OUP user menu

Mechanisms of egg contamination by Salmonella Enteritidis

Inne Gantois, Richard Ducatelle, Frank Pasmans, Freddy Haesebrouck, Richard Gast, Tom J. Humphrey, Filip Van Immerseel
DOI: http://dx.doi.org/10.1111/j.1574-6976.2008.00161.x 718-738 First published online: 1 July 2009


Salmonella Enteritidis (SE) has been the major cause of the food-borne salmonellosis pandemic in humans over the last 20 years, during which contaminated hen's eggs were the most important vehicle of the infection. Eggs can be contaminated on the outer shell surface and internally. Internal contamination can be the result of penetration through the eggshell or by direct contamination of egg contents before oviposition, originating from infection of the reproductive organs. Once inside the egg, the bacteria need to cope with antimicrobial factors in the albumen and vitelline membrane before migration to the yolk can occur. It would seem that serotype Enteritidis has intrinsic characteristics that allow an epidemiological association with hen eggs that are still undefined. There are indications that SE survives the attacks with the help of antimicrobial molecules during the formation of the egg in the hen's oviduct and inside the egg. This appears to require a unique combination of genes encoding for improved cell wall protection and repairing cellular and molecular damage, among others.

  • Salmonella Enteritidis
  • egg contamination
  • eggshell penetration
  • reproductive tract colonization
  • survival in the forming egg
  • growth in eggs post-lay

Eggs as the most important source of Salmonella Enteritidis (SE) infections in humans

The epidemiology of SE tells the story of a pathogen that has found a biological niche in table eggs. SE has caused the majority of food-borne outbreaks of salmonellosis reported worldwide since the mid-1980s. In the United States, 298 (80%) of the 371 known-source SE outbreaks from 1985 to 1999 were egg-associated (Patrick et al., 2004). In 2006, a total of 165 023 confirmed cases of human salmonellosis were reported in the European Union (EU) via the European Surveillance System (TESSy) (EFSA, 2007a). SE was identified as the cause of infection in 62.5% of the cases, and Salmonella Typhimurium in 12.9%. Other serotypes causing human illness are responsible for <2% of the human infections. Other serotypes among the top 10 causes of human salmonellosis cases in the EU are Infantis, Virchow, Newport, Hadar, Stanley, Derby, Agona and Kentucky. Eggs and egg products were the most often identified food vehicles in the Salmonella outbreaks (Braden et al., 2006). These findings clearly illustrate the link between eggs and human SE infections. An EU-wide study based on faecal and dust sampling from layer houses revealed that 30.8% of 5310 commercial large-scale laying hen holdings were Salmonella positive in 2006 (EFSA, 2007b). SE was the most common serotype in the laying flock environment (52.3%). Thus, almost 50% of the isolates were non-SE and the serotype distribution on layer farms does not match with that found in table eggs. The overall EU prevalence of Salmonella in table eggs was 0.8% in 2006, and >90% of all egg isolates were SE (EFSA, 2007a). These data should be interpreted with caution because the sampling site is not specified. The remaining 10% of the isolates belonged to different Salmonella serotypes, but these were mostly isolated in only one EU member state, indicating that the overall importance of non-SE serotypes in eggs is negligible. These data suggest that SE has some intrinsic characteristics that allow a specific interaction with either the reproductive organs of laying hens or the egg components.

Generally, there are two possible routes of egg contamination by Salmonella. Eggs can be contaminated by penetration through the eggshell from the colonized gut or from contaminated faeces during or after oviposition (horizontal transmission) (Messens et al., 2005a; De Reu et al., 2006). The second possible route is by direct contamination of the yolk, albumen, eggshell membranes or eggshells before oviposition, originating from the infection of reproductive organs with SE (vertical transmission) (Timoney et al., 1989; Keller et al., 1995; Miyamoto et al., 1997; Okamura et al., 2001a, b). Figure 1 shows a schematic representation of the egg pathogenesis. It is not yet clear as to which route is most important for SE to contaminate the egg contents. Although some authors claim horizontal transmission to be the most important way to contaminate eggs (Barrow & Lovell, 1991; Bichler et al., 1996), most authors claim that vertical transmission is the most important route (Gast & Beard, 1990; Miyamoto et al., 1997; Guard-Petter, 2001).

Figure 1

Pathogenesis of egg contamination by Salmonella. (a) Salmonella is orally taken up by the hen and enters the intestinal tract. Bacteria colonizing the intestinal lumen are able to invade the intestinal epithelial cells (gut colonization). As a consequence, immune cells, more specifically macrophages, are attracted to the site of invasion and enclose the Salmonella bacteria. This allows the bacteria to survive and multiply in the intracellular environment of the macrophage. These infected macrophages migrate to the internal organs such as the reproductive organs (systemic spread). In addition to systemic spread, bacteria can also access the oviduct through ascending infection from the cloaca. (b) One possible route of egg contamination is by Salmonella penetration through the eggshell and shell membranes after outer shell contamination. Surface contamination may be the result of either infection of the vagina or faecal contamination. (c) The second possible route is by direct contamination of the yolk, yolk membranes, albumen, shell membranes and egg shell originating from infection of the ovary, infundibulum, magnum, isthmus and shell gland, respectively. (d) Salmonella bacteria deposited in the albumen and on the vitelline membrane are able to survive and grow in the antibacterial environment. They are also capable of migrating to and penetrating the vitelline membrane in order to reach the yolk. After reaching this rich environment, they can grow extensively.

This review provides an overview of host–pathogen interactions in the hen reproductive tract and eggs at the cellular and molecular level. It aims to highlight potential differences between SE and other Salmonella serotypes that could allow SE strains to contaminate eggs more successfully than other serotypes.

Internal egg contamination after penetration of the eggshell

Outer shell contamination

Following oviposition, any contaminated environment in the area of the laid egg, such as the nest box, the hatchery environment or the hatchery truck, can lead to outer shell contamination. The presence of chicken manure and other moist organic materials facilitates the survival and growth of Salmonella by providing the required nutrients and a degree of physical protection. When eggs are artificially contaminated on the shell with faeces containing Salmonella and subsequently stored at 25 °C, numbers increase by 1–2 logs by day 1 and 4–5 logs by day 3 (Schoeni et al., 1995). Such a growth indicates that faeces can serve as a nutritional reservoir for Salmonella. However, Salmonella can also survive and grow on the eggshell in the absence of faecal contamination, especially at lower temperatures and a low relative humidity (Messens et al., 2006). Salmonella bacteria probably survive for a longer time at a low temperature due to the slower metabolism induced by the disadvantageous conditions on the dry eggshell surface (Radkowski et al., 2002). The egg surface can also be contaminated within the hen reproductive system after formation of the shell, but this will be discussed further in the text (Humphrey et al., 1991a). Presuming that no differences exist between different Salmonella serotypes in the interaction with the outer shell, more prevalent serotypes such as SE are more likely to contaminate egg surfaces. In order to reduce the risk of externally contaminated eggs in the food chain, the need to rapidly remove any faecal contamination should be emphasized. However, intensive control measures in the United States, such as examining eggs for cracks, and washing and disinfecting eggs, have not eliminated egg contamination with SE (Braden et al., 2006). It is, however, possible that penetration has occurred before examining and washing the eggs.

Factors influencing eggshell and membrane penetration

Bacteria can easily penetrate through a cracked egg shell (Fajardo et al., 1995). The intact egg, however, possesses three physical barriers to bacterial penetration (Fig. 2). These are the cuticle, which is a hydrophobic proteinaceous layer covering the eggshell and the pore openings, the crystalline eggshell and the shell membranes (Ruiz & Lunam, 2002). Shell membranes consist of three different layers, i.e. the inner and the outer membrane, consisting of a network of randomly oriented fibres, and a homogenous third layer of electro-dense material called the limiting membrane, demarcating the membrane at the interface with the albumen (Wong-Liong et al., 1997).

Figure 2

Schematic representation of the egg structure.

In addition to their function as a physical barrier, the eggshell and shell membranes also act as a chemical barrier. Although antibacterial proteins have been identified mainly in the albumen, proteins with well-known antibacterial properties have also been associated with the eggshell and shell membranes. Lysozyme is abundant in the limiting membrane and is also present in the shell membranes, the matrix and the cuticle of the eggshell (Hincke et al., 2000). Ovotransferrin has also been identified in the eggshell membranes and the basal calcified layer, possibly acting as a bacteriostatic filter (Gautron et al., 2001). Recently, ovocalyxin-36, a novel chicken eggshell and eggshell membrane protein, has been identified (Gautron et al., 2006). An antimicrobial role for ovocalyxin-36 was proposed because its protein sequence is highly similar to lipopolysaccharide-binding proteins, bactericidal/permeability-increasing (BPI) proteins and Plunc family proteins. These proteins are involved in antibacterial defence, and therefore it is believed that ovocalyxin-36 is of particular importance to keep the eggs free from pathogens. Protein extracts derived from the cuticle and the outer eggshell matrix indeed possess antimicrobial properties against both Gram-positive and Gram-negative bacteria (Hincke & Wellman-Labadie, 2007). Three bacterial species, Pseudomonas aeruginosa, Bacillus cereus and Staphylococcus aureus, were found to be inhibited in the presence of soluble eggshell matrix proteins, and it was demonstrated that these proteins might interact and disrupt the membrane integrity of the bacteria (Mine et al., 2003). On the other hand, Escherichia coli and SE were weakly inhibited only at an early stage of incubation time (up to 4 h).

