Identification and characterization

Identification and characterization of Vibrio species

by

René Erler

A thesis submitted in partial fulfillment of the requirements for the degree of

Doctor of Philosophy

Thesis Committee

Prof. Dr. Matthias S. Ullrich Molecular Microbiology Jacobs University

Prof. Dr. Frank Oliver Glöckner Microbial Genomics & Bioinformatics MPI for Marine Microbiology & Jacobs University Bremen

Dr. Antje Wichels Microbial Ecology Alfred Wegener Institute

Dr. Gunnar Gerdts Microbial Ecology Alfred Wegener Institute

Date of Defense: 19th January 2015

TABLE OF CONTENTS

2 TABLE OF CONTENTS

TABLE OF CONTENTS

SUMMARY ...... 1

INTRODUCTION ...... 3

RESEARCH AIMS ...... 13

OUTLINE ...... 15

CHAPTER I ...... 19

VibrioBase: a MALDI-TOF MS database for fast identification of Vibrio spp. that are potentially pathogenic in humans

EXCURSUS 1: Variability of MALDI-TOF MS measurements ...... 49

EXCURSUS 2: Application of VibrioBase in surveillance programs ...... 50

CHAPTER II ...... 51

Effective species identification of Vibrio spp. using mass spectrometric peaks as potential biomarkers

CHAPTER III ...... 77

Biogeographical mapping of V. cholerae, V. parahaemolyticus and V. vulnificus populations in the North and Baltic Seas using ERIC-PCR genotyping

GENERAL DISCUSSION ...... - 111

OUTLOOK ...... 121

REFERENCES ...... 123

ACKNOWLEDGMENTS ...... 141

STATUTORY DECLARATION ...... 142

4 SUMMARY

SUMMARY

Potentially pathogenic Vibrio spp. in the marine environment are regarded to as emerging health risk in Europe, particularly infections caused by V. parahaemolyticus, V. vulnificus and non-O1/non-O139 V. cholerae strains. Thus, the inclusion of Vibrio spp. in surveillance programs and studies about the distribution of Vibrio spp. are urgently needed. This thesis contributes both: the evaluation of Matrix-assisted laser / desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) as tool for species identification in environmental samples and the analysis of spatial patterns of Vibrio spp. populations in the North and Baltic Seas.

Regarding MALDI-TOF MS, the specific reference database VibrioBase was developed. VibrioBase reveals Vibrio spp. identification agreements of 96.9% compared to the approved rpoB-sequence-based identification method as well as higher identifications scores and a better discrimination for closely related Vibrio spp. compared to the general BiotyperTM database of the manufacturer. Thus, the cost-effective and rapid MALDI-TOF MS method achieves accurate Vibrio spp. identification when used in combination with VibrioBase. In addition, single mass peaks were used to allow species identifications based on sensitive peaks (SIBOPS). The SIBOPS approach achieves 98% identification agreements with the conventional MALDI-TOF MS / VibrioBase system and is independend of the respective mass spectrometric eqipment and software tools. Furthermore, mass spectrometric biomarkers were found, that might be used for the direct identification of Vibrio spp. in environmental samples. Using ERIC-PCR genotyping, general patterns were not observed for the distribution of V. cholerae, V. parahaemolyticus and V. vulnificus in the North and Baltic Seas. Particularly the latter two species were separated into distinct regions, which are however not linked to geographical aspects and, to a minor extent, are asscociated with environmental parameters such as salinity.

In consequence, MALDI-TOF MS was confirmed as a promising tool for species identifications from environmental samples. VibrioBase and SIBOPS are important steps towards the implementation of Vibrio spp. in surveillance programs executed by health authorities. Future reseach has to be done to acquire knowledge about ecological aspects of Vibrio spp. in temperate waters, in particular the linkage to plankton species in the North and Baltic Sea and the influence of bacteriophages for shaping Vibrio spp. populations.

- 1 - INTRODUCTION

- 2 - INTRODUCTION

INTRODUCTION

The genus Vibrio

The family Vibrionaceae within the class of γ-proteobacteria includes chemoorganotrophic bacteria of the genera Listonella, Photobacterium and Vibrio. The bacteria of this family are gram-negative and mostly appear as rod-shaped cells with a single flagellum. Vibrio is the largest genus and consists of more than 90 species (Thompson et al. 2004). A common characteristic feature of Vibrio spp. is the presence of two chromosomes (Heidelberg et al. 2000; Schoolnik et al. 2000). The larger chromosom I contains genes for essential cell functions whereas the smaller chromosom II carries genes that are typically found on plasmids. In general, vibrios reveal a high rate of recombination and are regarded as a phylogenetically diverse group of species (Harth et al. 2007; Vos et al. 2009).

Pathogenicity of Vibrio spp.

At least twelve species are known to cause illness in humans. All of them can be cultivated at 37°C and differ in their virulence potential. Three Vibrio species are of particular clinical importance: V. cholerae, V. parahaemolyticus and V. vulnificus (Blake et al. 1980). The remaining species are the causative agents for rare infections in humans and are considered as weak pathogens: V. alginolyticus, V. harveyi, V. mimicus, V. metschnikovii, V. fluvialis, V. furnissii, V. hollisae, V. damselae and V. cincinnatiensis (Janda et al. 1988; Thompson et al. 2004).

Globally, V. cholerae causes the most infections due to the consumption of contaminated water or food. According to the World Health Organization (WHO 2013), Cholera represents an estimated burden of 1.4 to 4.3 million cases, and 28,000 to 142,000 deaths per year, particularly on the African continent (Griffith et al. 2006; Mintz et al. 2013). The classical Cholera disease is associated with the toxin CTX that is responsible for the electrolyte imbalance and the secretion of water into the intestinal lumen, consequently leading to severe watery diarrhoea and dehydration (Muanprasat et al. 2013). The majority of Cholera cases is caused by two serotypes: O1 and O139 (Thompson et al. 2004). Infections with these highly pathogenic strains are rare in the northern hemisphere and only reported for travellers returning to America or Europe from Cholera-affected regions (Wittlinger et al. 1995). Other

- 3 - INTRODUCTION strains are commonly referred to as non-O1/non-O139 V. cholerae and cause mild-to- moderate diarrhoea, but also severe septicemia was reported (Namdari et al. 2000; Dutta et al. 2013). Infections with non-O1/non-O139 strains were reported from countries in the Baltic region such as Finland (Lukinmaa et al. 2006) and Poland (Stypulkowska-Misiurewicz et al. 2006).

In contrast, the main transmission route for pathogenic V. parahaemolyticus is the consumption of undercooked and raw shellfish. Infections with this species are the leading cause of bacterial seafood-borne gastroenteritis worldwide (Su et al. 2007). Two hemolysins are mainly responsible for the associated damage of intestinal epithelial cells: the thermostable direct hemolysin (TDH) and the TDH-related hemolysin (TRH) (Honda et al. 1985; Nishibuchi et al. 1995; Makino et al. 2003; Ritchie et al. 2012). The prevalence of the hemolysin genes tdh and trh is high in clinical (>85%) and low (<15%) in environmental isolates (Ottaviani et al. 2013; Li et al. 2014). The O3:K6 serotype of V. parahaemolyticus is of particular clinical importance (Okuda et al. 1997; Nair et al. 2007). During the last two decades, these tdh-positive strains dispersed over the world and even caused outbreaks in regions with temperate waters such as Alaska and Chile (Gonzalez-Escalona et al. 2005; McLaughlin et al. 2005). Recently, O3:K6 strains were isolated from oysters, harvested in Southern England (Powell et al. 2013). However, no V. parahaemolyticus associated infections were reported from the North and Baltic Seas, so far.

To some extend, gastroenteritis is also caused by V. vulnificus (De et al. 2011). But far more significant is the development of life-threatening septicemia from V. vulnificus, either caused by intestinal pathogenic strains or triggered by the sea-water-borne transmissions of V. vulnificus strains through open wounds. Such infections develop in people with weakened immune systems and can cause multiple organ failure with fatality rates greater than 50%; making this species to the leading cause of deaths from seafood-borne illness in the United States (Shapiro et al. 1998; Inoue et al. 2008; Jones et al. 2009). Deaths caused by V. vulnificus are sporadic reported from the Baltic Sea, particularly from the islands of Rügen and Usedom (Melhus et al. 1995; Dalsgaard et al. 1996; Ruppert et al. 2004). Although virulence-associated proteins were observed, the exact pathogenic mechanism remains unclear (Watanabe et al. 2004; Lee et al. 2007). Based on the absence/presence of tryptophanase referred to as indole test and based on DNA-based methods such as ribotyping, V. vulnificus strains are divided into two groups: the biotype I (human pathogen; indol

- 4 - INTRODUCTION positive) and the biotype II (eel pathogen, indole negative) (Tison et al. 1982; Amaro et al. 1992; Biosca et al. 1996; Arias et al. 1997).

Vibrio spp. in general are also common zoonotic agents and consequently important drivers within the marine ecosystem (Austin 2010). In particular, mass mortalities of pacific oysters are constantly reported (Garnier et al. 2007). Vibrio-related infections were also observed in marine mammals (Schroeder et al. 1985), aquatic birds (Buck 1990), fish (Grisez et al. 1995) and non-vertebrates such as shrimps (Karunasagar et al. 1994) and gorgonians (Martin et al. 2002). In consequence, infections with Vibrio spp. lead to economical losses in aquacultures (Martinez-Picado et al. 1996; Chatterjee et al. 2012). Some species such as V. mediterranei (Kushmaro et al. 2001) and V. coralliilyticus (Ben-Haim et al. 2003) cause the loss of zooxanthellae of corals. Thus, V. spp. are also associated with coral bleaching.

Occurrence of potentially pathogenic Vibrio spp. in the marine environment

Most marine bacteria belong to the uncultured microbial majority (Rappe et al. 2003). In consequence, Vibrio spp. represent only a minor fraction of culturable bacteria in seawater (Eilers et al. 2000). The highest concentration of Vibrio spp. is found in filter feeders such as clams; at least one potentially pathogenic Vibrio species was observed in every second mussel or oyster (Lhafi et al. 2007; Collin et al. 2011). Thus, filter-feeders can be considered as “hot spots”. Vibrio spp. can also be found as biofilms on the surface of zooplankton species such as copepods; an attachment process triggered by chemotactic chitin oligosaccharides and regulated by quorum sensing (Svitil et al. 1997; Cottrell et al. 2000; Heidelberg et al. 2002; Hammer et al. 2003; Gerdts et al. 2013). Furthermore, plankton particles are considered to serve as vectors for the dispersion of Vibrio spp. (Martinez-Urtaza et al. 2011). In this context, a rapid and global spread of potentially pathogenic species is also possible by the ballast water of ships (Drake et al. 2007; Emami et al. 2012; Dobbs et al. 2013).

The abundance of Vibrio spp. in the marine ecosystem is linked to environmental conditions. Particularly phytoplankton blooms and high levels of dissolved organic matter were observed to increase the density of Vibrio bacteria (Eilers et al. 2000; Heidelberg et al. 2002; Asplund et al. 2011; Oberbeckmann et al. 2011; Oberbeckmann et al. 2012). Vibrio spp. recquire salt for optimal growth and some species such as V. parahaemolyticus can even cope with higher salinities above 9% (Naughton et al. 2009). However, since the majority of species prefer

- 5 - INTRODUCTION low-to-moderate salinity levels, the highest abundances are reported from estuarine waters (Kelly 1982; Singleton et al. 1982). Furthermore, biotic components are considered to have a selective effect on the composition of Vibrio spp. populations: in particular the grazing of nanoflagellates and the lysis by bacteriophages (Beardsley et al. 2003; Garcia et al. 2013).

Nevertheless, temperature is referred to as the environmental factor with the most influence on the occurrence of certain Vibrio spp. (Kaneko et al. 1973; Kelly 1982; Kirschner et al. 2008; Oberbeckmann et al. 2011). Seasonality-dependent Vibrio assemblages were particularly observed in areas with temperate waters: the winter and spring population is dominated by non-human pathogenic cold-water species such as V. splendidus and during the summer, potentially human pathogenic warm-water species are highly abundant (Heidelberg et al. 2002; Thompson et al. 2004). Mesophilic potentially pathogenic Vibrio spp. endure low temperatures in wintertime by entering the viable-but-non-culturable state (VBNC) (Wong et al. 2004; Nowakowska et al. 2012). This starvation state is initiated by cold shock proteins and can even last for several years (Alam et al. 2007; Choi et al. 2012). Increasing temperatures in spring lead to a recovery of Vibrio spp. and, subsequently, the abundance of potentially pathogenic species is positively correlated with the sea surface temperature. Specific thresholds are suggested for increased Vibrio abundances and the related higher risk potential for infections: 15°C for V. parahaemolyticus and 20°C for V. vulnificus and V. cholerae (Kaneko et al. 1973; Depaola et al. 1990; Baker-Austin et al. 2013). In this context, global warming leads to an increase in the frequency of heat waves and hot summers (Kamae et al. 2014). In consequence, particularly Vibrio-related non-cholera infections are considered as emerging diseases for the near future (Colwell 1996; Lipp et al. 2002; Paz et al. 2007; Lindgren et al. 2012).

Particular risk areas for Vibrio infections are the and the Baltic Sea. These marginal seas of the Atlantic Ocean are of important relevance for the northern European economy and society. Both seas are characterized by low-to-moderate sea surface temperatures but are highly affected by global warming (Wiltshire et al. 2004; Belkin 2009). Longer periods with higher temperatures during the summer promote the growth of potentially pathogenic Vibrio spp. (Baker-Austin et al. 2013). Furthermore, favourable conditions for Vibrio spp. are provided by nutrient-rich freshwater influx from rivers in the North Sea and by the brackish waters of the Baltic Sea. Recent studies show, that V. alginolyticus and V. parahaemolyticus are common members of North Sea assemblages, while V. vulnificus and V. cholerae are more

- 6 - INTRODUCTION abundant in the Baltic Sea (Oberbeckmann et al. 2011; Boer et al. 2013). However, Vibrio related diseases are not considered as priority in Northern Europe and surveillance programs of health authorities contain no screening for Vibrio spp. in environmental samples. Thus, knowledge about the occurrence and distribution of potentially pathogenic Vibrio spp. is rare in these regions.

Recent methods for the isolation of Vibrio spp.

The clinical importance of Vibrio spp. and global warming driven changes necessitate surveillance programs to identify potentially pathogenic species in environmental samples. A standard protocol for the isolation from seafood is given by the U.S. Food and Drug Administration (FDA 2003); including the initial enrichment of Vibrio spp. in alkaline peptone water (APW) at 37°C to seperate potentially patgogenic species from other marine bacteria. Furthermore, the use of specific agar media is suggested to achieve separations at the species level, e.g. thiosulfate citrate bile salts sucrose (TCBS) agar and Vibrio- ChromagarTM (Pfeffer et al. 2003; Di Pinto et al. 2011).

ChromAgar was used for isolation in this thesis. This FIGURE 1: Plankton samples incubated on Vibrio-ChromagarTM. medium consists of galactose and glucose molecules linked Species-specific colonies appear red to indol, referred to as the chromogenic substrates X-Gal and (V. parahaemolyticus), green-to- turquoise (V. vulnificus and V. X-Glu. The reduction of these substrates by the enzymes cholerae), white-to-opaque (V. alginolyticus) or yellow (Shewanella beta-galactosidase and beta-glucosidase yields species- spp.). specific colony colours (Reissbrodt 2004). In consequence, the species V. cholerae/V. vulnificus (green-to-turqoise), V. parahaemolyticus (red) and V. alginolyticus (white-to-opaque) can be distinguished from each other (Figure 1). Nevertheless, other species can also appear such as V. mimicus (turquoise) and V. fluvialis (red) (Di Pinto et al. 2011). Thus, the colour of colony forming units on Vibrio-ChromagarTM is only an indication for the respective species identification.

- 7 - INTRODUCTION

Recent methods for the identification and characterization of Vibrio spp.

So far, numerous identification methods for Vibrio isolates were proposed (Table1). According to the FDA (2003), culture-dependent approaches are suggested such as biochemical and DNA-based. Well-established biochemical identification systems are commercially available, e.g. API20E and the automated Phoenix system (Smith et al. 1972; Donay et al. 2004). For Vibrio spp., however, the accuracy of these test systems is limited due to the high phenotypic diversity of Vibrio species (Alsina et al. 1994; Colodner et al. 2004). Thus, the better choice is to use DNA-based methods; in particular the toxR gene multiplex polymerase chain reaction (PCR) to distinguish the species V. cholerae, V. vulnificus, V. parahaemolyticus and V. alginolyticus from each other (Bauer et al. 2007). For the identification of pathogenic strains, several standard methods are recommended by the FDA (2003). Regarding V. cholerae, strains can be screened for the two serovars O1/O139 and the cholera toxin gene ctxA; either by serotyping or PCRs (Mantri et al. 2006). PCRs were also developed to detect the type III secretion system and the hemolysins TDH and TRH in pathogenic V. parahaemolyticus strains (Tada et al. 1992). For V. vulnificus, however, the PCR-based detection of virulence-related key genes allows no general assignments such as “pathogenic” and “non-pathogenic”. Nevertheless, based on sequence variations in the virulence correlated gene (vcg), V. vulnificus biotype 1 strains can be separated into two ecotypes: the C-types (clinical strains) and the E-types (environmental strains) (Rosche et al. 2005).

Due to the high effort, DNA-sequence based methods are not recommended for Vibrio spp. species identifications by health authorities. These methods are typically used as phylogentic tool or as reference method for checking the accuracy of other identification methods. The most general DNA-sequence based method for the identification of bacterial species is the comparative analyses of the 16S rRNA gene (Weisburg et al. 1991). However, with respect to the genus Vibrio, this gene lacks sufficient resolution for the accurate identification of closely related Vibrio spp.: particularly the congeneric species V. parahaemolyticus / V. alginolyticus and V. cholerae / V. mimicus exceed the similarity threshold for species identification of 99% (Kitatsukamoto et al. 1993). As an alternative, the nucleotide sequence data of other houskeeping genes can be used, e.g. gyrB (DNA gyrase subunit B), recA (RecA protein) or rpoB (RNA-polymerase subunit B) (Mollet et al. 1997; Thompson et al. 2004). Particularly rpoB sequence analysis was shown to reveal accurate identification results for Vibrio spp. (Ki

- 8 - INTRODUCTION et al. 2009; Oberbeckmann et al. 2011). In this method, sequence similarities lower than 85% indicate different genera, and sequence similarities between 96.5% and 98% can be used as species classification cut-offs. The best discriminatory power is, however, achieved using nucleotide data from different houskeeping genes, referred to as multiple locus sequence analyses (MLSA) (Maiden et al. 1998; Chowdhury et al. 2004). This method enables the identification of MLS-types that can be even used for phylogenetic relationship analyses below the species level (Urmersbach et al. 2014).

TABLE 1: Recent methods for the identification and characterization of Vibrio spp.

target method reference

culture-dependent methods phenotypic-based methods

enzyme reactions API 20E, Phoenix Alsina and Blanch, 1994 virulence-related serotypes serotyping Shimada et al., 1994 DNA-based methods

toxR gene multiplex PCR Bauer and Rorvik, 2007 virulence-related genes PCRs Tada et al., 1992 DNA sequence-based methods

16S rRNA gene single locus sequence analysis Kitatsukamoto et al., 1993 rpoB gene nucleotide sequence analysis Ki et al., 2009a housekeeping genes multiple locus sequence analysis Chowdhury et al., 2004 DNA genotyping-based methods

enterobacterial repetitive ERIC-PCR Keymer et al., 2009 sequences Proteom based methods

abundant cellular proteins MALDI-TOF MS Dieckmann et al., 2010

culture-independent methods

denaturing gradient gel DNA Eiler and Bertilsson, 2006 elctrophoresis (DGGE) Species- and virulence Real Time PCR Panicker and Bej, 2005 associated genes

- 9 - INTRODUCTION

Another approach is based on repetitive sequences, located in noncoding genome areas (Sharples et al. 1990; Hulton et al. 1991). Due to the genetic assortment over evolutionary time, the number and the length of repetitive sequences varies according to the similarity of strains (Wilson et al. 2006). Hence, amplification of these sequences by PCR leads to fragment patterns that are used as strain-specific profiles. Using genotyping based on these patterns allows intraspecific typing with a higher throughput than DNA sequence-based methods. Regarding Vibrio spp., the resulting genotypes are, to some extent, consistent with intraspecific groups, found by rpoB sequence analysis (Oberbeckmann et al. 2011). Enterobacterial repetitive sequences were already used by Keymer et al. for the analysis of geographical distribution for V. cholerae strains in California (Keymer et al. 2009). Thus, these sequences were also used in this thesis; in particular to obtain knowledge about the distribution of potentially pathogenic Vibrio spp. in the North and Baltic Seas.

Identification methods for surveillance programs

As appropriate, from a scientific perspective, environmental samples are to be regularly monitored for potentially pathogenic Vibrio spp.; ideally using culture-dependent DNA-based methods to reveal accurate species identification results and clues about the virulence potential of environmental isolates. However, these methods require significant effort, which precludes the use in large-scale monitoring programs. A higher throughput of environmental samples is provided by culture-independent methods such as DGGE and real-time PCRs (Campbell et al. 2003; Panicker et al. 2004; Eiler et al. 2006; Parsons et al. 2007) (Table 1). Such methods are even sensitive for Vibrio spp. in the VBNC-state (Gonzalez-Escalona et al. 2006). However, recent surveillance programs of public health authorities are based on culture-dependent methods; in particular for the identification of Enterobacteria linked to the fecal contamination. Regarding the strained financial situation of authorities, the costs are too high for an introduction of culture-independent approach, so far. In contrast, high throughput of samples and low effort is provided by the culture-dependent method MALDI-TOF MS. This species identification method needs no initial assessment like the choice of PCR primers. The speed and minimal costs of preparation and measurement for this method makes this method exceptionally well suited to be included in the general surveillance programs of health authorities; particularly with appropriate isolation methods for Vibrio spp.

- 10 - INTRODUCTION

MALDI-TOF principle

The principle of whole-cell matrix assisted laser desorption/ionization time of flight mass spectrometry (MALDI-TOF MS) is shown in Figure 2. Biomass from a liquid culture or a solid agar plate is transferred to a spot of the MALDI-target. The addition of matrix solution leads to the destruction of the bacterial cell wall and the formation of a crystal structure. Using a laser with a specific wave lengh triggers the transfer from protons (H+) of the matrix molecules to the embedded proteins in the crystal structure. Thus, proteins are ionised and released from the crystal structure at the same time (Karas et al. 2003). Within the flight tube, protein-ions are separated according to their mass-to-charge ratio (m/z) that is equal to the time-of-flight through the flight tube. A detector is measuring the ion impacts, leading to the subsequent generation of mass spectra. Each peak of the mass spectrum represents the singly- charged (H+) or multiple-charged (H+H+) ion of a protein. Acquired spectra are aligned with a reference database to calculate matching scores. In case of the BiotyperTM system of Bruker, these scores are based on m/z values of peaks and peak intensities; respective MALDI-TOF MS results rely on the species assignment of the reference entry with the best score (Maier et al. 2007). The other MALDI-TOF MS-based identification system is VitekTM of bioMerieux and comprises genus as well as species-specific peaks for the calculation of matching scores (Lartigue 2013).

Most of the detected protein-ions can be assigned to abundant cellular proteins, particularly to those involved in the process of translation such as constantly expressed ribosomal proteins (Ryzhov et al. 2001). Differences between mass spectra from distinct species rely either on the absence/presence of proteins or on the variable amino acid compositions of one protein, that, in consequence, result in distinguishable molecular masses and m/z values. Dependent on the equipment, the detection limit for whole-cell MALDI-TOF MS varies from 103 to 105 cells (Lasch et al. 2009).

- 11 - INTRODUCTION

biomass MALDI target matrix + proteins

laser shots flight tube

detector desorption/ionization H+H+ mass spectrum H+ H+H+

intensity H+ H+H+ mass/charge (m/z) H+ peak table

matching scores with reference spectra and species identification

reference database

FIGURE 2: Whole-cell MALDI-TOF MS principle. Different ionzed proteins (blue, orange, green) are separated in a flight tube according to their mass-to-charge ratio. Alignments of acquired spectra with reference spectra lead to matching scores and species identifications.

Particularly Hazen et al. (2009) and Dieckmann et al. (2010) showed the usefulness of this method for Vibrio spp. identifications (Hazen et al. 2009; Dieckmann et al. 2010). Nevertheless, two prerequisites are necessary for accurate species identification: the generation of spectra with a reasonable reliability and the establishment of high quality spectra libraries for environmental Vibrio spp.. The first requirement is, to some extent, reached by a consistent preparation of samples: especially the bacterial growth phase and the matrix mixture are important because these settings influence the proteome composition and the ionization process (Williams et al. 2003). The second requirement was part of this PhD thesis.

- 12 - RESEARCH AIMS

RESEARCH AIMS

The primary objective of this thesis was to implement the whole-cell MALDI-TOF MS method for Vibrio spp. species identification within future surveillance programs. Therefore, this fast and cost-effective species identification system needed to be improved by the following steps:

. Cultivation of potentially pathogenic Vibrio spp. from the North and Baltic Seas . Generation of a Vibrio specific MALDI-TOF MS reference database . Evaluation of this database with an approved species identification method . Validation of the discriminatory power for closely related species

Furthermore, the MALDI-TOF MS method was explored for additional improvements:

. Screening for species-specific peaks to allow software-independent identifications . Search for biomarkers to achieve direct identification from mixed samples

Another aim was to analyse the environmental isolates at the strain level. Therefore, ERIC- PCR genotyping was chosen as tool to answer the following questions:

. Are there any significant differences concerning the diversity of Vibrio assemblages from the North Sea and the Baltic Seas? . Are geographical aspects (such as distances or barriers) of decisive importance to the similarity of site-specific populations?

- 13 - OUTLINE

- 14 - OUTLINE

OUTLINE

The present thesis consists of a general introduction, three chapters representing one manuscript each and a general discussion.

Manuscript I

R. Erler, A. Wichels, E. Heinemeyer, G. Hauk, M. Hippelein, N. Torres Reyes and G. Gerdts “ VibrioBase: a MALDI-TOF MS database for fast identification of Vibrio spp. that are potentially pathogenic in humans “

published by Systematic and Applied Microbiology (2014) doi:10.1016/j.syapm.2014.10.009

This manuscript comprises the development and evaluation of the Vibrio-specific MALDI- TOF MS database VibrioBase. Reference main spectra were generated from 997 Vibrio isolates and validated using rpoB sequence analyses species identification results. The planning, the laboratory investigations and the manuscript writing were accomplished by René Erler with the assistance of Antje Wichels, Gunnar Gerdts, Nadja Torres-Reyes, Sarah Dehne and Hilke Döpke. The Vibrio isolates were provided by Ernst-August Heinemeyer, Martin Hippelein, Ciska Schets, Edda Bartelt, Christine Eichhorn, Annette Neumann, Craig Baker-Austin, Eckhard Strauch, Ralf Diekmann and Florian Gunzer.

Manuscript II

R. Erler, A. Wichels, I. Hartmann and G. Gerdts “Effective species identification of Vibrio spp. using mass spectrometric peaks as potential biomarkers “

This manuscript needs further editing for submission to a peer reviewed journal in 2015

Based on the data of Manuscript I, species-associated peaks were used for a simplified species identification. Furthermore, biomarkers are suggested for a direct species identification in environmental samples. The planning, data evaluation and manuscript writing was carried out by René Erler under the guidance of Antje Wichels and Gunnar Gerdts. Laboratory investigations were performed by René Erler and Ilka Hartmann.

- 15 - OUTLINE

Manuscript III

R. Erler, A. Wichels, L. Dlugosch and G. Gerdts “Biogeographical mapping of V. cholerae, V. parahaemolyticus and V. vulnificus populations in the North and Baltic Seas using ERIC-PCR genotyping “

This manuscript needs further editing for submission to a peer reviewed journal in 2015

ERIC-PCR genotypes were generated to reveal relationships between Vibrio populations in the North and Baltic Seas. The laboratory investigations were performed by Leon Dlugosch, Sarah Dehne and Hilke Döpke under the supervision of René Erler, Antje Wichels and Gunnar Gerdts. The data evaluation and manuscript writing was carried out by René Erler under the guidance of Antje Wichels and Gunnar Gerdts.

In addition, contributions to the following papers were given:

. S Boer, E. Heinemeyer, K. Luden, R. Erler, G. Gerdts, F. Janssen and N. Brennholt “ Temporal and Spatial Distribution Patterns of Potentially Pathogenic Vibrio spp. at Recreational Beaches of the German North Sea ”

Microbial Ecology 65(4): 1052-1067 (2013)

This publication comprises investigations about the linkage between environmental parameters and the distribution of potentially pathogenic Vibrio spp.. The PCR-based species identification and the screening for Vibrio spp. virulence factors of the isolates was performed on Helgoland, under the guidance of René Erler.

. E. Krause, A. Wichels, R. Erler and G. Gerdts “ Study on the effects of near-future ocean acidification on marine yeasts: a microcosm approach ”

Helgoland Marine Research: 1-15 (2013)

This paper investigates the impact of ocean acidification on the growth of marine yeasts. Laboratory assistance was given by René Erler; in particular for the MALDI- TOF based species identification and cluster analysis.

- 16 - OUTLINE

. S. Laakmann, G. Gerdts, R. Erler, T. Knebelsberger, P. Martinez Arbizu and M. Raupach

“ Comparison of molecular species identification for North Sea calanoid copepods (Crustacea) using proteome fingerprints and DNA sequences “

Molecular Ecology Resources 13(5): 862-876 (2013)

The species of copepods was analysed using DNA barcoding and MALDI-TOF MS. The latter investigations were carried out on Helgoland, under the guidance of René Erler.

- 17 - CHAPTER I

- 18 - CHAPTER I

CHAPTER I

VibrioBase: a MALDI-TOF MS database for fast identification of Vibrio spp. that are potentially pathogenic in humans

René Erler, Antje Wichels, Ernst-August Heinemeyer1, Gerhard Hauk2, Martin Hippelein3, Nadja Torres Reyes4 and Gunnar Gerdts

Alfred-Wegener-Institute, Helmholtz Zentrum for Polar and Marine Research, Biologische Anstalt Helgoland, Kurpromenade 201, 27498 Helgoland,

1 Governmental Institute of Public Health of Lower Saxony, Lüchtenburger Weg 24, 26603 Aurich, Germany 2 Regional Office for Health and Social Affairs of Mecklenburg-Western Pomerania, Gertrudenstraße 11, 18057 Rostock, Germany 3 University Medical Center Schleswig-Holstein, Central Facility: Medical Investigation Office and Hygiene, Brunswiker Straße 4, 24105 Kiel, Germany 4 GFZ German Research Centre for Geosciences, Helmholtz Centre Potsdam, Telegrafenberg, 14473 Potsdam, Germany

TABLE OF CONTENTS

Abstract ...... 20 Introduction ...... 21 Materials and Methods ...... 24 Results and Discussion ...... 29 Supplementary ...... 48

- 19 - CHAPTER I

Abstract

Mesophilic marine bacteria of the family Vibrionaceae, specifically V. cholerae, V. parahaemolyticus and V. vulnificus, are considered to cause severe illness in humans. Due to climate-change-driven temperature increases, higher Vibrio abundances and infections are predicted for Northern Europe, which in turn necessitates environmental surveillance programs to evaluate this risk. We propose that whole-cell matrix assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) profiling is a promising tool for the fast and reliable species classification of environmental isolates. Because the reference database does not contain sufficient Vibrio spectra we generated the VibrioBase database in this study. Mass spectrometric data were generated from 997 largely environmental strains and filed in this new database. MALDI-TOF MS clusters were assigned based on the species classification obtained by analysis of partial rpoB (RNA polymerase beta-subunit) sequences. The affiliation of strains to species-specific clusters was consistent in 97% of all cases using both approaches, and the extended VibrioBase generated more specific species identifications with higher matching scores compared to the commercially available database. Furthermore, the successful separation of two intraspecific V. alginolyticus groups indicates the immense potential of MALDI-TOF MS profiling. Therefore, we have made the VibrioBase database freely accessible, which paves the way for detailed risk assessment studies of potentially pathogenic Vibrio spp. from marine environments.

- 20 - CHAPTER I

Introduction

Vibrios are a heterogeneous group within the class Gammaproteobacteria and are typically found in marine and coastal environments throughout the world (Davis et al. 1962; Thompson et al. 2004; Farmer 2006). Well-known features of Vibrio bacteria include the formation of biofilms in locations such as the surfaces of marine eukaryotes and the accumulation in filtering organisms such as blue mussels and oysters (Lhafi et al. 2007; Yildiz et al. 2009). Some species are pathogenic in marine invertebrates, fish and mammals (Austin 2010). The main species associated with gastrointestinal illness, wound infection and septicemia in humans are Vibrio parahaemolyticus, Vibrio vulnificus and Vibrio cholerae (Daniels et al. 2000; Guerrant et al. 2006). The incidence of Vibrio infections can be increased by anthropogenic changes in the environment. One potential risk is the increasing marine traffic whereby epidemic strains can spread rapidly via ballast water (Ruiz et al. 2000; Drake et al. 2007; Emami et al. 2012). Another serious risk addressed recently by the IPCC is global warming, with a predicted temperature rise of at least 1.8°C by 2100 (IPCC 2013) This climate change can affect the spread and frequency of marine infectious diseases (Harvell et al. 2002). Consistent with this observation, the abundance of potentially pathogenic Vibrio spp. is positively correlated with seawater temperature (Oberbeckmann et al. 2012; Boer et al. 2013; Turner et al. 2013). This increase applies mainly to non-cholerae Vibrio spp., which are not considered a priority, especially in countries with temperate waters, and are therefore not included in surveillance programs (Paz et al. 2007; Baker-Austin et al. 2010; Lindgren et al. 2012). Due to global warming, increasing seawater temperatures in the form of heat waves (Wetz et al. 2013) might lead to a higher incidence of non-cholerae-related Vibrio spp. infections as proposed by Baker-Austin et al. (2013). Therefore, it is important to screen temperate waters for potentially pathogenic Vibrio spp., which would benefit the analysis of human health risks.

Sensitive and specific diagnostic tools for autonomous monitoring programs would facilitate such analyses (Burge et al. 2014). Culture-independent methods, such as quantitative real- time PCR, are important in such studies and are already used in Vibrio population analyses (Randa et al. 2004; Panicker et al. 2005; Zhou et al. 2007; Tall et al. 2012). The disadvantage of these methods is that they rely on the DNA sequences of the target genes. This dependence necessitates the design of species-specific primers for these target genes and the elimination of cross-reactions (Smith et al. 2009; Wetz et al. 2013). If these obstacles can be overcome,

- 21 - CHAPTER I culture-independent methods can be used in next-generation marine monitoring programs (Bourlat et al. 2013).

However, culture-dependent methods are currently used in routine surveillance programs because these methods involve well-established techniques such as biochemical identification systems and the enumeration of colony forming units (Tamelander et al. 2010). Biochemical methods have been used extensively for the classification of Vibrio spp. (Alsina et al. 1994; Noguerola et al. 2008); however, the accuracy of commercially available test systems is limited due to the high phenotypic diversity of Vibrio spp. (O'Hara et al. 2003). Therefore, new molecular identification methods based on conventional culture-dependent techniques would eliminate false-positive identification results (Croci et al. 2007). Oberbeckmann et al. (2011) used this approach and proposed polyphasic molecular species identification for Vibrio population analyses. This study confirms earlier observations that the 16S RNA ribosomal gene sequence-based method lacks sufficient resolution for the accurate identification of closely related Vibrio spp. and partial nucleotide sequence data of the functional gene rpoB is more suitable for identification of environmental Vibrio strains at the species level (Tarr et al. 2007; Ki et al. 2009). This is of particular importance for monitoring programs: species that are closely related but nonetheless differ with regard to their pathogenicity can be distinguished using rpoB sequence analysis and, therefore, risk assessment reports are more accurate. In this method, sequence similarities lower than 85% indicate different genera, and sequence similarities between 96.5% and 98% can be used as species classification cut-offs (Khamis et al. 2004; Adekambi et al. 2008). Other target genes or gene sets have been used to identify Vibrio strains at the species level (Thompson et al. 2005; Pascual et al. 2010). However, DNA isolation and sequencing require significant effort, which precludes the use of this method in large-scale monitoring programs.

Whole-cell matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) profiling allows a high sample throughput and measurements require less effort and cost (Gaillot et al. 2011). This technique has been used to distinguish species of the genus Vibrio (Hazen et al. 2009; Dieckmann et al. 2010) and species of other taxa despite high 16S rRNA similarity (Prisyazhnaya et al. 2012). The key aspect of this method is the rapid generation and comparison of mass spectra. The peaks of these mass spectra represent abundant cellular proteins. These proteins are separated during the passage through a flight tube in which the time-of-flight is equal to the mass to charge (m/z) ratio of each protein.

- 22 - CHAPTER I

Comparable mass spectra with species-specific peaks were initially generated by Claydon et al. (1996), Krishnamurthy et al. (1996) and Holland et al. (1996). The quality of these spectra was improved by the optimization of sample preparation procedure and matrix composition (Williams et al. 2003).

