Relatedness of type IV pilin PilA amongst geographically diverse Moraxella bovoculi isolated from cattle with infectious bovine keratoconjunctivitis

Introduction Moraxella bovoculi is frequently isolated from the eyes of cattle with infectious bovine keratoconjunctivitis (IBK; pinkeye). As with M. bovis, which has been causally linked to IBK, M. bovoculi expresses an RTX (repeats in the structural toxin) cytotoxin that is related to M. bovis cytotoxin. Pilin, another pathogenic factor in M. bovis , is required for corneal attachment. Seven antigenically distinct pilin serogroups have been described in M. bovis . Hypothesis/Gap Statement Multiple different serogroups exist amongst type IV pilin encoded by M. bovis , however, it is not known whether M. bovoculi exhibits a similar degree of diversity in type IV pilin that it encodes. Aim This study was done to characterize a structural pilin (PilA) encoded by M. bovoculi isolated from cases of IBK to determine if diversity exists amongst PilA sequences. Methodology Ninety-four isolates of M. bovoculi collected between 2002 and 2017 from 23 counties throughout California and from five counties in four other Western states were evaluated. Results DNA sequencing and determination of deduced amino acid sequences revealed ten (designated groups A through J) unique PilA sequences that were ~96.1–99.3 % identical. Pilin groups A and C matched previously reported putative PilA sequences from M. bovoculi isolated from IBK-affected cattle in the USA (Virginia, Nebraska, and Kansas) and Asia (Kazakhstan). The ten pilin sequences identified were only ~74–76 % identical to deduced amino acid sequences of putative pilin proteins identified from the previously reported whole-genome sequences of M. bovoculi derived from deep nasopharyngeal swabs of IBK-asymptomatic cattle. Conclusions Compared to the diversity reported between structural pilin proteins amongst different serogroups of M. bovis , M. bovoculi PilA from geographically diverse isolates derived from IBK-affected cattle are more conserved.


INTRODUCTION
Infectious bovine keratoconjunctivitis (IBK; pinkeye) is the most common eye disease of cattle and is characterized by the presence of corneal ulceration, corneal oedema, conjunctivitis, and eye pain. First reported in 2007 [1], M. bovoculi is now more frequently isolated from eyes of cattle affected with IBK compared to M. bovis [2,3]. While Koch's postulates were previously established for M. bovis and IBK [4], a direct link between the type strain of M. bovoculi (no. 237) and corneal ulceration in a scarification model of infection in dairy calves could not be established [5]. Two distinct OPEN ACCESS genotypes have been characterized in M. bovoculi; genotype 1 is associated with IBK-affected cattle while genotype 2 is associated with IBK-asymptomatic cattle [6,7]. Recent studies have also identified different matrix-assisted laser desorption/ ionization time-of-flight mass spectrometry (MALDI-TOF MS) profiles between the two genotypes [8].
From reports of isolations of M. bovoculi in different parts of the world, it is known that this organism has widespread geographic distribution amongst cattle as well as other ruminant [9,10] and non-ruminant [11] species. One specific pathogenic factor that may be important in the capacity for M. bovoculi to contribute to the pathogenesis of IBK is an RTX (repeats in the structural toxin) toxin [12] that has been shown to be similar to the M. bovis RTX toxin (cytotoxin) [13]. The role of this RTX toxin in the pathogenesis of M. bovoculi is not known and some genotype 1 strains have been identified that do not possess RTX toxin genes [6,7].
In addition to cytotoxin, the pathogenesis of M. bovis involves expression of pili that allow it to adhere to corneal epithelial cells [14][15][16]. For M. bovis it is thought that the presence of multiple M. bovis pilus serogroups [17] coupled with pilin gene inversions [18] increases antigenic variability and accounts for antigenic switching that may allow M. bovis to evade a host's immune response [19].
The purpose of this study was to characterize PilA from geographically diverse Western USA isolates of M. bovoculi from IBK-affected cattle. We also sought to compare deduced pilin amino acid sequences from these Western USA isolates with pilin-related sequences from isolates of M. bovoculi whose full-genome sequences were previously submitted to GenBank, pilin-related sequences from M. ovis, and previously defined pilins from characterized serogroups of M. bovis.