In spite of the protective physical and chemical barriers, numerous researchers have demonstrated rapid penetration into the egg by various bacteria, including Salmonella (Williams et al., 1968; Humphrey et al., 1989, 1991b). Miyamoto et al. (1998a) observed that after exposing freshly laid eggs to a Salmonella suspension for 2 h at 25 °C, the inner eggshell and egg contents were contaminated. Several studies investigated the various factors affecting the probability of bacterial penetration. Both intrinsic and extrinsic factors are highlighted in a review by Messens et al. (2005a). The eggshell appears to be more easily penetrated immediately after the egg is laid (Sparks & Board, 1985; Padron et al., 1990; Miyamoto et al., 1998a). It is suggested that for the first minutes after oviposition, the cuticle is immature and some pores may be open. Moreover, when the egg is exposed to an environment cooler than the chicken body temperature (42 °C), a negative pressure may develop and the bacteria migrate more easily through the eggshell and membranes (Board, 1966; Bruce & Drysdale, 1994). In addition, the cuticle in older eggs becomes dehydrated, resulting in its shrinkage, and the pores become more exposed to bacterial penetration (Mayes & Takeballi, 1983). In recent studies (De Reu et al., 2006; Messens et al., 2007), it was reported that cuticle deposition is important for the prevention of penetration, and in the absence of cuticle deposition, penetration is a frequent event. However, some research groups (Nascimento et al., 1992; Messens et al., 2005b) observed no correlation between cuticle deposition and penetration of Salmonella through the eggshell. Additionally, bacterial penetration was found to be independent of the pore number (Nascimento et al., 1992; Messens et al., 2005b; De Reu et al., 2006). As mentioned earlier, temperature is also an important factor affecting the penetration. Fast penetration is observed when a positive temperature differential is created between the egg (warm) and the bacterial suspension (cool) (Mayes & Takeballi, 1983; Bruce & Drysdale, 1994). It is believed that a positive temperature differential, combined with the presence of moisture, provides an ideal opportunity for the bacteria to penetrate the eggshell (Berrang et al., 1999). The use of different penetration models, differences in the bacterial strains used, differences in the number of bacteria inoculated, the temperature and relative humidity during storage and the egg characteristics (eggshell quality and egg age) may partly explain the conflicting results seen in the studies regarding eggshell penetration, as reviewed by Messens et al. (2005a).

Eggshell penetration by different Salmonella serotypes and other bacterial species

It has been well demonstrated that penetration of the eggshell and shell membranes is not a unique characteristic of SE and that other Salmonella serotypes, and even unrelated bacteria, are capable of passing through these barriers (Sauter & Petersen, 1969, 1974; Mayes & Takeballi, 1983; Jones et al., 2002; De Reu et al., 2006). In a comparative study, the penetration of seven selected bacterial species originally isolated from egg contents was assessed using two different egg penetration models (De Reu et al., 2006). The results indicate that Gram-negative, motile and nonclustering bacteria penetrate the eggshell most frequently. Using an agar model, i.e. filling eggs with agar and dipping in a bacterial suspension, Pseudomonas sp. (60%), Alcaligenes sp. (58%) and SE (43%) traversed the eggshell most frequently. However, using intact eggs dipped in a bacterial suspension, egg contents were most frequently contaminated by SE (33%), followed by Carnobacterium sp. (17.5%) and Acinetobacter baumannii (14.8%). The results obtained by the two experimental eggshell penetration assays suggest that shells can be penetrated by various bacterial species, but that SE has mechanisms to survive and/or grow in the internal egg contents, in contrast to the other bacterial species. In a study of naturally infected flocks, numerous Salmonella serotypes, such as Enteritidis, Typhimurium and Hadar, were isolated from eggshells, whereas only Enteritidis was isolated from egg contents (Humphrey et al., 1991b). Interestingly, only one egg was positive in both sites, suggesting that internal egg contamination is more likely to occur during formation of the egg rather than by penetration through the shell. Moreover, the relative prevalence of non-Enteritidis serotypes in faecal samples (measured by the overshoe method) of laying hen flocks (50%) is not consistent with the high prevalence of SE in table eggs (90%) (EFSA, 2007a). All these data support the idea that eggshell and egg membrane penetration are not a specific property of SE and that other characteristics of this serotype are related to egg contamination. These could include the ability to colonize the hen reproductive tract and survival and multiplication inside eggs, both of which could contribute to the epidemiological association of SE with eggs.

Contamination of eggs during egg formation

Colonization of the reproductive organs

Several lines of evidence support the view that egg contamination with SE is more likely to take place during the formation of the egg in the reproductive organs than by eggshell penetration. In several studies, SE was isolated from the reproductive tissue of infected birds, in the absence of intestinal colonization (Lister et al., 1988). Moreover, SE is capable of persistence in reproductive tissues of naturally and experimentally infected hens, even though the animals generate an innate and adaptive immune response to the infection, indicating that the bacteria can reside intracellularly and escape the host defence mechanisms. The deposition of Salmonella inside eggs is thus most likely a consequence of reproductive tissue colonization in infected laying hens (Keller et al., 1995; Methner et al., 1995; Gast & Holt, 2000a). Very little is known, however, about the exact site in reproductive tissues where the bacteria reside and the bacterial and host factors that play a role in the association between the reproductive tissue and Salmonella. The oviduct can be subdivided into five functional regions. Starting from the ovary, there are the infundibulum, magnum, isthmus, uterus and vagina. The infundibulum captures the ovulatory follicles, the magnum produces the albumen, the isthmus deposits the eggshell membranes, the uterus forms the eggshell and the vagina is involved in oviposition. Salmonella colonizing the oviduct could be incorporated into the albumen, the eggshell membranes or the eggshell itself, depending on the site of colonization (magnum, isthmus and uterus, respectively). Although SE has been isolated from both the yolk and the albumen, according to most authors, the albumen is most frequently contaminated, pointing to the oviduct tissue as the colonization site (Gast & Beard, 1990; Humphrey et al., 1991b; Keller et al., 1995; Miyamoto et al., 1997; De Buck et al., 2004c). However, some studies found the yolk to be primarily contaminated, suggesting the ovary to be the primary colonization site (Bichler et al., 1996; Gast & Holt, 2000a; Gast et al., 2002). One report indicated that several Salmonella strains colonized the ovary significantly more often than the oviduct, but were deposited at similar frequencies in the yolk and the albumen (Gast et al., 2007). Because of the very low incidence of egg contamination in natural infections and the fact that it is very labour-intensive to examine large numbers of eggs, not enough studies have been carried out to definitely establish the principal site of contamination. It is thus difficult, based on the contamination site in eggs, to predict the most important colonization site of Salmonella in the reproductive tract. However, it would be reasonable to suggest that, given that SE can be isolated from all sites in the hen reproductive tract, that contamination of any part of the egg is possible. An overview of all studies on internal egg contamination through reproductive organ colonization is presented in Table 1.

