pubs.acs.

org/biochemistry Article

N‑Terminal Extensions Appear to Frustrate HU Heterodimer

Formation by Strengthening Intersubunit Contacts and Blocking the
Formation of a Heterotetrameric Intermediate
Kanika Arora,† Bhishem Thakur,† Arpita Mrigwani, and Purnananda Guptasarma*
Cite This: Biochemistry 2021, 60, 1836−1852 Read Online

ACCESS Metrics & More Article Recommendations *

sı Supporting Information
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

ABSTRACT: HU is a bacterial nucleoid-associated protein. Two

homologues, known as HU-A, and HU-B, are found in Escherichia
coli within which the early, late, and stationary phases of growth are
Downloaded via IISER MOHALI on April 28, 2023 at 04:58:21 (UTC).

dominated by HU-AA, HU-BB, and HU-AB dimers, respectively.

Here, using genetic manipulation, mass spectrometry, spectrosco-
py, chromatography, and electrophoretic examination of gluta-
raldehyde-mediated cross-linking of subunits, in combination with
experiments involving mixing, co-expression, unfolding, and
refolding of HU chains, we show that the spontaneous formation
of HU-AB heterodimers that is reported to occur upon mixing of
wild-type HU-AA and HU-BB homodimers does not occur if chains possess N-terminal extensions. We show that N-terminal
extensions interfere with the conversion of homodimers into heterodimers. We also show that heterodimers are readily formed at
anticipated levels by chains possessing N-terminal extensions in vivo, when direct chain−chain interactions are facilitated through
production of HU-A and HU-B chains from proximal genes located upon the same plasmid. From the data, two explanations emerge
regarding the mechanism by which N-terminal extensions happen to adversely affect the conversion of homodimers into
heterodimers. (1) The disappearance of the α-amino group at HU’s N-terminus impacts the intersubunit stacking of β-sheets at
HU’s dimeric interface, reducing the ease with which subunits dissociate from each other. Simultaneously, (2) the presence of an N-
terminal extension appears to sterically prevent the association of HU-AA and HU-BB homodimers into a critically required,
heterotetrameric intermediate (within which homodimers could otherwise exchange subunits without releasing monomers into
solution, by remaining physically associated with each other).

U sing negative supercoiling and binding of nucleoid-

associated proteins (NAPs) such as HU to reduce the
hydrodynamic volume of DNA,1,2 the Escherichia coli cell
manages to effect a 1000-fold compaction of its ∼1.6 mm long
(∼4.5 Mbp) genome and pack it inside a rod-shaped cell that
is only ∼1−2 μm in length. Whereas most bacteria contain
only a single form of HU, enterobacteriaceae such as E. coli
possess two forms of HU, known as HU-A and HU-B, which
are encoded by the genes hupA and hupB, respectively, the
locations of which are separated by one-fifth of the genome’s
length.3,4 Much is known about HU-A and HU-B. However,
why enteric bacteria have two forms while other bacteria have
only one is still not fully understood. One suggestion is that
enteric bacteria live both within the body and outside the body
and could benefit from different types (or extents) of
packaging of DNA, during growth in different environments,
by mixing two different forms of HU. The two chains, HU-A
and HU-B, exist either as homodimers (HU-AA or HU-BB) or
as heterodimers (HU-AB), displaying an overall stronger
preference to form HU-AB heterodimers, as well as a tendency
for HU-BB (but not HU-AA) to associate into tetramers and
octamers.5,6
Here, we explore the mechanism by which HU heterodimers
are formed. In principle, from a purely theoretical viewpoint,
heterodimers of HU could potentially form through any one of
four different mechanisms: (1) de novo association of
monomeric chains of HU-A and HU-B during their synthesis
in vivo in close proximity, e.g., from two adjacent genes on the
same plasmid; (2) dissociation of preassembled homodimers
of HU-AA and HU-BB into folded/partially-unfolded mono-
mers through the operation of a natural dissociation−
association equilibrium that happens to then be “bled away”
toward the preferential formation of heterodimers (because
heterodimers are potentially more thermodynamically stable);
(3) formation of a heterotetrameric intermediate through
association of homodimers of HU-AA and HU-BB, leading to
swapping of subunits to generate a pair of dissociated
Received: January 29, 2021
Revised: May 12, 2021
Published: May 21, 2021

© 2021 American Chemical Society https://doi.org/10.1021/acs.biochem.1c00081

1836 Biochemistry 2021, 60, 1836−1852
Biochemistry pubs.acs.org/biochemistry Article

Figure 1. Structural similarities of E. coli HU homologues. (A) Structure of the HU-A chain (PDB entry 1MUL) showing side chains of
hydrophobic residues, phenylalanine (red), proline (blue), valine (magenta), leucine (orange), and isoleucine (copper). (B) Structure of the HU-B
chain (PDB entry 4P3V) showing hydrophobic residues with the same color scheme as in panel A. (C) Superimposed structures of HU-A and HU-
B showing conservation of backbone trajectory and hydrophobic side chain orientations. (D) Structure of the heterodimer formed by interacting
HU-A and HU-B chains (PDB entry 2O97). The N-terminal domain (NTD) consists of a short helix of three turns, beginning at the N-terminus
(halfway down the molecule), which is followed by a long helix of five and a half turns. The C-terminal domain (CTD) consists of an antiparallel β-
sheet, an intrinsically disordered loop that also adopts a twisted antiparallel β-sheet structure upon binding to DNA (not seen in this representation,
although the beginning and end of the loop are seen at the top of panel D), followed by two and a half turns of helix at the chain’s C-terminus. Note
the interactions between the antiparallel β-sheets across the subunit interface, in the CTD.

