Microbes respond to osmotic challenges in diverse, variable, and extreme natural environments
Exposure to diverse environments is a hallmark of microbial life. Microbes are everywhere;
collectively, microbes experience everything. They live inside and outside eukaryotic
hosts, in soil, water, and air at diverse planetary sites. They may exist as individuals
(planktonic growth) or aggregates, and form biofilms on biotic and abiotic surfaces.
Some survive gradual or abrupt, temporal or spatial transitions between different
environments.
Our understanding of microbial responses to osmotic challenges is based on intensive
studies of representative bacteria, archaea, and eukaryotic microbes. This Perspective
focuses on bacterial responses to osmotic challenges. Among the representative bacteria
for which the osmotic stress response is well characterized, Escherichia coli lives
in terrestrial and aquatic environments as well as in the meninges and the intestinal
and urinary tracts of mammals. Bacillus subtilis and Corynebacterium glutamicum are
soil bacteria (C. glutamicum is also used to manufacture fine chemicals), and Halomonas
elongata was isolated from a solar saltern (Wood, 2011a). Some bacteria can survive
in pure water and grow at a water activity (aW) near 1, many thrive within human tissues
(e.g., human blood, aW of 0.995) or in seawater (aW of 0.98), whereas others can only
inhabit hypersaline environments with water activities as low as 0.75. Further examples,
discussed below, illustrate the range of environments and environmental variations
to which bacteria respond.
Bacteria are bounded by semipermeable cytoplasmic membranes, often including aquaporins.
Most are also surrounded by a rigid, elastic, and porous cell wall (the murein or
peptidoglycan layer) that determines cell shape. The cell wall of Gram-negative bacteria
(such as E. coli) is bounded by an outer lipid membrane that includes porins like
those of mitochondria. The area between the outer and cytoplasmic membranes is denoted
the “periplasm.” The integrity and hydration of the cell and its compartments are
dictated by their solute contents and the osmotic pressures of their environments
(discussed in Altendorf et al., 2009). A decrease in external osmotic pressure causes
water influx and swelling or even lysis, whereas an increase in external osmotic pressure
causes water efflux and dehydration. Water fluxes simultaneously, and almost instantaneously,
perturb many cellular properties. These include cell volume (or the relative volumes
of the cytoplasm and periplasm); turgor pressure; cell wall strain; and cytoplasmic
membrane tension; as well as individual uncharged solute, salt ion, and biopolymer
concentrations. Cells exposed consistently to a very high osmotic pressure must maintain
correspondingly high cytoplasmic solute concentrations. Evidence suggests that the
regulation of cytoplasmic composition and hydration is a key objective of cellular
homeostasis (Wood, 2011b).
Common themes emerged as researchers characterized the osmoadaptive mechanisms of
bacteria representing diverse phylogenetic groups (Wood, 2011a, and references cited
therein). Cells respond to variations in external osmotic pressure by accumulating
or releasing solutes, thereby attenuating water fluxes. Those solutes include inorganic
ions (often K+), and organic molecules denoted “osmolytes” (Fig. 1). The latter are
selected to minimally perturb cellular functions, even after accumulating to high
(up to molar) concentrations. In turn, organisms have adapted to tolerate osmoregulatory
solute accumulation. In the extreme, some halophiles accumulate KCl to molar concentrations,
and their proteins function only in high salt environments. Osmoregulatory solutes
accumulate via active transport or synthesis if the osmotic pressure rises and are
released via mechanosensitive channels if the osmotic pressure falls. Multiple enzymes,
transporters, and channels with redundant functions and specificities mediate solute
accumulation and release from each organism (e.g., Fig. 2). The abundance of most
osmoregulatory systems is controlled transcriptionally (Altendorf et al., 2009; Krämer,
2010). Translational regulation, mediated by small regulatory RNAs, is emerging as
an important determinant of bacterial cell wall structure that may also influence
the levels of osmoregulatory systems.
Figure 1.
Osmolytes. These compounds accumulate in E. coli in high osmotic pressure media (Altendorf
et al., 2009).
