Characterization of Hemagglutinin Negative Botulinum Progenitor Toxins

Botulism is a disease involving intoxication with botulinum neurotoxins (BoNTs), toxic proteins produced by Clostridium botulinum and other clostridia. The 150 kDa neurotoxin is produced in conjunction with other proteins to form the botulinum progenitor toxin complex (PTC), alternating in size from 300 kDa to 500 kDa. These progenitor complexes can be classified into hemagglutinin positive or hemagglutinin negative, depending on the ability of some of the neurotoxin-associated proteins (NAPs) to cause hemagglutination. The hemagglutinin positive progenitor toxin complex consists of BoNT, nontoxic non-hemagglutinin (NTNH), and three hemagglutinin proteins; HA-70, HA-33, and HA-17. Hemagglutinin negative progenitor toxin complexes contain BoNT and NTNH as the minimally functional PTC (M-PTC), but not the three hemagglutinin proteins. Interestingly, the genome of hemagglutinin negative progenitor toxin complexes comprises open reading frames (orfs) which encode for three proteins, but the existence of these proteins has not yet been extensively demonstrated. In this work, we demonstrate that these three proteins exist and form part of the PTC for hemagglutinin negative complexes. Several hemagglutinin negative strains producing BoNT/A, /E, and /F were found to contain the three open reading frame proteins. Additionally, several BoNT/A-containing bivalent strains were examined, and NAPs from both genes, including the open reading frame proteins, were associated with BoNT/A. The open reading frame encoded proteins are more easily removed from the botulinum complex than the hemagglutinin proteins, but are present in several BoNT/A and /F toxin preparations. These are not easily removed from the BoNT/E complex, however, and are present even in commercially-available purified BoNT/E complex.


Introduction
Botulinum neurotoxins (BoNTs) are toxic proteins created by Clostridium botulinum and some related clostridia. Exposure to BoNT causes the disease known as botulism, typified by a descending flaccid paralysis which can be fatal if untreated. BoNT is comprised of two chains joined by a disulfide bond-a heavy chain which binds to receptors of neurons, and a light chain that cleaves proteins needed for nerve impulse transmission. BoNTs consist of seven confirmed serotypes, A-G, based upon antisera response. Most serotypes also display genetic and amino acid variance inside the serotype, and this variance is defined as subtype, designated by a number after the letter, as in subtype BoNT/E3. Most C. botulinum strains produce only one serotype of toxin; however, some strains of Our laboratory decided to utilize our experience in proteomic analyses of botulinum neurotoxins to examine the protein composition of several hemagglutinin negative botulinum progenitor toxin complexes. In this work, we examined the protein composition of BoNT/E3, /F1, /A1, /A2, and /A3 progenitor toxin complexes, all of which are orfX+. We also studied progenitor toxin complexes from three bivalent C. botulinum strains; BoNT/A1(B), BoNT/A2b, and /A2f, which each contain at least one orfX+ toxin gene cluster in combination with either a HA+ toxin gene cluster (BoNT/A1(B) and /A2b) or another orfX+ toxin gene cluster (BoNT/A2f). We found evidence for the existence of the proteins encoded by orfX1, orfX2, and/or orfX3 in many orfX+ progenitor toxin complexes that we studied.

Results
The first goal was to ensure that the proposed analytical technique would capture the true nature of a botulinum progenitor toxin complex. HA+ progenitor toxin complexes are well characterized, so the commercially-available BoNT/A1 Hall toxin complex was spiked into blank culture supernatant medium, then immunoextracted, the captured toxin complex was digested, and the resultant peptides were identified by LC-MS/MS analyses. Table 1 includes a list of all proteins identified through this process with their respective % coverages. NCBI accession numbers of the proteins identified are listed in Table S1 of the Supplementary Information. The PTC of C. botulinum HA+ A1 strains have been previously identified as containing BoNT/A1, NTNH, HA-70, HA-33, and HA-17 [24]. Our analyses demonstrated that all five of those proteins were present, with % coverages of 89, 90, 82, 49, and 46, respectively. Percent coverage refers to the number of amino acids from the identified peptides divided by the entire number of amino acids of the protein. Many peptides were identified which are unique to HA proteins. The MS/MS of three peptides unique to HA proteins originating from HA-70, HA-33, and HA-17 are in Figure 1.
