficient hemoglobin and with genetically unstable hemoglobins, The Journal of clinical investigation 49, 2008-2016. [35] Singer, K., Chernoff, A. I., and Singer, L. (1951) Studies on abnormal hemoglobins. I. Their demonstration in sickle cell anemia and other hematologic disorders by means of alkali denaturation, Blood 6, 413-428.

32

ACS Paragon Plus Environment

Page 32 of 52

Page 33 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

[36] Peisach, J., Blumberg, W. E., and Rachmilewitz, E. A. (1975) The demonstration of ferrihemochrome intermediates in heinz body formation following the reduction of oxyhemoglobin A by acetylphenylhydrazone, Biochim Biophys Acta 393, 404-418. [37] Waugh, S. M., and Low, P. S. (1985) Hemichrome binding to band 3: nucleation of Heinz bodies on the erythrocyte membrane, Biochemistry 24, 34-39. [38] Culbertson, D. S. (2011) Role of Heme in the Folding and Assembly of Globins, In Biochemistry and Cel Biology, p 177, Rice University, Houston, Texas. [39] Hargrove, M. S., Krzywda, S., Wilkinson, A. J., Dou, Y., Ikeda-Saito, M., and Olson, J. S. (1994) Stability of myoglobin: a model for the folding of heme proteins, Biochemistry 33, 11767-11775. [40] Hargrove, M. S., and Olson, J. S. (1996) The stability of holomyoglobin is determined by heme affinity, Biochemistry 35, 11310-11318. [41] Samuel, P. P., Smith, L. P., Phillips, G. N., Jr., and Olson, J. S. (2015) Apoglobin Stability Is the Major Factor Governing both Cell-free and in Vivo Expression of Holomyoglobin, J Biol Chem 290, 23479-23495. [42] Dumoulin, A., Manning, L. R., Jenkins, W. T., Winslow, R. M., and Manning, J. M. (1997) Exchange of subunit interfaces between recombinant adult and fetal hemoglobins. Evidence for a functional inter-relationship among regions of the tetramer, J Biol Chem 272, 31326-31332. [43] Adachi, K., Yang, Y., Lakka, V., Wehrli, S., Reddy, K. S., and Surrey, S. (2003) Significance of beta116 His (G18) at alpha1beta1 contact sites for alphabeta assembly and autoxidation of hemoglobin, Biochemistry 42, 10252-10259. [44] Adachi, K., Zhao, Y., Yamaguchi, T., and Surrey, S. (2000) Assembly of gamma- with alpha-globin chains to form human fetal hemoglobin in vitro and in vivo, J Biol Chem 275, 12424-12429. [45] Shen, T. J., Ho, N. T., Zou, M., Sun, D. P., Cottam, P. F., Simplaceanu, V., Tam, M. F., Bell, D. A., Jr., and Ho, C. (1997) Production of human normal adult and fetal hemoglobins in Escherichia coli, Protein Eng 10, 1085-1097. [46] Shen, T. J., Ho, N. T., Simplaceanu, V., Zou, M., Green, B. N., Tam, M. F., and Ho, C. (1993) Production of unmodified human adult hemoglobin in Escherichia coli, Proceedings of the National Academy of Sciences of the United States of America 90, 8108-8112. [47] Hoffman, S. J., Looker, D. L., Roehrich, J. M., Cozart, P. E., Durfee, S. L., Tedesco, J. L., and Stetler, G. L. (1990) Expression of fully functional tetrameric human hemoglobin in Escherichia coli, Proceedings of the National Academy of Sciences of the United States of America 87, 8521-8525. [48] Looker, D., Abbott-Brown, D., Cozart, P., Durfee, S., Hoffman, S., Mathews, A. J., MillerRoehrich, J., Shoemaker, S., Trimble, S., Fermi, G., and et al. (1992) A human recombinant haemoglobin designed for use as a blood substitute, Nature 356, 258-260. [49] Looker, D., Mathews, A. J., Neway, J. O., and Stetler, G. L. (1994) Expression of recombinant human hemoglobin in Escherichia coli, Methods in enzymology 231, 364374. [50] Maillett, D. H. (July, 2003) Engineering Hemoglobins and Myoglobins for Efficient Oxygen Transport, In Chemistry and Biochemistry and Cell Biology, p 276, Rice University, Houston, Texas.

