Overview of single-point mutations in MxA in different types of cancer
In the COSMIC database, 122 tumor-related mutations were identified in the human Mx1 gene from whole-genome, whole-exome, and transcriptome sequencing data of human cancers.
All 122 mutations reside in coding regions of the MxA protein. For these genetic aberrances,
frameshift mutations and termination codon mutations were excluded, as they lead to
obvious truncation of the polypeptide chain and therefore the obliteration of MxA
structure and function. In addition, silent mutations were also excluded because the
amino acid sequence of MxA remains intact in these cases. Consequently, 22 unique
cancer-related single amino acid mutations in MxA protein were selected and subjected
to subsequent analysis. These 22 MxA mutations, summarized in Table 1, were found across 12 types of human cancer, of which colorectal cancer accounted
for half of the mutations, followed by cutaneous squamous cell carcinoma (3 mutations)
and mantle cell lymphoma (2 mutations). The other 9 cancers each featured 1 mutation.
Two mutations, namely T651M and R655C, both appeared twice. T651M was found in colorectal
cancer and head and neck squamous cell carcinoma, whereas R655C was discovered in
two separate sequencing projects for colorectal cancer. Overall, MxA single-point
mutations are widely spread in different human cancers, and MxA mutates more frequently
in colorectal cancer according to current data.
Table 1. Summary of reported human myxovirus resistant protein A (MxA) single-point mutations
in human cancers
We then checked the conservation level of the amino acid residues mutated in cancers.
Seven residues are highly conserved in Mx proteins and dynamins; 4 residues also show
considerable overall conservation. In addition, 5 residues are conserved in Mx proteins
but not in dynamins, whereas 6 residues display no conservations at all (Fig. 1). Intriguingly, T651 and R655, although their corresponding mutations appeared more
frequently in sequencing data, belong to non-conserved group. This result suggests
that it is obscure to infer the physiological importance of these single-point mutations
only from their conservativeness in primary structure. To obtain more reliable information,
individual analysis of these mutations by structural means is indispensable.
Fig. 1. Amino acid sequence alignment of myxovirus resistance (Mx) proteins and dynamins.
Amino acid sequences of Mx proteins and dynamins from human, mouse, chicken, zebra
fish, fruit fly, C. elegans, and yeast are shown in the sequence alignment (see the “Methods†for details). Residues
with a conservation of greater than 70% are color-coded (negatively charged amino acid residues D and E in red; positively charged R, K, and H in blue; polar N, Q, S, and T in grey; weak or nonpolar A, L, I, V, F, Y, W, M, and C in green; and special P and G in brown). Numbers in square brackets in front of each 10-residue sequence fragment indicate the ordinal position of the
first residue of this fragment at the primary structure of the corresponding protein.
Tumor-associated mutations are indicated at the corresponding positions. Residues
highly conserved in Mx proteins and dynamins (L95, P96, P218, V263, L619, E632, and
L643) are shown in violet, residues showing considerable overall conservation (S134, R310, G392, and Y538)
in magenta, residues conserved in Mx proteins but not in dynamins (K326, V449, S572, R649, and
R654) in yellow, and residues with no conservativeness (T27, N491, R522, G540, T651, and R655) in
cyan
Distribution of cancer-related single-point mutations in MxA
Human MxA protein is a large dynamin-like GTPase composed of 5 structurally discriminated
domains plus an N?-loop (Fig. 2a). Recently, several crystal structures of MxA have been reported, which collectively
covered most of the molecule, except for the N?-loop and L4 4]–6]. We took advantage of these structures to perform an in-depth analysis of the 22
cancer-related mutations and their possible structural and functional outcomes on
the protein. First, we mapped the mutated residues on MxA molecule and found that
these 22 mutations are scattered throughout all regions (Fig. 2b). The G domain, which is largest in size, contains 7 mutation sites, whereas the
smaller N?-loop and Hinge 1 each contains 1 mutation site. Considering the size, however,
BSE and L4 are relatively more prone to mutation, where 5 and 3 mutation sites are
spotted, respectively (Fig. 2c–f). The distribution of the mutations is summarized in Table 2. Next, all single-point mutations were individually investigated according to the
domains.
