Genetics and pathology associated with Klebsiella pneumoniae and Klebsiella spp. isolates from North American Pacific coastal marine mammals Veterinary Microbiology

Klebsiella pneumoniae is occasionally isolated at necropsy, but not from live SSOs. Hypermucoviscous (HMV) K. pneumoniae strains are confirmed pathogens of Pacific Basin pinnipeds, but have not been previously isolated from SSOs. We characterized the virulence profiles of K. pneumoniae isolates from necropsied SSOs, evaluated killing of marine mammal K. pneumoniae following in vitro exposure to California sea lion (CSL: Zalophanus californianus ) whole blood and serum, and characterized lesion patterns associated with Klebsiella spp. infection in SSOs. Four of 15 SSO K. pneumoniae isolates were HMV and all were recovered from SSOs that stranded during 2005. Many K. pneumoniae infections were associated with moderate to severe pathology as a cause of death or sequela. All HMV infections were assessed as a primary cause of death or as a direct result of the primary cause of death. Klebsiella -infected SSOs exhibited broncho- pneumonia, tracheobronchitis and/or pleuritis, enteritis, Profilicollis sp. acanthocephalan peritonitis, septic peritonitis, and septicemia. All SSO HMV isolates were capsular type K2, the serotype most associated with HMV infections in CSLs. Multiplex PCR revealed two distinct virulence gene profiles within HMV isolates and two within non-HMV isolates. In vitro experiments investigating CSL whole blood and serum killing of K. pneumoniae suggest that HMV isolates are more resistant to serum killing than non-HMV isolates.


Introduction
Klebsiella pneumoniae is an encapsulated, gram-negative, non-motile bacillus in the order Enterobacterales that is commonly found in the environment and on mucosal surfaces (Bagley, 1985). It is a common nosocomial pathogen in human and veterinary medicine (Tsay et al., 2002;Brisse and Duijkeren, 2005). This bacterium has a broad host range and can rapidly develop antimicrobial resistance (Holt et al., 2015). Klebsiella spp. have been isolated from diverse terrestrial and marine mammals, but few reports exist for sea otters (Enhydra lutris) (Castinel et al., 2007;Whitaker et al., 2018;Miller et al., 2020).
Hypermucoviscous (HMV) K. pneumoniae are highly invasive, hypervirulent strains of K. pneumoniae that can cause severe localized or systemic, community-acquired infections in healthy hosts. HMV strains are more virulent than non-HMV K. pneumoniae isolates in animal models (Twenhafel et al., 2008;Shon et al., 2013). When grown in vitro on agar plates, HMV K. pneumoniae strains produce copious capsular mucopolysaccharides, as indicated by a mucoid string >5 mm long when stretched from the plate with an inoculating loop (Shon et al., 2013). HMV K. pneumoniae isolates commonly express multiple virulence genes, including rmpA (a regulator of the mucoid phenotype), magA (capsular serotype K1), K2wzy (capsular serotype K2), and siderophore genes among others, that allow the pathogen to infect host cells while avoiding host innate and adaptive immune responses (Shon et al., 2013;Holt et al., 2015;Struve et al., 2015;Soto et al., 2016). This emerging pathogen has been considered the next "superbug" (Shon and Russo, 2012).
Our composite microbiological data from assessment of >500 necropsied southern sea otters (SSO: Enhydra lutris nereis) to date suggest that Klebsiella spp. are not common commensals for these animals; all Klebsiella spp.-positive, necropsied SSOs from 1997 through 2020 are included in the current study. In addition, assessment of >120 apparently healthy, wild-caught SSOs were sampled, including swabs from the skin, upper respiratory tract, oropharynx, rectum, prepuce, or vagina were inoculated onto non-selective media (Blood and MacConkey agar) yielded no Klebsiella spp. isolates (M. Miller, unpub. data).
The objectives of this study were to evaluate the pathogenic profiles of K. pneumoniae isolates from SSOs, CSLs, and PHSs using a novel multiplex-PCR for known Klebsiella virulence genes. Stranding and necropsy records were reviewed to assess potential gross and microscopic pathology associations for Klebsiella spp.-positive SSOs. Phenotypic and genotypic variations of K. pneumoniae isolates from eight SSOs that stranded between 1998, eight CSLs examined between 1994 four PHSs examined between 2012-2016 were also compared. Finally, we challenged non-HMV and HMV K. pneumoniae isolates from SSOs, CSLs, and PHSs with whole blood and serum from healthy CSLs to assess bacterial persistence in vitro.