33

ACS Paragon Plus Environment

Biochemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

[51] Ascoli, F., Fanelli, M. R., and Antonini, E. (1981) Preparation and properties of apohemoglobin and reconstituted hemoglobins, Methods in enzymology 76, 72-87. [52] Gill, S. C., and von Hippel, P. H. (1989) Calculation of protein extinction coefficients from amino acid sequence data, Anal Biochem 182, 319-326. [53] Haouz, A., El Mohsni, S., Zentz, C., Merola, F., and Alpert, B. (1999) Heterogeneous motions within human apohemoglobin, Eur J Biochem 264, 250-257. [54] Graves, P. E. (2008) Enhancing Stability and Expression of Recombinant Human Hemoglobin in E. coli: Progress in the Development of a Recombinant HBOC Source, In Biochemistry and Cell Biology, p 75, Texas, Houston. [55] Scott, E. E., Paster, E. V., and Olson, J. S. (2000) The stabilities of mammalian apomyoglobins vary over a 600-fold range and can be enhanced by comparative mutagenesis, J Biol Chem 275, 27129-27136. [56] Manning, L. R., Jenkins, W. T., Hess, J. R., Vandegriff, K., Winslow, R. M., and Manning, J. M. (1996) Subunit dissociations in natural and recombinant hemoglobins, Protein science : a publication of the Protein Society 5, 775-781. [57] Manning, L. R., Dumoulin, A., Jenkins, W. T., Winslow, R. M., and Manning, J. M. (1999) Determining subunit dissociation constants in natural and recombinant proteins, Methods in enzymology 306, 113-129. [58] Julianne M. Yost, J. D. K., Don M. Coltart. (September 2008) Tris(2carboxyethyl)phosphine Hydrochloride, In e-EROS Encyclopedia of Reagents for Organic Synthesis, Duke University, Durham, NC, USA. [59] Kelly, S. M., Jess, T. J., and Price, N. C. (2005) How to study proteins by circular dichroism, Biochim Biophys Acta 1751, 119-139. [60] David S. Culbertson, J. S. O. (2010) Folding and Stability of Myoglobins and Hemoglobins, In Protein Folding and Metal Ions: Mechanisms, Biology and Disease (Claudio M. Gomes, P. W.-S., Ed.), pp 97-125, CRC Press. [61] Greenfield, N. J. (2006) Determination of the folding of proteins as a function of denaturants, osmolytes or ligands using circular dichroism, Nature protocols 1, 27332741. [62] Williams, T., and Kelly, C. (2015) Gnuplot 5.0: an interactive plotting program. [63] Bevington, P. R. (1969) Data reduction and error analysis for the physical sciences, McGraw-Hill, New York. [64] Akaike, H. (1974) New Look at Statistical-Model Identification, Ieee T Automat Contr Ac19, 716-723. [65] Schwarzinger, S., Mohana-Borges, R., Kroon, G. J., Dyson, H. J., and Wright, P. E. (2008) Structural characterization of partially folded intermediates of apomyoglobin H64F, Protein science : a publication of the Protein Society 17, 313-321. [66] Smith, L. P. (2003) The effects of amino acid substitution on apomyoglobin stability, folding intermediates, and holoprotein expression, In Biochemistry & Cell Biology, p 199, Rice University, Houston, TX. [67] Kroeger, K. S., and Kundrot, C. E. (1997) Structures of a hemoglobin-based blood substitute: insights into the function of allosteric proteins, Structure 5, 227-237. [68] Brucker, E. A. (2000) Genetically crosslinked hemoglobin: a structural study, Acta crystallographica. Section D, Biological crystallography 56, 812-816. [69] Robinson, C. R., and Sauer, R. T. (1996) Equilibrium stability and sub-millisecond refolding of a designed single-chain Arc repressor, Biochemistry 35, 13878-13884.