Fig. 2. Overview of the distribution of cancer-related mutations within the human myxovirus
resistant protein A (MxA) domains. a schematic representation of the domain structure of human MxA. N?-loop N-terminal disordered loop; B bundle-signaling-element domain (BSE), G G domain; H Hinge 1, S Stalk. Borders of the domains are indicated by corresponding residue numbers. b overview of the position of all mutations in MxA. The G domain is colored in orange, BSE in red, Hinge 1 in sky-blue, and Stalk in green. The missing N-terminal 44 residues (shown in magenta) and L4 (shown in cyan) are indicated as dashed lines. Mutations that are included by the reference model are illustrated as yellow spheres. Residues that are missing in this reference model but are present in other reference
models are illustrated as filled yellow circles. Residues missing in all reference structures are shown as yellow stars. c–f overview of the mutations in individual domains of MxA, as outlined by dashed rectangles in Fig. 2a at the corresponding areas: c G domain, d BSE, e Hinge 1, and f Stalk. Note that the representations of e Hinge 1 and f Stalk representations were rotated counter-clockwise 90° from those in Fig. 2b
Table 2. Domain distribution of cancer-related MxA single-point mutations
Single-point mutations in G domain
The G domain of MxA is responsible for GTP hydrolysis, as well as for the inter-ring
homo-dimerization via the nucleotide-binding pocket 39]. Both actions are crucial for the mechano-chemical coupling of the entire oligomer
and thus for the function of the protein 5], 6].
Cutaneous squamous cell carcinoma-associated single-point mutations L95P and P96S
sit within Switch I, a key component of the active site which harbors guanine nucleotides and exists
in all known GTPases 40]. L95 is deeply buried in a comprehensive hydrophobic pocket composed of L87, V93,
L107, L109, I143, and L164 (Fig. 3a). Its mutation to proline is less favored in this hydrophobic network and thereby
affecting the stability of the whole domain. Besides, as proline and glycine are chemically
different from other amino acids in ? and ? dihedral angles about the peptide bond,
L95P mutation may lead Switch I to a different bending direction, and this will hinder the binding of nucleotides
(Fig. 3b). Similarly, P96S mutation tends to generate more freedom for Switch I to swing, which is also unfavorable for nucleotide binding and hydrolysis (Fig. 3b).
Fig. 3. Cancer-related MxA mutations in G domain. a L95 is deeply buried in a hydrophobic cave. b P96 leads the direction of Switch I, which embraces guanine triphosphate nucleotide. c S134 is located on the surface of the G domain. d P218 sits at the end of ?-strand 4 (?4 G
). e V263 is loosely enwrapped by neighboring hydrophobic residues. f R310 is exposed to the solvent and takes two side chain conformations, whereas K326N
interacts with vicinal residues. All cancer-related mutations are shown as the original
(i.e., non-mutated) residues, and the post-mutation residues are included in the labels,
as L(original)95P(post-mutation), and so on. This scheme is also applied to all of
the following figures
S134L mutation occurs between the extra ?-strand 1 (?E1 G
) and extra ?-helix (?E G
) of MxA (Fig. 3c). As the polar S134 is located on the surface of the molecule, and its side chain
also protrudes outward, its mutation to hydrophobic leucine would not influence the
global folding of the G domain, but just slightly change the surface entropy of the
molecule.
P218S mutation emerges between ?-strand 4 (?4 G
) and ?-helix 3 of the G domain (?3 G
). P218 terminates ?4 G
and turns the polypeptide chain to the opposite direction (Fig. 3d). Analogous to P96S, P218S may change the original trajectory of the following loop
and therefore affect the folding of the protein.
V263 is surrounded by several other hydrophobic residues, including I223, L246, V264,
and V268 (Fig. 3e). Compared with the hydrophobic pocket engulfing L95, the hydrophobic environment
around V263 is much less extensive. As methionine is also a nonpolar residue, V263M
mutation makes no alteration of the hydrophobic property of this area. Although methionine
is physically a bulkier residue than valine, there is enough space at the top of V263
to accommodate a methionine side chain. Therefore, V263M mutation would cause very
limited influence to the folding and stability of the protein.
R310 sits on the surface of the molecule. The side chain of R310 points outward and
is quite flexible, as two conformations can be observed in the crystal structure,
which suggests that this residue is not bound by any side-chain interaction from other
parts of the protein (Fig. 3f). In this situation, its mutation to a polar serine residue will neither break any
intra-molecular associations nor drastically change the surface polarity of MxA. Although
R310S causes the loss of some positive charge, this mutation would not negatively
affect the function of the protein.