Care must be taken to differentiate osmotic stress from parallel, solute-specific
effects that dominate particular environments. For example, bacteria inhabiting seawater
face a higher osmotic pressure than those inhabiting most freshwater environments.
Salts predominate in seawater, and marine organisms simultaneously face both a high
osmotic pressure and a high Na+ concentration. Na+ fluxes are also implicated in pH
homoeostasis. Distinctions are also drawn between bacteria adapted to environments
with extreme and stable osmotic pressures (e.g., sea water, salt lakes) and those
experiencing osmotic pressure variations (e.g., those inhabiting estuarine waters
or colonizing mammalian intestinal tracts).
What cellular systems limit bacterial cell and population growth rates under osmotic
stress? How are osmotic stress responses orchestrated over time and space?
Solute accumulation powerfully stimulates bacterial growth at high osmotic pressure,
and solute release allows cells to survive osmotic downshocks. Thus, studies of bacterial
osmoregulation have focused on the enzymes, transporters, and channels mediating solute
accumulation and release (Krämer, 2010; Kung et al., 2010; Wood, 2011b) (Fig. 2).
However, we do not fully understand how increasing osmotic pressure would limit bacterial
cell or population growth in the absence of solute accumulation.
Figure 2.
Osmoregulatory systems of E. coli. In high osmotic pressure environments, solutes
accumulate in E. coli via synthesis (glutamate, trehalose, glycine betaine) or transport
from the external medium (e.g., others shown in Fig. 1; Altendorf et al., 2009). K+-H+
symporter Trk and P-type ATPase Kdp mediate K+ uptake. Major facilitator superfamily
member ProP, ABC transporter ProU, and betaine-carnitine-choline family members BetT
and BetU mediate organic osmolyte uptake. ProP and ProU are similarly broad in substrate
specificity, whereas BetT is choline specific (Murdock et al., 2014) and BetU is betaine
specific. Mechanosensitive channels, including MscS and MscL, release solutes from
the cytoplasm of osmotically downshocked bacteria. Aquaporin AqpZ exacerbates osmotic
stress by accelerating transmembrane water flux. BetT and BetU are homologues of BetP
from C. glutamicum, whereas ProU is a homologue of OpuA from L. lactis (see Fig. 3).
The evolution of bacterial cell and population size, protonmotive force, DNA replication,
protein synthesis, and solute content were documented both after osmotic shifts and
during steady-state culture of E. coli at various osmotic pressures, in the absence
or presence of osmoprotective solutes (Wood, 1999; Cayley and Record, 2004; Altendorf
et al., 2009). Such studies revealed that the population growth rate is directly proportional
to cytoplasmic hydration, and that accumulating solutes differentially affect cellular
rehydration and population growth. K+ glutamate accumulation partially rehydrates
cells and perturbs protein–nucleic acid interactions. It thereby offsets the impact
of increased macromolecular crowding on cellular processes but does not restore growth
to its pre-stress rate. In contrast, organic osmolytes rehydrate the cytoplasm and
restore growth to an extent that correlates with their preferential exclusion from
biopolymer surfaces (Cayley and Record, 2004) (discussed further below).
In contrast to our understanding of other stresses (e.g., oxidative stress; Imlay,
2013), we don’t know what cellular properties or processes limit population growth
rate when cells dehydrate. It was widely assumed that osmoregulation is necessary
because turgor pressure is essential for cell wall expansion and cell growth. However,
evidence contradicts that assumption (e.g., E. coli; Cayley and Record, 2003; Rojas
et al., 2014), and other cellular properties may be critical. Single-cell imaging
techniques are now elucidating how osmotic stress affects cell growth and development
(e.g., Pilizota and Shaevitz, 2013; Rojas et al., 2014), the composition and biophysical
properties of the cytoplasm and cell membranes (Mika and Poolman, 2011; Sochacki et
al., 2011; Wood, 2011b; Sévin and Sauer, 2014), and the subcellular locations of osmoregulatory
systems (Romantsov et al., 2010).