34

ACS Paragon Plus Environment

Page 34 of 52

Page 35 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

[70] Liang, H., Sandberg, W. S., and Terwilliger, T. C. (1993) Genetic fusion of subunits of a dimeric protein substantially enhances its stability and rate of folding, Proceedings of the National Academy of Sciences of the United States of America 90, 7010-7014. [71] Robinson, C. R., and Sauer, R. T. (1998) Optimizing the stability of single-chain proteins by linker length and composition mutagenesis, Proceedings of the National Academy of Sciences of the United States of America 95, 5929-5934. [72] Graves, P. E., Henderson, D. P., Horstman, M. J., Solomon, B. J., and Olson, J. S. (2008) Enhancing stability and expression of recombinant human hemoglobin in E. coli: Progress in the development of a recombinant HBOC source, Biochim Biophys Acta 1784, 1471-1479. [73] Barrick, D., and Baldwin, R. L. (1993) Three-state analysis of sperm whale apomyoglobin folding, Biochemistry 32, 3790-3796. [74] Hughson, F. M., Wright, P. E., and Baldwin, R. L. (1990) Structural characterization of a partly folded apomyoglobin intermediate, Science 249, 1544-1548. [75] Jennings, P. A., and Wright, P. E. (1993) Formation of a molten globule intermediate early in the kinetic folding pathway of apomyoglobin, Science 262, 892-896. [76] Harrington, J. P., Newton, P., Crumpton, T., and Keaton, L. (1993) Induced hemichrome formation of methemoglobins A, S and F by fatty acids, alkyl ureas and urea, The International journal of biochemistry 25, 665-670. [77] Manning, L. R., Russell, J. E., Padovan, J. C., Chait, B. T., Popowicz, A., Manning, R. S., and Manning, J. M. (2007) Human embryonic, fetal, and adult hemoglobins have different subunit interface strengths. Correlation with lifespan in the red cell, Protein science : a publication of the Protein Society 16, 1641-1658. [78] Higgs, D. R., Engel, J. D., and Stamatoyannopoulos, G. (2012) Thalassaemia, Lancet 379, 373-383.

35

ACS Paragon Plus Environment

Biochemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48

Page 36 of 52

Table 1: Fitted equilibrium unfolding parameters for apoHbA, apoHbF, and other rHb variants, using the 5 state, 4 step model shown in Figure 5. Parameters enclosed in parentheses are for rHb0.1 using the model shown in Figure 9. rHb β (H63L/V7F)

rHb α (H58L/V62F) β (H63L/V67F)

Parameters

HbA

HbF

rHb 0.0

rHb0.1

rHb α (H58L/V62F)

KID,D

164 ± 3

128±11

319 ± 68

(KIT,T) (158)

5,523 ± 154

586 ± 114

14,900 ±2,300

K2UM,ID

6.6 ± 0.1×108/M

83.6 ±21.5×108/M

7.99 ± 0.84 ×108/M

(K3UM,IT) (6.8×1016M-2)

6.5 ± 0.3 ×108 M-1

8.6±0.2 ×108 M-1

19 ± 5 ×108 M-1

1190

921

1190

(266)

1190

1190

1190

5.55 ×10-7M

0.165 ×10-7M

5.55 ×10-7M

(KU3,3UM) 5.55 (3.6 x 10-16M2) ×10-7M

5.55 ×10-7M

5.55 ×10-7M

16.22

17.48

16.22

(mIT,D) (20.1)

16.22

16.22

16.22

11.55

12.68

11.55

(m3UM,IT) (9.15)

11.55

11.55

11.55

5.39

5.39

5.39

(4.6)