Bacteria
Freshly dead or euthanized SSOs that stranded throughout central California, USA ( Fig. 1) or died in captivity were necropsied at the Marine Wildlife Veterinary Care and Research Center (MWVCRC; Santa Cruz, CA) between 1997 and 2020 using a standardized protocol (Miller et al., 2020). Sterile swabs inoculated from tissue or body fluids during necropsy were shipped overnight to the Microbiology Laboratory of the Veterinary Medical Teaching Hospital at the University of California-Davis for bacterial isolation and identification. Subsamples of all major tissues and body fluids were also cryopreserved at − 80 • C at MWVCRC.
Necropsied SSOs were enrolled in the study only if Klebsiella spp. were detected via bacterial culture. Detailed information for each SSO is available in Supplementary Material, Table 1. For any necropsied SSO with a previous Klebsiella spp. isolate that had not been cryopreserved, necropsy tissues cryoarchived at − 80 • C were thawed and fresh bacterial swabs were collected and submitted for isolation. The five most common sources of Klebsiella spp. isolation from necropsied SSOs (lung, kidney, lymph node, liver, and/or heart; see Supplementary Material, Table 1) were prioritized for bacterial isolation; inoculated swabs were streaked on SBA (BMS) and incubated at 37 • C in ambient atmosphere for 72 h. During incubation, bacterial colonies were evaluated for morphology consistent with K. pneumoniae and a string test was performed to identify southern sea otters (Enhydra lutris nereis) that were culture-positive for Klebsiella spp. Sea otter 3057-98 was not included as it came from a captive setting following stranding. Cases are indicated by red dots; parenthetical numbers represent multiple otters recovered in the same general area. Asterisks (*) represent the stranding locations of southern sea otters infected with biochemically confirmed hypermucoviscous Klebsiella pneumoniae (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.). the HMV phenotype (Fang et al., 2004). Cryopreserved K. pneumoniae strains obtained from stranded or moribund CSLs and PHSs examined at The Marine Mammal Center (TMMC, Sausalito, CA) from 1994 to 2016 were also revived following the same procedures mentioned above to compare with strains from SSOs.

Molecular characterization
Two multiplex PCR assays were designed to detect capsule serotypes and factors associated with hypervirulence. The capsule assays included detection of K1, K2, and K5 serotypes. Genes known to be associated with increased virulence in humans (Chen et al., 2004;Wu et al., 2009;Struve et al., 2015) were assessed, including the regulator of mucoid phenotype, the mucoviscosity-associated gene, the gene associated with allatonin metabolism, the genotoxin colibactin, three siderophores (aerobactin, salmochelin, and yersiniabactin), and two fimbrial adhesins (type 1 and type 3 fimbriae). Primers specific for the 16S-23S internal transcribed spacer in K. pneumoniae were also included for species verification (Turton et al., 2010). The two multiplex PCRs, including primer sequences and the specific genes targeted for each capsule serotype or virulence factor, are summarized in Table 1. Reference strains expressing capsule serotypes K1, K2 and K5, and selected K. pneumoniae strains with available whole genome sequences were used to validate the multiplex PCR assays (Struve et al., 2015).
Genomic DNA was extracted from bacterial isolates using a DNeasy Blood and Tissue Kit (Qiagen, Hilden, Germany) according to manufacturer protocols. All DNA samples were stored at − 20 • C until used. PCR assays were completed using a Multiplex PCR Kit (Qiagen) in a 50  Table 2 Lesion prevalence (defined by lesion presence/absence independent of the primary or contributing causes of death) in necropsied southern sea otters (Enhydra lutris nereis) with culture confirmed Klebsiella spp. and hypermucoviscous (HMV) and non-HMV K. pneumoniae infections. Total number of animals with each lesion is indicated; percentage of animals positive for the lesion noted in parentheses.

Observed Lesion
Non-HMV-positive K. pneumoniae (n = 11) μL reaction that included 2x Qiagen Multiplex PCR master mix, 0.2 μM forward and reverse primers, 5 μL DNA at concentration of ~50 ng/μL, and RNase-free water to volume using a SimpleAmp Thermal Cycler Applied-Biosystem (Thermo Fisher Scientific). PCR conditions included a 15 min denaturation at 95 • C, followed by 30 cycles of denaturation at 94 • C for 30 s, annealing at 59 • C for 90 s, extension at 72 • C for 90 s, and a final extension at 72 • C for 10 min. PCR products were visualized using a MyECL Imager (Thermo Fisher Scientific) after electrophoresis of the amplified DNA in a 2% agarose gel containing 1% SYBR safe DNA gel stain (Thermo Fisher Scientific).

Antimicrobial susceptibility testing
The minimum inhibitory concentrations (MIC) for 18 different antimicrobials were identified using Sensititre AVIAN1F Vet AST plates (Thermo Fisher Scientific) in cation-adjusted Mueller Hinton broth (Thermo Fisher Scientific) following Clinical and Laboratory Standards Institute (CLSI) protocols (CLSI, 2020). Escherichia coli strain ATTC 25922 was used as a quality control. Inoculated blood agar plates were incubated at 37 • C for 24 h, bacteria inoculated on the Sensititre plates according to manufacturer protocols, and bacterial growth was checked visually 24 and 48 h post-inoculation. The MIC was defined as the lowest concentration where no visible growth was exhibited. Breakpoints were interpreted following those for Enterobacterales in dogs (CLSI, 2018(CLSI, , 2020.