Respiration, the synthesis of precursor metabolites, replication, transcription, and
translation are obvious candidates for growth rate limitation (Wood, 1999). Individual
strains within a species vary widely in osmotic stress tolerance (e.g., Kunin et al.,
1992; Murdock et al., 2014). Analysis of new strains obtained via directed evolution
and of naturally occurring variants may reveal what modifications, to what systems,
extend the range of cytoplasmic hydration tolerated by an organism.
The application of high throughput “omic” technologies and cell sorting are also opening
new avenues of investigation. Such tools can elucidate the orchestration of osmoadaptive
mechanisms after an osmotic shift or during steady-state growth at various osmotic
pressures (Withman et al., 2013, and other studies cited therein). They can also show
how osmotic stress affects phenotypic variation within a microbial population. Analyses
of bacterial community composition suggest that the bacterial lineages inhabiting
marine and freshwater ecosystems are phylogenetically distinct, and that the capacity
for osmoadaptation may be a primary determinant of that divergence (Walsh et al.,
2013). Organisms adapted to a stable, high salinity marine environment may face particular
barriers when transitioning to a more variable estuarine or fresh water environment.
Such studies have relied heavily on genomic comparisons and annotations. Key tests
of these ideas may be devised by combining physiological experiments with phylogenetic
approaches.
How do proteins detect and respond to osmotic pressure variations?
Membrane proteins implicated in bacterial osmoregulation became the paradigms for
the study of osmosensing because they retain osmotic pressure–dependent activities
after purification and reconstitution in proteoliposomes (Poolman et al., 2004). Proteoliposome-based
studies provided critical evidence that mechanosensitive channels and osmosensing
transporters detect and respond to osmotic pressure changes in their phospholipid
environments, without input from other cellular components. Studies of bacterial systems
provided seminal evidence that mechanosensitive channels open in response to forces
exerted by the lipid bilayer (Teng et al., 2015). Analyses of MscL and MscS continue
to elucidate mechanosensory mechanisms (Iscla and Blount, 2012; Naismith and Booth,
2012). The signal(s) to which osmosensing transporters respond remains less clear,
however.
Osmosensing transporters.
ProP of E. coli, BetP of C. glutamicum, and OpuA of Lactococcus lactis serve as paradigms
for the study of osmosensing (Wood, 2011b) (Fig. 3). They represent different phylogenetic
groups and energy-coupling mechanisms. ProP is a proton symporter and a member of
the major facilitator superfamily, BetP is a Na+ symporter and a member of the betaine-choline-carnitine
transporter family, and OpuA is an ATP-hydrolyzing ATP-binding cassette (ABC) transporter.
Available data suggest that each is similar in structure and transport mechanism to
its paralogues that are not osmosensors.
Figure 3.
Structures of osmosensory transporters. The structures of BetP from C. glutamicum,
ProP from E. coli, and OpuA from L. lactis are illustrated. The protein backbones
are colored according to amino acid side-chain polarity unless otherwise indicated:
red for acidic residues Asp and Glu; blue for basic residues Arg, Lys, and His; green
for polar residues Ser, Thr, Cys, Asn, and Gln; and yellow for nonpolar residues.
BetP: a crystal structure of trimeric BetP (Protein Data Bank [PDB] accession no.
2WIT) as viewed from the cytoplasm (A) and of a single BetP subunit as viewed from
the membrane (B). In A, the three BetP subunits are colored black, gray, and by amino
acid. B shows a single subunit with residues from the N terminus through the end of
transmembrane helix II as strands and residues 313–324 as a trace to reveal glycine
betaine (space-filling, CPK coloring) within the substrate-binding site. ProP: a homology
model of a ProP monomer (PDB accession no. 1Y8S) (C and D, residues 4–236 and 246–452
of the 500-residue ProP protein) and a nuclear magnetic resonance (NMR) structure
of the C-terminal domain of ProP (PDB accession no. 1R48) (E, residues 468–497 of
the 500-residue ProP protein). ProP is viewed from the membrane with the cytoplasmic
surface down (C) and from the cytoplasm (D). The arrow in C marks the position of
a substrate analogue in the crystal structure of homologue LacY (PDB accession no.