Bruker Daltonics Inc. used this technique to develop the Biotyper™ system, which generates mass spectra from the biomass of bacterial colonies within minutes, followed by a calculation of matching scores with reference spectra filed in a database (Maier et al. 2007; Mellmann et al. 2008; Sauer et al. 2008). According to the manufacturer, species identification results with matching scores higher than 2.0 are probable and those over 2.3 highly probable, whereas scores between 1.7 and 2.0 only allow accurate identifications at the genus level. Bruker Daltonics Inc. provides a large Biotyper™ reference database for species classification. Each database entry contains a main spectrum generated from at least 20 summarized and processed single spectra of individual bacterial strains. This database was mainly generated for medical laboratory diagnostics because the main spectra largely correspond to infective bacterial species isolated from clinical samples. Therefore, this database lacks coverage of environmental strains, which has led to limited ecological studies based on whole-cell MALDI-TOF MS-data.

The current Biotyper™ database (Version 3.3.1.0) contains 94 Vibrionaceae main spectra, including seven V. parahaemolyticus and five V. vulnificus but no V. cholerae main spectra. Database extension with additional main spectra can increase the matching score and the discriminatory power of MALDI-TOF MS-based classification (Christensen et al. 2012; Lau et al. 2012; Sogawa et al. 2012; Wybo et al. 2012; Calderaro et al. 2013). In this study, we focus on the applicability of MALDI-TOF MS in monitoring programs by including a total of 997 largely environmental Vibrio strains originating from different sampling sites in the North and Baltic Seas. The goal of this study was to expand the Biotyper™ database with these additional strains for the analysis of Vibrio spp. in environmental samples. Therefore, we used a two-fold approach in which we acquired data using MALDI-TOF MS and rpoB sequence analysis. The rpoB sequences were used to assign novel MALDI-TOF MS main spectra to Vibrio species because these sequences have been extensively used to accurately differentiate Vibrio bacteria. The extended MALDI-TOF MS VibrioBase was examined by comparing the MALDI-TOF MS and rpoB data, which includes verification of species identifications by cluster analyses. VibrioBase was also compared with the Biotyper™ database to test whether

- 23 - CHAPTER I the MALDI-TOF MS species matching scores were improved. In addition, V. alginolyticus strains were analyzed in detail to test whether MALDI-TOF MS can distinguish intraspecific groups.

Materials and Methods

Strains

Table 1 provides an overview of the strains used in this study, their sampling locations, origins and times. All 1,036 environmental strains originated from surveillance programs, performed from 2001 to 2012 in Western and Northern Europe. The strains were isolated from water, plankton, seafood and sediment. The environmental Vibrio spp. strains were kindly provided by the Governmental Institute of Public Health of Lower Saxony (NLGA), Regional Office for Health and Social Affairs of Mecklenburg-Western Pomerania (LAGuS), Centre for Environment, Fisheries & Aquaculture Science (CEFAS), National Institute for Public Health and the Environment (RIVM) and Lower Saxony State Office for Consumer Protection and Food Safety (LAVES). Other strains were isolated from environmental samples obtained during a survey supervised by the University Medical Center Schleswig- Holstein (UKSH) and during a research cruise to the North and Baltic Seas. Strains from an in-house culture collection of the Alfred Wegner Institute were also included. Type strains and clinical strains were provided by the university hospital in Dresden and the hospital in Bremerhaven-Reinkenheide, or purchased from the German Collection of Microorganisms and Cell Cultures (DSMZ).

Cultivation

All strains were grown on marine broth 2216 medium (Lemos et al. 1985), containing 50% or 75% seawater (1.6% or 2.4% sodium chloride). The environmental strains were incubated at 37°C. The type strains were incubated at the temperature recommended by the DSMZ (http://www.dsmz.de/).

- 24 - CHAPTER I

TABLE 1: Strain provider, number, sampling location, sample origin and sampling date of strains used in this study.

strain provider No. sampling location sample origin sampling date

CEFAS1 23 South England water, mussels, crabs 2001-2006

RIVM2 28 Netherlands water 2009-2011 NLGA3 240 Lower Saxony, Germany water 2010-2011 LAGuS4 130 Mecklenburg-Vorpommern, Germany water 2010-2011 LAVES5 82 Lower Saxony, Germany mussels 2010-2012 AWI6/UKSH7 216 Schleswig-Holstein, Germany water 2011 AWI 110 North Sea, Kattegat, Baltic Sea water, plankton, sediment 2011 AWI 207 Helgoland Roads water, plankton, mussels 2008-2009 BfR8 7 type strains diverse TUD9 1 Dresden clinical samples 2011 BRHV10 2 Bremerhaven clinical samples 2012 DSMZ11 23 type strains diverse

1 Centre for Environment, Fisheries & Aquaculture Science (United Kingdom) 2 Rijksinstituut voor Volksgezondheid en Milieu (Netherlands; National Institute for Public Health and the Environment) 3 Niedersächsisches Landesgesundheitsamt (Germany; Governmental Institute of Public Health of Lower Saxony) 4 Landesamt für Gesundheit und Soziales Mecklenburg-Vorpommern (Germany; Regional Office for Health and Social Affairs of Mecklenburg-Western Pomerania) 5 Niedersächsischen Landesamtes für Verbraucherschutz und Lebensmittelsicherheit (Germany; Lower Saxony State Office for Consumer Protection and Food Safety) 6 Alfred-Wegener-Institut für Polar- und Meeresforschung (Germany; Alfred Wegener Institute for Polar and Marine Research) 7 Universitätsklinikum Schleswig-Holstein (Germany; University Medical Center Schleswig-Holstein) 8 Bundesinstitut für Risikobewertung (Germany; Federal Institute for Risk Assessment) 9 Technische Universität Dresden (Germany, University of Dresden) 10 Bremerhaven Hospital-Reinkenheide (Germany) 11 Deutsche Sammlung von Mikroorganismen und Zellkulturen (Germany; German Collection of Microorganisms and Cell Cultures)

rpoB sequence analysis for species identification

rpoB fragments were amplified using the primers rpoB458F and rpoB2105R, as described previously (Oberbeckmann et al. 2011), sequenced using the primers rpoB458F, rpoB1110F and rpoB2105R, as described by Tarr et al. (2007) and Hazen et al. (2009). The resulting sequences with a minimum length of 1,550 bp, including 13 sequences of representative type strains, were added to a phylogenetic tree constructed from 180 rpoB reference sequences obtained from GenBank using the ARB software package (version 5.5) provided by the technical university of Munich (Ludwig et al. 2004). The phylogenetic relationships were deduced by the Neighbor Joining method, as described by Oberbeckmann et al. (2011). - 25 - CHAPTER I

According to Adekambi et al. (2009) and Khamis et al. (2005), rpoB sequences with a minimum similarity of 96.5% were grouped into clusters. Reference sequences and sequences of type strains determined the assignment of resulting clusters to a species or species group. The environmental strains in this study were named according to their cluster assignment.

Nucleotide sequence accession numbers

The sequences obtained in this study are available from GenBank under the accession numbers KJ647447 - KJ648151 (https://www.ncbi.nlm.nih.gov/genbank).

MALDI-TOF MS

Strains were grown overnight on agar plates as described above. One loop of biomass from each culture was suspended in 300µL of LC-MS water. Nine hundred microliters of pure ethanol was added to sterilize the bacteria and denature the proteins. The samples were stored at -20°C and thawed before further MALDI-TOF MS sample preparation. The cellular proteins were extracted using a previously described formic acid/acetonitrile method (Mellmann et al. 2008). A 1.2µL aliquot of each protein extract was spotted onto the sample target in 8 replicates (MSP 96 target polished steel) and overlaid with 1.2µL of saturated CHCA matrix solution (alpha-cyano-4-hydroxycinnamic acid dissolved in 50% acetonitrile, 47.5% LC-MS water and 2.5% trifluoroacetic acid). Spectra were acquired using the microflex LT/SH system (Bruker Daltonics Inc., Bremen). Each spot was measured 4 times, resulting in 32 single spectra for each strain. According to the Bruker Daltonics Inc. specifications, low-quality spectra (e.g., outliers) were excluded using the flexAnalyses software (version 3.3, Bruker Daltonics Inc.). After this quality check, the single spectra of each strain were summarized and processed using the Biotyper™ software (version 3.1) to generate main spectra, each containing the 70 most prominent peaks of the single spectra. The main spectra were analyzed using BioNumerics™ (version 7.1, Applied Maths, Ghent). Similarity coefficients were calculated with the Jaccard Index (minimum height 1%, shift factor 1, constant tolerance 0.5, linear tolerance 400ppm), and UPGMA (unweighted pair group method with arithmetic mean) was used to perform hierarchical cluster analysis based on mass spectrometric information. Group-specific peaks were detected using the peak matching tool of BioNumerics™ (constant tolerance 0.5, linear tolerance 300ppm, maximum horizontal shift 1). Main spectra with equal rpoB species identification results were grouped

- 26 - CHAPTER I into clusters and rpoB sequences of strains determined the assignment of resulting clusters to a species or species group as well.

Database development rpoB sequence analysis data were available for many but not all main spectra. Initially, main spectra originating from strains with available rpoB sequences were used to compare the species identification results of MALDI-TOF MS and rpoB sequence analyses. The main spectra with deviant species identifications were excluded from the extended database. The final version of VibrioBase was developed in a four-step process: (1) main spectra with identical rpoB species identification cross-reference were included to form the core of VibrioBase; (2) cluster analysis was performed to classify species-specific clusters of main spectra; (3) the remaining main spectra with no rpoB species identification cross-reference were assigned based on their affiliation with species-specific MALDI-TOF MS clusters; (4) main spectra affiliated with MALDI-TOF MS clusters with no clear rpoB cross-reference were excluded.

VibrioBase analysis

Spectra of four selected strains representing the species V. alginolyticus, V. parahaemolyticus, V. cholerae and V. vulnificus were aligned with the respective VibrioBase main spectra of the same species to examine whether the database enhancement generated differences in the matching scores and to estimate the number of main spectra required for a comprehensive MALDI-TOF MS database. The alignments generated 252 (V. alginolyticus), 306 (V. parahaemolyticus), 64 (V. cholerae) and 216 (V. vulnificus) matching scores. Depending on the corresponding amount of main spectra, the expected highest matching scores were calculated using the following equations:

where p is the probability that the kx highest score (S) is the highest matching result in a database with d database entries, D is the total number of scores, and E is the expected highest score of a database with d database entries. These equations can be elucidated by hypothetical

- 27 - CHAPTER I matching scores; for instance, scores of 2.6, 2.4, 2.3, 2.2 and 2.0 (D=5) generate the following probabilities and expected highest scores:

d1 d2 d3 d4 d5 d1 d2 d3 d4 d5 k1 0.2 0.4 0.6 0.8 1.0 k1 0.52 1.04 1.56 2.08 2.60 k2 0.2 0.3 0.3 0.2 0 k2 0.48 0.72 0.72 0.48 0 k3 0.2 0.2 0.1 0 0 k3 0.46 0.46 0.23 0 0 k4 0.2 0.1 0 0 0 k4 0.44 0.22 0 0 0 k5 0.2 0 0 0 0 k5 0.40 0 0 0 0 E(d) 2.30 2.44 2.51 2.56 2.60

Correlations between the expected highest scores (E) and number of main spectra (d) were analyzed graphically using Sigma Plot (version 11.0). These four spectra were also aligned with the Biotyper™ database (version 3.3.1.0).

- 28 - CHAPTER I

Results and Discussion

rpoB sequence analysis and VibrioBase development

Because robust species identifications can be performed based on rpoB gene sequences (Mollet et al. 1997; Kupfer et al. 2006; Tarr et al. 2007; Ki et al. 2009; Ki et al. 2009), in this study, rpoB classification results were used to determine the classification of MALDI-TOF MS clusters to a species or species group. A total of 1,093 sequences were included in the rpoB sequence analyses (Table 2). Among these, 208 sequences were obtained from the data of Oberbeckmann et al. (2011), and 180 additional sequences were imported from GenBank. The remaining 705 sequences were generated in this study. Except the V. harveyi/campellii group, all rpoB species clusters reached the corresponding similarity threshold of 96.5% (Suppl. Table S-1). The phylogenetic position and identity of type strain sequences and of reliable GenBank entries (e.g., rpoB genes extracted from whole-genome sequences) determined the positioning and assignment of new strains by adding the rpoB sequences to the existing phylogenetic tree (Figure 3).

A total of 1,057 main spectra were created. Partial rpoB sequences and the corresponding species identification results were available for 903 strains (85%) in the MALDI-TOF MS VibrioBase (Table 2). The grouping of main spectra into species clusters was based on the available rpoB results and revealed intraspecific similarity values of 24.8% to 43.5% and interspecific similarity values of 2.0% to 21.2% (Suppl. Table S-2). However, high coefficients of variation reflect the higher variability of MALDI-TOF MS measurements compared to rpoB sequence analysis. Therefore, no general MALDI-TOF MS similarity thresholds can be defined for Vibrio species – a circumstance which, in turn, displays the need for more main spectra in the database to provide secure species identification results.

Table 2 contains detailed information on the main spectra: the majority of the analyzed strains (77%) were classified as V. parahaemolyticus, V. vulnificus, V. cholerae and V. alginolyticus, and the remaining strains mostly belonged to the species and species groups V. fluvialis, V. harveyi/campbellii, V. mimicus, V. navarrensis, V. aestuarianus, V. diazotrophicus, V. pacinii-related group, V. metschnikovii-related group and Listonella anguillarum.

- 29 - CHAPTER I

TABLE 2: Number of total rpoB sequences (internal constructed + external sequences from GenBank), total main spectra and number of rpoB sequenced strains, number of correct/incorrect species agreements (MALDI- TOF MS versus rpoB) and ratio between correctly identified main spectra and the number of rpoB sequenced strains.

species / groups according to the rpoB sequences main spectra species agreements MALDI-TOF MS cluster analyisis total internal external total rpoB correct incorrect ratio Listonella anguillarum 12 10 2 18 9 9 0 100 % Shewanella spp. 0 0 0 8 4 0 4 0 % V. aestuarianus 10 9 1 7 5 5 0 100 % V. agarivorans 1 1 0 1 1 1 0 100 % V. alginolyticus group 1+2 252 218 34 255 214 211 3 98.6 % V. cholerae 74 65 9 64 58 58 0 100 % V. coralliilyticus 1 1 0 1 1 1 0 100 % V. diazotrophicus 11 11 0 12 10 10 0 100 % V. ezurae 0 0 0 1 0 0 0 V. fluvialis 30 24 6 30 25 24 1 96.0 % V. fortis 1 1 0 1 1 1 0 100 % V. furnissii 6 1 5 1 1 1 0 100 % V. gazogenes 1 1 0 1 1 1 0 100 % V. gigantis 1 1 0 1 1 1 0 100 % V. harveyi/campbellii group 65 41 24 46 41 39 2 95.1 % V. kanaloae 1 1 0 1 1 1 0 100 % V. mediterranei 1 1 0 1 1 1 0 100 % V. metschnikovii-related group 9 6 3 6 6 6 0 100 % V. mimicus 28 22 6 28 22 21 1 95.5 % V. natriegens 11 1 10 1 1 1 0 100 % V. navarrensis 11 9 2 10 10 9 1 90.0 % V. pacinii- related group 0 0 0 6 2 0 2 0 % V. parahaemolyticus 353 281 72 311 276 271 5 98.2 % Vibrio spp. group / V. furnissii-related group 13 13 0 11 9 9 0 100 % V. superstes 0 0 0 1 0 0 0 V. tasmaniensis 1 1 0 1 1 1 0 100 % Vibrio spp. (not identified) 0 0 0 10 6 0 6 0 % V. vulnificus 1+2a group 196 190 6 219 192 189 3 98.4 % V. vulnificus 2b group 3 3 0 3 3 3 0 100 % V. xuii 1 1 0 1 1 1 0 100 % Total 1,093 913 180 1,057 903 875 32 96.9 %

- 30 - CHAPTER I

MALDI-TOF MS and DNA sequence-based species classifications are comparable (Mellmann et al. 2008; Rezzonico et al. 2010; Verroken et al. 2010; Alatoom et al. 2012); for instance Benagli et al. (2012) validated 93% of MALDI-TOF MS-based Aeromonas species classifications by gyrB sequence analysis; considering that the remaining 7% were not identified because of missing reference spectra. In this study, the majority of Vibrio strains (97%) identified by their species-specific MALDI-TOF MS cluster were consistent with the identification results based on rpoB sequence analyses (Table 2). In total, 28 species identification results were inconsistent. Only 5 of these discrepancies were observed within the V. parahaemolyticus/alginolyticus/harveyi/ campbellii group. One strain in the V. cholerae/mimicus group was not consistent with the classification by rpoB sequence analysis. No discrepancies were observed between the closely related species V. vulnificus and V. navarrensis. Therefore, only minor classification differences (MALDI-TOF MS vs. rpoB) were observed in congeneric Vibrio species. The main spectra with deviant species identifications were excluded from the extended database. VibrioBase main spectra with identical rpoB species identification cross-reference (n=875) formed the core of VibrioBase (Table 2). The remaining main spectra with no available rpoB species identification cross- reference (n=154) were assigned according to their affiliation with species-specific MALDI- TOF MS clusters followed by the removal of clusters with no clear species assignment (six clusters with 32 main spectra). Finally, a total of 997 main spectra were compiled in VibrioBase (Table 3).

TABLE 3: Number of main spectra filed in the Biotyper™ database and VibrioBase.

MALDI-TOF MS Biotyper™ VibrioBase MALDI-TOF MS Biotyper™ VibrioBase

species cluster main spectra main spectra species cluster main spectra main spectra

Listonella anguillarum 7 18 V. harveyi/campbellii 4 44 V. aestuarianus 1 7 V. kanaloae 1 1 V. agarivorans 1 1 V. mediterranei 1 1 V. alginolyticus 4 252 V. mimicus 1 27 V. cholerae 0 64 V. natriegens 1 1 V. coralliilyticus 1 1 V. navarrensis 1 9 V. diazotrophicus 2 12 V. parahaemolyticus 7 306 V. ezurae 1 1 V. superstes 1 1 V. fluvialis 3 29 V. tasmaniensis 1 1 V. fortis 1 1 V. vulnificus 5 216 V. furnissii 2 1 V. xuii 1 1 V. gazogenes 1 1 Total 49 997 V. gigantis 1 1

- 31 - CHAPTER I

VibrioBase evaluation

The main spectra in VibrioBase were aligned with the current Biotyper™ database (version 3.3.1.0) to examine whether the new main spectra from environmental strains complemented pre-existing MALDI-TOF MS species clusters in the Biotyper™ database (Figure 1). The highest matching scores of these alignments with the Biotyper™ database were divided into ranges: highly probable species identifications (> 2.3), probable species identifications (2.0- 2.3), reliable genus identifications (1.7-2.0) and unreliable identifications (< 1.7). Strains with Biotyper™ species classifications different from the species assignment of VibrioBase main spectra were integrated as “different identification”. 41.5% of the new main spectra had a matching score over 2.3 (“highly probable species identification”); 38.1% had a score between 2.0 and 2.3 (“probable species identification”); 10.9% had scores verifying only the genus level (1.7 to 2.0); 1.6% had score values below 1.7 (no reliable identification); 5.6% of the alignments led to a different species identification. The best results were obtained for V. parahaemolyticus and V. fluvialis with 88% and 89% matching scores, respectively, higher than 2.3 (Figure 1). Among the V. vulnificus main spectra 51% were classified with a score above 2.3 and 45% with a score between 2.0 and 2.3. Most of the V. alginolyticus main spectra (81%) had scores between 2.0 and 2.3. The majority of V. diazotrophicus (82%) and Listonella anguillarum (89%) main spectra were classified with a matching score between 2.0 and 2.3. Therefore, the existing Biotyper™ database provides high matching scores for these six species.

However, no probable identification was possible in case of species with only one Biotyper™ database main spectrum, such as V. mimicus, V. navarrensis and V. aestuarianus (Table 2, Figure 1). V. mimicus strains were identified either as V. mimicus (17 of 28) or V. albensis (V. cholerae biovar albensis; 10 of 28); all spectra except one had score values lower than 2.0. Six of ten V. navarrensis main spectra were classified correctly as V. navarrensis. The remaining four spectra of this species cluster were classified as V. vulnificus (Figure 1 “different identification”), with score values between 1.7 and 2.0. Only two of seven V. aestuarianus strains were classified correctly; three were identified as L. anguillarum, with scores lower than 2.0. Four Biotyper™ database main spectra are available for V. harveyi, (Table 1). However, all V. harveyi strains except one did not have probable species classification scores (> 2.0). A third of these strains were identified as V. harveyi by the Biotyper™ database. Most V. harveyi strains were misidentified as V. parahaemolyticus

- 32 - CHAPTER I

(n=18), V. mytili (n=5), V. rotiferianus (n=3), V. ponticus (n=2) or V. fortis (n=1). These results reveal that the available Biotyper™ database does not encompass the variability of some Vibrio species, suggesting that this database must be extended using the VibrioBase main spectra to perform highly probable species identifications.

FIGURE 1: Classification of new main spectra with the Biotyper™ database. Resulting highest matching scores of new main spectra were divided into ranges reflecting highly probable species identifications (> 2.3), probable species identifications (2.0 – 2.3), secure genus identifications (1.7-2.0) and unreliable identifications (< 1.7). Strains with Biotyper™ species classifications different from the species assignment for VibrioBase main spectra were integrated as “different identification”.

The comparison of the VibrioBase and Biotyper™ scores of randomly selected V. alginolyticus, V. parahaemolyticus, V. cholerae and V. vulnificus strains enabled the specification of two important values to emphasize the benefits of VibrioBase: 1) the difference between the highest matching scores of VibrioBase and the Biotyper™ database and 2) ratios of corresponding VibrioBase main spectra with matching scores higher and lower than the highest matching score of the Biotyper™ database. The alignment with the newly extended VibrioBase database revealed the following matching score differences in the corresponding highest main spectra: +0.290 (2.542 to 2.252; V. alginolyticus), +0.913 (2.628 to 1.715; V. cholerae), +0.228 (2.609 to 2.381; V. parahaemolyticus) and +0.310 (2.644 to 2.334; V. vulnificus) (Table 4). In case of V. parahaemolyticus and V. vulnificus, highly probable species identification (score value > 2.3) was possible using the Biotyper™ database, whereas alignment with VibrioBase was required to identify V. alginolyticus and V. cholerae at this level (Table 4). These results are consistent with Christensen et al. (2012),

- 33 - CHAPTER I who observed even 3.6-fold higher “highly probable” species identifications and similar score increases ranging from 0.038 to 0.527 using an extended database of Gram-positive bacteria.

TABLE 4: Strains used to evaluate VibrioBase analogous to Figure 2. Species, strain ID, number of the corresponding main spectra of VibrioBase and Biotyper™ database, best matching score values with VibrioBase and Biotyper™ database, best matching score differences, ratios between VibrioBase scores higher and scores lower than the highest Biotyper™ database score are shown.

VibrioBase Biotyper database database comparison number number of ratio of VibrioBase best score species strain of main best score main best score scores > best difference spectra spectra Biotyper score

V. alginolyticus VN-2973 252 2.542 4 2.252 0,290 73.5%

V. cholerae VN-3016 64 2.628 0 1.715 0,913 100%

V. parahaemolyticus VN-2809 306 2.609 7 2.381 0,228 79.4%

V. vulnificus VN-2970 216 2.644 5 2.334 0,310 74.4%

More than 70% of the new main spectra in VibrioBase generated matching scores higher than the best match with the Biotyper™ database (Table 4): 73.5% (V. alginolyticus), 79.4% for V. parahaemolyticus and 74.4% for V. vulnificus main spectra. These data indicate that addition of the main spectra of environmental Vibrio spp. strains to the database does not result in separate clusters but expands the range of variability of each species, which leads to a significant increase in matching score. However, V. cholerae is an exception to this rule because all new V. cholerae main spectra scores were higher than the best Biotyper™ database match (Table 4), which led to the delineation of a new MALDI-TOF MS cluster. Alignment of V. cholerae strains with the Biotyper™ database always resulted in V. cholerae biovar albensis identifications, with matching score values mainly (87%) between 1.7 and 2.0 (Figure 1); these data are consistent with the results of Khot et al. (2012) and can be explained by the absence of V. cholerae biovar cholerae main spectra in the officially available Biotyper™ database, which was previously reported by Saffert et al. (2011) (Saffert et al. 2011).

- 34 - CHAPTER I

The calculation of expected highest matching score values depending on the number of main spectra supports the hypothesis that a sufficient number of main spectra in the database is crucial for highly probable MALDI-TOF MS species classification (Figure 2). The expected highest matching score increases with the database size until it reaches saturation. V. vulnificus exhibited this saturation with 50 V. vulnificus main spectra, and V. parahaemolyticus and V. alginolyticus exhibited saturation only with 100 to 150 main spectra. These values represent the least number of required main spectra to account for the intraspecific variations.

2,7 2,7

2,6 2,6

2,5 2,5

2,4 2,4

V. alginolyticus strain VN-2973 2,3 2,3 V. cholerae strain VN-3016

0 50 100 150 200 250 0 20 40 60 number of V. alginolyticus main spectra in VibrioBase number of V. cholerae main spectra in VibrioBase 2,7 2,7

2,6 2,6

2,5 2,5

2,4 2,4

V. parahaemolyticus strain VN-2809 2,3 2,3 V. vulnificus strain VN-2970

0 50 100 150 200 250 300 0 50 100 150 200 number of V. parahaemolyticus main spectra in VibrioBase number of V. vulnificus main spectra in VibrioBase FIGURE 2: Correlations between the number of main spectra filed in VibrioBase and expected highest matching scores. Examples are illustrated for V. alginolyticus, V. cholerae, V. parahaemolyticus and V. vulnificus. Highest matching scores of the VibrioBase as well as the Biotyper™ database are given in Table 4.

- 35 - CHAPTER I

Phylogenetic and cluster analysis

The rpoB phylogenetic tree and the MALDI-TOF MS cluster analysis are illustrated in Figures 3 and 4. Overall both methods led to roughly equal clusters. However, no clear species assignment was possible for the following four MALDI-TOF MS clusters: a V. pacinii-related group, a V. metschnikovii-related group, a V. furnissii-related group and the V. vulnificus group 2b. The main spectra of these clusters were excluded from VibrioBase but can be included later if future analyses provide further information on their accurate affiliation with a species of the genus Vibrio.

In the V. pacinii-related group, six main spectra form a cluster in the MALDI-TOF MS tree (Figure 4). The best match for these main spectra in the Biotyper™ database is V. pacinii, with scores between 1.894 and 2.252. However, rpoB sequences are only available for two of these strains (Table 2), which were classified as V. alginolyticus and V. furnissii in the rpoB sequence analysis. A V. pacinii reference sequence was not available in GenBank, and amplification of the rpoB gene of the V. pacinii type strain (LMG 19999) was unsuccessful. Therefore, a V. pacinii-related cluster is absent in the rpoB tree (Figure 3), but present in the MALDI-TOF MS cluster analysis (Figure 4).

Six other strains were pooled in a V. metschnikovii-related group because rpoB sequences of these strains are placed on a branch near the group with the three V. metschnikovii rpoB reference sequences available in GenBank (Figure 3). Both groups had only 95.4% rpoB sequence similarity and therefore do not meet the species assignment described by Adekambi et al. (2008). Alignment of these six strains with the MALDI-TOF MS Biotyper™ database did not generate reliable species identifications, although V. metschnikovii main spectra are present in the Biotyper™ database. Eleven main spectra did not result in any reliable species identification with the Biotyper™ database and formed a distinct Vibrio spp. group in the MALDI-TOF MS cluster analysis (Figure 4). The rpoB sequence similarity to V. furnissii (92%) was insufficient for assignment of this species (Adekambi et al. 2008), but this similarity suggests the relatedness of this group to V. furnissii, which was indicated by naming this group V. furnissii-related group in the rpoB tree (Figure 3). Therefore, strains assigned to the V. furnissii-related group identified by rpoB sequencing analyses are identical to those classified in the Vibrio spp. group in the MALDI-TOF MS cluster analysis (Table 2). Three other strains formed a separate V. vulnificus 2b group in the rpoB phylogenetic tree and

- 36 - CHAPTER I in the MALDI-TOF MS tree (Figures 3 and 4; Table 2). No reliable species classifications were possible using the Biotyper™ database, and rpoB sequence analysis revealed an identity of only 95% with V. vulnificus. Biochemical analysis also revealed differences compared to the biovar pommeriensis of V. navarrensis (Jores et al. 2003; Jores et al. 2007) (data not shown). Further studies are required to elucidate the species assignments of these four groups.

For Vibrio species, which are difficult to resolve at the species level by 16S rRNA sequencing but are distinguishable by rpoB sequence analysis, it is important to consider whether MALDI-TOF MS analysis has the discriminatory power of rpoB sequence analysis. This is particularly important for the differentiation of potentially pathogenic Vibrio spp. from congeneric species with lower virulence potential such as V. parahaemolyticus and V. alginolyticus, V. vulnificus and V. navarrensis as well as V. cholerae and V. mimicus.

V. parahaemolyticus and V. alginolyticus are closely related, with 99.4% similarity at the 16S rRNA level (Kitatsukamoto et al. 1993) and an average rpoB sequence similarity of up to 96.8 % (Tarr et al. 2007). Both methods used in our study revealed the potential separation of these species, consistent with the results of Hazen et al. (2009); they used binary analysis of the absence or presence of MALDI-TOF MS peaks and observed 64% to 80% intraspecific similarity for V. parahaemolyticus (20 strains) and 56% to 80% interspecific similarity between V. parahaemolyticus and V. alginolyticus (3 strains). We analyzed a much larger number of strains (Table 3) compared to their study, and the intensity of peaks was considered for calculation of the similarity matrix, which revealed greater divergence: 0.8 to 97.1% for the intraspecific similarity among V. parahaemolyticus strains and 2.4 to 48.1% for the interspecific similarity between V. parahaemolyticus and V. alginolyticus. These data reveal the importance of sufficient main spectra to encompass the variability of Vibrio spp. However, despite these variable values, most V. parahaemolyticus strains were identified correctly by MALDI-TOF MS, with a specificity of 98.1% (Table 2) and a sensitivity of 96.1%. Therefore, highly probable V. parahaemolyticus MALDI-TOF MS species identifications can most likely be generated in future studies using the VibrioBase main spectra.

- 37 - CHAPTER I

FIGURE 3: Phylogenetic tree of Vibrio strains based on rpoB sequence analysis with a minimum length of 1,550bp. The phylogenetic relationship was deduced by neighbour joining method. The tree contains 1,113 sequences: 180 rpoB sequences from GenBank including Photobacterium (n=11), V. metschnikovii (n=9) and V. fischerii (n=1) sequences which are not listed in Table 2; and 913 in-house sequences including 13 type strain sequences. The scale bar represents 10 nucleotide substitutions per 100 nucleotides sequences.

- 38 - CHAPTER I

FIGURE 4: Phylogenetic tree based on MALDI-TOF MS data analyzed by BioNumerics™. The scale bar represents similarity coefficients calculated with the Jaccard Index (Minimum height 1%, shift factor 1, constant tolerance 0.5, linear tolerance 400 ppm). Cluster analysis was performed with UPGMA.

- 39 - CHAPTER I

V. navarrensis and V. vulnificus form another congeneric species pair. V. navarrensis is generally not widespread and was firstly isolated in Spain (Urdaci et al. 1991); however, Macián et al. (2000) observed a large number of V. navarrensis strains in seawater samples, and the strains analyzed in our study were also isolated from distinct areas such as Helgoland Roads, Kattegat and Baltic Sea. Incorrect identification of this species as V. vulnificus is highly likely because 1) V. navarrensis can grow on culture media optimized for V. vulnificus (Cerda-Cuellar et al. 2000), 2) the biochemical profiles of V. navarrensis and V. vulnificus are similar (Gladney 2014), 3) PCRs targeting the toxR gene of V. vulnificus yielded partially false-positive in our study (data not shown), 4) alignment with the MALDI-TOF MS Biotyper database resulted in 40% V. vulnificus matches (Figure 1). Therefore, these two species must be discriminated to avoid incorrect Vibrio vulnificus classifications, and this would enable more accurate risk evaluation in surveillance programs. VibrioBase provides sufficient main spectra for a clear differentiation of V. navarrensis and V. vulnificus, which is evident in the distinct clusters of these species in the MALDI-TOF MS tree (Figure 4).

The same principle applies to V. mimicus and V. cholerae. V. mimicus was previously categorized as biochemically atypical Vibrio cholerae, which is evident in the nearly complete identity of their 16S rRNA (Davis et al. 1981; Ruimy et al. 1994) and an average rpoB sequence similarity of 95.8% (Tarr et al. 2007). Alignment of V. mimicus strains with the Biotyper™ database yielded identifications only at the genus level (matching scores < 2.0) (Figure 1). Dieckmann et al. (2010) distinguished these two species clearly by MALDI-TOF MS, which is consistent with our results (Figure 4). Differentiation of these species is particularly important because of the CTXΦ bacteriophage in V. cholerae and V. mimicus; this bacteriophage carries the gene encoding the cholera toxin and can be transferred between strains (Boyd et al. 2000; Shinoda et al. 2004). Therefore, V. mimicus is also associated with clinical diarrhea cases (Shandera et al. 1983; Campos et al. 1996), and identification methods based on target genes must be evaluated continuously to avoid misidentification due to horizontal gene transfers. However, MALDI-TOF MS main spectra are based on the molecular masses of nearly 70 (mainly ribosomal) proteins (Ryzhov et al. 2001), which are unlikely to be transferred horizontally (Jain et al. 1999; Cohen et al. 2011). Therefore, VibrioBase is a powerful tool to reliably differentiate between V. mimicus and V. cholerae.

- 40 - CHAPTER I

In this study, three other pairs of closely related species were analyzed; these species have low or unknown pathogenic potential in humans: V. fluvialis and V. furnissii, Listonella anguillarum and V. aestuarianus as well as V. harveyi and V. campbellii.

V. fluvialis and V. furnissii are part of the cholera clade (Sawabe et al. 2007) and V. furnissii was initially assigned as V. fluvialis biotype II (Brenner et al. 1983). In contrast to Dieckmann et al. (Dieckmann et al. 2010), differentiation between V. fluvialis and V. furnissii was possible in our study with the limitation that only one V. furnissii main spectrum is available in VibrioBase (Figure 4). Therefore, this cluster must be expanded by the addition of new main spectra to achieve a reliable differentiation between V. fluvialis and V. furnissii. However, both species are not placed near V. cholerae in the MALDI-TOF MS cluster analysis (Figure 4) whereas the rpoB phylogenetic analysis in our study confirms the relationship with V. cholerae (Figure 3); more generally, pointing out that MALDI-TOF MS is a taxonomic tool rather than a phylogenetic method.

Listonella anguillarum and V. aestuarianus are included in the anguillarum clade (Thompson et al. 2005; Sawabe et al. 2007), which is consistent with the rpoB phylogeny and the MALDI-TOF cluster analysis in our study (Figures 3 and 4). Seven main spectra of Listonella anguillarum are present in the MALDI-TOF MS Biotyper™ database but only one exists for V. aestuarianus (Table 3). Therefore, clear Listonella anguillarum identification was possible, and 60% misidentification was observed for V. aestuarianus (Figure 1). However, a specificity of 100% was obtained for V. aestuarianus using VibrioBase (Table 2). Therefore, we showed that Listonella anguillarum and V. aestuarianus can be differentiated and identified by rpoB sequencing and MALDI-TOF MS.

V. harveyi and V. campbellii are very difficult to differentiate due to their high rpoB sequence similarity (99.77%) (Tarr et al. 2007). These species can only be differentiated using copious DNA sequence-based methods such as multi-locus-sequence-analyses (Thompson et al. 2007). Therefore, these species were assigned to the V. harveyi/campbellii group because the presence of V. campbellii strains in this group could not be excluded. However, the V. harveyi/campbellii group is characterized by high heterogeneity (Vandenberghe et al. 2003), and revealed a low intraspecific rpoB similarity (Table 2). Considering the 63% misidentified strains (Figure 1), the number of main spectra for V. harveyi in the Biotyper™ database (4 main spectra) (Table 3) is insufficient to account for the high intraspecific variability of this

- 41 - CHAPTER I species. However, using VibrioBase, 95% of the V. harveyi/ campbellii strains were identified correctly (Table 2), and placement of this cluster in the MALDI-TOF MS tree (Figure 4) reveals the affiliation of this group with the core clade of Vibrionaceae: V. parahaemolyticus, V. alginolyticus, V. harveyi and V. natriegens (Dorsch et al. 1992). V. natriegens, a more distant member of the Vibrio core clade (Pascual et al. 2010), can be separated from other species of this core clade in both, the rpoB sequence and MALDI-TOF MS analyses (Figures 3 and 4).

V. diazotrophicus forms an unique clade (Sawabe et al. 2007) due to its low 16S rRNA similarities (94 to 97%) with other Vibrio spp. such as V. fluvialis, V. cholerae, V. vulnificus and V. anguillarum (Aznar et al. 1994). The rpoB and MALDI-TOF MS analyses in our study reflect this uniqueness of V. diazotrophicus in the phylogeny of Vibrionaceae (Figures 3 and 4).