Potential lesion associations
Case records from all necropsied SSOs that were culture-positive for Klebsiella spp. in one or more tissues from 1997 through 2020 were assessed for possible bacterial-associated pathology. Associations were assessed in two ways: 1.) Klebsiella spp. detection relative to all primary or contributing causes of death and sequelae at necropsy, and 2.) Presence/absence of defined health conditions for all enrolled SSOs, regardless of whether each condition was considered a cause of death or sequela at necropsy (Table 2). Assessment #1 was performed to examine whether Klebsiella spp. detection was associated with the primary or contributing causes of death or sequelae, based on all findings from necropsy and histopathology. Assessment #2 was performed to identify health conditions that could indicate potential routes of Klebsiella spp. bacterial invasion, although these patterns would need to be verified or ruled out through subsequent research.
SSOs commonly exhibit multiple moderate to severe health conditions at necropsy (Miller et al., 2020), and careful assessment of the most common causes of death is important for prioritizing conservation efforts (Tinker et al., 2021), so this level of detail is important. Distinguishing core causes of death and sequelae also ensures that risk assessments are performed on animal groups with the highest possible similarity.
Klebsiella spp. infection was considered a potential contributor to animal morbidity and death (as a cause of death or sequela) by a veterinary pathologist based on review of all findings from gross necropsy, microbiology, and histopathology. Factors that were considered in this assessment included bacterial isolation from observed lesions and visualization of morphologically compatible bacteria (short, stout encapsulated or non-encapsulated bacterial rods) within lesions on histopathology (Fig. 2).
For each enrolled SSO, the primary cause of death was the most severe process that likely caused stranding. Secondary, tertiary and quaternary (cumulatively "contributing") causes of death represent independent processes that were also present at necropsy, were deemed moderate or severe, and likely contributed to stranding (Miller et al., 2020). Necropsy findings were defined as sequelae if they occurred as a direct consequence of a primary or contributing cause of death and contributed to the risk of death but were not independent processes. An example would be a white shark (Carcharodon carcharias) bite as a primary cause of death, with post-traumatic bacterial infection as a sequela (Miller et al., 2020). Miller et al. (2020) confirmed the importance of assessment of both causes of death and sequelae. For example, that study showed that most SSOs survive the acute traumatic effects of shark bite, but later succumb to bacterial infections that would not have occurred if the traumatic event had not happened, providing a portal for bacterial entry (Miller et al., 2020).
Although case detail varied, all available clinical, gross necropsy, histopathology, and diagnostic data were reviewed to identify Klebsiellaassociated lesion patterns. This included detection of morphologically compatible bacteria (short, stout, Gram-negative, encapsulated, or nonencapsulated encapsulated rods) on histopathology for Klebsiella-positive SSOs (see Figs. 2-4 for examples). Assessed lesions and conditions that could be potential indicators for routes of Klebsiella spp. bacterial invasion (Table 2) included presence/absence of septicemia, Profilicollis sp.-associated peritonitis, septic peritonitis, current or healed intestinal perforations, lymphadenomegaly, suppurative lymphadenitis, infected wounds or abscesses, pneumonia, potential bacterial gastroenteritis, enterocolitis, intestinal displacements, suppurative meningitis, and other suppurative conditions that could be attributed to bacterial infections. These conditions are part of standardized lesion documentation for SSOs and were assessed during necropsy and retrospectively via review of archival photographs and case records. Hypermucoviscous Klebsiella pneumoniae was isolated in pure culture from the fascial abscess, lymph node, brain and heart blood. At low magnification (A.), short, stout bacterial rods were visible throughout the section, including within the cytoplasm of macrophages (arrows). At higher magnification (B), thick, pale outer capsules were apparent around the bacterial rods (arrows). Hematoxylin and eosin stain. Bar = 12 μm (A), and Bar = 10 μm (B).
Animals were considered positive for the above conditions if documentation was found in the case record or identified via gross photographs and/or slide review by a veterinary pathologist. Negative cases were those where the condition of interest was reported as absent in the case record and/or could be excluded through pathologist review of the case record, photos, histopathology, or other means. Cases were noted as unknown if the record, available photographs, and histology lacked sufficient detail to confirm or exclude a defined condition (Miller et al., 2020).
Presence/absence of septicemia was diagnosed based on all findings from bacterial culture, gross necropsy, and histopathology. In addition to direct visualization of bacteria in tissues, lymphatics, and blood vessels, septic SSOs often exhibited increased neutrophil density on histopathology (e.g., in brain, lung, lymph nodes, spleen, kidney and liver), antemortem leukocyte toxicity, and intracytoplasmic bacteria within leukocytes.
Because one or more bacterial species can contribute to observed pathology, areas of mixed bacterial growth, and isolation of bacteria other than Klebsiella spp. were summarized in Supplementary Material, Table 1. Lesions were reported as Klebsiella spp.-associated when Klebsiella spp. was successfully isolated from the affected tissue and/or morphologically compatible bacteria (short, stout encapsulated or nonencapsulated bacterial rods) were observed in bacterial-mediated lesions on histopathology. Klebsiella spp. were considered a cause (or Fig. 3. Southern sea otter (Enhydra lutris nereis) 4486-05: A.) Gross view of bronchus and perihilar lung tissue; there is multifocal distension and yellow discoloration of perihilar alveoli (upper left), and moderate mucopurulent bronchial exudate (Lower right). Hypermucoviscous K. pneumoniae was isolated in pure culture from lung, hilar lymph node, and heart blood. B.) Gram stain of lung from the same sea otter; alveoli are filled with macrophages, neutropils and myriad free and intracytoplasmic Gram-negative bacterial rods.
Brown-Brenn stain. Bar = 25 μm (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.). potentially an additive cause in combination with other bacteria) of a lesion if Klebsiella spp. were isolated from tissues with lesions or adjacent tissues (such as a draining lymph node) and morphologically compatible bacteria were observed on histopathology. Given our preliminary state of knowledge and small sample size, additional studies will be needed to further refine and confirm the pathophysiology of Klebsiella spp. infections in SSOs.