5.39

5.39

5.39

-3.91

-6.81

-3.91

-3.91

-3.91

-3.91

S for D

1.003± 0.002

1.021± 0.001

1.008± 0.012

0.98± 0.002

0.989± 0.003

0.99± 0.005

S for ID

0.430 ± 0.002

0.452±0.002

0.44 ±0.02

0.414±0.002

0.443±0.005

0.429±0.002

S for UM

0.089±0.0001

0.11±0.01

0.095±0.005

0.081±0.001

0.088±0.005

0.118±0.009

S for U2

0.021 ±0.001

0.032±0.024

0.021±0.001

(S for U3) (0.035)

0.0174±0.0002

0.019±0.001

0.025±0.002

S for UC

0.0034±0.0002

0.008±0.008

0.0070±0.0007

(0.0042)

0.0049±0.002

0.006±0.003

0.0139±0.0036

KUC,UM KU2,2UM mID,D kJ mol-1M-1 m2UM,ID kJ mol-1 M-1 mUC,UM kJ mol-1 M-1 mU2,2UM kJ mol-1 M-1

(mU3,3UM) (-4.98) (S for T) (0.998) (S for IT) (0.698) (0.289)

36

ACS Paragon Plus Environment

Page 37 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

FIGURE LEGENDS Figure 1. Reverse phase HPLC chromatograms of native HbA, rHb0.1, and distal heme pocket rHb mutants. Protein at concentration of 310 µM was loaded onto HPLC column equilibrated with 0.1% TFA in 35% acetonitrile. The protein was eluted with an increasing gradient of acetonitrile from 35% to 47% at 1ml/min flow rate.

Figure 2. Analytical gel filtration column analysis of HbA and rHb0.1 without (A,B,C) and with(D,E,F) reducing agents. A, Elution profile of concentrated ferric HbA (K4,2=10uM) composed of a single peak. B, Elution profile of apoHbA sample showing multiple nonequilibrium peaks. C, Elution profile of aporHb 0.1 sample composed of two major nonequilibrium peaks. D, Elution profile of apoHbA following incubation with 50 mM TPEC. E, Elution profile of apoHbA sample following dialysis in 1mM DTT, 10 mM potassium phosphate, pH 7 buffer. F, Elution profile of apo-Hb0.1 sample following dialysis in 1mM DTT, 10 mM potassium phosphate, pH 7 buffer. All samples were loaded onto a 24 ml Superose-12 HR 10/30 GL column equilibrated with 200 mM potassium phosphate, pH 7 using a 100 µl sample loop. Final elution concentrations of (A) 82 µM ferric HbA; (B) 52 µM apoHbA ; (C) 23 µM apo-Hb0.1; (D) 54 µM apoHbA; (E) 38 µM apoHbA; and (F) 27 µM apo-Hb0.1. Samples were eluted with the equilibration buffer.

Figure 3. Basic models of equilibrium unfolding curves for ApoHb heterodimer (D). A, Model 1 (biphasic) – D unfolds through a dimeric intermediate folding state (ID) into 2 completely unfolded monomers (2UM). B, Model 2 (biphasic) – D dissociates into two monomeric molten globules (2IM) which then completely unfold into 2UM. C, Model 3

37

ACS Paragon Plus Environment

Biochemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(monophasic) - D directly dissociates into two unfolded monomers, 2UM, without any intermediates. These simple models for apoHb unfolding were originally proposed by Culbertson and Olson38, 60.

Figure 4. Fits of equilibrium unfolding curves for apoHbA to the simple 3-state, 2-step model 1. The solid circles and triangles are the observed data for fractional CD changes at 222 nm, and the solid and dashed lines are the fitted curves for the 2-step apoHb heterodimer unfolding mechanism (Figure 3A) obtained from global fitting of all the five curves to eq 4. These unfolding titrations were done in 200 mM potassium phosphate, 1mM DTT, pH 7 at 10° C and the total protein concentration (P0) is listed by symbol for the five different P0 titrations (i.e. P0 increased from 1.9 to 140 µM subunits). The observed data for P0 = 1.9, 12, and 61 µM represent the average of triplicate titrations. Note that the simple 2-step model cannot describe the final third phase for the loss of CD signal and that the observed data show a significantly smaller dependence on total protein concentration than is predicted by 3 state mechanism.