heterodimers; however, without releasing any folded mono-
mers into solution; or (4) involvement of an additional
protein, e.g., a chaperone that facilitates subunit dissociation
and reassembly into heterodimers, either independently, or in
conjunction with mechanism 2 or 3.
In the work presented in this paper, we have carried out a
systematic exploration of all of the possibilities mentioned
above. We conclude that mechanism 1 and mechanism 3 are
highly likely. We also conclude that mechanism 2 is highly
unlikely and that mechanism 4 is completely unlikely. These
conclusions are based on a combination of the data presented
in this paper, and what is already known about HU. Our
experiments use examination of the effects, upon heterodimer
formation, of modifying HU’s N-terminus through extension
by genetic fusion with either small peptide sequences (e.g.,
6xHis tags used for affinity chromatographic purification) or
protein sequences [e.g., the sequence of red fluorescent protein
(RFP) fused with the N-terminus of HU either with or without
a 6xHis tag].
Below, we summarize what is currently known about HU-A
and HU-B, to present the background that is necessary to
understand the work outlined here. HU-A and HU-B are
together present within E. coli at levels as high as ∼50000
dimers per cell,7,8 which amount to the production of ∼2.5 mg
of HU per liter of bacterial culture. Expression of hupA and
hupB is observed during every phase of growth of bacterial
cultures; however, individual levels of HU-A and HU-B chains
vary with the phase of growth, causing HU-AA homodimers to
dominate early exponential growth, HU-BB homodimers to
dominate late exponential growth, and HU-AB heterodimers to
dominate the stationary phase, even though all of these forms
exist with different relative abundances, at all times.5−7 HU
dimers bind nonspecifically to DNA to induce DNA bending
and display high affinity for bent, nicked, and gapped DNA, as
well as for Holliday (cruciform, four-way) junction DNA.9,10
DNA binding involves a pair of extended lysine- and arginine-
rich β-hairpin loops (one from each monomer) that embrace
the minor groove of DNA through a forceps-like action.11
Additional DNA contacts are made by lysine residues located
at the C-terminus of each monomer.12,13 Thermal unfolding of
homodimers has previously suggested that the native dimeric
1837
form (N2) unfolds into a partially unfolded dimeric
intermediate (I2) that subsequently transforms into dissociated
and unfolded monomers (2D).14 N2 is proposed to be the
preferred state of HU-AA homodimers, while I2 is proposed to
be the preferred state of HU-BB homodimers.14 Interestingly,
and somewhat counterintuitively (i.e., if the N2 model is true
for HU-AA and the I2 model is true for HU-BB), HU-BB is
also reported to display a stronger tendency than HU-AA to
form tetramers and octamers.5 Deletion mutants of hupA or
hupB display subtle growth defects,15 whereas deletion of both
hupA and hupB gives rise to profound defects.16,17 Such defects
are partially compensated by suppressor mutations in gyrase or
topoisomerase I.18 HU-A and HU-B display a high degree of
sequence identity (∼70%) and similarity (∼80%), as well as
very high structural similarity. The two homologues also
display similarities of hydrophobic side chain orientation but
different modes of DNA binding.19 Panels A and B of Figure 1
present the similar structures of the polypeptide chains of HU-
A [Protein Data Bank (PDB) entry 1MUL]14 and HU-B (PDB
entry 4P3V), respectively, as monomers. These monomers are
not known to actually exist in solution. Therefore, the
structures presented in panels A and B of Figure 1 use the
coordinates of the monomers that have been extracted from
the crystallized structures of dimers, merely for comparison.
Figure 1C shows that the two monomers are superimposable,
incidentally, to a root-mean-square deviation (RMSD) of 0.56
Å, as calculated by TM-ALIGN.20 Figure 1D shows HU-A and
HU-B chains associated into a heterodimer that has previously
been crystallized (PDB entry 2O97).21
HU-A and HU-B possess a helix-dominated N-terminal
domain (NTD) comprising residues 1−44 and a sheet/IDR-
dominated, highly basic, C-terminal domain (CTD) compris-
ing residues 45−90. Both the NTD and the CTD are engaged
in intersubunit as well as intrasubunit interactions, within both
homodimers and heterodimers.21,22 Intersubunit CTD-CTD
interactions are hydrophobic, involving stacking interactions
between antiparallel β-strands, as seen in Figure 1D, with
charge−charge interactions involving CTD occurring mainly
with DNA, and not between (or within) subunits. Intersubunit
NTD-NTD interactions involve two ionic interactions
(between the α-amino group of residue, M1, and the side
https://doi.org/10.1021/acs.biochem.1c00081
Biochemistry 2021, 60, 1836−1852
Biochemistry pubs.acs.org/biochemistry
chain of residue, D40, and between the side chains of residues
D8 and K18), in addition to hydrophobic interactions that
occur between abutting pairs of helices (a short helix of nearly
three turns and a long helix of five and a half turns comprising
a helix-turn-helix motif). Autonomous folding, assembly, and
DNA binding by the CTD of HU chains have recently been
established by us through the creation of a novel NTD-lacking
DNA-binding protein, which uses two fused CTDs derived
from the HU chains of two different organisms.22
HU homodimers and heterodimers are discernible through a
combination of ion-exchange chromatography and analytical
electrophoretic separation of HU-A and HU-B chains, which is
accomplished using AUT gels that incorporate acetic acid,
urea, and Triton X-100.5−7 It is reported that mixing of wild-
type HU-AA and HU-BB homodimers (independently created
through chromatographic separation of unfolded HU-A and
HU-B chains, followed by refolding into folded homodimers,
prior to mixing) leads to their spontaneous reorganization to
form HU-AB heterodimers, to a level of ∼90% of the
population.5−7 It has been suggested that HU-AB heterodimer
formation is owed to the formation of a heterotetrameric
intermediate created by the transient association of HU-AA
and HU-BB homodimers.23 HU purified from stationary-phase
E. coli is also reported to consist mainly of HU-AB
heterodimers, to a level of ∼90% of the population. This
suggests that the preferential association of chains into
heterodimers occurs both in vivo and in vitro,5−7 quite
independently of whether it occurs through (a) the de novo
association of HU-A and HU-B chains upon biosynthesis, to
form heterodimers, (b) the spontaneous reorganization of
folded and preassembled HU-AA and HU-BB homodimers
into HU-AB heterodimers, (c) the release of folded/partially
unfolded monomers by each homodimer to allow their
subsequent association into heterodimers, or (d) the formation
of a transient heterotetrameric intermediate.
Most interestingly, in an apparent case of contrasting data
that have been reported (but not yet subjected to detailed
analyses, or interpretation), it has been observed that when
either HU-A chains or HU-B chains are overexpressed in E. coli
with six-His tags fused to their N-termini, they are purified
through Ni-NTA affinity chromatography [or imidazole metal
affinity chromatography (IMAC)] as pure homodimers, with
no evidence of the formation of any heterodimers, or of the
presence of wild-type HU chains lacking the histidine tag.24
This is surprising for two reasons. (I) Abundant scope exists
in vivo for the occurrence of interactions of either histidine-
tagged HU-A chains or histidine-tagged HU-B chains with
HU-A chains lacking the histidine tag or HU-B chains lacking
the histidine tag. In cells overexpressing either of the two HU
chains with a histidine tag from any plasmid, overexpression
occurs at a level of only ∼10 mg per liter of culture. In such
cells, through natural expression from the genomic copies of
hupA and hupB, HU-A and HU-B chains lacking the histidine
tag are also produced at a combined level of ∼2.5 mg per liter
of culture. In other words, histidine-tagged HU chains and
histidine tag-lacking HU chains are produced within the same
cells, and within the same order of magnitude of expression.
This provides abundant scope for interactions between
homodimers of histidine-tagged HU chains and HU chains
lacking the histidine tag and ought to facilitate formation of
heterodimers in which one chain (expressed from a plasmid) is
histidine-tagged, while the other chain (produced from the
genomic copy of the hupA/hupB gene) lacks a histidine tag.
1838
Article
Despite the existence of such scope, however, heterodimer
formation between histidine-tagged chains, and chains lacking
the histidine tag, has not been observed or reported.24 (II)
When histidine-tagged HU-AA and HU-BB homodimers are
separately produced in different populations of overexpressing
cells, and the two homodimers are then purified, completely
unfolded, and allowed to co-refold within the same solution
(after being mixed together in equal proportions), it has been
reported that 90% of the population then consists of histidine-
tagged HU-AB heterodimers, rather than either HU-AA
homodimers or HU-BB homodimers.24 This demonstrates
that the formation of heterodimers by histidine-tagged HU
chains is not sterically forbidden, per se, as an outcome.
Rather, the nonformation of such heterodimers appears to be
something that does not ordinarily happen, as a process, for
histidine-tagged chains. In other words, for such chains, the
outcome of heterodimer formation is easily achieved through
one process (unfolding and co-refolding of histidine-tagged
chains), but not through another process (simple mixing and
spontaneous reorganization of homodimers already formed by
histidine-tagged chains), whereas both processes produce
heterodimers if chains lack histidine tags at their N-termini.
The fact that HU-AB heterodimers are not naturally, or
spontaneously, formed by (i) histidine-tagged HU-AA or HU-
BB homodimers or (ii) histidine-tagged HU-B or HU-A chains
through interactions with HU-B or HU-A chains lacking the
histidine tag, suggests that there is something that prevents
facile subunit exchange in dimers that contain chains
possessing N-terminal histidine tags. This is a highly intriguing
finding, and worthy of further investigation, from a structural
and biochemical viewpoint, for anyone interested in subunit
exchange and HU-AB heterodimer formation.
We have previously investigated whether the presence of a
histidine tag on HU-B interferes with its ability to interact with
an HU-A sequence lacking a histidine tag at its N-terminus.
We performed this study by creating a simulacrum of the
interaction between HU-B and HU-A chain sequences. The
simulacrum we created consisted of a genetic fusion of an N-
terminally histidine-tagged HU-B sequence with a HU-A
sequence lacking the histidine tag, to create a single chain
containing both HU sequences (6xHis-HU-B-linker-HU-A or
HU-B-A).25 We have reported that such a fusion (i.e., HU-B-
A) functions as a perfect simulacrum of the HU-AB
heterodimer, in that (i) it undergoes folding through
interactions between HU-B and HU-A chain sequences, akin
to any HU-AA, HU-BB, or HU-AB dimer, (ii) it binds to
DNA, (iii) it assembles further, to a limited degree, into chain
dimers (i.e., into a simulacrum of a heterotetramer consisting
of two chains, each with an HU-B sequence fused to an HU-A
sequence), and (iv) it is much more thermodynamically stable
than HU-BB homodimers (which, in turn, are more
thermodynamically stable than HU-AA homodimers). Ob-
viously, in such a simulacrum, there is no scope for any further
exchange of HU-B and HU-A chain sequences, because in HU-
B-A fusion-based chains, the HU-B and HU-A sequences exist
in genetic fusion and cannot undergo any exchange between
chains, even in higher-order oligomers. Even so, this work
clearly showed very recently that the presence of a histidine tag
at the N-terminus of the HU-B sequence in the HU-B-A fusion
does not interfere with further chain associations, confounding
further the issue of why histidine-tagged chains associate into
homodimers but not into heterodimers.
https://doi.org/10.1021/acs.biochem.1c00081
Biochemistry 2021, 60, 1836−1852
Biochemistry pubs.acs.org/biochemistry
Therefore, we carried out a far more detailed analysis of the
effect of N-terminal extensions on the folding- and assembly-
related transactions of HU-A and HU-B chains, as well as HU-
AA and HU-BB homodimers, as reported in this paper. The
results provide insight into the natural mechanisms of
formation of HU heterodimers.
■ MATERIALS AND METHODS
Cloning. The hupA and hupB genes of E. coli K-12 substrain
MG1655, encoding HU-A and HU-B, were cloned into the
multiple cloning site (MCS) of the pQE-30 vector (Qiagen)
between its BamHI and HindIII restriction sites. Plasmids
incorporating these genes were transformed into E. coli M15
cells for controlled expression of 6xHis-tagged versions of HU-
A and HU-B, through induction by 1 mM IPTG. In addition, a
gene encoding Tag-RFP fused to HU-A (i.e., RFP-HU-A)
without a 6xHis tag was cloned between the NdeI and XhoI
restriction sites of one MCS of the pET-Duet-1 plasmid vector,
while the hupB gene encoding HU-B with an N-terminal 6xHis
tag was cloned between the BamHI and HindIII restriction
sites of the second MCS in the same pET-Duet-1 vector (both
sites being under T7 promoters). Following cloning, the pET-
Duet-1 plasmid was transformed into BL21(DE3)pLysS* cells
for controlled protein expression through induction by 1 mM
IPTG. The following primers were used to clone hupA
(encoding HU-A) and hupB (encoding HU-B): HU-A BamHI-
Forward, AGCTACTGGATCCATGAACAAGACTCAA-
CTGATTG; HU-A HindIII-Reverse, ATATATAAGCTT-
TTACTTAACTGCGTCTTTCAGTGCCTTG; HU-B
BamHI-Forward, AGCTACTGGATCCATGAATAAAT-
CTCAATTGATCG; HU-B HindIII-Reverse, GAATACT-
AAGCTTTTAGTTTACCGCGTCTTTCAGT; HU-A NdeI-
Forward, AGCTACTCATATGAACACGACTCAACTG-
ATTGATG; HU-A XhoI-Reverse, GAATACTCTCTCGAG-
TTACTTAACTGCGTCTTTCAGTGC; HU-B Duet-1
BamHI-Forward, AGCTACTGGATCCAATGAATAAA-
TCTCAATTGATCG; HU-B Duet-1 HindIII-Reverse,
GAATACTAAGCTTTTAGTTTACCGCGTCTTTCAGT.
The sequences of primers for cloning Tag-RFP and fusing it
to HU-A, based on previous work,26 are given below: RFP-
NdeI-Forward, 5′-ATATATCATATGGCTAGCATGACTG-
GTGGACAGCAAATG-3′. All primers other than the one
described below may be found in ref 26.
Protein Purification. After expression, cells were subjected
to sedimentation, resuspension, and lysis through sonication in
the presence of lysozyme. HU-A or HU-B proteins were
purified from the supernatants of cell lysis extracts through use
of Ni-NTA affinity purification, or IMAC. Because HU-A and
HU-B tend to be co-purified with attached fragments of DNA
and other DNA-binding proteins, lysis was performed in 20
mM phosphate-buffered saline (PBS, pH 7.4), made with
sodium dihydrogen phosphate and disodium hydrogen
phosphate, and containing 150 mM NaCl, containing an
additional 1 M NaCl and 10 mM imidazole, with this
additional NaCl serving to dissociate HU from DNA.
Thereafter, the extract was loaded onto the Ni-NTA column,
and the column (now bound by 6xHis-tagged HU-A or HU-B)
was washed with 20 mM PBS (pH 7.4) containing 2 M NaCl
and 20 mM imidazole, to allow the removal of any DNA and
other proteins bound to the HU polypeptides, as well as any
contaminating proteins loosely bound to the column matrix.
Thereafter, PBS (pH 7.4) containing 250 mM imidazole was
used to elute the HU polypeptides from the Ni-NTA resin.
1839
Article
Following elution, HU polypeptides were dialyzed extensively
against 20 mM PBS alone, to remove imidazole, prior to
spectroscopic (or other) investigations. In individual prepara-
tions, if any faint impurities were detected by sodium dodecyl
sulfate−polyacrylamide gel electrophoresis (SDS−PAGE), a
further purification through gel filtration chromatography was
performed, although this was unnecessary in most instances.
Creation of the Four-Way Junction Holliday Inter-
mediate (4WJ-DNA). Four oligonucleotides designed to
associate to create the four strands of a Holliday junction
intermediate were mixed together in equimolar proportions
according to known protocols,27 heated to 90 °C, and cooled
from 90 to 25 °C over an extended period of time to allow
strands to anneal into 4WJ-DNA. The following sequences
were used to create 4WJ-DNA: strand 1, 5′-CCCTATAACC-
CCTGCATTGAATTCCTGTCTGATAA-3′; strand 2, 5′-
GTAGTCGTGATAGGTGCAGGGGTTATAGGG-3′; strand
3, 5′-AACAGTAGCTCTTAATTCGAGCTCGCGCCC-
TATCACGACTA-3′; strand 4, 5′-TTTATCAGACTGGAA-
TTCAAGCGCGAGCTCGAATAAGAGCTACTGT-3′.
Spectroscopic Studies of the Chemical and Thermal
Stability of HU-A and HU-B Homodimers and Their
Associations. A fixed concentration of HU (0.5 mg/mL HU-
A or HU-B) homodimers was incubated overnight with
molecular biology grade urea (MP Biotech catalog no.
821530), over a range of concentrations from 0 to 8 M, or
guanidium hydrochloride (Gdm.HCl) (Promega lot no.
0000150502), over a range of concentrations from 0 to 6 M.
On the following day, alterations in the secondary structure
owing to equilibrium unfolding were assessed through circular
dichroism (CD) spectroscopy, performed using a Biologic
MOS-500 CD spectrometer, and a quartz cuvette with a path
length of 1 mm. Thermal denaturation was carried out by
heating protein samples from 20 to 90 °C through use of the
Peltier block arrangement in a Chirascan Applied Photophysics
(UK) CD spectrometer, with data being collected at 2 °C
intervals, using a protein concentration of 0.2 mg/mL, and
cuvettes with a path length of 1 mM. The raw CD ellipticity
data were then converted into the mean residue ellipticity
(MRE) using the formula MRE [Θ] in degrees = [Θobserved (in
millidegrees) × 100 × MRW]/[1000 × concentration
(milligrams per milliliter) × path length (centimeters)].
Separately, for fluorescence resonance energy transfer
(FRET) studies to examine interactions between prefolded
HU-A and HU-B, the HU-A homologue was labeled with
fluorescein isothiocyanate (FITC) and the HU-B homologue
was labeled with tetramethylrhodamine isothiocyanate
(TmRITC), with purification of labeled protein away from
free label using a desalting Superdex G-25 (PD-10) column
(GE Healthcare). FRET was examined using 495 excitation for
FITC-HU-A, TmRITC-HU-B, and mixtures of the two in
equimolar amounts on an Agilent/Varian Cary Eclipse
spectrofluorimeter. It should be noted that the labeling was
carried out at pH 8.0, at which only the α-amino group at the
N-terminus is anticipated to be modified, and no ε-amino
groups in the side chains of lysine or arginine are expected to
be modified. While this restricts the number of labels per chain
to a single label, the placement of a label at the N-terminus
could be akin to an N-terminal modification affecting HU
behavior.
Gel Filtration Chromatography to Establish the
Formation of HU Heterodimers or to Examine Unfold-
ing or Refolding of HU Homodimers. Heterodimer
https://doi.org/10.1021/acs.biochem.1c00081
Biochemistry 2021, 60, 1836−1852
Biochemistry pubs.acs.org/biochemistry
Figure 2. Non-histidine-tagged HU is not co-purified with histidine-tagged HU. (A) MALDI-Q-TOF MS spectrum of histidine-tagged E. coli HU-
AA. The sequence of the histidine tag is shown in an inset, along with values of expected and measured molecular weights. (B) MALDI-Q-TOF MS
spectrum of histidine-tagged E. coli HU-BB. The sequence of the histidine tag (identical to that used for HU-A) is shown in an inset, along with
values of expected and measured molecular weights. No wild-type HU chain displaying a mass smaller than that of the histidine-tagged chain by
1416.50 Da is detected.
Formation. The eluent from the Ni-NTA purification of the
expression product of the pET-Duet-1 vector in BL21(DE3)-
pLysS* cells was loaded onto a 24 mL Superdex-200 10/300
GL (GE Healthcare) column on an AKTA Purifier-10
workstation, to monitor the presence of Tag-RFP-HU-A
along with 6xHis-tagged HU-B, by monitoring the absorbance
at 550 nm by RFP, and the absorbance at 215 nm by HU
(which naturally lacks Tyr or Trp residues and contains only
three Phe residues).
Chemical Unfolding of Homodimers. Gel filtration chromatog-
raphy was used to examine the effect of urea on the
hydrodynamic volumes of HU homodimers, by incubating
protein overnight in different molar concentrations of urea in
20 mM PBS, and through loading of such protein on the same
AKTA workstation using a Superdex-75 10/300 GL (GE-
Healthcare) gel filtration column, except that for each urea
concentration, the column was first pre-equilibrated with buffer
of the same composition and urea concentration as the sample.
Refolding of Homodimers after Thermal Unfolding. HU-A and
HU-B proteins (before, or after, being heated at 90 °C for 5
min and cooled to room temperature) were loaded onto the
same AKTA workstation using a 24 mL Superdex-75 10/300
GL (GE-healthcare) column mentioned above, pre-equili-
brated with 20 mM PBS buffer (pH 7.4), to compare elution
(and hydrodynamic) volumes of unheated and heated−cooled
protein. After elution, proteins were examined for their ability
to bind to 4WJ-DNA using the electrophoretic mobility shift
assay (EMSA) on 0.5% agarose gels, to test heated−cooled
HU protein for refolding to the DNA-binding-competent form.
Similarly, the ability of chemically denatured HU to refold was
assessed by dialyzing denaturants out and performing gel
filtration chromatography to monitor the formation of dimeric
HU. In all experiments, the standard buffer used as a base for
column equilibration and sample examination was 20 mM
phosphate-buffered saline (pH 7.4), made with sodium
dihydrogen phosphate and disodium hydrogen phosphate,
containing 150 mM NaCl, except where additional compo-
1840
Article
nents (e.g., urea) were also included, as specifically mentioned
above.
Electrophoretic Mobility Shift Assay (EMSA) to
Examine DNA Binding by HU. Agarose gels (0.5%) cast
in standard Tris-EDTA buffer were used for EMSAs with 2 μM
4WJ-DNA and varying HU protein concentrations in the range
of 0.5−15.0 μM, with binding of protein to DNA prior to
electrophoresis carried out at room temperature for 30 min in
PBS buffer (20 mM phosphate and 150 mM NaCl).
Glutaraldehyde Cross-Linking to Examine the Oligo-
meric Status during HU Unfolding or Refolding. The
oligomeric status of HU-A or HU-B protein was estimated
using protein cross-linking and SDS−PAGE analyses through
the use of glutaraldehyde as a chemical cross-linker present at a
concentration of 0.1% (v/v). After addition of glutaraldehyde,
proteins were incubated at room temperature for 5 min, and
5× SDS sample loading buffer was added to an appropriate
(standard) measure, followed by boiling of the sample and
loading on 15% SDS−PAGE. To examine the survival of
dimeric forms, or higher-order associations, after treatment
with a chemical denaturing agent capable of destabilizing
hydrogen bonding interactions, HU-A and HU-B proteins
were incubated overnight in different molar concentrations of
urea and subjected to chemical cross-linking and SDS−PAGE,
to investigate the survival of dimers or oligomers after
incubation with urea. Separately, to establish the role of
hydrophobic interactions in the survival of dimers of HU-A or
HU-B, proteins (0.3 mg/mL) were incubated overnight with a
nonpolar, water-miscible solvent, 1,4-dioxane (CDH fine
chemicals), in the concentration range of 0−50% (v/v)
dioxane in the presence or absence of a fixed urea
concentration of 3 M. In yet other experiments, glutaraldehyde
cross-linking was conducted at 90 °C, followed by SDS−PAGE
analysis, to examine the persistence of dimers in HU
populations at 90 °C.
Estimation of Protein Concentrations. Because wild-
type HU-A and HU-B chains both lack tryptophan and
tyrosine residues, we estimated the concentrations of
https://doi.org/10.1021/acs.biochem.1c00081
Biochemistry 2021, 60, 1836−1852
Biochemistry pubs.acs.org/biochemistry Article