Other Vibrio type strains were included in our analyses. Five of these strains, V. gigantis, V. kanaloae, V. tasmaniensis, V. fortis and V. agarivorans, can be pooled in a V. splendidus clade (Macian et al. 2001; Thompson et al. 2003; Thompson et al. 2003; Thompson et al. 2003; Le Roux et al. 2005). The nodes and branches of these five type strains are identical in the rpoB tree and MALDI-TOF MS cluster analysis (Figures 3 and 4), which reveals identical phylogenetic resolution using rpoB and MALDI-TOF MS analyses. Both analyses also reflect the close relationship of V. corallilyticus and V. mediterranei (Ben-Haim et al. 2003). V. xuii is related to these two species (Thompson et al. 2003) suggesting that the placement of V. xuii is phylogenetically correct in the rpoB tree and incorrect in the MALDI-TOF MS cluster analysis (Figure 3 and Figure 4). No close relationships were observed between V. gazogenes and other Vibrio species analyzed in this study which supports the hypothesis by Dikow et al. (2013) that V. gazogenes holds an unique position at the margin of the Vibrionaceae, near Salinivibrio spp., Grimontia spp. and Photobacterium spp.. Although no rpoB sequences were available for V. ezurae and V. superstes, their placement in neighboring branches in the MALDI-TOF MS tree (Figure 4) highlights their affiliation with the V. halioticoli clade (Hayashi et al. 2003; Sawabe et al. 2004). Therefore, although these observations are based on single type strains for each species, they reveal that MALDI-TOF MS analysis does not fully display the developed phylogenetic scheme of Vibrionaceae.

- 42 - CHAPTER I

Intraspecific typing of V. alginolyticus main spectra

Whole-cell MALDI-TOF MS analysis has been used to identify clinically relevant subspecies such as hyper-virulent Clostridium difficile strains (Reil et al. 2011), vancomycin-resistant Enterococci (Griffin et al. 2013) and methicillin-resistant Staphylococcus aureus strains (Wolters et al. 2011). In a study by Oberbeckmann et al., (2011) two distinct rpoB sequence- based groups of V. alginolyticus have been observed, and V. alginolyticus group 1 is considered more closely related to V. parahaemolyticus than V. alginolyticus group 2. Therefore, the V. alginolyticus strains in our study and the strains used in this previous study were analyzed to determine whether MALDI-TOF MS analysis can be used for intraspecific typing of vibrios.

Figure 3 shows that V. alginolyticus is still divided into two groups based on rpoB sequences. However, in contrast to the rpoB phylogenetic tree, MALDI-TOF MS cluster analysis did not clearly separate V. alginolyticus into two distinct groups (Figure 5). The MALDI-TOF MS cluster analysis revealed smaller sub-clusters, which were highly specific (94% to 100%) for the V. alginolyticus group 1 or V. alginolyticus group 2. Seibold et al. (2010) proposed a different approach to identify strains belonging to particular subspecies: analysis of the presence/absence of single “less conserved peaks” of spectra instead of all mainly “conserved peaks”. Therefore, the main spectra of V. alginolyticus sub-clusters were analyzed to determine whether V. alginolyticus group 1 and V. alginolyticus group 2 generated different mass spectrometric peaks. Four group-associated peaks were observed using this analysis, which can be assigned to these rpoB groups (Table 5, Figure 6): peak 5,642.2 +/- 1.7 m/z with an V. alginolyticus internal specificity of 99% and a sensitivity of 86% for V. alginolyticus group 2; peak 5,648.9 +/- 1.7 m/z for V. alginolyticus group 1 (93% internal specificity and 90% sensitivity); peak 11,278.6 +/- 3.4 m/z with an internal specificity of 99% and a sensitivity of 88% for V. alginolyticus group 2; peak 11,291.2 +/- 3.4 m/z for V. alginolyticus group 1 (93% internal specificity and 82% sensitivity). The V. alginolyticus group 1- associated peaks 5,648.9 and 11,291.2 were highly sensitive (85% and 77%, respectively) for Vibrio parahaemolyticus, and peak detection rates observed in the V. harveyi/campbellii group cluster are high for V. alginolyticus group 2-associated peaks (44% and 31%) and low for V. alginolyticus group 1-associated peaks (13% and 1%) (Table 5). The maximum peak detection rates in the remaining Vibrio spp. analyzed in this study is lower than 2%, indicating that these peaks are associated with the Vibrio spp. core clade, which consists of V.

- 43 - CHAPTER I alginolyticus, Vibrio parahaemolyticus and V. harveyi/campbellii. Furthermore, this analysis confirmed the close phylogenetic relationship between V. alginolyticus group 1 and V. parahaemolyticus, whereas V. alginolyticus group 2 appear more similar to V.

harveyi/campbellii.

85 90 95

80

65 70

40 45

60

75

50 55 100

96 % alg1_rpoB Cluster A (25 of 26)

100 % alg2_rpoB Cluster B (10 of 10)

Cluster C 100 % alg1_rpoB (4 of 4) 100 % alg1_rpoB Cluster D (13 of 13)

100 % alg2_rpoB Cluster E (10 of 10)

100 % alg1_rpoB Cluster F (12 of 12)

100 % alg2_rpoB Cluster G (16 of 16)

94 % alg1_rpoB Cluster H (45 of 48)

99 % alg2_rpoB Cluster I (67 of 68)

100 % alg2_rpoB Cluster J (2 of 2)

mainly V. alginolyticus rpoB group 1

mainly V. alginolyticus rpoB group 2

FIGURE 5: V. alginolyticus cluster analysis based on MALDI-TOF MS data analyzed by BioNumerics™. The scale bar represents similarity coefficients calculated with the Jaccard Index (Minimum height 1%, shift factor 1, constant tolerance 0.5, linear tolerance 400 ppm). Cluster analysis was performed with UPGMA. The amount of strains assigned as alginolyticus group 1 (alg1_rpoB) or alginolyticus group 2 (alg2_rpoB) by rpoB sequence analysis are given next

to each cluster.

- 44 - CHAPTER I

TABLE 5: MALDI-TOF MS peak detection rate of the V. alginolyticus rpoB group 1 and group 2 associated peaks at 5,648.9 +/- 1.7 m/z, 11,291.2 +/- 3.4 m/z, 5,642.2 +/- 1.7 m/z and 11,278.6 +/- 3.4 m/z. Peaks were detected with the peak matching tool of BioNumerics™. Peak detection rates are given for MALDI-TOF MS intraspecific clusters of V. alginolyticus (Figure 5) and for species or species groups.

V. alginolyticus rpoB group 1 V. alginolyticus rpoB group 2

associated peak detection rate associated peak detection rate

No. of 5,648.9 11,291.2 5,642.2 11,278.6 MALDI -TOF MS cluster strains +/- 1.7 m/z +/- 3.4 m/z +/- 1.7 m/z +/- 3.4 m/z

Cluster A (mainly alg1-rpoB) 26 81% 76% 0% 0%

Cluster B (mainly alg2-rpoB) 10 0% 0% 60% 60% Cluster C (mainly alg1-rpoB) 4 50% 0% 0% 0%

Cluster D (mainly alg1-rpoB) 13 92% 77% 0% 0% Cluster E (mainly alg2-rpoB) 10 0% 0% 100% 10% Cluster F (mainly alg1-rpoB) 12 92% 75% 0% 0%

Cluster G (mainly alg2-rpoB) 16 6% 0% 94% 94% Cluster H (mainly alg1-rpoB) 48 100% 100% 0% 0%

Cluster I (mainly alg2-rpoB) 68 5% 6% 92% 94% Cluster J (mainly alg2-rpoB) 2 0% 0% 50% 100%

No. of 5,648.9 +/- 1.7 11,291.2 +/- 3.4 5,642.2 +/- 1.7 11,278.6 +/- species / species groups strains m/z m/z m/z 3.4 m/z

V. alginolyticus rpoB group 1 100 90% 82% 1% 1%

V. alginolyticus rpoB group 2 107 7% 7% 86% 88% V. parahaemolyticus 306 85% 77% 5% < 1%

V. harveyi / campbellii group 44 13% 4% 44% 31% remaining V. spp 395 2% 1% < 1% < 1%

- 45 - CHAPTER I

Based on recent studies (Ilina et al. 2009; Hotta et al. 2011), we propose that these associated peaks represent the same protein. MALDI-TOF MS analysis enables their discrimination most likely because of single-nucleotide alterations in genes encoding abundant proteins in V. alginolyticus 1 and V. alginolyticus 2 strains. Furthermore, the associated peaks at 5,645 +/- 6 m/z and 11,285 +/- 10 m/z might represent the doubly and singly charged ions of this protein respectively. Hazen et al. (2009) observed a potential V. parahaemolyticus biomarker peak at almost the same position (11,294 m/z); suggesting, that this peak is consistent with the singly charged V. alginolyticus group 1-associated peak at 11,291.2 m/z.

These data are consistent with the hypothesis that proteome-based mass spectra reflect intraspecific groups identified using conventional DNA sequence-based methods. The absence/presence of group-specific peaks enabled the filtration of the mass spectrometric data for peaks associated with a species or an intraspecific group. The separation of V. alginolyticus into two phylogenetically distinct groups by MALDI-TOF MS analysis paves the way for future studies. These studies might enable the differentiation of Vibrio species with higher pathogenic potential from strains with lower pathogenic potential.

FIGURE 6: MALDI-TOF MS spectra of the strains VN-2653 (alginolyticus rpoB group 1; green) and VN-2677 (alginolyticus rpoB group 2; red). V. alginolyticus rpoB group 1 associated peaks are shown at 5,647.43 m/z (5,648.9 +/- 1.7 m/z) and 11,292.1 m/z (11,291.2 +/- 3.4 m/z). V. alginolyticus rpoB group 2 associated peaks are shown at 5,641.92 m/z (5,642.2 +/- 1.7 m/z) and 11,280.2 m/z (11,278.6 +/- 3.4 m/z).

- 46 - CHAPTER I

Database Download

The main contribution of this work is the MALDI-TOF MS VibrioBase database. All main spectra of VibrioBase are available as btmsp-file as supplement.

Acknowledgments

This work was supported by the Federal Ministry of Education and Research (German Research Platform for Zoonoses, VibrioNet, BMBF 669 grant 01KI1015A). The survey performed by the University Medical Center Schleswig-Holstein in 2011 was financially supported by the Ministry of Social Affairs, Health, Family and Equality of Land Schleswig- Holstein. The survey by the Governmental Institute of Public Health of Lower Saxony was financially supported by the research program KLIWAS of the German Federal Ministry of Transport, Building and Urban Development (BMVBS). The authors are grateful to Ciska Schets, Edda Bartelt, Christine Eichhorn, Annette Neumann, Craig Baker-Austin, Eckhard Strauch, Ralf Diekmann and Florian Gunzer for providing strains and Sarah Dehne, Hilke Döpke, Maike Berger, Simone Böer, Nicole Brennholt and Frederik Helmprobst for their valuable contribution to this study. We are also very grateful to the crew of the RV Heincke for assistance in sampling and to Bruker Daltonics Inc. as well as Applied Maths for their technical support.

- 47 - CHAPTER I

Supplementary

TABLE S-1: Minimum intraspecific rpoB sequence similarity values of rpoB species clusters and the maximum rpoB similarity of sequences between different species-clusters. The threshold for species cluster is 96.5%.

maximum rpoB similarity of sequences between different species-clusters

sequence similarity sequence

V. harveyi V.

V. fluvialis V.

V. cholerae V.

V. vulnificus V.

V. natriegens V.

V. navarrensis V.

L. anguillarum L.

V. alginolyticus V. aestuarianus V.

minimum intraspecific intraspecific minimum

V. diazotrophicus V.

rpoB rpoB

V. parahaemolyticus V.

V. parahaemolyticus 96.8 V. alginolyticus 97.9 94.9 V. natriegens 98.8 95.0 96.5 V. harveyi 94.7 92.1 93.6 95.1 V. vulnificus 98.2 84.5 86.3 87.9 85.0 V. navarrensis 97.3 81.7 83.2 84.7 81.9 88.4 L. anguillarum 98.9 78.4 79.9 81.4 78.5 81.1 78.0 V. aestuarianus 99.1 80.2 81.7 82.0 79.2 81.8 78.7 90.6 V. diazotrophicus 96.8 81.8 83.3 83.6 80.8 83.4 80.2 81.5 83.4 V. fluvialis 98.3 83.0 84.5 84.8 81.9 84.5 81.4 82.7 84.5 88.9 V. cholerae 97.6 78.3 79.8 80.1 77.2 79.8 76.7 78.0 79.8 84.2 85.8 V. mimicus 97.8 79.2 80.7 81.0 78.1 80.7 77.6 78.9 80.7 85.1 86.7 96.1

TABLE S-2: Mean intraspecific MALDI-TOF similarity values and the corresponding coefficients of variation for MALDI-TOF species clusters, and the mean intraspecific MALDI-TOF similarity values between MALDI- TOF species clusters.

mean MALDI-TOF similarity between species-clusters

TOF similarity TOF

-

. harveyi .

V

V. fluvialis V.

V. mimicus

V. cholerae V.

V. vulnificus V.

mean intraspecific intraspecific mean

V. navarrensis V.

L. anguillarum L.

V. alginolyticus V.

coefficient of variation of coefficient

V. diazotrophicus V.

MALDI

V. parahaemolyticus V.

V. parahaemolyticus 28.5 0.50 V. alginolyticus 35.8 0.42 17.4 V. harveyi 36.5 0.44 9.7 10.5 V. vulnificus 39.0 0.36 7.7 8.4 6.5 V. navarrensis 32.0 0.37 6.1 8.1 5.4 16.3 V. diazotrophicus 24.8 0.68 5.4 7.4 7.2 7.6 5.8 V. fluvialis 43.5 0.48 6.2 7.1 4.1 5.3 6.4 6.7 V. cholerae 42.6 0.32 6.5 4.9 5.0 6.4 5.1 6.8 5.4 V. mimicus 42.2 0.30 5.5 6.1 4.5 7.3 6.6 8.2 6.2 21.2 L. anguillarum 34.3 0.44 2.0 3.0 3.0 2.9 4.4 5.4 5.1 2.1 2.2 V. aestuarianus 26.2 0.63 4.3 3.4 4.2 4.9 6.2 4.6 5.8 2.6 2.1 11.1

- 48 - EXCURSUS I: Variability of MALDI-TOF MS measurements

EXCURSUS I: Variability of MALDI-TOF MS measurements

A test for reliability of Vibrio spp. spectra is shown in Figure 1: over a period of five days, three strains were inoculated daily on new medium, grown overnight and measured with MALDI-TOF MS. Based on the cluster analysis, the strain-related similarity varied between 54% and 61%. Thus, whole-cell MALDI-TOF MS reveals a large variation and provides no strain-specific profiles. Nevertheless, thresholds of 20% (highly probable identification) and 10% (probable identification) can be used to assign spectra to a certain species. To some extent, these thresholds correspond to the according Bruker BiotyperTM log-scores 2.3 and 2.0.

FIGURE 1: MALDI-TOF MS variability experiment. Three strains were measured separately from each other over period of five days. The scale bar represents similarity coefficients calculated with the Jaccard Index (minimum height 1%, shift factor 1, constant tolerance 0.5, linear tolerance 400 ppm). Cluster analysis was performed with UPGMA.

- 49 - EXCURSUS 2: Application of VibrioBase in surveillance programs

EXCURSUS 2: Application of VibrioBase in surveillance programs

During a two-weekly research cruise aboard the FS “Heincke”, environmental samples (water, zooplankton, phytoplankton) were concentrated on membrane filters, and subsequently transferred on CHROMagarTM Vibrio; without the alkaline peptone water enrichment step to allow direct quantifications of Vibrio spp.. Appearing colonies on CHROMAgarTM were analysed using MALDI-TOF MS in combination with the BiotyperTM database and VibrioBase. In total, 2,055 colonies were analysed: 16% revealed no reliable identification and 25% Vibrio species were detected (Figure 1). In conclusion, MALDI-TOF MS is an accurate tool for species identification fom marine environmental samples. However, CHROMAgarTM alone yield insufficient separation for the exclusion of non-Vibrio species. In consequence, no accurate quantification of Vibrio spp. was possible. Thus, enrichment in alkaline peptone water can be referred to as necessary for culture dependent surveillance programs.

Aeromonas Bacillus Aerococcus 3% Pseudomonas Halomonas Enterobacter Staphylococcus Micrococcus 4% Corynebacterium 11% L. anguillarium Acinetobacter V. alginolyticus 9% 5% 3%

V. diazotrophicus Vibrionaceae 2% Exiguobacterium 25% V. aestuarianus 36% 2% V. parahaemolyticus V. fluvialis V. cholerae 9% V. mimicus no reliable V. ordalii identification 16%

FIGURE 1: MALDI-TOF MS species identification results of 2,055 colonies from a surveillance program in the North and Baltic Seas.

- 50 - CHAPTER II

CHAPTER II

Evaluation of mass peak based for Vibrio spp. identification

René Erler1, Antje Wichels1, Ilka Hartmann1 and Gunnar Gerdts1

1 Alfred-Wegener-Institute Helmholtz Zentrum for Polar and Marine Research, Biologische Anstalt Helgoland, Kurpromenade 201, 27498 Helgoland, Germany

TABLE OF CONTENTS

Abstract ...... 52 Introduction ...... 53 Materials and Methods ...... 56 Results and Discussion ...... 60 Supplementary ...... 72

- 51 - CHAPTER II

Abstract

Bacteria of the genus Vibrio are common in coastal waters and, to some extent, cause severe infections. Thus, the analysis of environmental samples in respect to the occurrence of potentially pathogenic Vibrio spp. is of particular importance. The current bacterial identifications software tool BiotyperTM from Bruker was developed on the basis of whole- cell matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI- TOF MS) measurements and provides accurate identifications, particulary in combination with the reference database VibrioBase. We simplified this method using mass peaks with a high sensitivity towards certain Vibrio species and developed a software-independent species identification system based on the respective peak sensitivity values (SIBOPS). 194 spectra from mono-microbial Vibrio spp. cultures were analysed with this approach and 190 (97.9%) species identifications agreed with the results of the BiotyperTM software. A further check of mass peaks for exclusivity lead to the evaluation of one to six biomarkers for each species; allowing direct biomarker based species identifications (BIBSI) of bi-microbial Vibrio spp. cultures. Further surveillance programs are to be conducted to test these biomarkers for practical applicability.

- 52 - CHAPTER II

Introduction

Since the 1980ies matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) is used to generate peptide mass fingerprints for the identification of single proteins (Fenselau 1997). The key aspect of this method is the separation of ionized peptides during the passage through a flight tube in which the time-of-flight is equal to the mass-to-charge (m/z) ratio. Accumulations of equal m/z values are represented by peaks in the resulting mass spectrum and can be seen as mass spectrometric fingerprint for proteins. Each ionized molecule can be either singly or multiply charged and, therefore, is displayed at different m/z values. Advances in computer science and mass spectrometry technique allowed the analysis of multiple molecules in one sample and the generation of high-resolution mass spectra with numerous peaks. Comparable mass spectra from whole bacterial cells were initially generated by Claydon et al. (1996), Krishnamurthy et al. (1996) and Holland et al. (1996). These spectra contain peaks that can be assigned to abundant cellular proteins, particularly to those involved in the process of translation such as ribosomal proteins (Ryzhov et al. 2001). Differences between mass spectra from distinct species rely either on the absence/presence of proteins or on the variable amino acid compositions of one protein, that, in consequence, result in distinguishable molecular masses and m/z values (Holland et al. 1999; Lay 2001).

Bruker Daltonics Inc. used this whole-cell MALDI-TOF MS technique to develop the Biotyper™ species identification system that generates mass spectra from the biomass of bacterial colonies within minutes, followed by a calculation of matching scores with reference main spectra filed in a database (Maier et al. 2007; Sauer et al. 2008). Each main spectrum originates from a strain and contains the 70 most prominent mass peaks. The matching scores rely on alignments between peaks from main spectra and the m/z values as well as intensities of peaks from acquired spectra. A total agreement is reached by a score of 1000 which is expressed as log-score 3.0. According to the manufacturer, species classification results with matching scores higher than 2.0 are probable and those over 2.3 highly probable, whereas scores between 1.7 and 2.0 only allow accurate identifications at the genus level (Sauer et al. 2008). This identification system is typically used in combination with the provided BiotyperTM reference database from Bruker. Main spectra of the BiotyperTM database largely correspond to infective bacterial species. Therefore, the main field of application are medical laboratories where agar plates are incubated with clinical samples and

- 53 - CHAPTER II resulting spectra from bacterial colonies are checked for consistency with reference main spectra of the respective database (Sogawa et al. 2011). Nevertheless, MALDI-TOF MS is also applicable for environmental monitoring, particularly with a database that rather consists of main spectra from environmental isolates (Benagli et al. 2012; Koubek et al. 2012; Erler et al. 2014).

However, this current identification system has also some crucial limitations. For instance, acquired spectra of strains are aligned to each single main spectrum in the reference database. In consequence, the species assignation of the main spectrum with the highest score determines the identification result. Therefore, the BiotyperTM system provides no direct species identification and implies the possibility of probable identifications to multiple main spectra with different species assignations. These obstacles can be overcome by the use of mass spectrometric peaks that are sensitive or even exclusive towards species. The companies AnagnosTec and bioMérieux used this approach and implemented the concept of Super Spectra (Lartigue 2013). These spectra are filed in the SARAMISTM database that is part of the VITEK MS system. In contrast to main spectra of Bruker, Super Spectra contain mass spectrometric peaks that indicate bacterial families, genera and species. The respective identification results are based on the percentage of agreement with these peaks. Identifications based on mass spectrometric peaks were also introduced by the Andromas SAS company: “good identifications” are reached with a percentage of peak agreements of 67% and with a distance to the second species match of at least 10% (Bille et al. 2012). However, all these systems and the provided databases are only available for the respective customers and include inconsistent as well as complicated algorithms with empirical-based thresholds for species identification. Furthermore, these systems allow no reliable species identifications for bacteria in mixed cultures: alignments with the reference database lead either to less reliable matching scores or identification results with respect to the most abundant species in the mixture (Mahe et al. 2014). Thus, with the exception of clinical urine samples (Welker et al. 2011), isolation of bacterial colonies by plate streaking is still a necessary prerequisite for whole-cell MALDI-TOF MS measurements.

Direct species identifications are of particular importance for Vibrio spp. in mixed environmental samples. To some extent, species of this genus are able to infect seafood, fish, marine mammals and humans (Thompson et al. 2004). They are also common members of the marine bacterioplankton. In consequence, monitoring of environmental samples is crucial to

- 54 - CHAPTER II evaluate infection risks (Lindgren et al. 2012). Correlations between mass spectrometry peaks and Vibrio species were already observed by Hazen et al. (2009) and Dieckmann et al. (2010). These peaks were found on the basis of limited datasets; with only a few strains for each species. Variable ionization processes during MALDI-TOF MS measurements and intraspecific protein variations, however, necessitate the the use of a larger dataset that includes mass spectra from different species and strains. Regarding Vibrio spp., such data is provided by VibrioBase: a detailed MALDI-TOF MS database of main spectra from particularly 12 species (Erler et al. 2014). These species are either of clinical importance such as V. cholerae, V. parahaemolyticus and V. vulnificus or associated with zoonotic diseases e.g. V. aestuarianus and Listonella anguillarum (Thompson et al. 2004; Austin 2010). In context to this study, mass spectrometric data from VibrioBase provides two crucial facts for each peak: the sensitivity towards a certain species and the exclusivity against other species. In this study, VibrioBase peaks that exceed a certain sensitivity threshold for a species are referred to as “sensitive peaks”. These peaks were used to develop SIBOPS (species identification based on peak sensitivities): a simplified species identification system based on the comparison of peaks from acquired spectra with the m/z value of sensitive peaks. Peaks revealing also exclusivity are referred to as “exclusive peaks”. A main contribution of this study is the evaluation of SIBOPS and exclusive peaks using mono-microbial cultures with only one included species and bi-microbial cultures with two included species; in particular to obtain identification thresholds for SIBOPS and to test wheather exclusive peaks are still detectable in mixed cultures. Exclusive peaks, that passed this evaluation successfully, were assigned as “biomarkers” that can be used for biomarker based species identifications (BIBSI). Peak lists and Excel templates for both approaches are accessible for everyone and can be used, regardless of the respective MALDI-TOF MS equipment and software tools.

- 55 - CHAPTER II

Materials and Methods

Data source

The mass spectrometric data originates from main spectra that were acquired in course of the development of the database VibrioBase (Erler et al. 2014). According to the Bruker protocols, further main spectra were acquired for Vibrio spp. isolated from oysters at the island of Sylt. All main spectra were assigned in respect to the species identification results of the Biotyper™ software (version 3.1). 190 spectra achieved no reliable species identification but were assigned to the V. splendidus clade. 9 spectra from Shewanella spp. were included and defined as outgroup. In total, 1,267 main spectra were analysed. Each main spectrum contained a peak table with the m/z values of the 70 most prominent peaks. An overview for all main spectra is given in Table 1.

Peak table analysis

The workflow of the data analysis is shown in Figure 1. All main spectra were initially analyzed using BioNumerics™ (version 7.1, Applied Maths). Mass peaks were grouped using a linear m/z value tolerance of 400ppm and peak tables were generated with the respective mass peak intensities. Using Microsoft Excel, all peak tables were grouped and summarized in respect to their species assignation. Based on the absence/presence, peak sensitivity values were calculated for each species.

Species identification based on peak sensitivities (SIBOPS)

The SIBOPS approach relies on the comparison between m/z values of peaks from an acquired spectrum and m/z values of peaks that exceed the respective species threshold for “sensitive peaks” in VibrioBase (Figure 1). SIBOPS identification scores for each species reflect the ratio between the number of peaks consistent with sensitive peaks and the total number of sensitive peaks. To define a threshold for “sensitive peaks”, sensitivity values from VibrioBase were compared to the resulting SIBOPS score values; for the respective 1st species identification result and the distance to the 2nd species identification result. Therefore, the sensitivity value with the highest distance to the 2nd match reveals the best discriminatory power and was determined as threshold for “sensitive peaks”.

- 56 - CHAPTER II

VibrioBase data: 70 most prominent peaks for each main spectra / strain

species S sensitive peaks E = A N A B C D evaluation and P S peaks with high E I SIBOPS score sensitivity values to A X T X X species K I X threshold X V X T + E A X X X X determination B species P X with mono- identifications L E E A X X X based on peak K X microbial cultures sensitivity S (SIBOPS) SIBOPS score: 100 33 33 33 species species E E X A+B X A B C D A C A B C D C exclusive peaks P L P L E E X = A X U A U S S -- sensitive peaks with K I -- K I X -- exclusivity for one T X + V -- T + V A X E A X E species B B X -- L P -- L P E E E X E A A -- K -- K S S --- species A ------species A ------species B --- mono-microbial cultures bi-microbial cultures PEAK EVALUATION (sensitivity and exclusivity)

biomarker based Biomarkers: species identification (BIBSI)

FIGURE 1: Workflow scheme of the study. Mass spectrometric data from VibrioBase was screened for sensitive peaks. Species identification scores of the SIBOPS approach rely on the ratio between detected sensitive peaks in acquired spectra and the total number of sensitive peaks for the respective species. SIBOPS was evaluated, using 194 mono-microbial cultures. VibrioBase was further screened for exclusive peaks, that were evaluated using both, mono-microbial cultures and bi-microbial cultures. Peaks that remained highly sensitive and exclusive for a species can be used as biomarkers for the BIBSI approach.

SIBOPS was evaluated using spectra from mono-microbial cultures. This set of 194 Vibrio spp. strains was not part of the analysis of VibrioBase main spectra and originate from surveillance programs in the North and Baltic Seas as well as from clinical samples. All 194 acquired spectra were initially aligned with with VibrioBase to identify the species. The according peak tables were generated using flexAnalysis (version 3.3, Bruker Daltonics). Only mass peaks exceeding a peak-to-noise-ratio of 2.0 and a minimum intensity of 0.1% were displayed in the peak table. Using a linear tolerance of 1000ppm, these peaks were compared to the sensitive peaks of SIBOPS. Initially, the acquired spectra were assigned according to the species with the highest SIBOPS score. Thresholds for reliable SIBOBS

- 57 - CHAPTER II species identifications were determined based on the score value with the most single-species results and the distance towards the 2nd matching result.

Biomarker based species identifications (BIBSI)

All sensitive peaks from VibrioBase were tested for exclusivity (Figure 1). To match the criteria for exclusivity, peaks had to fulfil two requirements: sensitivity values above the respective sensitive peak threshold for a single species and low-to-moderate sensitivity values for maximal three other species. These exclusive peaks were evaluated using mono-microbial cultures (as stated for SIBOPS) and bi-microbial cultures. To obtain bi-microbial cultures, 10 Vibrio species were initially grown as mono-microbial culture in alkaline peptone water; for 24h at 37°C. Afterwards, two cultures were equally mixed to obtain 45 bi-microbial cultures. The acquired spectra from mono-microbial and bi-microbial cultures were screened for the exclusive peaks (as stated for the SIBOPS evaluation). An evaluation score for peaks was calculated using the median for sensitivity and exclusivity, with the following equation:

where E is the evaluation score for a certain peak, s is the sensitivity based on the ratio between the presence of the peak (p) and the total number of acquired spectra for the respective species (n1), and e is the exclusivity based on the ratio between the absence of the peak (a) and the total number of the remaining spectra (n2). Peaks passed this evaluation by exceeding an arbitrary evaluation score threshold of 75% for both, mono-microbial and bi- microbial cultures. Those peaks were assigned as biomarkers and, in consequence, can be used for BIBSI; either singly or in combination.

Protein assignment of biomarkers

Particularly potentially pathogenic Vibrio species were sequenced completely. The nucleotide sequence of genes and the respective amino acid sequences are filed in available databases. Thus, detected m/z values and molecular masses of biomarkers were assigned to proteins, according to a previous mass peak list of Dieckmann et al. (2010) and the database on the ExPASy proteomics server (http://web.expasy.org/tagident/) (Gasteiger et al. 2003).

- 58 - CHAPTER II

Results and Discussion

Peak table analysis

All main spectra from the 12 Vibrio species and from Listonella anguillarum as well as Shewanella spp. generated 798 different peaks. Putative thresholds were compared to the respective SIBOPS identification results to determine sensitive peaks (Figure 2). The difference between the 1st and the 2nd SIBOPS match decreases for thresholds higher than 80%. Therefore, peaks that reveal a minimum sensitivity of 80% were defined as sensitive peaks and used for SIBOPS. Based on the VibrioBase data, this threshold reveals a respective SIBOPS score of 88 to 98; with a difference to the 2nd result of 23 to 44. The cluster analysis in Figure 3 comprises all sensitive peaks and reveals that closely related species share sensitive peaks. Particularly V. alginolyticus and V. parahaemolyticus, V. vulnificus and Vibrio navarrensis, Listonella anguillarum and V. aestuarianus as well as V. cholerae and V. mimicus are grouped together. Furthermore, this cluster analysis reflects the phylogenetical relationships between these Vibrio species as revealed by approved DNA-sequence based methods (Thompson et al. 2005).

100 90

80 70

score 60 50

SIBOPS SIBOPS 40

30 20

10 0 0 10 20 30 40 50 60 70 80 90 100 VibrioBase sensitivity threshold (%) for sensitive peaks

1st SIBOPS identification result difference between the 1st and the 2nd SIBOPS identification result determined sensitivity threshold for sensitive peaks in SIBOPS

FIGURE 2: SIPOBS scores were calculated based on peaks exceeding a specific sensitivity value in VibrioBase. The curves represent the 1st species identification results (blue) and the distance to the 2nd identification results (black). The broken line represents the chosen threshold for sensitive peaks in SIBOPS.

- 59 - CHAPTER II

peak sensitivity

.

spp 20% - 50%

clade

pacinii

harveyi

fluvialis

mimicus species: cholerae

vulnificus

V. V. 50% - 80%

V. V.

V. V.

V. V.

V. V.

navarrensis

anguillarum

V. V.

aestuarianus alginolyticus

V. V.

diazotrophicus L. > 80%

V. V. V.

Shewanelle

splendidus

V. V.

parahaemolyticus

V. V.

V. V.

peaks: V. alginolyticus V. parahaemolyticus V. harveyi

V. navarrensis V. vulnificus

V. pacinii

Shewanella spp.

V. fluvialis

L. anguillarum V. aestuarianus

V. splendidus clade

V. diazotrophicus

V. mimicus V. cholerae

Vibrio

FIGURE 3: All peaks with a minimum sensitivity value for at least one species were clustered using the software tools Cluster (version 3.0) and treeview (version 1.1.6) (average distance similarity). The colours reflect sensitivity values: light green - low sensitivity (20-50%), light blue - moderate sensitivity (50-80%), dark blue - high sensitivity (>80%; SIBOPS sensitive peaks). Peak clusters that are dominated by certain species are listed on the right side.

- 60 - CHAPTER II

For each species, 12 to 38 sensitive peaks were observed and used for SIBOPS analysis (Table 1, Table S-1). Based on the mass peak table analysis, 139 sensitive peaks fulfilled the requirements for exclusivity; with a range of 4 to 19 exclusive peaks for each species (Table 1, Table S-1). TABLE 1: Main spectra, observed sensitive and exclusive peaks for each species, and available mono-microbial cultures for the evaluation of SIBOPS and exclusive peaks.

mass spectrometric data available mono- microbial cultures for the evaluation sensitive main exclusive of SIBOPS and peaks spectra peaks exclusive peaks (SIBOPS)

V. parahaemolyticus 310 30 12 30 V. alginolyticus 282 26 4 30 V. harveyi 57 12 3 V. fluvialis 30 31 14 31 V. vulnificus 246 25 9 45 V. navarrenis 10 21 10 V. cholerae 63 38 19 7 V. mimicus 28 34 10 V. pacinii 7 27 6 V. diazotrophicus 12 18 9 12 V. splendidus clade 190 14 8 V. aestuarianus 7 25 10 11 L. anguillarum 18 13 7 28 Shewanella spp. 7 22 18

Total 1267 336 139 0 194

SIBOBS evaluation using mono-microbial samples

In total, 190 of 194 (97.9%) spectra were identified correctly using the SIBOPS analysis; only 2 V. parahaemolyticus and 2 V. alginolyticus spectra were misidentified (Table 2). The minimum distance to the 2nd match is 11.5 for V. parahaemolyticus (to V. alginolyticus) and 20.6 for V. alginolyticus (to V. harveyi) (Table 2). A significantly better discrimination was achieved for V. vulnificus and V. navarrensis: all V. vulnificus strains were identified correctly, with a median distance of 26.9 to V. navarrensis (Table 2). However, since spectra of V. navarrensis were not available for this evaluation, there is a need to further analysis to check for the discriminatory power, vice versa.

- 61 - CHAPTER II

TABLE 2: Median SIBOPS score values, correct identifications, best 2nd match results and the respective median differences to the 1st match for all aquired spectra from mono-microbial cultures, and the expected SIBOPS score difference between species based on the mains spectromtric data from VibrioBase.

expected SIBOPS score correct score score best 2nd match difference identifications difference median median (VibrioBase)

26 x harveyi 20.6 41.4 V. alginolyticus 80.8 28/30a 1 x parahaemolyticus 32.9 37.9 1 x diazotrophicus 40.3 62.0

24 x alginolyticus 11.5 22.5 2 x harveyi 28.2 36.6 V. parahaemolyticus 81.0 28/30b 1 x vulnificus 31.9 61.6 1 x diazotrophicus 34.7 62.8

V. vulnificus 93.6 45/45 45 x navarrensis 26.9 30.3

V. cholerae 85.3 7/7 7 x mimicus 11.4 26.3

V. diazotrophicus 80.6 12/12 9 x vulnificus 26.9 52.4 1 x pacinii 38.0 66.6 1 x harveyi 38.9 60.8 1 x anguillarum 43.4 67.0

V. aestuarianus 88.4 11/11 11 x anguillarum 30.3 50.8

L. anguillarum 91.8 28/28 28 x aestuarianus 30.5 57.1

15 x anguillarum 39.3 58.8 3 x alginolyticus 43.0 64.1 6 x aestuarianus 44.3 68.7 V. fluvialis 76.8 31/31 5 x diazotrophicus 44.9 63.1 1 x vulnificus 48.9 73.6 1 x mimicus 55.2 77.3

a misidentified as V. harveyi (1x) and V. parahaemolyticus (1x) b misidentified as V. alginolyticus (2x)

All seven V. cholerae strains were identified correctly but revealed the lowest difference values to the 2nd match: 11.4 towards V. mimicus (Table 2). Three additionally acquired spectra from V. mimicus strains revealed a median SIBOPS score of 86.3 with a distance to V. cholerae of 23.1 (data not shown); suggesting that the separation of these both species has to be evaluated on the basis of even more spectra. The discriminatory power is high for the species V. aestuarianus and L. anguillarum: these species were identified correctly with a distance to the 2nd match of 30.3 and 30.5 respectively (Table 2). A clear separation towards

- 62 - CHAPTER II other Vibrio species was furthermore achieved for V. diazotrophicus (26.9 median distance) and V. fluvialis (39.3 median distance) (Table 2). However, latter species is closely related to V. furnissii (Brenner et al. 1983). Only one type strain of this species was available in this study, which achieved a SIBOPS score of 35.3 to V. fluvialis (data not shown). Thus, spectra from additional V. furnissii strains are to be tested for their discriminability towards V. fluvialis. For all Vibrio spp., the SIBOPS scores of the evaluation step are significantly lower than the expected scores based on the VibrioBase data; in particular the distance values between species (Table 2). This decreased discriminatory power is explained with the higher linear tolerance for peak detection in SIBOPS (1000ppm) compared to the linear tolerance for the initial main spectra analysis in BioNumerics (400ppm); leading to a simultanious detection of peaks with similar m/z values. In the course of the SIBOPS development, this increased linear tolerance in SIBOPS was found to be necessary to balance MALDI-TOF MS measurement variations. Nevertheless, the tolerance value might be adapted for single sensitive peaks to achieve a even higher discriminatory power between species.