Figure 5. Complete 3-phase mechanism for human apoHb unfolding. The apoHb heterodimer (D) initially unfolds into a dimeric, molten globule state (ID). ID dissociates into unfolded monomers (2UM), which still retain a small fraction of α-helicity. These unfolded monomers then either transiently form a heterodimer unfolded species (U2) through non-specific interactions or completely unfold into unstructured peptide chains (2UC).

Figure 6. Fits of equilibrium unfolding curves for apoHbA to the complete 5-state, 4-step model shown in Figure 5. The symbols represent observed data and were taken from Figure 4.

38

ACS Paragon Plus Environment

Page 38 of 52

Page 39 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

The solid and dashed lines are curves predicted from the 4-step apoHb heterodimer unfolding mechanism (Figure 5) and were obtained by global fitting of all the five curves to eq 9. As shown, this 4-step model is able to describe the smaller dependence on protein concentration of the second phase of unfolding and accurately represent the final third phase for loss of CD signal as described in the text.

Figure 7. GdnHCl induced equilibrium unfolding of apo- HbA and distal heme pocket rHb mutants followed by CD changes. The circles are the observed data, and the solid lines are the fitted curves to the 3-phase apoHb heterodimer unfolding mechanism (Figure 5) using eq 9. Unfolding measurements of these apoHbs were performed using 12 µM protein in 200 mM potassium phosphate, 1mM DTT, pH 7 at 10° C. The individual curves are labeled in the figure.

Figure 8. A, Comparison of GdnHCl induced equilibrium unfolding curves for apoprotein forms of HbA, rHb0.0, rHb0.1, and α(H58L/V62F)β(wt) rHb. The unfolding measurements were done at 12 µM protein subunits. B, GdnHCl induced equilibrium unfolding for apoHb0.1 measured at 5 different protein concentrations. The circles, squares, and triangles are the observed data, and the solid lines are the fitted curves. All five sets of unfolding measurements for aporHb 0.1 were fitted globally to the 3-phase apo-rHb0.1 unfolding mechanism (Figure 9) using eq 12. The unfolding measurements for the apoHbs without the genetic cross-linking were fitted to the 3-phase apoHb heterodimer unfolding mechanism (Figure 5) using eq 9. All measurements were done in 200 mM potassium phosphate, 1mM DTT, pH 7 at 10° C.

39

ACS Paragon Plus Environment

Biochemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 9. Apo-Hb0.1 equilibrium unfolding mechanism. Trimeric apo-rHb0.1 (T) unfolds initially into a trimeric molten globule state (IT). IT then undergoes dissociation into 3 unfolded monomers (3UM) that still retain a small fraction of α-helicity and are composed of 1 di-α monomer (UM(α)) and 2 β monomers (2UM(β)). These unfolded monomers then either transiently form a heterotrimer unfolded species (U3 ) through non-specific interactions or completely unfold into monomeric polypeptide chains (3UC ).

Figure 10. A, GdnHCl induced equilibrium unfolding of apoHbF as function of total protein concentration. B, Comparison between GdnHCl induced equilibrium unfolding curves for apoHbA and apoHbF both at 12 µM total protein subunits. The solid circles are the observed data, and the solid lines are the fitted curves to the 3-phase apoHb heterodimer unfolding mechanism (Figure 5) obtained from global fittings across varying apoHb protein concentrations using eq 9. Unfolding measurements of apoHbF and apoHbA were done in 200 mM potassium phosphate, 1mM DTT, pH 7 at 10° C.

Figure 11. GdnHCl induced equilibrium unfolding of apo- and holo- hemoglobins followed by CD changes. The circles represent observed data for apoHbA and apoHbF and the triangles represent data for metHbA and metHbF. The solid lines represent fits of the apoHbA and apoHbF data to the 3-phase apo-heterodimer unfolding mechanism (Figure 5) using eq 9. The dashed lines were just drawn through the holoprotein data to show the two phases. All four sets of measurements were taken using 12 µM protein subunits in 200 mM potassium phosphate, pH 7, 100C. 1mM DTT was present in the apo- but not the holoprotein unfolding experiments.

