AAC Accepted Manuscript Posted Online 26 May 2015 Antimicrob. Agents Chemother. doi:10.1128/AAC.00878-15 Copyright © 2015, American Society for Microbiology. All Rights Reserved.
1
Title: Characterization of poliovirus neutralization escape mutants of single-domain antibody
2
fragments (VHHs).
3 4
Running title: Poliovirus type 1 neutralization escape mutants against VHHs
5 6
Lise Schotte1, Bert Thys1, Mike Strauss2, David J. Filman2, Bart Rombaut1,* and James M. Hogle2,#.
7 8
1
9
Neurosciences, Vrije Universiteit Brussel, Laarbeeklaan 103, 1090 Brussels, Belgium
Department of Pharmaceutical Biotechnology and Molecular Biology, Center for
10
2
11
240 Longwood Avenue, Boston 02115, Massachusetts, USA
12
*
13
# Corresponding author:
[email protected] Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School,
Deceased
14 15
Corresponding author: James Hogle:
[email protected], Phone: 617-432-3918,
16
FAX: 617-432-4360
17 18
Word count: summary: 216
main text (excl. acknowledgements, excl. legends): 6248
19 20
Key words: poliovirus, VHH, nanobody, drug-resistance, escape-mutants
21 22
1|Page
23
Summary
24
To complete the eradication of poliovirus and to protect unvaccinated people subsequently, the
25
development of one or more antiviral drugs will be necessary. A set of five single-domain
26
antibody fragments (VHHs) with an in vitro neutralizing activity against poliovirus type 1 was
27
previously developed and their mechanisms of action have been studied. In this study,
28
neutralization escape mutants were selected for each VHH. Sequencing of the P1-region of the
29
genome showed that amino acid substitutions are found in the four viral proteins of the capsid
30
and that they are located both in proximity to the binding sites of the VHHs and in regions
31
further away from the canyon and hidden beneath the surface. Characterization of the mutants
32
demonstrated that they have similar single cycle replication kinetics as compared to their
33
parental strain and that they are all drug (VHH) independent. Their resistant phenotypes are
34
stable, as they do not regain full susceptibility to the VHH after passage over HeLa cells in the
35
absence of VHH. They are all at least as stable as the parental strain against heat inactivation at
36
44°C and three of them are even significantly (P 6014-fold (6A-mutants) when compared to Mahoney. It can also be seen
293
(Fig. 3) that the mutations are sufficiently stable to prevent reversion to their initial VHH-
294
susceptible state. However, four mutants, 6A_1, 6A_2, 19B_3 and 21E_5 did show an increase in
295
susceptibility to the VHHs, indicated by a decrease in their EC50 values.
296 297
3.5 Thermal stability
298
The thermal stability of the mutants was tested by heating of viral samples to 44°C. These
299
experiments were done because our previous paper investigating the mechanism of
300
neutralization showed that all five VHHs interfered with the uncoating of the virus and stabilized
301
the capsid (19). As represented in Fig. 4, all of the mutants showed a very good resistance
302
against heat inactivation. The titer at time 0 min was 4.2 x 107 ± 0.57 x 107 pfu/ml, and there
303
were no significant (α=0.05) differences between the titers of the mutants and Mahoney at time
304
0 (data not shown). After heating the virus samples at 44°C, the fraction of infectivity remaining
14 | P a g e
305
seemed to be higher for most of the mutants as compared to Mahoney when looking at the
306
graphs in Fig.4. However, when using an unpaired t-test to calculate the differences between
307
the fractions of infectivity remaining after heating the virus 60 minutes at 44°C (Fig.4), only
308
mutants 6A_4, 8A_1 and 19B_2 were shown to be significantly more stable than Mahoney to
309
thermal inactivation at 44°C.
