Identification of a PA-Binding Peptide with Inhibitory Activity against Influenza A and B Virus Replication

There is an urgent need for new drugs against influenza type A and B viruses due to incomplete protection by vaccines and the emergence of resistance to current antivirals. The influenza virus polymerase complex, consisting of the PB1, PB2 and PA subunits, represents a promising target for the development of new drugs. We have previously demonstrated the feasibility of targeting the protein-protein interaction domain between the PB1 and PA subunits of the polymerase complex of influenza A virus using a small peptide derived from the PA-binding domain of PB1. However, this influenza A virus-derived peptide did not affect influenza B virus polymerase activity. Here we report that the PA-binding domain of the polymerase subunit PB1 of influenza A and B viruses is highly conserved and that mutual amino acid exchange shows that they cannot be functionally exchanged with each other. Based on phylogenetic analysis and a novel biochemical ELISA-based screening approach, we were able to identify an influenza A-derived peptide with a single influenza B-specific amino acid substitution which efficiently binds to PA of both virus types. This dual-binding peptide blocked the viral polymerase activity and growth of both virus types. Our findings provide proof of principle that protein-protein interaction inhibitors can be generated against influenza A and B viruses. Furthermore, this dual-binding peptide, combined with our novel screening method, is a promising platform to identify new antiviral lead compounds.


Introduction
Influenza A and B viruses are closely related RNA viruses, which cause respiratory disease in humans and other animals. Both viruses belong to the Orthomyxoviruses, a family of enveloped, negative-sense RNA viruses that encode their genetic information in eight segments. While influenza A viruses (FluA) infect a broad variety of animals as well as humans, influenza B viruses (FluB) are predominantly found in humans [1]. Recently, FluA viruses have attracted the world's attention due to their ability to cause global pandemics, such as the 1918 ''Spanish'' flu or the novel H1N1 influenza virus of swine origin [2,3,4]. However, both FluA and FluB viruses are responsible for seasonal epidemics, which result in hundreds of thousands of deaths each year, since FluB can cause the same spectrum of symptoms that is observed with FluA [5,6,7,8,9]. In most years, only one type causes the majority of the cases, leading to the exclusion of the others [1]. For this reason, the annual influenza vaccination recommended by the WHO contains both FluA and FluB strains (http://www.who. int/csr/disease/influenza/vaccinerecommendations1/en/). Thus, both types A and B viruses pose a large threat to public health.
Since the current vaccines against influenza viruses offer only incomplete protection, antivirals are greatly needed. Presently, the only class of anti-influenza drugs available which are active against both FluA and FluB are neuraminidase inhibitors oseltamivir (Tamiflu) and zanamivir (Relenza) [10]. However, many influenza A strains, including the circulating H1N1 strains in Europe and the US, are already resistant to oseltamivir [11,12], suggesting a limited range of use for this type of drug. The resistance is not restricted to FluA, since the emergence of FluB viruses with reduced sensitivity to neuraminidase inhibitors is known [13,14]. In addition, the treatment of influenza B, particularly in young children, is associated with delayed antiviral effects and clinical resolution [15]. This might be related to the observation that oseltamivir carboxylate, the active metabolite of oseltamivir, is less active against the FluB rather than the FluA virus neuraminidase [13]. Other approved antivirals that target the viral M2 protein, such as amantadine, are only effective against FluA strains, since FluB strains lack a comparable M2 protein [16]. Additionally, the high prevalence of resistance to M2 inhibitors and the general rapid appearance of resistant strains during treatment restricts the therapeutic potential of this class of antiviral drugs [17].
Influenza virus replication and transcription is performed in the nucleus by the heterotrimeric viral polymerase complex consisting of the PB1, PB2, and PA subunits. This includes the synthesis of the genomic RNA from an intermediate copy RNA and the synthesis of viral mRNA transcripts. Synthesis of the latter transcripts is dependent on 8-15 nt long primers derived from the 59ends of capped endogenous mRNAs. PB1 possesses the RNA polymerization activity, whereas PB2 is known to bind capped mRNA [18,19,20]. Recently, RNAse activity could be identified in the N-terminal part of PA, which is most likely necessary to cleave the primer off the remaining host mRNA [21,22]. The structure of the trimeric complex at the atomic level is not yet resolved, but cryo-EM studies do exist [23,24]. Recent crystal structures have shown that PB1 binds to PA via a 3 10 -helix at its N-terminus comprised of amino acids (aa) 5-11 [25,26]. This finding is compatible with previous observations that the PA-binding site is localized to the extreme N-terminus of PB1 [27,28]. Since the association of these subunits is essential for viral replication [28,29,30,31], and since the sequence of this domain of PB1 is highly conserved among FluA strains [32], this interaction presents an attractive target for antiviral drugs. We have previously shown that a peptide derived from the first 25 amino acids of FluA PB1 can inhibit both the polymerase activity and viral spread of FluA but not FluB [32]. Here, we identify a chimeric FluA/FluB peptide which is active against both virus types.