Figure 3. Extent of heterodimer formation obtained through mixing of HU-A and HU-B, or through co-expression. (A) Fluorescence resonance
energy transfer (FRET) studies involving mixing of purified (histidine-tagged) FITC-labeled HU-AA, and TmRITC-labeled HU-BB, with
excitation at 495 nm. Emission from the FITC on HU-A chains (in the absence of TmRITC-labeled HU-B) is colored green, and that from the
TmRITC on HU-B chains (in the absence of FITC-labeled HU-A) is colored red. The numerical sum of the green and red spectra is colored black,
as the expected spectrum in the absence of FRET. The blue spectrum shows the measured spectrum, collected to detect reduced FITC emission
and increased TmRITC emission. The Rayleigh scatter peak is shown captured at 495 nm in the blue and red spectra. (B) SDS−PAGE profile of
histidine-tagged HU-B chains purified from E. coli co-expressing non-histidine-tagged RFP-HU-A chains. (C) Gel filtration chromatographic profile
of the histidine-tagged HU-B chain containing entities purified from E. coli co-expressing non-histidine-tagged RFP-HU-A with monitoring of
absorbance at 550 nm (RFP absorption) and 215 nm (peptide bond absorption).

tryptophan- and tyrosine-lacking, histidine-tagged HU-AA and
HU-BB homodimers by comparing their CD spectroscopic
signal strengths at 208 nm with the CD signal strength of a
tryptophan-containing HU mutant (HU-B F47W) of known
concentration, with an extinction coefficient of 5500 M−1 cm−1
at 280 nm. The mutant HU protein, HU-B F47W, was
available as it had been previously made and used by us, for a
different purpose, as reported in a different publication.28
wild-type (i.e., without the histidine tag) HU-A or HU-B
chains. It may be noted that the presence of any wild-type HU
chains would have been detectable in these purifications, even
if such chains were present at a level of 2% of the entire HU
population. Therefore, it appears that histidine-tagged HU-A
and histidine-tagged HU-B chains are indeed purified as HU-
AA or HU-BB homodimers, as previously reported, i.e., despite
the use of a more sensitive method of detection (mass