Toxins 2017, 9,192 3 of 14 /A2, and /A3 progenitor toxin complexes, all of which are orfX+. We also studied progenitor toxin complexes from three bivalent C. botulinum strains; BoNT/A1(B), BoNT/A2b, and /A2f, which each contain at least one orfX+ toxin gene cluster in combination with either a HA+ toxin gene cluster (BoNT/A1(B) and /A2b) or another orfX+ toxin gene cluster (BoNT/A2f). We found evidence for the existence of the proteins encoded by orfX1, orfX2, and/or orfX3 in many orfX+ progenitor toxin complexes that we studied.

Results
The first goal was to ensure that the proposed analytical technique would capture the true nature of a botulinum progenitor toxin complex. HA+ progenitor toxin complexes are well characterized, so the commercially-available BoNT/A1 Hall toxin complex was spiked into blank culture supernatant medium, then immunoextracted, the captured toxin complex was digested, and the resultant peptides were identified by LC-MS/MS analyses. Table 1 includes a list of all proteins identified through this process with their respective % coverages. NCBI accession numbers of the proteins identified are listed in Table S1 of the Supplementary Information. The PTC of C. botulinum HA+ A1 strains have been previously identified as containing BoNT/A1, NTNH, HA-70, HA-33, and HA-17 [24]. Our analyses demonstrated that all five of those proteins were present, with % coverages of 89, 90, 82, 49, and 46, respectively. Percent coverage refers to the number of amino acids from the identified peptides divided by the entire number of amino acids of the protein. Many peptides were identified which are unique to HA proteins. The MS/MS of three peptides unique to HA proteins originating from HA-70, HA-33, and HA-17 are in Figure 1.  Toxins 2017, 9,193 6 of 15 The correct identification of the proteins in the progenitor toxin complex of BoNT/A1 Hall demonstrated that this analytical technique was an appropriate tool to characterize progenitor toxin complexes. The next step was to examine a commercially-available orfX+ toxin complex. BoNT/E3 complex was spiked into blank culture supernatant medium, immunoextracted, the captured toxin complex was digested, and the resultant peptides were identified by LC-MS/MS analyses. Using this process, five proteins were identified as part of the toxin complex: BoNT/E3, NTNH from BoNT/E, Orf-X1, Orf-X2, and Orf-X3 with % sequence coverages of 88, 28, 74, 32, and 60, respectively, as listed in Table 1. The existence of these proteins is indicated through MS/MS analysis, and MS/MS of one peptide from each of the Orf-encoded proteins is in Figure 2. All three of these MS/MS analyses are from peptides which are unique to Orf-encoded proteins. Additionally, we looked at the only other commercially-available orfX+ botulinum, the BoNT/F1 complex. There were four proteins identified as part of the complex; BoNT/F1, NTNH, Orf-X2, and P47, as listed in Table 1. The correct identification of the proteins in the progenitor toxin complex of BoNT/A1 Hall demonstrated that this analytical technique was an appropriate tool to characterize progenitor toxin complexes. The next step was to examine a commercially-available orfX+ toxin complex. BoNT/E3 complex was spiked into blank culture supernatant medium, immunoextracted, the captured toxin complex was digested, and the resultant peptides were identified by LC-MS/MS analyses. Using this process, five proteins were identified as part of the toxin complex: BoNT/E3, NTNH from BoNT/E, Orf-X1, Orf-X2, and Orf-X3 with % sequence coverages of 88, 28, 74, 32, and 60, respectively, as listed in Table 1. The existence of these proteins is indicated through MS/MS analysis, and MS/MS of one peptide from each of the Orf-encoded proteins is in Figure 2. All three of these MS/MS analyses are from peptides which are unique to Orf-encoded proteins. Additionally, we looked at the only other commercially-available orfX+ botulinum, the BoNT/F1 complex. There were four proteins identified as part of the complex; BoNT/F1, NTNH, Orf-X2, and P47, as listed in Table 1. Next, toxins from culture supernatants were immunocaptured and investigated. The digest of C. botulinum strain A1 orfX+ CDC 297 contained two proteins: BoNT/A1 and NTNH with sequence coverages of 59% and 37%, respectively, as listed in Table 1. MS/MS data was able to differentiate the /A1 orfX1+ neurotoxin from other closely related BoNT/A1 neurotoxins. Similar procedures were followed for BoNT/A2 and /A3 with comparable results. In those cases, MS/MS data identified the presence of the neurotoxin and NTNH as the constituents of the progenitor toxin complex, as seen in Table 1. Sequence coverages of 59% and 56%, respectively, were obtained for BoNT/A2; whereas coverages were lower for BoNT/A3, with 35% and 18%, respectively. Nonetheless, the MS/MS data allowed for an identification of the correct neurotoxin in these cases. The digest of C. botulinum strain ATCC 9564 produced similar results to that of commercially-available complex toxin, namely, that BoNT/E1, NTNH, Orf-X1, Orf-X2, and Orf-X3 were all present with sequence coverages of 91.3%, 50.5%, 85.4%, 55.1%, and 70.3%, respectively. Next, toxins from culture supernatants were immunocaptured and investigated. The digest of C. botulinum strain A1 orfX+ CDC 297 contained two proteins: BoNT/A1 and NTNH with sequence coverages of 59% and 37%, respectively, as listed in Table 1. MS/MS data was able to differentiate the /A1 orfX1+ neurotoxin from other closely related BoNT/A1 neurotoxins. Similar procedures were followed for BoNT/A2 and /A3 with comparable results. In those cases, MS/MS data identified the presence of the neurotoxin and NTNH as the constituents of the progenitor toxin complex, as seen in Table 1. Sequence coverages of 59% and 56%, respectively, were obtained for BoNT/A2; whereas coverages were lower for BoNT/A3, with 35% and 18%, respectively. Nonetheless, the MS/MS data allowed for an identification of the correct neurotoxin in these cases. The digest of C. botulinum strain ATCC 9564 produced similar results to that of commercially-available complex toxin, namely, that BoNT/E1, NTNH, Orf-X1, Orf-X2, and Orf-X3 were all present with sequence coverages of 91.3%, 50.5%, 85.4%, 55.1%, and 70.3%, respectively.
Some of the bivalent toxin producers contain mixed toxin gene clusters. An interesting example is BoNT/A1(B), as it contains a BoNT/A1 with an orfX+ gene cluster and a BoNT/B which is not produced in its entirety due to a stop codon mutation in the gene sequence of the neurotoxin. The "silent" bont/B gene is part of a cluster which contains ntnhB and ha genes. Beads coated with anti-BoNT/A were added into the culture supernatant, and the extracted toxin was digested. BoNT/A1 was identified with 81% sequence coverage, along with a number of other proteins. NTNH associated with BoNT/A1 was identified (50% sequence coverage) as was NTNH associated with BoNT/B (71% sequence coverage). The sequence alignment and sequence coverage of these two proteins are seen in Figure 3. MS/MS of similar peptides YDQFYVDPALELIK from the BoNT/A NTNH ( Figure 4A) and YDEFYIDPAIELIK from the BoNT/B NTNH ( Figure 4B) demonstrate that both proteins are extracted with the immunocaptured BoNT/A1 and both of these peptides are similar, yet different enough for differentiation. Additionally, all three hemagglutinin proteins associated with BoNT/B were identified with the immunocaptured BoNT/A1 with specific sequence coverages: HA-70 (44%), HA-33 (28%), and HA-17 (19%), as well as Orf-X2 (14%) from the BoNT/A gene cluster.
Prior to this point, investigation of the hemagglutinin negative botulinum neurotoxin progenitor toxins consisted of immunoaffinity extraction of the progenitor toxin complex, followed by washes with PBST, a mild buffer. Previous work in our laboratory has established that washing the antibody-coated beads with 2 M NaCl, a stringent buffer, results in a decrease of proteins which are non-specifically bound [25]. Three hemagglutinin negative botulinum neurotoxin PTC (CDC 2357, SU1304, and ATCC 9564) were examined following washing with the stringent 2 M NaCl buffer, and the results were found to be different than progenitor toxins washes obtained previously under less stringent conditions for CDC 2357 and SU 1304. Specifically, in both cases, there is mass spectrometric evidence for the existence of the neurotoxin, the NTNH, and in the case of HA+ toxins, HA-70, HA-33, and HA-17 proteins, but there is no evidence for any of the proteins encoded by open reading frames Orf-X1, Orf-X2, or Orf-X3. All three of those proteins were present, however, in the progenitor toxin complex of ATCC 9564 even with stringent washing.