40

ACS Paragon Plus Environment

Page 40 of 52

Page 41 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

FIGURES

Figure 1. Reverse phase HPLC chromatograms of native HbA, rHb0.1, and distal heme pocket rHb mutants. Protein at concentration of 310 µM was loaded onto HPLC column equilibrated with 0.1% TFA in 35% acetonitrile. The protein was eluted with an increasing gradient of acetonitrile from 35% to 47% at 1ml/min flow rate.

41

ACS Paragon Plus Environment

Biochemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 2. Analytical gel filtration column analysis of HbA and rHb0.1 without (A,B,C) and with(D,E,F) reducing agents. A, Elution profile of concentrated ferric HbA (K4,2=10uM) composed of a single peak. B, Elution profile of apoHbA sample showing multiple nonequilibrium peaks. C, Elution profile of aporHb 0.1 sample composed of two major nonequilibrium peaks. D, Elution profile of apoHbA following incubation with 50 mM TPEC. E, Elution profile of apoHbA sample following dialysis in 1mM DTT, 10 mM potassium phosphate, pH 7 buffer. F, Elution profile of apo-Hb0.1 sample following dialysis in 1mM DTT, 10 mM potassium phosphate, pH 7 buffer. All samples were loaded onto a 24 ml Superose-12 HR 10/30 GL column equilibrated with 200 mM potassium phosphate, pH 7 using a 100 µl sample loop. Final elution concentrations of (A) 82 µM ferric HbA; (B) 52 µM apoHbA ; (C) 23 µM apo-Hb0.1; (D) 54 µM apoHbA; (E) 38 µM apoHbA; and (F) 27 µM apo-Hb0.1. Samples were eluted with the equilibration buffer.

42

ACS Paragon Plus Environment

Page 42 of 52

Page 43 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

Figure 3. Basic models of equilibrium unfolding curves for ApoHb heterodimer (D). A, Model 1 (biphasic) – D unfolds through a dimeric intermediate folding state (ID) into 2 completely unfolded monomers (2UM). B, Model 2 (biphasic) – D dissociates into two monomeric molten globules (2IM) which then completely unfold into 2UM. C, Model 3 (monophasic) - D directly dissociates into two unfolded monomers, 2UM, without any intermediates. These simple models for apoHb unfolding were originally proposed by Culbertson and Olson38, 60.

43

ACS Paragon Plus Environment

Biochemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 4. Fits of equilibrium unfolding curves for apoHbA to the simple 3-state, 2-step model 1. The solid circles and triangles are the observed data for fractional CD changes at 222 nm, and the solid and dashed lines are the fitted curves for the 2-step apoHb heterodimer unfolding mechanism (Figure 3A) obtained from global fitting of all the five curves to eq 4. These unfolding titrations were done in 200 mM potassium phosphate, 1mM DTT, pH 7 at 10° C and the total protein concentration (P0) is listed by symbol for the five different P0 titrations (i.e. P0 increased from 1.9 to 140 µM subunits). The observed data for P0 = 1.9, 12, and 61 µM represent the average of triplicate titrations. Note that the simple 2-step model cannot describe the final third phase for the loss of CD signal and that the observed data show a significantly smaller dependence on total protein concentration than is predicted by 3 state mechanism.

44

ACS Paragon Plus Environment

Page 44 of 52

Page 45 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

Figure 5. Complete 3-phase mechanism for human apoHb unfolding. The apoHb heterodimer (D) initially unfolds into a dimeric, molten globule state (ID). ID dissociates into unfolded monomers (2UM), which still retain a small fraction of α-helicity. These unfolded monomers then either transiently form a heterodimer unfolded species (U2) through non-specific interactions or completely unfold into unstructured peptide chains (2UC).