■ RESULTS
Histidine-Tagged HU-A and HU-B Do Not Detectably
Form Heterodimers with Wild-Type HU-A and HU-B In
Vivo. Panels A and B of Figure 2 show MALDI-Q-TOF MS
spectra of Ni-NTA affinity chromatography-purified 6xHis
(histidine-tagged) HU-A and HU-B, respectively. The
experimentally determined masses of both species can be
seen to be within a unit mass of the expected masses. Neither
spectrum shows detectable mass peaks corresponding to any
1841
spectrometry) than AUT gel electrophoresis, which was
employed earlier,24 and the purity of HU-AA and HU-BB
homodimers is confirmed to be both true and of a high order.
This suggests that histidine-tagged HU chains either do not
spontaneously form heterodimers with non-histidine-tagged
HU or do so extremely poorly in vivo (i.e., below the limits of
detection by mass spectrometry), despite the intracellular
existence of non-histidine-tagged HU chains at levels that fall
within the same order of magnitude as that of the
overexpressed histidine-tagged HU chains (note that the
https://doi.org/10.1021/acs.biochem.1c00081
Biochemistry 2021, 60, 1836−1852
Biochemistry pubs.acs.org/biochemistry
arguments supporting expression within the same order of
magnitude have already been pointed out in the introduction).
In other words, it is not as if any difference in the levels of
expression of deliberately overexpressed histidine-tagged HU
chains, on one hand, and naturally expressed, wild-type HU
chains lacking the histidine tag, on the other, could be used to
explain the nonformation of heterodimers, because the
overexpressed HU would certainly be expected to be
contaminated by wild-type HU, given their relative levels of
expression and the spontaneous nature of conversion of HU-
AA and HU-BB into HU-AB, when no histidine tag is present.
At the same time, it is necessary to note that the overexpressed
HU is produced from genes on plasmids, whereas wild-type
HU is produced from genes located on the genome; i.e.,
histidine-tagged HU chains and HU chains lacking the
histidine tag are not produced in close proximity. This could
also explain the lack of contaminating wild-type HU. What is
established, and confirmed, at this point is the observation that
preassembled homodimers of chains possessing a histidine tag
display no tendency to spontaneously reorganize into
heterodimers by associating with preassembled homodimers
formed by wild-type HU chains lacking histidine tags.
However, the formation of heterodimers through de novo
association of freshly synthesized HU chains during their
folding and assembly remains unaddressed.
Mixing of Prefolded Histidine-Tagged HU-AA and
HU-BB Homodimers Spontaneously Yields <5% (rather
than ∼90%) Heterodimers. Figure 3A shows the results of
FRET studies using histidine-tagged FITC-labeled HU-AA
homodimers and histidine-tagged TmRITC-labeled-HU-BB
homodimers, mixed in equimolar amounts and incubated at
room temperature under conditions identical to those
previously reported to lead to spontaneous formation of
heterodimers,5−7 to a level of ∼90% of the combined
population, between wild-type (i.e., lacking the histidine tag)
HU-AA and HU-BB. The FRET experiment was conducted to
examine the extent to which spontaneous subunit exchange
can give rise to a population of heterodimers, because HU-AB
heterodimers would be expected to display FRET between
FITC-labeled-HU-A chains and TmRITC-labeled HU-B
chains (manifesting as a decrease in FITC fluorescence and
an increase in TmRITC fluorescence), in proportion with the
extent of heterodimer formation. Figure 3A shows that 495 nm
light excites both FITC and TmRITC. Between 500 and 550
nm, the TmRITC emission (red curve) from labeled HU-B
chains shows almost no fluorescence signal in the peak
emission range of the FITC emission (green curve) from
labeled HU-A chains, whereas between 550 and 600 nm, the
FITC emission (green curve) from labeled HU-A chains
displays a significant signal in the emission range of TmRITC
from labeled HU-B chains. Upon digital addition of the green
and red curves, the resulting spectrum (black curve) overlaps
with the green curve between 500 and 550 nm but is distinct
from both green and red curves between 550 and 600 nm.
There is a marginal reduction in the donor (FITC)
fluorophore’s emission and a marginal increment in the
acceptor (TmRITC) fluorophore’s emission in the exper-
imentally obtained spectrum (blue curve), relative to the curve
obtained by digital addition of the FITC- and TmRITC-
derived spectra (black curve). The increase suggests that the
level of heterodimer formation is not ∼90%, as would be
expected if FITC-labeled HU-AA homodimers and TmRITC-
labeled HU-BB homodimers had interacted to exchange
1842
Article
subunits and reorganize into FITC- and TmRITC-labeled
HU-AB heterodimers. In place of the formation of ∼90%
heterodimers, what is observed is <5% formation of
heterodimers, indicating that histidine-tagged HU-A and
HU-B chains do not undergo facile or spontaneous subunit
exchange-based heterodimer formation.
It may be noted, in passing, that the HU chains are labeled
only at their N-termini due to the pH used for labeling, and
that the diameter of an HU dimer is on the order of 3 nm (or
30 Å), whereas the FRET efficiency of the fluorescein-
rhodamine FRET pair is 50% at a distance of 55 Å,29 which is
smaller than the Forster radius of the fluorescein fluorophore.
Thus, FRET could have been expected to result from every
heterodimer but was not seen, indicating that FITC-6xHis-
HU-A and TmRITC-6xHis-HU-B do not substantially form
heterodimers unlike their wild-type counterparts. It may also
be noted in passing that the fluorescence label is placed at the
N-terminus of the 6xHistag; i.e., the label affects the N-
terminus of HU-A and HU-B chains that are already modified
at their N-termini through the placement of histidine tags.
Thus, the labels cannot be taken to be agents modifying the N-
termini of HU chains at the location of the N-terminal
methionine residues of these chains, and the effects observed
must be interpreted as arising solely from the presence of the
histidine tags alone (and not from the labeling).
Co-expression of Non-Histidine-Tagged RFP-HU-A
with Histidine-Tagged HU-B Yields between ∼33%
and 50% (rather than ∼90%) Heterodimers. Figure 3B
shows an analytical SDS−PAGE gel electrophoretic profile of
Ni-NTA-purified histidine-tagged HU-B chains (∼10.6 kDa)
co-overexpressed with non-histidine-tagged RFP-HU-A chains
(∼38.6 kDa). Following bacterial cell lysis, and during Ni-
NTA-based affinity (IMAC) purification, RFP-HU-AA homo-
dimers elute in the (separately collected) flow-through of the
column, because they have no 6xHis tag. The two species that
are purified through affinity column chromatography are those
of (i) homodimers of histidine-tagged HU-BB and (ii)
heterodimers of histidine-tagged HU-B chains and non-
histidine-tagged, but RFP-fused, HU-A chains (i.e., RFP-HU-
A+6xHis-HU-B). In the gel, both polypeptides (i.e., 6xHis-
HU-B chains and RFP-HU-A chains) are seen to be present,
with the RFP-HU-A chains constituting <50% of the total
protein purified. It may be noted that the RFP-HU-A chain is
more than 3 times the length of the 6xHis-HU-B chain and
that this affects both (a) the interpretation of the intensity of
the stained gel bands for the RFP-HU-A and 6xHis-HU-B
chains, following Ni-NTA chromatography, and (b) the
interpretation of the area under the 215 nm elution absorbance
profiles of heterodimers (i.e., RFP-HU-A+6xHis-HU-B) and
homodimers of 6xHis-HU-BB, owing to absorption by peptide
bonds. RFP-HU-A chains in the column’s flow-through (i.e.,
homodimers of RFP-HU-AA, lacking a 6xHis tag) can be
viewed in the SDS−PAGE gel as RFP-HU-A chains, and
compared with the population of RFP-HU-A chains that is
separately, and specifically, eluted in association with 6xHis-
HU-B (i.e., as RFP-HU-A+6xHis-HU-B heterodimers).
For the corresponding gel lanes, equal volumes of protein
(16 μL) were loaded from a total flow-through volume of 9
mL, and an imidazole-based elution volume of 1 mL for the
affinity-purified protein. As the intensity of RFP-HU-A chains
in the column’s flow-through is approximately one-tenth of
that of RFP-HU-A chains present in the specifically eluted
fraction (in association with 6xHis-HU-B), it would appear
https://doi.org/10.1021/acs.biochem.1c00081
Biochemistry 2021, 60, 1836−1852
Biochemistry pubs.acs.org/biochemistry
Figure 4. Raw CD ellipticity of HU-A and HU-B in millidegrees as a function of denaturant concentration. (A) HU-AA and HU-BB variations
between 0 and 8 M urea. (B) HU-AA and HU-BB variations between 0 and 6 M Gdm.HCl.
that ∼50% of the total population of RFP-HU-A chains exists
in the form of homodimers. Given the expectation that 6xHis-
HU-B and RFP-HU-A populations are expressed at com-
parable levels (because they are under identical promoters in
the pET-Duet-1 plasmid vector that incorporates the two
genes at distinct multiple cloning sites separated by a few
nucleotides), these results suggest that heterodimers are
formed to a much higher extent (≲50%) through the in vivo
association of HU chains occurring during synthesis than upon
mixing (<5%) of chains that have been produced independ-
ently.
An essentially similar conclusion is reached from Figure 3C,
which shows the analytical gel filtration chromatographic
elution profile of the Ni-NTA (IMAC)-purified protein. As
already mentioned, the red-colored RFP-HU-AA homodimer
population is lost in the flow-through of the Ni-NTA column.
What is purified and loaded onto the gel filtration column, as
seen in the lane corresponding to the specific eluate in Figure
3B, is a mixture of the 6xHis-HU-B homodimer and the
heterodimer, i.e., RFP-HU-A+6xHis-HU-B. The gel filtration
profile shows that the peak eluting at ∼14 mL is the RFP-HU-
A+6xHis-HU-B heterodimer, which displays absorption at
both 215 and 550 nm (due to the presence of RFP). The other
population of homodimeric 6xHis-HU-BB, being smaller,
elutes at ∼16.2 mL, with no associated 550 nm absorption.
Comparison of the areas of the heterodimeric peak of RFP-
HU-A+6xHis-HU-B (1968 units) and the homodimer peak of
6xHis-HU-BB (1545 units) in the 215 nm absorbance profile
suggests a ratio of 1.28:1.00 for the elutions occurring at ∼14
and ∼16.2 mL.
If the difference between 1.28 and 1.00 is ignored
(suggesting that HU-AB and HU-BB are made in roughly
equal amounts, with the heterodimer being made more than
the homodimer) and the conclusion from the SDS−PAGE
data for the RFP-HU-AA homodimer’s presence in the flow-
through and elution is taken into account (suggesting that HU-
AA and HU-AB are made in equal amounts), it would appear
that the ratios of (1) the HU-AA (homodimer) form of RFP-
HU-AA, (2) the HU-BB (homodimer) form of 6xHis-HU-BB,
and (3) the HU-AB (heterodimer) form of RFP-HU-A+6xHis-
HU-B that are obtained through de novo association are
somewhere between 1:1:1 and 1:2:1. It may be noted that the
statistical expectation would be for a ratio of 1:2:1 but that
because the levels of the heterodimer seen are lower than
expected, the ratio is between 1:2:1 and 1:1:1. This result is
1843
Article
consistent with the statistical probability of encounters
between HU-A and HU-B chains, with the caveat that some
encounters involve HU-AA and HU-BB homodimers that have
already managed to form before chains containing HU-A and
HU-B sequences encounter each other, rather than through
encounters between HU-A and HU-B chains that have not yet
folded and assembled.
The results of this section suggest that de novo heterodimer
formation can occur in vivo between HU-A and HU-B
sequence-containing chains during their synthesis, and while
chains are still in a nascent state and being produced in close
proximity, from the same plasmid, using neighboring genes. At
the same time, we performed a separate experiment in which
6xHis-RFP-HU-AA (which is similar to RFP-HU-AA) and
6xHis-HU-BB were separately expressed, purified, and mixed.
Panels A−C of Figure S1 demonstrate that there is no
significant heterodimer formation under these conditions.
In other words, although RFP-HU-A chains and 6xHis-HU-
B chains can form heterodimers in vivo through de novo
association to a level of <50% of the population, due to their
being produced from proximal genes, they still cannot
spontaneously form heterodimers (or only do so to a level of
<5% of the population) if they are first synthesized, and folded,
independently (within cells) before they are independently
purified and then allowed to mix. Under no condition is the
population of heterodimers anywhere close to ≳90% of the
population. It may also be noted that the actual relative
amounts vary somewhat from experiment to experiment.
Chemical Denaturation Reveals the Persistence of a
Hydrophobically Stabilized Dimeric Interface in Histi-
dine-Tagged HU That Could Frustrate Subunit Ex-
change. We assessed the relative abilities of the three-
dimensional structures of histidine-tagged HU-AA and HU-BB
homodimers to withstand chemical unfolding and dissociation,
using a combination of spectroscopic, electrophoretic, and
chromatographic experiments.
Circular Dichroism Reveals Differences in the Unfolding of
6xHis-HU-AA and 6xHis-HU-BB by Urea and Gdm.HCl, with
Hints of Persistence of an Unfolding-Resistant Substructure. Far-
ultraviolet CD spectroscopy was used to assess changes in the
secondary structure content of HU as a function of denaturant
concentration, following overnight incubation in denaturing
buffers. For denaturation by urea, representative CD spectra
are shown in panels A and B of Figure S2. These (and the
corresponding spectra for denaturation by Gdm.HCl) reveal
https://doi.org/10.1021/acs.biochem.1c00081
Biochemistry 2021, 60, 1836−1852
Biochemistry pubs.acs.org/biochemistry Article