Results
The type-specificity of the PA-PB1 interaction is due to the first 25 amino acids of PB1 Previously, we have shown that the N-terminal 25 aa of FluA PB1 (PB1 1-25 A) are sufficient to bind FluA-PA and inhibit the polymerase activity of FluA [32]. However, this sequence was not able to block the polymerase activity of FluB. Since the first 25 aa of FluA-PA and FluB-PA differ at 8 positions (Fig. 1A), the failure to block FluB polymerase activity might be due to impaired binding of PB1 1-25 A to FluB PA. Alternatively, the PA-binding domain of FluB PB1 might extend beyond the extreme 25 amino acids. To test this, we generated FluA/B PB1 chimeras by substituting N-terminal regions of varying length from FluA-PB1 into FluB-PB1 (Fig. 1B). After transient co-expression in human 293T cells, we then used immunoprecipitation assays to test which of these chimeras could bind to FluA-PA. All PB1 chimeras tested, including a chimera containing only the first 25 aa of FluA-PB1, could bind FluA-PA equally, whereas wild-type FluB-PB1 could not (Fig. 1C). This implied that the few aa differences in this small domain are the major determinant in PB1 for the type-specific binding of PB1 to PA.

Quantification of the PA/PB1 interaction
To further characterize which amino acid positions are important for the type-specific binding of PB1 1-25 to PA, we developed an ELISA-based binding assay to quantify this interaction. For this purpose, we immobilized biotinylated peptides corresponding to PB1 1-25 on streptavidin-coated plates and subsequently added cell lysates containing HA-tagged PA. Bound PA was detected using HA-specific primary and peroxidase-coupled secondary antibodies. To determine the binding efficiency of variant peptides we performed the ELISA in a competitive manner. In this case, non-biotinylated competitor peptides were added simultaneously with HA-tagged PA to wells containing immobilized PB1 1-25 peptide.
The core-binding site of the PA-binding domain of FluA-PB1 was recently determined by co-crystalization with the C-terminus of PA and is comprised of a 3 10 -helix ranging from aa 5-11 [25]. We therefore first evaluated our FluA-binding assay by using a soluble wild-type PB1 1-25 A peptide, which resulted in a 50% inhibitory concentration (IC 50 ) of 1.8 nM ( Fig. 2A). A ''scrambled peptide'' harboring the same amino acid composition as PB1 1-25 A but with the order randomized, did not block PA-HA binding at highest concentrations (3000 nM, data not shown). Next, the minimal binding domain was determined by deleting aa from the N-and/or C-terminus of PB11-25A. Determination of the IC 50 using the truncated competitor peptides demonstrated that the 3-10 helix is absolutely essential for binding to PA, since any truncation of this sequence resulted in no competition (Fig. 2B). This is congruent with the results obtained by mutational analysis of the PA-binding domain [28]. Residual competition from partially C-or N-truncated peptides was observed, implying that some of the aa positions flanking the 3-10 helix may be dispensable for binding. Specifically, truncating the C-terminal 10 aa of FluA PB1 1-25 A resulted in an IC 50 of 43.3 nM compared to 1.8 nM for the full-length peptide (Fig. 2B). Since this truncated peptide can efficiently bind to PA, yet still contains many of the differing aa positions between FluA and FluB, we chose this sequence as a basis to test variant peptides for their binding to both FluA-and FluB-PA.
A single amino acid substitution allows PB11-25 to bind FluA PA and FluB PA As the PB1 1-15 sequence differs by 8 aa between FluA and FluB ( Fig. 1A), we decided to create chimeras containing combinations of FluA-and FluB-specific residues and test these 15mer peptides in the competitive ELISA using both FluA-PA and FluB-PA. For this assay, biotinylated PB1 1-25 A was used in combination with FluA-PA-HA, biotinylated PB1 1-25 B was combined with FluB-PA-HA, and the competitor peptides were tested with both combinations. Competitor peptides PB1 1-15 A and PB1 1-15 B, derived from the wild-type sequences, competed only for FluA PA or