View this table:
Table 1

Overview of studies carried out to analyse the internal egg contamination through infection of the reproductive organs

Inoculation routeStrainInoculation dose (log10 CFU mL−1)General resultEgg contamination rate
Timoney et al. (1989)OralSE PT46The relatively high frequency of internal egg contamination clearly demonstrates the potential for egg transmission of SEYolk: 9.6% Egg white: 3.6%
Shivaprasad et al. (1990)Oral, intravenous, cloacalSE PT88.3–8.6SE was only cultured from the yolk and egg white of a small number of eggs until 11 days postinfectionYolk: 0.4% Egg white: 1.5%
Gast & Beard (1990)Oral, contact transmissionSE PT13a9Although a high contamination rate of egg white and yolk was observed, Salmonella could not be recovered from any yolk content sample, suggesting that Salmonella is contaminating the vitelline membraneYolk: 18.5% (first week) Egg white: 20% (first week) Yolk contents: 0%
Humphrey et al. (1991a)OralSE PT43, 6, 8There was no relationship between the contamination of the egg contents and antibody status, faecal excretion or the dose administeredTotal egg content: Inoculum 3: 3.5% Inoculum 6: 0% Inoculum 8: 0%
Thiagarajan et al. (1994)OralSE PT8, SE PT28Not mentionedSE can colonize the preovulatory follicles at different stages of development. It is therefore suggested that SE remains attached to the vitelline membrane instead of contaminating the yolk contentPreovulatory follicle (16 birds): Membrane: 10 positive samples Yolk content: four positive samples Laid eggs: Yolk: eight positive yolks Egg white: three positive egg whites
Keller et al. (1995)OralSE8The contamination rate of forming eggs is much higher than the contamination rate of laid eggs, indicating that antibacterial factors within the egg may control the pathogen before the egg is laidForming eggs: 27.1–31.4% Freshly laid eggs: 0–0.6%
Methner et al. (1995)OralSE10No correlation was found between the contamination of the eggshell and that of the egg contentYolk: 0% Egg white: 0.4%
Bichler et al. (1996)OralSE10The hens produced SE-positive eggs at high frequencies in the first week postinfection.In the first week postinfection: Eggshell washing: 26.5% Egg content: 2.9% Egg white: 43% Yolk: 41%
Keller et al. (1997)OralSE, Salmonella Typhimurium8SE and Salmonella Typhimurium may be equal in their potential to colonize the tissues of the reproductive tract and forming eggs, but only SE was isolated from egg contents after ovipositionTotal egg content: Salmonella Typhimurium: 0%
Miyamoto et al. (1997)Intravenous, intravaginal, cloacalSE PT47Intravaginal and cloacal inoculation resulted in the colonization of only the lower oviduct whereas intravenous infection resulted in colonization of the entire oviductTotal egg content/shell: Intravenous: 11.5%/7.7% Intravaginal: 9.6%/12% Cloacal: 0%/4.6%
Williams et al. (1998)OralSalmonella Typhimurium DT1047These experiments have demonstrated that DT104 can contaminate the egg contents after oral infectionTotal egg content: 2.1%
Miyamoto et al. (1998b)Intravaginal, cloacalSE PT47After intravaginal inoculation, SE was recovered from the uterus and after cloacal inoculation SE was recovered from the vagina, indicating that SE only ascends to the lower parts of the oviductTotal egg content/shell Intravaginal: 5%/15% Cloacal: 0%/0%
Gast & Holt (1998)OralSE PT137This study has shown that infection of 1-day-old chicks can lead to frequent intestinal colonization and occasional egg contamination when these birds matureTotal egg content: 0.44%
Leach et al. (1999)Oral, aerosolSalmonella Typhimurium DT1047, 2–4The egg contamination rate in aerosol-infected birds was much higher compared with orally infected birdsTotal egg content: Oral: 1.7% Aerosol: 14–25%
Gast & Holt (2000a)OralTwo SE PT4 and one SE PT13a9For all three isolates, the incidence of yolk contamination was significantly higher than the incidence of egg white contamination and no significant difference was observed between the SE strainsYolk: 2.5% Egg white: 0.5%
Kinde et al. (2000)Oral, intravenousSE PT49, 6In the orally infected birds, 43% of the reproductive organs were positive, compared with 83% in the intravenously infected birdsTotal egg content: 2.6%
Okamura & Holt (2001a, b)IntravaginalSE, Salmonella serotypes Typhimurium, Infantis, Hadar, Heidelberg and Montevideo6This study suggests that SE has a specific advantage over the other Salmonella serotypes by its capacity to colonize the vaginal tissues of hensEgg content/shell SE: 7.5%/25% Salmonella Typhimurium: 3.1%/1.6% Salmonella Infantis: 0%/4% Salmonella Hadar: 0%/4.9% Salmonella Heidelberg: 0%/4.5% Salmonella Montevideo: 0%/1.9%
Okamura (2001a, b)IntravenousSE, Salmonella serotypes Typhimurium, Infantis, Hadar, Heidelberg and Montevideo6This study suggests that SE is the predominant serovar to colonize the reproductive organs of laying hens among the six serotypes testedYolk/egg white SE: yolk: 6.9%/2.3% Other Salmonella serotypes: 0%/0%
Wigley et al. (2001)OralSalmonella Pullorum9One-day-old chicks orally infected with Salmonella Pullorum produced contaminated eggs frequently during the period of sexual maturity as a consequence of reproductive tract colonizationTotal egg content: 6.5%
Gast & Holt (2001)OralSE PT13a9Deposition of SE within egg yolks appears to occur infrequently and SE is mostly deposited on the vitelline membraneTotal yolk: 4.3% Yolk content: 0.5%
Gast et al. (2002)Oral, aerosol, intravenousSE PT13a9, 9 and 5–7No significant differences were observed in egg contamination among the three inoculation routesYolk: 4–7% Egg white: 0–2%
Gast et al. (2003)OralSE PT13a wild type (WT), passaged SE PT13a (spleen, liver) passaged SE PT13a (oviduct and ovary)9Passaged SE strains recovered from ovaries and oviducts induced a significantly higher incidence of egg contamination than the WT SE strainTotal egg content: SE PT13a WT: 8.27% Passaged SE PT13a (spleen, liver): 10.41% Passaged SE PT13a (oviduct and ovary): 17%
Gast et al. (2004)OralSE, Salmonella Heidelberg9There was no significant difference in reproductive tract colonization between the two serotypes, but Salmonella Heidelberg was recovered from the eggs at lower frequencies than SETotal egg content: SE: 7.0% Salmonella Heidelberg: 1.1–4.5%
De Buck et al. (2004c)IntravenousSE PT47The infected birds produced the highest frequency of contaminated eggs in the first week postinfectionShell: 17.4% Yolk: 20.3% Egg white: 4.3%
Gast et al. (2005b)OralSE, 2 Salmonella Heidelberg strains and passaged variants of each WT strain9The Salmonella-passaged strains caused a significantly higher frequency of egg contamination than did the WT strains. Furthermore, no correlation was found between the duration of faecal shedding and the production of contaminated eggsTotal egg content: SE WT: 5% SE passaged strain: 8.84% Salmonella Heidelberg WT 1: 1.63% Salmonella Heidelberg passaged strain 1: 4.95% Salmonella Heidelberg WT 2: 3.14% Salmonella Heidelberg passaged strain 2: 5%
Gast et al. (2007)OralSE PT13a, SE PT14b, Salmonella Heidelberg9The frequency of ovarian colonization was significantly higher than the frequency of recovery from the oviduct for all three Salmonella strains, but no corresponding difference was observed between the incidence of deposition in yolk or egg white. The incidence of egg contamination with SE was higher than that of Salmonella HeidelbergSE: Yolk: 5.5% Egg white: 4.1% Salmonella Heidelberg: Yolk: 1.5% Egg white: 1.8%
Gantois et al. (2008c)Intravenous2 SE strains, Salmonella serotypes Typhimurium, -Heidelberg, Virchow, Hadar8The SE strains showed a higher colonization of the reproductive organs in comparison with the Salmonella serotypes Heidelberg, Virchow and Hadar. No significant difference was observed between the SE strains and the Salmonella Typhimurium strainTotal egg content (positive eggs/total eggs): SE1: 4/5 SE2: 3/5 Salmonella Typhimurium: 2/5 Salmonella Heidelberg: 0/6 Salmonella Virchow: 1/16 Salmonella Hadar: 0/16

It is generally believed that colonization of the reproductive organs is a consequence of systemic spread of Salmonella from the intestine (Vazquez-Torres et al., 1999). Invasion in the intestinal epithelial cells triggers infiltration of immune cells, mainly macrophages, resulting in the uptake of bacteria by these cells. Because of its capability to survive and replicate in the immune cells, bacteria carried in the macrophages are spread within the host, resulting in colonization of the reproductive organs (Keller et al., 1995; Miyamoto et al., 1997; Okamura et al., 2001a,b; Gast, 2007; Gantois et al., 2008c). Salmonella pathogenicity island-2 (SPI-2) is essential in the ability to spread within the host and to cause a systemic infection (Jones et al., 2001). Using a deletion mutant in the regulator of SPI-2 (ssrA), it was shown that after intravenous infection of laying hens, the bacterial numbers of the ssrA mutant were significantly lower in the oviducts and the ovaries as compared with the wild-type strain. These reduced ssrA colony counts in the reproductive organs point to a role for SPI-2 in the spread or the colonization of the reproductive tract tissues (Bohez et al., 2008).

Colonization of the reproductive organs has also been shown to be a consequence of systemic spread after airborne infections (Baskerville et al., 1992; Leach et al., 1999). It was even observed that the contamination rate of eggs was much higher following an aerosol challenge of the laying hens than following an oral challenge (Leach et al., 1999).