K326 is located in ?-helix 5 of the G domain (?5 G
) and also has an outward side-chain conformation. It forms a hydrogen bond with the
oxygen of K273 and additionally a weak salt bridge with E330 which also sits in ?5 G
(Fig. 3f). The K326 N mutation may not substantially affect these two interactions, as asparagine
also has a polar side chain that can form hydrogen bonds with K273 and E330. Therefore,
it is very likely that this mutation does not give rise to any major disruptions of
MxA structure.
Single-point mutations in BSE
The BSE is composed of three ?-helices which are widely dispersed at the very N-terminus,
middle, and the very C-terminus of MxA protein, respectively 5]. BSE is the pivot for transmitting the mechanical force generated from GTP hydrolysis
at the G domain to the stalk region, so as to regulate Stalk-dependent oligomerization
of the molecule 41]. Five tumor-associated single-point mutations were found in BSE and, more precisely,
the ?-helix 3 of BSE (?3 B
), which is also close to the end of the protein (Fig. 4a). According to the crystal structure of full-length MxA, all these 5 residues exhibit
an outward side chain conformation. L643 is hydrophobically linked to L357, but its
mutation to nonpolar valine would not substantially disrupt this interaction, and
thus does not adversely affect the structure of the protein. The side chains of R649,
T651, and R655 are all located within a spacious environment and have no contact to
other elements of the molecule. Therefore, their mutations, except for mutating to
glycine or proline, will lead to negligible interference in the folding of the protein.
However, since arginine and threonine are both strong polar residues, whereas methionine
and tryptophan possess bulky hydrophobic side chains, and cysteine is a weak-polar
residue prone to various post-translational modifications, the R649W, T651M, and R655C
mutations bear the possibility to interrupt the association between the protein and
other protein partners or incur unexpected modifications.
Fig. 4. Cancer-related MxA mutations in BSE. a surface-located mutations and a surrounding residue. b the full-length MxA oligomer in the crystal lattice represented by a linear hexamer.
All 6 monomers are distinguished by 6 different colors and indicted by corresponding numbers. c A magnified view from the dashed rectangle in Fig. 4b shows that the R654 is involved in inter-molecular Stalk-BSE interaction
In addition to its role as an intra-molecular messenger, BSE also contributes to the
formation of functional homo-oligomers, which is a fundamental feature of dynamin
superfamily members (Fig. 4b). R654 participates in BSE-Stalk interaction between parallel MxA monomers via a
charged interaction with D478 on the other molecule (Fig. 4c). Disruption of this salt bridge by a D478A mutation results in abnormal GTPase
activity and weakened oligomerization and antiviral abilities 4]. Not surprisingly, the colorectal cancer-associated R654Q mutation, which abolishes
the D478-R654 salt bridge, should have a similar effect. On the other hand, glutamine
still possesses the tendency to form a hydrogen bond with D478, and this can be deemed
as a compensation for the loss of the salt bridge. As a result, this mutant would
hardly cause negative consequence to MxA compared with the reported D478 mutant.
Single-point mutation in Hinge 1
Hinge 1 plays a crucial role in connecting BSE and the stalk region. Extensive interactions
between the two loops of Hinge 1 (L1 H
and L2 H
) stabilize the ambient region so that the relative position of BSE and Stalk is confined
from random movement 4]. The salt bridge formed by E632 on L1 H
and R640 on ?3 B
is the determining factor in this stabilization effect (Fig. 5). Destruction of this salt bridge by mutating either E632 or R640 to alanine leads
to tremendous change of GTPase activity, and almost completely abolishes the oligomerization
ability and antiviral effect of MxA. Therefore, it can be well expected that the melanoma-related
mutant E632K, which also disrupts the E632-R640 interaction, would be detrimental
to the integrity of Hinge 1, and thus the global stability of MxA molecule.