Whole blood killing assay
Whole blood bacterial killing assays were performed following protocols by Locke et al. (2007) with modifications. Briefly, whole blood from healthy, rehabilitated CSLs with no known history of K. pneumoniae infection was collected into EDTA vacutainer tubes (Greiner Bio-One, Kremsmünster, Austria) during routine release examination. Blood was used within 4 h in in vitro assays. Bacterial isolates (Table 3) were grown overnight on SBA at 37 • C. Bacterial concentration (colony forming unit; CFU/mL) was adjusted in phosphate-buffered saline (PBS) to ~1.5 × 10 5 CFU/mL and mixed in a 1:1 vol/vol ratio with EDTA whole blood. Inoculated blood was incubated at 37 • C for 1.5 h and then spot plated on SBA by serial dilution to determine counts of viable K. pneumoniae. Bacterial survival in whole blood was compared to a control group of 1.5 × 10 5 CFU/mL K. pneumoniae mixed in a 1:1 vol/vol ratio with sterile PBS. Whole blood experiments were performed in six replicates to confirm consistency of results and results were averaged among HMV and non-HMV isolates.

Serum killing assays
Complement-based killing (serum killing) of K. pneumoniae strains from SSOs, CSLs, and PHSs (Table 3) was assessed as described by Soto et al. (2010). Briefly, pooled serum from healthy, rehabilitated CSLs with no known history of K. pneumoniae infection was collected into serum separator tubes during routine release examination. Tubes were centrifuged at 3000 xg for 10 min to achieve adequate serum separation and sera from multiple animals and tubes were pooled. Bacterial isolates (incubated overnight on blood agar and suspended in PBS to 1.5 × 10 4 CFU/mL as above) were mixed in a 1:4 vol/vol ratio with serum that was passed through a 0.2 μm pore size filter and incubated at 37 • C for 3 or 24 h. Pooled heat inactivated serum (heated at 56 • C for 45 min to inactivate complement, but leaving other potential humoral components such as antibodies intact) inoculated with same amount of bacteria as the activated serum was used as a control. Colony counts were determined using a 6 × 6 drop plate method (Chen et al., 2003). Serum experiments were performed in five replicates for the 3 h incubation, and three replicates for the 24 h incubation period. Results were averaged among HMV and non-HMV isolates.

Statistical analysis
Bacterial persistence (culturable bacteria) in the different treatments compared in whole-blood and serum killing experiments were subjected to statistical analysis. Statistical significance was determined using twoway ANOVA tests with Tukey's multiple comparisons test (significance if p < 0.05) in GraphPad Prism (version 8.3.0, GraphPad Software, La Jolla, CA).

