SHOWCASE ON RESEARCH
When Proteostasis Goes Bad:
Protein Aggregation in the Cell
Mona Radwan, Rebecca Wood, Xiaojing Sui and Danny Hatters*
Department of Biochemistry and Molecular Biology, Bio21 Molecular Science and Biotechnology Institute,
University of Melbourne, VIC 3010
*Corresponding author: [email protected]
Protein aggregation is a hallmark of the major
neurodegenerative diseases including Alzheimer’s,
Parkinson’s, Huntington’s and amyotrophic lateral
sclerosis (ALS; also known as motor neuron disease) and is
a symptom of a breakdown in the management of proteome
foldedness (or proteostasis) (1). Indeed it is remarkable that
under normal conditions cells can keep their proteome in a
highly crowded and confined space without uncontrollable
aggregation. Proteins pose a particular challenge relative
to other classes of biomolecules because upon synthesis,
they must typically follow a complex folding pathway to
reach their functional conformation (native state). Nonnative conformations, including the unfolded nascent
chain, are highly prone to aberrant interactions, leading to
aggregation. Here we review recent advances in knowledge
of proteostasis, approaches to monitor proteostasis and the
impact that protein aggregation has on biology. We also
include a discussion of the outstanding challenges.
Nuts and Bolts of Proteostasis
To maintain proteostasis, the cell employs a proteostasis
network (PN) of quality control machinery that shield
proteins from non-native interactions, act as folding
checkpoints, and govern protein synthesis, turnover and
trafficking (shown conceptually in Fig. 1). During protein
synthesis, hydrophobic regions of nascent polypeptide
chains are shielded from non-specific interactions by
ribosome-associated chaperones including the nascent
chain associated complex, Hsp70 chaperones and Hsp40
co-chaperones. More than two-thirds of the proteome,
especially membrane proteins and other complex
multidomain proteins, require additional substantial
guidance to fold (2). Early steps of this guidance include
hsp70 family members acting as triage points connecting
to other branches of the PN, such as the calnexin cycle.
At the triage, proteins can be engaged with downstream
chaperones including Hsp90s and chaperonins, and
Fig. 1. Proteostasis
governs the life
of a protein from
synthesis to
degradation.
Shown is a simplified
diagram of key steps
in the life of a protein
and how this is guided
by the proteostasis
network (in green).
Management of
protein folding is
normally regulated by
a network of about 800
proteins in humans.
When proteostasis
fails, proteins can
misassemble into
aggregates (red). The
failure of proteostasis
correlates with a gain
of proteotoxicity;
some of the postulated
mechanisms covered
in this review are
indicated.
Vol 47 No 2 August 2016
AUSTRALIAN BIOCHEMIST
Page 11
SHOWCASE ON
RESEARCH
When Proteostasis Goes Bad:
Protein Aggregation in the Cell
in the case of being irreparably misfolded, targeted for
degradation by the ubiquitin proteasome system (UPS) or
autophagy. Hsp70 family proteins are also important in
preventing misfolding of nascent chains during translation
and ensuring optimum translation rates. Elements of the PN
also play critical roles targeting and translocating proteins
to cellular organelles including the nucleus, mitochondria
and the endoplasmic reticulum (ER), which contain their
own subnetworks of quality control machinery and
checkpoints for folding.
Failure of Proteostasis and Protein Aggregation
Unfolded states are permissive to entering misfolding
pathways that result in aggregation, which provides
a rival low energy state to the native fold (3). For this
reason, aggregation is often observed as symptomatic of
problems in maintaining proteostasis (1). For example,
proteins accumulate into punctate aggregate structures
inside neurons, known as inclusions, as a hallmark of
neurodegenerative diseases including Alzheimer’s disease,
ALS and Huntington’s disease (1). Proteins may also
naturally aggregate during the normal ageing process, with
the PN remodeling to accommodate the changes – a process
that may be dysregulated in neurodegenerative disease (4).
Mutations can also change the folding stability of individual
proteins, which may modify baseline proteostasis capacity
and influence disease risk (5).
One of the great challenges of current research efforts
is in understanding how protein aggregation relates to
toxicity. One of the longest standing theorems is that
soluble protein oligomers, as precursors to a larger
amyloid state, are directly proteotoxic and that this seems
to be independent of the protein sequence (6). However,
the mechanisms for toxicity, and the relevance to disease
remain to be robustly validated (6). Possible routes of
toxicity include physical disruption of membranes or
other cellular structures, or interference with synaptic
structure and plasticity (7). Larger aggregates seem to be
less toxic, and this may arise from a lower concentration of
effective reactive ‘ends’ of fibrils or cellular mechanisms
to sequester the dispersed oligomers together (8,9).