Discussion
Although the role of the classic neurotoxin-associated proteins (NAPs) is not completely understood, it is believed that these proteins assist in increasing the toxicity of botulinum neurotoxin when administered through the oral route [1]. These proteins may serve a role in shielding the neurotoxin from the severe conditions of the gastrointestinal tract and may aid in moving the neurotoxin across the intestinal epithelium through interaction with E-cadherin [2,3,7]. Accurate identification of the proteins present during each stage of intoxication would help to understand this process more fully. Additionally, because the composition of the neurotoxin-associated proteins differs between serotypes, and in some cases, subtypes, accurate identification of the neurotoxin-associated proteins can assist in further verification on the identification of the serotype/subtype of neurotoxin.
This study examined the composition of several hemagglutinin negative botulinum neurotoxin complexes, and in most cases, there was evidence for at least one of the proteins encoded by open reading frames. In the case of BoNT/E3, we observed proteins corresponding to orfX1, orfX2, and orfX3, all of the three open reading frames. The existence of a protein encoded by orfX1 in this complex is supported by a 2005 report of the same finding [14]. Advances in analytical techniques and instruments might explain why the 2005 report mentioned nothing about the existence of the proteins encoded by the other open reading frames, orfX2 and orfX3. Additionally, others have noted that the complex of BoNT/E although HA negative, is similar in size to HA positive complexes such as HA positive BoNT/A [17]. If the presence of a protein is not observed, this does not mean that the protein is not present. Indeed, although we did not observe orfX proteins in the extracted complexes in some of the strains studied here such as CDC 297, SU 1887, and Loch Maree, this does not mean that orfX proteins are not actually present in those complexes, but could not be detected at this point in time. Additionally, this is not the first report of the existence of a protein encoded by orfX2, although previous reports have found the Orf-X2 in the progenitor toxin complex [15,16], but absent in the purified toxin complex [15]. Our findings support the existence of the proteins encoded by open reading frames in hemagglutinin negative PTC and our findings from one experiment (exposure to 2 M NaCl) also support the absence of these proteins in purified toxin complex, as reported previously [15].
The creation of purified toxin complex from progenitor toxin is a rigorous process involving acid precipitation to concentrate the toxin, solubilization in sodium phosphate buffer, alkaline treatment, and precipitation with ammonium sulfate [26]. This purified toxin complex has been reported to be absent of any proteins encoded by open reading frames, at least in the case of BoNT/A2 [15], and it is likely that these proteins are removed from the toxin complex during its treatment with 3 N sulfuric acid and 1 N sodium hydroxide, in the same fashion that those proteins are removed from the toxin complex during exposure to 2 M NaCl in our experiments shown here. Of note, the wash steps are at neutral pH, and the complex is thought to dissociate at this pH [4]. However, in this study, the complex remained intact enough that the NAPs could easily be detected. This could be due to either the short time that the toxin is in the wash or due to steric constraints of having the toxin complex bound to antibody-coated beads.
Interestingly, the purification process of the toxin complex is the same for HA+ BoNT, yet the HA proteins remain as part of the purified toxin complex. A close examination of the NTNH proteins associated with BoNT might help to explain this phenomenon. The NTNH proteins associated with HA+ BoNT are longer than the NTNH proteins associated with HA− BoNT, containing 35 additional amino acids at a site near the N-terminus. A structural model of the BoNT/A PTC shows that this 35 amino acid sequence, known as the nloop, functions as an attachment site for HA-70 [6]. This loop binds three HA-70 molecules, which are in turn bound to HA-17. One HA-17 molecule binds two HA-33 molecules, so that the entire HA-70/HA-17/HA-33 structure is attached via this one site on the NTNH. In theory, any NTNH that does not contain this nloop would be unable to bind HA70, and thus would be devoid of HA proteins.
In fact, any NTNH that is missing this 35-amino acid sequence has been related to an orfX+ toxin cluster. What is less clearly understood is the deletion of the ha genes and their replacement with orfX genes in these toxin gene clusters. Perhaps the orfX+ gene cluster preceded the ha+ gene cluster chronologically, at least in the orfX+ BoNT/A1 gene clusters. It has been hypothesized that the ha+ BoNT/A1 clusters were the result of a recombination phenomenon within the ntnh genes that placed the BoNT/A1 neurotoxin gene and part of the BoNT/A1 ntnh gene within the BoNT/B (ha+) gene cluster [27]. The orfX genes were separated from the rest of the cluster as a result of this action.