100

90 80

70 60

score 50

40

SIBOPS SIBOPS 30 PAR – V. parahaemolyticus DIA – V. diazotrophicus 20 ALG – V. alginolyticus 1st – 1st identification result AES – V. aestuarianus 10 VUL – V. vulnificus 2nd – 2nd identification result ANG – L. anguillarum CHO – V. cholerae FLU – V. fluvialis 0 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd 1st 2nd PAR ALG VUL CHO DIA AES ANG FLU FIGURE 4: SIBOPS score values of aquired spectra from mono-microbial cultures. All boxplots refelect the medians, the interquartiles (25%/75%) and the standard variations st nd for the respective species identification result (orange, 1 ) and the best 2 matching result (grey).

In general, SIBOPS provided accurate species identification results (Figure 4). All species revealed median SIBOPS score values above 76 and more than 85% of all sensitive peaks were detected in the respective mono-microbial cultures (data not shown). However, these identification results only rely on the respective 1st match of the SIBOBS results and neclect the score of the 2nd match. Thus, the threshold of 77.5 was determined for reliable SIBOPS results; providing 80% single identification results and only 4% multiple species results (Figure 5). Another option is the use of the 67% identification threshold in combination with a

- 63 - CHAPTER II distance threshold to the 2nd species result of at least 10.0, as applied in the Andromas software for “good identifications” (Bille et al. 2012): 82% of the SIBOPS exceeded both requirements whereas 6% did not exceed the identification threshold and 10% not the distance threshold (Figure5, Table 3). In conclusion, either the 77.5 threshold or the Andromas thresholds can be used to assign “reliable species identification” whereas identifications based on the first match can be regarded as to “secure identifications”.

100

90 77.5% 80

(%)

70

results 60 no identification result single identificaton result 50 multiple identification results 40 threshold with the most identification single identification results

of 30

ratio 20

10

0 0 10 20 30 40 50 60 70 80 90 100 SIBOPS species identification threshold

FIGURE 5: SIBOPS scores thresholds and the respective species identification results. The thresholds with the most single species identifications ( 77.5%) is marked with a dotten line.

Both potential SIBOPS thresholds were compared with the VibrioBase / Biotyper identification thresholds for reliable (2.3) and secure species identification (2.0) (Table 3); showing an agreement of 72% between all reliable SIBOPS identifications and the reliable VibrioBase identifications. Thus, SIBOPS results are linked to the Bruker identification system and the respective matching scores. Furthermore, the number of detected peaks do not correlate with the quality of species identification results; no significant increasing reliable identifications were found for SIBOPS or the Bruker identification system (Figure 6). Nevertheless, a minimum number of at least 50 peaks is suggested to reveal accurate identification results with SIBOPS.

- 64 - CHAPTER II

TABLE 3: Agreements between SIBOPS and the Biotyper/VibrioBase identifications for reliable identifications (SIBOPS score > 77.5; SIBOPS score > 67 and distance > 10; Biotyper score > 2.3) and secure identifications (SIBOPS score < 77.5; SIBOPS score < 67 or distance < 10; Biotyper score between 2.0 and 2.3.

Biotyper /VibrioBase identifications score score score

> 2.3 2.0 - 2.3 < 2.0

86% 13% 1% SIBOPS identifications total (n = 167 (n = 25) (n = 2) 80% 72% 9% score > 77.5 0% (n =156) (n =139) (n =17) 18% 12% 4% 1% score < 77.5 (n =34) (n =24) (n =8) (n =2)

score > 67 and 82% 72% 11% 0% distance > 10 (n =160) (n =139) (n =21) score < 67 or 16% 12% 2% 1%

distance < 10 (n =30) (n =24) (n =4) (n =2) 2% 2% misidentifications 0% 0% (n =4) (n =4)

100 SIBOPS threshold (77.5) SIBOPS threshold (67 + 10 distance) 95 Biotyper / VibrioBase threshold (2.3)

(%) 90

85

identifications 80

reliable 75

70 40 50 60 70 80 90 100 110 120 number of minimum peaks in mass spectra

FIGURE 6: Number of minimum peaks in acquired mass spectra and the respective ratio between reliable identification results and the total number of identification results; for the respective two SIBOPS thresholds (blue - SIBOPS score > 77.5; orange - SIBOPS score > 67 and distance to the 2nd match > 10) and the Biotyper threshold (black - score > 2.3).

- 65 - CHAPTER II

BIBSI evaluation using mono- and bi-microbial cultures

74 exclusive peaks were evaluated using the 194 acquired spectra from mono-microbial cultures and the 45 spectra from bi-microbial cultures: 22 peakes passed this evaluation and were assigned as biomarkers (Figure 7B), 19 peaks failed the mono-microbial evaluation, 11 peaks failed the bi-microbial evaluation and 22 peaks failed both evaluations (Figure7A). B 100 A 90

80 75 70

score

60

50

evaluation 40

30

microbial

-

bi 20

10

0 0 10 20 30 40 50 60 70 75 80 90 100 mono-microbial evaluation score

V. parahaemolyticus V. alginolyticus V. fluvialis V. vulnificus L. anguillarum V. cholerae V. diazotrophicus

100 4,974.3 10,354.0 4,036.9 B 7,510.2 8,068.4 8,054.2 95 3,755.9 9,944.7 4,966.8 10,873.1 score 6,497.0 11,073.4 9,875.7 90 5,538.8

evaluation 10,310.5 7,968.7 85 9,033.3 5,496.0 5,438.7

microbial

- bi 80 10,543.7 4,958.0

4,939.1 75 75 80 85 90 95 100 mono-microbial evaluation score FIGURE 7: Mono-microbial and bi-microbial evaluation scores for exclusive peaks of Vibrio spp.. A) all exclusive peaks. B) exclusive that exceed the evaluation score threshold of 75.

- 66 - CHAPTER II

For V. parahaemolyticus, six biomarkers were found (Table 4, Figure 8). Four of them were already observed by Dieckmann et al. (2010): integration host factor subunit alpha (5,538.2 and 11,073.4; doubly- and singly-charged), 30S ribosomal protein S15 (4,966.8; doubly- charged) and 50S ribosomal protein L31 (8,054.2; singly-charged). The remaining two new biomarkers were found at the m/z values 3,755.9 and 7,510.2; representing the doubly- and singly-charged ion of a protein that is not filed in the ExPASy database. All these five biomarkers were detected at the same time in 22 of 30 mono-microbial cultures (73%) and in 8 of 9 (89%) bi-microbial cultures of V. parahaemolyticus.

Based on the study of Dieckmann et al. (2010), the presence of six biomarkers was confirmed; three for V. vulnificus and three for V. cholerae (Table 4, Figure 8). Biomarkers for V. vulnificus were assigned to the following proteins: the 30S ribosomal protein S15 as doubly-charged ion at 4,958.0, the 30S ribosomal protein L30 at 6,497.0 (singly-charged) and the 30S ribosomal protein S19 at 10,310.5 (singly-charged). These three peaks were observed in 39 of 45 (87%) mono-microbial cultures and in 7 of 9 (78%) bi-microbial cultures. Biomarkers for V. cholerae were assigned to the integration host factor subunit alpha at 5,596.0 (doubly-charged), the 50S ribosomal protein L31 at 7,968.7 (singly-charged) and the integration host factor subunit beta at 10,543.7 (singly-charged). 6 of 9 bi-microbial cultures (67%) and all V. cholerae spectra from mono-microbial cultures (7 of 7) revealed these three peaks. Nevertheless, 2 of 8 spectra from bi-microbial cultures that did not contain V. cholerae but V. mimicus also revealed these peaks too.

For V. alginolyticus, two biomarkers were evaluated according to Dieckmann et al. (2010): the singly- and the doubly-charged 50S ribosomal protein L31 protein at 4,036.9 and 8,068.4 (Table 4, Figure 8). Both biomarkers were detected at the same time in 24 of 30 mono- microbial cultures (47%) and in 9 of 9 (100%) bi-microbial cultures of V. alginolyticus.

Five peaks were assigned as biomarkers for V. fluvialis (Table 4, Figure 8). No V. fluvialis strain was sequenced completely so far. Thus, no data is provided in the ExPASy database and proteins can only be assumed: glutaredoxin-1 at 4,939.1 (doubly-charged) and 9,875.7 (singly charged) as well as the integration host factor subunit alpha at 5,438.7 (doubly- charged) and at 10,873.1 (singly-charged) whereas the peak at 9,033.3 revealed no protein assignment. Both biomarkers are present in 29 of 31 (94%) spectra from mono-microbial cultures and 7 of 9 (77%) spectra from bi-microbial cultures. Two new biomarkers were

- 67 - CHAPTER II found for L. anguillarum and assigned to the 30S ribosomal protein S15, represented by peaks at 4,974.3 (doubly-charged) and 9,944.7 (singly charged). Both biomarkers are present in 27 of 28 (96%) spectra of mono-microbial cultures and in all spectra from the bi-microbial cultures. For V. diazotrophicus, only one biomarker was found at 10,354.0. As already stated for V. fluvialis, the respective protein can only be assumed as 50S ribosomal protein L25 or S19.

Table 4: Biomarkers for Vibrio spp., respective protein assignments, charge (doubly and singly) and reference.

species m/z value protein charge reference

V. parahaemolyticus 3,755.9 2x V. parahaemolyticus 4,966.8 S15 2x d V. parahaemolyticus 5,538.8 IHF alpha 2x d V. parahaemolyticus 7,510.2 1x V. parahaemolyticus 8,054.2 L31 1x d V. parahaemolyticus 11,073.4 IHF alpha 1x d V. alginolyticus 4,036.9 L31 2x d V. alginolyticus 8,068.4 L31 1x d

V. vulnificus 4,958.0 S15 2x d

V. vulnificus 6,497.0 L30 1x d

V. vulnificus 10,310.5 S19 1x d

V. cholerae 5,496.0 IHF alpha 2x d

V. cholerae 7,968.7 L31 1x d V. cholerae 10,543.7 IHF beta 1x d

V. fluvialis 4,939.1 glutaredoxin-1 2x e V. fluvialis 5,438.7 IHF alpha 1x e V. fluvialis 9,033.3 V. fluvialis 9,875.7 glutaredoxin-1 1x e V. fluvialis 10,873.1 IHF alpha 1x e L. anguillarum 4,974.3 S15 2x e L. anguillarum 9,944.7 S15 1x e V. diazotrophicus 10,354.0 S19 1x e

S15/S19/L30/L31 – ribosomal proteins d – Dieckmann et al. (2010) IHF – integration host factor e – Expasy database

- 68 - CHAPTER II

FLU 4,939.1 ALG 4,949.8 S15 (doubly-charged) VUL 4,958.0 CHO 4,998.2 ANG 4,974.3 MIM PAR 4,990.8 4,966.8

L30 (singly-charged)

ALG 6,468.7 ANG VUL 6,478.4 6,497.0

DIA L31 (singly-charged) 7,982.6 CHO ALG 8,067.2 7,968.7

MIM PAR 7,952.4 8,054.2

VUL DIA ALG FLU 9,914.4 9,923.1 S15 9,899.4 9,875.7 (singly-charged)

ANG 9,944.7

FIGURE 8: Mass spectra of selected species with exclusive pekas and biomarkers (underlined). The protein with respect to the mass values is displayed in each right upper corner (S15, L30 and L31). The following species are displayed: V. parahaemolyticus (PAR, yellow), V. vulnificus (VUL, light green), V. cholerae (CHO, red), V. alginolyticus (ALG, light blue), V. fluvialis (FLU, brown), L. anguillarum (ANG, dark violet), V. diazotrophicus (DIA, dark blue), V. mimicus (MIM, dark green) and V. navarrensis (light violet).

- 69 - CHAPTER II

Genus and group-specific biomarkers

All Vibrio species revealed high sensitivity values for the peak at 4,278.3. 1,236 of 1,260 (98%) main spectra of VibrioBase and 191 of 194 (98%) acquired spectra (mono-microbial cultures) displayed this peak (data not shown). Furthermore, this peak was found in all spectra from mixed samples. It represents the 50S ribosomal protein L36. The screening of additional 68 spectra from non-Vibrio species yielded only 3 peak detections (4%). Thus, this peak can be regarded to as biomarker for the genus Vibrio.

Based on the main spectra analysis of VibrioBase, two peaks were observed for a separation of Vibrio spp. into two groups: potentially human pathogenic species (group I, e.g. V. parahaemolyticus, V. vulnificus and V. cholerae) and species that are not associated with human illness (group II, e.g. L. anguillarum, V. aestuarianus and the V. splendidus clade). Both peaks represent the doubly charged (4,178.6) and singly charged ion (8,354.6) of the 30S ribosomal protein S21; they are abundant in main spectra of group I (94%/95%) and almost absent in spectra of group II (10%/1%). However, the evaluation by mono-microbial cultures revealed significantly lower sensitivity values (73%/80%). Furthermore, less than 50% of all spectra from bi-microbial samples revealed these peaks. Thus, these two peaks failed the evaluation and cannot be assigned as biomarkers for the respective groups.

Intraspecific biomarkers

Ideally, each mass peak is either absent in mass spectra of species or sensitive towards a certain species. However, various peaks reveal low-to-moderate sensitivity values. These peak sensitivity values are, to some to extent, not the consequence of MALDI-TOF MS measurement discrepancies. They also reflect intraspecific variations. For instance, the peaks with the m/z value of 5,647.8 and 11,292.0 represent the doubly and singly charged 30S ribosomal protein S14 (RS14) protein (Dieckmann et al. 2010). Both peaks reveal a high sensitivity values towards V. parahaemolyticus (85%, 77%) and towards the intraspecific group I of V. alginolyticus (90%, 82%), but are almost absent in spectra from V. alginolyticus group II strains (7%, 7%) (Oberbeckmann et al. 2011). These variations of both intraspecific groups leads to moderate similarity values of 54% and 52% for all V. alginolyticus main spectra. Another example is the V. vulnificus 50S ribosomal protein L31 (RL31) (Dieckmann et al. 2010; Bier et al. 2013). We observed three variations of this protein at different m/z

- 70 - CHAPTER II values: at 8,068.4, at 8,080.5 and at 8,087.7 with respective sensitivity values of 18%, 39% and 42%. At least the first two variations are caused by different amino acids at position 31, aspartic acid and glutamic acid. Due to these low-to-moderate sensitivity values, these peaks cannot be used for the identification of species. Nevertheless, further studies are to be conducted to check whether these peaks allow identifications below the species level.

Conclusions

This study shows that mass spectrometry peaks are efficient tools for the identification and characterization of Vibrio spp.. Using sensitive peaks by SIBOPS allows accurate species identifications independently from the respective MALDI-TOF MS system and the according software equipment. Furthermore, SIBOPS enables the unrestricted exchange and comparison of mass spectrometric data. Therefore, a web-based database are to be developed such as GenBank and Expasy to share whole-cell MALDI-TOF MS data.

Biomarkers were found for certain species that might be used for the direct identification of Vibrio spp. in environmental samples, either singly or in combination. One prominent biomarker was found for the whole genus Vibrio. We anticipate the use of these peaks in further surveillance programs. However, the intensity of mass peaks is negatively correlated with the number of different species included in mixed samples. Thus, advanced MALDI- TOF MS technique, however, is needed to allow an accurate application of the BIBSI approach.

Acknowledgements

This work was supported by the Federal Ministry of Education and Research (German Research Platform for Zoonoses, VibrioNet, BMBF 669 grant 01KI1015A). Further isolates, not included in VibrioBase, were kindly provided by Matthias Wegener from the Alfred- Wegener-Institute.

- 71 - CHAPTER II

Supplementary

Table S-1: Sensitive peaks (s), exclusive peaks (e) and biomarkers (b) for the species V. parahaemolyiticus (PAR), V. alginolyticus (ALG), V. harveyi (HAR), V. fluvialis (FLU), V. vulnificus (VUL), V. navarrensis (NAV), V. mimicus (MIM), V. cholerae (CHO), V. pacinii (PAC), V. diazotrophicus (DIA), Listonella anguillarum (ANG), V. aestuarianus (AES), the V. splendidus clade (SPL) and Shewanella spp.

PAR ALG HAR FLU VUL NAV MIM CHO PAC DIA ANG AES SPL SHE

3043.07 s 3049.9 s 3061.3 e 3082.77 s s s s s 3089.3 s s 3094.42 s 3129.13 e 3135.87 s s s 3145.76 s s s 3164.55 e 3200.36 s s s 3224.03 s s 3229.43 s 3233.16 s s 3286.71 s 3422.21 s s 3428.15 s s 3437.93 s 3458.11 s 3566.75 s s 3568.0 e 3571.94 s s 3578.93 s s s 3589.51 s s 3595.3 s s 3601.54 s 3645.77 s 3661.22 s s 3668.62 e 3755.91 b 3827.02 s 3963.19 s 3969.35 e 3975.71 s 3983.9 e 4022.56 s 4036.86 b 4107.88 s 4178.6 s s s s s s s s s 4186.21 s 4278.34 s s s s s s s s s s s s s 4304.96 s s 4367.32 e 4379.09 s s 4406.82 s - 72 - CHAPTER II

Table S-1 (continued)

PAR ALG HAR FLU VUL NAV MIM CHO PAC DIA ANG AES SPL SHE

4425.4 s s 4443.57 e 4451.9 s s s 4484.0 s 4516.2 e 4532.08 s s s 4537.81 s s 4560.03 e 4570.92 s 4633.25 s 4639.98 e 4682.77 s 4687.73 s s s 4693.51 s 4698.6 s s 4704.63 s s 4716.3 s 4723.65 s s 4732.51 s 4746.0 e 4756.46 e 4809.91 s 4851.57 e 4939.13 b 4953.74 s 4958.0 b 4960.16 s s s 4966.81 b 4974.31 b 4990.81 s s 4998.2 e 5051.62 s 5084.22 s 5097.98 s 5124.45 s e 5137.34 s s 5147.98 e 5152.85 s 5158.78 s s 5181.4 s s s 5186.15 s 5207.82 s s s s 5253.29 s 5268.09 e 5276.7 e 5297.12 s 5368.1 s 5438.73 b 5461.94 s 5496.08 s b 5504.29 e

- 73 - CHAPTER II

Table S-1 (continued)

PAR ALG HAR FLU VUL NAV MIM CHO PAC DIA ANG AES SPL SHE

5538.83 b 5581.35 b 5630.97 s 5647.84 e 5677.2 s 5691.14 s s 5741.15 s s 6087.33 s 6103.93 s 6115.07 s 6167.2 s s s s s 6170.0 e 6179.91 e 6194.37 s s 6222.52 s 6258.91 e 6273.91 s 6294.63 s 6298.0 e 6401.72 s 6447.42 s s 6454.62 s s s s s s s 6468.66 s s 6478.39 s s 6494.55 s 6497.0 b 6721.52 e 6728.87 s s 6754.22 e 6782.79 e 6790.18 s s s 6814.72 e 7119.26 s 7134.45 s s 7147.7 s s 7160.69 s 7163.0 e 7179.63 s 7186.62 e 7192.6 s 7204.67 s 7230.99 e 7292.94 s 7432.75 s 7510.15 b 7925.59 s 7937.29 e 7952.38 s 7968.68 b 8014.36 s 8043.77 s

- 74 - CHAPTER II

Table S-1 (continued)

PAR ALG HAR FLU VUL NAV MIM CHO PAC DIA ANG AES SPL SHE

8054.21 b 8068.43 b 8128.62 s 8215.11 s 8325.72 s s 8354.62 s s s s s s s s s 8370.25 s 8736.62 s s s 8753.44 s s 8810.71 s 8853.37 s s 8884.51 e 8900.64 s 8908.49 e 8965.5 s 8976.96 s 9033.34 b 9060.59 s s s s 9070.8 s s 9079.97 s 9135.39 s 9261.31 s 9275.31 e 9325.67 e 9333.44 e 9362.96 s 9372.42 s s s 9381.78 s 9392.87 e s s 9406.21 s s 9427.17 s 9437.05 s 9449.96 s s s 9456.25 e 9463.46 e 9478.09 s 9491.38 e 9506.81 e 9544.04 s 9559.94 e 9615.07 s 9636.22 s s 9875.67 b 9899.42 s s 9914.42 s s 9930.66 e 9944.74 b 9976.27 s 9990.62 e 10061.4 s 10235.6 e

- 75 - CHAPTER II

Table S-1 (continued)

PAR ALG HAR FLU VUL NAV MIM CHO PAC DIA ANG AES SPL SHE

10264.3 s s s s 10283.6 s s 10293.9 s s s 10310.5 b 10354.0 b 10368.0 s 10543.7 b 10589.9 e 10816.2 s 10873.1 b 10911.7 s 10987.4 s s 11031.4 e 11073.4 b 11126.2 s 11161.9 s 11259.7 s 11262.2 e 11349.6 s 11377.3 s 11477.1 s s

- 76 - CHAPTER III

CHAPTER III

Biogeographical mapping of V. cholerae, V. parahaemolyticus and V. vulnificus

populations in the North and Baltic Seas using ERIC-PCR genotyping

René Erler1*, Antje Wichels1, Leon Dlugosch2 and Gunnar Gerdts1

1 Alfred-Wegener-Institute, Helmholtz Zentrum for Polar and Marine Research,

Biologische Anstalt Helgoland, Kurpromenade 201, 27498 Helgoland, Germany

2 Institute for Chemistry and Biology of the Marine Environment, Carl-von-Ossietzky-

Straße 9-11, 26111 Oldenburg, Germany

TABLE OF CONTENTS

Abstract ……………………….… 78 Introduction ……………………... 79 Materials and Methods …………. 84 Results …………………………… 86 Discussion ……………………..… 95 Supplementary ………………..… 99

- 77 - CHAPTER III

Abstract

Vibrio spp. are common members of the bacterioplankton occurring in low numbers in the marine environment. It is assumed, that global warming-driven changes will lead to an increase of potentially pathogenic Vibrio spp. and thereby increased incidences of Vibrio- associated infections. Hence, potentially pathogenic Vibrio spp. such as V. cholerae, V. parahaemolyticus and V. vulnificus pose an increasingly serious threat to the society, even in temperate regions. However, dispersal mechanisms and population structures of these species are poorly understood in the northern part of Europe. To describe current population structure and patterns of Vibrio populations in the North and Baltic Seas, 472 strains from distinct sampling sites were analyzed by enterobacterial repetitive intergenic consensus (ERIC) genotyping. Our results revealed the presence of rather ubiquitous than locally restricted genotypes; without any general patterns for all three Vibrio species. For V. cholerae populations, no spatial differences were observed, but higher genotype diversity was found within assemblages of the Baltic Sea. In respect to their genotype distribution, V. parahaemolyticus populations are distributed into geographically separated regions. For the distribution of V. vulnificus, two different Baltic Sea assemblages were observed; a distinct group is located in the western Baltic Sea and another group is located around the islands of Rügen and Usedom sharing a high similarity with populations in the North Sea.

- 78 - CHAPTER III

Introduction

Vibrios are culturable bacteria that are mainly found in marine and coastal environments (Farmer 2006). They occur worldwide, with higher abundances in tropical and subtropical waters (Wietz et al. 2010). Some species are classified as human pathogens and can be transmitted by contaminated water or seafood (Linkous et al. 1999; Reidl et al. 2002; Drake et al. 2007). Within the genus Vibrio, especially three species are of clinical importance: V. cholerae is divided into the Cholera disease causing pandemic strains of the serotype O1 or O139 and non-O1/O139 strains that are associated with mild-to-moderate diarrhea (Mosley et al. 1970; Morris et al. 1981), hemolytic strains of V. parahaemolyticus are one of the main causes for bacterial seafood-related illness (Su et al. 2007) and C-type strains of V. vulnificus are responsible for fatal wound infections and septicemias (Oliver 2005; Jones et al. 2009).

Vibrio spp. are spread by water currents, zooplankton particles or the ballast water of ships (McCarthy et al. 1992; Ruiz et al. 2000; Vezzulli et al. 2013). In theory, this long-range dispersal over seawater allows marine bacteria to be spread across the globe. However, environmental parameters and biogeographic constraints seem to select for the proliferation of particularly pathogenic Vibrio spp. strains. For instance, the highly pathogenic V. cholerae serotype O139 is restricted to South Asia and pandemic strains of the serotype O1 have not been isolated in the North and Baltic Seas, although these strains have already crossed oceans and caused outbreaks at locations far away from each other (Popovic et al. 1995; Gil et al. 2004). Furthermore, the proportion of hemolytic V. parahaemolyticus environmental strains is lower in waters of Northern Europe and higher in warmer regions such as the Gulf of Mexico and the coastal waters of Spain (Cabrera-Garcia et al. 2004; Rodriguez-Castro et al. 2009; Boer et al. 2013). Therefore, only a few clinical cases of seafood related illness were reported so far in western and northern Europe (Baker-Austin et al. 2010). Nevertheless, pandemic V. parahaemolyticus strains of the serotype O3:K6 can migrate into the North Sea and, in consequence, will cause outbreaks, as reported for other regions with temperate waters such as Alaska and Chile (Gonzalez-Escalona et al. 2005; McLaughlin et al. 2005; Nair et al. 2007). Most obviously, V. vulnificus poses the highest risk for northern Europe, because fatal clinical cases are currently reported from the Baltic Sea (Ruppert et al. 2004). However, no infections are reported so far from the North Sea, which might be explained by different environmental conditions such as salinity, that in turn are selective for pathogenic V. vulnificus strains in the Baltic region.

- 79 - CHAPTER III

Since seawater, plankton and marine sediments serve as sources and reservoirs for Vibrio spp., studies on their dispersal, distribution and migration patterns are crucial to evaluate the risk of outbreaks (Thompson et al. 2004; Ceccarelli et al. 2014). However, biogeographic studies are rare: they were mainly conducted for regions with higher current infection rates such as South Asia and America, and less information is available concerning the distribution of potentially pathogenic Vibrio spp. in mildly affected regions such as Europe (Alam et al. 2006; Okada et al. 2012). However, due to climate change, the North and the Baltic Seas warm up more quickly, and numerous studies point to a positive correlation between the sea surface temperature (SST) and the abundance of potentially pathogenic Vibrio spp. as well as to the prevalence of virulence associated factors (Chowdhury et al. 1990; Oberbeckmann et al. 2011). Thus, increased incidences of Vibrio infections in the Baltic Sea were linked to global warming and higher abundances of these pathogenic strains are also expected for the North Sea (Lipp et al. 2002; Lindgren et al. 2012; Baker-Austin et al. 2013; Vezzulli et al. 2013).

Further environmental parameters such as salinity, plankton and dissolved organic matter (DOM) are also linked to the occurrence of vibrios (Chavez et al. 2005; Eiler et al. 2006; Sedas 2007; Turner et al. 2009; Asplund et al. 2011). V. vulnificus and V. cholerae strains were mainly isolated in regions with brackish waters such as the Baltic Sea and river in the North Sea, whereas V. parahaemolyticus and other Vibrio spp. were also isolated in areas of higher salinity such as offshore areas of the North Sea (Eiler et al. 2006; Oberbeckmann et al. 2011; Boer et al. 2013). Thus, salinity seems to affect the species composition of Vibrio assemblages in parts of the North and Baltic Seas

To obtain biogeographical data about Vibrio populations, environmental strains from distinct sampling sites are to be analyzed. A number of methods are suggested for typing Vibrio spp. strains such as multi-locus sequence analyses (MLSA) (Chowdhury et al. 2004; Gonzalez- Escalona et al. 2008), multiple-locus variable-number tandem-repeat analysis (MLVA) (Danin-Poleg et al. 2007; Ansede-Bermejo et al. 2010), variable number of tandem repeats analysis (VNTR) (Garcia et al. 2012) and enterobacterial repetitive intergenic consensus sequence polymerase chain reaction (ERIC-PCR) (Colombo et al. 1997; Khan et al. 2002; Bhanumathi et al. 2003; Keymer et al. 2009; Oberbeckmann et al. 2011). The latter method was chosen for this study because it provides reasonable discriminatory power,

- 80 - CHAPTER III reproducibility and a high sample throughput rate. The ERIC-PCR generates imperfect palindrome sequences, which are located in noncoding genome areas of Enterobacteria and Vibrio spp. (Sharples et al. 1990; Hulton et al. 1991). Due to the genetic assortment over evolutionary time, the number and the length of the ERIC fragments varies according to the similarity of strains (Wilson et al. 2006). Hence, amplification of these sequences by PCR leads to band patterns that are used as strain-specific profiles. Closely related ERIC-profiles are grouped according to their similarity of fragment lenght (band patterns). In consequence, the occurrence or co-occurrence of genotypes at different sampling sites allows conclusions about the dispersal of strains and the comparison of site-specific Vibrio populations (Simpson et al. 2002). Particularly Keymer et al. used ERIC-PCR genotyping for the analysis of geographical distribution for V. cholerae strains in California and have yielded valuable clues about temporal and spatial distribution patterns as well as evidence for the influence of environmental parameters that shape populations (Keymer et al. 2009).

In the present study, ERIC-profiles were generated for V. cholerae, V. parahaemolyticus and V. vulnificus strains that were mainly isolated from the North and Baltic Seas. The respective strains found at each sampling site are described as site-specific populations. We focused particularly on the genotype-based comparison of these site-specific populations to answer the following questions:

(1) Are there any significant differences concerning diversity and genotypic richness between site-specific populations or Vibrio assemblages of the North Sea and the Baltic Seas?

(2) Are geographical aspects (i.e. distances or barriers) of decisive importance for the similarity of site-specific populations?

(3) Can ERIC-PCR genotyping be used to explain Vibrio spp. migration patterns?

- 81 - CHAPTER III

Materials and Methods

Strains

Environmental Vibrio spp. strains were kindly provided from Craig Baker-Austin (CEFAS), Ciska Schets (RIVM), August Heinemeyer and Katrin Luden (NLGA), Gerhard Hauk (LAGuS), Edda Bartelt (LAVES) and Martin Hippelein (UKSH). Table 1 gives an overview of strains used in this study. Strains from an in-house culture collection of the Alfred Wegner Institute were also included. In total, 470 environmental strains were used: 63 V. cholerae strains, 226 V. parahaemolyticus strains and 181 V. vulnificus strains whereas 21 strains originate from the English Channel and the river Thames, 226 strains originate from the North Sea, 7 from the Skagerrak, 20 from the Kattegat and 196 from the Baltic Sea. Furthermore, two clinical V. cholerae strains were provided by the hospital Bremerhaven-Reinkenheide: one strain was isolated from a patient with seafood-related illness caused by shrimps from the North Sea (“Clinical North Sea”) and the other strain was isolated from a patient who was infected with a non-pandemic strain while traveling in Africa (“Clinical Africa”).

ERIC-PCR

All strains were incubated at 37°C and grown on marine broth 2216 medium with (Lemos et al. 1985), and marine broth 2215E medium with reduced salinity (50% seawater). DNA was extracted as described earlier (Oberbeckmann et al. 2011). For each strain, three replicates were used. Two primers were used to amplify ERIC sequences: ERIC2 (5´-3´ AAGTAAGTGACTGGGGTGAGCG) and ERIC1R (5´-3´ ATGTAAGCTCCTGGGGATT CAC) (Versalovic et al. 1991). Each 25µl ERIC-PCR mixture contained 5µL Gitschier buffer

(1M (NH4)2SO4, 1M Tris-HCl, pH 8.8, 1M MgCl, 0.5M EDTA, pH 8.8, 1% mercaptoethanol), 2.5μL dimethyl sulfoxide, 1.25μL dNTPs (10mM), 0.4µL bovine serum albumin (20mg/mL), 2U Taq DNA polymerase (5Prime), 1μL of each primer ERIC2/ERIC1R and 80ng DNA. PCR reactions were carried out in a Eppendorf mastercycler with an initial denaturation at 95°C for 2min followed by 31 cycles (denaturation at 94°C for 3s, 92°C for 30s, annealing at 50°C for 1min and extension at 68°C for 8min). A final extension was performed at 68°C for 8min. The ERIC-PCR products were analyzed using the MCE-202 MultiNA microchip electrophoresis system (Shimadzu). The microchip electrophoresis was performed in the on-chip mixing mode with GelStar DyeTM (Biozym) and a DNA 2500 ladder

- 82 - CHAPTER III

(Promega) at 37°C. The gel images, electrophorograms and tables with the size of chromatogram bands were generated using the MultiNA viewer software (Shimadzu). The gel images were checked manually: if more than one profile of a strain had a rather weak resolution or various displaced bands, the PCR was repeated to obtain profiles of high resolution and homogeneous bands. For genotyping, only one profile was picked for each strain.

Genotyping

Tables with the ERIC bands were imported as fingerprinting profiles into BioNumerics (version 7.1; Applied Maths). A dice similarity matrix was generated for the profiles of each species. Based on this matrix, profiles were clustered using the unweighted pair group method with arithmetic means (UPGMA) according to the default settings of BioNumerics (version 7.1; Applied Maths). Three parameters determined the use of specific similarity cut-offs between 60% and 100% for genotyping: reliability, discriminatory power and genotype to profile ratio. To specify the reliability of ERIC-PCR genotyping, three replicates of 40 strains for each species were clustered and the minimum similarity values of the replicates were calculated (Figure S-1). For each cut-off, the index of discrimination was calculated to determine the discriminatory power of the method, which is in turn based on the probability that two ERIC-profiles are characterized as the same genotype (Figure S-3) (Hunter 1990). The ratios between genotypes and profiles for each similarity cut-off were calculated as well (Figure S-2). Because some site-specific populations consist of only one or a few profiles, we aimed to perform a hierarchical genotyping that displays the relationship of profiles more accurate on three different levels: the highest level should have a low genotype to profile ratio (<0.2) but a high reliability (>95%); the second level should have a moderate reliability (>85%) and a moderate index of discrimination (>85%); the lowest level should have a high index of discrimination (>90%) as well as an acceptable reliability (>80%) and genotype to profile ratio (<0.5). Therefore, three similarity cut-offs were chosen for genotyping: 80% for the highest level, 85% for the second level and 88% for the lowest level.

- 83 - CHAPTER III

TABLE 1: Strain provider, sampling location, sample origin, sampling date and number of strains for each species.

strain provider sampling location sample origin sampling date V. cholerae V. parahaemolyticus V. vulnificus

English Coast, water, CEFAS1 2001-2006 21 River Thames (UK) mussels, crabs

RIVM2 Netherlands water 2009-2011 28

Lower Saxony NLGA3 water 2010-2012 15 42 88 (Germany)

Mecklenburg- LAGuS4 Vorpommern water 2004-2011 37 1 84 (Germany)

Lower Saxony LAVES5 mussels 2010-2012 2 1 (Germany)

Schleswig-Holstein UKSH6 water 2011 8 68 3 (Germany)

water, North Sea, plankton, AWI7 Skagerrak, Kattegat, 2008-2011 1 65 6 sediment, Baltic Sea mussels Bremerhaven clinical BRHV8 2012 2 (Germany) samples

sampling location total strains V. cholerae V. parahaemolyticus V. vulnificus

English Coast and river Thames 21 21 North Sea (Netherlands, Germany, Denmark) 226 23 115 88 Skagerrak 7 4 3 Kattegat 20 1 16 3 Baltic Sea (Germany, Denmark) 196 39 70 87

total strains 470 63 226 181

1 Centre for Environment, Fisheries & Aquaculture Science (United Kingdom) 2 Rijksinstituut voor Volksgezondheid en Milieu (Netherlands; National Institute for Public Health and the Environment) 3 Niedersächsisches Landesgesundheitsamt (Germany; Governmental Institute of Public Health of Lower Saxony) 4 Landesamt für Gesundheit und Soziales Mecklenburg-Vorpommern (Germany; Regional Office for Health and Social Affairs of Mecklenburg-Western Pomerania) 5 Niedersächsischen Landesamtes für Verbraucherschutz und Lebensmittelsicherheit (Germany; Lower Saxony State Office for Consumer Protection and Food Safety) 6 Universitätsklinikum Schleswig-Holstein (Germany; University Medical Center Schleswig-Holstein) 7 Alfred-Wegener-Institut für Polar- und Meeresforschung (Germany; Alfred Wegener Institute for Polar and Marine Research) 8 Bremerhaven Hospital-Reinkenheide (Germany)

- 84 - CHAPTER III

Biogeographic analyses

Each site-specific population was characterized by the occurrence of the corresponding genotypes. To examine whether site-specific populations have a high or low diversity, the ratios of profiles to genotypes were calculated for all site-specific populations with more than one profile and plotted with a regression line according to Keymer et. al. (Keymer et al. 2007). Detailed biogeographical analyses were performed based on the distribution of genotypes along sampling sites. To eliminate artificial correlations between sampling sites and the number of profiles, a statistical method was conducted which relies on the relative genotype co-occurrences between site-specific populations. Therefore, the Dice's coefficient was used in following the equations:

S (a;b) = ∑ ((ax+bx)/2)

where S (a;b) is the similarity between the populations at the sampling sites a and b, ax and bx are the relative occurrences of genotypes, which are present in both populations, and g is the number of shared genotypes (genotype co-occurrences). This equation can be elucidated exemplarily by the following hypothetical relative genotype distributions for the populations at site A, B and C:

genotype 1 genotype 2 genotype 3 site A 0.8 0.2 0 S (site A, site B) = (0.8+0.6)/2 + (0.2+0.2)/2 = 90% site B 0.6 0.2 0.2 S (site A, site C) = (0.2+0.4)/2 = 30% site C 0 0.4 0.6 S (site B, site C) = (0.2+0.4)/2 + (0.2+0.6)/2 = 70%

The average similarities for each cut-off were calculated and a similarity matrix for each species was generated. Based on this similarity matrix, a cluster analyses was performed using PRIMER (group average clustering). Similarity cut-offs of 40% (V. cholerae, V. parahaemolyticus) and 50% (V. vulnificus) were used to group at least an average of two site specific populations into one cluster (cluster/population ratio = 0.5) (Figure S-9). Genotype co-occurrence similarities were calculated between each site-specific population and the North Sea as well as the Baltic Sea assemblages to generate and analyze data in a larger context. Relationships between the occurrence of dominant genotypes of the highest cut-off level and the year of isolation were further analyzed to reveal putative temporal shifts in site- specific populations.