1PV7). The stars in C and D mark the C-terminal amino acid of the model. (E) The structure
of a homodimeric peptide corresponding to residues 468–497 of ProP, determined by
NMR spectroscopy (PDB accession no. 1R48). This antiparallel α-helical coiled-coil
and transmembrane helix XII contribute to the ProP dimer interface in vivo (Wood,
2011b). OpuA: a schematic representation of transporter OpuA (F) and the structure
of periplasmic-binding protein domain OpuAC (G). In F, two cytoplasmic ATP-binding
OpuAA subunits, including C-terminal cystathionine-β-synthase (CBS) domains, are blue.
Two transmembrane OpuAB domains and the contiguous substrate-binding OpuAC domains
are yellow. G shows a crystal structure of domain OpuAC (yellow; PDB accession no.
3L6H) in complex with glycine betaine (spheres). The binding pocket includes three
Trp residues (W330, W377, and W484, shown as sticks) that coordinate the trimethylammonium
group of glycine betaine.
The rate of osmolyte uptake via each transporter (A) is a sigmoid function of the
osmotic pressure (Π) or osmolality (Π/RT, where R is the gas constant and T is the
temperature). Such data have been fit to an arbitrary relationship that implies no
particular activation mechanism:
(1)
A
=
A
max
[
1
+
exp
(
−
(
Π
-
Π
1
/
2
)
/
RTB
)
]
-
1
,
where Amax is the asymptotic uptake rate, B is a constant inversely proportional to
the slope of the response curve, and Π1/2/RT is the osmolality at which activity is
half-maximal. In this relationship, Π1/2/RT can be replaced with any property that
varies in parallel with the osmolality (e.g., the calculated concentration of a luminal
solute in proteoliposomes). Proteoliposome data have also been fit to the Hill equation:
(2)
A
=
A
max
[
1
+
(
K
ion
n
/
[
Ion
]
n
)
]
−
1
,
where Kion is the ion concentration required to attain half-maximal activity, and
n is a constant related to the slope of the curve (Mahmood et al., 2006).
To understand osmosensing, we must learn what cellular property is detected by an
osmosensor and understand how variations to that property modulate osmosensor structure
and function (Wood, 1999). In principle, an osmosensor would trigger a homeostatic
response upon detecting deviations from a “set point” of such a critical property.
Experiments performed with cells and proteoliposomes ruled out turgor pressure and
membrane strain as determinants of osmosensing transporter activity (Poolman et al.,
2004). Proteoliposome systems were then exploited to further assess the impacts of
the external and luminal solvents on the activity of each osmosensing transporter.
Merits and liabilities of proteoliposome systems.
The interpretation of proteoliposome data are supported by evidence that secondary
transporters ProP and BetP reconstitute predominantly with their cytoplasmic surfaces
facing the lumen, and the direction of transport is determined by an imposed ion motive
force. Studies of ABC transporter OpuA exploit the fact that the direction of transport
can be controlled by supplying ATP in either the external or the luminal medium (Wood,
2011b). To date, functional tests have been the primary indicators of osmosensing
(i.e., solute uptake assays as opposed to spectroscopic indicators of transporter
conformation). The requirements to maintain the membrane permeability barrier and
to meet energy requirements for transport restrict the range of luminal and external
solvent compositions accessible for these studies. The Amax values obtained with proteoliposomes
are variable because transporter purification, reconstitution, and solute loading
are intrinsically variable procedures. A recent comparison of the molecular activities
of BetP in cells and proteoliposomes indicated that only 2.4% of BetP molecules in
proteoliposomes were active (Maximov et al., 2014), reinforcing the need for careful
interpretation of proteoliposome data. In contrast, Π1/2/RT, B, and Kion values are
independent of transporter quantity, more reproducible, and hence presumed to be more
reliable indicators.
Solvent effects on biopolymer structures.