While the interactions leading to the structure of the HA+ toxin complex have been thoroughly investigated, the interactions of the toxin, NTNH, and the orfX proteins remain a mystery. It is obvious from our findings that these orfX protein interactions are not as secure as the HA protein interactions, and that they may consist of weaker binding, such as hydrophobic attractions only, that can be more easily disrupted by changes in pH or salt concentration. However, the constant presence of the orfX genes in ha− toxin clusters and identification of residual proteins in toxin complexes argue that they may serve some, as yet not understood, purpose. It may be possible that they serve as protectors during toxin expression or within natural environments, such as soil or water. However, the fragility of the orfX+ toxin complex makes further study challenging.
The orfX proteins are easily removed during toxin purification or washing steps for diagnostic or detection assays. An exception to this finding is BoNT/E, which is an orfX+ BoNT, yet the proteins encoded by open reading frames remain as part of the purified toxin complex within the commercially-available BoNT/E complex. One potential explanation for this difference might be the isoelectric point (pI) of the M-PTC. BoNT/E has the highest isoelectric point of all the BoNT, with a pI consistently at or above six, whereas the pI of all other BoNT are below six. Additionally, the NTNH associated with BoNT/E also have the highest pI, with a pI consistently above five, whereas the pI of the other NTNHs are all below five. Perhaps the differences in pI could account for increased stability of the bond between BoNT/E and the proteins encoded by open reading frames.
Another interesting observation of this work is that the immunoaffinity purification of the neurotoxin from a bivalent strain appeared to purify the NTNH associated with both gene clusters. It is important to note that the antibodies used for immunoaffinity purification are directed toward the neurotoxin itself, and that in all cases examined, there were no instances of both neurotoxins purified with the antibody designed for only one serotype. The presence of both NTNHs might be explained through an understanding of the interaction between the neurotoxin and NTNH. The interactions between BoNT/A1 Hall and its NTNH have been published and consist of 25 sites throughout the sequence of NTNH, but primarily in the C-terminal half of NTNH, with 21 of the 25 sites in the last third of the neurotoxin [4]. Table 2 consists of a list of these sites in the NTNH associated with BoNT/A1 Hall. Although the interactions between BoNT/A1(B) and its NTNH have not been published, the interactions are likely similar due to the high sequence similarity of BoNT/A1 Hall and BoNT/A1(B); in fact, these sequences are 99.8% identical, with only two residues that differ between the two proteins: P1Q and A26V. Of the 21 sites of BoNT/A1 Hall known to interact with its NTNH, 100% of those sites are identical to the sequence of BoNT/A1(B). Looking at the sites of the NTNH associated with BoNT/A1 Hall (NTNH-A) that are known to contact BoNT/A1 Hall listed in Table 2 and the sequence homology between the NTNH-A associated with the A1 cluster of BoNT/A1(B), it is apparent that there is a high degree of similarity; 19 of the 25 sites are 100% identical, and the remaining six sites are conserved mutations, yielding 76% identity and 100% similarity. Therefore, the interactions of NTNH-A associated with BoNT/A1(B) are likely nearly identical to the interactions published between BoNT/A1 Hall and its NTNH-A. Next, a comparison of the 25 sites in the NTNH-B presumed to interact with BoNT/A1(B) shows that all 25 of these sites are 100% identical in the amino acid sequences of the NTNHs associated with both the A and B clusters within BoNT/A1(B). Therefore, because the residues of NTNH-A presumed to interact with BoNT/A1 from the A1(B) cluster are 100% identical with those of NTNH-B, the neurotoxin apparently associates with each interchangeably, yielding our discovery here of the association of the neurotoxin with the NTNH from both gene clusters. Unfortunately, the interactions of the remaining toxins used in this study, BoNT/A2, /A3, /B5, and /F4, and their NTNH have not been published, and because there is a lower degree of amino acid similarity between these neurotoxins and BoNT/A1 Hall (39-90%), it is not possible to theorize about the interaction of these neurotoxins and their NTNHs.