Figure 5. Behavior of chemically unfolded and refolded histidine-tagged HU. (A) HU-AA incubated in urea and subjected to gel filtration
chromatography at the same urea concentration used for incubation. (B) HU-BB incubated in urea and subjected to gel filtration chromatography
at the same urea concentration used for incubation. (C) Gel filtration chromatographic behavior of HU-AA (red) and HU-BB (blue) after refolding
from 6 M urea. (D) Electrophoretic mobility shift assays of refolded HU-A chains or HU-B chains (10 μM HU chains; 5 μM HU dimers)
electrophoresed with or without 4WJ-DNA (2 μM) using an agarose (0.5%) gel: control 4WJ-DNA (lane 1), refolded HU-A (lane 2), refolded
HU-A with 4WJ-DNA (lane 3), refolded HU-B (lane 4), refolded HU-B with 4WJ-DNA (lane 5), and 100 bp resolution DNA molecular weight
marker ladder (lane 6).

that denaturation does not proceed to completion with either
urea, which predominantly destabilizes hydrogen bonding
interactions, or Gdm.HCl, which destabilizes both electrostatic
and hydrogen bonding interactions. With both denaturants,
residual ellipticity is observed at 222 nm even at very high
denaturant concentrations (note that >95% unfolding is seen
only above 7 M urea and 5 M Gdm.HCl). The raw ellipticity
values at 222 nm are turned into fractional unfolding values
and plotted in panels A and B of Figure 4, from which the
concentrations of denaturant required to achieve 50%
unfolding of the population were determined for histidine-
tagged forms of both HU-AA and HU-BB. With both HU-AA
and HU-BB, it was observed that the Gdm.HCl concentration
required for unfolding is roughly half the concentration of urea
required for unfolding. However, urea-mediated unfolding
concentrations for HU-AA (1.77 M) and HU-BB (2.14 M)
were more distinctly different than Gdm.HCl-mediated
unfolding concentrations for HU-AA (0.82 M) and HU-BB
(1.03 M), with HU-BB evidently being more stable to
chemical denaturation than HU-AA and with HU-BB also
requiring greater destabilization of hydrogen bonding to
undergo unfolding than HU-AA.
1844
Clearly, therefore, histidine-tagged HU-BB is more stable to
chemical denaturation than histidine-tagged HU-AA. This
result is at variance with the observation recorded in the
introduction that wild-type HU-BB has a partially unfolded
dimeric structure whereas wild-type HU-AA has a fully folded
structure. Of course, it is possible that the 6xHis tag
preferentially stabilizes the structure of HU-BB in relation to
wild-type HU-BB lacking the histidine tag, over and above its
effect in relation to HU-AA. Independently, the result also
supports the observation made in the introduction that HU-BB
has a stronger tendency to form oligomeric structures than
HU-AA, because a more stable protein can be anticipated to be
more likely to engage in oligomer formation and vice versa. The
spectra and plots further reveal that, in both HU-AA and HU-
BB, structural destabilization is initiated at extremely low
concentrations of denaturants, with a very high initial slope
with increasing of denaturant concentration (in the ranges <3
M urea and <1.5 M Gdm.HCl) preventing analyses of the
denaturation profiles as sigmoidal curves, with the slope
becoming much smaller only at higher denaturant concen-
trations (>3 M urea and >1.5 M Gdm.HCl).
With both HU-AA and HU-BB, the nature of the changes in
slopes of curves plotting unfolding as a function of increasing
https://doi.org/10.1021/acs.biochem.1c00081
Biochemistry 2021, 60, 1836−1852
Biochemistry pubs.acs.org/biochemistry Article

Figure 6. Glutaraldehyde cross-linking studies to estimate the strength of the dimeric interface of histidine-tagged HU. (A) HU-BB cross-linking
behavior at urea concentrations of ≤4 M. (B) HU-BB cross-linking behavior at urea concentrations of ≥4 M urea. (C) HU-AA cross-linking
behavior at urea concentrations between 0 and 6 M. (D) HU-BB cross-linking behavior in the presence of 1,4-dioxane alone or in the presence of
1,4-dioxane and urea. Details of glutaraldehyde concentrations used are mentioned above each lane. PMWM stands for protein molecular weight
markers.

denaturant concentrations above 3 M urea or 1.5 M Gdm.HCl
(Figure 4A,B), together with the shapes and relative intensities
of the CD spectra reported in panels A and B of Figure S2,
indicates that both histidine-tagged HU-AA and histidine-
tagged HU-BB lose the helical components of their structures
in a much more facile manner than they lose other (e.g., β-
sheet-based) structural components, during chemical denatu-
ration. This could explain why unfolding occurs in a less facile
manner at higher denaturant concentrations, than at lower
denaturant concentrations, in proteins in which unfolding does
not show much cooperativity. We have not been able to find
any evidence of published literature presenting chemical
denaturation studies of any form of HU, prior to these studies.
Therefore, we are unable to compare these results with any
other reports.
Analytical Gel Filtration Chromatography Reveals Differ-
ential Unfolding by Urea of HU-AA and HU-BB, with Hints
of Unfolding Preceding Dissociation (suggesting a persis-
tent substructure). Panels A and B of Figure 5 show variations
1845
in the gel filtration chromatographic behavior of HU-AA and
HU-BB, respectively, as a function of urea concentration,
following overnight incubation in different molar concen-
trations of urea, and assessed after equilibrating columns with
appropriate concentrations of urea in PBS buffer. The data
reveal the effect of urea upon the hydrodynamic volumes of
HU dimers. With an increase in urea concentration, the
hydrodynamic volumes of HU samples were initially found to
increase due to the formation of partially unfolded dimers
(with a smaller elution volume), prior to formation of
dissociated and substantively unfolded monomers (with a
larger elution volume). HU-AA, which had shifted to smaller
elution volumes (associated with larger hydrodynamic volumes
signifying partial unfolding) at urea denaturant concentrations
of ≲3.5 M, was observed to shift toward larger elution volumes
signifying dissociation of partially unfolded dimers into
substantively unfolded monomers. Such a shift was not seen
with HU-BB even at 4 M urea, supporting the result from CD
spectroscopy that HU-BB is more stable than HU-AA. In all
https://doi.org/10.1021/acs.biochem.1c00081
Biochemistry 2021, 60, 1836−1852
Biochemistry pubs.acs.org/biochemistry Article

Figure 7. CD spectral mean residue ellipticity (MRE) data monitoring thermal unfolding and refolding of histidine-tagged HU-AA and HU-BB.
(A) Changes in MRE of HU-AA at 222 nm during thermal unfolding (red) and refolding from a thermally unfolded state (blue). (B) Changes in
MRE of HU-BB at 222 nm during thermal unfolding (red) and refolding from a thermally unfolded state (blue). (C) Far-UV CD spectra of HU-
AA, before heating (black), in the heated state (red), and after cooling (blue). (D) Far-UV CD spectra of HU-BB, before heating (black), in the
heated state (red), and after cooling (blue).

likelihood, the shift occurs at even higher urea concentrations
that could not be accessed for HU-BB.
Gel Filtration Chromatography Reveals That Refolding
from a Chemically Denatured State Gives Rise to DNA-
Binding-Competent Soluble Higher-Order Oligomers of HU-
AA and HU-BB. Figure 5C shows the gel filtration behavior of
HU-AA and HU-BB exposed overnight to 6 M urea and
dialyzed to allow complete removal of the denaturant. The
figure shows that both homologues form large soluble higher-
order oligomers that elute in the void volume of the gel
filtration column, suggesting that refolding does not lead to
formation of dimers. However, the EMSA data in Figure 5D
establish that these soluble higher-order oligomers of HU-AA
and HU-BB, which are eluted in the void volume of the
column (and which were collected and tested), happen to be
DNA-binding-competent, because they slow the migration of
4WJ-DNA. Positive controls for the EMSA data shown in
Figure 5D are provided in panels A and B of Figure S3, in
which the EMSAs for 6xHis-HU-AA and 6xHis-HU-BB are
shown for a DNA concentration of 2 μM and varying protein
concentrations (0.5−45.0 μM).
1846
Glutaraldehyde Cross-Linking Reveals That the Persistent
Substructure Involves the Dimeric Subunit Interface. To
address the question of whether there is sufficient survival of
some core structure in HU-AA and HU-BB that could be
associated with intersubunit interactions, and with the
retention of dimeric, or tetrameric, states at high denaturant
concentrations, we performed cross-linking experiments using
protein samples that had previously been incubated overnight
in the presence of different concentrations of urea, with cross-
linking introduced by glutaraldehyde. The results are shown in
panels A and B of Figure 6 for HU-BB samples yielding HU-B
chains and in Figure 6C for HU-AA samples yielding HU-A
chains. From previous examination of Figure 4A, it is evident
that the half-unfolding value of HU-BB with urea is 2.14 M.
Panels A and B of Figure 6 show that dimeric species of HU-B
(i.e., nominal HU-BB homodimers) survive above a urea
concentration of 2.14 M, and even at 4 M urea, with even
tetrameric species surviving at 2.5 M urea (above the half-
unfolding value of 2.14 M). It may be noted from Figure 4A
that a concentration of 4 M urea is five-sixths of the
concentration required to elicit complete unfolding of HU-
https://doi.org/10.1021/acs.biochem.1c00081
Biochemistry 2021, 60, 1836−1852
Biochemistry pubs.acs.org/biochemistry Article

Figure 8. Behavior of histidine-tagged HU before and after thermal unfolding. (A) Gel filtration elution profile of unheated HU-AA (black) and
HU-AA refolded after exposure to 90 °C (red). (B) Gel filtration elution profile of unheated HU-BB (black) and HU-BB refolded after exposure to
90 °C (red). (C) Electrophoretic mobility shift assays of HU-AA and HU-BB following refolding after thermal unfolding, through electrophoresis
in the presence of DNA, using an agarose (0.5%) gel: 100 bp resolution DNA molecular weight marker ladder (lane 1), control 2 μM 4WJ-DNA
alone (lane 6), 2 μM 4WJ-DNA with refolded HU-A chains [0.5 μM (lane 2), 1.0 μM (lane 3), 1.5 μM (lane 4), and 2.0 μM (lane 5)], and 2 μM
4WJ-DNA with refolded HU-B chains [0.5 μM (lane 7), 1.0 μM (lane 8), 1.5 μM (lane 9), and 2.0 μM (lane 10)].