Bacterial isolate source and identification
A total of 94 isolates of M. bovoculi from cases of IBK in cattle from 28 counties were used for this study (see Table 1). Bacterial isolates were cultured from ocular swabs from eyes of cattle with IBK that had been collected by one of the authors (JAA), or submitted to the California Animal Health and Food Safety Laboratory (CAHFS), Davis, CA and provided by one of the authors (KAC Isolates were confirmed as M. bovoculi on the basis of biochemical testing as well as by blast analysis against the GenBank database of an amplified and sequenced ribosomal RNA gene (partial 16S ribosomal RNA gene and 16 S-23S ribosomal RNA intergenic spacer; ISR) (details provided below). A subset of isolates collected by one of the authors (KAC) was also subject to analysis by MALDI-TOF testing (see below). Ocular swabs were streaked onto trypticase soy agar with 0.5 % sheep blood plates (SBA) and incubated at 35 °C. Colonies with morphology consistent with Moraxella spp. after 24-48 h of incubation were subcultured for further characterization. Isolates that were catalase-positive, oxidase-positive, Gram-negative coccobacilli, negative for carbohydrate fermentation, able to reduce nitrate, negative for casein hydrolysis, and able to deaminate phenylalanine were characterized as M. bovoculi. Isolates were stored frozen at −70 °C until use.
The isolates provided by KAC (n=59) were also subject to MALDI-TOF testing by incubating isolates overnight on SBA at 35 °C in a 5-10 % CO 2 atmosphere. Isolates were tested according to the MALDI-TOF instrument manufacturer's recommended procedure for the direct smear method using α-cyano-4-hydroxycinnamic acid (Bruker Daltronics, Billerica, MA, USA), and subjected to automatic detection in a positive linear mode between 2 kDa and 20 kDa m/hz, with a laser frequency of 60 Hz (Microflex LT MALDI-TOF MS, Bruker Daltronics). The system was calibrated for reference masses of 3637-16 952 Da using the manufacturer's supplied bacterial test standard. Up to 240 spectrum profiles were obtained per colony, and all colonies were tested in duplicate. Bacterial identifications were determined using commercial software and the database provided by the manufacturer (Compass, 4.1, Bruker Daltronics). Identity scores >2.0 were considered very good to the genus and species level per the manufacturer's guidelines.

Genomic DNA and PCR
M. bovoculi isolates were thawed, streaked onto 5 % SBA, and incubated at 35 °C for 18-20 h. Genomic DNA was purified from whole bacteria using a commercial kit (DNEasy kit; Qiagen, Germantown, MD, USA). Further confirmation of all isolates as M. bovoculi was made by blast analysis of an amplified and sequenced ribosomal RNA gene (partial 16S

Impact Statement
Pilin (PilA) from M. bovoculi is conserved amongst geographically diverse isolates derived from cattle with IBK and displays considerably less variability amongst isolates compared to M. bovis pilins from different M. bovis serogroups. The significance of M. bovoculi pilin as it relates to the pathogenesis of IBK is presently unknown.  Accession numbers of sequences used for these comparisons are provided in Table 2.

Pilin sequence comparisons
The 94

DNA and deduced amino acid sequences of M. bovoculi pilin
A 459 bp ORF was identified in the sequenced amplicons of all 94 M. bovoculi isolates; 20 of these ORFs were unique. The deduced amino acid sequences of these 20 ORFs encoded ten unique PilA sequences that were designated M. bovoculi PilA groups A through J (Fig. 1). The number of isolates in these groups were 49, 19, 14, 5 and 2 for PilA groups A, B, C, D and E, respectively; one isolate each represented PilA groups F through J. The most frequently identified PilA sequence in this collection of samples was group A, which was identified in 22 of the 28 counties from which the 94 isolates originated ( Table 3).
The deduced amino acid sequences of these ten unique PilA sequences shared a high degree of sequence similarity with overall identical sites and pairwise identity of 92.8 and 98.1 %, respectively. Differences in deduced amino acid sequences between the ten groups included four conservative and eight radical amino acid replacements. At residue 61, the sequences were equally divided between those with serine versus asparagine residues. In one of the pairs of SFREC isolates from 2008 that originated from a single animal, one exhibited a PilA group A sequence (accession MT333669) while the second obtained from the opposite eye approximately 4 weeks later exhibited a group B sequence (MT333668). In the second pair of two isolates that originated from a single animal, the initial isolate exhibited a PilA group B sequence (accession MT333675), while the second isolate obtained from the opposite eye approximately 10 weeks later exhibited a PilA group A sequence (accession MT333674).