Colonization of the ovary

The extensive permeability of the vascular endothelia observed in the ovary may contribute to the high colonization rate at this site (Griffin et al., 1984). In the majority of experimental studies in laying hens, a higher frequency of ovary colonization is reported, compared with the frequency of recovery from the oviduct (De Buck et al., 2004b; Gantois et al., 2006; Gast et al., 2007). Therefore, it is strongly believed that SE must interact with the cellular components of the preovulatory follicles. It was indeed shown that SE can attach to developing and mature follicular granulosa cells exhibiting different attachment patterns (Thiagarajan et al., 1994). Higher bacterial numbers in the membranes of the preovulatory follicles than in the yolk itself suggest that during transovarian transmission, SE remains attached to the egg vitelline membranes. A previous study has also suggested that yolk contamination is more often associated with the vitelline membrane than with the interior yolk contents (Gast & Beard, 1990; Gast & Holt, 2000a). It has been noticed that in vitro attachment of SE to granulosa cells may involve binding to fibronectin (Thiagarajan et al., 1996a). Furthermore, a major role of the type 1 fimbriae in the attachment process was suggested because the in vitro attachment of SE to granulosa cells was inhibited by preincubation of the cells with purified fimbrial preparation (Thiagarajan et al., 1996a). There are also indications that Salmonella can invade and multiply in granulosa cells (Thiagarajan et al., 1996a). Howard et al. (2005) compared the ability of Salmonella to invade ovarian follicles at different stages of follicular maturity in vitro: the small white follicles (immature) were more susceptible to Salmonella invasion than the more mature small and large yellow ones. These authors believe that the penetration of immature follicles has practical implications because it can lead to contamination of eggs after maturation and can cause continuous transovarian infection of eggs throughout the reproductive cycle. This statement is, however, questionable because not all small white follicles will mature and because the extensive growth of Salmonella in the nutrient-rich follicles will most likely lead to their degeneration (Kinde et al., 2000).

The fact that Salmonella can interact with the cellular components of preovulatory follicles raises the question as to whether serotype Enteritidis harbours some intrinsic characteristics allowing it to specifically interact with these cells and, as a consequence, be transmitted to eggs. In a study by Okamura et al. (2001a, b), it was shown that among six different Salmonella serotypes, Enteritidis colonized ovaries and preovulatory follicles at significantly higher levels than five other serotypes after intravenous inoculation. Because samples in this study were only taken at 4 and 7 days postinfection, and bacteria were still persistent in the peripheral blood, it cannot be concluded, however, that SE displays a stronger interaction with follicles than other serotypes. Similar results were obtained by Gantois et al. (2008c) in an intravenous infection model, demonstrating a higher affinity of the serotype Enteritidis for the ovary compared with other Salmonella serotypes (Hadar, Virchow and Infantis), except for Typhimurium. The fact that SE and Salmonella Typhimurium may be equally capable of colonizing the ovary is in accordance with the data obtained by Keller et al. (1997). Studies comparing invasion of the serotypes Enteritidis and Typhimurium in ovarian follicles in vitro yielded conflicting results (Howard et al., 2005; Mizumoto et al., 2005). Based on the fact that systemic spread is a characteristic of most Salmonella serotypes, it is believed that ovarian colonization is not a specific trait allowing the serotype Enteritidis to contaminate eggs. However, the possibility that SE has a specific ability to interact and invade the preovulatory follicles cannot be ruled out. A large-scale study using multiple strains from different Salmonella serotypes should be carried out in order to provide more information regarding the serotype specificity of ovarian colonization and persistence. High levels of nutrients are available to bacteria invading ovarian follicles. Therefore, it is to be expected that this should lead to extensive replication of the bacteria, almost inevitably resulting in follicular degeneration. Because this is not a common phenomenon in naturally infected laying hens, as the laying percentage is usually not reduced, follicle colonization is not believed to be an important source of egg contamination, although this is under debate.

Colonization of the oviduct

Although several studies reported the vitelline membrane as the most common site of Salmonella contamination (Bichler et al., 1996; Gast & Holt, 2000a; Gast et al., 2002), other reports point to albumen as the principal site of contamination in eggs (Shivaprasad et al., 1990; Humphrey et al., 1991b; Keller et al., 1995), indicating that SE is colonizing oviduct tissues. Miyamoto et al. (1997) observed that developing eggs in a highly contaminated oviduct are likely to be Salmonella positive. Colonization of the reproductive tract can be the result of an ascending infection from the cloaca (Reiber et al., 1995; Miyamoto et al., 1997), a descending infection from the ovary (Keller et al., 1995) and/or a systemic spread of Salmonella. Depending on the site of contamination, i.e. the vagina, isthmus and magnum, Salmonella could be incorporated into the eggshell, the eggshell membranes or the albumen.

Vaginal colonization

Several authors have focused on the role of the vagina in the production of SE-contaminated eggs (Barrow & Lovell, 1991; Keller et al., 1995; Reiber et al., 1995; Miyamoto et al., 1999; Okamura et al., 2001a,b; Mizumoto et al., 2005). It is believed that intravaginal infection tends to ascend only to the lower parts of the oviduct because Salmonella is rarely recovered from the ovary and the upper oviduct in intravaginally inoculated hens (Miyamoto et al., 1997, 1998b; Okamura et al., 2001a, b). These studies obtained high egg contamination rates after intravaginal infection, indicating a high risk of contamination (primarily eggshell contamination) as the egg passes through a heavily colonized vagina. When the egg is laid, penetration through the eggshell can occur, due to suction of the organisms into eggs under the negative pressure caused by cooling of the egg (Schoeni et al., 1995; Miyamoto et al., 1998a). In spite of the fact that it is difficult to distinguish between contamination during formation of the egg or after oviposition, internal egg contamination after vaginal colonization most likely occurs after penetration of the eggshell and not by internal contamination following ascending infection of the upper oviduct, although this cannot be ruled out. In a comparative study with six different Salmonella serotypes, significantly higher numbers of SE were recovered from the vagina in comparison with strains belonging to other serotypes after intravaginal inoculation (Miyamoto et al., 1998a). The authors suggested a higher ability of the serotype Enteritidis to attach to the vaginal epithelium. It was also noticed that the rank order of the Salmonella invasiveness in vaginal epithelium was dependent on the lipopolysaccharide type, namely lipopolysaccharide type O9 (SE)>lipopolysaccharide type O4 (Salmonella Typhimurium, Salmonella Heidelberg and Salmonella Agona)>lipopolysaccharide type O7 (Salmonella Montevideo and Salmonella Infantis) and lipopolysaccharide type O8 (Salmonella Hadar) (Mizumoto et al., 2005).

Isthmus and magnum colonization

It is clear that different segments of the oviduct can be colonized by SE. Using different infection models, the tubular glands of the isthmus were identified as the predominant colonization site of SE in the oviduct by De Buck et al. (2004a). Colonization of the isthmus can result in contaminated eggshell membranes. These observations are in accordance with other experimental studies (Bichler et al., 1996; Miyamoto et al., 1997; Okamura et al., 2001a, b). In principle, eggshell membrane contamination can also be a consequence of penetration of Salmonella bacteria after deposition on the shell during the passage through the vagina rather than direct contamination of the eggshell membranes during passage through the isthmus. In addition, when culturing the eggshell and egg contents separately, some albumen sticks to the eggshell, making the interpretations of the shell membranes as the site of egg contamination even more complex.

Numerous studies suggest that SE most frequently migrates into eggs through the upper oviduct in association with the albumen (Gast & Beard, 1990; Hoop & Pospischil, 1993; Humphrey & Whitehead, 1993; Schoeni et al., 1995). Detection of SE associated with secretory cells of the upper and lower magnum by immunohistochemical staining is in agreement with the hypothesis that the pathogen may contaminate forming eggs through the albumen (Hoop & Pospischil, 1993; Keller et al., 1995; De Buck et al., 2004a). Recently, the abilities to invade and proliferate in isthmus and magnum oviduct cells of different Salmonella serotypes were assessed using a tubular gland cell primary culture model. All serotypes tested were equally able to invade and proliferate in the glandular epithelial cells, suggesting that invasion and proliferation in oviduct cells is most likely not a unique characteristic of the serotype Enteritidis (Gantois et al., 2008c). In the study of Gantois et al. (2008c), it was also shown that a Salmonella serotype Enteritidis and Typhimurium strain colonized the oviduct to higher levels than strains belonging to the serotypes Heidelberg, Virchow and Hadar, even if all serotypes invaded oviduct cells in vitro. This is in accordance with a previous intravenous infection study by Okamura et al. (2001a, b), demonstrating that of six serotypes, only Enteritidis and Typhimurium were able to colonize the reproductive organs at days 4 and 7 postinoculation. One-day-old chicks that were orally infected with the chicken-adapted Salmonella Pullorum produced a high amount of contaminated eggs (6.5%) during the period of sexual maturity as a consequence of reproductive organ colonization (Wigley et al., 2001). Isolates of SE and Salmonella Pullorum, together with isolates of Salmonella Gallinarum and Salmonella Dublin, form a related strain cluster that share the same lipopolysaccharide-based O-antigen structure (O-1, 9, 12, characteristic of serogroup D). Comparative genome analysis of SE and Salmonella Gallinarum indicated that these serotypes are highly related and that Salmonella Gallinarum may be a direct descendant of SE (Thomson et al., 2008). It can be speculated that these two serotypes harbour the same characteristics, allowing them to efficiently contaminate eggs, but this is not clear.