FluB PA, respectively, with no cross-type competition detectable (Fig. 3). We then tested competitor peptides containing combinations of FluBspecific aa in the PB1 1-15 A background. While most of these combinations showed no affinity for FluB, substituting positions 6 and 7 led to a significantly reduced IC 50 for FluB-PA as well as a slight decrease in the IC 50 for FluA-PA, suggesting that this peptide could bind to both proteins. By substituting only the FluB-specific tyrosine at position 6, efficient binding to PA of both types resulted. In contrast, changing only position 7 of a FluA-to a FluB-specific sequence ablated binding to PA of both types. Introduction of a phenylalanine or tryptophan at position 6 demonstrated that related residues can significantly increase the binding affinity to FluA PA, but only partially increase the affinity to FluB PA. A histidine or cysteine at position 6 led to peptides failing to bind to FluB PA, while binding to FluA PA was still possible, indicating that not all amino acid substitutions confer the dual-binding activity observed with tyrosine at this position.
To further confirm that tyrosine at position 6 of FluA PB1 can bind directly to both FluA-and FluB-PA, we used a biotinylated 25mer peptide PB1 1-25 A T6Y as well as wild-type peptides and cell lysates containing overexpressed PA-HA derived from several different FluA and FluB strains in the ELISA. As shown in Fig.  4A-C, PA-HA from several FluA strains of various subtypes, including a human H5N1 isolate, bound to both PB1 1-25 A and PB1 1-25 A T6Y , but not to PB1 1-25 B or a scrambled biotinylated peptide. Conversely, PA-HA derived from several different FluB strains bound to both the PB1 1-25 B and PB1 1-25 A T6Y (Fig. 4D). Thus, a FluA-PB1-derived sequence containing a single substitution at position 6 can directly bind to PA from different FluA and FluB viruses. Molecular modeling suggests that the FluB-derived tyrosine at position 6 of PB1 1-25 A T6Y fits into a hydrophobic pocket of FluA PA and displaces a water molecule (Fig. 3F). Both the increased hydrophobic interaction as well as the entropic effect of the water displacement might explain the enhanced binding of PB1-T6Y to PA of FluA and possibly of FluB.

PB 11-25 A T6Y can inhibit the polymerase activity and spread of FluA and FluB
To demonstrate that PB1 1-25 A T6Y can bind to both FluA-and FluB-PA in a cellular context, we performed immunoprecipitation experiments using overexpressed PA and PB1 1-25 peptides fused to GFP (Fig. 5B). This confirmed that PB1 1-25 derived from FluA or FluB can only bind to FluA-PA or FluB-PA, respectively. However, PB1 1-25 A T6Y could be immunoprecipitated with either FluA-PA or FluB-PA.
Next, we tested the ability of these fusion proteins to inhibit the polymerase activity of both FluA and FluB using a reconstituted viral ribonucleoprotein complex and a reporter gene. As previously shown [32], PB1 1-25 A-GFP inhibits the activity of FluA, but not FluB, while the converse is true of PB1 1-25 B-GFP (Fig. 5C). However, PB1 1-25 A T6Y -GFP inhibited both FluA and FluB polymerase activity to comparable levels as the wild-type-derived sequences (Fig. 5C), indicating that this variant peptide is also biologically active.
Finally, we tested the ability of these peptides to inhibit viral spread in cell culture. MDCK cells were infected with either A/ WSN/33 (H1N1), A/Thailand/1/2004 (H5N1), B/Yamagata/ 73, or vesicular stomatitis virus (VSV) as a specificity control from a different virus family. A plaque reduction assay was then performed in the presence of PB1 1-25 A, PB1 1-25 A T6Y , or a negative-control peptide fused to an HIV-Tat derived sequence [33] (Table 1) to allow the peptides to cross the plasma membrane. Similar experiments with a FluB-derived peptide were not possible due to its insolubility. The peptide containing the T6Y substitution inhibited the growth of all influenza strains tested, while PB1 1-25 A-Tat only inhibited FluA (Fig 6). Thus, a single mutation in the PB1 1-25 A sequence results in a peptide which is active against both FluA and FluB strains.