Identification of HMV K.pneumoniae
Klebsiella spp. isolates were obtained from 18 necropsied SSOs from 1997 through 2020, including 15 K. pneumoniae strains and three Klebsiella spp. strains (Table 3). Of the 15 K. pneumoniae isolates, 4/15 were HMV-positive, and all four HMV strains were from SSOs that died during 2005 (Supplementary Material, Table 1). Only one SSO case from 2005 was infected with a non-HMV K. pneumoniae strain. The four SSOs with HMV K. pneumoniae infections were of variable age: one pup, two subadults, and an adult (Supplementary Material, Table 1). Three of the HMV K. pneumoniae-infected SSOs stranded within 57 km of each other over an eight-week period (Fig. 1). The fourth SSO stranded approximately 203 km north of the other animals and six months later. The remaining 11 K. pneumoniae isolates were non-HMV strains. The three Klebsiella spp. isolates included one each of K. oxytoca, K. ozaenae-like and K. ornithium (Supplementary Material, Table 1).
HMV K. pneumoniae was the sole pathogen isolated in one case; all other HMV K. pneumoniae-positive cases yielded mixed bacterial growth, sometimes including other known or suspected bacterial SSO pathogens, including Streptococcus phocae, Vibrio parahemolyticus, Salmonella enterica, Proteus spp., Pseudomonas spp., and hemolytic E. coli (Miller et al., 2020).
Klebsiella spp. were commonly isolated from lymph nodes, heart blood, and peritoneal fluid (Table 3 and Supplementary Material,  Table 1). Six SSOs had K. pneumoniae recovered from more than one location. Among Klebsiella spp.-positive SSOs, three or more anatomical sites were culture-positive at necropsy for 2/4 HMV K. pneumoniae strains, 2/11 non-HMV K. pneumonia strains, and 0/3 isolates of other Klebsiella species. For SSO 4449− 05, HMV K. pneumoniae was isolated in pure culture from five different anatomic sites, and myriad short, stout, thickly encapsulated bacterial rods consistent with K. pneumoniae were identified within a K. pneumoniae-positive abscess and an adjacent lymph node on histopathology (Fig. 2). Six of the eight CSL K. pneumoniae isolates were HMV and two were non-HMV. All four PHS isolates were non-HMV K. pneumoniae.

Molecular characterization
Phenotypical and molecular characterization of K. pneumoniae isolates are presented in Tables 3 and 4. All isolates used for multiplex PCR were confirmed as K. pneumoniae based on amplification and sequencing of the 16S-23S internal transcribed spacer (Table 1) (Turton et al., 2010). All isolates were negative for magA (capsular serotype K1), K5wzy (capsular serotype K5), clbB (colibactin), and allS (allantoin metabolism island). Within each host species, there were multiple isolates that had identical virulence gene profiles, including three of the four SSO HMV isolates. The fourth SSO HMV isolate (4614− 05) only varied from the other three by absence of iroD, and was recovered from an animal that was spatially and temporally separated from the other three animals. HMV K. pneumoniae isolates obtained from SSOs and CSLs had similarities between virulence gene profiles, but none were identical. However, there were identical profiles for several non-HMV SSO and CSL isolates. Two of the non-HMV SSO isolates (4922− 07 and 4127− 04) were positive for iutA and fyuA but were otherwise identical to the other two non-HMV isolates (3908− 03 and 3057− 98).

Antimicrobial susceptibility
HMV and non-HMV isolates from SSOs were all susceptible to enrofloxacin and gentamicin and all resistant to amoxicillin (Table 5, CLSI, 2018CLSI, , 2020. There was variable resistance to ceftiofur, tetracycline, trimethoprim/sulfamethoxazole, and florfenicol (Table 5). Other antimicrobials did not have established breakpoints for Enterobacterales (CLSI, 2018(CLSI, , 2020, but all isolates had the same MIC for erythromycin, spectinomycin, sulphathiazole, novobiocin, and tylosin tartrate (Table 5). SSO 4486− 05, the SSO that grew pure K. pneumoniae on culture, had a streptomycin MIC of 256, while all other SSOs had an MIC of <8 for this antimicrobial.
K. pneumoniae strains (n = 11) were isolated from SSOs with acanthocephalan peritonitis. These results suggest, but do not confirm a possible enteric route for HMV and non-HMV K. pneumoniae exposure in SSOs (Table 2). In addition, perimortem septicemia was suspected for 100 % of HMV K. pneumoniae isolates, and 91 % of non-HMV K. pneumoniae strains, compared with 66 % for other Klebsiella spp. Our results, although preliminary, suggest that routes of bacterial exposure differ for K. pneumoniae and other, potentially less pathogenic Klebsiella spp. Potential portals for Klebsiella spp. invasion include traumatic, mating, or fight wounds, enteric disease, acanthocephalan-associated or other types of intestinal perforations, aspiration of material into the respiratory tract, and ascending urinary tract infection (Table 2).

Susceptibility of Klebsiella pneumoniae to serum and whole blood
All K. pneumoniae tested were resistant to CSL whole blood killing under the conditions investigated. There were no significant differences in survival of HMV compared to non-HMV isolates in whole blood from CSLs (p > 0.05) (Fig. 6).