Reproductive tract colonization: a matter of controversy

It is difficult to make comparisons between different experimental studies attempting to determine the preferred site of colonization or the strongest colonizing and persisting serotype. Indeed, experimental infection studies use different strains, inoculation methods, infection doses and laboratory techniques for bacteriological analysis. Individual strains of Salmonella (within and across serotype boundaries) can differ considerably in their ability to contaminate eggs (Gast & Holt, 2000a, 2001). Four Salmonella Heidelberg strains colonized the ovaries and oviducts of inoculated hens at frequencies similar to SE, but were found significantly less often inside eggs (Gast et al., 2004). Phenotypic attributes, such as the ability to produce high-molecular-mass lipopolysaccharide and the ability to grow to a high cell density, have been linked to an enhanced capability of egg contamination by SE (Guard-Petter et al., 1997; Guard-Petter et al., 1998). Recently, a set of small nucleotide polymorphisms (SNPs), which differ in two SE strains that vary in egg contamination, were identified (Guard-Bouldin et al., 2006). In addition, a high-throughput phenotype microarray assaying the growth of bacteria in response to 1920 different culture conditions revealed that these two strains show dramatic differences in amino acid and nucleic acid metabolism, which is most likely correlated to the SNPs (Morales et al., 2005). Furthermore, it was shown that serial passage through the reproductive organs also enhances egg contamination (Gast et al., 2003, 2005a). This indicates that the selective pressure in the reproductive tissues may promote the induction of specific bacterial properties, resulting in an elevated egg contamination. Numerous studies have also been performed to study the effect of the inoculation route on the production of contaminated eggs (Miyamoto et al., 1997; Gast et al., 2002). While Gast et al. (2002) reported that oral, aerosol and intravenous inoculations led to similar frequencies of egg contamination, Miyamoto et al. (1997) observed a higher contamination rate when birds were inoculated intravenously and intravaginally. Moreover, it should be taken into account that using different laboratory techniques to isolate the bacteria from eggs may have an impact on the outcome of the experiments. In some studies, the yolk samples are cultured together with the vitelline membrane and thus some albumen, while other studies extract yolk contents and thus do not culture the vitelline membrane. Furthermore, extending the incubation time from 24 to 48 h can increase the isolation rate of SE from eggs significantly (Humphrey & Whitehead, 1992), meaning that studies that do not take account of this may have underestimated the prevalence of egg contamination. The use of different pre-enrichment and enrichment media can also result in different outcomes (Humphrey & Whitehead, 1992). For isolation of Salmonella from whole eggs, it was found that the Rappaport Vassiliadis broth was superior to Selenite broth as a selective medium (Humphrey & Whitehead, 1992). Additionally, the outcomes of Salmonella infections may also be influenced by host susceptibility characteristics, such as the breed or the line of chickens (Beaumont et al., 1994, 1999; Keller et al., 1995; Kinde et al., 2000). Some studies reported that brown-egg layers are more susceptible than white-egg layers (Keller et al., 1995; Kinde et al., 2000), and on comparing four different lines of chickens, it was found that one line was more susceptible to SE than others (Protais et al., 1996). All these aspects indicate that care should be taken when interpreting data obtained from experimental infections.

Virulence factors associated with oviduct colonization

In order to gain a better understanding of the molecular mechanisms allowing the serotype Enteritidis to interact with the hen's reproductive tract and to adapt to this particular ecological niche, a genome-wide screen was carried out by Gantois et al. (2008b) to identify genes expressed in the oviduct, using in vivo expression technology. This study identified the genes involved in cell wall integrity, regulation of fimbrial operons, amino acid and nucleic acid metabolism, stress response and motility as being highly induced during colonization of the reproductive tract. This indicates that the oviduct is a stressful and damaging environment for Salmonella bacteria, but it also indicates that the bacteria can counteract this by stress-induced protective and reparative responses, enabling the bacteria to survive in the hostile environment and/or escape the host defence reactions.

Other SE factors that play a role in oviduct infections are fimbriae (De Buck et al., 2003, 2004b; Li et al., 2003). Li et al. (2003) were the first to identify binding sites for fimbriated SE in the chicken oviduct. The binding of type 1 fimbriae to glycosphingolipids and gangliosides from the oviduct mucosa is not uniform along this organ, and is mainly in the infundibulum. De Buck et al. (2003) clearly demonstrated that SE isolates are able to adhere to immobilized secretions of the oviduct. These authors showed that the receptor of adhesion is also localized inside the tubular gland cells of the isthmus and adhesion is blocked by the addition of mannose, indicating that the adhesion is mediated by type 1 fimbriae. Conversely, a study using SE isolates differing in their fimbrial expression found no difference in infection of the reproductive organs and eggs, making their role equivocal (Thiagarajan et al., 1996b). Nevertheless, a screening for promoters induced in the albumen showed that SE fimZ is highly induced during incubation at 42 °C. This may mean that when Salmonella resides extracellularly in the oviduct lumen, in the presence of albumen, the transcription of type 1 fimbriae will be activated, resulting in bacterial attachment to the secretory glandular cells (own unpublished data).

There is mounting evidence that lipopolysaccharide is also of particular importance for SE persistence in the reproductive tract tissues. Lipopolysaccharide is a major component of the outer membrane of Gram-negative bacteria and a prime target for recognition by the innate immune system. At least two different functions have been attributed to lipopolysaccharides with respect to persistent reproductive tract infection. It has been suggested that the composition of lipopolysaccharides is important in determining the survival of SE in avian macrophages and these cells may be a site where SE resides in the oviduct (He et al., 2006). Different levels of attachment of different Salmonella serotypes to chicken vaginal explants possibly also involve a role of the lipopolysaccharide structure (Mizumoto et al., 2005). Furthermore, it has been shown that high-molecular-weight lipopolysaccharide in SE is correlated with increased egg contamination (Guard-Petter et al., 1997). Although the exact role of high-molecular-weight lipopolysaccharide is not yet known, its presence has been correlated with an unusual pathology of the reproductive tract, although this was not reflected in higher egg contamination (Parker et al., 2002).

The type 3 secretion systems-1 and -2 (T3SS-1 and T3SS-2) may also play a role in egg contamination. T3SS-1 is mainly associated with bacterial invasion of the intestinal epithelium via the concerted action of effector proteins (Zhou et al., 1999), while T3SS-2 is responsible for the establishment of systemic infection by promoting the intracellular survival of Salmonella in macrophages, as mentioned earlier. Li et al. (2008) were the first to confirm the pathogenic role of T3SS-1 and T3SS-2 effectors in SE invasion and intracellular survival in chicken oviduct epithelial cells. It is believed that invasion and survival in tubular gland cells of the oviduct is not specific for serotype Enteritidis (Gantois et al., 2008c). Most likely, functions exerted by T3SS-1 and T3SS-2 are also required by Salmonella serotypes other than Enteritidis to invade and survive inside chicken oviduct epithelial cells (Jones et al., 2002). Furthermore, a recent study suggested that inactivation of ssrA, a regulator of T3SS-2, rendered SE unable to colonize the chicken reproductive tract successfully (Bohez et al., 2008).

Meanwhile, it has become clear that the process of oviduct colonization is complex and depends on many factors including fimbriae, flagellae, lipopolysaccharide, cell wall structure and stress tolerance. Although most, if not all, bacterial factors, shown to play a role in reproductive tract colonization, are not specific to the serotype Enteritidis, a unique regulation of these known virulence factors in the reproductive tract environment could be one plausible explanation for the epidemiological association with hen's eggs. This, however, has not been shown yet. It was demonstrated that repeated in vivo passages through the reproductive tissues of chickens increase the ability of an SE strain to induce internal egg contamination, whereas serial passage through the liver and the spleen did not affect the ability of the strain to cause egg contamination (Gast et al., 2003). This is an indication that interaction of SE with the reproductive tissues may either induce or select for the expression of microbial properties important for egg contamination. The complementarity between phenotypic traits with relevance for colonization and survival in different tissues may allow SE to traverse the complex series of events between the introduction of infection and the deposition inside eggs (Gast et al., 2002).