One of the more interesting recent hypotheses for
proteotoxicity of protein aggregates arises from the
observation that markers of stress granule abnormalities
appear in pathology (notably ALS) (10). Indeed several RNA
granule proteins, most notably trans-activation element
(TAR) DNA binding protein-43 (TDP-43), FUS and hnRNP
family proteins cause ALS when mutated. Stress granules,
and related structures, P-bodies (Fig. 1), are condensed
foci of mRNA and ribonucleoproteins that form under
translational stresses (10). They act as sites for temporal
translational repression and quality control of mRNAribonucleoprotein. Recent studies have suggested many
RNA granule proteins contain predicted prion-like domains
that mediate liquid:liquid protein phase separation and/
or ‘functional’ amyloid scaffolding (11). The presence of
‘functional amyloid’ indicates a necessity for these structures
to be rigorously tamed to avoid pathological aggregation
into amyloid fibrils. Indeed, mutations or conditions that tip
the balance to pathological aggregation have been proposed
as a basis for these neurodegenerative diseases (12).
Page 12
Mechanisms of Misfolded Protein Sequestration –
Cell Driven Inclusion Formation
While protein oligomers or nascent proteins may
themselves be toxic, there is also evidence that cells actively
cluster together misfolded proteins for sequestration.
The most compelling evidence for this model is from
a key study showing that in cells expressing mutant
Huntingtin exon 1, cells that formed inclusions had
greater rates of survival than cells than cells that did not
(13). The mechanisms for this effect with Huntingtin
exon 1 remain to be determined, however, similar ‘active’
mechanisms of inclusion building have been proposed
with other misfolded proteins. The original model was that
of the ‘aggresome’, which described a dynein-mediated
retrograde transport mechanism of misfolded DF508
Cystic Fibrosis Transmembrane conductance regulator
protein into the microtubule organising centre (14). The
aggresome model however, is problematic in that it does
not appropriately define the diversity of processes that
sequester aggregating proteins into a centralsed location.
Recent studies have partly addressed this deficiency by
unearthing two distinct aggresome-like compartments
for aggregating proteins. One is the juxtanuclear quality
control (JUNQ), which comprises reversibly-aggregated
proteins and the other is the insoluble protein deposit
(IPOD), which comprises irreversibly-aggregated proteins
(Fig. 1) (9). Different disease-associated aggregating
proteins seem to preferentially partition into either JUNQ
or IPOD exclusively, and that these compartments may
correlate with different mechanisms of aggregation (15).
Huntingtin exon 1 accumulates in IPOD-like structures.
By contrast, polyalanine, which aggregates via soluble
a-helical clusters, and SOD1 mutants, which also cluster
into soluble oligomers prior to aggregation, accumulate
in JUNQ-like structures (16). FUS and TDP-43 partitioned
into both structures and another unidentified structure
(17). These data indicate that aggregation into foci arises
via diverse processes, which are in part consistent with the
JUNQ and IPOD models, but which also seem to involve
a further uncharacterised layer of complexity. Hence,
critical questions remain as to what factors drive proteins
to different compartments, how the elements of different
models fit together, how the additional inclusion types
relate to the JUNQ and IPOD structures, and whether there
are further inclusion types that remain to be discovered.
An ‘Omics View of Proteostasis
‘Omics technologies have begun to transform our
understanding of what happens to the proteome under
proteostasis stress. Of note is a recent study showing that
the nuclear pore is fundamentally damaged in cells that
display extensive cytoplasmic aggregates of unrelated,
aggregation-prone proteins (18). Components of the
import and/or export machinery coaggregate with diseaseassociated mutant proteins, including polyglutaminecontaining huntingtin protein, mutant TDP-43 and the
C9orf72-associated polydipeptides (18). Indeed, the
progressive loss of proteostasis may account for the
observation of nuclear transport becoming increasingly
‘leaky’ with ageing (19).
‘Omics, along with computational approaches, has also
AUSTRALIAN BIOCHEMIST
Vol 47 No 2 August 2016
SHOWCASE ON
RESEARCH
When Proteostasis Goes Bad:
Protein Aggregation in the Cell
enabled insight to which proteins in the proteome become
more vulnerable to aggregation under proteostasis stress.
Indeed in different organisms and cell culture conditions,
hundreds of proteins aggregate under stress (20,21).