Discussion
Here we report that the PA-binding domains of the polymerase subunits PB1 of influenza A and B viruses are highly conserved within each virus type, and that these domains are not mutually exchangeable. Based on phylogenetic analysis and a biochemical ELISA-based screening approach, we were able to identify a PB1derived peptide of influenza A virus with a single influenza Bspecific amino acid substitution which recognizes PA of both virus types. Significantly, this dual-binding peptide blocked both the polymerase activity and spread of both viruses and raises the exciting possibility of developing new antivirals that specifically interfere with the polymerase complex assembly of both viruses.
The inability of PB1 1-25 A to efficiently bind to FluB-PA and vice versa strongly indicates that these PA-binding domains show limited functional compatibility between the two virus types. We speculate that this incompatibility is caused by an independent evolution of the FluA and FluB polymerase proteins. This is in accordance with a virus-type specific conservation of the PAbinding domain of PB1 (Fig. 1) and further highlighted by an overall aa identity of the viral polymerase subunits of less than 35% between FluA and FluB strains. Together, these differences may account for the failure to reconstitute an active trimeric polymerase complex by mixing FluA and FluB polymerase subunits ( [34,35,36], unpublished results). However, both FluA and FluB viruses have a common origin [1]. Thus, it is not unexpected that the aa residues of PA which are in close contact with PB1 1-25 are conserved between FluA and FluB [25,26], indicating that the PA domain of both virus types bind PB1 in a similar fashion. This similarity might explain how one single amino acid substitution (T6Y) in PB1 1-25 A is sufficient to allow binding to FluB-PA. Based on the available crystal structure of PA complexed with the N-terminus of PB1 and our modeling results with this dual-binding peptide PB1 1-25 A 6Y (Fig. 4F), we speculate that the tyrosine residue might exploit a hydrophobic pocket in the PB1-binding domain of FluB PA, thereby stabilizing the interaction. Such an occupation of a hydrophobic pocket may also be important to stabilize the interaction between PB1 1-25 A 6Y and FluB PA.
Both the identification of the FluA-and FluB-PA-binding peptide PB1 1-25 A T6Y and our novel ELISA-based screening assay provide a unique platform for future screening of small molecules targeting the PB1/PA interaction of both virus types. Such inhibitors represent attractive new tools to disrupt a variety of protein-protein interactions (PPI) [37,38]. Initial attempts to identify PPI inhibitors were challenged by the large contact surfaces of protein-protein interfaces. However, the identification of a small subset of aa residues, such as the PA-binding domain of PB1 that contribute to most of the free energy of binding provide excellent targets. Recent screening approaches led to the identification of potent PPI inhibitors that block dimerization between the human protein double mutant 2 (HDM2) and a 15-residue a-helical region of the tumor suppressor p53 [39], the basic helix-loop-helix leucine zipper (bHLHZip) transcription factor Myc and Max [40] or the interaction between paxilin and the integrin a4 [41]. Proof-ofprinciple that viral protein interactions are also attractive targets for the generation of PPI inhibitors was first obtained with human papilloma virus [42] and herpes simplex virus type 1 (HSV-1) [43]. In the case of HSV-1, the interaction between the catalytic subunit of the HSV-1 polymerase (UL30) and its cofactor, the processivity factor UL42 was targeted. Similar to the interaction of PB1 and PA, the extreme 18 C-terminal residues of UL30 are sufficient to bind to UL42. Using a biochemical screening assay based on the specific interaction of this C-terminal peptide and UL42, several potent small molecule inhibitors were identified [43]. It is therefore reasonable to assume that PPI inhibitors that block the dimerization of PB1 and PA can also be identified using the newly established ELISA-based screening assay. Indeed, recent results indicate that this approach has the potential to identify PPI inhibitors that prevent viral polymerase activity and replication of both types of influenza viruses (Kessler et al., unpublished results). We believe that the PB1/PA interface is an excellent target for PPI inhibitors, because the PA-binding domain within the extreme N-terminus of PB1 is crucial for the interaction with PA, since PB1 mutants lacking this domain fail to interact with PA [28]. Using PB1 1-25 A 6Y and FluA PA for screening, we expect that some PPI inhibitors possess a high specificity against both influenza virus types. This is supported by our observations that PB1 1-25 A 6Y blocks the growth of FluA and FluB but not VSV.
Besides the PA/PB1 complex, other targets of influenza viruses have become attractive for the development of PPI inhibitors, including the interaction interface between PB2 and PB1, for which the crystal structure was recently solved [44]. Additionally, based on the crystal structure of a fragment of PB2 bound to the cap analog m7GTP [20], non-PPI inhibitors might become feasible that compete with cap-binding properties of PB2. Furthermore, the  cap-snatching endonuclease of influenza virus polymerase localized in the N-terminal domain of PA [21,22] provides a further target for the development of effective inhibitors. However, it has yet to be shown whether these targets are suitable for the development of antivirals. Since the available information of these novel targets is restricted to FluA strains, it remains unclear whether such antivirals may also be active against FluB.
Attempts to substitute position 6 of the FluA PB1 1-25 A peptides with other residues than tyrosine did not result in improved FluB-PA binding in the competitive ELISA experiments. However, replacements with either phenylalanine or tryptophan residues enhanced binding to FluA-PA approximately 20 fold. This suggests that other substitutions may further enhance the binding affinity to FluA-PA and increases the antiviral activity of these peptides. In one study, random substitutions of single aa within an importin-alpha-binding peptide increased its affinity by several orders of magnitude [45], providing proof of principle that the development of potent peptide inhibitors from a peptide lead is possible. The antiviral activity of VIRIP, a peptide that blocks HIV-1 entry by interacting with the gp41 fusion peptide could be increased in its antiretroviral potency by two orders of magnitude [46]. Thus, random substitutions of the PA-binding domain may  reveal peptides with significantly enhanced binding properties for FluA-PA. This approach might also lead to the identification of peptides, which bind PA of both virus types, using PB1 1-25 A 6Y as the lead peptide. Most importantly, such peptides would be prime candidates for the development of antiviral peptidomimetics.