310 311
3.6 Cross-resistance
312
To determine whether a mutant resistant against one VHH also shows resistance to other VHHs,
313
we infected cells with a mutant and treated the cells with either a VHH different from the one
314
against which original resistance was established, or with the antiviral compound pirodavir, or
315
the poliovirus neutralizing antibody 35-1F4. The results of the mutants tested are represented in
316
Fig. 5. In these graphs, the percentage of cell viability measured in the well after infection of the
317
cells with the indicated mutant and treatment with the VHH or compound shown on the x-axis
318
is given. This percentage was calculated using the following formula: (signal of sample – signal of
319
0 % control)/(signal of 100 % control – signal of 0 % control) x 100, where the 0 % control is a
320
control where cells were infected but not treated and where the 100 % control is a control in
321
which cells were neither infected nor treated.
322 323
As a first observation, we see that cross-resistance does occur for most of the variants, except
324
for 6A_1, 6A_4, 21E_1 and both 29F_1 and 29F_5, which are still susceptible to neutralization by
325
the four other VHHs. Reversely, all of the tested mutants could still be fully neutralized by at
326
least two other VHHs at a concentration of 1 µM, e.g. the cells infected with variant 8A_1 and
15 | P a g e
327
treated with VHH PVSP6A or PVSP19B retain full viability after three days of incubation at 37°C.
328
Although all of the mutants showed cross-resistance to other VHHs, none of the aa-
329
substitutions interfered with the binding and neutralizing activity of the antiviral compound
330
pirodavir (except for 19B_1) or the monoclonal antibody 35-1F4.
331 332
When looking at the mutants resistant to PVSP6A, 6A_1 and 6A_4 do not show cross-resistance
333
to other VHHs. 6A_2 and 6A_3 are no longer susceptible to neutralization by PVSS8A and
334
PVSS21E, probably due to the VP2 H142R and VP1 V61M substitutions also found in 21E_2 and
335
21E_3, respectively. In the case of 6A_2 the resistance to neutralization by PVSS8A could also be
336
due to the VP1 V47L substitution. 6A_4 also contains the VP2 H142R substitution, but the VP3
337
D209V substitution might compensate for this and allow neutralization by PVSS21E.
338 339
The mutants against PVSS8A have a single, dominant substitution (VP1 V47L) that induces cross-
340
resistance against PVSS21E and PVSP29F.
341 342
The mutants against PVSP19B are all resistant against PVSS21E. Interestingly, mutants 19B_1
343
and 19B_3 only differ in the replacement of glutamic acid at position 168 of VP1 by either
344
glycine or lysine, respectively. This seems to influence the cross-resistance pattern, as 19B_1
345
shows resistance against pirodavir and 19B_3 does not. In contrast, 19B_3 is resistant to
346
neutralization by PVSS8A and 19B_1 is not. The glutamic acid to glycine substitution somehow
347
influences binding of pirodavir to 19B_1, while replacement of this same glutamic acid by lysine
16 | P a g e
348
does not interfere with neutralization by pirodavir in 19B_3, but does disturb neutralization by
349
PVSS8A.
350 351
Of the five selected variants resistant against PVSS21E, four have only a single substitution of
352
which one, in 21E_5, is located on VP4. 21E_1 (VP2 T139I) can still be neutralized by the other
353
four VHHs, while an extra substitution (VP1 V61M) introduces resistance towards PVSS8A and
354
PVSP19B.
355 356
The mutants against PVSP29F have a rather limited variation in their aa-substitutions and these
357
are not shared by any of the variants resistant against the other VHHs. They specifically
358
introduce resistance against PVSP29F and do not seem to interfere with the binding of and
359
neutralization by the other VHHs, pirodavir or the monoclonal antibody 35-1F4.
360 361
Considering the frequency of development of cross-resistance against a certain VHH, PVSP6A is
362
the best in class, with only one mutant out of ten selected by the other VHHs being cross-
363
resistant against PVSP6A. PVSS21E is the weakest since 60 % of the mutants that were selected
364
by other VHHs and that were tested for cross-resistance against PVSS21E were positive.
365
PVSP29F (17 %), PVSP19B (18 %) and PVSS8A (46 %) have percentages between these two
366
extremes.