More importantly, distinct stress types tend to induce the
same proteins to aggregate by reducing their aggregation
threshold, indicating the presence of a metastable
sub-proteome (22). This includes many regulators
of proteostasis, such as chaperones, and proteasome
subunits (23). Others have shown that in C. elegans,
highly abundant proteins become supersaturated with
respect to their solubility and most extensively contribute
to aggregate loads during ageing, despite their otherwise
low propensity to aggregate (24). A supersaturation score,
which combines experimental and bioinformatics data,
has been used to explain which proteins coaggregate with
amyloid plaques, neurofibrillary tangles, Lewy bodies
and artificial b-proteins, as well as proteins that aggregate
in C. elegans during ageing (25).
While many proteins are likely to be aggregating
‘inappropriately’ under stress, there is also evidence for an
adaptive aggregation response as well. For example, under
heat stress, many proteins form transiently aggregated
complexes that remain functionally competent, and these
are not degraded upon recovery (20). Such structures
were suggested to aid in the management of the stressed
proteome (20).
Quantification of Proteostasis
A major area of development in the study of proteostasis
involves building quantitative tools to mechanistically
probe the PN. One approach to do this relies on metastable
reporter proteins that engage with the PN and read out the
effectiveness of suppressing aggregation as an indicator of
proteostasis efficacy. This includes temperature-sensitive
endogenous proteins, which have been used to identify
novel regulators of the PN in C. elegans (26) and to probe
the collapse of proteostasis when disease-associated
proteins are expressed. More advanced strategies involved
designer ectopic reporters including destabilised mutants
of firefly luciferase, which is a known chaperone substrate
that requires Hsp70 and Hsp90 to fold (27), and the de novo
designed enzyme retroaldolase as a sensor of proteome stress
(28). These reporters offer the advantage of not interfering
with normal biological pathways and have been used in the
studies cited above to probe changes in proteostasis induced
by drugs, disease proteins, and ageing.
Another strategy for gaining insight into proteostasis
has been the systems approach of computationally
modelling the entire PN. This strategy offers mechanistic
insight into the connections between different modules
of the PN in a way that makes it possible to make
sense of experimental data and make predictions that
informatively guide experiments. FoldEco provides
a comprehensive computational model of the PN in
Escherichia coli (29). This platform provides the capacity
to model, and make predictions about, the kinetics of
chaperone-protein interactions as well as synthesis,
aggregation and degradation, in a fully integrated
system. It can also be ‘fitted’ to experimental data for a
mechanistic understanding of proteostasis (30). However,
Vol 47 No 2 August 2016
transferring this approach to eukaryotic cells remains an
ongoing challenge because of a vastly more complicated
proteostasis networks.
The Path Ahead
The true potential of proteostasis measurement will be
realised when computational models can be integrated
with powerful tools for measuring proteostasis activity.
Further work, which our lab is actively engaged in, will be
required to develop extremely well-characterised sensors
that are able probe the diverse arms of the proteostasis
network without disrupting it, and yield quantitative
information that can feed into network models that enable
information-rich analysis.
In conclusion, research into mechanisms of proteostasis
and their relationship to protein aggregation in disease
has progressed substantially in the last few years. We
see several key questions as immediate challenges to be
addressed. An important question is how generalisable
are the mechanisms of toxicity caused by aggregation of
different proteins? Are the mechanisms of proteotoxicity
non-specific? And are there certain machinery hotspots,
such as the nuclear pore, that need to be ‘hit’ to trigger
disease? Another gap is defining precisely which proteins
aggregate in the proteome under proteostasis collapse and
probing whether this explains the changes seen in disease.
Given that TDP-43 mislocalisation is observed as marker of
all forms of sporadic ALS, is TDP-43 simply a bellwether for
a broader subproteome highly prone to aggregation under
stress? A large part of our focus is on building new tools
and approaches that can get at such problems including
more quantitative measures of proteostasis.