BB homodimers. This indicates that HU-BB undergoes non-
two-state unfolding with subunit dissociation uncoupled from
the molecule’s unfolding, as there is clear evidence of the
persistence of sufficient structure in the “substantially
unfolded” state to facilitate retention of the dimeric or
tetrameric state(s) that shows up in the gel cross-linking
data. An entirely similar result is obtained with HU-AA, as one
can see in Figure 6C, in which dimeric species are observed
even at 2.5 M urea, despite HU-AA displaying a half-unfolding
concentration of only 1.77 M urea. Therefore, in both
homologues, 6xHis-HU-AA and 6xHis-HU-BB, glutaraldehyde
is able to cross-link monomers into dimers (suggesting the
persistence of dimers) at urea concentrations that are
significantly in excess of the half-unfolding concentrations of
urea.
A Hydrophobic Solvent (dioxane) Disrupts Cross-Linking,
Suggesting That Hydrophobic Intersubunit Interactions
Underlie the Persistent Substructure. The bulk of the
structures of HU-A, or HU-B, is stabilized by hydrogen
bonding. The entire NTD is stabilized by helices that are
susceptible to undergoing unfolding through a helix−coil
transition upon disruption of hydrogen bonding by urea. On
the contrary, significant portions of the CTD are stabilized by a
combination of hydrogen bonding and hydrophobic inter-
actions (notwithstanding the IDR regions that undergo
1847
electrostatic interactions with DNA to achieve structure;
otherwise remaining unstructured). We decided to examine
whether a nonpolar, water-miscible, solvent such as 1,4-
dioxane could be used to effect dissociation of HU-B in the
presence of urea. Lanes 5−10 in Figure 6D demonstrate that
the use of 1,4-dioxane reduces the rate of survival of dimeric
and tetrameric species in HU-B in a dose-dependent manner,
in the presence of 3 M urea. In addition, lane 4 demonstrates
that 50% dioxane substantially obviates survival of dimeric and
tetrameric species, even without the presence of urea and to
the same degree seen with use of 50% dioxane in the presence
of 3 M urea. This establishes that the dimeric interface holds
the key to the structural stability of HU, and the survival of a
largely unfolded dimeric state, with destruction of this interface
leading to unfolding and dissociation. In addition, Figure S4
presents data for HU-B refolded through dialysis of urea and
dioxane or dioxane alone. The figure shows that refolding
converts most of the protein into dimeric and tetrameric forms
(with glutaraldehyde adduction reducing staining by Coomas-
sie,30 as known) in both situations.
The Stability of HU Dimers to Thermal Unfolding
Also Reveals a Persistent Hydrophobically Stabilized
Dimeric Interface in Histidine-Tagged HU That Could
Frustrate Subunit Exchange. Circular Dichroism Reveals
Limited (partial) Thermal Unfolding of HU-A and HU-B,
https://doi.org/10.1021/acs.biochem.1c00081
Biochemistry 2021, 60, 1836−1852
Biochemistry pubs.acs.org/biochemistry
with Hints of the Persistence of Substructure Due to
Hydrophobic Interactions Stabilized Further by Heat. The
relative abilities of the three-dimensional structures of HU-AA
and HU-BB to withstand the application of heat were assessed
by spectroscopic means. CD spectra were recorded at
temperatures between 20 and 90 °C during heating of the
protein at a controlled rate (2 °C/min) and also during cooling
of the protein from 90 to 20 °C at the same rate. Panels A and
B of Figure 7 plot the changes in the mean residue ellipticity
(MRE) at 222 nm for HU-AA and HU-BB, respectively, during
heating (red curve) and cooling (blue curve). From the data, it
is clear that the magnitude of the reduction in helical content
with temperature and the temperature eliciting a 50%
reduction in CD signal strength at 222 nm (Tm) are different
for histidine-tagged HU-AA, i.e., 6xHis-HU-AA, and histidine-
tagged HU-BB, i.e., 6x-His-HU-BB. With histidine-tagged HU-
AA, the temperature of half-unfolding is ∼40.6 °C, whereas for
histidine-tagged HU-BB, the temperature of half-unfolding is
∼51.2 °C, establishing that histidine-tagged HU-AA is less
stable to thermal unfolding than histidine-tagged HU-BB,
exactly as observed for unfolding by chemical denaturants. In
addition, it is evident from the spectra that structural changes
in histidine-tagged HU-AA are initiated upon heating at a
temperature of 20 °C, whereas in HU-BB, no changes are seen
below 30 °C. Panels A and B of Figure 7 also support the
observation that HU-BB is more structurally stable than HU-
AA, because it is clear from these figures that, at 90 °C, there is
more residual structure in HU-BB than in HU-AA. In addition,
it must be noted that the amount of residual structure
persisting after thermal unfolding has saturated is greater than
the amount of residual structure seen after chemical unfolding
has saturated in 6 M Gdm.HCl or 8 M urea. This difference
supports the likelihood of the importance of hydrophobic
interactions to the persistence of a stable substructure in HU
that is resistant to unfolding, because the strength of
hydrophobic interactions increases with temperature in the
range of 40−140 °C, in aqueous solution, and this could be
expected to cause it to interfere to a greater extent with
thermal unfolding than with chemical unfolding. Indeed, as
shown in Figure S5, glutaraldehyde cross-linking at 90 °C
reveals the persistence of dimeric and tetrameric forms of HU
even at this temperature. Thus, it is clear that the largely
unfolded forms of HU obtained through heating remain
dimeric to a significant degree, even at 90 °C.
Refolding of HU-AA and HU-BB Occurs from the Partially
Thermally Unfolded State(s), Possibly Owing to Retention of
a Persistent Substructure. Panels A and D of Figure 7 show
the CD spectra of HU-AA and HU-BB, respectively, collected
at 20 °C, i.e., before heating (black curve), then at 90 °C in the
heated state (red curve), and once again at 20 °C after cooling
(blue curve). The data establish some hysteresis between the
heating (unfolding) and cooling (refolding) curves. HU-BB
returns to nearly the same secondary structural content after
cooling, but HU-AA displays a significantly less negative MRE
at 222 nm after cooling, suggesting a efficiency of refolding
somewhat poorer than that of HU-BB. The entire spectra
collected before heating, and after cooling, that are shown in
panels C and D of Figure 7 also support the occurrence of this
hysteresis, in terms of there being a larger difference between
initial and final states in HU-AA than in HU-BB.
Reassociation of Refolded HU-A and HU-B Chains into
DNA-Binding-Competent Dimers from a Partially Thermally
Unfolded State, Possibly Owing to Retention of the
1848
Article
Persistent Substructure. We then examined the protein
refolded from a thermally unfolded state to verify whether it
had formed predominantly dimeric HU. Panels A and B of
Figure 8 establish that the gel filtration chromatographic
elution profiles of unheated HU-A and HU-B, and their
heated−cooled forms, overlap sufficiently for one to conclude
that both homologues are capable of refolding to the dimeric
state. Figure 8C shows that these dimers obtained through
refolding of partially thermally unfolded HU (from 90 °C) are
capable of binding to 4WJ-DNA and displaying an electro-
phoretic mobility shift.
■ DISCUSSION
In E. coli, HU-A and HU-B chains vary in absolute amounts
and relative abundances with the stage of growth of the
bacterial population.6 As a consequence, levels of the
homodimers, HU-AA, and HU-BB, and of the heterodimer,
HU-AB, vary and the heterodimer dominates the stationary
phase.5 Heterodimers are thought to alter the extent of HU-
mediated compaction of DNA through effects upon the
formation of oligomers by HU.13 The quaternary structural
status of the HU-AB heterodimer has been reported to be
somewhere between that of the HU-AA homodimer (which
tends to be predominantly dimeric) and that of the HU-BB
homodimer (which tends to exist as dimers, tetramers,
hexamers, and octamers). It has been shown that chains
consisting of a fusion of HU-B and HU-A (which fold through
intrachain assembly of HU-B and HU-A sequences, into a
simulacrum of the HU-AB heterodimer) do associate to give
rise to a chain dimer that is a simulacrum of an HU
heterotetramer,25 but no further, in support of the dimerization
of HU-AB heterodimers. A possible suggestion, therefore, is
that homodimers of HU-AA keep DNA in a loosely compacted
state, favoring rapid bacterial growth and division, whereas
homodimers and higher-order oligomers of HU-BB cause
DNA to become much more compact, slowing growth and
division as available nutrients are diminished, through a sort of
“hard-braking” action. Following this, even as a culture enters
the stationary phase, it is possible that the availability of a
majority of heterodimers then facilitates a lower degree of
compaction of most parts of a bacterial nucleoid, to keep the
entire genome in a state of readiness for resumption of growth
upon the availability of fresh nutrients.21 The sequences of
HU-A and HU-B and of the regions constituting the N-
terminal domain (NTD) and C-terminal domain (CTD) of
both HU-A and HU-B are shown in Figure S6. From this
figure, it is clear that HU-A and HU-B differ sufficiently in their
sequences (∼69% identity and 80% similarity) for the two
forms of HU to display differences in their physicochemical
and structural−biochemical behavior.
In the existing literature, the mechanism that has been
suggested for formation of heterodimers, at least in vitro,
involves exchange of subunits between HU-AA homodimers
and HU-BB homodimers.5−7 Such exchange can occur, in
principle, through the dissociation of subunits, followed by the
release of folded monomeric subunits into solution, further
followed by the reassembly of such subunits in a heterodimeric
fashion without any interactions occurring between HU-AA
and HU-BB homodimers. However, the exposure of large
tracts of hydrophobic surface area upon subunit dissociation
(note the nature of the association of HU monomers into
dimers, which is evident from Figure 1) makes it extremely
unlikely that any HU chain could exist independently, in
https://doi.org/10.1021/acs.biochem.1c00081
Biochemistry 2021, 60, 1836−1852
Biochemistry pubs.acs.org/biochemistry Article

Figure 9. Proposed mechanism of HU unfolding and dissociation. (A) Ribbon diagram of the HU-AB heterodimer (PDB entry 2O97), shown here
as a representative of the generic HU homodimer, with positions of the N- and C-termini marked, and showing all intersubunit NTD-NTD and
CTD-CTD interactions, as well as hydrophobic residue side chains (shown using a stick model). The region marked IDR represents a schematic of
the unstructured DNA-binding loop that is flexible in the absence of DNA. (B) Schematic of the proposed substantially unfolded form of HU that,
however, remains nominally dimeric due to the persistence of intersubunit CTD-CTD interactions between antiparallel β-sheets engaged in
hydrophobic interactions, prior to any complete unfolding and dissociation. (C) Schematic of likely events resulting from mixing of wild-type (non-
histidine-tagged) HU-AA and HU-BB homodimers, as outlined previously,24 showing the proposed mechanism of spontaneous association into a
heterotetramer hosting subunit exchange before dissociation into heterodimers. (D) Schematic of likely events resulting from mixing of histidine-
tagged HU-AA and HU-BB homodimers, showing the poor association into heterotetramers and poor efficiency of further transformations
(denoted by red cross marks), with the majority of the population remaining unaltered and unable to form heterotetramers. (E) Schematic of likely
events resulting from encounters of incompletely folded, histidine-tagged HU-A and HU-B chains that form heterodimers during co-refolding (or
mutually proximate biosynthesis).