Current knowledge of solvent effects on biopolymer structures provides a useful context
for the analysis of osmosensory mechanisms. Soluble proteins and DNA have been the
primary foci of such studies, which explore the thermodynamic nonideality inherent
to physiological systems and their models (Record et al., 1998b, 2013). Solvent additives
can affect biopolymer processes by binding as ligands at specific sites, via preferential
interactions with buried or exposed biopolymer surfaces (Hofmeister effects, involving
both uncharged and charged solutes) and via conformation-specific, Coulombic interactions
with fixed biopolymer charges (charged solutes only). Thus, solutes may act individually
(high affinity ligand binding at one or a few specific sites) and/or collectively
(weak interactions at many sites).
Collective solute effects modulate the equilibrium constant (K) for any process that
changes the amount of biopolymer surface interacting with a solute. The magnitudes
and functional forms of these collective effects are determined by the nature of the
solute excluded from or concentrated at the biopolymer surface (particularly whether
it is charged or uncharged) and of the exposed or buried biopolymer surface (e.g.,
that of a high charge density polyelectrolyte like DNA or a low to no charge density
biopolymer like a typical protein). When salt concentrations are low, the collective
effects are primarily Coulombic (salts weaken charge–charge interactions). When salt
or uncharged solute concentrations are high, their contributions to Hofmeister effects
become dominant (e.g., Fig. 4; Record et al., 2013). The salt concentration ranges
over which Coulombic and Hofmeister effects dominate for DNA and protein processes
differ because DNA is a polyelectrolyte.
Figure 4.
Effects of Hofmeister salts on protein unfolding. The effects of low and high concentrations
of salts spanning the Hofmeister series on unfolding of the lac repressor DNA binding
domain at 40°C. Kobs is the unfolding equilibrium constant in the presence of salt
at concentration X, and Kobs,ref is the reference equilibrium constant in low salt
buffer. At low salt concentration, all salts exert similar stabilizing effects. These
are Coulombic in origin and vary nonlinearly with salt concentration. At high salt
concentration, different salts exhibit a wide range of stabilizing (e.g., (NH4)2SO4,
KF) to destabilizing (guanidinium HCl (GuHCl)) effects, which are linear in salt concentration
and follow the traditional Hofmeister series. The slopes of these high salt linear
regions (m-value/RT) correlate with the magnitude and chemical composition of the
protein surface that is exposed to the solution in unfolding and the chemical properties
of the salt (the places of the cation and anion in the Hofmeister series). Fitted
curves allow a separation of Coulombic and Hofmeister effects of these salts. Adapted
from Fig. 6 of Record et al. (2013) with permission of The Royal Society of Chemistry
(http://dx.doi.org/10.1039/C2FD20128C).
If a process changes the amount of uncharged or weakly charged biopolymer surface
exposed to a preferentially interacting solute, then the free energy (or the logarithm
of the equilibrium constant, K) for that process is a linear function of the solute
concentration with a proportionality constant (the thermodynamic m-value) that reflects
the properties of the solute and the magnitude of the exposed or buried biopolymer
surface. Such effects are very weak at low solute concentration. In contrast, if a
process changes the amount of a (polyanionic) DNA surface exposed to ionic solutes,
then the logarithm of the equilibrium constant (K) varies with a power of the logarithm
of the ion concentration. Such Coulombic effects are large even at low salt concentrations.
The latter analysis supersedes the Debye–Hückel approximation, based on ionic strengths
calculated as a function of ion concentrations and valencies, which has much more
limited application. Principles governing protein–membrane interactions have not been
analyzed in this way, but interactions of proteins with polyanionic membrane surfaces
can be expected to share characteristics with protein–DNA interactions.
The principles outlined above were established primarily with in vitro systems. There
is also evidence that cytoplasmic solutes collectively influence cellular processes,
particularly as osmotic pressure changes alter cytoplasmic hydration (Record et al.,
1998a,b). Small cytoplasmic solutes (e.g., K+, glutamate, and other metabolites) are
preferentially excluded from nonpolar biopolymer surfaces that become exposed in unfolding.