- 85 - CHAPTER III

Results

ERIC-PCR profiles and generation of genotypes

ERIC-PCR profiles of genomic DNA displayed 12 to 31 bands ranging in sizes between 64 to 3929 bp. As an example, Figure 1 shows the gel image, electrophoretogram and cluster analysis of 9 ERIC-profiles. Reliability values of 50%, 75% and 100% are reached with similarity cut-offs at 92%, 88% and 80% respectively (Figure S-4). With regards to reliability, genotype to profile ratio and index of discrimination, three similarity cut-offs were chosen for genotyping: 80%, 85% and 88% (Figure S-4, S-5 and S-6). The cluster analysis, the applied similarity cut-offs and the resulting genotypes for V. cholerae strains are exemplarily shown in Figure S-7. No significant correlation was observed between the similarity value of all pairwise site-specific population combinations and the according number of included ERIC- PCR profiles (data not shown). Thresholds of 40% (V. cholerae and V. parahaemolyticus) and 50% (V. vulnificus) were chosen for the cluster analysis of Vibrio spp. site-specific populations (Figure 2, Figure S-8).

Table 2 gives an overview for all ERIC-PCR profiles: for the 80%-similarity-cut-off, 71% genotypes have more than one profile whereas each genotype consists on average of 9.3 to 12.9 profiles; for the 85%-similarity-cut-off, 60% to 70% genotypes consist of more than one profile with an average profile to genotype ratio of 3.6 to 4.9; for the 88%-similarity-cut-off, 40% to 59% genotypes have more than one profile and on average of 2.2 to 3.0 profiles belong to each genotype. For all similarity-cut-offs, V. vulnificus has the highest number of profiles in one genotype (n=86) and the highest average profile to genotype ratio; V. parahaemolyticus displays the highest discriminatory indices (80.5%; 97.0%; 98.6%), followed by V. vulnificus with the second highest discriminatory indices (71.0%; 91.0%; 96.3%) and V. cholerae with the lowest discriminatory indices (64.5%; 87.9%; 93.9%). Further detailed information about the genotype assignment of each ERIC-PCR profile and about the distribution of genotypes for each sampling site can be found in the supplementary documents.

- 86 - CHAPTER III

A B

3516_B4

3516_F4

3516_B5

2907_C6

2907_C7

2907_C8

3537_C9

3537_C10

3537_C11

3516_B4 C 3516_F4 3516_B5

2907_C7

2907_C8

2907_C6

3537_C10

3537_C11

3537_C9

FIGURE 1: Microchip gel electrophoresis of three replicates of V. cholerae VN-3516, V. parahaemolyticus VN- 2907 and V. vulnificus VN-3537: A) gel image with the marker in lane 1 and the ERIC profiles in lane 2 to 10; B) electropherogram according to the gel image (MultiNA viewer); C) fingerprint profiles and cluster analysis with the similarity values (BioNumerics).

- 87 - CHAPTER III

TABLE 2: Total profiles, sampling sites, total genotypes, genotypes with more than one profile, the maximal number of profiles for one genotype, the ratios between profiles and genotypes, discrimatory indices and the total genotype co-occurrences between site-specific populations for all three similarity cut-offs (80%, 85%, 88%) and all three species. * 14 samling sites in the environment and two clinical strains.

V. cholerae V. parahaemolyticus V. vulnificus

total profiles 65 226 181 sampling sites 16* 31 22

80% similarity cut-off

total genotypes 7 24 14 genotypes with >1 profile 5 (71%) 17 (71%) 10 (71%) maximal profiles/genotype 29 70 86 profiles/genotypes 9.3 9.4 12.9 discriminatory index 64.5% 80.5% 71.0% genotype co-occurences 268 1348 670

85% similarity cut-off

total genotypes 17 63 37 genotypes with >1 profile 11 (65%) 38 (60%) 26 (70%) maximal profiles/genotype 17 20 41 profiles/genotypes 3.8 3.6 4.9 discriminatory index 87.9% 97.0% 91.0% genotype co-occurences 199 870 509

88% similarity cut-off

total genotypes 30 91 61 genotypes with >1 profile 12 (40%) 51 (56%) 36 (59%) maximal profiles/genotype 11 9 23 profiles/genotypes 2.2 2.5 3.0 discriminatory index (%) 93.9% 98.6% 96.3% genotype co-occurences 148 567 484

Biogeographic analysis

V. cholerae

One major cluster was identified and divided into the two sub-clusters C1a and C1b (Figure 2A). Each of them consists of 4 site-specific populations and 72% of all V. cholerae profiles belong to these two sub-clusters. The populations of these clusters belong to both, the North Sea and the Baltic Sea sites. Three neighboring populations at the peninsula Darss-Zingst belong to different (sub-)clusters: Darss-Zingst Bodden reveals the highest similarity to Darss-Zingst West and Darss-Zingst North (53 and 51% respectively) whereas the latter two populations reveal a lower similarity of 32% (Table S-1). A similar result has been achieved - 88 - CHAPTER III for populations at the island of Rügen in the eastern part of the German Baltic Sea coast: Dranske and Greifswald Bodden belong to the subclusters C1a and C1b respectively and reveal a similarity of 63%, but the similarity to Binz is low (16% and 0%) (Figure 2A, Table S-1). Only 3 of 10 similarity values between site-specific populations of the North Sea are above 40%, whereas 12 of 23 similarity values between Baltic Sea populations and 18 of 40 similarity values between North Sea and Baltic Sea populations reach this threshold (Table S- 1). In this context, all site-specific populations of the North Sea are shown to be more similar to the Baltic Sea than to the North Sea assemblages, and only two site-specific populations of the Baltic Sea (Bay of Wismar, Greifswald Bodden) are more similar to the North Sea assemblage (Figure 3A).

North Sea populations are less diverse compared to the Baltic Sea populations: 3 of 4 North Sea populations (except Ems ) fell below the regression line but 5 of 6 Baltic Sea populations (except Eckernfoerde Bay) fell above the regression line (Figure 4A). In this context, almost all V. cholerae genotypes are present in Baltic Sea populations (89% to 100%) whereas only 44% to 50% of all genotypes were observed in North Sea populations (Table S- 4).

The profile from the Kattegat is only connected with the population at Darss-Zingst-Bodden (32% similarity). The “Clinical Africa” profile is separated from all site-specific populations and reveals no genotype co-occurrences with any environmental profile of the North and Baltic Seas. In contrast, the “Clinical North Sea” profile has such genotype co-occurrences and is closer related to the V. cholerae assemblages of the North Sea (20% similarity) than to those of the Baltic Sea (14% similarity) whereas the Ems estuary has the highest similarity with “Clinical North Sea” (38%) (Figure 3A, Table S-1).

- 89 - CHAPTER III

A Dranske (DR) Meldorf Bay (MB) Cluster C1a 71% similarity Kattegat Eckernfoerde Bay (EB) V. cholerae (n=16) Darss Zingst Bodden (DB) Darss Zingst West (DW) KA Cluster C1b Estuary(WE) 67% similarity Greifswald Bodden (GB) (n=31) (EF) Darss Zingst North (DN) Clinical North Sea (CN) Clusters C2 C3 C4 Baltic Sea Ems Estuary (EE) DR Cluster C5 DN Bay of Wismar (BW) EB 100% similarity Jade Bight (JB) North Sea DW BI (n=2) GB MB DB Binz (BI) Kattegat (KA) Clusters C6 C7 C8 EF BW WE Clinical Africa (CA) JB EE CN CA 0 20 40 60 80 100 Ijsselmeer (IJ) Cluster P1 Skagerrak Brown Ridge North (BN) 59% similarity (n=9) SK English Channel (EC) Falmouth Bay (FB) Cluster P2 B Eckernförde Bay (EB) 66% similarity (n=8) Kattegat Schlei Estuary (SC) Cluster P3 V. parahaemolyticus North Frisia (NF) Elbe Estuary (EL) 45% similarity (n=33) KA Kiel Bight (KB) JU Scheldt Estuary (SE) Cluster P4 Rostock (RO) 54% similarity (n=58) Eastern North Sea Fehmarn Belt (FE) Baltic Sea Flensburg Firth (FF) Ems Estuary (EE) Cluster P5 SY FF Borkum (BO) DN RN 50% similarity (n=27) NF Brown Ridge South (BS) EB FE Lubec kBight (LB, Cluster P6) Southern North Sea HR TB LB Darss Zingst North (DN) Cluster P7 GE SC KB RO BO NO Sylt (SY) 49% similarity (n=11) EL River Thames (RT, Cluster P8) WF EF Norderney (NO) Cluster P9 WE West Frisia (WF) 49% similarity (n=19) EE Weser Estuary(WE) BN Ruegen North (RN) Cluster P10 Skagerrak (SK) 59% similarity (n=24) BS IJ RT Kattegat (KA) Tuemlau Bight (TB) East Frisia (EF) Clusters Jutland Bank (JU) P11 P12 SE EC German Bight (GE) P13 P14 P15 English Channel Helgoland Roads (HR) FB 0 20 40 60 80 100 Darss Zingst North (DN) Cluster V1 C Skagerrak SK Kiel Bight (KB) 57% similarity (n=8) Greifswald Bodden (GB) Cluster V2 V. vulnificus Darss Zingst Bodden (DB) 59% similarity (n=18) Pepelow Rerik (PR) Cluster V3 KA Lubec kBay (LB) 97% similarity (n=12) Binz (BI) Kattegat Dranske (DR) Rostock (RO) Cluster V4a Usedom North (UN) 65% similarity (n=43) Jasmunder Bodden (JA) Lubmin (LU) Weser Bremerhaven (WB) Eastern East Frisia (EF) Cluster V4b Weser Dedesdorf (WD) 63% similarity (n=79) DR Baltic Sea North Sea DN JA Weser Wremen (WW) Western Baltic Sea Bay of Wismar (BW, Cluster V4c) RO BI Darss Zingst West (DW, Cluster V5) KB BW DW DB UN EL PR GB LU Ems Estuary (EE, Cluster V6) EF LB Kattegat (KA, Cluster V7) WW Skagerrak (SK, Cluster V8) WB WD Elbe Estuary (EL, Cluster V9) EE 0 20 40 60 80 100

FIGURE 2: Map and cluster analysis based on the genotype co-occurrence similarities of site-specific populations. The similarity cut-off for V. cholerae (A) and V. parahaemolyticus (B) was 40% and for V. vulnificus (C) 50%. The resulting clusters were assigned to the respective species (C,P,V) with continous numbers. The cluster V4 was separated into sub-clusters (a,b,c).

- 90 - CHAPTER III

A V. cholerae B V. parahaemolyticus

Clinical North Sea B

Bay of Wismar Ijsselmeer

30 20 10 10 20 30 Scheldt Estuary similarity to the North Sea similarity to the Baltic Sea 20 10 10 20 C V. vulnificus similarity to the North Sea similarity to the Baltic Sea Clinical strains Dranske Eastern Baltic Sea Western Baltic Sea

Skagerrak and Kattegat Western North Sea Southern North Sea English Channel 25 15 5 5 15 25 similarity to the North Sea similarity to the Baltic Sea FIGURE 3: Comparison between the genotype co-occurrence based similarity of site-specific populations to the North Sea and the Baltic Sea overall population. A) V. cholerae B) V. parahaemolyticus C) V. vulnificus.

8 15 B A 12 6 9 4

genotypes 6 genotypes 2 3 V. parahaemolyticus V. cholerae 0 0 C 0 5 10 15 20 0 5 10 15 20 25 30 profiles profiles C 15 sampling site-specific populations 12

9 North Sea Baltic Sea genotypes 6 English Channel

3 V. vulnificus Skagerrak / Kattegat

0 0 5 10 15 20 25 30 35 profiles FIGURE 4: Relationship between the number of profiles and generated genotypes for each site- specific population with more than one profil. A polynomial regression was performed: y=- 0.0184x2+0.7519x and R2=0.5246 for V. cholerae (A), y=-0.0104x2+0.7905x and R2=0.9272 for V. parahaemolyticus (B) and y=-0,0076x2+0,6316x and R2=0,898 for V. vulnificus (C). Sampling site- specific populations above the regression line reveal higher genotype richness and a higher diversity respectively.

- 91 - CHAPTER III

V. parahaemolyticus

Different clusters with geographic associations were identified (Figure 2B): the three clusters P1, P5 and P7 are distinct from each other but consist exclusively of site-specific populations from the southern North Sea; the cluster P3 consists of populations located in the eastern North Sea and western North Sea; the cluster P4 has the most profiles (n=58), is closely related with the cluster P3 and, with the exeption of the Scheldt estuary, all site-specific populations of this cluster are located in the Baltic Sea. Populations in offshore areas of the North Sea such as Helgoland Roads, German Bight and Jutland Bank form the most separated clusters and reveal, with the inclusion of Rügen North, the lowest overall similarities (Figure 2B; Table S-2).

Except Scheldt estuary and Ijsselmeer, all populations from the southern North Sea area are more similar to the V. parahaemolyticus assemblage of the North Sea (Figure 3A). In contrast, sites located in the eastern North Sea at the coast of Schleswig-Holstein are more similar to the assemblage of the Baltic Sea. Except Rügen North, all site-specific populations of the Baltic Sea are closer related to the Baltic Sea assemblage. All site-specific populations reveal a similar diversity because they are all distributed along the regression line (Figure 4B). However, the genotype richness is higher in the assemblage of the North Sea than in those of the Baltic Sea: total genotypes (76%-90% / 42%-57%) and exclusive genotypes (50% / 20%) are more present there (Table S-4).

The northernmost populations (Kattegat and Skagerrak) belong to the cluster P8 and are slightly more similar to the assemblage of the Baltic Sea (Figures 2B and 3B). Although site- specific populations in England are located west of the North Sea, they are more similar to the assemblage of the further east located Baltic Sea (Figure 3B). Among these populations at the English coast, Falmouth Bay and English Channel form the separated cluster P2 whereas the River Thames population is closely related to populations from Sylt and Darss-Zingst North (Figure 2B).

Based on these data, all V. parahaemolyticus populations can be separated into four biogeographic regions: 1) southern North Sea; 2) offshore areas of the North Sea; 3) Skagerrak / Kattegat / eastern Baltic Sea; 4) England / Scheldt estuary / eastern North Sea / Baltic Sea.

- 92 - CHAPTER III

V. vulnificus

The majority of the profiles (69%) belong to the sub-clusters V4a, V4b and V4c (Figure 2C). Except Greifswald Bodden, all populations in the eastern Baltic Sea around the island of Rügen and Usedom belong to the subcluster V4a. In contrast, the subcluster V4b is exclusive for North Sea populations around the Weser estuary and has an average similarity of 63%. The other two North Sea populations, Ems estuary and Elbe estuary, form separated clusters and reveal a similarity of 38% and 25% respectively with the subcluster V4b (Figure 2C, Table S-3). Beside Elbe Estuary and Ems estuary, the populations at the Skagerrak and Kattegat also form separated clusters and reveal no similarity values higher than 50% with any other site-specific population. Neighboring populations at the peninsula Darss-Zingst in the western Batlic Sea belong to different clusters: Darss-Zingst Bodden and Darss-Zingst North are the closest populations with a similarity of 41% whereas Darss-Zingst West was strongly separated with a similarity of only 10% and 21% respectively to the other two populations.

Most of the site-specific populations are more similar to the V. vulnificus assemblage of the North Sea, in particular, all populations from the North Sea, from the Skagerrak / Kattegat and from the western Baltic Sea, except Dranske and Greifswald Bodden (Figure 3C). However, site-specific populations of the North Sea and the Baltic Sea reveal a comparable diversity (Figure 4C), and most of the genotypes are either exclusive for the North Sea (42%) or exclusive for the Baltic Sea (30%) (Table S-4). These data suggest a separation into three biogeographic regions: 1) North Sea and eastern Baltic Sea (Ruegen/Usedom), 2) Skagerrak/Kattegat and 3) western Baltic Sea (Kiel to Darss-Zingst).

Spatial and temporal distribution of genotypes

The 80%-similarity-cut-off analysis of all species revealed certain dominant genotypes. For V. cholerae, genotype 1 and 4 are the most dominant ones: 26 profiles and 29 profiles respectively are classified with these two genotypes. They reveal an almost equal temporal (2010/2011) and spatial (North Sea/Baltic Sea) distribution (Figure 5). For V. parahaemolyticus, three genotypes are dominant: genotype 3 (45 profiles), genotype 5 (36 profiles) and genotype 11 (70 profiles). Although no profiles of the southern North Sea strains from 2011 were classified as genotype 11, it seems to be a ubiquitous genotype because of its

- 93 - CHAPTER III presence in 24 of 31 populations (Figure S-6). Genotype 3 and genotype 5 reveal an equal distribution, spatially and temporally (Figure 5). For V. vulnificus, two genotypes are dominant: the genotype 4 (40 profiles) and the genotype 5 (86 profiles) (Figure5). 63% of all profiles from site-specific populations in the western Baltic Sea belong to genotype 4 whereas most of the North Sea profiles (63%) and profiles from the eastern Baltic Sea (53%) belong to genotype 5. Temporal shifts were recognized for these genotypes as well. North Sea profiles from 2011/2012 are classified as genotype 5 to a minor extent as compared to profiles from 2010 (27% / 61%). This is also true for profiles from the eastern Baltic Sea (genotype 5: 2004-2008 76%, 2010 57%). The profiles from the western Baltic Sea reveal decreased genotype 5 and genotype 4 classifications for strains isolated in 2010 (genotype 4: 31%; genotype 5: 3%) compared to strains isolated from 2004 to 2008 (genotype 4: 80%; genotype 5: 20%). However, the spatial difference for the dominance of genotypes for V. vulnificus remains resistant over time.

V. cholerae - North Sea - 2010 V. cholerae - North Sea - NS 2011 V. cholerae - Baltic Sea - 2011 V. parahaemolyticus - southern North Sea - 2010 V. parahaemolyticus - southern North Sea - 2011 V. parahaemolyticus - eastern North Sea - 2011 V. parahaemolyticus - Baltic Sea - 2011 V. vulnificus - North Sea - 2010 V. vulnificus - North Sea - 2011/12 V. vulnificus - western Baltic Sea - 2004-2008 V. vulnificus - western Baltic Sea - 2011/12 V. vulnificus - eastern Baltic Sea - 2004-2008 V. vulnificus - eastern Baltic Sea - 2010 0% 20% 40% 60% 80% 100% V. cholerae - genotype 1 V. parahaemolyticus - genotype 3 V. vulnificus - genotype 4 V. cholerae - other genotypes V. parahaemolyticus - genotype 5 V. vulnificus - other genotypes V. cholerae - genotype 4 V. parahaemolyticus - genotype 11 V. vulnificus - genotype 5 V. parahaemolyticus - other genotypes

FIGURE 5: Spatial and temporal distribution of specific genotypes (80% cut-off) that are dominant for V.cholerae, V. parahaemolyticus and V. vulnificus in the North and Baltic Seas.

- 94 - CHAPTER III

Discussion

Our study provides first information on the biogeography of the population structure of three potentially pathogenic Vibrio spp. in the North and Baltic Seas. Less than 27% genotypes are exclusively restricted to a site-specific population at the highest similarity-cut-off level. Not more than one strain is assigned to each of these genotypes and only 2.6% of all strains are classified as geographically restricted genotypes. These findings correspond to those of Keymer et al.: they found only 4% of geographically restricted V. cholerae genotypes (Keymer et al. 2009). Thus, each site-specific population reflects a diverse composition of, to some extent, frequently common genotypes.

For V. parahaemolyticus and V. vulnificus, this genotypic diversity within each site-specific population is largely the same throughout the North and Baltic Seas. For V. cholerae, however, the site-specific populations from the Baltic Sea are more diverse than those from the North Sea. Furthermore, the V. cholerae assemblage of the Baltic Sea includes almost all the genotypes that are present in the North Sea. On the basis of these findings, the Baltic Sea can be seen as natural reservoir of phylogenetic diversity for V. cholerae. Several studies show that V. cholerae strains are frequently attached to plankton and coherences were observed between site-specific Vibrio populations and the dominant plankton species at the respective sampling sites (Huq et al. 1983; de Magny et al. 2011; Kirschner et al. 2011). In this context, a higher plankton species diversity was observed in waters of low-salinity (Estrada et al. 2004). We assume, that the herein discovered higher genotype richness of V. cholerae strains in the Baltic Sea is linked to the attachment to more different plankton species. Recent studies about the microbiom of marine plankton cannot either confirm or exclude this hypothesis, because they are based on higher taxonomic classifications (Gerdts et al. 2013; Corte et al. 2014). Therefore, further studies are needed to analyze the affinity of V. cholerae to certain plankton species of temperate waters.

Despite the absence of locally restricted genotypes, the biogeographic analyses of genotype co-occurrences revealed important differences as well as common features of site-specific populations. Particularly striking is the separation of V. parahaemolyticus populations into different regions. This separation cannot be explained solely in terms of geographic barriers and distance. This is evident above all in the weak relationship between offshore populations in the North Sea and their neighboring nearshore populations in the southern and eastern - 95 - CHAPTER III

North Sea. The environmental conditions at these locations differ from each other. Offshore areas are influenced by rather seasonal fluctuations. In contrast, the continuously changing influx/efflux of seawater and the concominant supply of freshwater from rivers lead to variable salinity and temperature in coastal areas (Swertz et al. 1999). Therefore, we assume that these different environmental conditions play an important role in the process of shaping V. parahaemolyticus populations in the North Sea: either directly or by the occurrence of different plankton species. Furthermore, populations of the eastern North Sea coast are rather similar to assemblages of the Baltic Sea than to those of the southern North Sea coast. This finding is even more surprising since Vibrio populations of both regions, the eastern North Sea and the western Baltic Sea, have to cope with clearly different conditions and are hardly connected by the Skagerrak and Kattegat. Hence, the close relationship between sites in the eastern North Sea and the western Baltic Sea remains rather obscure. The same applies to V. parahaemolyticus strains from the English coast: despite of the continuously high salinity in this region, site-specific populations are more similar to assemblages of the Baltic Sea. In contrast, vibrios of the Ijsselmeer as well as from the Thames and Scheldt estuary are exposed to brackish and freshwater conditions. In this case, the close relationship of these populations to the V. parahaemolyticus assemblages of the Baltic Sea seems to be conclusive. However, other environmental parameters are to be considered for the formation of V. parahaemolyticus populations. In this regard, zooplankton was already mentioned as a reservoir and vector for offshore V. parahaemolyticus strains in Spain (Martinez-Urtaza et al. 2011). Thus, the affinity of V. parahaemolyticus to different temperate plankton species should be analyzed as well in the future (as already stated for V. cholerae).

For V. cholerae, a separation into regions was not possible and even most of the neighboring populations differed from each other. Since no clusters were associated to the North Sea or the Baltic Sea, we assume, that salinity has no greater influence on the formation of V. cholerae populations.

Given the different incidences of V. vulnificus in the North Sea and the Baltic Sea, the comparison of site-specific populations from both regions is of particular interest. Indeed, V. vulnificus populations can be separated into biogeographic groups. Most of the site-specific populations belong to locally restricted clusters or sub-clusters, for instance the closely related populations near to the Weser estuary in the North Sea and those in the eastern Baltic Sea. However, no clear separation was observed between V. vulnificus assemblages of the North

- 96 - CHAPTER III

Sea and assemblages of the Baltic Sea. Our results rather suggest connections of the populations of the North Sea and the eastern Baltic Sea than among site-specific populations of the Baltic Sea. This close relationship of populations from the North Sea and the eastern Baltic Sea is not linked with the different incidences in these regions. However, the strains used in this study were isolated during mild summers between 2010 and 2012, when no clinical V. vulnificus cases were reported. Thus, no environmental strains are included in this study that are linked to human infections. In consequence, strains are to be analyzed by ERIC- PCR genotyping during periods of high V. vulnificus incidences; in particular to observe whether spatial patterns are changing and Baltic Sea assemblages might be separated clearly from North Sea assemblages.

The absence of V. vulnificus and V. cholerae strains in offshore areas of the North Sea confirms previous studies, which stated that these species rather prefer environments of low salinity (Kelly 1982; Singleton et al. 1982). Nevertheless, both species are known to endure periods of disadvantageous environmental conditions such as decreased temperatures, low nutrient levels or high salinity by entering the protective viable-but-nonculturable (VBNC) survival state (Thomas et al. 2006; Nowakowska et al. 2013). Thus, V. vulnificus as well as V. cholerae strains do likely form VBNC stages while passing waters of high salinity between the river estuaries of the North Sea and the Baltic Sea. In contrast, V. parahaemolyticus is known to possess improved osmoadaptation (Naughton et al. 2009), which facilitates this species to grow in waters of high salinity. This allows pandemic pathogenic strains a faster colonization of new habitats. Further surveillance programs are to be conducted to reveal whether these highly pathogenic V. parahaemolyticus strains already reached the North and Baltic Seas; and if so, whether these strains become a permanent part of site-specific V. parahaemolyticus populations.

Since only two strains from clinical samples were included in this study, no statement can be made for the asscociation of pathogens with a site-specific population. However, general trends can be suggested, e.g. in case of the higher similarity to North Sea assemblages for the “Clinical North Sea” V. cholerae strain. Furthermore, newly imported strains might be distinguishable from strains of European waters, e.g. in case of the “Clinical Africa” V. cholerae strain.

- 97 - CHAPTER III

In this study, ERIC-PCR genotyping reveals a lower index of discrimination compared to other studies, e.g., 95% for Haemophilus parsuis (Moreno et al. 2011), 94% for Listera spp. (Laciar et al. 2006) and 83% for Escherichia coli (Mohapatra et al. 2007), with similarity cut- offs of 95%, 80% and 65% respectively. Even V. cholerae strains reached a discriminatory index of 97% using a 80% similarity cut-off (Keymer et al. 2009). The lower index of discrimination might be explained by the origin of the strains: they were obtained during different sampling campaigns with varying sampling sizes and variable sampling frequencies. These circumstances lead to highly variable numbers of isolated strains for each sampling site and probably to underrepresented or overrepresented genotypes that in turn decrease the index of discrimination. However, using three similarity cut-offs improved the analyses: the low cut-off (80%) yielded to comparable genotypes with a profile-to-genotype ratio and higher cut-offs (85%, 88%) revealed a higher discriminatory power for the genotype co-ocurrence similarity analyses.

Acknowledgments

This work was supported by the Federal Ministry of Education and Research (VibrioNet, BMBF 669 grant 01KI1015A). We would like to thank Sarah Dehne and Hilke Döpke for their valuable contribution to this study.

- 98 - CHAPTER III

Supplementary

TABLE S-1: Genotype co-occurrence similarity matrix of V. cholerae populations.

North Sea Baltic Sea

n

Binz

Zingst West Zingst

Zingst North Zingst

-

Dranske

Ranking Kattegat

-

Zingst Bodden Zingst

ß

East Frisia East Bight Jade -

Ems Estuary Ems

Meldorf Bay Meldorf

Weser Estuary Weser

Clinical Africa Clinical

Bay of Wismar Bay of

Eckernförde Bay Eckernförde

Dar

Clinical North Sea North Clinical

Darß

Greifswald Bodden Greifswald Darß Ems Estuary 3 10 27 16 20 38 50 0 38 16 48 32 40 38 8 31 0 East Frisia 8 3 27 40 61 18 44 0 66 40 58 44 54 61 34 86 0

Jade Bight 1 11 16 40 22 24 0 0 15 100 15 19 15 0 0 42 0 Weser Estuary 7 8 20 61 22 22 14 0 66 22 65 22 37 14 6 69 0

North Sea North Clinical North Sea 1 13 38 18 24 22 0 0 15 24 15 19 15 0 0 20 0 Meldorf Bay 4 7 50 44 0 14 0 0 77 0 50 43 55 84 16 50 0 Kattegat 1 15 0 0 0 0 0 0 0 0 0 0 32 0 0 0 0 Eckernförde Bay 4 1 38 66 15 66 15 77 0 15 69 37 66 82 13 82 0 Bay of Wismar 1 11 16 40 100 22 24 0 0 15 15 19 15 0 0 42 0

Darß-Zingst West 13 5 48 58 15 65 15 50 0 69 15 36 53 66 39 62 0 Darß-Zingst North 7 9 32 44 19 22 19 43 0 37 19 36 51 34 26 30 0 Darß-Zingst Bodden 7 4 40 54 15 37 15 55 32 66 15 53 51 63 27 49 0

Baltic Sea Baltic Dranske 1 6 38 61 0 14 0 84 0 82 0 66 34 63 16 67 0 Binz 3 14 8 34 0 6 0 16 0 13 0 39 26 27 16 0 0 Greifswald Bodden 3 2 31 86 42 69 20 50 0 82 42 62 30 49 67 0 0 Clinical Africa 1 16 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

- 99 - CHAPTER III

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Ems Estuary (n=3) 1 3 4 East Frisia (n=8) 1 4 Jade Bight (n=1) 1 Weser Estuary (n=7) 1 4 Clinical North Sea (n=1) 1 Meldorf Bay (n=4) 4 Kattegat (n=1) 5 Eckernförde Bay (n=4) 1 4 Bay of Wismar (n=1) 1 Darß-Zingst West (n=13) 1 2 3 4 Darß-Zingst North (n=7) 1 3 4 Darß-Zingst Bodden (n=7) 1 3 4 5 Dranske (n=1) 4 Binz (n=3) 2 4 7 Greifswald Bodden (n=3) 1 4 Clinical Africa (n=1) 6

80% similarity V. cholerae cut-off genotype (n=7) distribution among sampling sites (n=16) of ERIC-PCR profiles (n=65)

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Ems Estuary (n=3) 4 9 12 East Frisia (n=8) 1 3 11 12 Jade Bight (n=1) 1 Weser Estuary (n=7) 3 14 Clinical North Sea (n=1) 4 Meldorf Bay (n=4) 12 Kattegat (n=1) 15 Eckernförde Bay (n=4) 3 12 Bay of Wismar (n=1) 1 Darß-Zingst West (n=13) 2 3 8 9 11 12 13 14 Darß-Zingst North (n=7) 2 5 14 7 10 11 12 Darß-Zingst Bodden (n=7) 2 3 9 11 12 15 Dranske (n=1) 12 Binz (n=3) 8 11 17 Greifswald Bodden (n=3) 1 3 12 Clinical Africa (n=1) 16

85% similarity V. cholerae cut-off genotype (n=17) distribution among sampling sites (n=16) of ERIC-PCR profiles (n=65)

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Ems Estuary (n=3) 7 13 21 East Frisia (n=8) 1 5 16 17 18 Jade Bight (n=1) 2 Weser Estuary (n=7) 5 25 Clinical North Sea (n=1) 8 Meldorf Bay (n=4) 18 19 20 21 Kattegat (n=1) 26 Eckernförde Bay (n=4) 5 18 19 Bay of Wismar (n=1) 2 Darß-Zingst West (n=13) 4 5 12 13 17 18 22 23 24 Darß-Zingst North (n=7) 3 9 10 11 15 16 20 Darß-Zingst Bodden (n=7) 3 6 14 16 18 19 27 Dranske (n=1) 18 Binz (n=3) 12 16 29 Greifswald Bodden (n=3) 1 5 18 Clinical Africa (n=1) 28

88% similarity V. cholerae cut-off genotype (n=29) distribution among sampling sites (n=16) of ERIC-PCR profiles (n=65)

FIGURE S-1: Genotype distribution of V. cholerae strains based on Table-S1.

- 100 - CHAPTER III

TABLE S-2: Genotype co-occurrence similarity matrix of V. parahaemolyticus populations.

English Channel North Sea -West North Sea - East Baltic Sea

North

n Estuary

Sylt

ZingstNorth

Borkum Rostock

Ranking Kattegat -

Skagerrak

Ijsselmeer

Kiel Bight Kiel

Norderney Frisia East

Ems Estuary Ems

Rügen

JutlandBank

Lubeck Bight Lubeck FehmarnBelt

RiverThames

FalmouthBay Bight German

Weser Weser SchleiEstuary

TuemlauBight

Flensburg Firth Flensburg

Scheldt Estuary Scheldt

EnglishChannel

Eckernförde Bay Eckernförde

HelgolandRoads

Darß

Brown Ridge South BrownRidge North BrownRidge

West Frisia Wadden Sea Wadden Frisia West

North Frisia Wadden Sea Sea Wadden Frisia North

Meldorf Bay / Elbe Estuary Elbe /Bay Meldorf

Falmouth Bay 1 21 66 13 15 13 57 0 0 13 16 18 13 13 13 0 11 0 14 19 0 13 13 58 0 12 31 15 36 20 16 0

English Channel 7 15 66 7 14 20 39 12 0 30 26 12 11 16 14 13 26 0 29 19 45 7 26 26 0 23 32 42 38 14 10 0

River Thames 13 20 13 7 25 14 6 0 12 18 38 35 8 18 6 4 14 22 14 45 0 20 32 31 12 24 22 8 24 13 27 7

Scheldt Estuary 6 2 15 14 25 28 34 34 12 29 25 23 25 30 14 7 27 12 49 47 0 21 20 51 58 59 62 19 49 62 26 7

EnglishChannel Brown Ridge South 10 11 13 20 14 28 28 11 13 38 50 27 18 32 11 4 23 24 17 11 0 32 15 22 13 44 34 38 29 13 9 8

Brown Ridge North 8 18 57 39 6 34 28 59 0 22 36 10 21 22 11 7 6 0 22 11 0 5 9 25 0 28 25 41 39 13 9 0

Ijsselmeer 1 27 0 12 0 34 11 59 0 12 13 0 13 13 0 14 0 0 29 0 0 0 0 30 0 32 29 15 31 0 0 0

West Frisia Wadden Sea 3 26 0 0 12 12 13 0 0 13 12 62 0 56 0 0 13 20 11 0 0 15 13 12 20 15 54 0 0 0 0 15

Borkum 10 5 13 30 18 29 38 22 12 13 62 30 19 24 13 10 33 26 30 18 21 36 17 29 13 26 31 15 29 57 9 8

West

- Ems Estuary 7 4 16 26 38 25 50 36 13 12 62 61 14 21 8 6 26 22 23 27 0 37 17 33 12 50 34 44 57 16 17 7

Norderney 4 9 18 12 35 23 27 10 0 62 30 61 10 28 10 0 29 13 27 30 0 38 29 22 13 23 34 13 32 18 21 8

North Sea Sea North

East Frisia 8 25 13 11 8 25 18 21 13 0 19 14 10 21 11 18 13 0 15 16 0 9 11 10 0 10 19 15 14 13 9 6

Weser Estuary 12 10 13 16 18 30 32 22 13 56 24 21 28 21 30 17 14 22 35 14 34 18 24 24 13 26 26 15 16 13 11 8

Helgoland Roads 4 28 13 14 6 14 11 11 0 0 13 8 10 11 30 25 8 0 15 15 24 5 19 6 0 4 7 8 8 13 9 0

German Bight 6 30 0 13 4 7 4 7 14 0 10 6 0 18 17 25 5 0 10 5 27 0 16 5 0 7 5 8 7 0 0 0

Meldorf Bay / Elbe Estuary 8 16 11 26 14 27 23 6 0 13 33 26 29 13 14 8 5 22 27 25 21 41 23 18 57 31 24 6 10 11 12 8

East

- Tuemlau Bight 2 22 0 0 22 12 24 0 0 20 26 22 13 0 22 0 0 22 11 0 0 27 13 31 20 42 38 0 0 0 0 15

North Sea Sea North North Frisia Wadden Sea 18 6 14 29 14 49 17 22 29 11 30 23 27 15 35 15 10 27 11 26 19 24 34 26 53 38 36 16 36 14 24 6

Sylt 6 12 19 19 45 47 11 11 0 0 18 27 30 16 14 15 5 25 0 26 12 25 26 25 0 23 41 14 26 19 49 0

Jutland Bank 2 29 0 45 0 0 0 0 0 0 21 0 0 0 34 24 27 21 0 19 12 0 43 0 0 0 0 0 0 0 0 0

Skagerrak 4 14 13 7 20 21 32 5 0 15 36 37 38 9 18 5 0 41 27 24 25 0 57 26 15 27 25 8 13 13 16 68

Kattegat 16 13 13 26 32 20 15 9 0 13 17 17 29 11 24 19 16 23 13 34 26 43 57 15 13 17 19 8 11 13 13 51

Flensburg Firth 14 8 58 26 31 51 22 25 30 12 29 33 22 10 24 6 5 18 31 26 25 0 26 15 12 47 41 15 50 56 18 7

Schlei Estuary 1 24 0 0 12 58 13 0 0 20 13 12 13 0 13 0 0 57 20 53 0 0 15 13 12 62 54 0 0 0 0 15

Eckernförde Bay 6 3 12 23 24 59 44 28 32 15 26 50 23 10 26 4 7 31 42 38 23 0 27 17 47 62 61 42 50 12 14 10

Kiel Bight 25 1 31 32 22 62 34 25 29 54 31 34 34 19 26 7 5 24 38 36 41 0 25 19 41 54 61 38 55 58 51 8

Lubeck Bight 2 19 15 42 8 19 38 41 15 0 15 44 13 15 15 8 8 6 0 16 14 0 8 8 15 0 42 38 48 15 11 0

Baltic Sea Baltic Fehmarn Belt 11 7 36 38 24 49 29 39 31 0 29 57 32 14 16 8 7 10 0 36 26 0 13 11 50 0 50 55 48 59 16 0

Rostock 2 17 20 14 13 62 13 13 0 0 57 16 18 13 13 13 0 11 0 14 19 0 13 13 56 0 12 58 15 59 16 0

Darß-Zingst North 5 23 16 10 27 26 9 9 0 0 9 17 21 9 11 9 0 12 0 24 49 0 16 13 18 0 14 51 11 16 16 0

Rügen North 4 31 0 0 7 7 8 0 0 15 8 7 8 6 8 0 0 8 15 6 0 0 68 51 7 15 10 8 0 0 0 0

- 101 - CHAPTER III

FIGURE S-6: Genotype distribution of V. parahaemolyticus strains based on Table-S2.