The interaction between Salmonella and the forming egg

Transfer of Salmonella from the hen to the egg

The survival of SE in forming eggs has been considered crucial for internal egg contamination (Keller et al., 1995, 1997). Salmonella colonizing reproductive organs can potentially be incorporated into the forming egg, provided the contamination of the egg contents does not lead to an abortive egg formation and provided the bacteria are not killed by the albumen.

Yolk contamination can occur due to ovary colonization by Salmonella. Degeneration of follicles in the ovary has often been observed after experimental Salmonella infections, most likely caused by extensive growth in the nutrient-rich yolk at chicken body temperature, 42 °C (Kinde et al., 2000). Interestingly, extensive growth in whole eggs when stored at room temperature (20–25 °C) does not lead to changes in the colour, smell and consistency of the egg contents (Humphrey & Whitehead, 1993), suggesting that the process of yolk degeneration is dependent on physiological factors, such as the temperature. Degeneration of the ovarian follicles would result in a decline in the production cycle and thus no production of eggs containing contaminated yolks. The extent to which the yolk contents become positive after infection of the ovary is, however, not clear. After intravenous inoculation, SE cells are confined to the interstitial tissues and not to the yolk contained in the large follicles (Barrow & Lovell, 1991). Moreover, experimental studies have suggested that Salmonella are far more likely to be deposited on the outside of the vitelline membrane rather than inside the nutrient-rich yolk during ovary colonization (Gast & Holt, 2001). Inoculation of Salmonella onto the vitelline membrane in an in vitro egg contamination model demonstrated that some strains were capable of penetrating into the yolk contents at a low frequency during 24 h of incubation at 30 °C (Gast et al., 2005b), but a similar study reported no positive yolk contents samples after incubation for 24 h at 42 °C (Guan et al., 2006). Moreover, SE multiplication on the exterior vitelline membrane both preceded and exceeded multiplication resulting from penetration into the yolk contents during 36 h of incubation at 30 °C (Gast et al., 2008). This suggests a low invasion of yolk contents by Salmonella during egg formation. These data support the possibility of incorporation into the egg by carriage on the vitelline membrane.

Albumen or shell membrane contamination would occur when Salmonella colonizes the upper oviduct. According to Keller et al. (1995), the infection of the forming egg occurs at this site, before eggshell deposition. Indeed, after oral infection with SE, about one-third of the forming eggs were positive compared with 0.6% of the freshly laid eggs (Keller et al., 1995). This reduction clearly suggests that antibacterial factors within the albumen can exert a degree of control of SE in forming eggs. During the c. 26 h required for the formation of an egg, the ovum spends c. 5 h in the magnum, where it is surrounded by the albumen, followed by the addition of two shell membranes in the isthmus. The remaining 21 h are required for shell deposition in the uterus, after which the completed egg is moved through the vagina to pass through the cloaca as it is laid (Solomon et al., 1997). Survival in the forming egg could be a possible reason for selective isolation of the serotype Enteritidis in laid eggs, provided that this serotype harbours intrinsic or induced factors related to albumen resistance. Oral infection of laying hens with three different SE and Typhimurium strains revealed that both serotypes are equally able to colonize tissues of the reproductive tract and forming eggs in the oviduct before oviposition. However, only SE, but not Salmonella Typhimurium, was isolated from egg contents after oviposition (Keller et al., 1997), suggesting survival strategies of SE inside the forming eggs. Nevertheless, Salmonella Typhimurium DT104 was shown to contaminate the egg contents after oral infection of laying hens (Williams et al., 1998).

Antimicrobial components in the albumen and vitelline membrane

The reproductive tract produces antimicrobial components that are incorporated into the albumen, and that are growth restricting for Salmonella. The most well known are lysozyme and ovotransferrin. Lysozyme is a muramidase affecting the cell wall of Gram-positive bacteria (Hughey & Johnson, 1987), but that has also been shown to form pores in the cell wall of Gram-negative bacteria (Gast et al., 2005a). Ovotransferrin possesses two distinct mechanisms of bacteriostatic action against bacteria. The first is iron chelation, which creates an iron-deficient environment for bacteria (Mayes & Takeballi, 1983). The second is a direct interaction with the membrane and induction of damage to biological functions of the bacterial cytoplasmic membrane (Ibrahim et al., 1998, 2000).

Another group of antimicrobial proteins are those showing proteinase-inhibiting activity. They include ovomucoid, ovoinhibitors (serine protease inhibitors), cystatin (a cysteine protease inhibitor) and ovostatin (Stevens et al., 1991). Their function lies in inhibiting tryptic digestion of egg proteins by bacteria and thus protection of the antimicrobial activity of albumen proteins.

A recent study identified 11 types of gallinacins (β-defensins) expressed in the segments of the oviduct (Abdel Mageed et al., 2008). Defensins are antimicrobial peptides that play significant roles in innate immunity (Sugiarto & Yu, 2004; Higgs et al., 2005). The greatest expression of gallinacins was seen in the infundibulum and the vagina (Ohashi et al., 2005). Recently, Yoshimura et al. (2006) reported that the expression of gallinacin-1, -2 and -3 was increased within 24 h in response to SE infection or in response to purified lipopolysaccharides in cultured vaginal cells. The study by Abdel Mageed et al. (2008) confirmed that gallinacin-3 expression was enhanced by lipopolysaccharide in vivo. Escherichia coli lipopolysaccharide injection in laying hens also induced gallinacin expression in the ovarian follicles (Subedi et al., 2007). β-Defensin-11 was identified to be present in the chicken albumen as well (Mann, 2007). Remarkably, the albumen contains many proteins that are connected in some way to lipopolysaccharide binding or modification. One such albumen component is similar to the mammalian acyloxyacyl hydrolase, known to cleave acyl chains from bacterial lipopolysaccharide. Furthermore, proteins containing BPI domains have been identified. These usually occur in proteins binding and neutralizing lipopolysaccharide and thus eventually mediating the destruction of bacteria (Elsbach & Weiss, 1998). Such domains also occur in Tenp, a protein recently identified as an albumen component (Guérin-Dubiard et al., 2006), and in the eggshell-specific protein ovocalyxin-36 (Gautron et al., 2006). Recently, Silphaduang et al. (2006) were the first to report the presence of histones H1 and H2B as antimicrobial proteins in the avian reproductive system, but their functional significance in the chicken reproductive tract remains obscure. In the human placenta, histones H2A and H2B show a dose-dependent inhibition of the endotoxin activity of lipopolysaccharide by binding to and therefore blocking both the core and the lipid A moieties (Kim et al., 2002).

It is not known to what degree these antibacterial components affect different Salmonella serotypes. It is, however, striking that the function of most albumen proteins is linked to lipopolysaccharide binding. Given that the O-antigen structure of lipopolysaccharide is a major determinant of serotype specificity, it may be that the lipopolysaccharide structure plays a major role in the SE survival in forming eggs in vivo and that the lipopolysaccharide chemotype will affect the degree of binding with antimicrobial components and thus bacterial survival.

Besides the albumen, the vitelline membrane can also become contaminated (Bichler et al., 1996; Gast & Holt, 2000a; Gast et al., 2002). Recently, a proteomic analysis of the chicken egg vitelline membrane was carried out (Mann et al., 2008). Most of the components of the vitelline membrane that were known previously from other egg compartments, such as lysozyme, ovalbumin, ovotransferrin, ovomucin and lysozyme, also constitute c. 60% of the dry weight of the outer vitelline membrane. One outer vitelline membrane protein was identified as β-defensin 11.

Immunoglobulins are considered to belong to the antimicrobial defence system of avian eggs. Antibodies to Salmonella have been detected in the egg albumen and yolk from naturally and experimentally infected chickens (Schiemann & Montgomery, 1991; Desmidt et al., 1996). It has been suggested that antibodies transferred to the yolk after hyperimmunization of laying hens have no influence on the multiplication of Salmonella in the yolk (Takase et al., 1999; Gürtler & Fehlhaber, 2004). In contrast, Holt et al. (1996) described a significant difference in the growth behaviour of SE under the influence of antibodies. These authors, however, performed their experiments by inoculating Salmonella in a mixture of albumen and yolk, while the two previous studies were based on inoculations in separated yolk. Thus, the possibility exists that the antimicrobial components of the albumen had an additional inhibitory effect on Salmonella. The exact antimicrobial role of immunoglobulins in avian eggs thus remains to be defined.