Plasmid constructions
Plasmids pCA-Flag-GFP and pCA-PB1 1-25 A-GFP [32], pCA-PB1-HA [32], the FluA minireplicon plasmids [50] and the expression plasmids for the FluB minireplicon [32] were described elsewhere. The FluB minigenome expression plasmid, pPolI-lucRT_B, was obtained by cloning the firefly luciferase ORF (inverse orientation) flanked by the non-coding region of the segment 8 of the B/Yamagata/73 into the SapI-digested plasmid pPolI-SapI-Rib [50].  )) were PCR amplified with sense primers containing a NotI site (FluA strains) or a EcoRI site (FluB strains) upstream of the initiation codon and antisense primers with a deleted stop codon followed by an XmaI site, a coding sequence for an HA-tag and a XhoI site. The PCR products were cloned into a modified pCAGGs-vector [51] digested either with EcoRI/XhoI or NotI/XhoI, resulting in pCA-PB1-HA or pCA-PA-HA plasmids, coding for C-terminal tagged versions of the polymerase subunits. To obtain the pCA-PA A/SC35M -His plasmid, pCA-PA A/SC35M -HA was digested with XmaI/XhoI and the HA coding sequence was replaced by a 6xHis-linker. The A/B-chimeric expression plasmids were obtained by assembly PCR using the pCA-PB1-HA plasmids of SC35M and B/Yamagata/73 and by cloning the resulting PCR product in pCA-PB1 B/Yamagata/73 -HA digested with EcoRI/EcoRV.

Reconstitution of the influenza virus polymerase activity
293T cells were transiently transfected with a plasmid mixture containing either FluA-or FluB-derived PB1-, PB2-, PA-and NPexpression plasmids, polymerase I (Pol I)-driven plasmid transcribing an influenza A or influenza B virus-like RNA coding for the reporter protein firefly luciferase to monitor viral polymerase activity and with expression plasmids coding for the indicated GFP fusion proteins. Both minigenome RNAs were flanked by noncoding sequences of segment 8 of FluA and FluB, respectively. The transfection mixture also contained a plasmid constitutively expressing Renilla luciferase, which served to normalize variation in transfection efficiency. The reporter activity was determined 24 h post transfection and normalized using the Dual-GluH Lufierase Assay System (Promega). The activity observed with transfection reactions containing Flag-GFP were set to 100%.