Discussion
Our study goals were to investigate virulence gene profiles, in vitro virulence of K. pneumoniae isolates, and associations between specific lesions seen at necropsy and via histopathology and successful isolation of K. pneumoniae and Klebsiella spp. from stranded SSOs necropsied at MWVCRC over 24 years. Although precise enumeration is difficult, this amounts to >500 animals where aerobic bacterial culture was attempted from tissues or body fluids at necropsy. During routine necropsies, samples from grossly apparent lesions are typically submitted for bacterial culture; this selection bias may under-represent the actual prevalence of K. pneumoniae infections in the SSO population. Over 24 years of necropsy effort, only 18 SSOs were culture-positive for Klebsiella spp., with 15 of those isolates (83 %) being K. pneumoniae. Klebsiella spp. were usually isolated from SSO lymph nodes, heart blood, and peritoneal fluid. Relatively infrequent isolation of Klebsiella spp. may reflect relatively low infection prevalence in SSOs and/or postmortem proliferation of other bacteria that may outgrow Klebsiella spp. However, Klebsiella spp. detection is uncommon in live SSOs. Unpublished microbiological data from assessment of >120 live, ostensibly healthy SSOs sampled during routine capture and tagging activities suggest that Klebsiella spp. are uncommon commensals for SSOs (M. Miller, unpub. data).
Klebsiella spp. infection was considered a potential contributor to stranding and death for most culture-positive SSOs in the current study (Supplementary Material, Table 1). Our composite data, although preliminary, suggest that Klebsiella spp. isolation in necropsied SSOs is often associated with diagnosis of moderate to severe bacterial pathology (Supplementary Material, Table 1); subsequent studies should examine these associations in greater detail.
Interestingly, all four HMV K. pneumoniae-positive SSOs stranded in 2005, and three of the animals stranded within a 57 km stretch of coastline over an eight-week period. Two of these SSOs were recovered on the same day within 4.5 km of each other, suggestive of spatial and temporal clustering of HMV K. pneumoniae infection. The four HMV SSOs were of varying age; a similar age range has been reported for klebsiellosis in CSLs from California (Jang et al., 2010;Seguel et al., 2017;Whitaker et al., 2018) and PHSs (Whitaker et al., 2018), suggesting K. pneumoniae does not require age-associated factors (such as a developing immune system in young marine mammals) to be pathogenic. Of note, no HMV K. pneumoniae was recovered from CSLs or PHSs during 2005 (Whitaker et al., 2018;TMMC pers. comm.).
The epizootic potential of K. pneumoniae with the HMV phenotype and K2 serotype was first confirmed in marine mammals during multiple mortality events affecting endangered New Zealand sea lions (Phocarctos hookeri). Klebsiella pneumoniae was first detected as the cause of an epizootic in this colony during the 2001-2002 and 2002-2003 breeding seasons, prior to use of the phenotype string test and serotyping to discern HMV and non-HMV strains (Castinel et al., 2007;Roe pers. comm.). HMV K. pneumoniae was later identified as the cause of an epizootic that accounted for 58 % of pup mortality on Enderby Island during the 2006-2007 and 2009-2010 breeding seasons . Subsequent genetic characterization showed clonality of K. pneumoniae strains throughout many of New Zealand's offshore islands over multiple years (Pinpimai et al., 2018). In our study, we observed temporal and spatial clustering of three SSOs infected with HMV isolates with identical virulence gene profiles in 2005. These findings are suggestive of clonality and epizootic potential for HMV K. pneumoniae infection in SSOs; however, confirmation is not possible due to our small sample and additional genetic work, such as whole genome sequencing, is warranted to assess clonality.
Although virulence gene expression was not investigated in this study, and associations between gene presence and observed pathology were not assessed, confirmation of virulence gene presence can aid our understanding of the pathogenic potential of Klebsiella spp. The K1 and K2 capsular serotypes have all been associated with the HMV K. pneumoniae phenotype (Yu et al., 2006), and predominate in human infections (Fang et al., 2007;Yu et al., 2008). In combination with rmpA, the presence of these capsular serotypes is considered the best predictor for the HMV phenotype and hypervirulence in humans (Choby et al., 2020). Previous studies have found that the vast majority of marine mammal (CSLs, New Zealand sea lions, and harbor porpoises) HMV K. pneumoniae isolates have the K2 capsular serotype (Jang et al., 2010;Roe et al., 2015;Whitaker et al., 2018). All HMV K. pneumoniae isolates from SSOs and CSLs in this study had the K2 capsular serotype, while no non-HMV K. pneumoniae isolates had this serotype, confirming this serotype predominates in marine mammal HMV K. pneumoniae strains. Whole genome analysis and the addition of isolates recovered from California terrestrial mammals and humans may help clarify the significance of this finding.
None of the marine mammal isolates tested here were positive for magA, which encodes the K1 capsular serotype (Remya et al., 2018). This is consistent with other reports of HMV K. pneumoniae in marine mammals that found no magA positive isolates (Jang et al., 2010;Roe et al., 2015;Whitaker et al., 2018). Similarly, all marine mammal HMV K. pneumoniae isolates and only one non-HMV isolate were positive for rmpA, which supports associations between HMV and rmpA. One CSL non-HMV K. pneumoniae isolate from a prior study was also found to be positive for rmpA and rmpA2 (Whitaker et al., 2018). The lack of expression of the HMV phenotype could be due to functional mutations or lack of expression of the rmpA gene in the non-HMV K. pneumoniae isolates (Lin et al., 2020).
The multiplex-PCR methods developed during this study will facilitate future K. pneumoniae genotyping efforts and testing for virulence genes in Gram-negative bacteria. Three siderophores, salmochelin, aerobactin, and yersiniabactin, were investigated, all of which help bacteria acquire iron. Iron acquisition is a critical virulence factor for hypervirulent K. pneumoniae (Russo et al., 2015). In humans, K. pneumoniae strains possessing salmochelin often colonize the nasopharynx and are more virulent than strains lacking this virulence factor (Paczosa and Mecsas, 2016). Although salmochelin is present in only about 2-4 % of human nosocomial K. pneumoniae strains, it is much more prevalent in HMV K. pneumoniae, with one study reporting >90 % prevalence in HMV strains associated with pyogenic liver abscesses (Hsieh et al., 2008;Paczosa and Mecsas, 2016).