solution, in a folded form akin to that adopted by it within the
folded HU dimer. Therefore, it appears much more likely that
subunit exchange between HU-AA and HU-BB homodimers
occurs through the spontaneous rearrangement of a hetero-
tetrameric intermediate formed by interacting HU-AA and
HU-BB homodimers, into two HU-AB heterodimers, through
a mechanism of swapping of subunits (and without release of
any subunits into solution). Indeed, there is a proposal in the
literature suggesting that HU-AB heterodimer formation
occurs through a heterotetrameric intermediate.23 In addition,
at least in principle and in addition to the two mechanisms
mentioned above, it is also possible that the de novo and direct
association of freshly synthesized, virgin HU-A and HU-B
polypeptide chains produced in close proximity can occur
within the cell. In this third mechanism of formation of HU-AB
heterodimers, chains could associate directly, without first
passing through the stage of forming HU-AA and HU-BB
homodimers, with or without the assistance of some as-yet-
undiscovered chaperone.
As already suggested, examination of the structure of a
canonical HU dimer (e.g., the structure shown in Figure 1D)
and an analysis of the details of the intersubunit chain−chain
associations occurring within such a dimer suggest that the two
1849
subunits in an HU dimer interact extremely intimately and that
there could be problems associated with any spontaneous
dissociation of subunits, without an external trigger. The main
culprit in causing such problems could be the C-terminal
domain (CTD) of the HU chain that engages in a peculiar,
hydrophobicity-driven, intersubunit stacking of antiparallel β-
sheets between subunits. Such a stacking of β-sheets across the
subunit interface occurs in such a manner that it could
conceivably survive both high temperatures (e.g., 90 °C) and
destabilization of hydrogen bonding and electrostatic inter-
actions that otherwise cause unfolding of most of the HU
molecule, as demonstrated in this paper. In comparison with
those of the CTDs, the structures of the NTDs, which are
completely α-helical, appear to be capable of unfolding and
dissociating readily through helix−coil transitions triggered
simply by destabilization of hydrogen bonding and electrostatic
interactions. With the CTDs, on the contrary, it appears that
hydrophobically driven destabilization might also be required,
again as demonstrated in this paper. In terms of chain−chain
engagements amounting to “entanglement”, therefore, it could
probably be said that the helical NTDs of the two HU
monomers (whether in a homodimer or in a heterodimer)
happen to engage with each other without significant
https://doi.org/10.1021/acs.biochem.1c00081
Biochemistry 2021, 60, 1836−1852
Biochemistry pubs.acs.org/biochemistry
entanglement. In contrast, the CTDs of the two subunits
become significantly entangled, essentially because they
approach and traverse each other’s length from opposite
directions, to engage in hydrophobic β-sheet stacking.
The reasoning presented above suggests that dimers might
not be expected to dissociate on their own, to a significant
degree, i.e., that the natural equilibrium between monomers
and dimers of HU greatly favors dimers, both because of
problems associated with dissociation and because the
structures of the monomers themselves (Figure 1A,B) do not
encourage great confidence in their ability to exist
independently without undergoing dimerization (Figure 1D).
Against this background of reasoning, it is useful that the
literature already incorporates mention of a form of HU that
could potentially be blocked with respect to subunit exchange.
Using the prior observation that histidine-tagged HU-A and
HU-B tend to be purified as homodimers,24 we imagined that
these could be blocked with respect to subunit exchange and,
therefore, unable to spontaneously form heterodimers upon
mixing, especially because it has also been reported that HU-A
and HU-B chains do form heterodimers upon being co-
refolded (after first being completely unfolded).24
Therefore, we produced these forms ourselves, to first
confirm that histidine-tagged HU species are indeed purified as
HU-AA and HU-BB homodimers, as reported previously.20 To
do this, we used sensitive and accurate mass spectrometry
experiments to confirm that no chain masses could be detected
that would suggest the formation of any HU-AB heterodimers
in vivo within cells expressing either HU-A or HU-B
polypeptide chains (note that wild-type chains lacking histidine
tags existing in these populations, in heterodimeric association
with overexpressed histidine-tagged HU chains, would have
been clearly discerned in the mass spectra). Having thus
confirmed that histidine-tagged HU-AA and HU-BB homo-
dimers are deficient in their ability to naturally form HU-AB
heterodimers with wild-type HU chains lacking histidine tags
in vivo, we proceeded to examine whether histidine-tagged
HU-AA and HU-BB homodimers are able to spontaneously
form HU-AB heterodimers, following purification, labeling
with fluorophores, and mixing of the two homodimeric
populations in vitro. An insufficient occurrence of FRET,
suggesting the insignificant formation of heterodimers upon
such mixing, confirmed the suspicion that histidine-tagged
HU-AA and HU-BB homodimers are indeed incapable of
spontaneously transforming into a population dominated by
HU-AB heterodimers.
At the same time, examining whether virgin chains of freshly
synthesized histidine-tagged HU-B and non-histidine-tagged
but otherwise N-terminally modified HU-A could potentially
form HU-AB heterodimers in vivo, upon being synthesized at
comparable levels in the proximity of each other, from two
genes located upon the same plasmid remained necessary. To
do this, we produced histidine-tagged HU-B and RFP-tagged
HU-A from neighboring sites on plasmid pETDuet-1. The
result was that HU-AA, HU-BB, and HU-AB populations were
formed to comparable extents, in keeping with the statistical
probabilities of chain−chain encounters. This result served to
establish that heterodimers of HU chains possessing N-
terminal modifications/extensions can indeed be formed
through the de novo interactions of freshly synthesized chains,
even where such chains (i.e., histidine-tagged HU-B and RFP-
tagged HU-A) fail to spontaneously transform into a majority
population of HU-AB heterodimers (i.e., RFP-HU-A+6xHis-
1850
Article
HU-B) upon mixing. Thus far, all results confirmed the
previous report that mixing and co-refolding of fully unfolded
HU-AA and HU-BB produce substantial populations of HU-
AB heterodimers,24 because unfolded and co-refolded chains
could be conceived to behave like freshly synthesized chains
that happen to encounter each other, and assemble, during
synthesis.
The results with the formation of RFP-HU-A+6xHis-HU-B
thus suggested that N-terminal extensions on HU chains do
not sterically prevent the formation of HU-AB heterodimers
through direct chain−chain contacts; instead, they only appear
to prevent conversion of preassembled HU-AA and HU-BB
homodimers into HU-AB heterodimers upon mixing. There
could be two reasons for this. (1) There could be a problem
with the natural separation of the two chains in an HU dimer,
as suggested by panels A and B of Figure 9, which show
schematically that residual structure (in the form of
intersubunit β-sheet stacking contacts) could survive global
chain unfolding. Alternatively, (2) the presence of the N-
terminal extension could sterically affect the ability of HU-AA
homodimers to interact with HU-BB homodimers, as
suggested by Figure 9D, which schematically shows the N-
terminal extension preventing formation of a heterotetrameric
intermediate, critical for exchange of subunits without release
of monomers into solution. In this paper, we have provided
evidence in support of both of these possibilities. In addition,
we have also provided evidence for the formation of HU-AB
heterodimers through direct chain−chain contacts following
chain biosynthesis in vivo, as suggested by Figure 9E, which
schematically shows the formation of substantial populations of
HU-AB heterodimers through such contacts. Figure 9C
schematically shows the spontaneous formation of HU-AB
heterodimers by HU-AA and HU-BB homodimers lacking the
histidine tag; this is not something that we have shown in this
paper, but rather something for which evidence already exists
in the literature.5−7,24
To examine whether there is a problem with the natural
separation of chains, potentially owing to the persistence of a
substructure at HU’s subunit interface (which is further
stabilized by the N-terminal extension), as suggested by panels
A and B of Figure 9, we proceeded to destabilize purified HU-
AA and HU-BB homodimers. We examined the ease with
which the constituent chains of these homodimers tend to be
unfolded and separated from each other, holding separation of
chains to be a necessary prerequisite to (and rate determinant
for) any formation of HU-AB heterodimers from histidine-
tagged HU-AA and HU-BB homodimers. Through detailed
studies, we were able to confirm that helical structures are
unfolded with ease during chemical or thermal unfolding.
However, we also discovered that thermal destabilization does
not achieve complete unfolding of chains, with chemical
unfolding also achieving complete unfolding only at extremely
high denaturant concentrations. Both chemical and thermal
unfolding data, therefore, provided strong evidence of the
persistence of β-sheet-based substructure(s) in histidine-tagged
HU-AA and HU-BB that resist unfolding and chain
dissociation. The persistence of such substructure(s) was
then established to be a correlate of chains getting cross-linked
by glutaraldehyde even under extreme conditions of thermal or
chemical destabilization that otherwise elicit near-global chain
unfolding. This set of results demonstrated that HU chains are
capable of remaining “nominally dimeric” despite near-global
chain unfolding, due to the persistence of substructures
https://doi.org/10.1021/acs.biochem.1c00081
Biochemistry 2021, 60, 1836−1852
Biochemistry pubs.acs.org/biochemistry
engaged in hydrophobically stabilized, intersubunit contacts. In
other words, the data clearly established that there is a
significant problem in the dissociation of chains in an HU
dimer, a finding with important implications for subunit
exchange through the alternative route of association of
homodimers into a heterotetrameric intermediate, critical to
exchange of subunits without release of monomers into
solution.
Our suspicion that the persistent substructure under
question must be the hydrophobically stabilized, stacked pair
of antiparallel β-sheets (one contributed by each monomer)
formed between the CTDs of the two chains in a HU dimer
was then further confirmed by the finding that increasing of the
nonpolarity of the solvent (through addition of dioxane)
significantly relieves the persistence of glutaraldehyde-based
cross-linking of chains under denaturing conditions. This
finding suggested that nonseparation of chains could be one
reason for the nonformation of HU-AB heterodimers by
histidine-tagged HU-A and HU-B chains, as clearly already
evidenced by FRET experiments. It may be noted that the
modification of the N-terminus of HU through the addition of
a 6xHis tag, or through the fusion of an additional RFP-based
protein domain, effectively removes the charged α-amino
group from its location at the N-terminus of wild-type HU,
very close to the stacked β-sheets, and displaces the N-terminal
charge to a distal location placed far from the stacked
antiparallel β-sheets. As mentioned in the introduction, the α-
amino group upon the N-terminal methionine in any HU
chain, M1, normally engages in intersubunit charge−charge
interactions with an aspartate group (D40) across the subunit
interface. Quite obviously, therefore, displacement of this
charge to a distal location would destroy an important
electrostatic interaction that could ordinarily serve to
destabilize the intersubunit stacking of β-sheets (located
close to the N-terminus). It is possible, therefore, that an N-
terminal modification reduces the destabilizing effect(s) of this
charge, with this resulting in a greater resistance to dissociation
of the stacked β-sheets.
In addition to such a stabilization of the subunit interface
through ablation of the M1−D40 interaction in N-terminally
modified HU (especially in forms of HU possessing an N-
terminal extension), the very location of the N-terminus could
cause the 6xHis tag (or, indeed, any other N-terminal
extension, such as a fused RFP domain, or any other N-
terminal modification) to sterically interfere with the
association of HU-AA and HU-BB homodimers in a
heterotetrameric intermediate that is critical to swapping of
subunits without, however, sterically interfering with the
formation of HU-AB heterodimers per se. This suggests that
the interaction between HU-AA and HU-BB homodimers to
form the heterotetrameric intermediate must involve a region
of the surfaces of these dimers that contains the N-terminus,
such that the presence of any additional residues at the N-
terminus disallows formation of the heterotetramer.
■ ASSOCIATED CONTENT
* Supporting Information
sı
The Supporting Information is available free of charge at
https://pubs.acs.org/doi/10.1021/acs.biochem.1c00081.
In vitro mixing of separately produced 6xHis-RFP-HU-A
and 6xHis-HU-B for comparison with in vivo mixing
attempted using the pETDuet-1 vector producing RFP-
1851
Article
HU-A and 6xHis-HU-B (Figure S1), secondary structure
changes in HU-A and HU-B upon incubation in
different molar concentrations of urea, monitored
using CD spectroscopy (Figure S2), control experiments
showing an EMSA for binding of 6xHis-HU-A and
6xHis-HU-B to 4WJ-DNA, as adjunct to EMSAs for
DNA binding by HU refolded from urea-unfolded states
(Figure S3), glutaraldehyde cross-linking to assess
refolding of HU into a dimer from urea and/or 1,4-
dioxane (Figure S4), glutaraldehyde cross-linking to
assess the persistence of dimeric structure in histidine-
tagged HU-B before and after thermal unfolding (at 90
°C) (Figure S5), and sequences of wild-type HU-A and
HU-B lacking histidine tags (Figure S6) (PDF)
Accession Codes
The UniProt accession number for HU-A is P0ACF0 and for
HU-B is P0ACF4. The UniProt accession number for the HU-
AB heterodimer is P0ACF0.
■
AUTHOR INFORMATION
Corresponding Author
Purnananda Guptasarma − Centre for Protein Science, Design
and Engineering (CPSDE), Department of Biological
Sciences, Indian Institute of Science Education and Research
(IISER) Mohali, Punjab 140306, India; orcid.org/0000-
0002-4801-3180; Email: guptasarma@iisermohali.ac.in
Authors
Kanika Arora − Centre for Protein Science, Design and
Engineering (CPSDE), Department of Biological Sciences,
Indian Institute of Science Education and Research (IISER)
Mohali, Punjab 140306, India
Bhishem Thakur − Centre for Protein Science, Design and
Engineering (CPSDE), Department of Biological Sciences,
Indian Institute of Science Education and Research (IISER)
Mohali, Punjab 140306, India
Arpita Mrigwani − Centre for Protein Science, Design and
Engineering (CPSDE), Department of Biological Sciences,
Indian Institute of Science Education and Research (IISER)
Mohali, Punjab 140306, India
Complete contact information is available at:
https://pubs.acs.org/10.1021/acs.biochem.1c00081
Author Contributions
†
K.A. and B.T. contributed equally to this work.
Funding
K.A. thanks the University grant commission, Government of
India, for a doctoral fellowship. B.T. thanks the Council of
Scientific and Industrial Research (CSIR), New Delhi, for a
doctoral fellowship. A.M. thanks the Department of bio-
technology, Government of India, for a doctoral fellowship.
P.G. thanks the Ministry of Human Resource Development
(MHRD), Government of India, for a Centre of Excellence
Grant (MHRD-14-0064) in Protein Science, Design and
Engineering.
Notes
The authors declare no competing financial interest.
■
REFERENCES
(1) Dame, R. T. (2005) The role of nucleoid-associated proteins in
the organization and compaction of bacterial chromatin. Mol.
Microbiol. 56, 858−870.
https://doi.org/10.1021/acs.biochem.1c00081
Biochemistry 2021, 60, 1836−1852
Biochemistry pubs.acs.org/biochemistry Article