Peptide synthesis
The solid-phase synthesis of the peptides, including the peptides for the plaque reduction assay (Table 1), was carried out on a Pioneer automatic peptide synthesizer (Applied Biosystems, Foster City, USA) employing Fmoc chemistry with TBTU/diisopropylethyl amine activation. Side chain protections were as follows: Asp, Glu, Ser, Thr and Tyr: t-Bu; Asn, Gln and His: Trt; Arg: Pbf; Lys and Trp: Boc. Coupling time was 1 h. Double couplings were carried out if a difficult coupling was expected according to the program Peptide Companion (CoshiSoft/PeptiSearch, Tucson, USA). All peptides were generated as carboxyl amides by synthesis on Rapp S RAM resin (Rapp Polymere, Tübingen, Germany). Biotin was incorporated at the C-terminus of indicated peptides with Fmoc-Lys(Biotin)-OH (NovaBiochem/Merck, Nottingham, UK) and TBTU/diisopropylethyl amine activation for 18 h, followed by coupling of Fmoc-b-Ala-OH for 1 h. Peptides were cleaved from the resin and deprotected by a 3 h treatment with TFA containing 3% triisobutylsilane and 2% water (10 ml/g resin). After precipitation with t-butylmethyl ether, the resulting crude peptides were purified by preparative HPLC (RP-18) with water/acetonitrile gradients containing 0.1% TFA and characterized by analytical HPLC and MALDI-MS. Some peptides were synthesized by peptides&elephants (Nuthetal, Germany) and subsequently purified and characterized as described above.

Immunoprecipitation experiments
293T cells were transfected with the indicated plasmids in 6-well plates using Metafectene (Biontex, Martinsried, Germany). Cells were incubated 24 h post transfection with lysis buffer (20 mM Tris pH 7.5, 100 mM NaCl, 0.5 mM EDTA, 0.5% NP-40, 1% Protease inhibitor Mix G, (Serva, Heidelberg, Germany), 1 mM DTT) for 15 min on ice. After centrifugation by 13.000 rpm at 4uC supernatant was incubated with anti HA-specific antibodies coupled to agarose beads (Sigma) for 1 h at 4uC. After three washes with 1 ml of washing buffer (lysis buffer without protease inhibitor mix), bound material was eluted under denaturing conditions and separated on SDS-PAGE gels and transferred to PVDF membranes. Viral polymerase subunits and GFP fusion proteins were detected with antibodies directed against the HA-(Covance, Berkeley, California) or His-(Qiagen) or GFP-tag (Santa Cruz Biotechnology).
Plaque reduction assay. The experiments were carried out as described [52] with modifications. Confluent MDCK cells were infected with 80 PFU of A/WSN/33, B/Yamagata/73, A/KAN-1, or VSV/Indiana in PBS at room temperature. After removal of the inoculums, cells were overlaid with medium (DMEM with 20 mM Hepes) containing 1% Oxoidagar and peptides at the indicated concentrations. After incubation for 24 h (VSV), 48 h (A/ WSN/33, A/KAN-1) at 37uC with 5% CO 2 , or 72 h at 33uC with 5% CO 2 (B/Yamagata/73) respectively, cells were fixed with formaldehyde and stained with crystal violet. Plaques were counted and mean plaque number of the water control was set to 100%.

ELISA
Streptavidin-coated microwell plates were incubated with saturating concentrations of biotinylated peptides, washed and subsequently incubated at room temperature with HA-tagged PA. To obtain PA-HA, 293T cells were seeded into 94 mm-dishes, transfected with the respective plasmid and treated with lysis buffer 24 h post transfection as previously described [32]. After washing the microwell plates, the wells were incubated with an HA-specific primary antibody (Covance, MMS-101R), followed by three washes and an incubation with a peroxidase-coupled secondary antibody (Jackson Immuno Research, Newmarket, UK). After the final wash step, ABTS-substrate (Sigma, ready-to-use solution) was added and the optical density was determined at 405 nm.
The competition ELISA was carried out as described above with the exception that the competitor peptides were added to wells of the plate with bound peptides simultaneously with the addition of the cell extract containing HA-tagged PA subunits.
Modeling. Manual docking of the mutated peptide into the PA(C)-PB1(N) crystal structure [25] and subsequent minimization was performed with Accelrys Discovery Studio.