- 102 - CHAPTER III

TABLE S-3: Genotype co-occurrence similarity matrix of V. vulnificus populations.

North Sea Baltic Sea

line

-

rth

line

-

Rerik

- n

coast

Binz

Zingst West West Zingst

-

stuaryWremen North Zingst

-

Lubmin

Ranking Kattegat Dranske

-

Zingst Bodden Zingst

Skagerrak

East Frisia Frisia East

-

LubeckBay

EmsEstuary

ElbeEstuary

UsedomNo

Pepelow

Bay ofBay Wismar

Kiel

Darß

Rostockcoast

Darß

JasmunderBodden

GreifswaldBodden

Darß

Weser E Weser

Weser Estuary Weser Dedesdorf

Weser Estuary Weser Bremerhaven

Ems Estuary 6 15 20 50 38 42 30 37 20 9 0 14 0 27 18 33 0 14 44 20 36 38 38

East Frisia 1 10 20 66 68 61 18 23 18 0 0 47 0 61 39 16 0 47 51 71 33 72 68

Weser Estuary Dedesdorf 24 1 50 66 62 64 37 35 18 50 15 57 15 44 49 59 18 48 59 50 45 61 68 Weser Estuary Bremerhaven 33 11 38 68 62 58 14 38 14 10 15 50 15 41 40 26 8 50 43 49 23 57

NorthSea 58 Weser Estuary Wremen 21 5 42 61 64 58 29 20 15 19 16 48 16 48 51 46 9 48 39 48 53 51 59

Elbe Estuary 3 21 30 18 37 14 29 31 30 0 0 11 0 23 8 7 0 11 48 28 5 27 34 Skagerrak 3 16 37 23 35 38 20 31 44 0 0 16 0 28 12 12 0 16 46 39 10 46 45

Kattegat 3 22 20 18 18 14 15 30 44 0 0 11 0 23 8 14 0 11 36 35 5 36 34

Kiel - coast-line 3 17 9 0 50 10 19 0 0 0 23 37 23 16 20 57 37 16 0 10 48 0 11 Lubeck Bay 7 18 0 0 15 15 16 0 0 0 23 20 97 46 17 19 20 47 0 34 15 0 42

Bay of Wismar 2 3 14 47 57 50 48 11 16 11 37 20 20 49 48 51 50 64 51 56 47 50 54 Pepelow-Rerik 5 19 0 0 15 15 16 0 0 0 23 97 20 42 17 19 20 47 0 34 15 0 41 Rostock coast-line 10 7 27 61 44 41 48 23 28 23 16 46 49 42 33 20 14 73 58 71 22 51 62

Darß-Zingst West 4 13 18 39 49 40 51 8 12 8 20 17 48 17 33 21 10 48 38 40 46 37 38

Darß-Zingst North 5 14 33 16 59 26 46 7 12 14 57 19 51 19 20 21 41 22 16 18 46 20 37

Darß-Zingst Bodden 2 20 0 0 18 8 9 0 0 0 37 20 50 20 14 10 41 14 0 8 59 0 9

Baltic Sea Baltic Dranske 2 6 14 47 48 50 48 11 16 11 16 47 64 47 73 48 22 14 51 78 37 50 64

Jasmunder Bodden 3 9 44 51 59 43 39 48 46 36 0 0 51 0 58 38 16 0 51 62 29 76 78

Binz 10 4 20 71 50 49 48 28 39 35 10 34 56 34 71 40 18 8 78 62 29 64 61

Greifswald Bodden 16 12 36 33 45 23 53 5 10 5 48 15 47 15 22 46 46 59 37 29 29 28 27

Lubmin 7 8 38 72 61 57 51 27 46 36 0 0 50 0 51 37 20 0 50 76 64 28 74

Usedom North 11 2 38 68 68 58 59 34 45 34 11 42 54 41 62 38 37 9 64 78 61 27 74

- 103 - CHAPTER III

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Ems Estuary (n=6) 1 3 5 East Frisia (n=1) 5 Weser Estuary Dedesdorf (n=24) 1 3 4 5 7 11 Weser Estuary Bremerhaven (n=33) 2 3 4 5 6 8 Weser Estuary Wremen (n=21) 1 3 4 5 6 9 Elbe Estuary (n=3) 5 9 11 Skagerrak (n=3) 3 5 Kattegat (n=3) 5 9 10 Kiel - coast-line (n=3) 1 4 Lubeck Bay (n=7) 4 Bay of Wismar (n=2) 4 5 Pepelow-Rerik (n=5) 4 Rostock coast-line (n=10) 4 5 9 Darß-Zingst West (n=4) 1 4 5 14 Darß-Zingst Nord (n=5) 1 4 5 10 Darß-Zingst Bodden (n=2) 4 13 Dranske (n=2) 4 5 Jasmunder Bodden (n=3) 5 Binz (n=10) 4 5 9 12 Greifswald Bodden (n=16) 1 4 5 13 Lubmin (n=7) 5 12 Usedom North (n=11) 4 5

80% similarity V. vulnificus cut-off genotype (n=14) distribution among sampling areas (n=22) of ERIC-PCR profiles (n=181)

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Ems Estuary (n=6) 1 8 17 20 21 East Frisia (n=1) 18 Weser Estuary Dedesdorf (n=24) 1 3 7 14 15 18 20 21 25 32 Weser Estuary Bremerhaven (n=33) 4 5 6 7 8 14 16 18 19 21 22 23 24 26 Weser Estuary Wremen (n=21) 1 2 6 9 10 12 18 21 24 29 Elbe Estuary (n=3) 20 29 32 Skagerrak (n=3) 5 20 Kattegat (n=3) 20 27 30 Kiel - coast-line (n=3) 3 15 Lubeck Bay (n=7) 11 Bay of Wismar (n=2) 15 18 Pepelow-Rerik (n=5) 11 Rostock coast-line (n=10) 9 11 12 17 18 20 28 Darß-Zingst West (n=4) 2 14 18 37 Darß-Zingst Nord (n=5) 1 15 21 31 Darß-Zingst Bodden (n=2) 15 35 Dranske (n=2) 11 18 Jasmunder Bodden (n=3) 18 20 Binz (n=10) 11 18 20 27 28 34 Greifswald Bodden (n=16) 1 2 13 15 18 35 36 Lubmin (n=7) 18 20 21 33 Usedom North (n=11) 11 12 18 20 21

85% similarity V. vulnificus cut-off genotype (n=37) distribution among sampling areas (n=22) of ERIC-PCR profiles (n=181)

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Ems Estuary (n=6) 1 14 31 37 43 East Frisia (n=1) 33 Weser Estuary Dedesdorf (n=24) 1 2 7 12 25 27 28 33 34 36 40 41 42 49 57 Weser Estuary Bremerhaven (n=33) 8 9 11 13 14 25 26 30 33 34 35 40 41 43 44 45 46 48 50 Weser Estuary Wremen (n=21) 2 3 4 5 11 16 17 22 33 34 40 42 43 47 54 Elbe Estuary (n=3) 37 54 57 Skagerrak (n=3) 9 38 39 Kattegat (n=3) 38 51 55 Kiel - coast-line (n=3) 6 28 Lubeck Bay (n=7) 18 19 21 Bay of Wismar (n=2) 29 34 Pepelow-Rerik (n=5) 18 19 Rostock coast-line (n=10) 15 20 21 22 31 33 34 36 53 Darß-Zingst West (n=4) 5 27 34 64 Darß-Zingst Nord (n=5) 2 28 29 42 56 Darß-Zingst Bodden (n=2) 29 61 Dranske (n=2) 20 34 Jasmunder Bodden (n=3) 34 36 37 Binz (n=10) 20 33 34 39 52 53 60 Greifswald Bodden (n=16) 1 2 3 5 24 28 34 61 62 63 Lubmin (n=7) 33 34 36 38 43 59 Usedom North (n=11) 18 23 33 34 36 37 38 40 42 88% similarity V. vulnificus cut-off genotype (n=64) distribution among sampling areas (n=22) of ERIC-PCR profiles (n=181) FIGURE S-3: Genotype distribution of V. vulnificus strains based on Table-S3.

- 104 - CHAPTER III

replicates

-

ERIC

between

similarity

minimum

V. cholerae V. parahaemolyticus V. vulnificus

FIGURE S-4: Box-plot with the ERIC-PCR reliability values for 120 strains based on the minimum ERIC-profile similarity of three replicates. 100%

97.0%

94.0%

91.0% 89.5% 88.0% 86.5% 85.0%

offs - 82.5%

cut 80.0%

similarity 75.0%

70.0% V. cholerae V. parahaemolyticus V. vulnificus

60.0% 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 genotypes / profiles

FIGURE S-5: Genotype to profile ratios for each putative similarity cut- offs and species. Triangles represent the genotype / profile values for the putative similarity cut-offs and are connected with solid lines. The ratios of the chosen cut-offs are marked with broken lines.

- 105 - CHAPTER III

82.5% 100 85% 70% 75% 80% 90 86.5% 88% 80 89.5% 70 91% 60

50

reliability 40 94% 30 20 97% 10 V. cholerae 100% 0 0 10 20 30 40 50 60 70 80 90 100

index of discrimination

82.5% 100 85% 60% 70% 75% 80% 86.5% 90 88% 89.5% 80 70 91%

60

50

reliability 40 94% 30 20

10 97% V. parahaemolyticus 0 100% 0 10 20 30 40 50 60 70 80 90 100 index of discrimination

80% 82.5% 85% 100 86.5% 70% 75% 90 88% 80 89.5% 70

91% 60 50

reliability 40 94% 30

20

10 97% V. vulnificus 0 100% 0 10 20 30 40 50 60 70 80 90 100

index of discrimination FIGURE S-6: Reliability values and indices of discrimination for all putative similarity cut-offs. The choosen cut-offs (80%,85%,88%) are marked red.

- 106 - CHAPTER III

FIGURE S-6: similarity related genotypes

85 90

80 88

95 75 100 80% 85% 88% VN-2923 VN-3503 1 VN-3361 1 VN-3923 VN-3903 2 VN-0247 VN-3908 3 VN-3913 2 VN-3955 4 VN-3472 VN-3492 VN-3470 5 VN-3917 VN-3901 1 VN-3469 3 VN-3012 VN-3475 VN-3471 6 VN-3460 VN-3918 VN-3911 7 VN-3405 4 8 VN-4051 VN-3952 5 9 VN-3953 6 10 VN-3949 7 11 VN-3940 VN-3963 2 8 12 VN-3301 13 9 VN-3941 3 VN-3938 14 VN-3951 10 15 VN-10012 VN-10013 VN-3907 16 VN-0244 11 VN-0252 VN-3428 17 VN-3902 VN-0246 VN-0257 VN-3958 VN-3377 VN-2808 18 VN-3944 VN-2995 VN-3954 4 VN-0278 12 VN-3943 VN-2997 VN-3936 19 VN-3016 VN-2816 20 VN-3950 VN-2825 21 VN-3407 VN-3916 13 22 VN-3939 23 VN-3956 24 VN-3957 14 VN-3465 25 26 VN-3132 5 15 VN-3942 27 VN-4052 6 16 28 VN-3967 7 17 29

Reliability values and indices of discrimination for all putative similarity cut-offs. The choosen cut-offs (80%,85%,88%) are marked red. FIGURE S-7: Cluster analysis and genotyping of all V. cholerae ERIC-PCR profiles. Similarity cut-offs at 80%, 85% and 88% generate 7,17 and 29 genotypes respectlivly.

- 107 - CHAPTER III

TABLE 3: Comparison between ERIC-profiles from the North Sea and the Baltic Sea. The number of genotypes and profiles are listed which are present and exclusive for the North Sea and Baltic Sea overall population.

V. cholerae V. parahaemolyticus V. vulnificus North Sea Baltic Sea North Sea Baltic Sea North Sea Baltic Sea total profiles 62 185 175 profiles 23 (37%) 39 (63%) 115 (62%) 70 (38%) 88 (50%) 87 (50%) 80% similarity cut-off total genotypes 6 21 14 present genotypes 3 (50%) 6 (100%) 19 (90%) 12 (57%) 10 (71%) 8 (57%) exclusive genotypes 0 3 (50%) 9 (43%) 2 (10%) 6 (43%) 4 (29%) profiles in shared genotypes 58 (94%) 157 (85%) 149 (85%) 85% similarity cut-off total genotypes 16 58 36 present genotypes 7 (44%) 15 (94%) 44 (76%) 29 (50%) 26 (72%) 21 (58%) exclusive genotypes 1 (6%) 9 (56%) 29 (50%) 14 (24%) 15 (42%) 10 (28%) profiles in shared genotypes 48 (77%) 104 (56%) 117 (67%) 88% similarity cut-off total genotypes 27 85 59 present genotypes 12 (44%) 24 (89%) 65 (76%) 36 (42%) 39 (66%) 35 (59%) exclusive genotypes 3 (11%) 15 (56%) 49 (58%) 20 (24%) 24 (41%) 20 (34%) profiles in shared genotypes 41 (66%) 80 (43%) 98 (56%) profiles in exclusive genotypes 4 (7%) 17 (27%) 79 (43%) 26 (14%) 35 (20%) 42 (24%) Overall indices shared genotypes 40% 30% 28% exclusive genotypes 6% 54% 50% 20% 42% 30% profiles in shared genotypes 79% 61% 69% profiles in exclusive genotypes 3% 18% 30% 9% 15% 16%

- 108 - CHAPTER III

100

90

80

70

60

(%)

offs

- 50

cut

40

similarity

30

20 V. cholerae

V. parahaemolyticus 10 V. vulnificus

0

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 clusters / site-specific populations

FIGURE S-8: Similarity cut-offs and the according ratios between clusters and populations for the cluster analysis of site specific populations. A polynomial regression was performed: y=-0.0639x2+107.09x-12.536 and R2= 0.9891 for V. cholerae, y=-8.5998x2+78.673x+0.7041and R2=0.9911 for V. parahaemolyticus and y=- 26,689x2+101.4x +6.0288and R2=0,9936 for V. vulnificus. A cluster to population ratio of 0.5 had to be achieved. Therefore, a broken line marks the according similarity cut-offs at 40% (V. cholerae, V. parahaemolyticus) and 50% (V. vulnificus).

- 109 - GENERAL DISCUSSION

- 110 - GENERAL DISCUSSION

GENERAL DISCUSSION

The advances in medical science and sanitation during the 20th century lead to an immense decrease of severe bacterial infections. Particular in the western “developed world”, infectious diseases are rather associated with historical pandemic events than with actual health risks (Kass 1987). However, the widespread application of antibiotics favours the proliferation of multiple-resistent bacterial strains (Baquero et al. 2008). Furthermore, the globalized traffic and global warming enables pathogens to reach new habitats (Harvell et al. 2002; Burge et al. 2014). These anthropogenic influences lead to an increase of infections. In consequence, the surveillance of already known “old” pathogenic species and the risk evaluation of emerging “new” infectious diseases will be a major challenge for national and international health authorities.

Potentially pathogenic Vibrio spp. comprises both: already known pathogens such as O1/O139 V. cholerae strains as the causative agent for many severe Cholera pandemics (Barua 1992) and newly emerging diseases caused by V. parahaemolyticus, V. vulnificus as well as non-O1/non-O139 V. cholerae strains (Baker-Austin et al. 2010; Lindgren et al. 2012). Consistent with other clinical important species, virulence-related genes and genes for antiobiotic resistance can be transferred between Vibrio spp. strains (Beaber et al. 2002). Furthermore, climate changes triggered by global warming increase the abundance of potentially pathogenic Vibrio spp. in the marine environment (Colwell 1996; Paz et al. 2007).

In consequence, studies about the impact of Vibrio spp. are urgently needed, particularly the development of methods for the identification of potentially pathogenic species in environmental samples and investigations about the distribution of Vibrio spp. populations. This PhD thesis contributes both: the evaluation of whole-cell MALDI-TOF MS as an accurate tool for species identification and the ERIC-PCR genotyping based analysis of Vibrio spp. assemblages in the North and Baltic Seas.

- 111 - GENERAL DISCUSSION

Accuracy, discriminatory power and applicability of whole-cell MALDI-TOF MS

Bruker BiotyperTM is a widely distributed whole-cell MALDI-TOF MS species identification system. The discrimination of species relies on different molecular masses of abundant cell proteins that is reflected by respective mass-to-charge ratio difference of peaks in mass spectra. A large reference database is provided by the manufacturer Bruker, allowing the calculation of an identfication score between acquired spectra and reference main spectra entries of the BiotyperTM database (Maier et al. 2007; Sauer et al. 2008). This system was initially developed as clinical diagnostic tool to allow the fast and accurate identification of pathogens. Thus, the BiotyperTM database mainly consists of spectra that were generated from infectious bacterial species. Regarding Vibrio spp., this database is sufficient for the discrimination of the three main pathogenic species from clinical samples. However, environmental samples include a larger range of different Vibrio species: potentially pathogenic species as well as weak pathogens and non-human pathogenic species. Thus, VibrioBase was developed in the course of this PhD thesis, comprising 20-fold more reference spectra (997 entries) from Vibrio spp. isolates than the BiotyperTM database (49 Vibrio spp. entries). This VibrioBase database was checked for accuracy using the following three criteria:

(1) Number of species identification agreements between MALDI-TOF MS / VibrioBase and the approved DNA-based rpoB sequence analysis

(2) Identification score differences between the BiotyperTM database and VibrioBase

(3) Discriminatory potential for closely related Vibrio spp.

Based on the rpoB sequence analysis as reference method, MALDI-TOF MS / VibrioBase revealed Vibrio spp. identification agreements of 96.9%. Even higher species agreements were achieved for the three potentially pathogenic species: 100% for V. cholerae, 98.2% for V. parahaemolyticus and 98.4% for V. vulnificus. Several studies were conducted to compare MALDI-TOF MS with other species identification methods such as phenotypic-based (API) and DNA-sequence based methods (16S rRNA, gyrB); with respective species identification agreements ranging between 93% and 99% (Table 1). Thus, the results of this study are consistent with the results of previous research and the whole-cell MALDI-TOF MS method can be regarded to as accurate tool for potentially pathogenic Vibrio spp. identifications at the species level.

- 112 - GENERAL DISCUSSION

Based on the Bruker identification score, results are divided into “reliable identifications” (log-score > 2.0) and “highly reliable identifications” (log-score > 2.3). The assignment as “highly reliable identification” is linked to the respective size of the reference database. The results of this PhD thesis reveal that the BiotyperTM database does not encompass the genetic variability of certain Vibrio species. In contrast, VibrioBase revealed a log-score increase between 0.2 and 0.4 for potentially pathogenic species. Thus, alignments with VibrioBase lead to more highly probable species identifications than alignments with the BiotyperTM database. These values are consistent with the study of Christensen et al. (2012): using an extended database of Gram-postive bacteria, 3.6-fold more “highly probable species identifications” are reported with score increases ranging from 0.038 to 0.527. In consequence, VibrioBase is a useful specialised database for Vibrio spp. within the Bruker BiotyperTM system because it provides a high quality of species identification results.

TABLE 1: Studies regarding the comparison between reference methods (gyrB and 16S rRNA sequence analysis, API - analyte profile index) and MALDI-TOF MS systems (Bruker BiotyperTM, BioMerieux Saramis).

reference MALDI-TOF species respective organisms reference method system agreement

Aeromonas spp. gyrB Saramis 93% Benagli (2012)

16S Lactobacillus spp. Biotyper 93% Callaway (2013) rRNA

Enterococcus spp. PCR Biotyper 99% Griffin (2012)

acetic acid bacteria 16S (Acetobacter, Gluconacetobacter, Saramis 93% Andres-Barrao (2012) rRNA Gluconobacter)

clinical relevant bacteria (Escherichia coli, Pseudomoas, API Saramis 95% Benagli (2011) Staphylococcus)

gram positive and gram negative 16S Biotyper 97% Khot (2012) bacteria (Streptococcus, Bordetella) rRNA

Biotyper 96% clinical relevant yeast reviewed by Bader API (Candida, Aspergillus) (2012) Saramis 98%

- 113 - GENERAL DISCUSSION

Potentially pathogenic Vibrio spp. reveal a close relationship towards weakly pathogenic species: V. cholerae to V. mimicus, V. parahaemolyticus to V. alginolyticus and V. vulnificus to V. navarrensis. Therefore, MALDI-TOF MS has to distinguish these species pairs to avoid less reliable or even false positive species identification results. According to the cluster analysis of the MALDI-TOF MS variability test (Excursus I), a similarity threshold of 20% reflects the Bruker scores for “highly reliable identifications” (log-score 2.3). The similarity value for the two species pairs V. parahaemolyticus / V. alginolyticus (17.4%) and V. vulnificus / V. navarrensis (16.3%) drops below this threshold. Thus, these species can be clearly distinguished from each other. The species pair V. cholerae / V. mimicus, that reveals an almost identical 16S rRNA sequence, exceeded the MALDI-TOF threshold with a mean similarity value of 21.2%. In consequence, Bruker thresholds might be adapted as suggested by Bader (2012) for yeasts of clinical importance. Particularly for V. cholerae, an increased threshold of 25% (log-score 2.4) would lead to identification results of higher discriminatory potential towards closely related species such as V. mimicus. In contrast, a clear separation was not achieved for two species pairs V. harveyi / V. campbellii and V. fluvialis / V. furnissii. So far, these weakly pathogenic species can only be distinguished using DNA-sequence based methods. The same applies to non-human pathogenic species of the V. splendidus group. Apart from this exception, whole-cell MALDI-TOF MS reveals a phylogenetic resolution for potentially pathogenic Vibrio spp. comparable to rpoB sequence analysis and it achieves even more accurate identification results than 16S rRNA sequence analysis.

In contrast to the majority of slow-growing marine bacteria, mesophilic Vibrio spp. show fast growth at 37°C allowing the harvest of biomass within the required exponential growth phase for MALDI-TOF MS measurements after only one day of cultivation. Hence, whole-cell MALDI-TOF MS is naturally well-suited for species identifications of Vibrio spp.. Furthermore, surveillance programs demand a high throughput of samples, which can also be provided by this proteom-based method. A respective application of MALDI-TOF MS / VibrioBase was successfully accompanied in the course of a surveillance program (Excursus II): 84% of the analyzed colony forming units achieved a Bruker identification score of at least 2.0 (reliable identification) and 25% were assigned to Vibrio species. In conclusion, whole-cell MALDI-TOF MS in combination with VibrioBase can be regarded to as tool for accurate species identification for potentially pathogenic Vibrio spp. from environmental samples and, therefore, meets all requirements for an implementation in Vibrio surveillance programs carried out by health authorities.

- 114 - GENERAL DISCUSSION

Mass peaks for Vibrio spp. identification – the SIBOPS approach

Commercially available whole-cell MALDI-TOF MS species identification systems rely on two factors: the absence / presence of certain peaks and the peak intensity. It is therefore evident to screen mass spectrometric data from VibrioBase for species-associated peaks; in particular peaks exceeding the defined threshold of 80% VibrioBase species sensitivity for “sensitive peaks”. Between 12 and 38 sensitive peaks were found for each species and several sensitive peaks are shared between species (Figure 1). Based on these peaks, the SIBOPS (species identification based on peak sensitivities) method was developed. It relies on the alignment between mass spectrometry peaks of acquired spectra and sensitive peaks. An agreement between two peaks is reached if the respective mass-to-charge ratio difference is below 1000ppm. The resulting SIBOPS score represents the ratio between detected sensitive peaks of the acquired spectra and the total number of sensitive peaks (Figure 1).

peak table VibrioBase data: peak sensitivity value >80% to Vibrio spp. (m/z) species peaks species peaks 2133 2137 V. parahaemolyticus 30 V. mimicus 34 2165 V. alginolyticus 26 V. pacinii 27 2396 V. harveyi 12 V. diazotrophicus 18 2589 V. fluvialis 31 V. splendidus 14 2966 V. vulnificus 25 V. aestuarianus 25 3025 V. navarrenis 21 L. anguillarum 13 3083 V. cholerae 38 Shewanella spp. 22 3155 3203 3210 total detected ratio (SIBOPS 3226 peaks peaks score) 3310 1) V. parahaemolyticus 30 24 80.0 3361 2) V. harveyi 12 8 66.7 3389 3) V. alginolyticus 26 17 65.4 3433 3597 4) V. vulnificus 25 11 44.0 … 5) V. pacinii 27 11 40.7 total 88 peaks

FIGURE 1: Principle of SIBOPS. Peak tables with mass-to-charge ratios (m/z) are generated from acquired spectra. The alignment between these peaks and peaks sensitive for Vibrio spp. leads to SIBOPS scores and the Vibrio species with the highest scores are displayed in a matching score table.

- 115 - GENERAL DISCUSSION

198 Vibrio spp. strains were identified using both systems: SIBOPS and the combination of BiotyperTM and VibrioBase. In the course of this evaluation, 98% identifications were consistent with the best matching result of the BiotyperTM / VibrioBase system. Two thresholds were determined for “highly reliable identifications”: a general SIBOPS score of 77.5 and a SIBOPS score of 67.0 combined with a minimal distance to the 2nd matching result of 10.0. These thresholds are consistent with the Bruker BiotyperTM identification log-score threshold of 2.3.

In conclusion, SIBOPS achieves accurate identification results comparable to the Bruker BiotyperTM system. The most advantage of SIBOBS is transparency. Sensitive peaks can be re-evaluated indepently from the respective mass spectrometry eqipment and further sensitive peaks for other species can be included. The simplified matching score algorithm can be adapted and allows identifications with common software tools such as Microsoft Excel or Libre Office. Thus, SIBOPS does not entail additional costs by frequent software updates demanded by the manufacturers of commercially available systems. This point is of particular importance because federal and local health authorities have to deal with limited budgets and, therefore, SIBOPS increases the chance of implementation of MALDI-TOF MS in surveillance programs. In conclusion, the SIBOPS method provides an open access to whole- cell mass spectrometry data and the overriding objective should be the development of a worldwide MALDI-TOF MS database, comparable to GenBank for nucleotide and amino acid sequences.

Biomarker based species identification (BIBSI)

The largest limitation of the MALDI-TOF MS method is the requirement of mono-microbial cultures for species identification. Several studies, however, revealed the usefulness of MALDI-TOF MS for the direct identification of pathogenic species in clinical samples; particularly in bacteria-free fluids that are dominated by single bacterial species in the course of infections, e.g. blood cultures (Ferreira et al., 2010; Ferreira et al., 2011). In this context, the use of alkalinie peptone water as enrichment medium and a growth temperature of 37°C enables the exclusion of commonly slow-growing marine bacteria in the course the Vibrio spp. isolation process (Martinez-Urtaza et al. 2011). Thus, mass spectra can be acquired from samples with a limited range of Vibrio species and species might be directly identified using BIBSI (Figure 2).

- 116 - GENERAL DISCUSSION

environmental samples (water, plankton) FIGURE 2: Principle of BIBSI.

Vibrio spp. from environmental enrichment in alkaline peptone water samples are enriched using alkaline peptone water. Spectra are acquired sample preparation using the total biomass from the total biomass of the enriched samples. These spectra are screened for biomarkers that allow a generation of MALDI-TOF MS spectra direct species identification.

Biomarker Based Species Identification (BIBSI)

As with sensitive peaks for SIBOPS, potential biomarkers for Vibrio spp. were found based on the mass spectrometry data of VibrioBase. Regarding exclusivity, these peaks were evaluated using mono-microbial and bi-microbial cultures. 12 of these peaks passed this evaluation and are considered as biomarkers: three for each, V. cholerae and V. vulnificus, and six for V. parahaemolyticus. However, since these peaks were only tested for exclusivity against Vibrio spp. and the outgroup Shewanella, the applicability of these biomarkers for BIBSI is not guaranteed. Due to software restrictions, an alignment to the reference spectra of other species from the BiotyperTM database could not be executed; a requirement for accurate biomarker based species identifications in environmental samples. This limitation shows again the need for a freely accessible whole-cell MALDI-TOF database to allow the evaluation of biomarkers across bacterial species and genera. Furthermore, the number of different species included in environmental samples is negatively correlated with the peak intensities of biomarkers. In consequence, mass peak intensities fall below the noise-to-signal- ratio in spectra of mixed samples and biomarkers can not be detected anymore. Therefore, technical improvements are needed, particularly a more efficient MALDI-TOF MS ionization process leading to higher peak intensities. In conclusion, BIBSI and the direct MALDI-TOF- based identification of bacteria from environmental samples is a promising approach, which, however, needs further improvements for an application in surveillance programs.

Mass peaks for intraspecific typing

To some extent, whole-cell MALDI-TOF MS analysis can be used to identify mass peaks that correlate to intraspecific groups. Particularly in the field of clinical microbiology, pathotypes of species could be distinguished from each other and peaks were identified that are closely

- 117 - GENERAL DISCUSSION linked to antibiotic resistance (Table 2). Similar investigations were conducted using the mass spectrometry data of VibrioBase. Mass peaks were found allowing the differentiation of two distinct V. alginolyticus groups, that were initially described using rpoB sequence analysis by Oberbeckmann et al. (2011). These mass peaks represent the 30S ribosomal protein S14; with a sensitivity of 82% to 90% and an exclusivity of 93% to 99% towards both V. alginolyticus groups. Regarding V. vulnificus, several mass peaks were found by Bier et al. (2013) that correlate to highly-virulent multi-locus-sequence-types. For V. parahaemolyticus, no peaks were found so far, that are linked to trh-positive or tdh-positive strains respectively. The same applies to V. cholerae; no mass peaks correlate to the Cholera-associated serotypes O1/O139. Nevertheless, as already mentioned above for the BIBSI approach, technical improvements could enable the differentiation between strains of higher and lower virulence. Thus, the screening for these peaks might be also included in future MALDI-TOF MS-based Vibrio surveillance programs.

TABLE 2: Studies regarding the use of mass peaks for intraspecific typing.

species group intraspecific differentiation reference

Vibrio vulnificus multi-locus sequence types Bier (2013)

Escherichia coli pathotypes (STEC, EHEC, EPEC, EIEC) Clark (2012) Escherichia coli growth phases (exponential, stationary, decline) Momo (2013)

Enterococcus Vancomycin resistant strains Griffin (2013)

Staphylococcus Methicillin resistant strains Wolters (2011)

Neisseria meningitis sequence types and clonal complexes Suarez (2013) Campylobacter jejuni multi-locus sequence types Zautner (2013)

Yersinia enterocolitica biotypes Stephan (2011)

Clostridium difficile hyper-virulent ribotypes Reil (2011)

Distribution patterns of potentially pathogenic species in the North and Baltic Sea

Dispersal mechanisms and population structures of Vibrio spp. are poorly understood in the northern part of Europe. Hence, the results of the ERIC-PCR genotyping provide first information on the biogeography of three potentially pathogenic Vibrio spp. in the North and Baltic Seas. Only 2.6% of all strains were classified as geographically restricted genotypes. Thus, each site-specific population reflects a diverse composition of, to some extent,

- 118 - GENERAL DISCUSSION frequently common genotypes. These findings correspond to those of Keymer et al.: they found only 4% of geographically restricted genotypes for V. cholerae strains in California (Keymer et al. 2009). Regarding diversity, V. cholerae populations from the Baltic Sea are more diverse than those from the North Sea. In contrast, the genotypic diversity within each site-specific population of V. parahaemolyticus and V. vulnificus is largely the same throughout the North and Baltic Seas.

Despite the absence of locally restricted genotypes, the biogeographic analyses of genotype co-occurrences revealed important differences as well as common features of site-specific populations. Particularly striking is the separation of V. parahaemolyticus and V. vulnificus populations into different regions. V. parahaemolyticus populations are separated into the three regions southern North Sea, offshore areas of the North Sea and eastern North Sea / Baltic Sea whereas V. vulnificus populations are separated into the two regions North Sea / eastern Baltic Sea and western Baltic Sea.

These separations are not consistent with geographical aspects and allow no general genotypic differentiation between Vibrio spp. assemblages of the North Sea and Baltic Sea. Environmental parameters such as salinity are not associated with single genotypes and explain these separations only to a minor part. However, several studies show that Vibrio spp. strains are frequently attached to plankton and coherences were observed between site- specific Vibrio populations and the dominant plankton species at the respective sampling sites (Huq et al. 1983; de Magny et al. 2011; Kirschner et al. 2011). Thus, the herein discovered higher genotype richness of V. cholerae strains in the Baltic Sea and the biogeographic separation of V. parahaemolyticus and V. vulnificus might be linked to the attachment to more different plankton species.

During the last two decades, all clinical V. vulnificus cases at the German Baltic Sea were reported from the islands of Rügen and Usedom. Site-specific populations of these regions are closely related with each other and are separated from other populations in the western Baltic Sea. Thus, ERIC-PCR genotyping reflects the higher infection risk for people in Rügen and Usedom. Furthermore, populations of the Weser Estuary are closely linked to these assemblages from the eastern Baltic Sea. Hence, possible future V. vulnficus infections in the North Sea might be expected from this area close to the cities of Bremen an Bremerhaven.

- 119 - (Callaway et al. 2013)

(Callaway et al. 2013) (Benagli et al. 2011) (Clark et al. 2013) (Suarez et al. 2013) (Stephan et al. 2011) (Reil et al. 2011)

- 120 - OUTLOOK

OUTLOOK

Due to the global-warming-driven increase of heat waves, mesophilic potentially pathogenic Vibrio spp. are gaining in importance, particular in western and northern Europe (Lindgren et al. 2012). Thus, the society has to deal with the consequences of increased life-threatening Vibrio spp. associated diseases. Surveillance programs are important contributions risk evaluation reports for local beaches and, therefore, can reduce or even prevent such infections. Whole-cell MALDI-TOF MS combined with VibrioBase is a suitable method for these Vibrio surveillance programs; particularly because of the high-throughput of samples, the fast identification process and the high accuracy of identification results. The following suggested improvements will increase the efficiency of this method:

(1) A even higher accuracy of identification results can be achieved by an extension of VibrioBase, particular by reference spectra from weak pathogens or other Vibrio spp. that are closely related to potentially human pathogenic species such as V. natriegens, V. harveyi, V. metschnikovii and V. furnissii.

(2) Since more entries for each species lead to increased Bruker identification scores, threshold for “highly reliable identifications” should be raised to a log-score of at least 2.4, particular for the three major potential pathogenic species V. cholerae, V. parahaemolyticus and V. vulnificus.

(3) Reference spectra from databases and mass spectrometry data from Vibrio surveillance programs are to be filed in web-based platforms to allow an exchange of information as well as further validations of the whole-cell MALDI-TOF MS method for environmental studies.

Besides Vibrio spp., other specific reference databases can be developed, e.g. for non- pathogenic marine bacteria. Furthermore, MALDI-TOF MS is a promising tool to identify plankton species such as copepods.

The herein developed SIBOPS method is an important step towards the suggested exchange of mass spectrometry data. Further Vibrio spp. spectra acquired with the commercially systems of Bruker and BioMerieux should be simultaneously identified using the SIBOPS

- 121 - OUTLOOK approach. Therefore, sensitive peaks linked to each species can be re-evaluated and will promote acceptance of this optional identification method. Furthermore, more sensitive peaks are to be included in the SIBOPS method; those linked to Vibrio spp. as well as those linked to other bacterial species. A major challenge is the adjustment of detection limits between mass-to-charge values of peaks of acquired spectra and the respective values of sensitive peaks. These thresholds are to be adapted, for certain species and even for single sensitive peaks. In consequence, sensitive peaks with only minor mass-to-charge ratio differences will be better discriminated; leading to a higher accuracy of peak alignments and SIBOPS species identification results.

In contrast to the SIBOPS approach, BIBSI needs more improvements for a successful application in surveillance programs, particular open accessible databases for the biomarker screening/evaluation and technical MALDI-TOF MS measurement improvements to achieve a higher resolution of mass spectra. Nevertheless, investigations about the direct MALDI- TOF-based identification in mixed samples are currently being executed by mass spectrometry manufacturer and researchers in the field of clinical microbiology. Thus, it is only a question of time before systems similar to BIBSI will enter the market.

Besides methods for the accurate in-time identifications of potentially pathogenic species, in- deep analysis of Vibrio assemblages are still necessary, in particular to increase knowledge about linkages between environmental parameters and the distribution of certain Vibrio species. Studies about the influence of salinity and temperature were frequently published. Results of the ERIC-PCR genotyping suggest, that other parameters also might shape Vibrio spp. populations in the North and Baltic Sea. Thus, further studies are needed, in particular to analyze the affinity of potentially pathogenic Vibrio spp. to certain plankton species of temperate waters and to investigate the influence of bacteriophages in the marine environment. Such ecological analyses promote the understanding of distribution patterns and, therefore, provide useful information for further risk evaluation studies and Vibrio modeling studies.