Increasing concentrations of these solutes will favor conformational changes that
bury nonpolar surfaces (Record et al., 2013). At the same time, condensation of K+
as a DNA counterion impedes processes involving protein–DNA interactions. In addition,
increased concentrations of cytoplasmic biopolymers favor folding, especially if folding
is coupled to oligomerization, by an excluded volume effect (sometimes denoted as
“macromolecular crowding”; Cayley and Record, 2004).
Conceptual framework for the analysis of osmosensing.
In proteoliposomes, osmosensory transporters become active as luminal solute concentrations
approach 0.5 M. This suggests that both Coulombic and Hofmeister effects may participate
in transporter activation and that resolution of Coulombic and Hofmeister effects
will be challenging (c.f. Fig. 4). The sigmoid relationship between A and Π1/2/RT
implies that transporter molecules are systematically converted from an inactive to
an active conformation as the osmolality increases:
(3)
Transporter
I
⇔
Transporter
A
.
If so, the fraction of transporter active at a particular osmolality may be represented
by:
(4)
f
=
A
/
A
max
,
where A is the initial rate of substrate uptake at a given osmolality, and Amax is
the asymptotic initial rate approached at high osmolality. Then the equilibrium constant
K for this transition at a particular osmolality is:
(5)
K
=
f
/
(
1
−
f
)
.
If the activating conformational change were triggered only by solute exclusion from
nonpolar transporter surfaces that were exposed in the inactive and buried in the
active transporter (a Hofmeister effect), the logarithm of the equilibrium constant
K would be expected to vary linearly with the solute concentration X (Record et al.,
2013):
(6)
ln
K
=
ln
K
0
−
(
m
/
RT
)
X
or
K
=
K
0
exp
(
−
(
m
/
RT
)
X
)
.
In this equation, K0 would be the equilibrium constant at X = 0, where the transporter
activity is undetectably small, and m/RT would be a thermodynamic parameter (the thermodynamic
m-value) characteristic of the solute and the conformational change. To obtain m and
K0 for transporter activation, values of A at each X would be fit to the following
combined relationship:
(7)
A
=
A
max
exp
−
(
−
ln
K
0
+
(
m
/
RT
)
X
)
/
[
1
+
exp
−
(
−
ln
K
0
+
(
m
/
RT
)
X
)
]
.
The m-values for an array of solutes would follow the Hofmeister series (a ranking
of solutes according to their effects on diverse biopolymer processes; Record et al.,
2013). Eq. 7 has the same form as Eq. 1, but it provides a thermodynamic interpretation
of the resulting parameters. If the activating conformational change were triggered
only by interactions of ions with charged surfaces (a Coulombic effect), the logarithm
of the equilibrium constant K would be expected to vary with a power (n) of the logarithm
of the solute concentration X (Record et al., 2013):
(8)
ln
K
=
ln
K
0
−
(
m
/
RT
)
[
ln
X
]
n
.
A relationship analogous to Eq. 7 would then reflect the dependence of ln K on [ln
X]n, and ln K0 would be the value of ln K at an ion concentration (X) of 1 M. It is
critical to note that reliable estimates of m/RT, the most informative parameter,
can only be obtained from data that define the full range of f values.
Proteoliposome-based analysis of osmosensing by ProP, BetP, and OpuA (Wood, 2011b).
All tested membrane-impermeant solutes had similar effects on transporter activity
when applied to attain the same osmolality at the external transporter surface. This
response was phospholipid sensitive: the osmolality at which each transporter activates
was a direct function of the anionic lipid content of the host membrane (both in vitro
and in vivo). All three transporters became active as inorganic ions were concentrated
at their cytoplasmic surfaces from ∼0.1 to ∼0.5 M. Differences emerged when diverse
ions were used, however.
ProP activity correlated with luminal cation concentration but not luminal K+ concentration.
For proteoliposomes loaded with K phosphate plus the K salts of various anions, the
osmolality yielding half-maximal ProP activity (Π1/2/RT) followed the Hofmeister series.