367 368
Given the similar binding sites of the different VHHs, we were concerned that the presence of
369
the original selecting VHH could interfere with the binding of other VHHs in the cross-
17 | P a g e
370
neutralization experiments. Thus, the cross-resistance test was done in the absence of the first
371
‘selector’ VHH. To test whether the variant virus, which formed in the wells where cross-
372
resistance was seen, was still the original mutant and not a newly selected genotype, these
373
variant ‘multiresistant’ viruses were once more tested against the original VHH. The
374
percentages of viable cells infected with the multiresistant virus and treated with its original
375
VHH are shown in Fig. 6. For example, 6A_2/8A is the virus that replicated in the well infected
376
with 6A_2 and treated with VHH PVSS8A. As controls, the mutant virus from the same cross-
377
resistance experiment and treated with its original selector VHH (e.g. 6A_2/6A) was also
378
measured in this experiment. No significant difference (α=0.05) was seen between any of the
379
variants which replicated in the absence of the original selector VHH as compared to the one
380
which replicated in the presence of the same VHH.
381 382
4. Discussion
383
Genotypes of the neutralization escape mutants
384
The sequencing of the P1 genome region of five plaque purified neutralization escape variants
385
of poliovirus type 1 for each of five in vitro neutralizing VHHs (PVSP6A, PVSS8A, PVSP19B,
386
PVSS21E and PVSP29F) lead to some interesting observations. First, aa-substitutions in all of the
387
four viral proteins were found. Even though most of the substitutions are located nearby the
388
binding sites of the VHHs on the virion (which was shown to be the canyon), six different
389
sustitutions (VP1 A43T, VP1 V47L, VP1 V61M, VP2 A121T, VP3 D209V and VP4 I62T) were found
390
to be located in regions of the capsid that are further away from the canyon and are more
391
internal in the capsid. The VP1 V47L and the VP4 I62T substitutions even occur as single aa-
18 | P a g e
392
substitutions and thus are fully responsible for the resistant phenotype of variants 8A_1 and
393
21E_5, respectively. Mutants resistant against conventional antibodies tend to mutate in
394
epitopes that are more exposed onto the surface of the capsid, e.g. in the loop regions at the
395
five-fold axis of symmetry (BC- and EF-loop of VP1) or the loops surrounding the canyon (EF-
396
loop of VP2 and GH-loop of VP1) (28-36). The dimensions of conventional antibodies
397
(approximately 10 times those of VHHs) usually allow them to recognize only these epitopes on
398
the capsid that are easily accessible and escape mutations are therefore clustered in defined
399
regions that are consistent with the area of the antigen binding site at the tip of each Fab. In
400
contrast, the aa-substitutions selected by the VHHs that are in exposed residues are at the
401
periphery of the footprint of the VHH rather than in the center of the site. Note that the GH-
402
loop of VP1 and the EF-loop of VP2 are within the VHH footprint, but are also highly exposed
403
components that belong to major antigenic sites, and are therefore the locations of escape
404
substitutions from both conventional antibodies and VHHs. The conformation and dimensions
405
of VHHs make it advantageous for them to interact inside the canyon in order to maximize the
406
contact area between the VHH and the virus. Since the binding sites of the VHHs and that of the
407
natural poliovirus receptor (CD155) (19) (Strauss et al, in preparation) are so similar, the virus is
408
now more limited in the residues it can mutate without affecting its infectivity. Thus, changing
409
the residues at the center of the VHHs’ binding sites would be likely to also have a deleterious
410
effect on receptor binding. The occurrence of aa-substitutions in internal and non-accessible
411
regions of the capsid also contrasts with the almost exclusive surface localization of escape
412
mutations selected by monoclonal antibodies. It should be noted, however, that the presence of
413
aa-substitutions outside the immediate binding site is not unique to the VHHs. Thus, Blondel
19 | P a g e
414
and colleagues have previously shown that mutations affecting neutralization by antibodies
415
recognizing the N2 antigenic site can be located both outside and inside the binding site (37)
416
and must therefore act by a mechanism other than simple steric interference with antibody
417
binding. In our data, some substitutions (VP1 A43T, VP1 V47L, VP1 V61M, VP4 I62T) are located
418
remarkably far from the binding site on the inner surface of the virion, and several others, that
419
may map closer to the VHH binding site (including VP1 A150V, VP2 A121T, and VP3 D209V) are
420
in residues in intersubunit interfaces that are not exposed on the surface, and are therefore also
421
unlikely to seriously impact binding of the VHHs to the capsid. This type of mutation seems to
422
be sufficient to permit resistance against the VHHs with the lowest poliovirus neutralizing
423
capacity (and highest EC50 values: PVSS8A and PVSS21E). For the other three VHHs, which have
424
lower EC50 values, the substitution of surface-exposed residues at the periphery of the binding
425
site seems to be essential to escape neutralization. Although most of these aa-substitutions are
426
in very close proximity to the VHH binding site and may actually directly contact the selective
427
VHH (19) (Strauss et al., in preparation), we would argue that the location at the periphery of
428
the binding site would be expected to produce at most marginal effects on VHH binding affinity.