45

ACS Paragon Plus Environment

Biochemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 6. Fits of equilibrium unfolding curves for apoHbA to the complete 5-state, 4-step model shown in Figure 5. The symbols represent observed data and were taken from Figure 4. The solid and dashed lines are curves predicted from the 4-step apoHb heterodimer unfolding mechanism (Figure 5) and were obtained by global fitting of all the five curves to eq 9. As shown, this 4-step model is able to describe the smaller dependence on protein concentration of the second phase of unfolding and accurately represent the final third phase for loss of CD signal as described in the text.

46

ACS Paragon Plus Environment

Page 46 of 52

Page 47 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

Figure 7. GdnHCl induced equilibrium unfolding of apo- HbA and distal heme pocket rHb mutants followed by CD changes. The circles are the observed data, and the solid lines are the fitted curves to the 3-phase apoHb heterodimer unfolding mechanism (Figure 5) using eq 9. Unfolding measurements of these apoHbs were performed using 12 µM protein in 200 mM potassium phosphate, 1mM DTT, pH 7 at 10° C. The individual curves are labeled in the figure.

47

ACS Paragon Plus Environment

Biochemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 8. A, Comparison of GdnHCl induced equilibrium unfolding curves for apoprotein forms of HbA, rHb0.0, rHb0.1, and α(H58L/V62F)β(wt) rHb. The unfolding measurements were done at 12 µM protein subunits. B, GdnHCl induced equilibrium unfolding for apoHb0.1 measured at 5 different protein concentrations. The circles, squares, and triangles are the observed data, and the solid lines are the fitted curves. All five sets of unfolding measurements for aporHb 0.1 were fitted globally to the 3-phase apo-rHb0.1 unfolding mechanism (Figure 9) using eq 12. The unfolding measurements for the apoHbs without the genetic cross-linking were fitted to the 3-phase apoHb heterodimer unfolding mechanism (Figure 5) using eq 9. All measurements were done in 200 mM potassium phosphate, 1mM DTT, pH 7 at 10° C.

48

ACS Paragon Plus Environment

Page 48 of 52

Page 49 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

Figure 9. Apo-Hb0.1 equilibrium unfolding mechanism. Trimeric apo-rHb0.1 (T) unfolds initially into a trimeric molten globule state (IT). IT then undergoes dissociation into 3 unfolded monomers (3UM) that still retain a small fraction of α-helicity and are composed of 1 di-α monomer (UM(α)) and 2 β monomers (2UM(β)). These unfolded monomers then either transiently form a heterotrimer unfolded species (U3 ) through non-specific interactions or completely unfold into monomeric polypeptide chains (3UC ).

49

ACS Paragon Plus Environment

Biochemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 10. A, GdnHCl induced equilibrium unfolding of apoHbF as function of total protein concentration. B, Comparison between GdnHCl induced equilibrium unfolding curves for apoHbA and apoHbF both at 12 µM total protein subunits. The solid circles are the observed data, and the solid lines are the fitted curves to the 3-phase apoHb heterodimer unfolding mechanism (Figure 5) obtained from global fittings across varying apoHb protein concentrations using eq 9. Unfolding measurements of apoHbF and apoHbA were done in 200 mM potassium phosphate, 1mM DTT, pH 7 at 10° C.

50

ACS Paragon Plus Environment

Page 50 of 52

Page 51 of 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Biochemistry

Figure 11. GdnHCl induced equilibrium unfolding of apo- and holo- hemoglobins followed by CD changes. The circles represent observed data for apoHbA and apoHbF and the triangles represent data for metHbA and metHbF. The solid lines represent fits of the apoHbA and apoHbF data to the 3-phase apo-heterodimer unfolding mechanism (Figure 5) using eq 9. The dashed lines were just drawn through the holoprotein data to show the two phases. All four sets of measurements were taken using 12 µM protein subunits in 200 mM potassium phosphate, pH 7, 100C. 1mM DTT was present in the apo- but not the holoprotein unfolding experiments.

51

ACS Paragon Plus Environment

Biochemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

For Table of Contents (TOC) Use Only

The Mechanism of Human Apohemoglobin Unfolding Premila P. Samuel , William C. Ou, George N. Phillips, Jr., John S. Olson

52

ACS Paragon Plus Environment

Page 52 of 52