(2) Stuger, R., Woldringh, C. L., van der Weijden, C. C., Vischer, N. (23) Garnier, N., Loth, K., Coste, F., Augustyniak, R., Nadan, V.,
O., Bakker, B. M., van Spanning, R. J., Snoep, J. L., and Weterhoff, H. Damblon, C., and Castaing, B. (2011) An alternative flexible
V. (2002) DNA supercoiling by gyrase is linked to nucleoid conformation of the E. coli HUbeta(2) protein: structural, dynamics,
compaction. Mol. Biol. Rep. 29, 79−82. and functional aspects. Eur. Biophys. J. 40, 117−129.
(3) Kano, Y., Osato, K., Wada, M., and Imamoto, F. (1987) Cloning (24) Pellegrini, O., Oberto, J., Pinson, V., Wery, M., and Rouviere-
and sequencing of the HU-2 gene of Escherichia coli. Mol. Gen. Genet. Yaniv, J. (2000) Overproduction and improved strategies to purify the
209, 408−410. threenative forms of nuclease-free HU protein. Biochimie 82, 693−
(4) Kano, Y., Yoshino, S., Wada, M., Yokoyama, K., Nobuhara, M., 704.
and Imamoto, F. (1985) Molecular cloning and nucleotide sequence (25) Arora, K., Thakur, B., Gupta, A., and Guptasarma, P. (2021)
of the HU-1 gene of Escherichia coli. Mol. Gen. Genet. 201, 360−362. HU-AB simulacrum: Fusion of HU-B and HU-A into HU-B-A, a
(5) Rouviere-Yaniv, J., and Kjeldgaard, N. O. (1979) Native functional analog of the Escherichia coli HU-AB heterodimer.
Escherichia coli HU protein is a heterotypic dimer. FEBS Lett. 106, Biochem. Biophys. Res. Commun. 560, 27−31.
297−300. (26) Arora, K., Mangale, S. S., and Guptasarma, P. (2015) Single
(6) Claret, L., and Rouviere-Yaniv, J. (1997) Variation in HU cell-level detection and quantitation of leaky protein expression from
composition during growth of Escherichia coli: the heterodimer is any strongly regulated bacterial system. Anal. Biochem. 484, 180−182.
required for long term survival. J. Mol. Biol. 273, 93−104. (27) Ghosh, S., and Grove, A. (2004) Histone-like protein HU from
(7) Ali Azam, T., Iwata, A., Nishimura, A., Ueda, S., and Ishihama, A. Deinococcus radiodurans binds preferentially to four-way DNA
(1999) Growth phase-dependent variation in protein composition of junctions. J. Mol. Biol. 337, 561−571.
the Escherichia coli nucleoid. J. Bacteriol. 181, 6361−6370. (28) Thakur, B., Arora, K., Gupta, A., and Guptasarma, P. (2021)
(8) Kar, S., Choi, E. J., Guo, F., Dimitriadis, E. K., Kotova, S. L., and The DNA-binding protein HU is a molecular glue that attaches
bacteria to extracellular DNA in biofilms. J. Biol. Chem. 296, 100532.
Adhya, S. (2006) Right-handed DNA supercoiling by an octameric
(29) Merzlyakov, M., and Hristova, K. (2008) Forster resonance
form of histone-like protein HU: modulation of cellular transcription.
energy transfer measurements of transmembrane helix dimerization
J. Biol. Chem. 281, 40144−40153.
energetics. Methods Enzymol. 450, 107−127.
(9) Castaing, B., Zelwer, C., Laval, J., and Boiteux, S. (1995) HU
(30) Tiwari, P., Kaila, P., and Guptasarma, P. (2019) Understanding
protein of Escherichia coli binds specifically to DNA that contains anomalous mobility of proteins on SDS-PAGE with special reference
single-strand breaks or gaps. J. Biol. Chem. 270, 10291−10296. to the highly acidic extracellular domains of human E- and N-
(10) Kamashev, D., and Rouviere-Yaniv, J. (2000) The histone-like cadherins. Electrophoresis 40, 1273−1281.
protein HU binds specifically to DNA recombination and repair
intermediates. EMBO J. 19, 6527−6535.
(11) Swinger, K. K., Lemberg, K. M., Zhang, Y., and Rice, P. A.
(2003) Flexible DNA bending in HU-DNA cocrystal structures.
EMBO J. 22, 3749−3760.
(12) Hammel, M., Amlanjyoti, D., Reyes, F. E., Chen, J. H., Parpana,
R., Tang, H. Y., Larabell, C. A., Tainer, J. A., and Adhya, S. (2016)
HU multimerization shift controls nucleoid compaction. Sci. Adv. 2,
No. e1600650.
(13) Swinger, K. K., and Rice, P. A. (2004) IHF and HU: flexible
architects of bent DNA. Curr. Opin. Struct. Biol. 14, 28−35.
(14) Ramstein, J., Hervouet, N., Coste, F., Zelwer, C., Oberto, J., and
Castaing, B. (2003) Evidence of a thermal unfolding dimeric
intermediate for the Escherichia coli histone-like HU proteins:
thermodynamics and structure. J. Mol. Biol. 331, 101−121.
(15) Wada, M., Kano, Y., Ogawa, T., Okazaki, T., and Imamoto, F.
(1988) Construction and characterization of the deletion mutant of
hupA and hupB genes in Escherichia coli. J. Mol. Biol. 204, 581−591.
(16) Huisman, O., Faelen, M., Girard, D., Jaffe, A., Toussaint, A.,
and Rouviere-Yaniv, J. (1989) Multiple defects in Escherichia coli
mutants lacking HU protein. J. Bacteriol. 171, 3704−3712.
(17) Dri, A. M., Rouviere-Yaniv, J., and Moreau, P. L. (1991)
Inhibition of cell division in hupA hupB mutant bacteria lacking HU
protein. J. Bacteriol. 173, 2852−2863.
(18) Bensaid, A., Almeida, A., Drlica, K., and Rouviere-Yaniv, J.
(1996) Cross-talk between topoisomerase I and HU in Escherichia
coli. J. Mol. Biol. 256, 292−300.
(19) Pinson, V., Takahashi, M., and Rouviere-Yaniv, J. (1999)
Differential binding of the Escherichia coli HU, homodimeric forms
and heterodimeric form to linear, gapped and cruciform DNA. J. Mol.
Biol. 287, 485−497.
(20) Zhang, Y., and Skolnick, J. (2005) TM-align: a protein structure
alignment algorithm based on the TM-score. Nucleic Acids Res. 33,
2302−2309.
(21) Guo, F., and Adhya, S. (2007) Spiral structure of Escherichia
coli HUalphabeta provides foundation for DNA supercoiling. Proc.
Natl. Acad. Sci. U. S. A. 104, 4309−4314.
(22) Thakur, B., Gupta, A., and Guptasarma, P. (2021) A novel
protein-engineered dsDNA-binding protein (HU-Simulacrum) in-
spired by HU, a nucleoid-associated DNABII protein. Biochem.
Biophys. Res. Commun. 534, 47−52.

1852 https://doi.org/10.1021/acs.biochem.1c00081
Biochemistry 2021, 60, 1836−1852