Comparisons with previously reported M. bovoculi, M. ovis and M. bovis pilin-related sequences
Previously reported pilin-related sequences in M. bovoculi from IBK-affected cattle in Kansas (strain 57922), Virginia (strain 58086) and Kazakhstan (strain KZ-1) (respective GenBank accession nos.: AKG14945, AKG16627 and NSM11559) and the PilA protein from the type strain of M. bovoculi (237; KDN24455) were identical to the PilA group A deduced amino acid sequence. A previously reported hypothetical protein from a Nebraska isolate of M. bovoculi (strain 58069) from an IBK-affected cow (GenBank accession no.: AKG18359) was identical to the PilA group C deduced amino acid sequence. See Fig. 2.
Comparisons of the PilA groups A-J deduced amino acid sequences with putative pilin deduced amino acid sequences that were identified in M. bovoculi from deep nasopharyngeal swabs of cattle that did not have IBK (M. bovoculi strains 22 581, 23 343, 28 389 and 33 362 [7] with respective GenBank accession nos.: AKG07295, AKG10099, AKG12021 and AKG13989) revealed 74.3-75.7 % identity between sequence pairs (Fig. 2). Comparisons between previously reported M. ovis pilin sequences (GenBank accession nos.: WP_063514484, WP_112744298, ANB91903, SPX85670, A phylogenetic tree depicting relationships between the pilin sequences listed in Table 1 and the ten M. bovoculi PilA groups identified in this study showed distinct clustering of pilindeduced amino acid sequences from seven defined M. bovis serogroups, M. bovoculi isolated from cattle with IBK, M. ovis and M. bovoculi isolated from the nasopharynx of cattle without IBK (Fig. 3).

DISCUSSION
In this study we found ten unique structural pilin (PilA)deduced amino acid sequences amongst a collection of M. bovoculi that had been isolated from eyes of IBK-affected cattle throughout California and four other Western states (Idaho, Nevada, New Mexico and Arizona). Given the limited sample size and geographic distribution of isolates that we examined, however, it is impossible to say how likely or not it is that more PilA groups might exist amongst M. bovoculi.
Since two of these PilA group sequences (A and C) matched PilA sequences in M. bovoculi from IBK-affected cattle from other locations in the USA (Kansas, Nebraska, Virginia) and Asia (Akmola region of Kazakhstan), it is possible that the PilA groups identified in this study are representative of M. bovoculi PilA sequences in general. Of the ten PilA groups that we identified, three (A, B and C) were the most widely distributed over geography and time, however, it is likely that a larger sample size would have revealed additional locations of these less well-represented PilA groups.
Among the ten PilA groups that were characterized in this study population, the overall degree of difference was very small compared to the seven M. bovis serogroups that have been characterized [17,22]. Because M. bovoculi isolates that have been examined thus far via whole-genome sequencing have not displayed evidence for an I/Q pilin type phase shift [6] that was described in M. bovis [18,23], it seems likely that any additional PilA group types that may be identified in the future will probably exhibit a relatively low degree of variability from one another as compared to the variability that is observed in M. bovis exhibiting different pilin serogroups.  [26]. This suggests that pilin expression in M. bovoculi is likely involved in its ability to colonize the bovine eye. The fact that a putative PilA protein from M. bovoculi associated with the nasopharynx of IBK-asymptomatic cattle in Missouri, Tennessee, Kentucky and Kansas [7] and PilA from M. bovoculi from IBK-affected cattle exhibit only ~75 % identity between deduced amino acid sequences raises the possibility that differences in pilin sequence allow survival on different mucosal sites (for example, ocular surface versus nasopharynx).
Whether or not differences between the 10 PilA groups that we identified have any bearing on evasion of a host immune response during clinical IBK associated with the presence of M. bovoculi remains to be determined. For M. bovis strain Epp63 it is known that recombination events involving pilinexpressing genes allow different forms of pilin to be expressed [18], and that these different forms of pilin are associated with colonization versus maintenance of infection [27]. In the population of M. bovoculi isolates that we examined for this study we identified two PilA groups in each of the two animals (PilA groups A and B) where IBK developed in different eyes over a period of weeks. In one animal a PilA group A isolate was followed by identification of a PilA group B isolate, while in the other animal in the same herd the initial isolate was a PilA group B isolate followed by a group A isolate. Additional studies are needed to determine whether cattle develop immune responses to M. bovoculi pilin during ocular infections and whether the sorts of PilA group changes that we observed were reflective of host immune selection or just pure coincidence.

CONCLUSION
M. bovoculi PilA deduced amino acid sequences exhibit some diversity, however, overall, PilA sequences are relatively conserved across geographically diverse isolates from cattle with IBK and much more conserved relative to M. bovis pilin serogroups. The exact role that M. bovoculi PilA might play in the ability of M. bovoculi to exist in/around bovine ocular tissues remains to be determined.

Funding information
This work was supported by the USDA National Institute of Food and Agriculture, Animal Health project 1009855.

Conflicts of interest
The authors declare that there are no conflicts of interest.