In conclusion, we have examined the composition of the PTC of several hemagglutinin negative botulinum neurotoxin complexes and have found that most of the orfX+ PTC studied here contain at least some of the proteins encoded by these open reading frames. These proteins are more easily removed from the complex than the hemagglutinin proteins because these proteins were not observed following washing of immunoaffinity purified BoNT complexes in 2 M NaCl, with the exception of the toxin complex of BoNT/E. Commercially-available purified BoNT/E complex contains all three proteins encoded by open reading frames despite the harsh conditions used to purify the toxin complex. Additionally, several BoNT/A containing bivalent strains were examined, and the NTNHs and NAPs from both gene clusters were observed following immunoaffinity purification of each toxin component.

Preparation of Ab-Coated Beads
The Abs (20 µg) were biotinylated with 4 µL of the 300 µM sulfo-NHS-biotin in water which was prepared immediately prior to use. The Ab/biotin was incubated at room temperature overnight with no mixing. Biotin labeled Ab (2 µg total for each serotype) was added to washed streptavidin-coated beads (twice with 1 mL of Phosphate Buffered Saline and 0.05% Tween-20 (PBST); reconstituted in 100 µL PBST) and mixed, keeping beads in solution, at room temperature for 1 h. Finally, the beads were then washed two times in 1 mL each of PBST (neutral pH) and next the Ab bound beads were resuspended in 100 µL PBST. For a greater number of samples, the volumes were increased proportionally.

Production and Immunocapture of C. botulinum Culture Supernatants and BoNT Complexes
Biosafety Level-2 processes, practices, and facilities were used to ensure safe conditions while working with BoNT. Additionally, toxin stock material and samples containing BoNT were handled in a Class II biosafety cabinet with HEPA filters to minimize the risk of aerosol exposure. Crude culture supernatants representing C. botulinum strains A1 orfX+ CDC 297, A1(B) CDC 2357, A2f4 SU1304, A3 Loch Maree, A2b5 CDC 1436, A2 SU1887, and ATCC 9564 (additional information is available in Table 1) were produced by growing a subculture for five days at 35 • C in Tryptone Peptone Glucose Yeast Extract (TPGY) medium. After centrifugation, the supernatants were separated and filtered through 0.22 µm filters. An aliquot of the culture supernatant from 100 µL to 500 µL was positioned into a deep well plate containing PBST (either 1X or 10X) to constitute a final volume from 500 µL to 550 µL with an ultimate concentration of 1X PBST.
To extract BoNT/A, 20 µL of beads coated with 0.4 µg of RAZ1 and CR2 mAbs were added into the culture supernatant. The deep well plate was covered and positioned on a plate shaker at the lowest speed needed to keep the beads soluble for 1 h. The deep well plate was uncapped and placed in a KingFisher Flex magnetic particle processor (Thermo Fisher Scientific, Waltham, MA, USA) used for automated bead washing, including two washes of 1 mL each of 2 M NaCl (stringent wash) or 1X PBST (gentle wash) followed by two washes of 1 mL each of 1X PBST. The beads were eluted from the KingFisher Flex in 80 µL of water and next were removed from the KingFisher Flex. Extraction of BoNT/B /E, or /F proceeded in the same fashion, using specific mAbs, as indicated in the materials section, to capture the minor component of the strains in Table 1. Negative controls consisted of culture supernatants extracted with beads conjugated with an irrelevant antibody (i.e., BoNT/B for cultures absent of BoNT/B or ricin for cultures containing BoNT/B). Positive controls were culture supernatant medium spiked with 1 µg of BoNT/A, /B, /E, or /F complexes, and diluted in PBST to necessary levels immediately before spiking, with the rest of the extraction protocol as above.

Digestion and LC-MS/MS Analysis of BoNTs
The immunocaptured progenitor toxin complexes were first digested with trypsin and then chymotrypsin with the same protocol previously used [20]. Essentially, the toxin complexes were digested separately with trypsin and chymotrypsin for 5 min at 52 • C. Toxin complex peptides were identified with LC-MS/MS and database searching as described previously [20], with the peptides separated on a Waters NanoAcquity C18 1.7 µm UPLC column and then analyzed in MS and MS/MS modes on a LTQ-Orbitrap Elite instrument.