Fig. 5. Cancer-related MxA mutations in Hinge 1. E632 in Hinge 1 forms a salt bridge with
R640 to stabilize the local conformation. Part of the Stalk was removed for clarity
Single-point mutations in the stalk region
The stalk region is essential for the functional assembly of dynamin superfamily members
including MxA 5], 42], 43]. In crystal lattice, MxA stalk was found forming linear oligomers via several conserved
interfaces, which were then proven to also mediate the architecture of ring-like oligomers
in physiological conditions 4] (Fig. 6a). Therefore, it was necessary to first determine the mutations residing in these
interfaces, namely G392V and V449G. G392 is located on interface 3 and highly conserved
in dynamins and Mx proteins (Fig. 6b). The G392D mutation in MxA was demonstrated to disrupt the oligomerization of the
protein 4]. Moreover, its counterpart mutation (G385) in yeast dynamin leads to the breakdown
of tetramer into stable dimers 44]. It is therefore not astonishing that the G392V mutation found in renal cell carcinoma
deprives the oligomerization capability of MxA and thus its biological activity. In
addition, the V449 residue was found to be the interaction partner of G392 on the
parallel monomer in the MxA oligomer (Fig. 6b). Its mutation to glycine tends to affect the integrity of interface 3, as well
as the conformation of the subsequent Loop 2 on the stalk (L2 S
). Altogether, these two mutations (G392V and V449G) are likely to impair the physiological
function of MxA.
Fig. 6. Cancer-related MxA mutations in Stalk. a the MxA Stalk oligomer in the crystal lattice, as represented by a linear hexamer.
The monomers are color-coded and labeled in the same manner as in Fig. 4b. Note that compared with the relative direction of the full-length MxA hexamer in
Fig. 4b, the Stalk hexamer has been rotated clockwise for 90° along the X axis, and then 180° along the Y axis. b A magnified view from the dashed rectangle in Fig. 6a corresponding to interface 3 shows the interaction between G392 and V449 from parallel
monomers. The monomers 2 and 4 are removed for clarity. The invisible Loop 2 on the
stalk (L2 S
) in this model is indicated with a dashed line. c N491 forms a hydrogen bond with D385 from another ?-helix. d R522 is enveloped by three nearby glutamates. e L619 is surrounded by several hydrophobic residues
On the other hand, three additional single-point mutations that are not involved in
any oligomerization interfaces have been discovered in the stalk region. N491 is a
surface-located residue situated on the ?-helix 2 of Stalk (?2 S
), forming a hydrogen bond with D385 on the parallel ?-helix 1N-terminal part (?1N S
) (Fig. 6c). Its mutation to positively charged arginine may strengthen this interaction, as
a salt bridge can be thus introduced. Therefore, this single-point mutation would
play an insignificant role in tumorigenesis or in the development of colorectal cancer
22]. Another colorectal cancer-related mutation site, R522, is also a solvent-exposed
residue on ?3 S
. This residue, together with three neighboring negatively charged glutamates (E466
and E467 on ?2 S
, and E518 on ?3 S
), constitutes a vast network of charged interactions endorsed by the salt bridges,
in which R522 is at the center, to provide positive charge (Fig. 6d). It is imaginable that its mutation to weak polar cysteine causes the absence of
the pivotal positive charge and lead to interference of electrostatic balance on the
protein surface. In this case, the folding and stability of the whole protein would
become adversely influenced. Unlike N491 and R522, L619 located on ?4 S
possesses an inward side chain conformation. It is part of a local hydrophobic core
that includes L498 on ?3 S
, M616 on ?4 S
, and L629 on ?5 S
. When mutated to isoleucine, a derivative comparable to leucine itself in both size
and charge, the residue can still stably reside in and maintain this environment.
As a result, this ovarian carcinoma-associated mutation L619I will lead to hardly
detectable structural and functional aberrance to MxA.
Single-point mutations in the N?-loop and L4
The N?-loop and L4 are both intrinsically disordered loops that lack intra-molecular
interactions with other residues, and therefore, it is impractical to observe them
in crystal structures. The N?-loop is not conserved among Mx proteins, and its length
varies among different species. This region was recently reported to be involved in
the specification of viral targets 45]. Given the similar properties of threonine and serine, the T27S mutant would not
chemically differ from wild-type MxA, but it is unpredictable whether this mutation
would affect the protein’s interaction with viral structures. L4 is essential for
membrane binding and viral resistance by direct interaction 46], 47]. However, currently only several residues in the middle of L4, but not Y538, G540,
or S572, were proven to be functionally important for MxA 4], 47]. Therefore, it is difficult to predict the influence of these three mutations on
MxA function, although they are not likely to disrupt the flexible conformation of
L4.