References
1. Schneider, K., and Bertolotti, A. (2015) J. Cell Sci. 128,
3861-3869
2. Braselmann, E., Chaney, J.L., and Clark, P.L. (2013)
Trends Biochem. Sci. 38, 337-344
3. Jahn, T.R., and Radford, S.E. (2005) FEBS J. 272, 59625970
4. Brehme, M., Voisine, C., Rolland, T., Wachi, S., Soper,
James H., Zhu, Y., Orton, K., Villella, A., Garza, D.,
Vidal, M., Ge, H., and Morimoto, R.I. (2014) Cell Rep. 9,
1135-1150
5. Gidalevitz, T., Krupinski, T., Garcia, S., and Morimoto,
R.I. (2009) PLoS Genet. 5, e1000399
6. Fändrich, M. (2012) J. Mol. Biol. 421, 427-440
7. Haass, C., and Selkoe, D.J. (2007) Nat. Rev. Mol. Cell Biol.
8, 101-112
8. Bucciantini, M., Giannoni, E., Chiti, F., Baroni, F.,
Formigli, L., Zurdo, J., Taddei, N., Ramponi, G., Dobson,
C.M., and Stefani, M. (2002) Nature 416, 507–511
9. Kaganovich, D., Kopito, R., and Frydman, J. (2008)
Nature 454, 1088-1095
10.Ramaswami, M., Taylor, J.P., and Parker, R. (2013) Cell
154, 727-736
11.Molliex, A., Temirov, J., Lee, J., Coughlin, M., Kanagaraj,
A.P., Kim, H.J., Mittag, T., and Taylor, J.P. (2015) Cell 163,
123-133
References continued on page 7
AUSTRALIAN BIOCHEMIST
Page 13
SHOWCASE ON
RESEARCH
A Short History of Amyloid
References continued from page 13
12.Li, Y.R., King, O.D., Shorter, J., and Gitler, A.D. (2013) J.
Cell Biol. 201, 361-372
13.Arrasate, M., Mitra, S., Schweitzer, E.S., Segal, M.R., and
Finkbeiner, S. (2004) Nature 431, 805-810
14.Kopito, R.R. (2000) Trends Cell Biol. 10, 524
15.Polling, S., Mok, Y.-F., Ramdzan, Y.M., Turner, B. J.,
Yerbury, J.J., Hill, A.F., and Hatters, D.M. (2014) J. Biol.
Chem. 289, 6669-6680
16.Polling, S., Ormsby, A. R., Wood, R.J., Lee, K.,
Shoubridge, C., Hughes, J.N., Thomas, P.Q., Griffin,
M.D., Hill, A.F., and Bowden, Q. (2015) Nat. Struct. Mol.
Biol. 22, 1008-1015
17.Farrawell, N.E., Lambert-Smith, I.A., Warraich, S.T.,
Blair, I.P., Saunders, D.N., Hatters, D.M., and Yerbury,
J.J. (2015) Sci. Rep. 5, 13416
18.Woerner, A.C., Frottin, F., Hornburg, D., et al. (2016)
Science 351, 173-176
19.D’Angelo, M.A., Raices, M., Panowski, S.H., and Hetzer,
M.W. (2009) Cell 136, 284-295
20.Wallace, E.W., Kear-Scott, J.L., Pilipenko, E.V., et al.
(2015) Cell 162, 1286-1298
21.Narayanaswamy, R., Levy, M., Tsechansky, M., Stovall,
G.M., O’Connell, J.D., Mirrielees, J., Ellington, A.D., and
Marcotte, E.M. (2009) Proc. Natl. Acad. Sci. USA 106,
10147-10152
22.Weids, A.J., Ibstedt, S., Tamas, M.J., and Grant, C.M.
(2016) Sci. Rep. 6, 24554
23.David, D.C., Ollikainen, N., Trinidad, J.C., Cary, M.P.,
Burlingame, A.L., and Kenyon, C. (2010) PLoS Biol. 8,
e1000450
24.Walther, D.M., Kasturi, P., Zheng, M., Pinkert, S.,
Vecchi, G., Ciryam, P., Morimoto, R. I., Dobson, C. M.,
Vendruscolo, M., Mann, M., and Hartl, F.U. (2015) Cell
161, 919-932
25.Ciryam, P., Tartaglia, G.G., Morimoto, R.I., Dobson,
C.M., and Vendruscolo, M. (2013) Cell Rep. 5, 781-790
26.Silva, M.C., Fox, S., Beam, M., Thakkar, H., Amaral,
M.D., and Morimoto, R.I. (2011) PLoS Genet. 7, e1002438
27.Gupta, R., Kasturi, P., Bracher, A., Loew, C., Zheng, M.,
Villella, A., Garza, D., Hartl, F.U., and Raychaudhuri, S.
(2011) Nat. Meth. 8, 879-884
28.Liu, Y., Zhang, X., Chen, W., Tan, Y. L., and Kelly, J.W.
(2015) J. Am. Chem. Soc. 137, 11303-11311
29.Powers, E.T., Powers, D.L., and Gierasch, L.M. (2012)
30.Cho, Y., Zhang, X., Pobre, K.F.R., et al. (2015) Cell Rep. 11,
321-333
Vol 47 No 2 August 2016
AUSTRALIAN BIOCHEMIST
Page 7