In our study, 3/4 of SSO HMV isolates possessed the salmochelinspecific iroD gene, while it was absent in all CSL HMV strains. Other marine mammal HMV K. pneumoniae studies have not tested for this gene, so it is challenging to draw conclusions from these data alone (Jang et al., 2010;Roe et al., 2015;Whitaker et al., 2018). Interestingly, all three SSO HMV isolates that were temporally and spatially clustered presented this gene, in contrast with all other marine mammal HMV isolates; this finding further supports the epizootic potential of HMV K. pneumoniae in SSO, and the unique clonality of those three isolates. Although salmochelin expression may enhance virulence of HMV strains, it is not critical for phenotype expression or pathogenicity (Paczosa and Mecsas, 2016). In the current study seven SSO and CSL HMV isolates did not possess iroD, while it was detected in a single non-HMV isolate from a CSL.
All marine mammal HMV isolates in this study were positive for the aerobactin specific iutA gene, compared with 3/10 non-HMV isolates. The iutA-positive non-HMV CSL isolate was also the only non-HMV isolate positive for iroD. Aerobactin can facilitate the growth and survival of human K. pneumoniae strains ex vivo and in vivo (Nassif and Sansonetti, 1986;Russo et al., 2015). Disruption of aerobactin synthesis causes significant reduction in siderophore production, decreased growth/survival ex vivo, and decreased virulence in vivo, thus possessing iutA may help enhance virulence.
Disruption of the yersiniabactin specific fyuA gene does not decrease virulence of human HMV strains possessing multiple siderophores (despite higher prevalence in HMV strains than non-HMV strains), but may play a minor role in increasing bacterial growth (Hsieh et al., 2008;Russo et al., 2015). Although all SSO HMV isolates possessed the yersiniabactin specific fyuA gene, only 50 % (3/6) of CSL isolates were positive. Two SSO non-HMV isolates were positive for fyuA; interestingly, these were the same isolates that were also positive for iutA.
Having multiple siderophores provides functional redundancy due to the importance of acquiring iron for microbes and frequency of siderophore neutralization by hosts and is often linked to virulence (Miethke and Marahiel, 2007;Sheldon and Skaar, 2020). In general, the siderophores, while showing greater prevalence in marine mammal HMV isolates, were not solely associated with HMV strains, supporting that they promote the virulence of HMV strains, but are not restricted to HMV phenotype isolates.
All marine mammal HMV isolates and 9/10 of non-HMV isolates were positive for type 3 fimbriae, which promote biofilm formation (Schroll et al., 2010). As biofilm formation has been shown to enhance survival and persistence in vivo and ex vivo (Bjarnsholt et al., 2013;Carvalho, 2018), K. pneumoniae isolates expressing type 3 fimbriae may have greater survival in the harsh marine environment, possibly increasing opportunities to infect marine mammals and other hosts. Although one study found non-HMV isolates to have greater capacity to form in vitro biofilms than HMV isolates, HMV isolates required a higher disinfectant concentration to eradicate the biofilm than non-HMV isolates and comparison of in vivo biofilms remains unclear (Soto et al., 2020). Seventy percent (7/10) of marine mammal HMV and 90 % (9/10) of non-HMV isolates were positive for type 1 fimbriae, which mediates adhesion to host epithelial cells, a property that promotes pathogenicity (Schroll et al., 2010). Although type 1 fimbriae may affect pathogenicity, they are neither unique nor critical for HMV virulence. As type 1 and type 3 fimbriae are generally found in K. pneumoniae, including environmental isolates (Podschun and Ullmann, 1998), it is not surprising that non-HMV isolates also possessed the fimbriae.
Live-stranded California sea lions that subsequently died that were infected with HMV K. pneumoniae often had pleuritis and suppurative pneumonia (Jang et al., 2010;Seguel et al., 2017), while New Zealand sea lions with fatal HMV K. pneumoniae infection commonly had Fig. 7. Percent survival of K. pneumoniae strains with defined phenotype in normal (activated) and heat-inactivated (inactivated) serum from California sea lions (Zalophus californianus). Survival and growth of K. pneumoniae incubated in serum for 3 h (A) and 24 h (B). The error bars represent standard error for triplicate to sextuplet samples. Treatments with different letters are significantly different from one another at p < 0.05. septicemia, polyarthritis, peritonitis, cellulitis/dermatitis, and meningitis . Like the New Zealand sea lions, some K. pneumoniae-positive SSOs also had septicemia, polyarthritis, peritonitis, cellulitis/dermatitis, and meningitis (Table 2 and Supplementary  material, Table 1), however not all these sites had swabs submitted for bacterial culture. Evidence of Klebsiella spp.-associated septicemia in 89 % of necropsied, Klebsiella-positive SSOs suggests that bacterial invasion and dissemination can result from some, but not all Klebsiella spp. infections.
One interesting finding was that most Klebsiella spp.-infected SSOs (13/18) had suspected or confirmed Profilicollis sp. peritonitis at necropsy (Table 2). Profilicollis sp. are acanthocephalan parasites that commonly infect the intestinal tract and peritoneum of SSOs, often causing fatal peritonitis. Prior studies have identified acanthocephalan peritonitis as a common cause of death in necropsied SSOs, and septic peritonitis is a common sequela of acanthocephalan peritonitis (Kreuder et al., 2003;Miller et al., 2020). Our preliminary data suggest that co-infection with Profilicollis sp. may enhance the risk of fatal K. pneumoniae-associated peritonitis and sepsis in SSOs.
Acanthocephalan peritonitis in SSOs is positively correlated with bacterial infection (Miller et al., 2020). Acanthocephalans may alter the enteric environment, facilitating colonization by opportunistic bacterial pathogens (Miller et al., 2020), such as K. pneumoniae. Perforations that Profilicollis sp. make through the intestinal wall as they migrate may serve as a portal for bacteria to spread regionally and systemically (Fig. 4). Concurrent enteric bacterial infections and acanthocephalan peritonitis may also contribute to formation of larger intestinal perforations and omental abscesses (Figs. 4 and 5A, B) (Miller et al., 2020). Associations between an enteric metazoan parasite and K. pneumoniae peritonitis were also noted for CSL pups with hookworm (Uncinaria spp.) infection of the small intestine (Spraker et al., 2007).
Our study confirmed that HMV K. pneumoniae strains isolated from marine mammals can resist CSL complement-mediated bacterial killing more effectively than non-HMV strains. Although HMV isolates did not show greater resistance to CSL whole blood killing than non-HMV isolates, future studies should assess longer incubation times during in vitro experimentation, particularly since non-HMV isolates exhibit faster and greater in vitro growth in broth media (Soto et al., 2020). Additionally, in-vitro challenges using whole blood and serum from various marine mammals, including SSO, will help develop a better understanding of host-pathogen relationships.