Salmonella virulence factors affecting SE survival in the forming egg

The important role of lipopolysaccharides in conferring protection against the bactericidal component albumen was recently confirmed by Gantois et al. (2008a). Applying in vivo expression technology, the rfbH gene, involved in lipopolysaccharide O-antigen synthesis, was found to be transcriptionally induced during growth in whole eggs at room temperature. After inoculation of a SalmonellaΔrfbH strain in albumen at 42 °C, immediate killing was observed while the wild-type strain was able to survive in albumen during 24 h. Moreover, the ΔrfbH mutant was also unable to grow in whole eggs at room temperature. Lu et al. (2003) suggested that yafD and xthA play an essential role in the repair of DNA damage caused by the albumen and hence confer an advantage to SE to survive in forming chicken eggs. In a recent paper using transposon mutagenesis, it was found that the majority of genes associated with SE survival in albumen at 37 °C are involved in either cell wall structure/function or nucleic and amino acid metabolism (Clavijo et al., 2006). Two mutants had insertions in genes unique to SE. One is homologous to a restriction endonuclease and the other is the pef operon encoding a fimbrial biosynthesis gene. Both genes were transformed into a Salmonella Typhimurium strain, but only the former conferred an enhanced survival in albumen. The same study also demonstrated that survival in the albumen at 37 °C was higher for serotype Enteritidis compared with Typhimurium and E. coli. It is striking that many antimicrobial proteins in albumen bind to lipopolysaccharides while cell wall and lipopolysaccharide biosynthetic genes of Salmonella seem to play an important role in albumen survival.

Survival of different Salmonella serotypes in the forming egg

Some research groups have compared the survival capabilities of strains belonging to different serotypes in albumen at different incubation temperatures, yielding conflicting results. In one study, a similar survival for SE strains and Salmonella Typhimurium strains at 37 and 42 °C was shown (Guan et al., 2006). These findings are in contrast with earlier studies, demonstrating an enhanced survival in the albumen at 37 °C for the serotype Enteritidis (Lu et al., 2003; Clavijo et al., 2006). In the latter studies, strains were inoculated in the albumen of 1-week-old eggs, while in the study by Guan et al. (2006), the albumen of freshly laid eggs was used. It is assumed that fresh albumen (pH=8.16) enhances growth compared with stored albumen (pH=9.26), and this is most likely caused by the lower pH of the former (Messens et al., 2004). However, the results presented in a paper by Humphrey & Whitehead (1993) suggest that storage has little direct impact on the albumen with respect to the growth of SE. The lysozyme and ovotransferrin concentrations in the albumen increased with the hen's age throughout the laying period, which is reflected in an increased bacteriostatic effect of the albumen on SE at the mid and the final laying period, possibly influencing the data (Sellier et al., 2007).

It is striking that studies comparing strains belonging to different serotypes in their ability to survive in albumen at chicken body temperature are missing. Recently, the bactericidal effect of albumen at chicken body temperature was examined for five different Salmonella serotypes (Gantois et al., 2008c). Remarkably, the strains belonging to the serotypes Enteritidis, Typhimurium and Heidelberg were able to survive in the hostile environment of the albumen for 24 h, while the strains belonging to the serotypes Virchow and Hadar were very susceptible to the albumen, and after 24 h, almost all bacteria were killed. This could explain why Salmonella Virchow and Salmonella Hadar are almost never associated with eggs. However, not enough isolates were compared to draw general conclusions from this study.

Growth of SE in eggs post-lay

Growth patterns of SE in eggs

The risk of human infections following consumption of Salmonella-contaminated eggs depends on the bacterial numbers present. SE can grow in the contents of naturally contaminated eggs at room temperature (Humphrey & Whitehead, 1993). Cogan et al. (2001) observed growth after 8 days at 20 °C in 7% of whole eggs inoculated in the albumen near the shell with as few as 2 CFU. It is clear that this implies a serious threat to human health because extensive growth in eggs does not lead to changes in the colour, smell and consistency of the egg contents (Humphrey & Whitehead, 1993). After experimental and natural infections, some authors point to the albumen as being most frequently contaminated (Gast & Beard, 1990; Humphrey et al., 1991b), while others point to the vitelline membrane as the most common contamination site (Bichler et al., 1996; Gast & Holt, 2000a, 2001; Gast et al., 2007). The albumen is growth restricting for Salmonella because it contains multiple antimicrobial components, inducing bacterial cell wall and DNA damage (see Growth patterns of SE in eggs). At temperatures <10 °C, Salmonella bacteria are unable to grow in the albumen (Braun & Fehlhaber, 1995; Schoeni et al., 1995). At room temperature, data are conflicting and it is difficult to compare the various studies because the inoculation size, strains, incubation temperatures and period, age of eggs and many other factors vary (Humphrey & Whitehead, 1993; Braun & Fehlhaber, 1995; Schoeni et al., 1995; Gast & Holt, 2000b). Recent data showed that, at 20 °C, upon inoculation with 39 CFU mL−1 albumen, both SE and non-SE strains are able to grow in separated fresh albumen samples up to >106 CFU mL−1 (Clavijo et al., 2006) and, on extending the incubation time, the number of samples with pronounced growth increased further. Numerous other studies also observed the growth of SE in egg albumen at room temperature (Braun & Fehlhaber, 1995; Schoeni et al., 1995; Duboccage et al., 2001), indicating that the Salmonella bacteria harbour intrinsic characteristics to counteract the attacks of the antimicrobial components present in the egg albumen. Inoculation of bacteria in the egg albumen of whole eggs resulted in faster growth than separated egg albumen and also high numbers of Salmonella bacteria were detected in the yolk, indicating migration towards the yolk (Cogan et al., 2001; Messens et al., 2004).

It is believed that Salmonella cells that are deposited in the albumen are able to migrate to and penetrate through the vitelline membrane in the egg post-lay, in order to reach the yolk and thus gain access to a pool of nutrients that are necessary for its survival and growth. Rapid and extensive multiplication of SE in the nutrient-rich egg yolks at 25 °C has been reported (Gast & Holt, 2000b). This was confirmed in a recent study showing that all strains multiplied rapidly in yolk contents and reached c. 9.0 log cells mL−1 after 24 h of incubation at 37 or 42 °C (Guan et al., 2006). Data from contaminated eggs from either naturally (Humphrey & Whitehead, 1992) or artificially (Gast & Beard, 1992) infected hens suggest that there is a delay before yolk penetration and fast growth occurs in yolk, in eggs stored at room temperature. This is believed to be because the vitelline membrane in fresh eggs inhibits yolk invasion by Salmonella. Gradually, the integrity of the vitelline membrane will become lost during storage, resulting in leakage of nutrients into the albumen. This is considered to allow the bacteria to migrate to the vitelline membrane and multiply and invade the yolk (Humphrey & Whitehead, 1993).

Experimentally infected laying hens also often deposit SE on the vitelline membrane (Bichler et al., 1996; Gast & Holt, 2000a, 2001; Gast et al., 2007). A recent study has suggested that substantial growth occurred in association with the vitelline membrane before penetration through the membrane (Gast et al., 2008a). Contamination of the nutrient-rich yolk content is uncommon (Gast & Holt, 2001; Gast et al., 2003), but if it occurs, it leads to rapid bacterial multiplication (Gast & Holt, 2000b; Guan et al., 2006). There is considerable controversy on the major deposition site in eggs. Most likely, the use of different isolation techniques has a major impact on the outcome of the experiments. Therefore, the use of imaging such as 3D localization with bioluminescent or fluorescent Salmonella bacteria could help to unravel the preferred deposition site in eggs (Chen et al., 1996).

Studies on the growth of Salmonella in eggs have usually been performed after direct artificial infection of the eggs because the production of Salmonella-positive eggs is low after either natural (Humphrey et al., 1989; Humphrey & Whitehead, 1992) or experimental (Gast & Beard, 1992) infection of the hens. The Salmonella growth profiles seen in naturally contaminated eggs are different from those seen in eggs contaminated artificially. The latter suggests that growth is rapid in most eggs and that yolk invasion is common (Braun & Fehlhaber, 1995; Chen et al., 1996). The experimental studies, however, used either high contamination levels (104 cells per egg) (Chen et al., 1996), most likely not representative for naturally contaminated eggs, or buffered peptone water for the SE solution to be injected, which enhances bacterial growth in the albumen (Chen et al., 1996). Data obtained from artificial egg contamination models should thus be interpreted with caution. A model for artificial egg contamination mimicking the natural situation was developed by Cogan et al. (2001). Using low numbers of bacterial cells in a low-nutrient, low-iron suspension as inoculum, a low level of growth was detected in eggs, comparable to that seen in naturally contaminated eggs. When few Salmonella bacteria are deposited in the albumen, very little bacterial multiplication occurs and SE can persist there at supportive temperatures (Lock & Board, 1992; Hammack et al., 1993; Gast & Holt, 2000b). Humphrey & Whitehead (1993) observed that in artificially contaminated eggs, the inoculum increased c. 10-fold during the first 24 h postinoculation, as confirmed by Gast & Holt (2000b). The initial growth phase may involve the bacterium using its iron reserves, which appear to be sufficient to support about four generations. When the iron reserves are exhausted, cells enter a lag phase, where, in the majority of eggs, there is little or no change in the numbers of Salmonella organisms. It has been postulated that there may be leakage of nutrients from the yolk, leading to a bacterial attraction towards the yolk, some weeks after storage. Cogan and colleagues (Baron et al., 1997) provide evidence indicating that the bacteria will then (after weeks of storage) attach to and penetrate through the vitelline membrane and gain access to the yolk contents in order to grow (Baron et al., 1997).