ProP activity was enhanced when proteoliposomes were loaded with high molecular weight
polymers (polyethyleneglycols or bovine serum albumin) at concentrations that simulated
the volume exclusion occurring in the bacterial cytoplasm (Culham et al., 2012). Culham
et al. (2012) concluded that ProP activity is determined by the concentrations of
Hofmeister anions and macromolecular crowding.
Internal K+ phosphate, glutamate or chloride, Rb+, or Cs+ chloride activated BetP
to varying degrees, whereas Na+ (the coupling ion), NH4
+, or choline chloride did not. K+ salts yielded the strongest stimulations (Krämer,
2010). However, K+ dependence did not fully account for the osmotic activation of
BetP in vivo, leading Maximov et al. (2014) to conclude that BetP senses K+ concentration
and a signal from the membrane. The effects of crowding agents on BetP activity have
not been reported.
The rate of glycine betaine uptake via OpuA was enhanced similarly by K+, Na+, Li+,
or NH4
+ chloride. OpuA was further activated by MgCl2 and BaCl2 and inhibited by RbCl and
CsCl. Ions and a large polyethyelene glycol (PEG600) acted synergistically to stimulate
substrate-dependent ATP hydrolysis by OpuA in nanodiscs (Karasawa et al., 2013). Karasawa
et al. (2013) concluded that OpuA responds synergistically to the ionic strength and
macromolecular crowding.
These reports evoke critical roles for electrolytes, for a membrane with a polyanionic
surface, and possibly for cytoplasmic volume exclusion in transporter activation.
All are likely to result from some combination of collective Coulombic and Hofmeister
effects of luminal solutes on changes to cytoplasm-exposed membrane and transporter
surfaces. Unfortunately, the reported data are insufficient to clearly delineate the
relative contributions of Coulombic and Hofmeister effects.
It is challenging to deduce the structural mechanism of osmosensing because membrane
proteins are refractory to structural analysis. An impressive series of crystal structures
has made enormous contributions to our understanding of the transport mechanism for
BetP and related systems (Perez et al., 2014). However, conformational differences
between inactive and osmotically activated BetP conformers remain to be defined. Data
outlined above suggest that BetP is a chemosensor, possessing one or more cytoplasm-exposed,
K+-specific regulatory sites, but those sites have not been identified. By comparison,
our structural knowledge of ProP and OpuA is limited (Fig. 3).
We do know that each transporter is an oligomer (ProP and OpuA are dimers; BetP is
a trimer; Wood, 2011b). The role of oligomerization in osmosensing by BetP has been
explored experimentally but remains uncertain (Becker et al., 2014). The roles of
oligomerization for ProP and OpuA remain unknown. Each transporter possesses an extended
cytoplasmic C terminus (Fig. 3). The C termini of some ProP orthologues form antiparallel,
intermolecular coiled-coils, whereas the extended C termini of other orthologues do
not include coiled-coil motifs. The C terminus of BetP forms a long α helix that mediates
inter-monomer interactions within BetP trimers, and the C termini of the two ATP-binding
subunits of OpuAA include dual cystathionine-β-synthase domains with anionic tails.
Structural changes to the C-terminal domains modulate the osmoregulatory response
(they shift the osmolalities at which the transporters become active). It has been
proposed that the cytoplasmic C termini mediate osmosensing via salt-sensitive interactions
with other transporter elements (protein–protein interactions) and/or the polyanionic
membrane surface. Osmotically induced variations in membrane surface charge density
would also modulate protein–membrane interactions. Each of these interactions would
have a characteristic thermodynamic signature, and clear dominance of Coulombic or
Hofmeister effects would support distinct structural models. Thus, osmosensing may
provide a paradigm for the regulation of membrane protein structure and function through
protein–solvent interactions, involving solute exclusion from or accumulation at extensive
protein and/or membrane surfaces.
This Perspective series includes articles by Andersen, Sachs and Sivaselvan, and Haswell
and Verslues.