Conclusions
Our findings represent the first report of HMV K. pneumoniae detection and preliminary assessments of Klebsiella spp. associated pathology in SSOs. Based on our composite data from 20 years of testing, isolation of Klebsiella pneumoniae and other Klebsiella spp. is uncommon from SSOs. Although HMV and non-HMV K. pneumoniae strains were occasionally isolated at necropsy, no HMV or non-HMV K. pneumoniae strains have been isolated from live SSOs. In the current study, many SSO HMV and non-HMV K. pneumoniae infections were associated with moderate to severe pathology as a cause of death or sequela, and all HMV K. pneumoniae infections were assessed as a primary cause of death or occurred as a sequela to the primary cause of death.
All SSO HMV K. pneumoniae isolates were capsular type K2, the serotype most associated with HMV K. pneumoniae infections in CSLs. HMV K. pneumoniae isolates were more resistant to marine mammal serum killing in vitro than non-HMV K. pneumoniae isolates. HMV isolates from both SSOs and CSLs exhibited greater survival when exposed to CSL serum than non-HMV strains, indicating greater pathogenic potential. Temporospatial clustering and apparent clonality of some of HMV K. pneumoniae strains isolated in 2005 suggest epizootic potential, although our sample size was too small for confirmation. More virulence genes were detected in marine mammal HMV K. pneumoniae isolates when compared to non-HMV isolates, supporting potentially greater virulence as well as previously reported associations between the K2 capsular serotype and HMV phenotype for marine mammal K. pneumoniae strains. Future studies investigating bacterial killing by phagocytic cells and sampling additional tissues from SSOs that stranded in 2005 to screen for additional HMV cases are warranted. As more K. pneumoniae isolates are recovered from marine mammals, characterizing these isolates will help to clarify connections between HMV and non-HMV K. pneumoniae infections and host pathology as well as epidemiologic linkages among marine mammal host species.

Declaration of Competing Interest
The authors report no declarations of interest.