Salmonella virulence factors related to growth in whole eggs post-lay

It is logical to assume that all Salmonella factors that play a role in albumen survival or growth in forming eggs will also play a role in whole egg survival post-lay.

Using in vivo expression technology, it was demonstrated that expression of SE rfbH, a lipopolysaccharide O-antigen biosynthesis gene, was strongly induced in eggs at room temperature (Gantois et al., 2008a). The rfbH gene was shown to be crucial for growth in eggs at room temperature. This again demonstrates the importance of lipopolysaccharide in survival in the albumen. This may be important in the first weeks of egg storage.

During storage, the vitelline membrane gradually deteriorates, resulting in the release of nutrients into the albumen, possibly attracting bacteria that can penetrate the vitelline membrane and multiply in the nutrient-rich yolk. It is tempting to speculate that leakage out of the yolk into the albumen would generate a gradient of amino acids, sugars or other yolk components, triggering a chemotactic movement towards the vitelline membrane. This hypothesis is consistent with the fact that SE grows rapidly in eggs only after c. 28 days of storage at room temperature (Gantois et al., 2008a). Flagella are thought to be necessary components for bacterial migration towards the vitelline membrane in whole eggs (Baron et al., 1997). Nonmotile mutants, such as fliC and motAB mutants, were unable to move through the albumen towards the yolk; hence, proliferation did not take place (Baron et al., 1997). Moreover, the nonmotile serotypes Pullorum and Gallinarum were also not capable of growing in egg contents. Motility is indeed a significant factor for chemotactic bacteria to move towards higher concentrations of attractants and to avoid higher concentrations of repellents by sensing temporal changes in chemoeffector concentrations (Cogan et al., 2004). Chemotaxis in E. coli is the best-studied signal-transduction network of any living organism. It allows E. coli to sense amino acids, sugars, dipeptides and even redox, temperature and pH changes. In response to these chemical changes, a signal cascade of methylation/demethylation and phosphorylation/dephosphorylation is switched on. Binding of chemoeffectors to the transmembrane receptors triggers the Che operon, which transmits the signals to flagellar motors. The net result of this signal cascade is a change in the direction of flagellar motor rotation and thus induction of motility towards or away from a certain trigger. Salmonella harbours a similar chemotaxis system reacting to chemoeffector stimuli (Bourret et al., 1989; Sourjik et al., 2004).

The vitelline membrane comprises a collagenous matrix overlaid with a layer of glycoproteins (Mariconda et al., 2006). It is believed that curli fimbriae (Sef17) mediate bacterial adherence to these glycoproteins such as fibronectin (Bellairs et al., 1963). Yolk invasion and thus multiplication of a curli-deficient strain, an agfA mutant, occurred significantly less than that in the wild-type SE strain (Lock & Board, 1992). It is, therefore, suggested that curli fimbriae are needed to attach to the vitelline membrane, in order to facilitate yolk invasion and multiplication (Baron et al., 1997). The expression of curli fimbriae has been investigated under poor (stationary phase) and rich (exponential phase) nutrient conditions for 15 different Salmonella strains in a high-pH and iron-restricted medium at 20 °C (Baron et al., 1997). A correlation has been found between the expression of curli fimbriae during the late exponential phase and a high frequency of growth in eggs at room temperature. This suggests that when bacteria move closer to the yolk, they will start to grow exponentially, and thus, strains that show better growth in eggs are those that are able to express curli fimbriae under growth, rather than starvation. The fact that genes encoding curli fimbriae appear to be ubiquitous within the genus Salmonella (Collinson et al., 1991) does not mean that all Salmonella serotypes are equally effective at multiplying in eggs because the expression of curli fimbriae can be regulated differently depending on the environmental triggers and the bacterial growth phase.

Growth in eggs post-lay by different Salmonella serotypes

Numerous research reports have compared growth in eggs between different Salmonella serotypes using various artificial egg contamination models. Messens et al. (2004) showed that, at 20 °C, after inoculation with 39 CFU mL−1 in the separated albumen, both SE and non-Enteritidis serotypes were able to grow up to >106 CFU mL−1. These findings were in contrast to those of Lock and Board (Hammack et al., 1993). In the latter study, the SE strains showed slow growth in the albumen at 20 and 30 °C, while the majority of the strains belonging to other serotypes did not multiply, but only survived in the albumen. Data obtained from growth studies in albumen should be interpreted with caution because most studies report variable and inconsistent results for independent repeats of the experiments (Lock & Board, 1992; Hammack et al., 1993; Messens et al., 2004). Nevertheless, according to most studies, the capacity to persist and grow in the albumen at non-hen body temperatures is not a specific trait of serotype Enteritidis.

Penetration through the vitelline membrane provides an opportunity for extensive bacterial multiplication inside the yolk. Penetration of SE through the vitelline membrane has been reported in several in vitro egg contamination experiments (Humphrey & Whitehead, 1993; Gast & Holt, 2000b). Recently, a number of studies have compared penetration of other Salmonella serotypes using the in vitro contamination model described by Gast and colleagues (Doran et al., 1993; Gast et al., 2005b; Guan et al., 2006; Murase et al., 2006; Gantois et al., 2008c). In this model, the yolk and the albumen are separated and then the yolk is inoculated with c. 100 CFU of Salmonella onto the exterior surface of the vitelline membrane, after which the albumen of one single egg is gently poured onto the yolk. All reports agree that serotypes, other than Enteritidis, are able to invade the vitelline membrane and multiply in egg yolk. Furthermore, the multiplication of different Salmonella serotypes was assessed by inoculating very small numbers of Salmonella cells in fresh eggs according to the egg infection model described by Cogan and colleagues (Cogan et al., 2001; Gantois et al., 2008c). Except for the serotype Typhimurium, no significant difference was observed between the serotype Enteritidis and the strains belonging to the serotypes Heidelberg, Virchow and Hadar, suggesting that the multiplication strategies inside eggs at room temperature are not unique for the serotype Enteritidis. The Salmonella Typhimurium strain displayed the lowest frequency of yolk invasion (Gantois et al., 2008c). However, this finding is in strong contrast with a finding by Cogan and colleagues (Baron et al., 1997), showing the highest frequency of yolk invasion for the serotype Typhimurium. Despite the variability seen within and between experiments and the fact that even within the serotype Enteritidis different growth patterns have been observed in eggs post-lay (Baron et al., 1997), the epidemiological association of SE with eggs is assumed not to be caused by specific multiplication strategies in eggs post-lay because most Salmonella serotypes are equally effective in multiplying in eggs post-lay.


As opposed to mammals, the chicken embryo does not develop in the safe environment of the womb, continuously protected by the dam's immune system. Hence, it is not surprising that the egg has an impressive arsenal of antimicrobial protective mechanisms, including both nonspecific physical barriers and highly efficacious microbiocidal molecules. Although it is possible to infect eggs with various bacterial species under the artificial conditions of a laboratory experimental set-up, under natural conditions, this is a rare event. When it occurs, it usually causes so much damage that the egg will be easily identified as being infected. SE is unique in the way that it can pass into the egg and multiply inside it without inducing noticeable changes. Combining this exceptional trait with the pathogenicity for the human intestinal tract allowed this serotype of Salmonella to cause a pandemic that has lasted for more than a quarter of a century. Only now are we beginning to understand the mechanism by which SE contaminates chicken eggs much more successfully than any other Salmonella serotype. Evidence is accumulating that contamination of the eggs is not by penetration through the shell, but by passage from the hen's intestinal tract to the reproductive tract and from there incorporation into the forming egg on the vitelline membrane, in the egg white or the shell membranes. It turns out that many different Salmonella serotypes can pass from the intestine of the chicken into its blood stream. Even passage from the blood stream into the hen's reproductive tract is not a unique characteristic of SE. Apparently specific to SE, however, is its capacity to survive the attacks by antimicrobial molecules during the formation of the egg in the hen's oviduct. This appears to require a combination of genes or gene expression patterns encoding for improved cell wall protection and damage repair, among others. The exact reason for the epidemiological association of SE with eggs is, however, still undefined.


The authors would like to express their appreciation to Isabel de Smet, who designed the figures (http://www.isabeldesmet.be).


  • Editor: Simon Cutting


View Abstract