429
Interestingly, in every case the substitutions occur in residues where the structure of the
430
particle is known to change upon the receptor-induced conformational changes associated with
431
cell entry. Based on the observation that in all cases binding VHHs stabilizes the virion, and that
432
all of the VHHs neutralize at copy levels below those required to prevent receptor binding, we
433
have previously argued that the VHHs neutralize virions by preventing conformational changes
434
required for cell entry (19). Taken together the data presented here, it is suggested that both
20 | P a g e
435
the internal and surface aa-substitutions may confer resistance to the VHHs at least in part by
436
interfering with the ability of the VHH to prevent conformational changes needed for cell entry.
437
It is interesting that when primate antibodies were identified that cross-neutralize both
438
poliovirus serotypes 1 and 2 (and thus avoid the usual highly exposed antigenic sites), the
439
antibody binding sites overlapped that of the receptor, and the escape substitutions included
440
residues 65, 109 and 166 of VP1 (38, 39).Those findings are clearly very similar to our results
441
using neutralizing VHHs, and suggest that the mechanisms of neutralization may be similar.
442 443
Characteristics of the selected mutants
444
When testing antiviral compounds, the possibility of developing therapy resistant variants
445
should always be considered and knowledge of the implications of these resistant variants is
446
very important. The single cycle replication kinetics of the mutants are very similar to those of
447
their parental Mahoney strain. This shows that the aa-substitutions found in the variants offer
448
little advantage or disadvantage for the variants, and that the variants will replicate to the same
449
extent as the parental strain in the absence of the VHH. This has also been observed for other
450
neutralization escape variants against poliovirus type 1, 2 and 3 (40, 41).
451
Previously, a number of mutations in poliovirus type 2 (41) and type 3 (12) selected for
452
resistance to neutralization by an antiviral drug were found to be dependent on the drug against
453
which resistance was developed. They were called drug-dependent variants. The similarity
454
between the single cycle replication curves of Mahoney and the mutants in the absence of
455
compound provides evidence that none of the variants depends on the presence of its
456
corresponding VHH in order to induce infection. However, this is not conclusive, as a drug-
21 | P a g e
457
dependent mutant against poliovirus type 3 showed similar single cycle replication
458
characteristics in the presence and absence of a drug (12), due to problems at the level of virus-
459
release from the cells in the absence of compound. A more specific method to determine drug
460
dependency is to calculate the drug plating index (DPI). This is the ratio between the titers of
461
the variant measured in the presence and the absence of the drug. Variants with a DPI between
462
0.001 and 0.01 (41) are only slightly resistant to the drug, those with a DPI around 1 are non-
463
dependent on the drug for infectivity and those with a DPI of 100 or higher are drug-dependent.
464
All of the variants characterized in this study have DPIs between 0.23 (21E_5) and 1.7 (8A_1)
465
and have therefore no aa-substitutions that destabilize the capsid to such an extent that the
466
VHH is essential for infection. The DPIs of the wild type Mahoney strain are all under 0.0048.
467
We have shown that binding of the VHHs stabilizes the virions against heat degradation (19).
468
We therefore hypothesized that the VHH-resistant escape substitutions would be destabilizing,
469
thereby counterbalancing the effect of the VHHs. On the contrary, all of the tested variants
470
were at least as stable against heat inactivation as the parental strain and most of them
471
appeared to be even more stable at 44°C. The test was performed three times for each variant,
472
but due to rather large variations, the difference was significant (P