- 122 - REFERENCES

REFERENCES

Adekambi, T., M. Drancourt and D. Raoult (2009). The rpoB gene as a tool for clinical microbiologists. Trends in Microbiology 17(1): 37-45. Adekambi, T., T. M. Shinnick, D. Raoult and M. Drancourt (2008). Complete rpoB gene sequencing as a suitable supplement to DNA-DNA hybridization for bacterial species and genus delineation. International Journal of Systematic and Evolutionary Microbiology 58: 1807-1814. Alam, M., N. A. Hasan, A. Sadique, N. A. Bhuiyan, K. U. Ahmed, S. Nusrin, G. B. Nair, A. K. Siddique, R. B. Sack, D. A. Sack, A. Huq and R. R. Colwell (2006). Seasonal cholera caused by Vibrio cholerae serogroups O1 and O139 in the coastal aquatic environment of Bangladesh. Applied and environmental microbiology 72(6): 4096-4104. Alam, M., M. Sultana, G. B. Nair, A. K. Siddique, N. A. Hasan, R. B. Sack, D. A. Sack, K. U. Ahmed, A. Sadique, H. Watanabe, C. J. Grim, A. Huq and R. R. Colwell (2007). Viable but nonculturable Vibrio cholerae O1 in biofilms in the aquatic environment and their role in cholera transmission. Proceedings of the National Academy of Sciences of the United States of America 104(45): 17801-17806. Alatoom, A. A., C. J. Cazanave, S. A. Cunningham, S. M. Ihde and R. Patel (2012). Identification of Non- diphtheriae Corynebacterium by Use of Matrix-Assisted Laser Desorption Ionization-Time of Flight Mass Spectrometry. Journal of clinical microbiology 50(1): 160-163. Alsina, M. and A. R. Blanch (1994). A Set of Keys for Biochemical-Identification of Environmental Vibrio Species. Journal of Applied Bacteriology 76(1): 79-85. Amaro, C., E. G. Biosca, B. Fouz and E. Garay (1992). Electrophoretic analysis of heterogeneous lipopolysaccharides from various strains of Vibrio vulnificus biotypes 1 and 2 by silver staining and immunoblotting. Current Microbiology 25(2): 99-104. Ansede-Bermejo, J., R. G. Gavilan, J. Trinanes, R. T. Espejo and J. Martinez-Urtaza (2010). Origins and colonization history of pandemic Vibrio parahaemolyticus in South America. Molecular ecology 19(18): 3924-3937. Arias, C. R., L. Verdonck, J. Swings, E. Garay and R. Aznar (1997). Intraspecific Differentiation of Vibrio vulnificus Biotypes by Amplified Fragment Length Polymorphism and Ribotyping. Applied and environmental microbiology 63(7): 2600-2606. Asplund, M. E., A. S. Rehnstam-Holm, V. Atnur, P. Raghunath, V. Saravanan, K. Harnstrom, B. Collin, I. Karunasagar and A. Godhe (2011). Water column dynamics of Vibrio in relation to phytoplankton community composition and environmental conditions in a tropical coastal area. Environmental microbiology 13(10): 2738-2751. Austin, B. (2010). Vibrios as causal agents of zoonoses. Veterinary microbiology 140(3-4): 310-317. Aznar, R., W. Ludwig, R. I. Amann and K. H. Schleifer (1994). Sequence Determination of Ribosomal-Rna Genes of Pathogenic Vibrio Species and Whole-Cell Identification of Vibrio-Vulnificus with Ribosomal-Rna-Targeted Oligonucleotide Probes. International Journal of Systematic Bacteriology 44(2): 330-337. Baker-Austin, C., L. Stockley, R. Rangdale and J. Martinez-Urtaza (2010). Environmental occurrence and clinical impact of Vibrio vulnificus and Vibrio parahaemolyticus: a European perspective. Environmental Microbiology Reports 2(1): 7-18. Baker-Austin, C., J. A. Trinanes, N. G. H. Taylor, R. Hartnell, A. Siitonen and J. Martinez-Urtaza (2013). Emerging Vibrio risk at high latitudes in response to ocean warming. Nature Climate Change 3(1): 73- 77. Baquero, F., J. L. Martinez and R. Canton (2008). Antibiotics and antibiotic resistance in water environments. Curr Opin Biotechnol 19(3): 260-265. Barua, D. (1992). History of Cholera. Cholera. D. Barua and W. Greenough, III, Springer US: 1-36. Bauer, A. and L. M. Rorvik (2007). A novel multiplex PCR for the identification of Vibrio parahaemolyticus, Vibrio cholerae and Vibrio vulnificus. Letters in applied microbiology 45(4): 371-375. Beaber, J. W., B. Hochhut and M. K. Waldor (2002). Genomic and functional analyses of SXT, an integrating antibiotic resistance gene transfer element derived from Vibrio cholerae. Journal of Bacteriology 184(15): 4259-4269. Beardsley, C., J. Pernthaler, W. Wosniok and R. Amann (2003). Are Readily Culturable Bacteria in Coastal North Sea Waters Suppressed by Selective Grazing Mortality? Applied and environmental microbiology 69(5): 2624-2630. Belkin, I. M. (2009). Rapid warming of Large Marine Ecosystems. Progress In Oceanography 81(1-4): 207-213. Ben-Haim, Y., F. L. Thompson, C. C. Thompson, M. C. Cnockaert, B. Hoste, J. Swings and E. Rosenberg (2003). Vibrio coralliilyticus sp. nov., a temperature-dependent pathogen of the coral Pocillopora damicornis. International Journal of Systematic and Evolutionary Microbiology 53(Pt 1): 309-315.

- 123 - REFERENCES

Benagli, C., A. Demarta, A. Caminada, D. Ziegler, O. Petrini and M. Tonolla (2012). A Rapid MALDI-TOF MS Identification Database at Genospecies Level for Clinical and Environmental Aeromonas Strains. PloS one 7(10). Benagli, C., V. Rossi, M. Dolina, M. Tonolla and O. Petrini (2011). Matrix-assisted laser desorption ionization- time of flight mass spectrometry for the identification of clinically relevant bacteria. PLoS One 6(1): e16424. Bhanumathi, R., F. Sabeena, S. R. Isac, B. N. Shukla and D. V. Singh (2003). Molecular characterization of Vibrio cholerae O139 Bengal isolated from water and the aquatic plant Eichhornia crassipes in the River Ganga, Varanasi, India. Applied and environmental microbiology 69(4): 2389-2394. Bier, N., S. Bechlars, S. Diescher, F. Klein, G. Hauk, O. Duty, E. Strauch and R. Dieckmann (2013). Genotypic Diversity and Virulence Characteristics of Clinical and Environmental Vibrio vulnificus Isolates from the Baltic Sea Region. Applied and environmental microbiology. Bille, E., B. Dauphin, J. Leto, M. E. Bougnoux, J. L. Beretti, A. Lotz, S. Suarez, J. Meyer, O. Join-Lambert, P. Descamps, N. Grall, F. Mory, L. Dubreuil, P. Berche, X. Nassif and A. Ferroni (2012). MALDI-TOF MS Andromas strategy for the routine identification of bacteria, mycobacteria, yeasts, Aspergillus spp. and positive blood cultures. Clin Microbiol Infect 18(11): 1117-1125. Biosca, E. G. and C. Amaro (1996). Toxic and enzymatic activities of Vibrio vulnificus biotype 2 with respect to host specificity. Applied and environmental microbiology 62(7): 2331-2337. Blake, P. A., R. E. Weaver and D. G. Hollis (1980). Diseases of humans (other than cholera) caused by Vibrios. Annual Review of Microbiology 34: 341-367. Boer, S. I., E. A. Heinemeyer, K. Luden, R. Erler, G. Gerdts, F. Janssen and N. Brennholt (2013). Temporal and Spatial Distribution Patterns of Potentially Pathogenic Vibrio spp. at Recreational Beaches of the German North Sea. Microbial ecology 65(4): 1052-1067. Bourlat, S. J., A. Borja, J. Gilbert, M. I. Taylor, N. Davies, S. B. Weisberg, J. F. Griffith, T. Lettieri, D. Field, J. Benzie, F. O. Glockner, N. Rodriguez-Ezpeleta, D. P. Faith, T. P. Bean and M. Obst (2013). Genomics in marine monitoring: new opportunities for assessing marine health status. Marine pollution bulletin 74(1): 19-31. Boyd, E. F., K. E. Moyer, L. Shi and M. K. Waldor (2000). Infectious CTX Phi, and the vibrio pathogenicity island prophage in Vibrio mimicus: Evidence for recent horizontal transfer between V-mimicus and V- cholerae. Infection and immunity 68(3): 1507-1513. Brenner, D. J., F. W. Hickmanbrenner, J. V. Lee, A. G. Steigerwalt, G. R. Fanning, D. G. Hollis, J. J. Farmer, R. E. Weaver, S. W. Joseph and R. J. Seidler (1983). Vibrio-Furnissii (Formerly Aerogenic Biogroup of Vibrio-Fluvialis), a New Species Isolated from Human Feces and the Environment. Journal of clinical microbiology 18(4): 816-824. Buck, J. D. (1990). Isolation of Candida albicans and halophilic Vibrio spp. from aquatic birds in Connecticut and Florida. Applied and environmental microbiology 56(3): 826-828. Burge, C. A., C. Mark Eakin, C. S. Friedman, B. Froelich, P. K. Hershberger, E. E. Hofmann, L. E. Petes, K. C. Prager, E. Weil, B. L. Willis, S. E. Ford and C. D. Harvell (2014). Climate change influences on marine infectious diseases: implications for management and society. Annual review of marine science 6: 249- 277. Cabrera-Garcia, M. E., C. Vazquez-Salinas and E. I. Quinones-Ramirez (2004). Serologic and molecular characterization of Vibrio parahaemolyticus strains isolated from seawater and fish products of the Gulf of Mexico. Applied and environmental microbiology 70(11): 6401-6406. Calderaro, A., G. Piccolo, S. Montecchini, M. Buttrini, C. Gorrini, S. Rossi, M. C. Arcangeletti, F. De Conto, M. C. Medici and C. Chezzi (2013). MALDI-TOF MS analysis of human and animal Brachyspira species and benefits of database extension. Journal of Proteomics 78: 273-280. Callaway, A., M. Kostrzewa, B. Willershausen, F. Schmidt, B. Thiede, H. Kupper and S. Kneist (2013). Identification of Lactobacilli from deep carious lesions by means of species-specific PCR and MALDI- TOF mass spectrometry. Clin Lab 59(11-12): 1373-1379. Campbell, M. S. and A. C. Wright (2003). Real-time PCR analysis of Vibrio vulnificus from oysters. Applied and environmental microbiology 69(12): 7137-7144. Campos, E., H. Bolanos, M. T. Acuna, G. Diaz, M. C. Matamoros, H. Raventos, L. M. Sanchez, O. Sanchez and C. Barquero (1996). Vibrio mimicus diarrhea following ingestion of raw turtle eggs. Applied and environmental microbiology 62(4): 1141-1144. Ceccarelli, D. and R. R. Colwell (2014). Vibrio ecology, pathogenesis, and evolution. Frontiers in Microbiology 5. Cerda-Cuellar, M., J. Jofre and A. R. Blanch (2000). A selective medium and a specific probe for detection of Vibrio vulnificus. Applied and environmental microbiology 66(2): 855-859. Chatterjee, S. and S. Haldar (2012). Vibrio Related Diseases in Aquaculture and Development of Rapid and Accurate Identification Methods. J Marine Sci Res Dev S1:002.

- 124 - REFERENCES

Chavez, M. R. C., V. P. Sedas, E. O. Borunda and F. L. Reynoso (2005). Influence of water temperature and salinity on seasonal occurrences of Vibrio cholerae and enteric bacteria in oyster-producing areas of Veracruz, Mexico. Marine Pollution Bulletin 50(12): 1641-1648. Choi, S., K. Jang, H. J. Yun and D. H. Kang (2012). Identification of the Vibrio vulnificus htpG gene and its influence on cold shock recovery. Journal of microbiology 50(4): 707-711. Chowdhury, M. A. R., H. Yamanaka, S. Miyoshi and S. Shinoda (1990). Ecology and Seasonal Distribution of Vibrio-Parahaemolyticus in Aquatic Environments of a Temperate Region. FEMS microbiology ecology 74(1): 1-9. Chowdhury, N. R., O. C. Stine, J. G. Morris and G. B. Nair (2004). Assessment of evolution of pandemic Vibrio parahaemolyticus by multilocus sequence typing. Journal of clinical microbiology 42(3): 1280-1282. Christensen, J. J., R. Dargis, M. Hammer, U. S. Justesen, X. C. Nielsen, M. Kemp and D. M.-T. M. S. Grp (2012). Matrix-Assisted Laser Desorption Ionization-Time of Flight Mass Spectrometry Analysis of Gram-Positive, Catalase-Negative Cocci Not Belonging to the Streptococcus or Enterococcus Genus and Benefits of Database Extension. Journal of clinical microbiology 50(5): 1787-1791. Clark, C. G., P. Kruczkiewicz, C. Guan, S. J. McCorrister, P. Chong, J. Wylie, P. van Caeseele, H. A. Tabor, P. Snarr, M. W. Gilmour, E. N. Taboada and G. R. Westmacott (2013). Evaluation of MALDI-TOF mass spectroscopy methods for determination of Escherichia coli pathotypes. J Microbiol Methods 94(3): 180-191. Claydon, M. A., S. N. Davey, V. EdwardsJones and D. B. Gordon (1996). The rapid identification of intact microorganisms using mass spectrometry. Nature Biotechnology 14(11): 1584-1586. Cohen, O., U. Gophna and T. Pupko (2011). The Complexity Hypothesis Revisited: Connectivity Rather Than Function Constitutes a Barrier to Horizontal Gene Transfer. Molecular biology and evolution 28(4): 1481-1489. Collin, B. and A. S. Rehnstam-Holm (2011). Occurrence and potential pathogenesis of Vibrio cholerae, Vibrio parahaemolyticus and Vibrio vulnificus on the South Coast of Sweden. FEMS microbiology ecology 78(2): 306-313. Colodner, R., R. Raz, I. Meir, T. Lazarovich, L. Lerner, J. Kopelowitz, Y. Keness, W. Sakran, S. Ken-Dror and N. Bisharat (2004). Identification of the emerging pathogen Vibrio vulnificus biotype 3 by commercially available phenotypic methods. Journal of clinical microbiology 42(9): 4137-4140. Colombo, M. M., S. Mastrandrea, F. Leite, A. Santona, S. Uzzau, P. Rappelli, M. Pisano, S. Rubino and P. Cappuccinelli (1997). Tracking of clinical and environmental Vibrio cholerae O1 strains by combined analysis of the presence of the toxin cassette, plasmid content and ERIC PCR. FEMS immunology and medical microbiology 19(1): 33-45. Colwell, R. R. (1996). Global climate and infectious disease: the cholera paradigm. Science 274(5295): 2025- 2031. Corte, D. D., I. Lekunberri, E. Sintes, J. Garcia, S. Gonzales and G. Herndl (2014). Linkage between copepods and bacteria in the North Atlantic Ocean. Aquatic Microbial Ecology 72(3): 215-225. Cottrell, M. T., D. N. Wood, L. Y. Yu and D. L. Kirchman (2000). Selected chitinase genes in cultured and uncultured marine bacteria in the alpha- and gamma-subclasses of the proteobacteria. Applied and environmental microbiology 66(3): 1195-1201. Croci, L., E. Suffredini, L. Cozzi, L. Toti, D. Ottaviani, C. Pruzzo, P. Serratore, R. Fischetti, E. Goffredo, G. Loffredo and R. Mioni (2007). Comparison of different biochemical and molecular methods for the identification of Vibrio parahaemolyticus. Journal of applied microbiology 102(1): 229-237. Dalsgaard, A., N. FrimodtMoller, B. Bruun, L. Hoi and J. L. Larsen (1996). Clinical manifestations and molecular epidemiology of Vibrio vulnificus infections in Denmark. European Journal of Clinical Microbiology & Infectious Diseases 15(3): 227-232. Daniels, N. A. and A. Shafaie (2000). A review of pathogenic Vibrio infections for clinicians. Infections in Medicine 17(10): 665-+. Danin-Poleg, Y., L. A. Cohen, H. Gancz, Y. Y. Broza, H. Goldshmidt, E. Malul, L. Valinsky, L. Lerner, M. Broza and Y. Kashi (2007). Vibrio cholerae strain typing and Phylogeny study based on simple sequence repeats. Journal of clinical microbiology 45(3): 736-746. Davis, B. R., G. R. Fanning, J. M. Madden, A. G. Steigerwalt, H. B. Bradford, H. L. Smith and D. J. Brenner (1981). Characterization of Biochemically Atypical Vibrio-Cholerae Strains and Designation of a New Pathogenic Species, Vibrio-Mimicus. Journal of clinical microbiology 14(6): 631-639. Davis, G. H. and R. W. Park (1962). A taxonomic study of certain bacteria currently classified as Vibrio species. Journal of general microbiology 27: 101-119. De, A. and M. Mathur (2011). Vibrio vulnificus Diarrhea in a Child with Respiratory Infection. J Glob Infect Dis 3(3): 300-302. de Magny, G. C., P. K. Mozumder, C. J. Grim, N. A. Hasan, M. N. Naser, M. Alam, R. B. Sack, A. Huq and R. R. Colwell (2011). Role of zooplankton diversity in Vibrio cholerae population dynamics and in the incidence of cholera in the Bangladesh Sundarbans. Applied and environmental microbiology 77(17): 6125-6132. - 125 - REFERENCES

Depaola, A., L. H. Hopkins, J. T. Peeler, B. Wentz and R. M. Mcphearson (1990). Incidence of Vibrio- Parahaemolyticus in United-States Coastal Waters and Oysters. Applied and Environmental Microbiology 56(8): 2299-2302. Di Pinto, A., V. Terio, L. Novello and G. Tantillo (2011). Comparison between thiosulphate-citrate-bile salt sucrose (TCBS) agar and CHROMagar Vibrio for isolating Vibrio parahaemolyticus. Food Control 22(1): 124-127. Dieckmann, R., E. Strauch and T. Alter (2010). Rapid identification and characterization of Vibrio species using whole-cell MALDI-TOF mass spectrometry. Journal of applied microbiology 109(1): 199-211. Dikow, R. B. and W. L. Smith (2013). Genome-level homology and phylogeny of Vibrionaceae (Gammaproteobacteria: Vibrionales) with three new complete genome sequences. BMC microbiology 13. Dobbs, F. C., A. L. Goodrich, F. K. Thomson, 3rd and W. Hynes (2013). Pandemic serotypes of Vibrio cholerae isolated from ships' ballast tanks and coastal waters: assessment of antibiotic resistance and virulence genes (tcpA and ctxA). Microbial ecology 65(4): 969-974. Donay, J. L., D. Mathieu, P. Fernandes, C. Pregermain, P. Bruel, A. Wargnier, I. Casin, F. X. Weill, P. H. Lagrange and J. L. Herrmann (2004). Evaluation of the automated phoenix system for potential routine use in the clinical microbiology laboratory. Journal of clinical microbiology 42(4): 1542-1546. Dorsch, M., D. Lane and E. Stackebrandt (1992). Towards a Phylogeny of the Genus Vibrio Based on 16s Ribosomal-Rna Sequences. International Journal of Systematic Bacteriology 42(1): 58-63. Drake, L. A., M. A. Doblin and F. C. Dobbs (2007). Potential microbial bioinvasions via ships' ballast water, sediment, and biofilm. Marine pollution bulletin 55(7-9): 333-341. Drake, S. L., A. DePaola and L. A. Jaykus (2007). An overview of Vibrio vulnificus and Vibrio parahaemolyticus. Comprehensive Reviews in Food Science and Food Safety 6(4): 120-144. Dutta, D., G. Chowdhury, G. P. Pazhani, S. Guin, S. Dutta, S. Ghosh, K. Rajendran, R. K. Nandy, A. K. Mukhopadhyay, M. K. Bhattacharya, U. Mitra, Y. Takeda, G. B. Nair and T. Ramamurthy (2013). Vibrio cholerae non-O1, non-O139 serogroups and cholera-like diarrhea, Kolkata, India. Emerging infectious diseases 19(3): 464-467. Eiler, A. and S. Bertilsson (2006). Detection and quantification of Vibrio populations using denaturant gradient gel electrophoresis. Journal of microbiological methods 67(2): 339-348. Eiler, A., M. Johansson and S. Bertilsson (2006). Environmental influences on Vibrio populations in northern temperate and boreal coastal waters (Baltic and Skagerrak Seas). Applied and environmental microbiology 72(9): 6004-6011. Eilers, H., J. Pernthaler and R. Amann (2000). Succession of pelagic marine bacteria during enrichment: a close look at cultivation-induced shifts. Applied and environmental microbiology 66(11): 4634-4640. Emami, K., V. Askari, M. Ullrich, K. Mohinudeen, A. C. Anil, L. Khandeparker, J. G. Burgess and E. Mesbahi (2012). Characterization of Bacteria in Ballast Water Using MALDI-TOF Mass Spectrometry. PloS one 7(6). Erler, R., A. Wichels, E. Heinemeyer, G. Hauk, M. Hippelein, N. Torres-Reyes and G. Gerdts (2014). VibrioBase: a MALDI-TOF MS database for fast identification of Vibrio spp. that are potentially pathogenic in humans. Applied and Environmental Microbiology. Estrada, M., P. Henriksen, J. M. Gasol, E. O. Casamayor and C. Pedros-Alio (2004). Diversity of planktonic photoautotrophic microorganisms along a salinity gradient as depicted by microscopy, flow cytometry, pigment analysis and DNA-based methods. FEMS microbiology ecology 49(2): 281-293. Farmer, J. J. (2006). The Family Vibrionaceae. Prokaryotes: A Handbook on the Biology of Bacteria, Vol 6, Third Edition: 495-507. FDA (2003). available from: http://www.fda.gov/Food/FoodScienceResearch/LaboratoryMethods/ucm070830.htm. Fenselau, C. (1997). MALDI MS and strategies for protein analysis. Analytical chemistry 69(21): 661A-665A. Gaillot, O., N. Blondiaux, C. Loiez, F. Wallet, N. Lemaitre, S. Herwegh and R. J. Courcol (2011). Cost- Effectiveness of Switch to Matrix-Assisted Laser Desorption Ionization-Time of Flight Mass Spectrometry for Routine Bacterial Identification. Journal of clinical microbiology 49(12): 4412-4412. Garcia, K., R. Bastias, G. Higuera, R. Torres, A. Mellado, P. Uribe and R. T. Espejo (2012). Rise and fall of pandemic Vibrio parahaemolyticus serotype O3:K6 in southern Chile. Environmental microbiology. Garnier, M., Y. Labreuche, C. Garcia, M. Robert and J. L. Nicolas (2007). Evidence for the involvement of pathogenic bacteria in summer mortalities of the Pacific oyster Crassostrea gigas. Microbial ecology 53(2): 187-196. Gasteiger, E., A. Gattiker, C. Hoogland, I. Ivanyi, R. D. Appel and A. Bairoch (2003). ExPASy: The proteomics server for in-depth protein knowledge and analysis. Nucleic acids research 31(13): 3784-3788. Gerdts, G., P. Brandt, K. Kreisel, M. Boersma, K. L. Schoo and A. Wichels (2013). The microbiome of North Sea copepods. Helgoland Marine Research 67(4): 757-773.

- 126 - REFERENCES

Gil, A. I., V. R. Louis, I. N. G. Rivera, E. Lipp, A. Huq, C. F. Lanata, D. N. Taylor, E. Russek-Cohen, N. Choopun, R. B. Sack and R. R. Colwell (2004). Occurrence and distribution of Vibrio cholerae in the coastal environment of Peru. Environmental microbiology 6(7): 699-706. Gladney, L. (2014). personal communication. (CDC, Atlanta, USA). Gonzalez-Escalona, N., V. Cachicas, C. Acevedo, M. L. Rioseco, J. A. Vergara, F. Cabello, J. Romero and R. T. Espejo (2005). Vibrio parahaemolyticus diarrhea, Chile, 1998 and 2004. Emerging Infectious Diseases 11(1): 129-131. Gonzalez-Escalona, N., A. Fey, M. G. Hofle, R. T. Espejo and A. G. C (2006). Quantitative reverse transcription polymerase chain reaction analysis of Vibrio cholerae cells entering the viable but non-culturable state and starvation in response to cold shock. Environmental microbiology 8(4): 658-666. Gonzalez-Escalona, N., J. Martinez-Urtaza, J. Romero, R. T. Espejo, L. A. Jaykus and A. DePaola (2008). Determination of molecular phylogenetics of Vibrio parahaemolyticus strains by multilocus sequence typing. Journal of Bacteriology 190(8): 2831-2840. Griffin, P., G. Price, B. Hamilton, J. Schooneveldt, T. Urbanski, S. Schlebusch and D. Venter (2013). Rapid Identification of Vancomycin Resistant Enterococci with Maldi-Tof Mass Spectrometry. Internal Medicine Journal 43: 27-27. Griffith, D. C., L. A. Kelly-Hope and M. A. Miller (2006). Review of reported cholera outbreaks worldwide, 1995-2005. American Journal of Tropical Medicine and Hygiene 75(5): 973-977. Grisez, L. and F. Ollevier (1995). Comparative Serology of the Marine Fish Pathogen Vibrio anguillarum. Applied and environmental microbiology 61(12): 4367-4373. Guerrant, R. L., D. H. Walker and P. F. Weller (2006). Tropical infectious diseases : principles, pathogens & practice. Philadelphia, Churchill Livingstone. Hammer, B. K. and B. L. Bassler (2003). Quorum sensing controls biofilm formation in Vibrio cholerae. Molecular Microbiology 50(1): 101-104. Harth, E., J. Romero, R. Torres and R. T. Espejo (2007). Intragenomic heterogeneity and intergenomic recombination among Vibrio parahaemolyticus 16S rRNA genes. Microbiology 153(Pt 8): 2640-2647. Harvell, C. D., C. E. Mitchell, J. R. Ward, S. Altizer, A. P. Dobson, R. S. Ostfeld and M. D. Samuel (2002). Ecology - Climate warming and disease risks for terrestrial and marine biota. Science 296(5576): 2158- 2162. Hayashi, K., J. Moriwaki, T. Sawabe, F. L. Thompson, J. Swings, N. Gudkovs, R. Christen and Y. Ezura (2003). Vibrio superstes sp nov., isolated from the gut of Australian abalones Haliotis laevigata and Haliotis rubra. International Journal of Systematic and Evolutionary Microbiology 53: 1813-1817. Hazen, T. H., K. D. Kennedy, S. Chen, S. V. Yi and P. A. Sobecky (2009). Inactivation of mismatch repair increases the diversity of Vibrio parahaemolyticus. Environmental microbiology 11(5): 1254-1266. Hazen, T. H., R. J. Martinez, Y. Chen, P. C. Lafon, N. M. Garrett, M. B. Parsons, C. A. Bopp, M. C. Sullards and P. A. Sobecky (2009). Rapid identification of Vibrio parahaemolyticus by whole-cell matrix- assisted laser desorption ionization-time of flight mass spectrometry. Applied and environmental microbiology 75(21): 6745-6756. Heidelberg, J. F., J. A. Eisen, W. C. Nelson, R. A. Clayton, M. L. Gwinn, R. J. Dodson, D. H. Haft, E. K. Hickey, J. D. Peterson, L. Umayam, S. R. Gill, K. E. Nelson, T. D. Read, H. Tettelin, D. Richardson, M. D. Ermolaeva, J. Vamathevan, S. Bass, H. Y. Qin, I. Dragoi, P. Sellers, L. McDonald, T. Utterback, R. D. Fleishmann, W. C. Nierman, O. White, S. L. Salzberg, H. O. Smith, R. R. Colwell, J. J. Mekalanos, J. C. Venter and C. M. Fraser (2000). DNA sequence of both chromosomes of the cholera pathogen Vibrio cholerae. Nature 406(6795): 477-483. Heidelberg, J. F., K. B. Heidelberg and R. R. Colwell (2002). Bacteria of the -Subclass Proteobacteria Associated with Zooplankton in Chesapeake Bay. Applied and environmental microbiology 68(11): 5498-5507. Holland, R. D., J. G. Wilkes, F. Rafii, J. B. Sutherland, C. C. Persons, K. J. Voorhees and J. O. Lay (1996). Rapid identification of intact whole bacteria based on spectral patterns using matrix-assisted laser desorption/ionization with time-of-flight mass spectrometry. Rapid Communications in Mass Spectrometry 10(10): 1227-1232. Honda, T., M. Yoh, U. Kongmuang and T. Miwatani (1985). Enzyme-Linked Immunosorbent Assays for Detection of Thermostable Direct Hemolysin of Vibrio-Parahaemolyticus. Journal of clinical microbiology 22(3): 383-386. Hotta, Y., J. Sato, H. Sato, A. Hosoda and H. Tamura (2011). Classification of the Genus Bacillus Based on MALDI-TOF MS Analysis of Ribosomal Proteins Coded in S10 and spc Operons. Journal of Agricultural and Food Chemistry 59(10): 5222-5230. Hulton, C. S. J., C. F. Higgins and P. M. Sharp (1991). Eric Sequences - a Novel Family of Repetitive Elements in the Genomes of Escherichia-Coli, Salmonella-Typhimurium and Other Enterobacteria. Molecular Microbiology 5(4): 825-834. Hunter, P. R. (1990). Reproducibility and indices of discriminatory power of microbial typing methods. Journal of clinical microbiology 28(9): 1903-1905. - 127 - REFERENCES

Huq, A., E. B. Small, P. A. West, M. I. Huq, R. Rahman and R. R. Colwell (1983). Ecological relationships between Vibrio cholerae and planktonic crustacean copepods. Applied and environmental microbiology 45(1): 275-283. Ilina, E. N., A. D. Borovskaya, M. M. Malakhova, V. A. Vereshchagin, A. A. Kubanova, A. N. Kruglov, T. S. Svistunova, A. O. Gazarian, T. Maier, M. Kostrzewa and V. M. Govorun (2009). Direct Bacterial Profiling by Matrix-Assisted Laser Desorption-Ionization Time-of-Flight Mass Spectrometry for Identification of Pathogenic Neisseria. Journal of Molecular Diagnostics 11(1): 75-86. Inoue, Y., T. Ono, T. Matsui, J. Miyasaka, Y. Kinoshita and H. Ihn (2008). Epidemiological survey of Vibrio vulnificus infection in Japan between 1999 and 2003. J Dermatol 35(3): 129-139. IPCC. (2013). "Intergovernmental Panel on Climate Change. Working group I - Contribution to the fifth assessment report - Climate Change 2013 - the physical science basis. ." from www.ipcc.ch. Jain, R., M. C. Rivera and J. A. Lake (1999). Horizontal gene transfer among genomes: The complexity hypothesis. Proceedings of the National Academy of Sciences of the United States of America 96(7): 3801-3806. Janda, J. M., C. Powers, R. G. Bryant and S. L. Abbott (1988). Current perspectives on the epidemiology and pathogenesis of clinically significant Vibrio spp. Clin Microbiol Rev 1(3): 245-267. Jones, M. K. and J. D. Oliver (2009). Vibrio vulnificus: disease and pathogenesis. Infection and immunity 77(5): 1723-1733. Jores, J., B. Appel and A. Lewin (2007). Vibrio navarrensis biotype pommerensis: a new biotype of V. navarrensis isolated in the German Baltic Sea. Systematic and applied microbiology 30(1): 27-30. Jores, J., R. Stephan, D. Knabner, H. R. Gelderblom and A. Lewin (2003). Isolation of Vibrio vulnificus and atypical Vibrio from surface water of the Baltic Sea in Germany. Berliner Und Munchener Tierarztliche Wochenschrift 116(9-10): 396-400. Kamae, Y., H. Shiogama, M. Watanabe and M. Kimoto (2014). Attributing the increase in Northern Hemisphere hot summers since the late 20th century. Geophysical Research Letters 41(14): 2014GL061062. Kaneko, T. and R. R. Colwell (1973). Ecology of Vibrio-Parahaemolyticus in Chesapeake Bay. Journal of Bacteriology 113(1): 24-32. Karas, M. and R. Kruger (2003). Ion formation in MALDI: the cluster ionization mechanism. Chem Rev 103(2): 427-440. Karunasagar, I., R. Pai, G. R. Malathi and I. Karunasagar (1994). Mass mortality of Penaeus monodon larvae due to antibiotic-resistant Vibrio harveyi infection. Aquaculture 128(3–4): 203-209. Kass, E. H. (1987). History of the specialty of infectious diseases in the United States. Annals of Internal Medicine 106(5): 745-756. Kelly, M. T. (1982). Effect of temperature and salinity on Vibrio (Beneckea) vulnificus occurrence in a Gulf Coast environment. Applied and environmental microbiology 44(4): 820-824. Keymer, D. P., L. H. Lam and A. B. Boehm (2009). Biogeographic patterns in genomic diversity among a large collection of Vibrio cholerae isolates. Applied and environmental microbiology 75(6): 1658-1666. Keymer, D. P., M. C. Miller, G. K. Schoolnik and A. B. Boehm (2007). Genomic and phenotypic diversity of coastal Vibrio cholerae strains is linked to environmental factors. Applied and environmental microbiology 73(11): 3705-3714. Khamis, A., D. Raoult and B. La Scola (2004). rpoB gene sequencing for identification of Corynebacterium species. Journal of clinical microbiology 42(9): 3925-3931. Khamis, A., D. Raoult and B. La Scola (2005). Comparison between rpoB and 16S rRNA gene sequencing for molecular identification of 168 clinical isolates of Corynebacterium. Journal of clinical microbiology 43(4): 1934-1936. Khan, A. A., S. McCarthy, R. F. Wang and C. E. Cerniglia (2002). Characterization of United States outbreak isolates of Vibrio parahaemolyticus using enterobacterial repetitive intergenic consensus (ERIC) PCR and development of a rapid PCR method for detection of 03 : K6 isolates. FEMS microbiology letters 206(2): 209-214. Khot, P. D., M. R. Couturier, A. Wilson, A. Croft and M. A. Fisher (2012). Optimization of Matrix-Assisted Laser Desorption Ionization-Time of Flight Mass Spectrometry Analysis for Bacterial Identification. Journal of clinical microbiology 50(12): 3845-3852. Ki, J. S., R. Zhang, W. Zhang, Y. L. Huang and P. Y. Qian (2009). Analysis of RNA Polymerase Beta Subunit (rpoB) Gene Sequences for the Discriminative Power of Marine Vibrio Species. Microbial Ecology 58(4): 679-691. Ki, J. S., W. Zhang and P. Y. Qian (2009). Discovery of marine Bacillus species by 16S rRNA and rpoB comparisons and their usefulness for species identification. Journal of microbiological methods 77(1): 48-57. Kirschner, A. K., S. Schauer, B. Steinberger, I. Wilhartitz, C. J. Grim, A. Huq, R. R. Colwell, A. Herzig and R. Sommer (2011). Interaction of Vibrio cholerae non-O1/non-O139 with copepods, cladocerans and competing bacteria in the large alkaline lake Neusiedler See, Austria. Microbial ecology 61(3): 496- 506. - 128 - REFERENCES

Kirschner, A. K., J. Schlesinger, A. H. Farnleitner, R. Hornek, B. Suss, B. Golda, A. Herzig and B. Reitner (2008). Rapid growth of planktonic Vibrio cholerae non-O1/non-O139 strains in a large alkaline lake in Austria: dependence on temperature and dissolved organic carbon quality. Applied and environmental microbiology 74(7): 2004-2015. Kitatsukamoto, K., H. Oyaizu, K. Nanba and U. Simidu (1993). Phylogenetic-Relationships of Marine-Bacteria, Mainly Members of the Family Vibrionaceae, Determined on the Basis of 16s Ribosomal-Rna Sequences. International Journal of Systematic Bacteriology 43(1): 8-19. Koubek, J., O. Uhlik, K. Jecna, P. Junkova, J. Vrkoslavova, J. Lipov, V. Kurzawova, T. Macek and M. Mackova (2012). Whole-cell MALDI-TOF: Rapid screening method in environmental microbiology. International Biodeterioration & Biodegradation 69(0): 82-86. Krishnamurthy, T. and P. L. Ross (1996). Rapid identification of bacteria by direct matrix-assisted laser desorption/ionization mass spectrometric analysis of whole cells. Rapid Communications in Mass Spectrometry 10(15): 1992-1996. Kupfer, M., P. Kuhnert, B. M. Korczak, R. Peduzzi and A. Demarta (2006). Genetic relationships of Aeromonas strains inferred from 16S rRNA, gyrB and rpoB gene sequences. International Journal of Systematic and Evolutionary Microbiology 56: 2743-2751. Kushmaro, A., E. Banin, Y. Loya, E. Stackebrandt and E. Rosenberg (2001). Vibrio shiloi sp nov., the causative agent of bleaching of the coral Oculina patagonica. International Journal of Systematic and Evolutionary Microbiology 51: 1383-1388. Laciar, A., L. Vaca, R. Lopresti, A. Vega, C. Mattana and O. N. de Centorbi (2006). DNA fingerprinting by ERIC-PCR for comparing Listeria spp. strains isolated from different sources in San Luis, Argentina. Revista Argentina de microbiologia 38(2): 55-60. Lartigue, M. F. (2013). Matrix-assisted laser desorption ionization time-of-flight mass spectrometry for bacterial strain characterization. Infect Genet Evol 13: 230-235. Lasch, P., W. Beyer, H. Nattermann, M. Stammler, E. Siegbrecht, R. Grunow and D. Naumann (2009). Identification of Bacillus anthracis by using matrix-assisted laser desorption ionization-time of flight mass spectrometry and artificial neural networks. Applied and environmental microbiology 75(22): 7229-7242. Lau, S. K. P., B. S. F. Tang, S. O. T. Curreem, T. M. Chan, P. Martelli, C. W. S. Tse, A. K. L. Wu, K. Y. Yuen and P. C. Y. Woo (2012). Matrix-Assisted Laser Desorption Ionization-Time of Flight Mass Spectrometry for Rapid Identification of Burkholderia pseudomallei: Importance of Expanding Databases with Pathogens Endemic to Different Localities. Journal of clinical microbiology 50(9): 3142-3143. Lay, J. O., Jr. (2001). MALDI-TOF mass spectrometry of bacteria. Mass Spectrom Rev 20(4): 172-194. Le Roux, F., A. Goubet, F. L. Thompson, N. Faury, M. Gay, J. Swings and D. Saulnier (2005). Vibrio gigantis sp. nov., isolated from the haemolymph of cultured oysters (Crassostrea gigas). International Journal of Systematic and Evolutionary Microbiology 55(Pt 6): 2251-2255. Lee, J. H., M. W. Kim, B. S. Kim, S. M. Kim, B. C. Lee, T. S. Kim and S. H. Choi (2007). Identification and characterization of the Vibrio vulnificus rtxA essential for cytotoxicity in vitro and virulence in mice. Journal of microbiology 45(2): 146-152. Lemos, M. L., A. E. Toranzo and J. L. Barja (1985). Modified Medium for the Oxidation-Fermentation Test in the Identification of Marine-Bacteria. Applied and environmental microbiology 49(6): 1541-1543. Lhafi, S. K. and M. Kuhne (2007). Occurrence of Vibrio spp. in blue mussels (Mytilus edulis) from the German Wadden Sea. International Journal of Food Microbiology 116(2): 297-300. Li, W., L. Mei, Z. Tang, X. Yang, X. Li, X. Pei, G. Wang, P. Fu, Y. Wu and Y. Guo (2014). [Analysis of molecular features of clinical Vibrio parahaemolyticus strains in China]. Zhonghua Yu Fang Yi Xue Za Zhi 48(1): 44-52. Lindgren, E., Y. Andersson, J. E. Suk, B. Sudre and J. C. Semenza (2012). Monitoring EU Emerging Infectious Disease Risk Due to Climate Change. Science 336(6080): 418-419. Linkous, D. A. and J. D. Oliver (1999). Pathogenesis of Vibrio vulnificus. FEMS microbiology letters 174(2): 207-214. Lipp, E. K., A. Huq and R. R. Colwell (2002). Effects of Global Climate on Infectious Disease: the Cholera Model. Clinical Microbiology Reviews 15(4): 757-770. Ludwig, W., O. Strunk, R. Westram, L. Richter, H. Meier, Yadhukumar, A. Buchner, T. Lai, S. Steppi, G. Jobb, W. Forster, I. Brettske, S. Gerber, A. W. Ginhart, O. Gross, S. Grumann, S. Hermann, R. Jost, A. Konig, T. Liss, R. Lussmann, M. May, B. Nonhoff, B. Reichel, R. Strehlow, A. Stamatakis, N. Stuckmann, A. Vilbig, M. Lenke, T. Ludwig, A. Bode and K. H. Schleifer (2004). ARB: a software environment for sequence data. Nucleic acids research 32(4): 1363-1371. Lukinmaa, S., K. Mattila, V. Lehtinen, M. Hakkinen, M. Koskela and A. Siitonen (2006). Territorial waters of the Baltic Sea as a source of infections caused by Vibrio cholerae non-O1, non-O139: report of 3 hospitalized cases. Diagnostic microbiology and infectious disease 54(1): 1-6.

- 129 - REFERENCES

Macian, M. C., C. R. Arias, R. Aznar, E. Garay and M. J. Pujalte (2000). Identification of Vibrio spp. (other than V. vulnificus) recovered on CPC agar from marine natural samples. International microbiology : the official journal of the Spanish Society for Microbiology 3(1): 51-53. Macian, M. C., W. Ludwig, K. H. Schleifer, M. J. Pujalte and E. Garay (2001). Vibrio agarivorans sp. nov., a novel agarolytic marine bacterium. International Journal of Systematic and Evolutionary Microbiology 51(Pt 6): 2031-2036. Mahe, P., M. Arsac, S. Chatellier, V. Monnin, N. Perrot, S. Mailler, V. Girard, M. Ramjeet, J. Surre, B. Lacroix, A. van Belkum and J. B. Veyrieras (2014). Automatic identification of mixed bacterial species fingerprints in a MALDI-TOF mass-spectrum. Bioinformatics 30(9): 1280-1286. Maiden, M. C., J. A. Bygraves, E. Feil, G. Morelli, J. E. Russell, R. Urwin, Q. Zhang, J. Zhou, K. Zurth, D. A. Caugant, I. M. Feavers, M. Achtman and B. G. Spratt (1998). Multilocus sequence typing: a portable approach to the identification of clones within populations of pathogenic microorganisms. Proceedings of the National Academy of Sciences of the United States of America 95(6): 3140-3145. Maier, T. and M. Kostrzewa (2007). Fast and reliable MALDI-TOF MS-based microorganism identification. Chimica Oggi-Chemistry Today 25(2): 68-71. Makino, K., K. Oshima, K. Kurokawa, K. Yokoyama, T. Uda, K. Tagomori, Y. Iijima, M. Najima, M. Nakano, A. Yamashita, Y. Kubota, S. Kimura, T. Yasunaga, T. Honda, H. Shinagawa, M. Hattori and T. Iida (2003). Genome sequence of Vibrio parahaemolyticus: a pathogenic mechanism distinct from that of V cholerae. The Lancet 361(9359): 743-749. Mantri, C. K., S. S. Mohapatra, T. Ramamurthy, R. Ghosh, R. R. Colwell and D. V. Singh (2006). Septaplex PCR assay for rapid identification of Vibrio cholerae including detection of virulence and int SXT genes. FEMS microbiology letters 265(2): 208-214. Martin, Y., J. L. Bonnefort and L. Chancerelle (2002). Gorgonians mass mortality during the 1999 late summer in French Mediterranean coastal waters: the bacterial hypothesis. Water Research 36(3): 779-782. Martinez-Picado, J., M. Alsina, A. R. Blanch, M. Cerda and J. Jofre (1996). Species-specific detection of Vibrio anguillarum in marine aquaculture environments by selective culture and DNA hybridization. Applied and environmental microbiology 62(2): 443-449. Martinez-Urtaza, J., V. Blanco-Abad, A. Rodriguez-Castro, J. Ansede-Bermejo, A. Miranda and M. X. Rodriguez-Alvarez (2011). Ecological determinants of the occurrence and dynamics of Vibrio parahaemolyticus in offshore areas. The ISME journal. McCarthy, S. A., R. M. McPhearson, A. M. Guarino and J. L. Gaines (1992). Toxigenic Vibrio cholerae O1 and cargo ships entering Gulf of Mexico. Lancet 339(8793): 624-625. McLaughlin, J. B., A. DePaola, C. A. Bopp, K. A. Martinek, N. P. Napolilli, C. G. Allison, S. L. Murray, E. C. Thompson, M. M. Bird and J. P. Middaugh (2005). Outbreak of Vibrio parahaemolyticus gastroenteritis associated with Alaskan oysters. New England Journal of Medicine 353(14): 1463-1470. Melhus, A., T. Holmdahl and I. Tjernberg (1995). First documented case of bacteremia with Vibrio vulnificus in Sweden. Scandinavian Journal of Infectious Diseases 27(1): 81-82. Mellmann, A., J. Cloud, T. Maier, U. Keckevoet, I. Ramminger, P. Iwen, J. Dunn, G. Hall, D. Wilson, P. Lasala, M. Kostrzewa and D. Harmsen (2008). Evaluation of matrix-assisted laser desorption ionization-time- of-flight mass spectrometry in comparison to 16S rRNA gene sequencing for species identification of nonfermenting bacteria. Journal of clinical microbiology 46(6): 1946-1954. Mintz, E. D. and R. V. Tauxe (2013). Cholera in Africa: a closer look and a time for action. The Journal of infectious diseases 208 Suppl 1: S4-7. Mohapatra, B. R., K. Broersma and A. Mazumder (2007). Comparison of five rep-PCR genomic fingerprinting methods for differentiation of fecal Escherichia coli from humans, poultry and wild birds. FEMS microbiology letters 277(1): 98-106. Mollet, C., M. Drancourt and D. Raoult (1997). rpoB sequence analysis as a novel basis for bacterial identification. Molecular Microbiology 26(5): 1005-1011. Moreno, L. Z., K. S. Castilla, D. D. S. de Gobbi, T. A. Coutinho, T. S. P. Ferreira and A. M. Moreno (2011). Eric-Pcr Genotypic Characterization of Haemophilus Parasuis Isolated from Brazilian Swine. Brazilian Journal of Microbiology 42(4): 1420-1426. Morris, J. G., R. Wilson, B. R. Davis, I. K. Wachsmuth, C. F. Riddle, H. G. Wathen, R. A. Pollard and P. A. Blake (1981). Non-O Group-1 Vibrio-Cholerae Gastroenteritis in the United-States - Clinical, Epidemiologic, and Laboratory Characteristics of Sporadic Cases. Annals of Internal Medicine 94(5): 656-658. Mosley, W. H., K. M. S. Aziz and A. Ahmed (1970). Serological Evidence for Identity of Vascular Permeability Factor and Ileal Loop Toxin of Vibrio-Cholerae. Journal of Infectious Diseases 121(3): 243-&. Muanprasat, C. and V. Chatsudthipong (2013). Cholera: pathophysiology and emerging therapeutic targets. Future Med Chem 5(7): 781-798. Nair, G. B., T. Ramamurthy, S. K. Bhattacharya, B. Dutta, Y. Takeda and D. A. Sack (2007). Global dissemination of Vibrio parahaemolyticus serotype O3:K6 and its serovariants. Clinical Microbiology Reviews 20(1): 39-48. - 130 - REFERENCES

Namdari, H., C. R. Klaips and J. L. Hughes (2000). A cytotoxin-producing strain of Vibrio cholerae non-O1, non-O139 as a cause of cholera and bacteremia after consumption of raw clams. Journal of clinical microbiology 38(9): 3518-3519. Naughton, L. M., S. L. Blumerman, M. Carlberg and E. F. Boyd (2009). Osmoadaptation among Vibrio species and unique genomic features and physiological responses of Vibrio parahaemolyticus. Applied and environmental microbiology 75(9): 2802-2810. Nishibuchi, M. and J. B. Kaper (1995). Thermostable Direct Hemolysin Gene of Vibrio-Parahaemolyticus - a Virulence Gene Acquired by a Marine Bacterium. Infection and immunity 63(6): 2093-2099. Noguerola, I. and A. R. Blanch (2008). Identification of Vibrio spp. with a set of dichotomous keys. Journal of applied microbiology 105(1): 175-185. Nowakowska, J. and J. D. Oliver (2013). Resistance to environmental stresses by Vibrio vulnificus in the viable but nonculturable state. FEMS microbiology ecology 84(1): 213-222. O'Hara, C. M., E. G. Sowers, C. A. Bopp, S. B. Duda and N. A. Strockbine (2003). Accuracy of six commercially available systems for identification of members of the family Vibrionaceae. Journal of clinical microbiology 41(12): 5654-5659. Oberbeckmann, S., B. M. Fuchs, M. Meiners, A. Wichels, K. H. Wiltshire and G. Gerdts (2012). Seasonal dynamics and modeling of a Vibrio community in coastal waters of the North Sea. Microbial ecology 63(3): 543-551. Oberbeckmann, S., A. Wichels, T. Maier, M. Kostrzewa, S. Raffelberg and G. Gerdts (2011). A polyphasic approach for the differentiation of environmental Vibrio isolates from temperate waters. Fems Microbiology Ecology 75(1): 145-162. Oberbeckmann, S., A. Wichels, K. H. Wiltshire and G. Gerdts (2011). Occurrence of Vibrio parahaemolyticus and Vibrio alginolyticus in the German Bight over a seasonal cycle. Antonie Van Leeuwenhoek International Journal of General and Molecular Microbiology 100(2): 291-307. Okada, K., A. Roobthaisong, S. Hamada and S. Chantaroj (2012). Molecular and genomic investigations into spreading and diversification of Vibrio cholerae O1 in Thailand. International Journal of Infectious Diseases 16: E235-E235. Okuda, N., M. Ishibashi, E. Hayakawa, T. Nishino, Y. Takeda, A. K. Mukhopadhyay, S. Garg, S. K. Bhattacharya, G. B. Nair and M. Nishibuchi (1997). Emergence of a unique O3:K6 clone of Vibrio parahaemolyticus in Calcutta, India, and isolation of strains from the same clonal group from Southeast Asian travelers arriving in Japan. Journal of clinical microbiology 35(12): 3150-3155. Oliver, J. D. (2005). Wound infections caused by Vibrio vulnificus and other marine bacteria. Epidemiology and Infection 133(3): 383-391. Ottaviani, D., F. Leoni, E. Rocchegiani, R. Mioni, A. Costa, S. Virgilio, L. Serracca, D. Bove, C. Canonico, A. Di Cesare, L. Masini, S. Potenziani, G. Caburlotto, V. Ghidini and M. M. Lleo (2013). An extensive investigation into the prevalence and the genetic and serological diversity of toxigenic Vibrio parahaemolyticus in Italian marine coastal waters. Environmental microbiology 15(5): 1377-1386. Panicker, G. and A. K. Bej (2005). Real-time PCR detection of Vibrio vulnificus in oysters: comparison of oligonucleotide primers and probes targeting vvhA. Applied and environmental microbiology 71(10): 5702-5709. Panicker, G., M. L. Myers and A. K. Bej (2004). Rapid Detection of Vibrio vulnificus in Shellfish and Gulf of Mexico Water by Real-Time PCR. Applied and environmental microbiology 70(1): 498-507. Parsons, M. B., K. L. Cooper, K. A. Kubota, N. Puhr, S. Simington, P. S. Calimlim, D. Schoonmaker-Bopp, C. Bopp, B. Swaminathan, P. Gerner-Smidt and E. M. Ribot (2007). PulseNet USA standardized pulsed- field gel electrophoresis protocol for subtyping of Vibrio parahaemolyticus. Foodborne pathogens and disease 4(3): 285-292. Pascual, J., M. C. Macian, D. R. Arahal, E. Garay and M. J. Pujalte (2010). Multilocus sequence analysis of the central clade of the genus Vibrio by using the 16S rRNA, recA, pyrH, rpoD, gyrB, rctB and toxR genes. International Journal of Systematic and Evolutionary Microbiology 60: 154-165. Paz, S., N. Bisharat, E. Paz, O. Kidar and D. Cohen (2007). Climate change and the emergence of Vibrio vulnificus disease in Israel. Environmental Research 103(3): 390-396. Pfeffer, C. and J. D. Oliver (2003). A comparison of thiosulphate-citrate-bile salts-sucrose (TCBS) agar and thiosulphate-chloride-iodide (TCI) agar for the isolation of Vibrio species from estuarine environments. Letters in Applied Microbiology 36(3): 150-151. Popovic, T., P. I. Fields, O. Olsvik, J. G. Wells, G. M. Evins, D. N. Cameron, J. J. Farmer, C. A. Bopp, K. Wachsmuth, R. B. Sack, M. J. Albert, G. B. Nair, T. Shimada and J. C. Feeley (1995). Molecular Subtyping of Toxigenic Vibrio-Cholerae-O139 Causing Epidemic Cholera in India and Bangladesh, 1992-1993. Journal of Infectious Diseases 171(1): 122-127. Powell, A., C. Baker-Austin, S. Wagley, A. Bayley and R. Hartnell (2013). Isolation of pandemic Vibrio parahaemolyticus from UK water and shellfish produce. Microbial ecology 65(4): 924-927. Prisyazhnaya, N. V., E. G. Plotnikova, O. V. Bueva, E. S. Korsakova, L. V. Dorofeeva, E. N. Il’ina, A. T. Lebedev and L. I. Evtushenko (2012). Application of MALDI-TOF mass spectrometry for - 131 - REFERENCES

differentiation of closely related species of the “Arthrobacter crystallopoietes” phylogenetic group. Microbiology 81(6): 696-701. Randa, M. A., M. F. Polz and E. Lim (2004). Effects of temperature and salinity on Vibrio vulnificus population dynamics as assessed by quantitative PCR. Applied and environmental microbiology 70(9): 5469-5476. Rappe, M. S. and S. J. Giovannoni (2003). The uncultured microbial majority. Annual review of microbiology 57: 369-394. Reidl, J. and K. E. Klose (2002). Vibrio cholerae and cholera: out of the water and into the host. Fems Microbiology Reviews 26(2): 125-139. Reil, M., M. Erhard, E. J. Kuijper, M. Kist, H. Zaiss, W. Witte, H. Gruber and S. Borgmann (2011). Recognition of Clostridium difficile PCR-ribotypes 001, 027 and 126/078 using an extended MALDI-TOF MS system. European Journal of Clinical Microbiology & Infectious Diseases 30(11): 1431-1436. Reissbrodt, R. (2004). New chromogenic plating media for detection and enumeration of pathogenic Listeria spp.--an overview. International journal of food microbiology 95(1): 1-9. Rezzonico, F., G. Vogel, B. Duffy and M. Tonolla (2010). Application of Whole-Cell Matrix-Assisted Laser Desorption Ionization-Time of Flight Mass Spectrometry for Rapid Identification and Clustering Analysis of Pantoea Species. Applied and environmental microbiology 76(13): 4497-4509. Ritchie, J. M., H. P. Rui, X. H. Zhou, T. Iida, T. Kodoma, S. Ito, B. M. Davis, R. T. Bronson and M. K. Waldor (2012). Inflammation and Disintegration of Intestinal Villi in an Experimental Model for Vibrio parahaemolyticus-Induced Diarrhea. Plos Pathogens 8(3). Rodriguez-Castro, A., J. Ansede-Bermejo, V. Blanco-Abad, J. Varela-Pet, O. Garcia-Martin and J. Martinez- Urtaza (2009). Prevalence and genetic diversity of pathogenic populations of Vibrio parahaemolyticus in coastal waters of Galicia, Spain. Environmental Microbiology Reports 2(1): 58-66. Rosche, T. M., Y. Yano and J. D. Oliver (2005). A rapid and simple PCR analysis indicates there are two subgroups of Vibrio vulnificus which correlate with clinical or environmental isolation. Microbiology and immunology 49(4): 381-389. Ruimy, R., V. Breittmayer, P. Elbaze, B. Lafay, O. Boussemart, M. Gauthier and R. Christen (1994). Phylogenetic Analysis and Assessment of the Genera Vibrio, Photobacterium, Aeromonas, and Plesiomonas Deduced from Small-Subunit Ribosomal-Rna Sequences. International Journal of Systematic Bacteriology 44(3): 416-426. Ruiz, G. M., T. K. Rawlings, F. C. Dobbs, L. A. Drake, T. Mullady, A. Huq and R. R. Colwell (2000). Global spread of microorganisms by ships - Ballast water discharged from vessels harbours a cocktail of potential pathogens. Nature 408(6808): 49-50. Ruppert, J., B. Panzig, L. Guertler, P. Hinz, G. Schwesinger, S. B. Felix and S. Friesecke (2004). Two cases of severe sepsis due to Vibrio vulnificus wound infection acquired in the Baltic Sea. European journal of clinical microbiology & infectious diseases : official publication of the European Society of Clinical Microbiology 23(12): 912-915. Ryzhov, V. and C. Fenselau (2001). Characterization of the protein subset desorbed by MALDI from whole bacterial cells. Analytical Chemistry 73(4): 746-750. Ryzhov, V. and C. Fenselau (2001). Characterization of the protein subset desorbed by MALDI from whole bacterial cells. Analytical chemistry 73(4): 746-750. Saffert, R. T., S. A. Cunningham, S. M. Ihde, K. E. M. Jobe, J. Mandrekar and R. Patel (2011). Comparison of Bruker Biotyper Matrix-Assisted Laser Desorption Ionization-Time of Flight Mass Spectrometer to BD Phoenix Automated Microbiology System for Identification of Gram-Negative Bacilli. Journal of clinical microbiology 49(3): 887-892. Sauer, S., A. Freiwald, T. Maier, M. Kube, R. Reinhardt, M. Kostrzewa and K. Geider (2008). Classification and identification of bacteria by mass spectrometry and computational analysis. PLoS ONE 3(7): e2843. Sawabe, T., K. Hayashi, J. Moriwaki, Y. Fukui, F. L. Thompson, J. Swings and R. Christen (2004). Vibrio neonatus sp nov and Vibrio ezurae sp nov isolated from the Gut of Japanese Abalones. Systematic and applied microbiology 27(5): 527-534. Sawabe, T., K. Kita-Tsukamoto and F. L. Thompson (2007). Inferring the evolutionary history of vibrios by means of multilocus sequence analysis. Journal of Bacteriology 189(21): 7932-7936. Schoolnik, G. K. and F. H. Yildiz (2000). The complete genome sequence of Vibrio cholerae: a tale of two chromosomes and of two lifestyles. Genome Biol 1(3): REVIEWS1016. Schroeder, J. P., J. G. Wallace, M. B. Cates, S. B. Greco and P. W. Moore (1985). An infection by Vibrio alginolyticus in an Atlantic bottlenose dolphin housed in an open ocean pen. J Wildl Dis 21(4): 437- 438. Sedas, V. T. P. (2007). Influence of environmental factors on the presence of Vibrio cholerae in the marine environment: a climate link. Journal of Infection in Developing Countries 1(3): 224-241. Seibold, E., T. Maier, M. Kostrzewa, E. Zeman and W. Splettstoesser (2010). Identification of Francisella tularensis by Whole-Cell Matrix-Assisted Laser Desorption Ionization-Time of Flight Mass Spectrometry: Fast, Reliable, Robust, and Cost-Effective Differentiation on Species and Subspecies Levels. Journal of clinical microbiology 48(4): 1061-1069. - 132 - REFERENCES

Shandera, W. X., J. M. Johnston, B. R. Davis and P. A. Blake (1983). Disease from Infection with Vibrio- Mimicus, a Newly Recognized Vibrio Species - Clinical Characteristics and Epidemiology. Annals of Internal Medicine 99(2): 169-171. Shapiro, R. L., S. Altekruse, L. Hutwagner, R. Bishop, R. Hammond, S. Wilson, B. Ray, S. Thompson, R. V. Tauxe and P. M. Griffin (1998). The role of Gulf Coast oysters harvested in warmer months in Vibrio vulnificus infections in the United States, 1988-1996. Vibrio Working Group. The Journal of infectious diseases 178(3): 752-759. Sharples, G. J. and R. G. Lloyd (1990). A Novel Repeated DNA-Sequence Located in the Intergenic Regions of Bacterial Chromosomes. Nucleic acids research 18(22): 6503-6508. Shinoda, S., T. Nakagawa, L. Shi, K. Bi, Y. Kanoh, K. Tomochika, S. Miyoshi and T. Shimada (2004). Distribution of virulence-associated genes in Vibrio mimicus isolates from clinical and environmental origins. Microbiology and immunology 48(7): 547-551. Simpson, J. M., J. W. Santo Domingo and D. J. Reasoner (2002). Microbial source tracking: State of the science. Environmental Science & Technology 36(24): 5279-5288. Singleton, F. L., R. Attwell, S. Jangi and R. R. Colwell (1982). Effects of Temperature and Salinity on Vibrio- Cholerae Growth. Applied and environmental microbiology 44(5): 1047-1058. Smith, C. J. and A. M. Osborn (2009). Advantages and limitations of quantitative PCR (Q-PCR)-based approaches in microbial ecology. FEMS microbiology ecology 67(1): 6-20. Smith, P. B., K. M. Tomfohrde, D. L. Rhoden and A. Balows (1972). API system: a multitube micromethod for identification of Enterobacteriaceae. Appl Microbiol 24(3): 449-452. Sogawa, K., M. Watanabe, K. Sato, S. Segawa, A. Miyabe, S. Murata, T. Saito and F. Nomura (2012). Rapid identification of microorganisms by mass spectrometry: improved performance by incorporation of in- house spectral data into a commercial database. Analytical and bioanalytical chemistry 403(7): 1811- 1822. Stephan, R., N. Cernela, D. Ziegler, V. Pfluger, M. Tonolla, D. Ravasi, M. Fredriksson-Ahomaa and H. Hachler (2011). Rapid species specific identification and subtyping of Yersinia enterocolitica by MALDI-TOF mass spectrometry. J Microbiol Methods 87(2): 150-153. Stypulkowska-Misiurewicz, H., K. Pancer and A. Roszkowiak (2006). Two unrelated cases of septicaemia due to Vibrio cholerae non-O1, non-O139 in Poland, July and August 2006. Euro Surveill 11(11): E061130 061132. Su, Y. C. and C. C. Liu (2007). Vibrio parahaemolyticus: A concern of seafood safety. Food Microbiology 24(6): 549-558. Suarez, S., A. Ferroni, A. Lotz, K. A. Jolley, P. Guerin, J. Leto, B. Dauphin, A. Jamet, M. C. Maiden, X. Nassif and J. Armengaud (2013). Ribosomal proteins as biomarkers for bacterial identification by mass spectrometry in the clinical microbiology laboratory. J Microbiol Methods 94(3): 390-396. Svitil, A. L., S. M. N. Chadhain, J. A. Moore and D. L. Kirchman (1997). Chitin degradation proteins produced by the marine bacterium Vibrio harveyi growing on different forms of chitin. Applied and environmental microbiology 63(2): 408-413. Swertz, O. C., F. Colijn, H. W. Hofstraat and B. A. Althuis (1999). Temperature, Salinity, and Fluorescence in Southern North Sea: High-Resolution Data Sampled from a Ferry. Environ Manage 23(4): 527-538. Tada, J., T. Ohashi, N. Nishimura, Y. Shirasaki, H. Ozaki, S. Fukushima, J. Takano, M. Nishibuchi and Y. Takeda (1992). DETECTION OF THE THERMOSTABLE DIRECT HEMOLYSIN GENE (TDH) AND THE THERMOSTABLE DIRECT HEMOLYSIN-RELATED HEMOLYSIN GENE (TRH) OF VIBRIO-PARAHAEMOLYTICUS BY POLYMERASE CHAIN-REACTION. Molecular and Cellular Probes 6(6): 477-487. Tall, A., A. Teillon, C. Boisset, R. Delesmont, A. Touron-Bodilis and D. Hervio-Heath (2012). Real-time PCR optimization to identify environmental Vibrio spp. strains. Journal of applied microbiology 113(2): 361- 372. Tamelander, J., L. Riddering, F. Haag and J. Matheickal (2010). Guidelines for Development of a National Ballast Water Management Strategy. GloBallast Monograph Series 18. Tarr, C. L., J. S. Patel, N. D. Puhr, E. G. Sowers, C. A. Bopp and N. A. Strockbine (2007). Identification of Vibrio isolates by a multiplex PCR assay and rpoB sequence determination. Journal of Clinical Microbiology 45(1): 134-140. Thomas, K. U., N. Joseph, O. Reveendran and S. Nair (2006). Salinity-induced survival strategy of Vibrio cholerae associated with copepods in Cochin backwaters. Marine pollution bulletin 52(11): 1425-1430. Thompson, C. C., F. L. Thompson, K. Vandemeulebroecke, B. Hoste, P. Dawyndt and J. Swings (2004). Use of recA as an alternative phylogenetic marker in the family Vibrionaceae. International Journal of Systematic and Evolutionary Microbiology 54: 919-924. Thompson, F. L., D. Gevers, C. C. Thompson, P. Dawyndt, S. Naser, B. Hoste, C. B. Munn and J. Swings (2005). Phylogeny and molecular identification of vibrios on the basis of multilocus sequence analysis. Applied and environmental microbiology 71(9): 5107-5115.

- 133 - REFERENCES

Thompson, F. L., B. Gomez-Gil, A. T. R. Vasconcelos and T. Sawabe (2007). Multilocus sequence analysis reveals that Vibrio harveyi and V-campbellii are distinct species. Applied and environmental microbiology 73(13): 4279-4285. Thompson, F. L., T. Iida and J. Swings (2004). Biodiversity of vibrios. Microbiology and molecular biology reviews : MMBR 68(3): 403-431, table of contents. Thompson, F. L., Y. Li, B. Gomez-Gil, C. C. Thompson, B. Hoste, K. Vandemeulebroecke, G. S. Rupp, A. Pereira, M. M. De Bem, P. Sorgeloos and J. Swings (2003). Vibrio neptunius sp. nov., Vibrio brasiliensis sp. nov. and Vibrio xuii sp. nov., isolated from the marine aquaculture environment (bivalves, fish, rotifers and shrimps). International Journal of Systematic and Evolutionary Microbiology 53(Pt 1): 245-252. Thompson, F. L., C. C. Thompson, B. Hoste, K. Vandemeulebroecke, M. Gullian and J. Swings (2003). Vibrio fortis sp. nov. and Vibrio hepatarius sp. nov., isolated from aquatic animals and the marine environment. International Journal of Systematic and Evolutionary Microbiology 53(Pt 5): 1495-1501. Thompson, F. L., C. C. Thompson, Y. Li, B. Gomez-Gil, J. Vandenberghe, B. Hoste and J. Swings (2003). Vibrio kanaloae sp. nov., Vibrio pomeroyi sp. nov. and Vibrio chagasii sp. nov., from sea water and marine animals. International Journal of Systematic and Evolutionary Microbiology 53(Pt 3): 753-759. Thompson, F. L., C. C. Thompson and J. Swings (2003). Vibrio tasmaniensis sp. nov., isolated from Atlantic salmon (Salmo salar L.). Systematic and applied microbiology 26(1): 65-69. Thompson, J. R., M. A. Randa, L. A. Marcelino, A. Tomita-Mitchell, E. Lim and M. F. Polz (2004). Diversity and dynamics of a north atlantic coastal Vibrio community. Applied and environmental microbiology 70(7): 4103-4110. Tison, D. L., M. Nishibuchi, J. D. Greenwood and R. J. Seidler (1982). Vibrio vulnificus biogroup 2: new biogroup pathogenic for eels. Applied and environmental microbiology 44(3): 640-646. Turner, J. W., B. Good, D. Cole and E. K. Lipp (2009). Plankton composition and environmental factors contribute to Vibrio seasonality. Isme Journal 3(9): 1082-1092. Turner, J. W., R. N. Paranjpye, E. D. Landis, S. V. Biryukov, N. Gonzalez-Escalona, W. B. Nilsson and M. S. Strom (2013). Population Structure of Clinical and Environmental Vibrio parahaemolyticus from the Pacific Northwest Coast of the United States. PloS one 8(2). Urdaci, M. C., M. Marchand, E. Ageron, J. M. Arcos, B. Sesma and P. A. D. Grimont (1991). Vibrio- Navarrensis Sp-Nov, a Species from Sewage. International Journal of Systematic Bacteriology 41(2): 290-294. Urmersbach, S., T. Alter, M. S. Koralage, L. Sperling, G. Gerdts, U. Messelhausser and S. Huehn (2014). Population analysis of Vibrio parahaemolyticus originating from different geographical regions demonstrates a high genetic diversity. BMC microbiology 14: 59. Vandenberghe, J., F. L. Thompson, B. Gomez-Gil and J. Swings (2003). Phenotypic diversity amongst Vibrio isolates from marine aquaculture systems. Aquaculture 219(1-4): 9-20. Verroken, A., M. Janssens, C. Berhin, P. Bogaerts, T. D. Huang, G. Wauters and Y. Glupczynski (2010). Evaluation of Matrix-Assisted Laser Desorption Ionization-Time of Flight Mass Spectrometry for Identification of Nocardia Species. Journal of clinical microbiology 48(11): 4015-4021. Versalovic, J., T. Koeuth and J. R. Lupski (1991). Distribution of Repetitive DNA-Sequences in Eubacteria and Application to Fingerprinting of Bacterial Genomes. Nucleic acids research 19(24): 6823-6831. Vezzulli, L., R. R. Colwell and C. Pruzzo (2013). Ocean Warming and Spread of Pathogenic Vibrios in the Aquatic Environment. Microbial ecology. Vos, M. and X. Didelot (2009). A comparison of homologous recombination rates in bacteria and archaea. The ISME journal 3(2): 199-208. Watanabe, H., S. Miyoshi, T. Kawase, K. Tomochika and S. Shinoda (2004). High growing ability of Vibrio vulnificus biotype 1 is essential for production of a toxic metalloprotease causing systemic diseases in humans. Microb Pathog 36(3): 117-123. Weisburg, W. G., S. M. Barns, D. A. Pelletier and D. J. Lane (1991). 16S ribosomal DNA amplification for phylogenetic study. Journal of Bacteriology 173(2): 697-703. Welker, M. and E. R. Moore (2011). Applications of whole-cell matrix-assisted laser-desorption/ionization time- of-flight mass spectrometry in systematic microbiology. Systematic and applied microbiology 34(1): 2- 11. Wetz, J. J., A. D. Blackwood, J. S. Fries, Z. F. Williams and R. T. Noble (2013). Quantification of Vibrio vulnificus in an Estuarine Environment: a Multi-Year Analysis Using QPCR. Estuaries and Coasts. WHO. (2013). from http://www.who.int/gho/epidemic_diseases/cholera. Wietz, M., L. Gram, B. Jorgensen and A. Schramm (2010). Latitudinal patterns in the abundance of major marine bacterioplankton groups. Aquatic Microbial Ecology 61(2): 179-189. Williams, T. L., D. Andrzejewski, J. O. Lay and S. M. Musser (2003). Experimental factors affecting the quality and reproducibility of MALDI TOF mass spectra obtained from whole bacteria cells. Journal of the American Society for Mass Spectrometry 14(4): 342-351.

- 134 - REFERENCES

Wilson, L. A. and P. M. Sharp (2006). Enterobacterial repetitive intergenic consensus (ERIC) sequences in Escherichia coli: Evolution and implications for ERIC-PCR. Molecular biology and evolution 23(6): 1156-1168. Wiltshire, K. H. and B. F. J. Manly (2004). The warming trend at Helgoland Roads, North Sea: phytoplankton response. Helgoland Marine Research 58(4): 269-273. Wittlinger, F., R. Steffen, H. Watanabe and H. Handszuh (1995). Risk of Cholera Among Western and Japanese Travelers. J Travel Med 2(3): 154-158. Wolters, M., H. Rohde, T. Maier, C. Belmar-Campos, G. Franke, S. Scherpe, M. Aepfelbacher and M. Christner (2011). MALDI-TOF MS fingerprinting allows for discrimination of major methicillin-resistant Staphylococcus aureus lineages. International Journal of Medical Microbiology 301(1): 64-68. Wong, H. C. and P. Wang (2004). Induction of viable but nonculturable state in Vibrio parahaemolyticus and its susceptibility to environmental stresses. Journal of Applied Microbiology 96(2): 359-366. Wybo, I., O. Soetens, A. De Bel, F. Echahidi, E. Vancutsem, K. Vandoorslaer and D. Pierard (2012). Species Identification of Clinical Prevotella Isolates by Matrix-Assisted Laser Desorption Ionization-Time of Flight Mass Spectrometry. Journal of clinical microbiology 50(4): 1415-1418. Yildiz, F. H. and K. L. Visick (2009). Vibrio biofilms: so much the same yet so different. Trends in Microbiology 17(3): 109-118. Zhou, S., Z. Hou, N. Li and Q. Qin (2007). Development of a SYBR Green I real-time PCR for quantitative detection of Vibrio alginolyticus in seawater and seafood. Journal of applied microbiology 103(5): 1897-1906.

- 135 - REFERENCES

- 136 - REFERENCES

ACKNOWLEDGMENTS

First of all I would like to thank my PhD Thesis Committee, namely Dr. Gunnar Gerdts, Dr. Antje Wichels, Prof. Matthias Ullrich and Prof. Frank-Oliver Glöckner for great supervision, fruitful discussions and valuable input. I would like to thank the BMBF and the Alfred Wegener Institute for funding. I am also grateful for scientific and financial support from the International Max Planck Research School of Marine Microbiology (MarMic) and from VibrioNet. Special thanks to Sarah Dehne and Hilke Döpke for technical support. I also would like to thank the students Anna Cardenas, Frederik Helmprobst, Torben Schierhorn, Nadja Torres Reyes, Leon Dlugosch, Ilka Hartmann, Sidika Kirmici and Annabell Hillenbrand for their valuable contribution to this thesis and the work in VibrioNet on Helgoland. Special thanks to the whole working group “Microbial Ecology” for all the support, in particular my former office-mates Eva Maria and Rebecca. I further would like to thank my friends for their motivation and the supporters of the A.D.C. Vogtland: Monkey & Jacqui, Rosi & Zipfelklatscher & Richie junior, Thommy & Franzi & Emil & Elise, the family Schmidt/Holle, J.K., the Wombats and the Hobbits. Last but not least I would like to thank my family for their invaluable support: my nephews Paula and Frieda, my sister Gogs and my parents.

- 137 - STATUTORY DECLARATION

- 138 - STATUTORY DECLARATION

STATUTORY DECLARATION

I, René Erler, hereby declare that I have written this PhD thesis independently, unless where clearly stated otherwise. I have used only the sources, the data and the support that I have clearly mentioned. This PhD thesis has not been submitted for conferral of degree elsewhere.

Treuen, January 13, 2015

Signature:______

- 139 - STATUTORY DECLARATION

- 140 -